Split (1)

train · 100k rows

id stringlengths 3 9	source stringclasses 1 value	version stringclasses 1 value	text stringlengths 1.54k 298k	added stringdate 1993-11-25 05:05:38 2024-09-20 15:30:25	created stringdate 1-01-01 00:00:00 2024-07-31 00:00:00	metadata dict
247999908	pes2o/s2orc	v3-fos-license	Disparities in surgical outcomes for low socioeconomic status patients in Australia Abstract Background There are disparities in surgical outcomes for patients of low socioeconomic status globally, including in countries with universal healthcare systems. There is limited data on the impact of low socioeconomic status on surgical outcomes in Australia. This study examines surgical outcomes by both self‐reported unemployment and neighbourhood level socioeconomic status in Australia. Methods A retrospective administrative data review was conducted at a tertiary care centre over a 10‐year period (2008–2018) including all adult surgical patients. Multivariable logistic regression adjusting for year, age, sex and Charlson Comorbidity Index was performed. Results 106 197 patients underwent a surgical procedure in the decade examined. The overall adverse event rates were mortality (1.13%), total postoperative complications (10.9%), failure to rescue (0.75%) and return to theatre (4.31%). Following multivariable testing, unemployed and low socioeconomic patients had a higher risk of postoperative mortality (OR 2.06 (1.50–2.82), OR 1.37 (1.15–1.64)), all complications (OR 1.43 (1.31–1.56), OR 1.21 (1.14–1.28)), failure to rescue (OR 2.03 (1.39–2.95), OR 1.38 (1.11–1.72)) and return to theatre (OR 1.42 (1.27–1.59), OR 1.24 (1.14–1.36)) (P < 0.005 for all). Conclusions Despite universal healthcare, there are disparities in surgical adverse events for patients of low socioeconomic status in Australia. Disparities in surgical outcomes can stem from three facets: a patient's access to healthcare (the severity of disease at the time of presentation), variation in perioperative care delivery, and social determinants of health. Further work is required to pinpoint why these disparities are present and to evaluate the impact of strategies that aim to reduce disparities. Introduction There is a well-established link between low socioeconomic status and increased morbidity and mortality. [1][2][3] In the United States (US), patients with a low socioeconomic status have higher rates of postoperative complications, mortality, failure to rescue and readmissions. [4][5][6][7][8][9][10][11][12][13][14] Some studies suggest that these disparities may be attributable to inequalities in the US healthcare system, rather than differential treatment of patients within individual care settings. 9,15 Hospitals disproportionately treating patients with lower socioeconomic status have worse outcomes when compared to other hospitals. 9 One might expect that countries with universal healthcare settings may not encounter socioeconomic disparities in surgical outcomes. However, studies conducted in several countries with universal healthcare systems, including the United Kingdom, [16][17][18] Finland, 19 the Netherlands, 20 Italy 21 and New Zealand, 22 have found disparities in surgical outcomes for patients of low socioeconomic status. There are limited studies examining surgical outcomes by socioeconomic status in Australia. This study aimed to examine surgical outcomes by both self-reported unemployment and a neighbourhood level socioeconomic index at a regional tertiary care centre in Australia. Study design A retrospective study was conducted at a regional tertiary care centre in Queensland, Australia. Data were examined over a 10-year period between January 2008 and August 2018. All adult (ages >18 years) patients who underwent a surgical procedure at this centre were included. The Townsville Hospital and Health Service Human Research Ethics Committee in Australia granted ethics approval (HREC/QTHS/57820). This approval included a patient consent waiver due to the retrospective study design. Data source Two hospital-based databases were utilized. The Operating Room Management Information System (ORMIS) was used to extract operative details for all surgical procedures performed. Patient identification numbers were then matched to the Hospital Based Corporate Information System (HBCIS) to extract further administrative hospital data. Socioeconomic status variables HBCIS includes a free text self-reported occupation variable. This variable was used to identify responses indicating unemployment. Various synonyms for unemployment, spellings of unemployment and names of unemployment social security benefits (Job Seeker, working for the Dole and Newstart) were included. Only those in the labour force were coded as unemployed; those indicating that they were not in the labour force (e.g., disability support/pensions, stay at home parents, students, retirees or pensioners) were not coded as unemployed. To attain a neighbourhood marker of socioeconomic status, a patient's residential address was linked to 2016 Census tract Index of Economic Resources (IER) data. 23 This index focuses on financial aspects of socioeconomic advantage and disadvantage. For this study, the bottom three national IER deciles (those with the greatest relative lack of access to economic resources) were compared to the top three national deciles. Outcomes Operative mortality was defined as the rate of death occurring in the hospital or within 30 days of surgery. Complications included one or more occurrences of postoperative acute renal failure, acute myocardial infarction, bleeding requiring transfusion of four or more units of red cells within the first 72 h of surgery, cardiac arrest requiring cardiopulmonary resuscitation, coma of 24 h duration or more, deep vein thrombosis, fever, unplanned intubation or ventilation use for more than 48 h, pneumonia, pulmonary embolism, respiratory failure, major wound disruption, surgical site infection, sepsis or the systemic inflammatory response syndrome, septic shock, or return to the operating theatre. The occurrence of one or more of the above listed complications was counted as the total complication rate for each procedure. Failure to rescue is the inability to rescue a patient from death after a postoperative complication. 7 Failure to rescue was coded as those with both a postoperative complications and postoperative mortality. Rates of unplanned readmission within 28 days after discharge were also examined. Covariates The Charlson Comorbidity Index (CCI) was used to adjust for preexisting comorbidities. The index includes myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatic disease, peptic ulcer disease, mild liver disease, diabetes with/without chronic complications, hemiplegia or paraplegia, renal disease, any malignancy including lymphoma and leukaemia (except malignant neoplasm of the skin), moderate or severe liver disease, metastatic solid tumour and AIDS/HIV. These variables were derived using ICD 10-AM (Australian modification) diagnosis codes assigned during the episode of care. 24 A weighted score was then assigned to each comorbidity. A score of zero indicates that no comorbidities were found, the higher the score, the more comorbidities were identified. Statistical analysis For group comparisons, Pearson's chi-squared tests were used for categorical variables, while continuous variables were analysed with analysis of variance (ANOVA) tests. Multivariable logistic regression models based on a conceptual model were used. The first analysis included the independent variables procedure year, patient age, sex and CCI. The second analysis also included adjustment for emergency surgery status. These regressions were performed once for each outcome examined (e.g., mortality, all complications, FTR etc.). Area under the curve (AUC) with 95% confidence intervals were calculated for all regression models. All analyses were performed using Stata 14/MP statistical software package (StataCorp, College Station, TX). All tests were done two-sided. The level of significance was set to 0.05. Area under the curve The AUC was overall low in univariate regressions and much higher in the multivariate regression (Table 2). Socioeconomic status may impact healthcare access and health seeking behaviour, both in an acute illness and across a patient's lifespan. There are disparities in access to surgical care for low socioeconomic patients. 25,26 Delayed presentations, such as more advanced stage of cancer or more severe peripheral arterial disease, may necessitate higher risk procedures. 25 This analysis adjusted for rates of emergency surgery (a proxy for delayed surgical access). 25,27,28 Adjusting for rates of emergency surgery decreased the odds of adverse events in low socioeconomic patients for all outcomes examined; however the only outcome where the disparity was no longer statistically significant after adjustment was acute renal failure in unemployed patients. The impacts of a relative lack of access to healthcare across a lifespan cannot be adjusted away using one proxy for delayed surgical access. Variation in care may stem from within a hospital or from hospital level structural differences. This study involved only one hospital and found significant disparities in surgical outcomes for patients of low socioeconomic status. This eliminates the potential variation in care from hospital level structural differences and suggests potential variation in care within a hospital. In the United States, studies have identified several racial disparities in surgical process measures including the administration of best-practise venous thromboembolism prophylaxis 29 and prescription of beta blockers at discharge following cardiac surgery. 30 In contrast to this, a study in the United Kingdom examining disparities in 30-day laparotomy for low socioeconomic status patients did not find variations in patient level performance in standards of care (e.g., reviewed by a consultant surgeon within 14 h of admission, appropriate time from to arrival to theatre). 17 Further research is required to determine if potential disparities in processes of care may contribute to the disparities in surgical outcomes for patients of low socioeconomic status in Australia. Unconscious provider bias may contribute to healthcare disparities. International studies have shown that unconscious preferences for white race and upper social class are prevalent among registered nurses, medical students, trauma and acute care surgeons. 31-34 Unconscious preferences have not been shown to correlate with clinical decision-making. [31][32][33][34] However, clinician implicit bias is associated with markers of poor visit communication, poor patient ratings of care and patients' perceptions of recommended treatments. 35,36 These factors may have downstream effects on clinical care. Socioeconomic status is associated with health behaviours (smoking, obesity) and associated comorbidities (hypertension, diabetes) which may affect the risks of surgery. This study adjusted for the CCI but did not have data on individual health behaviours like smoking status, which may impact surgical outcomes. Future studies should examine the impact of individual health behaviours, such as smoking, on disparities in surgical outcomes for low socioeconomic patients. In this study prior to risk adjustment unemployed patients had lower rates of mortality and acute renal failure. However, unemployed patients were significantly younger and had significantly fewer comorbidities when compared to other patients. After adjustment for these variables unemployed patients had significantly higher rates of postoperative mortality and acute renal failure. This example highlights the importance of adjusting for relevant confounding factors. There are three proposed phases of health disparities research; detecting, understanding and reducing. 37 In the first phase, vulnerable populations are defined, the outcomes being examined are defined and these outcomes are measured for the vulnerable populations to detect potential disparities. The second phase is understanding why the disparities are present; determinants can be identified at several levels: the patient, clinical encounter, healthcare system or broader public policy. In the final phase, strategies that aim to reduce disparities are implemented and their impact is monitored utilizing the disparity sensitive metrics defined in the first phase. 37 This paper focuses on the first phase, detecting disparities. Further work is required to better understand the aetiology of these disparities and evaluate the impact of strategies that aim to reduce the disparities. Although the AUC was low in univariate regressions it was much higher in multivariate regressions indicating that our multivariate models managed to incorporate most of the important variables related to the outcomes. This study used residential postcodes to link to census tract data on residential socioeconomic status. There are more granular residential areas (statistical local areas) that can be linked to census tract data. Residential measures of socioeconomic status perform more consistently when more granular areas are used rather than postcodes. 38,39 The IER may have been a more reliable indicator of socioeconomic status if residential statistical local areas were linked to the IER data. Aboriginal and/or Torres Strait Islander patients may experience higher rates of postoperative adverse events. [40][41][42][43][44] Aboriginal and/or Torres Strait Islander patients have higher rates of unemployment and are more likely to reside in areas of socioeconomic disadvantage. 45 Socioeconomic disadvantage alone may be responsible for between one-third to one-half of the life expectancy gap for Aboriginal and/or Torres Strait Islander Australians. 46 The burden of disease rates for Aboriginal and/or Torres Strait Islander people are the highest in areas where the population is the most socioeconomically disadvantaged and falls with decreasing level of disadvantage. 47 Examining surgical outcomes for Aboriginal and/or Torres Strait Islander people, and exploring the confounding effects of socioeconomic status, is prudent. The data to examine surgical outcomes for Aboriginal and/or Torres Strait Islander people is readily available in hospital administrative databases. Despite acquiring the full Human Research Ethics Committee approval to include a variable on Aboriginal and/or Torres Strait Islander status in this project's dataset and examine for disparities in surgical outcomes for this group, the Queensland Government did not allow the acquisition of the Aboriginal and/or Torres Strait Islander status variable. Conclusion Despite Australia's universal healthcare system, there are disparities in surgical adverse events for unemployed patients and patients of low socioeconomic status. Disparities in surgical outcomes can stem from three facets: a patient's access to healthcare (the severity of disease at the time of presentation), variation in perioperative care delivery and social determinants of health or lifestyle factors. Further work is required to pinpoint the aetiology of these disparities, develop strategies to reduce the disparities and evaluate the impact of these strategies utilizing disparity sensitive surgical outcome metrics.	2022-04-08T06:22:45.149Z	2022-04-07T00:00:00.000	{ "year": 2022, "sha1": "dd8948035aa460153894938fa9459bef34be3452", "oa_license": "CCBYNC", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1111/ans.17675", "oa_status": "HYBRID", "pdf_src": "Wiley", "pdf_hash": "067de9e0969fed6a7d2802589195c83fc725d238", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
10494631	pes2o/s2orc	v3-fos-license	Many-Objective Pareto Local Search We propose a new Pareto Local Search Algorithm for the many-objective combinatorial optimization. Pareto Local Search proved to be a very effective tool in the case of the bi-objective combinatorial optimization and it was used in a number of the state-of-the-art algorithms for problems of this kind. On the other hand, the standard Pareto Local Search algorithm becomes very inefficient for problems with more than two objectives. We build an effective Many-Objective Pareto Local Search algorithm using three new mechanisms: the efficient update of large Pareto archives with ND-Tree data structure, a new mechanism for the selection of the promising solutions for the neighborhood exploration, and a partial exploration of the neighborhoods. We apply the proposed algorithm to the instances of two different problems, i.e. the traveling salesperson problem and the traveling salesperson problem with profits with up to 5 objectives showing high effectiveness of the proposed algorithm. Introduction Combinatorial optimization problems find numerous applications in transportation, logistics, scheduling, design, etc. (Yu, 2013). On the other hand, real-life optimization problems often require taking into account multiple points of view corresponding to multiple objectives. Since, multiple objective combinatorial optimization problems are usually NP-Hard, multiobjective metaheurics are often used to solve them approximately (Talbi et al., 2012). Effective single objective algorithms for combinatorial optimization problems usually use some kind of local search as one of their components. Pareto Local Search (PLS) (Angel et al., 2004;Paquete & Stutzle, 2006;Paquete et al., 2007) is a very natural extension of the concept of the local search to the multiobjective case. PLS works with a Pareto archive, i.e. a set of potentially Pareto-optimal solutions that are not dominated by any other solution obtained so far. PLS explores neighborhood of each solution in the archive and uses the neighbor solutions to update the Pareto archive. The effectiveness of PLS as a component of the hybrid algorithms may be explained by the fact that it has different characteristics than most other multiobjective metaheuristics. As noticed by Lara et al. (2010) each multiobjective metaheuristic should search both towards and along the Pareto front. The search towards the Pareto front means generating new solutions lying closer to the Pareto front than known solutions, and the search along the Pareto front means generating new potentially Pareto-optimal solutions improving the representation of the Pareto front. PLS is very effective in the search along Pareto front when started from a seed of high quality solutions. For example in the case of the bi-objective traveling salesperson problem such a seed may be generated with the use of the Lin-Kernighan (Lin & Kernighan, 1973) heuristic (Lust & Teghem, 2010;Lust & Jaszkiewicz, 2010;Liangjun et al., 2014;Cornu et al., 2017). The standard PLS algorithm becomes, however, very inefficient in the case of more than two objectives because the number of Pareto-optimal solutions grows very fast with the number of objectives. Thus PLS has to search neighborhoods of a huge number of solutions and the process of updating large Pareto archives with new solutions becomes very time-consuming. Indeed, with very few exceptions (e.g. Liangjun et al. (2014); Cornu et al. (2017)), majority of the publications on PLS concerns bi-objective optimization only. In this paper we propose a new Many-objective PLS algorithm (MPLS) dedicated to the problems with more than two objectives. Please note, that we use the term "Many-objective" in somehow non-standard way, since it usually describes problems with ≥ 4 objectives (Chand & Wagner, 2015). However, in the context of PLS, three objectives is already a high number. The proposed algorithm is based on the following three new mechanisms that differ it from the standard PLS: • The use of a recently proposed ND-Tree data structure (Jaszkiewicz & Lust, 2016) for the efficient update of the Pareto archive. • The use of a new mechanism for the selection of the promising solutions for the exploration of their neighborhoods based on the randomly selected weighted Chebycheff scalarizing functions. • The partial exploration of the neighborhoods, i.e. testing only some of the neighborhood solutions. The first mechanism reduces the time needed to update the Pareto archive, while the two latter mechanisms improve the quality of the archive with a limited number of the neighborhood solutions tested. We apply MPLS to two different multiobjective combinatorial optimization problems: the multiobjective traveling salesperson problem (MTSP) and the multiobjective traveling salesperson problem with profits (MTSPWP) with up to 5 objectives. Through a computational study we show that MPLS produces archives of much better quality than the standard PLS in the same time. We also show that each of the three new mechanisms described above contributes to the effectiveness of the proposed method. Furthermore, since for the bi-objective traveling salesperson problem the best results have been obtained by properly balancing the CPU time assigned to the Lin-Kernighan heuristic and to PLS (Lust & Teghem, 2010;Lust & Jaszkiewicz, 2010;Liangjun et al., 2014;Jaszkiewicz & Lust, 2017;Cornu et al., 2017) we test whether a similar situation holds in the case of MTSP with more than two objectives. The paper is organized in the following way. In the next section we describe the proposed Many-objective Pareto Local Search algorithm. Then the computational experiment is presented. In section 4, related works are discussed. Finally, conclusions and directions for further research are presented. Many-objective Pareto Local Search algorithm We start the description of the proposed MPLS algorithm with a short presentation of the standard PLS that is a basis for our algorithm. Then we describe the three new mechanisms used in MPLS. Finally, we present the whole algorithm of MPLS. Pareto Local Search PLS is a relatively straightforward adaptation of the idea of the single objective local search to the multiobjective case. In PLS the acceptance of a new neighborhood solution is based on the dominance relation. Consider a general multiobjective optimization problem with a feasible set X and d objective functions y k (x). Without loss of generality we assume that the objectives are maximized. The image of a feasible solution x in the objective space R d is a point y(x) = (y 1 (x), y 2 (x), . . . , y d (x)). We say that a point y 1 ∈ R d dominates a point y 2 ∈ R d if, and only if, y 1 j ≥ y 2 j ∀ j ∈ {1, . . . , d} ∧ ∃ j ∈ {1, . . . , d} : y 1 j > y 2 j . We denote this relation by y 1 y 2 . For the sake of simplicity we will use the dominance relation with respect to the corresponding solutions as well. The solutions not dominated by any other feasible solution are called Pareto-optimal and the image of the set of Pareto-optimal solutions in the objective space is the Pareto front. The set of distinct and mutually non-dominated solutions, i.e. solutions such that none of them dominates any other, is called a Pareto archive. In the context of multiobjective metaheuristics the Pareto archive is used to store the set of potentially Pareto-optimal solutions generated by the method till a given iteration. PLS works with a Pareto archive. It starts with some initial Pareto archive and then searches whole neighborhoods of all solutions whose neighborhoods have not been searched yet for new potentially Pareto-optimal solutions until no such new solution can be found in any of the neighborhoods (see Algorithm 1). 2.2. Mechanism I: Efficient update of the Pareto archive with the use of ND-Tree Whenever a new neighbor solution not dominated by the current solution is found, the Pareto archive needs to be updated. Updating a Pareto archive A with a new solution x means that: • x is added to A if it is not dominated nor equal to any solution in A, Algorithm 1 PLS Parameter : A: an initial Pareto archive where N (x) denotes the neighborhood of x and Update() updates the Pareto archive A. • all solutions dominated by x are removed from A. Since the neighborhood solutions for many combinatorial problems may be generated in a very short time and the number of Pareto-optimal solutions grows fast with growing number of objectives, the process of updating the Pareto archive may easily become the main factor influencing the running time of PLS. The simplest approach to the update the Pareto archive is to organize it as a simple list of solutions and compare the new solution to each solution in this list until a dominating or equal solution has been found or all solutions have been compared. This approach is, however, very time consuming for large Pareto archives. In the bi-objective case the Pareto archive may be efficiently updated using the binary search in an archive sorted according to the values of the objectives (Jaszkiewicz & Lust, 2016). This approach, however, cannot be extended to the case of more than two objectives. In MPLS we use a recently proposed ND-Tree data structure (Jaszkiewicz & Lust, 2016). In ND-Tree the Pareto archive is recursively split into disjoint subsets. For each subset approximate local ideal and approximate local nadir points, i.e. points in the objective space respectively dominating and dominated by all solutions in this subset, are kept. In other words, all solutions from each subset are contained in the hyper-box defined by the two points. These hyper-boxes may in general overlap, but the tree is built such that the subsets contain close solutions contained in small hyper-boxes. By comparing the new solution to the approximate local ideal and the approximate local nadir points many branches of the tree can be omitted. The computational complexity of updating the Pareto archive with ND-Tree is sub-linear with respect to the size of the archive for any number of objectives under mild assumptions (Jaszkiewicz & Lust, 2016). Thus the overall running time of MPLS can be significantly reduced with the use of this data structure. 2.3. Mechanism II: Selection of the promising solutions for the exploration of their neighborhoods The standard PLS algorithm ends-up with a Pareto-archive being locally optimal, i.e. when no solution being a neighbor of any solution in the archive can improve this archive. Obtaining a locally optimal Pareto-archive is, however, usually prohibitively time-consuming in the case of three and more objectives even with the use of ND-Tree. On the other hand, the algorithm when stopped earlier may produce an archive of a low quality, because there could be high variations in the CPU times spent in exploring various regions of the Pareto front. Dubois-Lacoste et al. (2011 proposed anytime PLS algorithm for the bi-objective case whose goal is to generate archives of a good quality at any iteration. One of the main mechanisms of that algorithm is the selection of solutions whose neighborhood exploration has a high potential to improve the current archive. Dubois-Lacoste et al. use the so-called optimistic hypervolume to select such solutions in the bi-objective case. This mechanism, however, cannot be directly extended to the case of of more than two objectives. Thus, we propose a new selection mechanism described below. In each iteration we would like to generate a new solution highly improving the quality of the Pareto archive. Since the neighbor solutions are usually similar to the original solutions, it is natural to expect that new good solutions may be found in the neighborhoods of known good solutions. Assume that an indicator of the quality of the Pareto archive is used. A good solution is a solution that substantially contributes to the value of this indicator, i.e. its removal would substantially deteriorate the value of this indicator. A number of indicators for the evaluation of the quality of Pareto archives have been proposed. An often used quality indicator is the hypervolume of the space dominated by the set of points (see e.g. (Zitzler et al., 2003)). Our approach is motivated by another quality indicator -the expected value of the weighted Chebycheff scalarizing functions (Jaszkiewicz, 2002). Such functions are defined in the following way: where y 0 is a reference point, Λ = [λ 1 , ..., λ d ] is a weight vector such that λ k ≥ 0 ∀k. Each weighted Chebycheff scalarizing function has at least one global minimum belonging to the set of Pareto-optimal solutions. For each Paretooptimal solution x there exists a weighted Chebycheff scalarizing function s ∞ such that x is a global minimum of s ∞ (see (Steuer, 1986), ch. 14.8). The quality indicator is defined as: where L is a set of all weight vectors. Although the two indicators, i.e. the hypervolume and the expected value of the weighted Chebycheff scalarizing functions, are defined in quite different terms, they are in fact closely related, which is illustrated in Figure 1. Consider the non-dominated border of the dominated area used in the hypervolume indicator. Each point lying on this border is optimum of one or more weighted Chebycheff scalarizing functions and has the same values of these scalarizing functions as some of the solutions in the archive. In other words, the hyper-facets of this border belong to the same isoquants of multiple Chebycheff functions as some solutions from the archive. In result, any improvement of the hypervolume implies an improvement of the best values of some weighted Chebycheff scalarizing functions. An important advantage of the expected value of the weighted Chebycheff scalarizing functions indicator is that to select good solutions we do not need to explicitly calculate the contribution of each solution to this quality indicator. Instead in each iteration we draw a random weight vector defining a weighted Chebycheff scalarizing function and select from the Pareto archive the solution that minimizes the value of this function. Thus, solutions being minima of many scalarizing functions, and so having high impact on the This mechanism not only selects promising solutions but also assures a uniform coverage of all regions of the objective space. If the solutions were selected at random with uniform probability, regions that have more dense representation in the current archive would have a higher chance to be explored further. Thus, any random differences in the density would tend to be further increased, since the dense regions would be explored more than regions with a low density of solutions. Naive approach for finding the solution minimizing a given scalarizing function through evaluation of each solution would be very time consuming. However, ND-Tree data structure used to update the Pareto archive may also be used to efficiently find the solution with the best value using Algorithm 2 proposed in this paper. Let us remind that in ND-Tree the Pareto archive is recursively split into disjoint subsets associated with approximate local ideal points. The algorithm is based on the observation that the value of each weighted Chebycheff scalarizing function for any solution in a subset is not lower than its value for the approximate local ideal point of this subset. Thus, if a value of this function for a solution is already known (upper bound) and this value is not worse than the value for the approximate local ideal point, the whole subset could be omitted (see Figure 2). If this subset corresponds to an internal node, its whole subtree could be omitted. To find quickly a solution with a good value of the scalarizing function defining a good upper bound at each internal node we select first the sub-node with the best value for its approximate local ideal point. Since Algorithm 2 is a new contribution reported in this paper, we analyze the computational complexity of this algorithm. This analysis is similar to the analysis of the algorithm for updating the Pareto archive with ND-Tree presented in (Jaszkiewicz & Lust, 2016). We start by formally defining the ND-Tree data structure : Definition 1. ND-Tree data structure is a tree with the following properties: 1. With each node n is associated a set of solutions S(n). 2. Each leaf node contains a list L(n) of solutions and S(n) = L(n). 3. For each internal node n, S(n) is the union of disjoint sets associated with all sub-nodes of n. 4. Each node n stores an approximate ideal point y * (S(n)) such that y * (S(n)) x ∀x ∈ S(n) and a approximate nadir point y * (S(n)) such that x y * (S(n)) ∀x ∈ S(n). 5. If n is a sub-node of n, then y * (S(n)) y * (S(n )) and y * (S(n )) y * (S(n)). In the worst case, at each intermediate node we need to analyze each subnode. Thus, all solutions will need to be evaluated and the computational complexity of Algorithm 2 is Θ(N ) where N is the number of solutions in the archive. In the best case, at each intermediate node only one sub-node has to be analyzed and the solutions are equally split into a predefined number of subnodes. In this case the number of evaluated solutions and points is described by the following recurrence: where C is the number of sub-nodes. Please note, that it is possible to consider even more optimistic scenarios, e.g. when only one solution needs to be evaluated, however, the scenario analyzed above is much more realistic. The most interesting but also the most difficult to analyze or even define is the average case. Consider the case when each internal node has exactly two sub-nodes. Assume that the number of solutions in one of the sub-nodes is drawn from the uniform distribution, i.e. each number of solutions is equally likely. The other sub-node will contain the remaining number of solutions. In an internal node, either one or both sub-nodes will be processed. Assume that the probability p 2 of processing both nodes is constant. In this case: Multiply both sides by N : Assume that N ≥ 2: Subtract equations 3 and 4: The above recurrence has the following solution: The Algorithm 2 is sub-linear for any p 2 < 1 under the above assumptions. Of course, it is still a very simplified analysis of the practical behavior of the algorithm. For example, the probability p 2 will not, in general, be constant. In fact, it may happen that none of the sub-nodes will need to be processed, since the ideal points of all sub-nodes will have the values of the scalarizing function above the current upper bound. Such situations, would further improve the practical efficiency of the algorithm. Mechanisms III: Partial exploration of the neighborhoods In the standard PLS the whole neighborhood of each solution is explored. This strategy is not well adapted, however, to the goal of obtaining a good anytime behavior since it may lead to the situations where some regions of the Pareto front will be intensively explored while other regions will be underexplored. Dubois-Lacoste et al. (2011 observed that the anytime On the other hand, the algorithm may become ineffective if too few neighbor solutions are tested. Please note, that before exploring a neighborhood, the solution for the exploration needs to be selected which takes non-negligible CPU time even with the use of ND-Tree. Thus the optimum strategy may be to test a randomly selected subset of neighbor solutions. Dubois-Lacoste et al. consider several options for partial exploration of the neighborhoods. Our approach is slightly different since we use the number of tested moves as a parameter of the method. MPLS algorithm The proposed method is summarized in Algorithm 3. As the reference point of the weighted Chebycheff scalarizing functions we use the point composed of the maximum values of particular objectives in the current archive increased by 10% of the range of a given objective in this archive. To draw a random weight vector we use the algorithm proposed in (Jaszkiewicz, 2002). Weighted Chebycheff scalarizing functions are applied to the objective values normalized on-line to the range [0, 1] based on the ranges of the objectives in the current archive, which is in fact achieved by dividing each individual weight by the range of the corresponding objective. where N (x) denotes the neighborhood of x and UpdateNDTree() updates the Pareto archive A using ND-Tree data structure. Computational experiment To test the effectiveness of the proposed algorithm we use two different combinatorial optimization problems. The source code, the instances, and the detailed numerical results are available on-line 1 . Multiobjective Traveling Salesperson Problem As one the test problems we use the Multiobjective Traveling Salesperson Problem (MTSP) which is a typical benchmark problem for multiple objective metaheuristics. Given a set {v 1 , v 2 , · · · , v N } of nodes and d costs c 1 (v i , v j ) . . . c d (v i , v j ) between each pair of distinct nodes {v i , v j }, the multiobjective traveling salesperson problem (MTSP) consists of finding an order π of the nodes, minimizing the following costs (k = 1, . . . , d): In the computational experiment, we used the symmetric multioobjective traveling salesperson problem with c We used Euclidean instances with 100 nodes in which the costs of the edges correspond to the Euclidean distance between two points in a plane, and each objective corresponds to a different plane. For the purpose of this experiment, we generated 10 instances for each number of objectives. In the first phase we used Lin-Kernighan heuristic (Lin & Kernighan, 1973). Precisely we used the efficient implementation of this heuristic from the Concorde project (http://www.math.uwaterloo.ca/tsp/concorde.html). In the first phase, we used 1000, 2000, and 3000 random weight vectors for d = 3, 4, 5, respectively. For each of the weight vectors the Lin-Kernighan heuristic was run optimizing a weighted sum of the objectives. We used the 2-edge exchange move in MPLS. Alike Lust & Jaszkiewicz (2010) and Jaszkiewicz & Lust (2017) we used the mechanisms of candidate moves to speed-up the calculations. Comparison of the various versions of PLS In the first experiment we compare MPLS to the standard PLS and to several other versions of PLS. Particular versions of PLS allow us to show not only that MPLS performs better than the standard PLS but also that each of the three new mechanisms used in MPLS indeed influences its effectiveness. The following methods are compared in this experiment: • "MPLS 100" described by Algorithm 3 with 100 random moves tested in each neighborhood, which is a relatively good value of this parameter according to our preliminary experiments. • "MPLS full neighborhood" -the algorithm equivalent to "MPLS 100" but with the full exploration of each neighborhood. This algorithm allows us to show that the partial exploration of the neighborhoods improves the performance of MPLS. • "MPLS 1" described by Algorithm 3 with just one random move tested in each neighborhood. We use this algorithm to show that too small number of tested moves deteriorates the effectiveness of MPLS. • "MPLS 100 Random" -the algorithm equivalent to "MPLS 100" but with the random selection of solutions with uniform probability. This algorithm allows us to show that the selection of solutions based on the randomly selected weighted Chebycheff scalarizing functions improves the performance of MPLS. • "MPLS 100 List" -the algorithm equivalent to "MPLS 100" but with the Pareto archive organized as a simple list. This algorithm allows us to show that the use of ND-Tree data structure improves the performance of MPLS. • "Standard PLS tree" described by Algorithm 1 but using ND-Tree data structure. This algorithm allows us to show that the selection of solutions based on the randomly selected weighted Chebycheff scalarizing functions and the partial exploration of the neighborhoods improves the performance of MPLS. • "Standard PLS List" described by Algorithm 1. This algorithm is the straightforward adaptation of the algorithm used in the bi-objective case to the many-objective case without any of the three new mechanisms used in MPLS. The results of this experiment are presented in Figures 3 to 5. The reported running times include also the running times of the first phase. For each instance, the maximum running time in the second phase was equal to the running time of the first phase. The values of the quality indicators were calculated in time steps equal to 10% of the running time of the first phase. The running times were averaged over 10 instances. Since the variations of the running times of the first phase were relatively low, we report just the average times. As the quality measure we report the hypervolume indicator, but we calculated also the average values of the weighted Chebycheff scalarizing functions (i.e. approximation of the expected value) with exactly the same conclusions. We show also the range of obtained values of the hypervolume indicator on 10 different instances for each number of objectives. The first point always corresponds to the Pareto archive obtained after the first phase. Since we used different instances we normalize the values of the hypervolume such that it is 1 after the first phase. The main observations are: • The relative differences between the methods grow with the growing number of objectives. In other words, the more objectives the more important are the three new mechanisms used in MPLS. • Each of the three new mechanisms of MPLS is important for the final performance of the method. The comparison of "MPLS 100" to "MPLS 100 List" indicates that the mechanism with the highest influence on the effectiveness of MPLS is the use of ND-Tree data structure to update the Pareto archive and to select the best solution for a given scalarizing function. The comparison of "MPLS 100" to "Standard PLS Tree" indicates that the second mechanism with the highest influence on the effectiveness of MPLS is the selection of solutions based on the randomly selected weighted Chebycheff scalarizing functions. Furthermore, the comparison to "MPLS 100 Random" indicates that the random selection of solutions with uniform probability is even worse that the selection mechanism used in the standard PLS. The comparison of "MPLS 100" to "MPLS Full neighborhood" and "MPLS 1" indicates that the partial exploration of the neighborhoods also substantially improves the performance of MPLS. • "Standard PLS List", i.e. the algorithm that uses none of the three new mechanisms used in MPLS is the worst algorithm in all cases. In order to test the significance of the differences we used the nonparametric statistical test of Mann-Whitney (Ferguson, 1967). The level of risk of the test has been fixed to 5%. The final results of "MPLS 100" were significantly better than the final results of all other methods in all cases except of "MPLS full neighborhood" with d = 3. MPLS with various number of runs of the Lin-Kernighan heuristic In the above experiment we have shown that MPLS performs better than other versions of PLS. The question remains, however, whether it is a good method in absolute terms. Since, in the bi-objective case the best results have been obtained combining the phases of the Lin-Kernighan heuristic and of PLS (Lust & Teghem, 2010;Lust & Jaszkiewicz, 2010;Liangjun et al., 2014;Jaszkiewicz & Lust, 2017;Cornu et al., 2017), we test whether MPLS can improve the results obtained with Lin-Kernighan heuristic standalone also in the many-objective case. Precisely, we ask the following question: Having some predefined CPU time, can we obtain the best result with the use of the Lin-Kernighan heuristic standalone, or by combining it with MPLS in two phases. Figure 4: Hypervolume indicator for 4-objective MTSP instances In the first phase, we used 2000, 4000, and 6000 as the maximum number of random weight vectors for d = 3, 4, 5, respectively. For each of the weight vectors the Lin-Kernighan heuristic was run optimizing a weighted sum of the objectives. The time needed for all runs of the Lin-Kernighan heuristic was then used as the maximum running time. Then, the number of Lin-Kernighan heuristic runs was reduced to 90%, . . . , 10% and the remaining running time was allocated to MPLS. The results are presented in Figures 6 to 8. The values of the hypervolume are again normalized such that it is 1 after the first phase with the maximum number of the Lin-Kernighan heuristic runs. Please note, that for the lower numbers of the Lin-Kernighan heuristic runs the quality after the first phase is not shown since it was much lower and the figures would become incomprehensible. The main observations are: • In each case, the results with the use of MPLS are better than with the use of the Lin-Kernighan heuristic standalone even with the lowest number of runs of the Lin-Kernighan heuristic. • The overall best results were obtained for 1000, 2400, and 4200 runs of the Lin-Kernighan heuristic, for d = 3, 4, 5, respectively, i.e. the best results were obtained with 50% or less of the CPU time allocated to MPLS. In other words, alike in the bi-objective case the best results are obtained by properly balancing the CPU time allocated to the phases of the Lin-Kernighan heuristic runs and MLPS. Please note, that the Lin-Kernighan heuristic, and especially its efficient implementation from the Concorde project, is an advanced method for TSP, while MPLS uses the very basic 2-edge exchange move. • The standard PLS algorithm was able to improve the results obtained with the Lin-Kernighan heuristic standalone only in the 3-objective case. In the 4-and 5-objective cases, the standard PLS algorithm was not able to improve the results obtained with the Lin-Kernighan heuristic standalone for any reduction of the number of runs in the first phase. Thus, the use of this algorithm in these cases would not be beneficial. between each pair of distinct nodes {v i , v j }, and d 2 profits pr 1 (v i ) . . . pr d 2 (v i ) associated with each node {v i }, the multiobjective traveling salesperson problem with profits (MTSPWP) consists of choosing a subset of nodes SN and finding an order π of these nodes, optimizing the following objectives: In other words, d 1 objectives correspond to the minimization of d 1 costs of the routes, while d 2 objectives correspond to the maximization of d 2 profits associated with the selected nodes. We used the symmetric costs where c For the purpose of this experiment we generated 10 instances with 100 nodes for each number of objectives. For d = 3, we used one cost objective and two profit objectives, for d = 4, we used two cost objectives and two profit objectives, and for d = 5, we used two cost objectives and three profit objectives. The cost objectives were generated using Euclidean distances, i.e. the costs of the edges correspond to the Euclidean distance between two points in a plane, and each objective corresponds to a different plane. Profits were drawn at random with a uniform distribution from a range [0,2000]. Please note, that despite of the similar names MTSPWP differs substantially from MTSP. In particular: • The two problems have different solution spaces. In MTSPWP the solutions are defined by both a choice of the nodes, and an order of the selected nodes. In result, MTSPWP requires a different set of the neighborhood moves described below. • MTSPWP exhibits a specific pattern of positive/negative correlations of objectives. The cost objectives are in general positively correlated because all of them improve with reducing the number of selected nodes. The profit objectives also are in general positively correlated because • The cost and profit objectives are defined by heterogeneous mathematical formulas. Because of the nature of MTSPWP, several types of moves were used to both select appropriate subset of nodes and optimize the route between these nodes (Jozefowiez et al., 2008): • "2-edge exchange move" is the same move as in the case of MTSP. • "Node delete move" removes one of the nodes from the current solution. • "Node insert move" inserts one of the nodes that is not selected at a given position in the route. • "Node exchange move" exchanges one of the nodes from the current solution for one of the nodes that is not selected. In the first phase we used the steepest local search algorithm that was testing all moves of all types before selecting the best neighbor. In MPLS the type of the move was drawn at random, and then the specific move was drawn. The steepest local search was applied to the optimization of the combined scalarizing functions being the weighted sums of the Chebycheff scalarizing function with weight 1 and the linear scalarizing function with weight 0.5, which was found the best option in the preliminary experiments. We have performed the two experiments analogous like in the case of MTSP with same numbers of local search runs like the numbers of the Lin-Kernighan heuristic runs. Despite of the different nature of the two problems the results of the experiments for MTSPWP are very similar to the results for MTSP. Because of the limited space we present only the results for the 5-objective instances in Figures 9 and 10. The main difference with respect to the results for MTSP is that for MTSPWP the "MPLS full neighborhood" method is not significantly worse than "MPLS 100", although the average values are worse for "MPLS full neighborhood". One of the main ideas of MPLS is to improve the anytime behavior of the algorithm, since obtaining the locally optimum Pareto archive is usually prohibitively time consuming in the many-objective case. An anytime PLS algorithm has been proposed by Dubois-Lacoste et al. (2011. This algorithm is, however, dedicated to the bi-objective case, and the authors define many-objective optimization as a direction for further research. At a general level some mechanisms used in MPLS are motivated by these works, but the specific techniques are very different. For example Dubois-Lacoste et al. use the optimistic hypervolume to select the promising solutions, which cannot be directly applied in the many-objective case. Liangjun et al. (2014) used PLS to solve the multiobjective knapsack problem instances with up to 4 objectives. Cornu et al. (2017) applied PLS to the bi-and 3-objective instances of TSP. These papers do not contain, however, information whether any specific mechanisms for the many-objective case, other than the reduction of the time allocated to PLS, were used. The idea of using the scalarizing functions with the randomly generated weight vectors has been proposed in some multiobjective evolutionary algorithms, e.g. by Ishibuchi & Murata (1998) and Jaszkiewicz (2002). The scalarizing functions were, however, used in a very different way to select solutions for recombination and to optimize the scalarizing functions with the single objective local search. The traveling salesperson problem with profits is a well-known combinatorial optimization problem, however, it is usually considered as a single objective problem with a single cost and a single profit aggregated to one objective with a weighted sum approach (Feillet et al., 2005). There are relatively few results for the bi-objective case (Jozefowiez et al., 2008;Zhu et al., 2011;Labadie et al., 2014;Bérubé et al., 2009). To our knowledge, multiobjective metaheuristics have not been applied yet to the multiobjective version of the traveling salesperson problem with profits with more than one cost and/or profit. Conclusions and directions for further research We have presented a new Many-Objective Pareto Local Search algorithm. The algorithm has been tested on two different multiobjective combinatorial optimization problems showing its effectiveness. We have also shown that each of the three new mechanisms used in MPLS significantly contributes to its effectiveness. Furthermore, the importance of the new mechanisms in relation to the standard PLS grows with the growing number of objectives. In particular, in the 4-and 5-objective cases, standard PLS without any of the new mechanisms was not able to improve the results obtained with the Lin-Kernighan heuristic (MTSP) or local search (MTSPWP) standalone for any reduction of the number of runs in the first phase. On the other hand, MPLS was able to improve these results in each case tested in the experiment. Since the standard PLS algorithm proved its high effectiveness in the bi-objective case, we believe that MPLS could be a very useful element in the toolbox for the general case of the multiobjective combinatorial optimization, especially when combined with other methods being very effective in the search towards the Pareto front. In fact, we expect that alike the standard PLS in the bi-objective case, MPLS could become a typical component of the state-of-the-art algorithms for the many-objective combinatorial optimization. The applications of MPLS to various benchmark and real-life problems, and combination with other multiobjective metaheuristics is a natural direction for further research. In particular MPLS could be combined with such concepts as the use of larger perturbation of solutions (Liangjun et al., 2014) or data perturbations (Cornu et al., 2017). MPLS introduces a new parameter, i.e. the number of the neighborhood moves. At present we do not have any guidelines for setting this parameter. Development of a mechanism for automatic adaptation of this parameter is an interesting direction for further research. MPLS generates a large number of potentially Pareto-optimal solutions. The maximum size of the Pareto archive in our experiment exceeded 2 millions of solutions. The final goal of the decision maker is to select the single solution being the best according to his/her preferences. If no information about these preferences is available, a multiobjective metaheuristic should generate a good solution for any possible preferences. However, often some partial preference information is known in advance or may be gathered during the run of the method. Thus, another interesting direction for further research is the incorporation of some partial preference information in order to limit the region of the Pareto front searched by MPLS.	2017-07-25T10:29:04.000Z	2017-07-25T00:00:00.000	{ "year": 2018, "sha1": "869949f95bc14f81dc4331c6f8e3d845ea4cd721", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1707.07899", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "1ee03ff748560037dc2b7028b143ff6d3fd58dcb", "s2fieldsofstudy": [ "Computer Science", "Mathematics" ], "extfieldsofstudy": [ "Computer Science", "Mathematics" ] }
256126403	pes2o/s2orc	v3-fos-license	Antimicrobial Prescribing Practices in Hospital Settings in Italy: A Retrospective Study Background: This study aims to evaluate the antimicrobial prescribing practices in hospital settings in Italy, focusing on the appropriateness of antibiotic use. Methods: This study was carried out through a retrospective review of medical records of patients admitted in three public hospitals located in Campania Region (Italy) between 1 January and 31 December 2018. Results: More than one third (34.2%) of patients received at least one inappropriate antibiotic prescription (antibiotic administered and not indicated). Being female, having a >1 Charlson comorbidity index score, and having a longer hospital stay were significant determinants of an inappropriate antibiotic prescription. Instead, patients who had had a non-urgent hospital admission, an infection on hospital admission, and a microbiological culture test during hospital stay were significantly less likely to have an inappropriate prescription. When the antibiotic prescriptions were analyzed, in 26.6% of cases they were not indicated, while among the 687 antibiotic prescriptions with indication, incorrect choice of antibiotics (36.8%) was the most common reason of the inappropriateness. Conclusions: The findings of the study indicate that the inappropriate use of antibiotics continues to be a relevant issue in the hospital setting and specific interventions are needed to encourage a wider utilization of diagnostic tools to practice targeted therapies and to counter the antimicrobial resistance. Introduction Antimicrobial resistance (AMR) represents a main health issue worldwide and the World Health Organization (WHO) has defined it as one of the greatest threats to public health due to its epidemiological relevance and economic burden [1]. The growing increase of antibiotic-resistant bacteria, which impairs the effective treatment of related infectious diseases, results in a sequence of negative effects on patients and the community. Indeed, AMR determines an increase in the morbidity, duration, severity, and mortality of infectious diseases, as well as failure of surgical interventions with a consequent rise in healthcare costs due to the duration of hospitalization, frequent diagnostic investigations, and longer and complex antimicrobial treatments [2]. Furthermore, the effects of infectious diseases caused by multi-resistant microorganisms impacts on the overall well-being of the community due to the indirect costs for patients and their families, such as the loss of working days and disability [3]. To contrast AMR, the WHO has approved a global action on AMR to ensure the successful treatment of infectious diseases through several objectives, such as the improvement of understanding of AMR, the optimization of the use of antimicrobial agents, and the increase in the investments in the research on new antibiotics, diagnostic tools, and vaccines [4]. The latest Italian Antibiotic Use Report has shown that Italy is among the countries with the highest consumption of antibiotics and spread of multi-resistant microorganisms in Europe [5]. This concerning scenario has prompted the approval of a National Antimicrobial-Resistance Contrast Plan (PNCAR) [6], which indicates objectives and actions to be implemented at national, regional, and local levels in order to promote an effective contrast to AMR in Italy. It is well known that the non-optimal and inappropriate antibiotic use is one of the main causes of AMR and although most antibiotics are used for treating infections outside hospitals, in the period 2016-2020, hospital consumption of antibiotics showed an increase in Italy [7]. This is very worrisome, since the PNCAR has set the goal of a 5% reduction of antibiotic consumption in the hospital setting. Instead, there is room for an increase in the incidence of healthcare-associated infections (HAI) caused by resistant microorganisms. Several investigations have been conducted to evaluate the prescriptions of antibiotics in the community and hospital settings [8][9][10] and to estimate the appropriateness of antibiotic prescription for infections caused by multidrug-resistant bacteria or for treatments of specific infectious diseases [11][12][13][14][15][16], while few experiences are available in Italy on antibiotic use for therapeutics reasons among hospital patients with or without infection on admission and the factors that can influence these antibiotic prescriptions [17,18]. The availability of data about the frequency and pattern of antimicrobial prescribing is crucial in order to achieve the success of action plans against AMR, to identify priority areas of intervention, and to implement training activities for healthcare workers to contrast the inappropriate antibiotic use and to provide useful information to improve antimicrobial prescriptions in the hospital setting. Therefore, this study has two main objectives. The first is to evaluate the antimicrobial prescribing practices in hospital settings in Italy, focusing on the appropriateness of antibiotic use, and the second is to investigate which determinants could predict hospital inappropriate antibiotic use. Results Overall, 723 patients were enrolled in the study, and they had received 936 antibiotic prescriptions for therapeutic reasons. The main demographic and anamnestic characteristics of the study population are described in Table 1. More than half (52.4%) were males, the average age was 63.5 years (range: 34-89), the large majority (90.5%) had at least one chronic medical condition, and approximately two thirds (61.4%) had a >1 Charlson comorbidity index score. Moreover, 26.8% had low serum albumin, 19.2% an immunosuppression status, and 26.6% a clinical infection on admission. Urgent admission to hospitals involved one in five of the study participants (21.7%), 40.8% were hospitalized in general medicine wards, 21.6% in surgical specialties wards, the average length of hospital stay was 8.9 days (range: 3-30), 21.4% had had at least one hospital admission in the previous 12 months, and 3.6% had had an HAI during the hospital stay. Respiratory, urinary tracts, and skin and soft tissue were the most common sites of infection, resulting in 40.3%, 21.5%, and 13.1% of all infections, respectively. A microbiological culture test was performed only in 94 patients (13%) and the more frequently isolated microorganisms were Escherichia coli (23.4%), Staphylococcus aureus (20.2%), and Klebsiella pneumoniae (11.7%). The results of stepwise logistic regression model performed to estimate predictors of inappropriate antibiotics prescribing are shown in Table 2. Being female (OR = 5.26; 95% CI 2.95-9.39), having a >1 Charlson comorbidity index score (OR = 2.36; 95% CI 1. 28-4.35), and having a longer hospital stay (OR = 1.24; 95% CI 1.16-1.33) were significant independent determinants of an inappropriate antibiotic prescription. Instead, patients who had had a non-urgent hospital admission (OR = 0.11; 95% CI 0.05-0.21), an infection on hospital admission (OR = 0.02; 95% CI 0.01-0.06), and a microbiological culture test during hospital stay (OR = 0.36; 95% CI 0.18-0.73) were significantly less likely to have an inappropriate antibiotic prescription. Table 3 shows the reasons of the inappropriate antibiotic prescriptions according to indication, drug choice, dose, duration, and route of administration. No indication for the antibiotic prescription accounted about a quarter of the reasons of the inappropriateness (26.6%), while among the 687 antibiotic prescriptions with indication, incorrect choice of antibiotics (34.3%) was the most common reason of the inappropriateness, followed by incorrect doses (18.6%) and excessive length of the antibiotic therapy (16.5%). Among the 236 inappropriate prescriptions due to an incorrect choice of antibiotic, the class of antibiotics more frequently used inappropriately were third generation cephalosporins (36.8%), fluoroquinolones (30.1%), and aminopenicillins (20.3%). Among the selected patients, 226 (31.2%) underwent a surgical procedure and 190 (87.6%) of them had received antibiotic prescriptions for prophylaxis purpose. All 178 patients eligible for the prophylaxis received antibiotics, whereas among the 48 procedures for which the surgical antibiotic prophylaxis (SAP) was not indicated, more than half (52.1%) received antibiotics. Overall, only 56 SAP (24.8%) were appropriate in accordance with indication, choice of the antibiotic, timing, and doses recommended by the guidelines. The antibiotics most frequently used inappropriately were ceftriaxone (17.9%) and ceftazidime (12.1%). Discussion This study provides relevant information on hospital antibiotic prescribing practices in Italy, where the levels of antibiotic utilization and spread of multi-resistant microorganisms are higher compared to other European countries. The results could be useful for the development of specific measures to contrast inappropriate antibiotic use and AMR in the hospital setting. This investigation has three key findings. A first key finding was that approximately one third of patients (34.2%) had received inappropriate antibiotic therapies since there was no indication. The comparison of this rate of inappropriateness with previous studies conducted in Italy and in other countries is difficult due to the different study criteria, healthcare setting, and characteristics of the study population. Despite these differences, similar results were found in two previous investigations where 33% of antibiotic prescriptions were not appropriate in a Swiss tertiary care hospital [19] and 32.7% of prescriptions were evaluated as inappropriate in emergency departments in Australia [20]. Lower rates of inappropriate prescriptions have been reported in a Dutch university hospital (29.3%) [21], whereas a higher frequency, ranging from 53.8% to 79.8%, were found in medical, surgical, and intensive care units in a previous investigation conducted in the same geographical area [22]. These data suggest that there is room for the improvement of adherence to antibiotic prescribing guidelines through the implementation of effective interventions. It is well known that antimicrobial stewardship programs (ASP) have a positive impact on the antibiotic use, and it could improve the hospitalization outcomes such as reduction of infectious diseases due to multidrug-resistant microorganisms, lengths of stay, readmission rate, and patients' disability and mortality [23][24][25][26]. Therefore, it is essential that healthcare services promote ASP in all healthcare facilities and settings, since several investigations demonstrated that the participation in ASP improves the appropriateness of antibiotic prescriptions [27][28][29]. A second key finding was that, when the single antibiotic prescriptions were analyzed, no indication for antibiotic use accounted for about a quarter of the reasons of the inappropriateness. Indeed, 26.6% of the antibiotics used inappropriately were prescribed to patients who did not have any evidence of infectious disease in the medical record. A possible reason is that physicians at the admission or during the hospitalization start antibiotic treatment even if they are unsure of the diagnosis and, therefore, they do not modify the treatment. This is a very unsafe practice because the unnecessary use of antibiotics may generate AMR, while this behavior should be limited given that Italy is one of the European countries where the consumption of antibiotics is higher, and inappropriate practices regarding antibiotic use has been observed both in hospital and in other healthcare settings [11,13,30,31]. The latest Italian Antibiotic Use Report has shown that inappropriate use of antibiotics was between 25% and 30% and the main reasons of inappropriateness were the treatment in the adult population of upper respiratory tract infections and uncomplicated lower urinary tract infections [7]. Moreover, other reasons for inappropriate antibiotic use in Italy have been found to be unnecessary antimicrobial prescription in hospital [30] and antibiotic use without the prescription of a physician [31][32][33]. A third key finding was that the results of multivariate logistic regression analysis showed that patients who had an infection on hospital admission and a microbiological culture test during the hospital stay were significantly less likely to receive an inappropriate antibiotic prescription. This result can be explained by the fact that physicians in our sample prescribe antibiotics when they suspect an infection, and they confirmed the infectious diseases and isolates with the microbiological testing. This finding underlines the usefulness of microbiological tests to implement a correct use of antibiotics, even if only 13% of patients with infections had had a test during the hospital stay. In the global action plan on antimicrobial resistance adopted by WHO [4], a key strategy was the optimization of the antimicrobial use and the increase in the investment on research on rapid diagnostic tools to contrast the AMR. Therefore, the microbiological tests are essential in the management of antibiotic therapies in healthcare facilities. In addition, the routine application of antibiograms is of help in identifying the levels of sensitivity and resistance to antibiotics of the microorganisms and for the choice of appropriate antibiotic therapy by physicians. In this study, the finding that only a small proportion of patients had received a microbiological test underscores the need to investigate the reasons for the poor use of diagnostic microbiological tools in order to provide useful information to implement more effective training for physicians to improve the appropriate use of testing and antimicrobials. Moreover, although evaluation of the SAP was not a priority objective of this study, the results showed that a large majority of SAP was inappropriate according to the Italian guidelines. Therefore, there is a need for health managers and healthcare professionals to be train in more careful antibiotic use according to the available guidelines in order to improve appropriate SAP. In conclusion, the results of the present study indicate that the inappropriate use of antibiotics continues to be a relevant issue in the hospital setting. Therefore, it is imperative to implement the adherence of physicians to the guidelines on antibiotic use and the participation in ASP in order to improve the appropriateness in antimicrobial prescription and to reduce the use of broad-spectrum antibiotics. Moreover, specific interventions are needed to encourage a wider utilization of diagnostic tools in order to practice targeted therapies and to counter the AMR. Limitations For a correct interpretation of the results, some limitations of the study should be considered. First, due to the retrospective study design, data were only retrieved by medical records. Indeed, it is possible that some information, such as medication prescriptions, reasons for the antibiotic use, and information on the patients' risk profiles, has not been accurately reported in the medical records by the healthcare professionals. Therefore, it cannot be excluded that the rate of appropriateness of antibiotic use may be underestimated due to the lack of information in the medical records and consultation with a physician that did not allow us to clarify the reason of several antibiotic prescriptions with no indication. Second, data were collected only in three hospitals in the Campania region. The characteristics of the hospitals may not be representative of all hospitals in Italy, and this may limit the generalizability of the study findings. Despite these limits, this investigation has several strengths, such as a large sample size and careful data collection for a one-year period in the selected hospitals. Therefore, we are confident that the results of this study are valid and provide important information on antimicrobial prescribing practices for inpatients. Setting, Study Design, and Sampling This study was carried out through a retrospective review of medical records of patients admitted in three public hospitals located in Campania Region (Italy) between 1 January and 31 December 2018. The minimum sample size consisted of 480 subjects, and it was calculated assuming a 30% prevalence of patients who have received inappropriate administration of antibiotics, a 95% confidence interval, and a 5% error. In order to increase the precision of the results, a larger sample of patients was recruited. Study Procedure, Recruitment, and Instrument for Data Collection Before starting the study, the Heads of the selected hospitals were contacted through a letter to obtain the approval to conduct the study and to describe the objectives of the investigation and the methods of collecting the information. After access to medical records was granted, the research team received the medical records in a secure electronic format from the medical records managers of the selected hospitals. Then, all information was reviewed and summarized on a standardized case report form by two authorized investigators not directly involved in patients care who consulted the medical records assuring the anonymity and confidentiality of the collected data. The medical records of patients fulfilling the following inclusion criteria were retrieved: (a) aged 18 years or above; (b) admitted for at least one day in medical or surgical wards; and (c) having received at least one antibiotic prescription for therapeutic reasons during their hospital stay. The following data were collected for each inpatient: age, gender, weight, height, date and diagnosis of hospital admission, ward and length of the hospital stay, hospitalization in the previous 12 months, smoking status, Charlson comorbidity index, type and number of chronic conditions, immunosuppression and nutrition status, previous allergies to antibiotics, clinical infection(s) on admission, microbiological culture test and antibiogram performed during the hospital stay, details about the antibiotic prescriptions (indication, type, timing, length, dose and route of administration of antibiotic therapy), and presence of HAI according to the criteria of the European Centers for Disease Control and Prevention [34]. Moreover, the following information were also collected for surgical patients: type of surgical procedure, surgical wound classification, ASA score, type of anesthesia, undergoing endoscopic surgery, implant of prosthesis, length of surgery, and details of SAP (type, timing, duration, dose, and route of administration). Appropriateness of antibiotic therapies was assessed according to the Guidelines for the implementation of Antimicrobial Stewardship programs and for the local implementation of antibiotic therapy protocols of the Campania region, and the national and international guidelines on the antibiotic treatments and preoperative prophylaxis [35][36][37][38]. The antibiotic prescriptions for therapeutic reasons and the SAP were evaluated as appropriate if the indication, choice of the antibiotic, the timing, the dose(s), and the length of administration were in accordance with the guidelines. Pilot Study and Ethical Statement The data collection instrument was pretested on a random sample of 25 medical records not included in the final sample, and the necessary changes were made before starting the study. The study protocol was approved by the Ethical Committee of the Teaching Hospital of the University of Campania "Luigi Vanvitelli" (approval number 215/2019). Statistical Analysis Statistical analyses were performed using Stata version 15 software [39] and were conducted in two stages. First, bivariate analysis was carried-out to evaluate the effect of the independent variables on the inappropriate antibiotic therapy in hospital patients using chi-square and Student's t-test for categorical and continuous variables, respectively. Subsequently, a multivariate stepwise logistic regression analysis was performed, including in the models the variables with a p-value ≤ 0.25 at the bivariate analysis according to Hosmer and Lemeshow's model building strategy [40], to investigate the independent characteristics associated with inappropriate antibiotic prescriptions for therapeutic reasons during the hospitalization (no = 0; yes = 1). In the stepwise logistic regression model the following independent variables were included: age (continuous), gender (male = 0; female = 1), type of hospital admission (urgent = 0; non-urgent = 1), ward of hospital stay (general medicine = 1; medical specialties = 2; general surgery = 3; surgical specialties = 4), length of hospital stay (continuous, in days), having a clinical infection on hospital admission (no = 0; yes = 1), immunosuppression status (chemotherapy and long term steroids use) (no = 0, yes = 1), low serum albumin (<3.5 g per deciliter) (no = 0, yes = 1), number of chronic medical conditions (0 = 0; 1 = 1; >1 = 2), Charlson comorbidity index score (0-1 = 0; >1 = 1), microbiological culture test (no = 0; yes = 1). The significance levels for the exclusion and inclusion of variables in the model were 0.4 and 0.2, respectively. All inferential tests were performed through the execution of a bilateral hypothesis test with the statistical significance level of p values equal to or less than 0.05. The results of multivariate regression analyses were reported as odds ratios (ORs) and 95% confidence intervals (CIs). Funding: This study was funded by the Department of Experimental Medicine of the University of Campania "Luigi Vanvitelli" (Funding of scientific research projects for the year 2019, protocol number 177358). The funder had no role in study design, in the collection, analysis and interpretation of data, in the writing of the report, and in the decision to submit the article for publication. Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of the Teaching Hospital of the University of Campania "Luigi Vanvitelli" (approval number 215/2019). Informed Consent Statement: The study protocol was submitted to the Heads of the selected hospitals to obtain the access to patients' clinical records, and complete anonymity and confidentiality of patients' data were guaranteed. Data Availability Statement: The data presented in this study are available on request from the corresponding author.	2023-01-24T16:23:15.707Z	2023-01-20T00:00:00.000	{ "year": 2023, "sha1": "b4cc89705aaa3d98d8010ff3b1934e2cab165fcf", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2079-6382/12/2/218/pdf?version=1674198482", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "b25bc035a740dc111436018385ac9568ec1a59a4", "s2fieldsofstudy": [ "Medicine", "Political Science" ], "extfieldsofstudy": [ "Medicine" ] }
43104074	pes2o/s2orc	v3-fos-license	The Two-Component Signal Transduction Protein Chk1p Regulates Quorum Sensing in Candida albicans ABSTRACT Regulation of hyphal morphogenesis in Candida albicans can occur through quorum sensing (QS). A QS signal, farnesol, is produced during high-density growth and inhibits morphogenesis. However, the signal transduction pathway that regulates QS is unknown. Here, we show that a C. albicans mutant lacking Chk1p but not either the Sln1p or the Nik1p histidine kinase is refractory to the inhibitory effect of farnesol both in cell suspension and during the formation of a biofilm. This study is the first to demonstrate a role for a two-component signal transduction protein in QS by a eukaryotic organism. Cell density is a critical factor in the regulation of Candida albicans hyphal morphogenesis. At a density of Ͼ10 6 cells/ml, yeast cells do not shift (germinate) to hyphae or do so at low frequencies, while at a density of Ͻ10 6 cells/ml, germination occurs (3). The relationship between cell density and new gene transcription (hyphal morphogenesis) resembles quorum sensing (QS) in some bacteria (14). Recent observations indicate that a QS system operates in C. albicans and that the isoprenoid farnesol is the QS autoinducer signal (12). Cells exposed to farnesol do not germinate, even at low cell densities. However, the regulatory and signal transduction events that direct QS are unknown, not only for C. albicans but for other fungi and eukaryotes in general. In some bacteria, two-component signaling regulates QS. Since C. albicans has several two-component signal proteins that are critical to a number of processes, including cell wall biosynthesis, adaptation to stress conditions, and virulence, our rationale was that farnesol sensing could be mediated through two-component proteins. C. albicans has three hybrid-histidine kinases, two of which have orthologues in Saccharomyces cerevisiae (Sln1p) and Neurospora crassa (Nik1p) that are presumed to play a role in an osmotic stress response (1,15,19,20). The third histidine kinase, Chk1p, has some similarity to two Schizosaccharomyces pombe proteins, Mak2p and Mak3p, which are known to function as sensors for oxidative stress (2,5). In addition to the histidine kinases, C. albicans has two response regulator proteins, Ssk1p and Skn7p, whose S. cerevisiae homologs act downstream of the Sln1p histidine kinase (11). In C. albicans, Ssk1p and Skn7p function in the adaptation of cells to oxidant stress, while Ssk1p, in addition, regulates the expression of structural cell wall proteins and negatively regulates Chk1p expression (7,18). The C. albicans strains used for this study have been described previously (4,6,9,20). Unless noted, cells were routinely cultured in YPD (1% yeast extract, 2% dextrose, 2% peptone) or YNB (0.67% yeast nitrogen base [pH 7.0], 50 mM glucose) at 30°C. To assess whether the two-component signal transduction proteins of C. albicans play a role in QS, all strains (see Table 1) were first cultured overnight at 30°C in YPD. Subsequently, the cells were washed twice and then inoculated into 10 ml of prewarmed medium 199 (pH 7.5) with or without 250 M trans,trans-farnesol (Sigma-Aldrich, St. Louis, Mo.) at a concentration of 5 ϫ 10 5 cells/ml. Cells were incubated at 37°C for 4 h, in order to allow germination to occur. Identical experiments were performed with 10% serum at 37°C as the inducing condition. The percentages of yeast cells and hyphae were then determined by light microscopy. Photographs were taken using a Canon digital camera, and figures were prepared with Adobe Photoshop 6.0. The C. albicans sln1, nik1, and chk1 histidine kinase mutants and the ssk1 response regulator mutant were compared to strain CAF2-1 (wild type) in hypha-inducing medium (10% serum or medium 199 [pH 7.5] with or without 250 M farnesol). In medium 199 (pH 7.5) lacking farnesol, germination proceeded normally (89 to 96%) for all strains (Fig. 1, left column; Table 1). In the presence of farnesol, the percentages of germination for CAF2-1 and for strains S (sln1/sln1), N (nik1/nik1), and SSK21 (ssk1/ssk1) decreased significantly to values ranging from 15 to 30%, while germination of the chk1 mutant (CHK21) was 84% of that of CAF2-1 ( Fig. 1, right column; Table 1). The germination of a strain reconstituted with a single copy of CHK1 (CHK23) was intermediate to that of CAF2-1 and the null counterpart (Fig. 1, right column, panel for CHK23; Table 1), indicating that the phenotype observed may be a result of the gene dosage. Similar results were seen when strains were grown in 10% serum (data not shown), indicating that the farnesol response is not medium dependent. An important aspect of the effect of farnesol on C. albicans is its influence on biofilm formation (8,16). C. albicans forms biofilms on a variety of substrates both in vitro and in clinical settings, such as indwelling intravenous catheters of patients (8). In the clinical setting, biofilm formation also represents a problem for therapeutic management of patients due to the resistance of the biofilm cells to antifungal therapy. In vitro studies indicate that farnesol inhibits biofilm formation, possibly by inhibiting the ability of the organism to shift to a filamentous morphology (16). Since our data indicate that the chk1 null mutant is not morphologically responsive to farnesol compared to parental and other mutants, the effect of farnesol on biofilm formation by this mutant was determined. C. albicans strains were grown overnight in YNB (pH 7) containing 50 mM glucose at 30°C, harvested, and washed twice in phosphate-buffered saline (PBS). The cell density was standardized to 10 7 CFU/ml, and cells (100 l of cell suspension) were allowed to adhere to the bottoms of 96-well microtiter plates. After 90 min of incubation at 37°C, the nonadhered cells were (16). After 2 h of incubation at 37°C, 100 l of each sample was transferred to a fresh plate, and the reduction in XTT was determined by measuring the optical density of the sample at 492 nm. In the absence of farnesol, biofilm formation occurred for all strains (Fig. 2). However, in the presence of subinhibitory concentrations of farnesol (25 M), CAF2-1 biofilm formation was reduced by 60%, while strain CHK21 (chk1/chk1) formed a biofilm that was approximately twofold greater than that of CAF2-1 (Fig. 2). At a 250 M concentration of farnesol, CAF2-1 again produced a biofilm that was approximately 40% of the size of that produced in the absence of farnesol by that strain. In contrast, the biofilm formation of the chk1 mutant exceeded that of CAF2-1. At a 25 M concentration of farnesol, strain CHK23, containing one copy of CHK1, formed a biofilm somewhat intermediate in size to those formed by strain CAF2-1 and the CHK21 null mutant (averaging 55% of the sizes of those formed by CAF2-1 and CHK21) (Fig. 2). With 250 M farnesol, biofilm formation by CHK23 was equal to that of CAF2-1. The ssk1, sln1, and nik1 null mutants resembled CAF2-1 and formed biofilms only in the absence of farnesol (data not shown). The data on biofilm formation by these strains support the observed effects of farnesol on germination ( Fig. 1; Table 1). Our observation of the role of the two-component signal transduction protein Chk1p in QS is the first for a eukaryotic organism. However, it is unclear how new gene transcription is initiated upon perception of the farnesol signal. Since Chk1p is a cytoplasmic protein, it is possible that another protein, which is upstream of Chk1p, perceives the farnesol signal and then activates a pathway that includes Chk1p. This hypothesis may be quite likely since no apparent motifs are found in Chk1p that would suggest a binding site for farnesol. The histidine box domain and the receiver domain of Chk1p contain residues of histidine and aspartate, respectively, that may be sequentially phosphorylated during signal transfer. In addition, a partial mitogen-activated protein kinase domain that may also be critical to signal transfer is located in the N-terminal half of Chk1p (5). The phenotype of the chk1 mutant in the presence of farnesol suggests that this kinase participates in a signal pathway. If so, this pathway would be unique among organisms that utilize two-component signal transduction. Our observation that QS is mediated through a two-component pathway may help in developing new strategies to inhibit hyphal morphogenic shifting in yeast cells. Furthermore, QS inhibitors in bacteria have been shown to be successful as antibiotics against biofilm formation (10). Since two-component signaling genes are not found in humans, it is likely that the development of two-component inhibitors may be useful in the treatment of candidiasis and candidal biofilms as well as in the dissection of the QS pathway. Such studies have recently been reported (13,17). This work was supported in part by Public Health Service grants NIAID-47047 and NIAID-43465 to R.A.C. and CA88456 and DE13478 to R.L.C. M.K. was supported in part by NIH training grant T32AI37251. B.P.K. was supported in part by a grant from The Netherlands Organization for Scientific Research (NWO). B.P.K. thanks Jesse Cohen for his technical assistance and Julia Douglas and her laboratory personnel for their valuable discussions. M.K. thanks Katie Kierpiec for assistance with the morphogenesis assays of strains grown in medium 199. Strains S and N (sln and nik mutants) were kindly provided by Mikio Arisawa, Nippon Roche, Kamakura, Japan. FIG. 2. Influence of farnesol on biofilm formation. CHK21 (chk1/ chk1⌬) is nonresponsive to farnesol. Biofilm formation was assessed using the XTT metabolic assay. The percentages of CHK21 and CHK23 cells that formed a biofilm are indicated relative to that of CAF2-1 without farnesol (black bars), with 25 M farnesol (white bars), and with 250 M farnesol (grey bars). The data represent averages of the results from four experiments.	2018-04-03T05:29:04.229Z	2004-08-01T00:00:00.000	{ "year": 2004, "sha1": "0c3d56a26b6cf6880019db0ff740a3190f8c8a7c", "oa_license": null, "oa_url": "https://ec.asm.org/content/3/4/1062.full.pdf", "oa_status": "GOLD", "pdf_src": "Highwire", "pdf_hash": "032da304aebf5d8f8b09e123391534a03bfc223c", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
269625110	pes2o/s2orc	v3-fos-license	Risk Factors for Intracerebral Hemorrhage: Genome-Wide Association Study and Mendelian Randomization Analyses BACKGROUND: The genetic and nongenetic causes of intracerebral hemorrhage (ICH) remain obscure. The present study aimed to uncover the genetic and modifiable risk factors for ICH. METHODS: We meta-analyzed genome-wide association study data from 3 European biobanks, involving 7605 ICH cases and 711 818 noncases, to identify the genomic loci linked to ICH. To uncover the potential causal associations of cardiometabolic and lifestyle factors with ICH, we performed Mendelian randomization analyses using genetic instruments identified in previous genome-wide association studies of the exposures and ICH data from the present genome-wide association study meta-analysis. We performed multivariable Mendelian randomization analyses to examine the independent associations of the identified risk factors with ICH and evaluate potential mediating pathways. RESULTS: We identified 1 ICH risk locus, located at the APOE genomic region. The lead variant in this locus was rs429358 (chr19:45411941), which was associated with an odds ratio of ICH of 1.17 (95% CI, 1.11–1.20; P=6.01×10−11) per C allele. Genetically predicted higher levels of body mass index, visceral adiposity, diastolic blood pressure, systolic blood pressure, and lifetime smoking index, as well as genetic liability to type 2 diabetes, were associated with higher odds of ICH after multiple testing corrections. Additionally, a genetic increase in waist-to-hip ratio and liability to smoking initiation were consistently associated with ICH, albeit at the nominal significance level (P<0.05). Multivariable Mendelian randomization analysis showed that the association between body mass index and ICH was attenuated on adjustment for type 2 diabetes and further that type 2 diabetes may be a mediator of the body mass index-ICH relationship. CONCLUSIONS: Our findings indicate that the APOE locus contributes to ICH genetic susceptibility in European populations. Excess adiposity, elevated blood pressure, type 2 diabetes, and smoking were identified as the chief modifiable cardiometabolic and lifestyle factors for ICH. study (GWAS) of 1545 ICH cases and 1481 controls identified 1 susceptibility locus for nonlobar ICH (chromosomal region 12q21.1) in European ancestry individuals. 7Most other genetic studies on ICH have used a candidate gene approach [2][3][4][5][6] and have reported a strong association with APOE genotypes. 5,6Among modifiable risk factors, hypertension and high systolic blood pressure have been strongly related to increased ICH risk in both conventional observational studies 1,8,9 and Mendelian randomization (MR) studies based on genetic variants associated with blood pressure as instrumental variables. 10,11Evidence from observational and MR studies further suggests that smoking 8,9,12 and alcohol consumption, 8,9,[13][14][15] 2 interrelated lifestyle behaviors, increase ICH risk.3,[18][19][20] Here, we performed a metaanalysis of genome-wide genotype data from 3 largescale studies and used the derived data set, comprising 7605 ICH cases and 711 818 noncases, to identify the modifiable cardiometabolic and lifestyle factors for ICH through MR analysis.We explored the independent associations for the identified risk factors and potential metabolic pathways from the risk factor to ICH through multivariable MR analysis.Although the focal point of this study was to uncover the major modifiable risk factors for ICH, the main results from the GWAS analysis are also reported. METHODS Data Availability Summary statistics data for the GWAS meta-analysis are available at the OSF data repository (https://osf.io/s69wd/). Study Design An overview of the study design is presented in Figure 1.We commenced this study by meta-analyzing GWAS data on ICH from the FinnGen study, the UK Biobank, and a previous GWAS by Woo et al 7 in 2014.We thereafter conducted univariable MR analyses to uncover the potential causal associations of conventional cardiometabolic risk factors (ie, obesity-related traits, type 2 diabetes and major glycemic traits, blood lipid traits, and blood pressure) and major lifestyle factors (ie, smoking, alcohol and coffee consumption, physical activity and inactivity, and sleep traits) with ICH risk, using our GWAS meta-analysis data for ICH and genetic variants associated with the exposures from previous GWASs.Finally, we performed multivariable MR analyses to explore independent causal associations and mediating pathways between the identified risk factors and ICH. FinnGen FinnGen is an ongoing project aimed at improving human health and promoting the discovery of therapeutic targets through genetic research. 27The project uses national health databases and broad genotypic data that have undergone thorough quality control measures, described in detail elsewhere. 27Briefly, related individuals, those with uncertain sex, non-Finnish genetic ancestry, high genotype missingness, or excess heterozygosity were removed.In the genetic variant quality control phase, variants with a missingness rate >2%, a Hardy-Weinberg equilibrium P<1×10 −6 , or a minor allele count <3 were excluded.Here, we utilized the ninth release of the FinnGen GWAS summary statistics data, which included 3749 ICH cases and 339 914 noncases.Cases were defined based on the International Classification of Diseases (ICD) codes: 431 for ICD, Eighth Revision and ICD, Ninth Revision and I61 for ICD, Tenth Revision, with data obtained from inpatient, outpatient, and social insurance reimbursement records.The association tests were adjusted for genotyping batch, age, sex, and top-10 principal components.The FinnGen study has received ethical approval, and written informed consent was obtained from all participants. UK Biobank The UK Biobank is a database and research initiative aiming to support scientific discoveries to improve human health. 28The cohort enrolled ≈500 000 UK adults between 2006 and 2010 and collected genetic data and information on a broad range of phenotypes.For the present GWAS, we excluded individuals lacking genetic data and those of non-British ancestry.At the sample quality control phase, we removed individuals with uncertain sex, related individuals (third-degree or closer), individuals with high missingness on genotype, and individuals with marked deviations in heterozygosity.At the variant quality control stage, we removed variants with a missingness rate >10%, variants with a Hardy-Weinberg equilibrium P<1×10 −6 , and variants with a minor allele frequency <0.01.ICH cases were identified from inpatient and outpatient records using ICD codes 431 for ICD, Eighth Revision and ICD, Ninth Revision and I61 for ICD, Tenth Revision.Our GWAS included 2311 ICH cases and 370 423 noncases.Associations were estimated using a logistic mixed model implemented in the REGENIE software, v2.0, 29 Woo et al GWAS The discovery phase of the Woo et al GWAS included individuals of European ancestry from 6 studies. 7Spontaneous ICH cases were defined as a new and acute neurological deficit with matching brain imaging displaying the existence of intraparenchymal bleeding.The exclusion criteria included trauma, hemorrhagic transformation of ischemic stroke, vascular malformation, tumors in the brain, and other causes of secondary ICH.The control group encompassed individuals free from ICH and were recruited from the same source population as the cases.Details on quality control and imputation procedures have been described elsewhere. 7 total of 1545 ICH cases and 1481 controls were included in the discovery phase, which was publicly available and included in the present GWAS meta-analysis.The association tests were adjusted for age, sex, and principal components.All studies were approved by the institutional review board or ethics committee at the participating site.Informed consent was obtained from participants themselves or from their legal proxies. GWAS Meta-Analysis We meta-analyzed data from the FinnGen, UK Biobank, and Woo et al GWAS under a fixed effect model and evaluated heterogeneity in the single-nucleotide polymorphism (SNP)-ICH association estimates across studies using METAL. 30Linkage disequilibrium (LD) score regression was applied to quantify the genomic inflation factor (λ GC ). 31 We used Functional Mapping and Annotation of Genome-Wide Association Studies 32 to identify independent risk loci, performing LD-based clumping using the 1000 Genomes Phase 3 data set as the reference by setting the thresholds (the GWAS significance level at P<5×10 −8 , the physical distance threshold at window >500 kb, and LD thresholds at r 2 =0.6 and r 2 2 =0.1).The lead SNP presenting the locus is defined as the variant with the lowest P value.The Multivariate Analysis of Genomic Annotation was used for gene mapping. 33For functional predictions, the tools incorporated the combined annotation-dependent depletion score, which is a metric quantifying the deleteriousness of single-nucleotide variants. 34Variants with elevated combined annotation-dependent depletion scores are generally considered more deleterious.We considered a locus to have a potentially deleterious functional impact when it had a combined annotation-dependent depletion score exceeding 10.To explore the pleiotropic effects associated with the identified locus/loci, we cross-referenced our findings using the GWAS Catalog database. 35We applied LD score regression to assess the genetic correlation between ICH and ischemic stroke and its subtypes, 36 subarachnoid hemorrhage, 37 and Alzheimer disease. 38 MR Analysis To identify the potential causal cardiometabolic and lifestyle risk factors for ICH, we performed 2-sample MR analyses using genetic variants associated with the exposures in previous GWASs as instrumental variables and ICH outcome data from the present GWAS meta-analysis.Data sources for the exposures are detailed in Table S1.This MR investigation followed the STROBE-MR guidelines (Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization). 39We constructed the genetic instruments by choosing SNPs associated with each exposure at the level of genome-wide significance (P<5×10 −8 ) in the relevant GWAS.For each exposure, we computed the LD matrix for selected SNPs using the 1000 Genomes European reference panel.For SNPs in high LD (r 2 >0.01), we kept only the SNP with the strongest association (lowest P value) with the exposure.We estimated the average F-statistic of used SNPs for each exposure to measure the strength of the genetic instruments.The SNPs used as instrumental variables for the exposures along with the summary statistics are provided in Table S2. We used the multiplicative random effects inverse variance weighted method to obtain the primary MR estimates.To evaluate the robustness of these estimates, we conducted sensitivity analyses using the weighted median, 40 MR-Egger, 41 and MR-PRESSO (Mendelian Randomization Pleiotropy Residual Sum and Outlier) 42 methods.We tested for heterogeneity among SNP estimates using the Cochran Q value and assessed horizontal pleiotropy using the MR-Egger intercept test and MR-PRESSO global test.To explore the independent and potential mediating effects of the exposures on ICH, we performed multivariable MR analyses.The mediation effect was calculated using the formula: (total effect−direct effect)/ total effect, and the SE of the mediation estimate was calculated using the propagation of error method.Given the strong associations with ICH for several SNPs associated with the lipid traits and located near the APOE locus, we conducted a sensitivity analysis for major lipid traits where we excluded SNPs ±1 Mb up/down from the APOE locus.We corrected for multiple tests using the Benjamini-Hochberg false discovery rate approach.All MR analyses were performed with the TwoSampleMR and MendelianRandomization packages in R, version 4.1.1. GWAS Results This GWAS meta-analysis, comprising a total of 7605 ICH cases and 711 818 noncases, resulted in 1 genome-wide significant locus (P<5×10 −8 ), located at the APOE locus (Figure 2).The genomic inflation factor (λ GC ) was <1.04 in each study and 1.06 in the meta-analysis, indicating minimal inflation due to population stratification or other confounding factors.The lead variant (chr19:45411941, rs429358) associated with an odds ratio of ICH of 1.17 (95% CI, 1.11-1.20;P=6.01×10 −11 ) per additional C allele in a meta-analysis of FinnGen and UK Biobank GWAS data (the SNP was unavailable in the GWAS by Woo et al and had no suitable proxy [LD r 2 >0.8]).The lead SNP had a combined annotation-dependent depletion of 12.6, suggesting its potential deleteriousness in ICH.A search in the GWAS Catalog database revealed pleiotropic associations of the APOE variant with lipid traits and Alzheimer disease, and a multitude of other phenotypes (Table S3).After correction for multiple testing, ICH showed a significant genetic correlation (r g ) with all ischemic stroke subtypes and subarachnoid hemorrhage but not with Alzheimer disease (Table S4). MR Results In univariable MR analysis, genetically predicted higher levels of body mass index, visceral adiposity, diastolic blood pressure, systolic blood pressure, and lifetime smoking index, as well as genetic liability to type 2 diabetes, were associated with increased odds of ICH after correction for multiple testing (Figure 3).There was weak to moderate heterogeneity between SNP association estimates for visceral adiposity, blood pressure traits, and type 2 diabetes, but these and the associations for body mass index and lifetime smoking index remained S5).Genetically predicted higher levels of waist-to-hip ratio, triglycerides, alcohol consumption, and smoking initiation liability were associated with higher odds of ICH at nominal significance (P<0.05; Figure 3); associations for waistto-hip ratio and smoking initiation remained consistent in sensitivity analyses (Table S5).Results for smoking initiation were consistent in a sensitivity analysis using effect size estimates (β coefficients) for the exposure from a GWAS analysis without the UK Biobank (Table S6).The association between genetically predicted alcohol consumption and ICH differed across studies, with a significant positive association in the UK Biobank, a suggestive positive association in the Woo et al GWAS, and no association in FinnGen (Table S7).We did not detect a significant association of any lipid fraction (except for the borderline association with triglycerides), fasting glucose, insulin, coffee consumption, physical activity, leisure-screen time, or any sleep traits with ICH (Figure 3).The lack of significant associations for the lipid traits persisted after excluding SNPs ±1 Mb up/ down from the APOE locus (Table S8). Multivariable MR analysis that included various combinations of the studied exposures suggested independent associations of genetically predicted diastolic blood pressure and smoking initiation with ICH (Table S9).The multivariable MR analysis that included body mass index and type 2 diabetes revealed an attenuated association for body mass index.The mediation analysis suggested that the association between genetically predicted body mass index and ICH is to a large extent (>50%) mediated by type 2 diabetes (Figure 4) although this analysis was underpowered to detect a significant mediation effect (Table S10).There was suggestive evidence that the association between genetic liability to smoking initiation may be mediated in part by diastolic blood pressure ORs are scaled per SD increase in genetically predicted body mass index, waist-to-hip ratio, blood lipids, and lifetime smoking index; per kilogram increase in genetically predicted visceral adiposity; per log-odds in the prevalence of type 2 diabetes and smoking initiation; per 1 mmol/L in genetically predicted fasting glucose; per 1 log-transformed picomole per liter in genetically predicted fasting insulin; per 10-mm Hg increase in genetically predicted blood pressure; per SD increase in the genetically predicted log-transformed alcoholic drinks/wk; per 50% increase in genetically predicted coffee consumption; per log-odds in moderate-to-vigorous physical activity (active vs inactive), short sleep (<7 vs 7 to 8 h/d), long sleep (≥9 vs 7 to 8 h/d), and insomnia; and per hour/day increase in sleep duration.FDR indicates false discovery rate; and OR, odds ratio. (but not systolic blood pressure) and type 2 diabetes (Figure 4) but with broad CIs including the null (Table S9).Multivariable MR analysis of smoking initiation and alcohol consumption found that smoking was the primary trait associated with ICH in the GWAS meta-analysis of all 3 studies, but that alcohol consumption was the major trait associated with ICH in UK Biobank (Table S9).The association between alcohol consumption and ICH in UK Biobank was attenuated and appeared to be mediated to some degree by diastolic and systolic blood pressure (Table S10).The lack of association for lipid traits remained consistent in the multivariable MR analysis with mutual adjustment for the 3 lipid traits (Table S9). DISCUSSION In this hitherto largest published GWAS of all ICH in individuals of European ancestry, comprising 7605 cases and 711 818 noncases, APOE was identified as the major ICH genetic susceptibility locus.Our MR analyses based on outcome data from this GWAS provided evidence that high blood pressure, smoking, excessive adiposity, and type 2 diabetes are risk factors for ICH. The APOE gene encodes apolipoprotein E, which has a central role in lipid metabolism, both in plasma and in the brain, and is a strong genetic risk factor for Alzheimer disease. 43Our GWAS and previous studies based on a candidate gene approach 5,6 indicate that the APOE locus also contributes to the genetic susceptibility to ICH.For example, a candidate genes study showed that the APOE ε2 and ε4 genotypes were associated with higher odds of lobar ICH at the genome-wide significance threshold, and the APOE ε2 genotype was associated with higher odds of nonlobar ICH at the level of nominal significance (P<0.001). 5A transethnic case-control study found that the APOE ε4 genotype was associated with higher odds of lobar ICH in white and Hispanic participants but not in black participants. 6The lead APOE variant identified in the present GWAS meta-analysis was not available in the GWAS by Woo et al, 7 which may explain the lack of genome-wide significant association at the APOE locus in that study.Our GWAS meta-analysis of all ICH did not detect genome-wide significant associations with other possible risk loci (eg, COL4A1, COL4A2, 12q21.1,1q22, GPX1, CR1, ITGAV, and PRKCH) for all ICH or its subtypes suggested by previous studies in European populations. 2,3,7,44As we only studied all ICH, our findings are diluted for subtype-specific loci.Given the differences in histopathology for lobar versus nonlobar (deep) ICH, further large-scale GWASs with data on ICH subtypes are needed to decipher the significance of non-APOE loci. With respect to modifiable risk factors, our MR findings confirm previous data linking high blood pressure 1,[8][9][10][11] and smoking 8,9,12 to increased risk of ICH.In our multivariable MR analysis including both diastolic and systolic blood pressure, only diastolic blood pressure remained associated with ICH.Our results provided suggestive evidence that the associations between smoking and ICH might be mediated to a minor extent by diastolic blood pressure (a nonsignificant ≈11% mediating effect) but not systolic blood pressure.Observational studies of the association between current smoking or consistent smokers and diastolic blood pressure levels have provided mixed results, with a positive, 45 inverse, 46,47 and no association 48 reported.Taken together, the effect of smoking on blood pressure is at most small, and the association between smoking and ICH is likely driven by other mechanisms than by diastolic blood pressure.For example, the smoking-ICH association might, in part, be mediated by type 2 diabetes, as suggested by our MR analysis. Self-reported high and heavy alcohol consumption, which commonly coexist with current smoking, has been associated with an increased risk of ICH in observational studies. 8,9,14,15In addition, genetically predicted alcohol consumption was associated with higher odds of ICH in previous MR analyses based on data from the China Kadoorie Biobank, 14 the UK Biobank, 13 and the Woo et al GWAS (nonsignificant association). 13The present MR analysis confirmed the association between genetically predicted higher alcohol consumption and ICH in the UK Biobank and the nonsignificant association in the Woo et al GWAS data set.The association between alcohol consumption and ICH in UK Biobank was independent of smoking but seemed to be partly mediated by blood pressure, as expected given the strong association between alcohol consumption on blood pressure. 13,49The lack of consistent association between genetically predicted alcohol consumption and ICH based on our GWAS metaanalysis data set was driven by the null association in the FinnGen study.This lack of association may suggest that the genetic instrument used for alcohol is not a good predictor of alcohol-related harm in FinnGen participants. Obesity and type 2 diabetes are interrelated metabolic traits associated with many chronic diseases, but previous results of these traits in relation to ICH are conflicting.The INTERSTROKE case-control study that involved 663 ICH cases and 3000 controls from 22 countries found that a high (versus low) waist-to-hip ratio was associated with an odds ratio of 1.41 (95% CI, 1.02-1.93). 8Consistently, an observational analysis of data from the UK Biobank, including 1391 incident ICH cases ascertained among 490 071 participants, showed that waist circumference adjusted for body mass index was positively associated with ICH risk. 50Additionally, a positive association of genetically predicted body mass index 25 or waist-to-hip ratio 22 with ICH was reported in previous MR studies.In contrast, a borderline inverse association between body mass index and ICH was observed in the Million Women Study of 712 433 women (n=2032 ICH cases). 9Nevertheless, this study found that diabetes that required treatment was associated with a 31% higher risk of ICH. 9 Our MR analyses found that all studied adiposity measures and type 2 diabetes liability were associated with higher odds of ICH, and the association for body mass index may be largely mediated by type 2 diabetes. Findings for blood lipids in relation to ICH are equivocal, and if anything, the direction of associations is opposite to those for atherosclerotic diseases, such as ischemic stroke. 8,51Results of the INTERSTROKE study (n=469 ICH cases; n=2127 controls) showed that high levels of HDL (high-density lipoprotein) cholesterol were associated with higher odds of ICH, whereas high levels of non-HDL cholesterol were associated with lower odds of ICH. 8 In a nested case-control study within the China Kadoorie Biobank (n=4776 ICH cases), no association was observed for HDL cholesterol, but a 1-mmol/L lowering in LDL (low-density lipoprotein) cholesterol was associated with a significant 16% higher odds of ICH in the observational analysis and a nonsignificant 13% higher odds of ICH in the MR analysis. 52Consistent with this, an MR study based on some studies included in the Woo et al GWAS (n=1286 ICH cases of European ancestry) reported a significant association between genetically predicted higher LDL cholesterol and lower odds of ICH. 20In contrast, other MR studies based on ICH data from the Woo.et al GWAS (n=1545 cases) did not detect any association for non-HDL cholesterol or triglycerides, 26,53 but one of these studies found that higher HDL cholesterol levels were associated with significantly higher odds of ICH in a multivariable MR analysis adjusted for non-HDL cholesterol and triglycerides. 53n MR study that included 1064 ICH cases ascertained in the UK Biobank (n=367 703 participants) found that genetically predicted higher levels of HDL cholesterol and triglycerides were significantly associated with lower odds of ICH. 19In our MR analysis of blood lipids and risk of ICH, contamination of ICH with secondary hemorrhage may have biased the results towards the null. Physical activity was not associated with ICH risk in previous observational studies, such as the INTER-STROKE study 8 and a British cohort study, 9 consistent with our MR finding.Studies on sleep traits in relation to the risk of ICH are scarce.We previously found that short sleep duration (<7 versus 7-9 hours/day) was associated with a 21% higher risk of incident ICH in a Swedish cohort. 21The present MR result for short sleep was null, but the precision of the estimate was low, with a 95% CI ranging from 0.71 to 1.31.We also found no association between genetically predicted long sleep or insomnia with ICH.As in previous observational 54,55 and MR studies, 56 we found no evidence of an association between coffee consumption and ICH.Hence, physical activity, sleep, and coffee consumption are not likely to be important risk factors for ICH. Strengths of this study include the large sample size and the comprehensive exploration of the causal role of potential cardiometabolic and lifestyle risk factors for ICH.A shortcoming is that we were unable to study ICH subtypes.We, therefore, may have overlooked ICH risk loci and modifiable risk factors related to only lobar or nonlobar ICH.Another limitation is the ICD-based definition of ICH and the lack of case adjudication in FinnGen and UK Biobank.This likely increased study heterogeneity through the inclusion of at least some degree of miscoded cases with secondary hemorrhage due to head trauma, brain tumor, subarachnoid hemorrhage, and hemorrhagic transformation of ischemic strokes.Another weakness is that we were unable to explore sex-specific associations due to a lack of sex-specific data on ICH.Furthermore, as we confined our study sample to individuals of European ancestry (to diminish population stratification bias), our results may not be transferable to non-European populations.Finally, we had no replication sample to validate the identified risk locus.However, the genomic region identified in the present GWAS has been reported in previous candidate genes studies, 3 and our study may, thus, be viewed as a replication of prior findings. Interpretations Our study corroborates the reported association between APOE and the risk of ICH in populations of European ancestry.Maintaining a healthy blood pressure throughout life and avoidance of excessive adiposity, type 2 diabetes, and smoking were prioritized as the chief modifiable risk factors for ICH. Figure 2 . Figure 2. Manhattan plot of genome-wide association meta-analysis of intracerebral hemorrhage.ICH indicates intracerebral hemorrhage; OR, odds ratio; and SNP, single-nucleotide polymorphism. Figure 3 . Figure 3. Univariable Mendelian randomization analysis of the associations between genetically predicted cardiometabolic and lifestyle factors and intracerebral hemorrhage risk.ORs are scaled per SD increase in genetically predicted body mass index, waist-to-hip ratio, blood lipids, and lifetime smoking index; per kilogram increase in genetically predicted visceral adiposity; per log-odds in the prevalence of type 2 diabetes and smoking initiation; per 1 mmol/L in genetically predicted fasting glucose; per 1 log-transformed picomole per liter in genetically predicted fasting insulin; per 10-mm Hg increase in genetically predicted blood pressure; per SD increase in the genetically predicted log-transformed alcoholic drinks/wk; per 50% increase in genetically predicted coffee consumption; per log-odds in moderate-to-vigorous physical activity (active vs inactive), short sleep (<7 vs 7 to 8 h/d), long sleep (≥9 vs 7 to 8 h/d), and insomnia; and per hour/day increase in sleep duration.FDR indicates false discovery rate; and OR, odds ratio. Figure 4 . Figure 4. Metabolic pathways mediating the associations of genetically predicted body mass index (BMI) and smoking initiation with intracerebral hemorrhage (ICH) risk.DBP indicates diastolic blood pressure; SBP, systolic blood pressure; and T2D, type 2 diabetes. and adjusted for genotyping batch, age, sex, and top-10 principal components.The study received approval from the North West Multi-Centre Research Ethics Committee.All participants provided written informed consent.	2024-05-09T06:16:34.895Z	2024-05-08T00:00:00.000	{ "year": 2024, "sha1": "fa7b3cefdadbb89780dceb85c854e7aaacefaaea", "oa_license": "CCBY", "oa_url": null, "oa_status": "CLOSED", "pdf_src": "PubMedCentral", "pdf_hash": "411d668360358a908fd3f980cc75be8e8de0fe38", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
270981531	pes2o/s2orc	v3-fos-license	Regular HbA1c examination and health outcomes in adult patients with type 2 diabetes in South Korea: A retrospective cohort study Concerns have been raised about the effectiveness of using process-centered indicators to assess the quality of diabetes care in Korea. This study aims to examine the factors influencing the performance of regular HbA1c testing and to explore its association with health outcomes, including hospitalization and mortality. We utilized a retrospective cohort design with a 4-year follow-up period, involving 159,452 adult patients newly diagnosed with type 2 diabetes (E11 in International Classification of Diseases, 10th Edition) in 2011. We established a national population database by merging the Korea National Health Insurance (KNHI) claims database and the KNHI Qualification Database of South Korea. The proportion of diabetic patients who underwent regular HbA1c testing at least once a year in the first 3 years was determined to be 33.8%. In comparison, patients who did not receive regular tests during the same period exhibited significantly increased odds of hospitalization (diabetes/CVD/renal, OR, 1.23, 95% CI, 1.12–1.34; diabetes, OR, 1.36, 95% CI, 1.17–1.57). Additionally, this nonpatient group experienced a higher risk of mortality (OR: 1.56, 95% CI: 1.36–1.80). This study supports the positive impact of regular HbA1c testing on health outcomes for individuals with type 2 diabetes. To increase the current 33% rate of regular HbA1c testing, developing patient-customized management policies is essential. Priority should be given to diabetic patients aged 65 or older, living in rural areas, and those belonging to low-income families (medical aid). Introduction Diabetes is one of the most common chronic diseases, which is experiencing a rapid global rise.Over the past decade, the number of diabetic patients has more than tripled, escalating from 151 million in 2000 to 463 million in 2019.It is projected to increase further up to 700 million by 2045. [1]Notably, 3 out of 4 individuals with diabetes mellitus falls within the workingage group, and each year, approximately 4 million individuals between the ages of 20 and 79 succumb to diabetes-related complications, resulting in a mounting social and economic burden. [1]n Korea also, the prevalence rate of type 2 diabetes among adults was 10.4% as of 2018, with approximately 190,000 new cases diagnosed annually.The total cost of type 2 diabetes treatment, including medication expenses, has been increasing by about 110 billion won ($84 million) each year. [2][5][6] The Organization for Economic Cooperation Development (OECD) also has been comparing the quality of diabetes care among member countries using quality indicators. [7]lycated hemoglobin A1c (HbA1c) plays a vital role in diabetes management, and thus, is widely used as a representative indicator for the quality of diabetes care in many countries. [8,9]orea has been conducting quality assessment initiatives for diabetes care based on HbA1c test indicators merely since 2011. Despite the regular HbA1c testing rate increasing year after year, the hospitalization rate (per 100,000 people) for diabetes patients has not decreased.This has raised concerns that focusing solely on the process indicators of diabetes management would lead to only increased spending without corresponding improvements in health outcomes. [2]his study examined the status of regular HbA1c tests, analyzing the factors influencing regular testing and the relationship between regular HbA1c testing and overall health outcomes for newly diagnosed type 2 diabetes patients in 2011 within the National Health Insurance program.Additionally, it aims to identify patient characteristics associated with poor adherence to regular diabetes management.Through this analysis, we intend to assess the effectiveness of diabetes care quality assessment initiatives in Korea and provide evidence to update the existing policies. Study design and data source This retrospective cohort study examined the relationship between regular HbA1c tests and health outcomes over a 4-year-follow-up period for patients newly diagnosed patients with type 2 diabetes (E11, International Classification of Diseases, 10th Edition (ICD-10)) in 2011. [10]were investigated for regular HbA1c examinations during the first 3 years of the 4-year follow-up, with hospitalization and mortality assessed in the final year. [3]Newly diagnosed diabetes was defined as the absence of any prior record of a diabetes diagnosis (ICD-10 code, E10-14) within the preceding 6 years (2005-2010).Based on the International Diabetes Federation's (IDF) Adult Diabetes Survey of OECD countries, the age of 20 years and older was used as the reference age for study subjects. [1]he study data utilized the Korea National Health Insurance (KNHI) Claims Database and the KNHI qualification database covering the period from 2005 to 2015.The data were linked using a separate patient number created by encrypting patients' personal identification numbers to ensure anonymity.Through the health insurance claims data, we were able to identify the date of diabetes diagnosis, whether HbA1c testing had been performed, the number of outpatient visits, the primary medical institution attended, the Charlson Comorbidity Index (CCI), and hospitalization information.In addition, using the KNHI qualification data, we obtained information on patients' sex, age, place of residence, income level, and date of death. [3]This study was approved by the Institutional Review Board of the Korea Institute for Health and Social Affairs (KIHASA) (Sejong, Republic of Korea) (IRB No. 2019-44). Study population In 2011, there were 361,617 newly diagnosed cases of type 2 diabetes in adults aged 20 or older.However, concerns about the accuracy of diagnoses based on health insurance claims data prompted the exclusion of certain patients from the analysis.Specifically, individuals who were diagnosed with type 2 diabetes but used outpatient services fewer than 4 times within the 3 years following their diagnosis (a total of 169,361 individuals) were excluded from the final analysis. [3,11]In addition, patients (a total of 15,975 individuals) who had been diagnosed with severe illnesses such as ischemic heart disease (IHD), cardiovascular disease (CVD), and cancer before their 2011 type 2 diabetes were also excluded.This exclusion was to ensure the clearer investigation into the impact of regular HbA1c testing on diabetes management for future hospitalization and mortality. [12]inally, after excluding patients (a total of 4613 individuals) who had died and patients (a total of 12,216 individuals) who had been hospitalized during the 3-year period in which regular HbA1c testing was evaluated, [3,12] a final study population of 159,452 patients was selected. Study variables 2.3.1.Independent variable: regular HbA1c examination.The study's independent variable is whether or not regular HbA1c testing was conducted.According to the Korean Diabetes Association's guidelines, HbA1c testing should be performed at least once per year. [2]For this study, patients who underwent HbA1c testing at least once a year for 3 consecutive years, in line with this standard, were chosen as regular testers. Dependent variable: health outcomes (hospitalization and mortality). The health outcomes analyzed in this study were hospitalization and mortality.Hospitalization was measured as a binary variable, indicating whether a patient was admitted to the hospital with a primary diagnosis code linked to diabetes (ICD-10, E10-E14), cardiovascular disease (CVD, ischemic heart disease (ICD-10, I20-I25), stroke (ICD-10, I60-I64)), or renal disease (ICD-10, N10-N12, N15-N19) during the last year of the 4-year follow-up period. [3,13]We defined mortality, including all causes of death, based on the information in the KNHI qualification database. [3,10]3.3.Covariates.Sex (male, female), age (20-44, 45-54, 55-64, 65+), Income level (Medical Aid, Decile 1 and 2, Decile 3 and 4, Decile 5 and 6, Decile 7 and 8, Decile 9 and 10), place of residence (Seoul, metropolitan, city, and county), number of ambulatory care visits for first 3 years (4-12, 13-24, 25-36, 37+), main attending medical institution (tertiary general hospital, general hospital, hospital, clinic, public health center), and Charlson Comorbidity Index (CCI, 0, 1, 2+) were controlled as covariates.[3] Patients insured by the National Health Insurance (NHI) have their income levels divided into ten equal parts (deciles), where the highest decile represents the highest income level, except for medical aid beneficiaries.The healthcare institution that patients visit the most frequently for ambulatory care for their diabetes during the first 3 years is considered their main attending medical institution.[3] The Charlson Comorbidity Index (Charlson index) is calculated based on all diagnostic information for 1 year prior to the initial diagnosis of type 2 diabetes in 2011.[3] Statistical analysis We conducted a chi-square test to examine the differences in test rates and to understand how patient characteristics are associated with regular HbA1c test rates.Additionally, we performed multiple logistic regression analyses to identify the factors influencing the performance of regular HbA1c testing.Finally, we conducted multiple logistic regression analyses, adjusting for several confounding factors, to determine the relationship between regular HbA1c testing and health outcomes such as hospitalization and mortality.These analyses were conducted separately for diabetes-related hospitalization (diabetes/CVD)/ renal disease) and diabetes-caused hospitalization.All statistical analyses were performed using SAS statistical software (version 9.3 for Windows; SAS Institute, Cary, NC). Differences in regular HbA1c examination according to patient characteristics Of all the patients involved in the study, the percentage of male patients (54.1%) was higher than that of female patients (45.9%).The largest age group was patients aged 45 and 54, accounting for 29.3% of participants (Table 1).Additionally, 55.4% of the patients resided in city and county areas, and those in the highest income bracket (9th-10th percentile) accounted for 26.8%.Most patients (25.5%) had 13 to 24 outpatient visits over 3 years, and 66.8% preferred clinics as their main attending medical institution for diabetes treatment.Furthermore, 45.8% of the patients had a CCI score of 0. During the 3-year period, 33.8% of patients received regular HbA1c tests at least once per year (Table 1).The rate of regular testing varied depending on patient characteristics, with males (35.5%) having a higher rate than females (31.8%).The highest testing rate was observed in the 20 to 44 age group at 40.9%, compared to other age groups.In terms of residence, patients living in Seoul had the highest testing rate at 38.6%, and those who primarily used tertiary general hospitals had a higher testing rate (43.1%).The regular testing rate increased with higher income levels and more frequent outpatient visits.Patients with a CCI score of 0 had a higher regular testing rate (38.0%).2).Patients residing in Seoul (OR: 1.52, 95% CI: 1.47-1.57)and metropolitan areas (OR: 1.03, 95% CI: 1.00-1.06)were more likely to receive regular HbA1c testing than those residing in other regions (city and county). Regular HbA1c examination and health outcomes We conducted a multiple logistic regression analysis to examine the relationship between regular HbA1c testing and health outcomes (Table 3).The analysis was adjusted for sex, age, residence, income, ambulatory care visits, main attending medical institution, and Charlson comorbidity index.The results revealed that patients who did not receive regular HbA1c testing over 3 years had higher odds of hospitalization after 3 years compared to those who did receive testing (diabetes/CVD/renal, OR, 1.23, 95% CI, 1.12-1.34;diabetes, OR, 1.36, 95% CI, 1.17-1.57).Furthermore, the odds of death were 1.56 times higher (95% CI: 1.36-1.80)for patients who did not receive regular testing. Discussion Well-structured medical services are essential for diabetic patients to maintain healthy lives.Implementing standardized clinical protocols strengthens diabetes management. [14]Many countries, including the United States, the United Kingdom, and Australia, have introduced active management programs to improve health outcomes for diabetes patients.These programs emphasize continuous patient management by placing primary care at the center and assessing the quality of the care provided. [3]egular testing of Hemoglobin A1c (HbA1c) is one of the most commonly used indicators for evaluating diabetes management.HbA1c testing is widely used in many countries as a key assessment of diabetes management.HbA1c indicates the average blood glucose control over the past 2 to 3 months, encompassing both fasting and postprandial blood glucose levels.It provides information about long-term blood glucose status and can reliably predict the risk of diabetes-related complications.Therefore, regular HbA1c measurement is strongly recommended for diabetic patients to evaluate their blood glucose control.The HbA1c test should be performed appropriately according to the clinical situation and treatment strategy.The American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) recommend that patients with stable blood glucose levels (HbA1c < 7%) should undergo testing at least once every 6 months.In contrast, patients with unstable or unmet blood glucose targets (HbA1c ≥ 7%) should be tested every 3 months. [8,9,15]ince 2011, Korea has evaluated the appropriateness of diabetes management in medical institutions using 7 process indicators.Among these, HbA1c testing is considered appropriate if performed at least once a year.The rate of appropriate HbA1c testing (at least once a year) in diabetes patients has consistently increased since the implementation of appropriateness evaluation, rising from 69.0% in 2011 to 83.1% in 2017. [2]However, while the appropriate testing rate is monitored yearly, the regular testing rate over the medium to long term is not tracked in relation to patients' health outcomes.It is essential to manage diabetes adequately from the early stages of diagnosis and maintain continuous management over the long term. According to this study, 33.8% of newly diagnosed diabetes patients received regular HbA1c testing at least once a year for 3 years.The regular testing rate varied according to patient characteristics; younger age, urban residence, higher income level, frequent outpatient visits, larger primary care facility, and lower disease severity were associated with higher rates of regular HbA1c testing.The regular HbA1c testing rate was notably higher among diabetic patients who resided in densely populated urban areas, faced no economic burden for regular outpatient visits and testing, and utilized medical institutions equipped with major testing equipment and relatively large-scale facilities.These findings suggest that social and economic conditions, as well as geographical accessibility, influenced the regular testing rate.18] HbA1c is a crucial test for managing diabetes.It helps detect prediabetes and take measures to prevent it from progressing into type 2 diabetes.It also aids in diagnosing type 2 diabetes and making appropriate treatment decisions. [19]The ADA and EASD recommend regular HbA1c testing to maintain HbA1c levels below 7% for effective diabetes management.Studies show that patients who adhere to the guidelines for HbA1c testing frequency are 2.43 times more likely to achieve the target HbA1c level than those who do not. [9]The HbA1c value is a surrogate marker for glycemic control and a key risk indicator for diabetesrelated microvascular and macrovascular complications and mortality.Therefore, regular HbA1c testing is crucial to reduce hospitalizations and mortality risk in patients with diabetes. [20]s HbA1c testing is being used as a significant evaluation indicator to assess the adequacy of diabetes management in Korea, the number of diabetes patients undergoing regular HbA1c testing has steadily increased yearly.However, the hospitalization rate (per 100,000 people) for diabetes patients remains very high compared to the OECD average (129.3people in the OECD average and 245.2 people in Korea in 2017). [2]his discrepancy has raised concerns that Korea's approach of focusing on process indicators of diabetes management may lead to increased medical expenses due to more frequent testing, without necessarily improving health outcomes.This study confirms that regular HbA1c testing can significantly reduce hospitalization and mortality rates among patients with type 2 diabetes.Patients who did not receive regular HbA1c tests for 3 years were found to have a 23% higher risk of being hospitalized for diabetes-related conditions such as diabetes, CVD, and renal disease compared to those who received regular tests.Moreover, they had a 36% higher risk of hospitalization for diabetes alone.The study also revealed that patients who did not receive regular testing had a 56% higher mortality risk.These findings emphasize the importance of regular HbA1c testing in improving the future health outcomes of patients with type 2 diabetes. This study had a few limitations.Firstly, it did not include all diabetes patients.The analysis focused solely on newly diagnosed type 2 diabetes patients.To examine the impact of appropriate management on health outcomes, it was necessary to align the disease characteristics of the study subjects.Differences in disease characteristics between the management and non-groups could obscure the effects of appropriate management intervention.Since type 1 diabetes has different mechanisms and causes of onset compared to type 2 diabetes, and since disease severity can vary significantly with the duration of illness in existing diabetic patients, only newly type 2 diabetic patients were selected to match the disease characteristics of the study subjects as closely as possible. Second, due to the limitations of the information available in the claims data, the frequency of HbA1c tests was determined, but changes in HbA1c levels could not be confirmed.This made it impossible to ascertain whether HbA1c was well controlled in patients with type 2 diabetes who received regular HbA1c testing.While Korea recommends receiving HbA1c testing at least once a year, the ADA and EASD recommend testing every 6 months (twice a year).Additional research is needed to determine whether the current testing standards in Korea are appropriate and to identify the most suitable testing standards by examining whether differences in HbA1c levels are based on the interval or frequency of HbA1c testing. Despite these limitations, this study is significant as it is the first to examine the association between regular HbA1c testing and health outcomes over a 4-year follow-up period among all newly diagnosed type 2 diabetes patients in Korea.Particularly for type 2 diabetes, which requires long-term management, regular testing over the mid-to-long term is necessary rather than short-term testing.This study found that regular testing over 3 years effectively improved health outcomes, highlighting the importance of consistently evaluating process measures, such as HbA1c testing, to enhance diabetes management outcomes in various countries. In conclusion, this study has confirmed that regular HbA1c testing can significantly improve the health outcomes of patients with diabetes.While many countries are working towards developing diabetes treatment guidelines and management programs, it has been reported that many diabetes patients fail to follow these guidelines properly. [9]In Korea, 83.1% of diabetes patients received HbA1c testing at least once in 2017.However, only 33.8% of patients received regular testing according to the standard over 3 years.It is crucial to narrow this gap in diabetes patient management in the future.Management policies need to be developed, especially for elderly patients (aged 65 and above), patients in rural areas, and low-income patients (Medical aid) who still have low rates of basic testing. Table 1 Differences in regular HbA1c examination according to patient characteristics. Table 2 Odds ratios of regular HbA1c examination by patient characteristics. Table 3 Analysis of the relationship between regular HbA1c examination and health outcomes.	2024-07-07T05:10:18.982Z	2024-07-05T00:00:00.000	{ "year": 2024, "sha1": "76f8a7c29cca1f3b358dc4b28cb98424a6ccb679", "oa_license": "CCBYNC", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "76f8a7c29cca1f3b358dc4b28cb98424a6ccb679", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
244883313	pes2o/s2orc	v3-fos-license	Macaque-human differences in SARS-CoV-2 Spike antibody response elicited by vaccination or infection Macaques are a commonly used model for studying immunity to human viruses, including for studies of SARS-CoV-2 infection and vaccination. However, it is unknown whether macaque antibody responses recapitulate, and thus appropriately model, the response in humans. To answer this question, we employed a phage-based deep mutational scanning approach (Phage-DMS) to compare which linear epitopes are targeted on the SARS-CoV-2 Spike protein in humans and macaques following either vaccination or infection. We also used Phage-DMS to determine antibody escape pathways within each epitope, enabling a granular comparison of antibody binding specificities at the locus level. Overall, we identified some common epitope targets in both macaques and humans, including in the fusion peptide (FP) and stem helix-heptad repeat 2 (SH-H) regions. Differences between groups included a response to epitopes in the N-terminal domain (NTD) and C-terminal domain (CTD) in vaccinated humans but not vaccinated macaques, as well as recognition of a CTD epitope and epitopes flanking the FP in convalescent macaques but not convalescent humans. There was also considerable variability in the escape pathways among individuals within each group. Sera from convalescent macaques showed the least variability in escape overall and converged on a common response with vaccinated humans in the SH-H epitope region, suggesting highly similar antibodies were elicited. Collectively, these findings suggest that the antibody response to SARS-CoV-2 in macaques shares many features with humans, but with substantial differences in the recognition of certain epitopes and considerable individual variability in antibody escape profiles, suggesting a diverse repertoire of antibodies that can respond to major epitopes in both humans and macaques. 2 23 11 Center for Innate Immunity and Immune Disease, University of Washington, Seattle, WA, USA 24 25 * Corresponding author 26 Email: joverbau@fredhutch.org 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 130 a Within each group of macaques, subgroups received slightly different treatments (described in Table S1). 131 Enrichment of wildtype peptides 132 To compare which regions of Spike protein are recognized by human and macaque antibodies, 133 we examined the enrichment of wildtype peptides by antibodies from each individual ( Fig 1A). (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 9 142 visualized on the structure of a Spike protein monomer in Fig 1B. In addition to these defined 143 regions, we noted that one convalescent rhesus macaque appeared to weakly recognize an 144 epitope at the beginning of the S2 subunit (amino acid 686-710, Fig 1A). 145 In general, we did not detect responses in the RBD because many epitopes in this region are 146 known to be conformational, and Phage-DMS only has the power to detect epitopes that 147 include linear sequences. Epitopes in the RBD have been extensively detailed elsewhere [62, 148 63]. However, we did detect strong binding to an RBD epitope in some vaccinated pigtail 149 macaques (Fig 1A). This same region was enriched in samples from before vaccination in four of 158 convalescent humans (Fig 2). 162 Meanwhile, convalescent macaques recognized the following epitope regions more than 163 convalescent humans: CTD-N' (p ≤ 0.01), pre-FP (p ≤ 0.001), and post-FP (p ≤ 0.01) (Fig 2B). All 164 groups consistently recognized the SH-H epitope region (Fig 2). While vaccination appeared to 165 induce a stronger response against HR2 than infection (Fig 1A), there were no significant 166 differences in response driven by species (Fig 2). Within each group of macaques (vaccinated 167 and convalescent), subgroups received slightly different treatments (Table S1) 170 Taken together, these findings indicate: 1) vaccinated humans were the only group to 171 consistently recognize peptides from both the NTD and CTD-C' epitope regions, which are in 172 close physical proximity to one another ( Fig 1B); 2) convalescent humans had a limited 173 response to the CTD-N'; 3) compared to other groups, convalescent macaques had a notably 174 more robust response to regions upstream and downstream of the main FP epitope region; 4) 175 vaccinated macaques did not recognize the FP as strongly as other groups; and 5) vaccination 176 seemed to induce a stronger response against HR2 than infection in both macaques and 177 humans. 178 Defining and comparing escape pathways 179 To assess differences in the binding characteristics of human and macaque antibodies on a 180 more granular level, we next examined the mutations in Spike that reduced antibody binding in 181 each epitope region of interest. Because the antibody escape pathways for vaccinated humans 182 have been described previously [60], we did not examine the NTD and CTD-C', which are 183 exclusively recognized by this group. Instead, we focused on comparing escape profiles 184 between groups in the following epitope regions: CTD-N', FP, and SH-H. We first represent the 185 data as scaled differential selection values in logo plot form, as commonly shown in previous 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 12 207 To summarize the trends observed in the individual findings, we calculated summed differential 208 selection values for each individual at each site and generated boxplots by group (Fig 3A). In 209 addition to the aforementioned regions of escape common to all groups, convalescent 210 macaques also showed considerable escape between AA 529-535, with vaccinated macaques 211 also showing a less consistent response in this area (Figs. 3A and S5). The complexity and 212 variability of the escape pathways also prompted us to quantify the similarity in escape 213 between and within groups. Escape similarity scores largely corresponded to areas of high 214 magnitude of escape. Sites with low-magnitude summed differential selection values indicate 215 loci where mutations have no notable impact, and therefore those escape profiles reflect 216 fluctuations in peptide enrichments due to noise, which drives a lower escape similarity score 217 at those sites ( Fig 3A, lower panel). At some sites (e.g., 560, as described above), low scores 218 were also the result of some samples showing negative differential selection and others 219 showing positive differential selection, a comparison that was assigned the highest cost in our 220 escape similarity score algorithm. 221 To test the similarity of escape profiles across the CTD-N' epitope region, escape similarity 222 scores were aggregated across the region and computed both within and between groups. 223 These are shown as boxplots, with each point representing a pairwise comparison between 224 individual samples ( Fig 3B). For example, every vaccinated macaque was compared to every 225 other vaccinated macaque (a within-group comparison) and to every vaccinated human (a 226 between-group comparison). We included a comparison of convalescent macaques and 227 vaccinated humans, given visual similarities between their patterns of escape ( Fig 3A). 228 Convalescent macaques showed the highest within-group similarity in escape profiles, meaning 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 13 229 their escape profiles were more consistent than those of the vaccinated macaques or 230 vaccinated humans ( Fig 3B). Between-group escape similarity scores were on par with the 231 within-group scores for the vaccinated macaques and humans, indicating that although there 232 was substantial variability in individual profiles, this was not driven by sample groups. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 14 251 escape similarity scores for vaccinated humans and convalescent macaques were also higher in 252 the SH-H than in the CTD-N' or FP, confirming a more concordant response. The median 253 between-group escape similarity score for vaccinated humans and convalescent macaques was 254 on par with their median within-group scores, indicating that the escape profile of a vaccinated 255 human looks as similar to that of a convalescent macaque as it does to another vaccinated 256 human (Fig 5B). The similarity between these two groups was higher than the similarity 257 between convalescent macaques and humans, as well as between vaccinated macaques and 258 humans (Fig 5B). Despite this overall trend, two vaccinated humans had more unique escape 285 Discussion 286 In this study, we aimed to assess whether the antibody binding specificities to SARS-CoV-2 287 Spike in macaques are a useful model for the human response. Our results indicate important 288 similarities between macaques and humans; for example, both have antibodies that recognize 289 major epitopes in the CTD, FP, and SH-H. However, many differences are also apparent, with 290 some groups showing responses to unique epitopes, such as two physically proximal epitopes in 291 the NTD and CTD that are recognized by antibodies from vaccinated humans but not macaques. 292 Additionally, epitope regions flanking the FP were recognized by antibodies from convalescent 293 macaques, while antibodies from convalescent humans did not recognize the flanking regions 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 16 294 but showed a strong response within the FP itself. We found considerable diversity in the 295 pathways of escape between individuals, and this was not specific to either macaques or 296 humans, suggesting a diverse repertoire of antibodies that can respond to the major epitopes in 297 both groups. Overall, these results suggest that macaques and humans share recognition of 298 certain major epitopes. The differences that exist could be due to species (macaque vs. human), 299 but could also be influenced by differences in the specific type and number of exposures to 300 antigen in each group. 353 While our focus was on understanding how macaques and humans respond to a similar 354 exposure (i.e., vaccination or infection), we also noted similarities in response between re-355 infected macaques and vaccinated humans. These groups both exhibited the broadest 356 recognition across Spike, although the epitope regions they targeted were somewhat different. 357 As described above, these groups also had highly similar antibody escape profiles in the SH-H. 358 The vaccinated humans and re-infected macaques both received two exposures to high doses 359 of antigen. It is plausible that re-exposure directed initially diverse antibodies to converge on a 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 365 Additionally, epitopes that may normally be glycosylated are exposed for antibody binding in 366 Phage-DMS. There also are known germline-encoded differences in the properties of 367 immunoglobulin subclasses and Fc receptors between macaques and humans, leading to 368 differences in antibody function that cannot be assayed using Phage-DMS [90]. Additionally, 369 our sample set includes variables that limit our ability to draw conclusions about species-370 specific (macaque vs. human) differences in antibody response. The vaccinated macaques and 371 humans both received RNA vaccines encoding full-length Spike protein, but there were 372 differences in vaccine technology, including: 1) the use of mRNA in the human vaccine vs. 373 repRNA in the macaque vaccine, 2) the stabilization of Spike in its pre-fusion state in the human 374 vaccine, 3) the dosage and number of doses delivered, and 4) the formulation used to deliver 375 the RNA. Despite these differences, we found commonalities in some of the epitopes targeted 376 by antibodies from both groups. Additionally, the convalescent rhesus macaques were 377 experimentally infected twice with high titers of virus, compared to the convalescent humans 378 who were naturally infected once. This important discrepancy could be the reason why the 379 response in re-infected macaques aligned more closely with vaccinated humans than 380 convalescent humans. Studies of re-infected humans would help address this possibility. 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint 20 381 Our findings suggest that while vaccinated and convalescent macaques and humans share 382 recognition of some major epitopes, each group has a unique antibody binding profile. 383 Antibody escape profiles suggest a diversity of individual responses to most epitopes. 384 Important avenues for future study include comparing macaque and human responses to the 385 RBD and evaluating species differences in antibody function. Continued investigation of 386 immunogenic epitopes in conserved regions of Spike is also warranted to inform the 387 development of immunity that is more robust in the face of viral escape. 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 454 The effect of a mutation on antibody-peptide binding was quantified as "differential selection," 455 which is the log fold change in the enrichment of a mutation-containing peptide compared to 456 the wildtype peptide. This number is multiplied by the average of the wildtype peptide 457 enrichments at that site and its two adjacent sites to get a "scaled differential selection" value, 458 as described previously [60]. The enrichment values of the adjacent wildtype peptides are 459 included in this calculation to make the analysis less susceptible to noise. Negative differential 460 selection values represent reduced binding compared to wildtype, while positive differential 461 selection values indicate that the mutation enhanced binding. "Summed differential selection" 462 is the sum of the 19 scaled differential selection values for all mutations at a site, and gives a 463 sense of the overall magnitude of escape at that site. 24 464 The comparison of two escape profiles is quantified by an escape similarity score computed in 465 the framework of an optimal transport problem [93]; this algorithm was described in detail at 466 https://matsengrp.github.io/phippery/esc-prof.html. An overview of the method is shown in S4 105 and is also made available for use under a CC0 license. (which was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC The copyright holder for this preprint this version posted December 3, 2021. ; https://doi.org/10.1101/2021.12.01.470697 doi: bioRxiv preprint	2021-12-05T16:16:39.032Z	2021-12-03T00:00:00.000	{ "year": 2021, "sha1": "21b0a5b0a2e8b618faa7bfa039e336787220bf40", "oa_license": "CC0", "oa_url": "https://www.biorxiv.org/content/biorxiv/early/2021/12/03/2021.12.01.470697.full.pdf", "oa_status": "GREEN", "pdf_src": "PubMedCentral", "pdf_hash": "2cc1215ee2e4c016d5cc13c3e1fa91d091107fdf", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
197879925	pes2o/s2orc	v3-fos-license	Corporate Social Responsibility and Financial Reporting Quality: Evidence from Korean Retail Industry* Purpose – We investigate whether a firm’s engagement in socially responsible activity affects the quality of financial reporting within the retail industry of Korean market. Recent studies argue that more socially responsible firms tend to show a better quality of financial reporting. Research design, data, and methodology – We use a variety of proxy variables related to the use of discretionary accruals and real activity manipulation to measure the quality of financial reporting. The total of environmental, social and governance score is used to represent the degree of socially responsible activity in the retail industry. We use regression models to examine whether more socially responsible firms show a higher quality of financial reporting. The sample of publicly traded Korea retail firms is analyzed from 2011 to 2016. Results – Our analysis finds supporting evidence for limited earning management via the use of discretionary accruals. We find, however, no significant relationship between the degree of social responsibility and the quality of financial reporting within chaebol affiliates unlike non-chaebol affiliates. Conclusions – Our results weakly support a better quality of financial reporting for more socially responsible firms. The results highlight the importance of firm characteristics in deciding the effect of socially responsible activity on corporate policies. Introduction Socially responsible activity of a corporation (CSR) has become one of major managerial objects recently. Thus, it has received a wide range of attentions how a firm's CSR practice is related with various corporate policies. Retail firms care more significantly about their CSR performances because these firms face the needs of individual customers which is graded by the Korean Corporate Governance Service. A set of cross-sectional regression models are adopted to test the above hypothesis. Recent studies also emphasize that the effect of CSR activity on corporate policies is not equally applicable for firms with different characteristics. Specifically, it is widely known that the CSR performances within the group of family owned large conglomerates, chaebol affiliates have different implications on corporate policies such as Yoon, Lee, and Byun (2018). Thus, we also conduct sub-sample analysis based on the categorization of chaebol and non-chaebol affiliates in the Korean retail industry. The main findings of our examination could be described as follows. Most of all, we find evidence for supporting a higher quality of financial reporting for more socially responsible firms in the Korean retail industry. Specifically, retail firms with more active CSR engagements are less likely to manage earnings via the use of discretionary accruals. Yet, our estimation results show no significant relationship between the ESG score and the proxy variables related to real activity management. This result seems more closely related to the characteristics of retail industries. In contrast to manufacturing firms, retail firms may have rather simple production procedures, which makes the use of real activity manipulation including production cost or cash flow generation more difficult. Hence, retail firms have tendency to rely on discretionary accruals in their earning management tool. More socially responsible firms within the retail industry turn out to be less likely to use discretionary accruals to manage earnings but these firms may not have incentives to reduce real activity manipulation, which is already at a low level. Our sub-sample analysis provides a set of novel results as well. More socially responsible firms within chaebol affiliates do not reduce abnormal discretionary accruals as well as the level of real activity manipulations. This finding contrasts the estimation results of firms not belonging to chaebol affiliates, which indicates a higher quality of financial reporting at least in terms of using discretionary accruals. Economic environments of chaebol affiliates appear to be closely associated with the above findings. Chaebol affiliates are subject to continuous monitoring from the Korean Supervisory Service. Chaebol affiliates have limitations in their debt issuance, cross-guarantee, and intra trading. The Korean Supervisory Service also requires more strict auditing for chaebol affiliates as well. In fact, Park, Shin, Kang, and Kwon (2004) indicated that chaebol affiliates have better financial transparency compared to non-chaebol affiliates. Thus socially responsible firms within the chaebol affiliates may not have strong incentives to additionally enhance the quality of financial reporting because the overall standard of financial reporting is already high within chaebol affiliates. This work contributes to the previous literature in a large number of aspects. Most of all, we find generally supporting evidence for the hypothesis predicting a higher quality of financial reporting quality within the firms actively engaging in CSR practices. This hypothesis is partly supported by the case of using discretionary accruals in the Korean retail industry. The evidence from real activity, at least, does not argue against the hypothesis. Furthermore, our work emphasizes the importance of industry characteristics or firm characteristics in shaping the effect of CSR performances on corporate policies. The characteristic of retail industry in the use of real activity manipulation appears to be closely associated with the weak evidence for a higher quality of financial reporting within socially responsible firms. Furthermore, severe supervision from the Korea Supervisory Service appear to drive statistically insignificant relationship between CSR practice and the quality of financial reporting. Such a heterogenous effect of CSR practice is consistent with recent studies such as Yoon et al. (2018) and Yoon and Lee (2019). The organization of this paper is as follows. Section 2 reviews extant literature. Section 3 summarizes the sample and our empirical strategy. Section 4 provides the main empirical estimation results. Section I5 concludes. Related Literature Extant studies concerned about CSR performance have mainly focused on relation between CSR and corporate performance. Many studies have recently analyzed the effect of firm's CSR practices on the quality of financial earnings management. Specifically, Aguilera, Rupp, Williams, and Ganapathi (2007) argue that managers with strong CSR commitments firm do not have incentives to manipulate earnings. Kim et al. (2012) provide evidence that CSR practices are negatively associated with firm's earnings management. Their study finds an enhanced transparency of financial reporting within more socially responsible firms in the U.S. financial market. Similarly, Kim and Venkatachalam (2011) find that the financial reporting quality of firms within 'sin'-industry that engaged in tobacco, gambling, and alcohol industries is superior compared to various control groups. However, some studies also argue a poor quality of financial reporting for more socially responsible firms. This prediction relies on the argument that CEOs implement CSR practices for their self-serving purpose to advance their careers and reputation (McWilliams & Siegel, 2000). Specifically, a CEO engages in CSR practice more actively to cover up their unethical practices including earnings management. (Hemingway & Maclagan, 2004). In Korean market, the relationship between CSR performances and earnings management has been recently discussed. Choi and Moon (2013) argue that firms under the assesment of Korea Economic Justice Institute (KEJI) are less likely to engage in earning management and hence have a better quality of financial reporting compared to the other group of firms. Chun and Cho (2017) find that a firm's implementation differentiation strategy is negatively associated with real activities manipulation and great CSR performances strengthen the negative relation between differentiation strategy and real activity manipulation. This study is closely associated with a branch of literature emphasizing the firm/industry level heterogeneity in deciding the relationship between CSR and corporate policies. Moreover, Lin, Chang, and Dang (2015) argue that each individual firm confronts heterogenous CSR requirements in accordance with the characteristics of firms. Miralles-Quirós, Miralles-Quirós, and Valente Gonçalves (2018) empirically show that the valuation effect of CSR performance is significantly more substantial for environmentally sensitive industries. In Korea, such an emphasis on firm characteristics is tightly associated with the literature related to the corporate policies of chaebol affiliates. For instance, Yoon et al. (2018) argue that the valuation effect of CSR performance differ across the categorization of chaebol and non-chaebol affiliates. This categorization is in line with the result of Yoon and Lee (2019) as well. Park et al. (2004) show that the transparency of financial reporting is higher for chaebol affiliates probably because of their continuous monitoring from the Korea Supervisory Service. Hypothesis Development As highlighted in Kim et al. (2012), if managers are ethical and engaged in CSR practices based on moral imperative, these managers are more likely to constrain earnings management, which results a more transparent financial reporting. Accordingly, more socially responsible firms are less apt to participate in earning management. H1: More socially responsible retail firms show a higher quality of financial reporting. However, as emphasized in Park et al. (2004), the sample of chaebol affiliates already has a higher quality of financial reporting due to the strict supervision from the Korean Supervisory Service. In particular, firms belonging to chaebol affiliates have restrictions in their debt issuance, cross-guarantee, and intra trading; they are also required a higher standard of financial reporting. Thus socially responsible firms within the chaebol affiliates may not have strong incentives to improve the quality of financial reporting because the overall standard of financial reporting is already higher within chaebol affiliates. H2: More socially responsible retail firms within chaebol affiliates may not show a higher quality of financial reporting in the retail industry. We test these two hypotheses in the sample of Korean retail industry. Data and Empirical Models We use two different sources of datasets to test a higher quality of financial information for socially responsible firms. The first source of data is associated with a firm's financial information to measure the quality of financial reporting from the perspective of earning managements. By following Kim et al. (2012), the annual cross-sectional industry regression model is to estimate a firm i's discretionary accruals, which calculates the abnormal discretionary accruals as the residual of the following equation: where DA represents discretionary accruals; REV represents revenue; REC represents receivables; PPE represents fixed capital; and IBIX represents income before extraordinary items. All variables are normalized by the value of lagged book assets. Abnormal level of real activity manipulation variables are also constructed from the method of Kim et al. (2012) by obtaining the residuals from the following regression: where the set of Y variables includes operating cash flow, cost of goods sold, changes in inventory and discretionary expenses. The S variable indicates a firm's sales. All variables are normalized by the lagged book value of asset. The second source of data is used to proxy the degree of CSR practice in the retail industry. A firm's CSR performances in terms of environmental, social and governance practices are evaluated by the Korean Corporate Governance Service, mostly for the firms listed in the Korean Exchange. We select the total score of environmental, social and governance practices to represent a firm's overall level of engagements in CSR practices. To be precise, the environmental performance assess the managerial activities related to green marketing, pollutants from production, and sales of environmentally friendly goods. The social score measures a corporation's attitude and investment toward business ethics and sustainable growth. A firm's governance structure is related to well-functioning board structures, the quality of auditing, shareholder protections and so on. We combine these three scores to measure the overall effectiveness of CSR performance within a corporation. We adopt the following empirical models to test our hypothesis. The following equation illustrates our empirical models: where the Quality variable represents the quality of financial reporting. ESG here indicates the score of ESG, which is defined as the total of environmental, social and governance scores. The quality variable includes the absolute value of abnormal discretionary accruals (AB_DA) and the positive (P_DA) and negative values of abnormal discretionary accruals (N_DA). Kim et al. (2012) predict a significantly negative β on the ESG score for a higher quality of financial reporting with absolute and positive abnormal discretionary accruals and a significantly positive β with negative abnormal discretionary accruals. The quality variable also contains the proxy variables for real activity manipulations. Real activity manipulation variables are the level of abnormal operating cash flows, A_CFO, the level of abnormal production costs, A_PROD, and the level of abnormal discretionary expenses, A_EXP. The production cost refers to direct production cost (cost of goods sold) and inventory stock changes. The discretionary expenses indicate the total of advertising expenses, R&D expenses, and other indirect costs. C_RAM indicates its combination which is defined as A_CFO-A_PROD+A_EXP. Kim et al. (2012) expect positive coefficients on the ESG score for the case of abnormal operating cash flow, and the case of abnormal discretionary expenses and a negative coefficient for the case of abnormal production costs. We use the following control variables for our empirical examinations. A firm's SIZE is the natural logarithm of a corporation's book value of assets. The market equity to book equity ratio is the ratio between the market value of a corporation's equity and its book value of equity. ROA is measured by operating cashflow, scaled by the lagged book value of assets after industry adjustments. BIG indicates that a firm-year observation is audited by major auditing companies. Leverage, LEV is the debt to total asset ratios. RD represents a corporation's R&D intensity, where the measure is represented by R&D expenditure divided by current sales. ADV represents the advertising intensity for the two digit industry codes. AGE is firm age variable, which is natural logarithm of firm age. The sample of Korean retail firms mostly listed in the Korea Exchange is analyzed because the evaluation of CSR performance is restricted to these firms. The sample period covers from 2011 to 2016. All firm-level variables are winsorized at 1% level to mitigate outlier problems in our empirical examinations. For the estimation of these model, we use the method of cross-section regression models by adopting the ordinary least square method. Table 1 presents the summary statistics for the variables used in our empirical models. The ESG score and the proxy variables related to the quality of financial reporting are our main variables of interests. The table further contains summary statistics for the set of firm control variables used in our empirical models. Table 1 documents the mean of each variable with its standard deviation. It also contains the first quartile, the second quartile and the third quartile for each variable. Descriptive Statistics Table 1 documents significant variations in the proxy variables for the transparency of financial reporting. For instance, the mean of absolute value of abnormal discretionary accruals is 3.77 but its standard deviation is quite large at 4.18. Such significant variations are also observed even when we split the sample with positive and negative abnormal discretionary accruals. The proxy variables related to real activity manipulations also provide large variations as well. Such a large variation helps us examining the relationship between the ESG score and the quality of financial reporting. Table 1 also implies that almost all of the control variables related to firm characteristics have right skewed distributions. The average value of these control variables are generally larger than the corresponding median values. If we exclude the binary variables of indicating the auditing company and equity offerings, only one exception is the logarithm of firm age variables. This implies a large proportion of relatively young firms in the retail industry, which consistent to the survivorship patterns of a company. Table 2 provides correlation coefficients among the variables used in our examinations. The set of variables encompasses the ESG score, proxy variables related to the transparency of financial reporting, and other control variables in the empirical model. The table excludes correlations related to positive and negative abnormal discretionary accruals and combined real activity manipulation because these values are calculated from the construction of other variables of interests. Table 2 presents a wide range of interesting findings. Most importantly, the ESG score shows negative relationship with absolute value of abnormal discretionary accruals, which implies a higher quality of financial reporting. In contrast, the correlation coefficients of ESG scores with real activity manipulation variables are inconsistent with the prediction of Kim et al. (2012). For example, the ESG score has a negative correlation with the abnormal operating cash flow and positive correlations with the abnormal production costs, both of which are not in line with the hypothesis of Kim et al. (2012). Finally, we also observe relatively weak correlations between the proxy variables related to discretionary accruals and real activity manipulations. Such weak correlations are also in line with the contrasting correlation patterns between the quality of financial reporting and the ESG score, as documented above. Table 3 documents our estimation results for the entire firm-year observations in the Korean distribution industry. The table tests H1 from the perspective of the use of discretionary accruals as earning management tool. The use of discretionary accruals are divided into three subcategories: the absolute value of abnormal discretionary accruals, the positive abnormal discretionary accrual case, and the negative abnormal discretionary accrual case. The total of environmental, social and governance score, the ESG score is employed as the benchmark CSR measure. A higher quality of financial reporting implies negative coefficients on the absolute and positive values of abnormal discretionary accruals and positive coefficients on the negative value of abnormal discretionary accruals. The set of control variables that are depicted in the empirical model is included as well. The table reports the estimated coefficients with their corresponding standard errors from the model. For all following tables. the symbols of , , and ** captures the level of statistical significance at 10%, 5% and 1%, respectively. Table 3 reports a negatively significant relationship between CSR performance and the absolute value of abnormal discretionary accruals, which supports a higher quality of financial reporting for socially responsible firms. Specifically, the estimated coefficient on the ESG score is significantly negative at -1.793. Such a negative coefficient implies a more transparent financial reporting in socially responsible firms. Entire Sample The next two columns indicate that CEOs in the retail industry constrain to manage earnings significantly if they need to use discretionary accruals negatively. The ESG score shows significantly positive correlation with the dependent variable, when we consider the negative abnormal discretionary accrual case. Such a significantly positive coefficient implies more transparent financial information for socially responsible firms. The coefficient on the main proxy variable (ESG score) is not significant relationship in case of the positive discretionary accrual case. The results of Table 3 generally support our hypothesis H1 predicting a higher quality of financial reporting for socially responsible firms via the traditional way of earning management. The use of discretionary accruals for earning management is generally considered as a traditional and easier way in earning management. This finding is also consistent with the U.S evidence of Kim et al. (2012) as well. Note that the set of control variables have limited power to explain the use of discretionary accruals in our examination. For instance, except the market to book ratio, the coefficients on these firm characteristic variables are statistically insignificant. Such insignificant results rather highlight the importance of the ESG score in shaping the use of discretionary accruals. Table 4 provides our estimation results for the entire sample of Korean distribution corporations with a different measure of financial reporting quality. The table considers the proxy variables for real activity manipulation as the dependent variables. The abnormal cashflow, abnormal production costs, abnormal discretionary expense, and their combinations are examined separately for each column. The total of environmental, social and governance score, the ESG score is employed as the performance measure of CSR practices. A better quality of financial reporting implies significantly positive coefficients on the ESG score all the empirical models, except the case of abnormal production costs; a higher quality of financial information implies a negative coefficient on the ESG score. and positive coefficients on the negative value of abnormal discretionary accruals. The set of control variables of Table 3 is included as well. Table 4 contains the estimated coefficients from the cross-sectional model and their standard errors. Table 4 shows statistically insignificant relationship between our measure of CSR performances and the proxy variables for real activity manipulations. For all cases of abnormal cashflow, abnormal production costs, and abnormal discretionary expenses, the estimation does not find any significant relationships. Accordingly, the combined measure of real activity manipulation does not have a significant relationship with the ESG score, either. The results of Table 4 do not support our hypothesis H1 predicting a higher quality of financial reporting for socially responsible firms in terms of real activity manipulation. These results are also in line with the pairwise correlation coefficient reported in Table 2; the signs of correlation coefficients are all inconsistent with the prediction of Kim et al. (2012). The contrasting results in between Table 3 and Table 4 are tightly related to the characteristics of retail firms. In contrast to ordinary manufacturing firms, retail firms tend to focus on the distribution of products, which makes the use of real activity manipulation by using production cost or cash flow generation difficult. Retail firms may have tendency to rely on the traditional way of earning management, the use of discretionary accruals. Thus, more socially responsible firms within the retail industry may focus on the use of discretionary accruals rather than real activity manipulation, which increases the transparency of financial reporting only in terms of discretionary accruals. Subsample Analysis: Chaebol and Non-Chaebol Affiliates Now, we conduct a subsample analysis based on the categorization of chaebol and non-chaebol affiliates. As highlighted in recent studies such as Yoon et al. (2018), firm level heterogeneity affects the effectiveness of CSR performance on various corporate policies. Table 5 presents the examination results from our empirical model when we use the subsamples of chaebol affiliates. Table 5 examines the use of discretionary accruals as a measure for financial reporting quality. Similar to Table 3, Table 5 separately examines the absolute abnormal discretionary accruals, positive abnormal discretionary accruals, and negative abnormal discretionary accruals cases. The ESG score is also used to capture the degree of engagements in socially responsible activities. The set of control variables are identical to those of Table 3. The table shows the estimated coefficients, the standard errors, the total number of observations and the estimated R-square value. The estimation results for EO and BIG variables are missed because chaebol affiliates did not conduct equity offerings and were audited by major auditing companies during the sample periods. Table 5 clearly indicates that the retail firms within chaebol affiliates do not increase the transparency of financial reporting even in the use of discretionary accruals. For all three cases of abnormal discretionary accrual values, the coefficients on the ESG scores are not statistically significantly. The estimated values for the absolute and negative abnormal discretionary accrual cases are quite smaller than the corresponding values from the entire sample regressions documented in Table 3. Table 6 further shows thee examination results from our empirical model when we use the subsamples of chaebol affiliates with regard to the real activity manipulations. The table considers the 4 proxy variables for real activity manipulation as the dependent variables including the abnormal cashflow, abnormal production costs, abnormal discretionary expense and their combinations. The ESG score is still used as the proxy variable for a firm's CSR performances. The set of control variables remain the same as those of previous tables. The table includes the coefficients, corresponding standard errors, the number of observations and the adjusted R-square as well. The estimation results for EO and BIG variables are omitted because chaebol affiliates did not conduct equity offerings and were audited by major auditing companies during the sample periods. Table 6 shows no significant relationship between a firm's CSR performances and real activity managements within chaebol affiliates. All of the coefficients on the ESG score are not statistically significant in line with the entire sample examination reported in Table 4. Such insignificant relationships are consistent with our hypothesis, H2, which predicts limited quality improvement of financial reporting for more socially responsible firms within chaebol affiliates. As highlighted in Yoon et al. (2018), this finding is closely associated with economic environments of chaebol affiliates. Chaebol affiliates are under continuous monitoring from the Korean Supervisory Service. For instance, firms belonging to chaebol affiliates have restrictions in their debt issuance, cross-guarantee, and intra trading. A better quality of financial reporting is required as well and a number of empirical studies verify a more transparent financial reporting within chaebol affiliates. Because these firms already have a great quality of financial reporting, more socially responsible firms within chaebol affiliates may not have strong incentives to raise the quality of financial reporting even in the case of using discretionary accruals. Tables 7 and 8 provide the estimation results from our empirical model when we employ the subsamples of non-chaebol affiliates. Tables 7 and 8 examine the use of discretionary accruals and the use of real activity manipulation to measure finanical reporting quality. The uses of dependant variables in Tables 7 and 8 are in line with those of Tables 5 and 6. The ESG score is still used to capture the degree of engagements in socially responsible activities. The table reports the estimated coefficients and their standard errors. Table 7 documents a negatively significant relationship between CSR practices and the absolute value of abnormal discretionary accruals, implying more transparent financial information for socially responsible firms. The estimated coefficient on the ESG score is negatively significant at -7.520. In line with the result of entire sample, the ESG score shows significantly positive relationship with the negative abnormal discretionary accruals. This finding suggests that the results of entire sample are mainly driven by the sample of non-chaebol affiliates within the Korean retail industry. Table 8 presents statistically insignificant relationship between the ESG scores and the proxy variables for real activity manipulations. For all of the cases of abnormal operating cashflow, abnormal cost of production, and abnormal discretionary expenses and their combinations, we are not able to find any statistically significant relationship between the ESG score and the proxy variables for real activity manipulation. These findings are consistent to the results of Tables 4 and 6, and probably capture the relatively insignificant role of real activity manipulation in the Korean retail industry. Conclusions This study investigated how a firm's CSR performance in the Korean retail industry is related to the quality of financial reporting. Specifically, Kim et al. (2012) argue that more socially responsible firms have strong incentives to improve the quality of financial reporting because they are committed to ethical behaviors to stakerholders. This paper tests whether this hypothesis applies well for the publicly traded firms in the Korean retail industry. For this purpose, we measure transparency of financial reporting in terms of the use of abnormal discretionary accruals and real activity manipulations. The effectiveness of CSR practice is proxied by the total of environmental, social and governance scores graded from the Korean Corporate Governance Services. All other control variables in shaping the quality of financial information are constructed based on a firm's financial statement information provided by FnGuide. The heterogenous effects of CSR practices on the quality of financial information are also evaluated in accordance with the categorization of chaebol and non-chaebol affiliates. Our estimation results generally implies a more transparent financial reporting within socially responsible firms in the Korean retail industry. Retail firms with better CSR performances are less likely to manipulate earnings via the variation of discretionary accruals. These firms at least provide a similar quality of financial information as ordinary retail firm does in terms of real activity manipulation. However, our estimation results further confirm that chaebol affiliates do not tend to provide a greater financial reporting quality in both terms of discretionary accruals and real activity manipulation. Such insignificant results are quite distinctive from the estimation results in non-chaebol affiliates. Our findings are closely associated with the characteristics of retail industries and chaebol affiliates. Retail firms may have simple production procedures, which restricts the use of real activity manipulation to manage earnings. Thus more socially responsible firms within the retail industry only have substantial incentives to make a higher quality of financial reporting by controlling the use of abnormal discretionary accruals, not by restricting real activity manipulations. Chaebol affiliates are subject to strict monitoring from the Korean Supervisory Service, which make them produce high quality of financial reporting. Thus even socially responsibles firms may not have incentives to additionally improve the quality of financial reporting within Chaebol affiliates. Our work contributes to the extant literature from a number of perspectives. Most of all, we find generally supporting evidence for the hypothesis predicting a more transparent financial reporting in socially responsible firms. This finding is in line with the U.S. evidence of Kim et al. (2012) Furthermore, we also emphasize that the characteristics of retail industry and chaebol affiliates significantly affects the testing results consistent with recent studies such as Yoon et al. (2018). However, our work still does not perfectly resolve the issue of estimation biases from the adoption of ordinary least square methods. Furthermore, our sample is limited to the firms within retail industry and we have not conducted similar analysis for other industries, while our estimation results emphasize the importance of industry characteristics in shaping the effectiveness of CSR performances on corporate policies. These topics are left for future researches.	2019-07-22T06:02:17.622Z	2019-06-01T00:00:00.000	{ "year": 2019, "sha1": "9b3917c11ea687c848985dc51a4751329627871a", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.15722/jds.17.6.201906.33", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "ed3be52fed65b46067a169400a4ce42f6c68cfe0", "s2fieldsofstudy": [ "Business" ], "extfieldsofstudy": [ "Business" ] }
202889122	pes2o/s2orc	v3-fos-license	Communication-Constrained Routing and Traffic Control for Autonomous Vehicles Autonomous vehicles (AV) is an advanced technology that can bring convenience, improve the road-network throughput, and reduce traffic accidents. To enable higher levels of automation (LoA), massive amounts of sensory data need to be uploaded to the network for processing, and then, maneuvering decisions must be returned to the AV. Furthermore, passengers might have a higher transmission rate demands for various data-hungry and delay-sensitive applications. I. INTRODUCTION The technology of autonomous vehicles (AVs) has the potential to drastically reduce the energy consumption, traffic congestion, and collisions of vehicles [1], [2]. In [3], five levels of automation (LoA) of AVs are described: (0) no automation, (1) driver assistance, (2) partial automation, (3) conditional automation, (4) high automation, and (5) full automation. With different LoA, AVs can enable new applications, such as dynamic ridesharing, platooning, remote driving, to name a few [3]. To achieve a higher LoA, AVs need to, throughout the entire trip, constantly communicate with neighboring AVs, as well as with traffic control services running in the infrastructure that continuously monitor the status of each AV and its surrounding environment to make driving decisions. Specifically, because AVs might not have sufficient computing capacity for machine or deep learning based video/data processing, it has been proposed that portions of the data processing and driving decisions be delegated to the edge or to the cloud [3], [24]. Therefore, the availability, reliability, latency, and sustainable transmission rate of AV's connectivity with traffic control services running in the infrastructure are G. Liu, S. Salehi, C.-C. Shen, and L. J. Cimini are with the University of Delaware, Newark, DE (e-mail: guangyi, salehi, cshen, cimini@udel.edu). E. Bala is with InterDigital (e-mail: Erdem.Bala@interdigital.com). critical to achieve a high LoA [3]- [6]. However, with AV's high mobility and the associated dynamic environment, these communication requirements bring significant challenges to the cellular infrastructure [7]- [9]. To provide wireless connectivity with higher quality-ofservice (QoS) and reliability, fifth generation cellular systems (5G) wil support use cases such as enhanced mobile broadband (eMBB), massive machine type communications (mMTC), and ultra-reliable low-latency communications (URLLC) for mission-critical applications [7], [11]. To enable these use cases, disruptive technologies, such as massive multiple-input and multiple-output (MIMO) and millimeter wave (mmWave), are expected to be deployed [7]. In [14], it is shown that four out of the nine 5G deployment scenarios support vehicular communication. These are 1) Broadband Access in Dense Areas -providing 300 Mbps for vehicles with speeds up to 60 mph; 2) 50+ Mbps Everywhere -providing 50 Mbps with speeds up to 60 mph; 3) Mobile Broadband in Vehiclesproviding 50 Mbps with speeds up to 300 mph; and 4) Ultra-Low Cost Broadband Access for Low Average Revenue Per User (ARPU) Areas -providing 10 Mbps with speeds up to 30 mph. The actual deployment of 5G is likely to support a mixture of these scenarios across all regions. Among the LoA, high and full automation would heavily rely on constant communication between an AV and the backend services running in the infrastructure. Typically, 50+ Mbps high-reliability low-latency communications is required during the entire trip to facilitate higher LoA and to accommodate any higher data rates demanded by passenger applications [3]. However, even when 5G is fully deployed, among the nine 5G deployment scenarios, only the "Broadband Access in Dense Areas" use case can guarantee the sum transmission rate requirement imposed by the LoA of high and full automation. In addition, it could take several years to cover all the major areas with 5G, and the quality of communication is often worse around the cell boundaries. Therefore, without proper route planning, there is no guarantee that the sumrate requirement imposed by the high LoA of the AVs can be sustained during the entire trip, especially when the density of AVs is high along certain road paths. In order to facilitate high LoA for the AVs during the entire ride, here, we propose a communication-constrained routing (CCR) problem, where, given the finite and different amounts of communication resources along the different road segments (RSs) and intersections (RIs), the routes of AVs are deliberately selected, such that the minimum sum transmission rate requirements of the AVs can be satisfied at all times. In addition, we also propose a communication-constrained traffic control (CCTC) problem, where the communication resources available over different roads can be utilized to maximize road-network throughput. To the best of our knowledge, CCR and CCTC are new classes of routing and resource allocation problems, respectively, that have not yet been explored. Only a recent 3GPP document [3] elaborating the QoS aspects of automated driving (Section 5.27.2.1) and remote driving (Section 5.28.2.2) describe selecting routes that can support the required communication QoS. Previous work related to this topic includes energyconstrained routing and traffic optimization problems discussed in the context of electric vehicles [15]- [17]. However, the graph-based models adopted for road networks in electric vehicle routing cannot be properly adapted to either CCR or CCTC. The difficulties can be attributed to the fact that (1) there are no detailed models connecting the key performance indexes (KPIs) of wireless communication with those of AV movement; and (2) the edge weights of any routing graph computed from the sum transmission rates of wireless communication may not be stable due to the fast time-varying nature of wireless channels. These two technical issues will be specifically addressed in this paper. From the aspect of shortest path routing, one major challenge is to reduce the time complexity for large networks [10], [15], [18]- [20]. In practice, a subset of critical nodes are often selected to divide the network into multiple layers, such that hierarchical routing can be applied with some preprocessing [2], [18]- [20]. However, the performance of hierarchical routing is strongly dependent on the selection of these critical nodes, which is often suboptimal. For CCR in large road networks, computing the theoretical minimum-triptime route is a critical issue that is also addressed in this paper. In terms of resource allocation, traffic control for AVs is an important issue that has been studied [21], [22]. Without taking the communication constraints into consideration, backpressure-based traffic control, which maximizes the roadnetwork throughput, is proposed in [21]. However, coupled with the CCR problem, when the density of AVs is high, the limited communication resources may not be able to satisfy the communication needs of all AVs. Under this constraint, we also study how to optimally utilize the available resources to maximize the road-network throughput. In this paper, we first describe an approach to modeling both the communication constraints for AVs and their effects on providing reliable wireless connectivity for higher LoA and reliable control. Then, by focusing on one individual AV, and without considering the impacts from other AVs, a nonconvex optimization problem is formulated to minimize that AV's trip duration, subject to a transmission rate requirement. For faster routing computation in large road networks, a new hierarchical routing scheme is proposed, which divides the computed routing graph into a top layer containing all the base stations (BSs) and a bottom layer containing local graphs made up of RIs and RSs. The optimality of this hierarchical scheme is demonstrated. In addition, simulation results show that, compared with certain greedy algorithms, this two-layered routing scheme provides longer communication coverage durations and acceptable source-to-destination trip duration. For optimal traffic control, we focus on maximizing the road-network throughput subject to a constraint on the communication resources within a certain area. This is due to the fact that the complexity of performing city-wide, joint traffic control and route planning of all AVs is high. Therefore, we allow AVs to be navigated in a non-cooperative manner and focus on improving the road-network throughput for a given area. Specifically, to characterize the traffic control performance of wireless communication for AV traffic control, we propose and investigate a new KPI for traffic control in cellular systems. Using this KPI, two lemmas are proved to determine the optimal speed of each AV and optimal frequency channel allocation across multiple cells. Numerical results are provided to show the performance gain. The rest of the paper is organized as follows. In Section II, the impacts of communication on AV routing and traffic control are discussed, motivating the introduction of two concepts to facilitate the study of CCR and CCTC. Section III studies the CCR problem, where a road network is modeled as a two-layer graph to facilitate a two-layered communicationconstrained routing scheme. Simulation results are presented to demonstrate the effectiveness of the proposed scheme. Section IV investigates the CCTC problem by first computing AV speed within a single cell to maximize road-network throughput. Then, to match traffic flow across adjacent cells, a spectrum re-balancing solution is introduced to maximize AV road-network throughput across multiple cells. The effectiveness of the solution is validated via numerical studies. Section V concludes the paper and suggests future research directions. II. IMPACTS OF COMMUNICATION ON ROUTING AND TRAFFIC CONTROL AVS This section discusses the impacts of communication on the routing and traffic control of AVs, and introduces effective speed map (ESM) and γ-rate cell of a BS. The speed of the AVs is a critical parameter for both routing and traffic control. For routing, given the distance of a road path between the source and destination, the speed determines the driving time for an AV to complete its route. For traffic control, the throughput of the road network directly depends on the speed of the AVs. To decide the speed of the AVs, the reliability, transmission rate, and latency of communication between AVs and the traffic control service running in the infrastructure are critical, in addition to the physical conditions of the roads, the density of vehicles, and the speed limit (v l ) imposed by the transportation authorities [3], [5], [23]. For instance, if the AVs have faster reaction times, they might be able to move faster and travel at closer distances with each other without compromising safety. This reaction time strongly depends on the reliability, transmission rate, and latency of the AV communication. In this paper, we propose an approach that uses the speed values, computed by maximizing the throughput of a road network (from Lemma 1 in Section IV), as the effective speed (values) of AVs for route planning (in Section III). Specifically, these speed values are associated with RSs of a road network to form its effective speed map (ESM) to be used by the routing function. For computing the trip duration and planning the routes, as with current vehicle navigation systems, we assume that the ESM does not change for the time period that a target AV is routed, and this AV travels exactly at this effective speed. B. γ-Rate Cell of a BS In [11], BSs would control the AVs. For a BS, geometrically, we define its γ-rate cell to be the largest area surrounding the BS such that (1) the probability of having a downlink transmission rate 1 more than γ Mbps at any location within the area, is higher than a threshold of 1− , where is a very small number, and (2) for any two locations on the roads within this area, there exists a road path within the same area connecting them. Thus, each γ-rate cell is one contiguous region. Based on this definition, with rates γ 1 < γ 2 , for a specific BS, its γ 1rate cell contains its γ 2 -rate cell. The two γ-rate cells of two adjacent BSs are connected if their overlapping area covers at least one common RI or one common RS. For example, owing to the channel hardening effect of massive MIMO [12], smallscale fading is mitigated and the magnitude of variations of transmission rates is small, so that the boundaries of γ-rate cells are stable over time even when goes to 0. Although it is not necessary to keep track of the exact boundary of a γ-rate cell, for AV navigation, it is good enough to keep track of all the RIs and RSs within the γ-rate cell of a BS. In order to compute the γ-rate cells for the BSs, the probability that the transmission rate is above a given threshold γ should be recorded every few meters along the RSs. To obtain this probability corresponding to this distance, all the rate measurements within this interval should be reported through the BSs to the route planning function, together with the corresponding GPS locations. III. COMMUNICATION-CONSTRAINED ROUTING (CCR) The CCR problem can be stated as follows: find the shortesttime path from the source location to the destination location where the given minimum transmission rate can be sustained along the entire path at all times. In this section, we focus on routing one AV in a non-cooperative (selfish) manner. The AV is fully autonomous with collision avoidance, so that the time spent by this AV on waiting at RIs and yielding to pedestrians and other AVs is considered negligible [2], [5], [21]. A road network with deployed BSs over a given area, as depicted in Fig. 2(a), can be modeled as a graph G road (V, E, M) [16], where node set V denotes the set of RIs, edge set E represents the set of RSs connecting adjacent RIs, and M is the set of BSs deployed in the given area. For BS m ∈ M, G m road denotes the graph representing the part of the road network located within m's γ-rate cell, as depicted in Fig. 2(b). For both G road and G m cell , the weight of an edge (RS) is the travel time of an AV over this edge, which is computed as the length of this edge divided by the corresponding effective speed obtained from the ESM, which was discussed in Subection II-A. A. Two-Layered AV Routing Scheme In [18], a road network was also modeled by a graph similar to G road but without the component M. To reduce routing complexity, [18] selected a subset of critical RIs to perform hierarchical routing. However, the performance of this approach strongly depends on the RIs selected and is usually suboptimal. As in urban scenarios where the number of BSs is much smaller than the number of RIs (for instance, some cells have radii of as much as 1-2 km [27]), in contrast to [18], we propose to divide G road (V, E, M) into two layers to facilitate a two-layered routing scheme. The top layer is denoted by graph G BS representing the connectivity among γ-rate-cells, and the bottom layer consists of all the graphs of γ-ratecells (G m cell ) of these BSs (as shown in Fig. 2(c)). Over these two layers, inter-γ-rate-cell routing and intra-γ-rate-cell routing are conducted in sequence followed by a process of dynamic programming to compute the shortest path between the source and destination locations. The optimality of the proposed scheme is demonstrated at the end of this subsection. Specifically, in G BS , there exists an edge between two γ-rate cells only when these two γ-rate cells are connected. Given that two γ-rate cells may be connected in the road network, via multiple common RIs and/or RSs, the exact travel time (i.e., weight) over a (top-layer) edge in G BS between any two adjacent γ-rate cells is initially undefined. Therefore, for interγ-rate-cell routing, Breadth-First Search (BFS) is applied to find the set of all possible (top-layer) paths from the source γ-rate cell, through zero or more intermediate γ-rate cells, to the destination γ-rate cell 2 . Let P = {P 1 , P 2 , · · ·, P \|P\| } be the set of (top-layer) paths found, and \|P i \| be the number of γ-rate cells traveled in path P i . For path P i , [γC i 1 , γC i 2 , · · · , γC i \|Pi\| ] denotes the sequence of γ-rate cells traveled (see the schematic diagram in Fig. 3) , where γC i 1 and γC i \|Pi\| are the source and destination γ-rate cells in G BS , respectively. For path P i , among the common RIs and RSs only between two adjacent γ-rate cells, the AV must pass through only one of them. We denote the common RIs between two adjacent γ-rate cells, as well as the mid-points (as indicated in Fig. 2(b)) of the common RSs that have no RIs on them, as core nodes. The set of core nodes between γC i j and γC i j+1 is denoted as K i,j = {k i,j 1 , k i,j 2 , ..., k i,j Ij }. By using the Dijkstra shortest-path algorithm, the intra-γ-rate-cell routing computes the minimum travel time T i,j n,m from core node k i,j−1 n (between γC i j−1 and γC i j ) to core node k i,j m (between γC i j and γC i j+1 ), for all k i,j−1 n in K i,j−1 and k i,j m in K i,j on G j cell 3 . In addition, let T i,j opt denote the optimal travel time from the source location to core node k i,j m , through path i, and T i,\|Pi\| 0 the optimal travel time from the source location to the destination location, through path i. After inter-γ-rate-cell and intra-γ-rate-cell routing, dynamic programming is applied to compute T i,j opt , and eventually T i,\|Pi\| 0 . Overall the two-layered routing scheme is summarized as follows: Step 1: Inter-γ-rate-cell routing. Find the set of all possible (top layer) paths on G BS , P, using BFS. Step 2: Intra-γ-rate-cell routing. Use the Dijkstra shortestpath algorithm to compute the minimum travel time T i,j n,m , ∀ i, j, m, n and record the time with the corresponding path on the road network. Step 3: n,m ). The shortest duration trip is then chosen to be min i T i,\|Pi\| 0 , with P i * being the chosen top-layer path. The bottom-layer road paths can be obtained from the recorded results in Step 2. Remark 1: The necessary and sufficient condition for the two-layered routing scheme to have a solution is that the path set P obtained in the inter-γ-rate-cell routing in Step 1 is not empty. If no result is obtained at Step 1, one option is to reduce the rate threshold γ until a path can be found. Remark 2: The proposed scheme is optimal, as shown next. Each route from the source location to the destination location corresponds to one distinct top-layer path (P i ), and in Step 1, all such top-layer paths, P, are found. For P i , in Step 3, the shortest travel time T i,\|Pi\| 0 is obtained using dynamic programming. Comparing T i,\|Pi\| 0 for all i, all results are exhaustively examined, and the route with minimum T i,\|Pi\| 0 is the optimal solution to the problem. B. Simulation Results We consider a Manhattan-like area as shown in Fig. 4, where A = 11 avenues and S = 51 streets divide the area into a grid topology of identical rectangles 4 . The corners of these rectangles are at the centers of the road intersections, the lengths and widths of which are L = 250 m and W = 100 m, respectively. K = 21 BSs are uniformly deployed in this area and all are equipped with N t = 128 transmit antennas. Different from the realistic traffic patterns in Manhattan, we assume that all streets and avenues allow bidirectional traffic flows. Given that there exists no other AV routing scheme subject to communication constraints, we compare the proposed two-layered AV routine scheme against two greedy routing 1) Greedy routing without communication constraints: In the scenario of a grid topology, typically, when the destination is located, say, northwest of the source, at each RI, an AV would choose the next RS that is (close to) either north bound or west bound, but neither east nor south bound, in order to travel the shortest distance. Specifically, at each RI, for all available RSs that lead closer to the destination, one is randomly chosen as the next RS. 2) Greedy routing subject to communication constraints: In this scheme, while driving, the AV learns the availability of local RSs that could satisfy its communication constraints (from the information broadcast by BSs). Upon arriving at an RI, the AV will choose, among all the available local RSs that satisfy its communication constraints, the RS that leads closer to the destination. In the worst case, the AV has to backtrack (i.e., make a U-turn) to a previous RI to choose a different RS (that satisfies the communication constraints). Using specific values of ESM and γ, Fig. 4 shows the routes computed by different routing schemes. The 'circle' around each BS represents its γ-rate cell. "Greedy w/o CC" and "Greedy s.t. CC" are the greedy methods introduced above, and "Shortest Time" chooses the shortest-time path using the Dijkstra algorithm without communication constraints. For the route computed by "Greedy s.t. CC", although it is mostly covered by γ-rate cells, its track shows backtracking behavior. In the simulation, we randomly generate the effective speeds for all RSs 10,000 times, from the set {10 m/s, 20 m/s, 30 m/s}. These speeds are consistent with those used in [25]. The slot length in an LTE-A system, τ = 1 ms, is used as the channel measurement interval. Also, the Winner II path loss model for urban areas [28] is applied, and the transmission rates are computed using the method in [26]. Fig. 5 depicts the CDF of the resulting trip duration for the different routing schemes with γ = 55 Mbps. The results show that the proposed two-layered routing scheme can achieve a trip duration close to the shortest-time routing while satisfying the communication requirements. Note that, due to the potential backtracking in "Greedy s.t. CC," an upper bound on the maximum number of RSs traveled is set. When this limit is reached, to reach the destination, the AV switches from "Greedy s.t. CC" to "Greedy w/0 CC", as noticed by the "stepping" appearance of the yellow curve in Fig. 5. that the proposed two-layered routing provides good communication coverage. In addition, because the "Shortest Time" scheme usually chooses RSs with higher effective speeds, the Doppler effect affects the transmission rates, resulting in worse coverage than "Greedy w/o CC." Furthermore, we compute the metric"successful trip percentage" defined as the probability that the target AV can find a route that satisfies the rate requirement along the entire trip. By varying the value of γ, the results, as shown in Fig. 7, show that, with increasing γ, it becomes more difficult to find a satisfactory route; however, the two-layered routing scheme always outperforms the other methods using this performance metric. In addition, we vary the number of BSs in the Manhattan area. The average P c (average percentage of trip duration covered by the required transmission rates), as well as the successful trip percentage, are plotted in Figs. 8 and 9, respectively. These two figures show that, to achieve a specified communication requirement, with CCR of AV, fewer BSs are required for deployment, but with a longer trip duration. In the last section, we considered the routing of a single AV and ignored the impacts of other AVs, even though a BS can serve multiple AVs simultaneously. From the viewpoint of traffic control, maximizing the throughput capability of a road network is one primary goal of adopting AVs [21]. However, to achieve the optimal CCTC of maximizing the road-network throughput subject to communication constraints, several interrelated issues remain to be studied. In this paper, we study two specific problems: (1) in one-cell scenarios, given the constrained communication resources, we investigate the number of AVs that can be served simultaneously by the BS as a new KPI for traffic control, and use the result to derive the optimal AV speed for maximizing the "sum traffic flow" into/out of the cell, and (2) in multi-cell scenarios, we study a traffic flow capacity mismatch problem across different cells along the same road, and propose a spectrum re-balancing solution to address the issue. Numerical results demonstrate that the one-cell and multi-cell solutions can be applied in sequence to maximize the road-network throughput. A. Speed Optimization within Single Cell As discussed in [3], to enable high LoA for AVs, certain minimum communication requirements of reliability, transmission rate, and latency, denoted by Γ = [R 0 , G 0 , A 0 ], must be satisfied for each AV 5 . We propose the following KPI, connecting Γ with AV traffic control, to indicate the traffic control performance of a single cell. N v Γ (B), which denotes the largest number of AVs that a BS can control simultaneously subject to the communication requirements of Γ, where AVs are uniformly distributed within the BS's coverage area, all AVs move at speed v, and B is the number of available frequency channels in this cell. Note that N v Γ (B) monotonically increases with B and decreases with v, given the same Γ. The decrease with speed results from the fact that more communication resources would need to be consumed for each AV traveling at higher speeds to satisfy Γ. Given the the same amount of available resources, fewer AVs could be controlled when traveling at higher speeds. For simplicity, we do not consider optimal channel allocation across channels; then, for identically distributed frequency channels within a cell, there exists a value B m such that the following linearity property holds. When there is only one frequency channel, a BS can still control multiple AVs by using time-division multiplexing (TDM) or frequency-division multiplexing (FDM) to divide the channel into multiple subchannels, as well as by using MIMO. Here, we assume this linearity property always holds. For simplicity, we assume that, within a cell, there is no lane merging and no dead ends, so that the numbers of lanes, L, into and out of a cell are identical. (For instance, in Fig. 10, for the red cell, L = 2; and ,for the green cell, L = 4.). We define BS Coverage C as the total surface area of the roads that the BS covers. The width of each lane is assumed to be W; then, the total length of all lanes in this cell is C W . Let D(v) denote the minimum distance between two AVs within the cell set by the transportation authority subject to effective speed v. For a steady flow of traffic, we consider that AVs are uniformly distributed in the BS's coverage area with speed v. Then, in theory, the distance between two consecutive AVs would be max{D(v), }. The result of dividing this distance by v becomes the average inter-arrival time between two consecutive AVs entering (or leaving) the cell. We define cell sum traffic flow, F m , to be the maximum number of AVs that is allowed to enter or leave a cell per minute. Then, by applying the equation on p. 319 in [33], In practice, D(v) is small and can be eliminated from Eq. (1). Therefore, the average distance between consecutive AVs be- 5 The higher the LoA, the more stringent the communication requirements Γ. . The result of dividing this average distance by v is the average inter-arrival time between two consecutive AVs entering or leaving the cell. Then, Note that (2) holds even for non-steady-state flows, where the inter-arrival time between AVs entering a cell is not uniformly distributed. Then, for the one-cell case considered here, the following lemma shows that each cell has an optimal speed for maximizing the road-network throughput in that cell. Lemma 1: Given the fixed amount of communication resources in a cell and the minimum communication requirements Γ for a specified LoA, there exists an optimal speed, and hence an optimal number of AVs, such that the roadnetwork throughput in this cell is maximized when all AVs travel at this speed. Proof. Given a fixed amount of communication resources in a cell, the relationship between N v Γ (B) and the AV speed v is a unique function, and N v Γ (B) monotonically decreases with v. N v Γ (B) peaks at v = 0. Also, there exists a speed threshold v m such that, when all the AVs in the cell travel at a speed above this threshold, they become uncontrollable, i.e., N v is the corresponding optimal number of AVs. For this cell, this value of sum traffic flow F * m is its maximum road-network throughput. Lemma 1 only proves the existence of an optimal AV speed. To derive the optimal AV speed, it becomes necessary to obtain the unique function between N v Γ (B) and AV speed v. This function depends on the technology employed by the BS, which can be obtained by using either statistical methods or model-based methods. In Subsection IV-C, a model-based method is introduced for a Time Division Duplex (TDD) BS. Also note that the optimal AV speed can be used for constructing the ESM for the CCR problem from the previous section, and this speed could differ in different cells. B. Spectrum Balancing across Multiple Cells For reasons such as severe channel fading and/or a high number of AVs, it is possible that the available communication resources within a cell are being exhausted so that it will not be able to support the desired LoA when more AVs enter the cell. Fig. 10 depicts such a scenario where the green cell does not have enough communication resources to accommodate more AVs coming from the red cell, so AVs may need to back up at the cell boundary. For a road crossing multiple cells, (communication constrained) traffic flow capacity mismatch between adjacent cells can lead to traffic congestion in a cell that supports less traffic flow. Among all the cells that a road travels across, we term the cell with the least amount of traffic flow capacity bottleneck cell. The overall traffic flow of a road is then constrained by the capacity of the bottleneck cell. Therefore, to avoid congestion, ideally, all the cells along a road should support the same traffic flow capacity. We provide Lemma 2, which shows that dynamic channel allocation across different cells provides a solution to mitigate the negative effects of the bottleneck cells. For instance, in current cellular systems, adjacent cells often use different frequencies to avoid inter-cell interference. Therefore, given the total number of channels B 0 over an area, more channels should be assigned to the bottleneck cells to increase their allowed traffic flows, and hence the overall throughput of the road crossing multiple cells. Consider the following scenario for Lemma 2. An area containing L r one-way roads (a two-way road can be considered as two one-way roads) is fully covered by N e hexagonal cells that do not overlap. Within this area, in total, there are B 0 frequency channels, and within a cell, each channel can control the same number of AVs. No two cells in this area share the same channel. BS i assigns B ij frequency channels for the AVs on road j, such that the maximum traffic flow on this section of road j within cell i is F ij . i,j B ij = B 0 . We consider FDM and assume that each frequency channel can be divided into a large number of subchannels to control multiple AVs. Also, it is allowed to assign a fraction of the channels to a cell. The proof for TDM is similar. Lemma 2: Given B 0 frequency channels to control AVs over an area containing L r one-way roads, a necessary and sufficient condition for deciding assignments B ij to achieve Pareto-optimal road throughput among the L r roads in this area is that, for each road j, F i1j = F i2j for all BSs i 1 and i 2 whose cells road j travels across. Proof. Randomly assign all B 0 channels across different roads and cells. Suppose road j crosses a sequence of cells S j = < s 1 , s 2 · · · s I(j) > in turn. Then, the traffic flow capability on this road should be F j = min si in Sj F ij , which is constrained by the cell with the smallest F ij . After the initial assignment, repeatedly shift channels or their subchannels from the cell with highest F ij to the cell with lowest F ij , one subchannel at a time, until F i1j = F i2j (if F i1j = F i2j is not achievable due to the fact that the number of AVs has to be an integer, then at least F i1j ≈ F i2j ), ∀ s i1 , s i2 in S j . Repeat this process for all L r roads in this area. With these adjustments, the road throughputs of L r roads all become higher. To further improve the throughput for road j, channels or subchannels need to be reassigned from the other roads. Because each channel can control the same number of AVs within a cell, if channels are switched between two roads, the number of channels for each of the roads stays the same, so that their allowed traffic flows stay the same. However, borrowing channels or subchannels is always at the cost of the other roads' throughputs. Thus, as long as F i1j = F i2j , ∀ s i1 , s i2 ∈ S j , and ∀ j, it is impossible to improve the throughput of a certain road without compromising another, and hence the throughputs for the L r roads are Pareto-optimal. Note that when frequency reuse is allowed across cells such that two cells not adjacent to each other may share the same spectrum, or when different subchannels can control different numbers of AVs, "F i1j = F i2j for all BSs i 1 and i 2 " in Lemma 2 is only a necessary condition for achieving Paretooptimal road throughput. Remark 3: From the two lemmas, it is possible that two adjacent cells may have different optimal speeds and different optimal distances between AVs. In practice, the movement control services running in the infrastructure would foresee such situations to gradually slow down or speed up AVs to avoid any abrupt change of speed when crossing cell boundaries along the road. C. Illustrative Scenarios 1) Derivation in one-cell scenarios: We consider the speed optimization problem in a single-cell slotted TDD MIMO scenario, where the duration of each slot is T slot . By periodically sending T pilot -length (in sec) pilots to measure the channels, the BS can obtain accurate channel information; this information is critical for satisfying the downlink communication requirements Γ. In the downlink, the BS sends periodic messages to each AV at a repetition rate of λ m , which means λ m messages are sent per second. The duration of one message is T m , in sec, which is an integer multiple of T slot . Because of the linearity property discussed above in Subsection IV-A, we focus on one frequency channel in the analysis, and use TDM for sharing the channel. Given accurate channel information, and using one frequency channel to send the T m -length messages, L-user-MIMO technology is applied in the BS, such that, at every downlink time slot, Γ can be satisfied for a maximum of L AVs. Note that, in order for the repetitive message transmissions to be successful, the message inter-arrival time, 1/λ m , should be larger than the duration of one message, T m ; then, λ m T m < 1. To obtain channel information that is sufficient and accurate enough to satisfy the Γ requirements, each AV must periodically send pilots to measure its channel at a maximum , where f D (v) is the AV's Doppler frequency and α is a scaling factor larger than 1. where f c is the carrier frequency and c is the speed of light. The '≈' is because in practice T v should be an integer multiple of T slot . In the frequency domain, one channel contains multiple subchannels, and during the T pilotlength measurement of one AV, this AV's pilots are sent over all these subchannels; however, since these pilots are sent at the same time during T pilot , we focus only on the time domain in the analysis below. Note that, within one frequency channel, TDM is used for multiplexing; dividing the frequency channel into multiple subchannels is for OFDM rather than for FDM. From the discussions in the two paragraphs above, in the time domain, the slots are occupied by the AVs, sending periodic channel measurement pilots in the uplink, in turn, as well as receiving messages from BSs periodically in the downlink, with at most L AVs sharing one downlink slot. Furthermore, within a time period T (T T m ), each AV would require roughly T /T v pilots, such that N v γ (1) AVs would require N v γ (1)(T /T v ) pilots, which take Compared with former generations of cellular systems, in 5G, more flexible TDM is allowed, as shown in Fig. 2 in [35]. Specifically, adjacent slots can be independently used for uplink and downlink transmissions. With such flexibility, the numbers of uplink slots and downlink slots can be arbitrarily specified. As shown in Fig. 11, with the increase in the number of AVs, eventually, it reaches a point where the time slots in T are compactly filled by pilots and messages of 3L AVs, and no additional AVs can be accomodated. Thus, with flexible TDM in 5G, the sum lengths of all uplink pilots and downlink messages (both in sec) within duration of T , can occupy a duration close to T . For the TDD MIMO scenario in this subsection, let When T slot is small, where N 0 is the set of non-negative integers. And when is an integer, the theoretical maximum number of AVs in the cell, from the viewpoint of filling up time slots, is Given (5), it can be proved that F m in (2) reaches its peak Therefore, the best strategy to maximize where v l is the speed limit imposed by the transport authority. The corresponding theoretical maximum cell sum traffic flow is 2) Example in a two-cell scenario: To demonstrate Lemma 2, we consider a frequency channel allocation problem using the two-cell scenario. As shown in Fig. 12, there are two horizontal lanes in the red cell, and two horizontal lanes and two vertical lanes in the green cell. In the red (green) cell, the AVs move at identical speed and the distances between two adjacent AVs on each lane are identical, thus the lanes' corresponding AV throughputs in the red (green) cell are also identical. This means, for the red cell, the cell sum traffic flow is the same as the sum traffic flow of the cell's horizontal lanes; and for the green cell, the former is twice the latter. Then, to achieve Pareto-optimal road-network throughput, intuitively, the number of frequency channels assigned to the green cell, B g , should be larger than that of the red, B r , to make the traffic flows on the horizontal road smooth. Denote the tuple of the number of lanes, the cell coverage, the optimal AV speed, and the maximum number of AVs, for the green cell, as (L g = 4, C g , v g , N vg g,Γ (B g )), and for the red cell as (L r = 2, C r , v r , N vr r,Γ (B r )). Then, using Lemma 2, the sum traffic flow of the red cell's horizontal lanes is the same as that of the green cell', and the cell sum traffic flow of the green cell is twice that of the red For the rest of this section, we consider the special case when the geometries of the two cells are identical; specifically, C g = 2C r , v g = v r , and the two functions N v g,Γ (B) = N v r,Γ (B). Then, solving Eq. (8), the optimal number of allocated channels are B g = 2B r = 2/3B 0 . Compared with equal channel allocation B r = B g = 1/2B 0 , it can be verified that B g = 2B r = 2/3B 0 brings an additional road-network throughput gain of 33.3% for this two-cell scenario. 3) Numerical results: Here, we use the one-cell scenario derivations in this subsection for both the red and green cells in Fig. 12. Let W = 3 m and the BS coverage of the red cell C r = 12, 000 m 2 . L = 10. Since 5G allows mini-slot of duration 0.1 ms, for the length of uplink channel measurement pilot, we let T pilot = 0.5 ms [35]. According to [3], [30], λ m ∈ [10, 100] Hz and T m ∈ [1, 100] ms. For simplicity, it is assumed that v l is large enough to be neglected. Then, for a total of B 0 = 10 frequency channels, the theoretical roadnetwork throughputs for the scenario in Fig. 12 are plotted in Figs. 13 and 14, for different values of α, f c , λ m T m , and different traffic control approaches. "Naive Approach" is the benchmark when neither of the optimizations in Lemma 1 and 2 are used; specifically, in both cells, AVs move at a suboptimal speed v * /2 and B 0 is equally allocated to the two cells. In contrast, "Use Lemma 1" uses optimal speed v * and "Use Lemma 1 & 2" utilizes the optimization results of both Lemma 1 and 2. From Fig. 13, we can see that, the theoretical roadnetwork throughput decreases with f c and α. It means that the throughput decreases with the required amount of channel measurements (the increase of channel measurements is caused either by faster-varying channel as indicated by f c , or by more stringent channel accuracy requirement as indicated by α). From Fig. 14, we can see that the theoretical roadnetwork throughput decreases with λ m T m . It means that this throughput decreases with the amount of downlink messages required by AVs (the increase of downlink messages is caused either by being sent more frequently as indicated by λ m or by longer messages as indicated by T m ). The unsmoothness of "Naive Approach" curves in Fig. 14 V. CONCLUSION AND FUTURE WORK In this paper, the problems of CCR and CCTC are motivated and investigated. For CCR, a two-layered routing scheme, including both inter-γ-rate-cell and intra-γ-rate-cell routing, is proposed, and its performance is compared with two greedy solutions. For CCTC, optimal effective speed and optimal frequency channel allocation across adjacent cells are derived to maximize theoretical road-network throughput within each cell and Pareto-optimal road-network throughputs across multiple cells. The CCR and CCTC problems have variations that remain to be formulated and resolved. For instance, in CCR, except rate requirement, AV applications may have various communication QoS and edge computing requirements [36]. The co-existence of AVs and other cellular users can change the communication resources in each cell available for AVs, in which case the graph of road network is time-varying [31]- [32]. For CCTC, except maximizing road-network throughput, there are other goals in AV traffic control, for instance, minimizing trip duration and increasing energy efficiency. Also, in practice, more advanced technologies, like optimal subchannel allocation for broadband communication and the collaboration of multiple BSs, can be used to increase the number of AVs one BS (or BSs) control simultaneously. In addition, some practical factors might be involved for more accurate modeling in CCR and CCTC, for instane, acceleration (deceleration) and group behavior in AV movement control; different AVs may have heterogeneous communication requirements and LoAs, and different BSs may provide heterogeneous communication capabilities; and the road network may be complex involving road signs, lane merging, roundabout, and parking capability.	2019-09-26T17:49:33.000Z	2019-09-26T00:00:00.000	{ "year": 2019, "sha1": "21f6a4f92a093af5977d08c5c561a3a0141b2ebc", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "21f6a4f92a093af5977d08c5c561a3a0141b2ebc", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Engineering", "Computer Science" ] }
237777505	pes2o/s2orc	v3-fos-license	Teachers' Self-efficacy for Using Information and Communication Technology: The Interaction Effect of Gender and Age The purpose of the study is to examine the moderating effect of age on gender differences in teachers’ self-efficacy for using information and communication technology (ICT) in teaching as well as possible variables underlying this effect. Following Bandura’s conceptualisation of self-efficacy, we defined teachers' self-efficacy as their confidence in performing specific tasks that require the integration of ICT into the teaching practice. The study was conducted via an online questionnaire on a sample of 6613 elementary and upper secondary school teachers in Croatia. The hierarchical multiple regression analysis was applied. The findings indicate minor gender differences in self-efficacy for using ICT that are more prominent among older teachers and practically non-existent among younger teachers. These effects remain statistically significant after controlling for the type of school where the teacher works, perceived technical and professional support for using ICT in school, and frequency of use of computer programmes in teaching. The interaction effect ceases to be statistically significant after the introduction of length of computer use in teaching and/or attitudes towards computers in the model, indicating that these two variables have a role in low self-efficacy for using ICT among older female teachers. A similar level of self-efficacy for using ICT among young male and female teachers is an encouraging finding which could hopefully be followed by gender equality in other aspects of ICT use. The findings suggest that strategies for enhancing ICT self-efficacy should be particularly targeted at older female teachers. This study contributes to a better understanding of the underresearched topic of gender differences in teacher’s ICT self-efficacy. Introduction Today's society witnesses a fast development and implementation of information and communication technology (ICT) in all areas of societal life, including education. Research indicates that teachers play a key role in the successful integration of ICT in schools. In this context, teachers' computer (ICT) self-efficacy and their self-confidence to successfully use ICT in teaching practice are particularly important (Albion et al., 2011;Hong et al., 2014;Krause et al., 2017). Teachers' ICT self-efficacy gained such a prominent role because it has a powerful influence on the teachers' behaviour and achievements. It determines whether the teachers will use ICT, in what way, to what extent and how successful they will be in their use of ICT for instructional purposes. According to Joo et al. (2018:50), teachers' selfefficacy can significantly motivate them to use new technologies in the classroom. Research also indicates that teachers' ICT self-efficacy depends on a number of factors, such as the 1 teachers' age and gender (Scherer and Siddiq, 2015), technology-related attitudes and affect (Cai et al., 2017;Gudek, 2019), teachers' computer experience (Sarfo et al., 2017) and school support (Hatlevik and Hatlevik, 2018). Gender differences in ICT use and related constructs have received a considerable research attention and have been widely studied in recent years. However, the existing empirical studies show inconsistent and even conflicting results in terms of gender differences in ICT attitudes and behaviours. Meta-analyses (Cai et al., 2017;Whitley, 1997) summarising the findings of individual studies conducted over a period of about 30 years report that this inconsistency in findings blurs the clear understanding of gender differences and makes it difficult to draw reliable conclusions about them. Cai et al. (2017) and Whitley (1997) explain this ambiguity of results as a function of various definitions and operationalisations of the researched constructs and diversity of studied populations. Despite the ambiguity of findings, meta-analyses indicate that gender differences were minimally reduced, especially regarding self-efficacy and affect dimension of attitudes towards ICT use. Males still harbour more favourable attitudes towards ICT use and report a higher level of ICT selfefficacy than females, although the effect sizes of these differences are small (Cai et al., 2017;Whitley, 1997). Such findings emphasise the research and practical need for a continuous exploration of gender differences in ICT use, including the self-efficacy of teachers. This study is aimed at exploring effects of gender and age on teachers' self-efficacy in using ICT for teaching purposes, while controlling for other variables which influence teachers' ICT self-efficacysuch as technology-related attitudes and affect (Cai et al., 2017), teachers' computer experience (Sarfo et al., 2017) and school support (Hatlevik and Hatlevik, 2018). The examination of gender differences in teachers' ICT self-efficacy is an important research topic for many reasons. According to available statistics, women across the world predominate in teaching, especially in primary education. According to OECD data (OECD.Stat, 2018), female teachers in Finland account for 79.8% of a total number of teachers in primary education and 67.7% in secondary education. In United Kingdom, the respective percentages are 86.2% for primary education and 61.2% for secondary education, while in Singapore there are 80.7% female teachers in primary education and 63.7% in secondary education (Ministry of Education of Singapore, 2020). Similarly, female teachers are also prevalent in the Croatian primary and secondary education. Their share in primary education is 86.0% and 67.1% in secondary education (Croatian Bureau of Statistics, 2020). Consequently, a successful integration of ICT in education significantly depends on female teachers and their self-efficacy in using ICT for teaching purposes. This feminisation of education emphasises the need to explore female teachers' self-efficacy and gender differences in teachers' ICT self-efficacy. Researchers also indicate teachers' age as one of key factors influencing their teaching strategies and use of new technologies. Klassen and Chiu (2010), as well as Hoy and Spero (2005), found that teachers` self-efficacy increases from their early career to mid-career and declines afterwards. In their meta-analysis, Cai et al. (2017) indicate that older female teachers tend to have a lower level of ICT self-efficacy than male teachers. The mentioned studies highlight the importance of researching age and gender-age interaction effect on teachers' ICT self-efficacy. Even more so if we consider the age structure of female teachers who, as we previously stated, account for the majority of the teacher population. The statistics indicate a relatively considerable share of female teachers over 40 years of age. In 2 Finland, 64% of primary school teachers are 40 years of age or older; in the United Kingdom, 52% (OECD.Stat, 2018) and in Singapore 53% (Ministry of Education of Singapore, 2020). In Croatia, as many as 82% of primary and secondary school teachers are older than 30 years (Markočić Dekanić et al., 2019). In summary, the mentioned trends of feminisation of education and the large share of teachers older than 30 and 40 require a deeper insight into teachers' ICT self-efficacy and factors influencing gender and age effects on their ICT self-efficacy. Addressing these issues in the Croatian context, this research aims to contribute to a more comprehensive picture of gender differences in teachers' ICT self-efficacy, which is still an underresearched topic that lacks consistency in research findings. In the Croatian context, the study also has practical importance. It provides empirically grounded insights which are useful for developing policy measures for enhancing ICT self-efficacy of teachers, which is an important prerequisite for the effective integration of ICT into education. The study is structured as follows: 1) review of relevant studies in the field of gender differences in ICT self-efficacy and the theoretical background of this study; 2) detailed elaboration of research aims and hypotheses; 3) description of research methodology, including a description of the research sample, procedure and instruments; 4) description of the conducted statistical analyses, namely hierarchical multiple regression analysis conducted on the whole sample of teachers and regression analyses performed for males and females separately; analyses providing descriptive statistics; 5) elaboration of main results, which includes presentation of descriptive statistics of the regressors and their relationship with gender and teachers' ICT self-efficacy, as well as description of the results of multiple regression analyses; 6) discussion of the results in the context of the research aims, hypotheses and other studies presented in the research overview; 7) the text ends with the main conclusions drawn from the research results, followed by indicating the limitations of research and its implications for further research and practical implications related to improving teachers' ICT self-efficacy in the Croatian context. Research overview and theoretical background As the literature overview shows, many studies exploring teachers' self-efficacy have used Bandura's concept of self-efficacy (Poulou, Reddy and Dudek, 2019;Hatlevik and Hatlevik, 2018;Krause et al., 2017;Scherer and Siddiq, 2015;Klassen and Chiu, 2010). According to Bandura, self-efficacy "refers to a belief in one's capabilities to organise and execute the courses of action required to produce given attainments" (Bandura, 1997, p. 3). Bandura (1997) also indicates that self-efficacy is a domain-specific construct. Self-efficacy researchers in education (Perera, Calkins and Part, 2019;Hatlevik and Hatlevik, 2018;Klassen and Chiu, 2010;Krause et al., 2017;Scherer and Siddiq, 2015;Tarhini et al., 2014), who follow Bandura's self-efficacy conceptualisation, agree that teachers' self-efficacy should reflect their confidence in performing specific tasks in their teaching practice by integrating ICT. It should be noted that, according to Bandura (1997), self-efficacy is a subjective estimation of one's own capabilities which is nonetheless affected by other variables and prior experience of individuals. Based on Bandura's theory, the concept of self-efficacy is often used in studies exploring gender differences in teachers' use of ICT for teaching purposes, providing a theoretical basis for conceptualising ICT teachers` self-efficacy in empirical studies, including this 3 study. Scherer and Siddiq (2015) conducted a study on teachers' computer self-efficacy using the data obtained from the Norwegian secondary school teachers who participated in the 2013 International Computer and Information Literacy Study. Teachers' computer selfefficacy was conceptualised as their confidence in performing basic and advanced skills in using computers and the use of computers for instructional purposes (Scherer and Siddiq, 2015, p. 48). The results revealed that male teachers, compared to female teachers, showed higher levels of computer self-efficacy in basic and advanced operational skills. However, there was no statistically significant gender difference in the third dimension of computer self-efficacyusing computers for instructional purposes. As the authors noted, the use of computers for teaching purposes may include aspects of pedagogical competence which makes this construct different from the constructs related to the competence of operational use of computers. This finding points to the relevance of exploring differences in teaching practices and pedagogical knowledge and their effects on gender differences in ICT selfefficacy. In Siddiq and Scherer's (2016) research on the relationship between the teachers' emphasis on the development of students' ICT skills and computer self-efficacy, gender and age were investigated as moderators. The findings show that, contrary to age, the moderation effect of gender was limited, relating only to the use of computers for instructional purposes. The interaction between gender and age did not show a statistically significant moderation effect. The research points to the important role of age in teachers' self-efficacy, particularly in older teachers. According to the authors, further research should be carried out to obtain a more comprehensive and clearer picture of moderation effects of age and gender on the relation between ICT-related constructs. The effect of age on gender differences in ICT self-efficacy of teachers was rarely a subject of studies. Studies often perceive age differences as a generational issue where older teachers are less exposed to the ICT experience, have lower ICT self-efficacy and generally a less favourable attitude towards ICT use (Guo et al., 2008;Prensky, 2001;Salajan et al., 2010). Sarfo et al. (2017) investigated gender differences in teachers' computer self-efficacy in relation to their age and computer experience on a sample of senior high school teachers in Ghana. The data show that there was no gender difference in the overall computer selfefficacy. However, when self-efficacy was itemised into specific skills, it appeared that male teachers had a higher level of self-efficacy in web-based skills compared to female teachers, while gender differences were not found in basic computer skills and media-related skills. Such findings suggest that gender differences in ICT self-efficacy are associated with the use of different computer programmes whereby men are more likely to use advanced and more complex computer programmes than women. Regarding the effect of age, the findings also point out that its effect on ICT self-efficacy varies according to the type of computer applications. More precisely, there was no statistically significant effect of age for basic and web-based applications, only for media-related skills in favour of the younger teachers aged 20-30 years. Interestingly, the interaction effect of age and gender on ICT self-efficacy was not identified, which supports the previously mentioned findings by Siddiq and Scherer (2016). Some researchers associate the impact of age on ICT self-efficacy with emotional attitude towards ICT use, pointing to a positive relation between age and computer anxiety (Chaffin and Harlow, 2005;Czaja et al., 2006;Turner et al., 2007). Concerning the studies on the relationship of ICT self-efficacy and computer anxiety of teachers, the studies provide 4 different findings. Sarfo et al. (2017, p. 21), referring to the research findings by Compeau and Higgins (1995), state that the teachers with a lower level of self-efficacy are more anxious about working with computers than the teachers with a higher level of computer self-efficacy. The relationship between anxiety and ICT self-efficacy was also explored from the gender perspective. Papanastasiou and Angeli (2008) found that male Greek Cypriot teachers had a higher level of computer anxiety than their female colleagues. Contrary to this, in their meta-analyses of 50 papers on gender differences in attitudes towards ICT use published from 1997 to 2017, Cai et al. (2017) report a tendency of women to show greater anxiety than men. However, they point out that, despite maintaining this tendency, the gender gap is narrowing. From the point of view of the present research, these studies are instructive because they conceptualise anxiety as a component of computer attitude or, more precisely, as an emotional component of computer attitudes, what was also covered in this study. Based on Bandura's (1994Bandura's ( , 1997) statement on the sources of self-efficacy, the relationship between the teachers' computer experience and contextual factors and their ICT selfefficacy has also been explored (Hatlevik and Hatlevik, 2018;Papanastasiou and Angely, 2008;Sarfo et al., 2017). Sarfo et al. (2017, p. 25) found that there was an interaction effect between male and female teachers with a low level of computer experience and male and female teachers with a high level of computer experience on their computer self-efficacy slightly in favour of male teachers with a high level of computer experience. Hatlevik and Hatlevik (2018) confirmed a positive relationship between teachers' ICT self-efficacy and their ICT use in the teaching practice. Said authors, like Papanastasiou and Angeli (2008), also found a positive relationship between teachers' ICT self-efficacy and their work context, i.e., collegial and school management support. This points to the relevance of school's support in exploring teachers' ICT self-efficacy with respect to their age and gender differences. In summary, the review of the studies shows that a considerable attention has been attributed to the research on teachers' ICT issues in general, but not to gender differences in ICT selfefficacy, especially with respect to age, as noted by Siddiq and Scherer (2016) and Sarfo et al. (2017). The cited studies confirm that individual effects of age and gender, as well as their interaction effect on gender differences in teachers` self-efficacy, exist in different educational contexts and societies. The scarcity of empirical studies and differences in their findings point to the need for a further examination of the role of gender and age in teachers' ICT self-efficacy and variables that influence their effects on gender differences in different educational and societal contexts. This study will contribute to the field by providing empirical insights into the role of gender and age in teachers' self-efficacy in using ICT for instructional purposes in primary and secondary education in the Croatian context. Aims and hypotheses The Croatian educational system consists of compulsory elementary schools (primary and lower secondary schools), upper secondary education (gymnasiums and vocational schools) and tertiary education (polytechnics and universities). After finishing compulsory elementary school at the age of 14 or 15, the students in Croatia enrol into upper secondary schools. They can choose between academically oriented gymnasiums that prepare students for higher education and vocational schools that are more oriented on preparing the students for the labour market. This research focusses on the elementary and upper secondary education. 5 The digitalisation of elementary and secondary schools in Croatia was intensified by the Government Strategy of Education, Science and Technology (Government of the Republic of Croatia, Expert Committee, 2014) and the national project e-Schools within which this research was conducted. The process of school digitalisation raised the questions of teachers' preparedness to integrate ICT into their teaching practice, including ICT gender differences. Gender differences are even more important since in the school year 2018/2019 the share of female teachers in Croatia was 86% in elementary schools and 67.1% in secondary schools (Croatian Bureau of Statistics, 2020). Additionally, the age composition of the teacher population, which includes all subject teachers in elementary and upper secondary schools, indicates that over the last 5 years the number of teachers under the age of 30 has decreased by 5%, while the number of teachers aged 30 to 50 has decreased by 11%. As a result, the current teacher population has only 8% of teachers under the age of 30, 67% of teachers between the ages of 30 and 49, and 24% of teachers over the age of 50 (Markočić Dekanić et al., 2019). The stated gender composition and changes in the age structure of the teacher population raise the questions of how the teachers` gender and age affect the integration of ICT into their teaching practice and of the extent of teachers` ICT self-efficacy. The review of the literature indicated that in this context, gender differences in ICT self-efficacy are one of the vital but unexplored problems. This is especially true for Croatia, where, to the best of the knowledge of the authors of this study, gender differences in teachers' ICT self-efficacy have not been the subject of research so far but have been marginally addressed in only a few studies. This study has three aims: 1) To examine the possible moderating effect of age on gender differences in teachers' self-efficacy for using ICT in teaching Following the example of other studies that have used Bandura's conceptualisation of selfefficacy (Hatlevik and Hatlevik, 2018;Sarfo et al., 2017), we understand teachers' selfefficacy as their confidence in performing specific tasks that require the integration of ICT into the teaching practice. Consistently with the findings from the earlier studies that men are more self-efficient for using ICT than women (Cai et al., 2017;Whitley, 1997), we expect that male teachers will report higher levels of self-efficacy in comparison with their female counterparts (H1). We further expect higher levels of self-efficacy for using ICT among younger teachers, who had better opportunities to embrace digital tools at an earlier age, than among older teachers (H2). Finally, we expect the gender-age interaction effect (i.e., the moderating effect) on teachers' self-efficacy for using ICT (H3). More specifically, we anticipate smaller gender differences in self-efficacy among younger teachers who were obliged to take mandatory courses in ICT during their upper secondary education and/or to use ICT in their studies regardless of their gender. 2) To identify other variables that explain gender-age interaction effect on teachers' selfefficacy in using ICT in teaching Furthermore, the aim of this study is to inspect the possible reasons for the aforementioned interaction effect and a relatively low level of self-efficacy for using ICT among older female teachers. In doing so, we will control for the effects of the following variables: the type of school where the teachers work (elementary vs. upper secondary school: i.e., gymnasium or vocational school), teachers' perceived level of technical and professional support in their school, frequency of the teachers' use of simple and complex computer 6 programmes, length of computer use in teaching, and teachers' attitudes towards computers. In general, we expect that upper secondary school teachers will be more self-efficient regarding the use of ICT than their elementary school counterparts, as a result of a more complex usage of technology in teaching older cohorts of students (H4). Furthermore, we expect higher levels of self-efficacy among the teachers who experience higher levels of technical and professional support in their schools (H5) (Hatlevik and Hatlevik, 2018), who use computer programmes in teaching more frequently (H6) (Hatlevik and Hatlevik, 2018), who have been using a computer in teaching for a longer period of time (H7) (Sarfo et al., 2017), as well as among the teachers who have more positive attitudes towards computers (H8) (Cai et al., 2017). By adding these variables successively into the statistical model, we aim to determine how they affect the main effects of gender and age on self-efficacy for using ICT in teaching, as well as the gender-age interaction effect. 3) To identify the strongest regressor of teachers' self-efficacy in using ICT in teaching In addition, we aim to explore the effect sizes of individual regressors of teachers' selfefficacy for using ICT to find the strongest regressor. Although this aim is exploratory, we expect that at least some of the regressors will yield stronger effects in comparison with both gender and age (H9). This expectation is in line with the previous findings that gender differences in ICT self-efficacy show small effect sizes (Cai et al., 2017;Whitley, 1997), while statistical significance of age effects on ICT self-efficacy depends on how ICT is operationalised (Sarfo et al., 2017). Therefore, other variables could probably provide a better explanation of variations between teachers in ICT self-efficacy. For example, Gudek (2019) found a substantial relationship between computer self-efficacy of teacher candidates and their attitudes towards digital technology, while Krause et al. (2017) concluded that teachers' attitudes towards using ICT are one of the essential predictors of successful integration of ICT into teaching. Sample and procedure This study was based on a broader research project "The use of ICT in learning, teaching and assessment in Croatian elementary and secondary schools" that was conducted by the Croatian Academic and Research Network -CARNET and the Institute for Social Research in Zagreb. The data for this study were collected in September 2018 using the online questionnaire that was distributed to the principals of all elementary and upper secondary schools in Croatia. The school principals were asked to forward the questionnaire link to all teachers employed in their respective schools. The questionnaire was created using the LimeSurvey platform (LimeSurvey GmbH, 2018). It consisted of multiple-choice and Likert-type items. The participants were not able to skip items and they had to answer each item in order to move onto the next item. The completion of the questionnaire was anonymous and voluntary. About 13% of all elementary school teachers (N = 4395; 85.2% females) and 8% of all upper secondary school teachers (N = 2218; 69.6% females) in Croatia completed the questionnaire (total N = 6613; 80% females). Among the elementary school teachers from the sample, the dominant age categories were 31-40 years (36.1%) and 41-50 years (30.7%). Upper secondary school teachers were dominantly 31-40 years old 7 (30.7%) or 41-50 years old (27.1%). The distribution of teachers in the sample with respect to gender and type of school differed negligibly from the distribution in the population (χ² = 467.10; p <.01; Cramér's V = .084). Instruments Among others, the questionnaire contained different constructs that were in line with the aims and hypotheses of this research. We used the scale of teachers' self-efficacy for using ICT in teaching and the scale of computer attitudes, both devised by Papanastasiou and Angeli (2008). Rogošić (2015) translated these scales and used them in her research on the ICT use among the Croatian elementary school teachers in Zagreb (the capital of Croatia). Furthermore, the frequency of use of computer programmes was measured using the list of computer programmes by Papanastasiou and Angeli (2008) that was expanded for the purpose of this study. The questionnaire also contained the original scale of perceived technical and professional support for using ICT in school, items on demographics, as well as one item about the length of computer use in teaching. Teachers' self-efficacy for using ICT in teaching. This construct is operationalised as the teachers' self-confidence in using computer programmes/applications of different complexity in their work and in teaching the students to apply ICT in their learning. The scale consisted of seven Likert-type items (e.g., "I feel confident that I can use email to communicate with my students"; "I can teach my students how to make their own web pages") on a five-point continuum ranging from "completely disagree" to "completely agree" (α = .88). The average result on all seven items was used as an outcome in the analyses. The principal axis factoring supported the unidimensionality of the scale. Computer attitudes. This construct was operationalised as the teachers' cognitive, affective and behavioural evaluation of computers in the context of teaching and learning. The scale consisted of six Likert-type items (e.g., "The computer helps students understand concepts in a more effective way"; "The use of computers in teaching and learning stresses me out"; "If something goes wrong, I will not know how to fix it") on a five-point continuum ranging from "completely disagree" to "completely agree" (α = .91). The average result on all six items was used as a regressor. The scale was recoded so that higher scores imply a more positive attitudes towards computers. The principal axis factoring supported the unidimensionality of the scale. Frequency of use of computer programmes. Teachers were asked to estimate how often they used different computer programmes in teaching on a five-point continuum from "never" to "every day". The list contained 11 computer programmes. Two factors were extracted: the frequency of use of simple computer programmes (text editors, presentation programmes, websites/web portals, specialised educational websites; α = .65) and the frequency of use of complex computer programmes (multimedia tools, webpage editors, concept map tools, database software, publishing software, modelling software, computer simulation software; α = .82). The average results on two sets of items were used as regressors. Perceived technical and professional support for using ICT in school. This construct is operationalised as the teachers' perceived level of school's ICT equipment adequacy, of its availability to both the teachers and their students, as well as the teachers' opportunity to seek help regarding ICT use. A new scale was constructed for this research project that 8 consisted of seven Likert-type items ("I have internet access at school", "Internet access is available to my students at school", "At school, I have the necessary ICT equipment for teaching", "At school, my students have the necessary ICT equipment for classes in my subject", "At school, I get the necessary professional help and support for ICT use in teaching", "At school, I have the support of other teachers to use ICT in teaching", "The school enables the necessary professional development for the application of ICT in teaching") on a five-point continuum ranging from "completely disagree" to "completely agree" (α = .86). A pilot study was conducted on a small sample of elementary and upper secondary school teachers to check the comprehensibility of the items. The average result on all seven items was used as a regressor. The principal axis factoring supported the unidimensionality of the scale. Items on demographics. Teachers were asked to specify their gender (female or male), age (30 or less, 31-40, 41-50, 51-60, 61 or more) and the type of school where they work (elementary school or upper secondary school; if they chose upper secondary school, they were asked to further specify whether it was a gymnasium or a vocational school). Gender and type of school were used as categorical regressors and age was treated as interval regressor in the analysis. Length of computer use in teaching (in years). Teachers were asked to indicate how many years they had been using computer in teaching. Possible answers were: 5 years or less, 6-10 years, 11-20 years, 21-30 years, and 31 years or more. This variable was used as an interval regressor in the analyses. Statistical analysis The hierarchical multiple regression analysis was conducted in order to address the study aims. Teachers' self-efficacy in integrating ICT into teaching was used as the outcome variable. The following variables were introduced as regressors in the successive steps of the model -the first step: teachers' gender, age, and gender-age interaction; the second step: type of school (elementary, gymnasium or vocational school) and perceived technical and professional support for using ICT in school; the third step: frequency of use of simple and complex computer programmes; the fourth step: length of computer use in teaching and computer attitudes. The first step of the model related to the first study aim and tested the main effects of gender and age as well as gender-age interaction effect, i.e., the moderating effect of age on gender differences in teachers' self-efficacy for using ICT. Steps 2 to 4 related to the second study aimto identify the possible reasons of gender-age interaction effect on teachers' self-efficacy for using ICT. In other words, we wanted to check if the interaction effect remains statistically significant after the introduction of different regressors. Furthermore, steps 2 to 4 related to the third aim as wellto identify the strongest regressor of teachers' self-efficacy for using ICT. In addition, hierarchical multiple regression models were tested for females and males separately in order to check if the regressor variables (age, type of school, perceived technical and professional support for using ICT, frequency of use of computer programmes, length of computer use and computer attitudes) display similar effects across genders. Since the number of uncompleted questionnaires was negligible, we conducted a complete 9 case analysis. All variance inflation factor values (VIF) were smaller than two, meaning that there were no signs of multicollinearity. Descriptive statistics Descriptive statistics of the regressors and the outcome are presented in Table 1. A large majority of participants in this sample are females. Almost two thirds of teachers are in their thirties or forties. About two thirds of teachers work in elementary schools, while the rest of them work in upper secondary schools (gymnasium and vocational schools). On average, teachers experience high levels of technical and professional support for using ICT in their schools. Likewise, they report a frequent use of simple computer programmes. On the other hand, teachers on average rarely use complex computer programmes. More than 75% of teachers have been using computer in teaching for 10 years or less. On average, they have positive attitudes towards computers and high levels of self-efficacy for using ICT in teaching. There are more female teachers in elementary schools, while males are more prevalent in upper secondary vocational schools. On average, males give higher estimates of technical and professional support for using ICT than females. Furthermore, they more frequently use complex computer programmes and use computer in teaching for a somewhat longer time than females. Gender difference in the frequency of using simple computer programmes is negligible. Males have more positive attitudes towards computers and higher self-efficacy for using ICT than females. However, it should be noted that the aforementioned gender differences are small and that both males and females have relatively high levels of selfefficacy for using ICT. On a bivariate level, all regressors are associated with the outcome in the expected direction. Hierarchical multiple regression analysis In accordance with the hypotheses related to the first aim, the hierarchical multiple regression analysis shows that males and younger teachers report higher levels of selfefficacy for using ICT in teaching in comparison with females and older teachers, respectively (Table 2, column "Whole sample"; H1 and H2 confirmed). The effects of gender and age are small but statistically significant in all steps of the model, i.e., after the introduction of all regressor variables. As expected, gender differences in self-efficacy for using ICT are greater among older teachers and practically non-existent among the youngest ones (Table 2; Figure 1; H3 confirmed). In the context of the second aim, it can be noted that gender-age moderation effect on teachers' self-efficacy for the ICT use holds for both elementary and upper secondary school teachers (Figure 1). The gender-age interaction effect is relatively robust ( Table 2, column "Whole sample"), in a way that it remains statistically significant even after the introduction of the type of school and the estimates of technical and professional support in the regression model (step 2), as well as after the introduction of the frequency of use of simple and complex computer programmes (step 3). .12 Note: p<.01, * p<.05; a females = 0, males = 1. 11 The interaction effect ceases to be statistically significant after the introduction of the length of computer use in teaching (in years) and the computer attitudes in the model (step 4). A more detailed analysis with the separate introduction of these two variables in the model (not reported here) shows that both the length of computer use and the computer attitudes cancel the effect of gender-age interaction. Apparently, older female teachers on average have more negative attitudes towards computers in comparison to other teachers. Moreover, they use computers in teaching for a shorter period of time than other teachers, which also affects their self-efficacy for using ICT. It should be noted that even older female teachers report average levels of self-efficacy for using ICT (value 3 -"I neither agree nor disagree"; Figure 1). In other words, their scores are low only relatively but not absolutely. In addition, teachers who work in upper secondary vocational schools have a higher level of self-efficacy for using ICT than their colleagues from elementary schools (H4 partially confirmed). However, no difference in self-efficacy was found between gymnasium teachers and elementary school teachers. Teachers provided higher estimates of self-efficacy for using ICT if they received higher levels of technical and professional support in their schools (H5 confirmed), if they use computer programmes more often (H6 confirmed), if they use computers in teaching for a longer time (H7 confirmed) and if they have more positive attitudes towards computers (H8 confirmed). Concerning the third aim, positive computer attitudes were the single strongest regressor of the teachers' self-efficacy for using ICT, which is consistent with the results of bivariate analyses (Table 1). Furthermore, the model with only gender, age and gender-age interaction ( Table 2, column "Whole sample", step 1) explains only 4.3% of variance in self-efficacy for using ICT. In comparison, the full model (step 4) explains nearly 40% of variance of the outcome, which suggest that other variables are more important than gender and age for explaining self-efficacy for using ICT (H9 confirmed). Additional hierarchical regression models (Table 2, columns "Females" and "Males") show that individual regressors have similar effects on self-efficacy for using ICT for teachers of both genders, i.e., both directions and values of regression coefficients are rather similar for females and males. The only exception from this pattern is the absence of age effect in male teachers' data in step 1, which indicates a greater importance of age for self-efficacy for using ICT among females. In step 4, both models explain a similar percentage of variance (40.4% and 32.6% for females and males, respectively). Table 2. Hierarchical multiple regression models of teachers' self-efficacy for using ICT (standardised coefficients β). Discussion As we have already mentioned, this study has three aims: the first aim is to examine the moderating effect of age on gender differences in teachers' self-efficacy for using ICT in teaching, the second aim is to identify variables that explain gender-age interaction effect on teachers' self-efficacy in using ICT in their teaching, while the third aim is to identify the strongest regressor of teachers' self-efficacy in using ICT in their teaching. The findings relating to the first aim reveal that the effects of gender and age on teachers` ICT self-efficacy are statistically significant but minor. As hypothesised, they indicate minor differences in self-efficacy for using ICT between male and female teachers in favour of male teachers (H1 confirmed). This finding is consistent with the results of metaanalyses of individual empirical studies conducted over the last several decades (Cai et al., 2017;Whitley, 1997). According to them, despite the reduction of gender differences, men still have higher ICT self-efficacy than women. The research confirms that this is true for gender differences in the Croatian educational context where male teachers still have greater self-efficacy than female teachers, although this difference is small. The effect of age indicates that younger teachers are more self-efficient in using ICT than the older ones (H2 confirmed). It should also be indicated that gender differences in teachers' self-efficacy for using ICT in teaching are smaller among younger teachers than among older teachers, indicating the relevant role of age in gender differences in teachers' ICT self-efficacy (H3 confirmed). More precisely, gender differences are practically nonexistent among the youngest teachers (Figure 1). Similarly, studies carried out by Sarfo et al. (2017) and Siddiq and Scherer (2016) also found effects of gender and age on teachers' computer or ICT self-efficacy. Their research indicates that the effects of gender and age on teachers' ICT self-efficacy vary depending on the teachers` computer operational skills (Siddiq and Scherer, 2016) and the type of computer programme they use (Sarfo et al. 2017). This is suggestive for further research on gender differences in teachers' ICT selfefficacy in the Croatian context, pointing to the need for a more thorough research into the relationship between ICT self-efficacy and constructs such as the complexity of computer programmes and the diversity of teachers' computer skill for ICT integration into teaching. 14 Concerning the second study aim, which is focused on the explanation of gender-age interaction effect on teachers' self-efficacy in using ICT, the following findings are relevant. First and foremost, the finding that all regressor variables used for explaining the gender-age interaction effect are associated with teachers' ICT self-efficacy in the expected direction on the bivariate level and in the regression analysis (Table 1 and Table 2; H4 partially confirmed; H5, H6, H7 and H8 confirmed). The only exception is that the teachers from gymnasiums did not differ from elementary school teachers in the average levels of self-efficacy for using ICT, as we anticipated. The hypothesis was that the upper secondary school teachers, i.e., vocational school teachers and gymnasium teachers, would have a higher level of ICT self-efficacy than their elementary school counterparts, as a result of a more complex use of ICT in teaching older cohorts of students. The reason why vocational school teachers, unlike gymnasium teachers, have a higher level of ICT self-efficacy and differ from elementary school teachers, could be found in their technologically-oriented professional profile and vocational school curricula that require the application of more complex ICT programmes. In contrast to this study focused on teachers' ICT self-efficacy, the studies on the association of teachers' self-efficacy for classroom management and student engagement with grade levels found that teachers in higher grade levels had lower self-efficacy compared to teachers in lower grade levels (Klassen and Chiu, 2010;Wolters and Daugherty, 2007). This difference suggests that the relationship between grade levels and teachers' self-efficacy varies depending, among other factors, on the type of teachers' self-efficacy. Regarding the gender-age interaction effect on teachers` ICT self-efficacy, the findings of this study show that it is relatively robust. They indicate that the gender-age interaction effect ceased to be statistically significant only in the last step of the regression analysis, after the introduction of the length of computer use in teaching and the computer attitudes into the model. This suggests that these two variables play a role in the low level of ICT self-efficacy among older teachers, in particular among older female teachers. The importance of the role of age in ICT self-efficacy of female teachers was confirmed by the findings of separate hierarchical regression analyses conducted on a subsample of male and female teachers indicating the absence of age effect in male teachers' data in step 1 (Table 2). In this context, it should be noted that older female teachers used computer in teaching for a shorter period of time than their male counterparts and that they tend to have a more negative attitude towards computer than the older male teachers. By reviewing these findings in the context of other studies, it should be noted that the finding on the effect of gender-age interaction on teachers' ICT self-efficacy differs from the findings by Sarfo et al. (2017) and Siddiq and Scherer (2016). They did not identify the effect of gender-age interaction on teachers' self-efficacy in the use of ICT, hence their studies do not support the findings of this study. This inconsistency in the findings may stem from the differences in the operationalisation of major concepts (e.g., self-efficacy in using different computer programmes) and differences in research samples (primary school teachers, lower secondary school teachers and senior high school teachers). As the literature overview indicates, the gender-age interaction effect on the gender differences in teachers' ICT selfefficacy is an underresearched and undertheorised research problem. In the context of scarce research, the mentioned variability and inconsistency in the findings confirms the need for further research of the moderating effect of age on gender differences in teachers' selfefficacy for using ICT in teaching. However, the findings of this study regarding the significant effects of previous experience with the use of ICT on gender differences in teachers' ICT self-efficacy are consistent with the findings of other studies. For example, Hatlevik and Hatlevik (2018) as well as Papanastasiou and Angeli (2008) also found a positive relationship between teachers' selfefficacy and the length of ICT use in their teaching practice, while Sarfo et al. (2017) identified the interaction effect of gender and computer experience of teachers on their computer self-efficacy in favour of males with a higher level of computer experience. As previously mentioned, in this study, computer attitudes are an important variable for explaining the gender-age effect on teachers' ICT self-efficacy, especially in terms of older female teachers who tend to have more negative attitudes towards computers compared to the older male teachers. Related to the third aim, it shows that computer attitudes, as the single strongest regressor of ICT self-efficacy (H9 confirmed), explain gender differences in self-efficacy for using ICT in teaching more than the other variables (type of school, technical and professional support in school, frequency of use of simple and complex computer programmes and number of years of computer use in teaching). This is not surprising if we consider that the teachers' attitudes are one of the essential prerequisites for the successful integration of ICT into teaching and into the classroom in general (Krause et al., 2017). Some studies suggest that more negative attitudes of female teachers towards computers could be accounted for by their higher level of computer anxiety, which is negatively associated with computer self-efficacy (Cai et al., 2017;Sarfo et al., 2017;Whitley, 1997). This points to the relevance of a more detailed research into the role of computer anxiety as a separate construct in gender differences in teachers` ICT self-efficacy in the future. Finally, the variance of teachers' ICT self-efficacy shows that computer attitudes and other variables (type of school, technical and professional support in school, frequency of use of simple and complex computer programmes and number of years of computer use in teaching) are more important than gender and age in explaining the differences in ICT selfefficacy, confirming thus the small effects of age and gender on teachers' ICT selfefficacy. Conclusions, limitations and implications of the research Overall, these findings show that gender differences in teachers' ICT self-efficacy exist, but they are minor. As a result, two findings are particularly important: 1) males and younger teachers report higher levels of self-efficacy for using ICT in teaching than females and older teachers, respectively; 2) gender differences in self-efficacy are greater among older teachers, while they are practically non-existent among the youngest ones. This moderation effect reveals an important role of age in gender differences in ICT selfefficacy addressing lower self-efficacy of older female teachers compared to other teachers. In this regard, it is important to emphasise that both male and female teachers have a relatively high level of self-efficacy in the use of ICT in their teaching. The findings also suggest that ICT attitudes and previous experience of teachers in the use of ICT play an important role in their ICT self-efficacy, especially among older female teachers who have more negative ICT attitudes and who have been using ICT in teaching for a shorter period of time than their male colleagues. As noted by other researchers (Siddiq and Scherer, 2016;Sarfo et al., 2017), gender 16 differences in teachers' ICT self-efficacy have previously been researched but very rarely, especially in terms of the role of age in ICT self-efficacy. Scarcity and inconsistencies in research findings call for further research into gender differences in teachers' ICT selfefficacy in different societal contexts and educational settings. Even more so as this is an important topic in many countries, which has been shown in the overview of previous research. This research contributes to the existing studies by highlighting the role of age of primary and secondary school teachers in Croatia in gender differences in ICT selfefficacy. In this respect, the finding which points to the greater importance of age for ICT self-efficacy among females as compared to males is quite significant. However, the study has some limitations. This research design was transversal, which allowed us to identify gender differences in teachers' ICT self-efficacy. A longitudinal study is required to test whether these differences increase with age within the same sample of teachers. Furthermore, taking into account the impact of school subjects taught by teachers would contribute to a more comprehensive and deeper understanding of gender differences in ICT self-efficacy. Although the sample size was relatively large, due to a large number of school subjects in Croatian elementary and secondary education, the dataset had a too small number of participants per subject for a more detailed analysis. The research was conducted before the outbreak of the COVID-19 pandemic. Its consequences and, in particular, the closure of schools and the transition to online teaching and learning will emphasise the role of ICT use in education and the importance of gender differences in teachers' ICT self-efficacy as a subject of research. Recent research indicate that COVID-19 pandemic measures in education are particularly difficult for women and female teachers due to their overload with school obligations and household chores (EIGE, 2020;UNESCO, 2020). The extent to which the everyday life and teaching experience during COVID-19 pandemic affects teachers' ICT self-efficacy is a topic that has yet to be explored more intensively. This research also has some practical implications. Pointing to the relatively lower ICT self-efficacy of older female teachers, the findings suggest that strategies to enhance ICT self-efficacy of teachers should be particularly targeted at them. The relevant roles of previous shorter computer experience and more negative computer attitudes of older female teachers in their ICT self-efficacy indicate the need for their empowerment through high-quality trainings in computer skills as well as school management and collegial support within the school institution. A similar level of self-efficacy for using ICT among the younger generations of teachers of both genders is an encouraging finding which could hopefully be followed by gender equality in other aspects of ICT use. Notes An early version of findings in this paper was presented at the conference ECER 2019, 2-6 September 2019, in Hamburg, Germany.	2021-09-01T15:03:03.882Z	2021-06-30T00:00:00.000	{ "year": 2021, "sha1": "28b4508b69064e80bfb38149109145556ee67fa6", "oa_license": "CCBY", "oa_url": "https://infedu.vu.lt/journal/INFEDU/article/699/file/pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "ddcf055eaa406e5fb3db66d6893ce03b02681546", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [ "Psychology" ] }
265512325	pes2o/s2orc	v3-fos-license	Study on the deterioration characteristics of aeolian sand concrete under the coupling effect of multiple factors in harsh environments This study takes the aeolian sand concrete as a research object, uses the relative dynamic elastic modulus to study its macro characteristics, and combines nuclear magnetic resonance、scanning electron microscope to study its pore characteristics and micro morphology under the action of prestress, freeze-thaw and salt intrusion. The results show that with the increase of the amount of aeolian sand, the dynamic elastic modulus of aeolian sand concrete shows a pattern of first decreasing, then increasing, and then decreasing; when no prestress is applied, the porosity of aeolian sand concrete first increases, then decreases, and then continues to increase. Among them, the porosity of aeolian sand concrete with a 40% content of aeolian sand decreases by 0.06% compared to that with a 0% content of aeolian sand, and decreases by 0.003% compared to that with a 60% content of aeolian sand; with the increase of prestress, the porosity of aeolian sand concrete with the same amount of aeolian sand increases gradually with the increase of damage degree. The porosity of concrete with 40% aeolian sand content increases by 0.33% when the damage degree is 0.0 compared to 0.3, with a 6.31% increase in the number of multi damage pores; under the coupling effect of multiple factors, when the amount of aeolian sand is 40%, the damage degree of the four groups of aeolian sand concrete before and after the coupling effect is 0.0, 0.1, 0.2, and 0.3, respectively, increases by 25.8%, 32.2%, 73.8%, and 85.8%, respectively; under the coupling effect of multiple factors, the content of aeolian sand is 60%, the damage degree is 0.2 and 0.3 groups, and the content of aeolian sand is 20%, the damage degree is 0.3 groups, which does not meet the standard requirements; under the coupling action of stress, freeze-thaw, salt intrusion and the amount of aeolian sand, the filling effect of aeolian sand on the internal pores of aeolian sand concrete decreases first, then increases, and then decreases with the increase of the amount of aeolian sand. The filling effect is further weakened after the action of stress. After the superposition of freeze-thaw and salt intrusion, the coupling effect of water and salt solution in frost heaving medium makes the variation law and range of physical and chemical characteristics of aeolian sand concrete show a great difference. Introduction The mechanics, frost resistance and other corrosion resistance characteristics of concrete under prestress are quite different from those under conventional service environment.Domestic and foreign scholars have also conducted a large number of experimental and theoretical studies on the deterioration characteristics and mechanism of concrete after stress damage, such as: Zhou Zhijun et al [1].Studied the carbonation resistance of concrete under the action of bending stress and nitrate erosion, and found that with the increase of nitric acid concentration and stress level, the carbonation depth increased, the content of carbonated substances decreased, and nitric acid consumed hydration products, weakened the filling of pores by CaCO 3 , formed microcracks, and increased the porosity, while bending stress promoted the development of microcracks to form tensile cracks, and formed microcracks in coordination with carbonation; Liu Yan et al [2].Studied the carbonation performance of recycled concrete under the action of axial stress, and pointed out that the axial compressive stress suppressed the carbonation damage of recycled concrete, and the axial tensile stress aggravated the carbonation damage of recycled concrete.Under the action of axial compressive stress carbonation, the internal structure of recycled concrete is relatively dense, and under the action of axial tensile stress carbonation, through cracks are formed in recycled concrete; Duan shaozhen [3] and others pointed out in their research on the mechanical properties of concrete-filled steel tubular short columns under axial compression under the action of initial construction stress that the simultaneous existence of initial construction stress of outer steel tube and section steel will delay the interaction time between concrete and steel tube and section steel, weaken the hooping effect of steel tube on concrete and the local constraint of concrete on section steel, make the composite members enter the elastoplastic stage in advance and expand the scope of this stage, the ultimate bearing capacity of composite columns decreases with the increase of initial stress; Tang Guanbao et al [4].Studied the influence of stress on concrete air permeability coefficient and CO2 diffusion coefficient, and established a prediction model of concrete carbonation depth under stress based on air permeability coefficient CDEP GPL model; Wang Jiabin et al [5].Studied the chloride ion diffusion performance in the tensile and compressive areas of shotcrete under the action of bending stress, and pointed out that the stress changes the pore structure distribution of shotcrete and the number and depth of microcracks, which has an impact on the chloride ion diffusion rate.On the side of the tensile area, the chloride ion content increases with the increase of the bending stress ratio; on the side of compression zone, the chloride ion content decreases with the increase of bending stress ratio; Zeng Yitao [6] and others studied the hydraulic fracturing problem when the dam heel has initial cracks due to construction stress based on the combination of different bending moments and water pressure values, and pointed out that a small load increment will break the steady state and promote the unstable expansion of cracks; there is a superposition effect between splitting water pressure and tensile stress.If the maximum action position is the same, the stress concentration at the crack tip is obvious, which is easy to cause hydraulic splitting damage.If the action positions are different, the strain distribution of the tensile section is more uniform, which greatly gives play to the splitting tensile resistance of the concrete tensile section and weakens the hydraulic splitting effect; Tran T.T et al [7].Studied the relationship between water permeability and chloride diffusion coefficient of concrete under compressive stress, proposed a fine-scale hydrodynamic lattice model to simulate the fluid flow and chloride entry in concrete under stress, used a softening damage model to describe the behavior of cement matrix and ITZ, and assumed that the aggregate was elastic, according to the experimental results of water permeability and chloride ion diffusion test, the hydraulic mechanical parameters of concrete members are calibrated, and the relationship between water permeability and diffusion coefficient is obtained; Liu Jin et al [8].Studied the multiscale analysis theory of concrete diffusion coefficient under mechanical stress, proposed a multi-scale analysis model, predicted the chloride ion diffusion rate of concrete under mechanical stress, and evaluated the quantitative relationship between the apparent diffusion coefficient of concrete and mechanical stress (here is volume strain), as well as the current porosity of cement paste and ITZ; the research on the characteristics of ultrasonic wave velocity of concrete under the action of Ling Zhang [9] and other stresses points out that under the same loading conditions, the ultrasonic velocity has an increasing trend with the increase of age for C10 and C20 concrete samples, but for C30, C35, C40 and C45 concrete samples, the overall growth trend of velocity is not obvious; Guo Jun liu et al [10].Studied the failure criterion of concrete under multiaxial stress, analyzed its limitations, and proposed the research direction of concrete strength failure criterion; Yu Chuan Jiang et al [11].Tested and evaluated the chloride permeability of concrete under compressive stress, pointing out that the electric flux of concrete under compressive stress is affected by the stress level.With the increase of stress level, the electric flux of concrete first decreases and then increases; the electric flux of concrete with different strength grades has different sensitivity to compressive stress.The lower the strength grade of concrete, the higher the sensitivity to compressive stress; the electric flux of concrete after compressive stress is related to crack recovery, and only the stress level of more than 60% has a significant impact on the electric flux of concrete; the load should be considered in the evaluation of chloride permeability of concrete structures; Jian Hui Yang et al [12].Studied the constitutive relationship of concrete under plane stress based on the generalized octahedral theory, and pointed out that the plane constitutive relationship of concrete can be understood through the plane problem of stress-strain space transformation.According to the symmetry of an orthogonal octahedral, one of the arbitrary inclined planes is parallel to one of the three rectangular coordinate axes. To sum up, scholars at home and abroad have studied the deterioration process of concrete under uniaxial, biaxial and bending stresses, and made clear its superimposed influence on the coupled erosion effects such as salt intrusion and carbonation, and pointed out that when stress is coupled with other erosion effects, concrete deterioration presents phased changes, and the promotion or suppression of concrete deterioration also depends on the superimposed effect, based on damage theory-Fick's law Octahedral theory established the deterioration model under the coupling effect of stress and erosion.However, the existing research often focuses on the traditional concrete materials, and the coupling factors are also common in the two factors coupling.The damage mechanism research is also dominated by the deterioration mechanism under the action of a single factor.There are still significant deficiencies in the research and development of green new building materials, as well as in the study of the deterioration mechanism of concrete materials under the coupling of multiple factors, which is in line with the harsh environment in the western region. In view of this, this study uses aeolian sand widely distributed in Western China as raw material to replace river sand to prepare a new green building material, namely aeolian sand concrete, and discusses its degradation process under multiple factors in the harsh environment of the western region (stress, freeze-thaw, and salt intrusion), so as to clarify its degradation characteristics and mechanism, which is not only conducive to the development of aeolian sand resources, the economic development of aeolian sand, and the research and development of new building materials, it is also conducive to protecting river sand resources, promoting social development and environmental protection, which is of great social, economic and practical significance. Introduction to raw materials The fine aggregate used in the study is from the sand field around Hohhot, with a fineness modulus of 2.91 and a particle size range of 0.075~4.75mm;aeolian sand is taken from the Kubuqi Desert in Inner Mongolia, China, with a fineness modulus of 0.72.The coarse aggregate used in the test is egg gravel, with an apparent density of 2669kg/m 3 , a bulk density of 1650kg/m 3 , and a particle size range of 4.75~20.0mm;the mixing water is ordinary tap water; the water reducing agent adopts polycarboxylic acid sc-40 high-efficiency water reducing agent of Inner Mongolia Rongshengda new materials Co., Ltd., with a water reducing rate of 26%; the air entraining agent is sj-3 high-efficiency air entraining agent, grade II fly ash of Inner Mongolia Jinqiao power plant and Jidong P•O42.5 cement.Some physical and chemical property indexes are shown in Tables 1-4 below. Mix proportion design of aeolian sand concrete In accordance with the relevant provisions of concrete mix proportion design in the code for hydraulic concrete construction (SL677-2014), code for mix proportion design of ordinary concrete JGJ55-2011) and ACI method of proportion, and in order to ensure that the aeolian sand concrete can obtain high slump and fluidity for engineering practice, equivalent replacement shall be carried out according to the aeolian sand content of 0%, 20%, 40%, 60%, 80% and 100%, the strength grade of the benchmark aeolian sand concrete is generally C25, and its mix is shown in Table 5. Introduction to coupling conditions According to the standard for test methods of long-term performance and durability of ordinary concrete > [13] (GBT50082-2009), aeolian sand concrete with different amounts of aeolian sand is selected to design the coupling conditions of stress, freeze-thaw and salt intrusion, as follows: ① Prestressing: select three groups of 100 mm×100 mm×400mm aeolian sand concrete prism specimens with different amounts of aeolian sand (specifically increase or decrease the number of groups according to the mechanical property test results), measure the initial relative dynamic elastic modulus, and define the damage degree at this time as 0.0.Apply axial loads respectively by universal testing machine.When the relative dynamic elastic modulus decreases by 5%, stop loading, and define the damage at this time as 0.1, and so on, the relative dynamic elastic modulus of aeolian sand concrete with damage degree of 0.2 and 0.3 is 90% and 85% of the initial value respectively. ② Apply freeze-thaw and salt intrusion: select concrete specimens of aeolian sand with different amounts of aeolian sand and different degrees of damage.After 24 days of standard curing, immerse them in (full immersion method) the freeze-thaw medium (5.0%NaCl solution) with a temperature of 15 ~20˚C.After 4 days, wipe the surface moisture of the soaked specimens, and further determine the relative dynamic elastic modulus at this time. Then, use the "quick freezing method((Frozen at -20~-10 for 2h, melted at 15-20 for 2h))" to carry out the freeze-thaw and salt intrusion coupling test of aeolian sand concrete. ③ Test and analysis: under the coupling effect of stress, freeze-thaw and salt intrusion, the relative dynamic elastic modulus of aeolian sand concrete samples with different amounts of aeolian sand is measured with 25 freeze-thaw cycles as a cycle.When the relative dynamic elastic modulus drops to 60% of the initial value, the test is terminated, and then the micro pore structure after prestressing and the coupling effect of stress, freeze-thaw and salt intrusion is measured by NMR, Sigma500 field emission scanning electron microscope was used to measure the micro morphological characteristics before and after the coupling condition. 3 Test results and analysis Mechanics and pore characteristics of aeolian sand concrete before and after stress action The compressive strength and splitting tensile strength of aeolian sand concrete are measured according to the standard for test methods of mechanical properties of ordinary concrete GB/ T50081-2002, as shown in Figs 1 and 2 below.According to Figs 1 and 2, when the content of aeolian sand is 0%, 20%, 40%, 60%, 80% and 100%, the 7d unconfined compressive strength and 28 d splitting strength first increase and then decrease with the increase of the content of aeolian sand, and the 28 d unconfined compressive strength gradually decreases with the increase of the content of aeolian sand.When the content of aeolian sand is 80%, the compressive strength is close to the standard value, and if the content of aeolian sand continues to increase, the compressive strength has been slightly lower than the standard requirements.In view of this, according to the analysis results of mechanical properties, the aeolian sand concrete with the content of 0%, 20%, 40% and 60% (groups I, II, III and IV) is selected for the subsequent stress, freeze-thaw and salt invasion coupling test, and its elastic modulus, porosity and other relevant physical and chemical indicators are measured, as shown in Fig 3 and Table 6 below. It can be seen from Fig 3 that with the increase of the amount of aeolian sand, the relative dynamic elastic modulus of aeolian sand concrete decreases first, then increases, and then decreases, but all meet the requirements of C25 concrete standard.According to the results of pore characteristics in Table 6, when prestress is not applied (the damage degree is 0.0 group), the porosity of aeolian sand concrete first increases, then decreases, and then continues to increase.The porosity of aeolian sand concrete with 40% of aeolian sand content is 0.06% lower than that with 0% of aeolian sand content, and 0.003% lower than that with 60% of aeolian sand content, and the pore changes are mainly concentrated in the harmless pores of 20-50nm [14], the content of aeolian sand with 40% is the highest, reaching 27.05%, and the number of harmful holes above 200nm is the least, only 44.79%.This is because the fineness modulus of aeolian sand is 0.72, which is far lower than 2.91 of ordinary sand.When the amount of aeolian sand is small, finer aeolian sand can fill the internal pores of aeolian sand concrete, reduce the porosity, increase the number of harmless pores, and reduce the number of harmful pores.However, with the increase of the amount of aeolian sand, the filling effect of aeolian sand decreases, and finer aeolian sand reduces the workability of aeolian sand concrete and increases the water consumption of standard consistency, at this time, the density of the aeolian sand concrete formed by mixing decreases, and the porosity increases, and most of them are harmful pores above 200nm.At the same time, it can be seen from Table 6 that when prestress is applied, with the increase of prestressing, the porosity of aeolian sand concrete with the same amount of aeolian sand presents a gradual increasing trend.When the damage degree is 0.0, the porosity of aeolian sand concrete with 40% of aeolian sand is 1.598%, while when the damage degree is 0.3, the porosity increases to 1.928%, with an increase of 0.33%, in which the number of harmful holes increases by 6.31%, and the irreducible fluid saturation decreases by 2.371%.At the same time, according to the indoor macro test, the compressive strength of aeolian sand concrete with 40% content of aeolian sand also decreases after prestress, of which the compressive strength is 25.2MPa when the damage degree is 0.3, which is 9.6% lower than the initial value of 27.9MPa.This is because under the action of prestress, the small harmless and less harmful holes in the aeolian sand concrete are destroyed with the increase of stress, and then transformed into harmful and more harmful holes.At this time, the NMR test results show that the irreducible fluid saturation also gradually decreases, which further proves that the proportion of small holes in the aeolian sand concrete decreases; the mechanical properties of aeolian sand concrete decrease with the increase of porosity. Test results and analysis of aeolian sand concrete under the coupling effect of stress, freeze-thaw, salt intrusion and aeolian sand content On the basis of the study on the mechanical properties of aeolian sand concrete and the pore characteristics after prestress, the concrete with damage degrees of 0.0, 0.1, 0.2 and 0.3 and the content of aeolian sand of 0%, 20%, 40% and 60% was first soaked in 5.0% sodium chloride solution for 4 days, and then put into the freeze-thaw test box with 5.0% sodium chloride solution as the freeze-thaw medium for stress, freeze-thaw and salt intrusion coupling tests, taking 25 freeze-thaw cycles as a cycle, measuring the relative dynamic elastic modulus of aeolian sand concrete.When the relative dynamic elastic modulus drops below 60% of the initial value or the freeze-thaw cycle is 200 times, the test is terminated, and samples are taken for NMR, 7. It can be seen from Figs 4 and 5 that the compactness of aeolian sand concrete under different damage degrees and the same amount of aeolian sand gradually decreases.After 200 times of stress, freeze-thaw and salt intrusion coupling, the relative dynamic elastic modulus of aeolian sand concrete with the same damage degree and different amounts of aeolian sand shows a downward trend as a whole, further indicating that the internal compactness of aeolian sand concrete under coupling conditions has decreased, the relative dynamic elastic modulus of aeolian sand concrete with 60% aeolian sand content decreased to 58% and 54% of the initial value respectively when the damage degree was 0.2 and 0.3, and the damage degree was 0.3.The relative dynamic elastic modulus of aeolian sand concrete with 20% aeolian sand content also decreased to 59% of the initial value, which was lower than the standard requirements, while other groups were higher than the standard value.At the same time, after the coupling working condition, the relative dynamic elastic modulus of aeolian sand concrete under the same damage degree shows a change law of first decreasing, then increasing, and then decreasing with the increase of the content of aeolian sand.For different damage degrees, with the increase of the damage degree, the relative dynamic elastic modulus after the coupling working condition decreases significantly, among which, when the content of aeolian sand is 40%, before and after the coupling, the relative dynamic elastic modulus of aeolian sand concrete with damage degree of 0.0, 0.1, 0.2 and 0.3 decreased by 23.8%, 24.4%, 26.0% and 38.8% respectively, and the relative dynamic elastic modulus of the damage degree of 0.0 group was 5.8%, 17.4% and 31.7% higher than that of 0.1, 0.2 and 0.3 groups respectively.It can be seen from Tables 6 and 7 that after 200 times of stress, freeze-thaw and salt invasion, the porosity of aeolian sand concrete increases and the irreducible fluid saturation decreases compared with the initial value.After the coupling effect, under the same damage degree, the porosity of aeolian sand concrete as a whole shows a trend of first increasing, then decreasing, and then increasing with the increase of aeolian sand content, and the main body of its pore size distribution shows that the proportion of harmless and less harmful pores decreases, and the proportion of harmful and more harmful pores increases, while the bound fluid saturation first decreases, then increases, and then decreases with the increase of aeolian sand content.Under different damage degrees, with the increase of damage degree, the porosity of aeolian sand concrete shows a significant increasing trend.When the content of aeolian sand is 40%, the concrete porosity of the four groups of aeolian sand with damage degrees of 0.0, 0.1, 0.2 and 0.3 before and after coupling increases by 25.8%, 32.2%, 73.8% and 85.8% respectively, while the irreducible fluid saturation decreases by 5.3%, 25.8%, 27.7% and 41.2% respectively; after the coupling effect, the porosity of 0.0 group is 11.8%, 49.8% and 77.8% lower than that of 0.1, 0.2 and 0.3 groups respectively, and the irreducible fluid saturation of 0.0 group is 20.3%, 22.4% and 40.8% higher than that of 0.1, 0.2 and 0.3 groups respectively.This is due to the unique physical and chemical characteristics of aeolian sand concrete different from ordinary concrete under the coupling action of stress, freeze-thaw, salt intrusion and the change of aeolian sand content.When aeolian sand replaces river sand, if the replacement proportion is less, aeolian sand, as a super fine sand, requires more water in the concrete slab and process, and because the amount of aeolian sand is less, its effect on filling the pores in the concrete is poor, resulting in an increase in the overall porosity.Therefore, when the amount of aeolian sand is 20%, the porosity of aeolian sand concrete increases, and the relative dynamic elastic modulus and irreducible fluid saturation decrease within the same damage degree; with the increase of the replacement proportion of aeolian sand, the filling effect of aeolian sand on the internal pores of concrete increases, so when the content of aeolian sand is 40%, the porosity of aeolian sand concrete in the same damage degree decreases, and the relative dynamic elastic modulus and irreducible fluid saturation increase; at the same time, when the amount of aeolian sand is large, the filling effect of aeolian sand on the pores in the concrete has reached saturation, and the excess aeolian sand needs more water during the preparation of concrete, resulting in the formation of aeolian sand particles including incomplete mixing in the concrete.Therefore, when the amount of aeolian sand is 60%, the internal porosity of aeolian sand concrete increases significantly, and the relative dynamic elastic modulus and bound fluid saturation decrease significantly. When the stress is superimposed, it is only the mechanical damage to the original concrete, which reduces the physical and chemical properties of aeolian sand concrete to a certain extent, but does not change the change rules of the physical and chemical properties such as the relative dynamic elastic modulus and porosity of aeolian sand concrete caused by the change of the content of aeolian sand; when the freeze-thaw and salt intrusion are superimposed, the change law and range of physical and chemical properties of aeolian sand concrete show great differences.This is because under the action of freeze-thaw and salt intrusion, the freeze-thaw medium gradually erodes into the interior of aeolian sand concrete, and the water in the freeze-thaw medium freezes and melts under the action of freeze-thaw, while the volume of water expands when it freezes, destroying the internal pores of aeolian sand concrete, make it transform from harmless and less harmful pores to more harmful and harmful pores with larger pores.At the same time, the salt in the freeze-thaw medium enters the interior of aeolian sand concrete with the pores generated by water frost heaving and melting under the action of capillary, infiltration and diffusion [15][16][17][18][19], and produces Friedel salt and other crystallization products, which accelerates the generation and destruction of interface cracks between cement paste and aggregate, further expands the pores generated by frost heaving, and starts again and again, the concrete compactness of aeolian sand decreases, the relative dynamic elastic modulus decreases, the porosity increases, and the irreducible fluid saturation increases. Conclusion 1.With the increase of the amount of aeolian sand, the relative dynamic elastic modulus of aeolian sand concrete decreases first, then increases, and then decreases.The porosity increases first, then decreases, and then continues to increase.The porosity of aeolian sand concrete with 40% of aeolian sand is 0.06% lower than that with 0% of aeolian sand, and 0.03% lower than that with 60% of aeolian sand, and the pore changes are mainly concentrated between harmless pores of 20-50nm, up to 27.05%. 2. With the increase of prestress, the porosity of aeolian sand concrete with the same amount of aeolian sand presents a gradual increase trend.When the damage degree of aeolian sand concrete with 40% of aeolian sand is 0.0, the porosity is 1.598%, while when the damage degree is 0.3, the porosity increases to 1.928%, with an increase of 0.33%, in which the number of more harmful pores increases by 6.31%, and the irreducible fluid saturation decreases by 2.371%.This indicates that pre-stressing causes the transformation of less harmful pores with smaller internal pore sizes into harmful pores with larger pore sizes, resulting in larger pores and a decrease in compactness in aeolian sand concrete. 3. Under the action of stress, freeze-thaw and salt intrusion, when the content of aeolian sand is 60%, the damage degree is 0.2 and 0.3, it decreases to 58%, 54% .And when the content of aeolian sand is 20%, the damage degree is 0.3, it decreases to 59% of the initial value, and the other groups meet the standard requirements; when the content of aeolian sand is 40%, the damage degree is 0.0, 0.1, 0.2 and 0.3.The relative dynamic elastic modulus of aeolian sand concrete decreases by 23.8%, 24.4%, 26.0% and 38.8% respectively, the porosity increases by 25.8%, 32.2%, 73.8% and 85.8% respectively, and the irreducible fluid saturation decreases by 5.3%, 25.8%, 27.7% and 41.2% respectively, meet the standard requirements. Fig 1 . Fig 1. Relationship curves of 7 d and 28 d of the aeolian sand concrete, the unconfined compressive strength and the volume of the aeolian sand.https://doi.org/10.1371/journal.pone.0289847.g001 Fig 5 . Fig 5. Micromorphology of aeolian sand concrete with different damage degrees after the coupling action of stress, freeze-thaw and salt intrusion.a.The damage degree after stress, freeze-thaw and salt intrusion is 0.0, and the content of aeolian sand is 40%.The results of electron microscope test of aeolian sand concrete.b.The damage degree after stress, freeze-thaw and salt intrusion is 0.1, and the content of aeolian sand is 40%.The results of electron microscope test of aeolian sand concrete.c.The damage degree after stress, freeze-thaw and salt intrusion is 0.2, and the content of aeolian sand is 40%.The results of electron microscope test of aeolian sand concrete.d.The damage degree after stress, freeze-thaw and salt intrusion is 0.4, and the content of aeolian sand is 40%.The results of electron microscope test of aeolian sand concrete.https://doi.org/10.1371/journal.pone.0289847.g005 Table 4 . The physical and dynamic parameters of experiment fly ash. https://doi.org/10.1371/journal.pone.0289847.t004field emission scanning electron microscope.The test results are shown in Figs 4 and 5 and Table	2023-12-02T05:08:40.975Z	2023-11-30T00:00:00.000	{ "year": 2023, "sha1": "9739aa816fc60089a899bfb2950754270b57ee70", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "9739aa816fc60089a899bfb2950754270b57ee70", "s2fieldsofstudy": [ "Engineering", "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }
24840222	pes2o/s2orc	v3-fos-license	Kringle domains of human angiostatin. Characterization of the anti-proliferative activity on endothelial cells. Recently we have identified angiostatin, an endogenous angiogenesis inhibitor of 38 kDa which specifically blocks the growth of endothelial cells (O'Reilly, M. S., Holmgren, L., Shing, Y., Chen, C. , Rosenthal, R. A., Moses, M., Lane, W. S., Cao, Y., Sage, E. H., and Folkman, J. (1994) Cell 79, 315-328; Folkman, J. (1995) Nat. Med. 1, 27-31). Angiostatin was shown to represent an internal fragment of plasminogen containing the first four kringle structures. We now report on the inhibitory effects of individual or combined kringle structures of angiostatin on capillary endothelial cell proliferation. Recombinant kringle 1 and kringle 3 exhibit potent inhibitory activity with half-maximal concentrations (ED50) of 320 nM and 460 nM, respectively. Also, recombinant kringle 2 displays a significant inhibition, although decreased compared with both kringle 1 and kringle 3. In contrast, kringle 4 is an ineffective inhibitor of basic fibroblast growth factor-stimulated endothelial cell proliferation. Among the tandem kringle arrays, the recombinant kringle 2-3 fragment exerts inhibitory activity similar to kringle 2 alone. However, relative to kringle 2-3, a marked enhancement in inhibition is observed when individual kringle 2 and kringle 3 are added together to endothelial cells. This implies that it is necessary to open the cystine bridge between kringle 2 and kringle 3 to obtain the maximal inhibitory effect of kringle 2-3. An increased (<2-fold) inhibitory activity is observed for the kringle 1-3 fragment (ED50 = 70 nM) compared with kringle 1-4 (ED50 = 135 nM). These data indicate that the anti-proliferative activity of angiostatin on endothelial cells is shared by kringle 1, kringle 2, and kringle 3, but probably not by kringle 4 and that more potent inhibition results when kringle 4 is removed from angiostatin. Thus, in view of the variable lysine affinity of the homologous domains, it would appear that lysine binding capability does not correlate with the relative inhibitory effects of the kringle-containing constructs. However, as we also demonstrate, appropriate folding of kringle structures is essential for angiostatin to maintain its full anti-endothelial activity. The formation of new blood vessels occurs as a result of the growth of capillaries by vascular sprouting from preexisting vessels, a process called angiogenesis (1,2). Upon growth stimulation, quiescent endothelial cells can enter into the cell cycle, migrate, degrade the underlying basement membrane, and form a lumen. Angiogenesis is required for a variety of physiological processes such as embryonic development, wound healing, and tissue as well as organ regeneration (3,4). These processes depend upon the tightly regulated growth of endothelial cells which can be switched on and off within a short period. Abnormal growth of new blood vessels can lead to the progression of many diseases such as diabetic retinopathy and tumor growth (4). Direct experimental evidence shows that tumor growth and metastases are angiogenesis-dependent (1)(2)(3)(4). The switch to the angiogenic phenotype requires both upregulation of angiogenic stimulators and down-regulation of angiogenesis inhibitors (5,6). A variety of growth factors can stimulate angiogenesis in vitro and in vivo (4). Of the known angiogenic factors, fibroblast growth factors (FGFs) 1 and vascular endothelial growth factor (VEGF)/vascular permeability factor are most commonly expressed in tumors (7)(8)(9)(10)(11). Tumor cells may overexpress one or more of these angiogenic factors that may function synergistically in promoting tumor growth. To date, VEGF has been reported as an endothelial cell-specific mitogen (12). Although overproduction of angiogenic stimulators is necessary for the angiogenic switch, it is not sufficient for a tumor to become angiogenic. Expression of angiogenesis inhibitors must be simultaneously down-regulated (13,14). An increasing number of negative angiogenesis regulators (nine presently known) have been identified (4). Among these endogenous suppressors, angiostatin (15) is produced in association with a murine Lewis lung carcinoma growth. Angiostatin has been characterized as a potent inhibitor of angiogenesis which specifically targets proliferating endothelial cells (3,15). Angiostatin is a circulating angiogenesis inhibitor that has been purified from serum and urine of mice bearing a murine Lewis lung carcinoma (15). Amino acid sequence analysis reveals that angiostatin is identical to an internal fragment of mouse plasminogen beginning with amino acid residue Val 79 (Val 98 including the signal peptide). Based on its molecular weight, the carboxyl terminus of angiostatin is predicted to be * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 residue 440 of plasminogen. Thus, angiostatin contains the first four triple loop disulfide-linked structures of plasminogen, known as kringle domains. Proteolytic fragments of human plasminogen comparable to the murine angiostatin, but not intact plasminogen, inhibit the neovascularization and growth of lung metastases in mice (15). In vitro, human angiostatin specifically inhibits endothelial cell proliferation but fails to inhibit proliferation of other cell types, including Lewis lung tumor cells. In vivo, it inhibits neovascularization in the chick chorioallantoic membrane assay and in the mouse corneal assay (15). We report here the characterization of the inhibitory activity of individual and multiple kringle fragments of angiostatin. This study was made possible by the availability of each of the individual angiostatin kringle modules as recombinant proteins in their properly folded and "active" (i.e. endowed with their characteristic ligand binding properties) forms (16 -19). Our data suggest that a functional difference is present among individual kringle structures and that a comparable, if not more potent, anti-endothelial activity can be obtained from a smaller fragment than from intact angiostatin. Integrity of the kringle structures of angiostatin is required to maintain its inhibitor potency. EXPERIMENTAL PROCEDURES Proteolytic Fragments of Plasminogen-Fragments containing kringles 1-3 (K1-3), kringles 1-4 (K1-4) and kringle 4 (K4) (Fig. 1) were prepared by digestion of human plasminogen (HPg) with Lys 78 carboxyl terminus (Lys-HPg) (by methods used in Abbott Laboratories) with porcine elastase (Sigma) as published (20,21). Briefly, 1.5 mg of elastase was incubated at room temperature with 200 mg of HPg in 50 mM Tris-HCl pH 8.0 overnight with shaking. The reaction was terminated by the addition of diisopropyl fluorophosphate (Sigma) to a final concentration of 1 mM. The mixture was rocked for an additional 30 min at room temperature and dialyzed overnight against 50 mM Tris-HCl, pH 8.0. Gene Construction, Expression, and Purification of Recombinant Kringles-A polymerase chain reaction-based method was used to generate the cDNA fragments coding for K1, K2, K3, K4, and K2-3 of HPg ( Fig. 1). Recombinant K1 (rK1) and K4 (rK4) were expressed in Escherichia coli strain DH5␣ using a pSTII plasmid vector as published (16,17). The expressed proteins are obtained periplasmic, i.e. oxidatively refolded. Purification to homogeneity was achieved by chromatography on lysine-Sepharose 4B (Pharmacia Biotech Inc.) and Mono Q (Bio-Rad) resins (16,17). Recombinant K2, K3, and K2-3 were expressed in E. coli bacterial cells strains BL21 (rK2) and M15 (rK3, rK2-3) using the expression vector pQE8 (Qiagen), as reported (18,19). To avoid homodimerization by formation of interkringle disulfide bridges, the cysteine residues 169 in rK2 and 297 in rK3 were mutated to serines (19). The rK2, rK3, and rK2-3 contained an amino-terminal hexahistidine tag, with an inserted factor Xa cleavage site. Initial purification was by chromatography on a Ni 2ϩ -nitrilotriacetic acid-agarose column (18,19). In the case of rK2-3, ␤-mercaptoethanol was added to all extraction buffers to a final concentration of 5 mM. To achieve refolding (22), the purified proteins were adjusted to pH 8.0, and dithiothreitol was added to a final concentration of 5 mM. Following an overnight incubation, the solution was diluted gradually with 5 volumes of 50 mM Tris-HCl, pH 8.0, containing each reduced and oxidized glutathione at a concentration of 1 mM. The renatured proteins were dialyzed against H 2 O and lyophilized. Final purification of rK2 and rK2-3 was achieved by affinity chromatography on a lysine-Bio-Gel column (2 ϫ 13 cm) and eluted with phosphate buffer containing 50 mM 6-aminohexanoic acid. After proteolytic removal of the hexahistidine tag from the rK2, the proteins were desalted by gel filtration on Sephadex G-50f. Recombinant K3 was purified by reverse-phase HPLC on an Aquapore Butyl column (2.1 ϫ 100 mm, widepore 30 nM, 7 m; Applied Biosystems) using a Hewlett Packard liquid chromatograph 1090 and acetonitrile gradients. The fact that rK1, rK2, rK2-3, and rK4 are all retained upon affinity chromatography on lysine-conjugated support gels affords good indication that the recovered proteins are functional. Further evidences of proper folding were provided by their thermal melting behaviors in the presence and absence of specific lysine analog ligands (16,17), NMR spectra by reference to intact modules obtained via proteolytic fragmentation of plasminogen (16 -19), and, in the case of rK1, independent x-ray crystallographic structures determined ligand-free (23) and complexed (24) with specific -aminocarboxylic acid ligands, analogs of lysine. Reduction and Alkylation-The reduction and alkylation of kringle fragments were performed according to a standard protocol, as published (25). Approximately 20 -80 g of purified protein in 300 -500 l of DMEM in the absence of serum were incubated at room temperature with 15 l of 0.5 M dithiothreitol for 15 min. After incubation, 30 l of 0.5 M iodoacetamide was added to the reaction. The protein solution was dialyzed at 4°C overnight against 20 volumes of DMEM. The solution was dialyzed further at 4°C for additional 4 h against 20 volumes of fresh DMEM. After dialysis, the samples were analyzed on a SDS gel and assayed for their inhibitory activities on endothelial cell proliferation. Endothelial Proliferation Assay-Bovine capillary endothelial (BCE) cells were isolated as described previously (26) and maintained in DMEM supplemented with 10% heat-inactivated bovine calf serum, antibiotics, and 3 ng/ml recombinant human bFGF (Scios Nova, Moun- tainview, CA). Monolayers of BCE cells growing in six-well plates were dispersed in a 0.05% trypsin solution. Cells were resuspended with DMEM containing 10% bovine calf serum. Approximately 12,500 cells in 0.5 ml were added to each well of gelatinized 24-well tissue culture plates and incubated at 37°C (in 10% CO 2 ) for 24 h. The medium was replaced with 500 l of fresh DMEM containing 5% bovine calf serum, and samples of individual or tandem kringle fragments in triplicate were added to each well. After 30 min of incubation, bFGF was added to a final concentration of 1 ng/ml. After a 72-h incubation, cells were trypsinized, resuspended in Hematall (Fisher Scientific), and counted with a Coulter counter. Purification and Characterization of Kringle Fragments of HPg-Recombinant proteins expressed from E. coli were refolded in vitro and purified to homogeneity using HPLC (Fig. 2). Under reducing conditions, rK2, rK3, and rK4 migrated with molecular masses of 12-13 kDa ( Fig. 2A, lanes 3-5), corresponding to the predicted molecular mass of each kringle fragment. Surprisingly, recombinant K1 migrated with a higher molecular mass of 17 kDa on SDS gel electrophoresis ( Fig. 2A, lane 2). This has been observed previously for a rK1 segment expressed as a different construct (27), which suggests oligomerization of the unfolded K1 unit. The fragments K1-4 (angiostatin) and K1-3, obtained via elastolytic digestion of Lys-HPg, migrated with the predicted molecular masses of 43 and 35 kDa, respectively (Fig. 2B, lanes 2 and 3). Aminoterminal amino acid sequence analysis of the purified fragments yielded an identical sequence, YLSECKTGNGK, for K1-3 and K1-4. The amino-terminal sequence for K4 produced VVQDCYHGDG, with approximately 20% VQDCYHGDG, as predicted from the expected sequence beginning with Val 354 and Val 355 of HPg, as observed previously (28). Recombinant K2-3 (21,564 kDa, determined by mass spectroscopy) also yielded a well defined band on SDS gel electrophoresis, albeit somewhat shifted from the 21.5-kDa standard marker (Fig. 2B, lane 4). Anti-endothelial Cell Proliferative Activity of Individual Kringles-Individual recombinant kringle fragments of angiostatin were assayed for their inhibitory activities on BCE cell growth stimulated by bFGF. As shown in Fig. 3A, rK1 inhibited BCE cell proliferation in a dose-dependent fashion. The concentration of rK1 required to reach 50% inhibition (ED 50 ) was about 320 nM (Table I). In contrast, rK4 exhibited a significantly less marked effect on endothelial cell proliferation. The anti-endothelial cell proliferation activity of rK4 was directly compared with that of native K4 derived from proteolytic plasminogen. The lack of inhibitory activity of these two fragments was virtually identical. A dose-dependent inhibition of endothelial cell proliferation was also observed for rK2 and rK3 (Fig. 3B). However, the inhibitory potency of rK2 was consistently lower than that of rK1 or rK3 (ED 50 ϭ 460) ( Fig. 3 and Table I). No cytotoxicity or distinct morphology associated with apoptotic endothelial cells such as rounding, detachment, and fragmentation of cells could be detected, even after incubation with a high concentration of these kringle fragments. These data suggest that the anti-endothelial growth activity of angiostatin is shared by fragments of K1, K2, and K3, but not, or only marginally by K4. Anti-endothelial Cell Proliferative Activity of K1-3 and K1-4 Fragments-To evaluate the anti-endothelial cell proliferative effect of combined kringle fragments, purified proteolytic fragments of human K1-4, K1-3, and rK2-3 were assayed on BCE cells. In agreement with our previous finding (15), BCE cell proliferation was inhibited significantly by angiostatin (ED 50 ϭ 135 nM) ( Fig. 4 and Table I). An increase of anti-endothelial growth activity is suggested for the K1-3 fragment (ED 50 ϭ 70 nM) (Table I). In both cases, the inhibition of endothelial cell proliferation occurred in a dose-dependent manner. These results indicate that the presence of K4 is not required for anti- endothelial growth activity and that its removal from angiostatin is likely to potentiate its activity. Additive Inhibition by rK2 and rK3-The rK2-3 fragment displayed only weak inhibitory activity that was similar to that of rK2 alone (Fig. 4). However, both rK2 and rK3 inhibited endothelial cell proliferation (Fig. 3B). This finding suggested to us that the inhibitory effect of K3 might be hidden in the structure of K2-3 as an interkringle disulfide bond links Cys 169 in K2 to Cys 297 in K3 (19,20) (Fig. 5B). To test this hypothesis, we examined the inhibitory effect of rK2 and rK3 in combination. Interestingly, a marked inhibition was seen when individual rK2 and rK3 fragments were added together to BCE cells (Fig. 5A). Inhibitory Activity of Combined Kringle Structures-To determine whether various fragments have enhancing or antagonistic effects on inhibition of endothelial cell proliferation, various kringle structures of human angiostatin at the concentration of 320 nM were tested on BCE cells. We first analyzed the two lysine-binding kringles rK1 and K4. As shown in Fig. 6A, K4 did not interfere with the inhibitory activity of K1, suggesting that the lysine binding site of rK1 might not be directly involved in the anti-endothelial growth activity. To investigate further if the lysine binding site is involved in the inhibitor activity, we studied the effect of 6-aminohexanoic acid (a ligand specific for the kringle lysine binding site) on kringle 1 and found no difference in its anti-proliferative activity. Similarly, rK2, which also exhibits measurable lysine binding ca-pability, 2 does not interfere with the inhibitory effect of rK1 (Fig. 6A). However, the inhibitory activity of rK1 was decreased significantly by the addition of rK2-3 (Fig. 6B). This finding suggests that the fragment of rK2-3 could compete for the binding site of K1 on endothelial cells. In contrast, the intact K1-3 fragment exerted a potent inhibitory effect on BCE cells. Thus, integrity of K1 and K2-3 within the same fragment is required to achieve maximal inhibition. Appropriate Folding of Kringle Structures Is Required for the Anti-endothelial Activity of Angiostatin-To study whether the folding of kringle structures is required for the anti-endothelial proliferation activity, native angiostatin (K1-4) was reduced with dithiothreitol, alkylated with iodoacetamide, and assayed on BCE cells. As shown in Fig. 7A, on SDS gel electrophoresis, the reduced/alkylated protein migrated to a higher position, with a molecular mass of about 42 kDa (lane 2) compared with the native angiostatin with a molecular mass of 33 kDa (lane 1), suggesting that the native angiostatin cystine bridges were significantly disrupted. The anti-proliferative activity of angiostatin was largely abolished after reduction/alkylation (Fig. 7B). Thus, appropriate folding of kringle structures as tandem domains held together by intrachain or interchain (K2-3) disulfide bonds is required for angiostatin to display its potent inhibitory activity. DISCUSSION Angiostatin is a proteolytic fragment of plasminogen processed in mice bearing a murine Lewis lung primary tumor. This fragment mediates the suppression of metastatic growth (3,15) and primary tumor growth (55) by blocking tumorinduced angiogenesis. At present, the origin of angiostatin in vivo is not known. It is unlikely that tumor cells produce angiostatin directly because they lack a detectable amount of mRNA for plasminogen. 3 Although elastase cleaves plasminogen into an active form of angiostatin in vitro, it is not yet clear which protease(s) is involved in conversion of plasminogen to angiostatin in vivo. Our data demonstrate that smaller kringle fragments of human angiostatin contain inhibitory activity on endothelial cell proliferation. Amino acid sequence alignment of the kringle domains of HPg shows that K1, K2, K3, and K4 display considerable sequence similarity (48 -50% identity, Fig. 8). NMR (19,29,30) and x-ray crystallographic (23,24,31,32) studies have demonstrated that the homology translates into a remarkable conformational uniformity. Among these structures, the lysine-binding K1 is the most potent inhibitory segment of endothelial cell proliferation. However, K3, which lacks lysine binding potential, exhibits higher inhibitory potency than either K2 4 or K4 (20,31,33), which show definite lysine binding capability. Indeed, K4, which manifests comparable or even higher affinity for lysine than does K1, exhibits marginal or no inhibitory activity. Hence, the lysine binding site of the kringle structures may not be directly involved in the inhibitory activity. In view of the functional divergence of the homologous structures, the kringle domains provide a unique system to study the role of mutations caused by DNA replication during evolution (34). Similar divergent activities related to the regulation of angiogenesis exhibited by a group of structurally related proteins are also found in the -CXC-chemokine and prolactingrowth hormone families (35)(36)(37)(38)(39). K4 is unique among the HPg kringles in that it contains two sets of positively charged lysine pairs, adjacent each to Cys 22 and Cys 80 , respectively (Fig. 8). Inspection of the three-dimensional structure (29,31) shows that these four lysines, together with Lys 59 , configure an exposed, positively charged area in K4, whereas other kringle structures lack such a cationic cluster. Whether this lysine-enriched domain contributes to the loss of inhibitory activity of the HPg K4 remains to be established. K4 was previously reported to stimulate proliferation of other cell types and to increase the release of intracellular calcium (40). The fact that removal of K4 from angiostatin somewhat potentiates its inhibitory activity on endothelial cells suggests that this structure may prevent some of the inhibitory effect of K1-3. Because proteases such as elastase can convert plasminogen to K1-3 in vitro, it would be interesting to investigate if such a conversion also occurs in vivo. The mechanism underlying how angiostatin and its related kringle fragments specifically inhibit endothelial cell growth is unclear. It is not known whether the inhibition occurs directly at the cell surface or whether angiostatin is internalized by endothelial cells and subsequently inhibits cell proliferation. Both mechanisms could involve a receptor specifically expressed in proliferating endothelial cells. Alternatively, angiostatin may interact with an endothelial cell adhesion receptor such as integrin ␣ v ␤ 3 , thus blocking integrin-mediated angiogenesis (41). Of interest, Friedlander et al. (42) reported recently that in vivo angiogenesis in cornea or chorioallantoic membrane models (induced by bFGF and by tumor necrosis factor) was integrin ␣ v ␤ 3 -dependent. However, angiogenesis stimulated by VEGF, transforming growth factor-␣, or phorbol esters was dependent on integrin ␣ v ␤ 5 . Antibodies to the individual integrins specifically blocked one of these pathways, and a cyclic peptide antagonist of both integrins blocked angiogenesis induced by each cytokine (42). Because bFGF-and VEGFinduced angiogenesis are inhibited by angiostatin, it may block a common pathway for these integrin-mediated angiogenesis. An increasing number of endogenous angiogenesis inhibitors have been identified in the last few decades (4). Of the characterized endothelial cell suppressors, several inhibitors are proteolytic fragments. For example, the 16-kDa amino-terminal fragment of human prolactin inhibits endothelial cell proliferation and blocks angiogenesis in vivo (43). In a recent paper, D'Angelo and co-workers reported that the anti-angiogenic 16-kDa amino-terminal fragment inhibited the activation of mitogen-activated protein kinase by VEGF and bFGF in capillary endothelial cells (44). Similar to angiostatin, the intact parental molecule of prolactin does not inhibit endothelial cell proliferation, nor is it an angiogenesis inhibitor. Platelet factor 4 inhibits angiogenesis at high concentrations (35,36). However, the amino-terminally truncated proteolytically cleaved platelet factor 4 fragment exhibits a 30 -50-fold increase in its antiproliferative activity over the intact platelet factor 4 molecule (45). Smaller peptide fragments of fibronectin, murine epidermal growth factor, and thrombospondin have also been shown to inhibit endothelial cell growth selectively (46 -48). When combined with our observation that the proteolytically processed fragments of K1-4 and K1-3 inhibit endothelial cell growth, these findings lead us to suggest that proteolytic processing of a large protein may change the conformational structure of the original molecule or expose new epitopes that are anti-angiogenic. Thus, protease(s) may play a critical role in the regulation of angiogenesis. To date, little is known about the regulation of these protease activities in vivo. In this paper we also show that the disulfide bond-mediated folding of the kringle structures in angiostatin is essential to maintain its inhibitory activity on endothelial cell growth. Kringle structures analogous to those in plasminogen are also found in a variety of other proteins. For example, apolipoprotein(a) has as many as 37 repeats of plasminogen K4 (49). The amino-terminal portion of prothrombin also contains two kringles that are homologous to those of plasminogen (50). The tissue-type and the kidney-type (urokinase) plasminogen activators possess kringle modules that share extensive homology with those of plasminogen (51,52). In addition, TrK-related tyrosine kinases (53) and hepatocyte growth factor also carry kringle structures (54). It remains to be seen whether these kringle proteins or kringle fragments can also inhibit endothelial cell growth. Endothelial cell-specific inhibitors such as angiostatin or related fragments may be useful for long term therapy in suppression of both primary and metastatic tumor growth. Therapeutic use of endogenous inhibitors is less likely to cause side effects such as suppression of hematopoiesis and gastrointestinal symptoms (4). The combination of various angiogenesis inhibitors may have a synergistic inhibitory effect on tumor growth because they probably block angiogenesis via different mechanisms.	2018-04-03T01:29:20.290Z	1996-11-15T00:00:00.000	{ "year": 1996, "sha1": "5c87d7c2ee2f4e1fad5d320a7e9999988b53318e", "oa_license": "CCBY", "oa_url": "http://www.jbc.org/article/S0021925818350889/pdf", "oa_status": "HYBRID", "pdf_src": "Highwire", "pdf_hash": "b8d27e10000300048eb9493a5ee42320ffe59f9b", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }
11348651	pes2o/s2orc	v3-fos-license	Folate deficiency presenting as pyrexia: a case report Folate deficiency is an uncommon cause of pyrexia. We describe the case of a 29-year-old male who presented with a pyrexial illness subsequently attributed to megaloblastic anaemia secondary to severe folate deficiency, after exclusion of other infective or inflammatory causes. A temperature chart documenting the course of the patient's pyrexia is presented and potential pathophysiological mechanisms are proposed. Folate deficiency is a reversible cause of pyrexia that should be considered in any patient who presents with a pyrexial illness of unknown cause. A 29-year-old caucasian male presented to our medical admissions unit complaining of a 6 week history of dyspnoea on exertion, recurrent fevers and sweats. He had a 7 year history of alcohol excess, drinking 3-4 litres of cider per day. He was a smoker of 10 cigarettes per day, but was otherwise fit and well with no significant past medical or family history. The patient reported no other symptoms and did not take any regular medication. Examination revealed a pulse rate of 124 beats/minute and a blood pressure of 120 mmHg/48 mmHg. He was noted to be febrile with a temperature of 38.8°C. Oxygen saturations were 100% on room air with no signs of respiratory distress. He had visibly pale conjunctiva and a lemon yellow tint to the skin (but no evidence of icteric sclerae). There were no signs of chronic liver disease and no peripheral stigmata of bacterial endocarditis. Cardiovascular examination revealed a collapsing pulse and a soft systolic murmur audible at the left sternal edge. Auscultation of the chest revealed vesicular breath sounds with no added sounds. On abdominal examination there was palpable splenomegaly but not other organomegaly or ascites. Neurological examination was normal with was no evidence of neck stiffness or skin rashes. Admission blood tests revealed a severe pancytopenia with a raised mean cell volume (see table 1). Also notable was a mild hyperbilirubinaemia but otherwise normal liver function with no evidence of impaired hepatic synthetic function (normal serum albumin and normal prothrombin time). Subsequently, serum folate levels were found to be low at 1.2 ug/L (normal range 4-24 ug/L) and vitamin B 12 levels were low-normal at 202 ng/L (normal range 180-900 ng/ L). Iron studies were normal. Initial blood film and subsequent bone marrow examination confirmed a severe megaloblastic picture, consistent with folate deficiency. Urine dip test and subsequent urine cultures were negative. Three sets of blood cultures were taken from different sites (before commencement of antibiotics) but were all negative after 5 days culture. Chest radiograph was nor-mal with no evidence of focal consolidation. Abdominal ultrasound confirmed splenomegaly but liver size and architecture was normal. A full auto-antibody screen was sent but found to be negative (see table 2). Clinical progression Initially, the patient was started on intravenous broad spectrum antibiotics as he was presumed to be pyrexial secondary to underlying sepsis of unknown origin. However, despite antibiotics, he did not improve rapidly and continued to remain intermittently pyrexial. Therefore, in the absence of any positive microbiological tests, antibiotics were stopped with continuation of B 12 and folate supplementation alone. Pyrexia settled on day 4 without any further antimicrobial or anti-inflammatory therapy (see figure 1). Symptomatically, the patient improved gradually after commencement of B 12 and folate supplementation on day 2 of admission. Vital signs remained stable throughout the admission. Peripheral blood counts also improved thereafter (see table 3). Subsequent measurement of B 12 and folate levels at follow-up outpatient appointment showed normalisation. Discussion The case presentation and results of microbiological, immunological and radiological investigations described above support our hypothesis that the occurrence of pyrexia in this patient is attributable directly to the presence of megaloblastic anaemia secondary to folate deficiency. We extensively investigated for other possible infective and inflammatory conditions but found no other contributory cause. Fever is a feature of megaloblastic anaemia that has been described previously in the literature [1,2] and is thought to be more common in patients with moderate to severe anaemia and thrombocytopenia [3] The level of pyrexia usually correlates with degree of anaemia and resolves one to three days after adequate vitamin supplementation [4], as illustrated by the case we present (see figure 1). Failure of resolution after vitamin supplementation should, however, suggest the probability of an alternative cause for the pyrexia [1]. The exact cause of pyrexia in megaloblastic anaemia is not known and previously it has been hypothesised that it may reflect a defect in oxygenation to the temperature regulatory centres in the brain [5]. However, this theory does not explain why the fever seen in patients with megaloblastic anaemia is not a recognised feature of other forms of anaemia. Another proposed mechanism is that megaloblastic anaemia leads to hyperplasia and thus increased activity within the bone marrow leading to systemic pyrexia [1,5]. The mechanism of how pyrexia is induced by an over-productive marrow is, however, unclear. In previous case series, treatment with B 12 and folic acid has been shown to cause rapid resolution of fever and this is felt to be due to immediate improvement in ineffective erythropoiesis [1]. This rapid resolution of fever was also seen in the case we describe and further supports our hypothesis of pyrexia caused directly by folate deficiency. Conclusion Megaloblastic anaemia is a rare but reversible cause of pyrexia that should be considered in any patient who presents with a pyrexial illness without other infective or inflammatory source. This is a particularly important consideration as such patients with fever and a pancytopenic peripheral blood picture may initially be wrongly treated with broad spectrum antibiotics as part of a neutropenic sepsis protocol (as demonstrated by the case we describe). Measurement of B 12 and folate levels should be requested as part of a screen sent for any patient who is pyrexial without an obvious cause.	2014-10-01T00:00:00.000Z	2008-10-26T00:00:00.000	{ "year": 2008, "sha1": "e89594b2102288c67120e0543eb0f7ce6ab18c52", "oa_license": "CCBY", "oa_url": "https://casesjournal.biomedcentral.com/track/pdf/10.1186/1757-1626-1-275", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "e89594b2102288c67120e0543eb0f7ce6ab18c52", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
211269169	pes2o/s2orc	v3-fos-license	Comparative study of intrathecal nalbuphine versus clonidine as adjuvants to 0.5% isobaric levobupivacaine for elective infra umbilical surgeries Background and objectives: Various intrathecal adjuvants have been clinically tried for the prolongation of intraoperative and postoperative analgesia. This study aims at evaluating the effects of intrathecal nalbuphine and clonidine as adjuvants to isobaric levobupivacaine in subarachnoid block. Methodology: 60 patients scheduled for elective infra umbilical surgeries were allocated into two groups of thirty each to receive 15 mg of 0.5% isobaric Levobupivacaine with either 1 mg nalbuphine (Group LN) or 30 μg clonidine (Group LC) intrathecally. Characteristics of spinal anesthesia in terms of sensory analgesia and motor blockade were noted. Hemodynamic parameters and adverse effects if any were recorded. Data obtained was compiled and statistically analysed with appropriate tests. Results: Onset of sensory and motor blocks was faster in group LN (2.43 ± 0.93 and 2.2 ± 0.9 min) compared to group LC (3.26 ± 1.04 and 3.13 ± 1.0 min). However, time to two segment regression (186.8 ± 24.5 vs 146.5 ± 21.4), total duration of effective analgesia (384.1 ± 56.6 vs 292.1 ± 40.9) and total duration of motor block (345.3 ± 41.7 vs 235.6 ± 29.5 min) were significantly prolonged in group LC than in group LN. There was no significant difference in hemodynamic changes and adverse effects between the groups. Conclusion: The addition of 30 μg clonidine to intrathecal 0.5% isobaric levobupivacaine as adjuvant, is associated with prolonged sensory and motor blockade with better perioperative analgesia compared to 1 mg nalbuphine. INTRODUCTION Spinal anesthesia though introduced more than 100 years back is still the standard anesthetic technique practiced for infra umbilical surgeries but the main drawback is its limited duration of action and does not provide prolonged postoperative analgesia if only local anesthetic is administered. 1 Alleviation of pain is of principal importance for the anesthesiologist in perioperative period, as it helps in smoother postoperative course and earlier discharge from hospital. 2 Intrathecal adjuvants increase the speed of onset (reduce latency), improve the quality and prolong the duration of neural blockade. There are myriad choices of neuraxial adjuvants like opioids (morphine, fentanyl, buprenorphine, nalbuphine), N-methyld-aspartate (NMDA) antagonists (ketamine), Citation: Shalini A, Kokila N, Manjunatha HG, Supriya L. Comparative study of intrathecal nalbuphine versus clonidine as adjuvants to 0.5% isobaric levobupivacaine for elective infra umbilical surgeries. Anaesth pain & intensive care 2019;23(4):370-376. DOI: https://doi.org/10. 35975/apic.v23i4.1169 alpha 2 adrenoceptor agonists (clonidine and dexmedetomidine), vasoconstrictors (adrenaline), GABA receptor agonists (midazolam), cholinergic agonists (neostigmine) and sodium bicarbonate but no drug inhibits nociception without its associated adverse effects. 3 Intrathecal opioids are synergistic with local anesthetics, thereby intensifying sensory block without affecting the sympathetic blockade. 2 However, opioids may produce side effects like pruritus, nausea, vomiting, respiratory depression, sedation, urinary retention and shivering. 4 Most of these side effects are mainly due to their action on mu opioid receptor. Nalbuphine is a synthetic, mixed agonist-antagonist opioid analgesic with agonistic action at kappa receptor and antagonism at mu receptor. The main advantage is its increased margin of safety due to ceiling effect on respiratory depression and sedation. 5 Despite these good qualities of nalbuphine, its use is not widespread for regional anesthesia. By using non-opioid drugs as an alternative to opioids the side effects of opioids can be avoided. Clonidine is a selective partial alpha 2 adrenergic agonist, and its intrathecal administration prolongs sensory as well as motor block when used as an adjuvant to local anesthetic. Analgesic action is due to its binding on alpha 2 receptors located in brainstem nuclei and spinal substantia gelatinosa. Clonidine also has antiemetic, anxiolytic, antishivering, sedative properties without respiratory depression. 6,7,8,9,10 Levobupivacaine, a levo-isomer of bupivacaine which is available as an isobaric preparation produces sensory and motor blockade similar to bupivacaine but has less cardiotoxic potential and lesser motor blockade leading to early mobilization of the patient. Its decreased toxicity is attributed to its less affinity and inhibitory effect on cardiac sodium channels and faster protein binding rate. 7,10,11 There are various studies conducted to compare the efficacy of intrathecal nalbuphine and clonidine at various doses with hyperbaric bupivacaine. 1,5,12 The present study was designed to evaluate the sensory and motor block characteristics, hemodynamic changes and any adverse effects of nalbuphine and clonidine when used as adjuvants to 0.5% isobaric levobupivacaine in patients undergoing elective infra umbilical surgeries under subarachnoid block. METHODOLOGY After institutional ethical committee approval, a prospective, double blind, randomized clinical study was conducted on 60 patients who gave a valid informed written consent and belonged to American Society of Anesthesiologists (ASA) class I and II, in the age group of 18-60 y, with a height 150-170 cm, undergoing elective infra umbilical surgeries under subarachnoid block at our tertiary care hospital. Patients with systemic diseases like diabetes mellitus, hypertension, cardiovascular, respiratory, hepatic, renal or neurologic disorders, and posted for emergency surgeries were excluded from the study. Unwilling patients, patients with history of known hypersensitivity to study drugs, or any contraindication to subarachnoid block, were also excluded from the study. Sixty patients were randomly allocated into two groups of thirty each by shuffled sealed opaque envelope method. A thorough pre-anesthetic checkup was carried out for each patient with relevant laboratory and radiological investigations. All patients were visited a day prior to the surgery and explained in detail about the anesthetic technique and visual analogue scale (VAS). Tablet ranitidine 150 mg and tablet alprazolam 0.5 mg orally was given at night as pre-medication and patients were kept nil orally from 12 midnight. On the day of surgery, after securing an intravenous access with 18G cannula patients were preloaded with Ringer lactate solution 10 ml/kg before the initiation of spinal anesthesia. Ampoules containing drugs 0.5% isobaric levobupivacaine (LEVO-ANAWIN 0.5% by Neon Laboratories limited), inj. clonidine hydrochloride (CLONEON 150 µg/ml by Neon laboratories limited) and inj nalbuphine hydrochloride (NACPHIN 10 mg/ml by Neon laboratories limited) were used for the study. The study drugs were loaded in a 5 ml syringe by an anesthesiologist who was not a part of the study, just before intrathecal injection. Intraoperative monitoring was done with baseline recording of heart rate (HR), oxygen saturation (SPO2), non-invasive blood pressure (NIBP) and electrocardiography (ECG) using automated multiparameter monitor. Under aseptic precaution, subarachnoid block was performed at L3-L4 interspace using 25 G Quincke's spinal needle with patient in lateral position by using midline approach after local infiltration of skin with lignocaine 2%. All the subarachnoid blocks were performed and monitored by the same anesthesiologist, who was also the observer of the study. Thus, both the observer and the subjects were blinded to the study drugs achieving double blinding. The patients received either of the drug solution below Group LN -3 ml of 0.5% isobaric levobupivacaine + 0.1 ml of nalbuphine (1 mg) + 0.1 ml normal saline to make a total volume of 3.2 ml or Group LC -3 ml of 0.5% isobaric levobupivacaine + 0.2 ml of clonidine (30 µg) to a total volume of 3.2 ml Time of intrathecal injection was noted. Sensory testing was assessed by loss of pinprick sensation to original article a blunt 27G hypodermic needle and degree of motor block assessed by modified Bromage scale. Onset of sensory block was considered as the time taken from intrathecal injection of drug to loss of sensation to pinprick at T10 dermatome. Maximum level of sensory block attained, time taken for maximum dermatomal level, two segment sensory regression time, total duration of effective analgesia (the time from intrathecal injection of drug to the first request of analgesia with VAS > 3) were noted. The time needed for the onset of motor block (Bromage 1), time taken for maximum motor block and total duration of motor block (time taken for complete motor recovery to Bromage 0) were also noted. Hemodynamic parameters like heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), mean arterial blood pressure (MAP) and respiratory rate were monitored every 2 min for the first 10 min, every 5 min for the next 30 min and then every 15 min till the completion of surgery. Mephentermine 6 mg was given to treat arterial hypotension (fall in SBP > 20% from basal value or < 90 mmHg). Atropine was used if HR went below 55 bpm. Incidence of side effects like nausea, vomiting, respiratory depression (respiratory rate < 10/min and SpO 2 < 90%), shivering, pruritus etc. were monitored throughout the procedure and postoperatively. Sedation was assessed by Ramsay sedation scoring. After the surgery patients were shifted to post anesthesia care and recovery unit and monitored until complete recovery from spinal anesthesia. Postoperative pain was assessed by VAS, where patients were asked to mark on the scale the degree of pain experienced every half an hour. When they complained of pain with VAS > 3, inj. diclofenac 75 mg intramuscular was given and the study ended. Statistical analysis: Sample size calculation was based on previous studies. An estimated 26 patients per group were necessary in order to detect at least clinically significant difference of 30 min in mean duration of analgesia between the groups for achieving type 1 error of 0.05 with 80% power and 95% confidence interval for group comparison. To compensate for the dropouts from the study and to make sure that the sample size is adequate, we selected 60 subjects. Continuous variables were summarized in the form of mean ± standard deviation (SD), median (range) and categorical variables as percentages. The results were analyzed statistically using Student's t-test (for continuous variables) and Chi-square test (for categorical variables). All variables are presented in the form of tables and graphs Data were entered in a spreadsheet and then exported to data editor of Statistical Package for Social Sciences (SPSS; Windows version 22.0) for analysis. p < 0.05 was considered to be significant, and < 0.01 as highly significant. RESULTS The two groups of patients enrolled in the study did not differ significantly with respect to age, sex, body weight, height, ASA status and duration of surgery as shown in Table 1. The salient features regarding sensory and motor block characteristics in both the groups are tabulated in Tables 2 and 3 respectively. In our study time to achieve T10 dermatomal level of analgesia was faster in group LN when compared to group LC which was statistically highly significant (p < 0.01). The maximum sensory level achieved in both the groups were comparable and statistically not significant (p= 0.293). The time taken to achieve this maximum sensory level was also statistically insignificant with a p value of 0.226. The time taken for sensory level regression by two dermatomes in group LC was significantly longer than in group LN (p < 0.01). The total duration of effective analgesia was significantly longer in group LC when compared to group LN (p < 0.01). In our study onset of motor block was significantly faster in group LN. Complete motor blockade of Bromage 3 was achieved in all 30 patients in group LC and 28 patients in group LN (p=0.15). The difference in time taken to achieve Bromage 3 was not significant (p > 0.05). The total duration of motor block was significantly prolonged (p<0.01) in group LC. Regarding hemodynamic changes there were no significant alterations in the measured parameters between the two groups (p > 0.05) at various time intervals. Variations in mean HR and MAP are shown in Figures 1 and 2 respectively. nalbuphine vs clonidine as adjuvants to levobupivacaine in bpm) at various time intervals Ramsay sedation score used for sedation assessment was < or =2 and was comparable in both the groups (p > 0.05). Incidence of adverse effects observed in both the groups are tabulated in Table 4 below. DISCUSSION Spinal anesthesia is the most commonly used technique for infra umbilical surgeries because it is easy, reliable, provides good analgesia and muscle relaxation, cost effective and analgesia extends into the postoperative period too. 13 To improve the subarachnoid block characteristics adjuvants like opioids and alpha 2 agonists are often being used with intrathecal local anesthetics. The rationale for combining opioids with local anesthetics intrathecally is that two different types of drugs eliminate pain by acting at two different sites in spinal cord, local anesthetics at the nerve axonal level (by blocking voltage gated sodium channels) and opioids at the receptor level (substantia gelatinosa to modulate the function of afferent pain carrying nerve fibers). Some of this intrathecal opioid gets absorbed into the systemic circulation and acts on the opioid receptors at brain. 13 Degree of this absorption depends upon the lipophilicity of the drug. Nalbuphine a synthetic opioid has agonistic action at kappa receptor and antagonistic action at mu receptor. It stimulates the kappa receptor and produces analgesia without the undesirable side effects of mu receptor agonism. It inhibits the release of substance P, which is a neurotransmitter that mediates pain. In addition, it acts as a post synaptic inhibitor of interneuron and ascending nociceptive spinothalamic tract. 12,13 Since it is highly lipid soluble it diffuses into systemic circulation fast unlike hydrophilic opioid like morphine, thereby producing short duration of action. It is safe to use intrathecally unlike morphine which can cause delayed respiratory depression due to rostral spread in CSF. 13 Clonidine is the most commonly used alpha 2 agonist in neuraxial anesthesia with well-established record of safety and efficacy. Clonidine is a partial alpha2 agonist with α1: α2 receptor affinity at a ratio of 200:1. 12 It acts synergistically with local anesthetics by opening potassium channels. Analgesic effect of clonidine is attributed to its blocking of C and A delta fibers. 14 Spinal clonidine binds to postsynaptic α2 receptors in substantia gelatinosa of spinal cord. 9 It also augments acetylcholine release due to its cholinergic activity, which increases the amount of acetylcholine available for modulating pain at the level of substantia gelatinosa. 12 The appropriate dose of intrathecal nalbuphine has been debated. It has been used as an additive with bupivacaine intrathecally in several clinical settings in doses ranging from 0.2 mg-2.4 mg. 3,15,16,17,18,19 Rawal et al 20 showed in a sheep model that intrathecal nalbuphine is not neurotoxic even in large doses 15-24 mg and is not associated with histopathological changes in spinal cord. From previous studies we found that increasing the dose to more than 2 mg showed no benefit, but increased the incidence of side effects. 3,18,19 This is because nalbuphine exerts ceiling effect, increase in the dose increases analgesic effect upto a certain point beyond which there is no further increase in analgesia. So based on these studies we have chosen 1 mg nalbuphine for our study like various researchers. 11,21,22,23 The optimal dose of clonidine for intrathecal use remains unknown, ranging from 15 µg to 150 µg with variable results. Walker et al. 25 have shown the absence of neurotoxicity with intrathecal clonidine. Various authors have studied different doses of clonidine intrathecally and concluded that better prolongation of analgesia and motor block with minimal hemodynamic changes and sedation is seen when 30 µg clonidine was used. 2,5,10,12,26,27,28,29 Based on these studies we chose 30 µg clonidine for our study. In our study, the onset of sensory block and motor block was faster in nalbuphine group compared to clonidine group. However the maximum sensory level attained and the time taken to achieve this peak sensory level were comparable and statistically not significant. Our results correlate to other studies, 1,5,11 The time taken to sensory level regression by two dermatomes and the total duration of effective analgesia were significantly longer in clonidine group compared to nalbuphine group (p < 0.01) as shown by other studies. 5,12 Our findings also correlate with the findings by Das T et al 11 who studied levobupivacaine with various doses of nalbuphine (0.5 mg, 0.75 mg and 1 mg) in patients receiving spinal anesthesia. The time taken to achieve maximum motor blockade (Bromage 3) was comparable between the two groups. The duration of motor block was significantly prolonged in clonidine group when compared to nalbuphine group, as shown by various researchers. 5,12,16,24 Hemodynamic parameters revealed no statistically significant difference. Incidence of adverse effects were comparable between the groups. There was no incidence of pruritus, respiratory depression and desaturation in both the groups. Low dose clonidine is not associated with hemodynamic instability and opioid related side effects were not observed since nalbuphine is a mu receptor antagonist. Our findings correlate well with previous studies. 5,12,13 CONCLUSION On the basis of the results of our study, we conclude that addition of 30 µg of clonidine to intrathecal 0.5 % isobaric levobupivacaine as adjuvant is preferable to 1 mg of nalbuphine, as it provides comparatively more prolonged sensory and motor blockade with better perioperative analgesia. Authors' contribution: SA: Concept and design, conduct of study, review of articles, preparation of manuscript KN: Conduct of study, preparation of manuscript, review of article MHG: Manuscript editing SL: Conduct of study, data collection nalbuphine vs clonidine as adjuvants to levobupivacaine  nalbuphine vs clonidine as adjuvants to levobupivacaine	2020-02-13T09:22:54.918Z	2020-01-14T00:00:00.000	{ "year": 2020, "sha1": "440e75bfbfc6da3820be6871a7c60c5bac137de3", "oa_license": "CCBYNC", "oa_url": "http://www.apicareonline.com/index.php/APIC/article/download/1169/1928", "oa_status": "GOLD", "pdf_src": "MergedPDFExtraction", "pdf_hash": "02d7fc1768cbdffc46a83b91a8ed2a2e470ab707", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
1931076	pes2o/s2orc	v3-fos-license	Heat Kernel Asymptotics on Symmetric Spaces We develop a new method for the calculation of the heat trace asymptotics of the Laplacian on symmetric spaces that is based on a representation of the heat semigroup in form of an average over the Lie group of isometries and obtain a generating function for the whole sequence of all heat invariants. Introduction The heat kernel is one of the most powerful tools in mathematical physics and geometric analysis (see, for example the books [27,20,28,17,29] and reviews [2,22,13,16,19,33]). In particular, it is widely used to study the propagators of quantum fields and the effective action in quantum field theory and the correlation functions and the partition function in statistical physics. The short-time asymptotic expansion of the trace of the heat kernel determines the so-called spectral invariants of the differential operator in question which is intimately related to the renormalization of quantum field theories [22,17,33,29], the high-temperature expansion in statistical physics [28], the dynamics of integrable systems, in particular, the Korteweg-de Vries hierarchy [28,18], as well as spectral geometry and index theorems [27]. There has been a tremendous progress in the explicit calculation of spectral asymptotics in the last thirty years [26,2,3,4,6,32,35] (see also [27,16,19,33,29] and the references therein). However, due to the combinatorial explosion in the complexity of the spectral invariants further progress in the "brute force" approach is unlikely, even if employing computer software for symbolic calculations. The results are so complicated that it requires about twenty pages to describe them [32,35]. It seems that the further progress in the study of spectral asymptotics can be achieved by restricting oneself to operators and manifolds with high level of symmetry, for example, homogeneous and symmetric spaces, which enables one to employ powerful algebraic methods. In some very special particular cases, such as group manifolds, spheres, rank-one symmetric spaces, split-rank symmetric spaces etc, it is possible to determine the spectrum of the differential operator exactly and to obtain closed formulas for the heat kernel in terms of the root vectors and their multiplicities [1,22,23,24,28,25]. The complexity of the method crucially depends on the global structure of the symmetric space, most importantly its rank. Most of the results for symmetric spaces are obtained for rank-one symmetric spaces only. However, it is well known that the spectral asymptotics are determined essentially by local geometry. They are polynomial invariants in the curvature with universal constants that do not depend on the global properties of the manifold. It is this universal structure that we are interested in this paper. We will report on our results obtained in the paper [12] (see also [8]). Related problems in a more general context are discussed in [9,11,14]. The metric also defines the torsion-free compatible connection ∇ T M on the tangent bundle T M, so called Levi-Civita connection. Using the Levi-Civita connection we naturally obtain connections on all bundles in the tensor algebra over T M and T * M; the resulting connection will be denoted just by ∇. It is usually clear which bundle's connection is being referred to, from the nature of the section being acted upon. With our notation, Greek indices, µ, ν, . . . , label the local coordinates x = (x µ ) on M and range from 1 through n. Let ∂ µ be a coordinate basis for the tangent space T x M at a point x ∈ M and dx µ be dual basis for the cotangent space T * x M. We adopt the notation that the Greek indices label the tensor components with respect to local coordinate frame and range from 1 through n. We also adopt the Einstein convention and sum over repeated indices. Let ∇ * : C ∞ (T * M) → C ∞ (M) be the formal adjoint of ∇ : C ∞ (M) → C ∞ (T * M) with respect to the L 2 inner product. A partial differential operator is called the scalar Laplacian. In local coordinates the Laplacian is a second-order partial differential operator of the form where \|g\| = det g µν . It is easy to see that the principal symbol of the operator (−∆) is where ξ ∈ T * x M is a covector. Heat Kernel The following is known about the operator (−∆) [27]. First of all, it is elliptic and self-adjoint. More precisely, it is essentially self-adjoint, i.e. it has a unique self-adjoint extension to L 2 (M). Second, the operator (−∆) has a positive definite leading symbol. The spectrum of such operators forms a real nondecreasing sequence, {λ k } ∞ k=1 , with each eigenspace being finite-dimensional. The eigenvectors {ϕ k } ∞ k=1 are smooth functions and form an orthonormal basis in L 2 (M). Moreover, as k → ∞ the eigenvalues increase as λ k ∼ Ck 2 as k → ∞, with some positive constant C. These facts lead to the existence of spectral asymptotics that will be discussed below. The heat semi-group is a one-parameter family of bounded operators U(t) = exp(t∆) : L 2 (M) → L 2 (M) for t > 0. The integral kernel U(t\|x, x ′ ) of this operator, called the heat kernel, is defined by where each eigenvalue is counted with multiplicities. It satisfies the heat equation with the initial condition where δ(x, x ′ ) is the Dirac delta-function. For t > 0 the heat kernel is a smooth function on M × M with a well defined diagonal Spectral Invariants For any t > 0 the heat semi-group U(t) = exp(t∆) is a trace-class operator with a well defined L 2 trace, called the heat trace, The heat trace is obviously a spectral invariant of the operator (−∆). It determines other spectral functions by integral transforms. In particular, the zetafunction, ζ(s, λ), is defined as the L 2 trace of the complex power of the operator where s and λ are complex variables with Re λ < λ 1 and Re s > n/2. The zeta function enables one to define, in particular, the regularized determinant of the operator (−∆ − λ), which determines the one-loop effective action in quantum field theory. Asymptotic Expansion It is well known that the heat kernel diagonal has the following asymptotic expan- The coefficients a k are called local heat kernel coefficients. They are scalar polynomials in the curvature and its covariant derivatives which are known explicitly up to a 5 . In particular, where R µναβ is the Riemann tensor, R µν = R α µαν is the Ricci tensor and R = R µν µν is the scalar curvature. The coefficient a 3 was computed in [26]. The coefficient a 4 was first computed in [2] and published in [3,4,6] (see also [17]). The coefficient a 5 was computed in [32,35]. The asymptotic expansion of the heat kernel diagonal can be integrated over the manifold to give the asymptotic expansion of the heat trace as t → 0 This is the famous Minakshisundaram-Pleijel asymptotic expansion. The coefficients A k are spectral invriants of the Laplacian. They are often called global heat kernel coefficients or Hadamard-Minakshisundaram-De Witt-Seeley (HMDS) coefficients. This expansion is of great importance in differential geometry, spectral geometry, quantum field theory and other areas of mathematical physics, such as the theory of Huygens' principle, heat kernel proofs of the index theorems, Korteweg-De Vries hierarchy, Brownian motion etc. (see, for example, [28]). The general structure of the heat kernel coefficients can be described as follows [6,16,17,19]. We define symmetric tensors K ( j) of type (2, j) (that we call symmetric jets of order ( j − 2)) as symmetrized covariant derivatives of the curvature where the parenthesis denote the complete symmetrization over all indices included. Next, we define all possible scalar (orthogonal) invariants of the form where n = (n 1 , . . . , n m ) is a multiindex, \|n\| = n 1 + · · · + n m , and tr g denotes contraction with the metric to get a scalar. The index A labels all possible contractions. All invariants J A m,n have m curvatures and \|n\| derivatives of the curvatures. The classification of these invariants is a separate interesting problem. Then the local heat kernel coefficients have the grading according to the number of derivatives, that is, where C A n are some universal constants. Notice that the dimension of the space of invariants J A m,n grows as a factorial for large k. This makes a direct explicit calculation of the heat kernel coefficients for large k meaningless. The leading terms in the heat kernel coefficients with the highest number of derivatives were computed in [2,5,6,21]. They have the form where the dots denote the terms with less derivatives. Notice that there are only two independent invariants. We are interested in this paper in the opposite case, that is, the terms 20) with no derivatives at all. More precisely, we assume that the curvature is parallel, in other words, we restrict ourselves to locally symmetric spaces. This is a much more complicated case due to the presence of more invariants and algebraic constraints on the curvature tensor. This is an essentially non-perturbative calculation. Explicit results exist only in some particular cases. Symmetric Spaces We list below some well known facts from the theory of symmetric spaces [34,30,31]. A Riemannian manifold with parallel curvature is called a locally symmetric space. A complete simply connected locally symmetric space is called a globally symmetric space (or, simply, a symmetric space). A symmetric space is said to be of compact, noncompact or Euclidean type if all sectional curvatures are positive, negative or zero. A product of symmetric spaces of compact and noncompact types is called a semisimple symmetric space. A general symmetric space is a product of a Euclidean space and a semisimple symmetric space. It should be noted that our analysis is purely local. We are looking for a universal local generating function of the curvature invariants in the category of locally symmetric spaces, that adequately reproduces the asymptotic expansion of the heat kernel diagonal. This function should give all the terms without covariant derivatives of the curvature a k,k in the asymptotic expansion of the heat kernel, in other words all heat kernel coefficients a k for any locally symmetric space. It turns out to be much easier to obtain a universal generating function whose Taylor coefficients reproduce the heat kernel coefficients a k than to compute them directly. It is obvious that flat subspaces do not contribute to the heat kernel coefficients a k . Therefore, it is sufficient to consider only semisimple symmetric spaces. Moreover, since the coefficients a k are polynomial in the curvatures, one can restrict oneself only to symmetric spaces of compact type. Using the factorization property of the heat kernel and the duality between the compact and the noncompact symmetric spaces one can obtain then the results for the general case by analytic continuation. That is why we consider below only compact symmetric spaces. Holonomy, Isometry and Isotropy Groups Let e a = e µ a ∂ µ be a basis for the tangent space T x M. Let e a µ be the matrix inverse to e µ a , defining the dual basis ω a = e a µ dx µ in the cotangent space T * x M. We extend these bases by parallel transport along geodesics to get local frames on the tangent bundle and the cotangent bundles, and, therefore, on any tensor bundle, that are parallel along geodesics. We adopt the notation that the Latin indices from the beginning of the alphabet, a, b, c, . . . , label the tensor components with respect to this local frame and range from 1 through n. Then the frame components of any parallel tensor (such as the curvature tensor) are constant. The components of the curvature tensor of a compact symmetric space can be always presented in the form where E i ab , (i = 1, . . . , p), with p ≤ n(n − 1)/2 being a positive integer, is a set of antisymmetric n × n matrices and β ik is some symmetric nondegenerate positive definite p × p matrix. In the following the Latin indices from the middle of the alphabet, i, j, k, . . . , will be used to denote such matrices; they should be not confused with the Latin indices from the beginning of the alphabet which denote tensor components. The Latin indices from the middle of the alphabet will be raised and lowered with the matrix β ik and its inverse (β ik ) = (β ik ) −1 . Next we define the traceless n × n matrices D i , by ( The matrices D i are known to be the generators of the holonomy algebra, H, i.e. the Lie algebra of the restricted holonomy group, H, where F j ik are the structure constants of the holonomy group. The matrix β ik plays the role of the metric of the holonomy group with the scalar curvature The structure constants define the traceless p × p matrices F i , by (F i ) j k = F j ik , which generate the adjoint representation of the holonomy algebra, Now, we go back to the equation (2.21). It is an overdetermined system of partial differential equations. By taking the commutator of covariant derivatives we obtain the integrability condition of this equation By using the decomposition of the Riemann tensor introduced above we obtain from this equation This is the most important equation that holds only in symmetric spaces; it is this equation that makes a Riemannian manifold the symmetric space. To explore further consequences of the equation (3.7) we introduce a new type of indices, the capital Latin indices, A, B, C, . . . , which split acording to A = (a, i) and run from 1 to N = p + n. We define a symmetric nondegenerate positive definite N × N matrix (3.8) This matrix and its inverse (γ AB ) = (γ AB ) −1 will be used to lower and to raise the capital Latin indices. Next, we introduce a collection of new quantities C A BC with the non-vanishing components Let us also introduce rectangular p × n matrices T a by (T a ) j c = E j ac and the n × p matricesT a by (T a ) Then the matrices C A satisfy the commutation relations 11) and generate the adjoint representation of the Lie algebra G of some Lie group G with the structure constants C A BC . In more details, the commutation relations have the form which makes it clear that the holonomy algebra H is the subalgebra of the Lie algebra G. The matrix γ AB plays the role of the metric on the group G with the scalar curvature It can be expressed in terms of the scalar curvature R of the symmetric space M and the scalar curvature R H of the isotropy subgroup H It is well known that for compact symmetric spaces the group of isometries is isomorphic to the Lie group G defined in the previous section (for more details see [34,12]). The generators of isometries are the Killing vector fields (ξ A ) = (P a , L i ), which form the Lie algebra of isometries The vector fields L i form the isotropy subalgebra of the isometry algebra, which is isomorphic to the holonomy algebra H. Thus, a compact symmetric space M is isomorphic to the quotient space of the isometry group by the isotropy subgroup M = G/H. We will need the explicit form of the Killing vectors fields in symmetric spaces. Let us fix a point x ′ in the manifold M. Let d(x, x ′ ) be the geodesic distance between a point x and the fixed point x ′ and Then the derivative ∇ µ σ(x, x ′ ) is the tangent vector to the geodesic connecting the points x and x ′ at the point x. We let and K a b = R a cbd y c y d . where K is a n × n matrix with the entries K a b . Heat Semigroup Let k A be the canonical coordinates on the isometry group G so that each isometry is represented in the form exp k, ξ , where k, ξ = k A ξ A . Then the left-invariant vector fields on the isometry group G are given by and C(k) = k A C A . The metric on the group G is given by Then it is easy to see that the determinant of the metric is where \|γ\| = det γ AB . Let X 2 be the Casimir operator on the group G defined by Lemma 1 Let Φ(t; k) be a function on the isometry group G defined by where k, γk = γ AB k A k B . Then Φ(t; k) satisfies the heat equation and the initial condition Proof. First, we notice that the function \|G\| −1/4 satisfies the following equations and where (4.11) By using these equations we show that eqs. (4.7) and (4.8) hold by a direct calculation. One can show that the Laplacian on a symmetric space is simply the Casimir operator of the isometry group, 12) and, therefore, belongs to the center of the enveloping algebra, i.e., (4.14) Then Ψ(t) satisfies the heat equation with initial condition Ψ(0) = 1 , (4. 16) and, therefore, Ψ(t) = exp(t∆) . Proof: We have By using the previous Lemma we obtain Now, by integrating by parts we get Next, we show that and, therefore, Thus, the function Ψ(t) satisfies the eq. (4.15). The initial condition (4.16) for the function Ψ(t) follows from the initial condition (4.8) for the function Φ(t; k). Heat Kernel Diagonal The heat kernel diagonal can be obtained by acting by the heat semigroup on the delta-function, To be able to use integral representation for the heat semigroup (4.14) obtained above we need to compute the action of the isometries exp k, ξ on the deltafunction. Lemma 2 Let ω i be the canonical coordinates on the isotropy group H and (k A ) = (q a , ω i ) be the natural splitting of the canonical coordinates on the isometry group G. Then where D(ω) = ω i D i and \|η\| = det g ab . Proof. We choose the normal coordinates y a with the origin at x ′ defined above and consider the equation of characteristics Up to quadratic terms we obtain In particular, we find the Jacobian Then we have By noticing that f a (1, 0, ω) = 0 , (4.31) we finally obtain which proves the lemma. One remark is in order here. We implicitly assumed here that q = 0 is the only solution of the equation f a (1, q, ω) = 0 . This is not necessarily true. This is the equation of closed orbits of isometries and it has multiple solutions on compact symmetric spaces. However, these global solutions, which reflect the global topological structure of the manifold, will not affect our local analysis. In particular, they do not affect the asymptotics of the heat kernel. That is why, we have neglected them here. Now by using the above lemmas and the theorem we can compute the heat kernel diagonal. Corollary 1 The asymptotic expansion of the heat kernel diagonal as t → 0 is given by the formal asymptotic expansion of the function U diag (t) ∼ (4πt) −n/2 exp one can obtain now all heat kernel coefficients in terms of various contractions tr β F ⊗ · · · ⊗ Ftr g (D ⊗ · · · ⊗ D) (4.42) of the matrices D a ib and F j ik with the matrices β ik and g ab . All these quantities are curvature invariants and can be expressed directly in terms of the Riemann tensor. There is an alternative representation of the Gaussian average in purely algebraic terms. Let b j and b * k be operators acting on a Hilbert space, called creation and annihilation operators, that satisfy the following commutation relations where b * , βb * = β jk b * j b * k . This should be computed by socalled normal ordering, that is, by simply commuting the operators b j through the operators b * k until they hit the vacuum vector giving zero. The remaining non-zero commutation terms precisely reproduce the eqs. (4.40), (4.41).	2014-10-01T00:00:00.000Z	2006-05-30T00:00:00.000	{ "year": 2006, "sha1": "2867434427649a2afe59ee6103efe6f390984428", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "05dd0b84b6e3fc30925b447a4b3054f7d4559eb4", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics", "Physics" ] }
259526249	pes2o/s2orc	v3-fos-license	The evaluation of the delayed swollen breast in patients with a history of breast implants Breast implants, whether placed for reconstructive or cosmetic purposes, are rarely lifetime devices. Rupture, resulting from compromised implant shell integrity, and capsular contracture caused by constriction of the specialized scar tissue that normally forms around breast implants, have long been recognized, and remain the leading causes of implant failure. It is apparent, however, that women with breast implants may also experience delayed breast swelling due to a range of etiologic factors. While a majority of delayed seromas associated with breast implants have a benign etiology, this presentation cannot be ignored without an adequate workup as malignancies such as breast implant associated anaplastic large cell lymphoma (BIA-ALCL), breast implant associated diffuse large B-cell lymphoma (BIA-DLBCL), and breast implant associated squamous cell carcinoma (BIA-SCC) can have a similar clinical presentation. Since these malignancies occur with sufficient frequency, and with sometimes lethal consequences, their existence must be recognized, and an appropriate diagnostic approach implemented. A multidisciplinary team that involves a plastic surgeon, radiologist, pathologist, and, as required, surgical and medical oncologists can expedite judicious care. Herein we review and further characterize conditions that can lead to delayed swelling around breast implants. Introduction Breast implants remain the most common implanted medical devices in plastic surgery operating rooms. Over 350,000 women underwent cosmetic breast augmentation in the Unites States in 2021, making it the second most popular aesthetic procedure next to liposuction (1). Breast implants also represent the most common form of post-mastectomy reconstruction for the 1 in 8 women in the United States who will be diagnosed with breast cancer during their lifetimes. Though breast implants are approved by the Federal Drug Administration (FDA) for the purposes of breast augmentation or reconstruction, they are not without risk. This has come to light more recently with the discovery of breast implant associated anaplastic large cell lymphoma (BIA-ALCL) and the considerable attention it has garnered over the last decade. BIA-ALCL is a rare T-cell lymphoma that most often presents as a delayed seroma surrounding a textured breast implant. A mass originating in the implant capsule may develop concurrently, or as the sole finding, with most cases presenting 8-10 years postimplantation, with earlier and later cases also reported. The first case of anaplastic T-cell lymphoma in proximity to a saline-filled breast implant was described as early as 1997, although recent literature identifies a possible earlier case description in 1996 (2)(3)(4). The first FDA safety communication regarding breast implants in 2011 and the recognition of BIA-ALCL as a separate category of malignancy by the World Health Organization (WHO) in 2016 sparked heightened awareness (5). Shortly thereafter, the National Comprehensive Cancer Network (NCCN) published guidelines for the diagnosis and treatment of BIA-ALCL, emphasizing early intervention and surgical treatment (6). As of April 1, 2022, the FDA has received a total of 1,130 United States and global medical device reports (MDRs) of BIA-ALCL (7). Despite its recent notoriety, BIA-ALCL represents only a small fraction of delayed complications associated with breast implants. Specifically, making an accurate diagnosis when a patient presents with delayed breast swelling can be challenging. Development of a swollen breast one or more years after implantation carries with it a lengthy differential diagnosis representing a wide range of potential morbidity and mortality. In addition to BIA-ALCL, other malignancies associated with breast implants have been recognized, including breast implant-associated squamous cell carcinoma (BIA-SCC) and breast implant associated diffuse large B-cell lymphoma (BIA-DLBCL). In this review, we focus on the etiology of a swollen breast that develops in a delayed manner following placement of breast implants. We expand upon each of the breast implant-associated malignancies, including a discussion on the varied presentation, etiology, diagnostic algorithm, findings, and treatment modalities for each disease. We also review common findings and treatment modalities for benign causes of delayed breast swelling, including infection, benign seroma, trauma, hematoma, double capsule, and capsular contracture. Diagnostic evaluation The NCCN has recently standardized the evaluation of the delayed swollen breast in patients with a history of implants ( Figure 1) (6). Patients should first be assessed with ultrasound to assess for fluid collection, breast masses, and lymphadenopathy. Complete ultrasound evaluation should include the implant; chest wall; axillary, internal mammary, and supraclavicular lymph nodes; and contralateral breast implant. If ultrasound is equivocal, breast magnetic resonance imaging (MRI) may aid in diagnosis for fluid collections or soft tissue masses. Fine needle aspiration is the standard for sampling periprosthetic fluid collections. Ultrasound guidance is recommended to obtain an appropriate sample and avoid implant injury. Any suspicious masses found during initial imaging should be biopsied and sent for histopathologic analysis. Specimens should be sent for cytology to evaluate cell morphology, immunohistochemistry (IHC) for immune cell markers, and flow cytometry to evaluate cells within the specimen. Cytologic and cell block preparations are utilized to identify neoplastic cells in aspirated effusion fluid. If an implant-associated malignancy is established, a multidisciplinary team including pathologists, medical and radiation oncologists, surgical oncologists, radiologists, plastic surgeons, and the patient should be leveraged to help stage and treat disease. Pre-operative laboratory studies should include a comprehensive metabolic panel, complete blood count with differential, coagulation studies, and lactate dehydrogenase (6). Hepatitis B testing should be performed if the patient may need chemotherapy. A preoperative positron emission tomography computed tomography (PET/CT) scan is recommended for evaluating associated capsular masses, chest wall involvement, lymph node spread, and organ metastases (8). One must keep in mind that a PET/CT performed for the first three months after surgical intervention is unreliable due to post-operative inflammatory changes within the tissue (6). Treatment modality is determined by the malignancy and extent of disease (Table 1). Breast implant associated large cell lymphoma BIA-ALCL is a CD30-positive T-cell lymphoma that arises around textured breast implants. This disease is distinct from a primary breast lymphoma, which is typically a B-cell lymphoma that arises within the breast parenchyma. The etiology of BIA-ALCL is unknown but likely triggered by chronic inflammation. Implant texturization is indisputably a driver, while host genetic factors, and time likely play a role in tumorigenesis. Bacterial infection and biofilm formation, specifically from Raltosonia spp, or perhaps the lipopolysaccharide coat of Gram-negative bacteria was thought to a play a primary role in the pathologic inflammation leading to BIA-ALCL (9). However, more recent research suggests other potential inciting events that may obscure the exact etiologic pathway, including mechanical stress, implant toxins, and surface tribology (10-14). Curiously, a relative attenuation of circulating T-helper cells may occur in the first couple days following placement of a textured, but not smooth breast implant (15). While each of these studies proposes a different "trigger," chronic inflammation is the common thread, and is the most likely facilitator of malignant transformation to BIA-ALCL (16). All verified cases of BIA-ALCL with complete implant history have been exclusively discovered in patients with a history of textured implants, many of which have been recalled, or in some countries banned, due to this association (26-28). Of the 1,130 medical device reports of ALCL, 37 cases were found in patients with smooth implants. However, these patients either previously had textured implants or insufficient implant history (29). By contrast, implant fill (saline or silicone) and reason for implantation (augmentation or reconstruction) has no defined relationship to BIA-ALCL. Therefore, a thorough surgical history including all previous implanted devices should be obtained in all patients presenting with delayed seroma. The incidence of BIA-ALCL in patients with a history of textured implants varies widely, reported as high as 1:355 patients to as low as 1:40,000 patients (26, [30][31][32]. Age of diagnosis is also varied, with an median of 53 years old (ranging 24-90 years old), while time from implantation to diagnosis is consistently prolonged, with an average timeframe of 7-10 years (6,16,32,33). BIA-ALCL manifests as a mass or, more commonly, a rapidly enlarging periprosthetic fluid collection surrounding an implant years after implantation (34). Other local and systemic symptoms reported in patients diagnosed with BIA-ALCL include pain, lymphadenopathy, skin rash, fevers, and capsular contracture (6,35). In the case of advanced disease, BIA-ALCL typically spreads to the ipsilateral axillary nodes, with supraclavicular, internal mammary, or mediastinal nodal involvement occurring less commonly (36). Intraoperative findings may demonstrate intracapsular periprosthetic fluid collection containing fibrinous material with or without extracapsular spread. Herein, we describe a patient with disseminated BIA-ALCL that presented with a swollen left breast (Figure 2A). PET scan reveals T4 lesion with extracapsular disease on the left side ( Figure 2B), mandating excision of adjacent axillary soft tissues in conjunction with en bloc capsulectomy ( Figure 2C). Importantly, this patient has a prophylactic total capsulectomy on the contralateral right side (ie. entire capsule and breast implant as a single unit, but not adjacent margin of soft tissue) and was noted to have an occult T1 luminal capsular BIA-ALCL. We recommend contralateral prophylactic total capsulectomy in patients undergoing therapeutic en bloc capsulectomy because of a 1-3% risk of bilateral disease (6,(37)(38)(39). The diagnosis of BIA-ALCL is made through cytologic and immunohistochemical analysis of periprosthetic fluid. The neoplastic cells of BIA-ALCL are large, pleomorphic cells with anaplastic morphology. The cell nuclei are large, oval or multilobulated, with dense chromatin, and have prominent nucleoli and frequent mitoses (40). Commonly described in association with ALCL are the "hallmark" cells, which have horseshoe-or kidney-shaped nuclei and are found in a majority of cases (41). BIA-ALCL is universally CD30 positive and ALK negative. Expression of CD30 is a fundamental finding in BIA-ALCL; however, it is important to note CD30 expression alone is not diagnostic for BIA-ALCL, as it is nonspecifically expressed by benign inflammatory cells (6,(42)(43)(44). Likewise, because other forms of ALCL also lack ALK expression, ALK negative IHC alone does not establish a diagnosis of BIA-ALCL. Additional biomarkers may be required to establish the diagnosis and exclude other malignancies, including CD2, CD3, CD4, CD5, CD7, CD8, and CD45 expression. The Lugano modification of the Ann Arbor staging system is traditionally used to stage all forms of Non-Hodgkin lymphoma (45). However, nearly all BIA-ALCL cases are staged as Stage IElymphoma with single extranodal involvement, or Stage IIEextranodal disease with local nodal involvement. Further, this staging system does not account for capsular involvement (46). Thus, MD Anderson Cancer Center developed a tumor, lymph node, metastasis (TNM) solid tumor staging system that was adapted by the NCCN to further characterize BIA-ALCL (Table 2). While many reported cases are categorized as stage I, over 25% of BIA-ALCL cases have extracapsular involvement at the time of diagnosis (6,33,46). This staging system has demonstrated improved efficacy in predicting survival and recurrence compared to the Ann Arbor staging system (46). Following diagnosis and pre-operative imaging, the mainstay of treatment for BIA-ALCL is total capsulectomy ( Figure 2C) (6,46). There is no proven role for mastectomy, sentinel lymph node biopsy, or axillary node dissection (6). Surgery should always be conducted with an en bloc capsulectomy to remove the implant in continuity with the capsule, any associated extracapsular masses, and a margin of contiguous healthy tissue. Complete surgical excision with negative margins is associated with long-term, disease free survival, and may be adequate for disease localized to the capsule (Stage IA-IIA) (46,47). However, disease recurrence is nearly 3-fold higher in Stage II and Stage III disease, and is more likely with incomplete resection, partial capsulectomies, or positive margins. The use of adjuvant therapy for BIA-ALCL is limited to patients with residual or disseminated disease. The NCCN recommends radiation therapy of 24 to 36 Gray (Gy) for any local residual disease or unresectable masses due to chest wall involvement (6). In patients with disseminated disease, first-line systemic therapy should include the combination regimen CHOP (Cyclophosphamide, Adriamycin, Vincristine, and Prednisone) and brentuximab vedotin, a CD30 monoclonal antibody that has recognized survivability benefits when treating peripheral T-cell lymphomas (48)(49)(50)(51)(52). Worldwide, there have been a total of 59 reported deaths related to BIA-ALCL up until April 1, 2022, with more under review (33). Overall survivability of early stage BIA-ALCL is reported as 94% at 3 years and 91% at 5 years, respectively (46). Later stage of presentation is associated with decreased survivability and higher risk of recurrence (46). Patients who reach remission should be monitored for recurrence every 3 to 6 months for 2 years, and radiologic imaging with CT or PET scan should be considered every 6 months for 2 years due to the high 3-year recurrence risk at all stages (6). Breast implant associated squamous cell carcinoma Breast implant associated squamous cell carcinoma (BIA-SCC) is a rare but aggressive malignancy that originates from the breast implant capsule. This entity was first proposed by Paletta and colleagues in 1992, who reported a case of BIA-SCC in a patient who underwent breast augmentation with silicone implants 16 years prior (53). Since this initial case report nearly 30 years ago, there have been 16 verified cases of BIA-SCC according to the American Society for Plastic Surgeons (ASPS), with more cases under review including one we present herein (34,(54)(55)(56)(57)(58)(59). Though rare, this malignancy has garnered attention recently due to its aggressive nature and high mortality. The origin of the squamous cell epithelium in this malignancy is unclear. Similar carcinogenic processes have been described with foreign bodies in other tissues, including bullet wounds and dental or orthopedic implants (60-63). It is proposed that ductal epithelium can be displaced at the time of pocket implantation, resulting in squamous epithelialization of the breast implant capsule (54,55,59). Additionally, macrophages and lymphocytes infiltrate the breast pocket and release cytokines to wall-off the foreign body. This may become exaggerated over a protracted course, such as in the case of a permanent breast implant. It is well-known that chronic inflammation can lead to an imbalance in inflammatory and apoptotic cell signaling pathways, resulting in tissue metaplasia (64). Such is the case for intestinal metaplasia of the esophagus and stomach in response to chronic acid exposure. Similar concepts may be applied to BIA-SCC: chronic inflammation and fibrosis surrounding the breast implant can inadvertently stimulate metaplastic squamous epithelium production within the capsule, a precursor to squamous cell carcinoma. Age of diagnosis is highly variable, ranging from 40-81 years old (34,(54)(55)(56)(57)(58)(59). The time interval from initial breast implant surgery to BIA-SCC diagnosis ranges from 11-41 years, and while consistently protracted, does overlap to an extent with the timeframe during which BIA-ALCL may develop. This dysplastic process, which may be indolent in nature, should therefore remain on the differential diagnosis along with BIA-ALCL when assessing patients who present with new breast swelling in the context of a breast implant multiple years after implantation. Presentation of BIA-SCC includes unilateral breast pain, erythema, and fluid collection. Patients may also present with some degree of capsular contracture and implant malposition, though this is not a uniform finding at time of diagnosis. Of the 16 reported cases, eleven occurred following breast augmentation and five occurred following breast reconstruction (34,(54)(55)(56)(57)(58)(59)65). Cases have been reported in all implant types, including saline and silicone implants with either textured or smooth surfaces (54-59, 65, 66). However, in a majority of these reports, implant surface and device history are not reported and so there is insufficient information to determine whether it has etiologic relevance. Future cases including completely documented device history will help determine if an association with implant texturing exists. The ASPS recommends FNA and cytology of any delayed seroma prior to surgical intervention (67). Most cases reported in the literature have not involved FNA of seroma fluid prior to operative intervention, as many of these cases were reported prior to new diagnostic recommendations (54-59, 65, 66). Seroma aspirates confirming BIA-SCC express epithelial carcinoma marker CK 5/6 and squamous cell transcription factor p63, and contain squamous cells and keratin. We report a case of a patient presenting with unilateral breast swelling 16 years following cosmetic breast augmentation with a macrotextured implant (Figures 3A, B). To properly evaluate for BIA-SCC, seroma fluid aspirate ( Figure 3C) should be sent for IHC looking for CD30 and ALK to evaluate for BIA-ALCL, along with CK 5/6 and p63. Flow cytometry should be employed to look for T-cells, squamous cells, and keratin. One should note that seroma aspirate is not completely comprehensive for detecting malignancy, and tissue biopsy is commonly required to achieve diagnosis. For the case we present herein, initial seroma aspirate failed to reveal malignancy; squamous cell carcinoma was only found on pathology after complete capsulectomy was performed. Pre-operative imaging facilitates surgical planning. Ultrasound is commonly used to guide aspiration of seroma fluid for analysis ( Figure 3D). Breast MRI with and without contrast can be employed to identify any masses. Findings consistent with BIA-SCC will demonstrate an ill-defined mass arising from the breast capsule, with possible extent into the chest wall. PET-CT should be employed prior to intervention to appropriately determine extent of disease ( Figure 3E). Intraoperative findings of BIA-SCC include fungating breast capsule masses with granulomatous and keratinized debris contained within a viscous, turbid seroma fluid. In a majority of reported cases, this malignancy arises from the posterior aspect of the implant capsule, with spread of keratinaceous material into the pectoralis muscle and axillary tissue ( Figures 3E-G) (53)(54)(55)(56)(57)59). Capsules are commonly found intact with a thickened appearance and yellow hue. Histology from these capsules and associated granulomatous material demonstrates invasive keratinized squamous cell carcinoma and metaplasia with evidence of acute on chronic inflammation. These findings further support the theory that chronic inflammation stimulates malignant transformation. The overall prognosis for BIA-SCC is grim, with a 6-month mortality rate of 43.8% (67). Early diagnosis and treatment can have life-lengthening benefits. Current treatment recommendations are likely to evolve as more diagnoses are described. The current treatment recommendation is surgical with en bloc capsulectomy and radical mastectomy (67). Surgeons should be aggressive, and should not hesitate to resect chest wall or axillary contents if there is suspicion for invasive malignancy. This is of utmost importance when treating BIA-SCC, as incomplete resection is associated with aggressive recurrence and increased mortality. There does not appear to be a role for adjuvant chemotherapy or radiotherapy, as the malignancy has thus far demonstrated little to no therapeutic response to either treatment modality. Breast implant associated diffuse large B-cell lymphoma Lymphomas associated with breast implants are rare and commonly have a T-cell origin. In a small minority of cases, though, delayed unilateral breast swelling years after breast implant placement ( Figure 4A) have been attributed to other lymphomas. Breast implant associated-B-cell lymphomas are characterized by a more heterogenous cellular origin that includes diffuse large B-cell lymphoma, follicular lymphoma, primary cutaneous lymphoma, intravascular large-cell lymphoma, splenic marginal zone lymphoma, and plasmablastic lymphoma (68-82). Of these diagnoses, breast implant associated diffuse large B-cell lymphoma (BIA-DLBCL) has been most commonly described in the literature, is associated with a delayed seroma ( Figure 4B), and has been associated, in several instances, with Epstein Barr Virus (EBV). Due to the broad array of reported B-cell lymphomas, patient presentation is wide-ranging. Of the 28 reported cases, a majority of patients reported breast pain and swelling or palpable mass, while fewer presented with capsular contracture, B symptoms, hepatosplenomegaly, and lymphadenopathy (68)(69)(70)(71)(72)(73)(74)(75)(76)(77)(78)(79)(80)(81)(82)(83). Herein, we present two cases, one of BIA-DLBCL ( Figure 4) and one of breast implant associated follicular lymphoma ( Figure 5). Age and time from initial implantation to diagnosis ranges from 34 -83 years old and 6 -44 years, respectively. Despite these differences, there are similarities, most notably the association with textured, silicone breast implants (70,(81)(82)(83). There have been fourteen reported patients with diffuse B-cell lymphoma, twelve of which were found to be positive for Epstein-Barr Virus. EBV-positive DLBCL has been implicated in states of immunosuppression and chronic inflammation, categorized by the WHO as diffuse large B-cell lymphoma associated with chronic inflammation (DLBCL-CI), of which pyothorax-associated lymphoma is the prototype (84). The presence of longstanding chronic inflammation associated with an indwelling implant may result in proliferation of EBV-transformed B-cells, such as in the case of DLBCL-CI (85). However, the fibrinous material associated with BIA-DLBCL lends itself to a diagnosis similar to fibrinassociated DLBCL (FA-DLBCL), an indolent form of EBVpositive large B-cell lymphoma that has been categorized by the WHO as a clinically distinct subtype of DLBCL-CI (86). This form of lymphoma has been reported in association with pathologic debris surrounding atrial myxomas, endovascular graft thrombi, metallic prosthesis, and pseudocysts. These cases have a much more favorable prognosis in comparison to DLBCL-CI, and may even represent a form of EBV-positive lymphoproliferative disease rather than a lymphoma (87). However, not all cases of implant-associated DLBCL are EBV-positive, including the case we present in Figure 4. Further case collection and pathologic investigation is required to characterize this novel lymphoma. Complete physical evaluation and diagnostic workup for the delayed swollen breast should be obtained to characterize this malignancy. Most cases are localized to the implant capsule, though few have been found to be invasive, mass-forming lymphomas. Gross pathologic findings of DLBLCL exhibit tan, thickened implant capsules with granular, gritty inner lining following en bloc capsulectomy ( Figure 4C). Reported microscopic examination demonstrates focal foreign body giant cell reactions and lymphoplasmacytic aggregates of pleomorphic lymphoid cells, which may have atypical nuclei with numerous mitotic figures, heterogeneous chromatin pattern, and/or prominent nuceloli (21,70). Immunohistochemical analysis exhibits a wide array of B-cell expression profiles, with inconsistent staining for B-cell markers CD20, CD19, CD79a, PAX-5, and BCL-6 ( Figure 4D) (70,72,83). En-bloc capsulectomy ( Figure 4C) and implant removal has proven to be adequate in treating localized breast implantassociated B-cell lymphomas (81,82). We have also used this approach to manage a case of follicular cell lymphoma associated with a breast implant capsule. Distinguishing this case as a follicular cell lymphoma were the findings of lymphoid aggregates seen in the breast implant capsule ( Figure 5A), atypia with irregular contours (Figure 5B), few T-cells ( Figure 5C), predominance of CD20+ Bcells ( Figure 5D), closely packed and expanded follicles without mantle zones ( Figure 5E), that were comprised of B-cells ( Figure 5F). Follicles were BCL2 ( Figure 5G) and BCL6 ( Figure 5H) positive. Given the few case reports and overall indolent nature of the disease, a consensus on the need for adjuvant chemoradiation or radiologic disease monitoring has not been reached. Regular breast imaging has been advocated to evaluate disease recurrence. Long-term follow up data is needed to determine disease prognosis and survivability benefit for each of these treatment modalities. Benign delayed seroma Benign delayed seromas are usually defined as serous fluid collections that develop around an implant more than one year after implantation. In accordance with the NCCN guidelines (6), these are diagnosed primarily by ultrasound, but breast MRI may be required in equivocal cases or when a greater level of sensitivity and specificity are warranted. It has been theorized that benign delayed seromas are the result of trauma, hematoma, subclinical infection, or implant rupture, though they can also occur without an identified precipitating cause. It has been theorized that benign delayed seromas are the result of trauma, hematoma, subclinical infection, or implant rupture, though they can also occur without an identified precipitating cause. Benign delayed seromas are rare events, occurring in less than 1% of subjects in large multicenter trials (88)(89)(90). Prior to the discovery of BIA-ALCL, a majority of early case reports describing delayed seromas occurred in patients who previously had macrotextured breast implants (12, 89,91,92). It is conceivable that early case reports of delayed seromas were undiagnosed BIA-ALCL, as all diagnosed patients had a history of textured implants. It is important to note that up to 10% of delayed fluid collections associated with breast implants are malignant on further diagnostic evaluation (42,88,89). Though similar in physical presentation, the pathogenesis of benign delayed seromas and BIA-ALCL are distinct. BIA-ALCL develops as a malignant effusion due to increased vascular permeability resulting from cellular production of interleukins and elevated oncotic pressure caused by the high cellularity and protein content of the fluid. Benign delayed seromas have a wide variety of purported etiologies, including an idiopathic one. Though associated with trauma and subclinical infection, benign delayed seromas typically lack a microbiologic or cytologic biomarker (92). The detection of CD30 positive cells on IHC aids the diagnosis of BIA-ALCL and helps differentiate a benign delayed seroma from a malignant one, drastically changing overall management (43). Treatment of benign delayed seroma varies based on surgeon and patient preferences, ranging from serial aspirations to complete capsulectomy. In their multicenter retrospective review of delayed seromas, Spear et al. expounded upon a graduated approach to treating delayed seromas that included antibiotics, serial aspirations, drain placement, and surgical resection (90). A majority of patients required surgical intervention to reach full resolution, while 28.5% of patients were able to be successfully managed with aspiration or antibiotics alone. In their series, all aspirate cultures were negative for planktonic bacteria identifiable with standard culture techniques. They did not perform advanced biofilm detection techniques such as 16S rRNA sequencing, immunohistochemistry for bacteria-specific probes, or scanning electron microscopy. Likewise, routine CD30 immunohistochemistry testing was not performed as BIA-ALCL was not a well-known diagnosis at this time. We believe a similar algorithm can be used to treat delayed seromas once they are determined to be benign and non-infectious. Double capsule Another rare but benign etiology of delayed breast swelling is the double capsule. This occurs when the inner capsule envelope adheres to the implant surface while a distinct outer capsule adheres to surrounding tissues, divided by an intercapsular space that may contain a seroma ( Figure 6A) (11,12). Typically, the outer capsule can be dissected from the surrounding soft tissues while the inner capsule remains intimately associated to the textured device ( Figure 6B) unless it is deliberately peeled off the implant ( Figure 6C). Hall-Findlay was the first to report on her experience with double capsule formation (12). Initial patient presentation is highly variable, including persistent seroma, capsular contracture, bottoming out, and asymmetry (12,93,94). Similar to BIA-ALCL, this phenomenon primarily develops in patients with textured implants. It is theorized that macrotextured surfaces induce some adherence of the capsule to the implant, which can result in mechanical shearing of the implant capsule from the implant surface and seroma formation (93, 95). Bacterial biofilms may contribute to double capsule formation by weakening capsule strength and facilitating extracellular matrix delamination and double-capsule formation (96). Diagnostic evaluation should follow the NCCN guidelines. The only surgical modality efficacious in treating double capsule includes inner capsulectomy FIGURE 6 Case of double capsule. (A) Patient with a macrotextured saline breast implant placed for cosmetic reasons 7 years prior presents with a rapidly developing seroma of the right breast. The forcep reflects the outer capsule and the hemostat penetrates the inner capsule that is intimately associated with the implant. Clear fluid was identified between these two layers in situ. (B) The outer capsule is free from the inner capsule that is in continuity with the textured implant shown here. (C) The inner capsule now dissected free from the implant surface. Infection When examining the delayed swollen breast, implant infection must be taken into consideration. Reported incidence of implant infection ranges from 0-2.5% following breast augmentation, and up to 35% following breast reconstruction after mastectomy (97)(98)(99)(100). Acute infections typically occur within weeks to months following implant placement. Patients commonly present with breast pain, drainage, and erythema, and systemic symptoms including fevers, nausea, and vomiting. Possible sources of infection include the patient's skin or breast microbiota, contaminated implant or irrigation fluid, surgical manipulation, and hematogenous spread. In addition to causing acute infections, many of these bacteria have evolved to adhere to implant surfaces, forming assemblages of surface-adherent bacteria encapsulated in extracellular polymers known as biofilms. The formation of these biofilms around an implant are implicated in subclinical infections, capsular contracture, and other systemic symptoms. Subacute infections, which can occur months to years after surgery, have a more indolent course, making them more difficult to distinguish from other diagnoses. Patients may present with chronic pain, persistent swelling and drainage, wound healing problems, or implant migration. Hematogenous spread of bacteria from distant sites play a crucial role for developing late onset breast implant infections. Initial evaluation for breast implant infection should rely heavily on the patient history and physical exam findings. Full history of recent illnesses or infections and surgical interventions should be reviewed. Providers should look for subtle signs of infection including fevers, nausea and vomiting, and new breast pain, erythema, or drainage. Laboratory tests should include a comprehensive metabolic panel and complete blood count. Diagnostic evaluation for subacute infections should include complete breast ultrasound to evaluate for drainable fluid collections, cultures, and bacterioscopic smear test may be considered to confirm and characterize infection. Malignancy should also be ruled out with imaging should new breast masses or lymphadenopathy be identified, and FNA performed of seroma fluid with histologic examination. While some investigators report successfully salvaging periprosthetic implant infections using negative pressure therapy with or without irrigation (101), treatment traditionally warrants surgical washout and implant explantation. Traumatic hematoma Early hematoma directly following breast implant placement, whether it be for reconstructive or aesthetic purposes, is a welldocumented post-operative complication occurring in 0.6-10.3% of all cases (102)(103)(104). Peri-prosthetic late hematomas that occur more than 6 months after surgery are considered a rare complication, many of which have unknown causes. Chest trauma is an acute inciting factor that can result in spontaneous capsular sheering and hematoma formation. Likewise, chronic inflammation or systemic therapies, such as corticosteroids, chemotherapy, or systemic anticoagulation, can damage peri-capsular arteries and lead to late capsular hematoma (105)(106)(107). In the absence of a clear inciting event, it is thought that mechanical friction between the prosthesis and the highly vascular capsule, with a consequent capsule microfractures, may play a role in delayed hematoma (104). In evaluating patients for delayed hematoma, MRI and ultrasonography may be performed, but are not helpful in distinguishing hematoma from implant rupture, and may lead to false positives. Treatment includes hematoma evacuation and implant exchange with or without capsulectomy in patients who want to maintain their breast size. Hematoma is also a risk factor for the development of capsular contracture and should be adequately addressed to avoid this latent complication. Breast cancer With nearly one in eight women diagnosed with breast cancer within their lifetime, the possibility that a new breast mass or fluid collection in a patient with breast implants is related to primary or recurrent breast cancer can occur (108). Thus, physicians must have a high index of suspicion for breast cancer when evaluating the delayed swollen breast. In particular, invasion of dermal lymphatics in inflammatory breast cancer can lead to rapid swelling, erythema, and pitting edema thus presenting as delayed swelling of the breast in a women with previous implant-based breast augmentation (108). Therefore, all patients who present with delayed breast swelling should undergo diagnostic mammography, breast MRI, and/or complete breast ultrasound to evaluate for cancerous lesions (109-111). Breast MRI is an important imaging modality that can be employed in younger patients with highly dense breast tissue. Any suspicious masses or should be evaluated by a radiologist and surgical oncologist to determine need for further evaluation and treatment. Other considerations In addition to complications associated with breast implants, patient medical comorbidities must be taken into consideration when evaluating the swollen breast. Previous case reports have demonstrated that unilateral breast edema may be a manifestation of congestive heart failure, specifically in elderly patients (112)(113)(114). Patients will present with signs and symptoms of heart failure on physical exam, including jugular venous distension, pretibital pitting edema, and pulmonary congestion. Chest radiographs will demonstrate cardiomegaly and pulmonary edema, and diagnosis will be made by decreased ejection fraction on electrocardiography. Conclusion The delayed presentation of a swollen breast in patients with a history of breast implants is a diagnostic challenge to all physicians. Though many cases are benign, one must carefully follow the NCCN guidelines to properly evaluate for the malignancy, including BIA-ALCL, BIA-SCC, and BIA-DLBCL, and recurrent or new primary breast cancer. All cases of malignancies associated with breast implants should be reported to the FDA's Manufacturer and User Facility Device Experience (MAUDE) database and the device manufacturer. To improve our understanding of these rare cancers, cases of breast implant malignancy from the United States should be reported to the PROFILE registry (https://plasticsurgery. formstack.com/forms/profile_case_submission) and equivalent registries in other countries. Moreover, genomics continue to play a critical role in the diagnosis and identification of targeted therapies to more effectively manage both breast cancers and breast-implant associated malignancies (24,25). Further, though these malignancy are rare, all patients receiving breast implants should be counseled pre-operatively on the risk of each of these cancers, and particularly BIA-SCC due to its severity and mortality (67). Author contributions TM takes full responsibility for the integrity and accuracy of this review articles. All authors approve the final article and agree to be accountable for all aspects of the work. Concept and design: GK, AK, TM. Drafting of the manuscript: GK, AK, TM. Critical revisions of the manuscript for important intellectual content: all authors. Supervision: TM. Acknowledgments Authors are grateful to Jill Guess for administrative assistance with pathology slides and Michelle Bingaman for collection of patient consent forms. Conflict of interest TM receives grant funding from Sientra and RTI Surgical and product development royalties from RTI Surgical. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.	2023-07-11T16:05:37.912Z	2023-07-05T00:00:00.000	{ "year": 2023, "sha1": "3f984f2e148974a5a31aa0ab3aba0acfd18a8de3", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fonc.2023.1174173/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "aa0aa0de72d840854be9cb25baa7558ea8fa4027", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
119227640	pes2o/s2orc	v3-fos-license	Degeneracy of doubly heavy baryons from heavy quark symmetry The spectroscopy of the doubly heavy baryons including different heavy quarks is studied based on the heavy quark symmetry of QCD. We point out that, when the two heavy quarks are in $S$-wave, these baryons with a certain spin $j_l$ of the light cloud can be classified into two sets: a heavy quark singlet with total spin of $j=j_l$ and a heavy quark multiplet with $j= (j_l+1) , j_l ,\ldots \vert j_l-1\vert$, all the baryons in these two sets have the same mass and, the baryons with the same quantum numbers in these two sets do not mix with each other. We finally point out that the strong decay of the first excited baryon with light spin $j_l = 1/2$ to the ground state and one-pion is determined by the mass splitting through the generalized Goldberger--Treiman relation. The spectroscopy of the doubly heavy baryons including different heavy quarks is studied based on the heavy quark symmetry of QCD. We point out that, when the two heavy quarks are in S-wave, these baryons with a certain spin j l of the light cloud can be classified into two sets: a heavy quark singlet with total spin of j = j l and a heavy quark multiplet with j = (j l + 1), j l , . . . \|j l − 1\|, all the baryons in these two sets have the same mass and, the baryons with the same quantum numbers in these two sets do not mix with each other. We finally point out that the strong decay of the first excited baryon with light spin j l = 1/2 to the ground state and one-pion is determined by the mass splitting through the generalized Goldberger-Treiman relation. The physics of heavy hadrons has become a hot subject in particle and nuclear physics because of the observations of a large amount of such states during the last decade in scientific facilities. With the collection of more data and updating of facilities, more and more states must be observed in the present and upcoming facilities such as BESIII, LHCb and Belle II. The observation of the hidden charm pentaquark state at LHCb [1] strongly indicates that it is the time to study baryons with two heavy quarks, the doubly heavy baryons (DHBs). In this work, we will focus on the spectroscopy of the DHBs with different heavy quarks based on the heavy quark symmetry (see, e.g. Ref. [28] for a review). In a DHB with different heavy quarks, say Q and Q ′ , the two heavy quarks in the S-wave can form a spin singlet and a spin triplet. For simplicity, we write the spin singlet and the spin triplet asΦ (QQ ′ ) andΦ being the color anti-triplet. Due to the spin-flavor symmetry in the heavy quark limit, Φ (QQ ′ ) andΦ (QQ ′ µ have the same mass, i.e., The interaction between the heavy diquarks, Φ (QQ ′ ) andΦ (QQ ′ ) µ , and gluons can be easily written down by considering that both Φ (QQ ′ ) andΦ (QQ ′ ) µ are the color anti-triplets. By taking the heavy quark limit, the effective Lagrangian for the diquarks is expressed as where v ν is the velocity of the diquarks, G ν is the gluon field and g is the gauge coupling constant of QCD. Effective Lagrangian (2) implies that the two diquarks have the same interaction with the gluon which combines them to a light degree of freedom ("Brown muck") to form two types of heavy baryons. Now, we are in the position to study the mass relation of the DHBs with quark content QQ ′ q where q stands for a light quark constituent. We first consider the DHBs in the ground state. In such a case, we schematically write the DHBs as D Q ≡ Φ (QQ ′ ) q and D µ Q ≡Φ (QQ ′ )µ q, where q symbolically denotes the Brown muck in the ground state which carries the spin-parity j P l = 1 2 + . Since the scalar diquarkΦ (QQ ′ ) carries spin zero, the spin-parity of the D Q is j P = 1 2 + . Therefore D Q is a heavy quark singlet. On the other hand, since the axial-vector diquarkΦ (QQ ′ ) µ carries spin one, the spin-parity of the D µ Q is either j P = 1 2 + or 3 2 + which forms a heavy quark doublet. With respect to Eq. (2), we see that the singlet D Q and doublet D µ Q have the same mass, i.e., Furthermore, the states with the same quantum number, explicitly those with j P = 1 2 + , in D Q and D µ Q cannot mix due to the difficulty of the heavy quark spin flipping. Then, we arrive at the conclusion that, in the heavy quark limit, the ground states of the DHBs with different heavy quarks form a heavy quark singlet and a heavy quark doublet which are classified by the total spin of the heavy diquark included in them and the DHBs in these two sets have the same mass. Next, we consider DHBs with the first orbital excitation with relative angular momentum between the light quark and heavy diquark source l = 1. In such a case, the quantum numbers of the Brown muck could be j P l = 1 2 − , respectively. In analogy to the discussion made to the ground states, the baryons in the heavy quark singlet N Q and those in the heavy quark triplet N µ Q have the same mass. When we combine the j P l = 3 2 − component to the heavy diquarks one gets a heavy quark singlet T µ Q with quantum numbers j P = 3 2 − and a heavy quark triplet T ′µ Q with quantum numbers j P = 5 2 − , 3 2 − , 1 2 − . Again, from the effective Lagrangian (2) we conclude that all the baryons in the heavy quark singlet T µ Q and heavy quark triplet T ′µ Q have the same mass and the two baryons with quantum numbers j P = 3 2 − decouple from each other. From the above discussion, we arrive at our conclusion that, for DHBs with total light spin-parity j P l , they can be classified into two sets: a heavy quark singlet with quantum numbers j P = j P l and a heavy quark multiplet with quantum numbers j P = (j l + 1) P , · · · , (\|j l − 1\|) P . These baryons have degenerate masses and the two baryons with quantum numbers j P = j P l in these two sets do not mix to each other due to heavy quark spin conservation in the heavy quark limit. Examples of the ground states and first excited states are summarized in Table. I. After the above discussion on the pattern of the mass degeneracy, we turn to the strong decays of the DHBs. Because the heavy quarks in a DHB have large masses, the light quark in the DHB sees the heavy diquark as a local source of gluon, which makes the picture of the DHB analogues to the heavy-light meson [4]. Then, to analyze the strong decays of the DHBs with j P l = 1 2 − , we use the chiral partner structure applied in the heavy-light meson sector [29,30]. There, the heavy-quark doublet including j P = 0 − and j P = 1 − heavy-light mesons is regarded as the chiral partner of the doublet including j P = 0 + and j P = 1 + heavy-light mesons. The coupling strengths of interactions between two doublets are determined from the mass differences through generalized Goldberger-Treiman relations, which are in good agreement with experiments [31][32][33]. The difference here is that, in the heavy-light meson sector the heavy component is a heavy quark but in the DHB sector the heavy component is a heavy diquark made of two heavy quarks. But this difference does not affect the chiral structure which is controlled by the light quark degree of freedom in a hadron. In addition, since it takes much more energy to excite the heavy diquark constituent in a DHB, we regard the excited DHB as those with the light quark excitation in it. Similar to the heavy-light meson sector, we regard the DHBs with the Brown muck of j P l = 1 2 − , i.e., the heavy quark singlet N Q and heavy quark doublet N µ Q , as the chiral partners to the ground states, i.e., D Q and D µ Q , respectively. Note that, as in the heavy-light meson sector [34], the extension of the present discussion to the excited states is straightforward. To accommodate the chiral dynamics in the DHB sector, along Ref. [19], we introduce the left-and righthanded DHB fields D where g L,R ∈ SU (2) L,R when we consider only the up and down quarks in the DHBs. In terms of D which transform as D where Ψ ′ QQ ′ and Ψ ′ * QQ ′ are Dirac spinors for the DHBs with j P = 1 2 + and 1 2 − , respectively. Note that the parity transformations of Ψ ′ QQ ′ and Ψ ′ * QQ ′ are given as The expressions of D µ Q and N µ Q in terms of physical states and their transformation are given in Ref. [19]. We will not repeat them here. In terms of the naming scheme in PDG, Ψ bc for the DHB including un-flavored quark and strange quark, respectively. Following the procedure used in Ref. [19] one can easily construct an effective Lagrangian of D Q . There is no coupling between the heavy-quark doublet and heavy quark singlet because of the heavy quark spin conservation. The chiral effective Lagrangian is simply a duplicate of the one given in Ref. [19] which is written in the chiral basis as where M is the light meson field which transforms as M → g L M g † R under chiral transformation. In terms of the scalar and pseudoscalar fields, one can make a decomposition M = S + iΦ = S + 2i (π a T a ) with π a being the pion fields and T a being the generators of SU (2) group with the normalization tr (T a T b ) = (1/2)δ ab . With a suitable choice of the Lagrangian of the light meson sector which will not be specialized here, the chiral symmetry can be realized in the Nambu-Goldstone mode. After chiral symmetry breaking, S, and therefore M field, acquire vacuum expectation value M = f π with f π being the pion decay constant. From the Lagrangian (8), the ∆ term shifts the masses of the DHBs in the same direction, therefore the mass difference between the chiral partners is provided by the g π term as The coupling constant g π in the model measures the magnitude of coupling between chiral partners which can be determined as g π = 4.65 from the heavy-light spectrum [19]. The relation (9) between the f π and g π is known as a generalized Goldberger-Treiman relation [29,30]. Then, by using f π = 92.4 MeV, the mass difference for the non-strange doubly heavy baryon is determined as In terms of the mass difference ∆M B;q , the intermultiplet one-pion transitions of the DHBs in the isospin symmetry limit can be studied. The relevant partial widths are expressed as where \|p π \| is the three-momentum of π in the rest frame of the decaying DHB. The channels including charged pions can be obtained by using the isospin relation. To provide some information for the experimental search of the DHBs, we give some explicit results of the masses and strong decay widths of the DHBs. Since the chiral partner structure only yields the mass difference between chiral partners (10), we need some ground state masses as reference values. Here, among the existing many calculations (see, e.g., those summarized in Refs. [2,3,8,14,16]) we choose the results calculated from the nonrelativistic QCD [6] in which the heavy diquark is treated in a similar fashion as the present work: Our numerical results are given in Table II. For the spectroscopy of the DHBs including a strange quark, by using the spectrum of the heavy-light meson including a strange quark and also the estimation for the DHBs including the same heavy quarks [19], we predict [19,[31][32][33] Q,s = m Gs − m Hs = 350 MeV. (13) Concerning this magnitude of the mass splitting, one concludes that the dominant decay channel of Ω * bc is not the isospin conserving process Ω * bc → Ω bc + η but the isospin violating process Ω * bc → Ω bc + π 0 arising from the η-π 0 mixing. The partial decay width is expressed as where ∆ π 0 η is the magnitude of the η-π 0 mixing which was estimated to be −5.32×10 −3 in Ref. [35] based on the two-mixing angle scheme (see, e.g., Ref. [36] and references therein). As the magnitude of the isospin breaking η-π 0 mixing is very small, the partial widths in Eq. (14) are expected to be small which are consistent with the numerical results given in Table. II. The explicit results of the masses and strong decay widths of the Omega DHBs in Table. II are obtained by chosen the typical reference values [7] m Ω bc = m Ω ′ bc = 6.89 ± 0.07 GeV. Our numerical results are given in Table II. Further, we want to make a comment on the mass splitting of the DHBs in a doublet arising from the heavy quark symmetry breaking effect that is beyond the scope of the present work. The results obtained in various models or methods listed in Ref. [3] can be averaged as: which is smaller than the pion mass. Therefore, the onepion transition between DHBs in the same heavy quark multiplet is forbidden for the kinetic reasons. This is dramatically different from the heavy-light meson sector in which, for example, the D * → Dπ could easily happen. In summary, the degeneracy of the doubly heavy baryons with different heavy quarks is studied based on the heavy quark symmetry of QCD. We point out that, for the DHBs with the same orbital excitation between II. Spectrum of the doubly heavy baryons with different heavy quarks and the partial widths of one-pion intermultiplet transitions. Here, we take mΞ bc = m Ξ ′ bc = 6800±50 MeV [6] and mΩ bc = m Ω ′ bc = 6890±70 MeV [7], and m π ± = m π 0 ≃ 140 MeV as input. The partial widths are obtained by using the central values of the baryon masses. Other partial widths of intermultiplet transitions can be obtained using the isospin relation. The first uncertainty is from the reference values in Eqs. (12) and (15), and the second one is from Eq. (16). heavy diquark and the light constituent, although they can be classified into different heavy quark multiplets, they have the degenerated masses. For the DHBs with the same quantum number but in different heavy quark multiplets, they do not mix to each other because of the heavy quark spin conservation. In addition, by using the chiral partner structure, we studied the mass splitting and the transition rates between the ground state DHBs and the first excited states of the light quark. The mass splitting is estimated to be about 430 MeV for the nonstrange DHBs and 350 MeV for the strange DHBs and, due to kinetic reason, the dominant decay channel of the parity odd strange DHB is an isospin-violating process which, therefore, has a small partial width.	2016-01-22T02:48:38.000Z	2015-10-26T00:00:00.000	{ "year": 2016, "sha1": "dbd3742c9057eb1df95ba477f838558932f505cb", "oa_license": "CCBY", "oa_url": "https://doi.org/10.1016/j.physletb.2016.01.011", "oa_status": "GOLD", "pdf_src": "Arxiv", "pdf_hash": "dbd3742c9057eb1df95ba477f838558932f505cb", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
253471731	pes2o/s2orc	v3-fos-license	Effect of Homo-Fermentative Lactic Acid Bacteria Inoculants on Fermentation Characteristics and Bacterial and Fungal Communities in Alfalfa Silage : We evaluated the effects of a homo-fermentative lactic acid bacteria (homo-LAB) inoculant on the fermentation and microbial communities of alfalfa ensiled at two dry matter (DM) contents of 38 and 46% DM. At both DMs, alfalfa was treated or not with an inoculant containing Pediococcus acidi-lactici , Enterococcus faecium and Lactobacillus plantarum at a targeted application rate of 165,000 cfu/g of fresh weight and stored for 3, 30 and 60 days. Treatment with the inoculant resulted in a lower drop in pH and, in general, higher lactic acid and lower acetic acid when applied to medium DM silage. For the four most abundant microbial genera, increased abundances of Bacteroides and Lactobacillus ( p < 0.05), as well as decreased abundances of Muribaculaceae were observed in high DM and inoculated silages. The abundance of Prevotellaceae-UCG-001 was lower in medium DM control silages than in high DM control silages. Inoculation and DM affected abundances of Vishniacozyma ( p < 0.05). Increased abundances of Vishniacozyma , as well as decreased abundances of Leucosporid-ium were observed in medium DM-inoculated silages. Changes in the relative abundance (RA) of the main populations of bacteria and yeasts did explain the fermentation and nutrition differences among treatments. Introduction Alfalfa is one of the popular fodder crops fed to ruminants, but high-quality alfalfa silage is difficult to make due to its high buffering capacity, low concentration of soluble carbohydrates and the presence of epiphytic lactic acid bacteria in the raw material [1].Epiphytic and endophytic harmful bacterial and fungi constitute the predominant microbiota as green fodder, and a substrate converted to silage is highly susceptible to decrease silage quality, negatively affecting the health status of animals [2].Silage inoculants containing lactic acid bacteria (LAB) are not always successful at improving silage quality because the degree of wilting, ensiling times and environmental temperatures vary greatly and can restrict the success of these inoculants [3,4].Better-adapted epiphytic microflora might outcompete organisms for inoculant supply and dominate the succeeding fermentation [5].Increasing the dry matter content of forage at ensiling can affect the ensuing fermentation, proteolysis and respiration in the silo.For example, in alfalfa silages wetter than 40% DM, Stallings et al. (1981) and Papadopoulos and McKersie (1983) reported increased proteolysis than in more highly wilted silages, while, in drier silages (greater than 40% DM), increased wilting has consistently reduced proteolysis [6,7]. Silages treated with homo-fermentative lactic acid bacteria (homo-LAB) have been shown to improve fermentation quality, indicated by a fast lowering of pH, high lactic acid content and low level of ammonia-N [8][9][10][11].These marked features of a silage inoculant arise from the proliferation of homo-LAB at the expense of hetero-LAB and other species [12].The use of homo-LAB such as L. plantarum, E. faecium and P. acidilactici is common [13].L. plantarum is widely used as a silage additive because the species effectively utilizes WSCs to produce lactic acid, which rapidly reduces the pH of the ensiled mass [14].Li reported that ferulic acid could be released from plant cell walls after treatment of L. plantarum A1; the release of ferulic acid from plant cell walls may increase the digestion of whole corn silage in the rumen [15]. Additives for ensilage, consisting of several homo-LAB strains, may have a different impact on the ensilage process; as we know, it is important to study the use of different combinations of LABs to determine the most beneficial effect on the different types of silages.More studies are needed to evaluate the effectiveness of homo-LAB combinations on affecting the fermentation of the kinds of silages under challenging management issues that can occur in the field, such as dry matter, storage times and any interactions among these factors. Next-generation sequencing (NGS) technology has become more sophisticated and easily accessible to characterize the microbial communities associated with silage fermentation [16].Characterizing the changes of the microbial community and its ultimate effects on inoculant interactions with pre-wilting might provide valuable information for improving silage fermentation with homo-LAB inoculation [8].The fact that most assessment of inoculants have focused only on fermentation products may lack insights into the bacterial and fungal microbiota associated with homo-LAB and any interactions with dry matter and storage times.The efficacy of homo-LAB inoculant use should be evaluated, not only for fermentation products, but also for the effects on microbial communities.The aim of this study was to evaluate the effect of three Lactobacillus species on the chemical and microbiological parameters of alfalfa silage by evaluating the effect of applying a homo-LAB inoculant on the fermentation and microbial community of alfalfa silage harvested at two DM contents. Ensiling On 26 June 2017, second-cut alfalfa (early bloom stage) from one field was harvested at the University of Delaware farm in Newark, Delaware and wilted to a DM content of 37.8% or 46.5%.Forage was chopped to 0.95 cm theoretical chop and five individually replicated piles of forage at each DM were treated with: a) no treatment (CTR) or b) treatment with Purina FI Enhance Inoculant (PEI, Purina Animal Nutrition LLC, Arden Hills, MN, USA) containing Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum, with a final targeted application rate of 165,000 cfu of total LAB/g of fresh forage weight.After treatment, about 1 kg of forage was packed into nylon-polyethylene vacuum barrier bags (3.5 mil thickness, 15.2 × 30.5 cm 2 , O 2 transmission rate of 97 cm 2 /m 3 /d at 22.8 • C, 0% rh; Doug Care Equipment Inc., Springville, CA, USA) for each treatment.Air was vacuumremoved from the bags and heat-sealed using a Best Vac vacuum sealer (distributed by Doug Care Equipment Inc., Springville, CA, USA).All silos were stored at 21 ± 0.5 • C for 3, 30 and 60 d. Analytical Procedures We measured the DM content and pH of fresh forages from four CTR piles at 0 d and of five samples from each silo after ensiling.The DM content was determined after 48 h of incubation in a 60 • C forced air oven.Weights of full and empty silos were measured at silo opening and with the DM content of fresh and ensiled samples, which were used to calculate DM recovery.Representative 25 g portions of fresh forage and silage were mixed with 225 mL of sterile, quarter-strength Ringers solution (Oxoid BR0052G, Oxoid, Unipath Ltd., Basingstoke, UK) and homogenized for 1 min in a Proctor-Silex 57171 blender (Hamilton Beach⁄Proctor-Silex Inc., Washington, DC, USA).The homogenate was filtered through four layers of cheesecloth.The pH was determined on the homogenized samples using an Oakton pH700 Meter (Oakton Instruments, Vernon Hills, IL, USA). Portions of the water extracts previously filtered through four layers of cheesecloth were further filtered through Whatman 54 filter paper (Whatman Inc., Clifton, NJ, USA).The extract was acidified with H 2 SO 4 50% vol/vol and frozen.Later, the extract was analyzed for the concentrations of lactic, acetic, butyric and propionic acids and 1, 2-propanediol and ethanol [17] by high-performance liquid chromatography using a Shimadzu LC-20AD (Shimadzu Scientific Instruments, Columbia, MD, USA).The concentration of NH3-N was determined in the water extracts by the phenol-hypochlorite method of Weatherburn (1967) [18], and water-soluble carbohydrates (WSC) were determined by the colorimetric procedure of Nelson (1944) [19]. Nutrient analysis of fresh forage described below was performed by Cumberland Valley Analytical Services (Waynesboro, WV, USA).A portion of each dried sample was ground using an Udy Cyclone Sample Mill (Udy Corp., Fort Collins, CO, USA) to pass through a 1 mm screen.Total N was measured by combustion of the sample following the AOAC method 990.03 (AOAC, 2010) [20] using a Leco CNS2000 Analyzer (Leco Corporation, St. Joseph, MI, USA), and the concentration of crude protein (CP) was calculated by multiplying the resulting total N concentration by 6.25.Soluble protein (SP) was determined by the method of Krishnamoorthy et al. (1982) [21] and expressed as % of CP.The concentration of acid detergent fiber (ADF) was quantified on dried ground samples using the AOAC method 973.18 [20], with the modification of using Whatman 934-AH glass microfiber filters with 1.5 µm particle retention instead of fritted glass crucibles.The concentration of neutral detergent fiber (NDF) was analyzed on dried samples according to the methodology of Van Soest et al. (1991) [22], with sodium sulfite and amylase and was not corrected for ash content.Another portion of the dry sample was ground to pass a 3 mm screen and analyzed for starch by the methodology of Hall (2008) [23]. Enumeration of Microorganisms on Agar Plates For each sample, a portion of the previously described, freshly prepared water extract was serially diluted in quarter-strength Ringers solution, and the numbers of LAB were determined by pour-plating in de Man, Rogosa and Sharpe (MRS) agar (CM3651, Oxoid, Unipath Ltd., Basingstoke, UK).Plates were incubated anaerobically at 35 • C in sealed plastic containers with anaerobic packs (Mitsubishi Gas Chemical Co., Tokyo, Japan) and an anaerobic indicator (BR0055, Oxoid, Unipath Ltd., Basingstoke, UK).Containers were vacuum-sealed in nylon-polyethylene bags (3.5 mil embossed pouches, 15.2 × 30.5 cm 2 ; Doug Care Equipment Inc., Springville, UT, USA) using a Best Vac vacuum machine (distributed by Doug Care Equipment Inc.).Plates with a number of colonies between 30 and 300 were counted after 48 and 72 h to obtain the number of colony-forming units.Numbers of yeasts and molds were determined on malt extract agar (MEA, CM0059, Oxoid, Unipath Ltd., Basingstoke, UK) acidified after autoclaving with lactic acid (85%) at a rate of 0.5% vol/vol.The plates were incubated aerobically at 30 • C and enumerated after 48 and 72 h.Plates with a number of colonies between 30 and 300 were counted to obtain the number of colony-forming units. Analysis of the Composition of the Microbial Communities by Next-Generation Sequencing Samples from five replicates of fresh forage and 90 d silages (CTR, PEI) were analyzed for the composition of their microbial community by NGS.Representative 25 g portions of the samples were mixed with 225 mL of autoclaved quarter-strength Ringers solution (Oxoid BR0052G, Oxoid, Unipath Ltd., Basingstoke, UK) for 2 min using a Colworth 400 stomacher (Seward, London, UK).The homogenates were filtered through four layers of sterile cheesecloth.Next, 2 mL of the filtrate were centrifuged for 3 min at 21,000× g using a Centrifuge 5424 R (Eppendorf AG, Hamburg, Germany).The supernatant was discarded, and 100 µL of autoclaved Ringers solution was used to resuspend the pellet.Samples were kept at −80 • C for further analysis. The procedure of Next-Generation Sequencing analysis was followed by E. B. da Silva (2020) [27].All sequencing data were received as FASTQ files and deposited in the NCBI Sequence Read Archive under BioProject accession PRJNA886387.We have released the above data. Statistical Analysis Agar counts of lactic acid bacteria and yeasts and molds were transformed to log10 before statistical analysis and were presented on a fresh weight basis.Data were analyzed using JMP version 12 (SAS Institute, Cary, NC, USA).Data on the chemical, microbial and physical parameters were analyzed as a 3 × 2 × 2 factorial arrangement of treatments in a completely randomized design.The model included the following main effects: day of ensiling (DAY), effect of inoculation (INO) and effect of DM (DRM).The interactions of DAY × INO, DAY × DRM, INO × DRM and DAY × INO × DRM were tested.The number of statistical repetitions for each group was 5. Differences were considered significant when p < 0.05 and trends towards significance were considered when p < 0.10.Pairwise mean comparisons were performed using Tukey's test at alpha = 0.05.[28].For microbial composition analysis, at each taxonomic level, we defined the taxa with relative abundance > 0.01% in at least one sample as identified, while those with relative abundance > 0.1% presented in more than half of the sample per group as detected. Targeted and Actual LAB Application Rate The targeted and actual final application rates of LAB added to the forage prior to ensiling are shown in Table 1.The amount of LAB applied was slightly lower than planned. Nutrient Composition and Numbers of Culturable Microorganisms in Fresh Forage The characteristics of the freshly chopped alfalfa before ensiling are presented in Table 2. Before ensiling, fresh and untreated alfalfa were wilted to a DM content of 378 g/kg (low DM) and 461 g/kg (high DM), and the pH value was 6.24 and 6.23.The average number of epiphytic microbes in fresh alfalfa such as LAB, yeasts and molds were about 5.86, 3.33 and 2.95 log10 cfu/g, respectively, and did not differ among the two DM level silage materials.Wilted alfalfa was low in water soluble carbohydrates and nitrates, high in protein and had a high buffer capacity at a fermentation coefficient (FC) below 58. Nutrient Composition, Fermentation Profile and Numbers of Culturable Microorganisms in Silages As expected, because fermentation end products change with the time of ensiling, there were numerous three-way interactions in this study.Results and discussion will focus primarily on the effects of inoculant treatments and any interactions between DM and days of ensiling. The chemical composition of silages, shown as the three-way interaction, is shown in Table 3, and only appropriate significant main and two-way interactions are shown (when there was no overlapping three-way or two-way interactions) in Table 4.As expected, the concentration of DM was not affected by inoculation, but it was different between DM (36.96 vs. 45.20%DM) and, on average, was slightly lower at 60 d (40.47%) compared to 3 (41.43%)and 30 d (41.44%) (Table 4).There were only main effects for the concentration of CP in silage.The concentration of CP was greater for PEI (average of 18.16%) vs. CTR (17.94%), lower for high DM silages (17.92%) compared to medium DM silages (18.25%) and increased with days of ensiling (17.87% at 3 d, 18.07% at 30 d and 18.26% at 60 d) (Table 4).There was a three-way interaction for Sol-CP (Table 3).Inoculation had no effect compared to CTR at 3 and 30 d.There was also a three-way interaction for NH 3 -N (Table 3).Similar to Sol-CP, inoculation had no effect when compared to CTR at 3 and 30 d.However, after 60 d in medium DM silage, inoculation resulted in less NH 3 -N (0.211% for PEI) compared to CTR (0.245%).In high DM silage at 60 d, inoculation numerically reduced NH 3 -N compared to CTR.INO and DRM affected ADF, but differences among treatments were not related to their interaction.The concentrations of NDF were higher in high DM than medium DM samples (Table 4).The concentration of WSC (Table 3) was higher in inoculated silages (PEI = 6.37%) than in CTR (4.44%) in the final 60 d in medium DM silage but was generally unaffected by inoculation at other times (three-way interaction).Fermentation end products are shown in Table 5.There was a three-way interaction for pH.Overall, the effect of inoculation on the drop in pH was faster in medium DM silages than high DM silages during the early stages of fermentation.After 3 d of ensiling, medium DM-inoculated silages had markedly lower pH (PEI = 5.17) compared to CTR (5.51), but pH was not affected by inoculation in the high DM treatment.In the current study, after 30 d, inoculated silages were still lower in pH (PEI = 4.08) compared to CTR (4.35) in medium DM silage.After 60 d of ensiling, pH was lower for inoculated silage (average pH = 4.08) vs. CTR (4.29) silage in medium DM.There was a three-way interaction for the concentration of lactic acid in silage (Table 5).Treatment with PEI resulted in consistent, numerically (but not always statistically) higher concentrations of lactic acid when compared to CTR silages at all DM and time points.Specifically, PEI was statistically higher in lactic acid (PEI = 7.04%) than CTR (5.89%) in medium DM silage at 30 d. Treatment with PEI had no effect on the concentrations of acetic acid in high DM silages after 3 d of ensiling (Table 5).However, inoculation resulted in consistently lower concentrations of acetic acid when compared to CTR in medium DM silages after 30 and 60 d.The inoculation of high DM silages numerically, but not statistically, resulted in lower acetic acid than CTR at 30 and 60 d.There was a three-way interaction for 1, 2-propanediol (Table 5). 1, 2-propanediol was present only after 3 d of ensiling.There was only an INO and INO × DAY effect for ethanol, as the PEI treatment (0.87%) was lower than CTR (1.13%)only in medium DM silage (Table 5).When compared to CTR, the numbers of LAB were not greater when inoculated.This is in contrast to the findings of lower pH in inoculated medium DM silage but suggests that while inoculation did not increase the numbers of LAB, it made fermentation more efficient.Low numbers of LAB in PEI silages in medium DM at 30 d of silage (between 3.6 and 3.94 log cfu/g) are not explainable and appears to be an anomaly.Inoculation had no primary or interaction effects on the numbers of yeasts. Analysis of the Composition of the Bacterial and Fungal Communities by Next-Generation Sequencing for Fresh Alfalfa and 90 d Silage The alpha diversity of bacterial and fungal communities for alfalfa silage is shown in Table S1.Fresh forage had a similar diverse bacterial population with silage, as indicated by a more or less consistent Shannon index with silages (Supplementary Table S1).Regarding the fungal community diversity, silages had a low Shannon and Chao1 index than fresh forage, indicating that ensiling reduced fungal diversity.For a better understanding of the microbial community structure in alfalfa silage, the 10 most abundant family and genus of bacteria and fungi in fresh forage and silages can be found in Figures 1 and 2. Figure 2 also shows the 10 most abundant fungal genera in fresh forage and silages.Silages have different predominant fungal genera compared with fresh material.Leucosporidium and Sclerotiniaceae were the most primary fungi in silages, while Filobasidium and Heterophoma dominated the alfalfa material.After 90 d ensiling, at the genus level, in medium DM CTR silages, the most predominant fungal populations were found to be Leucosporidium (37%) and Sclerotiniaceae (27%), which accounted for over 64% of total sequences (Figure 2b).For medium DM PEI silages, the most abundant fungal populations were Leucosporidium, Sclerotiniaceae and Vishniacozyma accounted for 20%, 18% and 17%, respectively.For high DM CTR silages, Leucosporidium (32%) was the most predominant fungal taxa, followed by Sclerotiniaceae (23%).For high DM PEI silages, Sclerotiniaceae was a dominant genus detected with 38% abundance, followed by Leucosporidium (35%) (Figure 2b), while other genera were detected below 5% of their reads.The predominant bacterial genera were Muribaculaceae and Bacteroides in fresh forage and silages, which were members of the order Bacteroidales (Figure 1).While Lactobacillus (6.04%) was the fourth predominant bacterial taxa in silages, traces of Lactobacillus were detected in material.At the genus level, the abundance of Muribaculaceae (20%) and Bacteroides (13%) was higher than other bacteria in fresh material.Sequences from Chryseobacterium, Parabacteroides and Muribaculum accounted for 9%, 7% and 4% abundance in fresh material, respectively.After 60 d ensiling, at the genus level, in medium dry matter silages, the most abundant bacterial populations were Muribaculaceae, Bacteroides, Alistipe, Parabacteroides, Prevotellaceae and Faecalibaculum accounting for 53%, 9%, 9%, 6%, 6% and 6% of all the bacterial reads, respectively.Moreover, the abundance of Lactobacillus (0.7%) was detected in medium DM CTR silage.For medium DM PEI silage, Muribaculaceae (35%), Bacteroides (19%) and Prevotellaceae (15%) were the three most abundant genera for the bacterial community, which accounted for over 70% of total sequences (Figure 1b).The percentage of Lactobacillus (9% for PEI) was distinctly higher than those of medium DM CTR silage.For the high dry matter alfalfa silage bacterial community, at the genus level, Muribaculaceae (34%), Bacteroides (15%), Prevotellaceae (20%) and Lactobacillus (6%) were the four most abundant genera in high DM CTR silage.Muribaculaceae (32%), Bacteroides (25%), Prevotellaceae (8%) and Lactobacillus (8%) were the most predominant bacterial taxa in PEI silages.Table 6 shows the effect of dry matter content and inoculation on the four most abundant bacterial and fungal genera in the alfalfa silages.For bacterial genera, the additive significantly affected the RA of Bacteroides and Lactobacillus (p < 0.05).The dry matter difference affected the RA of Bacteroides and Lactobacillus (p < 0.05) in silages.There was an interaction between dry matter content and additive for the RA of Muribaculaceae, Bacteroides, Prevotellaceae and Lactobacillus (p < 0.05) in silages.In contrast, the additive caused a marked decrease in the RA of Prevotellaceae in high DM PEI compared to high DM CTR.The addition of bacterial additives did not affect the numbers of cultivable LAB on MRS when compared to CTR in PEI silage, but it increased (p < 0.05) the RA of Lactobacillaceae in medium DM PEI compared to CTR.Table 6 shows the effect of dry matter content and additive treatment on the four most abundant fungal genera in the alfalfa silages.The addition of LAB and dry matter values affect the RA of Vishniacozyma (p < 0.01).There was an interaction between the dry matter content and additive for the RA of Leucosporidium, Sclerotiniaceae (p < 0.05) and Vishniacozyma (p < 0.01) in silages.Table 6.Relative abundance (%) of material, bacterial genus and fungal genus in alfalfa silages ensiled for 60 d, as analyzed by the sequencing of the V4-V5 region of the 16S rRNA, for bacteria, and ITS1, for fungi, using the Illumina MiSeq platform.Figure 2 also shows the 10 most abundant fungal genera in fresh forage and silages.Silages have different predominant fungal genera compared with fresh material.Leucosporidium and Sclerotiniaceae were the most primary fungi in silages, while Filobasidium and Heterophoma dominated the alfalfa material.After 90 d ensiling, at the genus level, in medium DM CTR silages, the most predominant fungal populations were found to be Leucosporidium (37%) and Sclerotiniaceae (27%), which accounted for over 64% of total sequences (Figure 2b).For medium DM PEI silages, the most abundant fungal populations were Leucosporidium, Sclerotiniaceae and Vishniacozyma accounted for 20%, 18% and 17%, respectively.For high DM CTR silages, Leucosporidium (32%) was the most predominant fungal taxa, followed by Sclerotiniaceae (23%).For high DM PEI silages, Sclerotiniaceae was a dominant genus detected with 38% abundance, followed by Leucosporidium (35%) (Figure 2b), while other genera were detected below 5% of their reads. Vishniacozyma Table 6 shows the effect of dry matter content and inoculation on the four most abundant bacterial and fungal genera in the alfalfa silages.For bacterial genera, the additive significantly affected the RA of Bacteroides and Lactobacillus (p < 0.05).The dry matter difference affected the RA of Bacteroides and Lactobacillus (p < 0.05) in silages.There was an interaction between dry matter content and additive for the RA of Muribaculaceae, Bacteroides, Prevotellaceae and Lactobacillus (p < 0.05) in silages.In contrast, the additive caused a marked decrease in the RA of Prevotellaceae in high DM PEI compared to high DM CTR.The addition of bacterial additives did not affect the numbers of cultivable LAB on MRS when compared to CTR in PEI silage, but it increased (p < 0.05) the RA of Lactobacillaceae in medium DM PEI compared to CTR.Table 6 shows the effect of dry matter content and additive treatment on the four most abundant fungal genera in the alfalfa silages.The addition of LAB and dry matter values affect the RA of Vishniacozyma (p < 0.01).There was an interaction between the dry matter content and additive for the RA of Leucosporidium, Sclerotiniaceae (p < 0.05) and Vishniacozyma (p < 0.01) in silages. Discussion The chemical and microbiological content of freshly chopped forages are shown in Table 2.The nutrient components were considered "normal" for alfalfa and, with the exception of DM, were very similar between the two DMs.The differences in alfalfa DM between treatments before ensiling were biologically significant enough to affect the silage characteristics (p < 0.05).Low numbers of lactobacilli and high numbers of aerobic bacteria were present in alfalfa material.This was observed previously by Cai et al. (2017) [29], who suggested that this may result in poor fermentation.Therefore, it is necessary to add fermentable substrates and LAB inoculants to control microbes in silage fermentation [30]. The benefits of LAB at ensiling were shown again, in this study, where the application of homo-LAB improved silage fermentation by lowering silage pH and the concentration of ammonia-N and increased WSC concentration and a higher concentration of lactic acid in medium DM silages.Lower CP and higher NDF in high DM silages is not surprising as there is greater potential for leaf loss during the harvesting of high DM alfalfa.Generally, the higher production of lactic acid via inoculation supports the findings that lower pH with inoculation is a strong indicator that the added lactic acid bacteria dominated the overall fermentation.These results indicated that application of homo-LAB resulted in silage with a more homo-lactic profile in medium DM silages [31].A faster drop in silage pH is desirable because low pH can inhibit undesirable Enterobacteria and Clostridia [32].The lack of consistent improvements in fermentation in high DM silages treated with homo-LAB is not unexpected because fermentation becomes more restricted as water activity declines with increasing concentrations of DM in silages [23].Previous work in our lab supports that, in high DM alfalfa silages, there is a lag of several days prior to active fermentation, most likely because of the low available water for microbial growth [30].Treatment with homo-LAB did not result in the classical effects of more lactic and less acetic acid and the suppression of proteolysis often reported in previous studies [10,30]. At the genus level, Muribacula (20.08-52.86%),Bacteroides (8.96-25.38%),Prevotellaceae (2.61-20.15%)and Lactobacillus (0.05-8.95%) were dominant in all samples.Other genera (Alistipes, Parabacteroides, Faecalibaculum, Chryseobacterium, Lachnospiraceae and Muribaculum) were represented with lower abundance but occupied the top 10 genera in material and silages (Figure 1b).Several predominant bacterial genera of epiphytic flora survived and were represented with a higher abundance.Inoculants comprising certain strains of LAB have been developed to reduce the reliance of epiphytic flora, but Lactobacillus was not the most abundant genus in all silages.Interestingly, in this study, the structure of the silage flora was similar with the flora in the rumen.To our knowledge, this is the first study to find that the abundance of rumen flora (like top 10 genus) was generally significantly higher in silages than in fresh material (p < 0.05).This indicated that these rumen bacterial genera can survive, reproduce and be the predominant bacterial genera in silages.In previous studies, the additives used generally have not had consistent effects on nutrient composition [33,34].In the current study, the NDF and ADF concentration of silages was altered by the homo-LAB treatment and dry matter change.This finding might have been a result of the main effect of these rumen bacterial genera in silages.It is known that Prevotellaceae, Bacteroides and Muribaculaceae, which belong to the phylum Bacteriodetes, responsible for hemicellulose, pectin and high carbohydrate levels, feed digestion and are capable of degrading cellulose and starch in the rumen [35,36].Bacteroides is an important anaerobic genus that plays a fundamental role in the gut ecosystem by breaking down polysaccharides [37].It consumes lactic acid and converts it to VFA and is able to degrade deoxy sugars via the propanediol pathway but produce the pathway intermediate 1, 2-propanediol as the final product [38].The appearance of 1, 2-propanediol and propionic acid in silages may be attributed to these rumen bacteria.The increased population of the genus Prevotellaceae in silages produced with LAB inoculants is another explanation for the higher molar proportion of propionic acid in medium DM PEI silages. The homo-LAB inoculant was successful in lowering pH and increasing lactic acid concentrations in silages but did not affect the yeast population and RA of fungi.The DM content also had no effect on the yeast exception of Vishniacozyma.Leucosporidium sp. was the first, most abundant feature in all the silage samples (Table 6).Leucosporidium is able to assimilate glucose [39].Sclerotiniaceae usually is found in alfalfa and causes disease because of its ability for long-term survival accompanied by great reproduction potential [40].Filobasidium is relatively common among the species of endophytic fungi of grasses and has been isolated from the rhizosphere of corn [41].The genus Filobasidium has been reported to be an abundant genus in fresh corn silage and declines with ensiling time [42].Vishniacozyma (also known as Cryptococcus) has been isolated from soil and wheat [43] while there are no reports of this genus in silage.Vishniacozyma was reported to be able to assimilate lactic acid and D-lactose [44].Most yeast comes from soil.Ash is a measure of the forage, with values > 10% for grasses reflecting soil contamination [45].Higher levels of ash in silages could be attributed to an increased risk of soil contamination when grasses are transported from the field to the farm in this study. Conclusions Silage is heterogenous in nature with differences in chemical and microbiological composition, as well as chemical properties within a silo [46], thus, different kinds of silage types are worth exploring in depth.The results from this study show that using a homo-LAB inoculant resulted in a lower drop in pH especially when applied to medium DM alfalfa silage and, in general, resulted in a more efficient silage fermentation characterized by higher lactic acid and lower acetic acid.Inoculating medium DM alfalfa also resulted in lower NH 3 -N and higher residual WSC concentrations compared to CTR after 60 d of fermentation.Overall, treatment with the inoculant in this study was highly beneficial as it caused a faster drop in pH and a more homo-lactic acid type of fermentation.The changes in the relative abundance of the main populations of bacteria and yeasts did explain the improvements in fermentation products and nutrition changes by the additive application.The finding that the rumen flora may survive and dominate in alfalfa silages needs to be further studied.The assessment and further study of homo-or hetero-LAB inoculants for Fermentation 2022, 8 , 16 Figure 1 . Figure 1.Relative abundances of the 10 most abundant family and genus of the bacterial microbiota of pre-ensiled crops and silages prepared from alfalfa silage.(a) Family-and (b) genus-level bacterial community in alfalfa silage.M-alfalfa ensiled at medium DM; H-alfalfa ensiled at high DM; CTR-no additive; PEI-inoculated with Land O'Lakes inoculant (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); M-medium DM alfalfa silage; and H-high DM alfalfa silage. Figure 1 . Figure 1.Relative abundances of the 10 most abundant family and genus of the bacterial microbiota of pre-ensiled crops and silages prepared from alfalfa silage.(a) Family-and (b) genus-level bacterial community in alfalfa silage.M-alfalfa ensiled at medium DM; H-alfalfa ensiled at high DM; CTR-no additive; PEI-inoculated with Land O'Lakes inoculant (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); M-medium DM alfalfa silage; and H-high DM alfalfa silage. Fermentation 2022, 8 , 16 Figure 2 . Figure 2. Relative abundances of the 10 most abundant family and genus of the fungal microbiota of pre-ensiled crops and silages prepared from alfalfa silage.(a) Family-and (b) genus-level bacterial community in alfalfa silage.M-alfalfa ensiled at medium DM; H-alfalfa ensiled at high DM; CTR-no additive; PEI-inoculated with Purina FI Enhance Inoculant (PEI, Purina Animal Nutrition LLC, Arden Hills, MN, USA) (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); M-medium DM alfalfa silage; and H-high DM alfalfa silage. Figure 2 . Figure 2. Relative abundances of the 10 most abundant family and genus of the fungal microbiota of pre-ensiled crops and silages prepared from alfalfa silage.(a) Family-and (b) genus-level bacterial community in alfalfa silage.M-alfalfa ensiled at medium DM; H-alfalfa ensiled at high DM; CTR-no additive; PEI-inoculated with Purina FI Enhance Inoculant (PEI, Purina Animal Nutrition LLC, Arden Hills, MN, USA) (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); M-medium DM alfalfa silage; and H-high DM alfalfa silage. Table 1 . Targeted and actual final lactic acid bacteria application rates. Targeted Final Lactic Acid Bacterial Targeted Application Rate-cfu/g of Fresh Forage Actual Final Lactic Acid Bacterial Targeted Application Rate-cfu/g of Fresh Forage 1 PEI-inoculated with Purina FI Enhance Inoculant (PEI, Purina Animal Nutrition LLC, Arden Hills, MN, USA).Actual final lactic acid bacterial targeted application rate based on plating the inoculant solutions actually applied on the day of the study. Table 2 . Chemical (% DM basis unless stated otherwise) and microbial analysis (log10 cfu/g fresh weight basis) of freshly chopped alfalfa before treatment. * SEM-the standard error of mean difference, n = 5. Table 3 . The DM and chemical analysis (% DM basis unless stated otherwise) of alfalfa silage ensiled for 3, 30 and 60 d.Means shown for three-way interaction.(ADF, NDF and ash data only for day 60).CTR-no additive; PEI-inoculated with Purina FI Enhance Inoculant (Purina Animal Nutrition LLC, Arden Hills, MN, USA) (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); and Day 3, 30 and 60days of ensiling.The model included the following main effects: day of ensiling (DAY), effect of inoculation (INO) and effect of DM (DRM).The interactions of DAY × INO, DAY × DRM, INO × DRM and DAY × INO × DRM were tested.SEM-the standard error of mean difference, n = 5.Values in the same column with different following letters (a-g) are significantly different.Means within columns with unlike superscript differ p < 0.05; nd-not determined. * Table 4 . The DM and chemical analysis (% DM basis unless stated otherwise) of alfalfa silage ensiled for 3, 30 and 60 d.Means shown for main effects and two-way interaction when significant.There were no significant INO × DAY or DRM × DAY interactions (data not shown).(ADF, NDF and ash data only for day 60). * CTR-no additive; PEI-inoculated with Land O'Lakes inoculant (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); and Day 3, 30 and 60-days of ensiling.The model included the following main effects: day of ensiling (DAY), effect of inoculation (INO) and effect of DM (DRM).The interactions of DAY × INO, DAY × DRM, INO × DRM and DAY × INO × DRM were tested.SEM-the standard error of mean difference, n = 5.Values in the same column with different following letters (a-c) are significantly different.Means within columns with unlike superscript differ p < 0.05; nd-not determined. Table 5 . The pH, fermentation end products (% of DM) and microbial populations (log cfu/g wet weight basis) of alfalfa silage ensiled for 3, 30 and 60 d. DAY × DRM, INO × DRM and DAY × INO × DRM were tested.SEM-the standard error of mean difference, n = 5.Values in the same column with different following letters (a-h) are significantly different.Means within columns with unlike superscript differ p < 0.05. Table 6 . Relative abundance (%) of material, bacterial genus and fungal genus in alfalfa silages ensiled for 60 d, as analyzed by the sequencing of the V4-V5 region of the 16S rRNA, for bacteria, and ITS1, for fungi, using the Illumina MiSeq platform.CTR-no additive; PEI-inoculated with Land O'Lakes inoculant (Pediococcus acidilactici, Enterococcus faecium and Lactobacillus plantarum); and Day 3, 30 and 60-days of ensiling.The model included the following main effects: effect of inoculation (INO) and effect of DM (DRM).The interactions of INO × DRM were tested.SEM-the standard error of mean difference, n = 5 or 4. Values in the same column with different following letters (a-c) are significantly different.Means within columns with unlike superscript differ p < 0.05. *	2022-11-12T16:11:23.785Z	2022-11-10T00:00:00.000	{ "year": 2022, "sha1": "7663dcd16ef55bfded38cec5b98938f7059cf2fa", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2311-5637/8/11/621/pdf?version=1668063992", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "a5279c23ccb84a32df3d2264e36964abf3b2f4ab", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "extfieldsofstudy": [] }
28261114	pes2o/s2orc	v3-fos-license	Brain grey matter volume alterations associated with antidepressant response in major depressive disorder Not all patients with major depressive disorder respond to adequate pharmacological therapy. Psychoradiological studies have reported that antidepressant responders and nonresponders show different alterations in brain grey matter, but the findings are inconsistent. The present study reports a meta-analysis of voxel-based morphometric studies of patients with major depressive disorder, both antidepressant responders and nonresponders, using the anisotropic effect size version of Seed-based D Mapping to identify brain regions correlated to clinical response. A systematic search was conducted up to June 2016 to identify studies focussing on antidepressant response. In responders across 9 datasets grey matter volume (GMV) was significantly higher in the left inferior frontal gyrus and insula, while GMV was significantly lower in the bilateral anterior cingulate cortex (ACC) and the right superior frontal gyrus (SFG). In nonresponders across 5 datasets GMV was significantly lower in the bilateral ACC, median cingulate cortex (MCC) and right SFG. Conjunction analysis confirmed significant differences in the bilateral ACC and right SFG, where GMV was significantly lower in nonresponders but higher in responders. The current study adds to psychoradiology, an evolving subspecialty of radiology mainly for psychiatry and clinical psychology. Discussion This meta-analysis is the first to analyze VBM studies in MDD antidepressant responders and nonresponders compared with healthy controls. GMV alterations in the cortico-limbic circuit, especially the prefrontal regions and ACC, were observed in both patient groups, implicating this in the intrinsic pathophysiology of major depressive disorder. However, there were notable differences between the patient groups. GMV in the bilateral ACC and right SFG was higher in responders, but lower in nonresponders. Lower GMV in the left insula and inferior frontal gyrus and higher GMV in right gyrus rectus were only observed in responders, and lower GMV in bilateral MCC was only observed in nonresponders, suggesting links to different mechanisms. Generally, only responders showed higher GMV compared with healthy controls, while nonresponders showed broadly lower GMV. Previous neuroimaging studies have reported lower GMV in MDD 17,24 , and postmortem studies have, accordingly, shown decreases in the density, number and size of neuronal and glial cells 25,26 . Antidepressant drugs may restore GMV in responsive patients 27 , possibly through synaptic plasticity and altered expression of neurotrophic factors 28,29 . This may explain why responders showed higher GMV while nonresponders showed lower GMV. We found GMV differences in the bilateral ACC and right SFG, which were higher in responders but lower in nonresponders. This is consistent with reports that GMV is lower in SFG and that this correlates with the severity of depression 30,31 and that remission of depressive symptoms is associated with higher GMV in right ACC [32][33][34] , higher functional and metabolic activity in ACC [35][36][37] and altered connectivity of the cingulate tracts 38 . As the major regions in the cortical-limbic network, bilateral ACC and right SFG are involved in dysfunctional mood and emotional regulation in MDD 39,40 . Given that antidepressant medication influences brain structure mainly through the serotonergic system 41,42 , GMV in bilateral ACC and right SFG might be especially sensitive to the clinical response to pharmacotherapy. Interestingly, a PET study found that nonresponders had lower serotonin transporter binding in ACC than responders 43 . A previous meta-analysis found the most robust grey matter reductions in a relatively focal region in rostral ACC, both in the pooled meta-analysis and in the subgroup analysis of multi-episode samples 44 , but not in subgroup analysis of first-episode studies 44 . A recent meta-analysis in first-episode depression also failed to find lower grey matter volume in the ACC 45 . As the symptoms of first-episode patients (and in our study, responders) are relatively lighter, this might suggest that ACC abnormalities are sensitive to the severity of symptoms in depressed patients. In responders, GMV was lower in the left insula and IFG. Consistent with this are previous reports that lower GMV in bilateral insula is correlated with severity of depression 46 , and that GMV is lower in the left insula in current and remitted MDD 47 . Functional neuroimaging studies in MDD patients have implicated the insular cortex in the emotional processing of guilt and sadness 48,49 . The lower GMV in the left insula in our study appears consistent with functional studies in MDD demonstrating lower activity in the bilateral or left-side insula 50,51 . Although the role of the insula in the pathophysiological processing in MDD still need be clarified, our results suggest a positive involvement in the treatment response of MDD patients, as was also observed in medication-free MDD patients in a previous meta-analysis 52 . Furthermore, meta-analysis of studies of medication-free patients with MDD has also shown lower GMV in the IFG 52 . In a longitudinal functional neuroimaging study using near infrared spectroscopy (NIRS), both untreated and remitted MDD groups showed significantly lower [oxy-Hb] activation during a verbal fluency task in the bilateral prefrontal cortices compared to HC 53 . These findings may indicate that brain structure and function in the left insula and IFG remains impaired in remitted patients even after improvement of depressive symptoms. The study has some limitations. First, the generalizability of the results was limited by the small sample size which combined only 5 nonresponder datasets and 9 responder datasets, which also meant that meta-regression could not be performed on the nonresponders. Second, important variables like intelligence quotient and handedness were not generally reported, precluding exploration of their impact on the results. Third, there is the potential for bias in the VBM method, whose relative insensitivity to spatially more diverse changes can lead to over-representation of group differences in regions of high anatomic variability. Fourth, the number of included studies was insufficient for an analysis of the effects of particular antidepressants or antidepressant classes. Fifth is the imprecision inherent in all methods depending on summarized coordinates, which are nevertheless necessary because published studies typically use different covariate models or raw statistics. Conclusion Taken together, the present findings demonstrate structural grey matter differences in regions involved in cortico-limbic networks in MDD patients. GMV of the bilateral ACC and right SFG was lower in nonresponders, but higher in the responders, suggesting that this might provide biomarkers associated with antidepressant response and prolonged remission. Furthermore, lower GMV in the left insula and IFG was only present in responders. Longitudinal studies will need to investigate the dynamic effect of antidepressant medication, and for a better understanding of the underlying cause of these GMV alterations. In particular, this study adds to Psychoradiology (https://radiopaedia.org/articles/psychoradiology), an evolving subspecialty of radiology, which is primed to play a major clinical role in guiding diagnostic and therapeutic decisions in patients with mental disorders 54,55 . Methods Inclusion criteria. We followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines (PRISMA) 56 to June 2016, using the keywords 'depressive disorder' or 'unipolar depression' or 'depression' or 'depress*' plus 'VBM' or 'voxel-based morphometry' or 'voxel' or 'morphometry' . We also checked the reference lists of those articles for further researches. Studies were included according to the following criteria: (1) a group of participants diagnosed as having MDD based on DSM criteria were compared with healthy controls; (2) VBM was used to analyze grey matter alteration in MDD patients; (3) coordinates were reported in a standard space like the Talairach space or the Montreal Neurological Institute (MNI) space; (4) the nonresponder group was defined as showing < 50% reduction in the 17-item Hamilton Depression Rating Scale (HDRS-17) total score (or Beck Depression Inventory or Montgomery and Åsberg Depression Rating Scale), and the responder group was defined as ≥50% reduction in the same scale, after treatment at a sufficient dose for 6 weeks. Studies were excluded if they met the following criteria: (1) MDD patients were not compared with healthy controls; (2) coordinates were not clearly reported; (3) VBM was not used; (4) comorbid panic disorder was not excluded; (5) late-onset MDD patients or adolescents with MDD were enrolled. Study selection. Two investigators independently examined abstracts from the initial search, and disagreements were discussed with a third author to reach a consensus. Authors were blinded to the articles' authors, their institutions and the source of funding in order to minimize potential bias. The full texts of studies thought to fulfill the inclusion criteria were assessed in detail to confirm eligibility. Data extraction. Two authors independently extracted data. Differences were resolved by discussion among the review authors. The following data were collected: first author's name, year of publication, details of study design, patient characteristics (including gender, age, illness duration, and disease severity at baseline), sample size, agent dose, duration of treatment and changes in VBM. From each included study we chose the statistically significant peak coordinates of GMV differences resulting from whole brain analysis. Higher grey matter Lower grey matter R SFG R GR R IFG R ACC L ACC L IFG L insula Table 3. Sensitivity analyses of voxel-based morphometric studies of grey matter in antidepressant responders compared with healthy controls. ACC = anterior cingulate cortex; GR = gyrus rectus; IFG = inferior frontal gyrus; L = left; r = right; SFG = superior frontal gyrus. Studies Lower grey matter Quality assessment. Based on previous studies 35,57,58 , the included studies were assessed for quality using a 13-point checklist, including clinical and demographic aspects as well as the imaging methodology. Two authors independently reviewed each paper and assigned a completeness rating for the following items: the quality of the diagnostic procedures, the demographic and clinical characterization, the sample size, the MRI acquisition parameters, the analysis technique and the quality of the reported results (see the Supplement, Table S1). Differences were resolved by discussion among the review authors and a consensus score was assigned, as presented in Table 1. Statistical Analysis. First, independent voxel-wise effects meta-analyses were conducted to investigate regional GMV differences within both responsive and non-responsive groups relative to controls. Second, subgroup analysis in responsive group was performed to investigate the GMV change after antidepressant treatment. Third, conjunction analysis was conducted to identify distinct brain regions where non-responders and responders differed from healthy controls; this used a multimodal analysis to compare the results from two independent meta-analysis 17,59 . The AES-SDM method has been described in detail elsewhere 15 , and we only describe it briefly here. First, the peak coordinates of the brain regions that were significantly different at the whole-brain level were selected. To avoid a potential bias toward liberally thresholded regions, we checked all the included studies to ensure that the same threshold was used throughout the brain. Second, we separately recreated a standard Talairach map of the differences in grey matter for each study by using a Gaussian kernel. The recreation of the peak coordinates was based on converting the peak t value to Hedges' effect size, and then applying a non-normalized Gaussian kernel to the voxels near the peak, which assigns higher values to the voxels closer to peaks. For null findings in the studies, the recreation was done with the same effect size, and all voxels in the effect size map were estimated to have a null effect size, which was the only difference. Similar to other effect sizes, the null effect size was also included in the random-effects meta-analytic models, thus modifying the meta-analytic effect size. Third, the mean of the study maps were analyzed using a voxel-wise calculation to generate a mean map, and this calculation was weighted by the square root of the sample size of each study, so a study with a larger sample size would contribute more. Finally, we used standard randomization tests to determine statistical significance, hence creating null distributions from which p values were directly obtained. The default AES-SDM kernel size and thresholds were used (full-width at half-maximum = 20 mm, voxel p = 0.005, peak height Z = 1, cluster extent = 10 voxels). Reliability analysis. The reliability of results was examined by using Jack-knife sensitivity analysis. The sensitivity analysis was repeated 5 times for nonresponder groups and 9 times for responder groups, to find highly replicable brain regions which were preserved throughout most combinations of the datasets. Meta-regression analysis and publication bias analysis. Meta-regression was used to explore which of the following moderator variables might be responsible for heterogeneity of the findings: mean age of patients, illness duration, and the percentage of female patients. In the absence of consistent depression-scale data, meta-regression to depression symptom severity was not feasible. The probability threshold was reduced to 0.0005 to decrease the finding of spurious associations as less as possible, and regions which are not in the main analysis were ignored. Effect size estimates of the significant clusters were extracted to examine publication bias by using Egger's test and funnel plots.	2018-04-03T01:10:36.741Z	2017-09-05T00:00:00.000	{ "year": 2017, "sha1": "2edd8beb8de291c3953fd8ddb0e72b4e040c3f3f", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-017-10676-5.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "2edd8beb8de291c3953fd8ddb0e72b4e040c3f3f", "s2fieldsofstudy": [ "Psychology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
3692119	pes2o/s2orc	v3-fos-license	Stabilizing the Retromer Complex in a Human Stem Cell Model of Alzheimer’s Disease Reduces TAU Phosphorylation Independently of Amyloid Precursor Protein Summary Developing effective therapeutics for complex diseases such as late-onset, sporadic Alzheimer’s disease (SAD) is difficult due to genetic and environmental heterogeneity in the human population and the limitations of existing animal models. Here, we used hiPSC-derived neurons to test a compound that stabilizes the retromer, a highly conserved multiprotein assembly that plays a pivotal role in trafficking molecules through the endosomal network. Using this human-specific system, we have confirmed previous data generated in murine models and show that retromer stabilization has a potentially beneficial effect on amyloid beta generation from human stem cell-derived neurons. We further demonstrate that manipulation of retromer complex levels within neurons affects pathogenic TAU phosphorylation in an amyloid-independent manner. Taken together, our work demonstrates that retromer stabilization is a promising candidate for therapeutic development in AD and highlights the advantages of testing novel compounds in a human-specific, neuronal system. INTRODUCTION Alzheimer's disease (AD) is a devastating disorder of the brain and a rapidly growing public health problem. Patients with rare mutations in the amyloid precursor protein (APP) and presenilin 1 and 2 genes (PSEN 1 and PSEN 2) make up about 5% of familial AD cases (FAD) while the vast majority of AD is late-onset, sporadic AD (SAD) (Avramopoulos, 2009). SAD is a complex and heterogeneous disorder with a clear heritable component that defines genetic risk (Gatz et al., 2006) but with unknown contributions from the environment. The lack of effective therapies for AD is, in large part, due to our lack of understanding of the cellular and molecular mechanisms that lead to the neuropathological outcomes of the disease. Reasons for this are several-fold. In particular, AD takes decades to manifest clinically, and irreversible cellular damage likely occurs before the overt clinical symptoms are detected. In addition, the human brain is an inaccessible organ, making it difficult to sample tissue from living patients. Human induced pluripotent stem cells (hiPSCs) have facilitated the development of human neuronal models for AD, as they can be differentiated into disease-appropriate cell types and maintain the unique patient genetic background. To date, AD has been successfully modeled using patient and isogenic cell lines, and disease-relevant phenotypes have been well-documented from stem cell-derived neurons for both FAD and SAD (Israel et al., 2012;Kondo et al., 2013;Muratore et al., 2014;Raja et al., 2016;Woodruff et al., 2013;Yagi et al., 2011;Young et al., 2015). We and others have demonstrated a role for endocytic genes in AD risk (Karch and Goate, 2015;Rogaeva et al., 2007;Young et al., 2015), and elegant microscopic analysis has documented enlarged endosomes and suggested endocytic dysfunction as an early phenotype in AD pathogenesis (Cataldo et al., 2000(Cataldo et al., , 2008Nixon et al., 2005). This work has been recapitulated in stem cellderived neurons (Israel et al., 2012;Raja et al., 2016); however, to date, no AD therapy targets the endosomal network, likely due to the complexity of its structure and regulation. Retromer is a multiprotein assembly with a primary role in the sorting and trafficking of plasma membrane proteins from the endosomes, to the trans-Golgi network, or directly back to the cell surface (Hu et al., 2015). APP is trafficked through the endocytic network via retromer through its interaction with SORLA, a member of the vacuolar protein-sorting VPS10 family of sorting receptors and a welldefined AD risk gene (Fjorback et al., 2012;Reitz et al., 2011Reitz et al., , 2013Rogaeva et al., 2007). The retromer assembly has two major components, including the cargo-recognition complex (CRC), a trimeric core composed of VPS35, VPS29, and VPS26 (Li et al., 2016). The second component of retromer comprises the membrane targeting specific nexins, chiefly SNX1, SNX3, SNX5, SNX6, and SNX27 (Burd and Cullen, 2014). Retromer deficiency, primarily loss of VPS35 and VPS26, has been described in SAD patients and increases Ab peptides in cell culture and cognitive decline in the mouse brain (Bhalla et al., 2012;Muhammad et al., 2008;Small et al., 2005;Wen et al., 2011). Recently, novel pharmacological chaperones were developed that stabilize the trimeric core complex of VPS35, VPS29, and VPS26. These molecules, R33 and R55, reduced Ab in mouse hippocampal neurons in a retromer-dependent manner (Mecozzi et al., 2014). Similarly, the retromer stabilization molecule R55 reduced Ab in in a murine neuronal cell line harboring the APP Swedish mutation (Chu and Pratico, 2017). These studies support the idea that increasing the levels and/or functions of endocytic trafficking components may be viable options for therapeutic development in AD. However, to move forward in investigation of therapeutic strategies for humans, it is critical to test candidates in a human-specific system. In addition, while Ab is a critical pathogenic readout for AD, how retromer stabilization affects other components of the disease process, such as TAU hyperphosphorylation, and if these two events are interdependent, is unknown. To address these questions, we used hiPSC-derived neurons from SAD and FAD patients and from genome-edited hiPSC lines. This allowed us to modulate the amounts of Ab generation and APP expression in neurons and test how retromer stabilization by the small molecule chaperone R33 affects endogenous levels of Ab and phospho-TAU (pTAU) in a human-specific system. RESULTS Retromer Pharmacological Chaperones Reduce APP Processing and Ab Peptides in hiPSC-Derived Neurons from Non-demented Controls, SAD Patients, and FAD Mutant Cells The pharmacological chaperone R55 was previously shown to stabilize the retromer core complex, reduce Ab peptides, and decrease APP in endosomes in mouse primary hippocampal neurons, suggesting that retromer stabilization is beneficial in terms of AD pathogenesis (Mecozzi et al., 2014). We extended these studies to test the effect of retromer stabilization in human neurons using an hiPSC approach. We tested both R55 and its analog R33 in FACS-purified neurons derived from non-demented control (NDC) hiPSCs. Our differentiation protocol has consistently been shown to recapitulate key cellular AD phenotypes in a purified system, allowing us to accurately quantitate analytes coming from a particular cell type (i.e., human neurons) (Israel et al., 2012;Young et al., 2015). We found that, in contrast to mouse neurons, R33 had a greater effect on Ab peptide reduction than R55 in human neurons differentiated from hiPSCs ( Figure S1). We then tested a cohort of cell lines derived from participants in the University of California, San Diego (UCSD) Alzheimer's Disease Research Center (ADRC), which included six NDCs and seven SAD patients (n = 13 patient cell lines) ( Figure 1A). We found that in cell lines derived from either NDC or SAD subjects, R33 treatment for 72 hr significantly reduced both Ab 1-40 and 1-42 peptide species ( Figures 1B-1E). Because this reduction occurred equally for both Ab species, we did not observe significant changes in the Ab 42:40 ratio ( Figures 1F and 1G). This is similar to what was previously observed in mouse hippocampal neurons, where retromer stabilization decreased Ab 40 and 42 peptides equally (Mecozzi et al., 2014). We observed variability among neuronal cultures in the amount of increase in VPS35 stability, which could be due to inherent variability between hiPSC lines, individual patient genomes, or the limited amount of material obtained from our purified neuron protocol ( Figure S2A). However, in representative experiments, we documented detectable increases in VPS35 stability from purified neuronal cultures ( Figure S2B). We next asked if retromer stabilization by R33 had an effect in genetic backgrounds of penetrant mutations leading to early-onset, FAD, either by increasing the levels of APP (APP duplication, APP Dp ) or by enhancing beta cleavage of APP (APP Swedish, APP Swe ). In either patient neurons from FAD APP Dp (Israel et al., 2012) or in isogenic neurons in which the Swedish mutation was introduced via CRISPR/Cas9 genome editing ( Figure S2D) (Woodruff et al., 2016), we observed a similar folddecrease in Ab peptides from APP Dp patients and the APP Swe isogenic neurons as with the NDC and SAD patient neurons (Figures 2A and 2B), suggesting that this small molecule chaperone is a potent reducer of Ab, despite patient or cell line genetic background. Indeed, we observed a significant decrease in all forms of APP processing ( Figure 2C) and observed a reduction in APP b-CTF in purified neurons treated with R33 ( Figure 2D). Interestingly, we did not observe a significant increase in FL APP ( Figure S2C), although this may be due to the variability of APP expression across all our different patient genetic backgrounds and the dynamic process between APP cleavage and recycling, where changes in APP cleavage may affect the overall expression of APP. Future experiments are needed to determine how retromer stabilization affects these processes. Taken together, these data corroborate previous work in the mouse and suggest that retromer stabilization keeps APP out of intracellular compartments that generate b-cleavage of APP and, subsequently, Ab. Israel et al. (2012) and Young et al. (2015). FACS purified neural stem cells and purified neurons were generated following previously published methods (Israel et al., 2012;Yuan et al., 2011). (B and C) Ab 1-40 (B) and Ab (Hardy and Higgins, 1992), while other evidence suggests that these pathways can occur independently (Small and Duff, 2008). Neurons purified from hiPSCs have detectable levels of phosphorylated TAU on Thr231 and this correlates with increases in Ab from APP Dp patients (Israel et al., 2012). We tested the effect of retromer stabilization by R33 on the pTAU/total TAU (tTAU) ratio on our cohort of patient neurons and found that R33 treatment significantly decreased the pTAU/tTAU ratio in all cell lines, whether they were derived from NDC or SAD individuals ( Figures 3A and 3B). Interestingly, this decrease in the ratio was due to an effect on pTAU ( Figure 3C), while the levels of tTAU in the cultures remained unchanged . R33 treatment had a similar effect on neurons derived from an APP Dp patient, with a reduction of pTAU ( Figure 3E), but not tTAU ( Figure 3F), leading to a lowered pTAU/tTAU ratio in these neurons ( Figure 3G). We probed a second pTAU epitope, paired helical fragment (PHF) TAU, and observed that in the presence of compound E, a gamma-secretase inhibitor, levels of PHF TAU increased, but when neurons were concomitantly treated with both compound E and R33, the PHF signal decreased ( Figure S3A). We tested whether R33 affected the activation of the TAU kinase GSK3b ( Figure S3B); however, we did not observe a significant effect of R33 treatment on the activity of GSK3b, suggesting that the decrease in pTAU by R33 is mediated through a different mechanism. Finally, we tested whether other molecules hypothesized to be neurotrophic or protective against AD phenotypes had a similar effect to R33 on the pTAU/tTAU ratio. We have previously shown that the neurotrophin brainderived neurotrophic factor (BDNF) reduced Ab peptides in stem cell-derived neurons and that this reduction was correlated with protective variants in the SORL1 gene, whose protein product, SORLA, is a receptor of the For each comparison, a two-tailed t test was performed. p < 0.05, p < 0.01, **p < 0.001. Error bars represent SD. See also Figure S2. retromer complex (Young et al., 2015). BDNF has been shown to reduce TAU phosphorylation, although at a different epitope than is analyzed in this study, in retinoic acid-differentiated SH-SY5Y cells (Chen et al., 2014). We treated hiPSC-derived neurons from three cell lines harboring SORL1 protective variants with BDNF and measured the pTAU/tTAU ratio. Under these conditions, however, we did not observe a significant change in the TAU ratio ( Figure S3C). Taken together, these data suggest that the decrease in the pTAU/tTAU ratio by R33 has a more specific effect on the stabilization of the trimeric core cargo-recognition complex than on the various retromer receptor or interacting proteins. Genetic Knockdown of VPS35 Increases Ab and Phospho-TAU in Stem Cell-Derived Neurons We next tested whether destabilization of the retromer complex by knockdown of VPS35 had the opposite effect of R33 treatment in hiPSC-derived neurons. Using lentiviral transduction, we expressed two independent short hairpin RNAs (shRNAs) against VPS35 mRNA (determined from a pool of four shRNAs, Figure S4A) in purified neurons and documented significant increases in Ab peptides similar to what has been previously reported in mice (Figures 4A and 4B) (Bhalla et al., 2012). We next measured phosphorylated TAU protein (Thr 231) from purified neurons transduced with two VPS35 shRNA and observed a Figure S3. small but significant increase in pTAU at Thr 231 (Figure 4C). The magnitude of the increase in pTAU after VPS35 knockdown is similar to that of the decrease in pTAU we observe after R33 treatment. Because shRNA VPS35-c gave the strongest effect, we confirmed the knockdown of VPS35 protein and VPS35 mRNA using that shRNA in multiple experiments ( Figures 4D and 4E). We consistently observed an approximately 50% knockdown of VPS35 protein in our neuronal cultures ( Figure 4D). Although we are only able to reduce the levels of VPS35 protein by half, it should be noted that germline deletion of VPS35 is results in embryonic lethality in mice and haploinsufficiency of VPS35 is sufficient to increase AD neuropathology in a transgenic mouse model (Wen et al., 2011). These data suggest that even modest changes in retromer subunit levels may have a large impact on cellular pheno-types. In terms of AD, we note that duplication of APP resulting in one extra copy of the gene and 50% more APP expression is sufficient to cause severe and early-onset AD (Rovelet-Lecrux et al., 2006). Thus, for factors that may increase or decrease risk but not the deterministic probability of AD, changes on the order of 20% are likely relevant in human disease. Phospho-TAU Reduction Correlates with, but Is Not Dependent on, Ab Peptide Levels Because the magnitude of decrease in pTAU after R33 treatment was similar to the decreases observed in APP processing, we performed a linear regression analysis on absolute levels of pTAU and absolute levels of Ab in the cultures across all 13 patient cell lines and found a positive correlation between the reduction of Ab and the reduction in pTAU ( Figure 5A). Previous work in hiPSC-derived neurons has suggested that beta cleavage of APP, rather than Ab, is more highly associated with TAU phosphorylation (Israel et al., 2012), so we also examined the levels of sAPPb compared with pTAU and observed a similar, but less robust, correlation between the reduction in sAPPb with pTAU ( Figure 5B). Taken together, however, these data suggest that the reduction in pTAU by R33 is highly correlative with a reduction in amyloidogenic processing of APP. We next asked if modulation of Ab levels affected the decrease in pTAU by R33. We measured pTAU from isogenic neurons harboring the APP Swedish mutation, which leads to a 2-to 3-fold increase in endogenous Ab in heterozygous and homozygous carriers, respectively ( Figure 5C). Interestingly, we did not observe increased levels of pTAU/tTAU ratio in cells harboring these mutations alone. Previous work using isogenic cell lines of another penetrant AD mutation, PSEN1DE9, also failed to show increased pTAU/tTAU ratios at baseline conditions (Woodruff et al., 2013). However, treatment with R33 reduced the pTAU/tTAU ratio in the isogenic neurons to the same extent across this allelic series, with no significant difference in R33-mediated TAU ratio reduction between the wild-type cells and the Swedish genotypes ( Figure 5D), suggesting that retromer stabilization reduces TAU phosphorylation regardless of genetic condition. To test whether modulating level of the APP b-CTF in our cell cultures affected the R33-mediated reduction of the TAU ratio, we treated wild-type purified neurons with the gamma-secretase inhibitor compound E, which inhibits the generation of Ab but increases the amount of the b-CTF of APP. Interestingly, R33 treatment further reduced the amount of Ab remaining after compound E treatment ( Figure 5E). The presence of compound E had no effect on the decrease pTAU/tTAU ratio after treatment with R33 ( Figure 5F). Taken together, these data suggest that while the pTAU/tTAU reduction by R33 is highly correlative with Ab peptide levels, modulation of APP processing, either up or down, does not change the R33-mediated reduction of pTAU. R33 Reduction of Phospho-TAU in hiPSC-Derived Neurons Is Independent of APP Expression Because hiPSCs are amenable to genome engineering, we generated an APP knockout (APP KO) hiPSC line from our parental APP duplication hiPSC line (APP Dp 1) using CRISPR/Cas9 genome editing. APP KO hiPSCs were generated from FAD APP Dp parental line by excision of one copy of APP and the introduction of two premature stop codons via non-homologous end-joining in the remaining copies (R. Van der Kant et al.,personal communication;Figures S5A and S5B). APP KO hiPSCs have no detectable APP protein by western blot analysis ( Figure S5C). We differentiated and purified neurons from APP KO hiPSCs and confirmed that these neurons do not have detectable levels of secreted Ab peptides by ELISA ( Figure 6A). We next treated APP KO neurons with R33 and observed decreased pTAU from APP KO neurons ( Figure 6B) with no change in tTAU ( Figure 6C), again leading to a significant reduction in the pTAU/tTAU ratio in APP KO neurons ( Figure 6D). When APP KO neurons were transduced with shVPS35-c, we documented a significant increase in the amount of pTAU ( Figure 6E), and this effect was confirmed with a second shRNA, shVPS35-d ( Figure S6A). Finally, we tested whether VPS35 knockdown would change the effect of R33 on pTAU levels in APP KO neurons. Interestingly, knockdown of VPS35 did not affect the R33-mediated reduction in pTAU in these cells ( Figure S6B). However, this result is not overly surprising, as we have an incomplete loss of VPS35 (50%) and the R33 chaperone is still stabilizing the remaining retromer. Indeed, previous work demonstrates that retromer stabilization chaperones continued to stabilize two of the three cargo-recognition core proteins (VPS29 and VPS35) under single knockdown conditions (shVPS26) (Mecozzi et al., 2014). Taken together, these experiments suggest that retromer stability influences TAU phosphorylation and that this is mediated through an APP-independent mechanism. Furthermore, these data imply that retromer enhancement may be beneficial even under less than optimal retromer conditions, such as loss of VPS35 expression, a condition documented in AD patient tissue (Small et al., 2005). DISCUSSION In this work, we demonstrate that a small molecule stabilizer of the retromer complex reduces endogenous cellular AD phenotypes in human neurons. This small molecule, R33, has been previously shown to reduce Ab peptides and decrease APP in early endosomes, consistent with its in vitro action as a pharmacological chaperone that stabilizes the retromer complex (Mecozzi et al., 2014) (Chu and Pratico, 2017). We extended this work in human neurons to test how retromer stabilization affects TAU phosphorylation, another critical pathologic phenotype in AD and one that correlates more highly with cognitive decline than the presence of amyloid plaques in patients (Arriagada et al., 1992). In this work, we demonstrate that retromer stabilization in human neurons reduces TAU phosphorylation at two epitopes, suggesting that this strategy may be beneficial for pathologies extending beyond amyloid. Indeed, while the focus on Ab as the premier pathological event has undoubtedly yielded insights into the disease, particularly for FAD, it is equally conceivable to envision a ''dual-pathway'' hypothesis where an upstream molecular driver influences Ab and TAU pathologies separately (Small and Duff, 2008). This may especially be the case for SAD, where immense genetic heterogeneity and complex environmental interactions have unknown influences on the course of the disease. Here we decouple the effect of retromer stabilization on both APP processing and TAU phosphorylation using gene-edited hiPSC-derived neurons. In both neurons engineered to have 2-to 3-fold higher amounts of Ab (APP Swe ) or neurons engineered to not express APP and which have no detectable Ab (APP KO), retromer stabilization decreases TAU phosphorylation to the same extent, suggesting that manipulation of endocytic trafficking via retromer may be an upstream driver of these two pathologies independently from one another. This approach of gene editing in pluripotent stem cells allows the dissection of relevant disease biology in a human-specific system. The link between retromer regulation of endosomal trafficking and TAU phosphorylation is currently unclear, although compelling studies are emerging to suggest that endocytic trafficking may directly affect TAU pathology. Michel et al., 2014, suggest that monomeric TAU is endocytosed from the extracellular space and its transit through endosomes facilitates TAU seeding and aggregation. Work in flies demonstrates that removing cathepsin D, a retromer cargo and integral enzyme in lysosomal degradation, exacerbates TAU-induced neurodegeneration (Khurana et al., 2010). Extracellular TAU can be taken up and propagated via exosomes in microglia, the resident immune cells of the CNS (Asai et al., 2015), and retromer deficiency impairs the recycling of TREM2 protein, a product of the TREM2 gene that is also associated with increased AD risk, which can inhibit microglial activation (Yin et al., 2016). Developing methods of improving endocytic trafficking and function in multiple CNS types will likely be protective in many neurodegenerative disorders, underscoring the importance of elucidating the role of the endo-cytic network in both Ab and TAU pathology. Finally, we highlight the potential of hiPSC models to confirm and elaborate on work done in non-human systems. We have replicated work previously published in mice to show that retromer stabilization decreases pathogenic APP processing (Mecozzi et al., 2014) and extended our studies to show a beneficial effect of retromer stabilization on TAU in human neurons, which, to our knowledge, has not been documented previously. Furthermore, we demonstrate that this process correlates with, but is not dependent upon, APP processing. Considering that the current failure rate of AD clinical trials is very high (Posner et al., 2017) early-and late-onset AD (Bellenguez et al., 2017;Kunkle et al., 2017;Rosenthal and Kamboh, 2014). For that reason, and because defects in the endocytic pathway occur early in AD pathogenesis, targeting this pathway represents a valid cell biological target for therapeutic development (Small et al., 2017). The use of hiPSCs, which can be differentiated into all cell types of the CNS, represents an ideal model to test novel and existing drugs and other therapies in a preclinical system. In particular, the advantage of a system that does not rely on overexpression of disease proteins has already been shown: patient-derived neurons behave differently from APP-overexpressing cell lines and animal AD model-derived neurons when exposed to drug candidates, such as gamma-secretase modulators (Liu et al., 2014). Recently, cortical neurons from hiPSCs have been used to screen modulators of Ab peptide length (Brownjohn et al., 2017), and work in scalable three-dimensional culture formats has shown detection of TAU aggregates in a quantifiable manner in 384-well plates (Medda et al., 2016), representing a growing technology involving complex cellular interactions that will likely be highly translational. Abnormalities in the endo/lysosomal network are a common theme in multiple neurodegenerative disorders (Hu et al., 2015;Schreij et al., 2016), and retromer-related mutations or defects are implicated not only in AD, but Parkinson disease and Down syndrome, as well as other nonneurological conditions (Berman et al., 2015). Our work, for the first time, examines a small molecule that enhances the function of retromer, which is an integral component of the endosomal network, using human neurons. This pharmacologic chaperone, R33, demonstrated pronounced and potentially beneficial effects on the two main pathological molecules in AD: Ab and pTAU. We predict that using hiPSCs, both from patients and isogenic lines with key mutations or variants, to test and screen novel therapeutic molecules and other therapeutic strategies will represent a major advance in therapeutic development for AD and other complex neurodegenerative disorders. Cell Lines and Cell Culture All patient-derived hiPSC lines were generated by four-factor reprogramming and have been previously characterized and published (Israel et al., 2012;Young et al., 2015). Briefly, hiPSCs were maintained on a mouse embryonic fibroblast (MEF) feeder layer in HUES medium and were routinely tested for mycoplasma. NSCs and neurons were generated using previously described methods (Israel et al., 2012;Yuan et al., 2011) and purified by fluorescence-activated cell sorting (FACS). After FACS purification, de-pending on the percentage of neurons in each differentiation, 2 to 4 independent purified cultures were re-plated for each experiment. From these independent cultures and for each assay, 2 to 4 technical replicates from each well were analyzed. CRISPR/Cas9 Gene Editing Genome-edited lines were generated using published protocols. Briefly, guide RNAs (gRNAs) to APP were generated using the Zhang Lab Crispr design Web site at MIT (crispr.mit.edu) and cloned into vector px458 (Addgene) that co-expresses the Cas9 nuclease and GFP. hiPSCs were electroporated with the px458 plasmid containing the gRNAs to target APP. We generated the APP KO cells from our parental APP Dp cell line (APP Dp 1), first described in Israel et al., 2012. To generate APP KO cells, no repair template was added. For generating the APP Swedish mutation, we targeted wild-type hiPSCs derived from J. Craig Venter (Gore et al., 2011) and 120-bp single-stranded oligonucleotide was designed to have the 2-bp Swedish mutation (GA to TC, corresponding to exon 16 of APP). Electroporated hiPSCs were sorted for GFP and plated a clonal density on 10-cm MEF plates. After 1.5 to 2.0 weeks, sub-cloned colonies were picked and transferred to a 96-well plate. Colonies were spit in duplicate and one set was maintained for cell line generation and the other was analyzed for the mutation of interest by Sanger Sequencing. All gene-edited lines were digitally karyotyped as euploid by the Illumina Infinium HumanCoreExomeBeadChip, as has been previously described (Young et al., 2015) Drug Treatment and ELISA Assays Pure R55 and R33 were obtained from MedKoo Biosciences (products TPT-260 and TPT-172) and re-suspended at designated concentrations in a solution of 50% DMSO/50% double-distilled H 2 O at low pH (pH 4.3). For these studies, R33 was used for every experiment. Purified neurons were treated with R33 or vehicle (DMSO/H 2 O) for 72 hr and then harvested for analysis. For Ab and TAU ELISA assays, purified neurons were seeded at 200,000 cells/well of a 96-well plate. After treatment, cell culture media and cell lysate were harvested from triplicate wells and run on Ab Triplex ELISA plates and pTAUThr231/tTAU ELISA plates (Meso Scale Discovery). Protein concentrations were determined by BCA assay (Pierce) and Ab levels were normalized to the amount of total protein. Western Blotting Cell lysates were run on NuPAGE 4% to 12% Bis-Tris Gels, transferred to nitrocellulose membranes, and probed with antibodies to VPS35 (Abcam #ab57632), APP (Life Tech #512700), TAU (Thermo #MN1000 and MN#1020). Quantification of western blots was performed on the Odyssey system and using ImageJ software. shRNA Design and Virus shRNAs against VPS35 were ordered from Origene (#TL300553). VPS35 and scrambled control shRNAs were packed into lentivirus using HEK293FT cells. Virus was purified by PEG-it and cells were transduced with lentivirus for 72 hr. Knockdown of VPS35 was confirmed by qRT-PCR and western blot analysis. gRNA, ssODN, and Primer Sequences gRNA to generate APP Swedish and APP KO cells: GGAGATCTCT GAAGTGAAGA TGG. ssODN for APP Swedish homologous recombination 2-bp change in bold: TGATGTAATACAGGTTCTGGGTTGACAAATATC AAGACGGAGGAGATCTCTGAAGTGAATCTGGATGCAGAATTCC GACATGACTCAGGATATGAAGTTCATCATCAAAAATTGGTACGT. Statistical Methods Statistics for all datasets were performed using GraphPad Prism. Normality for each dataset was assessed using the D'Agostino-Pearson test. When data were normally distributed, a parametric twotailed t test was used. The comparisons between multiple cells lines from different individuals usually resulted in non-normally distributed data and these data were analyzed using the Wilcoxon-signed rank test.	2018-04-03T02:27:38.914Z	2018-03-01T00:00:00.000	{ "year": 2018, "sha1": "928dc8f1469cf42378f6d5302ce0857a5ab65bfa", "oa_license": "CCBYNCND", "oa_url": "http://www.cell.com/article/S2213671118300572/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "928dc8f1469cf42378f6d5302ce0857a5ab65bfa", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
250588787	pes2o/s2orc	v3-fos-license	Friedelin Attenuates Neuronal Dysfunction and Memory Impairment by Inhibition of the Activated JNK/NF-κB Signalling Pathway in Scopolamine-Induced Mice Model of Neurodegeneration Oxidative stress (OS) and c-Jun N-terminal kinase (JNK) are both key indicators implicated in neuro-inflammatory signalling pathways and their respective neurodegenerative diseases. Drugs targeting these factors can be considered as suitable candidates for treatment of neuronal dysfunction and memory impairment. The present study encompasses beneficial effects of a naturally occurring triterpenoid, friedelin, against scopolamine-induced oxidative stress and neurodegenerative pathologies in mice models. The treated animals were subjected to behavioural tests i.e., Y-maze and Morris water maze (MWM) for memory dysfunction. The underlying mechanism was determined via western blotting, antioxidant enzymes and lipid profile analyses. Molecular docking studies were carried out to predict the binding modes of friedelin in the binding pocket of p-JNK protein. The results reveal that scopolamine caused oxidative stress by (1) inhibiting catalase (CAT), peroxidase enzyme (POD), superoxide dismutase (SOD), and reduced glutathione enzyme (GSH); (2) the up-regulation of thiobarbituric acid reactive substances (TBARS) in mice brain; and (3) affecting the neuronal synapse (both pre- and post-synapse) followed by associated memory dysfunction. In contrast, friedelin administration not only abolished scopolamine-induced oxidative stress, glial cell activation, and neuro-inflammation but also inhibited p-JNK and NF-κB and their downstream signaling molecules. Moreover, friedelin administration improved neuronal synapse and reversed scopolamine-induced memory impairment accompanied by the inhibition of β-secretase enzyme (BACE-1) to halt amyloidogenic pathways of amyloid-β production. In summary, all of the results show that friedelin is a potent naturally isolated neuro-therapeutic agent to reverse scopolamine-induced neuropathology, which is characteristic of Alzheimer’s disease. Introduction Neurodegenerative disorders (ND) are characterized by impaired structures and functions of neurons or neuronal cells. Parkinson's disease, multiple sclerosis, Huntington's disease and Alzheimer's disease (AD), along with associated dementia, are major examples of ND [1]. AD being the leading cause of memory impairment and dementia among geriatrics has the most devastating effect on cognitive skills leading to emotional distress. The etiological characterization of AD has yet to be completely explicated, however, the neuropathological effects of amyloid-β (Aβ) aggregates and neurofibrillary tangles (NFTs) owing to hyperphosphorylated tau proteins have been identified [2]. The Aβ aggregates are deposited on extracellular surfaces of the neuronal cells due to the activation of three mitogen activating protein kinases (MAPK) pathways in neurons i.e., c-Jun N-terminal kinase (JNK), p38, extracellular signal-regulated protein (ERK) and nuclear factor-κB (NF-κB) ultimately leads to neurodegeneration [3]. Specifically, the up-regulation of NF-κB activity not only exacerbates Aβ production and deposition, but also enhances pro-inflammatory cytokines and chemokine's expression inducing AD neurotoxicity [4]. NF-κB is found to be overexpressed in degenerated cholinergic neurons and the proximal glial nuclei of AD affected brains [5]. Scopolamine-induced neurodegenerative models are widely used for identification of neuroprotective agents, as it is associated with enhanced acetyl cholinesterase (AChE) activity, increased reactive oxygen species (ROS) levels in brain, increased the levels of inflammatory mediators such as COX-II in brain, and increased levels of thiobarbituric acid reactive substances (TBARS) [6]. With regard to AD, scopolamine induces reversible symptoms that mimic dementia by deregulating cholinergic function and promoting Aβ deposition, both being the hallmarks of the said disease [7]. As of today there are few FDA-approved drugs in the market for treatment, and several researchers have dedicated their research/studies on plant-derived or isolated compounds as a potential source of either novel or more effective neuro-therapeutic agents [8]. Pentacyclic triterpenes are the most abundantly found and versatile group of naturally occurring compounds. Both the anti-inflammatory and anti-oxidant activities of various pentacyclic triterpenes such as oleanolic acid, ursolic acid, maslinic acid, corosolic acid, erythrodiol and glycyrrhizic acid have already been established in murine models [9,10]. Therefore, in the current study, the aforementioned attributes of friedelin, a pentacyclic triterpenoid, were sought against neuropathologies specifically associated with AD. Friedelin belongs to class of pentacyclic triterpenes ( Figure 1) abundantly found in several species of plants. Friedelin is reported to possess anti-cancer [11], anti-ulcerogenic [12], anti-inflammatory, antipyretic, analgesic [13], in vitro antioxidant [14], hepatoprotective, anti-hyperlipidemic [15], and neuroprotective properties by hindering Aβ aggregation in in vitro models [16]. Owing to the necessity for the testing of this compound in in vivo models, it creates the baseline for the current investigation, in which the effects of friedelin was evaluated in an animal model of scopolamine-induced oxidative stress mediated phosphorylated JNK activation, memory impairment, neurodegeneration, neuroinflammation and synaptic dysfunction were appraised. This may be attributed to its antioxidant/antineuroinflammatory/anti-neuroapoptotic effects through the inhibition of phosphorylated JNK protein. Neurodegenerative disorders (ND) are characterized by impaired structures and functions of neurons or neuronal cells. Parkinson's disease, multiple sclerosis, Huntington's disease and Alzheimer's disease (AD), along with associated dementia, are major examples of ND [1]. AD being the leading cause of memory impairment and dementia among geriatrics has the most devastating effect on cognitive skills leading to emotional distress. The etiological characterization of AD has yet to be completely explicated, however, the neuropathological effects of amyloid-β (Aβ) aggregates and neurofibrillary tangles (NFTs) owing to hyperphosphorylated tau proteins have been identified [2]. The Aβ aggregates are deposited on extracellular surfaces of the neuronal cells due to the activation of three mitogen activating protein kinases (MAPK) pathways in neurons i.e., c-Jun N-terminal kinase (JNK), p38, extracellular signal-regulated protein (ERK) and nuclear factor-κB (NF-κB) ultimately leads to neurodegeneration [3]. Specifically, the up-regulation of NF-κB activity not only exacerbates Aβ production and deposition, but also enhances pro-inflammatory cytokines and chemokine's expression inducing AD neurotoxicity [4]. NF-κB is found to be overexpressed in degenerated cholinergic neurons and the proximal glial nuclei of AD affected brains [5]. Scopolamine-induced neurodegenerative models are widely used for identification of neuroprotective agents, as it is associated with enhanced acetyl cholinesterase (AChE) activity, increased reactive oxygen species (ROS) levels in brain, increased the levels of inflammatory mediators such as COX-II in brain, and increased levels of thiobarbituric acid reactive substances (TBARS) [6]. With regard to AD, scopolamine induces reversible symptoms that mimic dementia by deregulating cholinergic function and promoting Aβ deposition, both being the hallmarks of the said disease [7]. As of today there are few FDA-approved drugs in the market for treatment, and several researchers have dedicated their research/studies on plant-derived or isolated compounds as a potential source of either novel or more effective neuro-therapeutic agents [8]. Pentacyclic triterpenes are the most abundantly found and versatile group of naturally occurring compounds. Both the anti-inflammatory and anti-oxidant activities of various pentacyclic triterpenes such as oleanolic acid, ursolic acid, maslinic acid, corosolic acid, erythrodiol and glycyrrhizic acid have already been established in murine models [9,10]. Therefore, in the current study, the aforementioned attributes of friedelin, a pentacyclic triterpenoid, were sought against neuropathologies specifically associated with AD. Friedelin belongs to class of pentacyclic triterpenes ( Figure 1) abundantly found in several species of plants. Friedelin is reported to possess anti-cancer [11], anti-ulcerogenic [12], anti-inflammatory, antipyretic, analgesic [13], in vitro antioxidant [14], hepatoprotective, anti-hyperlipidemic [15], and neuroprotective properties by hindering Aβ aggregation in in vitro models [16]. Owing to the necessity for the testing of this compound in in vivo models, it creates the baseline for the current investigation, in which the effects of friedelin was evaluated in an animal model of scopolamine-induced oxidative stress mediated phosphorylated JNK activation, memory impairment, neurodegeneration, neuroinflammation and synaptic dysfunction were appraised. This may be attributed to its antioxidant/anti-neuroinflammatory/anti-neuroapoptotic effects through the inhibition of phosphorylated JNK protein. Materials and Methods Molecular docking methodology: Molecular docking studies were performed using AutoDock Vina in the PyRx 0.8 [17], to predict binding modes of friedelin in the binding pockets of JNK protein retrieved from the Protein Data Bank (PDB) having ID: 3V6S (chainA). The structure of Friedelin was created with ChemSketch and converted into PDB format using the Discovery Studio Visualizer (DSV), which was then transformed to Protein Data Bank, Partial Charge, and Atom Type format (PDBQT) for advance molecular docking evaluation. The 3D structure of the target protein, i.e., JNK, was downloaded from the PDB (PDB ID 3v6s) with a resolution 2.97 Å and R-Value 0.263 (Free), 0.216 (Work), 0.219 (Observed). The 3D structure of the target protein, JNK, was obtained experimentally by the X-ray diffraction method [18]. The downloaded protein was opened in DSV, and the H 2 O molecules and chain B were removed. Protein 3D protonation was carried out followed by the minimization of energy for the stability of the target protein by means of the default PyRx parameters. Default settings of AutoDock Vina were used for the docking studies. Twenty conformations were formed and top ranked conformation was selected on the bases of the docking score for additional analysis. Chemicals: Scopolamine, phosphate buffer saline tablets (PBS), sodium dodecyl sulphate (SDS), ammonium per sulphate (APS), acrylamide, bis-acrylamide, trizma base, potassium chloride (KCl), and sodium chloride (NaCl) were procured from Sigma Aldrich Chemical Co., St. Louis, MO, USA and monoclonal antibodies were purchased from Daejung Chemicals & Metals Co. Ltd., Siheung-si, Korea. The isolated test compound (friedelin) was provided by the courtesy of Natural Product Research Lab, Department of Pharmacy, Quaid-i-Azam University, Islamabad, Pakistan. Prior work has been carried out for the isolation of the subject compound from Quercus dilatata L., based upon the existing indigenous knowledge of medicinal uses of this plant specie. Friedelin was isolated from the n-hexane fraction of crude methanol extract of aerial parts of Quercus dilatata by bioassay guided separation techniques [19]. The purity of the compound was confirmed by HPLC-DAD analysis (15 mg, purity > 99%) carried out at the Natural Product Research Lab, Department of Pharmacy, Quaid-I-Azam University, Islamabad, Pakistan. Briefly, the RP-HPLC based purification of friedelin was checked on a Shimadzu HPLC system comprised of a SCL-10AVP controller, a DGU-14A degasser, a FCV-10ALVP low pressure mixer, an LC-10AT VP pump coupled with 3D-PDA detector (SPD-M10A VP), and LC solution software. The column used was an Agilent Zorbax C8 (5µm; 4.6 mm × 250 mm). The elution was done isocraticaly using the mobile phase which was composed of methanol and distilled water (9:1). The flow rate was 1 mL/min and the injection volume was 50 µL. The chromatogram was obtained at 207 nm (at λ max ). The retention time of the friedelin standard and sample was 4.1 min with a total run time of 10 min. The concentration of the standard and sample was 100 µg/mL and 400 µg/mL, respectively. The concentration of the sample was kept fourfold higher than the standard to detect the impurities. The standard and sample were compared using retention time, a 3D-graph and UV-spectrum. The comparative chromatograms and 3D graph are given in Table 1. The original chromatograms with additional parameters have also been given in the mentioned table. Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Standard Sample Molecules 2021, 26, x FOR PEER REVIEW 4 of 16 Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The Table 1. Chromatogram, 3D graph and UV spectrum of the friedelin sample and the standard. Spectrum Spectrum Mice grouping: To conduct these experiments, adult male Swiss albino mice were purchased from the NIH (National institute of Health) Veterinary sub division, and were brought to the Neuro Molecular Medicines Research Centre, (NMMRC), Peshawar. The number of animals in each group were calculated using the resource equation approach [20] and keeping the 3Rs (replacement, reduction and refinement) guidelines [21] for the use of animals in research in consideration. The experimental mice were placed in labelled Biobase cages with free access to food and water, and were allowed to acclimatize to the environment. The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of the University of Sargodha (protocol code SU/ORIC/2862 and date of approval 23 September 2021). The male mice of 30-32 g mean body weight were placed in separate cages with a controlled environment, i.e., light/dark cycle (12/12 h), 25 ± 01 • C temperature and ad libitum supply of clean water and standardized pellet food. As per animal ethics committee recommendations, all of the research animals were handled with care and empathy. The doses of scopolamine and friedelin were selected and optimized from the literature review [22]. Friedelin isolated from different plants had already been reported to have significant in vivo activities in animal models of hepatotoxicity [15], gastro-protective activities [12], and anti-inflammatory and antipyretic activities [23]. The mice were later randomly distributed into four groups (n = 5), C = control group (vehicle control group), S = scopolamine treated group (1 mg/kg), S + F = scopolamine (1 mg/kg) + friedelin (30 mg/kg) and F = friedelin treated group (30 mg/kg). Drug administration and experimental design: All experimental drugs were dissolved in normal saline and were dispensed interaperitonealy (i.p.). The animals were caught by tail with a hand wrapped around the neck region in a way to expose their ventral side upwards. The drug was administered in the left lower quadrant of the abdominal cavity with caution to avoid internal organ damage. Scopolamine was injected i.p. for 21 days, while the test drug, friedelin, was administered every day for the last 14 days. At the 10th day from the start of drug induction the training of mice (for behavioural tests) was started, and this continued for three days; after a rest period of two days the mice were subjected to the final tests for five consecutive days after 30 min of the drug's administration. After one day of rest, the final probe test was carried out and post-test the mice were subjected to anaesthesia for blood collection through cardiac puncture, and afterwards brains were collected. Behavioural Tests To demonstrate the therapeutic efficacy of friedelin on scopolamine-mediated memory dysfunction, two well-known behavioural tests were executed. Scopolamine was injected interaperitonealy (i.p.) to the mice of group II and III without and with friedelin, respectively. Later, all mice were tagged and randomly divided into four (04) groups, while the behavioural studies were carried out as a single blind trial. The researcher carrying out the behavioural tests was kept blind of the tags and the mice treatment group. Morris Water Maze (MWM) Test: To probe the hippocampus dependent long-term spatial learning capabilities of mice, the MWM behavioural task was performed. The design and dimensions of the MWM test apparatus were according to the previously reported details [24]. Before the actual commencement of the tests, mice were trained for swimming (two times/day) to acclimatize them with the water tank and platform for three days. Later, the escape latency for each mouse was calculated for a period of 60 s to find the hidden/submerged platform, and this practice continued for five days. If mouse failed to find the platform in given time (60 s) it was manually directed and placed on a platform for at least 10 s. The escape latency time (sec) for each day was recorded and maintained. Mice were placed on rest for a period of two days before final probe testing, i.e., the platform was hidden and the mean time spent by each mouse in the target quadrant was calculated. Y-Maze test: The Y-maze behavioural task was performed as executed earlier [25]. The Y-maze apparatus consists of three arms with dimensions of 50 × 10 × 20 cm 3 (L × W × H) connected at an angle of 120 • to each other. The mice were acclimatized to this new environment for 10 min each time. Afterwards, each mouse was placed in the middle of the maze one by one and was allowed to explore the maze spontaneously in all three arms for 8 min. Total arm entries of each mouse and successive triplet entries were computed employing software and the percentage of alternations was calculated using the following formula: The spatial memory function of mice was correlated positively with the percentage of alternations. Western Blotting Analysis All of the mice were euthanized after being given anesthesia by mild chloroform inhalation at the end of the 21st day of treatment, and behavioral studies mentioned in above sections. The mice were decapitated and the hippocampal part of the brain was carefully extracted, then quickly transferred to RNAlater (Sigma Ald. R0901) solution and PBS (1:1) and placed on ice. The hippocampal tissue was homogenized in total protein extraction reagent (T-PER by Thermo Scientific 1632086, Waltham, MA, USA) solution, and the tissue supernatant was collected and stored at −20 • C for future analysis. The total protein concentration was quantified employing a Bio-Rad protein estimation assay kit (5000001EDU) and absorbance was taken at 595 nm. All of the protein samples were normalized to 30 µg/group, and gel electrophoresis was performed using SDS-PAGE (12-15%). Running conditions were maintained at 50 mA for the first 20-30 min and then switched to 120 V for almost 1-1.5 h until the run was complete. Proteins from gel were trans-blotted to a PVDF membrane using the semi-dry transblott technique. Several mouse derived monoclonal primary antibodies procured from Santa Cruz Biotech Inc. (Dallas, TX, USA), such as anti-Iba (sc-32725), anti-GFAP (sc-33673), anti-p-JNK (sc-6254), anti-caspase-3 (sc-7272), anti-NF-κB (sc-8008), anti-PARP-1, anti-COX-II (sc-376861), anti-βactin (sc-47778), anti-Aβ (sc-28365), anti-SYP (sc-17750), anti-PSD95 (sc-71933), anti-BACE1 (sc-337) and antibodies of dilution 1:1000 in PBS were applied followed by the application of 1:2000 anti-mouse HRP conjugated secondary antibody (0000375517). The X-ray films were developed and analyzed using ImageJ software [26]. Biochemical Analysis of Plasma Upon the completion of drug treatment, the mice were euthanized and the collected blood was used for biochemical analysis of total cholesterol, high density lipoproteins (HDL), low density lipoproteins (LDL), very low density lipoproteins (VLDL), and triglycerides levels (TGL). Antioxidant Enzyme Profiling of Brain Homogenates Catalase assay (CAT): Catalase activity (CAT) was assessed through the method developed earlier with a few modifications [27]. The 3 mL of reaction mixture consisted of 2500 µL phosphate buffer (50 mM, pH 5.0), 400 µL of H 2 O 2 (5.9 mM) and 100 µL of brain homogenate supernatant. Variation in absorbance of the reaction mixture was taken at 1 min intervals at 240 nm. The change in absorbance of 0.01 units/min was considered as one unit of activity. Peroxidase assay (POD): The previously reported method with slight modifications was employed to measure the peroxidase activity [27]. The reaction mixture for peroxidase assay consists of 2500 µL phosphate buffer (50 mM) at pH 5.0, 300 µL H 2 O 2 (40 mM), 100 µL of guaiacol (20 mM) and 1000 µL supernatant of brain homogenate. The variation in reaction mixture absorbance was taken at one min intervals at 470 nm. One unit of POD activity was regarded as a change in absorbance of 0.01 units/min. Superoxide dismutase assay (SOD): This assay was performed with a slight modification, as reported earlier [27]. To estimate the superoxide dismutase (SOD) activity reaction mixture consists of 100 µL phenazine methosulphate (186 µM), 1200 µL sodium pyrophosphate buffer (0.052 mM, pH 7.0) and 300 µL brain homogenate supernatant. To initiate enzymatic activity, 200 µL of NADH (780 µM) was added to the reaction mixture, and then after 1 min, 1000 µL of the stopping agent, glacial acetic acid, was added. The absorbance of reaction mixture at 560 nm was measured to determine the amount of chromogen formed, and results were expressed as units/mg of protein. Reduced glutathione assay (GSH): For estimation of the reduced glutathione levels, the proteins were precipitated in 1000 µL of brain homogenate mixed with equal parts of 4% sulfosalicylic acid solution according to the method previously reported [28]. The reaction mixture was incubated at 4 • C for 1 h, then centrifuged for 20 min at 4 • C and 1200× g. The reaction mixture contained 2700 µL 0.1 M phosphate buffer of pH 7.4, 100 µL centrifuged aliquot and 200 µL (100 mM) DTNB solution. The immediate absorbance of the mixture was recorded at 412 nm. The reduced glutathione levels were stated as µM/g brain tissue. Estimation of lipid peroxidation (TBARS): A lipid peroxidation (TBARS) assay was carried out after slight modifications in the method developed earlier [29]. In the current assay, 1000 µL of the reaction mixture was comprised of 580 µL of 0.1 M phosphate buffer (pH 7.4), 200 µL 100 mM ascorbic acid, 200 µL brain homogenate supernatant, and 20 µL 100 mM ferric chloride. The final mixture was incubated for 1 h in a shaking water bath assembly maintained at 37 • C. 1000 µL of 10% trichloroacetic acid (TCA) solution was added as a stopping agent. Next, 1000 µL of 0.67% thiobarbituric acid (TBA) was added to the reaction tubes and these were placed in a boiling water bath for 20 min and then quickly shifted to a crushed ice bath and later centrifuged for 10 min at 2500× g. The amount of lipid peroxidation (TBARS) formed in all of the samples were quantified by taking the spectrophotometric absorbance of the supernatant at 535 nm. The results were expressed in units nM TBARS/min/mg of brain tissue at 37 • C (molar extinction coefficient of TBARS: Statistical Analysis: All data were represented as Mean ± SEM and evaluated using a one-way ANOVA followed by a Tukey's multiple comparison's posttest. The X-rays of the western blot were scanned, compiled and the statistical analysis of relative densities was carried out using imageJ, GraphPad Prism5, and Adobe Photoshop. The densities of the proteins were expressed in arbitrary units (A.U.s) given as the Mean ± S.E.M. The "#" symbol denotes the significant difference from normal control to treated groups, while "" denoted a significant difference from the scopolamine treated mice group, respectively; , # p < 0.05, , ## p < 0.01, , ### p < 0.001. Results Friedelin Reduced Scopolamine-induced Oxidative Stress Mediated Glial Cells Activation in Mice: Scopolamine was administered to the adult albino mice for three weeks once daily. After completion of the first week, friedelin was administered for the final two weeks and the brain homogenates of all mice were tested for the estimation of diverse antioxidant enzymes, i.e., SOD, POD, GSH, CAT and TBARS to establish the extent of oxidative stress induced by scopolamine. The results showed that scopolamine administration significantly amplified the oxidative stress in the mice brain by decreasing the SOD, POD, GSH, CAT levels and inducing TBARS activity in the adult mice brain. On the contrary, the administration of friedelin significantly increased the enzymatic activity of SOD, POD, GSH and CAT while TBARS activity declined, as shown in Figure 2a-e. The oxidative stress characterized by an upsurge in reactive oxygen species and the down regulation of the counterfeit antioxidant system in the body. One of the main reasons for progressive neurodegenerative diseases, specifically AD, is the state of chronic oxidative stress and the dysregulation of the inflammatory responses [30,31]. The glial cells' activation is considered as an important consequence in response to oxidative stress conditions [32]. Normally, glial cells including astrocytes and microglia are involved in the maintenance of homeostasis in the neuronal environment by regulating the movement of ions or neurotransmitters through the axons and synapses [33]. Scopolamine can lead to the activation of glial cells as well as impaired cellular antioxidant defence mechanisms, resulting in a progressed disease condition. In the current study, friedelin significantly inhibited the protein expressions of glial cells, which confirms its role as a counterfeit moiety for the management of oxidative stress, as shown in Figure 2f-h. to the activation of glial cells as well as impaired cellular antioxidant defence mechanisms, resulting in a progressed disease condition. In the current study, friedelin significantly inhibited the protein expressions of glial cells, which confirms its role as a counterfeit moiety for the management of oxidative stress, as shown in Figure 2f-h. Friedelin abolished the amyloidogenic pathway of Aβ production against Scopolamine: Research findings have confirmed the participatory role of scopolamine for producing Aβ in animal brain [7]. In this study we have investigated the level of expression of Aβ protein and BACE-1 in all groups of experimental animals using the western blotting technique. The immuno-blot results clearly demonstrate that scopolamine administration significantly triggered β-secretase activity to cut the amyloid precursor protein at the sites of Friedelin abolished the amyloidogenic pathway of Aβ production against Scopolamine: Research findings have confirmed the participatory role of scopolamine for producing Aβ in animal brain [7]. In this study we have investigated the level of expression of Aβ protein and BACE-1 in all groups of experimental animals using the western blotting technique. The immuno-blot results clearly demonstrate that scopolamine administration significantly triggered β-secretase activity to cut the amyloid precursor protein at the sites of formation of toxic Aβ fragments, leading to Aβ production via the amyloidogenic pathway. The Aβ and BACE-1 levels are shown in the figure with respective treatment (Figure 3a). Friedelin formation of toxic Aβ fragments, leading to Aβ production via the amyloidogenic pathway. The Aβ and BACE-1 levels are shown in the figure with respective treatment ( Figure 3a). Friedelin being a neuroprotective agent significantly inhibited the BACE-1, accompanied by less production of Aβ in the brains of experimental mice, as presented in Figure 3a-c. Friedelin improved synapse (pre and post) reverse memory dysfunction against scopolamine in mice: Scopolamine administration can significantly affect both pre-and post-neuronal synapse [34]. Neuronal pre-and post-synaptic proteins including synaptophysin (SYP) and PSD95 are of prime importance for neuronal communication. Their expression was evaluated through western blotting. The western blot results showed that scopolamine inhibited pre-and post-synapse protein expression in brains of male adult mice ( Figure 4a), while friedelin administration along with scopolamine significantly protected the neuronal plasticity at synaptic junction. The protein expression level of both pre-and postsynapse was observed to be upsurged with the administration of friedelin. Scopolamine is a chemical agent that affects the memory and behavior of the experimental animals [35]. In the current study the mice were injected with scopolamine and scopolamine + friedelin along with control and friedelin alone. The treated animals were subjected to two well-known behavioral tasks, the Morris water maze (MWM) and the Ymaze. In the MWM test the mice trained for two days (twice a day), then after giving them one day of rest, the data was collected for five (5) consecutive days. Their mean escape latencies were measured on a daily basis to reach the platform in one minute. On the first day, the mean escape latency of control animals was low, while scopolamine-treated mice have shown significantly higher mean escape latency (Figure 4d). In contrast, the group which received scopolamine along with friedelin showed significantly lesser mean escape latencies to reach the target platform. This trend was continued in all experimental groups, including the control animals and animals treated with only friedelin, which showed progressively less mean escape latencies. Although the scopolamine treated animals exhibited little improvement from day one to day five, overall their mean escape latencies were higher than that of the control animals. Moreover, in the probe test, after the removal of the hidden platform, the control and friedelin treated animals spent more time in the target quadrant, as given in the Figure 4e. In contrast, the scopolamine treated animals spent less time in the target quadrant. Interestingly, the mice which received Friedelin improved synapse (pre and post) reverse memory dysfunction against scopolamine in mice: Scopolamine administration can significantly affect both pre-and post-neuronal synapse [34]. Neuronal pre-and post-synaptic proteins including synaptophysin (SYP) and PSD95 are of prime importance for neuronal communication. Their expression was evaluated through western blotting. The western blot results showed that scopolamine inhibited pre-and post-synapse protein expression in brains of male adult mice (Figure 4a), while friedelin administration along with scopolamine significantly protected the neuronal plasticity at synaptic junction. The protein expression level of both pre-and post-synapse was observed to be upsurged with the administration of friedelin. Scopolamine is a chemical agent that affects the memory and behavior of the experimental animals [35]. In the current study the mice were injected with scopolamine and scopolamine + friedelin along with control and friedelin alone. The treated animals were subjected to two well-known behavioral tasks, the Morris water maze (MWM) and the Y-maze. In the MWM test the mice trained for two days (twice a day), then after giving them one day of rest, the data was collected for five (5) consecutive days. Their mean escape latencies were measured on a daily basis to reach the platform in one minute. On the first day, the mean escape latency of control animals was low, while scopolamine-treated mice have shown significantly higher mean escape latency (Figure 4d). In contrast, the group which received scopolamine along with friedelin showed significantly lesser mean escape latencies to reach the target platform. This trend was continued in all experimental groups, including the control animals and animals treated with only friedelin, which showed progressively less mean escape latencies. Although the scopolamine treated animals exhibited little improvement from day one to day five, overall their mean escape latencies were higher than that of the control animals. Moreover, in the probe test, after the removal of the hidden platform, the control and friedelin treated animals spent more time in the target quadrant, as given in the Figure 4e. In contrast, the scopolamine treated animals spent less time in the target quadrant. Interestingly, the mice which received friedelin in combination with scopolamine spent more time in the target quadrant, as presented in Figure 4f. group and the scopolamine treated groups showed the highest and lowest levels of locomotion, respectively. Friedelin treatment alone and in combination with scopolamine significantly enhanced the locomotion of mice in the respective groups. The percentage of spontaneous alternation of control animals was higher in comparison to scopolamine treated group of mice, while the group administered with friedelin + scopolamine displayed a higher percentage of spontaneous alternation, but overall it was less than the control animals, as shown in Figure 4f. Friedelin specifically inhibited phosphorylated JNK activation to abrogate scopolamine-induced neuroinflammation and neurodegeneration in mice: In the current study scopolamine administration caused up-regulation of phosphorylated JNK in adult mice brain. The administration of scopolamine daily for 21 days resulted in oxidative stress densities (b,c). The results were expressed in arbitrary unit (A.U) and were determined using Image J software and histogram indicates mean in A.U ± SEM. While results of behavioral tests are given as (d) mean escape latency in MWM (e) probe test, (f) %age spontaneous alteration in Y-Maze test and (g) Y-Maze number of arm entries. Significance of control vs. scopolamine is expressed as #, while denotes scopolamine vs. scopolamine + friedelin. Significance: # p < 0.05, , ## p < 0.01 and , ### p < 0.001. In the Y-maze behavioral task, the locomotor activity in terms of total arm entries made by mice in each group and the percentage of spontaneous alternations were observed to serve as an indicator of performance of spatial memory. The mice in the control group and the scopolamine treated groups showed the highest and lowest levels of locomotion, respectively. Friedelin treatment alone and in combination with scopolamine significantly enhanced the locomotion of mice in the respective groups. The percentage of spontaneous alternation of control animals was higher in comparison to scopolamine treated group of mice, while the group administered with friedelin + scopolamine displayed a higher percentage of spontaneous alternation, but overall it was less than the control animals, as shown in Figure 4f. Friedelin specifically inhibited phosphorylated JNK activation to abrogate scopolamineinduced neuroinflammation and neurodegeneration in mice: In the current study scopolamine administration caused up-regulation of phosphorylated JNK in adult mice brain. The administration of scopolamine daily for 21 days resulted in oxidative stress induced neuroinflammation and neurodegeneration, attributed to phosphorylation of JNK, resulting in its activation. Administration of friedelin specifically inhibited phosphorylated JNK according to the western blot results. This inhibition is accompanied by reduction in oxidative stress, neuroinflammation, neurodegeneration, neuronal synapse improvement along with memory and cognition reversal. So all these beneficial effects of friedelin can be attributed to its capability to inhibit the phosphorylated JNK. Scopolamine especially induced neuro-inflammatory markers such as NF-κB, by inducing its activation and translocation into the nucleus accompanied by the activation of COX-II (Figure 5a-d). Moreover this neuroinflammation was followed by neurodegeneration as scopolamine significantly triggered neuro-apoptotic markers such as caspase-3 (apoptotic executioner) leading to the damage of nucleus i.e., an increased expression of PARP-1 proteins to cause neuronal cell death as shown in Figure 5. Interestingly the administration of friedelin to the adult albino mice not only significantly abolished NF-κB and COX-II but also attenuated the neurodegenerative hallmarks comprising of caspase-3 and PARP-1 (Figure 5e,f). induced neuroinflammation and neurodegeneration, attributed to phosphorylation of JNK, resulting in its activation. Administration of friedelin specifically inhibited phosphorylated JNK according to the western blot results. This inhibition is accompanied by reduction in oxidative stress, neuroinflammation, neurodegeneration, neuronal synapse improvement along with memory and cognition reversal. So all these beneficial effects of friedelin can be attributed to its capability to inhibit the phosphorylated JNK. Scopolamine especially induced neuro-inflammatory markers such as NF-κB, by inducing its activation and translocation into the nucleus accompanied by the activation of COX-II (Figure 5a-d). Moreover this neuroinflammation was followed by neurodegeneration as scopolamine significantly triggered neuro-apoptotic markers such as caspase-3 (apoptotic executioner) leading to the damage of nucleus i.e., an increased expression of PARP-1 proteins to cause neuronal cell death as shown in Figure 5. Interestingly the administration of friedelin to the adult albino mice not only significantly abolished NF-κB and COX-II but also attenuated the neurodegenerative hallmarks comprising of caspase-3 and PARP-1 (Figure 5g-i). To validate the western blot results, the molecular docking studies were directed to elucidate the inhibitory potential of friedelin on p-JNK. Among the several conformations, To validate the western blot results, the molecular docking studies were directed to elucidate the inhibitory potential of friedelin on p-JNK. Among the several conformations, the most promising docking pose was detected inside the p-JNK binding pocket with appropriate orientation. The p-JNK binding site is comprised of both the hydrophilic and hydrophobic amino acids. The hydrophobic amino acids include Ile70, Val78, Ala91, Ile124, Met146, and 149, Ala151, Val196, and Leu206, while the hydrophilic amino acids include Gly71, Ser72, Gly73, Gln75, Gly76, Lys93, Glu147, Asp150, Cys154, Gln155, 158, Ser193, Asn194, and 255. The analysis of binding modes of the most favorite docked conformation exposed that friedelin interacted well over the binding cavity with hydrophobic amino acids (Ile70, Val78, Val196, and Leu206) through multiple alkyl interactions with minimum distance of 3.60 Å (shown by purple dashed lines). Figure 6 depicts the 3D interactions of the most favorable conformation. Thus, friedelin showed efficient binding in terms of good binding affinity (−7.9 Kcal/mol). Such a lower value indicated the good fitness of the compound in the binding pocket of JNK protein and a stable friedelin-protein interaction. These theoretical results confirmed that the friedelin showed much better activity by making numerous interactions with JNK protein key residues. the most promising docking pose was detected inside the p-JNK binding pocket with appropriate orientation. The p-JNK binding site is comprised of both the hydrophilic and hydrophobic amino acids. The hydrophobic amino acids include Ile70, Val78, Ala91, Ile124, Met146, and 149, Ala151, Val196, and Leu206, while the hydrophilic amino acids include Gly71, Ser72, Gly73, Gln75, Gly76, Lys93, Glu147, Asp150, Cys154, Gln155, 158, Ser193, Asn194, and 255. The analysis of binding modes of the most favorite docked conformation exposed that friedelin interacted well over the binding cavity with hydrophobic amino acids (Ile70, Val78, Val196, and Leu206) through multiple alkyl interactions with minimum distance of 3.60 Å (shown by purple dashed lines). Figure 6 depicts the 3D interactions of the most favorable conformation. Thus, friedelin showed efficient binding in terms of good binding affinity (−7.9 Kcal/mol). Such a lower value indicated the good fitness of the compound in the binding pocket of JNK protein and a stable friedelin-protein interaction. These theoretical results confirmed that the friedelin showed much better activity by making numerous interactions with JNK protein key residues. Friedelin reduced blood cholesterol and triglycerides abundance in adult mice: Scopolamine administration caused an increase in the cholesterol levels of adult mice. It is in accordance with a previous study in which scopolamine led to increased total cholesterol and triglycerides in experimental animals [36]. The results indicate that scopolamine significantly enhanced total cholesterol, TGL, LDL and VLDL, accompanied by the reduction in HDL in the serum of only scopolamine treated animals as shown in Figure 7. In contrast, the administration of friedelin significantly improved the blood parameters by lowering the total cholesterol, TGL, HDL, LDL and enhancing VLDL in the serum of the experimental animals, as shown in Figure 7. Friedelin reduced blood cholesterol and triglycerides abundance in adult mice: Scopolamine administration caused an increase in the cholesterol levels of adult mice. It is in accordance with a previous study in which scopolamine led to increased total cholesterol and triglycerides in experimental animals [36]. The results indicate that scopolamine significantly enhanced total cholesterol, TGL, LDL and VLDL, accompanied by the reduction in HDL in the serum of only scopolamine treated animals as shown in Figure 7. In contrast, the administration of friedelin significantly improved the blood parameters by lowering the total cholesterol, TGL, HDL, LDL and enhancing VLDL in the serum of the experimental animals, as shown in Figure 7. Discussion In the present findings, friedelin has shown neuroprotective effects against scopolamine-induced multiple neuropathological conditions of Alzheimer's disease. Friedelin, being a triterpenoid isolated from Quercus dilatata, displayed unique and potent characteristics in treating the neuronal communication gap attributed to the administration of scopolamine to adult mice. Additionally, it minimized not only the oxidative stress but also the neuroinflammation and neurodegeneration mediated memory dysfunction via phosphorylated JNK inhibition. Interestingly, it also significantly reduced the cholesterol and improved the lipid profile of the mice challenged with scopolamine. During the study we observed that friedelin had shown anti-oxidant capabilities, as it caused the up regulation of the levels of endogenous antioxidant enzymes i.e., SOD, POD, CAT, GSH; enzymes which are responsible for the decrease of oxidative stress. It also diminished the overexpressed TBARS. In the case of reduced expressions of antioxidant enzymes, oxidative stress may lead to neuronal damage and neuronal apoptosis [30]. A similar work was also reported on in which a nsuccinamide derivative significantly ameliorated scopolamine-induced antioxidant enzymes inhibition, which led to the recovery of the memory and relevant complications [33]. Yet another study reported the correlation of glial cells activation with the disease pathology and suggested that the molecules targeting the inhibition of glial cells and astrocytes can be a potential candidate for treatment of Alzheimer's disease [37]. Strong evidence suggests that increased cholesterol level is a potential risk factor in the progression of Alzheimer's disease complications. Increased cholesterol levels in neuronal membranes can facilitate Aβ production and its aggregation [38]. A recent study reported that LPS, a pro-inflammatory factor, significantly increased intracellular cholesterol accumulation by up-regulating the expression of NF-κB signalling by knocking-down overexpressed IKKα and TAB3. Interestingly, the increased levels The results are expressed as mean ± SEM employing one-way ANOVA followed by Tukey's multiple comparison post-test. Significance of control vs. scopolamine is expressed as #, while denotes scopolamine vs. scopolamine + friedelin/glutinol. Significance: ns = non-significant , # p < 0.05, *, ## p < 0.01 (n = 5). Discussion In the present findings, friedelin has shown neuroprotective effects against scopolamineinduced multiple neuropathological conditions of Alzheimer's disease. Friedelin, being a triterpenoid isolated from Quercus dilatata, displayed unique and potent characteristics in treating the neuronal communication gap attributed to the administration of scopolamine to adult mice. Additionally, it minimized not only the oxidative stress but also the neuroinflammation and neurodegeneration mediated memory dysfunction via phosphorylated JNK inhibition. Interestingly, it also significantly reduced the cholesterol and improved the lipid profile of the mice challenged with scopolamine. During the study we observed that friedelin had shown anti-oxidant capabilities, as it caused the up regulation of the levels of endogenous antioxidant enzymes i.e., SOD, POD, CAT, GSH; enzymes which are responsible for the decrease of oxidative stress. It also diminished the overexpressed TBARS. In the case of reduced expressions of antioxidant enzymes, oxidative stress may lead to neuronal damage and neuronal apoptosis [30]. A similar work was also reported on in which a nsuccinamide derivative significantly ameliorated scopolamine-induced antioxidant enzymes inhibition, which led to the recovery of the memory and relevant complications [33]. Yet another study reported the correlation of glial cells activation with the disease pathology and suggested that the molecules targeting the inhibition of glial cells and astrocytes can be a potential candidate for treatment of Alzheimer's disease [37]. Strong evidence suggests that increased cholesterol level is a potential risk factor in the progression of Alzheimer's disease complications. Increased cholesterol levels in neuronal membranes can facilitate Aβ production and its aggregation [38]. A recent study reported that LPS, a pro-inflammatory factor, significantly increased intracellular cholesterol accumulation by up-regulating the expression of NF-κB signalling by knocking-down overexpressed IKKα and TAB3. Interestingly, the increased levels of cholesterol contrariwise exerted an amplified pro-inflammatory effects by activating the NF-κB signalling pathway [39], as also confirmed by our results. These results may be suggestive of the correlation of serum cholesterol in correspondence with the overexpression of NF-κB and its downwards signalling pathways. The up-regulation of neuroinflammatory and neurodegenerative biomarkers is another hallmark of oxidative damage. Friedelin has shown anti-inflammatory and antineuroapoptotic effects by lowering these markers such as NF-κB. It also inhibited the neuro-inflammatory process by inhibiting p-JNK, which is activated in case of increased oxidative stress conditions [40,41]. The inhibition of JNK phosphorylation was also confirmed through molecular docking and western blotting analyses. The p-JNK is further responsible for the decreased level of pre-and post-synapse proteins; hence memory dysfunction induced in scopolamine-treated mice. Our results of MWM and Y-Maze clearly indicates that friedelin has the ability to improve both long and short term memory deficits in mice. The performance of the friedelin treated mice in MWM was far better than mice treated with scopolamine alone because they have shown reduced mean escape latencies on a regular interval throughout the testing period and displayed enhanced spatial memory, accompanied by more time spent in the target quadrant in the probe test. Although their performance was better than the scopolamine-treated mice, they still were less efficient than the control group mice, which suggests that these mice have partially recovered from the scopolamine-induced dysfunction in long-term memory. Furthermore, in a Y-Maze experiment, friedelin administered mice displayed a greater percentage of spontaneous alterations as compared to the mice that only received scopolamine. The production of Aβ from amyloid precursor protein (APP) protein through the amyloidogenic pathway is another well-established characteristic of Alzheimer's disease [42]. β-secretase (BACE-1) is a major secretase enzyme which causes the fragmentation of APP protein [43]. Interestingly, the current study demonstrated that triterpenoid friedelin efficiently inhibited β-secretase activity and abolished the production of Aβ. The reduction in Aβ production can be attributed to reduced serum cholesterol and triglyceride levels and enhanced antioxidant enzyme activity in friedelin-treated mice. Conclusions After summarizing the results, it can be suggested that scopolamine is neurotoxic, as it compels the normal brain to induce AD neuropathology in adult mice. Friedelin may be considered as a bioactive therapeutic agent that reverses all of the neuropathological symptoms induced by scopolamine. These results are of clinical importance because for the first time these findings are suggestive of a new investigational candidate as a potential lead for the treatment of AD by targeting multiple markers including those of amyloidogenic pathology. Informed Consent Statement: Not applicable. Data Availability Statement: Data will be available on request.	2022-07-17T15:10:25.997Z	2022-07-01T00:00:00.000	{ "year": 2022, "sha1": "fc8b28ba45e222f3c466a90ffdca2e381bee9465", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1420-3049/27/14/4513/pdf?version=1657882957", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "d4c99a9562c3ff49a30d77564619e4b49880fc26", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
33203552	pes2o/s2orc	v3-fos-license	Physical Proximity of Sister Chromatids Promotes Top2-Dependent Intertwining Summary Sister chromatid intertwines (SCIs), or catenanes, are topological links between replicated chromatids that interfere with chromosome segregation. The formation of SCIs is thought to be a consequence of fork swiveling during DNA replication, and their removal is thought to occur because of the intrinsic feature of type II topoisomerases (Top2) to simplify DNA topology. Here, we report that SCIs are also formed independently of DNA replication during G2/M by Top2-dependent concatenation of cohesed chromatids due to their physical proximity. We demonstrate that, in contrast to G2/M, Top2 removes SCIs from cohesed chromatids at the anaphase onset. Importantly, SCI removal in anaphase requires condensin and coincides with the hyperactivation of condensin DNA supercoiling activity. This is consistent with the longstanding proposal that condensin provides a bias in Top2 function toward decatenation. A comprehensive model for the formation and resolution of toxic SCI entanglements on eukaryotic genomes is proposed. In Brief Sister chromatid intertwines (SCIs) are produced during DNA replication and removed by Top2 before segregation. Sen et al. show that physical proximity between sister chromatids drives Top2 to introduce SCIs in already replicated chromatids. However, in anaphase, condensin supercoiling activity provides a bias in Top2 function to remove all SCIs. INTRODUCTION Unwinding of parental DNA strands by replicative helicases generates superhelical tension (in the form of positive supercoils) ahead of the replication fork. This tension can be removed either by topoisomerase-mediated relaxation of supercoils in the unreplicated parental region or by fork rotation, often referred to as fork ''swiveling'', relative to the unreplicated DNA (Champoux et al., 1984). A direct consequence of fork swiveling is the formation of crosses between replicated sister chromatids behind the forks; these become sister chromatid intertwines (SCIs), or catenanes, upon replication completion. During late stages of replication, as forks converge, swiveling is thought to be the preferred pathway to eliminate superhelical tension (Postow et al., 2001). Therefore, the formation of sister chromatid intertwines (SCIs) in replicated chromosomes is thought to be a direct consequence of DNA replication (Postow et al., 2001). Heavily intertwined sister chromatids exhibit difficulties during chromosome segregation. Cells use type II topoisomerases (like yeast Top2) to remove the bulk of SCIs between replicated chromosomes before anaphase. These enzymes work by cleaving both strands of one DNA duplex and passing a second DNA through the break before religation (Wang, 1996). When this strand-passage reaction involves cleavage of one chromatid and passage of a sister chromatid segment, the end product can be chromatid disentanglement (i.e., decatenation) or chromatid intertwining (i.e., catenation), depending on the direction of the passage. In theory, Top2 could be ''blind'' to the direction, and thus could introduce linkages between the sister chromatids as well as remove them. This might be a problem in the congested nuclear environment, where replicated chromosomes are woven together and sister chromatids are adjoined by cohesin complexes. Current thinking suggests that this is not an issue because Top2 action has an intrinsic bias toward decatenation. This view originates from the fact that, in vitro, type II topoisomerases decatenate plasmids below the expected thermodynamic equilibrium (Rybenkov et al., 1997). Moreover, modeling studies of Top2 action have suggested an ability to detect the chiral orientation of particular DNA crosses, leading to the proposition that Top2 is able to identify suitable substrates in a global context (Stone et al., 2003;Timsit and Vá rnai, 2010). Based on this, Top2 has been speculated to act as ''Maxwell's demon'' (Maxwell, 1871), acting only on a subset of possible substrates (DNA crossovers resulting in decatenation) which are required to obtain a desired global topological status, i.e., full decatenation of the genome (Pulleyblank, 1997). Ergo, the accepted (although not experimentally demonstrated) view is that, inside cells, Top2 has a bias in its strand-passage activity toward decatenation. Therefore, based on our present understanding, SCIs formed during DNA replication should be removed by Top2 during the period spanning S phase and mitosis (Stuchinskaya et al., 2009). SCI removal, however, might be more complex than this prediction, because recent data suggest that SCIs persist until mitosis in yeast minichromosomes when cells are arrested with the microtubule poison nocodazole (Farcas et al., 2011;Koshland and Hartwell, 1987). Maintenance of SCIs until metaphase requires cohesion (Farcas et al., 2011); this has led to the proposal that cohesin complexes ''protect'' SCIs by masking them from resolution by Top2 (Farcas et al., 2011). Condensin is not only required for metaphase chromosome condensation (Hirano, 2005), but is also necessary to prevent sister chromatids from being entangled during segregation; these two functions are likely to be linked (Koshland and Strunnikov, 1996). Therefore, it has been proposed that chromosome The wild-type strain bearing a 10 kb circular minichromosome was synchronized in G 1 and released into the cell cycle in the presence of nocodazole at 30 C. A second dose of nocodazole was added after 180 min. Samples were taken for analysis at 120 min and 240 min after release. Cell extracts were separated by differential sedimentation, heat denatured in 1% SDS, and analyzed by Southern blotting to reveal plasmid species. Plasmid species included monomeric forms (legend continued on next page) condensation, and condensin itself, might somehow facilitate Top2's role in removing SCIs from anaphase chromosomes (Koshland and Strunnikov, 1996;Nasmyth, 2001). Pioneering in vitro studies demonstrated that condensin overwinds DNA, generating positive supercoiling, in the presence of topoisomerases (Kimura and Hirano, 1997;Bazett-Jones et al., 2002). Recently, overwinding (leading to positive supercoiling) on yeast minichromosomes passing through mitosis was reported and was indeed shown to be dependent on condensin (Baxter et al., 2011). Importantly, Top2 showed a bias toward decatenation on catenated plasmids that were positively supercoiled but not negatively supercoiled (Baxter et al., 2011). These findings experimentally support the possibility that chromosome condensation facilitates Top2's role in removing SCIs during mitosis (Koshland and Strunnikov, 1996). The present study was designed to investigate why SCIs are maintained on chromosomes until metaphase and to test whether Top2 has a bias toward decatenation in its strand-passage activity during the G 2 /M period of the yeast cell cycle. Our findings reveal that SCIs are formed during G 2 /M by Top2dependent catenation, and also show that the physical proximity of sister chromatids facilitates SCI formation. Finally, we present data supporting the proposed role of condensin in promoting directionality in Top2's strand-passage activity to favor decatenation at the anaphase onset. SCIs Are Maintained in Prolonged Mitotic Arrests SCIs are thought to be formed during DNA replication (Postow et al., 2001) and removed progressively by Top2 during G 2 and early mitosis (Stuchinskaya et al., 2009). Analysis of yeast centromeric plasmids, however, has revealed that in cells arrested in metaphase in the absence of spindles (i.e., arrests mediated by the microtubule poison nocodazole), SCIs persist (Farcas et al., 2011;Koshland and Hartwell, 1987). First, we used differential sedimentation velocity and gel electrophoresis of large yeast centromeric plasmids (Farcas et al., 2011) to test whether catenated dimers are preserved in a prolonged mitotic arrest in the absence of spindle microtubules (nocodazole arrests). Cell extracts were sedimented in 10%-45% sucrose gradients, and the plasmid-containing fractions were denatured by heating to 65 C in 1% SDS before being electrophoresed through 0.5% agarose gels. Southern blotting with plasmid-specific probes revealed both catenated sister DNA dimers and monomeric DNAs ( Figure 1A). The electrophoretic mobility of monomers and dimers was similar to that observed in previous reports using this technique (Farcas et al., 2011). The identities of the electrophoretic forms were confirmed using digestion analysis ( Figure S1). To evaluate changes in SCI levels over time, cells were blocked in metaphase with nocodazole and either immediately processed for analysis or kept arrested for extended times. Sucrose gradient fractions were analyzed, and the quantity of dimers (Catc/bs) was calculated relative to either the total amount of monomers (supercoiled/relaxed) ( Figure 1A) or total DNA ( Figure S2). The results revealed an increase in SCI quantities in samples that had been blocked with nocodazole for extended times ( Figure 1A). Therefore, the level of catenated dimers did not diminish over time, as was expected from the assumption that Top2 progressively removes SCIs. Physical Proximity between Replicated Chromatids Is Required for the Maintenance of SCIs Previous studies have demonstrated that inactivation of cohesin during interphase prevents the presence of intertwining on yeast plasmids in nocodazole-arrested cells during metaphase (Farcas et al., 2011). This has led to the proposal that cohesin protects SCIs from Top2-mediated decatenation (Farcas et al., 2011). First, we measured the kinetics of SCIs in nocodazoleblocked cells upon cohesin inactivation. Following the temperature shift in cells which were already blocked in nocodazole, to inactivate scc1-73, we observed the disappearance of SCIs over a 4-hr period ( Figure 1B). Therefore, as shown previously for interphase (Farcas et al., 2011), inactivation of cohesin in metaphase-arrested cells also causes SCI loss. Next, we sought to investigate whether the maintenance of SCIs requires either physical proximity between sister chromatids or a function of the cohesin complex other than keeping chromatids together (i.e., direct protection of catenanes from Top2mediated decatenation). To address this, we decided to maintain physical proximity between sister plasmids in the absence of functional cohesin. Our minichromosomes carried a short stretch of lactose operators, and we expressed two forms of lactose repressors: one that binds a single operator (lacI-DC), and another capable of binding two operators simultaneously (lacI-t) (Straight et al., 1996). This approach has been previously used to hold sister chromatids in the absence of cohesin (Straight et al., 1996) ( Figure S3A). We compared SCIs on lacO-bearing plasmids in cells expressing either lacI-t or lacI-DC. Both lacI forms were expressed using the galactose-inducible promoter. The level of SCIs was greater in cells expressing lacI-t (Figure 2A), suggesting that increased closeness between chromatids might promote SCIs. Next, we inactivated cohesin using the scc1-73 allele in cells expressing lacI-t ( Figure 2B). scc1-73 cells carrying the lacI-t system were arrested in metaphase, and lacI-t was expressed as scc1-73 was inactivated ( Figure 2B). Note that inactivation of scc1-73 was achieved by raising the temperature to 34 C instead of 37 C because we found that (OCm, relaxed monomer; Lm, linear monomer; and CCCm, supercoiled monomer) and dimeric forms, including different sister chromatid intertwine (SCI) species (Catc-type catenanes and Catb-type catenanes). SCIs are indicated in the blots (dashed boxes). The mitotic arrest was stable during the experiment. SCI levels were maintained during the extended mitotic block. SCIs (including c-and b-type catenanes) were quantified using ImageJ software (bottom right-hand graph; showing the mean ± SD from three independent experiments). (B) The scc1-73 strain bearing the 10 kb circular minichromosome was synchronized in G 1 and released into the cell cycle in the presence of nocodazole at 25 C; upon metaphase arrest, the culture was shifted to 37 C to inactivate scc1-73. A second dose of nocodazole was added after 180 min. Samples were collected 0, 120, and 240 min after the temperature shift. Samples were analyzed as in (A). scc1-73 inactivation in metaphase-arrested cells led to the progressive disappearance of SCI species. SCIs were quantified as in (A) (bottom right-hand graph; showing the mean ± SD from three independent experiments). expression of lacI-t from the galactose promoter was not efficient at 37 C (data not shown); however, cohesion loss in scc1-73 cells was confirmed at 34 C ( Figure S3B). As before, inactivation of scc1-73 led to loss of SCIs ( Figure 2B, GLU); however, lacI-t expression reduced SCI loss significantly ( Figure 2B, GAL). From these results, we conclude that physical proximity between chromatids is a key parameter required for the maintenance of SCIs in cells arrested in metaphase by nocodazole treatment. Therefore, the loss of SCIs upon cohesin inactivation is a consequence of the separation between sister DNAs, rather than a loss of inhibition by cohesin on Top2 function. Increased Levels of Top2 Promote Formation of SCIs in Metaphase Our results show that the physical proximity of chromatids is important for the maintenance of SCI levels in nocodazole arrests ( Figure 2B). This suggests that, rather than blocking Top2-dependent decatenation, cohesin rings might facilitate the reverse reaction (i.e., catalysis of sister DNA catenation) by bringing about closeness of sister DNAs. Early in vitro studies using X. laevis egg extracts showed that increased levels of Top2 led to higher catenation of DNA substrates (Baldi et al., 1980). Therefore, we reasoned that increasing the amount of cellular Top2 might potentially alter SCI levels in chromosomes. It is noteworthy that overexpression of Top2 from the GAL1-10 promoter is detrimental to cell growth ( Figure 3A) (Vosberg, 1985;Worland and Wang, 1989). We arrested wild-type cells carrying an additional copy of Top2 under the GAL1-10 promoter in metaphase using nocodazole, and we induced Top2 expression with galactose. Analysis of sucrose gradient fractions showed that SCI levels increased after Top2 expression ( Figure 3B). This increase was not observed in cells carrying a control plasmid (Figure 3B). Therefore, we conclude that the quantity of Top2 protein inside cells affects the levels of SCIs in the minichromosomes. We noticed that Top2 overexpression caused shifting of plasmid distributions on the gradients ( Figure 3B), including monomer and dimer forms. It is possible that, in addition to altering SCI levels, additional Top2 leads to changes in the supercoiling status of plasmids, thus explaining the altered distribution of plasmids on the gradients. To investigate this, we compared the supercoiling of monomer plasmids in the presence or absence of Top2 overexpression ( Figure 3C). Cells were released from G 1 into metaphase in the presence or absence of Top2 overexpression. DNA was purified at both the G 1 and metaphase arrests and then analyzed in 2D chloroquine gels to reveal the supercoiling distribution of the plasmids. No change in the distribution of topoisomers was observed when cells did not overexpress Top2 ( Figure 3C, Control). In contrast, a small deviation toward a less negative topoisomer distribution was observed in cells expressing additional Top2 ( Figure 3C). Similar results were observed when Top2 expression was induced in cells arrested in metaphase ( Figure 3D). We conclude that changes in supercoiling contribute to the shifting of plasmid distributions on the gradients upon Top2 expression. Increasing Levels of Cellular Top2 Negatively Affect Sister Chromatid Segregation The changes observed in supercoiling and the increase in SCIs upon Top2 overexpression (Figures 3B-3D) prompted us to investigate effects on chromosome segregation. To test this, we expressed Top2 from the GAL1-10 promoter in G 1 -arrested cells before releasing them into the cell cycle, and we visualized the timing of sister chromatid segregation. We employed chromosome tags in the vicinity of the centromere and telomere of chromosome IV. Overexpression of Top2 caused delays in separation of both centromeric and telomeric regions ( Figure 4A). Stretched anaphase nuclei accumulated and persisted in cells overexpressing Top2 for longer times than in control cells (Figure 4A), demonstrating that Top2 expression adversely affected sister chromatid disjunction. However, cells eventually segregate nuclear masses normally. The chromosome segregation delay was not caused by delayed cell-cycle progression, as the kinetics of Pds1 (Securin) degradation was similar in Top2 overexpression and control samples ( Figure 4B). Furthermore, we did not observe activation of the DNA damage checkpoint kinase Rad53 in cells expressing Top2 ( Figure 4B), which shows that a DNA damage checkpoint response is not triggered when Top2 levels are increased. Nevertheless, a transient increase in Rad52 foci was observed during S phase in cells overexpressing Top2 ( Figure 4C), suggesting higher levels of ongoing DNA repair in these cells. These results raise the possibility that increased levels of Top2, despite not triggering a checkpoint response, might have detrimental effects on long-term genome stability. Condensin Is Required for SCI Resolution SCIs are maintained on cohesed metaphase minichromosomes because of their physical proximity ( Figure 2B). Condensin complexes promote mitotic chromosome condensation and Figure 2. Physical Proximity between Sister Chromatids Maintains SCIs Post-replication (A) Strains bearing a 10 kb circular minichromosome containing short lacO arrays and expressing a wild-type lacI (lacI-t; able to bind two operators simultaneously) or a lacI with a C-terminal deletion (lacI-DC; able to bind a single operator) under the galactose promoter were synchronized in G 1 and released into the cell cycle in the presence of galactose and nocodazole at 30 C. Cell extracts were separated by differential sedimentation, heat denatured in 1% SDS, and analyzed by Southern blotting to reveal plasmid species. Plasmid species included monomeric forms (OCm, Lm, and CCCm) and dimeric forms, including different SCI species (Catc-type catenanes and Catb-type catenanes). SCIs are indicated in the blots (dashed boxes). SCIs (including c-and b-type catenanes) were quantified relative to monomeric species using ImageJ software (right-hand graph; showing the mean ± SD from three independent experiments). Cells expressing lacI-t exhibited a small increase in SCI levels. (B) The scc1-73 strain bearing the 10 kb circular minichromosome containing the lacO array and the lacI-t construct under the galactose promoter was grown in raffinose media and synchronized in G 1 . The culture was divided in two. One half was released in galactose media (expressing lacI-t) while the other half was released in glucose media (repressing lacI-t). Both cultures were released from the G 1 arrest in the presence of nocodazole at 25 C, and upon metaphase arrest, the cultures were shifted to 34 C to inactivate scc1-73. Samples were analyzed at the metaphase block at 25 C and 3 hr after temperature shift. Analysis was done as in (A). scc1-73 inactivation led to reduced SCI levels in the absence of lacI-t expression, but not in its presence. SCIs were quantified as in (A) (bottom right-hand graph; showing the mean ± SD from three independent experiments). Figure 3. Increased Levels of Top2 Promote SCI Formation (A) A strain carrying a chromosomal insertion of GAL-TOP2 in the ura3 locus and a wild-type strain were grown and plated on media containing glucose (where GAL-TOP2 is not expressed) or galactose (where GAL-TOP2 is expressed) at 30 C. Images were taken after 3 days. Expression of TOP2 from the GAL1-10 promoter impaired cellular growth. (legend continued on next page) are required for sister chromatid resolution during segregation in anaphase (Hirano, 2012). Although the inactivation of condensin during interphase does not affect SCI levels on metaphase minichromosomes arrested with nocodazole (Farcas et al., 2011), it prevents the full removal of SCIs by telophase (Charbin et al., 2014). Surprisingly, in cells arrested in metaphase by the spindle poison nocodazole, SCIs are maintained on minichromosomes, while in cells arrested in metaphase through depletion of the anaphase-promoting complex (APC) activator Cdc20, SCIs are resolved (Charbin et al., 2014;Farcas et al., 2011). The difference between these two metaphase arrests lies in the presence of spindles in the Cdc20-mediated block. Incidentally, condensin binds to pericentromeric regions of replicated chromatids when nocodazole is present in Cdc20-mediated blocks, and spreads to chromosome arms as bipolar spindles are allowed to form (upon nocodazole removal in Cdc20-depleted conditions) (Leonard et al., 2015). Importantly, condensin relocalization temporally correlates with its function in promoting overwinding (positive supercoiling) of DNA (Baxter et al., 2011;Leonard et al., 2015), which requires Polo-and Aurora-kinasedependent phosphorylation (Leonard et al., 2015;St-Pierre et al., 2009). Since this function has been proposed to provide a bias for Top2 toward decatenation (Baxter et al., 2011), we considered the possibility that the hyperactivation of condensin supercoiling promotes SCI removal on cohesed chromatids despite their physical proximity. To test this, we investigated whether or not removal of SCIs occurs in metaphase arrests by Cdc20 depletion when condensin function is compromised. To this end, we used the conditional allele smc2-8. Analysis of sucrose gradient fractions confirmed that cdc20-td smc2-8 cells have higher levels of SCIs than cdc20-td cells do ( Figure 5A). This result shows a correlation between hyperactivation of condensin supercoiling and SCI resolution in metaphase blocks where physical proximity of chromatids due to cohesion is still present. Reversing Condensin-Dependent Overwinding Allows Top2-Dependent SCI Formation in Metaphase Cells Analysis of the genome-wide distribution of yeast condensin has shown that the complex binds to centromere regions of replicated chromatids in G 2 /M arrests with nocodazole (Leonard et al., 2015), while in Cdc20-mediated blocks condensin spreads to chromosome arms (Leonard et al., 2015). However, the addition of nocodazole to Cdc20-mediated blocks restores the centromeric localization of condensin (Leonard et al., 2015). Since condensin-dependent overwinding requires chromosomes to be attached to mitotic spindles (Baxter et al., 2011), we decided to test whether condensin overwinding is inhibited when nocodazole is added to Cdc20-mediated blocks. Overwinding can be detected as a change in the electrophoretic mobility of yeast minichromosomes in the absence of Top2 activity (Baxter et al., 2011;Leonard et al., 2015). The mobility shift is caused by a transition from negatively supercoiled (CatC) to positively supercoiled (CatC) catenanes (Baxter et al., 2011;Leonard et al., 2015). We released cells from G 1 in the absence of Top2 and Cdc20 ( Figure 5B, top2-td and cdc20-td) and analyzed the formation of CatCs (overwound, or positively supercoiled, dimers). We first detected CatCs (underwound, or negatively supercoiled, dimers) 40-60 min after release (Figure 5B). These CatCs transitioned to CatCs from 80 min onward as the cells reached metaphase arrests by Cdc20 depletion (Figure 5B). We performed two experiments in parallel, and added nocodazole to one of them 120 min after the G 1 release (a time when the CatC-to-CatC transition had taken place). CatCs reappeared when nocodazole was added back to Cdc20-blocked cells ( Figure 5B). Therefore, overwinding can be inhibited in Cdc20-blocked cells by the addition of nocodazole. Next, we reasoned that, if overwinding provides a bias toward decatenation for Top2 action, as predicted from our results (Figure 5A), then upon nocodazole inhibition of overwinding in Cdc20 blocks ( Figure 5B), Top2 should lose its directionality toward decatenation; as a consequence, new SCIs should be reformed between minichromosomes. First we confirmed that, in Cdc20 blocks containing nocodazole, SCIs are resolved when nocodazole is removed from media ( Figure 5C). Next, we added nocodazole back to Cdc20-blocked cells that had resolved SCIs ( Figure 5C). As predicted, nocodazole addition to Cdc20 blocks led to the reappearance of new SCIs ( Figure 5C), and this was dependent on Top2 ( Figure 5C). Therefore, our data suggest that condensin-dependent overwinding provides a bias for Top2 action toward decatenation when cohesed chromosomes are attached to spindle microtubules, and that inhibiting overwinding reverts Top2 strand-passage activity to work bidirectionally, as it does during the post-replicative period of G 2 , a (B) Wild-type and GAL-TOP2 strains bearing the 10 kb circular minichromosome were synchronized in G 1 in media containing raffinose and released into the cell cycle in the presence of nocodazole at 30 C. Following metaphase arrest, cells were transferred to new media containing galactose and nocodazole to activate expression of TOP2 from the GAL1-10 promoter. Samples were taken for analysis just before and 120 min after transfer to galactose. Cell extracts were separated by differential sedimentation, heat denatured in 1% SDS, and analyzed by Southern blotting to reveal plasmid species. Plasmid species included monomeric (OCm, Lm, and CCCm) and dimeric forms, including different SCI species (Catc-type catenanes and Catb-type catenanes). Running position for SCIs in the blots are indicated (dashed boxes). SCIs (including c-and b-type catenanes) were quantified using ImageJ software (right-hand graphs; showing the mean ± SD from three independent experiments). SCIs increased upon activation of GAL-TOP2. (C) GAL-TOP2 and control strain bearing a circular minichromosome (pRS316) were blocked in G 1 in media containing raffinose and released into the cell cycle in the presence of galactose and nocodazole at 30 C. DNA was purified from cells in the G 1 arrest (G 1 À Raffinose) and at the metaphase block (MÀGalactose) and analyzed by two-dimensional gel chloroquine analysis to assess the supercoiling status of monomer plasmids in the cells. A small distribution change toward a less negatively supercoiled population was observed upon TOP2 expression (bottom panels). A cartoon representation of how the plasmid distribution relates to supercoiling status is shown. (D) GAL-TOP2 bearing a circular minichromosome (pRS316) was blocked in G 1 in media containing raffinose and released to nocodazole in raffinose media at 30 C. Upon metaphase arrest, half of the culture was maintained in raffinose while galactose was added to the other half to induce TOP2 expression. Samples were taken 120 min after the galactose addition. DNA was purified and analyzed as in (C). A small distribution change toward a less negatively supercoiled population was observed upon TOP2 expression. time of the cell cycle when Top2 action not only removes SCIs between sister chromatids, but also promotes them. DISCUSSION Pioneering studies on SV40 replication termination first linked the formation of SCIs to the DNA replication process (Sundin and Varshavsky, 1980). Based on the intimate connection between DNA replication and SCI formation, and the fact that these structures can keep sister chromatids together, SCIs were the first mechanism proposed to explain sister chromatid cohesion (Murray and Szostak, 1985). However, the realization that in Cdc20 mutants yeast minichromosomes are cohesed in the absence of SCIs (Koshland and Hartwell, 1987) ruled out SCIs as the major cohesion force and provided the intellectual framework for the discovery of cohesins as the protein-mediated bridges that hold sister chromatids (Guacci et al., 1997;Losada et al., 1998;Michaelis et al., 1997). This was followed by a detailed description of the cell-cycle regulation behind cohesin removal in anaphase (Nasmyth, 2001), which implied that cells regulate cohesin to hold sisters and resulted in reduced interest in SCIs. In addition, in vitro studies of Top2 demonstrated that this enzyme decatenates plasmids below the expected thermodynamic equilibrium (Rybenkov et al., 1997), an observation that led to the proposal that Top2 has directionality toward decatenation. Therefore, SCIs were predicted to form exclusively during DNA replication and to be rapidly removed by Top2 during G 2 , with only a minor population of SCIs remaining until mitosis. Here, we have uncovered several unexpected facts about SCIs. First, we have shown that SCI formation is not limited to the process of DNA replication or the S phase period of the cell cycle. Second, we have demonstrated that Top2 does not have an intrinsic directionality toward decatenation. Third, we identified physical proximity between chromatids and Top2 cellular amounts as key parameters behind the replication-independent catenation of chromatids. And finally, we have shown that condensin overwinding, which requires mitotic spindles, correlates with the initiation of Top2 decatenation on cohesed chromosomes despite their physical proximity, suggesting that directionality toward decatenation relies on condensin overwinding. Moreover, our results fully explain why SCIs are present in metaphase arrests with nocodazole but absent in metaphase blocks mediated by Cdc20 depletion, in which spindles are present. The view that SCIs are rapidly dissolved by Top2 following DNA replication has been recently called into question by the observation that a significant percentage of SCIs persist in large yeast minichromosomes when cells are arrested in metaphase with the microtubule poison nocodazole (Farcas et al., 2011). In addition, studies on mammalian chromosomes using Top2 inhibitors have demonstrated that a significant number of catenations persist until mitosis, leading to the proposal that these catenations are removed during the process of chromosome individualization (Gimé nez-Abiá n et al., 1995; Losada et al., 2002). The persistence of SCIs after DNA replication in yeast minichromosomes was shown to depend on cohesin (Farcas et al., 2011). We considered two potential mechanisms to explain the requirement of cohesin for SCI maintenance: an ''SCI protection'' feature by cohesin in which the complex masks SCIs from Top2's resolution, or a ''stimulation of concatenation'' scenario brought about by the proximity of sister DNAs at cohesin sites. The fact that SCIs are maintained in minichromosomes held together by tetramerized lacI ( Figure 2B) and the observed de novo formation of SCIs when condensin-dependent overwinding is inactivated in mitosis ( Figure 5C) demonstrates that cohesin's role in maintaining SCIs is indirect, and simply a consequence of the fact that cohesion promotes physical proximity between chromatids. It is noteworthy that previous biochemical analysis of purified cohesin from HeLa cells showed that the complex stimulates intermolecular catenation of circular DNA molecules in the presence of Top2 (Losada and Hirano, 2001). Our findings extend these observations and show that in vivo cohesin sites represent regions where SCI links are formed by Top2 between sister chromatids. It is important to recognize that the formation of SCIs at cohesin sites does not represent an active, or cohesin-independent, mechanism to maintain sister chromatid cohesion, since SCIs depend on cohesin presence during G 2 . Several studies have proposed the existence of multiple mechanisms, including SCIs, to ensure cohesion along chromosomes (Díaz-Martínez et al., 2008). It is evident that at any given time the presence of SCIs provides additional physical links between sister DNA molecules; however, the ongoing formation and resolution of SCIs by Top2 at cohesin sites would result in the rapid disappearance of SCIs without cohesin's action in maintaining physical proximity. Consequently, SCIs would not be sufficient to maintain any form of sustainable cohesion between chromatids for (A) cdc20-td and cdc20-td smc2-8 strains bearing the 10 kb circular minichromosome were synchronized in G 1 and released into the cell cycle under conditions of Cdc20 depletion at 37 C. Samples were taken at the metaphase block. Cell extracts were separated by differential sedimentation, heat denatured in 1% SDS, and analyzed by Southern blotting to reveal plasmid species. Plasmid species included monomeric forms (OCm, Lm, and CCCm) and dimeric forms, including different SCI species (Cata-, Catb-, and Catc-type catenanes). Running position for SCIs in the blots is indicated (dashed boxes). The mitotic arrest was stable (legend continued on next page) extended periods of time. Therefore, in our view, SCIs represent a byproduct of cohesin function and not an active mechanism providing cohesion between chromatids. Previous arguments against a direct role for SCIs in cohesion often invoked the fact that SCI removal was not thought to be cell-cycle regulated, unlike cohesin cleavage by separase. Contrary to this view, our results show that mechanisms exist to promote removal of SCIs from cohesed chromosomes at the anaphase onset. SCIs are present in minichromosomes arrested in metaphase using the spindle poison nocodazole, but they are resolved when mitotic spindles reform ( Figure 5C). This demonstrates that removal of SCIs occurs after bipolar attachment of the mitotic spindles as the cells initiate anaphase. Indeed, we show that condensin-dependent overwinding, which is activated by spindle attachments (Baxter et al., 2011;Leonard et al., 2015), correlates with the removal of SCIs, suggesting a regulated mechanism for the clearance of SCIs from chromosomes at this stage. Pioneering studies with frog extracts showed that condensin overwinds DNA, generating positive supercoiling in the presence of topoisomerases (Kimura and Hirano, 1997). We recently demonstrated that yeast condensin promotes positive supercoiling of minichromosomes when spindle microtubules attach to kinetochores (Baxter et al., 2011;Leonard et al., 2015). To stabilize, and hence detect, the positive supercoiling status of monomer plasmids, Top2 inactivation was required (Baxter et al., 2011); this demonstrates that positively supercoiled plasmids are a strong substrate for Top2 relaxation activity. Importantly, human Top2 showed a bias toward decatenation when catenated plasmids were positively supercoiled, but not when they were negatively supercoiled (Baxter et al., 2011), suggesting that condensin-dependent overwinding promotes decatenation by Top2 through alteration of substrate topology (Baxter and Aragó n, 2012). Here, we have been able to confirm that condensin overwinding on cohesed chromosomes leads to SCI removal, and that inhibition of this activity results in new SCI formation in metaphase ( Figure 5C). Therefore, we propose that condensin overwinding upon bipolar attachment of sister chromatids to spindles represents a cell-cycle-regulated mechanism for SCI clearance in cohesed mitotic chromosomes. During our investigation we observed that increasing the amounts of Top2 in cells caused the accumulation of SCIs and altered the supercoiling status of minichromosomes; this suggests that enzyme levels are critical when considering Top2 activities on chromatin. The cellular amount of Top2 is clinically relevant because patients with amplification of Top2 respond favorably to anthracycline-based chemotherapy (O'Malley et al., 2011). Indeed, it has been known for some time that increased levels of Top2 in mammalian cells lead to hypersensitivity to Top2-targeting drugs, whereas reduced levels cause resistance to these poisons (Nitiss and Beck, 1996). Our findings demonstrating that cellular levels of Top2 correlate with Top2 activity are consistent with these observations, since higher amounts of drug-induced DNA lesions can be expected with increased cellular levels, and hence activity, of Top2. The importance of cellular mechanisms maintaining Top2 homeostasis has been underestimated; our data suggest that either too little or too much Top2 in cells is likely to generate genome instability and chromosome loss over generations. It follows that strategies to modulate cellular Top2 levels might present a novel therapeutic opportunity, because Top2 poisons are widely used in human cancer treatment. Our findings provide a new view for the mechanisms behind SCI formation and resolution ( Figure 6). We propose that, during DNA replication, the bulk of SCIs are produced as a consequence of fork swiveling, particularly during replication completion. Following replication, the proximity of sister DNAs brought about by their pairing at cohesin sites facilitates the formation of SCIs between sister chromatids by Top2 (Figure 6). Importantly, we predict that Top2 activity at sites of physical proximity, cohesin sites, is bidirectional: that is, both formation and removal of SCIs occur at these sites ( Figure 6), leading to a situation where the equilibrium results in a probability that the site will contain a given number of SCIs at any particular moment in time. If premature cohesion loss occurs, as is the case in cohesin mutants, a reduction in SCI formation would ensue because of the increased physical separation of chromatids; consequently, the global levels of SCIs would diminish. When cohesed sister chromatids become bioriented on the mitotic spindle, during the experiment. SCIs were increased in the cdc20-td smc2-8 strain. SCIs (including a-, b-, and c-type catenanes) were quantified using ImageJ software (bottom right-hand graph; showing the mean ± SD from three independent experiments). (B) A cdc20-td top2-td strain bearing the centromeric plasmid pRS316 was synchronized in G 1 , split in two, and released into the cell cycle under conditions of Cdc20 and Top2 depletion. Nocodazole was added to one of the samples after 120 min. Samples were collected every 20 min and DNA was resolved in agarose gels and probed for the circular minichromosome. Electrophoretic mobilities of monomers (OCm, Lm, and CCCm) are indicated. Dimeric forms for CatC-type catenanes (negatively supercoiled) and CatC-type catenanes (positively supercoiled) (Baxter et al., 2011) are also indicated. Addition of nocodazole led to the shift back from CatC to CatC. (C) A METp-CDC20 strain (Cdc20 under Methionine promoter) (top panel) bearing the 10 kb circular minichromosome was synchronized in G 1 and released into the cell cycle in the presence of nocodazole under Cdc20-depleting conditions (YPD + 5 mM methionine) at 36 C. Following metaphase arrest, cells were transferred to new media lacking nocodazole but maintaining the Cdc20 depletion (hence allowing formation of mitotic spindles) for 120 min before readdition of nocodazole. Samples were taken for analysis at the first nocodazole block (metaphase arrest À Cdc20 depletion + nocodazole), 120 min after nocodazole removal (metaphase arrest À Cdc20 depletion) and 120 min after nocodazole readdition (metaphase arrest À Cdc20 depletion + nocodazole readdition). A METp-CDC20 top2-4 (bottom panel) bearing the 10 kb circular minichromosome was synchronized in G 1 and released into the cell cycle under conditions of Cdc20 depletion (YPD + 5 mM methionine) at 25 C. Nocodazole was added to the culture, and the temperature was shifted to 36 C to inactivate Top2. Samples were taken at Cdc20 block and 120 min after nocodazole addition. Cell extracts were separated by differential sedimentation, heat denatured in 1% SDS, and analyzed by Southern blotting to reveal plasmid species. Plasmid species included monomeric (OCm, Lm, and CCCm) and dimeric forms, including different SCI species (Catc-type catenanes and Catb-type catenanes). Running position for SCIs in the blots is indicated (dashed boxes). The mitotic arrest was stable during the experiments. SCIs (including c-and b-type catenanes) were quantified using ImageJ software (bottom left-hand graph; showing the mean ± SD from three independent experiments). SCIs were reduced when nocodazole was removed, and reappeared upon its readdition in the presence of Top2, but not in its absence. condensin-mediated overwinding alters DNA topology, in ways we do not yet understand, so that Top2 gains a bias toward decatenation and SCIs are removed in cohesed chromatids despite their physical proximity (Figure 6), leaving cohesin cleavage by separase as the last cellular task needed to separate the two daughter genomes. Media and Cell-Cycle Synchronizations Yeast strains bearing the 10 kb plasmid were grown in synthetic media lacking tryptophan and supplemented with 2% glucose (ÀTrp+D) at 25 C. Exponentially growing cultures were subsequently diluted to OD 600 = 0.25 in YEP (2% dextrose) media and allowed to grow to attain OD 600 = 0.6-0.8 at 25 C. Cells were arrested at metaphase in the presence of 15 mg/mL nocodazole for 3 hr. Strains bearing the temperature-sensitive degron cdc20-td or top2td with or without the temperature-sensitive allele smc2-8 or top2-4 were grown overnight at 25 C in minimal media lacking tryptophan ÀTRP (2% dextrose). Exponentially growing cultures were diluted to OD 600 = 0.25 in YEP (2% raffinose) and allowed to grow for 8 hr before being diluted to OD 600 = 0.05 in YEP (2% raffinose) and allowed to grow overnight. On day 3, exponentially growing cultures were subsequently diluted to OD 600 = 0.25 and allowed to grow for 3 hr at 25 C. Temperature-sensitive alleles were then triggered by the addition of galactose to a 2% final concentration, the addition of doxycycline to a 50 mg/mL final concentration, and a temperature shift to 37 C. Where a release from an early metaphase arrest to a cdc20-td-mediated arrest was required, cells were first arrested with nocodazole; the nocodazole was then washed out and pellets were resuspended in at 37 C in 50 mg/mL doxycycline and either YEP (2% raffinose) or YEP (2% galactose). Cells remained in these conditions for 2 hr to ensure that passage to a cdc20-td-mediated arrest was attained. In experiments where Cdc20 was under control of the MET promoter, strains were cultured in minimal media lacking methionine ÀMET (2% raffinose), and Cdc20 inactivation was triggered by resuspension of the pellet in YEP (2% dextrose). For TOP2 overexpression, exponentially growing cells in YEP (2% raffinose) were arrested in G 1 by adding a factor, followed by release in YEP (2% glucose) or YEP (2% galactose) containing nocodazole (15 mg/mL). After 2 hr, cells reached G 2 /M. Samples were taken at that point and 2 hr later. Cells were kept arrested in nocodazole by the addition of a second dose of the drug 2 hr after the release from a factor. FACS Analysis Approximately 1 3 10 7 cells were sedimented at 9,500 rcf for 1 min, and pellets were resuspended and stored in 500 mL 70% ethanol at 4 C. Pellets were then resuspended in 250 mL 13 saline sodium citrate (SSC) supplemented with 0.1 mg/mL RNaseA and incubated at 37 C overnight. Then 50 mL 13 SSC supplemented with 0.75 mg/mL Proteinase K was added to the tubes, and cells were incubated at 50 C for 1 hr. A further 250 mL 13 SSC was added to the tubes, and samples were sonicated for 2 cycles (30 s on, 30 s off, low power) at 4 C. A total of 250 mL 13 SSC supplemented with 3 mg/mL propidium iodide was then added to the tubes and samples incubated at room temperature for 1 hr. Finally, 200 mL of each sample was read with a Becton Dickinson FACSCalibur, ensuring 20,000 events per sample. Genomic DNA Preparation Cells were harvested at 4,000 rpm, and pellets were washed in ice-cold H 2 O and stored at À80 C. Pellets were resuspended in 11 mM Tris-HCl and 10 mM DTT and incubated on ice for 20 min. Cells were washed in ice-cold H 2 O, resuspended in spheroplasting buffer (1 M sorbitol, 50 mM Tris-HCL [pH 7.5], 1 mM CaCl 2 , 1 mM MgCl 2 , 14 mM b-mercaptoethanol, and 2,200 units Zymolyase T100), and incubated for 30 min at 37 C with shaking. Figure 6. Diagrammatic Model for SCI Formation and Resolution during the Cell Cycle During DNA replication, pre-catenanes are formed as fork swiveling occurs, leading to the formation of SCIs upon replication completion. During G 2 , Top2's bidirectional activity acts on SCIs by removing them, but also by catalyzing their formation de novo at cohesin sites, where the physical proximity between chromatids offers this opportunity. Therefore, cohesin sites represent regions of the genome where SCI levels are at equilibrium, i.e., where Top2's SCI removal and catalysis co-exist. Catalysis of SCI formation at these regions is likely to be a consequence of the presence of intermolecular crossovers between sister chromatids at or adjacent to cohesin sites. If premature cohesion loss is induced by inactivation of cohesin, a reduction in SCI formation would ensue (because of the physical separation of chromatids); consequently, the global levels of SCIs would diminish. Upon bipolar spindle attachment, condensin action (promoting overwinding and chromosome compaction) enhances Top2's monodirectional activity toward SCI resolution, potentially by generating distinct chromatid identities and reducing the probability of intermolecular crossovers between sister chromatids, so that when SCI removal occurs SCI catalysis is not favorable. Following condensin activation and SCI removal, proteolytic cleavage of cohesin's subunit Mcd1 triggers anaphase chromosome segregation. Spheroplasts were sedimented in a fixed-angle rotor (Eppendorf F-34-6-38) at 6,000 rpm for 6 min, gently washed with 1 M sorbitol, transferred to 1.5 mL tubes, and sedimented for 1 min at 3,500 rcf at 4 C. Pellets were resuspended in 200 mL cold 0.4 M sorbitol, lysed on ice for 30 min by addition of 700 mL lysis buffer (25 mM HEPES/KOH [pH 8], 50 mM KCl, 10 mM MgSO 4 , 0.25% Triton X-100, 1 mM PMSF, 3 mM DTT, and protease inhibitor cocktail tablet [Roche]), and supplemented with 100 mg/mL RNaseA and 300 mM NaCl. Cell extracts were subsequently obtained by spinning the lysed spheroplasts at 12,000 rcf at 4 C for 5 min. Cleared lysates were loaded onto sucrose gradients prepared in a Biocomp gradient station and sedimented in a SW41 rotor (Beckman Optima L-100 XP Preparative Ultracentrifuge) at 18,000 rpm for 9 hr at 4 C. Gradients were fractionated into 30 equal fractions using a Biocomp gradient station. Southern Blot Analysis Gradient fractions were separated on 0.5% agarose gels (0.53 TBE) containing 0.5 mg/mL ethidium bromide at 1.15 V/cm for 40 hr at 4 C. Gels were transferred to positively charged nylon transfer membranes (Hybond-N+, Amersham Biosciences). Blots were hybridized with fluorescein-labeled probes against plasmid-specific regions, and membranes were visualized following incubation with CDP-Star detection reagent for 10 min at room temperature. Three biological replicas were done for the Southern blot experiments. SCI were quantified using ImageJ software. The total amount of SCIs (including Cata, Catb, and Catc species) and monomers (supercoiled, relaxed, and linear species) was quantified. Graphs show the ratio between SCIs and monomers. Fluorescence Microscopy A series of z-focal plane images were collected on Leica IRB using a Hamamatsu D742-95 digital camera and OpenLab software (Improvision). Images taken were phase, UV, and GFP filtered. To visualize the nuclei of intact cells, cells were resuspended in 1% Triton X-100 supplemented with DAPI. Sample Preparation and Western Blot Analysis Samples were prepared by TCA extraction. Extracts were prepared as follows. Cells were collected by centrifugation (4,000 rpm, 2 min) and washed with 20% TCA. The TCA was aspirated and the pellets frozen at À80 C. All of the following purification steps were performed on ice with pre-chilled solutions. Cells were resuspended in 250 mL 20% TCA, glass beads were added, and the cells were broken by one 40 s cycle (power 5.5 in a FastPrep FP120 [BIO 101] machine). Tubes were pierced with a hot needle, placed onto fresh Eppendorfs, and spun (1,000 rpm, 2 min) to collect lysate minus glass beads. The glass beads were washed with 1 mL 5% TCA; this was added to the lysate and mixed by pipetting. The precipitated proteins were collected by centrifugation (14,000 rpm, 10 min, 4 C), and then pellets were washed with 750 mL 100% ethanol. Proteins were solubilized in 50 mL 1 M Tris (pH 8) and 100 mL 32 SDS-PAGE loading buffer (60 mM Tris [pH 6.8], 2% SDS, 10% glycerol, and 0.2% bromophenol blue) and boiled for 5 min at 95 C. Insoluble material was removed by centrifugation (14,000 rpm, 5 min, room temperature), and the supernatant was either stored at À20 C or loaded immediately onto a SDS-PAGE minigel. Samples were either run on an 8% acrylamide gel in Tris-Glycine SDS or on running buffer using the Bio Rad Mini-PROTEAN 3 system. SDS-PAGE gels were transferred to polyvinylidene fluoride transfer membrane (Hybond-P, Amersham Biosciences) either in the Bio-Rad Mini Trans-Blot Electrophoretic Transfer Cell or by using the XCell SureLock Mini Cell Transfer module (Invitrogen). The Bio-Rad system was used in conjunction with Tris-Glycine blotting buffer (National Diagnostics) containing 20% methanol and run for 1 hr at 200 V or overnight at 30 V. Membranes were blocked in 5% skimmed milk powder in PBS with 0.1% Tween 20 (PBS-T) for 1 hr or overnight at 4 C, then incubated with anti-Rad53 antibody 12CA5 at a 1/5,000 dilution in blocking solution for between 1 hr at room temperature to overnight at 4 C. Following several washes in PBS-T, membranes were incubated with the sheep anti-rabbit IgG horseradish-peroxidase-linked antibody (GE Healthcare) at a 1/10,000 dilution in blocking solution. After several further washes in PBS-T, membranes were incubated with the ECL Plus Western Blotting Detection System (GE Healthcare) followed by exposure to ECL Hyperfilm (GE Healthcare) to detect the secondary antibody.	2018-04-03T02:21:06.027Z	2016-10-06T00:00:00.000	{ "year": 2016, "sha1": "6eb19e7a3d9ea7d9ba56444c406f13b55c50e615", "oa_license": "CCBY", "oa_url": "http://www.cell.com/article/S109727651630524X/pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "0735a94e082c664df18d64f9b4e9903570a3f5f8", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
270649921	pes2o/s2orc	v3-fos-license	Variation and convergence in the morpho-functional properties of the mammalian neocortex Man's natural inclination to classify and hierarchize the living world has prompted neurophysiologists to explore possible differences in brain organisation between mammals, with the aim of understanding the diversity of their behavioural repertoires. But what really distinguishes the human brain from that of a platypus, an opossum or a rodent? In this review, we compare the structural and electrical properties of neocortical neurons in the main mammalian radiations and examine their impact on the functioning of the networks they form. We discuss variations in overall brain size, number of neurons, length of their dendritic trees and density of spines, acknowledging their increase in humans as in most large-brained species. Our comparative analysis also highlights a remarkable consistency, particularly pronounced in marsupial and placental mammals, in the cell typology, intrinsic and synaptic electrical properties of pyramidal neuron subtypes, and in their organisation into functional circuits. These shared cellular and network characteristics contribute to the emergence of strikingly similar large-scale physiological and pathological brain dynamics across a wide range of species. These findings support the existence of a core set of neural principles and processes conserved throughout mammalian evolution, from which a number of species-specific adaptations appear, likely allowing distinct functional needs to be met in a variety of environmental contexts. Introduction Questions about the specific properties of the human brain originated in the debates that followed the publication of Darwin (1859)'s Origin of Species.Challenging long-held beliefs, largely inherited from the biblical Genesis account, that humans were by essence different from other animals, the application of the theory of evolution to humans soon gave rise to lively discussions about the biological traits likely to differentiate humans from apes and stimulated the first thoughts about the relationship between brain size and cognitive ability.Considered a good indicator of intelligence by most scientists in the second half of the 19 th century, overall brain size eventually proved to be an irrelevant measure of behavioural complexity, prompting the exploration of other levels of brain organisation.Thanks to in vitro investigations on postoperative tissue, our knowledge of the human neocortex-the cerebral structure whose functioning is critically involved in behavioural and cognitive abilities-has progressed considerably in recent years, providing the basis for inter-species comparisons at multiple levels.After tracing the historical origins of these questions back to the controversies surrounding the first discoveries of human fossils, this review is intended to provide an updated view of the functional organisation of the neocortex in representatives of the three major mammalian radiations, including humans.We will first analyze variations in brain size, number of cortical areas and number of neurons at a macroscopic level, before focusing on the anatomical and physiological features of neocortical pyramidal neurons.Adopting a reductionist approach, we will systematically compare (when data permit) the morphology, connectivity and electrical properties of pyramidal neuron subtypes between species, attempting to determine, where differences are observed, whether they are part of a continuum of variations or whether they represent genuine singularities leading to significant changes in the activity patterns of cortical circuits. Eugene Dubois and the missing link Proponents of Darwin's theory, led by the German biologist Ernst Haeckel, published essays in the 1860s discussing the status of the human species from an evolutionary point of view.According to Haeckel, the process of hominization was based on the acquisition of bipedalism, language and a large brain.In his family tree of the human species, which he attempted to reconstruct on the basis of comparative anatomical and embryological data, Haeckel inserted an intermediate evolutionary stage between the great apes and man, occupied by a hypothetical species called Pithecanthropus (ape-man) alalus (speechless).He imagined that this mute ape-man, originating from a continent now sunk in the Indian Ocean (Lemuria), could have spread and evolved in different parts of the world to give rise to different humanities and languages (Haeckel, 1868).Inspired by the work of Darwin and Haeckel and convinced of the necessity to support the theory of evolution with palaeontological evidence, in 1887 the young anatomist Eugène Dubois took the surprising decision to quit a promising academic career at the University of Amsterdam to mount an excavation campaign in the Dutch East Indies in search of the missing link between apes and humans.Accompanied by his wife Anna Lojenga and their daughter, Dubois enlisted as a medical officer in the Royal East India Army and set sail for the Indonesian archipelago aboard the Princess Amelia (Theunissen, 1988;Wood, 2020). After arriving on the island of Java, Dubois conducted extensive excavations in the summer of 1891 near the village of Trinil, along the Solo River (Figure 1).He initially uncovered a right maxillary third molar and a skull cap whose characteristicsa receding forehead, a supra-orbital torus and an estimated capacity of 700-750 cm 3 (about half the size of present-day humans)-suggested a large ape.However, the next year, Dubois unearthed a left femur some 15 metres upstream from the first remains, showing clear adaptations to upright posture and bipedalism (Theunissen, 1988).By discovering an individual whose skull and teeth displayed anthropoid ape characteristics but whose femur showed human-like features, Dubois had just uncovered fossil evidence of the hypothetical transitional primate envisioned by Haeckel, and at the same time provided one of the first material indications of human evolution.Initially named Anthropithecus erectus (an ape that stands and moves like a man) in the excavation reports, Dubois later renamed his new species Pithecanthropus erectus, emphasising its status as an upright "ape-man, " when he published his final manuscript (Figure 1).In good faith, he even revised the cranial capacity of his specimen to 850-900 cm 3 (Dubois, 1894(Dubois, , 1896;;Wood, 2020). The discovery of Pithecanthropus erectus sparked controversy, particularly over the attribution of the remains to an ape, a human, an intermediate being, or even to different individuals.Dubois had to face the scepticism of his prominent European colleagues who would have preferred an ancestor with a larger brain but less exotic origins (Leguebe, 1992), such as the Piltdown Man discovered in Sussex in 1912, which had a large human-like skull and an ape-like mandible, but which ultimately turned out to be one of the greatest paleoanthropological hoaxes (Stringer, 2012).In response to debates over the interpretation of his fossils, Dubois undertook research on the allometric relationship between brain weight and body weight.Extending earlier theoretical works (Snell, 1892), he established that brain size was not only related to body weight by a decreasing power function but also depended on a "coefficient of cephalization, " supposed to reflect the degree of development and complexification of the brain (Dubois, 1897).Applying this mathematical relationship to his fossils, Dubois calculated that the cephalization coefficient of Pithecantropus erectus was roughly half that of anatomically modern humans and double that of apes, further confirming the intermediate evolutionary position of the Javanese primate (Dubois, 1899). Pithecanthropus erectus is no longer considered the missing link.The idea of a hybrid creature, half ape and half human, making a direct transition between great apes and modern man, now belongs to the realm of fiction.The accumulation of fossil discoveries has led researchers to abandon the traditional vision of a linear and directed human evolution in favour of a more complex and diversified human lineage, made up of multiple species that most often coexisted.The concept of the missing link, which encompasses both the notion of continuity and rupture, remains pertinent as it questions the singularity of the "after" in relation to the "before."Man is often seen as a species apart, distinguished by the complexity of its cultures, social interactions and ability to communicate.A species whose considerable brain growth over time would have accompanied the emergence of remarkable cognitive capacities, enabling man to conquer and transform almost all of the terrestrial ecosystems.But the question remains: does our large brain possess truly distinctive properties?And if so, are these differences merely quantitative or do they represent a genuine qualitative leap? The long-standing question of size Questions of absolute or relative brain size seem to have preoccupied mankind long before Dubois and his contemporaries since they already appear in the writings of Aristotle, who notes that "of all animals, man has the largest brain in proportion to his size" (Aristotle, ca.335 BCE); a statement that is not entirely accurate, as we shall see below.However, it was mainly in the second half of the 19 th century that the relationship between brain size and human cognitive ability became a central theme of discussion among the scientific community, particularly within the Société d'Anthropologie de Paris, founded by the eminent neuroanatomist Paul Broca the year Darwin published his Origin of Species.A partisan of polygenist theories, which, in opposition to the creationist myth, favoured a multiple origin for the different human groups, but unfortunately mired in the prejudices of his time, Broca mistakenly thought that he could rank ethnic groups (and human beings in general) according to their level of intelligence by comparing the weight of their brains.By sorting individuals according to their ethnic origin, sex or profession, Broca came to the conclusion that white European men, whom he described as "distinguished" (as opposed to manual workers), were endowed with superior intelligence (Broca, 1861).In a similar vein, Francis Galton, an anthropologist renowned for his contributions to modern statistics and, more infamously, for his eugenics theories, later conducted a study into the brain size of Cambridge students.He also claimed, on the basis of measurements of questionable rigour, that those who graduated with honours had larger brains than those who did not receive such distinction (Galton, 1889). In the wake of phrenology, which claimed to determine character traits and mental faculties by inspecting the size of bumps on the surface of skulls, the underlying-perhaps somewhat simplistic-assumption behind these early attempts to explain differences in cognitive ability by brain size was that any increase in the size of an organ should correspond to an increase in its function.Craniometric studies based on the social or geographical origins of human beings, which served as scientific justification for the expansionist ambitions of many countries in the 19 th and 20 th centuries, fortunately declined after the Second World War.However, the search for principles behind the evolution of the mammalian brain has never ceased to intrigue scientists, who continue to study variations in absolute and relative brain size between species, as well as the evolution of its different parts, paying particular attention to the neocortex because of its essential role in the generation of complex behaviours. . Brain size in mammals As neuronal tissue does not fossilise, the formulation of general principles on the evolution of the brain requires a comparative analysis of cerebral organisation in living species (and, to a lesser extent, the study of endocranial casts of fossil specimens), based on the hypothesis that the characteristics present in the current members of a phylogenetic radiation can be explained more parsimoniously as being inherited from a common ancestor.The earliest mammals have likely evolved from mammal-like reptiles at the end of the Triassic over 200 million years ago, in the form of small-brained shrew-like creatures that probably laid eggs, like present-day monotremes (Kaas, 2011).Monotremes (prototherians), one of the three main extant mammalian groups, diverged from therians around 166-186 million years ago, while the marsupial (metatherian) and placental (eutherian) lineages are thought to have split around 147-160 million years ago (Bininda-Emonds et al., 2007;Phillips et al., 2009). The brain size of present-day mammals is extremely variable, ranging from <0.1 g in the Etruscan pygmy shrew to more than 9 kg in some large cetaceans (DeFelipe, 2011).In placentals, evolutionary processes have led to the emergence of large brains in several groups of species sometimes separated by a long independent phylogenetic history (Manger et al., 2013).These include a large proportion of whales (up to 9,200 g for the sperm whale) and dolphins (up to 2,900 g), elephants (up to 6,000 g), certain pinniped species (such as walruses or southern elephant seals ∼1,200 g), and members of the genus Homo (see Figure 2).Modern Homo sapiens, with an average brain mass of 1,350-1,400 g (varying from 1,100 to 1,800 g), are a long way behind elephants and cetaceans, but can nevertheless claim first place among primates, since the largest brains of the great apes do not exceed 500-600 g (Tower, 1954;Jerison, 1973;Haug, 1987;Roth and Dicke, 2005;Neubauer et al., 2018).Differences in brain size in present-day Homo sapiens do not appear to have functional significance, since substantial variations can be observed between individuals with apparently similar intellectual abilities (DeFelipe, 2011).The idea that a larger absolute brain size should necessarily confer more complex cognitive abilities or greater behavioural flexibility is also challenged by observations that animals with similar brain weights, such as gorillas and oxen (∼500 g) or elephant seals and some humans (∼1,200 g), have different behavioural repertoires, or that species with relatively small brains, such as dogs (60 g), rats (2 g) and mice (0.5 g) in mammals or corvids (6-15 g) in birds, can demonstrate sophisticated behaviours (Kaminski et al., 2004;Olkowicz et al., 2016). Another widely studied property is relative brain size.In mammals, brain size and body size are closely correlated, with a negative allometry.As a result, small species tend to have relatively larger brains in proportion to their body size than large species.In humans, for example, the brain represents around 2% of body mass, whereas this ratio can reach 10% in shrews and small rodents (Van Dongen, 1998;Roth and Dicke, 2005).It is this negative scaling relationship that led Dubois to introduce the "coefficient of cephalization, " later renamed the encephalization quotient (EQ), as a suitable measure for comparing relative brain size across species (Jerison, 1973).The EQ quantifies how much a species brain size deviates from what is expected based on its total body mass, using a standard species from the same taxonomic group as a reference.When calculating EQs in mammals, using the cat as the standard, humans emerge as the most encephalized species with an EQ of around 6-7, meaning that their brains are more than six times larger than expected.However, a high degree of encephalization is not exclusive to humans.Dolphins like the Tucuxi or the whitesided dolphin are not far behind humans with EQs around 4-5, followed by capuchin monkeys with EQs ranging from 2.4 to 4.8.Then come species like gorillas and chimpanzees (EQ = 1.5-3)-though often considered more cognitively able than capuchin monkeys (Deaner et al., 2007), elephants (EQ = 1.1-2.4),and rodents with relatively low EQ values of 0.5-1 (Jerison, 1973;Marino, 1998Marino, , 2002;;Roth and Dicke, 2005;Shoshani et al., 2006).However, the ability to predict the cognitive abilities of a species from EQ values remains quite limited, mainly due to the sensitivity of this metric to the choice of the reference taxonomic group and to the value of the exponent of the power law relating brain size to body size (typically between ∼0.55 and 0.8) used in the various studies (Harvey and Krebs, 1990;Marino, 1998;Charpier, 2008). . Brain size in hominins The cranial capacity of hominins has undergone a significant increase over the past 3 million years, evolving from 450 cm 3 in Australopithecus (a value similar to that of the great apes) to approximately 1,350-1,500 cm 3 in Homo sapiens and Homo neanderthalensis (that went extinct ∼40,000 years ago) (Figure 2A) (Jerison, 1973;Holloway, 2015;Neubauer et al., 2018).This cerebral expansion, particularly evident in Homo erectus (∼2 million years ago), was initially closely linked to changes in body weight and then progressed more independently over the last 500,000 years (Ruff et al., 1997;Hublin et al., 2015), explaining the high EQ values of Homo sapiens.It is interesting to note that Homo floresiensis, whose remains were discovered in the Liang Bua cave on the island of Flores, represents an exception to this brain size expansion in the Homo lineage.Despite its ability to create sophisticated tools and perhaps to navigate, Homo floresiensis (dated between ∼100,000 and 50,000 years ago) had unexpected morphological features such as a small stature (∼1 metre) and a small endocranial volume of around 430 cm 3 (Balzeau and Charlier, 2016;Sutikna et al., 2016).Similarly, Homo naledi, a new species recently discovered near Johannesburg in South Africa, which coexisted with Neanderthals and potentially with the first Homo sapiens, also possessed small brain capacities ranging from 465 to 610 cm 3 (Garvin et al., 2017).These fossil discoveries are profoundly challenging our knowledge of human diversity and the long-held idea of a continuous evolution towards ever-larger human beings with evermore-developed brains. The increase in size of the hominin brain over the course of evolution entails an increase in its energy cost.The metabolic requirements of a human brain are considerable: accounting for 50-55% of basal metabolism at birth, this proportion peaks at over 65% at the age of 4-5 years and remains high, at around 20%, in adulthood (Clarke and Sokoloff, 1999;Kuzawa et al., 2014).It is believed that the consumption of energy-rich foods like meat and marrow, and, later, their improved digestibility through cooking (Carmody and Wrangham, 2009), have enabled mothers to allocate more energy to their foetuses during pregnancy (and their newborns during nursing) and have thus favoured the development of larger brains (Martin, 1996).Thus, the large brains of early Homo may have emerged as an unintended byproduct of a change in maternal diet, perhaps initiated by a modification in climate and available resources.Another hypothesis suggests that a reallocation of energy to the brain may have been facilitated by a reduction in the size of other metabolically costly organs, such as the digestive system (Aiello and Wheeler, 1995).Finally, the social brain hypothesis proposes that the evolution of hominin encephalization could be the result of increasingly complex social demands in group-living species (Dunbar and Shultz, 2007). Compared with chimpanzees and macaques (∼40 and 70% brain growth in utero, respectively), a significant proportion of brain growth in humans occurs after birth.The size of the brain at birth is thought to be partly constrained by the anatomy of the woman's pelvis, whose dimensions are limited by biomechanical and postural factors associated with bipedalism (Hublin et al., 2015).From a size of around 400 cm 3 at birth (∼28% growth in utero), the brain of a young Homo sapiens undergoes rapid growth during the 1 st year, by the end of which it reaches 50% of its adult size, finally attained at the age of 5-6 years (Leigh, 2004;DeSilva and Lesnik, 2006).This perinatal phase of brain growth is accompanied by a change towards a more globular shape, typical of modern humans.This globularization process, which evolved progressively to reach the current variation between 100,000 and 35,000 years ago, is not observed in Neanderthals and chimpanzees (Figure 2B) (Bruner et al., 2003;Gunz et al., 2010;Neubauer et al., 2010Neubauer et al., , 2018)). . The neocortical expansion While a simple cortex is already present in the pallium of reptiles, the neocortex first appears as a complex multi-laminated structure in mammals (Nieuwenhuys, 1994;Rakic, 2009).Of modest proportion (20-48%) in small species of the three main radiations, such as platypuses, opossums, shrews and mice, it reaches considerable size in cetaceans and primates.Accounting for 75 to 84% (depending on the study) of the total mass or volume of the brain, the human neocortex is proportionally one of the largest among mammals.Humans are however closely followed (and sometimes equalled) by most odontocete cetaceans (∼72%), as well as several species of monkeys and apes, such as macaques (70-76%), grivets (79%), or chimpanzees (70-73%) (Pirlot and Nelson, 1978;Stephan et al., 1981;Hofman, 1988;Rilling and Insel, 1999;Manger, 2006).Within the neocortex, a pronounced enlargement of the frontal lobe, thought to be involved in higher cognitive functions, has long been regarded as a hallmark of human evolution.Brodmann (1912), shortly after completing his famous cytoarchitectural map of the cerebral cortex, published a comparative study of the frontal cortex surface in primates, demonstrating a progressive increase from prosimians to humans.While more recent studies have indeed demonstrated an increase in the absolute size of the frontal cortex in humans (Semendeferi et al., 1997(Semendeferi et al., , 2002;;Bush and Allman, 2004), this expansion does not seem to significantly differ from what would be expected from a great ape with a human-sized brain (Semendeferi et al., 2002;Barton and Venditti, 2013). Neocortex thickness generally correlates positively with brain size (Hutsler et al., 2005;Balaram and Kaas, 2014).However, this correlation does not uniformly apply to all taxa, as evidenced by the typically thin cortex (<2 mm) of cetaceans (Ridgway and Brownson, 1984).The average variation in cortical thickness between species (between 0.4 and 2.8 mm), which is comparable to the variability found between cortical areas in the same animal, remains relatively modest compared to the variation in overall brain size.This suggests that the expansion of the neocortex in large-brained mammals is mainly the result of an increase in its surface area (rather than its thickness), often resulting in the formation of convolutions, particularly visible in large primates and cetaceans (Hofman, 1985(Hofman, , 1988;;DeFelipe, 2011 for review).In primates, a gradual increase in neocortical thickness is observed from primary to more integrative sensory areas, a trend seemingly absent in motor and frontal association cortices (Wagstyl et al., 2020).Finally, the relative thickness of cortical layers also varies between species, with supragranular layers being proportionally thicker in primates than in carnivores and rodents, while infragranular layers show an inverted profile (Hutsler et al., 2005). Intrinsic organisation of the neocortex . Areas and columns The mammalian neocortex is typically subdivided into six layers defined by vertical differences in the size, shape, or density of neurons (Brodmann, 1909).There are, however, variations in the number, thickness or overall cytoarchitectonic organisation of the layers across the cortical mantle (Kaas, 1987;DeFelipe, 2011), which have formed the basis of its subdivision into distinct regions. The neocortex is classically regarded as a complex mosaic of anatomically and functionally specialised areas whose number increases with brain size (Brodmann, 1909).From about 10 to 15 cytoarchitectonic subdivisions mainly dedicated to sensory processing (primary somatosensory, visual, and auditory areas) and motor functions (although controversy remains over a clear separation of somatosensory and motor regions in some early mammals) in small-brained species (Krubitzer et al., 1995;Catania et al., 1999;Kaas, 2011), their number could approach 200 in humans (Glasser et al., 2016).In large-brained species, primary sensory and motor fields subdivide-more than 10 to 20 areas identified for the single visual cortex in cats and monkeys-, change in size and relative position and become separated by the inclusion of associative areas notably in the frontal and temporoparietal regions (Nieuwenhuys, 1994;Northcutt and Kaas, 1995).The small amount of cortical territory devoted to multimodal or association areas in monotremes suggests that unimodal sensory fields could constitute the core of the prototypical plan for neocortical organisation in mammals (Krubitzer et al., 1995).The developmental mechanisms responsible for this elaborate cortical parcellation have been debated as to whether the structural differences between areas are induced in a homogeneous population of cortical neurons by the patterned activity of thalamocortical projections, or whether the formation of the neocortical map is already genetically determined in the neural progenitors of the embryonic ventricular zone.In this later view, which seems to be gaining consensus, areas in the cortical plate would attract appropriate inputs rather than being specified by them.Activity-dependent mechanisms would then play an influential role at later stages in refining existing synaptic connexions (Rakic, 2002). The predominance of vertical over horizontal connexions in his anatomical reconstructions of rodent cortical neurons, led Lorente de Nó (1938) to suggest in the 1930s that cortical areas were composed of multiple 'elementary units' of information processing taking the form of vertical bands of interconnected neurons.Almost 20 years later, Mountcastle obtained persuasive evidence of a columnar segregation of sensory modalities in the cat somatosensory cortex by showing that neurons recorded along different vertical microelectrode tracks responded either to superficial or deep cutaneous stimulation.He introduced the term 'cortical column' , assigned them an average width of ∼0.5 mm and demonstrated that the different functional columns were intermingled in the manner of a mosaic (Mountcastle, 1957).Columnar organisation formed by groups of neurons activated more strongly by stimulation of one of the two eyes (ocular dominance column) or by stimuli having a common receptive field axis orientation (orientation columns) were later discovered in the cat primary visual cortex (Hubel andWiesel, 1963, 1969).Comparing data obtained in cats and macaques, Hubel andWiesel (1963, 1974) found that similar variations in orientation tuning were obtained with smaller electrode advances in monkeys, suggesting thinner columns or less defined borders in this species.Variations in the width of ocular dominance columns (from 200 to 800 µm) were also reported in subsequent studies on different primate species including humans (Bugbee and Goldman-Rakic, 1983;Horton and Adams, 2005), finally leading to the introduction of a new entity, the minicolumn, whose iteration and lateral combination through short-range horizontal connexions would form the basis of functional columns.Developmentally, minicolumns reflect the radial migration of neurons from the proliferative ventricular zone into narrow (30-50 µm) translaminar chain of cells separated by neuropil (Buxhoeveden and Casanova, 2002;Rakic, 2002).According to the "radial unit hypothesis, " the surface expansion of the neocortex during mammalian evolution (by ∼10,000 times from shrews to the largest cetaceans; Hofman, 1985;Manger, 2006), with no comparable variation in its thickness, could result from a change in the genetic mechanisms that control the timing and/or mode of cell division in the ventricular zone, leading to an increase in the pool of founder cells at the origin of radial columns (Rakic, 1995;Chenn and Walsh, 2002). Anatomical and functional evidence for a modular organisation of the neocortex has been obtained in a wide range of species from different mammalian radiations.Alternating bands of corticocortical projections related to monoaural or binaural responses are observed in the cat primary auditory cortex (Imig and Adrián, 1977), and patchy arrangements of axon terminal fields are apparent in the auditory area of the short-beaked echidna (Dann and Buhl, 1995).An additional example is the discrete architectonic units, known as "barrels, " formed by neurons preferentially activated by the same facial whisker in the primary somatosensory cortex of rodents and several other mammals with whiskers (Woolsey and Van der Loos, 1970).In the platypus somatosensory area, regions where neurons respond only to cutaneous stimulation of the bill are separated from regions where neurons process both tactile and electrical inputs (Krubitzer et al., 1995).Most frequently observed in primary sensory systems, the presence of columns is also attested in primary motor and association cortices (Bugbee and Goldman-Rakic, 1983;Amirikian and Georgopoulos, 2003).The existence of relatively similar organisational patterns in various areas and species led to idea that modular units could represent a fundamental principle of cortical function in mammals, important for perception, cognition and memory (Eccles, 1981;Mountcastle, 1997).In this context, it is expected that the subdivision of cortical regions into iterated computational units capable of operating in parallel should increase the number of possible spatio-temporal combinations of activity, and hence the processing capacities of large brains.However, the concept of cortical columns has also been contested based on an apparent intra-and inter-species inhomogeneity in size, shape, and expression without obvious differences in cortical function.For example, ocular dominance columns are well defined in Old World monkeys and remain rudimentary in most New World monkeys (Hendrickson et al., 1978;Adams and Horton, 2003), despite similar visual abilities.Similarly, barrels are not found in all the marsupial species that possess whiskers, and some rodents, like the chinchilla, have barrel fields without engaging in whisking behaviour (Purves et al., 1992).These findings raise the possibility that cortical modules may have emerged in different forms during areal specification in mammals, without acquiring an obvious function in all species (Horton and Adams, 2005). . Neurons and synapses in numbers The mammalian neocortex contains approximately 15-25% of the total number of brain neurons (Azevedo et al., 2009;Herculano-Houzel, 2012).Initial assumptions that cortical columns were composed of a constant number of neurons in all mammals (Rockel et al., 1980) have been challenged by subsequent studies showing, using stereological and non-stereological counting methods, variations in the density of neurons between species, areas and layers (DeFelipe et al., 2002;Herculano-Houzel et al., 2008).Neuronal density in the neocortex generally tends to be inversely correlated with brain volume, with different scaling rules applying to different orders of mammals.Thus, for a similar increase in neocortex mass, the corresponding decrease in neuronal density seems to be less pronounced in primates than in other placental mammals (Haug, 1987;Manger, 2006;Herculano-Houzel, 2012;Herculano-Houzel et al., 2014), and marsupials are reported to have fewer neurons than placentals of equivalent brain size (Haug, 1987;Seelke et al., 2014).The total number of cortical neurons in the human brain (12-16 billion) therefore exceeds that measured in other large-brained species such as whales or elephants (6-11 billion), but this number aligns with expectations for a primate with a human-sized brain (Haug, 1987;Azevedo et al., 2009;Herculano-Houzel et al., 2014).However, Pinson and colleagues recently discovered that expression of the modern human variant of transkelotase-like protein 1 (hTKTL1)-but not of the Neanderthal variant (which differs by a single amino acid substitution)-in the embryonic mouse neocortex can increase the abundance of a specific type of basal progenitors and promote neuron production, especially in the frontal lobe.These findings suggest that, even within primates, species with similar brain sizes, such as Homo sapiens and Neanderthals, may exhibit variations in the number of neurons (Pinson et al., 2022). Differences in counting methods, variations in age and number of samples, or in the amount of cortical volume examined make it difficult to compare calculations of synaptic density between laboratories (DeFelipe et al., 2002).Nevertheless, most studies agree that the mean synaptic density in the adult (defined as the total number of excitatory and inhibitory synapses per unit volume of cortical tissue) do vary across species (between ∼250 and 1,000 million/mm 3 ), but relatively independently of brain size (see DeFelipe et al., 2002;Karbowski, 2014 for reviews).For instance, a recent comparative study conducted in 25 primate species (including humans) found relatively constant synaptic densities in the primary visual and inferior temporal cortex of the different animals (∼256 million/mm 3 ), varying by only 1.9-fold despite brain weights differing by about 500-fold (Sherwood et al., 2020).Although a certain percentage of synapses continue to be remodelled in adulthood, synaptic density in the adult neocortex is globally stationary.This period of stable synaptogenesis is preceded by major changes in the rate of synapse production, which is particularly high during the perinatal period.The duration of this massive increase in synapse density around birth varies widely between mammals, ranging from 2 weeks in rats, 1 month in cats, 4 months in macaques, to around 3 years in humans (Bourgeois, 2008).The number of synapses is then maintained at a maximum until puberty (which is delayed in primates compared with rats and cats), during which synaptic density decreases markedly to levels comparable to those observed in adults (Huttenlocher, 1979;Bourgeois and Rakic, 1993;Bourgeois, 2008;Elston and Fujita, 2014).The maturation period for synaptic architecture is therefore considerably lengthened in primates, suggesting that the sensory environment could play an important role in the configuration and refinement of cortical circuits. The number of synapses per neuron, usually estimated by dividing the synaptic density by the neuronal density in a given layer, positively correlates with brain volume.In primates, the overall number of synapses per neuron (including both excitatory and inhibitory cells) in the inferior temporal cortex is thus higher in humans (∼4,850 synapses/neuron) than in gorillas (∼3,550 synapses/neuron), chimpanzees (∼2,885 synapses/neuron), and macaques (∼2,160 synapses/neuron) (Sherwood et al., 2020).However, this positive scaling does not universally apply to all species, as sensory cortex neurons in mice are reported to have more synapses than in rats (DeFelipe et al., 2002).The ratio of synapse density to neuron density remains a relatively coarse measure of connectivity because it does not differentiate between different types of neurons and overlooks the fact that dendrites, particularly those of pyramidal neurons, generally span several layers.A more accurate estimate of the number of synapses received by a given neuron could perhaps be obtained by quantifying the spine density along small dendritic segments (assuming that each dendritic spine is contacted by at least one synaptic input) and extrapolating these measurements to a cumulative number of spines, considering the length of the different dendritic compartments.Such analyses revealed that the density of spines on pyramidal neurons from the supragranular layers of the temporal cortex is higher in humans than in macaques (1.35 times), marmosets (1.9 times), or mice (1.3 times) (Elston et al., 2001;Benavides-Piccione et al., 2002).Based on these calculations and measures of total dendritic length, the total number of synapses received by a human temporal cortex L2/3 pyramidal cell has been estimated to be around 20,000 (Eyal et al., 2018). . Constituent cell types and functional microcircuit organisation Cortical circuit computations rely on the dialogic interaction of two main classes of neurons: the spiny glutamatergic excitatory neurons (comprising pyramidal and stellate cells), processing and transmitting information within and/or outside the neocortex, and the smooth or sparsely spiny GABAergic inhibitory interneurons, which finely regulate synaptic activity of local populations of excitatory neurons, shaping network dynamics. The following sections will primarily address the structural characteristics of the pyramidal neuron, accounting for ∼70-80% of the total population of neocortical neurons in placental mammals.Qualified as the "psychic cells" of the brain by Santiago Ramón y Cajal, pyramidal neurons are distributed across all cortical layers (except L1, where they still extend dendrites), and are regarded as the cornerstone of the cortical microcircuitry.The typical eutherian mammalian pyramidal neuron is distinguished by its prominent apical dendrite, radially oriented towards the pia, and its skirt of basal dendrites radiating from the soma (Figure 3).Pyramidal cells can be broadly classified as intratelencephalic (IT) or extratelencephalic (ET), depending on whether their long-range axons are confined to telencephalic structures (such as the neocortex, striatum, or claustrum) or whether they additionally establish connexions with brain structures outside the telencephalon (thalamus, tectum, pons, spinal cord).IT neurons are distributed throughout layers 2 to 6, while ET cells are confined to the deeper layers 5-6 (Harris and Shepherd, 2015;Baker et al., 2018a for reviews).IT neurons are the sole source of interhemispheric connexions, conveyed through the anterior commissure in monotremes and marsupials, as well as through the corpus callosum in eutherians (Suárez et al., 2018).Spiny stellate cells, lacking a prominent apical dendrite and instead featuring a star-like dendritic arbour, are predominantly localised in L4 of primary sensory cortices.Below I present an overview of the organisation of cortical circuits, outlining the main classes of neurons and their input-output connectivity patterns.This description is mainly based on findings obtained in eutherian mammals (in particular rodents, cats, and monkeys), with attempts to draw comparisons with the monotreme and marsupial literature where feasible.I will not cover here the properties of cortical astrocytes, which are now recognised as key contributors to various neuronal functions, including synaptic transmission, energy metabolism, and ion homeostasis.However, investigating their diversity and morpho-functional features within the main mammalian groups represents a promising direction for future research, given the reported variations in the number or size of protoplasmic astrocyte processes between humans and rodents, as well as the specific presence of certain astrocyte types in primates (see Oberheim Bush and Nedergaard, 2017 for review). . . Upper-layer pyramidal neurons in eutherians L2/3 pyramidal neurons, a subset of IT neurons, play a pivotal role in intra-cortical information processing through their local and long-range corticocortical connexions.These neurons receive inputs from specific thalamic nuclei on their basal dendrites, either directly or via ascending projections from L4 in sensory cortices or from upper L5 in cortices lacking a distinct granular layer, while non-specific thalamocortical and distant cortical inputs mainly terminate on their tuft branches in L1.L2/3 pyramidal neurons make numerous reciprocal connexions within their home column, Frontiers in Systems Neuroscience frontiersin.orgmostly on upper basal and apical oblique dendrites.Locally, L2/3 ITs send prominent descending axonal projections to L5 pyramidal cells (Weiler et al., 2008;Lefort et al., 2009;Petreanu et al., 2009;Harris and Shepherd, 2015).This robust output to L5 has been identified as an essential feature of neocortical microcircuitry, preserved in most regions and species (Thomson and Lamy, 2007;Weiler et al., 2008;Hooks et al., 2011).Long-range axons of supragranular pyramids establish connexions with ipsi-and contralateral cortical regions and the striatum (Petreanu et al., 2007;Anderson et al., 2010;Pidoux et al., 2011a).L2/3 pyramidal neurons show substantial regional and interspecies variation in their dendritic architecture.In primates, L2/3 ITs from higher integrative frontal cortices generally display a more extensive and branched basal dendritic tree than their counterparts from primary and secondary sensory areas (Elston and Rosa, 1998;Elston et al., 2001;Jacobs et al., 2001;Gilman et al., 2017;Galakhova et al., 2022).A similar size increase in basal arborization along the caudal-rostral axis has been reported in elephants (Jacobs et al., 2011) and rodents (Benavides-Piccione et al., 2006;Elston et al., 2006), although with less pronounced regional differences (Mohan et al., 2015;Gilman et al., 2017).A cross-species comparison indicates that upper-layer neurons in frontotemporal regions of macaques and humans have a larger total dendritic length (on average 1.5-fold for macaques and 3-fold for humans) and greater branching complexity than homologous neurons in mice (Mohan et al., 2015;Gilman et al., 2017), whereas such differences are not observed in primary visual cortices (Gilman et al., 2017).When comparing mammals with large brains, the length of the basilar dendritic arborization in elephants is reported to be slightly longer in both frontal cortex (∼7%) and occipital cortex (∼3%) than in humans, despite a lower degree of branching in the former species (Jacobs et al., 2011). Traditionally considered as a homogeneous cell population, there is increasing evidence for depth-dependent differences in the morpho-functional properties of L2/3 pyramidal cells in rodents (Brecht et al., 2003;Lübke et al., 2003;Staiger et al., 2015) and for the presence of neuronal subclasses distinguished by the expansion of their apical dendritic arborization, as observed in the rat medial prefrontal cortex (van Aerde and Feldmeyer, 2015).This morphological diversity of L2/3 pyramidal neurons appears to be even more pronounced in humans, with a significant increase in the extent of the horizontal field span of the apical tree, length of the basal arborization and mean radius of the cell body with increasing distance from the pia (Figure 3A) (Deitcher et al., 2017;Berg et al., 2021).In addition, recent RNAsequencing studies have revealed the existence of two additional transcriptomically defined subtypes of pyramidal neurons in deep human L3, apparently absent in mice (Hodge et al., 2019;Berg et al., 2021). The vast majority (70-95%) of excitatory synaptic inputs target the dendritic spines of neocortical pyramidal neurons (Nieuwenhuys, 1994).The elongated dendritic trees in multimodal and association cortices often correlates with an increase in spine density, particularly evident in frontal regions in primates (Elston et al., 2001;Jacobs et al., 2001).However, this trend does not hold true for all species.For examples, pyramidal neurons in the prefrontal cortex of the marmoset have fewer branches and spines than those in the temporal lobe (Elston et al., 2001), and the density of spines in the macaque prefrontal cortex does not exceed that observed in different regions of the mouse neocortex (Gilman et al., 2017).The length of dendritic spines (∼1.3 µm combining neck and spine head) does not seem to significantly change with cortical size (Karbowski, 2014), despite a slight tendency for spines in the human temporal cortex to have longer necks (0.9 vs. 0.7 µm) and larger heads (0.6 vs. 0.4 µm 2 ) compared to those in mice (Benavides-Piccione et al., 2002).Finally, in line with the idea that a dynamic balance between excitatory and inhibitory activities is a fundamental principle of cortical circuit function in physiological conditions, the ratio between excitatory (∼70-90%) and inhibitory (∼10-30%) synapses appears globally conserved in mammals, although laminar-dependent differences within and between species can be observed (for reviews DeFelipe et al., 2002;Karbowski, 2014). . . Deep-layer pyramidal neurons in eutherians Infragranular pyramidal neurons integrate inputs form virtually all neocortical layers thanks to their elongated apical dendrite and radiating basilar arborization.They in turn significantly influence cortical and subcortical operations through local connectivity and long-range output projections.L2/3 inputs to L5 cells are distributed along the dendritic tree, mainly on tuft branches, apical oblique and basal dendrites.Afferents from L4 mostly terminate on basal dendrites, which also receive axons from primary relay thalamic nuclei.Additionally, L5 pyramidal cells receive thalamocortical projections from higher-order thalamic nuclei on their apical tuft in L1 and basal dendrites (Weiler et al., 2008;Petreanu et al., 2009;Hooks et al., 2011).L5 ITs establish many reciprocal connexions and provide synaptic inputs to L5 ETs.Connexions from ETs to ITs are less numerous (Morishima and Kawaguchi, 2006;Brown and Hestrin, 2009;Kiritani et al., 2012); ET intracortical projections mainly contributing to inter-areal communications (Nelson et al., 2013;Ueta et al., 2013;Harris and Shepherd, 2015).Most of inputs to L6 ITs in sensory cortices originate from local deep-layer neurons, while L6 ETs are primarily innervated by axons from higher-order cortical areas (Zhang and Deschênes, 1998;Mercer et al., 2005;Feldmeyer, 2012;Vélez-Fort et al., 2014). Although IT and ET neurons are distributed across L5 and L6, they display laminar-dependent projection patterns.For instance, corticostriatal neurons projecting to ipsilateral and/or contralateral striatum are found throughout L5 (Anderson et al., 2010;Pidoux et al., 2011a), while corticospinal neurons seem to be confined to the deeper part of L5 (Anderson et al., 2010;Suter et al., 2013), and corticothalamic neurons predominate in L6 (Bourassa and Deschênes, 1995).Corticothalamic neurons from sensory areas typically project back to their primary relay thalamic nuclei, but those located in the deeper part of L6 may also project to higherorder thalamic nuclei (Bourassa and Deschênes, 1995;Chevee et al., 2018). The two subclasses of infragranular pyramidal neurons differ in their apical dendritic architecture, with great variability existing within each subpopulation in all species (for review, see Baker et al., 2018a).Traditionally, L5 ET neurons possess a more complex apical dendritic arborization with numerous branches and a crownshaped tuft that unfolds close to the pial surface (Figure 3B), whereas the apical dendritic tuft of IT neurons is more restricted with fewer side branches (Hattox and Nelson, 2007;Ramaswamy and Markram, 2015;Kalmbach et al., 2021).Morphological heterogeneity is also present within L6; L6 ETs exhibit a relatively compact apical dendritic arborization predominantly terminating in narrow tufts in L4, while corticocortical L6 neurons extend an untufted or sparsely tufted apical dendrite, rarely extending beyond L4-L5 border.By contrast, the apical dendrite of the corticoclaustral IT neurons in L6 can reach the lower boundary of L1 (Katz, 1987;Zhang and Deschênes, 1997;Kumar and Ohana, 2008;Oberlaender et al., 2012;Yang et al., 2022).Consistent with inter-areal variations observed in superficial layers, infragranular pyramidal cells from higher processing areas possess a more elaborate basal dendritic arborization in the macaque (Elston and Rosa, 2000).Despite the difficulty of unambiguously identifying ET neurons in humans (based on their axonal projections), a transcriptomic cell class sharing multiple distinctive marker genes and morphological attributes with the murine ET neuron subtype (Tasic et al., 2016) has been identified in different regions of the human neocortex, despite a relative lower abundance as compared to monkeys and rodents (Hodge et al., 2019;Bakken et al., 2021;Kalmbach et al., 2021). Certain subpopulations of ET neurons, with morphological features that deviate from the archetypal pyramidal neuron, are endemic to certain cortical areas and species.For instance, the corticospinal gigantopyramidal neuron (Betz cell in primates), with its very large cell body and extensive basilar dendrites, is exclusively found in the primary motor cortex of carnivores and primates (reviewed in Jacobs et al., 2018).The same applies to the large von Economo neuron, which is characterised by its spindle-shaped soma and thick, poorly branched apical and basal dendrites.Initially described in human anterior cingulate and frontoinsular cortices, they were first considered to be specifically human and identified as particularly prone to early loss and morphological alterations in various neuropsychiatric disorders (reviewed in Butti et al., 2013).Their existence, with similar cortical distributions and in fairly comparable numbers, was subsequently demonstrated in other large-brained species such as great apes, elephants and certain cetaceans, although their presence in nonprimate mammals is still debated (Butti et al., 2013;Banovac et al., 2019). Several key features of the eutherian pyramidal cell are retained in marsupials, including their presence throughout layers 2 to 6, an upward-projecting apical dendrite, elaborate basilar dendritic arborization, and recurrent excitatory synaptic connexions.Some variations, such as the more common bifurcation of apical dendrites into daughter branches, were however observed in wallabies, quokkas, and opossums, but not in dunnarts (Walsh and Ebner, 1970;Tyler et al., 1998;Jacobs et al., 2018).Our knowledge on neuronal classes in monotremes is still limited but the few existing studies suggest that pyramidal neurons in the short-beaked echidna represent a smaller proportion (35-50%) of the total population of cortical neurons compared to therian species.In addition, a substantial number of these pyramidal neurons (30-40%) display atypical attributes such as apical dendrites lacking a terminal bouquet or branching close to the soma, and poorly developed basal dendritic skirts.Monotreme pyramidal cells also appear to have a lower density of spines on apical and/or basal dendrites.However, the morphology of the different types of non-pyramidal neurons (spiny stellate cells and inhibitory interneurons) is very similar in monotremes, marsupials, and placentals.These observations led Hassiotis and colleagues to put forward the hypothesis that pyramidal and non-pyramidal neurons may have emerged as distinct morphological entities in the first mammals, while the entire set of typical pyramidal cell features would have appeared shortly after the split with the prototherian lineage, around 180 million years ago (Hassiotis and Ashwell, 2003;Hassiotis et al., 2005). In summary, the morphology of eutherian pyramidal neurons appears more diverse than previously thought, even within a single cortical area of a given species (see, for example, Figure 3A).Furthermore, it is interesting to note that the early bifurcation of apical dendrites in marsupials and monotremes is an anatomical feature that is also commonly observed in some placental species, such as hedgehogs or elephants (Valverde and Facal-Valverde, 1986;Jacobs et al., 2011).Thus, the canonical and non-canonical aspects of pyramidal cells seem to have been relatively well preserved during cortical evolution; a better understanding of neuronal properties in clades close to the root of the mammalian phylogenetic tree will help to clarify whether morphological heterogeneity is indeed greater in these species and whether it could translate into different cortical functioning. . . GABAergic interneurons Neocortical GABAergic interneurons constitute a highly heterogeneous set of cells that differ in their morphology, input-output connectivity, intrinsic electrophysiology, and expression of molecular markers such as calcium-binding proteins and neuropeptides.Although a consensus classification of interneurons is still debated (Ratliff and Batista-Brito, 2020), available data suggest a grouping into four main subclasses based on distinct immunohistochemical profiles.These include interneurons expressing the calcium-binding protein parvalbumin (PV), the neuropeptide somatostatin (Sst), and the ionotropic serotonin receptor 5HT3a (5HT3aR), with the 5HT3aR group being further divided into two subgroups based on vasoactive intestinal peptide (VIP) expression.Each of these molecularly identified subclasses encompasses several interneuron types, primarily defined by specific morpho-functional properties (Ascoli et al., 2008;Rudy et al., 2011;Tremblay et al., 2016).The different interneurons target preferential subcellular domains on neighbouring pyramidal neurons, presumably mediating specific functions within the cortical microcircuit.For instance, within the PV-positive group, fast-spiking basket cells mostly synapse on somatic and proximal dendritic regions, while chandelier (axo-axonic) cells innervate the axonal initial segment.In addition to their local axonal arborization, Sst-expressing Martinotti interneurons project to superficial layers, where they inhibit apical tuft dendrites.The vast majority of 5HT3aR/VIP cells located in superficial cortical layers have a bipolar-like dendritic morphology and vertically extending axons that exert a translaminar inhibitory influence on the basal dendrites of pyramidal cells, although they preferentially form synapses onto other interneurons.Finally, the 5HT3aR/non-VIP neurons mainly target dendrites of local pyramids via their dense axonal plexus (neurogliaform cells) or provide a perisomatic inhibition (5HT3aR/non-VIP, cholecystokinine-positive cells) (for comprehensive reviews, see Markram et al., 2004;Tremblay et al., 2016;Lourenço et al., 2020).Recent work combining analysis of transcriptomes, intrinsic electrical properties, and morphological features of GABAergic interneurons from the mouse visual cortex, has further refined the interneuron classification system, revealing an even greater diversity of cell types (Tasic et al., 2018;Gouwens et al., 2020). Following Ramón y Cajal (1917)'s belief that "the functional superiority of the human brain is intimately bound up with the prodigious abundance and the unusual wealth of forms of the so-called neurons with short axon", neuroanatomists have long speculated about an increased diversity of inhibitory interneurons in the primate brain (Nieuwenhuys, 1994).There is now growing evidence that the main subclasses of interneurons are shared by the three mammalian lineages, with substantial similarities in their anatomical and physiological properties.For instance, key features of neurogliaform interneurons described in rodents, including genetic marker expression, morphological characteristics, firing patterns, and neuromodulation, are conserved in macaques and humans (rodent: Tamás et al., 2003;rodent and primate: Oláh et al., 2007;Povysheva et al., 2007;Poorthuis et al., 2018), despite a seemingly higher intrinsic excitability in primates (Povysheva et al., 2007;Poorthuis et al., 2018).Species vary, however, in the overall percentage and laminar distribution of interneurons, as well as in the relative proportion of different subtypes (Fairén and Regidor, 1984;Kawaguchi and Kubota, 1997;Tyler et al., 1998;Hof et al., 1999;Hassiotis and Ashwell, 2003;Zaitsev et al., 2005;Krienen et al., 2020).Immunochemical and transcriptomic studies estimate the percentage of inhibitory interneurons in the neocortex to be approximatively 15% in rodents, compared to an average of 20-30% in humans, monkeys, and cetaceans (Glezer et al., 1993;DeFelipe et al., 2002;DŽaja et al., 2014;Krienen et al., 2020;Bakken et al., 2021). Some interneurons subtypes appear specific to certain species.Originally described by Ramón y Cajal in the human neocortex, calbindin-positive double bouquet cells, characterised by long-descending bundles of highly varicose axonal collaterals (the so-called "horse-tail" neurons), are mostly found in primates and, to a lesser extent, in carnivores (Ballesteros-Yáñez et al., 2005;DeFelipe et al., 2006).Even though neurons with radially oriented dendritic and/or axonal arborizations that resemble horse-tail neurons have been described in other placental and marsupial mammals (Somogyi and Cowey, 1984;Kawaguchi, 1995;Tyler et al., 1998), they display a less laterally confined axonal plexus with fewer varicosities in these species.Finally, a recent study has identified a group of L1 interneurons in humans distinguished by a specific immunochemical profile and anatomical features, including a compact bushy axonal arborization and large rosehip-shaped axonal boutons.These "rosehip cells, " which predominantly target the apical dendrites of pyramidal cells, are well positioned to modulate distal dendritic computation (Boldog et al., 2018). . . The sensory cortical microcircuit In the late 1980s, the observation of similarities in the composition and distribution of cortical neurons between species gave rise to the idea that a common canonical microcircuit might exist in mammals (Douglas et al., 1989;Douglas and Martin, 2004).Although the multidimensional nature of connectivity schemes rules out the possibility of a single cortical circuit, comparative analysis of the data accumulated in placental mammals does suggest the presence of shared principles of organisation and function (see Silberberg et al., 2002;Douglas and Martin, 2004;Thomson and Lamy, 2007;Harris and Shepherd, 2015 for reviews).The key stages in the integration of sensory input across cortical layers are described below, drawing primarily on research in rodents and cats. Sensory inputs from primary thalamic relay nuclei are routed to the cortex via modality-specific channels.These thalamocortical projections predominantly terminate on L4, but also at the boundary between L5 and L6.For their part, axons from higherorder thalamic nuclei primarily target L1 and L5, while avoiding L4 neurons (Thomson and Lamy, 2007;Petreanu et al., 2009;Constantinople and Bruno, 2013;Harris and Shepherd, 2015).L4 is considered as the main thalamorecipient layer and the starting point of intracortical sensory processing.L4 principal neurons comprise two types of glutamatergic cells: spiny stellate and star pyramidal cells, which have broadly similar functional properties but vary in proportion between areas and species.In rodents, spiny stellate cells are abundant in the primary somatosensory cortex, but rare in the visual cortex (Peters and Kara, 1985;Feldmeyer, 2012).Conversely, these cells are prevalent in the primary visual cortex of cats, monkeys and humans, while remaining sparse and randomly distributed among pyramidal cells in the auditory cortex (Lund, 1984;Meyer et al., 1989;Smith and Populin, 2001;Nassi and Callaway, 2009).Stellate cells are also found in the dunnart (Tyler et al., 1998) and in the echidna, where they exhibit a relatively wider distribution spanning L3 and L5 (Hassiotis and Ashwell, 2003). L4 spiny neurons receive dense intracortical excitatory inputs (Schubert et al., 2003;Thomson and Lamy, 2007;Lefort et al., 2009), which likely enhance the gain and duration of thalamocortical responses to ensure a robust representation of Frontiers in Systems Neuroscience frontiersin.orgsensory information (Lien and Scanziani, 2013;Li L. Y. et al., 2013;Li Y. T. et al., 2013).In the primary visual cortex of the cat, most of the glutamatergic connexions onto L4 spiny neurons originate from other L4 neurons and L6 pyramidal cells, whereas thalamocortical afferents account for only about 6% of the total synaptic contacts in L4 (Ahmed et al., 1994).The proportion of thalamocortical inputs to spiny neurons is larger in the mouse somatosensory cortex, reaching an average of ∼16% of excitatory inputs (Benshalom and White, 1986).Thalamic projections also terminate on L4 GABAergic interneurons, resulting in powerful feedforward inhibition (Cruikshank et al., 2007) that timely regulates the firing of excitatory cells and contributes to stimulus selectivity (Miller et al., 2001;Wilent and Contreras, 2005).An interesting example of consistency in sensory input processing across phylogenetically distant species is provided by studies on the short-tailed opossum, which show striking similarities in the pattern of thalamocortical connectivity and in the whiskerevoked synaptic responses of L4 neurons with those observed in mice and rats, despite the absence of barrels in these marsupials (Dooley et al., 2015;Ramamurthy and Krubitzer, 2016).From L4, sensory signals propagate to the upper layers via the prominent axonal projections between L4 spiny neurons and L2/3 pyramidal cells (Feldmeyer et al., 2002;Lefort et al., 2009).L2/3 ITs represent the second level of intracortical processing, distributing information within and beyond their home column through local and long-range corticocortical outputs.Depending on the behavioural context, sensory signals in L2/3 are further modulated by the integration of non-sensory information from other primary and/or associative cortical areas and from the thalamus (Feldmeyer, 2012;Harris and Shepherd, 2015).In vivo recordings revealed relatively low spontaneous and evoked firing rates in superficial pyramidal neurons, at least in rodents (de Kock and Sakmann, 2008;Sakata and Harris, 2009;O'Connor et al., 2010;Carton-Leclercq et al., 2023).This sparse firing, likely resulting from a hyperpolarized membrane potential keeping neurons away from their action potential (AP) threshold (Lefort et al., 2009;Carton-Leclercq et al., 2023) and from the recruitment of local inhibitory interneurons (Helmstaedter et al., 2008;Meyer et al., 2011;Haider et al., 2013), suggests that sensory information in superficial layers are encoded through the patterned discharge of fine-scale assemblies of neurons ("sparse coding") (Harris and Mrsic-Flogel, 2013).Supporting this notion, preferential connectivity between L2/3 cells dedicated to processing of related sensory information has been observed in rodent and cat primary visual cortex (Gilbert and Wiesel, 1989;Yoshimura et al., 2005;Ko et al., 2011), and spatially constrained firing activity has been recorded in response to modality-specific sensory stimulation in the superficial layers of the mouse auditory and somatosensory cortex (Sakata and Harris, 2009;O'Connor et al., 2010). Deep-layer neurons, which in rat, cat, and primate receive major inputs from L2/3 (reviewed in Thomson and Lamy, 2007), represent the last stage of signal processing within the cortical microcircuit.Engaged in complex computations, these cells combine the results of the successive integrations within the cortical column with converging long-range thalamic and cortical inputs to finally route the output message to specific sets of cortical and subcortical structures.L5 pyramidal neurons, particularly ET cells, exhibit a depolarized membrane potential and rather high spontaneous and sensory-evoked firing frequency in vivo (de Kock and Sakmann, 2008;Sakata and Harris, 2009;O'Connor et al., 2010;Carton-Leclercq et al., 2023), consistent with the dense excitatory inputs they receive and their more variable inhibitory innervation (Schubert et al., 2001;Thomson and Lamy, 2007;Petreanu et al., 2009;Feldmeyer, 2012).This suggests that the deep layers of the neocortex may employ a coding strategy based on global variations in firing rates ('dense coding') rather than on the implementation of discrete spatio-temporal dynamics of activity as in the superficial layers (Harris and Mrsic-Flogel, 2013). Unlike L5 ETs, L6 corticothalamic cells fire at low rate in vivo, even in response to various sensory stimuli (Sirota et al., 2005;O'Connor et al., 2010;Oberlaender et al., 2012).Through their projection to the thalamus and to L4, where they innervate GABAergic interneurons strongly in cats but more moderately in rats and mice, L6 corticothalamic cells are strategically positioned to modulate thalamocortical inputs (Thomson, 2010;Pichon et al., 2012).This was demonstrated in the mouse primary visual cortex, where optogenetic stimulation of corticothalamic neurons was shown to modulate the spiking level of upper-layer neurons, via the activation of intracortical and intrathalamic inhibitory circuits (Olsen et al., 2012).Moreover, by integrating longrange inputs from higher-order cortical areas, these cells are likely involved in the contextual processing of sensory signals (Zhang and Deschênes, 1998;Feldmeyer, 2012;Vélez-Fort et al., 2014). Overall, these findings suggest that the fundamental principles governing the functioning of cortical circuits are retained in the different eutherian species studied to date.It would be important to extend this research to a larger number of species, possibly from different mammalian radiations, to assess the extent to which these principles can be generalised to all mammals.Sensory processing in neocortical circuits sharing a relatively similar hodology may be differentially modulated during behaviour.In the mouse visual cortex, visually driven sensory responses in L2/3 pyramidal cells are scaled up as mice transitioned to locomotion (Niell and Stryker, 2010;Polack et al., 2013).This state-dependent facilitation of visual responsiveness has also been observed in monkeys during attention, and even in invertebrate species such as fruit flies during walking or flight (reviewed in Maimon, 2011).Conversely, in the mouse auditory cortex, active behaviour diminishes the gain of tone-evoked spiking responses in superficial layers (Zhou et al., 2014).Similar investigation in humans have produced heterogeneous results, with some studies showing a positive effect of exercise or natural walking on sensory gain and peripheral input processing (Bullock et al., 2015;Cao and Handel, 2019), while others report no modulation of visual sensitivity during treadmill walking (Benjamin et al., 2018). Electrophysiological properties of pyramidal neurons Synaptic potentials generated in the dendrites propagate to the soma and the proximal part of the axon, where they trigger APs when the voltage threshold is reached.The transduction of synaptic inputs into AP-encoded outputs depends not only on synaptic function, but also on the intrinsic membrane properties of neurons that shape the amplitude and kinetics of synaptic events and adjust their ability to elicit firing.Previous research in rodents has shown that the morphology of dendritic trees, together with their electrical properties, strongly influences synaptic integration, excitability, and firing patterns of pyramidal neurons (Mainen and Sejnowski, 1996;Yuste and Tank, 1996;Bekkers and Häusser, 2007;van Elburg and van Ooyen, 2010).This raises the question of whether such morpho-functional interactions are applicable to other species or whether the distinctive anatomical attributes of human neurons confer specific electrophysiological properties.Comparative analysis of the electrical properties of pyramidal neurons between species, and even within the same species, based on in vitro studies carried out by different laboratories can prove difficult due to the many experimental factors (e.g., the animal age, composition of intracellular and extracellular solutions, recording temperature or degree of damage to the dendritic tree during slices preparation) that vary from one study to another and are known to affect the synaptic and biophysical properties of neuronal membranes (Hardingham and Larkman, 1998;Zhu, 2000;Bekkers and Häusser, 2007).The task becomes even more complex when considering in vivo studies, as the background synaptic activity that characterises intact brain preparations is known to have a significant impact on the excitability and firing of cortical neurons (Destexhe et al., 2003;Altwegg-Boussac et al., 2014;Carton-Leclercq et al., 2021).The increase in the number of data sets and the integration of work combining experiments carried out on several species in a single study have nevertheless enabled a number of electrophysiological characteristics to be compared reliably. . Intrinsic membrane properties . . L / pyramidal neurons The intrinsic excitability of a neuron, which reflects its endogenous capacity to generate APs in response to a given input, is determined by its passive membrane properties, largely conferred by its physical structure, and by its array of non-synaptic voltagegated ion channels, which mediate a variety of active currents. According to the cable theory, the passive properties of the dendritic tree are predicted to impose a distortion on synaptic inputs, such that the most distal inputs will yield the most attenuated and prolonged excitatory postsynaptic potentials (EPSPs) at the soma (Rall, 1977;Spruston et al., 1994).Simultaneous somatic and apical dendritic whole-cell recordings from human and rodent L2/3 pyramidal neurons, together with morphologically realistic biophysical modelling, confirmed that somatic EPSP amplitude progressively decreased with increasing distance from the EPSP generation site (Mohan et al., 2015;Eyal et al., 2018;Gooch et al., 2022), and demonstrated that voltage attenuation followed a similar distance-dependence pattern in both species (Figure 4A) (Gooch et al., 2022).The extended and elaborate apical dendritic tree of human pyramidal cells is thus expected to result in strong attenuation of the most distal synaptic inputs.It has been proposed that the cable filtering of human dendrites may be partially compensated for by a lower specific membrane capacitance (∼0.5 µF/cm 2 vs. ∼1 µF/cm 2 in rodent neurons), a biophysical characteristic expected to increase the amplitude and reduce the peak delay of dendritic EPSPs at the soma (Eyal et al., 2016).However, this finding has not been confirmed by recent in vitro recordings of L2/3 (Gooch et al., 2022) and L5 (Beaulieu-Laroche et al., 2018) pyramidal cells, suggesting possible heterogeneity in the capacitive membrane properties of human neurons. In a passive neuronal model, distal EPSPs are expected to exhibit greater somatic temporal summation due to their greater half-width at the soma compared to more proximal EPSPs.However, trains of EPSPs generated at dendritic and somatic sites have been shown to summate similarly in human L2/3 pyramidal neurons (Gooch et al., 2022).This could result from a stronger dendritic expression of hyperpolarization-activated cyclic nucleotide-gated (HCN) channels mediating the h-current (I h ), which is known to reduce the duration of distally generated synaptic events in rat L5 neurons (Williams and Stuart, 2000).Although the spatial distribution of h-channels in L2/3 human neurons remains so far unknown, their presence was evidenced by a pronounced voltage sag (a slow partial repolarization) in response to long hyperpolarizing current pulses delivered at the soma (Deitcher et al., 2017;Kalmbach et al., 2018;Berg et al., 2021;Moradi Chameh et al., 2021), and further supported by in-situ hybridisation and single nucleus RNA sequencing studies (Zeng et al., 2012;Kalmbach et al., 2018).Sag ratio values (quantifying the amount of I h ) in human supragranular pyramids (11-17%) (Deitcher et al., 2017;Kalmbach et al., 2018) are comparable to those reported in macaques (12-15%) (Zaitsev et al., 2012;González-Burgos et al., 2019), but contrast sharply with their low levels (0.3-2.5%) in homologous neurons in rodents (Figure 4C) (Mason and Larkman, 1990;Larkum et al., 2007;Kalmbach et al., 2018;Berg et al., 2021; but see the relatively large sag ratio reported by van Aerde and Feldmeyer, 2015 in a subpopulation of rat L3 pyramidal cells and by Testa-Silva et al., 2022 in some L2/3 mouse neurons). The influence of distal dendritic inputs on AP output in human L2/3 neurons is therefore limited by the expansion of their apical dendritic tree and the impact of I h on the time course of synaptic potentials.However, recent studies indicate that, similar to rodent neurons (Larkum et al., 1999(Larkum et al., , 2007;;Schiller et al., 2000), human dendrites can initiate local dendritic spikes mediated by voltage-sensitive Na + or Ca 2+ channels, or dependent on Nmethyl-D-aspartate (NMDA) receptors, which can increase EPSP amplitude and drive firing at the axon initiation site (Figure 4B) (Gidon et al., 2020;Gooch et al., 2022;Testa-Silva et al., 2022).This suggests that active dendritic electrogenesis could provide a means to compensate for the normalisation of temporal summation and strong voltage attenuation of EPSPs in these neurons.The ability of human neurons to generate NMDA spikes in basal and oblique dendrites was, however, found to be significantly lower than in mouse neurons, probably due to the larger diameter of human dendrites (Testa-Silva et al., 2022).Other work on human neurons has reported briefer apical dendritic Ca 2+ spikes compared with rodents (Gidon et al., 2020), while a recent study suggested conversely that suprathreshold dendritic integrative operations were conserved between the two species (Gooch et al., 2022).These observations indicate that further research is required to gain a comprehensive picture of how the active properties of human dendrites differ from those of other species.The ability of EPSPs to initiate firing also depends on the excitability of the somatic membrane.Somatic excitability is generally assessed by measuring the current firing threshold, which is the minimum current of fixed duration required to trigger an AP (an approximation of the rheobase; Lapicque, 1926).This threshold current depends on a combination of membrane parameters, including the resting membrane potential, input resistance (which governs the capacity of a current to modify the membrane potential according to Ohm's law), and the value of the AP voltage threshold.In rodents, removal or occlusion of the dendritic tree increases the intrinsic excitability of individual neurons by increasing input resistance and lowering the AP threshold (Bekkers and Häusser, 2007).Such interactions between dendritic morphology and neuronal excitability seem directly transposable to other species, since pyramidal neurons in the superficial layers of the frontal cortex of the rhesus monkey, which have an elaborate dendritic tree, show reduced somatic excitability compared with their less branched counterparts in the primary visual cortex.Likewise, morphologically related pyramidal neurons, such as those in the visual cortex of mice and monkeys, show similar threshold current values (Amatrudo et al., 2012;Gilman et al., 2017).Finally, a direct comparison of the excitability of human and mouse L2/3 neurons in the same study revealed that human pyramids had a significantly lower input resistance and required the injection of a greater amount of current to be brought to spike (Figure 4D) (Kalmbach et al., 2018;Testa-Silva et al., 2022). . . L pyramidal neurons Important differences in subthreshold and suprathreshold intrinsic membrane properties distinguish ET from IT neurons in rodents (Figures 5A, B).Across areas, ET neurons typically exhibit a slightly more depolarized resting membrane potential, lower input resistance, faster membrane time constant and higher threshold current, and frequently (but not always) display reduced inputoutput gain (assessed by the slope of the relationship between injected current intensity and firing frequency) (Figure 5A) (Mason and Larkman, 1990;Hattox and Nelson, 2007;Dembrow et al., 2010;Oswald et al., 2013;Suter et al., 2013;Rock and Apicella, 2015;Anastasiades et al., 2018;Baker et al., 2018b;Kalmbach et al., 2021;but see Groh et al., 2010 for regional variations in biophysical parameters differentiating IT from ET neurons).These properties are largely conserved in human and non-human primates, where ET-like cells are found to be less excitable than their slender neighbours, both at rest and during integration of inputs of varying intensity (Figure 5A) (Beaulieu-Laroche et al., 2018, 2021;Bakken et al., 2021;Kalmbach et al., 2021;Moradi Chameh et al., 2021).Across species, the excitability of L5 ETs generally correlates negatively with their size, such that thicktufted pyramidal neurons of the Etruscan shrew are more easily driven to fire APs than their counterparts in rodents, macaques and humans (Beaulieu-Laroche et al., 2018, 2021;Bakken et al., 2021;Kalmbach et al., 2021) (Figure 5C).However, variations between ET and IT projection subtypes in a given species often appear to be greater than differences between species, at least for temporal cortex neurons (Kalmbach et al., 2021).Consistent with findings in other placental mammals, ET cells in humans have a more prominent voltage sag (∼25-30% sag ratio) than IT neurons (∼15%), associated with a progressive increase in h-channel density with increasing distance from the soma (Figure 5D) (Kole et al., 2006;Beaulieu-Laroche et al., 2018, 2021;Kalmbach et al., 2021;Moradi Chameh et al., 2021).The differences in the amount of I h between the two subtypes of L5 neurons are expected to contribute to their specific passive membrane properties and to shape their subthreshold integrative properties, like in rodents where h-channels impart ET neurons with a less pronounced temporal summation of EPSPs than in IT cells (Dembrow et al., 2010;Sheets et al., 2011;Kalmbach et al., 2015;Anastasiades et al., 2018).When activated by sustained depolarization, primate and rodent L5 ET neurons show less frequency adaptation (progressive increase in the inter-spike interval during a train of APs) than IT cells (Figure 5A).Individual APs in ETs also exhibit faster kinetics, shorter duration and a lower voltage threshold (Figures 5A, B) (rodent and primate: Bakken et al., 2021;Beaulieu-Laroche et al., 2021;Kalmbach et al., 2021;rodent: Mason and Larkman, 1990;Hattox and Nelson, 2007;Dembrow et al., 2010;Oswald et al., 2013;Suter et al., 2013;Pathak et al., 2016).The thinner APs of L5 corticofugal neurons could limit calcium influx and activation of calcium-dependent potassium channels responsible for spike frequency adaptation during AP trains (Faber and Sah, 2003), and thus contribute to their more regular firing pattern (Suter et al., 2013). Compared to L5, the intrinsic parameters that discriminate ET and IT populations in L6 are more difficult to identify.While some studies describe L6 ET neurons as having faster time constants, shorter APs and higher rheobase than IT cells as in L5 (Kumar and Ohana, 2008;Crandall et al., 2017), others report lower threshold currents, longer APs (Kinnischtzke et al., 2016), as well as higher input resistance values (Kinnischtzke et al., 2016;Crandall et al., 2017;Yang et al., 2022).However, weaker steady-state adaptation ratios and relatively larger voltage sags are biophysical properties that also consistently separate ETs from ITs in L6 (Kumar and Ohana, 2008;Vélez-Fort et al., 2014;Crandall et al., 2017;Yang et al., 2022; but see Cotel et al., 2018). A significant proportion (∼10-30%) of L5 ET neurons in cats and rats can fire in bursts (clusters of APs with short interspike intervals riding on a depolarizing envelope) in response to threshold current injections (rodent: Chagnac-Amitai et al., 1990;Mason and Larkman, 1990;Dégenètais et al., 2002;Paz et al., 2009;Groh et al., 2010;cat: Baranyi et al., 1993;Nuñez et al., 1993;Nowak et al., 2003).This firing behaviour is nonetheless not unique to ETs as burst firing can be observed in vivo and in vitro in L5 corticostriatal neurons in rodents (Yang et al., 1996;Pidoux et al., 2011b), and to a lesser extent, in the upper layers of sensorimotor and visual cortices in cats (Nishimura et al., 2001;Nowak et al., 2003).The traditional classification of L5 pyramidal cells into regular spiking or intrinsically bursting subtypes is further complicated by the fact that firing dynamics evolve in vivo with the level of background synaptic activity, which is function of vigilance state and behaviour (Mahon et al., 2001;Steriade, 2004;Altwegg-Boussac et al., 2014).Across species, in vivo recordings generally describe L5 ETs as more spontaneously active than ITs and more prone to generate spike bursts (monkey: Pasquereau and Turner, 2011; rat: Ushimaru and Kawaguchi, 2015;Rojas-Piloni et al., 2017;Saiki et al., 2018), in stark contrast to the sparse spiking activity of L6 corticothalamic neurons (rabbit: Swadlow, 1989;cat: Sirota et al., 2005;rat: Oberlaender et al., 2012).Burst firing is suggested to be less frequent in human and non-human primates than in rodents and carnivores (Avoli and Olivier, 1989;Foehring et al., 1991;Chang and Luebke, 2007;Beaulieu-Laroche et al., 2018, 2021;Moradi Chameh et al., 2021;Piette et al., 2021).This reduced bursting capacity of human neurons could result from an increased electrical isolation of distal dendrites due to their length (Beaulieu-Laroche et al., 2018, 2021), which is expected to limit electrical coupling between the different neuronal compartments and, thus, the influence of dendritic spikes on somatic output (Larkum et al., 1999(Larkum et al., , 2001;;Williams and Stuart, 1999).However, such a decrease in calcium dendritic electrogenesis in humans was not reported by another study (Kalmbach et al., 2021), suggesting variability in the ability of human dendrites to trigger regenerative potentials, as in rodents (Helmchen et al., 1999;Fletcher and Williams, 2019).As postsynaptic bursts are thought to influence synaptic integration, plasticity, and information signalling in cortical circuits (Lisman, 1997;Larkum et al., 1999;Williams and Stuart, 1999), the lower intrinsic burst capacity of human neurons could appear as disadvantageous to cortical computations.Alternatively, as suggested by theoretical studies, electrically isolated individual dendrites could function as separate processing units, increasing the possibilities for independent parallel operations and thus, the computational capabilities of cortical neurons (Häusser and Mel, 2003;Eyal et al., 2018). ET and IT pyramidal neurons also differ in their responses to neuromodulatory transmitters (for reviews, see Shepherd, 2013;Dembrow and Johnston, 2014;Radnikow and Feldmeyer, 2018).For instance, cholinergic inputs are more effective at promoting persistent spontaneous firing in deep-layer corticofugal neurons than in corticocortical cells (Dembrow et al., 2010;Joshi et al., 2016;Baker et al., 2018b), an effect likely involving inhibition of M-current mediating K + channels and activation of nonspecific cationic conductances (Baker et al., 2018b).Similarly, the increase in cortical excitability induced by activation of alpha2-noradrenergic receptors is significantly reduced in ITs compared with ETs (Dembrow et al., 2010).The responsiveness of pyramidal neurons to dopaminergic and serotoninergic afferents also varies according to their projection targets, owing to the differential expression of neuromodulator receptor subtypes (Avesar and Gulledge, 2012;Gee et al., 2012).Given the critical roles of neuromodulatory systems in regulating sleep, wakefulness, motor control, and various high cognitive functions such as conscious perception, attention, learning, and memory, a comprehensive knowledge of the neuromodulation of projection neuron subpopulations in humans will be essential for a cell-typespecific mechanistic understanding of numerous brain functions and dysfunctions, as well as for the development of potential treatments (Shepherd, 2013;Ma et al., 2018). . Excitatory synaptic function Recent cutting-edge studies, combining whole-cell recordings of pairs of L2/3 pyramidal neurons with the identification of putative synaptic contacts, have begun to characterise excitatory synaptic transmission in the supragranular layers of the human neocortex.Despite potential differences in the capacity to form synaptic connexions, given the higher density of spines on human neurons (Benavides-Piccione et al., 2002), the rate of local functional connectivity (around 12% in mice, 14% in rats, and 14% in humans) and the number of synapses per connexion (between ∼2 and 4) are reported to be comparable between rodents and humans.This similarity extends to the median amplitude (∼0.5 mV despite substantial differences between studies and cortical regions, see de Kock and Feldmeyer, 2023 for review), latency, and kinetics of unitary (evoked by the firing of single presynaptic neurons) EPSPs reaching the soma, which show remarkable consistency across a variety of species (human and mouse: Testa-Silva et al., 2014;Szegedi et al., 2016;Seeman et al., 2018;Campagnola et al., 2022;Hunt et al., 2023;mouse: Oswald and Reyes, 2008;Lefort et al., 2009;Ko et al., 2013;Jouhanneau et al., 2015;Luo et al., 2017;rat: Mason et al., 1991;Hardingham and Larkman, 1998;Holmgren et al., 2003;Yoshimura et al., 2005;Feldmeyer et al., 2006;Hardingham et al., 2010;macaque and rat: Povysheva et al., 2006;cat and rat: Thomson et al., 2002).It also predicted that a similarly small number of simultaneously active L2/3-L2/3 synapses (between ∼125 and 145) is required to generate a somatic AP in human and rat cell models (Eyal et al., 2018).This maintenance of the fundamental properties of excitatory synaptic transmission between L2/3 human neurons, despite a strong voltage attenuation of dendritic EPSPs and lower somatic input resistance, could result from compensatory mechanisms, such as initiation of dendritic spikes (see above) and/or an increase in AMPA and NMDA synaptic conductances, as predicted by modelling studies (Eyal et al., 2018;Hunt et al., 2023).L2/3 pyramidal-to-pyramidal synaptic connexions in humans are, however, considered to be more reliable than in rodents, with a failure rate between functionally connected cells ranging from 0 to 8% compared with 3.2 to 25% in mice and rats (human: Szegedi et al., 2016;human and mouse: Hunt et al., 2023;mouse: Oswald and Reyes, 2008;Jouhanneau et al., 2015;rat: Hardingham and Larkman, 1998;Koester and Johnston, 2005;Feldmeyer et al., 2006;Hardingham et al., 2010). Connexions between pyramidal cells and GABAergic interneurons generally produce larger synaptic responses than connexions between pyramidal cells, partly because of the higher input resistance of interneurons (rat and cat: Thomson et al., 2002;rat and macaque: Povysheva et al., 2006;rat: Holmgren et al., 2003;human: Molnár et al., 2008).The amplitude distribution of unitary EPSPs at this synapse often shows a long tail formed by large amplitude synaptic events, although this can also be observed at other types of connexions (Holmgren et al., 2003;Lefort et al., 2009).Interestingly, dual whole-cell patch clamp recordings in the human neocortex have identified a subset of supragranular pyramidal cells establishing particularly strong synapses on fast-spiking interneurons and producing excitatory events large enough to drive postsynaptic firing and initiate a complex sequence of polysynaptic activity in the local network (Molnár et al., 2008;Szegedi et al., 2016).The properties of human inhibitory synapses and their impact on local circuit function are still not fully established.Nevertheless, the consistency of longlasting inhibitory responses elicited by neurogliaform interneurons in rodent and human L2/3 pyramidal neurons (Tamás et al., 2003;Oláh et al., 2007), along with the comparable facilitatory effect of acetylcholine on Martinotti cell-mediated lateral inhibition in both species (Obermayer et al., 2018), suggest that these interneuron subtypes may similarly modulate synaptic input processing at distal dendrites in humans. Synaptic function in human L5 is just beginning to be uncovered.The estimated rate of functional connectivity between L5 pyramidal neurons is similarly low (∼10%) in humans and rodents (rat: Markram et al., 1997;Lefort et al., 2009;human: Seeman et al., 2018;Campagnola et al., 2022).Quantitative analysis of the structural correlates of synaptic transmission indicates that most L5 synaptic boutons in human temporal cortex present a single active zone comparable in size to those in L4 and L5B of the rat somatosensory cortex, despite substantial individual variability.The number of synaptic vesicles in the putative readily releasable pool of L5 neurons is similar in both species, but the resting pool is at least 2-fold larger in humans, which could facilitate rapid refilling of the releasable pool during sustained high-frequency activity (human: Yakoubi et al., 2019;rat: Rollenhagen et al., 2015rat: Rollenhagen et al., , 2018)).Finally, it is suggested that the mean amplitude of the unitary EPSP at the L5 pyramidal-to-pyramidal connexion (including ET and IT neurons) is comparable between species (∼0.6 mV), but further human studies are needed to confirm these results (rodent: de Kock and Feldmeyer, 2023 for review; see also Lefort et al., 2009;Hardingham et al., 2010;Kerr et al., 2013;human: Seeman et al., 2018). . . Plasticity of synaptic transmission Connexions between L2/3 pyramidal cells generally exhibit frequency-dependent short-term synaptic depression in vitro.This rapid decrease in the amplitude of the postsynaptic response during trains of presynaptic spikes, primarily mediated by presynaptic mechanisms involving a progressive inactivation of Ca 2+ channels in synaptic terminals and reduction of the pool of readily releasable synaptic vesicles (Zucker and Regehr, 2002;Nanou and Catterall, 2018), can be similarly observed in humans, cats, and rodents (mouse and human: Testa-Silva et al., 2014;Campagnola et al., 2022;Hunt et al., 2023;mouse: Oswald and Reyes, 2008;Luo et al., 2017;but see Jouhanneau et al., 2015 for an overall stability of excitatory synaptic responses in vivo; rat: Holmgren et al., 2003;Koester and Johnston, 2005;Feldmeyer et al., 2006;cat: reviewed in Thomson and Lamy, 2007).However, human L2/3 excitatory connexions have been shown to recover faster from short-term depression than synapses in rodents, a property that is expected to increase information transfer within the cortical circuit (Testa-Silva et al., 2014;Campagnola et al., 2022;Hunt et al., 2023). The timing-based learning rules governing the sign of longterm synaptic plasticity in L2/3 pyramidal cells diverge between humans and rodents.In juvenile rodents, pairing protocols in which an extracellularly-evoked EPSP precedes the current-evoked postsynaptic firing by a short time interval (pre-post intervals between 0 and +20 ms) typically induce long-term potentiation (LTP), whereas long-term depression (LTD) is triggered when the postsynaptic activity precedes the presynaptic stimulation for a broader range of time differences (pre-post intervals between 0 and −40 ms) (Feldman, 2000;Froemke and Dan, 2002).In contrast, LTP at adult human synapses is induced with a wider temporal window extending approximately from −100 to +5 ms, and LTD for pre-post intervals between +10 and +20 ms (Verhoog et al., 2013).Differences in developmental stage (i.e., juvenile vs. adult) may partly explain this disparity, as the ability of excitatory synaptic connexions to produce LTD has been shown to decrease with age in rodents (Banerjee et al., 2009;Verhoog et al., 2013).Beyond these differences in timing rules, the layer-specific cholinergic regulation of long-term synaptic plasticity in the rodent neocortex seems to be retained in humans (Obermayer et al., 2017 for review).Importantly, the variable firing patterns of cortical neurons in vivo (Shadlen and Newsome, 1998;O'Connor et al., 2010) are expected to give rise to multiple and complex spike-timing interactions between pre-and post-synaptic neurons, further complicating the rules for plasticity induction (Sjöström et al., 2001). Temporal organisation of cortical activity In an intact brain, individual neurons defined by their specific intrinsic membrane properties interact synaptically to give rise to the ongoing in vivo dynamics of cortical circuits (Llinás, 1988).Depending on the behavioural context, mammalian neurons can form cell assemblies of different sizes capable of generating oscillatory activities at various frequencies (Steriade, 2000;Buzsáki and Draguhn, 2004).These self-organised patterns of activity play a crucial role in determining the global functional state of the brain and contribute to a range of executive and cognitive functions, including selection of input signals, sensorimotor integration, regulation of information transfer across distributed networks, motor action, and contextual formation and storage of subjective percepts (Engel et al., 2001;Salenius and Hari, 2003;Buzsáki and Draguhn, 2004;Fries, 2015). By carefully placing electroencephalographic (EEG) electrodes on the skulls of rabbits and dogs, Caton (1875) was the first to discover, in 1875, that a living brain constantly generates fluctuating electrical activity, even in the absence of sensory stimulation or motor activity.A few decades later, the German psychiatrist Berger (1929), secretly conducting human EEG recordings at night in the basement of his university, understood that this ongoing cerebral activity was temporally structured in the form of oscillations whose amplitude and frequency depended on the behavioural or mental state, and introduced the idea that a given cerebral state could be defined by a specific profile of EEG activity.Since the seminal works of Caton and Berger, EEG recordings have been widely used in a variety of mammals to characterise brain dynamics during the sleep-wake cycle and their disruption in neurological disorders (Uhlhaas and Singer, 2006). . Physiological rhythms of sleep and wakefulness . . Slow-wave sleep All mammals examined so far exhibit some form of sleep, whose mean daily duration varies from 2 h in the African bush elephant to 20 h in the little brown bat (Siegel, 2022).Highamplitude, low-frequency (Blake and Gerard, 1937;Jouvet, 1967).Extracellular and intracellular recordings in cats and rodents have shown that EEG slow waves result from the rhythmic alternation between prolonged periods of membrane depolarization and firing, and episodes of silence and hyperpolarization in large ensembles of cortical neurons (Steriade et al., 1993a;Sanchez-Vives and McCormick, 2000;Mahon et al., 2001Mahon et al., , 2006;;Chauvette et al., 2011).This slow cortical rhythm and associated behavioural correlates are largely conserved in placental mammals (Figure 6A).However, while slow oscillations typically occur bilaterally in terrestrial mammals, cetaceans (whales and dolphins) spend 70-90% of their sleep time in unihemispheric SWS; a unique cortical state where one cerebral hemisphere exhibits slow-wave EEG, while the other shows wake-like EEG activity (Figure 6B) (Lesku et al., 2006;Lyamin et al., 2008).During unihemispheric SWS, the eye contralateral to the "asleep" hemisphere is usually closed and the eye contralateral to the "awake" hemisphere remains open (Lyamin et al., 2004).Cetaceans engaged in unihemispheric SWS can rest on the bottom, float at the surface or swim slowly in a single direction, with periodic fin movements aiding in posture stabilisation.It has been suggested that the unusual phenomenology of cetacean sleep may serve to maintain a certain level of motor activity, enabling them to come to the surface to breathe, to remain alert to potential threats, or for thermogenesis (Lyamin et al., 2008).Semi-aquatic fur seals represent a remarkable variant of this adaptation to environmental conditions, as they display large amounts of bilateral high-voltage slow activity when sleeping on land and large amounts of interhemispheric EEG asymmetry when sleeping in water (Figure 6B) (Lyamin et al., 2018).The neural mechanisms behind the lateralized nature of SWS in cetaceans remain a mystery.While the reduced interhemispheric synchronisation of SW activity in mammals with callosotomy or callosal agenesis (Singer and Creutzfeldt, 1969;Nielsen et al., 1994;Mohajerani et al., 2010) suggests that the relatively small corpus callosum of cetaceans may be involved, such callosal reduction has not been found in fur seals (Lyamin et al., 2008). Polygraphic correlates of SWS in marsupials closely resemble those of terrestrial placental mammals (Van Twyver and Allison, 1970;Zaid et al., 2022), despite variation in total sleep duration or SWS proportion between species (Siegel, 2005).Monotremes also show similar periods of quiet sleep with high-voltage EEG and reduced motor activity, for 6 to 8 h per day (Figure 6A) (Siegel et al., 1999;Siegel, 2022). . . Paradoxical sleep Episodes of rapid-eye-movement (REM) or paradoxical sleep usually follow periods of SWS.In marsupials and placentals, the slow, and globally synchronised EEG waves of the SWS are here replaced by rapid, low-voltage oscillations in the beta/gamma range (typically at 20-40 Hz) (Figure 6A) (Llinás and Ribary, 1993;Steriade et al., 1996Steriade et al., , 2001;;Zaid et al., 2022).These high-frequency cortical activities are accompanied by a regular hippocampal theta rhythm observable in the overlying neocortex, with variations in the duration and properties of theta periods across species (Robinson et al., 1977;Cantero et al., 2003).This sleep state is particularly intriguing because it combines reduced perceptiveness of the external world and almost complete muscular atonia with a high level of cerebral activity resembling that of wakefulness, irregular cardiac and respiratory rhythms and frequent phasic movements of the eyes and extremities (ears, vibrissae, fingers, beak, tail) (Aserinsky and Kleitman, 1953;Dement, 1958;Jouvet, 1967).Remarkably, this is also the period when our dreams are most vivid and take on a fantastic character.The sparse and often inconsistent reports of mammalian-like REM sleep in fish, amphibians, and reptiles suggests that this sleep state may have appeared with endothermy in birds and mammals (Lesku et al., 2006;Hartse, 2011).Sometimes described as a state of "heightened inward arousal, " one of the proposed functions of REM sleep could be to prepare the brain to wake up in a more alert state, thereby avoiding the poor sensorymotor function and high vulnerability associated with waking up during SWS, which would be detrimental for survival (Snyder, 1966). Monotremes appear to have an unusual form of REM sleep (Siegel et al., 1998).In platypuses, sleep phases characterised by rapid eye movements and intermittent twitching of the beak and head are associated with moderate-or high-voltage EEG waves (Figure 6A) (Siegel et al., 1999).Such an EEG profile, associated with an irregular firing of brainstem neurons typical of REM sleep in placentals and marsupials, has also been observed in the short-beaked echidna (Siegel et al., 1996(Siegel et al., , 1998)).However, these findings need to be confirmed, since another study conducted in the echidna has reported periods of REM sleep with EEG activities close to those described in therian mammals (Nicol et al., 2000).The daily duration of the REM sleep-like state in monotremes (∼7.5 h/day; the platypus is reputed to have more REM sleep than any other mammal) exceeds that of marsupials (∼4.4 h/day) and placentals (∼2 h/day) (Siegel, 2022).Given that monotremes have a lower body and brain temperature (∼31 • C) as compared to other mammals (∼35.5 • C in marsupials and 37 • C in placentals), it has been hypothesised that prolonged periods of REM sleep, which are associated with a warming of various brain regions (Wehr, 1992), may contribute to the maintenance of thermal conditions compatible with autonomic function and alert awakening (Siegel, 2022).It has also been suggested that the EEG characteristics of therian SWS and REM sleep may have evolved from a single, phylogenetically older sleep stage combining features of both states, and that cortical activation (fast and low-voltage cortical activity) during REM sleep may be a more recent acquisition (Siegel et al., 1998;Siegel, 2022). To date, there is no unequivocal proof of the existence of REM sleep in cetaceans.Behavioural signs typical of REM sleep in terrestrial mammals (muscle twitches, erections in males, eyelid movements) have been documented in several cetacean species but do not appear to be specific to this sleep state.However, it is possible that REM sleep episodes in cetaceans are too short to be reliably recorded, or that REM sleep is present in these species in a modified form (Lyamin et al., 2008).Again, it is interesting to note that the average daily amount of REM sleep in fur seals is much lower when they sleep in the water than when they sleep on land (Lyamin et al., 2018). . . Wakefulness Cortical activity during wakefulness continuously varies according to the level of attention, the behavioural context or overall motor activity.As first demonstrated by Hans Berger and Edgar Adrian, alpha-frequency oscillations (7-12 Hz) appear on the EEG during quiet wakefulness (especially when eyes are closed) and predominate in posterior brain regions in primates and carnivores (Berger, 1929;Adrian and Matthews, 1934;Lopes da Silva et al., 1973;Bollimunta et al., 2008).Not all human beings are equally endowed with an occipital alpha rhythm and a small proportion of individuals even show an EEG almost entirely devoid of alpha waves after childhood (Niedermeyer, 1997).A second form of alpha oscillations, occurring at 7-20 Hz (also known as alpha-type mu rhythm), is observed in the frontal and parietal regions of rodents and cats, as well as in the central regions of humans, during periods of motor idleness preceding the completion of a sensory-motor task (Gastaut, 1952;Rougeul et al., 1972;Buzsaki et al., 1988;Pfurtscheller et al., 1996). The transition from quiet to active wakefulness (i.e., associated with motor activity, attentional or behavioural processes) is manifested by a clear change in the EEG pattern in most cortical areas.This is characterised by a decrease in the proportion of cortical waves below 10 Hz, replaced by activities in the beta and gamma frequency bands (mainly in the 20-60 Hz range).These fast rhythms reflect changes in the ongoing subthreshold synaptic activity of cortical pyramidal neurons, which undergo a slight depolarization and shift from relatively slow and large oscillations to fast and small-amplitude synaptic potentials.The impact of membrane depolarization on cell firing varies by layer and cortical region, with pyramidal neurons in deeper layers showing a more pronounced increase in firing rate (reviewed in Poulet and Crochet, 2019).High-frequency activities are a hallmark of active waking in placentals (Steriade, 2003;Buzsáki et al., 2013), marsupials (Van Twyver andAllison, 1970;Zaid et al., 2022), and monotremes (Figure 6A) (Siegel et al., 1996(Siegel et al., , 1999)), although the latter two groups have been less extensively studied.Cortical activity during waking is regulated by several groups of neurons that release neuromodulatory substances, including acetylcholine, noradrenaline, histamine, and serotonin, among others.These neurons are active during wakefulness, while their firing is reduced or suppressed during SWS (Lin, 2000;Jones, 2005).Notably, ascending cholinergic inputs from the basal forebrain and pontine tegmental nuclei play an prominent role in promoting and maintaining arousal through direct effects on the excitability and activity of cortical cells, and indirectly, by activating thalamocortical glutamatergic neurons (Metherate et al., 1992;Steriade et al., 1993a,b;Pinto et al., 2013;Eggermann et al., 2014).High-frequency oscillations during sensory processing arise from small neural networks of interconnected inhibitory and pyramidal cells (Buzsáki and Wang, 2012), but locally generated gamma activities can transiently synchronise over long distances during periods of heightened attention or conscious perception in primates (Melloni et al., 2007;Gregoriou et al., 2009).One well-documented mechanism for preserving the frequency of physiological rhythms and their inter-areal coherence in brains of different sizes is the scaling of the diameter of a fraction of myelinated axons.This scaling is expected to accelerate the propagation of APs in large brains, ensuring that homologous brain regions in different species can communicate within roughly the same time frame (reviewed in Buzsáki et al., 2013). . Pathological oscillations in epileptic and dying brains If coordinated spatio-temporal activity patterns are the basis of normal brain functioning, their disruption is expected to lead to functional abnormalities.Indeed, Parkinson's disease, schizophrenia, epilepsy and certain forms of coma are all examples of pathologies whose clinical symptoms are associated with disturbances in cortical oscillatory activities (reviewed in Uhlhaas and Singer, 2006).Although there is substantial literature on a variety of brain disorders, I will focus here on two examples of abnormal neural synchronisation: spike-wave discharges in absence epilepsy and the altered cortical patterns observed in the dying brain. . . Absence epilepsy Childhood absence epilepsy, an epileptic syndrome characterised by recurrent generalised non-convulsive seizures, provides an interesting example of disturbed brain dynamics identifiable in several mammals, with fairly similar behavioural consequences.Typical absence seizures involve brief episodes of impaired consciousness, manifested by a sudden interruption of voluntary sensory-motor behaviour, staring, unresponsiveness and retrograde amnesia of the epileptic episode.Secondary clinical symptoms, such as automatisms or mild clonic movements of the eyes, eyelids, mouth or limbs, are also commonly observed.Occurring frequently, up to hundreds of times a day, these seizures can interfere with the normal cognitive and psychosocial development of children (Blumenfeld, 2005a;Cavanna and Monaco, 2009;Crunelli et al., 2020).Absence seizures result from abnormally synchronised oscillations in corticothalamocortical networks, evident as high-amplitude spike-waves at 3-4 Hz on human EEG (Depaulis and Charpier, 2018;Crunelli et al., 2020).Numerous investigations of absence seizure patient populations using high-density EEG recordings, MEG and functional MRI have demonstrated that epileptic oscillations are initiated in a circumscribed region of the neocortex before their generalisation (Holmes et al., 2004;Westmijse et al., 2009;Bai et al., 2010).The severity of clinical symptoms has been shown to correlate with the duration, intensity and inter-regional coherence of spike-wave discharges (Blumenfeld, 2005b;Guo et al., 2016). Spike-waves are observed not only in humans, but also in monkeys (Steriade, 1974), cats (Gloor and Fariello, 1988), and ferrets (Youngblood et al., 2015), where they have a similar frequency.The cortically-initiated spike-waves in rodents exhibit a relatively faster frequency of 5-10 Hz, despite similar behavioural correlates (altered sensory responsiveness, behavioural arrest, vibrissae or jaw muscle twitching, occasional chewing) and pharmacological sensitivity to anti-absence medications (Polack et al., 2007(Polack et al., , 2009;;Depaulis et al., 2016;Jarre et al., 2017).The reason for the higher frequency of spike-wave oscillations in rodents is not yet well understood, but experimental and theoretical work has proposed that it may involve species differences in the prevalence of GABAergic receptor subtypes affecting the duration of the oscillatory cycle (Destexhe, 1999;Destexhe et al., 1999;Sanchez-Vives et al., 2021).Consistent findings in cats and rats show that these abnormal oscillations dynamically alter the intrinsic excitability of cortical neurons, impeding their ability to integrate and process sensory information efficiently (Neckelmann et al., 2000;Williams et al., 2016Williams et al., , 2020)).It has therefore been suggested that such instability of sensory representations during seizures may contribute to impairing the cortical functions necessary for normal conscious experience (Williams et al., 2020).Other types of epilepsy are associated with abnormal cortical synchronisation in mammals (Uhlhaas and Singer, 2006).For example, short periods of highfrequency oscillations (100-500 Hz) are frequently observed near the time of seizure onset in rodent models and patients with mesial temporal lobe or neocortical focal epilepsy, where they may signal proximity to the epileptogenic focus (Bragin et al., 1999;Jirsch et al., 2006). . . Anoxia Disturbed cortical dynamics are also characteristic of cerebral ischaemia or anoxia following cardiac arrest, asphyxia or severe stroke.As a metabolically expensive organ, the mammalian brain is very sensitive to any interruption in its oxygenation, and prolonged global anoxia is often associated with a poor prognosis.Loss of oxygenation results in a rapid decrease in the amplitude and frequency of cortical activity, which gives way to a flat or isoelectric EEG.However, convergent findings from clinical investigations and studies on various mammals indicate that this sustained electrocerebral inactivity, marking the entry into a deep comatose state, is generally preceded by stereotyped changes in the frequency content of EEG signals (reviewed in Charpier, 2023). Experiments conducted on sedated or anaesthetised rodents, cats, and monkeys (rat : Hansen, 1978;Borjigin et al., 2013;Schramm et al., 2020;Carton-Leclercq et al., 2023;cat: Sugar and Gerard, 1938;Creutzfeldt et al., 1957;Hossmann and Sato, 1971;monkey: Myers and Yamaguchi, 1977) have shown that the onset of anoxia is associated with a transient increase in betagamma activities, often mixed with theta oscillations (Figure 7).Single-unit and intracellular recordings revealed that this period of high-frequency EEG activity correlates with rapid, low-amplitude synaptic fluctuations in deep-layer pyramidal neurons, a slight depolarization of their resting membrane potential and an increase in their firing rate (Figure 7) (Creutzfeldt et al., 1957;Carton-Leclercq et al., 2023).This cortical state is thought to result from an overactivation of synaptic receptors due to an excessive release of glutamate in the interstitial medium (Katchman and Hershkowitz, 1993;Fleidervish et al., 2001), although a complete understanding of the underlying cellular and network mechanisms will require further exploration.Fast electrical activity is also observed on the EEG of a number of critically ill patients after activation of the Do Not Resuscitate-Comfort Care protocol and discontinuation of life-sustaining treatment (Auyong et al., 2010;Norton et al., 2017;Xu et al., 2023).These brief occurrences of arousal-like activity, suggesting the emergence of consciousness without any outward signs of it, have been proposed as the origin of the vivid sensations and paradoxical lucidity associated with near-death experiences (Chawla et al., 2017;Parnia et al., 2023;Xu et al., 2023).However, high-frequency oscillations are a widespread rhythm not specific to conscious experiences, and the rather short duration of this episode of cortical activation has been deemed difficult to reconcile with the prolonged perceptions reported by some experiencers.The persistence of these rapid oscillations up to the moment of the isoelectric state and their resurgence at irregular intervals after its onset are also controversial.They have not been documented in all studies (see Chawla et al., 2017;Vincente et al., 2022;Xu et al., 2023 for report of such late surges of fast activity and Clute and Levy, 1990;Norton et al., 2017;Matory et al., 2021 for their nonoccurrence) and it has been suggested that they may, at least in part, reflect muscle contractions rather than brain activity (Greyson et al., 2022). The high-frequency cortical state is rapidly followed by a gradual decline of cortical activities in all frequency bands, briefly interrupted by a late surge of delta waves.This low-frequency EEG pattern is commonly observed in mammals exposed to anoxia and was first reported in humans by Berger during his pioneering experiments (human: Berger, 1934;Clute and Levy, 1990;Ammirati et al., 1998;Norton et al., 2017;rat: Borjigin et al., 2013;Lee et al., 2017;Schramm et al., 2020;Carton-Leclercq et al., 2023;cat: Sugar and Gerard, 1938;Creutzfeldt et al., 1957;rabbit: Colin et al., 1981).At this stage of the anoxic process, cortical pyramidal neurons show a reduced spontaneous firing and irregular synaptic depolarizations that progressively attenuate in amplitude and frequency with time.This dampening of cellular activity precedes the entry into the isoelectric state, marked by a cessation of spontaneous firing, and a loss of EEG and neuronal voltage fluctuations (Figure 7) (Creutzfeldt et al., 1957;Fujimura et al., 1997;Schramm et al., 2020;Carton-Leclercq et al., 2023).When ATP reserves reach a critical threshold and metabolic impairment is near-complete (85-90% decrease of the normal level), a large amplitude EEG wave, referred to as the "wave of death", suddenly emerges on the flat EEG.This wave reflects a massive depolarization (anoxic depolarization) of cortical neurons following the inhibition of Na + /K + ATPase pumps and the loss of homeostatic control of transmembrane ionic concentrations (Pietrobon and Moskowitz, 2014).Except for its onset time that may differ among species (but direct comparison is difficult due to variations in experimental conditions; Charpier, 2023 for review), anoxic depolarization has been observed with many phenomenological and mechanistical similarities in most mammals (mouse: Takano et al., 2007;rat: van Rijn et al., 2011;Schramm et al., 2020;Carton-Leclercq et al., 2023;human: Carlson et al., 2018;Dreier et al., 2018;rabbit: Leão, 1947;cat: Hossmann, 1971;monkey: Harris et al., 1981), as well as in the central nervous system of several invertebrates (Spong et al., 2016).This suggests that the neural correlates of near-death processes have also been well preserved throughout evolution.It is important to note that the wave of death is no longer considered terminal.Evidence from multi-scale electrophysiological recordings in rodents has shown that it can be reversed by a timely reoxygenation and replaced by its mirror wave, the "wave of resuscitation", which reflects the progressive repolarization of cortical neurons (Schramm et al., 2020;Carton-Leclercq et al., 2023).Although the wave of resuscitation has not yet been identified in humans, there is good reason to believe that the use of suitable electrodes and broadband recordings that preserve low frequencies will soon enable its discovery. Discussion The data reviewed above highlight the many anatomical disparities that distinguish the neocortex of the different mammals.Evolution has led to variations in the overall size of the brain, the number of cortical areas, the number of neurons, the size of their dendritic trees and the estimated number of synapses, and some of these differences appear to be particularly pronounced in humans.Does this mean that we can talk about unique properties that would set us apart from other species?Probably not.Firstly, because most of these changes are not exclusive to our species but are shared by most large-brained mammals.Secondly, because evidence suggests that the mammalian neocortex is organised into basic computational circuits, with common neuronal types, distribution profiles and synaptic connectivity patterns.This conservation extends to the electrophysiological properties of pyramidal neurons, whose biophysical characteristics (apart from a few specificities such as the expression of the h-current in the supragranular layers), excitability rules and firing profiles are similar across species.Furthermore, the consistency of unitary excitatory synaptic potentials suggests that, despite variations in neuronal architecture, the fundamental transfer of synaptic information between neurons remains unchanged, probably as a result of compensatory processes.Additionally, large-scale brain dynamics, which underpin integrative functions and dysfunctions of the brain, also appear to be largely preserved, pointing to a degree of determinism in the structure-function relationship of neocortical networks.However, variations in the morphology and proportion of certain neuron subtypes between species, individuals or cortical regions argue against the existence of a single, stereotyped mode of operation of the mammalian cortical circuit.To borrow a concept dear to the field of comparative anatomy, it would be more appropriate to refer to a "unity of plan" or "archetypal scheme" from which, on a continuum of possibilities, a number of specialisations emerge to meet different functional requirements.Thus, compared with other species, there is no revolution in the properties of human neurons and neocortical circuits; there are certainly differences, but there is also a marvellous resemblance, reflecting the continuation of an evolution that began with the first mammals and will probably continue beyond our species.In this respect, we could transpose to the study of the organisation and functioning of the neocortex what Darwin (1871) wrote in The Descent of Man, and selection in relation to sex: "Whether primaeval man, when he possessed very few arts of the rudest kind, and when his power of language was extremely imperfect, would have deserved to be called man, must depend on the definition which we employ.In a series of forms graduating insensibly from some ape-like creature to man as he now exists, it would be impossible to fix on any definite point when the term 'man' ought to be used."Echoing Darwin's thoughts on language evolution, it is interesting to note that left hemispheric asymmetries in regions homologous to human language areas have been observed in non-human primates (Gannon et al., 1998;Cantalupo and Hopkins, 2001).Moreover, research on baboons and chimpanzees has revealed that the direction and degree of lateralization of gestural communication are linked to contralateral asymmetry of part of the cortical region homologous to Broca's area (Taglialatela et al., 2006;Meguerditchian et al., 2012;Becker et al., 2022).These findings suggest a possible evolutionary continuity between primate gestural communication lateralization and human language hemispheric specialisation, implying that the cerebral organisation of communication areas could have been inherited from a common ancestor, before the emergence of speech (Becker et al., 2022).Furthermore, the recent discovery of a complex structure in sperm whale vocalisations (Sharma et al., 2024), as well as the detection of a lateralized arcuate fasciculus (a white matter fibre tract involved in human language) in the bottlenose dolphin (Wright et al., 2018), raise intriguing questions about the existence and function of such hemispheric asymmetry in species that have long diverged from primates. The analysis of the basic components of the circuit could, however, be considered insufficient to account for the complexity of functioning of the neocortex as a whole.Indeed, the emergentist approach to complex systems postulates that at each level of complexity, entirely new properties-not present in the system's constituents-appear, governed by new laws, new concepts and new generalisations; a theory that could be summed up by the physicist Anderson's eloquent formula "more is different" (Anderson, 1972).A well-known example of this theory is that of the properties of water, such as vorticity or entropy, which cannot be deduced from knowledge of the characteristics of individual H 2 O molecules.Applied to the brain, the concept of emergence implies a dynamic and highly integrated functioning of the cerebral cortex, in which the coordinated activity of groups of neurons distributed in primary and associative cortical areas allows the emergence of coherent behavioural and cognitive acts.Direct (though only correlative) evidence for the formation, mediated by synchronisation mechanisms over multiple frequency bands, of such coherent ensembles has been obtained in humans, cats, monkeys, and rats, during a variety of active perceptuomotor behaviours (for a review, see Varela et al., 2001).It seems reasonable to assume that the iteration of an archetypal cortical circuit (considered here as a cognitive module or unit) could increase the number of possible combinations of the different sub-circuits and favour the behavioural diversity of the system.We also need to consider the evolutionary aspect of these complex brain figures, which are likely to change as a function of the physical, biological and social contexts of each species or individual.These dynamic and reciprocal links between the cortex and its environment will thus lead to the emergence of diverse and characteristic experiences, underpinned by common elementary cerebral processes from which they are indissociable. A balanced synthesis between reductionist and emergentist approaches could postulate that the cerebral activities underlying mental processes are shared by different species and that cognitive performance is the result of complex cerebral dynamics.However, this perspective does not answer the fundamental question of how specific spatio-temporal brain dynamics translate into conscious experiences, such as sensations, reasoning or memories.Like the physiologist du Bois-Reymond (1882) (the first to describe AP in nerve fibres), who identified this issue as one of the seven world enigmas in his famous 1882 lecture on Ignorabimus, modern neurophilosophers continue to assert that bridging the explanatory gap between brain activity and subjective phenomena, regardless of the species, will remain beyond the reach of scientific investigation (Levine, 1983;Chalmers, 1995).A more optimistic view would be to consider that this intricate relationship between neural and mental events, which defies the intrinsic limits of our knowledge but whose analysis is a real driving force for questioning, is a missing link whose contours and principles have yet to be defined. FIGURE FIGURE Top left, The Island of Java in the Indonesian archipelago.Right, The skull cap of Pithecanthropus erectus.Middle, Excavations at Trinil in s on the left bank of the Solo River.The arrow indicates the approximate discovery site of the skull cap.Bottom left, Profile drawing of the Trinil site showing the position of the fossil remains in the sediment (level D).Right, Handwritten page from Dubois' article published in and photograph (John Reader/Science Photo Library) of his reconstruction of Pithecanthropus erectus, shown at the Universal Exhibition in Paris.Adapted with permission from Wood, , and Dubois, . FIGURE FIGURE Evolution of hominin brain size.(A) Evolution of hominin endocranial volume over time.Adapted with permission from Hublin et al. ( ). (B) Di erences in brain size and shape of a modern human (blue), a Neanderthal from La Chapelle-aux-Saints (red) and a chimpanzee (green), visualised by computer tomography.Adapted with permission from Neubauer et al. ( , ). FIGURE FIGURE Variability and evolution of pyramidal neuron morphology in mammals.(A) Database of reconstructed human temporal cortex L / neurons, classified according to their somatic depth, indicated in µm at the top of each cell.Two examples of cells whose soma was located (top) and (bottom) µm from the cortical surface are enlarged on the left.Adapted with permission from Deitcher et al. ( ). (B) Examples of supragranular (L / , top) and infragranular (L , bottom) pyramidal neurons from the visual (L / ) and motor (L ) cortices of the short-beaked echidna, the primary motor cortex of Benett's wallaby, the sensorimotor cortex of the American opossum, the primary motor cortex of the northern gira e, the primary somatosensory cortex of the Sprague-Dawley rat, the primary motor cortex of the chacma baboon and the human temporal cortex.Adapted from Dann and Buhl ( ) (echidna L / ); Hassiotis and Ashwell ( ) (echidna L ), Jacobs et al. ( ) (wallaby, baboon, gira e), Boyer et al. ( ) (rat L / ), Mahon and Charpier ( ) (rat L ), Deitcher et al. ( ) (human L / ), and Kalmbach et al. ( ) (human L ).Neurons were redrawn from the sources mentioned.Scale bars ( µm) are species-specific and apply to supra-and infragranular neurons.Estimates for monotreme-therian divergence and marsupial-placental separation are taken from Bininda-Emonds et al. ( ) and Phillips et al. ( ). FIGURE FIGURE Intrinsic membrane properties of L / pyramidal neurons.(A) Simultaneous somatic (Soma, black) and dendritic (Dend., blue) recordings of an EPSP generated at dendritic site by the injection of uniform current waveform (Inj.current, bottom traces) in a human and a rat temporal cortex L / pyramidal neuron.The location of recording electrodes is shown on the reconstructed neurons.The right graph compares the degree of dendro-somatic (D-S) voltage attenuation of simulated EPSPs as they spread from their increasingly remote apical dendritic site of generation to the soma, in human (filled symbols) and rat (open symbols) neurons.(B) Evoked dendritic spikes (asterisks) driving somatic AP firing in human and rat L / neurons.(C) Voltage responses to hyperpolarizing current steps recorded from a mouse temporal association cortex (left) and a human middle temporal gyrus (right) L / pyramidal neuron.Double arrow indicates the amplitude of the voltage sag.(D) Example voltage responses to depolarizing current steps of + and + pA for a representative mouse (left) and human (right) L / pyramidal neuron.(E) From top to bottom, Current-evoked firing responses recorded from a rat medial prefrontal cortex, a macaque prefrontal cortex, and a human temporal cortex L / pyramid.Adapted with permission from Gooch et al. ( ) (A, B); Kalmbach et al. ( ) (C, D); van Aerde and Feldmeyer ( ) (rat), González-Burgos et al. ( ) (rhesus monkey), and Deitcher et al. ( ) (human) (E). FIGURE FIGURE Intrinsic membrane properties of L pyramidal neurons.(A, B) Electrophysiological properties of infragranular ET and IT neurons in rat medial prefrontal cortex (top) and human temporal cortex (bottom).(A) Representative voltage responses of ET (green) and IT (red) neurons to depolarizing current steps of increasing intensity (left), and corresponding relationships between injected current intensity and number of evoked spikes (right).(B) Example APs in ET and IT neurons and corresponding summary data of their voltage threshold and half-width duration.Data are presented as mean ± s.e.m. in the top graph.Box plots in the bottom graph denote the median and -th percentiles.(C) Input-output function of thick-tufted L neurons across placental mammalian species.Left, Firing frequency as function of injected current intensity in Etruscan shrew, mouse, rat, rabbit, macaque, and human neurons.Lines illustrate population medians.Two-photon images of example somas are shown on top.Right, Population data of somatic surface area (mean ± s.e.m.), current firing threshold (median and -th percentiles), and input-output neuronal gain (median and -th percentiles).(D) h-channel-related membrane properties of L thick-tufted neurons.The graphs illustrate the sag ratio as a function of distance from the soma in mouse, rat, rabbit, and human neurons.Triangles represent somatic medians and lines are exponential fits.(A, B) are adapted with permission from Dembrow et al. ( ) (rat) and Kalmbach et al. ( ) (human), and (C, D) are adapted with permission from Beaulieu-Laroche et al. (). FIGURE FIGURE Conservation of sleep-wake EEG patterns across mammals.(A) Example traces of EEG activity recorded from a monotreme (platypus), a marsupial (dusky antechinus), and two placental (rat and human) species during active wakefulness (wake), deep stages of slow-wave sleep (SWS), and rapid eye movement sleep (REM).Note the similarity of morphology and frequency of EEG waveforms across species.A moderate-voltage EEG recording is illustrated for platypus REM sleep (see text).For each species, the EEGs in the three states of alertness are presented on the same scale.Data are adapted from Siegel et al. ( ) (platypus); Zaid et al. ( ) (dusky antechinus); Mahon et al. ( ) (rat); Stickgold and Walker ( ), Liu and Dan ( ) (human).(B) Bilateral EEG recordings (L: left, R: right) from a semi-aquatic northern fur seal during SWS, showing bilateral high-voltage slow waves when animals are sleeping on land (BSWS, top) and unihemispehric slow-wave activity when sleeping in the sea (USWS, bottom).Belugas spend most of their sleeping time in a unihemsipheric SWS (right).Adapted with permission from Lyamin et al. ( ) (fur seal) and Lyamin et al. ( ) (beluga). cortical oscillations are the most prominent pattern of electrical activity during the deep phases	2024-06-22T15:04:15.827Z	2024-06-20T00:00:00.000	{ "year": 2024, "sha1": "8084379532b610f4b9ff2b95ef6c0a25590ac948", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fnsys.2024.1413780/pdf?isPublishedV2=False", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "1efb0fcaeba355e7ba3ec1215ea6203661db7655", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [] }
113760943	pes2o/s2orc	v3-fos-license	Radial Strains of Double-layer Cylinders in Hydraulic Props of Powered Supports At present a lot of efforts are made to use double-layer power cylinders in hydraulic props of powered supports. To study the response of these cylinders to loads a special finite-element model has been developed and used for investigations into tension effect and double-layer cylinder thickness – radial strain relation under pressure of hydraulic liquid 50 МPа. It has been revealed that double-layer cylinders are distinguished by much lower radial strains in the zone of cup-like sealing elements as if compared with one-layer cylinders, as well as equivalent stresses are lower, and safety factor is higher. The data of the study can be recommended to calculate appropriate geometrical parameters of hydraulic props with respect to lower radial strains of a hydraulic cylinder, which improve its leak-tightness and functioning of cup-like sealing elements. The obtained results can be useful for design and construction of powered supports. Introduction A hydraulic prop is a hydraulic power cylinder; when it is exposed to loads its radial strains specify the cylinder -piston gap to be sealed affecting efficiency of a seal [1], as the consequence, leaktightness. A numerical value of this gap is a sum total of a fit up gap, calculated on the base of piston and cylinder fabrication tolerances, and radial strains of cylinder inner surface (dR) under hydraulic liquid pressure, dependent on the value of this pressure (P), manufacturing technology [2][3][4], hydraulic prop structure [5,6], its hydraulic extensibility (lp) [7], as well as on the structure of a support [8] and its external environment [9,10]. Methods A parameter-oriented model on the base of finite-element method was used to calculate radial strains and stresses in elements of the cylinder. This method is an up to date computational approach enabling precise and fast calculating complex designs by means of numerical engineering. Work Description The results of earlier investigations into radial strains of the main cylinder [7] are as follows: the curve shape is of the form depicted in Fig. 1, which becomes wider at the head end and narrows at the rod end. The following reference points were used to compare assessment criteria of strains lengthwise the cylinder. P. 1 has maximal radial strains at the piston (dR 1 ). P. 2 corresponds with the zone of stable radial strains in the cylinder body (dR 2 ). P. 3 agrees with maximal radial strains in the piston zone (dR 3 ) when compressing the cylinder, which can't exceed the minimal piston -cylinder fit up gap. P. 4 is in the zone of the first seal at the head end. Radial strains in this point (dR 4 ) determine the gap to be sealed, which affects sealing efficiency and leak-tightness of the hydraulic prop [1]. P. 5 has maximal radial strains at the bottom of the hydraulic cylinder (dR 5 ). It was proposed to fabricate the main cylinder of two layers coupled to each other by tension in order to reduce radial strains of the cylinder internal surface, which is necessary to improve operating conditions of cup-like sealing elements and leak-tightness of hydraulic props. A special finite-element parameter-oriented model was developed to study strain in the main cylinder of the hydraulic prop in the powered support, layout and parameters of this model are outlined in Fig. 2. Steel 9HF was admitted as material for outer cylinder, while steel 30HGSA was thought as material for inner cylinder. Properties of these materials are given in Table 1. Geometrical parameters of the model are presented in Table 2. Radial strains of double-layer cylinders were calculated according to the developed parameteroriented finite-element model on the base of planar linear 4-unit forming elements with axis symmetry and contact elements between layers. The following restrictions were accepted for the research purpose: total thickness of walls -25 mm; wall thickness relation of inner cylinder to the outer one -10/15, 12.5/12.5, 15/10, 20/5 mm; fit up tension -0.104 mm. As the result, radial strains of inner surface in a single-layer cylinder of a powered support OKP 70 were calculated provided that a hydraulic prop is fully extended (Fig. 3), the same strains were also computed for a double-layer cylinder (Fig. 4). Moreover, strains for various tensions between inner and outer layer of a cylinder are depicted in Fig. 4. The obtained radial strains in reference point 3 and reference point 4 (dR 3 and dR 4 ) are given in Table 3. The following criteria are proposed to measure strains in the main cylinder. Safety factor with respect to maximal tolerance by bulging cylinder, which is to be more than one to keep operating capacity of a hydraulic prop (a gap to be sealed doesn't exceed the safe one because of leak-tightness). Safety factor with respect to minimal tolerance by narrowing cylinder at the rod end provided there is no scuffing This criterion is to be more than one (in most cases) or negative. When 1   min n compression strains of a cylinder in reference point 3 are equal to the minimal possible gap dependent on piston and cylinder manufacturing tolerances, therefore, scuffing is possible on the cylinder face if units to be assembled are combined this way. If the values are negative, cylinder -piston gap gets broader relative to the fit up tolerance, as the result, scuffing is less probable than in an unloaded hydraulic prop. The system of proposed criteria enable assessing operational capacity of a hydraulic prop under critical pressure of hydraulic liquid. Values of criteria n Δmax and n Δmin for powered support OKP 70 with single-layer and double-layer cylinder are given in Table 3. Radial strains of the inner surface in the cylinder, pressure of hydraulic liquid is P=50 МPа, inner cylinder thickness is 15 mm, outer cylinder thickness is 10 mm, tension is various layer cylinder, wall thickness is 25 mm and Radial strains of the inner surface in the cylinder, pressure of hydraulic liquid is P=50 МPа, r cylinder thickness is 10 mm, tension is various Double-layer cylinders are advantageous over single-layer ones in view of data presented in Figure 3 and Figure 4, as well as in Table 4, due to the increasing safety factor measured relative to maximal equivalent stresses; dimensions of the hydraulic prop and its steel intensity are equal. In this case either reliability of the hydraulic prop can be improved (e. g. under dynamic load in a longwall, as the consequence, under pressure far beyond the rated one in the head end), or wall thickness can be reduced keeping the same safety factor. In general, double-layer cylinders used in the hydraulic prop of a powered support ОКP70 enabled increasing safety factor measured according to the maximal gap as if compared with the single-layer structure, and reducing radial strains of the cylinder near the first seal (Table 3). 4.Conclusions The following conclusions can be drawn on the base of comparative study on double-and single-layer cylinders in the powered support ОКP70. All structures in double-layer cylinders, even those with minimal tension, have lower radial strains if compared with the single-layer cylinder, all other parameters are equal. It supports the seal as it operates in a lower gap to be sealed, therefore, leaktightness of the prop gets better.	2019-04-15T13:07:31.495Z	2016-04-01T00:00:00.000	{ "year": 2016, "sha1": "f1ea330db1bb0cec99cbe1c8ec83a39f975c70fe", "oa_license": null, "oa_url": "https://doi.org/10.1088/1757-899x/127/1/012034", "oa_status": "GOLD", "pdf_src": "IOP", "pdf_hash": "8394a7be67a3b16466ef3bf9d5d4ec63460c4458", "s2fieldsofstudy": [ "Engineering", "Materials Science" ], "extfieldsofstudy": [ "Physics", "Engineering" ] }
3294242	pes2o/s2orc	v3-fos-license	MiR-361-5p suppresses chemoresistance of gastric cancer cells by targeting FOXM1 via the PI3K/Akt/mTOR pathway Gastric cancer is a prevalent cancer and chemotherapy is a main treatment for patients. Docetaxel is commonly used as a chemotherapeutic drug for gastric cancer patients. With the increasing emergence of docetaxel resistance, exploring the mechanism of chemoresistance may improve prognosis of patients. In this study, we found that overexpressed miR-361-5p suppressed chemoresistance to docetaxel of gastric cancer cells (SGC-7901, MKN-28) by decreasing IC50 values of docetaxel while increasing cell apoptosis rate, especially in docetaxel resistant SGC-7901 cells. Further researches revealed that overexpressed miR-361-5p inhibited chemoresistance through inhibiting autophagy with a characteristic of declined number of LC3+ puncta, decreased expression of Beclin-1 and the ratio of LC3 II/I and increased expression of p62. Bioinformatics study and Luciferase reporter assay indicated that FOXM1 was a target of miR-361-5p and FOXM1 was negatively regulated by miR-361-5p in gastric cancer. Simultaneously, overexpression of FOXM1 counteracted the inhibitory effects of miR-361-5p on chemoresistance of gastric cancer cells through activating autophagy, further certifying the targeting relationship between the two. Moreover, overexpressed miR-361-5p activated the PI3K/Akt/mTOR pathway. The adding of PI3K inhibitor LY294002 played an opposite role to miR-361-5p mimic by inducing autophagy and chemoresistance to docetaxel of gastric cancer cells compared with docetaxel + miR-361-5p mimic group, indicating that miR-361-5p suppressed autophagy-induced chemoresistance via the PI3K/Akt/mTOR pathway in gastric cancer cells. In conclusion, we found that miR-361-5p suppressed autophagy-induced chemoresistance of gastric cancer cells through targeting FOXM1 via the PI3K/Akt/mTOR pathway, providing a foundation for the mechanism research and treatment of gastric cancer. INTRODUCTION Gastric cancer (GC) is one of the most prevalent cancers around the world, accounting for almost 10% of all cancer death [1]. Because the clinical symptom is not obvious at early stage, patients with GC are often detected at late stage, thus causing an extremely low 5-year survival rate [2]. Surgery is considered as the main treatment for patients with GC. However, for patients who miss the best time for surgery, palliative www.impactjournals.com/oncotarget/ Oncotarget, 2018, Vol. 9, (No. 4), pp: 4886-4896 Research Paper chemotherapy is the main choice of treatment. During the past decades, several new-generation cytotoxic agents including docetaxel have been used for the treatment of GC [3]. Docetaxel has been reported to promote the assembly and stabilization of microtubules to inhibit the depolymerization in the treatment of GC [4,5]. However, obvious resistance to docetaxel has been reported in some types of tumors including prostate cancer, breast cancer, GC and nasopharyngeal carcinoma, seriously limiting the treatment effect of docetaxel [4,[6][7][8][9]. Therefore, inhibiting chemoresistance to docetaxel in GC may be a key role for improving therapeutic effects for GC patents. Autophagy is a cellular process which delivers intracellular aggregated or misfolded proteins to the lysosomal compartment where they are degraded and recycled [10]. Previous studies reported that activation of autophagy enhanced chemoresistance in various cancers. Fukuda's study indicated that inhibition of autophagy suppressed proliferation and overcame Cisplatin resistance of endometrial cancer cells [11]. A study by Zhang revealed that inhibition of TRPC5-induced autophagy suppressed drug resistance to Adriamycin in breast carcinoma [12]. However, detailed contributions of autophagy to chemoresistance of GC cells are still limited and require for further investigations. MicroRNAs (miRNAs), a group of small non-coding RNAs with 20-22 nucleotides, are reported to regulate the expression of specific genes [13]. Accumulated evidence indicated that miRNAs were involved in a great number of cellular processes including proliferation, apoptosis, migration and invasion during tumor progression [14,15]. Previous studies showed that some miRNAs such as miR-608 and miR-204 suppressed chemoresistance of tumor cells [16,17]. MiR-361-5p, one of the miRNAs, was found to function as a tumor suppressor in various tumors. Previous study showed that overexpressed miR-361-5p inhibited tumor growth of hepatocellular carcinoma in nude mice significantly [18]. However, whether miR-361-5p can regulate chemoresistance of tumor cells is still unclear. MiRNAs are reported to regulate the expression of their target genes to accomplish their effects. In our presents study, we found that FOXM1, a newly unified family member of Forkhead transcription factor, was one of the target genes of miR-361-5p [19]. FOXM1 was reported to confer resistance to herceptin and paclitaxel by altering microtubule dynamics to protect tumor cells from paclitaxel-induced apoptosis in breast cancer [20]. As both docetaxel and FOXM1 affected the microtubules dynamics, we suggested that FOXM1 might involve in chemoresistance to docetaxel, however, little research has been carried out. In our present study, we explored the mechanism of miR-361-5p/FOXM1 axis in regulating the chemoresistance of GC cells. We found that miR-361-5p suppressed autophagy-induced chemoresistance of GC cells through targeting FOXM1 via the PI3K/Akt/mTOR pathway, trying to find a new direction for the treatment of GC. MiR-361-5p suppresses chemoresistance of gastric cancer cells to docetaxel We initiated our study by investigating whether there was an association between miR-361-5p and chemoresistance to docetaxel of GC cells. SGC-7901 cells which were resistant to docetaxel and MKN-28 cells which were sensitive to docetaxel were treated with different concentrations of docetaxel and then cell viability was detected by MTT assay. We observed that docetaxel worked as an anti-tumor regent by suppressing cell viability significantly in a dose-dependent manner and the cell viability of SGC-7901 was higher than docetaxel sensitive cell line MKN-28 under the same concentration of docetaxel. Moreover, the half maximal inhibitory concentration (IC 50 ) values of docetaxel were approximately 0.025 mg/L and 0.015 mg/L in miR-361-5p mock transfected SGC-7901 cells and miR-361-5p mimic transfected SGC-7901 cells, respectively. However, the IC 50 in miR-361-5p mock transfected and miR-361-5p mimic transfected MKN-28 cells were approximately 0.015 mg/L and 0.013 mg/L respectively. These results suggested that overexpression of miR-361-5p increased the sensibility to docetaxel of gastric cancer cells and the promoting role was more obviously in docetaxel resistant cells ( Figure 1A-1B). Apart from that, results from cell apoptosis assay indicated that overexpression of miR-361-5p increased cell apoptosis rate compared with docetaxel + mock group in both docetaxel treated SGC-7901 and MKN-28 cells ( Figure 1C-1D, P * < 0.05, P # < 0.05). Docetaxel resistant cells SGC-7901 were chosen for following experiments. In conclusion, our results suggested that overexpressed miR-361-5p inhibited chemoresistance of GC cells to docetaxel. Overexpressed miR-361-5p suppresses chemoresistance of gastric cancer cells by inhibiting autophagy Accumulating evidence supported that chemoresistance of cancer cells was correlated with autophagy [21]. Therefore, we investigated whether miR-361-5p suppressed chemoresistance of GC through autophagy in our present study. The localization of LC3 to autophagosome formation was assessed through GFP-LC3 expression. Our data showed that the number of LC3 + puncta significantly increased after docetaxel treatment compared with control group. However, overexpressed miR-361-5p reduced the number of LC3 + puncta induced by docetaxel treatment compared with docetaxel + mock group markedly (Figure 2A-2B, P < 0.01, P # < 0.05). Then the expression of autophagy-related proteins was measured by western blot. Docetaxel treatment increased the expression of Beclin-1 and the ratio of LC3 II/I while decreased the expression of p62 compared with control group. Overexpressed miR-361-5p decreased the expression of Beclin-1 and the ratio of LC3 II/I while increased the expression of p62 compared withdocetaxel + mock group significantly, suggesting that miR-361-5p suppressed chemoresistance of GC cells through inhibiting autophagy ( Figure 2C-2D, P < 0.01, P # < 0.05). In order to verify our conjecture, autophagy inducing reagent rapamycin (Rapa) was used in our study. Because Rapa was dissolved in DMSO, thus DMSO group was used as control and we found that DMSO didn't have significant effect on cell viability and apoptosis rate. We observed that the IC 50 values were approximately 0.015 mg/L and 0.025 mg/L in miR-361-5p mimic group and mimic + Rapa group respectively, indicating that activation of autophagy by Rapa decreased the sensibility to docetaxel of overexpressed miR-361-5p GC cells ( Figure 2E). Besides that, activation of autophagy abolished the promoting effect of miR-361-5p on cell apoptosis compared with docetaxel + mimic group remarkably ( Figure 2F, P * < 0.05, P # < 0.05), suggesting that activation of autophagy increased chemoresistance of GC cells to docetaxel. Taken together, these results supported that overexpressed miR-361-5p suppressed chemoresistance of GC cells through inhibiting autophagy. FOXM1 is a target of miR-361-5p in gastric cancer cells In order to explore the molecular mechanisms of miR-361-5p in chemoresistance of GC cells, putative miR-361-5p targets were predicted through bioinformatics analysis. The predicted results showed that FOXM1 was one of the potential targets of miR-361-5p ( Figure 3A). SGC-7901 cells were transfected with pcDNA3.1-FOXM1 for overexpression of FOXM1 ( Figure 3B). Results from western blot showed that the expression of FOXM1 was increased by docetaxel treatment compared with control group and was decreased by the adding of miR-361-5p mimic compared with pcDNA3.1-FOXM1 group under the application of docetaxel, indicating that the expression of FOXM1 was negatively regulated by miR-361-5p ( Figure 3C-3D, P * < 0.05, P # < 0.05, P $ < 0.05). Moreover, the luciferase reporter assay showed that co-transfection with miR-361-5p mimic and FOXM1 WT in SGC-7901 cells led to a significant decrease in luciferase activity. However, co-transfection with FOXM1 MUT and miR-361-5p mimic didn't show significant difference compared with cells transfected with FOXM1 MUT ( Figure 3E, P ** < 0.01), confirming the targeting relationship between FOXM1 and miR-361-5p. In summary, our data elucidated that FOXM1 was a target of miR-361-5p and the expression of FOXM1 was negatively regulated by miR-361-5p in GC cells. DISCUSSION In gastric cancer treatment, surgery combined with chemotherapy is one of the most common methods to improve prognosis of patients. Docetaxel, one of the chemotherapy agents, is commonly used as a single agent or in combination with other agents such as latinum and fluoropyrimidine (DCF regimen) for the treatment of cancer patients [23,24]. Docetaxel is an active agent in advanced GC and it has also been shown to lack crossresistance with other drugs in GC [5]. However, the resistance to docetaxel did occur and more investigations on the mechanism of chemoresistance to docetaxel of tumor cells are urgently required. Our study presented the role of miR-361-5p/FOXM1 axis in the chemoresistance to The bars showed means ± SD of three independent experiments. P * < 0.05, P ** < 0.01 compared with control group, P # < 0.05 compared with docetaxel treated group, P $ < 0.05 compared with docetaxel + mimic group. www.impactjournals.com/oncotarget docetaxel of GC cells and tried to find a better therapeutic strategy for overcoming the resistance to docetaxel in GC. MiRNAs have gained increasing attentions as they appeared to be involved in tumor progression. Increasing evidence supported that miRNAs acted as tumor suppressors or oncogenes and involved in post-translational regulation of gene expression [25]. Previous studies reported that a great number of miRNAs were related to chemoresistance in various cancers. For example, Rajabpour's study revealed that miR-608 suppressed gemcitabine chemoresistance through regulating the resistance genes in pancreatic cancer cells [16]. Besides that, microRNA-204 was reported to increase chemosensitivity of prostate cancer cells [17]. Recently, accumulating evidence supported that miR-361-5p acted as a tumor suppressor in many kinds of tumors. Zhuang's research indicated that down-regulation of miR-361-5p associated with aggressive clinicopathological features and unfavorable prognosis in non-small cell lung cancer [26]. It was also reported that miR-361-5p inhibited epithelial-to-mesenchymal transition through targeting Twist1 in glioma cells [27]. However, whether miR-361-5p could regulate chemoresistance of tumor cells still remained unknown. In our present study, SGC-7901 cells which were resistant to docetaxel and MKN-28 cells which were sensitive to docetaxel were chosen for our experiments. Results indicated that overexpressed miR-361-5p in SGC-7901 and MKN-28 cells decreased the IC 50 value of docetaxel and increased cell apoptosis rate, suggesting that miR-361-5p inhibited chemoresistance to docetaxel of SGC-7901 and MKN-28 cells and the inhibiting effect was more obviously found in SGC-7901 cells. Our results elucidated that the inhibitive effect of miR-361-5p on chemoresistance in GC cell lines. After learning the effect of miR-361-5p on chemoresistance of GC cells, we then set to explore its potential mechanism. Autophagy was a programmed cell survival mechanism which was reported to enhance chemoresistance of tumor cells. Hu's data showed that p53 was important for the activation of autophagy and suppression of p53 potentiated chemosensitivity in nutrient-deprived cholangiocarcinoma cells [28]. Activation of autophagy by Fusobacterium nucleatum promoted chemoresistance of colorectal cancer cells [29]. Similarly, our results showed that overexpressed miR-361-5p decreased the number of LC3 + puncta which was widely used as a marker for autophagosomes. Moreover, the expression of Beclin-1 which was necessary for the formation of autophagosomes and the ratio of LC3 II/I were decreased while the expression of p62 which cleared the autophagosome was increased by overexpressed miR-361-5p, indicating that miR-361-5p inhibited autophagy in docetaxel treated SGC-7901 cells. Besides that, activation of autophagy by Rapa counteracted the inhibitory effect of miR-361-5p on chemoresistance to docetaxel through decreasing docetaxel sensibility and cell apoptosis rate in GC cells. Our data revealed that miR-361-5p suppressed chemoresistance through inhibiting autophagy in GC cells. Furthermore, our bioinformatics study revealed that FOXM1 was one of the potential targets of miR-361-5p. FOXM1, a proliferation specific oncogenic transcription factor, was reported to regulate microtubule dynamics to mediate chemoresistance of tumor cells [20]. Previous research discovered that inhibition of FOXM1 could reverse docetaxel resistance in GC [4]. FOXM1 also mediated resistance to herceptin and paclitaxel in breast cancer [20]. Another study from Hou elucidated that overexpression of miR-361-5p suppressed lung cancer proliferation and invasion by targeting FOXM1 [30]. Consistent with previous researches, we found that the expression of FOXM1 was negatively regulated by miR-361-5p. Overexpressed FOXM1 enhanced autophagy and chemoresistance to docetaxel of GC cells, suggesting that overexpression of FOXM1 weakened the inhibitory effects of miR-361-5p on autophagy-induced chemoresistance in gastric cancer cells. PI3K/Akt/mTOR pathway was a well-known pathway involved in the regulation of autophagy. Inhibition of PI3K/Akt/mTOR pathway was reported to enhance autophagy [31]. Previous study showed that autophagy was negatively regulated by the activation of mTOR in APP/PS1 double transgenic mice [32]. In accordance with these studies, overexpressed miR-361-5p was found to activate PI3K/Akt/mTOR pathway while overexpressed FOXM1 suppressed this pathway. Apart from that, PI3K inhibitor LY294002 abolished the effect of miR-361-5p on autophagy and chemoresistance to docetaxel of GC cells compared with docetaxel + miR-361-5p mimic group, indicating that miR-361-5p suppressed autophagy-induced chemoresistance by targeting FOXM1 via the PI3K/Akt/ mTOR pathway in GC cells. Taken together, our present study demonstrated that miR-361-5p/FOXM1 axis played an important role in regulating chemoresistance to docetaxel of GC cells. MiR-361-5p suppressed autophagy-induced chemoresistance of GC cells by targeting FOXM1 via the PI3K/Akt/mTOR pathway, providing a foundation for the mechanism research and treatment of GC. The human FOXM1 expression vector pcDNA3.1-FOXM1 was transfected into cells using Lipofectamine 2000 reagent in accordance with the manufacturer's protocol. MTT assay For MTT assays, 5x10 3 cells were seeded into 96well plates in triplicate and cultured with 100 μL of fresh medium which contained different concentrations of docetaxel for 48 h. Then, MTT solution (20 μL, 5 mg/ml in PBS) was added to each wells and incubated at 37°C for 4 h. After that, the solution was removed and 200 μL of DMSO was added to each well. The optical density at 490 nm was measured by a microplate reader (Bio-Rad, Hercules, CA, USA) after 10 min of vibration mixing. Cell apoptosis assay For cell apoptosis analysis, Annexin V-FITC/ propidium iodide (PI) apoptosis detection kit (Multisciences, Shanghai, China) was used in our present study. Firstly, different groups of cells (3x10 5 ) were collected and washed in ice-cold PBS. Then cells were resuspended and incubated with 5 μL of Annexin V-FIFC and 10 μL of PI. Cell apoptosis was analyzed in a flow cytometer (BD Biosciences). Fluorescence microscopy MiR-361-5p mimic or mock transfected SGC-7901 cells were transfected with GFP-LC3 plasmid by Lipofectamine 3000™ (Invitrogen). After untreated SGC-7901 cells or transfected cells treated with or without docetaxel, the number of puncta formation of GFP-LC3 was determined under fluorescent microscopy as described before [33]. Cells with more than 5 puncta counted were considered to have accumulated autophagosomes. Western blot Proteins were extracted from cells using RIPA lysis buffer (Beyotime Institute of Biotechnology, Shanghai, China) according to the manufacturer's protocol. Total proteins were separated by SDS-PAGE gel and then transferred into PVDF membranes (Millipore). After incubating with 5% skim milk for 1 h at room temperature, the membranes were incubated with primary antibodies overnight at 4˚C. The primary antibodies (Cell Signaling Technology, Beverly, MA., USA) used in this study include anti-Beclin Luciferase reporter assay The fragment of FOXM1 containing the target sequence of miR-361-5p was inserted into a pmirGlO Dual-luciferase miRNA Target Expression Vector (Promega, Madison, WI, USA) to form the reporter vector FOXM1-wild-type (FOXM1-WT) while FOXM1-mutated-type (FOXM1-MUT) contained mutated binding site. Cells at 5x10 4 were seeded into 96-well plate to reached 60% confluence. Then cells were co-transfected with FOXM1-WT or FOXM1-MUT and miR-361-5p mimics using Lipofectamine 2000. The Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) was used for testing the luciferase activity. Statistical analysis All results were presented as mean ± standard deviation (SD). Student's t test was performed to analyze the difference between groups. Statistical analysis was performed with SPSS software (version 19). Values of P < 0.05 were considered statistically significant. Author contributions All the authors conducted the conception and design of the study, acquisition and interpretation of data, drafting the article, final approval of the version to be published equally.	2018-04-03T02:29:19.431Z	2017-12-20T00:00:00.000	{ "year": 2017, "sha1": "ce7a398e33f88ff42b2d6ddcaf66e51aaac15e1c", "oa_license": "CCBY", "oa_url": "http://www.oncotarget.com/index.php?journal=oncotarget&op=download&page=article&path[]=23513&path[]=74054", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "ce7a398e33f88ff42b2d6ddcaf66e51aaac15e1c", "s2fieldsofstudy": [ "Biology", "Chemistry", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
260415225	pes2o/s2orc	v3-fos-license	Identification of an adverse outcome pathway (AOP) for chemical-induced craniofacial anomalies using the transgenic zebrafish model Abstract Craniofacial anomalies are one of the most frequent birth defects worldwide and are often caused by genetic and environmental factors such as pharmaceuticals and chemical agents. Although identifying adverse outcome pathways (AOPs) is a central issue for evaluating the teratogenicity, the AOP causing craniofacial anomalies has not been identified. Recently, zebrafish has gained interest as an emerging model for predicting teratogenicity because of high throughput, cost-effectiveness and availability of various tools for examining teratogenic mechanisms. Here, we established zebrafish sox10-EGFP reporter lines to visualize cranial neural crest cells (CNCCs) and have identified the AOPs for craniofacial anomalies. When we exposed the transgenic embryos to teratogens that were reported to cause craniofacial anomalies in mammals, CNCC migration and subsequent morphogenesis of the first pharyngeal arch were impaired at 24 hours post-fertilization. We also found that cell proliferation and apoptosis of the migratory CNCCs were disturbed, which would be key events of the AOP. From these results, we propose that our sox10-EGFP reporter lines serve as a valuable model for detecting craniofacial skeletal abnormalities, from early to late developmental stages. Given that the developmental process of CNCCs around this stage is highly conserved between zebrafish and mammals, our findings can be extrapolated to mammalian craniofacial development and thus help in predicting craniofacial anomalies in human. Craniofacial anomalies comprise over one-third of all congenital birth defects and over 700 disorders are associated with craniofacial features (https://www.ncbi.nlm.nih.gov/omim).These anomalies are thought to be caused by genetic and environmental factors, including pharmaceuticals and chemical agents, during embryonic development (Beaty et al., 2016;Edison and Muenke, 2003;Poswillo, 1988;Yoon et al., 2016).Such pharmaceuticals and chemical agents are called teratogens.The teratogenic potential, or teratogenicity, is evaluated and assessed by developmental toxicity tests.Recently the growing number of produced chemicals demands their evaluation and assessment with high-throughput and greater accuracy.Although mammals such as rodents and rabbits have been traditionally used for the tests, current trends toward the 3R principles (replacement, reduction, and refinement) and saving resources have encouraged the development of alternative teratogenicity testing methods (Str€ ahle et al., 2012).To develop an alternative testing strategy for teratogenicity, we definitely need an adverse outcome pathway (AOP) framework which would improve prediction including cross-species extrapolation, and contribute to human health risk assessment.Therefore, a central issue is identification of AOPs that are relays of sequential events with a causal relationship at different levels of biological response leading to a toxic effect (Ankley et al., 2010;Burden et al., 2015;Knapen et al., 2015;Tollefsen et al., 2014;Villeneuve et al., 2014).However, the AOP for chemical-induced craniofacial anomalies has not been fully described. In the last decade, zebrafish has been considered as a promising model of an alternative method of teratogenicity testing (Bambino and Chu, 2017;He et al., 2014;Nishimura et al., 2016;Teixid o et al., 2019;Yamashita et al., 2014).Zebrafish has several experimental advantages for high-throughput genetic and chemical screening, including evolutionarily conserved developmental programs, various tools for genetic manipulation and rapid external development during the embryonic stage.Accumulating evidence has indicated that the zebrafish model has a high predictive ability for teratogenicity (Augustine-Rauch et al., 2016;Brannen et al., 2010;Inoue et al., 2016;Selderslaghs et al., 2012).Thus, the zebrafish model enables us to describe an AOP of developmental anomalies precisely and develop alternative methods for evaluating teratogenicity based on the AOP. Recently, by using zebrafish, we demonstrated that craniofacial anomalies caused by teratogens are associated with defects in neural crest cells (NCCs) and their derivatives.Furthermore, these craniofacial anomalies phenocopied neurocristopathy which is a pathology in human caused by the defects in development, migration, or differentiation of NCCs (Liu et al., 2020;Narumi et al., 2020).These results strongly suggest that the AOP of craniofacial teratogenicity is conserved between zebrafish and mammals.Thus, we hypothesized that the developmental process of NCCs would include key events for describing the AOP of teratogenicity.To examine the behavior of NCCs, visualization of these cells with a genetic marker is crucial.sox10 has been characterized as a common neural crest marker in vertebrates, including mammals and fish (Aoki et al., 2003;Carney et al., 2006;McKeown et al., 2005;Meulemans and Bronner-Fraser, 2004;Simões-Costa and Bronner, 2015;Southard-Smith et al., 1998).Thus, we expected that we can monitor NCC development with transgenic zebrafish lines expressing a fluorescent protein driven by the sox10 promoter. Here, we developed a series of transgenic zebrafish lines, sox10:EGFP, sox10:EGFP-CAAX, and sox10:Dendra2, which visualize NCCs.In these lines, the fluorescence demarcated cranial neural crest cells (CNCCs), which migrate from the neural tube along the stereotypical pathway to form the first pharyngeal arch (PA1) and differentiate into craniofacial cartilages such as the ethmoid plate (zebrafish palate) and upper and lower jaws.We administered teratogens known to cause craniofacial anomalies in mammals to the transgenic embryos, and examined the teratogenic effect on CNCC development and craniofacial development.All of the teratogens induced craniofacial anomalies such as cleft palate and micrognathia at 96 hours post-fertilization (hpf), as highlighted by the EGFP fluorescence.We further found that impaired migration of CNCCs and PA1 formation during the first 24 h of development are the early major phenotypes associated these later craniofacial anomalies.Considering that zebrafish embryos around these developmental stages have similar morphological and molecular characteristics compared with mammalian embryos, we postulate that the CNCC migration period is critical for inducing craniofacial anomalies across vertebrates.Based on these results, we propose the AOP of craniofacial anomalies centered by CNCC development.Furthermore, sox10 transgenic lines are ideal for an AOP-based teratogenicity model for evaluating and predicting craniofacial anomalies. Egg production and embryo exposure Adult male and female zebrafish (4-10 months after fertilization) were placed in a breeding tank with a separator in the late afternoon of the day before spawning.The separator was removed in the morning and spawning was stimulated when the light was turned on.Fertilized eggs were collected within 1 h after removal of the separator.The eggs were incubated in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 mM MgSO 4 , 0.1 mM NaOH) at 28 C and exposed to test compounds at 4 hpf.The exposure medium was replaced daily and samples were collected from the 5-6 somite stage (ss) until 96 hpf. For time-lapse imaging, samples were pretreated with VPA (30 mM) for 30 min prior to sample embedding.The samples were anesthetized with 0.02% MS-222 and embedded in 1% lowmelting agarose containing 0.02% MS-222 and the teratogens on a 4-chambered glass bottom dish (Greiner).Subsequently, the E3 medium containing MS-222 and teratogens at the same concentrations as above was additionally applied onto the samples. Transgenesis The Tol2 transposon system was used for the transgenesis (Suster et al., 2009).The promoter region of sox10 was isolated from RW strain.The promoter region located in the genome 5.0 kb upstream from the translation initiation site was amplified by PCR.The isolated promoter region and EGFP, EGFP-CAAX, or Dendra2 were cloned into pT2AL200R150G by In-Fusion cloning (Takara) to load a transposon cassette according to the manufacturer's protocol.For Tol2 transposase synthesis, pCS-zT2TP was linearized by Not I restriction enzyme (Takara) and Tol2 transposase mRNA was synthesized using the mMESSAGE mMACHINE SP6 Transcription Kit (Invitrogen).Each transposon construct and Tol2 transposase mRNA were coinjected at the 1-cell embryo stage. Lineage tracing For lineage tracing of neural crest cells, Tg(-5.0sox10:Dendra2) was used.The samples were anesthetized with 0.02% MS-222 and embedded in 1% low-melting agarose containing 0.02% MS-222.To photoconvert the frontonasal region, maxillary region and mandibular region in Tg(-5.0sox10:Dendra2) at 10 ss and 24 hpf, each ROI (region of interest) was selected using ZEN blue software on a Zeiss LSM800 confocal microscope and exposed to 10% UV (405 nm) laser for 30 s. Successful photolabeling was confirmed by specific photoconversion of Dendra2 from green to red in each ROI. Quantification of CNCC migration, proliferation, and apoptosis To quantitatively assess the migration of CNCCs, we measured the linear distance between the midbrain-hindbrain boundary (MHB) and the anterior border of the CNCC population migrating via the frontonasal pathway at 10 ss (for details, see Figure 3A) using Fiji measurement plugin (National Institutes of Health).To quantify mitotic CNCCs, we counted the number of the cells that displayed positive signals for both pH3 and EGFP anterior to the MHB at 10 ss (magenta region in Figure 6A) or within the PA1 at 24 hpf.Similarly, for quantifying apoptotic CNCCs, we counted the number of the cells that displayed double-positive signals for both active-Caspase 3 and EGFP in the same regions as above.In both experiments, we used z-stack images obtained by confocal microscopy and counted the number of mitotic and apoptotic cells distributed on the left side of the embryos. Statistics Multiple comparison tests were performed using Graph Pad Prism version 8 software for Windows (La Jolla).p-values were calculated using a 1-way ANOVA followed by Dunnett's multiple comparison tests for quantification of the migration distance of the pH 3-or Active Cas3-positive CNCCs.p-values < .05were considered statistically significant.All data are presented as the mean 6 SD unless otherwise specified. Next, we examined whether these lines are useful for detecting craniofacial anomalies (Figure 2).For this, we tested them with 5 teratogens (VPA, WAF, SA, CAF, and MTX).These teratogens are known to cause craniofacial defects in mammals and zebrafish (Liu et al., 2020;Narumi et al., 2020).To visualize cartilage, we immunostained treated embryos using both an anti-coll2 antibody and PNA lectin, which recapitulates Alcian blue staining (Figs. 2A-C), and compared the fluorescence pattern with EGFP expression pattern.All 5 teratogens induced craniofacial defects in cartilage at 96 hpf, such as cleft palate and small jaws (micrognathia), as reported in our previous studies (Figs.2D-I; Liu et al., 2020;Narumi et al., 2020).The craniofacial anomalies showed similar patterns among the teratogens; however, the phenotypic severity was dependent on the teratogen.Importantly, the EGFP patterns mostly overlapped with the cartilage staining in treated embryos, and thus these defects were also detected by EGFP expression (Figs. 2D 0 -I 00 ). Taken together, our results show that sox10:EGFP can visualize the anomalies of the craniofacial structure as precisely as the conventional cartilage staining.The same craniofacial defects were observed in sox10:EGFP-CAAX and sox10:Dendra2 embryos (Supplementary Figs. 2 and 3).Remarkably, the transgenic lines have a great advantage that every key event of CNCC development in an individual embryo can be traced by live imaging, starting from the emergence of CNCCs, which is essential to identify AOPs.Using these transgenic lines, we will describe possible key events for the CNCC-related AOP. Inhibitory effect of the teratogens on neural crest cell migration Our previous study suggested that craniofacial defects induced by teratogens are consequences of their adverse effects on CNCC development (Liu et al., 2020;Narumi et al., 2020).To elucidate the key events that trigger craniofacial malformation, we focused on the earliest phase of CNCC development, when the cartilages are yet to be formed.We focused on the migration period and analyzed the effects of the teratogens on migratory CNCCs by live imaging of sox10:EGFP embryos at 10 ss.In control embryos, CNCCs migrated from the midbrain and anterior hindbrain region at 10 ss toward the dorsal side to the eye along the frontonasal pathway or toward the pharyngeal region along the maxillary pathway.However, CNCC migration via both pathways was inhibited by all 5 teratogens tested (Figs.3B-F 0 ).To quantify inhibition of cell migration, we focused on the CNCCs migrating via the frontonasal pathway and measured the distance between the MHB (yellow arrowheads in Figure 3) and the anterior border of the migratory CNCCs (white arrowheads in Figure 3).The migration distance of migratory CNCCs was significantly reduced in the teratogen-treated groups (Figure 3G).Furthermore, all teratogens exhibited dose dependency in their inhibitory effects on CNCC migration (Supplementary Figs.4A-E).Indeed, time-lapse imaging of zebrafish embryos treated with a higher dose of VPA, as an example, revealed a pronounced inhibition of CNCC migration (Supplementary Movies 1 and 2).Interestingly, there was a correlation between the extent of the inhibitory effect on migration and the morphological severity of craniofacial anomalies, ie, cleft palate and micrognathia, at 96 hpf (Figs.2D-I 00 ).Regarding MTX treatment, however, the migration phenotype was milder compared with the morphological phenotype, especially for micrognathia, at 96 hpf (Figs.2I-I 00 ) (Liu et al., 2020). We have previously reported that the expression level of sox10 was disturbed by teratogen exposure, as determined by RT-PCR analysis (Liu et al., 2020).Consistently, as shown in Supplementary Figure 4, the EGFP signal tends to decrease as the concentration of teratogens increase.In spite of this, the intensity of the fluorescence signals at high-dose concentrations was sufficient enough for us to investigate CNCC behavior. These results suggest that a defect in CNCC migration is the earliest key event caused by the teratogens and that this inhibitory effect on CNCC migration account for disruption of or abnormal CNCC development in the later events. Developmental fate of the CNCCs in the PA1 and frontonasal prominence Most of the above affected migratory CNCCs are known to colonize the PA1 under normal conditions.To relate the migratory defects with late craniofacial anomalies, we examined the fate of CNCCs in the PA1 NCCs at later stages by lineage tracing.For this, we utilized a sox10:Dendra2 transgenic line (Figure 4A and Supplementary Figs.1C-D 0 ).The photoconvertible fluorescent protein Dendra2 irreversibly changes color from green to red upon photoactivation by UV light (405 nm), and thus a specific CNCC population can be labeled and traced chronologically (Figure 4A). We first labeled the maxillary prominence of the PA1 at 24 hpf and analyzed their descendants at 48 and 72 hpf (Figs.4B-B 000 ).The maxillary prominence eventually became the lateral part of the palate (trabeculae) and pterygoid process (upper jaw) (Figure 4B 000 ). Next, we performed the same analysis for the mandibular prominence of the PA1 (Figs. 4C-C 000 ).The mandibular prominence differentiated into the Meckel's cartilage (lower jaw) (Figure 4C 000 ). Furthermore, when we labeled the frontonasal prominence at 24 hpf (Figure 4D), the labeled cells gradually occupied the medial part of the palate at 48 and 72 hpf (Figs.4D 00 and 4D 000 ). Thus, the transgenic lines were able to identify the morphological changes that lead to the craniofacial anomalies (Figs.4E-E 00 ), and revealed the lineage-relationship between CNCCs in the PA1 and later craniofacial skeletons. Morphological changes in the PA1 are correlated with craniofacial anomalies in sox10:EGFP embryos To obtain the direct relationship between PA1 and teratogeninduced craniofacial phenotypes, we examined how craniofacial anomalies emerge by focusing on the PA1 formation period and outgrowth period.We treated sox10:EGFP embryos with the teratogens and performed live-imaging analysis of the PA1 morphology at 24 hpf, craniofacial placode formation at 48 hpf and craniofacial morphology at 72 hpf (Figs.5A-R).Teratogen-exposed sox10:EGFP embryos displayed PA1 defects such as truncated maxillary and mandibular prominences at 24 hpf (Figs.5A-F 0 ).At 48 hpf, the primitive palate (white arrowhead) and lower jaw (yellow arrowhead) were disrupted by the teratogens, leading to hypoplasia or aplasia of the structures (Figs.5G-L).At 72 hpf, the treated embryos exhibited the palate hypoplasia with a cleft (cleft palate) and shortening in Meckel's cartilage (micrognathia) (Figs.5M-R).WA-, SA-, and CAF-treated embryos showed mild phenotypes, namely, hypoplasia of the palate and mandible, whereas VPA-and MTX-treated embryos showed more severe phenotypes, namely, aplasia of the palate and mandible.These results are in agreement with the severity of craniofacial defects in each teratogen (Figs. 2 and 3). Liu et al. \| 43 Taken together, our data show that the teratogenic effects observed at 96 hpf originated from aberrant morphological changes observed during the migration (10-21 ss.) and PA1 formation (24 hpf) periods. Altered cell proliferation and apoptosis in the CNCCs during the migration period What cellular defects underlined teratogen-induced craniofacial anomalies?Proper regulation of cell proliferation and death is crucial in neural crest development during vertebrate embryogenesis (Kee and Bronner-Fraser, 2005;Kulesa et al., 2004).Thus, we hypothesized that teratogen-induced disruption of cell proliferation and apoptosis partially account for the observed migration defects of CNCCs during the migration period.These defects could be key events for the PA1 dysplasia, finally leading to craniofacial anomalies as an adverse outcome.We thus examined cell proliferation and apoptosis in premigratory and migratory CNCCs during the first 24 h of development. To precisely locate premigratory and migratory CNCCs at 10 ss which are to populate the PA1, we again performed the lineage tracing experiments using sox10:Dendra2 embryos.Because the CNCCs positioned anterior to the MHB migrated to the craniofacial regions by 24 hpf (Figure 3), we reasoned that the CNCCs The above CNCC populations at 10 ss and in the PA1 at 24 hpf were subjected to cell proliferation assay, ie, immunofluorescence staining against pH3 as a mitotic marker (Figs. 6C-N 0 ).Quantification of the number of the pH3-positive premigratory and migratory CNCCs revealed that the mitotic activity was significantly decreased by treatment with any of the teratogens at 10 ss (Figure 6O).Moreover, the mitotic activity at 10 ss exhibited dose-dependent decrease following treatment with all the teratogens (Figs.6Q-U).These defects in migration behavior and mitotic activity account for the subsequent key event, PA1 formation (Figure 6P). Next, we examined apoptosis by active Caspase-3 (Cas-3) antibody staining (Figs. 7A-L 0 ).Quantification of the number of active Cas-3-positive CNCCs indicated that apoptotic cells significantly increased at 10 ss (Figure 7M).Moreover, the number of apoptotic cells exhibited a dose-dependent increase following treatment of all teratogens at 10 ss (Figs.7O-S).However, at the highest dose of VPA and WA, embryos did not show a significant increase in apoptotic cells (Figs. 7O and 7P).One reason for this result could be that excessive cell death was induced at earlier stages by the high-dosage teratogen treatment.In contrast, apoptosis of CNCCs in the PA1 at 24 hpf was not significantly induced by the teratogen exposure (Figure 7N).Thus, the number of apoptotic CNCCs in premigratory and migratory CNCCs was in agreement with the phenotypic severity at later stages. Collectively, the data indicated that decreased mitotic ability and increased apoptosis in CNCCs during the migration period could be major key events for teratogen-indued craniofacial anomalies via PA1 dysplasia. Discussion In the present study, we established transgenic lines driven by the sox10 promoter to visualize CNCCs and examined their cellular behaviors in detail to identify an AOP for chemical-induced craniofacial anomalies.Teratogens affected CNCC migration, mitotic ability, and apoptosis, leading to the PA1 morphological defect.These defects finally resulted in craniofacial anomalies such as cleft palate and micrognathia as adverse outcomes.Our lineage tracing experiments showed that, like in mammals, CNCCs in the zebrafish PA1 differentiated into craniofacial skeletal elements.Thus, the AOP of craniofacial anomalies could be conserved between fish and mammal. The craniofacial morphology of mammals is apparently different from that of zebrafish at the late embryonic (fetal) stages; however, the pharyngeal period (24 hpf for zebrafish and E9.5 for mouse), when PAs are being formed, is conserved at the morphological and transcriptome level among vertebrates (Irie and Kuratani, 2011;Kuratani, 2005).During this period, one of the crucial events is the development of NCCs (Etchevers et al., 2019;Hockman et al., 2019;Simões-Costa and Bronner, 2013).CNCCs originate from the midbrain and anterior hindbrain region during neurulation (Chai et al., 2000;Rocha et al., 2020;Figs. 1B and 1B 0 ). Similarities between fish and mammalian craniofacial development are also observed even after the pharyngeal period.In mammals, one of the critical events causing craniofacial anomalies is a failure of the primary and secondary palate formation (palatogenesis) during late craniofacial morphogenesis (E11.5 for mouse embryos).The primary palate develops from the frontonasal prominence and is transiently formed anterior to the secondary plate (Figure 8A).The secondary palate develops from the maxillary prominence and is finally fused with the primary palate to complete the process of craniofacial morphogenesis (Bush and Jiang, 2012).Any disturbance of this process in mammals can cause orofacial clefts (Hammond and Dixon, 2022;Lan et al., 2015).In zebrafish, in spite of the absence of the secondary palate, the palate is also formed in the anterior neurocranium and is composed of the ethmoid plate and trabeculae, which originate from the frontonasal and maxillary prominences, respectively (Figure 8A).Clefting, shortening, or the absence of these structures represent prototypical orofacial abnormalities in zebrafish (Dougherty et al., 2012;Eberhart et al., 2008;Swartz et al., 2011;Wada et al., 2005).Thus, palatogenesis following the PA1 formation in zebrafish and mammals is thought to progress in similar manners. In addition to the morphological similarity, zebrafish and mammals share similar teratogenic responses (Liu et al., 2020).Furthermore, they utilize the same molecular signaling pathways, including Fgf, Pdgfr, Bmp, Tgfb, Wnt, and Shh pathways, in palatogenesis, and disruption of any of these pathways could result in craniofacial defects in both zebrafish and mammals (Bush and Jiang, 2012;Schilling and Kimmel, 1997).Indeed, in our previous research, cleft palate in zebrafish induced by teratogens was triggered by canonical Wnt pathway inhibition, which is the same etiology of cleft palate as that found in mammals (Narumi et al., 2020).These lines of evidence support the idea that zebrafish share conserved craniofacial morphogenesis and teratogenic responses with mammals.Therefore, zebrafish is considered to be a useful model for evaluating and predicting chemicalinduced craniofacial anomalies across vertebrates. Identification of an AOP has been essential for developmental toxicity evaluation as well as cross-species extrapolations (Ankley et al., 2010;Knapen et al., 2015).Recently, an AOP of developmental vascular toxicity was identified (Saili et al., 2017); however, there is no report of an AOP causing craniofacial anomalies.Developmental toxicity manifests during morphogenesis, and thus teratogenicity should be monitored and evaluated in a chronological manner.Craniofacial anomalies have been investigated mainly by observation of gross external morphology or the pattern of Alcian blue cartilage staining (Brannen et al., 2010(Brannen et al., , 2013;;Liu et al., 2020).Although these strategies have been successful in detecting the readout of teratogenicity, the developmental process for the craniofacial anomalies could not be traced with these methods.To solve this problem, we established a series of sox10-reporter transgenic lines in zebrafish, which enables us to monitor for CNCC-based chronological teratogenic response and thereby to elucidate key events along with the causal relationship inducing craniofacial anomalies (Table 1).Several genes including foxd3 have been identified as markers for CNCCs (Drerup et al., 2009).Among these markers, sox10 is the most suitable in that it continues to be expressed in craniofacial cartilage after 48 hpf (Carroll et al., 2020), and thus can monitor both CNCC development and craniofacial morphogenesis. Our sox10:EGFP zebrafish line enables us to monitor the gross morphology of CNCC populations during craniofacial morphogenesis (Figs.1A and 1A 0 ).The sox10:EGFP-CAAX line demarks the cell membrane of each CNCC (Supplementary Figs.1B and 1B 0 ) so 1C and 1C 0 ).These lines can also detect cartilage anomalies at later stages (Figure 2 and Supplementary Figs. 2 and 3).The expression patterns of these fluorescent reporter genes are in agreement with the previously reported sox10 reporter lines (Askary et al., 2015;Dougherty et al., 2012).Therefore, our lines are suitable for examining the early events of teratogenicity such as migration and apoptosis through live imaging and monitoring the development of CNCCs.Gene expression analysis will be needed to further confirm the usefulness of this system, but it will be a next step to be done in the near future.Taken together, these transgenic lines provide the powerful platform with which to analyze the early events causing craniofacial anomalies and to evaluate AOP-based teratogenicity chronologically. CNCC development is classified into 3 periods: CNCC migration, PA1 formation and outgrowth of craniofacial placodes.Live-imaging with sox10:EGFP lines detected defects in CNCC migration, PA1 formation, and craniofacial morphology as an adverse outcome.When exposed to teratogens, premigratory and migratory CNCCs at 10 ss exhibited decreased proliferation and increased apoptosis, but CNCCs in the PA1during the pharyngeal only showed decreased proliferation at 24 hpf.These results suggest that migratory CNCCs are more vulnerable and/or sensitive to teratogens.Consistent with these results, when zebrafish embryos were treated with the teratogens for 4-24 hpf or 24-96 hpf, craniofacial anomalies occurred more frequently in the 4-24 hpf treatment than in the 24-96 hpf treatment embryos (Supplementary Figure 5).Taken together, we conclude that the migration period is critical for chemical-induced craniofacial anomalies and that the resulting PA1 morphology at the pharyngeal period 24 hpf is a reliable endpoint for predicting chemical-induced craniofacial anomalies. Finally, we provide the following AOP for chemical-induced craniofacial anomalies (Figure 8B): 1. Molecular response: Disturbance of gene expression during CNCC development Our previous study showed that teratogen-exposed embryos showed disturbance of the expression of genes involved in CNCC development (tfap2a, zic2a, pax3, snail2, sox10, and sox9a) (Liu et al., 2020).Genetic manipulation of some of these genes also induces defects in CNCCs leading to craniofacial anomalies in zebrafish (Knight et al., 2003;McKeown et al., 2005;Minchin and Hughes, 2008;TeSlaa et al., 2013;Yan et al., 2005).2. Cellular response: Inhibition of CNCC migration, decreased mitotic ability, and increased apoptosis in premigratory and migratory CNCCs 3. Tissue response: Disturbance of PA1 morphogenesis at the pharyngeal stage 4. Adverse outcomes: Craniofacial anomalies such as cleft palate and micrognathia Taken together, we propose a method for an alternative teratogenicity assay using transgenic zebrafish lines (Figure 8C).Teratogens are administered to sox10 transgenic lines at 4 hpf.The first endpoint is the PA1 formation at the pharyngeal period.This deformity predicts later craniofacial anomalies.The second endpoint is craniofacial morphology at 96 hpf, which is a readout of the PA1 defect. In sum, we have revealed the AOP for craniofacial anomalies caused by teratogens using zebrafish transgenic lines and identified cellular key events.Based on highly conserved developmental mechanisms and teratogenic responses between mammals and fish, we assume that our findings can be directly applicable to mammalian embryos.Zebrafish is thus a promising model for evaluating cross-species chemical-induced developmental toxicity. 0 ).The migratory CNCCs via the frontonasal pathway reached the ventral side of the eye at around the 21 ss (Figure1E0 black arrow).The migratory CNCCs expressing EGFP reached the ventral side of the pouches and began to form the pharyngeal arches (PAs) at 24 hpf (Figs.1F and 1F 0 ).The PA1, which give rise to a large part of the craniofacial bones, were EGFPpositive at this stage (Figure1F 0).The migratory CNCCs via the frontonasal pathway moved toward the PA1 (Figure1F0 black arrow).sox10:EGFP continuously visualized the developmental process of CNCCs and craniofacial morphogenesis from 48 to 96 hpf (Figs.1G-I 0).The EGFP-expressing CNCCs were present in the developing craniofacial skeletal primordium; trabeculae (developing lateral part of the ethmoid plate), median element (developing medial part of the ethmoid plate) derived from frontonasal process, and Meckel's cartilage (lower jaw) at 48 hpf (Figs.1G and 1G 0 ).At this stage, the EGFP-expressing CNCCs were also present in olfactory placodes and gonadotropin-releasing hormone cells(Figs.1G and 1G 0 ).The skeletal primordium differentiated into the ethmoid plate (zebrafish palate) consisting of the trabeculae and median elements and Meckel's cartilage at 72 hpf (Figs.1H and 1H 0 ).These structures grew and each skeletal element was firmly formed at 96 hpf (Figs.1I and 1I 0 ). Figure 4 . Figure 4. Lineage tracing of CNCCs in PA1 and frontonasal prominence using sox10:Dendra2.(A) Photoconversion was performed in sox10:Dendra2 transgenic zebrafish.Photoconvertible fluorescent protein Dendra2 is photoactivated by UV light (405 nm) from green to red.Specific Dendra2expressing cells were labeled irreversibly and traced.(B-D 000 ) Lineage tracing of the PA1 components: maxillary prominence (B, B 0 ), mandibular prominence (C, C 0 ), and frontonasal prominence (D, D 0 ).Magnified view of the photoconverted each prominence was displayed in panel B 0 , C 0 , and D 0 .(B-B 000 ) The maxillary prominence was labeled at 24 hpf and differentiated into lateral part of the ethmoid plate (zebrafish palate) and the pterygoid process (ptp), which is the upper jaw at 72 hpf.(C-C 000 ) The mandibular prominence was labeled at 24 hpf and differentiated into Meckel's cartilage of the lower jaw.(D-D 000 ) The frontonasal prominence was labeled at 24 hpf and differentiated into the medial part of the palate at 72 hpf.(E-E 00 ) The schematic diagram of the lineage tracing of the PA1 and frontonasal prominence.Each color in the panel E represents the following: yellow indicates the photoconverted region of the maxillary prominence, orange indicates the photoconverted region of the mandibular prominence, and blue indicates the photoconverted region of the frontonasal prominence.ch, ceratohyal; e, eye; gc, gill cartilages; m, Meckel's cartilage; mx, maxillary; p, palate; ptp, pterygoid process; tr, trabeculae.Scale bar: 100 mm. Figure 5 . Figure 5. Changes in the morphology of craniofacial anomalies in teratogen-treated sox10:EGFP embryos.(A-F) Lateral live view of the first pharyngal arch (PA1) at 24 hpf.(A 0 -F 0 ) Enlarged view of the white rectangle in panel A-F.Abnormal morphology of the PA1 was observed in the teratogen-treated embryos.(G-L) The developing palates (white arrowheads) and mandibles (yellow arrowheads) were disrupted in the teratogen-treated embryos at 48 hpf.(M-R) The PA1 defects resulted in a small-sized palate with clefting (cleft palate) and shortening in the Meckel's cartilage at 72 hpf.e: eye.Scale bar: 100 mm. Figure 8 . Figure 8. Conserved craniofacial morphogenesis during the pharyngeal stage and the AOP of craniofacial anomalies identified in the present study.(A) Schematic comparison of craniofacial morphogenesis between mammals (mouse) and zebrafish.Zebrafish and mammals share conserved craniofacial morphogenesis.At the pharyngeal stage, the frontonasal prominence (FNP), maxillary prominence (MxP) and mandibular prominence (MdP) are formed in both species.FNP and each region in the PA1 differentiate into craniofacial elements.(B) Identified AOP of chemical-induced craniofacial anomalies.(C) AOP-based teratogenicity assay utilizing the sox10 transgenic lines.PA1 morphology at 24 hpf and craniofacial morphology at 96 hpf are practical endpoints of teratogenicity evaluation and prediction. Table 1 . Established transgenic lines Three transgenic lines were established.Each line showed distinct fluorescence localization and utility for a specific purpose.	2023-08-03T15:21:22.122Z	2023-08-02T00:00:00.000	{ "year": 2023, "sha1": "403a5700ef62c792042cc99ff28ed88aad54f7b4", "oa_license": "CCBY", "oa_url": "https://academic.oup.com/toxsci/advance-article-pdf/doi/10.1093/toxsci/kfad078/51024160/kfad078.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "ca80f298b25f8a642b3e419d889c626d8d77397d", "s2fieldsofstudy": [ "Biology", "Environmental Science", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
187796341	pes2o/s2orc	v3-fos-license	Current Knowledge of TB Patients on Initiation of Treatment in Mettu Town Background: Knowledge on causes & initiation time of TB treatment of the community is critical and improve or worsen the transmission of the disease and treatment outcomes. Lack of awareness about the disease contribute to how compliance to chemotherapy and preventive measures of the disease. The objectives of this study were: To assess timeline of initiation of the treatment and current knowledge of patients in Mettu Town. A facility based descriptive cross-sectional study was conducted from April 28/2016 –May 29/2016 and all patients attending Mettu Karl hospital and Mettu health center were included. Data was compiled and analyzed by using Statistical package for Social Sciences software (SPSS) version 20.0. Statistical significance was defined at a level of 0.05 and data was described with a confidence interval of 95%. Results: Forty-six (52.87%) of respondents were female and 48.5% of the patients belong to age group of 15-29. Fifty-four (62.07%) of respondents residing in Urban setting and 54.02% were Oromo and 43.68% of respondents can read and write. Thirty-three (37.93%) of respondents recognized the symptoms one month before they presented to health facility for the first time. Conclusion and Recommendation: Significant numbers of the respondents were not knowledgeable and delayed in the initiation of the treatment. Appropriate intervention should be taken to reverse this problem. Open Access Background TB is a common and deadly infectious disease caused usually by mycobacterium tuberculosis in human. About 10% of latent infection eventually progress to active TB disease which, if left untreated kills about half of those infected. Once person develop active tuberculosis diseases, the symptoms of this disease may be mild for months and this leads to delays in seeking care and results in the transmission of bacteria from infected person to the others. Lack of understanding about tuberculosis, stigma toward the disease, inaccessibility of treatment is the main reason for delayed presentation; while lack of information and dissatisfaction of the treatment are the main reason of non-completion of the treatment. Perceived knowledge on the causes of tuberculosis and health seeking behavior among community members may reduce or increase the disease transmission. The spread of the disease increases due to failure to recognize the symptoms early, hence theses may delay the diagnosis of tuberculosis and treatment [1][2][3][4][5]. TB is re-emerging as a public health concern because of the low practices of peoples to prevent the transmission of TB, inadequate health coverage, impact of HIV epidemic and negative attitude of patients on its transmission and treatment. Countries with high particularly sub Saharan Africa have witnessed profound increase in the number of tuberculosis cases [6][7][8][9]. In Ethiopia TB is major public health problem with increase rate of TB cases of 2.6% each year. Among these 97.1% are new cases of all forms TB. Smear positive cases account 32% and smear negative and extra pulmonary TB are 34% and 32% respectively. Lack of adequate knowledge about TB and noncompliance to treatment result in treatment failure. The likelihood of successful treatment of tuberculosis depends on the extent to which patient complete the pressured treatment regimen (adherence and noncompliance) [10][11][12]. Better understanding of level of the knowledge and time of initiation of the treatment about TB patient help for care provider and administrate to identity the gap and to take relevant action. Therefore, the purpose of this study was to assess timely initiation of the treatment and their current knowledge toward TB for patient on treatment in Mettu Town. General objectives To assess the initiation time of the treatment and current knowledge of patients in Mettu town Specific Objective a. To assess the time line of initiation of the treatment of tuberculosis patient b. To determine current knowledge of tuberculosis patient c. To explore factors affecting time of initiation of TB treatment Study Area The study was conducted in health facilities of Mettu town, from April 2016 to May 2016. Mettu town is found in south west Ethiopia, in Ilu Aba Bor zone 600km away from Addis Ababa, the capital of the country. Sample Size Determination All TB patient attending Mettu Karl Hospital and Mettu health center during the study period was included as convenient sample size. Data Collection Procedure Standard self-administered questionnaire and interview was developed in English and then changed to Afaan Oromo (Oromo Language) and then to English language). Questionnaires were further modified from guidelines and previous study before actual data collection. Data collection was face to face interview by principal investigator as it gets direct information from respondents. Data Quality Assurance Measures The appropriately designed data collection instrument was used. Every day the collected data was reviewed and checked for completeness and consistently of the respondents. Pre-test was done on 5% of total population out of the study and necessary correction was made on the language clarity sequencing and workability of questionnaires. Data Analysis The collected data was analyzed manually after it was edited and checks for completeness and consistency, for further analysis, it was analyzed using SPSS. Descriptive analysis was used of describe the percentage of knowledge of TB patient and treatment initiation of TB patient to the area. Frequency distribution was using for qualitative data. Ethical Consideration Ethical clearance was obtained from ethical review board of Mattu university, college of public health and medical science department of nursing. Letter of permission present to Mattu Karl hospital and health center TB follow up clinic and verbal informed consent was obtained from each study is explain to respondent. Confidentiality of information was assured, and privacy of the respondent was maintained. Limitation of the Study Since the patients on follow up attend health center by appointment some TB patients were not involved in the study during the study period. Shortage of time of data collection and lake of similar literature in the study area to compare and contrast the result Time Lines of Treatment About timelines of treatment initiation, the time when patients recognized the first symptoms, frequency of health facility visit before diagnosis, delays for diagnosis, and delays for treatment initiation and treatment phase were interviewed. Concerning of the time when they recognized first symptoms, thirty-three (37.93%) of respondents recognized the symptoms one month before they presented to health facility for the first time. Fourth eight (55.17%) of respondents visit health facility only once before diagnosis of TB, 37(42.53%) were diagnosed for TB within a week after the first visit and 62.07% were start treatment immediately after diagnosis of TB. Of total of respondents, 62.07% were in the continuation phase (Table 1). Knowledge about TB Concerning knowledge of respondent's questions about causes, signs and symptoms, transmission, curability, side effects of anti TB drugs, prevention and importance of completion of the treatment were asked. About cause of TB 47 (54.02%) of respondents were respond that TB caused by bacteria (Figure 1). Concerning perceived knowledge about sign and symptoms, transmission, curability, anti TB drugs treatment, prevention of TB was also asked and analyzed. Eighty-one 93.1% of respondents were respond as they know TB can transmitted and how it can be transmitted. Concerning anti TB drugs treatment 85.06% of respondents were know as they may die if they don't treat and 91.95% of them know anti TB treatment was taken for 6 months. All the respondents had been taking the drugs once per day. About preventability of TB 86.21% know as TB is preventable disease. To sum up 13 knowledge-based question were asked and 82.76% were answered the question above mean (knowledgeable) and 17.24% were not knowledgeable ( Table 2). Factor Affecting Time of Initiation of TB Drug The adjusted odd ratio (AOR) was calculated to identify associated factors and their significant. The analyzed risk factors are sex, age, patient residence, marital status, occupation and educational status of the Patients. Age of the patients, marital status and occupation had no such significant effect on knowledge and TB treatment dalliance. Sex, patient living, and educational status are significantly affecting patient TB treatment knowledge and they take lion's share for dalliance of TB diagnosis and treatment. The odds of Females delayed from seeking TB treatment is 2.4(1.9-2.10) times than male in Mettu town with P-value of 0.003. Patients residing rural areas are delayed from seeking TB diagnosis and treatment 4.00(2. 57-6.45) times than urban resident patients with p-value of 0.001. Illiterate patients were 10.76(5.019-13.09) times delayed from TB diagnosis and treatment than educated patient up to grade ten and above, with significant values (P-0.001) (Tables 2 & 3). Discussion This study provided information regarding TB patients, timelines of treatment initiation, knowledge about tuberculosis and its treatment and factor affecting time of initiation of TB drug. In this study 88.5% of TB patients belong to economically productive age group (15-54years) which is slightly higher than the finding documented in WHO report which was 82% of cases. This difference might show that the finding of this study is recent and it was also done in specific area and my total population or respondents also not comparable with that study since WHO report is global study [13] and 54.02% of respondents responded that 5/6 bacteria as the cause of tuberculosis whereas study done in south west Ethiopia in 2010 showed that 33.7% of Tb suspected patients had knowledge on the cause the disease [14]. This study shows that 95.4% respondents mentioned that cough is one sign and symptom of TB which is slightly higher than the study done in Tanzania which is 72% of them know cough as one sign and symptom of TB [15]. Knowledge about route of the transmission of the disease is another important factor in TB prevention and control problem. In our study 93.1% of respondents responded as they know as TB can be transmitted, which is also slightly higher than study done in southwest Ethiopia which indicate 83.8% of patients have knowledge about route of transmission [16]. This difference is may be due to our study was conducted in the place where majority of study population were from urban area and they may had access of health information more. About curability of the disease 97.7% of respondents were know that as TB is curable disease and 91.95% of them know anti TB treatment was taken for 6 months. This finding was the same with study conducted in Tanzania in which all respondents knew that TB was curable disease [17]. In this study 86.21% of respondents knew that as Tb is preventable disease with preventive method. This is slightly higher than study conducted in south west Ethiopia in which 82.3% of respondents know as Tb can be prevented and 69.9% know that not coughing /sneezing in front of other people as the main preventive measure [18]. Sex, Patient Residence and Educational status are significantly affecting Patient TB treatment knowledge and are significant factors for dalliance of TB diagnosis and treatment Conclusion More than half of patient visited health facility only once before diagnosis. Majority of them started the treatment immediately after diagnosis and were in continuation phase [19]. Most of them know as Tb can be transmitted and coughing in front of people and using the same utensil were common made of transmission known by respondents. Almost all respondents know as TB is curable disease and requires 6 months treatment more than half know that as anti TB treatment has no side effect. Majority of respondents know as TB is preventable disease and almost all of respondents respond avoidant direct coughing in front of people as the common preventive methods [20]. Patients living rural areas and illiterate patients were delayed from seeking TB diagnosis and treatment. Recommendation Based on the finding of the study the following recommendation was given. The Mettu health center and Mettu Karl hospital responsible bodies should make an effort to implement planned and well-organized health information. As early diagnosis and treatment initiation is important in TB control, attention should be given in all health facility. The focus of health education for public should mainly on cause, route of transmission, side effects of anti TB treatment, sign and symptoms and preventive methods. In service training for health care provider on TB prevention and control should be strengthened to update their knowledge and skills. Further study should be done across the country on related topics. Focus should be given more for rural residence and un educated patients during health education	2019-03-28T13:33:23.161Z	2018-09-07T00:00:00.000	{ "year": 2018, "sha1": "9e2ba19c6bc844ef58489aeaf7ac36399bf08fb0", "oa_license": "CCBY", "oa_url": "https://biomedres.us/pdfs/BJSTR.MS.ID.001709.pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "f6bdee95566c59026a763ab8a8855d3482dadf7c", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
233279877	pes2o/s2orc	v3-fos-license	Bioinformatic analysis identifying FGF1 gene as a new prognostic indicator in clear cell Renal Cell Carcinoma Background Clear cell renal cell carcinoma (ccRCC) has been the commonest renal cell carcinoma (RCC). Although the disease classification, diagnosis and targeted therapy of RCC has been increasingly evolving attributing to the rapid development of current molecular pathology, the current clinical treatment situation is still challenging considering the comprehensive and progressively developing nature of malignant cancer. The study is to identify more potential responsible genes during the development of ccRCC using bioinformatic analysis, thus aiding more precise interpretation of the disease Methods Firstly, different cDNA expression profiles from Gene Expression Omnibus (GEO) online database were used to screen the abnormal differently expressed genes (DEGs) between ccRCC and normal renal tissues. Then, based on the protein–protein interaction network (PPI) of all DEGs, the module analysis was performed to scale down the potential genes, and further survival analysis assisted our proceeding to the next step for selecting a credible key gene. Thirdly, immunohistochemistry (IHC) and quantitative real-time PCR (QPCR) were conducted to validate the expression change of the key gene in ccRCC comparing to normal tissues, meanwhile the prognostic value was verified using TCGA clinical data. Lastly, the potential biological function of the gene and signaling mechanism of gene regulating ccRCC development was preliminary explored. Results Four cDNA expression profiles were picked from GEO database based on the number of containing sample cases, and a total of 192 DEGs, including 39 up-regulated and 153 down-regulated genes were shared in four profiles. Based on the DEGs PPI network, four function modules were identified highlighting a FGF1 gene involving PI3K-AKT signaling pathway which was shared in 3/4 modules. Further, both the IHC performed with ccRCC tissue microarray which contained 104 local samples and QPCR conducted using 30 different samples confirmed that FGF1 was aberrant lost in ccRCC. And Kaplan–Meier overall survival analysis revealed that FGF1 gene loss was related to worse ccRCC patients survival. Lastly, the pathological clinical features of FGF1 gene and the probable biological functions and signaling pathways it involved were analyzed using TCGA clinical data. Conclusions Using bioinformatic analysis, we revealed that FGF1 expression was aberrant lost in ccRCC which statistical significantly correlated with patients overall survival, and the gene’s clinical features and potential biological functions were also explored. However, more detailed experiments and clinical trials are needed to support its potential drug-target role in clinical medical use. Supplementary Information The online version contains supplementary material available at 10.1186/s12935-021-01917-9. Background Rising from renal tubular epithelial cells, renal cell carcinoma has been a common malignant tumor, which ranks only second to bladder carcinoma in adults urinary tract malignant tumors [1]. And within RCC, 65 ~ 70% is clear cell renal cell carcinoma (ccRCC), which possess specific microscopic appearance other than the other RCC subtypes. Attributing to the rapid development of molecular pathology, the classification and diagnosis of renal cell carcinoma have been increasingly evolving [2,3]. Currently, many molecular genetic abnormalities have been reported to exist in RCC, including chromosome number or structural abnormality, genes mutation, amplification or fusion genes resulting from chromosome translocation [4]. In ccRCC, the most classic molecular genetic characteristics are the changes of related genes on the short arm of chromosome 3 (3p), especially the VHL gene. The "first hit" of cancer usually comes from the change of VHL gene (gene mutation or promoter methylation), followed by "second hit"-3p chromosome deletion, which leads to tumor occurrence, and the 3p variation occurs in nearly 90% of ccRCC cases [5]. Besides the VHL gene, some other 3p gene variations have also been reported in ccRCC, for instance SETD2 [6] and BAP1 [7], whose mutation have been reported to be related with worse patients prognosis, as well as PBRM1 [8], which was associated with better patients survival. As for the clinical cure methods, besides the traditional surgery and stereotactic body radiation therapy, great improvements have been taking place in molecular targeted therapies. At present, 13 drugs in 6 categories have been approved for metastatic ccRCC, including VEGFR, mTORC1, c-Met and FGFR inhibition, as well as cytokines and anti PD-1/PD-L1 immune checkpoint inhibitors [9][10][11]. These drugs have been showing promising curative effects and increasing the median patients survival time from 15 to 30 months in the past 10 years [4]. However, the curative effective obviously vary among different individuals indicating the heterogeneity in drug mechanisms, tumor molecular genetic changes and host immune situations. It is of great importance to keep identifying new potential prognostic biomarkers as well as probable drug targeting genes thus aiding more precise understanding of the disease. Currently, with the gradual maturity and promotion of molecular pathological detection technologies, for instance tissue microarray, protein chip, next generation sequencing (NGS) and single cell sequencing which have been bringing in tremendous molecular data, it is more convenient for us to identify more potential disease-causing gene alterations and better understand the molecular basis of cancer development [12][13][14][15]. Gene Expression Omnibus(GEO) has been a widely used online cancer research database for providing highthroughput genes expression data submitted by research institutions all over the world. In the study, different GEO datasets were used to screen the differently expressed genes (DEGs) in ccRCC comparing to normal kidney tissues, followed by series of bioinformatic analysis, for instance protein-protein interacting (PPI) network construction, function modules analysis and Kaplan-Meier survival analysis to identify the key genes that potentially regulate ccRCC development. Further, local hospital patents samples were used to explore the potential clinical significance of the key gene. The results shall provide useful insights to the unearth of potential new prognostic biomarkers and drug targeting gene candidates for clinical ccRCC patients. Data source: cDNA expression profiles from GEO database From GEO online database [16], we picked four ccRCC cDNA expression profiles GSE53757 [17], GSE53000 [18], GSE71963 [19] and GSE68417 [20] based on the sample number (only the profiles that contain at least 40 samples covering both cancer and normal tissues were considered functions were also explored. However, more detailed experiments and clinical trials are needed to support its potential drug-target role in clinical medical use. Keywords: GEO database, Protein-protein interaction network (PPI), Clear cell renal cell carcinoma (ccRCC), FGF1 gene, PI3K-AKT signaling pathway, Molecular pathology Screen the DEGs in ccRCC comparing to normal renal tissues After the four cDNA expression profiles being downloaded from GEO database, GEO2R [21], which has been a widely used genes expression analyzing tool and commonly provided paired with GEO profiles online was used to screen the DEGs between ccRCC versus normal tissues. The criteria for DEGs identification were set as adjusted P value < 0.05 and \|log2FC\|≥ 2. Further, Venn diagram [22] was used to identify the DEGs that were shared in all four cDNA profiles followed by the shared DEGs' basic interpretation including their main biological processes, molecular functions and the signaling pathways they mainly enriched in using Gene ontology analysis (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) software [23]. DEGs PPI network construction and key genes identification To search the association between different genes, STRING [24], which is short for the Search Tool for the Retrieval of Interacting Genes was used to construct the PPI network among shared DEGs, and the construction criteria was set as confidence score ≥ 0.4 and maximum interactors number = 0. Followed the DEGs PPI network construction, Molecular Complex Detection (MCODE) plug-in of Cytoscape3.6.0 software [25] was used to analyze the gene function modules based on the network and the analysis cut-off values were set as degree = 2, node score = 0.2, k-core = 2, and max.depth = 100. Using MCODE analysis, we identified the top four gene module (gene clusters sharing similar function) and analyzed the signaling pathways module genes mainly enriched in, meanwhile, the genes that were shared in different modules indicating their probable connecting core genes' role in the network were highly focused, and the potential core gene's connectivity degree with surrounding genes was also validated by the Cytohubba plug-in of Cytoscape3.6.0 software. Kaplan-Meier survival analysis and clinical pathological features exploration Kaplan-Meier plotter [26], which contains a total of 54,000 genes in 21 types pan-cancers has been a widely used online service for assessing various genes' overall survival correlation. In the study, Kaplan-Meier plotter was used to validate the probable core gene's correlation with ccRCC patients survival and draw the survive curve. Meanwhile, UALCAN [27], which has been an openly accessed online service based on TCGA data was applied to explore gene's association with ccRCC clinical parameters. GEPIA and Oncomine gene expression analysis GEPIA [28] has been a commonly used online software for worldwide researchers to explore certain genes' expression and perform survival analysis in various cancers based on the sequencing databases of 9736 cancer and 8587 normal samples from TCGA and GTEx programs. In the study, GEPIA was used to preliminary explore the expression change of FGF1 gene in ccRCC comparing to normal renal tissues. Besides GEPIA, Oncomine database is also a widely used web-based data mining platform for genes expression analysis. In the study, we additionally used Oncomine to explore the FGF1 expression in broad spectrum human cancers. CcRCC tissue microarray production The ccRCC patients tissues used for microarray production were all collected from local hosptital surgeries at General Surgery Department and sent for pathology examination at our Pathology Department and then stored at Pathology Department Biobank. The Informed consent from the patients as well as the approval by the Hospital Institutional Board were both obtained (Second Hospital of ShanXi Medical University, China). Further, 104 ccRCC patients samples were picked from the biobank after HE staining confirmation of the disease diagnosis and evaluation of cancer percentage by two local hospital pathologists. Meanwhile, four areas in each sample (including two cancerous and two paracancerous normal areas. Two independent cancerous areas were used to eliminate tumor heterogeneity) were circled under microscope for further study and the receptor wax block were made with 1.5 mm needle according to operating instructions (Chloe, BeiJin, China). Further, tissue microarray was obtained by serial sectioning of the receptor wax block and stored at 4 °C refrigerator (Department of Pathology, Second Hospital of ShanXi Medical University, China). Regents and tissue samples IHC experiment was conducted using the ccRCC tissues microarray to validate the gene's expression difference between cancer and paracancerous normal renal tissues. And it was performed on VENTANA platform (Roche) in local hospital Pathology Department. The primary antibody of FGF1 gene was purchased from abcam (ab179455), and the secondary antibody (Envision /HRP kit) and DAB detection kit were from ZSBG-Bio. Other reagents including H2O2, antigen retrieval citrate solution, phosphate-buffered saline (PBS) and hematoxylin stain were from local hospital Supply Department. IHC experimental protocol The ccRCC tissue microarray slides were firstly taken out of 4 °C refrigerator and rewarmed at room temperature for 30 min. And then the slides were dewaxed and rehydrated with gradient ethanol followed by antigen retrieval using 10 mmol/l citrate solution. Meanwhile, to inhibit the activity of endogenous peroxidase, the slides were maintained in 0.3% H2O2 containing methanol for 20 min. Further, the slides were soaked in bovine serum albumin for 30 min and then incubated with primary FGF1 antibody (dilution 1:200) overnight at 4 °C followed by a 40 min secondary antibody incubation at 37 °C. Finally, the slides were processed with horseradish peroxidase (HRP) and visualized in DAB for results evaluation. IHC results evaluation The IHC result was evaluated based on both the microarray tissue cores' staining intensity and staining area which were scored by two experienced pathologists registered in local hospital Pathology Department with no prior information of the clinical or pathological details of the patients. The staining intensity was scored with the criteria set as: None (0), mild (1), moderate (2) and strong (3), meanwhile, the staining area was classified as: < 5% (0), 6-25% (1), 26-50% (2), 51-75% (3) and > 75% (4). The section's final score equals the multiplication of staining intensity and staining area, and the final result of each patient's cancer or paracancerous normal tissue was recorded as the average of two independent microarray cores' scores, and if the final score < 4, the result was defined as negative, meanwhile, if final score ≥ 4, the result was classified as positive. Quantitative real-time PCR (QPCR) experiments The total mRNA of 30 ccRCC cancer tissues and adjacent paracancerous normal renal tissues (independent of the 104 cases used for microarray production) were extracted using RNAiso-Plus (TAKARA, DaLian, China). And then1 μg extracted mRNA was used for cDNA synthesis with cDNA synthesis kit (TAKARA, DaLian, China) according to operating instruction. Further, qPCR was performed on Roche z 480 and the primers used were listed as below: The PCR cycling condition was set as: 95 °C 5 min for 1 cycle; 95 °C 5 s, 62 °C 30 s, and 72 °C 34 s for 35 cycles followed by the melting curve stage. And the relative gene expression in each sample was recorded as the average 2^ − ΔΔCT calculation result of three replicates. Further, T-test was used for detailed statistical analysis. P < 0.05 was considered statistically significant. Gene's physicochemical properties ProtParam [29] is a newly developed online software and it is commonly used for computing the physical and chemical parameters of certain proteins including their molecular weights, theoretical isoelectric point, aminoacid composition, extinction coefficient, estimated protein half life, protein instability index and grand average of hydrophilicity. Besides ProtParam, ProtScale [30] is also a widely used online service for computing the aminoacid scales on a selected protein, and the most frequently used scales are the hydrophobicity, hydrophilicity and secondary structure conformational parameters. In addition to ProtParam and ProtScale, Human Protein Atlas [31] is also an openly accessed online service for targeting proteins information. Using integration of various technologies, including antibody-based imaging, mass spectrometry-based proteomics, transcriptomics and systems biology, Human Protein Atlas is aiming to map various human proteins in cells, tissues and human organs. In the study, we used Human Protein Atlas to preliminary explore the cellular location of FGF1 protein in ccRCC cells. And ProtParam and ProtScale were used to interpret the gene's basic physicochemical parameters. Related signaling pathways and potential biological functions analysis Gene ontology analysis (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) have been two effectively used online services for annotating lists of genes and interpreting networks of signaling pathways. In the study, to explore the potential biological functions and probable signaling pathways of FGF1 gene, STRING was firstly used to reveal the surrounding genes that relate mostly with FGF1. And then, GO and KEGG were used to annotate the the signaling pathways that centered on FGF1, and the gene's potential biological functions were also preliminary explored. Identification of 192 DEGs in ccRCC comparing to normal renal tissues Four cDNA expression profiles from GEO database were used to screen the DEGs in ccRCC vs. normal renal tissues, and eventually a total of 1286 ( Fig. 1a), 1142 (Fig. 1b), 437 (Fig. 1c) and 257 (Fig. 1d) DEGs were identified in GSE53757, GSE71963, GSE68417 and GSE53000 profiles respectively. And of all the DEGs, 192 genes were shared in all four profiles including 39 genes that were shown to be up-regulated (Fig. 1e) and 153 down-regulated genes (Fig. 1f ) in cancer comparing to normal tissues (Additional file 1: Table S2). Basic interpretation of 192 DEGs by GO and KEGG To preliminary explore the biological functions of the 192 DEGs, GO and KEGG analysis were performed. Excitingly, GO results showed that the biological (Fig. 2c), and the molecular functions were primary oxidoreductase and receptor activities related (Fig. 2b). Further, KEGG/ biological pathway analysis showed the up-regulated DEGs were mostly enriched in HIF-1α related hypoxia and oxygen homeostasis regulating signaling pathways (Fig. 2d). And as for the 153 down-regulated genes, the cellular component were mainly enriched in plasma membrane (Fig. 2g), the molecular function were mostly focused on catalytic and auxiliary transport protein activities (Fig. 2f ), and the KEGG signaling were mainly transmembrane transport of small molecules, for instance glucose, bile salts and organic related (Fig. 2h). FGF1 gene works as a core gene in DEGs PPI network To further scale down the "candidate" genes and identify the potential key genes regulating ccRCC development, we construct the PPI network of 192 shared DEGs for further function modules analysis, thus understanding the interaction between different genes (Fig. 3a). And based on the PPI network, four modules were identified revealing signaling pathways that DEGs were mainly enriched in, interestingly, an FGF1 gene involving PI3K-AKT signaling was identified in 3/4 modules suggesting it's potential "core" position in the network (Fig. 3b-i). Additionally, to validate the "core" position of FGF1 gene in the network, the connectivity between different genes in the PPI were also explored, and the result supported FGF1 as one of the top 30 genes with high connectivity with surrounding genes (Fig. 4a, b). Moreover, Oncomine analysis revealed that although FGF1 expression various in different human cancers, multiple previous studies supported the FGF1 loss of expression in kidney cancers (Fig. 4c, d). And another analysis performed by GEPIA also showed consistent results that FGF1 expression various in different human tumors, for instance, the expression was higher glioblastoma and brain lower grade glioma comparing to paired normal tissues, but its expression in other tumors including ccRCC is aberrant lost (Fig. 4e). Additionally, not only in the solid tissues, FGF1 expression was significantly lower in ccRCC cell lines comparing to other cancer cells (Fig. 4f ). Both GEPIA and Oncomine results supported the aberrant loss of expression of FGF1 in ccRCC comparing to normal kidney in both solid tissues and cell lines. Aberrant FGF1 loss of expression in ccRCC To reveal the clinical value of FGF1 loss of expression in ccRCC. Kaplan-Meier plotter survival analysis was firstly conducted. And the overall survival analysis based on 530 kidney renal clear cell carcinoma samples revealed that FGF1 statistical significantly correlates with patients overall survival (OS), but not recurrence free survival (RFS), higher FGF1 gene expression directly associated with better patients overall survival indicating its potential tumor inhibitor function in ccRCC (Fig. 5a, b). To reveal the expression of FGF1 in ccRCC, besides the previous online analysis, IHC as well as QPCR experiments using local hospital patients tissues were also conducted (Detailed samples information see Additional file 2: Table S4). Consistent with the GEPIA online analysis (Fig. 5c), the result of qRT-PCR which was conducted using 30 local hospital ccRCC and paired normal renal tissues also supported the FGF1 loss of expression in cancer (Fig. 5d). Meanwhile, the immunohistochemistry (IHC) carried out in 104 local hospital ccRCC and paired normal renal tissues (different from the 30 samples used in qRT-PCR experiment) also revealed that FGF1 expression was significantly lower in cancer comparing to normal tissues. Significant loss of expression (less than 1%) was observed in ccRCC versus the much higher expression ratio (48.7%) in normal tissues (P < 0.01) (Fig. 5e). The association between FGF1 gene and ccRCC clinical features The association between FGF1 expression and ccRCC clinicopathological parameters was analyzed using Ualcan, which is an public online service based on TCGA data containing a whole of 533 ccRCC and 75 normal renal samples. The analysis result showed not only that FGF1 expresses much less in cancer comparing to normal renal tissues (Fig. 6a), but also the expression tends to decrease as the cancer grade and stage advancing although the difference was not statistical significant. Also, the expression of FGF1 tends to be lost more in older patients with lympho nodes metastasis than patients with younger age and no lympho metastasis, but the difference was not statistical significant. Meanwhile, no significance relationship was found between FGF1 expression and patients race and gender (Fig. 6b-g). Besides Ualcan online analysis, we also downloaded the original patients data from TCGA website(containing 539 ccRCC and 72 normal renal samples, detailed TCGA patients barcods in Additional file 1: Table S3) radiation therapy, which was consistent with the blood test which showed that white cell count was much higher in patients with high FGF1 expression (Table 1). Physicochemical properties of FGF1 gene Two online services ProtParam and ProtScale were used to predict FGF1′s physicochemical properties, Fig. 6 The association between FGF1 expression and ccRCC clinical parameters. a Relative FGF1 expression in ccRCC versus normal renal tissues. And the association between FGF1 expression and ccRCC b patients age, c gender, d race, e tumor grade, f lymph node metastasis and g tumor stage. (p < 0.05, p < 0.01, *p < 0.001. The first layer which is right above the error bar representing comparison to normal group, and the above layers * which were above a secondary line represent the comparison between corresponding groups that were covered by the line). h The hydrophilcity/hydrophobicity analysis of FGF1 protein. i FGF1 centered PPI network representing the genes that were mostly related to FGF1 and the results revealed that FGF1 protein is composed of 155 amino acids, including 19 negatively charged amino acid residues (ASP+Glu) and 18 positively charged amino acid residues (Arg+Lys). The molecular formula of FGF1 protein is C777H1208N210O238S5, the molecular weight is 17.5KD, and the theoretical isoelectric point is 6.51. Meanwhile, the estimated half-life of FGF1 protein is 30 h in mammals and the instability index is computed to be 40.67 indicating the protein tends to be cellular unstable. Additionally, ProtParam computed the hydrophobic value of FGF1 is 73.61 and the average hydrophilicity is − 0.620. ProtScale also revealed that FGF1 protein harbors several hydrophilic regions and shall be classified as a hydrophilic protein (Fig. 6h). Also, the result of Protein Atlas analysis supported FGF1 locating both in the nucleoplasm and is predicted to be secreted, suggesting its potential biological function as a hydrophilic signaling pathway particle. FGF1 gene centered biological functions and related signaling pathways To further explore the potential biological functions of FGF1 gene in ccRCC and the probable signaling pathways involved, GO and KEGG analysis were performed. And GO results showed that the biological processes FGF1 gene participated in were mainly focused on fibroblast growth factor receptor activities, phosphatidylinositol-3-phosphate biosynthetic processes and phosphatidylinositol-3-phosphate biosynthetic associated processes. And the molecular functions FGF1 played were most enriched in fibroblast growth factor receptor binding, 1-phosphatidylinositol-3-kinase activity, phosphatidylinositol-4,5-bisphosphate 3-kinase activity and protein tyrosine kinase activities (Fig. 6i, Table 2). Meanwhile, KEGG analysis revealed the signaling pathways FGF1 gene involved were mainly RAS signaling, Rap1 signaling, PI3K-AKT signaling and MAPK signaling pathways related (Table 3). Considering our previous gene module analysis based on PPI network which showed that FGF1 involved PI3K-AKT signaling shall play a core role in the network, it's of potential clinical value to further investigate the potential drug-targeting role of FGF1 gene or other FGF1 interacted PI3K-AKT signaling proteins in the development of ccRCC, thus aiding more precise understanding of the disease. Discussion Renal cell carcinoma has been a common malignant tumor of urinary tract, ranking only second to bladder carcinoma in morbidity of all adults urinary tract malignant tumors. Although the clinical treatment situation is still challenging given the tumor heterogeneity and evolutionary nature of cancer [32,33], the increasing developing molecular pathology has been bringing promising effect for ccRCC in both molecular diagnosis and targeting treatment. Especially in current precise medicine era, the various bioinformatic analysis tools has been making it more practicable for worldwide researchers to explore the molecular genetic abnormalities in cancers [34,35]. In the study, we combine used four different GEO cDNA expression profiles together with multiple bioinformatic analysis methods to explore the potential new prognostic indicators in ccRCC development, and we identified a specific FGF1 gene which was proved to be aberrant lost expression in ccRCC comparing to normal renal tissues, and the FGF1 lose of expression was indicated to be a worse overall survival indicator in ccRCC patients. GEO database has been one of the most commonly used public databases for worldwide researchers to explore the genetic abnormalities in various cancers [36][37][38][39]. In the study, we firstly picked four different cDNA expression profiles GSE53757, GSE53000, GSE71963 and GSE68417 from GEO database to analyze the differently expressed genes in ccRCC comparing to normal renal tissue, and the result revealed 192 genes that were shared in four profiles including 39 up-regulated and 153 down-regulated genes. Interestingly, although mainly enriched in different cellular locations and involved in various signaling pathways, both the up and down regulated DEGs were mostly participated in metabolism and energy regulation related biological processes. The mainly focus of DEGs on the metabolism related biological processes in the study supported the importance of the elaborate network of energy consuming in cancer development. Metabolomics has been a classic theory in cancer research based on the well known fact that even in the presence of oxygen, cancer cells perform less energy-efficient glycolysis process termed as aerobic glycolysis or Warburg effect [40,41]. Although the detailed reasons for Warburg effect are still unclear, one of the theories is that increased glycolysis may provide cancer cells easier access to accumulation of essential metabolic precursors they need for rapid cell proliferation [42][43][44]. To further scale down the "candidate" responsible genes and identify the potential "key" gene in ccRCC development, the PPI network of 192 DEGs was constructed to visualize the relationship between genes, and then gene function module analysis was successively performed. As a result, four gene modules involving various signaling pathways were identified based on the PPI network, excitingly, a FGF1 gene involving PI3K-Akt signaling pathway was shared in 3/4 modules indicating its potential core position in the network. What's more, the connectivity degree analysis between DEGs with surrounding genes also supported FGF1 gene as one of the top 30 genes with highest connectivity with other DEGs in the network. FGF1 gene, which is short for fibroblast growth factor 1, is one of the members of fibroblast growth factor (FGF) family and it has been reported to play important roles in the regulation of cell survival, cell division, angiogenesis, cell differentiation and migration [45]. In the study, the potential function of FGF1 gene in ccRCC development was explored. Firstly, Kaplan-Meier survival analysis based on TCGA data revealed that FGF1 gene expression statistical significantly correlates with ccRCC patients overall survival, higher FGF1 expression was associated with better survival, suggesting its potential tumor suppressor function. Then, to explore the expression of FGF1 in ccRCC comparing to normal renal tissues, both online database analysis and experiments based on local hospital samples were conducted. Both online GEPIA and Oncomine analysis indicated that although FGF1 expression various in different cancers, it was aberrant lost in ccRCC. Meanwhile, our IHC experiments conducted on tissue microarray which was produced using 104 local patients samples supported the loss of expression ratio (less than 1%) in ccRCC comparing to normal renal tissues (48.7%). What's more, QPCR experiment performed using 30 different patients samples also validated that FGF1 expressed less in cancer comparing to matched normal tissues. Since the sample number being used for our IHC and QPCR experiments was relatively low (104 cases for IHC experiment and 30 for QPCR experiment), and the patients with greater than 2, 3, 4 and 5 years follow-up was 70, 34, 17 and 13 respectively, the medial follow-up of the 134 patients was 33 months. To avoid the limitations of relatively small number samples and short duration of follow-up, an online service UALCAN which is based on TCGA data containing a total of 533 primary ccRCC and 72 normal renal samples was used for further analyzing the association between FGF1 and ccRCC clinical parameters. And the result showed that FGF1 loss expression in broad-spectrum ccRCC patients despite of the race, age, cancer grade and stage, and no significance relationship was found between FGF1 expression and patients gender. Further, to explore the potential biological function of FGF1 in ccRCC development, we computed the basic physicochemical parameters of the protein, which result revealed that FGF1 is a hydrophilic protein weighting 17.5KD, and the protein mainly locates in the nucleoplasm or to be secreted out of cells, the estimated halftime is 30 h and tend to be unstable. Meanwhile, the biological processes FGF1 gene participated in were mainly focused on fibroblast growth factor receptor activities and phosphatidylinositol-3-phosphate biosynthetic related processes, and the FGF1 centered signaling pathways were mostly RAS signaling, PI3K-AKT signaling and Rap1 signaling pathways. Given the result of our function module analysis which indicated that FGF1 gene involved PI3K-AKT signaling shall be in the core position of the DEGs PPI network, it's of potential clinical value to further investigate the detailed function and the mechanism behind FGF1 related PI3K-AKT signaling pathways in the regulation of ccRCC development. Actually, PI3K-Akt signaling has been commonly known to regulate insulin-based glucose metabolism and mutations of the pathway genes resulting in aberrant signaling activation, thus leading to higher amount of glucose uptake [46]. And activation of PI3K-Akt signaling provokes the expression of HIF-1α, which is a transcription factor generally known be involved in the cellular adaption to hypoxia and modulates cellular anaerobic metabolism [47]. Moreover, FGF1 expression was reported to be inhibited in diabetic nephropathy, and exogenous recombinant FGF1 protein not only has excellent function of reducing blood glucose level in type 2 diabetes mellitus, but also has a very obvious improvement effect on recovering the impaired diabetic renal function [48]. Interestingly, although FGF1 has no hypoglycemic effect on type 1 diabetes mellitus, it can also improve the renal function of type 1 diabetes mellitus indicating the improvement function of FGF1 on diabetic nephropathy exists independently of the hypoglycemic effect [49]. What's intriguing is that there's currently no evidence of association between ccRCC and diabetic nephropathy, sharing a similar genetic abnormality (loss of FGF1 expression) might provoke worldwide renal disease researchers' interest for further analysis. However, although above results shall provide meaningful insights into better understanding of ccRCC, it's not yet enough to classify FGF1 or other PI3K-AKT signaling proteins as new potential drug targets in ccRCC. To distinguish gene aberrations that can cause the disease and may serve as drug targets with those being closely linked to the disease and consequently are associated with the disease development, further comprehensive experiments and clinical trials are needed to be performed. Conclusion In conclusion, based on GEO database, we analyzed 192 DEGs in ccRCC comparing to normal renal tissues, and FGF1 gene and PI3K-AKT signaling was identified as a core signaling in DEGs' PPI network. Both online public data analysis and local hospital IHC as well as QPCR experiments validated the aberrant loss expression of FGF1 in ccRCC comparing to normal tissues. Kaplan-Meier overall survival analysis revealed that low FGF1 expression was associated with worse patients survival. Additionally, FGF1 centered biological processes and signaling pathways were preliminary explored. Comprehensive studies and clinical trials are needed to confirm the findings before promoting the clinical utility of FGF1as a new drug target and prognosis indicator in ccRCC.	2021-04-18T13:27:38.192Z	2020-12-18T00:00:00.000	{ "year": 2021, "sha1": "30cee9cd66b227dba8ba1b07ff710aba537740b4", "oa_license": "CCBY", "oa_url": "https://cancerci.biomedcentral.com/track/pdf/10.1186/s12935-021-01917-9", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "5380dab1b28a429f7ecde5fbce7817ac3fea3e4d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
3246166	pes2o/s2orc	v3-fos-license	Increased Serum Levels of Anti-Carbamylated 78-kDa Glucose-Regulated Protein Antibody in Patients with Rheumatoid Arthritis The objective of this study was to investigate the presence and titer of anti-carbamylated 78-kDa glucose-regulated protein (anti-CarGRP78) antibody in serum from controls, and patients with rheumatoid arthritis (RA), systemic lupus erythematosus (SLE) and primary Sjögren syndrome (pSS). Thirty-three RA patients, 20 SLE patients, 20 pSS patients, and 20 controls were enrolled from our outpatient clinic. GRP78 was cloned and carbamylated. Serum titers of anti- cyclic citrullinated peptides (anti-CCP), anti-GRP78, and anti-CarGRP78 were measured with an enzyme-linked immunosorbent assay. No differences in serum titers of anti-GRP78 antibody in patients with RA, SLE, or pSS compared with the controls were observed. Serum levels of anti-carGRP78 antibody in patients with RA, but not SLE or pSS, were significantly higher compared with the controls (OD405 0.15 ± 0.08 versus 0.11 ± 0.03, p = 0.033). There was a positive correlation between the serum levels of anti-GRP78 antibody, but not anti-CarGRP78 antibody, with the levels of anti-CCP antibody in patients with RA. Both anti-GRP78 and anti-carGRP78 antibodies failed to correlate with C-reactive protein levels in patients with RA. In conclusion, we demonstrated the presence of anti-CarGRP78 antibody in patients with RA. In addition, the serum titer of anti-CarGRP78 antibody was significantly elevated in patients with RA compared with the controls. Anti-CarGRP78 antibody could also be detected in patients with SLE or pSS. Introduction The 78 kDa glucose-regulated protein (GRP78), also known as binding immunoglobulin protein (BiP), has been shown to be an important autoantigen for rheumatoid arthritis (RA). GRP78 could stimulate the proliferation of synovial T cells from patients with RA [1]. Anti-GRP78 antibody can be found in the serum from patients both after and before their diagnosis of RA [2]. Anti-citrullinated protein antibody (ACPA) is known to be a very specific diagnostic marker for RA. In our previous study, we found that GRP78 was citrullinated (citGRP78) and became recognizable by ACPAs, and ACPAs could directly stimulate monocytes to secrete tumor necrosis factor α (TNF-α) via binding to citGRP78 [3]. In addition, anti-citGRP78 antibody was also widely detected in serum from patients with RA [4]. Moreover, parenteral administration of GRP78 could also ameliorate joint inflammation in mice with collagen-induced arthritis [5]. Carbamylation is a post-translational modification in which cyanate binds to primary amino or thiol groups. Recently, autoantibodies recognizing carbamylated proteins (anti-CarP antibodies) were found to be present in the sera from patients with RA and were predictive for joint damage [6]. In addition, the presence of anti-CarP antibodies could predict the development of RA [7] and serve as a poor prognostic factor for long-term disability in patients with RA [8]. Although the initial antigen used for detecting anti-CarP antibodies was carbamylated fetal calf serum [6], other carbamylated antigens such as α-enolase or vimentin [9,10] were found to be part of the anti-CarP antibodies. It is conceivable that other proteins could be carbamylated and recognizable by anti-CarP antibodies. Therefore, we hypothesized that the anti-CarGRP78 antibody might be present in the serum from patients with RA. The serum titers of anti-CarGRP78 antibody among patients with other systemic autoimmune diseases, including systemic lupus erythematosus (SLE) and primary Sjögren syndrome (pSS), were also investigated in this study. Patients and Controls We recorded the demographic and clinical characteristics of controls and patients with RA, SLE or pSS, including age, sex, anti-cyclic citrullinated peptides (anti-CCP) positivity, and rheumatoid factor positivity (Table 1). Validation of Carbamylaion of GRP78 The carbamylated GRP78, but not the native form GRP78, was recognizable by anti-modified citrulline antibody (Figure 1). with RA [4]. Moreover, parenteral administration of GRP78 could also ameliorate joint inflammation in mice with collagen-induced arthritis [5]. Carbamylation is a post-translational modification in which cyanate binds to primary amino or thiol groups. Recently, autoantibodies recognizing carbamylated proteins (anti-CarP antibodies) were found to be present in the sera from patients with RA and were predictive for joint damage [6]. In addition, the presence of anti-CarP antibodies could predict the development of RA [7] and serve as a poor prognostic factor for long-term disability in patients with RA [8]. Although the initial antigen used for detecting anti-CarP antibodies was carbamylated fetal calf serum [6], other carbamylated antigens such as α-enolase or vimentin [9,10] were found to be part of the anti-CarP antibodies. It is conceivable that other proteins could be carbamylated and recognizable by anti-CarP antibodies. Therefore, we hypothesized that the anti-CarGRP78 antibody might be present in the serum from patients with RA. The serum titers of anti-CarGRP78 antibody among patients with other systemic autoimmune diseases, including systemic lupus erythematosus (SLE) and primary Sjögren syndrome (pSS), were also investigated in this study. Patients and Controls We recorded the demographic and clinical characteristics of controls and patients with RA, SLE or pSS, including age, sex, anti-cyclic citrullinated peptides (anti-CCP) positivity, and rheumatoid factor positivity ( Table 1). Validation of Carbamylaion of GRP78 The carbamylated GRP78, but not the native form GRP78, was recognizable by anti-modified citrulline antibody ( Figure 1). Figure 1. Verification of the carbamylated GRP78 (CarGRP78) by Western blotting. Cloned GRP78 and CarGRP78 were modified by 2,3-butanedione monoxime and antipyrine in a strong acid solution. The modified citrulline residues were detected with rabbit polyclonal anti-modified citrulline antibodies. Only the CarGRP78 reacted with anti-modified citrulline antibodies. Serum Titers of Anti-GRP78 and Anti-CarGRP78 Antibody among Controls and Patients with Rheumatoid Arthritis, Systemic Lupus Erythematosus, or Primary Sjögren's Syndrome There were no statistically significant differences in the serum titers of anti-GRP78 antibody in patients with RA, SLE, or pSS, compared with the controls (Figure 2A). The percentage of positive The modified citrulline residues were detected with rabbit polyclonal anti-modified citrulline antibodies. Only the CarGRP78 reacted with anti-modified citrulline antibodies. Serum Titers of Anti-GRP78 and Anti-CarGRP78 Antibody among Controls and Patients with Rheumatoid Arthritis, Systemic Lupus Erythematosus, or Primary Sjögren's Syndrome There were no statistically significant differences in the serum titers of anti-GRP78 antibody in patients with RA, SLE, or pSS, compared with the controls ( The serum levels of anti-CarGRP78 in patients with RA was significantly higher compared with the controls (OD405 0.15 ± 0.08 versus 0.11 ± 0.03, p = 0.033; Figure 2B). On the other hand, no significant differences were observed when comparing the serum levels of anti-CarGRP78 antibody in the controls with SLE or pSS patients ( Figure 2B). The percentage of positive of anti-CarGRP78 antibody was 0% (0/20) in controls, 27% (9/33) in patients with RA, 35% (7/20) in patients with SLE, and 15% (3/20) in patients with pSS. The serum levels of anti-CarGRP78 in patients with RA was significantly higher compared with the controls (OD 405 0.15 ± 0.08 versus 0.11 ± 0.03, p = 0.033; Figure 2B). On the other hand, no significant differences were observed when comparing the serum levels of anti-CarGRP78 antibody in the controls with SLE or pSS patients ( Figure 2B). The percentage of positive of anti-CarGRP78 antibody was 0% (0/20) in controls, 27% (9/33) in patients with RA, 35% (7/20) in patients with SLE, and 15% (3/20) in patients with pSS. Correlation with Levels of Anti-GRP78 or Anti-CarGRP78 Antibody with Anti-CCP Antibody We found a positive correlation between the serum levels of anti-GRP78 antibody and the levels of anti-CCP antibody in patients with RA ( Figure 3A). However, there were no significant correlations between the serum levels of anti-CarGRP78 antibody and the levels of anti-CCP antibody in patients with RA. Correlation with Levels of Anti-GRP78 or Anti-CarGRP78 Antibody with Anti-CCP Antibody We found a positive correlation between the serum levels of anti-GRP78 antibody and the levels of anti-CCP antibody in patients with RA ( Figure 3A). However, there were no significant correlations between the serum levels of anti-CarGRP78 antibody and the levels of anti-CCP antibody in patients with RA. Correlation with Levels of Anti-GRP78 or Anti-CarGRP78 Antibody with CRP Levels We also found that there were no significant correlations between the serum level of anti-GRP78 antibody or anti-CarGRP78 and the CRP levels in patients with RA ( Figure 4). Correlation with Levels of Anti-GRP78 or Anti-CarGRP78 Antibody with CRP Levels We also found that there were no significant correlations between the serum level of anti-GRP78 antibody or anti-CarGRP78 and the CRP levels in patients with RA ( Figure 4). Discussion In this study, we demonstrated the presence of anti-CarGRP78 antibody in patients with RA. We did not find a significant difference between patients with RA and controls in the serum titer of anti-GRP78 antibody as previously described [2]. We noted that the titers of anti-GRP78 were high in few controls, which led to an elevated cutoff value. This could be explained by the source of our controls, which was rheumatology outpatients rather than healthy controls. In addition, the presence of high anti-GRP78 levels could be ethnicity-related. Compared with controls, the titer of anti-CarGRP78 antibody was significantly increased only in patients with RA (p = 0.033), but not in patients with SLE or in patients with pSS. We found that Discussion In this study, we demonstrated the presence of anti-CarGRP78 antibody in patients with RA. We did not find a significant difference between patients with RA and controls in the serum titer of anti-GRP78 antibody as previously described [2]. We noted that the titers of anti-GRP78 were high in few controls, which led to an elevated cutoff value. This could be explained by the source of our controls, which was rheumatology outpatients rather than healthy controls. In addition, the presence of high anti-GRP78 levels could be ethnicity-related. Compared with controls, the titer of anti-CarGRP78 antibody was significantly increased only in patients with RA (p = 0.033), but not in patients with SLE or in patients with pSS. We found that the percentage of positive anti-CarGRP78 antibody was 27% in patients with RA, 35% in patients with SLE, and 15% in patients with pSS. In comparison with other studies, anti-CarP antibodies were found in 45% of Japanese patients with RA [11], 11% of patients with SLE [12], and 27% of patients with pSS [13]. Furthermore, anti-CarP antibodies could be found in serum from patients with gout, osteoarthritis, spondyloarthritis, and psoriatic arthritis, though the percentage of anti-CarP antibodies positivity was below 10% among these patients [14,15]. Thus, compared with anti-CCP, anti-CarP antibodies appeared to be less specific for a differential diagnosis of RA. Further investigations are warranted to examine the presence of anti-CarGRP78 in other systemic autoimmune diseases and arthritides. Currently, there are still doubts about the specificity of anti-CarP antibodies. Cross-reactivity between the anti-CarP antibodies and anti-CCP might exist [9,16]. Our study showed that there was a positive correlation between the titer of anti-CCP and anti-GRP78, but not anti-CarGRP78, suggesting that anti-CarGRP78 might be different from anti-CCP, as previously reported [17]. GRP78 is an important autoantigen for RA. GRP78 on its own or GRP78-specific regulatory T cells have been considered as new therapeutic strategies for treating RA [18,19]. In this study, we did not observe any significant associations between the CRP levels and anti-GRP78 or anti-CarGRP78. Nevertheless, it is difficult to rule out the effects of potential confounders such as disease duration and use of medication in our cross-sectional study. With the elevated anti-CarGRP78 antibody levels observed in patients with RA, we still speculated that carmabylated GRP78 could play a role in the immunopathogenesis of RA. Patients In this study, 33 patients satisfying the 1987 American College of Rheumatology (ACR) revised criteria for the classification of RA [20], 20 patients satisfying the 1982 ACR revised criteria for the classification of SLE [21] and 20 patients satisfying the American-European Consensus Group Criteria for pSS [22] were recruited. For a strict evaluation of the clinical usefulness of the autoantibodies in the differential diagnosis of RA, instead of healthy individuals from the community, we randomly recruited 20 patients from the rheumatology outpatient clinic of our hospital with no known systemic autoimmune diseases as controls. All participants signed informed consent. The study was approved by the institutional review board of the Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, Taiwan (No. B10403022, 27 October 2015). Preparation of GRP78 and Carbamylated GRP78 Cloning of GRP78 was based on the methods described previously with modifications [23]. Full-length GRP78 cDNA was amplified by polymerase chain reaction (PCR). The resulting products were digested with NcoI and EcoRI (New England Biolabs, Ipswich, MA, USA) and then subcloned into pET32a. E. coli BL21 (DE3) was then transformed with recombinant pET-32a encoding GRP78. The transformed bacteria were grown in LB broth with ampicillin (0.1 g/L) and induced with 1 mM IPTG for 4 h at 37 • C. Bacteria were harvested by centrifugation, resuspended in 100 mL of 20 mM Tris-HCl buffer (pH 7.9) containing 0.5 M NaCl, 0.2 mM PMSF, 0.02% NaN 3 , 4 mM benzamidine and 0.5 mM imidazole, and lysed using a French press. GRP78 was purified from the crude lysate by sequential separation on nickel Sepharose column. The purified protein fractions were pooled, concentrated by ultrafiltration and dialyzed against polymerization buffer (5 mM Tris-HCl (pH 7.5), 2 mM CaCl 2 , 0.1 M KCl, 1 mM MgCl 2 , and 1 mM ATP). The purified GRP78 was stored at 4 • C before use. The cloned GRP78 was carbamylated by the method described by Shi et al. [6]. The CarGRP78 was extensively dialyzed against PBS and then stored. Western Blotting CarGRP78 were detected by Western blotting using a modified anti-citrulline detection kit (Upstate Biotechnology, Lake Placid, NY, USA) [24]. Briefly, 20 ng CarGRP78 and GRP78 were immobilized on PVDF membrane after 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by electrotransfer. Then, immobilized CarGRP78 and GRP78 were modified by 2,3-butanedione monoxime and antipyrine in a strong acid solution, according to the manufacturer's instructions. The modified citrulline residues were detected by the rabbit polyclonal anti-modified citrulline antibodies and goat anti-rabbit IgG-horseradish peroxidase (HRP) antibody conjugate. 4.5. Enzyme-Linked Immunosorbent Assay (ELISA) for Anti-Cyclic Citrullinated Peptides (Anti-CCP), Anti-78-kDa Glucose-Regulated Protein (Anti-GRP78) Antibody, and Anti-Carbamylated 78-kDa Glucose-Regulated Protein Anti-CarGRP78 Antibody Anti-CCP was detected by Quanta Lite CCP3 enzyme-linked immunosorbent assay (ELISA) kits (Inova Diagnostics, San Diego, CA, USA) in accordance to the manufacturer's recommendations. ELISA plates (BD Biosciences, San Jose, CA, USA) were coated for 2 h at 37 • C with 1 µg/mL GRP78 or CarGRP78 in coating buffer containing 35 mM NaHCO 3 and 15 mM Na 2 CO 3 , pH 9.6. The plates were washed with phosphate-buffered saline (PBS) for four times and blocked with PBS containing 2% skim milk overnight. Then, Quanta Lite reagents (Inova Diagnostics, San Diego, CA, USA) were used for performing ELISA according to the supplier's protocol. The plates were read at 405 nm on an Anthos Zenyth 3100 multimode fluorometer (Anthos Labtec Instruments GmbH, Salzburg, Austria). All samples were analyzed at the same time. The absorbance values of control wells were low in all cases. The cutoff values for the anti-GRP78 and anti-CarGRP78 antibody ELISA was defined as the mean plus two standard deviations of the controls. Statistical Analysis All data were represented as mean ± standard deviation. Statistical significance was assessed with non-parametric Mann-Whitney U test. Linear regression analysis was conducted to evaluate the correlations between levels of anti-GRP78 or anti-CarGRP78 antibody with the levels of CRP or anti-CCP antibody. Two-tailed p < 0.05 was considered statistically significant. Conclusions In conclusion, we demonstrated the presence of anti-CarGRP78 antibody in patients with RA, SLE, and pSS. The serum titer of anti-CarGRP78 antibody was significantly elevated in patients with RA compared with controls.	2017-05-06T13:22:36.820Z	2016-09-01T00:00:00.000	{ "year": 2016, "sha1": "7aad7753657034011afb4c30227ab3194f08e803", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1422-0067/17/9/1510/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "7aad7753657034011afb4c30227ab3194f08e803", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
199585681	pes2o/s2orc	v3-fos-license	Combination of Blowfish Algorithm First of File Algorithm and End of File Algorithm on Image Security In this age, there are many types of information. An image is a form of information that can be sent from anyone to anyone in various ways. But in the shipping process, there is often a hijacking or theft by an unwanted party. So that information must be secured by cryptography to avoid this. To add a sense of security between the sender and recipient, cryptography will be combined with steganography. In this study, the Blowfish Algorithm was combined with First Of File and End Of File steganography methods. Information that has been encrypted will be inserted into other media. Likewise for the recipient, can extract information then decrypt information. Based on experiments conducted on the Blowfish algorithm, the time of encryption and decryption needed when the resolution of the image gets larger is linearly proportional to the value. MSE gets smaller, and the PSNR value gets higher, it shows better image quality. Cover image will increase in size after the embedding process so that it becomes a stego image. Introduction The image was a form of data or information that exist in visual form. Nowadays, information or data transmission could be done by anyone and anyway. However, not all of the information tended to be common. There were classified messages which the originality must be kept along the transmission. A way to secure the data could be done by using cryptography and steganography. Cryptography comes from the words crypto and graphia. Crypto means secret, and graphia means writing [1]. Thus, cryptography is the science of maintaining the security of messages when messages are sent from one place to another place. The Blowfish algorithm was created by a cryptanalyst named Bruce Schneier and published in 1994 [2]. Blowfish is a 64-bit cipher block with variable key lengths and consists of two parts: key expansion and data encryption [3]. Steganography is the development of cryptography. Steganography is the hiding of messages in a storage medium and is not scrambling the contents of the file [4]. Steganography protects messages from unwanted people so that others can only see stego images without knowing the secret message [5]. There are several methods in steganography such as End of File (EOF) and First of File (FOF). In the End of File (EOF) method, the message is inserted at the end of the value. While First of File (FOF), the message is inserted at the beginning of the value [7]. Method In this implementation there are4 main menus, they are the blowfish encryption, FOF, and EOF Embedding, FOF and EOF extraction and blowfish decryption. Encryption The passkey used in this system test is "password" where the passkey will be processed in the Expansion Key on Blowfish. Passkey "password" will be XORed with P-Array which consists of 18 32-bit subkeys as follows: Change "password" to the binary number so that we can get a binary from "password" is: 01110000 01100001 01110011 01110011 01110111 01101111 01110010 01100100 XOR the initial 32 bits of the passkey with P [0] from the subkey and the next 32 bits with P [1] from the subkey repeat the process until all the P-arrays are XOR. P[0] 01110000 01100001 01110011 01110011 00100100 00111111 01101010 10001000 01110111 01101111 01110010 01100100 11110010 11001100 01111010 10110111 P [2] 01110000 01100001 01110011 01110011 0 11000010 00101000 01111011 01110011 P[16] 01110000 01100001 01110011 01110011 11111110 00010110 10001001 01111111 From the calculation above, the new key-key value is obtained, then the encryption string is entirely zero with the subkey that has been obtained from the process above, then replace all P-arrays with the result that the string encryption is entirely zero so that the output is continuously changing. So we can get 1 0 1 0 1 1 1 0 1 1 1 1 0 0 1 0 1 1 1 1 0 1 1 0 1 0 0 0 0 1 0 0 P[17] = 1 1 1 1 1 0 0 1 1 1 0 0 0 0 0 1 1 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 After getting the value in the subkey, we will encrypt the plain image, for the RGB value of the initial pixel of the plain image, this test is: 76 54 50 77 53 51 86 61 1. Convert the pixel value that will be encrypted to binary so that it becomes: Embedding In this process, the plain image that has been encrypted into a cipher image (cipher.bmp) will be inserted into the cover image measuring 500x500 pixels. The cipher image will be divided into two parts so that one section is inserted in the top pixel of the image and one part is inserted into the bottom pixel of the image. So that it produces a stego image. Extracting In this process, the stego image will be extracted so that the cipher image will be removed from the cover image. Cipher image will return to its original size as before Embedding. Table 3 Table of In this test stego image is the result of insertion from cipher.bmp to sunlight.bmp. Then after the Extracting process, cipher.bmp will return separately from sunlight.bmp. Tabel 4 Table of Decryption In this process, the cipher image that has been generated from the extracting process will be decrypted. So that the cipher image will return to the plain image as before. The decryption process is the same as doing blowfish encryption, only begins with P [17] and ends with P [0]. The key used is "password." Decryption steps are the same as encryption only if encryption starts from P [0] then the decryption will start from P Results and Discussions The experiments were performed on ASUS X452C with Intel-Core i3 processor and 2GB RAM. This study uses Eclipse Java Oxygen and uses the Java programming language. The results of the experiments for each set are presented in Tables 5 and 6. In table 5, the graph of the results of encryption testing with variations in plain image resolution shows the greater the image resolution, so encryption process longer, if the image resolution smaller, the encryption process time be faster. table 6, the graph of the results of decryption testing with variations in cipher image resolution shows the greater the image resolution, so decryption process longer, if the image resolution smaller, the decryption process time be faster. Conclusions Based on the experiments conducted in this study, it can be concluded that the process of running time of encryption and decryption blowfish algorithm is time needed when the resolution of the image gets bigger is relatively stable.	2019-08-15T22:24:43.393Z	2019-06-01T00:00:00.000	{ "year": 2019, "sha1": "b5c1b974e334d15ed323cf0c82c8377bf2d4e974", "oa_license": null, "oa_url": "https://doi.org/10.1088/1742-6596/1235/1/012078", "oa_status": "GOLD", "pdf_src": "IOP", "pdf_hash": "cd978b882b8eb9f6dec073075f940776f6ce6988", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
234669426	pes2o/s2orc	v3-fos-license	Single cell transcriptomics of human PINK1 iPSC differentiation dynamics reveal a core network of Parkinson’s disease Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder, characterized by the loss of dopaminergic neurons (mDA) in the midbrain. The heterogenous pathology and complex underlying mechanisms are only partly understood and there is no treatment able to reverse PD progression. Here, we targeted the disease mechanisms by focusing on the ILE368ASN mutation within the PINK1 (PARK6) gene and systematically characterized midbrain dopaminergic neurons obtained from human induced pluripotent stem cells (iPSCs). Single-cell RNA sequencing (RNAseq) and pairwise analysis of gene expression identied genes consistently differentially expressed during the mDA neuron differentiation process. Subsequent network analysis revealed that these genes form a core network, which interacts with all known 19 protein-coding Parkinson’s disease-associated genes and includes ubiquitination, mitochondrial, protein processing, RNA metabolism, and secretory pathways as important subnetworks. Our ndings indicate a unied network underlying PD pathology and offers new interpretation of the phenotypic heterogeneity of PD. Introduction Parkinson's disease (PD) is one of the most prevalent neurological disorders, second only to Alzheimer's disease, with a prevalence of 1.8%, among persons over the age of 65 and 2.6% in the 85 to 89 age group [1][2][3] . As the average age of the population increases, PD is expected to pose an increasing and signi cant burden to society. PD is characterized by the presence of motor symptoms, including bradykinesia, rigidity and tremor, but many patients also develop non-motor symptoms, such as depression or dementia 4 . Unfortunately, there are no treatments to slow down or reverse the progression of the disease. Current treatments only temporarily ameliorate the motor symptoms, but do not slow down the progression of PD 5 . Most of our understanding of PD pathology is based on the identi cation of mutations that lead to PD, although these account for only 3-5% of PD cases, with the remaining cases being idiopathic 2 . Despite the small fraction of cases they explain, these mutations provide an important window into the underlying molecular mechanisms of PD because they identify pathways which, when disrupted, are able to cause the disease. Many of these mutations converge on mitochondrial homeostasis, repair and mitophagy. Hence, mitochondrial dysfunction likely plays a key role in the pathophysiology of PD 6 . An important group of these mutations lies within the PINK1 gene, a nuclear gene that codes for a kinase important for mitochondrial function 7 . PINK1, also known as PARK6, clearly plays a role in mitophagy, but its function is much broader. The targets of this kinase are involved in many cellular functions, including neuronal maturation 8,9 . The impact of the loss-of-function mutations in this important kinase has not yet been fully elucidated 10 . One key characteristic of PD is the death of the midbrain dopaminergic (mDA) neurons, but until recently, it was impossible to study them since 60% of these neurons have disappeared by the time of diagnosis, and about 90% by the time the patients die 11 . As a result, research was limited to animal models 12 , but human-like mutations in animals often do not lead to the development of comparable pathology 13 . The development of cellular reprogramming allows nowadays for the conversion of somatic cells into induced pluripotent stem cells (iPSCs), which can subsequently be differentiated into neurons. This enables us to generate iPSCs from the skin cells of PD patients 14 and differentiate them into mDA neurons carrying the disease 15,16 . Furthermore, by using cells from patients with a known PD-associated mutation, we can analyze mutation-speci c pathology in the correct, permissive, genetic background. This is important because genetic background is known to signi cantly in uence the age of onset and severity of the disease 7,17 . Differentiating mDA neurons from iPSCs provides an almost unlimited source of neurons that allow for deep phenotyping and the elucidation of the cellular mechanisms underlying PD pathology. Here, we generated iPSCs from the broblasts of a patient homozygous for the PD-associated mutation ILE368ASN in the PINK1 gene 2 . We used an optimized differentiation protocol to speci cally generate mDA neurons, as this cell type displays a unique susceptibility to cell death in PD 15,18,19 . In contrast to mDA neurons, the effect of PD on other types of DA neurons is variable, hence their study would not be as pertinent to the elucidation of mechanisms causing PD-induced cell death 11,20 . The development of mDA neurons diverges from other DA neurons even before they commit to neural fate. During early neural development, neural tube stem cells generate neurons and glia, the two basic building blocks of the brain. While other DA neurons follow this direct path from neural stem cells to neurons, determined by the expression of the Pax6 transcription factor, mDA neurons develop from radial glial cells of the oor plate and are exposed to high levels of the SHH transcription factor 21 , which prevents expression of Pax6 22 and sets these cells on an entirely different developmental path 18,23 . Hence, they follow a very different signalling cascade, leading to the expression of a unique transcriptome that signi cantly differs from that of other DA neurons 18,20,24,25 . Their distinct identity is re ected in their function and leads to their unique susceptibility to death in PD, which in turn leads to the classic movement symptoms of the disease 11,19,20,23,26 . To investigate the disease mechanisms linked to the PINK1 mutation, we performed extensive single-cell RNA expression analysis using DropSeq 27 at four different timepoints during mDA neuron differentiation 15,16,18 . Differential expression analysis between four pairs, each consisting of a PINK1 and a control cell line, identi ed genes that were strongly and consistently dysregulated at all timepoints. Based on known protein-protein interactions, we show that these genes form a network and that its members directly interact with all 19 protein-coding PARK genes associated with PD. This suggests that other PD-associated mutations may also be acting through this common network of genes. Overall, our results indicate the existence of a common disease mechanism that potentially underlies idiopathic PD as well and may represent a unifying perspective on PD progression that will guide future intervention strategies. Results We performed a systematic differential expression analysis at single-cell resolution between an iPSC line carrying the PD-associated ILE368ASN mutation in the PINK1 gene and an age-and sex-matched control cell line (control 1-2 in ref 28 ) during their differentiation into mDA neurons (Fig. 1). After preprocessing and quality ltering, we used 4495 cells and 18,097 genes in our downstream analysis (Methods). Expression of key genes was also con rmed by qPCR and corresponding marker staining. For data integration, we performed a network analysis to identify the underlying key mechanisms of PD progression. Single-cell RNAseq analysis reveals gene expression panel for direct classi cation of iPSC-status The standard accepted procedure for determining iPSC status involves staining for iPSC markers and the use of gene expression platforms, such as Scorecard. However, we show that a standard panel of genes readily detectable by single cell analysis can be used to con rm iPSC status directly in the cells used for the single cell experiment, rather than by staining or expression analysis of an independent sample, which in some cases may not re ect the iPSC status of the experimental sample. Fibroblasts were isolated from a 64 year-old male with PD symptom onset at 33 years of age who was homozygous for the ILE368ASN (P.I368N/P.I368N) mutation in the PINK1 gene (Coriell Institute, Cat. No. ND40066). The broblasts were con rmed to have a normal karyotype ( Supplementary Fig. 1) and an episomal reprogramming method (Epi5™ Episomal iPSC Reprogramming Kit, Invitrogen USA, cat. # A15960) was used to avoid unwanted genetic modi cations of the target cell. The karyotype of iPSCs was con rmed (Supplementary Fig. 2) and their iPSC status was ascertained by standard methods, including staining for the POU5F1 marker (also known as Oct4) and by using the TRA-1-80 antibody (Fig. 2a), as well as by a TaqMan iPSC Scorecard Assay, which also con rmed their trilineage potential (Fig. 2b). In vitro differentiation of iPSC-derived mDAs recapitulates the in vivo process To con rm that our differentiation protocol (Supplementary Table 1) is recapitulating the in vivo mDA differentiation path, the expression of key genes (OTX2, EN1, LMX1B, LMX1A, and FOXA2) known to drive mDA differentiation in vivo 19,23,37 was con rmed using our single-cell RNAseq data or by qPCR (Fig.3b, c, Table 1). The absence of the early, non-mDA neuron marker PAX6 21,22 was further validated by staining (Supplementary Fig. 3 and Supplementary Table 2). In vivo, the development of mDA phenotype depends on the early high expression of Sonic Hedgehog (SHH), followed by the induction of Wnt signaling and the expression mDA-speci c downstream pathways 18,19,23 . Consistent with these in vivo differentiation steps, among the highest-expressed genes on day 6 (D6) of the differentiation protocol were PTCH1, a receptor for SHH, and FZD7, a receptor for Wnt proteins (Fig. 3b). Following the differentiation process from iPSCs to D21 (Fig. 4a), we could see the onset of expression of the mature mDA markers TH and KCNH6 (also known as GIRK2) in the D21 cluster (Fig. 4b). By day 21, many factors that are speci c to the mDA differentiation path, such as TCF12, ALCAM, PITX2, ASCL1, and DDC 20,38-41 , were among the most highly expressed genes (Fig. 4c). The onset of TH expression indicated that the cells achieved the state of early postmitotic mDA neurons 18 (Fig. 4b), which was con rmed by staining (Fig. 4d) and qPCR (Fig. 4e, Supplementary Table 3). In addition, markers speci c to mature mDA neurons including TH, DDC, ALDH1A1, and KCNJ6 (also known as GIRK2) also started to become expressed ( Fig. 4 b-e, Table 1). The PINK1 ILE368ASN mutation is associated with persistently dysregulated expression of nearly 300 loci To analyze mutation-induced changes in gene expression, we performed single-cell RNA expression analysis, at four important timepoints during the mDA neuron differentiation process, using DropSeq 27 - (Fig. 1). After preprocessing and quality-ltering (Methods and Supplementary Fig. 4), a total of 4495 cells (2518 control and 1977 PINK1 cells) and 18,097 genes were included in our analysis). Control and PINK1 cells co-clustered together based on their differentiation stage, from iPSCs, to day 6 (D6), D15 and D21 (Fig.4a), indicating that RNA expression was speci c to differentiation stages, rather than to a cell line. The PINK1 cells at D10 showed low viability, hence D10 timepoint was not included in the pairwise analysis (Fig. 5). The analysis of pairwise differential expression at each time point with adjusted p-values (p adj ) <0.01 fold changes (FC) > 0.1 resulted in 14 genes that were upregulated and 13 genes that were signi cantly downregulated in the PINK1 cell line compared to control (Table 2). Because iPSCs are very different from differentiating neuronal precursors, we next tested whether inclusion of iPSCs had disproportionately affected the results through the exclusion of neuron-speci c genes. Repeating the analysis using D6, D15 and D21 identi ed 28 genes that were upregulated and 27 genes that were downregulated at all four timepoints, including all genes previously identi ed ( Table 2, Fig. 5, Group A). As expected, excluding iPSCs resulted in the identi cation of a broader range of genes because genes that are differentially expressed only in the neuronal lineage were previously excluded due to the absence of their expression in iPSCs and the requirement that DEGs have to be dysregulated at all timepoints. However, both sets are equally valuable, as genes dysregulated even in iPSCs are more likely to be genes that participate in systemic PD pathology, regardless of cell type, and may be relevant to a broader spectrum of PD pathology than the death of mDA neurons. Interestingly, most of these genes are already linked to PD, other PD mutations, or neurodegeneration ( Table 2). For an alternative de nition of differentially expressed genes (DEGs), we used the maximum adjusted pvalue in a pairwise combinations as adjusted p-value, and the average fold change that occurred in the pairwise comparison as fold change threshold. With this approach we retained only genes dysregulated in the same direction at all timepoints. This analysis led to 151 DEGs, including the previously identi ed genes of Group A, of which 65 were upregulated and 86 downregulated compared with controls (p adj < 0.01 and FC > 0.1) (Group B, Supplementary Table 4). Taking the mean of FC of the different time points enhanced the identi cation of DEGs because it reduced the effect of the variability between pairs due to their different differentiation states. Repeating the same analysis for timepoints iPSCs, D6, D15 and D21, but taking into account only the absolute degree of change in iPSCs, yields 172 genes (Group C, Supplementary Table 5). Repeating the analysis using only timepoints D6, D15 and D21 identi ed a total of 285 DEGs (Group D) (Supplementary Table 6 and Supplementary Fig. 5). Together, when all analyses were pooled, we obtained 291 DEGs (6 genes in Group C depended on the inclusion of iPSCs and did not appear in Group D, see Supplementary Table 6). Data integration reveals a common PD network. To integrate the expression analysis and identify underlying disease mechanisms, we generated a network of interactions between the DEGs via Gephi 42 , using protein-protein interaction information obtained from the STRING and GeneMANIA databases 43,44 . The network we obtained includes 246 of the 291 DEGs, since pseudogenes and non-coding RNAs could not be integrated into a protein-protein intearction network (Supplementary Table 7), and 2122 interactions (Fig. 6, Supplementary Fig. 5c). The curated network only considers DEGs and any genes automatically added by the databases were excluded to ensure a reliable core network based solely on data. Based on known protein-protein interactions, the DEGs integrate into a close-knit core network in which several DEGs form central nodes (Fig. 6a, Supplementary Fig. 5). To evaluate the signi cance of the DEG-based PPI (protein-protein interaction) network produced by STIRNGdb (v10), we compared the DEG-based network with corresponding random networks generated from sets of 292 randomly chosen genes excluding DEGs. Based on 50 random networks, we show that the DEG-based network includes signi cantly more proteincoding genes and interactions than by chance (Fig. 6b) and that the network structure in term of degree distribution is signi cantly distinct as evaluated by Wilcoxon test (p = 2.22e-16) and indicates the mechanistic character of the network (Fig. 6c). The network of genes dysregulated by the presence of the PINK1 mutation includes genes related to other PD-associated pathways, which is intriguing, since it was generally assumed that each PD-associated mutation leads to PD pathology via an independent, characteristic path. For example, two DEGs, GOPC and GPC3 45,46 , interact with the PD-associated gene DJ-1 (PARK7) 2,47 . The DEG network also includes genes of the LRRK2 (PARK8) network 2,47 , namely ENAH, HSPA8, MYL6, MALAT1, and SNHG5 (Table 2). SNHG5 and MALAT1 interact with LRRK2 via miR-205-5p 44,45 . DLK1 and MALAT1 mediate a-synuclein accumulation 48,49 . In fact, the DLK1-NURR1 interaction involved in this process may be mDA neuronspeci c 50 , highlightihe necessity to use mDA neurons for the study of PD-related pathways. Additionally, MALAT1 was shown to increase a-synuclein protein expression 51 . In short, this suggests that interactions leading to PD pathology are more complex than one mutation leading via one path to PD, as generally thought, but it also indicates that there are likely many druggable targets that may be useful in treating PD, and that these may be universally effective for PD caused by several different mutations, and perhaps even for idiopathic PD. For example, terazosin, which is already in clinical use, was found to be associated with slower disease progression, likely by enhancing the activity of phosphoglycerate kinase 1 (PGK1) 52 , one of the top DEGs identi ed in our study. For the evaluation of the relative importance of each node within the network, we applied betweenness centrality 42 ( Supplementary Fig. 5a-c). The major nodes of these networks are formed by genes that were already shown to play an important role in PD pathology (Table 3). Next, we built a correlation network (pvalue < 0.05, r > 0.1) of the 246 DEGs based on the normalized counts (Fig. 6e). By extracting the common interactions of these two networks, we obtained a network with 297 interactions (Fig. 6d, e and Supplementary Table 7), which highlights protein-protein connections that correlate with differential expression of the genes. This analysis further supports the role of the connections between these genes in mediating the resulting differential expression in the presence of the PINK1 mutation. STRING was subsequently used to highlight functional pathways represented within the DEG network (Suppplementary Fig. 6 and Supplementary Table 8). Several pathways known to play a role in PD pathology are signi cantly represented within the network, notably ubiquitination 12,53 , mitochondrial pathways 6,54 , cellular response to stress 55 , lysosomal proteins 56 , protein metabolism (localization, modi cation, transport, folding and stability), RNA processing 57 , aromatic compound metabolism 58-61 , vesicle mediated transport and exocytosis 62 , and cellular catabolic processes 55,56 . Interestingly, the DEGs include genes of the KEGG-PD 43 pathway and the CHCHD2 gene, which was recently identi ed as a PDassociated gene and named PARK22 47,63,64 . To investigate further how the identi ed network relates to other known PD mechanisms, PD-associated genes, also known as PARK genes (Supplementary Table 9, Fig. 7), were added to the DEG network. Next, PARK-PARK interactions were removed and only PARK-DEG interactions were retained to test how PARK genes integrate into the network. All 19 protein-coding PARK genes 2,47 interact directly with at least one, but usually several DEGs ( Supplementary Fig. 7). The degree of interaction of PARK genes with the DEGs of the network is illustrated by coloring (in pink) DEGs that directly interact with a PARK gene. The darker the color, the greater the number of PARK genes the DEG interacts with. The central nodes of the network generally interact with several PARK genes, suggesting that they play a central role in linking the PARK genes to the network, but also that PARK genes may mediate PD pathology through a few central pathways of this network, and that the effects of different PARK genes converge on the same set of pathways. Further analysis revealed that a large number of the DEGs interact with genes associated with mitochondria or ubiquitination. For this analysis, we used BioGRID 44,65 to identify interactions with mitochondrial or ubiquitination proteins for the top 172 DEGs of groups A-C (Fig. 6f). These interactions were used to create a network illustrating that many of the DEGs in our study directly interact with genes involved in mitochondrial function and in ubiquitination. Thereby, only direct DEG -mitochondrial gene or DEG -ubiquitination gene interactions were included and PARK genes were added for reference ( Supplementary Fig. 8). Based on manual literature search, we determined that at least 68% of the DEGs (174 of 255 genes, not including pseudogenes and RNA genes) are already directly associated with PD, either experimentally, or as signi cant links in GWAS-PD, or by PD expression studies (Fig.7, Supplementary Table 10). This is particularly true for the major nodes of the network (Supplementary Fig. 9). Discussion The aim of this study was to identify genes that were differentially expressed as a result of a mutation in the PINK1 gene, using mDA neurons differentiated from patient-derived iPSCs, a model relevant to PD. We focused on cells undergoing neural differentiation, as these are not expected to display the activation of damage control pathways induced by neurotoxicity, but are likely to identify pathways that lead to primary pathology of PD. Gene expression alterations in aging neurons as a result of the modi cation identi ed here will be the next important step to investigate the manifestation of neurodegeneration. The single-cell expression data were analyzed in several layers. First, we identi ed the most strongly DEGs (group A), consistently altered in the same direction at all four timepoints including iPSCs (Fig.6, Table 2, Supplementary Fig. 5). By choosing four different timepoints along the differentiation path and selecting only genes whose differential expression was consistent at all timepoints, we excluded pathways associated with mDA differentiation, as these were expected to change signi cantlyuring the different steps of the differentiation process. In our rst analysis, we included iPSCs, in order to identify strongly DEGs due to the presence of the PINK1 mutation, regardless of cell type ( Table 2, Group A "incl. iPSCs"). Our reasoning was that since iPSCs differ signi cantly from neural precursors and are not expected to express neuronal pathways, their inclusion would bias against the selection of neuronal pathways. A corresponding approach in which iPSCs were excluded resulted in an expanded gene list that included genes more likely dysregulated speci cally in a neural cell type ( Fig. 5). Creating a protein-protein interaction network based on these groups of DEGs demonstrated that genes in Group Dlso formed important nodes within the interaction network and were frequently associated with PD ( Supplementary. Fig. 5). The resulting network illustrates that in spite of the very different nature of PD-associated genes, there is interconnectedness among the molecular pathways through which they mediate PD pathology. It also suggests that any one mutation leads to pathology via several molecular paths. This underscores the role the genetic background plays in PD penetrance and severity, as alleles of several network genes may reduce or amplify the effect of a mutation 7,17 . The complexity of genetic interactions in PD is well illustrated by the interaction of PINK1 and Parkin. PINK1 is known to interact with Parkin directly, however, our data indicates that the presence of the PINK1 mutation results in the dysregulation of several other genes that are possibly upstream of Parkin 66 , including HNRNPC 67 , MTRNR2L1 68 , MYL12A and SLC25A4 69 , as well as LMAN1, a membrane mannosebinding lectin, which was shown to play a role in Parkin translocation 70 . This suggests that the direct interaction between PINK1 and Parkin is not the only means by which PINK1 interacts with the Parkin pathway. Analysis of the networks shows that certain DEGs are a point of convergence between pathways and form major nodes. These DEGs seem to play a central role within the network, speci cally CUL3, HSPA8, EEF1A1, UQCRFS1, CNTNAR2, PSMA4, HNRNPC, and PLCB4, but also EGLN3, IPO5, IPO7, PALLD, PGD, RALGPS2, CYCS, SHH, BRCA2 and others ( Supplementary Fig. 5). In fact, these genes play key roles in PD pathology (Table 3, Supplementary Fig. 9). Hence, the network derived from our analysis of a PINK1 mutation is revealing the convergence of many known, key PD-associated pathways. This convergence suggests that different mutations may feed into the same PD pathology-associated routes. These central pathways include several genes (listed in Table 3), such as CUL3, which has been linked to PD by GWAS studies and is considered a potential PD drug target 71 . HSPA8 (also known as HSP73 and HSC70), which disaggregates α-synuclein amyloid brils and plays a role in autophagy and the catabolic pathway for αsynuclein, mediates mitophagy by regulating the stability of PINK1, and its expression was shown to be impaired in sporadic PD [72][73][74][75] . It is also one of the most highly dysregulated genes in our dataset. EEF1A1 mediates activation of heat-shock transcription factor HSF1, a key player in PD 76 , and prevent asynuclein aggregation, as well as interacts with Parkin (PARK2) and HTRA2 (PARK13) 65,77 . UQCRFS1 is a mitochondrial electron transport chain ubiquinol-cytochrome c reductase 78 , a member of the KEGG-PD pathway (Entry K00411 79,80 ), and has been identi ed as a PD risk gene 81 . CNTNAP2, which belongs to the neurexin superfamily, plays a role in triggering protein aggregates 82,83 , was found to be signi cantly differentially expressed in the blood of PD patients with LRRK2 mutation 84 , and was also associated with PD by GWAS 46 . PSMA4, a proteasome subunit, is part of the KEGG-PD pathway (hsa05012, bta05012, K02728) 79,80 and is a member of the ubiquitin-proteasomal pathway, which plays a key role in Parkinson's disease 85 . It also interacts with Parkin (PARK2) and FBXO7 (PARK15) 65 . HNRNPC interacts with both PARK2 and members of the Poly (ADP-ribose)-dependent cell death pathway implicated in PD 67 . PLCB4 has been linked to PD 46 and knock-out mice show motor defects consistent with ataxia 86 . These central nodes interact with several PARK genes (Supplementary Table 9), so it is possible that PDassociated mutations converge on the same main pathways, which may play a central role in PD pathology. Currently, mutations in 23 genes or loci are linked to PD. In addition to their o cial names, they are named PARK1-PARK23 (Supplement Table 9). Of these, 19 are protein-coding genes. The DEGs we have identi ed in this study directly interact with all 19 protein-coding PARK genes (Supplementary Fig. 7). In addition, the CHCHD2 gene (PARK22) was identi ed as one of the DEGs. This further supports the hypothesis that perturbations due to the mutations in PARK genes may converge on the same network, which may then be responsible for the PD phenotype. This would explain why mutations in so many genes lead to a common or at least similar PD pathology 87 . It will be of great interest to see if cells from idiopathic patients show dysregulation of this integrated network. In fact, our analysis has identi ed genes which had no known connection to molecular mechanisms underlying PD pathology, though some of themare known to be involved in sporadic PD. Knowing how they integrate into the network may point to the underlying mechanism of how they cause PD pathology. For example, one of the top DEGs is LGI1 88 . The development of antibodies to this protein leads to immunomodulated parkinsonism, yet there is no known mechanism linking it to PD pathology 88 . In the network LGI1 has several nearest DEG neighbors it interacts withSupplementary Fig. 10). Its most signi cant interaction is its co-expression with CNTNAP2, which is part of the neurexin family and is required for axon organization, and MGMT, which repairs the methylated nucleobase in DNA 43 . From GeneMANIA alone, the strongest evidence is for interaction with GOLT1B, which plays a role in Golgi transport 89 . Hence, LGI1-associated pathology leading to PD symptoms may be mediated through pathways which are also dysregulated by the presence of the PINK1 mutation. CNTNAP2 is another very good candidate, as it was shown to be dysregulated in PD patients carrying a mutation in the LRRK2 gene, providing additional evidence that it likely plays a role in PD pathology 84 . Another DEG in our dataset is CHCHD2, the PARK22 gene. Mutations in this gene are linked with autosomal dominant PD, although the precise mechanism is unknown 90 . One hypothesis was that CHCHD2 colocalizes with the mitochondrial contact site and cristae organizing system (MICOS) 90 . However, in the DEG network, CHCHD2 has strong interactions with at least three other DEGs, SLC25A4/ANT1 (STRING 43 ), GHITM (STRING and GeneMANIA), and NME4 (GeneMANIA 44 ). Evidence suggests that GHITM plays a role in PINK1-mediated neurodegeneration 91 and NME4 was shown to be downregulated in PD 58 . SLC25A4 (also known as ANT1) plays an essential role in mitophagy and has been linked to PD pathology 69,92 . Hence, the interaction of CHCHD2 with SLC25A4 (ANT1), GHITM, and NME4 may be relevant in mediating pathological changes in CHCHD2-associated PD. It is of note that pathways known to play a key role in PD are profoundly and consistently dysregulated at all time points examined, far before the onset of PD pathology. For example, CHCHD2 is part of the purine metabolic pathway that produces DNA, RNA, nucleosides and nucleotides and has been shown to be altered in PD [58][59][60][61] . The DEG network illustrates that the expression of a large number of interconnected aromatic compound metabolic pathway genes is altered in cells carrying the PINK1 mutation ( Supplementary Fig.6b). In total, 39 genes of the nitrogen compound metabolic process (Ncmp) and 88 genes speci c to the aromatic compound metabolic process (Acmp, a subgroup of the Ncmp) are differentially expressed in our dataset (Supplementary Table 8). Many of the DEGs identi ed in our study are part of more than one pathway and, therefore, interconnect the various pathways known to play a role in PD, including stress and catabolic processes 55,56 , aromatic compound metabolism 58 , vesiclemediated transport and exocytosis 62 , RNA metabolism 57 , protein transport, localization, folding, stability, and ubiquitination 53 (Supplementary Fig. 6 and Table 8). This con rms observations that PD pathology involves many different pathways 93 . Additionally, when a manual literature search was performed, 68% of the DEGs were found to be associated with PD (Fig. 7, Supplementary Table 10), which is particularly true for the central nodes of the network ( Table 3, Supplementary Fig. 9). This indicates that these genes may be key points of integration of the effects of PD pathology, an idea further substantiated by the convergence of the added PARK genes on these nodes ( Supplementary Fig. 7). Furthermore, these nodes form a link between different functional pathways. In particular, this is true for CUL3, HSPA8 and PSMA4 (Supplementary Fig. 6 and Table 8). We have also analyzed the correlation of expression between various gene pairs. This correlation may indicate that they are a target of the same regulatory pathway or otherwise related. In our dataset, the expression of several interaction partners shows high correlation, namely PLCB4-RALGPS-TTC3-ZNF37A, EIF3B-HSPA8 (major DEG network node, ubiquitination pathway)-PCBP1 (ubiquitination pathway). Another cluster centers on the MT-CYB gene and involves both mitochondrial and ubiquitination pathways by NME1-MT-CYB -MT-ND5 -MT-CO3 -MRPS21 interactions. Among the top pairs are also PSMD7-PSMB5, TAGLN-MYL9 and VWA5A-ZCCHC11 (Figure 6d). The interactions of these genes may, therefore, play a key role in PINK1-mediated PD pathology. The fact that so many genes which belong to other PD mutation-related pathways were dysregulated by the presence of the PINK1 mutation suggests that pathways involved in PD pathology are far more interconnected than previously thought. It is likely that PD pathology is more a disease with a characteristic network ngerprint, than a disease caused by independent defects (mutations) (Fig. 6). This and future studies will, hopefully, provide a picture of how various mutations feed into this network and cause its dysregulation. If idiopathic PD is shown to also be based on dysregulation of this network, then we may nally be able to understand the cause of idiopathic PD, which represents 80-85% of all PD cases 2 . Methods Generation of iPSC cell lines Fibroblasts (cat. No. ND40066) isolated from a 64-year-old male with PD symptom onset at 33 years of age who carried a homozygous mutation for the ILE368ASN (P.I368N/P.I368N) mutation in the PINK1 gene were obtained from the Coriell Institute (Cat. No. ND40066). Samples were collected in accordance with the US Government guidelines and are subject to an MTA issued by Coriell Institute for Medical Research NINDS Cell Repository. Fibroblasts were cultured on gelatin-coated plates (10% gelatin in PBS, coated for 10 min at room temperature) in KO DMEM +10% FBS + 1% penicillin/streptomycin (stock was 10,000 units penicillin and 10 mg streptomycin/mL) at standard culture conditions (37°C, 5% CO2). Live adherent broblasts in culture media were sent to be karyotyped by Cell Line Genetics, Madison, WI, USA (Supplementary Fig. 1) Analysis of iPSC status and trilineage potential by TaqMan iPSC Scorecard Assay. We also performed a TaqMan iPSC Scorecard Assay, which con rmed their iPSC status and their trilineage potential (Fig. 2b). The protocol was followed as described by the manufacturer of the TaqMan hPSC Scorecard Assay (ThermoFisher Scienti c). Immunocytochemistry Adherent colonies were xed in 4% PFA for 10 min, washed and permeabilized with 0.1% Triton X-100 in 1X PBS for 15 min, then washed and incubated in a blocking solution of 2% BSA in 1X PBS for 1 hour. Differentation of iPSCs into mDA neurons The protocol used to differentiate iPSCs into mDA neurons was modi ed from 16,94 (Table 1). The iPSCs were grown to 95% con uence, dissociated using accutase, and 1.5 wells were combined into one well at day -1. They were allowed to recover in the presence of ROCK inhibitor for about 8 hours and then in mTeSR without ROCK inhibitor for about 16 hours. After this, day 0 media were applied (Table 1). Cells were fed fresh media daily, 36 ml per 6-well plate or as needed, judging consumption from media colour and replacing media whenever it turned yellow, using the appropriate media and factor mix for that day. Real-time quantitative PCR of mDA and non-mDA markers. Total RNA was extracted from a cell pellet of a 12-well plate well using the RNeasy Plus Universal Kit Mini (2×) (cat. #K0223). This was followed by a dissociation curve to con rm that only one PCR product was present. The values were standardized per expression of the housekeeping gene GAPDH. Statistics In real-time qPCR graphs, each timepoint consists of three independently differentiated samples of the ND40066 (PINK1) cell line (n=3) and the control (n=3). Each of the samples was ampli ed in triplicate. Each sample value was an average of the experimental triplicate. Standard error was calculated using GraphPad, based on n=3 independently differentiated samples for each cell line at each timepoint. Single-cell RNA seq On the day of collection, cells were dissociated using accutase. The single-cell suspension was spun down and cells were washed with (PBS, 2%BSA) twice, then passed through a 40mm lter to remove larger cell clumps. The sample was then counted and viability was determined using (Vi-CELL XR, Cell Counter, Beckman Coulter). Cells were required to have at least a 90% viability. Most samples showed 95% viability. They were then diluted in PBS with 2%BSA to a nal concentration of 190,000 cells/ml. 3ml were then used for single-cell analysis. Subsequently, cells were processed by the DropSeq approach as described previously 27,96,97 and sequenced. NGS preparation for Drop-seq libraries The 3' end-enriched cDNA libraries were prepared by tagmentation reaction of 600 pg cDNA library using the standard Nextera XT tagmentation kit (Illumina). Reactions were performed according to the manufacturer's instructions. The PCR ampli cation cycling programme used was 95°C 30 s, and twelve cycles of 95°C (10 s), 55°C (30 s) and 72°C (30 s), followed by a nal extension step of 72°C (5 min). Libraries were puri ed twice to reduce primers and short DNA fragments with 0.6× and 1× Agencourt AMPure XP beads (Beckman Coulter), respectively, in accordance with the manufacturer's protocol. Finally, puri ed libraries were eluted in 10 μl Molecular Grade Water. Quality and quantity of the tagmented cDNA library were evaluated using Bioanalyzer High Sensitivity DNA Chip. The average size of the tagmented libraries prior to sequencing was between 400 and 700 bps. Puri ed Drop-seq cDNA libraries were sequenced using Illumina NextSeq 500 with the recommended sequencing protocol except for 6pM of custom primer (GCCTGTCCGCGGAAGCAGTGGTATCAACG CAGAGTAC) applied for priming of read 1. Paired-end sequencing of 20 bases (covering the 1-12 bases of random cell barcode and the remaining 13-20 bases of random unique molecular identi er (UMI)) was performed for read 1, and of 50 bases of the genes for read 2. Bioinformatics processing and data analysis The FASTQ les were assembled from the raw BCL les using Illumina's bcl2fastq converter and run through the FASTQC codes (Babraham bioinformatics; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to check for consistency in library qualities. The monitored quality assessment parameters monitored were (i) per base sequence quality (especially for the read 2 of the gene), (ii) per base N content, (iii) per base sequence content, and (iv) over-represented sequences. The FASTQ les were then merged and converted into binaries using PICARD's FastqToSam algorithm. The sequencing reads were converted into a digital gene expression matrix using the Drop-seq bioinformatics pipeline 27 . Single-cell RNA-seq data analysis The identi cation of low quality cells was done separately for each data set. In order to select only the highest quality data, we sorted the cells by their cumulative gene expression. Only cells with the highest cumulative expression were considered for the analysis 98 . In addition to this ltering we de ned cells as low-quality based on three criteria for each cell. The number of expressed genes must be more than 200 and 2 median-absolute-deviations (MADs) above the median; the total number of counts has to be 2 MADs above or below the median, and the percentage of counts to mitochondrial genes has to be 1.5 MADs above the median. Cells failing at least one criteria were considered as low quality cells and ltered out from further analysis. Similar to the cell ltering, we ltered out low quality genes, identi ed by being expressed in less than 10 cells in the data. The integration of the ltered matrices of the different datasets was performed using scTransform 99 was computed using the 5000 most variable genes of the integrated data. The clustering of data was performed using Louvain clustering. The resolution of the clustering was selected based on the best silhouette score of the different resolutions 101 . A short list of manually curated markers was used to validate the different stages of the differentiation process. We then performed differential expression analysis between the two treatments (control and PINK1) at each time point. The differential expression analysis was done using MAST 100 (default parameters) on the normalized counts using the total number of transcripts in each cell as a covariate and the Bonferroni correction to correct for multiple hypothesis testing (Padj). In addition, we tried to nd conserved markers among the different time points using MAST again and the total number of transcripts in each cell as a latent variable. Genes with fold changes of the same sign in the fold change were then identi ed across the different time points and the average fold change was calculated. The genes with average fold change > 0.1 and maximum adjusted p-value < 0.01 were selected as differentially expressed. Network analysis We extracted protein-protein interaction information between the DEGs from STRING 43 and from GeneMANIA 44 . We excluded indirect association, such as text mining, co-occurrence and neighborhood from STRING, and coexpression, colocalization, shared protein domains and predicted interactions from GeneMANIA, retaining only genetic interactions, pathways and physical interactions (2122 interactions in total). We deletes any any genes or interactions that were added by these databases, to focus only on DEGs and interactions among them. The network diameter was calculated and betweenness centrality was used to illustrate the relative importance of each node within the network. As a control, we selected the same number of genes at random, using the list of genes detected by our RNA-seq analysis, excluding DEGs. This control set did not produce a network and led to a mostly disconnected array of genes ( Supplementary Fig. 11). Networks were also generated using the STRING and GeneMANIA inputs independently ( Supplementary Fig. 5 a, b). We constructed a correlation network based on the correlation of expression of DEGs (p-value < 0.05, correlation > 0.1) and identi ed edges that are common to the two networks. This network consisted of 860 interactions (Fig. 6). We then extracted shared interactions of these two networks, which amounted to 297 interactions (Fig. 6). In order to validate the PPI network produced by STIRNGdb (v10), we created 50 PPI (protein protein interaction) networks using 292 random genes (same as the number of DEGs). We then compared the number of detected proteins, the number of interactions between the genes and the distribution of the node degrees. We performed the Wilcoxon test to access if the two-degree distributions are different from one another in a statistically signi cant manner, which showed a statistically signi cant difference (p=2.22e-16) (Fig.5b,c.). Data availability. Single cell RNAseq data will be uploaded to GEO upon acceptance of the manuscript. Table 3. Central nodes of the DEG network are associated with PD ( Supplementary Fig. 9). Central nodes were determined using the Gephi visualization platform. They represent points of convergence of the network ( Supplementary Fig. 5). Since these nodes have already been linked to PD pathways, many more DEGs might also contribute to PD pathology through these pathways. These nodes not only provide a point of convergence for DEGs identified in our study, but they also interact with several PARK genes, suggesting that PARK proteins may also converge on the pathways identified here (Supplementary Fig. 7). Figure 1 Experimental design. Fibroblasts were used to generate human induced pluripotent stem cells (iPSCs), which were then used to generate mDA neurons. Differentiation was initiated at three different times to obtain cells at different stages of differentiation that could all be collected and analyzed by single-cell RNA-seq on the same day, thus avoiding batch effects. "P+1" indicates that the iPSCs were passaged before new differentiation was initiated. Since D10 was not used in pairwise analysis, we indicated "P+2" between D15 and D6 differentiation initiation Generation and classi cation of iPS cell lines. a. Immunocytochemistry. Staining for the iPSC markers Oct3/4 and TRA-180. DAPI was used to stain cell nuclei as a reference. b. Results of Scorecard analysis of iPSCs and embryonic bodies (EBs). iPSCs are expected to show high expression of self-renewal genes (Self-renew +) and low expression of mesoderm, ectoderm and endoderm markers (Ecto -, Meso-, Endo -). EBs include cells at an early stage of spontaneous differentiation. Scorecard analysis of EBs determines the iPSC line's potential to differentiate into the three germ layers: ectoderm, mesoderm, and endoderm. EBs are expected to express few or no self-renewal genes (Self-renew -) and to show expression of some mesoderm, ectoderm and endoderm markers: Ecto +/-, Meso +/-, Endo +/-. Table 1). b. Expression of key iPSC markers and transition to the expression of genes belonging to pathways characteristic of early differentiation. c. Expression of OTX2 was con rmed by absolute quantitative real-time PCR (qPCR) (bars represent SE, n=3 independently differentiated samples, qPCR of each sample was done in triplicate). EN1, an important marker of mDA differentiation, was measured by qPCR (n=3, same qPCR procedure) in bulk samples, as its expression is too low to be detected by single cell RNA seq. Primers are listed in Supp. Table 3.	2020-10-28T18:33:31.444Z	2020-09-05T00:00:00.000	{ "year": 2020, "sha1": "427419d5720725c00367403c100cd520fb68fe7e", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s42003-021-02973-7.pdf", "oa_status": "GREEN", "pdf_src": "ScienceParsePlus", "pdf_hash": "f89577de9b3e845c8d53c2ea73c8e61f2e097c12", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [ "Biology" ] }
119288269	pes2o/s2orc	v3-fos-license	Millisecond Pulsars and Black Holes in Globular Clusters Over a hundred millisecond radio pulsars (MSPs) have been observed in globular clusters (GCs), motivating theoretical studies of the formation and evolution of these sources through stellar evolution coupled to stellar dynamics. Here we study MSPs in GCs using realistic $N$-body simulations with our Cluster Monte Carlo code. We show that neutron stars (NSs) formed in electron-capture supernovae (including both accretion-induced and merger-induced collapse of white dwarfs) can be spun up through mass transfer to form MSPs. Both NS formation and spin-up through accretion are greatly enhanced through dynamical interaction processes. We find that our models for average GCs at the present day with masses $\approx 2 \times 10^5\,M_\odot$ can produce up to $10-20$ MSPs, while a very massive GC model with mass $\approx 10^6\,M_\odot$ can produce close to $100$. We show that the number of MSPs is anti-correlated with the total number of stellar-mass black holes (BHs) retained in the host cluster. The radial distributions are also affected: MSPs are more concentrated towards the center in a host cluster with a smaller number of retained BHs. As a result, the number of MSPs in a GC could be used to place constraints on its BH population. Intrinsic properties of our model pulsars, such as their magnetic fields and spin periods, although hard to determine precisely, are in good overall agreement with observations. Interestingly, our models also demonstrate the possibility of dynamically forming NS--NS and NS--BH binaries in GCs, although the predicted numbers are very small. INTRODUCTION Globular clusters (GCs) are known to be highly efficient at producing millisecond pulsars (MSPs). Since the discovery of radio MSPs in GCs in the 1980s (Lyne et al. 1987), multiple pulsar surveys have found 150 pulsars in 28 GCs 1 (for reviews, see Camilo & Rasio 2005;Ransom 2008), including 38 in Terzan 5 and 25 in 47 Tuc. Although GCs make up only about 0.05% of stars in the Milky Way, collectively, GCs contain more than one third of the total number of known MSPs in our Galaxy (Manchester et al. 2005) GCs also contain a large number of low-mass X-ray binaries (LMXBs; Clark 1975) with neutron star (NS) accretors. The low surface magnetic fields (∼ 10 7 − 10 9 G) and short spin periods ( 30 ms) of MSPs suggest that they are "recycled" pulsars (Alpar et al. 1982) with LMXBs as their likely progenitors. Indeed some "transitional" MSPs, providing the link between LMXBs and MSPs, have recently been detected; these are observed to switch back and forth between phases of accretion-powered X-ray emission and rotation-powered radio emission. At present, there are three confirmed transitional MSPs, with one in GC M28 (Papitto et al. 2013) and two in the Galactic field (Archibald et al. 2009;Bassa et al. 2014;Roy et al. 2015). There are also a few additional candidates in GCs and the Galactic field (Bahramian et al. 2018, and references therein). While the physics is far from being understood in detail, it is plausible that mass transfer onto slow, dead pulsars with fields 10 10 G can "bury" their magnetic fields while at the same time spinning them up (e.g. Bhattacharya & van den Heuvel 1991;Rappaport et al. 1995;Kiel et al. 2008;Tauris et al. 2012, and references therein). In addition, MSPs have low spin-down rates and thus long lifetimes (∼ 10 10 yr, compared to ∼ 10 7 yr for young pulsars), which makes them easier to observe in old stellar systems like GCs. The large numbers of MSPs and NS LMXBs in GCs suggest that Galactic GCs on average contain at least a few hundred NSs (e.g., Ivanova et al. 2008). However, observations show that the majority of NSs in the Galactic field are born with velocities 200 km s −1 due to natal kicks associated with core collapse supernovae . Because of these large natal kicks, the majority of NSs born in GCs (where escape velocities are generally < 50 − 100 km s −1 even at early times when the clusters may have been more massive than at present) are ejected from the cluster at birth. This is seemingly at odds with the large numbers of NSs inferred to be present in GCs, which is often referred to as the NS "retention problem" (e.g. Pfahl et al. 2002b). However, the discovery of high-mass X-ray binaries with long orbital periods (P orb > 30 days) and low eccentricities (e 0.2) (HMXBs; Pfahl et al. 2002a) suggests that some NSs must be born with very small natal kicks. Additionally, NSs born in massive binaries may be easier to retain in GCs, but Pfahl et al. (2002b) showed that the retention fraction of NSs formed through core collapse supernovae in massive binaries is still at most a few percent, not enough to explain the large populations of NSs in GCs. Later studies suggested that electron-capture supernovae (ECSNe) can solve the retention problem by producing many NSs with small natal kicks (Podsiadlowski et al. 2004;Ivanova et al. 2008). When electron capture occurs onto Mg 24 and Ne 20 in a degenerate ONeMg core, it triggers the core to collapse to a NS (Miyaji et al. 1980;Nomoto 1984Nomoto , 1987. The explosion energy of ECSNe is much lower than that of iron core-collapse supernovae (CCSNe). As a result, NSs formed in ECSNe receive an order of magnitude smaller natal kicks compared to NSs formed in CCSNe (Podsiadlowski et al. 2004). Therefore a much larger fraction of ECSN NSs can be retained in GCs after formation (Kuranov & Postnov 2006;Ivanova et al. 2008), in contrast to CCSN NSs. The high stellar densities in GCs lead to frequent dynamical encounters and high formation rates of MSPs and LMXBS (Clark 1975;Hut et al. 1992;Pooley et al. 2003;Hui et al. 2010;Bahramian et al. 2013). For example, dynamical interactions can enhance the WD-WD merger rate (e.g. Shara & Hurley 2002), which can lead to a higher merger-induced collapse rate of WDs to NSs (Saio & Nomoto 1985). NS binaries are also created at an increased rate since single NSs can acquire companions through exchange interactions (Sigurdsson & Phinney 1993Rasio et al. 2000;Ivanova et al. 2008). Subsequent stellar evolution of the companion and/or other hardening encounters can then trigger Roche-lobe overflow (RLOF) and mass transfer, possibly recycling the NS. More than half of the observed MSPs in GCs are known to be in binaries. Several studies have shown that the number of MSPs in GCs is correlated with the stellar encounter rate, ∝ ρ 2 0 r 3 c /σ 0 , where ρ 0 is the central luminosity density, r c is the core radius and σ 0 is the central velocity dispersion (Verbunt & Hut 1987;Hui et al. 2010;Bahramian et al. 2013). In clusters with larger encounter rates, WDs and NSs undergo more dynamical interactions and produce more NS binaries. For example, while double NSs (DNSs) are very rare, the only confirmed GC DNS is in M15, which is a core-collapsed cluster with extremely high central density. It has been suggested that the frequent stellar encounters in M15 led to the formation of this DNS (Anderson et al. 1990;Prince et al. 1991;Phinney & Sigurdsson 1991;Deich & Kulkarni 1996;Jacoby et al. 2006). Additionally, BHs could have a strong influence upon the formation of NS binaries by altering the evolution of their host GCs. Once formed, BHs will quickly mass-segregate to the cluster core through dynamical friction. Several recent studies have shown that the retention fraction of BHs in GCs strongly affects the GC core densities and the ability of other compact objects to participate in the core dynamics (Mackey et al. 2008;Morscher et al. 2015;Chatterjee et al. 2017;Kremer et al. 2018a;Fragione et al. 2018b). Very few previous works have attempted to study the formation and evolution of MSPs in the context of fully realistic N -body simulations of GCs. Ivanova et al. (2008) modeled the formation and evolution of NSs in GCs using the population synthesis code StarTrack (Belczynski et al. 2002b to follow single and binary star evolution in a fixed cluster background, and the small N -body integrator Fewbody ) to compute dynamical interactions in the cluster core. They included ECSNe for NS formation and showed that the low natal kicks associated with these NSs were crucial to matching the observed number of MSPs in GCs. Here, we build upon this previous study by performing full, self-consistent N-body simulations for the cluster dynamics, and also incorporating the effects of magnetic field and spin period evolution during mass-transfer on the formation of MSPs. In Section 2, we describe how we model MSPs in GCs. We show the full grid of models used for this study, methods for simulating the magnetic fields and spin periods of NSs, and the formation and dynamical evolution of NSs in our models. In Section 3 we explore the anti-correlation between the number of MSPs and the number of BHs and we compare our results with pulsar observations. We discuss selection effects and summarize our results in Section 4. Models We use our Cluster Monte Carlo code (CMC) to simulate a grid of models for this study. CMC is a parallelized Hénon-type Monte Carlo code (Hénon 1971a,b) which has been developed and rigorously tested over many years Fregeau et al. 2003;Chatterjee et al. 2010;Umbreit et al. 2012;Pattabiraman et al. 2013;Chatterjee et al. 2013a;Rodriguez et al. 2018). CMC incorporates all the relevant physics for GC evolution, including two-body relaxation, three-body binary formation, strong threeand four-body interactions, and some post-Newtonian effects (Rodriguez et al. 2018). Updated versions of the SSE and BSE packages are used for single and binary stellar evolution in CMC and Fewbody ) is used to directly integrate all three-and four-body gravitational encounters, with post-Newtonian effects (Antognini et al. 2014;Amaro-Seoane & Chen 2016). In this study, we consider 26 total independent cluster models. In models 1-25 (which serve as our main grid of models with present-day properties typical of Milky Way GCs), we fix a number of initial cluster parameters: total star number N = 8 × 10 5 , binary fraction f b = 5%, virial radius r v = 1 pc, King concentration parameter W o = 5, galactocentric distance r g = 8 kpc and metallicity Z = 0.001 Chatterjee et al. 2017). We use the initial mass function given in Kroupa (2001) ranging from 0.08 to 150 M to sample the initial stellar masses. NS remnants from CCSNe at formation receive natal kicks drawn from a Maxwellian distribution with a standard deviation σ N S = 265 km s −1 . Following Kremer et al. (2018a), the only parameter varied in our models is the natal kicks for BHs, which allows us to easily isolate and understand the effects of BH retention on long-term GC evolution and dynamical evolution of MSPs. Varying the BH natal kicks to alter the retained number of BHs at present times has a similar effect on the GC properties as varying more physically motivated initial conditions, such as the initial virial radius (Kremer et al. 2018b). With all other parameters being fixed, all differences between models clearly originate from the differences in the fraction of BHs retained in the cluster upon formation (see Kremer et al. 2018a, for more details). All models were evolved for 12 Gyr. The natal kicks for BHs are assumed to be independent of BH masses and are drawn from the same Maxwellian distribution as the NSs. Their magnitudes are determined by the ratio σ BH σ N S which is varied between the models. We set different ratios from 0.005 to 1.0 for the models as shown in Table 1. The introduction of a parameter controlling the natal kicks received by the BHs gives us some benefits for the purpose of this study: it allows us to control the BH retention fractions of our models without the necessity to change any other initial parameters. As a result, the changes between the properties of these models can unambiguously be attributed to the difference in BH retention fractions. In addition, we simulate one model (model 26 in Table 1) meant to represent a very massive GC. This model has initial N = 3.5 × 10 6 , binary fraction 10% and BH natal kick σ BH σ N S = 1.0. We choose a higher binary fraction of 10% to maximize the number of MSPs in the large-N model. Other initial conditions for this model are the same as models 1-25. The final mass of the model is about 1.2 × 10 6 M , which is close to the masses of 47 Tuc and Terzan 5. This model is an extreme case designed to form many MSPs as a result of the large initial N and large BH kicks. In this paper, we define models with number of BHs greater than 200 as "BH-rich" models, models with 10 BHs as "BH-poor" models and the others as "BHintermediate" models. Note-Column 1-3: model number, initial number of stars and BH natal kick, respectively. Columns 4-13 (model properties at 12 Gyr): projected core radius, projected half-light radius, total cluster mass, number of BHs, number of retained NSs, number of retained dynamical-formed NSs, total number of pulsars including MSPs, total number of MSPs, number of single MSPs and number of binary MSPs, respectively. The last two columns are number of NS-BH binaries and DNSs that appear between 9 − 12 Gyr. Simulating NS evolution For this work we have updated the prescriptions for the evolution of NSs in SSE and BSE in CMC. These updates include changes to the magnetic field and spinperiod evolution for single and binary NSs, and to the natal kick prescriptions for NSs formed in ECSNe (Kiel et al. 2008;Kiel & Hurley 2009). Magnetic Field and Spin-Period Evolution The evolution of NS magnetic fields and spin periods has long been a topic of debate and still remains uncertain (e.g., Faucher-Giguere & Kaspi 2006, and references therein). In our models, NS remnants, when formed, are assigned randomly sampled magnetic fields (range 10 11.5 − 10 13.8 G) and spin periods (range 30 − 1000 ms) to match observations of young pulsars. To model NS evolution, we follow the prescriptions described in Hurley et al. (2002) and Kiel et al. (2008). As outlined in Kiel et al. (2008), we assume that the dominant spin-period evolution mechanism for single NSs is dipole radiation, and NSs are treated as solid spheres. The spin-down rate of single NSs iṡ where K = 9.87×10 −48 yr/G 2 , B is the surface magnetic field and P is the spin period 3 . Additionally, magnetic fields of single NSs are assumed to decay exponentially, on a timescale τ = 3 Gyr (Kiel et al. 2008). Here B 0 , and T are the initial magnetic field, and the NS's age, respectively. Faucher- Giguere & Kaspi (2006) shows that there is no significant magnetic field decay of single radio pulsars in a timescale of ∼ 100 Myr. Recent magnetothermal models of NS magnetic field also show that the magnetic field evolution of single radio pulsars is compatible with no decay or weak decay (e.g., Popov et al. 2010;Viganò et al. 2013). The timescale we adopted for magnetic field evolution of single radio pulsar is compatible with a very weak field decay. Further studies of magnetic field decay of GC pulsars will be discussed in future works. For NSs in binaries, binary evolution is also taken into account. The evolution of NSs in detached binaries is the same as for single NSs. On the other hand, during masstransfer episodes, the magnetic fields and spin periods of NSs can change significantly on a short timescale. During accretion, the magnetic field is assumed to decays as where t acc is the duration of the NS accretion phase (during RLOF) and ∆M is the mass accreted. The NS spin periods decrease accordingly due to angular momentum transfer. Wind mass loss, tidal evolution, magnetic braking and supernova kicks have only small effects on the magnetic field and spin period evolution. Only stable mass accretion can spin up NSs to MSPs in our models. We do not consider mass accretion during common envelope phases . Occasionally, through dynamical or binary evolution, NSs merge with main-sequence (MS) stars, giants and WDs. If the outcome of this merger is a NS (as opposed to a BH; see Hurley et al. 2002, Table 2), the magnetic fields and spin periods for these NSs are reset by drawing from the same ranges of initial values as above. The properties of magnetic fields and spin periods of newborn NSs are independent of their formation. However, if a MSP is involved in such a merger, different initial magnetic fields and spin periods are assigned (relative to regular NSs, as described above) so that the new magnetic fields and spin periods match the values of observed MSPs. For MSPs involved in mergers, magnetic fields are drawn from the range 10 8 − 10 8.8 G and spin periods are drawn from the range 3−20 ms. In other words, we assume that a MSP involved in a merger, remains a MSP after the merger. In addition, we assume the lower limit for NS magnetic fields to be 5 × 10 7 G (Kiel et al. 2008). We do not set a lower limit for the spin period. Electron-capture Supernovae In our models, NSs formed in ECSNe are the dominant type of retained NSs and in NS-LMXBs and MSPs. In our models we assume that an ECSN happens when an ONeMg WD reaches a critical mass M = 1.38 M . At that time electron capture is triggered on Mg 24 and Ne 20 and the WD goes through collapse from the lacking of electron pressure support (Miyaji et al. 1980;Nomoto 1984Nomoto , 1987. We give small natal kicks to ECSN NSs where the kicks are drawn from a Maxwellian distribution with a dispersion σ ECSN = 20 km s −1 (Kiel et al. 2008). In CMC we have three different evolutionary paths that lead to ECSNe (Ivanova et al. 2008). The first one is evolution-induced collapse (EIC) of a single star with initial mass in the range 6 − 8 M (Nomoto 1984(Nomoto , 1987Kiel et al. 2008). For main-sequence (MS) and giant stars, if their initial masses are smaller than the carbon ignition mass and larger than the critical NS formation mass, they can form NSs in ECSNe. Helium stars with masses between 1.6 M and 2.25 M also go through ECSNe. The second path is accretioninduced collapse (AIC) (Nomoto & Kondo 1991;Saio & Nomoto 2004). If an ONeMg WD accretes ONe or CO material from its companion during RLOF, the WD will collapse to a NS when its mass is larger than the ECSN critical mass and smaller than the maximum NS mass set by BSE, above which it will become a BH. The last is the merger-induced collapse (MIC) of merger or collision between two WDs (Saio & Nomoto 1985). They can be either ONe WDs or CO WDs. In the three cases, AIC generally produces NSs in binaries, which can easily lead to subsequent RLOF and the emergence of LMXBs and MSPs. It is important to note that the mass range for EC-SNe progenitors is under debate (see, e.g., Poelarends et al. 2017, and references therein). However, changing the mass range for EIC in our models has only a small effect on the number of retained NSs and MSPs at late times. This is because, by assumption, the ECSN kicks are not mass dependent. In general, since the mass range for helium cores to collapse in ECSNe is narrow, small changes to the maximum or minimum of the EIC mass range will not affect the overall number of EIC NSs and MSPs much. Thus we simply adopt the SSE prescription as described above. Influence of Dynamics on MSP Formation After their formation at early times, most of the NSs in our models remain as young pulsars in isolation until 12 Gyr. Their magnetic fields and spin rates slowly decrease until they die as pulsars. Dynamical interactions are essential in producing MSPs throughout the evolutionary histories of GCs (e.g., Ivanova et al. 2008 high stellar density regions. During frequent stellar encounters, single NSs can acquire companions through exchanges, and wide binaries can be hardened by repeated interactions. NSs in binaries can then be spun up to MSPs through mass transfer. Figure 1 illustrates two examples of the dynamical interaction histories of MSPs. The progenitor NS of a MSP on the right of the figure was formed in EIC at 0.11 Gyr. It had a magnetic field of 6 × 10 12 G and a spin period of 320 ms at birth. The pulsar experienced its first encounter at 5.3 Gyr and acquired a WD companion through a binary-single exchange. A second binary-single exchange replaced the WD companion to another WD. The binary then experienced a few binarysingle and binary-binary scatterings. During this time, there was no mass transfer, and the magnetic field and spin rates of the pulsar were decreasing (solid black line in Fig. 2). A third binary-single exchange encounter gave the pulsar a MS star companion. The MS star spun up the pulsar via RLOF-driven mass transfer (the first red-dashed line in Fig. 2). At this time it was not yet a MSP (not all mass transfer leads to MSPs, only extended periods of stable mass transfer can produce MSPs). The MS star later evolved to a giant star and the binary underwent a common-envelope phase (blue line in Fig. 2), which circularized and shrank the binary orbit. The remnant of the MS star was a WD which continued to fill its Roche lobe and spun up the pulsar to a MSP (The second red-dashed line in Fig. 2). The final system has a MSP with a 5.9×10 7 G magnetic field and a 1.57 ms spin period, and a companion with mass 0.0075 M at 12 Gyr. Note that the companion mass is very small because it has been depleted through the extended period of mass transfer that spun up its pulsar companion, as in "black widow" type binary MSPs (e.g., Rasio et al. 2000). The MSP on the left of the figure has a different evolutionary path. The NS was not formed until about 6.6 Gyr in AIC as a member of a double WD binary, which itself was formed through a series of dynamical interactions, including both binary-single scatterings and exchanges. After the formation of the NS there were no stellar encounters and the WD companion kept transferring mass to spin up the NS to a MSP. The MSP at 12 Gyr has a magnetic field of 7.5 × 10 7 G and a spin period of 1.56 ms, again with a low-mass companion of mass about 0.0075 M . Figures 2 and 3 show the evolution of magnetic fields and spin periods of the same two MSPs as in Figure 1. The black dashed line is the death line, below which radio pulsars do not have enough energy to support pair production. We use B P 2 = 0.17 × 10 12 G s −2 (4) (Ruderman & Sutherlandt 1975;Bhattacharya et al. 1992) for the death line. Note that the exact location of the death line is still under debate (see, e.g., Zhang et al. 2000;Zhang 2002;Zhou et al. 2017). In Figure 2, for the first 6 Gyr the magnetic field of the pulsar decreased due to magnetic field decay, and the spin period increased slowly due to magnetic dipole radiation. The pulsar exchanged into a binary at 6.31 Gyr with a MS star companion, and was accreting material from the MS star for about 200 Myr, causing it to spin up from angular momentum transfer. During this time the pulsar temporarily went below the death line. The MS star then evolved to a giant star and went through a common envelope phase, during which the pulsar was not accreting mass. The binary orbit was circularized and shrunk by the common envelope. At about 6.7 Gyr, the MS star became a WD and started mass transferring, continuing to spin up the pulsar. The pulsar emerged out of the "graveyard" and was spun-up to a MSP at late times. Figure 3 is similar to Figure 2 but for the other MSP on the left side of Figure 1. There were no dynamical interactions between the time the NS formed and 12 Gyr. In this case the NS-WD binary stayed intact after AIC and the WD continued through RLOF and spun the pulsar up to a MSP. A few studies have suggested that NSs formed in AIC are unlikely to be spun up to MSPs in the same binary (e.g., Verbunt & Hut 1987), because MSPs formed in this way cannot explain the overabundance of LMXBs and MSPs in GCs. However, this is in contrast to the results from our models. In the progenitor binaries of AIC NS, the companions may fill the Roche-lobe and transfer mass onto the ONe WD, which likely leads to common envelope evolution in BSE. The common envelope evolution circularizes and shrinks the binary orbits, producing tight (semi-major axis about 0.001 AU) WD-WD or NS-WD binaries. Because the binaries are so tight, the same WD companions can continue to fill the Roche lobe and spin up the NSs to MSPs. RESULTS In total, there are 208 pulsars in all our models. At birth, about 10% of all NSs are retained in N = 8 × 10 5 models, and about 20% are retained in the N = 3.5×10 6 model. Of all the retained NSs, about 5% are CCSN NSs in N = 8 × 10 5 models; while more than 90% are from ECSNe. For the large-N model, about 15% of all the retained NSs are formed in CCSNe, and about 80% are formed in ECSNe. For models with fewer retained BHs, a larger fraction of NSs retained at late times are formed through dynamically influenced channels (AIC, MIC, merger): about 5% of the NSs that are retained in BH-rich (models 1-10) and BH-intermediate models (models 11-19) and more than 10% in BH-poor models (models 20-26) are formed through AIC, MIC or merger. Because the CCSN NSs make up a small fraction of all NSs retained at late times, this class of NSs does not contribute significantly to the MSP population at late times. In contrast, about 26% and about 67% of all MSPs are formed through EIC and AIC, respectively. Metallicity has a minor effect on the number of retained pulsars in clusters and is unlikely to affect the anti-correlation between the number of retained BHs and MSPs (Sec. 3.1). We intend to perform a more detailed study of the effects of metallicity in future works. BH-MSP Anti-correlation We find a clear anti-correlation between the number of retained BHs and the number of MSPs in our models. Figure 4 shows the number of BHs and MSPs between 9 to 12 Gyr in models 1-25. We include multiple points from the same model sampled at different times, so a model can have different number of BHs and MSPs in the figure. For example, there are 4 models at N BH = 1, showing a range of N M SP the models can have for the same BH number. For cluster models with only a few retained BHs, there can be as many as 16 MSPs in these models; while for cluster models with more than 200 BHs, there are at most 2 MSPs. In the large-N model (model 26) there are 85 MSPs and 20 BHs at 12 Gyr, which we do not include in the figure, simply so that we can isolate the BH-MSP relation for models with similar N . It is worth noting that if the AIC MSPs are excluded from Figure 4, the anti-correlation between the number of BHs and MSPs still holds. This anti-correlation was anticipated, based on our understanding of BH populations in GCs. Figure 5 shows how the BHs dynamically influence the NSs including the MSPs in the clusters. As long as many are present, BHs dominate the cluster cores because of mass segregation and they prevent the NSs from concentrating in the high-density central region. Furthermore, GCs with more retained BHs have lower core densities due to the heating of the cores from BH interactions (Fragione et al. 2018b). The upper panel of Figure 5 shows the distribution of the 2D-projected radii of all NSs (step histograms) and only MSPs (filled histograms) in models 1-25. For models with fewer than 10 BHs, the NSs are located closer to the GC centers; while for models with a large number of BHs, the projected radii for most of the NSs are about an order of magnitude larger. This also affects the number of encounters the NSs can have during the cluster evolution. The stellar densities are higher towards the cluster centers, where the average stellar densities within the median 2D-projected radii (Fig. 5) for the BH-rich, BH-intermediate and BH-poor models are about 6.6 × 10 3 pc −3 , about 8.9 × 10 4 pc −3 and about 1.0 × 10 6 pc −3 , respectively. As a result, NSs located closer to the centers go through more dynamical interactions. Therefore NSs in the BH-poor models are more likely to acquire companions and accrete mass, and thus, a larger chance to become MSPs at late times. The numbers of encounters of all NSs and MSPs are shown in the lower panel of Figure 5. This trend can also be seen in the upper panel, where NSs in the BH-poor models are scattered more and have a wider radial distribution, in contrast to NSs in the BH-rich models. Spin Period and Spin Period Derivative We plot all the pulsars in our models on top of all the observed pulsars in the GC pulsar catalog 5 in Figure 6. The intrinsic spin period derivatives of the model pulsars are derived from their magnetic fields usingṖ = K B 2 P (see Sec. 2.2.1). In the figure, the blue dots show the spin periods and spin period derivatives of the observed GC pulsars and the orange dots show the intrinsic P anḋ P of model pulsars. We calculate the maximum accelerations the pulsars can have from the cluster potentials using Eq.(2.5) in Phinney (1992). The upper and lower limits of the"observed"Ṗ of model pulsars with accel- Figure 5. Top: 2D-projected radial distribution of NSs in models 1-25. The yellow, red and blue distributions mark the BH-poor, BH-intermediate and BH-rich models, respectively. The step histograms are the radial distribution of all the NSs in these models. The medians of the NS radial distribution are 0.25, 0.88, 2.86 pc for the three histograms. It is clear that, for models with a large number of BHs, the NSs are prevented from segregating into the cluster centers by BHs through dynamical interactions; while for models with a small number of BHs, the NSs are much more centrally concentrated. The filled histograms are the radial distribution of the MSPs in the models. The MSP radial distribution shows similar pattern as the radius distribution of all NSs. Bottom: Number of encounters of NSs in models 1-25. Again, step histograms are for all the NSs, and filled histograms are for MSPs only. For models with a smaller number of BHs, the NSs experience more stellar encounters because they are closer to the center, where stellar densities are high. This can also be seen in the top figure, where the histogram of the BH-poor models are wider, showing the NSs in these models get scattered more. erations are shown as green circles. We also include a few pulsars in 47 Tuc with derived intrinsicṖ and in Terzan 5 with inferred magnetic fields in Figure 6, shown by red stars and triangles, respectively Prager et al. 2017 There are both MSPs (P 30 ms) and young pulsars in our models (Fig. 6), as are observed in GCs. Most of the pulsars observed in Galactic GCs are MSPs, since MSPs have long lifetimes and can exist in GCs for more than a few Gyr. In contrast, young pulsars have relatively short lifetimes, and those that formed at early times of the GC histories are already dead. However, through dynamical interactions such as collision between a MS star and a WD, young pulsars can be formed at the present time in GCs. Almost all young pulsars in our models were formed by collisions at late times ( Fig. 6), including newborn NSs formed by collisions between WD-MS star or WD-WD, and old NSs that partially accreted during the collisions between NS-MS star or NS-giant star. It is suggested that young pulsars in GCs are formed by partial recycling of NSs in binaries (Verbunt & Freire 2014). However, partiallyrecycled young pulsars are not happening at a significant rate in our models: only 3 young pulsars in our models were formed by partially spinning up of NSs in binaries. The spin periods and spin period derivatives of the model pulsars overlap with the GC pulsars, which demonstrates that the evolution of magnetic fields and spin periods in our models match well with the observations. DNSs and NS-BH Binaries DNSs and NS-BH binaries may serve as valuable probes for general relativity and the mergers of these binaries can now be detected as gravitational wave sources and can power short gamma-ray bursts visible throughout the universe (e.g., Belczynski et al. 2002a;Clausen et al. 2014). Such systems may form in a variety of astrophysical environments, including through isolated binary evolution in the galactic field (e.g., Belczynski et al. 2002b), as well as dynamical evolution in galactic nuclei (e.g., Fragione et al. 2018a) and star clusters (e.g., Sigurdsson & Phinney 1995). So far only one DNS system has been identified in a cluster, PSR 2127+11 C in M15 6 . M15C has a spin period of 30.5 ms and a derived magnetic field of 1.2×10 10 G. Both the pulsar and its NS companion have masses 1.35 M . This DNS system has an 8-hour orbital period and a highly eccentric orbit with e = 0.68 (Anderson et al. 1990;Prince et al. 1991;Deich & Kulkarni 1996;Jacoby et al. 2006). The projected radius of M15C from the center of M15 is 2.7 pc, which, combined with its eccentricity, indicates that it was most likely formed by an exchange interaction in the cluster core with recoil to its current location (Phinney & Sigurdsson 1991). DNSs are rare in clusters because of natal kicks of NSs that breaks NS binaries or ejects them out of the clusters. One way of producing these systems in GCs is through stellar encounters. M15 is "core-collapsed", i.e., has an extremely high stellar density at the center (Harris 1996(Harris , 2010, which provides a good environment for DNS formation. NS-BH binaries should be even more rare than DNSs in GCs. Indeed, in clusters with many BHs, mass segregation and the strong heating by BH interactions prevent the NSs from interacting with the BHs (see Fig. 5), while in clusters with few BHs, NSs have few interactions with BHs compared to other stars. It is therefore not surprising that no NS-BH binary has ever been detected. Based on multimass King model simulations of the dynamical evolution of BH binaries in fixed GC backgrounds, Clausen et al. (2014) estimate that there are at maximum ∼ 10 BH-MSP binaries in the entire Milky Way GC system. We plan to explore more of the formation and merger rates of NS-BH binaries with our full and self-consistent Nbody simulations of GCs in future studies. We found that several NS-BH binaries and DNSs formed in our models. The last two columns of Table 1 show the total number of each that appears between 9 and 12 Gyr in our models. Two of these DNSs contain MSPs. 32 DNSs and 6 NS-BH binaries contain young pulsars. Most of the NS-BH binaries and DNSs in our models appeared for a short time (< 100 Myr) and were disrupted by dynamical encounters. However there are also a few that survived for longer times. For example, in the large-N model there is a long-surviving DNS formed through binary-binary exchange with a MSP of 1.32 M and a NS of 1.24 M . The binary has eccentricity of 0.05 and orbital period of about 8 min. In total, we identify 40 unique NS-BH binaries and 93 unique DNSs in our models. Of these, 11 NS-BH binaries and 22 DNSs have gravitational-radiation inspiral times less than a Hubble time. If such binaries merge in the local universe, they may be observed as gravitational-wave sources by detectors such as LIGO/Virgo (Abbott et al. 2017) and as sources for short-hard gamma-ray bursts (Meszaros 2006;Lee & Ramirez-Ruiz 2007). A more detailed study of the merger rates of NS-BH binaries and DNSs in GCs will be presented in a later study. DISCUSSION AND SUMMARY It is well understood that pulsar observations are limited by selection effects. The large distances of most GCs from Earth, Doppler shifts in short-period binaries, and dispersion of signals by the interstellar medium all make it difficult to detect radio pulsars. The luminosity function of GC pulsars, d log N ≈ −d log L (Hessels et al. 2007;Hui et al. 2010), suggests that some lowluminosity pulsars may remain unobserved. Given these observational biases, the true number of pulsars in each cluster is somewhat uncertain. Many studies have attempted to constrain the true population of radio pulsars in GCs Hui et al. 2010;Bagchi et al. 2011;Chennamangalam et al. 2013;Turk & Lorimer 2013). For example, the predicted numbers of potentially observable pulsars in Terzan 5 and 47 Tuc are about 150 and 80, respectively, with large error bars (∼ 50%) (Bagchi et al. 2011;Chennamangalam et al. 2013). Assuming the beaming fraction to be about 50%, there are around 70 potentially observable pulsars in our model 26 (Table 1), a very promising result since this model has a mass at 12 Gyr close to that of 47 Tuc (e.g., Giersz & Heggie 2011). There are 5 conspicuously sub-millisecond pulsars (sub-MSPs) in our models, as shown in Figure 6. The progenitor NSs were exchanged into binaries with eccentric orbits and with MS star or giant star companions in dynamical interactions, and they were spun up all the way to these very short spin periods through mass transfer during the evolution of their companion stars. However, there is no sub-MSP observed in any GC or in the Galactic field. Since we do not assume a lower limit for the NS spin period, sub-MSPs can appear in our models through continued mass transfer. The lack of sub-MSPs in Nature indicates that some physical mechanism not included in our models, such as gravitational radiation (see e.g. Bildsten 1998;Chakrabarty et al. 2003), may set constraints on how fast a pulsar can be spinning. Additionally, many of our model MSPs are still accreting (at very low masstransfer rates;Ṁ 10 −11 M yr −1 ) from extremely low-mass ( 0.01 M ) WD companions at late times. These binaries could potentially be observed as either radio pulsars or LMXBs. Since we cannot really distinguish between these two phases or model in any detail the transition with BSE, we have treated them as radio MSPs in this paper. A more detailed treatment of mass transfer onto NSs from low-mass companions, together with better constraints on sub-MSPs and selection effects, will be discussed in future works. In general, about 20% of all MSPs in our models are single. Of the remaining ∼ 80% found in binaries, about 75% of MSPs have WD companions and about 5% have MS star companions. Observationally, there are 17 black widow systems and 12 red back systems in the GC pulsar catalog. Assuming the ratio between the number of black widows and red backs is not affected by the selection effects, the fraction of MSP binaries with H-rich companions (red backs) in our models are low, and the fraction of MSP binaries with low-mass WD companions (black widows) are high, compared to observations. As this paper is our first attempt at studying the formation of MSPs within our CMC code, for simplicity, we do not explore the various uncertainties associated with the treatment of binary evolution in BSE, and instead focus on general trends (for example the anti-correlation between MSPs and BHs). We do get good agreement with observations, in the sense that our models produce reasonable numbers of all observed type of systems (single vs. binary MSPs, very low-mass companions vs. MS star companions, slow vs. fast pulsars). More closely matching observations would require a more sophisticated treatment of the binary evolution physics, which is beyond the scope of this analysis, but will be considered in future works. To summarize, in this study, we have simulated pulsars in GCs using the CMC code. We updated BSE to incorporate ECSN for NSs which includes three formation mechanisms: EIC for single stars and AIC and MIC for binary stars. For ECSN NSs, we applied low natal kicks with σ ECSN = 20 km s −1 . They are the major source for MSPs in our GC models. We also incorpo-rated magnetic field and spin period evolution of NSs in BSE to model realistic pulsars. In our models, magnetic field evolution follows simple prescriptions for magnetic field decay or "burying" of the field from mass accretion. Spin periods then evolve according to the change of magnetic field and angular momentum of the NSs. Dynamical interactions are clearly the key to understanding the formation of MSPs in GCs. Stellar dynamics can enhance the formation of NSs through AIC and MIC, which are likely to be retained in GCs, and enhance mass transfer in binaries with NS primaries. Furthermore, binaries can be harden or get to higher eccentricities through scatterings, which leads to mass accretion and spinning up of NSs. We have studied the pulsar population in 26 globular cluster models. We found that the number of MSPs is anti-correlated with the number of BHs in GCs. This results from the dynamical coupling of BHs and NSs in our models, where most of the pulsars are formed dynamically, including the few young pulsars that are still active at late times. Additionally, NS-BH and NS-NS binaries are also more readily formed in models without too many BHs. According to our study, the large specific abundances of MSPs in GCs relative to the Galactic field can be explained naturally by dynamical formation. This also provides a way to estimate the number of BHs in GCs given the number of observed pulsars. Furthermore, the spin periods and spin period derivatives of our model pulsars agree reasonably with observations, showing that CMC is able to model realistic pulsar evolution. Our code can also produce a reasonable number of pulsars in the large-N model, opening the door to future detailed modeling of very large GCs such as 47 Tuc (Giersz & Heggie 2011). Our work was supported by NASA ATP Grant NNX14AP92G and NSF Grant AST-1716762 at Northwestern University. Computations were supported in part through the resources and staff contributions provided for the Quest high-performance computing facility at Northwestern. C.S.Y. acknowledges support from the NSF GK-12 Fellowship Program under Grant DGE-0948017. K.K. acknowledges support from the NSF Graduate Research Fellowship Program under Grant DGE-1324585. S.C. acknowledges support from CIERA, and from NASA through Chandra Award TM5-16004X/NAS8-03060, issued by the Chandra Xray Observatory Center (operated by the Smithsonian Astrophysical Observatory for and on behalf of NASA under contract NAS8-03060). C.L.R. is supported by a Pappalardo Postdoctoral Fellowship at MIT. F.A.R. acknowledges support from NSF Grant PHY-1607611 while at the Aspen Center for Physics.	2019-02-15T19:00:36.000Z	2019-02-15T00:00:00.000	{ "year": 2019, "sha1": "7ac4b162029356319c9277d7b8251b327c6858b8", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1902.05963", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "ab3a9969596f8cafa96d00275d78b916e6ce2bf6", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
53558096	pes2o/s2orc	v3-fos-license	(2+)-replication and the Baby Monster The definitions of replicable and completely replicable functions are intimately related to the Hecke operators for the modular group. We define the notions of"$(2+)$-replicable"and"completely $(2+)$-replicable"functions by considering the Hecke operators for $\Gamma_0(2)^+$. We prove that the McKay-Thompson series for $2\cdot\mathbb{B}$, as computed by H\"ohn, are completely $(2+)$-replicable. Introduction The monstrous moonshine conjectures of Conway and Norton [8], led, via rapid developments in VOA theory [2,16] and generalized Kac-Moody algebras [3], to Borcherds' proof of the conjectures [4]. In this paper our aim is to generalize one aspect of this proof to the case of the baby monster, B, namely the connection between an appropriate generalization of "complete replication" of modular functions and the power map structure of 2 · B. To explain this connection, recall the monstrous case. For this, Norton [22], introduced the idea of replicable and completely replicable functions. A formal q-series f = q −1 +c 1 q +c 2 q 2 +· · · with rational coefficients is said to be replicable if and only if there exist formal q-series f (a) = q −1 + c (a) 1 q + c (a) 2 q 2 + · · · , a = 1, 2, 3, . . . such that ad=n 0≤b<d The Hauptmoduls for genus zero groups with rational integer coefficients are known to be replicable functions [11]. The genus zero congruence subgroups of PSL(2, R) have been classified in [9] and the completely replicable functions (with rational integer coefficients) were computed in [1]. The connection with the monster, M, is that the trace functions of the action of M on V are completely replicable as a consequence of the twisted denominator formula for the monster Lie algebra [4,8,15]. Thus we have the surprising fact that the relatively simple characterization of completely replicable functions in a sense captures the power map structure of the monster. These facts have been generalized. In particular Norton [23] introduced generalized moonshine by associating functions to pairs of commuting elements in M. There has been considerable progress in understanding this phenomenon and Carnahan has announced a proof of the generalized moonshine conjectures [5,6,7]. Here we take a complementary approach which aims to understand and extend complete replicability to the case of the baby monster. The idea is that the replication equations (1.1) were discovered by Conway and Norton by modifying the Hecke operators for PSL(2, Z) to take into account the conjectured trace functions on classes of M other than the class of the identity. Instead we start here with the Hecke operators for Γ 0 (2) + and then look for a natural way to introduce replication and complete replication in this context (see [8] for notation). The motivation is that 2·B is the centralizer in M of an element from the conjugacy class labeled 2A in the atlas and in monstrous moonshine the class 2A is associated with the Hauptmodul for Γ 0 (2) + . This is done in Section 2 and we find a different form of replicability and prove that it reflects the power map structure in 2 · B. We call it (2+)-replicability as it is motivated by the Hecke operators of Γ 0 (2) + (which is denoted by (2+)in Conway and Norton's notation). Although the functional equations we use are implicit in the work of Carnahan [5,6,7] and Borcherds [4, Section 10], we find that they have a natural interpretation as (2+)-replication identities. Our main motivation for this approach is that Höhn [18] has given a proof of the baby moonshine conjectures which closely parallels Borcherds' monstrous proof, but with a method which uses replicability rather than complete replicability. Introducing complete (2+)-replicability allows us to modify part of Höhn's proof so as to clone Borcherds' use of complete replicability in his monstrous proof. To do this, in Section 3 we show that a large class of Hauptmoduls are completely (2+)-replicable by working out candidate replicates. This class includes all but 13 of the Hauptmoduls which occur in moonshine for the baby monster. In Section 4 we prove that a completely (2+)-replicable function is completely determined by the first 5 coefficients of the function and its replicates. In the final Section 5 we prove that the baby monster McKay-Thompson series are completely (2+)-replicable with (2+)-replicability respecting the power map structure in 2 · B. This is sufficient to establish that McKay-Thompson series are Hauptmoduls once their first 5 coefficients are calculated, except for the 13 exceptions noted above. For these 13 cases our proof is similar that of Höhn, but simplified somewhat since we have additional recurrence relations satisfied by the McKay-Thompson series. This establishes, incidentally, that these exceptions are also (2+)-replicable -however we have not found a way to include them in the cases covered in Section 3. More generally, it is our belief that the notions of (2+)-replicability and complete (2+)replicability will lead to a better understanding of the connections between moonshine and Hecke operators. Let G be a discrete subgroup of PSL(2, R) which is commensurable with PSL(2, Z). We consider F(G) the set of functions meromorphic on the upper half plane and at cusps that are invariant under the action of G. For each element α ∈ PSL(2, R) in the commensurator of G we define a Hecke operator T α : F(G) −→ F(G) in the following way. Consider a decomposition GαG = n j=1 Gγ j as a disjoint union, guaranteed to be finite as α is in the commensurator of G. Define An important example is given by the Hecke operators of G = Γ = PSL(2, Z) (see, for example, [25, pp. 60-63]. Let ∆ = {α ∈ M 2 (Z) \| det(α) > 0} and letT m be the sum over all double cosets ΓαΓ with α ∈ ∆ and det(α) = m. Also define T a,d to be the double coset Γ a 0 0 d Γ and set T m = T 1,m . From the theory of elementary divisors we then have formulaT m = T a,d where the sum is over all positive a and d such that ad = m such that a divides d. As explained above, the Hecke action of these (distinct) double cosets on modular functions is obtained by expressing the double cosets as a union of left cosets (by a slight abuse of notation we use the same symbol for both a sum of double cosets and the corresponding operator). For example, if f is a modular function for Γ theñ The relationship to the Conway-Norton replication formulas of "classical" moonshine is as follows. When f = f g is a Thompson-McKay series for an element g in the monster, the Conway-Norton replication formula corresponding to (2.1) is where P 4,f (f ) is the 4th. Faber polynomial of f and the functions f (4) = f g 4 and f (2) = f g 2 are the Thompson-McKay series for the monstrous elements g 4 and g 2 respectively. Thus the left side of this replication formula is a modification of the Hecke action ofT 4 . So in this sense the Conway-Norton replication identities say that the result of a modified Hecke operator acting on a Thompson-McKay series is equal to a Faber polynomial in that series. We will show that a similar phenomenon occurs if we replace the monster by 2 · B and PSL(2, Z) by Γ 0 (2) + . For the Γ 0 (2) + case we set Also define T a,d to be the double coset Γ 0 (2) + a 0 0 d Γ 0 (2) + and T m = T m,1 . As we will seẽ T m = T a,d where in this case the sum is over all positive integers a and d such that a divides d and either ad = m or ad = m/2 (the second option being absent if m is odd). As for the modular group we have an action on functions. For examplẽ where we have used Proposition 2.1 below. As we will see, there is a "replication formula" for 2 · B which involves a modification of (2.4) in the same way that (2.2) involves a modification of (2.1). It is The novelty here is not in the existence of these identities, as mentioned above they are implicit in previous work. Nor is it the existence of the (2+)-replicates, which, as we shall see, are trace functions on appropriate VOA-modules. Rather we believe that the key points are firstly that a form of complete replication is restored by introducing "half step" replicates as [2] (which explains our chosen normalization for the exponents of (2+)-replicates). The absence of complete replicability is a complicating factor for groups other than the monster and introducing complete (2+)-replicability means that we can give a modified version of Höhn's 2 · B proof which is closer to Borcherds' monstrous proof. The second key point is that this approach emphasizes a close connection between replication identities and Hecke algebras of groups other than PSL(2, Z). A connection which, we believe, may prove fruitful. The purpose of the next section is to define (2+)-replicability based on the Hecke operators for Γ 0 (2) + . We then define complete (2+)-replicability and we will see that the set of completely (2+)-replicable functions includes the McKay-Thompson series for the baby monster group. Hecke operators for Γ 0 (2) + In this subsection we find expressions for the Hecke operators of Γ 0 (2) + . These will be used in Section 2.2 to motivate the definition of (2+)-replication. Let m be a positive integer. We define the following sets The main result of this subsection is the following Proposition 2.1. Let m be an even positive integer then we have the following The proof will proceed by several lemmas. Until Proposition 2.8, m will always be a positive even integer. Proof . For any γ and γ in M m we have that r = γ γ −1 fixes ∞. So if r is in Γ 0 (2) + it must be of the form ± 1 k 0 1 . A short calculation now shows that γ = γ . In what follows, for a positive integer n we define ψ(n) = n p\|n p prime Proof . First observe that each element of M m 1 has the form d y where d is a divisor of b (and hence odd) and 0 ≤ y < 2 a (b/d). The number of such matrices is 2 a times the number of matrices of the form By the standard theory of Hecke operators for the modular groups, the number of these matrices is ψ(b), the index of Γ 0 (b) in the modular group, and so the first equation follows. For the second equation, if w is in Γ 0 (2) + we have that if and only if w is in Γ 0 (2m). Since conjugation preserves areas of fundamental domains, it preserves indexes and so the required index is that of Γ 0 (2m) in Γ 0 (2) which is ψ(2m)/3 = ψ(2 a+1 b)/3 = 2 a ψ(b) as required. For the latter we have If g = gcd(x+2y, 2z) then g is odd since x is odd. Let a and b be such that a2z +b(x+2y) = g. Then r = b a −(2z/g) (x+2y)/g is in Γ 0 (2) and we have Up to left multiplication by a translation this last element is in M m 1 , as required. Lemma 2.5. Part 1 of Proposition 2.1 holds: Proof . By Lemma 2.4 and the fact that is a union of Γ 0 (2) + − Γ 0 (2) double cosets which includes the double coset Γ 0 (2) + [ 1 0 0 m ] Γ 0 (2). According to [25,Proposition 3.1], if G 1 and G 2 are commensurable subgroups of a group G then for each g in G the double coset G 1 gG 2 is a union of disjoint single cosets with S m 1 and S m 2 defined as in (2.6). We will find the sizes of these two sets. A typical element of S m 1 has the form 2 a (b/s) y 0 s where s is a divisor of b and hence odd. The elements of this matrix are coprime if and only if the elements of the matrix (b/s) y 0 s are coprime and there is a one to one correspondence between these two sets of matrices. But the number of matrices of the latter form is ψ(b) and so this is the size of S m 1 . To count the elements of S m 2 consider a typical element, which has the form 2 −1/2 (2m/z) y 0 z with z an even divisor of 2m such that 2m/z is even and y is such that 0 ≤ y < z and the entries are coprime. If we ignore the condition that z and 2m/z are even, there are ψ(2m) such matrices. But this over-counts by matrices of the form 2 Proof . By Lemmas 2.2 and 2.6 it suffices to show that each γw 2 with γ in M m 2 is equal, up to left multiplication by elements of Γ 0 (2) + , to an element of M m 1 . Case I: Suppose γ = [ x y 0 z ] is in S m 1 . Then w −1 2 γw 2 = z 0 −2y x . Let g = gcd(z, 2y) which is odd since z is odd, and let a and b be such that az − b2y = g. Then r = a b 2y/g z/g is in Γ 0 (2) and r z 0 −2y x = g bx 0 m/g which, up to a left translation, is in M m 1 . . Let g = gcd(z, y) which is odd since y is odd, and let a and b be such that az + by = g with a even (which is possible since y/g is odd). Then r = b a −z/g y/g is in Γ 0 (2) and r y −x/2 z 0 = g −bx/2 0 m/g which again, up to a left translation, is in M m 1 . Proof of Proposition 2.1. Part 1 of the Proposition was proved in Lemma 2.5. Part 2 follows from where the first equality follows from Lemma 2.7 and the fact that w 2 2 = −1 2 . Finally, part 3 follows from (2+)-Replication and the Baby Monster where we have used parts 1 and 2. For the case where m is odd we have from the definition (2.6) above that We then have the following: Proposition 2.8. Let m be an odd positive integer then Proof . The proof is similar, but simpler, than the proof of Proposition 2.1 and we omit it. We now wish to introduce Hecke operators for Γ 0 (2) + . In the introduction to this section, we defined Hecke operators by starting with double cosets and then writing these double cosets as single cosets. We will find it convenient here to reverse this ordering and start by defining operatorsT m and T m by their actions and then deducing their characterizations as double cosets given in (2.3). Moreover, to further simplify the presentation we only consider actions on functions (i.e., weight zero modular forms) so that the action of T a,d is equal to that of T a/d,1 which, as mentioned earlier, we write as T a/d . Thus we make the following definitions: Definition 2.9. Let m be a positive integer x,z even 0≤y<z f xτ + y z , m even. By Proposition 2.1 this definition of the action T m is the same as the action of the double coset Γ 0 (2) + [ m 0 0 1 ] Γ 0 (2) + as required. Moreover, if we observe that every element of ∆ 2 is, up to left multiplication by an element of Γ 0 (2) + , equivalent to one of the matrices which occur on the right side of part 2 of Definition 2.9 and that these matrices represent distinct left cosets of Γ 0 (2) + , then it is straightforward to verify that this definition ofT m is equivalent to that give in (2.3). The relationship between T m andT m is given by the following: Proposition 2.10. Let m = 2 α β be a positive integer where β is odd, theñ Proof . Define first the following operators Then we havẽ since when m is even the second term ofT m excludes the cases when x is odd and when z is odd which are disjoint. We also have since the cases when x, y and z are all even, when x is odd and when z is odd must be excluded from the sum in R 4m and these are disjoint. If m is odd then since in this case either x or z is odd (but not both). For T m we have the following This follows from the terms in M m in each case and the fact that the condition that the gcd of the entries is 1 for each element of M m . In particular, this establishes the required result if m is odd. We also havẽ This follows from (2.7) as follows. If m is even theñ using (2.9). To establish the proposition if m is even, we have from (2.10) The result now follows by induction on α where m = 2 α β with β odd. (2+)-replicability The operators T m andT m just defined map the field of modular functions for Γ 0 (2) + to itself. If T 2A is the normalized Hauptmodul of Γ 0 (2) + this means that T m (T 2A )(z) is a rational function of T 2A (z) and since T m (T 2A )(z) has no poles in the upper half-plane this rational function is actually a polynomial. From the power series expansion we can see that it has to be the m-th Faber polynomial of T 2A . We have just said that As discussed in the introduction, we will show that the situation is entirely analogous to the situation with PSL(2, Z), the monster and "ordinary" replication. For Γ 0 (2) + and the baby monster we will show that the appropriate generalization is the following: We make a few remarks on this definition. Remark 2.12. Given a (2+)-replicable function f , its (2+)-replicates are not determined uniquely. For example, for m = 2, equation (2.12) becomes and we can see that f [2] is known when f and f [ √ 2] are known. Also, the odd-power coefficients of the replicates f [ √ 2n] can be changed freely and identity (2.12) is still true. We will see instances of (2+)-replicable functions that are (2+)-replicable in different ways, i.e., have different (2+)replicates. For example, in Table 3 the trace functions for the 2·B classes 2e and 4d are both T 4C , the Hauptmodul for Γ 0 (4) which corresponds to the class 4C in the monster. However, for the classes 2e and 4d, f [ √ 2] is given by the monstrous functions T 2B and T 4C respectively. In other words there are two (2+)-replication structures which are compatible with the Hauptmodul T 4C and both of these occur for the baby monster. One might think that for "ordinary" replication there is in principle a similar lack of uniqueness in the definition of replicates. For example, the monstrous classes 27A and 27B have the same trace functions (in fact they are the only rational classes with this property). But both cube to 9B. More generally, as pointed out by Norton [22], the "ordinary" replicates of a replicable function are unique. So for "ordinary" replication we have a unique "power map" structure. Remark 2.13. If a function f is (2+)-replicable then it is replicable with n odd, Also, we can see that if f is replicable and if, for every n even, we can write ] , then f is (2+)-replicable. For n odd we obviously have f [n] = f (n) . This is shown by the following simple manipulation ad=n 0≤b<d Finally, if f is replicable then f is always (2+)-replicable by taking, for example, f [n √ 2] = 0. Note that this is in contrast with ordinary replication for which the replication equations uniquely fix the replicates. Remark 2.14. Because of Remark 2.13 what we are actually interested in is finding the possible (2+)-replicables for a given (2+)-replicable function. For example, we could ask if a (2+)replicable function is completely (2+)-replicable in the sense given below. There are examples of Hauptmoduls that are not complete replicable functions but are completely (2+)-replicable. As we will see, some examples of these are T 4∼b , T 12∼d among many others (see [24] for the notation). For the monster, since replicates corresponds to power maps, the replicable functions satisfy a stronger property called complete replicability. Namely that the replicate functions are themselves replicable and that taking replicates is "commutative". We make a corresponding definition for (2+)-replicability. In the following section we show that certain Hauptmoduls are completely (2+)-replicable. The list will include almost all the Hauptmoduls which occur in moonshine for the baby monster. The proof of (2+)-replicability for the remaining baby monstrous moonshine functions will be given at the end of Section 5. Complete (2+)-replicability and Hauptmoduls The following result is Theorem 5.15 in Ferenbaugh's Ph.D. Thesis [13] and will be useful in proving that certain Hauptmoduls that are not completely replicable are completely (2+)replicable. We refer to [8] and [14] for the notation on the Atkin-Lehner involutions W e and on groups of the form n \| h + e 1 , e 2 , . . . and n h + e 1 , e 2 , . . .. Remark 2.13 gives motivation to find formulas of type This will help us finding (2+)-replicates for certain Hauptmoduls. Before we proceed, we need some notation in order to separate even Atkin-Lehner involutions from the odd ones. From now on, any group n \| h + e, f, . . . will be written as n \| h + O 1 + 2 k O 2 where 2 k is the highest power of 2 that divides n h , the set O 1 is the set of odd Atkin-Lehner involutions of the group and O 2 is the set of even Atkin-Lehner involutions of the group divided by 2 k . We always include 1 in O 1 . For example, the group 30 + 6, 10, 15 will be written as 30 We should also note that the elements of both O 1 and O 2 are odd, and these sets can satisfy does not normalize the genus zero group If it normalizes, we have and this happens exactly when either h is even or h = 1, k ≥ 2 and O 2 is empty. Also, this matrix conjugates the groups 2N + O 1 and 4N + O 1 + 4O 1 onto each other, when (2, N ) = 1. We make the following observation which will be used several times in what follows. Lemma 3.4. We have the following identities: 1. If (2, N ) = 1 then or, equivalently, In particular, if k = 1 and Proof . The first identity is Theorem 3.1 applied to f = T 2N +O 1 +2O 1 and p = 2. The third identity is also a consequence of Theorem 3.1 applied to f = T 2 k+1 N +O 1 +2 k+1 O 2 and p = 2. Since 2 k N , for k ≥ 1, we have where e = 2 k e for some e ∈ O 2 . But and this proves the third identity. The second identity is the third one written in a different way. Proof . We start by proving the result for completely replicable functions and then use this to complete the proof for conjugates of completely replicable functions. For the cases where the function is completely replicable we know how replication works and for each g (2 m ) we will either apply Lemma 3.4 or, in view of Remark 2.13, take g (2 m ) as g [2 m ] and any function satisfying the condition in Remark 3.2 as g ] . This defines the (2+)-replicates g [m] , with m a power of · · · · · · · · · Table 1. (2+)-replicates of Hauptmoduls. [16] · · · · · · · · · Having done so, and because of Remark 2.13 again, it is enough in order to finish the proof in this case to show that, for all positive m and odd n, we have This is a case by case check on every g (2 m ) . 1. Our first case is when we are using some identity from Lemma 3.4 to decompose g (2 m ) . In this case, we must have g (2 m ) = 2 k N + O 1 + 2 k O 2 , i.e., h = 1, and g (2 m n) = g (2 m ] when we apply Lemma 3.4 to decompose g (2 m ) are Applying (n)-replication to every entry in the table we obtain We can see that every row in the table corresponds to some identity from Lemma 3.4 and identity (3.1) is satisfied in this case. ii) If O 2 = ∅ and O 2 = ∅ then the possibilities, depending on k, for g [2 m ] and g Applying (n)-replication to every entry in the table we obtain We can see that every row in the table corresponds to some identity from Lemma 3.4 and identity (3.1) is satisfied in this case. Applying (n)-replication to every entry in the table we obtain We can see that the first row corresponds to an identity from Lemma 3.4 and in the last row the Hauptmodul 2 k+1 N + O 1 satisfies the condition of Remark 3.2. Identity (3.1) is satisfied in both cases. 2. If we are taking g [2 m ] = g (2 m ) and g This completes the proof in the case g is completely replicable. The proof also shows that in Tables 1 and 2 we can take any 2 k N h + O 1 + 2 k O 2 instead of 2 k N \| h + O 1 + 2 k O 2 and the result is still valid. For the remaining cases in the tables, g is not completely replicable, but its invariance group is conjugate by [ 1 α 0 1 ], with α = 1 2 , 1 4 or 1 8 , to some group whose Hauptmodul f is completely replicable. We shall give the proof for the case α = 1 2 as the other two cases are similar. Let g = T G and f = T G α so that g(z) = −f z + 1 2 . Since we are assuming f is completely replicable it follows, as remarked above, that f is (2+)-replicable. So for some -not necessarily unique -(2+)-replicates we have Our aim is to use the relationship between f and g to show that (3.2) implies that g is (2+)-replicable after a suitable choice of (2+)-replicates for f . Now by the definition of the Faber polynomials, for any series h(z) = 1 q + · · · we have P n,h h z + 1 In particular applying this to f and g gives P n,f f z + 1 2 = (−1) n P n,g (g). So for n odd, substituting z by z + 1 2 in (3.2) gives So these replication identities for g are satisfied by choosing g [a] = f [a] 1 2 for a odd. For n even, we again substitute z by z + 1 2 in (3.2). Arranging the resulting right hand side by the highest power of 2 which divides a gives P n,g (g(z)) = ad=n a odd 0≤b<d 2 m] for n ≥ 2 these replication identities for g will be satisfied provided we can show that the i = 0 term ad=n a odd 0≤b<d In particular, substituting z by 2az+b d and summing over b we have This finishes the proof for functions of the form T We now consider g of the form T Substituting z by 2az+b d and summing over b we obtain As mentioned above, the (2+)-replicates of f in (3.2) are not necessarily unique and we are free to make any convenient choices. So we now choose f [ For the second part of the theorem we just note that when n ≥ 2 we have g [2 ] (m) and when n = 0 we have . The remaining cases can be done similarly. For Hauptmoduls of groups of the form G 1 4 (resp. G 1 8 ) it is the summand corresponding to i = 1 (resp. i = 2) that matters. We just note that the replicates f [ 2a] ) with a odd, have a power series expansion with coefficients of even powers of q equal to zero. This means that any other function with the same property will work as well. We just chose those particular ones because they give complete (2+)-replicability. 2 n] and apply the (2+)-replication rules from the theorem/tables we will necessarily have For doing this, we just have to see that after removing the first few entries of any column or applying (m)-replication, with m odd, to a full column, the list we obtain is still a full column in the tables. This is a case by case check and is what guarantees that any replicate g [2 Complete (2+)-replicability and generalized Mahler recurrence relations In this Section we extend some of Martin's results from [21] to (2+)-replication. Namely, we adapt his proof to show that the coefficients of completely (2+)-replicable functions satisfy recurrence relations very similar to those of completely replicable functions. We consider L a field extension of Q containing all roots of unity, the ring K = L . . . , x for r ∈ N ∪ √ 2N, and the polynomials P k,r (t) defined inductively by P 1,r (t) = t and P k,r (t) = k−1 . These are the same recurrence relations that Faber polynomials satisfy. We fix r ∈ N ∪ √ 2N and consider the set of equations indexed by k ≥ 1 ad=k 0≤b<d k , whose kernel contains I. For u ∈ N ∪ √ 2N we define a L-algebra endomorphism ψ u of K letting it fix every element of L and mapping x and also We define an action of ∆ in R M in the following way. For α = [ u v 0 y ] ∈ ∆ we set e α = e, x Definition 4.5. Let n be a positive integer and h(q) ∈ R 1 . We define T n by and call T n a generalized Hecke operator. Both sums are over positive integer numbers which makes the second sum be zero if n is odd. Proposition 4.7. Let l 1 and l 2 be relatively prime positive integers and h(q) ∈ R 1 . Then In particular, T l 1 and T l 2 commute. Proof . This comes from the fact that Proposition 4.8. If p is an odd prime then T p n T p (h(q)) = T p n+1 (h(q)) + pT p n−1 Ψ p (h(q)). If p = 2 then Proof . The case where p is odd is true as the operators T p n agree with the Hecke operator for PSL 2 (Z). For the case p = 2 we see that T 2 n T 2 (h(q)) equals Now, the first and fourth summands equal the second, third and last summands equal and this shows that and the theorem is proven. Corollary 4.9. The algebra generated by the operators T n , for n ∈ N, is commutative. Let l be a fixed prime. We set Q k = T l h(q) k for k ≥ 1, and for b ∈ K I we use b [l] to denote b l * 0 * . We set also Q 0 = 2l + 1, if l = 2, l + 1, if l = 2 and define T k l as the operator that sends R 1 to zero if l does not divide k. We use the same notation to denote both P k,r (t) and its image in K I [t]. If l is odd then and if l = 2 we have Proof . The case where l is odd can be found in [21]. When l = 2 we apply T 2 to both sides of equation (4.2) and the following manipulation If (2, k) = 1 then T 2 T k (h(q)) = T 2k (h(q)) = P 2k,1 (h(q)) and if 2 r is the exact power of 2 that divides k we have that This proves the assertion of the theorem. When l = 2 the Q k are simply the power sum symmetric functions on h [2] . By induction, using the previous result, one shows that every Q j is a polynomial in h [2] (q), h [2] x i 1 · · · x in be the elementary symmetric functions in the indeterminates x 1 , x 2 , x 3 , x 4 , x 5 . We know that the elementary symmetric function are polynomials in the power sum symmetric functions and from this we conclude that the elementary symmetric functions on h [2] ] (−q) and h(q). We use these facts in the proof of the following proposition. Applying the homomorphism E f defined at the beginning of this section we get 1 . Equating the coefficient of q 2 in both sides of the equation we see that a 2 1 + 2a 3 − a k q k , for n ∈ N ∪ √ 2N, then their coefficients satisfy the following recurrence relation: k+1 + a 2 2k + a [2] 2k , Proof . This is a consequence of the following two identities , The baby monster Lie algebra and (2+)-replication The moonshine module V was constructed in [15] by Frenkel, Lepowsky and Meurman. This is a vertex operator algebra that has M, the monster group, as symmetry group A vertex operator algebra is an intricate algebraic structure and we refer to [20] for the definition and the basics of its theory. The moonshine module has a grading V = n≥−1 V (n) and its graded dimension Tr g\|V (n) q n are completely replicable functions. To do this, he uses V to build a generalized Kac-Moody algebra, the monster Lie algebra, whose twisted denominator identity is essentially the statement that the n-th replicate of a T g is T g n . Knowing that Hauptmoduls for genus-zero congruence groups are also completely replicable functions and that completely replicable functions satisfy some recurrence relations that determine a function from the first 5 coefficients of the function and its replicates, he was able to show that every McKay-Thompson series is indeed a Hauptmodul for some genus-zero congruence groups by just comparing the first few coefficients of the functions involved. In [18], Höhn shows that there is a vertex algebra W where 2 · B acts as a symmetry group. This group is a central extension of B, the baby monster group, and it arises as the centralizer of an element of class 2A in M. This vertex operator algebra plays for 2 · B the role that V plays for the M and it was used to prove the generalized moonshine conjectures for the case of the baby monster, i.e., when g in T g,h (see [23] for a precise statement of the generalized moonshine conjectures and what T g,h is) is an involution of type 2A in M. In this section, we use t to represent a (fixed) element in class 2A in M. More precisely, what Höhn states in [18] is the following. If is the t-twisted module and h is an element in the centralizer of t in M then the McKay-Thompson series Tr g\|V ( n 2 ) (t) q n is the Hauptmodul for some genus zero congruence subgroup. We give a very brief sketch of the results from [18]. From the decompositions of V = V 00 V 01 and V (t) = V 10 V 11 of +1 and −1 eigenspaces for t, Höhn builds the vertex algebra W mentioned above on which 2 · B acts. Using this vertex algebra W a Lie algebra g B , the baby monster Lie algebra, is constructed too. This is a 1 2 Z × 1 2 Z-graded Lie algebra that has an action of 2 · B in it that respects the grading and its m 2 , n 2 piece is isomorphic to V [m,n] (2mn) ([·, ·] represents reduction mod 2). Also, V 10 is isomorphic to V 01 as 2 · B-modules. Tr g\|V [1,n] ( n 2 ) We can now state our main result. t,g = T t,g n and T Proof . From the twisted denominator identity for the baby monster Lie algebra (5.1) we have, after substituting p p m − n∈Z Tr g\|V [1,n] ( n 2 ) The right side of this equation is p −1 exp(Z), where Because of the isomorphism between V 01 and V 10 , Tr g i \|V [1,mn] ( mn 2 ) p im q in i − i>0 m∈Z + n∈Z m and n even Tr g i t\|V 00 ( mn 2 ) Tr g i \|V [1,mn] ( mn 2 ) p im q in i − i>0 m∈Z + n∈Z m and n even Tr g i \|V [1,mn] ( mn 2 ) p im q in i − i>0 m∈Z + n∈Z m and n even Tr g i t\|V ( mn 2 ) Tr g i \|V [1,mn] ( mn 2 ) Tr g a \|V [1,kd] ( kd 2 ) Since where P n is the n-th Faber polynomial, we conclude that, for all n ∈ N, ad=n 0≤b<d T t,g a aτ + b d + ad=n d even 0≤b<d and we get that T t,g is (2+)-replicable with (2+)-replicates given as stated in the theorem. By substituting g by g n we see that T t,g n is (2+)-replicable with replicates T [m] t,g n = T t,g mn = T [mn] t,g and T To complete the proof it remains to see that T Equivalently, what we need to prove is that, for every m ∈ N, ad=n 0≤b<d T 1,g ma t aτ + b d + ad=n d even 0≤b<d T t,g 2ma 2aτ + b d = P n T 1,g m t (q) . But since T 1,g m t is a monstrous function we know that ad=n 0≤b<d But now, This identity is clearly true for n odd and for n = 2n even it becomes, because of the isomorphism between V 01 and V 10 , a\|2(n ,k ) , and the theorem is proven. We can now use Theorem 5.1 to reprove Höhn's result Theorem 5.2. The McKay-Thompson series T t,g (z), for g ∈ 2 · B, are the q-expansions of Hauptmoduls. Proof . We know that the McKay-Thompson series T t,g are completely (2+)-replicable and consequently satisfy the recurrence relations from Theorem 4.12. We know that T [n √ 2] t,g = T 1,tg n is a monstrous function and therefore its coefficients are known once we know in what class in the monster the element tg is, for every g ∈ 2 · B. This can be done with GAP [17]. Hence, the first five coefficients of every T t,g determine all the coefficients of the T t,g completely. From Section 3 we also have some completely (2+)-replicability results for some Hauptmoduls and thus these Hauptmoduls satisfy the same recurrence relations from Theorem 4.12. To prove that every McKay-Thompson series is a Hauptmodul it would be enough to compare, for every g ∈ 2·B, the first five coefficients of T t,g , T 1,tg , T t,g 2 , T 1,tg 2 , . . . with those of f, f [ √ 2] , f [2] , f [2 √ 2] , . . ., respectively, for some Hauptmodul f in Tables 1 and 2. However, not all McKay-Thompson series correspond to Hauptmoduls listed in Tables 1 and 2 and, because of that, this method for proving that the McKay-Thompson series are Hauptmoduls works for all 247 classes in 2 · B with 13 exceptions. This happens because the Hauptmodul associated to each of these 13 classes is neither a completely replicable function nor a dash (see [14] for the definition of the dash operator) of a completely replicable function and our Tables 1 and 2 only contain such functions. For those 247 − 13 = 234 classes covered by Tables 1 and 2 we use the decomposition of the first five head characters given in [18]: 2) H 2 = χ 185 , 3) H 3 = 2χ 1 + χ 2 + χ 3 + χ 4 , 4) H 4 = 2χ 185 + χ 186 , 5) H 5 = 3χ 1 + 3χ 2 + 2χ 3 + χ 4 + χ 6 + χ 7 to make the comparison and we obtain a proof of Höhn's results which is analogous to Borcherds proof of the original moonshine conjectures. For the remaining 13 classes (their names are, using GAP notation: 12h, 12i, 20h, 20i, 24i, 24m, 24n, 36d, 36e, 40f , 40g, 60d, 60e) we use again the recurrence relations from Theorem 4.12 to find the first 23 coefficients of their McKay-Thompson series. Since we know that a replicable function (we recall from Remark 2.13 that a (2+)-replicable function is replicable) is completely determined by its first 23 coefficients, a simple check among the power series expansions of the 616 Hauptmoduls for genus zero groups with rational integer coefficients allows us to match every such class with some Hauptmodul (see references [9,14,24] for details of these functions and the corresponding groups). This is analogous to Höhn's proof but the recurrence relations from Theorem 4.12 simplify the computations greatly. A list of 2 · B classes with their corresponding McKay-Thompson series and √ 2 -replicates is given in Table 3 (see [24] for the labelling of the Hauptmoduls). Class of g (of g 2 ) T t,g T t,g = T 1,gt Class of g (of g 2 ) T t,g T t,g = T 1,gt Class of g (of g 2 ) T t,g T	2018-06-19T08:28:12.000Z	2017-10-03T00:00:00.000	{ "year": 2017, "sha1": "de2c4f0f1f415042ef9d271a206cf1605a02a07a", "oa_license": "CCBYSA", "oa_url": "https://www.emis.de/journals/SIGMA/2018/060/sigma18-060.pdf", "oa_status": "GOLD", "pdf_src": "Arxiv", "pdf_hash": "de2c4f0f1f415042ef9d271a206cf1605a02a07a", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
233565342	pes2o/s2orc	v3-fos-license	Formulation and Evaluation of Taste Masking Lercanidipine Hydrochloride Oral Disintegrating Tablets Lercanidipine is an antihypertensive drug. It is a dihydropyridine class of calcium channel blockers. It is extremely bitter. The reason for this exploration was to build up a non-bitter orally breaking down the tablet of inadequately solvent medication viz Lercanidipine. The bitterness of drug, masked through complexing Tulsion 339 in various ratios. Sodium starch glycolate, crospovidone, low substituted hydroxypropyl cellulose selected as super disintegrants in the formulation. The formulated tablets were assessed for various properties like Drug content, crushing strength, friability, wetting time, water retention proportion, breaking downtime and in-vitro disintegration time and dissolution studies. The disintegration time obtained in the range between 38.46-51.40 seconds. Release studies observed between 5 to 30minutes. From the prepared formulations, formulation using Low substituted hydroxypropyl cellulose with 5% concentration showed 98.89% drug release within 30minutes. Thus F9 was considered as best among the other formulations With effective dissolution and improves patient intake. Drug release Kinetic analysis (r) based on best curve itting method for optimized lercandipine formulation showed irst order kinetics proves that the drug release depends upon its concentration. Lercanidipine, Tulsion 339, super disintegrants, oral disintegrating tablets, disintegration time A Lercanidipine is an antihypertensive drug. It is a dihydropyridine class of calcium channel blockers. It is extremely bitter. The reason for this exploration was to build up a non-bitter orally breaking down the tablet of inadequately solvent medication viz Lercanidipine. The bitterness of drug, masked through complexing Tulsion 339 in various ratios. Sodium starch glycolate, crospovidone, low substituted hydroxypropyl cellulose selected as super disintegrants in the formulation. The formulated tablets were assessed for various properties like Drug content, crushing strength, friability, wetting time, water retention proportion, breaking downtime and in-vitro disintegration time and dissolution studies. The disintegration time obtained in the range between 38.46-51.40 seconds. Release studies observed between 5 to 30 minutes. From the prepared formulations, formulation using Low substituted hydroxypropyl cellulose with 5% concentration showed 98.89% drug release within 30minutes. Thus F9 was considered as best among the other formulations With effective dissolution and improves patient intake. Drug release Kinetic analysis (r 2 ) based on best curve itting method for optimized lercandipine formulation showed irst order kinetics proves that the drug release depends upon its concentration. INTRODUCTION The oral route of drug administration has been generally accepted and up to 50-60% of total dosage forms are administered orally. Solid dosage forms viz tablets and capsules are worldwide accepted dosage forms due to its precise dose, self medication, a non-invasive route which makes the solid dosage forms as patient user-friendly. However, the substantial drawbacks of these traditional dosage formulations include dysphagia for pediatric and geriatrics patients. This problem mainly encounters 35% of the general population. These traditional tablets need water for administration. This issue causes dif iculty in swallowing when water is not available. Hence Dispersible tablets plays a dominant role for these purposes, which can quickly dissolve or disintegrate in the oral cavity and have drawn a good interest to the patients (Saini and Garg, 2019). The word "orodispersible tablet" was adapted by European Pharmacopoeia as a tablet to be inserted in the mouth where it easily disappears before swallowing, suggesting maximum DT of 3 min as calculated in a conventional disintegration test apparatus. Other synonyms of ODT includes quick melts, rapid melts, fast dissolving, fast disintegrating, rapid dissolve or mouth dissolving tablets (Mohanachan-dran et al., 2011). The bitter taste of orally administered medicinal products often results in patient non-compliance with the use of medicinal products, especially for children and the elderly. Sadly, most medicines have a natural, bitter taste that can cause a burning sensation in the throat or mouth. In particular, a bitter taste can reduce patient compliance and thus reduce the ef iciency of pharmacotherapy (Suryadevara et al., 2017). The Drug Lercanidipine HCl used in the present study is a type-II biopharmaceutical classi ication system since it has low solubility and high permeability. Its recommended for relief of seasonal allergic rhinitis related symptoms in adults and children 2 years of age and is intended for chronic idiopathic urticaria therapy in adults and children 6 months of age and older (Suresh et al., 2007). Lercanidipine HCl shows low bioavailability so its aqueous solubility should be targeted by a Bioavailability Improvement strategy. It is used in the treatment of Hypertension, due to its selectivity and speci icity on the smooth vascular cells. MATERIALS Lercanidipine and Polacrallin potassium (Tulsion 339) was obtained from Spectrum pharma research solutions, Mumbai, Sodium starch glycolate, Crosspovidone, Low substituted hydroxypropyl cellulose, Sodium hydroxide and Sucralose were obtained from SD ine chemicals, Mumbai, Microcrystalline cellulose, Magnesium stearate, Talc were obtained from Central drug house (p) Ltd, New Delhi, potassium dihydrogen phosphate and sodium hydroxide were obtained from Finar chemicals ltd, Ahmedabad. Drug and Excipient Compatibility by using FTIR The interaction study between the drug and Tulsion 339 and other excipients were performed using FTIR. The pellets were prepared on KBR press. The spectra were recorded over the wavenumber range of 3500 cm −1 .The pictorial optimized formulation shown in Figure 2. Standard calibration curve of pureLercanidipine using U.V. spectroscopy Preparation of standard stock solution Standard stock solution of Lercanidipine was prepared by dissolving accurately weighed 100mg of Lercanidipine in the little quantity of phosphate buffer pH-6.8 in 100ml volumetric lask. The Volume was made up to the mark using the same buffer. From this 10ml was pipette out and Volume was made up to 100 ml with phosphate buffer pH-6.8 to get standard stock solution containing drug 100µg/ml. Spectrophotometric scanning of Lercanidipine From the stock solution, the ultraviolet scan was taken between the wavelength 200-400nm. Which gave the highest peak at 240nm and the same was selected for Lercanidipine estimation. Preparation of standard plot of Lercanidipine From the standard stock solution series of dilution were made to 5, 10, 15, 20, 25, 30 µg/ml solution using phosphate buffer pH-6.8 and corresponding absorbance was measured at 240 nm in a U.V spectrophotometer. Results are depicted in Figure 1. Formulation development As the drug is highly bitter irst attempt was made to mask the bitterness of the drug by using ion exchange resin such as Tulsion 339. Several trails were carried out with different ratios such as 1:1, 1:2, and 1:3, respectively. Preparation of drug resinate complex Lercanidipine was complexed with ion exchange resin using polacrillin potassium (Tulsion 339) to masks the taste with the following procedure. Step-II Drug resin complexation was prepared by a simple aqueous binding process. The ion-exchange resin particles were uniformly dispersed in a drug ethanolic solution with a mass ratio under magnetic stirring to achieve an equilibrium state. Step-III The complexes were subjected to iltration and cured with deionized water to decant the unbound drug and other ions. The complexes further dried in a hot air oven for 4 h at 40 • C to get powdered mass and stored in a tight glass vial. Step-IV From the above complexes, the best complex is selected based on Drug loading ef iciency. Characterization of the complex for drug content From the prepared Drug resonated-complex, equivalent to 8mg of drug was stirred through magnetic stirrer until the entire drug was leached out from the complex using 100ml of 6.8-phosphate buffer for 60min. The inal solution was iltered through Whatman's ilter paper after serial dilutions using pH-6.8 phosphate buffer and the drug content was assayed spectrophotometrically at 240 nm. From the observations Drug & Tulsion 339 complex ratio, as shown in Table 1, 1:2 used for study due to the high percentage of drug content in the complex. Interpretation of drug-resinated complex palatability Palatability was determined by time intensity method. Here ive human volunteers were selected and suf icient quantity of sample was placed in the mouth for 10 sec. to determine any bitter levels from the given resin complex based on 0-3 scale. A higher value is the sign of strong bitter taste. Angle of repose A glass funnel was selected with a stem of 15-30 mm and ixed to the funnel stand; Below which a graph paper was placed to determine the lowability of granules. Prepared Granules were assessed to form a heap. Heap circumference was marked and the pile height was measured using two rulers. The height was measured and noted it as (h). The area (πr 2 ) was determined, radius(r) was calculated and substituted in the formula (θ =tan −1 h/r), to obtain the angle of repose. Repeated the experiment twice more and calculate the average angle of repose (Kaur et al., 2020). Bul k density Lubricated blend sieved through #20 was correctly weighed to 25 g and transferred to a graduated cylinder of 100ml. Level the Powder carefully and read the unsettled apparent Volume (V0) without compacting. The apparent bulk density in gm/ml can be depicted as follows, Bulk density = W eight of powder Bulk volume Tapped densit y It is determined using the standard procedures and calculated as follows, Carr's Index It is an essential parameter to determine the powder compressibility and packing characterstics before the compression process. It can be determined as follows, Hausner's Ratio Hausner's Ratio is a number of co-related to a powder's lowability. The Hausner Ratio formula is as shown in the equation below, Preparation of tablet Using the lubricated blend as shown in Table 2, Lercanidipine Orally disintegrating tablets were compressed on 16 stations cadmach rotary compression machine equipped with 9mm biconcave punches and constant hardness is maintained for all the tablets. Five tablets were randomly selected, checked for color, odor and shape and the data was noted Thickness Thickness and diameter was measured for ive tablets from all the batches using vernier calipers. Hardnes s test It was tested using 'Monsanto' Hardness tester. Five tablets were randomly selected and placed between the two plungers using a compressible spring on a stainless steel barrel. The initial reading was noted when the lower plunger was in tablet contact and subjected forcilbly to move the upper plunger until the tablet breaks by appling compressional force. Barell containing Pointer on the guage indicates the force, which is a measure of the hardness of tablet strength. Weight variation test 20 tablets were randomly selected and determine the individual weights and their average Weight. Calculate the percentage deviation of IP standards. Friability test For this test, Roche friabilator was used to assess the friction and shock to overcome chipping and breaking of tablets during compression and handling. It has a plastic chamber spins at 25rpm y lowering the tablets from a distance of six inches for each revolution. Usually, preweighed tablets are placed in the friabilator subjected to 100 revolutions. The tablets are then de-dusted and reweighed (Panigrahi et al., 2010). Compressed tablets weigh less than 1.0% of their initial Weight are acceptable for consideration. In-Vitro Dispersion Time It is the time takenfor the tablet to fully disintegrate into ine particles. Three tablets were randomly chosen from each batch and In-vitro dispersion time was performed using 6.8 phosphate buffer (Liberman et al., 1987). In-Vitro D isintegratio n Test It is one of the most important criteria for the prepared tablets to meet out the needs. It can be performed using disintegration Test apparatus IP. Each tablet is placed into one tube of the glass assembly. The entire assembly is suspended in the beaker containing distilled water and subjected for running until the tablet disintegrates, time was noted. Standard The tablets must disintegrate within 30seconds when subjected to a disintegration test examination. Taste evaluation Taste assessment using time intensity process conducted in 6 volunteers. A tablet was placed in the oral cavity for 10 seconds and the recorded bitterness levels and continued the test for different time intervals and repeated the test for comparison (Yunxia et al., 1996). Water absorption ratio A tissue paper of the desired size is taken, folded twice and kept on the surface of a small Petri dish illed with 6 ml of distilled water. The single tablet was taken and placed on the Petri dish containing paper. Complete wetting time was observed and recorded. The last tablet was then reweighed (Kuchekar et al., 2003). Water absorption ratio= f inal weight − initial weight initial weight × 100 Wetting time Tissue papers cut down to a circular shape of 10cm diameter in size and dipped in 6ml(w/v)of Methylene blue dye solution in a Petri dish. A tablet was carefully placed on the surface. The time at which colour development on the upper tablet surface is observed (Yunxia et al., 1999). Drug content uniformity test From each formulation, 5 tablets were randomly selected weighed and powdered. Equivalent quantity of 100mg drug was transferred to 100 ml volumetric lasks. The powder substance is solubilised with small pH-6.8 phosphate buffer volume and subjected for sonication for half an hour. Later solution was iltered and the desired Volume was made using pH-6.8 phosphate buffer. The inal concentration was diluted to 10µg/ml and absorbance was observed at 240nm (Seager, 1998). Invitro drug release Drug release studies carried out using USP type-(paddle) apparatus at a speed of 50 rpm for a speci ied time period of 30mins at temperature (37±0.5 0 c) (Mallet, 1996). The samples were iltered and the amount of drug measured at 240 nm using UV Visible Spectrophotometer (Pahwa and Gupta, 2011). All the post compression parameters were tabulated in table Tables 4, 5 and 6 respectively. Kinetic study The dissolution data was subjected to various kinetic models such as zero order, irst order, Higuchi, Korsmeyer Peppas etc. to know the drug release kinetics of the optimized formulation (Figures 3 and 4). From the above observations, Kinetic analysis (r 2 ) for the optimized formulation of Lercanidipine shows First order kinetics (R 2 =0.94) indicated that the drug release depends upon its concentration (Tables 7, 8 and 9). CONCLUSIONS In the present research investigation, an attempt was made to explore by use of cation exchange resins i,e: Tulsion-339 as a taste masking agent in the formulation of oral disintegrating tablets of Lercanidipine Hcl. Drug resin complex was prepared in the ratio of 1:1,1:2, 1:3 among them maximum drug content was observed for1:2 ratio i,e.. 96.88% which was further inalized for formulations using super disintegrants sodium starch glycolate, cross povidone and L-HPC. From the above results, F9 with L-HPC showed maximum drug release 98.89% in 30 min and showed less disintegration of 38.46 seconds. There is no signi icant change in stability studies. Lercandipine tastes masked ODT are successfully prepared using minimum excipients and a simple method of manufacture.	2021-08-20T18:53:58.601Z	2021-04-07T00:00:00.000	{ "year": 2021, "sha1": "37330db24eb18b63a37566e0c771bd8f2dc7a9c5", "oa_license": "CCBY", "oa_url": "https://doi.org/10.26452/ijrps.v12i2.4633", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "10186a6d30965f5eea3fbca4914ddc62d8ebab93", "s2fieldsofstudy": [ "Chemistry" ], "extfieldsofstudy": [] }
10572586	pes2o/s2orc	v3-fos-license	Super-Resolution Dynamic Imaging of Dendritic Spines Using a Low-Affinity Photoconvertible Actin Probe The actin cytoskeleton of dendritic spines plays a key role in morphological aspects of synaptic plasticity. The detailed analysis of the spine structure and dynamics in live neurons, however, has been hampered by the diffraction-limited resolution of conventional fluorescence microscopy. The advent of nanoscopic imaging techniques thus holds great promise for the study of these processes. We implemented a strategy for the visualization of morphological changes of dendritic spines over tens of minutes at a lateral resolution of 25 to 65 nm. We have generated a low-affinity photoconvertible probe, capable of reversibly binding to actin and thus allowing long-term photoactivated localization microscopy of the spine cytoskeleton. Using this approach, we resolve structural parameters of spines and record their long-term dynamics at a temporal resolution below one minute. Furthermore, we have determined changes in the spine morphology in response to pharmacologically induced synaptic activity and quantified the actin redistribution underlying these changes. By combining PALM imaging with quantum dot tracking, we could also simultaneously visualize the cytoskeleton and the spine membrane, allowing us to record complementary information on the morphological changes of the spines at super-resolution. Introduction Fluorescence microscopy using genetically encoded fluorescent proteins has greatly advanced our understanding of many functional biological systems over the last decade. However, the precision at which cellular structures can be visualized has been limited by the spatial resolution imposed by the diffraction limit of light (,250 nm). Novel super-resolution imaging methods using photoactivatable proteins or photoswitchable fluorophores bypass this limitation and have the potential to revolutionize the experimental scope of light microscopy [1]. The application of these methods in living cells is far from trivial, although recent work has achieved a remarkable progress in this direction [2][3][4][5][6][7]. A general problem is the time needed to acquire a super-resolution image and the associated bleaching of fluorophores, which limits the temporal resolution as well as the use of these techniques for long-term imaging. Dendritic spines are small cellular structures that compartmentalize the sites of excitatory neurotransmission in neurons [8]. The small dimensions of spines (,500 nm in diameter) make superresolution methods ideally suited to image the morphology of spines at much greater detail than that achieved by conventional fluorescence microscopy. The structure of dendritic spines is defined by the F-actin cytoskeleton and can undergo fast dynamic morphological changes that are believed to contribute to the plasticity of synaptic transmission [9]. While physiological and morphological aspects of synaptic plasticity are under certain conditions independent from one another [10], the enhancement of synaptic transmission by long-term potentiation (LTP) appears to be generally associated with an increase in spine volume [4,11]. In line with these observations, the polymerization state and the dynamic properties of the actin cytoskeleton in dendritic spines have been shown to change during synaptic plasticity [12,13]. These studies aimed at identifying different actin pools by distinguishing between populations of actin molecules using photoactivatable fluorophores and FRET, respectively. Morphological changes associated with synaptic activity occur on relatively slow time scale of tens of minutes [4], however the tools to image these changes continuously at high spatial resolution are limited. Therefore the goal of our study was to develop a strategy that would enable us to visualize the dynamic changes of the spine morphology for long periods. Recent studies have combined the use of photoactivatable probes and single particle tracking (SPT) to study the kinetics of the actin cytoskeleton of spines [14,15]. The short-range motion of individual actin molecules in these experiments suggested a complex meshwork of actin filaments in dendritic spines, in line with observations made by electron microscopy (EM) [16,17]. However, SPT-based experiments lack the strength of a direct visualization of the changes of the spine morphology. In this project, we have therefore chosen a very different approach. Rather than inferring the structure from the tracking of single molecules, we aimed at reconstructing the structure itself, meaning that we tried to visualize the sub-spine actin organization. To this aim, we have chosen to use photoactivated localization microscopy (PALM) [18,19]. The principle of PALM is to sequentially activate, image and bleach a sparse subset of fluorescent molecules. By accurately localizing individual activated molecules in each frame, images can be obtained with sub-diffraction resolution. We have generated a probe that binds reversibly to actin, thus acting as a 'scanner' for the distribution of actin within spines. The low-affinity of this probe allows for the replenishment of the bleached fluorophores and thereby overcomes the limitation of long-term PALM imaging of fluorescently tagged fusion proteins. Here, we have used this actin probe to determine the morphology of individual spines and record their baseline dynamics and their evolution during pharmacologically induced synaptic activity. PALM imaging of the actin cytoskeleton using a low affinity photoconvertible probe To study the organization of the actin cytoskeleton we designed an actin probe that combines an actin-binding peptide (ABP) sequence [20] and tandem Eos fluorescent protein (tdEosFP). This construct is designed to bind reversibly to the F-actin cytoskeleton and can be used for PALM imaging due to its photoconversion from a green to a red fluorescent form upon illumination with 405 nm light. When expressed in COS-7 cells, ABP-tdEosFP labels filamentous actin structures with high specificity, as judged by the colocalization with phalloidin labeled F-actin (Fig. 1A). The close correspondence between the two labels also suggests that ABP-tdEosFP has no preference for different pools of F-actin. Sparse subsets of individual ABP-tdEosFP molecules were photoconverted, imaged and bleached under continuous illumination with the 405 nm and 561 nm lasers. The position of each individual fluorophore was determined by fitting the point spread function (PSF) of their fluorescent signal to a two-dimensional Gaussian distribution. By superimposing the positions of all the detected fluorophores a super-resolution image was obtained, in which the actin cytoskeleton can be seen in much greater detail (Fig. 1B). This revealed that some of the observed thick fibers are clearly composed of several thinner structures, while others represent densely packed bundles of actin filaments. The spatial resolution of PALM imaging depends on the ability to accurately determine the position of each single molecule, and ultimately on the number of detected photons emitted by the fluorophore, for EosFP typically in the range of hundreds of photons [21]. In order to experimentally determine the resolution of our system, we analyzed the spatial distribution of the positions from isolated ABP-tdEosFP fluorophores, recorded in consecutive frames in a fixed sample (Fig. 1C). The standard deviation s of the X and Y positions of isolated fluorophores ranges from 10 to 15 nm. The spatial resolution of an image that is reconstructed from individual molecules can be regarded as the full width at half maximum (FWHM) of the Gaussian curve representing the localization uncertainty s. Therefore, in our case, the spatial resolution lies between 25 and 35 nm. Indeed, cross section profiles across filamentous actin structures reveal that these have a thickness of at least 50 nm, well above the limit of our spatial resolution (Fig. 1B,D). The structures are therefore likely to represent bundles of several actin filaments, since individual filaments are known to measure only 8-10 nm in diameter [17]. To confirm the reversibility of the binding of ABP-tdEosFP to F-actin we also performed photoactivation experiments in live COS-7 cells (Fig. 1E,F). A small area of the cell was activated with a brief 405 nm pulse, converting the local population of ABP-tdEosFP into the red form of the fluorophore. By time-lapse imaging we then followed the exchange of this pool of fluorophores with the unconverted bulk of the ABP-tdEosFP pool. We observed that the ABP-tdEosFP signal quickly diffused with a time constant of about 40 s. Taken together, these experiments show that 1) ABP-tdEosFP has a high specificity for F-actin, 2) that its low affinity allows ABP-tdEosFP to exchange between different F-actin binding sites and 3) that PALM imaging of ABP-tdEosFP yields super-resolution images of the actin cytoskeleton down to a spatial resolution of ,25 nm. Super-resolution imaging of dendritic spines in hippocampal neurons We then transfected rat hippocampal neurons with ABP-tdEosFP and fixed the cultures at 3 to 4 weeks in vitro. By this time, the neurons had formed dense networks and exhibited mature dendritic spines. As expected, ABP-tdEosFP was accumulated in spines as a result of the high local concentration of actin (Fig 2A). Long-term expression of ABP-tdEosFP did not affect the spine density (average of 13.1 spines per 10 mm dendrite, N = 8 fields of view from 3 independent experiments), which was found to be comparable to the synapse density of untransfected neurons of similar age [22]. Using PALM image reconstruction we obtained high-resolution images of mature dendritic spines (Fig. 2B). Dendritic protrusions displayed a great variety of shapes and sizes, and included thin, stubby, cup-or mushroom-shaped spines as well as filopodia (see Fig. S1). Given the high spatial resolution of PALM microscopy we could measure the dimensions of spines in great detail as revealed by intensity profiles across different dendritic structures (Fig. 2C). In this context it should be mentioned that our images represent a 2D projection of an optical slice with a depth of about 500 nm. To accurately measure the described parameters, we have therefore focused on spines protruding from the dendrite parallel to the focal plane. We measured the length and the width of the neck of thin, cup-and mushroom-shaped spines, as well as the width of the spine heads (Fig. 2D, E). The quantification of these data shows that the width of the spine neck is relatively homogeneous with about 90640 nm (mean 6 standard deviation, N = 48 spines from 3 independent experiments) in diameter. In contrast, the length of the spine neck was much more variable, ranging from about 250 to 800 nm (N = 48). The average width of the spine head was 6706250 nm (N = 48). These values are similar to those obtained by EM [23,24]. We did not observe any correlation between the dimensions of the spine head and the length and thickness of the spine neck (Fig. S1), suggesting that these parameters are regulated independently [23]). The fact that our photometric measurements are in line with EM findings implies that long-term expression of ABP-tdEosFP does not affect the organization of the spine cytoskeleton. One of the limitations of EM, however, is that it cannot be applied to living cells, which is why PALM imaging has the great advantage that it can give access to dynamic processes at super-resolution, such as the temporal fluctuations of the spine morphology (see below). Large spine heads were either cup-shaped or mushroomshaped, in which case they frequently enclosed a region in which no ABP-tdEosFP signals were detected (Fig. 2C panel (c) and Fig. 2D). We hypothesized that this 'hole' represents a depression in the surface of the spine head viewed from above and possibly formed by a presynaptic bouton. The quantification of the cupshaped structures and the holes gave a mean diameter of about 3806160 nm (N = 21; Fig. 2E), similar to the dimensions of the postsynaptic density (PSD) of 200-500 nm [25]. In order to characterize these actin-free regions, we labeled the PSD with a specific antibody against the scaffold protein Shank2 and a secondary antibody coupled with Alexa647. This fluorophore can exchange between off and on states under reducing conditions, thus lending itself to direct stochastic optical reconstruction microscopy (dSTORM) imaging [26]. Dual-color super-resolution imaging of mature dendritic spines revealed that Shank2 is located at the tip of the spines, overlaps only partially with the ABP-tdEosFP signal, and was occasionally associated with cup shapes as well as with actin-free areas of the spine head (Fig. 2F). Closer inspection of the high-resolution PALM images shows that actin is not homogeneously distributed within the spine (arrowheads in Fig. 2D). Rather, we noticed a filamentous network that appeared to reticulate the spine head. This was most noticeable in the regions of the spine with a relatively low actin concentration. While the thickness of the observed structures is below the limit of our resolution, their length reached up to 100 nm. Optimization of long-term dynamic imaging Given the role of the actin cytoskeleton in morphological aspects of synaptic plasticity [9], our next aim was to measure temporal changes of the structural parameters described above. We therefore adapted our strategy for super-resolution long-term dynamic recording of the dendritic spine morphology. In order to determine the best conditions for live PALM imaging, we adjusted the image acquisition time to optimize the signal-to-noise-ratio (SNR) of the individual fluorophores. To ensure the fastest possible sampling of the structures, we chose the maximal excitation laser power (561 nm at 4 kW/cm 2 ) to bleach the activated fluorophores effectively. We then recorded individual fluorophores at different frame acquisition rates and measured their SNR (Fig. 3A). We found that the mean SNR was highest at 25 ms and 50 ms of acquisition (8.3 and 7.9, respectively). We therefore performed our live experiments consistently at 25 ms exposure time. For the reconstruction of live super-resolution images the spatial and the temporal resolution are intimately correlated. The number of frames for the reconstruction of the PALM image (that determines the temporal resolution) must be adjusted such that the Nyquist-Shannon criterion is satisfied. This criterion states that the sampling frequency must be at least twice as fine as the desired spatial resolution [27]. Taking this into account, we measured the number of detected fluorophores per unit of time in a given area of our structures. We found that in different cellular compartments the typical density of detections within 2000 frames of 25 ms ranged from r = 2610 23 to 8610 23 nm 22 , which results in a theoretical maximal spatial resolution of 44 to 90 nm (or 62 to 126 nm for 1000 frames) considering a constant density probability [28]. To test this concept, we reconstructed PALM images with 500, 1000, 2000, and 5000 frames of 25 ms from a fixed dendritic segment (Fig. 3B). Judging from these images, we found that 1000 to 2000 frames are sufficient to reach a density of reconstruction that satisfies the Nyquist-Shannon criterion. We then imaged live hippocampal neurons expressing ABP-tdEosFP for periods of up to 30 min of constant illumination. Since cellular structures move during live experiments, the time required for the reconstruction of one PALM image must not exceed the time scale of the morphological changes. Hence, there is a compromise between the spatial and the temporal resolution of the recording. In order to determine the best conditions for PALM image reconstruction, i.e. the highest possible temporal and spatial resolution, we chose cultured neurons at day in vitro (DIV) 9. At this developmental stage, the dendritic network is still very immature and displays thin filopodia-like protrusions that undergo relatively fast movements (Fig. S2). In Figure 3C we measured the standard deviation of the recorded signals along a profile across a filopodium (from Fig. S2). As can be seen, there is an optimum reconstruction time of ,50 s (2000 frames of 25 ms), with a standard deviation of 27 nm (spatial resolution of ,65 nm). Choosing a larger number of frames for reconstruction reduces the spatial resolution of the image due to the cell dynamics, whilst a smaller number of frames has the same effect due to the low density of detected fluorophores. It is noteworthy that this measurement has been performed on a thin structure (#65 nm) with a low density of single molecule events per image frame. In practice, this means that the chosen number of frames for PALM reconstruction not only depends on the biological system and photostability of the fluorophore, but also on the region of interest. In our case, 50 s (2000 frames) are necessary to image thin structures such as filopodia or spine necks, whereas 25 s (1000 frames) may be sufficient for the spine head. During acquisition, ABP-tdEosFP molecules are continuously photoconverted, imaged and finally bleached, meaning that the pool of unconverted fluorophores is slowly depleted. This effect causes a reduction of the number of events that are detected per image frame at constant illumination (Fig. 3D). However, the reduction is relatively slow and can be compensated by an increase in the 405 nm laser intensity used for photoconversion of ABP-tdEosFP (see arrow in Fig. 3D), which permitted us to do longterm recordings (up to 30 min of constant illumination). In contrast, a b-actin construct tagged with photoactivatable mCherry [29] decayed much faster during continuous imaging and could not be compensated by an increase of the activation laser. The likely cause of this difference is the reversibility of the ABP-tdEosFP binding to actin that allows the replenishment of the fluorophore (Fig. 1E,F), in combination with the high number of binding sites for ABP-tdEosFP. It is noteworthy that only molecules attached to F-actin are being detected as well-defined PSF signals. Unbound ABP-tdEosFP or fluorophores attached to G-actin diffuse too rapidly to be detected with our 25 ms acquisition time [30], and therefore only contribute to the background noise. In summary, ABP-tdEosFP serves as an ideal tool to achieve a high coverage of the actin cytoskeleton for relatively long periods of recording despite ongoing photobleaching. The reconstruction of PALM images from 2000 frames (50 s at 40 Hz) was generally sufficient to achieve a good sampling while providing an adequate temporal resolution to visualize the movements of dendritic spines or filopodia. Dynamic imaging of mature dendritic spines Having established the best conditions for live PALM imaging, we then recorded the baseline dynamics of mature dendritic spines in hippocampal cultures at DIV 20 to 30 under reduced levels of synaptic activity (MEM-based imaging medium containing 1 mM TTX). These recordings were done at constant illumination for up to 15 minutes at 40 Hz and images were reconstructed from 2000 individual frames (Fig. 4A). For a smoother display, movies of the F-actin dynamics were generated using a temporal sliding window (see Movie S1). We then quantified morphological parameters of the spine actin cytoskeleton in living neurons (Fig. 4B). The parameters that were measured are the same as those quantified in Fig. 2E. In living hippocampal neurons, the length of the neck varied from about 200 to 1000 nm, with a mean diameter of 140645 nm (mean 6 standard deviation, N = 236 spines from 3 independent experiments). The average width of the spine head was 6006180 nm (N = 385), while the diameter of the actin-free regions (holes) and the cup-shaped structures measured 340685 nm (N = 81). Altogether, the spine parameters obtained in the live experiments are in line with the data from fixed neurons, suggesting that the fixation did not noticeably alter the spine cytoskeleton. We also measured the evolution of the morphology of individual spines over time (Fig. 4C). Interestingly, the width of the spine head, the Dual super-resolution imaging of the spine cytoskeleton and plasma membrane In the experiments described thus far we inferred the morphology of the spine from the distribution of the actin cytoskeleton as judged by the localization of the ABP-tdEosFP probe. We therefore sought to relate our measurements of the actin cytoskeleton to the position of the plasma membrane, in order to study how temporal changes of the actin cytoskeleton are translated into a modification of the spine membrane. We co-expressed a GFP-tagged membrane construct (GFP-GPI) together with ABP-tdEosFP in hippocampal neurons. By attaching quantum dots (QDs, 705 nm emission) to the extracellular GFP domain using specific antibodies, we could visualize, in parallel and with a localization accuracy below the diffraction limit of light, the ABP-tdEosFP labeling the actin cytoskeleton and the position of the QDs at the cell membrane. The lateral diffusion of the QDtagged GPI was determined by single particle tracking (SPT) and served as a readout for the plasma membrane (Fig. 5A, Movie S2). We could then measure the distance between the plasma membrane outlined by the QD trajectories and the actin cytoskeleton, as reconstructed by the PALM images (Fig. 5B). In both the neck and the spine head, this was achieved by measuring the diameter across the structure in the two channels and dividing the difference by two. The diffusion of the QDs within the spine determined the sampling rate at which dual images could be reconstructed; typically 2000 frames of 25 ms were enough to obtain a complete outline of the plasma membrane. It should be kept in mind that QD diffusion may be restricted in crowded membrane compartments such as synapses, and that the QDs may thus not always explore the entire membrane surface (see Fig. 5A). Furthermore, the QDs used in this study are relatively large (,25 nm). We have therefore excluded spines with incomplete QD sampling from the analysis. We found that in the spine neck, the distance between the two structures was generally below 50 nm (24628 nm, mean 6 standard deviation, N = 16); in contrast, in the spine head we observed a larger apparent distance between the cytoskeleton and the membrane (46640 nm, N = 19). Nonetheless, in most of the measured spine heads the membrane was within 50 nm of the cytoskeleton. These data show that the super-resolution imaging of the actin cytoskeleton is an accurate readout of the spine morphology. Although both approaches are to a certain extent complementary, our measurements of the actin cytoskeleton provide additional information on the internal organization of synaptic spines. Activity-dependent remodeling of the dendritic spine cytoskeleton The strength of synaptic neurotransmission is regulated by activity-dependent processes, and includes morphological alterations of dendritic spines. These dynamic changes are brought about by rearrangements of the actin cytoskeleton [9]. To investigate the early effects of synaptic activity on the actin distribution within dendritic spines, we performed live PALM imaging during pharmacological activation of AMPA (a-amino-3hydroxy-5-methyl-4-isoxazolepropionic acid) receptors. In these experiments, the glutamate receptor agonist AMPA was added to the bath after recording of the baseline dynamics, and remained present throughout the measurement (up to 10 min). In response to the application of 10 mM AMPA we observed systematic changes in the distribution of actin within the spine head ( Fig. 6A upper panels, Movies S3, S4). Within approximately 2 min after application, the central regions of the spine head were more enriched in ABP-tdEosFP at the expense of the peripheral regions of the spine head. As a result, the size of the spine head was drastically reduced after AMPA application. Concurrently, the spine head lost actin molecules during AMPAR activation, as judged by a reduction in the ABP-tdEosFP fluorophore detections in the spine head over time and the increase of the actin levels in the shaft. However, in the PALM images only the bound pool of ABP-tdEosFP contributes to the image reconstruction, since these fluorophores produce well-defined PSF signals. Nevertheless, the actin redistribution was also apparent in the diffraction-limited images, where the diffuse pool of ABP-tdEosFP contributes to the fluorescence signal ( Fig. 6A lower panels; see Discussion). We quantified the AMPA-induced morphological changes of the spine cytoskeleton to determine the temporal profile and the magnitude of these events. When we measured the spine head area as a function of time, we observed a pronounced reduction of the spine size after AMPA treatment, with an average decay time of 67612 s (N = 5 spines from 3 independent experiments) (Fig. 6B). Similarly, the bath application of AMPA caused a transfer of actin from the spine head into the dendritic shaft (Fig. 6C). Measuring the ratio of detected ABP-tdEosFP fluorophores in the spine head relative to those detected in the shaft, we saw a decay on a time scale of 120670 s. These observations are in line with the depolymerization of F-actin during synaptic activity [31] and suggest that the reduction of the spine size precedes the loss of actin from the spine head into the dendritic shaft. Both effects were not observed in control neurons measured under baseline conditions. We also conducted experiments using a lower AMPA concentration of 2 mM that had been reported to lead to rounding and the loss of motility of dendritic spines [31]. Whereas we observed the described rounding in some of the spines (Fig. 6D, arrow), the general trend was a reduction of the spine head size similar to that seen with 10 mM AMPA application (compare Fig. 6B and Fig. 6E). To probe whether the AMPA induced morphological changes affect the plasma membrane in the same way as the spine cytoskeleton, we performed dual PALM/QD imaging to simultaneously visualize the two structures at superresolution (Fig. 6F). The quantification of these measurements shows that the membrane and the cytoskeleton after the 2 mM AMPA treatment remain closely related (mean distance 48637 nm, N = 13) and not significantly different from the baseline measurements (P = 0.8, Mann-Whitney U test). This demonstrates that the spine membrane adapts dynamically to the changes of the spine morphology. Discussion In this study, we have implemented PALM microscopy for longterm super-resolution dynamic imaging. The necessary tools and techniques are widely available, which makes this a relatively lowcost approach that may be easily set up in many laboratories using fluorescence microscopy. We initially validated our experimental strategy by studying the organization of the actin cytoskeleton in dendritic spines in fixed neurons. From the high-resolution PALM images of mature dendrites it was possible to quantify morphological parameters of spines that are comparable with the results obtained by electron microscopy [23,24]. We did not observe any correlations between the size of the spine head and the dimensions of the neck, in line with previous findings [23]. Upon closer inspection, the sub-spine distribution of actin is not homogeneous. The actin density within the spine head was generally higher in the center compared to the outer regions. These low-density regions appeared to consist of a filamentous mesh. The thickness of these structures is below the pointing accuracy of our detection and could thus not be determined accurately. However, it would be tempting to speculate that these represent actin filaments that have so far only been observed by electron microscopic analysis of the spine cytoskeleton [16,17]. Subsequently, we performed live PALM experiments to observe the baseline dynamics of the spine morphology over tens of minutes. The quantification of the morphological spine parameters in living neurons is in very good agreement with the values obtained in fixed neurons, suggesting that the chemical fixation with paraformaldehyde did not affect the cytoskeletal structure noticeably. Furthermore, we observed that the temporal rearrangement of the spine cytoskeleton was more pronounced in the spine head and the length of the neck, while the width of the neck remained remarkably constant. This could reflect the fact that the actin filaments in the spine head are organized in a complex manner in contrast to the neck where the F-actin appears to have a linear organization [32]. Changes of the spine morphology may thus be expected to occur mainly in the head width and the length of the spine rather than the width of the neck structure. Since the neck represents a bottleneck to the diffusion of synaptic proteins and ions between the head and the shaft, the precise determination of its dimensions in live systems is important to understand its involvement in regulatory mechanisms [8]. We found that the neck opening (width of around 150 nm) occupies as little as 1% of the surface area of the spine head (mean diameter of 600 nm). In this context it should be noted that our measurements refer to the dimension of the cytoskeleton as determined by the detection of the ABP-tdEosFP expression construct. In order to relate our findings to the localization of the plasma membrane, we combined PALM imaging with QD tracking of a diffusing membrane construct (GFP-GPI). This technique allowed us to visualize both, the actin cytoskeleton and the plasma membrane simultaneously at super-resolution. Our measurements showed a close correspondence between the cytoskeleton and the membrane, typically within a range of 50 nm. This distance includes the width of the plasma membrane (5-10 nm) in addition to the size of the antibodies (,5 nm) and the streptavidin-coupled QDs (,25 nm) used for labeling. The measured distances in the spine heads had a broader distribution compared to the neck. The likely cause of this difference is that F-actin filaments are distributed very unequally within the spine head and that consequently low density regions of the cytoskeleton at the periphery of the spine are sampled less efficiently than the core of the spine. To provoke morphological changes of the spine cytoskeleton, we induced a sustained synaptic depolarization by bath application of AMPA and measured the concomitant alterations of the spine shape. We found that this treatment reproducibly led to the shrinkage of synaptic spines, followed by a reduction of the actin levels in spines. This is in agreement with the data by Halpain and colleagues, who found that glutamate receptor activation triggered a loss of F-actin and the collapse of the spines [33]. A different report has suggested that AMPA application causes a rounding of the spine heads and the loss of spine dynamics [31]. This difference may be explained by a concentration-dependence of the glutamatergic activation in the two studies (i.e. 10 mM versus 1-2 mM AMPA). We have therefore reassessed the effect of AMPA application at 2 mM and 10 mM concentration. While we occasionally observed the rounding of the spine head in response to glutamate receptor activation with 2 mM AMPA, the most consistent effect was the reduction of the size of the spines at either concentration, with or without the rounding effect. Our findings are thus suggestive of a depolymerization of the F-actin cytoskeleton during depolarization. In order to explore how the depolymerization of the actin cytoskeleton translates into changes of the position of the plasma membrane, we also performed dualcolor PALM/QD tracking. We found that the correspondence between the two structures was equally close before and after the 2 mM application of AMPA. In other words, the position of the plasma membrane adapts dynamically to the rearrangement of the actin cytoskeleton. In addition to the morphological changes, the levels of actin in the spine head decreased during AMPA application. This redistribution of actin could also be observed in the diffractionlimited images, where the fluorescence intensity of diffusing fluorophores contributes to the overall fluorescence. In contrast, in the PALM image reconstruction only fluorophores that are bound to F-actin are likely to be detected as well focused spots (with 2D-Gaussian shape) and thus represented in the image. Therefore, the decrease in the signal intensity of the spine head relative to the shaft is due not only to depolymerization of actin but also to an actual redistribution of actin through the spine neck. Since the observed redistribution of depolymerized actin from the spine head into the dendritic shaft occurred gradually and on a somewhat slower time scale, our findings also imply that the spine neck represents a barrier that limits the diffusion of actin monomers to the dendrite, and underlines the importance of the spine neck for the compartmentalization of the spine head as the site of synaptic neurotransmission. The use of a low-affinity probe (ABP-tdEosFP) for live PALM imaging confers several beneficial features: 1) bleached fluorophores are replenished, in contrast to probes that are fused to the protein of interest, thereby enabling long-term recording of a given cellular structure; 2) the probe is not directly incorporated into the actin filaments, reducing the risk of altering the cytoskeletal organization; 3) the cell may tolerate higher levels of expression, meaning that there is a larger pool of photoactivatable probes. This also implies an effective increase of the sampling of the structure since a large number of binding sites are available for ABP-tdEosFP. Our experimental strategy also represents a step forward in comparison to the indirect measurements achieved by SPT techniques [14,15], since we reconstruct the entire spine cytoskeleton itself at a relatively high temporal resolution. Our approach can therefore be used to study the long-term morphological rearrangement of actin during synaptic plasticity. Sub-diffraction dynamic imaging (20 s per frame) of dendritic spines in organotypic slices at less than 100 nm resolution was previously shown by Nä gerl et al. using stimulated emission depletion microscopy (STED) [4]. However, in this study the authors made use of a diffusive cytosolic fluorophore (YFP) to visualize the shape of the spine, without resolving the internal organization of the cytoskeletal structure. Furthermore, in comparison to this approach we managed to reduce the high density of photons irradiating the sample by five orders of magnitude (from ,400 MW/cm 2 at 598 nm to ,4 kW/cm 2 at 561 nm) thus reducing the risk of phototoxicity. Indeed, we did not observe any consistent changes of any of the measured spine parameters or their dynamics over time (see Fig. 4C, movie S1). In conclusion, our work brings together several innovative approaches to study the dynamics of cellular structures at superresolution, namely long-term dynamic PALM imaging and the use of a low-affinity probe. Furthermore, we have combined live PALM imaging with QD-based single particle tracking in order to visualize two cellular structures (membrane and cytoskeleton) simultaneously at super-resolution. We believe that this novel approach is a powerful combination of techniques that can be applied to correlate single molecule dynamics with cellular structures at the nanometer scale, such as the movement of neurotransmitter receptors in relation to the postsynaptic density [34]. Materials and Methods Cell culture and transfection COS-7 cells (ATCC, cat. No. CRL-1651, [35]) were kept in DMEM medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine and antibiotics (50 U/ml penicillin/50 mg/ml streptomycin) at 36uC and 5% CO 2 . For transfection, cells were grown on glass coverslips to a density of 2610 4 / cm 2 and transfected with FuGENE 6 (Roche) according to the manufacturer's protocol (2 mg DNA and 6 mg transfection reagent in 2 ml volume per 6 cm Petri dish). Cells were maintained for 24 h in culture medium containing 2% FBS and then fixed or used for live imaging (see FLAP experiments). Rat primary hippocampal neurons were prepared as previously described [36], in accordance with the guidelines issued by the French Ministry of Agriculture and approved by the Direction départamentale des services vétérinaires de Paris (Ecole Normale Supérieure, Animalerie des Rongeurs, license B 75-05-20). All efforts were made to minimize animal suffering and to reduce the number of animals used. Briefly, hippocampi were dissected from Sprague Dawley rats at embryonic day 18, dissociated by trypsinization, plated at a density of 6610 4 /cm 2 on 18 mm glass coverslips that had been successively coated with 80 mg/ml poly-D,L-ornithine (Sigma) and 5% FBS (Invitrogen), and maintained in Neurobasal medium (Invitrogen) containing 2% B27 (Invitrogen) and 2 mM glutamine at 36uC/5% CO 2 . Neurons were generally transfected between day in vitro (DIV) 8 and 12 with Lipofectamine 2000 (Invitrogen) as suggested by the supplier (0.5 mg DNA and 2 ml transfection reagent in 600 ml volume per coverslip for 30 min). Mature hippocampal neurons were generally used for imaging experiments between DIV 20 and DIV 30. Expression constructs The actin-binding sequence of ABP140 from S. cerevisiae (ABP, [20]; MGVADLIKKFESISKEE) was subcloned into the Nhe I/EcoR I restriction sites of the pDendra2-N vector (Clontech) using an overlapping primer pair (ctagccaccATGGGCGTGGCCGACCT-GATCAAGAAGTTCGAGAGCATCAGCAAGGAGGAGactagt-Agaatt). The coding sequence of tandem Eos fluorescent protein (tdEosFP, containing the T158R point mutation, [37]) was amplified by PCR (cDNA kindly provided by Mike Davidson) and shuttled into the ABP-containing construct with Spe I and Not I, yielding the pABP-tdEosFP expression construct. A photoactivatable mCherry-tagged b-actin construct was provided by Vladislav Verkhusha [29]. A GFPtagged GPI expression construct was used to visualize the spine shape using QD-tracking [38]. Fixation and immunocytochemistry Cells were fixed in 0.1 M phosphate buffer (pH 7.4), containing 4% paraformaldehyde and 1% sucrose at 36uC for 10 min and rinsed in phosphate buffered saline (PBS, pH 7.4). The F-actin cytoskeleton in COS cells was labeled using 13 nM Alexa Fluor 647-phalloidin (Invitrogen, 1:500 dilution) in PBS containing 1% bovine serum albumin (BSA, Sigma) for 1 h and washed in PBS. For Shank2 immunolabeling, neurons were post-fixed in methanol for 5 min at 220uC and blocked over night at 4uC in PBS containing 3% BSA. A polyclonal rabbit antibody directed against the PSD protein Shank2 (ProSAP1, [39]) was applied in blocking solution for 1 h at a dilution of 1:500, followed by secondary Alexa Fluor 647-labeled donkey anti rabbit IgG (Invitrogen, 1:500, 1 h). Fixed (labeled or unlabeled) COS-7 cells or neurons were imaged in PBS, after attaching multicolored 0.1 mm beads (TetraSpeck microspheres, Invitrogen) to the coverslips for drift correction (,4610 7 beads in 20 ml PBS for 15 s, rinsed with PBS). Single molecule imaging Super-resolution imaging was performed on an inverted Nikon Ti Eclipse microscope. Activation (405 nm/532 nm) and excitation (561 nm/641nm) lasers for PALM/dSTORM were combined in an external platform; the combined beam was expanded and re-collimated with a beam expander, directed through free air into the microscope and focused in the rear plane of a 1006 objective (N.A. 1.49) using an achromatic converging doublet-lens with a focal length of 500 mm. Images were taken in wide-field configuration and, unless otherwise stated, the experiments were acquired under continuous illumination of both, activation and excitation laser light. Activation laser power was finely tuned during acquisition, in order to maintain the density of activated fluorophores constant; typical values of activation and imaging laser power densities on the sample were 7610 23 to 3610 22 kW/ cm 2 (405 nm), 4 kW/cm 2 (561 nm), 1 kW/cm 2 (532 nm), and 2.7 kW/cm 2 (641 nm). Single-molecule tdEosFP signals were separated with a 561 nm dichroic (Di01-R561-25636) and a single band 617 nm emission filter (FF01-617/73), expanded through a 1.56 lens in the tube-lens of the microscope and acquired with an Andor iXon EMCCD camera (5126512 pixels, pixel size 16 mm) at frame rates of 25 or 50 ms for up to 30 minutes. The low-resolution, conventional fluorescence images of the pre-converted form of tdEosFP were taken using a mercury lamp for illumination (excitation: Semrock FF01-485/20, emission: FF01-525/30). To avoid a substantial degradation of the spatial resolution, sample drift needed to be corrected. The position along the optical axis (Z axis) was actively stabilized during the acquisition by the Nikon perfect focus system (PFS) integrated in the scope. Drift in the XY plane, which may be as high as 50-100 nm in 10 min, was corrected in the post-processing of the images using the multicolor beads described above as fiducial markers. Live dual-color measurements of the actin cytoskeleton and the spine membrane were done by combining PALM imaging of ABP-tdEosFP with QD-tracking of a GFP-GPI membrane expression construct with a Photometrics dual-view detection system, using the 561 nm laser as excitation for both, the converted form of tdEosFP and the QDs. Co-transfected neurons were labeled with a rabbit antibody directed against GFP (Synaptic Systems, 1:10 4 in imaging buffer, 4 min), followed by a secondary biotinylated antirabbit Fab fragment (Jackson ImmunoResearch, 1:2000 in imaging buffer, 4 min) and streptavidin-coupled QDs emitting at 705 nm (Invitrogen Q10161MP, 1 nM in QD binding buffer, 1 min; see [40]). Note that these conjugated quantum dots are relatively large, with approximately 25 nm in diameter. The emitted light was then separated with a 633 nm dichroic and filtered (593/40 nm and 692/40 nm for tdEosFP and 705 nm QDs, respectively). To quantify the fluorescence loss after photoactivation (FLAP), the 405 nm laser beam was focused in the center of the field of view, covering an area of ,4 mm 2 . ABP-tdEosFP transfected COS-7 cells (in MEM imaging buffer at 35uC) were exposed to a short beam of the 405 nm laser, while images were taken every 30 s for 18 min using the mercury lamp (560/25 nm excitation and 614/70 nm emission filters). Image analysis Raw images of single molecule signals were analyzed with an adapted version of the Multi-Trace Tracking MTT algorithm [41]. In practice, we run the algorithm developed by Serge et al. [41], toggling off its tracking features, and using the detection and fitting parts of the algorithm without further modification. For each individual frame, spatially separated point spread function (PSF) signals emitted by single fluorophores were detected and fitted with a 2D Gaussian distribution. Subsequently, superresolution images were rendered by superimposing the position coordinates of the detected single molecules, represented with a 2D Gaussian curve with the same intensity value, using a standard deviation s that had been previously determined by the localization accuracy of single fluorophores (typically 10 nm). In the case of live-cell PALM imaging, the number of frames used for reconstruction was chosen for the best compromise between temporal and spatial resolution, as described in the results section (data shown in Fig. 3C). Figure S1 Sample images of dendritic protrusions, showing the morphological variety of spines described in [42], as seen by PALM imaging (A). In panel (B), neck length and width of dendritic spines show no correlation with the spine head diameter (for quantification see Fig. 2E). (TIF) Figure S2 Pointillist representation of live PALM imaging of a dendritic segment of an immature hippocampal neuron (DIV 9) at time 0 (black points), 12 min (blue) and 25 min (red) under continuous illumination and recording. The 405 nm laser power was continuously adjusted to yield a constant number of single molecule events. The chosen time-window for image reconstruction was 50 s (2000 frames). The data plotted in Fig. 3C were measured in the boxed region. (TIF) Movie S1 Live PALM imaging of a dendritic segment of a mature hippocampal neuron (DIV 27) under baseline conditions. The total length of the movie is 12.5 minutes. Each PALM frame was reconstructed from 2000 image frames of 25 ms, hence the temporal resolution is 50 s. The movie is rendered with a temporal sliding window of a step of 2.5 s. The width of the field of view is 12 mm. The drift of the movie was corrected with fiducial markers, therefore the actual movement of the dendrite is visualized. (AVI) Movie S2 Simultaneous PALM imaging and QD tracking. ABP-tdEosFP (top) and QD signals (bottom) were simultaneously imaged in a hippocampal neuron at DIV 27. The detected ABP-tdEosFP fluorophores were used to reconstruct the actin cytoskeleton of the dendritic spine (cumulative movie, middle panel); the QD trajectory (bottom) served to outline the shape of the spine membrane. Field of view is 5 mm63 mm.	2016-04-30T01:05:23.138Z	2011-01-17T00:00:00.000	{ "year": 2011, "sha1": "4cf03be8c705900aeb60c8221f11580acf681840", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0015611&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "4cf03be8c705900aeb60c8221f11580acf681840", "s2fieldsofstudy": [ "Biology", "Physics" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }
14079657	pes2o/s2orc	v3-fos-license	Brain Activation in Response to Visceral Stimulation in Rats with Amygdala Implants of Corticosterone: An fMRI Study Background Although visceral pain of gastrointestinal (GI) origin is the major complaint in patients with irritable bowel syndrome (IBS) it remains poorly understood. Brain imaging studies suggest a defect in brain-gut communication in IBS with a greater activation of central arousal circuits including the amygdala. Previously, we found that stereotaxic implantation of corticosterone (CORT) onto the amygdala in rats induced anxiety and colonic hypersensitivity. In the present study we used functional magnetic resonance imaging (fMRI) to identify specific brain sites activated in a rat model characterized by anxiety and colonic hypersensitivity. Methodology/Principal Findings Anesthetized male rats received micropellets (30 µg each) of either CORT or cholesterol (CHOL), to serve as a control, implanted stereotaxically on the dorsal margin of each amygdala. Seven days later, rats were anesthetized and placed in the fMRI magnet (7T). A series of isobaric colorectal balloon distensions (CRD - 90s ‘off’, 30s ‘on’, 8 replicates) at two pressures (40 and 60 mmHg) were performed in a standard block-design. Cross correlation statistical analysis was used to determine significant differences between distended and non-distended states in CORT and CHOL-treated animals. Analysis of the imaging data demonstrated greater overall brain activation in response to CRD in rats with CORT implants compared to CHOL controls. Additionally, CORT implants produced significant positive bilateral increases in MRI signal in response to CRD in specific nuclei known as integration sites important in anxiety and pain perception. Conclusions and Significance These data indicate that chronic exposure of the amygdala to elevated levels of CORT enhances overall brain activation in response to CRD, and identified other specific brain regions activated in response to mechanical distension of the colon. These results demonstrate the feasibility of performing fMRI imaging in a rodent model that supports clinical observations in IBS patients with enhanced amygdala activation and symptomatology of abdominal pain and anxiety. Introduction The role of the brain in the processing of information from the viscera, including the gastrointestinal (GI) tract, remains poorly understood. Clinical observations suggest that episodes of stress worsen symptoms associated with irritable bowel syndrome (IBS), a functional GI disorder characterized by episodes of abdominal pain, diarrhea and/or constipation and there is a close overlap between IBS and anxiety disorders [1][2][3][4]. Studies in patients with IBS suggest that heightened pain sensation to intraluminal distension of the colorectum may play a significant role in IBS symptomatology [5][6][7]. Altered perception of visceral stimuli due to afferent sensitization can occur due to abnormalities in ascending visceral signal processing from the gut to the brain [8,9]. In addition, activation of central mechanism(s) resulting in colorectal hypersensitivity due to descending facilitation from the brain induces remodeling of colorectal responsiveness to distension and involves activation of supraspinal nuclei leading to sensitization of spinal dorsal horn neurons [10][11][12][13]. Clinically, activation of descending facilitatory pathways may play a role in the exacerbation of IBS symptoms in response to stress and anxiety [14,15]. The amygdala, as part of the limbic system, is involved in the processing of visceral information, attention, emotion and integrating the physical and psychological components of the stress response. Additionally, the amygdala plays a crucial role in the generation and development of fear and anxiety [16,17]. Specifically, the central nucleus of the amygdala (CeA) has been shown to facilitate the activation of the hypothalamic-pituitary-adrenal axis in response to stress and increase the release of corticotrophin releasing factor, adrenocorticotropic hormone, and corticosterone (CORT) [18,19]. The CeA is also a major source of efferent pathways from the amygdala [20], and electrical stimulation of the CeA has been shown to modulate cardiovascular [21], respiratory [22], and GI function [5]. In recent years to investigate how the brain controls the GI tract, imaging studies using both functional magnetic resonance imaging (fMRI) and positron emission tomography have demonstrated that, in IBS patients, colorectal distension (CRD) activates areas of the brain involved in emotional sensory processing, particularly the amygdala, insula, and prefrontal cortex [23][24][25]. Although fMRI is a powerful tool for noninvasive imaging of brain function studies, neuroimaging in small animals such as rats are limited. However, brain imaging in rodent models offer many advantages such as being able to track the course of a diseases and examine the efficacy of novel treatments with control over relevant factors and experimental conditions that are uncontrollable in human studies. Previous studies to examine brain-gut communication and visceral pain pathways have used techniques ranging from electrophysiology [26][27][28] and c-fos detection [29][30][31] to behavioral analyses [32,33]. One study using fMRI was able to detect brain areas activated by distension of the colorectal region in normal adult anesthetized rats [30]. The aim of our study was to use MRI signal to investigate the effect on supraspinal neuronal processing of manipulation of the amygdala via exposure to elevated levels of CORT shown previously to induce anxiety-like behaviors and increase the sensitivity of the colorectum to luminal distension after 7 days. Blood oxygenation level dependent (BOLD) fMRI in the brain was measured using T 2 * -weighted fast low angle shot (FLASH) imaging at 7T and was performed before, during and after CRD in rats with micropellet implants of CORT or cholesterol (CHOL) as a control. To our knowledge this is the first study to image specific brain regions in response to a defined manipulation of the amygdala in an animal model that mimics aspects of IBS and is characterized by heightened anxiety and abdominal pain. Our results showed that there are significant differences in brain activation in rats in response to modulation of the amygdala via elevated levels of CORT. These results complement recent studies from our group and support clinical observations in which abnormalities in amygdala activation appear common in IBS patients [24,32,34]. Effect of CRD on Total CNS Activation in Rats with Elevated Levels of Amygdala CORT Initially we investigated whether repetitive CRD at 40 mmHg and 60 mmHg causes any disruption of the colonic mucosa. In this part of the study we used the same experimental design as that employed in the fMRI investigation, and examined the histological appearance of the colonic mucosa from rats that underwent CRD and compared the findings to naïve rats. Our results showed that 8 cycles of CRD at 40 mmHg and then 8 cycles at 60 mmHg caused no significant alteration in the architecture of the colonic mucosa compared to naïve controls ( Figure 1) in any colonic region examined and confirmed that there was no change in the histological score between control and treated rats (Score = 0 for both groups in all regions). In the current study we investigated total brain activation in response to 40 and 60 mmHg of CRD in anesthetized rats with bilateral implants of CORT placed stereotaxically on the dorsal margin of the amygdala (Figure 2). At 40 mmHg, CRD induced increases in CNS activation in control rats as illustrated in Figure 3A. The data also showed increases in the total number of activated pixels in CORT-implanted rats in response to CRD at 40 mmHg ( Figure 3B). At a higher distension pressure of 60 mmHg there was an additional increase in the total number of activated pixels in CHOL controls ( Figure 4A). The 60 mmHg CRD pressure also induced more total brain activation in rats with CORT implants ( Figure 4B). A careful comparison of the average number of fMRI activated pixels between the 4 groups (CHOL, 40 mmHg -28.2614.3 pixels or 60 mmHg -173.0675.4 pixels vs. CORT, 40 mmHg -201.8682.3 or 60 mmHg -317.06115.5 pixels, n = 6 per group) indicated a significant effect of group (2-Way ANOVA, F = 4.442, P,0.05) and pressure (2-Way ANOVA, F = 4.548, P,0.05) on brain activation. However, post-hoc tests failed to demonstrate further significant interactions between the groups. In addition, the data demonstrated a similar maximum signal intensity change between the CORT and CHOL groups for both the 40 and 60 mmHg pressures of luminal distension. A typical time curve of fMRI signal intensity of all activated areas is shown in Figure 5. Specific Brain Nuclei Activated in Response to CRD in Rats with Elevated Levels of Amygdala CORT In view of our initial observation of a greater increase in total brain activation in response to CRD in rats with elevated amygdala CORT for 7 days, we attempted to systematically examine specific brain nuclei that showed activation in CORTtreated rats but not in CHOL-implanted rats to gain a more complete picture of the central nuclei activated by CRD in the 2 groups. We focused our analysis to structures identified by fMRI as activated in response to 40 mmHg CRD in CORT-implanted rats but not activated in CHOL controls as these brain sites may be pivotal for the sensitization of visceral responses (Table 1 and 2). Of particular interest, we found that areas involved in the processing of emotion and pain showed activation at 40 mmHg in CORT-implanted animals but no activation in CHOL-implanted controls. In CORT implanted rats, the 40 mmHg visceral stimulus activated specific limbic structures including the hippocampal complex and the hypothalamus. Additionally, brain regions that form descending efferent pathways such as the basal ganglia exhibited activation in MRI signal in CORT but not CHOLtreated rats in response to 40 mmHg. CORT-treated rats also displayed activation in the sensory processing areas such as the somatosensory cortex during the 40 mmHg CRD that were not activated at the same distension pressure in CHOL controls ( Figure 6A-B). As illustrated in figure 6C-D we show the nuclei that were activated at 60 mmHg CRD following amygdala CORT treatment but showed no activation in response the same stimuli in CHOL-treated rats. In response to 60 mmHg animals receiving CORT showed activated pixels in limbic areas such as the cingulate cortex and sensory integration areas including the thalamus. Additionally, brainstem structures involved in sensorymotor integration including the pontine and reticular nuclei also showed activation in response to 60 mmHg CRD in CORTtreated animals but not CHOL controls (Table 1 and 2). Our final analysis examined specific central sites that were not activated in response to CRD at 40 mmHg but showed activation at the higher distension pressure of 60 mmHg in both CHOL and CORT treated rats (Table 1 and 2). This analysis demonstrated that the amygdala, hippocampus and hypothalamus, important regulators of autonomic and neuroendocrine function, showed activation only at 60 mmHg in CHOL-treated rats ( Figure 7A-B). The pontine nuclei and thalamus showed activation at 60 mmHg but not 40 mmHg in CORT-implanted animals ( Figure 7C-D). Interestingly, in CORT-treated rats, 60 mmHg distension also correlated with increased activation of the raphe nuclei which regulate serotonergic activity in the CNS. Discussion Stress and anxiety alter visceral perception and may play a role in central pain amplification and the pathophysiology of IBS [15]. The current study used fMRI to identify areas of brain activation in response to distension of the colorectum. Specifically we investigated brain activation in response to CRD between rats with bilateral amygdala implants of CORT and age and weightmatched CHOL control rats. We provide evidence that elevated CORT on the amygdala significantly increases brain activation in response to CRD compared to that seen in CHOL-treated controls. Taken together the results of the present study support our hypothesis that exposure of the amygdala to elevated levels of CORT leads to colonic hypersensitivity via enhanced activation of specific central circuitry. Specifically, we showed that fMRI detects areas of the brain activated by CRD. In the current study we were able to show CNS activation in the CORT-treated animals in response to distension pressure (40 and 60 mm Hg) was significantly greater than that in CHOL controls. In light of the recent findings by Lazovic et. al. [30], also in anesthetized rats, where 40 mmHg distension pressures did not produce a significant activation of the brain, we opted not to attempt lower distensions pressures. The reason for the depressed threshold in the neuronal activation in response to CRD between the two studies is unknown but is unlikely to be related to the level of anesthesia since in both studies the rats were anesthetized with chloral hydrate. Another possible explanation may be related to the resolution of the images since we used a 7T magnet whereas the earlier study employed a 3T magnet. In the current study, the highest distension pressure that we employed was 60 mmHg, and we did not investigate CNS . Total brain activation at 60 mmHg CRD. A) Activation in all rats with CHOL-containing micropellets. B) Activation in all rats with CORT-containing micropellets. For both A) and B), the total number of pixels per slice is shown in the histogram below the slice. The scale for r (0.4, blue to 0.7, red) is also provided. As illustrated, there was greater activation in rats with CORT micropellets in each slice. doi:10.1371/journal.pone.0008573.g004 activation at higher distension pressures of 80 mmHg since there have been some reports that this level of CRD leads to sensitization of a visceromotor behavioral response, indicative of colorectal hypersensitivity that may be related to a local inflammatory response due to mucosal damage [35]. We found in the current study that repetitive CRD at 40 and 60 mmHg does not induce damage to the colonic mucosa thus eliminating the possibility that peripheral afferent sensitization was responsible for the enhanced brain activation. A significant advance made in the current study was the enhanced brain activation in response to CRD in rats with CORT-containing micropellets placed stereotaxically on the dorsal margin of the amygdala. Thus, when taken together, we believe that this study provides pivotal data supporting the hypothesis that the amygdala is an important nucleus for the sensitization of neuronal sensory circuits that are involved in the induction of colonic hypersensitivity. The findings from the current study demonstrate a positive BOLD signal change in specific brain regions in response to elevated levels of CORT on the amygdala. Despite the size of the rat brain compared the human brain the use of a 7T magnet allowed us to distinguish specific brain nuclei but not sub-nuclei within that structure as seen with techniques such as c-fos and 2-deoxyglucose [36]. Here we show that CRD activated specific nuclei in both the CORT and CHOL controls. These areas include a collection of cortical and subcortical structures identified in human brain imaging studies as the pain matrix. This central network includes the thalamus, amygdala, cingulate cortex, basal ganglia, cerebellum and somatosensory cortex [37][38][39]. The current study demonstrates that, not only does luminal distension activate these structures in rodents, but also that CORT administration leads to a greater increase in the activation of these structures as well as the recruitment of other brain areas that were not activated in CHOL controls. A specific brain area activated in our study with particular relevance to visceral sensitivity is the hypothalamus which is known to be involved in emotional expression and the regulation of behavioral, neuroendocrine, and autonomic functions; however, this structure also contains visceral motor areas and plays an important role in modulating ascending nociceptive transmission. Specifically, the hypothalamus has diffuse connections to brainstem areas regulating visceral pain processing including the periaqueductal gray, raphe nuclei, and locus coeruleus [40]. Another structure important for visceral sensory processing is the cingulate cortex, which is involved in the awareness and perception of internal bodily states and contains specific subareas for integrating visceral sensation [41]. The cingulate also has direct connections to the amygdala, periaqueductal grey, and orbitofrontal cortex [42]. Taken together our observations in rodents [32,34] support clinical observations in which abnormalities in amygdala activation appear common in IBS patients [24]. In the current study negative BOLD signal changes have not been analyzed since the interpretation of deactivation remains rather controversial, however they represent an area for future research investigation in the current animal model. Although real-time fMRI has been reported in conscious rats by the team of Ferris and co-workers [43][44][45] who acclimated rats to the experimental conditions within the fMRI magnet as demonstrated by a reduction in the elevation in plasma CORT and motor movements, many technical and ethical challenges exist in performing experiments in unanesthetized rats using our experimental paradigm. To perform CRD in awake rats, they must be restrained in the fMRI magnet which is problematic in Fischer-344 (F344) rats for two important reasons: firstly there is a strong relationship between stress and the stimulation of visceral hypersensitivity in response to the stress provoking situations, and secondly F344 rats display a vulnerability to repeated stress since they do not habituate to the stressor whereas Sprague Dawley rats show a gradual decrease in the HPA axis following repeated restraint stress [46]. In light of these significant issues, we performed experiments in anesthetized animals, and given the robust nature of the BOLD signal intensity in our anesthetized animals in response to CRD we consider that studies on awake, restrained rats will likely yield data that is confounded by the effects of motion artifact during the distension procedure, stress due to the restraint and the inability to habituate F344 to the stressor. Recent studies claimed to have performed functional brain activation in unrestrained conscious rats in response to CRD [47], however the results from these studies must be treated with caution. Specifically, a limitation in the interpretation of these studies is that although [ 14 C]-iodoantipyrine was injected after the onset of CRD in the conscious rats, the brain images were taken from autoradiographic analysis from the same animals following euthanization, which may have had an effect on the distribution of this marker. In the current study our experimental procedure of performing real-time imaging rather than post-mortem analysis of brain slices offers many significant and obvious advantages, however there are factors that may affect interpretation of our findings including potential changes in cerebral metabolism and blood flow due to the anesthetic that is required to avoid movement artifacts. Moreover, BOLD imaging is not a direct measure of neuronal activity; instead a measure of oxygen utilization, which we assume correlates with metabolic activity. We consider our findings with 40 mmHg to have produced some pivotal data related to the central sites that may be involved in central sensitization despite the absence of data using lower distension pressures. However, in our model it is important to mention that central disinhibition of afferent signaling can also occur which would result in the same 'hypersensitive' response. Furthermore, a limitation of the current study is that although our study provides objective data for the existence of neuronal hypersensitivity, the structure-function relationship of these activated regions was beyond the scope of the current study. There are a number of important aspects of our model that support the validity of our conclusions. The first relates to the specificity of the observed responses to the amygdala. Although the micropellet was placed stereotaxically on the dorsal margin of the amygdala and ensures that the CORT bathes the central amygdala without physical damage to the structure, the possibility exists that other areas adjacent to the amygdala that express steroid receptors such as the hippocampus and caudate putamen, could be involved in our responses due to spread of the CORT from the micropellet. However, in our previous study we systematically performed a series of 'off-site' controls to examine the effect of elevated CORT in areas adjacent to the amygdala and found that the effects were specific to the central amygdala and not due to diffusion to adjacent brain regions [32]. Furthermore, to substantiate our claim that the steroid micropellet remained confined to the area of the implant, others have shown a diffusion distance of 750 mm for a 30 mg micropellet [19] and 900 mm for a 150 mg cannula of CORT [48]. Moreover, although there is a close relationship between the amygdala and the ventrocaudal caudate putamen, previous studies using autoradiography have shown limited binding of CORT [49], due to minimal expression of GR with no appreciable MR in the ventrocaudal caudate putamen [50]. Finally, although CRD can trigger contraction of the anal sphincter, and contraction of this sphincter also induces brain fMRI activity, in the current study external anal sphincter contraction was not measured but the level of fMRI activity was likely identical in both CORT and CHOLtreated rats in response to CRD. In conclusion, the present study provides evidence of increased brain activation, measured as increased MRI signal, in response to CRD in rats, following exposure of the amygdala to elevated levels of CORT for 7 days. Our findings demonstrate that CRD lead to activation of the brain in specific regions known to be involved in somatosensory processing, and that exaggerated brain activation in CORT-implanted animals is suggestive of heightened sensitivity of the brain-gut axis in response to modulation of the amygdala. In light of data from IBS patients demonstrating increased activation in many of the same brain regions that we have observed in the current study, we speculate that our findings in a rodent model may imply that abnormal amygdala activity is a key component to enhanced brain activation involved in the development of anxiety and gut hypersensitivity in IBS. Animals Experiments were performed on male F344 rats weighing 250-310 g and 13-15 wk of age purchased from Charles River Laboratories (Wilmington, MA) and housed under standard conditions with 12-h light/dark cycle with unrestricted access to chow and water. Rats remained in the animal facility for at least 7 days before the experimental procedure, and all experiments were performed at the same time each day. Experiments were performed in a total of 6 rats per experimental group (CORT 40 and 60 mmHg; CHOL 40 and 60 mmHg). A sample size of 6 in each group was determined to have 90% power to detect the expected difference between means (two group Satterthwaite t-test with a 0.05 two-sided significance level) [32]. Stereotaxic Surgery Animals had micropellets (30 mg -diameter of 0.3 mm with a height of 1.0 mm) of either CORT or CHOL placed bilaterally on the dorsal margin of the CeA under ketamine (100 mg/kg intraperitoneal (IP)) and xylazine (10 mg/kg IP) anesthesia as previously described [32,34]. Animals were then allowed seven days for recovery, during which time their behavior was observed to ensure that they were not in distress. Localization of Stereotaxic Implant During the imaging protocol, high-resolution anatomical images (RARE imaging sequence, field of view 2.462.4 cm 2 , matrix size 2566256) were obtained to illustrate the position of the micropellets of CORT or CHOL, as shown in Figure 2B. Following the imaging protocol, animals were euthanized and brains rapidly removed and frozen in chilled 2-methylbutane (Fisher Scientific, Fair Lawn, NJ). Brains were then stored at 280uC until cryosectioning. Serial coronal cross-sections (50 mm) were cryosectioned (Bright OTF, Fairfield, NJ) at 220uC and mounted onto slides followed by verification of micropellet placement by light microscopy. Colorectal Distension A 5-cm latex balloon catheter was constructed as described previously [32]. The balloon was then inserted via the anal canal 8 cm into the colon and secured with surgical tape around the tail following choral hydrate anesthesia. CRD was performed by inflating the balloon using a constant pressure barostat (Model IIR, G&J Electronics, Toronto, Canada) synchronized to the fMRI magnet. A series of 8 phasic CRD was performed at distension pressures of 40 mmHg and then 8 phasic CRD at 60 mmHg for 30 s each ('on') with a 90 s deflation (0 mmHg -'off') between distension periods. Histology To determine whether CRD had any effect on the colonic mucosa, we performed the identical distension protocol in a subgroup of rats (n = 3). Naive rats (n = 3) in which a balloon was neither inserted nor distended served as controls. Three 1 cm tissue segments from each rat were collected from the proximal (1 cm distal to cecum), medial (8 cm proximal to rectum) and distal (2 cm proximal to rectum) colon. Tissues were fixed in 4% buffered formalin, processed, embedded in paraffin and sectioned at 5 mm thickness. They were stained with haematoxylin and eosin and assessed for inflammation by a blinded observer under a light microscope. Inflammatory parameters were scored using the following criteria: 0-2 for ulceration and fibrosis and 0-3 for inflammatory infiltration and depth of lesion [51]. A score of 0 designated no visible signs of pathology in the tissue. Functional Magnetic Resonance Imaging Imaging was performed on a 7T Bruker MRI spectrometer (Biospec, Bruker Biospin, Billerica, MA) with a gradient coil (S116, Bruker Biospin). Prior to imaging, rats were anesthetized with chloral hydrate (400 mg/kg IP). To ensure a similar level of anesthesia across animals and within the same animal during the course of a single experiment an IP Teflon catheter (BD Insyte Autoguard, BD Medical Systems, Sandy, UT) was attached to an injection line for delivering additional anesthetic (40 mg/kg) if indicated by an increase in respiration rate by more than 20 breaths/min from the start of the experiment via monitoring respiration on a monitor (Model 1025, SA Instruments, Inc., Stony Brook, NY) via a transducer placed under the rat. The balloon catheter was then inserted and attached to the barostat with tygon tubing (4/160 ID, 5/160 OD). Following balloon catheter insertion, the rats were placed in a positioning frame and in the imaging coil (receive only quadrature surface coil and multirung volume coil for excitation). The barostat was interfaced to the MRI system console (MRI Interface, G & J Electronics, Inc.) to synchronize distention and deflation cycles with image acquisition. The fMRI paradigm consisted of 8 inflation and deflation cycles consisting of a 90 s baseline period with the balloon deflated ('off'), during which time 9 images were acquired, followed by a 30 s activation or inflation period ('on'), during which 3 images were obtained, for a total of 16 min and 96 images for each pressure (40 or 60 mmHg). The desired pressure of CRD was achieved by rapidly inflating (,5 s) the balloon controlled by the barostat. fMRI Image Selection To minimize signal loss due to susceptibility artifacts at air-tissue interfaces (sinus cavities), the axial plane was chosen. Eight axial (1.2 mm thick) slices were positioned relative to bregma at the following dorsoventral distances: 29.6, 28.4, 27.2, 26.0, 24.8, 23.6, 22.4 and 21.2 mm. Anatomical T 2 -weighted images (effective TE/TR = 50.78/2500 ms, field of view 464 cm 2 , matrix size 1286128, 8 averages) were acquired in 5.3 min using the rapid acquisition with relaxation enhancement imagining sequence. Functional images were obtained using a T 2 * -weighted FLASH (effective TE/TR = 14/160 ms, 64664 matrix, 1 average) 10.24 s for one image, with the same slice position, slice thickness, and field of view as the anatomical images. Data and Statistical Analysis Post processing and analysis of fMRI data were performed using the CCHIPS software [52]. Image co-registration and motion corrections were achieved with a pyramid co-registration algorithm [53]. Areas of activation in different regions acquired during 8 'off' cycles and 8 'on' cycles was cross-correlated with an idealized wave representative of the stimulation paradigm [54]. A cross-correlation coefficient (r) value of the pixels higher than r$0.4 is considered statistically significant (corresponding to P,0.0001), and is shown as different color-coded pixels representing activation on the image. Brain areas that contained activated pixels were identified based on a stereotaxic atlas [55], i.e. images were overlaid with the neuroanatomical image for nuclei identification. Total numbers of activated pixels for brain structures that extended over more than one slice were determined by adding activated pixels per each slice. Total brain activation in CORT and CHOL-treated animals was assessed at each pressure in terms of the total number of activated pixels per animal. Since a Bartlett's test for variance indicated a significant difference in variance between groups, to compare total brain activation between the CORT and CHOL groups in different brain nuclei, the total pixels per nuclei were logarithmically transformed (with no activation treated as 0). The transformed data was then subjected to a two-way analysis of variance (ANOVA) (group and pressure) followed by a Bonferroni post-hoc analysis (GraphPad Prism v. 4.0c for Mac, San Deigo, CA). The non-transformed data for individual nuclei were expressed as mean6standard error of the mean (SEM). Drugs and Chemicals CHOL, CORT and choral hydrate were obtained from Sigma-Aldrich (St. Louis, MO). CHOL and CORT were placed in the brain as micropellets of the dry powder of the chemical. Choral hydrate was dissolved in sterile saline to a concentration of 100 mg/mL (wt/vol). Ketamine (100 mg/mL vial) was obtained from Phoenix Pharmaceutical (St. Joseph, MO) and administered in combination with xylazine (20 mg/mL vial) acquired from Hospira (Lake Forest, IL).	2014-10-01T00:00:00.000Z	2010-01-05T00:00:00.000	{ "year": 2010, "sha1": "ed5bed8ce3ac188b7fa2b6d3e465f35ae7f77eea", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0008573&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "ed5bed8ce3ac188b7fa2b6d3e465f35ae7f77eea", "s2fieldsofstudy": [ "Biology", "Medicine", "Psychology" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
210695900	pes2o/s2orc	v3-fos-license	An Intelligent Scheduling of Non-Critical Patients Admission for Emergency Department The combination of the progressive growth of an aging population, increased life expectancy and a greater number of chronic diseases all contribute significantly to the growing demand for emergency medical care, and thus, causing saturation in Emergency Departments (EDs). This saturation is usually due to the admission of non-urgent patients, who constitute a high percentage of patients in an ED. The Agent-based Model (ABM) is one of the most important tools that helps to study complex systems and explores the emergent behavior of this type of department. Its simulation more accurately reflects the complexity of the operation of real systems. Our proposal is the design of an ABM to schedule the access of these non-critical patients into an ED, which can be useful for the service management dealing with the actual growing demand for emergency care. We suppose that a relocation of these non-critical patients within the expected input pattern, provided initially by historical records, enables a reduction in waiting time for all patients, and therefore, it will lead to an improvement in the quality of service. It would also allow us to avoid long waiting times. This research offers the availability of relevant knowledge for Emergency Department managers in order to help them make decisions to improve the quality of the service, in anticipation of the expected growing demand of the service in the very near future. I. INTRODUCTION Currently, the demand for emergency medical care is increasing and therefore, a hospital's management of Emergency Departments (EDs) becomes more important. In particular, one of the most important issues to deal with is the number of patients who enter the service daily. This is becoming a major problem in EDs worldwide, since it requires the use of a large amount of resources, both human and material, which unfortunately are often very limited. In addition to the above, it should be stressed that the degree of coordination between these resources must be very high [1]. The main consequence of this situation is that the increase in the number of patients entering the service is at saturation point [2]. The result of this is that there has been an increase in the total time a patient spends in the service, from the moment The associate editor coordinating the review of this manuscript and approving it for publication was Rashid Mehmood . the patient enters until they are discharged, which is called Length of Stay (LoS). This is the most accepted parameter in the literature, being used as an indicator of service quality. The aforementioned point of saturation can trigger a general discontentment among patients, including the feeling of being abandoned without care, limited access to emergency care and an increase in patient mortality [3]. One of the most complex areas of the hospital is the ED due to the dynamism and variability of the healthcare as well as the attention times.The operation of the service is the result of the interaction among the elements (agents) of which it is composed. The modeling and simulation of these systems, such as an ED, is one of the most powerful tools for their characterization. The simulation provides a better knowledge of their operation and it can assist decision-making to set up techniques for an optimal system operation [4], [5]. The ultimate objective of modeling and simulating a real system is to know more or find additional information VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/ about it. It is possible to achieve this through inference processes in the system's variables of interest to make predictions about the behavior of these variables under different conditions, based on the information obtained from the new data generated [6], [7]. In previous studies, we developed an ED simulator based on an Agent-Based Modeling (ABM). The ED simulator has been validated within our research group in collaboration with Tauli Hospital, Sabadell. This hospital is one of the most important in Spain, attending over 160,000 patients per year. The model developed describes the behavior of ED services from the interactions between agents, as well as their interaction with the physical environment. It has been developed in NetLogo, a multi-agent programmable modeling environment for complex systems [8], [9]. Based on historical records of non-critical patients from Tauli Hospital, these types of patients do not require urgent attention and they represent a high percentage of the patients. It has been observed that saturation in the ED service is mostly due to the admission of these patients [10]. We propose scheduling policies for the entry pattern of non-critical patients consisting of relocation. From the patients expected arrival time in the input pattern, initially predicted by the historical records, this relocation will provide an improvement in the waiting time affecting all patients, and therefore, an improvement in service quality from the point of view of the patient. The simulator is used as an evaluation tool of the result of the application of the proposed model. In fact, simulation is the only way to generate as much data as we want corresponding to different situations generated by the application of the model, regarding the scheduled input of patients into the system and the healthcare staff configuration. The analysis of this data is the way to show the improvement achieved. It would not be feasible to validate the model for the different possible situations in the real system. This is the main reason why we use simulation to obtain the necessary information for the model validation. As a first contribution of our research, we have developed an analytical model [10], [11] to compute the throughput of the healthcare staff in an ED service, which is defined as the number of patients it can attend to per unit of time. The goal of the analytical model is to use the theoretical throughput as an index to evaluate the responsive capacity of the Staff, taking into account the configuration of the staff (admission and triage staff, number of doctors and nurses), and patient flow throughout the ED. The analytical model is based on a set of equations which takes into account not only the number of personnel, but also their level of experience and the care provided (additional tests or treatments). This model has also been validated by analysis based on records generated by simulation of the real ED. The second contribution of this research is a proper definition of the scheduling model for non-critical patients' admission into the service [12], as well as by their relocation with respect to the initial flow pattern supplied by the hospital's records. The analytical model provides a way to measure the responsiveness of the healthcare staff, which will be used as a reference to carry out patient relocation. We validated the efficacy of the scheduling model based on the information from the actual records from Tauli Hospital and using the simulation to evaluate the results of its implementation. The described research contributions offer the ED managers new knowledge about the behavior of the service, which may be relevant in decision-making, and of great interest with regards to the improvement of service quality, taking into account the expected growth in demand of the service in the very near future. The paper is organized as follows: Section 2 presents the research objectives. In Section 3 we describe the related works. Section 4 gives a description of the main features of the ED and the simulator's capabilities. In Section 5 we describe the analytical model for the calculation of the response capacity in the ED. Section 6 describes the scheduling model for non-critical patients' admission. In Section 7 we present the experimental results. Finally, Section 8 closes the paper with our conclusions. II. RESEARCH OBJECTIVES The overall objective of the research we propose is to develop a methodology that enables us to enhance the best of attention provided in a medical ED, with a view to lessening an affected person's length of stay inside the service, via a model for programming the admission of non-critical patients within the service. We outline this objective in the following specific objectives: • Set suitable indicators of system behavior which allow us to assess the impact of the proposed model. • Typify the system in line with the demand for attention at every moment and for each viable situation. Decide on the responsiveness of the healthcare staff configuration in relation to this particular demand. • Develop a system to permit programming the admission of non-critical patients into the service, consistent with attention capacity, and analyze the effect produced on the quality of the service based on patients' length of stay prediction obtained by using simulation. We will be able to analyze the effect of the model's application, and corroborate its effectiveness, based on the values of the defined indicators or different variables of interest such as patients' length of stay inside the service, calculated from the data generated by ED simulation. III. RELATED WORKS In the related literature there are studies which propose reducing the LoS, and therefore, the time that the patient is waiting to be attended(LoW). Some of the solutions that have been implemented are defined as Fast Tracks [13], [14], or other solutions known as See and Treat [15]. Rachuba et al. [16] show that diagnostic imaging services are essential to the diagnosis pathway for many patients. 9210 VOLUME 8, 2020 Commonly, these patients need to be seen again by a doctor or emergency nurse after an X-ray has been taken to complete the diagnosis and determine the next stage in the patients' pathway. Finally, Rachuba proposes process mapping and assessing the reduction in patients' length of stay in ED. Many studies try to find the elements that impact patients' long periods of stay inside the service and its saturation [17], [18]. Others show that saturation and long waits increase the proportion of patients who leave the service without being attended by a doctor (LWBS) [19], [20]. In our research group, we looked for the best healthcare staff configuration to decrease the patients' LoS in the service, considering constraints related to the cost of the configurations and the amount of available resources [21]. We highlight those references as well as the use of simulation to test the effectiveness of the proposed measures for improvement in the LoS of patients within the service [22]- [25]. Our proposal attempts to move a step forward by obtaining a particular method to reduce the LoS within the ED service. In this way, the improvement in the quality of the ED is achieved by changing the manner in which non-critical patients arrive at the ED and not as much by way of modifying the healthcare staff nor the physical resources. As explained in related papers, simulation gives us a way to measure quality improvement inside the ED by the application of the model. IV. EMERGENCY DEPARTMENT MODEL As seen in Figure 1, the ED operation consists of different stages which every patient goes through from their entry into the ED service until these patients are discharged from the service or admitted into the hospital. For the triage stage, patients are classified according to their acuity level and they are assigned a priority. The scale of urgency implemented in the Spanish Triage service is based on the Andorran Triage model (MAT) [26] as described in Figure 2. A statistical evaluation of these records validates that the great number of patients attending the ED are noncritical patients who do not need immediate attention or they can even be outpatients. The patients' distribution by acuity level from historical records is described in Figure 2. A statistical evaluation of these records validates that the majority of patients attending the ED service are non-critical patients who do not need immediate attention or they can even be outpatients. If it were possible to give information about when it was more advisable to go to the ED service, non-critical patients would probably do it when the likelihood of long waits was lower. The main beneficiaries would be these types of patients, who could take advantage of a scheduling model for admission in the ED service. A. THE FLOW OF PATIENTS INSIDE OF THE ED SIMULATOR We developed the ED simulator including the following agents in Netlogo: admissions and triage staff, doctors for the different hospital areas and assistant nurses for the additional tests and treatments. The agents have their own state that will be modified by the interaction with other agents, these interactions will be reflected in the global operation of the ED service. As shown in Figure 3, when the patient arrives at the service, the simulation runs according to the flow of patients. The admissions and triage stages are the same for all patients entering the ED service. Then, patients with acuity levels 1, 2 and 3 are treated separately from the non-critical patients (levels 4 and 5), which means that they are attended by different doctors and nurses. There is also a low percentage of patients being referred to other hospital areas after the triage and others who leave the service without being seen by a doctor (due to waiting time). In our proposal, we're interested in analyzing the noncritical patients, patients who can be relocated in time for their arrival at the ED service. We will take into account all patients for admissions and triage stages, but only patients with acuity levels 4 and 5 for the diagnosis and treatment stage. In this stage, the patients go through an initial medical exploration (IE) phase. A percentage of them are immediately discharged and leave the ED service after this phase, as we have illustrated by a dashed line in Figure 3. The remaining patients in the ED services go through a section of complementary examinations and/or treatment carried out by technical personnel and/or assistant nurses. After this, they return to visit the same doctor, who analyzes the additional tests or/and treatments. Finally, the patients are discharged from the ED, as shown in the same Figure 3. Each scenario of the ED simulation is identified through a healthcare staff configuration and an input pattern of patients in the ED service (quantity and acuity level of the patients per hour). We add sensors to the ED simulator in order to obtain completely temporalized information about the output for each scenario, taking as a result information regarding the number of attended patients, Patient attention Time (PaT) and waiting time (LoW) for each patient in any stage in the ED service. V. ANALYTICAL MODEL The level of satisfaction with emergency care is mostly conditioned by the perception of waiting time. Furthermore, from the standpoint of service operation. The throughput of the ED is directly related to the number of patients attended and an efficient use of the ED facilities. We are proposing a model for service characterization. This model should provide knowledge and information to make changes in the system in order to improve it. Such a model is based on the definition of a set of service quality indicators as well as a set of equations, allowing us to measure intrinsic elements of the system, such as staff activities and patient flow, as shown in Figure 3. The equations described in this section will enable us to gain information (knowledge) about the capacity of the system with regards to the service resources. The objective is to use this knowledge to design a method for the distribution of non-critical patients, changing their current entry pattern, so that their arrival at the service should be in accordance with the computed capacity of the service. A. QUALITY SERVICE INDEXES We start defining an index named Patient attention Time (PaT) as the total time a patient is receiving attention throughout the stages in the service (admissions, triage, doctors, additional tests or treatments) for a specific member of staff. The PaT is calculated from the sum of the times in attention in each stage, which is obtained from the ED simulator based on the statistical records provided by the hospital (Eq. 1): PaT stage i defines the Patient Attention Time in stage i, and it is independent from the number of patients arriving at the ED service. It's important to know that PaT is not a static value for all patients, as it depends on the path followed by each patient (not all patients need additional tests or treatment). It's important to note that the parameter commonly used in the literature as an indicator of service quality is the Length of Stay (LoS), which is defined as the total time a patient spends in the ED service. Unlike the previous one, the LoS not only depends on the Staff configuration but also on the number and type of patients admitted to the ED service, including the waiting time. Finally we use the Length of Wait (LoW) as an indicator. This is defined as the total waiting time of a patient throughout the service. Therefore: Additionally, the Equivalent Patient attention Time (EPaT) for stage i (EpaT stage i ) is defined as the attention time of a patient, taking into account the possibility of working in parallel in that stage, and Equation 3 shows how it is computed. The SS i and JS i in Equation 3 and 5 stand for the total number of senior/junior staff in stage i respectively, and the calculation corresponds for parallelization on a pipeline model. The slowest stage of the staff configuration will represent the capacity at which patients can be attended in the ED service, as well the stage which can saturate the service. It is, therefore, the inverse of the equivalent attention time of the slowest stage, which will determine the number of patients that a given configuration can treat per unit of time. We define this index as Theoretical Throughput (ThP), which is the indicator for measuring patient attention capacity, which is its response capacity for specific scenarios. The ThP is represented by Equation 4. Indeed, the ThP for a specific stage i will be obtained by the inverse of Equation 3. B. THEORETICAL THROUGHPUT FOR THE DIAGNOSIS AND TREATMENT STAGES Unlike previous stages, the diagnosis and treatment stages are the most complex stages due to their non-linearity. The patients first go through an initial medical exploration (IE), which is their first contact with the doctor. There is a percentage p 1 of patients who need more tests after the IE phase and also a percentage p 2 of patients who need some treatment, which is applied and monitored by the nurses. After the test or treatment phase, the patients return to the doctor for a final diagnosis, after completing any additional examinations (AR). The remaining patients will be discharged from the ED service directly after their first visit with the doctor. Figure 4 details the flow of patients in this phase, according to the abovementioned considerations. The number of assistant nurses, both Senior and Junior, in the staff service is presented as SN/JN respectively. The number of doctors, also Senior or Junior, is represented by SD/JD, and it is important to distinguish between: • SD IE /JD IE : Senior / Junior doctors in the Initial Exploration. • SD AR /JD AR : Senior / Junior doctors in the Analysis of Results. In the ED services, the doctors prioritize the attention of patients after the IE, therefore, these patients will be treated when the doctor is available for the analysis of results (AR). This prevents queues on the return of patients after the additional tests or treatments. As we described before, the ThP has been defined as an index of the response capacity for each stage. In order to calculate the ThP in the diagnosis and treatment stage, it is important to consider the mean attention time of each type of doctor. This time will depend on experience (Junior or Senior), and also on the type of healthcare they are providing, the first step (IE), or the second, consisting of the AR of a requested additional treatment. These times are obtained by the calibration of the ED simulator, and denoted by PaT j i , which represents the mean of the Patient Attention Time for a doctor type i doing j. Then we consider: • PaT IE SD : Mean attention time of a Senior Doctor (SD) in the IE phase. From the historical records provided by the hospital, we see that the patients can go once or more times for the tests or/and treatments, and so they visit the doctor more than once, as shown in Figure 5. The hospital statistics for non-critical patients show that the percentage of patients who need more than one test or/and treatment is very low. There are two percentages for the complementary examinations. The first percentage p 1 of patients who, after their first visit with the doctor, require tests and the second percentage p 2 of patients that require treatment. Finally, there is a percentage 1−(p 1 +p 2 ) of patients who are discharged from the ED service directly after their IE. From the data presented in Figure 5, we can see that around 70% of non-critical patients leave the ED service directly after the IE. Therefore, 30% of patients require additional tests or treatment (p 1 + p 2 ). Thus, given these percentages and the flow of patients in Figure 4, we obtain the following relations of continuity: SD IE + SD AR = SD (8) The result of this linear system of equations gives us the values for SD IE , SD AR , JD IE , JD AR , and therefore, the values for P IE SD , P AR SD , P IE JD , P AR JD , for the doctor staff considered. Once the diagnosis and treatment phases have been characterized, we can obtain the ThP for the doctors' stage in the test and treatment phases by the sum of patients who have only been visited once by the doctor (P onlyIE ), those who have been required for complementary testing (P Test ), and the patients needing additional treatments (P Treat ) (Eq. 10). ThP Doctors = P onlyIE + P Test + P Treat (10) where: We introduce equations (11) to (13) on (10) and we find: so, In addition, the ThP for the nurses in the treatment stage, within the diagnosis and treatment phase, will be calculated using Equation 16: Finally, the ThP for the diagnosis and treatment will be the lowest value of the Equations 15 and 16, and the values of the ThP will be the indicators for the capacity of the ED service for patient attention, supposing that the admission and triage stages do not limit the ThP. According to the actual patient flow in the ED service, we validated [10] the analytical model to calculate the values of the ThP for admissions, triage, nursing and medical exploration stages. For the validation of the model, we used the ED simulator based on an agent-based model of the system as a sensor of the real system. The output information from ED simulation using different possible scenarios has been analyzed in order to get the information for the model validation. The validation of the analytical model consisted of using the ED simulator to see if the obtained values for the ThP for each stage in the ED service were in accordance with the data generated by the ED simulator. The obtained values for the ThP calculated from the equations are shown in Table 1. Once the parameters for the Staff configuration have been configured, and according to the results in the same Table 1, a constant and homogeneous input of patients was created to ensure that the system was continuously running homogeneously to guarantee that the system has been in a steady state, after a warm-up period. We run the simulation for three different numbers of patients entering the ED service per hour, around the theoretical ThP obtained as reference for the ThP for each stage from the system equations proposed. Afterwards, we carried out an analysis of the effect of the number of patients arriving into the service every hour using as an index the percentage of Staff time spent on attending or treating patients (which we defined as Occupancy) for each stage of the service. The obtained results are also shown in Table 1. These simulation results are in accordance with the ThP obtained with the analytical model (Table 1) for all stages of the service. The analytical model for the ThP calculation presented in this section will give us information for relocating noncritical patients, so that the theoretical ThP will be a mandatory indicator for the relocation of non-critical patients. VI. SCHEDULING MODEL FOR NON-CRITICAL PATIENTS The intention of the scheduling model presented in this section is to dynamically adjust the current patient pattern coming into the service to its attention capacity, in order that the flow of patients inside the service shall be in conformity with the healthcare staff available in the service at any time. The admissions scheduling model proposed must enhance quality of care, optimize the quality perception of the attention service in relation to the population, and contribute to the sustainability of the current ED, ensuring better use of available resources. Therefore, the idea aims to improve the ED service, which is the principal entrance of patients in the healthcare services with regards to quality and user satisfaction. The scheduling model for the entry of non-critical patients into the service is built based on the records extracted from the historical records of the hospital. On the other hand, we have defined as theoretical system throughput (ThP) the service characterization in terms of its response capacity to patients' attention. Its value is an indicator of the system capacity to absorb the demand for the service and it is a constraint in our model. It is important to keep in mind that physical space could become a bottleneck in the service, which is why we propose that, based on the throughput of the medical staff, we are able to define the capacity of the waiting rooms. The diagram in Figure 6 gives a global view of the cycle required to dynamically acquire an appointment scheduling for the admission in the service, in keeping with the present hourly demand. The idea is to have a recommendation plan concerning the admission of non-critial patients into the ED service. We based the model on a patient scheduling algorithm. Furthermore, we take into account the attention capacity of ED staff as a limitation, as well as the maximum delay time for relocation of patients as a second restriction. We based the knowledge of the system state hour-by-hour dynamically, which in turn is generated from the data obtained from the historical records and the changes made to it according to the actual demand of the ED, in relation to the access of patients and the kind of care received. A. DISTRIBUTION OF NON-CRITICAL PATIENTS: SYSTEM STATE We define the System State (SS) as the number of patients within the service each hour and we focus on the Patient attention Time (PaT). PaT is given as the staff time (doctor, assistant nurse or technician) spent attending the patient and the time for additional tests and/or treatments. Therefore, the SS does not take into account the Waiting Time for patients (LoW). This information provides us with a more realistic representation of what occurs hour-by-hour in the ED service. From the hospital's historical records, we obtain the distribution of patients arriving in the ED service, which gives us an initial approximation of the number of patients arriving each hour (historical entry patients in Figure 7). The patients that do not need additional tests or/and treatments are discharged by the doctor after the Initial Exploration (IE) and the patients need less than 1 hour to be attended (Direct patients), as shown in Figure 7). Besides that, there's a mean of 4 hours for patients who receive additional treatment, and 2 hours for those who need some tests. The SS is represented hour-by-hour, and for each hour we obtain the number of these types of patients, (test patients and treatment patients) as shown in Figure 7. The distribution probabilities of tests and treatments are provided by the hospital's historical records. The values of the PaT have been calculated from the calibration of the ED simulator using actual historical records from Tauli Hospital. Furthermore, it is important to note that the ED simulator contemplates a random exponential distribution to model the behavior of attention time by the staff, as well as the acuity level and the age of each patient [27]. We carried out a statistical analysis using the output of the simulation, which is given as the result of the mean PaT value for patients. This depends on whether or not they require additional treatments or tests, as we describe in Table 2. In accordance with the values of PaT in Table 2, we have to contemplate the propagation of the patients through the following hours after their arrival in the system because they will be receiving or wating for attention occupying an area in the ED service. Figure 7 shows test and treatment propagated patients for specific input of patients in the Historical Entry Patients row, corresponding to input pattern for a particular day from the statistical records of Tauli Hospital. For each patient we calculate the propagation time taking into consideration the mean PaT (Table 2) and for each hour i Propagated Patients i = TestPatients i−1 + TreatPatients i−1 + Finally, the sum of Entry Patients and Propagated Patients is represented in the System State row, which is the hourly distribution of patients in the system. Therefore, the number of patients inside the system in hour i is the corresponding value for SystemState i = EntryPatients i + PropagatedPatients i . A graphical representation of the SS in Figure 7 is illustrated in Figure 8. Here we consider that: The bars of the graph represent the patients in the system at the arrival time as well as the hours of stay in the service. We divided the patients that arrive and are discharged after the visit with the doctor (PaT: less than one hour) from the patients who arrive and require additional treatments or tests, so we defined the latter as propagated patients according to the corresponding PaT, as shown in Table 2. As we mentioned, the entry hour for non-critical patients is illustrated in the bars, first in the bar representing their arrival and the following bars representing the hours while they are being attended by the doctor or performing the tests or/and treatments. We use the representation hour : patients in Figure 8. It also helps to follow these types of patients in the bar chart. The horizontal non-contiguous lines seen in the Fig.8 indicate three values that represent the attention capacity of the system (ThP), which in turn show the ideal scenario regarding patient care. If patients per hour do not surpass the ThP, then these patients should not have to wait to obtain medical attention. The value of the ThP also indicates whether it is possible or not to enhance the actual scenario by relocating patients, so that the number of patients to attend per hour becomes as homogeneous as possible and does not exceed the ThP limit value. Three possible scenarios may occur: • If the value of the ThP is above the attention necessities, the system is oversized, then there should be no saturation and no modifications are needed. • If the value of ThP is below minimal service necessities, there is no alternative for patients' relocation under ThP and the system cannot avoid saturation without modifying resource availability. However, a possible option for improving the scenario slightly is by trying to flatten VOLUME 8, 2020 the curve, therefore reducing patients' LoW inside the system. • An intermediate case in which there is the option of patient relocation is when we are able to act to enhance system attention. In the last scenario, we use the ThP value as an index for the scheduling model and it will be a constraint in the patient scheduling model, enhancing their current arrival pattern in the system, so that their arrival at the service should result in an SS in accordance with the calculated system capacity. B. NON-CRITICAL PATIENTS SCHEDULING MODEL The relocation of the patients consists of the movement of the patient with respect to their arrival hour in the ED service to minimize the waiting time of the patient and consequently the LoS for all patients in the service. Only non-critical patients can be relocated and we must take into account that ThP is the maximum number of patients that the system can attend per hour without waiting times. The maximum delay time for patient relocation is 6 hours with respect to arrival hour. When the ThP has been calculated, the algorithm begins in hour i = 23 and goes backwards until the initial Hour, which is the first hour in the SS with a number of patients that surpasses the ThP. The algorithm performs the following steps for patient relocation: • It determines holes (free locations for patients to move to). Move backwards hour-by-hour, until identifying the first Critical Hour (the first hour in the day with a number of patients which surpass the value of the ThP) in the calculated System State. We also identify Tentative Patients to be relocated ( Figure 9). • Depending on the relocation range (6 hours) and the restriction of ThP, the algorithm removes the tentative patients from the Critical Hour to the corresponding hour for relocation and it creates holes, removing more patients if possible. Then, the algorithm recalculates patients' propagation for the new scenario and updates the SS, as illustrated in Figure 10. • Once the Initial Hour is reached, the algorithm generates an update of SS in all the hours, and also of Entry Patients, which is the Schedule Entry Patients in Fig.11. In the same Figure the Schedule Patient Limits indicates the number of patients that should be admitted into the ED system per hour. • The Appointment Scheduling Table is created (Figure 12), which describes the appointment hour for the relocated patients depending on their arrival time. The characterization of the SS and the creation of the Appointment Scheduling Table have been developed as a module for ED simulation. Therefore, they are new functionalities of the ED simulator, and thus the patient relocation algorithm runs in a dynamic way. If there are any changes in the entry pattern of patients, initially predicted by the historical records modifying the patient patterns, then the SS is updated again. At the same time, when the SS is updated, the Appointment Scheduling Table is updated. As shown in Figure 12, the Appointment Scheduling Table contains recommendations for patient relocation, as well the Patient Limits Schedule. However, the policy of appointment assignment is necessary and it will be based on a first come, first served basis. In Figure 13, it can be seen how the ED system is able to admit patients per hour, as they arrive, up to the limit of admission in the schedule. The remaining patients will be advised to remain at home until their visiting time, taking it as a new admission into the Scheduling Table. VII. EXPERIMENTAL RESULTS In this section we present the results of the simulation experiments carried out and we summarize the information obtained from these results. average LoW reduction values of the patients arriving each hour into the ED are calculated for each case study considered (called cases A, B and C), once the model for the admission has been carried out, that is to say, through the relocation of the patients, based on the System State calculated in each case. It is important to note that Area B in Tauli Hospital (the waiting room for non-urgent patients) was modeled according to the capacity of the room, in all cases defined by 52 patients (always greater than the ThP of the medical staff). In Tables 3 to 5, the average reduction of LoW in minutes (time) is specified, as well as the corresponding reduction percentage foreseen by the simulation data, with respect to the original LoW without relocation, for all the hours that have been affected by the relocation in which a minimum reduction of 10% has been obtained, which is more than 10 minutes. LoW reduction values are shown separately for patients who are discharged after their first consultation with a doctor and do not require any test or treatment (Direct Patients), patients who need additional tests (Test Patients) and those who receive some type of treatment (Treatment Patients) respectively. The initial relocation hour is marked in ocher in each table. The relocation effect can be extended to the previous hours, in addition to the following hours. This is because relocation can affect the waiting times of the test and treatment patients who arrived in the hours previous to the initial hour, due to its propagation. If this propagation reaches the hours in which other patients have been relocated. Some hours with atypical values in the LoW have been detected for treatment patients. Due to the outliers, the results with a reduction in the LoW have been rejected. Finally, the expected patient input (historical input) and the input programmed by relocation (programmed input) are shown graphically for each case (Figures 14, 15 and 16). On the same graphs, the corresponding values of the average patient LoS (direct, test and treatment patients) in the service at each hour are represented, without relocation and after patient relocation, respectively. A. CASE A This case is characterized by a distribution of patients entering the ED corresponding to that determined by the current data from the historical records of Tauli Hospital. Table 3 contains the details of the reduction in the LoW per hour for patients due to the relocation proposed by the model in this case. The data analyzed to obtain these values are those represented in Figure 14. In this case, the initial hour for relocation, is hour 10 and relocation has an effect on hour 8 to hour 18. The effectiveness of relocation is noticeable during all these hours, reaching the maximum reduction value in hour 12 for all patients. The reduction in the case of direct patients only exceeds the minimum value by 10 minutes in hour 12. Even so, those hours with a reduction of 9 minutes are also shown, with these reductions, in all cases, very close to 50% compared to the original LoW value. For test patients, the relocation of patients has an effect in the LoW value until hour 16 and for treatment patients the effect is extended until hour 18. In both cases, the effect of relocation on the previous hours to the initial hour is also shown. The LoW for these test or treatment patients who arrive in those previous hours could be reduced due to the fact that the subsequent hours have been modified by relocation of other patients, since they may be propagated in the system for two, three or four hours. This effect is observed in Table 3 for hours 8 and 9. Finally, the graphs in Figure 14 show the effect of relocation on the entry of patients and their length of stay (LoS). The relocation of patients and the consequent scheduling of their admission entails a redistribution in the arrival of these patients to the service according to the system's attention capacity. This new situation results in an improvement in the length of stay of the patients in the service, as can be observed in the LoS lines represented on the graphs of the Figure 14. B. CASE B Case B is characterized by an input distribution of patients with a high concentration of patients in the central hours of the day. Table 4 contains the details of the reduction in LoW per hour for patients due to the relocation proposed by the model. The effect of relocation focuses on the 6 hours following the initial relocation hour, and it also affects the previous hours for treatment patients admitted to the system during those hours. As already mentioned, the effect for treatment patients in these previous hours to the initial relocation hour is due to the fact that these patients remain in the service during the hours in which relocation of other patients has been carried out and that reduces their wait. An average reduction of more than 50% is achieved in all hours between hour 9 and hour 12 for the three types of patients. Once again, the positive effect of patient relocation is confirmed in this case, precisely in the central hours of the day for which the entry of patients has been considered greater. The graphs in Figure 15 show the effect of relocation on the entry of patients and on their total time in the service (LoS). In this case, the effect of relocation is more noticeable than in TABLE 5. Case C: Time (minutes) and percentage of average reduction in LoW due to patient relocation. FIGURE 16. Case C: Entry of patients with 17 more patients due to an accident at hour 13 without relocation (historical input) and with relocation (scheduled input). Average LoS per corresponding hour for direct, test and treatment patients without relocation and with patient relocation. the previous case, given the greater saturation of the service in the central hours of the day without relocation. In the case of direct patients, the LoS curve flattens out almost completely. For test patients, LoS with patient relocation is maintained between 150 and 200 minutes from hour 9 (initial relocation hour), and for treatment patients, LoS with patient relocation remains below 300 minutes. C. CASE C In this case, we have experimented with a scenario characterized by an influx of patients with an unexpected increase in patients at a certain time of day, simulating a possible accident. The scenario contemplates an entry of 17 patients in hour 13. The graphs in Figure 16 show the effect of relocation on the entry of patients and the total time of their stay in the service (LoS) for this case. The relocation of patients, before and after the arrival of the patients by the simulated accident, manages to improve the times of all patients in the hours affected by the accident. In return, the hours furthest from the hour of the accident, used for scheduling relocated patients to avoid saturation in critical hours, suffer an increase in waiting times. Even so, the results show a global improvement in patient waiting times. These experimental results, obtained through the analysis of simulation data of the three case studies considered, demonstrate the positive effect of the patient relocation method and, therefore, validate the scheduling model for the admission of non-critical patients into the ED. With its application through the relocation of these patients, the LoW improves in all cases, and consequently the LoS of the patients and, therefore, the quality of care provided in the ED. Table 5 contains the details of the reduction in LoW per hour for patients due to the relocation proposed by the model for this case. VIII. CONCLUSION The main contribution of this research has been the definition of an analytical model for the calculation of the response capacity of a certain healthcare staff configuration in an ED. We have seen how this characterization of the system, through its ability to absorb the demand of the service, indicates the ideal situation of the state of the system in each hour and it is a reference for measuring system performance. We proposed a scheduling model for non-critical patients which provides a method to enhance the quality of the healthcare service and it is a useful tool for the service administration in order to attend the current growing demand for emergency services. The implementation of the model in the ED simulator and the results produced by the analysis of simulation data validate the effectiveness of the model and improve patient waiting times in the system globally and, as a result, improve the quality of healthcare. It is also important to mention that the implementation of the model inside a real ED service will be efficient to the extent that the proposed scheduler system performs on the real entry of patients, which will be dependent on the patients' or health service users' decisions.	2020-01-02T21:47:19.710Z	2020-01-01T00:00:00.000	{ "year": 2020, "sha1": "89ebf71fafd4fed7ed1a5ac9222cee3867acf857", "oa_license": "CCBY", "oa_url": "https://ieeexplore.ieee.org/ielx7/6287639/8948470/08945359.pdf", "oa_status": "GOLD", "pdf_src": "IEEE", "pdf_hash": "877a0eea3dd1377488f6d5eda2b4bbdba2ad618a", "s2fieldsofstudy": [ "Medicine", "Computer Science", "Engineering" ], "extfieldsofstudy": [ "Computer Science" ] }
17569316	pes2o/s2orc	v3-fos-license	Disruption of the Lipid-Transporting LdMT-LdRos3 Complex in Leishmania donovani Affects Membrane Lipid Asymmetry but Not Host Cell Invasion Maintenance and regulation of the asymmetric lipid distribution across eukaryotic plasma membranes is governed by the concerted action of specific membrane proteins controlling lipid movement across the bilayer. Here, we show that the miltefosine transporter (LdMT), a member of the P4-ATPase subfamily in Leishmania donovani, and the Cdc50-like protein LdRos3 form a stable complex that plays an essential role in maintaining phospholipid asymmetry in the parasite plasma membrane. Loss of either LdMT or LdRos3 abolishes ATP-dependent transport of NBD-labelled phosphatidylethanolamine (PE) and phosphatidylcholine from the outer to the inner plasma membrane leaflet and results in an increased cell surface exposure of endogenous PE. We also find that promastigotes of L. donovani lack any detectable amount of phosphatidylserine (PS) but retain their infectivity in THP-1-derived macrophages. Likewise, infectivity was unchanged for parasites without LdMT-LdRos3 complexes. We conclude that exposure of PS and PE to the exoplasmic leaflet is not crucial for the infectivity of L. donovani promastigotes. Introduction Most eukaryotic cells exhibit an asymmetric distribution of phospholipids in their plasma membrane, with aminophospholipids phosphatidylserine (PS) and phosphatidylethanolamine (PE) restricted to the cytosolic leaflet [1]. This asymmetric lipid distribution is maintained by ATP-dependent processes, suggesting that it is critical to normal cell function. Indeed, if cells fail to engage mechanisms to maintain lipid asymmetry, PS and PE appear at the cell surface, which leads to dramatic changes in cell function [2]. In multicellular organisms, one of the important consequences of altered membrane lipid asymmetry is the recognition and engulfment of PS exposing apoptotic cells by mononuclear macrophages. Recently, PS exposure has also been implicated in the infectivity of Leishmania, an obligate, intracellular parasite that infects cells of the mononuclear phagocyte lineage in their vertebrate hosts [3][4][5]. The parasite has a digenic life cycle, residing as flagellated extracellular promastigotes in the gut of the insect vector and as obligatory intracellular aflagellated amastigotes found in the parasitophorous vacuoles of mammalian macrophages. The observation of PS exposure on the cell surface of Leishmania parasites led to the hypothesis that the protozoan parasite might take advantage of a regulated loss of lipid asymmetry and PS presentation to invade and survive in host macrophages [6]. However, it remains open whether (i) PS exposure is mandatory for invasion of host cells and (ii) exposure of other lipids may facilitate infectivity as well. In fact, exposure of PS is typically accompanied by a loss of transbilayer asymmetry of other lipids. In particular, PE redistributes to the exoplasmic leaflet as well. Hence, this lipid may contribute to or even be critical for infectivity of parasites. Maintenance and regulation of membrane lipid asymmetry is governed by the concerted action of ATP-driven translocases that move specific lipids against a concentration gradient across the bilayer. Prime candidates for outward-directed lipid translocases are members of the ATP binding cassette (ABC) transporter family [7], whereas potential inward-directed lipid translocases belong to the P4 subfamily of P-type ATPases [8]. Members of both transporter families have also been implicated in the development of drug resistance in Leishmania, which includes resistance to alkyllysophospholipids [9][10][11][12][13][14]. This suggests that the mechanism by which the drugs are extruded from cells is closely related to the flippase mechanism by which lipids are translocated across membranes, and that these processes involve structurally similar, if not identical transporters. Studies aimed at identifying the molecular basis of miltefosine (hexadecylphosphocholine) resistance led to the identification of two Leishmania membrane proteins, LdMT, a member of the P4 subfamily of P-type ATPases, and LdRos3, a potential noncatalytic subunit of LdMT related to Cdc50 family [11,12,14]. Both proteins are primarily localized to the Leishmania plasma membrane and required for the rapid intracellular uptake of alkylphosphocholine drugs. In the budding yeast Saccharomyces cerevisiae, members of the two protein families have been found to form stable transporter complexes that function in the translocation of phospholipids from the exoplasmic to the cytoplasmic leaflet of cellular membranes [15,16]. Likewise, LdMT and LdRos3 have been implicated in the uptake of fluorescent 7nitrobenz-2-oxa-1,3-diazol-4-yl (NBD)-labelled phospholipids, suggesting a role for these proteins in controlling lipid asymmetry in the parasite plasma membrane [12,17]. However, lipid asymmetry of parasite plasma membranes is not well defined and the proteins involved in its regulation remain to be identified. In this work, we study the regulation of phospholipid asymmetry in promastigotes of Leishmania donovani and the relevance of asymmetry loss for infection. In particular, we show that LdMT and LdRos3 form a stable protein complex that is essential for inward translocation of NBD-PE and -PC across the parasite plasma membrane. Loss of either LdMT or LdRos3 leads to an increased cell surface exposure of endogenous PE. We therefore propose that the LdMT-LdRos3 complex plays an essential role in maintaining phospholipid asymmetry in the parasites plasma membrane. Since, to our surprise, promastigotes of Leishmania donovani lack any detectable amount of PS, we could study specifically the role of PS and PE exposure in host cell invasion. Control cells with functional LdMT-LdRos3 complexes were able to invade host cells showing that PS is not mandatory for infection. Furthermore, infection was not enhanced for parasites without LdMT-LdRos3 complexes indicating that redistribution of PE to the exoplasmic leaflet does not facilitate invasion. Leishmania LdMT and LdRos3 form a stable complex To investigate the ability of LdMT and LdRos3 to form a stable complex as their homologues in yeast, a C-terminally green fluorescent protein (GFP)-tagged version of LdRos3 (LdRos3-GFP) was expressed in parasites lacking either LdRos3 (DLdRos3) or LdMT (DLdMT). Tagged LdRos3-GFP was functional, since it could suppress resistance of DLdRos3 parasites to the alkylphospholipid derivate miltefosine [12]. Total membrane preparations from the LdRos3-GFP expressing parasites were subjected to solubilisation in the presence of mild detergent and analysed by non-denaturing polyacrylamide gel electrophoresis (native-PAGE), in which complex formation between two proteins gives a new band with mobility different from that of either protein alone. Fluorimaging of the native gels identified one prominent fluorescent band of low molecular size for protein preparations obtained from DLdMT parasites expressing LdRos3-GFP (Fig. 1A). In protein samples obtained from DLdRos3 parasites expressing LdRos3-GFP, additional fluorescent bands of retarded mobility were detected (Fig. 1B, bands II-III). To test for the presence of LdMT in the GFP-labelled complexes, the fluorescent bands were excised from the native gel and subjected to SDS-PAGE followed by western blot analysis. Immunoblotting with antibodies to LdMT and LdRos3 showed that the fluorescent band of low molecular size (band I) only contained LdRos3-GFP while the more slowly migrating bands (band II-III) corresponded to a complex of LdMT and Ros3-GFP. To corroborate that LdMT and LdRos3 form a stable complex, a C-terminally GFP-tagged version of LdMT (LdMT-GFP) was introduced into DLdMT parasites, and co-immunoprecipitation experiments were performed using anti-GFP-MicroBeads. Tagged LdMT-GFP has been shown to be functional, since it can rescue the miltefosin uptake defect in DLdMT parasites [12]. Nontransfected wild-type parasites served as a control. As shown in Figure 1C, the anti-GFP antibodies efficiently precipitated LdMT-GFP and LdRos3 from detergent-solubilised membrane preparations obtained from parasites expressing LdMT-GFP. In contrast, an unrelated integral plasma membrane protein (metalloprotease gp63) was not present in the immunoprecipitate. As a control, a parallel immunoprecipitation from an extract obtained from a wild-type strain lacking the LdMT-GFP fusion did not precipitate LdRos3, indicating that the co-immunoprecipitation was specific. Analysis of the LdMT-GFP immunoprecipitates by mass spectrometry revealed that the preparations did not contain other Cdc50 members than LdRos3 (data not shown). Taken together, these results provide direct evidence that LdMT and LdRos3 reside in a stable complex in the membrane. We infer that the stoichiometry of both proteins in the complex is one because we did not detect corresponding fluorescent bands of retarded mobility for larger complexes. LdMT and LdRos3 are required for the ATP-dependent inward translocation of NBD-PC and -PE To characterize the lipid transport activity of the LdMT-LdRos3 complex in the plasma membrane we examined in wild type, DLdMT and DLdRos3 parasites the uptake of NBD-lipids. This can be monitored by flow cytometry and by fluorescence microscopy. Experiments were performed at 2uC to suppress endocytosis and suppress degradation of the NBD-lipids (consistently ,10%). Under these conditions, wild-type parasites efficiently internalized NBD-PC and NBD-PE, while NBD-PS and NBD-sphingomyelin (NBD-SM) were hardly taken up at all. The efficient uptake of NBD-PC and NBD-PE was significantly inhibited following ATP depletion by preincubation with sodium azide and 2-deoxyglucose or collapse of the proton electrochemical gradient with the protonophore carbonyl cyanide mchlorophenylhydrazone (CCCP) ( Fig. 2A). Consistent with these results fluorescence microscopy of wild-type parasites labelled with NBD-lipids revealed an intensive labelling of intracellular membranes with NBD-PC and -PE but not with NBD-PS and -SM (Fig. 2D). In contrast to wild-type parasites, DLdMT and DLdRos3 parasites were found defective in the low temperature uptake of NBD-PC and NBD-PE (Fig. 2B). These defects were due solely to the loss of LdMT and LdRos3, as lipid uptake was restored by re-expression of LdMT-GFP and LdRos3-GFP in DLdMT and DLdRos3 mutants, respectively (Fig. 2C). Based on these findings, we conclude that the LdMT-LdRos3 complex is essential to sustain an energy-dependent influx of NBD-PC and -PE across the parasite plasma membrane. LdMT and LdRos3 sustain plasma membrane PE asymmetry in L. donovani We next tested whether the LdMT-LdRos3 complex helps to maintain a tight asymmetric distribution of aminophospholipids at the plasma membrane. Promastigote stages of Leishmania wild type, DLdMT and DLdRos3 lines were incubated with different concentrations of duramycin and papuamide B, cytolytic peptides that requires binding to cell surface-exposed PE and PS, respectively, to exert their cytotoxicity [18]. As shown in Figure 3A, both DLdMT and DLdRos3 lines were more sensitive to duramycin-induced cytolysis as compared to the wild-type line with an EC50 of duramycin that was approximately 1.8-fold lower than that for wild-type cells. Restoration of LdMT and Ros3 expression returned the duramycin sensitivity profile back to the wild-type pattern ( Figure 3A). There was no significant difference between the DLdMT and the DLdRos3 Leishmania lines as compared to wild-type cells with respect to their sensitivity to papuamide B and amphotericin B, a polyene macrolide antibiotic that binds to membrane ergosterol and induces cellular leakage [19]. To more directly visualize that deletion of LdMT or LdRos3 affects the lipid asymmetry in the plasma membrane, we analyzed the exposure of PE by labelling with biotinylated Ro09-0198, a peptide that specifically binds to PE [20,21]. As shown in Figure 3B, both the DLdMT and the DLdRos3 Leishmania lines bound more PE-sensing biotinylated Ro09-0198 peptide which was visualized with Streptavidin-FITC. For a quantitative assessment of Ro09-0198 peptide binding, the FITC fluorescence associated with mutant and wildtype cells was measured by flow cytometry. As shown in Figure 3A, deletion of LdMT or LdRos3 caused a 10-fold increase in PE exposure when compared to wild type parasites based on Ro09-0198 peptide binding. We could, however, not detect in any of the lines labelling of propidium iodide (PI) negative parasites with annexin V-FITC, a protein that preferentially interacts with membranes containing PS [22]. To rule out the possibility that the observed differences in PE exposure resulted from substantial increase in the PE level in the DLdMT and the DLdRos3 lines as compared to wild-type parasites, we analyzed the total phospholipid composition in Leishmania parasites labelled to steady state with 32 P-phosphate. Four major phospholipids could be identified with comparable levels in all of the strains: PC (44%-49%), PE (27%-29%), phosphatidylinositol (PI, 9%) and inositolphosphorylceramide (IPC, 4-5%) (Fig. 4A). Thus, the possibility that increased cell surface exposure of PE was caused by increased PE synthesis can be ruled out. Notably, we could not detect significant levels of PS by this method. When stained with ninhydrin, a reagent labelling lipids with free amino groups such as PS and PE, the chromatograms exhibited only one spot corresponding to PE (Fig. 4B). Likewise, lipid analysis by HPLC coupled ESI-MS did not reveal the presence of PS, suggesting that L. donovani does, if it The extract obtained from DLdMT parasites also contains slow-migrating fluorescent protein bands (band II and III). Regions of the gel corresponding to the fluorescent bands were excised and loaded onto SDS-PAGE gels which were subsequently immunoblotted with polyclonal antibodies against LdRos3 (a-LdRos3) and LdMT (a-LdMT). Size markers indicate relative mobility of proteins in kDa. (C) Immunoblots from coimmunoprecipitation assays. LdMT-GFP was immunoprecipated from a detergent-solubilised membrane fraction (load) obtained from DLdMT parasites expressing LdMT-GFP as well as non-transfected wild-type parasites (wt) using anti-GFP-MicroBeads. Immunoprecipitates (IP) were subjected to immunoblot analysis using antibodies that recognize LdMT (a-LdMT), LdRos3 (a-LdRos3) and metalloprotease gp63 (a-gp63). doi:10.1371/journal.pone.0012443.g001 all, synthesize very low amounts of this phospholipid (Fig. 4C). The presence of only very low amounts of PS explains the equal sensitivity of the parasite lines to papuamide B and the lack of FITC-annexin V binding on intact parasites (see above). Phagocytosis of DLdMT and DLdRos3 Leishmania lines in THP-1-derived macrophages is unchanged Exposure of PS on the exoplasmic leaflet of the Leishmania plasma membrane has been implicated in entry of the parasite into host cells [3][4][5]. The availability of two Leishmania lines strains displaying an altered plasma membrane lipid asymmetry and apparent lack of PS provided an opportunity to test whether other phospholipids are relevant for the recognition of Leishmania parasites by macrophages using the human monocytic cell line THP1. THP1 cells have many characteristics of human monocytes, including morphology, surface-membrane receptor, and the capacity to undergo maturational changes when induced with phorbol esters [23,24]. We used early log phase promastigotes to evade apoptotic subpopulations which increase during late logarithmic and stationary phase. Parasites and THP-1-derived macrophages were prelabelled with CellTrackerTM Green and CellTrackerTM Dil, respectively, and co-incubated at a ratio of 1:10 (cells:parasitetes). 16 h post infection, 30 to 70% of the THP-1-derived macrophages were found to be infected. At this moment the number of intracellular parasites per infected macrophage ranged from 1 to 19 (Fig. 5A). The mean numbers of intracellular parasites per infected cell were similar: 2.6 for wt, 2.5 for DLdMT and 2.7 for DLdRos3 parasites, respectively. The total numbers of THP-1-derived macrophages infected by DLdMT and DLdRos3 parasites were not markedly different from that of wild-type parasites (Fig. 5B), indicating that at least in this infection model, changes in the plasma membrane PE arrangement of Leishmania parasites do not trigger specific recognition and removal of the parasites by macrophages. Discussion Previous studies revealed a critical role for both the miltefosine transporter LdMT and the LdRos3 protein in the uptake and potency of alkylphosphocholine type drugs and in phospholipid translocation at the plasma membrane of Leishmania parasites [11,12,14]. The present study provides direct evidence that both proteins form a stable complex that is essential for maintaining the normal asymmetric lipid distribution of at the parasite plasma membrane. Loss of plasma membrane PE asymmetry by disruption of the LdMT-LdRos3 complex does not alter the infectivity in macrophages. We conclude that in L. donovani loss of the normal asymmetric lipid distribution does not facilitate host cell invasion. Coming challenges are to identify whether all of these P4-ATPase require a Cdc50 binding partner for proper functioning and whether they can interact with several Cdc50 proteins as shown for the Arabidopsis P4-ATPase ALA3 [26]. In light of the results presented in this study, it seems that Leishmania LdMT does not associate with other Cdc50 members than LdRos3 and that the stoichiometry of both proteins in the complex is one. Incubation of parasites with fluorescent phospholipid analogues (PC, PE, PS and SM) served as an approach to analyze lipid transport activity of the LdMT-LdRos3 complex at the plasma membrane. We found that L. donovani parasites efficiently internalize NBD-PC and NBD-PE even at 2uC while NBD-PS and NBD-SM are hardly taken at this temperature. Since endocytosis is blocked under this condition [27], uptake of NBD-PC and NBD-PE must involve a translocation step across the plasma membrane. Loss of either LdMT or LdRos3 completely blocked the ATP-dependent uptake of NBD-PC and NBD-PE, indicating that the LdMT-LdRos3 complex facilitates the energy-coupled transport of natural PE and PC from the outer to the inner leaflet of the parasites plasma membrane. In support of this notion, loss of either LdMT or LdRos3 led to an increased cell surface exposure of natural PE, as evidenced by enhanced hypersensitivity to PE-binding peptide and labelling by biotinylated PE-binding peptide. These results are in line with a direct role of the LdMT-LdRos3 complex in pumping natural PE to the cytosolic leaflet to generate an asymmetric membrane as recently suggested for P4-ATPases from other organisms [28,29]. The ability of wild-type parasites to take up NBD-PC suggests that natural PC is restricted to the inner leaflet of the parasites plasma membrane and that the outer leaflet is primarily composed of sphingolipids, as NBD-sphingomyelin is not taken up. The substrate specificity of the inward translocation machinery appears to vary between Leishmania species. In contrast to L. donovani, L. infantum displays not only an ATP-dependent internalization of NBD-PC and NBD-PE but also an active transport of NBD-PS across its plasma membrane [30]. Thus, this parasite might express one or more P4-ATPases of broader or different substrate specificity at the plasma membrane, respectively. The physiological relevance of these species-specific differences in substrate specificity remains to be established. Previous studies suggested that the LdMT-LdRos3 complex of L.donovani also recognizes NBD-PS, however, as a poor substrate [11,12]. Significant uptake of NBD-PS in this species was only observed when higher label concentrations were used as compared to NBD-PC and NBD-PE. It is noteworthy that in agreement with former reports [31,32] we could not detect endogenous PS in total lipid extracts prepared from L. donovani promastigotes, suggesting that this parasite does not synthesise considerable amounts of PS. Thus, an inward-directed transport activity for PS at the plasma membrane might not be required in L. donovani. However, all experiments undertaken in the present work were performed on the promastigote early stage. It will be interesting to determine whether L. donovani amastigotes synthesise PS and regulate its distribution in the parasite plasma membrane. A stage specific regulation of the plasma membrane asymmetry is conceivable since amastigotes up-regulate the expression of two P4-ATPases (LmjF30.2260, LmjF34.3220) and a plasma membrane ABC transporter associated with lipid export (LmjF11.1260) [33]. Leishmania amastigotes replicate within the mature phagolysosome compartment and have complex nutritional requirements which must be scavenged from the host cell [34]. It is therefore tempting to speculate that the LdMT-LdRos3 transporter complex or related P4-ATPases are involved in the acquisition of host phospholipids. The surface of the Leishmania parasite is a major point of interaction with the host throughout the infectious cycle. A number of surface glycoconjugates such as lipophosphoglycans, glycosylphosphatidylinositol (GPI)-anchored proteins (e.g. the metalloprotease gp63), and a heterogeneous group of small glycosylinositolphospho-lipids have been implicated in the ability of the parasite to infect and survive in host macrophages [35]. In addition, surface exposed PS has been associated with parasite infectivity [3][4][5]. However, these results must be interpreted with caution because PS exposure was concluded on basis of annexin V and anti-PS antibody binding in these studies. Despite being used extensively to label PS, neither annexin V nor anti-PS antibodies are specific for this lipid and also bind phosphatidic acid, phosphatidylglycerol and phosphatidylinositol-4,5-bisphosphate [36]. Thus, a direct proof for PS exposure by Leishmania parasites is currently lacking. In fact, we observed that L. donovani promastigotes lack significant amounts of PS, and annexin V-FITC does not label living, PI negative L. donovani parasites but parasites positive for PI staining. Conceivably, the various Leishmania spp. might expose other lipids than PS that contribute to the infectivity. Our results on DLdMT and the DLdRos3 Leishmania mutant lines that display an altered PE asymmetry did not reveal major differences in terms of invasion efficiency into the human monocytic cell line THP-1. In line with these results, the DLdMT mutant remains infective and maintains virulence in cultures of primary isolated mouse peritoneal macrophages [37]. These results suggest that disruption of the LdMT-LdRos3 transporter complex and changes in the transbilayer distribution of PE, and probably PC, are not crucial for the infectivity of L. donovani promastigotes. Likewise, in the THP-1 infection model used here PS exposure is not mandatory for invasion of host cells. Future studies on other Leishmania species and mutants lacking the ability to synthesized PS may help to define in more detail the lipid types exposed on the cells surface and involved in parasite internalization and phagocyte inactivation. Protein analysis and immunoprecipitation For preparation of membrane proteins, parasites were harvested, resuspended in ice-cold hypo-osmotic lysis buffer (5 mM Tris-HCl pH 7.4) containing 1 mM phenylmethylsulfonyl fluoride, and then broken by vortexing with glass beads. The cell lysate was clarified by centrifugation at 5006g (10 min, 4uC). Subcellular membranes were collected by centrifugation at 100,0006g (1 h, 4uC) and resuspended in sample buffer (750 mM 6-aminocarproic acid, 50 mM Bistris pH 7.0, 20% glycerol) to a protein concentration of 5 mg/ml. Protein concentration was measured using the BCA protein assay kit (Pierce Chemical Company, Rockford IL, USA). Membranes were solubilised by adding 20 ml of ndodecyl-b-D-maltopyranoside (DDM, 10%) to 100 ml suspended membranes, corresponding to a DDM/protein ratio of 4 (w/w). After incubation for 60 min on ice, insoluble material was removed by centrifugation (100,0006g, 1 h, 4uC). GFP-co-immunoprecipitation assays were performed using anti-GFP microBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Per immuno-isolation, a 1 ml reaction was prepared in lysis buffer containing 150 ml of anti-GFP microBeads slurry, 250 ml of detergent-solubilised protein. The reactions were rotated gently at 4uC for 30 min. Beads were separated from the supernatants using mColumns with a mMACS separator (Miltenyi Biotec) and washed three times with lysis buffer containing 0.05% DDM. Membrane protein extracts and immuno-isolated membrane proteins were subjected to western blot analysis. Immunoblots were probed with polyclonal antibodies against LdMT and LdRos3 [14], and gp63 (kindly provided by Robert McMaster). Horseradish peroxidase-conjugated secondary antibodies were from Bio-Rad (Hercules, CA). Blots were developed using enhance chemiluminescence (ECL plus Kit, GE-Healthcare, Freiburg, Germany). Protein mass spectrometry of the immunoprecipitates was performed as described elsewhere [38]. Promastigotes in logarithmic phase of growth were harvested by centrifugation (10006g, 10 min, 4uC), washed twice with HPMI. To block the catabolism of NBD-lipids the cell suspension (10 7 parasites/ml) was pre-incubated with 5 mM 3-(4-octadecyl)benzoylacrylic acid (Biomol, Hamburg, Germany) and 1 mM phenylmethanesulphonylfluoride for 30 min at 28uC. To study the lipid uptake, the parasites were then incubated at 4uC with 5 mM NBD-lipid. After 30 minutes, cells were washed twice in HPMI containing 4% (w/v) fatty acid free bovine serum albumin to extract NBD lipids from the cell surface. Flow cytometry analysis was performed on a Becton Dickinson FACS (San Jose, CA) equipped with an argon laser (488 nm) using Cell Quest software. One ml of 1 mg/ml PI in H 2 O was added 200 ml cell suspension just before analysis. Ten thousand cells were analyzed at room temperature with gating during the acquisition. Live cells were selected based on forward/side-scatter gating and propidium iodide exclusion. The following fluorescence channels (log scale) were used: FL1 (530/30 nm, FITC, NBD, CellTrackerTM Green), FL2 (585/42 nm, PI, CellTrackerTM Dil). Data were analyzed by Cyflogic software. Annexin V und Bio-Rho Assay To visualize endogenous PE on the cell surface, 5610 6 promastigotes in logarithmic phase of growth were incubated in 20 ml HPMI containing 38 mM biotinylated Ro09-0198 (provided by Kazuma Tanaka). After 1 h at 4uC, cells were washed with phosphate buffered saline (PBS) containing 0.5% (w/v) bovine serum albumin and then fixed with PBS containing 5% (w/v) formaldehyde for 1 h at 30uC. Promastigotes were then washed in PBS, resuspended in 250 ml PBS containing 5 mg/ml Streptavidin-FITC and incubated for 30 min at 25uC prior to microscopy analysis. To measure exposure of endogenous PS on the cell surface, about 5610 5 parasites were incubated on ice for 10 min in the dark with 125 ng annexin V-FITC and 1 mg PI in 0.5 ml of binding buffer (10 mM HEPES pH 7.4, 140 mM NaCl, 2.5 mM CaCl 2 ). Cells were washed, resuspended in 0.5 ml of binding buffer and subjected to microscopy. Fluorescence Microscopy Epifluorescence microscopy and image acquisition were carried out using an inverse Axiovert 100 standard fluorescence microscope (Carl Zeiss, Oberkochen, Germany) equipped with a cooled CCD camera (Coolsnap, visitron systems, Puchheim, Germany) driven by Metamorph software (Universal Imaging, Downingtown, USA). NBD fluorescence was viewed using a Plan-APO 1006/1.3 NA oil objective with the following filter set: BP 450-490, FT 510, and BP 512-542. Confocal laser scanning microscopy was performed using an inverted Fluoview 1000 microscope (Olympus, Tokio, Japan) and a 606 (N.A. 1.35) oilimmersion objective. Fluorescence of FITC and CellTracker Green was excited with a 488 nm argon laser and recorded between 500 and 530 nm. Fluorescence of PI and CellTracker CM-Dil was excited with a 559 nm argon laser and recorded between 570 and 600 nm. Images with a frame size of 2566256 pixels were acquired.	2014-10-01T00:00:00.000Z	2010-08-26T00:00:00.000	{ "year": 2010, "sha1": "71e25e2b2c02063ed869e1ed08528414e3214911", "oa_license": "CCBY", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0012443&type=printable", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "a00dc6df85c00d69b63287db8abdd9da9c6df7e4", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
249138858	pes2o/s2orc	v3-fos-license	Monitoring of the Drying Process by Direct Solid-Phase Microextraction : SPME is known as a rapid and convenient method for analyzing air-borne organic compounds in closed systems ('headspace SPME'). In this investigation we have applied this technique to the monitoring of the drying process of chemicals containing small amounts of organic solvents under specific conditions. Interesting results were found and some of them are presented and discussed in this paper. Introduction The drying of solid materials (chemicals, foods, pharmaceuticals) is a major industrial unit operation. It consists of removing a liquid from the solid material up to an acceptably low value. For pharmaceutical products the limits are generally in the mglkg or even in the Ilglkg range. For the control of the process, it is necessary to continuously monitor the moisture content during the operation. This is usually done by sampling and off-line analysis in the laboratory, which is time consuming and unsuitable for the control of rapidly changing processes. On-line monitoring is obviously of greatest interest. In this paper we present some results obtained in our laboratories using headspace solid-phase microextraction (SPME) as a monitoring method in the drying of pharmaceuticals. [2]. It is a valuable technique for analyzing air-borne organic compounds. The method is a headspace sampling technique which concentrates volatile organic compounds (analytes) by adsorption onto a selective polymer-coated silica fiber prior to thermal desorption in the injection port of a gas chromatograph. It is especially suitable and straightforward when concentrations are constant over time i.e. for systems which are fully equilibrated. Quantitation in the absence of full equilibrium is much more complex because the kinetics of extraction are limited by the diffusion within the fiber coating and through a boundary layer. Some models have been developed recently [3]. Experimental Design and Conditions The laboratory drying equipment used was a modified Biichi Rotavapor ( Fig. 1) from which the cooling system had been removed in order to keep the headspace inside the apparatus as small as possible. The solid material to be dried was placed in the 250 ml rotating round-bottom flask and the nitrogen flow was introduced through a Teflon tubing insert into the core of the solid. The temperature of the process was regulated with a thermostated water bath. Dry nitrogen from a tank was directed through a rotameter and the flow periodically checked for accuracy with a bubble flowmeter. Vacuum was produced by a water aspirator and maintained at the desired value by a pressure regulator. The wet material used in this investigation was a pharmaceutical intermediate containing -4% of isopropanol. The process was monitored using a SPME fiber from Supelco with a 65 11m CW/DVD coating. The volatiles emitted were collected at different times by inserting the needle of a SPME fiber through a silicone rubber septum, located at the outlet of the equipment. The total amount of volatile analytes was measured by direct injection in a simplified gas chromatograph (TV 9000; Brechbi.ihler AG) developed in our laboratories and operating without a separating column [4]. It consists of an injector for a wide-bore column linked to a standard FID detector by a short capillary of 30 cm length. The signal was interpreted using the integration program Chrom-Card (Fig. 2). The sampling time was 2 min and the injection time I min. The drying experiments were carried out under following experimental condi- ..... General Procedure The material to be dried was weighed (-60 g) and placed in the apparatus and the experimental conditions set up. After a stabilization time of -1 h, the experiment was started and SPME analyses were made at different times. Periodically, the dryer was opened and a solid sample (-lg) was withdrawn for analysis. After recording the weight of the residual solid (mass balance), the equipment was rapidly reset for further measurements. The analyses of solid samples were made using an internal headspace GC-procedure. and the partition coefficients between the that we could obtain reliable results withfiber and the gas phase, K fg , depend on out reaching full equilibrium by using many parameters such as the type of ana-consistent sampling time and keeping the lytes, fiber type and operating conditions operating parameters constant. For a (temperature and pressure). Knowing sampling time of 2 min one can consider the proportionality constants and the the measured concentrations as being amounts captured by the fiber, concentra-'quasi' steady-state concentrations, which tions in the gas phase can be easily calcu-simplifies calculations greatly. The related. Unfortunately, very few prop or-producibility of the results was excellent tionality constants have been published for experiments carried out at atmospherand general models to calculate them are ic pressure but poor for the work carried not readily available. A viable practical out under reduced pressure. The latter is approach to overcome this difficulty is to . primarily due to the bad regulation of the calibrate the fiber for the given analytes pressure (stroke regulation). and to measure long enough to ensure complete equilibration. In our study the analyte concentration changed constantly during the drying process, rapidly at the beginning and more slowly at the end. Fortunately, the equilibration time of the fiber for our solvent was fairly short (within minutes), so Adsorption of volatiles onto the SPME fiber has been shown to be dependent on the temperature, pressure, and gas circulation in the drying apparatus. The most efficient parameter in the drying process proved to be the gas circulation. Thermodynamic Considerations A fundamental property of SPME is that the analytes partition between the fiber coating and the gas and calculation of the concentrations always involves equilibrium relationships. At equilibrium, the concentration of the analytes in the fiber coating is directly related to the concentration in the gas phase: order of the process [-]. Values of a between 0.9 and 1.1 were found for all our experiments, which indicates that drying is a first-order process (Fig. 4). Calibration of the SPME fiber Response factors were determined by preparing samples of isopropanol volumetrically (0-5 Ill) in 22 ml vials with screw and PTFE septum and injecting them into the TV 9000 (three replications). Responses varied with the fiber type, the age of the fiber and the experimental conditions. Correlations were excellent. On-line Monitoring of the Drying Process by SPME By using appropriate calibration curves, it was possible to monitor with a time interval 5' at a frequency of ] in 5' the amount of residual solvent in the solid at any time by direct SPME, without solid sampling (Fig. 5). Thus, the end-point of the dJ:ying process can be predicted with a good accuracy, which saves a great deal of time. Experiment 20 6. Conclusions The on-line monitoring of the drying process by using direct headspace SPME has been investigated. An accurate prediction of the end-point of the process can be made, provided adequate calibration curves are available. We are extending our investigations to other operating conditions, especially to the work under reduced pressure, with and without gas circulation.	2022-05-29T15:14:59.047Z	2001-09-26T00:00:00.000	{ "year": 2001, "sha1": "59ba5b628ae2685aadb43cfc563e9566cfc4f5a6", "oa_license": "CCBYNC", "oa_url": "https://www.chimia.ch/chimia/article/download/2001_741/2764", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "dbae848073d7ece3daf32949eae1b0037f768203", "s2fieldsofstudy": [ "Chemistry", "Environmental Science" ], "extfieldsofstudy": [] }
96446870	pes2o/s2orc	v3-fos-license	Large Binocular Telescope observations of PSR J2043+2740 We present the results of deep optical imaging of the radio/$\gamma$-ray pulsar PSR J2043+2740, obtained with the Large Binocular Telescope (LBT). With a characteristic age of 1.2 Myr, PSR J2043+2740 is one of the oldest (non recycled) pulsars detected in $\gamma$-rays, although with still a quite high rotational energy reservoir ($\dot{E}_{\rm rot} = 5.6 \times 10^{34}$ erg s$^{-1}$). The presumably close distance (a few hundred pc), suggested by the hydrogen column density ($N_{\rm H} \lesssim 3.6 \times 10^{20}$ cm$^{-2}$), would make it a viable target for deep optical observations, never attempted until now. We observed the pulsar with the Large Binocular Camera of the LBT. The only object (V=25.44$\pm$0.05) detected within ~3"from the pulsar radio coordinates is unrelated to it. PSR J2043+2740 is, thus, undetected down to V~26.6 (3-$\sigma$), the deepest limit on its optical emission. We discuss the implications of this result on the pulsar emission properties. INTRODUCTION The launch of the Fermi Gamma-ray Space Telescope has spurred on the search for pulsars in γ-rays (Grenier & Harding 2015), yielding over 200 1 detections and triggering multi-wavelength observations. While pulsars are common targets in the X-rays, they are very challenging targets in the optical and very few of them have been identified (see Mignani et al. 2016 and references therein). Here we report on Large Binocular Telescope (LBT) observations of an isolated pulsar, PSR J2043+2740, detected by both AGILE (Pellizzoni et al. 2009) and Fermi (Abdo et al. 2010;Noutsos et al. 2011). It was discovered as a radio pulsar (Ray et al. 1996) and later on as an X-ray source by XMM-Newton (Becker et al. 2004), although X-ray pulsations have not yet been found. PSR J2043+2740 is one of the very few non-recycled pulsars older than The LBT is an international collaboration among institutions in the United States, Italy and Germany. LBT Corporation partners are: The University of Arizona on behalf of the Arizona (Ray et al. 1996). This also yields a rotational energy loss rateĖrot = 5.6 × 10 34 erg s −1 and a surface dipolar magnetic field Bs = 3.54 × 10 11 G 2 . Although PSR J2043+2740 does not have a very large spin-down power compared to young (∼ 1-10 kyr) pulsars (∼ 10 36 -10 38 erg s −1 ), it is still a factor of two larger than that of middle aged γ-ray pulsars (∼ 0.1-0.5 Myr), such as Geminga, PSR B0656+14, and PSR B1055−52, all detected in the optical thanks to their distances 500 pc (Abdo et al. 2013). The distance to PSR J2043+2740 is uncertain owing to the lack of a radio parallax measurement. The radio dispersion measure (DM=21.0±0.1 pc cm −3 ; Ray et al. 1996) gives a distance of 1.8±0.3 kpc from the NE2001 model of the Galactic free electron density (Cordes & Lazio 2002). A slightly smaller distance (1.48 kpc) is inferred from the model of Yao et al. (2017). The hydrogen column density towards the pulsar obtained from the X-ray spectral fits (NH 3.6 × 10 20 cm −2 ; Abdo et al. 2013) suggests a distance of a few hundred pc (He et al. 2013), although these estimates depend on the model X-ray spectrum. Such a distance would make PSR J2043+2740 a viable target for deep optical observations, never carried out until now, and might be compatible with the debated association (Noutsos et al. 2011) with the Cygnus Loop supernova remnant (SNR) at 540 +100 −80 pc (Blair et al. 2005). The structure of this manuscript is as follows: observations and data reduction are described in Sectn. 2, whereas the results are presented and discussed in Sectn. 3 and 4, respectively. OBSERVATIONS AND DATA ANALYSIS The PSR J2043+2740 observations were carried out on July 5th, 2016 with the LBT at the Mount Graham International Observatory (Arizona, USA) and the Large Binocular Camera (LBC; Giallongo et al. 2008). The Camera's field of view is 23 ×25 , with a pixel scale of 0. 2255. The images were taken through the filters SDT-Uspec, V-BESSEL, and i-SLOAN, closely matching the Sloan filters u and i (Fukugita et al. 1996), and the Johnson V filter. For each filter, three sets of exposures were acquired with exposure times of 20s, 60s and 120s, for a total integration of 5887 s (Uspec and i-SLOAN) and 5376 s (V-BESSEL). Sky conditions were nonphotometric owing to the presence of cirri and the average seeing was around 1. 2. The target was observed with an average airmass around 1.01 and 1.09, and with a lunar illumination of ∼ 1%. Images were reduced with the LBC data reduction pipeline, correcting raw science frames for bias, dark and flat fields. A further low-order flat-field correction was obtained from the night sky flats to remove large-scale effects. We, then, corrected the images for geometrical distortions, applying a linear pixel scale resampling. Finally, we stacked all images taken with the same filter and used the master frames to compute the astrometric solution (∼ 0. 1 rms). Since the night was non-photometric, we performed the photometric calibration directly on the science frames by matching stars in public source catalogues. In particular, for the SDT-Uspec filter we used a source list extracted from the XMM-Newton Serendipitous Ultra-violet Source Survey Catalogue version 3.0 (XMM-SUSS3 3 ), built from observations with the XMM-Newton Optical Monitor (OM; Mason et al. 2001), whereas for both the V-BESSEL and i-SLOAN filters we used a source list from the American Association of Variable Stars Observers (AAVSO) Photometric All-Sky Survey 4 (APASS). All magnitudes are in the AB system (Oke 1974). For all filters, we computed object photometry with the DAOPHOT II software package (Stetson 1994) following a standard procedure for source detection, modelling of the image point spread function (PSF), and multi-band source catalogue generation (see, e.g., Testa et al. 2015). After accounting for photometric errors, the fit residuals turned out to be ∼ 0.01 magnitudes in all filters, to which we must add the average absolute photometry accuracy of SUSS3 and APASS, which is ∼ 0.05 magnitudes. al. (2015). PSR J2043+2740 has not been observed by Chandra, so that we cannot rely on an accurate, model-independent position. No proper motion has been measured for PSR J2043+2740. Therefore, to account for its unknown angular displacement between the epoch of the reference radio position and that of our LBT observations (MJD=57574), we looked for candidate counterparts within a conservative search region of 3 radius. This is three times as large as the formal radio position uncertainty and roughly equal to the angular displacement expected for a pulsar moving with an average transverse velocity of 400 km s −1 ) at a distance as close as the Cygnus Loop SNR (540 +100 −80 pc; Blair et al. 2005). Only one object is detected within the search region (3 radius) defined above (Fig. 1). The object is barely visible in the V band and not in the U band, whereas it is clearly detected in the i band. Its magnitudes have been computed following the same procedure as described in Sectn. 2 and are V=25.44±0.05, i=25.08±0.08, U>26.5 (AB). To investigate the characteristics of the object, we built a U−V vs V−i colour-colour diagram (CCD) of all objects within 5 from the pulsar position and compared its colours with respect to the main sequence (Fig. 2). Since the field stars are, presumably, at different distances with respect to the pulsar, the diagram is uncorrected for the reddening. The object's colours are V-i = 0.36 ± 0.09, U-V > 1.06 and are close to those of the main sequence. This means that it does not stand out for having peculiar colours, as one would expect for a pulsar, which is usually characterised by blue colours (e.g., Mignani & Caraveo 2001;Mignani et al. 2010). We compared the observed CCD to a synthetic one computed with the Besançon Model of Stellar Population Synthesis to simulate the Galactic stellar population within a 5 radius around the direction of PSR J2043+2740. As shown in Fig.2, the main sequence of the observed CCD is consistent with the model Galactic stellar population, supporting the conclusion that the object is a field star rather than the pulsar. Estimated 3σlevel limiting magnitudes are 26.5, 26.6, and 26.2 in the U, V and i bands, respectively, which we assume as upper limits on the pulsar fluxes. RESULTS LBT observations of PSR J2043+2740 3 dist <= 5 kpc 5 < dist <= 10 kpc 10 < dist <= 15 kpc 15 < dist <= 30 kpc dist > 30 kpc Figure 2. Synthetic CCD for the Galactic stellar population (in colour for different distances) in the pulsar direction simulated with the Besançon model (Robin et al. 2004), superimposed to the observed one (black). The spread in the synthetic CCD is the effect of the distance-dependent reddening correction, applied by the model, and the simulated photometric error. DISCUSSION Our observations of PSR J2043+2740 are much deeper than those obtained by Beronya et al. (2015) with the BTA (Bolshoi Teleskop Alt-azimutalnyi) 6m telescope, which only yielded a 3σ limit of R ∼ 21.7 (Vega) on the pulsar flux. The pulsar is obviously too faint to have been detected in the XMM-Newton OM images (Becker et al. 2004), with 3σ limits of B≈21.5 and U≈20.9 (Vega). We checked whether our limits on the pulsar flux could help to prove or disprove the association with the Cygnus Loop SNR. In general, the non-thermal optical luminosity Lopt of rotationpowered pulsars scales with a power of the rotational energy loss rate (see, e.g. Mignani et al. 2012) as Lopt ∝Ė 1.70±0.03 rot (1σ statistical error). From this relation, we estimate a luminosity of ∼ 3.16×10 27 erg s −1 for PSR J2043+2740, corresponding to a magnitude V ∼ 26.2-26.9 at the distance of the Cygnus Loop SNR, after accounting for the interstellar reddening E(B − V ) 0.06, inferred from the NH (Predehl & Schmitt 1995). Therefore, our detection limit (V∼ 26.6) does not determine whether the pulsar is at the distance of the Cygnus Loop SNR, and their association remains uncertain. Pushing the limit on the pulsar brightness down to V∼28 would imply a distance larger than ∼ 1 kpc for the same predicted optical luminosity and would disprove this association. Given the lack of a counter-evidence, we assume the pulsar DMbased distance (Yao et al. 2017) as a reference. The γ-ray spectrum of PSR J2043+2740 is described by a PL with an exponential cut off, where photon index Γγ = 1.44 ± 0.25 and cut off energy Ec = 1.34 ± 0.37 GeV (Acero et al. 2015). The XMM-Newton spectrum can be fit by a PL with ΓX = 2.98 +0.44 −0.29 (Abdo et al. 2013). The addition of a blackbody component is compatible with the counting statistics, but an f -test (Bevington 1969) shows no improvement in the fit significance. We compared our optical flux measurements with the extrapolations of the high-energy spectra, after correcting for the reddening using the extinction coefficients of Fitzpatrick (1999). The spectral energy distribution (SED) of PSR J2043+2740 is shown in Fig. 3. As seen in other cases, the extrapolations of the two PL spectra are not compatible with each other, implying a turnover in the γ-ray PL at low energies. This is also observed in, e.g. the middle-aged pulsar PSR The blue and red lines are the extrapolation of the X and γ-ray PL spectra, respectively (Abdo et al. 2013;Acero et al. 2015), with their 1σ errors (dashed lines). The blue and red regions mark the range where the X and γ-ray spectra were measured. B1055−52 (Mignani et al. 2010), although there is no apparent correlation between the presence of a turnover and the pulsar characteristic age. The optical flux upper limits are below the extrapolation of the assumed X-ray PL spectrum but are not deep enough to rule out that the optical emission might be compatible with the γ-ray PL extrapolation. This could be a rare case where the γ-ray and optical spectra are related to each other.	2017-09-26T16:42:18.000Z	2017-09-26T00:00:00.000	{ "year": 2018, "sha1": "2200d1852f3ab2134f64335b63b9020bff72ab4d", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1709.09169", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "2200d1852f3ab2134f64335b63b9020bff72ab4d", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
83947807	pes2o/s2orc	v3-fos-license	http://iaesjournal.com/online/index.php/IJAAS Self-fertility of Four Female Parent Clones of Ananas comosus L., involved in a 6x6 Complete Diallel Mating System with Selfings using the Typological Info 2013 To determine the cropping type to apply to four clones of Ananas comosus in farms, their behaviour under hand selfings was analysed. 103-104-6, 410-106-33 and 410-200-15 hybrid female clones and RE43 Queen Victoria clone as well as HA10 and HA25 as controls were involved in a 6 x 6 complete diallel crossing system with selfings. The total seeds number derived from self hand-pollinations per week, mean seeds number obtained per self-pollinated flower and per week, weight of ripe fruit and bloomed flowers number per week were measured. The Anova, Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) were successively run. Results showed that 410-200-15 was revealed self-incompatible, while RE43 is found to be self-sterile. In the same way, 410-106-33 expresses self-sterile behaviour, whereas 103-104-6 shows the self-fertile one. The behaviour under selfings of both 410-200-15 and 410-106-33 comes from their HA10 Smooth Cayenne female parent which was previously characterised like self-incompatible. The structurings provided by the Anova and HCA are globally comparable. The 410-200-15 hybrid clone can already be recommended for on-farm trials under mono-crop. Nonetheless, the multi-crop will be envisioned once panmixia results will have demonstrated its inter-sterility. The 410-106-33 and 103-104-6 must be first subjected to successive back crosses before their on-farm trials. The RE43 clone must be cultivated in one crop. The morphological relatedness of five clones is INTRODUCTION Ananas comosus L. Merr, fruits king, is a perennial, monocotyledonous, herbaceous, diploid (2n=2x=50) plant of the Bromeliaceae [1]. In world, its cultivation covers all tropical and subtropical areas. It is used as textile, ornamental plant, among others... Its edible fruit is the object of a very active international trade with an annual world production over 14.6 million tons [2], [3]. In Furthermore, the seedy fruits are not appreciated by the consumers and traders. Indeed the presence of seeds in fruit depreciates its quality [4]. A breeding programme of pineapple was initiated and conducted at the Station of Anguededou / DFA (Département des Fruits et Agrumes) from the IDEFOR (Institut des Forêts) at the present time CNRA (Centre National de Recherche Agronomique) [5]. This programme aimed to create some varieties intended for export of either fresh fruits or canning. In the first time, apart from the traditional criteria of yielding, quality and spineless, this programme also aimed to prevent the internal fruit browning in increasing its content in ascorbic acid. In the second phase, it concerned adaptation at warm and dry climates, resistance to diseases and pests, especially nematodes and Phytophthora sp., as well as resistance to yellowing. The latter is a physiological disorder of the ripening. This programme consisted in hybridizing two cultivars Smooth Cayenne and Mordilona and achieving the multi-characters selection of 40,000 hybrids created [6]. From these created hybrids, nineteen were selected of which three 410-106-33, 410-200-15 and 103-104-6 identified like the best [5]. Until now, we do not know the behaviour of these three hybrids in relation to the self-fertility or self-sterility. It seems that the three created hybrids are selfincompatible like one of their parents namely HA10 clone [7]. The knowledge of their either self-fertility or self-sterility could help to envision the cropping type either one or mixed cropping which could suit them. The aim of this to determine the cropping type to apply to four female clones 103-104-6, 410-106-33 and 410-200-15 as well as RE43 Queen Victoria involved in a 6 x 6 complete diallel crossing system with selfings. RESEARCH METHOD 2.1. Experimental site, plant material and self hand-pollinations The experiment was conducted from September 1997 to April 1998. It was carried out in the Research Station of the Anguededou at DFA/IDEFOR (Département des Fruits et Agrumes /Institut Des Forêts) in Côte d'Ivoire. This Station is located in latitude 5°25', north and longitude 4°08', west as well as 25 m above sea. Four common tester and parent clones 103-104-6, 410-106-33, 410-200-15 and RE43 as well as two others HA10 and HA25 as controls were used. The first three are of the recently created hybrids [5]. They were identified as superior hybrids. The clones 410-200-15 and 410-106-33 come from Smooth Cayenne x Perolera inter-variety crosses. The clone 103-104-6 comes from Perolera x Perolera intra-variety cross [5]. The Queen Victoria clone RE43, more sugared than cultivar Smooth Cayenne, is very valued on the international market. The clones HA10 and HA25 belong to cultivar Cayenne. They are also well valued on the international market. Moreover, the calcium carbide solution was applied in the leafy crown of each plant chosen as parents to force the flowering. Two months after such an applying the inflorescences emerging from plant were bagged with the isolation bags. Every day, in the morning the anthers surmounting the stamens were collected in Petri dish by means of tongs. They were manually used either to self-pollinate or cross-pollinate the bloomed flowers. These ones were scored then marked with red oil painting. At maturity, the fruits were harvested and weighed. The seeds contained in individual fruits were also counted after dissection. Experimental design, mating system and measured variables All of parent clones already belonged to pre-existent design. The latter consisted of plant of pineapple laid out on two rows. Six plants per clone were chosen as a function of their vigour and used as both common parents and testers. One for selfings and the five others for crossings. They were planted in two rows on the ridges. Plants had 40 x 25 cm spacings. A gap of 90 cm was maintained among ridges. Treatment consisted of a pineapple plant laid out on a ridge. In all, six treatments including crossings and selfings were tested. The clone RE43 from cultivar Queen Victoria was planted in the collection. The 410-200-15, 410-106-33 and 103-104-6 clones were set in plant multiplication fields. The HA10 and HA25 control clones were placed in monthly plantation. The six clones were involved in a complete diallel crossing system with selfings. Sole selfings were analysed in this paper. Four variables were measured: 1) the total seeds number deriving from self hand-pollinations per week (Nbseed), 2) mean seeds number obtained per self hand-pollinated flower and per week (Seedflow), 3) weight of ripe fruit (Weigfruit) and 4) bloomed flowers number per week (Blooflow). Data analysis The data set was processed by Xlstat 2007.6 software. The Anova incorporating the means separation, Principal Component (PCA) and Hierarchical Cluster Analyses (HCA) were run to interpret the variability. The means were separate in two times. First, Dunnett's test was applied to identify at more three classes. These are : 1) the class of female clones of which the means were on this side of that of controls, 2) class of the female clones of which the means were comparable to controls and 3) that of parent clones whose the means were beyond controls. Within each class the Newman-Keuls or Student t tests at 5% threshold were performed. The total seeds number deriving from self hand-pollinations per week (Nbseed) and mean seeds number obtained per self hand-pollinated flower and per week was subjected to the square root transformation, because they were not normally distributed. The PCA allowed the structuring of descriptors and individuals represented by the four female parent clones. The PCA also allowed the analysing of the proximity among identified groups. Prior, the number of factors or components used to interpret the variability was determined by means of Kaiser and angle criteria. The most relevant descriptors were selected from their representation quality namely QLT kl and Pearson's linear correlation coefficient. RESULTS AND ANALYSIS 3.1. Variability of the self-sterility of four female clones by descriptor With both the total seeds number deriving from self hand-pollinations per week (Nbseed) and mean seeds number obtained per hand self hand-pollinated flower and per week (Seedflow), three groups were identified in relation to Dunnett's test. First, composed of 410-200-15 female clone, was characterized by the lowest seeds production potential by self hand-pollinations per week. Second, consisted of RE43 female parent clones including HA10, HA25 control clones, was marked by mean production ability of seeds by self hand-pollinations per week. Third, constituted of 103-104-6 and 410-106-33 hybrid clones, was distinguishable by high seeds production potential deriving from hand self-pollinations. Within G2 group, no significant difference was noted among RE43, HA10 and HA25 using the Newman-Keuls' test. In the same way, within G3 group, no statistical difference was evidenced between 103-104-6 and 410-106-33 resorting to Student's t test. In all, three groups were observed: 1) G1 constituted of 410-200-15 female clone, 2) G2 group composed of RE43 clone with two HA10 and HA25 control clones and 3) G3 group comprising 103-104-6 and 410-106-33 hybrid female clones. Variability on both sides of average varied from 3.82 to 19.08% for Nbseed, as against from -∞ to 18.97 for Seedflow. The Untransformed averages fluctuated from 00 to 21.846 seeds for Nbseed, while those of Seedflow varied from 00 to 0.943 (Table 1). Regarding the weight of ripe fruit (Weigfruit), after Dunnett's test, three groups were identified. First composed of 410-106-33 and RE43 was distinguishable by averages on this side of control. Second constituted of HA10 control. The weight of its fruits was significantly higher than that of two abovementioned clones. Third consisted of 410-200-15 and 103-104-6 including control HA25. They produced the heaviest fruits. Within the first group, two sub-groups were observed after Newman-Keuls test: 1) represented by 410-106-33 was characterized by fruits of weak weight and 2) comprising clone RE43 was marked by fruits of elevated weight. Likewise, within the third group, three sub-groups were noted. First comprising hybrid clone 410-200-15 differed from two others by low weight of fruit. Second composed of second control HA25. It was distinguishable by fair weight. Third consisted of hybrid clone 103-104-6, was distinguishable from two previous by the highest fruit weight. The gaps between the average and each of observations oscillated from 0.05 to 0.11%. The untransformed averages of this variable stretched out from 0.000 to 0.943 Kg. (Table 1). Before interpreting results, self-incompatibility and self-sterility were defined. Self-incompatible clone is defined as the one which by self hand-pollination does not produce any seeds, but produces them by cross hand-pollination. Likewise the self-sterile one is the one which by hand selfing shows a mean seeds production per self pollinated flower null [8]. In sum, The HA10, HA25, RE43 and 410-200-15 female parent clones both expressed low production potential of seeds through the Anova (Table 1). Except for 410-106-33 parent clone, such an observation was confirmed by the HCA searching for proximity among three a posteriori groups (Table 5). According to previously given definitions, 410-200-15 hybrid female clone was classified like self-incompatible. Our works also showed that HA10 and HA25 control clones are self-sterile. In contrast, Cabot (1989) both ranked HA10 and HA25 clones like self-incompatible. Thereby, the self-incompatibility of 410-200-15 hybrid clone finds an explanation through self-sterility of its parent HA10. The RE43 clone was identified like self-sterile (Table 1). Moreover, the fruits weight of 410-200-15 and HA25 clones ranked them in grade C (0.9 to 1.1 Kg). They could be intended to export in this grade. Weight of fruit from RE43 classes it in grade lower than D for export. Consequently, the new 410-200-15 hybrid clone can already be recommended for on-farm trials under monocrop. Nonetheless, the multi-crop will be envisioned once panmixia results will have demonstrated its intersterility. Likewise, As regards the two others represented by 103-104-6 and 410-106-33, the former can be considered as a self-fertile, while the latter like self-sterile ( Table 2). The self-fertility is due to the lack of recognition between stigma and pollen proteins, while self-sterility is caused by the existence of such a recognition [9]. At the moment where pollen makes contact with stigma, if it does not exist no recognition between proteins of two organs, this pollen germinates and issues pollen tube. This one lengthens in stylar canal up to ovary. The vegetative nucleus weathers, whereas the reproductive one divides into two antherozoa. They fertilise eggs contained in embryo bag of ovule. The incompatibility in pineapple is under gametophytic control with either S or S/Z polymorphic loci according to authors. Thereby, Majumber et al., (1964), Brewbaker and Gorrez (1967) and Coppens d'Eeckenbrugge et al. (1997) defended the hypothesis of an only one locus [6], [10], [11]. In contrast Hayman (1956), Cardin (1990) and Issali (1998) postulated the assumption of S and Z independently segregating loci [8], [12], [13]. . They will be intended to export in this grade, on condition of reducing self-fertility of 103-104-6. In brief, 401-106-33 female clone can be cultivated in only one crop. However, by reason of the presence of lots of small corms at fruit basis and multiple crowns on fruit (data not shown), 410-106-33 should be subjected to successive back-crosses with 103-104-6. The former will use as a HA10 recurrent parent clone, whereas the latter will use its Perolera female parent. 3.2. Variability of the self-sterility of four female parent clones with the descriptors as a whole F1 and F2 principal components synthesised the information as a whole contained in the four initial descriptors. They were used in the course of the study to describe and interpret the variability (Figure 1; Table 2). From the PCA, the four initial descriptors allowed the structuring of four female parent clones in two a priori groups. First, composed of 103-104-6 female clone. Second, constituted of RE43, 410-106-33, 410-200-15 parent clones with the control ones namely HA25 and HA10. This same PCA showed that among the four initial descriptors, sole two were relevant. These are the mean seeds number obtained per self hand-pollinated flower and per week (Seedflow) as well as weight of ripe fruit (Weigfruit). Therefore, they were used in the rest of the study to cluster a posteriori studied clones via the HCA. This, to search for relationship between the a priori groups and the a posteriori ones. In the opposite, the total seeds number deriving from self hand-pollinations per week (Nbseed) and bloomed flowers number per week (Blooflow) was eliminated from the study (Figures 2 and 3; Tables 2, 3 and 4). In brief, F1 and F2 were the best selected principal components using Kaiser's and angle criteria ( Table 2). They synthesised the essential of information contained in the four initial descriptors. Such an approach was also been used in Yao (2012) [14]. Moreover, the second a priori group consisting of RE43, 410-106-33, 410-200-15 parent clones including the control ones namely HA25 and HA10 provided the least seedy fruits. The mean of group seem to be widely influenced by self-incompatibility of 410-200-15 clone ( Table 1). Concerning descriptors choice, the mean seeds number obtained per self hand-pollinated flower and per week (Seedflow) as well as weight of ripe fruit (Weigfruit) were found to be relevant, hence selected (Table 3). They expressed better representation quality (QLT kl ) on the 1-2 plane. Legend Cos 2 : square cosine on F1 and F2 factorial axes of used descriptors. QTL Kl : representation quality of used descriptors on the 1-2 plane of the correlation circle of the PCA. The Nbseed, Seedflow and Weigfruit expressed a good representation quality. This one was greater than or equal to 0.9. Footnote relating to the Principal Component Analysis According to Kaiser's criterion, sole F1 and F2 factorial axes recorded eigenvalues higher than 1 (F1 eigenvalue = 2.424; F2 eigenvalue = 1.237; F3 eigenvalue = 0.335; F4 eigenvalue = 0.004). After angle criterion, the frequencies histogram at point (F3;3) recorded brutal falling of the eigenvalue. In figure 1, this point corresponds is the point of inflection. Beyond this point, there is not more information but, on this side of this point, F1 and F2 axes contain the essential information. In sum, F1 and F2 factorial axes were retained for the rest of the study to analyse the variability. These two factorial axes described 91.54% total variation. F1 factorial axis explained 60.61% total variation. The seeds number deriving from cross-pollinations per week (Nbseed) and seeds number obtained per self-pollinated flower and per week (Seedflow) were well represented there (Table 3). This axis represented the ability of tested parents to produce seeds in selfings. F2 factorial axis was accounted for 30.93% residual variation unexplained by F1 factorial axis. This axis stated the rhythm of flowers issue. So it represented the potential of tested parents to yield flowers ( Figure 2). The projection of four female parent clones on the 1-2 plane of the factorial map with the four descriptors as a whole allowed the observing of some groups. On the principal plane, two a priori structured groups were observed. These are : 1) G1, composed of 103-104-6 clone. 2) G2-G3 consisting of RE43, 410-106-33, 410-200-15 clones with HA25 and HA10 controls ( Figure 3). Hierarchical Cluster Analysis The HCA performed with the two most relevant descriptors which are Seedflow and Weigfruit provided three a posteriori groups as against two a priori identified through the PCA. G1 group, consisted of the only 103-104-6 clone, was both characterised by high mean seeds number obtained per self handpollinated flower and per week as well as weight of ripe fruit. G2 group, constituted of RE43 and 410-106-33 clones including the control one namely HA10, was both singularisable by low mean seeds number produced per self hand-pollinated flower and per week as well as weight of ripe fruit. G3 group, composed of 410-200-15 clone with the control one termed HA25, both stood out from two aforementioned clones by very weak mean seeds number obtained per self hand-pollinated flower and per week, but mean weight of ripe fruit ( Figure 4; Table 4). Calculated genetic distance, by means of euclidian distance, showed that G2 and G3 groups would be related. It could belong to the same group. Finally, two big groups would exist from the six studied clones. The former named G1 goup, comprising 103-104-6 clone, was both distinguishable by strong mean seeds number obtained per self hand-pollinated flower and per week as well as weight of ripe fruit (Seedflow = 1.373; Weigfruit = 1220). The latter termed G2 2-3 group, consisting of 410-106-3, 410-200-15 and RE43 including the control ones namely HA10 and HA25, was both singularisable by low mean seeds number obtained per self hand-pollinated flower and per week as well as weight of ripe fruit (Seedflow = 0.1765; Weigfruit = 766.125; Table 5). Proximity among the identified groups G2 and G3 groups would be morphologically related. In short, two big groups could be identified from the six clones parents typed with two out of four initial descriptors : 1) G1 group composed of sole 103-104-6 clone and 2) G2 2-3 group, constituted of 410-106-3, 410-200-15 and RE43 with the control ones termed HA10 and HA25 both produced the lowest mean seeds number obtained per self hand-pollinated flower and per week as well as weight of ripe fruit (Table 5). In short, G3 group constituted of 410-200-15 and HA25 parent clones showed the weakest seeds production potential, but mean weight of ripe fruit (Table 4). Nonetheless, at this group can be added G2 group composed of RE43 and 410-106-33 clones as well as HA10 control. The big G2 2-3 group is the same than the one previously identified via the factorial map of the PCA (Figure 3). This clustered G2 2-3 group could be morphologically related. Indeed, HA25 and HA10 belong to Smooth Cayenne cultivar. The latter originated from Venezuela where it was selected and cultivated by Indians. Thereafter, it was introduced from Cayenne, in French Guyana, in 1820 (http://en.wikipedia.org/wiki/Pineapple). RE43 clone would be originated from South Africa. Queen Victoria and Smooth Cayenne clones are considered as varieties, but not as different cultivars [6]. They come from accumulation of minor somatic mutations. Hence, they would be very genetically related. 410-106-33 and 410-200-15 are descended from ♀HA10 x ♂Perolera crosses. They bring the genes inherited from their mother. This could explain the relatedness among clones constituting G2 2-3 group. In contrast, 103-104-6 hybrid clone comes from ♀Perolera x ♂Perolera cross. It might bring infrequent alleles justifying its distance in relation to clones from G2 2-3 group. Nevertheless, they belong to the same species complex [6], and hence are all cross-fertile.	2019-03-20T13:08:03.932Z	2013-06-01T00:00:00.000	{ "year": 2013, "sha1": "c50b4afbc87cb8b57619504ef5bc08c4d9276002", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.11591/ijaas.v2i2.5902", "oa_status": "HYBRID", "pdf_src": "Adhoc", "pdf_hash": "b137140b3be7317036be924dcdbea7d2ce4031ae", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "extfieldsofstudy": [ "Biology" ] }
18663445	pes2o/s2orc	v3-fos-license	Quark and Lepton Masses from a {\boldmath $U(1)\times Z_2$} Flavor Symmetry We show that solutions for the masses and mixings of the quarks and leptons based on a $U(1)\times Z_2$ horizontal symmetry are possible. The seesaw mechanism is shown to work consistently in the presence of the discrete symmetry. The discrete symmetry results in the phenomenologically useful suppressions of elements of the Yukawa matrices. The quark and lepton masses, the CKM mixing angles, and the neutrino mixing angles are accomodated at the order of magnitude level. I. INTRODUCTION Experimental data on the quark and lepton sector masses and mixings may provide a clue to the nature of new physics beyond the Standard Model (SM). Masses and mixings are experimentally accessible, but as far as the SM is concerned, these parameters can be adjusted at will without destroying the consistency of the theory. Therefore any relationships between them must come from theoretical ideas beyond those already contained in the SM, and the experimental data can guide us in narrowing down the choices and freedom in these ideas. The recent data on the mixing of neutrinos has rekindled interest in models of fermion masses and mixings since it supplements the existing data from the quark and charged lepton sectors. The neutrino observations have some intriguing features that one might hope to explain. First of all, the neutrinos are very light in comparison to the other fermions. This suggests that a heavy mass scale may be involved that is providing a small dimensionless number that is responsible for the small neutrino masses. Secondly, the atmospheric neutrino data [1] indicates that there is a large mixing angle involved. This is in contrast to the small mixing (Cabibbo-Kobayashi-Maskawa of CKM) angles of the quark sector. Since in grand unified theories (GUTs) the quarks and leptons are unified in representations of the larger gauge theory, this dichotomy of small CKM angles with large mixing in the lepton sector provides hints as to how the fermion masses might arise. In fact the quark and charged lepton sectors show large hierarchies of masses. This seems to indicate that there might be a flavor symmetry whose spontaneous breaking might result in the generation of naturally small contributions resulting in the hierarchical pattern of masses. One hope is that such a flavor symmetry can be implemented to understand the masses and mixings detailed above as well as the long-standing evidence for solar neutrino oscillations. In this paper we show solutions to the quark and lepton masses and mixings based on a U(1) × Z 2 Abelian flavor symmetry are possible. A particular solution (with nontrivial Z 2 charges) we detail is entirely consistent with all the phenomenological constraints, and one can implement the seesaw mechanism to explain the light neutrino masses. The paper is organized as follows. In Section II we briefly review the approach of supersymmetric Abelian flavor (or horizontal) symmetries, and present the phenomenological requirements that must be satisfied in both the lepton and quark sectors. In Section III we discuss how a discrete symmetry can be used to overcome the constraints implied by a U(1) symmetry, and show how the discrete symmetry can suppress an entry to the extent that it has no impact on the leading order predictions for the masses and mixing angles. In Sections IV and V we review how the light neutrino mass matrix is independent of the horizontal charges of the singlet neutrinos for the case where the symmetry is U(1). In Section V we also generalize the derivation of the light neutrino mass matrix for the case where the horizontal symmetry is U(1) × Z 2 . In Section VI we present a solution for the quark sector that satisfies all the phenomenological requirements. Finally in Section VII we present our conclusions. II. FLAVOR SYMMETRIES The hierarchical structure of the fermion mass matrices strongly suggests that there is a spontaneously broken family symmetry responsible for the suppression of Yukawa couplings. In this paper we employ supersymmetric Abelian horizontal symmetries. These flavor symmetries allow the fermion mass and mixing hierarchies to be naturally generated from nonrenormalizable terms in the effective low-energy theory. The idea is quite simple and easily implemented [2]. There is some field Φ which is charged under a U(1) family symmetry, and without loss of generality, we can assume that its charge is -1. There are terms contributing to effective Yukawa couplings for the quarks, and the integer exponents m ij and n ij are easily calculated in terms of the horizontal symmetry charges of the quark and Higgs fields. For example, if we choose to have the Higgs fields to be uncharged, then the exponent m ij is just the sum of the horizontal charge of the fields Q i and d j . The hierarchy is generated from terms in the superpotential that carry integer charges m ij , n ij ≥ 0. If we call the small breaking parameter λ, then the generated terms for say the down quark Yukawa matrix will be of order λ m ij . The holomorphy of the superpotential forbids terms from arising with m ij , n ij < 0. A nice analysis of the possible approaches to explaining the neutrino masses and mixings using U(1) symmetries only is given in Ref. [3]. In the remainder of this section we outline the experimental constraints that a solution employing the above idea must satisfy. A. Phenomenological requirements for leptons The first phenomenological constraints we consider involve the charged leptons whose masses are the most precisely measured parameters of the quark-lepton sector. For the experimental values for the masses, we require that where the small parameter is identified as the Cabibbo angle, i.e. λ ∼ 0.22. The remaining constraints on leptons involve the neutrino masses and mixings. The most interesting aspect of the neutrino data is that the atmospheric neutrino mixing appears to be large, perhaps even maximal. It is then hard to understand a hierarchical pattern for the neutrino masses, since large mixing should result when the neutrino masses are of roughly the same order of magnitude. The Super-Kamiokande data [1] suggest that where the subscripts indicate the generations of neutrinos involved in the mixing (here we assume the mixing is between ν µ and ν τ ). The solar neutrino flux can be explained by one of three distinct solutions. Two of these involve matter-enhanced oscillation (MSW), while the third involves vacuum oscillations (VO). The two MSW solutions are differentiated by the size of the mixing angle, so one is usually called the small mixing angle (SMA) solution, and the other is called the large mixing angle (LMA) solution. The values required for the mixing parameters in each of these three cases are shown in the table below. For example, consider the small mixing angle (SMA) solution of the solar neutrino problem. Combining the solar neutrino data with the atmospheric neutrino data, one requires then the following when the small parameter is taken to be the Cabibbo angle. As explained by Grossman, Nir, Shadmi [4] and Tanimoto [5], one can accommodate the phenomenological constraints on the neutrino masses and mixings as well as the charged lepton masses by postulating that there is a U(1) × Z 2 horizontal symmetry. We show how such a solution can be obtained in the seesaw mechanism in Section V. B. Phenomenological requirements for quarks Again taking the expansion parameter to be the Cabibbo angle, λ = \|V us \|, then the experimental constraints [6] \|V us \| = 0.2196 ± 0.0023 , \|V cb \| = 0.0395 ± 0.0017 , on the CKM matrix can be identified in terms of powers of λ by the following The constraint on \|V ub /V cb \| can be expressed in a stronger way at 90% confidence level as 0.25λ − 0.5λ. One also has a constraint on the CKM elements from B 0 d − B 0 d mixing [6], which implies that It has been argued that \|V ub \| is more accurately given as ∼ λ 4 and taking it to be ∼ λ 3 (as we will do) requires an unnatural cancellation. However, in our opinion, requiring \|V ub \| ∼ λ 4 is too restrictive for two reasons. Firstly, the fine-tuning required is much less if one uses an expansion parameter λ somewhat less than 0.22, say 0.18. Secondly, since there are four parameters in the CKM matrix we are trying to predict, it is not unnatural that one of these would appear mildly fine-tuned, given λ ∼ 1/5. One can appreciate the nature of the cancellation in terms of the unitarity of the CKM matrix. This constraint requires so to leading order in λ one has the relation 1 Since \|V cd \| ∼ λ, \|V cb \| ∼ λ 2 and \|V td \| ∼ λ 3 , unitarity implies that to leading order one might expect \|V ub \| ∼ λ 3 whereas the experimental data yields a value somewhat suppressed 2 . One can show [7] that with U(1) or Z 2 as components of the horizontal symmetry, one can suppress \|V ub \| (or \|V td \|) relative to λ 3 only by even powers of λ, so \|V ub \| ∼ λ 4 is not possible 3 . The additional suppression one might desire to attribute to an additional power of λ must in fact be resulting from a mild cancellation. There is a universal scaling factor associated with the renormalization group evolution of the CKM angles \|V ub \| and \|V cb \| from the high (grand unified) scale to the electroweak scale [8], but this scaling is not enough to change the expectations for the relevant exponents of λ. The mass ratios should satisfy To compare the predictions of flavor symmetries to these phenomenological constraints, one has to relate the CKM elements to the entries in the Yukawa matrices. The Yukawa matrices U and D can be diagonalized by biunitary transformations The CKM matrix is then given by The left-handed transformation matrices V L u and V L d can be defined in terms of three successive rotations in the (2,3), (1,3) and (1,2) sectors. These rotation angles of the transformation matrices can be expressed in terms of the elements of the Yukawa matrices as follows [7,9] s u 12 = where u ij is the i, jth component of the up quark Yukawa matrix, U/(U) 33 , andũ 22 = u 22 u 33 − u 23 u 32 . There are corresponding expressions for the s d ij in terms of the components of the down quark Yukawa matrix, D (which are slightly more complicated due to the fact that the (2,3) sector mixing in V R d might be of order one). Clearly contributions to the CKM matrix elements can come from a number of terms. We shall be interested in what follows in determining the leading order contribution(s) to the CKM angles and the fermion masses. III. TEXTURE ZEROS The procedure of adopting a U(1) horizontal symmetry introduces nontrivial relationships between the parameters in the quark and lepton mass matrices. This results because previously undetermined entries in the matrices are described in terms of a few parameters. For example, the quark (up and down) mass matrices are described by nine parameters, namely the U(1) charges of the fields Q L , u R , and d R . Relationships between the masses and mixing angles are then obtained. The motivation for including texture zeros in mass matrices was to derive more relationships between the masses and mixings. The earliest of these was the relationship between the Cabibbo angle and the first and second generation down quark masses, V us ≃ m d /m s . The texture zeros responsible for this relation can be obtained in models where there is an additional discrete symmetry that forbids their occurrence, for example. Furthermore, these relationships might also include Clebsch-Gordon factors that allow one to obtain phenomenologically acceptable relationships: one of the earliest and most successful of these was the Georgi-Jarlskog model [10], which was shown later [11] to be successful in the case of electroweak scale supersymmetry (with the experimental data available at that time). The guiding principle for the case where the Yukawa matrices are symmetric is this: the mass hierarchy is of order λ 4 in the up quark matrices, and is of order λ 2 in the down quark matrices (c.f. Eq. (11)). So the dominant contribution to the Cabibbo matrix should come from the down quark matrices (If the down quark matrix is symmetric and the 1-1 component is suppressed, then one has the relation \|V us \| ≃ m d /m s ), while the dominant contribution to \|V cb \| ∼ λ 2 should come from the diagonalization of the up quark matrices. Furthermore the relation \|V ub /V cb \| ≃ m u /m c follow from suitable texture zero patterns [9]). When the theory is supersymmetric, one can obtain zero entries in the mass matrices that are sometimes called holomorphic or supersymmetric zeros. They arise because the superpotential must be holomorphic, so entries that would get a contribution from Φ † are absent. In terms of the mass matrices, this simply means that there are no entries with the small parameter λ raised to a negative power. Rather entries, for which the quantum numbers would seem to require a negative exponent, are simply zero. In this paper we want to introduce another concept that we will call a texture zero by flavor suppression. The horizontal symmetry we are considering here does not by itself give us any information on the order one coefficients of the entries in the mass matrices. When certain entries are suppressed because of the discrete symmetry, it can result that the entry is sufficiently suppressed that it does not affect the leading order of the phenomenology. Equivalently this entry could be replaced with an exact zero, and the leading order expectation for the masses and mixing angles would remain the same. Consider a simple 2 × 2 example of quark mass matrices where there is a horizontal U(1) symmetry, and where the phenomenological constraints (listed below) are motivated by the (2,3) sector of the quark mass matrices. We require that the mixing angle be ∼ λ 2 and the quark mass ratios satisfy m c /m t ∼ λ 4 and m s /m b ∼ λ 2 . This can be obtained by assuming the charges Q L : 2, 0, u R : 2, 0, d R : 0, 0: This is a unique solution of U(1) charges, and thereby determines already relationships between the mixings and masses of the first generation. The procedure for determining the exponents of λ in a model with a U(1) solution, clearly leads to the relations between exponents, for all i, j = 1, 2, 3. However there is something more that adding a discrete symmetry can do. Notice that in Eq. (18), the mixing V cb arises from contributions from diagonalizing U and from diagonalizing D, since both of these contributions are of order λ 2 . We can however find stronger relationships if we can suppress one of these contributions. For example, if the mixing contribution from the (2,3) block of the up quark matrix is suppressed and the (D) 22 entry is suppressed, then one has that the mixing angle is the same power of λ as the square root of the mass ratio \|V cb \| ∼ m c /m t . (If the up quark matrix is known to be exactly symmetric, then one even knows that the order one coefficient of the leading contributions in λ is the same, \|V cb \| ≃ m c /m t .) Just this kind of suppression of the element (U) 23 can be obtained by employing a discrete symmetry. So if the exponent of λ in (U) 23 is sufficiently large that it plays no role in determining the phenomenological predictions (to leading order), then we say it is a texture zero. Returning to our example, take the U(1) × Z 2 charges to be Q L : (3, 1), (0, 0), u R : (1, 1), (0, 0), d R : (1, 0), (0, 1). Then one obtains the Yukawa matrices for which the phenomenological predictions in terms of powers of λ are the same, but the mixing comes at leading order from diagonalizing D. We see that the relations, Eq. (19), need not necessarily hold if one adds a Z 2 symmetry. Since it is the off-diagonal entries that are suppressed in U, we can define the texture pattern in the following schematic way, where X denotes a phenomenologically relevant entry, and the 0 entries are suppressed sufficiently that the exponent that appears there is irrelevant. Even though the entries in the first column of the up and down quark matrices are not suppressed by the discrete symmetry, we denote these as zeros because they do not contribute at leading order to either the phenomenologically relevant (left-handed) mixing angles or the quark masses. We leave the categorization of what patterns of texture zeros one can employ in 3 × 3 matrices to a future work [12]. IV. NEUTRINO MASSES Assume that the lepton fields have charges under a U(1) family symmetry We assume here that the quantum numbers satisfy the hierarchies In the lepton sector, the effective Yukawa couplings come from nonrenormalizable terms, giving where M R is the relevant high mass scale at which the light neutrino masses are generated by the effective (nonrenormalizable) operator in the second term. There are two cases we can consider: (1) the mechanism that gives rise to the light neutrino masses generates all possible contributions to the nonrenormalizable terms ℓ Li ℓ Lj H u H u . In this case the exponent q ij that makes the second term an invariant under the horizontal symmetry (before symmetry breaking) is just q ij = L i + L j . So the light neutrino mass matrix is simply where v 2 is the electroweak scale vev of H u (and v 1 is the vev of H d ). (2) The horizontal symmetry can play a role in the generation of the second term in which case it is not necessarily the case that the exponent q ij is given by a naive inspection of the charges L i and L j . The seesaw mechanism for the generation of the light neutrino masses is such an example, and we explore it further in the next section. In particular we show that with the appropriate assignment of heavy neutrino charges, N i , one can enhance the (m ν ) 33 entry of the light neutrino mass matrix. V. NEUTRINO SEESAW The neutrino seesaw mechanism is a popular way to explain the lightness of the observed neutrino masses. It follows naturally from the group theory structure of the Standard Model (SM). We have left-handed neutrino doublets in the SM we can supplement by adding a right-handed neutrino singlet. In fact we have three generations, so the resulting masses will involve mass matrices. The neutrino doublets can pair up with the singlets to form a Dirac mass matrix, m D , coming from the breakdown of the electroweak symmetry. The neutrino singlets can pair up with themselves to form a 3 × 3 Majorana matrix, M N . This mass matrix is expected to be superheavy; it is not generated by the electroweak symmetry breaking. The group theory dictates that the neutrino doublets cannot pair up with each other. So the result is a 6 × 6 mass matrix of the form and upon diagonalization of the 3 × 3 light neutrino mass matrix is given by the seesaw formula In the rest of this section we will explore the implications of assuming the mass matrix entries are governed by an Abelian horizontal symmetry. A simple result emerges for the case where the symmetry is U(1), and some interesting enhancements (or suppressions) can occur when the symmetry is extended to U(1) × Z 2 which have some phenomenological usefulness. A. U (1) Given lepton doublet charges L i and right-handed neutrino charges N i one has the following pattern for the neutrino Dirac mass matrix and the following pattern for the Majorana mass matrix Then one obtains the same form for the light neutrino mass matrix via the see-saw formula m ν = m D (M N ) −1 m T D that was presented in the previous section in Eq. (24). It is easy to see that if one wants to have a hierarchy in light neutrino masses m νµ /m ντ ∼ λ 2 , and large mixing in the (2,3) generation in the lepton sector, one cannot rely on a U(1) symmetry alone. From Eq. (24) one sees that L 2 = L 3 + 1 is required to give the proper mass ratio. This then immediately prevents the large mixing from arising in the neutrino sector, because the mixing is necessarily of order λ. However we must still examine the diagonalization of the charged lepton matrix. Let the U(1) charges of the charged lepton singlets be E i , then the charged lepton matrix is The relevant mixing comes from the right hand side of this matrix, λ L 2 +E 3 /λ L 3 +E 3 . So the mixing parameter here is also order λ, since L 2 = L 3 + 1. Therefore the atmospheric neutrino mixing is necessarily of the order sin θ ν 23 ∼ λ in contradiction to the order one mixing observed (c.f. Eq. (3)). We consider next the changes that can occur when a discrete symmetry is utilized. This avenue has been exploited to understand the large mixing in the atmospheric neutrino oscillations together with a hierarchy in the neutrino masses, m νµ /m ντ << 1 [5,4]. It can also lead to an enhancement in the production of a matter/antimatter symmetry in the early universe [13], if the heavy neutrinos are assumed to be decaying asymmetrically. In the rest of this section we explain in detail how to implement the discrete symmetry with the seesaw mechanism so that the m νµ /m ντ ∼ λ 2 , and large mixing results. Take the following U(1) × Z 2 charges for the lepton fields . Assume the symmetry breaking is characterized by the single expansion parameter λ. The formulas given above for the heavy neutrino mass matrix, M N , the neutrino Dirac mass matrix, m D , and the resulting light neutrino mass matrix, m ν are modified. With the above assignments one finds that so that So the effect of the discrete symmetry in our case is to enhance the 3-3 entry of the M N matrix, and thereby alter the results for the third row and the third column on the inverse matrix, (M N ) −1 . With our charge assignments, one also has an enhanced entry in the 3-3 component of the neutrino Dirac mass matrix, It is easy to see then that the light neutrino mass matrix m ν is modified so that only the 3-3 entry is enhanced as desired, A phenomenologically viable solution has been presented in Ref. [4]: taking L 1 = 3, L 2 = L 3 = 1, E 1 = 5, E 2 = 4, and E 3 = 2 yields mass matrices of the form which give the correct orders of magnitude for the small mixing angle MSW solution and the correct orders of magnitude for the charged lepton mass ratios, Eq. (2). It also gives sin θ ν 13 ∼ λ 2 . In fact it has been shown [4] that one can obtain a VO solution as well by a proper quantum number assignment to the lepton fields. Without introducing more theoretical input (e.g. grand unified theories) to relate the quark and lepton charges, we cannot say more about which of the solutions is the correct one. We have shown here that the lepton sector phenomenology and the neutrino seesaw mechanism is consistent with a U(1) × Z 2 symmetry, and in fact a hierarchy in the neutrino parameters m νµ << m ντ requires the extra Z 2 . Furthermore we have shown in detail how to implement the neutrino mass enhancement in the seesaw mechanism. In the next section, we show that the quark sector phenomenological constraints also admit a solution consistent with Eqs. (6) and (11), and with a U(1) × Z 2 symmetry. Our solution to the quark Yukawa matrices is based upon the Elwood-Irges-Ramond (EIR) solution [14] obtained with a U(1) flavor symmetry. Here we show that one can impose a texture pattern by choosing quark fields to be charged under the new Z 2 symmetry. EIR obtained the Yukawa matrices The EIR solution was obtained by the U(1) charges (after adding appropriate overall constants to each field which don't affect the hierarchy pattern) The CKM elements can be expressed in terms of the Yukawa matrix elements, where it is understood that there are possible phases associated with each term on the right hand sides of the equations. From Eq. (40), one sees that \|V ub \| is receiving contributions in the EIR solution of order λ 3 from both u 13 and d 13 , as well as from the final term Then the experimental value for \|V ub \| must arise from a partial cancellation of these three contributions. Tanimoto showed that a solution involving a U(1) × Z 2 symmetry is not possible if the Cabibbo mixing, \|V us \|, arises from the diagonalization of the down quark Yukawa matrix [5]. The Cabibbo mixing in our scheme comes from diagonalizing the up quark matrix U. Our first attempt has the following assignments for the quantum numbers of the quark fields i = 1 2 3 Q L : (4, 0) (2, 1) (0, 1) u R : (4, 0) (1, 0) (0, 1) d R : (0, 0) (0, 1) (0, 1) , which is easily related to the EIR U(1) assignment above by replacing the Z 2 charge with +1 for Q L and with −1 for u R and d R . One can always substitute 0 ↔ 1 for Z 2 charges without affecting the results. We obtain the following Yukawa matrices for the up and down quarks The mass matrices can be obtained by multiplying these Yukawa matrices by the relevant vev (v 1 for D and v 2 for U). The intergenerational hierarchy, m b /m t ∼ λ 3 , is then accounted for either by tan β = v 2 /v 1 , and/or by increasing the U(1) charges for d R . It is straightforward to check that these matrices have the texture pattern We refer to this type of suppression as a 3-texture zero solution, since three entries are suppressed by the charge assignments in the discrete symmetry. Referring back to Eq. (40), one can see that the dominant contribution to \|V ub \| comes only from the third term and is of order λ 3 . This solution was motivated by the desire 4 to derive that \|V ub \| is proportional to \|V cb \|; this can be achieved if the first term in parentheses in Eq. (40) is suppressed, and this requires a texture zero in the (1,3) position of both U and D. Consequently one only has a contribution from the final term (u 13 ∼ λ 5 and d 13 ∼ λ 5 ). But then \|V td \| ∼ λ 5 is inconsistent with Eq. (8), and unitarity (Eq. (10)) requires the relation which is also not supported by the experimental data, Eq. (5). Clearly the 3-texture zero pattern in Eq. (44) is too restrictive. We can relax the problematic constraint, Eq. (45), by removing the texture zero in the (1,3) position of the up quark matrix. Consider the following texture zero pattern from i = 1 2 3 Q L : (4, 0) (2, 1) (0, 1) u R : (6, 1) (2, 0) (0, 0) d R : (0, 0) (0, 1) (0, 1) 4 Since the experimental data for \|V cb \| = aλ 2 already requires the order one coefficient to be less than one, a ≃ 0.6 [5], it is more likely that this final term (which includes another order one coefficient we will call b) will give agreement with the experimental data, \|V ub \| = abλ 3 with b somewhat smaller than one. Furthermore it is then consistent with the experimental data on \|V ub /V cb \| = 0.08 ± 0.02 = bλ for b ≃ 0.4. So the combination of order one coefficients satisfy the required relation, ab ≃ λ. As is clear from the texture pattern in Eq. (46), the Cabibbo angle is arising in the up quark matrix U. However one avoids the uncomfortable relation \|V us \| ∼ m u /m c because the matrix is not symmetric. All phenomenological constraints are satisfied by this solution with \|V ub \| receiving contributions of order λ 3 from only the u 13 term and the last term in Eq. (40). The matrices in Eq. (35) and (47) show that nontrivial Z 2 charges can be assigned to the quark and lepton fields, and that all phenomenological requirements can be met. VII. CONCLUSION We have shown that one can explain all the masses and mixings of the quarks and leptons with a U(1) × Z 2 symmetry. The phenomenological requirements can be met if the Cabibbo mixing in the two light generations is generated in the up quark mass matrix. This runs counter to the bias of assuming that the Cabibbo mixing is coming from the down quark mass matrix so that the relation \|V us \| ≃ m d /m s is obtained. This prejudice should be abandoned in the context of these Abelian flavor symmetries, because the resulting Yukawa matrices need not be symmetric. and m u /m c ∼ λ 4 is a reasonable hierarchy even with the leading contribution to the Cabibbo angle \|V us \| coming from the up quark matrix. The advantages of employing the U(1) × Z 2 as a flavor symmetry are the following: • One can understand a mass hierarchy m νµ /m ντ ∼ λ 2 and large atmospheric neutrino mixing sin θ ν 23 ∼ 1, without invoking unnatural cancellations. • The discrete symmetry can be implemented consistently with the neutrino seesaw mechanism to give the necessary neutrino mass hierarchy. • The source for CKM mixing angles in terms of the original parameters in the Yukawa matrices is reduced via the presence of texture zeros. For example \|V ub \| arises from a single contribution in Eq. (40), since the other contributions are suppressed by a flavor suppression. The EIR model has the Cabibbo angle, \|V us \| ∼ λ, arising at leading order from both the up and down quark matrices (c.f. Eqs. (38) and (37)). Our solution suppresses the contribution from the down quark Yukawa matrix, and thus the leading contribution arises entirely in the up quark matrix. • The discrete Z 2 symmetry can enhance the lepton asymmetry generated by the decay of heavy right-handed neutrinos [13]. This can be exploited to straightforwardly explain the baryon asymmetry of the Universe. A discrete flavor symmetry offers some attractive features for generating phenomenologically reasonable models. We leave to future work the question of whether these models can be embedded in a larger theoretical structure.	2014-10-01T00:00:00.000Z	2000-03-13T00:00:00.000	{ "year": 2000, "sha1": "41c0b40b93499c2a46bf5e2c717e2ce7880d1d53", "oa_license": null, "oa_url": "http://arxiv.org/pdf/hep-ph/0003121", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "41c0b40b93499c2a46bf5e2c717e2ce7880d1d53", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
57757953	pes2o/s2orc	v3-fos-license	Clinical Pharmacology of Citrus aurantium and Citrus sinensis for the Treatment of Anxiety Objective The aim of this review is to analyze preclinical and clinical studies investigating the anxiety effects of Citrus aurantium or Citrus sinensis essential oils (EOs). Design The bibliographic research was made on the major scientific databases. Analysis included only articles written in English and published on peer-reviewed scientific journals describing preclinical experiments and clinical trials carried out to investigate the antianxiety effects of Citrus aurantium or Citrus sinensis EOs on anxiety disorders. Clinical studies reporting the antianxiety effects of products containing Citrus aurantium or Citrus sinensis EOs in combination with other active substances, including medicinal plants, were excluded. Nine clinical studies fulfilled the criteria adopted for analysis. Results Data show that Citrus aurantium or Citrus sinensis EOs produce anxiolytic effects both in preclinical experiments and in different clinical conditions. Citrus aurantium EO aromatherapy reduced anxiety level in the great part of stress conditions studied (subjects affected by chronic myeloid leukemia and preoperative patients) except for a sample of patients subjected to colonoscopy. Exposition to Citrus sinensis EO in clinical studies shows to be positive in reducing anxiety level in patients waiting for dental treatment as well as in healthy volunteers submitted to an anxiogenic situation. Conclusions Overview of clinical trials conducted with Citrus aurantium or Citrus sinensis on people with anxiety showed that inhalation or oral administration of Citrus aurantium and inhalation of Citrus sinensis can exert beneficial effects on anxiety; however, because of incomplete accuracy in the reporting of methodology, further more complete clinical studies are warranted. Introduction Citrus plants derived from the single genus Citrus are largely interbreedable. Among the common names given to the various members of the citrus family, orange often refers to the most popular Citrus sinensis and Citrus aurantium [1]. Chemical composition of Citrus plants is characterized by the presence of several polyphenolic classes, including flavones, flavanones, flavonols, flavans, and anthocyanins [2,3]. In folk medicine, products derived from the peel and/or whole dried immature fruit of orange plants have been eaten fresh or drunk as juice. Juice can be consumed directly or further processed into concentrate, and both derivatives are used in soda and cocktail drinks, punches, and liqueurs. Orange fruits and peels are also used in desserts, jams and marmalades, and candied peels, as well as cookies, cakes, and candies. EO derived from orange peels, flowers, leaves, and twigs is used in perfumes; orange seed oil may also be used in cooking or as a component in plastic industry [30]. The sweet orange tree is found more or less in the same places as the bitter orange tree. Sweet orange oil is extracted from the fruit of the tree via cold pressure; it is also possible to distill sweet orange oil [31]. Citrus sinensis contains several active secondary metabolites contributing to the pharmacological activities of the plant. In Citrus sinensis fruits, peel, leaves, juice, and roots, several types of chemical compounds including flavonoids [2,32], hydroxyamides, steroids, alkanes and fatty acids, coumarins, carbohydrates, peptides, carbamates and alkylamines, carotenoids, volatile compounds, and minerals such as potassium, magnesium, calcium, and sodium have been identified [12,33]. C. sinensis is a rich source of vitamin C, a natural antioxidant that support the immune system activity [33,34]. C. sinensis has been used traditionally, to treat intestinal disorders (such as cramps, constipation, colic, and diarrhea), respiratory disorders (such as cough, cold, bronchitis, and tuberculosis), obesity, menstrual disorder, cardiovascular disease (angina, hypertension), anxiety, depression, and stress [35]. Anxiety disorders are among the leading prevalent causes of global mental disorders [36,37]. They contribute to favour poor compliance with therapy [38] and insufficient patient adoption of healthy behaviors [39]. On the basis of their negative effects on the results of therapy, it is necessary to find effective interventions. It is known that inhalation of volatile components of Citrus EO is able to influence the activity of brain areas such as the hypothalamus, hippocampus, and pyriform; preclinical and clinical research showed that citrus fragrance can restore stress-induced cortex [40,41] and immunosuppression [42] and may have potential antidepressant effects in rats [43,44]. In light of the abovementioned findings, we tried to assess if the potential health effects of both Citrus aurantium and Citrus sinensis are really effective in the treatment of anxiety conditions. With this aim, we summarized the published reports of preclinical and clinical studies regarding the use of Citrus aurantium-or Citrus sinensis-based products in conditions related to anxiety disorders. Research Method and Inclusion Criteria of Clinical Trials. A bibliographic research was carried out independently by two researchers (blinded to the authors and initially on results) in the major scientific databases and search engines of peer-reviewed literature from 2000 to July 2018, on life sciences and biomedical topics (PubMed, Scopus, Embase, Web of Science, and Google Scholar). The following Evidence-Based Complementary and Alternative Medicine 3 Numbers of records identified through database searching n = 94 Number of additional records identified through other sources n = 0 Records a er duplicates removed n = 84 Records excluded for title and abstract n = 63 Records excluded a er full review n = 4 (association products) Full text articles assessed for eligibility n = 21 Articles included in the study n = 17 Pre-clinical studies n = 8 Clinical studies n = 9 keywords or combination of keywords "Citrus anxiety", "Citrus aurantium axiety", "Citrus sinensis anxiety", and "clinical trials" were used. [46] and the Jadad Scale [47]. From the eligible articles, two investigators independently extracted data by using a standard data extraction form. Data were considered for therapeutic indication, design of the study, number, sex, and age of subjects, endpoints, adverse effects, and outcome. In all the studies, reporting of adverse reactions was monitored. The absence of adverse reactions was defined as "not reported." All the authors reviewed all the eligible articles and resolved by discussion any uncertainty regarding the statistical method used to handle the missing data. Results A collection of 94 scientific articles was selected from our bibliography research. Only 17 articles describing effects of Citrus aurantium or Citrus sinensis treatment for anxiety were corresponding to the inclusion criteria. 63 articles were excluded for title and abstract and 4 were excluded because the products object of the studies were combinations including Citrus aurantium. Nine clinical studies were included in the review (Figure 1). In eight clinical studies, Citrus aurantium or Citrus sinensis were administered for inhalation as aromatherapy, and in one study Citrus aurantium was orally administered. Table 1 summarizes for each preclinical study the authors, the route of administration, animal species, experimental model to study anxiety, dose, and observed effects of Citrus aurantium or Citrus sinensis administration. Tables 2 and 3 summarize authorship of the paper, therapeutic indication, study design, subjects involved, endpoints, adverse effects, and outcome of all the clinical trials regarding use of Citrus aurantium or Citrus sinensis, respectively. Tables 4 and 5 report CONSORT items for trials with herbal medicine interventions applied to clinical studies, for Citrus aurantium and Citrus sinensis, respectively. Table 6 reports quality assessment of randomized controlled trials by the Jadad scoring system. Rotarod test Citrus aurantium essential oil 0.5 or 1.0 g/kg in a volume of 10 ml/kg. In light-dark box test, single treatment with essential oil (0.5 or 1.0 g/kg) increased the time spent by mice in the light chamber and the number of transitions between the two compartments. Single and repeated treatments with essential oil (0.5 or 1.0 g/kg) were able to suppress marble-burying behavior. No impairment on rotarod procedure after both single and repeated treatments with essential oil was observed, denoting absence of motor deficit. Evidence-Based Complementary and Alternative Medicine 5 Participants were randomly divided into three groups. Group 1 received 10 mg diazepam as oral dose; Group 2 received C. aurantium essential oil (EO) 10 mL diffused in the room through an electric dispenser. Group 3 (placebo) was exposed to the vaporization of saline solution. In the last two groups, the exposure lasted 30min. Patients of the fragrance condition were exposed to 5 drops (0.25 mL) of C. aurantium essential oil poured in 5 L of water and diffused using an electrical dispenser. Patients in the control condition were exposed in the same environment to diffusion of water without fragrance. All the patients (control and experimental) waited about 10 min in the waiting room. The dental anxiety scale (DAS) questionnaire was used to determine the anxiety level of the patients prior to surgery Mean blood pressure, respiratory rate, and pulse rate were also evaluated. Not reported Orange fragrance is effective in reducing anxiety linked to surgical removal of impacted mandibular third molar. Mean blood pressure, respiratory rate, and pulse rate, during surgery, were significantly reduced. 2 ml of orange essence was placed in a dispenser activated for 2 min every 10 min. Anxiety of children was assessed with salivary cortisol level and pulse rate before and on completion of each of two dental treatment appointments. Not reported Statistically significant reduction of salivary cortisol level and pulse rate in aromatherapy group compared to control group. Not reported Individuals exposed to the test aroma (2.5 and 10 drops) presented a lack of significant alterations (p> 0.05) in state-anxiety, subjective tension and tranquillity levels throughout the anxiogenic situation, revealing a dose-dependent anxiolytic activity of sweet orange essential oil. [48]. The acute administration induced an anxiolytic-like effect in the light/dark transition tests (increased time spent in the light side, and in the number of transitions) and in marble burying (decreased number of marbles buried) without any motor impairment, while repeated administration showed effects in the marble-burying test only. Repeated diazepam administrations did not increase light/dark transitions. Thus, the results suggest an anxiolytic-like effect following acute and repeated Citrus aurantium EO administration [49]. Similar results were obtained in another experiment showing that anxiolytic-like effect of oral administration of Citrus aurantium EO was reversed by 5-HT1A antagonist WAY100635 but not by flumazenil, a benzodiazepine antagonist, suggesting serotonergic mediation [44]. In another study, inhalation of Citrus aurantium EO increased social interactions (time spent in active social interaction) in rats and increased exploration time in the open arms of the elevated plus-maze, suggesting an anxiolytic-like effect at a dose that did not impair motor activity in the openfield test [50]. Acute intraperitoneal administration of Citrus aurantium EO, similar to fluoxetine, increased open arm explorations (percentage of time spent and percentage of entries) in the elevated plus-maze in male mice. The authors suggested that the effect of Citrus aurantium L. is linked to serotonergic transmission based on a fluoxetine + Citrus aurantium EO interaction. However, Citrus aurantium EO did not change the anxiolytic effect of fluoxetine in the elevated plus-maze, suggesting no drug interaction [51]. Khosrovi et al. investigated the effect of intraperitoneal injection of Citrus aurantium EO on anxiety and its interaction with GABAergic pathways, evaluating the coadministration effects of Citrus aurantium and diazepam. Evidence-Based Complementary and Alternative Medicine 13 The JADAD scoring system was used for the assessment of randomized controlled trials with the following 5 items: Was the study described as randomized? ( Flumazenil did not alter the effect of Citrus aurantium; however, authors suggest that Citrus aurantium L. may exert an anxiolytic-like effect acting as partial agonist at the GABA-A receptor/benzodiazepine site [52]. Preclinical antianxiety effects of Citrus aurantium are summarized in Table 1. Complex six studies investigating Citrus aurantium EO effects on anxiety levels during different medical conditions were found. All of the six studies in which Citrus aurantium was investigated were randomized/controlled clinical trials. One randomized clinical trial described the effects of Citrus aurantium EO in reducing anxiety during labor in a group of Iranian pregnant women [53]. Before the aromatherapy, both groups had the same levels of anxiety; the levels of anxiety evaluated at dilations of 3-4 and 6-8 cm were significantly lower in the aromatherapy group with Citrus aurantium compared with the control group, thus suggesting that aromatherapy with Citrus aurantium EO could reduce anxiety during labor [53]. A trial was carried out on patients proposed for colonoscopy and divided into two groups. Aromatherapy was performed by inhalation of Sunflower oil (control group) and Neroli oil (experimental group). Results showed that there was no significance difference of procedural anxiety measured by State Trait Anxiety Inventory state (STAI-S) score and procedural pain evaluated by visual analogue scale (VAS), before and after aromatherapy in patients subjected to colonoscopy. Significant lower pre-and postprocedural systolic blood pressure in Neroli group than control group were observed. The authors concluded that the inhalation aromatic agent had an effect on lowering procedural anxiety-related (excessive fear of medical, dental, or surgical procedures that results in acute distress or interference) blood pressure [54]. Patients with a mandibular third molar with B II classification of impacted teeth and American Society of Anaesthesiologists (ASA) class I patients (healthy subjects with no organic pathology), with moderate and high anxiety levels measured through the dental anxiety scale (DAS) questionnaire, were included in another randomized and controlled study. The ASA clinical status classification system assessed the fitness of patients before surgery. The outcome variables were physiologic measures related to anxiety, including mean blood pressure, respiratory rate, and pulse rate. After aromatherapy, mean blood pressure, pulse rate, and respiratory rate were significantly lower in the fragrance group during surgery (from the time of sitting in the dental chair to the end of surgery). Vital signs measurements showed significant differences between the two groups, which may be due to the sedative effect of the orange fragrance during surgery. Results suggest that ambient orange fragrance could be helpful in reducing dental anxiety during dental surgical removal [55]. Another study was carried out on a Brazilian volunteers' cohort with chronic myeloid leukemia (CML), treated with standard therapy. The study compared Citrus aurantium EO with diazepam anxiolytic effects, in the moment that precedes the collection of medullary material in patients with CML. Systolic and diastolic blood pressure and cardiac and respiratory frequencies were measured. C. aurantium EO decreased both systolic and diastolic blood pressure while with diazepam, only systolic blood pressure decreased. The Citrus aurantium EO did not decrease the respiratory frequencies. The use of STAI-S revealed an anxiolytic effect of Citrus aurantium EO group of patients with CML but not in the CML diazepam and placebo groups. In conclusion, study results showed an anxiolytic effect of Citrus aurantium EO in patients with CML with an improvement of psychological and physiologic parameters. For the authors, this effect has a great clinical relevance, because patients with cancer go through various stressful phases during the disease, and the standard therapy significantly contributes to improve the anxiety level and physiological parameters during a procedure that is a cause of distress [13]. Chaves Neto et al. (2017) studied the anxiolytic effects of Citrus aurantium EO in patients experiencing crack withdrawal. Based on the fact that individuals who experience crack withdrawal present a high anxiety trait, anxiety status was induced with the Simulated Public Speaking (SPS) method (subject is requested to deliver a speech in front of a video camera with its image being displayed on a TV screen). Anxiety levels were assessed by the Inventário de Ansiedade Traço-Estado (IDATE), the Brazilian version of STAI-S. The results demonstrated that subjects in the groups treated with Citrus aurantium EO maintained controlled anxiety levels during SPS when compared to the control group (no treatment). Subjects who were administered Citrus aurantium EO also maintained states of discomfort and cognitive impairment often associated with anxiety during SPS. However, nebulization of the EO of Citrus aurantium provided an acute anxiolytic effect in crack cocaine users exposed to SPS. Authors' conclusions were that Citrus aurantium EO, administered by nebulization, produced anxiolytic effects in crack users in abstinence thus indicating the possibility of Citrus aurantium EO use as an alternative complementary therapy in the control of anxiety in users who are abandoning drugs abuse [56]. In a randomized double-blind design, the effect of oral administration of Citrus aurantium blossom distillate (CABd) on preoperative anxiety was evaluated. Preoperative anxiety was assessed using STAIS and Amsterdam Preoperative Anxiety and Information Scale (APAIS). The main finding of this study was the confirmation of the anxiolytic effect obtained with oral administration of blossom distillate Citrus aurantium. Both STAI-state and APAIS scores were decreased by CABd. On the other hand, neither STAI-state nor APAIS was changed in the placebo group [57]. Results of the studies using Citrus aurantium EO showed that inhalation of the oil produces significant anxiolytic effects. Methods for diffusion of EO used in the studies are by direct inhalation with hand-hold nebulizers generally with doses of EO ranging 10 -50 mL or by dilution of EO in distilled water and successive diffusion through an electric dispenser or by gauzes impregnated with 4 mL of Citrus aurantium EO. Citrus sinensis (Sweet Orange) Pre-Clinical and Clinical Anti-Anxiety Effects. In comparison to findings published about investigation on Citrus aurantium, a reduced number of preclinical and clinical experiments were conducted on Citrus sinensis. Inhalation of Citrus sinensis EO (sweet orange, containing 97% limonene) in rats submitted to elevated plus-maze followed by the light/dark paradigm produced an increase in open arm exploration (% time spent and % number of entries) in the elevated plus-maze, and of the time spent in the lit chamber of the light/dark test. No effect on motor activity, as measured by the total distance travelled in the elevated plus-maze, was detected [58]. Anxiolytic-like, sedative, and antidepressant-like potential effects of inhalation of both Citrus aurantium and Citrus sinensis EOs were evaluated through behavioral tests and measurement of corticosterone and melatonin plasma levels in mice. Results of behavioural tests showed an anxiolytic-like and sedative effect of Citrus sinensis EO 10% inhalation, without affecting melatonin and corticosterone physiological levels. Inhalation of 10% Citrus aurantium EO did not show neither anxiolytic-like effects nor change in melatonin and corticosterone plasma levels [59]. Three randomized and controlled studies investigating on the effects of Citrus sinensis were found. Two studies describe the effects of Citrus sinensis EO on anxiety in adults and children, respectively, subjected to dental treatments. A randomized controlled trial was carried out on adult patients waiting for dental treatment to evaluate the anxiolytic effects of Citrus sinensis. Orange odor was diffused in the waiting room of odor group through an electrical dispenser whereas in the control group no odor was diffused. For assessing cognitive functions, patients completed the Wortschatz test (WST) [60]. For assessing trait and state anxiety, patients were given the State-Trait Anxiety Inventory (STAI) [61]. The Mehrdimensionale Befindlichkeitsfragebogen (MDBF) [62] were used for assessment of current mood, alertness, and calmness. Results of the study showed that exposure to ambient orange odor has a relaxant effect compared to control group. Results reported also a sex difference of Citrus sinensis effects, showing that women exposed to sweet orange EO had a lower level of state anxiety, a more positive mood, and a higher level of calmness [63]. The effects of orange odor (Citrus sinensis) on child anxiety during dental treatment were also evaluated in another study by using salivary cortisol and pulse rate as index of patients' anxiety. The results of this study showed that the salivary cortisol level and pulse rate significantly decreased in intervention groups by using aromatherapy with Citrus sinensis EO [64]. The potential anxiolytic effect of Citrus sinensis EO was evaluated in a double-blind, randomized, placebo-controlled clinical trials carried out on healthy volunteers submitted to an anxiogenic situation. Immediately after inhalation, each subject was submitted to the video-monitored version of the Stroop Color-Word Test (SCWT), a method to induce anxiety. During the text, a board with 100 of the color-naming words blue, yellow, red, green, and violet organized randomly in a 10 ⋅ 10 matrix is presented to each participant. The color of each word is different from its own meaning (for example, the word "green" printed in red). The participant has to say quickly (in two minutes) the sequence presented, the color of the ink, but not the colors described by the single words. The whole test is recorded and presented to the subject on a monitor during the test [65]. State-Trait Anxiety Inventory (STAI) and Visual Analogue Mood Scale (VAMS) were used to evaluate psychologic parameters. Physiologic parameters were evaluated before the inhalation period and before, during, and after the SCWT. Results showed dose-dependent anxiolytic properties of sweet orange EO [65]. These three studies indicate that Citrus sinensis EO exerts anxiolytic effects in anxiogenic situations. One of the studies show that Citrus sinensis EO exerts its anxiolytic effects also in children. Discussion EOs physiological and psychological effects are known in folk medicine and aromatherapy for a long time [14,66]. Aromatherapists have proposed that Citrus fragrances have mood-enhancing properties. These effects were confirmed by successful clinical study carried out with citrus fruits oils on patients affected by stress symptoms or depression [67,68]. Anxiety disorders are the most common type of psychiatric disorders in the general population [69]. Their treatment is difficult because of the important side effects of the drugs used to improve anxiety symptoms, which generally promote dependence [70]. Moreover, common treatments do not benefit all patients and only few of them have a slight resumption [71]. These conditions justify the increasing interest and search for alternative or complementary procedures aimed at improving anxiety symptoms. One of these is aromatherapy, a procedure that uses EOs by inhalation as a treatment for medical purposes [72]. Citrus aurantium and Citrus sinensis are rich in flavonoids and polyphenolic compounds with numerous pharmacological properties, such as the inhibition of the oxidation of lowmolecular weight proteins and platelet accumulation thus contributing to immune cell stability. They are also used in treatment of mental disorders, inflammation, viral infections, and allergies [73,74]. Flavonoids effects in reducing anxiety are due to their action as benzodiazepine receptor agonists [57]. A series of review articles regarding the safety and efficacy of bitter orange has been published [75][76][77][78]. However, this is the first analysis of clinical studies using Citrus aurantium or Citrus sinensis for the treatment of anxiety. We analyzed all clinical studies according to the recommendations described in the checklist developed by the Consolidated Standards of Reporting Trials [46] for the reporting of clinical trials using herbal medicinal products (Tables 4 and 5). The clinical trials included in this review have different levels of methodological accuracy. All of studies were randomized and controlled. Only one study regarding Citrus aurantium use was carried out with a double-blind methodology ( Table 2). Two of the studies regarding Citrus sinensis use were carried out with a double-blind methodology, and only one was performed as a crossover study (Table 3). Since it is a common source of selection bias, a point of weakness in all the studies is the lack of description of the methods adopted to generate random allocation sequences [78] (Table 6). Moreover, all the studies were poor in number of subjects recruited. In four of them, the sample of people recruited did not exceed 50 patients, in four studies, the samples exceeded 50 participants, and only in one study the sample was more than 100 patients (Tables 2 and 3). All the studies reported statistical data analysis but no one reported sample size calculation. Latin binomial name of the plant was always correctly reported, although not all articles provided an exhaustive description of the characteristics of the product. Procedure adopted to obtain the EO and description of the raw material used to produce the herbal preparations were reported only in one study (Tables 4 and 5). The highest standards in methodology for clinical trials in herbal medicine strongly recommend the reporting of the characteristics of the product to facilitate the reproducibility of the studies. The presence of these data are fundamental to establish a link between the putative efficacy and safety to the single product [46]. Only few studies report qualitative testing producing the chemical fingerprint of the herbal products (Table 4). Finally, one study [13] compares the effects of Citrus aurantium against diazepam, a well-established drug prescribed for the treatment of anxiety (Table 2). Conclusions Antianxiety effects of Citrus aurantium and Citrus sinensis EOs were previously demonstrated through behavioral experiments carried out on laboratory animals. Complex clinical studies considered for this overview suggest that Citrus aurantium or Citrus sinensis EOs, used for anxiolytic therapy of people prevalently in conditions in which stress is dominating, produce positive effects against anxiety. In particular, Citrus aurantium EO aromatherapy reduced anxiety level in the great part of stress conditions studied (subjects affected by chronic myeloid leukemia and preoperative patients) except for a sample of patients subjected to colonoscopy. Exposition to Citrus sinensis EO in clinical studies is shown to be positive in reducing anxiety level in patients waiting for dental treatment as well as in healthy volunteers submitted to an anxiogenic situation. However, a definitive conclusion needs further studies because of the reduced number of studies and the small number of patients for each study. Regarding data reporting of clinical studies with Citrus aurantium, they appear without description of the characteristics of the herbal product and lack qualitative testing. From the methodological point of view, most of the clinical studies show the lack of important items such as treatment allocation conceal and description of double-blind procedure. On the basis of the present overview, we can conclude that the use of Citrus aurantium and Citrus sinensis EOs could be useful to reduce anxiety caused by clinical stress conditions; however, results are impaired by the poor accuracy and methodology applied in clinical studies. EO: Essential oil STAI-S: State Trait Anxiety Inventory score VAS: Visual	2019-01-22T22:29:33.570Z	2018-12-02T00:00:00.000	{ "year": 2018, "sha1": "073f5d47da8b5a46189da4e69b9811eeb8377662", "oa_license": "CCBY", "oa_url": "https://downloads.hindawi.com/journals/ecam/2018/3624094.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "073f5d47da8b5a46189da4e69b9811eeb8377662", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
259260021	pes2o/s2orc	v3-fos-license	Constructing Randomly Lamellar HKUST–1@Clinoptilolite through Polyethylene Glycol—Assisted Hydrothermal Method and Coordinated Complexation for Enhanced Adsorptive Separation for CO2 and CH4 Clinoptilolite (CP) was successfully synthesized via a hydrothermal route in the presence of polyethylene glycol (PEG), and it was then delaminated by washing using Zn2+ containing acid. HKUST-1, as one kind of the Cu-based MOFs, showed a high CO2 adsorption capacity owing to its large pore volume and specific surface area. In the present work, we selected one of the most efficient ways for preparing the HKUST-1@CP compounds via coordination between exchanged Cu2+ and ligand (trimesic acid). Their structural and textural properties were characterized by XRD, SAXS, N2 sorption isotherms, SEM, and TG-DSC profiles. Particularly, the effect of the additive PEG (average molecular weight of 600) on the induction (nucleation) periods and growth behaviors were detailed and investigated in the hydrothermal crystallization procedures of synthetic CPs. The corresponding activation energies of induction (En) and growth (Eg) periods during crystallization intervals were calculated. Meanwhile, the pore size of the inter-particles of HKUST-1@CP was 14.16 nm, and the BET specific area and pore volume were 55.2 m2/g and 0.20 cm3/g, respectively. Their CO2 and CH4 adsorption capacities and selectivity were preliminarily explored, showing 0.93 mmol/g for HKUST-1@CP at 298 K with the highest selective factor of 5.87 for CO2/CH4, and the dynamic separation performance was evaluated in column breakthrough experiments. These results suggested an efficient way of preparing zeolites and MOFs composites that is conducive to being a promising adsorbent for applications in gas separation. Introduction Clinoptilolite (CP), as one of the heulandite (HEU) zeolites [1], is widely used for gas adsorption, industrial catalysis, and environmental treatment [2][3][4][5][6] due to its adjustable acidity, thermal stability, and unique micropore networks. It consists of channel A (0.44 × 0.72 nm) with 10-member rings, and channel B (0.40 × 0.55 nm) with 8-member rings that is parallel to channel A, as well as C (0.41 × 0.47 nm) with 10-member rings that intersect with both channel A and B. CP is widely applied in CO 2 and CH 4 separation because of its low cost, high productivity, large adsorption, and easy purification [7]. The synthesized CPs with the HEU structures presented the layered accumulations containing the ordered and expandable lamellar structures [8], which are conducive to not only improving the mass transport efficiency for the guest molecules but also for providing the more accessible surfaces adsorption sites for functional assembly. However, the steric hindrances of their HEU structures, caused by small interlayer spaces, strongly limit their wider applications. In this regard, we proposed a synthesis method of adding a variety of surfactants to the synthesis system for facilitating the realization of layered structure expansions and the lamellar disorder arrangements of the synthesized CPs [9,10]. As reported by Iwakai et al. [11], the silicate-1 with smaller particle size (about 40~60 nm) could be synthesized by the addition of polyoxyethylene polyether in the conventional hydrothermal system, while the particle size of conventional silicate-1 was larger than 500 nm. Silva et al. [12] provided a synthesis method of zeolite Y by adding surfactant cetyltrimethylammonium bromide in the hydrothermal system. Compared with commercialized USY zeolite, their products showed the same structure characteristics with relative lower crystallinity (about 33% of USY zeolite) but a much higher external surface area (661 cm 3 /g) than that of USY zeolite (208 cm 3 /g). Koohsaryan and Anbia proposed an efficient way of synthesizing zeolite 13X via additive polyethylene glycol (PEG), which was beneficial to synthesize the well-developed zeolite crystals that have a mesopore size (around 2-10 nm) with a shorter growth period and a relative greater crystallinity [13]. Therefore, we believe that the polymer polyols are beneficial to the layered expansion of synthetic CPs, owing to their good water solubility and greater amount of hydrogen bonds in aqueous solution. The traditional method of zeolite delamination is usually carried out under ultrasonic conditions. Corma et al. [14] reported an exfoliated method for layered ITQ-2 by cetyltrimethylammonium bromide and tetrapropyl ammonium hydroxide under ultrasonic condition. After exfoliation, the sheets of the zeolite became a random arrangement with an external surface area of about 700 m 2 /g, almost twice as much as the raw ITQ-2. However, the high energy cost of the ultrasonic process of zeolites is one shortcoming. Gorgojo et al. [15] achieved a direct exfoliation method by cetyltrimethylammonium ion without ultrasonic treatment, and the exfoliated Nu-6 presented the larger surface areas of around 300 m 2 /g, six times that of raw Nu-6. As demonstrated by Okrut et al. [16], the layered silicate magadiite was delaminated via Zn 2+ modification and acid treatment without ultrasound under mild conditions, resulting in that its external surface area enlarging from 10 to 25 m 2 /g. The delamination process usually follows the synthesis process, so combining two processes may improve the delamination performance. Roth et al. [17] proposed a delamination method by calcinations based on the as-synthesized MCM-22 with additive hexamethylenediamine, showing larger c-axis distance from around 0.25 nm to larger than 0.50 nm. However, rare reports proposed delamination methods regarding the abovementioned CP. Hopefully, the delamination of the synthetic CPs can not only open the lamellar structures from one direction of the crystal planes but also improve the adsorption capacity and the catalytic performance of the gases. As is well known, the metal-organic frameworks (MOFs), as kinds of multi-dimensional structure materials with high specific surface areas and abundant functional sites [18], have recently presented strong potential applications in adsorption, separation, and catalysis. HKUST-1, consisting of two channels, namely, a 0.9 nm cage and a 0.35 nm pore [19], exhibited a high CO 2 adsorption capacity (up to 4.20 mmol/g at 27 • C and 1 bar) owing to its large pore volume (0.69 cm 3 /g) and specific surface area (1615 m 2 /g) [20], but large-scale production and industrial application are difficult to achieve due to its low productivity, poor hydrothermal stability, and higher cost. In order to avoid these disadvantages, at least to some extent, one of the most efficient methods is the preparation of the composites of zeolites and MOFs. The composites are usually prepared by such a method that the MOFs are synthesized from the precursor solution with the addition of certain amounts of the synthesized zeolites. For example, the core-shell structured Zeolite-5A@MOF-74 was obtained from the MOF-74(Ni) precursor solution with the addition of Zeolite-5A, showing higher thermal stability [21]. The CO 2 /H 2 , CO/H 2 , CH 4 /H 2 , and N 2 /H 2 selectivity was 8659, 2375, 114, and 45, respectively, at 25 • C and 1 bar, higher than that of Zeolite-5A or MOF-74. Using a similar method, MIL-100 (Fe)@SBA-15 was successfully synthesized by Mahmoudia et al. [22], and the adsorption efficiency for methylene blue, rhodamine B, methyl orange, and alizarin yellow R dye solutions was higher than that of pure MIL-100 (Fe), due to the generations of the pore structures in the MIL-100 (Fe). As demonstrated by Tari et al. [23], the MCM-41/Cu(BDC) was prepared via the same method mentioned above, showing a higher CO 2 /CH 4 selectivity (the separation factor of around 4.9 at 303 K and 4 bar) as compared with pure MCM-41. In addition, transition metal cations can be introduced onto CP through ion exchange, which can not only balance the anions of the aluminosilicate frameworks but also provide the central ions for preparing MOFs. However, the composites between CP and MOFs are rarely reported. More recently, we proposed a novel constructed strategy of the disorderly layered UiO-66-on-clinoptilolite heterostructures through the assistance of PEG and polyvinylpyrrolidone, and the prepared hybrid materials exhibited an excellent adsorption selective performance for CO 2 and CH 4 [9]. Herein, we developed a preparation method of the composites of CP and HKUST-1 via delamination, ion exchange, and coordination. On the basis of the abovementioned PEG-assisted hydrothermal method [9] and using alkali, aluminum, and silicon sources as raw materials, the synthesized CP was delaminated in the Zn-containing HNO 3 solutions, and then Cu 2+ -exchanged after ammonization. Finally, the obtained Cu-CP was coordinated with trimesic acid ligand to form HKUST-1@CP composite without additional metal sources. Meanwhile, X-ray diffraction (XRD), Fourier transform infrared (FT-IR) spectra, thermogravimetric (TG) analysis, N 2 sorption isotherms, and scanning electron microscopic (SEM) images were used to elucidate the structure features and physicochemical performances of the resultant HKUST-1@CP composites. Meanwhile, the application of the composites in the CO 2 /CH 4 adsorptive separation and the column breakthrough performance were preliminarily explored. The novelty and contribution of the present work are the successful preparation of HKUST-1@CP composites by stratification, ion exchange, and coordination, especially the addition of PEG in the hydrothermal synthesis system being conducive to the layered expansion of CPs. The other contribution is that the exchanged Cu 2+ located in the exchange position of CP can be used not only as cations to balance the negative charge of CP skeleton but, also, as central ions to complex with the ligands. As an effective adsorbent, hopefully, the CO 2 /CH 4 adsorptive separation and the column breakthrough performance could be enhanced. Synthesis of CP Firstly, NaOH, KOH Al(OH) 3 and deionized water were mixed and transferred into a Teflon-lined stainless steel autoclave, and then stirred at 150 • C for 3 h to get an alumina sol. Subsequently, silica sol, deionized water, 3 wt% of seed, and 7 wt% of PEG were mixed and added into the alumina sol. The molar ratios were as follows: 1.38 Na 2 O: 1.38 K 2 O: 11.18 SiO 2 : Al 2 O 3 : 294 H 2 O. The mixture solution was stirred at room temperature for 2 h and continually transferred into a Teflon-lined stainless steel autoclave at 150 • C for 72 h to crystallize. Afterwards, the products were washed with enormous amounts of deionized water and dried at 120 • C overnight, and named CP-X, where X is the crystallization time (h). In comparison, pure CP was synthesized without any additive PEG. Delamination, Ammonization and Ion Exchange of CP-X Firstly, 0.5 g of the synthesized CP-X was added into 30 mL of 0.5 mol/L Zn(NO 3 ) 2 solution, and after stirring for 2 h at room temperature, 15 mL of 2 mol/L HNO 3 solution was added and then stirred for 1 h continually. After being washed with deionized water and dried at 120 • C for 12 h, the obtained products were delaminated, and named CP-X-E. Next, the delaminated CP-X-E was added into 1 mol/L NH 4 Cl solution and stirred at 80 • C for 2 h, following: 1 g CP-X-E: 100 mL NH 4 Cl solution. After being washed with deionized water and then dried at 120 • C overnight, this procedure was repeated 10 times, and the final product was named Ammonized CP-X-E. The ammonized CP-X-E was then added into CuCl 2 solution with a concentration of 0.0005, 0.003, and 0.06 mol/L, and then stirred at 80 • C for 2 h, following: 1 g ammonized CP-X-E: 200 mL CuCl 2 solution. After being washed with deionized water and dried at 120 • C overnight, this procedure was repeated 1-10 times, and the final product was named Cu-CP-X-E-Y, where Y represents the Cu content (wt%) in the samples. For example, when the X (the crystalline time) equals 72 h, the Y equals 7.3, and the corresponding sample was named Cu-CP-72-E-7.3. Various Cu-CPs and their copper contents are summarized in Table S1 of the Electronic Supplementary Information (ESI) section. Synthesis of HKUST-1 HKUST-1 was synthesized on the basis of the reported procedure [24] as follows: 1.539 g CuCl 2 ·2H 2 O and 1.26 g trimesic acid were dissolved into 75 mL deionized water and 75 mL ethanol for stirring at 50 • C, and, after that, 2.5 mL triethylamine was added into the above solution and stirred for 3 h continually. The powder was cooled down and washed with ethanol to remove unreacted reactant, and it then was centrifuged and dried at 120 • C overnight. Synthesis of HKUST-1@CP Composites A total of 0.3 g Cu-CP was added into 15 mL deionized water and sonicated for 30 min. Meanwhile, the desired amount (according to the Cu contents in Cu-CP at the same molar ratio of synthesis of HKUST-1) of trimesic acid was dissolved into 15 mL ethanol and continually stirred at 50 • C for 30 min. The desired amount of triethylamine was then added into the solution and stirred at 50 • C for 3 h. Finally, the HKUST-1@CP composites were obtained after cooling down, washing, centrifuging, and drying at 120 • C overnight, and named HKUST-1@CP-X-E-Y, correspondingly. The amounts of the used ligands and the triethylamine were summarized in Table S2, and the schematic diagram for CP preparation was shown in Figure S1 of the ESI section. Characterizations XRD patterns were recorded using the Beijing Purknje General Instrument Corporation XD-6 X-Ray diffractometer with Cu Kα for 4 • ·min −1 in 2 θ of 5-50 • at 36 kV and 20 mA. According to the relative value of the sum of ten diffractive peak intensities in the XRD patterns of the synthetic CPs, namely, indexed as (020), (200), (111), (13-1), (131), , , (350), (530), and (061), the crystallinity of the CP synthesized with the crystallized time of 72 h at 150 • C without any additive PEG and normalized as 100%. The relative crystallinity of the other samples could then be calculated on the basis of Equation [25], as follows (1): where ∑ 10 i=1 I i is the sum of the intensities of the ten peaks of the XRD pattern of the samples synthesized with different crystallization time and ∑ 10 j=1 I j which refers to the sum of the intensities of ten peaks of the XRD pattern of the CP. I is the intensity of each diffractive Nanomaterials 2023, 13, 1860 5 of 21 peak, and i or j is a number from 1 to 10, corresponding to diffractive peaks, indexed as (020), (200), (111), (13-1), (131), , , (350), (530), and (061), respectively. The metal ion contents were determined using a Hitachi ZA3300 Polarized Atomic Absorption Spectrometer (AAS). The samples were dissolved in HNO 3 solution as follows: 0.02 g sample was dissolved in 0.5 mL 23 mol/L HF solution, and then diluted to 10 mL with HNO 3 (Volume fraction 2%). Subsequently, 1 mL of the above solution was diluted to 10 mL with HNO 3 (Volume fraction 2%). The morphologies were obtained using SEM (JEOL JEM-220) images with the microscopes at 15.0 and 200 kV and the energy dispersive X-ray spectroscopy (EDS). The IR-Prestige-21 FT-IR spectrum was used to measure the functional groups of the obtained samples in the wavenumber range of 400-4000 cm −1 . The TG profiles were received using Perkin-Elmer Pyris Instruments TG-DSC thermal analyzer at a heating rate of 10 • C·min −1 and a N 2 flow rate of 20 mL·min −1 . The N 2 adsorption-desorption isotherms at 77 K as well as the CO 2 and CH 4 adsorption at 273 K and 298 K were both collected using a Beijing JWGB JW-BK300 gas adsorption instrument. Each sample was outgassed at 120 • C for 6 h before adsorption. The Small Angle X-ray Scattering (SAXS) patterns were performed at the 1W2A station of the Beijing Synchrotron Radiation Facility with the wavelength of the X-ray source of 0.154 nm. The distance from sample to detector was 1600 mm, which was the calibrated reference of the diffraction ring of a standard sample. The scattering vector magnitude was around 0.09 to 3.04 nm −1 , and the detector readout noise (dark current) of the Mar165 CCD measured with a mask before the sample measurements was approximately 10 counts per second. The sample was sealed with Scotch tape into a sample cell with 1 mm thickness. The scattering images were collected through the single-frame mode with a "multi-read" of 2 times, and the exposure time was 5 min. The two-dimensional SAXS images were transformed to the onedimensional data by the Fit2D software (http://www.esrf.eu/computing/scientific/FIT2D, accessed on 25 January 2016) and the S program package [26]. Breakthrough Experiments The breakthrough experiments were tested by a BSD-MAB Multi-Component Adsorption Breakthrough Curve Analyzer containing a cylindrical quartz column, a mass flow controller, pressure-control valves, and a mass spectrograph that was used to evaluate the dynamic separation performance of HKUST-1@CPs. The two adsorbing gases had a ratio of 50/50 vol%, and they were introduced into the controlled environment with the total flow rate of 5 mL/min. The desorption gas flow was detected with a BSD-MAB multi-component mass spectrograph. The breakthrough column with the bore size of 4.0 mm containing the adsorbate sample of about 0.1 g was installed inside the ceramic oven, which was located inside the convection oven. The pressure drops across the column and in the outlet of the column were both monitored. The gas flow sent to the analysis section could be adjusted with the valve. The sample was outgassed at 393 K for 2 h under a nitrogen atmosphere with the flow rate of 20 mL/min, and it was then cooled down before the breakthrough experiment. Figure 1 shows the XRD patterns of the HKUST-1@CPs, and the synthetic CPs and Cu-CPs are shown in Figure S2. As can be seen in Figure S2(Ab,c), both the CP-0 and the CP-24 presented diffractive peaks that were not obvious when the crystallization time was less than 24 h, while Figure S2(Ad) shows that CP-72 exhibited the same intensively diffractive peaks as the conventional CP [27], namely, indexed as (020), (200), (111), (13-1), (131), , , (350), (530), and (061), which is indicative of the HEU structures [28]. However, its characteristic peaks ( Figure S2(Ad)) not only slightly decreased in intensity but also shifted to a more or less high region in the 2 theta position, as compared with that of the pure CP ( Figure S2(Aa)). In contrast, the weakened intensity (200) of the CP-72-E (as shown in Figure S2(Ae)) implied decreases in its long-range order, possibly due to the delamination of the HEU lamellas [29]. Meanwhile, the intensity of other diffractive peaks before and after delaminated CP-72 can be well maintained. As a result, CP lamellas would be disordered in a specific crystal plane after delamination, which would lead to the achievement of delamination. In other words, delamination obviously reduced the long-range order of the CP, and, thus, more surface cations were exposed [29], which may facilitate the formation of MOF with more ligands through coordination in the following section. less than 24 h, while Figure S2(Ad) shows that CP-72 exhibited the same intensively diffractive peaks as the conventional CP [27], namely, indexed as (020), (200), (111), (13-1), (131), , , (350), (530), and (061), which is indicative of the HEU structures [28]. However, its characteristic peaks ( Figure S2(Ad)) not only slightly decreased in intensity but also shifted to a more or less high region in the 2 theta position, as compared with that of the pure CP ( Figure S2(Aa)). In contrast, the weakened intensity (200) of the CP-72-E (as shown in Figure S2(Ae)) implied decreases in its long-range order, possibly due to the delamination of the HEU lamellas [29]. Meanwhile, the intensity of other diffractive peaks before and after delaminated CP-72 can be well maintained. As a result, CP lamellas would be disordered in a specific crystal plane after delamination, which would lead to the achievement of delamination. In other words, delamination obviously reduced the long-range order of the CP, and, thus, more surface cations were exposed [29], which may facilitate the formation of MOF with more ligands through coordination in the following section. Meanwhile, the role of Zn 2+ in the delamination process was similar to the principle of the reference [16], namely, the formation of ZnOx(OH)y in the interlayer of CP. Briefly, a consequence of a single Zn 2+ inserted interlayer and complete exchange with two Na + or K + . A significant amount of Zn 2+ remained in CP-72-E (0.96 wt%), which is a result of Zn 2+ intercalation as well as nucleation and growth of small Zn(O)x(OH)y colloids in between the layer spaces. Beyond that, the (200) diffractive peak located at around 11.17 o in the XRD pattern revealed the presence of a HEU structure. After the delamination in the Zncontaining HNO3 solutions, the (200) peak position was not shifted and its corresponding d value was 0.79140 nm, indicating the Zn-containing HNO3 modification had no significant impact on (200) interplanar spacing. Structure Characterizations As can be seen in Figure S2B, the not obvious change of the (200) peak position after ammonization and the subsequent Cu 2+ exchange indicated that the ion exchange process had no significant influence on (200) interplanar spacing. The presence of the extra diffractive peaks at 2 theta of 16.10 and 15.40 suggested that the appearance of the atacamite and botallackite in Cu-CP-0-E-10.5 and Cu-CP-24-E-10.1 ( Figure S2(Ba,b)) were hardly removed via washing and drying [30,31]. Table S1 shows that the Cu content was around 1.64 mmol/g for Cu-CP-0-E-10.5 and 1.58 mmol/g for Cu-CP-24-E-10.1, which is more than the exchangeable cations (equal to total of Na content and K content), meaning that the Cu cations were partially located at the skeleton outside the CP with the low crystallinity. Regarding the Cu-CPs with different Cu contents ( Figure S2(Bc,d)), the position of these diffractive peaks was not a significant change, indicating that the HEU structures of the CPs were well maintained after several ion exchanges. Compared with that of the Cu-CPs with low crystallinity ( Figure S2(Ba,b)), the extra diffractive peaks of atacamite and Meanwhile, the role of Zn 2+ in the delamination process was similar to the principle of the reference [16], namely, the formation of ZnO x (OH) y in the interlayer of CP. Briefly, a consequence of a single Zn 2+ inserted interlayer and complete exchange with two Na + or K + . A significant amount of Zn 2+ remained in CP-72-E (0.96 wt%), which is a result of Zn 2+ intercalation as well as nucleation and growth of small Zn(O) x (OH) y colloids in between the layer spaces. Beyond that, the (200) diffractive peak located at around 11.17 o in the XRD pattern revealed the presence of a HEU structure. After the delamination in the Zncontaining HNO 3 solutions, the (200) peak position was not shifted and its corresponding d value was 0.79140 nm, indicating the Zn-containing HNO 3 modification had no significant impact on (200) interplanar spacing. As can be seen in Figure S2B, the not obvious change of the (200) peak position after ammonization and the subsequent Cu 2+ exchange indicated that the ion exchange process had no significant influence on (200) interplanar spacing. The presence of the extra diffractive peaks at 2 theta of 16.10 and 15.40 suggested that the appearance of the atacamite and botallackite in Cu-CP-0-E-10.5 and Cu-CP-24-E-10.1 ( Figure S2(Ba,b)) were hardly removed via washing and drying [30,31]. Table S1 shows that the Cu content was around 1.64 mmol/g for Cu-CP-0-E-10.5 and 1.58 mmol/g for Cu-CP-24-E-10.1, which is more than the exchangeable cations (equal to total of Na content and K content), meaning that the Cu cations were partially located at the skeleton outside the CP with the low crystallinity. Regarding the Cu-CPs with different Cu contents ( Figure S2(Bc,d)), the position of these diffractive peaks was not a significant change, indicating that the HEU structures of the CPs were well maintained after several ion exchanges. Compared with that of the Cu-CPs with low crystallinity ( Figure S2(Ba,b)), the extra diffractive peaks of atacamite and botallackite were hardly observed in Figure S2(Bc,d). The largest Cu content was around 1.14 mmol/g for Cu-CP-72-E-7.3, which was less than that of the exchangeable cations. In addition, the crystallinity decreased with an increase in Cu contents (shown in Table S1), showing 93.3% for Cu-CP-72-E-3.4 and 87.8% for Cu-CP-72-E-7.3. The XRD patterns of the HKUST-1@CPs with different crystallinity of the CP and different Cu contents are shown in Figure 1. As can be seen in Figure 1a,b, the XRD pattern of HKUST-1@CPs exhibited the structural features of HKUST-1 and the CPs, such as (222) for HKUST-1 and (131) for CP, in which the (222) crystal plane of HKUST-1 (as shown in Figure S2(Af)) nearly overlapped with the (200) of CP [32]. Beyond that, the disappearances of the diffractive peaks of atacamite and botallackite occurred in HKUST-1@CP-0-E-10.5 and HKUST-1@CP-24-E-10.1 (Figure 1a,b), compared with the corresponding Cu-CPs ( Figure S2(Ba,b)). These observations suggested the formation of partial HKUST-1 mixture in the CP with low crystallinity. The diffractive peaks of HKUST-1, meanwhile, were relatively weak, and they cannot even be observed in HKUST-1@CP-72-E, which may be due to its low intensity. Herein, the formula was defined as follows: where I represents the intensity of the diffractive peak, and I r is the specific value of the intensity of (200) and (131), which denotes the relative intensity of (200). As a result, the I r value of CP-72 was 40.0%, lower than that (50.8%) of the pure CP synthesized without additive PEG, implying that the PEG effect was somewhat beneficial to reducing the peak (200) intensity of CP. The interplanar spacing (d) value of (200) crystal plane of CP-72 was 0.7934 nm, as well, which was larger than that of the pure CP (0.7921 nm), further verifying that the PEG effect during the synthesis process of CP was conducive to enlarge the interplanar spacing of CP. We noted that the Zn content determined by AAS was nearly zero for Cu-CP-72-E-7.3, suggesting that the generation of Zn-MOFs was well avoided on HKUST-1@CPs. The interlayer Zn 2+ were fully exchanged after ammonization and Cu 2+ exchange, owing to cationic sites of Zn 2+ exposed to the outside layers. The abovementioned samples were further characterized via the SAXS patterns, and their scattering curves are shown in Figures 2 and S3. As can be seen, the linearity of the ln[I(q)] versus ln(q) scattering profile revealed the fractal property of the synthesized CPs [33], whereas the linear range was determined in the range of −1.90 < ln(q) < 0.00, and their corresponding slopes were between −4 and −3. 1.14 mmol/g for Cu-CP-72-E-7.3, which was less than that of the exchangeable cations. In addition, the crystallinity decreased with an increase in Cu contents (shown in Table S1), showing 93.3% for Cu-CP-72-E-3.4 and 87.8% for Cu-CP-72-E-7.3. The XRD patterns of the HKUST-1@CPs with different crystallinity of the CP and different Cu contents are shown in Figure 1. As can be seen in Figure 1a,b, the XRD pattern of HKUST-1@CPs exhibited the structural features of HKUST-1 and the CPs, such as (222) for HKUST-1 and (131) for CP, in which the (222) crystal plane of HKUST-1 (as shown in Figure S2(Af)) nearly overlapped with the (200) of CP [32]. Beyond that, the disappearances of the diffractive peaks of atacamite and botallackite occurred in HKUST-1@CP-0-E-10.5 and HKUST-1@CP-24-E-10.1 (Figure 1a,b), compared with the corresponding Cu-CPs ( Figure S2(Ba,b)). These observations suggested the formation of partial HKUST-1 mixture in the CP with low crystallinity. The diffractive peaks of HKUST-1, meanwhile, were relatively weak, and they cannot even be observed in HKUST-1@CP-72-E, which may be due to its low intensity. Herein, the formula was defined as follows: where I represents the intensity of the diffractive peak, and Ir is the specific value of the intensity of (200) and (131), which denotes the relative intensity of (200). As a result, the Ir value of CP-72 was 40.0%, lower than that (50.8%) of the pure CP synthesized without additive PEG, implying that the PEG effect was somewhat beneficial to reducing the peak (200) intensity of CP. The interplanar spacing (d) value of (200) crystal plane of CP-72 was 0.7934 nm, as well, which was larger than that of the pure CP (0.7921 nm), further verifying that the PEG effect during the synthesis process of CP was conducive to enlarge the interplanar spacing of CP. We noted that the Zn content determined by AAS was nearly zero for Cu-CP-72-E-7.3, suggesting that the generation of Zn-MOFs was well avoided on HKUST-1@CPs. The interlayer Zn 2+ were fully exchanged after ammonization and Cu 2+ exchange, owing to cationic sites of Zn 2+ exposed to the outside layers. The abovementioned samples were further characterized via the SAXS patterns, and their scattering curves are shown in Figures 2 and S3. As can be seen, the linearity of the ln[I(q)] versus ln(q) scattering profile revealed the fractal property of the synthesized CPs [33], whereas the linear range was determined in the range of −1.90 < ln(q) < 0.00, and their corresponding slopes were between −4 and −3. [34], showing an increasing D s value from 2.14 for CP-72 ( Figure S3d) to 2.33 for CP-72-E ( Figure S3e) before and after delamination, but larger than that of pure CP (2.06, as shown in Figure S2a). These results implied that the surfaces of the synthetic CP-72-E became rougher with a higher surface activity [35]. Moreover, an increasing Ds Nanomaterials 2023, 13, 1860 8 of 21 value appeared from 2.07 to 2.33 with an increase in the crystallization time from 0 to 24 h (shown in Figure S3b,c), suggesting the transformation from a smooth surface to a rough and open structure for the resultant CP [10]. After that, the Ds value decreased to 2.14 when the crystallization time was prolonged to 72 h ( Figure S3d). Similar results were demonstrated by Zhao et al. [36], indicating that the fractal structures were strongly related to the surface property of aluminosilicate species, which supplied active precursors or colloidal nuclei for promoting crystal growth of the CPs. Radlinski et al. [37] demonstrated that the scattering occurred predominantly between the structure of the sheets and the interlayered matters, while the scattering-derived porosity confirmed a minor contribution from micropores. As can be seen, the Ds value increased from 2.07 for CP-0 ( Figure S3b Figure S4A, the PDDF curves of various CPs lacked the perfect symmetry, particularly that of CP-72 ( Figure S4(Ad)), indicating that their particles probably had flake-like morphologies, which may be related to the thinning of the lamellar thickness [38,39]. However, the PDDF curves of the CPs with a short crystallization time of less than 24 h (as shown in Figure S4(Ab,c) had relatively poor symmetry compared with the pure CP ( Figure S4(Aa)) and CP-72 ( Figure S4 Accordingly, Figures 2 and S3 indicate that all of the samples possessed surface fractal (Ds) features [34], showing an increasing Ds value from 2.14 for CP-72 ( Figure S3d) to 2.33 for CP-72-E ( Figure S3e) before and after delamination, but larger than that of pure CP (2.06, as shown in Figure S2a). These results implied that the surfaces of the synthetic CP-72-E became rougher with a higher surface activity [35]. Moreover, an increasing Ds value appeared from 2.07 to 2.33 with an increase in the crystallization time from 0 to 24 h (shown in Figure S3b,c), suggesting the transformation from a smooth surface to a rough and open structure for the resultant CP [10]. After that, the Ds value decreased to 2.14 when the crystallization time was prolonged to 72 h ( Figure S3d). Similar results were demonstrated by Zhao et al. [36], indicating that the fractal structures were strongly related to the surface property of aluminosilicate species, which supplied active precursors or colloidal nuclei for promoting crystal growth of the CPs. Radlinski et al. [37] demonstrated that the scattering occurred predominantly between the structure of the sheets and the interlayered matters, while the scattering-derived porosity confirmed a minor contribution from micropores. As can be seen, the Ds value increased from 2.07 for CP-0 ( Figure S3b) to 2.10 for HKUST-1@CP-0-E-10.5 (Figure 2a), and from 2.33 for CP-24 ( Figure S3c) to 2.40 for HKUST-1@CP-24-E-10.1 (Figure 2b). Meanwhile, the scattering profiles of Cu-CPs with different Cu contents are shown in Figure S3f and S3g. As can be seen, the Ds value increased with the increase in the Cu contents, such as 2.38 for Cu-CP-72-E-3.4 ( Figure S3f) and 2.44 for Cu-CP-72-E-7.3 ( Figure S3g). Obviously, these results suggested the successful Cu 2+ -incorporation on CP and further coordination with the ligands. The variation trend of the Ds value of HKUST-1@CPs with different Cu contents was similar to that of the Cu-CPs (as shown in Figure S3f The PDDF curves of all related samples are shown in Figure 3 and Figure S4. As shown in Figure S4A, the PDDF curves of various CPs lacked the perfect symmetry, particularly that of CP-72 ( Figure S4(Ad)), indicating that their particles probably had flakelike morphologies, which may be related to the thinning of the lamellar thickness [38,39]. However, the PDDF curves of the CPs with a short crystallization time of less than 24 h (as shown in Figure S4(Ab,c) had relatively poor symmetry compared with the pure CP ( Figure S4(Aa)) and CP-72 ( Figure S4(Ad)), possibly due to the formation of the flake-like particles. The PDDF curves of HKUST-1@CPs with a different crystallization degree of CP and different Cu contents are shown in Figure 3. As can be seen, the symmetry of PDDF The PDDF curves of HKUST-1@CPs with a different crystallization degree of CP and different Cu contents are shown in Figure 3. As can be seen, the symmetry of PDDF profiles of the HKUST-1@CPs were almost the same as that of the synthetic CP (as shown in Figure S4(Ab,c,e)), suggesting that the morphologies of CP were not significantly varied during the ammonization Cu 2+ exchange process and even during the formation of HKUST-1. The PDDF curves of Cu-CPs with different Cu contents, such as Cu-CP-72-E-3.4, and Cu-CP-72-E-7.3, as shown in Figure S4B, show almost the same symmetries of each sample with the same Cu content, namely, HKUST-1@CP-72-E-3.4 and HKUST-1@CP-72-E-7.3, as shown in Figure 3. These results indicate that the ion exchange and the coordinated process Nanomaterials 2023, 13, 1860 9 of 21 had a not obvious effect on the morphologies of the lamination but that they remarkably influenced the layered stacking mode. The FT-IR spectra of all related samples in the scanned regions between 400 and 4000 cm −1 are shown in Figure S5. As can be seen in Figure S5c,d, the bands appeared at 1638, 1205, 1062, 793, 605, and 465 cm −1 , and were attributed to the features of the synthetic CPs as follows: the band at 1638 cm −1 was due to deformation vibrations of H 2 O molecules [40]. The peaks around 1062 cm −1 with a shoulder at 1205 cm −1 were assigned to the asymmetric internal T-O-T (T = Si or Al) stretching vibrations of the tetrahedral [41]. The band around 465 cm −1 was associated with internal T-O bending vibrations of the tetrahedral, while the band located at 605 cm −1 was associated with an external tetrahedral double ring [42]. Meanwhile, the peak that appeared at 793 cm −1 represented the stretching vibration modes of the O-T-O groups [41], suggesting the presence of the quartz impurity. Figure S5a,b showed that no absorption peaks at 1205, 1062 and 605 cm −1 were observed within the crystallization time of 0 and 24 h, indicating the absence of the tetrahedral structures in a short crystallization time. The FT-IR spectra of CP-72-E ( Figure S5d) was the same as that of CP-72 ( Figure S5c), meaning that the skeletal structure of the CP was well maintained after delamination and even by acid washing. As can be seen in Figure S5g, the HKUST-1 had obvious adsorption peaks in the ranges of 1700-1500 cm -1 and 1500-1300 cm −1 , which are attributable to the asymmetric and symmetric stretching modes of the carboxylate functional groups, respectively. In particular, the band at 1451 cm −1 was related to the stretching and deformation modes of the benzene ring, while the C-H bending mode of the benzene ring appeared at around 729 cm −1 [43]. The HKUST-1@CP-72-E-7.3 ( Figure S5f) had the additional weak signals at around 1451 and 729 cm −1 , belonging to Cu-O stretching vibration, in which the oxygen atom of the ligand was coordinated with the Cu 2+ [44]. As shown in Figure S5e, the Cu-CP-72-E-7.3 had no adsorption peaks around 1500 cm −1 , and other signals of the CP were almost the same. However, the band at around 1400 cm −1 indicated the unexchanged NH 4 + of the CP [45]. A weak adsorption peak at around 1400 cm −1 of Cu-CP-72-E-7.3 ( Figure S5e) still appeared, as compared with that the CP-72 ( Figure S5c). These observations implied that a fraction of NH 4 + can hardly be exchanged, although the Cu 2+ exchange reached the maximum exchange capacity. On the basis of the abovementioned descriptions, the HKUST-1@CP composites were successfully synthesized. The N 2 adsorption-desorption isotherms of all the related samples are shown in Figure S6. As can be seen, the isotherms of all the CPs exhibited the characteristic type-I curves with an H3-type hysteresis loop at 0.80 < P/P 0 < 0.98, indicating the appearances of the silt-shaped mesopores, probably originating from the lamellar accumulation of the CPs [46,47]. Meanwhile, the specific surface area of CP-72 ( Figure S6(Ad)) was around 33.9 m 2 /g, lower than that (42.0 m 2 /g) of pure CP ( Figure S6(Aa)), but almost same as that (35.9 m 2 /g) of CP-72-E ( Figure S6(Ae)) because the introduction of the additive PEG in the synthesis system was beneficial to expanding the lamellar structures of the synthesized CP, while the effect of the delamination process was unobvious. As can be seen in Figure S6(Af,g), all of the Cu-CPs presented almost same the isotherm profiles as CP-72-E (shown in Figure S6(Ae)). The HKUST-1 (shown in Figure S6(Ba)) showed the adsorption capacity plateau close to zero, indicating the non-porous nature of the sample. The adsorption equilibrium was reached at low pressure, and the hysteresis loop did not occur at high pressure [48], indicating type I isotherm. Meanwhile, the HKUST-1 presented a higher surface area (516.1 m 2 /g) and total pore volume (0.28 cm 3 /g) [49]. Comparably, the surface area of the HKUST-1@CP-72-E-7.3 was about 55.2 m 2 /g, higher than that (39.5 m 2 /g) of Cu-CP-72-E-7.3 due to the incorporation of HKUST-1 onto CP [50,51]. However, Table S3 shows that the surface area of HKUST-1@CP-72-E-3.4 was lower than that of Cu-CP-72-E-3.4, and the possible reason for this is the formation of the sub-units of HKUST-1 as well as the blockage of the micropores in the CP. Crystallization Kinetics of the Synthesized CP with the Additive PEG In order to explore the role of the additive PEG in the synthesis of the CPs in detail, the effects of the crystallization temperature and the time on the crystalline phase are further elucidated via crystallization kinetics. On the basis of the crystallinity of the CP synthesized with and without additive PEG at different temperatures, the crystallization kinetics performances of the related samples synthesized at 140, 150, and 160 • C are shown in Figure S7. As can be seen, there were three distinct regions in the crystallization kinetics curves, namely: (a) an induction period including the nucleation phase, (b) a growth period involving the rapid growth of crystallites, and (c) a stable period comprising the deceleration of the growth process, similar to the reported demonstration [52]. Herein, the induction time (t 0 ) is defined as the crystallization time that has elapsed in order to achieve the crystallization degree of about 15% [53], as summarized in Table S4. It should be noted that a crystallinity of 15% was taken into account when determining the induction period [54]. Accordingly, the t 0 value of the synthesized CP with additive PEG gradually decreased with the enhanced temperature from 50 h at 140 • C ( Figure S7(Ba)) to 24 h at 150 • C ( Figure S7(Bb)), and 12 h at 160 • C ( Figure S7(Bc)). Similar phenomena were also observed in the crystallization kinetics of CP synthesized without PEG ( Figure S7(Aa-c)). Based on the Arrhenius equation, the activation energy (E) during the induction and growth processes was calculated to clarify the crystallization mechanism. The E values and frequency factor (lnA n ) for the induction stage (E n ) values are determined from the nucleation rate (1/t 0 ) and calculated using Equation (3): where R is the ideal gas constant, T is the absolute temperature (K), and A n is the frequency factor for the induced stage [55]. The growth period (E g ) values are also obtained via Arrhenius Equation (4): where A g is the frequency factor for the growth stage and the growth rate (k max ) is taken as the steepest slope of crystallization curves. As shown in Table S4, the E n values of the synthetic CPs were all larger than their E g values, suggesting that the nucleation process was a controlled step during the hydrothermal crystallization procedure. Meanwhile, the E n value of CP-72 synthesized with additive PEG was 106.2 kJ/mol, higher than that (63.0 kJ/mol) of pure CP obtained without PEG, and we can conclude that there is a restraining effect of the PEG on the formation of the crystal nucleus, possibly due to the PEG encapsulation for silicate and aluminate species hindering their polycondensations. Meanwhile, the E g value was 31.3 kJ/mol for CP-72, less than that for pure CP (59.8 kJ/mol), indicating the promoting effect of the PEG on the growth procedure of the CP. Morphologies and Thermal Analysis The morphologies of the related samples are shown in Figure 4. As can be seen in Figure 4b, the dispersive CP precursor solution exhibited the colloidal particles with irregular shapes that had a size of around 70 nm can be assigned to the nanoaluminosilicate species [56]. Subsequently, the particle sizes of CP-24 were increased to 3 µm, besides the appearances of the almost flake-like morphology (Figure 4c). When the crystallization time was further prolonged to 72 h, the random and lamellar accumulations of CP-72 were present with a flower-like morphology in sizes of around 5 µm (Figure 4d), which is a noticeable difference from that of pure CP (Figure 4a). growth procedure of the CP. Morphologies and Thermal Analysis The morphologies of the related samples are shown in Figure 4. As can be seen in Figure 4b, the dispersive CP precursor solution exhibited the colloidal particles with irregular shapes that had a size of around 70 nm can be assigned to the nanoaluminosilicate species [56]. Subsequently, the particle sizes of CP-24 were increased to 3 µm, besides the appearances of the almost flake-like morphology (Figure 4c). When the crystallization time was further prolonged to 72 h, the random and lamellar accumulations of CP-72 were present with a flower-like morphology in sizes of around 5 µm (Figure 4d), which is a noticeable difference from that of pure CP (Figure 4a). CP-72 (Figure 4d), meanwhile, had an irregularly lamellar structure in length of 1.5 µm and width of 1.1 µm, which was smaller than that (length of 3.0 µm and width of 2.0 µm) of pure CP (Figure 4a). These results suggested that the PEG effect was useful to shrinking the lamellar size of the CP. Additionally, the lamellar size of CP-72-E (Figure 4e) was 1.2 µm in length and 0.8 µm in width, smaller than that of CP-72 (Figure 4d). The possible reason for this is that the acidity of the Zn-containing HNO 3 solution caused the dissolutions of the lamellas of the synthetic CPs. Additionally, Figure 4j shows the agglomerated particles of HKUST-1 of a size around 0.5 µm, similar to the result reported by Majano et al. [57]. As can be seen in Figure 4f,g, HKUST-1@CPs presented an obvious HKUST-1 and the amorphous particles, which again confirms demonstrations of XRD patterns (as shown in Figure 1a (Figure 4j). Therefore, we speculated that HKUST-1 may grow mainly along the highly dispersed lamellar surfaces of the CP, which depend on the Cu 2+ location distributed in CP. As described in the Experimental section, after each ion exchange process between the CPs and Cu 2+ was finished, an enormous amount of deionized water was used to wash the obtained solids so as to remove the excess Cu 2+ cations adsorbed on the CP surfaces. However, the adsorbed Cu 2+ ions on the low crystallinity CP were still not completely removed, leading to the generations of HKUST-1 deriving from coordination between the ligands and part of Cu 2+ ions in HKUST-1@CPs (as shown in peak (222) of Figure S8(Aa,b)), besides the appearances of atacamite and botallackite in the Cu 2+ -exchanged CPs (as shown in peak (011) of Figure S8(Ba) and peak (100) of Figure S8(Bb)) [30,31]. However, HKUST-1 in the HKUST-1@CPs with a high crystallization degree was difficult to observe in the XRD patterns and SEM measurements, and one of the main reasons was the fact that the exchanged Cu 2+ ions, as the central ion, were located in the equilibrium cation position of the CP, and, therefore, the particle size of the synthesized HKUST-1 was very small. A similar report was made by Wu et al. [58], who synthesized the MOFs@zeolites (NaY and HZSM-5) via coordination between the exchanged Zn 2+ and the ligands (2-methylimidazole and its derivatives), though the sub-units of the MOFs distributed in Zn-exchanged Zeolites were difficult to observe via XRD patterns, SEM, and TEM measurements. As described in the Introduction, the delamination process usually follows the synthesis process, which improves the delamination performance. In the present work, the layered structure expansions and the lamellar disorder arrangements of the synthesized CPs were firstly prepared via adding PEG in the hydrothermal synthesis system, and then the delamination process was performed via Zn-containing acid washing under mild conditions. As can be seen in Figure S9, the weight loss profiles could be divided into three stages: the first one occurred when the temperature was up to 300-500 K, which was attributed to desorption of the physisorbed water, and the second happened during the temperature range of from 500-800 K, which was assigned to the chemisorbed water and possible decompositions of the ligands of HKUST-1 [59]. The weight loss of pure CP ( Figure S9a Figure S9d) at 300-500 K was 9.1, 8.1, 8.8, and 7.7%, respectively. Those at 500-800 K were 0.6, 0.8, 2.0, and 2.3%, respectively. The last one of nearly 1% at 800-1000 K was due to dihydroxylation and decomposition of the residual ligands of HKUST-1 [60]. These observations suggested that the impacts of the Cu 2+ exchange and post-treatment (delamination and acid washing) on the thermal stability of the CP were not obvious [61]. The TG curve of HKUST-1, meanwhile, as shown in Figure S9f (insert), showed that the first weight-loss stage appeared between 300 and 400 K was 21.8%, due to desorption of the physisorbed water. The second one between 400 and 550 K belonged to the dehydration of hydrated copper cations and the physical desorption of the solvent [60]. The weight loss in the temperature range of around 550-700 K was attributed to decompositions of the ligands of HKUST-1 [59]. In particular, it was found that the weight loss of CP-72, HKUST-1, and HKUST-1@CP-72-E-7.3 at 500-800 K was increased from 0.6% to 2.3%, while that of the first stage (300-500 K) was nearly unexchanged, compared with that of CP-72. Similarly, the weight loss of HKUST-1@CP-72-E-7.3 ( Figure S9e) at 300-500 K was approximately 8%, almost the same as that of pure CP (Figure S9a), and its weight loss of around 2.3% at 500-600 K was ascribed to the dehydration of hydrated copper cations and the physical desorption of the solvent. The weight loss of around 5.5% at 600-1000 K belonged to the decompositions of the ligands of HKUST-1. It is actually difficult for us to verify that the thermal stability of HKUST-1 located on the CP was improved. As in the abovementioned demonstrations, the crystal size of the synthesized HKUST-1 was very small, resulting in its amount being difficult to quantify. In this regard, the amount of the synthesized HKUST-1 was estimated on the basis of the TG data of Cu-CP-72-E-7.3 ( Figure S9d), HKUST-1 ( Figure S9f), and HKUST-1@CP-72-E-7.3 ( Figure S9e), and the HKUST-1 amount in HKUST-1@CP-72-E-7.3 was estimated to be about 8.3%, much lower than the theoretical value (about 23%, assuming that all of the Cu 2+ ions distributed on CP were coordinated with ligands). The probable reason for this is that the steric hindrance effect strongly limited the coordination of most of Cu 2+ ions on CP with the ligands, even though the CP lamellas were disordered after delamination. The effective mechanism of PEG essentially restrained the formation of the crystal nucleus but promoted the growth procedure of the CPs, leading to the reduction of their peak (200) intensity and the decrease in the lamellar size of the synthesized CP. Figures 5, 6 and S10 all show the CO 2 and CH 4 equilibrium adsorption capacities and isotherms of the related samples. As can be seen in Figure 5A,C, the CO 2 and CH 4 adsorption capacity of the HKUST-1@CPs increased gradually with the increase in the crystallinity of the CP, namely, HKUST-1@CP-0-E-10.5 < HKUST-1@CP-24-E-10.1 < HKUST-1@CP-72-E-7.3, showing 0.68, 0.78, and 1.14 mmol/g for CO 2 at 273 K ( Figure 5(Aa-c)) and 0.06, 0.11, and 0.23 mmol/g for CH 4 at 273 K ( Figure 5(Ca-c)), respectively, which is the same phenomena at 298 K as the one shown in Figure 5B,D. Figure S10A,C presented the CO 2 and CH 4 adsorption capacity of the CPs synthesized at 273 K, showing the enhanced tendencies with the increased crystallinity, from 0.80 and 0.02 mmol/g for CP-0 to 1.39 and 0.31 mmol/g for CP-72, respectively. The similar phenomena at 298 K were also shown in Figure S10B,D. Meanwhile, as shown in Figure S10A-D, we noticed that the CO 2 and CH 4 adsorption capacity at 1 bar at both 273 K and 298 K was decreased in the following order: pure CP > CP-72 > CP-72-E, suggesting that the effects of the disordered lamellar morphologies of the synthetic CPs on the CO 2 and CH 4 adsorption performances were not obvious. Figure S10A,C presented the CO2 and CH4 adsorption capacity of the CPs synthesized at 273 K, showing the enhanced tendencies with the increased crystallinity, from 0.80 and 0.02 mmol/g for CP-0 to 1.39 and 0.31 mmol/g for CP-72, respectively. The similar phenomena at 298 K were also shown in Figure S10B,D. Meanwhile, as shown in Figure As is well known, the dynamic diameter is 0.38 nm for CO2 and 0.33 nm for CH4, besides their almost same polarizability. However, the quadrupole of CO2 is 4.30 × 10 −26 cm 2 , while the quadrupole of CH4 is almost zero. Obviously, the interaction force of CO2 with the synthesized CPs is stronger than that of CH4, which is conducive to the enhancements of its adsorption capacity and the improvement of its adsorption selectivity [62]. Meanwhile, Xin et al. proposed the possible CO2 adsorption mechanism on HKUST-1 [63], suggesting the appearance of two adsorption sites: one is the small pore cages composed of a ring of six metal dimers and six trimesate groups, and the other is the open Cu ions. Spanopoulos et al. further elucidated that the adsorption performances of CO2 mainly occurred at the two sites stated above, while that of CH4 only occurred at the open Cu structures [64]. Koyama et al. proposed four kinds of the exchangeable cation positions, namely, M(l), M(2), M(3), and M(4), which are distributed in different channels [65]. In detail, M(l) is located in channel A and is coordinated by oxygen atoms of two frameworks and five water molecules. M(2) appeared in channel B and is adjacent to the oxygen atoms of three frameworks and five water molecules. M(3) centered on channel C is nearly the center of its eight-member ring. Similar to M(l), M(4) appeared in channel A at a center of inversion, and can be coordinated with six water molecules occupying the vertices of an octahedron. The M(4) position may accommodate these excess atoms [65][66][67]. As reported by Garcia-Basabe et al. [68], most of the Cu 2+ on CP were located in the center of channel A and channel B of two extra framework sites, indicating that the Cu 2+ presence was mainly distributed at M(1) and M (4). They further demonstrated that an additional new site was found in channel A at distances of 0.145 and 0.165 nm from the M(1) and M(4) sites, originating from the Cu 2+ coordination with water molecules. CO 2 and CH 4 Adsorption Performances In the present work, our results seem to be related to the Cu 2+ positions distributed on the synthetic CPs and the adsorption sites, depending on the crystallization of the synthetic CPs and the possibilities of the coordinated HKUST-1. However, as shown in Figure 6A,C, the CO 2 and CH 4 adsorption capacity of HKUST-1@CP-72-E-3.4 presented 1.39 and 0.38 mmol/g at 273 K, more than that (1.14 and 0.23 mmol/g) of HKUST-1@CP-72-E-7.3. The similar results at 298 K were also shown in Figure 6B,D. Meanwhile, Figure S10E,G showed that the CO 2 and CH 4 adsorption capacity of Cu-CP-72-E-3.4 at 273 K were 1.25 and 0.28 mmol/g, lower than that (0.84 and 0.21 mmol/g) of Cu-CP-72-E-7.3, respectively. The similar observations at 298 K were also shown in Figure S10F,H. As is well known, the dynamic diameter is 0.38 nm for CO 2 and 0.33 nm for CH 4 , besides their almost same polarizability. However, the quadrupole of CO 2 is 4.30 × 10 −26 cm 2 , while the quadrupole of CH 4 is almost zero. Obviously, the interaction force of CO 2 with the synthesized CPs is stronger than that of CH 4 , which is conducive to the enhancements of its adsorption capacity and the improvement of its adsorption selectivity [62]. [65]. In detail, M(l) is located in channel A and is coordinated by oxygen atoms of two frameworks and five water molecules. M(2) appeared in channel B and is adjacent to the oxygen atoms of three frameworks and five water molecules. M(3) centered on channel C is nearly the center of its eight-member ring. Similar to M(l), M(4) appeared in channel A at a center of inversion, and can be coordinated with six water molecules occupying the vertices of an octahedron. The M(4) position may accommodate these excess atoms [65][66][67]. As reported by Garcia-Basabe et al. [68], most of the Cu 2+ on CP were located in the center of channel A and channel B of two extra framework sites, indicating that the Cu 2+ presence was mainly distributed at M(1) and M (4). They further demonstrated that an additional new site was found in channel A at distances of 0.145 and 0.165 nm from the M(1) and M(4) sites, originating from the Cu 2+ coordination with water molecules. In the present work, our results seem to be related to the Cu 2+ positions distributed on the synthetic CPs and the adsorption sites, depending on the crystallization of the synthetic CPs and the possibilities of the coordinated HKUST-1. Additionally, the CO 2 and CH 4 adsorption capacities of Cu-CP-72-E-7.3 at 273 K and 298 K (shown in Figure S10(Eb,Fb,Gb,Hb)) were even lower than that of CP-72-E (shown in Figure S10(Ae,Be,Ce,De)). These phenomena were mainly because of the hydrochloric acid generation deriving from the Cu 2+ hydrolysis leading to the destruction of the adsorption sites, which is consistent with the decrease in CP crystallinity that occurs with the increase in Cu content (as shown in Table S1). On the basis of the abovementioned results, the CO 2 and CH 4 isosteric adsorption heat of all related samples was calculated via the Clausius-Clapeyron equation, seen here in Equation (5): where P 1 and P 2 represent the pressure under the temperatures of T 1 and T 2 , ∆H vap represents the isosteric adsorption heat of CO 2 , and R represents the gas constant (8.314 J·mol −1 ·K −1 ) [69]. The Freundlich-Langmuir (F-L) equation, which is Equation (6) below, was used to fit the relationship between adsorbed amounts and relative pressure under the same adsorbed amounts: where Q is the adsorbed amount under different relative pressure (mmol/g), q s is the molar adsorption capacity of the system when the Q is nearly equal with an increase in relative pressure (mmol/g), K and n are F-L constants, and C is the relative pressure. The isosteric heat of CO 2 and CH 4 adsorption for various samples is shown in Figure 7. As can be seen there, the isosteric heats of CO 2 and CH 4 adsorption were less than zero, indicating that the CO 2 and CH 4 adsorption belonged to an exothermic process [70]. In detail, the CO 2 adsorption heats of HKUST-1@CP-0-E-10.5 (Figure 7 Additionally, the CO2 and CH4 adsorption capacities of Cu-CP-72-E-7.3 at 273 K and 298 K (shown in Figure S10(Eb,Fb,Gb,Hb)) were even lower than that of CP-72-E (shown in Figure S10(Ae,Be,Ce,De)). These phenomena were mainly because of the hydrochloric acid generation deriving from the Cu 2+ hydrolysis leading to the destruction of the adsorption sites, which is consistent with the decrease in CP crystallinity that occurs with the increase in Cu content (as shown in Table S1). On the basis of the abovementioned results, the CO2 and CH4 isosteric adsorption heat of all related samples was calculated via the Clausius-Clapeyron equation, seen here in Equation (5): where P1 and P2 represent the pressure under the temperatures of T1 and T2, ΔHvap represents the isosteric adsorption heat of CO2, and R represents the gas constant (8.314 J·mol −1 ·K −1 ) [69]. The Freundlich-Langmuir (F-L) equation, which is Equation (6) below, was used to fit the relationship between adsorbed amounts and relative pressure under the same adsorbed amounts: where Q is the adsorbed amount under different relative pressure (mmol/g), qs is the molar adsorption capacity of the system when the Q is nearly equal with an increase in relative pressure (mmol/g), K and n are F-L constants, and C is the relative pressure. The isosteric heat of CO2 and CH4 adsorption for various samples is shown in Figure 7. As can be seen there, the isosteric heats of CO2 and CH4 adsorption were less than zero, indicating that the CO2 and CH4 adsorption belonged to an exothermic process [70]. In detail, the CO2 adsorption heats of HKUST-1@CP-0-E-10.5 (Figure 7 Figure 7(Ac,d,Bc,d)) with different Cu contents was higher than that of CH4, indicating a stronger CO2 adsorption. Obviously, these results suggest that the CO2 adsorption is preferential compared with that of CH4. Meanwhile, the absolute values of the isosteric heat of all samples were less than 40 kJ/mol, indicating that the behaviors of the CO2 or CH4 adsorption belonged to the physical adsorption [71]. The absolute value of the CO 2 isosteric heat of HKUST-1@CP-72-E-3.4 and HKUST-1@CP-72-E-7.3 (shown in Figure 7(Ac,d,Bc,d)) with different Cu contents was higher than that of CH 4 , indicating a stronger CO 2 adsorption. Obviously, these results suggest that the CO 2 adsorption is preferential compared with that of CH 4 . Meanwhile, the absolute values of the isosteric heat of all samples were less than 40 kJ/mol, indicating that the behaviors of the CO 2 or CH 4 adsorption belonged to the physical adsorption [71]. To further verify the separation performance of CO 2 /CH 4 of the samples, the selectivity at 273 K and 298 K was shown in Figure 8. As can be seen in Figure 8A, HKUST-1@CP-0-E-10.5 and HKUST-1@CP-24-E-10.1 with low crystallinity of CP (Figure 8(Aa,b)) showed higher selectivity owing to their low CH 4 adsorption capacity, but their low CO 2 adsorption capacity may limit their applications in CO 2 and CH 4 separation. The similar results of their CO 2 /CH 4 selectivity at 298 K were also shown in Figure 8(Ba,b). To further verify the separation performance of CO2/CH4 of the samples, the selectivity at 273 K and 298 K was shown in Figure 8. As can be seen in Figure 8A, HKUST-1@CP-0-E-10.5 and HKUST-1@CP-24-E-10.1 with low crystallinity of CP (Figure 8(Aa,b)) showed higher selectivity owing to their low CH4 adsorption capacity, but their low CO2 adsorption capacity may limit their applications in CO2 and CH4 separation. The similar results of their CO2/CH4 selectivity at 298 K were also shown in Figure 8(Ba,b). Meanwhile, the effect of the coordination of HKUST-1 with Cu-CPs on the separation performance of CO2/CH4 and the corresponding selectivity at 273 K and 298 K was evaluated. As shown in Figure 8(Ac-e,Bc-e), the selectivity of HKUST-1@CP-72-E-3.4 was lower than that of HKUST-1@CP-72-E-7.3 but higher than that of the prepared HKUST-1 at both 273 K and 298 K. Obviously, one of the main reasons for this is due to the generation of HKUST-1 in the HKUST-1@CPs, with an increase in the Cu content. The CO2/CH4 selectivity of various samples at different temperatures under the pressure of 1 bar was collected in Table 1. Comparably, the reported selectivity of HKUST-1 depended on the prepared methods [24,59,72]. Meanwhile, the effect of the coordination of HKUST-1 with Cu-CPs on the separation performance of CO 2 /CH 4 and the corresponding selectivity at 273 K and 298 K was evaluated. As shown in Figure 8(Ac-e,Bc-e), the selectivity of HKUST-1@CP-72-E-3.4 was lower than that of HKUST-1@CP-72-E-7.3 but higher than that of the prepared HKUST-1 at both 273 K and 298 K. Obviously, one of the main reasons for this is due to the generation of HKUST-1 in the HKUST-1@CPs, with an increase in the Cu content. The CO 2 /CH 4 selectivity of various samples at different temperatures under the pressure of 1 bar was collected in Table 1. Comparably, the reported selectivity of HKUST-1 depended on the prepared methods [24,59,72]. [72] 273 K 7.19 HKUST−1 2 cited in references [24,59] 273 K 4.00 Breakthrough Performance Taking HKUST-1@CP-72-E-7.3 as an example, Figure 9 showed the breakthrough profiles of the binary mixtures (CO 2 and CH 4 ) at 50% CO 2 concentration (by volume) through the fixed bed packed with the sample pellets at 298 K. As can be seen, the relationships between C/C 0 versus time, where C and C 0 were the volumetric concentration of CO 2 or CH 4 in the outlet and inlet stream, showed that the purified CH 4 was detected in the outlet gas flow but that CO 2 was still trapped in the column when the mixture of CO 2 and CH 4 flowed into all of the column, thereby showing the highly adsorbed CO 2 performance of the prepared HKUST-1@CP as well as the adsorptive separation properties between the CH 4 and CO 2 [73,74]. through the fixed bed packed with the sample pellets at 298 K. As can be seen, the rela tionships between C/C0 versus time, where C and C0 were the volumetric concentration of CO2 or CH4 in the outlet and inlet stream, showed that the purified CH4 was detected in the outlet gas flow but that CO2 was still trapped in the column when the mixture of CO and CH4 flowed into all of the column, thereby showing the highly adsorbed CO2 performance of the prepared HKUST-1@CP as well as the adsorptive separation properties between the CH4 and CO2 [73,74]. Conclusions The HKUST-1@CPs with randomly lamellar morphologies were successfully synthe sized via the PEG-additive hydrothermal route and the Cu 2+ coordinated routes with tri mesic acid. Various characterizations demonstrated that the additive PEG had a strong hindering effect on the formation of the crystal nucleus of CPs but a promoting effect on their growth, although their nucleation process was a controlled step in the crystallization duration. HKUST-1@CP with various Cu 2+ contents or different crystallizations presented the enhancements of the adsorptive separation for CO2 and CH4 as compared with Cu 2+ exchanged CPs and synthesized HKUST-1. In particular, the higher CO2/CH4 selectivity and better breakthrough performances demonstrated that the obtained HKUST-1@CPs may be a good candidate for further study on potential applications in gas separation. Supplementary Materials: The following supporting information can be downloaded at www.mdpi.com/xxx/s1. Figure S1. Schematic of the PEG-assisted synthesis of CP via crystallization and delamination process. Figure S2 Figure S10. Equilibrium Conclusions The HKUST-1@CPs with randomly lamellar morphologies were successfully synthesized via the PEG-additive hydrothermal route and the Cu 2+ coordinated routes with trimesic acid. Various characterizations demonstrated that the additive PEG had a strong hindering effect on the formation of the crystal nucleus of CPs but a promoting effect on their growth, although their nucleation process was a controlled step in the crystallization duration. HKUST-1@CP with various Cu 2+ contents or different crystallizations presented the enhancements of the adsorptive separation for CO 2 and CH 4 as compared with Cu 2+ exchanged CPs and synthesized HKUST-1. In particular, the higher CO 2 /CH 4 selectivity and better breakthrough performances demonstrated that the obtained HKUST-1@CPs may be a good candidate for further study on potential applications in gas separation. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nano13121860/s1, Figure S1. Schematic of the PEG-assisted synthesis of CP via crystallization and delamination process; Figure Table S1. Summaries of the prepared conditions and various parameters of the obtained Cu-CPs; Table S2. Summaries of the amounts of the used ligand and triethylamine; Table S3. Summaries of the textural parameters of the related samples; Table S4. Summaries of various kinetic parameters during CP crystallization. Data Availability Statement: The data presented during the study is available on request from the corresponding author.	2023-06-28T06:17:26.620Z	2023-06-01T00:00:00.000	{ "year": 2023, "sha1": "3f50ffe11bca1e511abed41f57aeafd67c4b7df0", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2079-4991/13/12/1860/pdf?version=1686748960", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "874f9d07f6fff88ec9a37832d722dcb9648ed995", "s2fieldsofstudy": [ "Materials Science", "Chemistry" ], "extfieldsofstudy": [ "Medicine" ] }
118902098	pes2o/s2orc	v3-fos-license	Bouncing cosmology in $f(R,T)$ gravity A cosmological model with a specific form of the Hubble parameter is constructed in a flat homogeneous, and isotropic background in the framework of $f(R, T)$ gravity, where $R$ is the scalar curvature and $T$ is the trace of the stress-energy-momentum tensor. The proposed functional form of the Hubble parameter is taken in such a way that it fulfills the successful bouncing criteria to find the solution of the gravitational field equations provided the Universe is free from initial singularity. The various constraints on the parameters are involved in the functional form of the Hubble parameter which is analyzed in detail. In addition, we explore physical and geometrical consequences of the model based on the imposed constraints. Furthermore, we demonstrate the bouncing scenario which are realized in our model with some particular values of the model parameters. As a result, we find that all of the necessary conditions are satisfied for a successful bouncing model. Introduction In the present scenario, we are familiar with the accelerating expansion of the Universe in the early phase of evolution (inflation) as well as in the late Universe. Various cosmological observations give evidence of late time cosmic acceleration. Some of them are observations of high red-shift supernovae [1], cosmic microwave background radiations (CMBR) [2,3], supernovae of type Ia [4,5], Planck data [6] and baryon acoustic oscillations [7]. The fundamental pillar of the modern cosmology is the general theory of relativity (GTR). As of now GTR explains the large-scale structure of the Universe very well theoretically. In order to explain the late-time cosmic acceleration, different approaches have been developed in the past few decades e.g. modifying the energy momentum tensor or modifying the geometry in the Einstein's field equations (EFEs). The inclusion of the matter with highly negative pressure termed as dark energy (DE) in the right side of EFE is much successful in explaining the puzzle of the late-time Universe as observations reveal. But the nature of this mysterious DE is unknown to us as there is no any direct evidence of DE. Generally, it is believed that DE is a homogeneous fluid that permeates all over the space contributing almost 2/3 of the total volume of the Universe. However, this is matter of a great debate on the candidature of DE among the theorists. Many theoretical cosmological models of the Universe have been suggested to examine the behavior of the DE. Out of the numerous candidates of DE, the most efficient and prospective version is the cosmological constant introduced by Einstein [8,9]. However, it is afflicted by most familiar cosmological constant problem [10] which can be soothed by assuming a dynamical decaying (Λ). In another way, early inflation plays a major role in understanding the anisotropies in the CMBR and formation of large scale structures. Literature includes the list of DE candidates, such as quintessence, f-essence, k-essence, spintessence, tachyons, phantom, Chaplygin gas etc. (for a brief review see [11]). Despite of these excellent DE cosmological models, we lack behind on some issues and the hunt for a concrete model is still open. This motivates the theorists to consider alternative theories of gravity. The modification of GTR in which the origin of DE is associated by the rearrangement in gravity as it is described by Riemannian geometry and is without torsion. Although in many literatures, a number of gravitational theories have been studied in which the torsion effects appear in the extension of GTR. Some of the alternating theories are f (R) gravity [12,13,14,15,16], f (T ) gravity [17], where T is the torsion scalar in teleparallel gravity, f (G) gravity [18], where G is the Gauss-Bonnet invariant, etc. (for recent reviews on the dark energy problem and modified gravity theories, see [19]). However, many alternative theories have been studied in the past few decades such as Brans-Dicke theory, Einstein-Cartan theory, Loop quantum theory, Kaluza-Klein theory and many more to overcome the cosmological issues. Another prospective and efficient theory among the alternative theories is f (R, T ) gravity, where R and T are the Ricci scalar curvature and trace of stress-energy-momentum tensor (SEMT) respectively [20]. According to this theory there is an arbitrary coupling constant between matter and geometry which is responsible for a source term, performing the matter-stress-energy tensor with respect to the metric. The different choice of the matter Lagrangian L m would cause an explicit set of field equations. So many new investigations have come on the surface to describe the present cosmic acceleration in f (R, T ) gravity [21,22,23,24,25,26,27,28,29,30,31]. Several homogeneous isotropic and anisotropic cosmological models have been constructed in f (R, T ) theory of gravity for the past few years [32,33,34,35,36,37,38,39,40]. Houndjo et al. [41] has reconstructed a cosmological model in f (R, T ) gravity which is able to discuss the expansion history of the model in GTR by dark matter as well as by holographic dark energy. Barrientos et al. [42] has studied the f (R, T ) theories of gravity and its application with affine connection. According to the standard cosmological model, the Universe came into existence from a singularity (big bang) in the space-times which has some shortcomings e.g. horizon problem, flatness problem, transplanckian problem, entropy problem, original structure problem, and singularity problem. To resolve these issues, a sudden growth after the big bang was necessary to constitute a uniform, flat and smooth universe. Inflationary theory was developed by Alan Guth to solve a bunch of these standard cosmological problems which proved to be a quite successful to describe the various observational properties of the Universe. However, one of the fundamental questions in modern cosmology still remains unsolved i.e. the initial singularity problem. One of the attractive possible alternatives to the inflationary model was developed as a bouncing model of the Universe which solves the initial singularity problem during mid 1980's. The specification of this bouncing scenario is that the Universe may have emerged from a prior contracting phase and capable of being expanding without singularity or it experiences a bouncing process [43,44]. Some excellent research works on a nonsingular bouncing cosmology have been carried out by Cai et al. [45,46,47,48,49] and Brandenberger et al. [50] in recent years in which they studied the various phenomenological aspects of the bouncing scenario e.g. the cosmology of a Lee-Wick type scalar field theory, single scalar field matter containing a potential and kinetic term, a contracting universe that consists of radiation, cold dark matter (CDM) and a positive cosmological constant, bounce model with dark matter (DM) and DE, observational bouncing cosmologies such as the Planck and BICEP2 data, and the role of bouncing cosmologies as alternative theories to the cosmological inflation which are consistent with present day observational data. Bamba et al. [51,52,53] have discussed bouncing cosmological models in f (R) gravity by reconstructing a method, in Gauss-Bonnet gravity where the Gauss-Bonnet invariant couples to a dynamical scalar field, in f (G) gravity with the Gauss-Bonnet invariant G by reconstructing a method to search the bouncing scenario in the early Universe as well as examine the stability conditions for its solutions, and a bouncing inflationary model with a graceful exit into the Friedmann-Lemaitre-Robertson-Walker (FLRW) model in f (T ) gravity, T being the torsion scalar respectively. Bamba et al. [54] have also explored a bouncing inflationary model with a graceful exit into the FLRW model in f (T ) gravity. de la Cruz-Dombriz et al. [55] has discussed the bouncing cosmological model in the extended theory of teleparallel gravity. Cai et al. [56] have examined the matter bounce cosmological models in f (T ) gravity. In this work, we propose a new form of the parametrization of HP which varies with cosmic time t and study the evolution of the Universe in f (R, T ) gravity in the framework of a flat FLRW metric. Different cases arise while imposing the restrictions on model parameters, which involved the functional form of the Hubble parameter H. This leads to some accelerating expansion without singularity in a bouncing scenario. The physical consequences of the model have also been discussed. The outline of the work is as follows: In Sect. II, we present a brief review on f (R, T ) gravity and discuss the metric and its field equations. In Sect. III, we study the necessary conditions in order to accomplish a successful bouncing model of the Universe in standard cosmology and discuss the Hubble parameter (HP), Deceleration parameter (DP), equation of state (EoS) parameter for a bouncing model, and analyze the HP and DP mathematically on various constraints. In Sect. IV, we take a new parametrization of HP for some specific values of the model parameters c and λ 2 to understand our proposed model through various plots. In Sect. V, we examine the scalar field and self interacting potential by adopting Barrow's scheme [57]. Finally, we discuss and summarize the results in section VI. OVERVIEW OF f (R, T ) GRAVITY AND FIELD EQUATIONS The f (R, T ) theory is the modification of GTR in which a matter Lagrangian L m can be described by the combination of R and T , where R and T are the scalar [20]. The total gravitational action of f (R, T ) gravity becomes where g and G indicate the metric determinant and the gravitational constant respectively. Several forms of f (R, T ) function are given in literature; here we are assuming the coupling between R and T in the form f (R, T ) = R + 2f (T ). Furthermore, we assume f (T ) = λ 1 T , where λ 1 is an arbitrary constant. The above functional form of f (R, T ) is designed in such a manner that GTR can be obtained for λ 1 = 0. By varying the action (1) with respect to the metric tensor components yield where prime denotes differentiation with respect to the argument. We consider, here, a perfect fluid as the matter source filled in the Universe and therefore, SEMT of L m can be taken as where ρ and p are the cosmic energy density and isotropic pressure of the fluid respectively, and u i = (0, 0, 0, 1) represents the four velocity vector components in the comoving coordinate system which satisfies the conditions u i u i = 1 and u i ∇ j u i = 0. We choose the perfect fluid matter as L m = −p in the action (1). We consider a flat FLRW background geometry with a metric where a(t) is the scale factor; the gravitational field equations in f (R, T ) gravity (2) are where an overhead dot indicates the differentiation of the quantity with respect to cosmic time t. In the present scenario, we need one more constraint equation to solve the system of field equations completely. This constraint equation is the state equation of a fluid in general. But, here we are interested in a bouncing scenario which is not a new concept and has a long history for which we put a constraint on the Hubble parameter [58,59]. However, one can also consider a parametrization of physical or geometrical parameters to get a constraint equation which is consistent with the system. This is generally called the model independent way to study the DE models without violating the background theory. For a brief review for various parametrizations of cosmological parameters, see [60,61,62]. BOUNCING SOLUTION AND PARAMETRIZATION OF H In recent times, the bouncing scenario has gained popularity where the big bang is replaced by big bounce. The big bounce cosmological model can be interpreted as an oscillatory universe or the cyclic universe where one cosmological event was the outcome of the collapse of another or previous universe. A bouncing universe which contracted to a finite volume initially, then expanded subsequently provides a possible solution to the singularity problem of the standard big bang theory within GTR. For a successful bounce, it can be observed that the violation of a null energy condition (NEC) is required for a period of time in the vicinity of the bouncing point within a FLRW background geometry. Moreover, the EoS parameter ω of the matter content present in the Universe must undergo a phase transition from ω < −1 to ω > −1, to enter into the hot big bang age after the bounce [60,61]. The observational data [63] support the quintom model [64], which has been proposed to explore the behavior of the DE with an EoS parameter w > −1 and w < −1 in the past and at present respectively. The quintom model is a nonstatic model of DE which behaves distinctively from the other DE models e.g. cosmological constant, quintessence, phantom, k-essence etc. in the determination of the cosmic evolution. The detailed description on the necessary conditions in order to accomplish a successful bouncing model of the Universe in standard cosmology are as follows [60]: • In the contracting universe, a scale factor a(t) is decreasing i.e.ȧ(t) < 0, and in the expanding universe, the scale factor a(t) is increasing, i.e.ȧ(t) > 0. The cosmic scale factor reaches to a nonzero minimum value at the transfer point. This kind of bouncing scenario can avoid the singularity naturally which is inevitable in the standard model. In other words, during the bouncing point,ȧ(t) = 0 andä(t) > 0 for some period of time in the neighborhood of a bounce point. • Equivalently the HP passes through zero from H < 0 when the Universe contracts to H > 0 when the Universe expands and H = 0 when the bouncing point occurs. A successful bouncing model in standard cosmology requiresḢ = 4πGρ(1 + w) > 0 in the neighborhood of a bouncing point which is equivalent to the violation of the null energy condition (NEC) in Einstein gravity. From this equation, one can observe that w < −1 around the bouncing point. • The EoS parameter ω crosses the phantom divide line (quintom line) ω = −1 which is the remarkable feature of the quintom model. Motivated by the above bouncing scenario and the model independent way to study cosmological models, here in this paper, we would like to emphasize on the cosmographic parameter H that describes the expansion of the Universe and helps us to achieve some impressive bouncing solutions to the EFEs. We parametrize the functional form of the Hubble parameter as a product function given by where α is any arbitrary constant and h(t) is any analytic function. Looking at the proposed form of the Hubble parameter, we can observe that the algebraic function t present in the functional forces to vanish H at t = 0 implying that the scale factor function must take a constant value at t = 0 [the second function h(t) being nonvanishing at t = 0]. We have a choice on h(t) which could be any rational or transcendental, or periodic function to get another bounce in the future. Here, in this study, we consider a specific functional form of h(t) as so the parametrization of HP takes the form where λ 2 and c(> 0) are arbitrary constants and have time dimensions (we call them model parameters), which describe the dynamics of the Universe. In the following sections, we discuss the cosmological parameters for our model. Hubble parameter The aforesaid form of the HP in Eq. (9) is bouncing in nature and have some specific features. The sign of Hubble parameter decides the expansion, contraction and bounce of the Universe. In our case • The Universe is expanding if both α and h(t) have same signature. • The Universe is contracting if α and h(t) have opposite signature. • The Universe bounce i.e. H = 0 in the case either α or h(t) vanishes. Here, bounce occurs at t = 0 (initial bounce) and t = 3(1+λ2)2 We can discuss the future bounce w.r.t. λ 2 and c through the following plots shown in Figs. 1a and 1b. In these figures, we have fixed α = 5 (α can be treated as scaling constant) and plotted H(t) for different values of model parameter c by taking a fixed value of λ 2 in Fig. 1a. Similarly, we have fixed the value of c in Fig. 1b and plotted H(t) for different values of λ 2 . We observe that the future bounce depend on the value of model parameters c and λ 2 . The future bounce is delayed by increasing the value of c (see Fig. 1a). Similarly the future bounce is delayed by decreasing the value of λ 2 (see Fig. 1b). This behavior of future bounce is analyzed with the constrain c > t > 0 (see case 1 of Table 1). The different inequalities on the model parameters α , λ 2 , c leading to some cases of expansion of the Universe are given in the following Table 1. constraints on c Deceleration parameter The deceleration parameter q is given by Using the functional form of H(t) into Eq. (10), the expression for the deceleration parameter q is obtained as which depends on cosmic time t, and the signature of q depends on the model parameters that describes the dynamics of the Universe. The Universe exhibits accelerated expansion according as the current observations. Hence, we emphasize on accelerating cases only. The constraints on model parameters for the accelerating cases is estimated in Table 2. From the plot of deceleration parameter (see Fig. 2), we can easily inspect that the Universe exhibits acceleration throughout the evolution under the restrictions on the model parameters given in Table 2. EoS parameter Here, we have two constants λ 1 and λ 2 involved in our model. Without the loss of generality, we assume that λ 1 = λ 2 = λ to get a deterministic solution. With the help of assumption of Hubble parameter given in Eq. (9), the EFEs (5) and (6) can now be solved explicitly. The expressions for the energy density ρ and the pressure p of the cosmic fluid are Table 2. The EoS parameter is whose behavior can be analyzed on the various constrains given in Table 1 (see Fig. 3). Table 1. In case 1 of Table 1, when the coupling constant of f (R, T ) gravity λ is positive, the energy density ρ and the pressure p satisfies the EoS parameter ω with ω ≥ 0 at initial stage. In the late time ω transits its phase from perfect fluid to the quintessence region −1 < ω < 0 and approaches to the phantom divide line (quintom line) but never enters in to the phantom region. In case 2, 3 and 5 of Table 1, energy density and pressure satisfies the EoS ω < −1 due to the negative coupling constant of f (R, T ) gravity and in the late time approaches to the quintom line ω = −1. In case 4 of Table 1, EoS parameter decreases promptly from perfect fluid region to phantom phase, and approaches to quintom line in late times. Exemplification We take some particular values of the model parameters c and λ 2 to have a concrete understanding of our proposed model. We consider a more concise form of the function h(t) given in Eq. (8) by restricting the term tan −1 t up to third order only and by taking the model parameter c = 1 and λ 2 = 1. The new parametrization of HP in Eq. (9) takes the specific form as which is bouncing in nature having bounce at t 0.618 (we have neglected the coefficient of t 3 for mathematical ease). From Fig. 4 and 5, we observe that (i) the Hubble parameter H < 0 in the interval 0 < t < 0.618, H > 0 in the interval 0.618 < t < 1 and H = 0 at t 0.618 for α < 0 (see Fig. 4a), (ii) During the contracting universe, the scale factor a(t) is decreasing i.e.ȧ(t) < 0, and the scale factor a(t) shows increasing pattern, i.e.ȧ(t) > 0 during the expanding phase of the Universe. The scale factor of the Universe reaches to a non-zero minimum value a 0.348 at the transfer point t 0.618 for α < 0 (see Fig. 4b), (iii)Ḣ > 0 in the interval 0.2295 < t < 0.8167 i.e. in the neighborhood of bouncing point at t 0.618. Therefore the null energy condition (NEC) is violated in same interval (see Fig. 5a), (iv) Our obtained model is a Quintom model as the EoS parameter ω of the matter content undergoes a phase transition from ω < −1 to ω > −1 in the neighborhood of bouncing point at t 0.618. Therefore, in this scenario our Universe enters into the hot big bang age after the bounce. [60,61] (see Fig. 5b). Since all the above criterion are fulfilled by our derived cosmological model. Therefore, we can say that our model is a non-singular bouncing model within FLRW Universe in the background. Moreover, this model behaves like a Quintom model [64] which is supported by the observational data [63]. On the basis of above observations, we can predict that this model is very helpful to study the behaviors of the DE with an EoS parameter w > −1 in the past and w < −1 at present. Scalar field description In the recent years, the Quintom model have earned a great popularity to study the bouncing cosmological model within GTR. The simplest Quintom model contains two types of scalar fields: one is the quintessencelike and other is the Phantomlike. However, it is not easy to constitute a Quintom model theoretically since ω = −1 is not consistent with observations. If we want our model to be consistent with observations, we need to take ω −1. Thus we needφ 2 << V (φ) i.e. the kinetic energy of the scalar field is negligible in comparison to the potential energy. If ω −1, there are many models to explain acceleration. We can use exactly the same model for inflation. Here, we are interested to study the non-singular bouncing cosmological model using scalar fields in the background of f(R,T) gravity. The Einstein theory of gravity is defined by the following action where S m is the action for the quintessencelike and phantomlike scalar field denoted by S mq and S m ph respectively. Here we use normalization by taking c = 1. The action for the quintessencelike and phantomlike scalar field are given by and respectively. As the scalar field φ is time dependent, therefore it can be considered as perfect fluid with energy density ρ φ and pressure p φ . We assume that if the scalar field φ is the only source of DE having potential V (φ) which interacts with itself, so we can consider energy densities ρ φq , ρ φ ph and pressures p φq , p φ ph for the quintessencelike and phantomlike scalar fields in the framework of FLRW cosmology as Here, the suffixes q and ph correspond to quintessencelike and phantomlike scalar field respectively. The kinetic energies 1 2φ 2 q , 1 2φ 2 ph and potential energies V (φ q ), V (φ ph ) of quintessencelike and phantomlike scalar field correspondence are 1 2φ . show the DE models due to repulsive force in the interval 0.2252 < t < 0.8151 (Fig. 3a) and 0.08995 < t < 0.857 (Fig. 3b) in the neighborhood of bouncing point at t 0.618. From Eq. (23), we find that the potential energy V (φ q ) and V (φ ph ) for quintessence and phantomlike scalar field are equal and positive in the interval 0.08798 < t < 0.859 in the neighborhood of bouncing point at t 0.618. In case of phantomlike and quintessencelike scalar field, the EoS parameters ω are given by In case of Quintom behavior of the model when the EoS parameter ω crosses over the line ω = −1 then from Eqs. (21), (22) and (23), we have which is the necessary condition to the model having bouncing behavior and is consistent with the results of Cai et al. [60]. Hence, we conclude that our model is a non singular bouncing model in f (R, T ) gravity. Discussions and Conclusions In this paper, we have studied the flat FLRW model with a specific form of HP, which is a function of time t. Several different forms of HP have already been proposed in literature, but our parametrization possesses some specific features. We have obtained the deterministic solution to EFEs under our parametrization scheme (9) by assuming λ 1 = λ 2 = λ. Further, we have investigated some restrictions on the model parameters α, λ 2 , c leading to some cases of expanding Universe (H > 0) in Table 1. Also, in Table 2, we have found restrictions on model parameters which shows eternal acceleration (q < 0) . We have examined the physical behavior of the deceleration parameter, energy density, matter pressure, and the EoS parameter for the model. In order to have a concrete understanding of the bouncing scenario for our parametrization, we have considered a more concise form of the function h(t), given in (8) by providing some particular values to the model parameters and have discussed all the necessary conditions for a successful bouncing model. Lastly, we have also discussed the self interacting potential V (φ) and kinetic energyφ 2 2 for quintessence and phantom scalar field correspondence in the presence of f (R, T ) gravity and compared it with GR by considering scalar field φ as the source of DE. • In order to study the bouncing nature of the model, various restrictions have been imposed on the model parameters. Under some restrictions, our model shows bouncing behavior at t = 0 and at t = (approx.). We have examined the future bounce by varying the model parameters c and λ 2 (see Fig. 1a, 1b), and it is observed that the future bounce is delayed by rescaling the model parameters c and λ 2 (either by increasing c or decreasing λ 2 ). The model parameters have been taken in such a way that some specific features of our proposal could be studied. • The model exhibits eternal acceleration throughout the evolution of the Universe with some restrictions on model parameters estimated in Table 2. The physical behavior of EoS parameter under some restrictions on H > 0, mentioned in Table 1 is shown in Fig. 3. In our parametrization of H in Eq. (9), ω 1 and ω 4 shows transition from perfect fluid (0 < ω < 1) to DE region (ω < 0) at t 0.825 and t 0.26 respectively. At this time the model represents dust Universe (p = 0), whereas in all the other cases, the model represents DE only. The EoS parameter ω approaches to quintom line in all the cases at late time. • In order to have a concrete understanding of our proposed parametrization of H in Eq. (9), and to explain the bouncing process more precisely, we generate a new parametrization of H in Eq. (15) in a specific form with some particular values of the model parameters c and λ 2 . The bouncing scenario can be accomplished for both negative and positive scaling constant α but here to get an expanding Universe from the prior period of contraction, we have chosen α with negative value1. Some plots have been presented in order to achieve necessary conditions for a successful bouncing model (see Fig. 4, 5). In the parametric form of H (15), the bouncing point is attained at cosmic time t 0.618, which leads to a minimum, non vanishing value of scale factor a(t) (see Fig. 4a, 4b). • For the specific form of H in Eq. (15), the model depicts the bouncing process (i.e. expansion before bounce and contraction after bounce for α > 0, and contraction before bounce and expansion after bounce for α < 0). The first derivative of Hubble parameterḢ > 0 during the period t ∈ (0.2295, 0.8167) leads the violation of null energy condition (NEC), which is the compelling condition for a bouncing scenario in our model (see Fig. 5a). Our model is a Quintom model in which the EoS parameter of the matter content ω transits from phantom phase ω < −1 to quintessence phase ω > −1 in the neighborhood of bouncing point at t 0.618. Therefore, we can see that the Universe enters into the hot big bang era after the bouncing (see Fig. 5b). • In section 5, we have discussed the quintessence-like and phantom-like scalar fields in f (R, T ) gravity and GTR for the parameterization of H stated in Eq. (15). We have observed that both K.E. and P.E. exhibit similar pattern with different scaling in f (R, T ) gravity and GTR (see Fig 6). In this case, our model behaves as Quintom model provided the condition given in Eq. (25) is satisfied, which is the necessary condition to the model having bouncing behavior. This condition is also consistent with the results of Cai et al. [60]. Therefore, we say that our model is a non singular bouncing model in f (R, T ) gravity. • Thus we conclude that the bouncing scenario of the cosmological model have been discussed in f (R) gravity where the scale factor is taken in the forms of exponential and power-law, Gauss-Bonnet gravity, f (G) gravity where the Gauss-Bonnet invariant is taken as G, and f (T ) gravity where T is the torsion scalar in the teleparallelism but we have studied a non-singular bouncing cosmological model in f (R, T ) gravity within a flat FLRW background metric with a specific parametrization of the Hubble parameter which is the main difference of my research work and may also be useful for further investigation.	2018-06-30T16:32:37.000Z	2018-06-26T00:00:00.000	{ "year": 2018, "sha1": "41abbc974e93ec9d4a808fe85999b6878dd777a7", "oa_license": null, "oa_url": "http://arxiv.org/pdf/1807.01157", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "41abbc974e93ec9d4a808fe85999b6878dd777a7", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
14678242	pes2o/s2orc	v3-fos-license	Flips of moduli spaces and transition formulas for Donaldson polynomial invariants of rational surfaces We study the change of moduli spaces of Gieseker-semistable torsion free rank-$2$ sheaves on algebraic surfaces as we vary the polarizations. When the surfaces are rational with an effective anti-canonical divisor, the moduli spaces are linked by a series of flips (blowups and blowdowns). Using these results, we compute the transition formulas for Donaldson polynomial invariants of rational surfaces. Part of the work is also obtained independently by Matsuki-Wentworth and Ellingsrud-G{\"o}ttsche. Introduction. In [7], Donaldson has defined polynomial invariants for smooth simply connected 4-manifolds with b + 2 ≥ 3. These invariants have also been defined for 4-manifolds with b + 2 = 1 in [24,17,18], along lines suggested by the work of Donaldson in [5]. In this case, however, they depend on an additional piece of information, namely a chamber defined on the positive cone of H 2 (X; R) by a certain locally finite set of walls. Explicitly, let X be a simply connected, oriented, and closed smooth 4-manifold with b + 2 = 1 where b + 2 is the number of positive eigenvalues of the quadratic form q X when diagonalized over R. Let be the positive cone. Fix a class ∆ in H 2 (X, Z) and an integer c such that d = 4c − ∆ 2 − 3 is nonnegative. A wall of type (∆, c) is a nonempty hyperplane: in Ω X for some class ζ ∈ H 2 (X, Z) with ζ ≡ ∆ (mod 2) and ∆ 2 − 4c ≤ ζ 2 < 0. The connected components of the complement in Ω X of the walls of type (∆, c) are the chambers of type (∆, c). Then the Donaldson polynomial invariants of X associated to ∆ and c are defined with respect to chambers of type (∆, c). The invariants only depend on the class w = ∆ mod 2 ∈ H 2 (X; Z/2Z) and the integer p = ∆ 2 − 4c, and we shall often refer to walls and chambers of type (w, p) as well. We shall write D X w,p (C) for the Donaldson polynomial corresponding to the SO(3) bundle P with invariants w 2 (P ) = w and p 1 (P ) = p, depending on the chamber C. A basic question is then the following: Suppose that C + and C − are separated by a single wall W ζ . Here there may be more than one class ζ of type (∆, c) defining W ζ . Then find a formula for the difference δ X w,p (C + , C − ) = D X w,p (C + ) − D X w,p (C − ). We shall refer to such a difference as a transition formula. The first author was partially supported by NSF grants DMS-90-06116 and DMS-92-03940. The second author was partially supported by a grant from the ORAU Junior Faculty Enhancement Award Program and by NSF grants DMS-91-00383 and DMS-94-00729. Typeset by A M S-T E X 1 There has been considerable interest in the above problem. The first result in this direction is due to Donaldson in [5], who gave a formula in case ∆ = 0 and c = 1. Kotschick [17] showed that, on the part of the symmetric algebra generated by 2-dimensional classes, δ X w,p (C + , C − ) = ±ζ d for ζ 2 = −(4c − ∆ 2 ) = p, and that δ X w,p (C + , C − ) is in fact always divisible by ζ, except when p = −5 and ζ 2 = −1 (cf. also Mong [24] for some partial results along these lines). For a rational ruled surface X, all the transition formulas for ∆ = 0 and 2 ≤ c ≤ 4 have been determined in [24,33,22]. Using a gauge-theoretic approach, Yang [35] settled the problem for ∆ = 0 and c = 2, and computed the degree 5 Donaldson polynomials for rational surfaces. The known examples and the work of Kotschick and Morgan [18] raise the following rather natural conjecture: Conjecture. The transition formula δ X w,p (C + , C − ) is a homotopy invariant of the pair (X, ζ); more precisely, if φ is an oriented homotopy equivalence from X ′ to X, then δ X ′ φ * w,p (φ * (C + ), φ * (C − )) = φ * δ X w,p (C + , C − ). We remark that this conjecture is essentially equivalent to the following statement: the transition formula δ X w,p (C + , C − ) is a polynomial in ζ and the quadratic form q X with coefficients involving only ζ 2 , homotopy invariants of X (i.e. b − 2 (X)), and universal constants. Our goal in this paper is to study the corresponding problem in algebraic geometry. More precisely, let X be an algebraic surface (not necessarily with b + 2 (X) = 1) and let L be an ample line bundle on X. We can then identify the moduli space of L-stable rank two bundles V on X with c 1 (V ) = ∆ and c 2 (V ) = c with the moduli space of equivalence classes of ASD connections on X with respect to a Hodge metric on X corresponding to L. Let M L (∆, c) be the Gieseker compactification of this moduli space. It is known that M L (∆, c) changes as we change L, and that M L (∆, c) is constant on a set of chambers for the ample cone of X which are defined in a way analogous to the definition of chambers for Ω X given above. Using the recent result of Morgan [25] and Li [21] that the Donaldson polynomial of an algebraic surface can be evaluated using the Gieseker compactification M L (∆, c) of the moduli space of stable bundles, we shall work on M L (∆, c) for suitable choices of L and in particular analyze the change in M L (∆, c) for L ∈ C + or L ∈ C − , where C ± are two adjacent chambers. It turns out that we can obtain M L+ (∆, c) from M L− (∆, c) by a series of blowups and blowdowns (flips). Our results are thus very similar to those of Thaddeus in [31]. Thaddeus [32] and also Dolgachev-Hu [3] have developed a general picture for the variation of GIT quotients after a change of polarization, and although our methods are somewhat different it seems quite possible that they fit into their general framework. We have also found it convenient to borrow some of Thaddeus' notation. Next we shall apply our results on the change in the moduli spaces to determine the transition formula for Donaldson polynomials in case X is a rational surface with −K X effective. We shall give explicit formulas for δ X w,p (C + , C − ) in case the nonnegative integer ℓ ζ = (ζ 2 − p)/4 ≤ 2. These formulas are in agreement with the above conjecture, in the sense that the transition formula is indeed a polynomial in ζ and q X with coefficients involving only ζ 2 , K 2 X , and universal constants. We shall also give a formula in principle for δ X w,p (C + , C − ) in general (see Theorem 5. 4), but to make this formula explicit involves more knowledge of the enumerative geometry of Hilb n X than seems to be available at present. In case −K X is effective, the moduli spaces are (essentially) smooth and the centers of the blowup are smooth as well; in fact they are P k -bundles over Hilb n1 X × Hilb n2 X for appropriate k, n 1 and n 2 . In this way, we obtain general formulas which can be made explicit for low values of n. For instance, we show the following (see Theorem 6.4 for details): Theorem. Assume that the wall W ζ is defined only by ±ζ with ℓ ζ = 1 and that C ± lies on the ±-side of W ζ . Then, on the subspace of the symmetric algebra generated by H 2 (X), δ X w,p (C − , C + ) is equal to Along the direction of the work of Kronheimer and Mrowka [19,20], we also consider the difference of Donaldson polynomial invariants involving the natural generator x ∈ H 0 (X; Z). More precisely, let ν be the corresponding 4-dimensional class in the instanton moduli space. For α ∈ H 2 (X; Z), we give a formula for the difference δ X w,p (C − , C + )(α d−2 , ν) in Theorem 5.5. It is worth to point out that the similarity between Theorem 5.4 and Theorem 5.5 may indicate that there exists a deep relation between δ X w,p (C − , C + )(α d ) and δ X w,p (C − , C + )(α d−2 , ν), and suggest a way to generalize the notion of simple type in [19,20] from the case of b + 2 > 1 to the case of b + 2 = 1. For instance, modulo some lower degree terms, δ X w,p (C − , C + )(α d−2 , ν) can be obtained from (−1/4) · δ X w,p (C − , C + )(α d ) by replacing d by (d − 2) (see Theorem 5. 13 and Theorem 5.14). In fact, based on some heuristic arguments, it seems reasonable to conjecture that δ X w,p (C − , C + )(α d−2 , ν) is a combination of δ X w,p ′ (C − , C + )(α d−4k ) for various nonnegative integers k if the degrees are properly arranged. We hope to return to this issue in future. Our paper is organized as follows. In section 2, we study rank two torsion free sheaves which are semistable with respect to ample divisors in C − but not semistable with respect to ample divisors in C + . When the surface X is rational with −K X effective, these sheaves are parametrized by an open subset of a union of projective bundles over the product of two Hilbert schemes of points in X. More precisely, if ζ defines the wall separating C − from C + , define E n1,n2 ζ be the set of all isomorphism classes of nonsplit extensions of the form where F is a divisor class such that 2F − ∆ ≡ ζ and Z 1 and Z 2 are two zerodimensional subschemes of X with ℓ(Z i ) = n i such that n 1 + n 2 = ℓ ζ . In case X is rational, E n1,n2 ζ is a P N bundle over Hilb n1 X × Hilb n2 X, and the set of points of The main technical difficulty is that it is hard to control the rational map from E n1,n2 ζ to M L− (∆, c), and in particular this map is not a morphism. The general picture that we establish is the following: first, the map E 0,ℓ ζ ζ M L− (∆, c) is a morphism, and it is possible to make an elementary transformation, or flip, along its image. The result is a new space for which the rational map E 1,ℓ ζ −1 ζ M L− (∆, c) becomes a morphism, and it is possible to make a flip along its image. We continue in this way until we reach M L+ (∆, c). It seems rather difficult to see that the above picture holds directly. Instead we shall proceed as follows. We define abstractly a sequence of moduli spaces, indexed by an integer k with 0 ≤ k ≤ ℓ ζ + 1, such that the moduli space for k = 0 is M L− (∆, c), the moduli space for k = ℓ ζ + 1 is M L+ (∆, c), and moreover the k th moduli space contains an embedded copy of E k,ℓ ζ −k ζ such that the flip along this copy yields the (k + 1) st moduli space. Thus the picture is very similar to that developed independently by Thaddeus in [31]. To define our sequence of moduli spaces, we define (L 0 , ζ, k)-semistability in section 3 for rank two torsion free sheaves, where L 0 is any ample divisor contained in the common face of C + and C − , ζ is the set of classes of type (∆, c) defining the common wall of C + and C − , and k is a set of integers. We show that M L− (∆, c) and M L+ (∆, c) are linked by the moduli spaces M (ζ,k) 0 where the data k is allowed to vary. When the surface X is rational with −K X effective, we can obtain M L+ (∆, c) from M L− (∆, c) by a series of flips. The fact that all (L 0 , ζ, k)-semistable rank two torsion free sheaves do form a moduli space M (ζ,k) 0 in the usual sense is proved in section 4 where we introduce an equivalent notion of stability called mixed stability. Our method follows Gieseker's GIT argument in [13]. Roughly speaking, the goal of mixed stability is to define stability for a sheaf of the form V ⊗ Ξ, where V is a torsion free sheaf but Ξ is just a Q-divisor. To make this idea precise, given actual divisors H 1 and H 2 and positive weights a 1 and a 2 , we shall define a notion of stability which "mixes" stability for V ⊗ H 1 with stability for V ⊗ H 2 , together with weightings of the stability condition for V ⊗ H i . The effect of this definition will be formally the same as if we had defined stability of V ⊗ Ξ, where Ξ is the Q-divisor In section 5, using our results on flips of moduli spaces, we give a formula for the transition formula of Donaldson polynomials when X is rational with −K X effective, and compute the leading term in the transition formula. In section 6, we obtain explicit transition formulas when ℓ ζ ≤ 2. Some of the material in our section 2 has been worked out independently by Hu and Li [16] and Göttsche [14]. Moreover Ellingsrud and Göttsche [8] have recently studied the change in the moduli space by similar methods and have obtained results very similar to ours. Using very different methods, the results in Section 4 have also been obtained by Matsuki and Wentworth [23], who also consider the case of higher rank. They use branched covers of the surface X to study the change in the moduli space. We expect that a minor modification of the arguments in Section 4 of this paper will also handle the case of higher rank. Conventions and notations We fix some conventions and notations for the rest of this paper. Let X be a smooth algebraic surface. We shall be primarily interested in the case where X is simply connected and −K X is effective and nonzero. Thus necessarily X is a rational surface. However much of the discussion in sections 1-4 will also apply to the general case. Stability and semistability with respect to an ample line bundle L will always be understood to mean Gieseker stability or semistability unless otherwise noted. We shall not mention the choice of L explicitly if it is clear from the context. Recall that a torsion free sheaf V of rank two is Gieseker L-stable if and only if, for every rank one subsheaf W of V , either µ L (W ) < µ L (V ) or µ L (W ) = µ L (V ) and 2χ(W ) < χ(V ), where µ L is the normalized degree with respect to L. Semistability is similarly defined, where the second inequality is also allowed to be an equality. For a torsion free sheaf V , we use V ∨∨ to stand for its double dual. For two divisors D 1 and D 2 on X, the notation D 1 ≡ D 2 means that D 1 and D 2 are numerically equivalent, that is, D 1 · D = D 2 · D for any divisor D. For a locally free sheaf (or equivalently a vector bundle) E over a smooth variety Y , we use P(E) to denote the associated projective space bundle, that is, Fix a divisor ∆ and an integer c. Let C − and C + be two adjacent chambers of type (∆, c) separated by the wall W ζ . We assume that ζ · C − < 0 < ζ · C + . Let L ± ∈ C ± be an ample line bundle, so that L − · ζ < 0 < L + · ζ, and denote by M ± the moduli space M L± (∆, c) of rank two Gieseker semistable torsion free sheaves V with c 1 (V ) = ∆ and c 2 (V ) = c. Let L 0 be any ample divisor contained in the interior of the intersection of W ζ and the closures of C ± . Let ζ = ζ 1 , . . . , ζ n be all the positive rational multiples of ζ such that ζ i is an integral class of type (w, p) which also defines the wall W ζ . In sections 5-6, we will assume that n = 1 for notational simplicity. Finally, we point out that our µ-map is half of the µ-map used in [17,18] (see (viii) and (ix) in Notation 5.1). Thus our transition formula differs from the one defined in [18] by a universal constant. Acknowledgements. We would like to thank Hong-Jie Yang for invaluable access to his calculations, which helped to keep us on the right track. The second author would like to thank Wei-ping Li and Yun-Gang Ye for helpful discussions, and the Institute for Advanced Study at Princeton for its hospitality and financial support through NSF grant DMS-9100383 during the academic year 1992-1993 when part of this work was done. Preliminaries on the moduli space. In this section, we study rank two torsion free sheaves which are related to walls. These sheaves arise naturally from the comparison of L − -semistability and L + -semistability. We will show that when the surface X is rational with −K X effective, the moduli spaces M ± are smooth at the points corresponding to these sheaves. We start with the following lemma, which for simplicity is just stated for L − -stability. Lemma 2.1. Let V be a rank two torsion free sheaf on X with c 1 (V ) = ∆ and c 2 (V ) = c. If V is L − -semistable, then exactly one of the following holds: where 2F 1 ≡ ∆ ≡ 2F 2 , and Z 1 and Z 2 are zero-dimensional subschemes of X such that ℓ(Z 1 ) ≥ ℓ(Z 2 ). Moreover in this case V is L-semistable for every choice of an ample line bundle L and V is strictly L ± -semistable if and only if ℓ(Z 1 ) = ℓ(Z 2 ). Proof. Suppose that V is (Gieseker) L − -semistable. The vector bundle V ∨∨ satisfies c 1 (V ∨∨ ) = ∆ and c 2 (V ∨∨ ) ≤ c. Standard arguments [10] show that V ∨∨ is Mumford L − -semistable. If V ∨∨ is strictly Mumford L − -semistable, then by [10,30], either L − must lie on a wall of type (∆, c) or if O X (F 1 ) is a destabilizing sub-line bundle then ∆ ≡ 2F 1 . Since by assumption L − does not lie on a wall of type (∆, c), either V ∨∨ is Mumford L − -stable or there is an exact sequence The last sentence of (ii) is a straightforward argument left to the reader. If V satisfies the conclusions of (2.1)(ii), we shall call V universally semistable. Next we shall compare stability for L − and L + . Lemma 2.2. Let V be a torsion free rank two sheaf with c 1 (V ) = ∆ and c 2 (V ) = c. (i) If V is L − -stable but L + -unstable, then there exist a divisor class F and two zero-dimensional subschemes Z − and Z + of X and an exact sequence . Moreover the divisor F , the schemes Z − and Z + , and the map F ⊗ I Z− → V are unique mod scalars, and ζ = 2F − ∆ defines a wall of type (∆, c). (ii) Conversely, suppose that there is a nonsplit exact sequence as above. Then V is simple. Moreover, V is not L − -stable if and only if it is L − -unstable if and only if there exist subschemes Z ′ and Z ′′ and an exact sequence Proof. We first show (i). Suppose that V is L − -stable but L + -unstable. Then by (2.1) V ∨∨ is also L − -stable and L + -unstable. By [30], there is a uniquely determined line bundle O X (F ) and a map O X (F ) → V ∨∨ with torsion free quotient such that agrees with it away from finitely many points. for some zero-dimensional subscheme Z, and agrees with O X (∆ − F ) away from finitely many points. Thus the quotient is of the form O X (∆ − F ) ⊗ I Z+ for some zero-dimensional subscheme Z + . The uniqueness is clear. To see (ii), suppose that V is given as a nonsplit exact sequence for some zero-dimensional scheme Z. Moreover, by [30], V ∨∨ is L − -unstable if and only if the above exact sequence splits, and in particular if and only if Z = ∅ and We may clearly assume that the quotient is torsion free, in which case it is necessarily of the form we see that there is an inclusion I Z ′ ⊆ I Z+ ; moreover this inclusion must be strict since the defining exact sequence for V is nonsplit. Thus Z ′ strictly contains Z + and in particular ℓ( There is an inclusion Hom(V, V ) ⊆ Hom(V ∨∨ , V ∨∨ ). If V ∨∨ is split, then Hom(V ∨∨ , V ∨∨ ) = C ⊕ C. In this case, using a nonscalar endomorphism of V , it is easy to see that we can split the exact sequence defining V . For the rest of this section, we shall assume that −K X is effective and nonzero and that q(X) = 0. Thus X is a rational surface. Proof. Suppose that M ± is nonempty, and let V correspond to a point of M ± . Then by general theory (e.g. Chapter 7 of [10]), M ± is smooth of dimension 4c−∆ 2 −3 = −p − 3 at V if V is stable and Ext 2 (V, V ) = 0, since h 2 (X; O X ) = 0. Moreover, setting W = V ∨∨ , there is a surjection from H 2 (X; Hom(W, W )) to Ext 2 (V, V ). Thus to show that Ext 2 (V, V ) = 0 it suffices to show that H 2 (X; Hom(W, W )) = 0. Now H 2 (X; Hom(W, W )) is dual to H 0 (X; Hom(W, W ) ⊗ K X ). Since −K X is effective, there is an inclusion of H 0 (X; Hom(W, W )⊗K X ) in H 0 (X; Hom(W, W )). If W is stable, then H 0 (X; Hom(W, W )) ∼ = C and H 0 (X; Hom(W, W ) ⊗ K X ) = 0. Thus M ± is smooth at V . Standard theory [1,10] also shows that every torsion free sheaf V for which V ∨∨ is stable is smoothable. Thus the set of locally free sheaves is nonempty and dense in the component containing V in this case. Now consider a V such that W = V ∨∨ is not stable. Using the exact sequence for W which was given in the course of the proof of (2.1), it is easy to check that there is an exact sequence Since −K X is effective and nonzero, Thus Hom(W, W ⊗ K X ) = 0 as well. Once again V is smoothable. Now we claim that a general smoothing V ′ of V is Mumford stable. For otherwise by the proof of (2.1) there is an exact sequence It is natural to make the following conjecture, which is true for geometrically ruled X by [29] and is verified in certain other cases by [34]. Let us fix some notations for the rest of this paper. Definition 2.5. Let X be an algebraic surface (not necessarily rational), and let ζ be a fixed numerical equivalence class defining a wall of type (∆, c). Set ℓ ζ = (4c − ∆ 2 + ζ 2 )/4 = (ζ 2 − p)/4. Choose two nonnegative integers n − and n + with n − + n + = ℓ ζ , and let E n−,n+ ζ be the set of all isomorphism classes of nonsplit extensions of the form We remark that since ζ ≡ ∆ (mod 2) and ∆ 2 − 4c ≤ ζ 2 < 0, ℓ ζ is a nonnegative integer. If V corresponds to a point of E n−,n+ ζ , then V is L + -unstable since L + ·ζ > 0. By (2.2)(ii), V is simple, and if it is L − -semistable then it is actually stable. By (2.3), if X is a rational surface with −K X effective, then M − is smooth in a neighborhood of a point corresponding to a sheaf V lying in E n−,n+ ζ for some ζ, n − , n + . We shall now study E n−,n+ ζ in more detail for rational surfaces. Lemma 2.6. Suppose that −K X is effective and that q(X) = 0. For Z − and Z + two fixed zero-dimensional subschemes of X of lengths n − and n + respectively, . Now a standard argument [27] shows that Here given a class a = a i ∈ i A i (X), we denote by a ∨ the class i (−1) i a i . An easy computation gives Reversing the above argument, we see that Putting these together we see that dim Let us describe the scheme structure on E n−,n+ ζ more carefully. For Z − and Z + fixed, the set of extensions in E n−,n+ ζ corresponding to Z − , Z + , is equal to To make a universal construction, let H n± = Hilb n± X. Let Z n± be the universal codimension two subscheme of X×H n± . Let π 1 , π 2 be the projections of X × H n− × H n+ to X, H n− × H n+ respectively, and let π 1,2 , π 1,3 be the projections of X × H n− × H n+ to X × H n− , X × H n+ respectively. Define The previous lemma and standard base change results show that E n−,n+ ζ is locally free of rank h(ζ)+ℓ ζ over H n− ×H n+ . We set E n−,n+ ζ = P((E n−,n+ ζ ) ∨ ), if h(ζ)+ℓ ζ > 0. Moreover by standard facts about relative Ext sheaves there is an exact sequence Finally this last case arises if and only if ζ 2 = p and ζ · K X = ζ 2 + 2 = p + 2. If h(ζ) + ℓ ζ = 0, then by Lemma 2.2 there is a rational map from E n−,n+ ζ to the moduli space M − which is birational onto its image. However this map will not in general be a morphism if n − > 0 (see [16]). We shall study this more carefully in the next sections. Let us also remark that standard theory gives a universal sheaf V over E n−,n+ ζ : Proposition 2.8. Let ρ : X × E n−,n+ ζ → X × H n− × H n+ be the natural projection, and let π 2 : X × E n−,n+ ζ → E n−,n+ ζ be the projection. Then there is a coherent sheaf V over X × E n−,n+ ζ and an exact sequence Remark 2.9. Very similar results hold in the case where −K X is effective and nonzero (corresponding to certain elliptic ruled surfaces) or K X = 0 (corresponding to K3 or abelian surfaces). For example, in the case of a K3 surface X, the moduli space is smooth of dimension −p − 6 away from the sheaves which are strictly semistable for every ample divisor (although there exist components consisting entirely of non-locally free sheaves for small values of −p). In this case however h(ζ) = −ζ 2 /2 − 2 and N ζ + N −ζ + 2ℓ ζ = −p − 6, which is equal to the dimension d of the moduli space instead of to d − 1. For example, if ℓ ζ = 0, then N ζ = N −ζ = d/2. In this case E 0,0 Flips of moduli spaces. In this section, we begin by assuming again that X is an arbitrary algebraic surface. Let ζ = ζ 1 , . . . , ζ n be the positive rational multiples of ζ such that ζ i is an integral class also defining the wall W ζ . Our goal in this section is to deal with the problem that there is only a rational map in general from E n−,n+ ζi to M − . We shall do so by finding a sequence of spaces between M − and M + , each one given by blowing up and down the previous one, such that for an appropriate member of the sequence the rational map E n−,n+ ζi M − becomes a morphism (and a smooth embedding in the case of rational surfaces). Throughout the rest of this paper, L 0 shall denote any ample divisor contained in the interior of the intersection of W ζ and the closures of C ± . Recall that we have defined universal semistability after the proof of (2.1). Proof. If k i ≥ ℓ ζi for all i, then the condition that ℓ(Z 2 ) ≤ ℓ ζi and ℓ(Z 1 ) ≥ 0 are trivially always satisfied and the conditions ℓ(Z 2 ) ≤ −1 and ℓ(Z 1 ) ≥ ℓ ζi + 1 are vacuous. A similar argument handles the case k i ≤ −1 for all i. It is easy to see that this implies (i). Statement (ii) follows from (i), and (iii) follows from the definitions. As for (iv), let V ∈ E , we look for potentially destabilizing subsheaves with torsion free quotient. Similar arguments as in [30] show that the only potentially destabilizing subsheaves with torsion free quotient must be either Next suppose that we are given two integral vectors k and k ′ and a subset I of follows by symmetry. We shall now describe a sequence of actual moduli spaces M (ζ,k) 0 for which the integral vector k change in the way described before the statement of (3.3). is not an integer for any i. In this case, define where [x] is the greatest integer function, and define k(t) to be the vector formed by We shall also say that t i is ζ i -critical. Finally t is ζ-critical if it is ζ i -critical for some i. Note that there are only finitely many such t. Then we clearly have: We then have the following theorem, whose proof will be given in the next section: for which it is a coarse moduli space. The proof of (3.5) will also show that M (ζ,k(t)) 0 has the usual properties of a coarse moduli space: all sheaves corresponding to points of M (ζ,k(t)) 0 will turn out to be simple (as they will turn out to be stable for an appropriate notion of stability), a classical or formal neighborhood of a point of M For the rest of this section, we shall again restrict to the case where X is a rational surface with −K X effective, unless otherwise noted. Let ζ = ζ i for some i and let k = k(t) for some t which is not ζ-critical. The first step is to make some infinitesimal calculations concerning the differential of the map E ℓ ζ −k,k ζ → M (ζ,k) 0 and the normal bundle to its image. is actually stable and therefore simple (which was also proved in (2.2)) we may identify an analytic neighborhood of V ∈ M (ζ,k) 0 with the germ of the universal deformation space for V , i.e. with Ext 1 (V, V ). Let us now calculate the tangent space to E Then there is the following exact sequence for the tangent space to E ℓ ζ −k,k ζ at ξ: Note further that the tangent space to Hilb n X at Z is equal to Hom(I Z , O Z ), which we may further canonically identify with Ext 1 (I Z , I Z ) since X is rational and by a local calculation. We then have the following: . We claim that this map is surjective and will describe its kernel in more detail. The map factors into two maps: The cokernel of the first map is contained in Ext . To see that this group is zero, apply Serre duality: it suffices to show that Hom(O X (F ) ⊗ I Z1 , V ⊗ K X ) = 0. From the defining exact sequence for V , we have an exact sequence The first term is just H 0 (K X ) = 0 and the third is contained in If K is the kernel, then arguments as above show that there is an exact sequence The last term is Ext 1 (I Z1 , I Z1 ) which is the tangent space to H ℓ ζ −k at Z 1 , and the first two terms combine to give Ext 1 /C · ξ. Thus the kernel K looks very much like the tangent space to E ℓ ζ −k,k ζ at ξ and both spaces have the same dimension. Let us describe the tangent space to E ℓ ζ −k,k ζ at ξ and the differential of the Conversely such a choice of Z 1 , Thus there is a commutative diagram with exact rows and columns: at ξ, use the arguments above which show that there is a surjection from K to necessarily a surjection. The kernel of this surjection then defines an extension It follows that K is in the image of the tangent space to E ℓ ζ −k,k ζ at ξ. By counting dimensions the map on tangent is an immersion and identifying the normal space at ξ. Let us continue the proof of Proposition 3.6. To give a global description of the normal bundle to E , recall by standard deformation theory [10] that the pullback of the tangent bundle of M is the second projection. Moreover the calculations above globalize to show that the normal bundle is exactly Using standard base change results and the projection formula, we see that this sheaf is equal to , we shall need the following result which is a straightforward generalization of (A.2) of [11]. Proposition 3.8. Let X be a smooth projective scheme or compact complex manifold, and let T be smooth. Suppose that V is a rank two reflexive sheaf over X × T , flat over T . Let D be a reduced divisor on T , not necessarily smooth and let i : D → T be the inclusion. Suppose that L is a line bundle on X and that Z is a codimension two subscheme of X × D, flat over D. Suppose further that V → i * π * 1 L ⊗ I Z is a surjection, and let V ′ be its kernel: Then there is a line bundle M on X and a subscheme Z ′ of X × D codimension at least two, flat over D, with the following properties: (i) V ′ is reflexive and flat over T . (ii) There are exact sequences which restrict for each t ∈ D to give exact sequences Here Z is the subscheme of X defined by Z for the slice X × {t} and Z ′ t is likewise defined by Z ′ . (iii) If D is smooth, then the extension class corresponding to is defined by the image of the normal vector to D at t under the composition of the Kodaira-Spencer map from the tangent space of T at t to Here V ′ is called the elementary modification of V along D. This construction has the following symmetry: if we make the elementary modification of V ′ along D corresponding to the surjection Here is the typical way that we will apply the above: given X, let M be a smooth manifold and Y a submanifold of M . Let T be the blowup of M along Y and let D be the exceptional divisor. Let π : T → M be the natural map. Then, given ξ ∈ D, the image in the normal space to π(ξ) of the normal direction at ξ to D under π * may be identified with the line in the normal space corresponding to ξ. We can now state the main result as follows: is obtained as follows. For every i, fixing ζ i = ζ and . Then the exceptional divisor D is a P N ζ × P N −ζ -bundle over Hilb ℓ ζ −k X × Hilb k X. Moreover this divisor can be contracted in two different ways. Contracting the P N −ζ fibers for all possible ζ gives M (ζ,k(t+ε)) 0 . Contracting the P N ζ fibers for all possible ζ gives M Finally the construction is symmetric. Similar statements hold if h(±ζ i ) + ℓ ±ζi = 0 for some i, where we must also add in or delete an extra component coming from ±ζ i . Proof. Begin by blowing up for all possible ζ. For simplicity we shall just write down the argument in case there is only one ζ; the general case is just additional notation. Let M (ζ,k(t+ε)) 0 denote the blowup and D the exceptional divisor. Note that the normal bundle N , and an easy calculation using (2.7) shows that N = N −ζ + 1. It follows that the fibers of the induced map from D to Hilb for some a and the fact that For the rest of the argument, we assume that there exists a universal family on X × M (ζ,k(t+ε)) 0 . Of course, such a family will only exist locally in the classical orétale topology, but this will suffice for the argument. Let U be the pullback of the universal family to X × M (ζ,k(t+ε)) 0 . Locally again we may assume that the restriction of U to X × D is the pullback of the universal extension V of (2.8): Now consider the effect of making an elementary transformation of U on X × M (ζ,k(t+ε)) 0 along the divisor D, using the morphism from U to the pullback of given by considering the pullback of the universal extension. Applying (3.8) to the elementary transformation U ′ , we see that the fiber of U ′ at a point of the fiber P N ζ × P N −ζ lying over a point (Z 1 , Z 2 ) ∈ Hilb ℓ ζ −k X × Hilb k X is given by a nonsplit extension of the form Moreover the extension class corresponding to U is given by the projectivized normal vector in P N −ζ . Thus it is independent of the first factor P N ζ and the set of all possible such classes is parametrized by the second factor P N −ζ . There is then an induced morphism from M (ζ,k(t+ε)) 0 to M (ζ,k(t−ε)) 0 and clearly it has the effect of contracting D along its first ruling and has the property that the image of D is exactly E We leave the symmetry of the construction to the reader. This concludes the proof of (3.9). Remark 3.10. In the K3 or abelian case, the arguments of this section show that the rational map M is a Mukai elementary transformation [26,28]. We can also use (3.8) to analyze the rational map from E n−,n+ ζ to M − , in the case where it is not a morphism. For simplicity we shall only consider the case of Moreover, for p fixed, the extensions V corresponding to a split extension for V ∨∨ are exactly the kernel of the map from This way of realizing V as an extension gives a surjection , and we must look at the image of the normal space H 1 (O X (2F − ∆)) in this extension group. On the other hand, we have an exact sequence coming from the long exact Ext sequence, and it is an easy diagram chase to see that the induced map Ext The above has the following geometric interpretation: the locus U in E 1,0 ζ of L − -unstable sheaves is in fact a section of E 1,0 ζ . If we blow up this section and then make the elementary transformation, the result is exactly the set of elements of E 0,1 ζ corresponding to nonlocally free sheaves. This set is already a divisor in E 0,1 ζ . There is thus a morphism from the blowup of E 1,0 exactly along the divisor in E 0,1 ζ of nonlocally free sheaves. We can now give a picture of the birational map from M − to M + in this case. Begin with the subvariety E 0,1 ζ in M − and blow it up. Let D 0,1 be the exceptional divisor, ruled in two different ways. As E 0,1 ζ meets (E 1,0 ζ ) ′ along a divisor, the proper transform of (E 1,0 ζ ) ′ in the blowup is again (E 1,0 ζ ) ′ . Making the elementary modification along D 0,1 , we then blow down D 0,1 to get a new moduli space. This moduli space then contains E 1,0 ζ . At this point we can then blow up E 1,0 ζ and contract the new exceptional divisor D 1,0 to obtain M + (a few extra details need to be checked here concerning the Kodaira-Spencer class). Note again the symmetry of the situation. In principle we could hope to carry through this analysis to the case where ℓ ζ > 1 as well, but we run into trouble with the birational geometry of Hilb n X. Somehow the construction of our auxiliary sequence of moduli spaces has eliminated the necessity for understanding this birational geometry in detail. Mixed stability and mixed moduli spaces. Our goal in this section is to give a proof of Theorem 3.5 (for an arbitrary algebraic surface X). By way of motivation for our construction, let us analyze Gieseker semistability more closely. In the notation of the last section, we suppose that L 0 is an ample line bundle lying on a unique wall W of type (w, p), and let ζ 1 , . . . , ζ n be the integral classes of type (w, p) defining W . Let V be an L 0semistable rank two sheaf. Thus either V is Mumford L 0 -stable or it is Mumford strictly semistable. In the second case, let O X (F ) ⊗ I Z1 be a destabilizing subsheaf and suppose that there is an exact sequence Let ζ = 2F − ∆. We shall assume that ζ = ζ i for some i, or equivalently that ζ is not numerically equivalent to zero (i.e., V is not universally semistable). By , we may rewrite this last condition as Now from the exact sequence . By Riemann-Roch, Thus we have the following conditions on Z 1 and Z 2 : and so 2ℓ(Z 2 ) ≤ ℓ ζ + t. Setting k = ℓ ζ + t 2 , we have ℓ(Z 2 ) ≤ k. Applying a similar analysis to a subsheaf of the form O X (∆ − F ) ⊗ I Z1 shows that, if there is such a subsheaf, with a torsion free quotient O X (F ) ⊗ I Z2 , then In particular, if ℓ ζ + t 2 is not an integer, then this condition becomes ℓ(Z 2 ) ≤ ℓ ζ − k − 1. Thus, provided ℓ ζ + t 2 is not an integer for every ζ defining the wall W (i.e. t is not ζ-critical for every ζ), V is (L 0 , ζ, k)-semistable for k = ℓ ζ + t 2 and indeed V is (L 0 , ζ, k)-semistable, where k is defined in the obvious way. Conversely, assuming that t is not ζ-critical for every ζ, V is Gieseker L 0 -semistable, indeed Gieseker L 0 -stable, if it is (L 0 , ζ, k)-semistable for k as above. We would like to produce a similar condition where t is allowed to be any rational number which is not ζ-critical. One way to think of this problem is to consider the analogous problem where we replace ∆ by ∆ + 2Ξ and make the corresponding change in c, so that ∆ and p remain the same. This corresponds to twisting V by O X (Ξ), and t is replaced by t − ζ · Ξ. In particular, we see that the notion of Gieseker stability is rather sensitive to twisting by a line bundle. Moreover if W is defined by exactly one ζ such that there exists a divisor Ξ with ζ · Ξ = 1, for example if ζ is primitive and p g (X) = 0, it is easy to see that we can construct the appropriate moduli spaces as Gieseker moduli spaces corresponding to twists of V by various multiples of Ξ. In general however we will need to consider a problem which is roughly analogous to allowing twists of V by a Q-divisor Ξ. This is the goal of the following definition of mixed stability: A torsion free sheaf V is (H 1 , H 2 , a 1 , a 2 ) L 0 -stable if, for all subsheaves W of V with 0 < rank W < rank V and for all n ≫ 0, p V ;H1,H2,a1,a2 (n) > p W ;H1,H2,a1,a2 (n). The usual arguments show the following: In the case of rank two on a surface X (which is the only case which shall concern us), V is (H 1 , H 2 , a 1 , a 2 ) L 0 -stable if and only if, for all rank one subsheaves W , and for all n ≫ 0, we have In particular, if V is (H 1 , H 2 , a 1 , a 2 ) L 0 -stable then either V ⊗ H 1 or V ⊗ H 2 is stable, and a similar statement holds for semistability. A short calculation shows that the coefficient of n in the above expression (which is a degree two polynomial in n) is (a 1 + a 2 )(L 0 · (c 1 (V ) − 2F )) and that the constant term is Thus V is (H 1 , H 2 , a 1 , a 2 ) L 0 -stable (resp. semistable) if and only if it is either Mumford L 0 -stable or Mumford strictly semistable and the above constant term is positive (resp. nonnegative). It is easy to see, comparing this with the discussion at the beginning of this section, that formally this is the same as requiring that V ⊗ Ξ is (Gieseker) L 0 -stable or semistable, where Ξ is the Q-divisor Thus for example taking H 2 = 0 and replacing H 1 by a positive integer multiple we see that we can take for Ξ an arbitrary Q-divisor. Let us explicitly relate mixed stability to our previous notion of (L 0 , ζ, k)semistability: Lemma 4.3. Given ∆ and c and the corresponding w and p, let L 0 be an ample divisor lying on a unique wall of type (w, p) and let V be a rank two torsion free sheaf with c 1 (V ) = ∆ and c 2 (V ) = c. Let Ξ be the Q-divisor a 1 a 1 + a 2 H 1 + a 2 a 1 + a 2 H 2 and suppose that the rational number Proof. Using the additivity of the polynomials p V ;H1,H2,a1,a2 over exact sequences, it is easy to check that V is (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable if and only if it is Mumford L 0 -semistable, and for every Mumford destabilizing subsheaf of the form O X (F ) ⊗ I Z1 , either V is universally semistable or we have where ζ i = 2F − ∆. Using our calculations above, this works out to Equivalently since ℓ(Z 1 ) + ℓ(Z 2 ) = ℓ ζi , this becomes ℓ(Z 2 ) ≤ ℓ ζi + t i 2 . Thus V is (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable if and only if it is (L 0 , ζ, k(t))-semistable. Moreover, since t is not ζ i -critical, the inequalities are automatically strict, so that V is also (H 1 , H 2 , a 1 , a 2 ) L 0 -stable. Now choosing a Ξ 0 such that ζ 1 · Ξ 0 = 0, every rational number t is of the form 1 2 ζ 1 · (K X − ∆) − ζ 1 · rΞ 0 for some rational number r. Thus Theorem 3.5 will follow from Lemma 4.3 and from the more general result below: Proof. The argument will follow the arguments in [13] as closely as possible, and we shall assume a familiarity with that paper. Suppose that V is (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable. Then either V ⊗ H 1 or V ⊗ H 2 is L 0 -semistable, and thus by [13], Lemma 1.3 the set of all such V is bounded. We may thus choose an n such that, for all V which are (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable, V ⊗ H i ⊗ L n 0 is generated by its global sections and has no higher cohomology, for i = 1, 2. Fix such an n for the moment, and let d i = h 0 (V ⊗ H i ⊗ L n 0 ). Then d i is independent of V and V is a quotient of (H −1 i ⊗ L −n 0 ) ⊕di . Let Q i be the open subset of the corresponding Quot scheme associated to (H −1 i ⊗ L −n 0 ) ⊕di consisting of quotients which are rank two torsion free sheaves V i with c 1 (V i ) = ∆ and c 2 (V i ) = c, and such that V i ⊗ H i ⊗ L n 0 is generated by its global sections and has no higher cohomology. We will write a point of Q i as V i , suppressing the we have the closed subscheme I 0 consisting of quotients V 1 and V 2 such that dim Hom(V 1 , V 2 ) ≥ 1. There is also the open subvariety I ′ 0 of I 0 consisting of (V 1 , V 2 ) with dim Hom(V 1 , V 2 ) = 1. Using the universal sheaves U i over X × Q i , we can construct a C * bundle I over I ′ 0 whose points are (V 1 , V 2 , ϕ), where ϕ : V 1 → V 2 is a nonzero homomorphism, unique up to scalars. For i = 1, 2, let E i be a fixed vector space of dimension equal to d i = h 0 (V ⊗H i ⊗ L n 0 ). Fix once and for all an isomorphism (H −1 ) and via such a surjection a basis v 1 , . . . , v d1 of E 1 gives d 1 sections of V 1 ⊗ H 1 ⊗ L n 0 and similarly for a basis w 1 , . . . , w d2 of E 2 . Moreover GL(d i ) acts on (H −1 i ⊗ L −n 0 ) ⊕di and on Q i . By the universal property of the Quot scheme, this action extends to a GL(d i )-linearization of the universal sheaf U i over X × Q i . Thus there is a right action of GL(d 1 ) × GL(d 2 ) on I, and it is easy to see that the elements (λ Id, λ Id) act trivially. Let F i be the fixed vector space (The factor Hom(E 1 ⊗E 2 , F ) is there to make sure that the destabilizing subsheaves for V ⊗ H 1 and V ⊗ H 2 are in fact the same.) Note that GL(d 1 ) × GL(d 2 ) operates on the right on U and PU . For example, the pair (λ Id, µ Id) acts on the triple (T 1 , T 2 , T ) ∈ U via (T 1 , T 2 , T ) → (λ 2 T 1 , µ 2 T 2 , λµT ). Thus (A 1 , A 2 ) acts trivially on PU if and only if (A 1 , A 2 ) = (λ Id, λ Id). Given a quintuple V = (V 1 , V 2 , ψ 1 , ψ 2 , ϕ), is an isomorphism, and ϕ : V 1 → V 2 is a nonzero map, we will define a point (T 1 (V ), T 2 (V ), T (V )) ∈ PU . To do so, fix an isomorphism α 2 : det V 2 → O X (∆), and set α 1 = α 2 • det ϕ. (Thus α 1 = 0 if ϕ is not an isomorphism.) Given v, v ′ ∈ E 1 and w, w ′ ∈ E 2 , identify v, v ′ with their images in H 0 (V i ⊗ H 1 ⊗ L n 0 ) and similarly for w, w ′ , and let Changing α 2 by a nonzero scalar λ multiplies (T 1 (V ), T 2 (V ), T (V )) by λ, so that the induced element of PU is well defined. Similarly, if we replace ϕ by λϕ, then (T 1 (V ), T 2 (V ), T (V )) is replaced by (λ 2 T 1 (V ), T 2 (V ), λT (V )). It is easy to check that the map V → T (V ) induces a morphism from I to PU which is GL(d 1 ) × GL(d 2 )-equivariant. Further note that we can define (T 1 (V ), T 2 (V ), T (V )) more generally if we are given the data V of two rank two torsion free sheaves V 1 and V 2 with det V i = ∆, a morphism ϕ : V 1 → V 2 , and linear maps ψ i : , not necessarily isomorphisms, although it is possible for (T 1 (V ), T 2 (V ), T (V )) to be zero in this case. We have not yet introduced the extra parameters a 1 and a 2 . To do so, define G(a 1 , a 2 ) ⊂ GL(d 1 ) × GL(d 2 ) as follows: Thus unlike Thaddeus we don't change the polarization or the linearization but the actual group which we use to determine stability; still our construction could probably be interpreted in his general framework. Fixing a 1 and a 2 for the rest of the discussion, we shall denote G(a 1 , a 2 ) by G. Since a 1 and a 2 are positive, the matrix (λ Id, λ Id) lies in G if and only if λ is an m th root of unity, where m = a 1 d 1 + a 2 d 2 . Thus a quotient of G by a finite group acts faithfully on PU . Moreover, the problem of finding a good quotient of PU (for an appropriate open subset of PU ) for G is the same as that of finding a good quotient of PU for This last statement follows since G clearly contains SL(d 1 ) × SL(d 2 ) and since C * × C * is generated by its diagonal subgroup and by the subgroup We may thus apply the general machinery of GIT to the group G acting on PU . A one parameter subgroup of G is given by a basis {v i } of E 1 , a basis {w k } of E 2 and weights n i , m k ∈ Z, such that v λ i = λ ni v i , w λ k = λ m k w k , and We shall always arrange our choice of basis so that n 1 ≤ n 2 ≤ · · · ≤ n d1 and m 1 ≤ m 2 ≤ · · · ≤ m d2 . Given (T 1 , T 2 , T ) ∈ U and a one parameter subgroup of G as above, we see that lim λ→0 (T 1 , T 2 , T ) λ = 0 if and only if T 1 (v i ∧ v j ) = 0 for every pair of indices i, j such that n i +n j ≤ 0, T 2 (w k ∧w ℓ ) = 0 for every pair of indices k, ℓ such that m k + m ℓ ≤ 0, and T (v i ⊗ w j ) = 0 for every pair i, k such that n i + m k ≤ 0. Likewise the condition that lim λ→0 (T 1 , T 2 , T ) λ exists is similar, replacing the ≤ by strict inequality. Finally note that if n i + n j ≤ 0, then n 1 + n j ≤ 0, if m k + m ℓ ≤ 0 then m 1 + m ℓ ≤ 0, and if n i + m k ≤ 0 then n 1 + m k ≤ 0 and n i + m 1 ≤ 0. We then have the following: Lemma 4.5. (i) Suppose that we are given the data V of two rank two torsion free sheaves V 1 and V 2 with det V i = ∆, a morphism ϕ : V 1 → V 2 , and a linear map is not injective for some i or if ϕ is not an isomorphism, then (T 1 (V ), T 2 (V ), T (V )) is either zero or G-unstable. (ii) For n sufficiently large depending only on ∆ and c and for V a rank two torsion free sheaf with det V = ∆ and c 2 (V ) = c, V is (H 1 , H 2 , a 1 , a 2 (H 1 , H 2 , a 1 , a 2 ) L 0 -strictly semistable if and only if (T 1 (V ), T 2 (V ), T (V )) is G-strictly semistable for all such V . Thus V is (H 1 , H 2 , a 1 , a 2 Proof. First let us prove (i). We may assume that ( Complete v 1 to a basis of E 1 and choose a basis {w k } for E 2 . Then T 1 (V )(v 1 ∧ v i ) = 0 for all i and T (V )(v 1 ⊗ w k ) = 0 for all k. Define a one parameter subgroup of G as follows: , T (V )) λ = 0 provided that a and b are positive, so that (T 1 (V ), T 2 (V ), T (V )) is G-unstable provided that the one parameter subgroup so constructed lies in G, or on other words provided that a 1 (−N + a(d 1 − 1)) + a 2 bd 2 = 0. It thus suffices to take a an arbitrary positive integer, b = a 1 , and N = a(d 1 − 1) + a 2 d 2 . The argument in case ϕ has a kernel is similar: in this case let v 1 ∈ Ker ϕ. Then T 1 (V ) = 0 and T (V )(v 1 ⊗ w k ) = 0 for all k, so that the previous argument handles this case also. Next we show (ii). Let p V ⊗Hi be the usual normalized Hilbert polynomial of V ⊗ H i , and similarly for p W ⊗Hi , where W is a rank one subsheaf of V . Thus p V ⊗Hi and p W ⊗Hi have the same leading term. Given a polynomial p, let ∆p denote the difference polynomial. In our case, all of the polynomials p that occur are quadratic polynomials with the same fixed degree two term. Thus if p 1 and p 2 are two such polynomials, then p 1 (n) > p 2 (n) for all n ≫ 0 if and only if the linear term of p 1 is greater than or equal to the linear term of p 2 , and if the linear terms are equal then the constant term of p 1 is greater than the constant term of p 2 . In this last case, where the linear terms are also equal, we see that p 1 (n) > p 2 (n) for all n ≫ 0 if and only if p 1 (n) > p 2 (n) for some n. Finally the linear term of p 1 is greater than or equal to the linear term of p 2 if and only if ∆p 1 (n) ≥ ∆p 2 (n) for all n, which we shall write as ∆p 1 ≥ ∆p 2 . Thus if ∆p 1 ≥ ∆p 2 and p 1 (n) > p 2 (n) for some n, then p 1 (n) > p 2 (n) for all n ≫ 0. If ∆p 1 = ∆p 2 , then p 1 (n) > p 2 (n) for some n if and only if p 1 (n) > p 2 (n) for all n. We shall show that, for sufficiently large n, if V is (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable and V corresponds to data where E i → H 0 (V ⊗ H i ⊗ L n 0 ) is injective and ϕ is an isomorphism, then (T 1 (V ), T 2 (V ), T (V )) is G-semistable. Note that V is Mumford semistable. First we may choose n so that V ⊗ H i is generated by its global sections and has no higher cohomology, and so is injective, it is an isomorphism. Let W be a rank one subsheaf of V . Since V ⊗ H i is Mumford semistable, ∆p W ⊗Hi ≤ ∆p V ⊗Hi . Now the proof of (3) of Lemma 1.2 in [13] shows that there exists an N so that, for all n ≥ N , with d i as above, if W is a rank one subsheaf of V and such that h 0 (W ⊗ H i ⊗ L n 0 ) ≥ d i /2 for at least one i (i = 1, 2), then in fact ∆p V ⊗Hi = ∆p W ⊗Hi for all such W , and thus µ L0 (V ) = µ L0 (W ). It is then easy to see that there is a twist W ⊗ H i ⊗ L −k 0 , depending only on L 0 and ∆, such that The proof of Proposition 3.1 in [13] shows that in where Q is some universal bound for the numbers h 1 (V ⊗ H i ⊗ L −k 0 ) as V ⊗ H i ranges over the appropriate set of L 0semistable sheaves Thus by (4) of Lemma 1.2 in [13], the W satisfying the condition that h 0 (W ⊗ H i ⊗ L n 0 ) ≥ d i /2 for at least one i form a bounded family, and we may choose n so large, depending only on L 0 , ∆, c, such that h j (W ⊗ H i ⊗ L n 0 ) = 0 for j ≥ 1 and i = 1, 2. Now suppose that ( Then there exists a one parameter subgroup of G as above such that lim λ→0 (T 1 (V ), T 2 (V ), T (V )) λ = 0. Let Since a 1 i n i + a 2 k m k = 0, at least one of n 1 , m 1 is negative. By symmetry we may assume that n 1 is negative, and that n 1 ≤ m 1 . Since for j ≤ s 1 , v 1 ∧ v j is zero as a section of det(V ⊗ H 1 ⊗ L n 0 ), the sections corresponding to v j , 1 ≤ j ≤ s 1 , are all sections of a rank one subsheaf W 1 of V . Likewise the sections w ℓ , 1 ≤ ℓ ≤ s 2 , if there are any such, are all sections of a rank one subsheaf W 2 of V . The condition that T (V )(v 1 ⊗ w 1 ) = 0 insures that W 1 and W 2 are contained in a saturated rank one subsheaf W , if s 2 = 0, otherwise we shall just take for W the saturated rank one subsheaf containing W 1 . Moreover h 0 (W ⊗H 1 ⊗L n 0 ) ≥ s 1 and h 0 (W ⊗H 2 ⊗L n 0 ) ≥ s 2 . Suppose that we show that Thus in particular s i ≥ d i /2 for at least one i. By our choice of n and the previous paragraph, if s i ≥ d i /2 for at least one i, then h 0 (W ⊗ H i ⊗ L n 0 ) = χ(W ⊗ H i ⊗ L n 0 ) and furthermore µ L0 (V ) = µ L0 (W ). Thus for i = 1, 2 and so p V ;H1,H2,a1,a2 (n) < p W ;H1,H2,a1,a2 (n). On the other hand, p V ;H1,H2,a1,a2 and p W ;H1,H2,a1,a2 are two quadratic polynomials with the same linear and quadratic terms (since µ L0 (V ) = µ L0 (W )), and p V ;H1,H2,a1,a2 (n) < p W ;H1,H2,a1,a2 (n) for one value of n. Thus the constant term of p W ;H1,H2,a1,a2 must be larger than that of p V ;H1,H2,a1,a2 . This contradicts the (H 1 , H 2 , a 1 , a 2 ) Here t 1 ≤ s 1 since T (V )(v j ⊗ w 1 ) = 0 implies that v j and w 1 are contained in a rank one subsheaf of V , necessarily W , and thus that v 1 ∧ v j = 0. Let We have assumed that n 1 ≤ m 1 . Then consider the expression a 1 j (n 1 + n j ) + a 2 ℓ (n 1 + m ℓ ). On the one hand from the definition of the one parameter subgroup we have On the other hand, to estimate j (n 1 +n j ), we can ignore the positive terms where n 1 + n j ≥ 0 and each of the s 1 negative terms are at least n 1 + n 1 ≥ 2n 1 . Thus j (n 1 + n j ) ≥ 2s 1 n 1 . Since n 1 < 0, this term is ≥ 2s 1 n 1 . Also this inequality is strict or n 1 + n i ≤ 0 for every i, which would say that every section of V ⊗ H 1 ⊗ L n 0 is really a section of W ⊗ H 1 ⊗ L n 0 contradicting the fact that V ⊗ H 1 ⊗ L n 0 is generated by global sections. So j (n 1 + n j ) < 2s 1 n 1 . Likewise we claim that ℓ (n 1 + m ℓ ) ≥ 2s 2 n 1 . Here, to estimate ℓ (n 1 + m ℓ ), we may ignore the terms with n 1 + m ℓ positive, leaving t 2 terms n 1 + m ℓ which are ≤ 0, and moreover each such term is at least n 1 + m 1 ≥ 2n 1 . Thus ℓ (n 1 + m ℓ ) ≥ 2t 2 n 1 , and since t 2 ≤ s 2 and n 1 < 0, we have 2t 2 n 1 ≥ 2s 2 n 1 . We have thus shown that, if (T 1 (V ), T 2 (V ), T (V )) is G-unstable, then V is (H 1 , H 2 , a 1 , a 2 ) L 0 -unstable. A very similar argument handles the G-strictly semistable case. Now we turn to the converse statement, that if V is (H 1 , H 2 , a 1 , a 2 ) L 0 -unstable then (T 1 (V ), T 2 (V ), T (V )) is G-unstable. Suppose instead that is G-semistable. Let W be a rank one subsheaf of V such that p W ;H1,H2,a1,a2 (m) > p V ;H1,H2,a1,a2 (m) for all m ≫ 0. We may assume that the quotient W ′ = V /W is torsion free. Thus p W ′ ;H1,H2,a1,a2 (m) < p V ;H1,H2,a1,a2 (m) for all m ≫ 0, and so for i ≤ s 1 , and similarly choose a basis w 1 , . . . , w d2 for We will try to find a one parameter subgroup of G of the form and similarly It is easy to check that lim λ→0 (T 1 (V ), T 2 (V ), T (V )) λ = 0 if and only if n > m. What we must arrange is the condition Now consider the linear function with rational coefficients Since the coefficient of t is strictly positive f (t) is increasing, and Thus there is a rational number t = n/m > 1 such that f (t) = 0, and this gives the desired choice of n and m. Thus if The other possibility is that a 1 (d 1 − 2s 1 ) + a 2 (d 2 − 2s 2 ) ≥ 0. In this case d i ≥ 2s i for at least one i. Recalling that we have the quotient W ′ of V by W , it then follows that for such an i the image of E i in H 0 (W ′ ⊗ H i ⊗ L n 0 ) must have dimension at least d i /2. Arguing as in Proposition 3.2 of [13], it then follows from Lemma 1.2 of [13] that ∆p W ′ ⊗Hi = ∆p V ⊗Hi and so that V is Mumford L 0 -semistable and µ L0 (V ) = µ L0 (W ). Moreover, after enlarging n if necessary (independently of V ) we may assume that h j (V ⊗ H i ⊗ L n 0 ) = 0 for j > 0. In , the polynomials p W ;H1,H2,a1,a2 and p V ;H1,H2,a1,a2 have the same terms in degree one and two, and thus since p W ;H1,H2,a1,a2 (m) > p V ;H1,H2,a1,a2 (m) for some m the same is true for all m, in particular for m = n. Moreover, for a general choice of a smooth curve C in the linear system corresponding to L 0 , there is a fixed bound on the line bundle W ⊗ H i \|C. A standard argument as in the proof of (2) of Lemma 1.2 of [13] shows that, for n sufficiently large but independent of V , we have = 2(p V ;H1,H2,a1,a2 (n) − p W ;H1,H2,a1,a2 (n)) < 0. This contradicts the assumption that a 1 (d 1 −2s 1 )+a 2 (d 2 −2s 2 ) ≥ 0. It then follows that ( The strictly semistable case is similar. We may now finish the proof of Theorem 4.4. Let PU ss be the set of G-semistable points of PU . Let I ss be the inverse image of PU ss under the morphism I → PU . Since every semistable sheaf is stable, I ss is a C * -bundle over its image in Q 1 × Q 2 . Moreover the representable functor corresponding to I ss is easily seen to be formally smooth over the moduli functor. Arguments very similar to those for Lemma 4.3 and 4.5 of [13] show that the morphism I ss → PU ss is one-toone and proper, and thus in particular finite. Thus we may construct a quotient M L0 (∆, c; H 1 , H 2 , a 1 , a 2 ) of I ss by G. This quotient maps in a one-to-one and proper way to the GIT quotient of PU ss and is therefore projective. By the discussion at the beginning of the proof of Theorem 4.4 the points of M L0 (∆, c; H 1 , H 2 , a 1 , a 2 ) may be identified with isomorphism classes of (H 1 , H 2 , a 1 , a 2 ) L 0 -semistable rank two sheaves. Standard arguments then show that M L0 (∆, c; H 1 , H 2 , a 1 , a 2 ) has the usual properties of a coarse moduli space. The transition formula for Donaldson polynomial invariants. From now on, we will assume that the surface X is rational with −K X effective, and will study the transition formula of Donaldson polynomial invariants: where C − and C + are two adjacent chambers separated by a single wall W ζ of type (w, p) or equivalently of type (∆, c). For simplicity, we assume that the wall W ζ is only represented by ±ζ since the general case just involves additional notation. We use M . When ℓ ζ = 0, we also assume that (see Corollary 2.7). The special case when ℓ ζ = h(ζ) = 0 will be treated in Theorem 6.1. By Theorem 3.9 and Lemma 3.2 (ii), we have the following diagram: Next, we collect and establish some notations. Recall that in section 2 we have constructed the bundle E Notation 5.1. Let ζ define a wall of type (w, p). (i) λ k is the tautological line bundle over E ℓ ζ −k,k ζ = P((E ℓ ζ −k,k ζ ) ∨ ); for simplicity, we also use λ k to denote its first Chern class; (−D k )\|D k is the tautological line bundle on D k ; again, for simplicity, we also use ξ k to denote its first Chern class; is the natural generator. Let ν (ℓ ζ ) = ν − and that ν (−1) = ν + . Note that, in (viii) and (ix) above, the sheaf U (k) is only defined locally in the classical topology. However, since it is defined on the level of the Quot scheme a straightforward argument shows that p 1 (U (k) ) is a well-defined element in the rational cohomology of X × M (k) 0 , at least in the complement of the universally semistable sheaves. In case there are universally semistable sheaves, then the work of Li [21] extends the µ-map to M (k) 0 , at least for the two-dimensional algebraic classes. We can then extend the µ-map to the 4-dimensional class via a blowup formula due to O'Grady (unpublished). Moreover, there is a universal sheaf V k over X × E In what follows, we shall work as if there were a universal sheaf U (k) , and leave it to the reader to check that our final Chern class calculations can be verified directly even when no universal sheaf exists. Lemma 5.3. For α ∈ H 2 (X; Z) and the natural generator x ∈ H 0 (X; Z), we have Proof. From the construction, the sheaf (Id 0 is the elementary modification of (Id ×p k ) * U (k) along the divisor X×D k , using the surjection is a sheaf supported on X × D k , and that its first and second Chern classes are equal to (X ×D k ) and (X ×D 2 k )−π * 1 (∆− F ) · (X × D k ) respectively. It follows that . Now the conclusions follow from some straightforward calculations. In the next two theorems, we will give formulas for the differences [µ terms of the intersections in H ℓ ζ −k × H k and the Segre classes of the vector bundles E where k = 0, 1, . . . , ℓ ζ . The arguments are a little complicated, but the idea is that we are trying to get rid of the exceptional divisors D k as well as the Chern classes of the tautological line bundles ξ k and λ k . Proof. For simplicity, on the exceptional divisor D k , we put So we must compute σ d−1−j . Notice the relation Thus for 0 ≤ t ≤ s, we have is a hyperplane. Therefore, Now, we shall simplify (ξ k −λ k ) (d−1−j)−i . Since ξ k is the tautological line bundle on D k = P(N ∨ k ), the line bundle (ξ k ⊗ λ −1 k ) is the tautological line bundle on One verifies that in general, for u ′ ≥ N −ζ , one has , we see as before that Putting all these together, we conclude that σ d−1−j is equal to This completes the proof of the Theorem. Remark 5.6. For the sake of convenience, we record here the following relation among the Chern classes and the Segre classes of a vector bundle: with the convention that s 0 = 1. We refer to [12] for details. In the next section, using Theorem 5.4 and Theorem 5.5, we shall compute In principle, Theorem 5.4 and Theorem 5.5 give formulas for these differences in terms of certain intersections in H ℓ ζ −k × H k . However, it is difficult to evaluate these intersection numbers in general. In the following, we shall compute the term in the special cases when j = 2ℓ ζ and 2ℓ ζ − 1. We start with a simple lemma. Lemma 5.8. Let α, β ∈ H 2 (X; Z). Then Proof. The first equality is well-known (see [28] for instance). The other statements follow from the first one by considering and formally equating the terms involving (2k − 1) copies of α and one β or (2k − 2) copies of α and two copies of β. The next result computes the term (5.7) when j = 2ℓ ζ . Proposition 5.9. Let ζ define a wall of type (w, p), and α ∈ H 2 (X; Z). Then, Proof. This follows in a straightforward way from Lemma 5.8 (i): To compute the term (5.7) when j = (2ℓ ζ − 1), we study E k,ℓ ζ −k −ζ and E ℓ ζ −k,k ζ , and evaluate their first Chern classes. We begin with a general lemma. Lemma 5. 10. Let Z, W be codimension 2 cycles in a smooth variety Y . (i) If Z ⊆ W , then Hom(I W , is open and dense in Z, then Hom(I W , I Z ) = I Z ; (iii) If Z and W are local complete intersections meeting properly, then there is an exact sequence: (iii) We begin with the local identification: let R be a regular local ring, and let Z and W be two codimension 2 local complete intersection subschemes of R meeting properly. Applying the functor Hom R (·, I Z ) to the Koszul resolution of W Here (I W + I Z ) corresponds to the intersection W ∩ Z. The identification of Ext 1 R (I W , I Z ) and I Z /(I Z ∩ I W ) is not canonical. Globally we must correct by det N W . Thus globally we have an exact sequence: (iv) It is clear that Ext 1 (I W , I Z ) is a sheaf supported on W . To show that it has rank 1 as a sheaf on W , it suffices to verify that it has rank 1 at a generic point w of W . Since Z ∩ W is nowhere dense in W and W is smooth at a generic point, we may assume that w ∈ Z and that w is a smooth point of W . Then it follows from (iii) that Ext 1 (I W , I Z ) is of rank 1 at w. (i) There exist a row exact sequence and a column exact sequence: Proof. (i) Note that the bundle E ℓ ζ −k,k ζ is defined as Since R 2 π 2 * (π * 1 O X (ζ) ⊗ Hom) = 0, the row exact sequence follows from standard facts about relative Ext sheaves. To see the column exact sequence, we use Lemma 5.10 (ii) and apply the functor π 2 * to the exact sequence (ii) Note that Hom = I Z ℓ ζ −k and that R i π 2 * (π * 1 O X (ζ) ⊗ Hom) = 0 for i = 0, 2. By the Grothendieck-Riemann-Roch Theorem, we have Now, the conclusion follows by comparing the degree 1 terms and by the fact that 2 + (terms with degree ≥ 4). Proposition 5.12. Let α ∈ H 2 (X; Z) and a = (ζ · α)/2. Then, Proof. By the symmetry between k and (ℓ ζ − k), we see that S 2ℓ ζ −1 is equal to From Lemma 5.11, we conclude that c 1 (E Since , we see that where the c 3 's are cancelled out. Therefore, by Lemma 5.8, It is possible, but far more complicated, to compute (5.7) for j = 2ℓ ζ − 2. Next, we shall draw some consequences from our previous computations. Recall that q X denotes the intersection form of X, and that is the difference between the complex orientation and the standard orientation on the instanton moduli space (see [6]). Theorem 5.13 below has already been obtained by Kotschick and Morgan [18] for any smooth 4-manifold with b + 2 = 1. Theorem 5.13. Let ζ define a wall of type (w, p), and d = −p − 3. Then, for α ∈ H 2 (X; Z), where a = (ζ · α)/2. In other words, Proof. By Theorem 5.4 and our notation (5.7), we have By Proposition 5.12, S 2ℓ ζ −1 is divisible by a. Therefore, Now, our conclusion follows from Proposition 5.9 and the fact that The following is proved by using a similar method. Theorem 6.5. Let ζ define a wall of type (w, p) with ℓ ζ = 1, Proof. By Theorem 5.5, By Proposition 5.9, Proposition 5.12, and Lemma 6.3, we have In the rest of this section, we assume that ℓ ζ = 2. The following standard facts about double coverings can be found in [2,10]. Lemma 6.6. Let φ : Y 1 → Y 2 be a double covering between two smooth projective varieties with φ * O Y1 = O Y2 ⊕ L −1 where L is a line bundle on Y 2 . (i) K Y1 = φ * (K Y2 ⊗ L) and L ⊗2 = O Y2 (B) where B is the branch locus in Y 2 and is the image of the fixed set of the involution ι on Y 1 ; (ii) If D is a divisor on Y 1 , then φ * (O Y1 (D)) is a rank 2 bundle on Y 2 with c 1 (φ * (O Y1 (D))) = φ * D − L and Next, we recall some standard facts about the Hilbert scheme H 2 = Hilb 2 (X). Let ∆ 0 ⊂ X × X be the diagonal, and let ι be the obvious involution onH 2 = Bl ∆0 (X × X), the blowup of X × X along ∆ 0 . Let E be the exceptional divisor of the blowup inH 2 . Then, H 2 =H 2 /ι and the branch locus lies under E. Let Z 2 ⊂ X ×H 2 be the pullback of the codimension 2 cycle Z 2 ⊂ X × H 2 . Then, Z 2 splits into a union of two cyclesH 12 andH 13 in X ×H 2 , which are the proper transforms in X ×H 2 of the two morphisms of X × X into X × (X × X): the first maps the first factor in X × X diagonally into X × X which is the product of the first and second factors in X × (X × X), while the second maps the first factor in X × X diagonally into X × X which is the product of the first and third factors in X × (X × X). Thus eachH 1j is isomorphic to Bl ∆0 (X × X), and the projection of each toH 2 is an isomorphism. If α ∈ H 2 (X; Z), then [Z 2 ]/α = α ⊗ 1 + 1 ⊗ α = α ⊗ 1 + ι * (α ⊗ 1) (6.7) where α⊗1 and 1⊗α are the pull-backs of α by the two projections ofH 2 to X. Fix x ∈ X. LetX x be the pull-back of X × x ⊂ X × X toH 2 . Then,X x is isomorphic to the blow-up of X at p with the exceptional divisor (X x ∩ E); moreover, It is known (see p. 685 in [9]) that Z 2 is smooth. Let B be the branch locus of the natural double covering from Z 2 to H 2 . Then, B ∼ 2L for some divisor L on H 2 , and the pull-back of B ⊂ H 2 toH 2 is 2E. Let i : Z 2 → X × H 2 be the embedding, and π 1 and π 2 be the natural projections of X × H 2 to X and H 2 respectively. In the following, we compute the Chern and Segre classes of E 2−k,k ζ for k = 0, 1, 2. The method is to use Lemma 5.11 together with Lemma 6.6. We start with E 2,0 ζ . Lemma 6.9. Proof. Let notations be as in Lemma 5.11, and let ℓ ζ = 2 and k = 0. Then, Ext 1 = 0. By Lemma 5.11 (i), E 2,0 ζ sits in an exact sequence 0 → (π 2 · i) * (π 1 · i) * O X (ζ) → E 2,0 ζ → [O H2 ] ⊕ h(ζ) → 0. The following follows from Lemma 6.9 and Remark 5.6. Corollary 6. 10. The Segre classes of the bundle E 2,0 ζ are given by Here we have identified degree 4 classes with the corresponding integers. The following follows from Lemma 6.11 and Remark 5.6. Corollary 6.12. The Segre classes of E 0,2 ζ are given by Proof. The calculation of s 4 (E 0,2 ζ ) is similar to that of s 4 (E 2,0 ζ ) in Corollary 6.10. Note that s 4 (E 0,2 ζ ) may be obtained from s 4 (E 2,0 ζ ) by replacing ζ by K X − ζ, and indeed this holds more generally for s i when we add the sign (−1) i . Now we compute the Chern and Segre classes of E 1,1 ζ on X × X. Lemma 6.13. Let τ 1 and τ 2 be the two natural projections of X × X to X, let ∆ 0 be the diagonal in X × X, and let j : ∆ 0 → X × X be the inclusion. Then Proof. Let ℓ ζ = 2 and k = 1 in Lemma 5.11. Recall that π 1 and π 2 are the natural projections of X × (X × X) to X and (X × X) respectively. Claim 1. π 2 * π * 1 O X (ζ) ⊗ Ext 1 ∼ = τ * 2 O X (ζ − K X ) ⊗ I ∆0 . Proof. Let ∆ 12 be the diagonal in X × X which is formed by the first and second factors in X × (X × X), and let ∆ 13 be the diagonal in X × X which is formed by the first and third factors in X × (X × X). Then, ∆ 12 × X and ∆ 13 × X are smooth codimension 2 subvarieties in X × (X × X). Here it is understood that the factor X in ∆ 13 × X is embedded as the second factor in X × (X × X). Moreover, ∆ 12 × X and ∆ 13 × X intersect properly along the diagonal ∆ 123 in X × X × X. Thus, from Lemma 5.10 (iii), we conclude that where N is the normal bundle ∆ 13 × X in X × (X × X), and I is the ideal sheaf of ∆ 123 in ∆ 13 × X. Now, the restriction of π 2 to ∆ 13 × X gives an isomorphism from ∆ 13 × X to X × X. Via this isomorphism, ∆ 123 in ∆ 13 × X is identified with the diagonal ∆ 0 in X × X, det N is identified with τ * 2 (−K X ), and the restriction π * 1 O X (ζ)\|(∆ 13 × X) is identified with τ * 2 (ζ). Therefore, π 2 * π * 1 O X (ζ) ⊗ Ext 1 ∼ = π 2 * (π * 1 O X (ζ) ⊗ I ⊗ det N ) = τ * 2 O X (ζ − K X ) ⊗ I ∆0 .	2014-10-01T00:00:00.000Z	1994-10-12T00:00:00.000	{ "year": 1994, "sha1": "b381d9b777411625c388f87d2d11ae3198982867", "oa_license": null, "oa_url": "https://www.intlpress.com/site/pub/files/_fulltext/journals/cag/1995/0003/0001/CAG-1995-0003-0001-a002.pdf", "oa_status": "BRONZE", "pdf_src": "Arxiv", "pdf_hash": "b381d9b777411625c388f87d2d11ae3198982867", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
221129165	pes2o/s2orc	v3-fos-license	Upregulation of miR-382 contributes to renal fibrosis secondary to aristolochic acid-induced kidney injury via PTEN signaling pathway Acute kidney injury (AKI) has a critical role in the development of chronic kidney disease (CKD). Building on our previous findings, we explored the role of miR-382 in facilitating the transition of AKI to CKD using the Aristolochic acid (AA) nephropathy model, which was induced by intraperitoneal injection of aristolochic acid I salt (10 or 20 mg/kg). The effects of genetic depletion, pharmacologic inhibition, or overexpression of miR-382 on the PTEN/AKT signaling pathway were examined in vivo and in vitro. Changes in renal pathology and renal epithelial polarity were evaluated. A luciferase reporter assay was performed to investigate the reciprocal suppression relationship between miR-382 and PTEN. Renal fibrosis developed 14 d after AA exposure in a dose- and time-dependent manner. Renal abundance of miR-382 was upregulated following AA treatment, while genetic depletion or pharmacological inhibition of miR-382 partially reversed renal tubulointerstitial fibrosis. Expression of PTEN, a target of miR-382, was downregulated and subsequently its downstream AKT signaling pathway was activated during AKI to CKD transition induced by AA. Inhibition of PTEN in vitro resulted in the acquisition of the EMT phenotypes. Furthermore, upregulation of miR-382 in renal epithelial cells was partially mediated by the activation of NF-kB signaling, with a substantial elevation of proinflammatory cytokines. An in vivo study revealed that either miR-382 knockdown or miR-382 knockout was pivotal for inflammatory suppression, while an in vitro experiment confirmed that upregulation of miR-382 in cultured MTEC cells under AA exposure was remarkably reversed by NF-kB siRNA. These data indicated a novel role for the NF-κB/miR-382/PTEN/AKT axis in the pathogenesis of tubulointerstitial fibrosis following AA-induced acute renal tubular epithelial injury. Targeting miR-382 may lead to a potential novel therapeutic approach for retarding the AKT to CKD transition. Introduction Acute kidney injury (AKI), a syndrome defined by rapidly declining renal function, has a high worldwide incidence. Although AKI was considered a reversible disease, recent clinical research has revealed that even patients who have completely recovered from an episode of AKI are at a higher risk for chronic kidney disease (CKD) 1,2 . The proximal tubular cells are reported as the primary target as well as a dominant trigger of injury and progression of kidney disease 3,4 . Tubular epithelial cells are the determinant in the development of interstitial inflammation 5 , which can be observed in the early stage of many renal diseases 6 . However, mechanisms underlying the transition from AKI to CKD remain largely unclear. Aristolochic acid nephropathy (AAN), a common model of AKI to CKD transition secondary to renal toxicity 7,8 , is an important drug-associated renal injury disease 9 and AA mainly targets tubular epithelial cells and promotes death of tubules via formation of DNA adducts 9,10 . MicroRNAs are endogenous, small non-coding RNAs, 22 nucleotides in length, which mostly regulate gene expression negatively at the post-transcriptional level and have a critical role in kidney diseases 11,12 . MiR-382, located at chromosome 14q32. 31, is a member of the miR-17-92a cluster. Microarray analyses have revealed that miR-382 is upregulated during TGF-β1-induced epithelial to mesenchymal transition (EMT) in human renal epithelial cells 13 and also contributed to the development of renal fibrosis in mouse unilateral ureteral obstruction (UUO) models 14,15 . Thus, although it appeared that miR-382 has a critical role in CKD progression, whether miR-382 contributes to AKI-induced CKD remained unclear. Phosphatase and tensin homologs deleted on chromosome 10 (PTEN)/protein kinase B(AKT) pathway has been reported in tumors as a tumor suppressor 16,17 . In addition, depletion of PTEN was characteristic of renal fibrosis and overexpression of PTEN expression attenuated tubulointerstitial fibrosis [18][19][20][21][22] , inhibited macrophage polarization from M1 to M2 23,24 and suppressed inflammation responses 25,26 . Given that PTEN has been identified as a target of miR-382 in liver generation, acute promyelocytic leukemia, and tumor angiogenesis as well as oxidative stress of the tubular epithelium [27][28][29][30] , which remains unclear in kidney fibrosis. NF-κB is an important transcription factor that mediates the expression of various miRNAs [31][32][33][34] and is also a pivotal mediator of inflammation 35 . Therefore, we hypothesized that activation of NF-kB may induce transcription of miR-382, which promotes the AKI to CKD transition by targeting PTEN/AKT signaling. AA induces miR-382 overexpression and AKI-CKD transition in mice In order to observe the natural course of AKI to CKD transition, a time-and dose-course study was performed in mouse models of Aristolochic acid Nephropathy (1, 3, 7, 14, and 28 days; moderate versus severe). AA related kidney injury was characterized by a significant loss of proximal tubule brush, severe atrophy of the proximal tubular epithelium (PTE), acute tubular necrosis and obvious extracellular matrix (ECM) deposition in the mesenchyme. In comparison with the SAAN group, more proliferative PTE cells were seen in the MAAN group (Fig. 1a, b). EMT-related mesenchymal biomarkers Vimentin, α-smooth muscle actin (α-SMA) and Ncadherin 36 were increased while epithelial biomarkers E-cadherin was decreased on days 7, 14, and 28, as detected by immuno-histological staining and western blot. (Fig. 1c, e, p-s) Fibrotic related proteins fibronectin and collagen IV were increased in SAAN group on days 14 and 28 (Fig. 1d, t). Serum creatinine (Scr) ascended significantly on days 7 (MAAN group) and 3 (SAAN group), compared with that in the control group (Fig. 1f). Renal abundance of miR-382 was elevated early on day 1 (MAAN group), while it did not reach statistical significance until day 7 in the SAAN group, indicating that abundance of miR-382 might not be associated with the dosage of AA (Fig. 1g). Concomitantly, mRNA expression of NF-κB was elevated since day 1 and peaked on day 7 (MAAN group), and on day 14 in the SAAN group (Fig. 1k). Activity of NF-κB phosphorylated p65 in renal section was enhanced significantly on days 14 and 28 in the SAAN group (Fig. 1c, o). A similar tendency was found in the mRNA expression of fibrotic genes, such as α-SMA, Collagen I, and Collagen III, as well as inflammatory cytokines including interleukin-6 (IL-6), interleukin-10 (IL-10), and tumor necrosis factor-α (TNFα); (Fig. 1i-n). These results demonstrated that activation of NF-κB and upregulation of miR-382 was accompanied by the aggravation of inflammatory infiltration, progression of EMT and development of renal fibrosis. Knockout of miR-382 reverses renal inflammation and fibrosis in mice Relevance of miR-382 upregulation and development of renal fibrosis was further investigated in miR-382 knockout mice. A dose of 20 mg/kg AA was intraperitoneally administered to miR-382 knockout (KO) mice as well as wildtype (WT) mice (n = 3 per group) and mouse were killed at day 14. Firstly, miR-382 KO mice showed significantly less renal lesions, a mild loss of proximal tubule brush, less proximal tubule atrophy and less extracellular matrix (ECM) deposition than WT mice (Fig. 2a). Compared with WT mice, expression of α-SMA and Vimentin were decreased and loss of epithelial biomarker ZO-1 was attenuated in miR-382 KO mice after AA treatment (Fig. 2a, i-k). MiR-382 upregulation was significantly abolished in miR-382 KO mice compared to WT mice (Fig. 2b). Moreover, increased mRNA expression of α-SMA, Collagen I and Collagen III, and protein expression of fibronectin, collagen IV, N-cadherin, α-SMA and vimentin as well as loss of E-cadherin secondary to AA stimulation was found to be markedly attenuated by miR-382 KO ( Fig. 2f-h, l-q). In addition, compared to the anti-scramble group, anti-miR-382 Oligos treatment were also proved to be reno-protective ( Supplementary Fig. S1). MiR-382 promotes renal fibrosis via PTEN-mediated AKT signaling Given that transition of AKI to CKD was induced by the upregulation of miR-382 in the renal tissue of mouse AA models, we explored the mechanisms underlying miR-382 mediated renal fibrosis. Reciprocal suppression between miR-382 and PTEN was verified both in human and mouse kidneys. Notably, depletion of PTEN expression and upregulation of miR-382 were found in IgAN patients with tubulointersitial fibrosis (TIF), compared with IgAN patients without TIF (Fig. 3a, b). In mouse model of AA, PTEN expression was inhibited and its downstream AKT was activated through phosphorylation of AKT on Ser473. However, phosphorylation of AKT on Thr308 was significantly downregulated ( Fig. 3c-g). A dual-luciferase reporter assay was performed (Fig. 3h). Luciferase activity of the 293T cells was significantly suppressed when cotransfecting the miR-382 mimics and pMIR-PTEN-plasmid, but not with the pMIR-PTEN-mut plasmid (mutation has occurred in the selected sequence), suggesting that miR-382 inhibits PTEN by combining with 3′UTR of pten (Fig. 3i). Furthermore, inhibition of PTEN and activation of AKT through Phosphate-AKT Ser 437 were attenuated in miR-382 KO mice, which provided robust evidence of reciprocal suppression between miR-382 and PTEN ( Fig. 3j-m). Consider together, PTEN/AKT signaling might get involved in the contribution of miR-382 in driving AKI-to-CKD after AA treatment. AA induces miR-382 expression and promotes EMT in cultured mouse tubular epithelial cells To confirm miR-382 localization in the kidney, we performed in situ hybridization (ISH) of miR-382 in mouse kidney sections. miR-382 is widely distributed in the cortex and the medulla of the kidney, especially in the PTE while expression in the glomeruli were relatively sparse (Fig. 4a). As mentioned above, renal tubules are dominant in response to injuries 5 . Therefore, we performed a time course study and a dose-response study in mouse renal tubular epithelial cells (mTECs). Activity of NF-kB phosphorylated p65 in mTECs was increased in a dose-dependent way after AA stimulation (Fig. 4b, c). The abundance of miR-382 increased either in a timedependent manner (Fig. 4d) or in a dose-dependent manner (Fig. 4h). A significant increase in mRNA expression of IL-6, TNF-α, and NF-κB were detected after treating with AA ranging from 5 to 20 μg/ml for 48 h ( Fig. 4e-g) or with AA (10 μg/ml) for 12 h (Fig. 4i-k). Above all, AA stimulation in mTECs resulted in activation of NF-κB, upregulation of miR-382, and increased expression of inflammatory cytokines. PTEN/AKT signaling and phenotype of EMT were examined in AA-induced tubular epithelial cell injury. In consistent with in vivo study, protein expression of PTEN was significantly inhibited when mTECs were incubated with AA (10 and 20 μg/ml) either for 24 or for 48 h. Phosphorylation of AKT on Ser473 was increased in a dose-and time-dependent way while phosphorylation of AKT on Thr308 was decreased (Fig. 5a, j, c-f, l-o). Furthermore, E-cadherin was downregulated while α-SMA was upregulated after treatment of AA with indicated concentration for 48 h (Fig. 5b, k, g, h, p, q). As detected by immunofluorescence staining, expression of epithelial biomarker ZO-1 was declined and mesenchymal biomarker Vimentin and α-SMA were upregulated after AA (10 μg/ml) treatment for 48 h ( Fig. 5r-w). Overall, AA activated inflammation and subsequently upregulated miR-382, thereby promoting PTEN/AKT signaling and EMT in a time-and dose-dependent manner. This was consistent with the results of the in vivo study. AA induces miR-382 overexpression via NF-κB activation and promotes EMT mediated by PTEN signaling According to previous studies, NF-κB is essential to mediate inflammation response and also acts as a transcription factor of a variety of miRNAs. NF-κB siRNA treatment significantly attenuated AA-induced activation of NF-κB, and markedly downregulated miR-382, which provided evidence that NF-κB might regulate miR-382 (Fig. 6a). Furthermore, overexpression of miR-382 inhibited PTEN abundance and upregulated α-SMA expression, it did not exert any effects on NF-κB p65 expression, which indicated NF-κB did not respond to miR-382 overexpression (Fig. 6b). In addition, Expression of PTEN, which was consistently reduced following treatment with AA (10 μg/ml for 48 h), was reversed following anti-miR-382 treatment, which verified reciprocal suppression between miR-382 and PTEN in vitro ( Fig. 6c-e). (see figure on previous page) Fig. 1 AA-induced miR-382 expression and development of AKI to CKD in mice. a Representative images of mice kidney sections from Moderate AAN group administered 10 mg/kg of AA with H&E and Masson staining. Scale bars, 100 μm. b Representative images of mice kidney sections from the Severe AAN group administered 20 mg/kg AA with H&E, Masson staining. Scale bars, 100 μm. c Immunostaining for α-SMA and Vimentin from control mice, days 7, 14, and 28 in the severe AAN group. Scale bars, 100 μm. Immunofluorescence for p-NF-κB p65 from control mice, days 7, 14, and 28 in the severe AAN group. Scale bars, 50 μm. d, e Representative immunoblot analysis of Fibronectin, Collagen IV, E-cadherin, Ncadherin, Vimentin, and α-SMA in kidney tissues from the severe AAN group. GAPDH served as the standard. f Renal dysfunction of AAN was determined in AA-induced AKI-to-CKD models. Serum creatinine (Scr) was measured in sera; n = 6 per group. g Relative abundance of miR-382 was examined in mice kidney tissues. h-j Renal fibrotic manifestation was examined in the models. Relative mRNA levels of α-SMA, Collagen I and Collagen III in mice kidney tissues were examined; n = 6 per group. k Relative mRNA levels of NF-κB in mice kidney sections from AA-induced nephropathy; n = 6 per group. l-n Renal inflammation manifestation was examined in the models. Relative mRNA levels of IL-6, IL-10, and TNF-α in mice kidney tissues were examined; n = 6 per group. o Quantification of mean Fluorescence intensity for p-NF-κB p65 from control mice, day 1, 7, 14, and 28 in severe AAN group. We randomly choose five microscopical field per sections, n = 3 mice per group. p-t Quantification of western blot analysis for Vimentin, α-SMA, E-cadherin, N-cadherin, and Fibronectin from control mice, days 1, 3, 7, 14, and 28 in the severe AAN group. P < 0.05; P < 0.01; ANOVA. Moreover, PTEN was inhibited with PTEN inhibitor OV-Ohpic trihydrate prior to AA treatment. Inhibition of PTEN followed by AA treatment, the effects on inhibition of PTEN, phosphorylation of AKT on Ser473 and downregulation of p-AKT Thr308 were enhanced, compared with DMSO + AA group. Besides, downregulation of E-cadherin and ZO-1 and upregulation of α-SMA were aggravated in the OV-Ohpic + AA group, which indicated that PTEN/AKT signaling pathway is involved in the development of EMT in renal tubular epithelial cells ( Fig. 6f- Discussion This study provided novel evidence showing that the non-coding small RNA, miR-382, mediates TIF in mouse AA-induced AKI-to-CKD models. We have shown that miR-382 acted as a critical regulator of the inflammation response and progression of EMT and accumulation of ECM, both in vivo and in vitro. Based on the results of a dual-luciferase assay, we proposed a molecular mechanism whereby PTEN, an important factor that inhibits renal fibrosis, serves as a downstream target of miR-382. Furthermore, our findings indicated that AKI-to-CKD transition induced by AA was partially mediated by the activation of miR-382/PTEN/AKT signaling. In vivo and in vitro inhibition of miR-382, via genetic or pharmacological methods, rescued the loss of PTEN, negatively regulated the immune response, reversed EMT, and attenuated the accumulation of ECM. In contrast, overexpression of miR-382 resulted in the converse effect. Moreover, our result revealed the molecular mechanism by which NF-κB acts as an upstream transcription factor of miR-382 during AA-induced AKI-to-CKD transition. The close association between Chinese herbs containing aristolochic acids (AA) and rapidly progressive interstitial renal fibrosis in humans has been reported in several countries during the last decade 37 . This murine AAN model was characterized by the development of significant interstitial fibrosis, prominent tubular atrophy, and necrosis, which is identical to AAN seen in humans 38,39 and shares some features, such as upregulation of miR-382, loss of PTEN expression, and impaired renal function. As mentioned above, renal miR-382 upregulation was accompanied by the development of interstitial fibrosis as well as degeneration and regeneration of tubular epithelium, which is consistent with previous studies [13][14][15] . However, circulating miR-382 in mice seems to be significantly decreased in drug-or vascular calcificationinduced kidney injury models 28,40 , which opposes the expression of miR-382 in renal tissue. To avoid renal toxicity caused by microRNA inhibitors, our study is the first one to establish miR-382 knockout mice and reveal effect of miR-382 on AA-induced chronic kidney injury. However, miR-382 expression is not specific to kidneys and is also seen in various other tissues such as liver, heart, and brain 15 . We performed ISH of miR-382 in mouse renal sections and found that miR-382 is expressed globally in the cortex and the medulla, especially in the proximal tubular epithelium, which supports our in vitro studies using mouse tubular epithelium cell lines. In addition, our study demonstrated that the loss of PTEN expression in renal fibrosis, which has since been substantiated by many recent studies 14,18,41,42 . Both in vivo and in vitro, phosphorylation of AKT on Ser473 was increased as along with AA-induced kidney injury, which was consistent with previous studies 43 . In addition, we found impaired AKT activation with difference between Ser473 and Thr308 AKT phosphorylation during AKI-CKD transition, which also contributed to myocardial infarct size in 5/6 nephrectomy rats 43 . Therefore, we speculated that mutual regulation between p-AKT Ser473 and Thr308 might exist and phosphorylation of AKT on Ser473 would have a dominant role in PTENloss-induced AKT activation. In our in vitro study, inhibition of PTEN increased p-AKT Ser473 and decreased p-AKT Thr308 but had no effect on total AKT. Therefore, other mechanism might exist in the regulation of total AKT. The reciprocal relationship between miR-382 and PTEN has been reported in acute promyelocytic leukemia, (see figure on previous page) Fig. 2 Knockout of miR-382 contributes to reverse renal inflammation and fibrosis in mice. a Representative images of kidney sections from miR-382 WT or KO mice 14 d after the administering of 20 mg/kg AA (SAA group) with H&E and Masson staining, immunostaining for α-SMA as well as immunofluorescence for α-SMA, Vimentin and ZO-1. Scale bars, 100 or 50 μm. b Quantification of relative mRNA levels of miR-382 in mice kidney from miR-382 WT or KO mice 7, 14, and 28 days after the administering of 20 mg/kg AA; n = 6 per group. c-e Relative mRNA levels of IL-6, IL-10, and TNF-α in mice kidney tissues were examined in these groups. n = 6 per group. f-h Inhibition of miR-382 in mice reversed fibrotic manifestation of kidney significantly. Relative mRNA expression of Collagen I, Collagen III, and α-SMA in mice kidney tissues were examined between miR-382 WT and KO group 7 and 14 d following the administering of 20 mg/kg AA. n = 6 in each group. i-k Quantification of mean Fluorescence intensity for α-SMA, Vimentin, and ZO-1 between miR-382 WT and KO group 14 d following the administering of 20 mg/kg AA. We randomly choose five microscopical field per sections, n = 3 mice per group. l, n, o Representative and quantification immunoblot analysis of Fibronectin and Collagen IV in renal tissue from miR-382 WT or KO mice 14 d following 20 mg/kg AA treatment. GAPDH served as the standard; n = 6 in each group. m, p-s Representative and quantification immunoblot analysis of E-cadherin, N-cadherin Vimentin, and α-SMA in renal tissue from miR-382 WT or KO mice 14 d following 20 mg/kg AA treatment. GAPDH served as the standard; n = 6 in each group. P < 0.05; *P < 0.01; ANOVA. infantile hemangioma, and liver regeneration 27,29,44 In our study, we performed a dual-luciferase reporter assay, which confirmed that miR-382 negatively regulated PTEN by combining the 3′UTR of gene pten in 293T cells. Further, immunohistochemical staining of PTEN and RT-qPCR of miR-382 in renal biopsy sections from patients with IgA nephropathy (with or without TIF) as well as protein expression of PTEN and its downstream AKT in kidney from mouse AAN model both in WT and miR-382 KO mice also substantiated this relationship. Furthermore, we found inhibition of PTEN in vitro partially accelerated EMT progression, which has been reported in diabetic nephropathy 19 or cancers 45 . However, intervention of PTEN in vivo to further examine its effect on renal fibrotic outcome is absent in our study. Considerable evidence indicates that renal inflammation has a central role in the initiation and progression of CKD 46 . AA-induced inflammation in kidneys is a type of sterile inflammation, which is defined as inflammation in the absence of infectious agents or specific immunogens 47 . During this process, leukocytes, fibrogenic cells, and resident kidney immune cells are recruited to the renal interstitium, which lead to increased production of proinflammatory cytokines. Administering of a single dose of AA leads to a gradual increase in proinflammatory renal cytokines until day 28, which may account for the pivotal role had by inflammation in the progression from AKI to CKD. Compared that of with WT mice, the prominent expression of inflammatory cytokines in miR-382 KO mice was attenuated following treatment with 20 mg/ kg AA, indicating that miR-382 was a proinflammatory microRNA. However, in our in vitro study, the elevation of inflammatory cytokines started 12 h post AA treatment which was prior to the upregulation of miR-382, showing that the inflammation response possibly promoted miR-382 expression during the early stage of AA-induced kidney injury. Thus, AA-activated inflammatory reaction led to the upregulation of miR-382, a proinflammatory microRNA, which further exacerbated kidney inflammation, causing relentless interstitial fibrosis. NF-κB, which functions as a transcription factor for many miRNAs, has an important role in the regulation of countless cellular functions, including the cell cycle and apoptosis, among others 34 . in vitro inhibition of NF-κB significantly suppressed miR-382, indicating that NF-κB possibly targets miR-382. However, more in vivo intervention of NF-κB as well as RNA chip assays have to be conducted in the future. Moreover, whether miR-382 regulates the activation of inflammatory cells, such as macrophages, may demand further studies. In conclusion, our study proposes a novel mechanism wherein overexpression of miR-382, which occurs subsequent to proinflammatory cytokine NF-κB activation, enhances AKI-to-CKD transition in mouse AA-induced kidney injury models (Fig. 7a, b). Mice Male C57BL/6J mice (8-to 10-week-old; 20-25 g) were obtained commercially (SLAC LABORATORY ANIMAL CO. LTD, Shanghai, China). miR-382 −/− mice were bred in the BIORAY Lab, from C57BL/6 J mice back-crossed to miR-382 −/− mice, and used for experiments. All animal experiments were approved by the Institutional Animal Care and Use Committee of Fudan University and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Aristolochic acid-induced AKI-to-CKD transition was conducted using an intraperitoneally injected single dose of Aristolochic acid I sodium salt (A9451, Sigma). All the mouse were randomly assigned to experimental groups. Samples were performed in biological replicates. The sample size was estimated based on the need for statistical power. We divided mouse into moderate AAN group (10 mg/kg) and severe AAN group (20 mg/kg) and mouse were separately killed at 1, 3, 7, 14, and 28 days (n = 6 mouse per group). Control mice were treated with the same dosage of saline via intraperitoneal injection. Locked nucleic acid-(LNA-) modified anti-miR-382 oligonucleotides or anti-scramble (Exiqon, Shanghai, China) at a dose of 10 mg/kg/7 d diluted in saline (5 mg/ml) was (see figure on previous page) Fig. 3 MiR-382 promotes renal fibrosis via PTEN-mediated AKT signaling in AA reduced AKI to CKD. a Representative image of immunohistochemical staining for PTEN of human renal biopsy specimens from patients with IgAN with or without tubulointerstitial fibrosis. b Relative expression of miR-382 in human renal biopsy tissues. n = 6 per group. c-g miR-382 suppressed PTEN and activated AKT signaling pathway. Representative and quantification immunoblot analysis of PTEN, AKT, p-AKT Ser473, and p-AKT Thr308 in time course study of SAAN. GAPDH served as standard of PTEN; AKT served as standard of p-AKT; n-6 per group. h The targetScan bioinformatics website was used to predict the possible sequence of PTEN 3′UTR combined with miR-382. Mutation was induced in the PTEN 3′UTR sequence. i luciferase activity was quantified in 293 T cells between the Control group, NC+pMIR-PTEN-plasmid group, miR-382 mimic+ pMIR-PTEN-plasmid group, NC+pMIR− PTEN-mut plasmid group and miR-382 mimic+pMIR-PTEN− mut plasmid group. Dual-luciferase activity was measured via a Dual-Glo Luciferase Assay System. n = 3 per group. j-m Representative and quantification immunoblot analysis of PTEN, AKT, p-AKT Ser473, and p-AKT Thr308 in mice renal from miR-382 WT or KO mice 14 d following severe administering of AA. GAPDH served as standard of PTEN; AKT served as standard of p-AKT; n-6 per group. P < 0.05; *P < 0.01; ANOVA. intraperitoneally administered to inhibit miR-382 expression in mice. The first injection was performed within 30 min prior to AA-I injection and the treatment was repeated every 7 d. All the experiments were replicated at least twice. Cell culture and siRNA transfection Mouse renal tubular epithelial cells (mTECs) were purchased from the American Type Culture Collection (ATCC) and recently authenticated and tested for mycoplasma contamination. Cells were cultured in medium as indicated in Supplementary Table 2 and in a humidified chamber with 5% CO2 at 37°C. Cells were cultured in 6-wells plates and randomly assigned to experimental groups. Cells were treated with AA (1, 2, 5, 10, or 20 μg/ml) for 48 h for concentration course study and with AA (6, 12, 24, 48, or 72 h) for 10 μg/ml for time course study. Transgene expression was induced with Fig. 4 Distribution of miR-382 in mice kidney and AA-induced inflammation in mTECs. a In situ hybridization of miR-382 in mice kidney samples was performed. Representative images of miR-382 ISH in tubule and glomerulus. miR-382 was visualized by staining with Dye3 (Red). Nuclei were counterstained with DAPI (blue). Scale bars, 100 μm. b, c Immunofluorescence images and quantification of mean Fluorescence intensity for p-NF-κB p65 from control cells or after treatment with 5, 10, and 20 μg/ml AA for 24 h in mTECs. We randomly choose five microscopical field per sections, n = 3 holes per group. Scale bars, 50 μm. d Relative abundance of miR-382 in mTECs following treatment with 10 μg/ml AA for 0, 6, 12, 24, and 48 h. U6 served as the standard. n = 6 per group. e-g Quantification of relative mRNA levels of IL-6, TNF-α, and NF-κB in vitro in these groups; n = 6 per group. h Relative abundance of miR-382 in mTECs following treatment with 0, 1, 2, 5, 10, and 20 μg/ml of AA for 48 h. U6 served as the standard; n = 6 per group. i-k in vitro Quantification of relative mRNA levels of IL-6, TNF-α and NF-κB in these groups. n = 6 per group. P < 0.05; *P < 0.01; ANOVA. 100 nM anti-miR-382 (also referred to as anti-scramble) or 100 nM miR-382 mimic (also referred to as negative control) (Exiqon, Shanghai, China). Inhibition of NF-κB was performed by transfecting NF-κB siRNA or negative control (100 nM, Ribo Life Science, Shanghai, China). All the experiments were replicated at least twice. Human kidney tissue samples We studied 12 patients with IgA nephropathy (IgAN) between 2017 and 2018 and performed renal biopsy in Zhongshan Hospital, Fudan University in order to evaluate tubulointerstitial fibrosis (TIF) in these patients using IgAN. Patients were divided into IgAN with TIF and IgAN without TIF groups (n = 6 per group) according to evaluation of a pathologist. The investigator was blind to the group allocation when immunohistochemical staining of PTEN and RT-qPCR of miR-382 were performed on human renal specimens. This study was approved by the Clinical Research Ethical Committee of the Zhongshan Hospital, Fudan University. Written informed consent was obtained from all patients. Serum creatinine Serum creatinine levels of mice were determined using a QuantiChrom TM Creatinine Assay Kit (BioAssay Systems, Hayward, CA, USA) following the manufacturer's instructions. Immunohistochemical staining Renal tissues were fixed with 10% formalin, embedded in paraffin wax and sliced into 4-μm-thick sections for hematoxylin-eosin (H&E) staining, Masson staining or immune-histochemical staining. Immunohistochemical staining was performed as described previously 13 . The primary and secondary antibodies were showed in Supplementary Table 2. Sections were evaluated via microscopy (×200 magnification, Leica DM 6000B; Leica Microsystems, Wetzeler, Germany). Immunofluorescence After treatments, frozen kidney sections or mTECs, which crawled on the slide were fixed in 4% paraformaldehyde for 15 min. Cells were permeabilized by 0.5% Triton X-100 in PBS for 10 min. After blocking with 5% BSA in PBS, sections or cells were incubated with antibodies followed by secondary antibodies. Antibodies were presented in Supplementary Table 2. Images were acquired using Olympus FV1000 confocal microscope. Dual-luciferase assay The targetScan bioinformatics website (http://www. targetscan.org/vert_71/) was utilized to predict targets of miR-382 and the possible sequence of the target binding sites with miR-382. Next, pMIR-PTEN-3′UTR-wt and pMIR-PTEN-3′UTR-mut were cloned into the pMiR dual-luciferase reporter plasmid vector. The recombinant pMiR dual-luciferase reporter plasmid was co-transfected with miR-382 mimics or negative control into 293T cells via lipo3000. Dual-luciferase activity was measured using the Dual-Glo Luciferase Assay System. Western blotting Western blot was performed as previously described 14 . The primary antibodies were incubated followed by HRP-conjugated secondary antibodies, which was showed in Supplementary Table 2. Protein levels were quantified using the Image Lab software, version 3.0 (Bio-Rad, USA). (see figure on previous page) Fig. 5 AA stimulation regulated PTEN/AKT signaling pathway and promoted EMT in mTECs. a, c-f Representative and quantification immunoblot analysis of PTEN, AKT, p-AKT Thr308, and p-AKT Ser473 in mTECs from administration of 0, 1, 2, 5, 10, and 20 μg/ml AA for 48 h. GAPDH served as the standard of PTEN, p-AKT served as the standard of AKT; n = 6 per group. b, g-h Representative and quantification immunoblot analysis of E-cadherin, Ncadherin, and in mTECs from administration of 0, 1, 2, 5, 10, and 20 μg/ml AA for 48 h. GAPDH served as the standard of PTEN, p-AKT served as the standard of AKT; n = 6 per group. i, k-n Representative and quantification immunoblot analysis of PTEN, AKT, p-AKT Thr308, and p-AKT Ser473 in mTECs following the administering of 10 μg/ml AA for 0, 6, 12, 24, and 48 h separately. GAPDH served as the standard of PTEN, p-AKT served as the standard of AKT; n = 6 per group. j, o-p Representative and quantification immunoblot analysis of E-cadherin, N-cadherin, and in mTECs following the administering of 10 μg/ml AA for 0, 6, 12, 24, and 48 h separately. GAPDH served as the standard; n = 6 per group. q-v Immunofluorescence images and quantification of mean fluorescence intensity for ZO-1, Vimentin, and α-SMA from control group and after treatment with 10 μg/ml AA for 48 h in mTECs. We randomly choose five microscopical field per sections, n = 3 holes per group. Scale bars, 10 μm. P < 0.05; **P < 0.01; ANOVA. RNA isolation and real-time RT-PCR Total RNA from mTECs, kidney tissues and biopsy specimens were extracted using Trizol, and reverse transcribed into cDNA using PrimeScript TM RT reagent kit. The 18S rRNA gene was used to normalize gene expression. Expression of miR-382 was detected using Taqman probes, using U6 to normalize miR-382 expression. Primer sequences used for PCR are shown Supplementary Table 1. Statistical analysis All in vivo and in vitro experiments were performed in biological replicates. Data were analyzed using GraphPad Prism Software, and expressed as the mean ± standard error of the mean. Two-tailed, unpaired Student's tests were performed to detect differences between two groups. The variance between the groups that are being statistically compared is similar. Statistical significance was set at P < 0.05. Fig. 7 Possible working model of NF-κB/miR-382/PTEN/AKT signaling pathway in the progression of AA-induced AKI-to-CKD. a AA stimulation activated NF-κB, which acts as a transcriptional factor of miR-382 and promoted miR-382 expression. MiR-382 negatively regulated PTEN by combining with 3′UTR of pten. Inhibition of PTEN-activated AKT signaling pathway and then played roles in the progression of EMT, release of inflammatory cytokines and deposition of ECM. b Proposed the possible mechanism of AKI-to-CKD. AA treatment induced tubule injury, which would recovery after moderate tubule redifferentiation or would regulate NF-κB/miR-382/PTEN/AKT signaling axis resulting in sustained redifferentiation, EMT and expression of profibrotic signaling. In the process, inflammation response also promoted EMT process and prevented tubule recovery and finally led to tubule atrophy, deposition of ECM and renal interstitial fibrosis.	2020-08-15T14:40:39.941Z	2020-08-01T00:00:00.000	{ "year": 2020, "sha1": "82f86b4630e2d08e9a500c4f6e6dc0c4366c48b5", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41419-020-02876-1.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "82f86b4630e2d08e9a500c4f6e6dc0c4366c48b5", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
24709646	pes2o/s2orc	v3-fos-license	Baseline mineralizing surface determines the magnitude of the bisphosphonate effect on cortical bone mineralization in postmenopausal osteoporotic patients. PURPOSE To determine the effect of short- or long-term bisphosphonate treatment on cortical bone mineralization density distribution (BMDD). METHODS BMDD was assessed by quantitative backscatter electron imaging in postmenopausal osteoporosis: in paired transiliac biopsy samples (n=36) at baseline and after 3 years risedronate treatment from a clinical study, in transiliac biopsy samples from patients who were treated with either risedronate (n=31) or alendronate (n=68) for 3 to 7 years from an observational study. Outcomes were related to premenopausal reference data (n=73) and to histomorphometric mineralizing surface per bone surface (MS/BS). RESULTS In the clinical study, patients with lower (below cohort median) MS/BS had normal cortical CaMean at baseline. After 3 years risedronate, their CaMean was not different versus baseline but increased versus reference (+2.9%, p=0.003). Among the groups of the observational study, CaMean did not exceed reference level, was similar for alendronate versus risedronate and similar between 3 to 5 years versus longer than 5 years treatment duration. CONCLUSION Baseline bone mineralizing surface appears to be important for the effect of bisphosphonate on cortical bone mineralization. In patients with lower baseline MS/BS, level of mineralization after treatment can exceed reference level. Whether this is beneficial in the long-term is unknown. Introduction The fracture risk reduction and increase in bone mineral density (by dual X-ray absorptiometry) resulting from bisphosphonate treatment are well known [1][2][3] . Although high bone turnover appears to be the typical situation immediately after menopause 9,10 , there is variability in bone turnover later in the postmenopausal period ranging from highly increased to even decreased bone turnover [11][12][13] . The effect of the level of bone turnover before treatment on outcomes in cortical bone matrix mineralization in postmenopausal patients has not been studied previously. About 80% of the entire skeletal bone mass is cortical bone, which supports most of the mechanical load applied to bone 17 . Loss of cortical bone plays a pivotal role in bone fragility 18 . Material characteristics of cortical bone and potential treatment changes thereof are essential for the mechanical integrity of the skeleton. Bone mineralization density distribution (BMDD) is a bone material quality variable which reveals the degree and the distribution of the mineral content in bone 16 that in turn is known to influence the stiffness/elasticity of mineralized tissues 14,15 . The aim of the present work was to study the effects of 3 years treatment with risedronate on cortical BMDD measured in paired transiliac bone biopsy samples in postmenopausal osteoporotic patients from the clinical VERT-NA (Vertebral Efficacy with Risedronate Therapy North American) trial 2 in relation to baseline bone formation by MS/BS. In addition, cortical BMDD was assessed in transiliac biopsies from postmenopausal osteoporotic patients in a separate observational study in which the patients had been treated with risedronate or alendronate for more than 3 years 19 . Patients We analyzed paired transiliac bone biopsy samples from the biopsy cohort of the clinical VERT-NA 2 trial. The patients were postmenopausal osteoporotic women who had either two prevalent vertebral fractures at baseline or one prevalent vertebral fracture and a lumbar spine mineral density T-score of -2 or less. Paired bone biopsies were obtained at baseline and after 3 years with risedronate treatment. As from these patients baseline histomorphometric indices of bone formation/turnover obtained from cancellous bone (mineralizing surface per unit bone surface MS/BS; mineral apposition rate MAR; bone formation rate per bone volume BFR/BV) were available, we studied their relationship with BMDD outcomes. Furthermore, we separated the biopsy cohort into two sub-cohorts with higher turnover (HT, n=10) and with lower turnover (LT, n=8) according to their baseline MS/BS above or below the cohort median of the original study (MS/BS of 6.25%) 20 . Considering these subgroups enabled us to study the effects on cortical BMDD dependent on baseline MS/BS. Cancellous BMDD outcomes from the present HT cohort have been published previously 4,6 . Additionally, we analyzed transiliac biopsies from an observational study 19 in which postmenopausal osteoporotic women had been treated with either alendronate (n=68) or risedronate (n=31). In this study, groups were separated by treatment duration, either 3-5 years (alendronate 3-5, risedronate 3-5) or >5 years (alendronate >5, risedronate >5). For mean duration of treatment and clinical characteristics of the groups see Table 1. In this study biopsies were taken only after treatment, not at baseline. Cortical BMDD outcomes were correlated with MS/BS (standard histomorphometry on undecalcified biopsies was performed as previously described 22,23 ) and compared to previously published healthy premenopausal reference data of cortical bone (n=73) 21 . All studies were conducted according to the Declaration of Helsinki and approved by the local ethics committees or institutional review boards (IRB). Quantitative backscatter electron imaging (qBEI) In calibrated qBEI images of bone (as shown in Figure 1) the backscatter electron intensities/pixel grey levels are correlated with the local amount of calcium (Ca) in bone. Thus the grey level histograms derived from calibrated backscatter electron images of bone tissue provide information about the frequency distribution of pixels with a certain Ca weight% content (so called bone mineralization density distribution, BMDD), described in detail in a previous work 24 . Calibrated qBEI images were acquired with a DSM 962 digital scanning electron microscope (Zeiss, Oberkochen, Germany, equipped with a four-quadrant semiconductor backscatter electron detector) based on the following microscope settings: an accelerating voltage of 20 kV, a probe current of 110±0.4 pA (measured with a Faraday cup connected to a picoamperemeter), a working distance of 15 mm, and a scan speed of 100 seconds per frame. The entire bone tissue area was recorded in a series of images with a 50x nominal magnification corresponding to a scanned bone area of about 2 mm x 2.5 mm. BMDD was obtained from cancellous and cortical bone separately ( Figure 1). We used five parameters to characterize the BMDD curves: CaMean= the weighted mean Caconcentration of the bone area; CaPeak= the peak position of the histogram, indicating the most frequently measured Ca concentration; CaWidth= the full width at half maximum of the distribution, describing the heterogeneity in matrix mineralization; CaLow= the percentage of lowly mineralized bone areas (lower than 17.68 weight % Ca 25 ); CaHigh= the percentage of highly mineralized bone areas (higher than 25.30 wt% Ca, corresponding to fully mineralized bone areas, mainly interstitial bone 16 ). For cortical BMDD, the entire areas of the two cortical plates were analyzed separately and subsequently the arithmetic mean of the BMDD of the two plates was calculated for each sample (i.e. cortical BMDD was evaluable in intact biopsy samples), as described previously 21 . In the present work, we focussed on cortical BMDD, thus all presented BMDD outcomes are referring to cortical bone, except for the correlation between cortical CaMean versus cancellous CaMean (in this case "cortical" and "cancellous" is explicitly indicated). Statistical analysis The data are presented as mean (SD) (normally distributed data) or median (25 th ; 75 th percentiles) (non-normally distributed data). Statistical analysis was performed using SigmaStat for Windows Version 2.03 (SPSS Inc.). Normality of the data was analyzed by the Kolmogorov-Smirnov test. Comparison between HT and LT at baseline or after 3 years risedronate treatment, as well as comparison of each HT and LT versus premenopausal reference was analyzed by t-tests. Effect of treatment with risedronate in HT and LT patients was analyzed by paired t-tests. Comparison among patients from the observation study as well as comparison of them with clinical HT and LT groups is based on ANOVA or ANOVA on ranks (Kruskal-Wallis One Way Analysis of Variance on Ranks) for nonnormally distributed data. Differences among alendronate or among risedronate as well as comparison of alendronate or risedronate with premenopausal reference are based on t-tests or Mann-Whitney rank sum tests for non-normally distributed data. Correlation analyses of cortical CaMean versus cancellous CaMean and cortical Ca Mean versus MS/BS are based on Pearson correlation. Cortical BMDD from risedronate treated clinical study patients i) BMDD and its relation to MS/BS at baseline At baseline, BMDD outcomes were significantly correlated with MS/BS in the total patient cohort. CaMean (ii) Comparison between HT and LT subgroups at baseline CaMean in HT was significantly lower compared to LT (-5.0%, p<0.001) and compared to reference (-4.5%, p<0.001). Similarly, CaPeak and CaHigh were significantly lower versus LT and versus reference, while CaLow was significantly higher compared to both LT and reference (comparison between HT and LT is summarized in Figure 2, comparison to reference is shown in Figure 3). CaWidth in HT was not significantly different from LT or reference. At baseline, BMDD from LT was not significantly different from reference. After 3 years risedronate treatment, CaMean in HT was still lower than in LT (-3.5%, p=0.010) but was not different from reference. CaPeak, and CaHigh from HT remained lower than in LT ( Figure 2) and reference (Figure 3), while CaWidth and CaLow from HT were similar to those in LT (Figure 2) but were significantly lower than reference (Figure 3). After 3 years risedronate, BMDD in LT was significantly different from premenopausal reference. In particular, CaMean was higher (+2.9%, p=0.003), and CaPeak, and CaHigh were also higher (Figure 3), while CaWidth and CaLow were significantly lower than reference (Table 2 and Figure 3). For comparisons of 3 year risedronate treatment with reference see Table 2. Cortical BMDD from observational study patients i) Effect of short-or long-term risedronate or alendronate treatment on BMDD BMDD variables were available in n=27 alendronate 3-5, n=26 alendronate >5, n=21 risedronate 3-5, and n=5 risedronate >5. Among the 4 treatment groups there were no significant differences in BMDD outcomes ( Figure 3). Also, there was no significant difference due to treatment duration with either alendronate or risedronate treatment (3-5 versus >5 within drug treatment groups). Compared to premenopausal reference, alendronate 3-5 had lower CaMean, CaPeak, and CaHigh, and alendronate >5 had lower CaPeak. The only difference in risedronate from reference was lower CaPeak in risedronate 3-5 (see Table 2; Figure 3). Considering all observational study groups together, CaMean was significantly correlated with MS/BS (R=-0.31, p<0.01), showed a near to significance correlation with BFR/BV (R=-0.23, p=0.06), and it was not significantly related to MAR (p=0.74, n.s.). (ii) Comparison of observational study groups with clinical study groups after risedronate treatment Comparison of the observational study groups with HT and LT from the clinical study after 3 years with risedronate showed that although the patients from the observational study were treated for a longer time, they had similar CaMean and CaPeak compared to HT. Moreover, they had significantly lower CaMean and CaPeak (ANOVA p<0.01, post-hoc tests p<0.05) compared to the LT patients after 3 years with risedronate. All four treatment groups from the observational study had larger CaWidth compared to HT and LT from the clinical trial (ANOVA p<0.001, posthoc tests <0.01). Differences were also found for CaLow (ANOVA p=0.025), which was larger in alendronate 3-5 and risedronate 3-5 compared to both LT and HT. CaHigh was similar among the groups. . Cortical BMDD outcomes from HT, LT (baseline and risedronate 3 yrs) and from the observational study (alendronate and risedronate 3-5 and >5 years). White dotted lines and grey areas indicate premenopausal reference mean (or median) and reference range (±1 standard deviation or interquartile range) respectively. Reference data are from a previous work 21 . *p<0.001, p<0.01, *p<0.05 versus premenopausal reference. LT base= lower turnover patients baseline, HT base= higher turnover patients baseline, LT 3= lower turnover patients after 3 years with risedronate, HT 3 = higher turnover patients after 3 years with risedronate, ALN 3-5= patients from the observational study after 3-5 years with alendronate, RIS 3-5= patients from the observational study after 3-5 years with risedronate, ALN >5 = patients from the observational study after more than 5 years with alendronate, RIS >5 = patients from the observational study after more than 5 years with risedronate. iii) Relationship between cortical CaMean and cancellous CaMean in observational and clinical study groups Cortical CaMean was significantly correlated with cancellous CaMean in both HT and LT patients at baseline and after 3 years risedronate (clinical study) as well as in patients after 3 or more years with risedronate or alendronate (observational study). All correlations were similar (correlation coefficient and slope) and statistically significant (R ranging from 0.77 to 0.98, p ranging from 0.003 to <0.001), see Figure 4 a, b, c. Discussion The biopsies from the clinical study gave evidence for the dependency of the degree of bone mineralization on MS/ BS and BFR/BV at baseline. Additionally, the level of MS/ BS (lower or higher than cohort median) at baseline was decisive of the magnitude of the bisphosphonate treatment effect. The biopsies from the observational study (where no baseline biopsy was available), revealed the effect of shortto long-term alendronate or risedronate on cortical BMDD in a non-controlled clinical setting. We found that the degree of cortical bone mineralization remained unchanged betwen short-and long-term treatment and did not exceed reference levels even after long-term bisphosphonate treatment. High bone turnover in osteoporosis leads to reduced average tissue age which is associated with decreased degree and increased heterogeneity of cancellous bone matrix mineralization 7,26 . This is consistent with the observed correlations of BMDD outcomes with MS/BS and BFR/BV in the present work. Calcium and vitamin D insufficiency might further contribute to reduced degree of mineralization 4,6 . Consistent with aforementioned findings in cancellous bone, HT patients revealed a clear shift of the cortical BMDD to lower calcium concentrations and an increase in mineralization heterogeneity compared to both LT patients and premenopausal reference data. In contrast, the LT patients at baseline were not different from reference cortical BMDD. At first glance this was somewhat surprising, because the MS/ BS in LT was lower than previously reported MS/BS for healthy Table 2. Summary of deviations in cortical BMDD from reference values after 3 or more years with bisphosphonate treatment. Clinical trial patients Observational study patients 3yrs RIS in HT 3yrs RIS in LT 3-5yrs ALN 3-5yrs RIS >5yrs ALN >5yrs RIS CaHigh ↓ ↑ ↓ normal normal normal ↓ = significantly lower, ↑ = significantly higher compared to premenopausal cortical reference from 21 . Normal = no significant difference to premenopausal cortical reference 21 . postmenopausal women 27 . However, when compared to more recently reported other cohorts of healthy premenopausal or postmenopausal women 28,29 , our LT patients seem to have bone turnover in the lower normal range. The normality of BMDD in LT suggests also that bone fragility in this patient cohort is caused by factors (e.g. microarchitecture) other than deviations in bone matrix mineralization. It is well known that in high bone turnover patients the sudden reduction of turnover by bisphosphonates leads to an increase in the degree and a decrease in the heterogeneity of mineralization 5,7,8,30,31 . The latter is a transient effect which is observed after short term (about one to three years) therapy 6,30 . The treatment changes in cortical bone from the HT patients are in line with the aformentioned reported observations. The average degree of mineralization was even normalized in HT. The effects of further bone turnover reduction on mineralization in patients with low baseline level of bone turnover (by MS/BS) has not been previously reported. In the present study, we observed a smaller, nonsignificant increase in CaMean in LT. This is consistent with the observation that the increase in bone matrix mineralization after bisphosphonate treatment is proportional to the deviation of baseline level of mineralization from reference 4,32 . However, the increase in mineralization was sufficient to raise the average and typical degree of mineralization as well as the percentage of highly mineralized bone areas in LT beyond reference levels. The only previously reported observation of mineralization beyond the reference level was in a study after zoledronic acid treatment 33 . Fracture risk and bone mineral density studies have shown that bisphosphonates are beneficial to both higher and lower turnover postmenopausal patients 20,35,36 , with greatest relative increases in bone mineral density in patients who had the highest bone turnover at baseline 34,36 . As bisphosphonate treatment was generally not observed to cause substantial increases in bone volume, changes in the bone material quality in addition to the increase in bone mineral density might contribute to the decrease in fracture risk 37 . A previous work, which focussed on cancellous bone, suggested that the main contribution to the increase in bone mineral density was the increase in mineralization 4 . Thus the observed increases in cortical bone matrix mineralization in HT and LT patients in the present study are in line with the previously reported increases in bone mineral density in both groups 20 . Although fracture risk reduction has been reported after bisphosphonate in patients with lower turnover before treatment 35 , it remains an open question whether the moderate increase in bone mineralization above reference levels after bisphosphonate treatment in the LT patients is beneficial in the long-term or indicates unfavourable changes in the bone matrix. In the observational study, we found that BMDD outcomes were dependent on MS/BS after 3 to 7 years antiresorptive treatment, but were not correlated with MAR. This latter finding is in agreement with the VERT-NA trial patients at baseline. There was no increase in the degree of mineralization beyond reference even after longer treatment duration. In contrast, after long-term alendronate treatment, CaPeak was below the reference level. An explanation would be a lower compliance to treatment in this non-clinical setting or alternatively, the majority of patients had high bone turnover with low degree of matrix mineralization at baseline and the treatment has shifted the degree of mineralization towards reference values. The similarity of BMDD between the 3-5 years and >5 years groups in the present data and previous observations suggests that no further substantial changes in bone mineralization and other material characteristics are expected to occur after 3-5 years of bisphosphonate treatment 6,38,41 , which is the typically recommended duration of bisphosphonate treatment for patients at low or moderate risk of fragility fractures 39 . An interesting finding in the present work was the nearly 1:1 correlation of cortical CaMean versus cancellous CaMean in all study groups. This indicates a close relationship between the mineralization of the cancellous and the cortical compartments also in osteoporotic individuals similar to that found in healthy individuals 21 . The striking similarity in average degree of matrix mineralization/tissue age between cortical and cancellous bone of the iliac crest in each patient was neither influenced by the level of bone turnover (as the correlation was similar in LT and HT patients before treatment) nor by bisphosphonate treatment. Summary: The effects of bisphosphonate treatment on cortical bone mineralization were greater in patients who had higher turnover and lower bone mineralization at baseline. After 3 years of bisphosphonate treatment BMDD parameters were either shifted toward reference or even normalized in higher turnover patients. These outcomes after 3 years bisphosphonate treatment in HT patients were similar to the BMDD outcomes in the observational study patients after 3 to 7 years bisphosphonate treatment. In contrast, in patients with lower bone turnover at baseline, the small increase in mineralization due to treatment led to cortical bone mineralization above reference levels.	2017-10-28T02:39:12.095Z	2017-09-01T00:00:00.000	{ "year": 2017, "sha1": "884498a9fbeb75815d9ead93f491ac3d1d1319a7", "oa_license": "CCBYNCSA", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "884498a9fbeb75815d9ead93f491ac3d1d1319a7", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
221614455	pes2o/s2orc	v3-fos-license	Long-term clinical outcomes after Descemet Membrane Endothelial Keratoplasty (DMEK) in Irido-Corneal Endothelial Syndrome Purpose To evaluate the long-term clinical outcomes after Descemet Membrane Endothelial Keratoplasty (DMEK) in Irido-Corneal Endothelial Syndrome (ICE). Observation Four eyes of four patients diagnosed with ICE syndrome were treated with DMEK. Postoperatively, best corrected visual acuity (BCVA) and central endothelial cell density (ECD) were documented at 6, 12, 24 and 36 months for all the cases. All procedures were uneventful. Average follow-up time was 36 months. BCVA improved in all eyes. Mean BCVA improved significantly from 1.54 ± 0.71 log MAR preoperatively to 0.11 ± 0.14 logMAR at the final follow-up. Average donor ECD was 2895 ± 357 cells/mm2 preoperatively and 1992 ± 321 cells/mm2, 1816 ± 395 cells/mm2, 1571 ± 299 cells/mm2 and 1305 ± 246 cells/mm2 at 6, 12, 24 and 36 months after DMEK surgery respectively. This represented an average endothelial cell loss (ECL) of 31.3%, 37.7%, 46.8% and 55.1% at 6, 12, 24 and 36 months respectively. Postoperative intraocular pressure (IOP) rise was seen in 3 eyes at 1 month which normalized under topical antiglaucoma medications. Conclusion DMEK is a relatively safe procedure providing favourable clinical outcomes in eyes with ICE syndrome. Since angle closure is progressive in these condition, regular IOP monitoring and glaucoma control is critical for long term survival of the graft. Importance Till date management of ICE syndrome has always been a great challenge due to its varied presentation and complex anatomical abnormalities. Replacing the endothelial cells in an irregular anterior chamber poses additional difficulty. Even well-trained DMEK surgeons find it difficult to appose the Descemet's Membrane (DM) in such a scenario and we in this article provide key surgical tips for successful long term management of these cases. Introduction The iridocorneal endothelial (ICE) syndrome is a rare ocular disorder characterized by proliferative and structural abnormalities of the corneal endothelium, progressive obstruction of the iridocorneal angle, and iris anomalies such as atrophy and hole formation. 1 It comprises a spectrum of clinical entities: Progressive Essential Iris Atrophy, Cogan-Reese Syndrome, and Chandler's syndrome. 2 The surgical management of ICE syndrome has always been a challenge in the past years. Outcomes with Penetrating Keratoplasty (PK) in ICE syndrome have been relatively poor with high rates of rejection and endothelial failure. 3 Endothelial keratoplasty (EK) is a surgical procedure that selectively replaces dysfunctional endothelium, sparing the corneal stroma and epithelium. This surgical technique offers several advantages for the treatment of corneal edema in ICE syndrome when compared with PK, such as rapid visual recovery with minimal refractive changes, avoiding the use of sutures and better maintainencei of corneal recipient integrity and innervation. Price et al. found that selective replacement of dysfunctional endothelium with Descemet's Stripping Endothelial Keratoplasty (DSEK) can successfully treat corneal edema and associated visual loss and pain caused by ICE syndrome. 4 DSEK has the potential to provide good short-term visual outcomes in eyes with ICE syndrome. However, long-term graft survival beyond 2 years is poor because of late endothelial failure. 5 Descemet membrane endothelial keratoplasty (DMEK) allows for selective replacement of damaged endothelial cells, using only donor Descemet's membrane with endothelium. 6 There is sufficient evidence to demonstrate that DMEK is superior to DSEK in achieving a faster visual recovery, a better visual outcome and a lower immune rejection rate. Currently, DMEK is performed mostly in eyes with Fuchs' endothelial dystrophy and uncomplicated bullous keratopathy. 7 Even though DMEK has been shown to be performed successfully in eyes with prior vitrectomy and trabeculectomy or glaucoma drainage device placement, anterior chamber intraocular lens, large iris defect, or absence of lens support, still DSEK remains the preferred procedure to treat eyes with complex endothelial dysfunction and abnormal anatomy. Except for Sorkin et al. who first reported the applicability of DMEK in cases of ICE syndrome and Posterior Polymorphous Corneal Dystrophy (PPCD) with excellent early outcomes, there are no studies establishing the long term outcomes of DMEK in eyes with ICE syndrome. 8 In our series, we report the long term clinical outcomes of Descemet membrane endothelial keratoplasty (DMEK) for the treatment of corneal decompensation in patients with ICE syndrome. Four eyes of four patients diagnosed with ICE syndrome were treated with DMEK. Their case summary is as follows: Case 1 A 38 year old female presented with complaint of gradual diminution of vision in right eye with blurring more in the waking time. Her BCVA was 20/80. Ocular examination showed a mild microcystic corneal edema with bullae and shallow anterior chamber with irregular depth due to peripheral anterior synechiae (PAS) and corectopia suggestive of ICE Syndrome (Fig. 1a). Other eye was normal. DMEK was done and postoperatively her BCVA was 20/20 at 3 months. At about the end of 2 years she developed mild anterior uveitis, which resolved after hiking the steroid levels for a month. Clinically cornea was clear (Fig. 1b) till her recent 3 year follow up with Endothelial cell density (ECD) of 1621 cells/mm 2 (Fig. 1c). Case 2 A 55 year old male was referred after phacoemulsification and intraocular lens implantation in right eye and the referring surgeon had noted down patches of iris atrophy and PAS preoperatively. On presentation to us, his BCVA was 20/600 in right eye with cornea showing signs of corneal decompensation with extensive PAS and iris atrophic patches suggestive of ICE syndrome (Fig. 2a). Other eye was normal. We performed DMEK for the right eye. Postoperatively, on day 1, the AC was shallow as the air had gone behind the iris. Air bubble was released back into the AC and 1% pilocarpine was applied to constrict the pupil. Subsequently the cornea cleared ( Fig. 2b) and at 3 years the patient retained a BCVA of 20/20 and ECD of 1302 cells/mm 2 (Fig. 2c). Case 3 A 52 year old female presented with recurrent episodes of pain, redness and defective vision in right eye with BCVA of 20/600. Ocular examination showed diffuse microcystic corneal edema with peripheral anterior synechiae, peculiar aberrant iris vessels and complicated cataract suggestive of Chandler's Syndrome (Fig. 3a). We planned for DMEK with phacoemulsification and intraocular lens implantation. Iris vessels were cauterized with a pencil tip wet cautery used in retinal surgeries prior to phacoemulsification to prevent any inadvertent bleeding. Postoperatively patient had mild bleeding till 1 month, which spontaneously resolved and she was followed up for 3 years ( Fig. 3b) with no further episodes of bleeding and a BCVA of 20/40 and ECD of 1020 cells/mm 2 (Fig. 3c). Case 4 A 50 year old female was referred for corneal decompensation in left eye post cataract surgery. In the left eye visual acuity was Hand Movements only and the ocular examination showed severe corneal edema with scarring and iris atrophy, PAS and corectopia (Fig. 4a). Her right eye also showed similar findings confirming the rare diagnosis of bilateral ICE syndrome (Fig. 4b). DMEK was done in left eye following which her BCVA improved to 20/30 at 6 months and remained stable at 3 years follow up ( Fig. 4c) with an ECD of 1278 cells/mm 2 (Fig. 4d). Methods and Materials All the four patients of ICE syndrome underwent DMEK at our institute. Diagnosis was based on clinical slit-lamp findings and since the corneas were hazy preoperative specular was not possible in all the four cases. Preoperatively all patients underwent the Best Corrected Visual Acuity (BCVA) using the Snellen chart, slit lamp examination, Goldman's Applanation Tonometry (GAT) and dilated fundus examination. Ultrasonography (USG) B-scan was done in eyes where the fundus details were not clearly visible. All DMEK grafts were harvested at Sankara Eye Bank, Coimbatore and stored in organ culture medium (Cornisol) until transplantation. Postoperatively, BCVA and central ECD were documented at 6, 12, 24 and 36 months. Donor tissues above 50 years were selected and we strictly adhered to our regular DMEK surgical protocol to ensure a smooth transfer and correct attachment of DM in these complicated cases. Our protocol included donor DM preparation in the operating room (OR) by the surgeon using a single pull technique with a curved non toothed forceps to shorten the peel time (3-5 minutes) and reduce the endothelial cell loss. The stripped DM was marked on the stromal side with an L-shaped stamp to identify the endothelial side during unfolding in the AC. The prepared DM was injected into the recipient using a self-made injector which was assembled using a 1 cc syringe, regular C-type intraocular lens (IOL) cartridge and a sterile IV tubing. After unfolding DM in the correct orientation the AC was completely filled with air for 1 hour and was subsequently burped till the eccentric pupillary margin was reached to avoid pupillary block glaucoma. Results All four patients were diagnosed with ICE syndrome. 2 eyes (50%) were phakic and 2 eyes (50%) were pseudophakic. One of the phakic eyes had cataractous lens, for which DMEK was combined with phacoemulsification with implantation of an intraocular lens. Rest 3 patients underwent only DMEK. All procedures were uneventful. Average follow-up time was 36 months. BCVA improved in all eyes (Table 1). Mean BCVA improved significantly from 1.54 ± 0.71 LogMAR preoperatively to 0.11 ± 0.14 LogMAR at the final follow-up. Discussion Long term outcomes of DMEK is well established in most endothelial conditions, but to the best of our knowledge, this is the first reported series with long term outcomes after DMEK in ICE syndrome. All four patients achieved an excellent visual outcome with BCVA of 20/20-20/40 at 6 months after DMEK. Fajgenbaum et al. 5 showed an improvement of BCVA from 20/20-20/40 in 7 of 9 grafts at 6 months post op following DSEK in ICE patients. But subsequently they reported a significant fall in BCVA with only 12% of grafts having BCVA of 20/40 or better at 24 months follow up. Again, their ECL over 1 year was 78% and seven of the nine grafts failed because of late endothelial failure after a mean of 18 months. They proposed that this increased perioperative and late endothelial cell loss observed in eyes with ICE to be multifactorial. They attributed that excessive perioperative endothelial cell loss which occurred during insertion and positioning of the DSEK tissue as the most significant factor which limited the graft survival in their series. Also, associated glaucoma surgeries and a fragile blood-aqueous barrier which increases the tendency to inflammatory and rejection episodes can cause disturbances in ICE syndrome and contribute to medium to long-term DSEK failure. We feel that the insertion of an additional tissue of 150-200 μm into an already irregular and cramped anterior chamber promotes iris touch, PAS formation and subsequent inflammation and glaucoma also a contributing factor for graft failure in DSEK. Price et al. described 3 cases of ICE syndrome that were managed with Descemet's Stripping Automated Endothelial Keratoplasty (DSAEK) surgery, 4 showing excellent visual outcomes till 14 months. All the patients achieved a BCVA of 20/20 to 20/30 within 6 months with no complications. However, they pointed out that DSAEK can be more challenging to perform in eyes with ICE syndrome and PAS, because the shallow anterior chamber and iris abnormalities may make it more difficult for the donor tissue to be inserted through a small incision and unfolded correctly. Buxton et al. 9 stated that keratoplasty, rather than filtering surgery, is probably the initial surgical procedure of choice when the corneal edema is extensive and the intraocular pressure moderately increased and not associated with any objective glaucomatous changes. The largest series of PK grafts in ICE syndrome by Alvim et al. reported that overall prognosis of PK was favourable, but the patients required multiple corneal and glaucoma procedures. Also the allograft rejection episodes were found to be very frequent after PK for ICE syndrome (79%), especially after glaucoma surgery. 3 However in our study, there were no primary graft failure or rejections episodes during the study period. Also we used Prednisolone acetate 1% suspension in tapering doses in the postoperative period during the first 6 months and later switched over to Loteprednol Etabonate 0.5% gel which is equally effective in preventing immunologic graft rejection episodes after DMEK and was significantly less likely to raise IOP. 10 Postoperative IOP rise was seen in 3 eyes (up to 25 mm Hg) at 1 month which normalized under topical medical treatment with Timolol Maleate 0.5% eyedrops given twice daily. In eyes with pre-existing glaucoma or in high-risk eyes with complex anterior segment changes such as eyes with ICE syndrome there is a higher risk of uncontrolled IOP after EK which in turn also has adverse effects on the donor endothelial cells and graft survival. 11 Whenever DMEK is performed in such complex eyes, re-bubbling rates may be as high as 50%, and secondary graft failure occurs in up to 75% these eyes. 12 Unfortunately, data on the outcomes of DMEK in eyes with glaucoma are limited. Treder et all 13 found that there was no significant difference in BCVA and IOP during the early post-operative period (3 months) in patients with and without a pre-existing glaucoma who had received DMEK surgery. Aravena et al. 14 published the largest series of consecutive DMEK in patients with previous glaucoma surgery. They reported that DMEK provides excellent visual improvement without an increase in early postoperative complications in eyes who had undergone previous trabeculectomy and glaucoma drainage device implantation and there may be an advantage performing DMEK in these cases, but suggested further long term studies investigating the impact of glaucoma surgery on the endothelial cell loss and graft survival of DMEK is required. DMEK provides near perfect anatomical replacement of tissue and as there is not much change in the AC and angle dynamics, our cases remained quiet all through with IOP under control with medications. Surgical outcomes in ICE syndrome is summarized (Table 3). Till date published data on DMEK in ICE syndrome is limited. Weller et al. 12 in 2015 studied the feasibility and outcomes of DMEK in complex anterior segment and vitreous diseases. DMEK was performed in 3 eyes with ICE syndrome with a follow up of 6 months. Sorkin et al. 8 recently published their clinical outcome of DMEK in cases of corneal decompensation secondary to ICE and PPCD. Both the studies found favourable short term outcomes in their series and they proposed future prospective studies to compare DMEK, DSAEK, and possibly also pre-Descemet's endothelial keratoplasty for the treatment of ICE or PPCD patients, including long-term follow-up. DMEK is equally tough to perform in such complex anterior segment situations, but we still feel it is easier than DSAEK in ICE syndrome. DMEK offers 3 distinct advantages that render it an appealing procedure to treat patients with ICE syndrome. First, section in DMEK is small (2.8mm-3.00mm) so that the incision can be planned in an area avoiding the PAS. Second as the AC is already shallow, unfolding the DM in the AC is easier than in routine DMEK cases. Thirdly, the need for stronger dosage and the frequent usage of steroids is reduced and the chance of IOP rise in these ICE patients is less, who are otherwise predisposed to develop glaucoma. However, we propose that surgery should be done only by an experienced surgeon who has previously performed DMEK successfully in complex anterior segment situations. We also suggest a few adaptations in the donor selection and sizing and the surgical technique which will help the surgeons to perform DMEK without complications in these complex cases. The donor age should above 55 years, as it will unfold easily without much manipulation in an irregular AC. The donor DM sizing is critical and should not be based on the recipient's corneal diameter and should solely depend on the extend of PAS and the nonadherent central cornea. If the donor DM diameter is more, then the peripheral DM scroll will not be able to unfold due to the PAS, leading to detachment in the immediate post-operative period. Never attempt to release any adherent iris or perform gonio-synechiolysis, as there are high chances of recurrences due to iris memory and also bleeding can be profuse such that the surgery may need to be abandoned. Intracameral Pilocarpine 0.2% can be used to constrict the pupil to prevent lens damage while manipulating the graft. Postoperatively after 1 hour, the pupil has to be dilated and air is burped to cross the eccentric pupil to prevent acute spike in IOP and prevent further PAS formation. Conclusion DMEK is a safe and successful procedure providing favourable clinical outcomes in eyes with ICE syndrome. A few surgical adaptations in the technique yields results comparable to DMEK performed in variable indications. Since angle closure is progressive in these condition, regular IOP monitoring and glaucoma control is critical for long term survival of these grafts. Patient consent Consent to publish the case report was not obtained. This report does not contain any personal information that could lead to the identification of the patient. Funding No funding or grant support Authorship All authors attest that they meet the current ICMJE criteria for Authorship. Declaration of competing interest None of the authors have any financial disclosures	2020-08-27T09:13:38.381Z	2020-08-27T00:00:00.000	{ "year": 2020, "sha1": "b5c91af895c96087dccebff09d456555359f802b", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1016/j.ajoc.2020.100894", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "c037fb78249e457828bd005066e07e11e433db0b", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
268462263	pes2o/s2orc	v3-fos-license	Research on the Digital Protection and Inheritance Path of Liao Dynasty Ancient Pagodas . Under the background of the digital intelligence era, this paper discusses the significant importance of the digital protection and inheritance of cultural heritage of ancient pagodas from the Liao Dynasty, expounds the guiding ideology of digital protection and inheritance, and proposes the method of comprehensive utilization of digital technologies such as data collection, database construction, digital modeling, digital maps, and digital museums. These methods aim to promote the digital protection and inheritance of Liao Dynasty ancient pagodas and provide innovative approaches for the dissemination of their history, art, and culture. Introduction The ancient pagodas of the Liao Dynasty represent the cultural heritage of the ground structures from the Liao period in Liaoning Province.They hold an important position in the study of history, ancient architecture, and multi-ethnic culture in the Northeast and North China regions.The pagoda buildings in the Liaoning area, where the Khitan people once lived, are included as cultural relics protection units of various levels, with a total of 40 Liao pagodas.As an outstanding example of ancient Chinese architecture, Liao pagodas possess profound artistic value.Their architectural style, decoration, structure, and cultural background all endow them with unique artistic charm.However, with the rapid development of internet thinking and digital technologies, the period for the digital protection and inheritance of Liao Dynasty ancient pagodas has arrived. Significance of the Digital Protection and Inheritance of Liao Dynasty Ancient Pagodas The Liao Dynasty was an important period in the history of Chinese pagoda and temple construction with great historical influence.Not only were a vast number of pagodas and temples built during the Liao ruling period, but their artistic style also bridged that of the Tang and Song dynasties, featuring reasonable structures and smooth lines in architectural molding which, to some extent, influenced the style of subsequent pagoda and temple construction.Digitally protecting the ancient pagodas of the Liao Dynasty in the Liaoning region, forming complete digital archives, and disseminating and exhibiting them provides the most direct approach and the most important innovative form for the transmission of historical culture.Their research value and significance are mainly reflected in the following aspects: Ancient Pagodas of the Liao Dynasty are Precious Historical and Cultural Heritage Ancient pagodas of the Liao Dynasty are precious historical and cultural heritage.the Shifo Temple pagoda located in Shenbei New District, Shenyang City, which was built during the Xianyong period (the tenth year) as a hexagonal seven-layered dense eave brick pagoda.In the past, it was severely damaged, with only the base and part of the pagoda body remaining in the south, southeast, and northwest.Digital technology ensures that these precious cultural heritages are preserved and can be passed down even in the face of natural disasters, man-made destruction, or changes over time.Digital means are used for the digital collection, storage, restoration, display, and dissemination of ancient buildings, avoiding the damage caused by direct physical contact with historical monuments and reflecting the important manifestation of long-term and effective protection and inheritance of cultural heritage. Liao Dynasty Ancient Pagodas Contain Tremendous Economic and Cultural Values Liao Dynasty ancient pagodas can be developed as local tourism brands, attracting visitors to experience history and culture.At the same time, cities or villages around the pagodas often develop tourism infrastructure, such as hotels, restaurants, and travel agencies, to meet the needs of visitors.These investments help improve the city's infrastructure-driving tourism and economic development while also promoting cultural relics protection.They play an active role and have important significance in accelerating the inheritance of excellent historical culture, enhancing the confidence of cities and ethnicities, and promoting the cultural and tourism industry's development. 'Internet Plus' Digital Protection is the Information Bridge between Audiences and Monuments Through the internet, people directly obtain the information they want, and interesting interactive designs with digital products also enhance users' experience and attention.This breaks through the time and space limitations people face when visiting immovable cultural relics, allowing the cultural value of Liao Dynasty ancient pagodas to be shared by people worldwide. Guiding Ideology of Digital Protection and Inheritance of Liao Dynasty Ancient Pagodas By conducting resource surveys and sorting out information on the architecture, art, history, and cultural relics of the Liao Dynasty ancient pagodas, this integrates advanced digital technology organically into pagoda research work, aiming to improve the level and quality of research.Digital technology planning will cover multiple aspects such as digital collection, database construction, digital mapping, and digital display, constructing a complete digital knowledge system and technical framework.It merges the latest research findings from various disciplines and fields, thereby reducing the project's investment and risk and creating a digital work of Liao Dynasty ancient pagodas with high protection and research value.This aims for breakthrough progress in the digitalization of ancient architectural heritage, mastering a series of digital construction technologies, and promoting the progress of digital protection and inheritance of cultural heritage to ensure better protection and inheritance of these precious pagodas. Digital Protection Strategy of Liao Dynasty Ancient Pagodas Digitalization refers to the process where digital technologies in the information field comprehensively advance into all areas of human life, including the transition from traditional analog to digital modes in the communication and media fields.The Liao Dynasty ancient pagodas are massive and spectacular, inheriting the majestic style of Tang Dynasty architecture while exuding the boldness of northern ethnicities.For the digital protection and construction of Liao Dynasty ancient pagodas, the starting point should be dynamic development, holistic protection, and mutual interconnectivity.By collecting, organizing, digitalizing, disseminating, and utilizing detailed information resources, it is necessary to follow the principles of unified leadership and graded management and further promote the development of digital protection technology for cultural relics. Digital Collection Data collection, also known as data acquisition, is an important foundation and premise of the digitalization process, with the main purpose of collecting signals recognizable by computers and converting analog signals into digital signals through input devices.A comprehensive and multidimensional collection and recording of information on the historical background, roles, and humanistic environment of ancient buildings are conducted.The collected data are then sorted, classified, generalized, and analyzed.On-site data measurements of the current state, dimensions, and carvings of Liao Dynasty cultural relics are performed using three-dimensional scanning technology, and accurate data are obtained directly through research and measurement, ensuring the comprehensiveness and integrity of resource collection, which is beneficial for the value research, protection and restoration, and exhibition utilization of the pagodas. Database Construction Establish a database containing relevant information on Liao Dynasty ancient pagodas, facilitating research and management for researchers and cultural heritage conservationists.Collect relevant information on Liao Dynasty ancient pagodas, including geographical location, architectural features, historical background, etc. Set up a database structure, including the design of data and fields, data integrity constraints, to accommodate different types of information.Enter and organize previously collected Liao pagoda data, ensuring the accuracy and completeness of the data.Use the database to manage and analyze the information systematically, improving the efficiency and quality of research and protection work on Liao Dynasty ancient pagodas. Digital Modeling The production of the 3D digital model of the ancient pagodas from the Liao Dynasty is based on the data collected in the early stage, such as the dimensions and styles of various parts of the ancient buildings, as well as data on the surrounding terrain and architectural complexes.The base model can be completed using 3dsMax and ZBrush software, where 2D lines are drawn using software tools and extruded to create a three-dimensional effect.Using the Poly modeling method, commands such as insertions and extrusions are executed to create models of each part of the building.After the main building model is established, models of the surrounding scenes such as nearby buildings, terrain, and plants are also produced.In addition, a large number of pictures are collected and taken on site as material for texture maps, making the restoration work appear more realistic and achieving the goal of architectural restoration and preservation. Research on Digital Display and Dissemination of Ancient Pagodas from the Liao Dynasty Digital display and dissemination extend and expand the digital storage of ancient pagodas from the Liao Dynasty.By utilizing modern digital technology, the "productivity" and "dissemination power" of digitalization are fully harnessed.Through establishing digital maps, digital museums, digital films, digital games, etc., with a brand-new perspective and interactive experience, digital, artistic, narrative, and interactive methods are used to embark on an exploratory journey into society, history, culture, and art.The history, architectural style, sculptural art, and stories related to the Khitan culture of the Liao Dynasty pagodas are communicated globally, providing strong technical support for the inheritance of Chinese culture. Establishing a Digital Map A digital map is a map recorded and stored in digital form based on a map database with geospatial encoding.Since the pagodas from the Liao Dynasty are scattered across different regions, establishing a digital cultural map of the Liao Dynasty pagodas can present this significant cultural heritage to the public more effectively and promote the understanding and research of Liao Dynasty history and culture.GIS (Geographic Information System) software is used to create cultural maps, marking the position of each pagoda on the map and using different icons or colors to represent the types or characteristics of pagodas and the topological relationships in map space.Interactive elements are added to the map, allowing users to move, rotate, and zoom interactive objects in the scene and have an immersive experience.It is a visualization of an on-the-ground map that can comprehensively and vividly describe the terrain and landforms.It displays to the public the true geographical information of the distribution of Liao Dynasty ancient architectural resources, cultural background, visitor information, and other related information, fully showcasing the historical and cultural value of the Liao Dynasty pagodas. Establishing a Digital Museum A digital museum is a virtual museum based on digital technology, which digitizes the collections, exhibitions, and cultural resources of the museum with the help of digital media technology, the Internet, and virtual reality technology, making them more open and accessible to the public.Internet products such as websites, animations, 360 virtual panoramic museums, multimedia publications, WeChat platforms, APPs, etc., are used as dissemination forms, and multimedia materials such as audio, video, and animations are provided to enrich the content of exhibitions, breaking through spatial-temporal limitations.Transitioning from a single media display to an integrated multimedia performance enhances the presence and immersion, allowing visitors to visit the pagodas as if they were on-site, understanding their architectural style, historical background, and cultural value.Through the digital museum, the cultural heritage of the Liao Dynasty pagodas will be better protected and inherited. Creating Digital Films Digital films are created using digital cinematography, special effects, composition, and other technical means to tell stories about the history, architectural features, and legends of the Liao Dynasty pagodas.They are presented in digital media formats such as videos, animations, or anime, creating scenes with coexisting narratives to break away from the conventional presentation of historical relics, making the dissemination of the Liao pagoda's cultural heritage more vivid, lively, and touching.These short films can be played on various platforms, such as museum touch screens, social media, and entertainment systems, making the display and dissemination of cultural heritage more vivid and multidimensional.It meets the audience's multi-sensory experience with a visual core, enhancing the cultural infection power in all directions, and strengthening the observability of civilization and the sensibility of culture. Developing Digital Games The unique value of digital games in promoting excellent traditional culture is recognized.Digital media technologies such as 3D modeling, virtual reality, and augmented reality are utilized to develop digital games incorporating elements of the Liao Dynasty pagodas.Core gameplay is designed, including puzzle-solving, exploration, and educational elements.Engaging game plots are created, and the interactive experience parts of the game are developed to enable players to interact with the pagoda and its environment within the game, such as exploring the secrets within the pagoda and unraveling historical mysteries.These games help players understand and learn this part of history and experience the historical ambiance of the pagodas in a virtual environment.They are released and promoted through various channels, allowing more players to learn about and experience the game, thus achieving the purpose of disseminating the culture of the Liao Dynasty pagodas.This is also an innovative way of inheriting traditional culture. Conclusion In the digital age, multidisciplinary integration has become the mainstream trend in the protection and inheritance of cultural heritage.By recording and protecting the Liao Dynasty pagodas through digital technology and integrating the rich history and ancient architecture of the Liao era, they are being spread and displayed globally, promoting the inheritance and prosperity of ancient architectural culture.Innovating through the integration of culture and technology to promote China's excellent historical, architectural, and landscape culture is of great importance in promoting the Internet+ Chinese civilization and strengthening cultural confidence.	2024-03-17T16:04:17.842Z	2024-03-12T00:00:00.000	{ "year": 2024, "sha1": "73a27ae19260959e55cd6e66dd16751e07fea956", "oa_license": "CCBYNC", "oa_url": "https://wepub.org/index.php/TSSEHR/article/download/652/643", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "db7183d2dfb1657a9de5775ee8f54c2486f4fd02", "s2fieldsofstudy": [ "History", "Computer Science", "Art" ], "extfieldsofstudy": [] }
40768481	pes2o/s2orc	v3-fos-license	Dengue fever in a south Asian metropolis: a report on 219 cases Background and Objectives: Yearly epidemics of Dengue fever occur post-monsoon in India’s capital, Delhi. A prospective observational study was conducted during the outbreak months to understand the epidemiology and outcome of this infection and its economic impact. Materials and Methods: Febrile hospitalized (n=219) patients with dengue fever diagnosed by a combination of MAC-ELISA, GAC-ELISA and NS1Antigen-ELISA were enrolled. Epidemiologic (including economic) parameters, clinical, radiological and laboratory manifestations were noted and patients followed up over the period of hospital stay. Patient management means and outcome were recorded and analysed. Results: As per WHO-2009, 153 (69.9%) and 27 (12.3%) patients were classified as dengue with warning signs and Severe Dengue respectively while according to WHO-1997 guidelines 39 (17.8%) and 18 (8.2%) patients were classified as DHF and DSS respectively. 216 patients were from the city while three were travellers; hospitalization was more frequent among the young and male gender. Fever, vomiting, aches and abdominal pain were the most common troublesome manifestations; classical dengue triad was present in 55 (25.1%) patients; hemorrhagic, neurologic and mucocutaneous manifestations were present in 44 (20.1%), 8 (3.7%) and 70 (32%) patients. Ascitis, pleural effusion, and Gall bladder wall oedema was found in 53 (24.2%), 31 (14.1%) and 45 (20.5%) patients respectively. Mortality was 1.4% (3 deaths); in addition there was an intra-uterine fetal death; mean expenditure per patient during the illness was US$ 377.25. Conclusion: Dengue virus infection results in immense morbidity and substantial mortality. INTRODUCTION Dengue, a mosquito borne Flavivirus infection, leads to recurrent epidemics in urban agglomerates causing immense morbidity and occasional mortality (1). Though world-wide in distribution, dengue poses a bigger challenge in resource poor countries, like India, especially for the urban and rural poor. Recently, World Health Organization (WHO) has modified guidelines for management of dengue infection (2). Previous classification of Dengue Hemorrhagic fever (DHF) and Dengue shock syndrome (DSS) have been replaced by the new triage of Dengue without warning signs, Dengue with warning signs and Se-vere Dengue to expedite adequate therapy (2). Criteria for presumptive diagnosis, hospital admission, fluid and blood component administration, and discharge have been forwarded. These may significantly impact dengue mortality; however their usefulness is not validated in India (2). Repeated dengue epidemics occur in Delhi, National Capital Region, India (1,(3)(4)(5) where a large proportion of people live under the poverty line (6) or in unhygienic conditions (6). We planned this study to highlight the clinical and laboratory features of acute dengue infection and its financial impact on the affected patients. MATERIALS AND METHODS A prospective observational study was undertaken at St Stephen Hospital, Delhi from August 2013 to January 2014. A total of 918 suspected dengue patients were tested with for NS1 antigen, anti-dengue IgM and IgG antibodies by ELISA. Antibody testing was done with commercially available Panbio Dengue IgM capture ELISA (Inverness Medical Innovations, Cat No. E-DEN01M/EDEN01M05) (MAC-ELISA) and Dengue IgG capture ELISA (Inverness Medical Innovations, Cat No. E-DEN02G) (GAC-ELISA). One Negative control, one Reactive Control and Callibrators in triplicate was used in each ELISA test as per manufacturer instructions. Index value [ratio of sample absorbance and cut-off value (calculated from calibrator ODs and calibration factor given in each kit)] and Panbio units were calculated for each sample and results are classified as positive [Index value >2.2], negative (Index Value<1.8) and equivocal (Index Value~1. 8-2.2) according to the manufacturer's instructions. Any initial equivocal result was retested to confirm the result. NS1 antigen (NS1Ag) testing was done with commercially available Platelia TM Dengue capture 96-well sandwich format ELISA (Bio-Rad, Cat. No.72830). One Negative control, two calibrators, and one Positive Control were used in each run, and the cut-off was calculated as mean of calibrators. All ELISAs were performed on a fully automated EVOLIS platform. Inclusion criteria for observational study. Febrile hospitalized patients with either i) MAC-ELISA and GAC-ELISA reactive, or ii) MAC-ELISA alone reactive, or iii) NS1Ag-ELISA and MAC-ELISA re-active or iv) NS1Ag and GAC-ELISA were included in the study (n=219). Exclusion criteria for observational study. 1) Patient tested with rapid Immunochromatographic test for Dengue antibodies or Antigen, 2) Patient treated on outpatient basis (not admitted), 3) Patient MAC-ELISA and NS1Ag-ELISA negative, GAC-ELISA alone reactive, 4) Febrile patient with blood culture positive bacterial sepsis or significant pyuria and bacteriuria or malaria parasite documented by thick, thin films and antigen detection test. Informed consent was taken from each patient enrolled in the study. Institutional ethics committee approval was taken for conducting this study. Patient's occupation, daily income, loss of income if any, total expenses due to the illness were recorded. Meticulous clinical examination was carried out daily; all patients underwent the following investigations as per the clinical requirement-complete blood count (CBC), blood culture, urine culture, thin and thick smears for malarial parasite, rapid malarial antigen testing (Sure Test, pan-LDH-2 antigen detection, Microgene Diagnostics), Erythrocyte Sedimentation Rate (Westergren's method)), liver function tests (LFT), renal function tests, chest X-ray (CXR) and ultrasonography (USG). CBC was done daily for the first 4 days of hospital stay and then as and when required depending on the clinical situation. Clinically, presence of tachypnea, chest retractions, decreased breath sounds and decreased vocal resonance were considered signs of pleural effusion. Presence of abdominal distension with fullness of the flanks and presence of shifting dullness or fluid thrill was taken as evidence of ascites. The extent of hemoconcentration was quantitated by taking a difference between the maximum hematocrit at admission or anytime during the hospital stay and the minimum hematocrit recording at convalescence or discharge (7). Dengue with warning signs was defined as laboratory confirmed dengue with any of the following: abdominal pain or tenderness, Persistent vomiting, clinical fluid accumulation, mucosal bleed, lethargy, restlessness, Liver enlargement >2 cm and increase in hematocrit (>20%) concurrent with rapid decrease in platelet count (to below 40000/ul) (2). Severe dengue was defined as patients with any of the following features: severe plasma leakage with shock and/or fluid accumulation with respiratory distress, severe bleeding, or severe liver, renal, cardiac, and pulmo-IRAN. J. MICROBIOL. Volume 9 Number 3 (June 2017) 174-185 http://ijm.tums.ac.ir nary or central nervous system impairment (2). DHF was diagnosed as per the older WHO guidelines as a probable case of dengue fever with hemorrhagic tendencies and thrombocytopenia along with the presence of evidence of plasma leakage manifested by any one or more of the following i.e., a rise in the average hematocrit for the age and sex by >20%; a >20% drop in the hematocrit following volume replacement compared to the baseline; signs of plasma leakage i.e., pleural effusion, ascites, hypoproteinemia (7). The area specific hematocrit cut off values for hemoconcentration was defined as >36.3% in less than 12 years age group (4) and >37.5% in those more than 12 years (8). Criteria for splenomegaly on USG was span more than 11 cm on greatest dimension or weight greater than 250gm (9) and that for hepatomegaly was span at mid-clavicular line greater than 15.5 cm (10). Guidelines of the National Vector Borne Disease Control Program (NVBDCP) in India were used as criteria for assessing appropriateness of platelet transfusion (11). Each patient's total expenditure was determined from the final bill generated for each patient. Statistical analysis was carried out on the data in Microsoft Excel software. DISCUSSION Outbreaks of dengue fever with consequent hospitalization of severe cases are becoming frequent since 1980s (12) especially in the south East Asian region. With no specific treatment available, prevention remains a cornerstone. Preventive measures are not effectively applied in developing countries due to lack of funding, trained manpower, community ignorance and non-participation (13). The morbidity and financial cost of this infection is tremendous on the afflicted subjects (14). In addition fatalities due to dengue fever are not uncommon as seen in our study (3 deaths, one IUFD). Majority of those hospitalised and two of those who died were young adults (20-40 years). The male predominance among those hospitalised may be due to rampant gender discrimination in India (15). The seasonal distribution of dengue cases is well documented, with a peak incidence after the heaviest monsoon rainfall (12). The exact month of highest incidence may differ according to the timing of the rainfall. Mudpools, open water reservoirs and stagnant water in houses and roofs are breeding sites for Aedes (Stegomyia) aegypti mosquitoes (2,12). Most affected patients were residents of NCR, mainly school and college going students and homemakers; however three travellers were also among those hospitalised. This finding highlights the serious situation of dengue in the capital; complicated by the lack of vaccines and chemoprophylaxis against the disease. Travellers and residents alike must strictly guard against mosquito bites through use of mosquito nets and repellents. Public health education regarding protection against mosquito bites is important but the growing resistance of mosquitoes to insecticides is a concerning factor (16). Fever with chill and rigor, and aches were the commonest symptom. These apart vomiting and lack of appetite were very commonly noted. Many patients had troublesome persistent vomiting as their main complaint; abdominal pain was another significant trouble for some patients. Vomiting and abdominal pain have been previously documented to be particularly troublesome symptoms (17). Dengue triad was documented only in a quarter of the patients (25.1%). The tourniquet sign (11 out of 155 patients tested) was even more infrequently documented. The absence of the classical dengue triad or the tourniquet sign in many patients discourages the use of these for diagnosis or triage. In another two studies, one from Eastern India (17) and another from the north (4), positive tourniquet sign were found in 31% and 25.5% of dengue patients. Though hemorrhagic manifestations were noted in 44 patients, no individual site or lesion was responsible for majority of the cases; so it is important to be aware and on the look out of the myriad hemorrhagic presentations. Other studies have found similar wide distribution of hemorrhage patterns in dengue patients (4,(17)(18)(19). Of cutaneous manifestations, maculopapular or petiechial secondary skin rash was the most common manifestation, however initial flushing on the face or limbs or trunk was seen in minority (9.2%) of patients. Proportion of peteichial rash is similar to that seen patients from Lahore; howev-er that with initial flushing is very low compared to same study (20). The macular or maculopapular rash occurred commonly between day-4 and day-6 of fever (74%), started distally (36, 72%) and progressed centrally or remained localized. Itching on the rash was reported only in minority (13.6%) unlike Afzar NA et al. who reported pruritus on 62% of rashes (20). In DSS patients' onset of shock is usually around day 3 of fever (2,12); in our study shock occurred most commonly between day 2 and day 4 of fever (61.1%), mean duration between onset of fever and shock being 3. (21). Other markers of plasma leakage (2,12) were also detected better by radiological methods including ascitis and pleural effusion in 24.2% and 14.1% patients respectively, as noted in other studies (17,21). Further we noted that right sided pleural effusion was earlier to appear and pleural effusion was often unilateral. Many such minimal effusions were detected more frequently by USG than the conventional CXR. Thickened and oedematous gall bladder wall changes are characteristic of the plasma leakage syndrome in dengue (22) and may be marker of severe dengue (22). Significant thrombocytopenia, marked leucopenia (50.6%) and lymphocytosis (32.8%) with presence of many reactive lymphocytes in the peripheral smear (93%) are commonly noted features in dengue fever. In our laboratory we often utilize these parameters to suspect a dengue case during outbreak periods. Together with significant transamnitis (83.4% patients) and hemoconcentration (19.8%), these findings are a valuable clue to the laboratory regarding the diagnosis. Interestingly, in addition a rise in eosinophil count is often documented later in the course of dengue as seen in 32.8% of our patients. The cut-off levels of Hb% as a marker for hemoconcentration devised previously (4,8) seems to need revision as were many as 96.9% dengue patients were above cut-off at admission. In contrast, however, hemoconcentration in excess of 20% and 15% was documented in only 19.8% and 32.4% patients only. The cut-off levels need to indicate only the at-risk patients (those who will go on to develop severe dengue or DHF) (8) and not all cases of dengue. Leukopenia occurred in more than half of dengue patients; in contrast Shukla V et al. found leucopenia in only 10% of cases (21). As noted in our study, transamnitis with ALT and AST greater than double the upper normal range is documented in majority of patients (16,21,23). The AST: ALT ratio 1.42 which is similar to that of 1.8 and 2.0 seen in other studies (21,23). More severe liver dysfunction as indicated by hyperbilirubinemia, deranged coagulation profile, raised alkaline phosphatise and significant hypoproteinemia (and hypoalbuminemia) are found in only a minority of cases and indicates severe disease (24). Reversible renal dysfunction was also found in 24 (11.1%) patients. A leukoerythroblastic blood film and smear cells are found in even fewer patients of dengue hemorrhagic fever. Leukoerythroblastic blood film probably represents a bad prognostic factor in dengue patients as 3 out of the 5 patients died of the disease. A combination of dengue antigen and antibody detection by ELISA tests was used in our study to confirm the diagnosis. However, MAC-ELISA and GAC-ELI-SA can show serological cross-reactivity among flavivirus infections like Murray Valley fever, Japanese Encephalitis, St. Louis Encephalitis, Yellow fever and West Nile (2,25). Rheumatoid factor, malaria, Chikungunya, Lyme disease, Scrub Typhus, Hanta virus infection and leptospirosis are additional scenarios producing false positive MAC-ELISAs (2,25,26). A persisting IgM against Dengue virus originating from a previous infection will result in false diagnosis (27,28). IgM anti-Dengue usually persists for 2-6 months (28, 29) with a median time period of 179 days for primary and 139 days for secondary infections (29). Anti-Dengue IgG antibodies persists for an even longer time even up to years (2,30). False positive anti-Dengue virus IgG detection has been documented in bacteremia, leptospirosis, Q fever, and other viral infections like Chikungunya, Tick-borne encephalitis, Varicella, Cytomegalovirus, and Epstein-Barr infections (28,31). False negative GAC-ELISA can occur in primary dengue infections where these antibodies are slow to appear (32). Testing GAC-ELISA on paired samples is thus more worthwhile (2,27). NS1Ag is often only transiently present in secondary dengue infections and may be masked by antigen-antibody complexes. False positive NS1Ag tests have been detected in acute Zika virus infection, Cytomegalovirus infection and in patients with haematological malignancies (33)(34)(35). In light of such discrepancies, clinical diagnosis in our hospital depends on a combination of antigen and antibody detection rather than a single assay. Close observation and symptomatic management is all that is required in most cases of simple dengue (2,12). Anti-pyretic were nearly universally prescribed for these patients while anti-emetics were required in considerable number of patients. Careful intravenous fluid replacement remains the cornerstone for managing severe cases of dengue (2,12,17). 167 of our patients received intravenous crystalloids as therapy; fluid overload was monitored for in these cases. Fluid overload can lead to serious consequences (17). Two of our patients required intravenous colloid solutions while four patients were treated with fresh frozen plasma for refractory shock. Most authors report no relation between platelet counts and clinical bleeds, however the duration of shock has important implications regarding severe bleeding (11). Thus appropriate monitoring of shock and sequential hematocrit monitoring may have been more appropriate (2,11,12). Active bleeding may be a more appropriate indication platelet transfusion rather than low platelet count. Only 29% of dengue related platelet transfusions in our study population had active bleeding while 40.6% patients had no appropriate indications as per NBVD-CP guidelines (11). Other hospitals in Delhi also report high usage of platelet transfusions, often inappropriately (3,5). WHO as well as the Indian NBVDCP authorities recommend packed cell transfusion in only those cases who show 10% or more loss of blood volume and in those with refractory shock with declining hematocrit (11,12). In addition to platelets, four patients required such packed red cell transfusion while three patients were given fresh frozen plasma in view of coagulopathy. Mortality in our study was 1.2%; additionally there was a case of intra-uterine fetal death. All three dead patients were male, bread-winners of their family and belonged to 20-55 years working age group. One patient died due to refractory shock along with myocarditis and severe hepatic involvement; the second due to refractory shock with severe blood loss and hepatic involvement superadded on chronic renal failure and diabetes mellitus; and the third due to myocarditis, cellulitis and staphyloccal sepsis. Severe hepatic involvement with transaminases greater than 1000 IU/dl and myocarditis have previously also been associated with poor prognosis. The morbidity due to dengue is no less, mean duration of hospital stay was more than 6 days in our study, mean expenditure per patient due to the illness was greater than US$377; and further nosocomial infections complicated the course in as many as 5 patients. Shepard DS et al. in their pan-India study conducted from 2006-2012 who noted an average expenditure of $235.20 in a hospitalised patient in private set-up (14). The higher expenditure in our study may reflect inflationary changes and the fact that Delhi is one of the costliest cities in India. A major limitation in our study is the dependence of dengue diagnosis on serological and antigen detection assays only. Due to resource limitations, viral RNA detection and virus culture was not possible. Further, serotype / genotype identification could not be done during the study.	2018-04-03T04:52:36.170Z	2017-06-01T00:00:00.000	{ "year": 2017, "sha1": "ae7d4189fe2b668139688b474ccf0958f0673082", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "dbfa10760922779f9eeadf5fd8eb7bb6bf4c45fe", "s2fieldsofstudy": [ "Environmental Science", "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
236283871	pes2o/s2orc	v3-fos-license	Effects of Topography on Planted Trees in a Headwater Catchment on the Chinese Loess Plateau : The Chinese Loess Plateau (CLP) is known for its complex topography of hills and gullies, and lots of human land-use management activities have been put into practice to sustain the soil, water and other natural resources. Afforestation has been widely applied on the CLP and it’s important to understand the effects of topography on these planted trees. However, the coarse spatial resolution of remote sensing data makes it insensitive to local topography, and the traditional in-situ measurements would consume vast amounts of time and resources. In this study, a small headwater catchment of the CLP was selected to study the effects of topography on the planted trees. Low altitude unmanned aerial vehicle based light detection and ranging (UAV-based LiDAR) technology was utilized to obtain high-resolution topography and vegetation structure data. Results showed that the middle transition zone (mid-transition, slope > 45 ◦ ) was an important boundary of topography in the gully area of the CLP. In the forested catchment, the area of the mid-transition zone had the lowest of tree density, canopy coverage and leaf area index due to steep slope gradient. The tall trees ten to twenty meters high were concentrated in the downhill area, which had the highest canopy coverage and leaf area index. Elevation had signiﬁcant linear relationships with canopy coverage and leaf area index ( p < 0.001), which revealed the impact of topography on the forest indexes of the afforestation catchment. We concluded that the high-resolution LiDAR technology facilitated the research of topography and forest interactions in land surface. Introduction The topography has a significant influence on the growth and distribution of trees and they co-evolve under the joint influences of climate and the local ecosystems [1]. The differentiations of energy and moisture controlled by topography are the core factors driving the evolution of trees distribution under natural conditions [2,3]. Previous studies that characterized the interactions between topography and trees have been mostly conducted at basin or plot scale [4][5][6]. Headwater catchment is the basic geomorphic unit in studying the interactions of Earth surface processes, climate change and human impacts [2]. Thus, precise characterization of the relationship between topography and trees in headwater catchments can provide an in-depth understanding of microtopography impacts on trees growth and distribution. Afforestation is widely used to restore deteriorated ecosystems and combat land desertification in many places of the world. However, if looking back upon the arduous development of afforestation in the past twenty years, it has been and continues to be a controversial topic that keeps sparking academic and social debates [7][8][9][10][11][12]. The afforestation of drylands is criticized for its negative impact on water resources. On the other hand, many studies have shown that afforestation is more efficient in flood control and soil erosion reduction than the natural regrowth of grassland in some arid and semiarid regions, especially during the early stages of ecosystem rehabilitation [13][14][15]. Moreover, properly managed forest ecosystems play an important role in the progress towards meeting carbon neutrality targets, not just in China, but all over the world [16][17][18][19]. The satellite data (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017) reveals that the afforestation and cultivation of China and India lead the greening of the world, and the afforestation in China alone accounts for 10.5% of the global net increase in leaf area [20]. It is known that the afforestation work should be managed according to the topographic diversity, local climate, soil conditions and available resources. In reality, the topographic diversity plays an important role in the evolution of habitat diversity with its power in the determination of local thermal, humidity, and soil fertility conditions, and that should be taken into account when selecting reasonable species for afforestation, considering its potential strength to increase the ecological stability as well as to support the forestry productivity [21]. Also, the topographic concavity creates climatic microrefugia for local species, and the differentiation of the microtopographic thermal and moisture conditions can impact the species redistribution [3,22,23]. In the past, forest inventories were generally used to obtain quantitative information about trees. However, such manual methods are not efficient due to their need for a large labor force. Moreover, it is also difficult to obtain high-resolution topography data through traditional methods, and thus, it is a challenge to quantitively and precisely characterize the relationship between topography and trees. In recent years, high technologies, such as the small low-altitude unmanned aerial vehicle (UAV), digital aerial photogrammetry (DAP), light detection and ranging (LiDAR) and aerial laser scanning (ALS) techniques have been widely applied to quantify the highresolution data of topography and trees due to their advantages of high flexibility and high efficiency [24][25][26]. It's suggested that affordable and portable UAV-borne LiDAR system can generate very high resolution three-dimensional (3D) habitat information, including terrain and vegetation structure [27]. For example, Grijseels et al. [28] utilized LiDAR imaging and found that the riparian vegetation structure follows the patterns of topography related to energy and water subsidies; Cucchiaro et al. [29] combined the 3D photogrammetry and terrestrial LiDAR scanning (TLS) techniques and concluded that this data fusion can be utilized to create highly accurate digital terrain models (DTMs) in the context of complex and unreachable topography and vegetation cover. The high-resolution UAV and LiDAR techniques facilitate the observation practice of geomorphic change characterization [30]. Meinen and Robinson utilized UAV collected time-series data in the evaluation of soil erosion models in field experiments [31]. Gong et al. [32] applied the UAVs and structure from motion (SfM) technology to obtain high-resolution three-dimensional models of the erosion gully in an open-pit coal mine, and the freeze-thaw cycle-caused soil erosion was clearly captured. Thus, the integrated method of LiDAR and 3D photogrammetry has provided new possibilities to comprehensively quantify the relationship between the topography and vegetation distribution. The Chinese Loess Plateau (CLP) is a semi-arid region characterized by extensive loess distribution, low vegetation coverage, and severe soil erosion [33][34][35]. Since 1991, the Chinese government has launched the 'Conversion of Cropland to Forest Program' (CCFP), also known as the 'Grain for Green' project, and this project pays farmers to restore their farmland and the degraded land to forest, shrub and grassland [36,37]. After nearly 20 years of rehabilitation efforts, 16,000 km 2 of rain-fed cropland has been converted back into planted vegetation, which has caused a 25%-28% increase in vegetation coverage on the CLP [10,38,39]. Some studies have demonstrated that this large-scale afforestation and crop plantation on the CLP has induced a decline in both blue and green water resources, which would result in the degradation of forest ecosystems and the mosaic distribution of vegetation due to the microtopography impacts [11,[40][41][42]. However, the precise relationship between topography and trees in headwater catchments on the CLP is underestimated due to the limitations of current technology. The lack of such information will hinder our understanding of the evolution of planted trees and their relationship with the landform in the catchment. We hypothesize that topography should have a significant impact on the spatial distribution of planted trees at catchment scale. To test this hypothesis, a planted headwater catchment was selected to study the influence of topography on the evolution of trees. The catchment was planted in 1954 [5] and almost 90% was replanted with the single tree species Robinia pseudoacacia Linn. A small number of other tree species were planted along the edge of the catchment. Thus, the catchment can be treated as a monoculture catchment with singular tree species of R. pseudoacacia. Moreover, the catchment has more than 60 years of afforestation history, which can provide a predictive information for the evolution and adaption of trees at catchment scale on the CLP. In this study, the high-resolution data of forest indexes and topography in the catchment were obtained using the techniques of DAP and UAV-borne LiDAR and their relationships were deeply analyzed. The objectives of this study were to: (1) quantify the high-resolution spatial information of the landform and tree distributions, and (2) characterize the relationships between the topography and forest indexes at catchment scale. Study Area This study was conducted in a headwater catchment (Yangjiagou) on the southern CLP (107 • 33 19" E, 35 • 41 52" N). The small headwater catchment has an area of 0.87 km 2 , and it locates in the Nanxiaohegou basin of Xifeng district, Qingyang city, Gansu province, China ( Figure 1). It is a typical loess tableland and gully area, and the altitude ranges from 1180 m to 1365 m based on the ASTER GDEM V2 data. The mean annual temperature is 9.3 • C and the mean annual precipitation is 556.5 mm, and more than two thirds of annual precipitation occurred from June to September [43]. The loessal soil in the catchment derives from the Quaternary dust and it's about 200-250 m thick with the texture mostly of silt-loamy [5,[44][45][46]. The studied (Yangjiagou) catchment is generally composed of three basic landform units, including the level ground, upper hillslope, and down slope gully. The level ground is a flat area with a slope lower than 5 degrees, and it's used for cultivation which include wheat and corn plantation. The upper hillslope with a gentle slope gradient (10 • to 20 • ) connects the tableland and the down slope gully. The down slope gully with steep gradient is covered with high density of trees and understory vegetation [43]. Before 1954, the soil erosion processes in the catchment were dominated by hillslope water erosion and gully gravity erosion due to farming activities. In 1954-1958, the afforestation project was deployed in the catchment and surface and gravity erosion significantly decreased. After 65 years of vegetation restoration, soil erosion is completely controlled. Although gully gravity erosion occasionally occurred, the sediments cannot deliver to the outlet of the catchment due to the strong control effects of gully trees and understory. In 1954-1958, almost the total area of the catchment was planted with the single tree species of R. pseudoacacia. In the later time, a small number of other tree species, e.g., Armeniaca sibirica (L.) Lam, Pinus tabuliformis Carr., Salix matsudana Koidz., and Platycladus orientalis (L.) Franco, are planted in the edge of the catchment and their distribution area is less than 10%. Since 1954, cultivating, grazing, and lumbering were forbidden in the area until present day [5,43]. After 65 years of evolution, the catchment now forms a stable forest ecosystem [5,44,47]. In April of 2019, we conducted a carefully field investigation in the catchment and found that the planted trees of R. pseudoacacia account for almost 90% area of the catchment, and most of them are big trees with the height of 10 to 20 m and prosperous canopy. The tree species of A. sibirica, P. tabuliformis, S. matsudana, and P. orientalis are scattered along the edge of the catchment. Thus, the studied catchment can be treated as a monoculture catchment. During the past 60 years, the planted catchment received no large-scale forestry work, such as thinning. Currently, the trees (including understory vegetation) and hydrology in the gully have evolved into a natural equilibrium [5,43,48], (including understory vegetation) and hydrology in the gully have evolved into a natural equilibrium [5,43,48], thus making it an ideal case to study the effects of topography on the spatial distribution of trees at catchment scale on the CLP. accessories. The Phantom 4 Pro was used to take series of flight attitude informationembedded high-resolution photos, with 40% coverage of the adjacent pictures in both the vertical and horizontal directions. Those pictures were spatially aligned, orthorectified, mosaic processed, and transformed into digital surface models (DSM) and orthoimages by using the Agisoft PhotoScan 1.4 software (now renamed Agisoft Metashape, https: //www.agisoft.com/, accessed on 1 January 2018). The precise boundary of the study area was obtained based on the high-resolution orthoimages and DSM and careful field investigation, which facilitated the consequent ALS flight plan. Field Investigation and Aerial Scanning The ALS flight was deployed in July 2019, when the rainy season was about to begin and the vegetation started to prosper. The UAV-based LiDAR system is an integrated lightweight aerial Light Detection and Ranging (LiDAR) synthesis, consisted of a DJI M600 Pro drone with the real-time kinematic (RTK) mobile station and a LiAir 220 LiDAR system from Beijing Green Valley Technology. Co., Ltd. (https://www.lidar360.com/, accessed on 1 January 2018). This platform could work 30 min in the air and the longest range was about 200 m with a resolution of ±2 cm at the 20% reflectivity. Since the high-density laser beam of LiDAR could penetrate the plant canopy and both canopy and understory landform information can be obtained. The LiDAR360 software, which is a professional LiDAR point cloud processing and analyzing software developed by Beijing Green Valley Technology. Co., Ltd., were used to process the high-resolution data of both forest and landform. After a preliminary work of point cloud outlier removement, ground points classification, noise filter, ground points normalization, boundary extraction and subsampling, the resulting point cloud was utilized in the subsequent calculation of topographic and forest indexes. Topographical and Forest Indexes 2.3.1. Topographical Indexes A digital elevation model (DEM) of the Yangjiagou forest catchment was generated from the preprocessed point cloud by using the terrain toolbox in LiDAR360, with a pixel size of 0.5 × 0.5 m. The basic topographical indexes, including elevation, slope, and aspect, were calculated using spatial analyst tools in ArcGIS 10.6, and the other two indexes, topographic wetness index (TWI) and direct insolation (DI) were computed in SAGA 7.5.0 (http://www.saga-gis.org/, http://saga-gis.sourceforge.net/en/, accessed on 1 January 2018) [49]. TWI is one of the most widely used topographic indexes, which is related to the spatial distribution and size of saturation zones or variable source areas for runoff generation [50][51][52]. TWI is calculated from specific catchment area (α) of upper slope and local slope (tanβ) [53]. In this study, the standard computing method of TWI was applied, which calculated based on the equation of ln(α/tanβ) [50,52]. Direct insolation (DI) is the potential incoming solar radiation of a certain area in a specified time range, and the index of DI in this study was calculated using the Terrain Analysis tools in SAGA 7.5.0. The scenario period was set from February 1st, 2019 to December 31st, 2019 with the resolution of 7 days, and the whole day was set as the time span with a resolution of 0.5 h. Forest Indexes Canopy height model (CHM) is defined as the average height of the top of the vegetated canopy. Canopy cover (CC) is defined as the vertical projection of the tree canopy onto a hypothetical horizontal surface ground. Leaf area index (LAI) is the ratio of total upper leaf surface area to ground area (for broadleaf trees), or total projected conifer needle surface area to ground area (for coniferous plants). LAI directly represents the canopy structure and impacts the energy exchange, evapotranspiration and carbon dioxide be-tween plants and atmosphere, which can be used to predict primary productivity and crop growth [54]. Tree density (TD) is the number of trees in a given area, which is convenient and intuitive to display the plantation status. Since the default and smallest output pixel size of LAI was 15 m × 15 m, the TD, CHM, and CC were resampled and adjusted to the same pixel size as above. In this study, the minimum tree height was set as 2 m and crown base height threshold was set 0.8 m. The indexes of CHM, CC and LAI were calculated using the ALS Forest toolbox in LiDAR360 [55] and TD was calculated based on the spatial coordinates of trees from the forest segmentation data. Statistical Analysis The topographical and forest indexes were spatially aligned and converted from raster into matrix data, so that the significance test (Wilcoxon test), correlation analysis (Pearson method), and linear regression can be performed to quantify the relationships between the landform and vegetation. All the statistical analysis was conducted in the RStudio software (version1.2.5019, http://www.rstudio.com/, accessed on 1 July 2018) with R (version 3.6.2, https://www.R-project.org/, accessed on 1 July 2018) [56,57]. The topography and forest indexes were point-to-point aligned and transformed into unified raster data with the pixel size of 15 m × 15 m, then converted into points (observations). The elevation of the forest catchment was replaced with the relative elevation, which was the result of elevation minus the lowest altitude in Yangjiagou forest catchment. The observations were then utilized in the subsequent significance test, correlation analysis, and linear regression. The Wilcoxon test (α = 0.05) [58,59] were used and the comprehensively integrated observations were measured to verify the differences of topographical and forest data among the paired slope classes. The null hypothesis was that the paired slope classes have the same data distribution of topographical or forest index, and the alternative hypothesis was that the paired slope classes are different in the topographical or forest data distribution. Topographical Features of the Headwater Catchment Compared with the images of QuickBird, ASTER GDEM and DAP, the LiDAR imagery showed the highest accuracy of vegetation and landform information (Figure 3). As a result, the spatial distribution of elevation, aspect, slope, TWI and direct insolation was calculated based on the ALS point cloud data (Figure 4). The elevation of the Yangjiagou forest catchment ranges from 1126.3 to 1317.6 m, which can be categorized into six classes: (Figure 4c) and the simulated TWI also showed slope-dependent patterns (Figure 4d). Figure 4c revealed an obvious boundary between the upper part and the lower part, which were significant different in slope and elevation. These three sections of particular topographic positions could be regarded as three different landform units: the downhill slope, the middle transition zone (slope > 45 • ), and the uphill slope, hence it was named as the DTU landform system (Figure 4f). Forest Indexes of the Headwater Catchment Based on the field investigation results, we assume that the other tree species located in the catchment boundary cannot have obvious impacts on the distribution of forest indexes at catchment scale. Moreover, the planted catchment did not receive any forestry work, such as thinning. Thus, the characterized relationships between topography and forest indexes in the catchment is reliable. Figure 5 shows the distribution of forest indexes of the studied catchment. The average density of trees in the forest catchment was 8.6 stems per 225 m 2 . The downstream Forest Indexes of the Headwater Catchment Based on the field investigation results, we assume that the other tree species located in the catchment boundary cannot have obvious impacts on the distribution of forest indexes at catchment scale. Moreover, the planted catchment did not receive any forestry work, such as thinning. Thus, the characterized relationships between topography and forest indexes in the catchment is reliable. Figure 5 shows the distribution of forest indexes of the studied catchment. The average density of trees in the forest catchment was 8.6 stems per 225 m 2 . The downstream revealed a relatively higher density of trees than that in the upper stream. Moreover, the upper hillslope showed relatively higher density of trees than that in the down slope, especially in the downstream of the catchment. The canopy height of trees ranged from 3 to 10 m in the upper hillslope and the middle transition zone areas. Most of the tall trees concentrated in the down slope of the catchment, which shows the height of 10-20 m. The diameter at breast height of those trees ranged from 0.3 to 0.6 m. The CC and LAI showed significant higher values in the down slope than that in other areas. The average CC of the catchment was about 60-70%, and the down slope showed obviously higher CC than the other landforms. The values of LAI in the down slope were generally higher than 1 and the other areas were mostly lower than 1. Forests 2021, 12, x FOR PEER REVIEW 10 of 18 revealed a relatively higher density of trees than that in the upper stream. Moreover, the upper hillslope showed relatively higher density of trees than that in the down slope, especially in the downstream of the catchment. The canopy height of trees ranged from 3 to 10 m in the upper hillslope and the middle transition zone areas. Most of the tall trees concentrated in the down slope of the catchment, which shows the height of 10-20 m. The diameter at breast height of those trees ranged from 0.3 to 0.6 m. The CC and LAI showed significant higher values in the down slope than that in other areas. The average CC of the catchment was about 60-70%, and the down slope showed obviously higher CC than the other landforms. The values of LAI in the down slope were generally higher than 1 and the other areas were mostly lower than 1. Slope-Dependent Differences of Topography and Forest Indexes In this study, the topography and forest indexes in the Yangjiagou catchment were converted into 2540 vector points and the Wilcoxon test (α = 0.05) was used to verify the differences of topographical and forest indexes among the three different landform units, uphill slope, middle transition zone, and the downhill slope (Figures 6 and 7.). The distributions of elevation, aspect, slope and direct insolation were significantly different among the three different landform units (p < 0.0001, Figure 6). Moreover, the TWI was also different between the uphill and downhill slope (p < 0.0001, Figure 6). In the Yangjiagou forest catchment, the average elevation differences of the downhill, mid-transition and uphill slope were 53.0, 108.8 and 133.8 m, respectively. The aspect of the uphill slope was significantly different from that of the mid-transition and downhill slope, which showed a wide area of insolation in the upper hill. The average slopes of the downhill, mid-transition, and uphill, were 31.7°, 49.3° and 28.3°, respectively. Slope-Dependent Differences of Topography and Forest Indexes In this study, the topography and forest indexes in the Yangjiagou catchment were converted into 2540 vector points and the Wilcoxon test (α = 0.05) was used to verify the differences of topographical and forest indexes among the three different landform units, uphill slope, middle transition zone, and the downhill slope (Figures 6 and 7). The distributions of elevation, aspect, slope and direct insolation were significantly different among the three different landform units (p < 0.0001, Figure 6). Moreover, the TWI was also different between the uphill and downhill slope (p < 0.0001, Figure 6). In the Yangjiagou forest catchment, the average elevation differences of the downhill, mid-transition and uphill slope were 53.0, 108.8 and 133.8 m, respectively. The aspect of the uphill slope was significantly different from that of the mid-transition and downhill slope, which showed a wide area of insolation in the upper hill. The average slopes of the downhill, mid-transition, and uphill, were 31.7 • , 49.3 • and 28.3 • , respectively. The distributions of TD, CHM, CC and LAI were significantly different among the three different landform units (p < 0.0001). TD in the middle transition zone was significantly different from the uphill and downhill slopes (p < 0.001, Figure 7). However, no significant differences were found for TD between the upper and down slope areas. The average TD in the middle transition zone was 8.1 stems per 225 m 2 , which was lower than 9.1 in the downhill slope and 8.8 in the uphill slope. The CHM, CC and LAI showed a decreasing trend from the gully bottom to the upper hills ( Figure 7). The average canopy heights in the three different landform units (downhill, mid-transition, uphill) were 8.5, 3.3 and 2.4 m, respectively, and the highest tree of 25.3 m occurred in the down slope area. The average CC in the three different landform units were 79.5%, 50.7%, and 45.4% and the average LAI were 0.99, 0.42, and 0.35, respectively, indicating that the downhill slope area has good moisture conditions for tree growth. Figure 8 shows the results of Pearson correlation analysis between the topography (elevation, aspect, slope, TWI, DI) and forest indexes (TD, CHM, CC, LAI). No significant relationships between the topography and forest indexes were found, except elevation. Elevation was negatively correlated with TD, CHM, CC and LAI. The linear regression analysis between elevation, TD, CHM, CC and LAI showed that CC and LAI had significant linear relationships with elevation (p < 0.001), while no significant linear relationships between elevation, TD and CHM were found (Figure 9). . The size of ellipse in plot cells is proportional to the Pearson correlation coefficients (r). The scale bar extends from perfect negative correlation (dark bule, r = −1) to perfect positive correlation (dark red, r = 1). Black rectangles categorize the indexes into groups based on the Hierarchical clustering method. In the plot, SLO, ELE and ASP indicate slope, elevation and aspect, respectively. The hierarchical clustering results indicated that the slope, elevation, and aspect were categorized into one group, and the TWI and direct insolation were classified together, and the rest of the forest indexes were categorized into the same group. A faint correlation relationship (r = 0.22) was found between TWI and DI, but no obvious correlations were found among the slope, elevation, and aspect indexes. Slope was negatively correlated with TWI (r = −0.52) and DI (r = −0.53), and the elevation and aspect showed faintly correlations with the DI index. There were relatively high correlation relationships among the forest indexes. TD was positively correlated with CC (r = 0.53), and CHM was strongly positively correlated with CC (r = 0.63) and LAI (r = 0.7). A significant positive correlation existed between CC and LAI, which the coefficient was 0.87. There was hardly no obvious correlation relationship between the topography and the forest indexes, except for elevation. Elevation was negatively correlated with TD (r = −0.23), CHM (r = −0.43), and LAI (r = −0.56), and it was strongly negatively correlated with the CC index (r = −0.6). Therefore, we performed the linear regression analysis between the forest indexes (TD, CHM, CC, LAI) and the relative elevation to quantify the relationships and effects. Correlation and Linear Regression Analysis of Topography and Forest Indexes Linear regression analysis showed that the forest indexes of TD and CHM had weak negative relationships with the relative elevation (Figure 9a,b, R 2 = 0.054 and 0.182, respectively). However, the forest indexes of CC and LAI had more significant linear and negative relationships with the relative elevation (Figure 9c,d, R 2 = 0.359 and 0.318, respectively). Effects of Topography on the Distribution of Trees The results of this study demonstrated that the microrefugia derived from the microtopography has fulfilled its expected influences on the distribution of local species. Thus, we can conclude that topography has significant effects on the distribution of trees, i.e., the distribution of TD, CHM, CC, and LAI. In this study, the middle transition zone (slope > 45 • ) was recognized as a remarkable boundary between the uphill and downhill slope area in the forested catchment ( Figure 4). The forest indexes showed significant differences among the three different landform units. The CHM, CC and LAI showed a decreasing trend from the gully bottom to the upper hillslope ( Figure 7). However, the TD showed a V-shape changing curve from the down slope area to the upper hillslope area, with the smallest number of trees distributed in the middle transition zone. The low density of vegetation in the middle transition zone (slope > 45 • ) is assumed to be a typical phenomenon in the complex landform on the CLP due to the steep gradient constraining [60]. In this study, elevation showed the most significant impact on the distribution of trees. As suggested by Lin et al. [61], surface soil moisture would concentrate in the bottom of the valley and deplete in the upper hillslope. In the field investigation of this study, shallow narrow streams were observed in the lowest bottom of the catchment, and the nearby understory bushes and grasses were intensely growing. Most of the tall trees concentrated in the gully bottom and the canopy cover in the down slope area was higher than 80%. Due to the decreases of soil moisture and the increases of elevation, CHM, CC and LAI all declined from the gully bottom to the upper hillslope. As a result, elevation was recognized as the critical driving force of trees distribution in the catchment of the CLP. In this study, the main planted species is black locust (R. pseudoacacia). Overall, this is an effective tree species to restore vegetation on the CLP. However, black locust plantations are suffering degradation, especially in the landform of uphill slopes due to water limitation. Thus, more diversified and appropriate vegetation species are needed on the CLP, not only for the eco-environment restoration but also for the preservation of biodiversity under climate change conditions. In this study, we found that the forest indexes of CHM, CC and LAI in the uphill slopes were lower than that in the gully area. Thus, more drought-tolerant shrubs should be selected for the upper hillslopes or the strategy of natural restoration if the black locust plantations have died should be implemented. The Application of ALS in Geomorphology and Vegetation Structure Observation and quantification of land surface is progressing in revolutionary fashion due to the increased spatial resolution and scope provided by LiDAR technology, which has been widely applied in geomorphic change research, ecosystem monitoring, engineering investigation, environmental planning and forest inventory [24][25][26][27]62]. With the advances of drones, LiDAR sensors and point cloud software, the low altitude aircraft-based ALS is proved to be practical in complex landform research. Moreover, the DAP technique could be utilized as the auxiliary options in the high-resolution and high-accuracy survey. In this study, we found that LiDAR technology facilitated the research of topography and land cover interactions by providing precise, high-resolution 3D information of the Earth's surface features. The Loess Plateau is characterized by gullies and hills and known for its complexity of landforms [12]. It is estimated that the gully area occupies 40-45% of the whole plateau area [63,64]. In this study, ALS devices were deployed in the Yangjiagou forest catchment and obtained high resolution data of topographical features and forest structures (0.5 m × 0.5 m). The economic DAP technique could achieve the same level resolution as the ALS method, while the DAP optical sensors (or cameras) could not observe the understory land surface. The high-density laser beams from LiDAR could penetrate the canopy layers so that it allows for the simultaneous measurements of above-ground vegetation structure and anthropogenic facilities, as well as landform of the earth's surface [25]. Moreover, the ALS set is deployed with a real-time kinematic (RTK) mobile station, which advances the LiDAR survey to a higher accuracy level of centimeter positioning. Conclusions This research was carried out in a forested catchment with 60 years of afforestation history. It revealed that the high-resolution data from the economic UAV-based LiDAR technology facilitated the quantitative study of topography and forest relationships. The middle transition zone in the catchment (slope > 45 • ) was an important demarcation of landforms in the gully landform of the CLP. Slope gradients and forest indexes showed significant differences among the landform units of uphill slope, middle transition zone and downhill slope. Elevation was inferred to be the critical driving force of the tree distribution in the catchment. In the future, models should be developed to simulate the evolution of woody plant cover/tree stands in the complex landforms on the CLP.	2021-07-26T00:05:24.009Z	2021-06-16T00:00:00.000	{ "year": 2021, "sha1": "44010fa998588209e5f656fc71bff666187022a0", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1999-4907/12/6/792/pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "b0e0721264b590cbb0c9e14668857ca9c34811ab", "s2fieldsofstudy": [ "Environmental Science", "Geography" ], "extfieldsofstudy": [ "Environmental Science" ] }
232293725	pes2o/s2orc	v3-fos-license	Verbal threat learning does not spare loved ones Significant others provide individuals with a sense of safety and security. However, the mechanisms that underlie attachment-induced safety are hardly understood. Recent research has shown beneficial effects when viewing pictures of the romantic partner, leading to reduced pain experience and defensive responding. Building upon this, we examined the inhibitory capacity of loved face pictures on fear learning in an instructed threat paradigm. Pictures of loved familiar or unknown individuals served as signals for either threat of electric shocks or safety, while a broad set of psychophysiological measures was recorded. We assumed that a long-term learning history of beneficial relations interferes with social threat learning. Nevertheless, results yielded a typical pattern of physiological defense activation towards threat cues, regardless of whether threat was signaled by an unknown or a loved face. These findings call into question the notion that pictures of loved individuals are shielded against becoming threat cues, with implications for attachment and trauma research. Seeing your loved ones has particular benefits to human well-being and health. Going beyond the advantage of having a supportive social network, the presence of attachment figures has been shown to enhance life expectancy, physical health, and psychological resilience 1,2 . In addition, the mere vicarious presence of loved ones (e.g., by looking at pictures) is related to reduced pain and defensive behaviors [3][4][5] . However, attachment figures may also become a source of grief and misery, and recent translational research started examining the involved severe neurobiological and psychosocial deficits in humans and animals [6][7][8][9][10] . As a highly social species, humans' survival depends on the quality of their social network, and attachment figures provide a sense of safety and security. Looking at pictures of beloved faces evokes a variety of (emotional) memories and draws attention to certain situations that are difficult to ignore. On the psychophysiological level, a pattern of changes occurs that is distinctive of a positive emotional state 5,11,12 . This is shown, for instance, by a biphasic modulation of the heart rate (deceleration-acceleration), inhibition of defensive reflexes (e.g. startle reflex) and the corrugator muscle (frowning), and increases of zygomaticus muscle activity (smiling). In addition, activating a mental representation of attachment figures and supportive others has been shown to reduce pain experience 4,13,14 . For instance, the physical presence of the partner reduced pain, even without a need for interaction 3 . Similarly, Master et al. 15 found that viewing a partner photograph and holding the partner's hand while receiving thermal stimulations reduce pain perception more than holding an object or the hand of an unknown individual. Thus, viewing attachment figures or even their photograph is beneficial for coping with pain and stress, but little is known about social modulators of aversive learning. As an experimental model to investigate affective learning, much research used experiential learning paradigms such as Pavlovian conditioning. In this procedure, a previously neutral stimulus (conditioned stimulus, CS) acquires an affective value by being paired with an appetitive or aversive event (e.g., electric shock serving as unconditioned stimulus, UCS). Importantly, this association leads to conditioned responses to the CS when it is presented alone, as reflected by enhanced autonomic arousal, primed defensive reflexive motor responses, and activation of a neural fear network (e.g., amygdala, anterior cingulate cortex) 16 . Some stimuli, which evolutionary threatened survival (e.g., snakes), have been proposed to be more readily conditioned as aversive, and such prepared fear associations are harder to extinguish 17,18 . Recent studies suggested a parallel notion of prepared safety stimuli, which evolutionary benefited survival and thus be more readily learned as safety cue inhibiting fear responses [19][20][21] . However, humans do not only learn by means of first-hand experiences but through observation and verbal instructions 22,23 . Despite their broad relevance for educational and clinical phenomena, for example, affective and expectancy learning, racial discrimination or phobias [24][25][26] , such social learning processes are still hardly understood. In the present research, we examined the impact of verbal threat/safety learning while viewing loved and unknown faces serving as instructed cues for shock threat or safety. Moreover, instructional learning was used to Materials, design, and experimental presentation. Face photographs of four loved familiar (romantic partner, father, mother, best friend) and four unknown people (another participant's loved ones) were used. The selection of four loved identities was chosen based on previous research showing pronounced patterns of both central and peripheral responses (i.e., increased heart rate, zygomaticus muscle activity, SCRs, and P3/ LPP components), that is distinctive of positive emotions and not attributable to familiarity or undifferentiated emotional arousal alone 5,11,12,34,36 . Moreover, with four identities per category, we were able to achieve a sufficient number of trials for our psychophysiological measurements (e.g., startle EMG) without excessive repetition of single face identities causing habituation effects. Finally, the used partial reversal design in the second experimental block requires at least four stimuli (i.e. maintain threat cue, maintain safety cue, reversed threat-to-safe cue, and reversed safe-to-threat cue; e.g. 27 ). All face pictures were Caucasian, originated from Spain, and were matched for gender and age. For instance, if the participants own romantic partner was male, the corresponding picture of a friend had to be a female face (and vice versa). In addition, participants were asked to provide recent pictures of their mother and father. Picture materials were then matched for size (886 × 886 pixels), color (black and white), and background (light-colored). In a first block, half of the pictures of each face category were instructed as signals for either threat of electric shocks (e.g., mothers and romantic partners) or safety (e.g. fathers and best friends). In a second block, instructed threat and safety associations were partially reversed, in that two faces of each category maintained their original meaning (e.g. loved/unknown mother signaling threat, and loved/unknown best friends signaling safety), and two other faces were reversed (e.g. now fathers cue threat-of-shock and romantic partners signal safety). The assignment of face identities to threat and safety condition was counterbalanced across participants. However, to reduce the impact of within-category variability on threat/safety learning (e.g., due to familiarity or age) 5 , we applied the restriction of having each one high-and one less-familiar person as threat/safety cue in each experimental block (see "Supplementary materials S1"). Thus, the core experimental design (2 × 2 × 2) depicted Face Category (loved ones vs. unknown people), Cue (threat vs. safety) as repeated measures factors in the instantiation block, and in addition Contingency (maintained vs. reversed threat/safety) for the reversal block. In both blocks, threat and safety contingencies were verbally instructed and counterbalanced across participants. The sequence of stimulus presentation was pseudorandom with the restrictions that the same identity could not appear in more than two consecutive trials, and only three consecutive picture-startle or no-startle trials were presented in a row. Importantly, to focus on the impact of aversive anticipations (rather than experiences) no shocks were administered during the experiment. However, to enhance credibility of threat-of-shock instructions, a brief shock work-up procedure was carried out before the experiment started. The experiment began with a 2 min baseline period, followed by two blocks of 64 picture trials each, with every picture being presented 16 times throughout the experiment. Individual trials consisted of 4 s baseline period, 6 s picture presentation, a 4 s post-picture period and a varying inter-trial interval from 2 to 4 s (see Fig. 1). Pictures were presented at approximately 60 cm in front of the participants on a 19″ flat screen monitor. Auditory startle probes were delivered at either 4, 4.5, 5 or 5.5 s after picture onset in half of the picture trials (i.e. 32 probes per block) and were equally distributed across picture categories; four startle probes were also presented during the inter-trial intervals. Startle probes (105 dB, 50 ms) were produced by Coulbourn S81-02 noise generator, gated by a Coulbourn S82-24 audio-mixer amplifier (Coulbourn Instruments, Whitehall, PA) Procedure. An initial telephone interview served to clarify inclusion criteria: (1) having a highly positive relationship with their parents, romantic partner and best friend, (2) having a romantic relationship for at least 6 months up to 6 years (but not living together), and (3) having lived together with their parents at least until the age of 18 years. These latter criteria served to control for the duration of familiarity with regard to instructed threat/safety cues (i.e., parents are more familiar relative to romantic partner and best friend; for a discussion see 5 . Subsequently, instructions for preparation of picture materials were provided: frontal view of the face with a neutral expression, light-coloured background without objects behind, and the picture being taken by someone else other than the participant, to avoid background knowledge about the situational context of the picture. Upon arrival in the laboratory, participants completed a picture familiarity rating to ensure that control pictures were unknown (if not, a different set of control faces was used), and scored relationship quality to their loved ones on a five-point Likert scale "How would you currently define your relationship with your father/mother/ partner/friend on a scale ranging from 1 (very unsatisfactory) to 5 (very satisfactory)?" with 3 as a cut-off. Given the pre-selection and inclusion criteria, relationship quality with the romantic partner (M = 4.5, SD = 0.56), best friend (M = 4.24, SD = 0.54), mother (M = 4.42, SD = 0.64), and father (M = 4.39, SD = 0.64) was rated as very good. In addition, questionnaires on positive/negative affectivity (PANAS 38 ; asking how much participants currently feel e.g., active, distressed) and general social support (MOS 39 ; asking for e.g. the "availability of someone to help if confined to bed") were completed. However, these questionnaire measures were not specifically related to the relationship with their loved ones and assessed for exploratory reasons only. Subsequently, participants were seated in a sound-attenuated room, sensors were attached, and a shock work-up was carried out 40 . To this end, electrical stimulation was increased in steps of 0.1 mA until participants perceived stimuli (M = 0.28 mA, SD = 0.16) and reported shocks as "maximally unpleasant but not painful" (M = 1.34 mA, SD = 0.78). On average, 10.55 stimulations (SD = 6.55) were needed from the perceptual to the unpleasantness threshold. Key instructions were then given verbally about which face identities served as threat and safety cues (i.e. threat/safety contingencies) and the corresponding faces were shown on the instruction sheet. "If you see one of these four pictures, there is always a possibility of receiving an electric shock as long as the picture is present" (i.e. threat cues), while on the contrary "if you see any of these other four pictures, you will not receive any electric shock" (i.e. safety cues). In addition, the participants had the task of looking at all the pictures during the entire time they were on the screen. Following the first block, participants rated all faces regarding perceived threat. (a) An initial shock work-up procedure was carried out to ensure credibility of the threat-of-shock instructions. The first experimental block started with verbal instructions regarding which face identity (ID) is cueing threat or safety (instantiation). To this end, two loved and two unknown face identities were pointed out as cues for aversive shocks (e.g. both loved and unknown fathers and best friends), whereas the other four identities served as instructed safety cues (e.g. mothers and partners). In the partial reversal block, threat and safety associations were partially changed. Each one loved and unknown identity maintained cueing threat and safety, the associations of the other two identities were reversed. Note, the instructed contingencies between face identity and threat or safety were counterbalanced across participants. (b) For each block, all face identities were presented eight times (64 trials) and auditory startle probes were presented in half of the picture trials, four additional probes were presented during ITI. In order to focus on the impact of aversive anticipation (but not experience), no shocks were applied throughout the experiment. www.nature.com/scientificreports/ Before the second block, threat and safety associations were partially reversed. Instructions were the same as for the initial instantiation of threat/safety contingencies but with the changed threat/safety pictures. By the end of the experiment, participants completed the Self-Assessment Manikin (SAM 41 ) to rate all photographs as well as threat and safety conditions in terms of perceived valence, arousal, and dominance. After completing additional questionnaires on empathy and attachment style (Interpersonal Reactivity Index, IRI 42 ; Experience of Close Relationship, ECR 43 ), participants were debriefed and received course credits for participation. Data recording and reduction. To get a comprehensive picture of somatic and autonomic nervous system activation, we assessed a broad set of psychophysiological measures, which had been shown to be sensitive to threat instructions and pictorial stimuli (e.g. 30 ). Skin conductance responses were recorded using Ag/AgCl electrodes with isotonic gel (Biopac Systems) placed on the hypothenar eminence of the left hand and was recorded using a Coulbourn V71-23 coupler module with a sampling rate of 50 Hz. The electrocardiogram was measured at lead II using two standard Ag/AgCl electrodes filled with hyper-conductive gel (Parker Laboratories, Inc, New Jersey, U.S.A.). A Coulbourn V75-04 bio-amplifier, connected to a V75-48 high performance band-pass filter, was used for signal conditioning. Frequencies below 1.5 and above 20 Hz were cancelled out and the electrocardiogram was acquired at 1000 Hz. All EMG activity was recorded by means of miniature In Vivo Metrics electrodes filled with gel and separate Coulbourn V75-04 bioamplifiers. The raw signals were band-pass filtered (28-500 Hz) and subsequently rectified and integrated using a Coulbourn V75-24 integrator. Time constants and sampling rates were 500 ms and 20 ms for the zygomaticus and corrugator, as well as 100 and 1000 Hz for orbicularis muscles activity. Startle responses were scored with an automated detection algorithm 44 , verified by visual inspection. The startle amplitude was defined as the difference between the peak and the onset of the response, in a time window between 20 and 120 ms after stimulus onset. To control for between-subject variability, startle amplitudes for each participant were transformed to T-scores. Skin conductance responses, heart rate, zygomaticus, and corrugator activity were calculated by averaging across each half-second for the duration of the picture display and by subtracting the activity within 1 s prior to the picture onset. Self-report data. As a manipulation check, perceived threat was examined with a repeated measure ANOVA depicting the factors Cue (threat vs. safety), Face Category (loved vs. unknown), and Block (instantiation vs. reversal). Moreover, valence, arousal, and dominance ratings of the face pictures were analyzed by means of repeated measures ANOVAs including the within factors Cue (threat vs. safety) and Face Category (loved vs. unknown). Because these ratings were obtained only once at the end of the experiment, the factor Cue (threat vs. safety) could be tested only for those face pictures that maintained cueing threat or safety throughout the experiment. Finally, the credibility of threat/safety instructions during the instantiation and reversal block (asked during debriefing) was tested with a paired sample T-test. Peripheral measures. For all peripheral measures, repeated-measures ANOVAs were calculated separately for each experimental block (instantiation and reversal) including the factors Faces Category (loved vs. unknown), Cue (threat vs. safety), and additionally Contingency (maintained vs. reversed) for the reversal block. The factor Time (12 half-seconds) was included to examine the temporal development of skin conductance, heart rate, zygomaticus, and corrugator EMG responses. A significance level of p = 0.05 was used, partial eta square (η p 2 ) was used as measure of effect size, and 95% confidence intervals are reported. Greenhouse-Geisser corrections were applied when necessary, and Bonferroni corrections were applied for post-hoc analyses. Results Self-report data. The perceived threat was rated after both instantiation and reversal block (see Figs. 2A and 3A). As predicted, instructed threat cues were more threatening than safety cues in the instantiation block, After reversal learning, all threat cues were perceived as more threatening than safety cues regardless of face category, all ps < 0.001, but this threat effect was more pronounced for unknown compared to loved people. At the end of the experiment, the pictures were rated once in terms of valence, arousal and dominance (see Table 1). For valence ratings, loved faces were more pleasant relative to unknown faces, Face Category Startle reflex. For the instantiation block, the startle reflex was potentiated when viewing instructed threat relative to safety cues, Cue F(1,43) = 39.05, p < 0.001, η p 2 = 0.48 (see Fig. 2B and Table 2). Interestingly, no difference was observed between loved and unknown faces, Face Category Table 2). Planned comparisons revealed these threat effects significant between time points 3.5-6 s after picture onset (all ps < 0.026). Moreover, the non-significant interaction Cue × Face Cat- (Fig. 2), and a marginal interaction Face Category × Cue × Contingency was observed, Fs(1,43) = 3.99, p = 0.052, η p 2 = 0.09. Follow-up analyses indicate that SCRs were more pronounced to unknown faces that were newly learned as cues for threat relative to safety, p = 0.004. This was not observed for unknown faces which maintained cueing threat/safety, p = 0.213, and no differences emerged www.nature.com/scientificreports/ for loved faces, all ps > 0.663. While no further two-or three-way interaction approached significance, Fs < 1.23, p > 0.30, η p 2 < 0.03, however, the overall four-way interaction Cue × Contingency × Face Category × Time was significant, F(11,473) = 3.82, p < 0.023, η p 2 = 0.08, indicating that instructed threat and reversal effects evolved over time specifically for unknown face pictures (Fig. 3B). Discussion The present study examined whether pictures of significant others-the romantic partner, parents, or best friends-are more resistant to becoming threat cues than pictures of unknown people 20 . We further predicted that unknown faces would more readily acquire aversive qualities when threat-associations were reversed. A broad set of psychophysiological measures showed pronounced defensive responding towards face identities, which served as instructed threat relative to safety cues. This differential response pattern emerged for measures of the somatic nervous system (threat-potentiated startle reflex and corrugator EMG), the autonomous nervous system (enhanced SCRs and heart rate deceleration), as well as for self-report (threat and arousal ratings). Interestingly, the zygomaticus muscle was the only measure insensitive to threat instructions. Participants smiled more when viewing their loved ones, regardless of whether they cued threat or safety. Importantly, for the instantiation of threat-associations, no interaction effects were observed between face category and threat/safety instructions for none of the dependent variables. Thus, pictures of loved people became threat cues as easily as it was observed for pictures of unknown people. Regarding reversal learning, however, some indications suggest that changing safety to threat worked better with unknown faces. Taken together, no evidence was found that pictures of loved familiar faces were resistant against becoming threat cues, but unknown faces may be more easily learned as new threat cues. Learning about potential threats by means of social communication is highly beneficial, because an individual does not need to undergo aversive experiences him or herself 22,45 . This notion has received much support by research showing that the mere verbal instruction about the occurrence of threats is sufficient to provoke a pronounced psychophysiological pattern of defensive responding 30,31,46 . The present study replicates these findings within the domain of face and person perception. When viewing face identities that were associated with shock threat (relative to safety), participants were more aroused (enhanced SCRs and arousal rating), oriented towards the threat cue (heart rate deceleration), and defensive reflex activity was potentiated (startle reflex). Moreover, participants tended to frown more towards threat relative to safe identities (enhanced activity of the corrugator muscle). Thus, the mere verbal statement that a person might be dangerous primed defensive psychophysiological responding when viewing these individuals. Knowledge about other people, however, is malleable and can be flexibly updated based on new information. Verbal instructions are particularly effective in changing affective associations [47][48][49][50] . For instance, Costa et al. 27 showed that neutral stimuli associated with threat-of-shock or safety can be reversed from cueing threat to safety and vice versa. Similarly, verbal threat instantiation and reversal instructions can readily override the implicit affective meaning of emotional facial expressions (e.g. a smile may also signal threat 28,33 ). Importantly, however, reversal learning implicates the workings of (at least) two concurrent processes: the inhibition of previously learned threat-associations, while a new threat-association is established 51 . As indicated by self-reported threat (and, on an exploratory basis, for startle reflex and SCR 33 ), the present data provide some indication for the notion that new threat-associations are more readily acquired when threat is linked to unknown people, while concurrently loved people become new safety cues. While encounters with the 'unknown' may be more likely to involve a risk of danger, on the contrary, social relationships with romantic partners and good friends are important health factors 2,12,52 . Here, recent conditioning research suggested significant others as prepared safety cues 19,21 . For instance, using a fear conditioning procedure with pictures of supportive others, unknown people, and neutral objects as conditioned stimuli (100% reinforcement schedule), the authors reported no differential fear conditioning, as measured by skin conductance responses, towards social-support figures serving as CS+ compared to CS− 20 . The present data do not support this notion. During instantiation, we did not find differential threat/safety learning towards pictures of loved www.nature.com/scientificreports/ compared to unknown face pictures, for none of the psychophysiological response measures (ratings, startle EMG, SCR, heart rate, and facial EMG). Moreover, for reversal learning, threat rating and SCR data point to the notion that unknown people may act as prepared fear stimuli relative to loved ones. While several methodological differences may explain the divergent findings (e.g., dependent variables, number of trials, selection of stimuli 3,53 ), several theoretical aspects are of particular interest to further our understanding of the social factors involved in associative threat and safety learning. First, we employed instructional learning, which establishes an association between a particular face identity and UCS by means of verbal instructions but not own experiences. Thus, threat learning occurs with a 0% reinforcement rate and, accordingly, the absence of shocks during the experiment does not necessarily lead to quick extinction learning, as it usually occurs in classical conditioning designs (depending on reinforcement schedule). Such instructed threat associations have been shown to persist within and even across repeated test days without experiencing the aversive events 31,54 , reflecting the workings of worries and apprehensions in anticipatory anxiety. On the other side, instructions can critically shape the impact of previous learning history of allegedly threatening or safe persons 49 . For instance, instructed information has been shown to change feedback-driven aversive learning and still little is known about the combined effects of different learning pathways and prior knowledge (e.g. 48,55 ). Focusing on the neurobiological mechanisms involved in the social acquisition, maintenance and extinction of rather cognitive aspects of fear and anxiety may be particular informative. Second, the use of pictures displaying loved people may interfere less with threat learning compared to pictures of supportive-others. In the present study, we selected participants solely based on their reported high relationship quality but not on perceived social support. Thus, even attachment figures with whom perceived relationship quality is very high, do not necessarily imply helpful support in a threatening situation. Here, the physical presence or absence as well as the type of prosocial or helping behavior might be a more relevant factor than the person offering support 13,15,56,57 . For instance, holding hands with a loved one reduces reported unpleasantness during the anticipation of shocks relative to no hand holding (in happily married women 56 ) or holding hands with a stranger 58 . Moreover, this social regulatory process was associated with inhibition of a threat-related neural network (involving lateral prefrontal, cingulate, as well as posterior parietal cortices), which has been associated with salience detection, vigilance, and emotion regulation (e.g. 58,59 ). Following on from this, the direct comparison of more or less familiar or supportive individuals (e.g., romantic partners, parents, siblings, friends, or fugitive acquaintances) may also be of interest for examining different attachment types (e.g., stable vs. unstable relationships; filial vs. romantic love 5 ) and their relevance as social buffers in the face of immediate and/or prolonged periods of threat and stress (e.g., 60,61 ). Another noteworthy aspect regards the lack of predicted main effects of face category on defensive responding. In a previous study, we observed that viewing loved faces inhibited the defensive startle reflex 5 . However, this was not replicated in the present study. Whereas divergent findings may relate to different tasks (passive viewing vs. instructed threat) and/or reduced trial numbers, other alternative hypotheses are of interest. Specifically, an over-generalization of threat might have occurred across face categories 62 , and/or overwritten the implicit affective picture qualities through verbal instructions 28,33 . This also relates to clinical phenomena, which are observable, for example, in the emergence and treatment of phobias, panic, or trauma-related disorders. While the physical presence of loved ones may help patients to undergo exposure sessions, however, this accompanied exposure could also reinforce fears 'of not making it alone' . Thus, the present findings do not support the notion that loved ones may act as implicit safety cues, nor evolutionary prepared safety signals. In summary, this study shows that pictures of loved familiar people readily acquire threatening qualities. The mere verbal instruction about shock threat was sufficient to provoke a pronounced pattern of defensive physiological responding, even when loved ones served as instructed threat cues. Moreover, language information was highly effective to reverse such threat/safety association. Thus, the present data do not support the notion that loved people are per se safe or resistant to becoming threat cues. In contrast, as we know from the clinical domain (e.g., familial abuse and neglect 6 ), specifically loved ones can become a source of harm and grief. From a developmental perspective, future research could focus on the accelerating and buffering aspects of interpersonal relationships in modulating (mal-) adaptive social threat and safety learning to cope with adverse life events, sensitive transition periods, and challenging environmental conditions (e.g. 8,63,64 www.nature.com/scientificreports/	2021-03-22T17:34:06.834Z	2021-03-09T00:00:00.000	{ "year": 2021, "sha1": "93a11c177b64e973e8016a0698a9513249eb8f74", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-021-84921-3.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "93a11c177b64e973e8016a0698a9513249eb8f74", "s2fieldsofstudy": [ "Psychology" ], "extfieldsofstudy": [ "Medicine" ] }
257861385	pes2o/s2orc	v3-fos-license	Explainable Artificial Intelligence Approach to Identify the Origin of Phonon‐Assisted Emission in WSe2 Monolayer The application of explainable artificial intelligence in nanomaterial research has emerged in the past few years, which has facilitated the discovery of novel physical findings. However, a fundamental question arises concerning the physical insights presented by deep neural networks; the model interpretation results have not been carefully evaluated. Herein, explainable artificial intelligence and quantum mechanical calculations is bridged to investigate the correlation between light scattering and emission in a WSe2 monolayer. Convolutional neural networks using light scattering and emission data are first trained, while expecting the networks to determine the relationships between them. The trained models are interpreted and the specific phonon contribution during the exciton relaxation process is derived. Finally, the findings are independently evaluated through quantum mechanical calculations, such as the Born–Oppenheimer molecular dynamics simulation and density functional perturbation theory. The study provides reliable fundamental physical insight by evaluating the results of neural networks and suggests a novel methodology that can be applied in materials science. Introduction The tungsten diselenide (WSe 2 ) monolayer is a promising candidate for future semiconductor applications [1][2][3] due to its outstanding optoelectronic properties. [4,5] The optical properties of WSe 2 are dominated by comprehensive phonon-exciton interactions, particularly the subsequent energy relaxation pathways, which play an important role in charge-carrier transport. [5,6] Phonon-induced relaxation processes in WSe 2 monolayers have been investigated using various spectroscopic methods, [7,8] and it has been suggested that the LA(M) phonon acts as a carrier relaxation assistant. The zone-edge acoustic phonon facilitates efficient energy relaxation (2p/2s ! 1s) in the WSe 2 monolayer with minimal loss of coherence, [9] and it appears in the form of periodic Raman peaks that have energy distances of 15 meV. [10,11] In contrast, the recent theoretical calculations indicate that the out-of-plane A 0 1 mode promotes chargecarrier recombination in the pristine WSe 2 monolayer, [12] and pump-probe experiments revealed that the phonon mode of the periodic Raman peaks can be generated for A 0 1 (Γ) phonon mode through an anharmonic process. [13][14][15] In particular, because the early stage of charge carrier relaxation has an ultrafast timescale (100 fs), [16] it is difficult to directly observe the optical signatures using a specific analytical technique. Therefore, to reveal which phonons dominantly contribute to carrier relaxation, a comprehensive analysis of the hidden phonon-exciton correlation while considering a large amount of correlative spectroscopic data (e.g., one-to-one corresponding Raman and photoluminescence images) is required. The objective of deep learning (DL) is to extract knowledge from data by training deep neural networks without explicit human instruction. [17] DL enables high-throughput analysis in materials science using large amounts of experimental data. [18,19] In many cases, the output of the DL model provides scientifically The application of explainable artificial intelligence in nanomaterial research has emerged in the past few years, which has facilitated the discovery of novel physical findings. However, a fundamental question arises concerning the physical insights presented by deep neural networks; the model interpretation results have not been carefully evaluated. Herein, explainable artificial intelligence and quantum mechanical calculations is bridged to investigate the correlation between light scattering and emission in a WSe 2 monolayer. Convolutional neural networks using light scattering and emission data are first trained, while expecting the networks to determine the relationships between them. The trained models are interpreted and the specific phonon contribution during the exciton relaxation process is derived. Finally, the findings are independently evaluated through quantum mechanical calculations, such as the Born-Oppenheimer molecular dynamics simulation and density functional perturbation theory. The study provides reliable fundamental physical insight by evaluating the results of neural networks and suggests a novel methodology that can be applied in materials science. intuitive results that can be readily applied to practical tasks. For example, DL models can quickly and accurately determine the number of layers of 2D materials and efficiently detect abnormal atomic configurations using transmission electron microscopy images. [20,21] In this case, the expected output of the model can be compared with the actual experimental data obtained via analytical measurements, and researchers are more interested in the practical aspects of DL models. The advantage of this approach is that the application of DL models can facilitate (or accelerate) the experiment-oriented process, which can otherwise be laborious and time-consuming, as it requires a routine laboratory workflow for the collection of preliminary data. In addition, DL can be used to obtain scientific insights from experimental data. For example, Yang et al. reported that the concentration of valence electrons is the most important component for improving the hardness of as-cast high-entropy alloys. [22] Recently, Lu et al. reported that lattice deformation is a more important factor than the doping effect in the emission process by interpreting a machine learning model. [23] In these cases, the models are trained to comprehensively consider the input and output of the data and are expected to discover hidden correlations inherent in the data. Also, the interpretation of the trained model provides evidence of new physical findings in materials science by evaluating the contribution of the input data during model prediction or by observing the changes in the output data while the input data varies. These methods are also called explainable artificial intelligence (XAI), and they include a class activation map, layer-wise relevance propagation, and Shapley additive explanations. [24][25][26] However, a challenge has arisen in the application of current XAI; for the reliable discovery of physical insights, a careful evaluation of the XAI results is required to be followed. [27] In other words, it is crucial to evaluate the XAI results using reliable external methods although its evaluation is difficult. Herein, we introduce a new methodology by bridging XAI and the first principles, that is, a combination of inductive and deductive analyses. To the best of our knowledge, our study is the first to use quantum mechanical calculations as an external evaluation process in addition to XAI to discover novel physical insights in the field of materials science. In this work, we utilize XAI and density functional theory (DFT) to investigate the origin of the phonon-assisted relaxation pathway of hot excitons. We first measured the Raman scattering (RS) and photoluminescence (PL) spectra and then preprocessed the data using an in-house fitting algorithm, which automates the labor-intensive fitting process. Subsequently, convolutional neural networks (CNNs) that predict PL features from the corresponding RS spectrum were trained to comprehensively consider the phonon and carrier interaction. We then interpreted the trained DL models using intuitive XAI methods, partial dependence plot (PDP), [28] and individual conditional expectation (ICE), [29] which can directly describe the correlation between the RS and PL data. By interpreting the DL model, we discovered that A 0 1 phonon is strongly related to the charge-carrier emission in the pristine region. Finally, we confirmed the coherence of the correlation between the pristine A 0 1 phonon and carrier emission using DFT molecular dynamics and linear response theory calculations. Figure 1 illustrates the overall process of our XAI-applied spectroscopic study. First, we obtained the RS and PL mapping data from several monolayer WSe 2 flakes. Each mapping data contains thousands of spectra, and we treated each pair of RS and PL spectra corresponding to the same location as a singledata instance. Furthermore, each RS and PL spectrum is a mixture of Lorentzian or Gaussian distributions, that is, a combination of multiple peaks, which are known to be related to specific physical properties. To analyze the physical properties of WSe 2 , it is necessary to separate the spectrum into multiple peaks and examine the peak parameter values. The process of separating the peaks from the original spectrum is known as fitting or deconvolution. Because a Lorentzian or Gaussian distribution can be represented by three parameters, that is, position (photon energy) p ∈ ℝ, full width at half maximum (FWHM) w ∈ ℝ, and height h ∈ ℝ, the result of the deconvolution is a set of these three parameters fðp i , w i , h i Þg n i¼1 , which can fully represent the original spectrum using a Lorentzian or Gaussian mixture with n components. Pipeline of XAI-Applied Correlative Spectroscopy However, realizing the deconvolution of a large number of spectra is challenging for human researchers because it is a laborious task. To consider the RS and PL spectra comprehensively, we developed an automated deconvolution algorithm based on the gradient descent method. [30] Our algorithm accurately deconvolutes thousands of RS and PL spectra in a few minutes, thereby enabling a high-throughput analysis of the WSe 2 monolayer. In this study, we considered eight Lorentzian and one Gaussian to fit the RS and PL spectra, respectively, following previous studies; [31,32] our algorithm outputs a vector of RS parameters r ¼ ½p i ; w i ; h i ∈ ℝ 24 (three parameters per peak) and a vector of PL parameters p ¼ ½p; w; h ∈ ℝ 3 . Figure 1a presents an example of the fitted RS and PL spectra. In the next stage, the deconvoluted spectra were used to train and evaluate the CNN models, as shown in Figure 1b. First, we randomly split the data into a training and a validation set in the ratio of 7:3. We then trained the CNN models using the training set to predict the PL p when the corresponding RS r was provided. Here, r is converted into the length-l spectrum R ∈ ℝ 8Âl , which independently describes eight Lorentzian and is fed to the model. After the training process, we evaluated the trained model using the validation set. If the prediction accuracy of the validation set is significantly lower than that of the training set, we can conclude that the model memorizes the training set instead of learning physical insights from it. We repeated the training process by varying the model architectures with different hyperparameters (e.g., the number of training epochs) and observed the same tendencies in the training results. Examples of the model interpretation using PDP and ICE are presented in Figure 1c. The fundamental idea of both interpretation methods is to observe the changes in the prediction while perturbing the input data. The degree of change that occurs during the perturbation indicates the relationship between the input www.advancedsciencenews.com www.advintellsyst.com variable (e.g., a specific RS peak parameter) and the output variable (e.g., the PL peak parameter). Thus, the PDP and ICE evaluate the contribution of each RS peak when the model predicts the PL. For example, if an output variable (e.g., a PL feature) increases, while an input variable (e.g., A 0 1 phonon mode) increases on average during the perturbation, we can interpret that there exists a proportional relationship between these two variables. The PDP shows the average contribution of the input variable to the output variable for the entire dataset (i.e., the entire measurement of the WSe 2 flake). In contrast, ICE exhibits a local (i.e., per-spectrum) correlation between the input and output variables. We repeated the above procedure for several WSe 2 flakes and conditions and derived the results described in the following sections. Furthermore, we evaluated our XAI findings through a DFT-based external analysis, which is independent of the XAI. Detailed explanations of our fitting algorithm, CNN model architecture and training procedure, and interpretation methods (PDP and ICE) are available in Section 4. Result of XAI-Applied Correlative Spectroscopy In this study, we trained CNN models using the experimental results obtained from resonance (2.33 eV) and nonresonance (1.96 eV) conditions to consider the exciton-phonon coupling differences, which vary depending on the excitation energy. In the case of the resonance condition, two different chemically grown WSe 2 monolayers (star-and triangle-shaped, as shown in Figure 2a,d, respectively) were used to train the models. For each experimental data, we split the data into two subsets: training and validation sets. We trained the models using the training set and tested them with both the training and validation sets to determine whether the models were overfitted. On comparing the mean average percentage error (MAPE) and R 2 of the two subsets, we assumed that the models trained on the resonant WSe 2 monolayer found the physical relationship between RS and PL ( Figure S2a,c, Supporting Information). In addition, the images of the PL prediction (Figure 2b,e) accurately describe the nonuniform PL properties derived from various defects on the surface of the chemically grown WSe 2 monolayer. In contrast, the nonresonant WSe 2 monolayer tends to be overfitted ( Figure S2b,d, Supporting Information). We assume that the models simply memorized the training sets because the exciton relaxation pathway is mediated by many long-wavelength phonons instead of specific optical phonons in the nonresonance condition. Therefore, we assumed that the models trained on resonant WSe 2 monolayer data learnt physical insights about the RS and PL and selected this configuration for further analysis. Subsequently, we applied the XAI to the selected DL models (resonance condition) to determine the hidden correlations between the RS and PL. To examine the contribution of each phonon mode, we drew the PDP using the DL model on all phonon modes of the WSe 2 monolayer as parameters (shown in Figure 3a,b). As briefly introduced in the previous section, the Figure 1. Overview of XAI-applied correlative spectroscopy. a) Experimental results of the RS and PL of the WSe 2 monolayer are presented in contour images, and both spectra were calibrated and deconvoluted using an in-house automatic fitting algorithm. b) Resulting fitted RS and PL spectra were used to train the CNN models. The various phonon modes in the WSe 2 monolayer were considered to analyze their independent effects. c) Trained CNN model was interpreted by using XAI methods, PDP, and ICE. www.advancedsciencenews.com www.advintellsyst.com PDP shows changes in the PL prediction when the specific phonon mode varies. The gray region of each spectrum represents the range of 95% of the ground truth PL distribution (y-axis), and the A 0 1 phonon mode describes the widest region of the PL intensity data. This indicates that the A 0 1 phonon mode is the dominant factor for the phonon-assisted carrier relaxation pathway among the various phonon modes in the WSe 2 monolayer. The slightly increasing PDP curve of the A 0 1 phonon mode indicates a positive relationship between the RS and PL intensities. Because the RS and PL intensities imply the frequency of exciton-phonon interaction and the number of emitted photons, the aforementioned proportional relationship is strong evidence that A 0 1 phonon dominantly contributes to the relaxation process of the hot exciton. The vacancy of the chalcogenide atom (V se ) and its oxygen substitution (S O ) are the most frequently observed in chemically grown WSe 2 , and they suppress or increase the RS and PL processes due to lattice distortion and doping effects. These defectinduced physical properties appeared in the inhomogeneous RS and PL images of the star-and triangle-shaped WSe 2 monolayer flakes (Figure 3c,f,g,j). In general, the presence of V se suppresses the first-order optical phonon modes, and PL quenching effects occur due to nonradiative recombination from localized energy states in the vicinity of the conduction band edge. In the case of S O , the intensity of the RS also decreases due to structural deformation, as in the case of V se , but the exciton peak increases because the intrinsic n-type charge is compensated. We plotted the spatially resolved PL ICE slope (Figure 3e,i) to evaluate the degree of proportionality of the A 0 1 phonon according to the type of defects. The inhomogeneous ICE slope of the A 0 1 phonon mode image means that the DL model strongly considers the specific regions of the WSe 2 monolayer because the value of the PL ICE slope indicates the average change in the PL prediction. Thus, the correlation between the PL intensity and A 0 1 phonon is stronger in the bright region of the ICE image (i.e., large ICE slope). To examine the reason for the inhomogeneous slope of the ICE, we performed a comprehensive comparison using not only the A 0 1 phonon (Figure 3d,h) but also the PL (Figure 3f,j) images with the slope of the ICE. The A 0 1 phonon mode exhibits a high ICE slope in the region with strong intensity (pristine dominant), whereas the ICE slope is relatively low in areas with strong PL (S O rich) due to the charge compensation effect. Consequently, we confirmed that the DL model strongly considers the pristine A 0 1 phonon mode due to the higher ICE slope value than the defective regions such as V se and S O . These pristine-dependent properties provide additional insights that the A 0 1 phonon mainly contributes to the charge-carrier relaxation process instead of the LA(M) phonon because the in-plane acoustic phonon is a representative defect-related indicator with a long recombination rate caused by the defect-induced zone-edge renormalization effect. [33,34] Thus, the pristine A 0 1 phonon dominantly contributes to the efficient charge-carrier relaxation process under resonance conditions as compared to longitudinal acoustic phonons. Evaluation of XAI-Applied Correlative Spectroscopy via Quantum Mechanical Calculations Quantum mechanical calculations provide a theoretical basis without empirical elements and are used to explain physical properties by modeling actual circumstances. [35,36] Several recent studies have reported that DFT-based molecular dynamics and www.advancedsciencenews.com www.advintellsyst.com linear response theory facilitate the comprehensive study of charge-carrier dynamics and phonon fluctuations for various defective structures. [37,38] In this work, to efficiently evaluate the contribution of the previous pristine A 0 1 phonon, we performed Born-Oppenheimer molecular dynamics (BOMD) and density functional perturbation theory (DFPT) calculations on three different structures (Pristine, Se vacancies, and oxygen substitution structures, which are shown in Figure 4b; detailed calculation information is described in Section 4.6). The normalized velocity autocorrelation functions are obtained using the BOMD calculation as its Fourier-transformed functions (also known as spectral intensity) enable the identification of the phonons that promote the nonradiative relaxation process of excited carriers, resulting in the energy loss as heat. Figure 4a illustrates the spectral intensity of pristine (shaded gray) and defective WSe 2 with a V se (green) and S O (red). Two strong phonon modes (220 and 270 cm À1 ) exist in the spectral intensity of the pristine structure, and the frequency of 270 cm À1 corresponds to the A 0 1 phonon mode with Raman shift of 250 cm À1 . [39] From the spectral intensity of the V se and S O -containing structure, the A 0 1 À related peak located at the 270 cm À1 peak is reduced, which indicates that the A 0 1 phonon mode induces a excited carrier relaxation in the pristine WSe 2 . Subsequently, we performed the DFPT on pristine and defective structures to confirm the defect-dependent variation of the RS intensity. The presence of the defect induces a structural deformation, which reduces the intensity of the pristine Raman modes (e.g., E 0 , A 0 1 ) and produces abnormal phonon modes (shown in Figure 4c). The noticeable decrease in RS intensities within the defective structures (V se , S O ) corresponds to the correlation described earlier, which is associated with the interpreted DL model and A 0 1 phonon mode. This not only proves that the previous PL ICE is related to the strong A 0 1 phonon mode but also that the pristine A 0 1 phonon contributes to efficient carrier relaxation processes because the defect interferes with the electron-phonon interaction. Throughout the theoretical calculations, we Figure 3. Interpretation of the trained models using XAI methods. The PDPs are shown in a,b). The grey regions in the PDPs represent 95% of the real data (PL intensity). Among all the phonon modes, the A 0 1 phonon mode best describes the PL feature. c,d,f,g,h,j) RS and PL mapping images of chemically grown WSe 2 monolayer are shown, and all colour bar units are divided by 10 3 values. e,i) To confirm the spatially resolved gradients, the ICE curves are visualized in an image form. On comparing the RS and PL images with the ICE, we can observe a positive /inverse correlation between them. www.advancedsciencenews.com www.advintellsyst.com evaluated evidence of pristine A 0 1 phonon contributions for the hot exciton relaxation process, which is consistent with the results of the XAI method. Conclusion In this study, we investigated the correlation between light scattering and emission by a WSe 2 monolayer using XAI and quantum mechanical calculations. The measured spectroscopic data obtained from the monolayer WSe 2 were used to train the DL models. To obtain comprehensive insights into the physical mechanisms of the phonon-assisted charge-carrier relaxation pathway, we interpreted the DL model using PDP and ICE. As a result, we revealed that the pristine A 0 1 phonon mode is strongly related to the PL features, which indicates that the A 0 1 phonon mainly governs the hot exciton relaxation as compared to other phonons. We further evaluated the defect dependence of the optical phonon mode using quantum mechanical calculations, such as DFT-based molecular dynamics and linear response perturbation theory. Through the application of BOMD calculation, we were able to establish the role of the A 0 1 phonon in facilitating charge-carrier relaxation, while the utilization of DFPT calculation confirmed the notable reduction in the A 0 1 phonon mode within defective structures. Our work not only provides a fundamental understanding of the early stages of carrier relaxation in WSe 2 monolayers but also demonstrates an advanced methodology for nanomaterial research. Substrate Preparation: A 300 nm-thick oxide layer deposited on a Si substrate was treated with a 0.5 M NaOH aqueous solution for 30 min, followed by rinsing with deionized water. The treated SiO 2 /Si substrate was coated with the W-precursor solution at 3000 rpm for 1 min. Growth of Monolayer WSe 2 Flakes: A two-inch quartz-tube-equipped twozone furnace system was used to grow monolayer WSe 2 flakes. A quartz boat containing Se pellets and a W-precursor-coated SiO 2 /Si substrate were loaded into the center of the upstream zone (zone 1) and downstream zone (zone 2), respectively. The temperature of each zone was increased to 385 and 765°C for 7 min and maintained for 10 min. Subsequently, the quartz tube was cooled naturally to room temperature. The entire growth process was conducted under Ar and H 2 atmospheres with flow rates of 550 and 5 sccm, respectively, at atmospheric pressure. Measurement: The synthesized monolayer WSe 2 was transferred to a SiO 2 /Si (300 nm) substrate using the wet transfer method before performing the RS and PL measurements. Both the RS and confocal PL mapping data were measured using a multifunctional microscope system (NTEGRA, NT-MDT, Zelenograd Co., Russia & WITec Alpha 300, Oxford Instruments, Germany). The excitation sources were 532 and 633 nm, and high-magnification objective lenses (100x, NA = 0.9) were used. Throughout the experiments, gratings with 150 and 1800 grooves per mm were used for the PL and RS, respectively. Fitting Algorithm: The fitting process was started by classifying the background noise and signal of the actual WSe 2 flake. We assumed that the flake was not placed at the corner of the mapping image and computed the mean www.advancedsciencenews.com www.advintellsyst.com μ and variance σ of the spectra measured from the corner. Subsequently, we removed the signal if the mean of the signal was less than μ þ k Â σ. Thus, if the average of an arbitrary signal was equal to or greater than μ þ k Â σ, it was considered to originate from the actual WSe 2 flake. Next, we eliminated the noise and conducted baseline corrections for each spectrum. To detect noise, we computed the first-order difference in the spectrum. If the difference was larger than the threshold, the corresponding signal was removed and the removed values were linearly interpolated. We divided the spectrum into several areas by a specific wavelength and Raman shift and applied different thresholds to each area. Finally, we performed baseline correction by subtracting a piecewise linear function from the original spectrum. Similar to noise removal, we set several areas along the spectrum and defined the piecewise linear function by connecting the average signal values near the boundary of each area. It should be noted that k and the boundaries of the areas were empirically selected and were different for each type of spectroscopy and WSe 2 flake. After the preprocessing, deconvolution was conducted based on the RAdam optimizer. [30] Starting from the initial peak parameters, the RAdam optimizer searches for the optimal parameters by minimizing an objective function. The objective function was the mean-squared error between the original and fitted spectra. We also introduced two regularization techniques to increase the accuracy of the deconvolution. We enforced that the height values should be positive numbers and regularized the position of each peak such that it did not shift beyond the initial value. The initial peak parameters were selected based on previous studies. [31,32] In the case of RS, fitting was done separately for the first-order and second-order Raman regions (183-300 and 314-410 cm À1 , respectively) because the intensity of the former region was much higher than that of the latter region. Hyperparameters, such as the number of iterations and the learning rate, were also empirically selected and were different for each flake and area. Finally, we skimmed the fitted results and removed the abnormal results or refitted them with other hyperparameters. Training the Convolutional Neural Network Model: Both the deconvolution algorithm and training code were written in Python and PyTorch. [40] Figure S1, Supporting Information, shows the model architectures used in this study. Basically, the models consisted of ten 1D convolutional layer blocks and three linear (i.e., fully-connected) layers. Each convolutional layer block ('conv block' shown in Figure S1a, Supporting Information) is a stack of a convolutional layer, leaky rectified linear unit (LeakyReLU; α = 0.01), and instance-batch normalization. [41] We also tested a residual block ( Figure S1b, Supporting Information), [42] which consisted of three convolutional layer blocks and one LeakyReLU (α = 0.01) for evaluating the effect of model architectures. After two or three blocks, we used average pooling with kernel size and stride of two to reduce the size of intermediate features. The average pooling layer after the last convolutional block output a feature whose shape was (b, c, l), where b is the batch size, c is the channel size of the last convolutional layer, and l is the feature length. To guarantee a fixed-size feature vector for the linear layers, we used an additional average pooling layer that makes l equal 4; therefore, the shape of the resulting feature was (b, c, 4). The feature was flattened before the first linear layer. The linear layers, except for the latest one, used ReLU activation and dropout [43] with a probability of 0.5. We used the Adam optimizer [44] with a learning rate of 0.001. Each training and interpretation were conducted on a single NVIDIA GeForce RTX 3090 GPU with a batch size of 128. We set the number of training epochs to range from 1000 to 10 000, and the training results showed the same tendency (see Figure S2, Supporting Information). We split the entire dataset into training and validation sets, and the ratio of the training to the validation set was 7:3. Subsequently, we trained the models using the training set and tested the model on both the training and validation sets. By comprehensively considering the qualitative (scatterplot) and quantitative (mean absolute error and R 2 ) results obtained from the trained models, we determined whether the model was overfitted and selected configurations (e.g., resonance condition) to be interpreted. For the selected configurations, we trained each model again with the entire dataset (both the training and validation sets) and interpreted the resulting model. We used the model using the convolutional layer blocks (not residual blocks) with training epochs of 10 000 for drawing the figures in the manuscript except for Figure S2c,d, Supporting Information. Model Interpretation: One of the mandatory criteria for interpreting the DL model is high prediction accuracy. When we trained DL models that predicted PL features with RS features for several WSe 2 flakes, the accuracy of PL intensity prediction was satisfied with the criterion, while that of PL position and FWHM did not. Thus, we focused on explaining the correlation between RS features and PL intensity, where the DL model showed superior accuracy in this work. However, it would be too naive to conclude that there was no correlation between RS features and PL position or FWHM. Nonetheless, our observation was that the correlation between RS features and PL intensity was significant. In order to analyze the correlation between RS features and PL intensity, we used PDP and ICE. The basic idea of PDP [28] and ICE [29] is to analyze how the model prediction changes when a portion of the input changes in order to evaluate the contribution of each input variable to the model prediction. The analysis of both methods began with the selection of the input and output features to be evaluated. We assumed that f θ was the model parameterized by θ, X ¼ fx d g D d¼1 is the set of input features where x ¼ ½x 1 , x 2 , : : : , x n ∈ ℝ n is n-dimensional input feature, and D is the number of data. To observe the changes in the model prediction when varying the i-th input variable of j-th data x ðjÞ i , we computed the PDP as follows: In this work, the input X is the RS data obtained from the WSe 2 flakes. Under the same condition, the ICE is computed using Here, ε is a small number for perturbation. Because the ICE is computed for each spectrum, we obtained Figure 3e,i by assigning the average slope of the ICE curve to each pixel value of the corresponding location. In detail, we computed two ICE values by setting ε to AEσ i , the standard deviation of x i in X and computed the pixel value as follows: Theoretical Calculation: Quantum mechanical simulations were performed using the plane-wave method (CASTEP) as implemented in the BIOVIA Materials Studio platform. [45] During the calculation, the WSe 2 monolayer was modeled and geometry optimized with a local density approximation (LDA) functional. To achieve sufficient computational precision, the self-consistent function, the convergence tolerance of the force criteria, the sampling of k points, and the cut-off radius were implemented as 1.0 Â 10 À10 eV atom À1 , 0.01 eV Å À1 an actual spacing of 0.02 Å À1 , and 500 eV, respectively. The norm-conserving pseudopotential and Koelling-Harmon relativistic treatment were used for the entire calculation process. [46,47] Subsequently, we added a 20 Å vacuum slab, representing an isolated layered structure in a 3D periodic system. The proposed optimized pristine WSe 2 monolayer matched well with a previous report (our result: a = 3.268 Å, d WÀSe ¼ 2.518 Å; reference: a = 3.34 Å, d WÀSe ¼ 2.55 Å). [48] To identify the effects induced by the Se vacancy (V se ) and oxygen substitution (S O ), we used the size of the supercell structure (4 Â 4), and quantum mechanical calculations were performed on the V se -, S O -implied WSe 2 monolayer structure. We performed Born-Oppenheimer molecular dynamics simulations [49] using a microcanonical ensemble (constant number of atoms N, volume V and energy E) at 273.0 K with a timestep of 10 fs. In these simulations, we used the LDA functional for electron-exchange correlation, a plane-wave www.advancedsciencenews.com www.advintellsyst.com basis kinetic energy cut-off of 500 eV, and only the Γ-point in the Brillouin zone to enhance the simulation speed. Linear response theory was used to calculate the phonon and Raman tensors of each structure. [50] The electric eigenvalue and phonon energy tolerances were set as 1.0 Â 10 À10 eV atom À1 and 1.0 Â 10 À5 Å 3 , respectively. After determining the Raman activity of all the normal modes, the Raman intensities were obtained by calculating the tensors based on the exposed light wavelength (2.33 eV) and temperature (300 K) with a Lorentzian smearing of 5 cm À1 . Supporting Information Supporting Information is available from the Wiley Online Library or from the author.	2023-04-01T15:02:45.980Z	2023-03-29T00:00:00.000	{ "year": 2023, "sha1": "feedc1f63a93411f0ee76665759d42d6a97f66bc", "oa_license": "CCBY", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/aisy.202200463", "oa_status": "GOLD", "pdf_src": "Wiley", "pdf_hash": "fb6ad77d4c65f4e065658bc20e8a1a8de635c705", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [] }
36162393	pes2o/s2orc	v3-fos-license	Update on Staphylococcal Superantigen-Induced Signaling Pathways and Therapeutic Interventions Staphylococcal enterotoxin B (SEB) and related bacterial toxins cause diseases in humans and laboratory animals ranging from food poisoning, acute lung injury to toxic shock. These superantigens bind directly to the major histocompatibility complex class II molecules on antigen-presenting cells and specific Vβ regions of T-cell receptors (TCR), resulting in rapid hyper-activation of the host immune system. In addition to TCR and co-stimulatory signals, proinflammatory mediators activate signaling pathways culminating in cell-stress response, activation of NFκB and mammalian target of rapamycin (mTOR). This article presents a concise review of superantigen-activated signaling pathways and focuses on the therapeutic challenges against bacterial superantigens. regions of Vβ chains of the T-cell receptor (TCR), leading to activation of both APC and T-cells [7,11,[14][15][16][17]. The term "superantigen" was coined by Kappler and colleagues in 1989 to describe the novel hyper-stimulatory properties of these bacterial toxins [16]. A decade of crystallographic and structural studies revealed their common molecular structure and binding motifs [18], paving the way for investigations of their signaling mechanisms and the way in which these superantigens exert their potent immunological effects. Unlike conventional antigens, superantigens bypass normal "processing" by APC and induce a large proportion (5%-30%) of T-cells to proliferate at picomolar concentrations [7,16]. The excessive release of proinflammatory cytokines and chemokines from APC, T-cells, and other cell types mediate the toxic effects of staphylococcal superantigens [19][20][21][22][23][24][25]. The proinflammatory cytokines, tumor necrosis factor α (TNFα), interleukin 1 (IL-1) and gamma interferon (IFNγ) have tissue damaging effects [26] and together with matrix metalloproteinases (MMPs) and tissue factor produced by superantigen-activated host cells [27], activate both the inflammatory and coagulation pathways. The increased expression of adhesion molecules and chemokine gradient changes direct leukocyte migration to sites of tissue injury [28]. IL-2 from superantigen-activated T-cells causes vasodilation, vascular leak, and edema [29]. Toxic reactive oxygen species (ROS) from activated neutrophils increase vascular permeability and cause acute lung injury [28]. These molecular changes occur rapidly upon superantigen exposure and progress to hypotension, multi-organ failure and death. In addition to inflammatory pathways activated by staphylococcal superantigens, S. aureus also produces numerous virulence factors that aid in its survival and subsequent dissemination in the host. For example, staphylococcal extracellular adherence protein [30] and superantigen-like protein 5 [31] as well as two other staphylococcal surface proteins (the clumping factors A and B) [32] stimulate platelet aggregation which leads to disseminated intravascular coagulation. Targeting the inflammatory and coagulation pathways/molecules represent widely diverse strategies to prevent toxic shock and organ damage resulting from superantigens and various virulence factors [33]. SEB is considered a Category B select agent by the Centers for Disease Control and Prevention (CDC) as it is extremely toxic to humans and can be used as an air-borne, food-borne, and water-borne toxicant. The biodefense objective of mitigation of SEB toxicity in the absence of staphylococcal infection seems simpler when compared to the scenario of replicating pathogens with other virulence factors they produced. Recent efforts have been directed at preventing superantigenic shock, acute lung injury and organ damage resulting from the cumulative biological effects elicited by proinflammatory cytokines. Many reviews and books on superantigens have been published and I will present a concise review on the signaling pathways and give a perspective on the therapeutic modalities for counteracting superantigen-induced shock. Staphylococcal Superantigen Structure and Binding to Host Cells Staphylococcal superantigens are stable, single-chain proteins of 22-to 30-kD that are highly resistant to proteases and denaturation. Despite differences in sequence homology among staphylococcal enterotoxins (SEs) and the streptococcal pyrogenic exotoxins, they have similar protein folds and conserved receptor binding sites [5,15]. These bacterial toxins are classified into five distinct homology groups based on amino acid sequence and similarities in modes of binding to MHC class II molecules [13,15]. Among the different SE "serotypes", SEA, SED, and SEE share the highest amino acid sequence homology, ranging from 53%-81%, whereas SEB is 50%-66% homologous with SECs. TSST-1 has only a limited sequence homology with other SEs. It has a shorter primary sequence of 194 amino acids with no cysteines, and binds TCR Vβ differently than other SEs [17]. TSST-1 lacks enterotoxicity in non-human primates [34] and has a missing "disulfide loop", which may be responsible for the emetic activity of SEs, as mutation of residues in this loop abolishes the emetic activity of SEC2 [35]. There is a separation of the emetic and superantigenic domains of SEs since carboxymethylation of histidine residues of SEB resulted in the loss of emetic activity but not superantigenity [36]. Despite varying sequences, structural and crystallographic analysis of SEA, SEB, and TSST-1 show a conserved conformation with two tightly packed domains containing β-sheets and α-helices [18], separated by a shallow groove representing the TCR-binding site [37,38]. The C-terminal domain has a β-grasp motif found in other unrelated proteins. The N-terminal domain contains an oligosaccharide/oligonucleotide-binding (OB) fold, characterized by the presence of hydrophobic residues in the solvent-exposed regions [18]. Superantigens bind to common, conserved elements outside the peptide-binding groove on MHC class II molecules with relatively high affinity (K d = 10 −8 -10 −7 M) [3,39]. Structural analysis shows at least two distinct binding sites on MHC class II molecules for superantigen. A common, low-affinity binding site involving the invariant α-chain of MHC class II and a high-affinity, zinc-dependent binding site on the polymorphic β-chain [39][40][41][42][43][44][45][46]. SEA can cross-link MHC class II molecules on APC by binding to both sites, and persists longer on the cell surface of APC, prolonging its biological effects [47]. The groove formed between the conserved N-and C-terminal domains of staphylococcal superantigens represents an important interaction site for the TCR Vβ chain [48][49][50][51]. Each superantigen binds to a distinct repertoire of Vβ-bearing T-cells, revealing a unique biological "fingerprint" which might be useful for diagnosing toxin exposure [51,52]. The binding of superantigens to the Vβ chain of TCR is of low affinity (K d = 10 −4 -10 −6 M), with contacts mostly between the side-chain atoms of the superantigen and the complementarity-determining regions 1 and 2 and the hypervariable region 4 of the Vβ chain. Studies with mutants of SEB and SEC3 indicate that a small increase in the affinity of a superantigen for MHC can overcome a large decrease in their affinity for the TCR [48]. Thus, the multiple modes of superantigen binding to MHC and TCR indicate a cooperative effect of interactions in the formation of the trimolecular complex, hyper-activating the host immune system. The superantigen/MHC interactions strengthen their binding to TCR such that they mimic TCR binding to a conventional MHC-peptide complex [49]. Other co-stimulatory receptors on both cells also interact to further stabilize superantigen binding to many cell types [53,54]. A direct binding of SEB to the T-cell co-stimulatory receptor CD28 was reported recently [55]. Peptides derived from the CD28 binding region protected mice from SEB-induced lethality and reduced TNFα, IL-2 and IFNγ expression [55]. This correlates with previous reports of the resistance of CD28-deficient mice to superantigen-induced shock and the lack of serum TNFα and IFNγ after toxin challenge in these mice [56,57]. Three Signals Synergize to Sustain Cell Activation The three signals required for T-cell activation by superantigens and conventional antigens are similar even though superantigens bind outside the peptide-binding groove of MHC class II molecules. The first signal is induced upon the binding of superantigen with TCR-CD3 complex, which activates the Src family of protein tyrosine kinases (PTKs) [58][59][60]. The engagement of co-stimulatory molecules on APC and T-cells, subsequent to superantigen binding, results in a second signal that optimizes and sustains T-cell activation [61][62][63]. The interactions between the adhesion molecules LFA-1 with intercellular adhesion molecule 1 (ICAM-1), and the co-stimulatory molecules CD28 with CD80 on T-cells and APC, respectively, promotes stable cell conjugates. Co-ligation of receptors results in extensive cytoskeletal remodeling and the formation of immunological synapse, initiating signaling cascades [61,64]. PTKs, including Lck and ZAP-70, phosphorylate tyrosine-based motifs of the TCR intracellular components and other adaptors [58,59,65]. The TCR-induced kinases activate phospholipase C gamma (PLCγ) resulting in the generation of second messengers and increase in intracellular calcium levels. One specific second messenger, diacylglycerol (DAG), subsequently activates protein kinase C (PKC) and the proto-oncogene Ras [64,66]. PKC activates downstream signaling pathways including the mitogen-activated protein kinase (MAPK) and the NFκB cascade [67]. Many proinflammatory cytokine genes contain NFκB binding sites in the promoter region and are activated by NFκB [68]. The cytokines IL-1, TNFα, IFNγ, IL-2 and IL-6, and chemokines, in particular, MCP-1, are induced directly by superantigens in vitro and in vivo. The inflammatory environment provided by these proinflammatory mediators represents the third signal for T-cell activation. IL-1 and TNFα activate many other cell types including fibroblasts, epithelial, and endothelial cells to produce other mediators, cell adhesion molecules, tissue protease MMPs, and ROS. IFNγ from superantigen-activated T-cells activates expression of MHC class II and adhesion molecules, and synergizes with IL-1 and TNFα to promote tissue injury, specifically in the gut [10]. Collectively and individually, these mediators from superantigen-activated cells exert damaging effects on the immune and cardiovascular systems, culminating in multi-organ failure and lethal shock. Cross-Talk among Key Signaling Pathways The three signals of T-cell activation exert their potent effects by activating the phosphoinositide 3 kinase (PI3K)/mammalian target of rapamycin (mTOR), NFκB and MAPK pathways [67][68][69]. A description of these signal transduction pathways upon superantigen binding to host cell receptors was presented recently (Figure 1) [70]. Phosphorylation and dephosphorylation events modulate all three cascades with specific kinases and phosphatases. PTKs and lipid molecules from PLCγ activation trigger the PI3K pathway upon specific ligand binding to a number of receptors besides the TCR. Co-stimulatory receptor CD28, IL-2 receptor (IL-2R), IFNγR, growth factor receptor, and G-protein-coupled receptor (GPCR) all activate PI3K [69]. Different PTK inhibitors including genistein, tyrphostin, and herbimycin A, reduced IL-1 levels in TSST-1-stimulated cells [65]. PI3K activates Akt (also known as PKB) and mTOR downstream and modulates many biological processes including cell growth, differentiation, proliferation, survival, migration and metabolism [71][72][73][74][75][76]. The importance of the PI3K/mTOR pathway is shown by the efficacy of rapamycin, a specific inhibitor of mTOR complex 1 (mTORC1), in protecting mice from SEB-induced lethal shock [77]. Rapamycin inhibited SEB-stimulated T-cell proliferation and reduced SEB-induced IL-2 and IFNγ in vitro and in vivo. An alternative pathway of T-cell activation by SEE bypasses PTK tyrosine phosphorylation and uses PLCβ to activate PKC, ultimately activating extracellular signal-regulated kinase 1 and 2 (ERK1/2), NF-AT and NFκB [78]. The MAPK pathway is induced by mitogens, superantigens, cytokines, chemokines, growth factors, as well as environmental stress, and comprises of three major kinase cascades, ERK1/2, c-Jun-N-terminal kinase (JNK), and p38 MAPK. These MAP kinases control fundamental cellular processes to signal cell stress, culminating in the activation of transcription factors NFκB, NF-AT and AP-1 [79], affecting proliferation, differentiation and apoptosis. One common upstream activator of the MAPK pathway is PKC which is activated by TCR, co-stimulatory receptors and GPCR. MAPK promotes inflammation by targeting NFκB to promoters of inflammatory genes [80]. IL-1 and TNFα are both activators and effectors of MAPKs, as these mediators both activate MAPK via various intracellular TNF receptor-associated factors (TRAFs), and are themselves induced by MAPK activation. The proinflammatory cytokines IL-1 and TNFα can directly activate the transcriptional factor NFκB in many cell types that include epithelial and endothelial cells. IL-1 interacts with IL-1 receptor 1 (IL-1R1) and receptor accessory protein, uses signaling molecules, the adaptor myeloid differentiation factor 88 (MyD88), IL-1R-associated protein kinase 1 (IRAK1), and TRAF-6 to activate IκB kinases (IKK), leading to NFκB activation [81]. Phosphorylation of the inhibitory protein IκBα by IKK leads to IκBα degradation and release from cytoplasmic NFκB. This allows NFκB to translocate to the nucleus where it binds to promoter regions of various inflammatory genes [82]. Activation of NFκB leads to induction of many proinflammatory and anti-apoptotic genes. IL-1R1 has structural homology to toll-like receptors (TLRs) which use similar intracellular adaptors and molecules as those used for IL-1R1 for signaling ( Figure 1). TLRs are receptors used by the host to sense pathogen associated molecules such as lipoprotein, peptidoglycan, lipopolysaccharide, flagellin, dsRNA and viral RNA to activate a rapid innate response [83]. Recently, SEB was shown to upregulate the expression of TLR2 and TLR4, thereby enhancing the host response to other microbial products [84][85][86]. This might partially account for the synergistic effects of LPS and SEB in mouse models of SEB-induced shock [87][88][89]. TNFα binds to TNF receptor 1 and 2 (TNFR1, TNFR2) and signals with different intracellular TRAFs, ultimately activating MAPK and NFκB, and results in the expression of other cytokines, adhesion and co-stimulatory molecules [26,90]. An important and damaging component of signaling by the TNFR superfamily which includes various death receptors is caspase activation via the intracellular death domains of these TNFRs. Receptors in this superfamily use intracellular adaptors, TNFR-associated death domain (TRADD) and Fas-associated death domain (FADD) to activate the caspase 8 cascade, JNK, and NFκB. These multiple pathways account for the pleiotropic effects of TNFα including apoptosis, cell activation, coagulation, inflammation, and host defense [90]. The synergistic effects of TNFα and IFNγ on epithelial cells increase ion transport, leading to cell damage and epithelial leakage [10]. The critical role of TNFα in mediating lethality was shown by anti-TNFα antibodies protecting mice from SEB-induced shock in a D-galactosamine (Dgal)-sensitized model [91]. Chemokines, and T-cell cytokines, IFNγ and IL-2, bind to their respective receptors and activate the PI3K/mTOR and MAPK pathways with diverse signal transducers. IFNγ binds to IFNγR, uses Janus kinase 1 and 2 (JAK1, JAK2) to phosphorylate the signal transducer and activator of transcription 1 (STAT1) [92,93]. The main function of IFNγ is in antimicrobial defense as it activates antiviral genes, adhesion molecules, immunoproteasome, and E3 ligase. The IFNγ-activated JAKs also activate PI3K/mTOR independent of STAT1 [94]. Additionally, IFNγ induces the expression and activation of death receptors including Fas (CD95), leading to cell apoptosis [95]. Thus, IFNγ-induced immunoproteasome and CD95 death signaling pathways contribute to vascular cell apoptosis and cardiovascular inflammation [95]. The death receptors use intracellular death domains to activate FADD and caspase 8, resulting in mitochondrial cytochrome c release and DNA fragmentation. IFNγ disrupts ion transport and barrier function in superantigen-activated epithelial cells and these biological effects are amplified by TNFα [96]. However, anti-IFNγ had no effect on mortality and only reduced SEB-induced weight loss and hypoglycemia in the Dgal-sensitized mouse model of lethal shock [97]. A recent study suggests that IFNγ from SEB-stimulated cells plays an important role in autoimmunity in HLA-DQ8 transgenic mice [98]. IL-2 binds to the IL-2R and signals through JAK1 and JAK3 to activate PI3K and Ras, affecting proliferation, growth, and differentiation of many cell types [99]. Ras signals through the MAPK pathway to activate AP-1, cJun/Fos and NFAT. IL-2 increases microvascular permeability and induces vasodilation, resulting in perivascular edema in SEB-induced lung injury [100,101]. IL-2-deficient mice are resistant to SEB-induced toxic shock [102]. IL-6, from both macrophages and activated T-cells, has some overlapping activities with IL-1 and TNFα, and activates JAK3 and Ras upon binding to IL6R [103]. Additionally, IL-6R also signals through PI3K/mTOR to promote cell survival. The Ras pathway used by IL6, IL2, IFNγ, TCR and co-stimulatory receptors results in MAPK activation whereas the alternate PI3K pathway activates mTOR. The chemokines IL-8, MCP-1, MIP-1α, and MIP-1β, are induced directly by SEB or TSST-1 and are potent chemoattractants and leukocytes activators [22,26,104]. Chemokines bind to seven-transmembrane GPCR, induce early calcium flux, activate PLC and signal via the PI3K/mTOR pathway [26,104,105]. Chemokines orchestrate leukocyte migration to promote inflammation and increase tissue injury. Exudates from superantigen-injected air pouches contained predominantly neutrophils with few macrophages [22]. Recruited-and activated-neutrophils produce cytotoxic superoxide and MMPs, contributing to organ damage. Systemic or intranasal exposure to SEB resulted in acute lung injury characterized by increased expression of adhesion molecules ICAM-1 and VCAM, increased neutrophil and mononuclear cell infiltrate, endothelial cell injury, and increased vascular permeability [28,106]. TCR, co-stimulatory receptors and cytokines signal with diverse intracellular molecules to activate PI3K/mTOR, MAPK, and IKK/NFκB cascades. There is cross-talk among these pathways as the MAPKs cascade downstream from TCR, co-stimulatory receptors and T-cell cytokines all activate NFκB, whereas TRAFs from IL-1 and TNFα signaling activate MAPK and NFκB independently. There is some overlap and redundancy of these activation pathways as multiple receptors activate PI3K/mTOR, MAPK and NFκB. However, specificity exists as illustrated by the different classes of MAPKs and their targets. JNK regulates c-Jun and AP-1, and has detrimental effects in the liver whereas p38 MAPK has an additional effect on the phosphorylation of eukaryotic initiation factor (eIF-4E) and promotes translation [79]. The cellular responses to individual cytokines are also different and specific with IFNγ increasing cellular permeability in activated epithelial and endothelial cells whereas IL-1 has prothrombotic effects on the endothelium through the increased production of tissue factor and prostaglandins. Mouse Models of Superantigen-Induced Shock Superantigens from S. aureus and Streptococcus pyogenes are the causative agents of serious life threatening toxic shock syndrome (TSS) and the excessive release of cytokines contributes to the pathogenesis of TSS [1][2][3]33]. SEB has historically been used as a prototype superantigen in biological and biodefense research investigations, as humans are extremely sensitive to SEB especially by inhalation. An obvious step in developing new therapeutic approaches for SEB-induced toxic shock is finding relevant models that mimic human disease. Mice are often used as a model to study the immunological mechanisms of superantigen mediated shock [21,22,25,28,[55][56][57][87][88][89]101]. Although these animals lack an emetic response, they are ideal to work with as immunological reagents are available, the strains and genetic backgrounds including specific MHC class II are well-defined, and the low cost of maintenance allows more animals to be used in experimental protocols. However, mice are naturally less susceptible to SEs when compared to humans because of the lower toxin affinity to murine MHC class II [88,107]. As a result, mice do not develop lethal SEB shock and potentiating agents such as Dgal, actinomycin D, LPS, or viruses are used together with toxin to induce toxic shock [88,91,[108][109][110][111]. Depending on the injury model, sensitizing agents and route of delivery, the severity of disease may involve different organs and distinct profile of mediators. Both Dgal and actinomycin D induced TNFα-dependent hepatotoxicity, and SEB-induced shock models using these potentiators showed much higher serum levels of TNFα not present when SEB was used alone and liver injury was a key feature in these models. Both IL-2 and IFNγ are also critical in Dgal-sensitized models of superantigen-induced shock as IL-2 deficient mice were resistant to SEB-induced shock and antibodies to IFNγ inhibited SEB-induced weight loss and hypoglycemia [97,102]. LPS, a cell wall component of gram negative bacteria, is the most frequently used synergistic agent in mouse model of SEB-induced shock [25,[87][88][89]111]. Relatively high doses of SEB or LPS are used together in these models, usually with Balb/c mice. The shock syndrome induced by superantigens in these models results from the culmination of the biological effects of elevated levels of IL-1, TNFα, and IFNγ, not seen in the absence of LPS [88]. An analysis of the interdependent effects of various doses of SEB used alone and together with LPS in different dose combinations indicated that prolonged levels of certain cytokines were necessary to induce lethal shock in Balb/c mice [25]. Non-survivors in SEB plus LPS groups have significantly higher levels of TNFα, IL-6, MIP-2, and MCP-1 early (eight hours) after SEB exposure [25]. In addition to these mediators, non-survivors showed higher levels of IFNγ and IL-2 later at 24 h. In this LPS-sensitized shock model, lethality and cytokine response were both influenced mostly by the LPS dose and not by SEB. The early TNFα release and sustained levels of IFNγ, IL-2, IL-6, MIP-2 and MCP-1 later correlated with acute lethal shock at 48 h. Since LPS and SEB activate similar cytokines and cells, although using different receptors, it is difficult to compare the molecular mechanisms of shock in this traditional model with human TSS. To avoid the confounding effects of potentiating agents, a "double-hit" low dose SEB model was developed in a LPS resistant mouse strain C3H/HeJ to simulate human TSS [101]. This model mimics human TSS as intranasal delivery of SEB triggers lung inflammation, systemic release of cytokines, and hypothermia that culminate in death at later time points unlike the various potentiated models with much earlier lethal end points. An alternative model using transgenic mice expressing human HLA class II molecules was established to recapitulate human TSS, with different susceptibility to various superantigens dependent on the inserted human HLA class II type, DQ or DR [112][113][114][115][116]. Transgenic mice expressing HLA-DQ6 still required Dgal to potentiate the effects of SEB [112] whereas mice with HLA-DR3 were sensitive to SEB alone [116]. A recent study revealed multiple organ inflammation in lung, liver, kidney, heart and the small intestine that accompanied lethal shock in HLA-DR3 transgenics [116]. Moreover, intestinal absorptive functions were also interrupted in this transgenic model of SEB-induced shock. Vaccines and Therapeutic Antibodies There is currently no available treatment for superantigen-induced shock except for the use of intravenous human immunoglobulin [117]. Antibodies to superantigens can provide broad spectrum protection as neutralizing antibodies against one superantigen cross-react and block the biological effects of a different superantigen [118]. Naturally protective antibodies against superantigens can be found in S. aureus bacteremia and increase in neutralizing titers during infection correlated with recovery [119]. Other studies showed that there is a correlation of lower serum antibodies to TSST-1 in patients with recurring TSS [120,121]. Both monoclonal and human-mouse chimeric antibodies against SEB with high affinities in the picomolar range have been used effectively to target SEB-induced host responses [122][123][124]. The use of antibodies has certain limitations since neutralization of toxins is effective only at the early stages of exposure as it blocks the first step of host receptor interaction before cell activation. Vaccination is a proven method to prevent SEB-induced shock and attenuated mutants of SEB with defective MHC class II binding which lack superantigenicity were efficacious against toxin challenge in mice, piglets and monkeys [125,126]. Inhibitors of Cell Receptor-Toxin Interaction A number of small overlapping peptides, encompassing a conserved region of SEB (residues 150-161), bind to host cell receptors and have been tested to block superantigen-induced effects both in vitro and in vivo with contradictory results using the same peptide [111,113,127]. Although the dodecapeptide prevented transcytosis of various SEs across human intestinal epithelial cell monolayer and may block co-stimulatory signals [128], this and other "SEB peptide antagonists" failed to block SEB-induced T-cell proliferation in human peripheral blood mononuclear cells (PBMC) and had no effect on SEB-induced lethal shock in HLA-DR3 transgenic mice [113]. Blockade of the CD28 co-stimulatory receptor by its synthetic ligand, CTLA4Ig (also known as abatacept) prevented TSST-1-induced shock, and reduced the serums TNFα, IL-2, and IFNγ [129]. Peptide mimetics of CD28 also prevented co-stimulatory receptor interaction with SEB and inhibited SEB effects in vitro and in vivo [55]. Furthermore, a specific CD28-peptide mimetic blocked superantigen-binding to CD28 and attenuated toxic shock and necrotizing soft-tissue infection induced by Streptococcus pyogenes [130]. Anti-CD44 reduced the binding of SEB-activated leukocytes to lung epithelial cells and prevented acute lung injury [131]. Bi-specific chimeric inhibitors composed of MHC class II and TCR Vβ domains competitively blocked SEB binding and cell activation in human PBMC [132]. A soluble TCR Vβ mutant also neutralized the effects of superantigens in vitro [133]. Blocking receptor interaction has many limitations as blockade of host receptors might adversely affect immune function and the inhibitor has to be administered pre-exposure or soon after exposure to toxins for it to be effective. Aside from cross-reactive antibodies, receptor blockade inhibitors are usually specific and have to be tailor-made for a specific superantigen [132]. A list of the therapeutics used in mouse SEB-induced toxic shock models is shown in Table 1, arranged in order of their description in the text. Inhibitors of Signal Transduction and Cytokines Targeting host responses after superantigen exposure is an attractive strategy as these events occur later and may be more amenable to interruption. An important class of therapeutic compounds to be considered is inhibitors that can block signal transduction pathways activated by superantigens. Inhibitors of signal transduction are often cytokine inhibitors as cytokines are the best known "signalers", acting both as activators and effectors. An obvious advantage of signal transduction inhibitors is that they can be administered post-exposure and are likely broad spectrum, inhibiting many different superantigens or even pathogens that trigger similar host responses and signaling pathways. The pathways central to superantigen activation are PI3K/mTOR, MAPK, and NFκB, which are also used by other pathogens, TLRs, and cytokines. Various animal models indicated TSS results from the cytokine signals activating host pathways and inducing damage in multiple organs. The convergence of multiple receptor signaling allows activation of innate host pathway signals to persist and dominate, and these "signals" are likely potential therapeutic targets. An example is NFκB, which binds to the promoter regions of many inflammatory genes implicated in TSS including cell adhesion molecules, cytokines, chemokines, acute phase proteins, and inducible nitric oxide synthase [82]. The downstream activation of NFκB leads to the inducible expression of mediators involved in inflammation and tissue injury seen in SEB-induced lung injury and toxic shock models. A cell-permeable cyclic peptide targeting NFκB nuclear transport prevented lethal shock in Dgal-sensitized mice accompanied by attenuation in liver apoptosis and hemorrhagic necrosis [100,134]. This NFκB inhibitor also reduced SEB-induced inflammatory cytokines and T-cell responses [100]. Table 1. Therapeutics tested for efficacy in murine models of staphylococcal enterotoxins (SEB)-induced toxic shock. * indicates drug is FDA-approved. Pharmacologic agent Target Biological effects against SEB Anti-SEB monoclonal antibodies SEB Neutralized mitogenic activity of SEB in vitro. Prevented SEB-induced shock in HLA-DR3 transgenic mice [124]. Mimetic peptides of CD28 Dexamethasone * NFκB Inhibited SEB-induced proinflammatory cytokines and chemokines in human PBMC. Reduced serum levels of cytokines, attenuated hypothermia due to SEB, and protected mice 100% in both SEB-induced and SEB + LPS-induced shock models [106,135]. Pentoxifylline * Phosphodiesterase Attenuated SEB-induced cytokines in vitro and in vivo. There are other NFκB inhibitors which are FDA-approved for treatment of inflammatory diseases and cancers [148,149]. Dexamethasone, an immunosuppressive corticosteroid, potently attenuated superantigen-induced T-cell proliferation, cytokine release, and cell activation marker expression in human PBMC [150]. Dexamethasone also prevented lethal shock accompanied by attenuation of the hypothermic response, weight loss and serum cytokines in the LPS-potentiated SEB model and the SEB "double-hit" model of toxic shock [106,135]. The pulmonary lesions were reduced by dexamethasone treatment only at later time points (96 to 168 h) and resolution of lung inflammation lagged behind the reduction in cytokines such that a long course of steroid treatment was necessary to rescue mice from lethal shock [106]. Bortezomib, another inhibitor of NFκB, and a proteasome inhibitor, blocked SEB-induced cytokine release but had no effect on lethality or liver necrosis in transgenic mice [136]. Natural products such as epigallocatechin gallate (EGCG) from green tea, and resveratrol (RES) from red wine, are also NFκB inhibitors that separately reduced superantigen-induced T-cell proliferation and cytokine release from human PBMC [151]. EGCG attenuated IFNγ-induced epithelial permeability increases and suppressed T-cell activation and cytokines from SEB-stimulated human PBMC and murine lymph node cells [152]. RES reduced lung injury by blocking SEB-induced T-cell activation, pulmonary permeability increases, and caspase 8-dependent apoptosis [153]. Another upstream inhibitor of NFκB, a synthetic mimetic (EM-163) to the BB-loop of MyD88, reduced multiple cytokines in superantigen-stimulated human PBMC and protected mice from lethal shock in the LPS-sensitized model [137,138]. However, the complete or long-term blockade of NFκB would likely produce adverse side effects as NFκB is essential in maintaining normal host defense and homeostasis [68,154]. Other pathway inhibitors include those directed against the various kinases, PKC, MAPK and PTK. Genistein, a tyrosine kinase inhibitor, and H7, a PKC inhibitor, separately reduced TNFα but not IL-1 from TSST-1-stimulated PBMC [155]. A selective inhibitor of p38 MAPK (SB203580) and an inhibitor of ERK (PD098059) each partially blocked TNFα production from SEB-stimulated human T cell clones [156]. D609, an inhibitor of PLC, which is downstream from superantigen binding to TCR and CD28, blocked SEB-induced effects both in vitro and in vivo [139,140]. SOCS3, an intracellular feedback inhibitor of the various STATs used by IFNγ and IL-2 signaling, reduced the effects of these two cytokines [92]. A cell-permeable form of SOCS3 reduced the lethal effects of SEB and LPS by inhibiting the production of inflammatory cytokines and attenuating liver apoptosis and hemorrhagic necrosis [141]. Immunosuppressive drugs are also good candidates to block superantigen-induced immune responses as they are potent inhibitors against many cell types including T-cells and macrophages. Three FDA-approved drugs for preventing transplant rejection have been used in three different animal models of SEB-induced toxic shock. Cyclosporine A (CsA) inhibited SEB-induced T-cell proliferation in vitro, reduced serum cytokines, and attenuated pulmonary inflammation, but has no effect on lethality in monkeys [157]. Rapamycin, a specific inhibitor of mTORC1, was efficacious even when given 24 h after SEB in the SEB "double-hit" model [77]. Rapamycin blocked SEB-induced T-cell proliferation, reduced serum cytokines, and prevented hypothermia and weight loss induced by SEB. Intranasal rapamycin also protected mice against SEB-induced shock when administered as late as 17 h after toxin exposure, providing a practical route of drug delivery against SEB [142]. Another structural analog of rapamycin, tacrolimus, suppressed superantigen-induced T-cell activation in vitro but did not reduce lethality in HLA-DR3 transgenic mice [143]. Another hallmark of SEB-intoxication is acute lung injury which is most likely a result of oxidative stress inducing damage in the lung. Acute lung injury arises as SEB-, cytokine-and chemokine-activated neutrophils migrate into lung areas producing high levels of superoxide, which is capable of inducing vascular permeability and apoptosis [28]. The anti-oxidants N-acetyl cysteine (NAC) and pyrrolidine dithiocarbamate (PDTC) each mitigated NFκB signaling and T-cell proliferation, and blocked cytokine production in superantigen-activated human PBMC [144]. However, NAC has only a minor effect in vivo, reducing lethality by 30% in the SEB "double-hit" model [145]. Dexamethasone, although effective against SEB-induced shock, required a prolonged dosing of up to four days, which might not be ideal in a clinical setting as dexamethasone is immunosuppressive. Treatment with a short course of dexamethasone (up to five hours post-SEB) provided only 20% protection. Importantly, the combined effects of a short treatment course of intranasal dexamethasone followed by NAC prevented SEB-induced shock, hypothermia and weight loss [145]. Both dexamethasone and NAC are FDA-approved drugs that act distal to toxin binding. Another combination treatment, using a human-mouse chimeric anti-SEB antibody and lovastatin concomitantly and immediately after toxin exposure, also protected transgenic mice from SEB-induced shock [158]. Most therapeutic testing in animal models of SEB-induced shock have targeted proinflammatory cytokines as there is a strong correlation between toxicity and elevated serum levels of these mediators [21][22][23][24][25]. Inhibitors aimed at blocking proinflammatory mediator release overlap with inhibitors of signal transduction triggered by superantigens. The critical role of TNFα in lethal shock was established by the prevention of SEB-induced lethality with neutralizing antibodies against TNFα in Dgal-sensitized mice [91]. IL-10, an anti-inflammatory cytokine, prevented superantigen-induced toxic shock by reducing the production of the proinflammatory mediators IL-1, TNFα and IFNγ [159,160]. The nitric oxide inhibitor niacinamide improved the survival of mice given LPS plus SEB by attenuating serum IL-2 and IFNγ [146]. Doxycycline, an antibiotic, inhibited SEB-induced T-cell proliferation, proinflammatory cytokines, and chemokines in human PBMC [161]. Recently, a panel of different antibiotics was tested for inhibitory effects on cytokine release from SEA-and TSST-1-stimulated human PBMC [162]. Tigecycline decreased IL-6 and IFNγ whereas trimethoprim increased IL-8 and TNFα from superantigen-stimulated cells. Clindamycin, daptomycin, vanomycin and azithromycin had no effect on cytokine release in these stimulated cells. Another study showed that azithromycin suppressed TSST-1-induced T-cell proliferation by blocking ERK and JNK activity [163]. Pentoxyfylline, a phophodiesterase inhibitor used clinically to treat peripheral vascular disease, reduced cytokines and T-cell proliferation in SEB-or TSST-1-stimulated cells [147]. Pentoxyfylline prevented lethal shock accompanied by a reduction in serum cytokines in the LPS plus SEB mouse model [147]. Caspase inhibitors have also been used to attenuate the toxic effects of superantigens as caspases initiate cellular apoptosis and the release of certain cytokines from inactive precursors. The release of IL-1β is dependent on caspase 1, a proteolytic enzyme that cleaves pro-IL-1 into active IL-1β [26]. The caspase 1 specific inhibitor, Ac-YVAD-cmk, attenuated both IL-1β and MCP-1 production in SEBand TSST-1-stimulated PBMC cultures but had no effect on other cytokines or T-cell proliferation [164]. Repurposing of FDA-Approved Drugs for Biodefense Agents As seen from the above studies, FDA-approved drugs currently used for other indications including dexamethasone, rapamycin, cyclosporine A, tacrolimus, bortezomib, doxycycline, pentoxyfylline, NAC, PDTC, have been tested as therapeutics against superantigens with varying degree of success since the 1990s. The testing of FDA-approved drugs for preventing superantigen-induced shock should speed up the approval process for biodefense use in case of exposure. However, as seen from the various FDA-approved drugs tested, even knowing the mechanism of action of these drugs is no guarantee for success as in vivo dosages, dosing routes and schedules as well as animal models all affect the outcome. Rapamycin, by decreasing the levels and effects of IL-2 and IFNγ through mTOR inhibition, is proven to be the most effective single agent to counter both intranasal and systemic exposure to SEB [77,142]. Repurposing of FDA-approved therapeutics makes sense for biodefense use as the therapeutics approval process for human use requires resources and time that might not work for biodefense-related agents. Currently, the approval rate for therapeutics through the FDA is low with 90% of drugs rejected due to safety concerns, inadequate bioavailability or lack of efficacy [165,166]. Intuitively, drug repurposing makes use of the drug's mechanism of action and applies it to diseases or bioterror agents with known or putative pathogenic effects. It bypasses the usual time and resource consuming process of target discovery, optimization, preclinical development and clinical safety testing, and might possibly obtain faster regulatory review by the FDA. This fast track method of repurposing FDA-approved drugs is especially suited for biodefense agents as clinical evaluation of efficacy is usually not possible or ethical. The development of animal models that simulate human diseases by bioterror agents is of critical importance in this non-traditional route of drug repurposing for biodefense use. New avenues to be considered include the use of FDA-approved drugs singularly, in combination, or in tandem. In the case of simultaneous dosing, a lower dose of each drug with different mechanisms of action might produce synergistic beneficial effects and limit individual drug toxicity. Drugs used in tandem will likely be cooperative with the first drug muting out the early host inflammatory response and the second drug acting on secondary signals. Systematic identification of novel synergistic drug combinations will be beneficial to treat a multi-system and complex disease such as TSS. Summary Significant advances have been made in cell activation signals and pathways induced by staphylococcal superantigens. The superantigenic properties of SEB make it an "ideal" toxin to study the cellular interactions, biological effects and therapeutic interventions. Newer mouse models of toxic shock using human HLA class II transgenic mice or SEB un-potentiated mice can better define the systemic effects of SEB and aid in the therapeutic discovery to prevent TSS. Targeting proinflammatory mediators and T-cell cytokines appears to be most beneficial yet not all anti-inflammatory drugs are effective in preventing shock. The use of FDA-approved drugs, rational combinations of FDA-approved drugs, and changing treatment modality are avenues to fast-track and repurpose old drugs for biodefense use. Immunosuppressants, combinations of an immunosuppressant with an anti-oxidant and other carefully tailored combinations hold promise as treatment options for TSS. Disclaimer The views expressed in this publication are those of the author and do not reflect the official policy or position of the Department of the Army, the Department of Defense, or the U.S. Government.	2017-04-05T07:48:39.859Z	2013-09-01T00:00:00.000	{ "year": 2013, "sha1": "b57cb855cfb2d2f21affb4f18b794b75d4a2d33e", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2072-6651/5/9/1629/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "b57cb855cfb2d2f21affb4f18b794b75d4a2d33e", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
258592220	pes2o/s2orc	v3-fos-license	Deficient GABABergic and glutamatergic excitability in the motor cortex of patients with long-COVID and cognitive impairment Objective Attention, working memory and executive processing have been reported to be consistently impaired in Neuro-Long coronavirus disease (COVID). On the hypothesis of abnormal cortical excitability, we investigated the functional state of inhibitory and excitatory cortical regulatory circuits by single “paired-pulse” transcranial magnetic stimulation (ppTMS) and Short-latency Afferent Inhibition (SAI). Methods We compared clinical and neurophysiological data of 18 Long COVID patients complaining of persistent cognitive impairment with 16 Healthy control (HC) subjects. Cognitive status was evaluated by means of the Montreal Cognitive Assessment (MoCA) and a neuropsychological evaluation of the executive function domain; fatigue was scored by the Fatigue Severity Scale (FSS). Resting motor threshold (RMT), the amplitude of the motor evoked potential (MEP), Short Intra-cortical Inhibition (SICI), Intra-cortical Facilitation (ICF), Long-interval Intracortical Inhibition (LICI) and Short-afferent inhibition (SAI) were investigated over the motor (M1) cortex. Results MoCA corrected scores were significantly different between the two groups (p = 0.023). The majority of the patients’ performed sub-optimally in the neuropsychological assessment of the executive functions. The majority (77.80%) of the patients reported high levels of perceived fatigue in the FSS. RMT, MEPs, SICI and SAI were not significantly different between the two groups. On the other hand, Long COVID patients showed a reduced amount of inhibition in LICI (p = 0.003) and a significant reduction in ICF (p < 0.001). Conclusions Neuro-Long COVID patients performing sub-optimally in the executive functions showed a reduction of LICI related to GABAb inhibition and a reduction of ICF related to glutamatergic regulation. No alteration in cholinergic circuits was found. Significance These findings can help to better understand the neurophysiological characteristics of Neuro-Long COVID, and in particular, motor cortex regulation in people with “brain fog”. Introduction As new coronavirus disease (COVID19) variants keep scientists worried, more and more evidence suggest the existence of a postinfectious state known as Long COVID, or, more recently, Post-Acute Sequelae of COVID-19 (PASC, Groff et al., 2021). This clinical entity is defined by symptoms persisting more than four to twelve weeks after recovery from COVID-19 infection (Callard and Perego, 2021;Huang et al., 2021;Nalbandian et al., 2021). Shortness of breath, fatigue, hair loss, autonomic dysfunction, neuromuscular disorders, headache, and attention deficits have been frequently described (Manganotti et al., 2021Cares-Marambio et al., 2021;Lopez-Leon et al., 2021;Michelutti et al., 2022). Other psychiatric and cognitive disorders have also been reported, with descriptions of low mood and ''brain fog", i.e., minor memory impairments and deficits in focusing (Kingstone et al., 2020;Maury et al., 2021;Michelutti et al., 2022). A proper clinical definition of a Long COVID Syndrome has not met experts' consensus yet. However, this cluster of symptoms has been recently referred to as ''Neuro-Long-COVID" (Ferrarese et al., 2020;Helbok et al., 2020). The degree of the impairment has been described to follow a gradient correlated to the severity of the acute phase of the infection (Blomberg et al., 2021). Nevertheless, the chance of developing Neuro-Long COVID has been demonstrated not to be strictly linked to the course of the acute phase of the infection (Carod-Artal, 2021;Su et al., 2022). The cognitive deficits involve mainly executive functions such as working memory, attention, parallel processing, planning and problem-solving. The cognitive deficits have been described both in the acute stage of the COVID-19 infection, with severe involvement of the central nervous system (CNS) (Versace et al., 2021), and in Neuro-Long COVID, with differences in the presentation according to sex (Becker et al., 2021;Hosp et al., 2021;Michelutti et al., 2022). The cognitive deficits reported are typical of syndromes characterized by functional or structural impairment of the frontal and prefrontal lobes (Henri-Bhargava et al., 2018;Jones and Graff-Radford, 2021;Ajčević et al., 2023). These are crucial hubs for working memory, inhibition, cognitive flexibility, planning and problemsolving (Henri-Bhargava et al., 2018;Jones and Graff-Radford, 2021). ''Paired-Pulse" Transcranial Magnetic Stimulation (ppTMS) is a research tool used to investigate and measure the cortical excitation and inhibition in neurological and neuropsychiatric disorders including neurodegenerative disorders such as Mild Cognitive Impairment due to Alzheimer's Disease (Benussi et al., 2021) and frontotemporal dementia (Benussi et al., , 2017(Benussi et al., , 2016Padovani et al., 2018a). Short-latency Afferent Inhibition (SAI) is another TMS parameter that has been largely described to be sensitive to alterations caused by neurodegenerative diseases (Benussi et al., 2021;di Lazzaro et al., 2008di Lazzaro et al., , 2005di Lazzaro et al., , 2002Padovani et al., 2018b;Tokimura et al., 2000). On the basis of a possible alteration of the ''cortical neurophysiology", both SAI and ppTMS have been investigated in patients with neuro-long COVID. These parameters were found to be abnormal both after severe acute COVID-19 infection with neurological complications (Versace et al., 2021) and in Neuro-Long COVID (Ortelli et al., 2022) patients. The aim of this study was to investigate cortical excitability in sensorimotor areas by SAI and ppTMS in a group of Long COVID outpatients complaining of cognitive impairment who did not require hospitalization in their acute infection phase. Participants Participants who were referred to the Neuro-Long COVID ambulatory service of the University Hospital of Trieste from January, the 1st, 2021 to April, the 1st , 2022 were screened for the presence of self-reported cognitive impairment in the post-acute COVID-19 period (diagnosis and recovery confirmed by positive and negative SARS-CoV-2 nasopharyngeal swab, respectively). The Montreal Cognitive Assessment was administered as a cognitive screening test (Nasreddine et al., 2005). Subsequently, a neuropsychological assessment was performed (Michelutti et al., 2022). All the patients underwent a Magnetic Resonance Imaging (MRI) scan. Inclusion criteria for the study were: the persistence of selfreported cognitive impairment at least after 12 weeks from acute COVID-19 symptoms manifestation; age > 18 years and <75 years, positivity of a nasopharyngeal swab, full recovery from COVID-19 at the moment of the assessment (i.e., the absence of any acute COVID-19 symptom). Exclusion criteria for the study were: a history of COVID-19-related respiratory insufficiency or hospitalization; clinical and/or radiologic evidence of acute phase COVID-19 related pneumonia; the presence of severe psychiatric diseases that required management by a psychiatrist in the past; the presence of cardiologic, endocrine and neurologic major comorbidities, i.e. dementia or Mild Cognitive Impairment (MCI) due to other documented neurological diseases; current treatment with benzodiazepines, steroids, antidepressants and other drugs altering the brain cortex excitability (Robol et al., 2004;Ziemann et al., 2015), the presence of cortical atrophy and/or Fazekas score > 1 (Fazekas et al., 1987) on MRI. No restriction was considered in respect to the time between the onset of the symptoms and the time of the evaluation. All procedures were performed according to the declaration of Helsinki and the study was approved by the regional ethics board (CEUR FVG number 007-2021). All the subjects gave verbal consent for the procedures described. The total number of patients enrolled was 20. Two patients were discarded due to the finding of previously not-diagnosed psychiatric comorbidities. Eighteen patients (50 ± 11 years, 100% right-handers; 12/18 females) underwent the neurophysiological assessment of cortical excitability and regulation. Sixteen healthy controls (HC) were recruited for comparison (50 ± 71 years, 100% right-handers; 10/16 females). All of them were recruited between hospital personnel undergoing weekly SARS-CoV-2 screening nasopharyngeal swabs (resulted negative). The recruited subjects underwent all the TMS protocols of this study. Clinical assessment Demographic characteristics, presence of neurological, psychiatric, cardiovascular, respiratory, metabolic, neoplastic, endocrine comorbidities and both acute and chronic (lasting for more than 12 weeks) COVID-19 symptoms were collected. We collected the presence of acute upper respiratory symptoms, fever, dyspnoea, headache, myalgia or joint pain, hyposmia or anosmia, palpitations, diarrhea or gastrointestinal tract symptoms and fatigue in the COVID-19 acute phase. Additional acute phase data were collected for the requirement of ventilation for respiratory failure (Michelutti et al., 2022). The presence of Long COVID was extensively studied, screening for symptoms lasting for more than 12 weeks after the infection onset: the patients were investigated for the presence of persistent fatigue, respiratory symptoms, palpitations, gastrointestinal tract symptoms, myalgia or joint pain, tinnitus, vertigo, visual disturbances, persistent fever. The patients were especially investigated for the presence of persisting neurological symptoms, such as paraesthesia, hyposmia or anosmia/ ageusia, cognitive deficits, mood disturbances, headache, weakness of the arm/leg and insomnia (Michelutti et al., 2022). Cognitive assessment Cognitive impairment was screened by means of the Montreal Cognitive Assessment -MoCA test. The cut-off used was that sug-gested by Aiello (Aiello and Depaoli, 2022;, corrected for age and scholarity. A neuropsychological assessment by trained neuropsychologists was performed on all the patients. It was made of a series of psychometrically validated tests investigating attention and executive functions domain. The Fatigue Severity Scale (FSS) investigated the burden of fatigue on the daily activities of the subjects (Krupp et al., 1989): the subjects had to give a rating from 1 to 7 for every item, with 1 representing full disagreement and 7 representing full agreement. The pre-established cut-off was >4.67 (Krupp et al., 1989). Neuropsychological evaluation of attention and executive functions The Trail Making Test (Siciliano et al., 2019) is a neuropsychological evaluation tool consisting of two parts: part A of the test requires joining a series of numbers arranged in various spatial positions on a sheet of paper as quickly as possible; part B requires quickly joining numbers and letters variously arranged on a sheet of paper alternating between the two categories of stimuli. This test is designed to assess visual-spatial search ability, psychomotor speed and set-shifting skills, i.e. the ability to alternate attention between two different categories of visual stimuli and to perform two cognitive tasks simultaneously. The Symbol Digit Modalities Test (SDMT) (Nocentini et al., 2006) requires quickly associating symbols with numbers, and measures the speed of processing visual information, oculomotor coordination, sustained attention and learning new visual information. The Paced Auditory Serial Addition Task (Saetti et al., 2021) consists in a task in which numbers are presented orally at a rate of 3 seconds and the person must quickly perform serial additions, adding each number to the previous one. The test measures the speed of processing auditoryverbal information, sustained attention in auditory mode, divided attention and verbal working memory. The Stroop test (Brugnolo et al., 2016) assesses attentional control and inhibition, i.e., the ability to inhibit interfering or irrelevant information in order to select the information that is relevant to the task goals. Finally, the graphical fluency tests (Cattelani et al., 2011) assess mental productivity and creativity, the ability to monitor one's own behaviour and to inhibit actions that have already been implemented but are no longer appropriate. Transcranial magnetic stimulation (TMS) protocol The subjects were instructed to sit in a quiet room in a resting position with eyes open. Stimuli were delivered with a stimulating figure-of-eight coil by using a MagPro Ò magnetic stimulator (MagVenture Inc., Alpharetta, GA, USA) delivering monophasic pulses. This was connected to an electromyography device (Synergy Ò , Natus Ò , Middleton, WI, USA). The electromyography signals were recorded with a bandpass of 10 to 1000 Hz. Ten stimuli were delivered for each ISI and protocol in a pseudo-randomized sequence. For all the protocols, the amplitude of the conditioned responses was expressed as a percentage of the corresponding mean unconditioned response. Motor Evoked Potentials (MEPs) were recorded from the First Dorsal Interosseus (FDI) muscle of the dominant side with Ag/AgCl surface electrodes attached in a belly-tendon montage. We used a 7 cm figure-of-eight coil, tangentially oriented over the optimum scalp position to elicit MEPs in contralateral FDI, with the induced current flowing in a posterior-anterior direction (Rossini et al., 2015). Intensities were expressed as a percentage of maximum stimulator output (% MSO). The coil, whose position was continuously monitored during the entire experiment, was placed over the optimal site for eliciting MEPs in the contralateral FDI muscle. The optimal scalp position was determined by moving the coil around the area corresponding to the left M1 (approximately between C3 and P3) in 0.5 cm steps. Then, the optimal scalp position where the stimulation constantly produced the largest MEPs was marked in a tight-fitting plastic swimming cap. For each assessment peak-to-peak amplitude was measured and averaged offline for each participant. No substantial difference in the latencies of the MEPs of all the subjects was recorded. Short afferent inhibition (SAI) Short-latency Afferent Inhibition (SAI) was assessed to evaluate sensory afferents-mediated M1 inhibition. SAI is caused by the excitatory effect of cholinergic thalamocortical projections onto the inhibitory GABAergic cortical network (Tokimura et al., 2000;Valls-Solé et al., 1992). The conditioning stimulus was delivered to the ulnar nerve at the wrist (at an intensity just above the motor threshold for evoking a visible twitch in FDI) and preceded the TMS by an ISI of N20 À4 +0,4,8,ms (hence SAI 16,20,24,28) relative to the latency of the N20 component of the ulnar nerve somatosensory evoked potentials (Alle et al., 2009;. The intensity of the TMS test pulse over M1 was adjusted to elicit stable MEPs of more than 1 mV peak-to-peak amplitude in the relaxed FDI. Statistics and data analyses All statistical analyses were performed with SPSS version 23 (IBM). This is the primary analysis of these data. Data are reported as the medians, (25th-75th percentile) or counts and proportions (%) as appropriate. Two-tailed testing was performed. Mann-Whitney U test was used to assess differences between people with Long COVID and HC. To account for differences between groups in SAI, SICI and ICF considering the different ISI applied, the independent and interactive effect of health status (2 levels between subjects: Long COVID vs. healthy controls) and ISI (4 levels repeated measures for SAI: 16, 20, 24, 28 ms; 2 levels repeated measures for SICI: 3 ms and 5 ms; 2 levels repeated measures for ICF: 10 ms, 15 ms) was performed with a two-way mixed ANOVA. These analyses established the generalized effect of Long COVID MEPs over the different paired pulse protocols (different ISI), and their interaction. In the event of a statistically significant main group effect, a Bonferroni's correction for multiple testing was performed for each ISI. Greenhouse-Geisser correction was applied in case of lack of sphericity. Normality testing using the Shapiro-Wilk test was performed for all datasets. Significance was set for p < 0.05. Data availability The authors confirm that the presented data of this study are saved at the Clinical Unit of Neurology, Trieste University Hospital ASUGI, Italy. They are available upon reasonable request and according to the local institutional and ethics regulation. Results The demographic and clinical features of our patients both for acute and Long COVID stage of the infection are shown in Table 1. The average time from the diagnosis of COVID-19 infection to the evaluation was 165 ± 45 days. The time to access our outpatient service was always shorter than one year from SARS-COV2 nasopharyngeal swab positivity. Cognitive assessment The results of the cognitive assessment are shown in Table 2. All the patients self-reported being cognitively impaired at the time of the examination (18/18, 100%). Regarding the MoCA none of the patients totalized a score lower than the cut-off for pathological impairment, according to the normative data. The patients' median MoCA corrected score was significantly lower than the healthy controls' median score (p = 0.023). All the patients underwent a neuropsychological assessment of attentive and executive functions. The prevalence of the patients performing sub-optimally (Equivalent Score, ES < 3.00, 2.00 and 1.00) in at least one of the items of the assessment are shown in Table 2. All the patients were administered the Fatigue Severity Scale: 14 patients totalized a score higher than the cut-off for the scale. Discussion The main outcomes from this study suggest that LICI and ICF can be lower in people with Neuro-Long COVID with subjectively reported cognitive impairment than in healthy controls, whereas SICI and SAI are not significantly different. The finding of LICI dysregulation with suboptimal executive performances is partially in (Versace et al., 2021) and in patients with Long COVID (Ortelli et al., 2022) affected by a dysexecutive syndrome and an alteration in the perception of fatigue (Ortelli et al., 2022). Both our study and the study from Ortelli et al. (2022) did not find an alteration of GABAa circuits. Differently from them, we did not find any alteration in SAI; furthermore, we did find a reduction in ICF. The neurobiochemical circuits investigated by the TMS paired-pulse protocols (LICI, SICI, ICF) are regulated by GABAb, GABAa (Kujirai et al.,1993;Ziemann et al., 1996) and glutamate (Ziemann et al., 2015(Ziemann et al., , 1996, respectively. SAI is regulated by a thalamocortical cholinergic circuit (Tokimura et al., 2000;Valls-Solé et al., 1992). These TMS parameters have been often considered as early biomarkers for neurodegenerative diseases. Indeed, SICI (Benussi et al., 2016(Benussi et al., , 2021, and ICF (Benussi et al., 2016(Benussi et al., , 2017(Benussi et al., , 2021 have been correlated with frontotemporal neurodegeneration. SICI is also reduced in central nervous system disorders inducing chronic fatigue (Liepert et al., 2005;McDonald et al., 2010;Vucic et al., 2011). In particular, reductions in LICI have been described in patients affected by frontotemporal dementia (FTD, Benussi et al., 2020), and could be related to working memory in healthy middle-aged adults (Redondo-Camós et al., 2022). In addition, a trend for decreased LICI, although not significant, has also been observed in Mild Cognitive Impairment (MCI) due to FTD (Benussi et al., 2021) and in pre-symptomatic genetic FTD (Benussi et al., 2016). Abnormal LICI has also been considered a possible biomarker in different neuropsychiatric disorders such as mood disorder, schizophrenia, epilepsy, and amyotrophic lateral sclerosis (Fatih et al., 2021). GABAa and GABAb receptors are both involved in the spontaneous fluctuation of the persistent activity, i.e., a sustained change in action potential discharge that long outlasts a stimulus (Major and Tank, 2004) in cortical networks (Mann et al., 2009) Mixed-way ANOVA was used. Regarding SICI no significant ISI effect (F 1,32 = 7.299, p = 0.011, p g 2 0.186), group effect (F 1,32 = 2.054, p = 0.161, p g 2 0.060), nor ISI Â group effect (F 1,32 = 2.913, p = 0.098; p g 2 0.083) was found. Regarding ICF no significant ISI effect (F 1,32 = 3.795, p = 0.060, p g 2 0.106), whereas a significant ISI Â group effect (F 1,32 = 4.290, p = 0.046; p g 2 0118) and a significant group effect were found (F 1,32 = 20.215, p < 0.001, p g 2 0.387). . GABAa and GABAb receptor-mediated inhibition would have distinct roles in respectively balancing and terminating ''Up" and ''Down" states of activity (Mann et al., 2009). Disruption of GABAb signaling in the prefrontal cortex has been linked to the impairment of many executive functions such as working memory (Redondo-Camós et al., 2022, Bañuelos et al., 2014Luo et al., 2016;Major and Tank, 2004;Mederos et al., 2021) and goal-directed behavior (Mederos et al., 2021). In our study, the pp-TMS findings could highlight possible alterations of GABAb-mediated inhibition in Long COVID patients. Such alterations could underlie an inefficient regulation of persistent activity states, as GABA represents one of the main inhibitory neurotransmitters in the brain. The decrease in its levels has been observed in patients with cognitive deficits (Porges et al., 2017;Sumner et al., 2010) and behavioral disinhibition in FTD (Murley et al., 2020). A rationale could be suggested considering persistently disinhibited up states of activity due to non-effective GABAb signaling. In our study we also found an abnormal ICF pattern in Long COVID patients. This has not been shown in previous studies (Ortelli et al., 2022;Versace et al., 2021). ICF is related to glutamatergic excitatory signaling (Ziemann et al., 2015(Ziemann et al., , 1996, and it has been reported to be impaired in dementia (Benussi et al., 2020(Benussi et al., , 2017 and MCI (Benussi et al., 2021) due to frontotemporal degeneration. Our observation could be linked to the hypothesis of a neuroinflammation-induced increase in glutamate levels, which could be considered among the putative mechanisms of cognitive deficits in Long-COVID (Mohamed et al., 2022). Our findings show alterations in the TMS-elicited output of the motor cortex in subjects reporting persisting cognitive impairment months after COVID-19 infection. However, our assessment of motor descending pathways by means of single-pulse TMS showed no differences between the two groups of subjects. This indicates the integrity of motor signaling in our subjects. The finding of abnormal GABAb and glutammatergic (investigated by LICI and ICF respectively), as opposed to normal cholinergic (investigated by SAI) regulation of the excitability of M1 likely reflects a more widespread alteration in the regulatory crosstalk between a distributed network of regions that includes frontal and pre-frontal cortical hubs (Balzekas et al., 2018;Du et al., 2019). This taps into previous stages of motor processing such as planning and executive programming. The strength of the connections between motor, pre-motor and pre-frontal cortices during cognitive executive processing has been extensively supported by structural and functional connectivity data (Higashihara et al., 2021;Leisman et al., 2016;Mendoza and Merchant, 2014). For this reason, alterations in the regulation of the M1 output measured by ppTMS could be interpreted as a surrogate marker of activity happening elsewhere in the brain. In fact, alterations of ppTMS parameters in neurodegenerative and psychiatric diseases reflect alterations of cognitive rather than motor functions (Benussi et al., 2020(Benussi et al., , 2017(Benussi et al., , 2016Fatih et al., 2021). Given the correlation between LICI and executive functions (Redondo-Camós et al., 2022), whose heavy reliance on frontal processing has been extensively demonstrated (Peters, 2006;Sanger et al., 2001), one could suspect that the Long COVID dysexecutive deficits that have been reported in our sample of patients could be partially explained by pathogenetic processes involving the frontal areas of the brain. Unfortunately, to obtain a measurable TMS output from regions different than the motor cortex is technically demanding. Future studies including a TMS-EEG in their design would be useful to better localize the dysfunctions in the processes that underlie motor cortical regulation in Long COVID. A frontal epicentre for the deficits in executive functions found in COVID-infected patients has also been suggested by frontoparietal hypometabolism revealed by PET imaging in a study conducted on patients in an early post-infectious time frame (Hosp et al., 2021) and in Long COVID patients complaining of persistent fatigue (Guedj et al., 2021). A recent study has also indicated structural damage to frontal lobe-located, olfactory network-adjacent areas in Long COVID patients (Douaud et al., 2022). These areas are also thought to be involved in executive processing (Arnold et al., 2020;Jones and Graff-Radford, 2021). Preferential involvement of frontal lobes and frontal-adjacent hubs of executive processing has been putatively explained by means of trans-synaptic direct viral transfer (Baig et al., 2020), or immune-mediated damage due to micro-glial reactive activation (Matschke et al., 2022;Poloni et al., 2021) and leakage of pro-inflammatory cytokines through the ematoencephalic barrier caused by the proximity of the olfactory bulb, where a high amount of the virus-targeted ACE2 receptor are expressed (Douaud et al., 2022). Alterations in regulatory circuits in the brain might reflect a mild, often transitory, encephalopathy caused (directly or indirectly) by SARS-CoV2 (Buoite Liotta et al., 2020;Manganotti et al., 2021;Michelutti et al., 2022). TMS could be a useful and interesting tool to test intracortical excitability in Long COVID patients complaining of cognitive impairment. We have described a neurophysiological phenotype corresponding to sub-optimal executive processing in Long COVID patients. Our findings attribute this to the sole glutamate and GABAb-dependent circuits' dysfunction. We aimed to exclude the involvement of cholinergic circuits through an extensive investigation of SAI at multiple ISIs. These findings contribute to provide a more detailed and circumscribed picture of the cognitive Long Fig. 3. Long interval intracortical inhibition (LICI) at 100 ms interstimulus interval in Long COVID (n = 18, violet) and healthy controls (HC, n = 21, blue). Motor evoked potentials (MEP) expressed as the percentage of baseline MEP (i.e., the MEP measured without any previous conditioning stimulus). The difference was tested with the Mann-Whitney U test (p = 0.003). P. Manganotti, M. Michelutti, G. Furlanis et al. Clinical Neurophysiology 151 (2023) 83-91 COVID syndrome. Interesting implications for GABAb-dependent physiopathological mechanisms and therapeutic applications are warranted. Findings from the present study should be cautiously considered and could represent a preliminary investigation due to the limited sample size of both Long COVID and healthy controls; nevertheless, the Neuro-Long COVID participants who volunteered in this study were selected among the people who present to our Neuro-Long COVID ambulatory service and are characterized by symptoms persisting several months after the infection and therefore could be considered as chronically affected by this condition. In addition, to provide a fast and easily implementable protocol in this category of patients, the protocol used in this study applied 10 trials for each condition, and this might have affected the findings, recommending protocols considering a larger number of trials. Future studies on larger samples could not only confirm current literature, including the present findings, but could also assess if associations are present between the neurophysiological alterations and symptoms' severity. Conclusions Neuro-Long COVID patients with self-reported persistent cognitive deficits and suboptimal executive functions could present a reduction of LICI and of ICF, related to GABAb inhibitory and glutamatergic facilitatory circuits, respectively. Conversely, GABAa and cholinergic-dependent regulation seems not to be significantly impaired. TMS might be useful in individuating pre-clinical cognitive impairment in Long COVID. Future studies could provide therapeutic suggestions targeted at the specific neurotransmitters involved in such condition. 6. Author contributions P. Manganotti, MD PhD: conception and design of the study, interpretation of the data, revision of the manuscript, responsibility for the integrity of the work as a whole. M. Michelutti, MD: writing of the manuscript, acquirement of the neurophysiological data, interpretation of the data. G. Furlanis, MD: screening of the patients and inclusion to the study; theoretical input. M Deodato, PT PhD: methodology and statistical analysis, revision of the manuscript. A. Buoite Stella, PhD: methodology and statistical analysis, revision of the manuscript. Conflict of Interest The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Figures are original and not previously published. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.	2023-05-11T13:06:53.659Z	2023-05-01T00:00:00.000	{ "year": 2023, "sha1": "c33c6b0805595a725e22bab1648274642bed6c84", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "c0103fd0f82640891cc19236416424af9992545d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
257804012	pes2o/s2orc	v3-fos-license	The patients’ experience of neuroimaging of primary brain tumors: a cross-sectional survey study Purpose To gain insight into how patients with primary brain tumors experience MRI, follow-up protocols, and gadolinium-based contrast agent (GBCA) use. Methods Primary brain tumor patients answered a survey after their MRI exam. Questions were analyzed to determine trends in patients’ experience regarding the scan itself, follow-up frequency, and the use of GBCAs. Subgroup analysis was performed on sex, lesion grade, age, and the number of scans. Subgroup comparison was made using the Pearson chi-square test and the Mann–Whitney U-test for categorical and ordinal questions, respectively. Results Of the 100 patients, 93 had a histopathologically confirmed diagnosis, and seven were considered to have a slow-growing low-grade tumor after multidisciplinary assessment and follow-up. 61/100 patients were male, with a mean age ± standard deviation of 44 ± 14 years and 46 ± 13 years for the females. Fifty-nine patients had low-grade tumors. Patients consistently underestimated the number of their previous scans. 92% of primary brain tumor patients did not experience the MRI as bothering and 78% would not change the number of follow-up MRIs. 63% of the patients would prefer GBCA-free MRI scans if diagnostically equally accurate. Women found the MRI and receiving intravenous cannulas significantly more uncomfortable than men (p = 0.003). Age, diagnosis, and the number of previous scans had no relevant impact on the patient experience. Conclusion Patients with primary brain tumors experienced current neuro-oncological MRI practice as positive. Especially women would, however, prefer GBCA-free imaging if diagnostically equally accurate. Patient knowledge of GBCAs was limited, indicating improvable patient information. Supplementary Information The online version contains supplementary material available at 10.1007/s11060-023-04290-x. Introduction Patients with primary brain tumors, especially with gliomas, usually receive multiple MRI scans per year as standard care. Patients with a slow-growing low-grade glioma (LGG) may undergo dozens of MRI scans due to their chronic condition. Research endeavors towards faster and more informative MRI protocols are ongoing [1][2][3][4], but glioma MRI protocols remain lengthy and include gadolinium-based contrast agents (GBCA). The patient opinion on radiological care is largely unknown, despite the vulnerability of glioma patients and the relevant implications for patients and physicians. MRI is a crucial pillar of therapy planning and response evaluation in neuro-oncology [5]. Brain tumor MRI protocols tend to adhere to consensus recommendations [6]. Most guidelines include an initial follow-up interval between three to six months after the completion of therapy, depending on the tumor histology. The scanning interval should be decreased to four to eight weeks in case of possible disease progression [7]. However, patients with brain tumors undergo particularly long and frequent MRI scans, while many low-grade brain tumors remain stable for long periods [8][9][10]. Furthermore, the benefit of fixed interval imaging remains unclear [11,12]. Contrast-enhanced T1 weighted imaging (CET1w), often including contrast-enhanced dynamic susceptibility contrast perfusion imaging (DSC), is considered invaluable to the toolbox of neuroradiologists and is standard of care during the follow-up of brain tumors. However, research has shown long-term GBCA deposition, and current patient claims of GBCA-induced side effects are under investigation [13,14]. Therefore, American and European pharmaco-safety agencies urge clinicians only to use GBCA when strictly necessary, but risk-benefit analyses for GBCA are awaited [15][16][17]. Many lesions never enhance or enhance without being high-grade brain tumors. Against this backdrop, it becomes clear why neuroradiological research focuses on strategies to optimize imaging intervals. It also explores using advanced MRI sequences and artificial intelligence to shorten scan protocols and gain deeper insight into tumor biology [18][19][20][21][22][23][24]. This includes imaging without or with reduced GBCA, particularly for the low-grade tumor follow-up in the pediatric population [25][26][27]. The patient opinion on radiological care in brain tumor management is mainly unknown despite a general acknowledgement of the value of patient-centered research and shared decision-making [28,29]. This includes patient opinions on GBCA use. There is a knowledge gap regarding the opinion of the patient on neuro-oncological MRI and research developments in particular, which also has a negative impact on the planning of future MRI research lines. To gain more insight into the patient perspective on neuro-oncological MRI, its follow-up, and the use of GBCAs and to draw conclusions on the patient-perceived urgency of current research lines, we performed a cross-sectional survey on patients with primary brain tumors. Study design and participants The local ethics committee approved the study. A questionnaire was designed in collaboration with our patientreported experience measures department. One hundred patients were estimated as a sufficient sample size following the COSMIN study design checklist for patientreported outcome measurement instruments [30]. Questionnaire targets were adult primary intra-axial brain tumor patients with at least 1 year of known diagnosis who had regular follow-up at our institution. The diagnosis had been histopathologically confirmed or was based on radiological phenotype and multidisciplinary consensus ("scan and wait"). Impairments due to tumor therapy were not an exclusion criterion, nor a selection criterion. All patients that had a neuro-oncological MRI scan and met all the inclusioncriteria were consecutively approached before their clinical MRI scan between 01-09-2021 and 04-08-2022. Patients needed to give informed consent before the clinically scheduled regular MRI scan and were interviewed directly after the scan. Patients with acute impairment, e.g. due to recent brain surgery (early postoperative MRI), were excluded from recruitment, as were patients under legal guardianship. The questionnaire was conducted on Dutch-speaking patients only. Participation was voluntary, and patients did not receive any compensation. Additional information on sex, age at study participation, tumor type and therapy course were added based on medical records. Grade 4 and 3 lesions were classified as high-grade gliomas (HGG), as most lesions were glioma-type. All others were in the LGG category. The WHO classification at the time of surgery defined the diagnosis. The number of previous MRI scans was registered from electronic hospital notes. Questionnaire Originally, the questionnaire contained ten questions plus one open comment space. It was extended by one additional question (question 11) during the course of the study in order to gain additional information on patient knowledge of GBCA. First, the patient was asked to estimate their total number of tumor-related brain MRIs received until present. This was to evaluate if patients realistically assess the burden of MRI during the disease. The other questions required singlechoice categorical and ordinal tick-box answers. Eight out of eleven questions allowed multiple answer options. Questions 2 and 10 were general questions on the burden of undergoing radiological follow-up as a patient with a primary brain tumor. Questions 3-6 covered the dimension of burden due to GBCA injection. Questions 7-9 covered the burden of the MRI scan procedure beyond GBCA injection. An open comment section ended the questionnaire. These comments were categorized into four groups: general burden of MRI, attitude towards GBCA injection, the burden due to follow-up/scan interval, and others. The English version of the PENGUIN questionnaire can be found as Online Resource 1. Analysis Data was analyzed with descriptive statistics. Subgroup analyses were performed for sex, age, tumor grade, and the number of follow-up scans. Question 10 was transformed to a binary metric during the analysis, as patients were allowed to give multiple answers. If patients left all the boxes empty, they were, after confirmation by the patient, categorized as not having any stress. The subgroup comparison involved parametric testing with the Pearson chi-square test and the Mann-Whitney U-test for categorical and ordinal questions, respectively. We also performed the Pearson correlation test. Bonferroni correction was performed for multiplicity. The significance level was set by dividing the significance threshold (0.05) by the number of subgroups (4) at p < 0.01. Patient demographics Of the one hundred filled-in questionnaires, 61 were from male patients (mean age ± standard deviation (SD): 44 ± 14 years) and 39 from female patients (mean age ± SD: 46 ± 13 years, Table 1). The age difference was insignificant between male and female patients (p = 0.4). Ninety-three patients had a histopathologically confirmed diagnosis. The other seven patients were considered to have a slow-growing low-grade lesion based on their radiophenotype and growth rates. In total, 41 patients had a HGG. Online Resource 2 shows the number of patients for each tumor entity. General burden of MRI Patients systematically underestimated the number of scans they had undergone (49 underestimations vs 39 overestimations). An increase in the number of follow-up scans was associated with an increased underestimation of MRI burden (Fig. 1). If overestimation occurred, it was more marked than in cases of underestimation: 5 + 6.57 scans overestimated compared to 3 + 4 scans (median + interquartile range) Patient estimate on confirmed minimum number of MRI brain scans for glioma diagnosis ≥ 1 0 s c a n s ≥ 1 5 s c a n s ≥ 2 0 s c a n s ≥ 2 5 s c a n s ≥ 3 0 s c a n s ≥ 3 5 s c a n s ≥ 4 0 s c a n s Number of scans Percentage of people Fig. 1 Percentage of patients overestimating, underestimating, or guessing correctly the number of MRI for glioma they had undergone until questionnaire session (y-axis) as sorted by the number of MRI they had truly undergone (x-axis). Underestimation of scan burden increases with the number of MRI undergone underestimated, respectively. There was no significant difference in over-or underestimation between men and women (p = 0.83). The general trend was that patients did not consider MRI scans burdensome, as shown in Fig. 2. It appeared that older patients experienced lying in the MRI as longer than younger patients, albeit not significantly (Q9, Online Resource 3). The most frequent stress factor was fear of outcome/bad news (Fig. 2). Five patients found the noise annoying or suggested extra hearing protection beyond the existing double layer. Women reported more symptoms during or directly after the MRI (Q7) and found it more annoying than males (Q8). Women also experienced more stress from fearing bad news (Q10.1) and the travel times to the MRI unit (Q10.2) and experienced more stress in general (Q10.6). Patients with less than ten previous MRIs were significantly more dissatisfied regarding the scan follow-up interval than patients with 30 or more scans. The same group (< 10 scans) showed significantly less fear towards receiving an intravenous (IV) cannula. Finally, LGG patients experienced more claustrophobia than HGG (Online Resource 3). Attitude towards GBCA injection Most patients did not experience GBCA injections as burdensome (Fig. 3). However, 40% described at least some irritation when receiving IV administration. 63% preferred an MRI scan without GBCA if considered diagnostically equivalent. Fifty-eight patients answered the additional question about possible adverse effects from GBCA, but only three patients were aware of these. Nearly all patients found the wait between placing the cannula and taking the MRI perfect or short. Several patients wrote that they found cannulas annoying and painful and would prefer no cannula if diagnostic performance was maintained. One patient also mentioned that patients should receive more information regarding the use of GBCA agents. Female patients found cannula placement significantly more unpleasant than males (Q4). There were no significant differences between males and females regarding preferences for MRI options without GBCA (Q5). Discussion Patients with primary brain tumors expressed a generally positive attitude towards the current neuro-oncological MRI follow-up scheme in this monocentric survey at a tertiary academic center. However, GBCA-free MRI protocols would be preferred, provided their diagnostic non-inferiority. Importantly, patient knowledge about any potential adverse effects of GBCA was rare, and we identified women as less satisfied. At the same time, age, diagnosis and number of previous scans had no impact on satisfaction. Patients underestimated the number of scans they had undergone, with underestimation being positively correlated with the number of previous scans. Scan burden underestimation can be explained by 'positivity bias in memory'-a phenomenon describing a person's inclination to remember pleasant events more vividly and favorably than unpleasant ones [31,32]. Patients with 30 scans or more were significantly more satisfied with the number of scan follow-ups than patients with ten scans or fewer. We hypothesize that patients with more scans are more likely to think they are in a stable phase of their disease than patients who only recently got diagnosed. There is a tendency in the medical community to reduce both MRI frequency and protocol duration, scan time. Arguments are costs, waiting lists, and the assumption that patients find the MRI uncomfortable and have difficulty complying [33][34][35]. However, our results showed that most patients did not experience MRI as burdensome and that the follow-up intervals are perceived as appropriate-even by frequently scanned glioblastoma patients. The debate about whether scan frequency and protocol duration need to be reduced should include the patients, as they might oppose longer control intervals. On the other hand, data implies that most patients will tolerate moderate scan duration extension for imaging research. However, certain patients experienced at least moderate discomfort during the MRI scan. It is worthwhile to study this group in more detail. Our most remarkable and also most consistent finding is the role of sex in the perception of MRI. Overall, women found the MRI procedure more uncomfortable than men, which was characterized by experiencing the MRI procedure as more unpleasant, more often being afraid of bad news, and having a tendency to be more stressed about the travel times. These findings align with literature suggesting that women experience more stress and anxiety also when confronted with a brain tumor diagnosis [36,37]. Women also found receiving a cannula more unpleasant. Research has reported that sex is a risk factor for difficult venous access and that catheter insertion in women is more difficult, explaining the difference in comfort [38]. We conclude that sex, and most likely gender, is not sufficiently reflected in the current MRI workflow of brain tumor patients despite indicators for relevant differences between male and female perception. According to our results, women will benefit from shorter MRI protocols-and should innovation permit it-even GBCA-free ones. The discussion between patient welfare and patient clinical needs should therefore be carefully balanced. The age, number of previous scans, and diagnosis had surprisingly little impact on patients' MRI perception. Increasing age is a known stress factor for patients and MRI [39]. In our study, we could only confirm a tendency below the significance threshold regarding age: older patients tended to experience the MRI scan as longer and less comfortable than younger patients. This is relevant as brain tumor MRI protocols are particularly lengthy. Patients with ten or fewer scans showed less fear of receiving a cannula than patients with 30 scans or more, at a significant level before and a possibly still relevant level after Bonferroni correction. While the probability of a negative experience with cannulas increases with the growing number of scans, this contrasts with what would be expected as dictated by exposure therapy. With exposure therapy, frequent engagement with anxiety-provoking stimuli, such as cannulas, can reduce and disconfirm a person's fearful projections towards the respective stimulus [40,41]. Patients with a low-grade lesion experienced more claustrophobia than patients with a high-grade lesion. Patients in the low-grade group had a mean age of 41 years, while patients in the high-grade group had a mean age of 49, as expected. Even though the age differences between the two groups were normally distributed, the age difference could explain this finding as younger patients tend to be more stressed [42]. As younger patients with low-grade lesions will likely receive more follow-ups during their life time, any scientific innovation towards shorter protocols will be particularly in their favor. Clinically, the time between scans, the number of included sequences, and the decision of administering contrast is generally based on the lesion type. However, our results show that tumor type, a reflection of disease severity, does not play a relevant role in the perception of MRI. While our research did not focus on the patient-disease relationship, there may be a link between the patients' tolerance for number of follow-up scans and scan duration and the type of disease. An estimated ~ 40% of all MRI scans in neuroradiology are GBCA-dependent [43]. Therefore, patient opinion on gadolinium should be considered with the aim of shared decision-making [44]. Our results show that most patients would opt for an MRI scan without GBCAs if considered diagnostically non-inferior, supporting research in that direction. However, patients showed a profound lack of knowledge of GBCAs, including insufficient knowledge regarding possible adverse effects despite being a frequently prescribed diagnostic agent. Patients seemed to be poorly informed and could thus not make optimal decisions about their welfare. At this point, it must be understood that patients in the Netherlands will usually never meet with a radiologist, nor is written informed consent for MRI examinations with GBCA mandatory. Potential contraindications for MRI are ruled out by the clinician ordering the MRI scan. A detailed procedure description for the patient is usually not part of this conversation. Especially considering that pharmaco-safety agencies urge clinicians to reduce the use of gadolinium in the clinical workflow of glioma patients [16,45], patients should be well-informed about the added value of GBCAs and their possible harm to the human body [13]. Beyond considerate use of GBCA and conciseness of scan protocols, there are other factors which may be relevant to increase patient comfort such as an acoustic optimization of sequences, as was confirmed by the comments of five of our participants [46]. There are several limitations to this study. First, this is a monocentric study with Dutch patients only, with consequences for data interpretation. The sample was, however, representative of the disease's general prevalence regarding diagnosis, age, and sex distribution [47]. Second, some of our patients have outdated diagnosis without an IDH classification which may result in high grade glioma patients being considered low grade glioma. Patients for whom the questionnaire was too much of a burden were excluded, as were non-Dutch-speaking patients due to the language barrier. This biases the study, potentially underestimating the MRI burden by excluding the sickest patients and patients with a different cultural background. Further, the reference number of MRI scans derived from hospital records is a minimum estimate. Patients could also have been scanned elsewhere. However, patients in the Netherlands usually adhere to one clinic only for treatment and follow-up-making a relevant deviation in the correct total number of scans unlikely. In summary, this study finds that patients with primary brain tumors generally have positive experiences with neuro-oncological MRI. Especially women, however, would support endeavors towards GBCA-free MRI diagnostics and shorter protocols. Approaches to reduce imaging frequency are neither a patient priority, nor preference. A lack of knowledge on GBCA indicates that shared decision-making remains an unreached goal in glioma imaging. Acknowledgements The authors do not have any acknowledgements. Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by IJHGW, HLH, and VCK. The first draft of the manuscript was written by IJHGW and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Competing interests The authors have no relevant financial or nonfinancial interests to disclose. Ethical approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Amsterdam University Medical Center, location VUmc Ethics Committee (Date 26-08-2021/No 2021.0384). Consent to participate Informed consent was obtained from all individual participants included in the study. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.	2023-03-30T06:16:30.606Z	2023-03-28T00:00:00.000	{ "year": 2023, "sha1": "ecf39cd12ab7a93cb8f60882ff573b0e72e1ec5d", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s11060-023-04290-x.pdf", "oa_status": "HYBRID", "pdf_src": "Springer", "pdf_hash": "d72db7457962c665cc0a2ef12e7130f5fcb1aedb", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
145292752	pes2o/s2orc	v3-fos-license	The Role of Teachers in College English Classroom — From the Perspective of Affect College English teaching in China has been paid great attention since the language of English, as a world-wide one, is of great importance in international communication. Following the traditional teaching strategy, most people regard language learning as a result of imitating and practicing. However, more and more teachers and researchers emphasize that affective factors of students in English learning should be concerned in the college English classroom. Teachers should develop students’ positive affection to learn English, make sure full participation of students. Since then, the teaching strategy is changed from teacher-instructing to student-centered teaching. After taking research of previous studies and practical teaching, this paper points out affective factors, language teaching methodology, and teacher’s task to boost college English teaching in China. Introduction College English teaching in China has been paid great attention since the language of English, as a world-wide one, is of great importance in international communication.Following the traditional teaching strategy, most people regard language learning as a result of imitating and practicing.However, more and more teachers and researchers emphasize that affective factors of students in English learning should be concerned in the college English classroom.Teachers should develop students' positive affection to learn English, make sure full participation of students.Since then, the teaching strategy is changed from teacher-instructing to student-centered teaching.After taking research of previous studies and practical teaching, this paper points out affective factors, language teaching methodology, and teacher's task to boost college English teaching in China. The Definition of Affect Actually, it is hard to give an accurate definition of affect as Fehr and Russell (1984) once noted "Everyone knows what an emotion is, until asked to give a definition".Affect refers to emotion or feeling (H.D. Brown, 1987).Dickinson (1987) describes it as being concerned with the learner's attitude towards the target language and users of it, and with his/her emotional responses.Damasio(1994) differentiates the term emotions as "changes in body state in response to a positive or negative situation" from the term feelings as "perceptions of these changes.Arnold (1999) defines affect as "aspects of emotion, feeling, mood or attitude which condition behavior".It can be seen that the term "affect" sometimes replaced by "emotions", "feelings" or "affectivity" was defined from several perspectives such as educational one, psychological one.However, from the aspect of language teaching, Walter Apelt and Heike Koering (1997) demonstrate that affectivity is the totality of all components of foreign-language instruction that influence the emotional attitude toward learning a foreign language and toward using it, as well as the foreign language atmosphere in general and the success of the learning and teaching process in particular.Not only those components that promote intensive and language-activating emotions, but also those that inhibit or evoke negative emotions, should be given special consideration.This definition is more reasonable, because it covers both learning and teaching, clarifies positive affect and negative one and connects affect with cognition. Affective Factors Though affect cannot be separated as a complex system in some sense, studies tend to name different aspects of affect in order to carry out specific research of each aspect.In general, emotions which affect language acquisition can be classified as personality factors and factors between learners themselves and their relationship with teachers.Personality factors involve self-esteem, motivation, anxiety, and inhibition while the other involves empathy, classroom transactions and cross-cultural processes.Among these factors there are positive ones which can encourage learners and negative ones as well which will be hindrance to English learning such as anxiety, sadness. The Role of Teachers in College English Classroom As Pine and Boy (1997) express "pupils feel the personal emotional structure of the teacher long before they feel the impact of the intellectual content offered by that teacher".It is evident that teachers' performance at class will give an influence for their students.A teacher who lacks self-esteem will find it difficult to make self-esteem of his students.A teacher who does not lead a warm atmosphere at class will find low spirits of students to learn.So the teachers' role is very critical in language teaching.According to Yan & Zhang (2002), there are three basic roles for teachers: "lecturer", "teacher", and "facilitator".They made distinctions among the three based on the theory of affective factors."Lecturers" are those who solely consider their professional skills but neglect teaching methodology."Teachers" here refer to those who possess professional skills and teaching methodology but seldom care students' affective experience."Facilitators" are like those who not only take characters of the former two but also care students' affective state and learning process to help them in language learning by self-consciousness-raising.Through comparison, it reflects that lecturers lack flexibility during their teaching and there are less interaction between students and lecturers as if there exists an invisible wall.It is better for teachers to realize how to turn burdensome language class into dynamic one but they pay less attention to the students' involvement or do not perform enough to release the capacity of students because of individual differences.As facilitators, they try to break the invisible wall and communicate with students in time so that the previous opposite two sides can be changed into a harmonious group.Both learning and teaching go on smoothly without tension.It is no doubt that such facilitators can be successful ones for they explore attentively students' psychological feelings, skillfully manipulate students from loving language classes and attract students to participate actively. Being a Humanistic Teacher To be a humanistic teacher means a teacher's behavior should be democratic and his teaching aims at student-centered classroom in which a teacher acts as an organizer, encourager, and guide.Rogers (1969) argued that learning that combines intelligence and affection would promote the whole-person development.He regarded himself as a facilitator and asserted that significant learning can be facilitated by establishing an interpersonal relationship between the facilitator and the learner.Teachers should treat their learners as individuals with specific needs to be met and provide them with trust and emphatic understanding.Through the understanding and promotion of inner factors, students' learning strategies and even their learning outcomes would be of much distinction compared with what they did before.According to Gage and Berliner (1991), feelings are as important as facts.Much work from the humanistic view seems to validate this point and is one area where humanistically-oriented educators are making significant contributions to our knowledge base. Motivating in English Teaching Process It is well acknowledged that motivation is very crucial in language learning as Dornyee(1998) mentioned that motivation has been widely accepted by both teachers and researchers as one of the key factors that influence the rate and success of second foreign language learning.Motivation provides the primary impetus to initiate learning the foreign language and later the driving force to sustain the long and often tedious learning process.Therefore, at class teachers should give their priority to motivation.Students will learn from their experience and keep collecting to develop their full potential.Otherwise, classroom realities increasingly ask awareness to an overlooked aspect, i.e., the motivational needs from teachers, since teachers' motivation has direct influence to students' language learning efficiency.However, ways to motivate language teachers take little amount of research but it will broaden the research area. Designing affective course Affective course requires teachers' large amount of time and energy contributed to make a good arrangement of class which not only develop students' academic skills but also stimulate their inner motivation.Two activities are taken from Moskowitz's illustration here: Activity 1: I like you; you're different For homework, students write three positive and unique things about themselves on a card provided by the teacher, e.g.: 1.I am a good cook; 2. I was on TV when I was eight; & 3. I was a tennis champion in my hometown.The teacher reads each card aloud.Students suggest three possible student identities and the whole class votes on the one most likely.The mystery students reveal themselves and answer a few questions from classmates related to the card.This activity helps students learn about each other and promotes self-esteem. Activity 2: Fortune cookies In groups of four, each student writes a positive fortune for another group member, folds it with the student's name on the outside of the paper and puts it in the center of the group.In turn, eacherstudent takes the designated fortune, reads it aloud and reacts to it.To end, each group chooses one fortune to read to the class.Laughter and good feelings toward classmates ensue from these positive wishes.(Moskowitz, 1982) By these activities the whole class integrates together with each member shows his self-esteem and empathy instead of anxiety.Sometimes teachers can arrange interesting programs, like group-discussion, role-playing and video-watching, to make full practice of students and at the same time the ability of speaking, writing, listening and reading can be developed. Creating a Psychologically Secure Learning Environment for the Students In the process of English teaching, teachers should pay more attention to establishing certain relationship with their students.A harmonious and pleasant climate in the classroom can help to reduce the anxiety of students, ask the focus of students when learning English and form emotional bonds between students and teachers at the same time.Teachers can create the classroom a welcoming and relaxing place where psychological needs are met and language anxiety is kept to a minimum.(Oxford & Shearin, 1994) According to Gage and Berliner (1991), students learn best in a non-threatening environment.This is one area where humanistic educators have had an impact on current educational practice.The orientation espoused today is that the environment should be psychologically and emotionally, as well as physically, non-threatening. Conclusion To summarize, affective factors cannot be neglected in English language teaching.Among the three basic roles, "facilitators" are the most ideal one the responsibility of which is to organize effective and efficient teaching strategies for arousing students' awareness of affect and take their needs into consideration in order to equip them to learn with self-consciousness and efficiency.As Pine & Boy (1997) mentioned: the more the teacher humanizes his teaching, the more teaching humanizes him.The more the teacher cares for his students, the more they will care for him.The more the teacher frees his students to grow, the more he frees himself to grow.	2017-08-28T03:53:03.136Z	2010-07-12T00:00:00.000	{ "year": 2010, "sha1": "aaf8ddd582886c67b2302827d46a10ebec068f0d", "oa_license": "CCBY", "oa_url": "https://ccsenet.org/journal/index.php/ies/article/download/6793/5324", "oa_status": "HYBRID", "pdf_src": "ScienceParseMerged", "pdf_hash": "aaf8ddd582886c67b2302827d46a10ebec068f0d", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [ "Psychology" ] }
19147419	pes2o/s2orc	v3-fos-license	A corrected quadrature formula and applications A straightforward 3-point quadrature formula of closed type is derived that improves on Simpson's rule. Just using the additional information of the integrand's derivative at the two endpoints we show the error is sixth order in grid spacing. Various error bounds for the quadrature formula are obtained to quantify more precisely the errors. Applications in numerical integration are given. With these error bounds, which are generally better than the usual Peano bounds, the composite formulas can be applied to integrands with lower order derivatives. Introduction In recent years some authors have considered so called perturbed (corrected) quadrature rules. For example, the corrected midpoint and trapezoid quadrature rules are considered in [1] and [2]. As a specific introductory example, consider a correction to the midpoint rule. The classical midpoint rule has the form where R(f ) is the remainder term (error) of this rule. A perturbed (corrected) rule involves the derivative at the endpoints: where R 1 (f ) is the remainder term (error) of this rule. The following properties have analogues in the work we report on Simpson's rule: 1. the original rule (1) is exact for polynomials of degree 1, while the perturbed rule is exact for polynomials of degree 3, 2. estimates of the errors are respectively where \|f ′′ (t)\| ≤ M 2 , t ∈ [a, b] -the bound (4) is better than (3); 3. a corresponding composite quadrature formula, for the corrected rule, has only one additional term, with respect to a composite formula for the original rule, 4. the corrected composite formula has a better estimation of error than the original composite formula (a consequence of 2). Indeed another bound is The above properties are valid for similar corrected rules (for example, for the corrected trapezoid rule) [2]. However, we cannot correct all quadrature rules such that all the properties 1-4 hold. In Section 2 we show that the well-known Simpson's rule does have a simple endpoint correction, but that the quadrature weights have to be modified as well, see (8). We highlight some advantages of the corrected rule over the Simpson's rule. In Section 3 various error bounds of this rule are obtained. These error bounds are generally, but not always, better than the usual Peano error bounds. In Section 4 applications in numerical integration are given. An illustrative example demonstrates that the modified rule gives better results than Simpson's rule. Finite differences derive the modified Simpson's rule We modify Simpson's rule for integration. First, we consider integration rules formed over just two consectutive subintervals, each of length h, and derive endpoint modifications. This is analogous to the improvement to (1) achieved by the inclusion of endpoint derivative information in (2). Second, this modified Simpson's rule is straightforwardly summed to apply to an integration over many subintervals. In later sections we rederive these formula with less restrictions on the integrand and with error bounds rather than just leading order estimates. where the error term is, to a leading order estimate, Example: As a simple illustrative example, consider 1 −1 e x dx = e − 1/e = 2.3504 . Simpson's rule estimates the integral as approximately (e + 4 + 1/e)/3 = 2.3621 , in error by about 0.01 . However, the modified Simpson's rule (5) estimates the integral as (6e + 16 + 8/e)/15 = 2.3502 , which has an error about two orders of magnitude smaller. This modification to Simpson's rule can be very effective. Proof: Consider integrating f over two consectutive intervals in a regular grid of points x j with grid spacing h. Identify a = x j−1 , b = x j+1 and hence x j = (a + b)/2 is the midpoint. Following [3, p65], we write the analysis in terms of centred difference and mean operators, δf j = f j+1/2 + f j−1/2 and µf j = (f j+1/2 − f j−1/2 )/2 respectively, and the differentiation operator denoted by ∂. Then the intergral [3, p69] 1 2h x j+1 To derive a three point integration rule with endpoint corrections such as (5), the above right-hand side must be in the form [1 + αδ 2 + µδβ(h∂)]f j for some constant α and some function β(h∂): [1 + αδ 2 ]f j symmetrically involves f j and f j±1 alone; and µδβ(h∂)f j only involves the derivatives of f at the endpoints x j±1 . Thus we rearrange the operator equation as δ = 2 sinh 1 2 h∂ and µ = cosh 1 2 h∂ [3, p65]. Choosing this particular function β(h∂) would generate a rule in the interior together with end point corrections that would give an exact quadrature formula (for all α). However, the infinite derivatives required are not practical. We chose α to generate an accurate rule that only needs to know function values and the end point derivatives. The approach is to expand this function β(h∂) in powers of small h∂ to see Observe as an aside that choosing α = 1/6 eliminates the first derivative term in β leading to the familiar Simpson's rule with an error determined by the neglected parts of β, namely the end-point contributions −µδ(h∂) 3 /180+ · · · . 1 Instead we choose α = 7/30 to eliminate the third derivative term in β: x j+1 Substitute h = (b − a)/2 to reproduce (5) and its leading order error (6). ♠ Corollary 2 Apply (7) to n consecutive pairs of intervals from say a = x 0 to b = x 2n and sum to immediately deduce that for (8) where the error is to leading order See that these simple modifications to Simpson's rule generate an integration method with error O h 6 . Theorem 3 Let f ∈ C k (0, 1) and let γ k , Γ k be real numbers such that where R k (f ) are defined by (12), C k are defined by (15) and B k are defined by (18), for k = 2, 3, 4, 5 . In a similar way we can prove that (21) holds. ♠ The estimations (19) are Peano-like bounds and they are generally (but not always) better than the usual Peano bounds. Namely, we know that T k (x), k = 2, . . . , 6 are Peano kernels. The usual Peano error bounds are ∞ . Thus, in this case, the error bounds given by (19) are better than the Peano error bounds. In fact, they are equal if and only if Γ k = −γ k . This case (Γ k = −γ k ) is very rare in practice. Theoretically, we can derive better error bounds. Let us say something about the last assertion. We can verify that where p j (x) is any polynomial of degree ≤ j . Thus, The above estimations are theoretically better than the corresponding estimations in (19). We now give the above obtained results for an arbitrary interval [a, b]. If we use this bijection then we find that the polynomials P k and Q k on the interval [a, b] have the forms:P 0 (t) = 1 ,Q 0 (t) = 1 ,P 1 (t) = t − (23a + 7b)/30 , Q 1 (t) = t − (7a + 23b)/30 , etc. The polynomials can be also obtained by simple integration. For example,P 2 can be obtained by integratingP 1 and determining an additional constant such that (13) holds. (They are additionally normalized.) We define the functions for k = 2, . . . , 6 . Here we choose a = x i−1 , b = x i+1 and use the notations introduced in Section 2. Using these functions we derive the following results. Theorem 7 Under the assumptions of Corollary 4 suppose that π andT k are given as above. Proof: The proof follows immediately from Theorem 7 and Corollary 5. ♠ Theorem 9 Under the assumptions of Theorem 7 and Corollary 6 we have Proof: The proof follows immediately from Theorem 7 and Corollary 6. require a same amount of calculations. The rule (31) has only one additional term with respect to the rule (33), namely Hence, the amount of calculations is approximately the same for both formulae. (Recall that function evaluations are generally considered the computationally most expensive part of quadrature algorithms.) On the other hand, the rule (31) is exact for polynomials of degree ≤ 5, while the rule (33) is exact for polynomials of degree ≤ 3 . Furthermore, from Theorem 8 we have while the standard estimation for the Simpson's rule is Since, (34) is better than (35), the rule (31) has better approximation properties than Simpson's rule. Thus, we expect that it will give better results in practice (in most cases). Example: here we show errors in estimating Using just two subintervals, h = 1/2, our formula (8) computes I ≈ 0.746795 , whereas with four subintervals, that is h = 1/4, (8) gives I ≈ 0.746824 which is correct to six decimal places. Figure 1 shows our rule converges quickly with decreasing grid spacing h, and is essentially exact to double precision with just 64 subintervals.	2014-10-01T00:00:00.000Z	2003-03-26T00:00:00.000	{ "year": 2003, "sha1": "cae75ffcc17b70b6987d64078c6e3d83d6200b4e", "oa_license": null, "oa_url": "https://journal.austms.org.au/ojs/index.php/ANZIAMJ/article/download/499/369", "oa_status": "BRONZE", "pdf_src": "Arxiv", "pdf_hash": "611d047a446e4a916d37f2459e18183249d2883c", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics" ] }
259038219	pes2o/s2orc	v3-fos-license	Augmented or Mixed Reality Enhanced Head-Mounted Display Navigation for In Vivo Spine Surgery: A Systematic Review of Clinical Outcomes Background: This research paper provides a systematic literature review (SLR) on the current status of augmented-reality head-mounted devices (AR-HMDs) that guide and navigate spine surgeries and pedicle screw placement. Methods: Embase, Scopus, PubMed, Cochrane Library and IEEE Xplore databases were screened for the systematic literature search to collect and statistically analyze live patient clinical, procedural and user experience data. Multi-level Poisson and binominal models were used for analysis. Results: In vivo patient data, only the clinically widely used Gertzbein–Robbins Scale, were published as an outcome in the recent heterogeneous literature. The statistical analysis supports the hypothesis that using AR-HMDs has the same clinical outcomes as using more expensive robot-assisted surgical (RAS) systems. Conclusions: AR-HMD-guided pedicle screw insertion is reaching its technology readiness, providing similar benefits to RAS. Further meta-analysis is expected in the future from higher case-numbered and standardized randomized clinical trials. Introduction The recent medical and surgical advancements of image guided surgery (IGS) and computer integrated surgery (CIS) have reached spine surgery as well [1][2][3][4][5]. Intelligent imaging technologies, such as augmented reality (AR) or mixed reality (MR) enhanced navigation can support various spine surgery interventions, including vertebroplasty, kyphoplasty or tumor surgery. Nonetheless, AR/MR enhanced navigation has been most used in pedicle screw (PS) placement. IGS naturally supports PS placement with intraoperative navigation, which can be further leveraged by enhancing the registration and control of both robotic and AR/MR systems [6]. The collective term of extended reality involves three main dimensions of how digital information can be applied as an addition to the real environment. Virtual reality (VR) encloses all the computer-generated information in a headset for the user, making them unable to interact with the real environment, which is why the solution is mainly aimed at gaming purposes. AR and MR are different technologies in complexity that enable the user to interact with the virtual content without canceling out the physical world through a transparent head-mounted display (HMD). A slight difference between the two is that AR technology overlays the digital content (text, images, etc.) to the world, but MR allows the user to interact with these (such as 3D holographic models [7]). The authors recognized through the literature search that using the two terms in the medical domain is not 100% consistent and is frequently mixed up. Nowadays, still only a few available AR/MR solutions can be found for intra-operative use, mainly not even commercially available systems, just prototypes. However, as the technology is constantly and rapidly advancing, it is becoming clearer that augmented reality and mixed reality navigation can allow safer and more accurate navigation and guidance in the field of spine surgery [8]. Using a system such as xVision by Augmedics (the most widely used and only FDA-approved navigation known by the authors till the publishing date), the guidance is based on pre-operative diagnostic CT or MRI scan images and intra-operative cone-beam CT scans. These examinations can be segmented into 3D models manually, semi-automatically or via artificial intelligence (AI) and deep-learningdriven methods through image computing platforms (e.g., 3D Slicer [9,10]). Those models can be projected as monitor based, microscope based, or HMD AR or MR holograms [7] to the surgical site after the image-patient registration procedure using infrared cameras or electromagnetic tracking. The preference for the see-through AR-HMD navigation is that the holographic model is projected into the user's (surgeon's) field of view, preventing the further disruption of the surgical workflow. This could decrease the procedures' overall time and altogether lower the risk of more prolonged anesthesia ( Figure 1). One of the most popular medical fields where AR and MR are used is spine surgery and pedicle screw placement. PSs are used for stabilizing potential (such as degenerative spinal diseases and spinal stenosis) and existing instabilities (such as post-laminectomy syndrome and pseudoarthrosis) of the spinal column, as well as spinal traumas and fractures, tumor surgeries and spinal deformities (such as kyphoscoliosis) [11,12]. The widely used free-handed (FH) percutaneous PS placement procedure involves the insertion of pedicle screws into the pre-formed bone pathways and fixating the spine segment with metal rods with 2D fluoroscopy images as guidance. Despite the advantages of minimal invasive-percutaneous-PS placement (e.g., smaller incisions, less pain and faster recovery), with the best intent for patient care, a meta-analysis by Staartjes et al. showed a 3.3% need for secondary revision surgery because of the misplaced screw implantation of FH spine fixations. Using new technologies could decrease the incidence of misplacement complications, leading to a reduction in patient morbidity and mortality, reducing the cost of care USD 23,865-32,915per revision surgery [13,14]. The effect of HMD-based navigation on screw implantation accuracy, operating time, overall radiation dose and cost benefits are not yet proven because of the low number and heterogeneous literature discussing these data. A systematic literature review in 2020 by Bursröm et al. gathered and analyzed 28 articles focusing on AR navigation in spine surgery. Still, limited clinical data were presented, and meta-analysis could not be performed [16]. As the technology is rapidly advancing and new FDA (USA Food and Drug Administration) approved AR-capable navigation HMD systems are being used lately, the re-screening of the literature has become necessary. The authors aimed to systematically review the literature for living human studies reporting clinical outcomes (including procedural and user experience data) with AR/MR-HMDs used for navigation in spine surgery. For the period up to 27 November 2020, we relied on the systematic literature review by Bursröm et al. based on their search in PubMed and Web of Science databases. Our focus is on the detailed analysis of studies that have been published since then. Secondarily, we investigated whether these more recent studies allow the meta-analysis of the available data, and whether these confirm or contradict the existing knowledge. Search Strategy This review is following the PRISMA guidelines, but was not registered. The search of the systematic literature was performed for the period between 27 November 2020 and 1 May 2023. The search was designed to identify articles in which AR-or MR-guided (in HMD form) spine surgery was used in humans for navigation purposes. PubMed, Scopus, Embase, Cochrane Library and IEEE Xplore databases were screened. The following keywords were applied: "augmented reality", "mixed reality" and "spine surgery". The exact search terms are presented in the Appendix A. Patients Only studies involving human patients with a minimum sample size of five patients or a minimum of five implanted screws were included. Intervention AR/MR-HMD navigation in PS spine surgery. Comparator Any. Outcome Outcomes comprise any clinical outcome (e.g., recovery rates, length of hospitalization, visual analogue scale, and post-operative follow-up), accuracy data (linear tip error, angular trajectory error, and Gertzbein-Robbins scale), complications, procedural data (operating time and radiation dose) and user experience measured by any standard validated method. Setting Both experimental and non-experimental settings were included. Publication Types Articles written in English were included. The authors excluded research papers describing teleconsultation, telemedicine and educational use of AR and MR technologies, any abstracts, opinions, letters, reviews, SLRs, conference papers and single case reports. Screening Records identified during the search were screened in two steps. First, the hits were screened by their title and abstract by two reviewers (KM and AH) independently. Disagreements were solved by discussions and wherever it was needed, a third researcher was involved. Second, the selected articles were downloaded in full text and screened by applying the eligibility criteria, using the same independent review method. Data Extraction From the included articles, clinical, procedural and accuracy data were collected, analyzed and narrated by one reviewer (KM). Data extraction included the number of patients who went over spine surgery, the number and accuracy of inserted pedicle screws with accuracy data such as the Gertzbein-Robbins scale (GRS), linear tip error (LTE) and angular trajectory error (ATE), operating time (OT) and surgical complications as the occurrence of specific intra-operative and post-operative ones. Demographic data and the proportion of affected patients were analyzed as well. The user experience of surgeons and criticism about HMDs and AR/MR navigation were collected too, highlighting the presence of situation awareness, technical challenges and limitations of the different manufactured HMDs. As secondary data, where those were available, the region of spine operation (collar, thoracic, lumbar or sacral), disease etiology (trauma, compression fracture and tumor) and patient-reported outcomes were all gathered. Analyses A descriptive analysis of included studies was performed. Studies fulfilling our eligibility criteria in the systematic review by Bursröm et al. were analyzed to assess whether those together with the more recent studies were suitable for analysis. For this, multi-level Poisson and multi-level binomial models were used. Statistical analysis and meta-analysis were performed with STATA 17. This SLR followed the preferred reporting items on systematic reviews and meta-analysis (PRISMA) guidelines [17]. Results From the previous systematic literature review by Bursröm et al., one article fulfilled our eligibility criteria [18]. From the new search time period of 27 November 2020 to 1 May 2023, 392 publications were identified. With Mendeley Desktop 1.19.8 software [19] 190 duplicates were removed, and 202 records were screened by title and abstract. During this screening, 176 articles were excluded, as those were not eligible for inclusion criteria. The full-text review was performed on the remaining 26 papers, where 19 publications were excluded, as those did not contain valid clinical outcome data on live patients. Altogether, 7 articles were included in the analyses. Before the PS insertion, the most common to the rarest symptoms were lumbar back pain, radicular pain, weakness of limb, loss of sensation and urinary retention. The preoperative visual analog scale (VAS) and Oswestry disability index (ODI) were measured as an average of VAS 6.7 ± 1.8 and ODI 82.7 ± 6.2. All the papers added information about the operation-performing surgeons, who were trauma/orthopedics specialists or senior (chief) medical doctors. The pedicle screw fixations were performed on the lumbar, thoracic, thoraco-lumbar and Sacral 1 vertebrae. Accuracy and Tracking Remarkable data description heterogeneity was recognized through the literature review. A major reason could be the lack of standardized requirements for data publication in the field of spine surgery or, more precisely, AR-navigated spine surgery. On this basis, no information was identified in the four analyzed articles regarding linear tip error (LTA) and angular trajectory error (ATE). However, the clinically widely used Gertzbein-Robbins (GRS) classification score was presented in all of the screened papers based on intra-operative C-, O-arm or CT scans. GRS has 5 grades from A to E, based on the pedicle cortex breaching of the implanted screws [27]: • A-no breach detected in intrapedicular screw position; • B-screw exceeding the pedicle cortex is maximum 2 mm; • C-screw exceeding the pedicle cortex is 2-4 mm; • D-screw exceeding the pedicle cortex is 4-6 mm; • E-screw exceeding more than 6 mm or outside of the pedicle. Only Grades A and B can be considered satisfactory operation results, as in cases C to E, mild-to-severe neurological symptoms could occur during the post-operative follow-up [28]. Table 2 shows the clinical accuracy data of the reviewed publications. Altogether, 1258 pedicle and cortical screws were implanted into the 272 patients, an average of 4.6 screws/patient with a range of 2-15. The overall weighted average of GRS A and B was calculated as 98.69%. Bhatt et al. discussed 4 and Butler et al. discussed 3 misplaced screws out of their 218 (1.8%) and 606 (0.49%) placed ones through the PS implantation, which were identified, replaced and revised intra-operatively [20]. Where the few GRS C or D PS grade were recognized through the post-operative follow-up (mainly in Lumbar 4 and 5 vertebrae), patients were 100% asymptomatic [22,25]. According to the screened publications, in 6 out of 7 centers (85.7%), the xVision Spine AR navigation system, approved by the U.S. Food and Drug Administration (FDA), was used by Augmedics Ltd. (Chicago, IL, USA) [29]. One center used an MR-based intra-operative three-dimensional image-guided navigation system (MITINS), including HoloLens by Microsoft (Redmond, WA, USA) [30]. For image-patient registration, xVision uses registration clamps and infrared light-reflecting optical markers, while MITNIS is based on electromagnetic tracking and navigation. No significant difference in clinical accuracy was recognized between the two systems. Operation Time, Complications and Outcomes Only Bhatt et al. presented specific data about operation time (OT) as an average of 3.6 ± 1.7 h. They added information on intra-operative blood loss of 224.0 ± 332.5 mL and the use of mean 3 packets of blood transfusion in their 32 patients. The length of hospital stay of their patients was 4.1 ± 1.6 days [20]. Specifically for PS insertion, Butler et al. added information about the average placement time of 3 min and 54 s per screw with a median of 4 min and 8 s (1 min 10 s to 6 min 30 s). They also measured the learning curve through their data collection, differentiating the experience of the first and final 20 patients' procedures, where the mean insertion times were 4 min 1 s and 3 min 52 s per screw (p = 0.48) [24]. None of the 7 articles presented any surgical complications or the need for revision surgeries through the hospital stay or post-operative follow-up (from 2 weeks to 24 months). Clinical symptomatic reduction was noted in all patients through questionnaires by Yahanda et al., as well as significant improvement of VAS (18.4 ± 2.9) and ODI (16.4 ± 2.6) scores by Li et al. [21,23]. Bhatt et al. also added the mean total 3D imaging radiation dose for AR-navigated PS implantation, which was 576.8 ± 368.8 mGycm, and the average fluoroscopy time was 25.7 ± 29.8 s with a mean radiation dose of 0.3 ± 0.4 mGym 2 [20]. Advantages and Limitations Through the review, several comments were identified from the authors, as they expressed their experience with the use of AR-HMDs. Advantages • Bhatt et al. [20] mentioned that the navigation system with AR technology is highly effective in real-world patient-care scenarios, without a significant learning curve needed for using it. He also added that, with just a limited disruption in workflow, AR-HMDs are simple to integrate, and with it, the implantation accuracy is elevated, and the overall radiation dose is decreased through the procedures compared with the FH technique. • Yahanda et al. [23] commented on a similar or superior implantation accuracy of AR-HMDs compared with the most commonly used RASs (e.g., SpineAssist platform by Mazor, ExcelsiusGPS, ROSA or TianJi). He also added that the fluoroscopy time and radiation dose decreased through the surgeries. • Liu et al. [22] added their accuracy and surgical workflow data to highlight the similarities with RASs too (Mazor X, ROSA, TianJi). It was noted as a strong benefit that using AR-HMDs minimizes the attention shift, as the user can simultaneously visualize the operation field and the image guidance too. With this, cognitive and motor task performance are increased. Another comment was that any instrument can be universally navigated with the AR-HMD system, causing only a minimum interruption in the line of sight. Additionally, the technology's cost is not prohibitive to the distribution to patients. • Harel et al. [25] described the collected data on their user experience questionnaire (UEQ, 1-7 numbered scale on 26 clinical usability questions) regarding the xVision system. All the scores were higher than 6 on average in the dimensions of the clarity of navigation display, the fit into the surgical workflow and the reliability of the system. The lowest score they noted was about the HMD ergonomics (5.9-point average). Limitations • Li et al. [21] drew up the cost of AR-HMD systems compared to traditional FH/MIS methods. Additionally, they felt the disturbing defects of soft tissue simulation, the contrast of AR models, and images were limited by bright light and through the use of it, eye strain and visual discomfort may occur on the user, which needs training. For patients, unusually different laying posture was needed on the OR table because of the use of EM trackers and navigation tools. • Liu et al. [22] commented on the limited use of the technology on obese patients (as the image-patient registration marker clamps needed to be fixed on rigid locations were too short, causing four patients to be excluded from the use case study). The disadvantages of the HMDs were described as mechanical and visual discomfort, visual obstruction and sensory overload, and lastly, the prevention of using headlights through the procedures. • Yahanda et al. [23] gave the same notes about the learning curve of visual discomfort and disorientation of HMDs and the difficulties caused by the patient's obesity. Joint Analysis of All Available Studies From the previous SLR by Burström et al., only one article was identified as eligible for our inclusion criteria. Abe et al. published a cohort study on their experience on vertebroplasty guided by the Epson Movero AR-headset in 2013 [18]. Altogether, five osteoporotic vertebra-fractured (Th.12-L3) patients were operated on without any pedicular breach (100% GRS A), having only 2.09°± 1.3°axial and 1.98°± 1.8°sagittal trajectory error of the 10 implanted screws. No complication was identified through their follow-up. As there were no complications described in either of the investigated articles, a metaanalysis could not be performed. For further analysis, a multi-level Poisson model was used for the hypothesis, which measures and explains the incidence rate of "screw error odds" for every single screw insertion [31]. The confidence intervals were calculated with the exact Poisson method [32]. The examination was made in two scenarios, highlighting only GRS A class and GRS A and B rates together, as those are still clinically completely acceptable. Figure 3 contains the results of the analysis. A multi-level binominal model was used in the ideology to measure the "screw error odds" for every single new screw insertion as further attempts for error. The confidence intervals were calculated with the binomial (Clopper-Pearson) method [33]. The same two scenarios were used as given above; Figure 4 shows the results. Discussion As highlighted from this review, standardized studies and reports for AR-HMDs navigated spine surgeries cannot yet be found in great numbers, the examined articles are greatly heterogeneous in describing objective outcomes. However, the authors presented evidence about the benefit of using this system compared to free-handed screw insertion. The reviewed recent studies discussed low case-numbered results, and no exact complication rates could be collected. A further meta-analysis may not be accurately performed and would identify a high level of risk of bias. The authors strongly suggest that future studies and reports on the topic should be planned to contain standardized clinical, accuracy and procedural data for living patient care too. Linear tip errors and angular trajectory errors are measured only in phantom and cadaver studies, but those would include statistically more objective values for in vivo use rather than the clinically used Gertzbein-Robbins Scale. It is also clear that the relevance of such measurements in the clinical field is not intense, as, clinically, the main goal is to reduce pain and stabilize the spinal column or treat the morbidity without causing any neurological complications. We encourage researchers to consider our eligibility criteria (2.2.4 point) in the design of future studies. It would be desirable to develop points to consider for conducting clinical trials and observational studies in AR-HMDs guided spine surgery, specifying basic requirements, such as for study design and end points. Both of this article's multi-level statistical models resulted in approximately the same outcome, without significant differences from each other, showing that using AR-HMDs for spine surgery has only 1.2% GRS C-D-E grade. This value supports the hypothesis that the technology reached a higher clinical readiness level, as it confirms the existing knowledge in this research topic. According to the reviewed articles, the integration of this new technology was easy and time efficient, did not disrupt the clinical workflow, and all the clinical outcomes are similar or better compared to robot-assisted spine surgeries, which might make the AR-HMDs a cost-saving alternative method [15]. Applying to the surgical workflow, the AR-HMD system would not elongate significantly the operation time as well, which would elevate the risk of surgical site infection [34]. For future advancements, some recommendation was also mentioned: HMD-built-in light source, magnification lens and the system's complete integration with RAS. User satisfaction was clear in real-world scenarios; the system increased the pedicle screw placement accuracy and decreased the overall radiation dose needed for screw implantation. However, visual discomfort and eye strain may happen through use, and the use of AR-HMD guidance has limited possibilities for obese patients. Further development should take into account the importance of ergonomics and the comfort of long usability. As seen across various domains, the recent pandemic accelerated the adaption of robotics in telemedicine and surgery as well [35]. One great advancement in the research topic was the FDA clearance of xVision by Augmedics [29]. According to the FDA regulatory clearance, the requirement was a mean position accuracy error of under 2 mm and a mean trajectory error of 2°for the new system based on X-ray imaging. In the investigation, the overall system accuracy, image registration accuracy and tracking accuracy were tested. Technical performance, user need and software validation, electrical safety and electromagnetic compatibility, headset cleaning, disinfection, reusability, and biocompatibility tests also passed [36]. These preferences could stand as industry standards for future surgical-use AR-HMDs development, and moreover, it could spin out to collateral domains, such as smart farming, where the revolution via the internet-of-everything concept has already begun [37], incorporating the sustainability aspects of such innovation programs [38]. The community has already started to align with these requirements (e.g., https://www.sustainablerobotics.org/ accessed on 1 May 2023). Nevertheless, the ethical and regulatory aspects of the technology have to be managed in parallel to the technical advancement [39]. It is crucial to improve the transparency of the regulatory environment of AR/XR and RAS in medicine, streamline the standardization framework and increase the social acceptance, which is currently served by the standard family IEEE 700X (ethicsinaction.ieee.org/p7000 accessed on 1 May 2023), primarily to the IEEE 7000-2021-Model Process for Addressing Ethical Concerns During System Design [40]. Beyond this, the most recent IEEE 7007-2021 Ontologies for Ethically Driven Robotics and Automation Systems can also contribute to the field, which will be applied to the digital surgery domain as well. Conclusions The statistical analysis of the reviewed articles using AR-HMDs guidance on Pedicle Screw insertion in spine surgery showed the occurrence of 1.2% (95%CI: 1-3.5%) non-GRS A and B (clinically unacceptable grades) of all the screw implantations. The benefit of the system is clearly measurable compared to the free-handed implantation technique, yet the heterogeneity of published data prevents further meta-analysis. The authors used multi-level Poisson and binominal models for statistical analysis, and the results strongly support the claims of reviewed articles, that using the AR-HMD is as accurate as using the more expensive robot-assisted surgical systems. Following a standardized methodology for future case studies or randomized clinical trials would help with a low-bias statistical analysis. Further development of the AR-navigated surgical systems is needed based on the experience of end users, aiming specifically for use in operating rooms. Conflicts of Interest: The authors declare no conflict of interest. Abbreviations The following abbreviations are used in this manuscript: Appendix A Search terms and search strategy for the SLR: All papers published until 1 June 2022, with inclusion restriction of publication date after 27 November 2020.	2023-06-03T15:05:14.467Z	2023-05-31T00:00:00.000	{ "year": 2023, "sha1": "d483b3d9a852d9aca859a62f449c4783ac49382c", "oa_license": "CCBY", "oa_url": "https://doi.org/10.3390/jcm12113788", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "02c14720b3cf6562ce7bb79f8aea91eb338020ec", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [] }
53170410	pes2o/s2orc	v3-fos-license	No-Touch Radiofrequency Ablation of VX2 Hepatic Tumors In Vivo in Rabbits: A Proof of Concept Study Objective In a proof of concept study, we compared no-touch radiofrequency ablation (NtRFA) in bipolar mode with conventional direct tumor puncture (DTP) in terms of local tumor control (LTC), peritoneal seeding, and tumorigenic factors, in the rabbit VX2 subcapsular hepatic tumor model. Materials and Methods Sixty-two rabbits with VX2 subcapsular hepatic tumors were divided into three groups according to the procedure: DTP-RFA (n = 25); NtRFA (n = 25); and control (n = 12). Each of the three groups was subdivided into two sets for pathologic analysis (n = 24) or computed tomography (CT) follow-up for 6 weeks after RFA (n = 38). Ultrasonography-guided DTP-RFA and NtRFA were performed nine days after tumor implantation. LTC was defined by either achievement of complete tumor necrosis on histopathology or absence of local tumor progression on follow-up CT and autopsy. Development of peritoneal seeding was also compared among the groups. Serum hepatocyte growth factor (HGF), vascular endothelial growth factor (VEGF) and interleukin-6 (IL-6) were measured via ELISA (Elabscience Biotechnology Co.) after RFA for tumorigenic factor evaluation. Results Regarding LTC, there was a trend in NtRFA (80%, 20/25) toward better ablation than in DTP-RFA (56%, 14/25) (p = 0.069). Complete tumor necrosis was achieved in 54.5% of DTP-RFA (6/11) and 90.9% of NtRFA (10/11). Peritoneal seeding was significantly more common in DTP-RFA (71.4%, 10/14) than in NtRFA (21.4%, 3/14) (p = 0.021) or control (0%). Elevations of HGF, VEGF or IL-6 were not detected in any group. Conclusion No-touch radiofrequency ablation led to lower rates of peritoneal seeding and showed a tendency toward better LTC than DTP-RFA. INTRODUCTION Radiofrequency ablation (RFA) is currently widely used technique for curative treatment for early-stage hepatocellular carcinoma (HCC) (1).The most frequently utilized RFA technique is a monopolar technique with intra-tumorous electrode placement (2).However, the major limitation of RFA using the conventional monopolar direct tumor puncture technique is relatively high local tumor progression reported between 25-53%, which is related to the limited volume of tumor necrosis (3)(4)(5).Therefore, overlapping ablation techniques are frequently used to create sufficient safety margin around the target tumor (6).Also, conventional tumor penetration technique poses potential risks for unwanted tumor seeding with increasing risk of multiple placements of electrodes within the target tumor.Indeed, the achievement of large peritumoral ablation zone while decreasing the risk of intraprocedural tumor cell seeding could be ideal for improving the therapeutic efficacy of RFA for liver malignancies.In that sense, a no-touch technique with multi-bipolar techniques has been suggested to ensure maximum ablative area which creates high-density electrical fields between several pairs of independent electrodes (2,5,7,8).Although there have been sporadic studies which showed promising results of no-touch RFA technique in their retrospective review, only one recent retrospective study (2) compared no-touch RFA with conventional direct tumor puncture technique. Recently, RFA has been recognized to induce cytokines such as interleukin-6 (IL-6), hepatocyte growth factor (HGF) and vascular endothelial growth factor (VEGF) which promote liver regeneration but potentially facilitating unwanted tumor recurrence or distant metastasis (9)(10)(11)(12)(13).The cytokines are termed as tumorigenic ２ factors and may be produced in the residual incompletely treated tumor at the ablative margin and in the healthy liver surrounding the targeted tumor (12,13). However, there is no study to clarify whether the violation of tumor during the procedure influences the tumorigenic factor release. One critical application of no-touch RFA using a bipolar mode that has not been explored so far is whether no-touch RFA can provide less development of peritoneal seeding and whether it can show lower tumorigenic factor release compared with conventional monopolar tumor puncture RFA.If these were confirmed, no-touch bipolar RFA could become an attractive alternative to conventional monopolar RFA with direct tumor puncture, especially peripherally located liver malignancies.Thus, in this proof of concept study, we attempted to compare no-touch radiofrequency ablation (RFA) technique in a bipolar mode with conventional direct tumor puncture technique regarding local tumor control (LTC), peritoneal seeding and release of tumorigenic factors in rabbit VX2 subcapsular hepatic tumor model. MATERIALS AND METHODS The experimental design is summarized in a flowchart (Figure 1). Animal care This study was approved by the Institutional Animal Care and Use Committee at our hospital.Total of 62 adult New Zealand white rabbits weighing 2.5-3 kg were used.During all procedures including sampling, RFA, and computed tomography (CT) follow up, the rabbits were anesthetized with an intravenous injection of 5mgkg -1 bodyweight tiletamine-zolazepam (Zoletil 50; Virbac, Carros, France). VX2 Liver Implantation In addition to 62 rabbits with the hepatic VX2 tumor for the study, 10 rabbits were used as VX2 tumor donor.The VX2 tumor is a virus-induced anaplastic squamous cell carcinoma characterized by hypervascularity with rapid growth (14,15) and the rabbit VX2 tumor model has been widely employed in numerous RFAbased studies for the treatment of HCC (16)(17)(18)(19).The VX2 tumor was transplanted into the hind thigh muscle of donor rabbit and left in situ for 4-6 weeks.When the tumor had grown to more than 5cm, the tumor was harvested and was cut into 4mm 3 tumor chips.After the tumor chips preparation, midline subxiphoid laparotomy was performed in a total of 62 rabbits for liver tumor implantation.The tumor chip was directly implanted into the subcapsular parenchyma in the left medial lobe of the liver.All rabbits were taken appropriate postoperative care ４ including analgesics and antibiotics. Ex vivo test in Bovine Liver for RFA Optimization using Dual Bipolar We performed an ex vivo test for RFA optimization using the dual bipolar technique for ablation of 1cm size tumor.RFA in a total of 27 bovine liver blocks (size, 5 x 5 x 5 cm 3 ) which were immersed in a 50 x 20 x 25 cm 3 saline-filled bath. The RFA system utilized in the study was a multichannel radiofrequency (RF) generator system (VIVA multi-RF generator: STARmed, Goyang, Korea) and separable dual electrodes (Dual® electrode; STARmed) with two internally cooled electrodes with a 1-cm active tip each.As each electrode of the dual electrode is connected to a flexible, 50-cm-long cable, those electrodes could be placed in the liver with a diverse inter-electrode distance determined by the size of the tumor According to our previous studies where bipolar RFA induced confluent ablation zone within relatively short time than switching monopolar RFA with less thermal damage to the surrounding structures (20)(21)(22), we used bipolar mode for RFA procedure. In vivo Radiofrequency ablation procedure Dual RFA electrodes with a 1-cm active tip (STARmed) were placed percutaneously under ultrasonography (US) guidance.For direct tumor puncture (DTP)-RFA group (n=25), one of two electrodes directly penetrated tumor, whereas the other electrode was inserted at the periphery of the tumor.For notouch RFA group (n=25), both electrodes were inserted at the periphery of the ５ tumor, not violating the tumor.The only difference between two groups regarding RFA procedure was whether one of the electrodes penetrated the tumor or not. After the RFA procedure, cauterization of the electrode track was performed in both groups.Medical illustration of the in vivo RFA procedure is drawn in Figure 2. Radiological study The CT examinations were performed on a multi-detector CT scanner (Discovery CT 750 HD; GE Healthcare, Pewaukee, WI) with 2mL/kg of nonionic contrast medium (Ultravist 370, Bayer, Wayne, NJ, USA) through an auricular vein in the supine position.CT scans included pre, arterial (15 seconds) and portal (30 seconds) phase after contrast injection rate of 1ml/sec with 1mm slice thickness from head to upper thigh.The following CT parameters were used: 150 mA, 140 kVp, 1.0 pitch, and 214 x 214 mm 2 field of view.All 62 rabbits with VX2 tumor implantation underwent CT scans before the RFA (median nine days after implantation (range, 7-12)).In subgroups for pathologic diagnosis (11 rabbits in DTP-RFA and no-touch RFA respectively, two rabbits for control), CT scans were performed after three days post RFA before sacrifice.In subgroups for CT followup (14 rabbits in DTP-RFA and no-touch RFA respectively, ten rabbits for control), CT scans were performed every week until 6 th week after RFA.Local tumor progression was defined as the appearance of nodular, mass-like, or thick irregular tissue with enhancement adjacent to previous RFA treated site (23).Peritoneal and skin seeding was determined as the appearance of enhancing nodular or thick irregular shaped lesion with interval increment attached to the peritoneum or subcutaneous and skin, respectively (24).Euthanasia and autopsy were also ６ performed in rabbits for CT follow up to correlate with the CT data after completion of 6 weeks follow up. Pathologic analysis Euthanasia was performed on rabbits for pathologic examination three days after RFA.The liver was harvested, and multiple slides of 1cm thickness were cut perpendicular to the RFA needle insertion direction.The sliced tissues were embedded in optimal cutting temperature compound (Tissue Tek; Sakura Finetek, Tokyo, Japan), quenched in isopentane and frozen in liquid nitrogen before storage at -80 for nicotinamide adenine dinucleotide (NADH) diaphorase activity evaluation, which reflects viability of the tumor (25).The remnant specimen was fixed in 10% neutral buffered formalin, embedded in paraffin and sliced into 5-um sections for hematoxylin and eosin (H&E) staining for histologic structural evaluation.Pathologic analyses were performed on main tumor nodule and perinodular satellite nodule, respectively.Complete local necrosis was defined as entire involvement of both main tumor and all peri-nodular satellite nodules within the ablative zone on H&E staining with no NADH staining (25).To assess the tumorigenic effect, immunohistochemistry was performed to quantify activated Ki-67 positive hepatocytes by using anti-Ki-67 (Ab155580; Abcam, Cambridge, Mass) in the ablated lobe at high-power (x40) microscopy (10). Local tumor control was determined by either achievement of complete local necrosis on histopathologic examination or absence of local tumor progression (LTP) on follow-up CT and autopsy. ７ Biochemical analysis for tumorigenic factor Serum level of IL-6, HGF and VEGF on the pre, 24h, 48h, and 72h after RFA were measured using enzyme-linked immunosorbent assay (ELISA) by using a rabbit kit (Elabscience Biotechnology Co., Wuhan, China).Untreated rabbits served as a control.A total of 14 rabbits (6 rabbits per no-touch RFA and DTP-RFA, respectively, two rabbits for control) were used for the analysis.ELISA assays were performed according to manufacturer's instruction. Statistical analysis Data were reported as means ± standard deviation (SD), median (range) or number (percentage, %) as appropriate.Comparison between DTP-RFA and notouch RFA was performed using Fisher's exact test or Chi-squared test for categorical variables and unpaired Student-t test or Mann-Whitney U test for continuous variables.For continuous variable comparison among three groups (DTP-RFA, no-touch RFA, control), analysis of variance (ANOVA) was performed.All statistical analyses were performed using commercially available statistical software (SPSS software for Windows, version 21.0, SPSS-IBM).Pvalues of <0.05 were considered statistically significant. Ex vivo test for RFA optimization In the ex vivo test, we aimed to find optimal distance and energy where ablation volume reached more than 4000mm 3 with minimum ablation length of 12mm.To satisfy the desired ablative area, when distance was 10mm or 13mm, both 0.5 Kcal and 0.6 Kcal were sufficient.In case of 15mm distance, 0.6 Kcal was required.We applied either 10mm or 13mm distance with 0.5 Kcal for in vivo RFA experiment. Supplementary material 1 summarizes ex vivo test result. VX2 carcinoma and In vivo RFA procedure The intrahepatic VX2 tumor was grown for median nine days (range, 7-12), which reached the size of median 8mm in axial longest diameter (range, 5-12) and volume of median 315 mm 3 (range, 75-960) on pre-RFA CT.On pre-RFA CT, there was no significant difference regarding longest mean tumor size between DTP-RFA and no-touch RFA group (DTP RFA,7.9mm;No-touch RFA, 8.3mm; P=0.604).All 62 rabbits were confirmed to have successful VX2 tumor implantation in left medial lobe subscapular area with no iatrogenic tumor seeding on pre-RFA CT. We applied the result from the ex vivo test for the in vivo RFA procedure and specification as follows; power, 50 watts; the distance between two RFA needle, 10-13mm; energy, 0.51 Kcal (standard deviation, SD, ± 0.07); ablation time, 307.34 seconds (SD, ± 110.34).The number of punctures per procedure between ９ two groups differed with no-touch group requiring more punctures (DTP-RFA; 2 (median) (range: 2-3), no-touch RFA; 3 (2-5), P<0.001).On post three day post RFA CT, there was no significant difference regarding ablation zone seen as low attenuation in the liver between DTP-RFA and no-touch RFA group (DTP RFA, 7045 mm3; No-touch RFA, 7789 mm3; P=0.517).The tumor characteristics and in vivo RFA specifications are summarized in Table 1. Local tumor control Table 2 summarizes the result of the local tumor control in DTP-RFA, and notouch RFA in both pathologic analysis and 6-week post-RFA CT follow up.There was a tendency of a better result in no-touch RFA than that of DTP-RFA (DTP-RFA 56% (14/25) vs. no-touch RFA 80% (20/25), P=0.069).Regarding complete local necrosis (including main tumor and satellite nodules) on pathologic assessment, DTP-RFA reached 54.5% (6 out of 11) whereas no-touch RFA achieved 90.9% (10 out of 11) (P=0.148).Furthermore, in three of the five rabbits with viable satellite nodules in DTP-RFA group, viable satellite nodules were located more than 5mm distance from the main tumor (Fig. 3 CT follow-up for peritoneal seeding Table 3 summarizes the results of 6-week post-RFA CT follow-up in DTP-RFA, no-touch RFA and control groups.Regarding peritoneal seeding, DTP-RFA had significantly higher incidence than that of no-touch RFA (71.4% vs. 21.4%,P=0.021).Control group showed no evidence of peritoneal seeding or skin seeding. Lymph node metastasis and lung metastasis varied among three groups without statistical significance.Representative cases are presented in figure 4 and figure 5. Tumorigenic factor Serum level of HGF, VEGF, and Il-6 was not detectable due to a minimal increase in all three groups (DTP-RFA, no touch RFA and control).Additionally, measurement of the Ki-67 proliferation marker (at three days after ablation) in ablated lobe per high-power microscopy frame did not differ among three groups (DTP-RFA;7.5±3.0,no-touch RFA;8.1±4.1, control; 3.0±1.6,P=0.494) (supplementary material 3). DISCUSSION In the current study, we proved that no-touch RFA developed less peritoneal tumor seeding than DTP-RFA in the subcapsular VX2 tumors of the liver in rabbits. No-touch RFA showed the tendency of better local tumor control than DTP-RFA (DTP-RFA 56% (14/25) vs. no-touch RFA 80% (20/25), P=0.069) based on either pathologic assessment or combination of contrast-enhanced CT and autopsy findings.Also, no difference was found regarding the tumorigenic factor elevation between DTP-RFA and no-touch RFA.Considering these results of our study, notouch RFA may provide better clinical outcome for bridge therapy to liver transplantation for subcapsular HCC when comparing to dtp RFA. In fact, peritoneal tumor seeding is one of the most unfavorable complications after RFA for liver malignancies such as hepatocellular carcinoma or colorectal liver metastases.Many previous studies reported the various incidence of tumor seeding after RFA for HCC ranging from less than 1.5% (26-29) to 4.0% (30), although a high rate of 12.5% has been reported by one center (31).Most of the previous studies dealt with DTP-RFA technique and suggested various risk factors such as subcapsular location (30,31), a prior biopsy (27,29), poorly differentiated tumors (31) and no cauterization of the electrode track (28,29).Possible mechanisms for peritoneal tumor seeding after DTP-RFA include facilitation of viable cancer cell dissemination by increased intratumoral pressure (24,26), direct tumor implantation through a needle (28), or direct migration of tumor cells with a little bleeding into the peritoneal cavity (26,28).As no-touch RFA alternately places probes at the surrounding liver parenchyma to the tumor margin, there must １２ be reduced risk of direct tumor implantation through the needle or along the bleeding from the tumor.Also, no-touch ablation can induce vessel coagulation around the tumors first, and therefore, we may expect that it causes relatively less intratumoral pressure than DTP-RFA (2) leading to lower peritoneal seeding. Accordingly, recent clinical studies of no-touch RFA reported no peritoneal seeding (2,5,32), although most of them are retrospective studies with small sample size. The incidence of peritoneal seeding in our study was higher (71.4% in DTP-RFA, 21.4% in no-touch RFA) than previous studies on human (1.5% ~ 12.5%).We thought that it might attribute to the subcapsular location of the tumor and the virulence of VX2 carcinoma.Although there is a controversy over the risk of tumor seeding in RFA for subcapsular tumor, there are several studies proving the higher risk of tumor seeding in subcapsular tumors than that of the non-subcapsular tumors (30,31,33).Also, VX2 carcinoma is initially developed from a virusinduced anaplastic-squamous cell carcinoma and has been propagated by numerous serial transplantations into different tissues over the years resulting in high virulence and possible sarcomatous change.Given the innate different tumor characteristics of VX2 carcinoma in a rabbit model and those of HCC in human, caution is warranted when interpreting the incidence of peritoneal seeding between studies on rabbits and human. No-touch RFA is known to improve the rate of complete necrosis resulting from larger ablation volume with ablation forming centripetal direction inducing safety ablative margins.Seror et al. (7) reported enhanced complete necrosis in no-touch RFA (26 out of 29, 89.6%) than in DTP-RFA (14 out of 30, 46.6%) in patients with １３ HCC.In our experiment, we found a similar trend with no-touch RFA for the better local tumor control rate than DTP-RFA but failed to reach statistical significance( P=0.069).This was probably attributed to the biological features of VX2 tumors developing multiple tiny satellite nodules around the main implanted tumor when they reach more than 2 cm in diameter.On follow-up CT and autopsy, despite complete necrosis of the target tumor, local tumor progression was found from satellite nodules which might be related to the biological aggressiveness of VX2 tumors having an infiltrative growth. At the same time, of interesting note is that in three rabbits having viable satellite nodules of DTP-RFA group, the satellite nodules were located more than 5mm from the main tumor and one of them developed intravascular tumor emboli.On the contrary, no-touch RFA and control group had satellite nodules within the 2mm distance from the main tumor.This may reflect the facilitated local dissemination and subsequent trend toward poor local control rate DTP-RFA resulting from intratumoral placement of electrodes, intratumoral pressure increase and centrifugal ablation with late vascular perfusion blockage during ablation, recapitulating previous observations that manipulation of the tumor during surgery aggravates postoperative recurrence and distant metastasis (34-36) and that early vascular flow control before the manipulation of tumor improves the survival outcome (37).Further study with larger study volume is needed to clarify whether no-touch RFA can provide a better efficacy of complete local necrosis compared with conventional DTP-RFA. It is interesting to note that in our study, the number of total punctures per RFA procedure between two techniques differed significantly in our experiment (DTP-RFA; 2 (median) (range: 2-3), no-touch RFA; 3 (2-5), P<0.001).Also, of interesting observation is that all of three cases with peritoneal seeding in no-touch RFA in our study required more than four punctures due to the repositioning of unsatisfactory initial electrode insertions without intervening RF heating (38).We assume during the repositioning, peri-nodular satellite tumor nodules may have been manipulated resulting in tumor seeding facilitation in the three cases.Albeit with the technical and experimental difficulty, it is noteworthy that statistical difference was still found between no-touch RFA and DTP-RFA.Nonetheless, notouch RFA is assumed to be a more demanding technique that requires longer learning curve than DTP-RFA (2).Considering that radiologists are accustomed to placing electrode to the central portion of the tumor when an ultrasound-guided biopsy or ablation procedure, it might be expected that there might be a longer learning curve for no-touch RFA placing electrodes in the peritumoral zone with favorable geometry. Recently, there are increasing experimental evidence that hepatic RFA contribute to the release of the tumorigenic factors which stimulate tumor development, growth, or more aggressive biology at separate sites within the same organ or in the distant tumor.However, in our study, there was no increase of tumorigenic factors including IL-6, HGF and VEGF and no different Ki-67 proliferation index in both RFA groups (no-touch RFA and DTP-RFA) and control group.We contribute the contrary results in our study to the study design on rabbit model, considering previous studies have been conducted on the small animal model (mouse (12,13) and rat (10,11,39)). There are several limitations to our study to be acknowledged.First, there was technical challenge performing US-guided RFA of VX2 tumor in rabbit liver due to the anatomy of rabbit liver which has five lobes of racemose type, and small size of the tumor.Nevertheless, we strived to mimic real clinical practice and found a statistical difference between DTP-RFA and no-touch RFA regarding the development of peritoneal seeding.Second, another limitation would be a relatively small sample size.Thus, additional study in an expanded number and clinical scenarios is required to acquire high statistical significance.Finally, as noted earlier, using a small animal model with VX2 tumor could be a limitation in extrapolating our data to a clinical setting, as the biological behavior of VX2 tumors may differ from that of human hepatocellular carcinoma.Nonetheless, we believe this study presents a reasonable preclinical proof to show the improved safety with treatment efficiency of no-touch RFA in subcapsular hepatic tumors. CONCLUSION In conclusion, no-touch RFA has lower peritoneal seeding rate and tendency to induce better local tumor control compared to those of DTP-RFA which may contribute to better treatment in a subcapsular hepatic tumor and serve as an attractive bridge therapy to liver transplantation (b), supplementary materials 2 and 4) whereas all of satellite nodules in both No-touch RFA (n=1) and control group (n=2) were within 2mm distance from the main tumor.Also, intravascular tumor emboli were noted in one of the rabbits with the satellite nodule in DTP-RFA.(Supplementary material 4).On CT follow up and autopsy, DTP-RFA showed local tumor progression in 42.8% (6 out of 14) while no-touch RFA showed local tumor progression in 28.6% (4 out of 14) LTP, respectively (P=0.694). Fig 2 . Fig 2. Illustration of DTP-RFA and no-touch RFA Ex Vivo test in Bovine Liver for RFA optimization using Dual Bipolar. ３２Supplementary material 1.	2018-11-15T16:40:35.277Z	2018-10-18T00:00:00.000	{ "year": 2018, "sha1": "c63244c8c7eeadb4192cb3df283ee154102f7187", "oa_license": "CCBYNC", "oa_url": "https://europepmc.org/articles/pmc6201983?pdf=render", "oa_status": "GREEN", "pdf_src": "ScienceParseMerged", "pdf_hash": "c63244c8c7eeadb4192cb3df283ee154102f7187", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
212681214	pes2o/s2orc	v3-fos-license	Short-term effects of carbohydrates differing in glycemic index (GI) consumed at lunch on children’s cognitive function in a randomized crossover study Background Intervention studies suggest an influence of breakfast dietary glycemic index (GI) on children’s cognition. The Cognition Intervention Study Dortmund-GI-I study examined whether lunch dietary GI might have short-term effects on selected cognitive parameters. Methods A randomized crossover study was performed at a comprehensive school on 2 test days. One hundred and eighty-nine participants (5th and 6th grade) were randomly assigned to one of the two sequences, medium-high GI (m-hGI) or high-medium GI (h-mGI), following block randomization. In the first period, one group received a dish containing hGI rice (GI: 86) ad libitum, the other mGI rice (GI: 62)—1 week later, in the second period, vice versa. Tonic alertness, task switching, and working memory updating were tested with a computerized test battery 45 min after beginning of lunch break. Treatment effects were estimated using the t test for normally distributed data or the Wilcoxon rank-sum test for non-normally distributed data. Results The crossover approach revealed no effects of lunch dietary GI on the tested cognitive parameters in the early afternoon. However, we determined carryover effects for two parameters, and therefore analyzed only data of the first period. The reaction time of the two-back task (working memory updating) was faster (p = 0.001) and the count of commission errors in the alertness task was lower (p = 0.04) in the hGI group. Conclusion No evidence of short-term effects of lunch dietary GI on cognition of schoolchildren was found. Potential positive effects on single parameters of working memory updating and tonic alertness favoring hGI rice need to be verified. Introduction Children attending all-day schools are particularly challenged to maintain cognitive performance until the afternoon. A proper lunch might help to achieve this. However, it remains to be established which lunch composition is most suitable. Glucose is the main energy source of the brain and its continuous supply is crucial [1]. Although glucose has short-term positive effects on cognitive performance [2,3], high consumption of carbohydrates leading to fast and high increases in postprandial blood glucose might restrain cognitive potentials [4]. Instead, carbohydrate-rich foods causing a slower and prolonged rise of blood glucose were shown to improve attention and memory compared with carbohydraterich foods causing postprandial peaks [5][6][7]. The glycemic index (GI) ranks available carbohydrates provided by carbohydrate-rich foods by their effects on postprandial blood glucose concentrations. Consumption of breakfast with lowdietary GI resulted in better cognitive function in children than skipping breakfast or consuming breakfast with high-dietary GI [8][9][10]. However, a repeated-measures study did not report effects on cognition by breakfasts differing in their dietary GI [11]. Possible explanations are inconsistent methodologies and confounding factors of the studies [12]. To the best of our knowledge, studies on lunch dietary GI and short-term cognitive effects in children are not available up to now. Within the Cognition Intervention Study Dortmund (CogniDo) series, we observed no negative short-term effect of having or skipping lunch on the cognitive performance of schoolchildren, contrasting to studies among adults showing postprandial fatigue after having lunch [13][14][15][16][17]. In fact, for working memory updating, having lunch might be temporarily beneficial for children [16]. Since lunch composition was not considered in these interventions, the aim of this study was to elucidate short-term effects of lunch dietary GI on the cognitive performance of schoolchildren. Study design and participants In accordance with previous CogniDo studies [15][16][17], this study was designed as a randomized, single blind 2 × 2 crossover intervention. The all-day "Comprehensive School Berger Feld" in Gelsenkirchen, Germany was chosen as venue, for a duration of 10 weeks (April-mid-June 2016). Participants were recruited from the 5th and 6th grade (in total 12 classes). The students from the 6th grade had already participated in the previous CogniDo study (as 5th grade students). Children with a metabolic disorder, epilepsy or on a special diet were excluded. Children diagnosed with a learning disorder or insufficient knowledge of the German language (reported by the class teacher) were excluded post hoc from the analyses. Each study arm consisted of a sequence of two treatments. In one arm, participants received a lunch with medium GI (mGI) rice in period 1 and 1 week later, in period 2, a lunch with high GI (hGI) rice (sequence m-hGI). In the other arm, the sequence was reversed (sequence h-mGI). Participants were randomly assigned to one of the two sequences following block randomization per class. Block sizes ranged from two to four participants. For allocation, a computer-generated list of random numbers was used. The 2 test days per individual were scheduled on the same weekday 1 week apart (except one class with 3 weeks in between due to lesson cancellation at short notice). Lunch consisted of rice with ground beef sauce. Beforehand, commonly available rice and pasta products were tested by our staff for sensory properties and with regard to the acceptance by children. Hereupon, the GI of the four most promising products (two types pasta and rice, respectively) were analyzed by Sidney University's Glycemic Index Research Service (SUGiRS), a certified laboratory for GI testing (ISO 26642:2010). Briefly, ten healthy adult volunteers received glucose and four test foods containing 50 g of digestible carbohydrates on different occasions after overnight fasting. Capillary blood samples were obtained for glucose measurements before and at regular intervals during 2 h after ingestion. Using the incremental area under each 2-h plasma glucose response curve (iAUC), GI values for each test food relative to the reference were calculated. Since the two rice types (62 vs. 86) differed more in their GI than the two pasta types (48 vs. 54), rice was chosen for intervention. Study schedule The study design was integrated in the regular school routine. Each test day started at 09:15 a.m. with a standardized breakfast for each participant (bread from wholemeal flour, margarine, salami or Gouda cheese, and carrot sticks) ad libitum. At 12:25 p.m., the start of the regular lunch break, subjects received either lunch with mGI or hGI rice and ground beef sauce (Fig. 1). The amount of consumed rice and sauce was individually assessed by weighing plates before and after the meal. The estimated meal glycemic load (GL) of the consumed rice portion was obtained by multiplying the amount of rice-carbohydrates consumed by the GI of the respective rice (GL = GI × carbohydrate content (g) per portion/100). Carbohydrates provided by the sauce were ignored. Water was available at any time. After finishing lunch and a short break, cognitive assessment started at 1:15 p.m. in the school's computer room. Between the morning and the lunch break participants were asked to refrain from eating and drinking (except for water/unsweetened tea). The children were supervised during the breaks by the study staff. In addition, the participants completed a questionnaire on their food and beverage consumption on each intervention day. Cognitive assessment Cognitive functions were assessed using a computerized test battery consisting of three tasks designed by the Institute for Work, Learning and Ageing (ALA Institute) in Bochum, Germany as described before [17]. Briefly, each task was explained by the study personal and the participants had the chance to practice once in advance in a training mode. After a 5 min break with low-physical activity, the actual cognitive testing started. Subjects were asked to respond as quickly as possible without neglecting accuracy. The cognition tasks were applied in consistent order: task switching, working memory updating, and tonic alertness. Task switching Visual attention and task switching were measured using an alternative version of the Trail Making Task in three sections: the first two sections (numbers and letters) in a nonswitch condition, and the third section (letters and numbers) in a switch condition (Fig. 2a). In the first section, black numbers from 1 to 26 in white squares were presented in an irregular order on a computer Fig. 2 Computerized cognitive tasks. a Visual attention and task switching measured by switch task. The task comprised of three sections. 1 First section, numbers (nonswitch). Numbers had to be clicks in ascending order with the mouse curser. 2 Second section, letters (nonswitch). Letters from A to Z had to be clicked alphabetically. 3 Third section, number and letters (switch). Numbers and letters had to be clicked alternately in ascending order (i.e., 1- Working memory updating measured by two-back task. Fruits and vegetables were displayed on a computer screen. A predefined key had to be pressed when the current image was the same as the image two trials back. c Tonic alertness measured by alertness task. A predefined key had to be pressed as soon as a white circle appeared on the screen. Appearance of a white cross required no reaction. Fig. 1 Study schedule of the cross over study. Each test day started at 9:15 a.m. with a standardized breakfast; on the first test day at 12:25 p.m. participants from group A received high GI rice, group Breceived medium GI rice (period 1); on the second test day vice versa (period 2); cognitive assessment was performed between 1:15 p.m. and 2 p.m. screen ( Fig. 2a-1). Numbers had to be clicked in ascending order. When numbers were clicked correctly the squares turned green, otherwise red. Correctly processed squares faded out. The maximum time to finish was 3 min. The second section used letters from A to Z instead of numbers ( Fig. 2a-2). In the final section, the 26 squares contained numbers from 1 to 13 and letters from A to M (Fig. 2a-3). Participants had to alternately click numbers and letters in ascending order (i.e. 1-A-2-B-3-C). Outcomes included total reaction time (RT) for numbers (items 2-26), total RT for letters (items 2-13), and switch costs, i.e. the processing time of the third section minus the first section minus the difference between the first 12 items of the second section and the first 12 items of the first section. Negative switch costs were regarded as implausible and excluded. Working memory updating (two-back task) To assess capability in holding and manipulating information for a short time, the n-back task was used in a two-back condition. Subjects were asked to monitor a sequence of 106 consecutive pictures of fruits and vegetables presented in the middle of the screen (Fig. 2b). When the current picture matched the picture presented two trials earlier (n − 2), participants had to press a key. The stimuli were presented for 500 ms (interstimulus interval: 2100 ms, maximal RT: 1400 ms). No feedback was given. Twentyone pictures were targets (same picture as two trials before). The outcome variables were: ratio of missings (no reaction while reaction was required), ratio of false alarms (reaction while no reaction was required), and mean RT while reaction was required. Tonic alertness A simple reaction task was used to measure tonic alertness. A white fixation cross was presented on a black screen (Fig. 2c). In a response stimulus interval of 3300 ms (±20%), a circle followed the cross and the subjects had to press a button as soon as the circle appeared (maximal RT 1500 ms). The test included 50 items. The outcome variables were the mean RT (ms), the deviation of RT (ms), the number of omission errors (no reaction after 1500 ms), and the number of commission errors (reaction during the presence of the fixation cross). Statistical analyses All analyses were performed using the statistical software package IBM® SPSS® Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA). p < 0.05 was considered statistically significant. The parameters of the cognitive tasks were used as outcome variables, all interval-scaled. Statistical analysis was performed in accordance with the Consolidated Standards of Reporting Trials statement and Wellek and Blettner [18,19]. The sums of the two individual values of the outcome variables of period 1 and 2 were compared between both arms using an unpaired t test for normally distributed data and the Wilcoxon rank-sum test for nonnormally distributed data to examine potential carryover effects. If no carryover effects were observed, results from both days were considered for the treatment effect. Briefly, individual differences of the particular outcomes of both test days (test day 1-test day 2) were compared between both sequences (hGI-mGI vs. mGI-hGI). In case of carryover effects, only results from day 1 were considered. Associations of GI with cognitive parameters were adjusted for GL using a linear mixed model. In addition, period effects were determined with this model. GL, period, and GI were treated as fixed effects, subjects as random. Participants Out of 343 students in the 5th and 6th grade, 193 students (56%) with informed written consent participated. Of these, four were excluded due to diagnosed learning disorder, one student did not eat lunch on one of the test days (Fig. 3). Thus, intention-to-treat analysis was performed using cognitive performance data of 188 subjects. Characteristics of the 188 included participants are shown in Table 1. Lunch dietary GI and cognition Statistical analyses revealed no significant differences between lunch based on medium or high GI rice for most parameters of the selected cognitive outcomes ( Table 2). For one outcome of the two-back task (RT) and one of the alertness task (count of commission error) carryover effects were detected. Thus, statistical analysis was only performed for period 1. Eating lunch with hGI rice resulted in shorter RT compared with lunch with mGI rice. Similarly, number of errors of the alertness task were lower after lunch with hGI rice. There were period effects for some parameters: switch costs improved in period 2 compared with period 1 in both groups while the RT for visual search numbers slowed down. Within the two-back task and the alertness task, RT and the ratio of false alarms decreased in both groups in period 2 while the ratio of missings increased. Effect adjustment for estimated lunch GL Estimated meal GL differed significantly between both periods, while the amount of lunch consumed did not differ (Table 1). Including the estimated meal GL as covariate in the analysis revealed no significant association with cognitive parameters for either GI or GL (Table 3). Additional analyses for dietary GI and GL effects were performed considering only participants who had fully adhered to the study protocol (no eating and drinking except for lunch and water/unsweetened tea). The outcomes were not different from the results described above (data not shown). Discussion In our previous CogniDo studies, lunch caused no decline in cognitive performance, even indicating small positive effects [15,16]. In the present study, we investigated for the Sequence medium-high GI: participants received lunch with medium GI rice in the first period and high GI rice in the second period; sequence high-medium GI: vice versa, paired t test, mean ± SD, p < 0.05. GI glycemic index. first time short-term cognitive effects of carbohydrate-rich lunch differing in GI. Our results indicate small but significant effects favoring hGI rice with regard to two of the tested cognition parameters: working memory updating (two-back task: RT) and alertness (alertness task: count of commission errors), respectively, although only results from GI glycemic index, GL glycemic load, RT reaction time. a First 12 reactions; switch costs = (mean RT switch task) − (mean RT number task) − (mean RT first 12 reactions of letter task) − mean RT first 12 reactions of number task; analyzed with linear mixed model. period 1 could be considered for this analysis due to carryover effects. A carryover effect is defined as the effect of the treatment from the first period on the response at the second period. Accordingly, rice GI would have a lasting impact on cognition for over a week. Considering, that food digestion takes~6-8 h and the fact that numerous different carbohydrate sources were consumed by the children during the week between the two test periods, a carryover effect in the original sense appears rather implausible. Thus, our results suggesting that hGI rice would improve these two parameters of cognitive performance in period 1 should be interpreted cautiously. It may be possible that the cognitive parameters of the two groups differed from the start, independent of the consumed rice. However, cognitive improvements by a dish based on a high GI food are in line with findings from Micha et al. showing an improved shortterm memory and vigilance in 11-14-year-old children after eating a hGI breakfast [20,21]. On the contrary, Cooper et al. showed that a low-GI breakfast (GI = 48) improves the response time and accuracy of working memory and attention in 12-year-old children across the school morning as compared with hGI breakfast (GI = 72) and breakfast omission, probably by causing lower peak blood glucose concentrations [8]. Similarly, hGI breakfast has been associated with a significant decline in accuracy of attention compared with low-GI breakfast [9]. Apart from the differences, we found for the two cognitive parameters in period 1, none of the other parameters were affected by lunch composition within this crossover study. Some studies investigating the influence of GI in breakfast found positive cognitive effects for either low or hGI. This inconsistency compared with our study could be due to the prolonged fasting before breakfast. In our study, all participants were offered a standardized breakfast to ensure the same conditions for the cognition tasks. Consequently, the fasting phase took merely 3 h. Blood glucose concentrations usually start to increase 15 min post prandial, reaching peak concentrations after 30-60 min and return to baseline (or below) within 3-4 h. In addition, even though breakfast was standardized with regard to the food ingredients, the amount of breakfast consumed was not controlled. This might have influenced the outcome after lunch. Moreover, it has to be considered that rice was served with ground beef sauce, which makes the prediction of the glycemic response difficult. Principally, the simultaneous intake of rice with meat and fat leads to a prolonged digestion and a reduced glycemic response [1,22]. Thus, blood glucose concentrations may not have differed sufficiently between our intervention conditions. Measuring glucose responses with continuous glucose monitoring would have been helpful to differentiate between both test meals but was impossible in healthy children. The timeframe between eating and cognitive testing might be of importance as well. Others reported significant effects of a low-GI breakfast 2-3 h after ingestion on the grounds of lower but prolonged blood glucose concentrations that prevent a decline in cognitive performance later in the morning [6,8,9]. The timeframe in our study was only 45 min. Possibly, differences would have been visible later in the afternoon. In addition to the GI, we considered meal sizes and estimated the GL of the consumed rice portion. Although GL differed significantly between groups, adjusting the effects of dietary GI for GL showed no association with the cognitive performance. This is in line with findings from Brindal et al. reporting no changes in speed of processing, working memory, short-term memory, perceptual speed, and inspection time over a time course of 3 h in 10-12-yearold children after eating breakfasts with different GLs [23]. On the contrary, high-GL breakfasts have been shown to improve working memory and speed of information processing 90 min after breakfast [20], further highlighting the discrepancies between the studies investigating the effect of GI and GL on cognition. Finally, it must be mentioned that we discovered period effects within our crossover design. Independent of the rice GI, children showed either improved or impaired results for some cognitive parameters on the second test day, may be due to learning effects. Learning effects appear independently of carryover effects, which in turn would be a metabolically based cognitive consequence of the dietary intervention lasting until the second cognitive test. On the other hand, it is quite challenging to motivate children throughout the testing, especially when a task proceeds rather monotonously. A strength of the present study is that it was not performed under clinical conditions but tested schoolchildren in their everyday school environment. Thus, our results indicate effects of carbohydrates differing in GI consumed at lunch under real-life conditions. Considering large interindividual differences in cognition, the crossover design is another strength of our study, as every subject acts as his/her own control. In addition, the GI of the rice used for our dietary intervention was assessed in a certified lab according to ISO standards. However, our approach under real-life conditions is vulnerable to confounding factors. Schoolchildren and their peers tend to distract or influence each other when a whole class is tested simultaneously. Although, the children were supervised during the cognitive testing to assure they stay focused, this could be one explanation for the deterioration in some cognitive parameters. Another limitation is that the GI difference between the rice types may not have been sufficient. Since our aim was to examine our question within an everyday school life, we chose a lunch from which we know that children like it. Measuring glucose responses with continuous glucose monitoring was not possible. In conclusion, the present study provides no clear evidence that the dietary GI of lunch influences the short-term cognitive performance of schoolchildren 45 min after beginning of the lunch break. Potential beneficial effects of lunch based on hGI rice on working memory updating and alertness warrant attention. It is also of interest whether extending time between lunch and testing might influence the cognitive performance. Funding This study was supported by a grant from the Uniscientia Foundation, Vaduz, Liechtenstein. Uniscientia was not involved in the design, analysis or writing of this paper. Open Access funding enabled and organized by Projekt DEAL. Author contributions KJ and AD analyzed the data and wrote the paper. JT conducted the intervention. MF contributed the cognition analysis tool. HR analyzed the data. LL and MK conceived the study design. AEB, MG, and TL contributed to data interpretation. All authors reviewed and revised the paper and approved the final version. Compliance with ethical standards Conflict of interest AEB is a member of the ILSI Europe Carbohydrate Task Force and a member of the International Carbohydrate Quality Consortium (ICQC). None of the authors have any personal or financial conflicts of interest. Ethics The study was approved by the Ethics Committee of the University of Bonn and registered at clinicaltrials.gov (NCT02763371). All assessments were made in adherence to the Declaration of Helsinki. Informed conset Written informed consent was obtained from all participants and their parents or legal guardians. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.	2020-03-13T14:41:56.673Z	2020-03-12T00:00:00.000	{ "year": 2020, "sha1": "3f00cccc4609b4c69c525756f588c4e8602d4532", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41430-020-0600-0.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "3194c17170d940503a6c8671492dfb3c012f6c55", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
119107996	pes2o/s2orc	v3-fos-license	Exchange interaction and its tuning in magnetic binary chalcogenides Using a first-principles Green's function approach we study magnetic properties of the magnetic binary chalcogenides Bi2Te3, Bi2Se3, and Sb2Te3. The magnetic coupling between transition-metal impurities is long-range, extends beyond a quintuple layer, and decreases with increasing number of d electrons per 3d atom. We find two main mechanisms for the magnetic interaction in these materials: the indirect exchange interaction mediated by free carriers and the indirect interaction between magnetic moments via chalcogen atoms. The calculated Curie temperatures of these systems are in good agreement with available experimental data. Our results provide deep insight into magnetic interactions in magnetic binary chalcogenides and open a way to design new materials for promising applications. Using a first-principles Green's function approach we study magnetic properties of the magnetic binary chalcogenides Bi2Te3, Bi2Se3, and Sb2Te3. The magnetic coupling between transition-metal impurities is long-range, extends beyond a quintuple layer, and decreases with increasing number of d electrons per 3d atom. We find two main mechanisms for the magnetic interaction in these materials: the indirect exchange interaction mediated by free carriers and the indirect interaction between magnetic moments via chalcogen atoms. The calculated Curie temperatures of these systems are in good agreement with available experimental data. Our results provide deep insight into magnetic interactions in magnetic binary chalcogenides and open a way to design new materials for promising applications. Tetradymite chalcogenides, in particular Bi 2 Te 3 , Bi 2 Se 3 , and Sb 2 Te 3 , are of great interest due to their outstanding structural and electronic properties. These compounds consist of repeated blocks of five atomic layers (quintuple layers) separated by a van der Waals gap. The electronic structure features a narrow band gap and strong spin-orbit coupling, which are responsible for the inverted band structure at the Brillouin zone center Γ. Tetradymite chalcogenides are attractive for thermoelectric applications [1] because of their high figure of merit at room temperature. Recently, topologically protected surface states have been observed in all of these chalcogenides, which makes them subject of intense research [2,3]. Especially, these bichalcogenides serve as a basis for new materials with desired properties [4]. This is feasible by stacking of different compounds or specific doping. In particular, doping with magnetic impurities can open new perspectives for spintronics and spin caloritronics applications [5][6][7][8]. Magnetic properties of magnetic chalcogenides can be efficiently described by first-principles methods. One of the first comprehensive studies was carried out by Larson and Lambrecht [27], who investigated the electronic and magnetic properties of bulk Bi 2 Te 3 , Bi 2 Se 3 , and Sb 2 Te 3 doped with 3d transition metal atoms; their results for magnetically doped Bi 2 Se 3 were confirmed by several groups [28,29]. Recently, it was shown that the Dirac surface state of the topological insulator Bi 2 Te 3 survives upon moderate Mn doping of the surface layer, but can lose its topological nontrivial character depending on the magnetization direction [30,31]. However, critical magnetic properties and the exchange interaction in magnetic chalcogenides were not studied in detail on a theoretical ab initio level and, thus, are still under debate. In this work, doping bulk tetradymite chalcogenides with transition metals by means of a first-principles Green's function method, we show that the exchange interaction in these materials can be either long-range ferromagnetic, antiferromagnetic or paramagnetic, depending on the host and the impurity atoms. We identify in particular two main types of magnetic interactions and discuss ways to manipulate the magnetic properties of these systems. The calculations were performed within the density arXiv:1306.6590v1 [cond-mat.mtrl-sci] 27 Jun 2013 functional theory using the local spin density approximation (LSDA) and the generalized gradient approximation (GGA) [32,33]. A self-consistent Green's function method in both relativistic and scalar-relativistic implementations was used to compute electronic and magnetic properties of the magnetic chalcogenides. Substitutional disorder was treated within the coherent potential approximation (CPA; e. g. [34]). Heisenberg exchange constants J ij were obtained using the magnetic force theorem as it is implemented within the multiple-scattering theory [35]. Inclusion of spin-orbit coupling leads to minor changes in the magnetic interaction (about 3-5% with respect to the scalar-relativistic case). Therefore, for the sake of clarity, here we present only exchange constants calculated within the scalar-relativistic approximation. According to the available experimental data [9-20, 24, 25], 3d transition metal impurities in bulk tetradymite chalcogenides substitute typically cation atoms (Bi and Sb) and can supply 1-3 electrons for the bonding. The comparably smaller size of transition metal ions may lead to substantial relaxations of the underlying crystal structure [27]. We did not account for such structural deformations in our CPA calculations but investigated their impact on the magnetic interaction using a supercell approach, and found only minor changes of the exchange constant values. Therefore, the discussion below reports results from CPA calculations. We performed extensive Green's function calculations of Bi 2−x TM x Se 3 , Bi 2−x TM x Te 3 , and Sb 2−x TM x Te 3 (TM = Ti, V, Cr, Mn, Fe, Co, Ni) for the range of concentrations 0 < x < 1.0. The electronic structures of these compounds calculated within the CPA agree for low and medium concentrations (x < 0.3) with those of previous supercell calculations by Larson and Lambrecht [27] (see the Supplementary Material). The self-consistently obtained Green's function was further used to calculate the magnetic exchange constants J ij . Their relevant directions are depicted in Fig. 1 on top of the lattice structure of Sb 2−x TM x Te 3 for clarity, where we distinguish among in-plane (within the Sb or Bi plane) and an out-of-plane coupling. Although experimental data are available for a large concentration range [15,16], we first focus discussion on a representative value of x = 0.2. The results presented in Fig. 2 can be summarized as follows. (i) The effective exchange interaction is reduced with increasing number of d electrons per TM, going from positive to negative values. The strongest ferromagnetic interaction is found between Ti atoms, which is explained by the local density of states (DOS) of the impurities magnetization is 1.0 µ B /atom, indicating a valency of 3+ for each Ti atom. The strongly dispersed DOS at the Fermi energy leads to a large ferromagnetic coupling between the nearest magnetic moments within the Sb plane (Fig. 2). (ii) With increasing number of d electrons per TM atom, the spectral weight at the Fermi level is reduced. This leads to a strong decrease of the exchange interactions in Sb 2−x Mn x Te 3 , in which the majority and minority spin d electrons are well separated in energy and show minor dispersion. In the case of Fe impurities, due to an occupied minority spin d state at the Fermi level, the magnitude of the exchange interaction increases but becomes negative because of the large exchange splitting and the isolated impurity-like character of the occupied d orbitals. For Co and Ni impurities, the valency changes from 3+ to 2+ and 1+, respectively, reducing, thereby, the magnitude of the exchange interaction, which remains negative. The exchange interaction of Sb 2−x Ni x Te 3 , Bi 2−x Ni x Te 3 , and Bi 2−x Ni x Se 3 is very weak and is not discussed here. (iii) For almost all cases, the strongest exchange interaction is found between magnetic moments located in different Sb (Bi) planes but within the same quintuple layer (panel J 01 2 in Figs. 1 and 2). The coupling weakens systematically with the number of d electrons. This interaction occurs via a Te (Se) atom lying between two impurities and is of double exchange type. In addition, the magnitude of J 01 1 is as large as for the in-plane interaction between the nearest magnetic moments, which is an indirect exchange interaction mediated by free carrier sp states [36]. We thus conclude that two different exchange mechanisms, the double exchange interaction via an anion and the indirect exchange coupling via free carriers, determine the magnetic order in the TM doped chalcogenides. (iv) The size of cation atoms is crucial for the exchange interaction. The large size of atoms and, thus, the more spatially extended wave functions, can lead to a strong hybridization with the electronic states of the neighboring atoms. On the one hand, this can increase the number of free carriers within the cation layer, favoring the indirect exchange of Zener type [36]. On the other hand, the strong binding between a cation (e. g. Bi) and an anion (e. g. Te) reduces the number of valence electrons of the anion and, thereby, reduces the strength of the double ex-change interaction. Therefore, in the case of Sb 2 Te 3 , the double exchange interaction via Te atoms is significantly larger than that in Bi 2 Te 3 and Bi 2 Se 3 . (v) Surprisingly the exchange interaction between magnetic moments located in neighboring quintuple layers does not vanish (see J 02 1 and J 02 2 in Figs. 1 and 2). (where ρ ↑ (z) and ρ ↓ (z) stand for the spin-up and -down charge densities, respectively, integrated over the lateral coordinates x and y) "bridges" the van der Waals gap and is responsible for the "interquintuple layer" magnetic interaction (Fig. 3). The spin density in anion layers is negative and has a magnitude comparable with that of the spin density in the van der Waals gap. Considering a wider range of concentrations of the TM atom, we have estimated the critical temperatures T C using a Monte Carlo method [37][38][39]. To treat both ferromagnetic and antiferromagnetic materials, we investigate the spin-spin-correlation function where m i and Ω i are the magnetic moment and the interaction sphere around site i, respectively. We also account for percolation effects, using pair potentials, and compared estimated critical temperatures with the available experimental data. The results for ferromagnetic Sb 2−x TM x Te 3 (TM = Ti, V, Cr, Mn; Fig. 4) show a systematic increase of the T C with the concentration of dopants. Percolation effects do not affect strongly the behavior of T C at low concentrations; except in the case of Ti, for which percolation lowers T C . Calculations for Cr reproduce the experimentally measured trends for concentrations up to x = 0.6 [16]. For higher concentrations, we found a transition to antiferromagnetic order (area with a light red background in Fig. 4), which is understood as the results of an increasing antiferromagnetic interaction between magnetic moments from nearby quintuple layers. This explains why experimental data is unavailable for concentrations larger than x = 0.6. Concerning the Sb 2−x V x Te 3 case, we reproduce the trend found in experiments for a broad concentration range [15]. However, the theoretical absolute values are underestimated by about a factor of 1/2. One could speculate that either structural imperfections due to sample preparations or limitations in the ab initio description could cause this mismatch. For the reported cases of Mn doping at low concentration regimes (x ≤ 0.1), our calculations are, again, in qualitative agreement with experiment [10,14,19,[40][41][42]. The systematic study of exchange interaction discussed for TM-doped magnetic chalcogenides elucidated in this Letter can open new possibilities for a specific design of their magnetic properties. We infer for instance that one way to control the magnetic interaction and the Curie temperature is to replace particular atoms or sheets of Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Te Sb The critical temperature TC is determined from Monte Carlo simulations with randomly distributed impurities (black filled triangles) and with cluster percolation (red filled circles). TC is compared to experimental data (green and blue markers) [15,16]. In the case of Cr, there is a transition from a ferromagnetic state for x < 0.8 (light blue background) to an antiferromagnetic state at higher concentrations (light red background). atoms, in order to tune the strength of the electronic hybridization; the latter is responsible for the exchange interaction mechanisms in this class of materials. Here, it has been shown that the overlap between electronic wave functions of anions and cations is crucial. In an experimental realization, one could replace the anion layer between two cation sheets by atoms of the same group in the periodic table. As an example, we consider the layer between two Sb 1.80 Cr 0.20 sheets in the Sb 1.80 Cr 0.20 Te 3 alternated with various chalcogen atoms. We observe an increase of the associated Curie temperature with ionic size and spatial extension of the electronic wave function of the chalcogen ion: S, Se, Te, and Po yield T C = 69 K, 71 K, 76 K, and 79 K, respectively. A similar effect can be achieved with specific co-doping of the corresponding anion layer. The in-plane exchange interaction can instead be tuned by co-doping of the cation layers in accordance with the exchange constant behavior presented in Fig. 2. To illustrate this aspect, we calculated the exchange interaction in Bi 2−x−y Cr x Sb y Te 3 for x = 0.2 and y = 0.0, 0.2, and 0.4 (detailed results are presented in the Supplementary Material). Here, the Curie temperature increases with Sb concentration from 35 K at y = 0.0 to 39 K at y = 0.2, and to 42 K at y = 0.4. Another way to tune the magnetic interaction is to insert impurities into the van der Waals gap [43]. This can change the size of the van der Waals gap and, by a proper choice of the impurities, can supply free carriers, which are important for the indirect exchange of Zener type (see the Supplementary Material). In summary, we have studied the exchange interaction in the Bi 2 Te 3 , Bi 2 Se 3 , and Sb 2 Te 3 tetradymite chalcogenides doped with transition metals. Our first principles calculations have shown that the magnetic interaction is long-range and is mainly mediated out-of-plane by the double exchange mechanism via an anion and in-plane by the indirect exchange coupling via free carriers. The calculated Curie temperatures as a function of the concentration are found in qualitative agreement with available experimental data. Finally, we presented several ways to tune the magnetic interaction in these systems: (i) replacing the anion layer between two cation sheets by atoms of the same group, (ii) co-doping of the cation sheet, and (iii) inserting impurities in the van der Waals gap. These results provide deep insight into the magnetic interactions in the magnetic binary chalcogenides, and open new ways to design new materials for promising applications. We acknowledge support by the Ministry of Education and Science of Russian Federation (state task No.	2013-06-27T18:12:23.000Z	2013-06-27T00:00:00.000	{ "year": 2014, "sha1": "541865fb287a43ccf6c270bd239c4ab12d47ac59", "oa_license": "CCBY", "oa_url": "https://digital.csic.es/bitstream/10261/136940/1/binary%20chalcogenides.pdf", "oa_status": "GREEN", "pdf_src": "Arxiv", "pdf_hash": "541865fb287a43ccf6c270bd239c4ab12d47ac59", "s2fieldsofstudy": [ "Materials Science", "Chemistry" ], "extfieldsofstudy": [ "Physics" ] }
269572052	pes2o/s2orc	v3-fos-license	OV Modulators of the Paediatric Brain TIME: Current Status, Combination Strategies, Limitations and Future Directions Oncolytic viruses (OVs) are characterised by their preference for infecting and replicating in tumour cells either naturally or after genetic modification, resulting in oncolysis. Furthermore, OVs can elicit both local and systemic anticancer immune responses while specifically infecting and lysing tumour cells. These characteristics render them a promising therapeutic approach for paediatric brain tumours (PBTs). PBTs are frequently marked by a cold tumour immune microenvironment (TIME), which suppresses immunotherapies. Recent preclinical and clinical studies have demonstrated the capability of OVs to induce a proinflammatory immune response, thereby modifying the TIME. In-depth insights into the effect of OVs on different cell types in the TIME may therefore provide a compelling basis for using OVs in combination with other immunotherapy modalities. However, certain limitations persist in our understanding of oncolytic viruses’ ability to regulate the TIME to enhance anti-tumour activity. These limitations primarily stem from the translational limitations of model systems, the difficulties associated with tracking reliable markers of efficacy throughout the course of treatment and the role of pre-existing viral immunity. In this review, we describe the different alterations observed in the TIME in PBTs due to OV treatment, combination therapies of OVs with different immunotherapies and the hurdles limiting the development of effective OV therapies while suggesting future directions based on existing evidence. Introduction Among childhood cancer, tumours of the central nervous system (CNS) remain the leading cause of death [1].According to World Health Organisation (WHO) Classification for Paediatric Brain Tumours (PBTs), the main tumour entities consist of high-grade gliomas (HGGs), low-grade gliomas (LGGs), medulloblastomas (MBs), ependymomas (EPNs), atypical teratoid/rhabdoid tumours (ATRTs) and embryonal tumours with multilayered rosettes (ETMRs), each demonstrating a very distinct molecular background and cell of origin and each being further divided into additional subtypes [2].Overall survival rates (OSRs) of PBTs are highly variable and depend on the type of tumour, grade, location, size and age of the patient and the ranges can vary from <30% at 24 months for HGGs to >80% at 60 months for LGGs [3,4].The current standard of care for these brain tumours comprises a combination of surgical resection (if operable), radiotherapy (RT) and chemotherapy [1].Every PBT type is treated differently depending on the location and grade; however, in all cases, clinicians choose maximal resections when possible, higher doses of RT and chemotherapy, leading to increased survival of the patients [4].However, the increased survival comes at the cost of debilitating long-term side effects such as endocrine dysfunction and a decline in neurocognitive development for patients at a younger age in particular [4,5].The limited success and the high risk of long-term neurological effects are all indicative of the need for more effective and targeted forms of treatment. Cancer immunotherapies have been rapidly developing in terms of efficacy and are becoming crucial in the treatment of tumours due to their specificity to tumour cells while leaving healthy cells intact.These immunotherapies entail all forms of treatment that use the immune system to target malignant cells and have been successfully used in the treatment of multiple hematologic malignancies and solid tumours [6].Examples include antibodies, immune checkpoint inhibitors (ICIs), vaccines and chimeric antigen receptor (CAR)-T cells [6].Considerable obstacles in the use of immunotherapies in PBTs have been the low tumour mutational burden (TMB), the blood-brain barrier (BBB) and the common presence of a 'cold' tumour immune microenvironment (TIME) [7][8][9]. A major obstacle to effective immunotherapy in the paediatric TIME is the presence of a low TMB with a mutation frequency of 14 times lower than in adult tumours and with 0-6 somatic coding point mutations per megabase of targeted territory [8,9].This is a major disadvantage for designing effective immunotherapy as the amount of potentially druggable tumour-specific antigens (neoantigens) is expected to be much lower. The BBB is a physical barrier that greatly limits vesicular transport while allowing only passive transport by small (<400 Da) or lipid-soluble molecules in an effort to protect the brain from toxins and pathogens [10].This barrier in PBTs demonstrates a great heterogeneity in integrity among the different subtypes of tumours while remaining less permissive than adult counterparts, which can greatly reduce the influx of systemically administered therapeutic drugs, their bioavailability and clinical response [11]. A suppressive or 'cold' TIME consists of low expression of immunogenic markers on the tumour cells, low infiltration of proinflammatory lymphocytes, an abundance of regulatory immune cells, and the presence of tumour-associated macrophages/microglia (TAM) that hamper effective immunotherapeutic strategies [3,12].Provided that an immunosuppressive microenvironment in cancer patients correlates with poor prognosis and limits the effectiveness of immunotherapies, strategies to switch from an immunosuppressive to proinflammatory microenvironment may be key for robust and long-term anti-tumour responses required for therapeutic success [13]. One potential strategy is virotherapy where the use of oncolytic viruses (OVs) is harnessed to specifically target and kill tumour cells [14].A limited number of PBT clinical trials using OVs resulted in a number of cases with improved median overall survival (MOS), low-grade adverse events (AEs) and lack of neurological deterioration, thus sparking an interest in their potential therapeutic value [15][16][17][18].In addition, OVs have the innate ability to induce a systemic and local immune response.This ability can be further enhanced through genetic modifications [14].The OV-driven immune response is able to elicit an anti-tumour effect by stimulating the TIME [14].The shift towards a pro-inflammatory TIME by the OVs creates windows of opportunity for effective treatment with other immunotherapeutic modalities.Therefore, coupling OVs with immunotherapies like ICIs, adoptive cell therapies or tumour antigen vaccines may be the future of effective therapeutics in PBTs.Early investigations of these combinations have demonstrated the synergistic effects of OVs with immunotherapies, leading to improved immunomodulation and survival benefits in adult clinical trials [19,20]. Even with the accumulation of positive clinical responses, investigations on the immunomodulatory effects of OVs remain unexplored either due to the limitations of existing pre-clinical models or ineffective monitoring.This review aims to provide an overview of the current knowledge on the modulation of the TIME by OVs in PBTs, modifications employed for an enhanced immunomodulatory effect and combination strategies with existing immunotherapies.Furthermore, we discuss the current barriers to the development and investigation of effective OV therapies and subsequent combinations while providing suggestions to overcome those.Lastly, we provide suggestions on different clinical aspects affecting OV clinical trials in need of optimisation. Paediatric Brain Tumour Immune Microenvironment All PBTs were considered highly immunologically 'cold' in the past, with new insights demonstrating that each entity of a PBT is characterised by a varying immunophenotype that results in heterogenous tumour microenvironments (TME) (Figure 1) [12].More investigations associated the TIME heterogeneity as a predictor of prognosis and survival [21].These differences were not only observed between entities of tumours but also extents among molecular subgroups within one tumour entity [22,23].Briefly, LGGs and ATRTs bear higher levels of infiltration of myeloid cells and lymphocytes than MBs and HGGs [23,24], whereas in HGGs and diffuse midline gliomas (DMG) lymphocyte infiltration is very low compared to other PBTs [25].In this review, the most commonly occurring characteristics spanning the multiple PBT types will be discussed. The lymphocyte compartment of the PBTs is characterised by a low infiltration of B cells, CD4+ helper T cells and CD8+ cytotoxic T cells [25].Both T cell subsets are crucial in assisting and executing an adaptive immune response against tumour cells.Concurrently, regulatory T cells (Tregs) that release and maintain immunosuppressive cytokines like interleukin 10 (IL-10) and transforming growth factor-β (TGF-β) remain more prevalent in the tumour than in the surrounding healthy tissue [12,25].Furthermore, increased levels of TGF-β and increased expression of immune checkpoint molecules like the programmed cell death ligand 1 (PD-L1) and B 7 homolog 3 protein (B7H3) on the surface of the tumour cells, especially on HGGs, result in the inhibition of effector T cells and, subsequently, in further Treg infiltration [25]. In PBTs, natural killer (NK) cells and other innate lymphocytes (NKT and γδ) that recognise target cells independently of human leukocyte antigen (HLA), have hindered effector functions due to the extremely low expression of activating receptors necessary for a NK cell-mediated anti-tumour response.Such receptors are the natural killer group 2D (NKG2D) receptor and UL-16 binding proteins (ULBP) 1 through 6 on NK cells within the TIME of MBs and HGGs [26].One potential mechanism behind this low surface expression is thought to be induced by excessive TGF-β present in the TIME [27]. The myeloid compartment of the paediatric brain TIME is composed of regulatory cell types, such as myeloid-derived suppressor cells (MDSCs) and M2 macrophages (antiinflammatory and tumour supporting) and effector cell types, such as dendritic cells (DCs) and M1 macrophages (pro-inflammatory and tumour inhibiting) [28,29].The balance in the immunosuppressive TIME of PBTs sways more towards higher levels of MDSCs and M2 macrophages than to DCs and M1 macrophages [12]. A unique myeloid cell population of the brain known as microglia, also described as the macrophages of the brain, are multifunctional cells responsible for debris removal, release of neurotrophic and growth factors as well as the resident innate immune cells of the brain [30].Like macrophages, microglia demonstrate an M2 tumour-supporting and anti-inflammatory state and an M1 pro-inflammatory and anti-tumour state [31].Their phenotype seems to be correlated by the tumour entity with EPNs, with ATRTs having microglia with more proinflammatory phenotype while MBs and HGGs have an antiinflammatory phenotype [24]. DCs function as antigen-presenting cells (APCs) as they can capture antigens, originating from tumours or infected cells and effectively "cross-present" them to other immune cells, especially to CD8 T cells [32].However, in PBTs the limited presence of tumourassociated antigens (TAAs) due to the low TMB and the increased release of CD47 often detected in MYC-amplified tumours that reduce phagocytosis lead to low levels of antigen presentation and, consequently, lower anti-tumour immunity [33]. Characteristics of OVs OVs employ naturally occurring or genetically engineered viruses that selectively infect, replicate in and kill tumour cells while healthy cells remain unaffected [14].They originate from different viral families.The most basic oncolytic virus is composed of a core and a capsid.The core of an oncolytic virus comprises either DNA or RNA, which can be either double-stranded (ds) or single-stranded (ss), linear or circular, storing their genetic information whilst concurrently driving characteristics.DNA oncolytic viruses have more efficient replication abilities due to high-fidelity polymerase [34].On the other hand, they are less immunogenic as the hosts are usually equipped with deficient DNAsensing machinery and present a small risk of viral DNA integrating into the host genome, potentially exacerbating tumour mutations (e.g., activating oncogenes) [34].RNA oncolytic viruses have enhanced delivery efficiency and a distinct advantage in targeting CNS tumours [34].This is usually attributed to their size being considerably smaller than DNA viruses, enabling more efficient crossing of the BBB [35].The most extensively studied viruses that have been evaluated in clinical trials of brain cancers are described in Table 1 and Figure 2A [36][37][38].Wild-type and genetically modified OVs recognise overexpressed or tumour-specific surface receptors and abnormal upregulated or downregulated signalling pathways and their related cellular products unique to tumour cells [36].Moreover, the high rate of metabolism prevalent in tumour cells provides an attractive target for efficient viral replication [37].Some common modifications for improved viral entry include (a) replacement of the original viral fibre with a viral fibre of the desired tumour tropism (serotype switching), (b) modification of viral capsid proteins or viral envelope glycoproteins for improved viral entry and (c) insertion of genes encoding a single-chain antibody specific to a known tumour-surface antigen [36].Different strategies to improve tumour specificity based on intracellular tumour features are proposed.Insertion of tumour-specific promotors in the viral genome increased specificity by allowing selective expression of viral genes in tumour cells.Mutation or knock-out of specific viral genes deprives OVs of their ability to replicate/virulence in normal cells.This can be used to achieve selective OV replication in tumour cells that are characterised by loss of, e.g., tumour-suppressive genes (Figure 2B) [36,37]. Mechanism of Action of OVs After successful entry into the tumour cells, the anti-tumour effect of OVs can be divided into two main mechanisms.First, OVs can replicate and lyse tumour cells to release new viral particles into the TME, leading to successive cycles of infection [37,49].Second, tumour cell lysis results in the release of TAAs, viral pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs) as well as chemokines and toll-like receptor (TLR) agonists [49]. PAMPs consist of common motifs expressed by infectious agents (e.g., viral capsids or nucleic acids), whereas DAMPs are usually derived from the host cell [49].Examples of DAMPs include heat shock proteins, high mobility group box 1 (HMGB1) protein, calreticulin and ATP [49].Members of both groups can bind to pattern recognition receptors such as TLRs on macrophages and NK cells and instigate type I interferon (IFN) signalling along with further innate immune responses [49].Furthermore, the release of PAMPs, DAMPs, cytokines and TAAs into the TIME results in the increased recruitment of NK cells and APCs such as macrophages and DCs into the tumour environment [50].The secretion of pro-inflammatory mediators leads to the maturation and activation of APCs [50].The presentation of TAAs by APCs can activate adaptive immune cells either locally or in the lymph nodes, resulting in lymphocyte infiltration of activated CD4+ and CD8+ T cells into the tumour [50].In the tumour, the cytotoxic CD8+ T cells can specifically eliminate tumour cells that express the TAA via major histocompatibility complex (MHC) class I, regardless of whether they were infected with the OV [49,50].Moreover, memory T cells are formed, improving anti-tumour immunity against future tumour challenges.Thereby, OVs can kickstart a broad anti-tumour immune response by providing immunogenic TAAs and inducing increased levels of infiltrating immune cells (Figure 2C) [50]. Preclinical Evidence Preclinical evidence in different PBT models demonstrates the ability of OVs to modulate the brain TIME.Here, we provide a summary of preclinical findings from PBT models (Figure 3, Table 2). Immunocompetent MB mouse models treated intratumorally with C134, a modified HSV-1, demonstrated improved survival coupled with an increased influx of M1 macrophages, DCs, CD4+ and CD8+ T cells and NK T cells through single-cell analysis.Additionally, a significant increase in programmed cell death protein 1 (PD-1) and lymphocyte-activation gene 3 (LAG-3) cell expression was observed on CD4+ and CD8+ T cells as well as in the percentage of LAG3+ cells of both respective T cells, while PD-L1 was increased on the surrounding cells of the TME [51]. Further investigation of the gene expression of these infiltrating cells revealed altered expression of antigen presentation, IFN signalling, cytokine signalling and cytotoxicityrelated genes.More specifically, in lymphocytes (T, B and NK) expression of genes coding for granzyme A, B and natural killer cell granule protein 7 (NKG7) was increased.In monocytes, macrophages and microglia, an increased expression of cytokines, chemokines and MHC I genes was observed, while genes related to IFN-response and chemokines in DCs were increased [51]. A prior study investigating malignant glioma treatment with C134 also demonstrated a significant increase in the number of CD8+ T cells, but not CD4+ T cells in the CSF of both models.Interestingly, in T cell-depleted mice C134 treatment did not increase survival.In addition, re-challenge for 70 days led to inhibition of tumour growth, indicating that a circulating memory immune response had developed in the mice after treatment with C134 [52]. A similar increase in CD4+ and CD8+ T cells post-treatment with an oncolytic AdV named Delta-24-RGD (DNX-2401) within glioma tumours of immune-competent Syrian hamsters was observed [53].The effect of increased CD8+ T cells by Delta-24RGD was also demonstrated in immunocompetent humanised mouse models bearing ATRT tumours, though in this model a reduced presence of macrophages was observed [54].Interestingly, in the same study, the ETMR-bearing mice demonstrated a pronounced activation and recruitment of ionised calcium-binding adapter molecule 1 (Iba-1)-positive cells (a marker for proinflammatory macrophages and microglia) at the tumour margins; however, no active viral infection or T cell infiltration was detected, indicating fast clearance of Delta-24RGD [54]. Preclinical Evidence Preclinical evidence in different PBT models demonstrates the ability of OVs to modulate the brain TIME.Here, we provide a summary of preclinical findings from PBT models (Figure 3, Table 2).Immunocompetent MB mouse models treated intratumorally with C134, a modified HSV-1, demonstrated improved survival coupled with an increased influx of M1 macrophages, DCs, CD4+ and CD8+ T cells and NK T cells through single-cell analysis.Additionally, a significant increase in programmed cell death protein 1 (PD-1) and lymphocyteactivation gene 3 (LAG-3) cell expression was observed on CD4+ and CD8+ T cells as well as in the percentage of LAG3+ cells of both respective T cells, while PD-L1 was increased on the surrounding cells of the TME [51].Regarding the pre-clinical evidence of interactions between OVs and macrophages and microglia, there is an extreme lack of research investigating this in paediatric tumours such as PBTs [55].However, based on what has been studied in adult brain tumours such as glioblastoma (GBM), the role of macrophages and microglia on OV anti-tumour efficacy is multifaceted.Due to their traditional activation towards one of two polarisation states, the M1 pro-inflammatory and the M2 immunosuppressive state, M1 and M2 microglia and macrophages are generally considered to be anti-and pro-tumorigenic, respectively.In line with this, there is evidence that OV therapy can induce microglia and macrophage switching from an M2 to M1 phenotype in the GBM microenvironment [56].As a result, proinflammatory M1 microglia or macrophages can produce soluble factors such as reactive oxygen species (ROS), nitric oxide (NO), tumour necrosis factor-alpha (TNF-α) and IL-1β.These factors have the potential to induce apoptosis, DNA damage or cytotoxicity, resulting in the direct elimination of tumour cells [57].Indirectly, M1 TAMs can also recruit and activate other immune cells such as DCs, NK and T cells [57]. However, simultaneously, M1 pro-inflammatory microglia and macrophages form the primary anti-viral immune response through the production of type I IFNs and phagocytosis of OVs [58].This leads to the accelerated clearance of OVs and thereby a decreased viral infection, replication and lysis of tumour cells.Thus, while M1-like macrophages are expected to boost virus-induced activation of the anti-tumour immune response, they may also facilitate early virus clearance.In contrast, M2-like macrophages, linked to tumour angiogenesis, metastasis and suppression of the anti-tumour immune response may also suppress the anti-viral immune response and support oncolysis [55]. Based on these findings, we cannot determine an exact mechanism for OV-mediated immune modulation that results in a potent anti-tumour response.Nevertheless, it appears that OVs can elicit durable innate and adaptive immune responses across various in vivo models of PBTs, often resulting in extended overall survival, which is lost in the absence of a functional TIME, indicating therapeutic efficacy and high dependency on the immune system. Clinical Evidence There have only been a limited number of clinical trials involving oncolytic viruses in PBTs, some of which are still ongoing and pending results.However, here we review what has clinically been established in paediatric human trials so far, with a focus on OVs reported to be modulating the TIME (Table 3, Figure 2) [60].Increased levels of monocytes RV (Pelareorep) GBM, DMG, MB [17] In a dose-escalation study of DNX-2401 in patients with newly diagnosed DMG, it was found in tumour tissue from patients that immune cell infiltration of CD4+ and CD8+ T cells was scarce while CD11b+ myeloid cells were the most abundant immune population upon diagnosis [15].Tumour tissue from one patient demonstrated increased numbers of CD4+ and CD8+ T cells along with an absence of Treg cells and a reduced number of CD11b+ cells after administration of DNX-2401.While upregulation of gene ontology (GO) terms related to viral processes and immune response were demonstrated in TAMs.However, upon later autopsy of this patient, the infiltration of effector lymphocytes had decreased and was replaced by an increased number of immunosuppressive macrophages [15].The patients from this study demonstrated an MOS of 17.8 months and OSR of 50% at 18 months and their progress-free survival was linked to the T-cell receptor (TCR) clonality in their peripheral blood mononuclear cells [15]. A similar strong increase in infiltrating lymphocytes was seen in a Phase 1 study in children and adolescents with recurrent or progressive HGG after treatment with the genetically engineered HSV G207, leading to a median OSR of 12.2 months [16].More specifically, matched resection tissue from pre-treatment and 2-9 months post-treatment showed a significant increase in CD4+ and CD8+ T cells within the tumour as well as areas adjacent and distant from where G207 was inoculated [16].Interestingly, the tumourinfiltrating lymphocytes were observed at later timepoints, coinciding with the absence or lack of G207 replication, indicating a persistent immune response after OV clearance [16].Furthermore, the presence of B lymphocytes and plasma cells pre-treatment was low, whereas post-treatment, levels of B lymphocytes were increased in one patient and levels of plasma cells in two patients [16].Interestingly, when comparing pre-treatment and 3 months post-treatment tissue biopsies from a patient from the same trial, significantly increased expression of cytotoxic T lymphocyte-associated protein 4 (CTLA-4) and PD-1 checkpoint molecules was observed on the infiltrating CD8+ T cells [61].Moreover, local tumour cells significantly expressed increased amounts of PD-L1 and IDO after treatment with G207 [61].These data suggest that along with OV-induced immune cell infiltration of the TME, there is an increase in immune checkpoint molecules on tumour and immune cells [61]. A significant increase in tumour infiltrating CD8+ T cells was also observed in one patient when comparing pre-and post-treatment tissue, in a phase 1b trial studying the safety of recombinant PV lerapolturev (formerly known as PVSRIPO) in recurrent paediatric HGG [18].However, the patient was pre-treated with an inactivated PV vaccine booster 1 week prior to lerapolturev [18].As a result, it was unclear whether this CD8+ T cell infiltration was the effect of the pre-treatment booster or lerapolturev itself [18]. Pelareorep is an oncolytic RV which had shown preclinical efficacy in immunocompetent glioma models after intravenous injection when combined with preconditioning with granulocyte-macrophage colony-stimulating factor (GM-CSF), which recruits APCs to the TIME [62].In a phase 1 trial of Pelareorep in paediatric patients with recurrent or refractory high-grade brain tumours led to a MOS of 11.7 months with increased levels of monocytes observed in 60% of patients post-treatment [17].Moreover, all patients showed drops in white blood cells, platelets and neutrophils in the first week after therapy, which recovered in week two [17].Serum cytokine levels showed no significant differences between patients and it is unclear whether these findings were the result of preconditioning with GM-CSF or the treatment with pelareorep [17].The findings of this study, however, were rather limited by the level of pretreatment and prognostic uncertainty of the included patient cohort, as well as the fact that only serum samples were taken for assessment of immune response [17]. In the paediatric clinical trials described above, OV treatment was related mainly to Grades 1, 2 and in some rare cases Grade 3 AEs and with 1 Grade 4 case (pelareorep) and thus did not lead to any serious neurological deterioration and decline of quality of life (QoL) by the time the clinical trials concluded [15][16][17][18].However, most OV clinical trials do not have data available on QoL, neurological changes or long-term effects from long-term survivors due to how recently those trials were initiated, and future clinical trials on OVs should accommodate for this. In light of these observations, these clinical trials suggest that oncolytic viruses may modulate the TIME by increasing immune cell infiltration, particularly of CD4+ and CD8+ T cells.Especially in these cases where immune modulation was successful, the survival benefit was significant.However, further research is needed to better understand the specific mechanisms involved and to determine whether these effects are primarily due to the viruses themselves or other factors such as pre-treatment interventions. Armed OVs As more and more evidence demonstrates the immunomodulatory actions of OVs and their importance for improved survival, genetic modifications began focusing on arming OVs with transgenes to enhance their effector functions [50].These types of enhanced OVs have not been extensively studied in PBTs yet; therefore, the following data will be based mainly on adult brain tumour models or patients in order to demonstrate their immunomodulatory potential. Immunomodulatory Cytokines In order to expand the innate capacity of OVs to stimulate an anti-tumour immune response, arming them with immunostimulatory cytokines to be expressed and released during infection and/or lysis of the tumour cells increases their concentration in the TME in order to help recruit immune cells and boost their anti-tumour efficacy [36]. For example, M032 a modified HSV expressing IL-12 is currently under investigation in an ongoing clinical trial in adult patients with recurrent or progressive malignant glioma (NCT02062827) [20].The secretion of IL-12 by the infected tumour cells, led to IFN-γ production, enhancing the anti-tumour effect of cytotoxic T cells and NK cells [20].R-115, an HSV, was also modified to express IL-12, inducing the same effects in the glioma mouse model [63].Another engineered OV, Ad-TD-nsIL12, is an Adv engineered with a non-secretory form of IL-12 to localise its presence in the TME, thereby limiting adverse effects [64].Ad-TD-nsIL12 is currently under investigation in primary and progressive paediatric DMG patients in a Phase 1 trial (NCT05717712 and NCT05717699). OV-IL15C is an HSV that induces the expression and secretion of a soluble human IL-15/IL-15Rα and was successful in improving the survival in vitro and in vivo in models of adult GBM while enhancing NK-cell and CD8+ T cell cytotoxicity [65]. Co-Stimulatory Molecules Similar to the function of cytokines effective T cell activation requires interaction and stimulation of co-stimulatory molecules.Therefore, the modification of OVs to express immune co-stimulators could enhance the antigen presentation of tumours and lead to increased activation of tumour-specific T cells.Delta-24-RGDOX also known as DNX-2440 is an example of such an OV [66].It is an oncolytic Adv that expresses OX40 ligand (OX40L), on the surface of a panel of human and mouse tumour cell lines [66].Moreover, Delta-24-RGDOX led to higher levels of CD4+ and CD8+ T cell infiltration and activity in the tumour environment than Delta-24-RGD.This improvement was negated by OX40L blocking [66].Importantly, treatment with Delta-24-RGDOX demonstrated anti-glioma activity and a longer median survival rate than Delta-24-RGD-treated mice, demonstrating the advantage of boosting the immunomodulating effect of Ovs [66].DNX-2440 is currently under a Phase I clinical trial (NCT03714334) in patients with recurrent GBM. Tumour Suppressor Genes Similarly, common tumour suppressor genes can be engineered into Ovs to be synthesised and secreted by the infected tumour cells.An example of this is HSV-P10, which expresses the gene for phosphatase and tensin homolog deleted on chromosome 10 alpha (PTENα) [67].IC treatment of mice with this OV led to increased immune cell recruitment of microglia, DCs, NK cells and CD8+ T cells into the tumour [67].Treatment with HSV-P10 in comparison to the non-PTEN expressing HSVQ in a panel of tumour cell lines showed a reduction in cell surface PD-L1 expression on tumour cells, thereby helping to overcome tumour immune escape [67].The therapeutic efficacy of HSV-P10 has additionally been validated preclinically in a model of GBM stem-like cells [68].However, in this study, the interactions with the TME were not investigated yet. Combinational Treatment of OVs with Immunotherapies Over the years, immunotherapies have reached the status of being an important modality in the fight against different types of cancer.However, due to challenges such as the suppressive TME present in many solid tumours, including PBTs, the efficacy of immune monotherapy has not been optimal.The ability of oncolytic virotherapy to modulate the TIME towards a more active, immune 'hot' state provides potential for improved efficacy of other immunotherapies if used in combination with OVs (Figure 4). Immune Checkpoint Inhibitors The objective of immune checkpoint blockade (ICB) therapy is to disrupt immunosuppressive tumour signals and reinstate robust anti-tumour immune responses by the tumour-specific T cells.ICIs are antibodies that bind to checkpoint receptors and ligands, limiting their interaction, and, consequently, the downstream immune response inhibition.Examples of such receptors and ligands include PD-1 and its ligand PD-L1, CTLA-4 and its ligands CD80/86, LAG-3 and its ligands galectin-9 and MHC-II, and T-cell immunoreceptor with Ig and ITIM domains (TIGIT) and its ligands CD155/112.ICIs have been studied as monotherapy in many tumours before; however, low levels of tumour infiltrating lymphocytes (TILs), low mutational burden and low expression of TAAs hinder their efficacy in PBTs [69]. As mentioned before, treatment with the OV G207 induced the expression of CTLA-4 and PD-1 on infiltrating CD8+ T cells as well as PD-L1 and IDO on tumour cells [61].This represents the basis of a potentially synergistic combination of OV therapy and ICIs.As of yet, there have not been any paediatric clinical trials using combinational therapy of both ICIs and OVs in brain tumours.However, there is much preclinical evidence supporting this combination as well as (pre-)clinical evidence from adult brain tumour patients. Recently, a multicentre Phase 1/2 study evaluated the combination of intra-tumoral delivery of DNX-2401 and the intravenous anti-PD-1 antibody pembrolizumab in recurrent adult GBM.The overall survival at 12 months in this trial was 52.7%, whereas the overall survival at 12 months in a previous trial of DNX-2401 monotherapy was 32% [19].Of note, no monotherapy comparator cohort for DNX-2401 or Pembrolizumab alone was included in this study [19].The combination of both treatments was very well tolerated in all patients with AEs of Grade 3 and lower, demonstrating the safety of such combinations [19]. The therapeutic modalities of ICIs and OVs can also be combined into one therapeutic modality.OVs that encode for and secrete immune checkpoint inhibitor antibodies within the TME have been engineered.Moreover, OVs can be engineered to secrete soluble forms of immune checkpoint ligands, to act as decoy ligands and sterically hinder the binding of the receptor to its primary ligand on the effector immune cells [70,71].Engeland et al. constructed an attenuated MV vector expressing the genes for antibodies against both CTLA-4 and PD-L1 [72]. Another alternative could be the targeting of "do not eat me" molecules expressed by the tumour cells (e.g., CD47 and CD24) to increase phagocytosis and antigen presentation for improved adaptive immunity [73].An oncolytic AdV armed with SIRP-a-Fc or Siglec-10-Fc was able to express the extracellular domain of those ligands that can bind to the receptors CD47 and CD24, respectively [73].Glioma mouse models rich with macrophages in the TME were treated with these oncolytic AdV, which led to improved anti-tumour activity [73].This activity was attributed to the increased levels of phagocytosis and antigen presentation while reducing the proportion of MDSCs, TAMs and Tregs [73]. CAR-T CAR-T cell therapy is a form of adoptive cellular therapy where the patient (or donor)derived T cells are modified to express a chimeric receptor, the CAR, capable of recognising specific surface tumour antigens and amplified ex vivo [74].Upon reinfusion, these engineered CAR-T cells enter the TME, recognise and kill tumour cells that express the specific antigen [74].However, CAR-T cell monotherapy encounters multiple barriers that can restrict its efficacy causing limited expansion and persistence in vivo [75].Briefly, in solid tumours such as PBTs, the CAR-Ts have to overcome restricted tumour infiltration and low levels of chemokines.The presence of immunosuppressive cytokines and inhibitory ligands, as well as low expression of TAAs of the "cold" TME usually result in an impaired effector CAR-T cell phenotype [76].As OV-mediated infection and lysis of tumour cells can lead to the release of type I IFNs, DAMPs, cytokines and chemokines, this may improve the aforementioned issues related to ineffective CAR-T cell therapy [77]. In addition to the favourable virus-intrinsic effects of OVs on the TIME for CAR-T cell therapy, OV engineering can be used to extend this favourable profile.As previously discussed, armed OVs can carry transgenes for chemokines and cytokines that promote CAR-T cell infiltration, proliferation and activity [78].Moreover, in situ transgene production of ICI antibodies carried by OVs could limit local CAR-T cell inhibition in the TME [79].An additional strategy under investigation involves using OVs for the targeted delivery of surface antigens to tumours.In this approach, the OV delivers the transgene coding for the antigen recognised by the CAR exclusively to tumour cells, leading to de novo cell surface expression of the antigen.This addresses the challenge of limited TAA expression and tumour recognition by CAR-T cells [80]. The use of Ad5delta-24 in combination with GD-2 CAR-T cells in PDX neuroblastoma mouse models improved survival while enhancing the persistence of the CAR-Ts in vivo compared to the GD-2 monotherapy [78].When the OV was further modified to release RANTES and IL-15 from the infected cells and combined with GD-2 CAR-Ts, the survival benefit and CAR-T persistence were significantly higher compared to the combination of GD-2 CAR-Ts and Ad5delta-24 [78]. However, OV-activated immune responses do not necessarily all positively impact CAR-T cell therapy.In a study using a B16EGFRvIII tumour mouse model, the combination of a vesicular stomatitis virus (VSV) with IFNβ transgene VSV-mIFNβ, along with an EGFRvIII CAR-T cell therapy, was studied [81].It was observed that VSV infection generated chemokine expression of CXCL10 and CCL5, which should be favourable for CAR-T cell recruitment [81].However, against expectations, no therapeutic effect was observed, along with a large reduction in CAR-T cell numbers.A similar but more moderate reduction in CAR-T cells was observed after combining CAR-T cell therapy with RV [81].It was discovered that the exposure of CAR-T cells to high levels of type I IFN (such as IFNβ) released upon OV infection led to the upregulation of PD-1, T-cell immunoglobulin and mucin domain-containing protein 3 (TIM-3), LAG-3 and FAS, leading to apoptosis due to exhaustion.Contrastingly, the combination of VSVmIFNβ with type I IFN-insensitive CAR-T cells (IFNAR1-knockout) led to reduced tumour growth and increased survival in a lymphodepleted mouse model [81].Thus, high levels of type I IFN may be both beneficial for recruitment and activation of T cell responses as well as detrimental to T cells through activation of negative feedback mechanisms to prevent autoimmunity.Therefore, the combination of OVs with CAR-T cell therapy should be well balanced in order to ensure the resulting TIME is neither suppressive to T cells nor incendiary, leading to T cell damage [77]. For effective combined OV and CAR-T cell therapy in PBTs, suitable CAR targets should be evaluated.Multiple CAR-T cell therapies are currently under clinical investigation for PBTs.GD-2 targeted CAR-T cells are studied in DMG (H3K27M mutant) [82], HER2 CAR T cells in recurrent and progressive EPN [83] and B7-H3 CAR T cells in paediatric HGG [84].Furthermore, preclinical studies are looking into targeting GPC-2, which is highly expressed in MB and HGG [85]. NK and CAR-NK NK cells act as a first line of response against viruses as part of the innate immune response and are equipped with receptors, namely NKp46, NKp30, NKp44, NKG2D and DNAX accessory molecule 1 (DNAM-1), which recognise ligands usually expressed on virus-infected cells [86].Clinical data have demonstrated the activation of NK cells induced by the OVs, and combinations of OVs and adoptive NK cell therapy are currently being investigated [87].In the case of NK and CAR-NK cells, the anti-viral response and thus preference for virally infected cells might lead to fast viral clearance before they are able to provide significant oncolysis and immune modulation.In light of this, many strategies are investigated.An example would be to deplete NK cells prior to OV treatment followed by adoptive NK treatment, a therapeutic sequence that significantly improved the survival of GBM-bearing mice [88]. Similar to CAR-T cells, CAR constructs can be expressed by NK cells to enhance their specificity to certain antigens.CAR-NK cells are not HLA alloreactive and thus can be derived from any donor without prior HLA matching enabling an off-the-self allogeneic therapy approach [89].A study by Ma et al. compared the efficacy of the combination of a modified HSV, OV-IL15C and EGFR-CAR-NK cells in immunocompetent GBM-bearing mice [65].The rationale was that alongside the virus-intrinsic effects of the OV, IL-15 would contribute to the development, homeostasis, activation and survival of the CAR-NK cells [65].The combined strategy suppressed tumour growth and led to significantly longer survival than either therapy alone [65].Additionally, the combination induced increased levels of infiltrating NK and CD8+ T cells and longer persistence of the CAR-NK cells [65].Many CAR-NK cell therapies are preclinically investigated for GBM (HER-2, GD-2, c-Met, EGFR, EGFRvIII, B7-H3, AXL, CD73 and NKG2DL) with specifically GD-2-CAR-NK cells being evaluated against multiple paediatric DMGs [90].The first clinical trial with CAR-NKs against GBM employs the HER-2 CAR (NCT03383978). CAR-Macrophages In the case of macrophages, a CAR redirects the phagocytotic action of the macrophages towards specific antigens, and the CAR-macrophages (CAR-MΦs) continue to act as APCs, resulting in stimulation of the adaptive immune response.Recently, a study by Chen et al. showed that co-delivery of anti-CD47 and CD133-CAR-MΦs in GBM mouse models could significantly improve tumour regression and survival compared to the untreated mice [91].The combination treatment increased the percentage of DCs, CD8+ and CD4+ T cells in the TME, increased expression of TNFα and IFN-γ, followed by an enrichment of M1 macrophages [91].No published study has investigated the combination of CAR-MΦs with OVs; however, modified OVs could be used to express the targets for the CAR-MΦs or ligands that could block the tumour escape from phagocytosis such as CD47 and CD24 [71].This would allow unhindered phagocytosis of the tumour and consequent antigen presentation from the CAR-MΦs and activation of adaptive immunity. Cancer Vaccines Cancer vaccines take various forms, such as DCs, peptides, nucleic acids or viral vectors used to stimulate or amplify an immune response against a specific tumour antigen by inducing immune memory against subsequent tumour growth.The low immunogenicity of PBTs is considered one big hurdle required to be overcome for effective action of cancer vaccines.This mainly stems from the limited epitopes presented on MHC II complexes on APCs in the TIME and the subsequent failure to recruit enough T (helper) cells in order to enact the cytotoxic action of anti-tumour T cells [92].As such, the use of OVs alongside cancer vaccines is being developed in multiple directions.First, as immunoadjuvants, OVs can be used as a primary immune boost, which preconditions the tumour for the effective application of a tumour vaccine [93].Second, OVs can become 'oncolytic vaccines' when one or more TAAs are encoded into the OV that will be presented later by the APCs to the T cells [93].However, the concurrent endogenous expression of TAAs in healthy tissues poses an important concern for the safety of this type of treatment.Therefore, to optimise this treatment modality, tumour-specific antigens (neoantigens) should be identified. Antibody Therapeutics Bispecific T cell engagers (BiTEs) consist of two antibodies connected with a flexible linker.One of the two usually targets a TAA on the tumour surface, whereas the other targets CD3 or a different cell-surface molecule on T cells (preferably an activator) [94].Consequently, BiTEs bring tumour cells and T cells together, leading to polyclonal T cell activation, independent of MHC/TCR presentation and co-stimulation [94].The poor delivery and penetration into solid tumours demonstrated by the BiTEs limit their therapeutic success [94]. One solution was to combine CARs with BiTEs into a single modified T cell product, the CAR.BiTE, which improves the specificity of the CAR-T towards the tumour.In vitro preclinical testing demonstrated favourable T-cell differentiation and phenotype compared to those activated by the CAR or BiTE alone [95].The CAR-EGFRVIII.BiTE-EGFR were able to induce cytotoxicity in multiple PDX GBM cell lines expressing EGFRVIII and EGFR as well as EGFRVIII negative cells expressing only EGFR [96]. The other solution would be to use OV genetic carriers to secrete BiTEs upon arrival and infection of the tumour, bypassing delivery issues and promoting T cell activation in the TIME [97].Little research has been conducted into the combined use of OVs and BiTEs in both paediatric and adult brain tumours.However, the continuous search for paediatric brain antigen targets suggests such a strategy would be possible in the future. Overcoming Limitations and Obstacles of OV Therapy 6.1. Preclinical Barriers Brain tumours in children exhibit distinct molecular and immunological characteristics compared to those found in adults.Thus, the extensive knowledge about adult brain cancers cannot be translated completely to PBTs [12].Therefore, the applicability of effective immune-based therapies originating from findings in adult disease models should be carefully assessed in representative models of paediatric disease [25].This task is more challenging than it may initially appear. Brain tumours are commonly studied in human or patient-derived xenograft (PDX) mouse models due to their translational potential.Orthotopic PDX brain tumour models develop within an intracranial environment that closely resembles the physical and chemical matrix found in human brains.However, these tumours grow in immunodeficient mice lacking a host immune system.Thus, it is impossible to study an endogenous T cell response in such models.Even though PDX models have facilitated the study of various immunotherapies, the absence of key components within the TIME raises concerns regarding the translational validity of these studies.Alternative models, such as nude athymic mice, are incapable of mounting an adaptive immune response but retain functional subsets including B cells, DCs, macrophages, NK cells and innate immune responses [28].Moreover, some immunocompetent models have been developed by introducing driver mutations that were identified in human tumours [28,98].For example, an immunocompetent mouse model of DMG was successfully generated by Du Chatinier et al. [99].They established tumour cell lines from DMG mouse models that had been induced through intra-uterine electroporation of DMG-specific mutations into the embryonic brainstem [99].Hereafter, these primary murine tumour cells were orthotopically implanted as allografts in syngeneic mice [99].This led to the generation of secondary brain tumours that accurately reflected the morphology and growth pattern of human DMG [99].Importantly, the TIME of these model tumours mirrored the immune 'cold' TME that is observed in DMG patient material [99].Humanised mouse models, on the other hand, are engrafted with brain tumours that originate from patient samples in a similar manner to PDX models, yet also engrafted with human immune cells [100].Though this model can recreate most TIME interactions, they are unable to fully develop mature innate cells, they lack functional lymph node structures and germinal centres and their antigen-specific antibody response is greatly curbed [100]. Another issue related to the preclinical modelling of PBTs and therapeutic interventions is the differences observed between animal and human-derived models.For instance, in a study of oncolytic HSV in MB by Hedberg et al., major differences between animalderived cell lines and human cell lines of MB were observed in terms of viral toxicity and viral recovery [51].Human-derived cell lines were more sensitive to OV infection and OV-mediated killing than the mouse cells.Moreover, substantial increases in viral yield in the human cells were observed [51].This highlights the need for accurate translational models, especially between species, to avoid cases of early discontinuation of potential therapeutics, due to underperformance demonstrated in murine cells as opposed to their intended human target. In an attempt to avoid this underperformance and reduce the use of animal models, multiple ex vivo models are being created using patient-derived cells.These models vary in complexity from simple mono-culture spheres to complex organoids, with their common trait being the maintenance of the phenotype and characteristics of the tumour of origin [101,102].The use of those in vitro techniques has enabled the investigation of the oncolytic efficacy of OVs against multiple types of PBTs and enable the investigation of OV interactions with a limited number of immune cell populations [103,104].However, with the current technology and knowledge, those models remain insufficient at fully recreating all the events that occur during the treatment and cannot accommodate the multiple factors that would, in practice, affect the efficacy of the treatment. As a solution, some scientists suggest that a variety of models should always be used complementarily as factors that are absent or non-reactive in one model could be assessed in another to obtain a more complete image of the ongoing interactions e.g., the use of multiple animal species or combinations of in vivo and in vitro models [77].In an ideal scenario, the advancement of preclinical models would encompass a wide spectrum of PBTs, including their subtypes, various disease stages, and diverse local TIME. Monitoring Barriers In the field of OV therapy, accurate monitoring of viral delivery and characterising the anti-tumour immune responses triggered by the virus in a reliable, non-invasive manner is difficult. Currently, clinical efficacy is determined using a combination of magnetic resonance imaging (MRI) and blood tests.However, in MRI imaging the OV-induced influx of immune cell infiltration can increase the tumour volume, leading to a perceived pseudoprogression on the imaging, a problem reported by clinicians in all main paediatric clinical trials [15][16][17][18].This compromises our ability to accurately track the extent of progression-free survival as well as the progression of disease based on imaging [15].One proposed solution to this issue is the implementation of the durable response rate (DRR) as an endpoint in cancer immunotherapy, especially in clinical studies.As a response may occur after an initial increase in tumour size, the need for a durability dimension in the measurement of response became apparent.The DRR is defined as the rate of objective response (complete response/partial response) by WHO criteria, lasting ≥6 months continuously beginning within the first 12 months after initiation of treatment.Importantly, disease progression is allowed prior to the onset of response [105].Attaining a lasting response correlated with clinical advantages, including improved overall MOS, improved QoL, and an extended treatment-free interval.This underscores the significance of durable response as a valuable endpoint in immunotherapy clinical trials. In terms of monitoring the virally induced immune response in vivo, more challenges remain.To follow the viral spread, currently clinical samples collected through random biopsies under image guidance can later be analysed for the presence of the virus and immune cell infiltrates in a laboratory.However, this provides a limited representation of the viral replication in the tumour as the sample is taken from one location and the virus can replicate and spread unevenly to distant sites.Frequent sampling or sampling from distinct locations of the tumours was suggested, though these methods would not be ethically appropriate due to the high patient burden resulting from the invasiveness of the methodology [92].With the advent of single-cell technologies, multiplexed-immunofluorescent images or in situ spatial transcriptomics the sampling burden and the need for the repetitive sampling of the tumour, blood and CSF could be reduced as more detailed information about the viral dynamics, distribution and spatial interactions could be identified with the use of those techniques as they require smaller and fewer samples compared to the information they provide [106][107][108]. Furthermore, monitoring the virally induced immune response through systemic measurement of increases in specific immune cell types does not necessarily reflect the immune status in the TME.Solely quantifying the infiltrating immune cells in the TIME as a marker of effective anti-tumour response is inadequate, due to the diverse pro-tumour and anti-tumour functions of different immune cell populations [109].As an example, the distribution of virusspecific versus tumour-specific infiltrating T cells is not well characterised, nor investigated during the different stages of treatment in the patients, despite the fact that this ratio could substantially alter the outcome in tumour response [77,109]. Hence, to comprehensively understand the antitumor impact of oncolytic virotherapy without the need for invasive tissue sampling, non-invasive imaging should encompass viral monitoring, assessment of immune checkpoint expression, and the tracing of immune cells infiltrating tumours.To combat these issues, OVs encoding transgenes for reporter genes to enable real-time tracking of viral replication, radiolabelled antibodies to a variety of checkpoint proteins and radiolabelled CAR-T cells are all examples of techniques currently under development to improve our understanding of how virotherapy affects the TIME in vivo [110].Until, their successful implementation, collection of cerebrospinal fluid (CSF) and tumour biopsies pre-and post-treatment along with several blood samplings during the treatment is highly recommended for the investigation of the OV progression and its effects on the immune modulation as data of this type are severely missing or lacking in most OV clinical trials. The Interplay between Administration, Neutralising Antibodies and Anti-Viral Response Multiple factors surrounding the context of administration can additionally affect whether OV treatment is successful in tumour eradication or not.IV administration is always preferred over IT due to the reduced invasiveness and strain it induces on the patient.Upon administration of OVs, the conflict between the host immune system and the virus begins.During OV-infected tumour cell death, along with TAAs, viral antigens are released instigating a parallel anti-viral response.Thus, rapid clearance of the therapeutic OVs by the innate immune system can minimise viral replication and continuous infection in tumour cells [111].Antiviral cellular immunity and neutralising antibodies constitute the natural immune response against a perceived pathogen. To partly mitigate this conflict, direct intratumoral administration can be used to maximise OV tumour distribution.However, in cases of metastatic disease or tumours where intratumoral administration is completely or almost completely inaccessible, another strategy is required.To limit the host's response against the OV and enable repeated administrations, an immunosuppressive agent such as cyclophosphamide can be administered for a short term to suppress innate immune responses and reduce the number of neutralising antibodies [112].However, neutralising antibodies play a significant role in OV safety as in exceptional cases the OVs can be distributed to organs or body parts away from the tumour area and cause toxicity due to dissemination of the OV in the healthy cells [112]. Interestingly, seroconversion of neutralising antibodies against the OV is also an important factor capable of impacting the efficacy of OV therapies.This often occurs if the patient has had previous exposure to the virus type of the OV but can occur during OV treatment as well.One small benefit is that pre-existing neutralising antibodies (for some virus serotypes) do occur a lot more in adult patients as opposed to paediatric patients [113].When comparing the MOS of OV-treated DMG patients with low versus high neutralising antibody titres, the patients with low antibody titres had a much more favourable prognosis [15].A recent clinical trial using an HSV-1 named CAN-3110 demonstrated the opposite results.During that trial patients with recurring GBM were treated with a single dose of CAN-3110 and, interestingly, existing HSV-1 seropositivity prior to treatment was correlated to higher CD4 and CD8 T cell infiltration intratumorally and a higher clinical benefit when compared to seronegative patients prior to treatment [44].In this trial, it was highlighted that the responders with seropositivity had no positive staining for HSV-1 intratumorally, while the non-responders with negative HSV-1 seropositivity had positive staining for HSV-1 [44].This indicated that the tumour in the responders was cleared due to an active anti-viral response, thus the lack of HSV-1 staining, which led to an anti-tumour response [44], whereas the non-responders had weak to no anti-viral response that allowed the OV persistence [44]. Potential ways to circumvent the mechanisms of OV clearance are to (a) introduce multiple-timepoint OV injections to increase the chance of the OVs inducing their effect and switch their seropositivity status, or (b) alter the virus packaging to conceal the OV from the host immune system until tumour cell infection.One of such shielding mechanisms is to package the virus inside carrier cells such as neural stem cells (NSCs) or mesenchymal stem cells (MSCs).These cells have inherent tumour tropism, can cross the BBB, and distribute easily throughout the tumour, whilst enabling the OVs to evade immune cells and neutralising antibodies [42,114].A clinical trial for multiple adult tumours, including two paediatric MB patients, employing the oncolytic virus Icovir-5 loaded into autologous MSCs (NCT01844661) and a paediatric clinical trial loading Icovir-5 into heterologous MSCs (NCT04758533) already exist; however, the unknown serological status of the patients and a small number of paediatric patients do not provide sufficient evidence for the implication of an anti-viral response in the responders/non-responders [115].Other strategies to camouflage the OVs are using polymer coatings such as polyethylene glycol (PEG) to limit antibody binding or modifying the virus by switching viral envelopes or capsids of multiple virus serotypes [116].All these measures would serve as a temporal form of protection for the OVs to be able to initiate their manipulation of the TIME without the need to improve their virulence. While the magnitude of the innate immune response triggered by OVs is crucial for initiating an adaptive immune response to the virus and potentially in combating tumours, it can also be detrimental to pre-existing effector and memory T cells, including adoptively transferred CAR-T cells.This occurs because these cells are dependent on a specific balance of immunomodulatory mediators for their proliferation and activation in the TIME [77].Although the combinational therapy of OVs with other modalities of immunotherapies has generated promising preliminary results, the timing and dosage of OV therapy need to be carefully optimised.For instance, as opposed to co-administration of the two therapies, OVs could be used as an 'immune-priming' treatment followed up by the second immunotherapeutic modality after the first surge of the inflammatory response has subsided.This model of administration could be used with the multiple potentially synergistic combinations of OV therapy with immunotherapies such as ICIs, adoptive cellular therapy or cancer vaccines that we previously described [13,77]. Conclusions Considering the small number of patients and clinical trials tested, the data currently available on OV therapy have demonstrated improved MOS with limited implications; however, an investigation is still required to monitor any long-term effects of this therapy in survivors, a factor that should be implemented in future clinical trials.Based on the preclinical and clinical observations, the highlight of OV action has switched from oncolysis to immunomodulation and thus, the development of OVs that effectively but precisely modulate the TIME to facilitate optimal tumour killing is well underway.Combinations of various immune therapies with OVs are conceived frequently, producing favourable effects, but some roadblocks still exist that hamper their successful implementation.Improvements should be made in terms of dose/effect response, administration routes, number of injections and timing of OV therapy, especially when used alongside other immunotherapy modalities.To facilitate this, pre-clinical models should be developed that accurately reflect the TIME in paediatric patients.Last, the markers of progression should be developed to track effective treatment, to reflect OV persistence/clearance, OV distribution and immunomodulatory effects in the least invasive manner (Figure 5).The first steps towards improvement have been taken and we should soon begin to see the full potential of OV therapy in the treatment of PBTs. alongside other immunotherapy modalities.To facilitate this, pre-clinical models should be developed that accurately reflect the TIME in paediatric patients.Last, the markers of progression should be developed to track effective treatment, to reflect OV persistence/clearance, OV distribution and immunomodulatory effects in the least invasive manner (Figure 5).The first steps towards improvement have been taken and we should soon begin to see the full potential of OV therapy in the treatment of PBTs. Figure 5. Overview of future suggestions to improve OV therapeutic efficacy, safety and monitoring.OV characteristics affecting entry and safety can be optimised through serotype switching, altering fibre knobs, capsid proteins or targeting surface antigens.Moreover, OVs can be genetically modified to only allow replication in tumour cells.To enhance OV-induced tumour cell eradication, OVs can carry transgenes inducing the expression of costimulatory molecules or secretion of pro-inflammatory cytokines by infected tumour cells.Current preclinical models of PBTs can be optimised by developing specific paediatric models (as opposed to adult), immunocompetent models, and by using of a variety of models to improve the translation of preclinical data to the clinic.During clinical use of OVs, modifications can be made to improve the OV efficacy.The perseverance of OVs in the body can be enhanced by using polymer coatings or packaging the OVs in carrier cells that can cross the BBB.Moreover, the manner and timing of administration can be altered as well as the OV characteristics affecting entry and safety can be optimised through serotype switching, altering fibre knobs, capsid proteins or targeting surface antigens.Moreover, OVs can be genetically modified to only allow replication in tumour cells.To enhance OV-induced tumour cell eradication, OVs can carry transgenes inducing the expression of costimulatory molecules or secretion of pro-inflammatory cytokines by infected tumour cells.Current preclinical models of PBTs can be optimised by developing specific paediatric models (as opposed to adult), immunocompetent models, and by using of a variety of models to improve the translation of preclinical data to the clinic.During clinical use of OVs, modifications can be made to improve the OV efficacy.The perseverance of OVs in the body can be enhanced by using polymer coatings or packaging the OVs in carrier cells that can cross the BBB.Moreover, the manner and timing of administration can be altered as well as the consideration of pretreatment with an immunosuppressive agent to minimise the innate immune response targeting the OVs.OV therapy can additionally be combined with other immunotherapeutic modalities, potentially leading to synergetic therapeutic effects.The monitoring of OV therapy can be improved by focusing on alternative endpoints to avoid unnecessary early termination of treatment, by increasing sampling intervals, and by improving methods of continuous non-invasive monitoring.MSC: mesenchymal stem cell, NSC: neural stem cell, IV: intravenous, IT: intratumoral. Figure 1 .Figure 1 . Figure 1.Immune cold versus immune hot tumour microenvironment.The TIME of PBTs is variable depending on the type of tumour; however, it is generally considered to be immunologically cold.This is characterised by low infiltration of CD4+ and CD8+ T cells combined with high infiltration of regulatory T cells and MDSCs.The Tregs and MDSCs secrete immunosuppressive mediators such as IL-10 and TGF-β.The CD4+ and CD8+ T cells that are present in the TIME express high levels of Figure 1.Immune cold versus immune hot tumour microenvironment.The TIME of PBTs is variable depending on the type of tumour; however, it is generally considered to be immunologically cold.This is characterised by low infiltration of CD4+ and CD8+ T cells combined with high infiltration of regulatory T cells and MDSCs.The Tregs and MDSCs secrete immunosuppressive mediators such as IL-10 and TGF-β.The CD4+ and CD8+ T cells that are present in the TIME express high levels of immune checkpoint molecules such as PD-1.Moreover, the TMB in PBTs is low.This results in less tumour-associated and specific antigens being presented on the tumour cell surface for immune cell recognition.Macrophages and microglia present in the TIME are predominantly M2 polarised and, therefore, are tumour supporting.Conversely, in a hot TIME the infiltration of effector T cells, DCs, macrophages and microglia is high.Moreover, a high TMB leads to a higher neoantigen load, resulting in more tumour recognition and immune cell activation. Figure 2 . Figure 2. Oncolytic viruses (OVs) for brain tumour treatment.(A) There is a broad array of OVs that have been studied as anticancer agents in brain tumours.Different subtypes include RNA and DNA viruses with either single-stranded (ss) or double-stranded (ds) nucleic acids and the presence or absence of a viral envelope.(B) Tumour tropism and entry targeting of OVs can be enhanced in multiple ways: (1) 'serotype switching', (2) modifying targeting peptides of fibre knob domain, (3) insertion of viral envelope glycoproteins from alternate viral families and (4) insertion of genes encoding for a single-chain antibody specific to a known tumour-surface antigen.Moreover, post-entry tumour specificity can be enhanced through insertion of tumour-specific promotors, which only allow for expression of viral genes in tumour cells or through mutation or deletion of specific genes to deprive OVs of the ability to replicate in normal cells and reduce their virulence.(C) OVs' anti-tumour effect is accomplished through two mechanisms.First, OV infection into tumour cells, resulting in replication and cell lysis to further spread newly synthesised viral particles.Second, the lysed tumour cells release TAAs and damage signals which, along with the virus antigens, induce local APCs to activate and, in turn, activate effector T cells such as CD4+ and CD8+ T cells. Figure 3 . Figure 3.The OV-mediated modulation of the tumour immune microenvironment.(1) Upon OVinduced tumour cell lysis, TAAs, cytokines, DAMPs and PAMPs are released into the TME.(2) Innate immune cells such as NK cells, macrophages and DCs get activated by these released particles.NK cells will directly lyse tumour cells.(3) Upon activation, DCs will act as APCs and migrate to the lymph nodes.There, they can present the TAAs to T cells, leading to the activation of CD4+ and CD8+ T cells.(4) The activated effector T cells travel to the tumour site with the help of secreted cytokines and chemokines by macrophages and DCs.(5) Upon arrival at the tumour site, CD8+ T cells selectively target and kill tumour cells that present the TAAs. Figure 3 . Figure 3.The OV-mediated modulation of the tumour immune microenvironment.(1) Upon OVinduced tumour cell lysis, TAAs, cytokines, DAMPs and PAMPs are released into the TME.(2) Innate immune cells such as NK cells, macrophages and DCs get activated by these released particles.NK cells will directly lyse tumour cells.(3) Upon activation, DCs will act as APCs and migrate to the lymph nodes.There, they can present the TAAs to T cells, leading to the activation of CD4+ and CD8+ T cells.(4) The activated effector T cells travel to the tumour site with the help of secreted cytokines and chemokines by macrophages and DCs.(5) Upon arrival at the tumour site, CD8+ T cells selectively target and kill tumour cells that present the TAAs. Figure 4 . Figure 4. Strategies for combination therapy of OVs with immunotherapies.The combined tumour targeting with both OVs and different immunotherapy modalities could lead to more efficient tumour killing.OV therapy has been shown to lead to upregulation of immune checkpoint molecules on both tumour and immune cells, thereby priming the tumour for effective ICI therapy.Furthermore, OV infection incites the release of cytokines and chemokines.This can help recruit CAR T and NK cells to the tumour site and stimulate proliferation.Modified OVs can express target ligands for CAR macrophages or ligands to block tumour escape from phagocytosis.Moreover, the OV-induced influx of cytotoxic T cells and T helper cells can improve the efficacy of cancer vaccines.Last, BiTEs can be produced at the tumour site by encoding their transgenes into OVs, providing improved delivery.CD80: cluster of differentiation 80, MHC-II: major histocompatibility complex II, TIM-3: T cell immunoglobulin and mucin domain-containing protein 3, TIGIT: T-cell Figure 4 . Figure 4. Strategies for combination therapy of OVs with immunotherapies.The combined tumour targeting with both OVs and different immunotherapy modalities could lead to more efficient tumour killing.OV therapy has been shown to lead to upregulation of immune checkpoint molecules on both tumour and immune cells, thereby priming the tumour for effective ICI therapy.Furthermore, OV infection incites the release of cytokines and chemokines.This can help recruit CAR T and NK cells to the tumour site and stimulate proliferation.Modified OVs can express target ligands for CAR macrophages or ligands to block tumour escape from phagocytosis.Moreover, the OV-induced influx of cytotoxic T cells and T helper cells can improve the efficacy of cancer vaccines.Last, BiTEs can be produced at the tumour site by encoding their transgenes into OVs, providing improved delivery.CD80: cluster of differentiation 80, MHC-II: major histocompatibility complex II, TIM-3: T cell immunoglobulin and mucin domain-containing protein 3, TIGIT: T-cell immunoreceptor with Ig and ITIM domains, CD155: cluster of differentiation 155, CAR: chimera antigen receptor, TCR: T cell receptor, CD3: cluster of differentiation 3. Figure 5 . Figure 5. Overview of future suggestions to improve OV therapeutic efficacy, safety and monitoring.OV characteristics affecting entry and safety can be optimised through serotype switching, altering fibre knobs, capsid proteins or targeting surface antigens.Moreover, OVs can be genetically modified to only allow replication in tumour cells.To enhance OV-induced tumour cell eradication, OVs can carry transgenes inducing the expression of costimulatory molecules or secretion of pro-inflammatory cytokines by infected tumour cells.Current preclinical models of PBTs can be optimised by developing specific paediatric models (as opposed to adult), immunocompetent models, and by using of a variety of models to improve the translation of preclinical data to the clinic.During clinical use of OVs, modifications can be made to improve the OV efficacy.The perseverance of OVs in the body can be enhanced by using polymer coatings or packaging the OVs in carrier cells that can cross the BBB.Moreover, the manner and timing of administration can be altered as well as the consideration of pretreatment with an immunosuppressive agent to minimise the innate immune response targeting the OVs.OV therapy can additionally be combined with other immunotherapeutic modalities, potentially leading to synergetic therapeutic effects.The monitoring of OV therapy can be improved by focusing on alternative endpoints to avoid unnecessary early termination of treatment, by increasing sampling intervals, and by improving methods of continuous non-invasive monitoring.MSC: mesenchymal stem cell, NSC: neural stem cell, IV: intravenous, IT: intratumoral. Table 1 . Oncolytic viruses investigated in clinical trials for the treatment of paediatric and adult brain cancers.IT: intratumoural administration, IV: intravenous administration, IA: intra-arterial administration, ICT: intracavitary administration, LB: lumbar puncture, MOS: median overall survival, OSR: Overall survival rate.Paediatric trials are indicated in orange, adult trials are indicated in green. Table 2 . Overview of the OV-induced immune cell modulations reported in preclinical paediatric brain tumour models. Table 3 . Overview of the reported OV-induced immune cell modulations in the paediatric tumour microenvironment in a clinical setting.This table only reported OVs with innate immunomodulatory effects.	2024-05-05T15:05:42.727Z	2024-05-01T00:00:00.000	{ "year": 2024, "sha1": "bba1eacfeccf30a918b594a0c630a0a3502ba4f6", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/1422-0067/25/9/5007/pdf?version=1714729560", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "cbb0b7ae104c6efbce3583739dabd9c4537be76d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
258891686	pes2o/s2orc	v3-fos-license	Editorial: Roles of flavonoids in crop quality improvement and response to stresses COPYRIGHT © 2023 Zhang, Gangurde, Yang and Zhao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. TYPE Editorial PUBLISHED 26 May 2023 DOI 10.3389/fpls.2023.1210666 Introduction Flavonoids play a variety of biological functions in plants. More than anthocyanins pigments displaying red, blue and purple, in flowers, fruits and leaves, which determine economic traits of crops and ornamental plants, they play multiple functional roles in plant-environment interactions which protect plants from biotic and abiotic stresses (Zhang et al., 2014;Landi et al., 2015;Qi et al., 2020). Importantly, flavonoids have critical functions in antioxidant and antitumor, as well as promoting blood circulation, thus it is benificial to health of human from eating edible seeds and fruits rich in flavonoids (Zhou et al., 2004;Li et al., 2018). Here, this Research Topic focus on progress on the function and role of flavonoids in plants encountering abiotic stresses, and the genetic engineering for manipulation of the flavonoids biosynthesis pathways to enhance flavonoid content and the abiotic stress tolerance in plants, especially crops. Interaction between flavonoids of plants and their environments Wine grapes are often grown in regions characterized by dry and warm summers, and a proper irrigation strategy is essential for obtaining high-quality berries and wines (Buesa et al., 2021;Palai et al., 2022;Romero et al., 2022). Palai et al. studied that effect of the regulated deficit irrigation (RDI) on berry flavonoid content and its related biosynthetic pathways in Sangiovese grapevines, including the moderate (RDI-1M) or severe (RDI-1S) water deficit from pea-size berry to veraison, moderate (RDI-2M) or severe (RDI-2S) water deficit from veraison through harvest, and severe during the lagphase (RDI-LS). They found a highest accumulation of anthocyanin in berries from both RDI-1 treatments, and expression of many genes in the flavonoid pathway was upregulated from beginning of veraison until harvest. Both post-veraison water deficit also increased anthocyanin concentration, but to a lesser degree than RDI-1. Particularly, the moderate deficit irrigation regardless of the applied periods in pre-or post-veraison, enhanced anthocyanin accumulation compared with that in severe water-deficit. Flavonol concentration was higher in RDI-1S berries, due to the upregulated expression of genes encoding flavonol synthases and the flavonol-3-O-glycosyltransferases. Overall, the study highlighted that the timing and intensity of deficit irrigation can modulate berry flavonoid accumulation as well as phenylpropanoid/flavonoid pathways, suggesting a proper management of deficit irrigation is critical for berry to enhance accumulation of anthocyanin and/ or flavonol. Alkali stress is highly destructive to the ecology, seriously affecting the soil structure and the growth of plants (Fancy et al., 2017). Nitraria tangutorum had strong alkaline resistance, accompanying a higher accumulation of flavonoid and anthocyanin during the period. Zhang et al. examined the role of the exogenous application of ABA and sodium nitroprusside (SNP) on defensive response of N. tangutorum to alkaline stress. They found that exogenous ABA and SNP significantly increased the plant height, fresh weight, relative water content and degree of succulency, as well as reduced the growth inhibition and physiological damage caused by alkali stress in N. tangutorum seedlings. However, SNP has a better effect on the improvement of photosynthetic efficiency and regulation of carbohydrate accumulation, while ABA has a more obvious effect on regulating flavonoid accumulation. Alkali stress also induced accumulation of endogenous NO and ABA, which might play a positive regulated role on defensive response of N. tangutorum to alkaline stress. Powdery mildew is a fungal disease devastating to wheat, causing significant loss in quality and yield (Fofana et al., 2005;Bajpai et al., 2019). Xu et al. conducted the combined analysis of transcriptome and metabolome in susceptible Jimai229 wheat and resistant HHG46 with and without powdery mildew inoculation. They revealed that the flavone and flavonol biosynthesis pathways were significantly enriched in both cultivars following powdery mildew inoculation, which was consistent with the upregulated flavonoid biosynthesis genes and increased accumulation of total flavonoid. Moreover, exogenous flavonoid treatment of inoculated plants led to fewer and smaller powdery mildew spots on the wheat leaves. Thus flavonoids is suggested to confer resistance to powdery mildew in wheat. Identification of function of candidate genes/loci involved in flavonoid accumulation, growth and development of plants Flavonol synthases play important role in regulating flavonoid metabolism, specifically the flavonol and anthocyanin branching pathways (Vu et al., 2015). Guo et al. used the targeted LC-MS to determine flavonoid-related substances in overexpression of GbFLSa in Populus poplar and CK, and revealed the content of proanthocyanins including catechin, epicatechin, epigallocatechin and gallocatechin, as well as expression levels of two DFRs, three ANSs and two LARs in transgenic poplars were significantly lower than that in nontransgenic plants. The study indicates that GbFLSa overexpression plays a negative regulatory role in biosynthesis of proanthocyanins. Peanut (Arachis hypogaea), with variegated testa color representing a distinct regulation pattern of anthocyanin accumulation in integument cells, has attractive appearance and higher market value. Chen et al. constructed two populations from the crosses between Fuhua 8 (pure-pink testa) and Wucai (red on white variegated testa), Quanhonghua 1 (pure-red testa) and Wucai, respectively, and identified the genetic locus underlying variegated testa color in peanut. They revealed that the pigmentation of colored region in red on white variegated testa was controlled by a previous reported gene AhRt1, while the formation of white region in variegated testa was controlled by genetic locus AhVt1 (Arachis hypogaea Variegated Testa 1). The molecular markers closely linked to the AhVt1 were developed, and the marker-assisted selection was used to develop new variegated testa peanut lines. The findings accelerate the breeding program for developing new peanut varieties with "colorful" testa colors, and laid a foundation for map-based cloning of genes responsible for variegated testa. Ginkgo possessing its distinctive branching and fan-shaped leaves, has become one of the world's most popular ornamental street trees (Crane, 2019). Flavonoids have been associated with the regulation of auxin transport (Santelia et al., 2008). Ni et al. investigated the phenotypic changes in transgenic tobacco (Nicotiana tabacum) overexpressing ginkgo GbDFR6. Pleiotropic defects in root and leaf development indicated an impaired auxin transport in transgenic tobacco plants, and eight flavonoids identified are ideal candidates as novel regulators of auxin transport by UPLC-ESI-MS/MS analysis. Delayed flowering in transgenic tobacco plants indicated anthocyanin or flavonoid is involved in the regulation of flowering time, which provide a valuable cue to produce the late flowering tobacco variety. The research suggested the novel and multiple roles of GbDFR6 in ginkgo. Author contributions QZ and CZ drafted the manuscript. SG and XY provided input and comments to the draft. All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication. Funding This research is supported by Shandong Province Natural Science Foundation (ZR2022MC201), National Natural Science Foundation of China (32072090), Shandong Province Natural Science Foundation (ZR2020MC104, ZR2020MC105), Agricultural Scientifc and Technological Innovation Project of Shandong Academy of Agricultural Sciences, and Taishan Scholar Project of Shandong Province.	2023-05-26T13:16:22.982Z	2023-05-26T00:00:00.000	{ "year": 2023, "sha1": "a601efac04b65cbfa37b2079c8e24d820e538e85", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Frontier", "pdf_hash": "a601efac04b65cbfa37b2079c8e24d820e538e85", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "extfieldsofstudy": [ "Medicine" ] }
266224843	pes2o/s2orc	v3-fos-license	Prognostic Value of Antithrombin Activity Levels in the Early Phase of Intensive Care: A 2-Center Retrospective Cohort Study To investigate the relationship between antithrombin (AT) activity level and prognosis in patients requiring intensive care. Patients whose AT activity was measured within 24 h of intensive care unit (ICU) admission were enrolled for analysis. The primary endpoint was mortality at discharge. Prognostic accuracy was examined using receiver operating characteristic (ROC) curves and cox hazard regression analysis. Patients were divided into 6 groups based on predicted mortality, and a χ2 independence test was performed on the prognostic value of AT activity for each predicted mortality; P < .05 was considered significant. A total of 281 cases were analyzed. AT activity was associated with mortality at discharge (AT% [interquartile range, IQR]): survivor group, 69 (56-86) versus nonsurvivor group, 56 (44-73), P = .0003). We found an increasing risk for mortality in both the lowest level of AT activity (<50%; hazard ratio [HR] 2.43, 95% confidence interval [CI] 1.20-4.89, P = .01) and the middle-low level of AT activity (≥ 50% and < 70%; HR 2.06, 95% CI 1.06-4.02, P = .03), compared with the normal AT activity level (≥ 70%). ROC curve analysis showed that the prediction accuracy of AT was an area under the curve (AUC) of 0.66 (cutoff 58%, sensitivity 61.4%, specificity 68.2%, P = .0003). AT activity was significantly prognostic in the group with 20% to 50% predicted mortality (AUC 0.74, sensitivity: 24.0%-55.5%, specificity: 83.3%-93.0%). An early decrease in AT activity level in ICU patients may be a predictor of mortality at discharge. Background Antithrombin (AT), an anticoagulation inhibitory factor synthesized in the liver, binds 1:1 to thrombin and factor XA and inhibits their coagulation activity.AT activity is recognized as an important blood test for the diagnosis of disseminated intravascular coagulation (DIC) syndrome and as an indicator of AT administration. 1 In DIC, AT activity decreases based on decreased synthesis, 2 increased loss, 3 and drug-induced 4 factors, in addition to increased consumption.][7] Patients admitted to intensive care units (ICUs) are known to have higher severity and mortality rates than those admitted to general wards. 8Although recent updates to scoring systems using big data and machine learning techniques have improved the predictive accuracy of case groups with complex patient backgrounds, 9 the amount, type, and time course of data used daily in the ICU environment changes rapidly.This makes the routine use of highly accurate prognostic scores are not easy. 10][13] Against this background, this study examines the potential of AT activity level as a predictor of prognosis in patients admitted to ICUs.Since there have been no studies that have examined the application of AT activity values to all patients admitted to the ICU, the hypothesis of this study is that AT activity values at ICU admission may be used in a simple and timely manner to predict prognosis.The purpose of this study is to analyze the prognostic accuracy of AT activity values and compare it with existing prognostic scores.It is hoped that the results will provide a new method to support decision making regarding treatment strategies and resource allocation for ICU patients.It also aims to improve the quality of care through common understanding and communication with medical staff, patients themselves, and their families. 14,15 Study Design This was a 2-center retrospective cohort study with extended follow up from ICU admission to discharge, conducted at Yamagata University Hospital and Nihonkai General Hospital.Ethics committee approval was obtained from all institutions (approval numbers, Yamagata University Hospital: 2020-312; Nihonkai General Hospital: 2020-29), and all methods were performed in accordance with the guidelines of the Declaration of Helsinki.The need for informed consent was obtained by the opt-out method, and the lead author is responsible for the completeness of the study data and the accuracy of data analysis. The inclusion criteria were all patients admitted to the ICU of the target facility from January 2018 to December 2019.Patients whose AT activity was measured within 24 h of ICU admission were included in the statistical analysis.Exclusion criteria were death within 24 h of ICU admission at age 18 years or younger, use of AT preparations prior to measurement, pregnancy, congenital AT deficiency, human immunodeficiency virus infection or acquired immunodeficiency syndrome. 16There was no upper age limit. AT activity was measured using ACL TOP 750 CTS (Instrumentation Laboratory, Bedford, MA, USA) or CP3000 (Sekisui Medical Company, Tokyo, Japan).The normal values of AT activity provided by the manufacturers were 83% to 128% for ACL TOP 750 CTS and 80% to 130% for CP3000.The same instrument was used for other coagulation markers, including FDP.Arterial blood is used in cases in which observation arterial manometry is being performed, and venous blood is used in all other cases.Specimens are stored in test tubes containing 3.2% sodium citrate, and plasma components are extracted by centrifugation at 1500 rpm for 15 min and measured with the above measuring instrument.These series of measurements are performed quickly after blood collection and specimens are processed at room temperature.Measurements are based on the guidelines of the Clinical Laboratory Standards Institute. 17he primary endpoint was mortality at discharge, and secondary endpoints were mortality at 7 and 28 days.The study complies with the guidelines for diagnostic and prognostic research (STARD and TRIPOD), but does not comply with the part of the study that separates train set, test set, and verification set. Statistics Continuous variables were evaluated for normality of distribution using the Kolmogorov-Smirnov test.Categorical data are presented as frequencies and percentages.Comparisons between 2 groups were made with the Mann-Whitney U test or Student's t-test for continuous variables; comparisons between 3 groups were made with the analysis of variance (ANOVA) test or Kruskal-Wallis test.Categorical variables were tested for χ 2 independence, and Fisher's exact test was adapted when there were 2 or more cells with an expected value <5.Diagnostic accuracy was assessed using the area under the curve (AUC) of the receiver operating characteristic curve (ROC); the optimal cutoff value for the ROC curve was calculated using the Youden index. We assessed the prognostic utility of AT activity levels using Cox proportional hazards regression models.Two distinct analyses were conducted: one examined the hazard ratio for each 1% increase in AT activity levels, and the other stratified patients into 3 groups based on their AT activity levels for a more granular understanding of the risk.Patients were divided into 3 groups [group H (AT activity: ≥70%), group M (AT activity: ≥50% and <70%), and group L (AT activity: < 50%)]. 18Results are presented with hazard ratios, 95% confidence intervals, and P values.Model counts for each variable were also listed.Schoenfeld residuals test was used to verify the assumptions.The Schoenfeld residuals for each variable were plotted over time to confirm that they were independent of time.For multicollinearity, the variance expansion factor (VIF) was calculated and a VIF >10 was considered to be collinear.Model evaluation was performed using the concordance index, log likelihood, partial log likelihood, partial Akaike information criterion, and log likelihood ratio test.Additional adjustment variables were age, body mass index, organ failure assessment (SOFA) score, and presence of surgery.The number of independent variables was set so that the number of independent variables was less than the number of the lesser outcome divided by 10, with death in hospital as the primary endpoint.In our pilot study, sample size calculations based on AT activity values in the surviving and nonsurviving groups were 10 cases for survival group and 10 cases for nonsurvival group each for a 50% mortality rate and 19 cases for survival cases and 4 cases for nonsurvival cases for an 18% mortality rate.These numbers were found to be less than the former. For preprocessing, missing data were compared with 3 groups divided by AT activity values (AT activity <50, ≥50 and <70, ≥70) and the distribution of missing data with the χ 2 test; if there was no association between AT activity values and the distribution of missing values, the analysis was performed without excluding missing values in multivariate analysis.After confirming that no differences were observed between the counties, multiple assignment methods were used to supplement the data.The multiple assignment method using Bayesian theory was used to supplement the results. 19ategorical variables were made into dummy variables of 0 and 1. As a sensitivities analysis, analyses on a subset of the data were performed as follows.A random sample was selected from the overall dataset and the same analysis was performed on that subset.This was repeated a total of 5 times to ensure that the results were consistent. All reported P values were 2-sided, with a P value of <.05 indicating a statistically significant difference.Statistical analysis was performed using Python (Python Software Foundation). Results There were 1984 patients admitted to the intensive care unit during the study period.Out of these, a total of 281 patients were included in the study, focusing on those who had their AT activity levels measured and met the inclusion criteria (Figure 1).Most of the patients were male (60.4%) and the age of eligible cases were 69 (58-77) years.The reasons for admission were postoperative cases (55.2%) and emergency admission (74.4%) (Table 1).Additional demographic details for each hospital are provided in Supplemental Table 1. Univariate analysis showed significant differences in AT activity levels between survivor and nonsurvivor groups for mortality after 28 days and at discharge (Table 1).The Acute Physiology And Chronic Health Evaluation II (APACHE II) score was calculated from data within 24 h of admission, and the predicted mortality rate was calculated from the APACHE II score.Age and sex did not differ between the 2 groups.FDP was the most common missing data (13.4%).The other missing data were included in a 3-group test based on AT activity values.Because of the low association between missing data and AT activity, cases with missing data were also included. ROC curve analysis showed that the AT activity level had a prognostic value of AUC 0.66 (cutoff: 58%, sensitivity: 61.4%, specificity: 68.2%) (Figure 2).This was statistically as accurate as the APACHE II and SOFA scores (APACHE II score P = .68,SOFA score P = .18). The subjects were divided into 3 groups according to the AT activity values [group H (AT activity: ≥70%), group M (AT activity: ≥50% and <70%), group L (AT activity: <50%)] (Figure 3 and Supplemental Table 2).The severity of patients differed between the groups.The SOFA score increased with decreasing AT activity values among the 3 groups (Figure 3B).Furthermore, Dunn's multiple comparison test showed that group L had a significantly higher SOFA score than group H (P = .0063).Survival rates differed significantly between groups L, M, and H (Figure 3C).By Bonferroni correction, group H had significantly higher survival rates than group L (log-rank test P = .0069,Gehan-Breslow-Wilcoxon test P = .011).The findings for group M's SOFA score and survival were better than those for group L and worse than those for group H, but they were not statistically different from one another. In multivariable-adjusted Cox-proportional hazards regression models, the risk of death is 2.43 times higher in patients with group L (≤50%) and 2.06 times higher in those with group L (≥50% and <70%) (Table 2).Another multivariate analysis, which examined each disease separately, showed that there was a significant hazard risk for different levels of AT activity (Supplemental Tables 3 and 4).The analysis revealed that the L group had a higher hazard ratio compared to the H group, and a similar trend was observed in the M group.After statistical adjustment, significant differences emerged in the non-DIC and nonsurgery categories.Additionally, ROC curve analysis for the most common diseases yielded AUC values ranging from 0.520 to 0.721 (refer to Supplemental Table 5).Specifically, in cases of sepsis with DIC, the AUC was 0.72, with a cutoff value of 53.5%, a sensitivity of 60.0%, and a specificity of 73.7%. We stratified patients by predicted mortality and evaluated accuracy (Table 3, Supplemental Figure 1, the number of cases divided by the AT activity value [cutoff value 58%], which forecasts hospitalization mortality obtained from the APACHE II score, stratifies the predicted mortality).The χ 2 test with an AT activity cutoff of 58% significantly predicted the outcome in the ≥20% and <50% predicted mortality group (AUC: 0.74, sensitivity: 24%-55.6%,specificity: 83.3%-93.0%). Discussion In our study, we found that the level of AT activity within 24 h of ICU admission serves as an independent predictor of discharge prognosis, comparable to APACHE II and SOFA scores in ROC curve analysis.The predictive accuracy varies based on the underlying disease and predicted mortality, having higher discriminatory power in cases with predicted mortality ≥20% and <50%. Contrasting results have been reported in previous studies regarding AT's role in prognosis.While some past research associated low AT activity with poor prognosis, [20][21][22][23][24][25][26] others found its predictive value limited 6,[27][28][29][30][31][32][33] Our results align with the former, our subgroup analysis revealed that AT activity demonstrated a prognostic significance in line with previous findings in open cardiac operations and septic conditions, 27,28 excluding illnesses with small sample sizes, such as acute pancreatitis and trauma.This observation was made with a demarcation threshold of 50% to 60% AT activity, which is consistent with prior documentation. There are several reasons why AT activity values are prognostically useful.It may indicate increased consumption or decreased production in severe cases as an indicator of organ damage, and it may produce an anti-inflammatory response through syndecans. In our study, the SOFA score showed the highest accuracy in the predictive model, emphasizing the grim prognosis in severe cases with multiple organ failures, consistent with prior studies. 34AT also showed comparable predictive accuracy, as AT is a marker used in DIC diagnostic criteria, it may serve Cox proportional hazards regression analysis was performed to ascertain prognostic determinants.In addition to a single model, multivariable models adjusted for age, presence of surgery, body mass index, and sequential organ failure assessment score were created.Abbreviations: CI, confidence interval; AT, antithrombin activity.Based on threshold values for AT activity of 58%, for each anticipated mortality stratified using the APACHE II score, hospital mortality was investigated using a chi-square test.Abbreviations: AT, antithrombin; APACHE, Acute Physiology and Chronic Health Evaluation. as an organ damage marker. 1 Yet, platelet count alone did not independently predict prognosis, pointing to the potential association of other SOFA score parameters with prognosis and AT activity levels (Supplemental Table 2 and Figure 2).As other coagulation factors, synthesis occurs in the liver and may reflect liver function. 35As an overall indicator of multiorgan failure, we consider that it may reflect prognosis.AT is not only an anticoagulant but also binds to syndecans and provides vascular-endothelium protection, 36 myocardial protection, 37,38 and renoprotection. 33Therefore, low AT activity levels may lead to organ damage that progresses over time and a poor prognosis.The anti-inflammatory activity of AT is mediated directly via syndecans and indirectly via the coagulation pathway. 39Syndecans are transmembrane proteoglycans that play a major role in inflammation by regulating leukocyte extravasation and cytokine function. 40In addition, syndecan is a major source of cell surface heparan sulfate and regulates inflammatory cell maturation, leukocyte-endothelial interactions, and chemotaxis. 41The coagulation pathway is also involved in inflammation, and excessive fibrin deposition in the lung is a prominent feature of lung injury. 424][45] For our study group, predicted mortality for the entire subject population was 26.2%, with actual mortality rates of 6.0% for 7-day mortality, 13.8% for 28-day mortality and 22.1% for mortality at discharge.][48] However, long-term prognosis, dictated by the late phase transition, remains a challenge. 49Our analysis showed that the group with predicted mortality of 20% to 50% had significant discriminative power.This suggests that even patients with mild-to-moderate declines in AT activity, below normal, are more likely to be discharged alive.Our study also allowed us to identify the patient group for whom AT is likely to be prognostic.The association between AT activity level and prognosis was found to be low in mild cases and very severe cases. The reason for the lack of significant difference in AT activity between survivors and nonsurvivors in cases with predicted mortality of 50% or higher, is likely due to the high proportion of severe cases in this group, particularly among those who died within 2 days of admission.The time of lowest AT activity value may vary depending on the pathological condition and time course. 31ooking at specific diseases, the AUC was lower in patients with DIC, with a ROCAUC of 0.52.However, considering only septic DIC cases, the AUC demonstrated moderate accuracy with a ROCAUC of 0.72.The implications of assessing AT activity levels per predicted mortality may have important implications for AT use.In the KyberSept trial, 50 improved survival was observed in patients treated with AT, excluding patients with predicted mortality of 60% or greater.Another RCT using AT agents found no significant difference in mortality for septic DIC with AT activity levels of 50% to 80%. 51n sepsis, 4 phenotypes correlating with host response patterns and clinical outcome have been noted. 52This suggests that AT preparations may be most effective when selectively administered to patients for whom reduced AT activity levels are associated with prognosis.In our study, an AT activity level of approximately 50 in sepsis was a useful cutoff value for prognosis, and prognosis was useful in cases with a predicted mortality rate of 20% to 50%.Therefore, we can infer that the benefit of AT therapy is relatively small in the groups with particularly high or low severity of illness and relatively high AT activity, since the AT activity value has little impact on prognosis. The strengths of this study include the high variety of the patients, as it examines all admissions from multiple institutions, the sufficient sample size for multivariate analysis and the short 2-year time period covered. There are 6 limitations of the study: first, because it is a retrospective observational study, the reason for testing AT activity levels may vary from one physician to another; measurement of AT activity levels is likely to be for diseases such as sepsis or macrovascular disease, which can cause abnormalities in the coagulation-fibrinolytic system.It is also possible that protocols for measuring coagulation time and treatment follow up varied from case to case, which may have also influenced the results.Second, while our sample size was statistically robust, it featured a skewed distribution due to the small number of participating facilities and their overwhelming representation of postoperative surgical cases in the ICU.Even though we evaluated the accuracy for both surgical and nonsurgical cases separately, it is essential to interpret our findings within the context of the specific characteristics of these facilities.Third, FDP was the most common missing data (13.4%).We performed a test including other missing data, dividing the patients into 3 groups according to AT activity values.The fourth limitation was the use of fresh frozen plasma (FFP) products: cases of AT agents use were excluded, but FFP products were not excluded from the administration.The more frequent use of FFP during open heart surgery may have led to increase AT activity levels.On the other hand, unfractionated heparin is used intraoperatively during open heart surgery.Unfractionated heparin binds tightly to AT and exerts anticoagulant effects.Fifth, the treatment and medical histories before ICU admission were not recorded.In sepsis, for example, the AT activity level may be modified because blood samples are collected differently in the early stage of the disease, when the inflammatory response has not yet increased, compared to the middle stage, when the inflammatory response is high.In addition, transfusions and invasive procedures in the emergency room or surgical department may result in different AT activity levels.Lastly, the sixth limitation involves the different instruments used to measure AT activity values.However, previous studies have reported minimal discrepancies between the ACL TOP 750 and CP3000 models used in this study, suggesting a negligible effect on the results. 53 Conclusions In an analysis of cases in which AT activity values were measured within 24 h of ICU admission, AT activity values were an independent prognostic predictor of mortality at discharge. Figure 1 . Figure 1.Study flow diagram of the patient selection process. Figure 2 . Figure 2. The ROC curves predicting inpatient mortality.The ROC curves reveal sensitivity/specificity relationships based on AT activity (A), APACHE II score (B), SOFA score (C), and acute DIC score (D), respectively.The cutoff value and P value for AT activity (A) was 58% and P = .0003,for APACHE II score (B) was 19 and P ≤ .0001,for SOFA score (C) was 8 and P ≤ .0001,for acute DIC score (D) was 2.5 and P = .0002.Abbreviations: APACHE, Acute Physiology and Chronic Health Evaluation; AT, antithrombin; AUC, area under the curve; 95% CI, 95% confidence interval; DIC, disseminated intravascular coagulation; ROC, receiver operating characteristic; SOFA, sequential organ failure assessment. Figure 3 . Figure 3. Distribution of AT activity levels in relation to the SOFA score and prognosis in participants.(A) Box-and-scatterplots according to their AT activity values, the subjects were split into 3 groups (groups L, M, and H).(B) The SOFA values for the 3 groups were different, with group L scoring higher than group H. (C) Kaplain-Meier curves reveal that the survival rates of the 3 groups varied, with group H showing a greater survival rate than group L. Abbreviations: AT, antithrombin; SOFA, sequential organ failure assessment. Table 1 . Characteristics of the Study Population. Table 2 . Cox Proportional Hazard Analysis to Distinguish Mortality at Discharge. Table 3 . Accuracy of Prognosis Using Stratified Values of AT Activity for Projected Mortality.	2023-12-16T12:45:58.036Z	2023-01-01T00:00:00.000	{ "year": 2023, "sha1": "bc3731d54d91c7e74617f026bc87658a43d1cb9e", "oa_license": "CCBYNC", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "ac5a4649e62f19f409608a256c3555a08be4a3d1", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
232387350	pes2o/s2orc	v3-fos-license	A Swollenin From Talaromyces leycettanus JCM12802 Enhances Cellulase Hydrolysis Toward Various Substrates Swollenins exist within some fungal species and are candidate accessory proteins for the biodegradation of cellulosic substrates. Here, we describe the identification of a swollenin gene, Tlswo, in Talaromyces leycettanus JCM12802. Tlswo was successfully expressed in both Trichoderma reesei and Pichia pastoris. Assay results indicate that TlSWO is capable of releasing reducing sugars from lichenan, barley β-glucan, carboxymethyl cellulose sodium (CMC-Na) and laminarin. The specific activity of TlSWO toward lichenan, barley β-glucan, carboxymethyl cellulose sodium (CMC-Na) and laminarin is 9.0 ± 0.100, 8.9 ± 0.100, 2.3 ± 0.002 and 0.79 ± 0.002 U/mg, respectively. Additionally, TlSWO had disruptive activity on Avicel and a synergistic effect with cellobiohydrolases, increasing the activity on pretreated corn stover by up to 72.2%. The functional diversity of TlSWO broadens its applicability in experimental settings, and indicating that it may be a promising candidate for future industrial applications. INTRODUCTION Lignocellulosic biomass or plant dry matter, has been considered an alternative to fossil fuels. However, the growth and processing of biomass feedstocks for conversion into fuels, or for other chemical production purposes, remain challenging (Galbe and Wallberg, 2019). A successful pretreatment process is measured in two ways. Firstly, by the highly efficient recovery of carbohydrates from raw materials. Secondly, by minimizing the formation of toxic and inhibitive compounds to prevent unwanted health risks and environmental hazards (Bhatia et al., 2020). The enzymatic deconstruction of biomass that follows pretreatment can be enhanced using a number of non-glycoside hydrolase accessory proteins. These enzymes include, but are not limited to, expansins, loosenins, cerato-platanin proteins and certain other types of carbohydrate binding modules (CBM) (Gourlay et al., 2013;Payne et al., 2015;Guo et al., 2017;Luti et al., 2020). Among these proteins, expansins have been widely found in plant cell walls where they function to loosen the cellular wall (Marowa et al., 2016). Mcqueen-Mason proposed that expansins can disrupt hydrogen bonding between plant cell wall polysaccharides without hydrolyzing them (Mcqueen-Mason and Cosgrove, 1994). In 2002, the first expansin-like protein SWOI from fungi was discovered in Trichoderma reesei (Saloheimo et al., 2002). Fungal swollenins have sequence similarity to expansins and are often referred to as expansin-like proteins. Indeed, SWOI was able to disrupt cotton fibers and filter paper structures on a microscopic level without detectable reducing sugars (Saloheimo et al., 2002). Over the past decade, more than 10 types of swollenins have been identified (Yao et al., 2008;Zhou et al., 2011;Kang et al., 2013;Santos et al., 2017) and the Expansin engineering Database (ExED 1 ), has recently been released to the public (Lohoff et al., 2020). In general, expansins are no longer than 250 amino acids and have a two-domain structure. The primary domain of expansins resembles the glycoside hydrolase family 45 (GH45) and this homology preserves certain sequence features of the GH45 catalytic site (Bharadwaj et al., 2020). The second domain has a characteristic flat aromatic-rich surface and is homologous to group-2 grass pollen allergens. Some studies have proposed that this domain functions as a CBM (Cosgrove, 2000;Georgelis et al., 2012). Within the expansin, the two domains are interconnected by a short linker, and both domains are required for plant cell-wall loosening activity. However, the structural discrepancies between swollenins and expansins lead to functional differences. For example, swollenins have an additional CBM domain, making them homologous to fungal cellulases in the N-terminal (Eibinger et al., 2016). In other cellulases, CBMs direct the binding of the enzymes to cellulosic surfaces and enhance lignocellulose degradation (Hoffrén et al., 1995;Velikodvorskaia et al., 2013;Maharjan et al., 2018). Additionally, the O-glycosylation of linkers in swollenins may also have an effect on the enzyme binding to surfaces (Amore et al., 2017). Previous studies have also described the synergy between swollenins and glycoside hydrolases in releasing soluble sugars from substrates. Zhou et al. successfully expressed swollenin SWO2 in Aspergillus niger, showing that the simultaneous incubation of SWO2 with cellulases results in a significant synergistic increase in cellulose hydrolysis activity. This synergy was further improved upon pretreatment of cellulose with swollenin (Zhou et al., 2011). Other investigations have focused on the synergistic relationship between swollenin and xylanase. Santos et al. (2017) found that the TlSWO swollenin from T. harzianum creates a rough and amorphous surface on Avicel and has a highly synergistic effect in combination with a commercial xylanase from T. viride, enhancing its hydrolytic performance by up to 147%. Furthermore, Anthony et al. (2006) showed that a chimeric enzyme with T. reesei swollenin fused with A. niger feruloyl esterase A could significantly increase ferulic acid release from lignocellulose samples. Other studies have demonstrated that swollenins can release reducing sugars from cellulosic materials. For example, swollenin SWO2 from T. pseudokoningii and Af Swo1 from Aspergillus fumigatus exhibit very low levels of endoglucanase activity (Chen et al., 2010;Zhou et al., 2011). Recently, Andberg et al. (2015) demonstrated that swollenin SWOI from T. reesei had activity on substrates containing β-1,4 glycosidic bonds, and hypothesized a unique mode of mechanistic action with similarities to both endoglucanases and cellobiohydrolases (Andberg et al., 2015). Previously, we identified the thermophilic T. leycettanus strain JCM12802 which is an excellent CAZyme source (Wang et al., 2015(Wang et al., , 2016aXia et al., 2016). In this study, we present a swollenin protein, TlSWO, that was cloned from T. leycettanus JCM12802, and successfully expressed in T. reesei and Pichia pastoris. We examined the TlSWO enzymatic performance on various substrates and showed that TlSWO can release reducing sugars from barley β-glucan, lichenan, laminarin and carboxymethyl cellulose sodium (CMC-Na). Additionally, we showed that TlSWO can reduce Avicel particle size and surface structure, and significantly increase synergistic activity by up to 72.2% on pretreated corn stover (PCS). Strains and Plasmids Talaromyces leycettanus JCM12802, the donor strain, was purchased from Japan Collection of Microorganisms RIKEN BioResource Center (Tsukuba, Japan). Escherichia coli Trans I-T1 (TransGen, Beijing, China) was used for routine gene cloning. T. reesei AST1116 and P. pastoris GS115 (Invitrogen, Carlsbad, CA, United States) were used as hosts for gene expression. pPIC9 (Invitrogen) and pTrEno plasmids of were used to drive Tlswo gene expression in P. pastoris and in T. reesei, respectively. The pTrEno plasmid was constructed described by Linger et al. (2015). Sequence Analysis TlSWO DNA and amino acid sequences were analyzed using BLASTx and BLASTp programs 2 , respectively (Johnson et al., 2008). TlSWO introns and exons were predicted using the GENSCAN Web Server 3 (Burge and Karlin, 1997). SignalP 4.0 was used to predict the signal peptide sequence 4 (Petersen et al., 2011). Potential N-glycosylation sites were predicted online 5 . Sequence assembly and estimation of the molecular mass and pI of the mature peptide were achieved using the Vector NTI Suite 10.0 software (Invitrogen). Protein molecular weight and molar extinction coefficients were estimated at the ExPASy tools page 6 . Multiple sequence alignments were performed using the Clustal W program from MEGA software 4.0. PROSITE 7 was used to analyze protein domains and functional sites (Nicolas et al., 2006). The DiANNA web server 8 was used to predict protein disulfide bond topology (Ferrè and Clote, 2005). When using P. pastoris GS115 as the expression host, pPIC9-Tlswo recombinant plasmids were linearized with BglII (New England Biolabs) and transformed into the expression host via electroporation. Positive transformants were screened on minimal dextrose medium at 30 • C for 3 or 4 days until single colonies appeared. Single colonies were placed into shaking tubes for enzyme production using protocol provided in the Pichia Expression Kit (Invitrogen). Large-scale fermentation was performed as previously described (Zheng et al., 2018). The recombinant P. pastoris GS115 transformant containing pPIC9-Tlswo was grown at 30 • C in 400 mL BMGY medium in a 1 L flask with shaking at 200 rpm for 48 h. Cells were collected and resuspended in 200 mL buffered methanol-complex (BMMY) medium with 0.5% (v/v) methanol and cultured at 30 • C for 72 h with shaking (200 rpm). Methanol was added into the medium every 24 h. When using T. reesei AST1116 as the expression host, recombinant pTrEno-Tlswo plasmids were linearized with Sbf I (New England Biolabs, United Kingdom) and used to transform T. reesei AST1116 via electroporation. Potato dextrose (PD) plates were used for spore production and PDHX plates (PD plates with hygromycin and TritonX-100 at final concentrations of 100 µg/mL and 0.1%, respectively) were used for screen potential T. reesei transformants which were grown for 2 to 3 days at 30 • C. Mandels and Andreotti medium with 5% glucose (MAG) was used as the growth medium for TlSWO expression. Subsequently, complete medium lactose (CML) was used for overexpression of the transformants. MAG and CML medium protocols were previously described by Linger et al. (2015). For large-scale fermentation, positive transformant spore stocks were streaked on potato dextrose agar plates and allowed to grow for 2 to 3 days until a well-developed plate of spores was formed. The wide end of a sterile 1.0-mL pipette tip was used to extract an approximately 0.5-cm plug from the plate and transferred into 1.0 L of MAG medium in a 2.8-L shake flask. The culture was grown at 28 • C with 225 rpm shaking for 24 h, after which the entire 1.0 L was transferred to 7.0 L of the same medium in a bioreactor. The entirety of 8.0 L of medium were mixed at a 200 rpm and grown, after which a filtered air of 1.0 vol * vol −1 * min −1 was used to purge while the system was kept at a constant temperature of 28 • C, and a pH of 4.8 for 48 h by using 2M KOH and HCl (Linger et al., 2015). Then the culture broths were extracted for SDS-PAGE and activity assay analyses. Culture broths were clarified via centrifugation and transferred to microcentrifuge tubes. Broths were diluted 3:1 in 4 × LDS (Lithium dodecyl sulfate) sample buffer (Life Technologies Corp., Carlsbad, CA, United States) with 50 µL/mL β-mercaptoethanol as a reducing agent. Samples were incubated at 95 • C for 5 min before loading onto NuPAGE SDS gels with MOPS buffer, and proteins were electrophoresed at 200 V for approximately 40 min. Protein Purification Fermentation broths were harvested and sequentially vacuumfiltered. Filtered broth was then concentrated by tangential ultrafiltration with a 10 kDa MWCO (GE Healthcare, Chicago, IL, United States). The broths were roughly concentrated to volumes of 100 mL. The final concentrated volume was exchanged with at least 2.0 L of 20 mM Bis-Tris pH 6.5 to remove residual peptides and other low molecular weight debris. The following purification steps were then performed as previously described (Linger et al., 2015). The crude enzyme was purified through hydrophobic interaction chromatography (HIC) using a 26/10 Phenyl Sepharose Fast Flow column (GE Healthcare). Then the protein was subjected to anion exchange chromatography using a 10/100 anion exchange column packed with Source 15Q (GE Healthcare), HIC using a Source 15 iso 10/100 column (GE Healthcare), and size exclusion chromatography (SEC) using a 26/60 Superdex 75 column (GE Healthcare). The mobile phase was 20 mM acetate buffer pH 5.0, 100 mM NaCl. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed to assess purity of the TlSWO protein. Proteins were separated on a 12% gel and visualized by Coomassie Blue staining. Protein concentration was measured using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific Inc., Rockford, IL, United States) and the Bradford protein assay kit (Bio-Rad). Effect of pH and Temperature on TlSWO Activity The effects of pH and temperature on TlSWO activity were measured and compared. To determine the optimum pH of TlSWO, enzyme activity was assayed using 1% lichenan (w/v) in a buffer of different pH, 100 mM glycine-HCl (pH 1.0-3.0), McIlvaine buffer (pH 3.0-8.0) and glycine-NaOH (pH 9.0-12.0). For pH stability, TlSWO was preincubated at 37 • C for 1 h in buffers of different pH (1.0-12.0) and subjected to the residual activity assay. The optimum temperature of TlSWO activity was determined in a reaction solution of pH 4.0 using a temperature range from 30 to 80 • C. The TlSWO thermostability assay (100 µg/mL) was performed out by preincubating the enzyme at 37, 50, 60, or 70 • C for 0-60 min, and performing residual activity assays on 100 µL aliquots withdrawn at different time points. Light and Scanning Electron Microscopic Analyses Avicel PH-101 was used as a solid cellulosic substrate. Avicel (10 mg) was incubated with different amounts of purified TlSWO in 100 mM citric acid-Na 2 HPO 4 buffer (pH 4.0). The experiment was performed on a rotary shaker at 40 • C for different time intervals. Control experiments without TlSWO were also performed under the same conditions. The physical structure of Avicel fibers was initially observed using light microscopy (Olympus TH4-200, Japan). Subsequently, photomicrographs of the samples were captured using a scanning electron microscope (Hitachi SU8010, Tokyo, Japan) at a voltage of 15 kV. Synergism Between TlSWO and Cellulases The substrates used in this work were National Renewable Energy Laboratory (NREL) dilute acid PCS P120927, cellulose nanocrystals (CNCs) and phosphoric acid swollen cellulose (PASC). Each substrate was equivalent to 8.5 mg of glucan. For CNC preparations from Avicel, about 2 g of Avicel was added to HCl pre-heated at 80 • C. Then, the acid hydrolysis was run for 4 h, and stirred every 15 min using a glass or Teflon rod, followed by centrifugation several times at 1,600 × g for 10 min. The supernatant was then decanted and the pellet resuspended in deionized (DI) water until the pH reached 5.0. CNCs were pelleted by centrifugation, collected and resuspended in DI water. Each substrate was suspended in 20 mM sodium acetate buffer, pH 5.0 and reactions were performed in triplicate in vials at 40 • C. Avicel PASC was prepared as described by Zhang et al. (2006). The enzyme cocktail comprised endoglucanase I from Trichoderma longibrachiatum (Megazyme Co.), cellobiohydrolases Cel7A from Penicillium funiculosum and β-glucosidase from Aspergillus niger (Megazyme Co.) at concentrations (mg protein/g of glucan) of 2, 10, and 1, respectively. The reaction was carried out for 120 h, with sampling every 24 h. Samples of 100 µL, containing both solids and liquid were removed and diluted for HPLC sugar analysis using a Bio-Rad HPX-87H column. Control experiments using bovine serum albumin were also performed under the same conditions. Identification and Characterization of the Tlswo Gene The putative Tlswo open reading frame consists of six exons and encodes a 503 amino acid protein (TlSWO) and a signal peptide at the cleavage site between amino acids 20 and 21. The cloned gene sequence encoding a putative TlSWO was submitted to NCBI GenBank as MT180127. Deduced TlSWO shared 71.5% sequence similarity to the swollenin (BAI83433.1) from A. fumigatus, 71.2% to the swollenin (ADZ74267.1) from P. oxalicum. Further analysis using PROSITE demonstrated that TlSWO consists of three domains, fungal-type carbohydratebinding module family 1 (CBM1) (amino acids 23−59), family 45 endoglucanase-like domain of expansin (Expansin_EG45) (amino acids 206−388) and a cellulose-binding-like domain of expansin (Expansin_CBD) (amino acids 400−492), all of which are typical of swollenins from fungi (Figure 1). The six cysteines in the CBM1 of TlSWO were highly conserved. DiANNA disulfide bond prediction identified three disulfide bonds in the CBM1 of TlSWO (Cys4-Cys21, Cys11-Cys28, and Cys22-Cys28). CBM1 and Expansin_EG45 are connected by a Serine-Threonine rich linker domain. Although the function of the linker has been studied in other cellulases, it is not clear if the linker plays a similar role in swollenins (Kolaczkowski et al., 2020;Nakamura et al., 2020;Pena et al., 2020). Sequence alignment of swollenins revealed that TlSWO maintained the conserved HMD (histidine, methionine, aspartic acid) catalytic motif of the GH45 cellulase (HFD, histidine, phenylalanine, aspartic acid), which is part of the active site (Figure 1). The aspartic acid of GH45 HFD is the proton donor during the catalytic process. However, the other residue of the catalytic active site, aspartic acid, is absent from both swollenins and some GH45 cellulases. The expansin CBD in the TlSWO C-terminal region is homologous to the pollen allergen. There are also several conserved aromatic amino acids in the TlSWO sequence, including Y400, Y401, F402, W429, Y447, W450, Y496, and F503, which may play key roles in substrate binding (Figure 1). Expression of TlSWO in P. pastoris GS115 and T. reesei AST1116 TlSWO (483 amino acids, theoretical molecular weight 51.1 kDa) was expressed in P. pastoris and T. reesei using the aox1 and eno promoters, respectively (Supplementary Figure 1). Based on the results of SDS-PAGE ( Figure S1), the quantity of TlSWO expressed was higher in T. reesei, thus all the TlSWO characterized in this study was expressed and purified from T. reesei. The purified swollenin protein migrated as a protein of ∼80 kDa. This single band was analyzed by MALDI-TOF MS because of the difference between the observed and predicted protein size, and the trypsin-digested peptide sequences were matched to the deduced TlSWO amino acid sequence (Supplementary Figure 2). Sequence prediction results indicate that TlSWO has five N-glycan sites (Asn35, Asn154, Asn249, Asn366, and Asn436). After Endo H digestion, the molecular weight of TlSWO decreased to ∼72 kDa, which is still higher than theoretical MW (Supplementary Figure 1A). We speculate that the remainder of the molecular weight increase was caused by heavy O-glycan glycosylation in the linker region which is rich in serines and threonines, as was reported for other proteins expressed in these hosts (Bai et al., 2019). TlSWO Activity on Different Substrates TlSWO cellulolytic activity was measured using the substrates lichenan, barley β-glucan, CMC-Na, laminarin, Avicel and glucomannan. Xylanase activity was measured with birchwood xylan, and mannase activity was measured with locust bean gum. All reactions were carried out overnight. Our results show that TlSWO has significant activity on lichenan, barley β-glucan, glucomannan and CMC-Na, and a very low activity on laminarin. TlSWO showed the highest activity on lichenan (9.0 ± 0.100 U/mg) and barley β-glucan (8.9 ± 0.100 U/mg), followed by CMC-Na (2.3 ± 0.002 U/mg). In contrast, a very low level of activity was observed with laminarin substrate (0.79 ± 0.002 U/mg) (Figure 2). Together, these results suggest that TlSWO mainly acts on cellulose rich substrates and shows a preference toward substrates with 1,4 linkages. Effect of Temperature and pH on TlSWO The effect of pH and temperature on TlSWO activity were investigated with lichenan as a substrate. Although the enzyme displayed activity across a broad pH range (2.0-12.0), we determined that the optimal pH for TlSWO is 4.0 ( Figure 3A). TlSWO retained more than 80% of its activity within the pH range 2.0-9.0 after incubation at 37 • C for 1 h. In contrast, TlSWO lost 30% of its activity after incubation for 1 h at 37 • C and pH 10.0-12.0 ( Figure 3B). Additionally, TlSWO reached optimal activity at 50 • C and retained more than 90% of its activity within the temperature range of 40-60 • C. However, after heating to 70 • C or above, TlSWO activity falls off very rapidly ( Figure 3C). Additionally, TlSWO maintained stable activity between 37 and 50 • C after 1 h of incubation, but its activity decreased to 40% after 10-min at 70 • C ( Figure 3D). TlSWO Mode of Action The TlSWO mode of action was assessed using lichenan, barley β-glucan, glucomannan and CMC-Na (Figure 4). As a result, CMC-Na was hydrolyzed into cellobiose and a small amount of cellotriose (Figure 4). Analysis of the hydrolysis products of lichenan and barley β-glucan showed that TlSWO preferentially hydrolyzed these two substrates into products with different degrees of polymerization, including cellobiose and cellopentose, followed by cellohexose and cellotetrose (Figure 4). We detected no sugar release after TlSWO incubation with laminarin (Figure 4). These results suggest that TlSWO may function as an endo-cellulase. Disruptive Action of TlSWO on Avicel The disruptive effect of TlSWO on Avicel was evaluated using light microscopy (LM) and scanning electron microscopy (SEM). LM analysis showed that after incubation with different amounts of TlSWO for 24 h, the Avicel's physical structure significantly differed from that of untreated Avicel (Figure 5). Avicel was disrupted into smaller particles with increasing amounts of TlSWO. Avicel pretreated with 300 µg of TlSWO for 12 h was subjected to further analysis using SEM. In this sample, TlSWO created a rough surface on Avicel when compared with the unpretreated sample (Supplementary Figure 3). Synergism Between TlSWO and Cellulases To test the capacity of TlSWO in enhancing biomass hydrolysis via an enzymatic cocktail, we hydrolyzed pretreated biomass using cellulases alone first, followed by treatment using both cellulases and TlSWO. Biomass degradation experiments were performed using β-glucosidase (EC 3.2.1.21), cellobiohydrolase (EC 3.2.1.91) and endoglucanase (EC 3.2.1.4) in the presence of TlSWO. Reactions with BSA and without TlSWO were used as controls. A total of 13 mg protein/g of glucan was used in all reactions. Endoglucanases randomly cleave internal β-1,4-glycosidic bonds to create new reducing ends. This allows cellobiohydrolases to continuously act on the chain termini to release cellobiose, and β-glucosidase then hydrolyzes cellobiose into glucose (Zheng et al., 2018). Therefore, the production of glucose, as the endpoint, was compared in the different reactions. When using PCS as the substrate, TlSWO exhibited significant synergetic effects in the presence of cellobiohydrolase Cel7A. No glucose was detected when PCS was reacted with TlSWO at 13 mg protein/g of glucan, suggesting that PCS could not be hydrolyzed to monomers by TlSWO alone (Figure 6). When the reaction contained Cel7A and β-glucosidase individually, PCS conversion increased from 8.9% at 24 h to 16.4% after 120 h. When Cel7A and β-glucosidase were supplemented with TlSWO at 2 mg protein/g of glucan, glucose increased from 11.2% at 24 h to 26.4% at 120 h. Although TlSWO alone did not produce detectable levels of released glucose, it significantly enhanced hydrolytic activity when added to Cel7A and β-glucosidase. Phosphoric acid swollen cellulose and cellulose nanocrystals exist in amorphous and crystalline forms, respectively, which may affect their binding with TlSWO. Therefore, the effect of TlSWO on these substrates was further examined. Similar to what we observed with PCS, TlSWO could not release any sugars from PASC and CNC without the presence of other cellulase/s (Figure 7). When Cel7A and β-glucosidase were utilized, the PASC conversion rate increased from 9.9% at 24 h, to 40.2% at 120 h ( Figure 7A). However, when Cel7A and β-glucosidase were supplemented with TlSWO, the conversion rates increased from 33.0% at 24 h to 72.2% at 120 h. The 120-h data shows that the conversion rate of PASC increased by approximately 32% following TlSWO supplementation, suggesting that TlSWO FIGURE 2 \| TlSWO substrate specificity. Hydrolysis reactions using 1% barley β-glucan, 1% CMC-Na, 1% laminarin and 0.5% lichenan substrates were performed overnight at pH 4.0 and 50 • C. The linkage type of each substrate is labeled. has significant synergetic effects with Cel7A. Additionally, the synergetic effects of TlSWO and endoglucanases were also explored. When using endoglucanase and β-glucosidase individually, the conversion rate increase was 51.4% at 24 h and 85.6% at 120 h. When used in combination with TlSWO, the conversion rate was 58.6% at 24 h and 85.7% at 120 h. Although these conversion rates are slightly higher than those observed using endoglucanase and β-glucosidase individually, the conversion rate at 120 h was 86.4% and was within the margin of error when compared to 85.6%. Therefore, we conclude that the addition of TlSWO did not significantly increase PASC enzymatic hydrolysis when used in combination with endoglucanase. The cellulose conversion rate of CNC was overall significantly lower than that of PASC. Using CNC as a substrate, Cel7A and β-glucosidase achieved a conversion rate of 28.2% at 24 h and 59.3% at 120 h (Figure 7B), which increased further when TlSWO was added. The glucose yields obtained by the enzyme cocktail systems containing TlSWO, Cel7A and β-glucosidase were 31.9% at 24 h and 68.9% at 120 h, which is higher than all other cases when compared with the control group. Therefore, we conclude that TlSWO has a synergistic effect with processive cellobiohydrolases. However, when used in combination with endoglucanases, TlSWO does not produce any significant synergistic effect. After 120 h of enzymatic hydrolysis, the glucose yield released from CNC was 16.4% with the combination of TlSWO, endoglucanase and β-glucosidase. This corresponds well with the hydrolysis rate of 14.9% observed using endoglucanase and β-glucosidase individually. Taken together, our results indicate that, TlSWO acts more efficiently on amorphous cellulose than on the respective crystalline forms. DISCUSSION Several previous studies have shown that SWOI can disrupt plant cell wall structures without leaving any traceable amounts of reducing sugars. However, subsequent research has confirmed that SWOI does exhibit some hydrolytic activity on cellulosic substrates with features of both endoglucanases and cellobiohydrolases (Saloheimo et al., 2002;Andberg et al., 2015). Similar to results observed for SWOI, the other two swollenins, Af SWO1 from A. fumigatus and SWO2 from T. pseudokoningii have shown hydrolytic activity on various substrates, suggesting that these proteins interact with cellulose or hemicellulose (Chen et al., 2010;Zhou et al., 2011). In this study, TlSWO from T. leycettanus JCM12802 was found to have similar functionalities as other fungal swollenins. TlSWO was also shown to share relatively high sequence identity with SWOI, Af SWO1 and SWO2 (64.5, 73.7, and 63.2%, respectively). TlSWO has the highest activity on lichenan and barley β-glucan substrates, both of which contain β-1,4 and β-1,3 linkages, and its activity is minimal on laminarin, a substrate that only contains β-1,3 linkages. This indicates that the primary mechanism of action of TlSWO is via activity on β-1,4 linkages. Expansins are more highly similar to GH45 subfamily C enzymes than to other members of the GH45 family (Igarashi et al., 2008;Godoy et al., 2018). These expansins have a HFD motif termed as part of their active site, and this motif is also present in TlSWO (HMD). The aspartic acid in this motif plays the role of proton donor in GH45. Nevertheless, other key residues that are critical for catalytic activities are absent in both expansins and swollenins. This difference suggests expansins and swollenins may use an inverted mechanism during the catalytic process. In 2015, Nakamura et al. (2015) proposed that PcCel45A, which belongs to GH45 subfamily C, uses an imidic acid form of asparagine residue as a general base in the "Newton's cradle" proton relay catalytic mechanism. This proposal sheds light on some potential mechanistic properties of expansin's catalytic process. Horizontal gene transfer drives sequence differences between fungal swollenins and plant/bacterial expansins (Nikolas et al., 2014). These differences could mean that the inverted catalytic mechanism theory of PcCel45A is unfeasible when applied to fungal swollenins. Fungal swollenins are roughly twice as large as plant and bacterial swollenins because their D1 and D2 domains contain extra sequence insertions including the additional N-terminal CBM with linkers. CBMs can increase the concentration of their parent enzyme substrate surface, leading to more rapid polysaccharide degradation (Bolam et al., 1998;Janne et al., 2003). Primary amino acid sequence analysis using BLAST indicates that TlSWO contains an N-terminal CBM region (amino acid residues 21-59) that shows the highest similarity toward fungal GH6 family 1 CBMs. Six typical conserved cysteines present in the sequence may form three pairs of disulfide bonds. The TlSWO CBM also contains three conserved aromatic residues (Trp28, Tyr54, and Tyr55), which are typical in GH6 and GH7 cellulase CBMs. These sequences are important for cellulase stability and activity during the reaction. Moreover, the linker length is crucial for cellulase activity (Srisodsuk et al., 1993). The linker region of TlSWO is over 140 amino acids, and is longer than most reported swollenins and fungal cellulases (Sammond et al., 2012). Although CBMs and linkers have been well studied in cellulases, little is known about the role of these two regions in swollenins. We have shown that when treated with TlSWO, the smooth surface structure of the microcrystalline Avicel transitions into a rough texture. This is consistent with previous studies suggesting that swollenin proteins function to modify the cell wall (Cosgrove, 2017;Javier and Daniel, 2005). Previous studies have also shown that swollenins can synergize with other enzymes including cellulases and xylanases (Chen et al., 2010;Kang et al., 2013;Santos et al., 2017). In this study, we explored the ability of TlSWO in boosting cellulosic substrate hydrolysis by different enzymes. When the cellulose was incubated with TlSWO and cellobiohydrolases, a greater increase in glucose yields was observed. However, we observed no significant synergistic effect between TlSWO and endoglucanases. These results differ from those of a previous report, in which swollenin exhibited strong synergistic interaction with endoglucanases (Gourlay et al., 2013). Therefore, we propose that TlSWO has better synergistic activity with cellobiohydrolases. Cellulose degradation is summarized by the classical C 1 -C x model (C 1 : non-hydrolytic components, C x : endo-or exo-acting cellulases). Using this model, one proposal hypothesizes that C 1 can disrupt cellulose by displacing hydrogen bonds in the microfibril, leading to a more available structure for C x (Liu and King, 1967;Payne et al., 2015). Eibinger et al., and Kang et al., speculate that swollenins, much like endoglucanases, can act as C 1 components because of their disruptive activity in the enzymatic saccharification of lignocellulosic substrates. However, the mode of action of various C 1 proteins needs to be further explored (Reese et al., 1950;Kang et al., 2013;Eibinger et al., 2016). Our results using PASC and CNC substrates with or without TlSWO were used to compare the effects of swollenin on different crystallinity materials. We showed that the total glucose concentration increased by 32% when PASC was incubated with TlSWO and cellobiohydrolases, compared to incubating with only cellobiohydrolases. However, when using CNC as the substrate, no significant difference between the groups with or without TlSWO was observed. These results suggest that swollenin has a greater ability to bind and disrupt amorphous cellulose than it does crystalline cellulose. This may be due to the endoglucanase-like activity of TlSWO, as demonstrated on the model substrates we previously used, or because it is able to bind amorphous substrates more easily than crystalline substrates due to its CBM. CONCLUSION Here, we report that TlSWO, from T. leycettanus JCM12802, is an acidic and mesophilic swollenin that has activity toward lichenan, barley β-glucan, carboxymethyl cellulose sodium and laminarin. A greater increase in glucose yield was observed when cellulose substrates were incubated with TlSWO and cellobiohydrolases. Moreover, TlSWO exhibited synergetic effects on cellobiohydrolase when using PCS and PASC as substrates. However, no significant synergistic effect was observed between TlSWO and endoglucanases, suggesting that TlSWO has better coordination with cellobiohydrolases. Compared to chemical pretreatment, biological pretreatment with enzymes has extensive research potential given the advantages of low energy consumption, environmental friendliness and lower production cost. Different lignocellulosic biomasses need different types of pretreatments because the structural features of cellulose play important roles in enzyme hydrolysis that affect outcomes such as polymerization degree, cellulose crystallization arrangement, surface area accessibility, particles size, and the existence of hemicelluloses and lignin (Bhatia et al., 2020;Sankaran et al., 2020). Our results showed that TlSWO directly alters the cellulose structure, which in turn increases its hydrolysis rate. Using optimized preconditioning and molecular design, TlSWO could be a promising additive for improving lignocellulosic biomass generation performance. AUTHOR CONTRIBUTIONS HZ and YW performed the experiments. FZ and RB designed and performed the synergism experiments and analyzed the data. BY and XX designed the research and participated in the bioinformatics analysis. FZ and HL revised the manuscript. All authors read and approved the final manuscript. ACKNOWLEDGMENTS We would like to thank the reviewers for their insightful comments on the manuscript, as their remarks led to an improvement of the work. This work has been published as a pre-print (https://www.researchsquare.com/article/rs-22793/v2) (Zhang et al., 2020).	2021-03-29T13:16:03.244Z	2021-03-29T00:00:00.000	{ "year": 2021, "sha1": "2ff80b783c0a6f1f89eaf47cc0c66b077c35ffd6", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fmicb.2021.658096/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "2ff80b783c0a6f1f89eaf47cc0c66b077c35ffd6", "s2fieldsofstudy": [ "Biology", "Environmental Science", "Chemistry" ], "extfieldsofstudy": [ "Medicine" ] }
86812177	pes2o/s2orc	v3-fos-license	Use of Programmed Multilevel Ventilation as a Superior Method for Lung Recruitment in Heart Surgery Objectives: During cardiac surgery, extracorporeal circulation (ECC) causes lung injury. In these inhomogenously affected lungs, the pressure control ventilation (PCV) cannot adequately ventilate differently damaged lung compartments. We invented and used original multilevel lung ventilation method named 3-LV based on alternating 3 or more pressure levels, ventilation frequencies and delivered tidal volumes. The goal of this article is to compare lung mechanics in cardiac surgery patient after ECC using standard PCV when compared to 3LV ventilation. Methods: This study was performed on 88 cardiac surgery patients after disconnection from ECC; the patients were randomly (allocation by weekdays) divided into two groups and ventilated by PCV and 3LV. Group 1 (n = 44) started with 1 hour PCV followed by a second hour with the 3-LV mode. Group 2 (n = 44) was ventilated in the reverse order. Measured parameters were statistically evaluated by the Student’s paired t-test. Results: The static compliance (Cst) and PaO2/FiO2 ratio in 3-LV ventilation mode improved by 25-32% (p < 0.01) and 31% (p < 0.01), in group actually ventilated by 3LV ventilation. The respiratory rate after weaning in Group 1 significantly decreased compared with that in Group 2 (p < 0.05). An improved CO2 washout was observed in each group after switching to 3LV. Conclusions: 3-LV showed a better lung recruitment ability compared with PCV in patients after cardiac surgery, without using high PEEP level. Introduction Anaesthesia and extracorporeal circulation (ECC) during prolonged cardiac surgery often cause lung injury, which is usually represented by grade 1 ARDS (PaO 2 / FiO 2 ratio = 200-300).This phenomenon is already well documented [1] and our own experience shows that the lung mechanics and transport of O 2 are significantly impaired during and after anaesthesia [2].Placing the patient in a supine position during operation increases the dead space over tidal volume ventilation (VD/ VT) by approximately 15-20%.The decrease in pulmonary compliance is 10-20% on average and up to 35% in obese persons.The geometry of the alveoli is usually impaired by their cyclical collapsing and reopening, which leads to surfactant damage [2,3].The typical EK no.VZ/7 / KardO/2011.Head of Committee Stanislav Juhas, MD, PhD.This prospective study was performed starting from February 2014 to January 2017.It comprises 88 cardiac surgery patients, with cardiac arrest lasting 50 min (79 ± 26 min) in ECC.After being disconnected from the ECC, the patients were randomly divided into two groups ventilated by alternating PCV and 3-LV techniques in different orders.Protocol: Group 1 (n = 44) starts with 1 hour of PCV followed by 3-LV mode in the second hour.Group 2 (n = 44) was ventilated in the reverse order, thus starting with 3-LV for 1 hour followed by PCV, as described in the time flow chart in Figure 1.Patients undergoing scheduled cardiac surgery were assigned.Patients undergoing operation on Monday, Wednesday and Friday were assigned to Group 1, while the ones in Group 2 underwent operation on Tuesday and Thursday.The exclusion criteria were a chronic lung disease (bronchial asthma, chronic obstructive pulmonary disease (COPD), ejection fraction of the left ventricle (LVEF) < 30% and low lung compliance at the beginning of anaesthesia with dynamic compliance (Cdyn) < 30 ml * cm H 2 O -1 .The demographic parameters can be found in Table 1. Anaesthesia: Identical anaesthesia was applied to complications on the level of alveoli are atelectasis, atelectotrauma, biotrauma, impaired ventilation-to-perfusion ratio (Qs/Qt), increased permeability of the alveolar-capillary membrane for liquids and the increase in ventilation work after disconnecting from artificial lung ventilation (ALV).The application of high concentrations of O 2 , i.e., FiO 2 > 0.6-0.9,leads to resorption atelectasis [4,5].Because of considerable effect on the alveolar geometry, this condition may lead to significant postoperative complications.In many cases, the complications mentioned above can be prevented by a reduced usage of high O 2 concentration, retaining the alveolar geometry by maintaining positive end expiratory pressure (PEEP).Applying the principles of "protective ventilation" by performing a lung recruitment manoeuvre during the postoperative period by using necessary PEEP may help.However, a new ventilation technique called programmed multilevel ventilation (PMLV), in this case, "programmed three-level artificial lung ventilation (3-LV)", could be an even better solution.3-LV is our original method developed especially for the recruitment and ventilation of inhomogeneously affected lungs, which is typical in ARDS of various origins.The pressure control ventilation (PCV) mode in ARDS appears to be less effective because even with the best frequency and pressure adjustment, it cannot achieve an optimal distribution of gases to the lung compartments, which are differently affected.Our objective was to determine the lung recruitment ability by using the 3-LV ventilation technique in comparison with the commonly used conventional PCV mode in postoperative cardiac surgery patients after weaning from ECC.In this paper, we evaluate the parameters of lung mechanics, oxygenation and postoperative adaptation to spontaneous ventilation in 3-LV compared with PCV. Study design and patients The study was approved by the ethics committee of the East Slovakian Institute of Cardiovascular Diseases: breathing was observed and afterwards compared between both groups. Explanation of basic principles of the three levels of programmed ventilation (3-LV) It is well known that in classical ALV regimes, even the best frequency and pressure optimization procedure of ALV parameters cannot optimally distribute gases to individual but differently damaged compartments [6].Hence, a single ventilation mode with the ventilation frequency (f), time of inspiration-to-expiration ratio (T I : TE), respiratory volume (V T ), positive end-expiratory pressure (PEEP), and ventilation support pressure (Ppc) in pressure controlled mode as the fixed parameters cannot optimally ventilate inhomogeneous lungs [6,7].Differently damaged compartments are dispersed in whole lungs.It is not possible to divide the flow of gases into each separate lung compartment using fixed ventilation parameters [2,5].Fixed ventilation parameters allow optimal ventilation for only some compartments of the lungs with well-suited mechanical properties.Other compartments will be ventilated more or less sub-optimally.One possible solution is to use a ventilator with programmed 3-LV to generate variable volume, pressure and time ventilation cycles.It can improve the distribution of gases into the variably damaged lung compartments.The setting of the ventilation parameters of 3-LV is derived from continuously measured lung mechanics by a computer in the Aura-V ventilator with the ProfiLungs system.The principle of pressure changes during 3-LV compared with PCV is schematically depicted in Figure 2. Statistics The acquired parameters of lung mechanics and oxygenation index in both groups were statistically evaluated by Student`s paired t-test (Microsoft Excel). Results Based on the data shown below, the 3-LV ventilation method achieved better parameters than PCV.We compared 3LV to PCV using the following parameters: Lung mechanics, oxygenation, carbon dioxide elimination and haemodynamic.For the assessment of recruitment efficiency, the adaptation to breathing after disconnecting from ALV, the oxygenation of arterial blood and Cst were evaluated.The effect of ALV on the circulatory system was also considered.After admission to the ICU, the acute lung injury score (LIS -Murray score) was in the range of 1.5 ± 0.25 points, confirming moderate lung impairment.The change in ALV methods does not cause a significant decrease in the mean arterial pressure (MAP).The average MAP before and after the application of 3LV was 79 ± 8 vs. 75 ± 6 mmHg and 83 ± 8 vs. 79 ± 7 mmHg in the first and second groups, respectively (p = NS.Cardiac output was measured using Monitor Vigileo (Arrow)), the influence both groups in the following combination: Propofol (Fresenius) at the introduction to anaesthesia at a dose of 150 ± 41 mg, a 2.5-to 5-mg bolus of Midazolam (Accord) at the introduction, 15 -24 µg of Sufentanyl (Janssen) at the introduction to anaesthesia, a continuous dose of Sufentanyl forte (Jansen) of 0.66 ± 0.12 µg * kg -1* h -1 , a 50-mg bolus of Atracurium (GSK) repeated as necessary, and a minimal alveolar concentration of Sevofluran (Abbott) of (MAC) = 0.6-1.1 or 0.6 ± 0.2 during ECC.The patients were intubated with an 8.0 mm orotracheal tube.Artificial lung ventilation (ALV) during anaesthesia was performed using the principles of protective ALV listed in Table 2. FiO 2 was maintained if possible at not higher than 0.5, and ETCO 2 was > 35 but ≤ 41 mmHg (5.6 kPa).Measured parameters: During anaesthesia and during ALV in the postoperative period, the ventilation parameters, parameters of lung mechanics, static compliance (Cst), airway resistance (Raw), oxygenation parameters, concentrations of blood gases and haemodynamic parameters were monitored.After admission to the intensive care unit (ICU), the patients were ventilated according to the study protocol.Ventilator Chirolog Servoventilator AURA-V with a computer assistance module ProfiLungs (Chirana Medical, Slovakia) was used.During ALV the changes in the parameters of lung mechanics, oxygenation index, values of ETCO 2 , and minute ventilation (MV) were monitored.The pressures applied and the extent of changes of ventilation volumes during 3-LV were also monitored.The measured values of specific tidal volume/ml * kg -1 / (VTspec) fluctuated depending on the individual transition pressure levels of the 3-LV within the ranges of 1.9-2.8ml * kg -1 , 2.9-4.9 ml * kg -1 and 5.0-7.2 ml * kg -1 in 7-8%, 60 -75% and 15 -20% of the breathing cycles, respectively.3-LV caused variable volume, pressure and time ventilation.After weaning from ventilation and extubation, the spontaneous breathing frequency as an indicator of adaptation to spontaneous vs. 321 mmHg (p < 0.01).During 3-LV, we observed in both groups an increased EtCO 2 in the first 25 minutes of using 3-LV, which indicated an increasing area of gas exchange as a sign of lung recruitment (Figure 3).The Cst was monitored every 10 minutes and showed major improvement in Group 1 compared with Group 2. (p < 0.01).In all cases when 3-LV was applied, regardless of whether it was applied for the first or second time, the Cst was improved (+ 30%) compared with PCV (p < 0.01).This effect was observed regardless of the order in which the 3-LV technique was applied (Figure 4).At 60 and 120 minutes after extubation, the breathing frequency as an indicator of adaptation to spontaneous breathing was assessed.In both cases, it was higher in Group 2 than in Group 1 (p < 0.05).In Group 1, Cst was higher than that in Group 2. The improved adaptation in Group 1 was probably caused by the lower ventilation work (Table 3). of ventilation on haemodynamics was assessed.In all cases, as an indirect indicator of peripheral perfusion and CO, we measured the central venous blood oxygen saturation (ScvO 2 ).After the patient was introduced to ALV and disconnected from extracorporeal circulation (ECC), the arterio-alveolar difference in oxygen (A-aDO 2 ) significantly increased 2 to 4-fold, (average 201 ± 42 mmHg, p < 0.01).After the first hour of ALV, the average SaO 2 and SvO 2 were within the physiological range 94-100% and 67-77%, respectively.This indicates adequate peripheral perfusion and sufficient CO.The value of the PaO 2 /FiO 2 ratio after weaning from ECC decreased significantly in both groups, to below 300 Torr, which corresponds to grade 1 ARDS (according to the Berlin classification).The PaO 2 /FiO 2 ratio after admission to the ICU was 249 ± 24 mmHg in Group 1 and 266 ± 25 mmHg in Group 2 (p = NS).ALV lasted 3.3 ± 0.4 hr and then the patient was disconnected from ALV.The PaO 2 in Group 1 vs. Group 2 was 384 mmHg Table 3: Frequency of spontaneous breathing 1 hour (panel A) and 2 hours (panel B) after disconnection from ALV in Group 1 and Group 2. A significant difference was found in the respiratory frequency of spontaneous breathing between groups 1 and 2 (p < 0.05).The frequency of spontaneous breathing was higher in Group 2 in which Cst was lower before disconnecting from a ventilator.The adaptation of patients to spontaneous inspiration within 120 minutes after extubation was faster in Group 1 than in Group 2. A prerequisite for improved adaptation is probably lower ventilation work.pressure and breathing frequency during each respiratory cycle.Non-constant respiratory volumes, inspiration times and ventilation pressures probably enhanced the opening of collapsed but recruitable compartments [12].The benefit of 3LV ventilation could be confirmed by the improvement of PaO 2 /FiO 2 and lung compliance as well as by a reduction of A-aDO 2 [13].According to Johnson [2], the increase in Cst is a marker of lung recruitment effectiveness.This finding concurs with our results.The frequency of spontaneous breathing after disconnecting from a ventilator is a good indicator of spontaneous breathing work.Increasing breathing frequency is the most significant predictor of respiratory failure [14].Therefore, we used this parameter to compare respiratory stability after disconnection from 3-LV or PCV. Conclusion 1. We compared the effectiveness of PCV with the 3-LV ventilation method in postoperative cardiac surgery patients with ECC.Based on the monitored Cst, PaO 2 / FiO 2 , EtCO 2 we assessed the lung recruitment, impact on circulation and spontaneous breathing stability. 2. The application of 3-LV compared with PCV during postoperative period improved arterial blood oxygenation PaO 2 /FiO 2 , the elimination of CO 2 , the stability of alveolar geometry (Cst) and the transition from artificial ventilation to spontaneous breathing. 3. The advantage of using 3-LV is effective lung recruitment without using a high PEEP.The mean airway pressure value was comparable with PCV group. 4. In the cases of inhomogeneous distribution of gases in the lungs, for example, ARDS, SARS -(Severe Acute Respiratory Syndrome), H1N1 pneumonia, and lung oedema, 3-LV could be successfully used for lung recruitment and the improvement of gas distribution in such damaged lungs. 5. The effectiveness of 3-LV compared with classic ventilation methods (PCV, PS, SIMV, etc.) will probably be higher. Discussion It is well known that use of ECC during cardiac surgery is followed by substantial signs of lung impairment.The PaO 2 /FiO 2 ratio decreases to values indicating the presence of grade 1 ARDS [1].The acute lung injury score (LIS) after the surgical procedure fluctuated on average at 1.5 points, which indicates moderate lung impairment.The mechanism of impairment is multifactorial [1].During anaesthesia, there is a decline in A-aDO 2 associated with moderate hypoxemia and caused by the increased disproportion of ventilationperfusion due to the decreased ventilation of perfused parts of lungs [2].A major disproportion of ventilationperfusion was observed after weaning from ECC. Seventy percent of resorption atelectases arise predominantly in dependent lung zones, which are correlated with the number of hospitalization days [3].The decrease in Cst in all monitored groups of patients after disconnection from ECC is a sign of development of atelectasis and impairment of alveolar geometry [3].Mols, et al. [5] define atelectasis as a condition of the absence of gas in an alveolus causing its long-term collapse.Atelectotrauma as a cyclically repeated collapse and filling (opening) of the alveolus during a ventilation cycle.Alveoli open during inspiration and close during expiration.Alveolar recruitment is characterized by the re-opening of a previously collapsed alveolus, whereas derecruitment is the collapse of an alveolus previously opened by recruitment.Ongoing recruitment (lasting recruitment) is defined as a condition of permanent alveolar recruitment, which does not allow derecruitment.From a mechanical-pneumatic point of view, it is the opposite of atelectasis. All groups of our patients demonstrated a decrease in Cst after being disconnected from ECC as a sign of atelectasis.Our opinion that early recruitment using adequate PEEP or another lung recruitment technique (in our case, 3-LV) has a beneficial effect was already described by Borges, et al. [8].With increasing the FiO 2 value we are able to improve the PaO 2 , but this may cause absorption atelectasis and hyperoxia; therefore, we did not use it [4,9].The high FiO 2 may have an adverse effect on CO and peripheral vascular resistance (PVR) [10].We applied this concentration of O 2 to maintain PaO 2 in a recommended range of 75 -160 Torr, (10.0 -21 kPa) [11].It is well known that increased FiO 2 and PaO 2 in ICU patients increases mortality compared with patients with adequate PaO 2 [9,11] .Peak pressure applied during 3-LV had a minimal impact on circulation if adequate PEEP was used.During 3-LV with the application of adequate PEEP and the minimum possible peak airway pressure (Paw max), a moderate or insignificant decrease in MAP and the heart rate was observed.Similarly, no significant desaturations of venous blood as an indicator of a decrease in cardiac output (CO) was observed during 3-LV.Lung recruitment by 3-LV is probably achieved by irregular changes in respiratory volume, Figure 1 : Figure 1: Time flow chart of study from start of anaesthesia to disconnection from ventilation.Average time of anaesthesia: 283 ± 47 min, PCV mode in Group 1: 60 minutes and 3LV mode in Group 2: 60 minutes.Next step 3LV mode in Group 1 and PCV mode in Group 2. Next step PSV mode in both groups: Approx 65 ± 18 min.Next step "T" trial and extubation.Measurement of the frequency of spontaneous breathing during the subsequent two hours. Figure 2 : Figure 2: Panel A: Diagram of the pressure curves in 3 level ventilation mode (3-LV).P -Pressure; Tcy PCV(PSV) -Time of Ventilation Cycle of PCV (PSV); Tcy PEEP high-Time of Ventilation Cycle of PEEPhigh; Te -Expiratory Time; Ti -Inspiratory Time; t-Time in Sec.Panel B: Diagram of the pressure curves in pressure control ventilation mode (PCV/PSV).P -Pressure; Ti -Inspiratory Time; Te -Expiratory Time; Tcy PCV(PSV) -Time of Ventilation Cycle of PCV (PSV); t -Time in Sec. Figure 3 : Figure 3: Changes in ETCO 2 during 3-LV and PCV in Group 1 and Group 2 during 2 × 60 minutes of ventilation. Figure 4 : Figure 4: Cst -static compliance changes in patients in Groups 1 and 2. Before ECC, After ECC, admission to intensive care unit (ICU), 1 hour after admission to ICU, Switch to another ventilation mode, and 2 hours after ICU (1 hour after switching of ventilation mode). spontaneous breathing (breath * min -1 ) 1 hour after disconnection from ALV The frequency of spontaneous breathing (breath * min -1 ) 2 hours after disconnection from ALV Table 1 : General demographic parameters as well as numbers of cardiac surgery performed by type: CABG, coronary artery bypass graft, AVR aorta valve replacement, MVR mitral valve replacement or ring, and TVR tricuspid valve replacement or ring.Surgical procedures are mostly combined.	2019-03-28T13:33:48.203Z	2019-06-30T00:00:00.000	{ "year": 2019, "sha1": "26aff57a270f11c7cf036c1971f611b489ecdce1", "oa_license": "CCBY", "oa_url": "https://www.clinmedjournals.org/articles/ijccem/international-journal-of-critical-care-and-emergency-medicine-ijccem-5-067.pdf?jid=ijccem", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "26aff57a270f11c7cf036c1971f611b489ecdce1", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
17883034	pes2o/s2orc	v3-fos-license	Pharmacokinetic interaction between udenafil and dapoxetine: a randomized, open-labeled crossover study in healthy male volunteers Background “Udenafil” is a phosphodiesterase-5 inhibitor indicated for erectile dysfunction. “Dapoxetine” is a serotonin transport inhibitor indicated for premature ejaculation. The aim of the study reported here was to investigate the pharmacokinetic drug interaction between udenafil and dapoxetine in healthy male subjects. Methods An open-label, three-treatment, six-sequence, three-period crossover study was performed in healthy male subjects. In varying sequences, each subjects received single oral doses of udenafil 200 mg, dapoxetine 60 mg, and both treatments. The periods were separated by a washout period of 7 days. Serial blood samples were collected up to 48 hours after dosing. The plasma concentrations of udenafil and dapoxetine were determined using a validated liquid chromatography-tandem mass spectrometry method. Pharmacokinetic parameters were obtained by non-compartmental analysis. Tolerability was assessed throughout the study. Results Twenty-three healthy subjects completed the study. The geometric mean ratios of the area under the plasma concentration–time curve from time 0 to last measurable time point and measured peak plasma concentration for udenafil were 0.923 (90% confidence interval [CI]: 0.863–0.987) and 0.864 (90% CI: 0.789–0.947), respectively. The geometric mean ratios of the area under the plasma concentration–time curve from time 0 to last measurable time point and measured peak plasma concentration for dapoxetine were 1.125 (90% CI: 1.044–1.213) and 0.837 (90% CI: 0.758–0.925), respectively. There were no serious adverse events reported, and none of the subjects dropped out due to adverse events. Conclusion Udenafil was found to have no clinically significant pharmacokinetic interactions with dapoxetine. The concurrent administration of udenafil and dapoxetine was generally well tolerated. Introduction Erectile dysfunction (ED) and premature ejaculation (PE) are the two most prevalent male sexual dysfunctions. 1,2 "ED" is defined as the inability to achieve or maintain an erection sufficient for satisfactory sexual performance, with a prevalence of approximately 5%-30%. 3,4 "PE" is defined as earlier-than-desired ejaculation resulting from minimal stimulation that causes bother or distress, with a prevalence of approximately 20%-30%. [5][6][7] ED and PE may be comorbid conditions in some men. 8 A large survey that included 12,134 men from the USA, Germany, and Italy reported that 7.2% of men met the criteria for both ED and PE. Overall, 44% of men with ED also reported PE, whereas 32% of men with PE also reported ED. 9 "Udenafil" is an oral phosphodiesterase (PDE)-5 inhibitor for the treatment of ED. It is absorbed with time to reach peak concentration (t max ) at 0. 8 Kim et al declined mono-exponentially with a terminal half-life (t 1/2β ) of approximately 7.3-12.1 hours. 10 The absolute oral bioavailability in humans is not known but is 38.0%-55.6% in rats. 11 It is metabolized primarily by cytochrome P450 (CYP) 3A4 into its active metabolite, DA-8164, which has approximately half the pharmacological activity of the parent compound 12 (Figure 1). A meta-analysis of five trials involving 1,109 patients showed that the change from baseline in International Index of Erectile Function erectile-function domain score in the udenafil group was significantly greater than in the placebo group (mean difference 5.65, 95% confidence interval [CI] 4.41-6.89). 13 "Dapoxetine" is a short-acting selective serotonin reuptake inhibitor marketed for the treatment of PE. It is absorbed with a t max of 1.0-1.3 hours, and elimination is biphasic, with an initial half-life of approximately 1.4 hours and a t 1/2β of approximately 20 hours. 14 Dapoxetine is metabolized by multiple CYP isoenzymes including CYP3A4 and CYP2D6 to its active metabolite, desmethyl dapoxetine, which has similar pharmacological potency to the parent compound ( Figure 1). 15 An integrated analysis of five trials involving 6,081 patients showed that the PE profile measures improved significantly with dapoxetine as compared with placebo. 16 Patients receiving udenafil for the treatment of ED could also potentially receive dapoxetine for the treatment of PE, and both of these molecules undergo CYP3A4 metabolism. Thus, the potential for interactions must be evaluated. The objective of this study was to investigate the pharmacokinetic interactions and tolerability of udenafil and dapoxetine in healthy volunteers. Materials and methods subjects Healthy male volunteers aged 20-45 years and with a body mass index of 19-27 kg/m 2 were eligible for this study. Volunteers were considered to be in good health based on medical history, physical examinations, vital-sign measurements (blood pressure, heart rate, and body temperature), 12-lead electrocardiograms (ECGs), clinical laboratory tests (hematology, blood chemistry, and urinalysis), serology (hepatitis B surface antigen, hepatitis C virus antibody, and HIV antigen/antibody), and urine drug screening (for use of amphetamine, methamphetamine, barbiturate, cocaine, opiate, benzodiazepine, cannabinoid, and methadone) within 4 weeks before the first administration of the study drug. All subjects with known allergy or hypersensitivity to udenafil or dapoxetine, or with a history of drug abuse were excluded from the study. study design The study was designed as a randomized, open-label, singledose, three-treatment, three-period, six-sequence, crossover clinical trial. Subjects were randomly assigned to one of six sequences and received three different treatments: udenafil 200 mg (Treatment A), dapoxetine 60 mg (Treatment B), and co-administration of udenafil 200 mg and dapoxetine • • 1211 Pharmacokinetic interaction between udenafil and dapoxetine 60 mg (Treatment C). All treatments were given under fasting state with 240 mL of water. After the drug administration, the subjects were required to fast for 4 hours. Following a 1-week washout interval, subjects received alternate formulations ( Figure 2). For each treatment period, subjects were admitted in the Clinical Trial Center (CTC) at the Asan Medical Center (AMC) from Day1 through Day 2 (24 hours after dosing). On Days 2 (32 hours after dosing) and 3 (48 hours after dosing), subjects visited the CTC to assess the tolerability and pharmacokinetics of udenafil or dapoxetine. The schedule for the second and third treatment period procedures was the same for the first period. Follow-up visits were performed within 5 to 9 days after the last treatment. Tolerability was assessed throughout the study using vital-sign measurements, 12-lead ECGs, clinical laboratory tests (hematology, blood chemistry, and urinalysis), physical examinations, and monitoring of adverse events (AEs). AEs were recorded in terms of symptoms and signs, duration, intensity, relationship to the study drug, action taken, outcome, and seriousness. The study protocol was approved by the Ministry of Food and Drug Safety and the institutional review board of the AMC, Seoul, Republic of Korea. The study was conducted at the CTC of the AMC from September to November 2013. All subjects provided written informed consent before screening tests. The trial was registered with the identifier number NCT01928563 at ClinicalTrials.gov. Determination of udenafil and DA-8164 concentrations Plasma concentrations of udenafil and its active metabolite DA-8164 were determined using a validated ultraperformance liquid chromatography (UPLC) coupled with tandem mass spectrometry (MS) method, with the internal standards Udenafil-d 7 and DA-8164-d 3 , respectively. The sample extracts were analyzed using an Acquity UPLC ® System (Waters Corporation, Milford, MA, USA) and an Acquity UPLC BEH (ethylene bridged hybrid) C18 column (1.7 μm, 50.02.1 mm; Waters Corporation) with the mobile phase consisting of distilled water with 0.1% formic acid and methanol with 0.1% formic acid (50:50, v/v). Using this assay, the accuracy of the calibration standard curve for udenafil was between 94.5% and 104.0%, and the coefficient of variation (CV) of the back-calculated concentration was 2.4%. The accuracy of the calibration standard curve for DA-8164 was between 98.5% and 102.0%, and the CV of the back-calculated concentration was 2.3%. Determination of dapoxetine and desmethyl dapoxetine concentrations Plasma concentrations of dapoxetine and its active metabolite desmethyl dapoxetine were determined using a validated UPLC-tandem MS method, with the internal standards dapoxetine-d 4 and desmethyl dapoxetine-d 7 , respectively. The sample extracts were analyzed using the Acquity UPLC System and an Acquity UPLC BEH C18 column (1.7 μm, 50.02.1 mm; Waters Corporation) with the mobile phase A Quattro Premier™ XE MS system (Waters Corporation) was operated in positive-ion electrospray mode with multiple-reaction monitoring. For dapoxetine and desmethyl dapoxetine, the precursor-to-production reactions monitored were m/z 306.20 → 157.02 and 292.17 → 157.04, respectively. Calibration curves covered the concentration range of 5.0-2,000.0 ng/mL dapoxetine (R 2 0.995) and 0.25-100.0 ng/mL desmethyl dapoxetine (R 2 0.995). Using this assay, the accuracy of the calibration standard curve for dapoxetine was between 97.5% and 102.0%, and the CV of the back-calculated concentration was 2.9%. The accuracy of the calibration standard curve for desmethyl dapoxetine was between 97.0% and 103.2%, and the CV of the back-calculated concentration was 6.6%. Pharmacokinetic assessment and statistical analysis The plasma concentration-time profiles of udenafil, dapoxetine, DA-8164, and desmethyl dapoxetine of each subject were analyzed by a non-compartmental method using WinNonlin ® software (v 6.1; Pharsight Corporation, Mountain View, CA, USA). All analyses were made using actual times of sampling. The peak plasma concentration (C max ) and t max were determined from the observed values. The terminal elimination rate constant (λz) was estimated by linear regression of the terminal log-linear portion of the plasma concentration-time curves. The area under the time-concentration curve from time 0 to the last measurable time (AUC last ) was calculated by the trapezoidal rule and the area under the time-concentration curve extrapolated to infinity (AUC 0~∞ ) was obtained AUC last + C last/λz (C last : the last quantifiable concentration). The apparent oral clearance (CL/F) was obtained as dose/AUC 0~∞ . The t 1/2β was calculated for each participant as ln(2)/λz. To evaluate the metabolites of each component, DA-8164 and desmethyl dapoxetine, the metabolic ratio was calculated according to the ratio of the metabolite AUC last to the parent AUC last . All statistical analyses were performed using SAS ® software (v 9.3; SAS Institute, Cary, NC, USA) and Phoenix ® WinNonlin (v 6.1; Pharsight Corporation). Demographic data and pharmacokinetic parameters were summarized using descriptive statistics. For the comparison of pharmacokinetic characteristics between monotherapy of udenafil or dapoxetine and co-administration of udenafil and dapoxetine, C max , AUC last , AUC 0~∞ , CL/F, t 1/2β , and the metabolic ratio of each formulation were log-transformed and tested by a mixed-model analysis of variance. The mean differences and 90% CIs were back-transformed to obtain geometric mean ratios and CIs for those ratios. In addition, paired t-tests or Wilcoxon signed-rank tests were also used to compare the pharmacokinetic parameters. study participants A total of 25 healthy Korean volunteers were enrolled, and 23 volunteers were administered the study drugs and completed the study. Two subjects were dropped by the principal investigator: one subject showed abnormal lab result on Day1 and the other subject experienced an AE before drug administration. The mean (standard deviation) age of study participants was 27.65±4.54 years, the mean weight was 67.06±7.76 kg, and the mean height was 173.79±4.87 cm. Pharmacokinetic analysis To evaluate the pharmacokinetic drug-drug interactions between udenafil and dapoxetine, the pharmacokinetic profiles of udenafil, dapoxetine, DA-8164, and desmethyl dapoxetine were separately assessed. Figure 3 shows the plasma concentration-time profiles for udenafil and DA-8164. Figure 4 shows the plasma concentration-time profiles for dapoxetine and desmethyl dapoxetine. Of the 23 subjects who completed the study, one subject was excluded from all pharmacokinetic analysis due to vomiting after co-administration of udenafil and dapoxetine, and two subjects were excluded from the pharmacokinetic analysis of dapoxetine due to vomiting after the administration of dapoxetine. Thus, 22 subjects were included in the pharmacokinetic analysis of udenafil, and 20 subjects were included in the pharmacokinetic analysis of dapoxetine. The pharmacokinetic profile of udenafil was similar when it was administered alone and when it was co-administered with dapoxetine. It was rapidly absorbed, with a t max of 1.0-2.0 hours, and then declined with a t 1/2β of 10.81-11.02 hours. Plasma concentration of dapoxetine peaked at 1.0 hours, and elimination was rapid and biphasic, with a t 1/2β of 15.34-16.23 hours. The mean pharmacokinetic properties and the geometric mean ratios (combined/monotherapy) and 90% CIs of the AUC last and C max for udenafil, DA-8164, dapoxetine, and desmethyl dapoxetine are shown in Tables 1 and 2 Notes: Data presented as means ± standard deviation, except for t max , for which median (min, max) is shown; a geometric mean ratio of pharmacokinetic parameters = dapoxetine with udenafil combined therapy/dapoxetine monotherapy; b metabolic ratio = metabolite aUc last /parent aUc last ; c determined using paired t-test except for t max , for which Wilcoxon signed-rank test was used. Abbreviations: aUc last , area under the plasma concentration-time curve from time 0 to last measurable time point; AUC 0~∞ , area under the plasma concentration-time curve from time 0 to infinity; CI, confidence interval; C max , measured peak plasma concentration; CL/F, oral clearance; t 1/2β , terminal half-life; t max , time to reach peak concentration; h, hours. Tolerability Overall, udenafil and dapoxetine were well tolerated when administered alone or concomitantly. Seventeen subjects experienced a total of 69 AEs, among which 63 events in 16 subjects were considered "possibly related" to the study drug (Table 3). All AEs were mild or moderate in severity, and resolved without sequelae. No death or serious AE occurred during the entire course of the study. Likewise Note: Data presented as number of subjects (number of events). there were no clinically significant abnormalities in laboratory tests, physical examinations, or ECGs. Discussion The study reported here investigated the potential for a pharmacokinetic interaction between udenafil and dapoxetine in healthy male subjects. In the study, a 200 mg dose of udenafil and 60 mg dose of dapoxetine were selected because they are the highest recommended dosage. Generally, these drugs are used on-demand; for this reason, this study was conducted after a single oral administration of each drug. 2 In this study, the primary pharmacokinetic parameters were the C max and AUC last of udenafil and dapoxetine. The C max of udenafil and dapoxetine after the co-administration of udenafil and dapoxetine was slightly decreased compared with the administration of udenafil or dapoxetine alone. In terms of the AUC last , udenafil did not affect the pharmacokinetics of dapoxetine, nor did dapoxetine affect the pharmacokinetics of udenafil. In addition, the other pharmacokinetic parameters of udenafil, dapoxetine, and their metabolites were also within 0.80-1.25 regardless of whether the drugs were administered alone or in combination. The exceptions were the t max of udenafil and DA-8164, a metabolite of udenafil. After the co-administration of udenafil and dapoxetine, the t max of udenafil and DA-8164 was prolonged by approximately 1 hour (from 1.0 and 2.0 hours) and 1.5 hours (from 2.0 and 3.5 hours), respectively. The slight reduction in C max and delay in t max indicate a decreased drug absorption rate of udenafil by dapoxetine. To evaluate these differences, the absorption rate constant (k a ) of udenafil was obtained from a two-compartment structural model with first-order absorption using WinNonlin software. The k a of udenafil was 0.84±0.29 hours -1 and 0.50±0.17 hours -1 after monotherapy and combined therapy, respectively. These values are similar to a previously reported k a value (0.70±0.26 hours -1 ). 17 As metabolism via CYP3A4 is the major elimination pathway for udenafil and one of the elimination pathways for dapoxetine, all inducers or inhibitors of CYP3A4 have the potential to interfere with systemic exposure and metabolism. 12,15 A previous study reported that "ketoconazole", a known CYP3A4 inhibitor, affects the pharmacokinetics of udenafil; the mean C max and AUC last of udenafil increased 1.9-fold and 3.2-fold, in the co-admiministration of ketoconazole, respectively. The metabolic area under the concentration-time curve (AUC) ratio was 1.71 when udenafil was administered alone, and the value decreased to 0.19 when udenafil was dosed in combination with ketoconazole. 18 Ketoconazole also increased dapoxetine exposure to a greater extent. There was a 23% increase in C max and 88% increase in the AUC of active moiety and a 35% and 99% increase in parent drug C max and AUC, respectively. 15 However, other drug interaction studies on dapoxetine and other PDE-5 inhibitors, including tadalafil and sildenafil, have reported no clinically significant pharmacokinetic interactions. 19 The most common AEs were nausea and dizziness in the dapoxetine alone and dapoxetine with udenafil groups. These are well-known AEs of dapoxetine. 20 Dapoxetine is absorbed rapidly and decreases to approximately 5% of its peak concentration by 24 hours after drug administration. 14 Likewise, these AEs were started about 1 hour after drug administration and resolved within next day. The findings from this study support the combined use of udenafil and dapoxetine for the treatment of PE and ED. Publish your work in this journal Submit your manuscript here: http://www.dovepress.com/drug-design-development-and-therapy-journal Drug Design, Development and Therapy is an international, peerreviewed open-access journal that spans the spectrum of drug design and development through to clinical applications. Clinical outcomes, patient safety, and programs for the development and effective, safe, and sustained use of medicines are a feature of the journal, which has also been accepted for indexing on PubMed Central. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials.php to read real quotes from published authors. Drug Design, Development and Therapy 2015:9 submit your manuscript \| www.dovepress.com Dovepress Dovepress Dovepress 1216 Kim et al Therefore, further long-term studies in patients are needed to evaluate the tolerability and clinical efficacy of combination treatment with udenafil and dapoxetine. Conclusion Our study indicates that udenafil has no significant pharmacokinetic interactions with dapoxetine. Dapoxetine did not affect the pharmacokinetics of udenafil. The concurrent administration of udenafil and dapoxetine was generally well tolerated.	2016-05-04T20:20:58.661Z	2015-02-23T00:00:00.000	{ "year": 2015, "sha1": "39c542e1e8eef8d344550d563e8d49ae4794ea88", "oa_license": "CCBYNC", "oa_url": "https://www.dovepress.com/getfile.php?fileID=23883", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "36c581e7f415c295a77fcebf78319936253ef9ce", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
17151586	pes2o/s2orc	v3-fos-license	Role of endoplasmic reticulum Ca2+ signaling in the pathogenesis of Alzheimer disease Alzheimer disease (AD) is a major threat of twenty-first century that is responsible for the majority of dementia in the elderly. Development of effective AD-preventing therapies are the top priority tasks for neuroscience research. Amyloid hypothesis of AD is a dominant idea in the field, but so far all amyloid-targeting therapies have failed in clinical trials. In addition to amyloid accumulation, there are consistent reports of abnormal calcium signaling in AD neurons. AD neurons exhibit enhanced intracellular calcium (Ca2+) liberation from the endoplasmic reticulum (ER) and reduced store-operated Ca2+ entry (SOC). These changes occur primarily as a result of ER Ca2+ overload. We argue that normalization of intracellular Ca2+ homeostasis could be a strategy for development of effective disease-modifying therapies. The current review summarizes recent data about changes in ER Ca2+ signaling in AD. Ca2+ channels that are discussed in the current review include: inositol trisphosphate receptors, ryanodine receptors, presenilins as ER Ca2+ leak channels, and neuronal SOC channels. We discuss how function of these channels is altered in AD and how important are resulting Ca2+ signaling changes for AD pathogenesis. INTRODUCTION Calcium (Ca 2+ ) is one of the most important second messengers in the nervous system. Ca 2+ -mediated signal transduction connects membrane excitability and biological functions of neurons ranging from proliferation, secretion, gene expression, ATP production, cell death to memory formation and its loss. Acting at the border of electrical and signaling "worlds" of the cell, Ca 2+ -permeable channels play a major role in many key aspects of neuronal functions. Due to the huge importance of the calcium as the second messenger neurons utilize many approaches to regulate intracellular Ca 2+ content, mainly via local signal transduction pathways. Neuronal Ca 2+ influx can be maintained by different Ca 2+ -permeable channels, such as voltage-gated Ca 2+ channels of plasma membrane, N -methyl-D-aspartate (NMDA) receptors, αamino-3-hydroxy-5-methyl-4-isoxazolepropanoic acid (AMPA) receptors, nicotinic receptors, store-operated Ca 2+ channels (SOC). Ca 2+ can also be released from intracellular stores of endoplasmic reticulum (ER) via inositol-1,4,5-trisphosphate receptors (InsP3R) and ryanodine receptors (RyanRs). Mitochondria also play an important role in intracellular Ca 2+ handling. Neurons are highly susceptible to any changes in intracellular Ca 2+ concentrations: insufficient intracellular Ca 2+ content lead to abnormal functioning of neurons, whereas excessive Ca 2+ levels cause cell death (Berridge, 1998). Therefore, even small fluctuations in Ca 2+ content can be very detrimental over long life of a neuron (Khachaturian, 1989). Alzheimer disease (AD) is the threat of twenty-first century that is responsible for the majority of senile dementia. AD progresses slowly and affects neurons in the brain. Currently there are two main proteins whose dysfunctions and accumulation in the brain are correlated with the disease progress. The first is 40-42 long beta-amyloid (Aβ) peptides that constitute a major part of neuritic plaques and cause excessive neurotoxicity. These peptides are cleaved from the amyloid precursor protein (APP) by βand γ-secretases (Hardy and Selkoe, 2002). The second protein is tau whose hyperphosphorylation results in misfolding and forming of proteolysis-resistant neurofibrillar tangles (NFTs). Aβ40, Aβ42, and NFT are synaptotoxic to neurons and facilitate cell death (Small and Duff, 2008). Although the exact mechanism how Aβ40, Aβ42, and NFT mediate AD pathogenesis is not fully understood, there are observations that link Aβ42 accumulation with elevated Ca 2+ levels in neuronal cytoplasm in vivo (Kuchibhotla et al., 2008). It has been shown that oligomers of Aβ is able to make Ca 2+ permeable channels in plasma membrane of neurons, therefore directly affecting intracellular Ca 2+ concentration (Arispe et al., 1993). Recent publications state that soluble oligomeric form of Aβ42 potentiate Ca 2+ liberation from the ER through the stimulated production of inositol trisphosphate (Demuro and Parker, 2013) and by stimulating synaptic mGluR5 receptors (Renner et al., 2010). There is another line of evidence coming from mouse models harboring presenilin's mutations that AD-like symptoms and synaptic dysfunction can occur due to Ca 2+ accumulation in the ER in the absence of Aβ pathology (Stutzmann et al., 2004;Chakroborty et al., 2009;Zhang et al., 2010b). Early changes in intraneuronal Ca 2+ regulation are common observations in AD patients (Emilsson et al., 2006;Stutzmann, 2007;Bezprozvanny and Mattson, 2008). All these observations support calcium hypothesis of AD. This hypothesis was first formulated in 1987 by Dr. Zaven Khachaturian who proposed that sustained changes in intracellular calcium homeostasis provide the final common pathway for AD and age-associated brain changes (Khachaturian, 1987). Since that time many advances in our understanding of Ca 2+ signaling in AD have been obtained. New Ca 2+ permeable channels have been identified, some of them directly linked to AD. For example, it has been demonstrated that presenilins encode passive ER Ca 2+ leak channels (Tu et al., 2006) and a novel Ca 2+ channel called Ca 2+ homeostasis modulator 1 (CALHM1) has been linked to late-onset AD by genetic evidence (Dreses-Werringloer et al., 2008). However, as it usually happens with new findings, the existence of these novel Ca 2+ channels and their role in AD has been challenged. The main purpose of the current paper is to review recent publications in the field of ER Ca 2+ signaling in the context of AD pathology. We will review the role of two well accepted ER Ca 2+ channels that release Ca 2+ out of the neuronal ER -InsP 3 R and RyanR. We will also discuss new findings about the role of presenilins and neuronal SOC in neuronal function. Our focus will be on potential role of these channels in AD pathology and as targets for development of disease-modifying therapies. INOSITOL TRISPHOSPHATE RECEPTORS The first observation of exaggerated InsP3R-mediated Ca 2+ release from ER in fibroblasts from AD patients has been obtained even before the identification of presenilins (Ito et al., 1994). It was later shown that these fibroblasts (from patients AG06840 and AG06848) harbor A246Q mutation in PSEN1 (description in Coriell Institute Cell Database). The studies with fibroblasts taken from PS1-M146V knockin mice and with Xenopus oocytes expressing human presenilin proteins 1 and 2 (PS1 and PS2) mutant constructs showed an upregulation of InsP3R-mediated Ca 2+ release (Leissring et al., 1999a(Leissring et al., ,b, 2000. Experiments in cortical neurons using whole-cell patch clamp and rapid Ca 2+ imaging in brain slices from mutant PS1-M146V mice also demonstrated almost threefold exaggeration of ER Ca 2+ liberation by photolysis of caged InsP3 and accompanying enhancement of Ca 2+ -evoked outward membrane currents (Stutzmann et al., 2004). Similar results of enhanced InsP3-evoked Ca 2+ signals were observed in 3xTg-AD mice (Stutzmann et al., 2006). Important to note that the Ca 2+ disturbances were already observed in the 3xTg-AD mice at the age of 4-6 weeks that precedes appearance of Aβ plaques and NFTs by several months (Oddo et al., 2003). Later on it has been reported that in non-neuronal DT40 and Sf9 cell models familial AD (FAD) associated mutations PS1-M146L and PS2-N141I interact with InsP3R and exert stimulatory effects on its gating activities (Cheung et al., 2008). In more recent study the same group has proposed that stimulation of InsP3R gating by expression of mutant PS1-M146L in DT40 and PC12 cells results in generation of reactive oxygen species (ROS; Muller et al., 2011). Authors report that exaggerated Ca 2+ signaling through InsP3R-PS interaction and generation of ROS may contribute to the pathology of AD (Muller et al., 2011). Important to note the recent study showing that intracellular application of Aβ oligomers into Xenopus oocytes stimulates G-protein-mediated InsP3 production and consequent Ca 2+ release from the ER, that is cytotoxic (Demuro and Parker, 2013; depicted in Figure 1). Also, it was reported that Aβ oligomers stimulate synaptic mGluR5 receptors linked with InsP3 production (Renner et al., 2010). Although detrimental effect of Aβ oligomers on neurons has been extensively studied and many publications demonstrated that Aβ aggregates promote the increase in cytosolic Ca 2+ content of neurons (Walsh et al., 2002;Demuro et al., 2005Demuro et al., , 2010Deshpande et al., 2006;Simakova and Arispe, 2007;Bezprozvanny and Mattson, 2008;Green and LaFerla, 2008;Kuchibhotla et al., 2008), the exact mechanism how Aβ contributes to disruption of Ca 2+ signaling is not known. Therefore, the studies of Demuro and Parker (2013) and Renner et al. (2010) could potentially provide a connection between amyloid and overactivation of InsP3R-mediated Ca 2+ signals. RYANODINE RECEPTORS AND EFFECTS OF DANTROLENE Ryanodine receptors are expressed in soma, proximal dendrites as well as in distal processes and spines. RyanRs activity is enhanced in dendrites and synaptic spines from presymptomatic 3xTg-AD and TASTPM (APPsw; PS1.M146V; Howlett et al., 2004) AD mice (Goussakov et al., 2010). RyanR-mediated Ca 2+ -induced Ca 2+ release (CICR) in 3xTg-AD mice is exaggerated in response to synaptic stimulation, including NMDAR-mediated Ca 2+ influx (Goussakov et al., 2010). These authors proposed that enhanced synaptic CICR may alter synaptic function and may be recognized FIGURE 1 \| Proposed role of ER Ca 2+ signaling in the pathogenesis of AD. The cartoon represents the Ca 2+ hypothesis of AD that places presenilins (PS) in the center of AD pathogenesis. Amyloid oligomers stimulate InsP3R-mediated Ca 2+ release from ER by activating synaptic mGluR5 receptors and by stimulating G-protein-mediated InsP3 production. Our laboratory has shown that presenilins function as ER Ca 2+ leak channels and FAD associated PS mutations disrupt this function, causing overloading of ER with Ca 2+ . Similar ER Ca 2+ overload occurs as a result of neuronal aging process. The first physiological response to ER Ca 2+ elevation is compensatory increase in expression and/or activity of inositol trisphosphate receptors (InsP3R) and ryanodine receptors (RyanRs). The second response to ER Ca 2+ overload is reduction in store-operated Ca 2+ (SOC) entry, a mechanism involved in refilling of ER Ca 2+ stores (mediated by Orai and TRP channels). We hypothesize that these initially compensatory and protective mechanisms of ER Ca 2+ signaling become pathogenic in aging neurons and eventually lead to synaptic dysfunction, synaptic loss and neurodegeneration. Frontiers in Molecular Neuroscience www.frontiersin.org as an early pathogenic factor in AD (Goussakov et al., 2010). Increased levels of RyanR are at least partially responsible for enhanced CICR in AD neurons. Increased expression of RyanR has been described in human AD cases and in patients with mild cognitive impairment (MCI; Kelliher et al., 1999;Bruno et al., 2012). Elevated RyanR2 expression, cognitive decline, and synaptic loss observed in MCI patients are mirrored by an increase in RyanR2 expression and Ca 2+ release in presymptomatic AD mice (Kelliher et al., 1999;Stutzmann et al., 2006;Chakroborty et al., 2009;Zhang et al., 2010b). Recently, it has been suggested that increased RyanR expression at early stages of AD might play a role as a compensatory mechanism to stabilize the preexisting synaptic deficits and normalize the depressed synaptic network (Chakroborty et al., 2012b). Similar idea of elevated RyanR3 expression as a neuroprotective response to Aβ1-42 toxic effects has been suggested before . Several studies addressed the role of RyanR in the context of AD by using pharmacological agent dantrolene. Dantrolene is an antagonist of the RyanR and is used clinically to treat malignant hyperthermia, neuroleptic malignant syndrome, and muscle spasms (Krause et al., 2004;Inan and Wei, 2010). In the first study the dantrolene was administered to 3xTg-AD mice by intracerebroventricular (ICV) injection for 3 months using an Alzet intracranial ventricular infusion system and then subcutaneously three times per week for 8 month (Peng et al., 2012). The authors state that dantrolene treatment significantly reduced both memory deficits tested by Morris water maze test and amyloid plaque load in the hippocampus in 13-month-old 3xTg-AD mice (Peng et al., 2012). The second work performed sub-chronically short-term (4 weeks) treatment of AD models (3xTg-AD and TASTPM) with dantrolene (Chakroborty et al., 2012a). Using two-photon Ca 2+ imaging and patch clamp recordings authors showed that dantrolene treatment normalized ER Ca 2+ signaling within somatic and dendritic compartment in early and late-stage AD mice in hippocampal slice experiments (Chakroborty et al., 2012a). The third study (Oules et al., 2012) was performed with transgenic mice expressing human APPswe mutation (Tg2576). These authors observed that dantrolene treatment diminished Aβ load, reduced histological lesions, and slowed down learning and memory deficits in Tg2576 mice (Oules et al., 2012). These studies suggested that inhibition of RyanR with dantrolene may exert beneficial effects in the context of AD pathology. However, opposite conclusion was obtained by our laboratory in experiments with APPPS1 transgenic mouse model (Thy1-APPKM670/671NL, Thy1-PS1L166P; Zhang et al., 2010b). In these studies we discovered that long-term (starting at 2 months of age) oral feeding of dantrolene exacerbated plaque formation and resulted in loss of hippocampal synaptic markers and neuronal deterioration in 8-month-old APPPS1 mice (Zhang et al., 2010b). How can these seemingly divergent observations that center on dantrolene be explained? It is difficult to directly compare these results due to different routes of dantrolene administration used in the studies, variability in duration of treatments, mice age groups, and different AD mouse models used in the studies. Another potential problem with interpreting these results is that specific RyanR inhibitors do not exist and the drug dantrolene used in most studies has additional targets such as store-operated Ca 2+ channels (Zhao et al., 2006). Moreover, dantrolene is specific for skeletal muscle RyanR1 (Krause et al., 2004), and does not block neuronal RyanR2 and RyanR3 subtypes effectively. To resolve this controversy, our laboratory is currently taking a genetic approach to evaluate a role of RyanRs in AD. Our initial results indicate that RyanR may play initially compensatory and later detrimental role in the context of AD pathology. Taking together, it is clear from multiple studies with various AD cellular and animal models that ER Ca 2+ signaling is disturbed in AD and that activity of both InsP3R and RyanR is enhanced. Increased expression of RyanRs at least partially responsible for enhanced CICR in AD neurons. The mechanisms responsible for enhanced activity of InsP3R are less certain and may involve direct gating of InsP3R by presenilins. It is also likely that increased ER Ca 2+ levels contribute to enhanced RyanR-mediated and InsP3R-mediated Ca 2+ release, as discussed in more details in the following section. It also appears that RyanR is a potential pharmacological target for AD treatment and that dantrolene may provide potential avenue for suppressing RyanR activity in AD. PRESENILINS There are mutations in presenilin 1 (PSEN1), presenilin 2 (PSEN2), and APP genes that are linked to early onset FAD. The majority, nearly 200, of these mutations are within PSEN1. To date many known PSEN1 mutations contribute to Ca 2+ disruptions in ER Ca 2+ signaling (Bezprozvanny and Mattson, 2008). PS1 and PS2 constitute the catalytic pore of the γ-secretase complex, other partner of the complex are nicastrin, aph-1, and pen-2 (De Strooper, 2003). The γ-secretase complex cleaves type-1 transmembrane proteins, including Notch receptor protein and APP. One of the main therapeutic approaches to AD is focused on development of γ-secretase inhibitors (GSIs) and modulators, however so far this approach has failed in phase III clinical trials of Eli-Lilly's Semagacestat, a non-selective GSI (Doody et al., 2013). Semagacestat treatment resulted in worsen cognition scores and increase in the risk of skin cancer (Doody et al., 2013), most likely due to inhibition of Notch processing. As a result, clinical trials of GSIs have been halted. In addition to contributing to altered γ-secretase function in AD pathogenesis, FAD PS mutations result in disturbed Ca 2+ signaling in neurons (reviewed in Stutzmann, 2007;Bezprozvanny and Mattson, 2008;Supnet and Bezprozvanny, 2010a,b). As discussed above, multiple studies demonstrated enhanced InsP3Rmediated and RyanR-mediated ER Ca 2+ release in PS-FAD cells. Presenilin mutations also affected SOC, a refilling mechanism for ER stores (Leissring et al., 2000;Yoo et al., 2000;Giacomello et al., 2005;Zhang et al., 2010b). To explain these findings, it was suggested that gating of InsP3R or RyanRs directly modulated by presenilins (Cheung et al., 2008(Cheung et al., , 2010Rybalchenko et al., 2008). It was also suggested that presenilins potentiate activity of sarco-/endoplasmic reticulum Ca 2+ ATPase (SERCA; , a mechanism that could contribute to the overfilling of ER Ca 2+ store. Our laboratory offered an alternative mechanistic explanation for most of these findings by demonstrating that wild type PSs Frontiers in Molecular Neuroscience www.frontiersin.org function as ER Ca 2+ leak channels (Tu et al., 2006), which function to maintain ER Ca 2+ homeostasis by constantly leaking Ca 2+ into the cytosol and balancing SERCA activity. Our results suggested that presenilin holoproteins function as low conductance passive ER Ca 2+ leak channel, and that ER Ca 2+ leak function of presenilins does not depend on their γ-secretase activity (Tu et al., 2006). Moreover, we found that some, but not all, FAD PS mutations disrupt Ca 2+ leak function (Tu et al., 2006;Nelson et al., 2007Nelson et al., , 2010, leading to the overfilling of ER with Ca 2+ and exaggerated ER Ca 2+ release observed in PS1/PS2 FAD mutants fibroblasts (Tu et al., 2006;Nelson et al., 2007Nelson et al., , 2010, cultured hippocampal neurons from 3xTg AD neurons (Zhang et al., 2010b), and primary lymphoblasts from FAD patients (Nelson et al., 2010). These data suggest that mutations in presenilins directly linked to deranged Ca 2+ signaling and neuronal dysfunction in AD by causing ER Ca 2+ overload. Our hypothesis has been directly challenged, in particular by the group of Dr Kevin Foskett (Shilling et al., 2012). These authors claimed that presenilin does not have a pore and cannot act as an ion channel (Cheung et al., 2008;Shilling et al., 2012). As we previously outlined, a number of serious technical and experimental issues exists with their negative arguments . Other experiments that oppose to our hypothesis have also been reported (Zatti et al., 2004(Zatti et al., , 2006. In contrast to our finding, the authors of these papers observed that FAD-PS expression lower the ER calcium content (Zatti et al., 2004(Zatti et al., , 2006. Despite existence of these controversial results independent experimental support for leak function of presenilin recently began to accumulate (Das et al., 2012). In a recent study, Bandara et al. (2013) performed an unbiased RNAi-based screen for modulators of calcium homeostasis in HEK293 cells. They transfected 250 candidate short-interfering RNAs (siRNAs) into the cells and used the mathematical model to quantify the effects of knockdown on calcium pump and leak rates, which resulted in the identification of proteins involved in the elusive ER Ca 2+ leak pathway. Knocking down presenilin-2 or ORAI2 dramatically reduced ER calcium leak rate, and knocking down PEN-2, encoded by PSENEN, greatly increased calcium leak rate (Bandara et al., 2013). Knockdown of PSENEN would inhibit proteolytic processing of presenilins and thus increase the holoprotein form of the protein, which is the form of presenilins that functions in ER calcium leak according to our previous experiments (Tu et al., 2006). Thus, enhanced ER calcium leak resulting from PEN-2 knockdown most likely reflects the accumulation of the presenilin holoprotein in the ER. As discussed in the recent review article (Bezprozvanny, 2013) these findings provide strong support to our hypothesis that presenilin holoprotein functions as ER calcium leak channel. Interestingly, Honarnejad et al. (2013) recently reported that there is PS holoprotein upregulation in human AD brain samples, suggesting a possibility of compensatory upregulation of leak pathway in AD neurons in order to reduce ER Ca 2+ overload. Where is an ion conduction pore of presenilin leak channel? From the structural-functional analyses we suggested that transmembrane domains 7 and 9 but not transmembrane domain 6 may play a role in forming the ion conductance pore of PS1 (Nelson et al., 2011). Recent publication reported the first crystal structure of archeal homolog of presenilin (PSH; Li et al., 2013). These authors discovered that PSH has a large hole that transverse the entire protein and is surrounded by transmembrane domains 2, 3, 5, and 7. These data are in good agreement with our mutagenesis mapping studies (Nelson et al., 2011). Moreover, these authors postulate that the hole is large enough to allow passage of the small ions (Li et al., 2013), suggesting that PSH may function as an ion channel. Importantly, the motifs that constitute catalytic core are conserved between PSH and PS1, therefore the structure of PS1 should be very similar to the structure of PSH. STORE-OPERATED CALCIUM CHANNELS Recent growing evidence suggests that SOC channels may be involved in AD pathogenesis. SOC channels are unique in the nature of their activation. They are activated in response to lowering of Ca 2+ content in ER. The first reports about role of SOC channels in the pathogenesis of AD have been published in 2000. Leissring et al. (2000) observed that fibroblasts isolated from PS1-M146V knock in mice exhibit significant impairments in store-operated Ca 2+ entry after stimulation of cells with bradykinin. These authors suggested that impaired SOC in these cells is due to elevated ER Ca 2+ levels in PS1-M146V fibroblasts (Leissring et al., 2000). In the same year Yoo et al. (2000) reported alteration in SOC activity in presenilin FAD mutant neurons. Two different mechanisms of mutant PS1-mediated dysregulation of SOC have been proposed (Herms et al., 2003). The first mechanism is linked to direct attenuation of SOC at the cell surface, the second mechanism evokes changes in processing of APP and generation of amyloid peptides (Herms et al., 2003). However, second mechanism cannot explain alterations of SOC observed in the absence of human APP and Aβ42 accumulation. TRP channels may play a role in disruption of neuronal SOC in AD (Yamamoto et al., 2007), but the mechanisms involved in changes in TRP channel expression or activity in AD are poorly understood. In addition to TRP channels, important players of SOC in excitable and non-excitable cells are stromal interaction molecule 1 and 2 (STIM2) proteins. STIM 1 and STIM2 protein reside in ER, and reduction in ER Ca 2+ levels causes oligomerization of STIMs, translocation to plasma membrane, and activation of SOC channels (Liou et al., 2005). The molecular identity of neuronal SOC is poorly understood, but most likely involves complex of STIMs with TRP channels and/or Orai proteins (Figure 1). Interestingly, changes in expression of STIM1 and STIM2 proteins were found in PS knockout and FAD mutant cells (Bojarski et al., 2009), suggesting a possible mechanism for SOC dysregulation. In recent review articles we suggested a possible connection between dysregulated neuronal SOC and synaptic spine maintenance in AD and aging brains (Popugaeva et al., 2012;Bezprozvanny and Hiesinger, 2013). These ideas are currently being tested experimentally in our laboratory. Another possibility involves potential connection between impaired neuronal SOC and abnormal synaptic vesicle recycling in PS mutant neurons (Zhang et al., 2009(Zhang et al., , 2010a. SUMMARY In the summary we would like to conclude with our working hypothesis for ER Ca 2+ dysregulation in AD (Figure 1). FAD Frontiers in Molecular Neuroscience www.frontiersin.org linked mutations in PS cause disruption of PS Ca 2+ leak function. As a result Ca 2+ is accumulating inside of the ER. Similar increase in ER Ca 2+ levels occur as a result of brain aging. In order to compensate for ER Ca 2+ overload neurons mount two physiological responses: (1) upregulate gating of InsP3R and expression/activity of RyanR, and (2) downregulate activity of neuronal SOC (Figure 1). We hypothesize that these initially protective responses with time become toxic and eventually lead to synaptic dysfunction, synaptic loss, impaired plasticity, and learning, loss of memories and neurodegeneration. The role of RyanR in these processes is likely to be more significant than the role of InsP3R, as InsP3R predominantly localized in the soma, whereas RyanR are abundant in the postsynaptic and presynaptic terminals. Dantrolene provides a possible way to suppress RyanR-mediated Ca 2+ release pharmacologically, but there are significant issues with specificity of dantrolene effects and its delivery to the brain. Neuronal SOC pathway provides a novel potential target for AD treatment that should be explored further.	2016-06-17T20:20:25.497Z	2013-09-18T00:00:00.000	{ "year": 2013, "sha1": "fb9d8c60b96a6840f22afa6f025dc59b1ec2482b", "oa_license": "CCBY", "oa_url": "https://www.frontiersin.org/articles/10.3389/fnmol.2013.00029/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "fb9d8c60b96a6840f22afa6f025dc59b1ec2482b", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Chemistry" ] }
218797393	pes2o/s2orc	v3-fos-license	The challenges of project management in small and medium-sized enterprises: a literature review based on bibliometric software and content analysis The purpose of this paper is to design an overview about Project Management (PM) in Small and Medium-Sized Enterprises (SMEs) by analysing the evolution of publications and the main topics since 1996 to 2016 to motivate future research that helps SMEs to apply PM practices more efficiently. This study performed bibliometrics associated with content analysis of publications collected in scientific bases Web of Science and Scopus and in the periodic International Journal of Project Management. For that, the software VOSviewer, Nvivo, Minitab, and Excel were used in the analyses. The scan of 235 papers about Project Management in SMEs supported a literature overview. Furthermore, four thematic categories are highlighted: Project Management Practices, Planning and Control Systems, Collaboration, and Knowledge Management. Moreover, it was observed that SMEs requires a lighter PM methodology, focused on people and flexible. Besides that, the results show that the main challenges involve a lack of resources and qualified people and the high turnover. However, overcoming these issues, PM can benefit growth and innovation in SMEs. Therefore, this study presents a conceptual framework of benefits and challenges in Project Management in SMEs, reducing the research gap. Furthermore, recommendations for future research, mainly in Brazil, are given. Introduction Small and medium-sized enterprises (SMEs) perform an important role in economic and social development. While in the European Union as well as the average member countries of the Organization for Economic Co-operation and Development (OECD), small and medium-sized enterprises represent more than 90% of establishments and contribute more than 60% of total turnover (Muller et al., 2015). In Brazil, however, they represent more than 98% of the companies and contributed only 27% of GDP in 2011 (SEBRAE, 2014). For Feldmann (2011), the lack of innovation, entrepreneurship training, and cooperation between companies are the main responsible for this low performance. The contribution of SMEs is not limited to the large number of companies, employees, and share of GDP, moreover, they are also responsible for innovation (Mannan et al., 2016). SMEs have a great innovative capacity, however, the lack of resources and knowledge hamper their development (Marcelino-Sádaba et al., 2014). Therefore, there is a need to accelerate the growth and competitiveness of these enterprises (Forsman, 2008) for the progress of the economy (Mannan et al., 2016). The study by Pollack & Adler (2016) demonstrates that project management (PM) improves the financial performance of SMEs. In addition, Turner et al. (2010) confirms that it can benefit operations, growth and innovation processes of SMEs. However, there are obstacles in adopting traditional project management guides (such as PMBoK and IPMA ICB) because they are bureaucratic and have little focus on personal skills (Turner et al., 2009(Turner et al., , 2010(Turner et al., , 2012. In addition, SMEs have unique characteristics, such as limited human and financial resources, few clients, non-specialized employees, and high turnover (Forsman, 2008;Wong, 2005). Despite their collaboration with the global economy and the key role in the development of nations, few publications about project management in SMEs are available (Juca et al., 2010;Pollack & Adler, 2016;Turner et al., 2010), and, according to Turner et al. (2012), for the PMI (Project Management Institute) interests SME-focused studies. To contribute to the literature, this article seeks to answer the following research question (RQ): What are the challenges and benefits of project management for SMEs? Facing this context, this study aims to outline the project management landscape in SMEs from 1996 to 2016, identifying the main topics, references and trends, through bibliometric analysis and content according to literature methodologies (Carvalho et al., 2013a;Iritani et al., 2015;Watanuki et al., 2014). In this way, aiming to contribute to the construction of the knowledge of the area and with future research focused mainly on how project management can benefit SMEs. The article follows this structure: section 2 presents the most relevant characteristics of SMEs and the main challenges. Section 3 explains the Research Methodology based on data of ISI Web of Science, Scopus and the International Journal of Project Management, on their statistical analyses (evolution of publications per year, main journals, and countries) and on bibliometrics (networks of keywords, co-citation, identification of outliers and hot topics). For this purpose, VOSviewer software (van Eck & Waltman, 2010), Nvivo (QSR International Pty Ltd., 2012) and Minitab (2010) are used. Section 4 shows the results of the previous analyses. Next, section 5 presents the content analysis of the main publications, and section 6 discloses the discussion. Finally, section 7 develops the conclusions, recommendations, and limitations. SMEs: profile and main challenges Different countries use diverse classifications of company sizes (OECD, 2016), as shown in Table 1. These divergences hamper the study of this area, so the commonly adopted classification is based on the number of employees (OECD, 2016), as in this study. The economic sector of SMEs also depends on the country. Among OECD members, SMEs are the majority of the construction and service sectors (OECD, 2016). The same report indicates that services and manufacturing predominate over construction in Brazil. In the European Union, the companies of service-related sectors are growing (Muller et al., 2015). In addition, the financial performance is also divergent, while the average OECD membership is 60% of the total economic value added; in Brazil, it is less than 50%; and, in Japan, it is practically 100% (OECD, 2016). (Brasil, 2006) does not address medium-sized enterprises. Despite these differences, SMEs have characteristics in common. In general, they have a small number of employees led by a visionary entrepreneur (Juca et al., 2010) or by founding partners, who also holds senior management roles, which means there are few hierarchical levels (Huin, 2004). Therefore, there is a close relationship between top-management and employees that provides informal communication (Huin, 2004;Malhotra & Temponi, 2010). In addition, employees perform various tasks (Malhotra & Temponi, 2010) and need multiple skills (Forsman, 2008) and there is no clear and formal division between departments (Huin, 2004). Moreover, SMEs do not usually have a product portfolio, hence they specialize in a niche market (Malhotra & Temponi, 2010). SMEs need to address several challenges to become more competitive and to enter the market. Globalization, rapid advances in technology, and shortening product lifecycle force SMEs to be more innovative (Mannan et al., 2016). However, government difficulties and access to funding may hinder this process (OECD, 2016;Mannan et al., 2016). Another challenge is to hire qualified employees (Malhotra & Temponi, 2010), as the lack of necessary skills and human resistance to change are obstacles to innovation (Mannan et al., 2016). SMEs have lower intellectual capital than large companies (Hsu & Fang, 2009). Moreover, SMEs are not able to manage innovation in a systemic way, resulting in indefinite and flawed projects (Marcelino Sádaba et al., 2016). Research methods Given the characteristics of the objectives and research, the methodology is a systematic literature review (Carvalho et al., 2013a). The systematic review of the literature is built on a clear question whose answer is developed based on a reproducible scientific method that determines rigorous criteria for information evaluation. Because it is broad, the systematic review can synthesize important concepts from a certain field of knowledge. Thus, indicating trends for future research and providing researchers with a concise overview of the area. The systematic review employs several tools, among them, bibliometry, content analysis and semantics (Carvalho et al., 2013a;Sampaio & Mancini, 2007). For this work, we selected bibliometrics and content analysis. Sampling process The sample was collected from ISI Web of Science (2016) and Scopus (2016) databases, since they are bases that include publications from other databases and allow the extraction of metadata (abstracts, references, number of citations, institutions, among others) that support the bibliometric study (Carvalho et al., 2013a;Iritani et al., 2015;Watanuki et al., 2014). In addition, the ISI Web of Science presents only papers published in journals with impact-factor (IF) calculated in the Journal Citation Report (JCR) (Carvalho et al., 2013a). Additionally, publications of the International Journal of Project Management (IJPM, 2016) were included due to the importance of this journal. The strings were "project management" and "small company or medium companion or small enterpris * or medium enterpris * or sme", the publications were filtered by year (1996 to 2016) and by type (article, review, and article in press). This process resulted in 408 identified publications. Of this sample, 33 publications were excluded because they were repeated and 140 because they did not approach Project Management in SMEs after reviewing titles, abstracts, and keywords. Most of the excluded were in the knowledge area, but not specifically for SMEs or were studies about large companies and SMEs. In these cases, an individual evaluation was executed to exclude only those whose main sample was composed of large companies and not SMEs. For example, Proverbs et al. (1999) presents the term SME only as a reference because the entire study was performed with projects of large companies in the construction sector. Hence, bibliometric and content analyses were performed as shown in Figure 1. Data analysis Bibliometry was conducted in two steps: metadata and content analyses. The first study, called Sample Demography, consisted of three stages: counting the number of publications per periodical per year, which identified the journals that generally approach the theme (Carvalho et al., 2013a); count of the total of publications per year, that delineated the evolution of the area; and stratification by country, which indicated the countries that contributed most to the construction of the area. The second analysis was based on bibliometric networks of the metadata using VOSviewer (van Eck & Waltman, 2010). The first network was that of keywords, which established the terms that occur concomitantly. In this study were excluded similar terms (e.g. "erp", that means Enterprise Resource Planning); the second was that of co-citation, which demonstrated the degree of similarity between articles that are cited together (Carvalho et al., 2013a). Nvivo software (Bazeley & Jackson, 2013) was used in content analysis and identification of hot topics. The hot topics indicate interesting topics in the area of knowledge, according to the methodology proposed by Banks (2006). In general terms, the calculated m-index, which results from the ratio of citations per author per year, indicates whether the topic is of wide scope and importance (m> 2), or has the potential to become a hot topic (0.5 < m <2), or if it is of restricted interest (m <0.5). Outliers are papers with high performance of the impact factor (IF), which results of the relation between number of citations per year (Cn) and the impact factor of the journal according to the JCR (JIF), as Equation 1 demonstrates (Carvalho et al., 2013a). This statistical analysis was conducted in Minitab software. Articles without citations were excluded. Finally, the second stage relied on content analysis of these outliers and other relevant papers to identify the main topics addressed and the main challenges involved in SMEs' projects (Carvalho et al., 2013b). (1) Sample demography The 235 valid publications belonged to 137 journals from various areas such as engineering, business, administration, computer science, environmental sciences, decision science, among many others. Therefore, the theme was relevant and multidisciplinary. Of the total periodicals, approximately 70% had only one publication. Almost 20% of the publications were concentrated in 4 journals, namely International Journal of Project Management (30), ZWF (8), Production Planning and Control (7), and Construction Management and Economics (7). Regarding the publications of these 4 journals, as is shown in Figure 2, the first one was published in 2001, and it is interesting to note that only the IJPM showed an increasing trend in the number of publications. Reminding that the sample was withdrawn in September 2016, thus this number may have increased. The analysis of publications, Figure 3, per year revealed that, on average, journals accepted 12 publications per year and there was a peak with 26 publications in 2010. About half the sample came from six countries: the United Kingdom (17%), Germany (10%), the United States (7%), Taiwan (5%), Spain (5%) and India (4%), as is shown in Figure 4. The prominence of European nations was due to the most European companies are SMEs, therefore the development of these countries depends on their growth and innovation (Marcelino Sádaba et al., 2016). In relation to Brazil, only 4 publications were found, although 98% of their companies are SMEs, demonstrating that there is space for research in the area. Figure 5 shows the keyword network designed in VOSviewer software, selecting keywords with at least 6 occurrences. The position reveals the proximity between the terms; and the lines, concomitant occurrences. Project management was the main theme that interconnects the others. Three major topics were: Manufacturing, Construction and Information, Innovation and Knowledge. Table 2 shows these categories. The first theme, Manufacturing, embedded in the SME research because of the pressure they underwent to meet the needs of large customers and become globally competitive (Huin, 2004;Malhotra & Temponi, 2010). Examples of topics were: the implementation of advanced manufacturing technologies in SMEs (Dangayach & Deshmukh, 2005); the application of the ERP system (Huin, 2004;Malhotra & Temponi, 2010;Mohammadjafari et al., 2011;Muscatello et al., 2003;Zafeiropoulos et al., 2005); and the development of information system for mass customization (Dean et al., 2009). Main themes Regarding the second theme, Construction & Information, approximately, 10% of the sample addressed it. One of the most cited articles in the sample (127 citations), by Sage et al. (2010), states that construction sector SMEs should adopt supply chain practices to become more reliable and productive. Another relevant article (51 citations), by the Sexton & Barrett (2004), presents a model to help construction SMEs to improve technology management to become more competitive. Implementing information systems was widely approached as a means to control, integrate and manage chains of small and medium-sized contractors (Ahuja et al., 2009a;Jacobsson & Linderoth., 2012;Lee & Egbu, 2007;Lin, 2010;Pellicer et al., 2009). About the third theme, it was assumed that innovation depends on knowledge management. There were researches that focused on developing specific methodologies for SMEs, such as innovation management (Maravelakis et al., 2006), risk management (Marcelino- Sádaba et al., 2014), and management of innovation portfolio (Von Ahsen & Heesen, 2009). Other studies addressed public policies to support innovation in SMEs (Egbetokun et al., 2009;Hwang & Ward, 2001). In addition, some researchers studied innovation projects between SMEs and research institutions (Bjerregaard, 2010;Davenport et al., 1998;Okamuro, 2007). The analysis of hot topics identifies topics within a research area that are interesting to the scientific community (Banks, 2006). In Figure 6, it can be seen from the m-index that there were no hot topics, however, there were topics likely to become hot topics in a larger scientific community (0.5 <m ≤2) (Banks, 2006), they were: Processes, Information, Technology, Development, Products, and Success. The trend was confirmed in 6 of the 9 articles published in 2016, as is shown in Table 3. On the other hand, Software, Business, Construction, Knowledge, and Tools were topics of interest to a smaller scientific community (m <0.5) and were not hot topics (Banks, 2006). (Dyerson et al., 2016) Among the published articles in 2016, the topic of success was presented in combination with processes and information technology. For example, according to Marcelino Sádaba et al. (2016) and Moreau (2016), the implementation of appropriate project management for SMEs is a critical success factor for the company's survival. Another success factor is the effective teamwork (Pons & Haefele, 2016). Regarding IT, Pollack and Adler cite "[…] using project management and IT professional skills to undertake core business activities make a significant contribution to improving the financial performance of small to medium enterprises." (Pollack & Adler, 2016, p. 836). Therefore, for researchers of SMEs, studies combining these topics are of interest to the scientific community. In addition, they are themes that are of interest to project management practitioners in SMEs, as they relate to the competitiveness and survival of these companies. Outliers -main authors and studies "While in many fields outliers can simply be discarded as being exceptions, in bibliometrics the extreme values represent the high-end of research performance and therefore deserve special attention", says Glänzel (2013, p. 13). In Figure 7, there are 17 outliers publications that cover mainly four themes: Project Management Practices, Control and Planning Systems, Collaboration, and Knowledge Management, which were discussed in the next session. Wong's (2005) study about the knowledge management in SMEs received 51% of outliers' citations. The 17 outliers applied different methodologies, 8 were qualitative studies (semi-structured interviews, case study), 4 used qualitative and quantitative methods (literature review and research), 4 were exclusively quantitative (research) and only 1 performed a simulation. The predominance of qualitative studies revealed that it is a developing area (Fleury et al., 2016). The co-citation network, Figure 9, shows the most cited references in the sample. This network represents the frequency that two documents were cited together (Small, 1973). Table 4 summarizes the main themes of the clusters. The criterion for selection of publications was a minimum of 7 times citation. (Turner et al., 2009(Turner et al., , 2010(Turner et al., , 2012) (Rostami et al., 2015) Based on the total link force calculated by VOSviewer; * Cited the central reference. Cluster 1 related to knowledge management presented the central reference with the highest number of citations. The most cited work was "The knowledge-creating company" (Nonaka & Takeuchi, 1995), whose main contribution was the definition of tacit and explicit knowledge. The study inspired articles about knowledge, risk management, obstacles to technology transfer and innovation in SMEs. Cluster 5 had as main references the work of Peter Love who dealt with management in the construction industry, an example was the article "An exploratory study of information technology evaluation and benefits management of SMEs in the construction industry" (Love & Irani, 2004, p.227). He contributed to articles approaching the adoption of IT, communication and supply chain tools in SMEs in the construction industry. Another highlight was the researcher Rodney Turner (cluster 6). His major contribution in the area was a series of three articles that compared the practices adopted according to the SMEs' size and nature. These papers supported a project management approach more appropriate to the characteristics and needs of these companies (Turner et al., 2009(Turner et al., , 2010(Turner et al., , 2012. Content analysis The content analysis, based on the main papers, led to four thematic categories: Project Management Practices, Planning and Control Systems, Collaboration, and Knowledge Management. They are presented in Table 5. Figure 10 summarizes the discussion. Project management practices SMEs implemented project management both to accomplish internal innovation and growth and to attend external customers' requirements (Turner et al., 2010(Turner et al., , 2012. Businesses that used PM practices commonly reported increased profitability (Pollack & Adler, 2016). However, very bureaucratic and formal methodologies, such as those proposed by PRINCE2, were not the most suitable, as best fit to large engineering and construction projects (Turner et al., 2010). SMEs required "light" versions of PM guides, aimed at non-specialized employees and involving informal communication. In addition, the practices required by smaller companies are different from those required by medium-sized enterprises (Turner et al., 2012). SMEs applied PM practices in their operations to provide customized products to their customers, and to manage growth and innovation (Marcelino-Sádaba et al., 2014;Turner et al., 2010). However, few companies followed rigorous methodologies (Currie, 2003). Consequently, SMEs became less competitive and innovative than large companies (Hsu & Fang, 2009;Wong, 2005). Therefore, SMEs used and needed project management practices, however, two key factors impacted their success: limited resources and a lack of skilled employees (Le Pochat et al., 2007;Turner et al., 2010;Wong, 2005). Hence, there is a great challenge to be overcome: developing efficient and feasible PM approaches to be performed by SMEs. In this context, the literature has focused mainly on developing procedures, guides and models to assist in their successful implementation, as an example they presented models for risk management (Currie, 2003;Marcelino Sádaba et al., 2014;Renna & Argoneto, 2010); resource management (Meade & Presley, 2002); and ecodesign (Le Pochat et al., 2007). Planning and control systems SMEs faced great challenges in entering global markets and becoming suppliers to large consumers (Malhotra & Temponi, 2010). Mainly, in the manufacturing sector, they required production planning and control to deal with large companies (Huin, 2004). However, most enterprises were not used to plan, instead, they had produced only upon request of the customer (Renna & Argoneto, 2010). To change this culture, SMEs adopted production planning tools already used by large companies, such as Enterprise Resource Planning (ERP) (Huin, 2004;Malhotra & Temponi, 2010;Muscatello et al., 2003) and Workload Control (WLC) (Stevenson et al., 2011). Hence, SMEs began to adopt systems that aid in the planning of production. Although these systems appeared as competitive advantages, their costly and complex implementation resulted in failures (Huin, 2004;Malhotra & Temponi, 2010;Muscatello et al., 2003). Thus, the literature focused on identifying the success factors and developing suitable models for SMEs to implement ERP (Huin, 2004;Malhotra & Temponi, 2010;Muscatello et al., 2003), WLC (Stevenson et al., 2011) and a new model for connecting commercial planning and negotiation through digital platforms (e-procurement) (Renna & Argoneto, 2010). Collaboration Collaboration projects were means of sharing knowledge and tools. They were important, especially, for SMEs to overcome the lack of resources and to remain competitive. These due to the synergy effect, such as scale and scope gain, and risk reduction (Kernel, 2005;Okamuro, 2007). Many nations promoted cooperation between SMEs and research institutes and universities to encourage commercially productive relations through technology transfer (Bjerregaard, 2010;Davenport et al., 1998). Moreover, co-operation among SMEs could lead to common goals that benefit all companies in the same sector that would be difficult to achieve individually (Kernel, 2005). Therefore, cooperation projects were important for SMEs to achieve strategic objectives, such as the development of a new product or sustainable development. The studies focused on the success factors for shared projects between SMEs and public research institutes (Bjerregaard, 2010;Davenport et al., 1998;Okamuro, 2007), and among SMEs and other business partners (Kernel, 2005;Okamuro, 2007). Collaboration with other organizations can be a strategy for innovation in SMEs (Bjerregaard, 2010;Davenport et al., 1998;Kernel, 2005;Okamuro, 2007). It was a logical response to turbulent conditions and was a mean of transferring knowledge and tools (Kernel, 2005), technology (Davenport et al., 1998) and gain of scale and scope (Okamuro, 2007). However, there are also difficulties in managing these types of projects. For their success, they required efficient stakeholder management (Kernel, 2005), the establishment of trust between stakeholders, and a clear division of costs, responsibilities, and results (Davenport et al., 1998). Hence, PM practices were essential for SMEs to manage their innovation processes that allow them to be more competitive in the market. Knowledge management in SMEs Project management also benefited innovation through knowledge management. Knowledge management is critical to business success (Wong, 2005). Most of the knowledge in SMEs was in the tacit form (Miranda et al., 2014), which was difficult to communicate and formalize (Sexton & Barrett, 2004). Tacit knowledge is highly personal, non-standardized and consists especially of know-how (Currie, 2003). While explicit knowledge is formal, which makes it easy to share information and procedures (Currie, 2003). Furthermore, another difficulty in managing knowledge was the employee turnover, which led to information loose (Wong, 2005). Successful implementation of knowledge management depends on the support of leadership, organizational culture, available technologies, and infrastructure, as well as strategic alignment and human resource management (Wong, 2005). In addition to inadequate knowledge management, other factors may have led to failures in innovation capacity, such as: the lack of investment in intellectual capital (human, relational and structural) (Hsu & Fang, 2009), difficulties in introducing a new concept in the market, and to establish partnerships with organizations capable of producing innovation (Mannan et al., 2016). Discussion Supported by the results of the bibliometrics (keywords and co-citation networks), hot topics and content analysis of outliers, it was possible to map which themes are trends. The topics that were identified in all the analyses and their respective numbers of related publications were: Product Development (33), Information Technology (23), and Project Management (208). The themes of Planning (74), Knowledge Management (45) and Construction (42) were not only identified as a hot topic, however, they appeared in all other analyses. Therefore, for researchers interested in SMEs, a trend is approaching the critical success factors and processes required for Project Management. A recommended source due to its prominence as a theoretical pillar (the network of co-citation) and as outlier are the articles of the group Turner, Ledwith & Kelly. This group has conducted a series of studies to identify the best practices already used and the needs of European SMEs (Turner et al., 2009(Turner et al., , 2010(Turner et al., , 2012. These surveys could be replicated in other countries because they detail the methodology. Another trend identified is the Knowledge Management. One of the theoretical pillars is the author Ikujiro Nonaka who popularized the classification of knowledge in tacit and explicit and demonstrated its importance for innovation (Nonaka & Takeuchi, 1995). Knowledge management is recognized as a way to innovation, new product development and competitive advantage (Hang Do et al., 2014;Hsu & Fang, 2009;Huang et al., 2008;Sexton & Barrett, 2004;Wong, 2005). Furthermore, it is imperative for risk management, since it is necessary to create a history of risk and to share information to develop a risk assessment model (Currie, 2003;Leopoulos et al., 2006;Marcelino-Sádaba et al., 2014). Moreover, the importance of the construction industry is highlighted. "Most construction companies are small and medium-sized enterprises (SMEs) that manage project-based and business-focused activities simultaneously" (Pellicer et al., 2009, p. 4). The construction sector experienced transformations to absorb tools developed in the manufacturing to become more productive and reduce costs and waste (Dainty et al., 2001;Taggart et al., 2014). This evolution was also drawn in the keyword network since terms related to industry, production, and planning (such as ERP) preceded those related to construction. Therefore, there was a transfer of knowledge and tools from the manufacturing to the construction. As construction was not indicated as a hot topic, other sectors should dominate research in the area, such as technology companies (startups) and services, because SMEs dominate these sectors (OECD, 2016). Regarding the Brazilian scenario, the sample had only four articles and, curiously, all dealt with technology-based SMEs. They are recent publications, after 2009, which supports the previous hypothesis that research on technology companies is a trend. Brazilian SMEs presented high levels of informality and a lack of knowledge and risk management (Juca et al., 2010;Miranda et al., 2014). The researches were performed in a qualitative way, demonstrating that the area is still in development. The themes were: project management maturity (Juca et al., 2010); risk management (Miranda et al., 2014); PMO implementation (Richter, 2009); and the development of software that could help SMEs to improve their practices (Pereira et al., 2013). Therefore, quantitative studies are needed to have a diagnosis of project management practices in SMEs in Brazil. Turner, Ledwith & Kelly publications can serve as a basis. Moreover, none addressed critical success factors, so identifying them for the national context would also be a relevant contribution. Although all four articles approached with technological SMEs, they focused on the development of software. Thus, studies in other sectors would be relevant due to their essential role in the social and economic development of the country through innovation (Juca et al., 2010). One example would be to study the relationship of cooperation between SMEs and public research institutes and universities in biotechnology projects, as they result in products of high added value that directly impact the population. Conclusions This article proposed to answer the RQ: What are the challenges and benefits of project management for SMEs? The survey of 235 publications on Project Management in SMEs from the Web of Science and Scopus databases and the International Journal of Project Management allowed us to answer the research question, contributing to the literature in presenting the challenges, characteristics, and benefits of project management in SMEs. Project management in SMEs is characterized by their requirement for lighter PM methodologies, associated with the flexibility, few resources, people-oriented, and for non-specialized employees. In addition, for a good PM, SMEs should seek collaboration with external stakeholders to mitigate resource shortages, looking for synergies. The benefits indicated increased profitability and innovation in SMEs. The results demonstrated that the main challenges involve the lack of resources and qualified personnel and high turnover. Four thematic categories appeared to be most promising for future work: Project Management Practices; Planning and Control Systems; Collaboration; and Knowledge Management. Through bibliometric and content analysis, it was observed that there was a transition of knowledge and practices between sectors. Initially, the studies focused on the manufacturing industry as small and medium suppliers of large companies needed to adapt to project management practices to become more competitive and serve global markets. The prominence in the area was the studies on implementing ERP (Enterprise Resource Planning) systems. These practices were adopted by the construction industry, also formed predominantly by SMEs, to become more productive. Finally, through the analysis of hot topics and the most recent articles, we noticed that the focus is becoming technology-based companies. Finally, we suggested a quantitative study to evaluate the extent of the use of good management practices and to identify the critical success factors for the projects in Brazil. In addition, it was encouraged to expand research on technology-based SMEs since innovation is the way to social and economic development (Juca et al., 2010). Moreover, it would be interesting to approach partnerships with universities, because collaboration helps to optimize the use of resources and results in innovation through technology transfer. The limitations of this article were associated with the methodology. Interesting articles may have been excluded from sampling, although the bases chosen are very inclusive. Additionally, important information may have been lost during data analysis as it is subject to human error. However, we believe that statistical and bibliometric analyses help to soften biased interpretations.	2020-04-23T09:09:39.887Z	2020-01-01T00:00:00.000	{ "year": 2020, "sha1": "a4c47988209ecb7f8ae090dbcd4b45a72d2929cc", "oa_license": "CCBY", "oa_url": "http://www.scielo.br/pdf/gp/v27n1/0104-530X-gp-27-1-e3768.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "250ded8689147d41210bf143b8255ae1bfc7d9fd", "s2fieldsofstudy": [ "Business", "Engineering" ], "extfieldsofstudy": [ "Business" ] }
9290551	pes2o/s2orc	v3-fos-license	The Impact of Polymer Grafting from a Graphene Oxide Surface on Its Compatibility with a PDMS Matrix and the Light-Induced Actuation of the Composites Poly(dimethyl siloxane) (PDMS)-based materials with improved photoactuation properties were prepared by the incorporation of polymer-grafted graphene oxide particles. The modification of the graphene oxide (GO) surface was achieved via a surface initiated atom transfer radical polymerization (SI ATRP) of methyl methacrylate and butyl methacrylate. The modification was confirmed by thermogravimetric analysis, infrared spectroscopy and electron microscopy. The GO surface reduction during the SI ATRP was investigated using Raman spectroscopy and conductivity measurements. Contact angle measurements, dielectric spectroscopy and dynamic mechanical analyses were used to investigate the compatibility of the GO filler with the PDMS matrix and the influence of the GO surface modification on its physical properties and the interactions with the matrix. Finally, the thermal conductivity and photoactuation properties of the PDMS matrix and composites were compared. The incorporation of GO with grafted polymer chains, especially poly(n-butyl methacrylate), into the PDMS matrix improved the compatibility of the GO filler with the matrix, increased the energy dissipation due to the improved flexibility of the PDMS chains, enhanced the damping behavior and increased the thermal conductivity. All the changes in the properties positively affected the photoactuation behavior of the PDMS composites containing polymer-grafted GO. Introduction Photomechanical actuators are materials that convert photons into mechanical motion. In rubbery elastomeric matrices, stress and displacement are generated upon irradiation of a pre-strained sample via an external light source, i.e., the thermoplastic effect. The material absorbs the energy that is transported via thermal convection to the pre-strained polymer chains that contract and generate mechanical motion. The proposed mechanism has been studied in-depth by Cohn et al. [1]. Elastomeric photomechanical actuators are lightweight and remotely controlled. They are utilized as tactile devices [2,3], artificial muscles [4], vascular stents [5], intravascular embolic coils [6], micro-grippers [7], and micro-motors [8]. Liquid crystalline elastomers (LCE) [9,10], thermoplastic elastomers [5,[11][12][13][14][15] (TPE) and cross-linked elastomers [6,16] have been reported as matrices for polymer composite-based photomechanical actuators. The negligible role of the host polymer on the actuation mechanism was revealed [16]. Poly(dimethyl siloxanes) (PDMS) are cross-linked elastomers that provide shape stability over time, biocompatibility and a variety of mechanical properties depending on the cross-linking density. Despite all their advantages compared to thermoplastic elastomers, their main drawback is that they are not re-processable. The photomechanical actuation response can be enhanced via the utilization of additives or functional fillers, such as carbon nanotubes and graphene oxide [17][18][19], graphene [15,18], carbon black [18], and molybdenum disulfide [20,21]. The fillers can facilitate the absorption of light energy and its transfer through a material to enhance the actuation response. The actuation response can be affected by the type of filler, the concentration and the arrangement. To maximize its efficacy, the filler should be well-dispersed in the matrix [18,22]. An arrangement in the actuation direction has been reported to promote the actuation as well [9]. Regarding the shape of the fillers, the effectiveness increases with the dimensionality, which facilitates the orientation of the polymer chains; i.e., one dimensional carbon nanotubes were more effective than carbon black, but not as effective as two dimensional graphene platelets [18], and this was ascribed to the better dispersion in PDMS. Surprisingly, graphite oxide flakes did not exhibit a good performance. Nevertheless, this drawback can be overcome using a suitable surface modification. The oxidized form of graphene, i.e., graphene oxide (GO), is rich in oxygen-containing functional groups that facilitate its further functionalization, and after reduction, it is electrically conductive [23]. Grafting polymer chains onto the surface of inorganic materials is a powerful approach to finely tune the surface chemistry of the particles. This approach enables control of the physicochemical properties, interfaces, and/or biochemical functionalities [24]. In addition, living/controlled polymerizations allow for tailoring of the molecular architecture of polymer brushes, e.g., the polymer chain uniformity (low dispersity), chain composition (homopolymers, block copolymers, gradient, periodic, and statistical), functionality and grafting density (sparse or dense). Our previous studies revealed that the suitable surface modification of carbon nanotubes (CNTs) can effectively tailor the photoactuation response through the selective location of CNTs in block copolymers exhibiting thermoplastic elastomer properties [14,22,25,26]. Here, GO prepared via the oxidation of expanded graphite was investigated instead of CNTs. The amount and price of the prepared GO are more acceptable for large-scale applications. The hydroxyl and carboxylic groups enabled simple immobilization of the initiators for subsequent atom transfer radical polymerization (ATRP) from the GO surface. The polymer brushes of poly(methyl methacrylate) (PMMA) and poly(n-butyl methacrylate) (PBMA), which have comparable molar masses and grafting densities, were grafted from GO. The study focused on the effect of polymer modification on GO surfaces on the photoactuation response of the PDMS composites. The surface modification of GO by methacrylate polymers affected the compatibility with the PDMS matrix and the physical and thermal properties of the PDMS composites. Because of this effect, a polymeric material with high and reversible response to light was obtained. Graphene Oxide Preparation and Initiator Immobilization Graphene oxide (GO) was prepared from a graphite powder using a modified Hummers method [27]. The raw graphite (5 g) was vigorously stirred with H 2 SO 4 (100 mL), and the mixture was cooled to 5 • C using an ice/water bath. Subsequently, NaNO 3 (2.5 g) and KMnO 4 (15 g) were gradually added. The mixture was stirred for an additional 6 h, and then, DW (300 mL) was added dropwise, while the temperature was kept below 40 • C. Finally, concentrated H 2 O 2 (40 mL) was added, and the solution turned a brilliant brown, which indicated the complete oxidization of graphite. The product was separated in a high-speed centrifuge (Sorvall LYNX 4000, Thermo Scientific, Waltham, MA, USA) operating at 10,000 rpm for 20 min at 25 • C. The cleaning routine was based on the dispersion of GO in 0.1 M HCl and re-separation in a centrifugal field. The procedure was repeated with DW several times until the pH reached a value of 7. Then, the particles were lyophilized to remove any residual water after the purification process. Finally, a brown powder was obtained. The presence of the reactive groups on the GO surface was used for the reaction with BiBB to immobilize the ATRP initiator on the GO surface and prepare GO-I using the method described elsewhere [28,29]. Grafting Polymer Chains from the GO Surface The GO sheets modified with the ATRP initiator (0.75 g) were transferred into a Schlenk flask equipped with a gas inlet/outlet and a septum. The flask was evacuated and replaced with argon several times. BMA (17.5 mL, 110 mmol) or MMA (11.8 mL, 110 mmol), EBiB (0.162 mL, 1.1 mmol), PMDETA (0.92 mL, 4.4 mmol), and anisole (15 mL) were then added to the flask, and the flask was degassed using several freeze-pump-thaw cycles. The CuBr catalyst (0.156 g, 1.1 mmol) was quickly added under an argon flow into the frozen system. Anisole was used as the solvent at an amount of 50 vol %. The polymerization mixture was stirred at 60 • C for 2 h, and then, the polymerization was stopped by exposure to air. The product was filtered, washed with DMF (2 × 200 mL) and acetone (2 × 200 mL), and dried with diethyl ether (2 × 100 mL). Composite Preparation An elastomeric matrix was prepared by mixing PDMS, silicone oil and a curing agent in a volume ratio of 8:2:1. The matrix was filled either with neat GO or GO with grafted PMMA or PBMA chains and was properly homogenized using mechanical stirring for 30 min at room temperature. The concentrations of the fillers were 0.1, 0.5 and 1 vol %. The mixture was poured into a Teflon-lined mold and evacuated at 60 mbar to eliminate the presence of air bubbles. Then, the mold was placed into an oven for 2 h at 60 • C to fully cross-link the PDMS-based composites. It should be stated that same procedure was used for the GO-I particles; however, even after 48 h, the composites were not fully cross-linked probably due to the presence of 2-bromoisobutyryl groups on the GO surface, which could eliminated the function of the cross-linker. General Characterizations Proton nuclear magnetic resonance ( 1 H NMR) spectra were recorded at 25 • C using an instrument (400 MHz VNMRS Varian, Tokyo, Japan) and deuterated chloroform (CDCl 3 ) as the solvent. The molar mass (M n ) and dispersity (Đ) of the polymer chains were investigated by gel permeation chromatography (GPC) using a GPC instrument (PL-GPC220, Agilent, Tokyo, Japan) equipped with GPC columns (Waters 515 pump, two PPS SDV 5 µm columns (diameter of 8 mm, length of 300 mm, 500 Å + 105 Å)) and a Waters 410 differential refractive index detector at 30 • C. The samples for NMR spectroscopy and GPC analyses were prepared by their dilution with CDCl 3 and THF, respectively, followed by the purification process, in which they were passed through a neutral alumina column. The neat GO and GO with a grafted polymer layer were observed using a transmission electron microscope (TEM, JEM-2100Plus, Jeol, Tokyo, Japan). The samples for the TEM analysis were prepared by dispersing the particles in acetone using mechanical stirring for 5 and 2 min of sonication and dropping the resultant suspension onto a copper grid. Fourier transform infrared (FTIR) spectra (64 scans, resolution of 4 cm −1 ) were recorded on a Nicolet 6700 (Nicolet, Madison, WI, USA) within a wavenumber range of 4000-600 cm −1 , and the ATR technique with a germanium crystal was employed. The spectra were recorded at room temperature. The Raman spectra (3 scans, resolution of 2 cm −1 ) were collected on a Nicolet DXR (Nicolet, Madison, WI, USA) using an excitation wavelength of 532 nm. The integration time was 30 s, and the laser power on the surface was set to 1 mW. The powders were compressed to form pellets (diameter of 13 mm, thickness of 1 mm) on a laboratory hydraulic press (Trystom Olomouc, H-62, Olomouc, Czech Republic). The pellets were used for electrical conductivity measurements, and a two-point method at room temperature was applied with the help of an electrometer (Keithley 6517B, Beaverton, OR, USA). The presented results are the average values from 10 independent measurements. The contact angle (CA) measurements were evaluated using the static sessile drop method on the pellets and were performed on a surface energy evaluation system equipped with a CCD camera (Advex Instruments, Brno, Czech Republic). A droplet (5 µL) of PDMS was carefully dropped onto the surface, and the CA value was recorded. The presented CA results are the average values from 10 independent measurements. To confirm that the contact angle results were not affected by the roughness of the GO surface of the investigated pellets, atomic force microscopy (AFM) was used to investigate the surface topography using an atomic force microscope (AFM), model Dimension ICON (Bruker, Billerica, MA, USA). Measurements were performed at a scan speed of 1 Hz with a resolution of 256 × 256 pixels in the tapping mode at room temperature in an air atmosphere. A silicone-nitride probe with a resonant frequency of (150 ± 50) kHz and a stiffness constant of 5 N/m (MPP-12120, Bruker, Billerica, MA, USA) was used. The image analysis and surface roughness (Sa) determination were performed using the program Gwyddion v.2.48 (Gwyddion, Brno, Czech Republic). The thermal conductivity was measured via a one-side contact method using the TCi model (C-term technologies, Fredericton, NB, Canada). The viscoelastic properties of both the nanocomposite and pure polymer matrix were studied via dynamic mechanical analysis (DMA) in the tensile mode. All the measurements were performed in the linear viscoelastic region. The measurements were performed at 1 Hz in the temperature range from −150 to 150 • C. Dielectric spectroscopy in the temperature range from −150 to 100 • C and in the frequency range from 10 −1 to 10 7 Hz was used to investigate the polymer chain dynamics. The glass transition process was evaluated using the activation energies calculated from the Arrhenius equation (Equation (1)) to see the effect of modification on the relaxation processes in the PDMS based composites. where E a is the activation energy, f ∞ is the pre-exponential factor, T is the temperature in Kelvin, and k B is the Boltzmann constant. Photoactuation The photoactuation ability of both the matrix and composite samples was investigated using a thermal mechanical analysis (TMA, Mettler Toledo, Langacher, Switzerland) similar to that previously published [14]. A red LED diode (Luxeon Rebell, Philips, Amsterdam, The Netherlands) was used for the irradiation. The irradiation was applied for 10 s at 627 nm with a 6 mW light source intensity under a 10% pre-strain of the samples. The maximum value of the actuation is characterized by a change in the sample length during the exposure to light, ∆L = (L 0 − L)/L 0 , where L 0 is the length of the non-irradiated sample, and L is the length of the irradiated sample. Synthesis of GO-Polymer-Modified Particles The modification of the GO particles with PMMA (GO-PMMA) and PBMA (GO-PBMA) polymer chains was performed via surface initiated ATRP, as previously described for other methacrylate-based monomers [29]. The MMA and BMA conversions, which were calculated from the 1 H NMR spectra, were 89% and 91%, respectively. Using a sacrificial initiator in addition to the initiator immobilized on the GO surface allowed for the determination of the molecular characteristics of the polymer chains with the consideration that the growth of both the free and GO surface-bonded polymer chains is comparable [30]. The molecular weight (M n ) and polydispersity index (Đ) determined for PMMA were 5620 g mol −1 and 1.18, respectively, and they were 5210 g·mol −1 and 1.21, respectively, for PBMA. The molar masses correlated well with the monomer conversions. The successful grafting process was confirmed by the FTIR investigations with on-line monitoring during the TGA measurements. In Figure 1a, a release of oxygen-containing groups from the neat GO surface can be seen in the temperature range of 150-300 • C. In the same temperature range, the FTIR spectrum in Figure 1b of the released compound was analyzed, and the vibrations from the hydroxyl groups at 3510 cm −1 , a small amount from the carbonyl groups at 1723 cm −1 and the stretching of the hydroxyl groups at 1423 cm −1 were found. The release of the groups from PMMA was observed in the TGA in the range of 220-380 • C, as seen in Figure 1c. The FTIR from the released groups showed an increase in the peaks of the carbonyl groups at approximately 1731 cm −1 in comparison to the neat GO, where this peak is almost negligible, and the appearance of peaks in the alkyl vibrations region of 2600-3000 cm −1 (Figure 1d), which confirmed the presence of PMMA on the GO surface. Similar results were obtained for the GO modified with the PBMA chains ( Figure 1e,f), but the release of the groups from PBMA occurred at higher temperatures. To quantify the amounts of the individual components in the GO, GO-PMMA and GO-PBMA particles, the TGA data were analyzed. It was found that, similar to our previous studies [28,29], the amount of the oxygen-containing groups for the neat GO was app. 30%. On the other hand, for both cases of polymer modification, the amount of polymer on the hybrid particles was app. 9% for PMMA and 11% for PBMA, and the amount of oxygen-containing groups was nearly 21% for PMMA and 19% for PBMA, which indicated partial reduction of GO surface during its modification as observed also previously [28]. Transmission Electron Microscopy In Figure 2, the TEM images of all the investigated GO particles are shown. In the case of neat GO, the proper exfoliation of the graphite powder using the Hummers method was achieved, and the neat GO sheet can be clearly seen on the TEM image ( Figure 2a). The sheet-like morphology was also observed in the case of the GO modified with the PMMA chains ( Figure 2b) when the 2D shape was sustained on the same level. The slightly darker contrast of the GO-PMMA sheet is due to the presence of the polymer layer on the surface of the GO. In the case of the GO-PBMA particles (Figure 2c), the 2D shape was again observed, which indicated they had the same morphology as the neat GO and GO-PMMA. Here, several layers of GO most likely lay on top of each other, and with the sustainable polymer layer of PBMA on the surface of the GO, they provide the higher contrast of the image. The sharp edges of the neat GO become smoother in the case of both the modified GOs due to presence of polymer chains, which again confirmed the successful modification of the GO surface. individual components in the GO, GO-PMMA and GO-PBMA particles, the TGA data were analyzed. It was found that, similar to our previous studies [28,29], the amount of the oxygen-containing groups for the neat GO was app. 30%. On the other hand, for both cases of polymer modification, the amount of polymer on the hybrid particles was app. 9% for PMMA and 11% for PBMA, and the amount of oxygen-containing groups was nearly 21% for PMMA and 19% for PBMA, which indicated partial reduction of GO surface during its modification as observed also previously [28]. Transmission Electron Microscopy In Figure 2, the TEM images of all the investigated GO particles are shown. In the case of neat GO, the proper exfoliation of the graphite powder using the Hummers method was achieved, and the neat GO sheet can be clearly seen on the TEM image (Figure 2a). The sheet-like morphology was also observed in the case of the GO modified with the PMMA chains (Figure 2b) when the 2D shape was sustained on the same level. The slightly darker contrast of the GO-PMMA sheet is due to the presence of the polymer layer on the surface of the GO. In the case of the GO-PBMA particles ( Figure 2c), the 2D shape was again observed, which indicated they had the same morphology as the neat GO and GO-PMMA. Here, several layers of GO most likely lay on top of each other, and with the sustainable polymer layer of PBMA on the surface of the GO, they provide the higher contrast of the image. The sharp edges of the neat GO become smoother in the case of both the modified GOs due to presence of polymer chains, which again confirmed the successful modification of the GO surface. Reduction of the GO Particles during the Synthesis In our previous work, we showed that the reduction of GO can occur during SI ATRP [28]. To confirm the reduction of the GO particles during the SI ATRP process, Raman spectroscopy was used as a useful tool to compare the broad D and G peaks corresponding to the sp 2 and sp 3 hybridized forms of the carbon atoms in the GO sheets. In Figure 3, a comparison among the neat GO sheets, GO-PMMA and GO-PBMA can be seen. A significant reduction in the GO sheets was achieved, and it can be observed from the significant change in the D/G peak intensity ratio (ID/IG) obtained after the SI ATRP process. Thus, a change from 0.90 to 1.05 and 1.08 was observed after the SI ATRP of MMA and BMA, respectively. It can also be seen that the 2D structure of GO created during the oxidation was nearly the same after modification with both the PMMA and PBMA brushes. The conductivity measurements confirmed the results obtained from the Raman spectroscopy, which indicated the reduction of GO after modification. The conductivity slightly increased from 1.2 × 10 −8 S cm −1 for the neat GO to 6.3 × 10 −8 S cm −1 for the GO-PMMA and 2.1 × 10 −7 S cm −1 for the GO-PBMA. Therefore, it can be stated that within the SI ATRP process, a partial reduction of the GO with the simultaneous grafting of PBA polymer brushes was achieved. Reduction of the GO Particles during the Synthesis In our previous work, we showed that the reduction of GO can occur during SI ATRP [28]. To confirm the reduction of the GO particles during the SI ATRP process, Raman spectroscopy was used as a useful tool to compare the broad D and G peaks corresponding to the sp 2 and sp 3 hybridized forms of the carbon atoms in the GO sheets. In Figure 3, a comparison among the neat GO sheets, GO-PMMA and GO-PBMA can be seen. A significant reduction in the GO sheets was achieved, and it can be observed from the significant change in the D/G peak intensity ratio (I D /I G ) obtained after the SI ATRP process. Thus, a change from 0.90 to 1.05 and 1.08 was observed after the SI ATRP of MMA and BMA, respectively. It can also be seen that the 2D structure of GO created during the oxidation was nearly the same after modification with both the PMMA and PBMA brushes. The conductivity measurements confirmed the results obtained from the Raman spectroscopy, which indicated the reduction of GO after modification. The conductivity slightly increased from 1.2 × 10 −8 S cm −1 for the neat GO to 6.3 × 10 −8 S cm −1 for the GO-PMMA and 2.1 × 10 −7 S cm −1 for the GO-PBMA. Therefore, it can be stated that within the SI ATRP process, a partial reduction of the GO with the simultaneous grafting of PBA polymer brushes was achieved. conductivity measurements confirmed the results obtained from the Raman spectroscopy, which indicated the reduction of GO after modification. The conductivity slightly increased from 1.2 × 10 −8 S cm −1 for the neat GO to 6.3 × 10 −8 S cm −1 for the GO-PMMA and 2.1 × 10 −7 S cm −1 for the GO-PBMA. Therefore, it can be stated that within the SI ATRP process, a partial reduction of the GO with the simultaneous grafting of PBA polymer brushes was achieved. Compatibility of the Particles with the PDMS Matrix The compatibility of the GO particles with the surrounding matrix when they are dispersed is a crucial part of their potential applications as light sensors or actuators [14,25]. Therefore, the compatibility between the neat GO or polymer-functionalized GO and PDMS was investigated via contact angle measurements. As seen in Figure 4a, the contact angle between the neat GO pellet and the PDMS droplet was 49.9 • ± 3.2 • , which indicated a relatively poor compatibility. The presence of the short polymer chains of PMMA improved the compatibility and decreased the contact angle to 38.7 • ± 2.7 • (Figure 4b). In addition, the longer aliphatic butyl chain present in PBMA decreased the contact angle to 28.7 • ± 2.7 • (Figure 4c). This result indicated significantly improved interactions between PDMS and the surface of the GO-PBMA in comparison with the surface of neat GO, which contains only hydroxyl, carbonyl, carboxyl or epoxy groups. To prove that these results were not affected by the surface roughness, an AFM investigation was performed. As seen in Figure 4d-f, the roughnesses were nearly identical for all the investigated samples, and the surface roughness (Sa) was found to be 154, 143 and 162 nm for GO, GO-PMMA and GO-PBMA, respectively. Therefore, it can be concluded that the contact angle was significantly affected by the chemical modification (coating) and that the roughness played a negligible role. Compatibility of the Particles with the PDMS Matrix The compatibility of the GO particles with the surrounding matrix when they are dispersed is a crucial part of their potential applications as light sensors or actuators [14,25]. Therefore, the compatibility between the neat GO or polymer-functionalized GO and PDMS was investigated via contact angle measurements. As seen in Figure 4a, the contact angle between the neat GO pellet and the PDMS droplet was 49.9° ± 3.2°, which indicated a relatively poor compatibility. The presence of the short polymer chains of PMMA improved the compatibility and decreased the contact angle to 38.7° ± 2.7° (Figure 4b). In addition, the longer aliphatic butyl chain present in PBMA decreased the contact angle to 28.7° ± 2.7° (Figure 4c). This result indicated significantly improved interactions between PDMS and the surface of the GO-PBMA in comparison with the surface of neat GO, which contains only hydroxyl, carbonyl, carboxyl or epoxy groups. To prove that these results were not affected by the surface roughness, an AFM investigation was performed. As seen in Figure 4d-f, the roughnesses were nearly identical for all the investigated samples, and the surface roughness (Sa) was found to be 154, 143 and 162 nm for GO, GO-PMMA and GO-PBMA, respectively. Therefore, it can be concluded that the contact angle was significantly affected by the chemical modification (coating) and that the roughness played a negligible role. Dielectric Investigation of the Polymer Chain Dynamics To investigate the influence of the GO surface modification on the polymer chain dynamics, dielectric spectroscopy measurements were performed. As seen in Figure 5, the presence of the neat GO in the PDMS matrix slightly affected the response of the PDMS polymer chains; however, in the case of GO-PMMA and GO-PBMA, the presence of short polymer grafts significantly shifted the glass transition temperature to lower values. To confirm this change, the activation energies of the glass transition process were calculated, and the effects of the filler nature and its loading were investigated. Within the range of the studied frequencies, a linear dependence of the Tg on the frequency was observed; therefore, a simple Arrhenius equation was used for the determination of the glass transition activation energies. The values are summarized in Table 1. The activation energy decreased with the increasing amount of filler. This indicated that the filler behaves as a softener in this case and makes the transition easier, and the overall movement of the main PDMS backbone Dielectric Investigation of the Polymer Chain Dynamics To investigate the influence of the GO surface modification on the polymer chain dynamics, dielectric spectroscopy measurements were performed. As seen in Figure 5, the presence of the neat GO in the PDMS matrix slightly affected the response of the PDMS polymer chains; however, in the case of GO-PMMA and GO-PBMA, the presence of short polymer grafts significantly shifted the glass transition temperature to lower values. To confirm this change, the activation energies of the glass transition process were calculated, and the effects of the filler nature and its loading were investigated. Within the range of the studied frequencies, a linear dependence of the T g on the frequency was observed; therefore, a simple Arrhenius equation was used for the determination of the glass transition activation energies. The values are summarized in Table 1. The activation energy decreased with the increasing amount of filler. This indicated that the filler behaves as a softener in this case and makes the transition easier, and the overall movement of the main PDMS backbone requires less exertion. Moreover, from the results, it can be clearly seen that the longer pendant aliphatic chain in PBMA caused a more significant softening of the polymer backbone, and, thus, the PDMS chains are more movable. This behavior should have a positive impact on the investigated photoactuation performance. Dynamic Mechanical Analysis Because photoactuation is a dynamic process, a dynamic mechanical analysis of the prepared composites is crucial to investigate their suitability for intended applications. Through this investigation, the interactions between the particles and the matrix can also be estimated. As seen in Figure 6, below the glass transition temperature, Tg, all the composites exhibited nearly the same mechanical properties (Figure 6a). However, due to the presence of various fillers, the Tg of the composites slightly changed. Moreover, another transition region connected to the melting of the small crystalline phase present in PDMS was observed, which was also observed elsewhere [31]. Only Dynamic Mechanical Analysis Because photoactuation is a dynamic process, a dynamic mechanical analysis of the prepared composites is crucial to investigate their suitability for intended applications. Through this investigation, the interactions between the particles and the matrix can also be estimated. As seen in Figure 6, below the glass transition temperature, T g , all the composites exhibited nearly the same mechanical properties (Figure 6a). However, due to the presence of various fillers, the T g of the composites slightly changed. Moreover, another transition region connected to the melting of the small crystalline phase present in PDMS was observed, which was also observed elsewhere [31]. Only a minor effect of the fillers was seen in the melting temperature region. In the region of utilization, the storage moduli were nearly independent of the temperature up to 40 • C. In this case, the best mechanical performance was observed for the neat GO. The reason for this could be possible covalent bonding between the OH groups and the PDMS polymer chains, which was observed by Bose et al. for OH groups of carbonyl iron and PDMS [32]. After coating of GO with the PMMA or PBMA chains, the partial physical entanglement provided an improved compatibility between PDMS and GO-PMMA or GO-PBMA compared with neat GO. However, at the same time, the short polymer chains at the GO surface can serve as plasticizers and thus the final PDMS material can be more flexible; but the PDMS was still reinforced in comparison to the neat PDMS matrix. The damping properties of the PDMS-containing polymer-grafted GO were slightly enhanced, as indicated by the higher tan delta values (Figure 6b). This is highly desirable from a photoactuation performance point of view because physical entanglements and GO-PBMA can provide the best flexibility in the prepared composite. In fact, this is in good agreement with the activation energies determined from the dielectric spectroscopy and will be confirmed by the photoactuation investigation (see below). Polymers 2017, 9, 264 9 of 13 a minor effect of the fillers was seen in the melting temperature region. In the region of utilization, the storage moduli were nearly independent of the temperature up to 40 °C. In this case, the best mechanical performance was observed for the neat GO. The reason for this could be possible covalent bonding between the OH groups and the PDMS polymer chains, which was observed by Bose et al. for OH groups of carbonyl iron and PDMS [32]. After coating of GO with the PMMA or PBMA chains, the partial physical entanglement provided an improved compatibility between PDMS and GO-PMMA or GO-PBMA compared with neat GO. However, at the same time, the short polymer chains at the GO surface can serve as plasticizers and thus the final PDMS material can be more flexible; but the PDMS was still reinforced in comparison to the neat PDMS matrix. The damping properties of the PDMS-containing polymer-grafted GO were slightly enhanced, as indicated by the higher tan delta values (Figure 6b). This is highly desirable from a photoactuation performance point of view because physical entanglements and GO-PBMA can provide the best flexibility in the prepared composite. In fact, this is in good agreement with the activation energies determined from the dielectric spectroscopy and will be confirmed by the photoactuation investigation (see below). To investigate the effect of the filler loading on the mechanical performance, the composites with the most promising properties, i.e., those containing GO-PBMA particles with various filler loadings (0.1, 0.5 and 1 vol %), were investigated (Figure 7). It can be clearly seen that below the Tg, the materials exhibited nearly the same behavior. Over the Tg, the samples with a higher filler loading (0.5 and 1 vol %) exhibited a significant drop in the storage moduli ( Figure 7a) with a subsequent increase to a maximum before melting. The drop can be caused by interactions between the GO-PBMA particles and the polymer matrix, which make the PDMS matrix more flexible, as described above. This behavior can also be connected to the decreased activation energy, which is in good agreement with our previous observations. After the melting of the crystalline phase, the present GO-PBMA particles still act as a reinforcing filler from a storage modulus point of view. The Tg slightly shifted to a lower temperature for all the filler contents (Figure 7b), which indicated the intervention of the GO-PBMA particles in the glass transition process. The right shoulder is most likely connected to the physical interactions of the polymer chains of GO-PBMA with the PDMS matrix, which were already described in Figure 7a. Moreover, the damping behavior of the composites was enhanced with an increase in the filler content, which indicated the relatively high energy dissipation caused by the more flexible structure present in the GO-PBMA/PDMS composites. It can be concluded that the modification of GO with various polymers can affect the mechanical performance; i.e., the storage moduli can be tailored using various modifications, and the polymer To investigate the effect of the filler loading on the mechanical performance, the composites with the most promising properties, i.e., those containing GO-PBMA particles with various filler loadings (0.1, 0.5 and 1 vol %), were investigated (Figure 7). It can be clearly seen that below the T g , the materials exhibited nearly the same behavior. Over the T g , the samples with a higher filler loading (0.5 and 1 vol %) exhibited a significant drop in the storage moduli ( Figure 7a) with a subsequent increase to a maximum before melting. The drop can be caused by interactions between the GO-PBMA particles and the polymer matrix, which make the PDMS matrix more flexible, as described above. This behavior can also be connected to the decreased activation energy, which is in good agreement with our previous observations. After the melting of the crystalline phase, the present GO-PBMA particles still act as a reinforcing filler from a storage modulus point of view. The T g slightly shifted to a lower temperature for all the filler contents (Figure 7b), which indicated the intervention of the GO-PBMA particles in the glass transition process. The right shoulder is most likely connected to the physical interactions of the polymer chains of GO-PBMA with the PDMS matrix, which were already described in Figure 7a. Moreover, the damping behavior of the composites was enhanced with an increase in the filler content, which indicated the relatively high energy dissipation caused by the more flexible structure present in the GO-PBMA/PDMS composites. It can be concluded that the modification of GO with various polymers can affect the mechanical performance; i.e., the storage moduli can be tailored using various modifications, and the polymer modification of GO particles in PDMS composites significantly contributes to the increasing energy dissipation due to the improved flexibility and provides an enhanced damping behavior. Thermal Conductivity According to a previous study [19], the thermal conductivity is another crucial parameter affecting the photoactuation performance. As seen in Table 2, the pure PDMS matrix has a thermal conductivity of 0.071 W mK −1 , and it increased with the increasing amount of filler for all the investigated samples. Due to the improved compatibility with the PMMA and PBMA polymer chains attached on the GO surface, these samples exhibited improved thermal conductivities compared to the neat GO. The best thermal conductivity was found for the composites containing the GO-PBMA particles. Thus, these composites can provide a significant contribution to enhanced photoactuation capabilities due to the significantly improved heat distribution within the samples. Photoactuation Performance The photoactuation performance was investigated for various composite compositions, and the impact of the filler modification and the filler loading on the final properties was elucidated. As seen in Figure 8, the pure matrix showed some actuation performance. The determined maximum value of actuation, ΔL, was 7.1 μm, and the recovery time was 30 s. The addition of 0.1 vol % of neat GO particles increased the maximum value of the actuation to 9.1 μm, and this was likely due to the improved heat transfer that was observed during the thermal conductivity measurement. However, the recovery time was the same as that for the pure PDMS matrix. In the case of the GO-PMMA- Thermal Conductivity According to a previous study [19], the thermal conductivity is another crucial parameter affecting the photoactuation performance. As seen in Table 2, the pure PDMS matrix has a thermal conductivity of 0.071 W mK −1 , and it increased with the increasing amount of filler for all the investigated samples. Due to the improved compatibility with the PMMA and PBMA polymer chains attached on the GO surface, these samples exhibited improved thermal conductivities compared to the neat GO. The best thermal conductivity was found for the composites containing the GO-PBMA particles. Thus, these composites can provide a significant contribution to enhanced photoactuation capabilities due to the significantly improved heat distribution within the samples. Photoactuation Performance The photoactuation performance was investigated for various composite compositions, and the impact of the filler modification and the filler loading on the final properties was elucidated. As seen Figure 8, the pure matrix showed some actuation performance. The determined maximum value of actuation, ∆L, was 7.1 µm, and the recovery time was 30 s. The addition of 0.1 vol % of neat GO particles increased the maximum value of the actuation to 9.1 µm, and this was likely due to the improved heat transfer that was observed during the thermal conductivity measurement. However, the recovery time was the same as that for the pure PDMS matrix. In the case of the GO-PMMA-based composite, the maximum actuation increased to 9.4 µm, and the recovery time was shortened to 25 s. A similar behavior was obtained for the sample containing GO-PBMA particles. The actuation performance was 11.8 µm, and the recovery time similar to that of the GO-PMMA composites was achieved. The main reason for this capability is the proper incorporation of the GO-PMMA and GO-PBMA particles into the PDMS matrix, which causes better shape recovery and improved heat transitions within the samples. The impact of the filler content on the photoactuation performance of the PDMS materials was investigated, and the results are plotted in Figure 9. In all cases, the ΔL increased with an increasing amount of filler. The best capability was found for the composites containing the GO-PBMA particles, and the maximum actuation value for a composite containing 1 vol % of GO-PBMA was approximately 5 times higher than that for pure PDMS. The main reason for this significant improvement was already proposed by previous investigations and is likely the enhanced flexibility of the polymer composite, which was confirmed by the lower activation energy, higher damping and enhanced thermal conductivity, improving the heat distribution within the composite samples. The polymer grafting from GO approach for the preparation of composites with photoactuation capabilities provides a system with an enhanced performance in comparison to other systems, including 2 wt % of neat CNTs, carbon black, GO or graphene nanoplatelets in the PDMS composites. The systems mentioned last showed a relative change in the length, below 25 μm, when recalculated for our conditions [18]. Conclusions In summary, this study investigated the influence of grafting polymer chains from a GO surface on the properties of the resulting PDMS-based composites. Short PMMA or PBMA polymer chains were grafted from the GO surface via SI-ATRP, and a slight reduction of the GO surface during the grafting was determined using Raman spectroscopy and conductivity measurements. The contact angle measurements confirmed the improved compatibility of the polymer-grafted GO particles with The impact of the filler content on the photoactuation performance of the PDMS materials was investigated, and the results are plotted in Figure 9. In all cases, the ∆L increased with an increasing amount of filler. The best capability was found for the composites containing the GO-PBMA particles, and the maximum actuation value for a composite containing 1 vol % of GO-PBMA was approximately 5 times higher than that for pure PDMS. The main reason for this significant improvement was already proposed by previous investigations and is likely the enhanced flexibility of the polymer composite, which was confirmed by the lower activation energy, higher damping and enhanced thermal conductivity, improving the heat distribution within the composite samples. The polymer grafting from GO approach for the preparation of composites with photoactuation capabilities provides a system with an enhanced performance in comparison to other systems, including 2 wt % of neat CNTs, carbon black, GO or graphene nanoplatelets in the PDMS composites. The systems mentioned last showed a relative change in the length, below 25 µm, when recalculated for our conditions [18]. The impact of the filler content on the photoactuation performance of the PDMS materials was investigated, and the results are plotted in Figure 9. In all cases, the ΔL increased with an increasing amount of filler. The best capability was found for the composites containing the GO-PBMA particles, and the maximum actuation value for a composite containing 1 vol % of GO-PBMA was approximately 5 times higher than that for pure PDMS. The main reason for this significant improvement was already proposed by previous investigations and is likely the enhanced flexibility of the polymer composite, which was confirmed by the lower activation energy, higher damping and enhanced thermal conductivity, improving the heat distribution within the composite samples. The polymer grafting from GO approach for the preparation of composites with photoactuation capabilities provides a system with an enhanced performance in comparison to other systems, including 2 wt % of neat CNTs, carbon black, GO or graphene nanoplatelets in the PDMS composites. The systems mentioned last showed a relative change in the length, below 25 μm, when recalculated for our conditions [18]. Conclusions In summary, this study investigated the influence of grafting polymer chains from a GO surface on the properties of the resulting PDMS-based composites. Short PMMA or PBMA polymer chains were grafted from the GO surface via SI-ATRP, and a slight reduction of the GO surface during the grafting was determined using Raman spectroscopy and conductivity measurements. The contact angle measurements confirmed the improved compatibility of the polymer-grafted GO particles with the PDMS matrix, especially when the GO was grafted with the PBMA chains. The polymer chain Figure 9. Dependence of the filler content on the change in the length for the pure PDMS matrix (black squares) and PDMS composites containing neat GO (red squares), GO-PMMA (blue circles) and GO-PBMA (green triangles). Error bars are not higher than the size of the symbols. Conclusions In summary, this study investigated the influence of grafting polymer chains from a GO surface on the properties of the resulting PDMS-based composites. Short PMMA or PBMA polymer chains were grafted from the GO surface via SI-ATRP, and a slight reduction of the GO surface during the grafting was determined using Raman spectroscopy and conductivity measurements. The contact angle measurements confirmed the improved compatibility of the polymer-grafted GO particles with the PDMS matrix, especially when the GO was grafted with the PBMA chains. The polymer chain dynamics were investigated using dielectric spectroscopy, and the Arrhenius equation was applied to calculate the activation energy of the relaxation around the glass transition temperature. It was found that the modifications caused a significant decrease in the activation energy and acted as a plasticizer for the PDMS matrix, which has a positive influence on the photoactuation behavior. This result was confirmed by the dynamic mechanical analysis. The thermal conductivity was also improved by the presence of the GO particles, and the polymer-grafted GO more significantly enhanced the thermal conductivity due to the better dispersibility of the particles in the PDMS matrix. Finally, the photoactuation performance was elucidated, and it was found that the composites including the GO-PBMA particles had the best capability due to the more flexible polymer chains and better heat distribution within the PDMS matrix.	2017-08-21T02:45:00.781Z	2017-07-01T00:00:00.000	{ "year": 2017, "sha1": "15db88a845d6a7978e04ff5500d223c0b8603955", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2073-4360/9/7/264/pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "15db88a845d6a7978e04ff5500d223c0b8603955", "s2fieldsofstudy": [ "Materials Science", "Engineering" ], "extfieldsofstudy": [ "Materials Science", "Medicine" ] }
2208810	pes2o/s2orc	v3-fos-license	Flag-symmetry of the poset of shuffles and a local action of the symmetric group We show that the poset of shuffles introduced by Greene in 1988 is flag-symmetric, and we describe a"local"permutation action of the symmetric group on the maximal chains which is closely related to the flag symmetric function of the poset. A key tool is provided by a new labeling of the maximal chains of a poset of shuffles, which is also used to give bijective proofs of enumerative properties originally obtained by Greene. In addition we define a monoid of multiplicative functions on all posets of shuffles and describe this monoid in terms of a new operation on power series in two variables. Introduction In [St2], Stanley initiated an investigation of posets which involve two algebraic objects related to the order structure of the poset -a certain symmetric function (flag symmetric function) and a certain associated representation of the symmetric group. In Section 1 we give precise definitions and summarize the results of [St2] that will be used later in this paper. Briefly, [St2] is concerned with classes of posets whose order structure leads to a symmetric function derived from the enumeration of rank-selected chains, and which turns out to be the Frobenius characteristic of a representation of the symmetric group, of degree equal to the number of maximal chains of the poset; moreover, this representation can be realized via an action of the symmetric group on the maximal chains of the poset, under which each adjacent transposition σ i = (i, i + 1) acts on chains locally, that is, modifying at most the chain element of rank i. The goal of this paper is to add a new infinite family of posets to the examples appearing in [St2] and [St3], namely, the posets of shuffles introduced and investigated by Greene [Gre1]. In the process, several general results emerge. In Section 1 we give the necessary background on locally rank-symmetric posets affording a local action of the symmetric group (based on [St2]). Section 2 contains the necessary preliminaries concerning the posets of shuffles (i.e., shuffles of subwords of two given words). In Section 3 we give a new labeling of the posets of shuffles and establish its properties which are instrumental in the remainder of the paper. In Section 4 we describe a local action of the symmetric group on the maximal chains of a poset of shuffles, such that the Frobenius characteristic for the corresponding representation character is (essentially) the flag symmetric function. The desired results regarding the posets of shuffles follow from more general results motivated by the properties of the new labeling of these posets. Section 5 is devoted to the enumeration of shuffles according to a natural notion of type. As a consequence we describe the monoid of multiplicative functions on the poset of shuffles in terms of operations on power series. As a by-product of the present investigation of the posets of shuffles, we obtain alternative, purely combinatorial, derivations of enumerative results obtained in [Gre1]. The present work parallels that of [St3] regarding the lattice of noncrossing partitions, thus adding to previously known structural analogies between the posets of non-crossing partitions and those of shuffles. It is hoped that this work will facilitate the development of a systematic general theory of the posets with a local group action concordant with the flag symmetric function. Preliminaries Let P be a finite poset with a minimum element0, and a maximum element1. Throughout this paper, we will consider only such posets that are ranked, that is, there exists a function ρ: P → Z such that ρ(0) = 0 and ρ(t ′ ) = ρ(t) + 1 whenever t< · t ′ (the notation t< · t ′ means that t is covered by t ′ , i.e., t < t ′ and there is no element u ∈ P such that t < u < t ′ ). The flag f -and h-vectors appear in numerous contexts in algebraic and geometric combinatorics; for instance, the values β P (S) have topological significance related to the order complex of the rank-selected subposet P S : = {0,1} ∪ {t ∈ P : ρ(t) ∈ S} (see, e.g., [St1,section 3.12] for additional information and references). Consider now the formal power series F P (x): = F P (x 1 , x 2 , . . .) = 0 ≤t 1 ≤t 2 ≤···≤t k <1 x ρ(t 1 ) 1 x ρ(t 2 )−ρ(t 1 ) 2 · · · x n−ρ(t k ) k+1 . This definition was suggested for investigation by Richard Ehrenborg [E] and is one of the central objects in [St2] and in this paper. Alternatively, F P (x) = S⊆[n−1] S={s 1 <s 2 <···<s k } α P (S) · 1≤i 1 <i 2 <···<i k+1 It is easy to see that the series F P (x) is homogeneous of degree n and that it is a quasisymmetric function, that is, for every sequence n 1 , n 2 , . . . , n m of exponents, the monomials x n 1 i 1 x n 2 i 2 · · · x n m i m and x n 1 j 1 x n 2 j 2 · · · x n m j m appear with equal coefficients whenever i 1 < i 2 < · · · < i m and j 1 < j 2 < · · · < j m . Through a simple counting argument and using the relation (1), the series F P (x) can also be rewritten as where the L S,n (x) are Gessel's quasisymmetric functions which constitute a basis for the (2 n−1 -dimensional) space of quasisymmetric functions of degree n (for more on quasisymmetric functions and symmetric functions we refer the interested reader to [Re] and [M]). A first question discussed in [St2] is that of conditions under which F P (x) is actually a symmetric function, in which case we refer to F P (x) as the flag symmetric function of P and to P as a flag-symmetric poset. An immediate necessary condition is that P be ranksymmetric (i.e., #{t ∈ P : ρ(t) = r} = #{t ∈ P : ρ(t) = n − r} for every 0 ≤ r ≤ n). A necessary and sufficient condition can be deduced readily from (2) Namely, for every S ⊆ [n − 1] the value of α P (S) depends only on the (multi)set of differences s 1 − 0, s 2 − s 1 , . . . , s k − s k−1 , n − s k and not on their ordering. If this is the case, then the symmetric function F P (x) can be expressed in terms of the basis of monomial symmetric where λ ⊢ n denotes a partition λ = (λ 1 ≥ λ 2 ≥ · · · ≥ λ l > 0) of n, and α P (λ): = α P ({λ 1 , λ 1 + λ 2 , . . . , λ 1 + λ 2 + . . . + λ l−1 }). The following sufficient condition for F P (x) to be a symmetric function introduces the class of locally rank-symmetric posets. This condition is not necessary for flag-symmetry, as shown by the poset of Figure 2, but it is necessary and sufficient for every interval of P to be flag-symmetric. Proposition 1.1 [St2,Theorem 1.4] Let P be a ranked poset with0 and1. If P is locally rank-symmetric, i.e., if every interval in P is a rank-symmetric (sub)poset, then F P (x) is a symmetric function. Locally rank-symmetric posets turn out to be a rich source of examples yielding flag symmetric functions. The examples of flag-symmetric posets provided in [St2] include products of chains (shown to be the only flag-symmetric distributive lattices, and identical to the class of locally rank-symmetric distributive lattices), and Hall lattices (a "q-analogue" of a product of chains), as well as a discussion of some other classes of posets. If F P (x) is a symmetric function, homogeneous of degree n, and if it turns out to be Schur positive (i.e., its expression in terms of the Schur functions basis has nonnegative coefficients only), then it follows from the general theory of representations and symmetric functions that it is the Frobenius characteristic of a character ψ of the symmetric group S n . In the preceding display line, λ runs over all partitions of n, ψ(λ) is the value of ψ on the conjugacy class of type λ, z λ = 1/(λ 1 λ 2 · · · m 1 !m 2 ! · · ·) with m i being the multiplicity of i as a part of λ, and p λ (x) is the power symmetric function indexed by λ (that is, p λ (x) = p λ 1 (x)p λ 2 (x) · · · with p j (x): = x j 1 + x j 2 + · · ·). It is known that when the Frobenius characteristic of a character ψ of S n is expanded in terms of Schur functions {s λ (x)} λ⊢n , then the coefficient of s λ (x) is the multiplicity with which the irreducible character χ λ of S n occurs in ψ. Thus, F P (x) describes a representation of S n , whose degree ψ(1 n ) can be recovered as the coefficient of m 1 n in F P (x). In view of (4), the degree of ψ is α P (1 n ), the number of maximal chains in P . The preceding discussion suggests seeking a natural action of S n on the complex vector space CM(P ) with the set M(P ) of maximal chains in P as a basis, giving rise to a representation of S n with character ψ as in (5). Of particular interest would be a local action with this property (defined in [St2] and motivated by the notion of local stationary algebra appearing in [V]); that is, an action such that for every adjacent transposition σ i = (i, i + 1) and every maximal chain m of P we have with nonzero coefficient c mm ′ only if m ′ differs from m at most in the element of rank i. Following [St2], we call such an action good. Good actions of the symmetric group are discussed in [St2] in the case of posets whose rank-two intervals are isomorphic to C 3 or C 2 × C 2 (where C i denotes an i-element chain), and for posets whose rank-three intervals are isomorphic to C 4 or C 3 × C 2 or C 2 × C 2 × C 2 . These results are based on work of David Grabiner [Gra]. Another illustration in [St2] gives a local action of the Hecke algebra of S n on CM(B n (q)), where B n (q) denotes the lattice of subspaces of an n-dimensional vector space over GF q . In [St3] a good action is exhibited for the lattice of noncrossing partitions. To these classes of examples this paper adds the posets of shuffles. We note that related results were recently obtained by Patricia Hersh [He] (generalizing the local S n -action on noncrossing partitions), and Jonathan Farley and Stefan Schmidt [FaSc] (generalizing the work of Grabiner [Gra]). Flag-symmetry of the posets of shuffles Let A = {a 1 , a 2 , . . . , a M } and X = {x 1 , x 2 , . . . , x N } be two (finite) disjoint sets which we will call the lower and the upper alphabets, respectively. Consider the collection of shuffles over A and X , that is, words w = w 1 w 2 · · · w k with distinct letters from A ∪ X satisfying the shuffle property: the subset of letters belonging to each alphabet appears in increasing order of the letter-subscripts in the appropriate alphabet. For instance, if M = 4 and N = 3, then w = x 2 a 1 a 3 x 3 is a shuffle word, but w = a 1 x 2 a 2 a 3 x 1 is not a shuffle word. Note that the empty word ∅ is a shuffle word. The poset of shuffles W M N consists of the shuffle words over alphabets A and X with #A = M and #X = N with the order relation given by w< · w ′ iff w ′ is obtained from w either by deleting a letter belonging to A or by inserting (in an allowable position) a letter belonging to X . In particular,0 = a 1 a 2 · · · a M and1 = x 1 x 2 · · · x N . Figure 1 shows the Hasse diagram of W 21 . Clearly, W 0N and W M 0 are isomorphic to the boolean lattices of rank N and M , respectively. We will write {w} for the set of letters of a shuffle word w. Greene [Gre1] investigated the posets of shuffles, whose definition was motivated by an idealized model considered in mathematical biology. The results established in [Gre1] include structural properties of W M N (e.g., W M N is a ranked poset; it admits a decomposition into symmetrically embedded boolean lattices and, hence, a symmetric chain decomposition; W M N is an EL-shellable poset), as well as expressions for key invariants of W M N (the zeta polynomial, the number of maximal chains, the Möbius function, the rank generating function, the characteristic polynomial). Two of the formulas in [Gre1] will arise later in this paper. Proposition 2.1. [Gre1,Theorem 3.4] The number of maximal chains in W M N is given by The Möbius function of W M N is We now turn to the interval structure of the posets of shuffles. Lemma 2.2. Every interval in a poset of shuffles is isomorphic to a product of posets of shuffles. Proof. Let [u, w] be an arbitrary interval in W M N , and write u = u 1 u 2 · · · u r , w = w 1 w 2 · · · w s . Let u i 1 u i 2 · · · u i t and w j 1 w j 2 · · · w j t be the subwords of u and w, respectively, formed by the letters common to the two words. Because u < w, the shuffle property implies u i p = w j p for each p = 1, 2, . . . , t. Moreover, the remaining letters of u belong to the alphabet A and the remaining letters of w belong to the alphabet X . Therefore the interval [u, w] is isomorphic to the product of the posets of shuffles W i p −i p−1 −1,j p −j p−1 −1 for p = 1, 2, . . . , t + 1, where we set i 0 = j 0 = 0, i t+1 = r + 1 and j t+1 = s + 1. For example, if u = a 2 x 3 a 4 a 5 a 10 x 6 x 8 and v = x 1 x 2 x 3 x 5 a 10 x 6 x 8 x 10 x 11 in W 10,15 , then r = 7, s = 9 and there are t = 4 letters common to the two words. These form the word Remark 2.3. Of course, factors of the form W 00 are singleton posets and can be discarded from the product, and W i0 ≃ W 0i ≃ B i , the boolean lattice with i atoms. Using the notation from the proof of Lemma 2.2, we will write [u, w]≃ c p W i p −i p−1 −1,j p −j p−1 −1 , the canonical isomorphism type of the interval [u, w]. The notion of canonical isomorphism type of an interval will be used in Section 5. leading to where e j = 1≤i 1 <i 2 <···<i j x i 1 x i 2 · · · x i j , the jth elementary symmetric function in variables x 1 , x 2 , . . ., and p 2 = x 2 i . Consequently, For example, the calculation of the flag f -vector of W 21 done in Section 1 gives F 21 (x) = We omit the details of this argument. Instead, we will obtain the expression (10) A labeling for posets of shuffles To describe an action of S M +N on the maximal chains of W M N , it would be natural to resort to a labeling of the chains and have the symmetric group act on the chains by acting on their label sequences simply by permuting coordinates. The poset W M N is already known to be EL-shellable [Gre1], through the labeling of each covering relation u< · w by the unique letter in the symmetric difference of the sets of letters {u} and {w}, and with the ordering a 1 < a 2 < · · · < a M < x 1 < x 2 · · · < x N for the labels. Under this labeling each maximal chain is labeled by a permutation in S M +N . However, this does not serve well the goal of describing an S M +N action on the maximal chains. A similar situation occurred in [St3], where the standard EL-labeling of the noncrossing partition lattice was not suitable for describing a local action of the symmetric group on the maximal chains, and a new EL-labeling was produced for this purpose. Here too, we will define a new labeling for a poset of shuffles which lends itself naturally to the description of the desired local action of S M +N . By a labeling we mean a map Λ: M(P ) → L n , written where n is the length of the maximal chains of P , and L is a totally ordered set. The labeling of interest in the present paper is a C-labeling, that is, for every maximal chain c = (0 = w 0 < · w 1 < · · · · < · w n =1) and every r ∈ [n], the label Λ r (c) depends only on the initial subchain (0 = w 0 < · w 1 < · · · · < · w r ). If the label Λ r (c) depends only on the covering w r−1 < · w r , and not on the maximal chain c itself, then Λ is an E-labeling. Three properties of labelings will play a role in this paper: the R * -, R-, and S-labeling has a unique completion by covering relations where c is any maximal chain beginning0 = w 0 < · w 1 < · · · · < · w s . (By the definition of Clabeling, the remaining elements of c do not affect the labels Λ i (c) for 1 ≤ i ≤ s.) In the same setting as for an R * -labeling, the requirement for an R-labeling is the existence of a unique weakly increasing completion of the chain: A labeling Λ of the maximal chains of a poset is an S-labeling if it is one-to-one and if for every maximal chain c = (0 = w 0 < · w 1 < · · · · < · w n =1) and for every rank i ∈ [n − 1] such that We now turn to the definition of a labeling Λ for the poset of shuffles, and then show that it has the properties R * and S. In the next section we will see the implications of an RS-or R * S-labeling with regard to a local action on the poset. To each maximal chain c by assigning a label from A ∪ X to each covering relation on c. In defining Λ we distinguish three types of covering relations, (x), (xa), and (a), as follows: where this is the first deletion along c, starting from0, of a letter (necessarily belonging to A) located immediately after x k ; then we set Λ i+1 (c) = x k . (a) w i < · w i+1 with w i+1 obtained from w i by deleting a letter a j ∈ A, and this deletion is not of type (xa); then we set Λ i+1 (c) = a j . Proof. From the definition of Λ it is clear that all letters in X appear in the label sequence of any maximal chain c and that for every a j ∈ A which does not appear in the label sequence, there is an x k ∈ X which appears twice. Thus, the label sequence of every maximal chain c is of the claimed form. Conversely, we claim that given a multipermutation σ of A ∪ 2X ∪ (X − X) for some A and X as in the statement of the Lemma, there is a unique maximal chain in W M N having label sequence Λ(c) = σ. Indeed, first note that if A = A and X = ∅ (that is, σ is a permutation of A ∪ X ), then only coverings of type (a) and (x) are possible. Thus, starting from0 = w 0 , σ dictates a sequence of deletions of letters from A and insertions of letters from X , each insertion being made in the rightmost possible position. This determines a unique maximal chain c as desired. For example, for W 23 , the permutation σ = a 2 x 3 x 1 a 1 x 2 determines the Observe that the shuffle condition implies that if the pairs x m , a p and x n , a q are involved in coverings of type (xa), then m = n and p = q, and m < n if and only if p < q. Therefore, in the multipermutation σ, the second occurrence of x j r must correspond to a covering of type (xa) involving the pair of letters x j r , a i r , for each r = 1, . . . , k. The first occurrence of x j r in σ is forced to correspond to the insertion of x j r immediately in front of a i r , and for each x t ∈ X, its unique occurrence in σ forces the insertion of x t in the rightmost position possible to the left of a i r and/or x j r , if j r = min{s > t : x s ∈ X} (if this set is empty, then x t is inserted in the rightmost position possible). As in the preceding case, a unique maximal chain c is determined by σ. For example, The coverings of type (xa) must involve the pairs x 2 , a 1 and x 4 , a 3 . From σ we reconstruct the chain The behavior of Λ on intervals of rank two can be easily described. Lemma 3.2. For every rank-two interval in a poset of shuffles W M N , the labeling Λ conforms to one of the following cases: 3.2.1) If a rank-two interval is isomorphic to C 2 × C 2 , then its two chains c 1 and c 2 have label sequences of the form Λ(c 1 ) = (l 1 , l 2 ) and Λ(c 2 ) = (l 2 , l 1 ), where l 1 and l 2 are distinct letters from A ∪ X . 3.2.2) If a rank-two interval is isomorphic to Π 3 , then its three chains γ 1 , γ 2 and γ 3 have Proof. Each rank-two interval of a poset of shuffles has either 4 or 5 elements. That is, each rank-two interval is isomorphic either to C 2 × C 2 or to the lattice Π 3 of partitions of a 3-element set. Specifically, an interval of rank 2 is of one of the following forms: ii) [ua p v, ux m v] for some words u, v. Such an interval is isomorphic to W 11 ≃ Π 3 . The definition of the chain labeling Λ and the two possible structures of the intervals of rank 2 yield the two cases in the desired conclusion. Proposition 3.3. The labeling Λ is an S-labeling on the poset of shuffles. Proof. This follows immediately from Lemma 3.1 and Lemma 3.2. Proposition 3.4. Consider the ordering a 1 < a 2 < · · · < a M < x 1 < x 2 < · · · < x N on the union of the two alphabets. Then the labeling Λ is an R * -labeling of the poset of shuffles. Proof. Let0 = w 0 < · w 1 < · · · · < · w r < v be a chain in W , [BjGaSt]). Indeed, the unique strictly increasing0-u-chain guaranteed by the preceding result can be taken as the "root" of each interval [u, v], and the labeling Λ * defined Remark 3.6. The proof of Proposition 3.4 shows that the R * -labeling Λ has a stronger property: the unique increasingly labeled extension of a chain0 = w 0 < · w 1 < · · · · < · w r < v depends only on w r and v. We will write γ(w r , v) to denote this chain. The remainder of this section is devoted to enumerative consequences of the labeling Λ, yielding combinatorial proofs of results from [Gre1]. We begin with a bijective proof of the local rank-symmetry of the posets of shuffles (an inductive proof was given in Proposition 2.4). In particular, this is a bijective proof of the rank-symmetry of a poset of shuffles. An alternative bijective proof of the rank-symmetry of W M N is implicit in the symmetric chain decomposition which appears in [Gre1]. Proof. Let v ∈ [u, w] be an element of rank ρ(u)+i. Consider the maximal chain c (u, v, w) formed by concatenating γ(0, u), γ(u, v), γ(v, w), and γ(w,1). Let c ′ (u, v, w) be the unique maximal chain whose label sequence is the concatenation of Λ(γ(0, u)), Λ(γ(v, w)), Λ(γ(u, v)), and Λ(γ(w,1)). Define ϕ(v) to be the element of rank ρ(w) − i on the chain c ′ (u, v, w). It is easy to see (from the definition and injectivity of Λ) that c ′ (u, v, w) contains the elements u and w and that ϕ establishes a bijection between the rank-(ρ(u) + i) and the rank-(ρ(w) − i) elements in the interval [u, w]. Proof. We may count the increasingly labeled chains γ(0, w) since they are in bijection with the elements w ∈ W M N . For a prescribed number k ≥ 0 of coverings of type (xa), the set of labels along such a chain is determined by the choice of k pairs from A × X for the coverings of type (xa), and an arbitrary subset of the complement in A ∪ X of the letters chosen for the k pairs. Lemma 3.1 yields readily the number of maximal chains in a poset of shuffles, giving a more direct derivation of the formula (6) due to C. Greene. Corollary 3.9. The number of maximal chains in the poset of shuffles W M N is Proof. By Lemma 3.1, we can count the maximal chains in W M N by counting the possible label sequences Λ(c). For each value k ≥ 0, the kth term in the sum gives the number of multipermutations Λ(c) in which k letters of X appear with multiplicity 2, while k of the letters of A do not occur in σ. From the R * -labeling Λ we can recover the formula (7) for the Möbius function of a poset of shuffles, which was obtained in [Gre1] using an EL-labeling as well as through an alternative computation. Corollary 3.10. The Möbius function of the poset of shuffles W M N is given by Proof. By the general theory of [BjW], the Möbius function is (−1) M +N times the number of maximal chains to which the R * -labeling Λ assigns weakly decreasing label sequences. From Lemma 3.1 it follows that such chains have label sequences of the form for some 0 ≤ k ≤ min{M, N }, where i M −k > i M −k−1 > · · · > i 1 and j N+k ≥ j N+k−1 ≥ · · · ≥ j 1 with k non-consecutive equalities. Therefore, the decreasingly labeled maximal chains correspond bijectively to the selections of M − k letters from A and k letters from X for some k. It is an easy exercise to show that the number of such selections is M +N M yielding the desired formula for the Möbius function. A local action of the symmetric group We begin with two general results which imply that the posets of shuffles have a local action of the symmetric group and establish the relation between the Frobenius characteristic of the corresponding character and the flag symmetric function of the poset. Theorem 4.1. Suppose P is a finite ranked poset of rank n, with0 and1. If P has an S-labeling Λ, then the action of S n on labels by permuting coordinates induces a local (permutation) action on the maximal chains of P . The adjacent transposition σ i = (i, i + 1) acts on Λ(c) by interchanging Λ i and Λ i+1 . By definition of S-labeling, there is a unique maximal chain c ′ such that Λ(c ′ ) = σ i · Λ(c). Since the σ i 's generate S n , we get an action of S n on the set of labels of maximal chains, and hence on M(P ). Moreover, this action is local by the definition of an S-labeling. Observation 4.2.2. A special case is when the maximal chains in each orbit form a subposet isomorphic to a product of chains, C ν 1 +1 × C ν 2 +1 × · · ·. It is not hard to show that this is the case for the posets of shuffles and the action discussed here, as well as for the lattice of noncrossing partitions discussed in [St3]. Thus, in addition to admitting a partition of the elements into boolean lattices (as shown in [Gre1] for poset of shuffles and in [SiU] for the noncrossing partition lattice), these posets also admit a partition of their maximal chains into products of chains. Figure where a ν ≥ 0. Proof. (a) Let θ ∈ stab(c), and let i be the least element of [n] for which θ −1 (i) = j > i. Note that only one factor of the last product above moves i 1 , namely, σ i 1 . Let t be the element of c of rank i. It follows from the definition of local action that t is also an element of the chain c ′ = σ i 2 · · · σ i r −2 σ i r −1 · c. Let s be the element of the chain c ′′ = σ i 1 σ i 2 · · · σ i r −1 · c of rank i. Then again by definition of local action, s is an element of the chain θ · c (since the factors to the left of σ i 1 in (11) can be written as products of σ p 's with p > i). Since θ · c = c, we have s = t. Thus σ i 1 · c ′ = c ′ , so we can remove the factor σ i 1 from the product (11) and still get a permutation θ ′ ∈ stab(c). But Hence (i 1 , i r ) ∈ stab(c), as claimed. It follows by induction on i (as defined above) that if for any a, b ∈ [n] we have θ(a) = b, then (a, b) ∈ stab(c). From this it is clear that stab(c) is a Young subgroup of S n . (b) By (a), the S n -action on M(P ), when restricted to an orbit O ∈ M(P )/S n , is equivalent to the action of S n on the set S n /S ν of cosets of some Young subgroup S ν = S ν 1 ×S ν 2 ×· · ·, where ν = ν O ⊢ n. If ψ ν is the character of this action of S n on S n /S ν , then ch(ψ ν ) = h ν . (b) If Λ is an R * S-labeling, then ch(ψ) = ωF P , where ω is the standard involution on symmetric functions [M, p.21]. Hence F P is an e-positive symmetric function. Since the action of S n by permuting coordinates of permutations of multiset of type ν has Frobenius characteristic h ν , we get F P = ch(ψ). Since there is a unique weakly increasing maximal chain of P from0 to1 (equivalently, since α P (∅) = 1), we get F P = h ν for some ν ⊢ n. The expression (10) In the remainder of this section we make comments regarding the preceding results and discuss other possible directions for generalizations. Remark 4.6. Theorems 4.1 and 4.4 apply to posets which are products of chains and also to the lattice of noncrossing partitions. The corresponding conclusions are established directly in [St2] and [St3]. Remark 4.7. The power series F P (x) may be viewed in a broader context. For a function ϕ in the incidence algebra (see, e.g., [St1, section 3.6]) of a ranked poset P , define where the sum ranges over the chains in P whose rank-support is the set S ⊆ [n − 1], and n is the rank of P . Now define Note that F P (ϕ, x) is a quasisymmetric function, homogeneous of degree n. In particular, if ϕ = ζ, the zeta function of P (i.e., ζ(u, v) = 1 if u ≤ v, and ζ(u, v) = 0 otherwise), we recover α P (ζ, S) = α P (S) and F P (ζ, x) is the function F P (x) of (2). We intend to pursue this generalization elsewhere, mentioning here only one result -the next proposition. We note that the same result holds for an arbitrary invertible element ϕ from the incidence algebra of P and its inverse. Here we present a direct proof for the special case ϕ = ζ and ϕ −1 = µ which is the instance occurring in the context of this paper. Proposition 4.7.1. Let P be a ranked poset of rank n, having elements0 and1. If ζ and µ are, as usual, the zeta function and the Möbius function of P , then where ω is the involution on quasisymmetric functions defined by ωL S,n (x) = L S,n (x), with S denoting the complement of S in [n − 1]. (When k = 0, the chain0 <1 gives the term µ(0,1).) By grouping the chains in P S according to their rank-support, U ⊆ S, we obtain Finally, by the definition of L V,n (x) and an inclusion-exclusion argument, the inner sum over Since ω restricted to symmetric functions agrees with the standard involution ω, the expressions for the characteristic ch(ψ) from Theorem 4.4 can be restated as (a) ch(ψ) = F P (ζ, x) = h ν when P is RS-labeled, and (b) ch(ψ) = (−1) n F P (µ, x) when P is R * S-labeled. Remark 4.8. The S M +N -action of Corollary 4.5 is a permutation action on the maximal chains of W M N , for which each orbit consists of those maximal chains whose label sequence Λ(c) is a permutation of the same multiset of letters. Thus, the explanation of formula (6) for the maximal chains of W M N provided in the proof of Corollary 3.9 amounts to counting the maximal chains according to their orbit, and grouping the orbits according to the type 2 k 1 M +N−2k of the multiset of the chain labels. Remark 4.9. An S-labeling does not ensure that each orbit of maximal chains is isomorphic to a product of chains. The poset shown in Figure 5, suggested to us by H. Barcelo, has a local action induced from the action of S 4 on the labels, but the orbit of chains labeled by the multiset 1122 is not a product of chains. In fact, no product of chains other than the trivial one, C 5 , occurs as a subposet of rank four in this poset. Remark 4.10. The converse of Theorem 4.1 does not hold. That is, a chain labeling such that the action of S n on labels induces a local action is not necessarily an S-labeling. The poset P shown in Figure 6(a) has a labeling of its maximal chains which, although not injective, gives rise to an S 3 local action on the maximal chains of P . The orbits are four copies of C 2 × C 3 , each labeled in the standard way with the multiset a, b, b. On the other hand, the noninjective labeling of Figure 6(b) does not produce an S 3 action on the maximal chains (e.g., σ 1 σ 2 σ 1 (0< · A< · B< ·1) = σ 2 σ 1 σ 2 (0< · A< · B< ·1)). a) The action on labels induces a local action. b) The action on labels does not induce a local action. Finally, we give a local condition which characterizes labeled posets with a local action induced from the action on labels. This, of course, can be seen to apply to the earlier examples. A local action is induced from the S n -action on labels if and only if Λ satisfies the following condition for every maximal chain c and every value of i ∈ {1, 2, . . . , n − 2}: The length-three chains δ from t i−1 to t i+2 with labels induced by restricting Λ(τ δθ) can be partitioned so that (a) each class is isomorphic to C 2 × C 2 × C 2 or C 3 × C 2 or C 4 , and (b) the labeling in each class coincides with the standard labeling of a product of chains by join-irreducibles. Proof. The conditions (a) and (b) on Λ imply readily the Coxeter relations for S n , showing that the local action is well-defined. Conversely, within each orbit of chains δ, the local action fixes the chains labeled aaa, so these form classes isomorphic to C 4 ; a chain δ labeled with aba is mapped under the local action to chains with the same τ and θ and label sequences aab and baa, structured as a copy of C 2 × C 3 and forming a class as claimed; similarly, a chain δ labeled as abc is mapped by the subgroup generated by σ i and σ i+1 to six chains forming a copy of C 2 × C 2 × C 2 , with labels as claimed. Multiplicative functions on the poset of shuffles Consider now the poset W ∞∞ whose elements are the shuffles of finite words using the lower alphabet A ∞ = {a 1 , a 2 , . . .} and the upper alphabet X ∞ = {x 1 , x 2 , . . .}. The comparability relation is as in the case of finite alphabets. A multiplicative function on W ∞∞ is a function f defined on the intervals in W ∞∞ for which f 00 = 1 and which has the property that if where we write f ij for the value of f on an interval canonically isomorphic to W ij (see Remark 2.3). Let f and g be two multiplicative functions on W ∞∞ , and let where f * g denotes convolution in the incidence algebra I(W ∞∞ ). The main result of this section, Theorem 5.2, shows how to express F * G in terms of F and G, and hence "determines" the monoid of multiplicative functions on W ∞∞ . We begin by establishing an expression for the number of elements w ∈ W M N of a given ij . Note that the canonical isomorphism (Remark 2.3) implies that 1 + ib ij = a ij and 1 + ja ij = b ij , and we can recover from the type of a word w the values m: and n: = #({w} ∩ X ) = i,j j (a ij + b ij ). Also, the type of w ∈ W M N determines M and N , so the enumeration of shuffle words by type can be done in W ∞∞ . Proposition 5.1. Let (a ij ) and (b ij ) be nonnegative integers such that Then the number of elements w ∈ W ∞∞ whose type is ((a ij ), (b ij )) is given by Proof. Each w ∈ W ∞∞ − {∅} is of the form U LU L · · · U L, or LU LU · · · LU , or LU · · · LU L, or U L · · · U LU , where each U is a nonempty factor whose letters are from the upper alphabet, and each L is a nonempty factor whose letters are from the lower alphabet. If the type of w is ((a ij ), (b ij )) then, for each j ≥ 1, the number of U -factors of length j is i a ij and, for each i ≥ 1, the number of L-factors of length i in w is equal to j b ij . The alternation of nonempty L-and U -factors imposes the condition ǫ ∈ {1, 0, −1} appearing in the hypothesis. To construct a word w of the prescribed type, we begin by deciding the length of each Uand L-factor. The number of possibilities is the number of (multi)permutations of the nonzero lengths, so that U -and L-factors alternate: r a 01 , a 02 , . . . , a 11 , a 12 , . . . s b 10 , b 20 , . . . , b 11 , b 21 , . . . (1 + χ(ǫ = 0)), where, following A. Garsia [Ga] (see also [Kn]), if p is a proposition then we write χ(p) = 1 if p is true, and χ(p) = 0 if p is false. To complete the construction of w, we need to choose the location of the W i0 's and W 0j 's required by entries a i0 and b 0j in the type of w. A factor W i0 in the canonical product for [0, w] must arise between two successive lower alphabet letters of w, or in front of the first L-factor if w begins with an L-factor, or after the last L-factor if w ends with an L-factor. Therefore such a factor W i0 occurs in one of m − s + 1 − ǫ positions. Similarly, a factor W 0j in the canonical product for [w,1] can arise from any of n − r + 1 + ǫ positions (between two successive letters from the upper alphabet, in front of the first U -factor if w begins with a U -factor, or after the last U -factor if w ends with a U -factor). In conclusion, the word w can be completed in m − s + 1 − ǫ a 10 , a 20 , . . . , m − s + 1 − ǫ − i a i0 n − r + 1 + ǫ b 01 , b 02 , . . . , n − r + 1 + ǫ − j b 0j ways. The remainder of the proof is a calculation. After multiplying (15) by the preceding expression, the relations r − s = ǫ, 1 + ib ij = a ij , and 1 + ja ij = b ij allow some The first case in the conclusion of the Proposition now follows from a simple manipulation with binomial coefficients. The case w = ∅ is trivial, so the proof is complete. Theorem 5.2. Let F 0 = F (x, 0), G 0 = G(0, y), and F (x, y) = F (x, G 0 y) G(x, y) = G(F 0 x, y). Then Proof. For fixed r, s, m, n ≥ 0 write where the sum ranges over all a ij and b ij satisfying By Proposition 5.1, the convolution F * G is given by If M is a monomial then write [M ]Q for the coefficient of M in the power series Q. We first consider (F * G) 0 . We have Note that r g 0r y r = G 0 and i,j j =0 z j f ij x i y j = F (x, yz) − F 0 , and similarly for F 0 and G(xz, y) − G 0 . Hence Exactly analogous reasoning applies to (F * G) −1 and (F * G) 1 . For instance, (F * G) −1 can be written (F * G) −1 = k,m,n k! (k + 1)! (m − k), (n + 1 − k)! [q m+1 r n+1 s m t n u k+1 v k ] · a ij ,b ij j =0 q t j u f ij x i y j a ij i q f i0 x i a i0 i =0 r s i v g ij x i y j b ij j r g 0j y j b 0j ( a ij !) ( b ij !) . Reasoning as above yields Similarly (or by symmetry), From this we easily obtain as desired. Example 5.3. In general it seems difficult to understand the operation F * G. It is not even obvious from (16) that * is associative! A special case for which F * G can be explicitly evaluated is the following. Let a 1 , b 1 , . . . , a k , b k be real numbers (or indeterminates). Then it is straightforward to prove from Theorem 5.2 by induction on k that 1 (1 − a 1 x)(1 − b 1 y) * · · · * 1 (1 − a k x)(1 − b k y) = 1 1 − ( a i )x − ( b i )y + ( (a 1 + a 2 + · · · + a i )b i )xy . For instance, if ζ denotes the zeta function of W ∞∞ (whose value is 1 on every interval of W ∞∞ ), then i,j ζ ij x i y j = 1 (1 − x)(1 − y) . Hence the left-hand side of (17) becomes the generating function for ζ k , whose value at W ij is the number Z ij (k) of k-element multichains0 = t 0 ≤ t 1 ≤ · · · ≤ t k =1 in W ij . Equivalently, Z ij (k) is the value of the zeta polynomial of W ij at k [St1]. We obtain from (17) that i,j Z ij (k)x i y j = 1 1 − kx − ky − k+1 2 xy , a result of Greene [Gre1]. The left-hand side provides a refinement of the flag f -vector in the sense that the coefficient of a i 1 1 b j 1 1 a i 2 2 b j 2 2 · · · x m y n , where m = i 1 + i 2 + · · · and n = j 1 + j 2 + · · ·, is the number of multichains0 = t 0 ≤ t 1 ≤ · · · ≤ t m+n−1 < t m+n =1 in W mn such that t r−1 and t r differ by i r letters from the alphabet A and by j r letters from the alphabet X . The right-hand side of (17) can be interpreted as the generating function for the language L over the alphabet {a 1 , a 2 , . . . , b 1 , b 2 , . . .} consisting of the words in which there is no occurrence of a letter a k immediately followed by a letter b l with k ≤ l. (This follows directly through a simple signreversing-involution argument or from the general theory of Cartier and Foata [CaFo].) The number of such words formed with the multiset of letters a i 1 1 b j 1 1 a i 2 2 b j 2 2 · · ·, where m = i 1 +i 2 +· · · and n = j 1 + j 2 + · · ·, is the coefficient of a i 1 1 b j 1 1 a i 2 2 b j 2 2 · · · x m y n on the right-hand side of (17). For example, the coefficient of a 2 1 b 1 b 2 x 2 y 2 is 2, accounting for the words b 1 b 2 a 1 a 1 and b 2 b 1 a 1 a 1 in L. Now each word w ∈ L consisting of m a i 's and n b i 's determines a shuffle-word s(w) in W mn by placing the alphabet A in the positions of the a i 's and the alphabet X in the positions of the b i 's. For example, w = a 2 b 3 b 3 a 1 a 3 b 5 b 1 b 2 a 3 gives rise to the shuffle word s(w) = a 1 x 1 x 2 a 2 a 3 x 3 x 4 x 5 a 4 in W 45 . Of course, w → s is not a one-to-one map. Given a word w ∈ L, we construct a unique multichain t(w) in W mn by starting with t 0 =0 and obtaining t r+1 from t r by removing the letters appearing in s(w) in those positions where a r occurs in w, and inserting the letters appearing in s(w) in those positions where b r occurs in s(w). For instance, the word w from the preceding example yields t(w) = (0 = t 0 < t 1 < t 2 < t 3 = t 4 < t 5 =1), where t 0 = a 1 a 2 a 3 a 4 , t 1 = a 1 a 3 x 4 a 4 , t 2 = a 3 x 4 x 5 a 4 , t 3 = t 4 = x 1 x 2 x 4 x 5 , t 5 = x 1 x 2 x 3 x 4 x 5 . Recall from Section 3 that there is a unique order in which the i r deletions and j r insertions can be performed such that t r+1 is reached from t r via the Λ-increasing chain γ(t r , t r+1 ). The condition defining the words w ∈ L ensures that each insertion of a letter from X is made in the leftmost position permitted by the shuffle word s(w), thus not multiply-counting chains which involve covering relations of the type (xa). From these observations it follows that w → t(w) is a bijection, completing a combinatorial proof of (17).	2014-10-01T00:00:00.000Z	1998-06-10T00:00:00.000	{ "year": 1998, "sha1": "fabe94953cd99476ad436e11d31b65d193af5107", "oa_license": "elsevier-specific: oa user license", "oa_url": "https://doi.org/10.1016/s0012-365x(98)00381-1", "oa_status": "BRONZE", "pdf_src": "Arxiv", "pdf_hash": "fabe94953cd99476ad436e11d31b65d193af5107", "s2fieldsofstudy": [ "Mathematics" ], "extfieldsofstudy": [ "Mathematics", "Computer Science" ] }
8865357	pes2o/s2orc	v3-fos-license	Aligning an Italian WordNet with a Lexicographic Dictionary: Coping with limited data This work describes the evaluations of two approaches, Lexical Matching and Sense Similarity, for word sense alignment between MultiWordNet and a lexicographic dictionary, Senso Comune De Mauro, when having few sense descriptions (MultiWordNet) and no structure over senses (Senso Comune De Mauro). The results obtained from the merging of the two approaches are satisfying, with F1 values of 0.47 for verbs and 0.64 for nouns. Introduction This work is situated in the field of word sense alignment, a research area which has seen an increasing interest in recent years and which is a key requirement for achieving semantic interoperability between different lexical-semantic resources (Matuschek and Gurevych, 2013).Our goal is to automatically import high-quality glosses in Italian in MultiWordNet (Pianta et al., 2002) (MWN) by aligning its synsets to the entries of a lexicographic dictionary, namely the Senso Comune De Mauro (SCDM), thus providing Italian with a more complete and robust version of MWN.For SCDM, the linking of the entries with MWN plays a double role.On the one hand, it will introduce lexical-semantic relations, thus facilitating its use for NLP tasks in Italian, and, on the other hand, it will make SCDM a structurally and semantically interoperable resource for Italian, to which other lexical-semantic resources (both in Italian, such as PAROLE-SIMPLE-CLIPS (Ruimy et al., 2003), and in English, such as VerbNet (Kipper Schuler, 2005), among others), sense annotated corpora (e.g. the MultiSemCor corpus (Bentivogli and Pianta, 2005)), and Web-based encyclopedia (e.g.Wikipedia) can be connected. At this stage of development we focused on the alignment of verbs and nouns.The remaining of this paper is organized as follows.Section 2 will state the task and describe the characteristics of the two lexica.In Section 3 some related works and the perculiarities of our work are discussed.The approaches we have adopted are described in Section 4. The evaluation is carried out in Section 5, including an error analysis.Finally, in Section 6 conclusions and future works are reported. Problem Description and Resources Following (Matuschek and Gurevych, 2013), word sense alignment (WSA) can be formally defined as a list of pairs of senses from two lexical-semantic resources.A pair of aligned senses denotes the same meaning. For instance, taken the two senses of the word "day" "amount of hours of work done in one day and "the recurring hours established by contract or usage for work" (taken from translated SCDM and MWN, respectively), they must be aligned as they are clearly equivalent. MultiWordNet MWN is a computational multilingual lexicon perfectly aligned to Princeton WN 1.6.As in WN, concepts are organized in synonym sets (synsets) which are hierarchically connected by means of hypernym relations (is a).Additional semantic relations such as meronymy, troponymy, nearest synonym and others are encoded as well.The Italian section of MWN is composed of 38,653 synsets, with 4,985 synsets for verbs and 28,517 synsets for nouns.Each synset is accompanied by a gloss describing its meaning and, when present, one or more examples of use.Only 3,177 glosses (8,21%) are in Italian and, in particular, 402 for verbs and 2,481 for nouns. Senso Comune aims at building an open knowledge base for the Italian language, designed as a crowd-sourced initiative that stands on the solid ground of an ontological formalization and wellestablished lexical resources.The lexicon entries have been obtained from the De Mauro GRADIT dictionary and consists in the 2,071 most frequent Italian words, for a total of 11,939 fundamental senses.As for verbs we have 3,827 senses, corresponding to 643 lemmas, with an average polysemy of 5.9 senses per lemma.As for nouns we have 4,586 senses, corresponding to 1,111 lemmas with an average polysemy of 4.12 senses per lemma.In SCDM, word senses are encoded following lexicographic principles and are associated with lexicographic examples of usage. Senso Comune comprises three modules: i.) a top level module for basic ontological concepts; ii.) a lexical module for linguistic and lexicographic structures; and iii.) a frame module for modeling the predicative structure of verbs and nouns.The top level ontology is inspired by DOLCE (Descriptive Ontology for Linguistic and Cognitive Engineering) (Masolo et al., 2002).All nominal entries have been manually classified according to the ontological concepts and an ontological classification of verb entries will start in the near future.With respect to MWN, word senses are not hierarchically structured and no semantic relation is encoded.Senses of polysemous entries have a flat representation, one following the other. Related Works Previous works in word sense alignment can be divided into two main groups: a.) approaches and frameworks which aim at linking lexica based on different models to WN synsets (Rigau and Eneko (1995); Navigli (2006); Roventini et al. (2007)) or language resources, such as Wikipedia (Ruiz-Casado et al. (2005); Mihalcea (2007); Niemann and Gurevych (2011)), and b.) approaches towards the merging of different language resources (Gurevych et al. (2012); Navigli and Ponzetto (2012)).Our work clearly fits into the first group.While different methods are employed (similaritybased approaches vs. graph-based approaches), common elements of these works are: i.) the extensive use of lexical knowledge based on the sense descriptions such as the WN glosses or an article first paragraph as in the case of Wikipedia; and ii.) the extension of the basic sense descriptions with additional information such as hypernyms for WN entries, domains labels or categories for dictionaries or Wikipedia entries so as to expand the set of available information, thus improving the quality of the alignments. As for our task, the most similar work is (Navigli, 2006) where entries from a lexicographic dictionary, namely the Oxford English Dictionary (OED), are mapped to WN.The author adopts and compares two methods: a.) a pure lexical matching function based on the notion of lexical overlap (Lesk, 1986) of the lemmas in the sense descriptions; and b.) a semantic matching based on a knowledge-based WSD system, Structural Semantic Interconnections (SSI), built upon WN and enriched with collocation information representing semantic relatedness between sense pairs.In this latter approach, first each sense description in WN and in the OED is disambiguated by means of SSI with respect to the WN sense inventory, thus obtaining a semantic description as a bag of concepts.Then, two senses are matched if a relation edge is identified between the concepts in the description of each sense in the two lexica.Both approaches are evaluated with respect to a manually created gold standard.The author reports an overall F1 measure of 73.84% for lexical matching, and of 83.11% for semantic matching. With respect to the SCDM, the OED has some advantages, namely i.) the distinction between core senses and subsenses for polysemous entries; ii.) the presence of hypernyms explicitly signalled; and iii.) domain labels associated with word senses.Such kind of information is not present in the SCDM where senses are presented as a flat list and no enrichment of the sense descriptions with additional information is available, except for the ontological tagging of nouns.Moreover, the low number of MWN glosses in Italian prevents a straightforward application of state-ofthe-art methods for sense alignment.MWN sense descriptions must be built up from other sources.Thus, the main issue we are facing is related to data sparseness, that is how to tackle sense alignment when we have few descriptions in Italian (MWN side) and few meta-data and no structure over senses (SCDM side). The automatic alignment of senses has been conducted by applying two approaches for constructing the sense representations of the resources and evaluation. Lexical Match In the first approach, Lexical Match, for each word w and for each sense s in the given resources R ∈ {MWN, SCDM} we constructed a sense descriptions d R (s) as a bag of words in Italian.Provided the different characteristics of the two resources, two different types of bag of words have been built.As for the SCDM, the bag of words is represented by the lexical items in the textual definition of s w , automatically lemmatized and partof-speech analyzed with the TextPro tool suite (Pianta et al., 2008) with standard stopword removal.On the other hand, for each synset, S, and for each part of speech in analysis, the sense description of each MWN synset was built by optionally exploiting: • the set of synset words in a synset excluding w; • the set of direct hypernyms of s in the taxonomy hierarchy in MWN; • the set of synset words in MWN standing in the relation of nearest synonyms with s; • the set of synset words in MWN composing the manually disambiguated glosses of s from the "Princeton Annotated Gloss Corpus"2 .To extract the corresponding Italian synset(s), we have ported MWN to WN 30; • the set of synset words in MWN composing the gloss of s in Italian (when available); • for verbs, the set of synset words in MWN standing in the relations of entailment/is entailed, causes/is caused with s; • for nouns, the set of synset words in MWN standing in the relations of part of /has part, has member/is member with s. The alignment of senses is based on the notion of lexical overlap.We used Text::Similarity v.0.09module3 , and in particular the method Text::Similarity::Overlaps, to obtain the overlap value between two bags of words of s w .Text similarity is based on counting the number of overlapping tokens between the two strings, normalized by the length of the strings. One of the well known limitation of the Lexical Match approach is the so called "lexical gap" problem (Meyer and Gurevych, 2011), i.e. a reduced number of overlapping words.To overcome this limit, we have exploited a newly developed multilingual resource, BabelNet (Navigli and Ponzetto, 2012), which has been obtained by merging together WN synsets and Wikipedia pages with an accuracy of 83%.It contains 4,683,031 nominal glosses (2,985,243 of which are in English).In BabelNet English WN 3.0 synsets have been aligned to their corresponding Wikipedia pages and then extended to other languages, including Italian, by exploiting Wikipedia language links and WN mappings.As for our task, we have retained only those BabelNet entries which have a corresponding synset word in MWN.In this way, we have extended the bag of words representation of nominal entries for MWN synsets by adding the Italian Wikipedia glosses from BabelNet. Sense Similarity In the second approach, Sense Similarity, the basis for sense alignment is the Personalized Page Rank (PPR) algorithm (Eneko and Soroa, 2009) relying on a lexical-semantic knowledge base model as a graph G = (V, E) as available in the UKB tool suite4 .As knowledge base we have used WN 3.0 extended with the "Princeton Annotated Gloss Corpus".Each vertex v of the graph is a synset, and the edges represent semantic relations between synsets (e.g.hyperonymy, hyponymy, etc.).The PPR algorithm ranks the vertices in a graph according to their importance within the set and assigns stronger initial probabilities to certain kinds of vertices in the graph.The result of the PPR algorithm is a vector whose elements denotes the probability for the corresponding vertex that a jumper ends on that vertex if randomly following the edges of the graph. To obtain the PPR vector for a sense s of the SCDM, we have translated the Italian textual definitions in English by means of a state-of-theart Machine Translation system5 , automatically lemmatized and part-of-speech analyzed with the TextPro tool suite, remove standard stopwords and applied the UKB tool suite.The PPR vector is a thus semantic representation overall the entire WN synsets of the textual definition of s in SCDM. As for the MWN synsets, we have exploited its conversion to WN 3.0.Instead of building the PPR vector by means of the lexical items, we have passed to the UKB tool suite the WN synset id, thus assuming that the MWN synset is already disambiguated.Given two PPR vectors, namely ppr mwn and ppr scdm for the MWN synset w syn and for the SCDM sense w scdm , we calculated their cosine similarity.On the basis of the similarity score, the sense pair is considered as aligned or not. Gold Standards To evaluate the reliability of the two approaches with respect to our data, we developed two different gold standards, one for verbs and one for nouns. The verb gold standard is composed by 44 lemmas selected according to corpus frequency (highly frequent lemmas in the La Repubblica Corpus (Baroni et al., 2004)) and patterns in terms of semantic and syntactic features6 .It is composed by 350 aligned sense pairs obtained by manually mapping the MWN synsets to their corresponding senses in the SCDM lexicon.These verbs corresponds to 279 synsets and 424 senses in the SCDM.Overall, 211 of the 279 MWN synsets have a corresponding sense in the SCDM (i.e.SCDM covers 84.22% of the MWN senses in the data set), while 235 out of 424 SCDM senses have a correspondence in MWN (i.e MWN covers 49.76% of the SCDM senses).Average degree of polysemy for MWN entries is 6.34, while for the SCDM is 9.63. The noun gold standard is composed by 46 lemmas selected according to frequency and polysemy with respect to the fundamental senses in the SCDM (each lemma must have at least two fundamental senses in the SCDM).On the basis of the manual alignment, we have obtained 166 aligned sense pairs.The noun lemmas correspond to 229 synsets and 216 senses in the SCDM.Overall, 134 of the 229 MWN synsets have a corresponding sense in the SCDM (i.e.SCDM covers 53.71% of the MWN senses in the data set), while 123 out of 216 SCDM senses have a correspondence in MWN (i.e MWN covers 62.03% of the SCDM senses).Average degree of polysemy for MWN entries is 4.97, while for the SCDM is 4.69.The difference in terms of coverage with respect to the verbs is clearly due to two aspects, namely i.) the restrictions of the SCDM entries to the fundamental senses; ii.) the higher coverage in terms of nouns synsets of MWN with respect to the verbal ones. Though small, the size of the gold standards is representative of the two lexica.In particular, the 279 verbs synsets yield 3,319 possible sense pairs, i.e. 11.8 SCDM senses per synset on average.As for nouns, the 229 nominal synsets yield 1,414 sense pairs, i.e. 6.13 SCDM senses on average. Results The evaluation has been performed by computing Precision (the ratio of the correct alignment with respect to all proposed alignments), Recall (the ratio of extracted correct alignment with respect to the alignments in the gold standard), F-measure (the harmonic mean of Precision and Recall calculated as 2P R/P + R) and Accuracy (the precentage of the correctly identifed alignments and non alignments).As baseline, we have implemented a random match algorithm, rand, which for the same word w in SCDM and in MWN assigns a random SCDM sense to each synset with w as synset word, returning a one-to-one alignment.The selection of the correct alignments has been obtained by applying two types of thresholds with respect to all proposed alignments (the "no threshold" row in the tables): i.) a simple cut-off at specified values (0.1; 0.2); ii.) the selection of the maximum score (either lesk measure or cosine; row "max score" in the tables) between each synset S and the proposed aligned senses of the SCDM.As for the maximum score threshold, we have retained as good alignments also instances of a tie, thus allowing the possibility of having one MWN synset aligned to more than one SCDM sense. Table 1: Results for automatic alignment based on Lexical Match for SYN and SREL sense representations. Lexical Match Results We have analyzed different combinations of the sense representation of a synset.We developed two basic representations: SYN, which is composed by the set of synset words excluding the target word w to be aligned, all of its direct hypernyms, the set of synset words in MWN standing in the relation of nearest synonyms and the synset words obtained from the "Princeton Annotated Gloss Corpus"; and SREL, which contains all the items of SYN plus the the synset words included in the selected set of semantic relations.The results are reported in Table 1.As the figures show, all synset configurations outperform the baseline rand for both parts of speech in analysis.However, it is interesting to observe that the alignment of noun senses performs much better than that for verbs in both sense representations and with all filtering methods.On the basis of the alignment method (i.e.lexical overlap) such a difference in performance provides interesting data on the two resources in analysis.A manual exploration of the data in the configurations both for verbs and nouns has highlighted that, on the one hand, we suffer from data sparseness on the SCDM side as no extension of the sense description of the glosses is possible, and, on the other hand, that senses are described in ways that are semantically equivalent but with different lexical items. As for verbs the Recall with no filtering (no threshold) has extremely low levels, ranging from 0.32 for SREL to 0.29 for SYN.The SREL sense representation outperforms SYN when no filtering is applied only in terms of Recall (+0.03), thus signaling that the additional semantic relations play a very limited role in the description of verb senses without providing real additional information to match data in the SCDM glosses.Furthermore, the difference in performance of the SREL configuration is not statistically significant with respect to the SYN configuration (p > 0.05). The situation looks different for nouns where, although low, the no threshold Recall values range between 0.60 (SREL) to 0.59 (for SYN).As for the two basic configurations, SYN and SREL, the results show that SYN is more accurate and that the impact of additional semantic relations, though it slightly improves the Recall, is not statistically signiticant (p > 0.05). Both for verbs and nouns we decided to select the SYN basic configuration as the best sense representation because it has a simpler bag-of-words and better Precision.To improve the results, we have extended this basic representation with the lexical items in the corresponding glosses of Ba-belNet (+BABEL) (only for nouns) and the lexical items of the MWN Italian glosses (+IT) (for verbs and nouns)7 .The results are illustrated in Table 2. In both cases, the extension of the basic sense representations with additional data is positive, namely for Recall.Notice that for verbs the presence of Italian MWN glosses improves the alignment results (for the no-threshold filter, F1=0.37 vs. F=0.35for SREL and F1=0.34 for SYN) as they introduce information which better represents the sense definition than the synset words in the bag of words representations and overcomes missing information in the WN 3.0 annotated glosses.For instance, consider the following example for the verb "rendere" [to make].In example 1a) the two senses are aligned with a very low lexical overlap score as there is only one word in com- The positive effect of the original Italian data for verbs points out a further issue for our task, namely that the derivation of sense representations of MWN synsets by means of synset words (including the sense annotated glosses of WN 3.0) is not as powerful as having at disposal original glosses. Similarly, for nouns we register an improvement in Recall at a low or null cost for Precision for all filtering methods, with the exclusion of the no threshold filtering.Precision for SYN+BABEL+IT with maximum score filtering is lowered with respect to the extension with the BabelNet data only (P=0.66 for SYN+BABEL+IT vs. P=0.69 for SYN+BABEL)8 .To better clarify these results, consider the following example for the noun "palla" [ball].In the example 2a) the two senses are not aligned as there are no matching words, while in 2b) the extension by means of the BabelNet data provides a sufficient number of matching items for aligning the two senses.As for the previous example, the lexical items of the sense descriptions are reported in Italian, matching words are in bold.Concerning the filtering of the proposed alignments, the maximum score filter provides the best results for Precision at a low cost in terms of Recall, with F1 scores for verbs ranging from 0.34 (SYN+IT) to 0.29 (SYN), and from 0.55 (SYN+BABEL) to 0.52 (SYN and SREL) for nouns.It is interesting to point out a further difference in performance between verbs and nouns.In particular, for verbs we can observe that the filtering based on maximum score has lower F1 values with respect to the no threshold baseline in all sense descriptions.As for nouns, on the contrary, both the two basic sense descriptions, SYN and SREL, and the SYN+BABEL configuration have comparable F1 values between the no threshold and the maximum score data.Nevertheless, the filtering based on the maximum score improves the quality of the proposed alignment by removing lots of false positives both for verbs and nouns (for verbs P=0.59 for SYN, P=0.60 for SREL, and P=0.63 for SYN+IT; for nouns, P=0.69 for SYN, SREL, and SYN+BABEL, P=0.66 for SYN+BABEL+IT) without impacting on the number of good instances retrieved (for verbs R=0.19 for SYN, R=0.20 for SREL, and R=0.23 for SYN+IT; for nouns R=0.42 for SYN and SREL, R=0.44 for SYN+BABEL; R=0.45 for SYN+BABEL+IT). Similarity Measure Results The results for the Similarity Measure obtained from the Personalized Page Rank algorithm on the basis of the vectors described in Section 4.2 are illustrated in Table 3. Similarly to the Lexical Match, the Personalized Page Rank approach outperforms the baseline rand.Overall, the differences in performance with the Lexical Match results are not immediate.In general, as the Recall values for no threshold filtering show, almost all aligned sense pairs of the gold are retrieved, outperforming the Lexical Match.Clearly, this difference is strictly related to the different nature of the sense descriptions, i.e. a semantic representation based on a lexical knowledge graph, which is able to catch semantically related items out of the scope for the Lexical Match approach. By observing the figures for verbs, we notice that the simple cut-off thresholds provide better results with respect to the maximum score.The best F1 score (F1=0.32) is obtained when setting the cosine similarity to 0.1, though Precision is less than 0.50 (namely, 0.47).When compared with threshold value of 0.1 of the Lexical Match, the Personalized Page Rank method yields the best Precision (P=0.47 vs. P=0.42 for Verb SYN, P=0.38 for Verb SYN+IT, and P=0.40 for Verb SREL).Similar observations can be done when the threshold is set to 0.2.In this latter case, Personalized Page Rank yields the best Precision score for verbs with respect to all other filtering methods and the Lexical Match results obtained with maximum score (P=0.66 vs. P=0.59for Verb SYN, P=0.63 for Verb SYN+IT, and P=0.60 for Verb SREL). The analysis for nouns is more complex.Apparently, the Personalized Page Rank approach has lower F1 scores with respect to all Lexical Match sense configurations and filtering methods, including the no threshold score of the basic sense descriptions (respectively, F1=0.55 for SYN, F1=0.54 for SREL, F1=0.21 for Personalized Page Rank).However, when maximizing Precision for the Personalized Page Rank (threshold 0.2), the algorithm provides better performances (F1=0.33)with respect to Lexical Match on the same filtering method, minimizing the drop of Recall (R=0.21;+0.09 with respect to SYN+BABEL with same threshold; + 0.08 with respect to SREL; +0.05 with respect to SYN, respectively). The better performance of the simple cut-off thresholds with respect to the maximum score is due to the fact that aligning senses by means of semantic similarity provides a larger set of alignments and facilitates the identification of multiple alignments, i.e. one-to-many. Merging Lexical Match and Sense Similarity As The combination of the best results yields the best performance for both parts of speech compared to the stand-alone approaches.In particular, for verbs we obtain an F1=0.47, with an improvement of 0.18 points with respect to SYN and of 21 points with respect to Personalized Page Rank with threshold 0.2.Similar improvements can be observed for nouns, where SYN+BABEL+ppr02 has an F1=0.64, with an improvement of 9 points with respect to SYN+BABEL and of 31 points with respect to Personalized Page Rank with threshold 0.2.In both cases the performance gains originate from the higher precision of the Personalized Page Rank approach which minimizes the data sparseness of the SCDM lexicon. Conclusion and Future Work This paper focuses on the automatic alignment of senses from two different resources when few data are available.In particular, the lack of Italian glosses in MWN and the absence of any kind of structured information in the SCDM dictionary posed a serious issue for the application of stateof-the-art techniques for sense alignment. We experimented with two different approaches: Lexical Match and Sense Similarity obtained from Personalized Page Rank.In all cases, when filtering the data we are facing low scores for Recall which point out issues namely related to data sparseness in our lexica.By comparing the results of the two approaches, we can observe that: i.) the Personalized Page Rank yields the best Precision with respect to Lexical Match; ii.) Lexical Match, with a simple sense description configuration (i.e. the SYN configurations for verbs and nouns), is still a powerful approach for this kind of tasks; the exploitation of additional semantically related items (e.g.SREL for verbs) or additional sense descriptors (e.g.SYN+BABEL for nouns), though good in principle, has a limited contribution to solve the "lexical gap" problem in our case and points out differences in the way word senses are encoded in the two lexica; and iii.)Personal-ized Page Rank vectors and Lexical Match appears to qualify as complementary methods for achieving reliable sense alignments, namely when dealing with few data.Our approach provides satisfying results both for verb and noun sense alignment, with an overall F1=0.47 for verbs and an F1=0.64 for nouns.The better results for nouns are strictly related to the definitions of the senses which mainly relies on synonym words and hypernyms.On the other hand, verbs tend to have more abstract definitions and the contribution of additional semantic relations (i.e. the SREL configuration) is poor. Future work will concentrate on two aspects by exploiting the sense alignment results.The aligned sense pairs will be used for sense clustering as a strategy to reduce the sense descriptions in MWN and in SCDM.Existing clustering of WN senses (e.g.Navigli (2006)) will be used as a starting point and for subsequent evaluation.Furthermore, we aim at importing the ontological classes of SCDM in MWN.This aspect will be useful for the identification of possible taxonomical errors in the MWN hierarchy and boostrap better sense alignments. Table 3 : Results for automatic alignment based on Similarity Score. Table 4 : Results for automatic alignment merging the best results from Lexical Match and Sense Similarity.	2015-08-11T20:29:18.000Z	2014-01-25T00:00:00.000	{ "year": 2014, "sha1": "90229e89feb2b5f7a5ba5e3b85d696eed92f7c63", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "ACL", "pdf_hash": "90229e89feb2b5f7a5ba5e3b85d696eed92f7c63", "s2fieldsofstudy": [ "Computer Science", "Linguistics" ], "extfieldsofstudy": [ "Computer Science" ] }
139339386	pes2o/s2orc	v3-fos-license	Carbonation Resistance of Alkali-Activated Slag Under Natural and Accelerated Conditions In this paper, carbonation resistance of alkali-activated slag (AAS) pastes exposed to natural and accelerated conditions up to 1 year was investigated. Two aspects of carbonation mechanism were evaluated. The first was the potential carbonation of the main binding phases in finely powdered AAS pastes. The second was the reactivity and diffusivity of CO2 within the bulk AAS paste. From Fourier transform infrared spectroscopy and thermogravimetric analysis coupled with mass spectroscopy time-series measurements, it was found that powdered AAS was largely carbonated within 28 days with a CO2 uptake of 14 wt%. The main carbonation products were calcium carbonates. Nevertheless, the bulk paste samples were highly resistant to carbonation, regardless of the exposure conditions. The findings showed that the pH value (initial pH > 12) and strength of the samples did not decrease under accelerated carbonation compared to those of the samples exposed under natural conditions. The mineralogy of the samples in these two carbonation exposures did not alter either, except for outdoor conditions. The gel pores were dominant in the pastes (pore size in range of 2–15 nm). The dense microstructure was the main barrier for CO2 to diffuse and further react with binding phases. Introduction Ground granulated blast furnace slag (GGBFS) is a byproduct from steel production that has been widely used as a supplementary cementitious material (SCM) in blended cements [1]. In the Netherlands, the building industry has almost a century of experience in the use of GGBFS cements with a high GGBFS content, comparable to the current CEM III/B (66-80% GGBFS) as defined in EN 197-1 (2011) for major infrastructure, including marine concrete [2]. The binders with higher replacement such as above 70 wt% of GGBFS, have a significantly lower amount of calcium hydroxide (CH) than that in the ordinary Portland cement (OPC)-based binders [3]. Consequently, the faster carbonation of calcium silicate hydrate (C-S-H) gel occurs in GGBFS-based binders [4,5]. Therefore, there is a certain level of OPC replacement by GGBFS (not more than 70 wt%), for which properties such as high mechanical strength and carbonation resistance still can be obtained [6]. On the other hand, the GGBFS valorization in the building industry can be achieved due to alkaline activation technology. The need for compelling utilization of byproducts and recovered waste materials, based on economic and environmental concerns, has led to the development of alkali-activated materials (AAMs). Alkaliactivated slag (AAS) has been recognized for decades [7] as an environmentally friendly alternative to OPC based on two aspects. The first is the use of byproducts instead of natural resources for synthesis of AAS [8] and the second is its comparable or even better short and long-term properties compared to the OPC-based binders [9]. However, durability of AAS-based systems is still under debate, which limits their application in engineering practice. In particular, carbonation is of great interest since it induces both chemical and physical changes. It leads to corrosion of the steel reinforcement such as in OPC-based concrete. Namely, the corrosion of the steel reinforcement is inhibited due to the protective oxide film on the surface of the steel which is chemically stable in an alkaline concrete pore solution (pH [ 12.5) [10]. However, the oxide film can be destroyed if the alkalinity of the pore solution drops. The CH buffers pH of the pore solution in OPCbased concrete, and it is consumed during the carbonation reaction with environmental CO 2 . As a result, the pH of the concrete pore solution drops from approximately 12.5 to \ 9 [10]. If the carbonation front reaches the reinforcement and there is sufficient moisture around the steel reinforcement surface, the corrosion is likely to be initiated, limiting the service-life of concrete. Researchers have been using different binder systems (AAS, and alkali-activated blended systems of GGBFS with fly ash or metakaoline) and exposures (CO 2 , RH) to study the carbonation mechanism of AAS in laboratory conditions [11][12][13][14][15][16]. On the other hand, the long-term properties of AAS binders in service conditions are rarely reported. For instance, Shi et al. [17] have presented a significant number of real applications of AAS concrete, however, the information about their long-term performance was limited. Xu et al. [18] by mechanical tests' results and microstructural characterization of the AAS concretes (activated by carbonates or carbonate/hydroxide solutions and cast between 1964 and 1982), demonstrated that the AAS concretes have served for prolonged periods, and their strength increased with time. Another study [16] showed that accelerated carbonation conditions did not replicate the carbonation rate in AAS concretes under natural conditions. Namely, lower carbonation depth was observed in AAS concrete in natural carbonation conditions after 7 years of exposure (with RH varying between 70 and 76% and temperature between 19 and 38°C) compared to the predicted carbonation depth in accelerated carbonation conditions (1% v/v CO 2 ). Contrary to excellent performance of AAS concrete in service, researchers have observed low carbonation resistance of AAS pastes, mortars and concretes in laboratory conditions [11,19,20]. To study the carbonation mechanism in AAS, generally pastes were ground to a fine powder and as such exposed to different carbonation conditions. In such a way more reaction sites would be provided for carbonation reaction. In all studies, it was found that AAS was carbonated regardless of the type and concentration of activator and exposure conditions. Furthermore, the carbonation resistance of AAS was found to be lower than that of OPC [19]. The general carbonation mechanism in AAS appears to be relatively well established by Bernal et al. [14,21]. First, the carbonation of pore solution leads to the formation of sodium carbonates, followed by decalcification of C-(N)-A-S-H gel and consequent decay of mechanical strength of the AAS systems. A detailed study by Bernal et al. [21] on the effect of binder content on the performance of AAS concretes revealed that carbonation of the concrete mixtures with GGBFS content 400 kg/m 3 or higher is initially a chemically controlled process [22] followed by diffusion. When the carbonation depth becomes greater (more than 20% of the total examined depth), it was observed that diffusion of dissolved CO 2 through the pore network of the carbonated layer becomes dominant in determining the rate of further carbonation. It was also found that the use of higher binder content (400-500 kg/m 3 ) resulted in a reduction of carbonation depth of AAS concretes. Furthermore, Bernal et al. [13] found that beside binder content, the binder chemistry significantly influences carbonation rate, in particular MgO content of raw GGBFS. Hydrotalcite-like phases (Mg-Al layered double hydroxides) have been seen to increase the resistance to carbonation of alkali-silicate activated GGBFS by absorbing CO 2 in their structure [13]. However, when GGBFS is blended with other precursors such as fly ash or metakaolin [11,12,15], these systems exhibited lower resistance to accelerated carbonation compared to pure AAS systems due to their different pore solution chemistry [14] and composition of the precursors. Although the studies on powdered pastes are of immense value in understanding the carbonation mechanism in the AAS, the effect of carbonation on pore structure and mechanical properties cannot be then captured at the powdered paste level. Pore structure is the most fundamental paste property. Grinding eliminates much of the capillary porosity and this removes the mechanism that collapses the large gel pores [23]. Any diffusion mechanism occurring in the porous media is directly or indirectly controlled by its pore structure. The influence of the pore structure of AAS bulk paste on the carbonation rate was barely discussed in the literature regarding CO 2 diffusion path. Therefore, the aim of this paper is twofold: (i) to study carbonation of powdered AAS pastes, with a focus on reactivity of CO 2 and maximum CO 2 uptake, and (ii) to investigate carbonation of bulk AAS pastes in terms of reactivity but also diffusivity of CO 2 , while capturing the effect of chemical degradation on the pore structure and mechanical properties, which is not possible within the first part. In this way, the paper may provide fundamental insight on the role of pore structure, and exposure conditions, such as relative humidity (RH) and CO 2 concentration on the carbonation mechanism at the paste level. Materials and Sample Preparation The GGBFS used in this study was supplied by ORCEM (the Netherlands). X-ray fluorescence (XRF) measurements were carried out using a PANalytical's Epsilon 3 XLE spectrometer equipped with a Rhodium X-ray source, the silicon-drift detector with 135 eV resolution at 5.9 keV/ 1000 cps. A 2-3 g of GGBFS powder was poured in a 32-mm spectro cup fitted with a bottom of stretched 4-lm prolene film held with a concentric ring. The chemical composition of the GGBFS determined by XRF is given in Table 1. The negative LOI value was related to the oxidation of sulfur-rich species in the GGBFS. It should be noted that the GGBFS was heated to 950°C in air to determine the LOI, which was not corrected in the XRF measurements. Figure 1 shows the particle size distribution of GGBFS, which was measured by laser diffraction analyzer. The instrument can provide particle size distribution ranging from 0.1 to 1000 lm. The chosen dispersant was ethanol. An external ultrasonic bath was used for de-agglomeration of the GGBFS particles. The average particle size of GGBFS, d 50 , was 19 lm. The X-ray diffractogram (XRD) of GGBFS was performed using a Philips PW 1830 powder X-ray diffractometer, with Co Ka (1.789 Å ) radiation, tube setting 40 kV and 40 mA, a step size of 0.030°, a rate of 2.0 s per step and a 2 Theta range of 10°-70°. The amorphous phase was dominant (Fig. 2). The alkaline activator was prepared by mixing anhydrous pellets of sodium hydroxide with deionized water and commercial sodium silicate solution (27.5 wt% SiO 2 , 8.25 wt% Na 2 O). The sodium silicate modulus i.e. SiO 2 / Na 2 O mass ratio of the activator prepared and used for alkali activation was 1.5. The liquid-to-GGBFS ratio was 0.5. The liquid includes the solid part of the activator as well. After mixing, the activator was cooled down at room temperature for 24 h prior to the preparation of the pastes. Before the preparation of paste samples for the carbonation study, the appearance of cracking and compressive strength was monitored at the certain curing age on prismatic AAS paste samples (160 9 40 9 40 mm 3 ) according to the NEN-EN 196-1. The test was carried out because of the sensitivity of AAS pastes to drying and autogenous shrinkage reported by a previous study [24]. Two groups of samples were investigated, unsealed and sealed prismatic samples. The specific aim of the test was to obtain sufficient curing time and avoid shrinkage cracks. The cracks can ultimately underestimate compressive strength of the samples at the later ages, when coupled effect of drying shrinkage and carbonation exposure on results of compressive strength would be difficult to interpret. Therefore, the samples were cured for 7, 14, 18, 21, 28, and 40 days. After each curing period, the samples were moved from the fog room to the laboratory conditions at 20°C and 55% RH. The cracks were observed after each curing age in the laboratory conditions, except for the 28-day-cured samples for both, unsealed (Fig. 3) and sealed samples (not shown in this study). At 28 days, the samples did not have any crack, nor did the samples after 40 days of curing, which was the extended curing time to ensure that even after 28 days, no cracking would appear. At 40 days, compressive strengths of all samples (with different curing durations within 40 days) were tested. For each curing regime, three samples were tested. The results are presented in Fig. 4. It can be seen that compressive strength did not increase significantly beyond 14 days of curing in the fog room. However, the drying shrinkage cracks were observed on the samples with curing time less than 28 days. Although cracking did not affect compressive strength results, these cracks could enable faster gas and water penetration into the samples, thereby reducing the durability of the material. Therefore, all samples for the carbonation study were cured for 28 days. The compressive strength during curing and preconditioning of AAS paste samples, as after different carbonation time exposures, was also examined in the last section of this study for unsealed cured samples. The curing regime for the paste samples for carbonation study, i.e. 28 days in a fog room and then 28 days of Journal of Sustainable Metallurgy preconditioning in the laboratory conditions, was chosen after a study on prismatic samples. The purpose of samples preconditioning in the laboratory conditions for an additional 28 days was to obtain nonsaturated conditions, so that the pores of the samples are partially filled with the moisture and that CO 2 molecules can diffuse. The pastes for the carbonation study were cast in cylinders with 54 mm diameter and height of 100 mm. The samples were sealed and cured for 28 days. The sealing of the samples was applied to prevent premature carbonation. Figure 5 shows typical cylindric paste sample after 28 days of sealed curing. The cylindric samples were further placed in the laboratory conditions at 20°C and 55% RH for additional 28 days (preconditioning of the samples). At the age of 56 days, the samples were exposed to different environmental conditions as described in Table 2. It should be noted that one group of the samples was sealed cured for one year (reference, noncarbonated samples) to provide a baseline for comparison with carbonated samples at the end of the carbonation exposure. Experimental Program Samples were divided into two groups (Table 3, Fig. 5). The first group of pastes was crushed and ground to a powder with an average particle size \ 75 lm and exposed to accelerated conditions, 1% v/v CO 2 , 60% RH. The average particle size of 75 lm was adopted in this study according to the literature [11,12,25], where this size was used or it was shown that it is the one providing the full carbonation in a practical timescale. The second group of the samples was of initial cylindric form (bulk samples) and as such was exposed to different conditions as described in Table 2. Carbonation capacity, carbonation products, and CO 2 uptake in powdered pastes, were studied with Fourier transform infrared spectroscopy (FT-IR), X-ray diffraction (XRD), thermogravimetric analysis coupled with mass spectroscopy (TG-MS), respectively. The powdered paste samples were exposed in a climate chamber with 1% v/v CO 2 concentration, 60% RH, and 20°C. It should be noted that the powdered samples with regard to carbonation capacity testing were analyzed without any defined predrying, while for all other tests samples were prepared as described later. The bulk AAS pastes were exposed in natural [laboratory and outdoor (unsheltered)] and accelerated carbonation conditions (Table 2). In addition to FT-IR, XRD, TG-MS study, pore solution analysis enabled pH measurements and quantitation of the main present ions (7) 40 (14) 40 (18) 40 (21) 40 (28) 40 (40) (Na ? , K ? , Ca 2? ) in bulk paste samples. Nitrogen-adsorption (N 2 -adsorption), ESEM, and compressive strength tests were used to study effects of different carbonation conditions on pore structure and mechanical properties of bulk paste samples. The silicate and carbonate chemical bands were identified in powdered pastes at different time intervals by attenuated total reflectance Fourier transformed infrared spectroscopy (FT-IR). Spectra were acquired using Perk-inElmer Spectrum 100, over the wavelength range of 2000-600 cm -1 with a resolution of 4 cm -1 . A total of 16 scans were collected per measurement. For XRD phase identification, the representative samples were gently crushed and then immersed in isopropanol for 1 week, by which period water is first replaced and then evaporated. Subsequently, samples were placed under vacuum at 25°C for a minimum of 3 weeks. XRD measurements were carried out using Philips PW 1830 powder X-ray diffractometer equipped with X'pert High Score Plus software. Thermogravimetric (TG) analysis was carried out using a heating rate of 10°C/min between 40 and 1000°C, with an argon purge at 70 ml/min. The sample predrying preparation was done in the same way as for XRD analysis. The thermobalance Netzsch STA 449 F3 Jupiter was coupled with a mass spectrometer (MS) Netzsch QMS 403 C, to identify more accurately the temperature range of H 2 O and CO 2 emissions and to quantitate CO 2 uptake. The well-dried samples were ground into powder. The mass of powdered samples for testing was approximately 35 mg. The microstructure of the samples from 56 days and 1 year of exposure was studied by ESEM. Similar to XRD analysis, the reaction of the pastes was stopped using isopropanol. The samples were crushed into small pieces with dimensions of 1-2 cm 3 and impregnated using a low-viscosity epoxy resin and polished down to lm. The internal RH of bulk AAS pastes was measured by Rotronic HygroLab C1 equipped with two HC2-AW RH station probes with an accuracy ± 1% RH. The samples were crashed in small pieces, and only core sample pieces were put in two plastic containers, in the measuring chambers. Before the measurements, RH probes were calibrated using saturated salt solutions with known constant RH in the range of 65-95%. More detailed information about the internal RH measurement procedure can be found in the study of Huang and Ye [26]. The RH was measured at 28 days, and for all exposure conditions at 1 year. The procedure for pore solution extraction from AAS pastes is adopted from [27]. Pressures of up to 750 MPa were used to extract pore fluid from the samples. The cylindrical paste samples with 35 mm diameter and 70 mm height were used for pore solution extraction. After extraction, the pore solution was filtered and half of each solution was diluted using nitric acid (0.2 vol%). The diluted solutions were analyzed in a Perkin Elmer Optima 5300 DV apparatus. The nondiluted solutions were left for measurement of OHconcentration in the pore solution by titration against HCl acid. It should be noted that it was not possible to obtain a sufficient amount of pore solution in all cases. In samples with sufficient extracted volume of pore solution, pH was determined through measurement of OHconcentration. For samples where this was not the case, the pH measurements were carried out on the simulated pore solution, by dissolving 1 g of powdered sample in 10 ml of deionized water. The suspension was magnet rotated and maintained during 15 min at ambient temperature. Subsequently, the pH of pore solution was measured by pH meter 827 Metrohm. To verify the suspension method, samples with sufficient extracted pore solution for analysis were also ground into powder and mixed with water to obtain suspension and measure their pH. This way, the suspension method was shown to be a good indicator of the pH value, in addition to the standard method (i.e. analysis of the extracted pore solution). The sample preparation for N 2 adsorption was the same as for ESEM analysis with the exception that samples were not impregnated, instead they were weighed and degassed under vacuum at 25°C. Approximately 1 g of sample was used for the analysis. The N 2 -adsorption tests were conducted by using Gemini VII 2390 with a relative pressure P/P 0 range from 0.05 to 0.99 (the relative pressure is defined as the equilibrium vapor pressure divided by the saturation vapor pressure). A relative pressure of 0.99 corresponds to a pore-size of 193.5 nm, which is a maximum pore size that can be assessed. The volumetric method was used to determine the molar quantity of nitrogen adsorbed on hardened AAS pastes from the pressure and temperature measurement, using the real gas equation of state. The Brunauer-Emmett-Teller (BET) [28] gas-adsorption method has been applied for determination of the surface area of samples. The pore size distribution is obtained by Barrett-Joyner-Halenda (BJH) [29], assuming cylindrical, nonconnecting pore geometry. Carbonation of Powdered Pastes FT-IR Study can be observed at t = 0 for AAS paste which is assigned to the bending of Al-O-Si bands in ring structures consistent with the study of Bernal et al. [32]. This band also appears in unreacted GGBFS (Fig. 6). The effect of accelerated carbonation on the local bonding environment in AAS powdered paste is shown also in Fig. 6. The AAS paste was ground to provide more reaction sites on a well distributed surface grain area of the paste powder. The FT-IR has different vibrational bands for carbonates due to their different local bonding environment [14]. They are marked in Fig. 6 as: m 3 CO for asymmetric stretching vibration of carbonates, m 2 CO for out of plane bending vibration of carbonates, m 4 CO for in plane bending vibration of bicarbonates, centered at 1471, 856 and 715 cm -1 , respectively. These bands correspond to calcium carbonates, aragonite and calcite, with an exception that band at 715 cm -1 can correspond also to the bending of Al-O-Si bands as observed for noncarbonated powder. The first significant change between reference and carbonated powder can be seen after 1 h of carbonation exposure. The peak at 856 cm -1 is formed and its intensity increases with longer exposure. The shoulder at * 840 cm -1 in the carbonated samples can be also observed. This band can be assigned to a bending mode in Fig. 6 FT-IR spectra of powdered AAS exposed to accelerated carbonation (Color figure online) Journal of Sustainable Metallurgy the HCO 3 ion [11]. The Si-O-T (with T = Al, Si) band shifts from lower toward higher wavenumbers (943 ? 1015 cm -1 ). This indicates polymerization of silicate gel due to decalcification of the main reaction product of AAS paste, C-(N-)A-S-H gel, such as found by Bernal et al. [11]. The m 3 CO band centered at 1471 cm -1 and Si-O-T band at 1015 cm -1 , did not change intensity after 393 days of exposure. However, intensity of the peaks at 715 and 856 cm -1 is more pronounced after 393 days, and the shoulder at 1409 cm -1 disappeared. Figure 7 compares XRD diffractograms of the reference raw GGBFS and the carbonated raw GGBFS. The reference sample is fully amorphous with regard to the large diffuse diffraction peak centered at 30°2 Theta, while the intensity of the same peak decreases in carbonated raw GGBFS. It seems that carbonation suppresses the amorphous phase in raw GGBFS. Nevertheless, no other phases were formed, suggesting that raw GGBFS is inert to carbonation. Figure 8 compares the XRD diffractograms of reference and carbonated powdered AAS paste after 28 days of accelerated carbonation. The powdered AAS paste shows no crystalline phases, compared to carbonated powdered paste in which two different types of carbonates, sodiumbased and calcium-based carbonates were identified. Nahcolite (NaHCO 3 ) and gaylusite (Na 2 Ca(CO 3 ) 2 Á5H 2 O) were identified as sodium-based carbonates, which originated from the carbonation of the Na ? and Ca 2? ions from the aqueous solution. The identification of nahcolite is consistent with the FT-IR data (Fig. 6). The formation of aqueous solution may be favored in the accelerated carbonation environment between moisture, CO 2 molecules and free alkalis from the powder surface. On the other hand, calcium carbonates, i.e. vaterite, aragonite and calcite were formed due to carbonation of C-(N-)A-S-H gel. Carbonation mechanism can be explained by the cationexchanging reaction of Ca 2? from the gel inter-or sheet layers with H ? or Na ? from the aqueous solution. Subsequently, reaction of Ca 2? with CO 3 2from aqueous solution results in calcium carbonates formation, when the solution is supersaturated with these species. Consequently, polymerization of decalcified gel occurs by condensation of neighboring groups Si-OH or Si-O-Na into silicate gel, as observed in FT-IR analyses. Carbonation products and CO 2 uptake were further studied by the TG-MS technique. TG-MS Study TG and MS results were combined to measure the maximum uptake of CO 2 from the carbonated powder. Considering both the MS curve for CO 2 and DTG curve (Fig. 9), the weight loss in the range of 230-730°C is indicated as the temperature range for decomposition of metastable (low-crystalline) and stable calcium carbonates [33,34] and this range was used for CO 2 uptake. The maximum CO 2 uptake was 14% by weight of dried powder, as calculated according to Taylor's method [35]. A similar value was reported for high strength Klockner Oxygen Blown Maxhutte steel slag binder activated by carbonation with a CO 2 uptake of 13% [36]. Furthermore, besides two sharp features in the MS curve, corresponding to different atomic structural forms of calcium carbonates present in the carbonated powder, one more carbonate phase can be observed at a very low temperature, a narrow shoulder around 140°C. The appearance of this shoulder in the MS curve indicates a rather low amount of this phase in carbonated powder, compared to calcium carbonates and Journal of Sustainable Metallurgy therefore it was not included for calculation of the CO 2 uptake. It is believed that the present phase is nahcolite (NaHCO 3 ), which was previously identified by the XRD and FT-IR analysis. Pore Solution Study The RH and pH measurements of the pastes in different exposures and time intervals are presented in Table 4. The internal RH results show that samples had 60-70% RH at which CO 2 diffusion is suggested to be the fastest [15], except for AAS that was exposed outdoor. In outdoor conditions, i.e. in climate conditions of the Netherlands, RH varies between 80 and 99% RH. Therefore, samples are expected to be saturated such as measured in this study (RH outdoor * 95%). However, wetting and drying of the samples during 1 year can provide perfect conditions for carbonation, which is reflected by element concentrations, primarily by reduction of Na ? and K ? . From Table 4 it can be clearly seen that Na ? content was reduced 10 times, and so were other elements, except Fe. Considering that the samples were exposed in the outdoor (unsheltered) conditions, the loss of Na ? can also be attributed to the leaching of unbonded Na ? . Although the alkalis (Na ? , K ? , Ca 2? ) in the pore solution were reduced significantly, the pH did not drop below 11. The Na content obtained in outdoor exposed AAS was similar to Na content in noncarbonated GGBFS cements, such as found by Kempl and Ç opuroglu [27]. The corresponding pH value was above 12 in the noncarbonated GGBFS pastes. The minimum pH detected in natural outdoor conditions was 12.18. The high pH values, regardless of the pH measurement method, suggested strong alkaline media under all exposures after 1 year, indicating slightly carbonated or noncarbonated pore solution of AAS pastes. FT-IR Study FT-IR spectra of the bulk samples are presented in Fig. 10. After 1 year, peak locations for chemical bands of sealed and exposed AAS pastes were similar, suggesting no carbonation effect on the molecular structure of the gel, except for the samples that were exposed outdoor. The band at 1471 cm -1 indicated the presence of carbonates, m 3 CO. However, this was a weak band compared to the bands of carbonates in spectra of powdered AAS (Fig. 6). In addition, the Si-O vibrational band also shifted to higher wavenumbers (943 ? 957), indicating that the degree of silicon polymerization increased. 123 XRD Study Figure 11 shows the XRD diffractograms of bulk AAS pastes from all exposure conditions after 1 year. Comparing the features of the XRD diffractograms, there is no significant difference between reference and exposed samples. Carbonation of the C-(N-)A-S-H gel was equally limited in accelerated conditions. These results indicate that the material should be very dense, such that the pore sizes are close to molecular dimensions. These pores would then restrict CO 2 gas diffusion on the basis of their molecular dimensions rather than CO 2 molecular weight and concentration. This will be further discussed in the pore structure section. The TG results combined with respective MS curves for release of H 2 O and CO 2 are presented in Fig. 12a, b. The TG-DTG curves of all the pastes exhibited a one-stage thermal degradation process, suggesting a homogeneous reaction product (Fig. 12a). The mass loss at 105-345°C corresponds to the emission of interlayer and chemically bound water from the C-(N-)A-S-H gel. From the TG curves, the highest mass loss is measured in the conditions of accelerated carbonation. It is assumed that in these conditions gel was continuously developing due to the constant control of relative humidity value ( Table 2). The MS curves clearly show that the release of water was dominant (Fig. 12b). Very little CO 2 was released for AAS paste in the natural outdoor conditions suggesting slight carbonation of the paste, as it was identified by FT-IR study. Microstructure Study by ESEM Figure 13 demonstrates microstructure morphology obtained by ESEM imaging. AAS pastes have similar microstructure either after 56 days of preconditioning or 1 year of natural laboratory carbonation where the gel is uniformly dispersed around the unreacted GGBFS particles. No evidence of carbonation was observed, which was confirmed by previous studies, such as with XRD study (Fig. 11). Similar microstructure was observed for water glass activated GGBFS containing 13.2% of MgO after 1 year by Haha et al. [37]. According to their results, such dense microstructure has very low porosity (max. 3 vol%) after 28 days of curing, which was also found in this study. Regarding microcracking of AAS such as often observed for dense and brittle materials, in this work no cracking was found, contrary to observations of Bernal et al. [13] and Brough et al. [38]. It can be concluded that the 28 days of sealed cured AAS pastes provided a volumetrically stable microstructural development. Pore Structure Study by N 2 Adsorption The pore structure was first tested by mercury intrusion porosimetry (MIP) method. The intrusion volume was extremely low so it was not possible to detect any characteristic pore diameter by MIP in AAS [39]. Therefore, N 2 -adsorption technique was applied for pore structure analysis. Before considering pore size distribution, adsorption isotherms should be considered first. Figure 14a shows N 2 -adsorption isotherms for four samples. The distinctive isotherm for the sample in the natural outdoor conditions has a steady increase in the adsorbed volume at approximately p/p 0 = 0.4 and a sharp increase in adsorbed volume from 0.4 to around p/p 0 = 0.96. The sharp increase is representative of bulk pore filling such as shown in [40]. Other curves indicated almost nonporous material property since the quantity of adsorbed N 2 was nearly 0.02 mmol/g. The shape of a N 2 -adsorption isotherm can be used as a source of qualitative structural information to determine whether the pores present in the sample are micropores (below 2 nm), mesopores (between 2 and 50 nm), or macropores (above 50 nm) [41]. The isotherms in Fig. 14a are identified as slit-shaped pores according to the H3 type [41], similar to the synthesized C-S-H and montmorillonite isotherms identified in study by Costoya [42]. Comparison of the quantities of adsorbed N 2 between AAS paste in outdoor conditions (0.138 mmol/g) and for C 3 S paste (* 2 mmol/g) [42], indicated a much denser AAS paste microstructure as demonstrated by ESEM-BSE images (Fig. 13). Converting the relative pressures (p/p 0 ) to pore diameters and obtaining the pore volume (Fig. 14b), it can be seen that a significantly higher volume is found for AAS in outdoor conditions (5 9 10 -3 cm 3 /g) compared to other exposures (\ 1 9 10 -3 cm 3 /g). This can be due to more severe outdoor weathering of the material. The visually observed cracking of the samples, beside gel porosity, might contribute to pore volume increase. The BET N2 surface areas of AAS in studied exposure conditions after 1 year follows the same trend as the pore volume change. The BET N2 surface area of AAS in outdoor conditions (1.72 m 2 /g) was much higher than found for AAS in natural laboratory conditions (0.25 m 2 /g) or for the reference sample (0.79 m 2 /g). The pore structure study shows that the pore volumes measured in AAS pastes are far lower than the pore volumes normally found in OPC-GGBFSbased systems [34]. Furthermore, pore size distribution (PSD) is in the range of mesopores with diameter of 2 nm and maximum pore diameter of 15 nm, as shown in Fig. 14c. The transport mode for this range of pores sizes conforms Knudsen diffusion, according to Houst and Wittmann [43]. This means that the pore sizes are smaller than the mean free path of the gas molecules (O 2 or CO 2 ). Therefore, it is believed that accessible PSD is the main reason for carbonation resistance of AAS bulk paste samples in this study. Beside PSD, the transport of gases across the nanoporous material is also controlled by the pore connectivity [44]. In this regard, more fundamental insight using molecular level characterization techniques is needed to test the mode of CO 2 diffusion and interaction within such a dense system, such as in AAS. Figure 15 presents compressive strength development in AAS during 56 days of curing and preconditioning. The results are higher than the results reported in previous studies [33,34]. This is mainly due to the difference in alkaline solution composition (% Na 2 O, SiO 2 /Na 2 O modulus ratio). The relatively low alkaline conditions used in this work (4.8% Na 2 O/100 g of GGBFS compared to the 5, 10, 15 wt% Na 2 O in [33]) provided effective dissolution of GGBFS and formation of C-(N-)A-S-H gel, which is responsible for the strength gain. Furthermore, the results from this study suggested that the maximum compressive strength gain for AAS paste is relatively low beyond 14 days. An extended curing time in the fog room does not contribute to further increase of the compressive strength. In contrast, curing time for obtaining the maximum compressive strength in OPC-based paste is around 60 days, as reported by Chindaprasirt et al. [45]. The comparison is based on similar w/b ratio (0.35 for OPC-based paste and 0.38 for AAS paste). Compressive Strength Regarding the effect of carbonation on compressive strength, accelerated carbonation can decrease, but also it can increase compressive strength of AAS mortars, depending on the type of alkaline activator used and exposure conditions [15]. Figure 16 compares the compressive strength results of AAS paste samples exposed in natural laboratory and accelerated carbonation conditions. It can be seen that there is no significant difference in strength results from two different exposure conditions in this study. Conclusions Carbonation resistance of AAS under natural and accelerated conditions was evaluated. Two aspects were studied, the potential carbonation of the main binding phases in finely powdered paste and the reactivity and diffusivity of CO 2 within the bulk paste. In general, the carbonation of raw GGBFS was not possible, while carbonation of AAS powdered paste involves one reaction: the carbonation of gel (C-(N-)A-S-H). This reaction causes the gel decomposition and formation of different calcium carbonates. The majority of the CO 2 uptake was occurring within the first 28 days of accelerated carbonation. Nevertheless, the bulk samples were highly resistant to carbonation, regardless of the exposure conditions. Carbonation of bulk AAS pastes was inhibited due to their dense microstructure, the presence of extremely small pores with sizes ranging from subnanometers to tens of nanometers, and high alkalinity (pH values [ 12.18). Nitrogen-adsorption tests identified the gel pores (\ 15 nm) to be dominant in the system, which blocked CO 2 to diffuse into the pastes during the exposure period. The dense AAS microstructure exhibited high mechanical strength (* 110 MPa). The results of this study shed light on the importance and the effects of the physical properties (density, pore size distribution) and alkalinity (pH) of AAS pastes on their long-term performance and durability. Materials innovation institute M2i (www.m2i.nl) and the Technology Foundation STW (www.stw.nl), which is part of the Netherlands Organisation for Scientific Research (www.nwo.nl). The authors gratefully acknowledge John van den Berg for his help with pore solution study. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://crea tivecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.	2019-04-30T13:06:55.456Z	2018-02-16T00:00:00.000	{ "year": 2018, "sha1": "cd957faccc03b23f79d84feb4ccc9087400bd22c", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s40831-018-0166-4.pdf", "oa_status": "HYBRID", "pdf_src": "ScienceParsePlus", "pdf_hash": "ab663d17a54e16db7f275ed5e5ad85b32c305614", "s2fieldsofstudy": [ "Materials Science" ], "extfieldsofstudy": [ "Materials Science" ] }
260227741	pes2o/s2orc	v3-fos-license	Demographic Factors Influencing Music Teachers’ Attitudes Toward World Music Teaching in China This study aims to identify the demographic factors affecting Chinese music teachers’ attitudes toward world music teaching in China. Through electronic questionnaires issued online, a total of 1,368 participants completed A Survey of Music Teachers’ Attitudes Toward World Music Teaching , which includes demographic information and twenty Likert-type items. The results showed that two independent variables of “gender” and “whether teachers had received training in world music after entry” did not influence Chinese music teachers’ attitudes and practices in world music teaching. Furthermore, it could be speculated that the teachers, especially those who were 20-34 years old with a master’s degree/Ph.D. and 1-10 years of teaching experience, as well as good at Western instruments, studied world music courses, graduated from a comprehensive university, and with teaching places located in the provincial capital cities, leaned more toward ethnomusicological and multicultural consciousness in world music teaching. I. INTRODUCTION As an important item in multicultural education, World Music Education has long been recognized for its value. In 1998, the International Music Council held a discussion on music education in a multicultural society. The conference pointed out that the teaching of cultural diversity in music education could help break down existing cultural boundaries, increase social respect and understanding across borders, reduce racial tensions in schools, build new cultural identities and social consensus, satisfy human curiosity, and bring happiness to participants (Traasdahl, 1998). Therefore, the broader goal of world music teaching is not only to understand music itself but also to care about people, society, and the world through music. With the development of music education, Western art music as the core and basis of the music curriculum is now generally questioned and even opposed (McPherson, 2006). The International Society for Music Education (ISME) also continues to question and adjust the old music education system and the music teaching model. Goldsworthy (1989) pointed out that listening to European classical music was largely considered an individual private experience-an enjoyment for aesthetic reasons, whereas world music was often seen as part of a collective experience in a culture based on auditory/oral traditions, serving everyday functions. In addition, when teaching Western classical music, teachers focused on introducing "classic" pieces of music from various periods, which were "frozen." In contrast, the literature on world music has shown that "classics" were challenged, while the transmission, flow, and change of music were valued. Many teaching concepts and methods of world music courses should be different from the appreciation and teaching modes of Western classical music. However, a monoculture centered on Western (art) music still pervades the thinking on music and music education in the Western world and most of its practice (Schippers, 2010). Zhuang and Pan's (2022) survey of Chinese music teachers' attitudes toward world music teaching also showed that Chinese music teachers were influenced more by Western centralist values than by multicultural music education and ethnomusicology, in which human music was interpreted from a cultural perspective. While teachers play an important role in the implementation of teaching, their attitudes toward teaching different music cultures will directly affect their teaching practices. In addition, many studies have confirmed the influence of teachers' demographic factors on their attitudes toward multicultural music teaching (Wong, 2014;Zhang, 2007;Lu, 2013). As such, this study aims to verify the impacts of the demographic variables of Chinese music teachers on their attitudes toward world music teaching. Clarifying these factors will help teachers' administrative departments to select target groups and the issues to be targeted in the training of world music teaching for music teachers. Besides playing a reference role in whether some universities should strengthen the teaching of world music courses for students, it also provides advice to the textbook publishing department. Hence, the study aims to address the following questions: 1) Does the "age" variable have significant differences in music teachers' attitudes toward world music teaching? 2) Does the "gender" variable have significant differences in music teachers' attitudes toward world music teaching? 3) Does the "educational level" variable have significant differences in music teachers' attitudes toward world music teaching? 4) Does the "years of teaching" variable have significant differences in music teachers' attitudes toward world music teaching? 5) Does the "musical skill" variable have significant differences in music teachers' attitudes toward world music teaching? 6) Does the "have taken ethnomusicology and world music courses" variable have significant differences in music teachers' attitudes toward world music teaching? 7) Does the "graduated from different types of colleges" variable have significant differences in music teachers' attitudes toward world music teaching? 8) Does the "teaching location" variable have significant differences in music teachers' attitudes toward world music teaching? 9) Does the "had participated in the training of world music teaching" variable have significant differences in music teachers' attitudes toward world music teaching? II. LITERATURE REVIEW Peppers' (2010) study examined teachers' attitudes toward assessment and their relationships with the demographic factors in Michigan elementary general music classrooms. Respondents who received their last music education degree more than 20 years ago experienced greater difficulty in assessment-related tasks and had more negative attitudes toward the assessment, while those with a moderate number of years (10-19) since their last music education degree believed that they had not been adequately prepared in college. Peppers believed that the relationship between teachers' attitudes toward assessment and their demographic backgrounds might reveal information that is useful for designing degree programs and workshops that are geared toward teachers' needs. Likewise, an examination of the relationship between teachers' attitudes toward suggested assessment improvements and demographic information could also allow colleges and administration to meet teachers' needs. Butler et al. (2007) suggested that, among other factors, preservice teachers' racial, ethnic, and cultural backgrounds and experiences influenced how they developed as teachers. Consequently, the components of the teacher dimension-to the extent that they are mediated by race, ethnicity, and culture-might impact teacher candidates' ability to develop knowledge, skills, and dispositions, thus contributing to cross-cultural competence that may influence their preferences for teaching in culturally diverse educational settings as in-service professionals. Meanwhile, McKoy (2013) investigated the effects of race/ethnicity and the school community setting for early field experience practice and student teaching on music student teachers' self-reported crosscultural competence. Participants (N = 337) from 36 colleges and universities across the United States completed the survey and the results indicated no significant main effect of school community setting on participants' cross-cultural competence; however, a significant main effect of race/ethnicity (p < .05) was observed for the "Constrain" subscale of the survey. Besides, participants in the racial-ethnic minority also held fewer beliefs and attitudes that would hinder their readiness to teach in culturally diverse educational environments. There are many studies on the factors that influence the attitudes and practices of general music teachers toward multicultural music education. Moore (1995), for instance, verified that there was no significant correlation between teachers' gender, education level, and attitudes toward multicultural music education. Meanwhile, Petersen (2005) found that age and teaching experience had no significant effect on the multicultural level of general music teachers. However, it is interesting to note that average music teachers from 45 to 54 years of age with more than 16 years of teaching experience had more positive attitudes toward multiculturalism. The results of Wong's (2014) study in Malaysia were also consistent with Moore's (1995) and Petersen (2005)'s in the USA. In addition, Wong found that although gender had no significant effect on teachers' levels of multiculturalism, the results (the average results) indicated that women had more positive attitudes toward multiculturalism than men. Moreover, in Wong's (2014) study, ethnic identity and religion seemed to be important variables affecting the level of multiculturalism among general music teachers. Educational attainment, especially among teachers with master's and doctoral degrees, was also more favorable toward multiculturalism. Additionally, Zhang (2007) investigated the attitudes of primary and secondary school music teachers in Lanzhou, China, toward multi-cultural music. Among the demographic factors affecting teachers' attitudes, the study found no significant correlation between educational background and teachers' attitudes toward multi-cultural music. The minority teachers did not show obvious demand for multiculturalism, but they showed more negative attitudes than the Han teachers. Furthermore, teachers' music learning experience also had a significant correlation with their attitudes toward multi-cultural music. For instance, teachers who only learned Western instruments had more negative attitudes toward multi-cultural music. Nonetheless, the study found that teachers who enjoyed listening to minority music were more positive about multi-cultural music, while those who enjoyed European classical music showed the opposite. Lu (2013) investigated the differences in music teachers' views of multicultural music teaching in Junior high schools in New Taipei City, Taiwan, with different background variables such as personal backgrounds (gender, highest education), teaching backgrounds (teacher status, teaching years), and curriculum experience (pre-service, in-service). The results showed that music teachers in the Junior high schools in New Taipei City held similar views of multicultural music teaching, and their views were not affected by different background variables. A. Respondents and Procedure The respondents of this study include music teachers in primary and secondary schools nationwide in China. Considering the representativeness of the samples, the researchers used the simple random sampling technique. Through electronic questionnaires issued online, a total of 1,396 questionnaires were collected from the respondents, of which 1,368 were valid questionnaires with an efficiency of 98%. The questionnaire collection period was from May to November 2019. B. Instrument A Survey of Music Teachers' Attitudes Toward World Music Teaching was developed by the researchers. The questionnaire can be divided into two parts. 1) Part one Your Basic Situation collects a total of 9 demographic information of respondents such as age, gender, education level, and years of teaching. Music teachers provided demographic information about themselves through self-reported items on the survey, and the distribution of the categories in the variables reflects an extensive sample. The characteristics of the respondents are described in Table I. In China, the source of music teachers in primary and secondary schools mainly comes from the Conservatory of Music, the Academy of Arts, the School of Music in Comprehensive Universities and Normal Universities, as well as various public and private Art Vocational Colleges. Currently, there are 12 Conservatories of Music in China, including the Central Conservatory of Music, which pay more attention to students' professional performance. Due to the small enrollment, it is very rare for students to be admitted to such a conservatory. Those who are strong performers and fail to be admitted to the Conservatories of Music may be admitted to the Academy of Arts. The Conservatory of Music is an institution of higher learning specializing in cultivating talents engaged in music art education and research, whereas the Academy of Arts can be understood as a comprehensive university of arts, offering not only music but also dance, fine arts, drama, media, and so on. The Schools of Music in Comprehensive Universities and Normal Universities recruit students with relatively weak professional abilities but high academic scores. Students with weaker professional ability and lower academic scores may enter various public and private Art Vocational Colleges. In this study, 69.4% of the respondents came from the third category --Schools of Music in Comprehensive Universities and Normal Universities. The administrative divisions of the People's Republic of China are composed of provincial-level administrative regions, prefecture-level administrative regions, county-level administrative regions, and township administrative regions. Beijing, Shanghai, Tianjin, and Chongqing are called municipalities. They have the same administrative status as provincial-level administrative regions, but they are directly under the central government. According to the administrative divisions of China, the respondents' teaching locations in this study were classified into Municipality, Provincial capital, Prefecture-level city, Countylevel city, and Rural area. 2) Part two The Attitude Survey includes 20 questions. Respondents provide their feedback to each statement using a five-point Likert Scale from "strongly disagree" to "strongly agree". To assess the internal consistency of the Attitude Survey, the researchers used Cronbach's alpha to measure the 20 items. Cronbach's alpha is a standard method for measuring the reliability of psychological or educational studies. Generally, the higher the score, the higher the reliability. In basic research, the reliability should reach at least 0.80 to be acceptable. In this study, Cronbach's alpha reached 0.872, thus indicating that the reliability level is good. To assess the construct validity of the Attitude Survey, the researchers conducted Exploratory Factor Analysis. The goodness-of-fit test results showed a KMO value of 0.884, while the Bartlett test of sphericity indicators reached the significance level. Since all indicators reached a statistically required standard, the factor analysis of the data was deemed appropriate. Additionally, Principal Component Analysis and Varimax Orthogonal Rotation were also used to extract the common factors. The number of factors was determined by the eigenvalues greater than 1, combined with the scree plot. Overall, the results showed a clear five-factor structure, which is consistent with the research assumption, and the Total Variance Explained was 62.248%. In the five-factor structure, the loading was higher than 0.60 and the proportion of each item determined by the common factor was more than 0.50. Therefore, the above data indicated that the questionnaire had a good structure. The five-factor structure is shown in Table II. C. Data Analysis According to the above five-factor structure, the 20 items in the Attitude Survey were divided into five dimensions: "behavioral intention toward teaching methods" (items 1-6), "evaluation of teaching materials" (items 7-10), "cognition of authenticity, context, notation/oral transmission" (items 11-14), "world view of music" (items 15-17), and "intention to implement teaching" (items 18-20). Each of the nine independent variables collected from Your Basic Situation was compared with the five dimensions. Subsequently, the data collected from the questionnaires were compiled and analyzed using quantitative measures. SPSS 26.0 software was used to process and analyze the collected data statistically, and the statistical methods used mainly include descriptive statistics, the independent sample T-test, analysis of variance, factor analysis, and the Chi-square test. IV. RESULTS AND DISCUSSIONS In terms of the comparison of the age variable with the five dimensions in the Attitude Survey, the results showed significant differences involving three dimensions, namely "evaluation of teaching materials," "cognition of authenticity, context, and transmission mode," and "intention to implement teaching" among teachers of different age groups. Overall, the scores for teachers aged 20-34 years were lower than those aged 35-44 and 45-54 years. Combining the statements of items in the three dimensions above, the researchers speculated that teachers aged 20-34 years were more ethnomusicological in world music teaching. There was no significant difference between teachers of different genders in each dimension; thus, gender was not a factor affecting attitudes toward teaching world music. This finding is also consistent with previous studies (Petersen, 2005;Wong, 2014). In terms of the comparison of the education level variable with the five dimensions of the Attitude Survey, the results showed significant differences involving three dimensions, namely "evaluation of teaching materials," "cognition of authenticity, context, and transmission mode," and "intention to implement teaching" among teachers of different education levels. The scores for teachers with a master's degree and above were lower than those who graduated from junior college and were Bachelor's degree holders. Combining the statements of items in the three dimensions above, the researchers speculated that teachers with a master's degree/PhD were more ethnomusicological in world music teaching. Besides, this finding is consistent with Wong's (2014) study in which the educational level, specifically those with a master's degree/PhD, showed more positive attitudes toward multiculturalism. In terms of the comparison of the years of teaching variable with the five dimensions in the Attitude Survey, the results showed significant differences in two dimensions of "evaluation of teaching materials" and "intention to implement teaching" among teachers. Regarding the "evaluation of teaching materials," the scores for teachers with a teaching experience of 1-5 years did not differ from those who have been teaching for 6-10 years but were lower than those who have been teaching for more than 11 years. Meanwhile, the scores for teachers with 6-10 years of teaching did not differ from those with 11-15 years but were lower than those with 16 years or more. As for the "intention to implement teaching," the scores for teachers with 6-10 years of teaching were lower than those in all other groups. Combining the statements of items in the two dimensions above, the researchers speculated that teachers with teaching experience of 1-10 years were more ethnomusicological in world music teaching. In terms of the comparison of the musical skill variable with the five dimensions of the Attitude Survey, the results showed a significant difference in the dimension of "intention to implement teaching" among teachers with different musical skills. Evidently, the score for teachers who were skilled at Western musical instruments was lower than those who were good at vocals and others. Combining the statements of items in the dimension above, the researchers speculated that teachers with the said musical skill, namely Western musical instruments, were more ethnomusicological in world music teaching. This is indeed an interesting finding that further research may examine and account for in the future. In terms of the comparison of "whether teachers took ethnomusicology and world music course" with the five dimensions of the Attitude Survey, the results showed significant differences involving the dimensions of "evaluation of teaching materials, " "cognition of authenticity, context, and transmission mode, " and "intention to implement teaching" among teachers. Unexpectedly, in the above three dimensions, the scores for teachers who had learned world music were lower than those who had learned ethnomusicology. Since the basic theory of world music comes from ethnomusicology, the researchers hypothesized that teachers who had learned world music scores should not be significantly different from those of teachers who had learned ethnomusicology. In this regard, the researchers boldly inferred that the reason for this phenomenon is attributed to the confusing understanding of the Chinese name for "Ethnomusicology." In addition to the small number of schools offering "Ethnomusicology," schools have mostly offered two more popular courses: "Ethnic and Folk Music" and "Traditional Chinese Music." These are different courses; while "Ethnomusicology" studies music in its social and cultural contexts, the latter still focuses on the study of music itself. Moreover, many teachers could not tell the difference between "Ethnomusicology" and the "Ethnic and Folk Music" or the "Traditional Chinese Music" because they thought they were the same course, and they only understood that "Ethnomusicology" was literally a subject of traditional and folk Chinese music. Although 257 teachers in the present survey claimed that they had taken the course "Ethnomusicology" and 607 teachers claimed that they had taken both along with world music, the researchers speculated that some teachers regarded the "Ethnic and Folk Music" or "Traditional Chinese Music" and other courses as "Ethnomusicology" because of the above reasons. In terms of the comparison of the types of schools graduated by teachers with the five dimensions of the Attitude Survey, the results showed significant differences in the "cognition of authenticity, context, and transmission mode" and "intention to implement teaching" among teachers who graduated from different types of schools. Based on the results related to the above two dimensions, the scores for teachers who graduated from universities were lower than those from conservatories of music. Essentially, this means that teachers who graduated from universities were more likely to be "multicultural" when teaching world music. In contrast, those who graduated from conservatories of music were more likely to be "monocultural." While teachers who graduated from conservatories of music are undoubtedly more professionally competent, in line with Schippers' (2010) assertion: "Ironically, as music students develop toward being professionals, they are likely to focus on a single culture." (p.32) In terms of the comparison of the teaching location variable with the five dimensions of the Attitude Survey, the results showed significant differences in the "evaluation of teaching materials," "cognition of authenticity, context, and transmission mode," and "intention to implement teaching" among teachers with different teaching locations. Overall, in the above three dimensions, the scores for teachers with teaching locations in provincial capitals were lower. Therefore, when teaching world music, teachers with teaching locations in provincial capitals were more likely to be "ethnomusicological." Based on the results, there were no significant differences in whether the teachers had participated in "world music" training in each dimension of the Attitude Survey. This finding, however, is not consistent with the expected assumption; thus, further research on this subject can be carried out in the future to reveal the possible reasons.	2023-07-28T15:09:51.987Z	2023-07-25T00:00:00.000	{ "year": 2023, "sha1": "f62cb56bf27ed5d1a165c33ad1c1a91905775b55", "oa_license": "CCBYNC", "oa_url": "https://ej-social.org/index.php/ejsocial/article/download/481/235", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "91892006368e62023ae717536bc2d6f595c19287", "s2fieldsofstudy": [ "Education" ], "extfieldsofstudy": [] }
259041799	pes2o/s2orc	v3-fos-license	New weighted BERT features and multi-CNN models to enhance the performance of MOOC posts classification Learning is an essential requirement for humans, and its means have evolved. Ten years ago, Massive Open Online Courses (MOOCs) were introduced, attracting many interests and learners. MOOCs provide forums for learners to interact with instructors and to express any problems they encounter in the educational process. However, MOOCs have a high dropout rate due to the difficulties of following up on learners' posts and identifying the urgent ones to react quickly. This research aims to assist instructors in automatically identifying urgent posts, making it easier to respond to such posts rapidly, increasing learner engagement, and improving course completion rate. In this paper, we propose a novel classification model for identifying urgent posts. The proposed model consists of four stages. In the first stage, the post-text is code-encoded and vectorized using a pre-trained BERT model. In the second stage, a novel feature aggregation model is proposed to reveal data-based relationships between token features and their representation in a higher-level feature. In the third stage, a novel model based on convolutional neural networks (CNNs) is proposed to reveal the meaning of a text context more accurately. In the last stage, the extracted composite features are used to classify the text of the post. Several experimental studies were conducted to get the best performance of the proposed stages of the system. The experimental results demonstrated the architectural efficiency of the proposed feature aggregation and multiple CNN models, as well as the accuracy of the proposed system compared to the current research. Introduction Massive Open Online Courses (MOOCs) date back to 2008, when videos were used as a medium of instruction. When Stanford University launched its free online courses in 2012, nearly 300,000 learners enrolled. Currently, the number of learners through MOOCs is more than 220 million worldwide [1]. Despite their popularity, MOOCs face many challenges, such as completion rate, accountability, accreditation, accessibility, and financial sustainability [2]. One of the most significant challenges facing MOOCs is the low completion rate, which represents a major obstacle to achieving the goal of the educational process itself. One of the main reasons for it is the lack of interaction between the instructor and the learner [3][4][5]. MOOCs provide forums for learners to communicate with one another and instructors [6]. Hence, it provides a means for the learners to express the problems and obstacles encountered during the learning process [7,8]. Despite numerous learners' posts during the course, only about 20% are urgent and require the instructor's attention [7]. As a result, the instructor is overwhelmed with identifying urgent posts that require a quick response. Automatically informing instructors in real-time of urgent posts is one of the most important tools for improving engagement, reducing dropouts, and increasing completion rates [9]. The identification of urgent posts is a text classification issue. Text classification problems are solved using traditional machine learning and deep learning techniques combined with natural language processing techniques. These technologies are employed in many applications, such as Twitter sentiment analysis [10], YouTube comments [11], emotion text classification [12], text plagiarism detection [13,14], and educational data mining [11]. To categorize MOOCs forum posts into urgent and nonurgent topics, extensive studies have been conducted based on word representation and classification approaches. In [7,8,15], statistical methods such as term frequency (TF), inverse document frequency (IDF), and term frequencyinverse document frequency TF-IDF are used to convert text terms into numerical representations. The text's word sequence is disregarded by these methods [16]. This leads to a defect in comprehending the document context, as the meaning of the context depends on the words' order and the relationship between each word and its neighbors. In [17][18][19], deep learning models are introduced for word representation, and each word is depicted by a dense vector that reflects its significance within the context of the document [20]. In [7,8,15], traditional classification algorithms, such as nearest centroid, SVM, and others, are constructed to classify MOOCs forum posts. These algorithms are straightforward and demand little computing power, but their performance largely depends on human correction, which helps in the feature selection process. When these algorithms give poor performance, humans intervene to redevelop experiments using feature selection techniques to select the effective features that can improve the performance. Additionally, these techniques can be learned from data sets that are comparatively small. Therefore, the best obtained F1-weighted score of these algorithms is 88% and 70% of the class of urgent posts. On the other hand, deep learning classification methods can improve their results through repetition without human involvement. They are also able to fit a large quantity of data and weigh the contribution of each feature to the decision-making process. In [17][18][19], deep learning algorithms are utilized for word representation and classification process. Pre-trained deep learning models such as Google News and GloVe are used for word embedding to extract more text numerical representation features. Multiple CNN, GRU, and attention layers are also constructed to develop the classifier. Their efforts focused on additional representational features, selecting the effective features, and giving the most significant features more weight. However, the best obtained F1-weighted score of these methods is 91.8% and 80.1% of the class of urgent posts, where intra-relationships between word features and inter-relationships among text words were not taken into consideration in revealing the post-context meaning that leads to detecting the urgent posts. In this study, a novel MOOC post-classifier is proposed to assist instructors in quickly accessing and responding to urgent posts. It considers disclosing post-context meaning, which helps classify urgent posts accurately. It is interested in embedding each word into effective numerical representational features, extracting intra-relationships between word features, and inter-relationships between text words that capture the post-context meaning. The proposed model consists of four stages: tokenizing and embedding, aggregation and weighting of token's features, extraction of post-context meaning, and classification. Generally, the main contributions of this research are summarized as follows: • A pre-trained BERT model is employed for tokenization and embedding processes; it can return different vectors for the same word depending on its context. • A novel feature aggregation model for token's features is proposed to uncover data-driven relationships between them and express them as a higher-level feature. It also considers the aggregated feature's weight with the word meaning's diversity within the posts. • A novel parallel CNNs model is constructed to extract the composite features of multiple words to reveal the post-context meaning. It consists of four different CNN architectures that can extract the different relationships between the sentence words and uncover the postcontext meaning. • The relevance of the post can be determined by analyzing the extracted composite features through the proposed neural network architecture in the classification stage of the proposed system. • Experimental comparative studies are conducted to evaluate the design efficiency of the feature aggregation model of token's features and the multi-CNN model to reveal the post-context meaning. The proposed system performance is evaluated and compared with state-of-the-art algorithms. The experiments were conducted using three groups of training and test datasets from the benchmark database of the Stanford MOOC post corpus, as predefined in [8,18,19]. The experimental results showed that the proposed model, compared with the other algorithms, achieved a significant improvement in the weighted F1 score and a balance between the precision and recall scores of urgent posts. The remainder of this paper is organized as follows: In Sect. 2, the related work will be explored. In Sect. 3, the proposed approach is explained in detail. In Sect. 4, the experiments and obtained results will be presented and discussed. Finally, in Sect. 5, the conclusion will be presented. Related work Due to the recent extensive use of MOOCs, several classification algorithms for MOOC post forums have been proposed. These algorithms can be divided into two categories based on the techniques employed: traditional machine learning and deep learning approaches [18]. Feng et al. [21] analyzed more than 100,000 discussion threads collected from Coursera. They found that most of the posts were unrelated to the course content. The posts were based on new features related to user interactions with different subforums. Linear regression combined with a gradient lifting decision tree (GBDT) is used to enhance the classification of a discussion thread. The advantage of this model is that it is based on features independent of the course content thread. It achieved an accuracy of 85%, which is an improvement of 12% compared to the baseline results [22]. Agrawal et al. [7] proposed a labeled MOOCs dataset. They proposed a two stages system, in the first stage, a classifier model is developed to identify confusing posts. In the second stage, a recommendation system is applied to recommend a short clip clarifying that confusion. The proposed solution achieved an F-score value of 77% of the confused label, depending on the course. Cui and Wise [23] used the binary support vector machines model to classify whether questions posts were related to course content or not. Bakharia et al. [15] conducted a comparative study to classify the Stanford MOOCs posts according to confusion, urgency, and sentiment labels, they compared the performance of three classifiers Naïve Bayes, support vector machine (RBF), and random forest. Almatrafi et al. [8] used a combination of metadata and linguistic features to build MOOCs urgent posts identification model. According to their results, the AdaBoost algorithm achieved the best results. All the previous methods used traditional machine learning algorithms, and even the best algorithm only managed to maximize the F-score value to 77% of the urgent label, which is insufficient. These results are particularly referred to the nature of MOOC posts, which are very short in most cases, contain spelling errors, and are very noisy. Therefore, the researchers used deep learning techniques to handle these problems. The recent improvements in hardware and computing power allow powerful deep learning algorithms to be applied and developed for large datasets. Classifying text using deep learning depends on embedding techniques to represent text. Generally, the text consists of words and characters, and different embedding techniques can represent both. Then, deep learning methods utilize the word and character embedding vectors to make the decision [18]. Ombabi et al. [24] proposed an opinion analysis algorithm using Twitter to summarize user interests. They used pre-trained Word2Vec as a word embedding technique and a combination of CNN and support vector machine (SVM) for opinion classification. SVM provided the final prediction based on the features and semantic information extracted by CNN. Sotthisopha et al. [25] proposed a short text classification algorithm based on multichannel CNN and the k-max-pooling layer. Also, they added preprocessing data module to maximize the coverage of word embedding. XI GUO et al. [18] proposed a hybrid model using features extracted from word embedding and character embedding. They evaluated the performance of the proposed model depending on the Google-news Vectors and GloVe. They reported that the proposed pre-trained word embedding model based on Google-news vectors offers a better result than that based on GloVe. Khodeir [19] used BERT as an embedding technique and Bi-GRU to build the classification model. However, the proposed solution slightly enhanced the results. Although the stateof-the-art algorithm achieved a weighted F1 value of 91.9%, this result does not accurately reflect the enhancement in the classification of urgent posts. The class ''not urgent'' achieved an F1 value of 94.8%, whereas the ''urgent'' achieved an F1 value of 81.2%. The results indicate that the model is still unable to identify urgent posts. Proposed system Identifying urgent posts in MOOCs becomes a significant challenge for instructors as the number of students and their posts grows. Therefore, an accurate model is proposed to classify MOOCs forum posts into urgent and non-urgent topics. It assists in prioritizing replies and managing many posts. The goal of the proposed model is to improve engagement, reduce dropouts, and increase completion rates. The proposed model is based on deep learning approaches for word representation and classification processes. It consists of four stages, tokenizing and embedding, aggregation and weighting of token's features, extraction of context meaning, and classification, as shown in Fig. 1. In the first stage, the proposed model takes a post-text as an input and tokenizes it into effective numerical representational features. Then, each token is represented by a vector. In the second stage, a novel features aggregation model is proposed to uncover data-driven relationships between token's features and represent them as higherlevel features. In the third stage, a novel multi-CNN model is constructed to reveal the context meaning of post-text, considering the discovery of the post meaning based on the different relationships between the sentence words. In the fourth stage, the extracted features are utilized to classify the post-text. Tokenizing and embedding stage The post-text for MOOCs is in an unstructured format. Therefore, the first stage of the proposed system is designed to preprocess the post-text and prepare it for use in the subsequent stages, which converts the post-text into numerical representation, as shown in Fig. 2. This stage employs the BERT model to perform tokenization and embedding processes on the input post, as shown in Fig. 3. The contextual token's embedding value depends on the token's position, segment's embedding, and token's embedding [26]. The main advantage of using the BERT model is that it can interpret the context of a word; it returns different vectors for the same word depending on the words around it. For example, the following text was tokenized and embedded by the BERT model. The BERT model first tokenized the text, and each token is represented by a predefined value depending on the words around it, as shown in Fig. 4. The word ''bank'' appeared in three positions with two different meanings. The BERT model represented each word with a vector proportional to its meaning, as shown in Table 1. The similarity between the word ''bank'' in the first two positions, ''bank vault'' and ''bank robber,'' equals 0.94, while the similarity in the last two positions, ''Bank robber'' and ''River bank,'' equals 0.69. The results Fig. 1 Overall structure of the proposed system demonstrate the advantage of using the BERT model to generate an embedding vector. This capability of the BERT model is utilized to improve token's representation and model performance. The proposed system is based on a pre-trained BERT-Base-Uncased model that is not sensitive to the case of the letter if it is a capital or small letter. The pre-trained BERT model transforms each token into f features. It has a fixed input length of n, where n is the number of input tokens in the text. A text with a larger number of tokens is terminated at token n. If the text length is shorter than n, the text will be padded. The input post will be transformed into a P tok array, as expressed in Eq. (1). Aggregation and weighting of token's features stage Feature aggregation is a technique that builds a global feature vector by integrating the various local features of a dataset instance to form the global features of data. In the pattern recognition field, the feature aggregation process is a method that takes many local features from an image and combines them into a single global feature vector. The purpose of aggregating features is to uncover data-driven relationships between instance features that may be difficult to detect. Each aggregated feature can be considered as a higher-level feature that epitomizes multiple lower-level features. Therefore, these higher-level features can reveal significantly more useful information than any single local feature. In this study, a novel feature aggregation model is proposed. The goal of this model is to uncover data-driven relationships between token's features and represent them as higher-level features. Furthermore, the proposed model considers the weighting of the aggregated feature ''token's global feature'' in different situations, especially when using the same token in different posts. This makes the model more flexible with the diversity of the word meaning within the posts, as each word may carry different meanings and weights in determining the importance of the post. The proposed model aggregates and weights each token's feature based on a deep learning approach. It consists of one convolution layer, with E filters of one size equal to 1 9 Fs. Each filter combines the f local features of a token and expresses them as a global feature. The proposed CNN model slides a filter over the input tokens. The token's feature values are multiplied by their corresponding values in the filter. Then, the result is summed up into a global value in the output channel/feature map to uncover data-driven relationships between the token's features. This global feature will be weighed into E values depending on the diversity of word meaning and importance of a word within the posts. It weights into E values using E filters that reflect the number of convolutional layer channels. In this stage, as shown in Fig. 5, the proposed aggregation and weighting model of token's features takes the P tok array as input and extracts the P wg array as expressed in Eq. (2). (4) illustrates how the number of output vectors Dn is computed, where f is the number of token's features, Fs is the filter width that has the same value of token's features, p is the padding value, s is the stride value, Fh is the filter height that equals to one, E is the number of channels, and each feature value F j of Tw i in P wg array is completed as indicated in Eq. (5), where g is a rectified linear unit (ReLU) function, and b ei is the bias. where f e ¼ W e Ã T i þ b ei Extraction of context meaning stage The text consists of related words. The context of the text is always based on a set of closely related words that sheds light not only on the meanings of single words but also on the meaning and purpose of the entire text. Therefore, the context of the text is the essence of the intended meaning in any textual or verbal structure. It sheds light not only on the word but also on the written text and the overall meaning through the relationship of the vocabulary to each other in any of the different contexts. The sentence meaning is revealed only through the contextualization of the linguistic unit, that is, placing it in different contexts. Therefore, in this study, a new proposed model aims at revealing the context of the post and recognizing its importance; it considers the discovery of the meaning based on the different relationships between sentence words. The purpose of this stage is to extract the composite features of multiple words to reveal the context of the post. It is based on a new multi-CNN model. This model is constructed with different convolutional filter sizes to extract the different relationships between the words in a text, which can reveal the post-context meaning. The proposed multi-CNN model consists of four parallel CNNs, as shown in Fig. 6. Each of them has two layers: convolutional and pooling layers. The first CNN, with a filter size of 1 9 F DE , is used to extract the best features of a single word that can be expressed within the context meaning of the post. The second CNN, with a filter size of 2 9 F DE , is used to find the set of two words features that can be highlighted in the context meaning of the post. The third CNN, with a filter size of 3 9 F DE , is developed to extract the expressive features of each of the three words in the context meaning of the post. The fourth CNN, with a filter size of 4 9 F DE , is structured to discover the best features of each of the four words that are prone to reveal the context meaning of the post. In this stage, the proposed model slides the convolutional layer filters of CNNs over the input tokens. In the multiple convolutional layer filters, the token's feature values are multiplied by their corresponding values. Then, the outputs of the convolutional layers are fed to the corresponding pooling layers to uncover the feature values of different groups of words, which can emphasize the context meaning of the post. The proposed multi-CNN model is input by the P wg array of the previous stage. Then, for each CNN, P Cm arrays are constructed as expressed in Eq. (6), where m is the number of CNN. Equation (7) illustrates how the number of output vectors Cm is determined, and Eq. (8) illustrates how the number of output features F C is computed, where f cl is the filter width that has the same value as token's features DE, Fh is the filter height size, p is the padding value that equals zero, and s is the stride value that equals one. where After constructing the output features of each CNN as shown in Fig. 6, the proposed model concatenates the P Cm arrays into a single P C array, as indicated in Eq. (9). Classification stage As previously indicated, in the third stage, the different relationships between the words in the text are extracted, which can reveal the post-context meaning. These relationships are expressed as composite features, which are computed and concatenated into a P C array. The feature values of the P C array are fed to the fully connected layer in the last stage of the proposed system, as shown in Fig. 7. The fully connected layer is contained by ReLU, equal to the number of features in the P C array. Then, the output of the ReLU is fed to the output unit of the sigmoid function. Equation (10) illustrates how the output of the proposed system that detects the importance of the post is determined, where g is a sigmoid activation function, and b o is the bias. Experimental results and discussion Several experiments were conducted to test and evaluate the proposed system and its stages. The first set of experiments evaluated the structural efficiency of the proposed system. The performance of the overall structure of the proposed system was also evaluated and compared to stateof-the-art algorithms in the second set of experiments. Dataset and evaluation metrics The experiments were conducted using a benchmark dataset of the Stanford MOOC post corpus proposed by Agrawal et al. [7]. It contains posts related to 11 public online classes from Stanford University. The dataset courses are categorized into three domains: Humanities/ Sciences, Medicine, and Education. Each domain contains nearly 10,000 posts, and the total number of posts is 30,002. Agrawal et al. [7] manually classified the data into six dimensions, namely question, opinion, sentiment, urgency, and confusion, on a scale of 1 to 7. Table 2 shows an example of the dataset. Agrawal et al. [7] excluded about 398 posts with malformed or missing scores, reducing the total number of posts to 29,604. For each post, metadata includes up-votes, number of reads, post position, etc. [7]. This study aims to classify the posts into ''urgent posts'' and ''not urgent posts.'' Therefore, the class labels of the corpus were modified to binary classification, with the urgent label approximated to 0 if it is lower than 4 and approximated to 1 if it is greater than or equal [7,8,18,19], and [13]. After the approximation, the urgent Figure 8 shows the number of posts per label in the original labeling. Figure 9 shows the number of posts after approximation. The state-of-the-art algorithms [8,18], and [19] used on the Stanford MOOC corpus dataset divided the posts into three different scenarios: Groups A, B, and C. • Group A: this group simulates the general case in which the training and test datasets were independent of the course or domain. • Group B: the data were split into training and test datasets depending on the course name; all posts related to some courses were selected as the training dataset, whereas the posts related to the other courses were selected as the test dataset. • Group C (a Domain out): the data were split into training and test datasets depending on the domain; Medicine and Education domains for the training and evaluation processes and Humanities domain for the test process. The proposed system performance was evaluated and compared with state-of-the-art algorithms based on recall, precision, accuracy, and F-score metrics. The recall was used to calculate the percentage of correctly classified posts, as indicated in Eq. (11). The precision was used to calculate the ratio of correctly classified posts, as shown in Eq. (12). The accuracy of the proposed system represents the percentage of the total classified posts, as shown in Eq. (13). The F1-score represents the relationship between precision and recall, as shown in Eq. (14). It is a good measure of unbalanced data. where TN, FN, FP, and TP are the true negative, false negative, false positive, and true positive, respectively. Structural efficiency evaluation of proposed system The proposed system consists of four stages. In the first stage, the post is preprocessed so that it can be fed into the subsequent stages, which transform the post-text into a numerical representation P tok array. It uses a pre-trained BERT model to perform tokenization and embedding processes on the input post. The pre-trained BERT model used a fixed input length of n = 512 of the input tokens in the text. In the second stage, the new proposed feature aggregation model of the token's features ''P tok array'' was applied, as shown in Fig. 1. It aimed to uncover data-driven relationships between the token's features and express them as higher-level features. It consists of one convolution layer, with E filters of one size equal to 1 9 768. Each filter combines the 768 local features of a token and expresses them as a global feature. Several experiments were conducted to determine the best values of the number of filters E. The best value of E is 350, as shown in Table 3, which has the highest accuracy of the post-classification compared with the other values of E. The experiments indicate that 350 is the best value of the number of filters representing the number of different weighting values of the aggregated feature ''token's global feature'', which made the model more flexible with the diversity of a word meaning within the posts. To evaluate the effect of the aggregation and weighting of the token's features stage on the proposed system performance, six experiments were developed. As a result, in Figs. 10 and 11, the area under the receiver operating characteristic curve (AUC) is shown. It can be seen in these figures that the feature aggregation stage has a positive effect, and the proposed system based on this stage achieved the highest AUC values on the datasets of groups A, B, and C. In the third stage, the post-context meaning was revealed. It is based on a novel proposed multi-CNN model that extracts the composite features of multiple words. It consists of four CNNs. Each of them contains two layers: convolutional and pooling layers. The first CNN, with a filter size of 1 9 350, is used to extract the best features of a single word that can be expressed within the context meaning of the post. The second CNN, with a filter size of 2 9 350, is used to find the set of features of each of the two words that can be highlighted in the post-context meaning. The third CNN, with a filter size of 3 9 350, is used to extract the expressive features of each of the three words in the context meaning of the post. The fourth CNN, with a filter size of 4 9 350, is used to discover the best features of each of the four words that could reveal the post-context meaning. Several experiments were conducted to evaluate and assess the structural efficiency of the proposed multi-CNN model to reveal the post-context meaning. The proposed multi-CNN model based on extracting the composite features of each word, two words, three words, and four words of the post-text is the efficient structure, as shown in Table 4. It can identify the most effective composite features, resulting in the best post-classification performance. Comparison with state-of-the-art algorithms The performance of the overall structure of the proposed system was evaluated and compared to state-of-the-art algorithms using groups A, B, and C datasets of the Stanford MOOC post corpus. These algorithms focused on additional data preprocessing and representational features, selecting the effective features, and extracting the longterm dependencies between the post words. Almatrafi et al. [8] proposed MOOCs posts classification model. It is based on TF, linguistic and metadata features, and the AdaBoost classification algorithm. It used the TF technique for word representation to convert text terms into numerical representations. TF ignores the word order of the text. The context's meaning relies on the words' order and the relationship between each word and its neighbors, which leads to a defect in understanding the document context. In addition, linguistic features are extracted using Linguistic Inquiry and Word Count (LIWC), a text analysis tool that depends on word count. The LIWC performance is negatively impacted by misspellings, symbols, and expressions, which limits the AdaBoost performance. Guo et al. [7] developed a model based on google-news, metadata features, and architecture of CNN and Bi-GRU layers. It used google-news for word embedding. Google- news created a vector that represents the word's absolute meaning, while ignoring the word's context meaning. The proposed CNN and Bi-GRU layers architecture emphasized the extraction of long-term dependencies between post words. Nabila [8] proposed model based on preprocessing, BERT, and Bi-GRU techniques. Text data preprocessing was used to eliminate stop words and special marks such as ''!''. These stop words and punctuation marks may be helping to better understand the context of the posts, which negatively affects the classification accuracy. Bi-GRU was used to build the classification model depending on the extraction of long-term dependencies between words. Despite these efforts of the state-of-the-art algorithms, the best obtained F1-weighted score of these methods is 91.9% and 81.2% of the class of urgent posts, as shown in Tables 5, 6, and 7. On the other hand, the proposed model takes into account uncovering data-driven relationships between the word features and weighting the extracted dependencies with the diversity of the meaning of a word within the posts. It also considers extracting the various relationships between the sentence words to disclose the post-context meaning. The proposed model achieved 83.6%, 83%, 83.3%, and 92.7% in precision, recall, F1 of urgent, and F1-weighted, respectively, as shown in Table 5. It obtained an improvement rate equivalent to 2.0%, 1.5%, 2.1%, and 0.8% over the best state-of-the-art algorithm results on group A dataset. It also obtained an enhancement of the F1-urgent scores by 0.8% on group B dataset compared with [7] and by 0.7% compared with the state-of-theart algorithm result [8], as shown in Table 6. In addition, the proposed model achieved an improvement on the overall F1-weighted score by 0.3% on group C dataset compared to the state-of-the-art algorithm result [8] and nearly maintains the same performance in the urgent detection. Precision and recall scores reflect the trade-off between quality and variation, and they depend on the intended Bold values are the highest scores Bold values are the highest scores application. In this study, the numerical analysis of the experimental dataset indicates that the urgent posts account for 20% of all posts, yet they should be the focus of the instructor's attention. Therefore, lower precision and a higher recall score mean that the instructor must manually filter many posts. In contrast, higher precision and a lower recall score mean that the system cannot identify a large portion of urgent posts. A balance between precision and recall scores must be maintained for the importance of accuracy and remembrance. The experimental results of the proposed model on different test scenarios demonstrate the efficiency of the proposed model compared to state-ofthe-art algorithms, which achieved a clear improvement in the weighted F1 score, and it achieved a balance between precision and recall scores of the urgent posts, as shown in Tables 5, 6 and 7. Conclusion In this paper, an accurate MOOC post-classifier is proposed to increase the interactivity between instructors and learners. It consists of four stages. In the first stage, the post-text was tokenized using a pre-trained BERT pretrained. In the second stage, a novel features aggregation model was proposed to uncover data-driven relationships between the token features and express them as higherlevel features. In the third stage, a novel multi-CNN model was constructed to reveal the context meaning of the posttext. In the last stage, the extracted features are utilized to classify the post-text. The proposed system performance was evaluated and compared with state-of-the-art algorithms using a benchmark dataset of the Stanford MOOC posts corpus. The experimental results showed the efficiency of the proposed model compared with the other algorithms, which achieved a clear improvement in the weighted F1 score, and it achieved a balance between the precision and recall scores of the urgent posts. Extending the research to improve the model's framework could be one aspect of future work. Sequential deep learning neural networks can be used to extract the semantic features that reflect the long dependencies between words, such as Bi-GRU and Bi-LSTM. Due to the data imbalance (urgent posts make up about 20% of all posts), there is a large discrepancy between the F1 value for urgent and non-urgent posts. Another direction for future research can be envisaged to address the imbalanced data problems using appropriate techniques such as oversampling and data augmentation. Future research could also mitigate the reasons for urgent posts, and whether they relate to the content, logistical, or technical, we suggest using topic modeling techniques to determine the origin of trending posts and extract relevant topics. Analyzing the learner's behavior and opinions to explore the hidden factors that cause dropout can also be addressed. Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). Data availability The datasets analyzed during the current study are available in the Stanford repository, http://datastage.stanford.edu/ StanfordMoocPosts/. Conflict of interest The authors have no affiliation with any organization with a direct or indirect financial interest in the subject matter discussed in the manuscript. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.	2023-06-03T15:09:57.022Z	2023-06-01T00:00:00.000	{ "year": 2023, "sha1": "6cc685cb1be686f5d883aa40f42a5da796e429ef", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00521-023-08673-z.pdf", "oa_status": "HYBRID", "pdf_src": "Springer", "pdf_hash": "bd85ce367ca4f7481063215c9b7d728afb26ec56", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [] }
259182453	pes2o/s2orc	v3-fos-license	Expression and Clinical Significance of TIGIT in Primary Breast Cancer Purpose The roles of T cell immunoreceptor with Ig and ITIM domains (TIGIT) in the diagnosis of primary breast cancer (PBC) are still unclear. This study was designed to investigate the expression of TIGIT in PBC patients, with an aim to analyze its diagnostic value in PBC. Patients and Methods We first explore the expression of TIGIT in cancer patients based on TCGA database, and then we analyzed its correlation with clinicopathological features. Afterwards, we compared the protein and mRNA expressions of TIGIT in two BC cell lines (MCF-7 and MDA-MB-231) and normal breast epithelial cell line (MCF-10A). Subsequently, 56 PBC female patients admitted to the Taizhou People’s Hospital from October 2018 to June 2021 were included in this study. Flow cytometry was used to detect TIGIT level on peripheral blood CD3+ T cells of PBC patients and healthy controls. TIGIT expression in PBC tissues was detected by immunohistochemistry (IHC) and immunofluorescence staining. Results TCGA database showed that compared with adjacent tissues, TIGIT was significantly upregulated in tumor tissues. High TIGIT expression was positively correlated with tumor stage and negatively correlated with recurrence free survival (RFS) and overall survival (OS). TIGIT level in BC cell lines, peripheral blood and tumor tissues of PBC patients was significantly higher than that of control (P < 0.05). TIGIT level was correlated with age (P < 0.05), rather than tumor size, pathological type, lymph node metastasis, ER, PR, HER-2, and P53. ROC curve showed that the optimal critical value of peripheral blood TIGIT for BC screening was 23.38%. Postoperative TIGIT level in peripheral blood was significantly decreased compared to the preoperative TIGIT level (P < 0.05). Conclusion TIGIT was upregulated in PBC and was correlated with age. It may be a potential target for the diagnosis and immunotherapy of PBC. Introduction According to GLOBOCAN estimates of cancer incidence and mortality produced by IARC, female breast cancer (BC) has surpassed lung cancer as the most diagnosed cancer in 2020 worldwide, with an estimated 2.3 million new cases (11.7%). 1 Current major treatment methods for early BC include surgical resection, adjuvant therapy, radiotherapy, chemotherapy, and hormone therapy. 2 Unfortunately, the therapeutic effects of these methods have been weakened due to early metastasis of tumors, side effects of drugs and drug resistance of tumors. BC is a highly heterogeneous disease at the molecular level. 3 Molecular targeted therapy is a method that specifically kills tumor cells without damaging normal tissues by delivering drugs that can bind to specific cancer-causing sites at the molecular level. 4 It is expected to be a more effective and less toxic therapeutic strategy for BC. Tumor cells usually escape from immune surveillance by activating various immunosuppressive pathways, including the activation of inhibitory receptors on tumor-infiltrating T cells. 5 Under normal circumstances, immune checkpoint molecules mainly maintain their own immune tolerance and protect normal tissues from the damage of the immune system. 6 However, lymphocytes will abnormally express co-inhibitory molecules due to long-term exposure to tumor antigens, which will eventually lead to the exhaustion of T cells in the tumor microenvironment. 7 One of the important manifestations of T cell exhaustion is high expression of immune checkpoint molecules such as programmed cell death protein 1 (PD-1), cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) and T cell immunoglobulin-3 (Tim-3). 8,9 Nowadays, monoclonal antibodies (mAbs) blocking PD-1, PD-L1, and CTLA-4 have been approved for multiple cancer indications, [10][11][12][13][14][15] however, only a minority of patients benefit from them. Therefore, increasing attention has been paid to the recognition of new inhibitory receptors. T cell immunoreceptor with Ig and ITIM domains (TIGIT), also known as VSig9, Vstm3 or WUCAM, is an inhibitory receptor expressed on the surface of T cells and NK cells, which is overexpressed in the presence of naive CD4 + T cell activation. 16 TIGIT shows a high affinity to CD155 rather than CD112. In a previous study, the expression of TIGIT on NK cells was highly correlated with its effectiveness in suppressing cytotoxicity. 17 It has also been shown that TIGIT could negatively regulate the anti-tumor response as its deficiency would lead to significant suppression of tumor growth. 18 In addition, increased TIGIT was correlated with parameters of HIV progression, and ex-vivo single or combinational antibody blockade of TIGIT and/or PD-L1 restored viral-specific CD8 + T cell effector responses. 19 Besides, TIGIT has been reported to be highly expressed in various tumor diseases such as ovarian cancer, liver cancer, and BC, which may be involved in the process of anti-tumor immune escape. 20,21 However, little is known on the diagnostic values of TIGIT in BC patients. This study was designed to investigate the expression of TIGIT in BC tumor cells, as well as tissues and T cells of peripheral blood in BC patients, which may provide helpful information for the diagnosis and immunotherapy of BC. TCGA Data Collection The expression of TIGIT in 33 types of pan-cancer samples and 1376 invasive BC samples (291 adjacent normal tissues and 1085 tumor tissues) was obtained from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.com). We divided the samples into high-and low-expression groups using the median of TIGIT expression in the TCGA-BC dataset and performed differential analysis between the two groups to determine the expression cutoff. The data were used to analyze the expression of TIGIT in various cancers and the relationship between TIGIT expression and overall survival (OS) and recurrence-free survival (RFS) in patients with invasive BC. Cell Culture Two human BC cell lines (ie MCF-7 and MDA-MB-231) and one human normal breast epithelial cell line (MCF-10A) were cultured in RPMI 1640 medium containing 10% fetal bovine serum (Gibco). All the cells were cultured in a humidified 37°C incubator supplemented with 5% CO 2 and collected or passaged when their growth reached about 80% confluency. All cell lines were purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China). Western Blot Total protein was extracted from the cell lines using RIPA lysis buffer (Beyotime, Shanghai, China). The protein content was evaluated by BCA method using a commercial kit purchased from Bioteke (Beijing, China). Protein (20 μg) was separated on 10% SDS-PAGE gels, and then was transferred onto PVDF membranes (Millipore, Bedford, USA) and blocked with 5% skimmed milk under room temperature. Subsequently, rabbit monoclonal primary antibodies against TIGIT (ab243903; 1:4000; Abcam) and against α-tubulin (66,031; 1:10,000; Proteintech, China) were added and incubated overnight at 4°C. PVDF membrane was washed four times with TBST for 5 min each time and then incubated with goat anti-mouse or anti-rabbit IgG horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h at room temperature. After washing PVDF membrane for four times, immunoreactive bands were displayed by enhanced chemiluminescence (ECL), photographed in grayscale and quantified using the Image J program (NIH, Bethesda, MD, USA). The α-tubulin was used as an internal reference to normalize bands. The ratio of IOD TIGIT/IOD α-tubulin indicated the relative expression of TIGIT protein. The qRT-PCR was carried out with reference to the previous method. 22 Briefly, TRIzol reagent (Takara, Dalian, China) was used to isolate total RNA from the cells. Reverse transcription of RNA into first-strand cDNA was performed using Rever Tra Ace ® qPCR RT Master Mix (FSQ-201; TOYOBO, Japan) according to the manufacturer's instructions. Then TIGIT mRNA level was measured by Power SYBR Green PCR Master Mix (Thermo Fisher Scientific, Woolston Warrington, UK) using the Light Cycler 480 II system (Roche diagnostics, Mannheim, Germany), with β-actin as an internal control. The TIGIT sequences were as follows: forward, 5'-TCTGCATCTATCACACCTACCC-3'; reverse, 5'-CCACCACGATGACTGCTGT-3'. The PCR reaction was initiated with a 10-min denaturation at 95°C. Amplification was carried out for 40 cycles of 30 s at 95°C, 30 s at 60°C and 30 s at 72°C. The 2 −ΔΔCt method was used to analyze the relative expression of TIGIT gene. Patients We included 56 females with pathologically diagnosed primary breast cancer (PBC) admitted to the Taizhou People's Hospital (Taizhou, China) from October 2018 to June 2021. The diagnostic criteria were in line with the guideline for BC (2017 edition). The inclusion criteria were as follows: patients with pathologically diagnosed PBC underwent surgical resection for the first time, and had not previously received antitumor therapy. Patients with chronic infections, infectious diseases, immune system diseases, or a history of cancer or receiving any antitumor therapy were excluded. Thirty-four healthy female individuals underwent regular physical examination with normal blood routine, lymphocytes, and tumor markers after flow cytometry, served as a control group. The study protocols were approved by the Medical Ethics Committee of Taizhou People's Hospital (No. KY201804801). Written informed consent was obtained from all subjects before the study. IHC TIGIT expression in tumor tissue and adjacent normal tissue was detected by IHC. Briefly, tissue sections were deparaffinized, hydrated, and antigen retrieved with a citrate buffer (pH 6.0). After washing with PBS buffer and blocking with 10% goat serum, 70 μL anti-TIGIT primary antibody (ab243903; 1:200; Abcam) was added, and the mixture was incubated at 37°C for 1 h. Then, a secondary antibody included in the Dako K5007 kit was added. The mixture was incubated at room temperature for 30 min and rinsed with PBS buffer. Afterwards, the sections were incubated with 3.3'-diaminobenzidine (DAB), counterstained with hematoxylin, rinsed with running water and differentiated with 0.5% hydrochloric acid alcohol for 3-5 sec. Then, the sections were washed again with running water for 15 min, dehydrated, and mounted, and finally, were observed under an inverted optical microscope (Olympus, XDS-1A). Immunofluorescent Staining Immunofluorescent staining was performed to further determine the location of TIGIT expression in BC tissues. Briefly, tissue sections were deparaffinized, hydrated, and antigen retrieved with a citrate buffer (pH 6.0). After washing with PBS buffer and blocking with 10% goat serum, anti-TIGIT primary antibody (E5Y1W#99567; CST) was added, and the mixture was incubated overnight at 4°C. Then, the TIGIT fluorescent secondary antibody (Thermo Fisher Scientific, Waltham, MA) was added to give a red reaction product. Afterwards, the mixture was incubated with mouse monoclonal primary antibodies against CD3 (17A2#24,265; CST) and CD56 (123C3#3576; CST) overnight at 4°C. Then, the CD3 (A11008) or CD56 (A11012) fluorescent secondary antibody (Thermo Fisher Scientific, Waltham, MA) was added to give a green reaction product. After washing with PBS buffer, the nuclei were counterstained with DAPI for 10 min in a dark environment. Fluorescence images were visualized with a fluorescence microscope (Leica Microsystems, Wetzlar, Germany). Flow Cytometry EDTA-K3 anticoagulated peripheral blood (5 mL) was collected. Briefly, 5 μL anti-CD3-FITC mAb (BD company), 2 μL anti-TIGIT-PE mAb (BioLegend company), and 50 μL EDTA-K3 anticoagulated blood were added to the bottom of flow tube. The mixture was shaken vigorously and left in the dark conditions for 15 min. Then, 500 μL hemolysin was added, and the mixture was vortexed and left in the dark for 10 min. Upon centrifugation at 1500 rpm for 5 min, the supernatant was discarded, followed by adding 500 μL normal saline. TIGIT expression on the T cell surface was detected using FACS Calibur flow cytometer (BD Biosciences). Lymphocyte populations were identified by forward and side scatter, gated on CD3-positive cells. Isotype control (negative control) was used to define the background staining. FlowJo software (Tree Star Inc.) was utilized to analyze the generated data. Statistical Analysis Normally distributed measurement data were expressed as mean ± standard deviation, and Student's t-test was used for comparison between two groups. The measurement data with skewed distribution were expressed as median (interquartile range), and nonparametric test (U-test) was used for comparison. Qualitative data were expressed in frequency, and a chi-square test was used for comparison. Receiver operating characteristic (ROC) curves were used to determine the optimal cut-off value for BC diagnosis. A p-value of <0.05 was statistically significant. TIGIT Was Upregulated in Invasive BC and Associated with Prognosis Based on TCGA Database TCGA database showed that TIGIT was significantly upregulated in 33 types of tumor tissues than those in normal tissues (P < 0.05, Figure 1A). The expression of TIGIT in BC tissues was significantly higher than that in adjacent normal tissues ( Figure 1B). TIGIT level in patients with stage II and stage III BC was significantly higher than those in patients with stage I BC ( Figure 1C). Kaplan-Meier survival plots revealed that patients with high expression of TIGIT exhibited better recurrence-free survival (RFS) and overall survival (OS) (P < 0.05, Figure 1D and Figure 1). TIGIT Was Upregulated in BC Cell Lines and Tumor Tissues Compared with the normal MCF-10A cells, the protein and mRNA levels of TIGIT were significantly upregulated in the MCF-7 and MDA-MB-231 BC cells (P < 0.05, Figure 2). TIGIT expressions in tumor tissues and adjacent tissues are shown in Figure 3 and Table 1. Compared with the adjacent tissues, the positive expression rate of TIGIT in 56 tumor tissues was significantly increased (67.86% vs 5.88%, P < 0.0001). In addition, the results of immunofluorescence staining showed that TIGIT expression was mainly localized in cancer cells rather than tumor infiltrating cells (Figure 4). Up-Regulated TIGIT in Peripheral Blood of PBC Patients Exhibited Diagnostic Value TIGIT level on peripheral blood T cell surface of PBC patients was significantly higher than that of healthy controls (27.56% vs 19.87%, u=318, P < 0.0001, Figures 5A-D and 6A). The ROC curve was drawn based on the percentage of TIGIT in peripheral blood T cells of PBC patients, and the area under the ROC curve was calculated as 0.833 (95% CI, 0.752-0.914, Figure 6B). The critical value for both optimum sensitivity (94.12%) and optimum specificity (69.64%) was 23.38%. This indicated that TIGIT in peripheral blood could distinguish healthy individuals from BC patients. Taken together, our data indicated that TIGIT showed diagnostic value for PBC. TIGIT on Peripheral Blood T Cells Was Downregulated in Postoperative PBC Patients The changes of TIGIT level on peripheral blood T cells in 12 PBC patients before and after operation were analyzed. We found that compared with preoperative patients, TIGIT level on peripheral blood T cells was decreased significantly in postoperative patients (36.96±11.07% vs 31.83±10.54%, P < 0.05, Figures 7 and 8). Correlation Between TIGIT Expression on Peripheral Blood T Cells and Clinicopathological Features We analyzed the correlation of TIGIT expression level on peripheral blood T cells based on age, tumor size, pathological type, lymph node metastasis, estrogen receptor (ER), progesterone receptor (PR), HER-2, P53 and Ki-67 ( Table 2). The results showed that TIGIT level in PBC patients aged >58 years was significantly higher than that in those aged ≤58 yrs (P = 0.0067). No significant correlation was observed between TIGIT level and other clinicopathological features. In this study, the correlation between TIGIT level and Ki-67 level was not shown to be attributed to the fact that the sample size with Ki-67 ≤14% (14% was used as a cut-off value) was too small (only 3) to perform a systematic analysis. Discussion BC is the most common cancer in females worldwide with complex etiologies, involving genetic inheritance, lifestyle and environmental factors. 23 Immune checkpoint molecules are a class of glycoproteins that exist on the cellular surface. The activated T cells can constitutively express positive costimulatory signaling molecules or negative coinhibitory signaling molecules. The activation, proliferation and differentiation of T cells can be regulated by promoting or inhibiting the activation signals of T cells. TIGIT, a coinhibitory immune checkpoint molecule with immunosuppressive effects, was discovered in 2008 by researchers from Genentech Inc. 24 It could bind to CD155/CD112 ligand and inhibit the activation and proliferation of T cells. Currently, it is generally believed that TIGIT is involved in the negative regulation of human immunity and is closely related to immune escape and carcinomas. Li et al 25 showed that TIGIT expression on the T cellular surface in peripheral blood was upregulated in various cancers such as lung squamous cell carcinoma and colon cancer. Using flow cytometry, He et al 26 found that the expression of TIGIT on peripheral blood T cells in 24 gastric cancer patients was significantly increased compared with 16 healthy controls. Tang et al 27 analyzed the expression of TIGIT on peripheral blood T cells in 23 patients with locally advanced gastric cancer and 8 healthy controls, and their data showed that in addition to the significant increase in TIGIT expression on T cells in gastric cancer patients compared to healthy controls, TIGIT level was significantly reduced in patients who underwent D2 gastrectomy. Therefore, we speculated that TIGIT may play an important role in the immune escape during the pathogenesis of BC. 2411 In this study, we compared the protein and mRNA expressions of TIGIT in two BC cell lines and normal breast epithelial cell lines. Besides, we detected the expression of TIGIT on the T cellular surface using flow cytometry, and TIGIT expression in tumor tissue and adjacent normal tissue was detected by IHC and immunofluorescent staining. Compared with the normal MCF-10A cells, the protein and mRNA levels of TIGIT were significantly upregulated in the MCF-7 and MDA-MB-231 BC cells. TIGIT on T cells in PBC patients was upregulated compared to that in normal controls, and the difference was statistically significant (P < 0.05). This may indicate depletion and dysfunction of T cells in the peripheral immune system of PBC patients, resulting in the inability of tumor cells to be effectively eliminated by the human immune system. In addition, compared with preoperative patients, TIGIT level on peripheral blood T cells decreased significantly in postoperative patients (P < 0.05). This suggested that TIGIT molecules may be involved in the pathogenesis and progression of BC. On the contrary, TIGIT was a safeguard molecule for the improvement of liver regeneration via negatively regulating NK-hepatocyte crosstalk. 28 Database analysis also showed that TIGIT was 2412 associated with better prognosis in BC. 29 These were consistent with the TCGA database results in this study showing that high expression of TIGIT was associated with a better prognosis in BC. Taken together, the role of TIGIT in human malignancies remains controversial. It may cooperate with different immune molecules to play completely opposite roles in tumor progression. Further studies are necessary for clarifying the function and underlying mechanism of TIGIT in BC. In a previous study, Zhang et al 30 found that age was one of the reasons for the different expression of TIGIT in tumors in their study on adult type B acute lymphoblastic leukemia; however, their study did not involve different age groups. Our study demonstrated that TIGIT expression in peripheral blood T cells of PBC patients was significantly correlated with age. Specifically, the expression of TIGIT on T cells in patients aged >58 (median age) was significantly higher than that in patients aged ≤58 (P < 0.05). This implied that TIGIT expression was positively correlated with age. Previous studies have shown that advanced age was a risk factor for thyroid cancer recurrence, 31 and age was also a major risk factor for BC. 32 Taken together, we suggested that TIGIT showed the potency to promote T-cellular depletion and apoptosis in the BC microenvironment with increasing age. In this study, the area under the ROC curve of TIGIT for PBC diagnosis was 0.833 (95% CI, 0.752-0.914), indicating moderate clinical diagnostic value of TIGIT. Besides, the critical value for both optimum sensitivity (94.12%) and optimum specificity (69.64%) was 23.38%, which effectively indicated the high sensitivity of peripheral blood TIGIT for PBC diagnosis. In the future, more studies are needed to confirm whether TIGIT can be used for early BC diagnosis, and whether it can improve the sensitivity and accuracy of diagnosis together with tumor marker CA153 that is considered to be the gold standard for diagnosis. Recently, immunotherapy has been considered to play a pivotal role in treating malignancies, and immune checkpoint inhibitors have become one of the most effective methods for the treatment of malignant tumors. Using multiple immunohistochemical staining techniques, Li et al 33 found that the expression of TIGIT was significantly different in 40 tumor tissues of Hodgkin's lymphoma patients and in 2 normal adjacent tissues (P < 0.05). Tang et al 27 reported that TIGIT level in 441 tumor tissues of patients with gastric cancer was significantly higher than that in adjacent tissues with IHC (P < 0.001). Besides, TIGIT level was significantly correlated with tumor size, histological type, clinical stage, and lymph node metastasis (P < 0.001). Similarly, the IHC results in this study showed that the positive expression rate of TIGIT in tumor tissues of PBC patients was significantly higher than that in adjacent tissues (P < 0.05). The mechanism by which TIGIT enhances immunotherapy sensitivity remains unclear. In a previous study, TIGIT exerted inhibitory functions in multiple processes of various tumor immune cycles. 34 It could inhibit the function of NK cells and inhibit the apoptosis of tumor cells. TIGIT on T cells could inhibit the co-stimulatory ability of dendritic cells, increase the generation of anti-inflammatory cytokines such as IL-10 and induce PVR signals in tumor cells. Besides, myeloid cells stimulated by TIGIT + Treg or PVR could suppress CD8 + T cell function and affect CD4 + T cell activation. Moreover, TIGIT could even directly suppress the roles of CD8 + T cells and protect cancer cells from elimination. Additionally, TIGIT was upregulated in activated T lymphocytes and showed a similar function to PD-1, which may suppress excessive immune responses. 35,36 TIGIT could also negatively regulate the immune activity of T cells via downregulating the level of T cell receptors. 35,37 Its blockade or deletion could promote NK cell-mediated anti-tumor responses and reduce the metastatic ability of tumor cells. 38,39 These findings support ongoing clinical trial of PD-1/TIGIT-dual blockades in tumor progression, which may restore the process of tumor suppression. 40,41 Overall, TIGIT showed a negative regulatory effect on tumor immunity, and immunotherapy targeting TIGIT will be a promising therapy for human malignancies. Conclusion In summary, TIGIT was significantly upregulated in BC tissues and peripheral blood T cells of BC patients. This suggested that TIGIT molecule may be involved in the regulation of immune response in tumor microenvironment, which may become a new target of immunotherapy for BC. Data Sharing Statement The data that support our findings of this study are available from the corresponding author upon reasonable request. Ethics Approval and Informed Consent This study was performed according to the convention of the Declaration of Helsinki. The study protocols were approved by the Medical Ethics Committee of Taizhou People's Hospital (No. KY201804801). Written informed consent was obtained from all subjects before the study.	2023-06-18T05:14:30.537Z	2023-06-01T00:00:00.000	{ "year": 2023, "sha1": "c1102917a7e67e1846b2e5d87c4f0999bf014bde", "oa_license": "CCBYNC", "oa_url": "https://doi.org/10.2147/ijgm.s407725", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "c1102917a7e67e1846b2e5d87c4f0999bf014bde", "s2fieldsofstudy": [ "Medicine", "Biology" ], "extfieldsofstudy": [] }
62930823	pes2o/s2orc	v3-fos-license	Realization of a Desktop Flight Simulation System for Motion-Cueing Studies Parallel robotic mechanisms are generally used in flight simulators with a motion-cueing algorithm to create an unlimited motion feeling of a simulated medium in a bounded workspace of the simulator. A major problem in flight simulators is that the simulation has an unbounded space and the manipulator has a limited one. Using a washout filter in the motion-cueing algorithm overcomes this. In this study, a low-cost six degrees of freedom (DoF) desktop parallel manipulator is used to test a classical motion-cueing algorithm; the algorithm's functionality is confirmed with a Simulink real-time environment. Translational accelerations and angular velocities of the simulated medium obtained from FlightGear flight simulation software are processed through a generated washout filter algorithm and the simulated medium's motion information is transmitted to the desktop parallel robotic mechanism as a set point for each leg. The major issues of this paper are designing a desktop simulation system, controlling the parallel manipulator, communicating between the flight simulation and the platform, designing a motion-cueing algorithm and determining the parameters of the washout filters. Abstract Parallel robotic mechanisms are generally used in flight simulators with a motion-cueing algorithm to create an unlimited motion feeling of a simulated medium in a bounded workspace of the simulator. A major problem in flight simulators is that the simulation has an unbounded space and the manipulator has a limited one. Using a washout filter in the motion-cueing algorithm overcomes this. In this study, a low-cost six degrees of freedom (DoF) desktop parallel manipulator is used to test a classical motion-cueing algorithm; the algorithm's functionality is confirmed with a Simulink real-time environment. Translational accelerations and angular velocities of the simulated medium obtained from FlightGear flight simulation software are processed through a generated washout filter algorithm and the simulated medium's motion information is transmitted to the desktop parallel robotic mechanism as a set point for each leg. The major issues of this paper are designing a desktop simulation system, controlling the parallel manipulator, communicating between the flight simulation and the platform, designing a motion-cueing algorithm and determining the parameters of the washout filters. Introduction Flight simulators are necessary for studies related to vehicle design and tests, and simulator systems help us understand pilot behaviour that is close to reality. Flight simulators provide an economical and secure way of testing new technologies for implementing in aircraft. Several universities and industrial laboratories are undertaking research for new generation prototypes and vehicle dynamic models. Simulators create a realistic flight feeling using motion information feedback [1], and are completely reliable, realizable and easy to implement. It is proposed to increase this reality by improving the software, hardware and physical abilities of simulators. To get closer to the real flight feeling, it is important to employ a robust motioncueing algorithm [2]. Simulators are generally human in the loop(HITL) processes and include an intensive code algorithm. For instance, in a Level-D simulator there are numerous code arrays, which are expensive software systems. Because of their complex structure and high cost, simulator design is quite a difficult task [3]. In simulator systems, examining which factors cause the differences between real and virtual data is needed. These factors are generally separated into three categories: a. human motion perception (vestibular) systems b. the problems that may occur in visual flight simulation c. audio feedback The questions possible in applications targeting human motion perception are: i. What does human perception depend on in real-time applications? ii. How is it possible to improve human perception by changing the parameters of a flight simulation system? [4] Human motion perception systems are the basis of motion perception studies. Humans sense motion through the vestibular system located in the inner ear, which consists of semi-circular channels sensing angular motion and otoliths sensing translational motion. Mathematical models of the vestibular system were first represented in [5] and simplified transfer function models were presented in [6]. Then simulation studies became possible. The otoliths are the human sensory system for translational accelerations and are stimulated by head position. There are three semi-circular canals for sensing angular acceleration or rotational motion. Corresponding signals are sent to the neural system through nerve fibres by the otolith organs' receptor cells and semi-circular canals. The utricle and saccule are the otolith organs in the ear. The utricle senses horizontal motion and the saccule senses vertical motion. The vestibular system is associated with other organs responsible for sensing motion and it plays a major role in the balance of a human body [7,8]. The otolith organs sense acceleration and the effect of gravity on the tilt of the pilot's head. Since the otoliths, acceleration and tilt are associated with one another, this term is called tilt coordination and it improves the accuracy of any simulation. The tilt effect is performed by passing the aircraft translational acceleration through a low-pass filter. Additional low-pass filters are added into the motion-cueing algorithm in the pitch/surge and roll/sway modes [9] to integrate the tilt coordination into the motioncueing algorithm. Visual scaling results in differences in image perception. The optimal level of visual scaling affects the image perception positively. Changes in visual feedback were studied under different visual scale factors in [10]. Additionally, image depth and slope play a significant role in motion perception. Studies performed by Jamson confirmed that to create an accurate motion perception, the horizontal size of the image needs to cover 120°. In [10], the pilot was asked to move at a certain speed and simulation accuracy was tested. Naturally, visual feedback in a simulation plays a significant role in motion perception. Image workflow, resolution or deflections in frequency adjustment create problems. Initially, it was discovered that the image a pilot senses is based on differences in the optical workflow, which is a dynamic model the brain uses to solve a specific motion and helps the pilot predict absolute velocity, distance and relative spatial range. This image signal is effective for understanding major tasks like velocity of the vehicle, distance from objects and lateral control [4]. Audio feedback is less effective in motion perception. Sound is based on simulation acoustics and the number of sounds supplied. Sound-emitting objects in the virtual environment make it easier to get an accurate space. Research has showed that sound is not effective in velocity control and it has been proved that rotation, lateral and yaw motion play a major role in sound vibration [4]. Inertial cues are very important for simulators. Without them, the pilot has no information about aircraft acceleration or rotation. Unlike real vehicles, motion platforms have limited workspace for rotation and acceleration. Motion-cueing algorithms are called washout filters and can be classified into several categories. These filters produce motion commands to be sent to the motion platform [11]. There are four algorithms in the literature for motion cueing: 1. classical approach optimal algorithm 3. adaptive base algorithm model-predictive approach The classical algorithm was founded in the NASA Ames Research Center by Conrad and Schmidt in 1969 and Conrad, Schmidt and Douviller created the rotating coordinate filter in 1973. Generally, the classical approach consists of a combination of high-and low-pass filters in which the cut-off frequency and damping ratio are initially set. While examining the acceleration effect, integrating the gravitational effect into the acceleration change in the vertical direction is also an important issue. First, translational accelerations and angular velocities are scaled and angular velocities are passed through high pass filters. To return the simulator to its initial position, second-and third-order filters are used. If the effects are scaled and limited, then just the second-order filter is sufficient [12]. Because of the workspace limitations of a simulator, an adaptive algorithm was developed by Parrish, Dieudonne, Bowles and Martin in 1975, which is implemented in the frequency domain to eliminate the limited workspace problem. In this algorithm the coefficients are different [13] and motion perception is considered a tracking problem [14]. The acceleration that is obtained from the simulator should be tracked [15]. By applying an adaptive filter in a classical algorithm, the adaptive classical motion-cueing algorithm was developed. The adaptive gain is obtained by minimizing a cost function [12] and the high-pass filter gain is changed. The purpose of this algorithm is to minimize the cost function. Defusing the wrong cues at the actuator extensions and providing a smooth simulation environment is an advantage of the adaptive algorithm but due to the varying structure of the algorithm, there are sometimes gaps in motion perception. Adaptive motion-cueing structures also provide dynamic stability to the system in case of exposure to any distorting effects like parameter changes [16]. In [17], the developed adaptive algorithm structure was also examined. This structure was implemented as an alternative method to keep the simulator in the limits of the workspace. It has a single DoF motion-limit calculator block, which calculates the kinematics and the central point coordinates of the platform together. Corner frequency, damping ratio, gain, cost and adaptive step size as filter parameters are adjusted to keep the translational motion in the limited workspace. Initial cost values are determined based on experience. After the classical filter design, implementing a trial-and-error method is sufficient to determine the necessary parameters [18]. There are several studies comparing use of classical algorithm on three, six and eight DoF simulation platforms. On three DoF platforms, the translational feedback of the system was quite good. On six DoF platforms, motion was felt more intensively because of its freedom; eight DoF platforms showed similar performance [19]. The results of comparing platforms with different DoF are: • The rotation effect must as small as possible. • If the rotation coordination effect is unavoidable, the pilot's head must be taken as the reference point. • A small washout corner frequency must be chosen. • Filter parameters must be chosen so to decrease the washout signal levels. To handle the phase delay between the visual simulation and the platform, the software must be flexible and cost effective. There are also methods of compensation available (i.e., fuzzy logic systems, adaptive models and nonlinear solving) but most require closed-loop control [20]. The model-predictive algorithm is a novel motion-cueing algorithm approach and is presented in [21]. Model predictive control (MPC) is an advanced control technique where the provided control actions are optimized according to a determined system model as well as set of constraints on state evolution, input and input rates. MPC computes the control action by solving a finite horizon open-loop optimal control problem. Finally, by using the current state of the system for the initial state of the plant, the optimization process gives an optimal control sequence [11]. Using an internal model of the plant in MPC is thought of as a limitation, but it can also be considered an advantage for allowing the controller to understand system dynamics [22]. The MPC algorithm structure was tested in a Loughborough simulator and the results of MPC and the classical algorithm were compared. The Loughborough simulator is an electrically actuated Stewart platform. It has sensory feedback, three computer screens, a wheel torque feedback system and three speakers. The first two screens create the image and the last provides a real-time simulation appli-cation. Actuators of the simulator in the study had 600 mm stroke and ±20° angular motion capabilities. In some cases, motion-cueing algorithms give good results but in others actuators may reach the maximum limits, which may disrupt the system that the adaptive algorithm has nonlinear filters for [23]. To reduce the cost function, the optimal control theory including the human vestibular system model was implemented by Sivan, Ishsalom and Huang in 1982. The optimal control theory is based on a problem of how to control the system as close to the desired value of specified vestibular system dynamics and constraints [24]. The linear motion perception model was suggested by Hosman and Van Der Vaart in 1981. In this algorithm, the Riccati equation is adjusted in real time using Newton Raphson method. It is observed that the optimal motion-cueing algorithm is more effective than the classical and adaptive algorithms [25]. Many types of washout filters are studied in the literature. In [26], the human vestibular system-based senseless manoeuvre optimal washout filter was proposed to decrease pilot sensation errors. The human motion perception threshold values are presented and the platform is similar to an inverted pendulum. The control strategy is separated into two controls: a conventional control such as PID and an advanced control such as MPC [27]. Studies on motion perception are based on linear filtering and optimal control. Recently, studies on the MPC algorithm have been performed and were associated with optimal control. It was observed that the MPC algorithm is more effective than the classical algorithm for similar conditions. In the MPC algorithm, the acceleration obtained from the simulation is converted to the acceleration sensed by the motion perception system. This acceleration value is used as a reference for operating the MPC algorithm. The MPC signals are obtained and transferred to the simulator. In the literature, the MPC algorithm was designed, separating the problem into two. Actuator lengths were used instead of Cartesian quantities to keep the platform inside the maximum limits of the workspace [28]. Pouliot suggested two approaches in looking at motion [29], the comparison of: • simulator and air vehicle in terms of angular velocity and some produced forces; and • angular velocity and jerk. In [29] suggested approach is to compare: • translational acceleration or angular velocity of simulator with vehicle data; or • rotational acceleration or angular velocity with vehicle data. In this study, a desktop flight simulation system is introduced with components including flight simulation software, a desktop-parallel robotic manipulator and a motion-cueing system. Human motion perception is summarized. Inverse/forward kinematics of the system and the real-time solution is explained. The communication problem between the simulation and the platform is addressed and, finally, the results from the hardware are discussed. The hardware used in the simulation and the real-time experiments is summarized in Figure 1. Flight Simulators and Components The parallel robotic system as a simulator has six DoFthree translational and three rotational motions. It was first designed in 1965 by D. Stewart, and mounts top and bottom plates and connects with six parallel legs that move linearly [30]. Motion simulator The system is on a 6x6 parallel desktop manipulator and is driven by linear DC motors (Linmot PS01-23X80 / 0150-1201). Linear motors have a motion capacity of 110 mm; multi-axis motion control of the motors is performed via Linmot E210-VF model motor drivers, Figure 2. Figure 3 illustrates the joystick used and the flight simulation software. Because the internal control algorithms of linear motor drivers are extensively active in velocity mode, the force mode of drivers is selected so the authors can apply external control algorithms effectively. Position data are obtained via internal linear encoders of the motors with a sensitivity of 10 µm. Platform parameters are listed in Table 1. Motion cueing In flight simulators, all linear actuators are driven simultaneously to move the six DoF (x, y, z-roll, pitch, yaw) parallel manipulator to any desired position. Required positions of actuators are recalculated at each sampling time and the position error of the legs is kept to a minimum. The major issue in flight simulators is that the software has infinite space and platform has a limited one. A washout filter used in the motion-cueing algorithm eliminates this issue, which constrains the position data the platform receives and, most importantly, attempts to move the platform to its initial position under the driver's perception of motion level. The motion perception algorithm consisting of high-pass and low-pass filters is applied to reduce the translational acceleration and angular velocity values obtained from the FlightGear flight simulation software via a communication protocol to an acceptable limit of the platform workspace. Translational accelerations are filtered from high-frequency data and noise using a third-order (first-order high-pass washout and second-order low-pass washout) filter and angular velocities are filtered from high-frequency data and noise by a second-order washout filter. As shown in Figure 4, the translational motion channel is responsible for simulating the aircraft translational motion. The channel's input is the translational acceleration of the aircraft in the flight simulation software. The rotational motion channel simulates the aircraft rotations. Its input is the angular velocity of the aircraft. The tilt-coordination channel uses a tilt-coordination algorithm to estimate aircraft inertia along x and y axis. The output of this channel is summed with the rotational motion channel establishing the final rotational cue. Platform velocity is kept lower than the human perception level during the washout process. Thus the pilot does not feel the washout motion presence. The combination of movement creates a real flight feeling. As shown in Figure 5, the classical algorithm has a simple structure, yet it is not adaptive due to its constant parameters. In the algorithm, the first-order high-pass filter is as follows: The second-order high-pass filter is: The first-order low-pass filter is: ζ is the damping ratio of the second-order filter. In this study, the damping ratio is one. If necessary, this factor can be reduced. ω is the filters' corner frequency. In this experimental study, the factor is chosen according to the system performance and platform workspace boundaries. Figure 6 illustrates the tilt effect used to simulate translation acceleration using the rotational motion of the platform while the platform is motionless. Since motion simulators tend to have limited workspace, it is necessary to simulate translational acceleration using the Earth's gravity and rotational motion of the platform when the platform is motionless. In a motion-cueing algorithm, the tilt coordination formulation can be shown as: Tilt-coordination effect , , x y x y a arctan g j In Figure 6, the aircraft accelerates along the x axis and is tilted about y axis. φ x and φ y are the tilt angles of platform. a x, y shows acceleration in the x and y directions of the platform and g is the Earth's gravity. Since the tilt effect is a low-frequency motion, the simulation of translational acceleration can be easily simulated by using the Earth's gravity [31]. However, high-frequency acceleration is simulated by the translational motion of the platform. The axes of the aircraft are represented in Figure 7. Human motion perception The human perception system is the basis for human perception studies. The vestibular system is shown in Figure 8. The dynamics between sensed angular velocities, ω i , and the input angular velocities, ω i , at each semi-circular canal with stimulus accelerations are given below. τ 1 , τ 2 , τ a , τ L are the time constants and G sc is the gain of the dynamic system. In the transfer function between translational acceleration sensed in the ear, â i , and stimulus acceleration, a i , 1/ τ L is Index i shows the rotational motions in Euler angles (roll, pitch, yaw). Human motion perception system parameters are shown in Table 2 and human motion perception threshold parameters are shown in Table 3. A human detects acceleration and angular motion through the inner ear's vestibular system. Because of insensitivity at low speeds, motion perception algorithms simulate the acceleration of the platform and attempt to return the platform to its initial position at these low speeds. When high frequency motion signals move the platform, low frequency motion signals are transformed into angular Kinematic Analysis of Six DoF Parallel Mechanisms As in serial robots, kinematic analysis of parallel mechanisms should be examined in terms of inverse and forward kinematics. Inverse kinematic analysis helps to find the leg lengths, (L 1 , L 2 , L 3 , L 4 , L 5 , L 6 ), given the position and orientation of the end effector when the position and rotation of top plate, (x, y, z, α, β, γ), is known. Forward kinematics is the problem of finding the position and orientation of the top plate with respect to the fixed base when the leg lengths are known. Forward kinematics analysis is important in terms of all the conditions the platform may have in specific leg lengths. The contrast between serial and parallel robots continues here. When a specific trajectory need to be followed, it is necessary to solve the inverse kinematics in real time. The forward kinematics solution of a parallel mechanism is not unique in that there are several possible configurations of leg lengths. In inverse kinematics, the solution is unique, which means one position/orientation of the top plate refers to certain leg lengths. Inverse kinematic analysis Inverse kinematic analysis for parallel mechanisms is simple to solve compared to forward kinematics. Inverse kinematics of a robotic system can be used to plan a specified trajectory that uses leg lengths by providing any specified position of the centre of the platform top plate. The platform is shown in Figure 9. The position of points P i are defined by Equation (7) Forward kinematic analysis There are several approaches on forward kinematics solutions in the literature. In this study, Newton-Raphson method is used and an equation in the form of F(x) = 0 is solved. In this equation, x 0 is considered as initial point and if F(x) has a derivative at x 0 , the tangent of the function at related point can be expressed as follows: Next, x 1 can be found by replacing x in the equation and zero instead of Y. x 1 in Equation (16) will be closer to the solution. To achieve a more absolute solution with the same method, iterations are applied and solutions of x 1 , x 2 , x 3 …x n can be obtained. When the desired solution sensitivity is reached, iterations are terminated. Implementation of forward kinematics For the forward kinematics solution of the Stewart platform mechanism, Newton-Raphson method is applied as follows: R is the rotational transformation matrix and the equation that will calculate leg lengths is defined in Equation (17 Because the function contains more than one variable, the derivative of the function used in Newton-Raphson method can be found by taking the partial derivative for each variable as follows: Then, the unknowns are calculated by the following equation: The initial conditions are: As a result of this iterative approach, a sufficiently accurate result is obtained. When approaching the solution, an ε value of sufficient accuracy should be defined as: When Equation (21) is confirmed, the iterations are terminated and last X value is considered the solution. The flowchart of forward kinematics is in Figure 10. Figure 10. Flowchart of forward kinematics analysis [30] Communication Software To maintain signal transmissions between control/flight computers and the simulator in real time, the system's communication is treated to preserve true motion feeling. Flight data is fed to the control computer to generate the necessary trajectories of the platform based on motioncueing algorithms [32]. Communication and software The mathematical model of the simulated aircraft is hidden in the visual simulation software, FlightGear, and the motion of the aircraft is commanded using an external professional joystick system. These mathematical models produce motion variables for the flight, which are transferred to the control computer. A generic protocol is used on the input/output port of the visual simulation software. The flight parameters produced by the flight simulation software's mathematical models are transmitted via generic protocol as data packages and delivered into the platform control computer employing the TCP/IP protocol. During this process, data to be transmitted are defined by creating an XML script file. Thus, six parameters including translational accelerations of the simulated aircraft at different pilot positions and angular velocities are transmitted to an external port (Table 4). The created XML script is kept in the visual simulation software's setup folder and is run here in real time. The XML script is shown in Figure 11. It is determined that a sampling frequency of 240 Hz is sufficient for visual simulation. Thus, the visual simulation external communication frequency is chosen to be 120 Hz based on the Nyquist criterion. Flight data obtained from the flight simulation software are transmitted to the defined port of motion control computer via a TCP/IP protocol ( Figure 12). A server in an m-File is created to receive incoming data in a Matlab environment. The final part of the protocol is the format variation code structure, which is shown in Figure 13. Control The parallel manipulator is controlled in two stages: 1. Force (current) control of linear actuator drivers 2. Individual leg position PID controllers implemented in a Simulink control model over the Q8 card The overall structure of the control model is represented in Figure 14, in which each leg's control signal is generated and fed separately. The output of the PID controller is an analogue input to the motor drivers as ±10 V, which is considered to be a current set point since the drivers are used in force-control mode. The linear actuator inner control model is shown in Figure 15. This input relationship is shown in Figure 16. Thus, ±10 V refers to ±3A in the driver controller, which results in force control over the linear actuator. Drivers of the linear actuators are set to provide auxiliary encoder output in 10 µ resolution (over the possibilities of 1, 5, 10, 20 and 50 µ), which provides the best performance for the setup used. Since set points regarding FlightGear data is compared with actual encoder readings, the difference between them is considered to be the error supplied to the controller structure. Flight Simulation by FlightGear 3.4 can be run in 5-10 Hz; 5 Hz is employed in this application because of the general acceptance by JAR standards for level D simulators. Because of the nonlinearity and complexity of the platform kinematics, PID parameters are selected based on Table 6. Results Washout filter parameters are chosen as in Table 7 using trial-and-error method in which 18 parameters are present and motion range, limited workspace, obtaining logical motion behaviour and the security of the platform are the considerations. By varying the parameters with small intervals, the motion of the platform has been observed and best results are chosen based on above criteria. Simulation tests are performed on the desktop simulator system. Boeing 777-300ER model aircraft is prepared for departure from a specified airport in FlightGear 3.4 visual simulation software. The acceleration effect on the aircraft during departure is simulated as an amount of motion in the x-axis (forward) direction ( Figure 17). The aircraft moved through x axis with 4 m/s 2 acceleration and the platform moved 0.08 metres through x direction. The platform rotated 8° about the y axis simultaneously In the second case, the aircraft performed roll motion and moved about the x axis through a CCW direction with an alpha angle of -5° (Fig. 19). Then the following motion about the clockwise direction with a 7° alpha angle (Fig. 20) is performed. Results are as shown in Figure 21 and Figure 22. In the third case, the aircraft descended with an angle of 10°a bout the y axis (Fig. 23.) and lifted its nose with 15° about the y axis (Fig. 24.). Results are as shown in Figure 25. In this motion, the platform moved through z axis for perceived g-force of aircraft. The platform moved -0.05 metres when the aircraft dived. Secondly, the platform moved 0.1 metres in a z direction when the aircraft lifted its nose. The results of the simulation on a desktop simulator are considered to be satisfactory and the desktop system is proven to be more economic to use in trials of various washout filters. As a further analysis, the pilot's perception of motion is to be analysed using the provided vestibular model and forward kinematics of the platform. Since the results obtained are based on a classical washout algorithm, further investigation is to be performed using extensive algorithms like adaptive, optimal and model predictive washout structures. Conclusions and Future Work This paper presents a novel and economical desktop flight simulator system. When compared to similar systems, it can be seen that the proposed system is more convenient for motion-cueing studies because it is simple and economic to apply developed motion-cueing algorithms. The FlightGear package programme uses XML as a data protocol to provide communication between FlightGear and MATLAB via TCP/IP protocol so an XML script file is necessary for this system. The development of other motion-cueing methods (optimal, model predictive, adaptive) is also being performed using this system. The main goal is to optimize the algorithm giving the best results to use in a full flight simulator system (FFS, Level D), which is already constructed and in use. Acknowledgements This article is a revised and expanded version of a paper	2019-02-16T14:19:48.068Z	2016-05-01T00:00:00.000	{ "year": 2016, "sha1": "8e010cf1a01a229b6f8e53958fd39399b5b82907", "oa_license": "CCBY", "oa_url": "https://journals.sagepub.com/doi/pdf/10.5772/63239", "oa_status": "GOLD", "pdf_src": "ScienceParsePlus", "pdf_hash": "88ac08fab9f5d5b656530c33439257512c5bf05e", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
4391207	pes2o/s2orc	v3-fos-license	Discovery of novel heart rate-associated loci using the Exome Chip Abstract Resting heart rate is a heritable trait, and an increase in heart rate is associated with increased mortality risk. Genome-wide association study analyses have found loci associated with resting heart rate, at the time of our study these loci explained 0.9% of the variation. This study aims to discover new genetic loci associated with heart rate from Exome Chip meta-analyses. Heart rate was measured from either elecrtrocardiograms or pulse recordings. We meta-analysed heart rate association results from 104 452 European-ancestry individuals from 30 cohorts, genotyped using the Exome Chip. Twenty-four variants were selected for follow-up in an independent dataset (UK Biobank, N = 134 251). Conditional and gene-based testing was undertaken, and variants were investigated with bioinformatics methods. We discovered five novel heart rate loci, and one new independent low-frequency non-synonymous variant in an established heart rate locus (KIAA1755). Lead variants in four of the novel loci are non-synonymous variants in the genes C10orf71, DALDR3, TESK2 and SEC31B. The variant at SEC31B is significantly associated with SEC31B expression in heart and tibial nerve tissue. Further candidate genes were detected from long-range regulatory chromatin interactions in heart tissue (SCD, SLF2 and MAPK8). We observed significant enrichment in DNase I hypersensitive sites in fetal heart and lung. Moreover, enrichment was seen for the first time in human neuronal progenitor cells (derived from embryonic stem cells) and fetal muscle samples by including our novel variants. Our findings advance the knowledge of the genetic architecture of heart rate, and indicate new candidate genes for follow-up functional studies. Introduction Increased resting heart rate (HR) is a known risk factor for cardiovascular morbidity and mortality (1)(2)(3), including stroke (4) and sudden cardiac death (5,6). Heart rate increased by 20 beats per minute (BPM) is associated with 30-50% higher mortality and appears to be independent of confounder factors (7). High HR increases myocardial oxygen consumption yet lessens oxygen delivery to myocardial tissue. It also increases arterial stiffness and risk of plaque rupture (8). Although HR can be influenced by many non-genetic factors (e.g. exercise, smoking and cardiovascular drugs), the heritability of resting HR is estimated to be 26-32% from family studies (9,10), and 55-63% from twin studies (11). Several meta-analyses of genome-wide association studies (GWASs) have been undertaken to detect genetic determinants of HR (12)(13)(14)(15)(16). There were 21 HR loci previously reported at the time of our study by den Hoed et al. (12) in a GWAS analysis of 180 000 individuals, predominantly of European ancestry. The study implicated 20 candidate genes from follow-up functional studies in Danio rerio and Drosophila melanogaster models. Smaller GWAS analyses have also been performed in Icelandic and Norwegian populations (15), African Americans (13) and genetically isolated European populations (16). The variants discovered by GWAS are common, and are mostly in introns or intergenic regions. Together the previous loci from GWAS at the time of our study only explain a small percentage [0.9% of the variability in HR (12,17)]. To increase our knowledge of genetic determinants influencing HR and discover novel loci, especially rare or low frequency coding variants with larger effects, we meta-analysed data from 104 452 individuals of European-ancestry using the Exome Chip, from cohorts that participated in the Cohorts for Heart & Aging Research in Genomic Epidemiology (CHARGE) EKG consortium. The Exome Chip permits a cost-efficient analysis of coding variants derived from sequencing of >12 000 individuals and includes many rare and low-frequency variants (18). We performed a validation experiment using independent replication samples from UK Biobank data, and bioinformatics investigations to gain an understanding of the new HR loci. Single-nucleotide variant analysis in individuals of European-ancestry In the discovery phase, association results of 235 677 single-nucleotide variants (SNVs) from 104 452 individuals were metaanalysed using a fixed-effects model (Supplementary Material, Fig. S1). Two analyses were performed. The first used RR-intervals (RR in milliseconds¼ 60 000/HR, in beats per minute, according to the inverse relationship between HR and RR). The second used the inverse-normalized residuals of the linear regression RR-interval adjusted for age þ sex þ body mass index (BMI) as covariates (denoted as RR-INVN). An overview of the study design is provided in Figure 1. We observed a high correlation of effect sizes and P-values between the RR-interval and RR-INVN meta-analyses (r 2 ¼ 0.99 and 0.98, respectively; Supplementary Material, Fig. S2). Furthermore, the RR-interval was near-normally distributed, so inverse normalization was deemed unnecessary (Supplementary Material, Fig. S3). Beta-blockers are clinically known to lower HR, therefore the phenotype measurements of beta-blocker users may be under-estimated, and hence the inclusion of beta-blocker users in our analysis may potentially bias our analysis results. We therefore performed a sensitivity analysis by also metaanalysing a subgroup of cohorts that provided beta-blocker data (N ¼ 48 347; 17 cohorts). Results including or excluding beta-blocker users were highly correlated (r 2 of the betas ¼ 0.97; r 2 of the P-values ¼ 0.74; Supplementary Material, Fig. S4), suggesting there is little or no bias from including beta-blocker users in the analysis. Therefore we report the meta-analysis results from the full dataset for the RR-interval, to maximize sample size and power. Replication and meta-analysis with the UK Biobank dataset To identify novel associated loci, we selected 12 variants with P < 1 Â 10 À5 that mapped outside the 21 HR loci reported in the previous GWAS (12) for follow-up in an independent dataset. Within each unknown locus, there were no potential secondary SNVs not in linkage disequilibrium (LD) with the lead SNV (r 2 < 0.2) and meeting our look-up significance threshold (P < 1 Â 10 À5 ). Hence only 12 new lead SNVs were carried forward. We also followed-up 12 potential secondary signals at 9 of the 21 previously reported HR loci (further details on selection criteria are provided in the Materials and Methods) (12). None of the selected variants was in LD (r 2 < 0.2) with each other, or with the published SNVs. Thus, a total of 24 variants were taken forward into replication. The UK Biobank dataset provided results for the selected genetic variants (N ¼ 134 251 individuals). Nine of the 12 previously unknown variants were validated based on exome-wide significance (P 2.12 Â 10 À7 ) in the combined meta-analysis of CHARGE and UK Biobank data, and on Bonferroni-adjusted significance (P 0.0042 for 12 tests) in the replication dataset alone, with concordant directions of effects taking into account the inverse relationship between the RRinterval from the discovery data and HR from the replication data (Table 1; Fig. 2). Indeed, all nine SNV associations were genome-wide significant in the combined meta-analysis (P < 5.0 Â 10 À8 ). Four of our nine validated novel loci were reported in a UK Biobank study (17) that was published after completion of our study (Table 1B). Hence, we present results here for five unreported novel loci (Table 1A; Supplementary Material, Figs S5 and S6). Twelve of the 21 HR-associated SNVs from the previously reported GWAS (12) were covered on the Exome Chip, either directly or by a proxy SNV in high LD (r 2 > 0.8). Our discovery metaanalysis showed strong support for the previous findings, with 11 of the 12 SNVs validated at Bonferroni-adjusted significance (P 0.0042 for 12 tests), of which nine were validated at exomewide significance (P < 2 Â 10 À7 ; Fig. 2). Only rs4140885 at the TFPI locus was not supported in our data (P ¼ 0.10; Supplementary Material, Table S1). Independent secondary signals at known loci All 12 potential secondary signals at loci previously reported by den Hoed et al. (12) were genome-wide significant in the combined meta-analysis (Supplementary Material, Table S2) and are independent to the known SNPs according to LD (r 2 < 0.2). We performed a conditional analysis using Genome-wide Complex Traits Analysis (GCTA) to formally identify secondary signals of association. Five of the 12 validated potential secondary SNVs (within CD46, CCDC141, SLC35F1, ACHE and KIAA1755 loci) were selected within the final GCTA model (Supplementary Material, Table S3). At four of the previously reported HR regions the secondary signals that we identified were confirmed to be statistically independent signals of association: CD46 (rs2745967), CCDC141 (rs10497529), SLC35F1 (rs12210810) and KIAA1755 (rs41282820) in addition to the known SNV, as both the published SNV and the new secondary SNV were present in the final GCTA model of jointly independent associated variants. Hence, we identified two distinct signals of association at each of these four known HR loci. However, the published SNV at the ACHE locus (rs13245899) is not covered on the Exome Chip, or by any proxies (Supplementary Material, Table S1), so the GCTA analysis does not include the known variant. As we are not able to condition on the unavailable published SNV and formally test association jointly with the known SNV, we are unable to statistically confirm the total number of independent signals at the ACHE locus. The secondary SNVs at CCDC141, ACHE and KIAA1755 are non-synonymous variants. Furthermore, the SNVs at CCDC141 and KIAA1755 are low-frequency with minor allele frequencies (MAFs) of 3.6 and 1.7%, respectively. Secondary signals have also recently been observed at four of the five loci (CD46, CCDC141, SLC35F1 and ACHE) in UK Biobank data (17), since completion of our meta-analysis. At CD46, our secondary SNV (rs2745967) is in high LD (r 2 ¼ 0.78) with the secondary SNV (rs2745959) reported in UK Biobank, so likely to be the same signal. At CCDC141 our secondary variant is exactly the same SNV as from UK Biobank (rs10497529). Similarly, at SLC35F1, our secondary SNV (rs12210810) is in very high LD (r 2 ¼ 0.98), so is likely to be the same signal. Hence at these three known loci (CD46, CCDC141, SLC35F1), all existing data suggest there are two independent signals of association. At the ACHE locus, our secondary SNV (rs542137; $38 kb and r 2 < 0.2 from the published SNV) is not in LD (r 2 < 0.2) with the secondary SNV from UK Biobank (rs140367586; $659 kb and r 2 < 0.2 from the published SNV). We are unable to clearly determine the number of distinct signals at the ACHE locus from our Exome Chip RR-interval discovery meta-analysis data, without the published SNV being covered on the Exome Chip. The low-frequency non-synonymous variant (rs41282820) at the known KIAA1755 locus is a new, secondary variant, with strong evidence of independent association, it does not overlap with other published findings. Variance explained Twelve of the 21 previously reported HR-associated SNVs (12) covered on the Exome Chip explain 1.14% of RR-interval variance (P ¼ 3.96 Â 10 À10 ) within the 1958 Birth Cohort study (see Materials and Methods). The added contribution of the lead SNVs at our five unreported novel loci, combined with the 12 previously reported SNVs, increases the variance explained to 1.28% overall (P ¼ 9.17 Â 10 À11 ). Comparison of results between European and non-European populations To investigate our data from non-European samples [9358 African Americans (AA), 1411 Hispanic (HIS) and 754 Chinese-Americans (CH); Supplementary Material, Table S4], we first extracted results for the 12 of the 21 previously reported HR-associated SNVs covered on the Exome Chip (12). In contrast to previous results for Europeans, only two known HR-SNVs showed evidence of association (P < 0.05), at the GJA1 and MYH6 loci, in the AA population only. This is likely due to a lack of power from the smaller non-European sample sizes, considering the power was calculated to be only 48, 11.7 and 8.5% for AA, HIS and CH, respectively. Concordance in the direction of effects compared with Europeans was only significant for AA, with 92, 64 and 50% concordance, corresponding to P-values of 2.9 Â 10 À3 , 0.16 and 0.23 from binomial tests for AA, HIS and CH, On the X axis, P-values are expressed as Àlog10(P) are plotted according to physical genomic locations by chromosome. The Y-axis is truncated to Àlog 10 (P) ¼ 20 with any variants with P < 1 Â 10 À20 displayed on the Àlog 10 (P) ¼ 20 line. The nine novel variants validated from the combined meta-analysis with UK Biobank data are represented by squares. Variants in linkage disequilibrium (LD; r 2 > 0.8) with published GWAS variants are highlighted with black circles (12). New secondary variants validated in our analysis are indicated as triangles. Locus names of the novel loci correspond to the nearest annotated gene, with 5p13.3 denoting an intergenic variant. The dashed line indicates a P-value threshold of 1 Â 10 À5 , corresponding to the lookup significance threshold and the continuous line indicates a P-value threshold of 2 Â 10 À7 , corresponding to exome-wide significance. Due to the inverse relationship between R-R interval and HR the opposite beta directions do relate to concordant directions of effect between discovery and replication. SNV, single-nucleotide variant; Chr:Pos, Chromosome:Position based on HG build 19; EA, effect allele; EAF, effect allele frequency from the discovery data; BETA-RR, beta effect estimate of RR-interval (milliseconds) taken from the ExomeRR discovery data; SE, standard error of the effect estimate; N, sample size analysed per variant (provided for genotyped discovery data only, as replication data was imputed so N ¼ maximum N for all variants); BETA-HR, beta effect for heart rate (in beats per minute) taken from the UK Biobank replication data; P, P-value from either the discovery meta-analysis, the replication data, or the combined meta-analysis of discovery and replication data. Locus name indicates the nearest gene to the HR-associated SNV. a Indicates that the lead or a proxy SNV ( Indicates if the lead SNV is predicted to be damaging. Mapping to more than 500 kb from either side of a previously reported HR-associated SNV. A novel locus is a genomic region with no SNVs in LD (r 2 < 0.2) with HR-associated SNVs. respectively. The lack of support of previous findings from the under-powered non-European data led us to restrict our primary discovery meta-analysis to Europeans only. We also performed a look-up of the nine validated SNV associations in the non-European samples. Due to the lack of power, and different allele frequencies compared with Europeans, none of the SNVs had results with P < 0.05 within any ancestry (Supplementary Material, Fig. S6), and there was little concordance in effect directions: 56% and P ¼ 0.246 for AA; 33% and P ¼ 0.164 for HIS and CH. Gene-based tests Gene-based testing was performed to identify genes which may have multiple rare variant associations. None of the gene-based test results was significant, after excluding the single most significant low-frequency variant from the tests (Supplementary Material, Table S5). Look-up of UK Biobank HR-SNVs Since completion of our meta-analysis of Exome Chip genotypes, a genome-wide scan for HR has been completed in UK Biobank (17). This study published 46 new HR loci. Four of these novel loci were simultaneously discovered in our analyses (RNF207, SCN10A, 5p13.3, KDELR3: Table 1B). Among the 42 remaining UK Biobank loci, only five of the lead SNVs were covered on the Exome Chip at r 2 ! 0.8. Results from our exome RR European-ancestry meta-analyses show support for all five of these loci (P < 0.01; Bonferroni-adjusted significance for five tests; Supplementary Material, Table S6). HR loci and association with other traits To provide insights into possible shared aetiologies or mechanisms of disease, we assessed association of our five unreported novel HR-SNVs (and their proxies, r 2 ! 0.8) with other traits. Genome-wide significant phenotype-genotype associations were observed for three novel loci (Supplementary Material, Table S7). The SNV at the DLRD3 locus was associated with age of menarche. The SNV at the JAZF1 locus was highly pleiotropic, as shown by associations with several autoimmune disorders (systemic lupus erythematosus, Crohn's disease and selective immunoglobulin A deficiency), height, type 2 diabetes and JAZF1 transcript levels in adipose tissue. The SNV at the SEC31B locus was associated with plasma palmitoleic acid levels and differential exon expression of SEC31B. Functional annotation of novel HR-SNVs and candidate genes Four of the five unreported novel HR-SNVs or their proxies (r 2 > 0.8) are non-synonymous SNVs in TESK2, DALRD3, C10orf71 and SEC31B (Table 1A). The non-synonymous SNV in SEC31B (rs2295774, c.1096T>G, p.Ser332Ala) is in a conserved region of the protein, and is predicted to be damaging using three different algorithms in ANNOVAR (19). We also investigated whether the novel HR-associated SNVs or their proxies (r 2 > 0.8) were associated with changes in expression levels of nearby genes (i.e. as expression quantitative trait loci, or eQTLs) in the Genotype-Tissue Expression database (GTEx) dataset (20). We observed a significant eQTL association at one novel HR locus (Supplementary Material, Table S8). Specifically, the HR increasing allele of the non-synonymous SNV at SEC31B was associated with increased levels of SEC31B in tibial nerves (P ¼ 8.08 Â 10 À33 ), lung (P ¼ 1.22 Â 10 À23 ), atrial appendage tissue (P ¼ 4.56 Â 10 À11 ) and the left ventricle (P ¼ 4.0 Â 10 À9 ), tissues which may be regarded as physiologically relevant for HR. We also observed HR loci to be significantly enriched for DNase I hypersensitive sites (DHSs; Fig. 3). We evaluated regions containing the five unreported novel HR loci and five independent secondary variants at previously reported HR loci (12) together with all 67 published HR-associated SNVs [21 loci reported from the original GWAS (12) plus 46 loci recently published from UK Biobank (17)]. Highest enrichment for DHSs in HR loci occurred within regions that are transcriptionally active in fetal heart tissue and fetal lung, as reported in the UK Biobank study. Moreover, for the first time we found significant enrichment for DHSs in human neuronal progenitor cells (derived from embryonic stem cells) and fetal muscle samples, with the inclusion of our novel loci. Pathway analyses We used Ingenuity pathway analyses to determine whether there was any increased enrichment in HR-associated pathways with the contribution of our five newly identified loci. We identified 16 significantly enriched pathways at P < 1 Â 10 À4 . Most of these pathways are related to the cardiovascular system and involve, for example, supraventricular arrhythmias, dilated cardiomyopathy and HR (Supplementary Material, Table S9). Coding variants at HR loci The Exome Chip provides a unique opportunity to search for coding variants within known HR loci. Although GWAS analyses typically identify intron or intergenic variants, Exome Chip analysis may identify HR-associated coding variants, which would point to candidate causal genes. We considered all 67 published HR loci [21 previously reported GWAS loci (12) plus 46 recently published loci from UK Biobank (17)] and extracted all SNVs in high LD with the lead variants (r 2 ! 0.8), tagging the same association signal, restricted to variants covered on the Exome Chip. We further filtered variants to obtain SNVs that reached exomewide significance for associations with RR-interval in our primary discovery meta-analysis, to ensure that variants have a highly significant association with the trait. Coding SNVs were identified, using the CHARGE Exome Chip annotation file. We only observed two such coding variants in two reported loci: CCDC141 and KIAA1755. The published CCDC141 coding variant was previously annotated as being non-synonymous (12), and is predicted to be damaging in our annotation (rs17362588; p.Arg935Trp). The coding SNV at KIAA1755 is the best proxy (r 2 $ 1) for the published non-synonymous SNV (rs6127471) covered on the Exome Chip (Supplementary Material, Table S1). The original GWAS (12) had reported this signal as non-synonymous. Therefore, our Exome Chip analyses do not reveal any new evidence of likely causal coding variants at well-established HR loci. Regulatory variants at HR loci Our analyses of coding variants at all known HR loci indicated that the majority of HR-associated SNVs and the variants in high LD with them are non-coding. We thus investigated which variants could have a causal effect through regulatory chromatin interactions, such as promoter-enhancer contacts. We considered all 67 published HR loci [21 previously reported GWAS loci (12) plus 46 recently published loci from UK Biobank (17)], and the five novel loci reported here. We found variants that potentially affect enhancer function using RegulomeDB (21) and found genes whose promoter regions form significant chromatin interaction with them from right ventricle Hi-C data (22). We found 64 potential target genes in 49 HR loci (4 new loci, 18 loci from the GWAS and 27 loci from the UK Biobank study; Supplementary Material, Table S10). Including these long-range interactors in the candidate causal genes list increased the significance of enrichment for many HR-related terms, such as arrhythmia and cardiac fibrillation in our IngenuityV R Pathway Analysis (IPAV R ; Supplementary Material, Table S11). For newly identified loci, the TESK2 promoter had a longrange interaction with the SNVs with highest regulatory potential in the locus, underlining it as a candidate. LOC441204, a gene of unknown function was found to interact with the JAZF1 locus. At the SEC31B locus, there were interactions with two genes, SCD and SLF2. At the C10orf71 locus, MAPK8 showed the most significant interaction. In the 21 loci from the previously published GWAS (12), we identified significant chromatin contacts for the regulatory SNVs of 18 loci. We found CALCRL, TTN, HTR2B, PLD1 and CHRM2 as strongest interactors at the TFPI, CCDC141, B3GNT7, FNDC3B and CHRM2 loci, respectively, out of these only CALCRL is in LD (r 2 > 0.8) with the lead SNV. The previous study (12) functionally tested 31 candidate genes, they found 20 of them to have an HR phenotype in either Drosophila melanogaster or Danio rerio experiments. All five of the strongest interactor genes were amongst the 20 genes with an HR phenotype. Finally, we found 41 potential causal genes that have not been implicated by previous GWASs. A few of these genes have a cardiac function, including RAPGEF4 (18) and PIM1 (23), whereas some are involved in neuronal development and function, e.g. PBX3, NRNX3. These candidates open up new avenues that may aid our understanding of HR biology. Discussion Our meta-analysis of Exome Chip genotypes yielded five unreported novel HR loci, and one unreported independent new secondary signal, which was a low-frequency non-synonymous SNV at the previously reported KIAA1755 locus. Our data strongly supported the association of SNVs at 11 of the 12 previously reported GWAS loci that were covered on the Exome Chip. All lead SNVs at all validated novel loci are common (MAF ! 5%) and have similar effect sizes, which are smaller than the effect sizes for the majority of previously reported SNVs (Supplementary Material, Fig. S7). Our study did not yield any rare SNV associations with HR, indicating that much larger sample sizes will be required in future studies to have sufficient power to detect effects of any rare variants and assess their contributions to HR heritability. The same observation of the need of larger sample sizes applies to the analysis of HR loci identified within Europeans in other ancestries, where the lack of significance and concordance in the results from non-European populations is most likely due to a lack of power, as well as differences in the allele frequencies and LD patterns between Europeans and non-Europeans. As the non-European samples were much smaller, we did not perform a comprehensive comparison across populations or a robust trans-ethnic meta-analysis. Annotation of novel HR-SNVs or their close proxies, eQTL analyses and long-range chromatin interactions in heart tissue reveal new potential causal candidate HR genes (Supplementary Material, Tables S10 and S12). At the SEC31B locus there is a predicted damaging non-synonymous variant in SEC31B, and SNVs at this locus are also significantly associated with SEC31B expression levels. Although its precise function is unknown, the SEC31B gene encodes SEC31 homolog B, a COPII coat complex component. SEC31B has been proposed to function in vesicle budding, and cargo export from the endoplasmic reticulum (24). The gene is ubiquitously expressed at low levels, but there are higher levels of expression in the cerebellum. There are 13 transcripts, and thus several predicted SEC31B proteins. The major isoform is 129 kDa, but the HR-associated nonsynonymous SNV maps to all SEC31B transcripts. There are no existing mouse models, and the predicted protein does not directly interact with other proteins or pathways currently recognized as being important to HR. Chromatin interactions in heart tissue indicate SCD and SLF2 as two other candidate genes for consideration at this locus. SCD encodes a stearoyl-CoA desaturase, which has a role in myocardial dysfunction (25) and SLF2 encodes the SMC5-SMC6 complex localization factor 2. TESK2, C10orf71 and DALRD3 can be considered as candidates for further analyses, based on the lead SNVs being nonsynonymous variants in each gene. TESK2 encodes a serine/ threonine protein kinase with an N-terminal protein kinase domain that is structurally similar to the kinase domains of testisspecific protein kinase-1 and the LIM motif-containing protein kinases. TESK2 is ubiquitously expressed, but its function is unknown (26). There is also support for TESK2 from the chromatin interaction analyses. C10orf71 encodes an open reading frame of unknown function that is highly expressed in heart and skeletal muscle. Chromatin interaction analyses indicate MAPK8 as a second candidate gene at the C10orf71 locus, MAPK8 is involved in formation of the heart as well as HR regulation (27,28). DALRD3 encodes a protein with a DALR anticodon-binding domain similar to that of class Ia aminoacyl tRNA synthetases (29). The conditional analysis results provided one new, unreported association at a previously reported HR locus, KIAA1755 (rs41282820; c.1528C>T or c.1528C>A; p. Arg510Ter, a loss of function variant). KIAA1755 is predicted to encode an uncharacterized protein, and is only characterized at the transcriptional level. The transcript is highly expressed in the brain and nerves, and it is also expressed in the heart. Our analyses and the recently published UK Biobank analyses (17) discovered a second low-frequency non-synonymous SNV at CCDC141 (rs10497529, c. 442C > T, P. Ala141Val). CCDC141 (also known as CAMDI) encodes the coiled-coil domain containing 141 protein and interacts with DISC1 (disrupted in schizophrenia 1) and MYL2 (phosphorylatable myosin light chain). CCDC141 is highly expressed in heart muscle (30). Knockdown of CCDC141 in neurons leads to abnormal cortical neuronal migration, but there are otherwise limited functional studies of CCDC141 (30). The CCDC141 locus includes TTN (titin), which encodes a major structural protein in striated muscle. TTN mutations are associated with a range of hereditary myopathies (31). Prior work (12) using RNA interference in Drosophila melanogaster has shown that knockdown of TTN leads to significant changes in resting HR and HR post tachypacing, supporting TTN is a causal candidate gene at this locus. The new data described here implicate CCDC141 as a second candidate gene at this locus for functional follow-up. Enrichment analysis of HR variants in DNase I hypersensitivity sites across nearly 300 tissue samples and cell lines indicated new candidate tissues, such as neuronal progenitors and fetal muscle as being functionally relevant. Our data suggest these tissues should be targeted for future functional studies. Our long-range regulatory chromatin interaction analyses provided additional support for some of the candidate genes have been experimentally tested previously (12) and shown to have an HR-related phenotype (CALCRL, TTN, HTR2B, PLD1 and CHRM2). By expanding the list of HR loci to include new and published, several new candidate genes are highlighted for functional studies in Supplementary Material, Table S10. The Exome Chip contains non-synonymous, splicing and stop-coding variants that are thought to alter protein expression and function. Our analyses discovered four novel coding variants, indicating potential candidate causal genes at these loci. Our two-stage study design permitted the robust validation of all our novel loci findings, with a large replication sample size from UK Biobank (N ¼ 134 251) to add together to our European discovery data (N ¼ 104 452) for a large combined meta-analysis. However, due to the Exome Chip covering mainly coding regions, we were not able to compare results with all previous GWAS findings. In conclusion, our results taken together with recent studies (12) indicate HR-associated SNVs are mostly common (MAF > 5%) and have relatively small effect sizes. The maximum effect sizes reported thus far are $0.70 BPM per allele and MAF of 1% for SNVs at CCDC141 (rs17362588) and GJA1 (rs1015451). An analysis of much larger sample sizes (1M and above) including rare and common SNVs, and samples across different ancestries may provide further information on the contributions of both coding and non-coding variants, and the importance of rare coding variants in HR. Study populations, phenotypes and exclusions Thirty cohorts contributed data to the discovery meta-analysis in individuals of European ancestry. Details of all participating cohorts are provided in Supplementary Material, Table S13, including phenotype, cohort ancestry, study design and key references. The UK Biobank study, which was only recently published since the completion of our meta-analysis (17), provided results for replication analyses. Details of this study are also included in Supplementary Material, Table S13. All participating cohorts either measured RR-intervals from the standard 12-lead electrocardiogram (ECG) or used HR measurements (in beats per minute) from peripheral pulse measurements (Supplementary Material, Table S14), which were converted to the RR-interval scale (in milliseconds) using the inverse relationship formula: RR (ms) ¼ 60 000/HR (BPM). The discovery analysis was undertaken using the RR-interval phenotype. The exclusion criteria included: extreme RR-intervals (< 600 or > 1500 ms), atrial fibrillation on the ECG, a history of myocardial infarction or heart failure, use of nondihydropyridine calcium-antagonists [Anatomic Therapeutic Chemical (ATC) code C08D], digoxin (ATC code C01AA5), second or third degree atrioventricular block and a pacemaker signal on the ECG. Local ethics committees approved the contributing studies from the CHARGE consortium, and all individuals provided their consent in writing. The UK Biobank study has approval from the North West Multi-centre Research Ethics Committee and has Research Tissue Bank approval. Study-level genotyping and quality control All discovery cohorts were genotyped using a human Exome Chip array (exact details of the chip for each study are provided in Supplementary Material, Table S15). Quality control (QC) was done according to CHARGE Exome QC guidelines, including joint variant calling with zCall (32). At the study-level, the samplelevel QC consisted of excluding samples of non-European ancestry (for European-ancestry cohorts), samples with call rates <95%, samples with sex discordance or related samples with an unexpected high identical by descent estimate. It was recommended that principal components (PCs) be obtained using variants with MAF ! 1%. The variant QC consisted of exclusion of SNVs with call rate < 95%, with Hardy-Weinberg equilibrium values of P < 1 Â 10 À6 , and of variants that were strongly associated with plate assignment. Study-level statistical analysis Each cohort performed two SNV association analyses using an additive model implemented with the R package SeqMeta, http://cran.r-project.org/web/packages/seqMeta/index.html. Analyses were stratified by ancestry. One SNV association analysis used an untransformed model with RR-interval as the outcome, adjusted for age, sex, BMI and cohort-specific adjustments. The other SNV association analysis was a model based on the rank-based inverse-normal transformed residuals (RR-INVN), with residuals taken from a linear regression RRinterval adjusted for age, sex and BMI covariates. The RR-INVN analysis was performed to check for potential sensitivity to deviations from normality within the analysis of rare variants. Additional cohort-specific covariate adjustments were also applied, which included for example PCs or family structure. Central QC and meta-analyses We performed additional QC checks centrally. For each study, we checked the sample size and the total number of SNVs (monomorphic and polymorphic) and assessed the beta distribution. Within each cohort's results, all monomorphic SNVs were checked to have non-available results. In order to detect potential strand-flip issues, the cohort-coded effect allele frequencies (EAF) of each SNV were compared with the metaanalysed EAF of a group of CHARGE cohorts (AGES, ARIC, CHS, FHS and WHI). Any discordant SNVs showing cohort-EAF $ 0 in at least one study, but meta-EAF $ 1, or vice versa, were excluded from the central meta-analysis. A set of approximately 11 000 SNVs that were known to have QC issues from central CHARGE QC were also excluded from the meta-analysis. Quantile-Quantile plots were produced to inspect each cohort. After all QC steps were completed 235 677 SNVs remained. The results from all cohorts were then combined into a discovery meta-analysis using the SeqMeta R package. Sensitivity analyses A sensitivity analysis was performed on the use of betablockers (ATC code C07) due to the recognized effects of betablockers on HR. All cohorts with data on beta-blocker use were re-analysed with exclusion of individuals using beta-blockers at the time of phenotype measurement. Results of this metaanalysis were compared with the results from the same subset of cohorts with beta-blocker users included. Selection of variants for replication All SNVs with P < 1 Â 10 À5 from the discovery meta-analysis in European individuals were considered for follow-up. As a QC step after meta-analysis, we excluded four SNVs with unrealistically high beta values, large standard errors and results that were reported in less than four studies. We defined a novel locus as a genomic region (i) with SNVs not in LD (r 2 < 0.2) with any well-established HR-associated SNVs from the previously reported GWAS (12) (Supplementary Material, Table S1), and (ii) mapping to more than 500 kb from either side of a previously reported HR-associated SNV. At the time of our study, there were 21 loci reported from GWAS analyses with HR-associated SNVs (12). A potential secondary signal within a previously reported locus was defined as being within a 1 Mb region centred around the published SNV, but not in LD (r 2 < 0.2) with the published SNV in that region. LocusZoom plots were produced for all selected SNVs. Only the lead SNV was carried forward, for each signal being followed up. Specifically, the most significantly associated SNV was selected for any SNVs in pairwise-LD (r 2 > 0.2). LD was calculated within UK Biobank genetic data, in order to calculate pairwise-LD for all 21 known SNVs (not only those covered on the Exome Chip). Replication analyses We used data from UK Biobank for replication of the selected SNVs (at the time of analysis genetic data were available for 150 000 individuals). The UK Biobank data were analysed with untransformed HR as the phenotype, with no exclusions for medication use. In UK Biobank resting HR was assessed by two methods: first, pulse rate using an automated reading during blood pressure measurement, and second, pulse rate during arterial stiffness measurement using the pulse wave form obtained of the finger with an infra red sensor. When multiple HR measurements were available during the first visit for an individual, these measurements were averaged. In 99.7% of participants at least one single measurement was available. Individuals were excluded with extreme (> 4 SD) values (N ¼ 818). Further details are provided (17). The results of our European exome discovery meta-analysis for RR were combined with the UK Biobank replication results for HR (N ¼ 134 251), and a combined meta-analysis, using sample-size weighted fixed effects meta-analysis in METAL was performed (33). All alleles were aligned between the discovery and replication data, and the inverse relationship between RR-interval and HR was taken into account, i.e. so that a negative beta direction from our discovery data for a decreased effect on RR-interval was made equivalent to a positive beta. A novel locus was declared if the lead SNV reached exomewide significance in the combined meta-analysis of discovery and replication data (P < 2.12 Â 10 À7 ) and replicated with Bonferroni-adjusted significance (P < 0.0042 for 12 tests) in the replication data alone. In addition, the directions of effect between the discovery and replication data were required to be concordant, taking into account the inverse relationship between RR from our discovery data and HR from the replication data. Potential secondary SNVs at known regions were declared as validated if there was an exome-wide significant association in the combined meta-analysis. Variants that validated were subsequently tested for independence from previously reported HR variants in a conditional analysis. Conditional analysis In order to determine whether the validated secondary signals at previously reported loci were independent of the published SNV, conditional analysis was performed within GCTA software (34) applying the -cojo method (consisting of conditional and joint analysis with stepwise model selection). The input data were the exome-wide summary statistics from the full discovery meta-analysis of RR-interval in Europeans. The 1958 Birth Cohort Study (1958BC; N ¼ 5815) dataset was used as the reference for genotype data, because it represents one of the largest discovery studies (See Supplementary Material, Table S13). LD was calculated between pairwise SNVs, but any SNVs further than 10 Mb apart were assumed to not be in LD. All autosomal chromosomes were analysed, with MAF restricted to ! 0.01%, to allow for low frequency secondary SNVs, whilst taking into account the statistical power achievable. To allow for secondary associations a P-value cut-off of 1 Â 10 À4 was used as the modelling selection threshold within the GCTA analysis. Results were then extracted for the nine previously reported regions, within which potential secondary signals had been validated from the combined meta-analysis. To be consistent with the look-up threshold for selecting SNVs to carry forward from discovery to replication, results were restricted to SNVs with a significance level of P < 1 Â 10 À5 from both the discovery meta-analysis and the joint association from GCTA. Gene-based testing Gene-based testing was conducted using the primary discovery data in Europeans. Analysis was performed using the SNV-set Kernel Association (SKAT) test within the seqMeta R Package. SKAT tests were performed according to two different MAF filters of 1% and of 5%, and three different levels of variant filtering, based on annotations within the CHARGE Exome SNP Info annotation file: (i) all variants, (ii) variants deemed predicted to be damaging (24) and (iii) variants that were non-synonymous or leading to abnormal splicing. For gene-based tests we adjusted for multiple testing using the Bonferroni correction, according to the number of genes tested. The gene-wide significance level was calculated as 1.98 Â 10 À6 for 25 241 tests (i.e. the number of genes on the Exome Chip). For any genes attaining significance, the gene-based tests were repeated with exclusion of the most significantly associated lead variant, in order to confirm that the association was due to multiple rare variants. Non-European ancestry analyses Association results were also received for non-European samples. Analysis and QC were performed as described for the European data. A meta-analysis was performed centrally in seqMeta for AA ancestry, combining data from the five AA cohorts. Study-level results remained for HIS and CH ancestries (from only the MESA cohort), in order to consider the three non-European ancestries (AA, CH and HIS) separately from stratified analyses. Due to the smaller sample sizes, power calculations were performed using the Genetic Power Calculator (35), based on the average percent trait variance explained per locus being 0.04%, according to the recently published results from 64 validated HR loci explaining $2.5% of HR variance (17). To assess the level of heterogeneity by ancestry in non-European data, we performed a look-up of SNVs at the 12 published HR loci covered on the Exome Chip, extracting results for these variants from each of the AA, CH and HIS results. We restricted our primary discovery analysis to Europeans only after finding a lack of significant validation and concordance between EUR and non-EUR data for previously reported HR variants. As a secondary analysis, we performed look-ups of all validated novel loci within the non-European data. The forest plots for all validated novel loci display non-European results, to serve as a comparison to results within Europeans. In addition to calculating the percentage of concordance of effect directions for each ancestry compared with Europeans, a Binomial sign test was also performed in R. This test was based on the number of SNVs with consistent effect directions, and it was done to determine whether the concordance was higher than expected by chance alone, using P < 0.05 to declare significant concordance. Variance explained The percentage variance explained for RR-interval was calculated using data from all subjects in the 1958BC study. The SNV genotypes were extracted from the 1958BC Exome Chip data and considered in two different sets: the 12 previously reported SNVs covered on the chip including proxies (r 2 ! 0.8; see Supplementary Material, Table S1); and the lead SNVs from the five unreported novel loci (see Table 1A). First, RR-interval was regressed in a linear model against the sex and BMI covariates (not age, as all 1958BC subjects are of same age). Then the trait residuals from this first model were used as the phenotype in a second linear regression model, with all SNVs in the given set analysed jointly as multiple predictors, and adjusted for the top 10 PCs. The percentage trait variance explained by the set of SNPs was estimated from this second model, according to the adjusted R 2 value. HR loci annotation For the purposes of annotation, all signals were expanded to include SNVs in LD. LD was calculated within the UK Biobank full genetic dataset using PLINK (v1.9). All variants with an r 2 ! 0.8 within 500 kb downstream or upstream of the SNVs of interest were identified. These variants were annotated using ANNOVAR [vJun2015 (19)]. ANNOVAR functionally annotates variants, provides their conservation score, identifies SNVs that may cause protein-coding changes and reports their damaging prediction scores. Various prediction scores are available in ANNOVAR, including SIFT, PolyPhen and MutationTaster, among others. We investigated the unreported novel SNVs and their proxies (r 2 ! 0.8) across 44 tissues available in the GTEx dataset (20) for eQTLs. We reviewed the results for SNV-eQTL associations across all tissues, focusing on the heart, nerve, lung, muscle, adrenal and brain tissues which may be relevant tissues for HR based on known physiology of HR and our results from the enrichment analysis. Genes reported as eQTLs are based on studyspecific significance thresholds (P-values < 10 À8 ) and r 2 ! 0.8 between HR-SNV and top-eQTL SNV (the SNV most significantly associated with transcript). PhenoScanner PhenoScanner (36) was used to identify variants that are associated with other traits. All proxy SNVs in high LD (r 2 ! 0.8) with the lead SNVs at our five unreported novel loci were investigated in the PhenoScanner 1000 Genomes reference dataset. Results were filtered to those reaching a genome-wide significance P-value 5 Â 10 À8 . Potential candidate genes at new HR loci Candidate genes at each locus were compiled using LD information, ANNOVAR-derived annotation and eQTL lookup results. A literature review was conducted for potential candidate genes at each new HR locus. Sources of information included: published articles, GeneCards, Online Mendelian Inheritance in ManV R , the Human Protein Atlas, STRING and UniProt. We searched for information on the corresponding mouse models via the International Mouse Phenotyping Consortium and the Jackson Laboratory online catalogue. URLs for each of the sources is provided in the URL section below. Pathway analyses Pathway analyses were performed using QIAGEN's IPA V R (QIAGEN Redwood City) software. In order to distinguish the pathway enrichment contribution of novel loci from known HR loci, two sets of analyses were conducted. The first analysis captured the total known signal to date, investigating all 67 loci currently published, which include the 21 loci from the previously reported GWAS (12) and the 46 loci recently published from UK Biobank (17) since the completion of our meta-analysis. The second analysis included our five unreported novel loci in addition to all the previously reported loci. In each case, the analysis included all genes annotated from the lead SNVs and their proxies (r 2 ! 0.8). Results were filtered for pathway enrichment of P-values 10 À4 . We specifically report the pathways for which enrichment is increased with the inclusion of genes from our novel loci. Enrichment in DHSs To identify the tissues in which HR-associated SNVs are active, we used FORGE to look for enrichment of DHSs in 299 tissue samples from the Roadmap Epigenome Project (37). FORGE calculates enrichment for overlap of HR variants with DHS by comparison with overlap of DHSs with 1000 matched background variant sets (matching distance to transcriptional start sites, GC content and MAF). We performed two different enrichment analyses. First, we did a 'known' analysis using all 67 currently published lead SNVs to date [21 previously reported from the original GWAS (12) and 46 new loci from the recently published UK Biobank study (17)]. Second, we did an 'all' analysis using the lead SNVs at our five unreported novel loci and the five independent secondary SNVs that we found at previously reported loci; together with the 67 known signals, denoted as the 'all' analysis. We compared the enrichment results of the two analyses, in order to identify any new enrichment due to the inclusion of our novel loci. The enrichment is expressed as Z-score statistics. A Z-score of 2.58 was used as a threshold for statistical significance, which corresponds to false discovery rate (FDR) < 1.5%. We calculated the Z-score all À Z-score known (DZ-score) for those tissue samples that were found statistically significant in the 'all' analysis in order to assess the effect of the 10 new, additional SNVs from our study. Regulatory potential of SNVs We selected the HR-associated SNVs and proxies in LD (r 2 ! 0.8; calculated using the UK Biobank full genetic dataset) that were identified in this study, and from the previous GWAS (16) and UK Biobank studies (17) for annotation. To identify the potential regulatory variants, we retrieve the functional confidence score for SNVs from the RegulomeDb database (21). RegulomeDb assigns a functional confidence score to each SNV by overlapping them with functional genomic data mainly from ENCODE (e.g. DNase I hypersensitivity, DNase I footprinting, ChIP-seq), with eQTL data and with computational prediction (e.g. TF-binding sites and their disruption). We considered any SNP with at least one functional annotation to have regulatory potential (this corresponds to functional confidence scores: 1a-6). Long-range regulatory contacts Using significant long-range chromatin interactions as identified by Fit-Hi-C in right ventricle Hi-C data [40 kb resolution (22)], we annotated the potential regulatory SNVs with potential target genes, whose promoter is in contact with the given SNV. Where the 40-kb genomic region containing the SNV had more significant promoter interactions, we show the genes in order of most significant interaction to least significant. For every locus, we took the gene that had the most significant promoter interaction with a regulatory SNV, and using IPAV R , we assessed which pathways were affected, and specifically those that were enriched compared to using only genes in LD with HR-SNVs. Supplementary Material Supplementary Material is available at HMG online. Scotland team, which includes interviewers, computer and la-the study participants, the staff from the Rotterdam Study and the participating general practitioners and pharmacists. We thank Ms Mila Jhamai, Ms Sarah Higgins and Mr Marijn Verkerk for their help in creating the Exome Chip database, and Carolina Medina-Gomez, MSc, Lennard Karsten, MSc and Linda Broer PhD for QC and variant calling. Variants were called using the best practice protocol developed by Grove et al. as part of the CHARGE consortium Exome Chip central calling effort SardiNIA: We thank all the volunteers who generously participated in this study and made this research possible. SHIP: We thank all SHIP and SHIP-TREND participants and staff members as well as the genotyping staff involved in the generation of the SNP data. WHI: The authors thank the WHI investigators and staff for their dedication, and the study participants for making the program possible. A full listing of WHI investigators can be found at: http://www.whi.org/researchers/Documents%20%20Write% 20a%20Paper/WHI%20Investigator%20Long%20List.pdf YFS: The expert technical assistance in the statistical analyses by Irina Lisinen is gratefully acknowledged. UK Biobank: This research has been conducted using the UK Biobank Resource Application Number 9628. Conflicts of Interest statement. Dr B.M.P. serves on the DSMB of a clinical trial funded by the manufacturer (Zoll LifeCor) and on the Steering Committee of the Yale Open Data Access Project funded by Johnson & Johnson. Dr P.T.E. is the PI on a grant from Bayer HealthCare to the Broad Institute focused on the genetics and therapeutics of atrial fibrillation. Dr N.P. has received financial support from several pharmaceutical companies that manufacture either blood pressure lowering or lipid lowering agents or both and consultancy fees. Dr P.S. has received research awards from Pfizer. Dr M.J.C. is Chief Scientist for Genomics England, a UK government company. Danish Pharmaceutical Association, the Augustinus Foundation, the Ib Henriksen Foundation, the Becket Foundation, and the Danish Diabetes Association. KORA: The KORA study was initiated and financed by the	2017-07-11T15:45:17.542Z	2017-04-03T00:00:00.000	{ "year": 2017, "sha1": "63e0b45461ec65c7a13dc3044c733066d624916f", "oa_license": "CCBY", "oa_url": "https://academic.oup.com/hmg/article-pdf/26/12/2346/17658020/ddx113.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "63e0b45461ec65c7a13dc3044c733066d624916f", "s2fieldsofstudy": [ "Biology" ], "extfieldsofstudy": [ "Medicine", "Biology" ] }
254564508	pes2o/s2orc	v3-fos-license	Neural Bandits for Data Mining: Searching for Dangerous Polypharmacy Polypharmacy, most often defined as the simultaneous consumption of five or more drugs at once, is a prevalent phenomenon in the older population. Some of these polypharmacies, deemed inappropriate, may be associated with adverse health outcomes such as death or hospitalization. Considering the combinatorial nature of the problem as well as the size of claims database and the cost to compute an exact association measure for a given drug combination, it is impossible to investigate every possible combination of drugs. Therefore, we propose to optimize the search for potentially inappropriate polypharmacies (PIPs). To this end, we propose the OptimNeuralTS strategy, based on Neural Thompson Sampling and differential evolution, to efficiently mine claims datasets and build a predictive model of the association between drug combinations and health outcomes. We benchmark our method using two datasets generated by an internally developed simulator of polypharmacy data containing 500 drugs and 100 000 distinct combinations. Empirically, our method can detect up to 72% of PIPs while maintaining an average precision score of 99% using 30 000 time steps. Introduction Polypharmacy is most often defined as the simultaneous consumption of five or more drugs at once by a patient [18] and is a prevalent phenomenon in the older population. In the USA, 65.1% of older adults experience polypharmacy, with most of them using more than 5 medications at once [26]. In Canada, older adults in long term care facilities use on average 9.9 drug classes, while older adults living outside of these facilities use on average 6.7 drug classes [2]. Some polypharmacies can be dangerous, in the sense that they can lead to to negative health outcomes, like death or hospitalization. Fortunately, screening tools exists to avoid prescription of potentially inappropriate drugs (e.g., opioids and benzodiazepines [21]). These tools essentially consist of finite lists of individual drugs and drug pairs that have been identified as dangerous by experts during pharmaco-epidemiological studies as well as from experience. By definition, this means that potentially dangerous combination resulting from more than two drugs interacting together cannot be identified using the screening tools, in addition to other currently unknown dangerous drug combinations. In order to prevent the prescription of potentially harmful polypharmarcies, it would be important to expand the screening tools until they ideally contain all potentially dangerous combinations. Unfortunately, given all the possible drug combinations as well as their varying effects depending on different patient characteristics, pharmaco-epidemiologists cannot investigate all of them. The goal of this work is therefore to build a predictive model able to identify drug combinations at risk of being harmful, so that they can be investigated further. We propose to achieve this by leveraging neural networks to predict an association measure to a health outcome given any input describing an arbitrary number of drugs. In practice, such model would be trained using historical data on drugs prescribed to patients, their clinical and sociodemographic characteristics, and their health outcomes. These datasets are typically very large, which makes the association measure expensive to compute, and highly unbalanced. We therefore tackle the general problem of efficiently mining historical data to train a generalizable and useful model. To achieve this, we formulate the problem under the neural bandit setting so that it can be addressed with the Neural Thompson Sampling (NeuralTS) [27] strategy. However, using this strategy on a very large action space (such as the one considered here) also raises challenges, which we address by combining NeuralTS with differential evolution (DE) [24]. The proposed OptimNeuralTS approach finally results in an ensemble predictor made of the evolving sequence of models trained on all the intermediate data subsets. We evaluate the potential of OptimNeuralTS in simulated experiments. Our results show that our approach can be used to iteratively build an information-rich dataset that can in turn be used for training a predictive model, resulting in an ensemble model capable of extracting new potentially inappropriate polypharmacies (PIPs). We finally provide an overview of related work in machine learning (in general) applied to polypharmacy discovery and bandit strategies applied for data mining. We highlight two contributions: 1. Tackling the problem of efficient creation of information-rich datasets under the contextual bandit setting. 2. Introducing the OptimNeuralTS approach to learn predictive models by mining relevant data from very large unbalanced datasets. Problem formulation Let D denote a historical dataset containing information about drug combinations and health outcomes. Table 1 shows a simple example of such a historical dataset, where each line corresponds to a drug combination identified in a binary format, (i.e. 0/1 indicate whether a drug was taken or not by an individual) along with a binary variable indicating whether the individual developed (or not) a given health outcome while consuming the drug combination. In the targeted application, the measure of association is called the relative risk (RR) and is described as the risks for a given health outcome in the exposed population over the risks in the unexposed population [23]. An exposed patient is a patient which takes a given drug combination. The exposed population is simply the portion of the population which consumes the combination of drugs. Mathematically, given Table 2, the RR is defined as: RR has the advantage of having an implicit threshold: if RR > 1 then a drug combination is associated with a given health outcome, therefore it is potentially harmful, while RR < 1 implies that a drug combination protects against a given health outcome, therefore it is safe. RR = 1 means that a drug combination is neither protective against nor associated with a given health outcome. Safe Harmful In practice, historical datasets are extracted from medico-administrative databases. Therefore, medication claims data typically contain hundreds of columns related to drug usage and millions of rows enumerating all possible simultaneous drug combinations taken by all patients in a large population, but no precomputed association measure. Unfortunately, computing the association measure for a given drug combination is computationally expensive since it requires enumerating every row in the dataset. Therefore, as the size of the data grows, it becomes harder to compute. This raises a challenge when aiming to train a model capable of predicting an association measure to the considered health outcome for any drug combination provided as input because it is not computationally realistic to compute the RR for the whole dataset D. The training of a predictive model must therefore be performed on a subset of at most samples from D, where \|D \|. This opens the question of how to sample such subsets from D. In addition, one must note that the historical dataset D is highly unbalanced. Indeed, prescribers usually prescribe drugs which do not interact together in a harmful way, thus most drug combinations have a low measure of association to health outcomes such as hospitalization and death. Figure 1 displays a simulated but typical distribution of the estimated association measures observed in the real data. Therefore, randomly sampling from D would yield a training dataset containing mostly combinations of drugs with low measures of association as PIPs associated with adverse health outcomes are rare. However, it is well known that the performance of a predictive model highly depends on the quality of the underlying data [10]. In other words, if the training dataset contains few to no PIPs associated with adverse health outcomes, the predictive model is very unlikely to learn to identify such PIPs. Therefore we need a strategy capable of sampling a training dataset with higher odds of containing PIPs associated with adverse health outcomes. Proposed Approach We tackle this challenge by formulating the training dataset creation problem as a contextual bandit problem [14], where we leverage the NeuralTS [27] action selection strategy combined with DE [24] to select drug combinations for which to compute the RR. Neural Contextual Bandits A contextual bandit environment is described by a collection of actions A, a feature space X, and an unknown reward function ℎ : X ↦ → R, such that each action ∈ A is associated with a feature vector, or context, ∈ X. At each time step = 1, 2, . . . , of the contextual bandit game, the player (agent interacting with the environment) is presented with a subset of actions A ⊂ A. It then selects an action ∈ A to play and observes a noisy reward = ℎ( ) + , where is a -sub-Gaussian noise (e.g., ∼ N (0, )). The goal of an agent playing this game is to maximize the cumulative reward, defined as: The obvious solution is to simply play the optimal action * at every round, which has the highest expected reward for this round. That is * = arg max ∈A ℎ( ). However, function ℎ is not known a priori, therefore, the agent must learn by trial-and-error in order to improve its behavior over time. In the training dataset construction problem from historical data, A corresponds to the set of all possible drug combinations, A corresponds to the set of drug combinations that are available to explore at time , the features correspond to a multi-hot representation of drug combination , and the reward function ℎ corresponds to the measure of association between a drug combination and a given health outcome, i.e., the RR. At each time step , the agent selects a drug combination for which to compute the RR. Since computing the RR on the historical dataset is computationally expensive, it is instead computed on a subset of the data, hence the noisy reward . Neural bandit strategies, such as NeuralUCB [28] and NeuralTS [27], rely on a neural network (·; ) : X ↦ → R to model the reward function ℎ in order to predict the expected reward given any feature ∈ X. More importantly, these approaches can estimate the confidence interval around the prediction of the neural network to guide the exploration. They achieve this by using the gradient on the activation, (·; ) : X ↦ → R \| \| . NeuralTS NeuralTS uses the gradient to estimate the distribution of reward for an action. Indeed, at step , the parameters of a normal prior N ( , ) are estimated as follows: is a design matrix containing the gradient computed for the inputs (selected actions) up to time , is a regularization parameter and is the number of parameters in the neural network. NeuralTS selects the action to play by sampling a value from N ( ( ), ( )), ∀ ∈ A and picking the action with the highest value. After an action is played, the associated context as well as the observed reward are added to the training dataset D = D −1 ∪ {( , )} used for computing the new network parameters , before moving on to the next time step. Since the bandit strategy will seek to play actions which yield high rewards, this should lead to a dataset D containing a reasonable amount of drug combinations with high measures of association to a given health outcome. We therefore hypothesize that such a dataset will make it possible to train a predictive model capable of identifying PIPs with high precision. However, for the bandit strategy to be able to recommend actions with high RR, such actions need to be contained in the set of available actions A . Generating relevant available action sets From the neural contextual bandit problem formulation, action is selected from the subset A ⊂ A containing all available actions at time . This is due to the fact that the bandit strategy must consider each action ∈ A in order to recommend , and that the complete action set A is typically too large to be entirely considered at every time step. This is also the case in the considered application due to the combinatorial nature of polypharmacy. For the same reason that the predictive model training dataset cannot be sampled at random from D, we cannot generate A by randomly sampling from A due to the highly skewed distribution of RRs. We must therefore generate subsets A such that the presence of potentially harmful drug combinations is favored. To achieve this, we propose to generate subsets of available actions A using differential evolution (DE) [24], an evolutionary optimization algorithm which does not rely on a gradient signal to converge to a solution. The general principle behind DE is to maintain a population and mutate its members, which are feature vectors, according to a strategy. Here, we consider the best/1/bin strategy [24] described in Algorithm 1, where the objective function : X ↦ → R corresponds to an action value function sampled from a neural network. The DE optimization process is therefore conducted on function . DE with best/1/bin therefore corresponds to considering \|A \| = × available actions at each time step . Parameters and are typically chosen such that × \|A\|. The best member returned after the steps of DE corresponds to the action features maximizing , which is a value function given by the neural network model. Therefore the best member would correspond to . For example, with NeuralTS, (·) corresponds to a sample from the distribution N ( −1 (·), −1 (·)) at Algorithm 1 DE best/1/bin Generate a mutated feature vector where components are computed as follows: end for 11: end for 12: ★ ← arg max ∈W ( ) 13: return ★ time step (see Eq. 2 and 3). Now, computing on-the-fly on D requires the selected action to be contained in D. However, DE (best/1/bin) is not constrained to D, so this condition may not be fulfilled. In order to account for this situation, we propose to select the action as being the 1-nearest-neighbor in D to the action returned by DE (best/1/bin) with ties broken arbitrarily. This ensures that the association measure for the drug combination can be computed. OptimNeuralTS Algorithm 2 describes the resulting OptimNeuralTS. The agent warms up by randomly sampling actions during the first steps, observing their rewards and updating the internal parameters (lines 3-7). The neural network is then trained for the first time on the random data using a standard gradient descent with the L2 regularization scheme of NeuralTS (line 8), before the agent starts playing actions according to the NeuralTS and DE strategy. We slightly abuse the notation when calling DE (line 10) to indicate that the objective function evaluated at features consists in a normal distribution centered at −1 ( ) with standard deviation −1 ( ) (see Eq. 2 and 3). The design matrix of the agent is then updated (line 11), the agent plays the transformed action , observes the reward , and updates the dataset D before updating the neural network parameters with the same procedure as previously stated (lines [13][14][15]. OptimNeuralTS finally returns the dataset generated by the algorithm as well as the ensemble model corresponding to all the intermediate models ( . . . ) encountered along the search (line 17). Indeed, as experiments will show, the subsets of data encountered along the neural contextual bandit game will result in neural network models that are specialized in different relevant regions of PIPs. Combining these models in an ensemble therefore results in a strong predictive model with a good coverage of the input space, which can then be used to predict a RR for any given drug combination. Detecting new PIPs is then only a matter of applying a threshold over the predicted RR's lower confidence bound, which can be computed from (·) and (·). Warming-up Previous work showed that non-informative priors can impact performance [17]. To mitigate this, we allow the agent to warm-up by selecting random actions ∈ D and observe their rewards as a way to initialize its belief about the data. This effectively creates a small randomly sampled dataset composed of the seen contexts and the observed rewards. This is helpful, as the data gathered after this point is dependent on the previously gathered data, therefore breaking the i.i.d. assumption of data in supervised learning. This in turn can lead to a failure in learning as an agent without any representation of the relationship in the data may sample it poorly when playing, leading to a poor representation and so on. Transforming recommended actions into playable actions As previously mentioned, our problem requires that we transformˆinto . However, is updated with (ˆ; −1 ) instead of ( ; −1 ) (line 12). Two facts motivate this choice of update. The first is that if the entirety of A was present in D, then no transformation would be needed. Indeed, the transformation is only implemented in order to train the neural network on a relationship existing in the historical data and soˆis the recommended action. Secondly, due to the distribution of RRs on our data, the gradient of often contains very little information. This is due to the fact that the RR is concentrated around the mean, which leads to the last bias term of the neural network being almost the only participating term in the prediction. The gradient vector ( ; ) is then almost barren, which in turn leads to a design matrix containing little information. In practice, updating with ( ; ) works, but we have found it leads to much less new PIPs detected much later during the bandit algorithm's training. Experiments Evaluating the proposed approach requires a dataset with a ground-truth and a structure similar to the real world data. As no such dataset is readily available, we first develop a simulator to generate synthetic data. Synthetic data generation Our main hypothesis guiding data generation is that a drug combination similar to a drug combination with a high RR should have a similarly high RR. Consequently, we generate by sampling from binomial distributions the set P of what we call "dangerous patterns", which are characterized by a high RR. Likewise, we randomly generate the set of distinct drug combinations C without attributing them a RR. The similarity between each drug combination ∈ C and each dangerous pattern ∈ P are then computed using the Hamming distance. Two cases can arise from this: 1) either drug combination has some drug(s) in common with the nearest pattern or 2) it does not. If it does, then the drug combination is said to intersect with the pattern and is attributed a RR proportional to its similarity to its nearest pattern . Alternatively, if a combination is disjoint of the nearest pattern, then its RR is sampled Fig. 2 Overview of the simulated assignment of RR to drug combinations. The RR attributed to a combination is proportional to its similarity to its nearest dangerous pattern , shown as the parent in the resulting tree. inter , disjoint and disjoint are user-defined parameters. Disjoint combinations from a normal distribution N ( disjoint , disjoint ). This procedure results in a dataset with only distinct combinations with a precomputed RR. The Hamming distance, by definition, favors dangerous patterns containing smaller subsets of drugs during the nearest pattern search. As a result, a combination's nearest dangerous pattern is not always the one with the biggest overlap in terms of drugs. This results in a dataset with a RR not necessarily increasing proportionally to the size of the intersection between combinations and patterns, which adds difficulty to the problem. Figure 2 gives the overview of the RR generation process. Experimental setup We devise one experiment on two datasets generated by the simulator. The goal of the experiment is to detect drug combinations with an RR above a certain threshold. In our work, we consider a threshold of 1.1 for the RR since we consider a RR between 1.0 and 1.1 to be too low to be significant. The most important aspect here is not to find every PIP but to detect them with as few false positives as possible. This is crucial, as in practice these findings need to be further studied by healthcare experts and it is laborious to do so. Furthermore, since we plan on using samples of the real dataset in our practical application, we also add a noise term ∼ N (0, 0.1) to the observed RR for a drug combination to simulate sample noise. Therefore, to ensure a low false positive rate, a drug combination is only classified as potentially harmful if ( ) − 3 ( ) > 1.1, to emulate a pessimistic 99% lower confidence bound. The choice of = 0.1 for the normal distribution is so the noise does not dominate the reward signal while still resulting in a challenging instance in the datasets described below. We generate two datasets each representing a different hypothesis on the effect of the consumption of drugs: a neutral effect instance,where most RRs are near 1, and a protective effect instance (where most RRs are concentrated near 0). The average RR in the latter is well below the dangerous RR threshold, as would be expected in a real dataset. However, to study the robustness of the proposed approach, we also consider a neutral effect dataset where the distribution of RRs is concentrated around the threshold. This leads to a more challenging instance as the noise is more likely to make a safe combination appear potentially harmful. Figure 3 shows the distribution of RRs in both datasets. Both datasets contain 100k distinct combinations of 500 possible drugs, with RRs computed from 10 random dangerous patterns that do not appear in the distinct combinations. The two datasets are highly unbalanced, with the neutral and protective datasets respectively containing 2082 and 7805 distinct dangerous drug combinations (RR > 1.1). Table 3 shows the OptimNeuralTS parameters used in this experiment. The number of time steps is chosen such that only a small fraction of the entire drug combination space can be investigated during the bandit game. Indeed, in real life applications, millions of drug combinations are typically available. Setting to a small number compared to the number of possible combinations thus requires Op-timNeuralTS to be efficient in its choice of drugs to investigate at every round in order to succeed. The results shown here are for a warm-up duration of = 10k samples and an exploration factor of = 1 taken as the best configuration from a grid search of the space ∈ {1k, 10k, 20k, 30k} and ∈ {1, 10} . All the configurations in the grid search succeed in finding variable amounts of PIPs with high precision, except when the warm-up phase is too long (e.g. = 30k), highlighting the need for a bandit strategy. Furthermore, the considered space for the grid search of is never below 1 to discourage greedy action (i.e. exploiting the already known PIPs). Furthermore, we schedule the learning rate = 0.01 to decrease as the loss reaches a plateau during training. This parameter was found by a hyperparameter search guided by OpTuna [1] and Tune [15]. As for DE, the parameters were selected The total number of configurations tried is thus 7, as = 30k is the same for any manually to maximize the mutation of the population at each optimization step while still maintaining a very small population and a quick runtime. As previously mentioned, by applying a threshold over the lower bounds on neural network predictions, the regression problem of learning a mapping from drug combinations to a RR can be turned into a binary classification problem where a prediction lower bound over a threshold (e.g. 1.1) corresponds to a PIP, else to a safe drug combination. Therefore, classical classification performance metrics such as precision and recall are used here to evaluate the model trained by OptimNeuralTS. In addition to these two metrics, we also report the ratio of dangerous patterns used to generate the data that were found (Ratio P), as well as the ratio of PIPs detected in D that are not in D (Ratio ∉ D ). These last two metrics aim to evaluate the generalization to PIPs unseen during training as the dangerous patterns P are not in D but are still known to have a high RR. All the results are reported for 25 repetitions of the experiments and every evaluation metric is computed at every 200 time steps of training. Furthermore, in order to quantify the benefits of using an ensemble model we compare the metrics of the ensemble approach to that of the latest trained model (i.e. the single model trained by OptimNeuralTS before the evaluation). Implementation details Since feature vectors are multi-hot vectors, the recommended drug combination (ˆ) is transformed into the most similar one in D ( ) using the Hamming distance. Furthermore, unlike the original NeuralTS training routine that trains for a set amount of gradient steps and then returns the last parameters computed with the gradient step, our implementation keeps the parameters associated with the lowest loss on the training dataset as this maximizes likelihood [9]. As training is time consuming, the neural network is retrained every 10 steps and uses the Adam [12] optimizer due to its faster convergence than regular SGD. Finally, in order to simplify and accelerate computation, we rely on the same tricks as the original NeuralTS implementation [27]: we approximate the matrix by taking only its diagonal, remove the division by (see Eq. 3) and only compute the L2 penalty on the current weights . Precision on the neutral and protective instances using the single latest model Fig. 7 Recall on the neutral and protective instances using the single latest model progresses, the agent must focus on drug combinations near the RR threshold, thus leading to slightly more false positives. Impact of ensemble To highlight the benefits of using the history of generated models as an ensemble rather than simply relying on the most recent model, Figures 6 and 7 respectively show the precision and recall obtained at each time step by using only the most recent model instead of the ensemble. We first observe (Fig. 7) that the most recent model is not consistently getting better at detecting PIPs on its own. Indeed, later models sometimes have worse recall than those of earlier iterations, suggesting that they did not retain knowledge acquired earlier. However, we observe (Fig. 6) that high precision is maintained throughout the time steps, although with more noise compared with the ensemble (Fig. 4). That is, every individual neural network trained by OptimNeuralTS has a low false positive rate. The ensemble leverages this fact efficiently by using a single vote to classify a combination as potentially harmful. Moreover, we observe in general that the precision for single models fluctuates more for the protective instance than for the neutral instance. This behavior is most likely due to the wider range of RR of the protective dataset which results in Ratio ∉ D on the neutral and protective instances using the ensemble predictive model drug combinations having similar components but very different RR. Even so, the precision remains high enough to have very little false positives in practice for both datasets when using ensembles. Indeed, the ensemble approach results in 819 ± 47 correctly detected PIPs for 20 ± 9 false positives on the neutral instance while it yields on the 1504 ± 481 correctly detected PIPs for 1 ± 0 false positives on the protective instance. The bigger fluctuation in the number of true positives for the protective instance can also be attributed to the bigger range of RR to cover. Figure 8 shows the ratio of dangerous patterns P identified as PIPs by the resulting model. As previously mentioned, the dangerous patterns used to generate data are not in D. However, although they can never be observed directly, the ensemble is capable of detecting dangerous patterns. Furthermore, we observe that a good percentage of the detections made by the ensemble are not contained in the training data D . Generalization We observe (Fig. 9) that on average up to 39% (on the protective instance) of detected PIPs were not even in D upon the completion of training. This is promising, as it indicates that the resulting ensemble can pick up on observed trends to predict unseen patterns. In practice, this would represent drug combinations that have never been prescribed together before, but whose combination could be dangerous, and so detecting them will prevent health risks for patients. Related Work There exists prior work on data mining in the polypharmacy context. Methods have been proposed to detect new potentially inappropriate medications or model the association of polypharmacy to side effects using small datasets [25,11]. General machine learning techniques have also been used to model polypharmacy and its side effects [30,19,13] by using complex drug data that are usually not contained in claims database. Therefore, efficiently learning from large datasets such as those considered in the current work require new approaches, hence motivating the proposed bandit angle. Several contextual bandit strategies have been proposed previously to handle combinatorial problems with linear rewards [3,7]. The linear reward assumption is common and allows for efficient computations of action recommendations. However, the tackled application requires the estimation of a (possibly) non-linear reward functions. Although there exists non-linear reward combinatorial bandit strategies [6], they rely on oracles to recommend an action to play. Such oracles are typically designed for specific problems and unfortunately, our problem is not one of them. Alternatively, strategies to extract the top-best actions [22] do not assume linearity of the reward function and do not rely on an oracle. However, they require to preset the -order magnitude of relevant actions, which is a priori unknown in the current application. Considering too low would result in missing PIPs, while setting it too high would result in false positives. As our objective is not to maximize the cumulative rewards (Eq. 1), but rather explore drug combinations in order to detect potentially dangerous ones, the pure exploration setting would also be a natural formulation for the current application. Several combinatorial pure exploration bandit strategies have been proposed [5,8,4] previously. However, due to their combinatorial nature, they all exhibit a dependency on an oracle, which makes them then unusable in the tackled problem. Pure exploration neural bandits have also been studied previously [29]. Although theoretically relevant, these methods bear important implementation challenges that prevent them from being used efficiently. The proposed OptimNeuralTS strategy is simpler to implement while still maintaining high precision and a capacity to generalize to unseen data. Finally, thresholding bandits [16] is another relevant setting where the objective is to extract actions with a mean value estimate over a certain threshold. However, proposed approaches for this setting [16,20] only maintain mean estimates of every actions encountered during the game and could therefore not lead to a model that can predict the association measure for any new drug combination. Without such a model it becomes impossible to generalize to unseen actions as required by the tackled application. Conclusion This paper introduces the OptimNeuralTS approach combining NeuralTS [27] and differential evolution [24] to data mine relevant data from very large unbalanced datasets. This method leverages the neural contextual bandit formulation to create an information-rich dataset on which to learn an ensemble predictive model. Optim-NeuralTS is a general method for data mining that can be applied to any unlabelled dataset with a combinatorial structure. We conduct experiments using simulated datasets representing both protective and neutral settings. Results show that the predictive model learned with OptimNeuralTS is empirically capable of detecting PIPs with high precision. More importantly, the model is able to identify underlying dangerous patterns that are not observed directly in the data. These encouraging results suggest that OptimNeuralTS is a promising approach for guiding pharmaceutical research by recommending potentially dangerous drug combinations to investigate further and therefore contribute to safer prescriptions. In future work, one could attempt to improve the sample efficiency using pure exploration neural bandits methods. Furthermore, while we do not take them into account, other important factors other than the presence of drugs (e.g. sex, age, medical conditions) contribute to whether a combination should be considered a PIP. Therefore, our simulation data can still be improved to portray a more complete setting.	2022-12-13T06:41:44.761Z	2022-12-10T00:00:00.000	{ "year": 2022, "sha1": "8783ced50bf8bf10775309ea325feda57860eaec", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "8783ced50bf8bf10775309ea325feda57860eaec", "s2fieldsofstudy": [ "Computer Science" ], "extfieldsofstudy": [ "Computer Science" ] }
119099212	pes2o/s2orc	v3-fos-license	Single Experimental Setup for High Sensitive Absorption Coefficient and Optical Nonlinearities Measurements Accurate knowledge of absorption coefficient of a sample is a prerequisite for measuring the third order optical nonlinearity of materials, which can be a serious limitation for unknown samples. We introduce a method, which measures both the absorption coefficient and the third order optical nonlinearity of materials with high sensitivity in a single experimental arrangement. We use a dual-beam pump-probe experiment and conventional single-beam z-scan under different conditions to achieve this goal. We also demonstrate a counterintuitive coupling of the non-interacting probe-beam with the pump-beam in pump-probe z-scan experiment. Introduction Development of high power laser sources has motivated an extensive research in the study of nonlinear optical properties and optical limiting behavior of materials. There exits a continued effort in making sensitive measurements on absorption coefficient and nonlinear coefficients, however, most of the experimental techniques are focused on measuring one or the other of these two important parameters. A variety of interferometric methods 1,2 , degenerate four wave mixing 3 , nearly degenerate three wave mixing 4 and beam distortion measurement 5 , have been used for measuring the nonlinear refractive index. One of the most important techniques to measure nonlinear refractive index was shown by Sheik Bahaei et al. 6 This technique is simple and versatile yet is highly sensitive. However, an accurate knowledge of absorption coefficient (α 0 ) is necessary for the use of this technique, which is a serious limitation for unknown samples. An easy way to measure α 0 is to use the Beer's law [7][8] , which operates in the linear absorption regime and has limited sensitivity. More sensitive methods have been developed, of which the technique 9-15 using thermal lens (TL) effect is perhaps the simplest and the most effective. In this method, a lens focuses the laser beam into the sample resulting in a temperature gradient, which in turn produces a spatial gradient in refractive index. The relative change in transmittance of the laser beam can then be measured, after passing through an aperture, with the help of a detector [9][10][11] . Shen et al. [12][13][14] introduced a pump probe laser scheme under mode-mismatched and mode-matched conditions to improve sensitivity of the TL method. More recently, Marcano et. al. 15 have used this method to measure the absorption coefficient of water with high accuracy. A single experimental technique to measure both the parameters, however, is yet to emerge, which could be of significance in the study of new materials. In this paper we introduce a single experimental technique to measure both the parameters. Our aim has been to measure the absorption coefficient (α 0 ) as well as the real and imaginary parts of third-order optical nonlinearity ( ) with high sensitivity in a transparent sample using a single experimental setup. We also show how mid-IR absorption in water (at 1560nm) manifests itself as a minute nonlinear absorption coefficient in the near-IR transmission window (at 780nm). Such high sensitive experiments have become possible due to the ultrahigh sensitivity provided by a stable ultrafast laser operating at both the above wavelengths. ) 3 ( χ Our technique is a modification of the well-known z-scan technique introduced by Shiek Bahaei et. al 6 in 1990, where one measures the change in transmittance of a focused laser beam through sample that is being moved through the focal point of the lens. Since we are dealing with Gaussian optics when the beam is passing through lens and sample, we will use the Gaussian optics formalism. Subsequently, we will discuss the z-scan theory and finally we will discuss the modifications that we have introduced in addition to the new experimental results and discussions. Background a. The Gaussian beam in a homogeneous medium In most laser application it is necessary to focus, modify or shape the laser beam by using lenses or other optical components. In general, laser beam propagation can be approximated by assuming that the laser beam has an ideal Gaussian intensity profile corresponding to TEM 00 mode. Using Maxwell equation in an isotropic, charge free medium one can derive the wave equation 16,17 : where k . Let us assume a solution whose transverse dependence is only on in Eq.(1). We consider nearly plane wave situation where the flow of energy is along a single direction (e.g. z), and therefore the electric field, E, is: Substituting these in Eq.(1) we derive where we have assumed that longitudinal variation is slow, such that k is valid. In the next step, we take Ψ of the form If this equation is to hold true for all r, then the coefficients of different powers of r must be equal to zero, which leads to: Q kQ For solving this differential equation we introduce a function S(z), such that Replacing the value of Q in Eq.(6) with the relation from Eq.(7), we get It is more convenient to deal with a parameter q, where q z k Q z ( ) ( ) = . So that we can rewrite Eq.(9) as: where q is a constant (q ). From Eq. (6) and (11) we have where the arbitrary constant of integration is chosen as zero. The constant of integration will modify the phase of the field solution, Eq.(2). Since the time origin is arbitrary, the phase can be taken as zero. Combining Eqs. (11) and (12) in Eq.(4), we obtain We take the q to be purely imaginary and express in terms of new constant w where z w 0 0 2 = π λ . We can write Eq.(2) as: and we can also write which is the fundamental Gaussian beam solution. The parameters w z w ( ), 0 are beam spot size and minimum spot size at z=0 and the parameter R(z) is the radius of curvature of the spherical wavefronts at z. Our aim is to calculate spot size of the beam when it passes through a thin lens of focal length f as shown in the Fig.1. Since, at the input plane (1) of Fig. 1, w w = 01 and R 1 = ∞, we can write using Eq.(18) the following relation: Similarly, at the output plane (2) of Fig.1, we get: and the minimum spot size in focal point is equal to: where z w The other parameter of interest is the Rayleigh range (RR), which is the axial distance from the point of minimum beam waist (w 0 ) to the point where the beam diameter has increased to 2 w 0 in the region of a Gaussian beam focus by a diffraction-limited lens. This is given by the expression: We are using a lens with focal length f=75cm and w 01 =2.77 mm (which is measured by integrating the residual intensity that is measured by translating a knife edge across the beam (Fig. 2). With this background on Gaussian optics, we now discuss the technique of Sheik Bahei et. al. 6 in the following section. b. The Z-scan technique The technique introduced by Bahaei et. al 6 is now popularly known as the z-scan technique 18 as it involves the motion of the sample in the sample across the focal point of laser beam along the direction of propagation of the laser beam (Fig. 3). Assuming Gaussian beam optics as discussed in the previous section, this experiment allows an intensity scan of the irradiated sample, and provides information about the nonlinearity in the sample. The typical z-scan is performed by translating the sample along the z axis from one side of the focus to the other ( fig.3). This results in changing the spot size of the incident beam on the sample to a minimum at the focus and then increasing again on crossing the focus. Correspondingly, the intensity of incident light increases on approaching the focus till a maximum at the focus is reached and then reduces on moving away from the focus. Thus, the overall purpose of the experiment is to determine the variation in transmission as the incident intensity changes by translation along the z-axis. The change in the transmittance of the focusing Gaussian beam in a medium is recorded as a function of position of medium. The transmitted beam is collected either completely (which is called the open aperture case) or through a finite aperture (A) as shown in Fig.3. Let us first discuss schematically, a simple case of a thin sample with negative nonlinear refractive index when the aperture is closed (A=0.5, which means just 50% of the beam passes through the aperture). When it moves in the z direction it can act as a thin lens with variable focal length. If we start the scan from -z (far from focal length), where the nonlinear refraction is negligible, the transmittance remains relatively constant. As the sample moves closer to the focus, the beam irradiance increases because of selffocusing of the beam will tend to collimate the beam and cause a beam narrowing at the aperture which results in an increase in the measured transmittance ( fig. 4a). As the scan continues and sample passes the focal plane, the self-defocusing phenomena will occur. This will broaden the beam at the aperture and a corresponding decrease in transmittance will continue until the sample reaches +z (that is sufficiently far from focus) such that the transmittance becomes linear. If we open the aperture (A=1) and do the same scan again from -z direction, the transmittance will increase till focal point and as discussed above, it will decrease to the linear case when the sample moves away from focal point to the +z direction (Fig.4c). Thus, the open aperture case scan gives information on purely absorption nonlinearity while a close aperture case scan contains information about the absorption and dispersion nonlinearity. In case of materials with positive refractive index the story is the reverse of the above cases ( Fig.4b and Fig.4d). Induced beam focusing and defocusing of this type have been observed during nonlinear refractive measurement of some semiconductors 10,11 . Let us now consider the above qualitative discussion mathematically. We consider a sample with third order nonlinearity where the index of refraction is equal to: where n 0 is the linear refraction index, E is the peak electric field (derived in Eq. (17) ). I is the irradiance of the laser beam within the sample, n 2 and γ are related through the conversion formula n esu n c m W Since our sample is thin we can approximate the Gaussian beam is parallel inside the sample. We want to calculate the phase shift of the beam when it passes through the sample. The amplitude I and phase of electric field in the slowly varying envelop approximation as a function of z′ (propagation depth in the sample), are given by two pair equation 6 : ∆Φ , the on axis (r =0) phase shift at focus (z =0) is defined as : where , L is sample length and α where E z r t in ( , , ) is the same as in Eq. (17). Now we are going to derive the electric field in aperture. A method which is called "Gaussian decomposition" (GD) and is given by Weaire et. al. 12 can be used to obtain the far field pattern of the electric field at the aperture plane. They decompose the E out into a summation of Gaussian beams through a Taylor series expansion therefore from Eq. The GD method is very useful for small phase distortions detected with Z-scan therefore only few terms of Eq.(32) are needed. Now we can calculate transmittance power through the aperture: Including the pulse temporal variation, the normalized Z-scan transmittance can be calculated as where P t w I t i ( ) ( ) / = π 0 2 0 2 is the instantaneous input power (within the sample) and is the aperture linear transmittance (w S r a a = − − 1 2 2 2 exp( / ) w a is the beam radius at the aperture). In above discussion we have assumed the effect of third order nonlinearity only and that no absorptive nonlinearity effects that arise from multiphoton or saturation absorption exist. Multiphoton absorption suppress the peak and enhance the valley, while saturation produce the opposite effect 6,9 . c. The Dual-Beam technique Shen et al. [12][13][14] introduced a pump probe laser scheme under mode-mismatched and mode-matched conditions to make sensitive TL measurements. In such dual beams experiments, one of laser beams is essentially probing the effect of the TL caused by the pump beam by scanning across its focus. This results in an effective z-scan of the probe beam across a TL generated by a focusing pump beam. The closed aperture case of this scenario is shown schematically in Fig.5 which has been used by Marcano et. al. 15 to measure the absorption coefficient of water with high accuracy. Mathematically, as in Ref. [15], we can also use the expression of Shen et. al. [12][13][14] , who have derived an expression for the TL signal using diffraction approximation for Gaussian beams in steady state case as: Both continuous and pulsed lasers have been effectively used for z-scan experiments that have relied on these mathematical principles discussed here 19 . These discussions in this section form the basis of our present work that we present hereafter. We explore the open-aperture dual beam TL experiments and achieve our single experimental setup to achieve high sensitive measurements. III. Present Work Our experiments are variation from the conventional z-scan discussed in the above section. We not only use the single beam technique as mentioned in the previous section, but with very simple changes in the experimental set-up, make measurements corresponding to the dual beam z-scan experiments. We will now concentrate more on our actual experimental scheme and the results and discussions arising thereafter. a. Experiment Our experimental scheme involves a sub-100 femtosecond mode-locked Er:doped fiber laser (IMRA Inc.) operating at a repetition rate of 50MHz and provides the fundamental (1560nm) and its second-harmonic wavelength (780nm) simultaneously as a single output. The pulse characteristics of the laser pulses are shown in Fig.6. Either we use both the wavelengths from the laser simultaneously or separate the two copropagating beams with the help of a dichroic beamsplitter and use each of them independently. We scan the sample through the focal point of a 75cm focusing lens and this allows a smooth intensity scan for either/or both of the wavelengths. Care has been taken to make sure that there is no effect of the laser beams on the cuvette alone by conducting an empty cuvette experiment. A silicon photodetector (Thorlab: DET210) is used for the 780nm beam detection, while an InGaAs photodetector (Acton Research) is used for the 1560nm beam detection. We find that the 1560nm beam produces changes in the relative transmission of the laser beam at different intensities as the sample is scanned through the lens focus depending whether we collect all the light or only central 40% of the transmitted light (Fig.7). These results in Fig.7 essentially represent the z-scan technique of Sheik Bahaei et al. 6 to measure the real and imaginary parts of the third-order optical nonlinearity ( ). However, the 780nm beam does not produce any effect even at our peak powers at the focal point of the laser as is expected from negligible absorption at 780nm (Fig. 7c). This enables us to use the 780nm wavelength as the non-interacting probe beam for the subsequent dual-beam experiments where we use both the wavelengths from the laser simultaneously. Since our 75cm lens focuses the 780nm probe beam to its minimal spot size position 0.4mm ahead of the pump beam of 1560nm, this is a mode-mismatched pump-probe experiment. However, the focal spot size of 9µm for 780nm is 15µm smaller than the corresponding 1560nm spot size at its own focus and from Eq.(24) the Rayleigh range for 780nm is 0.32mm and for 1560mm it is 1.15mm. Thus, the 780nm laser volume is always confined within the 1560nm laser beam volume when both the beams are used simultaneously from the laser and can act as an effective probe. Fig.8a shows the case when the entire transmitted probe (780nm) beam is being collected and this essentially depicts the saturation environment created by the pump (1560nm) beam. This statement is further reinforced by Fig.8b, where the case of 1560nm beam alone from Fig.7a is plotted along with the results in Fig.8a. Essentially, as the 1560nm beam starts to saturate the sample at its focal point, the 780nm beam also experiences a saturated environment, whereby its transmittance increases at its focal point and shows an identical transmission behavior although the signal level is two orders of magnitude lower. Such a result indicates the thermal capacity of water that can affect the spectroscopic behavior of water. Finally, Fig.8c represents the thermal lens effect of the pump beam resulting in a temperature gradient, which in turn produces a spatial gradient in refractive index which is depicted in the relative change in transmittance of the probe beam. Such thermal lensing (TL) effect can be used to determine the absorption coefficient (α 0 ) of the sample at the pump wavelength very accurately 15 . The solid line in the fig.8c is the result of a theoretical fit to Eq.(36). This fit gives the value of phase shift, θ = 9.957, which when substituted in Eq.(37) with the parameters and for pure water Our experimental results discussed above also enables us to determine the nonlinear absorption coefficients of water. For nonlinear materials the index of refraction n is expressed in terms of nonlinear n 2 or γ through the relation: In Fig.5b, the valley-peak structure representing the 40% closed-aperture data for 1560nm suggests a self-focusing effect inside the sample. The Ryleigh range (Z(r)) for 1560mm is 1.15mm. From fig. 5b, valley to peak separation at 1560nm is 4mm, which is 3.4×Z(r) indicating that all the effects at 1560nm are thermal in nature 24 . So we use the Gaussian Decomposition method to fit this closed aperture z-scan data quite convincingly ( fig. 5b, solid line fit to the raw data), and we obtain γ=1.57× 10 -3 cm 2 /GW, which is proportional to n 2 =4.9×10 -12 esu. Thus α 0 , β and n 2 values of the water sample at 1560nm wavelength are determined. Finally, we use the theoretical expression for the thermal lens of the pump beam given by Eq.(38) to fit the experimental data (Fig.6a) (a) A plot of the intensity of the light (along y-axis) coming to the photodiode as the translating knife-edge (distance along x-axis) lets out more and more of the incident laser beam into the photodiode. (b) A derivative of the data in Fig. 2a gives the beam size in the lower plot, which fits to a Gaussian beam waist of 2.77mm. The pulsewidth of the 780nm pulse measured through a non-collinear autocorrelation using a speaker as the delay arm into a secondharmonic BBO crystal, which is detected into a PMT (Hamamatsu 1P28). The Gaussian fit to the autocorrelation trace provides a pulsewith of 90fs for the 780nm laser pulses. (c) The spectra at the center wavelength of 1560nm measured through the SP-150 monochromator into an InGaS detector (both Acton Research Co.) (d) The pulsewidth of the 1560nm pulse measured through collinear cross-correlating with the 780nm beam into a second-harmonic BBO crystal, which is detected into a PMT (Hamamatsu 1P28). Using a Gaussian fit and deconvoluting the 90fs pulse of 780nm results in a pulsewidth of 120fs for the 1560nm pulse. (b) Measured z-scan of a 16mm thick double distilled water using 95fs pulses at λ=1560nm (diamond) and theoretical fit (solid line) for 40% closed aperture (experimentally measured transmittance is normalized to unity). (c) Water spectra covering the 780nm and 1560nm range of wavelengths. Figure 8. (a) Measured z-scan transmittance of 80fs pulses of λ=780nm as a probe through a 16mm thick double distilled water being irradiated with 95fs of λ=1560nm as pump (diamond) and theoretical fit (solid line) in fully open aperture (raw transmittance data is presented to illustrate sensitivity of our measurements). (b) Replot of Fig. 5a of 1560nm alone and Fig. 6a of 780nm probe measurements for 1560nm pump case to show that both essentially have same features except their two orders of magnitude difference in their signal levels. (c) Measured z-scan transmittance of 80fs pulses of λ=780nm as a probe through a 16mm thick double distilled water being irradiated with 95fs of λ=1560nm as pump (diamond) and theoretical fit (solid line) in 40% closed aperture (experimentally measured transmittance is normalized to unity).	2019-04-14T03:13:18.818Z	2004-01-01T00:00:00.000	{ "year": 2004, "sha1": "2ca5114f3a77ef26977f78b3d7e177f73da14757", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "2ca5114f3a77ef26977f78b3d7e177f73da14757", "s2fieldsofstudy": [ "Physics", "Materials Science" ], "extfieldsofstudy": [ "Physics" ] }
233668112	pes2o/s2orc	v3-fos-license	Self-Tuning Algorithm for Tuneable Clamping Table for Chatter Suppression in Blade Recontouring : The production and repair of blades for aerospace engines and energy turbines is a complex process due their inherently low stiffness and damping properties. The ﬁnal recontouring operation is usually performed by milling operations where regenerative chatter is one of the main productivity limiting factors. With the objective of avoiding speciﬁc stiffening ﬁxtures for each blade geometry, this paper proposes a semi-active tuneable clamping table (TCT) based on mode tuning for blade machining. The active mode of the device can be externally controlled by means of a rotary spring and eddy current damping modules. Its in-series architecture allows damping to be introduced to the critical mode of the thin-walled part without any direct contact in the machining area and enables a more universal clamping. Its chatter suppression capabilities are maximized by means of a novel self-tuning algorithm that iteratively optimizes the tuning for the measured chatter frequency. The beneﬁts of the iterative algorithm are validated through semidiscretization and initial value time-domain simulations, showing a clear improvement in blade recontouring stability compared to regular broad-bandwidth tuning methods. Introduction The manufacturing and repair of fan, compressor and turbine blades constitute very high added-value industrial processes with a high relevance in the aerospace manufacturing industry. Due to the continuous increase in raw material and manufacturing costs, the specific weight of maintenance, repair and overhaul (MRO [1,2]) processes of engine parts has been increasing in recent years, which in some applications can result in savings of up to 70% of the total costs compared to disposal and replacement with a remanufactured component [3]. Among the four main stages of the blade repair process (pre-inspection, welding, machining and post-inspection [4]), the machining or recontouring stage is of particular interest due to the specific problems that arise during these operations. The aim of the recontouring stage is to remove, by machining operations, the excess material added in the welding stage (see Figure 1a,b) using additive manufacturing technologies such as laser cladding [5] or laser metal deposition [6] and achieve the final blade geometry. Although these operations have traditionally been performed through manual belt grinding, the need to automate the process to the extent possible has encouraged the engine part maintenance industry to move towards milling-based recontouring (see Figure 1c), performed on 5-axis machines or hybrid cells in which welding and machining operations are performed in the same clamping [7,8]. In any case, milling operations on thin-walled blades are always a complicated task due to static deflections and, especially, the occurrence of regenerative chatter. The inherently low stiffness and damping properties of the thin-walled blades result in a large dynamic flexibility at the points to be recontoured and largely promote the onset of chatter. Chatter is a type of self-excited vibration caused by the regenerative effect, whereby each cutting edge cuts the wavy pattern left by the vibration in the previous pass [9,10]. Under certain cutting parameters, this closed-loop relationship can become unstable and result in deep surface marks that can lead to rejection of the machined blade, accelerated tool wear and damage to machine tool components. Chatter suppression can be tackled in different ways [11], such as spindle speed tuning based on a stability lobe diagram (SLD), optimized tool geometries [12] or continuous spindle speed variation [13]. However, the difficult-to-cut titanium, nickel and cobalt alloys commonly used for the blades and the uneven and varying dynamics of blades greatly restrict the applicability of process parameter tuning techniques. The optimization of tool or process parameters may not be effective when the dynamics of the part are altered, or a severe decrease in tool life may occur if the spindle speed is outside the narrow recommended cutting speed range. Increasing the dynamic stiffness of the blade by means of stiffening devices or dampers provides a different approach to reduce the tendency to regenerative chatter without actually changing the process parameters. [6]. (b) Excess of material to be removed [6]. The application of stiffening devices based on the clamping of the blade near the tip (see Figure 1d) is the most widespread technique at industrial level to prevent chatter in blade MRO processes. To cope with different blade geometries and dimensions, a wide set of fixtures is required, which must also have a certain level of flexibility to cope with the shape deviation of the damaged blade and not to cause an excessive static deflection. In this regard, the use of low melting point alloys as pinpoint has also been investigated as a more flexible stiffening alternative [14]. On the other hand, active and passive dampers can also be implemented to increase the dynamic stiffness of the blades. Active dampers can improve the dynamic stiffness of the part by counteracting the measured vibration by means of an actuator. This idea has recently been implemented to successfully damp flexible parts during heavy duty milling operations on thin walls [15]. Besides, passive damping solutions in the form of tuned mass dampers (TMDs) [16] or Lanchester dampers [17], have lately been applied for chatter avoidance in blade and similar thin-walled part milling. Nevertheless, the application of the technologies developed so far to blade recontouring may not be feasible, as they require direct attachment of the device close to the machining point, which may interfere with the machining trajectory or pose a major difficulty when dealing with blades with complex geometries. As opposed to the in-parallel architecture of regular TMDs, the tuneable clamping table (TCT [18]) is an application case of mode tuning with an in-series architecture, which can increase the damping of the critical mode by tuning a highly damped and controlled mode close to the targeted mode in the frequency band. Its in-series architecture provides the capability to achieve the tuning between the controlled mode and the critical mode of the blade with a universal clamping system and without any direct intervention in the cutting zone. Nevertheless, in mode tuning devices fine tuning is essential to achieve a maximum increase in dynamic stiffness. To this effect, the algebraic expressions for optimal H ∞ tuning for a TCT have already been developed [18], which are the equivalent-but different-to classical TMD amplitude tuning methodologies [19,20]. When the aim is to maximize milling stability, the aforementioned strategies are no longer optimal [21], and specific tuning procedures based on the optimization of the real part of the receptance function have to be applied [22,23]. In any case, whatever the objective function, the aim of these strategies is to achieve a wide-bandwidth tuning, considering that neither the part dynamics nor the frequency of the external excitation are subject to change during normal operation. In an industrial environment where a wide variety of parts-and therefore, dynamicsexists and therefore a high degree flexibility and automation is sought, the TCT provides a blade clamping solution to mitigate chatter vibrations with a universal clamping system and minimal interference with the cutting process. This paper presents a self-tuning algorithm for the tuneable clamping table for efficient blade manufacturing and repair. Benefiting from the possibility of externally tuning the frequency and damping of its active mode, its iterative nature allows the device's chatter suppressing capability to be exploited to the maximum extent and also provides the capacity to adapt to blades with uneven dynamics. The analytical expressions for iterative tuning are introduced, which are implemented in an algorithm running in real-time during the milling process. The algorithm is then validated through semidiscretization and initial value time-domain simulations, showing a superior performance compared to regular wide-bandwidth tuning methods. Semi-Active Tuneable Clamping Table: Concept and Benefits Similarly to TMDs, the tuneable clamping table proposed for blade recontouring is based on mode tuning. When tuning the controlled mode close in the frequency band to the lowly-damped critical mode of the blade, the two modes interact and the shapes of the neighboring modes are combined. Due to the existence of the damping of the device, a partial transfer of damping between the ideal (stand-alone) modes is achieved, which can be used to increase the dynamic stiffness of the system. Nevertheless, the classical TMD and TCT concepts present significant differences. Figure 2 shows the differences in architecture between the well-known tuned mass damper and the tuneable clamping table. As opposed to the TMDs, where the controlled mode is placed in-parallel to the excitation force flow (Figure 2a), the TCT approach adopts an in-series configuration (Figure 2b), clamping the blade from below. In this way, mode tuning and the subsequent increase of the dynamic stiffness at the machining point can be achieved without physically attaching any device to the blade near the machining zone that could interfere with the cutting process. Moreover, this permits the development of a universal clamping system for the entire blade set without the need for an array composed by specific fixtures or dampers for every blade geometry. In order to cope with a wide set of blades with different dynamics, the TCT concept is implemented in a semi-active form in this paper. Semi-active damping devices, first introduced in [24] for chatter suppression, require external energy to achieve the tuning, but unlike active devices, the dissipation of vibration energy takes place without any external energy input. The alteration of the stiffness, damping or mass properties of the device mode can be achieved by multiple means. Purely mechanical assemblies based on preloaded elastomers [24], position-adjustable moving masses [25,26] or rotary springs [27][28][29][30] have been proposed to control the stiffness of the damper. Regarding the damping, eddy current modules have been proven to be a reliable alternative for obtaining a nearly-ideal and adjustable viscous damping [18,28,29]. The use of electrorheological and magnetorheological fluids to simultaneously control the stiffness and damping of the device has also been investigated [31], with application over vibration attenuation in thin walled part milling as well [32,33]. Finally, it is also worth mentioning the use of piezoelectric shunts for semi-active damping purposes [34][35][36]. Indeed, regardless of how the mode is controlled, semi-active devices require a control strategy to achieve the tuning of the device mode to the critical mode to be damped. In this sense, velocity [34] or force [37] feedback control strategies, as well as fuzzy control algorithms [38,39], can be found in the literature. Nonetheless, when it comes to suppressing regenerative chatter -rather than general vibration problems-iterative algorithms based on the measured chatter frequency show a higher degree of effectiveness [29,40]. In the case of the semi-active TCT presented here, the controlled mode is materialized as a dominant translational mode with a fixed moving mass m 1 guided by flexures of stiffness k f . The table features a rotary spring (see Figure 3) with unequal stiffnesses k s,1 and k s,2 in its two main directions and driven by a motor-encoder assembly. This permits externally modifying the stiffness k 1 and, consequently, the natural frequency ω 1 of the device with the angular position of the spring θ as In this manner, blade modes within the frequency range [ω 1,min , ω 1,max ], as shown in Figure 3, can be tuned with the TCT. The viscous damping of the table is provided by the eddy current damping modules under the table, whose damping level can be adjusted by modifying the immersion of the conductive plates inside the magnetic field created by permanent magnets. Further details on its mechanical design and conception can be found in [18]. Finally, once the design and construction of the TCT has been defined, the objective is to develop a control strategy seeking for the maximization the chatter suppression capabilities of the TCT. Similar to the aforementioned strategies for TMDs [29,40], an iterative self-tuning algorithm for the in-series architecture based on the measured chatter frequency is proposed here. The development and implementation of said algorithm is explained in the following sections. Unidirectional Milling Model With Tuneable Clamping Table Tip recontouring processes mostly consist of one or several milling passes parallel to the so-called chord line of the blade. Like many other parts with thin wall characteristics, blades usually present a dominant modal direction normal to the chord line and perpendicular to the feed motion of the cutter, along which the active direction of the TCT is placed for maximum efficiency of the mode tuning. Accordingly, the whole milling dynamics can be accurately described in this case by a unidirectional milling model as in Figure 4. Structural Dynamics Similarly to other thin-walled parts, the presence of very lowly-damped modes that are well-separated over the frequency band is usually expected, which results in a negligible level of coupling between the different blade modes. Under these circumstances, the dynamic behavior of the blade at the machining point nearby a targeted (dominant) mode can be accurately represented by a 2-degree of freedom (DOF) lumped mass model as in Figure 4. m 1 , k 1 and c 1 represent the standalone active mode of the device and m 2 , k 2 and c 2 define the standalone targeted mode of the blade at the machining point, leading to where Regenerative Milling Force For simplicity, a cylindrical and straight fluted end mill with Z regularly spaced teeth rotating at a constant spindle speed Ω is considered. Assuming a cutting force model with linear characteristics with respect to the chip thickness h, the specific cutting force vector f = F/a in the tangential-radial plane (t, r) can be defined as where K e = [K te , K re ] and K c = [K tc , K rc ] are respectively the so-called edge and cutting coefficient vectors resulting from the linear fitting to the experimentally measured mean cutting forces [41]. Due to the regenerative effect, each cutting edge cuts the wavy surface left by the vibration of the previous tooth. Hence, the momentary chip thickness can be calculated by approximating the tool edge kinematics to a circular motion as where τ is the regenerative delay, which is equal to the tooth passing period T Z = 2π ZΩ in this case, and where stands for the angular position of the ith tooth at the instant t. The total cutting force F 2 to which the blade is subjected to for an axial depth of cut a can be hence calculated by summing the projections onto the flexible direction of all the Z teeth as where g(ϕ) represents the screen function which determines whether the tooth is in or out of cut. The total cutting force can be divided into two terms as The former is the time-periodic stationary forcing term which represents the cutting forces of the equivalent undeformed milling model. This term induces an unavoidable forced vibration x s (t). The latter is the state-dependent variational term of the cutting force. It arises due to the regenerative nature of the cutting force and is calculated as where is the time-periodic directional milling coefficient, which concentrates the projection of the cutting force onto the mode direction and the projection of the vibration onto the chip thickness direction. The presence of the regenerative-or dynamic-term of the cutting force makes the governing equation a time-periodic delay differential equation (DDE [42]). Under a certain set of cutting parameters, the DDE can undergo Hopf or period-doubling (flip) instabilities, which actually correspond to the large amplitude chatter vibrations. Zeroth-Order Milling Stability The asymptotic linear stability of the stationary solution of (2) can be assessed by introducing a perturbation p(t) around the stationary solution x(t) = x s (t) + p(t). For the case that the perturbed solution p(t) is not influenced by the stationary cutting solution, the stability of (2) can be evaluated by studying the stability of the perturbed equation: Numerical methods such as semidiscretization [43] or full-discretization [44] are wellestablished and efficient techniques for assessing the stability properties of time-periodic DDEs. However, like other numerical methods, they often hide the direct relations between process parameters and system stability that could serve as a core function for a stability maximization algorithm. Instead, the stability of (12) can be studied in the frequency domain by considering the marginal case where the system has a critical dominant vibration (chatter) frequency ω c and its modulations ω c + k ω Z related to the tooth passing frequency together with the Fourier decomposition of the time-periodic milling coefficient In case of only being interested in Hopf-type instabilities, (13) can be truncated to k = 0 and the directional milling coefficient approximated by is averaged value is the mean directional factor [45]. This approximation-commonly known as the zerothorder algorithm (ZOA [46])-leads to an eigenvalue problem from which the critical stability curves can be parametrically computed. ZOA provides accurate results in noninterrupted milling processes where only Hopf bifurcations are expected and the effect of machining mode coupling is negligible [47]. In a unidirectional milling case like this, the eigenvalue problem is reduced to the following scalar characteristic equation: where H 22 refers to the direct receptance function at the blade tip originated from In this way, a semi-analytical expression of the parametric root crossing (Ω l (ω c ), a(ω c )) curves can be raised as and where l = 1, 2, 3, ... stands for the lobe number and Re H 22 and ψ 22 are, respectively, the real part and phase of the aforementioned direct receptance function at the blade tip. Semidiscretization Based Milling Stability Alternatively to ZOA, the stability of the time-periodic DDE in (12) can be studied by means of time-discrete methods such as semidiscretization. Through the application of the Floquet theory, the semidiscretization method provides the finite-dimensional version of the monodromy matrix [48] for discrete values of Ω and a, whose eigenvalues-Floquet multipliers-are used as a measure of the system stability. If all characteristic multipliers remain inside the unit circle (modulus less than one), then the stationary solution is asymptotically stable, otherwise, it is unstable. In this manner, by setting a fine mesh of a and Ω pairs, stability maps of highly interrupted milling cases exhibiting Hopf and period doubling instabilities can be constructed. Chatter Suppressing Strategy for the TCT Architecture In the previous section, the semi-analytical expressions relating the system dynamics and milling stability for a blade recontouring case with unidirectional dynamics have been introduced. Taking this relation into account, the objective of the present section is to develop a tuning strategy for the tuneable clamping table to get the most out of the mode tuning for maximum milling chatter suppression. Dimensionless Formulation of the System Dynamics As observed in (18), in milling processes with a dominant flexible direction, the limiting depth of cut without chatter is approximately given by the inverse of the oriented receptance function β 0 Re H 22 [21]. Therefore, the mean directional factor β 0 determines whether the negative real part of H 22 (positive directional factor, β 0 > 0) or the positive real part (negative directional factor, β 0 < 0) describes the limiting depth of cut curves. In milling β 0 is most likely to be positive, although negative cases are also possible depending on the radial engagement limits [ϕ en , ϕ ex ]. In any case, the complex receptance function of the TCT-blade assembly at the machining point H 22 can be expressed as a function of the lumped-mass parameters of the 2-DOF system in Figure 4 as As can be noticed, the internal viscous damping of the blade c 2 has been neglected, which leads to a good approximation due to the inherently low damping properties of thin-walled parts. In order to obtain a general tuning formulation for whatever mass and stiffness of both TCT and blade, the receptance function in (19) is transformed into its complex dimensionless form through the dimensionless parameters collected in Table 1. Then, the real part of the dimensionless receptance function, which is the one driving the stability limit, can be obtained as follows: Excitation Frequency Dependent Tuning for the TCT At this point, there are different optimization procedures to maximize the negative part of (21)-or minimize its positive part. The traditional approach, in which constant device tuning is provided no matter what the frequency of the external excitation g is, benefits from the presence of three invariant points in (21). In this manner, a wide-bandwidth optimization of either negative or positive real parts can be achieved as follows: first, the two neighbouring invariants of the corresponding side of the curve are matched and, then, a device damping for a zero-tangent at the mentioned invariants is set. The application of this strategy to the TCT architecture leads to the analytical expressions for constant frequency ratio and device damping tunings, respectively. A more detailed explanation of the procedure for obtaining these expressions can be found in [23]. Constant parameter tuning strategies offer a good solution for fixed dampers or for dampers whose stiffness and damping cannot be externally altered in real-time. However, in semi-active devices like the TCT, the stiffness of the controlled mode can be externally controlled in real-time by means of a rotary spring driven by a motor and an encoder. Therefore, if the frequency of the external excitation g is known, a more efficient tuning strategy based on that frequency can be devised [29,40]. Attending to (18), the arising dominant chatter frequency ω c can be considered as sampling on the real part of the receptance function. Thus, a general chatter frequencydependent optimal tuning can be developed by optimising the negative or positive real part of h 22 for every dimensionless frequency g = ω c /ω 2 . This can be achieved by simply deriving Re h 22 with respect to the frequency tuning f and setting it equal to zero as for every value of the dimensionless frequency g and standalone table damping ζ 1 . Solving (23) leads to computing five roots: one at f 0 = 0 and four others symmetrically arranged with respect to f 0 . By just considering the positive tuning values, the dimensionless (chatter) frequency-dependent tuning formula for positive (β 0 > 0) and negative (β 0 < 0) directional factors is attained. The expression (24) presents the tuning limits g lim,− = 1 and g lim, outside of which the tuning can actually be applied. When machining close to the resonance of the standalone blade system, dimensionless chatter frequencies close to g = 1 are expected. This would require an unreachable frequency tuning in practice according to (24), or that may not even exist if g ∈ [g lim,− , g lim,+ ], which is more likely to occur under a positive directional factor. Nonetheless, it is important to remember that values of the real part of the receptance higher or lower than zero do not generate lobe structures for positive and negative directional factor cases, respectively. Based on this, an alternative chatter frequency dependent strategy can be devised by simply setting the real part of the receptance function to be zero for every dimensionless chatter frequency, that is, Hence, considering that each of the strategies has complementary tuning domains and that for g > 1 the directional factor must be positive and negative for g < 1, a frequency dependent tuning strategy can be raised by defining the following piecewise smooth function: The curves of the optimal frequency tuning ratios depending on the dimensionless frequency are depicted in Figure 5 together with the values of the broad-bandwith (frequency independent) strategies in (22). In practice, the tuning is only realizable in a region limited by f min and f max given by the stiffness range limits of the rotary spring (see Figure 3). Comparison with Equivalent Constant Parameter Tuning Strategy The effect of the proposed tuning strategy can be appreciated by plotting Re h 22 optimally tuned for every value of g according to f o (g) in (27), namely Re h 22 ( f o (g), g), and comparing it to the case with constant tuning f o,± from (22) as in Figure 6a. Here, for comparison purposes, the optimal damping ratios from the corresponding constant tuning strategy in (22) have been set. The excitation frequency dependent strategy further optimizes the wide-bandwidth tuning strategy proposed in [23], except for a single value of g where f o (g) and f o,± obviously coincide. frequency ratio, dimensionless spindle speed, ν(g) dimless depth of cut, σ(g) The effect of the introduced strategy on the milling stability can be observed by computing stability limit given by the parametric curves of the dimensionless ZOA case as where σ(g) and ν(g) are respectively the dimensionless depth of cut and spindle speed. Figure 6b shows the dimensionless stability limits given by the piecewise smooth function Re h 22 ( f o (g), g) and the ones subjected to constant tuning Re h 22 ( f o,± , g). As can be observed, the stability is enhanced for both positive and negative directional factor cases over the whole range of spindle speeds, except for the aforementioned points where the two strategies converge. As noticed in [29], the 'double lobe' shape of the stability diagram becomes a 'single lobe' structure. Finally, apart from the mentioned increase of the stability limit, a variation with respect the location of the maximum stability asymptotes g a is also noticed. These stability asymptotes are directly linked to the dimensionless spindle speeds at which the so-called 'sweets spots' take place by the relation ν a = g a /l, being l = 1, 2, 3, ... the lobe number. These dimensionless resonant frequencies can be numerically computed by simply Re h 22 (g a ) = 0. The curves of the optimal dimensionless frequency for the frequency dependent strategy presented in this paper are shown in Figure 6c for increasing mass ratio. Accordingly, by implementing the presented tuning and machining at spindle speeds calculated by g a , the maximum stability is guaranteed. Iterative Tuning Algorithm for the TCT The excitation frequency dependent tuning strategy leads to a theoretical increase of the stability limits by optimising the system dynamics for each chatter frequency ω c . Obviously, the direct application of this tuning would imply that the chatter frequency is known beforehand for every value of the spindle speed Ω. In practice, the tuning will only take place once the process is unstable, and for a certain spindle speed Ω the arising unstable frequency ω c will vary with the tuning f , since the dynamics of the TCT-blade assembly vary with every tuning. Consequently, the 'final' ω c cannot be known in advance. The practical realization of the previously developed tuning strategy must therefore be attained by an algorithm in which the optimal tuning is sought by iteratively measuring the chatter frequency and performing the tuning until the convergence is achieved. The flowchart of the algorithm proposed for the TCT is shown in Figure 7, which is implemented in a real-time controller. Regardless of the machining parameters, the TCT is initially tuned according to the broad-bandwidth tuning strategy in (22). Once the machining starts, the blade undergoes forced vibration even for the stable cutting situation due to the unavoidable stationary part of (7). The part vibration can be estimated by microphones or other contactless solutions, avoiding sensors that could interfere with the cutting process. Then, the vibration measurement is fed to the real-time controller for the continuous computation of the frequency spectrum. This forced vibration, which manifests itself in the form of harmonics at the tooth passing frequency ω Z = 2π/T Z , does not usually entail a real hazard for an acceptable part quality, except in finishing passes where high surface quality is required. When regenerative vibrations arise, a dominant high amplitude peak at a frequency higher or lower than the standalone natural frequency ω 2 (for β 0 > 0 and β 0 < 0 cases, respectively) can be noticed, leading to deep marks and bad surface quality. Therefore, the real-time controller detects the onset of chatter when the frequency of the maximum vibration peak is not an integer of the tooth passing frequency ω Z . If this is the case, the optimal natural frequency of the table ω 1 is determined according to the measured chatter frequency ω c by setting ω 1 = f o (ω c /ω 2 ) ω 2 , with f o (g) from the expression in (27). The position of the rotary spring is then modified according to (1) to match the required natural frequency. This procedure is iteratively carried out until the machining becomes stable or when a limit of six iterations is reached, if the process cannot be stabilized at all. chatter detection algorithm initial tuning: f i = f o and i = 1 device tuning: However, under certain cutting parameters, the tuning may oscillate and not converge to its optimal value even for a theoretically stabilizable point, as the changes on the assembly dynamics produced by the tuning lead to too large chatter frequency jumps. In these instances, the convergence of the algorithm can be improved as follows: after a threshold of three tuning iterations, the ideal objective tuning is calculated with the expression in (27), but in spite of directly tuning the device to that value, the device is tuned to the mean value of the actual tuning and the calculated ideal tuning, achieving a better convergence. Validation of the Concept The introduced self-tuning algorithm has theoretically demonstrated a higher performance compared to other wide-bandwidth tuning methods. It has been noted that ZOA leads to accurate stability predictions as long as the milling process is not interrupted or only lightly interrupted. Nonetheless, blade recontouring may involve low radial immersion operations performed with cutters featuring few teeth and negligible smoothing effect of the tool helix. This would lead to an interrupted cutting that could favour the occurrence of both flip instabilities and machining mode coupling effects that might compromise the effectiveness of the algorithm. For this reason, the effectiveness of the algorithm presented in this paper is validated in this section by means of the semidiscretization and initial value time domain simulations for the milling case in Table 2, which resembles an interrupted recontouring operation. Milling Stability through Semidiscretization Method The stability and chatter frequency charts for the standalone (stiff) clamping, the TCT under broad-bandwidth tuning and the self-tuned TCT obtained through semidiscretization [43] are respectively shown in Figure 8a,b. For the self-tuned case, the chatter frequency is computed for every unstable point of the Ω-a mesh as in [49], and optimal tuning is iteratively performed according to the algorithm in Figure 7 until the process is stable or a maximum of 6 iterations is reached. The results from the semidiscretization calculations confirm the superior performance of the iterative strategy over the fixed-frequency strategy, with the exceptions of the aforementioned spindle speeds at which the two strategies converge. Moreover, it can also be noticed that the introduced strategy also increases the stability limit for spindle speed ranges where a dominant period doubling behavior may be expected, even if the expressions driving their stability limit differ from those of the Hopf-kind instabilities [50]. Initial Value Time Domain Simulations with Fly-Over Effect So far, an ideal linear milling model has been considered in order to devise the tuning algorithm and validate it through semidiscretization calculations. However, the real milling cases are subdued to the fly-over non-smoothness [51], effect by which the cutting edge jumps out of the cut due to the vibration. Apart from being the cause of the vibration amplitude reaching a threshold instead of growing indefinitely, this strong non-linearity also induces a multi-regenerative effect than can make the chatter frequencies diverge from the ideal linear case [52] and, thus, compromise the effectiveness of the tuning strategy. In order to assess the robustness of the self-tuning algorithm against these deviations, initial value time domain simulations of (2) have been performed together with the selftuning algorithm. With this purpose, (2) is transformed onto its modal state-space form and numerically solved through an exponential time differencing forward Euler scheme [53]. For each discrete time step, the cutting force in (7) is calculated by considering a total of 20 prior tooth passes for the computation of the chip thickness and switching it off whenever fly-over takes place (h i (t) < 0). For each Ω, a point, the chatter detection algorithm is run every 100 tool rotation periods, of which the last 60 periods are taken for the FFT calculation. Chatter is detected and its dominant frequency ω c measured if any peak is noticed at frequencies not matching the tooth passing frequency ω Z and if its amplitude is higher than the 30% of the largest tooth passing harmonic. If this is the case, the tuning of the device ω 1 is computed according to Figure 7 based on the measured chatter frequency ω c , and the system matrices corresponding to the new tuning k 1 = m 1 ω 2 1 are recalculated. Here, it is considered that the tuning mechanism (rotary spring) is agile enough to perform the system retuning without affecting the system stability. Therefore, the dynamics of the tuning mechanism are omitted and the tuning is assumed to take place instantaneously. The process is iteratively repeated until the stability is reached or the tuning goes into an oscillatory situation (machining stability is not possible). The results of the time domain simulations for the A, B, C and D points from the stability chart in Figure 8 are shown in Figure 9. As predicted by the semidiscretization calculations, the algorithm successfully stabilizes the initially Hopf unstable cases. Point B (see Figure 9b), chosen according to the stability asymptote in Figure 6c, confirms that large stability gains can be attained with the TCT by integrating the self-tuning algorithm and adjusting the spindle speed to the stability asymptote. The capacity of the algorithm to enhance the stability in period doubling zones is also demonstrated in Figure 9c, as one case of flip chatter was successfully suppressed by the tuning algorithm. An 'unstabilizable' case can be seen in Figure 9d, where an oscillatory behavior between two unstable tunings is reached. Finally, once the capability of the algorithm to maximize the stability of the system has been demonstrated, it would also be important to study the robustness aspects of the algorithm itself. One of the main hazards for all the mode tuning strategies is the uncertainty with respect to the actual dynamic parameters of the system, namely, initial deviations of the part dynamics or deviations due to the material removal. In this sense, an in-depth analysis of the uncertainty and variability of the system parameters and their effect on the tuning robustness would be decisive for a proper design of devices based on the TCT architecture. ω Conclusions The wide variety of different blade geometries present in a single engine makes blade manufacturing and repairing processes a complex task, since specific stiffening fixtures are needed for each geometry in order to avoid machining chatter when performing the recontouring operations. Similarly to TMDs, the tuneable clamping table can damp the critical modes by means of mode tuning, but thanks to its in-series configuration, the tuning can be achieved without interfering with the cutting area, enabling a more universal clamping. Aiming at maximising its chatter suppression capabilities, a selftuning algorithm has been developed for TCT architecture. The algorithm iteratively measures the frequency of the arising chatter and optimally tunes the device for that frequency, outperforming other wide-bandwidth tuning strategies. The effectiveness of the algorithm has been validated through semidiscretization calculations and initial value time domain simulations, where the stability enhancement provided in the ideal linear case has been validated. The industrial application of the algorithm would require studying the robustness of the algorithm with respect the uncertainty of the part dynamics due to initial deviations or material removal. Conflicts of Interest: The authors declare no conflict of interest. Abbreviations The following abbreviations are used in this manuscript:	2021-05-05T00:09:29.812Z	2021-03-13T00:00:00.000	{ "year": 2021, "sha1": "e63b2be926eb8203329b2128f7cb8568ee696473", "oa_license": "CCBY", "oa_url": "https://www.mdpi.com/2076-3417/11/6/2569/pdf", "oa_status": "GOLD", "pdf_src": "Adhoc", "pdf_hash": "48d2dd29d6d15cb1429380819584195890937f65", "s2fieldsofstudy": [ "Engineering" ], "extfieldsofstudy": [ "Computer Science" ] }
55437685	pes2o/s2orc	v3-fos-license	Exact Electronic Potentials in Coupled Electron-Ion Dynamics We develop a novel approach to the coupled motion of electrons and ions that focuses on the dynamics of the electronic subsystem. Usually the description of electron dynamics involves an electronic Schr\"odinger equation where the nuclear degrees of freedom appear as parameters or as classical trajectories. Here we derive the exact Schr\"odinger equation for the subsystem of electrons, staying within a full quantum treatment of the nuclei. This exact Schr\"odinger equation features a time-dependent potential energy surface for electrons (e-TDPES). We demonstrate that this exact e-TDPES differs significantly from the electrostatic potential produced by classical or quantum nuclei. The theoretical description of electronic motion in the time domain is among the biggest challenges in theoretical physics. A variety of tools has been developed to tackle this problem, among them the Kadanoff-Baym approach [1], time-dependent density functional theory [2], the hierarchical equations of motion approach [3] as well as the multiconfiguration timedependent Hartree-Fock approach [4]. From the point of view of electronic dynamics all these approaches are formally exact as long as the nuclei are considered clamped. However, some of the most fascinating phenomena result from the coupling of electronic and nuclear motion, e.g., photovoltaics [5], processes in vision [6], photosynthesis [7], molecular electronics [8], and strong field processes [9]. To properly capture electron dynamics in these phenomena, it is essential to account for electron-nuclear (e-n) coupling. In principle, the e-n dynamics is described by the complete time-dependent Schrödinger equation (TDSE) HΨ(r, R, t) = i∂ t Ψ(r, R, t), with Hamiltonian H =T n (R) +V n ext (R, t) +Ĥ BO (r, R) +v e ext (r, t), (2) whereĤ BO (r, R) is the traditional Born-Oppenheimer (BO) electronic Hamiltonian, H BO =T e (r) +Ŵ ee (r) +Ŵ en (r, R) +Ŵ nn (R). ∇ 2 j 2m are the nuclear and electronic kinetic energy operators,Ŵ ee , W en andŴ nn are the electron-electron, e-n and nuclearnuclear interaction, andV n ext (R, t) andv e ext (r, t) are time-dependent (TD) external potentials acting on the nuclei and electrons, respectively. Throughout this paper R and r collectively represent the nuclear and electronic coordinates respectively and = 1. A full numerical solution of the complete e-n TDSE, Eq. (1), is extremely hard to achieve and has been obtained only for small systems with very few degrees of freedom, such as H + 2 [10]. For larger systems, an efficient and widely used approximation is the mixed quantum-classical description where the electrons are propagated quantum mechanically according to the TDSE which is coupled to the classical nuclear trajectories, R α (t), determined by Ehrenfest or surface-hopping algorithms [11]. The potential V (r, t) felt by the electrons is then given by the classical expression where R(t) denotes the set of classical nuclear trajectories, R α (t). A better approximation to the potential V (r, t) felt by the electrons is the electrostatic or Hartree expression [11]: where χ(R, t) represents a nuclear many-body wavefunction obtained, e.g., from nuclear wave packet dynamics. Clearly, Eq. (6) reduces to the classical expression (5) in the limit of very narrow wave packets centered around the classical trajectories R(t). The Hartree expression (6) incorporates the nuclear charge distribution, but the potential is still approximate as it neglects e-n correlations. In this paper we address the question whether the potential V (r, t) in the purely electronic many-body TDSE, Eq. (4), can be chosen such that the resulting electronic wavefunction Φ(r, t) becomes exact. By exact we mean that Φ(r, t) reproduces the true electronic N e -body density and the true N e -body current density that would be obtained from the full e-n wavefunction Ψ(r, R, t) of Eq. (1). We shall demonstrate that the answer is yes provided we allow for a vector potential, S(r, t), in the electronic TDSE, in addition to the scalar potential V (r, t). We will analyse this potential for an exciting experiment, namely the laser-induced localization of the electron in the H + 2 molecule [12]. We find significant differences between this exact potential and both the classicalnuclei potential Eq. (5) and the Hartree potential Eq. (6). Refs. [13,14] proved that the exact solution of the complete molecular TDSE Eq. (1) can be written as a single product, of a nuclear wavefunction χ(R, t), and an electronic wavefunction parametrized by the nuclear coordinates, Φ R (r, t), which satisfies the partial normalization condition (PNC) dr\|Φ R (r, t)\| 2 = 1. Here we instead consider the reverse factorization, It is straightforward to see that the formalism presented in Ref. [13] follows through simply with a switch of the role of electronic and nuclear coordinates. In particular, (i) The exact solution of the TDSE may be written as Eq. (8), where χ r (R, t) satisfies the PNC dR\|χ r (R, t)\| 2 = 1. (ii) The nuclear wavefunction χ r (R, t) satisfies with the nuclear Hamiltonian The electronic wavefunction Φ(r, t) satisfies the TDSE: (11) Here the exact TD potential energy surface for electrons (e-TDPES) e (r, t) and the exact electronic TD vector potential S j (r, t) are defined as where ...\|...\|... R denotes an inner product over all nuclear variables only. (iv) The wave functions χ r (R, t) and Φ(r, t) are interpreted as nuclear and electronic wavefunctions: \|Φ(r, t)\| 2 = \|Ψ(r, R, t)\| 2 dR is the probability density of finding the electronic configuration r at time t, and \|χ r (R, t)\| 2 = \|Ψ(r, R, t)\| 2 /\|Φ(r, t)\| 2 is the conditional probability of finding the nuclei at R, given that the electronic configuration is r. The exact electronic N e -body current-density can be obtained from We can regard Eq. (11) as the exact electronic TDSE: The time evolution of Φ(r, t) is completely determined by the exact e-TDPES, e (r, t), and the vector potential S j (r, t). Moreover, these potentials are unique up to within a gauge transformation (iii, above). In other words, if one requires a purely electronic TDSE (11) with solution Φ(r, t) to yield the true electron (N e -body) density and current density of the full e-n problem, then the potentials appearing in this TDSE are (up to within a gauge transformation) uniquely given by Eqs. (12) and (13). A formalism in which the nuclear wavefunction is conditionally dependent on the electronic coordinates, rather than the other way around, may appear somewhat non-intuitive. However, in many non-adiabatic processes, the nuclear and electronic speeds are comparable, and, in some cases, such as highly excited Rydberg molecules, nuclei may even move faster than electrons [15]. We shall show in the following that the present factorization is useful to interpret the dynamics of attosecond electron localization, and that it gives direct insight into how the e-n coupling affects nonadiabatic electron dynamics. For this purpose it is useful to rewrite the exact e-TDPES as e (r, t) = approx e (r, t) + ∆ e (r, t) where approx e (r, t) = χ r (t) Ŵ ee (r) +Ŵ en (r, R) +Ŵ nn (R) +v e ext (r, t) +V n ext (R, t) χ r (t) R and If the nuclear density is approximated as a deltafunction at R(t), then approx e reduces to the electronic potential used in the traditional mixed quantumclassical approximations: trad e (r, t) =Ŵ ee (r) +Ŵ en (r, R(t)) +Ŵ nn (R(t)) +v e ext (r, t) +V n ext (R(t)). This approximation not only neglects the width of the nuclear wavefunction but it also misses the contribution to the potential from ∆ e (r, t), Eq. (16). Methods that retain a quantum description of the nuclei (e.g. TD Hartree [11]) approximate Eq. (15), although without the parametric dependence of the nuclear wavefunction on r, and still miss the contribution from Eq. (16). In the following example, we will show the significance of the e-n correlation represented in the term ∆ e . Among the many charge-transfer processes accompanying nuclear motion mentioned earlier, here we study attosecond electron localization dynamics in the dissociation of the H + 2 molecule achieved by time-delayed coherent ultrashort laser pulses [12]. In the experiment, first an ultraviolet (UV) pulse excites H + 2 to the dissociative 2pσ u state while a second time-delayed infrared (IR) pulse induces electron transfer between the dissociating atoms. This relatively recent technique has gathered increasing attention since it is expected to eventually lead to the direct control of chemical reactions via the control of electron dynamics. Extensive theoretical studies have led to progress in understanding the mechanism [12], and highlight the important role of e-n correlated motion. Here we study the exact e-n coupling terms by computing the exact e-TDPES Eq. (12). We consider a one-dimensional H + 2 model, starting the dynamics after the excitation by the UV pulse: the wavepacket starts at t = 0 on the first excited state (2pσ u state) of H + 2 as a Frank-Condon projection of the wavefunction of the ground state, and then is exposed to the IR laser pulse. The Hamiltonian is given by Eq. (2), with R → R, the internuclear distance, and r → z, the electronic coordinate as measured from the nuclear center-of mass [16]. The kinetic energy terms areT n (R) = − 1 2µn . The IR pulse is taken into account using the dipole approximation and length gauge, asv e ext (z, t) = E(t)q e z, where E(t) = E 0 exp − t−∆t τ 2 cos(ω(t−∆t)), and the reduced charge q e = 2MH+2 2MH+1 . The wavelength is 800 nm and the peak intensity I 0 = E 2 0 = 3.0 × 10 12 W/cm 2 . The pulse duration is τ = 4.8f s and ∆t is the time delay between the UV and IR pulses. Here we show the results of ∆t = 7 fs. We propagate the full TDSE (1) numerically exactly to obtain the full molecular wavefunction Ψ(z, R, t), and from it we calculate the probabilities of directional localization of the electron, P ± , which are defined as P +(−) = z>(<)0 dz dR\|Ψ(z, R, t)\| 2 . These are shown as the black solid (P − ) and dashed (P + ) lines in Fig. 1. It is evident from this figure that considerable electron localization occurs, with the electron density predominantly localized on the left (negative z-axis). We now propagate the electrons under the traditional potential Eq. (17), employing the exact TD mean nuclear position R(t) obtained from Ψ(z, R, t) by R(t) = Ψ(z, R, t)\| R \|Ψ(z, R, t) , and calculate the electron localization probabilities, shown as red solid line (negative region) and dashed line (positive region) in Fig. 1. Comparing the red and black lines in Fig. 1, we find that the traditional potential yields the correct dynamics until around 5 fs, but then becomes less accurate: finally it predicts the electron to be almost perfectly localized on the left nucleus, while the exact calculation still gives some probability of finding the electron on the right. To understand the error in the dynamics determined by the traditional surface, we compute the exact e-TDPES (12) in the gauge where the vector potential S(z, t) is zero [17]. In the upper panel of Fig. 2, the exact 12)) is plotted (black line) at three times [18], and compared with the traditional potential (Eq. (17)) trad e (red line) evaluated at the exact mean nuclear position. In the lower panel, the electron densities calculated from dynamics on the respective potentials are plotted. A notable difference between e and trad e is an additional interatomic barrier which appears in the exact potential, and a step-like feature that shifts one well with respect to the other. These additional features arise from the coupling terms contained in ∆ , and are responsible for the correct dynamics, which is evident from the green curve in Fig. 1: this shows the results predicted by propagating the electrons on approx e . The result is close to that of the red traditional curve, and the potentials (not shown for figure clarity) are also close to the red potentials shown in Fig 2. A TD Hartree treatment is also close to the results from propagating on trad e . An examination of the different components in Eq. (16) shows that the additional interatomic barrier arises from the term 1 2m ∂ ∂z χ z \| ∂ ∂z χ z R , while the other two terms in Eq. (16) yield the step. The current understanding of the mechanism for electron localization is that as the molecule dissociates, there is a rising interatomic barrier from W en , which, when it reaches the energy level of the excited electronic state largely shuts off electron transfer between the ions [12]. The electron distribution is largely frozen after this point, as the electron can only tunnel between the nuclei. The additional barrier we see in the exact e-TDPES, leads to an earlier localization time, and ultimately smaller localization asymmetry. However, each of the three terms in Eq. (16) for ∆ play an important role in the dynamics: if the electronic system is evolved adding only the barrier correction to approx e the local-ization asymmetry is somewhat reduced compared to evolving on approx e alone but far more so when all three terms of ∆ are included. In conclusion, we have presented the exact factorization of the complete molecular wavefunction into electronic and nuclear wavefunctions, Ψ(r, R, t) = χ r (R, t)Φ(r, t), where the electronic wavefunction Φ(r, t) satisfies an electronic TDSE, and the nuclear wavefunction is conditionally dependent on the electronic coordinates. This is complementary to the factorization of Refs. [13,14,19], Ψ(r, R, t) = χ(R, t)Φ R (r, t), where instead the nuclear wavefunction satisfies a TDSE while the electronic wavefunction does not. The exact e-TDPES and exact TD vector potential acting on the electrons were uniquely defined and compared with the traditional potentials used in studying localization dynamics in a model of the H + 2 molecular ion. The importance of the exact e-n correlation in the e-TDPES in reproducing the correct electron dynamics was demonstrated. Further studies on this and other model systems will lead to insight into how e-n correlation affects electron dynamics in non-adiabatic processes, an insight that can never be gained from the classical electrostatic potentials caused by the point charges of the clamped nucleus nor the charge distributions of the exact nuclear density. Preliminary studies using the Shin-Metiu model [20] of field-free electronic dynamics in the presence of strong non-adiabatic couplings show that peak and shift structures in the exact e-TDPES, similar to those in the localization problem discussed here, appear typically after non-adiabatic transitions. Finally, we note that the exact TD electronic potentials defined in this study, together with the exact TD nuclear potentials derived in [13,14] establish the exact potential functionals of TD multicomponent density functional theory [21,22]. The study of these potentials may ultimately lead to approximate density-functionals for use in this theory, which holds promise for the description of real-time coupled e-n dynamics in real systems. Partial support from the Deutsche Forschungsgemeinschaft (SFB 762), the European Commission (FP7-NMP-CRONOS), and the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences and Biosciences under award DE-SC0008623 (NTM),is gratefully acknowledged.	2013-11-13T17:16:06.000Z	2013-11-13T00:00:00.000	{ "year": 2013, "sha1": "08d10ca7cb5ddc0670084ba47229428e85a50500", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "08d10ca7cb5ddc0670084ba47229428e85a50500", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
259412088	pes2o/s2orc	v3-fos-license	FAILURE HANDING STRATEGY FOR MICRO PEOPLE BUSINESS CREDIT (KUR) : The smooth operation of the loans released cannot be guaranteed since under certain circumstances, whether intentionally or inadvertently, the debtor breaks his word, making it difficult for Bank BRI Salobulo Kota Palopo to accept the return of the loans that have been given. This study aims to determine the strategy carried out by BRI Salobulo Unit in dealing with KUR Micro credit defaults and the causes and solutions for the handling strategy. The type of research used is qualitative research through interactive data analysis that produces descriptive data, in the form of a fact with the right interpretation to obtain a systematic, factual and accurate picture or painting of facts, characteristics, and the relationship between the phenomenons under study. The results showed that the default on Micro KUR Loans at the BRI Salobulo Bank, Palopo City occurred because the debtor was late in paying his credit bill when it was due. The strategy for handling default loans is adopted by Bank BRI Salobulo Unit through a risk mitigation strategy (preventive) by analyzing customer eligibility with the 5C principle. The rescue strategy (repressive) is carried out by restructuring credit, which is re-launched with the 3R technique (Rescheduling, Reconditioning and Restructuring). After mapping the SWOT matrix, several strategies are appropriate, and the strategy focuses on the ST strategy . BACKGROUND The problem of default in the provision of credit facilities is a significant problem that often occurs, especially in micro people's business credit (KUR), which targets new people in the business world and are vulnerable. The disbursed loans cannot be guaranteed to run smoothly as expected because, in a condition that is either intentional or unintentional, the debtor violates his promise so that Bank BRI Salobulo, Palopo City, has difficulty receiving installments for repayment of loans that have been given. Based on Indonesian banking statistical data released by the OJK, it is known that working capital category credit/financing in Indonesian commercial banks was recorded at 13,086 trillion, specifically for sharia banking itself of 8,845 trillion in MSME financing (Indonesian Banking Statistics, 2022). Many default debtor customers will impact the NPL percentage ratio (not performing loan) because the bank no longer accepts installments based on time. The table above illustrates that the Micro KUR NPL at Bank BRI Unit Salobulo from 2019-2021 has always increased. A significant increase from 0.56% in 2019 to 1.11% in 2021. Even though this amount is still within the safe limits of Bank Indonesia (BI), if it is allowed to continue without any effective handling action, it will affect credit quality and incur potential losses. If the distribution of Micro KUR loans is not accompanied by a good handling strategy, it will cause credit defaults to arise, which in turn will impact the poor quality of credit distribution due to a large number of nonperforming loans. Familial strategies can encourage customers to pay their debts without using coercion, thus increasing awareness of paying their obligations (Erawati & Atmadja, 2020). With this study, the causes of default can be identified, providing an overview regarding strategies in dealing with the problem of defaulting Micro KUR debtors, where credit that is considered to have the potential to default can be taken preventive measures, improving credit quality so that it runs smoothly and controlling credit restructuring so that it does not return to non-performing loans and formulating the right strategy formulation in dealing with Micro KUR defaults at Bank BRI Salobulo Unit, Palopo City. THEORY AND METHODS Default The term "Default" comes from the Dutch language, which refers to an unwanted result. Default is an act of someone not carrying out achievements or forgetting to carry out tasks made between the creditor and the debtor. Achievement is the content of an agreement, and the debtor is considered in default if he fails to fulfill the agreed obligations (Supriono, 2011). In contract law, the terms schuld and Haftung are known. Schuld is the creditor's obligation to present achievements to the debtor. While Haftung is the role of the debtor to hand over his assets to be accessed by creditors in the amount of a loan to pay off the debtor's debt if the debtor fails to fulfill the role of paying off his loan. People who do not achieve are those who do not meet the expected goals and cannot or fail to carry out everything that is their responsibility on purpose (Kotler et al., 2006). According to R. Subekti, nonperformance can be identified by several measurements, namely, when the debtor never performs his obligations, the debtor violates the terms of the agreement by not fulfilling his responsibilities properly or on time or by acting in a manner prohibited by the agreement. It is these variables in the banking system (internal factors) and those from debtors (external factors) that cause credit defaults to develop (Kasmir, 2015). Internal Factor Internal factors are factors caused by the banks themselves, in this case because of the human resources they have. The Human Factors Analysis and Classification System (HFACS) reveals that failures that occur in humans are caused by elements, namely violations and negligence. According to human error theory, this factor is due to human actions that cause the system to run unsatisfactorily. Debtor Factor Default has been seen from the debtor, namely first caused by factors force will where there is an element of intent on the debtor, intentionality means wanting and wanting what he does. In the theory of will (will theory) that someone deliberately ignores his promise based on his will to do the deed, while on the theory of knowledge or imagining (performance theory), intentional means that he already knows the consequences of his actions. The second factor is because of force majeure/force majeure, where circumstances force them not to carry out their achievements due to an event beyond the debtor's will (Amalia, 2012). Default Handling Strategy Strategy plays an important role in achieving goals because it determines the actions that must be taken and how they must be carried out to be effective. According to Stephanie K. Marrus, strategy is a concept designed and focused on what the organization wants to achieve in a long time, accompanied by methods, actions, and hard work. Meanwhile, strategy, in the words of Fred R. David, is the art and science of developing, practicing, and assessing cross-functional choices that enable an organization to achieve its goals (Rahim & Radjab, 2016). Useful strategies protect, maintain, and enhance an organization's capabilities and competitive advantages. In implementing strategy, some research results find that organizations with a systematic or directed strategy perform better (outperformed) than organizations without/not formulated with a systematic strategy. The right strategy will determine maximum success in resolving non-performing loans so that losses on non-performing loans can be minimized. Facing credit problems that are already problematic because customers deny their performance, banks can choose a strategy consisting of 2 (two) options, namely first, "ending the bond" through litigation, resolving disputes through a process carried out in court where the authority to regulate and decide cases is taken over by the judge. The second option is "continuation of the credit relationship," which means bad loans are resolved by a family process without going through the courts (non-litigation) where there is a renegotiation between the creditor and the debtor customer. This is an active alternative dispute resolution (ADR) in the language is called alternative dispute resolution (APS). Philip D. Bostwick defines APS as a rescue tool, and legal effort that transfers legal cases outside the court based on the agreement of the parties, reduces lawsuit costs and streamlines time and prevents legal problems which are generally brought to court. In dispute resolution theory, selecting one of these strategies must be supported by the implementation of the strategy (action plan) is good; before severing a relationship with a customer, it is better if the bank's management can try to save it to avoid bigger losses. Rescuing credit is an attempt to find a way out, both in a preventive and curative form, for credit defaults that may occur or have occurred. The description of the rescue or restructuring actions above is clearer as follows (Kasmir, 2015): 1. Rescheduling (Rescheduling) Rescheduling can be done in several ways, for example, by extending the debt repayment period, loosening the time or debt repayment schedule, in the sense that the bank opens the way to provide a longer time extension for customers to pay off their loans. Banks can also increase the length of customer installments, for example, from 18 to 26 times to reduce the monthly installments that must be paid. Return requirements (Reconditioning) Return requirements, namely changing some or all of the terms of the debt such as interest capitalization, temporary suspension of interest payments, subsidizing loan interest rates or even freeing credit from interest. Rearrangement (Restructuring) Rearrangement, namely changing credit terms in the form of increasing the amount of credit where the bank changes all or part of the interest that has matured into a new loan principal. Banks may carry out restructuring so that credit quality can return smoothly, banks cannot carry out restructuring if the aim is to reduce the classification of financing quality, to set a higher PPA, or to stop recognizing margin income on an accrual basis. Restructuring can only be carried out at the debtor's written request who meets the following requirements (Totok & Nuritmo, 2017): 1. The client is in a difficult financial situation. 2. The client has progressed in business and is still able to pay installments after restructuring. Restructured credit records are loans in the substandard, doubtful, and loss credit groups. The handling of problematic installments is not only through installment restructuring which has been discussed above; the execution of collateral objects can also be carried out either through direct sales by the debtor or through an auction. At least in executing credit guarantees, private sales must first be attempted if the debtor still wants to cooperate, but if private sales cannot be achieved, then the execution of collateral items through auctions can be carried out. 1. A direct sale is a sale of collateral or non-collateral voluntarily or voluntarily for guaranteed assets so that they can pay part or all of the loan. 2. Credit loan auctions are sales of credit guarantee goods open to the public with written and/or verbal price bids that are raised or lowered to reach the highest price preceded by an auction notification. As for the settlement of credit defaults through legal channels (litigation), namely the District Court / Prosecutor's Office, the Committee for State Receivables (PUPN), and through arbitration or Alternative Dispute Resolution Agency (Noor, 2013). Micro Business Credit Survey (KUR) Credit comes from the Greek word "believe," which means trust (truth or faith). Therefore the basis of credit is trust. The interpretation of a loan (credit) according to banking law number 10 of 1998 is the provision of money or an equivalent claim resulting from an agreement that requires the borrower to pay off his debt after a certain period with interest or profit sharing. To advance the community's economy, especially for people with middle to lower incomes, the government initiated a special subsidized credit program for MSME actors. MSME loans are loans or financing given to business actors who meet the criteria for Micro, Small, and Medium Enterprises. Article 1 point 1 of Law Number 20 of 2008 concerning Micro, Small, and Medium Enterprises explains that Micro Enterprises are productive businesses owned by individuals (individuals) or businesses belonging to groups. In essence, micro-enterprises are all kinds of businesses that are productive, have creditworthiness, and are worth paying for. So that business actors estimate and are trusted to be able to pay the amount of the installments until completion according to the contract with the creditor bank. These loans are channeled to MSME players who have productive and viable businesses but do not yet have additional or feasible collateral and are not yet bankable. Launching the KUR program aims to improve the economy, alleviate poverty and absorb labor (Tambunan, 2012). The main requirement for micro KUR recipients is that the prospective debtor is an individual who carries out a viable, productive business, is registered as a city where applies as evidenced by the ownership of a KTP, KK, and the length of time the business has been running for at least 6 months. Several requirements for applying for a Micro KUR loan must be met, namely having run a business for at least 6 months, having an original KTP and KK along with photocopies, attaching a business license, NPWP, and others. Research methods This study uses an approach with qualitative methods. Theoretically, qualitative/naturalistic research is research that seeks to uncover a case in its natural state so that it can provide valid facts. Unlike the case with quantitative research which is limited to testing hypotheses, qualitative research is discovery (Sugiyono, 2014). Based on its nature, the type of research used is descriptive qualitative research with a case study approach (case study), trying to understand a case by specifically incorporating perspective, knowledge, and creativity into the analysis process, discussing issues from a related theoretical and research point of view to design a realistic strategy through observations, interviews, impressions and opinions of others regarding the problem at hand analyzed. So that the data obtained is expressed symbolically in the form of sentences written to describe reality. The focus of this research is to find out the facts on the ground regarding the handling of default customer problems in the Micro KUR credit agreements that are carried out. Researchers focus on the credit prevention and control strategy carried out by the BRI Unit Salobulo and analyze the company's environment to plan a new strategy so that the bank can decide the right strategy for resolving credit defaults that occur. Data analysis is a technique of systematically finding and compiling data from observations, interviews, field notes, and others to find meaning (meaning). Data from various sources were observed, processed, filtered, and analyzed qualitatively through descriptive-analytical and SWOT analysis. A qualitative analysis was carried out with a descriptive-analytical characteristic to obtain an overview regarding the strategy carried out by the BRI Salobulo Unit during the study {expose the fact) or to see what conditions are in the situation. While the SWOT analysis is used to plan a strategy in dealing with non-performing Micro KUR loans. RESULTS AND DISCUSSION Results of Data Analysis 1. Analysis of the factors that influence the occurrence of default on Micro KUR at Bank BRI Salobolo Unit, Palopo City. Micro KUR credit defaults at Bank BRI Unit Salobulo occurred because there were factors that pushed that caused this to happen. Based on the interviews, it can be seen that the causes of defaults at Bank BRI Unit Salobulo Palopo City arose due to internal and external factors. External factors are caused by the debtor who does not carry out his obligations due to circumstances forcing him not to perform that is outside his will (force majeur) because customers experience economic difficulties due to failed businesses or even due to natural disasters or other disasters and other factors from customers who are absent from their obligations by intentionally committing default (force will). While the internal factors were caused by the bank itself not being good at analyzing the eligibility of prospective customers before the credit was decided, it was also explained that what made the default rate influenced was the target set every day so that some paramedics were too focused on pursuing targets which led to the birth of credits. that are not qualified and have the opportunity to become problem loans. Mr. Udi Pratwandi, one of the paramedics for KUR Micro BRI, the Salobulo Unit, said that the most common cause of complaints by default customers was because the business being run was declining. As happened with Mr. Akmal Syam as a defaulting people's business credit debtor, it is known that he has been in arrears for payments for 4 months starting from January. Mr. Akmal Syam took Micro KUR credit with a platform of Rp. 25,000,000 (twenty-five rupiahs) and a monthly installment fee of Rp. 1,108,100 (one million one hundred eight thousand one hundred rupiahs). Thus, the troubled people's business credit experienced by Mr. Akmal Syam can be classified as a substandard loan because it has exceeded 120 days. Mr. Akmal was conveyed to the bank because his cafe business had decreased turnover due to a lack of visitors, so he could not pay his credit installments. Another default customer affected by the customer's bad character and who is not responsible for his promise is Mrs. Amelia, who has been in arrears since the second month's payment and is now in arrears for 3 months by taking a credit of Rp. 10,000,000 when several friendly attempts were made, the customer was not at home. In general, BRI Unit PT. The BRI Salobulo Unit, Palopo City, in carrying out interim control based on the results of an interview with Mr. Suprayitro Arafah, in the process of preventing defaults or defaults on bank loans, can be done using careful and thorough credit analysis when predicting risk and return that will be obtained, does not necessarily provide credit to all customers who want Micro KUR credit. Mantri as a field officer, must have good analytical skills to take into account all aspects and indicators of eligibility assessment of prospective customers to maintain credit distribution according to their functions. always informing customers when the payment is before the due date; this was conveyed by Mr. Udi Pratwandi "Actually, we don't have to wait for the customer to pay late and then take action, we always do it before it's time for the customer to pay, we will remind him earlier because sometimes the customer forgets the installment date ". He further explained that in preventing credit defaults so after the credit is disbursed, the bank does not just let go. Various efforts were made by Bank BRI Salobulo Unit in analyzing the feasibility of prospective BRI Salobulo customers using steps according to the 5C principle. As for the use of this principle first character, namely by assessing the character or character of the prospective customer, Secondcapacity namely by assessing the customer's ability to pay back or settle installments as seen from the business to be financed. Third capital, namely by assessing the capital customers own in conducting business. Fourthcollateral, namely assessing the guarantee or collateral provided by the customer to the bank. The fifth condition of the economy is assessing the customer's economic condition. b. Strategy to save Default Credit on Micro KUR by Bank BRI Salobolo Unit, Palopo City. Various attempts were made to save credit so as not to harm the bank further. As for the results of interviews conducted by the author in the field, it can be seen that efforts to save Bank BRI Unit PT. The BRI Salobulo Unit, Palopo City, handles the distribution of default micro people's business loans, namely taking handling actions starting with This continuous or continuous billing method is applied to loans included in collectability under special mention and are substandard which are more than 90 days old. Billing done by the orderlies/AO is to visit the business financed by the bank or the customer's house to discuss problems that occur persuasively, through written billing letters or only through telephone media which are routinely carried out to save credit before it gets worse. Mr. Suprayitro said that in monitoring credit, it must be ascertained why the credit is problematic, whether due to business or other factors. one of the efforts made is to come directly with a visit form to the arrears. Banks can also relieve customers who have good intentions to complete installments. When interviewed, Mr. Hasbi Hatta, the mantra of KUR Micro, explained that the model or technique for rescuing default loans applied to Bank BRI's Salobulo unit uses the 3R technique (rescheduling, reconditioning, restructuring). In; conduct visits to customers' homes to find out what the obstacles are and find solutions by providing leeway in payment terms, lowering interest rates or waiving interest. Mr. Udi Pratwandi added that banks facing problem loans were resolved well, hoping that the debtor would pay off his obligations without having to take legal action for the good of the bank and the customer. c. Strategy for Settlement of Defaults on Micro KUR Loans at BRI Bank Salobulo Unit, Palopo City. Settlement of credit is done because of the failure of rescue efforts. The AO of Bank BRI Salobulo Unit explained that the resolution of bad loans in credit agreements could be reached in two ways, namely by means of litigation and non-litigation. Litigation efforts are also known as settlement efforts through legal channels. Litigation is a dispute resolution mechanism through the courts by filing a lawsuit. Considering that the settlement process through the courts usually takes a relatively long time, Bank BRI Unit Salobulo tends to choose the settlement of bad loans through efforts-litigation (outside court). Mr. Hasbi Hatta explained that if the credit has absolutely no way to save it, then it is forced to settle the relationship, the bank will give several options to the customer whether the guarantee will be sold himself, then the price will be used to cover the rest of the loan or pursued by court if the customer not willing to cooperate. We can conclude from the results of interviews with several informants, information is obtained where the steps to achieve settlement of non-performing loans are taken in the best way for both parties. Analysis of planning a new strategy that is appropriate in handling defaults on Micro KUR at Bank BRI Salobolo Unit, Palopo City. SWOT Analysis of Bank BRI Salobulo Unit, Palopo City in Handling Micro KUR Defaults can be seen in the following table: Causes of default on Micro KUR at Bank BRI Salobolo Unit, Palopo City From the results of the interviews and through the observations of the root cause researchers in improving the quality of non-performing loans at Bank BRI Salobulo Unit, Palopo City, it is divided into two interrelated factors, namely due to internal problems and influences from external companies. Mr. Hasbi Hatta as the KUR Micro Mantri explained that one of the reasons for a large number of credit defaults was influenced by internal factors, but the employees of Bank BRI Salobulo Unit had worked well and were responsible for their respective jobs so internal obstacles could be slightly avoided. The various causes of default from internal factors are as follows: a. The Mantri is not good enough to do a 5C analysis of prospective debtors to find out whether the prospective customer is eligible for Micro KUR. Providing credit is not according to the needs and abilities of the customer, the orderlies are not detailed during the survey, and not having sufficient information about the character of the prospective debtor can make the orderlies wrong in predicting creditworthiness before being approved. b. Mantri is too focused on pursuing targets so he doesn't pay attention to the quality of the credit he distributes. It was explained that what made the default rate more or less influenced by the existence of targets that had to be met so that some paramedics felt motivated to pursue targets which actually led to the birth of loans that were not of high quality and had the opportunity to become problem loans. c. Ineffective supervision and control due to the large number of customers that must be handled. According to the theory of human error, this factor is due to human errors, which cause the system to run less effectively and have a negative impact on bank performance. In addition to the company's internal problems, it can also occur due to external obstacles, namely originating from the Micro KUR debtors themselves who intentionally or unintentionally cause problematic Micro KUR Credit. The form of customer negligence is caused by several customers using Micro KUR capital for consumptive purposes or because of an element of fraud where KUR funds are divided between several people so that when the time comes for payment, the customer has difficulty collecting their installments and there are even cases of Micro KUR customers arguing that the credit was not used as additional business capital unless there are other parties who deliberately work together to take credit because the name is disabled. This moral hazard behavior is influenced by the character of the customer. The nature or character of a person can be used as one of the most important assessment indicators because, from nature, we can know the willingness to pay and be responsible (Kasmir, 2008). If the character of the customer is not good, where he is more concerned with other things than his credit, or there is no willingness to pay off the debt, it is very difficult to handle. According to Mr. Udi Pratwandi during an interview, other external factors were influenced by market changes, such as economic conditions that experienced an increase, causing a decrease in trader turnover so that many credit problems. The Strategy for Handling Micro KUR Credit Defaults has been carried out by Bank BRI Salobolo Unit, Palopo City in dealing with default debtors. a. The prevention strategy carried out by Bank BRI Salobulo Unit, Palopo City as a basis for other considerations before deciding on a loan, is guided by the 5C technique as a form of customer feasibility analysis. In its implementation, the bank, before making a decision, will see and assess directly the background, goals, capabilities, and effort to be financed, but the paramedics have not sought in-depth and thorough information on all aspects of the assessment so that it can be known properly and clearly. Sometimes the customer is smarter than the bank, so he is able to hide the real condition, making the paramedics unable to find out his true purpose. So a more careful character assessment is needed; the bank that applies it prudential banking principal can slightly minimize the risk of non-performing loans so that NPLs can be stable. The theory of prudential banking explains that banks provide loans in a more careful manner which is an important principle and must be implemented because it is the best solution in order to maintain a quality, resilient, and safe banking system (Hermansyah, 2018). Adequacy of information is needed to support and optimize prevention in risk mitigation strategies. As a distributor of Micro KUR, whether it is Islamic banking or conventional banking or Islamic financial institutions or other financial institutions, it is important to apply the principle of prudence starting from the beginning when the debtor submits a credit application, analysis of credit granting until credit is paid off is the key to success for healthy credit. b. The strategy for saving BRI's problem loans from the Salobulo Unit in a family way. Guided by Bank Indonesia Circular No. 26/4/BPPP dated May 29, 1993, which regulates the rescue of problem loans before they are resolved through legal institutions. According to Hanafi, for risks that have occurred in a company, it is important to carry out risk control because risks must be managed properly and determine the best way to save the company from losses (Arifudin, et al., 2020). This was also carried out at Bank BRI Salobulo Unit, Palopo City, where rescue efforts were taken after seeing signs that credit would become a bigger loss. With this rescue strategy, the bank will make approaches to customers. This approach is carried out by providing direction, coaching, and instructions that can help customers, it is hoped that this will awaken the debtor's ability to try optimally so that it does not turn into bad credit.. Efforts to save problem loans can be made using the method of restructuration; the strategy for rescuing default loans in Micro KUR loans at Bank BRI Salobulo Unit is with changes regarding payment schedules and credit terms. Based on the acquisition of data from the BRI Salobulo Unit in the process of rescheduling, BRI Salobulo Unit only rescues customer credit that meets certain requirements to be rescued, including that the debtor's business must have business prospects to be able to rise and the debtor still shows good faith in improving his credit so that there is still a possibility that the credit can be saved and the bank's losses are not getting bigger. Changing various existing requirements, such as interest capitalization, where interest is used as the principal debt. BRI Unit Salobulo seeks to save credit by delaying interest payments until a certain time while the principal must still be paid. By providing relief in which partial unpaid interest is released or stopping special interest calculations for honest, open, and cooperative customers, and the business still has a chance to return to operating as before and allows customers to complete their loans without going through litigation. c. The settlement strategy is taken by Bank BRI Salobulo Unit, Palopo City, if the customer's credit cannot be saved at all, the customer does not show a cooperative attitude anymore, then the litigation route is the last resort to resolve disputes that occur in the context of restoring rights and obligations between debtors and creditors. The aggressive efforts of banks to resolve bad loans are by taking firm action in the form of verbal and written warnings to customers and finally liquidating (selling collateral In table 3 above, the strength factor has a score of 1.92 while the weakness factor has a score of 1.51. Furthermore, in table 4 above, the value of the opportunity factors is 1.16 and the value of the threat factors is 2.11. These results indicate that the BRI division of Bank Salobulo, Palopo City, faces more threats than opportunities. The placement of internal and external factors above resulted in a range of scores of Strengths = 1.92, Weaknesses = 1.51, Opportunities = 1.16, and Threats = 2.11. The results can be seen in the following figure. Based on the results of the SWOT evaluation of diagrams and reflections on the ST matrix (strength-threats), now the strategy will focus on ST, as explained earlier in the SWOT Matrix table, will use a strategy to use resources to overcome the problems that have been identified by means of strategy diversification. performance to smooth, but if there is no improvement after the credit is restructured and the debtor is no longer cooperative, BRI Salobulo bank is forced to liquidate the credit guarantee either through court institutions or sales underhand. Planning a new strategy formulation with the SWOT analysis of Bank BRI Salobulo Unit shows the ST strategy which focuses on strengths to dispel threats which can be done by paying attention to early learning indicators and SID facilities to maintain and provide assistance and control period after the credit is disbursed. Abdurrasyid, Priyatna. (2002).Arbitration & Alternative Dispute Resolution.	2023-07-11T00:50:35.769Z	2023-06-16T00:00:00.000	{ "year": 2023, "sha1": "32d0966779abe4fe79b7d5c394c1dc2fbfa3f682", "oa_license": "CCBY", "oa_url": "https://journal.ar-raniry.ac.id/index.php/JoSE/article/download/2457/1376", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "4ac10485f1755e480e928330443a3b733bf4410a", "s2fieldsofstudy": [ "Business", "Economics" ], "extfieldsofstudy": [] }
230660141	pes2o/s2orc	v3-fos-license	Three‐year outcomes of peripheral blood mononuclear cells vs purified CD34 + cells in the treatment of angiitis‐induced no‐option critical limb ischemia and a cost‐effectiveness assessment: A randomized single‐blinded noninferiority trial Abstract For patients with angiitis‐induced critical limb ischemia (AICLI), cell transplantation, such as purified CD34+ cells (PCCs) and peripheral blood mononuclear cells (PBMNCs), is gradually being used as a promising treatment. This was the first randomized single‐blinded noninferiority trial (number: NCT 02089828) specifically designed to evaluate the therapeutic efficacies of the transplantation of PCCs vs those of PBMNCs for the treatment of AICLI. We aimed to compare the mid‐term safety and efficacy between the two groups and determine their respective advantages. From April 2014 to September 2019, 50 patients with AICLI were equally allocated to the two groups, except for 1 lost patient, 1 amputee, and 1 patient who died of heart disease. The other 47 patients completed the 36‐month follow‐up. The endpoints were as follows: major amputation‐free survival and total amputation‐free survival at 6 months, which were 96.0% and 84.0% in the PBMNCs group and 96.0% and 72.0% in the PCCs group, respectively. These rates remained stable at 12, 24, and 36 months. The PCCs group had a significant higher probability of rest pain relief than the PBMNCs group, whereas earlier significant improvements in the Rutherford classification were observed in the PBMNCs group. Accordingly, PCCs would be preferred for patients with significant pain, whereas PBMNCs may be a good option for patients with two or more critically ischemic limbs. Concerning cost‐effectiveness, PCCs are not more cost‐effective than PBMNCs. These outcomes require verification from long‐term trials involving larger numbers of patients. \| Design and participants Fifty patients were enrolled in this randomized single-blinded parallelgroup controlled trial from April 2014 to September 2019, and the study was approved by the Ethics Committee of Fudan University Affiliated Zhongshan Hospital. All the participants signed informed consent before enrollment. The study protocol was detailed previously. 9 In brief, the inclusion criteria were as follows: patients aged 18-80 years; the presence of stenotic or occlusive lesions in the limb arteries, as confirmed by magnetic resonance angiography, computed tomography angiography, or digital subtraction angiography; CLI with a Rutherford classification of 4-5 that was unsuitable for or not improved ≥3 months following surgery or an endovascular intervention; rest pain that was not alleviated after ≥1 month of conservative treatments; and an area of tissue loss that had not diminished in size after ≥1 month of these treatments. The exclusion criteria were cardiac-cerebral vascular events ≤3 months before admission; proliferative retinopathy; a life expectancy of ≤1 year; a diagnosis or suspicion of cancer ≤5 years before admission; or contraindications for the administration of recombinant human granulocyte colony-stimulating factor (rhG-CSF). \| Randomization and masking As we had described, 9 participants' demographic data, CLI etiologies, Rutherford classifications, comorbidities, other disease histories, and medical and surgical histories were recorded by the investigators before randomization. Then, the eligible patients were allocated 1:1 to the Lessons learned • This study demonstrated the satisfactory mid-term outcomes of transplantation of both peripheral blood mononuclear cells (PBMNCs) and purified CD34 + cells (PCCs) in the treatment of no-option angiitis-induced critical limb ischemia. • PBMNC transplantation was more cost-effective in most conditions and may be preferred for patients with ≥2 limbs needing transplantation at one stage. • PCC transplantation was advantageous in its ability to achieve earlier pain relief and thus was preferred for patients with severe pain. • In patients with obvious inflammation, PCCs could be chosen to avoid aggravation of inflammation for its lower injected volume and removal of most CD34 − cells. Significance statement In 2014, the first clinical trial specifically designed to evaluate the therapeutic efficacy and safety of transplantation of peripheral blood mononuclear cells (mixed cell types) with those of purified CD34 + cells (special cell type) for the treatment of angiitis-induced no-option critical limb ischemia was initiated in the authors' center. The mid-term outcomes of the study showed similar satisfactory efficacy and safety of the two therapies and provided evidence for more precise application of cell therapy under different conditions. PBMNCs or PCCs groups at random using a computer-generated randomization schedule (SAS, Proc Mixed, version 8.2; SAS Institute, Cary, North Carolina). All of the patients were masked before and after the interventions. The cell transplantations, the independent decisions about minor or major amputations after the transplantations and the patients' follow-up assessments, and data collection and analysis were completed by three separate groups of surgeons who were blinded to the work of other groups. The masking was removed if a serious adverse event occurred that was related to the trial, death, or loss to follow-up occurred or if an emergency required unmasking. When the 36-month follow-up period ended, the database was locked. \| Procedures RhG-CSF (Neupogen; Amgen, Thousand Oaks, California; 5-10 μg/kg per day) and enoxaparin (4000 IU/day) were subcutaneously administered in all patients for 4 days. Apheresis (COM.TEC; Fresenius Hemocare GmbH, Bad Homburg, Germany) was performed on the fifth day. Then, for the patients in the PBMNCs group, cells separated by apheresis were washed three times and resuspended in an ethylenediaminetetraacetic acid-phosphate buffered saline solution (200 mL) that contained 0.5% human albumin. For the patients in the PCCs group, CD34 + cells were purified using a magnetic cell sorting system (MiltenyiBiotec GmbH, Bergisch Gladbach, Germany) immediately after leukapheresis. The final cell products were assessed by flow cytometry and leukocyte counting. The surgeons implanted the cells into the calves/arms and feet/hands of the ischemic limbs via equidistant intramuscular injections (0.5 mL/site) under general anesthesia. Severely infected wounds were debrided. Additional details about the procedure have been previously described. 10 \| Outcomes All of the adverse events and all-cause mortality from mobilization to 2 weeks after injection, the leukocyte counts during hospitalization and at 1, 2, 3, 6, 12, 24, and 36 months during the follow-up, and pathological retinal angiogenesis were included in the safety outcomes. The primary efficacy outcomes included minor amputation (below the ankle), major amputation (above the ankle), and total amputation (both). The major amputation-free survival (MAFS) and total amputation-free survival (TAFS) rates were calculated. The secondary efficacy outcomes included complete wound healing, the Wong-Baker Faces Pain Rating Scale (WBFPS; a score of 0 represents no pain and a score of 10 represents the greatest pain), Rutherford classification, pain-free walking time (PFWT; at 2.5 km/h and at a 10% incline on a treadmill), the ankle-brachial index (ABI), the toe-brachial index (TBI), transcutaneous oxygen pressure (TcPO 2 ), and quality of life (QoL). 9,11,12 During the 3-year follow-up, recurrence (transplanted limb had CLI again) and new F I G U R E 1 Trial design. HES, hypereosinophilic syndrome; PBMNCs, peripheral blood mononuclear cells; PCCs, purified CD34 + cells; SLE, systemic lupus erythematosus; TAO, thromboangiitis obliterans lesions (untransplanted limb had CLI) were assessed. The recurrence rate (RR) and new lesion rate (NLR) were evaluated. By using incremental cost-effectiveness ratios (ICERs), cost-effectiveness was quantified as the ratio of the difference in cost and the effect between the two strategies. Quality-adjusted life-year (QALY), which ranges from 0 (death) to 1 (health), was used to evaluate the effect of the two strategies. We calculated the QALYs by transferring the scores of the 36-item Short Form Health Survey (version 2; SF-36 v2) to the Short Form 6-Dimension (SF-6D). 13,14 Any patient who was lost to follow-up or dead was considered the worst-case scenario. All outcomes at 24 and 36 months were observational. \| Statistical analyses The quantitative data are shown as the estimated margin mean (EMM) ± SE (for comparison between two groups), mean ± SD, or as the median with the interquartile range (IQR), depending on their distribution. The categorical data are presented as numbers with percentages. Pearson's chi-square test with or without Yete's continuity correction or Fisher's exact test was used to compare the groups in relation to allcause mortality, complete wound healing, rest pain alleviation, Ruther- were enrolled in the trial after screening. All patients were equally allocated to each group at random. Forty-seven patients completed the 36-month follow-up period. One patient (PCCs group) was lost to follow-up at 2 months after transplantation, one patient (PBMNCs group) underwent a major amputation within 6 months after transplantation, and one patient (PCCs group) underwent a major amputation at 26 months and died of cardiac disease at 27 months after transplantation ( Figure 1). The baseline characteristics of the patients were detailed in our previous study. 9 The mean age of all patients was 41.46 years, and all patients were male with unilateral ischemic lower limb. In both groups, most patients had a smoking history (92.0% of the PBMNCs group and 84.0% of the PCCs group), whereas only a few patients had risk factors for cardio-cerebrovascular disease, such as hypertension (4.0% of the PBMNCs group and 8.0% of the PCCs group), diabetes mellitus (4.0% and 8.0%, respectively), and hyperlipidemia (8.0% and 8.0%, respectively). All patients were in AICLI condition with a 4-5 Rutherford classification. Forty-seven patients had TAO, two had hypereosinophilic syndrome (HES), and one had SLE. At admission, two patients had only rest pain (Rutherford classification = 4), 18 patients had a nonhealing ulcer, and 30 patients had gangrene (Rutherford classification = 5). The mean ABI, TBI, and TcPO 2 were similar in both groups. No significant differences were observed between the groups among all the baseline characteristics. \| Efficacy evaluation Preoperatively, 2 patients were categorized as Rutherford classification 4 (1 in the PBMNCs group, 1 in the PCCs group), and 48 were categorized as Rutherford classification 5 (24 in the PBMNCs group and 24 in the PCCs group). During the 36-month follow-up, the Rutherford classification in the PBMNCs group improved significantly by 3 months (P < .05) and was sustained up to 36 months (Figure 2A), whereas in the PCCs group, the Rutherford classification improved significantly by 6 months (P < .001) and was sustained up to 36 months ( Figure 2C). F I G U R E 3 Longitudinal changes in blood perfusion restoration and functional improvement. The assessments of blood perfusion restoration included the A, ankle-brachial index, B, toe-brachial index, and C, transcutaneous oxygen pressure, and the functional improvement was assessed based on D, pain-free walking time. The values are presented in linear graphs that show the means and SDs. P < .05 vs baseline; P < .01 vs baseline. ABI, ankle-brachial index; PBMNCs, peripheral blood mononuclear cells; PCCs, purified CD34 + cells; PFWT, pain-free walking time; TBI, toe-brachial index; TcPO 2 , transcutaneous oxygen pressure F I G U R E 4 Longitudinal changes in the Wong-Baker Faces Pain Rating Scale. The longitudinal changes in the WBFPS in both groups are depicted as linear graphs that show the mean values and the SD bars. P < .05; **P < .01 (intragroup comparison with baseline, based on a general linear mixed model). PBMNCs, peripheral blood mononuclear cells; PCCs, purified CD34 + cells; WBFPS, Wong-Baker Faces Pain Rating Scale T A B L E 1 Comparisons of the groups using the intention-to-treat principle based on worst-case scenarios Figure 3A). \| Amputation Four patients (two in each group) underwent planned minor amputation during transplantation for severe infection. We did not include these events in the statistical analysis. Within 6 months after F I G U R E 5 Kaplan-Meier curves showing the probabilities of A, major amputation-free survival, B, total amputation-free survival, and C, allcause mortality in both groups. PBMNCs, peripheral blood mononuclear cells; PCCs, purified CD34 + cells transplantation, nine patients underwent unplanned minor amputations (three in the PBMNCs group and six in the PCCs group), and one patient in the PBMNCs group underwent major amputations. At 12, 24, and 36 months, only one major amputation was performed in the PCCs group at 26 months. The major amputation rates were 4% (PBMNCs group) and 4% (one patient in the PCCs group was lost to follow-up at 2 months) at 6 months, 12 months, and 24 months and 4% (PBMNCs group) and 8% (PCCs group) at 36 months. The total amputation rate of the PBMNCs group was 16% at 6 months, and this rate was sustained up to 12, 24, and 36 months. In the PCCs group, the total amputation rates were 28% at 6, 12, and 24 months and 32% at 36 months (Table 1). There were no significant differences between the two groups in terms of the probabilities of MAFS and TAFS (Breslow-Wilcoxon test: P = .161 and P = .529, respectively; Figure 5). \| Quality of life The \| Recurrence and new lesions By the 36-month follow-up assessment, two patients experienced recurrence (one in the PBMNCs group and one in the PCCs group; Figure S1; Table 1 \| DISCUSSION Stem cell transplantation has been considered a promising treatment for patients with AICLI owing to its effects on vasculogenesis and angiogenesis. [16][17][18][19] Both PBMNC and PCC transplantation have been shown to be effective. 10,20,21 Endothelial progenitor cells, of which CD34 + cells are a key component, are responsible for therapeutic angiogenesis. 22 We reported the 12-month follow-up of this study and found that similar results were observed between the PBMNCs and PCCs groups. 9 When we combined the similar 36-month results of the two groups with the significant differences of the absolute number of CD34 + cells and that/body weight between the two groups, we inferred that CD34 + cells play a predominant role in the treatment and that the high purity of CD34 + cells might compensate for the loss of CD34 − cells. Most patients in the current trial had TAO (24/25 in the PBMNCs group and 23/25 in the PCCs group). Accordingly, the two groups had similar and comparable preoperative demographic characteristics, such as a low rate of cardiovascular and cerebrovascular disease risk factors and a high smoking rate (92% in the PBMNCs group and 84% in the PCCs group). TAO is a kind of disease characterized by nonatherosclerotic, progressive vasculitis of the small and medium arteries. Conventional therapeutic approaches, including surgical and endovascular procedures, have shown poor mid-to long-term efficacy. The major amputation rate of patients with TAO has been reported to be as high as 12%-31%, and approximately 34.8% of patients lost their ability to work before 42 years of age. [23][24][25] The poor effect of conventional revascularization methods was ascribed to pathological features, such as involvement of the distal middle and small vessels, preservation of the internal elastic lamina of the involved vessels, superficial phlebitis, and inflammatory nature, which are detrimental to endovascular or surgical treatment. 21,26 The other three patients had other types of angiitis, such as SLE and HES. Although they were not as common as TAO, attention should also be paid to them. A study reported that the prevalence of SLE-induced CLI was 1.4% in a retrospective study, with 71.43% of patients suffering from digit loss, 27 which was similar to the findings of our previous study. 25 Furthermore, there were no related serious adverse events, such as death, cardio-cerebrovascular events, or hepatic or renal dysfunction, that occurred in any patient. In brief, similar satisfactory mid-term safety and efficacy were observed in the two groups. In the process of obtaining PCCs, along with CD34 − cells, approximately 73% of CD34 + cells were lost. Therefore, we generally In a recent report, 29 patients were divided into a BMMNCs group and a smoking cessation group, and the results showed that smoking cessation alone may not improve ischemia, but smoking cessation might be a critical factor enabling proper stem cell function. Fourth, because 47 patients of the study had TAO and only 3 patients had SLE/HES, larger trials with more patients without TAO will be needed to further evaluate the clinical outcomes of patients with AICLI treated with PBMNCs or PCCs. Finally, the relative high cell loss rate (73%) during the purification of CD34 + cells may result from some procedures before connecting to the magnetic cell sorting system. For example, the supernatant of centrifugation after leukapheresis still contained an amount of white blood cells because the transfer bag was not originally designed for centrifugation and was prone to deforming during the process. \| CONCLUSION The results of this study indicated that the use of two types of cells seemed to result in similar mid-term efficacy in treating AICLI, and each kind of cell had its own unique advantages. Although PCCs seemed to result in earlier pain relief, it is not a cost-effective treatment compared with PBMNCs. PBMNCs may be a better choice when two or more limbs require transplantation at one stage. Validation of the conclusions is pending more evidence on the basis of the long-term outcomes of a larger number of patients. CONFLICT OF INTEREST The authors declared no potential conflicts of interest. AUTHOR CONTRIBUTIONS H.L., T.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; Y.F., G.F., Y.L., X.J.: collection and/or assembly of data, data analysis and interpretation; B.C.: administrative support, provision of study material or patients; W.Z.: collection and/or assembly of data; S.G.: provision of study material or patients, collection and/or assembly of data; P.L.: administrative support; Z.D., W.F.: conception and design, final approval of manuscript. DATA AVAILABILITY STATEMENT The data used to support the findings of this study are available from the corresponding author upon request.	2021-01-06T06:18:56.324Z	2021-01-05T00:00:00.000	{ "year": 2021, "sha1": "5445dd730e37ee7fb269b640f8f063abd69284d1", "oa_license": "CCBYNCND", "oa_url": "https://stemcellsjournals.onlinelibrary.wiley.com/doi/pdfdirect/10.1002/sctm.20-0033", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "d7cd42fb46dbb3dfd6835bb31625b463151d247d", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
208857481	pes2o/s2orc	v3-fos-license	Quantifying the potential and flux landscapes for nonequilibrium multiverse, a new scenario for time arrow We propose a new scenario of nonequilibirum multiverse. We quantified the potential landscape and the flux landscape for the Bousso-Polchinski type of multiverse. The potential landscape quantifies the weight of each universe. When the terminal vacuum with zero (flat) or negative cosmological constant (AdS) have a chance to tunnel back to the normal universes with positive cosmological constant (dS) through the bounce suggested by the recent studies, the detailed balance of the populations of the multiverse can be broken. We found that the degree of the detailed balance breaking can be quantified by the underlying average flux and associated flux landscape, which gives arise to the dynamical origin of irreversibility and the arrow of time. We also showed that the steady state of the multiverse is maintained by the thermodynamic cost quantified by the entropy production rate which is associated to the flux. This gives arise to thermodynamic origin of time irreversibility. We show that terminal vacuum universes can have dominant weights or lowest potentials giving arise to a funnel shaped potential landscape, while terminal vacuum universes together with other normal universes including ours can form dominant cycles giving arise to a funnel shaped cycle flux landscape. This indicates that even our universe may not be distinct from others based on the probability measure, it may lie in the dominant cycle(s), leading to higher chance of being found. This may provide an additional way beyond the anthropological principle for identifying our universe. I. INTRODUCTION The conventional Big Bang cosmology theory faces difficulty for interpreting the homogeneity problem, the flatness problem, and the magnetic monopole problem and so on. Inflation theory was suggested to resolve these issues [1]. The essential picture of the inflation is that the universe went through a transient fast accelerating expansion phase soon after the Big Bang. The inflation is often thought as being driven by the vacuum energy/dark energy. If there are different vacuum states, different vacuum can tunnel to each other [2]. Following the inflation of the universe, the process of vacuum decay is reminiscent to the bubble formation in vaporization of liquid to gas phase transitions. However, there are issues associated to this old inflation model. Although the false vacuum can decay to the true vacuum through the bubble nucleation, They cannot have the chance to collide and reheat the universe since they cannot catch up with the expansion of the rest of the inflationary universe. For this, a new inflation theory was proposed to have a slow and continuous transition from the false vacuum to the true vacuum within a single bubble (universe) [3]. The inflation theory predicted the amplitudes and the fluctuation spectrum of the cos- * Email: jin.wang.1@stonybrook.edu mic background radiation as the seeds for large structure formation in our universe consistent with the observations. Furthermore, it was pointed out that the inflation can continue without ending in most part of the universe, giving arise to new bubble universes one after another. Thus, the inflation theory gives arise to a very different picture of the evolution of the universe, leading to a multiverse with eternal inflation [3,4]. On the other hand, in string theory, there are huge number of different kinds of universes with different vacuum and different coupling constants. In fact, there are estimated of 10 500 of such universes or vacuum. Some studies suggest that there is no preferred universe and all universes should be treated on an equal footing [4,5]. There are immediate questions about how our universe can be identified from the enormous amount of the possible other universes [4]. There is another closely related issue. According to the quantum field theory, the vacuum energy is huge. But according to the cosmological observation, the cosmological constant is tiny. The associated issue is why the cosmological constant of our universe is so small out of so many possibilities of the different universes [4]. Anthropological principle has been used to explain the observational existence of our universe [4]. For example, if we have somewhat different universe with different cosmological constant and coupling constants, the galaxies will not be formed properly, our human be-ing will not come to existence to be able to observe our current universe. Therefore, according to the anthropological principle, we happen to live in a universe which can produce ourselves and the observational universe we currently see. This has been used to explain why the cosmological constant of our universe is so small [4]. Despite of progresses being made, there are still challenging issues related to the multiverse picture. One is related to the observational evidences. Suggestions have been made to study the emergence of the black holes as the trace of the possible consequence of the collisions of different universes [6]. The clear evidences remain to be seen. Another important issue is related to the time arrow. How can we determine the direction of time in our universe or the multiverse? Suggestions have been made to explain the time arrow by including the terminal universes with negative or zero cosmological constants [7]. These terminal universes cannot convert to the universes through the vacuum tunneling. But the other universes can tunnel to these terminal universes [7]. Therefore, the terminal universes act like sinks in the multiverse. This produces the irreversibility and the arrow of time. However, the time arrow can only last at finite time during the multiverse evolution. In the long time limit, the time arrow ceases to exist. Recent studies suggested the possibilities of bounces avoiding singularities from a contracting universe (with negative cosmological constant) back to the expanding universe (with positive cosmological constant) [8]. Furthermore, there were also discussions on the possibility of expanding universe (with positive cosmological constant) born from the flat universe (with zero cosmological constant) [9][10][11][12][13][14]. If one takes this into consideration, the terminal vacuum (with zero or negative cosmological constant) will have small chances of coming back to the normal vacuum (with positive cosmological constant). The detailed balance can be broken in this case [8,14]. This suggests a need to study the multiverse problem from a nonequilibrium perspective. In this work, we will study the multiverse evolution from the perspective of nonequilibrium physics. Without the assumption of the detailed balance among the vacuum tunneling switching of the universes, we show that the multiverse evolution is driven by two forces. One force is the underlying probability landscape of the multiverse quantified by the steady state probability distribution in multiverse state space. The other is the steady state probability flux which quantifies the degree of the detailed balance breaking. While the landscape attracts the multiverse down to their steady state basins of attractions, the flux provides a driving force for the cycle flow in multiverse state space. The steady state probability forms a potential or weight landscape while the steady state flux forms a flux landscape of cycles. We found that while the terminal vacuum has largest weight and therefore lowest potential, the largest flux cycles can include not only terminal vacuum but also normal vac-uum with positive cosmological constant including our universe. Therefore, although our universe might not be the highest probability one, it may lie in the highest probable flux cycle, which still gives a higher chance to be observed. On the other hand, the explicit detailed balance breaking characterized by the magnitude of the flux through the possibility of terminal vacuum tunneling back to normal universe provides a source of irreversibility and therefore the arrow of time at all times (even at long time limit). Moreover, the dynamical driving force in terms of flux gives arise to the entropy production rate. In fact, the entropy production rate provides the thermodynamic cost for maintaining the steady state of the universe, which is also an indication of the time arrow. A. Multiverse description To explore the multiverse evolution, let us first consider the evolution of the universe. In general, we can use Einstein's general relativity to describe the evolution of universe. Let us assume that the universe is homogeneous and isotropic on the large scales and treat matter as the perfect fluid [15]. From the Einstein's equations, we can obtain the Friedmann Equations (flat universe) for the evolution of the universe [15]: where a is the expansion factor and H is the Hubble constant representing the rate of expansion, ρ is the proper energy density and P is the pressure. If we assume that the evolution of the universe has an inflation driven by a constant vacuum energy (Λ), then H = Λ/3, a(t)∝exp(H t). The universe rapidly expands. Suppose there are two adjacent places, Inflation will drive them away from each other. Beyond H −1 , there is no causal relation [3]. Within the universe (H −1 ), there are chances of the birth of another universe through quantum tunneling such as the ones illustrated from brown to yellow in Fig 1 [6]. The yellow universe also grows rapidly, but it never catches up with the brown universe. This process continues with the birth of more other universes represented by different colors shown in Fig 1 [6]. As a result, this eternal inflation leads to the multiverse landscape, a foam of expanding bubble universes within bubble universes. It was suggested that the cosmological constant Λ has a discrete spectrum based on the string theory [16], Here Λ bare is the bare cosmological constant. q m is the charge of the string compactified vacuum flux. The range of n m is an integer assumed to be within N is a positive integer. The state characterized by the cosmological constant Λ i has a degree of degeneracy D(i ) [17]: We often view the cosmological constant as the vacuum energy. Different cosmological constants Λ i correspond to different vacuum states. These different vacua can be classified into three categories. The vacuum with positive cosmological constant Λ i >0 is called the de-Sitter (dS) vacuum. The vacuum with zero cosmological constant Λ i =0 is called the Minkowski vacuum. The vacuum with negative cosmological constant Λ i <0 is called the antide-Sitter (AdS) vacuum. In this study, we use the Greek letters to label the dS vacuum and Latin letters to label any vacuum state. There are different definitions to describe the fractions of different vacua [19]. For simplicity, we consider the fraction of comoving volume (P i ). For the description by the fraction of proper volume, see [19,20]. The probabilistic evolution of the multiverse containing various vacua for dS vacua is determined by the master equation [14,17,21] where κ αβ is the transition rate or the transition probability per unit time for an observer who is currently in vacuum β to find himself in vacuum α, it is determined by [8,14,17,21] and Γ αβ is the tunneling transition rate per unit physical space time volume between two different dS vacuum, it is determined by [2,8,17,18] Γ αβ = e −S(β→α)+S β . Here S β = −8π 2 /H 2 β is the Euclidean action for the vacuum state Λ β and S(β → α) is the Euclidean action for the tunneling trajectory β → α which satisfied S(β → α) = S(α → β) [2,8,17,18]. The physical meaning of equation (7) is clear. The change in the chance or the probability of the universe α being observed will be determined by the input from the chances of other universes decaying to the current universe (α) subtracting the output from decaying of the current universe (α) to the other universes. Considering the state with cosmological constant Λ α under the degree of degeneracy of D(α), the equation (8) should be modified as One can easily show that is the steady state solution of the master equation (7) and From equation (11), we can see that the factor S β→α in Γ αβ acts as a scale factor constant and therefore is not important. In addition, we know that S β = −8π 2 /H 2 β and H β = Λ β /3 [17], So we can simply set If all the universes are dS vacuum, the detailed balance is preserved. The resulting steady state is an equilibrium state. Since this is an equilibrium state, there is no emergence of time arrow. To resolve this issue, one can explore the effects of the emergence of the Minkowski vacuum or AdS vacuum. The general master equation for the probability evolution of the multiverse involving all possible universes (dS, AdS and Minkowski) is given by In previous studies [7,17,21], the tunneling transition rates from Mikowski and AdS universes to dS universes are assumed to be zero. Recent studies show that there is a possibility of the tunneling transition from Minkowski or AdS vacuum to dS universes through the bounce [8][9][10][11][12][13][14]. For this study, we simply assume that: In equation (15), state "1" represents the AdS vacuum state. State "2" represents the Minkowski vacuum state. The Latin letters "i" represent any vacuum state. S 1,2,3,4 are certain small constants. In addition, we assume that the formula (13) can be generalized to III. RESULTS A. Potential landscape and flux as the driving forces for the evolution of multiverse 1. Potential landscape and flux decomposition for the driving force dictating the evolution of multiverse. We can write the master equation in (14) for determining the probabilistic evolution of the multiverse involving various vacua in the form of [22] d P dt where Here M is the transition rate matrix representing the transition rates from one state (universe) to another while P (i) represents the fraction of comoving volume of vacuum Λ i . For the steady state solution (P ss i ), M P ss = 0, P ss represents the steady state probability distribution of the vacuum states of the multiverse. We define the potential landscape as [22,23] The potential landscape U can attract the multiverse to the steady state. As seen clearly, the transition rate matrix M determines the evolution of the probability dynamics of the multiverse. We can decompose the driving force into the following form through the symmetrization and antisymmetrization decomposition [24]: and therefore where Here F ss ji = M ij P ss j − M ji P ss i is the local steady state probability flux between i and j. If all the local steady state flux is zero between any i and j states, then there is no net input or output to or from the system. Thus the detailed balance is preserved and the system is in equilibrium state. On the other hand, if any local steady state flux is not zero, then there is a net input or output to or from the system. Thus the detailed balance is broken and the system is in nonequilibrium state. From this decomposition, we can see that the driving force for the probability dynamics is determined by two parts. ∆ is time-reversible and the detailed balance preserved part of the driving force. It is determined by the potential landscape difference or gradient. This is analogous to the usual equilibrium dynamics where the driving force is dictated by the gradient of the potential. Θ is time-irreversible. It is determined by the steady state probability flux. The steady state probability flux can be used to measure the degree of the detailed balance breaking and thus the degree of the time-irreversibility. Thus the flux quantifies the nonequilibrium (detailed balance breaking) part of the driving force of the probability evolution dynamics. If the system is in equilibrium state, Θ must be zero and equilibrium dynamics is determined by the landscape gradient alone. However, if flux is not zero, then the nonequilibrium dynamics is determined by both the landscape gradient and flux. This decomposition is in the discretized representation of the state space. In the continuous state space, the probability evolution can be described by the Fokker-Planck equation, we can decompose the driving force into a landscape gradient and curl steady state probability flux [25][26][27]. Compared to the continuous representation, ∆ is related to the landscape gradient and Θ is related to the curl flux. These two parts determine the evolution of the multiverse. While the potential landscape attracts the multiverse into certain states and provides the stability of these attractors, the flux provides a driving force to form stable flow in the multiverse state space. We can also decompose the driving force of the probability dynamics for the transition matrix M into two components in another mathematical rigorous form as [22]: where One can easily prove that This demonstrates that we can decompose the driving force (M ) for the probability evolution into two parts. One part preserves the detailed balance (D), this force is time-reversible, another part breaks down the detailed balance (C ) and this part is time-irreversibility [22]. Let us define [22] F ji ≡ P ss j M ji − P ss i M ij ≡ P ss j C ji − P ss i C ij (30) as the steady state flux between state i and j. In the next, we will show that the steady state probability flux can be further decomposed to certain cycles to form the cycle landscape of the steady state probability flux. Cycle fluxes forming flux landscapes. Let us look at the flux component of the driving force for the evolution of the multiverse. We can define the flux directly from the definition of the master equation while the probability flux F ji is defined as The master equation can be interpreted as the local conservation equation for the probability. The evolution of the probability is equal to the net flux in or out. On the other hand, at steady state, dP i /dt = 0. If F ji = 0, the net flux is zero. This corresponds to the detailed balance and equilibrium situation. If the F ji is not equal zero at the steady state, then the presence of the local net flux to the system indicates that the detailed balance is broken. The steady state probability flux breaking the detailed balance becomes F ss ji = P ss j κ ij − P ss i κ ji . Since both F ss ij and F ss ji refer to the same net local flux, we can delete this kind of redundancy. For simplicity, one can delete the one which is smaller than zero to reach the following definition [28]: One can easily prove that The definition (34) included all net local fluxes without redundancy. One can easily prove that Except for the null matrix, for any matrix, if it has the property (36) and (37), then one can always decompose the flux into the flux cycles (or flux loops) [22,28]. The procedures are outlined as follows. Suppose J i1i2 > 0, then from (37), one can find One can easily prove that J ij still has the property (36) and (37). Therefore, one can repeat the above process and get α 2 , r (2) , J (2) , α 3 , r (3) , J (3) ,... until J (M +1) = 0, where M is a finite positive integer. We now have Therefore, from the global perspective, the steady state flux can be decomposed to many cycles or loops of circulation fluxes. This forms the nonequilibrium flux landscape. B. The potential landscape in the comoving coordinate of multiverse According to our definition of the potential landscape, the potential landscape spectrum is shown in Fig 2, Fig 2 is plotted under the choice of the parameters for the string vacuum and transition among AdS, Minkowski and dS vacuum as J = 7, q 1 = 2.5, q 2 = 5.185, q 3 = 5.155, The rational of the parameter choice is as follows: First, we need to specify AdS and Minkowski vacuum states, respectively. Second, the transition rates from AdS or Minkowski vacuum to the dS vacuum is assumed to be small compared with the transition rates in the opposite direction. Lastly, we assume that the transition rate from Minkowski vacuum to AdS vacuum is larger than the rate in the opposite direction. If we arrange all these states of the multiverse on a 2 dimensional plane, the potential landscape can be shown in We notice that there is a characteristic of the energy band (Fig 2). The reason for the presence of this band is that the string theory inspired parameters q 2,...,7 are close to each other. When these string theory implied parameters become closer, the resulting characteristics of the band become clearer. Under these string parameters, the cosmological constant spectrum has the characteristic of the band. This can lead to the band structure in the potential landscape. If there is a significant gap between the minimum and average of the potential landscape spectrum compared with the dispersion or variation, the potential landscape topography can be biased towards the dominant vacuum at the bottom. One can use a characteristic ratio for landscape topography to measure the degree of this bias [22], defined as Here < U > is the average value of the potential landscape, < U 2 > is the average value of the square of the potential landscape, and U m is the lowest potential energy value. While the numerator represents the gap, the denominator represents the fluctuation characterized by the standard deviation. RR thus represents the ratio of the gap between the lowest and the average potential of the universes against the fluctuations characterized by the standard deviation through the variances. While the gap can be viewed as the slope or the bias towards the dominant vacuum, the fluctuations through the variances can be viewed as the measure of the roughness or traps of the potential landscape. A large ratio of RR indicates a landscape with large bias towards the dominant universe against the roughness or traps. That is the landscape of the multiverse has a funneled shape towards the bottom terminal vacua (AdS and Mikowski). It also shows that the dominant state sitting at the bottom of the landscape is distinct and discriminant from the rest of the universes and therefore stable. In this sense AdS and the Minkowski vacua (universes) are usually more stable than the dS vacua (universes). We found that when N =2, RR(U ) = 3.8166 and when N =3, RR(U ) = 12.6087. These two values are significantly larger than 1. This indicates that the AdS vacuum is dominant and stable against others. In this figure, the blue curve is for N =2 and the red curve is for N =3. We can see that when S 2 becomes smaller, the RR(U) at first do not change significantly and then becomes larger. This indicates that when S 2 become smaller, the dominant vacuum becomes more distinct and discriminant from others and therefore stable. It is worthwhile to point out that our numerical simulation shows that when S 2 changes from 10 −6 to 10 −14 , the dominant vacuum is not always the AdS vacuum. In addition, we noticed that these two curves with N =2 and N =3 have similar trends against the vacuum transition rates from AdS to the other universes. Here, V f represents the value of the flux in a loop. The red line represents the cycle with the dominant flux. There are altogether 5073 flux loops or cycles in this multiverse connecting different universes together. The flux landscape is illustrated in the right panel of the Fig 5. For the purpose of clear view, we do not present all the flux loops but only a few dominant ones. Here, different nodes represent different states or universes. The thickness of the arrows represents the magnitude of the circulation flux in the loop. The size of node represents the weight or the steady state probability of the node or the specific universe. The dominant black node represents the AdS vacuum. The second dominant black node represents the Minkowski vacuum. Other small gray nodes represent the dS vacuum. We noticed that although there are 128 states (N =2), there are only 127 nodes. The third state does not appear in any loop. This is not strange since under these parameters,J i3 = 0, for any "i". Therefore, the third state cannot appear in any loop. The cycle involving the red arrows represents the dominant loop. Since the dominant flux loop stands out from the rest of the others, it represents a limit cycle oscillation in multiverse state space. The orange arrows represent the second dominant loop and the green arrows represent the third dominant loop, while the blue and purple arrows represent the fourth and fifth dominant loops, respectively. The second, third, fourth and fifth limit cycle oscillations can also emerge but with much less chances than the dominant one. This gives us a new angle of looking at the organization of the universes. The universes in the multiverse forms a network. Through the tunneling connections, the network of the universes is organized in a hierarchical fashion with the bottom layer or ground flux state as the dominant flux cycle connecting certain universes to- gether. Then the hierarchical structure towers are built up by the subsequent less dominant flux cycles layer by layer or excited flux states. This new structure of the network universe indicates that the universes may emerge or function in a clustered fashion in the form of cycles. Robustness of the flux landscape Similar to the definition (40), we can use the definition [22] to measure the shape of the flux landscape. < U F > represents the average value of the flux landscape while < U F 2 > represents the fluctuations as the average value of the square of the flux landscape. The U F m represents the smallest value of U F . The RR is defined as the ratio of the gap between the minimum (dominant) flux potential and the average of the flux potential against the standard deviation characterizing the fluctuations in the fluxes. A high value of RR indicates a high discrimination of the dominant flux cycle/loop against the other ones, leading to the limit cycle oscillations among the universes in the dominant flux cycle/loop. In Fig 5, RR(U F ) = 11.1334, giving arise to a distinct dominant oscillation flux cycle/loop. Although the numerical results are different with different input parameters, the flux landscape can always emerge, as long as the detailed balance is broken and S 2 = 0. If S 2 = 0, this corresponds to terminal vacuum without the transitions to the other universes. Then the irreversibility comes from the time evolution and ceases to exist at long time steady state. The flux landscape provides a new scenario and perspective for the evolution of our universe. First, due to the detailed balance breaking, the emergence of the flux gives arise to another driving force for the evolution dynamics of the multiverse. Second, the steady state flux can be decomposed to directional loops or cycles of fluxes. This generates the irreversibility and therefore the direction of time due to the intrinsic nonequilbrium nature of the detailed balance breaking. Note that this mechanism of time arrow is quite different from the one suggested before on the time evolution under terminal vacuum assumption where at the long time limit the evolution reaches the equilibrium preserving the detailed balance. Third, another distinct feature is that the dominant steady state probability flux cycles or loops with higher chances being seen not only involve the universes with negative or zero cosmological constant but also with the normal universes with positive cosmological constant. This indicates although the weight of our living universe may not stand out from the rest of the universes in the multiverse, it can be involved in a dominant flux cycle which increases its chance of being observed. This may help to pick up our living universe with less dependence on the anthropological principle for resolving the cosmological constant problem. Fourth, the global nature of the multiverse evolution dynamics is determined by both the underlying potential landscape and flux landscape. Fifth, the flux landscape gives arise a hierarchical structure of the organization of the universes in terms of flux cycles (or loops). Sixth, this gives a new picture of the universe evolution. From the viewpoint of the dominant cycle, the birth of our universe may be due to the transition from another universe and the death of our universe may be due to the transition to another universe. The birth and the death of the universe occur in a periodic fashion. In this sense our universe never dies. It repeats itself over and over again at a fixed amount of times by emerging from or transforming to the other universes on the dominant cycle, much like the case of a biological cell cycle [22]. From this perspective, our universe is eternal. D. The irreversibility, the thermodynamic dissipation and arrow of time To address the irreversibility or time arrow, let us study the irreversibility of a particular trajectory. If there exists the irreversibility of any one of the trajectories, then there is the irreversibility of the whole system. Given a particular trajectory specified as We consider the following functional [29] R = R({σ}, P (σ 0 ), P (σ n )) = ln[ κ σn,σn−1 ...κ σ2,σ1 κ σ1,σ0 P (σ 0 ) κ σ0,σ1 ...κ σ1,σ2 κ σn−1,σn P (σ n ) ] to measure the irreversibility of the trajectory (43) since this gives the ratio of forward transition probability against the backward transition probability. An equal forward and backward transition probability will result in an equilibrium with a zero value in R. This indicates the time reversibility. The magnitude of R gives a measure of how forward rates are different from the backward rates in time. Therefore, R is a measure of irreversibility or time arrow. Fig 6 shows the value of R or irreversibility versus the variation of the parameter S 2 for some trajectories where EPR is the entropy production rate [31]. the precise relation between EPR and the irreversibility revealed by the famous fluctuation theorem [32].We noted that This is the connection between the EPR and the steady state probability flux. The steady state flux gives the dynamical origin of the nonequilibriumness through the detailed balance breaking, while the steady state EPR provides the thermodynamic origin of the nonequilibirumness for maintaining the steady state. We can define that For the potential landscape in Fig 2, when N =2, < R > 1 = 2.3747 × 10 −9 ; when N =3, < R > 1 = 1.4419 × 10 −8 . Again, the entropy production rate is not zero indicating the nonequilibrium thermodynamic cost is needed to maintain the irreversibility and time arrow. Fig 7 shows the first (average) and second order (fluctuation) of < R > versus the variation of the parameter S 2 . Larger transition rate from AdS to other universes will give arise to more degree of detailed balance breaking The green curve is < R >2 versus the variation of -lg(S2) for N =2 multiverse. The red curve is < R >1 versus the variation of -lg(S2) for N =3 multiverse. The black curve is < R >2 versus the variation of -lg(S2) for N =3 multiverse. and therefore the arrow of time. The blue dotted curve and the green dash-dotted curve represent the < R > 1 and < R > 2 respectively for N =2. Similar two curves representing the < R > 1 and < R > 2 respectively are plotted for N =3. Fig7 shows that the following approximation is a good one: Therefore, we can use both < R > and EPR to describe the irreversibility of the system. In other words, < R > and EPR can provide the thermodynamic measure of the arrow of time in comoving coordinate. Here, the origin of the time arrow is the detail-balance breaking. This is different with the time arrow in [33] which comes from the dynamic evolution of our universe. We should point out that when S 2 is strictly equal to zero, the AdS vacuum and the Minkowski vacuum become the terminal vacuum. They cannot decay to another vacuum. In this case, the formula (51) or (52) cannot be directly used to describe the arrow of time. Because in this case, the fraction of the AdS vacuum is 1 and others are zero. The formula (51) and (52) are ill-defined. IV. CONCLUSIONS In this work, we studied the nonequilibrium evolution of the multiverse in the comoving coordinate in detail. We uncovered that the driving force of the multiverse evolution in general is determined by the global landscape quantified by the steady state probability distribu-tion and the steady state probability flux. While the landscape attracts the multiverse to the steady state basins, the flux gives arise to cycles and loops associating certain universes together. The emergence of the flux loops provides a signature and a quantitative measure of the degree of detailed balance breakingnonequilibriumness. In other words, the landscape of the multiverse is uncovered and quantified in this work. However, the multiverse evolution dynamics is driven by not only the landscape but also the flux. This is very different from the conventional picture of the dynamics determined by the landscape gradient for equilibrium systems under detailed balance. The vacuum of the universes with positive cosmological constant can be transformed from each other through tunneling. Recent studies show the possibilities of the bounce of contracting universe with negative cosmological constant back to the expanding universe with positive cosmological constant [9][10][11][12][13]. This can lead to the breaking of the detailed balance [8,14]. The flux associated with the detailed balance breaking provides the dynamical origin of the irreversibility and time arrow. On the other hand, the entropy production is associated with the flux. Thus, it provides the thermodynamic origin or cost for maintaining the irreversibility and time arrow. In contrast to the time evolution argument of the underlying detailed balance system with terminal vacuum for generating the irreversibility, the current approach emphasizes the possibility of the time arrow generated by the detailed balance breaking even at long time steady state. This provides a new mechanism for the time arrow for the multiverse. Furthermore, the potential landscape shows a funneled shape dominated by the contracting universes with negative cosmological constant and flat universes with zero cosmological constant. On the other hand, the flux loops form a flux landscape. The flux landscape also shows a funneled shape dominated by certain flux cycles or loops. The dominant flux loops or cycles can involve not only the contracting and flat universes, but also expanding universes with positive cosmological constant. The dominant loop gives arise to the associations of certain different kinds of the universes together and this leads to the oscillation cycles among these universes in the multiverse evolution. In other words, a universe on the dominant cycle can appear due to the transition from the other universes on the same cycle and can also disappear due to the transition to the other universes on the same cycle. This occurs periodically. If our universe is on this dominant cycle, then it can be born due to the transition from another universe and die due to the transition to another universe in a periodic manner. In this sense, our universe never dies. The birth and death of our universe go through cycles. Therefore, the universes on the dominant cycle oscillate from one universe to another and repeatedly appear and disappear under a fixed period of times. Moreover, although our living universe with small pos-itive cosmological constant does not necessarily have significantly higher probability compared to others in the multiverse, it can lie in the dominant flux cycles. If that is the case, then the chance of being observed can be significantly enhanced. Thus, this may provide a boost in addition to the anthropological principle for selecting our living universe. The presence of the steady state probability flux breaks the detailed balance. This leads to the multiverse as an intrinsic nonequilibirum system. Even in steady state, the equilibrium is not reached since the detailed balance is not preserved and there is a net flux. In other words, the time arrow originated from this intrinsic nonequilibriumness exists at all times (even at the long time limit). This is in contrast to the case where the time arrow is generated during the evolution of the multiverse and ceases to exist at the long time limit.	2019-12-06T15:13:50.000Z	2019-12-06T00:00:00.000	{ "year": 2021, "sha1": "24217204bf1ed8251760ed8b75705d6e6df0b796", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/JHEP02(2021)105.pdf", "oa_status": "GOLD", "pdf_src": "Arxiv", "pdf_hash": "6fa8e8f29bba1880b62d3440eaead91154f32c04", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
16472227	pes2o/s2orc	v3-fos-license	Mouse Oocytes and Embryos Cryotop-vitrification Using Low Concentrated Solutions: Effects on Meiotic Spindle, Genetic Material Array and Developmental Ability Objective(s): The examination of the possibility of applying lower CPA- concentrations and obtaining the similar results to those using higher concentrations; as it is shown, the toxicity of the CPAs used in vitrification approach will diminish. Materials and Methods: Following vitrification/warming, oocytes were subjected to PZD/ICSI. SRs, FRs, and DRs were recorded. SRs and DRs of the embryos were monitored after vitrification/warming. IHC studies were done. Data were analyzed in comparison to the data of Exp. (experimental groups) applying 1.5 M CPA- concentrations (largely-used concentration). Results: The data of oocytes exposed to 1.25 M concentrated CPAs were in consistency with those exposed to 1.5 M and fresh oocytes in terms of SRs, FRs and DRs. Normal spindle and chromatin configuration is in consistence between the two experimental groups, but lower in comparison with control group. The lower the concentrations were, the less SRs, FRs, DRs were. Also, spindle organizations were more normal in comparison with the experimental groups as the concentrations decreased. The results of DRs for embryos which were exposed to 1.25 and 1.0 M concentrated CPAs were close to those vitrified with 1.5 M and fresh embryos but IHC observations in the three Exp. were significantly lower than those of fresh embryos. The results of 7.5 M concentrated CPAs solutions were significantly lower than those of the control group 1.5, 1.25 and 1.0 M treated. Conclusions: Vitrification by cryotop technology using minimal volume approach increases both cooling and warming rates, therefore, the CPAs limited reduction to 1.25 and 1.0 M instead of using 1.5 M for oocytes and embryos cryotop-vitrification procedure, may be a slight adjustment. Introduction The risk of multiple pregnancies often presents in IVF programs and the existence of factors in stimulated uterine cycle which may jeopardize implantation, are important forces in perfecting embryo cryopreservation.On the other hand, the ethical restriction and assurance of plastic pathologic condition have led scientists to focus on female gamete preservation (1).During ultrarapid freezing (Vitrification), the liquid is so rapidly cooled that it forms into a glassy, vitrified solid state from the liquid phase at low temperature, not by ice crystallization, but by extreme elevation in viscosity during cooling (1).Basically, vitrification approach totally eliminates one source of chilling injury, ice crystal formation, and avoids zona fracture (2); however, it exposes cells to a considerably elevated toxic and osmotic effect (3). The normal physiology of the freezed-oocytes and embryos is detected by successful developmental consequences which in turn is dependent on having normal spindle apparatus and genetic material array in cells.The probable cause of reduced viability of cell during freezing and thawing is the disruption of cell organelles, particularly the CSK organization and genetic material array, as a result of intracellular ice formation (4).On the other hand, preserving freezed-gametes and qualified embryos until the time that they need to be used, is fundamental in all cryostorage methods (5).Researchers consider the integrity of the CSK structure and cell functionality as a valuable parameter for the quality of frozen-thawed cells (6).Obviously, decreasing CPA-concentrations and therefore the toxicity would be a step toward the higher security level of the cryopreservation technique (7). Researchers have introduced various cryopreserving methods and protocols through changing freezing solutions, introducing new different cryocontainer and examining various cooling/warming rates all over the world.Above all, preserving freezed-gametes and qualified embryos until they need to be used is fundamental in all cryostorage methods.Would it be possible to decrease cryoprotectants concentration and therefore the toxicity, in order to develop higher security cryopreservation technique? In the present study, through investigating developmental consequences, spindle apparatus morphology as well as the genetic material array of cryotop-vitrified mouse M-II oocytes and four cell stage-embryos using low concentrated CPAs solutions, we re-examine the vitrification protocol and improve the technique. Materials and Methods This was an experimental study.Animals were taken care of according to the university guide for the care and the use of laboratory animals, including 12/12 light/dark, 18-22°C and fed based on a diet containing straw and barley grains.All chemicals were purchased from Sigma unless otherwise stated. Reagents and media The Vitrification kit was purchased from Kitazato Biopharma, Mitojima, Japan.The medium for oocytes manipulation and embryos culture was Hypermedium, as Eroglu et al indicated (8).Before use, the drops of the Hypermedium were overlaid with embryo-tested mineral oil and equilibrated overnight under a humidified atmosphere of 6% CO 2 in air at 37 o C. Egg and embryo collection Oocytes and embryos were obtained from 8 to 10 wkold C57BL/6J mice; Pasteur Institute animal house, Iran.Superovulation was induced as described previously (9).M-II oocytes and four cell stage-embryos collecting techniques were detailed elsewhere (7). Vitrification/warming M-II oocytes and four cell stage-embryos were vitrified/ warmed by the minimum volume cooling method using cryotop as exactly as mentioned by Kyono et al (10).There were five or six oocytes/embryos on each cryotop.The oocytes/embryos with a poor grade, irregular contours, dark coloration, or fragmented unequal blastomeres were excluded from cryopreservation.Cryostorage was done in LN2 for 5 days.Re-expanded oocytes/embryos were considered to have survived and transferred to the Hypermedium for recovery before experimentation. The fertilization and development of cryopreserved oocytes Sperm preparation was performed as previously described (11).Oocytes PZD and ICSI were performed after 1 h-incubation period following the method of Balaban et al with brief modifications (12).The zona pellucida of cryosurvived oocytes (assessed by the translucent appearance of cytoplasm, the integrity of the plasma membrane and the zona pellucida, the size of the perivitelline space and extruded polar body) was dissected on 10-15% of its circumference with a fine glass needle, far from the polar body area at 37 o C. Morphologically normal, motile spermatozoa were randomly selected for ICSI.Separated sperm heads were injected into the PZD-oocytes at 17 o C on a pre-cooled inverted microscope.Injected oocytes were left to rest for 20 min.at 17 o C followed by 15 min.at room temperature (RT).All oocytes after ICSI were washed four times in Hypermedium 0.4% BSA (w/v).Fertilized oocytes were cultured in a 60 mm tissue culture dish in 20 µl of Hypermedium 0.4% BSA (w/v) under mineral oil.The presumptive zygotes were developed to two cell-stage embryos, which were scored 6 and 24 hr after insemination, respectively. The culture of cryopreserved embryos Cryosurvived embryos (assessed by morphologically normal blastomeres with apparent zona pellucida integrity) were cultured in a 60 mm tissue culture dish in 20 µl of Hypermedium 0.4% BSA (w/v) under mineral oil with incubation at 37 o C under a 6% CO 2 atmosphere.Within a few hr, the cryosurvival of the embryos was assessed by their morphological appearance.Then, the developmental competence of the embryos was evaluated by their ability to develop to 8-cell, morulla, young blastocyst and hatching/hatched-stage blastocyst after 48, 72, 96, 100-120 h in culture medium, respectively. Spindle and chromatin visualization Spindle and chromatin configuration were assessed using the fluorescent methods as follows: a number of both treated and fresh oocytes/embryos immediately after warming and the other ones after one hr-incubation period, were fixed in prewarmed (37 °C) 2% PFA in DPBS with 0.04% Triton X-100 for 1 hr.They were then washed in washing buffer DPBS containing 0.1% BSA (w/v) two times (15 min.each) and left in Triton X-100 0.1% solution for 30 min.for permeabilization.Non-specific binding was blocked using goat serum 10% for 1 hr.To spindle/chromatin visualization, fixed oocytes/embryos were incubated in mouse monoclonal anti-α-tubulin (1/100) in DPBS with 0.05% BSA (w/v) for 1 hr at 37°C.Oocytes/embryos were then washed and stained with FITC conjugated to goat anti-mouse (1/150) in DPBS + 0.05% BSA (w/v) for 1 hr in the dark.After the sec wash (30 min.each at RT), samples were stained for chromatin with DAPI (1/1000) in DPBS + 0.05% BSA (w/v) for 30 min, in the darkness at RT, mounted on glass slides.The localization of meiotic spindle and chromatin revealed by FITC and DAPI fluorescence using a fluorescence microscope (Leica DMIL; Germany) with optical filters specific for the wavelengths of 450-490 nm and 330-380 nm to detect the green signal of FITC and the blue signal of DAPI. Morphologically normal meiotic spindles and chromosome alignments in oocytes were barrel-shaped with two distinct poles, and metaphase plate-aligned compact group, respectively (Figure 1).Any other configurations were considered as abnormal (Figure 2). The evidences of the morphologically normal chromatin arrangement in vitrified embryos, in comparison with the images of fresh embryos were observed, in a way that the existence of four compact masses of chromatin, each in one blastomere (Figure 3) was considered normal and any other configurations were considered abnormal (Figure 4).Furthermore, according to the method described by Zenzes et al (13), the amount of fluorescent MTs in treated embryos being as near as possible to those of the control embryos (Table 4) provides explanation to consider MTs in vitrified embryos morphologically normal.In panels B and C, in addition to abnormal localization, the spindles appear not to have distinct poles.In panel B, in spite of one compaction, various compact masses of chromatin exist.In panel D, in contrast to complete depolymerization of spindle microtubules, it seems chromosomes did not become dispersed.Rather, they remained close together in disorganized bundles order.Similarly, vitrified embryos assigned to Exp. 1 to 5, were cultured in vitro after warming.Non-vitrified fresh oocytes and embryos were considered as control group.Experiments in all series were repeated at least seven times to evaluate developmental ability and four times to assess immunostained organelles. Statistical analysis The mean percentage of the differences in the rates of survival, fertilization, further development, having normal spindle apparatus and genetic material array between the control and treatment groups were tested for significance through one-way ANOVA and LSD as post hoc test.The level of significance was set at less than 0.05. Developmental competence Oocytes in Exp. 1 (larger-used CPA-concentrations ) showed a statistically non-significant difference in SR with Exp. 2 and significant higher SR with the other Exp.(P<0.001).Comparing the results of Exp. 3, 4 and 5 revealed statistically significant decreased SRs in proportion to using lower CPA-concentrations (P<0.001).In other words, the less CPAs concentrated solutions were used, the less SRs were obtained.When we compared FRs, the control group and Exp. 1 and 2 exhibited no difference, although FRs in Exp. 1 and 2 were lower than fresh oocytes.The FR in Exp. 3 was significantly lower than those recorded for fresh and Exp. 1 and 2 oocytes (P<0.001).None of cryosurvived oocytes in Exp. 4 were fertilized.The DRs to two cell-stage embryo were not different between the control group and Exp. 1 and 2 oocytes.DR to two cellstage embryo in Exp. 3 was significantly lower than the control group and Exp. 1 and 2 (P<0.001).Table 1 provides the details of our observations after using different freezing solutions in order to vitrify mouse mature oocytes. Embryos in Exp. 1 showed a statistically non-significant different SRs in Exp. 2 and 3 and significant higher SR in Exp. 4 and 5 (P<0.001).Cryosurvived embryos in Exp. 4 showed significant lower SR in Exp. 1, 2 and 3 (P<0.001).There were significant differences between embryos SR in Exp. 5 and the rest of Exp.(P<0.001).There were no statistical differences between DR to 8 cell-stage embryos of Exp. 1 and Exp.2 compared to the control group, whereas DRs of Exp. 1 and 2 were lower than the control embryos.DR to 8 cell-stage embryo in Exp. 3 was significantly lower compared to the control group (P<0.001) and Exp. 1 (P<0.001)and Exp. 2 (P<0.05).In addition to significantly lower results in comparison to Exp. 3, the description was the same for Exp. 4 (P<0.001).None of cryosurvived embryos in Exp. 5 were developed further.The rates of embryos developed to morulla were lower, but not-statistically different in control group and Exp. 1, 2 and 3. DR to morulla in Exp. 4 was significantly lower than that recorded for fresh and Exp. 1 (P<0.05),whereas, there was no significant difference compared to Exp. 2 and 3.The rates of embryos developed to young blastocyst-stage in all the vitrified groups were lower than those of the control group, however, the difference between the control group and Exp. 1 and Exp. 2 were not statistically significant.DR to young blastocyst in Exp. 3 and 4, were significantly lower compared to the control group, Exp. 1 and 2 (P<0.001).The rates of embryos developed to hatching/ hatched blastocyst-stage in all the vitrified groups were lower than those of the control group; however, the difference between the control group and Exp. 1, 2 and 3 were not statistically significant.DR to hatching/hatched blastocyst in Exp. 4 was significantly lower compared to the control and vitrified groups (P<0.001).Table 2 summarizes DRs of fresh versus vitrified embryos. Immunostaining observations Not all fresh oocytes showed normal spindle and chromatin configuration immediately after warming; but the number of fresh oocytes with normal spindle apparatus after 1 hr-incubation period was larger.There was no significant difference between these two records at various times after warming (P<0.001).This was in contrast with the number of oocytes with normal spindle organization in Exp. 1, 2, 3 and 4 at two different times, which displayed significant differences (P<0.001).When comparing the results of Exp. 1 and 2 oocytes, it was noted that the number of oocytes with normal spindle organization both immediately and after 1 hr were similar but significantly lower than those of control oocytes (P<0.001).The proportion of oocytes with normal spindle and chromatin b The fluorescence index is the value calculated for a group of embryos vitrified with indicated concentrated solutions; the value refers to the number of embryos rated to exhibit a given fluorescence intensity multiplied by the number of embryos with that rating; that quantity is divided by the total number of embryos in that group. There were significant decreases in fluorescence intensity of tubulin in vitrified embryos immediately after warming.As shown in table 4, the fluorescence intensity of tubulin increased following one hr-incubation period in 37 o C.Even after incubation, the fluorescence indices of tubulin in Exp. 4 and 5 were lower than those of control, Exp. 1, 2 and 3 which were not significantly different. Obviously time-dependent restorations in genetic material array were apparent in the control and Exp.groups.However, it were statistically significant for Exp.(P<0.001) and not significant for the control group.When comparing the results of Exp. 1, 2 and 3 embryos, it was noted that the number of embryos with normal status both immediately and after one hr were similar but significantly lower than those of the control embryos (P<0.001).The percentage of normal embryos in Exp. 4 and 5 were significantly lower than the control group, Exp. 1, 2 and 3 (P<0.001)at both different times.Table 5 shows the details of our observations. Discussion Although there have been numerous studies on vitrification of mouse oocytes and embryos, the majority of them have used at least 1.5 M concentrated CPAs as freezing solution.The purpose of the experiment described herein was to examine the possibility of the applying lower CPA-concentrations and obtain the similar results to those using higher concentrations.As it is, the toxicity of the CPAs used in oocytes/embryos vitrification approach will be diminished. In designing the experiment, we considered the earlier findings published by Tucker et al (1).The actual cooling rate during vitrification, and therefore, the efficiency, may still vary extremely depending on the device used (1).Regarding to the capability of the new tool, cryotop, to allow for an even smaller volume of vitrification medium (<0.1 µl) to be used and therefore yield quicker cooling and warming rate (23,000 o C/min and 42,000 o C/min) ( 14), it appears logical to assume that it is an adjustment to use CPA agents at lower concentration, while maintaining the necessary concentration to achieve vitrification. To avoid a degree of uncertainty surrounding the outcome of the IVF procedure and to achieve success to over- come infertility, using the most qualified gametes and embryos plays the central role in the ART program (15).Cryopreservation protocols' efficiency is evaluated by the fact that how much they are able to preserve the quality of the freezed-gametes or embryos (16).Regarding efficiency, assessing current vitrification protocols is not an exception.Therefore, we decided to investigate one of the qualified preserving indicators, spindle apparatus and chromatin array to focuse on cryoservived oocytes fertility and embryos developmental ability by the immunostaining technique.The biophysical detail of CPAs and the mechanisms of freezing/thawing rates are beyond the scope of this paper.Briefly, it is noted that CPAs are organic solutes that help to protect cellular organelles during cryopreservation although they may damage the CSK system as they can be toxic and cause disruptive osmotic damage to the cell (17).Novel approaches have been tested to reduce the toxicity of various solutions that are to be used to vitrify oocytes/embryos.One of the candidate CPA agents was EG, which was very effective and less toxic for mouse oocytes vitrification (18).Kartberg et al realized that vitrification with DMSO protects embryo membrane integrity better than solutions without DMSO (19).The incorporation of DMSO into an EG containing medium has at least two advantages: firstly, vitrification is facilitated because of the greater glass-forming characteristics of DMSO; secondly, the permeability of each CPA is enhanced in the presence of the other (20).Therefore, we were more attracted by the current mixed vitrification solution. Cobo et al obtained excellent 96.9 % SR after vitrification through applying the cryotop method and usual CPAconcentrations (1.5 M) to human oocyte (21).Kuwayama et al and Katayama et al have reported a 91% and 94% SRs, 81% and 90% cleavage rates (CRs), respectively (22)(23).Morato et al have scored 94.5% SR and 46.1% CR (24).100% morphologically survived and 93% CRs of human pronuclear stage vitrified embryo are the highest published results so far (14).Above-mentioned were the teams with dramatic improvement in their cryostorage method.In the current study, the SR, FR, and CR of the oocytes, which were subjected to 1.5 M and 1.25 M of CPAs (Exp.1 and 2) and further development of embryos in similar Exp., were near to those findings.This seems to support the claim that using 1.25 M DMSO+EG for vitrification medium containing 0.5 M sucrose and cryotop as cryocontainer, we are able to obtain the findings comparable with largely-used higher concentrations (1.5 M).We can take credit for our claim because of the immunostaining results of Exp. 2 oocytes/embryos following 1 h-incubation period, which were in consistency with those of Exp. 1. Comparing the IHC results of these two Exp.immediately after warming has shown remarkable spindle and chromatin organization anomalies which were effectively repaired after 1 h incubation at 37 o C, though they were small but significantly lower than those of control group. Following fresh oocytes collection, images showed 17% spindle organization anomalies.To find out the underlying cause of these, we came across the report which had explained adverse effects of hormonal stimulation, asynchrony in nuclear and cytoplasmic maturation under in vitro condition and oocyte manipulation in laboratory, on spindle apparatus abnormalities (13).The egg collection protocol, laboratory condition and used culture media for the present experiments were not excluded from the accepted items.Similar findings were obtained from embryonic control group stating adverse effects not only on oocytes but also on embryos. In addition to centrosomes associated with the meiotic spindle, mouse oocytes contain MTOCs in the cortical cytoplasm (25).Disruption of the meiotic spindle of mouse oocytes can be reversed after rewarming and incubation at 37 •C (25).The absence of centrosomes and MTs in the cortical cytoplasm (such as those that occur in mouse oocytes) is believed to be responsible for irreversible damage to the meiotic spindle after cooling of human, bovine and monkey oocytes to low temperatures (25).Chemical and physical stresses have been shown, in several species, to affect the MT structure of the oocyte meiotic spindle with deleterious consequences on chromosomal organization (26).Oocytes incubation following freezing/thawing for specific times relieves physical and chemical stresses and results in higher intracellular homeostasis (27). Restored spindle organization rates of all treated oocytes in current study were not alone in agreeing the positive effects of 1 h-incubation period after warming (28)(29), but what deserves attention is the differences in the rates of cryosurvived oocytes and those in similar experimental groups which had normal spindle configuration after incubation.Because of the technical variety of these results, we can not firmly conclude, but these findings argue that the oocytes viability after warming is not meant to be all intracellular structures, miotic spindle in particular, in healthy condition.These may evoke some researchers' emphasis on investigating morphological, ultrastructural and molecular status of vitrified cells as well as what commonly is done by evaluating developmental consequences (26,28).The positive effects of postwarming incubation period on embryos was revealed by rising fluorescent MTs amount particularly in Exp. 1, 2 and 3, which were close to those of the control group.As rising MTs amount after incubation, genetic material array returned to more normal status in all Exp.Embryos.Results were similar in Exp. 1, 2 and 3 but statistically lower than those of control group.Using low CPA-concentrations of 1.25 M and 1.0 M, results in similar findings of using 1.5 M CPAs at the embryo CSK level.This optimism takes credit from evaluating developmental competence of Exp. 3. The statistics showed small but statistically significant decrease in 8-cell stage (as resuming development after warming and prepare to compact) and young blastocyst stage (as preparing to produce inner cell mass) but the SRs and DRs to progressed stages did not show any significant trends between Exp. 3 and control, Exp. 1 and 2. According to the results of embryos treated by 1.0 M CPAs (Exp.3) and two early studies showing that mouse embryos can be frozen using lower concentrations of CPAs (i.e., 1.0 M DMSO and 1.2 M EG) with good success rates (9), further studies are stimulated. More research and study should be conducted to demonstrate the proof-of-principle that mature mouse oocytes and embryos cryopreserved using reduced concentration of CPAs can develop to term.Proving the applicability of any protocols needs the protocols to be tested on preserving other cryosensitive oocytes of mammalian species, stage-dependent sensitive embryos to damage during vitrification and develop the embryos in vitro (20,30). The results of developmental competence and spindle organization of Exp. 3 stated that improper CPA-concentration (1.0 M) for ideal oocytes vitrification was adopted.CPA-concentration reduction, lower than 1.25 M (Exp.4 and 5), breed irreversible effects on spindle apparatus and developmental ability in proportion to CPA concentrations.Although comparing the FRs and the percentages of oocytes with normal spindle apparatus after incubation of the Exp. 1 with the ones for Exp. 2 and 3 did not show significant differences; 63% of oocytes were with normal spindle structure in Exp. 3 in contrast to their fertility limited to 38%.As this data has been detected in lower CPA-concentration condition, appropriate CPA-concentration at freezing and sucrose concentration at warming procedure should be designed to head toward more effective cryopreservation.Mullen et al have shown M-II oocytes can maintain a normal spindle structure after exposure to widely range of osmotic conditions (31).On the other hand, developmental potential is reduced by about 50% across all test conditions.CPA and sucrose concentrations equilibration should not expose cells to osmotic changes more than they can tolerate.According to the other investigations (31)(32)(33) and the fact that reduction of CPA-concentration preserves oocytes CSK in a safer way (21), there are various organelles and processes that contribute to fertilization that is negatively affected in such a condition.Further studies may be conducted on removing CPAs procedure by lower concentrated sucrose solutions and numerous dilution steps, as per using lower CPA-concentrations as freezing media. Kim et al have noted that MTs and MFs are integrated during fertilization (34).These CSK element interactions are required for the union of sperm and egg nuclei and subsequent cell division.In mature mouse and rat oocytes, MFs are mainly located in the cell cortex overlying the meiotic spindle.This domain, rich in MFs, appears to be responsible for maintenance of the meiotic spindle and chromosomes in a peripheral position.Whenever MF-rich domain is segregated, the spindle and chromatin migrate toward the center the egg (34).Images were taken from Exp. 3 demonstrating migration of the spindle apparatus from the subcortical area to the center.This, as documented earlier, may be the major cause of the formation of two female pronuclei without polar body extrusion (34), which was considered as fertilization indicator in this study.Our results are in contrast with those presented by Paynter (35), absent or abnormal spindles were observed in all of the oocytes cryopreserved which survived but failed to fertilize.As Isachenko and Nayudu have suggested (36), we set out discrepancies between higher proportions of oocytes with barrelshaped spindles after freezing and subsequent low FRs suggesting that less obvious defects of spindle/chromatin, which may have not been detected by the staining technique, and oocytes fertilization are also expected to be influenced negatively.From an intracellular structural level, observations of different CSK elements simultaneously after cryopreservation should extend to larger numbers and more refined methods. Katayama et al have put forward an interesting theory of MC in oocytes, backing up their argument thorough data which would be helpful to explain the probable cause of the low fertility potential of oocytes after freezing (37).MCs are essential cell organelles, as well as ATP synthesis, involved in intracellular Ca 2+ homeostasis, a key ion for normal fertilization to happen.Perinuclear aggregation of MC rapidly facilitates the energy supply at fertilization and early embryonic stages may be positively correlated with the developmental ability of embryos.MC distribution pattern is altered in IVF oocytes.The extreme low temperature using in cryopreservation, is also another non-physiologic condition.Also, there isno physiologic protection against it in cells (38), which may be one of the reasons for low fertilization potential subsequent to the worsening alteration of the MC distribution pattern (39) or the MC structure disordering (28) and the Ca 2+ homeostasis disturbing (40) in oocytes vitrified by rather low protective level, as conducted in Exp. 3 oocytes. To focus on contributing causes of the low fertility rates of Exp. 3 oocytes, we should call attention to Ca 2+ and the main component in Ca 2+ releasing system, ER.Because Ca 2+ release is essential for several aspects of successful fertilization, it is critical for vitrified oocytes to preserve the ability to release Ca 2+ (40).Even if MTs are disrupted and a spindle is able to reform within a few hr after oocyte warming, it is possible that the association between the ER and MTs does not re-establish within this time (40). Using low CPA-concentrations, the toxicity on CSK will be reduced.Although this has been widely acknowledged, information gained by the fore mentioned studies provide insight into the other intracellular structures and functions that may be compromised by vitrifying using such low concentrations.Their susceptibility to temperature fluctuation more than that of spindle apparatus seems to be a subject of our near future examination. In consistent with the previous observations by many investigators of oocytes cryopreservation methods (13), the immunofluorescent technique that we used shows degraded displaced spindles, several chromatin masses in spite of unique compact chromatin arrangement.Similar to some other studies (25), we observed) complete depolymerized spindle while chromosomes tend to remain together.This suggests that kinetochores may remain associated with MTOCs (26).Gomes et al have mentioned the positive correlation between the incidence of spindle repolymerization after vitrification/ warming and the chromatin cohesiveness during cryopreservation (41).Larger majority of the oocytes in Exp. 1 and 2 in comparison with the other Exp.has displayed cohesive chromatin after 1 h-incubation period, which may validate this eventuality when incubation period longer than 1 h is appropriate to restore spindle organization; this issue is yet to be understood accurately (29).Observing similar images taken from the rest of Exp.showed that low success rates would encourage further examinations using the protocols with longer incubation period post-warming. Further developments and evidences based on SRs, FRs as well as ICC results of the Exp. 4 and 5, were totally disheartening and oocytes return to physiologic functional status was not confirmed using the protocol.Although the precise nature of the damage caused by cryopreservation remains to be exactly determined (9), findings of several studies suggest that the major obstacles in success-ful oocytes cryopreserving are the characteristics of the oolema (42), the presence of cortical granules, spindle system at the metaphase of meiosis II (13) and zona pellucida hardening (43).In addition, the oocytes must be fertilized by sperm at the appropriate time (18). Intracellular ice formation can be affected by the presence of the CPAs in the freezing solutions, and by the freezing and thawing rate (44).To consider important issues like temperature reduction rate, it should be noted that at cooling rates slower than the optimum rate, cell death is due to long periods of exposure to hypertonic conditions.At cooling rates faster than the optimum, cell death is associated with intracellular ice formation, which is inevitably lethal (45).The actual value of the optimum rate is determined by a number of biophysical factors: 1. Cell volume and surface area 2. Permeability to water 3. Arrhenius activation energy (temperature-dependent energy required for the rate of chemical reactions) 4. Type and concentration of CPA additives (45).The last item establishes a connection between the concentration of the CPAs used and the cooling/warming rate.Being able to successfully adopt the CPA-concentrations used for Exp. 4 and 5, a new cryodevice which can increase further cooling/warming rate than cryotop must be invented.Achieving optimum rate for oocytes (because of the small surface-to-volume ratio) and embryo (because of the large surface area and low water permeability) is the subject of numerous researches (46). It is good to mention draw attention to equilibration issue which is often critical in the case of oocytes cryopreserving.Since the oocyte is a large cell containing a large quantity of water, it requires time to reach adequate dehydration (osmotically balanced by the CPA solution) before lowering the temperature and thus it is more difficult to avoid ice crystal formation.The pretreatment (or equilibration) time before cooling might affect the viability and developmental ability of oocytes (18).Comparing the SRs of oocytes pretreated to 0.75 M (Exp.4) with those exposed to 0.375 M CPAs (Exp.5) lays emphasis on the vital effect of pretreatment in terms of CPA-concentration, exposure temperature and duration. Any intervention causing even temporary change in the equilibrium of the physiological state could potentially be toxic to cells, including ICSI, which was introduced as a unique technique to overcome cryo-induced zona hardening and sequential IVF failure (5).Due to mechanical stresses and small amounts of PVP into oocytes during procedure (9), the movement of the polar body in the perivitelline space or migration of the spindle deeper into the oocyte could be misleading as for the relation between the spindle and polar body (47), so that ICSI technique has fallen into harmful category.These can be the reason why none of the cryosurvived oocytes in Exp. 4 were fertilized.In addition to the immediate causes of the cryodamage explained above, we shall make reference to the work of Tucker and Libermann (16).who have provided evidence for the fact that although the cell nucleus has the ability to reassemble morphologically following cryopreservation, the future development of the embryo could be suboptimal.Spindle and chromatin re-organization may not be the whole point but it is a vital component in any dividing cells nucleus.The degree of CSK repolymerization was critically dependant on the method used to dilute or remove the CPAs (48).Post-warming incubation period brings about reassembled spindles in vitrified oocytes.Furthermore, controversy continues over the striking resemblance of the reassembled spindle with the original one (49).In order to guarantee the used protocol, one needs to find the answer of the raised question how to carry out warming procedure to make useful contribution in preventing original spindle depolymerization and grasping an opportunity to have the original spindle throughout vitrification/ warming procedure. Except for the results published by Schroeder et al who were able to successfully cryopreserve mouse oocytes using a slow-cooling protocol and 1.0 M DMSO (50), this appears to be the first time that 1.0 M and 0.75 M concentrated CPAs + 0.5 M sucrose have been evaluated for cryotop-vitrified mouse M-II oocytes (Exp.3 and 4).Experience led us to the expectation that in the course of cryopreservation with 1.0 M concentrated CPAs for oocytes freezing, the oocytes viability will be reduced, although not necessarily to an extent that makes them incapable of becoming fertilized or developed further.It is suggested that the current reduced concentrated solutions i) could be combined with other additives (such as CSK stabilizers (51), ice blocking polymers (52) and high concentration of the sugars (53) into, or ii) to deplete some of the supplements from freezing solutions [such as sodium (54)(55) and Ca 2+ ions (56)], in such circumstances with iii) different pretreatment temperature, duration and the concentration of the CPAs (45).And finally, we draw your attention to a large body of experimental evidences that indicate major positive impacts of cumulus cells on vitality of oocytes (51,(57)(58)(59).It would be effective to study the efficiency of the new protocol on cryopreserving cumulus oocyte complex; furthermore, because of the mechanical stress of PZD/ICSI procedures, monitoring oocyte parthenogenetic activity (18) or adopting conventional IVF following PZD are safer alternatives to scoring oocyte functioning after warming.In turn, it may lead to some improvements in cryopreservation unfertilized oocytes procedures. Conclusion From the results of this study it can be concluded that vitrification by cryotop technology using minimal volume approach increases both cooling and warming rates.Therefore, the CPAs (EG+DMSO) limited reduction to 1.25 and 1.0 M instead of 1.5 M for oocytes and embryos cryotop-vitrification procedure, which may be a slight adjustment. M -II oocytes and four cell stage-embryos were allotted randomly to one of the following control and five experimental groups (Exp.):The oocytes of Exp. 1 to 5 were subjected to PZD/ICSI procedures and in vitro culture after vitrification/warming, while 0.75 M ES and 1.5 M VS, 0.625 M ES and 1.25 M VS, 0.5 M ES and 1.0 M VS, 0.75 M ES and 0.75 M VS and 0.375 M ES and 0.75 M VS, were used in Figure 1 . Figure 1.The fluorescent micrograph of control oocyte shows the examples of normal spindle and chromatin configurations.A fresh oocyte stained immunocytochemistically with anti α-tubulin monoclonal antibody and fluorescein isothiocyanate to visualize the spindle (green), along with counterstained with Diamidino-2-phenylindole to visualize the chromosomes (blue).After freezing at the metaphase-II stage and warming, the subcortically localized normal spindle configuration appears barrelshaped with chromosomes equatorially arranged on a regular plate Figure 2 . Figure 2. The fluorescent micrographs of the treated oocytes show the examples of spindle and chromatin configurations abnormalities.Spindles were stained in green and chromosomes in blue.Arrow in panel A indicates chromosomes displaced out of the abnormally localized spindle.In panels B and C, in addition to abnormal localization, the spindles appear not to have distinct poles.In panel B, in spite of one compaction, various compact masses of chromatin exist.In panel D, in contrast to complete depolymerization of spindle microtubules, it seems chromosomes did not become dispersed.Rather, they remained close together in disorganized bundles Figure 3 . Figure 3.The fluorescent micrograph of control embryo shows the examples of normal genetic material array.A fresh embryo stained immunohistochemically with anti α-tubulin monoclonal antibody and Fluorescein Isothiocyanate to visualize the microtubules (green) and counterstained with Diamidino-2-phenylindole to visualize the chromosomes (blue).After freezing at the four cell stage-embryo and warming, four compact masses of chromatin, each in one blastomere are considered normal. Figure 4 . Figure 4.The fluorescent micrographs of treated embryos show the examples of chromatin configurations abnormalities.Microtubules were stained in green and chromatin in blue.Arrow in panel A indicates a separate part of chromatin displaced away from the rest of mass in one blastomere.The embryo in panel B seems to have unusually large and the small masses of chromatin in different blastomeres.The embryos in panel C and D show the several compact masses of chromatin, atypical in size and distribution. Table 1 . Metaphase II-oocytes survival, fertilization and developmental rates after using different concentrations of the cryoprotectants Data presented in parentheses as mean ± standard division * P<0.001 Table 2 . Four cell stage-embryos developmental rates after using different concentrations of the cryoprotectants Table3shows the details of our observations at different times after warming, and after using different freezing solutions in order to vitrify mouse mature oocytes. Data presented in parentheses as mean ± standard division * P<0.001 configuration in Exp. 3 was significantly lower than the control group, Exp. 1 and 2 (P<0.001) at both of the different times.While there were no oocytes with normal spindle apparatus immediately post-warming in Exp. 4, the proportion of oocytes with normal spindle status after 1 h-incubation period was larger, although they were significantly lower than the control group, Exp. 1, 2 and 3.The oocytes in Exp. 5 were detected with normal spindle apparatus neither immediately after warming, nor following 1 hr incubation period. Table 3 . Metaphase II-oocytes normal spindle organization rate immediately after warming and following one hr-incubation period after using different Table 4 . Four cell stage-embryos normal fluorescence index immediately after warming and following one hr-incubation period after using different concentrations of the cryoprotectants a Each figure refers to the number of embryos that were assigned a subjective rating for fluorescence intensity of the MTs: 0= little or no fluorescence; 1= weak; 2= moderate; 3= strong fluorescence. Table 5 . Four cell stage-embryos normal chromatin configuration rate immediately after warming and following one hr-incubation period after using	2018-04-03T01:33:33.549Z	2013-04-01T00:00:00.000	{ "year": 2013, "sha1": "82f24fbf6175da7d22582b7a77594fab9aef654e", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "82f24fbf6175da7d22582b7a77594fab9aef654e", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
15720439	pes2o/s2orc	v3-fos-license	Overexpression of STAMP2 suppresses atherosclerosis and stabilizes plaques in diabetic mice Our research aims to evaluate the function of the STAMP2 gene, an important trigger in insulin resistance (IR), and explore its role in macrophage apoptosis in diabetic atherosclerotic vulnerable plaques. The characteristics of diabetic mice were measured by serial metabolite and pathology tests. The level of STAMP2 was measured by RT-PCR and Western blot. The plaque area, lipid and collagen content of brachiocephalic artery plaques were measured by histopathological analyses, and the macrophage apoptosis was measured by TUNEL. Correlation of STAMP2/Akt signaling pathway and macrophage apoptosis was validated by Ad-STAMP2 transfection and STAMP2 siRNA inhibition. The diabetic mice showed typical features of IR, hyperglycaemia. Overexpression of STAMP2 ameliorated IR and decreased serum glucose level. In brachiocephalic lesions, lipid content, macrophage quantity and the vulnerability index were significantly decreased by overexpression of STAMP2. Moreover, the numbers of apoptotic cells and macrophages in lesions were both significantly decreased. In vitro, both mRNA and protein expressions of STAMP2 were increased under high glucose treatment. P-Akt was highly expressed and caspase-3 was decreased after overexpression of STAMP2. However, expression of p-Akt protein was decreased and caspase-3 was increased when STAMP2 was inhibited by siRNA. STAMP2 overexpression could exert a protective effect on diabetic atherosclerosis by reducing IR and diminishing macrophage apoptosis. Introduction The type 2 diabetes mellitus forms a growing public health problem worldwide. In Western countries, about 60% patients with coronary heart disease also have diabetes mellitus, and the number has reached to 80% in China. Even with many medical therapies, the mortality is still as high as 70%. Therefore, atherosclerotic cardiovascular disease has become the most serious complication of diabetes [1,2]. Patients with diabetes are at a high risk for developing acute coronary syndrome (ACS) [3]. Diabetic patients with ACS suffer from increased mortality due to rupture of vulnerable atherosclerotic plaques compared with their non-diabetic peers [4,5]. The patients with diabetes have more serious vulnerable plaques and higher incidence. However, the mechanism remains to be elucidated. Diabetes mellitus and atherosclerotic cardiovascular disease increase the medical cost directly, also increase the economic burden of society indirectly. With the increasing health consciousness and development of preventive medicine in recent years, the mortality of coronary heart disease has declined. However, the mortality of diabetic coronary heart disease increased year by year. Therefore, it is necessary to explore the mechanism for ACS treatment and take advantage of medical resources reasonably. Insulin resistance (IR) plays a causal role in the pathogenesis and development of type 2 diabetes [6,7] and atherosclerotic cardiovascular disease [8]. Studies have shown that IR is involved in impairment of glucolipid metabolism [9], thus increasing atherosclerotic plaque formation and decreasing plaque stabilization [10]. In advanced lesions, macrophage apoptosis promotes the development of the necrotic core, a key factor in rendering plaques vulnerable to disruption [11][12][13]. Analysis of culprit lesions from diabetic patients of sudden coronary death has shown that the majority of cells at the rupture site were macrophages and that the apoptosis in macrophages was more frequent at the rupture site as opposed to areas of intact fibrous cap [14]. Akt signalling pathway is a key step in signal transductions of apoptosis in macrophages [15]. Akt inhibition leads to macrophage apoptosis, abnormal glucose tolerance and IR [15][16][17]. Therefore, insulin-resistance macrophages are more susceptible to apoptosis in atherosclerotic lesions, probably predisposing to plaque rupture [9,18,19]. Carlos suggests that special treatment of activated Akt1 could increase stability and reduce formation of atherosclerotic plaques in vivo. On the other hand, Akt1 reduction could reduce the LDL intake of macrophages, suggesting Akt1 may have a role in atherogenesis [15]. PI3K/Akt activation enhances macrophage survival in lesions, and the persistent activation of Akt promotes cellular hypertrophy and hyperplasia, thereby promoting atherogenesis [15]. Therefore, exploring a new biomarker to regulate Akt signalling pathway exactly through adjusted insulin pathway is necessary. Studies have shown that six-trans membrane protein STAMP2 [20,21] may be the key molecule. STAMP2 is involved in regulation of glucose metabolism. Its deficiency induces impairment of Akt signalling pathway and insulin exocytosis that lead to glucose metabolism disorder and IR [22]. Moreover, STAMP2 is expressed in human and mouse macrophages, and regulates foam cell formation. STAMP2 is expressed in both human and mouse atherosclerotic plaques, and its deficiency promotes atherosclerosis in mice [23]. Therefore, we propose that STAMP2/Akt signalling pathway may be the common intermediary among macrophage apoptosis, IR and vulnerable plaques in the diabetic state. Here, we investigated the role of STAMP2 in macrophage apoptosis of vulnerable atherosclerotic plaques and determined the molecular mechanism whereby STAMP2 modulates Akt signalling pathways in RAW264.7. Regulating STAMP2 appears a promising novel therapeutic approach in patients with ACS. Mice All animal procedures were in accordance with the institutional guidelines of Qilu Hospital of Shandong University and approved by Shandong University Institutional Animal Care and Use Committee. Induction of diabetes and atherosclerosis in mice Three-week-old male ApoE À/À /LDLR À/À mice were fed a high-fat diet (HFD; 20% fat, 20% sugar, 1.25% cholesterol; Beijing HFK Bio-Technol-ogy, China). At the age of 9 weeks, any mouse who exhibited IR was injected once with low-dose streptozotocin (STZ, Sigma-Aldrich, St. Louis, MO, USA; 75-80 mg/kg i.p. in 0.1 mol/l citrate buffer, pH 4.5). After 2 weeks, most mice displayed hyperglycaemia, IR and glucose intolerance, as previously reported [24]. At age 11 weeks, similar animals were randomly divided into two groups, one will be treated with the STAMP2-expressing adenoviruses, referred to as the diabetes model (DM) STAMP2 group (n = 10), and the other treated with the expressing vector control adenovirus, referred to as the DM Vehicle group (n = 10; see later for the details of the adenovirus used). The mice fed a normal diet were used as non-diabetic controls, divided into two groups. One will be treated with the STAMP2-expressing adenovirus, referred to as the Control STAMP2 group (n = 10), and the other treated with the expressing vector control adenovirus, referred to as the Control Vehicle group (n = 10). Intraperitoneal glucose tolerance test (IPGTT) Glucose tolerance was assessed by IPGTT. Mice fasted overnight were challenged intraperitoneally with glucose at 1.5 g/kg bodyweight. Blood samples were collected sequentially from the tail vein at 0, 15, 30, 60 and 120 min. and tested for glucose. Plasma glucose of every animal was measured with the OneTouch SureStep glucometer (LifeScan Inc., Milpitas, CA, USA). The mean area under the receiver operating characteristic curve (AUC) was calculated for glucose. Construction of STAMP2-expressing adenovirus The recombinant pAdxsi adenovirus constitutively expressing STAMP2 was constructed using the pAdxsi Adenoviral System (SinoGenoMax, Beijing, China). The STAMP2 cDNAs from mouse were inserted into pShuttle-CMV-EGFP vector. The pAdxsi vector adenovirus was used as the control vehicle virus. The STAMP2-expressing adenovirus of 5 9 10 9 plaque-forming units (PFU) were injected to the mice (the DM STAMP2 group and the Control STAMP2 group) by jugular vein injection at 20 weeks and another 5 9 10 9 PFU of virus by jugular vein injection at 22 weeks. The control vehicle virus of the same PFU was also injected to the mice (the DM Vehicle group and the Control Vehicle group) by jugular vein injection. All mice were killed for further biochemical and histological study at 24 weeks. were embedded in optimal cutting temperature compound (OTC; Sakura Finetek, Beijing, China). Sections were cut every 5 lm along the BCA and stained with haematoxylin and eosin, Picrosirius red, Masson' trichrome staining and Oil Red O. A Nikon microscope (Nikon, Melville, NY, USA) was used to capture the images. The mean was obtained by quantitative morphometry with automated image analysis (Image-Pro Plus, Version 5.0; Media Cybernatics, Houston, TX, USA). For the detection of oligonucleosomal DNA cleavage, a stringent TUNEL (terminal deoxynucleotidyl transferase end labelling) technique was used (Roche, Mannheim, Germany). In brief, sections were immersed in PBS for 15 min. at room temperature, incubated in permeabilization solution for 2 min. on ice, then transferred into the TUNEL reaction mixture, and incubated for 60 min. at 37°C in a humidified atmosphere in the dark. Adhesion assay The Costar culture plates were precoated with poly-lysine (50 lg/ml, Sigma-Aldrich). Added 5 9 10 6 macrophages into plates, after 10 min. incubation at 37°C, the adherent cells were stained by DAPI and counted under a fluorescence microscope. Migration assay Cell migration assays were performed with 24-well Costar Transwell plates (5 lm). After overnight starvation in DMEM with 2% foetal bovine serum, 7.5 9 10 4 cells were loaded into the upper chambers. The lower chambers were filled with solution of monocyte chemoattractant protein-1 (100 ng/ml; PeproTech Inc, Rochy Hill, NJ, USA) in DMEM. After 6 hrs incubation at 37°C, macrophages were stained with crystal violet (Sigma-Aldrich). The macrophages on the upper side of the membrane were removed by a cotton swab. Then we counted the average number of macrophages on the lower side. Phagocytosis assay Cells (1 9 10 6 cells/well) were seeded to glass bottom microwell dishes (MatTek Corporation, Ashland, MA, USA). After macrophages were incubated 10 min. at 37°C in dark with Alexa Fluor 488 Ac-LDL (10 lg/ml; Molecular Probes, Invitrogen, Eugene, OR, USA), excessive Ac-LDL was washed away and cells stained by DAPI in 5 min. Then the cells were observed under a laser scanning confocal microscopy (Leica TCS SP2; Leica, Wetzlar, Germany) and analysed by a flow cytometry (FACSCaliber; BD Biosciences Franklin Lakes, NJ, USA). SiRNA transfection Transfection was performed with Lipofectamine 2000 reagent (0.5 ll, Invitrogen). For reporter assays, 1.5 9 10 5 cells were seeded in 12-well plates at least 12 hrs before transfection. Cells were transfected with 2 lg of siR-NA-STAMP2 or 2 lg of negative control (NC) siRNA. Transfection was performed using the following primers: STAMP2, forward 5′-GCA GCA UCC AAG UCU GAC ATT-3′ and reverse 5′-UGU CAG ACU UGG AUG CUG CTT3-3′; NC, forward 5′-UUC UCC GAA CGU GUC ACG UTT-3′ and reverse 5′-ACG UGA CAC GUU CGG AGA ATT3-3′. After incubation for 6 hrs, culture medium should be changed. Then the cells were observed under a laser scanning confocal microscopy (Leica TCS SP2; Leica) and the transfection efficiency was calculated. Twenty-four hours after transfection, cells were re-treated for 16 hrs with HG and harvested for further study. Transfection of STAMP2 overexpressing adenovirus The cells (1 9 10 6 cells/well) were administered virus in 200 MOI. The culture medium was changed after 12 hrs. Then the cells were observed under a laser scanning confocal microscopy. Twenty-four hours after transfection, cells were re-treated for 16 hrs with HG and harvested for further study. cycles; Bio-Rad, Hercules, CA, USA), using SYBR green as fluorescence dye. Relative expression analysis involved the 2 ÀΔΔCT method. Statistical analysis Data are presented as mean AE SD (n is noted in the fig legends), and the statistical significance of differences was evaluated with an ANOVA. Significance was accepted at the level of P < 0.05. Results Diabetic atherosclerosis is induced in ApoE À/À /LDLR À/À mice. We generated a non-genetic rodent model closely resembling human dia-betic atherosclerosis disease by feeding male ApoE À/À /LDLR À/À mice with high fat diet and by the STZ treatment (referred to as DM mice hereafter), and normal diet (chow diet) fed mice serve as controls, as previously reported [24] (Materials and methods). Multiple metabolic parameters of the mice were examined to confirm the model. Insulin resistance in the DM mice was confirmed by IPGTT. The levels of blood glucose in the DM group were similar to that of the control group at the age of 3 weeks (P > 0.05; Fig. 1A). However, at the age of 9 and 11 weeks, the levels of blood glucose in the DM group were significantly higher than the control mice at all of the time points tested (P < 0.05; Fig. 1A). Similarly, at the age of 9 and 11 weeks, the AUC for glucose level of the DM mice was higher compared with the control mice compared to that at baseline of 3 weeks (P < 0.05; respectively; Fig. 1B). Intraperitoneal glucose tolerance test results in the control mice had no significant changes between week-3 and week-11. As expected, the bodyweight was significantly higher in the DM group than in the control group at and after week-9 except at week-11 (P < 0.05 for all pair-wise comparisons; Fig. 1C). We also confirmed atherosclerosis as previously reported [27] in the DM mice at the age of 24 weeks after high fat diet and STZ treatment ( Fig. 1D). In summary, the diabetic model induced by a HF diet and STZ showed typical features of IR, hyperglycaemia, obesity and aorta lipid accumulation, resembling the state of human diabetic atherosclerosis disease. Endogenous STAMP2 expression is suppressed in atherosclerotic aorta To study the local functional role of STAMP2 in atherogenesis, we investigated the mRNA and protein expression levels of STAMP2 in the aorta of the diabetic mice and the non-diabetic mice. The mRNA level of STAMP2 in aorta was significantly decreased in the DM Vehicle group compared with that in the Control Vehicle group, as assessed by the RT-PCR (0.27 AE 0.08 versus 1.00 AE 0.00; P < 0.01; Fig. 2A). Consistent with the mRNA expression levels, the protein level of STAMP2 was also significantly decreased by 28%, in the DM Vehicle group compared with in the Control (non-diabetic) vehicle group, as assessed by Western blot analysis (0.55 AE 0.02 versus 0.76 AE 0.04; P < 0.01; Fig. 2B and C). These data demonstrate the correlation between the atherosclerotic aorta and low levels of STAMP2 expression. Overexpression of STAMP2 in aorta Although we showed a lower STAMP2 expression in atherosclerotic aorta as compared with normal aorta, it is not known if it contributes to atherogenesis or merely one of the consequences of atherogenesis. Thus, we asked if overexpression of STAMP2 in aorta would be sufficient to reduce atherogenesis. As expected, the mRNA expression of STAMP2 in aorta was upregulated in the DM STAMP2 mice group by 122% compared to the DM Vehicle mice group (0.60 AE 0.12 versus 0.27 AE 0.08; P < 0.05; Fig. 2A), and the STAMP2 mRNA level was increased in the Control STAMP2 mice group compared with the Control Vehicle mice group (1.85 AE 0.60 versus 1.00 AE 0.00; P = 0.22; Fig. 2A). Consistent with the STAMP2 mRNA level changes, the STAMP2 protein level was also increased in the DM STAMP2 mice group by 27% compared with the DM Vehicle mice group (0.70 AE 0.04 versus 0.55 AE 0.02; P < 0.001; Fig. 2B and C), and the STAMP2 protein level was increased in the Control STAMP2 mice group by 18% compared to the Control Vehicle mice group (0.90 AE 0.03 versus 0.76 AE 0.04; P < 0.001; Fig. 2B and C). The magnitude of protein level changes was smaller compared with that of the mRNA level changes. Overexpression of STAMP2 improves metabolism in ApoE À/À /LDLR À/À mice STAMP2 was shown previously to play an important role in regulating metabolism, IR and atherogenesis. We investigated the global effect of STAMP2 by systemic overexpression of STAMP2 in our mouse model of diabetes and atherosclerosis. STAMP2-expressing adenovirus was used to treat diabetic and non-diabetic ApoE À/À /LDLR À/À mice by jugular vein injection and the expression vehicle adenovirus was used as control. As expected from our previous results, several metabolic indices were substantially altered in the diabetic mice. Comparing the DM Vehicle group to the Control Vehicle group, the bodyweight, the level of fasting serum glucose and cholesterol were significantly increased (Table 1). After overexpression of STAMP2, the indices decreased numerically in the DM STAMP2 group versus DM Vehicle group, although for most part not reaching statistically significance ( Table 1). Calculation of the HOMA index as a measure of IR in the (Table 1). In contrast, HOMA decreased in the DM STAMP2 group versus the DM Vehicle group (Table 1). Taken together, these data suggest that systemic STAMP2 overexpression does not have obvious deleterious effect, and can modestly reverse the metabolic disease state of the diabetic ApoE À/À / LDLR À/À mice. Overexpression of STAMP2 stabilizes lesions in the brachiocephalic artery Spontaneous plaque rupture has been demonstrated in the BCA of mice in recent research [28]. We asked whether the overexpression of STAMP2 is able to affect the stability of atherosclerotic plaque. The area, lipid content and macrophage quantity in BCA were known previously to be associated with the plaque stability. We found that each of the three parameters in DM Vehicle group was significantly increased compared with those in the Control Vehicle group (Fig. 3A1-A3; Table 2). STAMP2 overexpression lead to significant less area, lipid content and macrophage quantity in the DM STAMP2 group compared with those in the DM Vehicle group (Fig. 3A1-A3; Table 2). Similarly, STAMP2 overexpression also reduced each of the three parameters in the non-diabetic mice, the Control STAMP2 group, compared with the Control Vehicle group (Fig. 3A1-A3; Table 2). Brachiocephalic arteries collagen content was known to affect the stability of atherosclerotic plaques. We used Masson's trichrome staining and Picrosirius red staining to measure BCA collagen content. Comparing to the BCA plaques of the Control Vehicle group, the BCA plaques of the DM Vehicle group had lower collagen (51% reduction), collagen I-III ratio (14% reduction, P = 0.15) and smooth muscle cell (SMC) content (30% reduction; Fig. 3A4-A6; Table 2), indicating a highly instable state. STAMP2 overexpression appears to be able to correct the situation. STAMP2 overexpression increased the collagen content by 29%, increased the collagen I-III ratio by 34% and SMC content by 16% (P = 0.21) in the BCA plaques, when comparing the DM STAMP2 group with the DM Vehicle group (Fig. 3A4-A6; Table 2). STAMP2 overexpression also increased the collagen content by 7% (P = 0.16), the collagen I-III ratio by 50% and SMC content by 41% in the BCA plaques of the non-diabetic mice, the Control STAMP2 group, when compared with the Control Vehicle group (Fig. 3A4-A6; Table 2). The vulnerability index (VI) was used to describe atherosclerotic plaque stability in previous studies [29]. In our study, VI was significantly higher, by about threefold, in the DM Vehicle group than in the Control Vehicle group (Table 2). STAMP2 overexpression decreased VI in the DM STAMP2 group by 41% compared with the DM Vehicle group (Table 2). Taken together, these results showed that the BCA plaques are destabilized in the diabetic mice and that STAMP2 overexpression is sufficient to increase the stability of the BCA plaques, thereby likely preventing plaque rupture. Overexpression of STAMP2 reduces macrophages apoptosis in the BCA Cell apoptosis, especially macrophage apoptosis, has significant roles in the atherosclerosis process. We investigated whether the STAMP2 overexpression exhibited an anti-apoptotic effect. Thin sections of mice BCA were stained by TUNEL which labels apoptotic cells. We found that cell apoptosis was significantly higher, by 86%, in the DM Vehicle group than the Control Vehicle group (174.16 AE 13.55 versus 93.71 AE 0.78 9 10 3 lm 2 ; P < 0.001; Fig. 3B). STAMP2 overexpression significantly reduced TUNEL staining in the BCA of the DM STAMP2 group compared with the DM Vehicle group (98.88 AE 12.52 versus 174.16 AE 13.55 9 10 3 lm 2 ; P < 0.001; Fig. 3B). The Control STAMP2 group had less cell apoptosis compared with the Control Vehicle group (82.78 AE 1.01 versus 93.71 AE 0.78 9 10 3 lm 2 ; P < 0.001; Fig. 3B). We further analysed the macrophage apoptosis in the BCA plaques by double staining the BCA plaques with TUNEL and an antibody against MOMA-2, a macrophage marker. There was a significantly higher number of TUNEL + and MOMA-2 + macrophages in the DM Vehicle group than the Control Vehicle group (136.50 AE 13.26 versus 52.63 AE 4.83 9 10 3 lm 2 ; P < 0.001; Fig. 3C). Overexpression STAMP2 significantly reduced macrophage apoptosis in the BCA plaques of the DM STAMP2 group compared with the DM Vehicle group (65.42 AE 6.26 versus 136.50 AE 13.26 9 10 3 lm 2 ; P < 0.001; Fig. 3C) and in the Control STAMP2 group compared with the Control vehicle (35.24 AE 7.45 versus 52.63 AE 4.83 9 10 3 lm 2 ; P < 0.01; Fig. 3C). These data showed that macrophage apoptosis is substantially increased in the atherosclerotic plaques of diabetic mice. The result suggested that one of the cellular mechanisms by which STAMP2 overexpression increases the stability of atherosclerotic plaques is to decrease the macrophage apoptosis. Overexpression of STAMP2 activates the Akt signalling pathway in aorta plaques of ApoE À/À / LDLR À/À mice It was known previously that STAMP2 regulates Akt signalling pathway [22] and Akt pathway controls cell survival by regulating the apoptotic machinery such BCL family of anti-apoptotic factors. Thus, we investigated whether one of the molecular mechanisms by which STAMP2 affects atherogenesis and cell apoptosis in plaques is through Akt pathway regulation. We have shown that the STAMP2 mRNA and protein levels in aorta were significantly decreased in the DM Vehicle group compared with the Control Vehicle group ( Fig. 2A-C). This appears to cause the down-regulation of Akt activity in aorta. Phosphorylation of Akt, the measure of Akt activation, was decreased in the DM Vehicle group by 28% compared with the Control Vehicle group (0.26 AE 0.01 versus 0.36 AE 0.00; P < 0.001; Fig. 4A and B). Since Akt phosphorylation activates the BCL family of anti-apoptotic factors, as expected, Bcl-2 was down-regulated by 32% in the DM Vehicle group compared with the Control Vehicle group (0.38 AE 0.01 versus 0.56 AE 0.03; P < 0.01; Fig. 4A and C). Consistent with the Bcl-2 level changes, caspase-3 protein level was up-regulated by 59% in the DM Vehicle group compared with the Control Vehicle group (0.89 AE 0.00 versus 0.56 AE 0.02; P < 0.001; Fig. 4A and D), which may explain the 86% increase in BCA cell apoptosis observed (Fig. 3B; see above section). STAMP2 overexpression significantly increased the aorta p-Akt level by 27% in the DM STAMP2 group compared with the DM Vehicle group (0.33 AE 0.02 versus 0.26 AE 0.01, P < 0.001; Fig. 4A and B). As expected, comparing the DM STAMP2 group to the DM Vehicle group, the Bcl-2 protein level was increased by 26% (0.48 AE 0.01 versus 0.38 AE 0.01, P < 0.001; Fig. 4A and C), but caspase-3 protein level was decreased by 13% (0.77 AE 0.02 versus 0.89 AE 0.00; P < 0.001; Fig. 4A and D). Similarly, p-Akt level was up-regulated by 8% in the Control STAMP2 group compared with the Control Vehicle group (0.39 AE 0.02 versus 0.36 AE 0.00; P < 0.05; Fig. 4A and B). Bcl-2 protein level was increased (0.75 AE 0.03 versus 0.56 AE 0.03; P < 0.001; Fig. 4A and C) and caspase-3 protein level was decreased (0.42 AE 0.01 versus 0.56 AE 0.02; P < 0.001; Fig. 4A and D) in the Control STAMP group compared with the Control Vehicle group. These results suggested that the STAMP2 reduces the pathogenesis of atherosclerosis by activating the Akt signalling pathway and preventing cell apoptosis, including macrophage apoptosis, in the plaques of diabetic mice. Macrophage cell line model in hyperglycaemic condition To further investigate the molecular mechanisms underlining STAMP2 function in diabetic mice, we established a macrophage cell line model in hyperglycaemic condition in vitro which recapitulates some of the characteristics of macrophages in the diabetic mouse model. RAW 264.7 is a mouse monocyte/macrophage cell line. The cells were cultured under a LG (5.5 mmol/l), HG (25 mmol/l glucose) and the control hypotonic (5.5 mmol/l glucose+19.5 mmol/l mannitol, HO) conditions. We found that cells from the HG treated group had significantly higher abilities of adhesion, migration and phagocytosis than the LG and the HO treated group (P < 0.05; Fig. 5A-C). High glucose treatment also increased RAW 264.7 cell apoptosis. Caspase-3 protein level showed a time-dependent increase with a maximum at 72 hrs in the HG treated cells compared with 0 hr (P < 0.01; Fig. 5D). Flow cytometry analysis based on Annexin-V and PI staining showed that the HG treatment induced higher level of apoptosis at different time points compared with the control cells (time 0 hr; Fig. 5E). High glucose treatment effect on cell apoptosis is mediated by the Akt signalling pathway. As western blot analysis showed, LY294002 inhibited Akt phosphorylation under HG treatment in 20 hrs (0.86 AE 0.01 versus 0.95 AE 0.00, P < 0.001), 24 hrs (0.84 AE 0.00 Fig. 5F), respectively. These data would suggest that HG could suppress Akt signalling pathway to induce macrophages apoptosis. We showed that in our diabetic mouse model, endogenous STAMP2 expression was reduced in aorta. However, it was not known if STAMP2 expression was directly affected in macrophage cells by hyperglycaemia and if downregulation of STAMP2 is sufficient to induce the phenotypic changes of macrophages seen in diabetic mice. To answer these questions, we studied STAMP2 function in the RAW264.7 cell line model under the hyperglycaemic condition. The relative mRNA expressions of STAMP2 were significantly increased by 16 hrs HG treatment compared with 0 hr HG treatment (9.90 AE 11.45 versus 1.00 AE 0.00, P < 0.01; Fig. 5G). Although LG treatment slightly up-regulated STAMP2 mRNA level with a maximum at 20 hrs (2.32 AE 0.98 versus 1.00 AE 0.00, P < 0.01; Fig. 5G), in comparison, the effect of HG treatment is much higher. As expected, there was no significant change STAMP2 expression by hypertonic (HO) treatment (P > 0.05). Consistently, the STAMP2 protein levels under HG treatment also increased at different time points compared with at 0 hr time point with maximum reached at 20 hrs treatment (P < 0.05; Fig. 5H). The LG treatment had minimal effect and HO treatment had no effect (P > 0.05; Fig. 5H). We investigated the role of STAMP2 in regulating the Akt signalling pathway to control cell apoptosis under HG treatment. Western blot analysis showed that the protein level of STAMP2 was 20% lower in the RNAi-STAMP2 group than the RNAi-Normal Control (RNAi-NC) group (0.78 AE 0.10 versus 0.97 AE 0.03; P < 0.05; Fig. 6A and B). After silence of STAMP2, the phosphorylated PI3K and phosphorylated Akt were reduced by 1% (0.81 AE 0.06 versus 0.82 AE 0.01; P = 0.84) and 11% (0.76 AE 0.06 versus 0.85 AE 0.03; P < 0.05; Fig. 6A, C and D), respectively, in RNAi-STAMP2 group compared with RNAi-NC group. However, the caspase-3 protein level resulted in a significant increase by 4-fold in RNAi-STAMP2 group compared with RNAi-NC group (0.90 AE 0.02 versus 0.65 AE 0.05; P < 0.001; Fig. 6A and E). The Bcl-2 protein level was reduced by 57% (0.30 AE 0.23 versus 0.70 AE 0.16; P < 0.05; Fig. 6A and F). Taken together, these data showed that STAMP2 silencing suppresses the PI3K/Akt signalling pathway and increases cell apoptosis. Overexpression of STAMP2 on cell apoptosis under HG treatment As seen in Figure 6A and G, western blot analysis showed that the protein level of STAMP2 was increased by 17% in the STAMP2expressing adenovirus (Ad-STAMP2) group compared to Normal Control-expressing adenovirus (Ad-NC) group (0.82 AE 0.06 versus 0.70 AE 0.04; P < 0.05). The phosphorylated PI3K was not changed in Ad-STAMP2 group compared with Ad-NC group (0.87 AE 0.03 versus 0.86 AE 0.03; P = 0.48; Fig. 6A and H). However, the phosphorylated Akt was significantly enhanced in Ad-STAMP2 group (0.76 AE 0.04 versus 0.69 AE 0.00; P < 0.05; Fig. 6A and I). Meanwhile, caspase-3 protein level resulted in a significantly decrease in Ad-STAMP2 group compared with Ad-NC group (0.67 AE 0.06 versus 0.76 AE 0.06; P < 0.05; Fig. 6A and J). The protein level of Bcl-2 was increased in the Ad-STAMP2 group (0.72 AE 0.02 versus 0.68 AE 0.02; P < 0.05; Fig. 6A and K). The results indicate that the activation of STAMP2/PI3K/Akt signalling pathway is responsible to reduce cell apoptosis. Overexpression of STAMP2 and macrophage functions Results have shown that cells from Ad-STAMP2 group showed significantly lower adhesion ability and higher phagocytosis ability compared with Ad-NC group (P < 0.01; Fig. 7A3, 4 and C3, 4). However, migration ability had no change in Ad-STAMP2 group compared with Ad-NC group (P = 0.09; Fig. 7B3, 4). The cells from Ad-STAMP2 group showed less apoptosis compared with Ad-NC group (P < 0.001; Fig. 7D). These results suggested that silence or overexpression of STAMP2 affect cells abilities of adhesion, migration, phagocytosis and apoptosis. Discussion The major finding of this study is that overexpression of STAMP2 could effectively improve metabolic indices, reduce IR and stabilize the atherosclerotic plaques in the diabetic ApoE À/À /LDLR À/À mice. Mechanistically, this phenotype is due to STAMP2 regulation of the Akt signalling pathway, thus diminishing macrophage apoptosis and stabilizing the vulnerable plaques. STAMP2 may be causally associated with IR, playing an important role in the regulation of glucolipid metabolism. For example, after STAMP2 is suppressed, glucose transport stimulated by insulin and Glut4 inversion to membrane mediated by insulin is significantly reduced [22]. In addition, STAMP2 À/À mice showed IR, glucose intolerance and lipid metabolism disorder [22]. We found a significant increase in bodyweight, blood glucose levels, total serum cholesterol and triglyceride levels in diabetic ApoE À/À /LDLR À/À mice. Overexpression of STAMP2 effectively improves metabolic indices and reduces IR in diabetic ApoE À/À /LDLR À/À mice, although there was no significant effect on serum insulin, FFA and TG level. Our data provide evidence that STAMP2 effectively reduces risk factors of atherosclerosis in the diabetic mouse model. The relationship between macrophage apoptosis and atherosclerosis is complex. Advanced macrophage apoptosis is an important contributing factor in vulnerable atherosclerotic lesions [13]. STAMP2 was expressed in both human and mouse atherosclerotic plaques [15]. In line with our observations in the brachiocephalic plaques of 745 ApoE À/À /LDLR À/À mice, as assessed by staining for TUNEL, macrophage apoptosis was significantly higher in diabetes group than in chow diet group. With the overexpression of STAMP2, the macrophage apoptosis in lesions of diabetic ApoE À/À /LDLR À/À mice was diminished. The lesions in brachiocephalic artery of diabetic ApoE À/À / LDLR À/À mice were demonstrated to be vulnerable. Some phenotypic characteristics of atherosclerotic plaques such as area of plaque, lipid content and macrophage quantity significantly increased. However, collagen content, SMC content in lesions were reduced. Usually increased content of lipid and macrophage or decreased content of collagen and SMC in patches have been widely used as indicators of plaque instability. We confirmed that overexpression of STAMP2 could lead to more contents of collagen, more SMC, less contents of lipid and less macrophage, thus increasing the stability of the plaques. Consequently, overexpression of STAMP2 could reduce the vulnerability of atherosclerotic plaques through diminishing macrophage apoptosis effectively. Akt signalling pathway involved in regulation of many cell functions is a key component in the regulation of macrophage apoptosis [30,31]. We showed in a cellular model that the expressions of STAMP2 mRNA and protein are increased initially by HG treatment but eventually decreased after extended treatment. We also showed in a diabetic mouse model where serum glucose level is persistently high, STAMP2 expression is also suppressed. In vitro, macrophage cell line RAW264.7 exhibited significantly increased adhesion, migration and phagocytosis abilities and apoptosis with HG treatment. Cell apoptosis induced by HG treatment is also induced by the PI3K inhibitor LY294002, indicating that the STAMP2/PI3K/Akt pathway play an important role in macrophage apoptosis under HG. We showed that in a macrophage cellular model, STAMP2 expression is necessary and sufficient to regulate the capabilities of migration, adhesion, phagocytosis and apoptosis by the loss-of-function experiment via STAMP2 siRNA and by the gain-of-function experiment via STAMP2 overexpression. Conclusions This study revealed that STAMP2 activity is important in increasing plaque stability in diabetic atherosclerotic mice, which is likely to function by activating the Akt signal pathway and suppressing apoptosis machinery to decrease the macrophage apoptosis in the atherosclerotic plaques. In addition to providing important information about the pathophysiology of atherosclerosis, the data from these studies might form the basis of novel therapeutic approaches to combat vulnerable plaque.	2017-04-04T16:42:21.131Z	2014-01-22T00:00:00.000	{ "year": 2014, "sha1": "0e1d0be88c4d47991bd2693d86e6ecdc8312fbae", "oa_license": "CCBY", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1111/jcmm.12222", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "0e1d0be88c4d47991bd2693d86e6ecdc8312fbae", "s2fieldsofstudy": [ "Biology", "Medicine" ], "extfieldsofstudy": [ "Biology", "Medicine" ] }
261996196	pes2o/s2orc	v3-fos-license	SUSTAINABLE DEVELOPMENT, AGENDA 2030 AND FOOD SECURITY IN HISTORICAL PERSPECTIVE : The sustainable development (SD) concept has won substantial popularity in recent decades. At the same time, neoliberalism (the socio-economic orthodoxy since the mid-1970s) is somewhat put in the shade. The paper attempts to find out whether the SD paradigm and its recent incarnation (Agenda 2030) constitutes the decisive break from the mainstream. The second aim is to assess whether the concept of food security is adequately addressed by the Agenda 2030. The study shows that Agenda 2030, with its 17 Sustainable Development Goals, is a much broader concept than preceding MDGs, but it still cannot guarantee the attainment of food security both in the short-term (the risk of commodity price bubbles) and in the longer-term (i.e. till 2030) due to the prevalence of extreme poverty Introduction Sustainable development (SD) is commonly referred to as meeting the needs of the present generation without compromising the ability of future generations to meet their own needs.SD goes along with heterodox economics in its holistic, pluralistic and interdisciplinary approach.The concept takes into account both time (i.e.present and future generations) and institutions which are often ignored by orthodoxy. Therefore, the concept of sustainable development has won great popularity in recent decades.It is esteemed not only by popular media and academic world, but also by many environmental activists.SD is often contrasted with the mainstream focus on economic growth measured by GDP dynamics (which represents a proxy of short term performance) and the level of GDP per capita (the proxy of the standard of living and/or even wealth of a country or a society).The SD concept (sometimes referred to as paradigm), however, often focuses on two pillars (i.e. economic and environmental).The third social pillar of the SD paradigm has been neglected in popular discourse.Mainstream economics aspired to be positive economics (i.e.value-free).That is the reason why the debates about income and wealth inequality or about social security and employment contracts were sporadic.The main goal was to safeguard economic growth via market mechanisms.These tendencies were further strengthened after the collapse of the communist Eastern Bloc.Hence, the so-called Washington Consensus reinforced free market imperialism.Capitalism (to be more precise, its neoliberal incarnation) was portrayed as the only viable socio-economic system.Hence, the popularity of the acronym TINA (there is no alternative).The free market was portrayed as a smoothly functioning fair mechanism.For mainstream economists, failures of this mechanism were minor (especially in comparison with government failures) and could practically emerge only in the environmental sphere in the form of negative externalities or insufficient provision of public goods.The remedy to these minor inefficiencies of the market system was to create yet another market (as in the case of the CO 2 emission trading scheme) and the reinforcement of the environmental pillar of SD. Yet, the edifice of mainstream economics was undermined by the global financial crisis that started in September 2008 with the collapse of Lehman Brothers -the fourth largest investment bank in the US.The contagion spread to Europe and resulted in the Great Recession (the biggest recession since the Great Depression in the 1930s).This financial and economic turmoil was a clear indicator that developed countries are not immune to serious, multifaceted crises (Szydło, 2013).The aim of the paper is to assess (mainly via the extensive review of literature, the UN documents and the analysis of FAO DOI: 10.34659/eis.2023.85.2.560 data) whether SD concept (especially Agenda 2030) represents a decisive break from current orthodoxy and whether it adequately addresses food security and poverty.To achieve this aim historical perspective is employed. Sustainable development The United Nations Conference on the Human Environment in Stockholm, June 5-16, 1972 (Stockholm Declaration) represented "a first taking stock of the global human impact on the environment, an attempt at forging a basic common outlook on how to address the challenge of preserving and enhancing the human environment" (Handl, 2012).Colander et al. (2004) optimistically claim that "economics is moving away from a strict adherence to the holy trinity -rationality, selfishness and equilibrium -to a more eclectic position of purposeful behaviour, enlightened self-interest and sustainability". The process of creation of the sustainable development concept coincided with the collapse of the Bretton Woods system on August 15th 1971, when the US unilaterally terminated the convertibility of USD to gold and the demise of the Smithsonian Agreement.It was soon followed by the ending of the "brief Keynesian experiment" in West Germany with the resignation of Schiller as minister of economics and finance in 1972 (Leaman, 2009). It is also worth remembering that the collapse of the Bretton Woods system briefly preceded the first oil crisis in the early 1970s.This paradigm shift is often referred to as the Hayekian counter-revolution, monetarist revolution or, as in the case of Harvey (2005), the Volcker shock. The creation of the SD theory can be treated as a progressive alternative to the Keynesian system, particularly in the environmental sphere or pillar.This springs from the fact that both SD and neoliberalism were being implemented in tandem.The rise to power of the new form of liberalism (i.e.neoliberalism) under Ronald Reagan, Paul Volcker (the chairman of the US Fed) and Margaret Thatcher in the early 1980s was happening practically at the same time when the SD concept was popularised by the UN conferences.In fact, Volcker replaced Miller at the Fed already in August 1979.Yet, monetarist money supply targeting, as suggested by Friedman, was first introduced in the German Bundesbank.This was preceded by "the revolt of the thirty-somethings", which initiated the monetarist anti-Keynesian revolution in German economics, especially from 1970from to 1976from (Janssen, 2006)).Also, Chile was an early laboratory for neoliberal ideas and policies.Augusto Pinochet had called upon a local group of Chicago-trained economists to propose a radically different economic program (Valdes, 1995).This group was actively supported by Hayek and Friedman.In April 1975, the former economist paid a controversial visit to Chile, governed by a brutal dictator (De Haan, 2016).Already by 1976, the year when Milton Friedman was awarded the Nobel Prize in economics, the "Chicago Boys" gained control of Chilean economic policy.Undoubtedly, neoliberalism is based on economic freedom (especially for the rich), while political freedom, democracy and human rights are rather forgotten.The dominant ideology has made great progress in depoliticising many spheres of human life. Hence, sustainable development might be viewed as a democratic, much broader and progressive approach towards human life on Earth.It became common knowledge that the term sustainable development was popularised by the so-called Brundtland report 'Our Common Future', published by the World Commission on Environment and Development (WCED) in 1987.The Brundtland report provides the classic definition of sustainable development: "development which meets the needs of the present without compromising the ability of future generations to meet their own needs".As noted by Tulloch and Neilson (2014), "the emphasis on 'needs' and 'development" in the same breath is significant as it positions economic development as the critical issue for meeting people's needs -both now and in the future -while ecological sustainability is only implicitly and indirectly identified and subtly cast as a problem of the future".According to Spash and Guisan (2021), "while needs to remain objective, how they are expressed, perceived, and fulfilled will always be subjective, conditioned by institutional arrangements and wider social and cultural contexts".Unquestionably, the definition of SD provided in Brundtland's report is vague, which makes it vulnerable to reinterpretations by the current orthodoxy.Indeed, according to Tulloch and Neilson (2014), the first step in the neoliberalisation of sustainable development was achieved in the Brundtland Report.The concept was also depoliticised, and "the power relations and historical specificity of the presently dominant capitalist mode of production are taken out of the account, and 'economic and social development' is ideologically neutralised" (Tulloch & Neilson, 2014). The earlier definition of SD, infrequently referred to even by the experts in the field, was formulated in the World Conservation Strategy already in 1980.This Strategy (IUCN, 1980) was prepared and advocated by the International Union for Conservation of Nature and Natural Resources (IUCN), UNEP, WWF, FAO and UNESCO.According to the World Conservation Strategy, "for development to be sustainable, it must take account of social and ecological factors, as well as economic ones; of the living and non-living resource base; and of the long term as well as the short term advantages and disadvantages of alternative actions" (IUCN, 1980).Paradoxically, this older definition by the IUCN (1980) provides a fuller and more progressive definition of SD than the younger formulated in the Brundtland report "Our Com-mon Future" in 1987.It needs to be emphasised that the classic Brundtland definition introduced a new term, 'needs'.At the same time, it got rid of references to social, ecological and economic factors and resigned from allusion to the living and non-living resource base.From this perspective, it could be argued that SD was dominated by new neoliberal orthodoxy already in the 1980s to become just a little more progressive offshoot of the all encompassing neoliberal paradigm. SD has been dominated by the new neoliberal system as it was not able to prevent a number of economic, social and environmental crises.The global financial crisis and consecutive Great Recession (a clear reference to the Great Depression of the 1930s) proved that neoliberalism, which led to financialisation and monopoly capitalism, was dangerous not only to people living in developing and least developed countries but also to the majority of those in developed countries.This proves that the SD concept (especially its economic pillar or dimension) was inept in sustaining development and standard of living after 2008. Both neoliberalism and SD gained popularity in the 1970s.Baldwin et al. (2019) claim that" the desire to "free up" the market to drive economic growth has been pursued in tandem with the aim of sustainable development" (United Nations, 2002;Fisher, 2006;Wagner, 2006;Bakker, 2010).But one cannot deny that neoliberalism won the battle of ideas and, for over four decades, has been the dominant paradigm which structures the functioning of individuals and societies in capitalist countries.Numerous financial and economic crises both in developing and developed countries did not undermine this dominance.Even the global financial crisis and subsequent Great Recession in 2009 did not lead to the paradigm shift.The sustainable development concept has been too weak to avert financial, economic, social and environmental crises. According to Newig et al. (2019), "current literature on sustainability governance and institutions is preoccupied with innovation, novelty, success and "best practice", but there is an emergent tendency to consider decline and failure as opportunities and leverage points to work towards and to achieve sustainability.Although failure, crisis and decay have been treated extensively, the link towards their productive potential has remained underdeveloped in the literature" (Newig et al., 2019). The concept of sustainable development is present in the popular debate, yet the discussions often concentrate on merely one aspect of environmental order, namely climate change resulting from greenhouse gas emissions.Other environmental problems are rarely addressed.The same applies to issues belonging to the economic and social pillars of SD.As the voice of the dissenters was intelligently reduced to activists focusing solely on averting climate change, one cannot expect a decisive break from current economic orthodoxy.Generally, for decades the actual implementation of the SD concept has been highly selective and mainly focused on the conservation of the environment and, at the same time, maintaining economic growth.This can be illustrated by the growing popularity of the terms "green growth", "greening the economy" and as in the case of (Köhn, 2012) "greening the financial sector".The social pillar of SD is not fashionable.The widespread popularity of micro analysis operating within the current economic architecture suits the interests of multinational corporations.Macroeconomic analysis is significantly reduced. GDP per capita and GDP dynamics were and still are the most important indicators in the economic pillar of SD.For example, Agenda 2030, in its first target of goal 8, aims to "sustain per capita economic growth in accordance with national circumstances and, in particular, at least 7 per cent gross domestic product growth per annum in the least developed countries".Other key measures of economic order assessing 21st-century capitalism are either rarely discussed or simply omitted (even in the case of supposedly broad and progressive SDGs).Such a shallow analysis cannot properly assess the following issues: asset and commodity price bubbles, financialisation, monopolisation, labour market, intra-and intergenerational income and wealth inequality, indebtedness, economic sectors, demography, spatial cohesion, leisure, well-being, etc. The UN's SDG indicators The UN's global indicator framework was adopted by the General Assembly on 6 July 2017 and is contained in the Resolution adopted by the General Assembly on Work of the Statistical Commission pertaining to the 2030 Agenda for Sustainable Development (United Nations, 2015).The annex to this resolutions entitled "Global indicator framework for the Sustainable Development Goals and targets of the 2030 Agenda for Sustainable Development" comprises 17 Sustainable Development Goals (SDGs) and a list of indicators to be refined annually and reviewed comprehensively by the Commission.At present, the official global indicator framework comprises 231unique SDG indicators.However, the total number of listed indicators is 247, as twelve indicators repeat under two or three different targets (United Nations, 2017).The global indicator framework was developed by the Inter-Agency and Expert Group on SDG Indicators (IAEG-SDGs) and later agreed upon by the Statistical Commission at its forty-eighth session, held from 7 to 10 March 2017, as a voluntary and country-led instrument. Classification into social, ecological and economic pillars or orders allows a clear assessment of the SD concept by the experts in particular fields.When the classification was erased from the classic definition of SD in Brundtland Report in 1987, officially referred to as "Our Common Future, Report of the World Commission on Environment and Development", the concept has become blurred.The same caveat applies to recent implementations of the concept in Millennium Development Goals (MDGs) and Sustainable Development Goals (SDGs).Some authors attempted to divide SDGs into four aspects (or spheres/pillars): economy (goals 8, 9, 10, and 12), society (goals 1, 3, 4, 5, 11, and 16), environment (goals 2, 6, 7, 13, 14, and 15), and governance (goal 17) (Lu et al., 2015).But the goals are not clearly grouped into pillars.This is one of the reasons why SD is still referred to as a 'contested concept' (i.e. that can be defined in more ways than one).It could well be argued that this post-modernist amalgamation has nothing to do with the heterodox noble postulates calling for broadening both the popular debate and academic analysis (as in the case of interdisciplinary studies and research)."Neoliberal articulation of sustainability with the broader field of contesting perspectives combined with a strategy of 'passive revolution', that are together summed up as the Rio process, has led earlier radical discourses being incorporated and subordinated to neoliberal hegemony" (Tulloch & Neilson, 2014).Weakening of sustainable development paradigm (especially its economic pillar) allows orthodoxy to replace it by the concept of 'green growth', 'greening the economy' and even 'greening the financial sector'.The focus on growth (measured by GDP) instead of development (especially sustainable development) is a central feature of neoliberalism.Hence, the edifice of the current orthodoxy remains intact.Naturally, sustainable development is not the only example of contested concepts.Söderbaum (2019) also adds 'democracy' and 'institution' to the list of terms whose definition is debatable and, therefore, vague."Neoclassical economists tend to limit attention to concepts that can be quantified and therefore avoid or reduce the role of contested concepts" (Söderbaum, 2019). At the same time, neoliberalism attempts to shape (and, if necessary, adjust) the definition of a "contested concept" so that it does not pose much threat to the orthodoxy.It could well be argued that this applies to the concept of "sustainable development" in economic and social pillars of Agenda 2030. " -Goal 1. End poverty in all its forms everywhere, -Goal 2. End hunger, achieve food security and improved nutrition and promote sustainable agriculture, -Goal 3. Ensure healthy lives and promote well-being for all at all ages, -Goal 4. Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all, -Goal 5. Achieve gender equality and empower all women and girls, -Goal 6. Ensure availability and sustainable management of water and sanitation for all, -Goal 7. Ensure access to affordable, reliable, sustainable and modern energy for all, -Goal 8. Promote sustained, inclusive and sustainable economic growth, full and productive employment and decent work for all, -Goal 9. Build resilient infrastructure, promote inclusive and sustainable industrialisation and foster innovation, -Goal 10.Reduce inequality within and among countries, -Goal 11.Make cities and human settlements inclusive, safe, resilient and sustainable, -Goal 12. Ensure sustainable consumption and production patterns, -Goal 13.Take urgent action to combat climate change and its impacts, -Goal 14.Conserve and sustainably use the oceans, seas and marine resources for sustainable development, -Goal 15.Protect, restore and promote sustainable use of terrestrial ecosystems, sustainably manage forests, combat desertification, and halt and reverse land degradation and halt biodiversity loss, -Goal 16.Promote peaceful and inclusive societies for sustainable development, provide access to justice for all and build effective, accountable and inclusive institutions at all levels, -Goal 17.Strengthen the means of implementation and revitalise the Global Partnership for Sustainable Development.As the MDGs practically applied only to developing countries, the SDGs are global and extended the focus of international development beyond poverty to sustainability (Adelman, 2017).Realization of previous UN sustainable development strategies (MDGs) proved challenging and met with mixed success, despite its limited scope. One cannot deny that SD (especially Agenda 2030 with its 17 SDGs) is a much broader concept than neoliberal focus on high economic growth, deregulation, privatisation and low inflation.However, the global financial crisis of 2007-2008+ and the subsequent Great Recession of 2009+ substantially undermined both neoliberalism and the current understanding of SD as big, private companies (dubbed too-big-to-fail) were bailed out by governments and central banks.It is argued that even the current understanding of SD (i.e. the 17 SDGs promoted by the UN and the EU) can be undermined or even falsified.The SDGs concept simply lacks predictive power; hence, according to the positive economics of Milton Friedman, it can be falsified.It is argued that a heterodox perspective would strengthen the informative and predictive power of SD indicators in the context of globalisation, financialisation and monopolisation.It is also claimed that the present set of SD indicators practically does not address the sources of the global financial crisis of 2007-2008+ and its repercussion in the real sphere via Great Recession in the following years. By using heterodox analysis (Post-Keynesian, institutional and evolutionary economics), it could be demonstrated that SDGs are properly structured in order to fit into neoliberal orthodoxy (not vice versa).This caveat also applies to the Millennium Development Goals (MDGs) that preceded the contemporary SDGs, as they "focused attention on the need to reduce absolute poverty" (Shafik, 2012).In contrast, heterodox scholars prefer relative measures of poverty and inequality.According to O'Grady (2016), the UN system has been under neoliberal assault for decades and is facing its own test of contemporary relevance.The characteristic of the 17 SDGs seems to confirm this statement. Food security The term "food security" was first defined at the World Food Conference held in Rome in 1974(5-16 November) by the United Nations.The conference was, in part, an answer to the challenges posed by two formidable food crises in Bangladesh in 1972 and 1974.Already these tragic events pointed to the multidimensional aspect of food security.While the government in Bangladesh succeeded in averting a widely predicted famine in the first case, it failed to prevent an actual famine in the later case when such a cataclysmic disaster was least anticipated (Dowlah, 2006).According to Dowlah (2006), "the 1974 famine was caused by successive onslaughts of natural disasters such as floods and droughts, and man-made disasters such as the government's inability to import foods, the directing of subsidised food to the politically vocal urban population, an abrupt fall in food aid and political and administrative corruption that encouraged massive hoarding and the smuggling of food grain". Nevertheless, the first definition of "food security" employed a narrow perspective which mainly focused on the concept of food availability (Simon, 2017).Yet, according to Caiafa and Wrabel (2019), "food security" encompassed the availability of food as well as the ability to access food.The 1974 World Food Summit defined food security as: "availability at all times of adequate world food supplies of basic foodstuffs to sustain a steady expansion of food consumption and to offset fluctuations in production and prices" (United Nations, 1975).The concept was further expanded by FAO in 1983 to include securing access by vulnerable people to available supplies, implying that attention should be balanced between the demand and supply side of the food security equation: "ensuring that all people at all times have both physical and economic access to the basic food that they need" (FAO, 1983). The definition was further adjusted by the World Bank in 1986.It introduced the distinction between chronic food insecurity (dealing with problems of continuing or structural poverty and low incomes) and transitory food insecurity (which involved periods of intensified pressure caused by natural disasters, economic collapse or conflict) "access of all people at all times to enough food for an active, healthy life" (World Bank, 1986). Later adjustments of the term in 1996 made it more comprehensive in order to address persistent global undernutrition and growing fear concerning worldwide agricultural capacity (a clear reference to Malthusian thinking).According to the World Food Summit (1996) declaration "food security, at the individual, household, national, regional and global levels [is achieved] when all people, at all times, have physical and economic access to sufficient, safe and nutritious food to meet their dietary needs and food preferences for an active and healthy life" (FAO, 1996). The next adjustment by FAO in 2002 introduced the social aspect of food security "Food security [is] a situation that exists when all people, at all times, have physical, social and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life" (FAO, 2002). The definition of "food security" further evolved building on the works of Indian economist and philosopher Amartya Sen, especially his influential text (Sen, 1981).Sen was analysing the entitlements of individuals and households rather than concentrating on the concept of food security.A new approach drawing from his research focused on consumption, the demand side and the issues of access by vulnerable people to food. For Sen, the poor lack many kinds or forms of freedom, which are perceived as obvious for the rich.A more recent confirmation of this finding is provided by Banerjee and Duflo (2012)."Development is something more than economic progress measured by the quantity of goods produced.(...) It is a social development: i.e. the increase of the number of people experiencing freedom springing from gaining the abilities indispensable to reach an adequate standard of living" (Kishtainy, 2017).Doubtless, the hierarchy of values favoured by Sen differs from the standard approach favoured by mainstream economists.Sometimes it could be easily recognised (Szydło, 2020a).For example, there is a clear contrast between "Development as freedom"the title of Amartya Sen's book published in 1999 and the titles of two books by Balcerowicz (a leading Polish free market economist) in which freedom plays the most important role: "Freedom and development.Economics of free market" (Balcerowicz, 1995) and "Freedom, development, democracy" (Balcerowicz, 2017).The approach favoured by the Indian economist, however, concentrates on development (i.e.'freedom to' as depicted by Berlin (1969)) rather than individualistic 'freedom from' (i.e.freedom from state coercion).Interestingly, according to Kowalik (2010), liberal Bochniarz and conservative Legutko expressed deep dissatisfaction when Sen was awarded the Nobel Memorial Prize in Economic Sciences in 1998. Recent understanding of the term "food security" incorporates four main dimensions (features, pillars): • Physical AVAILABILITY of food which addresses the "supply side" of food security and is determined by the level of food production, stock levels and net trade, • Economic and physical ACCESS to food -an adequate supply of food at the national or international level does not in itself guarantee household-level food security.Concerns about insufficient food access have resulted in a greater policy focus on incomes, expenditure, markets and prices in achieving food security objectives, • Food UTILIZATION -commonly understood as the way the body makes the most of various nutrients in the food.Sufficient energy and nutrient intake by individuals is the result of good care and feeding practices, food preparation, diversity of the diet and intra-household distribution of food.Combined with good biological utilisation of food consumed, this determines the nutritional status of individuals, • STABILITY of the other three dimensions over time -even if your food intake is adequate today, you are still considered to be food insecure if you have inadequate access to food on a periodic basis, risking a deterioration of your nutritional status.Adverse weather conditions, political instability, or economic factors (unemployment, rising food prices) may have an impact on your food security status (FAO, 2008).One approach to policies that encourage stability is to reduce the chances of shocks occurring in the first place.According to Caiafa and Wrabel (2019), "This can be achieved by adopting systems for monitoring and analysing food security risks to anticipate, and potentially attenuate, disruptions.Policies that support farmers' ability to produce food and contribute to national food stocks without stress or uncertainty about their income or livelihood are additional mechanisms for achieving this".More specifically, stability could be safeguarded, for example, by building buffers so that consumers can maintain their access to and use of food when the inevitable happens, rapid deployment of social safety net programs, plans for reintegrating refugees and displaced people, maintaining ecosystem integrity, mitigating the infrastructural and social effects of hazardous weather events, strengthening peacebuilding efforts to minimise conflict (Caiafa & Wrabel, 2019). The two additional dimensions of "AGENCY" and "SUSTAINABILITY" are proposed by the High Level Panel of Experts (HLPE) of the Committee on World Food Security (CFS)but are not formally agreed upon by FAO or other bodies, nor is there an agreed language on the definition (FAO, 2021).However, HLPE Report 14 and previous HLPR Reports recognised "agency" and "sustainability" as vital dimensions of food security that flow directly from the principle of the right to food.In a broader sense, "agency" is defined as "what a person is free to do and achieve in pursuit of whatever goals or values he or she regards as important" (Sen, 1981).According to Alsop and Heinsohn (2005), the agency goes beyond access to material resources in that it includes empowerment -the ability to take actions that help improve their own well-being, as well as their ability to engage in society in ways that influence the broader context, including their exercise of voice in shaping policies. In a narrow sense, connected with safeguarding food security, "agency implies the capacity of individuals or groups to make their own decisions about what foods they produce, how that food is produced, processed and distributed within food systems and their ability to engage in processes that shape food system policies and governance" (HLPE, 2020). The concept of "agency" has similarities with the notion of positive liberty described in Isaiah Berlin's seminal essay: "For the `positive' sense of liberty comes to light if we try to answer the question, not `What am I free to do or be?', but `By whom am I ruled?' or `Who is to say what I am, and what I am not, to be or do?" (Berlin, 1969).The 'positive' conception of liberty: freedom-to lead one prescribed form of life, is therefore contrasted with the 'negative' conception of liberty: the freedom which is involved in answer to the question "What is the area within which the subject -a person or group of persons -is or should be left to do or be what he is able to do or be, without interference by other persons?" (Berlin, 1969). Sustainability(the sixth overall dimension of food security, i.e. the second extra dimension) was initially defined as the sustainability of food systems in all three dimensions: economic, social and environmental, in their capacity to ensure good quality and adequate food for this generation and future generations (HLPE, 2014).After a minor refinement, it presently refers to "the long-term ability of food systems to provide food security and nutrition today in such a way that does not compromise the environmental, economic, and social bases that generate food security and nutrition for future generations" (HLPE, 2020).Despite long efforts to overcome hunger, in as many as 46 countries the prevalence of undernourishment exceeded 10% in 2019.Even in the years preceding the COVID-19 pandemic the undernourishment was on the rise since 2013-2015.The figures for Africa (especially Middle, Eastern and Sub-Saharan Africa) are alarming.The positive tendency was reversed in all African regions, Asia and South America.Although, the prevalence of undernourishment in Asia (especially Central Asia and South-Eastern Asia) has diminished in the last two decades. Unfortunately, more sophisticated and comprehensive food security strategies did not lead to lowering the prevalence of undernourishment in the recent 6-8 years. Interestingly, the commodity price boom (including food, fertilisers and crude oil) in 2004-2008 and the global financial crisis did not immediately translate into higher levels of undernourishment.The question remains whether it springs from measurement problems.It well might be that these economies are relatively closed (possibly more self-sufficient) and, therefore, better insulated from external shocks. Higher dynamics of fertiliser and oil prices in comparison with food prices (as in the case of recent two commodity price booms in 2004-2008 and 2021-2022) must lead to the extraction of income and wealth from the agricultural sector into oligopolies or monopolies producing fertilisers and oil.Agenda 2030 does not comprise indicators measuring price changes of these two vital types of commodities but narrowly concentrates only on food price anomalies in Goal 2. It is hardly surprising as economic orthodoxy has avoided addressing asset bubbles and recognising the existence of cost-push inflation. Forty years of neoliberal fixation on GDP growth coupled with successive UNSD strategies (including MDGs and SDGs) proved unsuccessful in eradicating hunger and poverty.The number of undernourished stubbornly hovers well above 785.4 million (the level in 2015) and has been on the rise in recent years.The prevalence of severe food insecurity in the world population increased from 7.7% in 2014 to 11.7% in 2021, while the prevalence of moderate or severe food insecurity in the total population (percent) expanded from 21.2% to 29.3% in the same period. Sustainable technologies in agriculture and sustainable development paradigm Undoubtedly, one has to agree with Singh et al. (2022) that "sustainable technology-led agriculture is the need of the hour to enhance and maintain the ecosystem not only for the present generation but also for future generations".Particular challenges are faced by small farms in developing countries.Implementation of sustainable agricultural technologies (Singh et al., 2021) may prove too expensive to implement, as in the case of drones (Singh et al., 2022).Naturally, it also applies to technologies connected with smart farming aiming to improve crop yield and product quality (such as GIS remote sensing, nanotechnology, and genome editing tools, including molecular biological techniques) (Singh et al., 2021).Development of these modern methods requires substantial funds, highly trained specialists and an adequate process of certification in line with the precautionary principle of the sustainable development paradigm.Given small spending on research and development (R&D) in the majority of countries, more advanced methods could only be developed in selected high-income countries (i.e. the USA, Canada and West European countries, such as Germany, the UK, France, Switzerland, etc.) and big developing countries such as China and India.Therefore, the majority of countries would have to import modern, often expensive technologies.Similarly, 'green' organic food may, unfortunately, prove too costly for most consumers in developing countries. A number of smaller developing countries even do not have the capacity to produce fertilisers (N, P, K).Hence, in order to improve yield to safeguard food security, they have to rely on imports.During asset price bubbles between 2005 and 2008 (i.e.prior to the Global Financial Crisis) and during conflicts or war (as in the case of the Russian-Ukraine war), this strategy poses a serious risk.The international prices of strategic commodities such as oil (extensively utilised by agricultural machines, especially in developed and developing countries), gas (used in the process of fertiliser production), and NPK fertilisers increase even faster than food.Paradoxically, least developed countries (LDCs) are not severely affected by this type of crisis as they could not even afford to import fertilisers prior to asset bubbles (i.e. when the prices of fertilisers were relatively low).The vulnerability of fertili ser--importing countries with floating exchange rates is further reinforced by adverse currency movements (i.e.depreciation of the local currency against USD), which became a norm during financial and economic crises. Therefore, the 21st-century crises prove that safeguarding food security cannot rest solely on the belief in the free-trade concept advocated by the WTO.Food security can only be maintained by the adequate implementation of broad sustainable development (SD) paradigm comprising at least five pillars or orders (i.e. economic, social, environmental, institutional and spatial).Global Financial Crisis and subsequent Great Recession also show that the inclusion of financial order as the sixth pillar of SD is indispensable.Limiting the understanding of the sustainable development paradigm to the environmental order (pillar) as in the Constitution of the Republic of Poland of 1997 would not effectively address various challenges faced by humankind in the 21 st century (Szydło, 2020b).To illustrate the point, article 5 of the Constitution states: "(t)he Republic of Poland shall (...) ensure the freedoms and rights of persons and citizens, the security of the citizens, safeguard the national heritage and shall ensure the protection of the natural environment pursuant to the principles of sustainable development".A similar approach employs by Singh et al. (2022), who define SD as "a set of principles that guide us to effective utilisation of natural resources without undermining their integrity and stability for future generations".It could well be argued that there is a need to return to the original definition of SD, which was presented in World Conservation Strategy (IUCN, 1980).Only broad cooperation of scholars from various scientific fields (i.e.agronomy, biology, chemistry, ecology, economics, geography, meteorology, physics, etc.) could safeguard a better future for present and future generations. At the same time, certain ecological ideas blaming agriculture for massive CO 2 emissions because of the huge livestock population (Warner, 2021) should be treated with caution.The data above clearly show that the share of agricultural methane emissions in the proxy of total anthropogenic CO 2 emissions fell from 12.95% in 1990 to 8.95% in 2019. Transforming Our World: the 2030 Agenda for Sustainable Development" (often referred to as Agenda 2030 or Sustainable Development Goals -SDGs) is the current United Nations (2015) sustainable development strategy which covers 15 years (2016-2030).It was adopted by 193 countries in the UN General Assembly on September 25, 2015.Agenda 2030 replaced the UN Millennium Development Goals (MDGs), which were set by the 189 UN member states following the Millennium Summit in New York, 6-8 September 2000 and Millennium Declaration (A/RES/55/2) and prior to the World Summit on Sustainable Development in Johannesburg (26 Aug.-4 Sept. 2002).Hence, the eight (rather narrow) Millennium Development Goals: • to eliminate extreme poverty and hunger, • to achieve global primary education, • to empower women and promote gender equality, • to reduce child mortality, • to promote maternal health, • to fight malaria, HIV/AIDS, and other diseases, • to promote environmental sustainability, • to develop a universal partnership for development, were superseded by the Agenda 2030 with the 17 Sustainable Goals (SDGs): Table 2 . Bank (2022)l methane emissions (metric tons of CO 2 equivalent per capita) as a share of total (anthropogenic) CO 2 emissions* [%] Proxy: the sum of CO 2 emissions (metric tons per capita) and agricultural methane emissions (metric tons of CO 2 equivalent per capita); Carbon dioxide emissions are those stemming from the burning of fossil fuels and the manufacture of cement.They include carbon dioxide produced during the consumption of solid, liquid, and gas fuels and gas flaring.Source: author's work based on WorldBank (2022). *	2023-09-17T15:09:27.728Z	2023-09-14T00:00:00.000	{ "year": 2023, "sha1": "6ca9700fd27766eb6293159e35176c99e71427fa", "oa_license": "CCBYSA", "oa_url": "https://www.ekonomiaisrodowisko.pl/journal/article/download/560/528", "oa_status": "HYBRID", "pdf_src": "Anansi", "pdf_hash": "71cc49514b6e8e9f8ab5da6df91d38a815190b12", "s2fieldsofstudy": [ "Environmental Science", "Economics", "Political Science", "History" ], "extfieldsofstudy": [] }
4412244	pes2o/s2orc	v3-fos-license	Atypical information-use in children with autism spectrum disorder during judgments of child and adult face identity ABSTRACT Unusual patterns of fixation behavior in individuals with autism spectrum disorder during face tasks hint at atypical processing strategies that could contribute to diminished face expertise in this group. Here, we use the Bubbles reverse correlation technique to directly examine face-processing strategies during identity judgments in children with and without autism, and typical adults. Results support a qualitative atypicality in autistic face processing. We identify clear differences not only in the specific features relied upon for face judgments, but also more generally in the extent to which they demonstrate a flexible and adaptive profile of information use. difficulties in the condition. Interestingly, older autistic children showed no particular preference for either half; with their use of the top and bottom face halves hinting at a combination of the profiles observed in children from the other two groups. This research interest may partly reflect that several conceptualizations of autism give cause to predict atypical processing strategies and information use during face judgments. Atypical looking at the eyes could stem, for example, from active avoidance of direct eye contact in the condition due to emotional arousal associated with this potent signal of social engagement (Kylliäinen & Hietanen, 2006;Tanaka & Sung, 2013). Equally, individual differences in scanning patterns could reflect atypical communicative skills in autism, driving variability in fixations to the eyes (particularly implicated in complex socio-emotional interactions) vs. the mouth (more critical for language and speech-related information) (Falck-Ytter, Fernell, Gillberg, & von Hofsten, 2010;Norbury et al., 2009). Finally, it could indicate an immature processing strategy associated with limited perceptual expertise with these social stimuli and a lack of appreciation of the importance/utility of this region (Itier & Batty, 2009). Irrespective of its origins, such an atypical profile is likely to negatively impact expertise because the eyes constitute a critical cue for face reading (Peterson & Eckstein, 2011) Eye-tracking studies have unquestionably provided helpful insights into how participants read information from faces (Boraston & Blakemore, 2007). Yet there are limits to the utility of gaze behavior as an independent index of information use and processing strategies. Fixations are just one of a series of processing subcomponents that culminate in a discrete social judgment. They reflect a range of bottom-up and top-down influences, and can vary independently of visual attention (Triesch, Ballard, Hayhoe, & Sullivan, 2003;Turatto, Angrilli, Mazza, Umità, & Driver, 2002). An alternative and complementary approach to revealing face-processing strategies and profiles of information use for face judgments is the "Bubbles" reverse correlation technique (Gosselin & Schyns, 2001). This experimental paradigm allows researchers to pinpoint the specific visual information associated with participant outcomes on a categorization task (e.g., accuracy, reaction time, EEG activation). In the context of face-processing tasks, test images are presented to participants for categorization (e.g., identity, emotional expression, gender) with the systematic addition of visual noise across trials: randomly positioned Gaussian apertures or "bubbles" that reveal only subsampled regions of each stimulus on each trial (Gosselin & Schyns, 2001). Upon completion of the task, when the entire stimulus space has been sampled, performance across trials is analyzed to generate classification images that reveal the visual features significantly associated with correct (cf. incorrect) performance. That is, the pixels/cues/features that participants relied upon significantly for their judgments. Spezio, Adolphs, Hurley, and Piven (2006) used this technique with autistic adults to reveal faceprocessing strategies during expression categorization judgments (fear vs. happy). Relative to a group of typical adults of similar cognitive ability, autistic individuals demonstrated equivalent levels of performance accuracy but with a distinctly atypical profile of information use. Classification images indicated that autistic adults relied more upon information from the mouth region and less on information from the eye region than did their typical counterparts. Concurrent eye tracking revealed close alignment between these patterns of information use and fixation behavior in each participant group. That is, the autism group consistently fixated more on the mouth and slightly less on the eyes than did the typical group. This association was interpreted as evidence that atypicalities in social gaze behavior in autism strongly contribute to differences in information use during face tasks. In a more detailed investigation of participants' saccade behavior during that original task, Spezio and colleagues (2007) also identified diminished specificity in this gaze to the mouth in autistic adults. Thus, this selective looking to the mouth was observed even when useful information was available in other regions (e.g., the eyes) that could have aided their categorizations. In line with this result, another study that used the same paradigm with a different group of adults confirmed consistent deviation from the pattern of fixation behavior predicted by computational models of stimulus salience and observed in the typical population, that is, in favor of looking at the mouth in the autistic group (Neumann, Spezio, Piven, & Adolphs, 2006). Together, these findings strongly suggest that reliance upon information from the mouth region during expression judgments may constitute a top-down driven information-processing bias in autism. Only one study has investigated face-processing strategies in children on the autism spectrum using Bubbles. This dearth of research is surprising; researchers are often particularly interested in children because they are presumed to be less likely than adults to have developed compensatory strategies. Song, Kawabe, Hakoda, and Du (2012) investigated information use in school-aged autistic and typical children of similar age and cognitive ability during emotion and identity categorizations (Song et al., 2012). The inclusion of an identity task allowed unique exploration of a judgment that enlists highly specialist processing resources in the typical population (see Schwaninger, Carbon, & Leder, 2003) and is known to be especially challenging for individuals with autism (Robel et al., 2004;Scherf, Behrmann, Minshew, & Luna, 2008;Serra et al., 2003). Results indicated that autistic children's information use during expression judgments closely resembled the typical children's profile. On the identity task, however, counter to expectations based on previous eye-tracking research with autistic children and previous Bubbles experiments with autistic adults, children with autism were not less reliant upon information in the eye region than were typical children (Song et al., 2012). Moreover, the key point of difference from the typical children, who consistently relied on both the eye and mouth regions, was that rather surprisingly, the autistic children relied almost exclusively on the eye region. This unexpected finding has yet to be replicated, and it remains unclear whether these intriguing results reflect a genuine atypicality in the salience and/or importance of the eye region for identity judgments in children with autism. Song and colleagues substantially adapted the Bubbles experimental paradigm to ensure that it was appropriate for their sample (Song et al., 2012). They kept trial numbers very low (80 per task), used child rather than adult face images in the task (a smiling and a neutral version of two identities) and presented these test images in pairs, rather than individually, to simplify the required participant response (i.e., "which of these faces is IdentityA/Happy?"). It is possible that these modifications contributed to their unexpected findings. For example, the limited number of trials might have prevented adequate sampling of the stimulus space; children's faces might have been less distinctive than adult faces, which could have distorted participants' profiles of information use, particularly for identity judgments; and presenting faces in pairs might have encouraged an atypical, low-level feature-matching processing strategy. The current study sought to rule out these possibilities by investigating information use during face recognition judgments in autistic children using a more traditional Bubbles experimental paradigm. Our bubbles identity categorization task included as many trials as possible for the participant groups employed (based on pilot testing) to ensure optimized sampling of the stimulus space (216 trials). We also used the traditional individual presentation of test faces for categorization to encourage high-level face processing and probed judgments of both child and adult face stimuli in separate versions of the task (administered on different days). Based on the findings of previous adult Bubbles research and other evidence suggestive of a qualitative autistic atypicality in the use of the eyes and mouth during face tasks, we hypothesized that autistic children would use information from the eye region selectively less than typically developing children for their identity categorizations. Typical adults were included as a secondary comparison group. Their inclusion allowed us to place results from the typical children in a developmental context and to relate findings from this abbreviated bubbles paradigm with those from more exhaustive investigations of identityrelated information use (e.g., Butler, Blais, Gosselin, Bub, & Fiset, 2010;Gosselin & Schyns, 2001;Schyns, Bonnar, & Gosselin, 2002). Predictions regarding autistic vs. typical participants' performance profiles for the child and adult faces were less straightforward. Given some previous reports of superior ability with own-relative to other-age faces (Anastasi & Rhodes, 2005;Hills & Lewis, 2011), it seemed possible that we might observe differences in face-processing strategies for these different categories in our typically developing participants. Such own vs. other-age face differences, however, may be absent (or be present to a lesser extent) for autistic children, who less reliably demonstrate in/out-group processing biases (see Chien, Wang, Chen, Chen, & Chen, 2014;Wilson, Palermo, Burton, & Brock, 2011;Yi et al., 2015). Ethics statement This study was approved by the Human Research Ethics Committee at the UCL Institute of Education, University College London. All adults and parents provided written consent prior to their child's participation in the project. All children also gave verbal assent before taking part. Participants Participants were eight cognitively able autistic children (five male), eight typically developing children (three male) and eight typical adults (three male). See Table 1 for detailed descriptive information. The sample size is similar to Bubbles studies investigating information use in adults with ASD (e.g., Neumann et al., 2006;Spezio et al., 2006Spezio et al., , 2007. These children were recruited from a primary school in London (with autism specialist and mainstream classrooms) and adults were personal contacts of the researchers. Both typical groups did not have any personal history of autism spectrum disorder or psychiatric disorders. Parents of the typical children also completed the Social (Raven, 1998)) and the British Picture Vocabulary Scale III (BPVS, (Dunn et al., 2009)). b Higher scores on the SCQ (Social Communication Questionaire -Lifetime form; Rutter et al., 2003) indicate a greater degree of autism symptomatology. c Face processing ability measures: Adults completed the Cambridge Face Memory Test (Duchaine & Nakayama, 2006), children completed the Cambridge Face Memory Test -for Children (Croydon et al., 2014). Scores = accuracy (total percentage correct). Communication Questionnaire (SCQ, Rutter, Bailey, Lord, & Berument, 2003), which revealed scores well below the cut-off for clinically significant autism symptoms (15), see Table 1. All autistic children had received an independent clinical diagnosis following DSM-IV criteria (American Psychiatric Association, 2000) and scored above 21 on the SCQ (Rutter et al., 2003). This group was significantly older than the typically developing comparison group of children, but otherwise did not differ with regards to verbal ability, as measured by the British Picture Vocabulary Scale (BPVS III, Dunn, Dunn, Styles, & Sewell, 2009) Raven, 1998) (see Table 1). They did, however, perform more poorly than the typical group on a standardized measure of face-recognition ability (Cambridge Face Memory Test for Children; CFMT-C, Croydon et al., 2014). Adults would perform at ceiling on these children's measures, so were assessed only for face-recognition ability (for completeness), using the adult form of the Cambridge Face Memory Test (CFMT, Duchaine & Nakayama, 2006). All participants had normal or corrected to normal vision, as reported by themselves and/or their parents. Stimuli Stimuli were grayscale photographs of neutral expression faces taken from stimulus databases with standardized pose and lighting conditions (see Figure 2). We used two adult male identities (from Schyns & Oliva, 1999) and two (approximately) 9 year-old male identities (JimStim database, University of Victoria). Hairstyle and feature locations were standardized within each age stimulus set using Adobe Photoshop. The Puzzle Bubble Game-child and adult versions In the Puzzle Bubble Game, learned face identities were presented individually for participants to identify with a verbal response or labeled key-press ("Bob or Ted" adult identities, "Guy or Max" child identities). The task was challenging because participants were provided with only subsets of information on a given trial, revealed through pseudo-randomly positioned circularly symmetric Gaussian apertures or "bubbles". The rest of the image was hidden from view (for full methodological details, see Gosselin & Schyns, 2001). To minimize trial numbers, only the central portion of the stimulus images -including the entire face stimulus itself but not the outer dark grey area surrounding the faces, was sampled with bubbles during the experiment. An adaptive staircase algorithm was used to adjust the sampling density (i.e., total number of bubbles) on each trial to target participants' accuracy at 75% correct (minimum 40 bubbles, maximum 250 bubbles). That is, when performance was low we presented participants with more visual information to guide their judgments, and when performance was high we presented less information. This personalized calibration of bubble numbers ensured that the task was comparably challenging across participant groups. Stimuli were projected on a light gray background to the center of the screen at a viewing distance of approximately 50 cm, subtending 6.5 × 6.5°of visual angle (similar to 5.7 × 5.7 in Gosselin & Schyns, 2001). The child and adult versions of the task were identically structured, differing only in the to-becategorized stimuli. Each game began with a training phase (12 trials), during which participants learned the names of the two test identities and practiced categorizing them. They were encouraged to look carefully at each picture and if they were unsure, to take their "best guess". In this training phase the identities first appeared intact for an unlimited amount of time (twice each), and then intact for 1000 ms (twice each) and then "with bubbles" for 1000 ms (twice each) to familiarize and prepare participants for the main test trials. Auditory accuracy feedback was provided during this training phase. A minimum 75% level of performance accuracy during this training was required in order to progress to the main task. The main test trials comprised nine blocks of 24 test trials (216 total). On each trial, a centrally presented test face appeared for 1000 ms, followed by a blank screen until the participant made their response. Between-block breaks provided participants with generic encouragement (e.g., screens saying "keep up the great effort", odd-numbered blocks) or an engaging task-irrelevant game (even-numbered blocks). In this game (The Puzzle Bubble Challenge) participants identified "bubbled" images of films, TV shows, or geographical locations (category = participant's choice) with as few added "clues" as possible, which each revealed more visual information to make their task easier. The visual information significantly associated with correct categorization performance for each stimulus face age in each participant group shown on a sample face image (red regions are significant p<0.01, green regions at p<0.05, blue regions at p<0.1). For the autistic group, the mouth significantly drives correct performance. Second row: Diagnostic images depicting only that information significantly associated with correct performance (p<0.05). Figure 2B. As Figure 2A for adult face stimuli. Figure 2C. Child faces. Top row: The visual information that is used significantly more by typical children than autistic children (left column) and more by autistic children than typical children (right column) on a sample face stimulus (red regions at p<0.01, green regions at p<0.05 and blue at p<0.1). Bottom row: Diagnostic images depicting only that information whose use differs significantly between the typical children and autistic children. Figure 2D. As Figure 2C for the adult face images. Figure 2E. Computationally determined salient visual information available to differentiate the two face identities in each task (red colour indicates greatest salience, blue colour indicates minimal values). A thresholded version applied to a sample stimulus image highlights the corresponding visual information for comparison. Figure 2F. As Figure 2A for the adult participants Procedure All participants completed both the child and adult face versions of the Puzzle Bubble Game along with our additional measures (children: CFMT-C, BPVS, RCPM; adults: CFMT) during two (children) or one (adults) 30-45 minute session/s. The order of the child and adult face versions was counterbalanced across participants. The computer tasks were run on a 13-inch Samsung Notebook computer. An experimenter sat alongside each participant at all times to monitor engagement and provide one-on-one encouragement. Participant performance metrics Our behavioral measure of performance accuracy during the categorization task was percentage correct (Figure 1(a)). Despite using a staircase algorithm to calibrate task difficulty and maintain performance at 75%, the use of an unbiased and equivalent "starting point" for all participants (125 bubbles) meant that accuracy was not necessarily matched perfectly across groups within the 216 experimental trials. A 2-way repeated-measures ANOVA investigating the impact of face stimulus age (child, adult) and participant group (autistic children, typical children, adults) on this variable revealed a significant effect only for participant group, F(2,21) = 7.09, p = 0.004, η p 2 = .40. The effects of stimulus age and the interaction were not significant, Fs < 2.31, ps > 0.13. Importantly, the effect of participant group did not reflect any difference in categorization accuracy between autistic children (M = 74.1, SD = 3.9) and typical children (M = 74.8, SD = 4.8), t(14) = 0.39, p = 0.69). Rather, it was that typical adults performed significantly better (M = 79.9, SD = 4.4) compared to both child groups (ts >2.89, ps < 0.01). A second 2-way repeated-measures ANOVA investigated the effects of face stimulus age and participant group on the amount of information (median number of bubbles) that participants required to reach these performance levels (see Figure 1(b)). Once again, there was a significant main effect of participant group (F (2,21) = 4.64, p = 0.02, η p 2 = .30) but again, this effect did not reflect differences between the two child participant groups. To achieve the comparable levels of categorization accuracy reported above, the autistic children (M = 55.8, SD = 33.0) needed numerically but not significantly more visual information than typically developing children (M = 49.7, SD = 39.5), t(14) = 0.45, p = 0.65. Unsurprisingly, adults required significantly fewer bubbles than both child groups to achieve their superior accuracy levels (M = 22.9, SD = 15.0), ts > 2.49, ps < 0.02). This analysis also identified a main effect of face stimulus age, F(1,21) = 4.64, p = 0.04, η p 2 = 0.18. Interestingly, this result did not reflect the child faces (Guy and Max) being more difficult to discriminate than the more mature adult faces (Bob and Ted). Instead, the reverse was true: overall participants needed fewer bubbles to identify the child faces (M = 34.2, SD = 26.8) than the adult faces (M = 51.5, SD = 37.8). There was no interaction with participant group, F(2,21) = 0.76, p = 0.47, η p 2 = 0.06. There was no significant association between either of these performance metrics and participants' age, verbal ability or non-verbal ability in the autistic or typical child groups (all ps > 0.08, Kendall's tau_b nonparametric correlations). Classification & difference images We followed standard approaches (e.g., Gosselin & Schyns, 2001) to determine the specific information associated with correct categorization performance in the child and adult stimulus versions of the task. Each trial was sorted as a function of whether the information presented to the participant resulted in a correct or incorrect identity categorization response. Observers tend to be correct if the information necessary to perform a task is available to them and incorrect if this information is missing. Thus, we summed together all of the bubble masks (pixel locations of available visual information) leading to correct categorizations in each case and divided this by the sum of all bubble masks presented during that task. The resulting classification images represented the probability that presenting visual information at each pixel location would lead to a correct response. The classification images were transformed into z-scores using the non-informative normalized hairstyle region at the top of the image space as a baseline. We established those regions that were statistically associated with correct categorization performance by applying a threshold criterion on the z-scores. The information found to be significantly associated with correct categorization performance, termed the diagnostic information, was superimposed in red (p < 0.01), green (p < 0.05), and blue (p < 0.1) on a representative face image to reveal its location (Figure 2(a) and 1(b), Significant Regions). We chose deliberately to include liberal (p < 0.1) as well as the more standard, conservative (p < .05, p < .01) thresholds when presenting the results to ensure that no important visual features were missed due to not quite reaching these (somewhat arbitrary) criteria. To create the diagnostic image, we displayed the information significantly associated with correct performance (at the p < 0.05 level) on a representative face (Figure 2(a,b), bottom row). These diagnostic images serve to further highlight those regions significantly driving performance in each group. This process was completed separately for the child and adult stimulus versions of the Bubbles task and separately for autistic children (Figure 2(a)), typical children (Figure 2(b)), as well as the adult comparison group (Figure 2(f)). To compare information use across autistic and typical children we also computed the difference of the z-scored maps. We used the un-thresholded z-score maps that included all pixel values greater than zero (indicating greater than average association with correct performance) rather than use thresholded images which may result in misleading findings if some features are associated with performance but do not quite pass a significance threshold. These differences were re-normalized to the baseline region and we applied the same probability threshold criteria. Figure 2(c) illustrates the visual information that is significantly more used by typical children vs. autistic children (left column), and the reverse (right column) when categorizing child faces, with regions highlighted in red (p < 0.01), green (p < 0.05), and blue (p < 0.1) on a sample face image. Revealing only this significant information (p < 0.05 level) on a sample face provides a clear indication of those facial regions used more by typical children (corner of the nose) and used more by autistic children (left side of the mouth). Figure 2(d) similarly indicates the group differences for the adult faces. Here, typical children make relatively more use of the left-sided eye. There is very little facial information that autistic children use more than their typical peers for these adult faces. During the child face task, the classification images also revealed some differences in the information used by autistic children compared to the typical children (Figure 2(a)). For example, the autistic children demonstrated a focused reliance upon one particular feature in isolation, whereas typical children drew upon a slightly broader set of face cues. Crucially, the singular, significant point of focus for the children with autism was the mouth region. This finding contrasts directly with the findings of Song and colleagues (Song et al., 2012) but fits well with the widely reported autistic bias to look relatively less at the eyes than typical individuals (Tanaka & Sung, 2013). Given this result, it is tempting to speculate about a possible link between differences or deficits in making use of visual information in the eye region in autism and atypical/impaired identity processing in autism. Yet any such association is invalidated by the parallel mouth-focus also observed in typically developing children, as well as adults (Figure 2(f)). Just like the children with autism, typical children also failed to make significant use of the eye region during their identity categorizations of these child faces. They relied instead upon the mouth and a slightly larger area of the face, encompassing also some of the nose and cheeks. Looking at the classification images for the adult face stimuli, we observe a different-more traditional profile of information use in typical children and adults (see Figure 2(b,f)). Here, they relied significantly upon information in the eye region (particularly the left side eye) and the mouth region, as reported in Butler et al. (2010), Caldara et al. (2005), Schyns et al. (2002). Small idiosyncrasies were observed between these two typical groups, for example, adults consistently also used a left side jawline cue. Generally, however, both showed a similar profile of information use for the adult faces, which differed distinctly from that we observed for their child categorizations. These results highlight that for typical participants, the most efficient face processing strategy can vary depending on the task and specific stimuli presented (Smith & Merlusca, 2014). It was interesting to note then, that the same was not true here for children with autism. Instead, this group used the same strategy with the adult faces as they had with the child faces. That is, they persisted in their strong reliance upon information in the mouth region. These strategy differences are borne out in the difference images where the typically developing children are shown to make more use of the eye area for adult faces, and the side of the nose for the child faces, whereas the autistic children consistently focus more on the mouth. Saliency model We used the Graph-Based Visual Saliency metric (Harel, Koch, & Perona, 2007) to identify the most salient cues available for the identity discrimination judgment in each face set from the stimulus images. We computed a saliency map of the visual difference between the two identities separately for the two child faces (i.e. child face 1 minus child face 2) and the two adult faces (adult face 1 minus adult face 2). These saliency map images highlight the pixel locations of the most objectively salient regions in our to-be-discriminated test stimuli using metrics based on biologically grounded models of the early primate visual system (Itti & Koch, 2001). Figure 2(e) provides the result of the saliency model for the child and adult face stimuli. Applying a threshold to the saliency metric allows the most salient regions to be visualized on a sample face stimulus to permit direct comparison with the visual information used by participants discriminating the images (Figure 2(e), bottom row). Importantly, these results indicate that the profile of information use that we observed for the child faces across all three participant groups (focused largely on the lower half of the face) is closely aligned with those features highlighted by the saliency model. Similar alignment of participants' behavior and test stimulus properties was observed for the adult faces, which proved to be particularly important for interpreting our results. The focus on the left eye observed in the two typical groups (and also reported in other studies) could have been viewed as a product of the lateralization of face processing (Meng, Cherian, Singal, & Sinha, 2012;Rhodes, 1985). Crucially, however, the saliency model results indicate that the left eye also happened to be objectively useful for discriminating between the two adult face identities used in this experiment. Figure 2(e) indicates also that the mouth region was not a particularly salient cue for these particular stimuli. Nevertheless, all three participant groups relied significantly on information in this region (as also found previously Gosselin & Schyns, 2001; though see Butler et al., 2010). The use of these less than salient cues in the typical children and adults suggests that a bias to sub-optimally encode redundant facial features is not unique to autism. Discussion The current study examined how autistic children go about the complex process of reading identity information from faces. Our carefully designed developmental adaptation of the Bubbles reverse correlation technique allowed us to identify the cues that autistic children, typically developing children, and adults rely upon during a challenging identity categorization task. Results revealed a consistent, atypical bias in autistic children to focus on the mouth region for these judgments. This bias held irrespective of the to-be-discriminated stimuli, that is, whether they were child or adult faces. This behavioral profile differed markedly from the more flexible profile observed with typical children, who demonstrated a qualitatively similar profile to adults. The atypical information use we observed in autistic children cannot be explained as the product of baseline differences between our participant groups. It is unlikely, for example, to reflect the increased age of the autistic children relative to the typical children because there was no significant correlation between age and either of our performance metrics during the Bubbles tasks. Our Bubbles design allowed us to equate categorization performance (percent correct accuracy for child and adult faces) and overall processing efficiency (number of bubbles) to highlight this qualitative difference in their face-processing strategy, which could be contributing to difficulties with face perception observed here and elsewhere. Between-group differences were also observed in the distinct response profiles generated across the child and adult face tasks. Though the stimuli in each version were similarly standardized, visual saliency maps highlighted distinct face regions that were more and less salient for categorizations of the identities in these child and adult face pairs. Specifically, the eyes were confirmed to be particularly discriminative for the adult faces, the mouth for the child faces. Our classification images signal that typical but not autistic children were sensitive to this variability. Categorization behavior revealed that for the autistic children, the mouth region was always diagnostic, irrespective of the test stimulus. Typical children, however, flexibly modulated their information use across tasks: broadly in line with the features that were most salient. Similarly strategic information use was observed for the adult participants, and has been reported in other adult studies exploring other categorical face decisions (e.g., Schyns et al., 2002). Autistic children demonstrated a fixed processing strategy for face identity that relied consistently on information in the mouth region. Such an approach fits with several eye-tracking studies that have similarly observed a particular focus on the mouth (Jones, Carr, & Klin, 2008;Klin, Jones, Schultz, Volkmar, & Cohen, 2002) and could reflect several different mechanisms. A bias toward the mouth (often also associated with a bias away from the eyes) could be associated with atypical communication in the condition (Langdell, 1978), an aversion to the socially intimidating eye region (Tanaka & Sung, 2013) or a failure to appreciate the utility of this information (Itier & Batty, 2009). Regardless of its origins, a mouth bias could negatively influence face-processing ability by preventing the exploitation of useful cues in the top half of the face (Peterson & Eckstein, 2011). Certainly, in the current study, performance in the autistic children was poorer than that of their more flexible and strategic comparison group of typical children. The fixed, particular reliance upon the mouth region that we observed in autistic children contrasts not only with the behavioral profile observed in our typical comparison groups, but also with the results of the only previous investigation of information use in autistic children. Song and colleagues (2012) reported a strong reliance upon the eyes, rather than the mouth, during identity judgments in their Bubbles study. It is difficult to draw strong conclusions about the cause of the discrepancy between these findings because many methodological differences distinguished the two experimental tasks. Our confidence in the current findings is drawn from the extent to which our task closely resembles the classic Bubbles experimental paradigm, for example, with individual presentation of test stimuli. Moreover, our results are consistent with other behavioral and eyetracking evidence that supports an autistic focus to fixate upon the mouth region (see Tanaka & Sung, 2013) and previous Bubbles research conducted with autistic adults reporting a strong reliance upon the mouth during fear vs happy judgments (see Spezio et al., 2006Spezio et al., , 2007. It is important to note that the face processing strategies observed in typical children, as well as adults, were also far from (objectively) perfect. Even though both these participant groups seemed to recognize the utility of the information in the eye region when categorizing adult face stimuli, they also continued to draw upon the mouth region. This information was used even though the mouth was not particularly helpful for discriminating between these two particular identities (confirmed by the saliency model). Such a bias to draw information from the mouth is very much in line with other published work on information use during face judgments, which is suggested to be broadly optimized to support face expertise (Smith, Cottrell, Gosselin, & Schyns, 2005). We speculate that across their extensive face experience accumulated from infancy, typical individuals might develop a "default" face processing strategy, which is then flexibly adapted to match stimulus characteristics and task demands but was not wholly recalibrated in the 216 trials of the current study. The reverse correlation technique provides a highly sophisticated means to pinpoint the specific information significantly driving categorization judgments. There are necessary limits, however, to the ecological validity of such tasks. We consciously avoided overloading our participants by asking them to learn a large set of identities. Yet outside of the experimental context, face identity judgments are clearly more complex than the two-choice categorizations assessed here. Moreover the flexible face profiles of typical information use observed across stimulus categories in the current study confirm that participants' experience with the test identities can impact upon performance outcomes, including face-processing strategies. The current study sought to characterize face-processing strategies in autistic children, typical children and adults by elucidating the visual information that drives identity judgments. Our results indicate that autistic children differ from typical children not only in the specific features that they rely upon for these judgments of child and adult faces, but also more generally in the extent to which they demonstrate a flexible and adaptive profile of information use in this domain. These results were striking, even in our small sample of autistic participants-which was comparable to most previous studies in this domain, e.g., Spezio et al. (2006) tested nine adults, Spezio et al. (2007) tested eight adults, and Neumann et al. (2006) tested ten adults. We acknowledge that the trial numbers were relatively small in the context of "classical" bubbles research (e.g., Gosselin & Schyns, 2001) but note that they were not far from the more modest numbers that have led to stable solutions in individual level analyses associated with EEG studies (e.g., Schyns, Petro, Smith, 2007). The profile of information use observed with typical adults converges nicely with solutions obtained in more exhaustive testing sessions and, perhaps most crucially, the total number of trials per participant was considerably higher than the only other published study conducted with autistic children. 1 Having identified a clear, potentially developmentally stable qualitative difference in autistic face processing strategies, an interesting future direction for this research will be to more directly investigate the functional consequences with respect to processing ability. There is a broad consensus that efficient (i.e., in some sense optimized) information use and flexible processing strategies support typical face expertise, but this link is yet to be empirically tested. It is true that evidence of atypical strategic information use in populations with face reading difficulties are consistent with this notion, for example, autism spectrum disorder (e.g., current study, also Neumann et al., 2006;Spezio et al., 2006Spezio et al., , 2007 and prosopagnosia (e.g., Caldara et al., 2005;Xivry et al., 2008). Still these groups demonstrate other, potentially influential visuoperceptual and/or social atypicalities, making it an interesting open question whether this association truly holds and/or extends to the typical population. Directly assessing and contrasting profiles of information use in high-and lowerperforming ability children and adults could highlight the functional consequences of qualitative differences in face processing strategy. Findings could provide an evidence base for training programs to improve skills in those with clinical and non-clinical difficulties in this domain. Note 1. In the current study: 8 participants 216 = 3456, that is, 1728 identity categorization trials with adult faces plus 1728 trials with child faces cf. Song et al. 15 participants 80 = 1200 identity categorization trials with child faces only.	2018-04-03T04:36:22.035Z	2018-03-20T00:00:00.000	{ "year": 2018, "sha1": "db4e5aeb6d64e0e25f642a430b92e173fa21b29d", "oa_license": "implied-oa", "oa_url": "https://europepmc.org/articles/pmc5964451?pdf=render", "oa_status": "GREEN", "pdf_src": "PubMedCentral", "pdf_hash": "3a8f2439d2f60e4e539f647472c9f906757eff9c", "s2fieldsofstudy": [ "Psychology" ], "extfieldsofstudy": [ "Psychology", "Medicine" ] }
117919884	pes2o/s2orc	v3-fos-license	Coherence properties of modeless lasers Most of classical light sources show a close similarity between their first and second order correlation functions (resp. g (1) and g (2) ) functions. We present here the original coherence properties of a peculiar type of laser named modeless laser or Frequency Shifted Feedback (FSF) laser where the g (1) and g (2) functions show a different behaviour. We calculate and evidence experimentally the first and second order correlation functions of modeless lasers, through measurements of the homodyne beat signal and interferometric autocorrelation of a dye FSF laser at the output of a Michelson interferometer. Whereas the degree of first-order coherence vanishes beyond the coherence length of the FSF source, the degree of second-order coherence exhibits periodic revivals far beyond the coherence length, with a period equal to the cavity roundtrip time. Our observations are in good agreement with the theoretical treatment of Yatsenko et al. (Opt. Comm. 282 (2009) 300) [1]. Introduction: field and intensity temporal correlations Coherence properties of light fields are of primary importance both from a theoretical and and experimental perspective in optics. The field temporal correlations -or degree of first order coherence-is linked to the phase properties of a light field whereas the intensity temporal correlation (or degree of second order coherence) is linked its amplitude properties. Classical imaging uses the degree of first order coherence but because of the strong influence of usual propagation media on the optical phase, resulting in the degradation of the optical wavefront in inhomogeneous media (atmosphere, water, biological tissues...), it could be highly interesting to perform imaging by using instead the degree of second order coherence. The famous Hanbury-Brown Twiss (HBT) stellar interferometer opened a new way of thinking coherence in optics and at the same time demonstrated new possibilities of imaging, that are currently investigated through the field of quantum imaging [2]. In the HBT experiment, the astronomers measure the angular diameter of bright stars by measuring the degree of second order coherence, and make the hypothesis that the degree of second order coherence is directly linked to the degree of first order coherence [3]. In fact most of usual light sources show this property: thermal sources, chaotic sources, single mode or multimode lasers. But here we show that this is not always the case and we study an original laser source, called modeless laser of FSF laser (for frequency shifted feedback), which shows decorrelated g (1) and g (2) functions. The g (1) function of modeless lasers is limited to the domain of coherence while the g (2) exhibits periodic revivals, far beyond the (first order) coherence time of the laser. We therefore demonstrate that modeless laser shows time delays for which the fields are mutually incoherent but the amplitudes are coherent. Definitions We restrict our study to classical light fields described by continuous functions of time, instead of quantum operators. Recall that the degree of first order coherence (or field temporal correlation function) is defined by [4]: (1) where τ is the time delay, E(t) is the complex electrical light field and the brackets denote ensemble averages. We implicitely assume a stationnary process, that is g (1) is independent from t. Defining the instantaneous intensity as I(t)=E(t)E * (t), we define similarly the degree of second order coherence (or intensity temporal correlation function) by: (2) 22 Coherence properties of modeless lasers H. Guillet de Chatellus Properties We consider a light field in the form where E 0 (t) and ϕ(t) correspond respectively to amplitude and phase fluctuations. It can easily been shown that g (1) depends on both phase and amplitude whereas g (2) depends only on amplitude fluctuations. It is also instructive to calculate the Fourier transform of g (1) and g (2) . In the case of g (1) , The Fourier transform of g (1) is equal to the optical spectrum, provided the Wiener-Khinchine theorem can be applied, which is the case for stationary processes. In the case of g (2) : Provided the Wiener-Khinchine theorem can be applied, the Fourier transform of g (2) is equal to the power spectrum of the time fluctuations of the photocurrent, called radio frequency spectrum. Therefore both g (1) and g (2) functions can be measured indirectly through their Fourier conjugate. 2.Coherence properties of usual classical light sources In this paragraph we review the first and second order coherence properties of classical usual light sources, to illustrate the fact that in most of cases, both the degrees of first and second order coherence show a large degree of similarity. Single emitter We consider here the case of a single light emitter, whose emission process is randomly dephased by collision. This is a case of a single atom emitter in a gas discharge lamp. We also assume that the lifetime duration of the excited level is long compared to the optical period to neglect the variation of the amplitude of the emitted wave. The electric field can be written as: Coherence properties of modeless lasers H. Guillet de Chatellus In that case the g (1) function is localized around zero with a width (coherence time) linked to the average time between successive collisions. The g (2) function is equal to 1 for all t. In the case of a single emitter, the g (1) and g (2) functions are not similar. Top left: electric field sequence emitted by a single atom submitted to dephasing collisions. Bottom left: theoretical g (1) and g (2) functions. Right: optical (top) and RF spectrum (bottom). N emitters We now turn to the case of N independent emitters. The electric field can be expressed as: The g (2) function can be calculated and after averaging out the different interference terms and one gets: Figure 2. Left: theoretical g (1) and g (2) functions of multiple independent emitters. Right: optical (top) and RF spectrum (bottom). Chaotic light sources We now turn to the case of a continuous distribution of emitters. Then the electric field can be described as a random process (chaotic). If one assumes gaussian statistics for E(t) the following factorization rule holds: Replacing this expression into the deginition of g (2) leads to the general relation: The g (2) function is therefore closely related to the g (1) function for a large variety of light sources: incandescent lamp, gas discharge lamp, thermal cavity, spontaneous emission, amplified spontaneous emission... This relation is at the basis of the HBT stellar interferometer for which the coherence length of intensity fluctuations has been identified to the coherence length of the electric field to deduce the diameter of the star. For a null time delay, the g (2) is equal to two. In term of photon statistics this relation is the signature of photon bunching. In terms of intensity fluctuations, the variance of the process obeys the relation: Figure 3. g (1) and g (2) functions of chaotic light sources. Single mode laser The case of a single mode laser is particularly simple. The amplitude of the electric field is stabilized by the gain. The electric field can be written as 0 0 () it Et Ee ω = and the related g (1) and g (2) functions are equal to 1, for any value of the time delay. The corresponding optical and RF spectrum are Dirac functions. (1) and g (2) functions of a single mode laser. Right: optical (top) and RF spectrum (bottom). Multimode laser We treat the case of a multimode laser with no specific phase relationship between the modes. Phases are assumed random and we neglect the competition arising in the laser gain medium and the related dynamic effects. We assume and electric field in the form: where ω 0 is the average angular frequency, ζ(n) is the amplitude of the n th mode, δω is the laser free spectral range and ϕ n is the phase of the n th mode. In that case, the g (1) and g (2) functions show periodic revivals at integer multiples of the cavity roundtrip time (τ r = 2π/δω). The temporal width of the peaks is the coherence time of the laser, whereas the amplitude of the peaks decay as a function of the coherence time of the phase ϕ n . The g (2) function at zero delay is equal to 2 (photon bunching). In the case where the phases of the modes have a definite relationship as in the case of mode-locked lasers, the output light field can be pulsed. In that case, the g (1) function remains the same but the g (2) function changes: the height of the peaks of the g (2) function increase and can reach very large values whereas the background between two peaks falls down to zero. 3.Theoretical coherence properties of modeless laser It appears from the review of the g (1) and g (2) functions of usual classical sources that there is a strong similarity between the g (1) and the g (2) function of a classical light source. But this is not always the case: in the following we focus on the coherence properties of a particular laser source called modeless laser and we show both theoretically and experimentally that the g (1) and g (2) functions are decorrelated. A modeless (or FSF for Frequency Shifted Feedback laser) is a laser cavity in which a frequency shifter has been inserted [5]. This shifter is most of the time an acousto-optics (AO) modulator where an acoustic grating is created in a crystal and diffracts the photons while increasing their frequency by the frequency of the acoustic modulation. The laser cavity is closed on the +1 order of the diffraction grating. Therefore a photon experiences a frequency shift Δ per roundtrip equal to twice the AO frequency. To express the output electric field we consider the model of the passive cavity: we neglect the spectral dependence of the gain. Therefore the cavity can be described as a regenerative cavity, seeded by the field resulting from the spontaneous emission of the gain medium and shifted in frequency at each roundtrip [1]. The electric field at the output of the laser can be defined by the recurrence relation: AO seeding output where R is the diffraction efficiency of the AOM (R is close to 1). In first approximation, () ( ) r It It τ − which shows the periodic nature of the light intensity of the FSF laser. This periodicity is therefore present in the g (2) function [6]. This property is not specific to modeless PoS(QQ09)008 Coherence properties of modeless lasers H. Guillet de Chatellus lasers and is also shared by multimode lasers. On the contrary, in the case of a modeless laser the electric field can be approximated by: Et Et e τ − Δ − which shows that the electric field of a modeless laser is intrinsically chirped. g (1) and g (2) functions By using the correlation function of the ξ field, one can calculate analytically the g (1) and g (2) functions of the modeless laser and one obtains respectively [6]: . Theoretical g (1) and g (2) functions of a modeless laser PoS(QQ09)008 Coherence properties of modeless lasers H. Guillet de Chatellus Therefore the modeless laser exhibits decorrelated g (1) and g (2) functions. The g (1) function is localized within the coherence time of the laser while the g (2) function exhibits periodic revivals, with a period equal to the cavity roundtrip time. 4.Experimental measurement of the degree of first and second order coherence of modeless lasers We turn to the experimental determination of the g (1) and g (2) functions of a dye modeless laser operating at 589 nm. The spectral width is equal to 85 GHz. The g (1) function is measured in a conventional manner by scanning an interferometer. Concerning the g (2) function, because of the short coherence time of our modeless laser (about 10 ps) we chose to perform second harmonic generation (SHG) at the output of the interferometer to measure the autocorrelation trace of the modeless laser, in the same way as the measurement of ultrashort pulses. We used a long interferometer where the path difference between both arms can exceed one cavity length, enabling to reach delays larger than the cavity roundtrip time τ r [6]. Figure 10. Experimental setup for the measurement of the g (1) and g (2) functions of the modeless laser. The g (2) is deduced from the collinear autocorrelation function measured by the frequency doubling in the nonlinear crystal at the output of the interferometer. The theoretical autocorrelation trace can be calculated using the expression of the modeless laser field. PoS(QQ09)008 Coherence properties of modeless lasers H. Guillet de Chatellus The experimental results show that the g (1) function is localized within the coherence length and vanishes elsewhere. On the contrary the autocorrelation trace exhibits a peak when the time delay is equal to the cavity roundtrip time. The measured ratio of the height of the peak to the background is close to 3:2, which is in good agreement with the theoretical prediction from the passive cavity model. PoS(QQ09)008 Coherence properties of modeless lasers H. Guillet de Chatellus 12 In conclusion, we have shown that contrary to most of usual light sources, the g (1) and g (2) functions of a modeless (or FSF) laser show a different bahaviour. In other words, the modeless laser exhibits intensity correlation where the field is no more (first order) coherent Possible applications concern quantum imaging [2]. This original behaviour is to be related to the very specific radio-frequency properties of modeless lasers [1,6].	2019-04-17T15:35:54.182Z	2009-12-01T00:00:00.000	{ "year": 2010, "sha1": "48ec1c5d59681cdc0ec857c25533334fc9adf5c7", "oa_license": "CCBYNCSA", "oa_url": "https://pos.sissa.it/101/008/pdf", "oa_status": "HYBRID", "pdf_src": "Adhoc", "pdf_hash": "2ba31c2497d4e6f8583c9587760d75125139c79d", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Mathematics", "Physics" ] }
220612902	pes2o/s2orc	v3-fos-license	Spectrum of genetic variants in moderate to severe sporadic hearing loss in Pakistan Hearing loss affects 380 million people worldwide due to environmental or genetic causes. Determining the cause of deafness in individuals without previous family history of hearing loss is challenging and has been relatively unexplored in Pakistan. We investigated the spectrum of genetic variants in hearing loss in a cohort of singleton affected individuals born to consanguineous parents. Twenty-one individuals with moderate to severe hearing loss were recruited. We performed whole-exome sequencing on DNA samples from the participants, which identified seventeen variants in ten known deafness genes and one novel candidate gene. All identified variants were homozygous except for two. Eleven of the variants were novel, including one multi-exonic homozygous deletion in OTOA. A missense variant in ESRRB was implicated for recessively inherited moderate to severe hearing loss. Two individuals were heterozygous for variants in MYO7A and CHD7, respectively, consistent with de novo variants or dominant inheritance with incomplete penetrance as the reason for their hearing loss. Our results indicate that similar to familial cases of deafness, variants in a large number of genes are responsible for moderate to severe hearing loss in sporadic individuals born to consanguineous couples. www.nature.com/scientificreports/ identified from various schools for deaf and special education schools in Punjab. The selected participants were born to consanguineous parents. The syndromic associations with hearing loss in this cohort were excluded by observation of the clinical features in individuals and detailed questioning regarding phenotypes. Written informed consents were obtained from participants or their legal guardians. Audiometry was performed to measure average hearing thresholds for all the participants at frequencies of 0.5, 1, 2, 4 and 8 kHz under ambient noise conditions. Romberg and tandem gait tests were completed to identify vestibular defects. Whole-exome sequencing and variants filtering. Whole blood up to 10 ml was drawn from the participants and the DNA was extracted using a standard method involving sucrose lysis and salting out. Wholeexome sequencing was performed at the Baylor-Hopkins Center for Mendelian Genomics (BHCMG). Exome capture was performed using the Agilent SureSelect Human All ExonV5 kit using a low input library preparation protocol 13 . Libraries were sequenced on the Illumina HiSeq2500 platform to generate 125 bp paired end runs. Reads were aligned with BWA mem 0.7.8 to the 1,000 genomes phase 2 (GRCh37) human genome reference sequence. Variant calling was performed using GATK 3.3-0 joint calling with HaplotypeCaller. The data after final output was analyzed independently at BHCMG, USA and SBS, University of the Punjab, Lahore. The program wANNOVAR (https ://wanno var.usc.edu/) was used for annotating the variant call files (VCF). Using either the PhenoDB Variant Analysis Tool 14 or manually, the output data from wANNOVAR was filtered against the population frequencies in the 1,000 Genomes database, genome Aggregation Database (gnomAD) and the Exome Aggregation Consortium (ExAC) database. Variants were retained for further evaluation if they had an allele frequency of less than 0.01 in these public databases. Homozygous, hemizygous and compound heterozygous exonic and splice site variants were examined. Large deletions and copy number variations (CNV) were detected using ExomeDepth 15 using read depth data from exome sequencing experiments. For these analyses, each test exome was compared to a matched, aggregate reference set of samples. CNV calls were annotated using AnnotSV 16 . Candidate CNVs were prioritized by minor allele frequency, exon number, Bayes factor (BF) and the ratio of observed/expected number of reads. The wANNOVAR files also included predicted pathogenic scores for these variants from Polyphen 2, Mutation Taster and SIFT along with the pathogenicity CADD scores indicating the probable impact of variation on the function of the encoded protein. In addition, REVEL pathogenicity scores for the variants were accessed (https :// sites .googl e.com/site/revel genom ics/). The conservation of selected amino acid residues affected by variants was checked across vertebrate species. For this purpose, multiple alignments were carried out on the protein sequences obtained from UniProt (https ://www.unipr ot.org/) using ClustalO (https ://www.ebi.ac.uk/Tool/msa/clust alo). Results Subjects and audiological phenotype. Twenty-one individuals including sixteen males and five females with ages ranging from 5 to 23 years participated in the study. The pure tone averages (PTA) for better hearing ears ranged between 65 to 88 dB HL (Fig. 1). Romberg and tandem gait tests were negative, which indicated normal vestibular function. All the participants exhibited no other phenotype except for hearing loss at the time of recruitment. Genetic findings after whole-exome sequencing. Whole-exome sequencing identified candidate causative variants in eleven genes in seventeen individuals (Table 1). Of the seventeen identified variants, fifteen were classified as pathogenic based on the predictions from in silico tools and pathogenicity scores from REVEL and CADD. According to the guidelines by ACMG 17 twelve were pathogenic, three were likely pathogenic and two variants were classified as variants of uncertain significance. Most of the variants except for one variant in OTOF (OMIM 603681) were unique to only one individual. All variants observed during this study have been deposited in ClinVar (SCV000924172. Pathogenic homozygous variants in genes known to cause nonsyndromic deafness. Variants in OTOF affected three individuals with deafness in this cohort. Two novel frameshift variants were identified for three unrelated participants. The variant c.4990_4991del (p.Tyr1497TyrfsTer10) was present in two individuals HLRBS13 and HLRBS14. Another novel frameshift variant c.2443delC (p.Gln815GlnfsTer1) was identified in individual SPK6. Variants in GJB2, SLC26A4 and OTOA (OMIM 607038) affected more than one individual. Among these, one missense variant c.158G > T (p.Cys53Phe) in GJB2 and one variant c.3188C > G (p.Pro1063Arg) in OTOA were identified for the first time. A large homozygous deletion including OTOA was identified in individual SPK7. Homozygous missense variants in RDX (OMIM 179410), CABP2 (OMIM 607314), and ESRRB (OMIM 602167) were found to contribute to hearing loss in three individuals. All these variants had high pathogenicity scores and the affected amino acids were conserved among different vertebrate species (Fig. 1B). www.nature.com/scientificreports/ variant, it is possible that p.Leu18Val is also a nonsyndromic deafness allele. However, some individuals with missense variants only manifest renal abnormalities as adults 19 . It is therefore possible that renal abnormalities could be manifested in future by the child with the p. (OMIM 613483) was identified in individual HLMS7 who had moderate to severe hearing loss. This variant was predicted to be disease causing by various online prediction tools but was of uncertain significance according to ACMG guidelines. The variant had a high CADD score of 22.9 and a relatively low REVEL score of 0.38. The variant was rare as it had a low allele frequency of 0.00008 and 0.0001 in ExAC and gnomAD respectively, with no homozygous individuals in the control population. It had a GERP++ score of 3.18. Discussion Consanguineous families have served as a rich resource for the identification of genetic causes of recessively inherited disorders. In Pakistan 40-60% marriages are among first cousins 20,21 which increases the risk of prevalence of recessive disorders, including hearing loss. According to the World Health Organization, the Pakistani population has a high prevalence of recessive disorders (2.4%) as compared to the incidence worldwide (1.7%). In this study, we explored the genetics of moderate to severe hearing loss for the first time in Pakistan in single individuals born to unaffected parents who were cousins. It was suspected that variants in few genes like STRC Table 1. Details of genes and variants obtained from the analysis of whole-exome sequencing. N/A Not applicable, D disease causing/deleterious/damaging, P probably damaging, N neutral, T tolerated, B benign, VUS variant of uncertain significance, REVEL rare exome variant ensemble learner, SIFT sorting intolerant from tolerant, MT mutation taster, PMID PubMed identifier, ExAC Exome Aggregation Consortium, gnomAD genome aggregation database. www.nature.com/scientificreports/ , GJB2, SLC26A4, OTOG or TECTA may explain the hearing loss for the majority of individuals in our cohort as is the case in many other world populations for the individuals with moderate to severe deafness. However, the identification of variants in multiple genes associated with hearing loss in our cohort of sporadic cases suggests a similar genetic heterogeneity in sporadic and familial cases in Pakistan. The combined contribution of genes involved in profound deafness is 52% to moderate to severe hearing loss in this cohort. The phenotypic variability due to variants in the same genes implicate the involvement of extrinsic factors or modifiers affecting the severity of hearing loss. The variants in OTOF were more frequent in our cohort as compared to the published data for different ethnicities or populations. Four other reports from Korea, Japan and China on the genetic predisposition of hearing loss in sporadic individuals included more than 60 participants with moderate to severe or profound deafness, in which they demonstrated SLC26A4 as the major contributor to hearing loss 6,22,23 . Case ID The GJB2 related deafness accounts for 10% cases in our cohort, which is similar to the reported incidence of GJB2 variants (9.5%) obtained from screening of large consanguineous families 11 . However, a recent research from Pakistan on 40 individuals with profound deafness from Bannu and Kohat districts indicated that GJB2 variants caused deafness in 37% of non-familial cases 9 . The small sample size, difference of ethnic background and less severe hearing phenotype may explain this lower contribution of GJB2 variants in the present study. A variant c.733G > C (p.Asp245His) in ESRRB was identified for moderate to severe hearing loss in one participant of this study. ESRRB is an estrogen related receptor beta gene which is known to cause hearing loss at DFNB35 (OMIM 608565) locus. The encoded protein consists of two domains; DNA binding domain (DBD) and ligand binding domain (LBD). Seven of the previously identified variants affect the ligand binding domain of the protein. The variant identified in this study also affects the ligand binding domain and the amino acid at this position is conserved among vertebrate orthologues (Fig. 1B). However, instead of profound deafness, the missense variant was observed to cause a moderate to severe phenotype in the affected individual in this study. It suggests that the severity of hearing loss caused by ESRRB can be modified by certain genetic or environmental factors. A missense variant c.187G > C (p.Gly63Arg) in BHLHE22 was potentially implicated for moderate to severe hearing loss. BHLHE22 has a single coding exon which encodes a class E basic helix loop helix protein 22 (BHLHE22). It is a small protein of 381 amino acids which serves as a sequence specific DNA binding transcription factor and mediates cell differentiation and proliferation. Mutant murine models have demonstrated that BHLHE22 is necessary for retinogenesis 24 and development of dorsal cochlear nuclei 25,26 . BHLHE22 has the highest expression in retina 26 however, it is also expressed in cochlear hair cells, supporting cells and utricle in the inner ear (umgear.org, https ://shiel d.hms.harva rd.edu/). The variant c.187G > C in BHLHE22 had a relatively low conservation score (3.18; only conserved among mammals and some reptiles) and high pathogenicity scores (CADD, 22.9). These scores may be explained by a previous study on transcriptional repressors, which suggested that nonconserved regions are vital for the DNA binding function of the proteins. They may also provide a drift during evolution for the correct folding and thus secondary structure of the respective protein 27 . Therefore, the identification of BHLHE22 variants in additional affected individuals or mice models will be useful to understand the role of this gene, if any, in hearing loss. Majority of variants in MYO7A primarily cause autosomal recessive nonsyndromic hearing loss (DFNB2) (OMIM 600060) and Usher syndrome 1B (USH1B) (OMIM 27690) 28,29 . A heterozygous variant identified in individual HLMS32 suggests that hearing loss is probably nonsyndromic dominant as observed for DFNA11 (OMIM 601317) instead of USH1B or DFNB2 which are caused by biallelic variants of MYO7A. However, we cannot exclude the possibilities that either the individual is a carrier for DFNB2/USH1B variant or the variant may be benign, in spite of its prediction to be damaging. The diagnostic rate for sporadic cases in this research was relatively high (80%) as compared to other studies. For instance, in a cohort of 63 simplex cases from China the successful diagnostic rate was of 12.7% 8 . Few other studies on sporadic cases from Korea, China and Japan have reported the pathogenic variant detection rates of 20% (from 92 cases) 30 , 23.1% (from 13 cases) 6 , 32% (from 34 cases) 31 and 45.4% (from 11 cases) 5 using wholeexome sequencing. The mutation detection rate in these studies is lower as they screened the individuals for common variants of GJB2 and SLC26A4 prior to whole-exome sequencing. However, even after excluding the GJB2 cases, the identification rate for the present study still remains as high as 71%. This may perhaps be due to the fact that we specifically studied hearing loss in individuals born in consanguineous unions. This increased the possibility that the disorder was recessively inherited. Variants of uncertain significance were identified in multiple participants (Table 1) while no potential pathogenic variant was identified for four individuals after whole-exome sequencing. For the latter, some pathogenic variants may have been overlooked due to the stringent criteria to classify a variant as pathogenic. Secondly, a few pathogenic variants may have been missed as they could be present in non-coding exons, introns or regulatory regions of the genes. Our study comprehensively evaluated the genetic cause of moderate to severe hearing loss in a cohort of sporadic individuals. Results show that a similar diversity of gene variants is responsible for sporadic deafness as seen for familial hearing loss. Therefore, such cohorts can serve as a rich source for the determination of genetic and molecular basis of hereditary deafness. These results also suggest that targeted sequencing of few common deafness genes prioritized according to the ethnicity, followed by whole-exome sequencing will be a simple and cost effective approach for the genetic diagnosis and management of isolated hearing loss.	2020-07-18T15:41:48.914Z	2020-07-17T00:00:00.000	{ "year": 2020, "sha1": "2bd430a7bda08d4709aa064686ebae97adb6df5f", "oa_license": "CCBY", "oa_url": "https://www.nature.com/articles/s41598-020-68779-5.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "2bd430a7bda08d4709aa064686ebae97adb6df5f", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
79543400	pes2o/s2orc	v3-fos-license	Reversal of pulmonary arterial hypertension in POEMS syndrome with thalidomide: a case report Abstract Introduction Pulmonary arterial hypertension (PAH) is a progressive and life-threatening disease characterized by the remodelling of distal pulmonary arteries in the absence of other cardiopulmonary disease, usually leading to right ventricular failure. Given the current European Society of Cardiology and the European Respiratory Society guidelines for the diagnosis and treatment of pulmonary hypertension (PH), most of the patients with severe PAH usually require to have a lifelong combination therapy that includes prostacyclin, phosphodiesterase-5 inhibitor, and endothelin receptor antagonist. However, the reversibility of PAH has been reported through the treatment of the underlying diseases or comorbidities. Case presentation We present a case of a 45-year-old woman with a chief complaint of dyspnoea, eventually diagnosed with severe PAH in the setting of POEMS (polyneuropathy, organomegaly, endocrinopathy, M-protein, and skin changes) syndrome that was successfully treated with thalidomide and dexamethasone. Discussion Our case suggests that it would be important to consider associated syndromes when a diagnosis of PH is made, because treatment of the underlying condition may lead to improvement in PAH. Introduction Despite enormous efforts in research and the development of available therapies over the last two decades, pulmonary arterial hypertension (PAH) remains a progressive and relatively incurable disease. 1 Pulmonary arterial hypertension is a disease characterized by the remodelling of distal pulmonary arteries in the absence of other cardiopulmonary disease, usually leading to right heart failure and subsequent death without appropriate intervention. 2,3 Given the current European Society of Cardiology/European Respiratory Society guidelines for the diagnosis and treatment of pulmonary hypertension (PH), 4 most of the patients with severe PAH usually require to have a lifelong combination therapy that includes Learning points • Early diagnosis and intervention with thalidomide and dexamethasone might contribute to a decrease of pulmonary vascular resistance, even in patients with severe pulmonary arterial hypertension (PAH) concomitant with POEMS (polyneuropathy, organomegaly, endocrinopathy, M-protein, and skin changes) syndrome. • It would be important to consider associated syndromes when a diagnosis of pulmonary hypertension is made, because treatment of the underlying condition may lead to improvement in PAH. . prostacyclin, phosphodiesterase-5 inhibitor, and endothelin receptor antagonist (ERA). However, the reversibility of PH has been reported through the treatment of the underlying diseases or comorbidities. [5][6][7] Here, we present a case of severe PAH concomitant with a syndrome of POEMS (polyneuropathy, organomegaly, endocrinopathy, M-protein, and skin changes) syndrome that was successfully treated with thalidomide and dexamethasone. Case presentation A 45-year-old woman was referred to our institution with a 3-month history of shortness of breath on exertion. The patient did not have significant past medical history. On physical exam, she was noted to have a palpable lymph node in the neck and axilla; skin pigmentation around the neck, chest, and back; significant peripheral oedema of the lower limbs; and clubbed fingers on both hands were noted ( Figure 1). Electrocardiogram showed right axis deviation, P pulmonale in lead II, V1R, and strain pattern in lead V1-3 ( Figure 2A). Chest X-ray revealed significant cardiomegaly ( Figure 2B). Pulmonary function test (PFT) was normal with the exception of impaired diffusing capacity of the lung for carbon monoxide (DLCO) at 30.4% (normal reference value: DLCO 80-120%) of predicted value. Transthoracic echocardiogram (TTE) revealed severe PH with an estimated pulmonary artery systolic pressure of 80-85 mmHg, significant right atrium (RA) and right ventricular (RV) enlargement, and RV impairment with a tricuspid annular plane systolic excursion (TAPSE) of 10.2 mm (normal reference value: TAPSE > 16 mm). Cardiac magnetic resonance (CMR) revealed mildly impaired RV function with right ventricular ejection fraction (RVEF) of 47% (normal reference value: RVEF 50-65%) (Supplementary material online). Moderate pericardial effusion was present ( Figure 3A, B). Ventilation perfusion scan indicated a low probability of pulmonary thromboembolism. Contrast-enhanced computed tomography (CT) did reveal significant dilatation of RA, RV, and pulmonary artery ( Figure 4D) with no evidence of pulmonary thromboembolism. Bilateral pleural effusion, pericardial effusion, and ascites were present. In addition, hepatosplenomegaly and lymphadenopathy in bilateral neck, axilla, mediastinum, and para-aorta were present ( Figure 4A-C). Right heart catheterization (RHC) confirmed the presence of PAH ( Table 1). To determine the aetiology of the patient's PAH, further evaluation was performed. Blood test revealed mild thrombocytosis with a platelet count of 388 Â 10 3 cells/lL (normal reference value: 150-350 Â 10 3 cells/lL). Serum HIV-1 titre and the antibody titre related to autoimmune disease were completely negative. Low free T3 and free T4 levels with high thyroid-stimulating hormone suggested hypothyroidism. Total serum immunoglobulin was normal with normal protein electrophoresis and a negative test of Bence Jones protein. There was elevation of j and k free light chain levels (55.0 mg/L and 33.2 mg/L, respectively) (normal reference value: 3.3-19.4 mg/L, 5.7-26.3 mg/L) with an increased light chain ratio (1.66) (normal reference value: 0.26-1.65). Notably, serum vascular endothelial cell growth factor (VEGF) level was significantly elevated. Computed tomography revealed multiple patchy osteosclerotic lesions in pelvic bone, vertebral body, costal bone, and sternal bone ( Figure 4E Bilateral papilledema was present. In addition, although there was no clinical history suggestive of neuropathy, nerve conduction studies confirmed peripheral polyneuropathy involving the reduction in amplitude and velocity of sensory potentials and clear slowing of motor velocities with a decrease in amplitude in excess of 50%. All of these findings supported a diagnosis of POEMS syndrome according to the diagnostic criteria. 8 As symptoms and haemodynamics have not been improved with the initiation of tadalafil for PAH, our patient was started on immunomodulatory agent using thalidomide and dexamethasone. A year later, as a consequence, her symptoms of dyspnoea, pericardial effusion, pleural effusion, and ascites have completely disappeared with significant recovery of RV function ( Figures 2C and 3C Discussion POEMS (polyneuropathy, organomegaly, endocrinopathy, M-protein, and skin changes) syndrome is a rare paraneoplastic disorder caused by plasma cell dyscrasia. Additionally, PAH is thought to be an uncommon feature of POEMS syndrome. Although there is weakness and/or numbness of the limb due to peripheral neuropathy, peripheral oedema and skin change were reported as the most common initial symptoms. 9 This is, to our knowledge, the first case reported in which PAH was the initial presentation in a patient with POEMS syndrome who was treated successfully with thalidomide and dexamethasone. Contrary to the current diagnostic criteria of POEMS syndrome, monoclonal plasma cell dyscrasia was not detected in the present case. However, in spite of the absence of this type of dyscrasia, we consider this case to be consistent with POEMS syndrome for the following reasons. The first reason is the previous finding that 25% of cases of POEMS syndrome lack a monoclonal plasma cell disorder. 10 The second reason is the presence of typical clinical findings and a clinical course specific for a favourable response to immunomodulatory agents. Based on these factors, we assume that the present case might be a very early stage of POEMS syndrome with a characteristic finding of PAH. The suggested prevalence of associated PH in POEMS syndrome is 27-48%, 5-7 although these figures were obtained using TTE to estimate pulmonary arterial systolic pressure (PASP), implying the prevalence of the PAH might be far less than these reports. A case similar to ours has been previously published, which mentions a PAH patient in the setting of POEMS syndrome, successfully treated with thalidomide and methylprednisolone. 11 However, contrary to our case, Reversal of pulmonary arterial hypertension PAH in this report was diagnosed by TTE, not with RHC which is the standard diagnostic modality, suggesting lack of definitive evidence of PAH. To date, despite the conflicting findings seen with anti-VEGF therapy, VEGF is the cytokine that correlates best with disease severity and treatment response of PH. 12 In our case, the serum level of VEGF was elevated at 4080 pg/mL (normal reference value; plasma VEGF 31-86 pg/mL, 13 serum VEGF < 1040 pg/mL 12 ) and dropped significantly to 240 pg/mL over the year with treatment, accompanied by the significant improvement of other symptoms. Vascular endothelial cell growth factor was postulated to mediate PH with its vascular permeability causing interstitial and perivascular oedema and subsequent hypoxaemia inducing endothelial VEGF, completing a vicious cycle, eventually leading to PAH. 14 However, VEGF level might not be the driving force of the disease based on the conflicting findings seen with anti-VEGF therapy. [15][16][17] With respect to available therapy, there are no standard treatments for POEMS syndrome. However, the patient was not considered as an ideal candidate for two recognized effective treatments. 18,19 Radiotherapy was rejected because of multiple bone lesions, and autologous stem-cell transplantation was rejected due to severe PAH. 7 More recently, thalidomide treatment has been shown to have decreased serum VEGF levels and improved clinical symptoms from randomized placebo-controlled trial. 20 Thalidomide has many effects that are potentially useful for the treatment of POEMS syndrome, including suppression of monoclonal plasma cell proliferation and modulation of upregulated cytokines 21 ; however, how it contributes to the treatment for PAH remains uncertain. Importantly, the effect on RV haemodynamics would be of profound significance in the recovery of RV failure. Before treatment, the patient was considered at high risk for PAH because of poor RV haemodynamics with relatively high right atrial pressure and low cardiac index (CI) in addition to poor RV function by TTE and CMR. After treatment with thalidomide and dexamethasone, reversal of PAH was obtained by RHC with a significant decrease of pulmonary vascular resistance (PVR) and mean pulmonary arterial pressure. The substantial improvement of haemodynamics presumably led to the subsequent recovery of RV size and function confirmed by CMR. 22 This recovery is most likely due to inhibition of pressure/volume overload secondary to vascular permeability by immunomodulatory therapy, which is consistent with the significant reduction of serum VEGF levels. Our report has a key limitation in its nature of single case study. Nevertheless, our case suggests that early diagnosis and intervention with thalidomide and dexamethasone have decreased PVR drastically in patients with severe PAH concomitant with POEMS syndrome and a subsequent improvement of symptoms. Accordingly, it would be important to consider associated syndromes when a diagnosis of Supplementary material Supplementary material is available at European Heart Journal -Case Reports online. RHC, right heart catheterization; RAP, right atrial pressure; PAP, pulmonary arterial pressure; PCWP, pulmonary capillary wedge pressure; PA saturation, pulmonary arterial saturation; PVR, pulmonary vascular resistance; CMR, cardiac magnetic resonance; RVEF, right ventricular ejection fraction; RVEDV, right ventricular end-diastolic volume; VEGF, vascular endothelial cell growth factor; BNP, brain natriuretic peptide; DLCO, diffusing capacity of the lungs for carbon monoxide. Reversal of pulmonary arterial hypertension	2019-03-17T13:12:34.218Z	2018-04-27T00:00:00.000	{ "year": 2018, "sha1": "f3bd0c88d7240431470758da9b3206bc4cb5005a", "oa_license": "CCBYNC", "oa_url": "https://academic.oup.com/ehjcr/article-pdf/2/2/yty051/28122441/yty051.pdf", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "f3bd0c88d7240431470758da9b3206bc4cb5005a", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
13782222	pes2o/s2orc	v3-fos-license	Determining the impact of 24/7 phone support on hospital readmissions after aortic valve replacement surgery (the AVRre study): study protocol for a randomised controlled trial Background Patients undergoing surgical aortic valve replacement (sAVR) have high rates of 30-day readmissions. They also report a low health-related quality of life (HRQOL) and elevated anxiety and depression. The aim of the AVRre study is to determine the efficacy and cost of a 24/7 phone-support intervention in reducing post-discharge readmissions after sAVR. The nature of the support is to help patients better understand and self-manage non-urgent symptoms at home. Methods/design AVRre is a prospective, randomised controlled study comprising 30 days of continuous phone-support intervention and then intermittent follow-up for the first 12 months. Phone call data from and to patients are evaluated qualitatively; thus, the study has a mixed-method design. Two hundred and eighty-six patients, aged >18 years, scheduled for a sAVR — singly or in combination with another procedure — are recruited from locations in southeast Norway. Patients are randomly assigned to the intervention group, who are purposively phone-called individually 2 and 9 days after discharge and offered on-demand 24/7 (around-the-clock) telephone support for 30 days post-discharge. The primary outcome variable is the number of 30-day hospital readmissions. Secondary outcomes are anxiety and depression symptoms, as measured by the Hospital Anxiety and Depression Scale, HRQOL and quality-adjusted life years, measured by the EuroQol (EQ-5D). Intervention and hospital readmission (diagnosis-related groups (DRGs)/length of stay) for the first year after initial discharge from hospital are used for a cost-utility analysis. Standard parametric and non-parametric tests are used for evaluations over time. Analysis of covariance is used to control for possible differences at baseline. Narratives from phone calls are transcribed verbatim and analysed using systematic text condensation. Discussion A complex ‘around-the-clock’ intervention within a university hospital-based setting could be an effective strategy to reduce the high readmission rates to hospital after sAVR. Furthermore, the AVRre 24/7 phone-support manual can be adapted to other high-risk surgery populations with high readmission rates. Trial registration ClinicalTrials.gov, NCT02522663. Registered on 11 August 2015. Electronic supplementary material The online version of this article (doi:10.1186/s13063-017-1971-y) contains supplementary material, which is available to authorized users. Background Severe aortic stenosis that demands surgical aortic valve replacement (sAVR) due to considerable morbidity and mortality is increasing in prevalence as the elderly population increases globally [1]. sAVR remains the definitive treatment for aortic stenosis (AS), and sAVR has an estimated annual incidence of 85,000 cases [1] in the USA and 1500 cases in Norway (unpublished data from Norwegian Heart Surgery). Irrespective of good immediate surgical outcomes, sAVR patients are characterised by high rates of 30-day readmissions to hospital after discharge. For example, the rates are 19.6% in a US population [2] and 26% in a Danish population [3], and from unpublished register data in Norway (Norwegian Patient Registry, AVR patients' readmission to hospital, 2011-2014, the Norwegian Directorate of Health 2016), it is estimated to be 22.4% in Norway. Reasons for 30-day readmissions after sAVR are available in two studies. In an American study, heart failure, cardiac rhythm disorders, stroke or transient ischaemic attack, pneumonia, pneumothorax/pleural effusion and gastrointestinal bleeding were reported [4]. In a Danish study, atrial fibrillation, pericardial effusion, congestive heart failure and pneumonia were the most dominant reasons for 30-day readmissions; these conditions occurred acutely in 25% of cases [3]. One in five patients in a study after major surgery was readmitted to a nonindex hospital. The use of an index hospital with specialised competence, versus non-index re-hospitalisation, resulted in significantly lower in-hospital mortality [5]. Readmission to hospital in Norway is defined as an unplanned, emergency admittance 8 hours to 30 days after discharge from hospital, accompanied by at least one overnight stay with the readmittance [6]. The majority of patients (96%) discharged approximately 1 week after complex sAVR return home intending to be responsible for their own physical and mental health and for arranging follow-up by their general practitioner (GP) when needed. However, following discharge, patients/inhabitants and partners experience insecurity and the psychological and physical burdens associated with potential readmissions. Moreover, the estimated cost of readmissions is 2 billion Norwegian kroner (NOK) each year, with an estimated readmission rate of approximately 20% [6]. The clinical experience of specialists and municipal healthcare services reveals that standard care at discharge does not typically include patient education. Two systematic reviews and meta-analyses of randomised controlled trials (RCTs) that sought to reduce 30-day hospital readmissions for different diseases concluded that no single intervention (e.g. education, telephone follow-up) was associated with reduced risk for 30-day re-hospitalisation [7,8]. For example, Melton et al. (2012) suggested a two-time telephone follow-up after discharge during office time [9]. More complex, highmethodological quality interventions, ones in which patients are educated and receive support for self-care, are recommended for preventing hospital readmission and increasing health-related quality of life (HRQOL) status, which otherwise is poorly self-reported [3]. Research on readmission after heart surgery highlights a great need for interventions to be implemented during the first 30 days after discharge to ensure that patients receive quality healthcare and engage in safe practices [3,5,[10][11][12][13]. In a Norwegian home-based intervention the first month after cardiac surgery (n = 185), patients and relatives pointed to several negative factors, including lack of information at discharge, insecurity and lack of a 'connection' to the index hospital. This was especially true in the first month after surgery, if complications such as pleural effusion and arrhythmias appeared post-discharge [14]. Furthermore, the Norwegian patient experience surveys (2016) report that almost 50% of patients received incomplete information related to discharge preparation, especially regarding what symptoms to expect after discharge, and how and whom to contact if complications occur [15]. These experiences may contribute to feelings of anxiety in patients. Indeed, approximately 29-61% of all patients experience moderate to severe levels of anxiety and depression during the first month after cardiac surgery, with symptoms remaining elevated up to 6 months following surgery [16,17]. These factors deserve our attention, because anxiety and depression are predictors of morbidity and mortality after heart surgery [18][19][20][21]. Therefore, one can hypothesised that interventions that target patients' and relatives' need for information and follow-up during the first month after cardiac surgery and the provision of these interventions around the clock could avoid unnecessary hospital readmissions. Indeed, a 24/7 follow-up service by phone goes beyond the results of regular telephone follow-up during office time. No study has tested the effect of an around-theclock follow-up intervention, where the patients' needs and symptoms are the base for the intervention. Expert healthcare professionals will be able to assess worsening of symptoms on the phone before a critical stage, and patients can be advised to contact a GP. Telephone follow-up also allows for inclusion of patients who live a long distance from both index and non-index hospitals. This paper presents the detailed protocol for the AVRre study, in which we aim to determine the efficacy and cost utility of 30-day around-the-clock, 24/7 phone-support intervention after discharge for sAVR. The study's design and protocol are in accordance with the current Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines [22]. A SPIRIT checklist is available online for this manuscript (see Additional file 1). Results will be reported following the CONsolidated Standards Of Reporting Trials (CON-SORT) guidelines for non-pharmacological interventions [23,24]. Primary objective The primary objective of this study is to determine whether a 30-day, around-the-clock, 24/7 phone-support intervention reduces the number of hospital readmissions 30 days after discharge from hospital. The intervention begins immediately after initial discharge, and the outcomes of patients in the intervention are compared to a control group, which receives usual care. Secondary objectives The secondary objectives of the study are as follows: To determine whether an around-the-clock, 24/7 phone-support intervention implemented within 30 days after discharge reduces objectively measured symptoms of anxiety and depression compared to a control group in the first year after discharge from hospital To determine whether the around-the-clock, 24/7 phone-support intervention implemented within 30 days after discharge improves HRQOL and quality-adjusted life years (QALYs) compared to the control group in the first year after discharge from hospital To perform an economic evaluation specifically to (1) determine the cost utility of the intervention compared to usual care in the study population and (2) assess the cost of readmission to hospital and the cost of GP consultations during the first year after discharge for the intervention and the control groups Study design AVRre is a prospective, randomised controlled trial (RCT), comprising 30 days of intervention and 12 months of follow-up. The main study began in August 2015. As the intervention consists of phone calls from patients to hospital and vice versa, the design of the study includes an explorative, qualitative component. Thus, this study employs a mixed-method study design. Supporting material for the AVRre study is provided in Additional file 2. Study population, recruitment, randomisation and follow-up Patients eligible for study participation are 18 years or older and are referred for sAVR surgery for the first time at Oslo University Hospital, at either the Ullevål or Rikshospitalet locations, the largest hospitals in southeast Norway. Consecutively admitted sAVR patients are asked by project nurses to participate, and they are included if they meet the following criteria: (1) the surgery is an elective treatment with a single AVR (biological (b) or mechanical (m), an AVR (b or m) + aortocoronary bypass or an AVR (b or m) + supracoronary tube graft; (2) the patient can understand, speak and write the Norwegian language and (3) can be contacted by phone after discharge from hospital. Exclusion criteria are the following: (1) the patient has been admitted to an intensive care unit (ICU) for more than 24 hours; and/or (2) has complications related to surgery (e.g. surgery caused cerebral insult with significant impact on cognitive functions). One to three days before the planned sAVR, patients arrive at hospital for preoperative preparations. During this time, the project nurse informs the patients about the aim and process of the study. The patients are then given the informed consent form and the baseline questionnaires for review and are given time to consider participation in the study. Patients are contacted a second time before surgery to answer any questions about the study and to deliver further information about the study. After the patient has provided written informed consent, patient assignment to either usual care (control) or intervention is accomplished by a web-based randomisation system developed and administered by the Unit of Applied Clinical Research, Institute of Cancer Research and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway. This system has been approved by the Data Protection Officer at Oslo University Hospital as complying with human experimental subject protections. Randomisation (1:1 ratio) is performed consecutively with block randomisation and varying size of the blocks to make it impossible to predict to which group the patients are likely to be allocated. Randomisation is done without stratification to the two cardiothoracic sites of the study hospital (i.e. Ullevål or Rikshospitalet). Before standard discharge from the university hospital to the patient's local hospital on the fourth day post-sAVR, the project coordinator (SOD) informs the patient verbally and in writing (with a leaflet) to which group he/she has been allocated. For both the control and the intervention groups, the follow-up assessment takes place 1 month (T1), 3 months (T2), 6 months (T3) and 12 months (T4) after discharge from hospital. Follow-up consists of mailing by postal questionnaires with prepaid stamps for return post after completion (see the patient flow and data collection chart of Fig. 1). Usual care Preoperatively, all patients recommended by a thoracic surgeon to have aortic valve replacement surgery receive information on expected HRQOL improvement, longer life expectancy and possible complications of surgery. Currently in Norway, at discharge, there is no standard information or post-discharge telephone support in usual care, not from nurses, doctors, university hospitals or local hospitals. There is no blinding in the study. Intervention: two components A brief description of the development of the 24/7 phone-support manual and an example section of it is available online (Additional file 1). The intervention consists of two components. Component 1 Patients in the intervention group are purposely called on days 2 and 9 after discharge by the project coordinator to proactively assess the patient's present condition and to determine if the patient has questions or if problems have emerged. Details in the patient's medical history are compiled in advance and reviewed prior to the phone calls, and each structured call relays reminder information to the patient about the availability of the 24/7 phone-call service as part of the intervention. The project coordinator emphasises the importance of daily physical activity [3] and how it has a positive effect on rehabilitation, morbidity and mortality after sAVR. When the patient receives a Short Message Service (SMS) one day ahead of the phone call, he/she has the opportunity to respond if the scheduled day or time is inconvenient for them. Data from the phone calls will be collected in a written, standard format. Also, patients will be encouraged to relate their individual responses/experiences/narratives, for example, when they experience anxiety symptoms. Component 2 The intervention group is also offered 24/7 around-theclock telephone availability during the entire first month post-discharge. Volunteer, expert intensive care nurses on duty in a cardiac ICU have been trained by an interdisciplinary team to answer the calls from sAVR patients during the first month after discharge. One aspect that the ICU nurses make extremely clear is that this 30-day 24/7 phone-support provision is not a replacement for emergency calls to 113 (911 in some countries). For ethical reasons and because of hospital responsibility to the patients, we include some 'red' responses (i.e. acute or emergency) in the manual. However, we expect that the sAVR patients will call the intervention phone line mostly for non-urgent health information. If a patient's problems demand advanced expertise, the project nurse will consult the thoracic surgeon or cardiologist on call in hospital to ensure that accurate diagnostics are completed and suggestions for treatment are made. The project coordinator is always available for the ICU nurses to consult, and will take initiative to arrange regular follow-up meetings and interdisciplinary discussion of challenging phone calls. Variables, sources and measurement Patients will be longitudinally assessed five times during the course of the project: before sAVR and 1, 3, 6 and 12 months after discharge for sAVR. Additional file 1 includes a SPIRIT checklist for the schedule of enrolment, intervention and assessments as presented in Fig. 2. The written informed consent form for the AVRre study is included as Additional file 3. Primary outcome Readmission Data from the Norwegian Patient Registry (NPR), the Norwegian Directorate of Health and patients' medical records will be used to gather the numbers of readmissions within 30 days after sAVR discharge. Moreover, data on causes of readmissions (ICD-10 codes), time and location (index and non-index hospital) will be collected. Secondary outcomes Anxiety and depression symptoms are measured using the Hospital Anxiety and Depression Scale (HADS) [25,26], a Enrolment Allocation Close-out Fig. 2 SPIRIT schedule of enrolment, interventions and assessments for the AVRre study standardised, self-report instrument consisting of 14 items in two subscales. The 14 items include seven items for anxiety (HADS-A) assessment and seven for depression (HADS-D). Patients rate themselves on each item from 0 (not present) to 3 (maximum), yielding a total possible score of 21 for each subscale. The psychometric properties of the HADS are well documented in research conducted in many different countries; this includes valid use in heart patients [26]. HRQOL and QALY are assessed using the internationally recognised EQ-5D instrument [27]. EQ-5D is a standardised instrument comprising five dimensions of selfreported health status for clinical and economic appraisal. These dimensions are mobility, self-care, usual activities, pain/discomfort and anxiety/depression. The respondent rates himself on each dimension for the degree (no problem, some problem, extreme problem) that best describes his/her present health status. Economic evaluation Economic analyses are performed for two reasons. (1) The time it takes to proactively call the patients, as well as the time needed to answer the patient on the intervention phone and the time needed for calling back if consulting the physician at hospital, will be measured and valued. (2) Register data on the cost of readmission to hospital (diagnosis-related groups (DRGs)/length of stay) and the number of GP consultations during 30 days and the first year after discharge will be used for the cost-utility analysis and will be reported as an incremental cost-effectiveness ratio (ICER). Sensitivity analyses will be conducted to measure uncertainty in the estimates. In addition, data from the patients' medical records are gathered, e.g. comorbidities. Data management and statistical analysis The first and second author have the daily responsibility for overseeing patient safety, study design, database integrity and study conduct and have access to the final study dataset. No data will be entered before the intervention is finalised, to make sure that the baseline data will not influence the intervention. A random check of at least 20% of entered data will be performed to ensure data quality before starting the full data analysis. Data are presented as means ± standard deviations for continuous variables and percentages for nominal variables. The primary outcome variable is measured using the chi-square test to evaluate group differences. The secondary outcome variables are measured longitudinally to assess changes over time. Symptoms of anxiety and depression (HADS) will be analysed in continuous-form variables before being transformed to a cut-off score ≥8 for anxiety and depression respectively. We will apply a multilevel logistic model with the time nested within the patient, and Hosmer's step-down procedure [28] to establish a final model. Analyses will be conducted in R version 3.3.2 (2016-05-03, R Core Team, 2016) (https://www.r-project.org/). Mixed model analyses will be applied for repeated measurement of anxiety (yes/no) or depression (yes/no) using HADS and EQ-5D [29]. Analysis of covariance (ANCOVA) is used to test mean changes between groups, controlling for possible differences at baseline [30,31]. A paired sampled t test is used to analyse mean changes within groups. A statistic will be considered significant when the corresponding P value is <0.05. The Statistical Package for Social Sciences (SPSS), version 21 (released 2012, IBM Corp., Armonk, NY, USA) is used for statistical analysis. Missing data The amount of missing data in the study and the methods used to handle missing data in the analysis will be reported [32]. Complete registry data will be available on primary outcome readmission 30 days after discharge for sAVR. If a patient dies within 30 days, it will be counted as readmission. Out of a total sample of 286 sAVR patients, we estimate 0-2 deaths. These numbers will not influence the power of the study. Regarding secondary outcomes, the guidelines in the article of Little et al. [32] will be followed; hence, we will perform multiple imputation analyses in analyses where missing data are not handled properly otherwise. In addition, sensitivity analyses will be performed to assess the robustness of assumptions made. Narrative data analysis Data/narratives from patients' phone calls to hospital and project coordinator phone calls to patients All qualitative data are transcribed verbatim and analysed in several steps using systematic text condensation in accordance with the approach of Malterud [33]. Experts in qualitative analysis in the research group responsible for the AVRre study will re-read the narratives independently before the subsequent data reduction into meaning units, condensed meaning units, subthemes and themes guided by the study's aim. Mixed methods Qualitative data as narratives from the patients are intended to complement and enrich the quantitative data from study measures. Using narratives from patients' phone calls will focus on the spontaneous needs and symptoms from the patients' perspective, thus avoiding recall bias that may occur during interviews at a later time. The two approaches are planned to be used in tandem to answer the research questions in this study [34]. One challenge that needs to be figured out is how to interpret conflicting results. Sample size and power calculation In 2013, a total of 503 patients had aortic valve replacement surgery at Oslo University Hospital. To estimate the sample size required to make confident conclusions about the primary outcomethe number of readmissions 30 days after discharge from hospitalwe used published data on readmissions in Norway for patients >65 years old. Seventeen percent of the patients are readmitted to hospital within 30 days from discharge [6]. A sample size of 286 patients with 143 patients in each group will achieve at least 80% power to detect an expected difference of 15% in the control group and 5% in the intervention group at the 5% significance level using the chi-square test. Ethical considerations: ethics and disseminations The study is conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the Regional Committees for Medical and Health Research Ethics (approval 2013/2031-3). All patients receive both verbal and written information about the aim of the study and are informed that they are free to withdraw from the study at any time. Patients sign an informed consent document prior to inclusion. The codebook with study numbers and person-sensitive information and data from phone calls is kept in a locked, firewall-secured cabinet. To be able to perform the cost-utility analysis, we included in the written, informed consent form the patients' permission to collect person-identifiable sensitive data from the medical record and from the Patient Registry Department at the Norwegian Directorate of Health. The results are presented so that the identity of the subjects cannot be identified, either directly or from derived information. Results from this study will be published in peerreviewed journals. Discussion This randomised controlled study, which we call AVRre, is the first programme to offer and test the effectiveness of a complex 24/7, around-the-clock intervention to optimise sAVR patients' safety and healthcare in the vulnerable readmission phase 30 days after discharge. The intervention is complex, because phone calls are made proactively to the patient 2 and 9 days after discharge, and because telephone support from expert healthcare professionals is made available day, evening and night during the first 30 days after hospital discharge. Combining experimental and explorative approaches results in mixed-method data, which will strengthen the conclusions we can draw and produce more solid information about sAVR patients' experiences at home. Analyses of patient narratives about the symptoms they experience and their needs during early rehabilitation will produce new insights for developing effective patient information systems and education programmes relevant for sAVR patients in the future. Moreover, symptom monitoring combined with evidence-based and clinical expertise advice can accommodate patients' desires to feel secure and to submit their requests for information after discharge from hospital [14]. Moreover, as we have hypothesised, this should reduce the number of 30-day hospital readmissions and reduce symptoms of anxiety and depression. Readmissions after sAVR are sparsely documented in the research literature, and reports of readmissions in RCTs, except for a few registry studies/observational studies, are almost unknown [8]. Mixing both quantitative and qualitative methods in this RCT increases the probability of obtaining valuable empirical knowledge from sAVR patients in addition to evidence of treatment effect [34]. First, triangulation generated by different data sources is possible; e.g. suppose a patient in the intervention group has a high score for anxiety on HADS, and that patient calls the AVR 24/ 7 phone to elaborate on and get advice for a case he felt anxious about after sAVR. This would validate information that stems from the instrument. Prevention of missing data to increase the representativeness of the sample in this trial is related to both designing and conducting the trial [32]. In designing the intervention, former patients and interdisciplinary specialists in the cardiac field revealed the themes for the intervention manual and 24/7 follow-up after discharge, in accordance with evidence-based literature. Moreover, the intervention is flexible, as it is based on when the patients need support. The patients in the control group receive information at discharge about group allocation and the importance of comparing the intervention and the control group in order to offer future sAVR patients a solid follow-up based on patients' needs. When conducting the study, the participants' burden and inconvenience of data collection is limited to only two questionnaires with a few items, to avoid missing values and drop-out. The project coordinator is dedicated to follow up the participants and the expert nurses responsible for the 24/7 intervention to limit missing data in the conduct of the study. It is time-consuming to carry out a 24/7 phone support service, and it requires expert healthcare professionals to be deeply and continuously motivated to seriously carry out the study. This is an ongoing challenge, and it is necessary to safeguard the strength of the study. The intervention is bolstered by experiencing and discussing patient cases and by the teaching of relevant themes during the intervention period. Moreover, assessment of the intervention's cost utility will provide valuable information for the healthcare system to develop ways to improve the transition of patient care to reduce readmissions [35]. Furthermore, knowledge from this study may add valuable information to optimize healthcare for future comparison to the emerging transcatheter aortic valve implantation (TAVI) patient population. Insight into an individual patient's pathway through the readmission process is made possible for the first time by patients' agreeing to allow researchers to gain access to register data. A normal pathway for a patient undergoing sAVR is to be transferred from an index, specialised hospital to a non-index hospital with a lower level of care at the fourth day after surgery. Fragmented care, which can occur when patients are transferred between hospitals at different levels in healthcare systems, increases the risk of mortality [5] and is a present challenge for patients and the healthcare system. This study is limited in that it includes patients only at one university hospital with two departments. Before surgery, the patients are informed about the expected increase in health status after surgery. Adding the QALY analysis takes into account both the quantity and quality of life generated by healthcare intervention and may add valuable preoperative information for future patients undergoing sAVR. The lack of masking in this study related to patients may have a potential influence on outcomes [24]. Patients are informed about group allocation 1-2 days before discharge from the University Hospital. If a patient from the control group and one from the intervention group by coincidence are in the same room, the project nurse has organised separate information about further follow-up in the study. The patient in the intervention group is encouraged not to share information about the intervention. The general information of the ongoing AVRre study at the Department might influence the patient in the control group and the caregivers, e.g. to offer more information than usual care and possibly threaten internal validity (the Hawthorne effect). In conclusion, the knowledge gained from this study will provide valuable insights for adjusting aspects of the healthcare system now to improve care for patients undergoing sAVR and will inform future studies on sAVR. The 24/7 phone-support manual has the potential to be modified and adopted for use by other surgical populations with high readmission rates.	2018-04-03T04:05:58.327Z	2017-05-30T00:00:00.000	{ "year": 2017, "sha1": "897c037bfd0c9555dbf48bcc8ad3b69f57e54327", "oa_license": "CCBY", "oa_url": "https://trialsjournal.biomedcentral.com/track/pdf/10.1186/s13063-017-1971-y", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "897c037bfd0c9555dbf48bcc8ad3b69f57e54327", "s2fieldsofstudy": [ "Medicine" ], "extfieldsofstudy": [ "Medicine" ] }
248133642	pes2o/s2orc	v3-fos-license	Ultra-sensitive ppb-level methane detection based on NIR all-optical photoacoustic spectroscopy by using differential fiber-optic microphones with gold-chromium composite nanomembrane In this paper, we propose and experimentally demonstrate an ultra-sensitive all-optical PAS gas sensor, incorporating with a near-infrared (NIR) diode laser, fiber-optic microphones (FOMs) and a double channel differential T-type photoacoustic cell. The FOM is realized by Fabry-Perot interferometry and novel gold-chromium (Au-Cr) composite nanomembranes. To meet the demand of high sensitivity and flat frequency response for the FOMs, the Au-Cr composite diaphragm is deliberately designed and fabricated by E-beam evaporation deposition with 330 nm in thickness and 6.35 mm in radius. Experimental results show that the FOM has a sensitivity of about 30 V/Pa and a flat frequency response from 300 to 900 Hz with fluctuation below 1 dB. Moreover, a double channel differential T-type photoacoustic cell is designed and employed in the all-optical PAS gas sensor, with the first-order resonant frequency of 610 Hz. The all-optical gas sensor is established and verified for CH4 detection and the normalized noise equivalent absorption (NNEA) is 4.42 × 10−10 W∙cm−1∙Hz−1/2. The minimum detection limit (MDL) of 36.45 ppb is achieved with a 1 s integration time. The MDL could be further enhanced to 4.87 ppb with an integration time of 81 s, allowing ultra-sensitive trace gas detection. Introduction Methane (CH 4 ), a colorless, odorless, flammable and explosive gas, plays an important role in environmental atmospheric monitoring, clinical diagnosis and industrial control [1][2][3]. According to the 2021 WMO Greenhouse Gas Bulletin [4], the concentration of methane in the atmosphere is about 1.8 ppm and arises yearly, which contributes to the greenhouse effect. Therefore, it is of great significance to implement accurate and sensitive CH 4 detection. Beyond conventional methods such as electro-chemistry and gas chromatography [5,6], optical methods allow improved selectivity and sensitivity as well as fast response. Optical methods using laser absorption spectroscopy (LAS) based on the 'finger-print' absorption lines of molecules, such as tunable diode laser absorption spectroscopy (TDLAS), cavity ring-down spectroscopy (CRDS) and photoacoustic spectroscopy (PAS) [7][8][9][10], have been widely investigated for trace gas detection. TDLAS-based sensors have been widely commercialized [11], and it fundamentally relies on the long optical path and superior photodetectors to achieve favorable sensing performance. Also, ultra-stable cavity mode is essential in CRDS, which limits practical implementations. PAS is acknowledged as an indirect method for trace gas sensing based on the photoacoustic effect. The photoacoustic effect refers to the principle that optical absorption of gas molecules results in localized heating and then generates gas vibration, which produces the acoustic pressure wave detected by an acoustic sensor i.e. microphone [12]. The PAS sensor has the special advantages of no wavelength selectivity and proportional to the laser power. Benefit from the development of laser technology, high power or mid-infrared laser is used in the PAS sensor to obtain high detection sensitive [13,14]. For example, Wu et al. reported a 3.3 µm inter-band cascade laser (ICL) in the quartz-enhanced photoacoustic spectroscopy (QEPAS) system to improve detection sensitivity [13]. Chen et al. reported a PAS sensor combined a high power near-infrared (NIR) laser with an erbium-doped fiber amplifier (EDFA), the output laser power up to 1 W [14]. Nevertheless, the use of ICL or EDFA made the sensor complex and costly, compared to single near-infrared diode laser. Beside optical sources, photoacoustic cells and microphones are the other key components in the PAS gas sensor. Initially, electrical microphones were widely used in PAS due to low cost and excellent stability [15]. However, there are restrictions in electrical microphones such as electromagnetic interference and low sensitivity. Attributed to compact size, high sensitivity, remote sensing and immunity to electromagnetic interference, fiber-optic microphone (FOM) have been investigated in past decades [16][17][18][19][20]. To be noted, reported diaphragms such as polymer and graphene composite used in optical microphones still have several limitations in the application [19,20]. The relatively low reflectivity of most non-metallic material diaphragm results in additional attenuation and noise. On the other hand, the unstable chemical properties of the diaphragm material bring the risk of performance degradation as working in harsh environment, which may restrict the gas sensing performance and could be enhanced. Moreover, considering relatively high resonant frequency of traditional H-type photoacoustic cell [21], the optical microphones often have a small area to obtain a larger flat response range to match the photoacoustic cell, but this would lead to lower sensitivity of the diaphragm. Taking account into the different boundary conditions of the T-type photoacoustic cell, lower resonance frequency and fast response can be obtained simultaneously compared to the traditional H-type cell [22]. However, existed acoustic wave antinodes at the end facet of the reported single channel T-type cell may lead to excessive signal noise due to optical window absorption and environmental interference. In this paper, we propose and demonstrate an ultra-sensitive alloptical PAS methane sensor incorporating with a pair of novel fiberoptic microphones (FOMs) and a double channel differential T-type photoacoustic cell. The acoustic signals are detected by two Fabry-Perot (FP) interferometric fiber-optic microphones with novel composite Au-Cr diaphragm and demodulated by the intensity-based quadrature point (Q-point) method. The proposed composite Au-Cr nanomembranes are designed by diaphragm forced vibration and fabricated by E-beam evaporation deposition with 330 nm in thickness (gold of 300 nm and chromium of 30 nm) and 6.35 mm in radius, achieving high acoustic sensitivity and flat frequency response. Moreover, a novel double channel differential T-type photoacoustic cell with the first-order resonant frequency of 610 Hz is designed and fabricated to match the proposed FOM for sensitivity enhancement. The wavelength modulation spectroscopy and second-harmonic detection (WMS-2f) with a low cost 1.65 µm DFB laser is employed to eliminate the fundamental frequency noise and increase signal noise ratio [23]. This all-optical PAS gas sensor is established and verified for CH 4 detection and the MDL of 36.45 ppb is achieved with a 1 s integration time. It could be further enhanced to 4.87 ppb with an integration time of 81 s, allowing ultra-sensitive trace gas sensing in versatile applications. Design of the composite diaphragm and FOM The essential acoustic detector directly determines the sensitivity and detection limit of the PAS sensor. For the diaphragm-based Fabry-Perot interferometric FOM, the vibration of diaphragm converts the acoustic signal into the variation of reflection spectrum. According to the membrane model for diaphragm vibration analysis, the tension of the diaphragm makes the diaphragm deformation to recover [24]. Based on the forced vibration of the diaphragm structure, the main influence on the sensitivity performance of the FOM is the deformation at the center of the circular diaphragm. By normalizing the deformation at the center of the diaphragm (r = 0), the vibration sensitivity S m (nm/Pa) and the first-order resonant frequency f 1 (Hz) can be calculated through the forced vibration equation by Eq. (1) and Eq. (2): (1) Where P, h, a, w, ρ, r is tensile stress, thickness, radius, angular frequency of acoustic, material density, radius in polar coordinates, respectively. k w is set as w/c = w/√P/ρ. J 0 is zero-order Bessel function [25]. It is implied from Eq. (1) that the vibration sensitivity S m could be directly improved by reducing diaphragm tension and thickness, as well as increasing material density and diaphragm radius. To be noted it could come across practical fabrication challenges, such as in the nanomembrane transferring process if the diaphragm is too thin. Gold diaphragm is primarily selected with high-density and chemical stability beyond other available materials. But in the metal nanomembranes deposition and transferring process, it is challenging to obtain thin layer and large proportion of pure gold nanomembrane simultaneously, due to tension and thermal effect. Here, a chromium substrate layer is introduced and a new type of Au-Cr composite diaphragm is proposed in our FOM. It is found that nanomembrane transferring process has been significantly upgraded by adding chromium as the substrate layer. The composite diaphragm can be considered theoretically as a uniform elastic diaphragm and deforms gradually in the process of diaphragm transfer. This approximation allows the interface force between the composite diaphragms around the free edge and the substrate approaching a shear force. Excessive shearing force will cause nanomembranes wrinkles and even ruptures. When the external pulling force t(x) works on the free end of diaphragm, it will produce a corresponding tension σ m . It is acknowledged that tension at the edge of the diaphragm σ m will evenly transfer to the entire interface. The governing equation q (x) can be calculated by Eq. (3): where q(x), h f , σ m , x is interface shear force with distance from diaphragm edge, thickness, free end of diaphragm tension and distance from diaphragm edge, respectively [26]. Where k can be described by Eq. (4): where E s,f , v s,f is Young's modulus and Poisson's ratio of substrate and film. Chromium has a higher Young's modulus and a lower Poisson's ratio than gold, therefore Cr-Si interface have less shear force according to Eq. (3). The composite diaphragm has higher chance of diaphragm formation, whose interface shear force distribution is shown in Fig. 1. Radius and thickness of the diaphragm play an important role in the FOM performance including frequency response and sensitivity characteristics. In terms of gold-chromium diaphragm, the bulk density and tensile stress of the diaphragm are set to 19,300 kg/m 3 and 50 MPa. Diaphragm response curve are obtained by changing radius and thickness of the diaphragm. As shown in Fig. 2, it is indicated that the diaphragm thickness merely affects the sensitivity with inverse proportion relationship, and the enlarged radius contributes to increase sensitivity but also brings lower resonant frequency. Flat frequency response and optimized sensitivity are both required around the photoacoustic cell resonant frequency. With comprehensive analysis of sensitivity and response flatness characteristics, the composite Au-Cr diaphragm is designed optimized and fabricated by Ebeam evaporation deposition. The composite nanomembranes of 330 nm thickness is realized with gold of 300 nm and chromium of 30 nm and 6.35 mm in radius. The detailed fabrication procedure for the metal nanomembranes and FOM package were discussed in our previous work [27]. The resonance frequency is calculated to be 3.1 kHz according to Eq. (2). The schematic structure and packaged samples of the proposed FOM is shown in Fig. 3. A low-finesse Fabry-Perot interferometer is established with the fiber facet and the Au-Cr diaphragm. When an acoustic pressure is applied to the diaphragm, the FP cavity length will change. Therefore, reflected intensity can be approximately described by Eq. (5): Where is the intensity of the incident light, the contrast of FOM, initial phase and phase to be measured [28]. The interference spectrum of the designed FOM (finesse number is about 2) was measured by the optical spectrum analyzer (AQ6370C-20, YOKOGAWA) as shown in Fig. 4. By choosing a proper wavelength, the FOM be operated in quadrature point (Q point, φ 0 = (m+1/2)π, m is an integer), which can make the FOM achieve maximum measurement sensitivity and dynamic range. To characterize the performance of the designed FOM, measured frequency response with the frequency range of 100-2500 Hz is shown in Fig. 5. The sensor exhibits a flat frequency range between 300 and 900 Hz. The sensitivity fluctuation is less than 1 dB while the average sensitivity is about 30 V/Pa. In addition, the first-order resonant frequency is about 1.9 kHz, which separates from frequency domain for acoustic detection, and experimental results show that the two proposed FOMs have similar sensitivity. Compared with the reported pure metal diaphragm, the composite diaphragm has a thinner thickness and larger radius to obtain ultra-high sensitivity and improved stability. In addition, benefits from the extremely stable physical and chemical properties of Cr and Au, the designed FOM can be applied in the measurement of corrosive or viscous gases for a certain long time. Double channel differential T-type photoacoustic cell A novel double channel differential T-type photoacoustic cell is designed and experimentally implemented in the PAS methane sensor. To optimize the performance of the system, a photoacoustic cell should be designed to match the FOM geometry and frequency. Compared with the conventional H-type cell, the T-type cell has a lower resonant frequency, a higher cell constant and faster response. For the T-type resonant PA cell, the 1st resonant frequency f can be calculated by Eq. (6): Where v is the speed of sound in the sample gas, L and R are the radius and length of the resonator [29]. Moreover, longer length and smaller radius of the photoacoustic cell could be properly considered to increase the cell constant and photoacoustic signals. The schematic design and 3D structure of the double channel differential T-type photoacoustic cell are shown in Fig. 6(a) and (b), respectively. The differential photoacoustic cell consists of two identical cylindrical channels as two photoacoustic resonators and one buffer volume at the near end of these channels. The inner radius and length of the identical channels are 4 mm and 120 mm. The inner radius and length of the buffer are 15 mm and 30 mm. A finite element model is constructed to simulate the acoustic field distributions at the first resonant frequency in the photoacoustic cell, as shown in Fig. 7(a). For the Ttype differential cell, the end facet of the photoacoustic resonators exhibits the acoustic antinode, thus a pair of identical FOMs is located at the end of each channel to detect the maximum acoustic pressure. In addition, the acoustic wave phase is reversed for the photoacoustic resonators, resulting the acoustic signals are doubled [30]. Since only one channel is excited by the laser source, the background noise caused by window absorption and external environment are effectively suppressed by using a differential amplifier to subtract the signals from the pair of FOMs. The simulated frequency responses of the designed photoacoustic cell are shown in Fig. 7(b). The first-order resonant frequency of the designed photoacoustic cell is about 629 Hz, which is located in the flat response region of the FOM. Experimental configuration The WMS-2f experiment of the PAS methane sensor system is depicted in Fig. 8. A DFB laser (NTT NLK1U5EAAA) with a central wavelength of 1.65 µm is used as the excitation source to generate the photoacoustic signals. A laser driver (Wavelength Electronics, model LDTC0520) is used to control temperature and provide injection current, and the laser temperature and current are set at 29.5 ℃ and 97.5 mA respectively to align with the selected methane absorption line at 1650.96 nm. The measured laser output power of the DFB laser is 14.7 mW. According to the HITRAN2012 database [31], the absorption coefficient of CH 4 gas molecule at 1650.96 nm is 0.37 cm − 1 , corresponding to line strength of 1.52 × 10 − 21 cm − 1 /(mol•cm − 2 ). An arbitrary waveform generator (Tektronix, Model AFG31102) is used to generate a sawtooth pattern with modulated sinusoidal wave. The sawtooth pattern is to sweep the laser central wavelength around the absorption line while the sinusoidal wave is used to modulate the laser wavelength for harmonics detection. The interaction of the modulated laser with the absorption line leads to the generation of signals at The output laser beam is horizontal collimated into one photoacoustic channel of the T-type differential photoacoustic cell through a fiber collimator (OZ optical, Model LPC-01). A CaF 2 window with diameter of 20 mm and transmissivity efficiency of > 95% is mounted on the one side of the photoacoustic cell and a silver-plated reflector is mounted on the other side to double the absorption path length. Two identical FOMs are located to detect acoustic signals from two channels. In the FOM signal demodulation part, a dual-channel narrow linewidth tunable laser (Alnair Labs, Model TLG-200) is used to connect to the fiber-optic microphones through optical circulators. The output wavelength of each channel is tuned to the Q-point of the interference spectrum of the two FOMs respectively. The interference spectrum is received by the photoelectric detector (New Focus, Model 1623) via the same optical circulator. The electrical signals are first processed by the differential pre-amplifier and then fed to a lock-in amplifier (LIA, Zurich Instruments, MFLI 500KHz) to demodulate the signals in a 2 f mode. The time constant and filter slope of the LIA are set to 1 s and 12 dB/Oct, corresponding to a detection bandwidth of 0.25 Hz. The demodulated signals are recorded and processed by the computer. The differential Ttype photoacoustic cell has gas inlet and outlet controlled by two electromagnetic valves. The gas inlet was located near the ends of buffer volume. To ensure the gas exchange and diffusion within the photoacoustic cell, two gas outlets were symmetrically positioned at the end facet of two identical cylindrical channels. The PTFE membrane of 3 μm thickness and desiccant are set in the inlet to filter particles and water when monitoring the CH 4 in air. Two bottles of CH 4 /N 2 standard gas (2% uncertainty) with 20 ppm and 500 ppm are diluted with pure N 2 gas to produce different concentrations by using a gas blender (MCQ Instruments, GB100, 1% uncertainty). A pressure controller and a flow meter are used to control the gas pressure and the flow rate within the gas pipeline. Optimization of the laser modulation frequency In the resonant PAS system, the acoustic signal maximizes when the laser modulation frequency agrees with the resonant frequency of the photoacoustic cell. The 1st resonant frequency of the designed photoacoustic cell is measured experimentally. The 50 ppm CH 4 is flowed photoacoustic the photoacoustic cell and the wavelength of the DFB laser is locked at 1650.96 nm for the CH 4 absorption line. The sinusoidal modulation frequency of the laser is scanned from 230 Hz to 1050 Hz, and the WMS-2f signal is recorded to measure the frequency response from 460 Hz to 2100 Hz. As shown in Fig. 9, there is a peak value at the modulation frequency of 305 Hz, therefore the 1st resonant frequency of the optimized T-type photoacoustic cell is about 610 Hz, close to the simulation value of 629 Hz and located in the flat response region of the employed FOM. Compared with the Fig. 5, it is known that a peak value of 950 Hz is generated at the first-order resonant frequency of the proposed FOMs. Optimization of the laser modulation amplitude In order to improve the intensity of the 2nd acoustic harmonic, the laser modulation amplitude is optimized in this experiment. The DFB laser is thermostatically controlled, and the bias current is adjusted to set the laser wavelength to 1650.96 nm. The 310 Hz sinusoidal modulation current ranging from 1.5 mA to 17.5 mA is then superimposed to the bias current. The photoacoustic cell is filled with 50 ppm CH 4 , and the output acoustic signal is recorded, as shown in Fig. 10. It is indicated that the optimal laser modulated bandwidth is about 11.78 GHz. Considering the full width at half maximum (FWHM) of the absorption line at 1650.96 nm being 4.64 GHz from the HITRAN database, the optimal laser modulation index is calculated to be 2.54 in the PAS sensor. Results and discussion In order to evaluate the performance of the PAS sensor for CH 4 detection, the parameters of this system are adjusted and setup properly. The CH 4 gases with different concentrations are obtained by the commercial gas blender. Various CH 4 /N 2 mixture from 2 ppm to 500 ppm are flowed into the photoacoustic cell. The laser bias current is increased from 86 mA to 110 mA. The relationship between single channel and differential double channels was investigated as shown in Fig. 11. It is shown that the magnitude of WMS-2f signals from the differential double channels basically doubled compared to that of the single channel. Furthermore, the inset shows that the phases of the acoustic waves from the two resonators are reversed, which are detected by two photoelectric detectors of the FOMs. The WMS-2f signals of CH 4 with various concentrations are shown in Fig. 12(a). The sensor response curve is obtained by imposing a linear fit to the experimental data-points, characterized by an R 2 value of > 0.999 and a slope of 57.26 μV/ppm. The background noise of the sensor is obtained by flushing pure N 2 into the photoacoustic cell with an integration time of 1 s, which is shown in Fig. 12(b). The background noise (1σ) is calculated to about 2 µV. The MDL could be estimated to be 36.45 ppb with the responsiveness of 57.26 μV/ppm, comparable with CH 4 sensors using mid-infrared lasers. The normalized noise equivalent absorption (NNEA), which is independent excitation source power and absorption strength, is calculated to be 4.42 × 10 − 10 W•cm − 1 •Hz − 1/2 with the condition of 0.25 Hz detection bandwidth and 14.7 mW optical power. For comparison, the performances of some state-of-the-art laser spectroscopy based CH 4 sensors are summarized in Table 1. To the best of our knowledge, the sensing NNEA and achieved MDL in this work are optimum among previously reported optical PAS methane sensors so far, even superior than those using mid-infrared laser sources. This gas sensor could be directly upgraded if using amplified near infrared laser or the mid-infrared laser source. An Allan-Werle deviation analysis was conducted in pure N 2 to evaluate the long-term stability of the PAS CH 4 sensor [32,33], and the results are shown in Fig. 12(c). It is observed that the Allan deviation is 11. The output WMS-2f signals from the single channel and from the differential double channel. In the inset are detected signals of the photoelectric detectors. 0.2789 µV for an integration time of 81 s. The corresponding MDL can be further improved to 4.87 ppb, extended Allan variance data can be obtained if further adding the PID feedback control to stabilize the Q point in the FOM demodulation [38]. Furthermore, the air in the laboratory environment is directly tested by the PAS sensor and the WMS-2f harmonic of CH 4 in air is shown in Fig. 12(d). It shows that the peak value of the 2 f harmonic is 0.225 mV, therefore the methane in the laboratory environment is measured to be around 3.365 ppm. With benefits of the optical fiber scheme, such as low transmission loss, intrinsic safety and immunity to electromagnetic interference, the demonstrated all optical PAS sensor shows potentials in the long-distance sensing for accurate trace gas detection. Conclusions In summary, a novel all-optical ultra-sensitive PAS CH 4 sensor is developed using a near infrared DFB laser. The essential composting diaphragmed FOMs and a differential T-type photoacoustic cell are designed and fabricated. Acoustic sensitivity and nanomembrane transfer success rate have be greatly improved by use of the novel Au-Cr composite diaphragm. The Au-Cr composite diaphragm is fabricated by E-beam evaporation deposition with 330 nm in thickness and 6.35 mm in radius. Compared with traditional electrical microphone, the FOM exhibits a high sensitivity of about 30 V/Pa and flat frequency response. Environmental and gas flow noise is inhibited and the acoustic signal is basically doubled by use of the unique multichannel differential T-type photoacoustic cell with the first-order resonant frequency of 610 Hz. 4 in the lab has also been monitored verified by the PAS sensor. The experimental results indicate that ultra-sensitive CH 4 gas monitoring can be achieved by the use of a low cost near infrared DFB laser. The developed all optical PAS sensor in this paper shows the features of ultra-sensitivity and optical fiber connection with the alloptical configuration, which pave the way for remote and longdistance trace gas detection. Declaration of Competing Interest The authors declare that there are no conflicts of interest.	2022-04-14T15:29:57.383Z	2022-04-01T00:00:00.000	{ "year": 2022, "sha1": "8afd5b0f23a6fb97dc8374d55215c5401dee6ba1", "oa_license": "CCBYNCND", "oa_url": "https://doi.org/10.1016/j.pacs.2022.100353", "oa_status": "GOLD", "pdf_src": "PubMedCentral", "pdf_hash": "9b581098b548268701eccbddae0101e18188d990", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Medicine" ] }
8085986	pes2o/s2orc	v3-fos-license	Heavy-Ion Physics with CMS This article presents a brief overview of the CMS experiment capabilities to study the hot and dense matter created in relativistic heavy-ion collisions. The CERN Large Hadron Collider will provide collisions of Pb nuclei at 5.5 TeV per nucleon. The CMS heavy ion group has developed a plethora of physics analyses addressing many important aspects of heavy-ion physics in preparation for a competitive and successful program. Introduction The CMS heavy ion program will study the QCD matter created in Pb+Pb collisions at the Large Hadron Collider (LHC), at CERN. The Pb nuclei will be collided at an energy of 5.5 TeV per nucleon, the highest ever reached by a particle accelerator. The energy density will surely exceed that currently accessible at the Relativistic Heavy Ion Collider (RHIC), allowing us to explore a novel QCD regime, where semi-hard and hard processes will be dominant. The CMS heavy ion program will study all aspects of particle production from soft physics at relatively low momenta, to hard probes and ultra-peripheral collisions. The broad range of physics topics and future analysis possibilities are summarised in the CMS Heavy Ion Technical Design Report on high-density QCD with heavy-ions 1 . Here, only a very brief summary of some of the planned physics analyses will be discussed. The CMS detector is well equipped to carry out a successful heavy ion program. The detector has a layered design with full azimuthal coverage over a wide pseudorapidity range. The main systems -the Silicon tracker, the electromagnetic (ECAL) and hadronic (HCAL) calorimeters, the muon systems and the solenoidal magnet -are detailed in the Technical Design Report 2 . The global event characterisation measurements will extensively rely on information from the Silicon tracker (both the pixel and strip layers), which covers the pseudorapidity window \|η\| < 2.5. The tracker has a momentum resolution of better than 2% for tracks with p T < 100 GeV/c (\|η\| < 0.5) thanks to the powerful 4 T magnetic field. This detector will provide Figure 1: Charged particle p T spectra expected for Pb+Pb collisions at 5.5 TeV with nominal integrated luminosity (0.5 nb −1 ), in 9 centrality bins, offset by factors of 100 for illustration purposes, only using the minimum bias triggered sample (left) and using the jet-triggered data sets (right). both charged particle tracking and vertex reconstruction. Given the high granularity of the pixels, the occupancy for the inner pixel layers is expected not to exceed the 3% level, even in the heavy ion environment. The CMS calorimetry and muon identification systems will provide information for the measurement of specific probes, such as identified γ and jets. CMS has a very extensive pseudorapidity coverage of the electromagnetic (\|η\|<3) and hadronic (\|η\|<5.2) calorimeters, useful, in particular, for the measurement of jets. This coverage is further extended by the Castor (5<\|η\|<6.6) and ZDC (\|η\|>8.1, for neutrals) forward calorimeters. The mid-rapidity muon chambers, interspersed in the return yoke magnetic field (2 T) will provide a very fast and precise measurement of muon position and (when coupled with the tracker) momentum. This clean sample of muons will be used to reconstruct quarkonia (J/ψ and Υ family) via the di-muon channel. CMS has an elaborate triggering system. The combination of fast Level-1 hardware systems and powerful computational High Level Trigger will be crucial for a successful heavy ion program. Soft physics The measurement of identified particles is essential to fully characterise the global properties of the created medium: the initial energy density, the freeze-out properties of the system, and its hadro-chemistry. For this purpose, tracking algorithms have been developed for low momentum (p T <1.5 GeV/c) charged particle track reconstruction, using the pixel detector. In this kinematic range, the ionisation energy loss in the pixel layers is exploited to identify π ± , K ± (0.2<p T <0.8 GeV/c), and p(p) (p T <1.5 GeV/c) with a resolution of 5-7%. Hard probes Hard probes are of special interest to probe deconfinement properties. Quarks and gluons, initially created via hard partonic interactions in the collision, are expected to interact strongly and lose energy in the resultant hot, dense QCD matter. Thus, they are used as a probe of early times in the collision evolution and a probe of the medium itself. The suppression of high-p T particles 4 (due to medium induced parton energy loss) and the disappearance of backto-back jets in central Au+Au collisions 5 are two of the main physics results connected with the properties of the hot, dense QCD medium created at RHIC energies. Neither of these effects are observed in cold nuclear matter -peripheral Au+Au or d+Au collisions. The transition from RHIC to the LHC is especially beneficial for hard probe measurements. Firstly, the increase in accelerator luminosities will allow access to more hard scattering events. Secondly, the vastly increased collision energy results in an increase in cross-section for hard interactions, that will yield a large statistical dataset of jets, not obtainable at RHIC. Thirdly, specific to CMS, a sophisticated high-level triggering capability will provide an excellent foundation for hard probe physics. It will extend the p T reach of the collected events (for charged hadrons and reconstructed jets) by a factor of three over the number of events recorded in the absence of the trigger system (see Figs. 1 and 2). The CMS 4 T magnetic field provides a good tracking efficiency (∼ 75%), resolution (∼ 1.5%) and low fake rate (< 5%) for central collision data in the mid-rapidity barrel region (η < 0.5) for tracks with transverse momenta of up to 100 GeV/c. Finally, the extensive calorimeter coverage in both η and φ will provide full jet reconstruction in heavy-ion collisions. Jet reconstruction utilises both the hadronic and electromagnetic calorimeters. The "iterative cone" algorithm with radius R=0.5 and background subtraction is used. An efficiency and purity of ∼ 100% is attained for jets with E T >75 GeV. The energy resolution for jets with E T >100 GeV is ∼ 15%. In the first year's running (0.5 nb −1 ), the expected reach for jets is up to a transverse energy of 0.5 TeV (see left panel of Fig. 2). A particular hard scattering interaction that results in back-to-back γ-jet production is considered a "golden" probe for the study of in-medium energy loss. The direct or prompt photon will not interact with the medium created, thus providing a measure of the initial jet energy scale (before medium interaction) on an event-by-event basis. Isolation of the γ is attained by using combined information from ECAL clusters (photon identification) and HCAL. Any surrounding jet activity in HCAL is used to veto jets in which (for example) leading π 0 (→ γγ) fragments also result in ECAL clusters. To construct the fragmentation function, the ECAL clusters from isolated photons are correlated with reconstructed calorimeter jets. A pair of back-to-back γjet is selected if ∆φ(γ, jet)>172 0 is fulfilled. The E T of the photon is used as an estimate of the original E T of the away-side parton. The jet fragmentation function (ξ = ln(E γ T /p T )) is then formed using the charged hadrons reconstructed in the tracker, on the away-side to the γ. partially translated into lower momentum particles (high ξ). Additional details can be found in Ref. 6 Heavy quarkonia measurements are the signature measurements of the CMS heavy ion program, through reconstructed J/ψ and Υ via di-muons using the muon chambers. Such measurements will provide crucial information on the many-body dynamics of high-density QCD matter. It is generally considered that the step-wise suppression of heavy quark-antiquark bound states will be one of the direct probes of Quark-Gluon Plasma formation 7 . Measuring quarkonia crosssections with an excellent signal-to-background ratio (the best at the LHC) and with a mass resolution of 1% of the quarkonium mass (for both muons in \|η\| < 2.4) will allow a clean study of the J/ψ and Υ families. The broad η and high-p T acceptance combined with HLT selection will provide a statistical sample of 1.8×10 5 J/ψ and 2.5×10 4 Υ from the first year of running (0.5 nb −1 ). Summary The design of the CMS detector is suitable to provide a broad range of physics measurements for the study of high-density QCD. The highly segmented Silicon pixel detector is instrumental in the identification of low momentum particles, in order to study the systematics of bulk particle production. The extensive coverage of the calorimeters will provide measurements of fully reconstructed jets up to E T ∼0.5 TeV in the first year of running (0.5 nb −1 ). The muon chambers, coupled with the tracker, will provide precision measurements of the J/ψ and Υ families which may shed light on whether a Quark-Gluon Plasma has actually been formed or not. A wealth of other important measurements, which include elliptic flow and ultra peripheral collisions, will lead to a competitive heavy ion program at LHC.	2008-06-06T15:32:45.000Z	2008-06-06T00:00:00.000	{ "year": 2008, "sha1": "cece6332a7fec957e4ae793c75cc2b65e700c04a", "oa_license": null, "oa_url": null, "oa_status": null, "pdf_src": "Arxiv", "pdf_hash": "cece6332a7fec957e4ae793c75cc2b65e700c04a", "s2fieldsofstudy": [ "Physics" ], "extfieldsofstudy": [ "Physics" ] }
239730657	pes2o/s2orc	v3-fos-license	Open-File Behavior and Survival of Hatchery Rainbow Trout ( Oncorhynchus mykiss ) in the Upper Cowlitz River Basin, Washington, 2013 and 2017 A two-year study (2013 and 2017) was conducted to determine if annual releases of hatchery rainbow trout (resi-dent Oncorhynchus mykiss ) in the upper Cowlitz River Basin, Washington adversely affected anadromous fish in the basin. Rainbow trout tagged with radio transmitters were monitored after release to describe movement patterns, entrainment rates at Cowlitz Falls Dam, and survival. Additionally, trout that were radio-tagged in 2017 were monitored during spring 2018 to determine if any moved upstream and entered tributaries where winter steelhead (anadromous Oncorhynchus mykiss ) spawning occurs. A total of 580 hatchery rainbow trout (122 in 2013 and 458 in 2017) were radio-tagged and released at three release sites: (1) Cowlitz Falls Campground on Cowlitz River Arm of Lake Scanewa river kilometer (rkm) 155, (2) Cispus River Arm of Lake Scanewa rkm 1, and (3) Day Use Park on Cowlitz River Arm of Lake Scanewa rkm 146. Most radio-tagged trout (70 percent) remained within 6.4 rkm of the release site but some fish moved at least 25.7 rkm from the release site. The predominant movement direction was downstream. More than twice as many fish released at Cowlitz Falls Campground in 2017 (compared to the other two release sites) remained in the Cowlitz River, where potential overlap with steelhead occurs. A total of 28.3 percent of the study fish were entrained at Cowlitz Falls Dam. Apparent survival (time until movement ceased) for most tagged trout was fewer than 100 days from release in both years and no fish were detected moving during the spring following their release. In summary, hatchery rainbow trout released upstream from Cowlitz Falls Dam seem to remain primarily in Lake Scanewa or entrained at Cowlitz Falls Dam with few fish surviving to winter months. We found no evidence of hatchery trout interacting with steelhead in spawning tributaries during spring months. These results suggest that trout stocking in the upper Cowlitz River Basin poses minimal threat to anadromous fish in the basin. Introduction Hatchery rainbow trout (Oncorhynchus mykiss) are commonly released into streams, lakes, and reservoirs to provide or supplement fisheries in various locations worldwide (MacCrimmon, 1971;Walters and others, 1997;Bettinger and Bettoli, 2002;Farrington and others, 2004;High and Meyer, 2009;Candiotto and others, 2011). In the Pacific Northwest, United States, rainbow trout are annually released in some rivers to mitigate for fisheries that were eliminated because of the construction and operation of hydropower dams. Dams and reservoirs blocked access to the habitat of native anadromous salmon, resulting in the extirpation of these fishes upstream from many dams. In some cases, mitigation fisheries have been established to offset these losses. In Lake Scanewa, upstream from Cowlitz Falls Dam, approximately 25,000 rainbow trout are released each year during June-August to mitigate for lost fishing opportunities when dams were constructed in the Cowlitz River during the 1960s and 1990s. If many unharvested trout remain in the upper Cowlitz River Basin, resident and anadromous fishes could potentially be impacted. Estimates from creel surveys conducted during 2000 and 2010 indicated that less than onehalf of the released hatchery trout were eventually captured by anglers (Tipping and Serl, 2000;Liedtke and others, 2011). Liedtke and others (2011) also found 2.6 percent of trout greater than (>) 30 centimeters (cm) and 0.2 percent of trout less than (<) 30 cm had salmonids in their stomachs. Because the stomach content sampling did not seasonally coincide with the peak abundance of salmon fry and parr, these were likely conservative estimates of predation. These results indicated that hatchery trout could have an impact on juvenile salmonid populations, but dispersal and survival of the trout are unknown. Studies of hatchery rainbow trout in the upper Cowlitz River Basin prior to 2013 have focused on creel surveys and stomach samples to estimate impact on anadromous salmonids. A substantial amount of research has assessed how hatchery rainbow trout respond after release in recreational fisheries. In general, studies have shown that rainbow trout tend to have low return-to-creel rates, limited dispersal from release sites, and low survival. Research on return-to-creel rates suggests that less than 50 percent of the released trout are eventually harvested by anglers (Cresswell, 1981;Heimer and others, 1985;Wiley and others, 1993). Previous studies on the Cowlitz River are consistent with this finding, with return-to-creel of 37.9 percent (Liedtke and others, 2011) and 41 percent (Tipping and Serl, 2000). In many studies, hatchery rainbow trout generally remained close to the stocking site. Post-release dispersal in rivers and streams was often within 5 river kilometers (rkm) (Shetter, 1947;Bjornn and Mallet, 1964;Bettinger and Bettoli, 2002;High and Meyer, 2009), but most fish were caught within 1 rkm of the stocking location (Trembley, 1945;Helfrich and Kendall, 1982;Heimer and others, 1985;Baird and others, 2006). However, some studies showed rainbow trout moved as far as 24 rkm away (Shetter and Hazzard, 1941;Bjornn and Mallet, 1964;Bettinger and Bettoli, 2002). The limited dispersal of stocked fish is beneficial for anglers (if access is not limited) and may help reduce potential interactions with native populations. Stocked hatchery rainbow trout can reduce native fish populations because of competition and predation (Miller, 1958;Beauchamp and others, 1995;Seiler and Keeley, 2009). Competition for resources such as space and food can contribute to low survival of resident fish, especially at high stocking rates (Petrosky and Bjornn, 1988). Hatchery fish tend to be more aggressive, more active, and are larger than wild conspecifics and therefore have a higher metabolic rate (Butler, 1975;Ersbak and Haase, 1983;Mesa, 1991;McMichael and others, 1999;Bettinger and Bettoli, 2002). Bachman (1984) found wild brown trout used a sit-and-wait approach for feeding, taking advantage of drifting food while hatchery fish moved more and fed less. The potential negative impacts that hatchery trout might have on resident populations would be short-term if the stocked fish died quickly or moved downstream. Research has shown that most rainbow trout survive less than 90 days following release (Shetter and Hazzard, 1941;Walters and others, 1997;Bettinger and Bettoli, 2002;High and Meyer, 2009) and few trout survive to the following fishing season after a winter (Shetter and Hazzard, 1941;Cooper, 1953;Heimer and others, 1985;Wiley and others, 1993;Meyer and Griffith, 1997). We conducted a study in two separate years (2013 and 2017), using radio telemetry to quantify movement and survival of hatchery rainbow trout from three release sites in the upper Cowlitz River Basin. The objectives were to assess postrelease movement, post-release survival, and potential spatial overlap with spawning steelhead. Study Area The Cowlitz River, in southwestern Washington, is a primary tributary to the lower Columbia River, the largest river in the Pacific Northwest. The Cowlitz River drains 6,698 square kilometers (km 2 ) on the western slopes of the Cascade Mountain range. Three dams are present: Mayfield (rkm 84) and Mossyrock (rkm 105) dams, constructed during the 1960s, and Cowlitz Falls Dam (rkm 143), which began operating in 1994. The upper Cowlitz River Basin, located upstream from Cowlitz Falls Dam, is drained by the Cowlitz (1,577 km 2 drainage area) and Cispus (1,121 km 2 drainage area) rivers ( fig. 1). They share a confluence at rkm 144, near the center of Lake Scanewa, a small reservoir which has a surface area of 264 hectares. Hatchery trout were stocked in three locations: (1) Cowlitz Falls Campground on Cowlitz River Arm of Lake Scanewa (Campground) at rkm 155, (2) Cispus River Arm of Lake Scanewa (Cispus Arm) at rkm 1, and (3) Day Use Park on Cowlitz River Arm of Lake Scanewa (Cowlitz Arm) at rkm 146. Radio-tagged fish were released at Campground, Cispus Arm and Cowlitz Arm ( fig. 1 Radio Telemetry Monitoring Radio telemetry monitoring sites (hereafter, fixed sites) were installed at several locations in the Cowlitz River basin during 2013 and 2017 to monitor movements of rainbow trout that were tagged and released for the study. ( fig. 1). In 2013, fixed sites were installed at the following locations: • Riffe Lake, Cowlitz River rkm 132.9 • Taidnapam Mobile tracking was conducted (by vehicle and boat) to augment detections from fixed sites. During the 2013 study year, mobile tracking occurred on three dates: September 7, 2014; November 16, 2014; and January 25, 2015. In 2017, mobile tracking occurred on 14 dates during July-November, 2017. Trout tagged in 2013 were monitored on fixed gear through June 30, 2014, and trout tagged in 2017 were monitored on fixed gear until July 25, 2018. Trout mobile tracking was supplemented by a concurrent radio telemetry study monitoring adult steelhead in 2017 and 2018 in Lake Scanewa and the Cowlitz and Cispus Rivers, upstream from the trout monitoring fixed sites (Liedtke and others, 2020). Fish Tagging and Releasing On each stocking date, hatchery rainbow trout (hereafter, rainbow trout) were transported by truck from a commercial facility near Orting, Washington (about 1-hour transport time) to release locations in the upper Cowlitz River Basin. Approximately 25,000 fish, averaging 2 fish per pound, were stocked bi-monthly during June-August per year. Stocked fish were released at three sites on each release date ( fig. 1). Study fish were subsampled from the stocked fish just prior to release or collected in the reservoir. During the 2013 study, approximately 40 fish were radio-tagged in July, August, and November for a total of 122 trout (table 1). During the July and August tagging events, fish were netted from the transport truck at the Cispus Arm stocking site located at rkm 1. Following tagging at the Cispus Arm site, one-half of the fish were released, and one-half were transported by boat to the Cowlitz Arm and released. During the November tagging event, fish were collected in Lake Scanewa using hook and line, transported to the PUD boat launch for radio-tagging, and then transported by boat to the Cowlitz Arm and Cispus Arm release sites. In 2017, fish were netted from the transport truck, tagged, and released at three stocking sites-Campground, Cispus Arm, and Cowlitz Arm. Approximately 50 tagged rainbow trout were released at each stocking site in June, July, and August, for a total of approximately 150 fish each month (table 1). Prior to tagging, fish were held in containers with flowthrough river water at densities less than 20 grams (g) of fish per liter of water. Collected fish were held from 1 to 6 hours until tagging, except for the fish collected by angling in November of 2013, which were held overnight to recover from collection then tagged and released the following day. Radio transmitters were surgically implanted using methods described by Liedtke and others (2012). Ten percent Aqui-S 20E ® (Aqua Tactics, Kirkland, Washington) was used as an anesthetic during tagging. Radio-tag model TX-PSC-I-450 (8.5 g in air; Sigma Eight, Inc.) was implanted during 2013. Tagged fish recovered for 10 minutes or until fish reached equilibrium in 68-liter plastic totes and then were released. There were no differences in fish size between release sites, but size varied by month ( fig. 2; table 1). In 2013, there were no differences in mean fork length or weight of fish between release sites (one-way generalized linear model [GLM]; fork length P=0.9754; weight P=0.3430). Fork lengths of tagged fish ranged from 21.8 to 30.8 cm (mean 26.7; standard deviation [SD] 1.4) and weights ranged from 168.9 to 331.0 g (mean 210.0; SD 28.3). Fork length increased each month (one-way GLM; P<0.0001) and fish were heavier in August than the other two months (one-way GLM; P=0.0219; fig. 2 Data Analysis Detection records from fixed telemetry sites and from mobile tracking were merged with tagging and release data to create a combined detection dataset. False positive detections were identified in the dataset using an automated filter described in Beeman and Perry (2012). The remaining data were reviewed by the authors and reconciled to create the final dataset. All Global Positioning System coordinates of fish locations were rounded to the nearest river mile in each river or creek and then converted to river kilometers. These data were managed, processed, and analyzed using SAS ® version 9.4 (Cary, North Carolina). R e l e a s e d a t e s F o r k l e n g t h , i n c e n t i m e t e r s Our goal was to determine the following: (1) how far fish moved away from release sites, (2) how long fish survived after release, (3) the proportion of released fish that were entrained at Cowlitz Falls Dam, and (4) if tagged rainbow trout were detected in steelhead spawning tributaries during the steelhead spawning period. The maximum distance moved was calculated to determine the maximum river kilometer detected upstream and downstream from each release site. Fish entrained at the dam were of interest because they were no longer available for harvest. Entrainment at Cowlitz Falls Dam occurred if a fish passed downstream from the dam through the turbines or into the fish collection facility. Percent entrainment was calculated by dividing the number of tagged fish collected at Cowlitz Falls Dam or detected downstream from Cowlitz Falls Dam by the total number of tagged fish released during each study year. To determine if a flow-entrainment relationship existed, mean daily flow data on the Cowlitz River downstream from Cowlitz Falls Dam were downloaded from U.S. Geological Survey National Water Information System (U.S. Geological Survey, 2020) and compared to daily observations of tagged fish in the collection facility at the dam and downstream from the dam. Apparent survival was calculated using fish movement data. Fish detected moving within the study area were assumed to be alive, whereas fish that stopped moving and never were detected moving again were assumed to be dead. Apparent survival was defined as the elapsed time from release until movement cessation. Elapsed time from release to movement cessation was plotted using the Kaplan-Meier survivorship function to illustrate apparent survival. Statistical significance was assessed using an alpha of 0.05 when comparing apparent survival. The fate of each fish was identified to estimate the percentage of fish that were in the Cowlitz River or Cispus River, that remained in Lake Scanewa, fell back downstream from Cowlitz Falls Dam, or were harvested or predated. Fish were assigned Cowlitz River or Cispus River fates if they were last detected in either river; fish were assigned a Lake Scanewa fate if they were last detected in Lake Scanewa; and fish were assigned an entrainment fate if they were detected downstream from Cowlitz Falls Dam or collected in the fish collection facility at the dam. Harvest was determined from angler reports. In addition, tags mobile tracked to residences were assigned to the harvest fate. Fish that were tracked to areas on land or near birds of prey nesting areas were assigned to a predation fate. Evaluation of Spawning Overlap with Steelhead During 2017 and 2018, a concurrent telemetry study was conducted in the upper Cowlitz River Basin that focused on adult steelhead (O. mykiss) movements (Liedtke and others, 2020). A total of 130 natural and 85 hatchery-origin steelhead were radio-tagged and released at the Cowlitz Arm site and at rkm 28 in the Cispus River from February through May of 2017 and 2018. From March 2017 to July 2018, we mobile tracked radio-tagged steelhead and trout in an area encompassing the trout study-45 rkm in the Cispus River near the confluence of Prospect Creek and the 72 rkm in the Cowlitz River upstream from Cowlitz Falls Dam. Nearby tributaries were also monitored; mobile tracking occurred in 18 Cowlitz River tributaries and 13 Cispus River tributaries. Results A total of 69.6 percent of tagged fish stayed within 6.4 rkm of the release site but some fish moved at least 25.7 rkm from the release site ( fig. 3). A total of 24.7 percent of fish moved between 6.5 and 12.9 rkm while few fish (5.5 percent) moved more than 12.9 rkm away from the release site ( fig. 3). Five percent of fish did not leave the release site and, of the fish that did move from the release site, movement was limited and predominantly downstream. The median maximum distance moved was 3.2 rkm downstream and 0 rkm upstream for both years. Tagged fish moved greater distances away from the release site in 2017 than in 2013 and were detected as far as 25.7 rkm upstream and 24.1 rkm downstream. There were few differences in maximum distance moved between release month in either year, but some fish in 2017 moved a greater distance upstream after release in July and August than in June. Most fish released at the Lake Scanewa release sites (Cispus and Cowlitz Arms) stayed in Lake Scanewa or were entrained (were collected at or passed through Cowlitz Falls Dam in Lake Scanewa). Similarly, most (67.3 percent) of the fish released at the river site (Campground) remained in the Cowlitz River. More than one-half (58.2 percent) of the 2013 trout and 30.8 percent of the 2017 trout released at Cispus or Cowlitz Arms had fates indicating their last detection was in Lake Scanewa compared to 0.7 percent of the Campgroundreleased fish (table 2). In 2013, few fish had river fates (Cowlitz or Cispus) while 3.3, 6.5, 15.7, and 20.4 percent of fish released at the Lake Scanewa release sites had river fates in 2017 (table 2). River fates suggest a potential overlap with steelhead who spawn in the upstream tributaries of the Cowlitz and Cispus Rivers and 67.3 percent of Campground fish remained in the Cowlitz River at the end of the study. Few tagged fish were reported as harvested in either year. Of the 122 radio-tagged trout released in 2013, four (3.3 percent) were known to be harvested, three of them between December 2013 and January 2014. Harvests occurred 22, 25, 61, and 86 days after release. In 2017, four tags were reported as harvested fish by anglers. These fish were released at the Cispus Arm (n=2) and Cowlitz Arm (n=2) sites. Of the reported harvest, fish were captured 14, 32, 32, and 52 days after release. Several radio tags were mobile tracked to areas considered to be evidence of harvest, such as near a residence. An additional 13 fish in 2017 were considered harvested for a total of 3.3 percent in 2013 and 3.7 percent in 2017 (table 2). Total harvest numbers are likely underestimated due to few reported harvested fish compared to other studies in the upper Cowlitz River Basin. Some fish were mobile tracked to locations indicating predation, including proximity to a bird of prey nest. Ten fish in 2013 (8.2 percent) and 32 fish in 2017 (7.0 percent) were assigned predation as their final fate (table 2). A total of 28.3 percent of fish released were entrained at Cowlitz Falls Dam by either collection at the Cowlitz Falls Fish Facility or fallback downstream into Riffe Lake. A nearly equal percentage of fish released in 2013 at the two release sites were detected as fallbacks in Riffe Lake (22.6 and 21.7 percent; table 2). No radio-tagged trout were recorded at the collection facility in 2013. In 2017, 7.6 percent of fish were collected at Cowlitz Falls Dam and 22.3 percent were detected as fallbacks in Riffe Lake. Fish released at the Cowlitz Arm site experienced substantially higher fallback than fish released at the Cispus Arm and Campground sites (table 2). Most entrainment occurred in August and early September when flows were low, but entrainment continued throughout winter in both years ( fig. 4). Median time to entrainment was 33.5 days in 2013 and 9.4 days in 2017. Most fish were entrained or stopped moving and were assumed to have died within 100 days of release. Median apparent survival (time from release to entrainment or movement cessation) was nearly double in 2013 (30.4 days) than in 2017 (16.4 days) but the survival trends were similar ( fig. 5). A total of 75 percent of fish stopped moving by day 57 in 2013 and day 45 in 2017 and less than 9.0 percent of fish were detected moving within the study area after 100 days in both years. None of the tagged trout were detected moving during spring months in the year following release. There were no differences in apparent survival between month of release in either year (Wilcoxon test, 2013(Wilcoxon test, : Z=4.3145, P=0.11562017: Z=3.7221, P=0.1555. In 2017, fish released at the Cispus Arm had slightly longer apparent survival (median Cispus Arm 26.2 days) compared to fish released at the other two sites (median Campground 13.7 days; Cowlitz Arm 12.8 days). Evaluation of Spawning Overlap with Steelhead The last detected movement of any rainbow trout was within 165 days of release and no movement was detected the following spring of either year. Some radio-tagged trout moved upstream and were last detected in the Cowlitz and Cispus Rivers during the year they were released. However, all those fish had detection histories which indicated that they died, and we never detected tagged rainbow trout in either river the following spring when steelhead spawning occurred. No tagged trout were detected during mobile tracking as part of the steelhead effort in 2018. No radio-tagged trout released in 2013 or 2017 were found in steelhead spawning tributaries. Discussion Tagged hatchery rainbow trout primarily remained near release sites and most had an apparent survival of less than 100 days. The results from this study are consistent with other studies. For example, Trembley (1945) and Baird and others (2006) found rainbow trout moved less than 1.6 rkm from release locations in Spring Creek, Pennsylvania and the Moose River, New York, respectively. In other studies, rainbow trout moved less than 5 rkm (Shetter, 1947;Bjornn and Mallet, 1964;High and Meyer, 2009). High and Meyer (2009) reported 85 percent of trout were presumed dead at 30 days after release in an Idaho River. In Tennessee, 93 percent of stocked rainbow trout died or moved out of the study area within 11 weeks of stocking (Bettinger and Bettoli, 2002). Similarly, in the Colorado River, trout were only caught within four weeks of stocking (Walters and others, 1997). Our findings suggest that rainbow trout survival in the upper Cowlitz River Basin is relatively short, providing angling opportunities in the summer months near the release sites, but long-term survival beyond the summer fishing season is limited. High harvest rates close to the stocking dates can influence potential low survival within a few months after stocking (Cresswell, 1981;Dillon and others, 2000;High and Meyer, 2009). There were few reports of study fish harvested in either year of our study, which could be a result of our limited presence at the boat ramps during fishing season, lack of creel surveys, unreported or illegal harvest, or reluctance of anglers to report tags. From creel surveys in 2010, Liedtke and others (2011) reported 1,214 anglers resided in the county encompassing the study area. Using mobile tracking, we detected several unreported radio tags in residences, providing evidence to categorize these fish as harvested. These residences were away from water bodies, reducing the possibility of errant radio signals. In addition, we mobile tracked some transmitters to areas on land. These land-based locations could have been a combination of avian predation, natural mortality of a trout followed by scavengers dragging the carcass to land, or angler harvest and tag removal. More than one-half of the known harvest was within 32 days of release with the remaining harvest at 2-3 months from release. When combining reported harvest with predation from our mobile tracking efforts, our harvest rates are still likely underestimated. We found 11.5 and 10.7 percent combined harvest and predation in our 2013 and 2017 study years, respectively. Creel surveys were not conducted in either study year. Using creel surveys, Tipping and Serl (2000) and Liedtke and others (2011) reported 38-41 percent of stocked rainbow trout were harvested in the upper Cowlitz River Basin, leaving many fish in our studies unaccounted for. Once released, hatchery fish have a hard time transitioning from manufactured pellets to variable and seasonal food sources in the wild. In the wild, hatchery fish concentrate on food that is similar in size, shape, and coloration to artificial food pellets in hatchery diets (Ersbak and Haase, 1983). In the first months after stocking, planted hatchery trout lost weight due to starvation (Miller, 1952;Ersbak and Haase, 1983). As food resources change throughout the year, hatchery rainbow trout had difficulty adapting to available food as evidenced by few prey items (Fenner and others, 2004) and limited variety in their diet (Weiland and Hayward, 1997). In addition to limited variety in their diet, hatchery trout tended to spend more time and effort actively feeding, thereby increasing their caloric requirements (Butler, 1975;Bachman, 1984). Although it was not a formal part of our study, we observed evidence of the feeding challenges for hatchery rainbow trout. Fish collected for tagging by angling in the upper Cowlitz River Basin in November 2013 were the survivors of the July and August releases. The fish collected in November had greater fork length but weighed less than those released in July and August, suggesting that trout released in the upper Cowlitz River Basin lost weight in the 3-4 months following release. Water temperature, food availability, and competition can affect growth rates (Miller, 1958;Beauchamp and others, 1995;Baird and others, 2006;Seiler and Keeley, 2009), but based on examination of stomach contents (Liedtke and others, 2011), hatchery trout in this study seemed to not be feeding effectively. We believe this ineffective foraging was the primary factor influencing lack of growth. The lack of adapting to effectively foraging in the wild may cause some of the low apparent survival beyond 100 days. Several studies have reported that hatchery rainbow trout consumed live prey. In laboratory trials, hatchery rainbow trout diets consisted of less than one-half of the prey than that of wild fish (Ward and others, 2018), suggesting that stocked trout may impact live prey less than native steelhead. In the upper Cowlitz River Basin, Liedtke and others (2011) found fish prey in 2.3 percent of stomachs sampled, but two-thirds of the prey were non-salmon species. Liedtke and others (2011) estimated their 2.1 percent of salmonid predation was an underestimate due to the timing of their creel census which coincided with fewer juvenile salmonids present. Predation was highest in the upper Cowlitz River Basin during February, March, and May (Liedtke and others, 2011), when few of the tagged trout were thought to be alive. Size of the fish can impact piscivory where larger fish consume more prey. Beauchamp (1990) reported rainbow trout >25 cm were piscivorous throughout the year in Lake Washington, Washington. In the upper Cowlitz River Basin, more than eightfold of salmonids were consumed by trout >30 cm long than trout less than 30 cm long (Tipping and Serl, 2000;Liedtke and others, 2011). Less than 1 percent of hatchery trout released in 2013 and 2017 were >30 cm, indicating that a small percentage of the stocked trout could be impacting live prey and juvenile salmonids could be at risk. However, few juvenile anadromous salmonids should be in the upper Cowlitz River Basin when trout are present, due to the short-term presence of most trout. Our trout telemetry results indicated that few hatchery trout survive more than 3 months after release, and mobile tracking during the steelhead spawning period later in the year also failed to identify the presence of tagged trout in steelhead spawning tributaries. These findings align with one of the assumptions of the trout stocking program: that unharvested fish will not interact with natural-origin steelhead (Mike Kohn, Lewis County Public Utility District, personal communication, 2016). However, in other basins, resident rainbow trout have been shown to spawn with steelhead. In the Yakima River Basin in Washington, 7-20 percent of steelhead had resident maternal life histories in two separate years (Courter and others, 2013). McMillan and others (2007) reported mating attempts between female steelhead and wild resident males in the Calawah and Sol Duc river basins, Washington. In the Babine River in British Columbia, Canada, Zimmerman and Reeves (2000) reported mixing of maternal genes between rainbow trout and steelhead life histories. Of concern is the evidence that indicates hatchery fish may negatively impact wild fish through competition (physical contact, displacement;McMichael and others, 1999). Hatchery trout do interact with anadromous steelhead, although there is little evidence of it occurring in the upper Cowlitz River Basin based on our evaluations with tagged trout. However, in 2017 we captured video of a natural-origin male rainbow trout courting a radio-tagged female hatchery-origin steelhead in a tributary of the Cowlitz River (h ttps://www .usgs.gov/ media/ videos/ anadromoussteelhead-and-resident-rainbow-trout-interactions), illustrating the potential for steelhead-rainbow trout progeny. Managers can improve the trout stocking program in the Cowlitz River Basin in several ways. They can continue to use several small releases of fish compared to one large one, allowing multiple periods when fish are available for harvest. Most of the angling is from June to September (Liedtke and others, 2011) and releases in these months will support angling opportunities. Fish released at the Cispus Arm site tend to remain near the release site for a longer period, allowing for greater opportunities of harvest. The falls just upstream from the Cispus Arm site may provide a partial natural barrier for rainbow trout because low numbers of tagged trout were detected upstream. Fish, including steelhead, can ascend the falls, but few trout do, keeping the trout mostly in the lake or the arms where they are less likely to interact with steelhead. While the Cowlitz Arm is convenient for anglers due to the proximity of a boat ramp, more fish released there are prone to entrainment at Cowlitz Falls Dam, making them unavailable for harvest. The risk of entrainment, although reducing angling opportunities, also reduces the risk of trout competing with anadromous fish in the upper Cowlitz River Basin. Entrainment is related to flows and lake drawdowns and is variable between years. Tagged fish moved a greater distance upstream after release in July and August than June 2017, likely seeking cooler water as Lake Scanewa warmed during the summer. Trout released at the Campground are more likely to head upstream or stay in the Cowlitz River, potentially interacting with steelhead. Eliminating the Campground release site, thereby reducing the total number of trout released, and continuing stocking trout at the lake release sites (Cispus and Cowlitz Arms) over a period of several months in the summer may be a strategy to provide continued angler satisfaction and minimize the likelihood of interactions with anadromous salmonid stocks. In summary, the short apparent survival and limited movement of radio-tagged rainbow trout indicate stocked rainbow trout have a low impact to anadromous steelhead, either during spawning periods or during the juvenile outmigration when predation is a concern. However, not all the release locations posed equivalent risk to impacting anadromous fish. Eliminating the Campground release site may reduce the risk of potential interaction of rainbow trout and adult steelhead in the Cowlitz River and still assist in meeting the goals of the trout release program.	2021-08-31T15:18:42.245Z	2021-01-01T00:00:00.000	{ "year": 2021, "sha1": "625fd608fd284c90af40465f2af06b10c0327521", "oa_license": null, "oa_url": "https://pubs.usgs.gov/of/2021/1085/ofr20211085.pdf", "oa_status": "GOLD", "pdf_src": "Anansi", "pdf_hash": "6f1452fd76a2764ba104f22ba301fd6b4380590f", "s2fieldsofstudy": [ "Environmental Science" ], "extfieldsofstudy": [ "Biology" ] }
258294703	pes2o/s2orc	v3-fos-license	Nanomaterial-decorated micromotors for enhanced photoacoustic imaging Micro-and nanorobots have the potential to perform non-invasive drug delivery, sensing, and surgery in living organisms, with the aid of diverse medical imaging techniques. To perform such actions, microrobots require high spatiotemporal resolution tracking with real-time closed-loop feedback. To that end, photoacoustic imaging has appeared as a promising technique for imaging microrobots in deep tissue with higher molecular specificity and contrast. Here, we present different strategies to track magnetically-driven micromotors with improved contrast and specificity using dedicated contrast agents (Au nanorods and nanostars). Furthermore, we discuss the possibility of improving the light absorption properties of the employed nanomaterials considering possible light scattering and coupling to the underlying metal-oxide layers on the micromotor’s surface. For that, 2D COMSOL simulation and experimental results were correlated, confirming that an increased spacing between the Au-nanostructures and the increase of thickness of the underlying oxide layer lead to enhanced light absorption and preservation of the characteristic absorption peak. These characteristics are important when visualizing the micromotors in a complex in vivo environment, to distinguish them from the light absorption properties of the surrounding natural chromophores. Supplementary Information The online version contains supplementary material available at 10.1007/s12213-023-00156-7. Introduction Micromotors can maneuver through the body to perform assigned medical tasks as they possess the capacity of reaching hard-to-access locations [1][2][3].Micromotorassisted targeted drug delivery [4][5][6][7], biopsy [8], blood clot removal [9], or cell transport [10] have shown promising results.However, there are still significant limitations when steering micromotors in living organisms [2,11], in particular when the intended application and micromotor type require high spatiotemporal resolution with precise anatomical positioning.Medical imaging or tracking of such robots in vivo is crucial for achieving effective control in a complex environment.Researchers have implemented numerous imaging techniques for microrobot monitoring, such as ultrasound (US), magnetic resonance imaging (MRI), positron emission tomography-computed tomography (PET-CT), and single-photon emission computed tomography (SPECT) [12][13][14][15][16]. US offers high penetration depth but low signal-to-noise ratio and spatial resolution.US Doppler and phase analysis techniques can further improve the contrast and spatial resolution in an echogenic and dynamic environment [17].MRI has better imaging contrast for soft tissues than other conventional techniques, but its spatiotemporal resolution is insufficient to visualize small robots.CT provides deep tissue penetration but poor temporal resolution and long-term exposure might harm living tissue.PET and SPECT provide high sensitivity and molecular information, but the radiation dose remains the foremost concern.Optical methods including fluorescence [18], reflectionbased IR imaging [19], or optical coherence tomography (OCT) [20], have been used to track microrobots below scattering tissues with good spatiotemporal resolution but with limited penetration depth [21,22].Spatial resolution degrades significantly with depth for optical methods due to pronounced light scattering by tissue. Photoacoustic imaging (PAI) combines the high spatial resolution of optics and the imaging depth of US in tissue.The use of PAI to track micromotors was first suggested by a part of us in 2017 [2].At that time, we visualized in realtime single magnetically-driven micromotors (up to 100 µm long) in 3D, below 1 cm thick chicken tissue [23][24][25].Later, PAI has been employed for guiding capsules containing catalytic micromotors in mice intestines [26], for monitoring swarms of magnetic spiral-like micromotors to treat induced subcutaneous bacterial infection [27], and multifunctional urease-based therapeutic nanorobots inside the bladder of the mouse [28].The tracking of cell-sized magnetic microparticles circulating in the mouse brain towards intravascular applications has also been reported [29].Different from tissue and body fluids, microrobots are often covered with absorbing metal layers, which produce an enhanced PA signal for tracking.A suitable contrast agent can thus further enhance the PA signal for the intended application in deep tissue where the PA signal degrades with depth.Actuation and closed-loop control are other prerequisites to maneuver the microrobot in deep tissue with the required precision to target the disease region. Here, we present the functionalization of Janus magnetic micromotors with gold nanorods (AuNRs) and gold nanostars (AuNSs) to enhance the PA contrast.Such Au nanoparticles have a better capacity to absorb light and provide thermoelastic expansion which features a superior PA response as compared to other nanomaterials.We also present the motion characteristics of micromotors in a parametric study using optical feedback control.Additionally, we show that by preventing plasmon coupling between Aunanoparticles and the underlying metal layer by increasing the thickness of the oxide layer in between them, the characteristic absorption peak of the employed Au nanomaterials was preserved and the contrast was further enhanced, which was also validated by 2D COMSOL simulations. Fabrication of micromotors and characterization of the motion behavior The micromotors were fabricated using drop-casting followed by thin metal layer deposition.First, glass slides (15 × 15 mm 2 ) were sonicated in acetone and isopropanol for 3 min each and subsequently dried with an N 2 gun.The substrates were further plasma-treated to remove impurities and contaminants and to obtain clean and hydrophilic glass surfaces.A monolayer of silicon dioxide (SiO 2 ) particles (⌀ = 100 µm) was then assembled.Briefly, SiO 2 particles were washed threefold with methanol, centrifuged for 1 min to remove the supernatant, and suspended again before usage.Silica particles were mixed thoroughly in methanol and ~ 15 µL of the particle-solvent dispersion was drop-casted on the edge of the cleaned glass slide in a tilted angle, to achieve a homogeneous layer.Microarrays were randomly formed in the direction of solvent evaporation.The resulting monolayer was dried in the air at room temperature.Finally, the samples were half-coated with Ti (10 nm), Fe (50 nm), and Ti (10 nm) by using electron beam physical vapor deposition, at a deposition rate between 0.5-1.0Å/s.Scanning electron microscopy (SEM) was performed after coating the sample with ~ 10 nm Pt to make the specimen conductive and to avoid charging effects during imaging.Figure 1a shows an SEM image of a metal-coated Janus particle before the deposition of Au nanoparticles.Afterward, the microstructures were coated with SiO 2 to facilitate surface functionalization with Au-nanoparticles.The motion behavior of the micromotors was characterized through a parametric study using closed-loop control with optical feedback and real-time magnetic actuation (Fig. 1b).The micromotors were immersed in DI water inside an enclosed channel and then actuated and steered by an external rotating magnetic field.The interaction of the Fe layer with the magnetic field results in torque and a respective change of orientation of the micromotor.Upon application of a rotating magnetic field, it will rotate and translate in the direction perpendicular to the rotational axis through a rolling-like interaction with the surface [30].The steering of micromotors is realized based on real-time imaging feedback: An inverted microscope (Eclipse Ti2-E, Nikon Corp., Japan) with a digital camera (a2A2590-60ucPRO, Basler AG, Germany) captures monochromatic bright-field image frames (2592 × 1944 pixels, 8 bit per pixel) at 20 Hz.Real-time image processing for binarization with a configurable threshold and calculation of the centroid was implemented based on CuPy [31] and performed on a GPU (GeForce RTX 3090, Nvidia, USA).The localized micromotors are linked between consecutive frames using TrackPy [32] running on CPUs (2 × Intel Xeon Gold 5217, Intel, USA).The difference vector to the target location was fed to the position controller, which calculates the magnetic field vector to achieve a rolling motion in that direction as shown in the time-lapse (20 × speed) of closed-loop control of a micromotor (Video S1).The magnetic field of 2 mT and varying rotational frequency was generated with a commercial 8-coil setup (MFG-100-i, Magnebotix AG, Switzerland).The motion planning determines the target location to follow a figure-of-8 shape (Fig. 1c). For the parametric study, the average velocity over one revolution was determined for rotational frequencies of 0.5 Hz to 20 Hz in 0.5 Hz steps, with 3 repetitions.The experiments were performed in a solution of PBS and 0.2% MC, which mimic the rheological properties of a variety of biological fluids, and under varying magnetic field strength (2, 3, 4 mT) (Fig. 1d and e).In all cases, a characteristic behavior emerges for the translational speed in relation to the rotational frequency.A monotonic increase was observed until a step-out frequency was reached, followed by a sharp decline in the speed.Both the step-out-frequency and the maximum speed were found to increase with increasing magnetic field strength.The micromotor's speed reduces by approximately a quarter in MC, due to the higher viscosity and increased drag of 0.2% MC compared to PBS. Synthesis of Au nanoparticles and deposition on micromotors Gold nanostars (AuNSs) were prepared using a surfactantfree method assisted by silver ions [33].~ 14 nm gold seeds (1.5 mL, [Au 0 ] = 0.5 mM) prepared by the Turkevich method [34] were added to an aqueous solution (200 mL) containing hydrogen tetrachloroaurate trihydrate (HAuCl 4 ) (1 mL, 50 mM) and HCl (0.2 mL, 1 M), followed by a fast addition of silver nitrate (AgNO 3 ) (0.6 mL, 10 mM) and ascorbic acid (AA) (1 mL, 100 mM) under vigorous stirring.After 30 s, an aqueous hexadecyltrimethylammonium bromide (CTAB) solution (4 mL, 100 mM) was added to the mixture to increase the colloidal stability of the AuNSs.Upon synthesis, the solution was centrifuged (3500 rpm, 30 min) to remove excess reactants and dispersed in CTAB solution (1 mM).The final gold concentration was 1 mM.The average diameter determined by measuring the dimensions from the transmission electron microscopy (TEM) images was 80 ± 3. Gold nanorods (AuNRs) were prepared using Agassisted seeded growth [35].Gold seeds were synthesized by fast reduction of HAuCl 4 with sodium borohydride (NaBH 4 )in CTAB solution.HAuCl 4 solution (0.025 mL, 50 mM) solution was added to a solution of CTAB (4.7 mL, 100 mM).Afterward, a freshly prepared NaBH 4 (0.3 mL, 10 mM) solution was rapidly injected under vigorous stirring.The solution color changed from yellow to brownish yellow and the stirring was stopped after 2 min.The gold seed solution was aged at room temperature for 30 min before use.To prepare the growth solution, 9.0 g of CTAB and 1.234 g of NaOH were dissolved in 500 mL of warm Milli-Q water (~ 50 ºC) in a 1 L Erlenmeyer flask.Once the sodium oleate was completely dissolved, the mixture was cooled down to 30 ºC and AgNO 3 (24 mL, 4 mM) under stirring.The mixture was kept at 30 ºC for 15 min after which HAuCl 4 was added (2.5 mL, 100 mM) under vigorous stirring.The mixture became colorless after 20 min at 30 ºC and after the introduction of HCl (2.1 mL, 37%).After 15 min of stirring, AA (1.25 mL, 64 mM) was added, and the solution was vigorously stirred for 30 s. Finally, the seed solution (0.8 mL, 0.25 mM) was injected into the growth solution under vigorous stirring for 5 min, and then the solution was left undisturbed at 30 ºC for 12 h.The solution was centrifuged twice (8000 rpm, 30 min) to remove excess reactants and dispersed in an aqueous CTAB solution (1 mM).The final gold concentration was 0.5 mM.The average length and diameter (in nm) determined by measuring the dimensions from the TEM images were 83 ± 5 and 18 ± 1, respectively. Mercapto poly (ethylene glycol) carboxylic acid (PEG) with a molecular weight of 10 kg/mol was used for ligand exchange [36].An aqueous solution of PEG (5 mL) containing 50 molecules/nm 2 was added dropwise to the solution of gold nanoparticles (50 mL, 0.5 mM) under vigorous stirring.The mixture was reacted for about 1 h.PEGmodified gold nanoparticles were centrifuged twice (previous conditions) and finally dispersed in water. TEM images of AuNRs and AuNSs were obtained with a JEOL JEM-1400PLUS transmission electron microscope operating at an acceleration voltage of 120 kV using carboncoated 400 square mesh copper grids as shown in Fig. 2a and b respectively.UV-Vis-NIR optical extinction spectra were recorded using a spectrophotometer in DI water as a medium (Fig. 2c), which featured intense localized surface plasmon resonance (LSPR) peaks at 840 nm for AuNSs and 870 nm for AuNRs.After synthesis, carboxyl-modified AuNRs/AuNSs were immobilized on SiO 2 -coated micromotors using a previously described protocol based on (3-aminopropyl)triethoxysilane (APTES) and carbodiimide chemistry [37][38][39].Briefly, oxygen plasma was applied (1 min) to activate the surface molecules of the deposited oxide, and APTES (2 wt%) solution was prepared with 5 wt% DI water and 93 wt% absolute ethanol beforehand.The micromotor sample was immersed into the APTES solution after plasma activation and incubated for 1 h at room temperature, to obtain amine groups on the micromotor surface.The samples were rinsed with absolute ethanol and phosphate-buffered saline (PBS).Then, N-ethyl-N-3-dimethylaminopropyl carbodiimide hydrochloride (EDC) at a concentration of 10 mM, and active ester compound N-hydroxy-succinimide (NHS) at a concentration of 5 mM, were prepared and used for coupling the carboxyl groups from AuNRs/AuNSs (with a concentration of 300 µg/mL) to the amino groups on the micromotor surface, forming covalent bonds.The samples were soaked for approx. 2 h at room temperature.Afterward, the samples were washed with PBS and DI water.As observed in Fig. 2d, the control sample showed no sign of Au-nanomaterial, while the functionalized samples were successfully labeled with AuNRs and AuNSs on the micromotors' surface, respectively (Fig. 2e and f). Dual US and PA imaging of micromotors Dual US and PA measurements were carried out by using the Vevo-LAZR X (FUJIFILM VisualSonics, The Netherlands) system, a multimodal platform that allows the simultaneous imaging of high-resolution US and PA.US provides anatomical and functional information while PA contributes to the molecular details.The system was equipped with a linear array US transducer at a central frequency of 21 MHz with a depth of 25 mm and fiber optic bundles on either side of the transducer for illumination.The fiber bundle was coupled to a tunable Nd: YAG laser (680 to 970 nm) with a 20 Hz repetition rate and the signals were collected by the 256-element linear array transducer (with an in-plane axial resolution of 75 µm).The pulsed laser generated a wavelength-tunable pulsed beam which was delivered by a bifurcated fiber bundle integrated with the transducer.Both US and PA signals were collected and reconstructed using onboard software.For laser spectral excitation, the PA images were acquired at a wavelength range of 680-970 nm with an increment of 5 nm over the entire scan range.All the measurements were performed in DI water. The imaging experiments were carried out using a dual US and PA setup, as schematically shown in Fig. 3a.A phantom setup was prepared including a water bath and enclosed tubing channel.For the tubing phantom, transparent intravascular polyurethane (IPU) tube (inner diameter ~ 380 µm, outer diameter ~ 840 µm; SAI Infusion Technologies, USA) was mounted in a water bath.The micromotors were inserted into the tube and immersed in the phantom chamber containing DI water for better acoustic coupling.The fluence was set below the Maximum Permissible Exposure (MPE) limit (20 mJ/cm 2 ), followed by the safe exposure guidelines [40]. The measurements were performed with a positionfixed high-frequency transducer to avoid image distortion.PAI relies on multiwavelength excitation and subsequent spectral processing to identify optical signatures of the specific contrast agents.First of all, the NIR spectrum of AuNRs and AuNSs was recorded using PA imaging mode (Fig. 3b).Absorption bands were recorded at 820 nm for AuNSs and 875 nm for AuNRs, in agreement with the UV-Vis-NIR spectra measured by optical spectrophotometry with slight differences.The PAI system was equipped with a NIR pulsed laser and all PA images were acquired over the entire scan range of 680-970 nm, with an increment of 5 nm.Altogether, four samples were prepared in PBS as a medium, including bare SiO 2 particles, halfmetal coated particles, AuNRs-coated, and AuNSs-coated micromotors (Fig. 3c, i-iv).Dual US and PA images of all samples were captured separately and no PA signal was recorded from SiO 2 particles, which exhibited US contrast only (Fig. 3ci).The half-metal coated particles (⌀ = 100 µm) provided both US and PA contrast because they are coated with thin absorbing adjacent metal layers (Ti = 10 nm, Fe = 50 nm, Ti = 10 nm, and SiO 2 = 30 nm) (Fig. 3cii) [41].Yellow arrows in the image indicate the position of a trail of micromotors, from a single to a swarm of them.The Janus micromotors inside the enclosed channel can also make small clusters in the form of dimer or trimer, which lead to enhanced PA signal intensity due to the increased IR light absorption surface.To observe the influence of particle clustering on the PA resulting signal, an additional experiment was performed by placing a single, dimer, and trimer inside the phantom tubing, and as expected there was a good agreement between the PA signal increase to the number of imaged Janus micromotors.(Fig. S1).The samples were further functionalized with AuNRs and AuNSs to enhance the PA signal contrast.Contrast agents with narrow absorption bands in the NIR are a suitable choice of exogenous contrast agents.AuNRs/AuNSs are known to be biocompatible and have successfully been implemented to improve PA contrast. The labeled micromotors were inserted into tubing for imaging and the labeled micromotors exhibited a stronger PA signal, as compared to metal-coated micromotors (Fig. 3c, iii-iv).By plotting the recorded PA data we indeed observed an enhanced PA signal from labeled micromotors (Fig. 3d), which is crucial for deep tissue tracking applications.Although both nanoparticle samples display absorption bands around 820-875 nm, it was not possible to observe such bands upon deposition on the half-coated SiO 2 particles.Both samples provided comparable PA signals over a broad spectral range, meaning that this approach can improve the PA signal of micromotors in hard-to-reach regions.All PA measurements were performed at a gain of 40 dB.The employed micromotors and the here-evaluated multimodal imaging setup are appealing for the supervised drug cargo-delivery towards urinary tract diseases [42], bladder cancer or infection, or towards in vivo assisted fertilization, where similar engineered parts can be used to guide or transport sperm [4,10]. It is worth noting that non-functionalized micromotors (Janus) also absorb light but their resulting PA signals do not exhibit strong absorption signals.The reason is that the micromotors are first coated with Ti and Fe layers for further magnetic manipulation, and such layers also possess plasmon resonances but with broader absorbance spectra.For AuNRs/AuNSs-coated micromotors, there is an increase in the PA signal, as expected.However, due to the small distance between the Au-nanomaterials and the micromotor surface, plasmon coupling results in the broadening and damping of the absorbance band.This effect can be reduced by introducing a transparent layer to IR light during the synthesis of Au-nanomaterials to preserve the optical properties of Au-nanoparticles and their PA response [43]. Effect of increased spacing between AuNRs and micromotor surface To study the interaction and effect of Au-nanoparticles on the absorption signal, we implemented a 2D COMSOL simulation model, where the geometry in the out-of-plane can be regarded as uniformly distributed.We simulated AuNRsfunctionalized to metal surface with varying thicknesses of transparent SiO 2 layer (30 nm and 1 µm) as shown in Fig. 4a and b.SiO 2 is transparent in the visible and NIR, which maintains the optical properties of AuNRs and hence their PA properties.Moreover, SiO 2 coating has been reported to enhance the photoacoustic signal due to its higher thermal conductivity [44].A plane TE-polarized electromagnetic wave is incident on the sample in water as a medium with a spacing distance of 30 nm between AuNRs and the metal layer.The model uses a refractive index of 1.33 for water and 1.5 interpolated wavelength-dependent refractive index for the dielectric and metal layers, which involve the COMSOL material library.The upper boundary defines the incident plane wave and the lower boundary satisfies the scattering boundary condition which absorbs the transmitted plane wave.The side boundary has Floquet conditions, meaning that the solution on one side of the geometry equals the solution on the other side multiplied by a complex-valued phase factor.This effectively turns the model into a section of a geometry that extends indefinitely in the XY plane.Initially, the AuNRs were not introduced, so the background plane wavefield was calculated.We then removed the incident wave and added a perfectly matched layer (for absorbing the incident wave to the boundary) in the four boundaries.Subsequently, AuNRs were included, and the field from the first calculation was set as the background field.We calculated the relative field from the AuNR (which can be regarded as a perturbation).If the field distribution extends to the inner side of the AuNR, it will cause more absorption.We calculated the field from 680 to 970 nm, to derive the absorption spectra, this wavelength range being identical to that in the PAI experiments.The resulting calculated field values show a strong absorption peak at 820 nm for the sample with a 1 µm spacing distance, whereas the sample with a smaller spacing distance showed a weak, broad absorption band (Fig. 4c).These results suggest that a sufficiently thick transparent layer can enhance the optical absorption signal.We, therefore, fabricated Janus microparticles with thin metal layers (Ti/Fe/Ti) and evaporated a thick layer of SiO 2 (1000 nm) using chemical vapor deposition (CVD) before labeling them with AuNRs.SEM imaging was used to demonstrate the labeling of AuNRs on the outer surface of the new Janus samples (Fig. 4d).The recorded Vis-NIR spectrum of AuNRs coated on a 1 µm thick SiO 2 layer shows strong absorption bands at 840 nm using the spectrophotometer (Fig. 4e).In PAI, a strong and broad absorption signal was recorded, peaking at 875 nm (Fig. 4f), in agreement with the PA spectrum of AuNRs only (see Fig. 3b). The thick silica coating reduces plasmon coupling between AuNRs and the metal layer, thereby improving the PA signal strength.The plasmon resonance band shifted up to 20 nm for the spectrophotometer study and 55 nm for PA data, as compared to the 2D simulation.This difference might be due to differences between simulation and experimental settings and the absorption band does not come from absorption only but also scattering.In PAI, it is only possible to detect absorption signatures and this might be the origin of the spectral differences.The simulation was performed with a simple 2D model to generate optical absorption signal, whereas PA takes into account different illumination and US detection mechanism.As a perspective of this work, a 3D simulation that considers different parameters such as AuNPs density, AuNPs shape, and the underlying surface geometry and composition will be realized and compared with the corresponding experimental data to evaluate their effect on the resulting PA signal strength and specificity. Conclusion This work presents the functionalization of Janus micromotors with plasmonic nanomaterials, in particular AuNRs and AuNSs, which are structures with higher photothermal conversion efficiency, to enhance the PA contrast.Such labeled micromotors can provide an improved signal in deep tissue due to the presence of PA agents.As expected, AuNRs/AuNSs-coated motors exhibited enhanced PA signals as compared to bare and thin metal-coated particles.Furthermore, 2D COMSOL simulation and experimental data show that increased spacing between the AuNPs and the underlying metal layer would lead to enhanced absorption spectra which is a crucial parameter while doing imaging of micromotors in deep tissue.We also show the motion behavior of the micromotors through a parametric study using closed-loop control with optical feedback. The average velocity over one revolution is determined for rotational frequencies and a characteristic behavior emerges for the translational speed to the rotational frequency.Such feedback control algorithms can also be implemented using other medical imaging modalities to increase the future targeting efficiency of those microrobots when performing a medical operation in vivo. Fig. 1 Fig. 1 Fabrication and characterization of the motion behavior; a) SEM image of a metal-coated Janus particle for the tracking experiments.b) Block diagram showing the closed-loop control optical planning.c) A single micromotor following a figure-of-8 shape through closed-loop control.The arrows show the direction of Fig. 2 Fig. 2 Fabrication and characterization of Au nanoparticle-coated micromotors; a, b) Representative TEM images of AuNRs (a) and AuNSs (b).c) UV-Vis-NIR spectra of AuNRs and AuNSs in aqueous dispersion.d) Sketch of a half-metal coated Janus particle and Fig. 3 Fig. 3 Dual US and PA imaging; a) Schematic of PA imaging principle with a different type of micromotors.b) PA spectra of AuNSs and AuNRs with absorption bands at 820 nm and 875 nm, respectively.c) Dual US and PA imaging of bare SiO 2 particles (⌀ = 100 µm) (i), Fig. 4 Fig. 4 Simulation and experimental results.a) The layer stacks implemented were simulated using COMSOL to calculate the absorption signal for AuNRs on a metal film.b) An increase in absorption spectra by increasing the gap between metal and AuNRs.c) Simulated spectra of AuNRs (on 30 nm and 1 µm thick SiO 2 layer) showing an enhanced absorption signal at 820 nm.d) Schematic represen-	2023-04-24T15:07:16.317Z	2023-04-22T00:00:00.000	{ "year": 2023, "sha1": "e5dcaba9c375f012ab8bc200107869f5e3037ba6", "oa_license": "CCBY", "oa_url": "https://link.springer.com/content/pdf/10.1007/s12213-023-00156-7.pdf", "oa_status": "HYBRID", "pdf_src": "PubMedCentral", "pdf_hash": "95d7dc4cf98a6c9747138fe277babfa8ba9503ba", "s2fieldsofstudy": [ "Engineering", "Medicine", "Materials Science" ], "extfieldsofstudy": [ "Medicine" ] }
245854151	pes2o/s2orc	v3-fos-license	Epigenetic Mechanisms in Parenchymal Lung Diseases: Bystanders or Therapeutic Targets? Epigenetic responses due to environmental changes alter chromatin structure, which in turn modifies the phenotype, gene expression profile, and activity of each cell type that has a role in the pathophysiology of a disease. Pulmonary diseases are one of the major causes of death in the world, including lung cancer, idiopathic pulmonary fibrosis (IPF), chronic obstructive pulmonary disease (COPD), pulmonary hypertension (PH), lung tuberculosis, pulmonary embolism, and asthma. Several lines of evidence indicate that epigenetic modifications may be one of the main factors to explain the increasing incidence and prevalence of lung diseases including IPF and COPD. Interestingly, isolated fibroblasts and smooth muscle cells from patients with pulmonary diseases such as IPF and PH that were cultured ex vivo maintained the disease phenotype. The cells often show a hyper-proliferative, apoptosis-resistant phenotype with increased expression of extracellular matrix (ECM) and activated focal adhesions suggesting the presence of an epigenetically imprinted phenotype. Moreover, many abnormalities observed in molecular processes in IPF patients are shown to be epigenetically regulated, such as innate immunity, cellular senescence, and apoptotic cell death. DNA methylation, histone modification, and microRNA regulation constitute the most common epigenetic modification mechanisms. Pathogenesis of IPF and COPD Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive disease of the lung with unknown etiology characterized by the excessive extracellular matrix (ECM) formation and remodeling of the lung architecture, and eventually death. According to the World Health Organization (WHO) mortality database, the mean age of the mortality rates from IPF across Europe in 2001-2013 were 3.75 per 100,000 for males and 1.50 per 100,000 for females [1]. The median survival among 65-year-old or older adults with IPF in the United States was 3.8 years [2]. The core pathogenesis of IPF seems to arise from the dynamic interplay between genetic, epigenetic, and environmental factors causing irritation, damage, senescence, and apoptosis of alveolar epithelial cells leading to the production of profibrotic factors responsible for activation of fibroblasts and their trans-differentiation into myofibroblasts [3][4][5][6]. Studies on the pathogenesis of the disease have mostly focused on the mechanisms that regulate proliferation, activation, and differentiation of alphasmooth muscle actin (α-SMA) producing myofibroblasts. However, there is a need for an integrated and comprehensive understanding of the molecular mechanisms of IPF in a cell type-specific manner to develop more efficient therapeutics for this complex disease. IPF is recognized as neoproliferative disorder of the lung and has been reported to have similarities and links to cancer biology. Epigenetic and genetic alterations, abnormal expression of microRNAs, gene mutations, activation of similar signaling pathways, apoptotic resistance, myofibroblast origin and behavior, altered cellular communications, intracellular signaling pathways, and unknown etiology are among various contributing pathogenic mechanisms common in both of these fatal disorders [7,8] (Figure 1). It is worth mentioning that these genetic and epigenetic mechanisms usually lead to aberrant activation of the epithelial cells in the first place, which produce the factors participating in the activation of the fibroblastic/myofibroblastic foci and the activation of a fibrotic response [9]. Moreover, the pathogenesis of other pulmonary diseases such as asthma and chronic obstructive pulmonary disease (COPD) has been already linked to epigenetics [10,11]. Chronic obstructive pulmonary disease (COPD) is a cigarette smoke-and agingrelated progressive disorder, characterized by lung inflammation, emphysema formation, abnormal tissue repair, and alveolar destruction ( Figure 2). Decline in lung function and airflow obstruction as characteristic of COPD is mainly connected to remodeling of small airways [12,13], which eventually leads to destruction of lung parenchyma. Airway wall thickening occurs progressively along with increased severity of the disease in patients with COPD [14]. COPD is the third leading cause of fatality worldwide, and the prevalence continued to increase by 2019, according to WHO records (WHO Global Health Estimates, 2019). To date, molecular characterization related to COPD has dominantly focused on oxidant-antioxidant balance, elastase-anti elastase hypothesis, chronic inflammation, apoptosis, and aberrant repair. Cigarette smoking is the main risk factor as well as other toxic inhalant and gases for developing COPD and inflammation [15,16]. Over 4700 chemical compounds, 73% of which have been identified as carcinogenic by IARC (International Agency for Research on Cancer), and a large number of free radicals activate transcription factors such as Nuclear factor kappa B NF-κB, leading to increased pro-inflammatory cytokine levels in COPD patients. Increased levels of reactive oxygen species (ROS) in lung airways and blood samples is a hallmark of impaired oxidative stress in COPD patients [17][18][19]. The damage due to generated free radicals/oxidants eventually results in alveolar epithelial injury, inactivation of pulmonary surfactants and anti-proteases, membrane lipid peroxidation, remodeling of blood vessels as well as extracellular matrix, dysregulated mitochondrial function, apoptosis, and lastly inflammation [20]. Inflammation itself is a major influencer of the ROS response in the cells [21]. Infiltrating cells such as leukocytes, macrophages, dendritic cells, and natural killer cells damage tissue by activating a persistent immune response due to reduced epithelial barrier integrity. Collectively, dysfunction of alveolar and bronchial epithelial cells, emphysema formation in gas exchange surface area in the lung, and severe tissue remodeling result in progressive airway wall thickening in COPD [22]. Currently, there is no effective clinical treatment available. Over the past two decades, anti-inflammatory inhaled corticosteroids in combination with bronchodilators have been the main therapeutic strategy to improve the health status of patients with COPD, and subsequently their quality of life [11,23]. However, large numbers of COPD patients are resistant to corticosteroids. Since COPD displays different phenotypes, new therapeutic approaches for personalized medicine are urgently requisite. In recent years, epidemiological studies indicate that up to 40% of all COPD patients are smoke-free yet suffer from the disease [24]. As COPD is not a monogenic disease and genetic contribution differs, epigenetic axis to COPD should be considered carefully. Thus, elucidating the epigenetic mechanisms underlying parenchymal remodeling in COPD and IPF are highly relevant to discover the novel treatments and regimes for these deadly diseases. This review summarizes the dysregulation of various epigenetic mechanisms (DNA methylation, histone modifications, and non-coding RNAs) and their impact on pulmonary fibrosis and COPD, as well as the elaborate in vitro and in vivo studies that brought the crucial role of epigenetic mechanisms in disease pathogenesis to light. We also discuss the current preclinical status of pan-and isoform-selective histone deacetylase (HDAC) inhibitors, DNA methylation, and microRNA modulators and propose new research areas that may facilitate locus-specific epigenome editing as a novel therapeutic strategy for IPF and COPD. Epigenetic Dysregulation in IPF and COPD Epigenetics generally refers to heritable changes in gene expression that occur independent of DNA sequence. Epigenetic alterations can be subdivided into three main classes: DNA methylation, post-translational histone modifications, and non-coding RNAs. These epigenetic marks are induced by environmental factors, diet, accumulated somatic mutations over aging, and a variety of diseases [25,26]. Collectively, studies have emerged with evidence showing that epigenetic mechanisms impact on phenotypic changes in chronic lung diseases. Thus, epigenetics enlightens and provides a comprehensive link for genotype-phenotype correlation. DNA Methylation in Lung Fibrosis Methylation of cytosine residues within the CpG islands of gene promoters is known to block RNA polymerase complex and therefore suppresses gene expression. DNA methylation is shown to associate with altered expression of genes and pathways important to lung diseases such as IPF [27,28]. In addition, DNA hypermethylation at promoter regions is considered a crucial contributing factor for downregulation of the IPF suppressor genes such as THY1 (Thy-1 antigen), CAV1 (Caveolin 1), PTEN (Phosphatase and tensin homolog), and PTGER2 (Prostaglandin E receptor 2), which are known to regulate important cellular processes. Moreover, MUC (mucins) are among key proteins involved in the regulation of cell growth and tissue remodeling processes whose genetic and epigenetic deregulation is associated with lung diseases such as IPF [29] and COPD [30]. Interestingly, it was shown that the hypermethylation of the MUC5 promoter region is associated with IPF and that there is a mutation within the same region surrounding the FOXA2 (Forkhead box A2) binding motif. This mutation can change the binding affinity for other transcription factors affecting the expression of MUC5. Genome-wide methylation profiling studies in both lung tissue [27,31] and isolated primary fibroblasts [32] identified extensive DNA methylation changes of patients with IPF compared with controls with substantial effect of these methylation changes on gene expression. DNA methyltransferases (DNMTs) are a family of enzymes responsible for maintaining DNA methylation. Recent studies show that there is an increased expression of DNMTs specifically for DNMT3a and DNMT3b, but no significant changes were observed in levels of DNMT1 in IPF lung tissues [24]. In addition, another study showed that application of TGF-β1 (Transforming growth factor beta 1), a major contributing factor in IPF, increased the protein levels of DNMT1 and DNMT3a in lung fibroblasts without altering their mRNA expression by distinct post-transcriptional mechanisms [33]. Production of DNMT3a was increased by TGF-β1 via an increase in its protein synthesis and translation. By contrast, TGF-β1 inactivates glycogen synthase kinase-3β that causes inhibition of DNMT1 ubiquitination and its proteasomal degradation in lung fibroblasts. These studies suggest that DNA methylation is a crucial factor in the pathogenesis of IPF and that targeting DNMTs should be applied with caution in an isoform-specific and cell-specific manner. DNA Methylation in COPD DNA methylation is commonly linked to gene repression even though its effect is dependent on the location and cell type [34] and regulates the important pathways in COPD [35,36]. Recent studies have shown that a great number of CpG methylation sites are associated with both occupancy of the disease and severity of the symptoms in COPD patients [37]. The change in DNA methylation patterns of the promoters of pro-inflammatory genes has been revealed in alveolar epithelial cells and alveolar macrophages of patients with COPD [38]. DNA methylation is a key factor for developing COPD pathogenesis since this epigenetic mark is largely altered by aging and cigarette smoke through triggering severe inflammatory response, at last leading to the disease development [39]. The connection between DNA methylation pattern and cigarette smoke exposure with or without COPD has widely been analyzed through epigenome-wide association studies (EWASs). The largest EWAS held in 2013 by Zeilinger identified 187 differentially methylated CpG sites between non-smokers and smokers, and the group reported that reduced methylation levels were linked to cigarette smoking actively [40]. The cigarette smoke exposure has been shown to alter methylation pattern and in the context of the termination of tobacco consuming allows restoration to the methylation pattern of the non-smokers [41]. In addition, aberrant DNA methylation status of GATA4 (GATA binding protein 4) and p16 promoters obtained from sputum samples has been connected with decline in lung function in COPD [42]. Another study using small airway epithelial cells discovered the differences of 1260 methylated CpG sites related to COPD [35]. Interestingly, since one result of a study clearly showed that cigarette smoke-related changes in DNA methylation is reversible after quitting smoking, it signifies that DNA methylation might be a useful biomarker for COPD [43]. Smoking-related differentially methylated genes in peripheral blood cells are known to result in the development of COPD and gradual decline in lung function [44,45]. Even though there is heterogeneity in the findings obtained from various studies, some epigenetic players have been identified so far. For instance, hypomethylation in SERPINA1 (Serpin family A member 1) gene at two CpG sites (cg02181506 and cg24621042 on chromosome 14) was linked to the COPD and smoker group, as well as AHRR (Aryl-hydrocarbon receptor repressor) gene hypomethylation in intron 3 (cg21161138) [43,46]. In contrast with this, GABRB1 (Gamma-aminobutyric acid type a receptor beta-1 subunit) (cg15393297), NOS1AP (Nitric oxide synthase 1 adaptor protein) (cg26663636), and TNFAIP2 (TNF-alpha-induced protein 2) (cg18620571) genes were reported to be hypermethylated in gene bodies of smokers as well as COPD when compared with a healthy group [47]. SULF2 (Sulfatase 2), lung cancer associate gene, has been shown to have increased total methylation status in sputum samples of ex-smokers showing lasting symptoms of COPD, which is chronic mucous hypersecretion (CMH) [48]. On the other hand, Armstrong et al. have reported that DNA methylation profile changes of CLIP4 (CAP-Gly domain-containing linker protein family member 4) (cg26118047), HSH2D (Hematopoietic SH2 domain-containing) in 5 -UTR region and SNX10 (Sorting nexin 10) gene in 5 -UTR region are associated with metabolic differences in lung macrophages [49]. In another study, hypermethylation of mtTFA (Mitochondrial transcription factor A) promoter is demonstrated to be triggered via elevated cigarette smoke and to result in COPD development [50]. Interestingly, the researchers point out that there is a strong correlation between aberrant DNA methylation signals and overexpression of DNA methyltransferases (DNMT1, DNMT3a, and DNMT3b). Liu et al. have reported that after exposure to cigarette smoke condensates for up to 9 months, DNMT3b expression significantly upregulated in respiratory epithelial cells, including small airway epithelial cells and HBECs (human bronchial epithelial cells), while DNMT1 is reduced [51]. The DNA methylation pattern of transcription factors plays a critical role in goblet cell differentiation as well. In this context, Song et al. identified the hypomethylation of FOXA2 (Forkhead box protein a2) and SPDEF (SAM-pointed domain-containing ETS-like factor) promoters, which are involved in inhibition of goblet cell differentiation and mucus production, respectively [52]. In a COPD small airway study group, Vucic et al. identified hypermethylation of genes involved in metabolism, glutathione S-transferases such as GSTT1, GSTM1, GSTP1) and cholinergic receptors such as CHRNB1, CHRNB2, CHRND [35]. In addition, a great number of genes encoding ECM proteins were differentially methylated in airway lung fibroblasts of COPD patients. Clifford et al. reported that several genes including WNT3a (WNT family member 3a), TMEM44 (Transmembrane protein 44), and HLA-DP1 (Major histocompatibility complex DR beta 5) were differentially methylated among 652 loci [53]. There are various studies indicating a biological link between COPD and lung cancer. One study among them demonstrated that hypermethylation of WIF-1 (WNT inhibitory factor 1) and IL-12Rbeta2 (Interleukin 12 receptor, beta 2 subunit) promoters commonly arise during the transition of COPD and add the risk for lung cancer development [54]. In addition to these findings, air pollution is also shown to alter DNA methylation. For instance, SYTL2 (Synaptotagmin-like 2) (cg11691844), WDR46 (WD repeat domain 46) (cg05454562), AKNA (AT-Hook transcription factor) (cg13999433) and NEGR1 (neuronal growth regulator 1) (cg07721244) are differently methylated in their CpG sites due to air pollution [37,55]. Lastly, hypomethylation of HDAC6 (Histone deacetylase 6) is reported, which leads to HDAC activity and causes epithelial dysfunction [56]. Overall, these findings suggest that altered DNA methylation is involved in cigarette smoke related to COPD pathogenesis, and further investigations need to be conducted for a better understanding the underlying mechanisms. Histone Modifications in Lung Fibrosis Aside from DNA methylation, post-translational modification (PTM) of histones plays an important role in most biological processes that impact gene expression by changing chromatin structure. This process is often linked to the development, differentiation, and pathogenesis of many diseases and disorders, including pulmonary diseases. The two most common histone modifications are namely methylation and acetylation. Methylation of histones can associate with either transcriptional activation such as H3K4 [57] and H3K36 [58] or transcriptional repression on H3K9 [59], H4K20 [60], and H3K27 [61,62]. Histone methylation is regulated by a dynamic interplay between two sets of enzymes: histone methyltransferases (HMTs) and histone demethylases (HDMs). Histone methyltransferases (HMTs) add a methyl group on the side chains of lysines and arginines of the H3 and H4 histones as well as non-histone proteins. While lysines are mentioned to be mono-, di-, or tri-methylated, arginines on the other hand are suggested to be mono-, symmetrically or asymmetrically di-methylated. In contrast, histone dimethyltransferases (HDM) are capable of removing methyl groups from both histones and other proteins. The disruption of a balance between the opposing activities of HDMs and HMTs contributes to the developmental defects and tumorigenesis observed in various organs [63]. The histone acetylation signature of a cell plays an important role in the modulation of chromatin structure and gene expression. This dynamic process is regulated by the balance between histone acetyltransferase (HATs) and histone deacetylase (HDAC) activities. Histone acetyltransferases (HATs) acetylate lysine amino acid on histone tails, which promotes open chromatin structure and hence leads to increased gene expression. Conversely, histone deacetylases (HDACs) remove acetyl groups from histone tails. This causes the histones to wrap the DNA more tightly, making it less permissive for transcription factors to bind to it, which results in decreased gene expression and transcriptional repression. Among HATs, p300 is perhaps the most widely studied protein that is associated with the transcriptional activation of numerous genes in response to cellular signaling. The increased activity and expression of p300 were shown to associate with different diseases, including pulmonary fibrosis [64] and acute respiratory distress syndrome [65], and reported in different types of cancers [65][66][67][68]. Moreover, the genetic deficiency and pharmacological inhibition of p300 were shown to abrogate pulmonary fibrosis both in vitro and in vivo [64,69]. The role of HATs is opposed by HDACs. In humans, there are 18 HDACs that are grouped into five different classes according to phylogenetic and sequence homology: class I, class IIa, class IIb, class IV HDACs, which are zinc-dependent, and class III or sirtuins [70]. Class I HDACs (including HDAC 1, 2, 3, and 8) contain a deacetylase domain and are primarily localized in the nucleus of cells. Members of this class are ubiquitously expressed throughout various developmental stages and different cell types. Class II HDACs on the other hand have tissue-and stage-specific expression and can shuttle between cytoplasm and nucleus in response to various regulatory signals [71,72]. Members of Class II HDACs are divided into two subclasses: subclass IIa includes HDAC4, 5, 7, and 9 and subclass IIb consists of HDAC6 and 10 based on their primary structure. Class IIa HDACs consists of a large N-terminal regulatory domain that is required for protein-protein interactions and has important roles in the recruitment of various cofactors in addition to a C-terminal catalytic domain. Class III HDACs or sirtuins (including all sirtuins from 1 to 7) and class IV (HDAC11) display enzymatic activity to a certain extent [73]. HDACs are involved in the deacetylation of not only chromatin proteins, which can lead to altered gene-transcription regulation, but also of various non-histone proteins, regulating their function, stability, cellular localization, and/or protein-protein interaction [74,75]. Moreover, HDACs are known to modulate the expression of a large number of genes in different ways. For instance, they can form corepressor complexes with the nuclear receptor in the absence of a ligand. Studies show that HDACs can also directly interact with various transcription factors such as oncosuppressor protein p53, Stat3, GATA2, E2f, HIF1α, the retinoblastoma protein, NF-κB, Hsp90, and TFIIE [75][76][77]. Many HDACs can associate with multiprotein corepressor complexes, such as the transcriptional corepressors N-CoR, mSin3, and SMRT. These multiprotein complexes can interact with other proteins such as nuclear receptors, transcription factors, and other epigenetic gene modifiers, such as histone methyltransferases (HMTs), DNA methyltransferases (DNMTs), and methyl-CpG-binding proteins (MBDs) to regulate gene expression [78]. HDACs are known to regulate different cellular processes and are closely linked with tumorigenic features such as proliferation, metastasis, differentiation, and apoptosisresistance phenotype of cells. Moreover, the aberrant expression or activity of HDACs is usually associated with human cancers and poor prognosis [79][80][81][82][83][84][85][86]. Aberrant HDAC activities are also observed in fibrotic diseases including renal [87], cardiac [88], liver [89], and pulmonary fibrosis [90]. In each study the specific mechanism of HDACs is different; however, based on the studies, the evidence suggests that HDACs usually trigger fibrogenesis in various ways and that increased expression of HDACs stimulates fibroblast to myofibroblasts trans-differentiation [91,92]. A wide-scale study examining the expression of HDACs in IPF demonstrated that nearly all class I and II HDAC enzymes are upregulated in IPF lung tissue. Moreover, upregulation of HDACs was predominantly observed in myofibroblasts of fibrotic foci region as well as in abnormal bronchiolar basal cells in areas of bronchiolization in IPF lungs [93]. Furthermore, they showed that HDACi such as LBH589 may be useful for the treatment of IPF, interfering with fibroblast to myofibroblasts differentiation, and more importantly leading to the downregulation of ACTA2 and ECM genes such as COL1A1, COL3A1, and FN in primary IPF fibroblasts. Histone Modifications in COPD Several studies demonstrated that the decrease in HDAC activity results in upregulated transcription status of pro-inflammatory cytokine genes, and in the case of HDAC overexpression, lung neoplasia is promoted as the transcription of tumor suppressor genes is decreased. Since histone acetylation and deacetylation are critical regulators of chromatin structure and gene expression, imbalance in between promotes susceptibility to the development lung diseases, including COPD and carcinogenesis in smokers [94,95]. Almost two decades ago, the role of HATs (CBP, GCN5, p300, and PCAF) was suggested in COPD pathogenesis, and it is demonstrated that HAT inhibitors targeting these molecules could be used in clinical applications as a therapeutic aspect [96]. In another study, Adenuga et al. identified that HDAC2 expression and enzyme activity in specimens of COPD patients were significantly reduced, which is thus correlated with excessive inflammation [97]. Decreased HDAC2 activity is associated with glucocorticoid resistance and elevated oxidative stress as well as with pro-inflammatory mediators in smoke-induced COPD [98]. Under normal conditions, HDAC2 plays a role in T cell differentiation and IL-17 production. Lai et al. reported that HDAC2 activity inhibits airway remodeling in lung tissue samples of COPD patients by repressing IL17A production. Hereby, this study confirmed that there is a strong correlation between the dysregulation of HDAC2/IL17A axis and bronchial wall thickening as well as collagen deposition [99]. A similar study using curcumin, an anti-inflammatory drug, in order to modulate HDAC2 expression indicated that inflammation is resolved and corticosteroid resistance is recruited in AECII (alveolar epithelial type II cells) in a COPD rat model [100]. One of the interesting effects of HDAC dysregulation is that it is capable of stimulating pathological responses. Ding et al. noted that the expressions of HDAC1 and HDAC2 were upregulated in mice exposed to cigarette smoke and led to skeletal muscle atrophy. The group also reported that the use of TSA (Trichostatin A), a global HDAC inhibitor, reserved skeletal muscle atrophy by inhibiting HDAC1 and 2 activities [101]. There is another study supporting the evidence that selective HDAC1-3 inhibitor MS-275 contributed to anti-inflammatory effect in a cigarette smoke mice model via stimulating IL-10 production [102]. One of the HDAC class enzymes, SIRT1 (silent inflammation regulator 1), is known to play role in inflammation, aging, cell senescence, and emphysema formation in COPD [103]. Ma et al. demonstrated that upregulated SIRT1 expression controlled NF-κB acetylation and repressed inflammatory responses through erythromycin [104]. Vincenzo et al. also noted that deficiency of SIRT1-FOXO3 interaction resulted in aberrant inflammatory responses in bronchial epithelial cells [105]. Apart from histone acetylation/deacetylation, histone methylation in COPD can regulate gene transcription based on the modified lysine and arginine residues, where the formation could be mono-, di-, or tri-methylated [106]. H3K4me3 is an important histone marker, linked with transcription activation [107]. Similar to DNA methylation, the H3K4me3 epigenetic marker plays a critical role in mammalian development [108]. Alteration of this marker is already associated with cancer and other diseases [109][110][111]. Since the histone methylation concept in COPD is a more complicated process than acetylation/deacetylation, this research field is still narrow. However, Yildirim et al. indicated that elevated mRNA and protein expression of PRMT2, a protein arginase methyltransferase, was a stimulus for COPD development in a hypoxia-exposed mice model [112]. Moreover, Andresen et al. noted that increased mRNA level of DEFB1 (Beta defensin 1) is correlated with H3K4me3 in progression of COPD [113]. Recently, He et al. demonstrated that PRMT6, a protein arginase methyltransferase induced by H3K4me3, regulates NF-κB activation negatively [114]. Taken together, histone modifications seem to have a crucial role in the development and progression of COPD and IPF. MicroRNAs in IPF MicroRNAs (miRNAs) are a class of small non-coding RNAs and key epigenetic regulators that can bind to the 3 untranslated region of mRNA targets and mediate their degradation. Several studies have mentioned the role of miRNA in lung development and pulmonary diseases, such as asthma, cystic fibrosis, chronic obstructive pulmonary disease, and pulmonary artery hypertension [115]. Moreover, miRNAs have been associated with almost every stage in the pathogenesis of IPF including lung epithelial repair; epithelial-mesenchymal transition (EMT) such as let-7d, miR-200, miR-26a, and miR-375; activation of fibroblast and their trans differentiation to myofibroblasts such as miR-21, miR-155, miR-26a, miR-27a-3p, miR-9-5p; AECII cell senescence; and regulation of collagen production such as miR-320a [116][117][118]. MicroRNAs are both upregulated and downregulated during the pathogenesis of IPF and were shown to have both pro-fibrotic and anti-fibrotic roles in the pathogenesis of the disease [119]. Interestingly a study identified 47 significantly differentially expressed serum miRNAs from IPF patients compared with healthy controls including 21 upregulated miRNAs and 26 downregulated miRNAs [120]. Moreover, these miRNAs were shown to regulate important biological processes that are known in the pathogenesis of the disease, such as TGF-β signaling, MAPK signaling, PI3K-Akt signaling, Wnt signaling, HIF-1 signaling, Jak-STAT signaling, Notch signaling pathway, and regulation of actin cytoskeleton. Finally, as the application of miRNAs as novel diagnostic and therapeutic tools in lung diseases has become more and more attractive, it still faces significant challenges, including non-specific targets, specific delivery to the targeted cells, activation of the innate and adaptive immune responses, and possible cytotoxicity. MicroRNAs in COPD As a research field, microRNAs (miRNAs) in COPD have been an exciting topic to explore for a better understanding of COPD pathogenesis so far. A growing number of studies indicate that miRNAs are involved in many pathways, such as tissue repair and inflammation, that have important roles in emphysema and COPD pathogenesis. The dysregulated miRNA expression has been observed in lung samples of COPD patients when compared with smokers as a control group [121]. High throughput platforms have enabled the investigation of 70 differentially expressed miRNAs in lung tissue samples of smokers with or without airflow limitations. According to this research study, miR-422a, miR-923, and miR-937 significantly downregulated in COPD patients, while miR-144, miR-223, and miR-1274a upregulated [121]. In another study, miR-203 was found to be downregulated in lung tissue while being upregulated in blood samples of COPD patients when compared with non-smokers [122]. Further, 56 differentially expressed miRNAs were identified in patient blood samples of smoke-induced COPD. Furthermore, it was demonstrated that the expression level of miR-26a-5p, miR-149-3p, miR-451b, and miR-3202 were significantly downregulated in smokers with or without COPD when compared with the non-smoker group. In addition, miR-149-3p was highlighted to control the NF-κB pathway by regulating TLR4 (Toll like receptor 4) response in THP-1 cells in a murine monocytic cell line [123]. In another study, TLR4 mRNA was shown to be regulated by miR-1236, which leads to risk of development of VAP (ventilator-associated pneumonia) in COPD [124]. In another study, Pottelberge et al. noted that miR-125b and let-7c in sputum supernatant were reduced in COPD patients when compared with the non-smoker control group. Interestingly, the group discovered that let-7c was inversely linked with its target TNFRII (tumor necrosis factor receptor type II) in the sputum of patients [125]. Some studies have demonstrated that the Notch signaling pathway has a critical role in COPD since Notch3 (Notch receptor 3) expression was shown to be inhibited by miR-206 in human pulmonary microvascular endothelial cells of smokers with COPD [126]. Long et al. revealed that miR-34a inhibits Notch1 (Notch receptor 1) gene expression in endothelial cells exposed to chronic smoke [127]. A number of papers noted that there is a strong correlation between COPD and lung cancer. Currently, COPD is suggested to be a driver of lung cancer since many lung cancerrelated death cases were associated with smoking [128]. Chronic exposure to cigarette smoke is known to cause oxidative stress through free radicals and epithelial-mesenchymal transition (EMT), these two changes of which are highly present in COPD as well as in lung cancer pathogenesis. For instance, miR-200 is found to suppress tumor progression via binding to ZEB1 (Zinc finger E-box-binding homeobox 1) and E-cadherin, eventually inhibiting the EMT process. The regulation of miR-200 expression is also linked with metastasis in lung cancer [129,130]. In addition, several miRNAs are identified to be involved in emphysema pathogenesis. Christenson et al. demonstrated alteration of 63 differentially expressed miRNAs such as miR-181d, miR-30c, and miR-638 in regional emphysema of COPD patients through comparing distinct locations of varying grades of emphysema. These miRNAs target corresponding genes associated with oxidative stress and progressive aging in emphysema [131]. Moreover, Savarimuthu et al. reported that miR-34c targets SERPINE1, a protease inhibitor, in emphysema via altering protease/anti-protease balance [132]. Strikingly, there is another emerging research field in COPD pathogenesis related to miRNAs, which is their transportation within EVs (extracellular vesicles). EVs are lipid bilayer structures released by all cell types and present in nearly all body fluids including BALF (broncho alveolar lavage fluid). EVs can be classified based on their size and origins. According to these criteria, exosomes are >100 nm while microvesicles are larger than 100 nm. EVs are responsible for cell-to-cell communication and cell homeostasis by transporting nucleic acids including miRNAs, proteins, lipids, and other bioactive molecules [133]. Since EVs are shown to contain epigenetic marks such as DNMTs and HDACs, it is most likely that they are involved in epigenetic regulation of lung diseases as well [134,135]. The exosomes released from BECs (broncho epithelial cells) exposed to chronic exposure are shown to bear extracellular matrix-associated CCN-1 (Cellular communication factor 1) protein, which induces MMP-1 (Matrix metalloproteinase 1) secretion leading to emphysema formation eventually [136]. Similarly, AMs (alveolar macrophages) release EV and MV when exposed to chronic cigarette smoke in order to upregulate pro-inflammatory mediators such as IL-8 (Interleukin 8), ICAM-1 (Intercellular adhesion molecule 1), and MCP-1 (Monocyte chemoattractant protein 1) in AECs [137]. Another study reported that EVs bearing miR-210 derived from COPD patients were found to silence ATG7 (Autophagy related 7), autophagy-related factor, during fibroblast to myofibroblast differentiation [138]. Finally, MVs released from endothelial cells after cigarette smoke were identified to be enriched by miR-125a, miR-126, miR-191, and let-7d. Once these miRNAs were transported as cargo by MVs, they influenced efferocytosis of recipient macrophages [139]. These results suggest that EVs have a crucial regulatory role in COPD. Thus, transportation of small RNAs through EVs might be a novel mechanism that could result in spreading altered bioactive molecules and epigenetic marks from a single cell to another. Further investigations are needed to elucidate the underlying mechanism in COPD. Conventional Medications and New Strategies for IPF and COPD There is no effective cure for IPF and COPD so far. However, medication and other treatment options can help improve patients' quality of life. A number of comorbidities are often linked to IPF, including lung cancer, pulmonary hypertension, emphysema, depression, cardiovascular disease, thrombosis, acute respiratory distress syndrome (ARDS), and respiratory failure, which causes further difficulties or delays in diagnosing and treating IPF [140][141][142][143][144][145][146][147]. Currently, there are two drugs approved by the FDA (U.S. Food and Drug Administartion) for the treatment of idiopathic pulmonary fibrosis (IPF). These include nintedanib and pirfenidone. However, as these agents do not show curative effects [148], new therapeutic approaches for patients are required. On the other hand, anti-inflammatory inhaled corticosteroids in combination with bronchodilators have been in use for patients with COPD [11,23] (Figure 3). However, large numbers of COPD patients are resistant to corticosteroids, as mentioned earlier. To improve the current therapeutic options, a considerable shift in efforts towards the identification of signaling mechanisms involved in the pathogenesis of IPF and COPD was observed. Since then, perturbations of several molecular mechanisms, including pathways involving growth factors, cytokines, metabolic signaling, and transcription factors and epigenome that may underlie the pathogenesis of the disease, have been described. Recently, HDACs have gained increasing attention, as HDAC-inhibiting compounds were shown to correct abnormalities in various cell process including proliferation, migration, vascularization, and death. Although many researchers have shown the valuable effects of HDAC inhibitors against multiple human diseases, they can simultaneously induce acetylation of histones as well as non-histone proteins involved in regulation of gene expression and in various cellular pathways. Thus, they can contribute to toxic and fatal side effects. Targeting HDACs in an isoform-specific manner could achieve enhanced clinical utility by reducing or eliminating the serious side effects associated with current first-generation non-selective HDAC inhibitors [149,150]. Similarly, conventional DNMT inhibitors (DNMTIs) lack specificity for gene(s) of interest and were shown to induce the demethylation of not only tumor suppressor genes but also oncogenes [151]. Thus, they are not capable of specifically reversing the aberrant DNA methylation regulation involved in the pathology of lung diseases. The current therapies with epigenetic modifying drugs (epi-therapies) to treat pulmonary diseases are still in their infancy. However, comprehensive and integrated studies are required to delineate specific epigenetic events and their consequences in each cell type and each lung disease. Recently, a new therapeutic era has emerged in the COPD field: so-called epigenetic editing. Epigenetic editing is a valuable tool for writing or erasing epigenetic modifications at the target gene of interest in order to modulate its expression. The epigenetic editing tool is a combination of an epigenetic effector domain with a DNA-binding domain coupled to it. The epigenetic effector domain is a catalytic domain of epigenetic enzyme that alters the epigenetic status of the target locus via writing or erasing the modifications. The DNA-binding domain coupled to an epigenetic effector domain recognizes the specific sequence at the desired gene. Here, the ultimate goal is to upregulate or downregulate the expression of the desired genes [152]. So far, numerous epigenetic modifier systems have been employed to alter gene expression via introducing epigenetic editors to the cells with various transfer methods. However, lung delivery of epigenetic modifiers has its own limitations in terms of successful applications. Since lung has both immune and physical barriers to keep the environment pathogen-free, delivering these epigenetic modifiers to the cells is a major challenge. The mentioned barriers consist of tight junctions among epithelial cells, alveolar macrophages, which are responsible for taking up and clearance of infectious and toxic particles, and lastly mucociliary escalator [153]. The other challenge is the broad actions of these epigenetic editors, as they mostly have multiple targets, which consequently might cause deleterious side effects. Targeting and modifying gene expression at the DNA level is more advantageous when compared with targeting RNA or protein. In order to inhibit RNA or protein effectively, continuous administration of epigenetic tools is necessary, while targeting DNA is to silence the source of expression directly. In addition, potential splicing isoforms of RNA and translated distinct proteins could be another compelling point to consider in applications for lung diseases. In the matter of upregulation of gene expression at the DNA level, this leads to increased gene expression due to all possible isoforms in natural ratios. In fact, ectopic cDNA delivery leads to over-expression of only the desired isoform of the gene of interest. For this purpose, ATFs (artificial transcription factors) have been used to modify genes at the DNA level. ATFs include DNA-binding domains coupled to transcription effector domains, which can be a transcription activator domain such as VP16 fused to VP64 (Herpes simplex virus protein VP16 and its tetramer VP64) or transcription repressor domains such as SKD (Super KRAB domain). TFO (triplex forming helix), ZFPs (zinc finger proteins), TALEs (transcription activator-like domains), and CRISPRs (clustered regularly interspaced short palindromic repeats) have been developed to target and modify the desired DNA sequences [154]. Up to now, a limited number of research studies have been reported regarding DNA targeting system in COPD. In order to study the roles of genes linked to inflammation, surfactant production, and epithelial cell senescence in the pathogenesis of COPD, knock-out cell and animal lines have been created by CRISPR-editing. For instance, Chu et al. demonstrated the role of MUC18 was found to be pro-inflammatory in CRISPR-edited primary human airway epithelial cells [155]. In another study, Zhang et al. studied pulmonary surfactant synthesis in CRISPR-mediated miRNA-26a-1/miRNA 26a-2 double knockout mice [156]. Moreover, GDF15 (Growth differentiation factor 15) production was shown to be involved in cigarette smoke-induced epithelial cell senescence by a CRISPR-mediated knock-out strategy [157]. The airway mucus hypersecretion was targeted by epigenetic editing since it contributes to COPD pathogenesis. Song et al. revealed that ZFPs and CRISPR-mediated gene silencing of SPDEF (SAM-pointed domain-containing Ets-like factor) reduced excessive mucus secretion in lung epithelial cells, suggesting a therapeutic strategy for COPD patients [158]. The researchers utilized a system including the addition of histone and DNA methylation as well as transcription repressors to silence the promoter of SPDEF gene. Gene transfer with non-viral and viral approaches has been investigated broadly in the respiratory diseases. Particularly, viral vectors are reported to be quite efficient, even in the clinical practice setting since they have a natural tropism to the respiratory system [159,160]. Alton et al. noted that F/HN-pseudo typed SIV vector was shown to produce gene expression in the lungs during their lifetime. Based on these promising data, the group built up the rSIV.F/HN vector into a first-in-man CF (cystic fibrosis) clinical trial [161]. However, there are a limited number of studies reported in the COPD aspect. For the non-viral approaches, commercially available liposomes such as lipofectamine, PEG (polyethylene glycol), GL67A) have been employed as a strategy for delivering a gene into a target cell, as well as lipid nanoparticles. Mastorakos and coworkers produced highly stable and non-toxic polymers, so called BPAEs (β-amino esters), in order to treat the mucosal layer above the respiratory epithelium [162]. In other studies, PBAE-MMPs (PBAE-based mucus-penetrating DNA nanoparticles) were highlighted to have a certain extent of success regarding transfection efficiency and long-term effect after repetitive administration [163]. Additionally, Mahiny et al. provided a successful example of gene editing in surfactant B (SP-B)-deficient mice via using nuclease-encoding mRNA coupled with chitosan-coated nanoparticles, a safe polymer [164]. Chitosan nanoparticles were also reported to be used as a protein delivery system [165]. Since there is a strong correlation between distinct miRNA signature patterns and disease progression in COPD, miRNA profiling has been studied widely. In one study, miR-146a incorporated with NCMPs (nanocomposite microparticles) was successfully delivered to reduce IRAK1 (IL-1 receptor-associated kinase) and TRAF6 (TNF receptor-associated factor 6) gene expressions for COPD treatment purposes [166]. Cadmium, one of the components of cigarette smoke, is known to trigger inflammation and contribute to COPD pathogenesis. Interestingly, miR-181a-2-3p expression was attenuated while inflammatory response was elevated in cadmium (Cd)-treated human bronchial epithelial cells [167]. Additionally, miR-197 expression was correlated with the contractile phenotype in SMCs (smooth muscle cells). The researchers showed that miR-197 inhibition provoked migration while preventing the acquisition of SMC contractile markers [168]. Clearly, these studies point out that some miRNAs might be considered as biomarkers and a crucial target for a therapeutic approach in COPD. For instance, downregulation of miR-3177-3p and miR-183-5p in peripheral leukocytes is an important biomarker for COPD [169]. Due to these promising studies, epigenetic editing opens a novel and exciting therapeutic avenue in lung disease therapy. Conflicts of Interest: The authors declare no conflict of interest.	2022-01-09T16:12:35.969Z	2022-01-01T00:00:00.000	{ "year": 2022, "sha1": "d5a8e164ef8930d2fb6a6dce2b163dd540c78c91", "oa_license": "CCBY", "oa_url": null, "oa_status": null, "pdf_src": "PubMedCentral", "pdf_hash": "b9fb41c125ce027c5c7cc7806cee44fd6057324d", "s2fieldsofstudy": [ "Medicine", "Environmental Science" ], "extfieldsofstudy": [ "Medicine" ] }

Subsets and Splits

No community queries yet

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