text stringlengths 1 1.05M | meta dict |
|---|---|
Introduction {#S1}
============
Epilepsy is one of the most prevalent and severe chronic neurological disorders, afflicting approximately 50 million people worldwide. It often features the generation of spontaneous and recurrent seizures in the affected brain area including the hippocampus ([@B9]; [@B20]; [@B19]), finally disrupting proper brain function. Nowadays, more than 20 anti-seizure drugs (ASDs) have been approved for the treatment of epilepsy. Nevertheless, nearly 30% of patients fail to achieve seizure control ([@B15]; [@B8]). Additionally, the traditional ASDs show various and serious adverse reactions as a result of their action on ubiquitously distributed targets that are involved in physical processes ([@B25]). This has spurred the identification of alternative targets to develop satisfactory anti-seizure therapies.
Ferroptosis, a recently discovered regulated cell death (RCD) which can be manipulated pharmacologically and genetically and under the control of intrinsic molecular mechanism ([@B27]), is characterized with iron-dependent lipid peroxidation ([@B3]; [@B18]; [@B21]). It is distinct from other cell death modalities including apoptosis, necroptosis and autophagy, at morphological, biochemical, and genetic levels. Specifically, mitochondrial shrinkage and a condensed outer membrane are the features of ferroptosis ([@B3]). It has been extensively reported that ferroptosis is involved in the etiology of diverse neurological disorders such as Alzheimer's disease, Parkinson's disease, stroke, and traumatic brain injury ([@B5]; [@B28]; [@B31]; [@B35]). However, the role of the ferroptosis process in the seizure genesis remains unclear, especially in pentylenetetrazole (PTZ) kindling and pilocarpine (Pilo)-induced seizures. In our present study, we provided the direct evidence for the occurrence of ferroptosis in murine models of PTZ kindling and Pilo-induced seizures. We found that treatment with ferroptosis inhibitor ferrostatin-1 (Fer-1) potently alleviated seizure severity and frequency.
Materials and Methods {#S2}
=====================
Animals and Ethics Statement {#S2.SS1}
----------------------------
Male C57BL/6J mice (8--10 weeks old) were provided from the Animal Centre of Central South University and maintained in a constant environment (24 ± 2°C, 12 h light/12 h dark, 50--70% humidity) with *ad libitum* access to standard food and water. All animal care and procedures throughout the study were approved by the Ethical Committee of the Animal Centre of Central South University.
Establishments of PTZ Kindling and Pilo-Induced Seizure Models {#S2.SS2}
--------------------------------------------------------------
### PTZ Kindling {#S2.SS2.SSS1}
The C57BL/6J mice were intraperitoneally injected with PTZ (35 mg/kg, Sigma-Aldrich, United States) once every other day for eleven injections ([@B23]) and mice exhibiting more than three consecutive stage 4 seizures were considered to be kindled. Behavioral seizures were analyzed for the subsequent 90 min after the last injection of PTZ according to a modified Racing scale ([@B14]; [@B22]): stage 0, no response; stage 1, immobility; stage 2, rigidity; stage 3, head bobbing and circling; stage 4, intermittent rearing and falling; stage 5, continuous rearing and falling; and stage 6, tonic-clonic convulsions and rapid jumping. Animals that died were assigned stage 6 during the experiments. Ferrostatin-1 (Fer-1, Selleck, United States) were administered intraperitoneally at the 8th day after PTZ injection with a dose of 2.5 μmol/kg for 2 consecutive weeks. The dosage of Fer-1 was selected according to a previous report ([@B29]).
### Pilo Model {#S2.SS2.SSS2}
The ramping-up dosing protocol was selected for the preparation of the Pilo-induced seizure model ([@B24]). In brief, the C57BL/6J mice underwent repeated low-dose treatment by intraperitoneal application of Pilo (100 mg/kg, Sigma-Aldrich, United States) every 20 min until onset of limbic seizure. Usually, three injections are sufficient for the induction of continuous seizure activity. Methylscopolamine (1 mg/kg, Sigma-Aldrich, United States) was injected intraperitoneally 30 min prior to Pilo in order to evade peripheral cholinergic side effects. And after 90 min of continuous limbic seizures, mice received diazepam (10 mg/kg, XiangYa Hospital, China) for the termination of seizures. The control group was injected with methylscopolamine and diazepam, like the Pilo-treated mice, except for three injections of saline instead of pilocarpine. Fer-1 was pretreated with a dose of 2.5 μmol/kg for 2 consecutive weeks before Pilo injection.
Transmission Electron Microscope (TEM) {#S2.SS3}
--------------------------------------
Mice from different groups were deeply anesthetized with 10% (g/ml) chloral hydrate and transcardially perfused with 0.1 M phosphate buffer saline (PBS, pH = 7.4), followed by the fixations of 4% paraformaldehyde (PFA), and 2% glutaraldehyde. The tissues were cut into 100 nm-thick sections and then were stained with uranyl acetate and lead citrate. Morphological mitochondrial features were observed under a JEM2000EX transmission electron microscope (TEM; JEOL, Tokyo, Japan).
Western Blot Assay {#S2.SS4}
------------------
Western blotting was conducted according to our previous descriptions ([@B17]). Briefly, hippocampal tissues were lysed in high KCl lysis buffer containing 10 mM Tris--HCl, pH 8.0, 140 mM NaCl, 300 mM KCl, 1 mM EDTA, 0.5% Triton X-100, and 0.5% sodium deoxycholate with 1 mM phenylmethylsulfonyl fluoride (Roche, United Kingdom). The supernatant was quantified using a commercial BCA kit (Beyotime Biotechnology Institute, China). Protein samples were separated by SDS-polyacrylamide gels (SDS-PAGE) and transferred electrophoretically to polyvinylidene fluoride membranes. After blocking, the membranes were incubated with 4-hydroxynonenal (4-HNE) (Mouse, 39--122 KDa, MAB3249, 1 μg/ml, Novus, United Kingdom), GPX4 (Rabbit, 22 kDa, ab125066, 1:5000, Abcam, United Kingdom) and β-actin (Mouse, 43 kDa, A5441, 1:10000, Sigma-Aldrich, United States) overnight. The next day, after washing, the membranes were then incubated with secondary IgG goat anti-rabbit (A9169, 1:10000, Sigma-Aldrich, United States) or rabbit anti-mouse antibody (A9044, 1:10000, Sigma-Aldrich, United States). Immunodetection was performed using an enhanced chemiluminescence kit and the intensity of protein bands was analyzed by Quantity One software (BioRad, United States).
Real-Time Quantitative PCR {#S2.SS5}
--------------------------
After drug treatment, total RNA from tissues or cell cultures was obtained using TRIzol reagent (Invitrogen, United States) following the manufacturer's protocols. Then, 1 μg total RNA for each sample was reverse transcribed into complement DNA using the SYBR Green RT Kit (Takara, Japan). Real-time PCR was carried out using SYBR Green PCR Master Mix (Takara, Japan). Samples were analyzed on a LightCycler Roche 480 qPCR instrument with absolute quantification settings. PCR conditions were displayed as follows: 30 s hot start at 95°C followed by 40 cycles of 5 s at 95°C, 30 s at 55°C, and 30 s at 72°C; 30 s melting curve at 95°C. All samples were determined in triplicate, and differences in mRNA levels were calculated using the δδCt method, with β-actin as an internal reference control. The following primer sequences were used: PTGS2: Forward: 5′-GGGAGTCTGGAACATTGTGAA-3′ and Reverse: 5′-GTGCACATTGTAAGTAGGTGGACT-3′. β-actin: Forward: 5′-GTGACGTTGA-CATCCGTAAAGA-3′ and Reverse: 5′-GCCGGACTCATCGTACTCC-3′.
Measurements of Malonaldehyde (MDA) and Glutathione (GSH) Levels {#S2.SS6}
----------------------------------------------------------------
Measurements of GSH and MDA levels were detected using their corresponding commercial kits (S0055 for GSH and S0131 for MDA, Beyotime Technology Institute, China) according to manufacturer's instructions.
Detection of Iron Content {#S2.SS7}
-------------------------
The iron concentration from each group was determined using an Iron Assay Kit (ab83366, Abcam, United Kingdom) following the manufacturer's protocols.
Data Analysis {#S2.SS8}
-------------
Experimental data was presented as mean ± SD and Prism 5.0 software (GraphPad Software, La Jolla, CA, United States) was used for analysis. A statistical significance was identified using Student's *t* tests between two groups and between three or more groups by analysis of variance, using One-Way ANOVA with the Bonferroni test. Statistical differences with *p* values less than 0.05 were deemed significant.
Results {#S3}
=======
Occurrence of Ferroptosis in Murine Models of PTZ Kindling and Pilo-Induced Seizures {#S3.SS1}
------------------------------------------------------------------------------------
First, we explored whether ferroptosis occurred in PTZ kindling and Pilo-treated mice through morphological observation of mitochondria using TEM and the detection of PTGS2 mRNA, a previously identified ferroptotic marker, using real-time quantitative PCR ([@B13]). Our results revealed that smaller mitochondria and the upregulation of PTGS2 mRNA were found in murine models of PTZ kindling or Pilo-triggered seizures ([Figure 1](#F1){ref-type="fig"}), indicating the presence of ferroptosis in these two models.
{#F1}
Fer-1 Attenuates Seizures in PTZ Kindling and Pilo-Treated Mice {#S3.SS2}
---------------------------------------------------------------
PTZ- and Pilo-treated mice both exhibited at least stage 5 seizures while mice in the control group did not show any signs of seizure ([Figure 2](#F2){ref-type="fig"}). We also observed an average of eight and nine seizures in PTZ- and Pilo-treated mice within 90 min of behavioral observation ([Figure 2](#F2){ref-type="fig"}). However, treatment with Fer-1, a specific ferroptosis inhibitor ([@B3]; [@B26]), significantly decreased seizure severity (decreased seizure score) and frequency (decreased number of seizures within 90 min) in murine models of PTZ kindling as well as Pilo-induced seizures. No significant difference was found among the three groups in terms of the latency to seizures.
{#F2}
Fer-1 Decreases Iron Accumulation and PTGS2 mRNA in the Mice Hippocampus of PTZ Kindling and Pilo-Induced Seizures {#S3.SS3}
------------------------------------------------------------------------------------------------------------------
The results of iron measurement indicated that iron accumulation was observed in the hippocampus of PTZ kindling and Pilo-induced seizures ([Figures 3A,B](#F3){ref-type="fig"}). Administration of Fer-1 remarkably reduced hippocampal iron content in PTZ- and Pilo-treated mice ([Figures 3A,B](#F3){ref-type="fig"}). PTGS2 mRNA was decreased in both seizure models after Fer-1 treatment ([Figures 3C,D](#F3){ref-type="fig"}).
{#F3}
Fer-1 Inhibits Lipid Peroxidation in the Mice Hippocampus of PTZ Kindling and Pilo-Induced Seizures {#S3.SS4}
---------------------------------------------------------------------------------------------------
The lipid peroxidation in the hippocampus in our current work was assessed by detecting the GPX4 protein expression, GSH level, MDA content, and 4-HNE level. As shown in [Figure 4](#F4){ref-type="fig"}, enormous lipid peroxidation was found, including decreased GPX4 protein expression, reduced GSH level, increased MDA content, and 4-HNE level in the hippocampus of mice subjected to PTZ kindling and Pilo-induced seizures, and this effect was reversed by Fer-1.
{#F4}
Discussion {#S4}
==========
Recurrent and spontaneous seizures is a common phenomenon in the etiology of epilepsy. Most of the traditional drugs exert neuroprotection or anti-epileptic potential by manipulating seizure severity and seizure frequency ([@B15]; [@B8]), indicating that figuring out the molecular mechanism on how a seizure generates is of vital importance. Hippocampal neuronal loss is a major pathological characteristic of human epilepsy. Multiple cell death modes including apoptosis, necroptosis, autophagy, and pyroptosis have been involved in hippocampal neuronal loss, subsequently aggregating epileptic progress ([@B2]; [@B30]; [@B16]; [@B33]). Suppression of these cell death processes can alleviate seizure-induced hippocampal damage and our present investigation provided direct evidence showing the presence of ferroptosis in PTZ kindling and Pilo-induced seizures in mice. More importantly, treatment with ferroptosis inhibitor Fer-1 ameliorated seizure severity and seizure frequency, highlights the potential therapeutic value for curing seizure-associated diseases such as epilepsy by targeting the ferroptosis process.
Pentylenetetrazole kindling and Pilo-induced seizures are ideal models for studying absence epilepsy and status epilepticus, respectively ([@B7]; [@B33]). These two models have been widely used for studying the process of epileptogenesis and develop novel anti-epileptic drugs. In our current work, PTZ kindling was established through the repeated injection with a sub-convulsive dose of PTZ in mice and almost all mice exhibited consecutive stage 4 seizures by the final dose, which was in line with previous investigations ([@B36]). Our current results first revealed the occurrence of ferroptosis in PTZ kindling, thus, future studies are essential to explore the regulatory mechanism of the ferroptosis process in this model. The ramping-up dosing protocol of the Pilo injection was also selected in our present study, as in our preliminary experiments, the high rate of lethality was observed by an injection with the dose of 300 mg/kg Pilo. It was found that three injections with the 100 mg/kg dose are sufficient for the induction of continuous seizure activity, which was consistent with a previous study ([@B24]).
Ferroptosis is a novel type of RCD and has been reported to be involved in multiple diseases including cancers ([@B1]), neurodegeneration ([@B11]), and renal failure ([@B10]). At least two key factors such as iron and manipulation of lipid peroxidation are indispensable for the execution of ferroptosis ([@B4]). Thus, we hypothesized that brain pathology including PTZ kindling or Pilo-induced seizures is vulnerable to ferroptosis, as the brain contains high-rich phospholipids which are easily subject to lipid peroxidation ([@B6]). We found that shrunken mitochondria and the upregulation of PTGS2 mRNA, two features of ferroptosis, in PTZ- and Pilo-treated mice, confirmed the presence of ferroptosis in epileptic seizures. A previous study reported the occurrence of ferroptosis in kainic acid-induced epileptic rats, which was similar to our current investigation ([@B34]). Additionally, in our current work, treatment with the specific ferroptosis inhibitor Fer-1 remarkably alleviated seizures in murine models of PTZ- or Pilo-injection. However, prior work did not observe the ameliorative effect of seizures after treatment with Fer-1 in kainic acid-injected rats. The discrepancy may be attributable to the different methods of seizure model preparation and the use of different species.
As a key regulator of ferroptosis, GPX4 inhibition was previously found to trigger renal failure and exacerbate cognitive deficits via inducing ferroptotic cell death in mice ([@B10]; [@B11]). Evidence for a critical role of GPX4 in epilepsy has arisen from the results showing that GPX4 is a selenium-dependent enzyme for interneuron development and prevention of epileptic seizures ([@B32]; [@B12]). Our present investigation demonstrated the reduction of GPX4 protein expression in mice treated with PTZ and Pilo. And Fer-1 restored the GPX4 protein level. Consistently, kainic acid-treated rats also exhibited GPX4 reduction and this effect was reversed by Fer-1 ([@B34]).
In summary, our present work uncovers a novel type of cell death mode, ferroptosis, in PTZ kindling and Pilo-induced seizures and treatment with ferroptosis inhibitor Fer-1 significantly mitigates seizures in PTZ- and Pilo-treated mice.
Data Availability {#S5}
=================
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.
Ethics Statement {#S6}
================
All animal care and procedures throughout the study were approved by the Ethical Committee of the Animal Centre of Central South University.
Author Contributions {#S7}
====================
X-YM and W-LJ designed the study. X-YM wrote the manuscript. W-LJ and H-HZ revised the manuscript.
Conflict of Interest Statement {#conf1}
==============================
The 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.
**Funding.** This work was partially supported by the National Natural Science Foundation of China (Nos. 81671293 and 81302750), the Natural Science Foundation of Hunan Province (No. 2017JJ3479), and the Hunan Provincial Department of Education Innovation Platform Open Fund Project (No. 17K100).
[^1]: Edited by: Mohammad Badruzzaman Khan, Augusta University, United States
[^2]: Reviewed by: Mohammad Farhan, Hamad Bin Khalifa University, Qatar; Syed Mohd Abbas Zaidi, Hakim Syed Ziaul Hasan Government Unani Medical College, India; Syed Shadab Raza, ERA's Lucknow Medical College, India
[^3]: This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neuroscience
| {
"pile_set_name": "PubMed Central"
} |
Background and goal of study
============================
This study reports of a new way to verify the recruitment of atelectatic areas with a mobile CT, in patients who are served superimposed high frequency jet ventilation (SHFJV) \[[@B1]\].
Material and methods
====================
After institutional approval four patients, three male and one female with a mean age of 52 years (ranging 21 to 83 years) received SHFJV for four hours by a prototype of a new electronic jet ventilator named ALEXANDER 1. Blood gas analyses were taken every half hour during the whole study period. A thorax CT (mobile CT, Philips Tomoscan M, Netherlands), was taken of every patient before and after the period of ventilation with the possibility of 3D reconstruction. The ventilated areas of each thoracic CT-picture were digitally remastered and volumetricly measured by a serial scan of the lungs (matrix 512 to 512, zoom factor 1.4; slicethickness 1.0 cm, pitch 1.0 cm, trash hold of -400 to -1000 HE). A high resolution CT (HR-CT) was performed of every patient\'s lungs (slicethickness 0.2 cm, pitch 1.5 cm, same trash hold and same matrix as in the serial scan) before and after jet-ventilation, to detect any alteration of the parenchyma or changing of (he diameters of the bronchi. Data export was performed via DICOM III on a Philips Easy Vision 2,1.2. Sun Sparc Station 5.
Results and discussion
======================
The situation of every lung improved. The average amount of the improvement of the ventilated areas were 329 cm, which represented an average increase of 11.7%. In analogy of these values, we saw that the PaO~2~ measurements of the arterial blood gas analyses were improving too. The average increase of the PaO~2~ value amounted to 40.4% within the period of SHFJV, meaning that the improvement of the ventilation (recruitment of the atelectatic areas) leads to a better oxygenation. Regarding the HR CT, we saw no dilatation of the bronchi, but a decrease of the atelctatic areas.
Conclusions
===========
The improvement of the ventilation and oxygenation can be umpirely objectified with this new method. This possibility improves the critically regarded situation of the SHFJV.
| {
"pile_set_name": "PubMed Central"
} |
Genetically identical cells have the capacity to stochastically differentiate into various phenotypes each with it own with unique attributes. This hedge survival strategy allows the population to continuously deploy specialized cells in response to, and in anticipation of, possible drastic changes in conditions[@b1][@b2][@b3][@b4][@b5][@b6][@b7][@b8][@b9][@b10][@b11]. The stochastic phenotype differentiations (or stochastic fate determinations) involve cell-cell communication and coordination and provide each cell with the flexibility and freedom to select its own phenotype according to the specific conditions it encounters but in harmony with the other cells. A variety of different phenotypes interact and contribute for the well-being of the colony by performing different tasks[@b12][@b13].
The phenotypical diversity arising from isogenic populations leads to the question, how random these individual decisions are. On one hand, neighbors exposed to the same environment need to make different decisions about their fates, in order to achieve diversity at the local level. On the other hand, the individual cell decisions must be collectivley regulated and coordinated carefully to garantee the optimal distribution of phenotypes for the colony as a whole. That means that even though there is need for randomness in the decision process to break symmetry, the probabilities of the possible outcomes must be carefully regulated by sensorial inputs and cell-cell communication.
The challenge is to reveal the principles governing how individual cells sense their environment and communicate with their neighbors before their own fate determination. And how, at the same time, these same individuals leave the final decision to chance in order to avoid the choice of the same phenotype by the whole population? To do so, the decision circuits must have a special capacity (yet not understood) for noise managment, allowing the bacteria to detemine fate by "playing dice with controled odds"[@b1]. Cellular capacity to manage the odds should entail both means to program and regulate the noise level and means to program the effect of the noise on the gene circuit performance[@b5][@b6][@b7][@b8][@b9][@b10][@b14]. Several studies have shown that circuit architecture (the connectivity map between the circuit genes) can encode distinct noise behaviors critical to the function implemented by the circuit[@b4][@b5][@b6][@b15][@b16][@b17].
A prototypic example of how genetic networks harness noise for performance of cellular differentiation is the fate determination between sporulation and competence in *Bacillus subtilis*. Many bacteria strains, in response to severe starvation, can form endospores -- dormant cells that are remarkably resistant to many hazards like heat, radiation and toxic chemicals. The process of sporulation is accompanied eventually by termination of metabolic activity in one daughter cell (the spore) and death by lysis of the other (the 'mother cell'). Sporulation is not initiated automatically upon nutrient limitation, but instead it is a last resort. Initially, a variety of other tactics to survive the stress can be employed. Up to eight different phenotypes have been identified in *Bacillus subtilis* when facing starvation, including differentiation into higher flagellated motile phenotype seeking new food niches, differentiation into phenotypes mastering in the secretion of hydrolytic enzymes to scavenge extracellular proteins and polysaccharides and differentiation into cannibal phenotype feeding on its peers[@b12][@b13]. When other tactics fail in lifting the stress, sporulation is the cell fate chosen by a majority of the cells. The material released by lysis of the sporulating cells is not wasted but can be taken up by a minority of competent cells. On the path towards sporulation, the individual cells can switch (escape) into competence and become able to uptake the genetic material from lysed cells which can be used as a food sources, for DNA repair and occasionally even as new genetic information to enable resisting the encountered stress. The competent cells can switch back (after about a day) into vegetative growth and proceed towards sporulation[@b18].
The network performing the decision between sporulation competence is a complex one, involving several modules and inputs. Many different stress signals are integrated into a phosphorelay leading to the phosphorylation of the sporulation master regulator Spo0A which have been show to act as a timer with adaptable clock rate - the production rate of Spo0A\* (phosphorylated Spo0A)[@b19][@b20]. The process terminate upon comitment to sporulation when Spo0A\* exceeeds a thereshold level. In order to regulate this process, pheromones are sent and received by the cell to indicate not only the local population density, but also information on stress levels sensed by the neighbors, which indicate their propensities of entering competence or sporulation[@b21]. The transition into competence requires noise in the expression of its master regulator ComK. A positive feedback loop on ComK is activated when fluctuations lead its concentration to cross a certain threshold. By interfering with the active degradation of ComK by MecA, a peptide ComS linked to the quorum sensing response sets the threshold for self-activation of ComK.
The stand-alone operations of the two modules -- the sporulation timer and the competence switch - are well understood. The next challenge requires understanding of the interplay between the operations of the two modules, which determines the way bacteria decide between the two phenotypes. The AbrB-Rock decison gate is at the core of the sporulation-competence interplay[@b22][@b23]. However, the operational principles of this gate, which couples the opening of the stochastic switch with the state of the timer, are still not understood and subject to an ongoing debate.
Here we reveal the connection between the distinct circuit architecture of this gate and its special dynamics and noise management characteristics. Our starting point is the realization that the gate has intriguing dynamics since AbrB forms, together with Spo0A\* and Spo0E, a special repressilator-like motif[@b24]. This leads to new operational principles: 1. "Inhibition of inhibition" -- inhibition of the gate by Spo0A\* and inhibition of ComK by the gate. 2. "Window of opportunity" with oscillating dynamics that generates a chain of short intervals with high probability of transition into competence -- a chain of opportunities. We propose that the "inhibition of inhibition" and the "chain of opportunities" are essential principles for collective decision-making in general.
Global view of the decision-making system: Considerable research effort has been devoted to untangling the components of the genetic decision--making system which allows an individual cell to decide whether to wait, go through a competence cycle or commit to sporulation. It is now understood that the decision follows an elaborate assessment of its individual stress level, the colony density, stress signals from other cells and a memory of previously encountered stress and colony state. Years of intensive experimental studies identified the tens of key regulatory genes and measured several of the associated physiological parameters that are involved in the sporulation-competence decision process of the domesticated *B. subtilis* 168. More recently, these findings led to the development of tractable quantitative model of some of the elements (or module circuits) of this highly interconnected genetic network shown in [Figure 1](#f1){ref-type="fig"}.
In this approach the operation of the competence module is modeled as a stochastic switch whose transition rate is controlled by a quorum sensing unit, and the operation of the sporulation module as an adaptable timer whose clock rate is adjusted by stress signals and signals sent from other cells. More specifically, Spo0A\* accumulation is determined by a cascade of kinases transferring phosphate to the sporulation master regulator Spo0A[@b25][@b26][@b27][@b28][@b29]. Phosphate is transferred down the relay, leading to the accumulation of Spo0A\*. The outcome is that the clock rate of the sporulation timer is adjusted by the cell stress. Spo0A\* acts as a transcriptional activator of both Spo0A and Spo0F via the sigma factor **σ**^H^.
The competence stochastic switch consists of a self-activator master regulator ComK and a degradation complex MecA/ClpP/ClpC which continuously acts to keep ComK at low levels[@b30][@b31]. This degradation is regulated by competitive binding of peptide ComS. It has been proposed that the ComK-ComS-MecA circuit can act as an excitable system, a bi-stable system or both, depending on parameters[@b3][@b4][@b6][@b7][@b32]. The transition probability is regulated by the cell density in response to the level of quorum sensing pheromone ComX[@b33] which activates the production of ComS via the competence and the ComP-ComA two component quorum sensing system.
The interplay between the timer and stochastic switch master modules is regulated by the operation of the Rap communication module and the AbrB-Rok decision gate. The Rap module[@b34][@b35] acts as the central cell-cell communication and information possessing system of the decision-making network. Generally speaking, Rap decreases the clock rate of the sporulation timer and increases the waiting time of the competence switch by dephosphorylation of Spo0F\* and inactivation of ComA, respectively. The module is up-regulated by the quorum sensing signal (via ComA) and down-regulated by the external peptide pheromones secreted by the neighboring cells and the cell itself. Rap is also regulated by Spo0A\* which enhances the production of some of the pheromones (e.g. PhrC) via **σ**^H^. More recently, it was discovered that the Rap system provides the means to prevent sporulation during competence. ComK activates RapH, which dephosphorylate phosphorelay component Spo0F[@b35].
The AbrB-Rok gate, described in greater details in the next section, acts as an inhibitory gate (repressor) of ComK, which blocks competence transitions, unless high levels of Spo0A\* make the levels of both AbrB and Rok sufficiently low, as detailed further below.
The AbrB-Rok decision gate: Previously, Schultz *et al*[@b14] proposed that the combined task of the AbrB-Rok module is gating the competence transitions to be allowed only between two values of Spo0A\* - the "window of opportunity". AbrB is repressed by the sporulation master regulator Spo0A\* and also by itself, in a negative feedback loop that prevents overexpression. Due to the instability of the protein, AbrB concentration responds quickly to transcriptional repression, dropping its levels quickly in the presence of Spo0A\*. Lower AbrB concentrations allow increase in the expression of Spo0E, a phosphatase that acts directly on Spo0A\*, slowing down its accumulation.
The Spo0A-AbrB-Spo0E part of the decision gate regulates the clock rate of the sporulation timer (the rate of accumulation of Spo0A\*). Since Spo0A\* is dephosphorylated by Spo0E which is inhibited by AbrB which is inhibited by Spo0A\*, these three genes form a special repressilator circuit[@b24]. Hence, the clock rate is regulated via a special repressilator-like dynamic. For some input signals, this repressilator can cause the concentration of the components of the decision module to oscillate. The oscillations of the protein levels lead to short intervals with elevated probability of transition into competence ("chain of opportunities") when the levels of both AbrB and Rock are below some threshold.
The repressilator is a well studied network motif consisting of three genes that repress each other in sequence and in a loop - A represses B, B represses C, C represses A (ABC for short). This circuit, when implemented experimentally in a cell, showed oscillatory behavior. The Spo0A\*-AbrB-Spo0E (ABE for short) circuit is a variant of the classical repressilator, where one of the repressions (Spo0E- Spo0A\*) is mediated via dephosphorylation, rather than a transcriptional repression, and two of the components show regulatory feedback loops, one negative (AbrB represses its own transcription), and one positive (Spo0A\* activates its own transcription indirectly through **σ**^H^ transcription activation of Spo0A). In addition, unlike the classical repressilator, the circuit studied here is driven by an input signal: the rate of phosphorylation of Spo0A that is determined by the stress level. As we show in the next sections, the outcome of this driven repressilator is not only directly responsible for entrance into sporulation, but it also intermediates the sporulation pathway and the competence pathway by translating oscillations in the driven repressilator into windows of opportunity.
Results
=======
The modeling approach and the definition of the activation and inhibition Hill functions are presented in the Methods section via the example of a variant of the classical ABC repressilator that is transcription driven by an external signal. To better understand the functional role of the various features of the Spo0A\*-AbrB-Spo0E circuit, we also inspect in the Methods section the dynamics of a variant of the classical ABC repressilator in which A (the gene stimulated by the external signal) is self-activated. The models studied here involve a large number of parameters (e.g. transcription and degradation rates of the various genes, the rank of the various inhibitory and excitatory Hill functions, etc.). Since only some of the parameters are known, and even those ones have uncertainty regarding the exact values, the challenge is to choose a "realistic" set of parameters for meaningful predictions and sound testing of the model predictions. The way the parameters were selected is described in the Methods section and [SI6](#s1){ref-type="supplementary-material"} where we also detail all the parameters. For the sake of clarity, we present the models with real rather than dimensionless units -- concentrations are represented by the number of molecules in the cell of each of the proteins, time is measured in minutes, and production and degradation rates are measured in number of molecules per minute.
Phosphorylation driven repressilator
------------------------------------
The Spo0A\*-AbrB-Spo0E circuit employed by the bacteria can be viewed as a phosphorylation driven self-activated repressilator in which A\* (Spo0A\*), the phosphorylated version of the protein A (Spo0A), inhibits B (AbrB) which inhibits E (Spo0E). E dephosphorylates A\* and the input signal phosphorylates A which is activated by A\*. The deterministic dynamics of the circuit is described by the following equations for A, A\*, B and E: where = 0.00375 (1/molecule)/minute is the rate constant of the phosphorylation of Spo0A by the input signal I~S~ = S is measured in number of molecules (see [SI6](#s1){ref-type="supplementary-material"}), and = 0.00155 (1/molecule)/minute is the rate constant of dephosphorylation of Spo0A\* by Spo0E. and () represent activation and inhibition, respectively, Hill function of gene (X) by protein (Y) (see Methods). is an inhibitory Hill function describing the AbrB self-inhibition. Comparison between the dynamics of the phosphorylation repressilator ([Equation 1](#m1){ref-type="disp-formula"} to [](#m2){ref-type="disp-formula"}[](#m3){ref-type="disp-formula"}[4](#m4){ref-type="disp-formula"}) and the classical repressilator with self-activation (See Method section) is shown in [Figure 2](#f2){ref-type="fig"}. Inspection of the role of the AbrB self-inhibition shows that it does not have significant effects on the dynamics. Therefore, it seems that the AbrB self-inhibition is relevant to other functions of this gene. However, we kept the effect in the model for the sake of completeness. Inspection of how changes in the circuit parameters affect the dynamics is included in [SI3](#s1){ref-type="supplementary-material"}. In particular it should be kept in mind that while the gate exhibits oscillations for a wide range of circuit parameters (Case I in [SI3](#s1){ref-type="supplementary-material"}), for some parameters there is a decline in AbrB but oscillations are not generated (Case II and Case III in [SI3](#s1){ref-type="supplementary-material"}). The midpoint of the Spo0A self-activation , is 180, 240 and 300 proteins for Case I, II and III, respectively. The additional parameters for these cases are listed in [SI6](#s1){ref-type="supplementary-material"}.
The AbrB-Rok gating characteristics
-----------------------------------
AbrB-Rok gate blocks competence transitions by transcription inhibition of ComK by both AbrB and Rok independently ([Figure 3a](#f3){ref-type="fig"}). These effects are incorporated in the dynamical equation for ComK (denoted by K) by multiplying the production rate of K by two corresponding inhibitory Hill functions so that (see [SI2 for more details](#s1){ref-type="supplementary-material"}): R represents the concentration of Rok and the last term on the RHS of [equation 5](#m5){ref-type="disp-formula"} represents the effect of ComS on the degradation of K (see Ref. [@b14] and [SI3](#s1){ref-type="supplementary-material"}). [Equation (5)](#m5){ref-type="disp-formula"} clarifies why the gate enables competence transition only when the values of both AbrB and Rok are below a certain threshold. The AbrB inhibition of Rok is incorporated by an inhibitory Hill function , which is multiplied by another inhibitory Hill function representing the Rok self-inhibition. Thus, the deterministic dynamics of Rock is described by Consequently, when AbrB oscillates it induces reciprocal oscillations in Rok with some small phase shift as is shown in [Figure 3b](#f3){ref-type="fig"}. Therefore, at each oscillation there is a short time interval during which the values of both AbrB and RoK are sufficiently low that the inhibition of ComK is reduced as is shown in [Figure 3c](#f3){ref-type="fig"}. Each inhibition reduction leads to a sharp increase in the concentration of ComK ([Figure 3d](#f3){ref-type="fig"}) which in turns elevates the transition probability per unit time into competence, as is shown in [Figure 3e](#f3){ref-type="fig"}. In other words, the AbrB-Rok competence gate opens a window of opportunity to escape into competence with the accumulated transition probability grows in steps as in shown in [Figure 3f](#f3){ref-type="fig"}.
In the example shown in [Figure 3](#f3){ref-type="fig"} the accumulated probability is 0.06. While the probability depends on the value of the parameters. We note that by using "realistic parameters" the probability can typically vary between 0.01 and 0.1, which is consistent with experimental observations. Also the gate function can change when the Rok transcription parameters change in a drastic manner -- insufficient inhibition prevents transitions into competence, while over inhibition enhance the competence transition ([SI6](#s1){ref-type="supplementary-material"}). This prediction can be directly tested experimentally.
Noise managment
---------------
So far, we analyzed the gate dynamics in the absence of noise. In this section we investigate, the dependence of the gate performance on its capacity to manage the effect of both external and internal noise and on its robustness with regard to changes in the circuit parameters. Our investigation are motivated by previous studies of the relationships between circuit architecture, noise behaviors and the circuit task performance[@b1][@b16][@b17].
### External noise
Motivated by the above we investigated the AbrB-Rok gate capacity to manage external and internal noise. In [SI4](#s1){ref-type="supplementary-material"} we show that the cell fate determination between sporulation and competence is not sensitive to external noise. More specifically, we found that noise added to the external input signal has only weak effect on the accumulated transition probability into competence. In other words, the circuit integrates the stress signal while filtering out noise, thus guaranteeing a robust response.
### Internal noise
A stochastic approach was used to investigate the AbrB-Rok gate capacity to manage internal noise. In this approach we no longer use the differential equations that describe the deterministic dynamics. Instead, we modeled the circuit considering protein synthesis, degradation, binding and unbinding as stochastic events. The different relative probabilities are set to match the corresponding rate constants in the deterministic equations and the noise level is set by the binding and unbinding rates, taken from[@b24]. The circuit was then simulated with the Gillespie algorithm[@b36].
For circuit parameters which give rise to oscillatory dynamics in the deterministic case (Case I in [Figure 2](#f2){ref-type="fig"}), the internal noise causes the oscillations to become less ordered, yet the accumulated transition probability is retained almost unchanged as is show in [Figs. 4](#f4){ref-type="fig"} and [5](#f5){ref-type="fig"} (see [SI5](#s1){ref-type="supplementary-material"} for more details). For some circuit parameters which do not give raise to oscillatory dynamics in the deterministic case (Case II and Case III mentioned earlier), the internal noise has a stronger effect -- it can induce oscillations in a manner similar to that of the cases where the deterministic model shows oscillatory behavior, as is shown in [Figs. 4](#f4){ref-type="fig"}. Yet, the effect of the noise on the accumulate transition probability is also weak as is shown in [Figure 5](#f5){ref-type="fig"}. Thus, the special architecture of the AbrB-Rok gate enables the system to manage noise in a way which leads to oscillatory dynamics even in the absence of oscillations in the deterministic limit. In addition the noise is "harnessed" to make the dynamics more robust with respect to variations in the circuit parameters.
Discussion
==========
The current post genomic era increased information about the interactions between gene circuits to such an extent that is now possible to map the architecture of consolidative multi-module systems. This new information is paving the way to develop new classes of integrative models to comprehend the operational principles and noise management of multi-module task performing cellular networks[@b14]. A special well-studied example of such cellular system is the intricate decision-making network used by bacteria for fate determination between sporulation and competence. While each cell has the freedom to determine its own fate, the ratio between sporulation and competence is collectively regulated to fit current conditions according to the needs of the colony as a whole. This implies that the stochastic cell differentiations are carried out with special capacity of noise managements to harness the noise at different modules and different times according to the function. For example, in the adaptable timer (the Spo0A phosphorylation pathway), noise is undesirable. Stress fluctuations experienced by the individual cell should not lead to decision to sporulate at inconvenient times as it could have negative effect on the colony. Therefore the system has evolved to integrate stress signals over time, filtering out transient activations and guaranteeing a robust response[@b14]. On the other hand, the ComK stochastic switch is driven by noise that is required for the transition into competence: the ComK positive feedback loop is activated when fluctuations lead the ComK concentration to cross a certain threshold. Our studies explain how the two opposite noise requirements can be satisfied by the decision gate which regulates the opening of the stochastic switch according to the state of the timer.
Most previous studies focused on either the competence transition or the sporulation pathway. An important, yet less studied, part of the sporulation/competence decision system is the interplay between the competence switch and the sporulation timer via the AbrB-Rock decision gate studied here. The task of this gate is to regulate the opening and closing of the ComK stochastic switch according to the Spo0A timer. We demonstrated that the distinct circuit architecture of this gate leads to special dynamics and noise management characteristics. These are required for efficient task-performance of coordination between the stochastic switch and the adaptable timer - two modules with opposite noise requierments. Our studies revealed the operation-architecture principles of the gate: 1. Inhibition of inhibition -- inhibition of the gate by Spo0A\* and inhibition of ComK by the gate. 2. Phosporylation driven repressilator-like motif. 3. "Window of opportunity" with oscillating dynamics. We showed that the advantage of "inhibition of inhibition" is that the gate is not sensitive to noise from Spo0A\* (high concentration - low noise) and at the same time adds to the noise in the regulation of ComK (low concentration -- high noise)[@b1]. The advantage of the oscillations is to increase the transition probability of each cell by steps which can improve coordination between cells as is discussed further bellow.
Thus, new principles were revealed that explain the intriguing interplay between the timer and the stochastic switch. Experimental verifications of the predicted gate oscillations require parallel monitoring of the time dynamics of the concentration of three genes within a single cell. Such measurements are feasible with current technology.
Different gene circuits can generate oscillations. As we have shown, the Spo0A\*-AbrB-Spo0E repressilator motif leads to oscillations that are less sensitive to the circuit parameters and less sensitive to internal noise. For circuit parameters whose deterministic dynamics is not oscillating, the noise generates oscillating dynamics. This way the noise management of the gate renders its operation to be less sensitive to cell-cell variations in the circuit parameters. Being phosphorylation (instead of transcription) driven, the repressilator has a narrow and well-defined "window of opportunity" and is less sensitive to fluctuations in the external stress.
Since neither sporulation nor competence is advantageous to a solitary cell, it is crucial that the option to escape into competence vs. commitment to sporulation is done within a time frame, which is synchronized with the other cells. The phosphorylation driven repressilator motif facilitates narrow windows with its oscillating dynamics, which is crucial for cell-cell coordination. At each cycle the cell sends out a pheromone signal when level of Spo0A\* (which regulates the Rap communication module) increases. At the same time Spo0A\* has a delayed positive feedback via its activation of Spo0F and the cell also receive signals from the other cells. As we will show in details in forthcoming publication, the cell-cell communication leads to harmonization of the oscillating dynamics between the communicating cells.
Looking ahead, the new principles found in the context of sporulation vs. competence fate determination are likely to be relevant to other cases of collective cellular decisions of stressed bacteria such as cannibalism and fratricide as well as spore germination when conditions are improved. Another and less expected possible direction is cancer. Tumorigenesis involves a variety of cellular decisions such as epithelial to mesenchymal transitions, transition into dormancy (transition into quiescent state analogous to sporulation), and relapse (germination of dormant cells). Recently it has been shown that the decision of micrometastases to grow is a complex process in which several nearby micrometastases have to perform a common decision to grow together for full metastasis maturation[@b37].
Methods
=======
Classical repressilator
-----------------------
To better understand the functional role of the various features of the Spo0A\*-AbrB-Spo0E circuit, we first inspect the dynamics of a classical ABC repressilator that is transcription driven along a variant in which A is self-activated.
The deterministic equations of a classical ABC repressilator are given by: The gene base production rates are g~X~ (X stands for A, B and C respectively) and the corresponding protein degradation rates are k~X~. The transcription inhibitions are included by the inhibition Hill functions represent inhibition of gene (X) by protein (Y) are given by: Where (n) is the rank of the Hill function (nonlinearity or cooperativity) and Y~0~ is the midpoint concentration. In [SI1](#s1){ref-type="supplementary-material"} we present the phase space analysis and the time dynamics of classical repressilator with different nonlinearity. We also show that the oscillatory behavior is retained when A is self-activated so that [equation (7)](#m7){ref-type="disp-formula"} is replaced by where is an excitatory (activation) Hill function given by:
Transcription driven repressilator
----------------------------------
The repressilator can be driven by an input signal \[I~S~ = S\] or by another gene D (whose level is S) incorporated by replacing [equation (7)](#m7){ref-type="disp-formula"} with [equation (13)](#m13){ref-type="disp-formula"} and (14) respectively: where I~S~ = S in [equation (13)](#m13){ref-type="disp-formula"}, is an excitatory Hill function ([equation 12](#m12){ref-type="disp-formula"}) and D = S in equation (14). We found, as is shown in [Figure 2](#f2){ref-type="fig"} and detailed in [SI1](#s1){ref-type="supplementary-material"}, that the oscillatory behavior is retained for both cases for a wide range of signal levels. For specific choices of the circuit parameters the oscillations start above a threshold signal level S~1~ and exist up to a second higher signal level S~2~. When the signal is increased in time, driving the repressilator via a Hill function (equation 14) introduces nonlinearity in the production rate, which enables to reduce the range of oscillations. However, it requires taking a non-realistic high level of g~AS~ (orders of magnitude higher than g~A~).
Inclusion of self-activation of A also enables to reduce the range of oscillations. This was shown by investigating the additional variants of the circuit in which we replaced [equations (13)](#m13){ref-type="disp-formula"} and (14) by [equations (15)](#m14){ref-type="disp-formula"} and (16) below:
Analysis of the variant described in [equation (16)](#m15){ref-type="disp-formula"} is detailed in [SI1](#s1){ref-type="supplementary-material"}. We found that as long as "realistic" parameters are used (see further below), the self-activation can lead to some reduction in the oscillation range but yet does not enable to generate a narrow window to limit the oscillations to be within a limited range. The reason of this lack of flexibility has to do with the inherent coupling between the effect of the driving signal on the transcription of A and the cascade of transcription inhibitions that lead to the oscillations. This coupling is relaxed when the repressilator is phosphorylation driven as is described in the Results section.
Selection of "realistic" parameters
-----------------------------------
To meet the challenge of selecting "realistic" parameters, we began with comparison of the model of the classical repressilator with the synthetic one studied in[@b24]. Based on the estimates of transcription rates of 0.5 mRNA/sec, translation rates of 0.167 mRNA/sec, mRNA half life of 2 minutes, protein half life of 10 minutes, and peak value of about 600 proteins, we chose the typical protein degradation rate as 0.1/minute, and the synthesis rate as 60 proteins/minute. Since the maximum protein levels are several hundreds, we selected the midpoint of the Hill functions to be at 100--200 proteins. Although much effort were used to determine these parameters, the features new mechanistic principles revealed by this work are robust and therefore not sensitive to precise values of these parameters.
In previous modeling of a classical repressilator, the rank of the Hill functions was taken as n = 2. We selected n = 4 to fit the rank of the Hill functions in the Spo0A-AbrB-Spo0E circuit. Consequently, the values were slightly adjusted to have the typical oscillation time of 30 minutes. The estimate of the rank of the Hill functions were based on the fact that AbrB forms a tetramer[@b38] and Spo0A forms a dimmer[@b39]. Consequently, we chose the rank of the Spo0A by Spo0A\* (that is mediated via **σ**^H^) as n~AA~ = 3 and the rank of the inhibition of Spo0E by AbrB as n~BA~ = 4. We note that since there is uncertainty regarding the proper ranks of the Hill functions, we tested the behavior for other sets of ranks and found that similar results can be obtained.
The value of the production rate for AbrB is higher to account for the fact that it is unstable[@b40], since production rates are normalized by degradation rates. In the case of self activation of Spo0A, we selected a base production of 12 proteins per minute and the self activation rate factor as slightly higher (than 60) -- 80 proteins per minute.
The relevant rate of the input signal phosphorylation of Spo0A and the Spo0E dephosphorylation of Spo0A\* were estimated based on the typical phosphorylation and dephosphorylation rates discussed in[@b41][@b42]. These rates are measured in [equations (1)](#m1){ref-type="disp-formula"} and (2) in units of (1/molecule)/minute since they are multiply by the number of molecules (the signal S for phosphorylation and the number of Spo0E molecules for dephosphorylation). To compare the phosphorylation and dephosphorylation rates with that of proteins\' synthesis and degradation we need to converted the (1/molecule)/minute units to units of (molecule/minute) and (1/minute) that are used to describe protein synthesis and degradation. Since the number of protein molecules is in the order of hundreds, the rates of phosphorylation and dephosphorylation can be several orders of magnitude higher than that of production and degradation in agreement with molecular information.
Final adjustments of the parameters were done to fit the experimental estimates of protein levels in the cells[@b4] (see the parameter list in [SI6](#s1){ref-type="supplementary-material"}). The estimations of the ComK parameters were taken from Refs [@b4],[@b6] and the estimations of the parameters related to the competence transition probability ([SI2](#s1){ref-type="supplementary-material"}) were done to fit the experimental observations that about 5--10% of the cells make the transition (see [SI6](#s1){ref-type="supplementary-material"}).
Author Contributions
====================
Authors\' contribution: D.S., J.O. and E.B.J. devised the model, D.S., M.L. and T.S. performed the simulations. All authors analyzed the results and wrote the article
Supplementary Material {#s1}
======================
###### Supplementary Information
Supporting Information
We are thankful to Herbert Levine and Gurol Suel for valuable and enlightening conversations. This research has been supported by the Center for Theoretical Biological Physics sponsored by the NSF (Grant PHY-1308264) and by the Cancer Prevention and Research Institute of Texas (CPRIT) at Rice University and the Tauber Family Foundation and the Maguy-Glass Chair in Physics of Complex Systems at Tel Aviv University.
{#f1}
{ref-type="disp-formula"} in the Methods section). (b) The corresponding bifurcation diagram. Y-axis is the range of the number of protein B for different values of the signal (X-axis) computed when the signal is constant at each value. When the signal level is close to zero or larger than a threshold value (below and above the two bifurcation points marked in brown and orange), the protein level is constant and marked in black (stable fixed point). Between these bifurcation points the protein levels oscillate - the maximum and minimum levels for the oscillation states are shown in solid blue line and the average levels are shown in dotted blue line. (c) and (d) are the same but for the phosphorylation driven circuit (B), [equation (1)](#m1){ref-type="disp-formula"}--[](#m2){ref-type="disp-formula"}[](#m3){ref-type="disp-formula"}[(4)](#m4){ref-type="disp-formula"} for circuit parameters corresponding to Case I -- the oscillation case. (e) and (f) are similar to (c) and (d) but for circuit parameters that do not yield oscillations -- Case II.](srep01668-f2){#f2}
{#f3}
{ref-type="fig"}. The two middle panels show the corresponding dynamics of *I* -- the ComK Inhibition. The two bottom panels show the corresponding accumulated transition probabilities.](srep01668-f4){#f4}
{ref-type="fig"}); Stars and red line denote simulations with parameters of the non-oscillation Case II (shown in [Figure 2](#f2){ref-type="fig"}).](srep01668-f5){#f5}
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Vertebral osteomyelitis is a bacterial infection of the bones of the spine resulting from hematogenous spread, direct inoculation, or contiguous spread from soft tissue infection \[[@CR1], [@CR2]\]. It is most often encountered at the lumbar levels, followed by thoracic and cervical levels \[[@CR1]--[@CR3]\]. The vast majority of bacterial species implicated in this infection are facultative anaerobes. The most common gram-positive species are *Staphylococcus aureus*, *Staphylococcus epidermidis, Streptococcus* species, and *Enterococcus* species while the most common gram-negative species are *E. coli*, *Pseudomonas aeruginosa,* and *Proteus mirabilis.*^4,5,6,7^ However, vertebral osteomyelitis is rarely caused by strictly anaerobic organisms, accounting for only 3--5% of all spondylodiscitis cases \[[@CR4]--[@CR6]\].
*Veillonella* is an anaerobic gram-negative bacterial species known to be found in the oral cavity, gastrointestinal tract, and female genital tract \[[@CR4], [@CR7]--[@CR9]\]. Thirteen species of the genus have been identified, with only six having been isolated in human oral cavities \[[@CR10]\]. Risk factors for *Veillonella* infections are typically immunosuppression, diabetes, malignancy, collagen disease, periodontal disease, open fractures, and instrumentation with endoscopy. Infections can also occur without any particular risk factors in healthy individuals. The most common ones are bone and joint infections, followed by endocarditis and bacteremia \[[@CR3], [@CR4], [@CR8]--[@CR12]\].
This case report received an exempt review status from the institutional review board and written consent was obtained from the patient allowing this information to be published.
Case presentation {#Sec2}
=================
Initial case presentation {#Sec3}
-------------------------
A 76-year-old male diagnosed with *Veillonella* bacteremia presented with a sudden onset of back pain, fever, and chills of four days duration. Although the fever and chills had since subsided, the back pain continued to persist and was exacerbated with positional change. CT scans revealed a psoas abscess with extradural extension, along with indications of lumbar osteomyelitis. MRI scans were unable to be obtained owing to the patient's pacemaker. Due to medication allergies and growing sensitivity to the infection, the patient was transferred to MICU for a short period to be desensitized to ampicillin-sulbactam. Interventional Radiology then placed a retroperitoneal drain for the psoas abscess as well as a PICC line for IV antibiotic administration at home.
First surgical operation {#Sec4}
------------------------
During the following week, the Neurosurgery service performed laminectomy at levels L4-S1. Bilateral medial facetectomies and foraminotomies were also performed at L4-L5 and L5-S1 to fully decompress the exiting nerve root and thecal sac. Significant facet arthropathy was noted. The extradural abscess, which was revealed to be a large synovial cyst with no pus, was resected. Cultures were taken and sent for laboratory analysis, revealing the presence of *Veillonella* species. Post-operatively, CT scans indicated a resolving abscess. The patient was discharged and instructed to complete his course of ampicillin-sulbactam at home and switch to metronidazole and levofloxacin for a total six-week course of antibiotics.
Second presentation {#Sec5}
-------------------
The patient re-presented to the hospital 18 days after the neurosurgical operation complaining of worsening back pain radiating into his posterior buttocks that began 10 days after the operation. CT scans revealed considerable interval decrease in the psoas abscess but extensive osseous destruction in the L5-S1 region, anterior to the laminectomy sites, thereby advancing concerns of osteomyelitis. CT scans revealed trace retrolisthesis of L2 on L3 and L3 on L4, as well as grade 1 anterolisthesis of L4 on L5. There was multilevel disc height loss, most severe at L3-L4 and L5-S1, and persistent narrowing of the thecal sac. Fluid was observed posterior to the surgical site along with phlegmonous changes extending from L4-S1. A small abscess was also seen in the S1 prevertebral soft tissues. Interval erosions of the L5 and S1 endplates were noted, compatible with discitis. All of these findings were consistent with progressive osteomyelitis. Furthermore, bone and tissue cultures revealed continued *Veillonella* infection, perhaps due to a combination of both insufficient debridement as well as inadequate duration of antibiotic treatment. The patient was instructed to begin IV vancomycin in addition to metronidazole and levofloxacin for an additional 6 weeks.
Third presentation {#Sec6}
------------------
The patient presented to the office 25 days later with unbearable, radiating lower back pain forcing him to use a wheelchair to move about. CT scans indicated progressive destructive changes of the L5-S1 vertebral body endplates. Two presacral abscesses measuring 4 mm and 7 mm were identified anterior to S1; additional microabscesses were found in the prevertebral phlegmon at the L5 level. A larger abscess measuring 15 mm was identified dorsal to the thecal sac at the previous L5 laminectomy site, leading to a narrowing of the dural sac. Progressive destruction of the vertebral bodies was noted with fragmentation of the inferior endplate of L5 and superior endplate of S1. Reflexes were absent bilaterally in the lower limbs and there was decreased sensation to light touch. Tissue cultures taken from the affected vertebrae were all negative at this time, and given the patient's debilitated status, reconstructive spine surgery was indicated.
Second surgical operation {#Sec7}
-------------------------
Preoperative X-ray revealed spondylolisthesis of L5 on S1, and a CT myelogram showed severe multilevel spinal stenosis from L2 to S1. Therefore, the patient underwent a L2-S1 lumbar laminectomy with posterior segmental instrumentation and posterolateral arthrodesis from the L2 to the ilium using pedicle screws and iliac bolts bilaterally. Decompression of spinal nerves was achieved by posterior lumbar laminectomy, bilateral medial facetectomies, and bilateral foraminotomies at all spinal levels between L2 and S1. Given the patient's history of infection, an allograft was used to form a final instrumented fusion mass. (Figs. [1](#Fig1){ref-type="fig"}, [2](#Fig2){ref-type="fig"}, [3](#Fig3){ref-type="fig"}, [4](#Fig4){ref-type="fig"} and [5](#Fig5){ref-type="fig"}). Fig. 1Lumbar CT Sagittal Reconstruction: progression of osteomyelitis at L4 and L5 with progressed anterolisthesis of L5 Fig. 2Lumbar CT Sagittal reconstruction: L5 with early osteomyelitic change Fig. 3Axial CT lumbar spine: note air within the abscess in the right iliopsoas muscle Fig. 4Lateral Lumbar X-ray: L2-Pelvis fusion construct with maintenance of lordotic alignment Fig. 5AP Lumbar X-ray: L2-Pelvis fusion construct
Post-operation {#Sec8}
--------------
The patient recovered well after surgery. Final blood and tissue cultures were all negative. CT scans indicated resolution of the remaining abscesses. He was discharged and instructed to complete the previously prescribed course of antibiotics and follow-up with rehabilitation facilities. After 12 months of rehabilitation therapy, the patient regained a majority of mobility and ambulated with cane. However, at the 12 month follow up visit, he complained of groin pain. Additional physical therapy including aqua therapy was ordered and followed. At the 18 month follow up visit, the patient is noted to be doing quite well with occasional back discomfort and continues to ambulate with a cane.
Discussion and conclusion {#Sec9}
=========================
Here we present the case of an elderly male that suffered from vertebral osteomyelitis secondary to *Veillonella* bacteremia that resulted from a tooth abscess. To the best of our knowledge, we could identify only one report that attributes its case of vertebral *Veillonella* osteomyelitis to the oral cavity \[[@CR9]\]. Additionally, there have been seven reports of vertebral *Veillonella* osteomyelitis and two reports of *Veillonella* foot osteomyelitis in diabetics. Other reported cases exist, but they could not identify an underlying source \[[@CR9], [@CR13]\].
*Veillonella* species are generally found in polymicrobial processes, and our case seems to adhere to this as no singular species could be identified. They have been implicated in dental infections, pulmonary infections, endocarditis, prosthetic joint infections, cardiac prosthetic valve infections and osteomyelitis \[[@CR4], [@CR7], [@CR9], [@CR10], [@CR13]\]. None of these predisposing risk factors could be identified in our patient. These bacteria are typically resistant to vancomycin, tetracycline, aminoglycosides, and ciprofloxacin, but infections are traditionally known to respond well to therapy with penicillin \[[@CR7], [@CR13]\]. In vitro, *Veillonella* is usually susceptible to cephalosporins, clindamycin, metronidazole, and chloramphenicol. Due to the scarce number of reports on *Veillonella* as a pathogen associated with invasive infection, there are no clear treatment recommendations in the literature. A few reports point to penicillin, cephalosporins, chloramphenicol, clindamycin, and metronidazole as providing positive responses \[[@CR7], [@CR13]--[@CR18]\].
In our unique case, the patient's allergies to penicillin and cephalosporin complicated his treatment course as extra steps were required to initially desensitize him to penicillin. For the first treatment course, the patient was given penicillin followed by a combination of metronidazole and fluoroquinolone for a total of 6 weeks. However, during this modified treatment period, the *Veillonella* species seemed to have invaded even more of the vertebral areas, despite the patient's initial spinal surgery to limit spread of infection and decompress the spinal nerves. This led to worsening symptoms, recurrent visits to the hospital, and continued persistence of *Veillonella* in blood and tissue cultures of the vertebrae. The patient was given a second course of treatment consisting of vancomycin, metronidazole, and fluoroquinolone. Two weeks into treatment, blood cultures and biopsy cultures of the bone tissue were all negative. Thus, posterior lumbar laminectomy and segmental instrumentation was performed.
It is strongly believed that the source of *Veillonella* infection in this patient was the oral cavity. Examination of the patient's past medical history revealed a tooth abscess which was incised and drained by an endodontist six weeks before the initial presentation. Due to the patient's travel plans, antibiotics were prescribed only after the first procedure. We believe that the delay in the onset of antibiotic treatment is what led to the initial bacteremia that ultimately took root in the lower lumbar vertebrae.
AP
: Anteroposterior
CT
: Computed Tomography
IV
: Intravenous
MICU
: Medical Intensive Care Unit
MRI
: Magnetic Resonance Imaging
PICC
: Peripherally Inserted Central Catheter
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Not Applicable.
AB, CM, JB and HH reviewed the case, collected the data and prepared the manuscript. ZH reviewed the manuscript and approved all aspects of project. All authors have read and approved the manuscript.
Authors' information {#FPar1}
====================
Not Applicable.
Not Applicable.
Not Applicable.
Participant consent was obtained. Drexel University Institutional review board waived review for this case report.
The patient gave written consent for clinical and personal information and any identifying images to publish in this study.
All authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .203}
{#sp2 .204}
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
A large number of poorly water-soluble drugs and drug candidates have significant bioavailability problems which limit their therapeutic efficiency or development beyond an early stage. For Biopharmaceutics Classification System II drugs, which are characterized by low solubility and high permeability, dissolution is the limiting step with regard to their oral absorption and bioavailability.[@b1-ijn-8-3119],[@b2-ijn-8-3119] The dissolution rate can be improved by reducing the particle size because of their increased surface area.
Nanosuspensions containing drugs (amorphous or crystalline) are nanoscale colloidal dispersions of pure drug particles stabilized by surfactants.[@b3-ijn-8-3119] Compared with other nanoscale drug delivery systems, nanosuspensions show promise for enhancing dissolution and hence the bio-availability of poorly water-soluble drugs, by eliminating the effects of food, allowing for dose escalation, and improving efficacy and safety.[@b4-ijn-8-3119]--[@b6-ijn-8-3119] Since the 1990s, many commercial products, including Rapamune®, Emend®, Tricor®, Megace ES®, Avinza®, Focalin XR®, Ritalin®, and Zanaflex Capsules™ have been marketed successfully.[@b7-ijn-8-3119] However, in spite of the advantages of nanosuspensions, there are many drawbacks associated with nanosuspension technology. A critical aspect concerns the poor stability of nanosuspensions in aqueous medium. Common processes, such as hydrolytic/oxidative degradation, particle growth, aggregation/agglomeration, change in crystallinity state, and sedimentation or creaming, may occur.[@b8-ijn-8-3119]--[@b10-ijn-8-3119] Moreover, nanosuspensions are always diluted, and have to be administered in large volumes to achieve therapeutic levels in the circulation.[@b11-ijn-8-3119] To overcome these problems, it is desirable to formulate nanosuspensions into solid dose forms.[@b3-ijn-8-3119],[@b12-ijn-8-3119] Freeze-drying and spray-drying are the two most commonly used solidification methods.[@b10-ijn-8-3119],[@b13-ijn-8-3119] However, the freeze-drying process is very costly and time-consuming, and spray-drying, although quite a robust process, requires manipulation at relatively high temperatures and is difficult to scale up. Other strategies, such as granulation, fluid-bed coating, and tableting, have not been widely investigated to date.[@b13-ijn-8-3119]
Fluid-bed coating is a "one-step" technique that is commonly used to add a film coating onto a substrate, and has widespread applications in the pharmaceutical industry. This technique involves solvent removal and simultaneous deposition of coating materials onto nonpareil pellets. In comparison with conventional spray-drying, fluid-bed coating is a much more efficient method of preparing solid formulations from bulk aqueous or organic solutions or suspensions. Moreover, fluid-bed coating is more scalable. Because of the advantages of improved drug dissolution, modified drug release, taste masking, and enhanced drug absorption,[@b14-ijn-8-3119]--[@b16-ijn-8-3119] solidification of nanosuspensions by fluid-bed coating appears attractive and promising. However, the process of solidification of drug nanosuspensions by fluid-bed coating is challenging because solidified formulations need to have the ability to reconstitute into their original nanosuspensions. Möschwitzer and Müller[@b17-ijn-8-3119] formulated drug nanosuspensions into pellets using a fluid-bed process, but did not report any data on the redispersibility of the nanocrystals upon reconstitution with water. Recently, Kayaert et al[@b18-ijn-8-3119] also reported on the feasibility of fluid-bed coating of nanosuspensions, but did not identify the solid state of the drug nanocrystals used. To date, there have been few reports on the art of solidification of nanosuspensions by fluid-bed coating.
Our previous findings[@b19-ijn-8-3119] indicated that amorphous indomethacin nanosuspensions could be prepared using a precipitation-ultrasonication method involving stabilization by biocompatible food proteins, eg, soybean protein isolate, whey protein isolate, and β-lactoglobulin. To improve the stability of such suspensions further for long-term storage, in this study we coated drug nanosuspensions onto pellets using fluid-bed coating technology. Specific attention was paid to the ability of the solidified nanosuspension to reconstitute by studying changes in particle size and distribution, drug particles in the solid state, and drug dissolution.
Materials and methods
=====================
Materials
---------
Indomethacin was purchased from Sine Pharmaceuticals (Shanghai, People's Republic of China), whey protein isolate from Davisco Foods International Inc (Le Sueur, MN, USA), soybean protein isolate from Hufeng Chemical Industry Co, Ltd (Shanghai, People's Republic of China), and β-lactoglobulin (from bovine milk, L3908, \>90% purity grade) from Sigma Chemical Co (St Louis, MO, USA). Polyvinylpyrrolidone (PVP) K30 was kindly supplied by International Specialty Products (Shanghai, People's Republic of China). Nonpareil pellets (sugar spheres 0.5--0.7 mm in diameter) were provided by Gaocheng Biotech and Health Co, Ltd (Hangzhou, People's Republic of China). Deionized water was prepared using a Milli-Q purification system (Millipore, Billerica, MA, USA). Other reagents were of analytical grade and used as received.
Preparation of nanosuspensions
------------------------------
An aqueous suspension of protein was obtained by dispersing 300 mg of protein (soybean protein isolate, whey protein isolate, or β-lactoglobulin) powder into 25 mL of water under magnetic stirring for one hour at 25°C, then adjusting the samples to pH 7 using 1 M NaOH. To denature the proteins and expose the nonpolar and disulfide bonds buried in the protein interior and thus increase the stabilizing capacity of the proteins, the soybean protein isolate, whey protein isolate, and β-lactoglobulin solutions were heated to 105°C, 85°C, and 85°C, respectively, in closed centrifuge tubes (50 mL, Corning Incorporated, Tewksbury, MA, USA) for 30 minutes.[@b19-ijn-8-3119]--[@b22-ijn-8-3119] The denatured protein solution was then cooled to 25°C for two hours.
The nanosuspensions were prepared according to the precipitation-ultrasonication method described previously by Mateucci et al and Xia et al, with some modifications.[@b2-ijn-8-3119],[@b23-ijn-8-3119] The aqueous dispersion of denatured protein functioned as the aqueous phase (30 mL), in which the concentrations of soybean protein isolate, whey protein isolate, and β-lactoglobulin were 8.5 mg/mL, 8.5 mg/mL, and 3.3 mg/mL, respectively. The drug-containing organic phase was prepared by dissolving indomethacin in 2 mL of acetone. The amount of indomethacin in the nanosuspensions stabilized by soybean protein isolate, whey protein isolate, and β-lactoglobulin was 150 mg, 150 mg, and 200 mg, respectively. The organic and aqueous phases were precooled to below 3°C in an ice-water bath. The organic phase was then added to the aqueous phase under mechanical stirring at 1,200 rpm. After the antisolvent precipitation process, the samples were immediately treated with an ultrasonic probe (20--25 kHz, Scientz Biotechnology Co, Ltd, Ningbo, People's Republic of China) at 500 w for 15 minutes. The probe, which had a tip diameter of 8 mm, was immersed 1 cm into the liquid, resulting in the wave traveling downwards and reflecting upwards. The period of ultrasound burst was set to 3 seconds, with a pause of 3 seconds between each burst of ultrasound. Temperature was controlled throughout using an ice-water bath.
Coating nanosuspensions onto pellets
------------------------------------
The layered pellets were produced by coating the indomethacin nanosuspensions onto nonpareil pellets using a fluid-bed coater (DPL1/3 Multi-processor, Jinggong Pharmaceutical Machinery Co, Ltd, Chongqing, People's Republic of China, [Figure 1](#f1-ijn-8-3119){ref-type="fig"}). Briefly, trehalose was first dissolved in the protein-stabilized nanosuspension, and PVP K30 was mixed with the aqueous dispersion of denatured soybean protein isolate under gentle stirring. Dispersion of the coating formulations was done by introducing the aqueous soybean protein isolate dispersion containing PVP into the nanosuspension slowly under gentle stirring. The dispersion was sprayed through a nozzle into the surface of the nonpareil cores in the fluid-bed coater. The operating conditions were as follows: inlet air temperature, 40°C--45°C; product temperature, 35°C; blower frequency, 12--26 Hz; rotational speed of peristaltic pump, 4--12 rpm; atomizing air pressure, 0.15--0.25 MPa; and spray nozzle diameter, 0.5 mm. After coating, the pellets were dried in the coating chamber at 35°C for a further 15 minutes.
The yield (%) was expressed as the theoretical percentage of the weight gained (TWG, %) to the weight of the layered pellets, and was calculated using the following formula:
TWG
\%
=
(
W
1
W
0
−
1
)
×
100
where W~0~ and W~1~ are the weights of the nonpareil pellets and layered pellets, respectively.
Redispersibility study
----------------------
The pellets (100 mg) layered by the nanosuspensions were dispersed in 10 mL of deionized water by shaking for about 1 minute. One milliliter of the resulting redispersed aqueous suspension was placed in a test tube and allowed to stand for a few minutes. The supernatant was withdrawn and characterized for mean particle (intensity weight) size and distribution by dynamic laser scattering.
Determination of particle size and zeta potential
-------------------------------------------------
The particle size and size distribution of the nanosuspensions were measured using a dynamic laser scattering instrument (380 ZLS, Nicomp Instruments, Santa Barbara, CA, USA). Raw data were collected over 5 minutes at 25°C and at an angle of 90 degrees, and processed further using the ZPW388 software program. The mean intensity--weight particle size was expressed as a volume-weighted Gaussian distribution (with a chi-squared value \<3). The surface charge of the nanosuspensions was determined by measuring electrophoretic mobility at 25°C with the Nicomp 380 ZLS. Nanosuspensions were diluted 50-fold in water before measurement.
Scanning electron microscopy
----------------------------
The surfaces of the nanosuspensions and pellets were studied using a scanning electron microscope (XL30, Philips, Eindhoven, the Netherlands). Prior to examination, the samples were fixed on a brass stub using double-sided tape and gold-coated in a vacuum by a sputter coater. The photographs were taken at an excitation voltage of 10 kV
Transmission electron microscopy
--------------------------------
A transmission electron microscope (JEM-1230, JEOL Ltd Tokyo, Japan) was used to determine the morphology of the nanosuspensions. The nanosuspensions and redispersed nanosuspensions were placed on copper grids and negatively stained with 1% (w/v) uranyl acetate for 5 minutes at room temperature.
In vitro dissolution
--------------------
The drug dissolution profiles for the raw crystals, nano suspensions stabilized using soybean protein isolate, whey protein isolate, and β-lactoglobulin, and the layered pellets were determined using US Pharmacopeia II apparatus (ZRS-8G release tester, Tianjin, People's Republic of China) at 100 rpm and a temperature of 37°C ± 0.5°C. Nanosuspensions, layered pellets, or raw crystals (25 mg) were added to 900 mL of fluid (phosphate buffer pH 6.8). Five milliliters of the samples were withdrawn at specific time intervals, filtered through a 0.2 μm filter, and the drug concentration was determined by ultraviolet spectrophotometry.
Differential scanning calorimetry
---------------------------------
Dried nanosuspension powder samples carefully peeled off from the outer layer of the glass pellets were used for physical characterization by differential scanning calorimetry (DSC) and subsequent powder x-ray diffraction analysis. Briefly, the indomethacin nanosuspension formulation was layered onto glass pellets (0.8--1 mm) using a fluid-bed coater under the conditions described above. The layered glass pellets were placed in a porcelain mortar (180 mm) and then gently ground to peel off the coating layer. About 5 mg of the samples (pure indomethacin, freeze-dried nanosuspension powder, and dried powder) were weighed into a nonhermetically sealed aluminum pan, and DSC analysis was performed using a 204A/G Phoenix 1 instrument (Netzsch, Selb, Bavaria, Germany). The samples were heated from 20°C to 250°C at a heating rate of 10 K per minute. The instrument was calibrated using indium. All DSC measurements were carried out in a nitrogen atmosphere at a flow rate of 100 mL per minute.
Powder x-ray diffraction
------------------------
Powder x-ray diffraction analysis of the samples (pure indomethacin, freeze-dried nanosuspension powder, and dried powder) was done using an X′Pert PRO diffractometer (Panalytical, Almelo, the Netherlands) over a 2θ range of 2.5--50 degrees at a scan rate of 3 degrees per minute, where the tube anode was Cu with Ka=0.154 nm monochromatized with a graphite crystal. The pattern was collected at 40 kV of tube voltage and 60 mA of tube current in step scan mode (step size 0.02 degrees, counting time 1 second per step).
Statistical analysis
--------------------
The results are expressed as the mean ± standard deviation. One-way analysis of variance was used to assess the statistical significance of differences between samples. Results with *P* \<0.05 were considered to be statistically significant.
Results and discussion
======================
Preparation and characterization of nanosuspensions
---------------------------------------------------
The food protein-stabilized nanosuspensions were successfully prepared using a precipitation-ultrasonication method. The particle size and size distribution were unchanged after the nanosuspensions were stored at 4°C for more than 30 days, indicating excellent stability. The particle sizes/zeta potentials of the nanosuspensions stabilized with soybean protein isolate, whey protein isolate, and β-lactoglobulin were 588 nm/−23.7 mV, 320 nm/−30.8 mV, and 243 nm/−25.9 mV, respectively. The polydispersity index is a measure of the homogeneity of dispersion, with values ranging from 0 to 1, where \<0.3 suggests a homogeneous dispersion.[@b24-ijn-8-3119] The polydispersity index for nanosuspensions based on soybean protein isolate, whey protein isolate, and β-lactoglobulin was 0.17, 0.17, and 0.21, respectively, indicating a narrow particle size distribution. Further, transmission and scanning electron micrographs of the nanosuspensions revealed a needle-like morphology with particle diameters of about 200--600 nm ([Figure 2](#f2-ijn-8-3119){ref-type="fig"}), corresponding closely to the results obtained by dynamic light scattering. The stabilization effects on the nanosuspensions were attributed to two factors. Firstly, adsorption of protein onto the drug particles produced effective steric stabilization,[@b3-ijn-8-3119],[@b25-ijn-8-3119] and secondly, the surface charge from the --COOH groups on the proteins generated electrostatic repulsion, with an absolute zeta potential value of 20 mV being sufficient to maintain a stable nanosuspension.[@b26-ijn-8-3119]
Coating nanosuspensions onto pellets
------------------------------------
Given that an increase in surface area results in an increase in free energy, the stability of drug nanosuspensions is a very challenging issue during pharmaceutical product development.[@b3-ijn-8-3119],[@b8-ijn-8-3119] An ideal way of addressing this problem is to convert the nanosuspension into a solid form.[@b10-ijn-8-3119],[@b13-ijn-8-3119] Therefore, in this study, we solidified the drug nanosuspensions using an easily scalable fluid-bed coating technology ([Figure 1](#f1-ijn-8-3119){ref-type="fig"}). The nanosuspensions coated into pellets were stabilized by soybean protein isolate-150 (containing 150 mg of indomethacin in acetone), whey protein isolate-150 (containing 150 mg of indomethacin in acetone), and β-lactoglobulin-200 (containing 200 mg of indomethacin in acetone). The coating formulations and their efficiency are shown in [Table 1](#t1-ijn-8-3119){ref-type="table"}. The yield of soybean protein isolate-150, whey protein isolate-150, and β-lactoglobulin-200 was 85.7%, 92.9%, and 88.3%, respectively, indicating excellent coating efficiency, mainly attributable to the excellent film-forming properties of PVP.[@b27-ijn-8-3119] Similar yields were obtained from these three formulations, likely because the three proteins used have similar structures and physicochemical properties.[@b28-ijn-8-3119]--[@b30-ijn-8-3119] The surfaces of pellets layered with nanosuspensions stabilized by soybean protein isolate, whey protein isolate, or β-lactoglobulin were smooth ([Figure 3A1](#f3-ijn-8-3119){ref-type="fig"}--[C1](#f3-ijn-8-3119){ref-type="fig"}) and each layer of solidified nanosuspension was tightly packed and could be distinguished easily from the pellet cores ([Figure 3A2](#f3-ijn-8-3119){ref-type="fig"}--[C2](#f3-ijn-8-3119){ref-type="fig"}, [3A3](#f3-ijn-8-3119){ref-type="fig"}--[C3](#f3-ijn-8-3119){ref-type="fig"}).
Redispersibility study
----------------------
The redispersibility of solidified nanosuspensions is the most important factor and the main challenge in product development.[@b31-ijn-8-3119] Therefore, we did a redispersibility study for the solidified nanosuspension pellets. The particle size and size distribution (intensity-weight Nicomp distribution) for the redispersed nanosuspensions are shown in [Figure 4A](#f4-ijn-8-3119){ref-type="fig"}--[D](#f4-ijn-8-3119){ref-type="fig"}. The particle size of redispersed nanosuspensions stabilized by soybean protein isolate was similar to that of the original nanosuspensions, but its size distribution showed a bimodal pattern and a slight shift into the smaller range. This may be because the denatured soybean protein isolate could not be rehydrated completely after the redispersion process. A similar phenomenon was observed with dried nanocapsules based on a soybean protein isolate in another report (He and Wu, unpublished data, 2013). The particle size and size distribution for the redispersed nanosuspension stabilized by whey protein isolate were almost identical to that of the original nanosuspension. The particle size of the redispersed nanosuspension stabilized by β-lactoglobulin was slightly increased to 289 nm from the original particle size of 243 nm, and the size distribution also shifted slightly towards the larger particle size range, which could be attributed to the smaller amount of protein used in formulation of the β-lactoglobulin-200 nanosuspension in comparison with that of the nanosuspensions containing soybean protein isolate or whey protein isolate. Our previous report suggested that, to some extent, food proteins with a globular structure could act as cryoprotectants, protecting the product from the stresses of freezing and drying.[@b29-ijn-8-3119] The increased polydispersity index was ascribed to the fact that not only the redispersed nanosuspensions but also the polymer PVP contribute to the intensity versus size distribution determined by dynamic light scattering.[@b32-ijn-8-3119] Preservation of the nanoparticle diameter size after the drying process indicates that nanosuspensions stabilized by soybean protein isolate, whey protein isolate, and β-lactoglobulin can be converted successfully into a solid dosage form by fluid-bed coating technology.[@b10-ijn-8-3119]
To confirm the results of dynamic light scattering, transmission and scanning electron microscopy were carried out to determine the morphology and particle size of the redispersed nanosuspensions. As shown in [Figure 4E](#f4-ijn-8-3119){ref-type="fig"}--[G](#f4-ijn-8-3119){ref-type="fig"}, redispersed nanosuspensions stabilized by soybean protein isolate, whey protein isolate, or β-lactoglobulin had a morphology and particle size similar to that of the original nanocrystals, consistent with the results obtained by dynamic light scattering.
The synergistic protection afforded by soybean protein isolate and trehalose ensured redispersibility of the dried nanosuspensions in water. Firstly, nanocapsules stabilized by soybean protein isolate can be freeze-dried and coated onto the pellets directly, maintaining their original particle size and size distribution.[@b29-ijn-8-3119] Secondly, it was shown that the soybean protein isolate-60, whey protein isolate-60, and β-lactoglobulin-30 nanosuspension formulations could be freeze-dried directly without addition of any other cryoprotectants, with the ability to reconstitute to their original particle size and size distribution (data not shown). Trehalose clearly acts as a common protectant, and can also protect nanosuspensions against aggregation during the drying process.[@b10-ijn-8-3119] However, the concentration of trehalose in our suspensions was too low to ensure redispersibility, and too high a concentration (\>2%, w/w) would have greatly hindered the process of film coating. Another feature of the soybean protein isolate was its behavior as an antisticking agent in the coating formulation.
In vitro dissolution
--------------------
Preservation of rapid dissolution is another important property of solidified nanosuspensions.[@b13-ijn-8-3119] The dissolution profiles for the raw drug powder, original drug nanosuspensions, and pellets layered by the nanosuspensions are shown in [Figure 5](#f5-ijn-8-3119){ref-type="fig"}. Compared with dissolution of the raw drug powder, both the original drug nanosuspensions and pellets layered with the nanosuspensions showed much faster dissolution, with near complete dissolution within 5 minutes. The nanosuspension pellets showed dissolution profiles similar to those of the original drug nanosuspensions, but with a delay of 3 minutes, which could be explained by the fact that reconstitution of the compactly coated pellets took longer.[@b33-ijn-8-3119],[@b34-ijn-8-3119] It is concluded that the dissolution capacity of the original nanosuspensions was preserved in the solidified nanosuspensions.
Powder x-ray diffraction and DSC
--------------------------------
Drug dissolution, absorption/bioavailability, and stability are influenced greatly by the form of the drug particle in the solid state. Herein we assessed the solid-state form of the drug particles by powder x-ray diffraction and DSC, by comparing the dried nanosuspensions prepared using fluid-bed coating and the original nanosuspensions obtained from freeze-drying. As shown in [Figure 6A](#f6-ijn-8-3119){ref-type="fig"}, the raw crystal powder showed diffraction peaks at 11.6, 17.3, 19.6, 21.8, and 26.6, ranging from 2.5°--50° ( 2 *θ*), suggesting that the drug was highly crystalline in nature.[@b35-ijn-8-3119],[@b36-ijn-8-3119] The same diffraction peaks were also observed in samples of the physical mixture. The diffraction peaks for the drug taken from samples of the original nanosuspensions disappeared, indicating that the drug particles were present in an amorphous state. Importantly, the powder x-ray diffraction patterns for the dried nanosuspensions prepared by fluid-bed coating did not show any diffraction peaks, suggesting that the drug particles were also present in an amorphous state. The DSC pattern ([Figure 6B](#f6-ijn-8-3119){ref-type="fig"}) confirmed the results obtained by powder x-ray diffraction. At about 160°C, the endothermic peaks of the active compound and its physical mixture were observed, whereas the melting peak was absent from the dried nanosuspensions prepared by fluid-bed coating and from the original nanocrystals. Based on the results of DSC and powder x-ray diffraction, the dried drug particles in nanosuspensions prepared by fluid-bed coating are present in an amorphous state, and are not changed by the drying process.
Conclusion
----------
Nanosuspensions of indomethacin stabilized by food proteins were converted into a solid dosage form by fluid-bed coating. The solidified nanosuspension pellets preserved the redispersibility, solid state, particle size, and size distribution of the original nanosuspensions. Drug dissolution from the dried nanosuspensions was much faster than that from the raw crystals. The dissolution profile for the dried nanosuspensions was similar to that of the original nanosuspensions, save a delay of a few minutes. In summary, fluid-bed coating technology shows potential for the solidification of drug nanosuspensions. It is expected that a variety of fluidic nanoparticle dispersions could be converted into solid dosage forms using this strategy.
This study was supported financially by the National Key Basic Research Program of China (2009CB930300). Wu W is grateful to the Shanghai Commission of Education (10SG05) and Ministry of Education (NCET-11-0114) for personnel-fostering financial support. He W is grateful for the support of the Innovative Personnel Training Plan of Fudan University States Key Disciplines.
**Disclosure**
The authors report no conflicts of interest in this work.
{#f1-ijn-8-3119}
{#f2-ijn-8-3119}
{#f3-ijn-8-3119}
![Particle size and polydispersity index data for original and redispersed nanosuspensions (**A**). Particle size distribution of original and redispersed nanosuspensions from pellets \[stabilized by soybean protein isolate (**B**), whey protein isolate (**C**), or β-lactoglobulin (**D**)\]. Optical photographs of the original (**E1**) and redispersed nanosuspensions (**E2**). Transmission (**F1--3**) and scanning (**G1--3**) electron micrographs of nanosuspensions redispersed from pellets. From left to right: soybean protein isolate, whey protein isolate, and β-lactoglobulin.\
**Abbreviations:** PI, polydispersity index; SPI, soybean protein isolate; WPI, whey protein isolate; β-LG, β-lactoglobulin.](ijn-8-3119Fig4){#f4-ijn-8-3119}
{#f5-ijn-8-3119}
{#f6-ijn-8-3119}
######
Coating formulations and coating efficiency
Formulation (F) Composition for coating Yield (%)
----------------- ------------------------- ----------- --- ---- ---- ---- ---- ------
F1 3 10 1 20 50 -- -- 85.7
F2 3 10 1 20 -- 50 -- 92.9
F3 3 10 1 20 -- -- 50 88.3
**Note:** Concentration of denatured SPI in aqueous dispersion was 67 mg/mL.
**Abbreviations:** SPI, soybean protein isolate; WPI, whey protein isolate; β-Lg, β-lactoglobulin; PVP, polyvinylpyrrolidone.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Amino acids are much more than mere building blocks of proteins: their different chemical properties dictate the catalytic activity of enzymes, protein half-life and a plethora of different post-translational modifications that govern protein function. This essay concentrates on the role of cysteine, a thiol containing amino acid that can participate in a variety of chemical reactions such as post-translational oxidative modifications. Many of them are reversible at physiological conditions, thereby allowing cysteine to act as a powerful molecular switch, akin to protein phosphorylation-dephosphorylation cycles. Hence, cysteine modifications are not limited to the well-known structural role of disulfide bonds in proteins synthesized in the endoplasmic reticulum (ER), but participate in fundamental intra- and inter-cellular signaling pathways. The downside of the pleiotropic reactivity of cysteines resides is their high susceptibility to undesired activation/inactivation in conditions of redox disequilibrium (either oxidative or reductive stress). In many genetic diseases and cancer, mutations can either directly target a cysteine or affect residues that contribute to maintain optimal cysteine pKa, accessibility and/or reactivity.
Here we briefly discuss redoxstasis in cell compartments and provide examples of disease---associated modifications/mutations of key cysteine residues.
Cysteines Reactivity and Redox Homeostasis {#s2}
==========================================
Redox reactions involve the gain (reduction) or loss (oxidation) of electrons in the reacting compounds. From its reduced form (SH), the sulfur atom of a cysteine residue can undergo a wide-range of oxidative modifications (Figure [1](#F1){ref-type="fig"}). Reactivity is greatly enhanced for cysteines whose thiol side chain is in the thiolate form, i.e., deprotonated at physiological pH (S^−^), and is influenced by structural factors (Ferrer-Sueta et al., [@B28]). Disulfide bonds stabilize the tertiary and/or quaternary structures of many proteins. They also serve as regulatory functional switches, a prototype being the activation of the bacterial transcription factor OxyR in response to oxidative stress (Zheng et al., [@B101]; Jo et al., [@B44]). Progressive cysteine oxidation by H~2~O~2~ leads to cysteine sulfenylation (SOH), sulfinylation (SO~2~H) and sulfonylation (SO~3~H). Among these, oxidation to SO~3~H is regarded as irreversible. S-sulfydration (also called persulfidation) can occur after reactions between derivatives of hydrogen sulfide (H~2~S) and thiols (Mishanina et al., [@B60]). Reactive nitrogen species (RNS) like nitric oxide (NO) react with some cysteines causing S-nitrosylation/nitrosation (Evangelista et al., [@B25]). Cysteines can also undergo lipid modifications including palmitoylation and prenylation or bind metals such as Zn, Fe and Cu. This latter property is crucial for formation of zinc fingers and iron-sulfur clusters (Oteiza, [@B70]; Rouault, [@B79]). Owing to their nucleophilic properties, thiolate groups also participate in non-redox reactions as in the catalytic groups of cysteine-proteases and ubiquitin ligases. For a detailed discussion on cysteine reactivity, its chemotypes and the methods for their detection, we refer to excellent reviews (Nagy, [@B65]; Paulsen and Carroll, [@B72]; Go et al., [@B34]).
![Main post-translational modifications of cysteines. Intermolecular disulfide bonds can be formed with another protein or low molecular weight thiols (like glutathione). In general, intramolecular bonds are inserted into a reduced protein by disulfide exchange with oxidized glutathione (GSSG) or another oxidized protein (e.g., Protein disulfide isomerase, PDI), through the formation of mixed disulfides. Oxidation by reactive oxygen species (ROS) initially leads to sulfenylation (SOH). Because of its relative instability, sulfenylated cysteine can promote intramolecular disulfide bond formation or additionally react with ROS leading first to sulfinylation (SO~2~H) and then to sulfonylation (SO~3~H). While SO~2~H can be reversed through the catalytic activity of the cytoplasmic enzyme sulfiredoxin-1 (SRXN-1; Biteau et al., [@B10]), SO~3~H is so far considered irreversible. Palmitoylation can also take place through creation of thioester bonds between palmitate and cysteine (Fukata and Fukata, [@B32]).](fnmol-10-00167-g0001){#F1}
Cysteine thiols are key players in conditions of oxidative stress. Most non-protein antioxidants as well as antioxidant enzymes are thiol based. Glutathione (GSH, γ-L-Glutamyl-L-cysteinylglycine) acts as a redox buffer and a cofactor of many enzymes including glutathione peroxidases (Gpx) that scavenge peroxides generating oxidized glutathione (GSSG). In humans, there are eight Gpxs, localized in different compartments (Brigelius-Flohé and Maiorino, [@B11]; Figure [2](#F2){ref-type="fig"}). Other key peroxide scavengers are peroxiredoxins (Prx; Perkins et al., [@B73]). Of the six human Prxs, two are localized in mitochondria and one in the ER. Thioredoxins (Trx) and glutaredoxins (Grx) reduce oxidized protein thiols. Oxidized Trx and Grx are reduced by Trx reductases (TrxR) and GSH, respectively (Holmgren, [@B40]; Mustacich and Powis, [@B64]; Fernandes and Holmgren, [@B27]). Glutathione reductase (GR) is also a key player for redox homeostasis, replenishing the GSH pool at the expense of GSSG. It is important to stress that both GR and TrxR rely on the NADPH/NADP system for their activity, thus establishing a link between the nicotinamide and thiol redox systems (Jones and Sies, [@B46]). Another important player is sulfiredoxin-1, an ATP-dependent enzyme capable of reducing sulfinylated proteins (Biteau et al., [@B10]; Mishra et al., [@B61]).
![Cellular compartments differ in their redox poise. The left part of the figure summarizes the data available in the literature concerning glutathione redox potential values (E~h~ GSSG) in intra- and extra-cellular compartments. Depending on the cell types, physiological conditions and methods used, the results can vary rather significantly. There remains no doubt, however, that mitochondria and cytosol are far more reducing than the endoplasmic reticulum (ER) and extracellular space. Noteworthy, Thioredoxin 1 (Trx1) and Trx2 display a more reducing Kox, confirming their pivotal role in maintaining a suitable redox in the cytosol. Owing to the permeability of nuclear pores, the nucleus is likely to have values similar to the cytosol (values are from Gutscher et al., [@B35]; Jones and Go, [@B45]; Kojer et al., [@B49]; Birk et al., [@B8]; Kirstein et al., [@B48]). The right part highlights instead the main redox control systems in the cytosol, mitochondria and ER. Note the presence in the ER of proteins promoting formation of disulfide bonds (endoplasmic reticulum oxidoreductin 1 (Ero1), quiescin sulfhydryl oxidase (QSOX), PDI, ERp44, etc.) and also the absence of Glutathione reductase (GR), thus contributing to higher GSSG/GSH ratios in the ER compared to the cytosol.](fnmol-10-00167-g0002){#F2}
Redox Compartmentalization {#s3}
==========================
The organelles of eukaryotic cells can differ dramatically with respect to the redox poise of their various redox couples (Figure [2](#F2){ref-type="fig"}). The chemistry of primordial cells evolved in an oxygen-free atmosphere, and cytosolic and mitochondrial cysteines tend to remain in the reduced states (Go et al., [@B34]). Incidentally, this is why most cell-free protocols that recapitulate nuclear or cytoplasmic reactions include the addition of DTT or other reductants to work efficiently. *In vivo*, the establishment of stable disulfide bonds in these compartments is extremely unfavorable because of the combined reducing power of GSH, Grxs and Trxs.
However, there are places where disulfide bonds need to be inserted into selected proteins. For example, the Mia40/Erv1 relay allows formation of disulfide bonds in proteins and their import into the inter-membrane mitochondrial space, where the low Grx pool kinetically favor this process (Mesecke et al., [@B59]; Kojer et al., [@B50]; Erdogan and Riemer, [@B24]).
The redox couples present in organelles of the exocytic pathway display redox poises similar to the extracellular space. In the ER, nascent membrane and secretory proteins form disulfide bonds, preparing for their adult life in the oxidizing extracellular environments. Thus, oxidative power is needed in the ER, but redox conditions must be tightly controlled to allow isomerization or reduction of non-native disulfides. The occurrence of opposite reactions is guaranteed by protein relays capable of selectively inserting or removing disulfides (Fassio and Sitia, [@B26]; Hagiwara and Nagata, [@B36]). Oxidative folding is catalyzed by enzymes of the protein disulfide isomerase (PDI) superfamily, which receive oxidative power from oxidases like endoplasmic reticulum oxidoreductin 1 (Ero1) and quiescin sulfhydryl oxidase (QSOX; Bulleid and Ellgaard, [@B12]; Hudson et al., [@B42]). The human genome encodes for numerous PDI-like enzymes whose activity largely depends on the number of Trx-like domains and the redox potential of their CXXC motifs (Hatahet and Ruddock, [@B39]; Okumura et al., [@B69]). Depending on the surrounding redox and ionic conditions, oxidoreductases may oxidize, isomerize or reduce disulfides. In this wide range of activities, ERdJ5 is most suited for reducing disulfides (Dong et al., [@B23]; Ushioda et al., [@B87]). TrxR and import of cytosolic GSH have been proposed as reducing powers to prevent ER hyperoxidation (Molteni et al., [@B62]; Appenzeller-Herzog, [@B3]; Poet et al., [@B74]).
Disulfide interactions with PDI-like enzymes provide key quality control of the secretome, preventing the release of immature proteins. For instance, ERp44, captures proteins with exposed thiols and redox-active enzymes lacking suitable ER localization signals (e.g., Ero1, Prx4, Sumf1), retrieving them to the ER (Vavassori et al., [@B200]; Anelli et al., [@B1]).
Cysteines as Redox Molecular Switches {#s4}
=====================================
Mechanisms ensuring tight redoxstasis control are present in the three cellular compartments where protein folding takes place (cytosol, ER and mitochondria) and they are intimately linked to protein quality control (Anelli et al., [@B1]). However, cells additionally exploit cysteine reactivity for purposes other than oxidative protein folding, namely as switches regulating signaling and adaptive responses.
A prototypic example is provided by the Nuclear factor erythroid 2-related factor 2 (Nrf2), a transcription factor whose nuclear translocation is prevented by interactions with Kelch-like ECH-associating protein 1 (Keap1). Upon oxidation, Keap1 dissociates from Nrf2, which can reach the nucleus and promote transcription of antioxidant response genes (Dinkova-Kostova et al., [@B22]). Recently, unexpected links between Nrf2, redox and ER stress emerged. Ire1 is a transmembrane protein that initiates the unfolded protein response upon accumulation of misfolded proteins in the ER lumen. Upon oxidative stress, Ire1 is sulfenylated and activates the Nrf2 pathway, abandoning its canonical ER stress sensing function (Hourihan et al., [@B41]). Thus, a subtle cysteine modification can shift the pathway to which a signal transducer is affiliated.
Redox modifications also play key roles in regulating protein tyrosine phosphorylation. Cysteine oxidative modifications such as SOH, disulfide formation and S-nitrosylation inhibit phosphatase and tensin homolog (PTEN) and other protein tyrosine phosphatases by interfering for example with their cysteine-dependent catalytic activity (Numajiri et al., [@B68]; Corcoran and Cotter, [@B18]; Pulido, [@B75]). As an example of physiological importance, abolishing this rheostat circuit dampens B lymphocyte activation and antibody production (Bertolotti et al., [@B5]). Moreover, a growing body of evidence supports the redox regulation of several tyrosine kinases, as described for c-Src (Giannoni and Chiarugi, [@B33]) and Janus kinase 2 (JAK2; Smith et al., [@B84]).
High mobility group protein B1 (HMGB1) is a DNA-binding nuclear protein that can be released by stressed cells. In the extracellular space, HMGB1 mediates inflammation or tissue repair, according to its redox state. If fully reduced, it binds to Advanced glycation end product-specific receptor (RAGE) and C-X-C chemokine receptor type 4 (CXCR4) and activates cell migration and autophagy. Upon formation of an intramolecular disulfide bond, HMGB1 binds Toll-like receptor 4 (TLR4)/MD-2 receptors complex and stimulates cytokine secretion. Sulphonylation then inactivates HMGB1, highlighting how a protein switches function depending on its cysteine redox state (Fiuza et al., [@B30]; Venereau et al., [@B89]; Vénéreau et al., [@B90]). Oxidation of a conserved cysteine residue also modulate the permeability of aquaporin 8 by reversibly inhibiting the transport of H~2~O~2~ and H~2~O across the membrane of stressed cells (Medraño-Fernandez et al., [@B202]).
Another example of redox-based functional re-targeting aimed to prevent protein aggregation is the induction of holdase activity in the ER chaperone Immunoglobulin heavy chain-binding protein (BiP) by cysteine oxidation (Wei et al., [@B97]; Wang et al., [@B94]). Similarly, Prx sulphinylation promotes formation of homo-oligomers endowed with chaperone activity (Jang et al., [@B43]; Hanzén et al., [@B37]).
Thus, cysteine modifications are key in many intra- and inter-cellular signaling and adaptive pathways. The sub-compartmental organization of redoxstasis, based on spatially constrained protein relays (Woo et al., [@B98]), and the low diffusibility of small redox active compounds such as H~2~O~2~ (Bienert and Chaumont, [@B6]) can explain how redox-dependent signals can propagate in the presence of powerful antioxidant systems.
Cysteines and Diseases {#s5}
======================
Secretory Proteins {#s5-1}
------------------
Owing to the importance of structural and regulatory disulfide bonds in membrane and secretory proteins, mutations in luminal cysteines generally have dramatic consequences. Hence, acquisition or loss of a cysteine often causes retention of the mutated protein in the ER by thiol-mediated mechanisms (Anelli et al., [@B1]), with consequent loss or gain of function. The difficulty in forming the proper array of disulfide bonds in the cysteine-rich domains of many membrane receptors can lead to ER retention and degradation (loss of function), but also gain of function by interchain disulfide bonding that chronically activates signal transduction. An astonishing example comes from type 2A multiple endocrine neoplasia (MEN2A). This severe condition is often due to mutations in the cysteine-rich luminal portion of a tyrosine kinase receptor, RET, with strong genotype-phenotype correlations. The oncogenic hit is the formation of ligand-independent, covalent homodimers that constitutively deliver growth signals (Asai et al., [@B4]; Mulligan, [@B63]). RET malfunction can also lead to congenital abnormalities characterized by failure of neuroblast migration and defective maturation of the enteric nervous system (Hirschsprung disease), a condition that in some families coexisted with MEN2A (Takahashi et al., [@B85]; Frank-Raue et al., [@B31]).
Many genetic diseases are caused by gain or loss of a cysteine in secretory or membrane proteins. Aberrant thiol-mediated interactions via unpaired cysteines can directly provoke ER retention and aggregation besides misfolding. In Pelizaeus-Merzbacher disease, a myelination defect, a subgroup of mutations affecting the extracellular loop of the PLP/DM20 protein impair formation of intramolecular disulfide bridges and cause abnormal protein cross-links, ER retention and oligodendrocyte death (Dhaunchak et al., [@B20]). Similar mechanisms have been demonstrated in some forms of autism (Comoletti et al., [@B17]), color blindness (Patel et al., [@B71]) and von Willebrand disease (Wang et al., [@B95]).
Mutations of uromodulin, causing medullary cystic kidney disease/familial juvenile hyperuricemic nephropathy most often affect one of the 48 conserved cysteine residues (Rampoldi et al., [@B76]; Scolari et al., [@B80]). Cysteine mutations and aberrant disulfide bonding underlie the pathogenesis of CD40 deficiency (Lanzi et al., [@B52]), TNFR1-associated periodic fever syndrome (Lobito et al., [@B54]) and MiDY insulin-deficient diabetes (Liu et al., [@B53]). Mutations causing conformational alterations of alpha-1-antitrypsin make its only cysteine more prone to form aberrant disulfide bonds in the ER, thus facilitating the intracellular retention and polymerization of alpha-1-antitrypsin in Alpha-1-antitrypsin deficiency (AATD; Ronzoni et al., [@B78]).
Marinesco-Sjogren is a syndrome causing ataxia, intellectual disability and muscle weakness. This rare disease is caused by mutations in Sil1, a cofactor of BiP (Anttonen et al., [@B2]; Krieger et al., [@B51]). In yeast recovering from stress, Sil1 reduces oxidized Kar2, the paralog of human BiP, restoring its normal foldase activity (Siegenthaler et al., [@B83]). It remains to be seen whether and how mutations in Marinesco-Sjogren patients also impact the reductase function of Sil1.
Cytosolic and Nuclear Proteins {#s5-2}
------------------------------
In cytosolic proteins, cysteines can be direct targets of mutations, but more frequently they are dysregulated or inactivated by oxidative stress or other environmental conditions. Both mechanisms have been shown in Parkinson's disease (PD). Parkin (PARK2) is an E3 ubiquitin ligase whose dysfunction causes accumulation of protein aggregates, endangering dopaminergic neurons (Charan and LaVoie, [@B13]). Parkin is highly expressed in the brain and frequently mutated in autosomal recessive juvenile PD (Biskup et al., [@B9]). These mutations often affect cysteines, causing loss of function and decreased stability of the enzyme (Wang et al., [@B93]; Seirafi et al., [@B81]). Parkin can also be inactivated by S-nitrosylation or sulphonylation (Chung et al., [@B16]; Meng et al., [@B58]). Recent studies describe interesting interplays between parkin oxidative modifications, its role in mitochondrial quality control and PD onset (Zhang et al., [@B99]). In dopaminergic neurodegenerative disorders, a key pathogenetic event is also the inactivation of tyrosine hydroxylase, a rate-limiting enzyme in dopamine and norepinephrine biosynthesis, by oxidative injury (Di Giovanni et al., [@B21]).
Mutations of the antioxidant superoxide dismutase gene (SOD1) are linked to about one fifth of the cases of familial amyotrophic lateral sclerosis, a degenerative disorder of motor neurons. Wild type SOD1 is a covalent disulfide-linked homodimer localized in part in the mitochondrial intermembrane space. Pathogenic SOD1 mutants form high molecular weight oligomers, inducing mitochondrial dysfunctions (Ferri et al., [@B29]; Magrané et al., [@B55]). Noteworthy, intermolecular disulfide cross-links and glutathionylation enhance mutant SOD1 aggregation (Cozzolino et al., [@B19]; Redler et al., [@B77]; McAlary et al., [@B56]), cysteine 111 being a key residue (Valle and Carrì, [@B88]).
In addition, Alzheimer's disease (AD) is associated with thiol modifications, in particular S-nitrosylation. NO is produced in the brain by neuronal NO synthase (nNOS) and serves as a key second messenger for instance, regulating neuronal plasticity and survival (Nakamura and Lipton, [@B67]; Chong et al., [@B15]). However, aberrant S-nitrosylation of proteins such as PDI and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) can occur in AD (Uehara et al., [@B86]; Zhao et al., [@B100]). S-nitrosylation of GAPDH enhances its binding to the ubiquitin ligase Siah1. GAPDH/Siah1 complexes accumulate in the nucleus triggering neuronal apoptosis via excessive protein degradation and trans-nitrosylation signaling cascades (Hara et al., [@B38]; Sen and Snyder, [@B82]; Nakamura and Lipton, [@B66]).
Numerous examples of "gain of cysteine" mutations are found in cancer, p53, KRAS and other oncogenes being preferred targets. The acquired cysteines cause decreased stability or impaired DNA binding of the tumor suppressor p53, while KRAS oncogenes are constitutively activated. Noteworthy, such acquired cysteines are potential targets for antitumor treatments (Visscher et al., [@B91]).
Concluding Remarks {#s6}
==================
The multiple chemical reactions of cysteines and their reversibility in physiological conditions make them ideal tuneable devices for regulating protein function. Indeed, evolution has increasingly exploited the regulatory potential of cysteine chemistry as atmospheric oxygen became more abundant and complex multicellular organisms evolved. The frequency and conservation of this amino acid is indeed higher in mammals (\>2% of the proteome) than in prokaryotes (0.5%). The examples provided in this essay reveal the pathophysiological relevance of cysteine redox modifications in the different compartments of human cells. Disulfide bonds prevail in the exocytic and endocytic compartments, organelles which are in direct contact with the oxidizing extracellular environment. These covalent bonds increase protein stability, facilitate quality control (Medraño-Fernandez et al., [@B57]) and underlie the functional regulation of many secreted proteins. A wider range of modifications acts in the cytosol and mitochondria, whose chemistry reflects their origin in an oxygen free atmosphere. Cysteine residues in these compartments are largely found in the reduced thiol/thiolate state, which permits regulation of protein function and activity by way of a wide-range of oxidative post-translational modifications. The redox gradients that form within and amongst cells hence provide ample opportunities to regulate signaling, transcription and other key biological processes. The price we pay is the many diseases caused by cysteine mutations or oxidative deregulation. Novel reagents (Chen et al., [@B14]; Kim et al., [@B47]; Bilan and Belousov, [@B7]; Wagener et al., [@B92]) are being developed to better understand cysteine modifications and their links with disease, ultimately offering ample practical exploitations (Nakamura and Lipton, [@B67]; Wani and Murray, [@B96]).
Author Contributions {#s7}
====================
AF, EDY and RS discussed the concepts and pitfalls, and wrote the manuscript.
Conflict of Interest Statement {#s8}
==============================
The 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.
The authors thank Dr. Mattia Laffranchi (University of Brescia, Italy), Dr. Luca Rampoldi (San Raffaele Scientific Institute, Milan, Italy), Dr. Bruce Morgan (University of Kaiserslautern, Germany) and the members of their labs for stimulating discussions. AF thanks Fondazione Cariplo (2013-0967), EDY (PCOFUND-GA-2010-267264 INVEST), RS AIRC (IG14559), Fondazione Banca del Monte di Lombardia Fondazione Cariplo (2015-0591), Ministero della Salute (PE-2011-02352286) and Telethon (GGP15059) for generous support. We apologize to the many colleagues whose fundamental work could not be cited for space constraints.
AATD
: Alpha-1-antitrypsin deficiency
BiP
: Immunoglobulin heavy chain-binding protein
CFTR
: Cystic fibrosis transmembrane conductance regulator
CXCR4
: C-X-C chemokine receptor type 4
ER
: Endoplasmic reticulum
ERAD
: ER associated degradation
Ero1
: Endoplasmic reticulum oxidoreductin 1
Gpx
: Glutathione peroxidase
GR
: Glutathione reductase
Grx
: Glutaredoxin
GSH/GSSG
: Reduced/oxidized glutathione
HMGB1
: High mobility group protein B1
KEAP1
: Kelch-like ECH-associated protein 1
MD2
: Lymphocyte antigen 96
Nrf2
: Nuclear factor erythroid 2-related factor 2
PDI
: Protein disulfide isomerase
Prx
: Peroxiredoxin
PTEN
: Phosphatase and tensin homolog
QSOX
: Quiescin sulfhydryl oxidase
RAGE/ARGE
: Advanced glycosylation end product-specific receptor
ROS
: Reactive oxygen species
SOD1
: Superoxide Dismutase 1
TLR4
: Toll-like receptor 4
TNF
: Tumor necrosis factor
TNFR
: TNF receptors
Trx
: Thioredoxin
TrxR
: Thioredoxin reductase
VkOR
: Vitamin K epoxide reductase.
[^1]: Edited by: Angelo Poletti, Università degli Studi di Milano, Italy
[^2]: Reviewed by: Annalisa Pastore, King's College London, United Kingdom; Maria Teresa Carri, University of Rome Tor Vergata, Italy
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
The lupus anticoagulant (LA), most commonly an immunoglobulin, is an immediate-acting coagulation inhibitor found in a variety of autoimmune disorders and sometimes found in otherwise healthy individuals.[@b1-jbm-2-087] It appears to be directed specifically against the phospholipids moiety of prothrombinase complex formed by the interaction of factors Xa, Va, platelet phospholipids, and calcium.[@b2-jbm-2-087] It was first described in 1952.[@b3-jbm-2-087] and its strong association with thromboembolic phenomenon, spontaneous miscarriage, and stillbirth was established.[@b4-jbm-2-087]--[@b6-jbm-2-087]
LA is an acquired autoantibody that binds to phospholipids' active coagulation factors, which slows down the rate of thrombin generation and therefore retards clot formation in vitro,[@b6-jbm-2-087] but promotes both venous and arterial thrombosis in vivo.[@b7-jbm-2-087] This paradoxical association between in vitro anticoagulant effect and in vivo prothrombotic state activity of this autoantibody is not fully understood.
While antiphospholipid syndrome (APS) is known to be one of the most important causes of acquired hypercoagulable states[@b8-jbm-2-087] and specifically causes late pregnancy loss, some studies found an association of 7% to 10% between recurrent spontaneous abortions in the first trimester and LA.[@b6-jbm-2-087]
Placental vessel thrombosis with ischemia,[@b9-jbm-2-087] which starts early in pregnancy, and spiral artery vasculopathy,[@b10-jbm-2-087] which begins at about 8 weeks and is complete by 16 to 20 weeks of gestation, are regarded as the causes of pregnancy loss.
Information about LA frequency in pregnant women with history of recurrent fetal loss is sparse in this part of the world, hence this study.
Materials and methods
=====================
Subjects
--------
Before the study began, approval was obtained from the ethical review committee of Oyo State Hospitals Management Board and informed consent from individual patients. Fifty women with intact pregnancy but who had had at least two previous unexplained pregnancy losses (ie, at least gravida 3), irrespective of their gestational age, were consecutively recruited into the study at the antenatal clinic of Adeoyo Maternity Hospital, Yemetu, Ibadan between January and April 1998. The age range was 26 to 39 years (average 31.6 years). Their antenatal records did not show comorbidities attributable to pregnancy loss.
Sample collection
-----------------
Venous blood (4.5 mL) was carefully collected through clean venupucture from each patient into a clean plastic tube containing 0.5 mL of 3.8% trisodium citrate (9 parts venous blood: 1 part anticoagulant). After thorough but gentle mixing, the sample was centrifuged for 15 minutes at 1500 g at 25°C and platelet-poor plasma (PPP) was immediately and carefully aspirated and stored on ice in the freezer at −20°C prior to use.
Using a similar method PPP was obtained from each of the individual's voluntary relative donors, which was pooled but discounted and stored in 2 mL aliquot at −20°C.
Coagulation tests
-----------------
Prothrombin time was determined by the original technique by Quick.[@b11-jbm-2-087] Also partial thromboplastin time with kaolin was performed using the original method by Proctor and Rapparport.[@b12-jbm-2-087]
A kaolin clotting time (KCT) test, which is essentially an activated partial thromboplastin test but without any added phospholipid, was performed as previously described by Exner et al[@b13-jbm-2-087],[@b14-jbm-2-087] by preincubating 0.2 mL of citrated plasma with 0.1 mL of kaolin suspension (20 g/L in tris buffer pH 7.4) for 3 minutes at 37°C.
Method
------
A mixing experiment was performed using dilutions of normal patient plasma prepared as shown in [Table 1](#t1-jbm-2-087){ref-type="table"}. Kaolin is added and then calcium to initiate coagulation. The KCT is the time from adding calcium to clot formation.
KCT ratio,[@b14-jbm-2-087] ie, a ratio of KCT at 20% test plasma to KCT at 100% normal control plasma, was then calculated for each of the samples $$\frac{\text{KCT}{({80\%\,\text{normal}:\, 20\%\,\text{test}})}}{\text{KCT}\, 100\%\,\text{normal}:} = \geq 1.2$$
A ratio of ≥1.2 is considered positive for a lupus anticoagulant.
Results
=======
Out of the 50 recruited pregnant women, 34 (68%), 7 (14%), and 9 (18%) were in first, second, and third trimester, respectively.
[Table 2](#t2-jbm-2-087){ref-type="table"} summarizes the outcome and interpretation of results in 50 pregnant women with history of fetal loss. The percentage dilution of neat/Test (N/T) plasma was plotted against time on a Lin-Lin graph for the 15 patients (30%) that had prolonged KCT; of these, 12 (24%) showed graphical evidence of LA in circulation and 3 (6%) showed graphical evidence of LA as well as deficiency in the cofactor necessary for full inhibitory effect.
Discussion
==========
The APS is defined by the presence of antiphospholipid antibody or LA, usually in high titer, and any or all of the following clinical events: recurrent thromboses, recurrent fetal losses, and thrombocytopenia.[@b15-jbm-2-087] Also Livedo reticularis is considered to be an additional marker for the disease.[@b16-jbm-2-087] Antiphospholipid antibody is identified either with an enzyme linked immunosorbent assay (ELISA), which commonly uses cardiolipin as the phospholipid's antigen or by finding an LA with clotting tests. While ELISA for antiphospholipid antibody is known to be sensitive, but not very specific for predicting clinical events, LA with clotting tests, on the other hand, are specific but less sensitive.[@b17-jbm-2-087] Although antiphospholipid antibody thus occurs in a variety of situations in healthy individuals, in association with infections such as syphilis, in HIV-1, hepatitis C, and cytomegalovirus and in relation to medications, autoimmune antiphospholipid antibodies have higher titers and are more commonly of IgG isotype (mainly IgG2 and IgG4 subclasses), have higher avidity, and require presence of a cofactor.[@b15-jbm-2-087],[@b17-jbm-2-087],[@b18-jbm-2-087] APS is a heterogeneous disorder both in terms of clinical manifestation and the range of autoantibodies.
Although antiphospholpid antibody causes prolongation of phospholipid-dependent clotting tests in vitro, it causes thrombosis in vivo. Different mechanisms thought to be responsible for this in vivo thrombosis and ischemia include inhibition by antibodies of antithromb-independent anticoagulant mechanisms, activated protein C, and inhibition of fibrinolysis.[@b19-jbm-2-087] Other mechanisms are increased plasma concentration of soluble tissue factor and tissue pathway inhibitor, increased monocyte expression of tissue factor, and procoagulant activity in some patients with the syndrome.[@b20-jbm-2-087] Also, the autoantibody can inhibit prostacyclin secretion and promote release of von Willebrand factor by endothelia cells in vitro.[@b21-jbm-2-087],[@b22-jbm-2-087] Finally, platelet activation also plays a role in APS, particularly in arterial thrombosis. Therefore a thrombotic basis for pregnancy failure in APS is highly supported by the finding of decidual vasculopathy and placental infarction observed by Rand et al.[@b23-jbm-2-087]
In the present study, 60% of the patients have prolonged prothrombin time (18% mildly prolonged and 42% severely prolonged). This phenomenon has been observed in the past that LA occasionally increases the prothrombin time, and in turn the international normalized ratio, and therefore monitoring anticoagulation with warfarin becomes difficult.[@b24-jbm-2-087],[@b25-jbm-2-087]
Also our results agree with those of Haywood and Brown[@b26-jbm-2-087] who observed approximately 10% to 15% presence of LA in all the patients undergoing recurrent pregnancy loss.
The diagnosis of LA requires a high degree of suspicion. Many other manifestations should arouse this high degree of suspicion for LA apart from recurrent fetal loss, including unexplained thrombocytopenia, pulmonary hypertension, history of thrombotic events, cerebral vascular diseases in the young, unexplained infertility, autoimmune hemolytic anemia, thrombotic endocarditic mimicking rheumatic heart disease without history, and other features of rheumatic fever.
This high prevalence cannot be solely attributable to primary APS bearing in mind the heterogeneous nature of the antibody and the protean manifestation of its clinical picture. As enumerated by Greaves[@b7-jbm-2-087] LA is also detectable in association with infections, ie, viral infections such as HIV-1, varicella, and hepatitis C, bacteria infections such as syphilis, and parasitic infections such malaria, which is endemic in this part of the world. Also various drugs such as guanidine, hydralazine, procainamide, and phenytoin may be responsible, as may be miscellaneous associations such as sickle cell disease, autoimmune hemolytic anemia, and autoimmune thrombocytopenia. Therefore strong relevant clinical information is required to make a meaningful decision.
**Disclosure**
The authors declare no conflicts of interest.
######
Dilutions of normal patient plasma
**Normal plasma** **Patient plasma**
------------------- --------------------
100% 0%
90% 10%
80% 20%
50% 50%
20% 80%
10% 90%
######
Outcome of partial thromboplastin time with kaolin (PTT~k~) and kaolin clotting time (KCT) in 50 pregnant women with recurrent pregnancy loss
**Tests** **n** **Within normal** **Mild prolongation** **Very prolonged**
----------- ------- ------------------- ------------------------ ------------------------------
PTT~k~ 50 14 -- 36 (72%)
KCT 50 35 3 (6%) (LA + cofactor) 12 (24%) (LA in circulation)
**Abbreviation:** LA, lupus anticoagulant.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-metabolites-10-00255}
===============
In view of the resemblance between mice (*Mus musculus*) and humans on genetic, anatomical, physiological, and pathophysiological levels, the mouse is nowadays considered to be the most common model in biomedical research, as well as the most published and well-characterized \[[@B1-metabolites-10-00255],[@B2-metabolites-10-00255]\]. According to the latest report concerning the number of animals used for scientific purposes in the Member States of the European Union in 2017, mice constituted 61% of all animals used in studies \[[@B3-metabolites-10-00255]\]. There currently exists hundreds of established inbred, outbred, and transgenic mouse strains with defined genetic backgrounds and unique features, such as coat color, behavior, metabolism, fertility, immune function, and other physiological traits \[[@B4-metabolites-10-00255]\].
Among the most commonly used strains, the inbred C57BL/6 and BALB/c lines, as well as the CD1 outbred stock, are particularly predominant in research. As these strains are considered to be general-purpose models, they are extensively and often interchangeably used in many different disciplines of biomedical research. The first two of the above-mentioned strains are considered 'classical' inbred strains, signifying that the animals are genetically homogeneous and individual mice within a strain are identical clones of their parents and siblings \[[@B5-metabolites-10-00255]\]. Generally speaking, the use of inbred strains has become standard for research in most areas of mouse biology, as well as for basic and preclinical investigations. Currently, the most common strain, C57BL6, is used in a wide variety of research areas, including cardiovascular and developmental biology, diabetes and obesity, genetics, immunology, neurobiology, and sensorineural research. BALB/c mice are currently among the five most widely used inbred strains in biomedical applications, and a particular favorite in immunology and infectious disease research \[[@B6-metabolites-10-00255]\]. CD1 mice, in contrast to the two strains mentioned above, is an outbred stock that is bred specifically to maximize genetic diversity and heterozygosity within a population. The CD1 line is a general multipurpose model that can be used in fields such as toxicology (safety and efficacy testing), aging, and oncology \[[@B5-metabolites-10-00255],[@B7-metabolites-10-00255]\].
Metabolomics, or metabolic phenotyping, encompasses the analysis of all low-molecular-weight (\< 1 kDa) components in a biological sample, and it is generally considered to be a field of research complementary to proteomics, genomics, and transcriptomics. The metabolome, representing the total number of metabolites within a given biological system, may consist of metabolites solely under endogenous control, and may also involve those originating from exogenous sources \[[@B8-metabolites-10-00255],[@B9-metabolites-10-00255]\]. Sets of small metabolites attainable in metabolomics are closely related to the phenotypes of living organisms and provide information on biochemical activities by reflecting the substrates and products of cellular metabolism \[[@B10-metabolites-10-00255],[@B11-metabolites-10-00255],[@B12-metabolites-10-00255]\]. The metabolome is dynamic and susceptible to many factors such as environment, genetic modifications, changes in gut microflora and altered kinetic activity of enzymes, and can thus provide insights into current cell status as well as describing an actual health condition of the system being studied \[[@B13-metabolites-10-00255]\]. One of the analytical chemistry strategies applied in metabolomics is to carry out untargeted analysis of the metabolism so as to investigate the system on a global, relatively unbiased scale, thus aiming to measure the broadest range of metabolites present in an extracted sample without a priori knowledge of the metabolome \[[@B14-metabolites-10-00255],[@B15-metabolites-10-00255]\]. Owing to its high sensitivity and high throughput, this method has gained wide attention as a method for profiling endogenous metabolites. Untargeted tissue metabolomics in particular can provide valuable insights into the physiological characteristics of the body and greatly enhance our understanding of the biochemistry and metabolism of biological systems \[[@B12-metabolites-10-00255],[@B16-metabolites-10-00255]\]. However, successful characterization of the metabolome of tissue is often burdened by challenging steps such as tissue collection, quenching of the metabolism, homogenization, and metabolite extraction \[[@B17-metabolites-10-00255],[@B18-metabolites-10-00255]\].
Solid phase microextraction (SPME) has been proven to be a powerful, low invasive tool for tissue analysis in untargeted metabolomics studies \[[@B19-metabolites-10-00255]\], overcoming all the challenges mentioned above while successfully capturing unstable and short-lived metabolites often undetectable via traditional methods. SPME is a relatively young sample preparation method with unique features that enable its successful application in animal studies as an alternative to standard protocols \[[@B20-metabolites-10-00255],[@B21-metabolites-10-00255]\]. Choices regarding pretreatment methodology play an extremely important role in metabolomic studies since sample pretreatment may affect not only the molecular features available, but also the biological interpretation of the obtained chromatographic data \[[@B17-metabolites-10-00255]\]. SPME technology uses special fibers coated with biocompatible sorbents (thin film of polymer) that can be inserted directly into the tissue (brain, muscle, liver, lungs, etc.) in order to extract small molecules in amounts proportional to their biologically active unbound concentrations. SPME has a number of advantages, including simplicity, high sensitivity and a relatively low invasive nature. Moreover, it is a non-exhaustive sample preparation procedure based on chemical biopsy that combines sampling, sample preparation, rapid metabolism quenching, and extraction into a single step \[[@B20-metabolites-10-00255],[@B21-metabolites-10-00255],[@B22-metabolites-10-00255],[@B23-metabolites-10-00255],[@B24-metabolites-10-00255]\].
Analysis of tissue as a primary site of any dysregulation has been one of the main goals of biomedical experiments using animal models. Therefore, the aim of the present study was to compare the tissue metabolome of selected organs (liver, kidney, brain, and thigh muscle) in the three mouse strains commonly used in biomedical research: C57BL/6, BALB/c and CD1. The outbred stock (CD1) was chosen for this study to find out to what extent an outbred line differs in its metabolome compared to inbred strains. We expect that having an insight into the metabolomic profile of different organs in intact animals will give us an opportunity to find out how much these three popular strains differ among each other in terms of the basic metabolic processes occurring in the body.
2. Results {#sec2-metabolites-10-00255}
==========
[Figure 1](#metabolites-10-00255-f001){ref-type="fig"} presents an ion map of molecular features after a data filtration step, whereas an ion map analysis of all molecular features detected in mice tissues is presented in [Figure S1](#app1-metabolites-10-00255){ref-type="app"}.
Principal component analysis (PCA) was used to confirm the quality of instrumental analysis ([Figure S2](#app1-metabolites-10-00255){ref-type="app"}) and investigate the differences in the metabolic profiles of organs for all (673) detected features among C57BL/6, BALB/c and CD1 mice. The two-dimensional score plots (PC1 vs. PC2) presented in [Figure 2](#metabolites-10-00255-f002){ref-type="fig"} revealed major differences in metabolomic patterns of brain ([Figure 2](#metabolites-10-00255-f002){ref-type="fig"}a) and liver ([Figure 2](#metabolites-10-00255-f002){ref-type="fig"}b) tissue between the examined strains. The metabolome of kidney tissue in BALB/c mice distinctly separated from C57BL/6 and CD1 mice, whereas data points in the score plots for C57BL/6 and CD1 mice had mostly overlapping distributions ([Figure 2](#metabolites-10-00255-f002){ref-type="fig"}c). Likewise, for muscle tissue, the BALB/c strain separated clearly from C57BL/6 and CD1 mice, whereas distribution differences between C57BL6 and CD1 mice were not so evident ([Figure 2](#metabolites-10-00255-f002){ref-type="fig"}d). Clusters were noticeably larger for BALB/c mice in kidney, liver, and muscle tissue, but similar for brain tissue among all examined strains. Additionally, three-dimensional (3D) PCA score plots (representing relationships between PC1, PC2, and PC3) are shown in [Figure S3](#app1-metabolites-10-00255){ref-type="app"}. On the whole, 3D scatter plots revealed similar cluster separations to those of two-dimensional plots. Clear separation of strains within the brain and liver tissues presented in 2D plots was confirmed by three-dimensional visualization. Moreover, many more scattered points for BALB/c compared to the other two mouse strains was also observed.
Identification of detected features was performed on Compound Discoverer 2.1 software, whereas classification of putative metabolites on the basis of their chemical taxonomy was done with the use of Human Metabolome Database HMDB. Selected compounds belonging to a variety of metabolite classes ([Table 1](#metabolites-10-00255-t001){ref-type="table"}) were subjected to analysis of variance (ANOVA), revealing that most of the presented metabolites differentiated (*p* \< 0.05) studied strains within a given organ ([Table 1](#metabolites-10-00255-t001){ref-type="table"}). Metabolites differentiating all the examined tissues were found to predominantly belong to alpha-amino acids and derivatives, purine nucleotides and fatty acid ester groups. In addition, metabolomic profiles of brain tissue differed between strains in the levels of metabolites belonging to purine derivatives, purine and pyrimidine nucleosides and derivatives, hydroxy fatty acids and fatty amides, as well as alcohols and polyols. Liver tissue profiles differentiated with respect to levels of purine derivatives, purine nucleosides, ceramides, benzoic acids and derivatives, as well as imidazoles. Benzoic acids and derivatives, as well as [N]{.smallcaps}-acyl-alpha-amino acids, differentiated kidney tissues whereas skeletal muscle profiles differed in [N]{.smallcaps}-acyl-alpha-amino acids, purine derivatives and alcohol and polyol levels.
[Table 1](#metabolites-10-00255-t001){ref-type="table"} also contains compounds that did not differentiate the targeted organs (*p* \> 0.05) but were introduced for further comparative analysis. [Figure 3](#metabolites-10-00255-f003){ref-type="fig"} presents the differences in the levels of metabolites that were common among all tissues.
The selected metabolites listed in [Table 1](#metabolites-10-00255-t001){ref-type="table"} were additionally subjected to the partial least squares discriminant analysis (PLS-DA). Results are presented in [Figure 4](#metabolites-10-00255-f004){ref-type="fig"} (scores plots), [Figure S4](#app1-metabolites-10-00255){ref-type="app"} (loading plots) and [Table S2](#app1-metabolites-10-00255){ref-type="app"} (validation metrics).
The PLS-DA method was used to more specifically model differences in the metabolome profiles of the targeted tissues of the compared strains of mice. Each model was validated (venetian blinds cross-validation) and refined using a permutation test (number of permutations = 100). This statistical analysis refined the number of features, yielding an optimal set of compounds from the selected metabolites (presented in [Table 1](#metabolites-10-00255-t001){ref-type="table"}) that successfully differentiated strains within each organ. For brain and liver tissues, the distinctions observed among BALB/c, C57BL/6 and CD1 mice were mainly similar to the results obtained in the PCA; however, for kidney and muscle, the separation between groups was much cleaner, as typically expected from supervised chemometric methods.
A metabolic pathway analysis was carried out to investigate the key biochemical pathways of the selected metabolites ([Table 1](#metabolites-10-00255-t001){ref-type="table"}). The results from the pathway analysis are shown graphically in [Figure 5](#metabolites-10-00255-f005){ref-type="fig"}, while the compounds involved in the respective metabolic pathways are presented in [Table S3](#app1-metabolites-10-00255){ref-type="app"}. The analysis uncovered a total of fourteen metabolic pathways mainly involved in the metabolism of amino acids (beta-alanine, valine, leucine, isoleucine, cysteine, methionine, tryptophan, arginine, proline, asparagine, glutamine), the biosynthesis of the aminoacyl-tRNA, terpenoid backbone, pantothenate and CoA, as well as the metabolism of glutathione, sphingolipids, pyrimidine, and purine.
3. Discussion {#sec3-metabolites-10-00255}
=============
Mammalian tissues, representing the sites of cellular metabolism, are a main target for metabolomics experiments, which can give us a 'snapshot' of a given tissue's temporary state when properly performed \[[@B25-metabolites-10-00255]\]. Considering that the extent to which genetics alters metabolism at the tissue level is still poorly understood, we conducted this study to find out whether two common mouse inbred strains (BALB/c, C57BL/6) and an outbred stock (CD1) are characterized by peculiar, strain-specific metabolite profiles, considering the selected organs (brain, liver, kidney and skeletal muscle) as optimal sites of extraction given their predominant use in research. The chosen lines of mice are general multipurpose models that have been used interchangeably for different scientific purposes. This particularly applies to the two inbred strains (BALB/c and C57BL/6) selected in this study. Inbred mice are defined as genetically identical within the strain and are used in studies in which isogenicity and homozygosity in the test population are desired. Conversely, in an outbred mouse stock (CD1), each animal is genetically unique and phenotypic variation in outbred stocks is usually greater than that of inbred strains due to both genetic and non-genetic factors \[[@B5-metabolites-10-00255]\].
Bearing in mind the fact that an individual's metabolome is sensitive to many internal and external variables, including age, gender, diet, environment, time of day, and even one's own genetics \[[@B12-metabolites-10-00255]\], the present study was performed on animals of the same age and sex that were kept in standard environmental conditions. Furthermore, we used intact mice (no experimental factors were applied) so we could compare the metabolomic profile of mouse organs under physiological conditions. The SPME technique was utilized as a sampling tool as it has been proven to be well suited for metabolomic analysis of tissue \[[@B19-metabolites-10-00255],[@B20-metabolites-10-00255],[@B21-metabolites-10-00255]\]. SPME additionally enabled extractions immediately after organ collection in our experiments, which is particularly desirable in rodent studies given that the metabolomic profile of tissues changes rapidly after cessation of blood circulation \[[@B25-metabolites-10-00255]\].
The PCA clearly revealed distinct differences in the entire metabolome of the brain and liver among all examined strains. The kidney and muscle metabolomic profiles of BALB/c mice distinctly varied from C57BL/6 and CD1, whereas these profiles in C57BL/6 and CD1 strains were, to some extent, similar. It should be stressed that the size of the clusters was larger for BALB/c in liver, kidney and muscle tissues, indicating greater variability in the metabolome, whereas in brain tissue, cluster size was similar for all the examined strains. In a previous study aimed at profiling metabolites in brain, heart, kidney, and liver tissues of 26 mammalian species representing ten taxonomical orders, it was suggested that brain metabolites are the most conserved among the examined organs, and have evolved largely according to the phylogeny. In contrast, the metabolites of other examined organs diverged to a much greater extent, possibly due to stronger environmental influences or other selection pressures \[[@B26-metabolites-10-00255]\]. Previous reports \[[@B27-metabolites-10-00255],[@B28-metabolites-10-00255]\] clearly indicate that genetic background has profound effects on the overall metabolite profiles of murine tissues, a factor that was confirmed in our experiment, especially for the brain and liver. It is also worth mentioning that significant phenotypic differences covering a number of physiological, biochemical, and neurobehavioral systems have been previously identified, even between very close mouse strains such as C57BL/6J and C57BL/6N \[[@B27-metabolites-10-00255]\].
In our work, amino acids and derivatives turned out to be the main group of compounds differentiating strains within individual organs. Considering brain tissue, the level of proline was significantly higher in C57BL/6 and CD1 mice compared to BALB/c. It must be stressed that a similar trend was also observed for all other examined organs. Proline is an endogenous amino acid in mammals that plays an essential role in primary metabolism and physiologic functions of living organisms. It can be endogenously synthesized either from glutamate or ornithine. Proline plays an important role in the synthesis and structure of proteins and in their metabolism (particularly the synthesis of arginine, polyamines, and glutamate via P5C), and can act as a direct substrate for ATP production \[[@B29-metabolites-10-00255],[@B30-metabolites-10-00255]\]. It has also been revealed that, under certain conditions, proline present in the brain can act as a neurotoxin that non-selectively destroys pyramidal and granule cells in rats \[[@B31-metabolites-10-00255]\]. The BALB/c strain had the lowest level of brain proline but was characterized by the highest levels of other amino acids such as asparagine, pyroglutamic acid, and [N]{.smallcaps}-acetylaspartic acid. These differences were significant compared to the levels of these metabolites in brain tissue of CD1 mice.
[N]{.smallcaps}-acetylaspartic acid (NAA), one of the most concentrated molecules in the central nervous system, is synthesized from aspartate and acetyl-coenzyme A in neurons. Its metabolic and neurochemical functions are still under investigation, but it is suggested that NAA is a direct precursor for the synthesis of the important dipeptide neurotransmitter [N]{.smallcaps}-acetylaspartylglutamate. It plays a role in neuronal osmoregulation and axon--glial signaling as well as in brain nitrogen balance \[[@B32-metabolites-10-00255]\]. Studies on mice and rats have shown that brain N-acetylaspartic acid concentrations remain relatively constant in different strains \[[@B33-metabolites-10-00255]\]. Considering mice, only two strains (DBA 2J/Sel and C57 BC/cdJ/Sel) had slightly higher levels of [N]{.smallcaps}-acetylaspartic than the other strains tested. We compared quite different strains, among which BALB/c mice were characterized by the highest content of this metabolite in the brain compared to C57BL/6 and CD1.
Valine, along with leucine and isoleucine, belongs to the branched chain amino acid (BCAA) group which cannot be synthesized de novo in mammals and must be supplied by a diet. In the central nervous system, BCAAs play a crucial role by providing nitrogen for the synthesis of the neurotransmitter glutamate \[[@B29-metabolites-10-00255]\]. In the present study, the level of valine in brain tissue did not differ significantly among the tested strains of mice. Previous research on (NIH) Swiss mice revealed that the levels of branched chain aliphatic amino acids (leucine, isoleucine, allo-isoleucine, and valine) detected in brain tissue were relatively low compared to other amino acids \[[@B34-metabolites-10-00255]\]. The same study showed that the most prevalent of all free brain amino acids was glutamic acid, followed by glutamine and aspartic acid. The present study did not show variability in the levels of the above mentioned brain amino acids, but we found differences in the levels of derivatives such as pyroglutamic acid (present in substantial amounts in the brain and other mammalian tissues \[[@B35-metabolites-10-00255]\]) and [N]{.smallcaps}-acetylaspartic acid. The highest levels of these amino acids were found in the brain tissue of BALB/c mice, whereas the lowest levels were present in CD1 mice, which suggests strain differences in the metabolism of glutathione and the neurotransmitter glutamate.
The results of our study suggest that purines (nucleotides, nucleosides and derivatives) are also among the main metabolites differentiating strains at the tissue level. They are crucial compounds for cell life. They are coenzymes, sources of energy, and direct precursors of DNA and RNA; moreover, they are involved in many other important biological processes \[[@B36-metabolites-10-00255]\]. A comparison of brain tissue in five different strains, including C57BL/6J and BALB/c, revealed that adenosine levels measured at basal conditions significantly varied with respect to mouse strain. Generally, levels of adenosine in the brain have been found to be low, with the exception of the BALB/c strain, which presented relatively higher adenosine levels \[[@B37-metabolites-10-00255]\]. In our study, the levels of adenosine monophosphate (AMP) (a direct precursor of adenosine in cells and a second messenger in many biological processes) detected in the brain tissue of BALB/c mice were higher compared to C57BL/6 and CD1 mice, which corresponds with the findings cited above. At the same time, levels of xanthine, an intermediate in the degradation of adenosine monophosphate to uric acid, were highest in the brain of CD1 mice in comparison to the other two mouse strains, which testifies to interstrain differences in purine metabolism.
Our research showed that in addition to amino acids and purines, compounds such as fatty acid esters, ceramides, benzoic acids, and imidazoles differentiated the liver tissue of the studied strains. Detected levels of valine, asparagine, pyroglutamic acid, AMP and 2-methylbutyrylcarnitine were highest in the liver tissue of BALB/c mice among the compared strains, thereby pointing to differences in relevant metabolic pathways (biosynthesis of panthotenate and CoA, aminoacyl-tRNA, biosynthesis and degradation of valine, leucine, and isoleucine, as well as metabolism of purines, glutathione, alanine, aspartate and glutamate) compared to C57BL/6 and CD1 strains. We did not see significant differences in the levels of all these metabolites between C57BL/6 and CD1 mice. Previous reports \[[@B26-metabolites-10-00255]\] have revealed that liver tissue is rich in a wide range of metabolites, including amino acids, glycerophospholipids, carbohydrates, and steroids likely indicative of liver-specific pathways. Moreover, it has been proven that the underlying metabolome of liver tissue in mice is highly sensitive to genetic differences, which is in agreement with the results of our study. A previous comparison of different mouse strains, among which C57BL/6J was present, showed that levels of metabolites involved in purine and pyrimidine metabolism, as well as pathways that play a role in amino acid metabolism in the liver, presented significant differences with respect to strain \[[@B28-metabolites-10-00255]\]. Although different mouse lines were compared in this study, amino acids, purine nucleotides, nucleosides and purine derivatives were among the main classes differentiating liver tissue. We also found that liver was the only tissue in which interstrain differences in allantoin levels were demonstrated. Allantoin and uric acid are the key compounds of purine nucleotide catabolism, which are formed in the liver as well as many other organs in rats \[[@B36-metabolites-10-00255]\].
The present study also revealed that strain type exerted a large effect on the selected metabolites of kidney tissue. We found interstrain differences in the levels of some amino acids, as well as in the levels of purines, fatty acids esters, and benzoic acids. The highest levels of valine and the lowest levels of proline, [N]{.smallcaps}-acetylasprtic acid, [N]{.smallcaps}-acetylvaline, AMP, and 2-methylbutyrylcarnitine were found in the kidney tissue of BALB/c mice, which generally presented a metabolome that differed considerably compared to C57BL/6 and CD1 mice. Ma et al. \[[@B26-metabolites-10-00255]\] found that most of the detected proteinogenic amino acids were present at moderate to high levels in kidney tissue relative to other organs, which can be interpreted as evidence that the metabolite profile of an organ reflects its biological functions.
Our results for skeletal muscle analyses revealed that strain affected the levels of some amino acids. The highest levels of asparagine and pyroglutamic acid and the lowest level of proline were typical of BALB/c, which according to our PCA analysis, separated best compared to C57BL/6 and CD1 mice. Interstrain differences were also found in the levels of pantothenic acid, a water-soluble vitamin; in mice, this compound plays a vital role in the growth of juveniles and in muscle maintenance of adults, as it regulates proper muscle mass among other functions \[[@B38-metabolites-10-00255]\]. A pathway analysis revealed that the metabolites differentiating the strains were involved in the metabolism of amino acids (arginine and proline) as well as in the biosynthesis of panthothenate, CoA, and aminoacyl-tRNA. Our results are in agreement with the previous research, which also demonstrated that strain affected the metabolomic profiles of skeletal muscle and pathways involved in energy metabolism (pantothenate and CoA biosynthesis and TCA cycle) in mice \[[@B28-metabolites-10-00255]\]. The same study proved that the examined tissues (muscle and liver) were largely unaffected by sex, suggesting that the tissue metabolome remains largely stable across sex.
4. Materials and Methods {#sec4-metabolites-10-00255}
========================
4.1. Chemicals {#sec4dot1-metabolites-10-00255}
--------------
External calibrant Pierce LTQ Velos ESI Positive Ion Calibration Solution was purchased from Thermo Scientific. All other chemicals were purchased from Sigma Aldrich (Poznan, Poland). Isopropanol, methanol, water, acetonitrile and formic acid were LC-MS grade. For SPME fiber preparation, N,N-dimethylformamide American Chemical Society grade (ACS) reagent and polyacrylonitrile were used.
4.2. Materials {#sec4dot2-metabolites-10-00255}
--------------
SPME fibers were manufactured in house, as described by Gomez-Rios et al. \[[@B39-metabolites-10-00255]\]. Probes with a 4 mm extractive phase coating were manufactured with the use of 5 μm hydrophilic-lipophilic balanced (HLB) particles provided by Waters (Wilmslow, U.K.).
4.3. Animal Handling and Tissue Collection {#sec4dot3-metabolites-10-00255}
------------------------------------------
BALB/c and C57BL/6 mice were purchased from the Experimental Medicine Centre of the Medical University in Bialystok, while CD1 mice were purchased from Jagiellonian University Medical College, Krakow. Experiments were performed on 15 adult males (aged 12 weeks) for a total of five mice per strain: BALB/c, C57BL/6 and CD1. The animals were housed in a controlled environment: temp. 22 ± 2 °C, 12 h light--dark cycle, humidity 55 ± 10%, with standard mouse chow and water available ad libitum. The mice were sacrificed by manual cervical dislocation, which resulted in euthanasia within approximately 10 s. Once euthanasia was confirmed, brain, liver, kidney and thigh muscle were immediately collected. According to European Union law, permission from the Local Ethical Commission is not required for the use of animal tissue or organs for scientific purposes.
4.4. Solid Phase Procedure and Sample Preparation {#sec4dot4-metabolites-10-00255}
-------------------------------------------------
SPME fibers coated with a biocompatible sorbent (4 mm length of HLB extraction phase) were used for the extraction of metabolites from the selected tissues. Before sampling, all fibers were conditioned overnight with methanol:water, 1:1, *v*/*v* solution. Prior to each extraction, fibers were rinsed for a few seconds in purified water to remove residues of organic solvents. A total of two fibers were inserted per organ for an extraction period of 15 min. Immediately after sampling, fibers were removed from the organ, quickly rinsed with water, then gently dried with wipes to remove any residue of the examined tissue and blood. Afterwards, fibers were placed into empty 0.3 mL polypropylene vials and stored in a freezer at −30 °C until analysis. Desorption was concurrently performed for all fibers directly before instrumental analysis. SPME fibers were placed into vials containing 200 µL of desorption solution consisting of acetonitrile:water (80:20, *v*/*v*) for 120 min with simultaneous vortex agitation (1500 rpm).
4.5. Liquid Chromatography--High Resolution Mass Spectrometry Analysis (LC--HRMS) {#sec4dot5-metabolites-10-00255}
---------------------------------------------------------------------------------
Samples were analyzed using a LC-HRMS procedure on an ultra-high performance liquid chromatograph coupled to a Q-Exactive Focus Orbitrap mass spectrometer. An instrumental method was adopted from Vukovic et al. \[[@B40-metabolites-10-00255]\]. Analytes were separated using a pentafluorophenyl column (Supelco Discovery HS F5, 2.1 mm × 100 mm, 3 μm). Phase A was water + 0.1% formic acid and phase B was acentonitrile + 0.1% formic acid. The gradient was as follows: 0--3 min 0% B, 3--25 min linear gradient to 90% B, 25--34 min 90% B, 34--40 min 0% B. Flow was set to 0.3 mL/min and injection volume was 10 μL. The column temperature was set to 25 °C and sample vials were held at 4 °C in the autosampler.
Mass spectrometer parameters in positive ionization mode were as follows: sheath gas flow rate: 40 a.u.; aux gas flow rate: 15 a.u.; spare gas flow rate: 0 a.u.; spray voltage 1.5 kV; capillary temp 300 °C; aux gas heater temp 300 °C, S-lens radio frequency (RF) level 55; S-lens voltage 25 V; skimmer voltage 15 V. Scan range was set on *m/z* 80--1000 with resolution 70,000. Acquisition was performed using automatic gain control (AGC) target 1E6 and inject time to C-trap was set on auto. The instrument was calibrated using external calibration every 72 h, resulting in mass accuracy \<2 ppm. Within-sequence samples were randomized. Pooled quality control (QC) samples composed of 10 µL of each sample were run every 12 injections to monitor instrument performance.
The structure of selected compounds was confirmed based on suitable LC retention time and \<3 ppm mass accuracy. Full MS/dd-MS2 confirmation mode was used for this purpose. Fragmentation parameters were as follows. Mass resolution: 35,000 full width at half maximum (FWHM), AGC target: 2E4, minimum AGC: 8E3, intensity threshold: auto, maximum IT: auto, isolation window: 3.0 *m/z*, stepped collision energy: 10 V, 20 V, 40 V, loop count: 2, dynamic exclusion: auto.
4.6. Data Processing and Statistical Analysis {#sec4dot6-metabolites-10-00255}
---------------------------------------------
Raw MS data was processed by Compound Discoverer 2.1 (Thermo Fisher Scientific) to putatively identify metabolites. QC-based area correction was set to min 50% QC coverage and max QC area relative standard deviation (RSD) 30%. Only features with min peak intensity 105 and signal to noise ratio \> 3 were taken into consideration. A data filtration step removed 8795 (93%) of the 9468 features. Putative identification of features was performed by searching for the exact molecular weights of identified features (3 ppm accuracy) in the Human Metabolome Database (HMDB) and Kyoto Encyclopedia of Genes and Genomes (KEGG) online databases. Annotations were verified according to the isotopic distribution of molecules using the ChemSpider Database as a reference ([Table S4](#app1-metabolites-10-00255){ref-type="app"}). The MetaboAnalyst 4.0 software and the KEGG pathway library corresponding to the *Mus musculus* metabolome were used to obtain metabolic pathways associated with significantly differential metabolites. Identification of spectra of fragmented compounds was done with the use of Thermo Scientific FreeStyle 1.4 software linked to online mzCloud database ([Table S4](#app1-metabolites-10-00255){ref-type="app"}).
The averaged peak areas (from two independent SPME fiber replicates) for the obtained compounds were analyzed using Statistica 13.3 PL software (StatSoft, Inc., Tulsa, Oklahoma, USA) via one-way analysis of variance (ANOVA). A post-hoc Tukey Honestly Significant Difference (HSD) test was used to determine the significance of differences among groups, where a *p*-value of \< 0.05 was considered to be significant. Resulting data were then exported to the (PLS)-Toolbox (Eigenvector Research Inc.) in Matlab^®^ version 2018b (MathWorks Inc., Natick, MA, USA) for multivariate statistical analysis. Data was log-transformed and mean-centered prior to principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). The PLS-DA model was cross-validated using the venetian blinds method and refined by random permutation (100 times) of the Y variable. Two-dimensional score plots were generated to visually assess separation between sample groups.
5. Conclusions {#sec5-metabolites-10-00255}
==============
The results of the present study clearly indicate that the baseline metabolomic profiles of organs, especially brain and liver tissue, as well as different metabolic pathways, vary widely among the laboratory mouse strains commonly used in biomedical research. Interstrain differences in the metabolome at the tissue level testify that even general-purpose models can give different answers to the same experimental factors, and as a result, yield contradictory outcomes in a study if such factors are not accounted for. For this reason, close attention should be paid when choosing a mouse strain for a particular purpose of scientific research, and considerations of strain variability should also be included in the interpretation of results. Further, the present study corroborates that SPME is an easy, quick and reliable sample preparation method that is ideally suited for tissue analysis in metabolomic studies. However, one needs to remember that the simplicity of the method compromises coverage of analytes compared to multi-step and time-consuming liquid--liquid extraction, and for in-depth investigation of all metabolites, a traditional approach should be considered. Also, it must be pointed out that our results are based mainly on the putative identification of metabolites, so further LC--MS/MS analysis with a high level of confidence is still required to confirm the identities of all metabolites.
The authors would like to thank Thermo Fisher Scientific for access to the Q-Exactive Focus orbitrap mass spectrometer.
The following are available online at <https://www.mdpi.com/2218-1989/10/6/255/s1>, Figure S1: Ion map (molecular weight versus retention time) of all extracted metabolites following LC-MS analysis in positive ionization mode, Figure S2: Principal component analysis (PCA) plot of all analyzed samples and extraction quality control (QC) samples, Figure S3: 3D PCA score plots of untargeted metabolomics data of tissues collected from three different mouse strains. Examined strains included: BALB/c (red), C57BL/6 (green), and CD1 (blue). Tissues analyzed included: brain (a), liver (b), kidney (c), and muscle (d), Figure S4: Loadings plots of PLS-DA. Tissues analyzed were: brain (a), liver (b), kidney (c), muscle (d), Table S1: Details of detected features, Table S2: Validation metrics of PLS-DA models, Table S3: Metabolic pathways and involved metabolites differentiating tissues of C57BL/6, BALB/c and CD1 mice strains, Table S4: Isotope pattern matching analysis of differential metabolites using the ChemSpider database.
######
Click here for additional data file.
Conceptualization, K.B., B.B.; SPME fibers manufacturing, K.B., I.S., J.B.; conducting the experiments, K.B., I.S., J.B., K.J.; instrumental analysis, J.B., K.B.; data processing, I.S., J.B.; data analysis, K.B., I.S., J.B.; statistical analysis, B.K., K.B.; writing, manuscript preparation, K.B.; review, funding acquisition, supervision, B.B.; All authors have read and agreed to the published version of the manuscript.
This research was funded by Nicolaus Copernicus University in Toruń, grant number 451.
The authors declare no conflict of interest.
{ref-type="app"}.](metabolites-10-00255-g001){#metabolites-10-00255-f001}
{#metabolites-10-00255-f002}
{#metabolites-10-00255-f003}
{#metabolites-10-00255-f004}
{#metabolites-10-00255-f005}
metabolites-10-00255-t001_Table 1
######
Chemical taxonomy of selected annotated features present in examined tissues of BALB/c, C67BL/6 and CD1 mice.
Organ Class Compound Molecular Weight Retention Time (min) *p*-Value (ANOVA)
---------------------------------------- ----------------------------------- ---------- ------------------ ---------------------- -------------------
Brain Alpha-amino acids and derivatives Proline 115.0636 1.80 0.0019
Valine 117.0792 1.28 0.6876
Asparagine 132.0536 1.19 0.0000
Pyroglutamic acid 129.0427 1.22 0.0104
Cystine 240.0237 1.17 0.0441
[N]{.smallcaps}-acetylaspartic acid 175.0481 2.12 0.0005
Tetrahydrodipicolinate 171.0532 1.36 0.0145
[N]{.smallcaps}-acyl-alpha-amino acids [N]{.smallcaps}-acetylvaline 159.0896 7.69 0.3731
Purine derivatives Xanthine 152.0334 3.23 0.0065
Purine nucleotides Adenosine monophosphate (AMP) 347.0629 1.34 0.0063
Purine nucleosides Inosine 268.0807 6.68 0.0070
8-hydroxydeoxyguanosine 283.0916 6.93 0.0005
Pyrimidine derivatives Uracil 112.0276 1.34 0.0091
Pyrimidine nucleosides 2'-deoxycytidine 227.0905 7.05 0.0002
Fatty acid esters 2-methylbutyrylcarnitine 245.1627 17.87 0.0277
Hydroxy fatty acids Mevalonic acid 148.0737 1.33 0.0483
Fatty amides Oleamide 281.2718 20.64 0.0079
Alcohols and polyols Pantothenic acid 219.1107 7.08 0.0273
Liver Alpha-amino acids and derivatives Proline 115.0636 1.80 0.0005
Valine 117.0792 1.28 0.0466
Asparagine 132.0536 1.19 0.0190
Pyroglutamic acid 129.0427 1.22 0.0083
Tetrahydrodipicolinate 171.0532 1.36 0.0185
[N]{.smallcaps}-acetylaspartic acid 175.0481 2.12 0.5917
[N-]{.smallcaps}acyl-alpha-amino acids [N]{.smallcaps}-acetylvaline 159.0896 7.69 0.3605
Purine derivatives Xanthine 152.0334 3.23 0.6444
5-hydroxyisourate 184.0233 1.34 0.0004
Purine nucleotides Adenosine monophosphate (AMP) 347.0629 1.34 0.0490
Purine nucleosides Inosine 268.0807 6.68 0.0120
8-hydroxydeoxyguanosine 283.0916 6.93 0.0033
Fatty acid esters 2-methylbutyrylcarnitine 245.1627 17.87 0.0235
Ethyl eicosapentaenoic acid 330.2557 21.65 0.0148
Ceramides Ceramide (d40:1) 621.6063 26.30 0.0175
Benzoic acids and derivatives 2-aminobenzoic acid 137.0476 1.33 0.0048
Imidazoles Allantoin 158.0439 1.15 0.0442
Kidney Alpha-amino acids and derivatives Proline 115.0636 1.80 0.0069
Valine 117.0792 1.28 0.0446
Asparagine 132.0536 1.19 0.7313
Pyroglutamic acid 129.0427 1.22 0.4256
[N]{.smallcaps}-acetylaspartic acid 175.0481 2.12 0.0080
[N]{.smallcaps}-acyl-alpha-amino acids [N]{.smallcaps}-acetylvaline 159.0896 7.69 0.0000
Purine derivatives Xanthine 152.0334 3.23 0.1554
Purine nucleotides Adenosine monophosphate (AMP) 347.0629 1.34 0.0176
Fatty acid esters 2-methylbutyrylcarnitine 245.1627 17.87 0.0003
Ethyl eicosapentaenoic acid 330.2557 21.65 0.0066
Benzoic acids and derivatives 2-aminobenzoic acid 137.0476 1.33 0.0067
Muscle Alpha-amino acids and derivatives Proline 115.0636 1.80 0.0015
Valine 117.0792 1.28 0.3540
Asparagine 132.0536 1.19 0.0236
Pyroglutamic acid 129.0427 1.22 0.0206
[N]{.smallcaps}-acetylaspartic acid 175.0481 2.12 0.3746
[N]{.smallcaps}-acyl-alpha-amino acids [N]{.smallcaps}-acetylvaline 159.0896 7.69 0.6755
[N]{.smallcaps}-tridecanoylglycine 271.2147 22.52 0.0407
Purine derivatives Xanthine 152.0334 3.23 0.0635
5-hydroxyisourate 184.0233 1.34 0.0038
Purine nucleotides Adenosine monophosphate (AMP) 347.0629 1.34 0.1549
Fatty acid esters 2-methylbutyrylcarnitine 245.1627 17.87 0.0051
Alcohols and polyols Pantothenic acid 219.1107 7.08 0.0305
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Attachment theory is an evolutionary-based theory of a specific type of intimate human social relationship conceived to have a major developmental influence from "the cradle to the grave" (Bowlby, [@B7], [@B8]). According to attachment theory, the foundation of the attachment relationship is a biologically based behavioral system that evolved in ways that influence and organize motivational, cognitive, emotional and memory processes. These processes are organized in early infancy with respect to significant caregiving figures that extend into adulthood. Bowlby ([@B9]) conceived attachment as a key mechanism related to maintaining biological homeostasis, including the modulation of physiological stress and mental health. Researchers have found physiological correlates of attachment and the affective components of relationships in nonhuman species and humans. Mental representations of early attachment relationships shape emotional and cognitive information, which affects our attention and memory. In order to maintain organization within the attachment system, emotional reactivity is then regulated within the central nervous system (Bretherton, [@B12]; Main, [@B47]). Over the decades, psychobiological attachment research with infants and adults has increased dramatically (Coan, [@B21]; Gander and Buchheim, [@B28]). Attachment patterns have been linked to different ways to emotion regulation processes and some researchers even argued that the attachment system is in itself an emotion regulation device (Vrtička and Vuilleumier, [@B62]).
Most recent findings on attachment and neurobiology in functional magnetic neuroimaging (fMRI) showed that researchers investigated very different systems, often by very different means and a variety of paradigms, ranging from the presentation of individual photos of loved and unknown faces to more complex approaches (e.g., reflecting on attachment-relevant events, priming experiments, talking about attachment-relevant situations; see overview Buchheim and George, [@B13]). At present, the delineation of a neuronal network of attachment is not possible yet. However, there is evidence across fMRI studies that brain regions like the amygdala and orbito/prefrontal cortices are involved in processing attachment-related stimuli. In addition, convergent research results suggest that when caregiving is addressed, dopamine-associated regions of the reward system are active that differ from the neural correlates of the postulated "attachment circuitry" (Buchheim et al., [@B18]).
Several neurophysiological studies of adult attachment assessing the autonomic nervous system, the hypothalamic-pituitary-adrenocortical axis or frontal electroencephalography (EEG) asymmetry used self-report measures (Carpenter and Kirkpatrick, [@B20]; Kim, [@B40]; Laurent and Powers, [@B44]; Rochman et al., [@B54]; Zhang et al., [@B67]; Kiss et al., [@B41]; Dan and Raz, [@B22]), while other studies used narrative interview measures of attachment such as the Adult Attachment Interview (AAI) and the Adult Attachment Projective Picture System (AAP; Beijersbergen et al., [@B5]; Buchheim et al., [@B17]; Fraedrich et al., [@B26]; Holland and Roisman, [@B39]; Behrens et al., [@B4]; Leyh et al., [@B45]).
The self-report questionnaire instruments are conceived as personality constructs and assess the subjective evaluation of attachment styles with reported patterns monitored by conscious processing of feelings and experiences related to desires and worries regarding a romantic partner; these measures typically differentiate secure from insecure avoidant or anxious attachment styles (Ravitz et al., [@B53]). By contrast, developmental attachment measures such as the AAI (George et al., [@B34]; unpublished manuscript) or AAP (George et al., [@B31]) are designed to activate the individuals attachment system by introducing attachment-related topics (e.g., separation, illness, abuse and death), and assess attachment representations (secure, insecure-dismissing, insecure-preoccupied and unresolved trauma) based on the analysis of discourse patterns of verbatim transcripts. Interview discourse analysis is less concerned with a specified response (as compared with attachment style measures) as how experiences and feelings are described.
In a very recent fMRI study, Yaseen et al. ([@B65]) investigated the comparison of brain activity correlating with self-report (Relationship Scales Questionnaire, RSQ) vs. a narrative attachment measure (AAI) during conscious appraisal of an attachment figure. Interestingly the two measures elicited different brain responses. While the AAI appeared to disproportionately correlate with conscious appraisal associated activity in Default Mode Network (DMN) and *subcortical* structures, the RSQ seemed to tap Executive Frontal Network (EFN) structures more extensively. The authors suggested, that the AAI captured more interoceptive, "core-self"-related processes, while the RSQ assessed higher-order cognitions involved in attachment. The authors recommended in their conclusions, that the AAP might be an appropriate alternative in this kind of research, since this measure consists of a set of pictures feasible to present during an experimental setting.
The feasibility of the AAP measure as an attachment-activating stimulus in a neurobiological context (fMRI, neuroendocrinology) has been established in diverse experimental settings in clinical and nonclinical groups (e.g., Buchheim et al., [@B16], [@B15], [@B17], [@B19]). Participants in these different settings were instructed to tell stories to the AAP picture stimuli in the fMRI environment (Buchheim et al., [@B16], [@B15]) or were presented individualized sentences constructed from their own AAP responses in the fMRI setting (Buchheim et al., [@B19]).
In the context of a double-blind study with a neuroendocrine research question, we modified the AAP task for a double blind controlled study comparing the effects of oxytocin to a placebo condition. The AAP picture presentations were accompanied by prototypical phrases constructed to represent one of the four established attachment categories (i.e., a generalized attachment-sentence schema for each attachment group). The participants were instructed to rank these phrases from the most to the least appropriate for each presentation. The most interesting finding from this study was that insecurely attached individuals at baseline decided that secure attachment sentences were most appropriate for them under the oxytocin condition (Buchheim et al., [@B17]). This attachment experiment was a first attempt to assess a combination of conscious and unconscious processes in a self-report perception task. In this present study, we sought to improve on this approach by using this methodology in a reaction time (RT) paradigm. The research question was to examine if the RTs differed with respect to an individual's attachment representation in order to provide a paradigm to use in a neurobiological setting, like an EEG experiment.
One interesting development in the past few years has been experimental research using the Implicit Association Task (IAT), the goal of which was to explore the domain of automatic cognitive processes concerning the evaluation of the self and others (Lane et al., [@B43]). The IAT task is based on the measurement of RTs and answers in combination with a target category, for example gender stereotyping and the self. Implicit measures have been successful in predicting verbal behavior, group membership, sexual behavior, and evaluative judgments (Gawronski, [@B29]) or personality (Grumm and von Collani, [@B35]).
RT research has a long experimental tradition in psychology, beginning with the experiments by Helmholtz ([@B100]). Helmholtz was interested in the time relations structured by the nervous systems of living beings not just from a physiological but also from a psychological point of view. In fact, at the time at which he performed his time experiments in frogs, Helmholtz carried out similar studies in human beings (Schmidgen, [@B57]).
RT experiments are relatively inexpensive to execute and results are easy to obtain, even though the conclusive interpretations are still under discussion. According to Harris et al. ([@B36]) RT experiments have become a standard paradigm for measuring behavioral reactions without taking into account underlying mental processes. Harris et al. ([@B36]) suggested a sophisticated way to improve the analysis and interpretations of RT paradigms.
The idea behind the measurement of RT is that it can be used as a measure in social cognition research, as an index of the complexity of the underlying mental processes. Results showed for example, that more complex processes are associated with longer elaboration/RTs (Rösler, [@B56]). Moreover, RT experiments have a predictive value for social decisions and have been used successfully in IAT clinical and social experiments (Lane et al., [@B43]). It was possible to differentiate groups with and without disorders using the IAT in reliable experiments about self-judgments (Gemar et al., [@B30]). RT measures were also used to understand semantic priming. Reactions were more quickly facilitated when categories were closely related and shared the same reaction as compared to categories that did not share the same reaction. Attribution measures can be interpreted as a measure of relative identity with the objects (Lane et al., [@B43]). The association of the self as a target category and an attitude dimension provides a measure of implicit self-esteem; it describes the strength between associations of the self and another category. Studies show that emotional relevant primes have an effect on memory performance. One study showed that memory performance was impaired in borderline patients when negatively valued interference was presented (Mensebach et al., [@B51]). A recent study on autobiographical memories reported that past intentions could be reliably identified with high accuracy using a RT measure (Zangrossi et al., [@B66]).
A central concept of attachment theory is that individuals develop internal working models, that include expectations about the self, and significant others outside of conscious awareness (Bowlby, [@B7], [@B8], [@B9]; Bretherton, [@B10]). Furthermore, internal working model content is believed to include knowledge about concrete details of interpersonal experiences as well as the associated affect (Bretherton, [@B10]). In general, psychoanalytic theory suggests to divide the mind into three different levels: the conscious mind includes everything we are aware of and represents our mental processing that we think of and talk about rationally. A part of this includes memory structures, which are considered not always to be part of consciousness, but can be retrieved and brought into awareness, called preconscious. The unconscious mind constitutes a reservoir of feelings, thoughts, urges and memories that exist outside of conscious awareness. From a psychoanalytic point of view, most of these contents are unacceptable or unpleasant and represent feelings of pain, anxiety or individual conflicts. Unconscious processes are considered to influence our behavior and experience, even though we are unaware of these underlying influences (Freud, [@B27]). As mentioned above internal working models are also thought to work primarily outside of conscious awareness (unconsciously) and guide attention, interpretation, and memory of attachment experiences and emotions. This allows individuals to generate expectations about the future concerning interpersonal interactions and to develop plans relating to them (Bowlby, [@B7], [@B8], [@B9]; Bretherton, [@B10], [@B11]). Bowlby ([@B9]) examined possible memory constructs and unconscious processes to explain misrepresentations of mental functioning and behavior, informed by mid-20th century advances in cognitive psychology (Bowlby, [@B9]). He suggested conscious representations of what parents made the child believe are stored in the semantic memory, while the defensively excluded and traumatic attachment experiences are stored in the episodic memory. Emotional schemata are part of episodic memories and, over time, these schemata can grow into explicit models of the self and the attachment figure (Liotti, [@B46]). According to information processing theory, the term "unconscious" describes the product of the perceptual systems that work unattended or unrehearsed. Thus from this perspective, nonconscious mental life is identified with early preattentive perceptual processes such as e.g., pattern or face recognition. One of the most common forms of preconscious processing is priming. When investigating the label "automatic", some processes are intended, others require recent conscious and intentional processing of related information (Bargh et al., [@B2]). In the following, we use the term "unconscious" in association with the internal working models of attachment and "preconscious" when relating to information processing theory or neurobiological models.
There are several recent studies investigating implicit aspects of romantic attachment using self-report measures as explicit instruments for assessing attachment style (Marks and Vicary, [@B48]; De Carli et al., [@B23]). In the present study, we were interested to capture automatic processes in the moment the attachment system is activated by using a narrative attachment measure with an implicit nature. The AAP is designed to activate the individual's attachment system and emphasizes the evaluation of unconscious defensive processes in the narratives. In this study, we intended to combine attachment methodology with knowledge from implicit measures by using a RT paradigm.
The general question for this study addressed how a person accepts or rejects prototypic sentences belonging to the two major attachment categories (secure and insecure) using a modified version of the AAP (Buchheim et al., [@B17]) in a RT paradigm. All participants were administered the standard AAP interview before the experiments started in order to assess their individual attachment representation. The participants did not get any information about their attachment representation during the whole assessments. The experimental design is described in the "Materials and Methods" Section in detail. In short, participants were presented pictures from the AAP accompanied with sentences representing different attachment patterns while assessing how long it took for them to make a decision (i.e., accept or not accept).
1. We expected that participants would accept the prototypic sentences from the experiment more frequently when these sentences match with their own adult attachment classification.
2. We expected group differences in reaction speed between participants with secure or insecure adult attachment classification assessed in the previous AAP interview. These expectations were differentiated for four possible configurations of the stimulus (secure, insecure) and the reaction (acceptance, rejection). Comparing both groups, we expected that (2a)secure sentences will be accepted faster by securely attached participants,(2b)secure sentences will be rejected faster by insecurely attached participants,(2c)insecure sentences will be accepted faster by insecurely attached participants,(2d)insecure sentences will be rejected faster by securely attached participants.
3. The preference of secure or insecure prototypes in the experimental procedure was expressed by the continuous Adult Attachment Projective Relationship Choices Version 2 (AAP-RC) security scale (see below). We expected following correlations of the security scale with the reaction speed: the higher the security scale, ... (3a)... the faster the acceptance of secure sentences,(3b)... the slower the rejection of secure sentences,(3c)... the slower the acceptance of insecure sentences,(3d)... the faster the rejection of insecure sentences.
Materials and Methods {#s2}
=====================
Adult Attachment Projective Picture System (AAP) {#s2-1}
------------------------------------------------
The AAP (George and West, [@B33]) assesses the attachment status in adults using a standardized set of eight picture stimuli. The stimuli are line drawings that include a warm-up scene and seven attachment scenes of individuals in conceptually-defined attachment situations. Four so called "alone pictures" depict scenes of a single person with no other persons visible in the picture. Three so called "dyadic pictures" depict scenes of two or more persons in a potential attachment dyad. The scenes portray characters in different age groups across the life span (e.g., child to old age). The drawings contain only as much details necessary to connote the situation. Features indicating details such as emotion, ethnicity and gender are obscure. Stimulus presentation is standardized so as to introduce increasingly distressing attachment scenes. Participants are asked to tell a story to each picture using a standardized set of instructions: "What is going on in the picture, what led up to this scene, what are the characters thinking or feeling, and what might happen next." AAP administration is done in person on an individual basis in a quiet location with no distraction and typically takes 30 min. The stories are audio-recorded and analyses are done from verbatim transcripts.
Each stimulus response is coded for attachment related content and defensive processes. Content coding evaluates representation of the presence and degree of integration (as defined by attachment research) of attachment relationships in the response, the actual coding dimensions of which are evaluated on different dimensions for the alone and dyadic pictures. Alone response content is evaluated on two dimensions. The agency of self is defined as the degree to which the character can seek and effectively use attachment figures. The connectedness is defined as the degree to which the character is portrayed as seeking proximity to others. Dyadic response content is evaluated for synchrony, a single dimension that captures the quality of agency of self and connectedness used for the alone pictures. Synchrony is defined as the degree to which responses depict attachment figure sensitivity in the context of distress themes (e.g., a child is sick) or mutual enjoyment in the context of togetherness themes (e.g., couple goes on a trip). Defensive processes are coded for the three standard attachment-defined defenses (Solomon et al., [@B61]): deactivation (distanced attention from attachment), cognitive disconnection (close attention to and confusion by attachment) and segregated systems (attachment fear and dysregulation).
The AAP designates four attachment classifications based on the evaluation of response content and defensive processes coding patterns across the entire set of seven attachment stories. Individuals are judged secure (F) when the coding patterns demonstrate that attachment figures are present and self and attachment figures manifest integrated interaction (sensitivity, relationship repair, thoughtful action and mutual enjoyment). Defensive processes, which can be depicted in any of the three defense, help integrate and maintain relationships, and manage attachment fears. Individuals with insecure-dismissing (Ds) or insecure-preoccupied (E) attachment are characterized by relative absence of integration, sensitivity and mutual enjoyment in their responses. Descriptions of the alone characters range from themes that portray taking simple action (reactive problem solving behavior without thoughtful consideration) to evidence, that characters cannot move forward. Attachment figures, if included, are described in ways ranging from functional roles without comfort (e.g., just give the sick boy soup), unable to respond (e.g., the mother refuses to hug the child), punishing and sometimes harsh (e.g., an enraged parent who is drunk and abusive). Connections with others, if described, are typically shift to interactions with non-attachment figures (e.g., police, nurse, soccer coach). The dismissing group is characterized by defensive processes, that deactivate attachment needs and shift attention away from attachment distress and themes (e.g., by rejection, power, achievement). The preoccupied group is characterized by defensive processes, that disconnect attachment needs and relationships (i.e., deconstruct the details) shifting attention to elements of frustration and anger, or distorting or blurring story characters and events (e.g., the child is waking up or going to bed; someone died---cannot specify who). Individuals are judged as insecure-unresolved with regard to trauma (U) when they remain dysregulated and overwhelmed by dangerous or threatening story elements (e.g., being frightened, assault, isolation, helplessness). One or more of their stories are void of the content and defensive processing features associated with integration, functional care, or attachment figure or other people providing care. For more complete details of the coding system and classification, see the monograph (George and West, [@B33]).
Multiple studies have demonstrated excellent convergent validity of the AAP with the AAI (George et al., [@B34]), test-retest reliability, inter-judge reliability, and discriminant validity in community samples and clinical patients. Results from a large-scale psychometric investigation, including 144 adult participants demonstrated excellent inter-judge reliability; the concordance rate for two judges on the four-group classifications were 90%, *κ* = 0.85, test-retest reliability (after 3 months 84% remained in the same attachment category; *κ* = 0.78) and discriminant validity. To evaluate the convergent validity, AAP classifications were compared to independent AAI classifications. The concordance rates for the four-group classifications were 90%, *κ* = 0.84, and for the two groups ("secure" vs. "insecure") even 97%, *κ* = 0.89 (George and West, [@B32], [@B33]; Buchheim and George, [@B14]).
Development of the AAP Reaction Time Paradigm {#s2-2}
---------------------------------------------
Buchheim et al. ([@B17]) developed and used the first experimental adaption of the AAP in a double-blind, placebo-controlled within-subject experimental design. These researchers developed the AAP-RC stimulus set, which is comprised of a set of statements that represent attachment-related sentences that describe the AAP picture stimuli. The statements were schematic descriptions of secure, dismissing, preoccupied, and unresolved attachment stories, as determined by two expert AAP judges (AB, CG) who collectively had experience with over 300 AAP transcripts. Attachment statements described common story response situations. The study compared participant responses to the statement in an oxytocin and a placebo condition. The eight AAP picture stimuli were presented over four sequences, always presented in each sequence in the standardized order. Each of the 32 picture presentations was accompanied by four prototype phrases each of them representing one of the four established attachment categories. The participants were instructed to rank these phrases from the most to the least appropriate for each presentation. The phrases were presented in a randomized balanced sequence in order to minimize simple memory effects across test sessions.
The present study used a modified version of the Buchheim et al. ([@B17]) prototype sentences. Sentences were revised to improve the content and to control for the sentence length. In each group of length-adjusted sentences, all four sentences consisted of the same number of German words in order to minimize the effect of the sentence length on RTs. The modified system of 128 sentences is called Adult Attachment Projective Relationship Choices Version 2 (AAP-RC 2.0). The AAP-RC evaluation procedure uses all eight AAP drawings, including the first dyadic "warm-up" picture with two playing children. Hence, 64 sentences relate to the alone pictures and 64 sentences to the dyadic ones. The revised sentences were rated for content by three certified AAP judges. Table [1](#T1){ref-type="table"} shows example sets of four sentences that represent four attachment categories for two selected AAP picture stimuli. Figure [1](#F1){ref-type="fig"} demonstrates an example how a stimulus sentence was presented on the PC screen to the participant. The experimental procedure consisted of 128 such screen sequences.
######
**Examples of prototypical sentences from the Adult Attachment Projective Relationship Choices Version 2 (AAP-RC 2.0) instrument**.
Prototypical sentences from the AAP-RC Attachment classification
--------------------------------------------------------------------------------------- ---------------------------
**AAP picture "Bed"**
A child is put to bed by his mother and she sings a nice comforting lullaby for him. **F**---secure
A child is put to bed by his mother and she gives him a toy and walks out. **Ds**---dismissing
A child is put to bed by his mother and she is angry because he was too naughty. **E**---preoccupied
A child is put to bed by his mother and she is helpless due to the child's nightmare. **U**---unresolved trauma
**AAP picture "Departure"**
A couple bids farewell and is looking forward to being together soon again. **F**---secure
A couple bids farewell and he is ruminating about his upcoming business meeting. **Ds**---dismissing
A couple bids farewell and she is very angry about his surprising departure. **E**---preoccupied
A couple bids farewell and she threatens to hurt herself if he leaves her. **U**---unresolved trauma
{#F1}
The RT paradigm used in this study followed Bem's ([@B6]) procedure for RT analysis for gender role schemas. The procedure was to record answers and to assess the time the participants needed to react. The Bem's ([@B6]) study showed that schema consistent judgments were more quick when the stimuli presented during a selection task matched participants' gender role schema. We chose this approach because gender role schemas, like attachment representations, are conceived as stable views of self that develop in early childhood that automatically monitor, modulate attentional shifts and appraise new experiences (Bretherton, [@B11]). Classical experimental designs of self-concept tests using randomized stimulus sequences like the IAT (Gawronski, [@B29]) could not be used because attachment assessments such as the AAP must adhere to the procedural order in which stimuli are presented (George and West, [@B33]).
In the current study in the context of a diploma thesis (Wichmann, [@B64], unpublished diploma thesis), we first administered the AAP and next presented the AAP-RC 2.0. We conducted an Attachment Reaction Time analysis (ART) for the experimental condition. The AAP was administered by a trained interviewer (TW). AAP verbatim protocols were coded by a certified AAP judge (AB). The structure of the administration procedure was as follows: the entire series of AAP picture stimuli were presented 16 times and each series was composed of eight pictures in the standardized AAP administration order. Each picture presentation was accompanied by one stimulus sentence, which was related to one of the four attachment representations. The experimental procedure contained a measure for the individual responses (yes/no) to the prototypical sentences and the recorded RTs.
The interview and the experimental task took place in the same office. The experimental condition was conducted using a computer. The computer was a table mounted Dell computer with no internet-connection and no additionally installed software. The program used for the presentation and RT measurement was *E-Prime* (Schneider et al., [@B58]). Participants were alone in the room. Room lighting was artificial and participants sat 0.5 m from the monitor. Answer responses were given via the computer keyboard. At the beginning of the experiment the participants were told to put the index fingers of their hands on the keys: Y for "yes" and the key N for "no". The keys were marked with a red label. The participants had to press a key to move on with the task. A short practice task was given before to ensure that the participants had understood the task. The practice task included three attachment neutral stimuli with drawings produced in a style similar to the AAP pictures. All instructions were given on the computer screen and, if necessary, explained a second time after the test run.
Participants were told that the task was a speed task so as to avoid participant reflection and distraction. It was emphasized that there were no right or wrong answers, and that responses were simply their preferences. The timing of the presented sentence order (Figure [1](#F1){ref-type="fig"}) was as follows: 1st a fixation cross (1 s); 2nd the sentence (2 s); 3rd a countdown (1.5 s) and 4th the picture along with the decision task. AAP RC sentences were shown one at a time, next showing an AAP picture, with a "3-2-1" countdown shown between the sentence and the picture. The labels "Yes" and "No" were presented on the side of the monitor, analogs to the keys, during the decision tasks. The picture was displayed on the screen until a decision was made. The experiment continued only after a decision was made. We presented the participants first the sentence and then the picture so as to eliminate bias produced by different reading speed. The four attachment categories were presented each with four prototype sentences per picture. The order of attachment representations within the sentence was randomized. In sum there were 128 choices to be made. The choices were forced choice subjective selections, representing their acceptance (yes) or rejection (no) of an AAP RC sentence (see Table [2](#T2){ref-type="table"}). Task scores are based on counts of agreements and rejections by four attachment representations.
######
**Schema of evaluation of the AAP-RC 2.0 by a test person**.
Test person F secure Ds dismissing E preoccupied U unresolved trauma Total
------------- ---------- --------------- --------------- --------------------- ---------
**Yes** a b c d r
**No** 32 − a 32 − b 32 − c 32 − d 128 − r
**Total** 32 32 32 32 128
Participants {#s2-3}
------------
Participants were asked for voluntary participation. The sample was comprised of 30 students from the University of Innsbruck (17 women, 13 men; sample mean age: 26.8, SD = 3.4). The participants reported no neurological conditions and were not in psychological or psychiatric treatment. All had normal or corrected eye vision. The study was conducted according to the Helsinki Declaration with informed consent received from all participants. All participants completed the study.
Results {#s3}
=======
The reported results concern three methodical approaches: the AAP attachment classification assessment; the computerized experimental method AAP-RC; and the ART experiment. The results first describe the findings associated with each of the measures used in the study and second report the relations among them. There were no missing data.
Adult Attachment Projective Picture System (AAP): Distribution of Attachment Classifications {#s3-1}
--------------------------------------------------------------------------------------------
The attachment classification distribution was as follows: 10 (33%) F, 12 (40%) Ds, 6 (20%) E, and 2 (7%) U. Because of the small frequencies in especially the preoccupied and unresolved groups, insecure classifications were collapsed together and data analyses compared only secure (*n* = 10, 33%) vs. insecure attachment (*n* = 20, 67%).
Relationship Choices, Version 2.0 (AAP-RC): Psychometrical Analysis and Security Index {#s3-2}
--------------------------------------------------------------------------------------
Reactions to AAP-RC stimuli in the ART test were coded dichotomously as yes (endorsement, acceptance) or no (rejection). The frequencies *a, b, c, d* shown in Table [2](#T2){ref-type="table"} represent numbers of accepted sequences belonging to the four attachment prototypes. The sets of 32 dichotomous items related to the attachment prototypes F, Ds, E, U, as well as the joint set of 96 insecure type items can be understood as a scale in the psychometric sense. These values of Cronbach α were satisfactory for the secure scale (0.77), for the U scale (0.82) for the joint insecure scale (0.88). However, they were *not* satisfactory for the Ds scale (0.64) and for the E scale (0.66). The correlation structure was investigated by means of the item-scale correlations and corrected item-scale correlations. The correlation structure was satisfactory for the system of two scales, secure and insecure. However, it was *not* satisfactory for the more detailed system of four scales F, Ds, E, U.
Guided by the referred psychometric results, we have decided to base the analyses of AAP-RC on the secure-insecure dichotomy. In respect of this aspect we have defined a *security index* expressing the degree of security vs. insecurity by the formula (see Table [2](#T2){ref-type="table"}): *a/r* = *a/(a + b + c + d)*. The index is a proportion of accepted secure sentences related to all accepted sentences, ranging from 0.00 (completely insecure) to 1.00 (completely secure). By the random answering independent of sentence prototype, it would oscillate around 0.25. The analogously constructed complementary insecurity index *(b + c + d)/(a + b + c + d)* is mathematically redundant; summing up to one, both indices contain the same information. Hence, the following analyses utilize the *security index* as a central measure.
Adult Attachment Reaction Time (ART): Reaction Time Analysis {#s3-3}
------------------------------------------------------------
The program *E-Prime* stored the dichotomous reaction and the needed RT in milliseconds. The hierarchically structured data sample consisted of 30 persons × 128 sentences = 3840 pairs of reactions and RTs.
Figure [2A](#F2){ref-type="fig"}: the average RT was about 1 s, ranging from 0 up to 15 s; exact values of measures and statistics see in Table [3](#T3){ref-type="table"}. As commonly experienced by the duration time data, the distribution was skewed and its normality was rejected by the Kolmogorov-Smirnov test.
{#F2}
######
**Kolmogorov-Smirnov test of normality of the reaction time (RT) distribution**.
Reaction time \[ms\] log~10~(time\[s\]) Speed pooled normalization Speed, intra-individual normalization
--------------------- ---------------------- -------------------- ---------------------------- ---------------------------------------
***N*** 3840 3840 3840 3840
***M*** 949 2.811 50.0000 50.0000
***SD*** 999 0.376 9.9996 9.9507
**KS statistic^1^** 0.195 0.060 0.002 0.005
***p*** *p* \< 0.001 *p* \< 0.001 1.000 1.000
^1^*The largest absolute difference between empirical and theoretical distribution, exact two-tailed test*.
Figure [2B](#F2){ref-type="fig"}: this was the case after applying the frequently recommended logarithmic transformation. Additional problems were caused by some extreme outlier values. Similarly, other transformations considered by Harris et al. ([@B36]) did not lead to satisfactory results in this case.
Figure [2C](#F2){ref-type="fig"}: transformation based on quantiles in the total pooled sample of 3840 measurements resulted in a close approximation to the normal distribution; the variable was transformed by the linear function *s*~(z)~ = 50--10*z*. The resulting variable is interpreted as the *speed* of the reaction. However, there were striking differences in the RTs between the 30 study participants, on an average ranging from 0.32 s up to 2.41 s. The reaction speed differed significantly by ANOVA (*F*~(29,3810)~ = 74.17, *p* \< 0.001; *η*^2^ = 0.36); a considerable portion of measurement variance was explained by the individual basic reaction speed.
Figure [2D](#F2){ref-type="fig"}: in regard to excluding the bias by individual basic reaction speed, we have normalized speed values intra-individually. The RTs for a test person were replaced by ranks 1 for the slowest reaction to 128 for the quickest one, and transformed to the *s* (speed) values according to the quantiles of the normal distribution *N*(50,10)[^1^](#fn0001){ref-type="fn"}. Because of the subsample sizes *n* = 128, the density curve of the obtained empirical distribution is less smooth than the previous one. Nevertheless, it is very close to the normal distribution *N*(50,10), and its normality in the sample of 3840 observations was *not* rejected by the exact Kolmogorov-Smirnov test.
The tests of study hypotheses (correlations, *t*-tests) were based on the last described intra-individually normalized speed values. The *N* = 3840 single values were aggregated to the intra-individual means for each of *N* = 30 study participants. Particularly, the following four aggregated values were relevant: speed of "yes" reactions to secure sentences; speed of "no" reactions to secure sentences; speed of "yes" reactions to insecure sentences; and, speed of "no" reactions to insecure sentences.
Convergent Validity Between the AAP Interview and the AAP-RC Security Index {#s3-4}
---------------------------------------------------------------------------
The convergent validity of the AAP-RC security index was examined by its comparison with the secure and insecure attachment classifications (Figure [3](#F3){ref-type="fig"}). Mean of the AAP-RC security index in the secure group (*n* = 10, *M* = 0.432, SD = 0.105) was greater than the mean in the insecure group (*n* = 20, *M* = 0.293, SD = 0.087); this difference was significant according to the two-sided two-group *t*-test: *t*~(28)~ = 3.866, *p* = 0.001, Cohen's *d* = 1.50 indicated a strong effect.
{#F3}
The predictability of attachment classifications on the basis of the security index in the attachment RT experiment was estimated by the discriminant analysis. The cross-validated classification was used, which is by small sample sizes particularly important (see "discriminant---cross-validation" in the IBM SPSS software system). The procedure recommended predicting AAP classification as secure when the security index exceeded the threshold 0.362 shown by horizontal line in Figure [3](#F3){ref-type="fig"}. Appling this threshold, 8 of 10 secure participants and 16 of 20 insecure participants were recognized correctly; the prediction was successful in 80% of cases in both groups.
Reaction Time to Accept or Reject Secure or Insecure Prototype Sentences {#s3-5}
------------------------------------------------------------------------
For each participant, the set of 128 measurements was divided by sentence prototype stimulus (secure, insecure) and his/her answer reaction (yes, no) into 2 × 2 = 4 subsets, as described above (see "Adult Attachment Reaction Time (ART): Reaction Time Analysis" Section). Within each subset, the intra-individual mean values of speed were computed, resulting in the speed values of the following four stimulus-reaction combinations: (1) accept secure sentences; (2) reject secure sentences; (3) accept insecure sentences; and (4) reject insecure sentences. These computations were based (a) on all 128 sentences and alternatively; (b) on 64 sentences relating to the alone pictures; and (c) on 64 sentences relating to the dyadic pictures.
These speed variables were compared by ANOVAs for 2 × 2 repeated measures in the whole sample of *N* = 30 participants. Results of analyses based on all, alone and dyadic stimuli are shown in Table [4](#T4){ref-type="table"}: (a) The analysis based on the complete material has shown that the *interaction of sentence prototype and answer* was significant (*p* = 0.011), whereas both main effects were not. As can be seen, participants answered more quickly to "yes to secure" and "no to insecure" and more slowly to "no to secure" and "yes to insecure". It means that the "secure-conform" answers were given more quickly than "insecure-conform" ones. (b) For the alone pictures, none of the three ANOVA effects was significant. (c) For the dyadic pictures, the *interaction* effect (*p* = 0.005) and the main effect *sentence prototype* (*p* = 0.020) were significant. The highest speed was observed for the combination "yes to secure"; the lowest speed and hence the highest time needed to answer was observed for the combination "yes to insecure".
######
**Speed of yes/no answers to secure/insecure prototype sentences**.
PROTOTYPE × ANSWER ANOVA
------------------- -------------------- -------- -------- -------- ----------- ------- -----------
All stimuli 51.25 48.89 48.85 50.05 2.290 1.026 **7.320**
(2.90) (2.90) (2.16) (1.30) 0.141 0.319 **0.011**
Alone stimuli 49.53 48.89 49.71 50.29 1.661 0.002 0.641
(4.17) (4.90) (3.55) (1.81) 0.208 0.968 0.430
Dyadic stimuli^x^ 52.14 49.40 47.50 49.81 **6.089** 0.278 **9.258**
(3.83) (5.71) (2.91) (1.84) **0.020** 0.603 **0.005**
*Repeated measures in the sample of N = 30 participants. *Bold*: significant effects. ^x^sample n = 28, df = 1, 27*.
Reaction Times in Secure and Insecure Attachment Groups According to AAP {#s3-6}
------------------------------------------------------------------------
The four speed variables (1--4) described in "Reaction Time to Accept or Reject Secure or Insecure Prototype Sentences" Section were compared between secure and insecure AAP attachment classification groups. As shown in Table [5](#T5){ref-type="table"}, the significant group differences were found for "RC-insecure" prototype sentences:
1. The insecure participants *accepted* the RC-insecure sentences more quickly than the secure participants.
2. The secure participants *rejected* the RC-insecure sentences more quickly than insecure participants.
######
**Speed of answers in participants with secure and insecure attachment according to the AAP classification**.
AAP-RC 2.0 AAP secure *n* = 10 AAP insecure *n* = 20 Cohen effect size Two-group t-test
-------------------- --------------------- ----------------------- ------------------- ------------------ -------- ----------- ------------ -----------
**All stimuli**
Secure Yes 51.34 (2.81) 51.20 (3.01) +0.05 +0.123 0.903
Secure No 47.65 (5.90) 49.51 (3.08) −0.44 −1.139 0.265
Insecure Yes 47.00 (2.46) \*49.77 (1.24) **−1.60** **−4.134** **0.000**
Insecure No \*50.73 (0.86) 49.71 (1.37) **+0.84** **+2.164** **0.039**
**Alone stimuli**
Secure Yes 49.80 (3.11) 49.39 (4.68) +0.10 +0.250 0.804
Secure No 47.06 (5.45) 49.81 (4.46) −0.57 −1.475 0.151
Insecure Yes 47.70 (3.68) \*50.72 (3.11) **−0.91** **−2.359** **0.026**
Insecure No 50.56 (1.60) 50.16 (1.94) +0.22 +0.565 0.576
**Dyadic stimuli**
Secure Yes 52.03 (3.77) 52.19 (3.95) −0.04 −0.108 0.915
Secure No^x^ 49.75 (7.01) 49.26 (5.3ß) +0.08 +0.201 0.842
Insecure Yes 45.91 (3.76) \*48.29 (2.07) **−0.87** **−2.256** **0.032**
Insecure No \*50.82 (1.47) 49.30 (1.82) **+0.88** **+2.277** **0.031**
*Group comparison. *Bold*: significant t-test. \*significantly higher group mean. ^x^sample n = 28, AAP secure n = 8, df = 26*.
The first result was also confirmed for both subsets of alone and dyadic sentences. The second result was confirmed for sentences connected to the dyadic pictures. In sum, the differences between the secure and the insecure attachment group according to the AAP were significantly manifest for the RC-insecure sentences.
Reaction Times in Correlation to the Security Index AAP-RC in ART {#s3-7}
-----------------------------------------------------------------
The AAP-RC security index (see "Relationship Choices, version 2.0 (AAP-RC): Psychometrical Analysis and Security Index" Section) ranges from completely insecure (0.0) to completely secure (1.0) reactions to the 128 stimuli. The correlations of the AAP-RC security index with variables concerning the reaction speed by four stimulus-reaction pairings are shown in Figures [4A--D](#F4){ref-type="fig"}.
{#F4}
\(A\) The RT speed of acceptance of secure sentences (Figure [4A](#F4){ref-type="fig"}) was *not* significantly correlated with the security index; according to ART both rather securely or insecurely attached persons reacted comparable quick in this case.
\(B\) The rejection speed of secure sentences (Figure [4B](#F4){ref-type="fig"}) was negatively correlated with the security index; the rather insecurely attached persons rejected the secure sentences more quickly.
\(C\) The acceptance speed of insecure sentences (Figure [4C](#F4){ref-type="fig"}) was negatively correlated with the security index; the rather insecurely attached persons accepted the insecure sentences more quickly.
\(D\) The rejection speed of insecure sentences (Figure [4D](#F4){ref-type="fig"}) was positively correlated with the security index; the rather securely attached persons rejected the insecure sentences more quickly.
Summarizing, the results of the experiment---especially for insecure prototype sentences---indicate a consistency between the attachment preferences and the higher speed of the corresponding reaction. The complete results of the experiment are summarized visually in Figures [4A--D](#F4){ref-type="fig"}, which underlines the consistency of the different results.
Discussion {#s4}
==========
Discussion of the Methodology and Results {#s4-1}
-----------------------------------------
Bowlby ([@B9]) proposed that inner working models of attachment function automatically and outside of conscious awareness. RT analyses are a way to observe the implicit automatic reactions. Pietromonaco and Barrett ([@B3]) recommend the use of implicit measures like RTs to capture the unconscious parts of attachment, which are obscured when relying on self-report measures. Therefore we chose a narrative attachment measure, the AAP, designed to elicit unconscious elements by focusing on defensive processes, in combination with a RT measure. Our results reflect that implicit methodology can bring new and interesting insights in attachment related processes in the domain of neuroscience of human attachment.
Participants in our study were tested in an attachment RT experiment using the AAP picture stimuli accompanied by prototypic sentences representing different attachment representations (AAP-RC). Participants were not informed about their attachment classification prior to beginning the experimental session; therefore their reaction to the sentences was considered to be automatic and outside of conscious appraisal (i.e., unconscious). We hypothesized that the participants would accept the prototype sentences in the attachment RT paradigm: (1) more frequently; (2) more quickly when sentences matched with their own representations of attachment classification; and (3) that this would correspond with their attachment prototype preferences in the RT experiment (ART). Overall most of our expected results were confirmed for secure vs. insecure attachment groups.
The distribution of attachment classifications in our sample showed an overrepresentation of dismissing participants as compared to the distributions in samples with healthy controls (Bakermans-Kranenburg and van Ijzendoorn, [@B1]). Therefore, one caveat of our study is that we did not have a representative distribution of attachment classifications. Another caveat is the small sample size. The consequence was that data analyses for separate attachment groups was not possible and we were confined to comparisons of participants with secure and insecure attachment representations. This remains a challenge for our next studies.
Item-scale analyses confirmed the internal consistency for the secure and insecure scales. Discriminant analysis showed that AAP-AAP RC 2.0 convergence prediction was successful in 80% of the cases in both groups; 8 of 10 secure and 16 of 20 insecure participants. Although a 100% correspondence was not reached, there was a sufficient agreement in this study to demonstrate the validity of the paradigm. This association was stronger, for example, than the results of studies that correlated narrative and self-report attachment measures (e.g., Roisman, [@B55]).
Given that participants did not know their own attachment classifications by the standard AAP procedure, we can assume that they were not guided by informed conscious appraisals of attachment while evaluating the prototype sentences, rather by unconscious processes. The fact that the different measures showed a considerable convergence supports the conclusion that we were able to capture both conscious and unconscious automatic reactions to attachment related stimuli. The average RTs differed significantly between the study participants. The considerable portion of variance of the originally measured RTs is explained by the individual basic speed of reactions to the presented stimuli. This empirically found fact can be caused by different plausible reasons, like overall speed or slowness of mental processes of the subject, extended rational reasoning on the presented sentences, or intensive imagination triggered by them.
With regard to the RT results, we found that all participants had a tendency to answer "yes" to secure and "no" to insecure sentences quickly and more slowly when the cases were inverted (i.e., "no" to secure and "yes" to insecure). ANOVA did not show significant results for both main factors sentence prototype stimulus and answer reaction; the *interaction effect prototype and answer* was significant however. The participants accepted secure prototype sentences and rejected the insecure prototype sentences more quickly. One possible explanation of this finding is social desirability, because the perception of secure sentences could be expected to be ideal. This is in line with findings by De Carli et al. ([@B23]). In their IAT study about caregiving and attachment, which they proposed as two different systems, the authors found that adult attachment style had a role in shaping the implicit attitude, but not the explicit attitude, concerning the category "mother." The explicit attitude did not appear to be influenced in that study by experimental manipulation or the participants' attachment style. The authors discussed that this can be explained by social desirability, because the perception of mother is expected to be mostly positive. In sum the IAT findings of De Carli et al. ([@B23]) in the context of the transmission of attachment are in line with our results by showing that the participants preferred a particular style of caregiving coherent with their own attachment style. However the authors pointed out that their attachment measure was a self-report instrument that captured explicit thoughts only. However a notable strength in our study is that we used a free-response narrative attachment assessment measure, which seems to be more appropriate for this kind of experimental approach because of its implicit nature. Yet the role of social desirability should be clarified in future studies.
Despite the results, that all participants in our study accepted the prototype secure sentences faster than insecure prototypes, there were significant differences between the two adult attachment groups. Secure participants accepted more prototype secure sentences and showed faster RTs than insecure participants. Insecure participants accepted more insecure sentences, and did so faster than secure participants. This result underscores the presence of automatic unconscious detection and appraisal processes when responding to attachment relevant information.
Parallel patterns were found in the AAP-RC with the RTs in the experiment. Participants with higher preference for secure prototype secure sentences rejected insecure sentences more quickly. Participants with higher preference for insecure prototype sentences accepted insecure sentences more quickly and rejected secure ones more quickly.
Our differential hypotheses addressed secure and insecure prototype sentences. Findings supported our hypotheses, and we confirmed all hypotheses concerning the insecure prototypes. In other words: "*accept secure and reject insecure"* goes *fast*, and "*reject secure and accept insecure"* goes *slow*. It seems that the "insecure-type" reactions demand more time.
In a study by Rösler ([@B56]), more complex processes took longer elaboration time than more simple ones. What makes the insecure reaction more demanding than the secure one in our study? We can nearly exclude that the linguistic or cognitive complexity would play a role: the grouped sentences had the same length and they were clear and understandable. However, the complexity of the relationship related decision processes might differ. It is reasonable then to conclude that the differences were due to story content. George and West ([@B33]) described, how different insecure attachment representations are connected to different defensive mechanisms. The insecure attachment prototypes have the potential to address inner conflicts (e.g., ambivalence or deactivation of attachment relevant information), which must be recognized first, and then accepted or rejected. This unconscious process might request the additional "working time"[^2^](#fn0002){ref-type="fn"}.
Our findings are in line with those of Vrtička et al.'s ([@B63]) study of attachment style. These researchers used an explicit choice paradigm and found distinct effects of attachment avoidance and anxiety on subjective emotional judgments. Their results supported the assumption that anxious attachment is associated with a hyperactivating tendency for the appraisal of social threat, but may also involve an ambivalence influencing the judgment of information. Although, the authors did not use a RT experiment, their results support thinking that proposes that insecure attachment seems to need more mental elaboration time.
Therefore we could have assumed in our study that individuals with preoccupied attachment representations associated with heightened emotional reactivity would show different RT patterns compared to dismissing individuals, characterized by deactivating attachment related emotions. This important differentiation should be the next step in future studies with a larger sample size.
According to previous research with the AAP we might have also expected particularly differentiated results for the analyses based on "alone pictures" compared to "dyadic" ones. Alone pictures represent scenarios of emptiness and loneliness and seem to elicit high affective arousal in participants (Buchheim and George, [@B14]). However the results of the present study showed that insecure individuals needed longer elaboration times confronted with the dyadic pictures. This type of sentences (like in the AAP picture of the couple in the scenario "departure") represents explicit attachment related scenarios between two or more persons (potential separation, need for care). We might conclude that insecure individuals needed more elaboration time for processing these attachment related conflicts. The observed differences should be verified in further investigations using a larger sample.
In sum, high security index scores were associated with prompt rejection of insecure prototype sentences. Lower security index scores were associated with prompt acceptance of insecure sentences, as well as rejection of secure sentences. Some other hypotheses could not be confirmed significantly; there were no contrary findings nevertheless. We might have demonstrated that the secure vs. insecure attachment classification groupings could be observed with the implicit measure, by observing the activation of inner working model in "real time."
Our results support the conceptualization of inner working models of attachment as guiding attention and interpretation outside of conscious awareness and the coherency of the association between mental representation and interpretation of attachment situations (Bowlby, [@B9]).
From a methodological perspective, we suggest that the observation of RTs is valuable to complement the spectrum of mainstream measures in human neuroscience, like brain mapping or EEG analyses. These highly advanced measures focus on brain localizations and processes associated with different psychological tasks and events. The RT approach investigates the overall time of participants to specific stimuli analogous to the time complexity theory in computer science (Sedgewick and Wayne, [@B60]). The more operations are needed for the problem solution, the more time is needed. The time needed for the problem solution might then constitute an operationalization of the problem complexity and depends on numerous biasing factors. Human processing time consists of the individual's basal or momentary reaction speed including external disturbing influences, which could cause long outlier RTs. The data-analytic procedure proposed in this article was designed with the aim to be robust against the mentioned biasing factors and could be a fruitful additional approach in an EEG analysis when using a similar paradigm.
Limitations {#s4-2}
-----------
The size of our sample of 30 participants was sufficient, albeit small, for the experimental investigation of the RT phenomena. The number of 128 attachment prototype sentences was considerably larger than the sample size; this circumstance limited the use of more advanced psychometrical analyses (e.g., factor analysis). Similarly, the sample sizes and the distribution of four particular attachment groups led us to the decision to confine the analyses to the two basic attachment classifications secure and insecure. In fact, secure vs. insecure analyses are often chosen as a comparison in the field of attachment. However as we discussed it would have been valuable to differentiate the insecure attachment groups and the individuals' RTs. This aspect should be tested in further research.
In sum the present study served as a pilot study to test its feasibility in healthy participants. The next steps are the application of the RT experiment in clinical studies with a larger sample. Moreover, the AAP measure is constructed and validated for adults and adolescents only, so the application is limited to that age group and not feasible for children, where other measures should be used, like the Separation Anxiety Test (Klagsburn and Bowlby, [@B42]).
Despite these limitations, the study has shown that the concept of immediate reactions to stimulus sentence could be beneficial for experimental attachment research contributing to measure the intensity of unconscious processes empirically. As a following research step, we intend using psychometric procedures to continue and improve the development of the AAP-RC instrument in order to implement it in a neurobiological setting.
Outlook: Neurobiological Model Using the Reaction Time Experiment on Adult Attachment {#s4-3}
-------------------------------------------------------------------------------------
In the presented study, we have seen that stimuli with more distressing attachment content might need a longer RT for its elaboration than stimuli with more harmonious content. Future studies need to replicate these findings using larger samples. A further next step is to adapt the experiment for an EEG setting, which could give further insight into the neural mechanisms of potential response delays during an implicit task.
One of the most interesting areas in the research of preconscious perception is the investigation of early brain potentials. Until now, there are only a small number of studies examining the perception of emotional stimuli in individuals with different attachment patterns. In an EEG setting the N1 potential, which is also called N170 component, is considered to be a very sensitive representation of early perceptual processing. Spatio-temporal analyses of brain activity patterns during the first 200 ms after stimulus presentation have characterized the timing of attentional selection processes and different stages of feature encoding and pattern analyses (Hillyard et al., [@B38]). In an attachment study on *face recognition* Zhang et al. ([@B67]) reported distinct differences in N1 activation using self-reports. The perception of angry faces was followed by high N1 amplitudes in anxious and secure individuals in contrast to the smaller amplitudes in avoidant individuals. Given that N1 is considered to be an index of the level of attention, the authors suggested that individuals with anxious attachment "use most, and avoidant individuals use least attentional resources to face stimuli than secure individuals". The authors considered these differences as the results of automatic processes in association with conscious and preconscious emotional information processing. In contrast to the latter study Fraedrich et al. ([@B26]) focused on event-related potentials (ERPs) in mothers during the perception of infant emotions by presenting positive, negative and neutral facial expressions as well as non-facial stimuli within an oddball paradigm. Dismissing mothers exhibited elevated N170 amplitudes for *facial* target stimuli within conditions that contained frequent non-facial stimuli. In summary, the findings suggested that insecure mothers require more cognitive resources to process infant faces, while secure mothers allocate more attention to infant faces and clearly show a perceptual bias toward social information. The differences between the study results of Zhang et al. ([@B67]) and Fraedrich et al. ([@B26]) might be due to the different stimulus material.
In a very recent study by Leyh et al. ([@B45]), the association between maternal attachment representation and brain activity (ERPs) underlying the perception of infant emotions was examined. Securely attached mothers recognized emotions of infants more accurately than insecurely attached mothers. ERPs yielded amplified N170 amplitudes for insecure mothers when focusing on negative infant emotions. Secure mothers showed enlarged P3 amplitudes to target emotion expressions of infants compared to insecure mothers, especially within conditions with frequent negative infant emotions. In these conditions, P3 latencies were prolonged in insecure mothers.
One potential limitation of attachment research of preconscious perception with the help of the early brain potentials so far might be the predominant focus on *face processing* as the stimulus material. Neural processing in secure and insecure subjects were not examined by attachment related material directly linked to the individuals' own attachment representations using a paradigm where spontaneous preferences had to be given in a defined time frame.
In a recently published article by Matheus-Roth et al. ([@B49]) early occipital ERP's (e.g., P100 and N170) have been shown to be sensitive for a "preference" for stimuli with alcohol association in patients with alcohol dependance. The authors used a Go-NoGo paradigm with three visual stimuli: tea, juice and beer. The N170 amplitudes were elevated in response to the alcohol-related (beer) stimuli in the NoGo condition in these patients compared to controls. The patients had to react to the frequent tea stimuli and ignore the beer and the juice stimuli. While the higher N170 component correlated with a relapse within the following 3-month, the shorter P100 latencies were related to higher depression scores. The latencies of these early ERPs represent the "RTs" of the brain, presumably independent of deliberate influence. In another study, the so called "mismatch negativity" (MMN) has been demonstrated to react pre-attentively to syntactic or semantic errors (Menning et al., [@B50]). The authors used an auditory oddball design with frequent standard sentences to elicit a memory trace, which was interrupted by rare deviant (erroneous) sentences. Moreover, Hietanen and Nummenmaa ([@B37]) revealed that N170 is sensitive to stimuli of naked bodies. In their studies it is even greater for nudes than to faces. Overall N170 seems to be an indicator for the preconscious individual importance of visual stimuli.
Finally the analysis of P300 component---an indicator for emotional operations---might reveal interesting results (Nieuwenhuis et al., [@B52]; Schupp et al., [@B59]; Flaisch et al., [@B25]). However, assuming that P300 is a correlate of conscious perception (Dehaene et al., [@B24]), more early EEG components like cited above should be considered first to capturing modes of more unconscious processes.
In sum these neurophysiological and the other cited attachment studies investigating implicit aspects of romantic attachment using self-report measures as explicit instruments for assessing attachment style (Marks and Vicary, [@B48]; De Carli et al., [@B23]) suggest that early visual and auditory stimuli could be used as a change detector of emotionally preferred stimuli. Thus, transposed to our tested and validated AAP RT paradigm, we would expect that the specific (secure or insecure) attachment system paves the way for a specific ERP, e.g., higher amplitudes or shorter latencies of the N170 or P300 to individual preferred stimuli which represent the own attachment representation. One advantage of our paradigm would be to use attachment related material linked to the individuals' inner working models of attachment in a RT setting. This might extend previous studies in healthy samples and may provide some feasibility for clinical studies.
The measures based on RT reflect the overall activity of the brain needed for the elaboration of different stimuli. The results of the referred study suggested that the overall time needed for the processing of "unpleasant", discomforting stimuli was higher than for "pleasant", comforting ones. The fact that RT showed convergence with the individual' inner working model of attachment in our study, has the potential to contribute to the validity of neurobiological experiments, like EEG. Therefore RT analysis with the proposed evaluation procedures might be of interest for a broader variety of questions concerning attachment in experimental and neurophysiological settings to capture automatic, unconscious processes in association with internal working models of attachment.
Author Contributions {#s5}
====================
The study was conceptualized by AB, CG, TW and DP. The attachment experiment was developed by AB. The study setup and data collection were organized and conducted by TW. Coding of attachment interviews were conducted by AB. DP performed the statistical data analysis and contributed substantially to the result interpretation. DP developed the statistical procedure for RT analyses. CG, DP, TW, HM, IS and AB provided important intellectual contribution in commenting and revising the manuscript. AB, DP and TW wrote major parts of the manuscript and edited its final version.
Funding {#s6}
=======
The publication is funded by the Faculty of Psychology and Sports Science, University of Innsbruck, Austria; Research Funding for Young Scientists.
Conflict of Interest Statement {#s7}
==============================
The 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.
This would not have been possible without the support of many colleagues. Many thanks also to Florian Juen and Ann-Christine Jahnke-Majorkovits for rating the diverse attachment sentences and to Stefan Fischer for methodological advice at the beginning of the study.
^1^Technical note on ties: the ties were resolved by the replacing values by the average quantile value. For instance, by the subject p09, the first three quickest reactions needed the same time 0.185 s. Without ties, the ranks 1, 2 and 3 would correspond to the speed scores 76.6, 72.7 and 70.6. Because of ties, the mean value of three of these scores 73.3 was considered, rather than 72.2 corresponding to the mean rank 2.0. Resolving ties in this way, the intra-individual mean value was exactly 50.000 for each participant, the intra-individual standard deviations were very close to the value 10.000 (9.981--9.989), depending on the number of ties.
^2^An analogous approach is being used in the computer science: the complexity of a problem is classically operationalized by the number of needed steps of the problem solving algorithm, and consequently by the time demanded for the problem solution; see Sedgewick and Wayne ([@B60]).
AAP
: Adult Attachment Projective Picture System
AAP-RC
: Adult Attachment Projective Relationship Choices Version 1
AAP-RC 2.0
: Adult Attachment Projective Relationship Choices Version 2
ART
: Attachment Reaction Times
Ds
: dismissing attachment
E
: preoccupied attachment
EEG
: Electroencephalography
ERP
: Event related potential
F
: secure attachment
fMRI
: Functional magnetic resonance imaging
RT
: reaction times
U
: unresolved trauma.
[^1]: Edited by: Mikhail Lebedev, Duke University, USA
[^2]: Reviewed by: Martin Brüne, Ruhr University Bochum, Germany; Saori Nishikawa, University of Fukui, Japan
[^3]: ^†^ These authors have contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION {#s1}
============
Breast cancer (BC) is the most frequently diagnosed cancer among women (American Cancer Society) \[[@R1]\]. Heterogeneity of BC has been documented in BC diagnoses and therapies over the past two decades. The identification of heterogeneous biomarkers and molecular-driven therapies benefits patients with individualized treatment \[[@R2]--[@R3]\]. Though outcomes of BC have been improved due to this understanding, challenges still remain because of the aggressive nature, lack of representative biomarkers and potentially curable targets \[[@R4]--[@R6]\].
Inter-independent genetic profiles reported that the gene amplification of EPHA10 is between 0.8%-17.2% in breast cancer, and a missense mutation in the RTK domain (TCGA data) in 0.2% (5/482) of BC cases \[[@R7]--[@R9]\]. However, the EPHA10 protein is observed to be overexpressed in all BC subtypes regardless of their hormone status \[[@R7], [@R10]\]. The function and mechanism of the EPHA10 protein remains unclear. EPHA10 belongs to the Eph receptor tyrosine kinase family which characters with dual-signaling transduction of forward signaling (tyrosine kinase activity via Eph receptor) and reverse signaling (receptor-ligand interaction via ephrin ligand) \[[@R11]\]. Importantly, several lines of researches demonstrate that the transduction approach is implicated in the cell-cell interaction, tissue development, and tumor progression \[[@R12], [@R13]\]. Due to an altered β7-asparagine in the receptor tyrosine kinase (RTK) domain, the RTK of EPHA10 protein is a pseudo kinase, and thus, phosphorylation of substrate is blunted \[[@R14]--[@R16]\]. Overexpression of EPHA10 protein can be annotated by two isoforms ([Supplementary Figure 1](#SD1){ref-type="supplementary-material"}). EphA10s (NP_775912.2) is a small protein with a single Eph receptor A10 domain, which is predicted as a secretary protein (UniProtKB-Q5JZY3). EphA10 (NP_001092909.1) is a type I membrane protein with Eph receptor A10 domain, Fibronectin type III domain, and the pseudo-RTK domain. In this study, we investigated the expression pattern of EPHA10 isoforms and their biological functions in BC development and progression.
In the malignant transformation of breast cancer, E-cadherin (ECAD) and β-catenin are classically regarded as key players in cell-cell adhesion, and regulator in dynamic behavior \[[@R17], [@R18]\]. Notably, their expression and distribution were identified as biomarkers of breast cancer stem cells \[[@R18], [@R19]--[@R20]\]. ECAD plays a role in mediating calcium-dependent cell-cell adhesion, where the intercellular domain of ECAD binds to components such as β-catenin and p120-catenin, and associates with reorganization of actin cytoskeleton \[[@R21]\]. Under certain conditions, ECAD is phosphorylated (pECAD) which activates ECAD endocytic pathways and results in disassociation of ECAD complex \[[@R22], [@R23]\]. Loss of ECAD expression in membrane is one of the characteristic events of epithelial mesenchymal transition \[[@R24]\]. β-catenin has dual roles in the regulation of cancer progression. On one hand, β-catenin binds to ECAD in membrane where it facilitates cell-cell adhesion \[[@R25]\]. On the other hand, isolated from the ECAD complex (non-canonical pathway) \[[@R26]\] or from the Wnt signaling pathway (canonical pathway) \[[@R27]\], unbounded β-catenin acts as a strong oncogenic transcription factor \[[@R28]\]. In this study, we demonstrated that the expression pattern from EphA10s and EphA10 in BC weakens the stability of the membrane complex of ECAD and β-catenin, thus, promotes dynamic behavior of cancer.
RESULTS {#s2}
=======
Expression of EPHA10 protein in breast cancer {#s2_1}
---------------------------------------------
Expression of EPHA10 protein was screened by IHC staining with an anti-Eph receptor A10 domain antibody (Figure [1A-1C](#F1){ref-type="fig"}). In invasive samples (n=325), EPHA10 protein is highly expressed in cytoplasm in compared to benign samples (n=76), and in lymph-node metastasis samples (n=50), a significantly additional expression of EPHA10 protein was observed ([Supplementary Table 1](#SD1){ref-type="supplementary-material"}). Moreover, our data supported that EPHA10 protein is over expressed in samples with TNM stage severity ([Supplementary Table 2](#SD1){ref-type="supplementary-material"}).
{#F1}
Measuring EPHA10 protein expression by immunostain could be oversimplified because it is difficult to differentiate isoforms as EphA10s and EphA10. Therefore, we performed qPCR to identify mRNA expression of EphA10s and EphA10 in qualified 42 benign samples and 245 invasive samples. EphA10s is down regulated (*P*\<0.001) while EphA10 is highly expressed (*P*\<0.001) in invasive samples, and expression pattern of EPHA10 isoforms shifts from EphA10s/EphA10 (42.9%/33.3%) in 42 benign samples to EphA10s (13.6%) with dominantly expression of EphA10 (70.6%) in invasive samples (Figure [1D-1E](#F1){ref-type="fig"}, and [Supplementary Table 3](#SD1){ref-type="supplementary-material"}). Correlating IHC staining with expression of EphA10s and EphA10, we found a closer relation with EphA10s (r=-0.502, *P*=0.006) in benign samples, and with EphA10 (r=0.603, *P*\<0.001) in invasive samples (Figure [1F](#F1){ref-type="fig"}).
EphA10s and EphA10 expression patterns associate with breast cancer outcomes {#s2_2}
----------------------------------------------------------------------------
In invasive samples, EphA10s is low expressed regardless of age, tumor size, lymph-node metastases, TNM stages, or cancer subtypes (Table [1](#T1){ref-type="table"}). Differently, EphA10 is generally up-regulated and significantly increased in tumor-size enlargement (*P*=0.042) and TNM stage severity (*P*=0.002). Moreover, the expression level of EphA10s and EphA10 was found to contribute to the outcome in 5-year follow-up ([Supplementary Figure 2](#SD1){ref-type="supplementary-material"}). Patients with higher Eph receptor A10 staining show a poor expectation in disease-free survival analysis and in overall survival (OS) analysis (*P*=0.015, Log Rank HR=1.760 \[95% CI: 1.071- 2.893\]). When detailed by EphA10s and EphA10, higher EphA10s expression associates with a better disease-free and OS period, while higher EphA10 expression suggests a poor disease-free survival period and a poor OS expectation (*P*=0.013, Log Rank, HR=1.975 \[95% CI: 1.157 -3.371\]).
###### Profile of EphA10s and EphA10 expression in breast cancer characters
Traits qPCR:EphA10s *P* value qPCR:EphA10 *P* value
---------------------- ----- -------------- ----------- ------- ------------- ----------- ---------- -----------
**AGE** 245 245
≥45 178 24(13.5) 154(86.5) 0.667 178 128(71.9) 50(28.1) 0.053
\<45 67 7(10.4) 60(89.6) 67 45(67.2) 22(32.8)
**pT** 245 245
T1 and T2 203 26(12.8) 177(87.2) 1.000 203 149(73.4) 54(26.6) **0.042**
T3 and T4 42 5(11.9) 37(88.1) 42 24(57.1) 18(42.9)
**pN** 245 245
N0 203 29(14.3) 174(85.7) 0.125 203 138(68.0) 65(32.0) 0.062
N1 and N2 42 2(4.8) 40(95.2) 42 35(83.3) 7(16.7)
**pSTAGE** 245 245
I and II 210 23(11.0) 187(89.0) 0.058 210 141(67.1) 69(32.9) **0.002**
III 35 8(22.9) 27(77.1) 35 32(91.4) 3(8.60)
**Molecular Typing** 236 236
Luminal A and B 134 12(9.00) 122(91.0) 0.159 134 98(73.1) 36(26.9) 0.207
HER-2 80 14(17.5) 66(82.5) 80 54(67.5) 26(32.5)
Basal-like 22 2(9.1) 20(90.9) 22 19(86.4) 3(13.6)
Tumor-node-metastasis (TNM) is classified according to the revised American Joint Committee on Cancer TNM classification. *P* value is evaluated with Pearson Chi-square, two tails.
Secretary EphA10s inhibits cell migration and invasion {#s2_3}
------------------------------------------------------
EphA10s was down regulated in MDA-MB-231 cells (invasive) in compared to MCF-10A cells (benign, [Supplementary Figure 3](#SD1){ref-type="supplementary-material"}). To verify EphA10s as secretary protein, cell lysate (CL) and supernatants (S) of MCF-10A cells, and MDA-MB-231 cells with lentiviral EphA10s overexpression, were subjected to Western Blot (Figure [2A](#F2){ref-type="fig"} and [Supplementary Figure 4A](#SD1){ref-type="supplementary-material"}). Both natural EphA10s (MCF-10A) and overexpressed EphA10s are detected in culture supernatants. Moreover, EphA10s expressing MDA-MB-231 cells displayed a concomitant decrease in migration by 0.46 fold and in invasion by 0.33 fold (Figure [2B](#F2){ref-type="fig"}). When culture MDA-MB-231 cells with supernatant from EphA10s-expressing cells, the similar dynamic decrease was observed ([Supplementary Figure 4B](#SD1){ref-type="supplementary-material"}).
{#F2}
EphA10s assists with β-catenin distribution and stability in membrane {#s2_4}
---------------------------------------------------------------------
Overexpression of EphA10s is found in resembling of F-actin filaments (Figure [2C](#F2){ref-type="fig"}). Molecular analysis in regulation of actin cytoskeleton was performed accordingly. EphA10s overexpression results in decrease of total β-catenin expression and the level of its phosphorylated version, while the similar findings were observed in cells cultured with supernatant from EphA10s-expressing cells (Figure [2D](#F2){ref-type="fig"}). Total β-catenin is explained by its subcellular accumulation. Immunocytochemistry staining shows that accumulation of β-catenin enhances in membrane in the present of EphA10s. The measurement of β-catenin in consecutive membrane staining shows a significant increase by 2.3 fold (*P*\<0.001, Figure [2E](#F2){ref-type="fig"}). Furthermore, we enriched protein from membrane (M) and cytoplasm (CP), respectively. Immunoblot shows that accumulation of β-catenin increases in membrane and apparently decreases in cytoplasm in the presence of EphA10s, in which the β-catenin ratio of membrane-associate over cytoplasmic is enhanced by 2.7 fold (*P*=0.034, Figure [2F](#F2){ref-type="fig"}), while the total β-catenin shrinks.
Ligand ephrin A5 participates in the redistribution of β-catenin {#s2_5}
----------------------------------------------------------------
The interaction of EphA10s needs to bind with its ligands \[[@R11]\]. Proteins from ephrin family were analyzed in affinity and kinetic activity with Eph receptor A10 \[[@R14], [@R29]\]. ephrin A3, ephrin A4 and ephrin A5 shows the best performance based on the evaluation in structural and chemical simulation. We, thus, screened expression of ephrins in benign and invasive samples. ephrin A5 presents a significantly decrease in cancer, in which it shows an intensive correlation with the co-expression of EphA10s (r=0.558, *P*=0.004; [Supplementary Figure 4C and 4D](#SD1){ref-type="supplementary-material"}). Therefore, we knocked down the expression of ephrin A5 in cells. Immunoblot shows that the inhibition of total β-catenin in present of EphA10s is rescued in low-ephrin A5 expressing cells (Figure [2G](#F2){ref-type="fig"}). To verify the interaction, membrane protein from EphA10s^−^/ EphA10^−^ cells were subjected to anti-Eph receptor A10 immunoprecipitation (IP, Figure [2H](#F2){ref-type="fig"}). The blots of ephrin A5 shows in cells both with endogenous EphA10s or extraneous EphA10s.
Overexpression of EphA10 accumulates in cytoplasm and promotes invasion {#s2_6}
-----------------------------------------------------------------------
Using anti-Eph receptor A10 antibody, IHC staining is strongly accumulated in cytoplasm of invasive and metastatic samples, where the staining is closely correlated with EphA10 expression. We investigated the distribution *in vitro* (Figure [3A](#F3){ref-type="fig"}). In MCF-10A cells, EphA10 is accumulated in the membrane after treated with proteasome inhibitor MG132. Differently, in MDA-MB-231 cells, it shows an apparently cytoplasmic distribution, which is even stronger after using MG132. The observation was verified in Western Blot (Figure [3B and 3C](#F3){ref-type="fig"}). Knocking down EphA10 expression, the capability of migration and invasion in invasive cells is significantly decreased by 1.6 fold (migration) and 1.3 fold (invasion), with a concomitant depolymerization and disruption of F-actin (Figure [3E](#F3){ref-type="fig"}). However, down-regulation of EphA10 in MCF-10A cells results in less change of similar dynamic behaviors (data not shown).
{#F3}
EphA10 associates with ECAD phosphorylation {#s2_7}
-------------------------------------------
For a better understanding the difference, both benign cells and invasive cells were subjected to further investigation. In Western Bolt (Figure [3F](#F3){ref-type="fig"}), knocking down EphA10 by lentiviral infection in MCF-10A cells leads to a slightly increase of pECAD, while in MDA-MB-231 cells results in a shrink of pECAD with a decrease of N-Cadherin (NCAD). The phosphorylation of ECAD complex involves the tyrosine kinase induced tyrosine phosphorylation \[[@R23]\]. We blocked the activity of tyrosine kinase by using inhibitors AG490 and Genistein \[[@R30]\] for 6 hours (Figure [3G](#F3){ref-type="fig"}). Immunoblot shows that both inhibitors notably blunt pECAD level without decreasing EphA10 expression.
EphA10 is redistributed to cytoplasm in invasive cells. We performed anti-Eph receptor A10 IP to understand the EphA10 interaction in either membrane or cytoplasm in both benign and invasive cells (Figure [3H and 3I](#F3){ref-type="fig"}). In both cellular fractionations, pseudo-RTK EphA10 shows a low activity of phosphorylation. In MCF-10A cells, membrane-associate EphA10 recruits the complex of ECAD and β-catenin, while low interaction is detected in cytoplasm. Knockdown EphA10 from membrane abolishes the interaction, in which ECAD associated with enhanced level of phosphorylation. In MDA-MB-231 cells, cytoplasmic EphA10 associates with pECAD, EphA10 knocking down weakens the interaction.
Up-regulating EphA10s and down-regulating EphA10 inhibits cell invasion {#s2_8}
-----------------------------------------------------------------------
We modified expression patterns of EphA10s and EphA10 in MDA-MB-231 cells by lentiviral infection for a further investigation. Expression patterns were identified as high/low EphA10s (EphA10s^+^/EphA10s^−^) and high/low EphA10 (EphA10^+^/EphA10^−^), respectively. Compared to EphA10s^−^/EphA10^+^ cells, silencing EphA10 results to 0.62-fold decrease in the ability of invasion, with additional expression of EphA10s, the invasive ability collapses to 0.33 fold (ANOVA, *P*=3.05E-04, Figure [4A and 4B](#F4){ref-type="fig"}).
{#F4}
Cells with different expression of EphA10s and EphA10 were subjected to Western Blots for the analysis of regulation in ECAD complex (Figure [4C](#F4){ref-type="fig"}). Overexpressing EphA10s weakens the total expression of β-catenin while down-regulating EphA10 results in a decrease of pECAD and total β-catenin. Moreover, anti-β-catenin IP was employed to investigate the decrease of total β-catenin in the scenario (Figure [4D](#F4){ref-type="fig"}). Overexpressing EphA10s and down-regulating EphA10 recruit more membrane-associate β-catenin and apparently decreases cytoplasmic β-catenin, in which the β-catenin ratio of (M/CP) increases to 5.6 fold in EphA10s^+^/EphA10^−^cells (Figure [4E](#F4){ref-type="fig"}). Meanwhile, changes in the expression of EphA10s and EphA10 did not have significant impact on cell proliferation and apoptosis (Figure [4F and 4G](#F4){ref-type="fig"}).
Higher EphA10s and lower EphA10 reverse lymph-node metastasis {#s2_9}
-------------------------------------------------------------
MDA-MB-231 cells with different EphA10s and EphA10 expression were employed into xenograft model (Figure [5A](#F5){ref-type="fig"}). Tumors with low EphA10s (EphA10s^−^) and high EphA10 (EphA10^+^) took a half of the time to grow until being sized as 8 mm × 8 mm than time that EphA10s^+^/EphA10^−^ tumors required (48 days, [Supplementary Figure 5](#SD1){ref-type="supplementary-material"}). Tumor mass with similar size was collected for weighting (Figure [5B](#F5){ref-type="fig"}), and re-identifying in expression pattern of EPHA10 isoforms (Figure [5C](#F5){ref-type="fig"}).
{#F5}
Tumors were trypsinized, then, single cancer cells were harvested by cell sorting. Different from the various growth rates *in vivo*, cells expressing different EphA10s and EphA10 show the similar capability to form clones *in vitro* (Figure [5D](#F5){ref-type="fig"}). Tumor slides were subjected to immunohistochemistry in pECAD and β-catenin (Figure [5E](#F5){ref-type="fig"}). Stronger membrane-associate staining of β-catenin was observed in higher EphA10s expressing tumors, while less accumulation of pECAD was found in lower Epha10 expressing tumors. Additionally, recipients with tumors were housed for overall survival analysis (Figure [5F](#F5){ref-type="fig"}). Mice with EphA10^−^ tumors show a significantly longer survival period, which is even better in the presence of additional EphA10s expression (*P*=0.01, Log Rank, HR=0.34 \[95% CI: 0.065- 0.52\], Figure [5F](#F5){ref-type="fig"}).
Measurement of lymph vascular invasion and lymphatic metastasis was performed when tumor had the similar size (Figure [6A](#F6){ref-type="fig"}). EphA10s^+^/EphA10^−^ tumors were found a 0.58-fold decrease in the formation of peritumoral lymph vessels than the ones in EphA10s^−^/EphA10^+^ tumors (Figure [6B and 6C](#F6){ref-type="fig"}). Visible (enlarged) lymph nodes from superficial or para-aortic area were measured in number and weight (Figure [6D-6F](#F6){ref-type="fig"}). In total, lymph nodes were found a decrease in terms of number (up to 0.53 fold) and average weight (up to 0.47 fold) from recipients with tumors expressing higher EphA10s and lower EphA10. Metastasis in enlarged lymph nodes, identified by staining with an epithelial marker pan-cytokeratin (PCK), was subjected to statistical measurement in number and weight, respectively (Figure [6G-6I](#F6){ref-type="fig"}). In axillary lymph nodes (sentinel), a decrease of 0.17 fold in number and 0.21 fold in average weight was observed in compared with ones from EphA10^+^ tumors. In para-aortic lymph nodes, tumor metastases show to be shrunken to 0.63 fold in number and 0.58 fold in average weight than the ones with EphA10s^−^/EphA10^+^ metastases. More details are presented in [Supplementary Figure 6](#SD1){ref-type="supplementary-material"}.
{#F6}
Expression pattern of EphA10s and Epha10 linearizes the progression of breast cancer {#s2_10}
------------------------------------------------------------------------------------
Our findings *in vitro* suggested the interaction of EPHA10 isoforms with the complex of ECAD. Having the expression data of pECAD and β-catenin in 76 benign samples, 325 invasive samples, and 50 metastasis samples, which were matched with Eph receptor A10 staining and the expression of EphA10s and EphA10 in qualified 42 benign samples and 245 invasive samples, respectively, we generated Spearman\'s correlation coefficients (SSCs) analysis to illuminate the co-expression pattern, and annotated in subcellular distribution (Figure [7A-7C](#F7){ref-type="fig"}). In invasive samples, IHC staining shows a negative SCC with membrane-associate β-catenin and a positive value with membrane-associate pECAD. EphA10s positively associates with β-catenin in membrane in both benign and invasive samples, while EphA10 shows a positive value with membrane-associate staining of pECAD in invasive samples. The co-expression pattern suggested the interaction of EPHA10 isoforms with the complex of ECAD involved into BC progression.
{#F7}
Furthermore, we compiled the EPHA10 isoform profile, and employed principle component analysis (PCA) in the scenario of benign, invasive and metastatic samples (Figure [7D and 7E](#F7){ref-type="fig"}). The percentage of each PC lesion indicates its representative performance in diverse regions of a 3-dimensional version. The plots show that the profile of EPHA10 isoform properly describes and identifies the variousness between either benign and invasive samples, or invasive and metastatic samples.
The same profile and related expression profile of ECAD complex were used in Lasso regression analysis for a statistically potential of its prediction capability in the outcomes in 5-year follow-up ([Supplementary Figure 7](#SD1){ref-type="supplementary-material"}). The component is better promised when it is isolated earlier and/or valued with a larger number. In this model, Eph receptor A10 staining and EPHA10 isoforms show the strongest potential. Then, overall receiver operator characteristic (ROC) curves were employed to evaluate EPHA10-Lasso models (Figure [7F](#F7){ref-type="fig"}). In EPHA10 profile, ROC suggests a strong prognostic value of area under the curve (AUC) as 0.89 (95% CI: 0.86-0.93) in breast cancer prediction (red). When compared to the model built on the profile of ER, PR and HER2 signatures (dashed), our profile (blue) shows a better predictive efficiency as an AUC value of 0.72 (95% CI: 0.81-0.62) in tumor relapse (Figure [7F](#F7){ref-type="fig"}).
Furthermore, classical survival analysis was performed in the context of EphA10s and EphA10 expression (Figure [7G and 7H](#F7){ref-type="fig"}). Patients with higher EphA10s and lower EphA10 expressing demonstrate an advantage in analysis of non-disease survival (*P*=0.027, Log Rank, HR=0.26 \[95% CI: 0.18-0.65\]), and OS (*P=*0.001, Log Rank, HR=0.32 \[95% CI: 0.19-0.66\]) than ones with lower EphA10s and higher EphA10.
DISCUSSION {#s3}
==========
We annotated the expression of EPHA10 protein with its isoforms as down-regulated EphA10s and high-expressed EphA10 in BC progression, where the stability of the complex of ECAD and β-catenin is disrupted.
Similar to Eph receptor family member \[[@R31]\], EPHA10 isoforms are suggested to participate in the cross talk between cells and the microenvironment through Eph/ephrin interaction and tyrosine kinase activity ([Supplementary Figure 8](#SD1){ref-type="supplementary-material"}). Secretary EphA10s stabilizes membrane-associated β-catenin and weakens cytoplasmic β-catenin (benign cell). Down-regulation of EphA10s results in weakening of β-catenin recruitment in membrane (cell), and progressing of cancer behavior (invasive cell). Eph receptor A10 domain is suggested to bind with ephrin A5 in this context. Activated ephrin A5 is functionalized to recruit membrane-associated β-catenin in epithelial cells \[[@R32], [@R33]\]. Our data suggested that overexpression of EphA10s lose to redistribute β-catenin when lack of ephrin A5. Whether the interaction of Eph receptor A10 domain and ephrinA5 requires another component as EphA2 is in research. EphA10 is redistributed in cytoplasm where it promotes tumor invasion and metastasis (invasive cell). EphA10 has a pseudo-RTK domain that is in low phosphorylation activity when binding with ECAD and pECAD. It is assumed to be a silencer to protect substrate from being phosphorylation by blocking the signal transduction. Membrane-associate EphA10 binds with ECAD (benign cell) while cytoplasmic EphA10 binds with pECAD (invasive cell). Down-regulated EphA10 results in the loss of protection, and a decrease of ECAD from membrane (benign cell), while a decline of cytoplasmic pECAD (invasive cell).
The diagnoses and therapies of BC are still in challenges \[[@R34]\]. Here, we highlighted the need of identification of EphA10s and EphA10 status. By envisaging a scenario whereby the expression patterns of EphA10s and EphA10 were altered in patients, their outcome was found variously. Tracing EPHA10 isoforms result in a potentially satisfied prediction of tumor development and progression. We reported the participation of EphA10s in lymph-vessel formation \[[@R35]\], and an anti-tumor role in the utility of endogenous or extraneous EphA10s. Moreover, cytoplasmic EphA10 played a strong role in promoting invasion and metastasis of cancer. A reduction of EphA10 *in vivo* significantly weakened tumor dynamic, and leads to a promising outcome. The signatures of EphA10s and EphA10 are, therefore, promising contributions to the modern diagnosis and drug development.
MATERIALS AND METHODS {#s4}
=====================
Patients and TMAs {#s4_1}
-----------------
A total of 76 mammary gland cystadenoma samples and 275 invasive ductal carcinoma samples originating from patients who were treated at Tongji Hospital in Wuhan, China, from November 2008 to August 2009, were retrospectively enrolled, tumor-tissue microarrays (TMA: BR10010a) were purchased from [Alenabio.com](http://Alenabio.com) ([Supplementary Table 4](#SD1){ref-type="supplementary-material"}). Sections (5 μm) were cut from specimens with new scalpel for RNA extraction. A complete follow-up was available for 220 patients, ranging from 5 to 69 months (median 53 months, mean 60 months). This study was approved by the Ethics Committee of Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, PR China.
Cells and lentiviral transfection {#s4_2}
---------------------------------
MCF-10A and MDA-MB-231 cell lines were purchased from the American Type Culture Collection (USA) and cultured according to their instructions. All of cell lines were authenticated at the Shanghai Paternity Genetic Testing Center in April, 2013 using short tandem repeat (STR) DNA profiling (ABI 3130xl Genetic Analyzer, Life Technologies, USA). The plasmid (OHu08987) containing the EphA10s coding sequence (RefSeq Accession: NM_173641) was purchased from Genscript (USA). Lentivirus with mCherry label and carrying EphA10s plasmid or small hairpin RNAs of EphA10 were constructed and provided by CHENCHEM (China). Cell Infection was according to the manufacturer\'s instructions. Infected cells with mCherry signal were sorted using flow cytometer (Becton Dickinson, San Jose, CA).
Migration and invasion assay {#s4_3}
----------------------------
Assays were performed in Transwell chambers (Corning, NY) for 24 hours. Transwell filters were pre-coated with Matrigel (BD Biosciences) in the invasion assay. A total of 1-2×10^4^ cells were seeded in standard condition. Cells passing the filter were fixed and dyed with 0.1% crystal violet. The circle invasive assay (CIA) was performed similarly to invasion assay.
RNA extraction and qPCR {#s4_4}
-----------------------
Using a Tissue RNA FFPE Purification Kit (Promega, Madison, WI), total RNA of selected samples was extracted from qualified sections according to the manufacturer\'s protocol. RNA was purified on a Maxwell^®^ 16 instrument using a Total RNA Purification Kit (Promega, Madison, WI). An internal reference of GAPDH was using as previously described \[[@R36]\] (all primer sequences are in [Supplementary Table 5](#SD1){ref-type="supplementary-material"}).
Western blot and co-IP {#s4_5}
----------------------
Standard Western Blot analysis was performed as previously described \[[@R37]\]. Three independent experiments in a certain condition were subjected to Western Blot analysis. Supernatant protein was extracted using an Amicon Ultra Centrifugal filter (Millipore). Membrane protein was extracted using a Mem-PER™ Plus Membrane Protein Extraction Kit (Life Technologies). Immunoprecipitation (IP) was performed using the Pierce Crosslink IP Kit (Thermo, USA) according to the manufacturer\'s protocol.
Orthotropic human tumor xenografts {#s4_6}
----------------------------------
Female BALB/c nude mice (4 weeks old) were prepared based on previous studies \[[@R38], [@R39]\]. 2×10^5^ cells of MDA-MB-231 cells modified with different EphA10s and EphA10 expression patterns were injected under the 2nd breast fat pads. Mice were euthanized when the primary tumor grew to 8 mm×8 mm. Tumor masses and all visible lymph nodes were collected for further analyses.
Data analysis {#s4_7}
-------------
The SPSS statistical software package was used for all statistical analyses. For each parameter, SCCs were assessed to determine the co-expression among parameters. Disease-free and over-all survival was expressed as the number of months from surgery/injection to the occurrence of distant relapse or breast-related death by using Kaplan--Meier method and Log Rank test. In grouped comparisons, ANOVA was used followed by the least significant difference test for each group, and groups with different lowercase letters indicate significant differences. A P value of \<0.05 was considered statistically significant. Mathematical models of PCA and Lasso linear regression \[[@R40]\] were performed in R.
Extended Materials and Methods with the associated references are in the Supplemental Material.
SUPPLEMENTARY MATERIALS AND METHODS FIGURES AND TABLES {#s5}
======================================================
This work was supported by the National Development Program (973) for Key Basic Research of China (grants 2009CB521800 and 2013CB911304), the National Science Foundation of China (grants 81072135, 81372801, 30901749 and 81272426).
**CONFLICTS OF INTEREST**
The author declares that there are no potential conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Clinical characterisation of fundus albipunctatus {#Sec2}
-------------------------------------------------
Fundus albipunctatus (FA; MIM 136880) is a rare, hereditary, in most cases stationary, retinal disease, which is characterised by impaired night vision and numerous small, white-yellowish retinal lesions placed throughout the retina, except the fovea (Sergouniotis et al. [@CR50]). FA belongs to a heterogenous group of genetically determined flecked retina syndromes. The symptoms of these disorders include conditions characterised by multiple retinal yellowish-white lesions of various sizes and configuration, without vascular or optic nerve abnormalities. The group of flecked retina syndromes encompasses FA, retinitis punctata albescens, fundus flavimaculatus (Stargardt disease), familial drusen and fleck retina of Kandori, but far more diseases correspond to the rather vague definition of fleck retina syndromes (De Laey [@CR8]; Walia et al. [@CR59]). Moreover, there is a collection of diseases called white dots syndromes that can also be misdiagnosed with flecked retina syndromes. White dots syndromes are characterised by white lesions in the retinal pigment epithelium (RPE) or choroidal layers. The aetiology of these disorders is unknown, but these syndromes are suspected to be inflammatory in nature and can be associated with uveitis (Matsumoto et al. [@CR33]).
FA is a form of congenital stationary night blindness. The symptoms of defective dark adaptation may not be perceptible to the affected person. The optic nerve heads and the retinal vessels show no signs of irregularity. The visual field and visual acuity examinations of patients suffering from FA do not detect any abnormalities unless a dim stimulus is used. Dim stimulus causes a worsening of visual acuity and a constriction of the visual field. The scotopic electroretinography (ERG) responses are reduced after a 30--40-min period of dark adaptation, but typically normalise after prolonged dark adaptation (Yamamoto et al. [@CR62]; Sergouniotis et al. [@CR50]; Wang et al. [@CR61]). The photopic responses are usually normal if FA is not accompanied by macular dystrophy. Although long-term follow-up usually shows no progression in rods dysfunction in patients with this form of night blindness, some patients, especially the elderly, reveal progressive cone dystrophy (Nakamura et al. [@CR36], [@CR37]; Wada et al. [@CR57]; Niwa et al. [@CR39]). Full-field photopic electroretinograms of these individuals are usually severely reduced, a bull's eye maculopathy is often identified, and visual fields and acuity are impaired (Nakamura et al. [@CR36][@CR37]). Recently, it has been estimated that cone dysfunction can affect more than 30 % of patients with FA (Niwa et al. [@CR39]; Sergouniotis et al. [@CR50]; Pras et al. [@CR42]). Lidén and coworkers suggested that cone dystrophy may be either the result of impaired function of the RPE caused by a *RDH5* gene mutation or a direct consequence of a decreased supply of 11-*cis* retinal to the cones (Lidén et al. [@CR28]).
Genetic background {#Sec3}
------------------
FA shows an autosomal recessive inheritance pattern. In one family with this retinal disease, autosomal dominant or pseudodominant inheritance was suggested (Kranias et al. [@CR25]). FA is caused almost exclusively by mutations in the 11-*cis* retinol dehydrogenase 5 (*RDH5*) gene (Yamamoto et al. [@CR62]). However, mutations in two other genes, retinaldehyde binding protein 1 (*RLBP1*) and RPE-specific protein (*RPE65*), are also known to be associated with FA (Naz et al. [@CR38]; Schatz et al. [@CR48]). Retinaldehyde binding protein 1 is expressed in the RPE and Müller cells of the neuroretina, where it carries 11-*cis* retinol and 11-*cis* retinaldehyde as ligands (Sparkes et al. [@CR54]). Only a few *RLBP1* gene mutations in patients with FA have been reported to date (Katsanis et al. [@CR23]; Naz et al. [@CR38]). Katsanis end coworkers ([@CR23]) found a p.Arg150Gln mutation in the *RLBP1* gene in a consanguineous Saudi Arabian kindred with a retinal dystrophy phenotype that fulfilled the criteria of FA in younger individuals and retinitis punctata albescens in older patients. Two homozygous *RLBP1* gene mutations (p.Arg156\* and p.Gly116Arg) have also been identified in two unrelated, consanguineous Pakistani families suffering from FA (Naz et al. [@CR38]). RPE-specific protein (*RPE65*) is the isomerase of the visual cycle, catalysing the conversion of all-*trans* retinyl ester to 11-*cis* retinol (Moiseyev et al. [@CR34]). To date, mutations in the *RPE65* gene associated with FA have only been reported in one paper. A compound heterozygote of IVS + 5G \> A and c.344 T \> C mutations in the *RPE65* gene was found in a patient with FA (Schatz et al. [@CR48]).
Most cases of FA are caused by mutations in the *RDH5* gene (Gonzalez-Fernandez et al. [@CR13]). The *RDH5* gene encodes the enzyme that is a part of the visual cycle, the 11-*cis* retinol dehydrogenase (Simon et al. [@CR51]). The retinoid (visual) cycle is an enzyme pathway that occurs to regenerate the visual chromophore following light exposure (Travis et al. [@CR56]). 11-*cis* retinol dehydrogenase (*RDH5*) is predominantly expressed in the smooth endoplasmic reticulum of the RPE of the eye (Simon et al. [@CR52]). RPE cells play multiple roles essential for visual function, such as involvement in the uptake and metabolic processing of retinoids in the visual cycle (Simon et al. [@CR53]). *RDH5* has an important role in the molecular background of vision, as it catalyses the final step in the biosynthesis of 11-*cis* retinaldehyde, the universal chromophore of visual pigments (Simon et al. [@CR51]). Absorption of a photon by an opsin pigment causes isomerisation of the chromophore from 11-*cis* retinaldehyde to all-*trans* retinaldehyde. After entering the RPE cell, all-*trans* retinol is transferred into all-*trans* retinyl esters, which are isomerised by *RPE65* (RPE-specific) protein into 11-*cis* retinol esters (11-*cis* retinol). Then, 11-*cis* retinol is transported through the subretinal space, where it is oxidated by the *RDH5* enzyme into 11-*cis* retinal (Simon et al. [@CR51]; Wang et al. [@CR61]).
Retinol dehydrogenase 5 protein consists of 318 amino acids and is a member of the short-chain dehydrogenases/reductases (SDR) superfamily (Simon et al. [@CR52]). This family encompasses at least 57 varied, well-characterised enzymes (Jörnvall et al. [@CR21]), which catalyse the metabolism of steroids, fatty acids, carbohydrates, amino acids and aromatic compounds (Marchler-Bauer et al. [@CR32]). Although the amino acids sequence identity between the members of this large protein family is only at the 15--30 % level, there are two well-conserved regions within the enzymes' sequences: the motif Gly-X-X-X-Gly-X-Gly, consisting of three glycines within a cofactor binding site for NAD(H) or NADP(H), and the amino acid motif Tyr-X-X-X-Lys, with an invariant tyrosine residue inside the active site. Like all the classical SDR enzymes, the *RDH5* amino acid sequence contains these two conserved domains: the motif Gly-Cys-Asp-Ser-Gly-Phe-Gly at the amino acid residues 35--41 and the sequence encompassing the invariant tyrosine Tyr-Cys-Val-Ser-Lys at the residues 175--179 (Persson et al. [@CR41]; Jörnvall et al. [@CR21]; Simon et al. [@CR53]). Retinol dehydrogenase 5 protein is highly conserved among species (Simon et al. [@CR52]). The amino acid conservation of part of the active site (encompassing three conserved residues: Ser-163, Tyr-175, Lys-179) among *RDH5* of several species and three other short-chain dehydrogenases is shown in Fig. [1](#Fig1){ref-type="fig"}. *RDH5* is an integral membrane protein (Simon et al. [@CR51]). It is composed of the N-terminus (18 amino acids) located within the membrane, the ectodomain encompassing the active site, which is present in the lumen of the smooth endoplasmic reticulum (SER) (residues 19--288), the C-terminal membrane-spanning domain (289--310 amino acids) and the C-terminal tail (311--318 amino acids) located in the cytosol of the RPE (Simon et al. [@CR53]; Ajmal et al. [@CR3]). The different localisation of the *RDH5* domains within the RPE cell suggests that biosynthesis of 11-*cis* retinaldehyde is a compartmentalised process (Simon et al. [@CR53]).Fig. 1The amino acid conservation of part of the active site (residues 157--196 according to the numbering system of the *RDH5* protein) among *RDH5* of several species and three other short-chain dehydrogenases. The red frame indicates the invariant tyrosine, while the black frame indicates two other highly conserved residues involved in the catalytic mechanism: serine-163 and lysine-179. The abbreviation 'DH' in the names of three aligned proteins' sequences stands for 'dehydrogenase'
The *RDH5* gene is mapped on the chromosome 12q13-q14. The transcript (ENST00000257895) spans 1,269 bp and contains five exons, including four coding exons (2--5). The lengths of the coding exons are as follows: exon 2--342 bp (32 bp of the 5′ untranslated region and 310 bp translated sequence), exon 3--259, exon 4--164 and exon 5--384 bp (the translated part is 221 bp) (Simon et al. [@CR52]).
Materials and methods {#Sec5}
=====================
This study was conducted in accordance with the tenets of the Declaration of Helsinki. A 16-year-old female patient of Polish origin with clinical signs of night blindness was examined. The patient underwent colour vision testing, fundus photography, automated visual field testing (Humphrey--Zeiss), full-field ERG and spectral optical coherent tomography (SOCT). The electrophysiological examinations included a full-field ERG protocol with the standard scotopic 20 min dark adaptation and extended protocol with prolonged 120 min of dark adaptation. The standard photopic ERG (30 Hz white flicker stimulation) was performed after 10 min of light adaptation.
Voluntary informed consent for genetic examination was obtained not only from the mother of the patient (as the patient was underaged), but also from both parents and two sisters, who had their blood taken for segregation analysis for the presence of the novel mutation. Genomic DNA was extracted from peripheral blood using the conventional salting-out procedure. The coding regions of the *RDH5* gene (exons 2--5) were amplified and sequenced to screen for disease-causing mutations in the patient. Eight primer pairs (including three pairs for amplifying exon 2, two for exons 3 and 5, and one pair for exon 4) were used following a previous report (Yamamoto et al. [@CR62]). A fragment of exon 3 (primer pair designed 3b) was also amplified in both parents and two sisters of the proband. The polymerase chain reaction (PCR) products were purified with the use of ExoSAP-IT (Exonuclease I and Shrimp Alkaline Phosphatase Cleanup for PCR products, Affymetrix) and directly sequenced using Dye Terminator chemistry (v3.1 BigDye® Terminator, Life Technologies). The sequencing products were separated on an ABI 3130xl capillary sequencer (Applied Biosystems). The obtained sequences were verified by comparing them to the reference sequence of the *RDH5* gene (GenBank NM_001199771.1) and screened for mutations. The in silico analysis using PROVEAN (Choi et al. [@CR6]), SIFT (Kumar et al. [@CR26]) and PolyPhen-2 (Adzhubei et al. [@CR2]) software was performed to assess the possible functional effect of the novel missense mutation.
Results {#Sec6}
=======
Family history {#Sec7}
--------------
The patient has three sisters. The younger sister suffers from astigmatism, while the mother and the two older sisters had no ophthalmologic problems. In both the proband's father's eyes, presenile cataract was revealed at the age of 38 years. He suffered from retinal detachment in the right eye later on. The first pregnancy of the patient's mother ended with a stillbirth (pedigree, Fig. [2](#Fig2){ref-type="fig"}).Fig. 2Pedigree and genotypes at the *RDH5* gene nucleotide position 524 of the family with the c.524A \> T mutation. The mutation is marked with red 'M' letter, while the blue '+' symbol indicates a wild-type allele. The parents and two sisters of the proband were involved in the exon 3 sequencing analysis
Clinical status {#Sec8}
---------------
Visual acuity and colour vision in the patient were normal. The examination of the anterior segment and the pupillary reflexes showed no abnormalities. An eye fundus examination revealed numerous small white-yellowish retinal lesions mainly in the upper quadrants of the retina (Fig. [3](#Fig3){ref-type="fig"}). Examination of the visual field revealed its peripheral constriction to approximately 10--20°. A full-field electroretinogram showed significantly reduced scotopic responses after the standard period of 20--30 min dark adaptation (Fig. [4](#Fig4){ref-type="fig"}a). However, after a prolonged 120 min of dark adaptation, rod responses normalised (Fig. [4](#Fig4){ref-type="fig"}b). Photopic responses in all examinations were normal. The high-definition SOCT showed no abnormalities of the central macular thickness in either eye, but local modulations of the RPE and IS/OS (inner segment--outer segment) junctions corresponding with retinal flecks were identified.Fig. 3Fundus examination. **a** The right eye of a healthy individual, **b** The right eye of the patient with the c.524A \> T mutation in the *RDH5* gene. Numerous small, white-yellowish retinal lesions are located in the upper segments of the retinaFig. 4Comparison of scotopic responses after 30 and 120 min of dark adaptation: *RE*, right eye; *LE*, left eye. **a** The reduction of scotopic responses (DA 0.009 cdxs/m^2^) (b-wave amplitude RE: 26.68 μV, LE: 26.53 μV, normal 260 ± 151.4 μV), standard electroretinography (ERG) response (3 cdxs/m^2^) on the borderline after 30 min of dark adaptation. **b** The normalisation of scotopic responses after 120 min of scotopic adaptation (b-wave amplitude RE: 259.1 μV, LE: 378.2 μV)
Genetic analysis {#Sec9}
----------------
As the clinical findings, especially characteristic eye fundus appearance, indicated a suspicion of FA, searching for a mutation in the *RDH5* gene seemed to be the most appropriate strategy. Bidirectional sequencing of the *RDH5* gene coding region (exons 2--5) revealed a homozygous mutation c.524A \> T in exon 3. This transversion changes codon UAC to UUC, which results in the substitution of polar tyrosine to non-polar phenylalanine at amino acid position 175 (p.Tyr175Phe) (Fig. [5](#Fig5){ref-type="fig"}). The in silico analysis of the predicted influence of the p.Tyr175Phe substitution on protein function with the use of PROVEAN software (the tolerance index score was −3.900), as well as SIFT software (tolerance index score 0.00), revealed that this amino acid change is deleterious. The in silico analysis using PolyPhen-2 software predicted the mutation to be probably damaging (score of 1). The c.524A \> T variant was not found in a control cohort annotated in the Exome Variant Server (EVS) database (Exome Variant Server [@CR11]) nor in the 1000 Genomes Project database (1000 Genomes Project Consortium [@CR1]). The segregation analysis of the mutation in the proband's family revealed that both parents and one of the proband's sisters are heterozygous carriers of the c.524A \> T substitution (pedigree, Fig. [2](#Fig2){ref-type="fig"}).Fig. 5A chromatogram showing the c.524A \> T mutation in the *RDH5* gene. **a** The wild-type nucleotide sequence and the wild-type protein sequence. The orange frame indicates the most conservative element between short-chain alcohol dehydrogenases, located within the active site of the enzyme. Invariant tyrosine is labelled blue and indicated with the red frame. **b** The nucleotide sequence of the heterozygous parent. **c** The sequence of the patient with c.524A \> T mutation and truncated protein sequence
Mutations in the *RDH5* gene: review of the literature {#Sec10}
------------------------------------------------------
Since 1999, when Yamamoto et al. described the mutations in the *RDH5* gene in two unrelated patients suffering from FA (Yamamoto et al. [@CR62]), there have been reports of missense, in-frame and frameshift mutations (Sergouniotis et al. [@CR50]). To date, more than 40 mutations in the *RDH5* gene have been reported, and most of them are missense variants (Nakamura et al. [@CR36]; Driessen et al. [@CR10]; Sergouniotis et al. [@CR50]; Ajmal et al. [@CR3]; Wang et al. [@CR61]; Waldron and Medefindt [@CR58]; Stenson et al. [@CR55]). The list of *RDH5* gene mutations identified in patients with FA together with the regions of the protein affected by these changes is shown in Table [1](#Tab1){ref-type="table"}. The *RDH5* gene mutations have been identified mostly in homozygotic or compound heterozygotic forms, but a few variants (p.Arg19Gly, p.Arg191Gln and p.Arg278Gln) have been found as single heterozygous mutations (Sergouniotis et al. [@CR50]; Pras et al. [@CR42]). There seems to be no hot spot in the gene, as the reported mutations are distributed across the entire *RDH5* coding sequence. The mutations identified affect the entire protein, except the N-terminus, with most mutations located in the longest domain: lumenal ectodomain (see Table [1](#Tab1){ref-type="table"}). The analyses of biochemical defects in *RDH5* mutants associated with FA revealed that all the mutations tested (marked with '+' in Table [1](#Tab1){ref-type="table"}) affect the stability and expression level of the protein and result in subcellular mislocalisation. Moreover, loss of enzymatic activity in vitro and in vivo has been observed for almost all the constructed mutants (except the mutant with amino acid change p.Ala294Pro located in the C-terminal domain) (Yamamoto et al. [@CR62]; Cideciyan et al. [@CR7]; Lidén et al. [@CR28]). Even in the absence of the enzyme activity caused by many *RDH5* mutations, night vision regenerates after prolonged dark adaptation in patients with FA. This fact can be explained by the results of the studies on Rdh knockout mice models (Driessen et al. [@CR9]; Kim et al. [@CR24]). Driessen and coworkers revealed that transgenic mice missing the *RDH5* gene display delayed dark adaptation, but only at a very high bleach level. The studies on Rdh5 and Rdh11 knockout mice revealed that one more enzyme, RDH11, appeared to have an important role in regenerating the chromophore. These results indicate that both RDH5 and RDH11 contribute to 11-*cis* retinal production (Driessen et al. [@CR9]).Table 1*RDH5* mutations identified in patients with fundus albipunctatus (FA)Exon/intronNucleotide positionAmino acid residueRegion of the proteinMutants analysis^a^ReferenceExon 2c.55A \> Gp.Arg19GlyEctodomain−Sergouniotis et al. ([@CR50])Exon 2c.71_74delTGCCp.Leu24Profs\*36Ectodomain−Pras et al. ([@CR42])Exon 2c.95delTp.Phe32Serfs\*29Ectodomain−Schatz et al. ([@CR47])Exon 2c.98 T \> Cp.Ile33ThrEctodomain−Sergouniotis et al. ([@CR50])Exon 2c.98 T \> Ap.Ile33AsnEctodomain−Rüther et al. ([@CR45])Exon 2c.103G \> Ap.Gly35SerEctodomain, the conserved cofactor binding motif+Nakamura et al. ([@CR36]); Wada et al. ([@CR57])Exon 2c.124C \> Tp.Arg42CysEctodomain−Niwa et al. ([@CR39])Exon 2c.129delTp.Leu44Trpfs\*17Ectodomain−Driessen et al. ([@CR10])Exon 2c.160C \> Tp.Arg54\*Ectodomain−Pras et al. ([@CR42])Exon 2c.175 T \> Ap.Cys59SerEctodomain−Wang et al. ([@CR61])Exon 2c.214insGTGGp.Val71fs\*86Ectodomain−Driessen et al. ([@CR10])Exon 2c.218C \> Tp.Ser73PheEctodomain+Yamamoto et al. ([@CR62])Exon 2c.285G \> Ap.Trp95\*Ectodomain−Wang et al. ([@CR61])Intron 2c.310 + 1G \> A--Ectodomain−Sergouniotis et al. ([@CR50])Exon 3c.319G \> Cp.Gly107ArgEctodomain−Nakamura et al. ([@CR36]); Sato et al. ([@CR46]); Hotta et al. ([@CR18])Exon 3c.346G \> Cp.Gly116ArgEctodomain−Sergouniotis et al. ([@CR50])Exon 3c.346_347insGCAp.Gly116_Ile117insSerEctodomain−Sergouniotis et al. ([@CR50])Exon 3c.382G \> Ap.Asp128AsnEctodomain+Iannaccone et al. ([@CR19]); Schatz et al. ([@CR47]); Pras et al. ([@CR42])Exon 3c.394 G \> Ap.Val132MetEctodomain−Nakamura et al. ([@CR36])Exon 3c.416G \> Tp.Gly139ValEctodomain−Sergouniotis et al. ([@CR50])Exon 3c.469C \> Tp.Arg157TrpEctodomain+Cideciyan et al. ([@CR7])Exon 3c.470G \> Ap.Arg157GlnEctodomain−Hajali et al. ([@CR15]); Sergouniotis et al. ([@CR50])Exon 3c.490G \> Tp.Val164PheEctodomain−Yamamoto et al. ([@CR63])Exon 3c.500G \> Ap.Arg167HisEctodomain−Sekiya et al. ([@CR49])Exon 3c.524A \> Tp.Tyr175PheEctodomain, the conserved motif within the active site, invariant tyrosine−This studyExon 3c.530 T \> Gp.Val177GlyEctodomain, the conserved motif within active site−Kuroiwa et al. ([@CR27])Exon 4c.572G \> Ap.Arg191GlnEctodomain−Pras et al. ([@CR42])Exon 4c.625C \> Tp.Arg209\*Ectodomain−Schatz et al. ([@CR47])Exon 4c.689_690delCTinsGGp.Pro230ArgEctodomain−Wang et al. ([@CR60])Exon 4c.710A \> Cp.Tyr237SerEctodomainSergouniotis et al. ([@CR50])Exon 4c.712G \> Tp.Gly238TrpEctodomain+Yamamoto et al. ([@CR62]); Gonzalez-Fernandez et al. ([@CR13]); Hajali et al. ([@CR15]); Iannaccone et al. ([@CR19])Exon 4c.718dupGp.Ala240Glyfs\*19Ectodomain−Nakamura et al. ([@CR36])Exon 4c.718delGp.Ala240Profs\*7Ectodomain−Makiyama et al. ([@CR31])Exon 5c.758 T \> Gp.Met253ArgEctodomain−Ajmal et al. ([@CR3])Exon 5c.791 T \> Gp.Val264GlyEctodomain+Hirose et al. ([@CR17])Exon 5c.801C \> Gp.Cys267TrpEctodomain−Driessen et al. ([@CR10])Exon 5c.824_825delGAp.Arg275Profs\*60Ectodomain−Sergouniotis et al. ([@CR50])Exon 5c.832C \> Tp.Arg278\*Ectodomain−Liu et al. ([@CR29])Exon 5c.833G \> Ap.Arg278GlnEctodomain−Pras et al. ([@CR42])Exon 5c.839G \> Ap.Arg280HisEctodomain+Gonzalez-Fernandez et al. ([@CR13]); Nakamura et al. ([@CR36]); Kuroiwa et al. ([@CR27]); Sato et al. ([@CR46])Exon 5c.841 T \> Cp.Tyr281HisEctodomain−Nakamura et al. ([@CR36]); Nakamura and Miyake ([@CR35])Exon 5c.880G \> Cp.Ala294ProC-terminal transmembrane domain+Gonzalez-Fernandez et al. ([@CR13]); Schatz et al. ([@CR47])Exon 5c.913_917delGTGCTp.Val305Hisfs\*29C-terminal transmembrane domain−Ajmal et al. ([@CR3])Exon 5c.928delCinsGAAGp.Leu310GluValC-terminal transmembrane domain+Nakamura et al. ([@CR36]); Nakamura and Miyake ([@CR35]); Sato et al. ([@CR46]); Wang et al. ([@CR60]); Liu et al. ([@CR29]); Makiyama et al. ([@CR31])Exon 5c.955 T \> Cp.\*319Argext\*32C-terminal cytosolic tail−Sergouniotis et al. ([@CR50])^a^'+' indicates that the mutants were constructed for this mutation and the analysis of biochemical defects was performed (Yamamoto et al. [@CR62]; Cideciyan et al. [@CR7]; Lidén et al. [@CR28]); '--' indicates that the mutants analysis have not been reported
High variability of the disease's phenotype is observed among patients with FA carrying *RDH5* mutations. They show a variable visual acuity and variation in the density of white flecks (from minimal white dots or even normal fundus to numerous larger coalescent spots) (Sergouniotis et al. [@CR50]; Ajmal et al. [@CR3]). Despite the observed phenotypic variability, the presence of white dots appeared to be a common feature in patients with FA. These retinal flecks are hypothesised to be the effect of an accumulation of toxic retinyl esters in the RPE as the result of 11-*cis* retinol dehydrogenase disruption (Driessen et al. [@CR9]). However, it is known that, with increasing age in the patients with FA or after uveitis, the dots may fade and become smaller and discrete, especially in the far periphery of the fundus (Yamamoto et al. [@CR63]; Imaizumi et al. [@CR20]; Sergouniotis et al. [@CR50]). Patients with mutations in the *RDH5* gene can manifest a non-progressive or progressive form of the disease. It has been reported that individuals with or without cone dystrophy also presented varying degrees of severity of FA (Nakamura et al. [@CR36], [@CR37]; Sergouniotis et al. [@CR50]; Ajmal et al. [@CR3]). Moreover, different phenotypes have been observed in patients with the same mutation, for example, c.928delCinsGAAG (p.Leu310GluVal), which is the most commonly identified *RDH5* gene mutation (Nakamura et al. [@CR36], [@CR37]; Nakamura and Miyake [@CR35]; Sato et al. [@CR46]; Pras et al. [@CR42]; Ajmal et al. [@CR3]).
Therefore, based on the complete review of the literature, it is difficult to establish any valid correlation between the *RDH5* variants and the disease. There is no significant association between the localisation or the type of *RDH5* mutation with the severity of the disease phenotype (including electrophysiological observations or the presence/absence of cone dystrophy) (Sato et al. [@CR46]; Niwa et al. [@CR39]; Sergouniotis et al. [@CR50]; Pras et al. [@CR42]).
Discussion {#Sec11}
==========
The differential diagnosis of flecked retina/white dots syndromes can be difficult using routine ophthalmological examination. In cases of FA with progressive cone dystrophy, signs and symptoms may be non-specific and lead to misdiagnosis. Small white-yellow retinal lesions could indicate the diagnosis of fundus flavimaculatus, familial dominant drusen or retinitis punctata albescens (De Laey [@CR8]; Walia et al. [@CR59]). Moreover, phenotypic variability in the fundus appearance of patients with FA has been described (Sergouniotis et al. [@CR50]; Ajmal et al. [@CR3]). Electrophysiological findings, together with the appropriate genetic analysis, appear to be crucial tools in the differential diagnosis of FA (Pras et al. [@CR42]). Although decreased scotopic ERG responses could appear in many different conditions (retinitis punctata albescens, FA, FA with progressive cone dystrophy and Stargardt disease), their normalisation after 120 min of dark adaptation is observed mostly in FA (Table [2](#Tab2){ref-type="table"}) (Yamamoto et al. [@CR62]; Kanski [@CR23]). However, retinitis punctata albescens due to *RLBP1* mutation (Bothnia dystrophy) may be more difficult to distinguish, as in the early stages, there is phenotypic overlap with FA. Some patients with Bothnia dystrophy show a dramatic improvement in electroretinograms after prolonged dark adaptation (Burstedt et al. [@CR5]; Gränse et al. [@CR14]), while some patients with FA may present a minimal change, even after several hours of dark adaptation (Sergouniotis et al. [@CR50]).Table 2Comparison of conditions with the symptom of small white-yellow retinal lesionsFundus albipunctatusFundus albipunctatus with progressive cone dystrophyRetinitis punctata albescensFundus flavimaculatus (Stargardt disease)Our patientEye fundusNumerous small white-yellow retinal lesionsNumerous small white-yellow retinal lesionsNumerous small white-yellow retinal lesionsNumerous fleck-like yellow retinal lesionsNumerous small white-yellow retinal lesionsVisual fieldNormalCan be constrictedConstrictedCan be constrictedPeripherally constrictedRetinal vesselsNormalNormalAttenuatedCan be attenuatedNormalERGDepressed rods responsesDepressed rods responsesDepressed rods responsesVariableDepressed rods responsesERG after prolonged dark adaptationProper scotopic responsesProper scotopic responsesDepressed scotopic responseVariableProper scotopic responsesmfERGNormalReduced cones densityNormalDecreased centralReduced cones density in peripheral rings
Optimistically, recent studies provide hope for the successful treatment of patients diagnosed with FA. Studies on mouse models of FA demonstrated a significant improvement in rod and cone visual function after treatment with 9-*cis* retinal (Maeda et al. [@CR30]). Moreover, the latest pilot clinical testing on a group of patients with FA revealed that treatment with 9-*cis*-β-carotene as a food supplement led to a considerable visual improvement. It is very promising, as there has been no reported treatment resulting in a significant improvement in the visual functions in patients with retinal dystrophy to date and, what is more, this approach will also be helpful for some patients with retinitis pigmentosa (Rotenstreich et al. [@CR43], [@CR44]).
Genetic analysis of the *RDH5* gene (exons 2--5) in our patient revealed a novel, homozygous mutation c.524A \> T in exon 3. The change of the chemical properties of the substituted amino acids and the mutation's predicted influence on the protein function indicate that the p.Tyr175Phe mutation is probably pathogenic. Moreover, tyrosine at position 175 of the *RDH5* protein is localised within the active site of the enzyme, and was described as invariant tyrosine (Simon et al. [@CR52]). It is known that invariant tyrosines are found in all short-chain alcohol dehydrogenases. The Tyr-X-X-X-Lys sequence motif, a part of the substrate binding (active) site, is the most conserved element in SDRs (Persson et al. [@CR41]; Jörnvall et al. [@CR21]). To date, only one mutation in this highly conserved motif of the human *RDH5* enzyme (Tyr-Cys-Val-Ser-Lys) has been identified. It was a substitution of valine to glycine at amino acid position 177. This variant was found in a boy with FA, who was a compound heterozygote of p.Val177Gly and p.Arg280His (Kuroiwa et al. [@CR27]), but also in a boy diagnosed with familial fleck retina with night blindness (a heterozygote of p.Val177Gly and p.Leu310GluVal) (Hayashi et al. [@CR16]).
The invariant tyrosine, together with the lysine at position 179 (of the human *RDH5*) within the conserved motif and serine-163 (shown in Fig. [1](#Fig1){ref-type="fig"}), are involved in the catalytic mechanism (as putative active site residues), but only tyrosine located within this sequence is rigidly conserved in the SDR superfamily (Jörnvall et al. [@CR21]; Filling et al. [@CR12]; Oppermann et al. [@CR40]). The role of the invariant tyrosine was analysed in the most studied member of the SDR superfamily: Drosophila alcohol dehydrogenase (ADH). Albalat and González-Duarte ([@CR4]) constructed a Drosophila alcohol dehydrogenase, in which the invariant tyrosine (at amino acid position 152) was substituted by phenylalanine. Drosophila alcohol dehydrogenase-phenylalanine-152 revealed no enzymatic activity. Therefore, it is very likely that substitution of the invariant tyrosine to phenylalanine in human retinol dehydrogenase (*RDH5*) protein would have a similarly damaging effect to that reported in Drosophila ADH.
The segregation analysis of the presence of the c.524A \> T mutation in the family studied was found to be consistent with the autosomal recessive mode of inheritance. It revealed that both the proband's parents are heterozygous carriers of this novel substitution. Therefore, it is highly probable that they are related. We did not confirm this assumption based on the exact pedigree data, but the parents' families come from villages located in very close proximity.
To conclude, we have presented a brief but complete review of the literature on FA, focusing on the genetic background of the disease. Our study expands the spectrum of *RDH5* mutations, as we also report the novel mutation in the 11-*cis* retinol dehydrogenase 5 gene. This study is the first report of a *RDH5* gene mutation that affects the invariant tyrosine, one of the most conserved amino acid residues in SDRs, crucial for these enzymes' activity. The location of the substitution, together with the mutation's predicted influence on protein function, indicate that the p.Tyr175Phe mutation is probably pathogenic and can be recognised as the cause of FA. Moreover, we have presented the first molecular evidence for 11-*cis* retinol dehydrogenase 5 (*RDH5*) gene mutation in a Polish patient with this rare retinal disease. This study may also help clinicians to improve the difficult process of FA differential diagnosis, in which genetic analysis is an indispensable element, which would enable the correct treatment of patients.
This study was partially supported by a grant from the Polish Ministry of Science and Higher Education (806/N-NIEMCY/2010/0).
Conflict of interest {#FPar1}
====================
None.
[^1]: Communicated by: Michal Witt
| {
"pile_set_name": "PubMed Central"
} |
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus.
Introduction
============
Atypical hemolytic uremic syndrome (aHUS) is an extremely rare etiology of microangiopathic hemolytic anemia (MAHA). In common, disorders with MAHA affect endothelial cells of small vessels and cause mechanical destructions of red blood cells. Other MAHA etiologies include hemolytic uremic syndrome, disseminated intravascular coagulations and thrombotic thrombocytopenic purpura (TTP). aHUS is 10 times less common than HUS, and only two cases of aHUS per million persons are reported each year in the United States \[[@REF1],[@REF2]\].
Unlike other etiologies of MAHA that develop in vascular system, aHUS is primarily a renal disorder. aHUS involves genetic defects of regulator genes of the alternative complement pathway (AP) and results in the overactivation of the alternative pathway of the complement system in the kidneys. Together, direct injuries from membrane attack complex of the complement (MAC) and injuries from endothelial dysfunction cause microvascular thrombosis, ischemia, glomerular dysfunction and acute renal failure \[[@REF2]\]. Extrarenal manifestations of aHUS are often due to an increased vascular permeability in organs secondary to the circulating anaphylatoxins C3a and C5a produced in the kidneys \[[@REF3]\].
aHUS can be familial or sporadic. Up to 20% of aHUS cases are familial. Familial aHUS has a poor prognosis with a rate of end-stage renal disease or death of 50% to 90%. Half cases of sporadic aHUS are idiopathic, the other half cases of aHUS patients have an identifiable trigger such as a pregnancy, an infection with human immunodeficiency virus, an organ transplant, or a malignancy. These events are considered to enhance AP activation \[[@REF4]-[@REF7]\].
Case presentation
=================
A 24-year-old G2 P0 patient with history of chronic hypertension and chronic unspecified kidney disease diagnosed in infancy and followed by a nephrologist was admitted for one-day history of epigastric and right upper quadrant pain. The pain was sharp, intermittent with an intensity of 9/10. On admission day, the pain increased two hours after eating spicy tacos. She denies any exacerbating or alleviating factor. The patient also vomited six times after eating.
The patient also reported painful uterine contraction. On physical exam, the patient was hemodynamically stable. The vaginal exam revealed 3--4 cm dilated and effaced cervix.
On admission, laboratory results were consistent with microangiopathic hemolytic anemia (severe thrombocytopenia, anemia, worsening kidney function) (Table [1](#TAB1){ref-type="table"}).
###### Laboratory results on admission.
WBC: White blood cell; RBC: Red blood cell count; MCV: Mean corpuscular volume; MCH: Mean corpuscular hemoglobin; MCHC: Mean corpuscular hemoglobin concentration; RDW: Red blood cell distribution width; MPV: Mean platelet volume; PT: Prothrombin time; INR: International normalized ratio; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; A/G: Albumin/globulin ratio; GFR: Glomerular filtration rate.
---------------------------------------- -------------------------- ---------------------- ------
Cell Blood Count Complete Metabolic Panel
WBC 13.5 Sodium 133
Hemoglobin 9.4 Potassium 5.1
Hematocrit 27.4 Chloride 111
Platelet count 52 CO2 11
RBC 3.09 Urea Nitrogen 88
MCV 88.6 Creatinine 5.58
MCH 30.5 Glucose 96
MCHC 34.4 Est GFR 9
RDW 12.3 Anion Gap 11
MPV 8.9 Calcium 7.8
Coagulation Studies Protein, Total 5.5
PT 13.9 Albumin 2.9
INR 1.1 A/G ratio 1.1
Antiphospholipid Antibody-Panel-Status Bilirubin, Total 0.4
Antiphospholipid IgG 6 Alkaline Phosphatase 82
Antiphospholipid IgM 5 AST 209
ALT 149
---------------------------------------- -------------------------- ---------------------- ------
The delivery team was consulted for dilation and evacuation for a possible pre-eclampsia/HELLP syndrome. Failure of thrombocytopenia to improve after dilation and evacuation led to consider TTP and aHUS. TTP is treated with plasma exchange, and aHUS is treated with anti-complement therapy. TTP was considered and treated presumptively given the severity of its complication: intracranial bleeding. Creatinine level was monitored to assess the response to treatment. A lack of improvement would change the presumptive diagnosis of TTP to aHUS. ADAMTS 13 level was sent prior to any transfusion. ADAMTS level was low at 44% favoring aHUS. The patient entered in remission when she was treated with Eculizumab.
Imaging and genetic tests
Chest X-ray showed basilar opacities, likely moderate pleural effusions (Figure [1](#FIG1){ref-type="fig"}).
{#FIG1}
The abdominal sonogram demonstrated an atrophic right kidney with increased echogenicity, slightly prominent tortuous veins at the splenic hilum which may represent focal perisplenic varices.
Genetic testing
DNA sequencing and multiplex ligation-dependent probe amplification (MLPA) analysis did not identify any significant variant of the genes tested: CPH, CFI, MCP, THBD, VFB, C3, DGKE, ADAMTS 13, C4BPA, C4BPB, LMNA, CFTHR1, CFHR3. Genetic variants identify only 50--60% of aHUS cases.
Discussion
==========
Genetics of the complement system regulation: the basis of aHUS pathogenesis, incomplete penetrance, and prognosis
The activation of the AP of the complement system is tightly regulated, and defects of the AP genes and regulatory genes are involved in the development of aHUS. The spontaneous hydrolysis of C3 into C3H20 in the plasma initiates the AP. C3H20 generates C3b that binds to the surface of plasma membranes. On the surface of plasma membranes, C3b interacts with Factor B. Factor D will later on cleave Factor B and conclude the formation of the C3 convertase of the AP (C3bBb). C3 convertase cleaves C3 into C3a and C3b. The newly generated C3b can interact with Factor B and create an amplification loop that will activate the AP until all the complement components are consumed. Complement factor I (CFI) inactivates C3b into C3bi and prevents C3b interaction with Factor B. The reaction of CFI is possible only if the Complement Factor H (CFH) is bound to both C3b and the host cell, and the co-factor membrane complement protein (MCP) is functional. CFH has a discriminative property. It can bind only to host cells and not to pathogens. CFH distinguishes host cells from pathogens and prevents host cells from undergoing complement-mediated cells lysis. Thrombomodulin (TM) enhances the activity of CFI. Complement regulatory proteins prevent the consumption of all complement substrates and restrict the activation of the complement system on the surface of microorganisms.
Loss of function mutations of regulatory genes
CFH, MCP, TM or CFI and increased function mutations of C3b and Factor B have been all identified in the pathogenesis of HUS \[[@REF4],[@REF7]\]. About 6 to 10% of aHUS patients have an anti-CFH antibody that prevents a proper binding of CFH to C3b. Interestingly, anti-CFH antibodies develop in young children who are homozygous for the deletion of CFHR1 and CFHR3 genes (Figure [2](#FIG2){ref-type="fig"}) \[[@REF8]\].
{#FIG2}
The alteration of a single gene alone cannot cause aHUS. In a study where patients and their relatives were screened for mutation and polymorphism for CFH, MCP, and CFI genes, patients did not develop aHUS unless they had also a combination of mutations, haplotypes and single nucleotides polymorphism (SNP). In a patient with MCP mutation, for example, MCPggaac haplotype increases two-fold the risk of aHUS compared with control. MCPggaac haplotype contains two SNPs in the MCP promoter which decreases the transcription of MCP \[[@REF9],[@REF10]\].
The genetic origin of aHUS also explains the incomplete penetrance. For mutation involving CFH, CFI, MCP, and CFB, the penetrance is ∼50%. Even when many altered genes are associated with the same patient, aHUS develops only after the occurrence of certain environmental factors. Pregnancy, HIV, and oral contraceptive pills have been reported to trigger aHUS in genetically predisposed patients \[[@REF4]\].
Genetic predispositions also determine the prognosis after the first episode and after renal transplantation. MCP patients have a better prognosis with a complete remission rate of 80 to 90%. CFH, CFI, C3 mutation and anti-CFH antibody are associated with poor outcome in the first episode. Renal transplantation is favorable for patients with MCP mutations as well, whereas it has a poor success rate for patients with CFI and CFH mutations \[[@REF4],[@REF11],[@REF12]\].
Conclusions
===========
aHUS is a disorder of the alternative complement pathway. Genetic defects of components that regulate the AP lead to unopposed activation of AP in the kidneys. Local endothelial cells lesions and MAC formation, and circulatory C3a and C5a explain renal and extrarenal manifestations, respectively. Clinically, aHUS may be challenging to distinguish from pre-eclampsia in a pregnant patient because they both share thrombocytopenia, and a pregnancy can trigger both diseases. It is not clear whether chronic kidney disease in the patient increased the likelihood of pre-eclampsia-like presentation. The initial management should favor urgency and consider pre-eclampsia. The lack of response to the pre-eclampsia treatment and the ADAMST 13 level will readjust the presumptive diagnosis to either HUS or aHUS. We report the first case of aHUS presenting as pre-eclampsia in a pregnant woman with severe kidney disease.
The authors have declared that no competing interests exist.
Consent was obtained by all participants in this study
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
As the impacts of human activities on our planet continue to increase^[@CR1]^, all ecosystems are profoundly altered. Aquatic systems, particularly rivers, are recognized as being highly degraded in terms of both species diversity and ecological functioning^[@CR2]^. The rate of decline in aquatic populations is almost twice that of terrestrial populations^[@CR3]^. Facing this dramatic development, increasing the scale and frequency of aquatic biodiversity observations is a priority^[@CR4]^. Among other bioindicators, fish are substantially valuable to water managers and the public. However, due to sampling difficulties, the probability of detecting rare species is low^[@CR5]^, and quantitatively monitoring fish assemblages remains a hard and costly task, particularly in large rivers^[@CR6]^.
Since its first application to macroorganisms^[@CR7]^, environmental DNA (eDNA) has increasingly appeared to be a promising non-invasive method for improving aquatic biodiversity monitoring^[@CR8]^. eDNA refers to DNA obtained from environmental samples without the prior isolation of any target organism^[@CR9]^. In the case of water samples, eDNA contains both intra-organism DNA (e.g., small planktonic organisms) and extra-organism DNA (e.g., from fish) which can be cellular or extracellular and degraded^[@CR9]^. While DNA can persist for long periods in dry, cold conditions and in the absence of light^[@CR7]^, eDNA persistence in water and normal temperature conditions varies from a few days to a few weeks^[@CR10]--[@CR12]^. With the emergence of next-generation sequencing (NGS) platforms and the use of universal PCR primers (eDNA metabarcoding), large collections of taxa can be identified via a single experiment^[@CR13]^. This not only offers the possibility to detect rare or evasive species^[@CR8]^ without a priori but also allows the rapid biodiversity assessment of large communities and the reconstruction of ecological and evolutionary processes from easy-to-collect samples^[@CR14]^.
Previous studies have shown that this approach is effective for inventorying fish and amphibian species in mesocosms^[@CR15],[@CR16]^, estuarine and marine systems^[@CR17]--[@CR19]^, lakes^[@CR20],[@CR21]^, rivers and ponds^[@CR22]--[@CR24]^. However, like other inventory methods, eDNA metabarcoding has its own biases. False-negative detections can occur, which are mainly due to failure of the method (low marker sensitivity, low eDNA quantity, ineffective sample preservation^[@CR25]^, PCR inhibitors^[@CR26]^ and/or insufficient sampling efforts^[@CR27]^). False-positive detections can be related to other failures of the method (e.g., contamination and lack of DNA marker species specificity) as well as to the contamination of the studied system by external sources of eDNA, such as sewage effluents and animal excrements^[@CR25],[@CR28]^.
In streams, the concentration of eDNA and its detectability are not only dependent on production and degradation rates but also on dilution, transport through the river network, deposition and resuspension^[@CR27]^. The observed eDNA detection distances vary between studies from less than one km in mesocosm and field experiments^[@CR27],[@CR29]^ to around ten km at the outlet of lakes when detecting eDNA from two lake-dwelling invertebrate species^[@CR30]^. This last result suggests that eDNA in rivers integrates diversity information over space on the catchment scale^[@CR31]^. In a similar study, the detection distance of eDNA from a lake-dwelling fish in the outlet stream was also high (up to 3.6 km)^[@CR22]^. Nevertheless, this study also demonstrated that eDNA metabarcoding could reveal changes in fish communities along a small river stretch of less than 7 km, proving that this method can highlight biodiversity patterns along the river. Other studies have also demonstrated such spatial changes in fish communities along environmental gradients^[@CR19],[@CR28],[@CR32]^. However, the methodological differences between these studies hamper the detailed comparison of their results. Nevertheless, the detection distance in rivers is a key point to analyze how eDNA can reveal spatial and temporal changes in biological community structures, and additional efforts are needed to understand the spatiotemporal dynamics of eDNA in aquatic systems^[@CR33]^.
Another major challenge of eDNA methods is whether they enable the estimation of animal abundance^[@CR8],[@CR15],[@CR34]^. Quantitative eDNA analyses (qPCR, ddPCR) allow investigations into the strength of the relationship between eDNA concentrations and the density or biomass of a single fish species in mesocosms^[@CR15],[@CR35],[@CR36]^ or under natural conditions^[@CR25],[@CR29],[@CR37],[@CR38]^. Nevertheless, the eDNA concentration often explains a moderate amount of fish abundance variability^[@CR39]^ and is either better correlated with the number of individuals or the biomass. When using metabarcoding methods, quantitative fish detection is accomplished by assessing relationships between the abundance estimations obtained from traditional sampling methods and the number of eDNA copies or the frequency of positive samples^[@CR27]^. The correlations proved to be significant, but their moderate strengths implied that these results are more of a first proof-of-concept rather than a demonstration of a quantitative relationship with fish stocks. Higher water temperatures improved this relationship^[@CR40]^. In addition to the numerous factors influencing eDNA concentrations and dispersion in water, biases along the analytical pipeline can strongly alter the relationship between the initial quantity of eDNA in the water sample and the final number of reads per species^[@CR14],[@CR33],[@CR41]^. Nevertheless, a recent experiment demonstrated a high correlation between the number of sequence reads and the relative amounts of MOTUs (metabarcoding operational taxonomic units) when primers were carefully designed^[@CR42]^.
In this paper, we analyzed the capacity of eDNA metabarcoding to reveal quantitative patterns of fish biodiversity along a major European river, the Rhône River. eDNA was regularly sampled (Fig. [1](#Fig1){ref-type="fig"}) from Lake Geneva to the Mediterranean Sea (524 km long), and we used a metabarcoding approach previously described^[@CR21]^. First, to evaluate the ability of number of reads of MOTUs to allow between sites comparison of species relative abundance, we investigated its relationship with the detection rate for each of the MOTUs. Second, by comparing our results with historical traditional electrofishing (TEF) surveys, we tested the ability of eDNA metabarcoding to i) estimate species richness and relative species abundances on the local scale (a few kilometers), and ii) to reveal quantitative patterns of fish biodiversity on the scale of the entire longitudinal course of the river. Third, by combining field data and previous knowledge related to the behavior of fine particulate organic matter (FPOM) in rivers, we simulated the detection distances of eDNA metabarcoding for a large range of hydraulic conditions. Finally, we herein discussed the possibility of using eDNA metabarcoding for river monitoring.Figure 1Longitudinal profile of the Rhône River from Lake Geneva to the Mediterranean Sea (540 km). Locations of the hydroelectric power schemes and river sections from A to T (i.e., A-GE to T-PA, see Supplementary Table [1](#MOESM1){ref-type="media"} for the associated names). Locations of the confluence points with the main tributaries (and names of the five most important). Box: locations of the 59 eDNA samples (in red) and 40 TEF long-term surveys (in blue).
Results {#Sec2}
=======
Relation between detection rate and number of MOTU's copies {#Sec3}
-----------------------------------------------------------
When each of the 24 selected MOTUs (taxa) was modelled separately, all the relationships between number of standardized reads and detection rate were significant (explained deviance range: 32.3% to 87.6%, P \< 10^−20^ for each species). When pooling all the MOTUs (Fig. [2](#Fig2){ref-type="fig"}), the global model was also highly significant (P \< 10^−15^) and explained 64.4% of the total deviance. The detection rates were 5% (close to the detection rate for one positive PCR in 24), 50% (12 positive PCR in 24) and 95% (close to 23 positive PCR in 24) when MOTU's relative abundance were respectively 0.02%, 0.30% and 5.49%. Adding MOTU's identity and the interaction with the number of standardized reads to the model increased the explained deviance up to 70.6% of the total deviance (P \< 10^−15^). MOTU's identity and the interaction had both a significant effect (P \< 10^−15^) and explained respectively 4.21% and 1.93% of the total deviance in addition to the deviance explained by the number of standardized reads (Fig. [2](#Fig2){ref-type="fig"}). For a predicted detection rate of 5%, 50% and 95% the relative abundance of MOTUs ranges were respectively 0.01--0.11% (median value: 0.02%), 0.15--0.71% (median value: 0.37%) and 1.56--25.68% (median value: 5.18%) with the last model.Figure 2Relationship between number of standardized reads (in log) and detection rate (frequency of positive PCR) of MOTUs among sites. Grey cross: observed values. Red line: predicted detection rate as a function of the number of standardized reads (glm model). Blue lines: predicted detection rate as a function of the number of standardized reads, MOTU's identity and their interaction (glm model). List of the 24 MOTUs considered in the second model: Abr_bra, Alb_alb, Alb_bip, Alo_spp, Ang_ang, Bar_bar, Bli_bjo, Cor_lav, Cot_sp, Cyp_car, Cypr_1, Eso_luc, Lep_gib, Leu_spp, Liz_ram, Onc_myk, Perc_flu, Pho_pho, Pse_par, Rho_ser, Sal_spp, Sal_tru San_luc, Sil_gla (see Supplementary Table [2](#MOESM1){ref-type="media"} for the species names corresponding to the abbreviations).
Comparison of eDNA metabarcoding and TEF samples on the sampling site scale {#Sec4}
---------------------------------------------------------------------------
Among the 48 taxa detected by eDNA metabarcoding, four species likely do not exist in the Rhône River^[@CR43]^. *Argyrosomus regius* and *Salmo salar* have never been recorded in the Rhône catchment. *Salvelinus_spp* includes cold water species inhabiting only lakes or small brooks, and *Oncorhynchus mykiss* is a farmed species^[@CR43]^. The relative numbers of reads for *A*. *regius* and *S*. *salar* were low (maximums of 0.08 to 0.78%, respectively), but *O*. *mykiss* and *Salvelinus spp*. had normal read numbers (maximums of 5.0% and 89%, respectively). Among the other 44 species detected by eDNA, four were not registered in our TEF dataset but are known to be present in the Rhône River^[@CR43]^ (*Chelon labrosus*, *Mugil cephalus*, *Zingel asper*, *Misgurnus fossilis*). Two species caught by TEF were also detected by eDNA, but the numbers of DNA copies were below our threshold, and the species were discarded (*Leuciscus delineatus*, *Lota lota*).
Of the 16 selected river locations (L1 to L16, see Supplementary Table [1](#MOESM1){ref-type="media"}) for which pairs of eDNA and TEF samples were compared, the mean species richness values of the annual TEF samples (12.3 to 20.6 species) were always significantly lower (P \< 0.001) than those of the paired eDNA samples (26 to 33 species), i.e., 28.6% to 57.7% fewer species were identified in the TEF samples (Fig. [3](#Fig3){ref-type="fig"}). Considering the species that were sampled at least once during the ten years of the TEF surveys, the total (or cumulative) richness per sampling site (21 to 30 species) did not differ from that estimated by eDNA metabarcoding in one sampling session (Wilcoxon signed-rank test, P \> 0.05). Furthermore, 73% to 93% of the total numbers of species detected in each river location were common to both the TEF and eDNA samples (Table [1](#Tab1){ref-type="table"}). Zero to ten species per river location were detected by only eDNA, and they represented less than 1% of the total number of reads per sample in 84% of the cases (maximum of 4.1% of the total number of reads). The most frequent taxa not detected by TEF were Cypr_2, *Salmo trutta*, *Lampetra spp* and *Salaria fluviatilis*. Zero to five species per river location were detected by only TEF. The most frequent taxa not detected by eDNA metabarcoding were *Anguilla anguilla*, *Scardinius erythrophthalmus*, *Alosa spp* and *Leuciscus spp*. These taxa represented less than 1% of the total number of fish caught except in one case (*Ameiurus melas*: 2.66%).Figure 3Comparison of the mean number of species (±95% confidence interval) caught annually with TEF (blue circle), the total number of species (blue triangle) caught during the 10-year survey (2006--2016), and the number of species detected in eDNA samples (red squares) for each of the 16 locations (L01 to L16) sampled with both methods (see Supplementary Table [1](#MOESM1){ref-type="media"}).Table 1Comparison of species detections in eDNA and TEF samples (accumulated over ten years) at 16 sampling locations along the Rhône River.Sampling locationsNb. of species in commonNb. of species in only eDNA metabarcoding dataNb. of species in only TEF dataSpearman's coefficient (p value)L0129410.261 (p = 0.1714)L0220910.653 (p = 0.0018)L0325420.512 (p = 0.0088)L0425140.45 (p = 0.0241)L0525250.618 (p = 0.001)L0625340.333 (p = 0.104)L0728210.474 (p = 0.0107)L0828230.814 (p \< 0.0001)L0925500.346 (p = 0.0905)L1028020.643 (p = 0.0002)L1128120.478 (p = 0.0101)L1226230.578 (p = 0.002)L1324430.594 (p = 0.0022)L1424810.563 (p = 0.0042)L1526630.726 (p \< 0.0001)L16221020.53 (p \< 0.0001)The numbers of species detected in both eDNA and TEF samples and in only one of the two methods are specified. The Spearman's correlation coefficients between the standardized number of reads and the number of individuals caught per species and per CPUE (see Methods for details) and the associated p-values are also given.
The correlations between the number of reads (eDNA) and the number of fish caught (TEF) were significant (Spearman's rank correlation test) in 13 of the 16 selected river locations (Table [1](#Tab1){ref-type="table"}). For six species (*Barbus barbus*, *Abramis brama*, *Gymnocephalus cernuus*, *Barbatula barbatula*, *Phoxinus phoxinus*, *Sander lucioperca*), the relative abundance in the eDNA sample was significantly higher than that in the TEF sample (non-parametric signed test, P \< 0.01 after Bonferroni correction for multiple comparisons). Only two species had significantly higher relative abundances in the TEF samples than in the eDNA samples: *Alburnus alburnus* (P \< 0.001) and *Squalius cephalus* (P \< 0.05).
The Shannon and Evenness indices from the eDNA samples (2.21 to 2.83 and 0.68 to 0.82, respectively) were significantly higher than the mean Shannon (Wilcoxon signed-rank test, P \< 0.001) and mean Evenness indices (Wilcoxon signed-rank test, P \< 0.001 to P \< 0.05) from the annual TEF samples (1.19 to 2.03 and 0.47 to 0.68, respectively) (Fig. [4](#Fig4){ref-type="fig"}).Figure 4Mean values (±95% confidence interval) of the (**A**) Shannon and (**B**) Evenness indices for the 10 annual TEF samples (blue circles) and the eDNA samples collected in 2016 (red circles) at each of the 16 sampling locations (L01 to L16).
Longitudinal patterns of fish biodiversity revealed by eDNA metabarcoding and TEF {#Sec5}
---------------------------------------------------------------------------------
According to eigenvalues associated with the PCA analyses of eDNA (number of reads) and TEF samples from 14 of the 16 successive river sections from Lake Geneva to the sea, the first two principal components (PC1 and PC2) explained most of the structures for both datasets (59.1% and 51.8% of the total inertia, respectively). Regarding the co-inertia analysis performed using these first two principal components (PC), the co-inertia criterion was highly significant (RV = 0.736, p \< 0.0001), which demonstrates a clear similarity between the fish assemblage structures described by eDNA metabarcoding and TEF in each of the 14 river sections. The first two co-inertia axes accounted for 85.4% of the total inertia. The first and second co-inertia PCs were highly correlated with the first and second PCs of the independent PCAs performed on the eDNA metabarcoding (R = −0.982 and R = −0.951, respectively) and TEF (R = −0.923 and R = −0.872, respectively) datasets. The coordinates of the eDNA and TEF samples on the first and second co-inertia PCs (Fig. [5](#Fig5){ref-type="fig"}) were highly correlated (R^2^ = 0.859 and R^2^ = 0.763, respectively, P \< 0.001). The slope and intercept of the two regression lines did not differ from zero and one, respectively (Student's t-test, P \> 0.05).Figure 5Relationships between the standardized scores of the eDNA and TEF samples for each of the 14 river sections on the first (**A**) and second (**B**) principle components of the co-inertia analysis. Solid line: regression of TEF samples on standardized scores of the eDNA samples. Dashed line: regression line with an intercept of zero and a slope of one (dashed line).
The first and the second principal components (PC1 and PC2) of the PCAs performed on all the pooled eDNA samples within each of the 20 successive river sections accounted for 37.4 and 15.4%, respectively, of the total inertia of the dataset (Fig. [6](#Fig6){ref-type="fig"}). Among the 42 species studied, 15 and 16 species had a loading \>3% on PC1 and PC2, respectively. PC1 ordered the sections from upstream to downstream, and PC2 differentiated the first two river sections (A-GE, B-SE, large impoundments), associated with typical lake species (*Coregonus lavaretus*, *tinca*, *Perca fluviatilis*), and the last deltaic river section (T-PA), associated with estuarine species (*C*. *labrosus*, *Liza ramada*, *Mugil cephalus*). The river sections from the upper Rhône (C-CH to G-MI) were mainly characterized by lotic species (*B*. *barbus*, *Leuciscus spp*, *Thymallus*, *Cottus sp*.). Downstream from confluence with the Saône River, the river sections (I-VG to P-DM) had similar scores on both axes, and river sections Q-CA through R-AV were associated with typical lentic species (*Silurus glanis*, *Cyprinus carpio*, *Carassius spp*.) and long-distance migratory species (*A*. *anguilla*, *Alosa spp*).Figure 6Longitudinal profile of fish diversity as determined by principal component analysis performed on the eDNA samples from the Rhône River. Scores of the 42 species (**A**) and 20 successive river sections (**B**) on the two first principal components, PC1 and PC2 (See Supplementary Table [1](#MOESM1){ref-type="media"} for the names of the successive 20 river sections (A-GE to T-PA) and Supplementary Table [2](#MOESM1){ref-type="media"} for the species names corresponding to the abbreviations).
When considering the longitudinal profile of the river, Mantel tests between Bray-Curtis dissimilarity matrices and the longitudinal distance matrix were significant for both eDNA metabarcoding data (P \< 0.01) and TEF data (P \< 0.01). Mantel correlograms performed on the eDNA and TEF samples showed significant spatial autocorrelations across spatial extents of 70 km and 40 km, respectively (Fig. [7](#Fig7){ref-type="fig"}).Figure 7Mantel correlation between dissimilarities in fish species compositions (Bray-Curtis β-diversity) and distances between sampling stations for the eDNA (N = 27, red bold line) and TEF samples (N = 30, blue dashed line) for each 10-km distance class (76.5 to 493 km downstream from Lake Geneva). The filled circles (eDNA) and filled triangles (TEF) represent significant spatial correlations between the fish assemblages on the correlogram.
eDNA metabarcoding detection distance {#Sec6}
-------------------------------------
Whitefish (*Coregonus lavaretus*) is a lake-dwelling species for which eDNA was detected in abundance up to 60 km downstream from Lake Geneva (river section B-SE, 8.9% of the total number of reads) and in the two tributaries connected to the Annecy and Le Bourget lakes (kilometer point 64 and KP 80). Whitefish had a positive detection for less than 9 PCR replicates when the distance from the DNA source (river section B-SE's dam) was higher than 50 km and was no more detected at KP 195. The eDNA travelling time was estimated to be 41.7 hours for the first 100 km. The detection threshold predicted by the glm model (Fig. [8](#Fig8){ref-type="fig"}) fell below 4% at 130 km (±std error: 121--141 km).Figure 8Observed and predicted detection rate of *Coregonus lavaretus* downstream from B-SE river section's dam. Glm model predictions (blue thick line) and prediction area delimited by the standard error (blue fine lines). Detection rate was determined as the number of positive amplifications of *C*. *lavaretus* DNA in 24 PCR replicates.
According to V~dep~ values reported in the literature for fine particle organic matter (FPOM) (Supplementary Table [4](#MOESM1){ref-type="media"}), the median V~dep~ value was approximately 0.180 mm.s^−1^ (interquartile range 0.08--0.43). Using this median V~dep~ value, our simulations allowed correct predictions of the eDNA detection distance ranges observed in experimental and observational studies previously reported in the literature^[@CR22],[@CR26],[@CR29],[@CR30]^ and in this study when the corresponding hydrological conditions were considered (Table [2](#Tab2){ref-type="table"}).Table 2Comparison of eDNA detection distances observed in caged fish experiments (Exp) or observational studies (Obs) and those predicted by simulation considering a value of the vertical transfer of FPOM from the water column to the riverbed (V~dep~) computed from selected publications (median value ± interquartile range, see Supplementary Table [4](#MOESM1){ref-type="media"}) The maximal predicted detection distance is defined as the distance for which only one mtDNA copy of a hypothetical MOTU quantity released upstream is still present in the water column (see Method section for explanation).ReferenceMethodDNA analysisWaterflow (m^3^.s^−1^)Wetted width (m)Observed distance of detectionPredicted distance of detection (±Interquartile Q1--Q3)Jane *et al*. (2015)ExpqPCR0.0031.19\>239.5 m128 m (54--283)0.0313.35\>239.5 m443 m (186--979)Wilcox *et al*. (2016)ExpqPCR0.0061.1941 to 222 m235 m (98--518)Exp0.0103.3572 to 1,459 m137 m (57--303)Obs.0.0592.88100 to 900 m965 m (404--2131)Civade et al. (2016)Obs.meta barcoding0.1701.801.7 to 3.6 km4.5 km (1.9--9.9)Deiner et Altermatt (2014)ObsqPCR3.52014.00\>9.1 km11.9 km (5--26.3)3.52014.00\>9.1 km12.8 km (5.4--28.3)3.79014.001.6 to 9.1 km11.9 km (5--26.3)This studyObsmeta barcoding436.00132.00130.0 km (120.7--141.1)156.1 km (65.4--344.9)The water velocity (u) and the water depth (h) used to compute Sp and the maximal detection distance are approximated from the mean water flow (Q in m^3^) and the mean wetted width (w) considering a channel with a rectangular cross section.
Discussion {#Sec7}
==========
Comparing the eDNA metabarcoding sampling campaign and the TEF long-term surveys demonstrated the capacity of this approach to qualitatively and quantitatively (relative abundance) reveal the longitudinal patterns of fish biodiversity along a large river. Combining field data and a literature review, we also showed that eDNAm behaves in the river water column similarly to FPOM and that its detection distance is mainly dependent on the hydraulic characteristics of the river channel.
Our reference database Teleostei^[@CR21]^ covers all the fish biodiversity in the Rhône catchment. However, some molecular marker did not discriminate between species (Supplementary Table [2](#MOESM1){ref-type="media"}). Diversity of some genera was reduced from 7 to 3 species (*Carassius*, *Leuciscus* and *Salvelinus*), and identifying new markers to overcome this limitation is important when markers detect phylogenetically distant species (Cypr_1 and Cypr_2).
Of the species that were detected by only eDNA and were previously unknown in the Rhône River, four (e.g., *S*. *salar*, *A*. *regius*, *D*. *labrax*) are consumed in large quantities, and their DNAs most likely originated from treatment plants from the cities (Geneva, Lyon). In addition to being marketed as food, *O*. *mykiss* and *Salvelinus spp*. are subject to fish farming^[@CR43]^, and their DNAs are also released into the river with fish farm effluents. The abundance of *O*. *mykiss* eDNA could also have been related to the release of individuals by fishery associations at the time of our eDNA sampling campaign, a few weeks after the fishing season opening. Such eDNA transfer in rivers limits the investigation of commonly consumed or manipulated species^[@CR28]^, and eDNA sampling too close to the outlet of a lake or confluence with a tributary should be avoided.
Excluding estuarine species, 41 of the 48 species occurring along the Rhône River^[@CR43]^ were detected by eDNA metabarcoding, whereas 43 species were detected by the 10-year TEF surveys. Interest in using the eDNA method to monitor rare or elusive fish species has already been emphasized^[@CR45]^. Two critically endangered species and difficult to capture by electrofishing (nocturnal benthic species), the Apron (*Z*. *asper*) and the weather loach (*M*. *fossilis*), were detected in the Rhône River directly downstream from respectively the confluence of the Ardèche River and backwaters (section F-SB) where they are known to be present^[@CR43]^. eDNA method was also more efficient than TEF to differentiate the thicklip grey mullet (*C*. *labrosus*) and the flathead grey mullet (*M*. *cephalus*) from the thinlip grey mullet (*Liza ramada*), another estuarine species present in TEF samples.
Detection rates of a given MOTU vary with the amount of eDNA analyzed and the corresponding species abundance^[@CR46]^ and are used to evaluate the frequency of a MOTU between sites^[@CR30]^. The significant relationships between the standardized number of reads and the detection rates for each of the 29 MOTU/taxa demonstrated the validity of between sites comparison of the relative abundance of a given MOTU/taxa. Furthermore, our glm model including all the MOTUs highlighted also that the relative abundance of MOTUs are globally comparable between them and between sites. However, the significance of the MOTU's identity effect showed that a given detection rate did not correspond to the same standardized number reads according to the considered MOTU. But the deviance explained by this fixed effect is limited and the discrepancies between species allow to distinguish between relative abundance classes, a sufficient level of precision for biomonitoring.
When comparing the eDNA metabarcoding and TEF methods on the local scale (few km), the eDNA samples produced a more comprehensive species list than the TEF samples. eDNA metabarcoding has already been recognized as a more sensitive method than electrofishing^[@CR22],[@CR29]^ or netting^[@CR20],[@CR21],[@CR47],[@CR48]^ for fish detection. The similarity between the faunistic list of an eDNA sample and the cumulative list of species caught during all TEF campaigns demonstrates that all species detected by eDNA could be locally present. The large homing range of most of fish species^[@CR49]^ in combination with species rarity, can lead to very low detection probabilities with TEF. Most importantly, traditional methods are insufficient for providing a comprehensive species list with a limited sampling effort in large rivers^[@CR6],[@CR50]^. When the sampling effort was higher and different fishing methods were used in addition to TEF, the number of species caught in one traditional sampling campaign was comparable to that detected by eDNA metabarcoding with our workflow (Danube River study, Supplementary Table [3](#MOESM1){ref-type="media"}).
In addition to comprehensively estimating species richness, we observed a significant correlation (of medium intensity) between species relative abundance (TEF) and the number of standardized reads. Similar results were previously obtained with the metabarcoding approach^[@CR12],[@CR16],[@CR20]^, but contrasting results were also observed^[@CR19],[@CR39],[@CR51]^. The differences between TEF and eDNA metabarcoding can be related to the fact that in rivers more than two meters deep, TEF is possible only along the bank. Comparisons with other traditional fishing methods (gill nets, longline) in large rivers^[@CR6],[@CR51]^ confirm our results: *A*. *alburnus* and *S*. *cephalus*, two surface-oriented species, are overrepresented by TEF, whereas several species are underrepresented by TEF because they preferentially position themselves in the bottom layer of the water column (*B*. *barbus*, *A*. *brama* and *S*. *lucioperca*) or on the bottom (*G*. *cernua* and *B*. *barbatula*)^[@CR43]^. eDNA gives probably a better integrated view of fish assemblages in large rivers than classical methods.
The Shannon index values computed from the eDNA samples were unsurprisingly higher than those obtained from the annual TEF samples because of the differences in species richness. However, the higher Evenness index values achieved with eDNA demonstrate a better repartition of individuals between species than that obtained from TEF. We hypothesize that due to the turbulent conditions prevailing in rivers, the homogenization of eDNA is easier and makes MOTU detection more stable between sampling occasions^[@CR15]^.
When eDNA and TEF samples were pooled on a larger scale (14 river sections), the species relative abundances of the fish assemblages revealed by the two methods were highly correlated. The succession of the dominant species along the 20 successive river sections as presented by the eDNA samples (Supplementary Figure 2) agrees with previous descriptions^[@CR52]^. Cool water species are located upstream (*Esox Lucius*, *Cottus sp*, *Leuciscus spp*) at the opposite of estuarine species (*L*. *ramada*) while *Rhodeus sericeus* is abundant in midstream river sections. eDNAm also highlights the upstream migration of twaite shad (*Alosa spp*) during the sampling period^[@CR43]^ and the regular decrease in eel (*Anguilla Anguilla*) upstream, demonstrating the potential of eDNA metabarcoding to monitor anadromous species. Notably, such a comprehensive description of large-scale fish biodiversity patterns obtained in less than two weeks is not feasible with traditional fish sampling techniques.
The significant spatial autocorrelation between TEF samples up to 40 km shows the similarity of fish assemblages on the river section scale. The low diversity of aquatic habitats in such a heavily channelized and impounded river tends to decrease the spatial variability of fish assemblages^[@CR52]^. However, the spatial autocorrelation between the eDNA samples was significant within 70 km, which indicates an influence of the upstream production of eDNA far downstream due to its transportation.
Studying the downstream transport of eDNA and the review of literature regarding FPOM transport in streams makes discussing this last question feasible. We used the detection rate decrease in whitefish downstream from the outlet of three alpine lakes to evaluate its detection distance, assuming a constant release of whitefish eDNA from these lakes. As we did not consider the potential degradation of eDNA during the two-day transport period^[@CR10],[@CR11]^, we most likely underestimated the potential transport distance (S~p~). Conversely, the presence of occasional whitefish individuals within the Rhône River might have led to a slight overestimation of this distance. Our detection distance (130 km) is much higher than those already estimated experimentally in small streams (less than one km)^[@CR26],[@CR29]^ but also from dwelling species at the outlet of lakes (1.7 km to more than 9.1 km)^[@CR22],[@CR30]^.
Using an estimation of deposition velocity (V~dep~) based on previous studies (Supplementary Table [4](#MOESM1){ref-type="media"}) allowed to predict correctly the range of eDNA's detection distance observed in previous works and in this study (Table [2](#Tab2){ref-type="table"}). The hypothesis that the downstream decrease in eDNA is comparable to that observed for FPOM and highly dependent of the local hydraulic characteristics is confirmed. eDNA behaves like FPOM not only in small streams^[@CR29]^ but also in large rivers. Deposition velocity in streams is a balance between upward turbulent mixing and gravitational settling^[@CR44]^. V~dep~ hardly varies among particle types and size of FPOM but tend to increase significantly in streams with flows below 100 L.s^−1^ in relation to the proximity and roughness of the sediment (shear stress, transport storage zone)^[@CR53]^. Unlike V~dep~, S~p~ is the product of both deposition and transport characteristics, and deep, fast rivers transport FPOM further than shallow rivers^[@CR44]^. The scale at which eDNA reveals fish assemblage structures seems highly dependent of the river size. Simulations of the detection distance (Supplementary Figure [3](#MOESM1){ref-type="media"}) show that in small streams, eDNA transportation distance are comparable to the scale at which traditional sampling technics are performed (from less to few km). As the river size increases, eDNA is conveyed further downstream and deliver a more spatially integrated measure of biodiversity which is decoupled from a physical local habitat^[@CR31]^. But, in a large catchment as the Rhône River, eDNA collected downstream does not detect anymore the fish species which are dominant upstream (Supplementary Figure [2](#MOESM1){ref-type="media"}). To our opinion, the statement that eDNA sampling allows "an estimate of catchment-level diversity, including both aquatic and terrestrial taxa"^[@CR31]^ must be relativized according to the size of the watershed. In addition, as eDNA is subject to deposition velocity, the relative amount of eDNA originating from upstream decreases faster in comparison with the amount of eDNA released locally by fish species. Then, when analyzing not only species occurrence but also relative species abundance based on the standardized number of reads, our study demonstrates that eDNA reveals a correct picture of fish assemblage structures in a large river despite its downstream transportation. However other catchment systems of various size and with different species pools must be investigated to better understand the complex spatiotemporal dynamics of the information delivered by eDNA in rivers^[@CR33]^.
To our knowledge, our results are the first demonstration of the capacity of eDNA metabarcoding to describe large-scale quantitative patterns of fish assemblages in rivers. Compared to the electrofishing technique, eDNA provides a comparable image of the fish assemblage but integrates a larger extent than the traditional sampling location both upstream and within the entire cross river section. Our eDNA workflow combined into one single sampling campaign of 12 days is as efficient as ten years of traditional sampling effort (approximately 300 days of field work) at obtaining an accurate image of fish biodiversity on the watershed scale. Making precise inferences on fish abundance or biomass from eDNA metabarcoding is unrealistic, and the quantity of eDNA released by the fish is most likely also related to specific metabolism rates^[@CR40]^. However, eDNA metabarcoding can deliver valuable information about relative species abundance and be useful for monitoring purposes^[@CR48]^. The large quantity of water sampled ensures the detection of most rare species with only two replicates, and *in situ* filtration prevents the risks of contamination and DNA degradation prior to analysis^[@CR21]^. In practice, to limit the potential noise of transported eDNA and to ensure independence between eDNA samples in large rivers, the distance between the eDNA samples should be around 70 km. This distance depends on the river size, and a few kilometers between sampling sites should be sufficient in a smaller river. It is also recommended to avoid sampling close to locations where the introduction of exogenous fish eDNA to the river is suspected (e.g., tributaries, effluents). Indeed, despite the ecological and societal importance of large rivers, fish sampling in these systems remains costly, time consuming, and limited to specific habitats. eDNA metabarcoding thus appears to be a reliable, cost-effective method^[@CR45]^ for screening quantitative fish biodiversity patterns on a large-scale.
Methods {#Sec8}
=======
Study area {#Sec9}
----------
From the France-Switzerland border (Rhône River kilometer point KP 24) to the Mediterranean Sea (KP 540), successive dams (Fig. [1](#Fig1){ref-type="fig"}) delimit 20 river sections^[@CR52]^ (Supplementary Table [1](#MOESM1){ref-type="media"}). The two first sections (A-GE, B-SE) are deep impoundments directly connected to Lake Geneva (KP 0 to 60). Downstream, seventeen hydropower schemes separated by free-flowing sections (main channel MC) follow each other (Fig. [1](#Fig1){ref-type="fig"}), and 16 consist of a bypass reach (BPS, former riverbed) and a diversion canal (MC) with a hydropower plant. The final river section (T-PA), including the main channel and a secondary arm (DELT, Deltaïc Rhône), delimits the Island of Camargue (see Supplement Note 2 for the main hydromorphological and physico-chemical characteristics of the river).
eDNA sampling, *in situ* filtration and treatment {#Sec10}
-------------------------------------------------
A total of 59 water samples (Supplementary Table [1](#MOESM1){ref-type="media"}) were collected (KP 24.5 to KP 548) in each of the 20 successive sections of the Rhône River (MC, BPR and DELT: respectively 31, 20 and 8 samples) from April 6th to May 13^th^, 2016. The average discharge was 2154 m^3^.s^−1^ (1184 to 3115 m^3^.s^−1^). eDNA sampling was performed using a filtration device (VigiBOAT, SPYGEN, le Bourget du Lac, France) composed of a peristaltic pump (nominal flow of 1.1 L.min^−1^), a VigiDNA® 0.45-µM cross flow filtration capsule (SPYGEN, le Bourget du Lac, France) and disposable sterile tubing for each filtration capsule (Supplementary Figure [4](#MOESM1){ref-type="media"}). Two filtrations were performed in parallel at each site, each timed at 30 min for a water volume of approximately 30 L. A preliminary study on the influence of the sampling effort on eDNA detection showed that two filtrations were sufficient to detect more than 95% of the local species richness (Supplementary Figure 1). At the end of each filtration, the water inside the capsule was emptied, and the capsule was filled with 80 mL of CL1 Conservation buffer (SPYGEN, le Bourget du Lac, France) and stored at room temperature.
DNA extraction, amplification, high-throughput sequencing, sequence analyzing, and taxa assignment are described in Supplementary Note 1.
For convenience, all MOTUs are referred to as species in the text (Supplementary Table [2](#MOESM1){ref-type="media"}). Species not previously recorded in the Rhône River^[@CR43]^ were not considered in the subsequent data analysis. Two implausible detections by eDNA of estuarine fish species downstream from Lake Geneva (*Atherina boyeri* and *Dicentrarchus labrax*) were also discarded.
As the total number of DNA copies varied between sites, we standardized the reads numbers to ensure that the numbers of reads per species were comparable between sites^[@CR32]^ and that they can be interpreted in terms of relative abundance. All eDNA samples were resampled to randomly select 163,121 reads per site (R package MASS^[@CR59]^, function sample without replacement), which was the smallest total number of reads found at one site. All species detected in the initial dataset were still found after resampling.
Relation between detection rate and number of MOTU's copies {#Sec11}
-----------------------------------------------------------
We modelled the relationship between the standardized read number of each MOTU (log-transformed) and its detection rate (proportion of positive amplifications in 24 PCR replicates per site) among sites using generalized models (R software, package MASS^[@CR59]^, function glm, binomial error). Model's residual deviance was used as the goodness-of fit criterion in the model evaluation. We considered the 24 MOTUs covering most of the detection rate's range (3 to 23 positive PCR) and present in more than 10 sites. We modelled separately the relationship for each of the 24 MOTUs. Then we performed the model on the complete dataset, adding MOTU's identity and the interaction with the number of reads.
Comparison of eDNA metabarcoding and TEF samples on the sampling site scale {#Sec12}
---------------------------------------------------------------------------
Traditional electrofishing (TEF) was performed every year or two years along the entire Rhône River course (Fig. [1](#Fig1){ref-type="fig"}) for one decade (2006--2016) as part of regular surveys operated by Electricité de France (EDF, 20 sites) and the National French Agency for Biodiversity (AFB, 20 sites). TEF sampling methods are described in Supplementary Note 2.
Preliminary analysis showed that ten annual TEF samples were necessary to not underestimate the species richness (species accumulation curves, not presented here). Thus, only the TEF sites sampled every year (n = 25) were considered to compare TEF and eDNA samples. For all TEF samples we performed generalized linear model to ensure that the species richness per site did not show any consistent trend during the studied decade and pooled all the sampling sessions per site.
On the local scale (sampling location), the pairs of TEF and eDNA samples to be compared were required to be from the same type of river reach (MC, BPS or DELT). When different pairs of TEF and eDNAm sites could be selected from the same river section, the least distant pair of sites was selected. Finally, a total of 16 river locations (L01 to L16) distributed along the entire river were retained (Supplementary Table [1](#MOESM1){ref-type="media"}). The distances between the eDNA samples and TEF occasions ranged from 0.2 to 7 km (average of 3.6 km), a distance comparable to the recommended fishing length for large rivers (20 times the river width, i.e.2.8 to 8.8 km for our sites)^[@CR55]^. The species richness, Shannon index and Evenness index were computed using the package vegan (R software, functions specnumber and diversity)^[@CR54]^.
Longitudinal patterns of fish biodiversity revealed by eDNA metabarcoding and TEF {#Sec13}
---------------------------------------------------------------------------------
To compare quantitatively (relative abundance) the fish assemblages described by metabarcoding and TEF methods on a large scale (from Lake Geneva to the sea), the eDNA and TEF samples were pooled within each of the successive river sections where the two methods were available (14 of the 20 river sections) and log-transformed.
The longitudinal patterns of fish assemblages revealed by eDNA and TEF were compared using co-inertia analysis (R software, package ade-4^[@CR57]^, functions dudi.pca and coinertia). This multivariate method allows the comparison of the ordinations of two datasets to find the orthogonal co-inertia principal components (PC) that maximize the co-inertia (or co-variance) between the two datasets^[@CR57]^. This method is especially suitable when many species are sampled from a few sites. First, we conducted separate analyses of the TEF and eDNA datasets with two centered principal component analyses (PCAs) and then processed the co-inertia analysis on the principal components of the first two PCAs of each dataset. The RV co-inertia criterion (0 to 1) measures the adequacy between the two tables^[@CR57],[@CR58]^ and its significance (Monte-Carlo test, 10,000 permutations).
To comprehensively describe the quantitative patterns of fish biodiversity along the Rhône River as revealed by eDNA, the eDNA samples were pooled in each of the 20 river sections and analyzed with a centered PCA.
For each sampling method, we performed a Mantel test (R software, package vegan^[@CR56]^, functions mantel and mantel.randtest) to evaluate whether the similarity between samples located on the main river course (TEF: 31 sites, eDNA: 39 sites) was dependent on the distance between the sampling sites. We used the Bray-Curtis index as a measure of the dissimilarity between assemblages and performed 999 permutations for 19 geographic distance classes (ten-km width) among the sites. For each sampling method, a correlogram (R software, package vegan^[@CR56]^, function mantel.correlog) was performed to determine the maximal geographic distance between samples for which the spatial autocorrelation remained significant.
eDNA metabarcoding detection distance {#Sec14}
-------------------------------------
Whitefish (*Coregonus lavaretus*) is abundant in the three Alpine lakes connected to the upper Rhône^[@CR43]^ but is only found occasionally in the Rhône River itself^[@CR52]^ downstream from section B-SE: only one individual reported in our TEF dataset, and less than one-thousandth of the total catch by fishermen^[@CR59]^. We used generalized linear model (R package MASS, function glm) to analyze detection rate of eDNA relative to the downstream distance of the sampling site from section B-SE. The maximal detection distance was defined as distance before detection drops below a 4% threshold (less than 1 PCR positive on 24)^[@CR30]^. We tested the hypothesis that the downstream decrease in eDNA was mainly due to a sedimentation rate similar to that observed for fine particulate organic matter (FPOM)^[@CR29],[@CR44]^.
The vertical transfer of FPOM from the water column to the riverbed (V~dep,~ deposition velocity) was expressed as V~dep~ = u.h/S~p~ (Eq. 1), with S~p~ the distance in m needed to retain 63.2% of the FPOM in the riverbed, u the mean water velocity (in m.s^−1^) and h the mean water depth (in m)^[@CR44]^. According to the V~dep~ values reported in the literature (Supplementary Table [4](#MOESM1){ref-type="media"}) we compared the eDNA's detection distance observed in experiments or observational studies (including this study) and the eDNA's detection distance predicted by V~dep~ for the hydrological conditions prevailing in each previous study. The predicted maximal detection distance was defined as the distance for which only one mtDNA (mitochondrial DNA) copy of a MOTU released upstream was still present in the water column, considering an initial released quantity of 2000 mtDNA copies/L multiplied by 2.5, to simulate taking a 2.5-L sample^[@CR29]^. All statistical analyses were done with the program R, version 3.3.3^[@CR60]^.
Data Availability {#Sec15}
-----------------
All Illumina raw sequences data are available on Dryad 10.5061/dryad.t4n42rr.
Electronic supplementary material
=================================
{#Sec16}
Supplementary Material
**Electronic supplementary material**
**Supplementary information** accompanies this paper at 10.1038/s41598-018-28424-8.
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Funding for this work was provided by the "Compagnie Nationale du Rhône" (CNR), SPYGEN, the Association Beauval Nature and the ANR AquaDNA (ANR-13-ECOT-0002). We thank Marie Pierron, Coline Gaboriaud, Jean-Baptiste Decotte, Lionel Meric and Christophe Mora for contributing to the field work and/or the laboratory analysis. We thank "Agence Française pour la Biodiversité" (AFB) and "Electricité De France" (EDF) for providing electrofishing survey data from the National Fish Monitoring Survey Database and the monitoring program carried out near EDF nuclear power plants, respectively. We also thank Thomas Changeux for his valuable contribution regarding fishing gear and fisheries statistics in the Rhône River. This work is part of the activities of the EU COST Action DNAqua-Net (CA15219) "Developing new genetic tools for bioassessment of aquatic ecosystems in Europe".
R.C., T.D., P.J., D.P., N.R. and A.V. performed the preliminary experiments to calibrate this study. M.R. and T.D. designed the study. T.D., P.J., M.R. collected the eDNA samples in the field. T.D., P.J. and A.V. conducted the laboratory and bioinformatics analyses. A.M. and N.R. prepared and provided the electrofishing survey data. M. S. and H. Z. collected and analyzed the data related to the comparisons between traditional sampling techniques and eDNA on the Danube River. D.P. performed the statistical data analysis, prepared the figures and wrote most of the manuscript with significant contributions from all the authors.
Competing Interests {#FPar1}
===================
'teleo' primers and the use of the amplified fragment for identifying fish species from environmental samples are patent. This patent only restricts commercial applications and has no impact on the use of this method by academic researchers. MR and AM are research engineers of two French electricity generation companies, AV and TD are research scientists in a private company, specialized on the use of eDNA for species detection. DP acts as a scientific adviser of SPYGEN.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#jve11-sec-0005}
============
HIV enters the central nervous system (CNS) in primary HIV infection (PHI), most likely via the migration of infected immune cells across the blood--brain barrier [@jve11-bib-0001]. In untreated patients, HIV can be detected in cerebrospinal fluid (CSF) throughout the course of infection [@jve11-bib-0003], which in part may reflect establishment of a local reservoir for HIV infection within the CNS [@jve11-bib-0004]. Early establishment of CNS HIV infection may contribute to neurocognitive impairment and neurological complications observed in chronic HIV despite combination antiretroviral therapy (cART) [@jve11-bib-0005].
In HIV-associated dementia (HAD), genetic compartmentalisation frequently exists between viral variants in the plasma and CSF [@jve11-bib-0006]. Additionally, the presence of discordantly resistant variants in CSF may be associated with virological 'escape' within the CNS in patients on ART, which may be more common than previously recognised [@jve11-bib-0008], and in rare cases associated with overt clinical neurological disease [@jve11-bib-0009]. Although biologically important compartmentalisation has been thought to arise late in the course of HIV, recent recognition that differences in the viral quasispecies in the CNS and blood compartments may arise in early infection [@jve11-bib-0011] may have implications for HIV treatment and eradication strategies.
We performed a pilot study employing deep-sequencing (DS) methods to assess HIV from paired CSF and plasma samples obtained during PHI to determine the feasibility of amplifying adequate sequences from CSF, to detect mutations within the low-level CSF viral variants, and to assess for differences in sequences between the CSF and blood compartments during early infection.
Methods {#jve11-sec-0006}
=======
Study design {#jve11-sec-0007}
------------
Samples were obtained in the context of an observational longitudinal neurological study of PHI in San Francisco, USA [@jve11-bib-0001]. Paired baseline plasma and CSF samples from five cART-naïve participants who enrolled between January 2008 and July 2010 were selected based on specimen HIV-RNA \>3,000 copies/mL to optimise success with DS. Confirmation of PHI (\<1 year after transmission) and estimated timing of infection were ascertained as previously described [@jve11-bib-0001]. Informed consent was obtained from participants and the study was approved by the involved Institutional Review Boards.
Specimen sampling and laboratory studies {#jve11-sec-0008}
----------------------------------------
CSF, blood and general medical and neurological assessments were obtained as previously described [@jve11-bib-0001]. Background CSF clinical values and blood CD4+ and CD8+ T lymphocyte counts by flow cytometry were measured in the San Francisco General Hospital Clinical Laboratory on fresh samples. CSF was spun at 1,200 × g for 10 min then supernatants and blood plasma were aliquoted and stored within 6 hours of collection in −70°C freezers.
Virological methods {#jve11-sec-0009}
-------------------
HIV-RNA levels were measured in previously frozen CSF and plasma using the ultrasensitive (20 copies/mL lower limit of detection) Amplicor HIV Monitor (version 1.5; Roche Molecular Diagnostic Systems, Branchburg, NJ) or the Abbott RealTime HIV-1 (Abbot Laboratories, Abbot Park, IL, USA) assays. Paired blood and CSF measurements were made in the same PCR run.
Deep-sequencing methods {#jve11-sec-0010}
-----------------------
Using SuperScriptII (Invitrogen, Waltham, MA, USA), first-strand cDNA was generated with three gene-specific oligonucleotides and treated with RNAse H (NEB). Subsequently, eight partly overlapping amplicons tagged with sequencing adaptors were generated from the cDNA by 40 cycles of PCR with FastStart HiFi Polymerase (Roche, Penzberg, Germany). The amplicons were purified with AMPure magnetic beads (Beckman Coulter, Brea, CA, USA) and quantified by use of PicoGreen (Invitrogen). After equimolar pooling of all 8 amplicons per sample, clonal amplification on beads (emulsion PCR) was performed using reagents that enabled sequencing in both the forward and reverse directions (kit II and kit III, respectively; Roche--454 Life Sciences, Branford, CT, USA). After bead isolation (breaking), enriched DNA-containing beads were counted on a Multisizer3 (Beckman Coulter). Thirty thousand beads per sample were prepared for ultra-deep sequencing and loaded in one region on a PicoTiter plate fitted with a 16-lane gasket. Sequencing was performed on a Genome Sequencer FLX (Roche--454 Life Sciences). The minimum accepted sequence coverage was 700 clonal reads for each position of interest, with average coverage for the sample set estimated at approximately 1,700 clonal reads per position. GS Amplicon Variant Analyzer (AVA) software was used to analyse the ultra-deep sequencing results. Amplicon nucleotide sequence reads were aligned to a consensus sequence generated from \>6,000 sequences obtained from the Los Alamos HIV sequence database ( [www.hiv.lanl.gov/](www.hiv.lanl.gov/)). The underlying flowgram signals were used in concert with each read\'s base-called nucleotides to facilitate alignment accuracy. The direct application of the flowgram signals, rather than quality scores, allowed the consideration of any potentially undercalled nucleotides during the alignment process -- bases for which no quality scores would exist. Interread similarity was also exploited for the alignment of any insertion bases relative to the reference sequence. Reads from both orientations and from overlapping amplicons were combined into a single alignment, and primer regions were automatically trimmed to avoid artefacts from the nucleotide content of the synthesised primers. AVA software was used with predefined codons conferring drug resistance based on the Stanford HIVdb Genotypic Resistance ( <http://hivdb.stanford.edu>) mutations.
Analysis {#jve11-sec-0011}
--------
Inclusion threshold was a variant frequency of at least 0.2%. Estimated mutational load (ML) was calculated by multiplying mutant variant frequency by HIV-RNA level, which is a method used frequently in similar studies [@jve11-bib-0014].
Results {#jve11-sec-0012}
=======
Subject characteristics are shown in Table [1](#jve11-tbl-0001){ref-type="table"}. DS generated abundant reads in all samples. Plasma samples averaged 7,124 reads of PR and 2,448 reads of RT, whereas CSF samples averaged 7,082 and 2,792 reads, respectively. In CSF samples of subjects 9024, 9039 and 9044, and plasma of 9024, RT codons 88--148 were not covered.
######
Background clinical characteristics and reverse transcriptase (RT) and protease (PR) mutations detected in paired CSF and plasma specimens in PHI participants
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Age\ Duration of HIV infection\ CD4 cell count\ HIV-RNA plasma\ HIV-RNA CSF\ CSF WBC\ Plasma RT\ CSF RT\ Plasma PR\ CSF PR\
(years) (days) (cells/μL) (log~10~ copies/mL) (log~10~ copies/mL) (cells/μL) %, Mutant variants (copies/mL) %, Mutant variants (copies/mL) %, Mutant variants (copies/mL) %, Mutant variants (copies/mL)
---------- --------- ---------------------------- ----------------- --------------------- --------------------- ------------ ---------------------------------------------------------------- --------------------------------------------------------------- -------------------------------- ------------------------------------------------------------
**9024** 33 222 974 4.75 4.29 12 *Wildtype*\ *Wildtype*\ 0.69 D30N (388)\ *Wildtype*
0.74 D67G (416) *c88-148 not covered* 0.30 V82A (169)\
0.30 N83D (169)\
0.27 I85V (152)
**9039** 33 37 539 5.57 4.30 53 0.33 M184V[^a^](#jve11-fn-0002){ref-type="table-fn"} (1226)\ 0.29 F77L (58)\ *Wildtype* 0.52 V82A (104)\
99.4 T215D[^a^](#jve11-fn-0002){ref-type="table-fn"} (369000)\ 99.4 T215D[^b^](#jve11-fn-0002){ref-type="table-fn"} (19838)\ 0.27 N83D (54)
0.42 T227L (1560) 0.26 T227L[^b^](#jve11-fn-0002){ref-type="table-fn"} (52)
**9044** 25 128 533 4.82 3.49 20 0.32 V75A (211) *Wildtype*\ 0.46 M46I (304)\ 0.39 V82A (12)
*c88-148 not covered* 1.16 I47V (766)
**9055** 32 106 619 5.38 3.51 12 0.33 D67G (791)\ *Wildtype*\ *Wildtype* *Wildtype*
*c88-148 not covered* *c88-148 not covered*
**9058** 29 110 237 5.18 3.67 4 0.34 G190E (515)\ 1.56 G190E (72)\ *Wildtype* 9.77 D30N (451)\
0.72 E138K (1727)\ 0.59 L210W (27) 8.10 M46I[^c^](#jve11-fn-0002){ref-type="table-fn"} (374)\
1.31 T69N (1983)\ 0.27 V82A[^c^](#jve11-fn-0002){ref-type="table-fn"} (12)\
0.44 D67G (666)\ 5.16 G73S (238)
0.27 T227L (409)
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
CSF, cerebrospinal fluid; WBC, white blood cells; RT, reverse transcriptase gene; PR, protease gene.
Denotes presence of linkage between indicated mutations within each specimen and region.
Highlighted text shows where \>1% mutant variants were found.
Mutations associated with transmitted drug resistance were detected in low abundance in CSF and blood in multiple participants (see Table [1](#jve11-tbl-0001){ref-type="table"}).
In subject 9058, sequencing of the RT region from both the CSF and plasma revealed NRTI and NNRTI drug resistance, while the PI region for plasma showed no significant drug resistance mutations. However, in the PI region for the CSF sample, D30N (9.8%, ML 451 copies/mL), M46I (8.1%, ML 374 copies/mL), V82A (0.27%, 12 copies/mL) and G73S (5.2%, ML 238 copies/mL) mutations were detected. In this sample, D30N, M46I and V82A mutations were found to be linked. Results of their interpretation through the Stanford HIVdb Genotypic Resistance Interpretation Algorithm are shown in Table [1](#jve11-tbl-0001){ref-type="table"}. Numerous other non-drug-resistance mutations that deviated from consensus sequence were found (not shown). Figure [1](#jve-1-264-g001){ref-type="fig"} shows longitudinal plasma and CSF HIV-RNA levels of participants prior to and after cART, indicating viral suppression to the level of assay detection in blood and CSF in all four subjects who initiated cART during follow-up. Rate of HIV-RNA decay after cART could not be compared between study subjects due to variable sampling intervals, but was rapid in CSF even in the participant harbouring compartmentalised PI mutations in the CSF, and, based on the available data, was similar in the blood and CSF compartments in all four subjects.
{#jve-1-264-g001}
Discussion {#jve11-sec-0013}
==========
In this pilot study, we demonstrate the feasibility of investigating compartmentalisation between CSF- and blood-derived HIV through detection of low-abundance HIV drug-resistant viral variants using DS. To our knowledge there has only been one other report of the use of DS to examine the CSF HIV quasispecies [@jve11-bib-0015]. In their report Tong *et al*. employed the same technique to examine the CSF HIV quasispecies of a limited number of subjects, without other confirmation.
In our study, this highly sensitive method for mutation detection revealed linked drug-resistance mutations present at significant levels (up to 9.77%) in CSF but which were not detected in plasma during PHI in one out of five subjects selected for our pilot study. DS is a recent addition to the armamentarium for detection of viral variants in HIV populations. Low-level variants can be detected through a high-throughput automated assay, as opposed to detection of predominant population sequences through Sanger sequencing. The aim of this pilot study was to determine the feasibility of applying HIV DS methods to CSF. The high number of reads in the RT and PR genes confirms the success of application of DS to previously frozen CSF samples. Our findings may be influenced by the fact that, as samples were selected based on HIV-RNA levels, our participants had median CSF HIV-RNA levels higher than those in the overall observational study cohort (3.67 log~10~ copies/mL *vs*. 2.62 log~10~ copies/mL). However, these data suggest that DS may be employed to examine CSF HIV in other relevant subject groups, provided CSF HIV-RNA levels are adequate for amplification.
The significance of early neuroinvasion for the course of CNS HIV infection remains unclear. One key question is when CNS compartmentalisation, measured through distinctions between CSF and plasma-derived HIV, is established during the course of disease. Compartmentalisation may result from independent replication of virus within the CNS, or a bottleneck or selection of HIV upon entry into the CNS. Previous studies have documented phenotypic or genotypic compartmentalisation between CNS-derived and systemic HIV populations in chronic infection [@jve11-bib-0016]. Specifically, CNS compartmentalisation of *env* correlates strongly with presence of dementia [@jve11-bib-0007]. Several studies have specifically detected discordant RT and PR resistance genotypes in CSF and plasma using conventional genotyping methods, also noting increased genetic distance between compartments in the setting of HAD [@jve11-bib-0006]. Furthermore, drug-resistance mutations using conventional genotyping methods have been associated with virological failure in CSF in chronic infection [@jve11-bib-0019]. The prospective association between low-level drug resistance and emergence of CSF HIV 'escape' has not been investigated. However, recent studies have demonstrated a high prevalence of asymptomatic CSF 'escape' [@jve11-bib-0008], or more rarely, CSF escape with progressive CNS disease in patients on systemically suppressive cART [@jve11-bib-0009], which has been associated in some cases with CSF HIV resistance detected by standard genotyping.
As a pilot study to determine the feasibility of performing HIV DS on CSF specimens, the small sample size was one limitation of our study. However, the main objective of this study was to prove the concept and determine the feasibility of performing HIV DS on CSF specimens and provide a rationale for further investigation with a larger number of participants. In the next phase of the study we plan to quantify the cDNA template input for all our samples. Additionally, some of the distinct variants in Table [1](#jve11-tbl-0001){ref-type="table"} had very low MLs and frequencies near the limit of detection for DS [@jve11-bib-0012]. Thus, some variant difference may have been due to the error rate in HIV, or enzymatic errors due to RT-PCR or DS. Furthermore, although DS can be performed on samples with a low copy number, the levels of mutations identified reflect mutations present in the RT-PCR amplicons and may or may not reflect the variant distribution in the sample [@jve11-bib-0020]. The significance of these mutations at a level below 1% is unknown. Construction of phylogenetic trees based on the numerous variants was beyond the scope of this analysis. Despite these caveats, one of the participants (9058) had clear discordances in the variant distribution between the two compartments at a level between 5--10%. In this participant, three of these PR mutations in the CSF were linked.
Our DS results are consistent with prior findings in this PHI cohort indicating limited compartmentalisation of CSF HIV detected by standard methods [@jve11-bib-0011]. The detection of frequent mutant variants by DS is consistent with a moderate prevalence of plasma-transmitted drug resistance by standard genotyping in the community from which the participants were enrolled [@jve11-bib-0021]. Low-level plasma drug-resistance mutations may impact long-term systemic treatment outcomes, since cART-naïve participants with NNRTI-resistance mutations detected by DS more frequently experience virological failure after initiating NNRTI and NRTI regimens [@jve11-bib-0022]. It is unclear what mutational load is required for these low-abundance drug-resistance mutations to have an impact on clinical outcome, and further investigation is necessary. The treatment outcome of subjects harbouring low-level drug-resistant variants may be successful if a boosted PI-based regimen with a higher resistance barrier is used [@jve11-bib-0020].
The clinical significance of CSF low-prevalence drug-resistant variants in PHI is unknown. In very limited clinical follow-up, initiation of PI/r or integrase-based cART in four of our study subjects led to successful viral suppression in both compartments (Figure [1](#jve-1-264-g001){ref-type="fig"}). Possibly, such variants have no impact on the eventual course of HIV infection provided treatment is initiated and suppresses high-level viral replication. All four subjects in our study who started cART appeared to have similar viral decay rates in both compartments, although sampling frequencies were not adequate to assess initial decay phases in both CSF and blood. Alternatively, low-abundance drug-resistant viral variants may be significant in the CSF if treatment is delayed allowing for further replication of resistant virus, or if regimens are selected with poor CNS penetration or targeting distinct resistance genotypes detected in plasma. Interestingly, the participant in this study (subject 9058) who had low-level resistant variants in CSF prior to cART was found in another study to have unique envelope sequences in CSF as compared to blood in the context of suppressive cART using single-genome sequencing, implying the possibility of persistence of compartmentalised HIV in the CNS on cART [@jve11-bib-0023]. It is possible that low-level drug resistance detected in this analysis may have had an impact on this persistence. Future studies dedicated to investigating low-abundance HIV variants in CSF in different stages of infection, assessing changes over time in dual compartments prior to therapy, as well as observing the response to cART initiation will be likely to yield a new understanding of the importance of early detection and characterisation of low-level CSF variants in individuals infected with HIV.
We sincerely thank the participants who volunteered for these studies. We also thank the San Francisco General Hospital (SFGH)/University of California--San Francisco (UCSF) Clinical Research Center, staff at the ARI-UCSF Laboratory of Clinical Virology, and staff at the UCSF Options Project and Magnet for their invaluable help. This work was supported by the National Institutes of Health (grants R01 MH081772, K23 MH074466, and M01 RR0008336), and was previously presented in part as a late-breaker platform presentation at the International AIDS Society Conference, Rome, Italy, July 2011.
{#jve11-sec-0014}
Conflict of interests statement {#jve11-sec-0015}
-------------------------------
MJK reports serving as the principal investigator on grants that Yale University received from Merck, Pfizer, Gilead Sciences, Hologic, Abbott, GlaxoSmithKline, Bristol-Myers Squibb, ViiV and Vertex. MJK also receives royalties from a patent owned by Stanford University for some HIV diagnostic tests. EPS-J and BBS are employees of 454 Life Sciences. RWP and SSS have received support for travel and an honorarium from AbbVie Inc., for presentations at a scientific meeting.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s0005}
============
Apical organs are found in the larval stages of phylogenetically diverse animal groups ([Fig. 1](#f0005){ref-type="fig"}A) such as anthozoan cnidarians, protostomes (e.g. molluscs and annelids) and deuterostomes (e.g. echinoderms and hemichordates). The functions of apical organs are not well understood and may vary between taxa. Given their temporal restriction to larval stages, apical organs have been proposed to play an important role in the detection of settlement cues and for the induction of metamorphosis, presumably acting as chemosensory and/or mechanosensory structures. Such a function has been confirmed by functional approaches in only few species ([@bib13; @bib29; @bib43; @bib77; @bib103; @bib104]). The regulation of this life cycle transition not only represents a key developmental, but also an important ecological role, since the recruitment to the substrate and the metamorphosis of swimming larvae contribute to the shaping of benthic communities (e.g. [@bib83; @bib84; @bib111]). Moreover, apical organs are thought to regulate the activity of ciliary bands and musculature in some bilaterians and thus likely contribute to their locomotion ([@bib16; @bib28; @bib81]).
Apical organs are often considered the only larvae-specific organ, and according to this assumption, the question whether they share a common evolutionary origin has important implications for the evolution of metazoan life cycles. It has been proposed that apical organs could represent the simple brain of a gastrula-like ancestor of cnidarians and bilaterians (see [@bib65]). If this holds true, this common ancestor would have developed through a larva-like stage and adult stages were either added multiple times independently or have diversified drastically during evolution. If instead apical organs are not homologous, this would lend support to the hypothesis that the development of the common ancestor did not comprise a larva-like stage and that larvae in different taxa are the product of convergent evolution ([@bib75; @bib87]). However, the distinction of direct vs indirect development (i.e. with or without larval stage) can be difficult and, accordingly, using potential homology of apical organs to infer the evolution of life cycles is not straightforward.
Two distinct aspects can be considered when discussing the potential homology of apical organs. The first one is their position: apical organs are generally located at the pole opposite to the gastrulation site ([@bib67]), in a specific territory of the embryo called the apical domain ([@bib46; @bib67]). This position tightly couples the development of apical organs to the patterning of the apical-blastoporal axis.
Besides this shared position, diagnostic morphological characters of bilaterian apical organs are rather limited: in general, a tuft of long cilia emerging from mono- or multiciliated cells is present, accompanied by flask-shaped receptor cells, often positive for serotonin or FMRFamide immunoreactivity and connected to the larval nervous system by a plexus of neurites ([@bib9; @bib31; @bib60; @bib78]). The morphology, number and arrangement of nerve cells associated with apical organs can vary considerably (e.g. [@bib1; @bib20; @bib31; @bib43; @bib92; @bib105]). In deuterostomes, the cell bodies of the apical organ-associated neurons are typically located in or close to the apical portion of the ciliary bands and can be organized into bilaterally arranged ganglia ([@bib9; @bib67]). Generally, larvae swim with their apical organs pointing forward, prior to settling on a substrate and undergoing metamorphosis---at which point the apical organ degenerates ([@bib27; @bib35; @bib66]).
Since Cnidaria are the sister group to Bilateria, their apical organs are particularly relevant for evolutionary comparisons ([@bib32; @bib72]). Based on similarities in the genetic control of the development of the bilaterian anterior/apical pole and the aboral pole of *Nematostella vectensis* homology of these two domains has recently been proposed ([@bib51; @bib80; @bib86]), suggesting that apical organs in a cnidarian and in bilaterians develop from homologous territories. However, these studies addressed the homology of the apical organ itself only to a very limited extent. Cnidarian apical organs with a ciliary tuft have only been identified in anthozoan, and not in medusozoan planulae. They have been described as a set of columnar cells with basal nuclei, with a cytoplasm filled with various undetermined vesicles, and an underlying plexus of neurites ([@bib11; @bib62]) but, in contrast to bilaterians, distinct nerve cells associated with the apical organ have not been identified. In addition to the long cilia-bearing cells, the apical organ of *N. vectensis* contains two types of gland or secretory cells ([@bib62]). Clearly, the paucity of comparable morphological characters both in bilaterians and cnidarians complicates inferences about the homology of apical organs. Moreover, possible homologies may not be found at the level of the whole apical organ, but rather at the level of individual cell types ([@bib3; @bib51])
Accordingly, a molecular characterization of the cells constituting the apical organ is necessary in order to improve our understanding of this structure from both a functional and evolutionary perspective. Ideally, such a molecular characterization would include genes which control the development of the apical organ cells (i.e. signalling molecules and transcription factors), and genes that define specific cellular structures and processes (e.g. cytoskeletal components or metabolic genes). Developmental regulators are frequently used to test potential homology of characters (e.g. body parts or cell types) in different organisms with the assumption that a common evolutionary origin (i.e. homology) should be reflected in a shared developmental program ([@bib18; @bib106]). However, substantial parts of a particular developmental program can be re-employed for the formation of evolutionarily unrelated characters. This is most evident for signal transduction pathways which often control the development of a plethora of characters in a spatially and temporally distinct manner (e.g. [@bib34; @bib44; @bib48; @bib76; @bib102]). Thus, information about the differentiated structure is an important aspect of a homology assumption. This information is typically derived from morphological analyses, but can also be provided by a molecular characterization, e.g. by the structural or physiological components that are present in the structure. For the comparison of cnidarian and bilaterian apical organs, information about their development and about the apical organ cells is currently scarce, mainly due to the absence of apical organs in the classical model organisms for molecular developmental biology, such as *Drosophila*, *Caenorhabditis elegans* or vertebrates. Within bilaterians, a set of conserved transcription factors and signalling molecules has been identified that is expressed in the apical organ regions in the trochophore larvae of the annelid *Platynereis dumerilii*, in sea urchin embryos and in the hemichordate *Saccoglossus kowalewskii*. These similarities have been used to propose the homology of bilaterian apical organ cell types and their development ([@bib51]).
In the present study we aim to provide a molecular fingerprint of the apical organ in the anthozoan *N. vectensis* ([Fig. 1](#f0005){ref-type="fig"}B)*,* using an approach which is not based on candidate genes. The starting point of this study is the involvement of FGF signalling in the formation of the apical organ, and in particular of two FGF ligands that are co-expressed at the aboral pole and which have opposite functions: *NvFGFa1* promotes the formation of the apical organ and the associated tuft of cilia, while *NvFGFa2* restricts the size of the apical organ, as demonstrated by gene knockdown experiments ([@bib77]). Importantly, *NvFGFa1* and *NvFGFa2* function specifically in the formation of the apical organ, but not in the definition of the aboral territory ([@bib86]). The phenotypes resulting from the injection of morpholino (MO) antisense oligonucleotides directed against the two FGF ligands are easily discernible at larval stage: *NvFGFa1* MO produces embryos without a ciliary tuft, while *NvFGFa2* MO produces embryos with a strongly expanded tuft ([@bib77]). This observation provided us with a unique tool to identify genes that are expressed in the apical organ by comparing the gene expression profiles of these two situations by microarray analysis (see [Fig. 1](#f0005){ref-type="fig"}C). Since previously identified transcription factors and signalling molecules with potential roles in the development and/or the differentiation of the apical organ remain expressed throughout planula stage (e.g. [@bib50; @bib52; @bib69; @bib86]) the expression analysis at this relatively late stage should in principle allow the identification of both developmental regulators and structural genes.
The 78 genes recovered with this approach provide a basis for a detailed characterization of the cells constituting the apical organ of *Nematostella*. Moreover, by analysing the expression of a subset of these identified apical organ genes in sea urchin, we demonstrate the utility of this dataset for comparative studies.
Materials and methods {#s0010}
=====================
*Nematostella* culture {#s0015}
----------------------
*N. vectensis* was maintained and induced to spawn in the Sars Centre facility, as described previously ([@bib25; @bib30]).
Design and analysis of microarray {#s0020}
---------------------------------
The experimental design is illustrated in [Fig. 1](#f0005){ref-type="fig"}C. *Nematostella* eggs were injected after fertilization, with either morpholino for *NvFGFa1* or for *NvFGFa2*, as in [@bib77]. Planulae with the expected phenotype and without general developmental abnormalities were selected after 72 h of development at room temperature, and placed in RNAlater Solution (Ambion) for subsequent extraction of RNA. Three conditions were isolated: embryos injected with *NvFGFa1* MO, embryos injected with *NvFGFa2* MO and non-injected embryos for control. Total RNA was then extracted with the RNAqueous kit (Ambion), following the manufacturer's protocol, and temporarily stored at −80 °C, in order to accumulate the quantity necessary for the hybridization (20 µg of total RNA for each condition). The microarray analysis was performed by Nimblegen (Roche), on a custom designed 385k cDNA chip based on the annotation of the *Nematostella* genome ([@bib74]) provided by the Department of Energy Joint Genome Institute (JGI, <http://genome.jgi-psf.org/Nemve1/Nemve1.info.html>). The probes used for detection were 60mer oligos, and (when possible) 7 of them were used per target. The array hybridizations were conducted in duplicate (technical). For normalization, the raw data (.pair file) was subjected to Robust Multi-Array Analysis ([@bib41]), quantile normalization ([@bib7]), and background correction as implemented in the NimbleScan software package (Roche NimbleGen, Inc.). Data were analysed with the J-Express 2009 software ([@bib22; @bib88]). Genes were selected on the basis of expression level fold change, when comparing control to injected conditions. The ao numbers (for apical organ) in this study are based on the 198 gene list identified by fold-change, not on the list of genes confirmed by in situ hybridization. Microarray data are deposited in ArrayExpress with accession number E-MTAB-3004.
*Nematostella* in situ hybridization {#s0025}
------------------------------------
Genes of interest obtained from the microarray were confirmed by in situ hybridization, performed as previously described ([@bib77]), in order to build a working dataset of apical organ genes.
Identification and in situ hybridization of putative sea urchin orthologs {#s0030}
-------------------------------------------------------------------------
The genes that were confirmed to be expressed in the *Nematostella* apical organ were used as query sequences for an orthology search in *Strongylocentrotus purpuratus*, based on an integration of a reciprocal BLAST method ([@bib107]) and best BLAST hit with sequences obtained from the SpBase website version 2.1 (<http://www.spbase.org>) and NCBI. The putative orthologous genes were investigated in sea urchin by in situ hybridization ([@bib55]).
Pictures were taken with Nikon Eclipse E800 and a Nikon AZ100M microscope, and processed with Adobe Photoshop CS5.
Identification of possible ciliary genes {#s0035}
----------------------------------------
AO dataset genes potentially related to cilia were identified by an alignment search against an assembled database of ciliary genes, obtained from the Cildb v2.1 database ([@bib4]). The assembled database is provided in Supplementary File 1. By using a Standalone BLAST obtained from NCBI, the AO dataset was blasted against the ciliary database, using a threshold *e*-value of 1*e*^−5^. To validate the results, a randomization search was performed: 10 random datasets, the same number of genes as the AO dataset, were generated from the *Nematostella* genome and blasted against the ciliary dataset using the same parameters. Results are provided in [Supplementary Fig. 1](#s0105){ref-type="sec"}.
Results {#s0040}
=======
Analysis of microarray data {#s0045}
---------------------------
The aim of the microarray analysis was to identify genes specifically expressed or enriched in the apical organ of *Nematostella*. The microarray was based on the annotated *Nematostella* genome and was used to compare the transcription profiles of three different conditions: a wild type situation with a properly formed apical organ, and two morphant situations, one with an expanded apical organ (*NvFGFa2* MO) and one without an apical organ (*NvFGFa1* MO, see [Fig. 1](#f0005){ref-type="fig"}C). The expression data were first evaluated for quality and reliability by analysing the expression profiles of previously described apical organ genes, e.g. *NvCOE, NvFGFa1*, *NvHoxF/Anthox1* ([@bib24; @bib69; @bib77]). This step confirmed the consistent up-regulation of apical organ genes in *NvFGFa2* MO injected animals, but revealed that these genes were not reliably down-regulated in the *NvFGFa1* MO injected animals. This might be due to the relatively small size of the apical organ in wild type planulae, which likely precludes the identification of significant gene expression changes when compared to planulae lacking the apical organ. In addition, non-specific effects of the *NvFGFa1* MO cannot be excluded. Therefore, the analyses were focused on the genes enriched in the expanded apical organ condition (when compared to wild type or no-apical-organ phenotype).
The selection of genes specifically up-regulated in the *NvFGFa2* MO condition, i.e. the possible candidates for apical organ specific genes, was based on a fold change criterion. The expression profiles of known apical organ genes were used to select a cut-off value of 1.8 fold change; for sequence-specific transcription factors this value was lowered to 1.5 fold.
With this approach, a set of 198 putative apical-organ (ao) genes was recovered. Classification by KOG terms (euKaryotic Orthologous Groups), which identify a possible biological function of the gene based on sequence similarities, was used for a first evaluation of the dataset ([Table 1](#t0005){ref-type="table"}). Similar to the complete genome annotation, 49.6% of the recovered genes had no clear KOG assignment (48.1% for whole genome) and of these 26.3% were totally uncharacterized (27% for whole genome). Some KOG classes of genes are particularly over-represented in the AO dataset (at least 25% increase): (i) cell motility, (ii) nucleotide transport and metabolism, (iii) secondary metabolites biosynthesis, transport and catabolism, (iv) cytoskeleton, (v) intracellular trafficking, secretion, and vesicular transport, (vi) signal transduction mechanisms.
Even though the microarray hybridizations were performed with only two technical replicates and therefore could not be analysed statistically, the over-represented classes of genes are consistent with previous descriptions of apical organ structure, i.e. being composed of ciliated cells, rich in vesicles, and with a possible sensory and/or secretory function.
A working dataset of apical-organ specific genes {#s0050}
------------------------------------------------
To confirm the expression of the AO dataset in the apical organ region in situ hybridization was performed for 100 genes that are evolutionarily conserved as indicated by BLAST hits with scores \>50 in other metazoan genomes. This systematic analysis showed that 78 genes were exclusively or predominantly expressed in the apical organ, while the remaining genes either had no detectable expression or weak uniform expression. A list of the newly identified apical organ genes is provided in [Table 2](#t0010){ref-type="table"}, expression patterns for all 78 genes at gastrula and planula stages are shown in [Fig. S2](#s0105){ref-type="sec"}.
This confirmed dataset of 78 genes constitutes the working set of AO genes, which were used in the subsequent analyses. The recovered genes are very diverse, including signalling pathway components, transmembrane molecules, transcription factors, and many uncharacterized genes.
Identification of cilia-related genes, up-regulated in the apical organ {#s0055}
-----------------------------------------------------------------------
As described in the introduction, the major discernible feature of the *Nematostella* apical organ is the tuft of long cilia, and long cilia are also a general distinctive trait of bilaterian apical organs. While different types of cilia share a core of some structural components, they differ with respect to other structural proteins, metabolic proteins, their developmental program and signalling factors ([@bib12; @bib85; @bib91]). The molecular composition of these cilia could thus be informative for the comparison of apical organs in different taxa. However, whether apical organ cilia have molecular characteristics that distinguish them from other cilia is currently unclear.
Due to the wide array of diseases caused by defects in cilia development or function, many studies have contributed to the generation of databases of genes and proteins involved in ciliary processes ([@bib4; @bib26; @bib71]). These databases and the rich literature allowed an in-depth search for putative ciliary genes within the apical organ dataset. For this purpose a dataset was assembled from ten species in which high throughput ciliary studies have been performed, including vertebrates (*Homo sapiens, Rattus norvegicus, Mus musculus, Xenopus laevis*), protostomes (*Drosophila melanogaster, C. elegans*), protozoans (*Trypanosoma brucei,* and the ciliates *Paramecium tetraurelia, Tetrahymena thermophila)* and the green alga *Chlamydomonas reinhardtii.* The analysis of the *Nematostella* AO dataset against this multi-species ciliary database (Supplementary File 1, see also the [Materials and methods](#s0010){ref-type="sec"} section) indicated that the AO dataset is indeed enriched in ciliary genes ([Fig. S1](#s0105){ref-type="sec"}). Specifically, it identified a possible cilia-related function for 52 genes of the working dataset (see [Table 2](#t0010){ref-type="table"}), however, this does not exclude the possibility that these genes are also involved in additional, non-ciliary cellular processes.
The 52 putative cilia-related genes can be considered in several categories. The first category includes genes involved in ciliogenesis. The forkhead domain containing gene *NvFoxJ1(ao194)* is among the few transcription factors that were recovered in the analysis. *FoxJ1* has been characterized as a key regulator of the development of motile cilia in vertebrates and this function has been suggested to be conserved across metazoans ([@bib93; @bib101]). The onset of *NvFoxJ1* expression at the aboral pole of *Nematostella* coincides with the differentiation of the apical organ at early planula stage ([@bib86]). The assembly and maintenance of cilia depends on conserved structural proteins and an intraflagellar transport (IFT) system. Genes falling into this category are *Nvβ-tubulin (ao162)*, *NvTektin (ao90)*, *NvPACRG*-*like* (*Parkin-co-regulared gene (ao101, ao137;* ([@bib19]), and the motor protein components *NvDynein heavy chain-like (ao35)* and *NvDynein light chain-like (ao77)* ([Fig. 2](#f0010){ref-type="fig"}A--D). Kinesins are microtubule-binding motor proteins that have diverse cellular functions including intraflagellar transport ([@bib37; @bib100]). Two *Nematostella* kinesins, *NvKif9-like* (*ao113*) and *NvKif16-like (ao74)*, are expressed in the apical organ; however, members of these particular kinesin families have not been implicated in intraflagellar transport (see the [Discussion](#s0075){ref-type="sec"} section).
Nima-related kinases 1 and 8 (NEK1 and 8) are associated with the cilium in mammals and in *Chlamydomonas* and they have been implicated in vertebrate ciliopathies ([@bib8; @bib49; @bib98]). Their exact function is not known and it is unclear whether they are restricted to particular subsets of cilia. In *Nematostella*, a *NEK8/9-like* (*ao18*) gene is expressed in the apical organ from early planula stage on ([Fig. 2](#f0010){ref-type="fig"}E).
Cilia constitute a separate compartment of the cell, distinct from the rest of the cytoplasm ([@bib59]). However, they require a constant provision of energy for mobility and to maintain the activity of the intraflagellar transport machinery ([@bib79]). This high demand for energy is supported by metabolic enzymes located in the cilium, as demonstrated in mammals and *Chlamydomonas* ([@bib56; @bib58]). Several genes involved in energy provision were recovered in the screen. For example, adenylate kinase (ADK) regenerates ATP from ADP and has been shown to localize and function in cilia in various organisms ([@bib23; @bib61]). Two *Nematostella ADK-like* genes (*NvADK-like 1* and *2, ao49* and *ao114*) were found to be expressed in the apical organ ([Fig. 2](#f0010){ref-type="fig"}F and [Fig. S2](#s0105){ref-type="sec"}PA)). There are also genes coding for enzymes involved in metabolic pathways, in particular enzymes producing NADH+H like malate dehydrogenase (*NvMDH, ao156*) and aldehyde dehydrogenase (*ao171*) ([Fig. 2](#f0010){ref-type="fig"}G).
Several studies have provided evidence that not only immotile, but also motile cilia can have sensory functions, as for example in the case of the motile cilia of the respiratory epithelium of mammals, which sense mechanical and chemical cues ([@bib42; @bib82]). Similarly, our screen revealed the presence of genes with a possible function in the sensing and transduction of signals. A direct connection between the apical organ of *Nematostella* and the nervous system has not been shown ([@bib62]), but two genes coding for nicotinic acetylcholine receptors (*ao19* and *ao145*) are expressed in the apical organ cells ([Fig. 2](#f0010){ref-type="fig"}H and [Fig. S2](#s0105){ref-type="sec"}) and may mediate signalling between the apical organ and the underlying plexus of neurites. Another gene potentially involved in a sensory function is *NvTRPV-like (ao151;* [Fig. 2](#f0010){ref-type="fig"}I**)**, which is related to calcium channels of the Transient Receptor Potential (TRP) family. Members of this group of ion channels are involved in mechano-, chemo- and thermo- sensation ([@bib47]) and interestingly, a *TRPV* gene is expressed in the apical tuft cells of *P. dumerilii* ([@bib51]).
Other genes that are predicted to localize to the plasma membrane or to the extracellular matrix (ECM) and are thus potentially involved in cell--cell or cell-ECM interactions include the transmembrane protein *NvTetraspanin33-like* (*ao130,* [Fig. 2](#f0010){ref-type="fig"}J), the astacin metalloprotease *NvMeprin-like (ao146,* [Fig. 2](#f0010){ref-type="fig"}K) ([@bib89]), and the ECM protein *NvSpondin1-like* (*ao147,* [Fig. S2](#s0105){ref-type="sec"}AB) ([@bib33]).
Nearly half of the potentially cilia-related apical organ genes are poorly characterized. In some cases a specific domain is recognizable, for example in the coiled-coil domain containing genes *NvCCDC74-like (ao60)*, *NvCCDC81-like* (*ao155*) and *NvCCDC121-like (ao95)* or in three Leucine-Rich Repeat containing genes (*ao43*, *ao46*, *ao62*; [Fig. 2](#f0010){ref-type="fig"}L--N). However, several other genes display sequence conservation but lack recognizable domains ([Table 2](#t0010){ref-type="table"}).
"Non-ciliary" genes {#s0060}
-------------------
As for the cilia-related genes, about half of the apical organ genes that are not found in the ciliary databases are evolutionarily conserved, but uncharacterized.
Regarding the characterized genes, the non-ciliary dataset contained two genes related to the Wnt signalling pathway, which determines the site of gastrulation in bilaterians and cnidarians and is involved in the patterning of the anterior--posterior axis of bilaterians. Both the Wnt receptor *NvFrizzled5/8* (*ao97*) and the putative Wnt antagonist secreted Frizzled-Related protein (*NvSFRP1, ao63*) are expressed in a broad aboral domain at gastrulation and at highest levels in the apical organ of the planula larvae ([Fig. 3](#f0015){ref-type="fig"}A--D**)**, and ([@bib45])). While the function of the two Fibroblast growth factor genes *NvFGFa1* and *NvFGFa2* in apical organ development has been described ([@bib77; @bib86]), there are 11 *Nematostella* FGFs for which expression data are not yet available. We identified *NvFGF1e (ao190)* ([@bib52]) as an additional apical organ related FGF. *NvFGF1e* is expressed at the aboral pole from gastrula stage on and its expression at planula stage is restricted to a subset of apical organ cells ([Fig. 3](#f0015){ref-type="fig"}E and F).
Among the metabolic genes, we recovered a gene for the catabolism of the amino acid taurine, a homolog of the *TauD/Tfd*A (*ao110*) gene of *E. coli* ([Fig. 3](#f0015){ref-type="fig"}G and H). Taurine is involved in a wide array of biological functions such as osmoregulation, antioxidation, modulation of neurotransmitters, stimulation of glycolysis, and maintenance of photoreceptors ([@bib39]). In particular, in marine invertebrates taurine is important for the development and settlement of larvae, and in cnidarian larvae it is probably functioning as an inhibitor of metamorphosis ([@bib5]), but see also ([@bib108]). The expression of the *NvTauD* gene in *Nematostella* could therefore support a role for the apical organ in metamorphosis, as previously suggested. Indeed, *Nematostella* larvae lacking an apical organ do not enter metamorphosis, as demonstrated by the *NvFGFa1* MO injected animals ([@bib77]).
Genes expressed in the apical organ and in specific cell-types {#s0065}
--------------------------------------------------------------
Some of the identified apical organ genes displayed additional expression that appeared to be restricted to particular cell-types. The O-linked-mannose beta-1,2-*N*-acetylglucosaminyltransferase gene *NvPOMGnT1-like* (*ao51*) is expressed in scattered cells in the ectoderm and in the pharynx ([Fig. 4](#f0020){ref-type="fig"}A). *NvCellulase*-positive cells (*ao132*) are enriched in a broad domain in the aboral ectoderm and in the pharynx ([Fig. 4](#f0020){ref-type="fig"}B). Compared to the often spindle-shaped *NvPOMGnT1-like* expressing cells, the *NvCellulase*-positive cells appear more compact. On the basis of the predicted gene function and the distribution of these cells, we assume that the *NvCellulase*-positive cells are a particular type of gland cells (gland cells with translucent vesicles ([@bib62])). One of the uncharacterized apical organ genes (*ao81*) is expressed in only a few ectodermal cells in the aboral half of the larvae, whereas the coiled-coil domain containing gene *NvCCDC81 (ao155)* can be detected in individual cells throughout the ectoderm ([Fig. 4](#f0020){ref-type="fig"}C and D).
A dataset for evolutionary comparisons: Sea urchin {#s0070}
--------------------------------------------------
The main goal of this study was to establish an improved basis for the comparison of cnidarian and bilaterian apical organs, which could allow a better understanding of their evolutionary histories.
As a test case for an initial comparative study we chose the purple sea urchin *S. purpuratus*, as it is an established model system for developmental studies with a sequenced genome and with an apical organ at embryonic stages. The development of the sea urchin apical territory has been addressed by different studies (reviewed in [@bib2]), and a recent paper showed similarities with the determination of the apical territory in *Nematostella* ([@bib86]). The sea urchin larva has two prominent ectodermal structures with long cilia: the apical organ and the ciliary bands. The presence of a second structure with long cilia could help to distinguish between genes that are generically involved in cilia formation and whose expression level simply reflects the length of the cilia, and genes that are specific to the apical organ tuft.
The orthology search identified 73 putative sea urchin genes homologous to the genes in the *Nematostella* AO dataset, of which several have previously been shown to be expressed in the apical territory (e.g. *frizzled 5/8, SFRP1, beta tubulin, foxJ1, Tektin* ([@bib14]; [@bib21]; [@bib40]; [@bib73]; [@bib96])). For an initial in situ hybridisation analysis 18 genes were selected for which no spatial expression data were available. Twelve of the 18 genes showed specific or enriched apical organ expression ([Fig. 5](#f0025){ref-type="fig"} and [Fig. S3](#s0105){ref-type="sec"}) sometime during the developments of the sea urchin larva, while for the others no expression was detected (with the exception of one gene expressed in the gut, *Aldh2*). Of the genes with apical organ expression some were specifically expressed in the apical organ area (e.g. *PACRG*, [Fig. 5](#f0025){ref-type="fig"}C and D), while others had later additional domains in the ciliary band and few in the ciliated gut (e.g. *Annexin*, [Fig. 5](#f0025){ref-type="fig"}I and J). The gene for an acetylcholine receptor had a patchy ectodermal expression domain, with an enrichment of cells in the apical organ ([Fig. S3](#s0105){ref-type="sec"}).
Discussion {#s0075}
==========
In this study we have identified a set of genes that are expressed in the apical organ of the anthozoan *N. vectensis*, providing a molecular signature for the apical organ cells. The recovered genes are very diverse and constitute a new and versatile tool for developmental, functional and comparative studies of apical organs across metazoans. However, only seven out of 15 previously described apical organ genes were recovered with the cut-off values chosen for the initial analysis (the 198 gene list, see [Table S1](#s0105){ref-type="sec"}) and our dataset therefore likely represents an underestimation of all apical organ specific or enriched genes.
A signature for "apical organ" cells {#s0080}
------------------------------------
Though a bundle of long cilia is considered the hallmark of apical organs both in bilaterian and cnidarian larvae, these aboral cilia can be more or less prominent in different larvae, in particular among cnidarians (e.g. [@bib110]). It is also not clear whether the long cilia are a prerequisite for the function of apical organs, or if functionally equivalent apical organ-like sensory structures with short cilia exist. For example, the morphology and cell type composition of the aboral pole of planulae of the scyphozoan cnidarian *Aurelia* is highly similar to the apical organ region of anthozoan planulae, but it lacks the prominent tuft of long cilia ([@bib11; @bib62; @bib63; @bib114]). Interestingly, the apical organ-like region of *Aurelia* contains taurine immunoreactive sensory cells ([@bib63]) and *NvTauD*, a gene involved in taurine catabolism, is expressed in the apical organ region of *Nematostella* (this study). The essential role played by cells localized in the aboral part of the larvae has been described also in several coral species, such as *Pocillopora damicornis,* where each sensory cell bears a single cilium surrounded by a collar of microvilli ([@bib95; @bib99]), or in *Acropora* species, in which the aboral pole has been shown to have a role in the recruitment to the substrate ([@bib68]). Given that the settlement of coral larvae is a highly specialized process with significant impact on reef communities (see for example [@bib57]), a deeper understanding of the molecular processes and of the cell types involved is of ecological importance. Our dataset could therefore help in identifying "apical organ" cell types, even in those cases in which a prototypical apical organ, with distinctive long cilia, is missing.
Molecular characteristics of apical tuft cilia {#s0085}
----------------------------------------------
An open question about the long apical organ cilia is whether they are molecularly distinct from the shorter cilia that cover other parts of the larval epidermis. Cilia are highly conserved structures of eukaryotic cells, with fundamental roles in locomotion, movement of fluid or in the sensing of external cues (for a recent review see [@bib12]). The core element of a cilium is the axoneme, consisting of nine doublets of microtubules emerging from the basal body, and surrounded by a membrane in continuity with the plasma membrane of the cell. Cilia are usually classified in two types according to their axonemal architecture: the 9+0 type cilia are sensory and immotile, while the 9+2 type cilia (with two central microtubule singlets) are considered as motile. However, there are many exceptions to this classification, and recent studies have suggested that all cilia could have a sensory role ([@bib6]). The majority of the conserved genes that we identified as enriched or specific to the apical organ of *Nematostella* are associated with the development, structure or function of cilia, although they can have additional, non-ciliary functions. While several classes of structural and motor proteins are required in all cilia, the evolutionary diversification of some of these components makes them informative for the comparison of apical organ cilia and other cilia. For example, at least eleven distinct families of Kinesin motor proteins were present in the last common ancestor of eukaryotes and 45 kinesin genes have been identified in mammals ([@bib54; @bib109]). Kinesins are involved in several intracellular processes and this is partially reflected in specific expression patterns ([@bib36; @bib37; @bib53]). Accordingly, the expression of *NvKif9-like* may reflect a specific function in apical organ cilia. *NvKif9-like* belongs to the Kinesin 9 family which is poorly studied, but the *Chlamydomonas* KIF9 protein Klp1 is unusual in that it localizes to the central microtubules of the cilium and is required for ciliary motility ([@bib112]). *Kif16* genes have not been related to ciliary functions. Instead, *Kif16b* has been shown to regulate endosome trafficking, including the transport of the FGF receptor to the cell membrane ([@bib38; @bib97]), whereas the Kif16a protein localizes to centrosomes and has been implicated in mitotic spindle formation ([@bib94]). Thus, the expression of *NvKif16-like* in the apical organ may not be related to the development or function of cilia.
A *caveat* for the interpretation of our data is that in situ hybridization cannot rule out a uniformly low level of expression of the identified genes in the ciliated ectoderm. This means that the strong expression in the apical organ cells might only reflect the length of the apical organ cilia and a correspondingly higher demand for the gene products in these cells. The analysis of larvae with ciliary bands will be particularly informative in addressing this problem. If the expression level of a particular gene is relative to the length of the cilia, then ciliary bands would be expected to display stronger signal than the remaining ectoderm. Indeed, while expression of the *Park2 co-regulated gene* (*PACRG*) in sea urchin is restricted to the apical pole, another homolog of a *Nematostella* apical organ gene, *AnnexinA*, is expressed in the apical organ and in the ciliary bands (see also below).
Apical organ-specific developmental regulators {#s0090}
----------------------------------------------
Transcription factors and signal transduction pathways regulate the development of body regions, organs and cell types and are frequently employed to address questions of homology. In this study we identified relatively few developmental regulators, probably due to the often comparably low expression level of these genes, which can hinder their identification in microarray experiments with low replicate numbers. Among the developmental genes that we found were two Wnt pathway components, the Wnt receptor *NvFrizzled5/8*, a putative positive regulator of the pathway, and *NvSFRP1*, a putative negative regulator. Both genes have been shown to be expressed at the anterior pole of some bilaterian larvae or embryos, e.g. in sea urchin, in the hemichordate *Saccoglossus kowalevskii* and the annelid *P. dumerilii* ([@bib15; @bib17; @bib40; @bib51; @bib70]). In bilaterians and cnidarians, Wnt signalling is required for the determination of the gastrulation site and for the patterning of the apical-blastoporal axis. However, the regulation by NvFGF signalling suggests that in addition to an early patterning function, *NvFrizzled5/8* and *NvSFRP1* may have separate, later functions in the development of the apical organ.
Of the transcription factors that we identified, *NvFoxJ1* is of particular interest. *FoxJ1* is a key regulator of motile cilia formation in vertebrates and it has been suggested to be a regulator of this process across the Metazoa ([@bib12; @bib90; @bib93; @bib101; @bib113]). In vertebrates, *foxJ1* has been shown to be regulated by FGF and Wnt signalling ([@bib10; @bib64]), a situation which might also be the case in *Nematostella* (see also [@bib86]). The strong expression of *NvFoxJ1* in the apical organ together with the expression of genes that have been associated with the high demand for energy of motile cilia (*NvADK-like, NvMDH*) indicates that the long cilia of the *Nematostella* apical organ are motile cilia. Functional studies will have to determine whether *NvFoxJ1* is specifically required for the development of the apical organ cilia or whether the motile cilia of the body surface also depend on the function of this gene.
Overall, functional studies of the developmental regulators identified in this study will help to characterize the developmental programs of apical organ cells in *Nematostella* and to compare it to other organisms, but their relatively small number does not allow a comprehensive understanding of apical organ formation and evolution.
Comparison to the apical organ of sea urchins {#s0095}
---------------------------------------------
Our experimental approach aimed at the identification of apical organ genes as a versatile tool for studying apical organs in different animal groups. The analysis of a subset of the identified genes in the developing larvae of the sea urchin *S. purpuratus* supports the suitability of this approach. In particular, we could discriminate between genes expressed solely in the apical organ (*Sp-Ajpx1*), or with additional domains in the ciliary bands (*Sp-Anxa7*) or even specific to particular ciliated territories, such in the case of the *Aldh* gene, which is expressed in the apical organ and the endoderm of *Nematostella* and in the mid-portion of the gut of sea urchin larvae ([Fig. S2](#s0105){ref-type="sec"}QB and [S3](#s0105){ref-type="sec"}). Furthermore, the usability of the dataset to describe conserved aspects of previously uncharacterized genes is demonstrated by the expression of Sp-005437 ([Fig. 5](#f0025){ref-type="fig"}A and B; most similar to *Nematostella ao154*) in the apical domain of the sea urchin. This gene encodes a short predicted protein (83 and 68 amino acids in sea urchin and *Nematostella*, respectively) that lacks annotated domains and is also conserved in the hemichordate *S. kowalewskii*.
In conclusion, we have identified a set of 78 new apical organ genes that are mainly related to structural and physiological aspects of the apical organ cells. These genes serve as an entry point for a better understanding of the structure and function of the apical organ of *Nematostella* and they enable comparisons of the molecular composition of apical organs within Cnidaria, and in different bilaterian taxa, as exemplified by the expression analysis of a subset of these genes in the sea urchin. Therefore, the results of this study provide additional characters to address the homology of apical organ cell types, but they also serve as a generic tool to identify aboral cells that may play a role in the sensing of environmental cues.
Appendix A. Supplementary materials {#s0105}
===================================
Supplementary data: **Table S1: A subset of previously identified apical organ genes is recovered in the microarray** Note that in Marlow et al., 2009, the sequence annotated as *NvNCAM2* is producing the expression pattern that is described as *NvNCAM3* (and vice versa).
Fig. S1**The AO gene set is enriched for ciliary genes.** The percentage of BLAST hits against the custom ciliary database (Supplementary File 1) is shown for the AO dataset (red square) and for 10 random *Nematostella* datasets. The values of the random datasets are lower than the value for the AO dataset.
Fig. S2-1**Expression patterns of*****Nematostella*****apical organ genes.** In situ hybridizations at gastrula and planula stages for the genes identified in this study. All images are lateral views with aboral pole to the left. The insets are: (D) aboral view, (DA, AB, VA) higher magnifications. *Note*: some pictures are also shown in the main figures. Scale bar=100 µm
Fig. S2-2
Fig. S2-3
Fig. S2-4
Fig. S3-1**Expression patterns of putative sea urchin apical organ genes.** In situ hybridizations for the sea urchin genes most similar to the indicated *Nematostella* genes (by reciprocal BLAST). For each gene from left to right: lateral views of mid-gastrula, late gastrula (with apical pole to the top) and pluteus (oral side indicated by asterisk) stages, respectively. For SPU_007284 an apical pole view is shown in the middle.
Fig. S3-3
We thank Kjell Petersen and Yvan Strahm (Computational Biology Unit, Uni Computing, Bergen) for processing of the microarray raw data, the Norwegian Microarray Consortium for access to and training in the J-Express software; Libero Petrone (University College of London) for help in the laboratory; Uli Technau (University of Vienna) for helpful discussions and Gemma Richards and Lucas Leclère for critical reading of the manuscript. We thank also Bård Giezendanner, Tessa Bargmann, Alina Rey and Monica Martinussen for taking care of the *Nematostella* culture. Work in F.R.'s lab is supported by the Sars Centre core budget, C.S.'s work in the P.O's lab was supported by a grant from the Meltzerfond. P.O.'s laboratory is supported by University College London and a Human Frontier Science Program Young Investigator Grant.
{#f0005}
{#f0010}
{#f0015}
{#f0020}
{#ir0005}; the corresponding *Nematostella* gene is also reported.](gr5){#f0025}
######
**KOG class analysis of putative apical organ genes**. Relative abundances of the various KOG classes of genes, in the whole genome and in the condition with an expanded apical organ (*NvFGFa2* MO). In both cases, nearly half of the genes are poorly characterized ("*Function unknown*", "*General function, prediction only*" and "*Uncharacterized- No KOG assigned*"). The last column indicates with the signs±the direction of the change, i.e. if a class is more or less abundant in the *NvFGFa2* MO injected when compared to the wild type (whole genome). The categories particularly overrepresented in the *NvFGFa2* MO condition are highlighted in red. The numbers displayed are percentages.
**KOG class** ***Nematostella*****genome (v 1.0)** **Genes upregulated in*****NvFGFa2*****MO** **Change compared to genome**
--------------------------------------------------------------- -------------------------------------- --------------------------------------------- -------------------------------
Cell motility 0.2 1.5 \+
Defence mechanisms 0.3 0.0 −
Nuclear structure 0.3 0.5 \+
Cell wall/membrane/envelope biogenesis 0.5 0 −
Coenzyme transport and metabolism 0.6 0 −
Nucleotide transport and metabolism 0.7 1.5 \+
Cell cycle control, cell division, chromosome partitioning 1.0 0.5 −
Secondary metabolites biosynthesis, transport and catabolism 1.3 2.0 \+
Replication, recombination and repair 1.3 0.5 −
Energy production and conversion 1.4 1.0 −
Chromatin structure and dynamics 1.4 1.0 −
Lipid transport and metabolism 1.7 0.0 −
Translation, ribosomal structure and biogenesis 1.8 0.5 −
RNA processing and modification 2.3 2.0 −
Carbohydrate transport and metabolism 2.3 1.5 −
Amino acid transport and metabolism 2.6 2.0 −
Extracellular structures 2.6 1.5 −
Inorganic ion transport and metabolism 2.7 2.5 −
Cytoskeleton 3.0 6.1 \+
Intracellular trafficking, secretion, and vesicular transport 3.2 4.0 \+
Transcription 3.6 3.5 −
Posttranslational modification, protein turnover, chaperones 6.0 4.0 −
Signal transduction mechanisms 11.1 14.1 \+
Function unknown 5.5 7.1 \+
General function prediction only 15.6 16.2 \+
Uncharacterized (No KOG assigned) 27.0 26.3 −
**Total number of genes** **25,911** **198**
######
**List of the genes with confirmed apical organ expression**. This gene list constitutes the AO working list. Domain assignment was performed with SMART, including Pfam domains. For conserved, uncharacterized genes the best hit in the human genome is indicated. Note that some of these genes have no homologs in humans. Classification as ciliary or non-ciliary is based on the dataset provided in Supplementary File 1. The last two columns report the results of the search for putative orthologs in sea urchin. The first symbol (+ or −) indicates whether a putative ortholog was identified, the second whether it is expressed at the aboral pole (+) or not (−), (?) indicates no data avalaible.
**No in this study** **Gene ID** **KOG** **KOG class** **Domains (SMART+Pfam)** **BLASTP result (provisional gene name)** **Cilia-related (BLAST hit)** **In sea urchin**
---------------------- ------------- ---------- --------------------------------------------------------------- -------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------- ------------------------------- ------------------- ------------
**2** 31543 KOG0167 Function unknown Armadillo repeats Arm-repeat containing 4, yeast vacuolar protein 8 X +/? SPU_007521
**6** 224064 *No KOG* \- - - - - - - - *None* Sperm associated antigen 17 X +/? SPU_013103
**13** 53190 KOG3627 Amino acid transport and metabolism MAM (Meprin, A-5 protein, Mu) MAM domain containing glycosylphosphatidylinositol anchor 1 +/? SPU_015727
**16** 172654 *No KOG* \- - - - - - - - THEG (Testicular Haploid Expressed Gene) Testicular haploid expressed gene product **+/+** SPU_011786
**18** 89190 KOG1426 Function unknown RCC1 (Regulator of Chromosome Condensation) NEK9 X +/? SPU_019063
**19** 110265 KOG3645 Signal transduction mechanisms Neurotransmitter-gated ion-channel Acetylcholine receptor, alpha subunit X +/? SPU_001774
**20** 240082 KOG0613 Cytoskeleton Ig-like, FN III NvNCAM2 X +/? SPU_005613
**29** 147392 KOG3627 Amino acid transport and metabolism MAM (Meprin, A-5 protein, Mu) MAM domain protein *(potentially part of ao146)* +/? SPU_004114
**35** 224165 KOG3595 Cytoskeleton *None* Dynein heavy chain, axonemal X **+/+** SPU_030227
**37** 96324 KOG0819 Intracellular trafficking, secretion, and vesicular transport Annexin Annexin A X **+/+** SPU_019139
**43** 24766 KOG4308 Function unknown LRR LRR protein X +/? SPU_005282
**44** 96179 KOG0819 Intracellular trafficking, secretion, and vesicular transport Annexin Annexin A X **+/+** SPU_019139
**46** 212934 KOG4308 Function unknown LRR (Leucine-Rich Repeat), EF hand *Conserved, uncharacterized* (C14orf166B) X +/? SPU_005282
**48** 233391 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* (C1orf177) +/? SPU_026146
**49** 98446 KOG3078 Nucleotide transport and metabolism Adenylate kinase (adk) Adenylate kinase-like protein +/? SPU_010767
**51** 181253 KOG1413 Carbohydrate transport and metabolism *N*-acetylglucosaminyltransferase *O*-linked-mannose beta-1,2-*N*-acetylglucosaminyltransferase +/? SPU_008219
**54** 163356 KOG0306 RNA processing and modification WD40, coiled-coil WD repeat domain 49 X +/? SPU_026011
**60** 237587 KOG0388 Replication, recombination and repair Coiled-coil Coiled-coil 74 X **+/+** SPU_018584
**62** 244599 KOG0531 Signal transduction mechanisms LRR (Leucine-Rich Repeat) Leucine rich repeat containing 48, phosphatase X +/? SPU_006598
**63** 200285 KOG3577 Signal transduction mechanisms Frizzled, C345C sFRP1 +/+ SPU_011271
**68** 237703 KOG4193 Signal transduction mechanisms GPCR proteolytic site, TM Latrophilin +/? SPU_012362
**69** 235461 KOG1399 Secondary metabolites biosynthesis, transport and catabolism Flavin-containing monooxygenase Flavin-containing monooxygenase X +/? SPU_007044
**71** 242796 *No KOG* \- - - - - - - - *None* Conserved, uncharacterized (C7orf31) +/? SPU_009640
**74** 30871 KOG0245 Cytoskeleton Kinesin motor catalytic domain, FHA Kinesin-like protein KIF16B X +/? SPU_026237
**75** 86487 KOG4713 Signal transduction mechanisms CDK2AP CDK2-associated protein 1 +/? SPU_004653
**77** 237378 KOG4115 Cell motility RobLC7 domain Dynein light chain roadblock-type 2 X +/? SPU_003137
**78** 237330 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* +/? SPU_028435
**79** 59658 *No KOG* \- - - - - - - - DUF4542 *Conserved, uncharacterized* (C17orf98) +/? SPU_004558
**80** 209931 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* (C9orf135) X +/? SPU_017778
**81** 239479 *No KOG* \- - - - - - - - DUF3504 *Conserved, uncharacterized* (KIAA1958) +/? SPU_021568
**84** 205233 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* (C9orf116) X +/? SPU_004486
**85** 120202 KOG0667 General function prediction only S/T protein kinase catalytic domain Dual-specificity tyrosine-(*Y*)-phosphorylation regulated kinase DYRK4 X +/? SPU_012899
**90** 195162 *No KOG* \- - - - - - - - Coiled-coil Coiled-coil domain containing 105, tektin X +/? SPU_002424
**92** 11327 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* (FLJ43738) X **+/+** SPU_005267
**95** 93809 KOG0161 Cytoskeleton Coiled-coil Coiled-coil 121 X +/? SPU_026895
**97** 183962 KOG3577 Signal transduction mechanisms Frizzled, FRI (CRD) Frizzled 5/8 +/+ SPU_022916
**101** 113661 KOG3961 Function unknown ParcG PARK2 co-regulated-like X +/? SPU_012917
**103** 93943 KOG0161 Cytoskeleton Coiled-coil, IQ motif (calmodulin binding) IQ motif containing D X +/? SPU_002424
**104** 218953 KOG1909 Signal transduction mechanisms LRR Ran GTPase-activating protein 1 +/? SPU_004276
**110** 242938 *No KOG* \- - - - - - - - TauD Taurine catabolism dioxygenase TauD/TfdA **−**
**113** 234547 KOG4280 Cytoskeleton Kinesin motor domain Kinesin-like protein KIF9 X +/? SPU_000875
**114** 232308 KOG3079 Nucleotide transport and metabolism *None* Adenylate kinase 5 or 8 X +/? SPU_019553
**117** 223606 *No KOG* \- - - - - - - - IQ motif, coiled-coil Spermatogenesis-associated protein 17 +/? SPU_023743
**124** 218233 KOG0274 General function prediction only F-box, WD40 F-box and WD repeat domain containing 7 X +/? SPU_015976
**125** 164988 KOG4682 General function prediction only *None* BTB (POZ) domain containing 16 +/? SPU_015356
**130** 171968 KOG3882 General function prediction only Tetraspanin Tetraspanin (33) X +/? SPU_027747
**132** 83869 *No KOG* \- - - - - - - - Glycoside hydroxylase Endoglucanase +/? SPU_021602
**137** 182272 KOG3961 Function unknown ParcG PARK2 co-regulated X **+/+** SPU_004619
**139** 160170 *No KOG* \- - - - - - - - Coiled-coil, THEG Testicular haploid expressed gene protein-like **+/+** SPU_026963
**142** 162410 *No KOG* \- - - - - - - - DUF3695 *Conserved, uncharacterized* (C1orf194) X **+/+** SPU_013076
**145** 199721 KOG3645 Signal transduction mechanisms Neurotransmitter-gated ion channel ligand binding domain, neurotransmitter-gated ion channel transmembrane domain (Pfam) Nicotinic Acetylcholine Receptor alpha **+/+** SPU_001774
**146** 131533 KOG3714 Posttranslational modification, protein turnover, chaperones Zn dependent metalloprotease, MAM Meprin X **+/+** SPU_004114
**147** 79471 KOG3539 Extracellular structures Spondin, TSP1 Spondin-1 X +/? SPU_009594
**148** 123439 *No KOG* \- - - - - - - - Scavenger Receptor Cysteine-rich Galectin 3 binding protein X −
**149** 243308 *No KOG* \- - - - - - - - Tubulin tyrosine ligase Tubulin tyrosine ligase-like 9 X +/? SPU_000277
**151** 81127 KOG3676 Inorganic ion transport and metabolism Ankyrin, PKD channel TRPV channel X **+/+** SPU_007504
**153** 240906 *No KOG* \- - - - - - - - Coiled-coil Stathmin 4, Nucleolar Protein 9 +/? SPU_008203
**154** 81173 *No KOG* \- - - - - - - - *None* *Conserved, uncharacterized* **+/+** SPU_005437
**155** 168814 KOG4364 Chromatin structure and dynamics Coiled-coil Coiled-coil domain containing 81 X +/? SPU_021664
**156** 90973 KOG1496 Energy production and conversion Coiled-coil, lactate/malate dyhdrogenase NAD binding and C-terminal domains Malate dehydrogenase X **+/+** SPU_015928
**159** 117995 KOG0032 Signal transduction mechanisms S/T kinase catalytic domain DAP kinase-related apoptosis-inducing protein kinase 1 X +/? SPU_028649
**161** 240545 *No KOG* \- - - - - - - - KIAA1430 *Conserved, uncharacterized* (C17orf105) X +/? SPU_010239
**162** 245773 KOG1375 Cytoskeleton Tubulin GTPase, Tubulin C-terminal Beta tubulin X **+/+** SPU_000062
**165** 238199 KOG4415 Function unknown Coiled-coil *Conserved, uncharacterized* +/? SPU_011316
**166** 245865 KOG3508 General function prediction only PKC conserved region 2, RasGAP, DUF 3498 Disabled homolog 2-interacting protein X −
**167** 34056 KOG0819 Intracellular trafficking, secretion, and vesicular transport Annexin Annexin A X **+/+** SPU_019139
**168** 41471 KOG0200 Signal transduction mechanisms Tyrosine kinase FGF receptor X −
**169** 208307 KOG0531 Signal transduction mechanisms LRR Protein phosphatase 1 regulatory subunit 7 X +/? SPU_012637
**171** 245626 KOG2450 Energy production and conversion Aldehyde dyhydrogenase ALDH, ALDH1b X +/− SPU_007284
**175** 143747 KOG3940 Function unknown C2HC Zn finger Zn finger C2HC domain containing protein 1 C X +/? SPU_007461
**180** 3074 KOG4308 Function unknown LRR *Conserved, uncharacterized* (C14orf166B); *potentially part of ao46* X +/? SPU_005282
**181** 245069 *No KOG* \- - - - - - - - NADH dehydrogenase, FAD-containing subunit *Conserved, uncharacterized* (C20orf26) X +/? SPU_018537
**182** 208725 *No KOG* \- - - - - - - - DUF4562 *Conserved, uncharacterized* (C4orf45) −
**189** 94003 KOG0490 General function prediction only HOX NVHD146-paired class homeobox protein OR Q50-6 \[*Nematostella vectensis*\] X −
**190** 212596 KOG3885 Signal transduction mechanisms FGF Fibroblast growth factor 1E \[*Nematostella vectensis*\] **+/+** SPU_006242
**193** 165815 KOG3585 Transcription T-Box Tbx4/5 protein \[*Podocoryne carnea*\] X **+/+** SPU_023386
**194** 65438 KOG2294 Transcription Forkhead Forkhead box J1b \[*Danio rerio*\] X **+/+** SPU_027969
**195** 153628 KOG0490 General function prediction only HOX K50-5 \[*Nematostella vectensis*\]; DMBXf-paired class homeobox protein \[*Nematostella vectensis*\], X −
[^1]: Present address: Developmental Biology Unit, Observatoire Oceanologique, 06234 Villefranche-sur-mer, France.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction
===============
In measuring γ-ray spectra it is frequently necessary to remove the effects of the resolution of the detector from an observed pulse height distribution. This is known as "unfolding", "unscrambling", or "unsmearing". To do this a matrix representing the response of the detector must be found. Let the incident spectrum be denoted by an *m* dimensional vector *N*: $$N = {\begin{pmatrix}
n_{1} \\
n_{2} \\
\vdots \\
n_{m} \\
\end{pmatrix}.}$$The response may be represented by an *m×m* matrix *R.* The detected pulse height distribution, *P*, is then given by $$P_{i} = {\sum\limits_{j}{R_{ij}N_{j}.}}$$Unfolding is the name given to the process of finding *N* such that $$N_{j} = {\sum\limits_{j}R_{ij}^{- 1}}P_{i},$$where *R~if~*^−1^ is the inverse to the matrix *R~tj~*.
It is frequently undesirable to obtain a solution *N~j~* by inverting the response function matrix. Usually the response function matrix is a very large square matrix. In this experiment one form of the matrix was 700×700. The inversion of such a matrix would be a formidable task, even when utilizing computer techniques.
For this reason, iterative approximations to solutions have been developed by Scofield \[[@b2-jresv68an4p401_a1b]\] and by Skarsgard, Johns, and Green \[[@b3-jresv68an4p401_a1b]\]. An iterative technique similar to that described by the latter has been developed independently in this laboratory. Convergence criteria for this technique have been discussed by Geiringer \[[@b1-jresv68an4p401_a1b]\]. In applying the technique, empirical evidence has been obtained for the validity of solutions obtained by this method. This evidence is discussed below. In addition, the propagation of error in the unfolding process is investigated in detail.
2. Iterative Solution
=====================
2.1. Procedure
--------------
[Equation (1)](#fd2-jresv68an4p401_a1b){ref-type="disp-formula"} may be written in matrix form as $$P = RN.$$Assume *N= U*~0~. Then this initial estimate will give $$P_{0} = RU_{0}.$$A measure of the closeness with which *U*~0~ represents the true *N* is given by the difference $$\Delta_{0} = P - P_{0}.$$*U*~0~ may be corrected to form $$U_{1} = U_{0} + \Delta_{0},$$and the new correction $$\Delta_{1} = P - RU_{1}$$is found. For the *n^th^* iteration $$\left. \begin{array}{l}
{U_{n} = U_{n - 1} + \Delta_{n - 1}} \\
{\Delta_{n} = P - RU_{n}} \\
{U_{n + 1} = U_{r} + \Delta_{n}} \\
\end{array} \right\}.$$
It has been found for the present work that it is satisfactory to use *P* itself as the initial estimate *U*~0~. The technique has been used primarily in unfolding pulse height distributions obtained with the NBS Two Crystal Pair Spectrometer. Details of the detector and its response are described by Ziegler, Wyckoff, and Koch \[[@b4-jresv68an4p401_a1b]\].
Various methods for arresting the iterative procedure may be used. In this work the data were unfolded using a predetermined number of iterations.
2.2. The Response Function Matrix
---------------------------------
This section will discuss the problem of finding a matrix representation for the assumed analytic form of the response function. The response at pulse height *ϵ* due to one incident photon of energy *k*~0~ may be written \[[@b4-jresv68an4p401_a1b]\] $$\begin{array}{l}
{R\left( {\mathit{\epsilon},k_{0}} \right)} \\
{\mspace{22mu} = {\int_{0}^{k_{0}}{dk\left\lbrack {C_{1}\delta\left( {k - k_{0}} \right) + {C_{2}e -}^{(\frac{k_{0} - k}{C_{3}k_{0}})}} \right\rbrack}} \cdot C_{4}e^{- \frac{1}{2}{(\frac{k - \mathit{\epsilon}}{\alpha})}^{2}},} \\
{\mspace{94mu}{\int_{0}^{k_{0}}R}\left( {\mathit{\epsilon},k_{0}} \right)dk_{0} = 1} \\
\end{array}$$where *C*~1~, *C*~2~, *C*~3~, *C*~4~, and *α* are constants characteristic of the detector. The pulse height distribution becomes $$P\left( \mathit{\epsilon} \right) - {\int_{0}^{\infty}{dk_{0}}}R\left( {\mathit{\epsilon},k_{0}} \right)N\left( k_{0} \right)$$where *N*(*k*~0~) is the continuous incident photon number spectrum. [Equation (4)](#fd12-jresv68an4p401_a1b){ref-type="disp-formula"} is the continuous form of [(1)](#fd2-jresv68an4p401_a1b){ref-type="disp-formula"}. Experimentally the vector $$P_{i} = {\int_{\mathit{\epsilon}_{i} - \Delta_{i}}^{\mathit{\epsilon}_{i} + \Delta_{i}}{P\left( \mathit{\epsilon} \right)d\mathit{\epsilon}}}$$is the quantity measured as counts per channel in a multichannel pulse height analyzer.
The integral [equation (4)](#fd12-jresv68an4p401_a1b){ref-type="disp-formula"} does not possess an exact solution. Integrating over *k* in [(3)](#fd10-jresv68an4p401_a1b){ref-type="disp-formula"}, [(4)](#fd12-jresv68an4p401_a1b){ref-type="disp-formula"} becomes $$P(\mathit{\epsilon}) = {\int{dk_{0}\left\{ {C_{1}C_{4}\exp\left\lbrack {- \frac{{(k_{0} - \mathit{\epsilon})}^{2}}{2\alpha^{2}}} \right\rbrack + C_{2}C_{4}\left( {\alpha\sqrt{\frac{\pi}{2}}} \right)\left\lbrack {\exp\left( {- \frac{1}{C_{3}} + \frac{\mathit{\epsilon}}{C_{3}k_{0}} + \frac{\alpha}{\sqrt{8}C_{3}^{2}k_{0}^{2}}} \right)} \right\rbrack\left\lbrack {\Phi\left( {\frac{k_{0}}{\sqrt{2\alpha}} - \frac{\mathit{\epsilon}}{\sqrt{2\alpha}} - \frac{\alpha}{\sqrt{2}C_{3}k_{0}}} \right) - \Phi\left( {\frac{\mathit{\epsilon}}{\sqrt{2}\alpha} + \frac{\alpha}{\sqrt{2}C_{3}k_{0}}} \right)} \right\rbrack} \right\} N(k_{0})}}$$where Φ(*x*) is the error integral \[[@b5-jresv68an4p401_a1b]\]. Thus [(5)](#fd13-jresv68an4p401_a1b){ref-type="disp-formula"} is seen to be an integral equation with a Gaussian kernel; such an equation does not possess a general unique solution \[[@b6-jresv68an4p401_a1b]\]. This is a manifestation of the inability to experimentally differentiate between a smooth spectrum and a spectrum containing a series of sharp spikes. The Gaussian broadening is responsible for this.
In order to obtain a matrix representation for *R*(*ϵ, k*~0~) a particular form must be assumed for *N*(*k*~0~). Two forms have been investigated. One may assume the incident spectrum to consist of a series of discrete steps so that over a fixed small energy width the spectrum is constant \[[@b7-jresv68an4p401_a1b]\]. Alternatively one may assume the spectrum to be composed of a sum of Dirac delta functions so that when an integration is performed over a small energy width the area of the delta function gives the number of photons in that width.
Both cases lead to essentially the same form for the matrix. The latter case will be carried through to obtain the matrix explicitly.
Let $N\left( k_{0} \right) = {\sum\limits_{j}{a_{j}\delta\left( {k_{j} - k_{0}} \right).}}$ Then [(4)](#fd12-jresv68an4p401_a1b){ref-type="disp-formula"} becomes $$P(\mathit{\epsilon}) = {\sum\limits_{j}{K_{j}f\left( {k_{j},\mathit{\epsilon}} \right)a_{j}}}$$where $K_{j} = C_{1}C_{4} + k_{j}C_{2}C_{3}C_{4}\left( {1 - e^{- 1/C_{3}}} \right)$ is a number, and $f\left( {k_{j},\mathit{\epsilon}} \right) = e^{- \frac{1}{2}{(\frac{k_{i} - \mathit{\epsilon}}{\alpha})}^{2}}$.
From the above the number of counts in channel *ϵ*~i~ is $$P_{i} = {\int_{\epsilon_{i} - \Delta_{i}}^{\mathit{\epsilon}_{i} + \Delta_{i}}{\sum\limits_{j}K_{j}}}f\left( {k_{j},\mathit{\epsilon}} \right)a_{j}$$which, after interchanging integration and summation becomes $$P_{i} = {\sum\limits_{j}{a_{j}\left\lbrack {K_{j}b} \right.\left( {k_{j},\mathit{\epsilon}_{i},\Delta_{i}} \right)\rbrack}}$$where $$b\left( {k_{j},\mathit{\epsilon}_{i},\Delta_{i}} \right) = {\int_{\mathit{\epsilon}_{i} - \Delta_{i}}^{\mathit{\epsilon}_{i} + \Delta i}e^{- \frac{1}{2}{(\frac{k_{i} - \mathit{\epsilon}}{\alpha})}}}^{2}d\mathit{\epsilon} = b_{ji}\left( \Delta_{i} \right).$$Identifying *a~j~* with *N~j~* and *K~j~b~ij~* with *R~ij~* [(7)](#fd16-jresv68an4p401_a1b){ref-type="disp-formula"} becomes identical with [(1)](#fd2-jresv68an4p401_a1b){ref-type="disp-formula"}.
3. Empirical Justification
==========================
3.1. Convergence
----------------
In setting out on this course there was no reason to believe the technique to be convergent. It has been shown \[[@b3-jresv68an4p401_a1b]\] that convergence is assured for a smooth function if the eigenvalues Λ*~i~* of the response function matrix satisfy the requirement $$0 < \Lambda_{i} < 2$$
This was not a useful test because the size of the matrices used made calculation of the eigenvalues impractical. Therefore the primary justification is empirical.
In analysis utilizing a 200 × 200 form of *R~ij~*, eleven iterations were ordinarily performed. However, as a check on convergence, as many as twenty-one iterations have been performed, during which Δ*~n~* of [equation (2)](#fd9-jresv68an4p401_a1b){ref-type="disp-formula"} continues to converge.
In [figure 1](#f1-jresv68an4p401_a1b){ref-type="fig"} a typical set of points to be unfolded is plotted. Let *A* denote this set. On the same figure is plotted *B*, the result of unfolding *A.* The set *A* contained points only up to 40 MeV. In order to avoid introducing a large discontinuity in the first derivative at 40 MeV, a straight line tail has been added. The work was done with an energy grid width of 0.5 MeV. Typical standard deviation is shown for a point of *A* at 16 MeV.
In order to compare *B* with *A*, the difference Δ~11~*=A*−*RB* (see [(2)](#fd9-jresv68an4p401_a1b){ref-type="disp-formula"}) is plotted in [figure 2](#f2-jresv68an4p401_a1b){ref-type="fig"}. If *B* is the "correct" unfolded set of points then Δ~11~ must vanish. Convergence requires that Δ*~n~* vanishes for increasing *n.*
Some values of the difference in percent are indicated on the plot. The very small (0.7%) difference at 19.5 MeV is at the peak of *A.*
In order to further check the convergence properties of the scheme a set of points with large uncertainties was unfolded, using twenty-one iterations. The set *C* and its "unfold" *D* are shown in [figure 3](#f3-jresv68an4p401_a1b){ref-type="fig"}. Again a straight line tail has been added to *C* at 40 MeV.
Because of the poorer statistics on *C* there are more fluctuations in *D.* The slope of *C* appears to have a large discontinuity at 32.5 MeV. A spike in *D* is observed to grow at this energy with successive iterations. This demonstrates that fluctuations are magnified as one approaches an exact solution.
The question of convergence is best illustrated by examining Δ*~n~*, for various values of *n.* [Figures 4](#f4-jresv68an4p401_a1b){ref-type="fig"} and [5](#f5-jresv68an4p401_a1b){ref-type="fig"} show Δ*~n~* for *n*=0,1,4,11, and 21. It is observed that Δ*~n~* converges rapidly for small *n.* The maximum of the ratio Δ~21~/Δ~0~ is approximately 10^−3^. The maxima of δ~21~ in percent of *C* are +0.4 and −0.5 percent.
A numerical criterion for testing convergence in this sense is suggested by Skarsgard, Johns, and Green \[[@b3-jresv68an4p401_a1b]\] for a pulse height distribution containing pure Poisson counting errors, i.e. the standard deviation on *P~i~* is ${\sqrt{P}}_{i}$.
If $\frac{\left( \Delta_{i} \right)^{2}}{P_{i}} < < 1$ then the deviation for the point *U~i~* is well within the limits of random measurement errors. Therefore if $${\sum\limits_{i = 1}^{M}\frac{\Delta_{i}^{2}}{P_{i}}} \leq M$$the unfold is regarded as satisfactory. This test was used in unfolding a pulse height distribution for which the errors on each point were purely counting errors. The results were similar to those found by Skarsgard, Johns, and Green, \[[@b3-jresv68an4p401_a1b]\] namely, convergence is rapid until [(8)](#fd19-jresv68an4p401_a1b){ref-type="disp-formula"} is satisfied (\~3 iterations). After this convergence proceeds slowly.
One might hope to be able to prove convergence from the classical theorems \[[@b1-jresv68an4p401_a1b]\]. It is easily shown that if one denotes (*I-R*) by *A*, then: $$U^{(n)} - U_{r} = A^{n}\left( {U_{0} - U_{T}} \right).$$Therefore \[[@b1-jresv68an4p401_a1b]\] *U*^(^*^n^*^)^ converges to the true solution *U~T~* if and only if the eigenvalues of *A* are less than one in modulus. From the rapid convergence which is observed empirically, one is led to believe that the eigenvalues of *A* are indeed less than one in modulus.
A sufficient condition for convergence is that the maximum of the absolute row sums *μ~i~* satisfy $$\mu_{i} = \left( {\sum\limits_{j = 1}^{n}\left| A_{ij} \right|} \right) < 1.$$However, this is not the case for the matrix *A=I-R* on which the present work is based.
4. Error Propagation
====================
4.1. Empirical
--------------
In order to demonstrate the effect of statistical fluctuations, two different experimental determinations of the same pulse height distribution have been unfolded. A portion of the unfolded spectra for both sets of data are presented in [figure 6](#f6-jresv68an4p401_a1b){ref-type="fig"}. The spectra are designated *U~a~* and *U~b~.* The two pulse height distributions are not presented because of the typographical difficulty in distinguishing the two sets of data on a meaningful scale.
Differences between the two spectra should be purely statistical. Let the measured pulse height distributions from which *U~a~* and *U~b~* were obtained be *P~a~* and *P~b~.* The ratio *ρ*=(*P~a~/U~a~*)*/*(*P~b~/U~b~*) has been plotted in [figure 7](#f7-jresv68an4p401_a1b){ref-type="fig"} for the region from 15 MeV to 25 MeV. One would expect this ratio to be randomly distributed about unity due to the statistical fluctuations in *P~a~* and *P~b~.* This is observed.
In addition, if the unfolding procedure does not introduce false structure, then, for *P~a~\>P~b~* one expects the relations *U~a~\>U~b~* and therefore *ρ*\>1, to hold approximately. Examination of [figures 6](#f6-jresv68an4p401_a1b){ref-type="fig"} and [7](#f7-jresv68an4p401_a1b){ref-type="fig"} will show that for *U~a~*\>*U~b~, ρ*\>1, and for *U~a~\<U~b~, ρ*\<1, except when *U~a~*≃*U~b~* where fluctuations in adjacent points become important. This indicates that the iterative procedure does not introduce false structure.
Some qualitative effects of error propagation are illustrated quite well by [figure 6](#f6-jresv68an4p401_a1b){ref-type="fig"}. The most pronounced effect is the increase of fluctuations in the unfolded curves with increasing energies. The reason for this will emerge from the discussion following.
4.2. Calculation of Error
-------------------------
The calculation of error for an individual point in an unfolded spectrum is made difficult, because, in folding, correlations arise between errors in adjacent points.
Assume that error in the detected pulse height distribution is known as a function of energy and denote it by *σ~k~*. The folded set of points, which are obtained in the first step, may be written $${\lbrack S\left( P \right)\rbrack}_{i} = {\sum\limits_{j}{R_{ij}P_{j.}}}$$The standard deviation of *S~i~* becomes $$\delta_{i} = \lbrack{\sum\limits_{j}{\left( {R_{ij}\sigma_{j}} \right)^{2}\rbrack^{\frac{1}{2}}}}.$$From [(2)](#fd9-jresv68an4p401_a1b){ref-type="disp-formula"} the solution after one iteration may be written $$U_{i}^{(1)} = 2P_{i} - {\sum\limits_{j}{R_{ij}P_{j}}}$$with corresponding error $$\delta U^{(1)}~_{i} = \lbrack\left( {2 - R_{ii}} \right)^{2}{\sigma_{i}}^{2} + {\sum\limits_{j \neq i}{\left( {R_{ij}\sigma_{j}} \right)^{2}\rbrack^{\frac{1}{2}}.}}$$The second iteration gives: $$\begin{matrix}
{U^{(2)} = U^{(1)} + P - 2RP + R^{2}P} \\
{= 3P - 3RP + R^{2}P} \\
\end{matrix}$$where $$R^{2}P = {\sum\limits_{jk}{R_{ij}R_{jk}P_{k}.}}$$Expanding *U*^(^*^2^*^)^ in the same way as *U*^(^*^1^*^)^, the error on *U*^(^*^2^*^)^ becomes $$\delta U_{i}~^{(2)} = {\lbrack{\left( {3 - 3R_{ii} + R_{ii}~^{2}} \right){\sigma_{i}}^{2} + \ \text{other}\ \text{terms}}\rbrack}^{\frac{1}{2}}.$$The "other terms" have numerous cross products. For example, the contribution to $\left\lbrack {\delta U_{i}^{(2)}} \right\rbrack^{2}$ from *σ~i~*~+1~ is \[−3*R~i,i~*~+1~+*R~ii~R~i,i~*~+1~+*R~i~*, *~i~*~+1~, *R~i~*~+1,\ i+1~\]^2^*σ~i~*~+1~^2^.
The general form for the solution after *n* iterations may be written symbolically as $$U^{(n)} = \frac{1}{R}\left\lbrack {I - {(I - R)}^{n + 1}} \right\rbrack P$$where is the identity matrix and *1/R=R*^−1^. If the variance Var (*P*) *= σ*^2^*I*, then the variance of *U*^(^*^n^*^)^ may be written formally as: $$\text{Var}\left( U^{(n)} \right) = R^{- 1}\left\lbrack {I - \left( {I - R} \right)^{n + 1}} \right\rbrack\ \left\lbrack {I - {(I - R)}^{n + 1}} \right\rbrack^{T}\left( R^{- 1} \right)^{T}\sigma^{2},$$where *T* denotes the transpose. Here we have used \[[@b8-jresv68an4p401_a1b]\] $$\text{Var}\left( {CP} \right) = C\left\lbrack {{var}\left( P \right)} \right\rbrack C^{T}.$$Since all elements of *R* are less than unity, it is evident that in the limit as *n* approaches infinity $$\lim\limits_{n\rightarrow\infty}\text{Var}\left( U^{(n)} \right) = R^{- 1}\left( R^{- 1} \right)^{T}\sigma^{2}.$$
In order to simplify the error calculation the response function and the error will be assumed to satisfy the following conditions: The half-width of the response function is narrow. This corresponds to good resolution in the detector.The shape of the response function does not change rapidly with incident photon energy. This is equivalent to assuming that $$\begin{matrix}
{R\left( {\mathit{\epsilon},k_{1}} \right) \simeq R\left( {\mathit{\epsilon} + \Delta k_{1},k_{1} + \Delta k_{1}} \right)} \\
{\text{or}\ R_{ij} \simeq R_{i + m,j + m\ }\ \text{and}\ R_{i - m,j} \simeq R_{i,j + m,}} \\
\end{matrix}$$where *m* is an integer. See [figure 8](#f8-jresv68an4p401_a1b){ref-type="fig"}.The error, *σ~k~*, is a constant, *σ*, over the half-width of the response function. Note that the first condition makes this more likely.
Let *R~ii~=w*~0~, *R~i−~*~1~,*~i~=w*~1,~ *R~i~*~−2~,*~i~=w*~2~, ... But, from condition [(2)](#fd9-jresv68an4p401_a1b){ref-type="disp-formula"} *R~i~,~i+k~*≃w*~k~.* Using this [(9)](#fd22-jresv68an4p401_a1b){ref-type="disp-formula"} may now be written $$\left\lbrack {RP} \right\rbrack_{i} = {\sum\limits_{k}{w_{k}P_{k + i},}}$$where the *w~k~* may be obtained from the response at one incident energy.
In the [appendix](#app1-jresv68an4p401_a1b){ref-type="app"} it is shown that the solution has the general form: $$U_{i}^{(n)} = a_{0}P_{i} + a_{i}P_{i + 1} + a_{2}P_{i + 2} + \ldots$$
If the error is assumed to be a constant, *σ*, this gives $$\delta U_{i}~^{(n)} = {({\sum\limits_{j}a_{j}^{2}})}^{\frac{1}{2}}\sigma$$for the error at point *i* in the *n*th iterated solution.
For the response described in \[[@b4-jresv68an4p401_a1b]\] the error after three iterations has been calculated at 18.5 MeV and 48.5 MeV. (See [appendix](#app1-jresv68an4p401_a1b){ref-type="app"}.) The results are ±2.36 *σ* at 18.5 MeV and ±3.23 *σ* at 48.5 MeV.
In both cases if all terms in (*A*3) which are cubic in *w* were omitted, the difference in *U~i~*^(3)^ would be small, and the difference in *δU~i~*^(3)^ would be negligible. The terms which are cubic in *w* are approximately an order of magnitude smaller than the quadratic terms. The conclusion is that *δU*^(*n*)^ has converged for n≥3.
From (*A*1), (*A*2), and (*A*3) it may be observed that after a large number of iterations the coefficient of *P*~o~ in (*A*3) will converge to: $$a_{0} = \left\lbrack {1 - \left( {1 - w_{0}} \right)^{n + 1}} \right\rbrack/w_{0}.$$
For the cases at 18.5 and 48.5 MeV this gives *a*~o~=2.21 and 3.104 respectively, for *n*=3. Comparison with the results for *δU~i~*^(3)^ above shows very close agreement. Thus one concludes that three iterations satisfy the large number criterion.
This coefficient then places a lower limit on the propagated error at each point of the solution. In general *w*~o~ decreases with increasing incident photon energy, for this experiment. Therefore the error must increase with energy. This is independent of the shape of the curve to be unfolded. [Figure 9](#f9-jresv68an4p401_a1b){ref-type="fig"} shows *a*~o~ plotted as a function of energy for *n*=3.
5. Summary
==========
Because of the uncertainties associated with any point on a measured pulse height distribution, any "solution" for the unfolded spectrum is acceptable, if the difference between the measured distribution and the fold of the solution lies within the uncertainty associated with the measured distribution. The additional requirement of smoothness is sufficient to ensure that the iterative process converges to a useful solution.
General error analysis is difficult. However, approximations may be made which become better as the resolution of the detector improves; these approximations make an error estimate possible.
Interesting results from the use of this technique may be seen in the work of Ziegler, Koch, Wyckoff, and Uhlig \[[@b9-jresv68an4p401_a1b]\].
The author gratefully acknowledges the help and encouragement of H. W. Koch, It. A. Schrack, and Michael Danos in the development of this method, and their suggestions for improvement of this paper. The original work in finding a suitable form for the response function and the fitting of that form to experimental data was done by R. A. Schrack.
Special thanks go to Joseph Cameron and Brian Joiner for their help in studying the propagation of error.
In this appendix it will be shown how the general form of the solution, [equation (14)](#fd35-jresv68an4p401_a1b){ref-type="disp-formula"}, may be found.
Using [(9a)](#fd34-jresv68an4p401_a1b){ref-type="disp-formula"} and expanding [(13)](#fd29-jresv68an4p401_a1b){ref-type="disp-formula"} the general solution becomes: $$U_{t}^{(n)} = nP_{i} - \frac{n\left( {n - 1} \right)}{2!}{\sum\limits_{j}{w_{j}P_{j + i} + \frac{n\left( {n - 1} \right)\left( {n - 2} \right)}{3!}}}{\sum\limits_{k}{\sum\limits_{j}{w_{k}w_{j}}}}P_{k + j + i} + \frac{n\left( {n - 1} \right)\left( {n - 2} \right)\left( {n - 3} \right)}{4!}{\sum\limits_{l}{\sum\limits_{k}{\sum\limits_{j}{w_{l}w_{k}w_{j}P_{l + k + j + i} + \ldots}}}}$$Choosing an arbitrary zero point for index *i* (Al) may be written: $$U_{0}^{(3)} = 4P_{0} - 6{\sum\limits_{j}{w_{j}P_{j} + 4{\sum\limits_{j}w_{j}^{2}}}}P_{2j} + 8{\sum\limits_{k}{\sum\limits_{< j}{w_{k}w_{j}P_{k + j} - {\sum\limits_{j}w_{j}^{3}}}}}P_{3j} - 3{\sum\limits_{k}{\sum\limits_{< j}{w_{k}^{2}w_{j}P_{2k +} - 3{\sum\limits_{j}{\sum\limits_{< k}w_{k}^{2}}}}}}w_{j}P_{2k + j} - 6{\sum\limits_{j}{\sum\limits_{< k}{\sum\limits_{< j}P_{l + k + j.}}}}$$
Expanding and collecting terms: $$\begin{array}{l}
{U_{0}^{(3)} = (4 - 6w_{0} + 4w_{0}^{2} - w_{0}^{3})P_{0} + ( - 6w_{1} + 8w_{0}w_{1} - 3w_{0}^{3}w_{1})P_{1}} \\
{\mspace{45mu} + ( - 6w_{2} + 4w_{1}^{2} + 8w_{0}w_{2} - 3w_{0}^{2}w_{2} - 3w_{0}w_{1}^{2})P_{2}} \\
{\mspace{45mu} + ( - 6w_{3} + 8w_{0}w_{3} + 8w_{1}w_{2} - w_{1}^{3} - 3w_{0}^{2}w_{3} - 6w_{0}w_{1}w_{2})P_{3} + ( - 6w_{4} + 4w_{2} + 8w_{0}w_{4} + 8w_{1}w_{3} - 3w_{0}^{2}w_{4} - 3w_{1}^{2}w_{2} - 3w_{2}^{2}w_{0} - 6w_{0}w_{1}w_{3})} \\
{\mspace{45mu} P_{4} + ( - 6w_{5} + 8w_{0}w_{5} + 8w_{1}w_{4} + 8w_{2}w_{3} - 3w_{0}^{2}w_{5} - 3w_{1}^{2}w_{3} - 3w_{2}^{2}w_{1} - 6w_{0}w_{1}w_{4} - 6w_{0}w_{2}w_{3})P_{5} + \ldots} \\
\end{array}$$
At 18.5 MeV the response function for the detection system described by Ziegler, Wyckoff, and Koch \[[@b4-jresv68an4p401_a1b]\] was given by $$\left( {w_{0},w_{1},w_{2},w_{3},w_{4},w_{5},\ldots} \right) = \left( {0.39,0.225,0.138,0.088,0.056,0.035,\ldots} \right).$$Using these values in (A3) one finds: $$U_{0}^{(3)} = 2.209P_{0} - 0.751P_{1} - 0.317P_{2} - 0.129P_{3} - 0.084P_{4} - 0.003P_{5} - \ldots,$$and from [(15)](#fd36-jresv68an4p401_a1b){ref-type="disp-formula"}: $$\delta U_{0}~^{(3)} = 2.36\sigma.$$Similarly the response at 48.5 MeV gives $$U_{0}~^{(3)} = 3.103P_{0} - 0.642P_{1} - 0.454P_{2} - 0.333P_{3} - 0.242P_{4} - 0.174P_{5} - \ldots,$$which leads to $$\delta U_{0}~^{(3)} = 3.23\sigma.$$
{#f1-jresv68an4p401_a1b}
{ref-type="fig"} and the folding of the unfolded curve of [figure 1](#f1-jresv68an4p401_a1b){ref-type="fig"}.\
X% indicates percentage difference Δ/A. Ordinate scale is in same units as [figure 1](#f1-jresv68an4p401_a1b){ref-type="fig"}.](jresv68an4p401_a1bf2){#f2-jresv68an4p401_a1b}
{#f3-jresv68an4p401_a1b}
{ref-type="fig"} and the folding of the unfolded curve after *n* iterations, for n=0, 1, and 4.\
Ordinate scale in [figures 4](#f4-jresv68an4p401_a1b){ref-type="fig"} and [5](#f5-jresv68an4p401_a1b){ref-type="fig"} is in the same units as [figure 3](#f3-jresv68an4p401_a1b){ref-type="fig"}.](jresv68an4p401_a1bf4){#f4-jresv68an4p401_a1b}
{ref-type="fig"}.](jresv68an4p401_a1bf5){#f5-jresv68an4p401_a1b}
{#f6-jresv68an4p401_a1b}
{#f7-jresv68an4p401_a1b}
{#f8-jresv68an4p401_a1b}
![The function *a*~0~ = \[1 − (1−*w*~0~)^*n*+1^\]/*w*~0~ as a function of energy for *n*=3.](jresv68an4p401_a1bf9){#f9-jresv68an4p401_a1b}
[^1]: Present address: Department of Physics and Astronomy, University of Maryland, College Park, Md.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Ovarian tumors are a group of molecularly and etiologically heterogeneous cancers \[[@CR1]\], and are the fifth most common cause of cancer related mortality among women. The current standard of therapy is debulking surgery and combination chemotherapy, consisting of a taxane (e.g. paclitaxel) and a platinum-based compound (e.g cisplatin or carboplatin) \[[@CR2], [@CR3]\]. Despite an initial high chemo-responsiveness, with response rates over 80 % \[[@CR4]\], most advanced epithelial ovarian cancer patients will relapse and ultimately die of drug-resistant disease \[[@CR5]\].
In the cellular response to cytotoxic substances, cell signaling pathways, such as the MAPK pathways, play a pivotal role. The classical MAPK pathway is activated by receptor tyrosine kinases (RTK), which bind GRB2 and SOS, after which the kinases RAS, RAF, MEK and MAPK are sequentially activated. Three major MAPK routes have been identified: the p38 MAPK, Jun kinase and ERK pathway. The MAPK pathways regulate numerous targets, including p90 ribosomal S6 protein kinases (RSKs), which in turn can activate downstream proteins. Together, MAPK pathways regulate the activity of genes involved in cell proliferation, DNA damage repair, cell cycle progression and apoptosis \[[@CR6]--[@CR10]\].
Although the mechanisms behind chemotherapy resistance in ovarian cancer have been studied extensively, the involvement of microRNAs (miRNAs), small RNAs that regulate gene expression, is just beginning to be unraveled \[[@CR11], [@CR12]\]. We have recently shown that *miR-141* targets *KEAP1* in ovarian cancer cells, repression of which results in enhanced cisplatin resistance \[[@CR13]\]. In the current study we aimed to identify additional miRNAs that play a role in cisplatin resistance. Here, we describe that *miR-634* can sensitize both ovarian cancer cell lines and primary ovarian cancer cell cultures to chemotherapy. We show that *miR-634* regulates cyclin D1 and several Ras-MAPK pathway components (GRB2, ERK2, RSK1 and RSK2), which may contribute to the effects of *miR-634* on ovarian cancer cell survival and chemotherapy response.
Results {#Sec2}
=======
Comparison of miRNA expression profiles of cisplatin sensitive and resistant cell line pairs {#Sec3}
--------------------------------------------------------------------------------------------
In order to find miRNAs that play a role in cisplatin resistance, we compared miRNA expression profiles of cisplatin sensitive/resistant cell line pairs (IC~50~ values in Additional file [1](#MOESM1){ref-type="media"}: Table S1A). We hypothesized that in different cell types the same miRNAs play a role in cisplatin sensitivity, as has been reported for other factors involved in drug resistance \[[@CR14]\]. Therefore, the miRNA expression pattern of an ovarian cancer cell line pair (A2780/A2780 DDP) was compared with expression patterns of a bladder cancer (T24/T24 DDP) and colon cancer (HCT8/HCT8 DDP) cell line pair. The only miRNA that showed a common pattern in all cell lines was *miR-634* (Additional file [1](#MOESM1){ref-type="media"}: Figure S1, FDR = 0.000), which was downregulated ≥1.5 fold in all cisplatin resistant cell lines (Additional file [1](#MOESM1){ref-type="media"}: Table S2). We further investigated the role of *miR-634* in ovarian cancer.
Effects of miR-634 overexpression on cell cycle and apoptosis {#Sec4}
-------------------------------------------------------------
Before examining the effects of *miR-634* on cisplatin sensitivity, we determined whether *miR-634* overexpression affects the cell cycle and cell survival of A2780 DDP cells, which have a low basal *miR-634* expression compared to the parental A2780 cells. Upon transfection of the *miR-634* mimic, a slightly higher percentage of cells was observed in the G1 phase (*p* = 0.04) accompanied by a lower number of cells in the G2/M phase (*p* = 0.04) (Fig. [1a](#Fig1){ref-type="fig"}). These effects, which were observed at 48 h after transfection in multiple experiments, suggest that *miR-634* overexpression may affect the G1-to-S phase transition. At 72 h after transfection, however, the cell cycle profile of *miR-634* overexpressing cells was comparable to cells transfected with scrambled mimic (Fig. [1a](#Fig1){ref-type="fig"}).Fig. 1*MiR-634* overexpression induces G1 arrest and causes cell death. **a** Percentage of A2780 DDP cells in G0/G1, S or G2/M phase 48 or 72 h after transfection with a *miR-634* mimic or a scrambled control (*n* = 3), \* = *p* \< 0.05. **b** A2780 DDP cells were stained with PI and Annexin V 48 or 72 h after transfection with *miR-634* mimic or a scrambled control. Depicted are viable (PI/Annexin V negative), early apoptotic (Annexin V positive/PI negative), late (Annexin V positive/PI positive) and dead (PI positive/Annexin V negative) cells (*n* = 3). **c** Viability of *miR-634* mimic transfected ovarian cancer cells compared to cells transfected with a scrambled mimic (set at 100 %), as determined by an MTT assay 72 h after transfection. Depicted are average values ± SD (*n* = 3)
Next, we examined if *miR-634* overexpression induces apoptosis. Whereas at 48 h after transfection the viability of control and *miR-634* mimic transfected cells was comparable, at 72 h the percentage of viable cells was significantly lower (*p* = 0.03) in *miR-634* transfectants, corresponding to increased numbers of apoptotic and dead cells (Fig. [1b](#Fig1){ref-type="fig"}). This effect of *miR-634* on apoptosis was also detected by MTT assay in five other ovarian cancer cell lines, A2780 (parental line), OV56, OAW42, TOV21G and TOV112D. In these cells *miR-634* gave rise to a 20--50 % reduction in viability, relative to control transfectants (Fig. [1c](#Fig1){ref-type="fig"}).
MiR-634 enhances cisplatin sensitivity of ovarian cancer cell lines {#Sec5}
-------------------------------------------------------------------
We next determined the effects of *miR-634* overexpression on cisplatin sensitivity using a previously developed assay \[[@CR13]\]. Briefly, cells were transfected with a *miR-634* mimic or a scrambled control, and after 48 h exposed to various concentrations of cisplatin. After another 24 h cell viability was determined using an MTT assay. Because miRNA transfection was transient we used 24 h drug exposure intervals. Note that the difference in IC~50~ values observed between drug sensitive and resistant cell lines was similar to IC~50~ values determined in assays with longer drug exposure times \[[@CR13]\] (Additional file [1](#MOESM1){ref-type="media"}: Table S1A, C). As is shown in Fig. [2a](#Fig2){ref-type="fig"}, transfection with *miR-634* mimic gave rise to a marked increase in sensitivity after treatment with 80 μM (*p* = 0.006) and 125 μM cisplatin (*p* = 0.002) in the A2780 DDP cell line. A possible explanation for this sensitization may be a higher intracellular cisplatin accumulation and concomitantly increased cytotoxic activity. In fact, cisplatin resistance is frequently accompanied by reduced intracellular cisplatin levels \[[@CR15]\] (Additional file [1](#MOESM1){ref-type="media"}: Table S1B). We measured the intracellular platinum content of *miR-634* and scrambled control transfected cells after exposure to 80 and 125 μM cisplatin for 2 h, 6 h and 12 h. *MiR-634* transfection did not affect the platinum uptake by A2780 DDP cells (Additional file [1](#MOESM1){ref-type="media"}: Figure S2) ruling out that *miR-634* sensitizes ovarian cancer cells for cisplatin by increasing its uptake and/or reducing its efflux.Fig. 2*MiR-634* mimic enhances sensitivity for cisplatin in ovarian cancer cell lines. Ovarian cancer cell lines were transfected with *miR-634* mimic or a scrambled control. 48 h after transfection, cisplatin was added and after 24H cell viability was determined with an MTT assay. **a** Overexpression of *miR-634* miRNA in the sensitive A2780 and resistant A2780 DDP cell lines. Error bars represent the standard deviation within one experiment (*n* = 5). \*\* = *p* \< 0.01. **b** Overexpression of *miR-634* miRNA in the OV56, TOV21G, OAW42, TOV112D ovarian cancer cell lines. The cisplatin concentration is based on the IC~25~ value of mock transfected cells). Error bars represent the standard deviation within one experiment (*n* = 3). \* = *p* \< 0.05
In contrast to the resistant A2780 DDP cell line, enhanced expression of *miR-634* in the sensitive A2780 cells did not alter the sensitivity for cisplatin. This might be due to the fact that *miR-634* levels are already high in A2780 cells relative to A2780 DDP. We next examined the effect of *miR-634* transfection on other ovarian cancer cell lines. Because the most significant effect was observed at the IC~25~ dose in the A2780 DDP cell line, the other ovarian cancer cell lines were treated with the IC~25~ dose of mock (transfection reagent only) transfected cells (Additional file [1](#MOESM1){ref-type="media"}: Table S1C). In two of the most resistant cell lines in this assay, OV56 and TOV21G, *miR-634* overexpression significantly reduced viability (Fig. [2b](#Fig2){ref-type="fig"}). In contrast, in the more sensitive cell lines OAW42 and TOV112D, overexpression of *miR-634* did not alter sensitivity. The effect of *miR-634* on cisplatin response thus appears to be strongest in resistant cell lines.
MiR-634 sensitizes tumor cells from patients with drug resistant ovarian cancer to chemotherapy {#Sec6}
-----------------------------------------------------------------------------------------------
Ovarian cancer is routinely treated with platinum (e.g. cisplatin, carboplatin) and taxane (e.g. paclitaxel) based combination chemotherapy. However, most tumors eventually become resistant. As *miR-634* overexpression can increase cisplatin sensitivity of cell lines, we examined whether this treatment could also sensitize chemotherapy resistant primary ovarian tumor cells. Ascites was collected from 6 patients with serous and 1 patient with clear cell ovarian cancer (for patient characteristics, see Table [1](#Tab1){ref-type="table"}). One patient was chemotherapy naïve, the others had received carboplatin/paclitaxel combination regimens and were or had become resistant. The ovarian cancer cells were isolated and cultured from ascites as described previously \[[@CR16], [@CR17]\], All cultures tested positive for the epithelial marker pan-keratin (Additional file [1](#MOESM1){ref-type="media"}: Figure S3). Another epithelial marker, EpCAM, was only positive in tumor cells from patient 2, however it is known that primary ovarian cancer cells may lose EpCAM expression in culture \[[@CR16]\]. To verify that the cultures contain tumor cells, p53 staining was analyzed as p53 is often mutated in high and intermediate grade serous ovarian cancer \[[@CR18]\]. A clear nuclear p53 staining was observed in all cultures, indicating the cultures do indeed consist of tumor cells.Table 1Ovarian cancer patient characteristicsHistological subtypeFIGO stageGradeCytotoxic agents usedResponse to last Pt containing treatmentPatient 1SerousIV2Carboplatin, PaclitaxelPD within 1 monthPatient 2SerousIIIcUKCarboplatin, Paclitaxel, GemcitabinePD within 4 monthsPatient 3SerousIIIc3Carboplatin, Paclitaxel, GemcitabinePD within 1 monthPatient 4SerousIIIc2Carboplatin, Paclitaxel, Olaparib, CaelyxPD during treatmentPatient 5Clear cellIIIc3Carboplatin, IfosfamidePD within 1 monthPatient 6SerousIIIc3Carboplatin, Paclitaxel, CaelyxPD within 5 monthsPatient 7SerousIIIc2Chemotherapy naive*UK:* Unknown, *Pt:* Platinum, *PD:* Progressive disease, Caelyx: liposomal doxorubicin
We first tested whether *miR-634* overexpression in primary tumor cells gave rise to a reduced viability, as was observed for the ovarian cancer cell lines. Remarkably, *miR-634* overexpression in primary cell cultures only mildly diminished (5--25 %) the number of viable cells (Fig. [3a](#Fig3){ref-type="fig"}).Fig. 3*MiR-634* mimic enhances chemotherapy sensitivity in primary ovarian cancer cell cultures. **a** Overexpression of *miR-634* in primary ovarian cancer cell cultures, derived from the ascites of ovarian cancer patients, has no or a moderate effect on cell viability 72 h after transfection. Indicated is the viability of *miR-634* transfected cells relative to the viability of cells transfected with scrambled mimic (set at 100 %). The numbers on the X-axis refer to the number of the culture. Depicted are average values ± SD (*n* = 3). **b** *MiR-634* enhances the sensitivity for cisplatin. Depicted are the results of a representative experiment for culture 3 (left, *n* = 4) and culture 7 (right, *n* = 3). An overview of the results obtained for all cultures is in Additional file [1](#MOESM1){ref-type="media"} Figure S4. \* = *p* \< 0.05, \*\* = *p* \< 0.01. **c** *MiR-634* enhances the sensitivity for carboplatin. Depicted are the results of a representative experiment for culture 3 (left, *n* = 3) and culture 7 (right, *n* = 4). An overview of all cultures that were treated with carboplatin is in Additional file [1](#MOESM1){ref-type="media"}: Figure S5. \* = *p* \< 0.05, **d** *MiR-634* enhances the sensitivity for doxorubicin. Depicted are the results of a representative experiment for culture 3 (left, *n* = 3) and culture 7 (right, *n* = 3). \*\* = *p* \< 0.01. **e** *MiR-634* does not significantly alter the sensitivity for paclitaxel. Depicted are the results of a representative experiment for culture 3 (left, *n* = 3) and culture 7 (right, *n* = 3)
*miR-634* overexpression gave rise to a significant and reproducible decrease in cisplatin resistance in five cultures (Fig. [3b](#Fig3){ref-type="fig"}, Additional file [1](#MOESM1){ref-type="media"}: Figure S4). Note that in cultures 1 and 2 *miR-634* overexpression gave rise to a similar phenotype, however, they could only be tested once due to low cell numbers. When the data of all primary cultures were combined, *miR-634* overexpression was associated with a significant reduction in cellular viability of cells treated with 15 or 30 μM cisplatin (*p* = 0.002, *p* \< 0.001, respectively).
Next, we tested if *miR-634* could alter the response towards other anticancer compounds. *miR-634* overexpression gave rise to significant increases in carboplatin sensitivity in cultures 3 and 7 (Fig. [3c](#Fig3){ref-type="fig"}, Additional file [1](#MOESM1){ref-type="media"}: Figure S5) as well as reduced resistance in two other cultures (which could only be tested once). We could not examine other cultures because of low cell numbers. In addition, *miR-634* overexpression gave rise to a significant increase in doxorubicin sensitivity (*p* = 0.004), whereas there was no significant reduction in sensitivity for paclitaxel (*p* = 0.335) (Fig. [3d, e](#Fig3){ref-type="fig"}). These data indicate that *miR-634* overexpression sensitizes resistant primary ovarian cancer cells to cisplatin, carboplatin and doxorubicin.
MiR-634 modulates the expression of key proliferation genes {#Sec7}
-----------------------------------------------------------
MiRNAs mainly act by downregulating the expression of protein-coding genes. In order to identify target genes, we performed a pathway analysis on genes predicted to be a *miR-634* target. Among the most enriched Gene Ontology (GO) terms were 'positive regulation of cyclin-dependent protein kinase activity' and 'regulation of the G1-S transition of the mitotic cell cycle' (Additional file [1](#MOESM1){ref-type="media"}: Figure S6, Table S3). Notably, Cyclin D isoforms CCND1 and CCND2 were associated with these GO terms, which activity is required for the G1 to S phase transition. Overexpression of *miR-634* resulted in a decreased CCND1 protein level in A2780, A2780 DDP and primary tumor cells (derived from patient 3) (Fig. [4a](#Fig4){ref-type="fig"}) and this may explain the slight increase in G1 cells at 48 h after transfection (Fig. [1a](#Fig1){ref-type="fig"}).Fig. 4*MiR-634* regulates genes involved in cell cycle regulation and the Ras-MAPK pathway. Ovarian cancer cell lines (A2780, A2780 DDP) and patient derived tumor cells (culture \#3) were transiently transfected with scrambled control (-) or *miR-634* mimic (+). 48 h after transfection protein lysates were analyzed for the expression of putative *miR-634* target genes by Western blotting (A, B) or RT-PCR (C). **a** Transfection with *miR-634* reduces Cyclin D1 protein levels in the A2780 cell lines and in primary ovarian cancer cell cultures. β-actin is used as a loading control. **b** *MiR-634* overexpression reduces protein levels of GRB2, ERK2, RSK1 and RSK2, components of the Ras-MAPK pathway in the A2780 and A2780 DDP cell lines and primary ovarian cancer cell cultures. β-actin is used as a loading control. **c** *MiR-634* overexpression lowers mRNA levels of CCND1, GRB2, ERK2, RSK1 and RSK2 in the A2780 DDP cell lines 48 h after transfection. Expression was normalized to HPRT and GAPDH expression. Depicted are average values ± SD (*n* = 2). The mRNA level in cells transfected with scrambled mimic is set at 100 %. **d** *MiR-634* binds directly to elements of the 3'UTR of CCND1, GRB2, ERK2 and RSK2. A2780 DDP cells were transfected with a scrambled mimic or a *miR-634* mimic. After 8 h, the same cells were transfected with a *Renilla* luciferase reporter construct (psiCHECK™-2) containing a region of 500 bp surrounding the predicted target sites (Additional file [1](#MOESM1){ref-type="media"}: Figure S7), or a luciferase construct in which the *miR-634* binding sites were mutated. Since RSK2 contains both a canonical and a non-canonical *miR-634* binding site, one construct was generated that contained a mutation in the canonical binding site (RSK2_2 mut 1) and another construct that contained mutations in both the canonical and non-canonical site (RSK2_2 mut 2). The Renilla luciferase activity was measured and normalized using the Firefly luciferase activity. For each construct, the relative luciferase activity of cells transfected with the scrambled mimic was set at 1. Depicted are the average values ± SD (*n* = 3). \* *p* \< 0.05, \*\* *p* \< 0.01
Another GO term that was enriched was 'positive regulation of proliferation'. Several proliferation pathways were predicted to be regulated by *miR-634*, most notably the Ras-MAPK pathway, of which *GRB2*, *N-RAS*, *RAF*, *ERK2*, *RSK1* and *RSK2* (Additional file [1](#MOESM1){ref-type="media"}: Table S3 and S4, Additional file [1](#MOESM1){ref-type="media"}: Figure S6) are potential targets. Concurringly, overexpression of *miR-634* gave rise to reduced protein levels of GRB2, ERK2, RSK1 and RSK2 in the A2780 cell lines and primary cell culture 3 (Fig. [4b](#Fig4){ref-type="fig"}). Since miRNAs can regulate target expression by translational repression and/or mRNA degradation, we also examined target mRNA levels. Whereas there was only a slight decrease in *CCND1* mRNA levels in *miR-634* transfected cells, we observed a 25--75 % decrease in *GRB2*, *ERK2*, *RSK1* and *RSK2* mRNA levels (Fig. [4c](#Fig4){ref-type="fig"}).
We next tested whether *miR-634* could directly regulate these genes. For each of the potential target genes, luciferase constructs were generated that express part of the 3'UTR containing the *miR-634* target site (Additional file [1](#MOESM1){ref-type="media"}: Figure S7). As *ERK2* and *RSK2* contain respectively three and two canonical *miR-634* target sites, the 3'UTR surrounding these sites was cloned separately. As expected, *miR-634* could inhibit the 3'UTR of *CCND1* and *GRB2*, the *ERK2* 3'UTR (constructs ERK2_1 and ERK2_2) and the *RSK2* 3'UTR (construct RSK2_2) (Additional file [1](#MOESM1){ref-type="media"}: Figure S8). In contrast, we observed no repression of the *miR-634* target site in the *RSK1* 3'UTR, suggesting that the effect of *miR-634* on *RSK1* protein and mRNA levels may be indirect.
To confirm specific binding, the *miR-634* binding site was mutated in the relevant constructs (CCND1, GRB2, ERK2_1, ERK2_2, RSK2_2). Since construct RSK2_2 contained two potential binding sites (Additional file [1](#MOESM1){ref-type="media"}: Figure S7), we created one construct with a mutation in the canonical site only (RSK2_2 mut 1) and another construct with mutations in both sites (RSK2_2 mut 2). As indicated in Fig. [4d](#Fig4){ref-type="fig"}, *miR-634* could no longer repress the mutated 3'UTRs of *CCND1*, *GRB2*, and *ERK2*. Interestingly, mutation of both sites in RSK2_2 was necessary to completely prevent *miR-634* binding.
As an alternative approach to inhibit *miR-634* function we assessed the effect of a specific antisense inhibitor on the luciferase activity of the wild-type 3'UTRs of *miR-634* target genes. As shown in Additional file [1](#MOESM1){ref-type="media"}: Figure S9, inhibition of endogenous *miR-634* led to increased luciferase activity with the RSK2_2 and, to a lesser extent, GRB2 constructs, indicating that endogenous *miR-634* regulates the expression of these target genes. In this setup we did not find an increased luciferase activity for the ERK2 or CCND1 constructs, and therefore we cannot confirm that ERK2 and CCND1 are endogenous *miR-634* targets in the A2780 DDP cell line. We next examined the relation between *miR-634* expression and expression levels of its target genes in a large cohort of ovarian serous cystadenoma's (TCGA dataset) (Additional file [1](#MOESM1){ref-type="media"}: Figure S10). Unfortunately, in the majority of tumors *miR-634* expression was not detected above background. However, when we examined the 23 tumors with high *miR-634* expression (expression \> mean + stdev) we found a significant negative correlation between *miR-634* levels and RSK2 (*p* = 0.002). We also found an inverse relation between *miR-634* and ERK2 (*p* = 0.091 or *p* = 0.310 for different probes), RSK1 (*p* = 0.372) and GRB2 (*p* = 0.614), but in this set of 23 expressing tumors this relation was not significant. It is possible that in a larger dataset this relation becomes significant. We did not observe an inverse relation between *miR-634* and CCND1 expression in this dataset.
Altogether, our findings show that overexpression of *miR-634* leads to direct repression of *RSK2*, *CCND1, GRB2* and *ERK2* in ovarian cancer cell lines and ovarian cancer cells derived from patients. Repression of the Ras-MAPK pathway may contribute to the decrease in cellular proliferation observed upon *miR-634* overexpression (Fig. [1b, c, d](#Fig1){ref-type="fig"}). In line with this, we observed that the MEK inhibitor PD0325901 reduced cellular viability at 48H after treatment (Additional file [1](#MOESM1){ref-type="media"}: Figure S11A). In addition, Ras-MAPK signaling may contribute to cisplatin resistance \[[@CR19]\]. Our previous work has demonstrated that RSK2 depletion enhances cisplatin sensitivity \[[@CR20]\]. Concurringly, PD0325901 augmented cisplatin toxicity at 24H after treatment (Additional file [1](#MOESM1){ref-type="media"}: Figure S11B).
Discussion {#Sec8}
==========
The aim of this study was to discover miRNAs that affect the response of ovarian cancer cells to cisplatin chemotherapy. We compared the expression of three cisplatin-sensitive and --resistant cell line pairs and identified *miR-634* as a miRNA that can modulate the sensitivity to various drugs in ovarian cancer cells.
*miR-634* is located on chromosome 17 within intron 15 of *PRKCA* (Protein Kinase C α), and is only conserved in primates. *miR-634* has been first detected in colon cancer cells \[[@CR21]\] via miRNA serial analysis of gene expression (miRAGE). Afterwards, *miR-634* has been identified as a miRNA able to regulate the expression of the androgen receptor (AR) in prostate cancer cells \[[@CR22]\]. Repression of AR resulted in a reduced viability, however, the effect of *miR-634* overexpression was stronger than the effect of AR siRNA, suggesting that *miR-634* may target other survival genes as well. We now report that *miR-634* overexpression results in downregulation of multiple genes of the Ras-MAPK pathway, an important cell proliferation pathway that is activated in many types of cancer \[[@CR8]\].
The observation that *miR-634* is able to regulate multiple Ras-MAPK pathway genes instead of one key mediator is thought-provoking. Interestingly, many miRNAs appear to synergistically regulate a set of genes that participate in similar processes, such as the *miR-17-92* cluster (regulates genes involved in growth control) \[[@CR23]\], *let-7* (represses Ras and its downstream target HMGA2) \[[@CR24], [@CR25]\] and the *miR-200* family (represses the epithelial-to-mesenchymal transition (EMT) \[[@CR26]--[@CR29]\]. MiRNAs may thus allow cells to effectively switch off similar signaling pathways in response to changing circumstances, and this may be especially relevant for proliferative pathways, which activity may need to be tightly controlled. In addition, it may be advantageous for cells to switch off production of proteins for signaling pathways that are not active.
Repression of oncogene activity often leads to cell death in cancer cells, a phenomenon known as 'oncogene addiction' \[[@CR30]\]. The reduction in cell viability observed in the ovarian cancer cell lines upon overexpression of *miR-634* may be caused by repression of the Ras-MAPK pathway. Interestingly, the effects of *miR-634* overexpression in primary cultures are much less pronounced, suggesting that these cells depend less on proliferation signaling. In support of this theory, the ascites derived tumor cells divided more slowly than the ovarian cancer cell lines, and the fastest growing cultures (cultures 3 and 7) showed the largest reduction in cell viability upon *miR-634* overexpression.
We describe that *miR-634* transfection results in enhanced cisplatin sensitivity. Intriguingly, this effect of *miR-634* overexpression is most apparent in resistant ovarian cancer cell lines, and also occurs in tumor cells derived from ascites. The *miR-634* mediated repression of the Ras-MAPK pathway might contribute to the sensitization \[[@CR19]\], which is supported by our finding that a MEK inhibitor enhances cisplatin sensitivity. Both ERK2 and RSK2 can inhibit several pro-apoptotic genes \[[@CR31]--[@CR35]\], and as a consequence of repression by *miR-634*, downregulation of ERK2 and RSK2 might lower the threshold for apoptosis upon treatment with cytotoxic therapy (Fig. [5](#Fig5){ref-type="fig"}). Indeed, our previous study shows that depletion of RSK2 leads to increased cisplatin sensitivity \[[@CR20]\].Fig. 5*MiR-634* enhances the sensitivity for chemotherapy. By regulation of Cyclin D1 and the Ras-MAPK pathway, *miR-634* overexpression may affect the cell cycle profile and apoptosis. Furthermore, repression of the Ras-MAPK pathway, which inhibits several pro-apoptotic factors (e.g. Bad \[[@CR55]\], BimEL \[[@CR56], [@CR57]\]), can indirectly sensitize ovarian tumors to chemotherapeutics, such as cisplatin/carboplatin (Pt) and doxorubicin (dox)
Because of the pivotal role of the Ras-MAPK pathway in cell proliferation and cancer, inhibitors have been developed that target this pathway \[[@CR36]\]. However, the objective response rates are modest \[[@CR37]\], perhaps because tumors may rewire their signaling pathways \[[@CR38]\]. Since *miR-634* can inhibit several key proliferation factors simultaneously, acquired resistance is less likely to be a problem for a *miR-634* based therapy than it is for targeted agents. A next step would be to study the effect of *miR-634* overexpression on tumor growth and therapy response in a genetic mouse model for ovarian cancer. However, such a study is complicated by the fact that there is no high grade serous mouse model for ovarian cancer \[[@CR39], [@CR40]\], and *miR-634* has no murine orthologue.
Furthermore, *miR-634* levels in tumors may correlate with response to chemotherapy. Although some people have detected *miR-634* expression using miRAGE and microarrays \[[@CR22], [@CR41]\], these studies do not report quantitative PCR validation. In addition, we and a few others were unable to detect endogenous *miR-634* by RT-PCR \[[@CR21], [@CR42]\], despite evidence that the assay is functional (e.g. housekeepers RNU43, RNU48 were readily detected in our samples and we could find a *miR-634* signal in mimic-transfected cells; Additional file [1](#MOESM1){ref-type="media"}: Figure S12). However, some publications describe successful *miR-634* detection in RT-PCR assays \[[@CR43]--[@CR45]\]. An explanation for the discrepancy between studies could be that *miR-634* is modified under certain cell-type specific conditions. Of note, modifications at the 3' end are not uncommon for microRNAs \[[@CR46], [@CR47]\] and may prevent *miR-634* detection by RT-PCR. Therefore, in order to monitor *miR-634* levels in tumors, a specific high throughput *miR-634* detection tool needs to be developed.
Conclusions {#Sec9}
===========
In summary our data indicate that *miR-634* is an important player in cisplatin-resistance. First of all, *miR-634* was the only miRNA that was commonly downregulated in three cisplatin/sensitive cell line pairs. Overexpression of *miR-634* transiently inhibited G1-S cycle progression and enhanced apoptosis of ovarian cancer cells. Furthermore, *miR-634* enhanced the chemotherapy response of cisplatin-resistant ovarian cancer cell lines and drug resistant patient-derived primary tumor cells. In addition, we observed that *miR-634* overexpression in ovarian cancer cell lines and patient samples negatively regulates important cell-cycle genes (CCND1) and Ras-MAPK pathway components (GRB2, ERK2, RSK1 and RSK2). Inhibition of the Ras-MAPK pathway resulted in increased sensitivity to cisplatin, suggesting that the *miR-634*-mediated repression of this pathway is responsible for the effect of *miR-634* on cisplatin resistance. In the future, therapeutic delivery of this miRNA to drug resistant ovarian cancer cells may help to resensitize patients to treatment.
Methods {#Sec10}
=======
Cell culture and reagents {#Sec11}
-------------------------
The ovarian carcinoma cell line A2780, colon carcinoma cell line HCT8, bladder carcinoma cell line T24 and their cisplatin-resistant derivatives A2870 DDP, HCT8 DDP, and T24 DDP10 have been described before \[[@CR14], [@CR15], [@CR48], [@CR49]\]. Resistant cell lines were routinely challenged with cisplatin. Ovarian cancer cell lines OV56 and OAW42 were purchased from the ECACC (Salisbury, UK) and TOV112D, TOV21G were obtained from the ATCC (Manassas, VA, USA). Characteristics of the ovarian cancer cells were published in \[[@CR50]\] and the authenticity of cell lines was verified by STR analysis. All cell lines were cultured in RPMI 1640 Glutamax (Invitrogen, Bleiswijk, The Netherlands) supplemented with 10 % FBS (Greiner, Alphen a/d Rijn, The Netherlands) and Penicillin/Streptomycin (final concentration 100 IE each, Sigma, Zwijndrecht, The Netherlands). Cisplatin, Carboplatin, Paclitaxel and Doxorubicin were obtained from Pharmachemie, Haarlem, the Netherlands. PD0325901 (cat \# PZ0162) was obtained from Sigma-Aldrich.
Isolation and culture of primary ovarian tumor cell cultures occurred as described before \[[@CR16], [@CR17]\]. Briefly, 25 mL ascites fluid was mixed 1:1 with MCDB105/M199 medium supplemented with 0.5 μg/mL Fungizone and 50 μg/mL Gentamycine (all obtained from Sigma-Aldrich), and incubated at 37 °C, 5 % CO~2~, in T75 flasks for 4 days. Afterwards, the medium was replaced with fresh MCDB105/M199 medium and replaced twice weekly until cells were confluent. Cells were split 1:2-1:3 for up to 6 passages. To verify that the cultures contain tumor cells, cytospins were analyzed for epithelial markers (pan-keratin and EpCAM) and p53.
The study was approved by the Medical Ethical Committee of the Erasmus University MC (MEC-2008-183).
Microarray and data analysis {#Sec12}
----------------------------
Total RNA was isolated with RNA Bee (BioConnect, Huissen, the Netherlands), and 1 μg RNA was labeled with the Cy3-TM ULS labeling kit (Kreatech Biotechnology, Amsterdam, The Netherlands), according to the manufacturer's instructions. The RNA was hybridized with the LNA-based capture probe set (Exiqon, Vedbaek, Denmark) version 10 (annotation version 13). This probe set consists of 1344 probes including 725 human miRNAs, which are spotted in duplicate. Spots were quantified with the Imagene software (BioDiscovery), obvious outliers were removed and quantile normalization was performed.
miRNA profiling experiments were performed with RNA isolated at two (HCT8 and T24) or three (A2780) different passages. The average miRNA expression in the sensitive cell line was compared with the average expression in the resistant cell line and the fold-change was calculated. The microarray expression data has been deposited to the Gene Expression Omnibus (GEO) data repository (accession number GSE54665).
RT-PCR {#Sec13}
------
0.5-1 μg RNA was reverse transcribed using random primers (Applied Biosystems, Bleiswijk, The Netherlands). 45 ng cDNA was used in a real-time PCR reaction using Taqman® assays-on-demand (*CCND1*, *ERK2*, *GRB2*, *RSK1*, *RSK2*). Expression was normalized to *HPRT* and *GAPDH* expression using the comparative C~T~-method \[[@CR51]\]. MiRNA expression was determined using Taqman® miRNA assays (Applied Biosystems) according to the manufacturer's protocol. In brief, 50 ng of total RNA was reverse transcribed using specific miRNA primers. The cDNA was used as input in a quantitative real-time PCR. RNU43 and RNU48 expression were used for normalization using the comparative C~T~- method \[[@CR51]\].
Transfection {#Sec14}
------------
A miRIDIAN mimic for *miR-634* (C-300961-01), a scrambled mimic (Mimic negative control \#1; CN-001000-01), a miRIDIAN hairpin inhibitor for miR-634 (IH-300961-03), a scrambled miRIDIAN hairpin inhibitor (negative control; IN-001005-01) and transfection controls (miRIDIAN mimic with Dy547; CP-004500-01/miRIDIAN hairpin inhibitor with Dy547; IP-004500-01) were obtained from Dharmacon (Epsom, UK). The seeding concentrations (cells/well) in 24 well plates were for A2780 and A2780 DDP 4.5\*10^4^, for OV56 and TOV21G 4\*10^4^, for OAW42 3\*10^4^, for TOV112D 6\*10^4^ and for primary ovarian cell cultures 4\*10^4^ in a final volume of 450 μL medium without antibiotics. On the day after seeding, 50 μL of a mixture of Dharmafect 1 (final concentration 0.3 % (v/v)) and mimic (final concentration 50 nM) or inhibitor (final concentration 25 nM) in serum-free medium was added dropwise to each well. Under these conditions, the transfection efficiency was over 90 % as determined using fluorescently labeled mimics. 48 h after transfection, drugs were added in varying concentrations. After 24 h of continuous drug exposure, an MTT assay was performed \[[@CR52]\]. psiCHECK™-2 constructs were transfected using Fugene HD (Promega) according to recommendations by the manufacturer.
FACS analysis {#Sec15}
-------------
Forty-eight and seventy-two hours after miRNA transfection, cells were harvested and stained with FITC-Annexin V and Propidium iodide (PI) (FITC Annexin V Apoptosis Detection Kit I (BD Pharmingen, Breda, The Netherlands)) or, for cell cycle analysis, fixed in 70 % ethanol on ice, washed with PBS, then stained with PI (20 μg/mL in PBS-0.5 mL).
Platinum measurements {#Sec16}
---------------------
Platinum measurements in cell lysates were carried out essentially as described before using Atomic Absorption Spectrophotometry (AAS) \[[@CR53]\].
Cytospins and immunohistochemistry {#Sec17}
----------------------------------
Primary ovarian cancer cell cultures were harvested using a cell scraper and collected in tubes, washed and resuspended in PBS containing 1 % BSA. 50-100 μL (depending on cell number) was used for cytospin preparation. Slides were air-dried for 30 min, fixed O/N in 10 % formalin, then stored in 70 % ethanol at 4 °C, until IHC staining. IHC staining was performed using Ventana Benchmark ULTRA automated slide stainers, and antibodies against pan keratin (Neomarkers, Fremont,CA, USA, MS-343-P), EpCAM (DAKO, Heverlee, Belgium, M0804) and p53 (DAKO, M7001).
Pathway analysis {#Sec18}
----------------
The entire list of predicted *miR-634* targets in Targetscan v6.0 ([http://www.targetscan.org](http://www.targetscan.org/)) was used for pathway analysis using the analysis wizard of DAVID Bioinformatics Resources (<http://david.abcc.ncifcrf.gov/tools.jsp>). Analyzed were enriched GO-terms (BP_fat; Additional file [1](#MOESM1){ref-type="media"}: Table S3) and enriched pathways (BIOCARTA, KEGG; Additional file [1](#MOESM1){ref-type="media"}: Table S4).
Western blot analysis {#Sec19}
---------------------
Fifteen to twenty μg of total protein of each sample was subjected to SDS-PAGE/Western blotting. Specific proteins were detected with antibodies against mouse GRB2 (BD Pharmingen, 610111), mouse RSK2 (Santa Cruz, Heerhugowaard, The Netherlands, SC-9986), rabbit ERK1/2 (Cell signaling, Leiden, The Netherlands, \#9102), rabbit RSK1 (Santa Cruz, SC-231), rabbit CCND1 (Thermoscientific, Etten-Leur, The Netherlands, RM-9104-S) and mouse anti-β-actin (Sigma, A5441). Secondary HRP-conjugated antibodies used were Goat-anti-mouse (Santa Cruz, SC-2005) and Goat-anti-rabbit (Jackson ImmunoResearch, Westgrove, PA, USA, 111-035-144).
Cloning {#Sec20}
-------
Parts of the 3'UTR of CCND1, GRB2, ERK2, RSK1 and RSK2 (see Additional file [1](#MOESM1){ref-type="media"}: Figure S7) were PCR amplified from human genomic DNA (Promega) introducing a XhoI (5'-end) and a NotI site (3'-end). The PCR products were cloned in pCR®-Blunt, followed by XhoI and NotI restriction digests and ligation in psiCHECK™-2 (Promega). The constructs were verified by sequencing. The putative *miR-634* target sites were mutated, forming a SmaI site to efficiently screen for mutants, in the CCND1, GRB2, ERK2 and RSK2 3'UTR constructs, using the Quickchange site-directed mutagenesis kit (Stratagene; Additional file [1](#MOESM1){ref-type="media"}: Figure S7).
Luciferase assay {#Sec21}
----------------
Forty-eight hours after transfection with the miRNA mimics/inhibitors and the psiCHECK2 constructs, a luciferase assay (Promega; Dual-Luciferase Reporter assay) was carried out according to the manufacturer's instructions. The *Renilla* luciferase expression was normalized on the Firefly luciferase signal.
Statistical analysis {#Sec22}
--------------------
A paired SAM analysis \[[@CR54]\] was performed to compare miRNA expression profiles of cisplatin sensitive and resistant ovarian (A2780/A2780 DDP), colon (HCT8/HCT8 DDP) and bladder cancer cell lines (T24/T24 DDP). Two-tailed paired sample T-tests were used to assess whether differences were consistent between the scrambled mimic and *miR-634* (Figs. [1](#Fig1){ref-type="fig"}, [2](#Fig2){ref-type="fig"}, [4d](#Fig4){ref-type="fig"}). Two way ANOVA's were used to examine whether *miR-634* overexpression had a significant effect on drug sensitivity, independent of the ascites batch. Furthermore, the effect of *miR-634* overexpression on drug sensitivity in individual ascites cultures was assessed using paired sample T-tests (Fig. [3](#Fig3){ref-type="fig"}, Additional file [1](#MOESM1){ref-type="media"}: Figure S3 and S4). *miR-634* and CCND1, ERK2, RSK1 and RSK2 expression was examined in 535 ovarian serous cystadenoma's (TCGA dataset, (<http://cancergenome.nih.gov/>). Spearman rank tests were used to correlate target gene expression in tumors with high *miR-634* expression (expression \> mean + stdev) (Additional file [1](#MOESM1){ref-type="media"}: Figure S10).
Additional file {#Sec23}
===============
Additional file 1: Table S1.Cisplatin sensitivity of cell lines. **Table S2.** miRNAs that show a ≥1.5 fold change of expression in cisplatin sensitive and resistant cell lines. **Table S3.** GO-terms that are enriched for miR-634 targets. **Table S4.** An overview of KEGG and Biocarta pathways that are enriched for miR-634 targets. **Figure S1.** miR-634 is the only miRNA showing consistent changes in cisplatin sensitive and resistant cell lines. **Figure S2.** Platinum levels in A2780 DDP cells transfected with miR-634 or scrambled controls. **Figure S3.** Immunohistochemical staining of primary ovarian cancer cell cultures. **Figure S4.** miR-634 enhances cisplatin sensitivity. **Figure S5.** miR-634 enhances carboplatin sensitivity. **Figure S6.** Predicted miR-634 targets. **Figure S7.** Overview of the location of the miR-634 target sites in the 3'UTR of its potential target genes. **Figure S8.** Effect of miR-634 overexpression on luciferase activity of psiCHECK™-2 reporter constructs containing part of the 3'UTR of miR-634 predicted targets. **Figure S9.** Effect of miR-634 inhibitors on luciferase activity of psiCHECK™-2 reporter constructs containing part of the 3'UTR of miR-634 predicted targets. **Figure S10.** Correlation between miR-634 expression and its target gene expression in 23 miR-634 expressing primary ovarian cancer samples (TCGA dataset). **Figure S11.** Effect of PD0325901 on cell survival and cisplatin sensitivity. **Figure S12.** Detection of miR-634, RNU43 and RNU48 by RT-PCR. (PDF 1609 kb)
**Competing interests**
The authors declare that they have no competing interests.
**Author's contributions**
MvJ designed and executed experiments, analyzed and interpreted the data and drafted the manuscript. PvK and AB performed experiments. WvIJ provided the micro-array facility. JH, RM, JP, EB and JV contributed to the design of the experiments and interpretation of the data. EW conceived, designed and coordinated experiments, helped interpreting the data and drafting the manuscript. All authors have read and approved the manuscript.
The authors are grateful to Dr. Seiji Nato (Kyushu University, Fukuoka, Japan) and to Dr. Kevin Scanlon (Keck Graduate Institute of Applied Life Sciences, Claremont, CA, USA) for the kind gifts of the cisplatin sensitive/resistant cell line pairs. Thanks are due to Mariël Brok for technical assistance in early experiments, Peter de Bruijn for platinum measurements, the department of Pathology for performing immunohistochemistry on primary ovarian cancer cells, Patricia Ewing for pathological analysis of the ovarian cancer tumors, and Wendy Onstenk and Bianca Mostert for providing us with the ascites samples. The results shown here are in part based upon data generated by the TCGA Research Network: <http://cancergenome.nih.gov/>. This project was financially supported by a grant from the Dutch Cancer Society EMCR 2007-3794.
| {
"pile_set_name": "PubMed Central"
} |
Bovine ovariectomy is considered a necessary farm animal management technique, not only for research on reproductive endocrinology \[[@r10], [@r13]\], but also for improving feeding efficiency in feeder cattle \[[@r15]\] and for treating ovarian diseases, such as ovarian tumors and granulosa cell tumors \[[@r3], [@r7], [@r15]\]. Ovariectomy used to be performed by colpotomy or laparotomy \[[@r2], [@r5], [@r12]\]. In these methods, the operator pulls the ovary into the vagina, where it is ligated and ablated under direct vision or with the aid of a Willis spay instrument or Kimberling-Rupp (K-R) spay instrument. During ovariectomy by colpotomy, postoperative bleeding can easily go unnoticed and can thus be fatal. Ovariectomy by laparotomy is performed under direct vision, making it possible to reduce the risk of complications; however, the size of the cow or a small uterus (such as during non-pregnant stages) may make it difficult to confirm the ovary position under direct vision, sometimes making it impossible to complete the desired treatment. Furthermore, ovariectomy by colpotomy or laparotomy increases the risk of adhesions or bleeding from the ovarian pedicle; in colpotomy, peritonitis may result in life-threatening complications \[[@r5], [@r12]\].
In contrast, laparoscopic ovariectomy offers various advantages, which are being increasingly recognized. Minimal invasion of the abdominal cavity is not only esthetically advantageous, but also therapeutically beneficial. Because this method causes little invasiveness and pain, patients are admitted to a hospital for shorter periods and return to normal production levels quicker \[[@r8], [@r9], [@r11]\]. In addition, the use of a laparoscopic camera and specialized forceps makes it possible to perform ovariectomy in a reliable manner.
However, in laparoscopic ovariectomy, the uterus and ovaries are difficult to visualize when over fattening results in pronounced fat deposits in organs or when insufficient withholding of food makes it impossible to secure sufficient intra-abdominal space; as a result, the surgery is prolonged, thereby placing significant burden on the cow.
Rectal palpation is a physical examination commonly used for the diagnosis of reproductive disturbance and pregnancy \[[@r6]\]. Using this technique, the ovary or uterus can be palpated within a few sec. Thus, we investigated the effect of transrectal guidance of the ovaries by an assistant on operative time during laparoscopic ovariectomy.
All animal experiments were performed in compliance with the Guide for the Animal Care and Use Committee at Azabu University, School of Veterinary Medicine (No. 160829-3). In total, 24 Holstein dairy cows (28.2--98.9 months, 1--5 parities, weight 406--698 kg, BCS 2.75--4.00) from a commercial dairy farm were included in the present study ([Table 1](#tbl_001){ref-type="table"}Table 1.Description of the experimental cows (n=24; 12 control cows and 12 guidance cows) used for laparoscopic ovariectomyNo.GroupBreedAge (Months)ParityBody weight (kg)BCS^a)^1ControlHolstein40.02406.02.752Holstein28.21454.02.753Holstein33.82478.03.004Holstein36.11488.02.755Holstein29.61555.03.506Holstein57.82559.03.007Holstein37.61582.03.508Holstein52.03587.03.259Holstein38.21593.53.5010Holstein92.93600.03.7511Holstein78.44643.03.7512Holstein40.81698.04.00Mean ± SD47.1 ± 20.11.8 ± 1.0553.6 ± 83.33.29 ± 0.413GuidanceHolstein29.31462.02.7514Holstein36.62413.02.7515Holstein39.71456.02.7516Holstein31.31481.02.7517Holstein58.62527.03.0018Holstein55.22530.02.7519Holstein66.13545.03.5020Holstein45.11560.03.5021Holstein71.53575.03.5022Holstein98.95595.53.7523Holstein46.81617.04.0024Holstein66.83645.03.75Mean ± SD53.8 ± 20.02.1 ± 1.2533.9 ± 70.23.23 ± 0.5a) Body condition score.). These cows were randomly divided into two groups: with (guidance group) and without (control group) transrectal guidance of the ovaries by an assistant (12 cows each). All cows were submitted to general status examinations and blood tests prior to surgery and on days 1 and day 14 after surgery. All surgeries were performed by a veterinary surgeon who performs laparoscopic examinations and surgeries on a daily basis, a camera assistant, and a surgical technician. Transrectal manipulation of the ovaries in the guidance group were also performed by a veterinary technician who performs transrectal examinations on a daily basis.
The cows were allowed free access to water at all times but were prevented from feeding for 24 hr prior to surgery. Although one study reported that a sufficient visual field could not be obtained if food was withheld in the rumen for less than 36 hr \[[@r1]\], we were able to procure sufficient intra-abdominal space by withholding food for only 24 hr prior to surgery. First, 5,000 IU/kg procaine penicillin G (Kyoritsu Seiyaku Inc., Tokyo, Japan) was injected into the muscles 1 hr before surgery and for 3 days after surgery. The cows were restrained in a standing position. Local anesthesia consisted of a lumbar epidural injection of 2% lidocaine hydrochloride (Pfizer Japan Inc., Tokyo, Japan) 0.2 mg/kg and 2% Xylazine hydrochloride (Selactar 0.2%; Bayer, Ltd., Tokyo, Japam) 0.05 mg/kg were administrated between the first and second lumber vertebrae using the epidural needle (16G, 12 cm in length; Hakko Syoji., Tokyo, Japan).
With a left abdominal incision, the risk of tissue damage is lower than if a right abdominal incision was utilized, and the ovaries and uterus can be viewed more easily \[[@r14]\]. An approach from the right abdominal wall is obstructed by the omentum majus; however, an approach from the left flank enabled sufficient visualization of the uterus and ovaries. We clipped the hair, cleaned and disinfected the region extending from the lumbar spine in the left abdomen to the lower abdomen across a width of 25 cm, from the last rib to the tuber coxae; at the site of the ports, an infiltration anesthesia with lidocaine 2% was performed. The laparoscopic camera port (port 1) site was approximately 15 cm rostral from the left tuber coxae and approximately 10 cm towards the lower abdomen from the transverse processes of the lumbar vertebrae. After inserting the trocar (11 mm in diameter, 20 cm in length; KARL STORZ GmbH & Co. KG, Tuttligen, Germany), CO~2~ gas was injected into the abdominal cavity with a pressure of 10 mmHg. There were two surgical access ports: one (port 2) situated 10--15 cm towards the lower abdomen from port 1 and the other (port 3) situated approximately 10 cm towards the lower abdomen from port 2 and approximately 5 cm cranial to the vertical line from the tuber coxae ([Fig. 1A and 1B](#fig_001){ref-type="fig"}Fig. 1.Images showing portals for laparoscopic ovariectomy via the left flank. Image (A) and illustration (B) showing the surgical site after insertion of the laparoscope and forceps. The laparoscope (①) and two forceps portals (②, ③) were established. The port (①) was approximately 15 cm rostral from the left tuber coxae (a) and approximately 10 cm towards the lower abdomen from the transverse processes of the lumbar vertebrae (b). The port (②) situated 10--15 cm towards the lower abdomen from port 1 (c) and the port (③) situated approximately 10 cm towards the lower abdomen from port 2 (d) and approximately 5 cm cranial to the vertical line from the tuber coxae (e). The ovary was grasped with forceps and cauterized by a vessel-sealing device (C). Cran=cranial, Caud=caudal, LUH=left uterus horn, RUH=right uterus horn, OP=ovarian pedicle, OV=ovary, TC=tuber coxae.).
First, a 30° laparoscope (10 mm in diameter, 57 cm in length; KARL STORZ GmbH & Co. KG) was inserted through port 1. In the control group, two pairs of grasping forceps (HOPKINS Forceps, 43 cm in length; KARL STORZ GmbH & Co. KG) were introduced through ports 2 and 3 to identify the uterus, which was then grasped with forceps; we then followed along the right and left uterus horn until we confirmed the ovaries. In the guidance group, the designated assistant grasped the ovaries via a transrectal approach and pulled them to a position where they could be visualized with the laparoscopic camera. In both groups, once the ovaries were identified with the laparoscopic camera, the left ovarian parenchyma was grasped with the grasping forceps inserted through port 2; an injection cannula (43 cm in length; KARL STORZ GmbH & Co. KG) was then inserted through port 3, and 10 m*l* of lidocaine was injected into the mesosalpinx and the mesovarium. The ovary was pulled with the grasping forceps, and the extended mesovarium was cauterized close to the ovary with a vessel-sealing device (Ligasure Maryland 44; Medtronic plc, Dublin, Ireland) inserted through port 3. In the present study, we used a vessel-sealing device that can seal blood vessels of up to 7 mm in diameter. The ovarian artery is expected to have a diameter of 2.5 mm \[[@r4]\]. Once the cauterization site turned white, the mesovarium was ablated with an organ cutter ([Fig. 1C](#fig_001){ref-type="fig"}); while confirming that there was no bleeding from the ablation surface, this process was repeated until the ovary was completely detached. While continuing to grasp the detached ovary with the forceps, the ovary was removed from the abdominal cavity through port 2. The right ovary was also resected with the same procedure.
In the control group, if more than 90 min passed after the skin incision, the designated assistant performed transrectal guidance of the ovaries. Following surgery, the abdominal cavity was deflated through port 1, and the skin incision sites were closed with a stapler (WiSM Skin Stapler; Keisei Ika KK, Tokyo, Japan). For cows with large ovaries which therefore required a longer incision in the muscle in access port 3 for removal from the abdominal cavity, the muscle layer was continuously sutured with synthetic absorbable suture material (USP 4) before stapling the skin incision.
Blood samples were obtained from the jugular vein prior to surgery and 1 day and 14 days after surgery. Collecting samples were used for counting of red blood cells (RBC) and white blood cells (WBC), Platlet (PLT) and hematocrit (Ht) analysis by automated cell counter (PCE-170, ERMA Inc., Tokyo, Japan) within 30 min after collection.
All statistical analyses were performed using the statistical software (Statcel, 4th edition, OMS Publishing, Saitama, Japan). Data were tested normality distribution and as this was confirmed, Student's *t*-test was used to compare the mean values resulting from both treatments groups (control and guidance groups). Data in the present study were expressed as mean ± standard deviation of the mean. Statistical significance was defined as *P*\<0.05.
The time required from skin incision to the completion of bilateral ovariectomy in the control group (63 ± 25.2 min) was significantly longer than that in the guidance group (24 ± 6.6 min) (*P*\<0.01; [Fig. 2](#fig_002){ref-type="fig"}Fig. 2.Bar graph (mean ± SD) showing the surgery time of laparoscopic ovariectomy (control vs guidance group). Statistical differences between the groups were determined by the Student's *t*-test (*P*\<0.01).). In a previous study, the duration of ovariectomy in cows was reported to be 120--150 min \[[@r1]\]. In the present study, however, transrectal examination enabled us to complete ovariectomy in a much shorter period of time.
None of the 24 cows demonstrated any abnormalities at general status examinations or blood tests prior to surgery or at day 1 or day 14 after surgery; furthermore, no differences were observed in blood tests between the groups ([Table 2](#tbl_002){ref-type="table"}Table 2.Profile of blood examinations prior to surgery and on day 1 and day 14 after surgeryPre1 day14 daysControl (n=12)Guidance (n=12)Control (n=12)Guidance (n=12)Control (n=12)Guidance (n=12)RBC×10^4^/*µl*697 ± 115609 ± 246679 ± 141726 ± 122653 ± 93631 ± 109WBC/*µl*8,308 ± 2,6278,392 ± 6,8299,042 ± 3,9098,325 ± 4,3368,517 ± 3,8528,217 ± 2,730PLT×10^4^/*µl*49 ± 1343 ± 3859 ± 2740 ± 2230 ± 2042 ± 22Ht%35 ± 933 ± 1031 ± 634 ± 533 ± 433 ± 4Fibrinogenmg/d*l*500 ± 148500 ± 341583 ± 199492 ± 198267 ± 98327 ± 237RBC, red blood cell; WBC, white blood cell; PLT, platlet; Ht, hematocrit.).
In general, it was necessary to use large traumatic forceps with relatively large teeth to avoid losing hold of the uterine horn; this procedure can cause bleeding and lacerations in the uterine serosa \[[@r1]\]. However, in the present study, an assistant grasped the ovaries via a transrectal approach, which enabled us to avoid injuring the uterine serosa. We can therefore confirm that the transrectal assistance for ovariectomy by laparoscopy in cows helps to protect the cow from physical damage.
The local anesthesia is necessary when the ovariectomy is done even the method excluding the laparoscopy. The ovariectomy by colpotomy or laparotomy increases the risk of adhesions or bleeding from the ovarian pedicle \[[@r5], [@r12]\]. Even if the laparoscopic ovariectomy need special instruments, it is reduce the risk of complications because performed under direct vision. Additionally the small incisions and little pain are return to normal activity level quickly \[[@r8], [@r9], [@r11]\].
In conclusion, our results indicate that the combined technique of using laparoscopy and a transrescal assistant is effective for ovariectomy in cows because it reduces operative time, physical damage, and the burden on the operator.
The authors thank Takashi Ikeuchi (Covidien Japan Inc.) for technical support and student assistant at Azabu University for collecting samples and caring cows during this experiment.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Thyroid carcinoma is the most common endocrine malignancy with a prevalence of 335,000 and incidence of 37,200 in the United States in 2009 \[[@B1]\]. Differentiated thyroid carcinoma, namely papillary and follicular thyroid carcinoma, makes up about 94% of these cases. Despite the generally good prognosis of thyroid carcinoma, about 5% of patients will develop metastatic disease which fails to respond to radioactive iodine, exhibiting a more aggressive behavior. These patients will die of their disease \[[@B1]--[@B4]\].
85% of patients with differentiated thyroid carcinomas are cured with surgery, radioactive iodine, and TSH suppression. Of those that recur, the vast majority will recur in the neck, and best treatment options are surgical with potential further radioactive iodine. A small percentage of patients will develop or present with metastases and are more difficult to treat. When metastases have radioiodine avidity, prognosis is better, and further radioactive iodine may be used. However, when multiple doses of radioactive iodine have been tried or the patient has nonradioactive iodine avid disease, other options need to be considered. This paper will aim to discuss the treatment options of those patients with nonradioiodine avid, recurrent, or metastatic differentiated thyroid cancer.
2. Diagnosis of Recurrent/Metastatic Disease Extent {#sec2}
===================================================
Screening ultrasound of the neck and tumor marker (thyroglobulin) should be performed on all patients with differentiated thyroid cancer,per accepted guidelines \[[@B5]\]. The finding of an elevated thyroglobulin or thyroglobulin antibody in the face of a negative radioactive iodine scan is indicative of non-radioiodine avid residual or recurrent disease. Both ultrasound of the neck and thin spiral CT of the chest should be performed for detection of disease. If symptoms occur or the thyroglobulin is out of proportion to the amount of disease seen, other imaging can be ordered as dictated by the clinical scenario. Other imaging modalities include MRI of the brain, spine, bone scan, and ^18^FDG-PET/CT scans. [Table 1](#tab1){ref-type="table"} summarizes the imaging modalities used in thyroid cancer surveillance.
2.1. Role of Ultrasound {#sec2.1}
-----------------------
Ultrasonography (U/S) of the neck (thyroid bed and cervical neck compartments), as opposed to RAI scans, is recommended in the followup of these patients. This shift in practice is due to the fact that many recurrent tumors lose the ability to capture iodine, leading to false negatives. As many as half of the patients with findings of recurrence on U/S may have no uptake on radioiodine scanning or may have an undetectable serum thyroglobulin \[[@B6]\].
Recurrence of papillary thyroid carcinoma is most commonly in the neck (thyroid bed and lymph nodes), and hence, ultrasound is the mainstay of routine followup of these patients. U/S can be used to accurately diagnose and identify lesions in the neck as small as 3 mm. Routine use of U/S in the 3- to 12-month monitoring of patients with extrathyroidal invasion or local-regional nodal metastases \[[@B7], [@B8]\] is now recommended as part of consensus guidelines \[[@B5]\]. Although U/S can aid in distinguishing benign lesions from malignant lesions, FNA (U/S guided) is most helpful to definitively prove recurrent cancer. Thyroglobulin can be measured in the washout of the needle taken from neck lymph nodes \[[@B9]--[@B11]\]. This is especially helpful in cases where the FNA specimen is nondiagnostic.
2.2. CT and MRI {#sec2.2}
---------------
Other imaging techniques that can be used in individual cases of thyroid cancer followup include CT scan of the neck with IV contrast, CT scan of the chest, and magnetic resonance imaging (MRI). MRI and CT scan of the neck play important roles in the detection of recurrent disease although the sensitivity of these is not as well established as ultrasound for the detection of true thyroid cancer recurrences in the neck. CT and MRI of the neck are not recommended for routine use in the detection of recurrent disease but have the advantage of being much less operator dependent. If good ultrasonography is not readily available or deep posterior neck disease is suspected, CT and MRI of the neck can be used for the detection of disease.
The most common place for papillary thyroid carcinoma to metastasize outside the neck is the chest. CT scan of the chest may show macro- and micronodular pulmonary metastases that do not routinely take up iodine. This cross-sectional imaging of the chest is used for long-term followup when lung metastases are known or suspected based on elevations in thyroglobulin.
CT and MRI scans of other less commonly found sites of distant metastases include imaging of the brain, spine, abdomen, and pelvis as the clinical scenario dictates based on symptoms, clinical suspicion, or prior to initiation of various therapies.
2.3. ^18^FDG PET-CT {#sec2.3}
-------------------
Fluorodeoxyglucose positron-emission tomography (FDG-PET or PET)/CT imaging is an increasingly more useful tool in the detection of radioiodine-negative, thyroglobulin-positive thyroid cancer \[[@B12]--[@B15]\]. Thyroid carcinomas with little to no iodine activity tend to have higher glucose metabolism and positive FDG-PET scans \[[@B12], [@B16], [@B17]\]. This tends to be representative of tumor dedifferentiation. Patients with larger volumes of FDG-avid disease or higher SUVs are less likely to respond to radioiodine and have a higher mortality over a 3-year followup compared with the patients with no FDG uptake \[[@B18], [@B19]\]. Tumors that take up radioactive iodine are less likely to yield positive FDG PET scans \[[@B20]\].
PET/CT scans can be used for detection of occult recurrences or metastases \[[@B16], [@B21], [@B22]\] or to provide information about the biology of the metastatic disease and prognostic information. The latter is not standard practice, but several studies have now shown that FDG-PET correlates with the overall survival \[[@B14], [@B15], [@B21]\]. This information may be helpful to decide which patients warrant systemic treatment for their metastatic disease if they are refractory/resistant to radioactive iodine or have reached the maximum benefit from this treatment.
^18^FDG-PET has been approved for reimbursement for the detection of occult thyroid cancer in patients who have a thyroglobulin greater than 10 ng/mL and have negative radioiodine imaging. ^18^FDG-PET CTs are also used in those patients whose cancers are very poorly differentiated and make no thyroglobulin.
The overall sensitivity, specificity, and accuracy of ^18^F-FDG PET/CT in one series of 59 patients with radioiodine-negative, thyroglobulin-positive, recurrent disease were 68.4%, 82.4%, and 73.8%, respectively \[[@B23]\]. Other studies have shown a sensitivity of 70--95% and a specificity of 77--100% \[[@B12], [@B13]\]. FDG-PET is not sensitive enough to detect subcentimeter metastases, as it is common in metastatic papillary thyroid carcinoma and should be used in conjunction with CT chest imaging.
TSH stimulates ^18^FDG uptake by differentiated thyroid carcinoma \[[@B24]\], suggesting that PET scans may be more sensitive after TSH stimulation with rhTSH or withdrawal of thyroid hormone \[[@B24]--[@B26]\]. While rhTSH-stimulated PET-CT identified more total FDG-avid lesions compared to nonstimulated FDG-PET CT in a large multicenter study, it changed treatment planning only 6% of the time \[[@B27]\].
False positives, such as infections or granulomatous diseases/sarcoid or postoperative changes due to inflammation, amongst others, have been reported in thyroid cancer about 11--25% of the time suggesting that the malignant nature of the disease should be confirmed prior to further therapy \[[@B27]--[@B30]\].
3. Treatment of Advanced or Metastatic Thyroid Cancer {#sec3}
=====================================================
Most patients with differentiated thyroid cancers are rendered free of disease after surgery, radioactive iodine, and thyroid hormone suppression. Approximately 15--20% of patients will recur locoregionally or have distant metastases. Although it is the most effective medical treatment for differentiated thyroid carcinoma, only about 50--80% of primary tumors and their metastases take up radioactive iodine \[[@B7], [@B31]--[@B34]\], rendering this therapy ineffective in most cases. Thus, other treatment modalities such as surgery, external beam radiation, percutaneous ethanol injection therapy (PEIT), and systemic chemotherapy are indicated. [Table 2](#tab2){ref-type="table"} summarizes the indications for each of these therapies.
3.1. Surgery {#sec3.1}
------------
Surgery in advanced thyroid carcinomas is most commonly used for recurrent neck metastases and metastasectomies in selected sites. Recurrences in the neck are most commonly seen in the thyroid bed or regional lymph nodes. Although most occur within the first five years after diagnosis, late recurrences do occur. In one study with 40-year followup, 35% of patients recurred. Two-thirds of them were within the first decade after initial therapy, and two thirds were locoregional \[[@B35]\].
Surgery is considered first-line therapy in patients with gross nodal or recurrent neck disease. This can be followed by further radioactive iodine (if the recurrent tumors took up radioiodine prior to surgery) and thyroid hormone suppression. One-third to one half of patients may be free of disease in short-term followup \[[@B36]\]. If the gross tumors do not take up radioactive iodine from previous posttreatment scans or preoperative radioiodine scans are negative, further postoperative radioactive iodine will be of limited benefit and may increase the side effects of further iodine. Adverse events from further radioactive iodine include xerostomia, nasolacrimal duct obstruction, and secondary malignancies \[[@B37]--[@B42]\].
Surgery alone with complete ipsilateral compartmental dissections of involved areas or modified neck dissections as opposed to "berry picking" or selective lymph node resection procedures or ethanol ablation may be of benefit \[[@B43], [@B44]\]. It is not evident that recurrent locoregional disease in the setting of distant metastatic disease should be resected, unless there is airway or an other vital structural compromise. If the tumors invade the upper aerodigestive tract, a combined treatment modality of surgery and ^131^I (if tumors take up radioactive iodine) and/or adjuvant external beam radiotherapy is advised \[[@B45]--[@B48]\]. Due to potential morbidity from surgical resection of recurrent disease, these patients should be referred to centers with expertise in this area.
Surgery is also considered for isolated metastases, metastases in bone (especially if long bones, spine, or weight bearing), and brain (see [Section 3.3](#sec3.3){ref-type="sec"} below).
3.2. External Beam Radiotherapy {#sec3.2}
-------------------------------
External beam radiotherapy (EBRT) has a very specific role in the treatment of papillary thyroid carcinoma. Although it is somewhat controversial, retrospective studies have shown that it may be an effective adjuvant therapy to prevent local-regional recurrence in patients 45 years of age and older with locally invasive papillary carcinoma after surgery \[[@B49], [@B50]\]. Ten-year local relapse-free rates (93% versus 78%) and disease-specific survival rates (100% versus 95%) were higher in a subgroup of patients with papillary histology and presumed microscopic disease treated with EBRT \[[@B49]\]. Doses in the range of 40--50 Gy may aid in local-regional control in patients with papillary thyroid carcinoma who are over 45 years of age and have incomplete resection near the aerodigestive tract and/or those with gross extrathyroidal invasion with presumed microscopic residual disease. EBRT is generally avoided in patients under 45 years of age both because of their good prognosis and the potential late side effects of therapy including secondary malignancies. External beam radiation may also preclude further surgery in the future if the tumor recurs.
Acute complications of external beam radiotherapy include esophagitis and tracheitis. Long-term complications include neck fibrosis, xerostomia, dental decay, osteoradionecrosis, and the risk of tracheal stenosis \[[@B34]\]. Newer techniques to deliver radiation with fewer adverse events are being used in the treatment of cancer, including intensity-modulated radiation therapy (IMRT). Limited data in thyroid cancer with short-term followup suggests similar outcomes and may reduce chronic morbidity relative to conventional EBRT \[[@B51]\]. Many centers currently use IMRT as the radiation treatment of choice for thyroid carcinomas requiring EBRT \[[@B52]\].
3.3. Metastatic Sites Requiring Special Attention {#sec3.3}
-------------------------------------------------
Although most patients with metastatic disease will need systemic therapy, metastatic disease to certain sites deserves special attention.
### 3.3.1. CNS Metastases {#sec3.3.1}
Brain metastases more often occur in elderly individuals with more advanced disease and have an overall poorer prognosis \[[@B53]\]. Surgical resection significantly improves median overall survival from four to 22 months in patients with 1 or more brain metastases \[[@B53]\]. Current guidelines recommend resection when one CNS lesion is present \[[@B54]\]. Radioiodine therapy and/or external beam radiotherapy (with steroids to minimize tumor swelling) should be considered after surgical resection \[[@B55]\]. If CNS lesions are not surgically resectable or the morbidity from surgery is unacceptable, whole brain radiotherapy for numerous lesions or gamma knife radiosurgery to selected lesions should be used in conjunction with radioiodine if the tumor concentrates iodine \[[@B56]\]. If radioiodine is to be used, prior radiotherapy and concomitant steroids should be strongly considered to decrease tumor swelling \[[@B57]\].
### 3.3.2. Bone Metastases {#sec3.3.2}
Although bone lesions tend to concentrate radioiodine as well as lungs, there is complete resolution in less than 10% of the time. Metastases that cause pain or compression of spinal cord or other vital organs necessitate treatment. Symptoms from painful bone lesions or spinal-cord-compressing lesions may be relieved by surgical treatment. External beam radiation therapy (EBRT) or gamma knife radiosurgery has also been used successfully to render bone lesions pain-free. Arterial embolization has been used with successful reduction in pain and neurologic symptoms and can be used in conjunction with external beam radiation \[[@B58]\]. ^131^I treatment may follow surgical resection of distant metastatic disease if the tumor takes up radioactive iodine. A recent study found that patients with solitary bony metastases treated with I131 and surgery had a better prognosis than those who did not \[[@B59]\].
Intravenous bisphosphonates (pamidronate or zoledronic acid) are prescribed for painful bony metastases with some success as well. Orita et al., retrospectively, examined 50 patients with bony metastases from DTC and found that those who had received monthly infusions of zoledronic acid had significantly fewer-skeletal related events (defined as fracture, spinal cord compression, and hypercalcemia) than those who did not receive this drug \[[@B60]\]. Whether this treatment slows progression of bone metastases is not known. The most common adverse event associated with intravenous bisphosphonates is a transient flu-like syndrome, usually associated with the first administration, with symptoms dissipating and disappearing with subsequent infusions. Osteonecrosis of the jaw is less common but serious adverse event associated with intravenous bisphosphonates.
3.4. Systemic Therapy {#sec3.4}
---------------------
Lack of RAI uptake by distant metastases confers a poor prognosis. For example, patients with no RAI uptake in the lungs have a 10-year survival rate of 25% compared with 76% in those whose lung metastases have RAI uptake \[[@B61]\]. Pulmonary metastases that do not take up radioactive iodine do not typically respond to that radionuclide therapy, and these patients are at high risk of death \[[@B62]\].
Most diagnostic scan-negative, thyroglobulin-positive patients who have disease seen on other imaging modalities are not rendered disease-free by repeat radioiodine treatments, although tumor burden may decrease \[[@B63]\]. No survival advantage nor decrease in morbidity has been seen with repeat radioactive iodine therapies. Repeated doses of radioactive iodine have been used in patients considered to have radioactive noniodine-avid, thyroglobulin-positive disease with little clinical benefit \[[@B64]\]. Although controversial, a single dose of 100--150 mCi of radioactive iodine therapy can be given to a patient with elevated thyroglobulin and negative diagnostic scan. A posttreatment scan should be done, and if negative, further radioactive iodine should be avoided.
Repeated radioiodine therapy has adverse events including xerostomia, nasolacrimal duct obstruction with epiphora, and secondary malignancies \[[@B37]--[@B42]\]. Further radioactive iodine therapy should generally be avoided in these patients, and the use of systemic agents should be considered \[[@B65]\].
Because metastatic differentiated thyroid cancer can be stable and quiescent for many years, only patients with progressive or symptomatic disease should be treated with other systemic treatments. Systemic therapy with targeted agents or cytotoxic chemotherapy is the usual treatment of choice for RAI-refractory, progressive distant metastatic disease.
Clinical trials should consider first-line therapy for those patients who do not take up radioactive iodine. If a clinical trial is not available or the patient is not suitable for one, then off-label use of targeted therapies such as pazopanib, sorafenib, sunitinib, or cytotoxic chemotherapy should be considered \[[@B54]\].
### 3.4.1. Cytotoxic Chemotherapy {#sec3.4.1}
Traditional cytotoxic chemotherapies such as doxorubicin, taxol, and cisplatin are associated with a 25--37% partial response rate with rare complete remission \[[@B66]--[@B69]\]. Due to toxic side effects, short duration of responses, and low response rates, systemic cytotoxic chemotherapy is reserved for patients with rapidly progressive metastatic disease that is not suitable or nonresponsive to surgery, radioiodine, and external beam radiotherapy and those who cannot enter into clinical trials or use targeted agents (see below).
Systemic chemotherapy is used in certain cases of widespread progressive disease that is radioiodine resistant although available regimens have not been well studied and are not very effective to date. Doxorubicin is associated with a response rate of up to 40% for progressive differentiated cancers that do not respond to radioactive iodine \[[@B70], [@B71]\]. The recommended dosage is 60--75 mg/m^2^ every 3 weeks. Combination therapies are also used, but data are limited because of the small number of patients in reported series. Doxorubicin, epirubicin, taxol, and cisplatin have all been used in various combinations; responses do not seem to be any better than single agent with increased toxicities \[[@B68], [@B69], [@B72]\]. Response rates vary from 25 to 37% with mostly partial responses. Doxorubicin has also been used as a radiation sensitizer with not much different results from radiation alone. Patients with advanced progressive radioiodine nonresponsive disease should be considered for participation in clinical trials.
### 3.4.2. Newer Targeted Therapies {#sec3.4.2}
Newer approaches to thyroid carcinoma therapy include inhibition of the various metabolic pathways found to be altered in these cancerous cells. Prior to the discussion of the available and tried targeted therapies below, we briefly summarize the known aberrant pathways.
\(1\) Mutations and Promotion of Tumor GrowthFollicular cell tumorigenesis pathways have been key to the development of clinical trials testing novel therapies in the treatment of thyroid cancer. Mutations in either BRAF, RAS or RET/PTC rearrangements are present in most differentiated thyroid cancers \[[@B73]\]. Chromosomal rearrangement of the gene encoding the transmembrane tyrosine kinase receptors *ret* and *trk* is one identified early step in the development of these tumors. RET/PTC genetic alterations have been found in 40% and 60% of papillary carcinomas in adults and children, respectively, and are the most common mutation found in the Chernobyl radiation-induced thyroid carcinomas \[[@B74]--[@B76]\].Mutations and constitutive activation of the MAP kinase pathway have been of interest of late. BRAF (in papillary thyroid cancer) and RAS genes in the MAP kinase pathway normally code for growth and function in normal and tumor cells. BRAF mutations have been identified in approximately 45% or more of clinically evident papillary carcinomas and may behave more aggressively \[[@B77]--[@B79]\]. Activating mutations of RAS are more common in follicular variant PTC and follicular thyroid cancer \[[@B80]\] and may be a marker of more aggressive disease \[[@B81]\].Other discoveries include the dependence of tumors on angiogenesis. Angiogenesis is important for tumor cell growth, promotion, and development of metastases \[[@B82]\]. Vascular endothelial growth factor (VEGF), an important proangiogenic factor, binds to VEGF receptors that in turn can further activate MAP kinase signaling and promote further tumor growth. VEGF receptors play a contributory role in the development and progression of thyroid cancer \[[@B73], [@B83]\]. VEGF expression is associated with higher risk of recurrence and shorter disease-free survival \[[@B84], [@B85]\]. Like in other tumors, epigenetic modifications of chromosomal DNA and histones, including the promoter gene of the sodium-iodine symporter, may also play an important role in promotion of tumor growth.
\(2\) Therapeutic Options and Clinical TrialsPatients with progressive or symptomatic metastatic thyroid cancer that is deemed nonradioiodine responsive should be considered for treatment on a clinical trial \[[@B5]\]. The recent identification of the molecular and cellular pathogenesis of both the development and progression of cancer has led to development of newer molecular-targeted therapies. Oncogenic mutations in the MAP kinase pathway (BRAF and RAS), as well as the importance of vascular endothelial growth factor receptors in thyroid cancer (mentioned above), have led to several clinical trials with small molecule inhibitors. These agents inhibit multiple kinases and can affect multiple signaling pathways. Tyrosine kinase inhibitors (TKIs) are orally administered and generally well tolerated. Immunomodulators, other oncogene inhibitors, and modulators of growth or apoptosis are all under investigation as well. Various clinical trials in the United States and Europe are recruiting differentiated thyroid cancer patients with radioiodine-negative progressive disease. The following are among the ones that have raised maximal interest and are summarized in [Table 3](#tab3){ref-type="table"}.
\(3\) Commercially Available TKIs(a) *Sorafenib*. Sorafenib is an oral, small molecule tyrosine kinase inhibitor that inhibits RET, BRAF, and VEGF receptors 2 and 3. It is currently approved in the United States for advanced renal cell carcinoma and unresectable hepatocellular carcinoma. Two phase II trials have been performed in patients with differentiated thyroid cancer that have both shown promise. In the larger, National Cancer Institute sponsored study with 58 patients with differentiated and anaplastic thyroid cancer (46 differentiated thyroid cancer patients evaluable for response). The partial response rate was 13%, and stable disease rate was 54% in patients with differentiated thyroid cancer; however, 6 patients who were deemed not assessable were excluded from this analysis \[[@B86]\]. A second phase II study reported partial remission in 7 of 22 (32%) patients with differentiated thyroid cancer \[[@B87]\]. With early data showing sorafenib to have promise in differentiated thyroid carcinoma (DTC) and sorafenib being a commercially available drug, many clinicians began using this drug in an off-label manner (not FDA approved for thyroid cancer). We recently reviewed our experience with off-label use of sorafenib in patients with DTC \[[@B88]\]. All patients had progressive nonradioactive iodine-avid disease to receive drug. Twenty percent of the patients developed a partial response, and 60% of patients developed stable disease achieving a clinical benefit rate of 80%. Progression-free survival was lengthened from four months predrug to 19 months. There was no difference in response based on BRAF mutational status in any of these studies thus far. A phase III international randomized controlled trial is underway currently evaluating sorafenib in progressive nonradioiodine-responsive metastatic differentiated thyroid cancer. This trial is randomized to placebo, and the primary endpoint is progression-free survival.(b) *Sunitinib*. Sunitinib is another oral small molecule tyrosine kinase inhibitor that is FDA approved for the treatment of metastatic renal cell carcinoma. This drug inhibits RET, RET/PTC subtypes 1 and 3, and VEGFR \[[@B89]\]. Two patients with differentiated thyroid carcinoma treated with daily sunitinib for four weeks and two weeks holiday have had prolonged partial responses and decreases in SUV on PET scan \[[@B90]\]. An ongoing open-label phase II study showed a 13% partial response rate, and in another 68% percent of differentiated thyroid carcinoma patients, disease stabilization was seen \[[@B91]\]. A second phase II study has reported partial remission or disease stabilization in two of 12 patients thus far \[[@B92]\]. Other case reports have described prolonged partial responses \[[@B90]\]. The study by Carr et al. is the largest published trial to date using sunitinib in patients with metastatic PET-positive, RAI-refractory differentiated thyroid cancer. 28% of patients with differentiated thyroid cancer achieved a response (complete or partial response) \[[@B93]\].(c) *Pazopanib*. Pazopanib is a small molecule inhibitor of all VEGFR subtypes and PDGFR. It is approved in the United States for the treatment of advanced renal cell carcinoma. Mayo Clinic recently reported on 37 rapidly progressive DTC patients on a phase II single agent trial \[[@B94]\]. 49% of patients had partial responses with a starting dose of 800 mg. Progression-free survival was 12 months. This drug carries a black box warning due to severe and fatal hepatotoxicity observed in a renal cell carcinoma patient. Liver transaminase levels should be monitored closely with this drug.(d) *Adverse Events Common to Sorafenib, Sunitinib, and Pazopanib*. Adverse events seen with these drugs are similar to those described with the treatment of advanced renal cell carcinoma and are summarized in [Table 4](#tab4){ref-type="table"}. The most common ones include diarrhea, hand-foot syndrome, rash, hypertension, and fatigue. Skin changes have also been noted including keratoacanthomas and squamous cell carcinomas in about 5--11% of patients taking sorafenib. Due to the promise shown by these agents in differentiated thyroid carcinoma and the availability and tolerability of these drugs, sorafenib, sunitinib, and pazopanib have been added as treatment options for patients with progressive nonradioactive iodine-avid thyroid cancer who cannot be enrolled in a clinical trial \[[@B5]\]. Due to significant potential toxicity, only clinicians versed in the management of the side effects of these therapies should use these drugs in a permissible off-label manner.
\(4\) Clinical Trials with TKIs(a) *Motesanib*. Motesanib is an oral tyrosine kinase inhibitor that inhibits VEGFRs 1, 2, and 3 \[[@B95]\]. It is currently not commercially available. A phase II trial was initiated based on responses in five patients with differentiated thyroid cancer in a phase I trial \[[@B96]\]. Patients with progressive differentiated thyroid cancer based on serial radiographic imaging in a six-month period were enrolled in this international phase II trial. 93 patients were enrolled, 14% had a partial response, and 35% of patients developed disease stabilization for 24 weeks \[[@B97]\]. One-third of the patients were still on drug after 48 weeks. The median progression-free survival was 40 weeks. Tumors with BRAF mutations responded better than those without, despite the lack of BRAF inhibition by motesanib, suggesting that these BRAF-positive tumors may be more dependent on VEGF-mediated angiogenesis. The most common adverse events noted include fatigue, nausea, diarrhea, and hypertension. This trial also noted the unanticipated side effect of a 30% increase in levothyroxine requirement to maintain the purposeful TSH suppression; two-thirds of the patients developed a TSH outside of the range.(b) *Axitinib*. Axitinib is an oral tyrosine kinase inhibitor that blocks VEGFRs and has been studied in thyroid cancer. It is not commercially available. A phase II, multicenter study was initiated based on the experience of five thyroid cancer patients in a phase I trial \[[@B98]\]. 31% of differentiated thyroid carcinoma patients had a partial response \[[@B99]\]. Many of the patients had been previously treated with various chemotherapeutic agents. Median progression-free survival for all histologic types of thyroid cancer was 18 months. The most common side effects seen in this trial include hypertension, stomatitis, fatigue, and diarrhea. A current trial is underway to evaluate axitinib\'s efficacy in doxorubicin-refractory metastatic thyroid cancer patients.(c) *Other Therapeutic Agents*. A small molecule inhibitor of the epidermal growth factor receptor, gefitinib, has been looked at in advanced thyroid cancer but had no complete or partial responses \[[@B100]\]. With increased expression of c-MET described in PTC, cabozantinib (XL184), an oral small molecule inhibitor of various tyrosine kinases including C-MET and VEGFR2, is being studied in differentiated thyroid cancer \[[@B104]--[@B103]\]. Results presented at the American Thyroid Association (ATA) were very encouraging, with a partial response rate of 53% in a cohort of 15 DTC patients \[[@B104]\]. A phase 2 trial with lenvatinib (E7080) in 58 patients with radioactive iodine refractory DTC showed 50% partial response and a progression-free survival of 13 months \[[@B105]\]. XL281, a small molecule inhibitor of BRAF kinases currently in a phase I trial, preliminarily shows five patients with papillary thyroid carcinomas with prolonged stable diseases. Two of these patients had V600E BRAF mutation \[[@B106]\]. Vemurafenib (also known as PLX4032, RO5185426, and RG7204), also a small molecule inhibitor of mutant BRAF kinase, has shown promise in thyroid cancer \[[@B107]\] and is currently being studied in patients with BRAF-mutated thyroid cancer in a phase II trial. An open-label phase II study in Kentucky examined the efficacy of thalidomide in progressive metastatic thyroid cancer of all histologies \[[@B108]\]. An 18% partial response rate with 32% stable disease as best response is described in the 28 evaluable patients of all thyroid histologies. Lenalidomide, a similar drug to thalidomide but less toxic, is being evaluated in a phase II open-label study in DTC patients currently \[[@B109]\]. Thus far, 39% of the patients developed a partial response and 50% in whom disease stabilized. Overall survival was shorter than in most trials with a median of 11 months.(d) *Agents to Restore Radioactive Iodine Uptake*. Researchers have been searching for ways to restore loss of radioactive iodine to nonavid tumors. 13 cis-retinoic acid partially restored radioactive iodine uptake in poorly differentiated follicular thyroid cancer cells \[[@B110]\]. Many clinical trials using retinoid receptors and other drugs have been focused on this restoration of iodine avidity with little success. Bexarotene, a synthetic agonist of the retinoid X receptor, was evaluated in a phase II trial to attempt to restore radioiodine activity, followed by treatment. After six weeks of therapy, eight of 11 patients had partial restoration of iodine avidity, but there was not much tumor reduction \[[@B111]\]. Rosiglitazone, a peroxisomal proliferators-activated receptor-gamma agonist, was evaluated for restoration of radioiodine uptake \[[@B112]\]. After eight weeks of treatment, although four patients had iodine avidity, clinical response was lacking. Depsipeptide, a histone deacetylase inhibitor, was evaluated in a phase II trial of patients with nonradioiodine metastatic DTC \[[@B113]\]. Only one patient out of 14 showed improvement in radioiodine uptake, but significant cardiac toxicities were seen, including sudden death. The most encouraging of these studies was presented at the American Thyroid Association (ATA) in 2011, evaluating single-agent MEK1/2 inhibitor, selumetinib (AZD6244), in 17 patients who were RAI refractory. In 11 patients, RAI uptake was restored. Information on best response was available in 7 patients, 6 of whom had a partial response to RAI \[[@B114]\].Other agents are under investigation in both phases I and II studies, including agents that inhibit the PI3kinase/aKt pathways, histone deacetylase inhibitors, and combinations of methylation inhibitors with histone deacetylase inhibitors.
4. Summary {#sec4}
==========
Differentiated thyroid carcinoma that is nonradioiodine avid is difficult to detect and treat. Clinically dictated selected imaging including ultrasound of the neck, CT imaging of the chest, MRIs of the spine and brain, bone scan, and ^18^FDG-PET-CT have all been useful in the detection of disease. Further radioactive iodine therapy in these patients tends to increase adverse events with minimal clinical benefit.
Recurrent neck disease is often treated with further surgery. In addition, patients may benefit from resection of specific symptomatic metastatic sites. Selected patients may benefit from external beam radiotherapy, radiofrequency ablation, or chemoembolization of other metastatic sites as well.
Patients with stable metastatic disease may be observed with thyroid hormone suppression therapy only. More advanced, progressive neck disease and progressive, distant metastatic disease require systemic treatment. Cytotoxic chemotherapy has limited response rates and significant toxicity and is therefore reserved for symptomatic progressive disease in a patient that cannot get on a clinical trial or tolerate antiangiogenic therapy. The advancement in understanding the molecular aberrations in thyroid cancer has led to an explosion of promising recent clinical trials. Targeted agents against the VEGF receptor and the MAP kinase pathway are amongst the most promising thus far. These agents have shown some of the most impressive responses and have fairly tolerable adverse effects.
Given our current knowledge and trial results, it remains difficult to choose the optimal therapy for selected patients. Many of the trials had varying entry criteria, many of which did not require progression. No trial thus far has overall survival as the primary endpoint, prolongation of this being the ultimate goal of patients. In addition, most patients eventually progress through these agents suggesting development of other pathways of resistance. Future trials will likely necessitate combination of therapy with minimal increased toxicity. Future trials may include agents that inhibit the PI3 kinase pathway in addition to the MAP kinase pathway or combination of cytotoxic chemotherapy with targeted agents. The main goal of all these trials should be to prolong life with minimal decrease in quality of life.
N. L. Busaidy has grant funding from Bayer and Novartis; M. E. Cabanillas has grant funding from Eisai and Exelixis.
######
Imaging modalities for RAI-refractory recurrent disease.
Imaging study Utility Pros Cons
----------------- ---------------------------------------------------------------------- -------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Ultrasound neck Detection of neck disease Sensitive; ability to biopsy Operator dependent; difficult to detect invasive disease and disease in the posterior neck
CT Detection of local and metastatic disease Sensitive; less operator dependent Radiation exposure; risk of renal injury with contrast; delays in radioiodine administration
MRI Detection of local and metastatic disease Sensitive for CNS disease; no radiation exposure Difficult to tolerate in some patients; risk of nephrogenic systemic fibrosis (NSF) in patients with renal failure; contraindicated in patients with certain metal devices or implants
FDG-PET scan Detection of metastatic disease and providing prognostic information Sensitive when used with CT Detects FDG-avid disease only
######
Therapeutic modalities for RAI-refractory recurrent disease.
Indication Pros Cons
---------------------------------------- -------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------- ------------------------------------------------------------------------------------
Surgery Surgically resectable local recurrences; metastasectomy Potential for cure Potential significant morbidity
External beam radiation Adjuvant: neck Therapeutic and palliative: metastatic sites Decrease in recurrence, progression, and pain May preclude future neck surgery; dysphagia and xerostomia; secondary malignancies
PEIT Locally recurrent disease in patients at high risk for morbidity and mortality from surgical resection Potential for avoidance of surgery Local pain; injury to local structures; unknown effect on survival and recurrence
Systemic chemotherapy (including TKIs) Unresectable, RAI-refractory, metastatic disease May slow progression of disease; may alleviate disease symptoms Significant adverse events; unknown effect on survival
PEIT: percutaneous ethanol injection therapy; TKI: tyrosine kinase inhibitors.
######
Targeted therapies evaluated in clinical trials for thyroid cancer.
Drug VEGFR1 VEGFR2 VEGFR3 RET BRAF Other Response; PFS Citation
-------------- -------- -------- -------- ----- ------ ------- ------------------------------- -----------------------------------------------------------------------------------------
Axitinib X X X 31% PR; 18 mos (MTC included) Cohen et al. \[[@B91], [@B99]\]
Motesanib X X X X 14% PR; 9 mos Sherman et al. \[[@B97]\]
Sorafenib X X X X 13--32% PR; PFS 10--21 mos Kloos et al. \[[@B86]\], Gupta-Abramson et al. \[[@B87]\], Cabanillas et al. \[[@B88]\]
Sunitinib X X X X 28% CR + PR; TTP 13 mos Carr et al. \[[@B93]\]
Pazopanib X X X X 49% PR; PFS 12 mos Bible et al. \[[@B94]\]
Lenvatinib X X X X FGFR 50% PR; PFS 13 mos Sherman et al. \[[@B105]\]
Cabozantinib X X X c-MET 53% PR; PFS n/a Cabanillas et al. \[[@B104]\]
PR: partial response, SD: stable disease, TTP: time to progression, PFS: progression-free surivival, n/a: not available, and mos: months.
######
Adverse events associated with the commercially available TKIs used in thyroid cancer.
Adverse event Sorafenib (%) Sunitinib (%) Pazopanib (%)
--------------------------------- --------------- --------------- --------------- ----- ------ -----
Hypertension 17 4 30 12 40 4
CHF or LVEF decline 1.7 NR 13 3 \<1% NR
Proteinuria NR NR NR NR 9 \<1
Hand-foot skin reaction 30 6 29 6 6 NR
Stomatitis NR NR 30 1 4 NR
Anorexia 16 \<1 34 2 22 2
Weight loss 10 \<1 12 \<1 52 3.5
Diarrhea 43 2 61 9 52 3.5
AST elevation NR NR 56 2 53 7.5
ALT elevation NR NR 51 2.5 53 12
Fatigue 37 5 54 11 19 2
Hypothyroidism NR NR 14 2 7 NR
Arterial thromboembolism 2.9 NR NR NR 3 2
Hemorrhage/bleeding (all sites) 15 3 30 3 13 2
CHF: congestive heart failure; LVEF: left ventricular ejection fraction; AST: aspartate aminotransferase; ALT: alanine aminotransferase; NR: not reported. table adapted from \[[@B115]\].
[^1]: Academic Editor: Mingzhao M. Xing
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
In some clinical situations, such as root canal overinstrumentation, apical resorption and teeth with open apices, there may be extrusion of the filling material, either gutta-percha, root canal sealer or both, due to the difficulty or impossibility to lock the master gutta-percha cone. In these situations, fabrication of an apical plug with calcium hydroxide and, more recently, with mineral trioxide aggregate (MTA) has been suggested[@B02],[@B04],[@B13],[@B21],[@B25],[@B26].
MTA was introduced in the early 1990\'s as an experimental material developed by Dr. Mahmoud Torabinejad at Loma Linda University, USA. This material was originally indicated as a retrograde filling material for use in endodontic surgery and cases of intraradicular and furcal perforations[@B18]. Since then, it has been used in different clinical situations, such as communicating internal and external resorptions, as capping material in mechanically exposed pulps, as intracoronal barrier during internal bleaching of endodontically treated teeth, and as apical plug in case of difficulty to lock the master gutta-percha cone[@B04]. These indications of MTA are related to the possibility of use in moist environments, as in most aforementioned indications, and mainly to its biocompatibility[@B04],[@B26].
The sealing ability of MTA apical plug and its thickness have been investigated by several authors[@B02],[@B13],[@B16],[@B17],[@B20],[@B21],[@B26]. Four-millimeter-thick plugs have been shown to be the most efficient with respect to root canal sealing ability and resistance to displacement[@B13],[@B20],[@B21],[@B26]. All of these studies have sought an alternative to MTA apical plug as well as its most appropriate thickness.
Wucherpfennig and Green[@B27] (1999) have called the attention to the fact that MTA and Portland cement were similar materials. Other studies were conducted and confirmed this similarity by means of microbiological, chemical, physical and biological behavior tests[@B01],[@B05],[@B08],[@B10],[@B11],[@B15],[@B23].
The purpose of this study was to evaluate the sealing ability of apical plugs made of white and gray MTA-Angelus® (Angelus Soluções em Odontologia, Londrina, PR, Brazil) and white Portland cement (Votorantim Cimentos, Votorantim, São Paulo, SP, Brazil) placed via the root canal and having different thicknesses (2, 5 and 7 mm).
MATERIAL AND METHODS
====================
This study was approved by the Institutional Review Board of the Dental School of Bauru, University of São Paulo, Brazil. Ninety extracted human single-rooted teeth with intact roots and completely formed apices were used. The teeth were obtained from the files of the Department of Endodontics of the Dental School of Bauru and were kept in 10 % aqueous formalin solution.
The dental crowns were sectioned at the cementoenamel junction with a low-speed diamond saw (KG Sorensen, São Paulo, SP, Brazil) under continuous water spray to obtain access to the root canal. The root canal length was determined by inserting a size 15 K-file (Dentsply-Maillefer Instruments SA, Ballaigues, Switzerland) into the canal until its tip reached the apical foramen. The working length was established by subtracting 1 mm from this measurement. The stepback technique was employed and the root canal was flared using a size 60 K-file to the working length. Instrumentation was aided by irrigation with 1 mL of 1% sodium hypochlorite solution (Biodinâmica Química e Farmacêutica Ltda, Ibiporã, PR, Brazil) alternated with the sequence of instruments, followed by a final flushing with 1 mL of sterile water.
The teeth were assigned to 3 groups (n=30), according to the material used for fabrication of the apical plugs: Group A = placement of gray MTA-Angelus® plugs; Group B = placement of white MTA-Angelus® plugs; Group C = placement of white Portland cement plugs. The groups were further subdivided into groups of 10 teeth each, according to the thickness of the apical plugs, namely 2, 5 and 7 mm ([Table 1](#t01){ref-type="table"}).
###### Distribution of teeth in the groups with respect to the plug thickness and material
Groups Material Plug Thickness (mm)
-------- ----------------------- --------------------- ---- ----
A Gray MTA 10 10 10
B White MTA 10 10 10
C White Portland cement 10 10 10
Before standardization of the foraminal opening, the roots were made impermeable by application of a layer of epoxy adhesive (Araldite-Ciba-Geigy S.A., Taboão da Serra, SP, Brazil), followed by two coats of nail polish (Cosbra Cosmeticos Ltda., São Paulo, SP, Brazil). The foramen diameter was standardized by inserting the 40 K-file 1 mm beyond the apical foramen, so that only the apical opening would not be impermeable.
For fabrication of the apical plugs, the tested materials were applied with a size 4 Lentulo spiral (Dentsply-Maillefer Instruments SA, Ballaigues, Switzerland) at the apical end of the root, trying to fill it completely. The material was condensed with the tip of a size 40 K-file involved in cotton for achievement of the plug[@B19]. Next, using a size 40 K-file with a rubber stop positioned 2, 5 and 7 mm shorter than the root canal length, the excess material was removed for fabrication of 2-, 5- and 7-mm-thick apical plugs, respectively. Finally, the root canal walls were cleaned with the tip of an instrument wrapped in moist cotton.
The materials were allowed complete setting (nearly 3 hours). Then, the canal entrances were sealed with epoxy adhesive and nail polish, and the roots were immersed in 0.2 % Rhodamine B solution (Labsynth Produtos para Laboratórios Ltda, Diadema, SP, Brazil) at pH 7.0 for 72 hours at 37°C. After this period, the roots were removed from the dye, washed in running water for 24 hours, had the impermeable coating scraped away and were washed for additional 12 hours.
The roots were then sectioned longitudinally in a buccolingual direction for exposure of the apical plugs, photographed with a digital camera (Canon EOS Rebel 300 D) and analyzed by Image Tool software (University of Texas Health Science Center, San Antonio, TX, USA). Data were analyzed statistically by Kruskal-Wallis and Dunn\'s tests. Significance level was set at 5%.
RESULTS
=======
[Table 2](#t02){ref-type="table"} shows the percent marginal leakage for the different materials for each plug thickness. [Tables 3](#t03){ref-type="table"}, [4](#t04){ref-type="table"} and [5](#t05){ref-type="table"} show the results of the Kruskal-Wallis statistical test for comparison among groups A (MTA-Angelus® gray), B (MTA-Angelus® white) and C (white Portland cement) with respect to the plug thicknesses (2, 5 and 7 mm).
###### Percent dye leakage for the tested materials and plug thicknesses
Material Plug Thickness (mm)
----------------------- --------------------- ------- -------
Gray MTA 82.20 68.76 44.07
White MTA 96.85 69.14 71.51
White Portland cement 91.85 66.80 42.44
###### Comparison among groups A (gray MTA), B (white MTA) and C (white Portland cement) for 2-mm-thick apical plugs
-----------------------------------------------------------------------------------
Stat\ Kruskal-Wallis ANOVA by Ranks\
Nonpar\ (dados_infil.sta)\
Stats Independent (grouping) variable: Group\
Kruskal-Wallis test: H (2; N= 30) = 6.609514\
Dunn test: p= 0.0367
-------------------- ----------------------------------------------- --------------
Depend.: INFILT\_% Valid (N) Sum of Ranks
A 10 108.0000
B 10 208.0000
C 10 149.0000
-----------------------------------------------------------------------------------
###### Comparison among groups A (gray MTA), B (white MTA) and C (white Portland cement) for 5-mm-thick apical plugs
------------------------------------------------------------------------------------
Stat\ Kruskal-Wallis ANOVA by Ranks\
Nonpar\ (dados_infil.sta)\
Stats Independent (grouping) variable: Group\
Kruskal-Wallis test: H (2; N= 30) = 0.0393611\
Dunn test: p= 0.9805
-------------------- ------------------------------------------------ --------------
Depend.: INFILT\_% Valid (N) Sum of Ranks
A 10 150.0000
B 10 157.5000
C 10 150.5000
------------------------------------------------------------------------------------
###### Comparison among groups A (gray MTA), B (white MTA) and C (White Portland cement) for 7-mm-thick apical plugs
-----------------------------------------------------------------------------------
Stat\ Kruskal-Wallis ANOVA by Ranks\
Nonpar\ (dados_infil.sta)\
Stats Independent (grouping) variable: Group\
Kruskal-Wallis test: H (2; N= 30) = 15.08129\
Dunn test: p=0.0005
-------------------- ----------------------------------------------- --------------
Depend.: INFILT\_% Valid (N) Sum of Ranks
A 10 117.0000
B 10 243.0000
C 10 105.0000
-----------------------------------------------------------------------------------
A statistically significant difference (p\<0.05) was observed between groups A and B as to the leakage in 2- mm-thick plugs, with better results for group A ([Table 3](#t03){ref-type="table"}). Materials did not show statistically significant difference (p\>0.05) when groups with 5-mm-thick plugs were compared to each other ([Table 4](#t04){ref-type="table"}). Regarding the 7-mm-thick plugs, there was statistically significant difference (p\<0.05) between groups A and B and between groups B and C, with worst results for group B in both comparisons ([Table 5](#t05){ref-type="table"}).
In [Figure 1](#f01){ref-type="fig"} shows the graphic presentation of percent marginal leakage (0.2% Rhodamine B) in the root canals, regarding the tested materials and plug thicknesses. It is possible to verify that 2-mm-thick plugs yielded the least satisfactory results in all groups, whereas the 7-mm-thick plugs yielded the best results for groups A and C, yet not for group B, in which the 5-mm-plugs had the best results as to dye leakage.
{#f01}
DISCUSSION
==========
This study investigated the sealing ability of gray MTA, white MTA and white Portland cement used for fabrication of apical plugs, given that this procedure is often required in the clinical practice[@B12],[@B14],[@B19], mainly in cases where appropriate adaptation of the master gutta-percha cone is difficult. Torabinejad and Chivian[@B25] (1999) pointed out that the goal of an apical plug is to induce hard tissue formation, in order to prevent filling material extrusion in teeth with open apices.
The use of methylene blue in marginal sealing studies has been questioned, due to its incompatibility with alkaline substances, which may induce discoloration of the dye. Calcium oxide is one of the components found in MTA. When calcium oxide is mixed with water, it results in the formation of calcium hydroxide, with a subsequent increase in pH, as previously demonstrated by Duarte, et al.[@B09] (2003). Thus, discoloration of the surfaces stained by methylene blue may occur. Therefore, Rhodamine B dye solution is more appropriate for evaluating the sealing ability of MTA[@B22],[@B24].
In the present study, 5- and 7-mm-thick plugs were more efficient for apical sealing than 2-mm-thick plugs, regardless of the material utilized ([Table 1](#t01){ref-type="table"} and [Figure 1](#f01){ref-type="fig"}), which is in agreement with the findings of previous[@B02],[@B13],[@B21],[@B26]. In group B (white MTA), the 5-mm-thick plugs had better performance than the 2- and 7-mm-thick plugs, yet without statistically significant difference. With regard to tested materials, no statistical difference was expected among them because the chemical components of MTA and Portland cement are the same, except for bismuth oxide, which provides radiopacity to MTA[@B08],[@B10],[@B11].
Camilleri, et al.[@B07] (2005) evaluated the chemical constitution and biocompatibility of white and gray Portland cement, white and gray MTA, Portland cement clinker without calcium sulfate and Portland cement clinker without calcium sulfate with addition of bismuth oxide. They concluded that the chemical composition of the tested materials is similar, primarily containing tricalcium silicate and dicalcium silicate. The white cements differ from the gray cements by the small quantity of iron oxide (FeO), while MTA differs from Portland cement due to the presence of bismuth oxide. There was no difference between white and gray MTA, and the addition of bismuth oxide did not interfere with the biocompatibility of cements. In the present study, white MTA showed poorer results than gray MTA and white Portland cement for all tested plug thicknesses ([Table 1](#t01){ref-type="table"}).
Asgary, et al.[@B03] (2005) observed significant differences between gray and white MTA, especially in the contents of aluminum trioxide (Al~2~O~3~), magnesium oxide (MgO) and iron oxide (FeO). However, these differences are not enough to explain the results of the present study, in which gray MTA showed better results than white MTA as to dye leakage. Matt, et al.[@B21] (2004) observed similar results in their study, where apical plugs made with gray MTA were more efficient than those made with white MTA.
As to the fabrication of apical plugs, Torabinejad and Chivian[@B25] (1999) recommended carrying the MTA with a large amalgam carrier to the root canal and then condensing the material to the apical end of the root with pluggers or paper points. Sometimes, this procedure is difficult due to the root canal diameter and anatomy. In the present study, MTA and Portland cement were carried to the root canal with a size 4 Lentulo spiral, according to the technique proposed by Bramante, et al.[@B06] (2004). The Lentulo spiral is used to carry the MTA in paste consistence more easily to the root canal end in a fast and correct manner. Another important factor is material condensation at the end of the root canal because the apical plug should resist the filling material. This condensation is more effective when performed with a K-file compatible with the root canal diameter and with the tip wrapped in a cotton mesh. Cleaning of the root canal walls and removal of excess material must be performed with the same K-file wrapped in moist cotton in sterile saline, not to interfere with the proper root canal filling.
CONCLUSION
==========
The findings of the present study showed that gray MTA and Portland cement had better sealing ability than white MTA when used as apical plugs. Dye leakage was smaller for 5- and 7-mm-thick plugs compared to 2-mm-thick plugs.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Soybean is ranked among the most important agricultural food crops, being an essential part of food security (Akibode and Maredia [@CR2]). Production of pulses, groundnuts and soybean have followed an increasing trend with the global production tripling from 148 million tons in 1980--1982 to 480 million tons in 2012--2014. The increase was led by soybean production which increased from 87 million tons to 214 million tons due to an increase in the demand for protein meals and oils. Such trends push soybean agriculture in to marginal semi-arid areas, where water-limiting conditions often constrain crop productivity (Darianto et al. [@CR14]).
Soybean plants are very sensitive to drought stress, especially during reproduction. According to Iowa State data (Lenssen [@CR29]), 4 days of visible moisture stress in the 3rd week of pod development results in about 36 % loss, increasing to 39--45 % in the 2nd--4th week of seed filling. Drought stress causes abortion of small pods, reduced seeds per pod, and reduced seed size. Currently, the economically viable approaches to support crop production under drought are still limited. Further research of physiological effects of natural extracts as a possible tool to improve plant resistance to abiotic stresses could be a viable strategy.
It is well-known that seaweed extracts improve the stress tolerance of agricultural crops. The unique properties and diverse functionality of seaweed extracts for agricultural applications have been extensively reviewed (Craigie [@CR13]; Arioli et al. [@CR3]). Numerous studies have shown the positive effect of Acadian^®^ extract of *Ascophyllum nodosum* (Acadian Seaplants, Nova Scotia, Canada) on plant resistance to drought stress (Neily et al. [@CR34]; Spann and Little [@CR39]). However, the modes of action of seaweed extract to improve stress tolerance are not fully understood.
The response of higher plants to drought stress is a complex and dynamic process (Chaves et al. [@CR9]). The plant leaf is a primary receptor of stress, triggering a chain of physiological responses from gene expression and hormone regulation (Peleg and Blumwald [@CR35]) to osmoregulation (Ahmad and Wani [@CR1]) and structural adaptation (Bacelar et al. [@CR4]). The plant leaf has an inherent mechanism of stomatal regulation, which controls gaseous exchange, photosynthesis and metabolic activities in response to environmental changes by maintaining a crucial balance between photosynthetic gains and water losses (Chaves et al. [@CR9]; Jones et al. [@CR26]; Costa et al. [@CR12]).
Abscisic acid (ABA)-mediated stomata closure is one of the first plant responses to drought stress (Hetherington and Woodward [@CR18]). Although guard cells can lose turgor as a result of a direct loss of water, stomatal closure in response to dehydration is always an active, energy-dependent process (Hetherington and Woodward [@CR18]). Stomatal closure results in reduction of stomatal conductance and CO~2~ availability, which directly affects rates of photosynthesis (Chaves et al. [@CR9]). It is accompanied by an increase in leaf temperature (Jones [@CR22]). If this temperature reaches a threshold, it often leads to irreversible leaf tissue damage. Hence, leaf temperature can be used as an indicator of plant stress (Jones [@CR22]; Jones et al. [@CR26]; Costa et al. [@CR12]). Crop water stress index (CWSI) is calculated, using equation (Idso et al. [@CR19]):$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$CWSI = \left( {T_{canopy} {-}T_{nws} } \right)/\left( {T_{{max} } {-}T_{nws} } \right)$$\end{document}$$where *T*~*max*~ is a temperature of the dry leaf surface and *T*~*nws*~ is the temperature under non-limiting soil water condition, when crop transpiration is at its maximum rate. Fuentes et al. ([@CR16]) developed automated methodology for measurements of water stress index of grapevine canopies, using thermal imaging. However, field application of this equation encountered several problems, including difficulty in separation of relevant crop canopy temperature from background and normalization of CWSI under changing climatic conditions. For practical purposes, Jones proposed simplified "index of relative stomatal conductance" *I*~*g*~, which assumes constant environmental conditions (Jones [@CR24]):$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$I_{g} = (T_{dry} - T_{leaf} )/(T_{leaf} - T_{wet} ) = g_{s} /G$$\end{document}$$where *g*~*s*~ is stomatal conductance (m/s), *G* is the constant, calculated as a slope of *g*~*s*~(*I*~*g*~) regression. The advantage of using thermal index is linear relationship between leaf temperature and stomatal conductance (Jones [@CR22], [@CR23]). It is important to note that (2) can be used to evaluate stomatal conductance in controlled environment, where leaf temperature is mostly dependent on water availability. However, field applications of this equation are limited because of the side effects of environmental factors, such as air temperature, light intensity, relative humidity, wind speed and soil conditions (Jones and Schofield [@CR25]; Leinonen et al. [@CR28]).
Thermal imaging was successfully used to measure temperature of single leaves (Chaerle and Van Der Straeten [@CR8]; Ribeiro da Luz and Crowley [@CR36]; Kapanigowda et al. [@CR27]) and whole plant (Grant et al. [@CR17]; Blonquist et al. [@CR7]; Zia et al. [@CR42]; Fuentes et al. [@CR16]). Theory and applications of thermal imaging to the study of plant water relations was reported by Jones ([@CR24]). The theoretical background of thermal imaging, in particular the relationship between leaf temperature and stomatal conductance, was thoroughly reviewed by Cohen et al. ([@CR11]) and Maes and Steppe ([@CR32]). An excellent review of digital image processing for detecting, quantifying and classifying plant diseases was presented by Barbedo ([@CR5]). Although imaging-guided expert systems are widely used in medicine (Egger [@CR15]) and agriculture (Ishimwe et al. [@CR20]), there is a gap in applications of real-time thermal imaging in plant physiology. As to our knowledge, thermal imaging has not been applied to study the stress-response of plants treated with biostimulants, such as seaweed extracts.
The initial hypothesis was that thermal imaging is able to detect early-stage physiological response of soybean plants to water stress "in-situ". To test this hypothesis, the temperature and leaf angle of individual plants were measured over a 5-day stress-recovery experiment. In order to understand the mode of action of Acadian^®^ seaweed extract, responses of treated and untreated control plants in the same stress conditions were compared. It was anticipated that imaging of early-stage plant stress response would significantly advance our knowledge about the mode of action of seaweed extracts. This research is an important step towards understanding the full benefits of seaweed extracts for improving crop yield under water stress.
Methods {#Sec2}
=======
Plant preparation {#Sec3}
-----------------
Soybean seeds (*Glycine max* (L.) Merr) variety Savana were planted in ProMix BX (Premier Tech Horticulture, Canada) in 200-seed trays, in a controlled environment room (27 °C 16:8 day/night). After 7 days, the plants were transplanted into 4.5 inch pots and placed in a Conviron environmental chamber (Winnipeg, Canada) with fluorescent lamps Sylvania Hg Pentron 4100 K, 39 W (Osram Sylvania, USA), After an additional 14 days, drought stress was initiated. The environmental chamber was set at 16:8 day/night with the temperature 27 °C, \~400 µmol/m^2^/s PAR intensity, 600 ppm CO~2~ and 60 % relative humidity. Due to additional heat being generated by the lights, the temperature reached 33 °C during the day. Twice a week the plants were treated with 100 mL solution of 0.5 g/L 20-8-20 (Plant Products, Canada) or 0.5 g/L 20-8-20 plus 7.0 mL/L Acadian^®^. On Day 21, plants were treated with 1.0 g/L 20-8-20 or 1.0 g/L 20-8-20 plus 7.0 mL/L Acadian^®^ until the soil was completely saturated and the excess liquid ran through the bottom of the pots (\~300 mL). This equalised the soil moisture content at \~70 % for the onset of drought stress and no further water was applied. Each experiment consisted of two plants (one treated, one control). Three separate experiments were conducted using a total of six plants (three treated, three control). A 5-day stress-recovery experiment with re-watering on the fifth day allowed observation of all stages of drought stress, i.e., stress response (Day 3), adaptation (Day 4) and recovery (Day 5). Data for leaf temperatures and angles were recorded from one pot as a biological replicate (two pots per experiment). Statistics were conducted on results from three experiments.
Thermal imaging {#Sec4}
---------------
A T440 IR camera (FLIR Systems Inc., North Billerica, MA, USA) with the focal plane array (FPA) uncooled microbolometer, 320 × 240 pixels and spectral range of 7.5--13 µm was used to take both thermal and RGB visible images automatically every 10 min. Camera settings for emissivity was constant 0.95 for the entire experiment. The imaging camera was installed in a Conviron environmental chamber at a distance of 1.1 m from the plants. In order to minimize the effects of wall reflectance and air temperature gradients due to vertical convective airflow, plants were shielded with 1.0×1.0 m Styrofoam sheets. This measure provided 0.1 °C temperature resolution in the field of view. One treated and one control plant were monitored in each experiment. Leaf temperature data and background temperature were measured in the following two modes:*Online (camera) measurements*: Two rectangular regions of interest (ROI) were set directly on the camera screen. Thermal images displayed maximum and minimum temperatures within the ROI for each image in continuous mode of operation. Before stress the minimum temperature corresponded to leaf temperature, while the maximum temperature reflected ambient air temperature. After stress the minimum temperature represented air temperature, while the maximum temperature represented leaf (or stem) temperature.*Offline measurements*: The data were recorded in form of radiometric images and analysed, using FLIR Research IR 4.1 software (FLIR Systems Inc., North Billerica, MA, USA).
The software offered computation of maximum and minimum temperatures from any pre-defined ROI by using rectangle (Fig. [1](#Fig1){ref-type="fig"}), oval or user-defined shape selection. Figure [1](#Fig1){ref-type="fig"} shows statistical data, which represent leaf temperature distribution for each plant. The air temperature was recorded from a ROI measurement cursor (3 × 3 pixels) at the point between the two plants. This technique allowed closer examination of the leaf temperatures and adjustment of the ROI if one plant grew into the ROI for the other plant. Also, as the apical point wilted, it would occasionally bend down into the ROI and interfere with the maximum temperature results. In this case the leaf temperature was corrected, using Research IR software. Online automated measurements were used as a primary source of information, whereas offline manual measurements were used to double-check measured values.Fig. 1Screenshot of FLIR Research IR 4.1 software (FLIR Systems Inc., USA) for thermal image analysis
Leaf angle and turgor calculations {#Sec5}
----------------------------------
Data for the leaf angle were obtained from the radiometric image by segmentation and using a digital protractor (IMAQ Vision Assistant, USA) with the vertical axis aligned to the stem. Leaf turgor was calculated from the leaf angle assuming a linear relationship. An angle of 90°, observed for horizontally oriented leaves, was associated with full turgor (100 %). An angle of 0°, observed for wilted plants, was interpreted as zero turgor (0 %). The thermal images collected allowed for measurements of leaf angle not only during the daytime, but also during some of the night periods.
Statistical analysis {#Sec6}
--------------------
Two statistical methods, namely "repeated measures analysis" and "non-linear regression modeling", were applied to the data sets (Montgomery [@CR33]). Repeated measures analysis was completed to determine the effect of the Acadian^®^ treatment on leaf temperature and turgor, and how this changed over time. The experimental design adopted was the Randomized Block Design, with three blocks (i.e., three different experiments), two treatments (i.e., Acadian^®^-treated vs. control), and the temperature and turgor responses averaged for every hour during 10 h of daytime period. Hourly estimates of responses were taken as a mean value of data from six images. Akaike Information Criterion (Littell et al. [@CR31]) was used to determine the most appropriate co-variance structure (compound symmetry) for both responses. The analysis was completed using the Mixed Procedure of SAS (SAS Institute Inc. [@CR37]), and further multiple means comparison was used for significant (p-value \<0.05) effects by comparing the least squares means of the corresponding treatment combinations. Letter groupings were generated using a 5 % level of significance for the main effects (treatment and time) and using a 1 % level of significance for temperature and turgor changes over time (treatment by hour interaction effect). For each response, the validity of model assumptions were verified by examining the residuals as described in Montgomery ([@CR33]).
Stress-responses of control and Acadian^®^-treated plants were analyzed, using the initiation of wilting at Day 3 as the initial point (hour 0). Leaf temperature kinetics was analyzed with asymptotic model (Eq. [3](#Equ3){ref-type=""}).$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$Y = \theta_{1} - \theta_{2} \exp ( - \theta_{3} X) + \varepsilon$$\end{document}$$where *Y* is the dependent variable (temperature), *X* is the independent variable (time), *θ*~1~, *θ*~2~, *θ*~3~ are model parameters, and the error term ε is assumed to have a normal distribution with constant variance.
The parameters of this non-linear regression model were estimated iteratively using the NLIN Procedure of SAS (SAS Institute Inc. [@CR37]) and the models were checked for adequacy (Bates and Watts [@CR6]). Validity of the normality, constant variance and independence assumptions on the error terms were verified by examining the residuals (Bates and Watts [@CR6]). Relationships between leaf temperature and turgor were explored for one experiment on Day 3 by using Lowess Smoother function of Minitab (Cleveland [@CR10]).
Results {#Sec7}
=======
Typical appearance of control and Acadian^®^-treated soybean plants over the 5-day stress-recovery experiment is shown in Fig. [2](#Fig2){ref-type="fig"}.Fig. 2Visible (**a**, **c**, **e**) and thermal (**b**, **d**, **f**) images of soybean plants (Acadian^®^-treated---*right*, control---*left*), taken by FLIR T-440 camera in environmental chamber Conviron, air temperature 30 °C, relative humidity 60 % on Day 1 (**a**, **b**), Day 4 (**c**, **d**) and Day 5 (**e**, **f**)
During the first 2 days of the experiment there was no statistically significant difference in leaf temperature and turgor between the treated and control plants (a, b). The difference between treated and control plants became significant on Day 3. After 5--6 h of light exposure control plants rapidly wilted, while treated plants still maintained turgor. After 24 h of stress, on Day 4, the turgor of Acadian^®^-treated plants was slightly higher, than the control (c, d). In the afternoon of Day 4, the moisture content in the pots was \~10--13 %. Re-watering on Day 5 showed significant visible difference between treatments. Within 2--3 h after re-watering, all Acadian^®^-treated plants recovered, while control plants were not able to fully recover (e, f). Consistent positive effect of seaweed treatment was verified in multiple experiments.
The changes of leaf temperature and leaf angle over 5-day stress-recovery experiment are presented in Fig. [3](#Fig3){ref-type="fig"}. Initial rapid decrease of leaf temperature on Day 1 (Fig. [3](#Fig3){ref-type="fig"}a) could be explained as a response to initial soil saturation. On Day 1 and at the beginning of Day 2 the temperature of the plant leaves was constant, at around 28--29 °C, which indicated that water was still available to the plants. The leaf temperature was 3.0--3.5 °C below air temperature, which corresponded to normal transpiration. Towards the end of Day 2 the temperature of the both plants increased, which indicated partial closure of the stomata due to water becoming the limiting resource. The difference between air and leaf temperatures decreased to 2.0--1.5 °C, which could be interpreted as a decrease in stomatal conductance. These changes in leaf temperature were accompanied by corresponding changes in leaf angle (Fig. [3](#Fig3){ref-type="fig"}b). At night, the leaves were in low position, increasing to 50°--60° during the day, which indicated an availability of water. During the daytime, the leaf temperature of the control plant was higher, which suggests better cooling ability of Acadian^®^-treated plants; however, visible images did not show any significant difference between the Acadian^®^-treated and control plants.Fig. 3An example of leaf temperature (**a**) and turgor (**b**) of Acadian^®^-treated plants versus control over a 5-day period. Experiment was carried out in Conviron chamber in Cornwallis with one control and one treated plant
From Fig. [3](#Fig3){ref-type="fig"}a, three distinct stages of plant response were recognized:
Stage 1: (Day 3): At the beginning of the day, there were no visual differences between plants. During the first 4--5 h of the daytime period, the temperature of both control and Acadian^®^-treated plants significantly decreased (i.e., the leaf-air temperature increased to 4.5--5.0 °C), probably due to increased stomatal conductance. It is interesting, that the leaf angle of Acadian^®^-treated plants demonstrated a short-term increase, as compared to a gradual drop of turgor pressure as observed in the control (Fig. [3](#Fig3){ref-type="fig"}b). Since leaf turgor is a good indicator of stomatal opening (Jones et al. [@CR26]), it could be postulated that the treated plants were able to control their stomatal apertures under water stress conditions.
On Day 3 after 5--6 h of light exposure the water stress caused wilting of control plants. The temperature of the control plants increased from 28.0 to 32.0 °C, approaching equilibrium with the air temperature (see Fig. [3](#Fig3){ref-type="fig"}a). This was accompanied by a rapid decrease in leaf angle (see Fig. [3](#Fig3){ref-type="fig"}b), which could be interpreted as a loss of control over stomatal opening during the water stress conditions. In contrast, the temperature of the Acadian^®^-treated plants increased slowly with a slow decrease of leaf angle, which could be explained by better stomatal control.
Stage 2 (Day 4): No differences in leaf temperature, but small differences between the turgor of the control and Acadian^®^-treated plants were observed. Turgor pressure of Acadian^®^-treated plants was slightly higher, than control (Fig. [3](#Fig3){ref-type="fig"}b). We could assume that Acadian^®^-treated plants were able to maintain functionality of stomata openings in stress conditions.
Stage 3: (Day 5): After a day with higher than normal leaf temperature, both plants were re-watered. The visual difference between treatments was apparent. Within 2--3 h after re-watering, all Acadian^®^-treated plants recovered turgor, while control plants were not able to fully recover (Fig. [3](#Fig3){ref-type="fig"}b). Further thermal imaging showed that the Acadian^®^-treated plants were able to gradually reduce their leaf surface temperature by 2--3 °C below ambient temperature, indicating a return to normal transpiration. In contrast, the leaf surface temperature of the control plants did not change, remaining in equilibrium with the ambient temperature. This can be explained by the fact that the control plants lost their ability to transport water and recover turgor in some leaves. The observed difference allowed us to conclude that Acadian^®^-treated plants were better able to withstand the types of drought stress as imposed under the experimental conditions.
To understand physiological aspects of the response, data from three different experiments were analyzed using the treatment (Acadian^®^-treated and control) as the primary factor of interest, experiment as a blocking factor (3 blocks), and time (10 h of daytime period) as the repeated measures factor. The statistical analyses of the data collected from the plant responses on Day 2, 3 and 4 are presented in Table [1](#Tab1){ref-type="table"} and Fig. [4](#Fig4){ref-type="fig"}.Table 1ANOVA p-values for the main and interaction effects of block, treatment and time on Day 2 temperature (°C), Day 3 temperature (°C), Day 3 turgor (%), and Day 4 temperature (°C)Source of variationDay 2 temperatureDay 3 temperatureDay 3 turgorDay 4 temperatureBlock0.7070.3490.4290.720Treatment0.8840.017*0.062*0.859Time*0.001*0.001*0.001*0.799Treatment × time0.992*0.001*0.7820.947Significant effects that required multiple means comparison are shown in italicsFig. 4Mean temperature on Day 2 (**a**), and mean temperature of the control and Acadian^®^-treated plants on Day 3 (**b**: stress). Within each day, means sharing the *same letter* are not significantly different
The results showed that the differences among the 3 Blocks (experiments) were not significant, which indicated a consistent stress-response in all experiments. In Day 1 the difference in leaf temperature and turgor between the two treatments was not significant (not shown in the Table). In Day 2 the difference between the leaf temperatures from the two treatments was not significant for the first 6 h, but become significant for the last 4 h of daytime period. Figure [4](#Fig4){ref-type="fig"}a shows significant increase of leaf surface temperature observed during the last 4 h of the daytime period. For Day 3 we observed a significant interaction effect of treatment and time (hour) on leaf temperature, which indicated that the plant responses to the applied stress between the Acadian^®^-treated and control plants changed over the period of observation (Fig. [4](#Fig4){ref-type="fig"}b). As a result, we separated the time series data in two clusters (−4 to 0 h, before wilting) and (0 to 6 h, after wilting), which allowed us to distinguish two phases of the stress response.
Analysis of the leaf turgor stress-response on the Day 3 showed significant difference between the Acadian^®^-treated and control plants. The turgor of treated plants was significantly higher (43.52 %) than control (23.31 %).
As shown in Table [1](#Tab1){ref-type="table"}, the interaction between Treatment and Time on the leaf turgor, calculated on Day 3 was not significant (p = 0.782), which implies that the difference between average turgor of treated and control plants was consistent during the 10 h. Since the interaction effect was not significant for each hour, the mean value shown in Fig. [5](#Fig5){ref-type="fig"} is average of the values obtained from treated and control treatments.Fig. 5Main effect plot of time on Day 3 leaf turgor (%) showing trends of the mean values for 11 h. Means sharing the *same letter* are not significantly different
Figure [5](#Fig5){ref-type="fig"} shows that leaf turgor increased marginally just before time point "0", which indicated the "critical point of stress". It is possible that this is an indication of a protective stress-response, such as short-term closing of the stomata in response to drought. However, further drought stress resulted in a gradual decrease of turgor.
The difference in early stress-response {#Sec8}
---------------------------------------
To determine the difference in stress-response of the Acadian^®^-treated plants versus control, the leaf temperature kinetics at the onset of physiological stress were analyzed with regression analysis. Fitted asymptotic models for the control and Acadian^®^-treated plants with 30 min increments are presented in Fig. [6](#Fig6){ref-type="fig"}. The leaf temperature of control plants followed asymptotic behavior, whereas Acadian^®^-treated plants demonstrated linear changes.Fig. 6Leaf surface temperature of the control (*left*) and Acadian^®^-treated (*right*) related to the time (8 h after wilting). Each experimental point is the average of three replications. **a** Temperature (control) = 31.83 − 4.0 × exp(−0.523 × time), **b** temperature (treated) = 82.81 − 55.1 × exp(−0.0098 × time)
A plausible explanation for these observations could be provided from thermodynamic principles. Water evapotranspiration from stomata provides heat balance between the plant leaf and the environment. The asymptotic model usually describes a transient, first-order process of equilibration with ambient conditions. From Fig. [6](#Fig6){ref-type="fig"}a, it follows that the leaf temperature of the control plants asymptotically reached equilibrium with the ambient temperature. The equilibrium temperature, predicted by the model, was 31.8 °C, which is close to the ambient air temperature (32 °C). The observed passive response could be interpreted as a failure of plant evapotranspiration.
In contrast, the almost linear response of the Acadian^®^-treated plants showed that their status was far from thermal equilibrium. This type of response could be interpreted as an active control of the plant thermal regime due to Acadian^®^ treatment.
Correlations between leaf turgor and temperature {#Sec9}
------------------------------------------------
To further explore if there was any relationship between turgor and leaf temperature, a correlation analysis was carried out. To understand if there were any differences in the stress response between the Acadian^®^-treated and control plants, leaf surface temperature-turgor data in the range of 28.5--31.0 °C for Day 3 were plotted (Fig. [7](#Fig7){ref-type="fig"}).Fig. 7Leaf turgor-leaf temperature relationship for Acadian^®^-treated (*solid*) and control (*broken*) plotted for Day 3
It was concluded that for both treatments, increasing leaf surface temperature was accompanied by corresponding decreased turgor pressure (i.e., a negative correlation). However, the difference in slope was significant (p \< 0.05). Acadian^®^-treated plants demonstrated higher sensitivity to changes of leaf temperature (i.e., better ability to regulate turgor with critical temperature increase). These results suggested that Acadian^®^-treated plants adjusted turgor with respect to leaf temperature changes, while control plants were trying to maintain constant turgor over the range of leaf temperatures.
Discussion {#Sec10}
==========
Results of our study demonstrated the positive effect of Acadian^®^ seaweed extract on stress resistance and recovery of soybean plants. The general tendency of increasing leaf temperature during the daytime period in Day 1 and 2 was observed for both Acadian^®^-treated and control plants. The observation could be attributed to the physiological phenomenon of midday depression of photosynthesis, caused by a decrease of stomatal conductance at high light intensity (Tenhunen et al. [@CR40]). This increase was even more evident on Day 2 (Fig. [3](#Fig3){ref-type="fig"}a), which indicated limited water availability and its effect on partial stomata closure. Based on previous research (Wally et al. [@CR41]), we could assume that stomata control in the first two days of drought stress was accompanied with the synthesis of endogenous ABA, especially in Acadian^®^-treated plants. This assumption, however, requires careful experimental verification.
On Day 3, the combination of water deficit with high irradiation significantly changed the physiological response of the leaves. In the first 7 h of the daytime period the leaf temperature decreased from 28.5 to 27.5 °C in the treated, and even to 26.5 °C in the control plants, which indicated progressive increase of transpiration. This phenomenon could be explained as a rapid initial stomata opening before the long-term closure in conditions of water deficit, known as "Iwanov" effect (Iwanov [@CR21]). Since soil water was not available at this point of the drought stress, we can speculate that plants gave up some intracellular water. Consequences of this were dramatic and different for the control as compared to Acadian^®^-treated plants. The next changes of leaf temperature showed different patterns: the temperature of the control leaves asymptotically approached thermal equilibrium (Fig. [6](#Fig6){ref-type="fig"}), while the Acadian^®^-treated plants demonstrated a gradual increase of leaf temperature, typical for ABA-controlled stomatal closure. It is obvious that the Acadian^®^ treatment improved the physiological ability to withstand the water stress. Three stages of plant stress-response, identified in our experiments, could be attributed to three stages of General Adaptation Syndrome (GAS): alarm, resistance and exhaustion/survival (Selye [@CR38]). There is still discussion whether or not GAS concept is applicable to plant physiology (Leshem and Kuiper [@CR30]). Regardless of the answer, our results indicate that treatment with Acadian^®^ seaweed extract provided successful adaptation and survival of treated plants as compared to control. However, the biochemical mechanism of the action of a commercial extract of *Ascophyllum nodosum* remain unclear and require further investigation.
Conclusions {#Sec11}
===========
Thermal imaging of leaf temperature can be used as an indicator of stomatal closure in response to soil water deficit. Turgor and leaf temperature changes provide vital information to enhance the scientific understanding of plant physiological stress. The early-stage response of plants to drought stress was modified by prior treatment with Acadian^®^ seaweed extract, which resulted in better adaptation and survival of Acadian^®^-treated plants.
Conceived and designed the experiment: AM, KS, ATC, WN. Performed the experiments: AM, KS, WN. Analyzed the data: AM, TA, KS, GP. Contributed reagents/materials/analysis tools: AM, KS, ATC, GP, WN. Contributed to the writing of the manuscript: AM, KS, TA, ATC. All authors read and approved the final manuscript.
Acknowledgements {#FPar1}
================
We would like to thank NSERC Engage Grant EGP 466287-14 for the financial support, provided to Dr. Alex Martynenko. Special thanks to Sarah Fancy for the processing thermal images.
Competing interests {#FPar2}
===================
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
In the past decade, there has been an increased interest in the photocatalytic degradation of various kinds of organic pollutants in water and soil \[[@R1]\]. Many of these pollutants, particularly dyes, are carcinogenic and mutagenic \[[@R2]\]. Thus, there is an urgent need for removal of these pollutants as these are harmful to both human and environment \[[@R3]\]. Previous studies have extensively explored the role of semiconductor oxides mainly ZnO, TiO~2~ in the photocatalytic decomposition of organic pollutants \[[@R4]--[@R8]\]. These semiconductor photocatalysts not only degrade the contaminants, but also cause their complete mineralization into CO~2~, H~2~O and mineral acids \[[@R9]--[@R10]\]. Thus, it is advantageous over physico-chemical methods such as flocculation--coagulation \[[@R11]\], ozonization \[[@R12]\] and adsorption \[[@R13]\], as these methods are unable to remove the contaminants completely. Some recent studies have reported ZnO as a better photocatalytic material in the degradation of organic dyes in aqueous solutions, because of high charge carrier mobility and significantly longer electron life time than TiO~2~ \[[@R14]--[@R16]\].
Zinc oxide is a well-known semiconductor with a band gap energy of 3.37 eV and has been widely explored as photocatalytic material due to its non-toxic nature, high exciton binding energy (60 meV), photosensitivity and stability on exposure to high energy radiation \[[@R17]\]. Due to this high band gap value, ZnO can only absorb ultraviolet (UV) light and this limits its practical applications \[[@R18]\]. Thus, in order to design more efficient photocatalysts, which are active in visible light, many research groups have devoted their studies towards dye sensitization \[[@R19]\], ion doping \[[@R20]\] and coupling of semiconductors \[[@R21]\]. Recently, coupling of the semiconductors have attracted much attention and it has been proved that this coupling efficiently increases the photocatalytic performance by reducing the recombination probability of photo-generated charge carriers, increasing the photo response range and enhancing the interfacial charge transfer \[[@R22]\].
We have focused our study on the coupling of ZnO nanorods with nanoparticles of CdS, a semiconductor active under visible light, to form coupled CdS--ZnO heterojunction nanostructures. The CdS nanoparticles are an attractive photocatalytic material for visible-light harvesting due to the narrow band gap (2.42 eV) \[[@R23]\]. But fast recombination of photo generated charge carriers and their aggregation to form large particles, limits the photocatalytic activity of CdS nanostructures \[[@R24]\]. Once these structures are coupled with other semiconductor materials to form nanocomposites, they turn out be an efficient photocatalyst \[[@R25]\]. Recently, there have been few reports available in literature on CdS--ZnO coupled photocatalytic systems with enhanced activity \[[@R26]\]. On illumination of light, charge transfer takes place from the conduction band (CB) of CdS to that of ZnO \[[@R22],[@R27]\].
The CdS--ZnO semiconductor nanostructures can be further supported on graphene/reduced graphene oxide (RGO) materials to improve their photocatalytic properties. Ideally, graphene is a single layer carbon sheet, which consists of a two dimensional (2D) network of sp^2^-hybridized carbon atoms with hexagonal packed lattice structure \[[@R28]\]. Graphene also possesses unique electronic, optical and mechanical properties such as high theoretical specific surface area (2630 m^2^·g^−1^) \[[@R29]\], chemical stability, high transparency and good thermal conductivity (5000 W·m^−1^·K^−1^) \[[@R30]\]. Its optical transmittance is about 97.7% and possesses superior electron mobility (200000 cm^2^·V^−1^·s^−1^), which makes it an ideal material for photocatalyst support \[[@R31]\]. Several semiconductor nanocomposites supported on graphene have been used as photocatalysts for the degradation of organic pollutants \[[@R32]--[@R35]\]. In one of our recent works \[[@R34]\], we have reported the synergistic effect of MoS~2~--RGO support to improve the photocatalytic performance of ZnO nanoparticles. However, the role played by RGO support in enhancing the photocatalytic performance of the nanocomposites has not been fully explored. Furthermore, the photocatalytic activity is also influenced by the irradiation source. With regard to this, we focused our studies on determining the role of the RGO support and the irradiation source on the photocatalytic activity of CdS--ZnO semiconductor nanostructures. Another unique aspect of this work is the formation of efficient binary and ternary heterojunctions having nanoparticles (NP), nanorods (NR) and nanosheets (NS), comprising of CdS, ZnO and RGO, respectively. In this work, , the preparation and detailed characterization of binary and ternary nanocomposites are presented and their photocatalytic activity have been demonstrated with respect to the degradation of methyl orange (MO) dye, both under visible light (only) irradiation from a solar simulator and natural sun light. The obtained results have been discussed in detail, and the role of RGO support and irradiation source on the photocatalytic activity of CdS--ZnO nanostructures has been elucidated.
Experimental
============
Materials
---------
For the synthesis of GO graphite powder (crystalline, −300 mesh, 99%) was purchased from Alfa Aesar, whereas sodium nitrate (NaNO~3~), sulfuric acid (H~2~SO~4~), potassium permanganate (KMnO~4~) and hydrogen peroxide (H~2~O~2~) were purchased from Merck. Zinc chloride (ZnCl~2~), sodium hydroxide (NaOH), cadmium acetate dihydrate (Cd(OOCCH~3~)~2~·2H~2~O), sodium sulfide (Na~2~S), ammonia solution and methyl orange were also supplied by Merck. Polyvinyl pyrrolidone (PVP) used in synthesis was purchased from Sigma-Aldrich. All chemicals were used as received without further purification. Deionized water (18.2 MΩ·cm) used in synthesis was obtained from a double-stage water purifier (ELGA PURELAB Option-R7).
### Synthesis of graphene oxide
Graphene oxide (GO) was synthesized from natural graphite flakes using a modified Hummers' method \[[@R36]\]. As described in literature \[[@R34]\], 1.0 g of graphite powder and 0.5 g of NaNO~3~ was stirred in 23 mL of concentrated H~2~SO~4~ in an ice bath to maintain a reaction temperature below 10 °C. This was followed by the slow addition of 3.0 g of KMnO~4~ to the reaction mixture with continuous stirring. Subsequently, the reaction mixture was stirred in an oil bath at 35 °C until a brown colored paste was formed after about 4 h. The reaction was terminated by slow addition of deionized water (90 mL), which increased the temperature to 95--98 °C and resulting suspension was maintained at this temperature for 15--20 min; subsequently, the suspension was then diluted to about 250 mL by the addition of deionized water. This is followed by the addition of 10 mL H~2~O~2~ to remove unreacted KMnO~4~ in the reaction mixture. In order to remove the ions of oxidant origin, the mixture was washed with 10% HCl and then with deionized water until pH value of the filtrate was neutral. Obtained graphite oxide was subjected to ultrasonication for its exfoliation followed by centrifugation at 4500 rpm for 15 min. The final product was obtained by drying with rotary evaporator at 40 °C followed by vacuum drying overnight at same temperature.
### Synthesis of ZnO nanorods
ZnO nanorods (NR) were synthesized through a previously reported solvothermal method \[[@R37]\]. In brief, 10 mL of 0.2 M zinc chloride (ZnCl~2~) solution in ethanol was added into 70 mL of 0.5 M sodium hydroxide (NaOH) solution dropwise under vigorous stirring. This was followed by ultrasonic treatment of the solution for 30 min for homogenization. Then, this homogenous solution was transferred to a 100 mL teflon-lined stainless steel autoclave, sealed tightly and maintained at 180 °C for 12 h. White precipitates of ZnO NR were collected by centrifugation and washed several times with deionized water and ethanol and finally dried at 60 °C.
### Synthesis of CdS nanoparticles
CdS nanoparticles were synthesized as per a previously reported method \[[@R38]\]. In a typical procedure, about 20 mL of cadmium acetate dihydrate (Cd(OOCCH~3~)~2~·2H~2~O) (0.2 M) was prepared in deionized water. To this solution, 20 mL of sodium sulfide (Na~2~S) solution (0.2 M) was added dropwise under continuous stirring. After 10 min of stirring, 0.5 g of PVP was added as capping agent under vigorous stirring. The pH value of the solution was maintained at around 10 by adding ammonia solution. The resultant solution was refluxed for about 1 h at 70 °C. Upon completion of the reaction, the product was washed with deionized water and ethanol thrice and finally yellow colored CdS NP were formed after drying in an oven at 80 °C for 2 h.
### Synthesis of CdS--ZnO binary nanocomposite
CdS--ZnO binary nanocomposite was prepared by employing a reported hydrothermal strategy \[[@R39]\]. In short, 0.2 M ZnCl~2~ was dispersed in 40 mL deionized water, followed by the addition of 0.5 M NaOH solution dropwise with continuous stirring. Aqueous ammonia was added to maintain the pH value around 8. Finally, Cd(OOCCH~3~)~2~·2H~2~O (0.2 M) and 4 mL of thioglycolic acid (0.2 M) was added into 40 mL of above solution under vigorous stirring. Subsequently, this homogenous suspension was transferred to 50 mL teflon-lined stainless steel autoclave and kept at 140 °C for 48 h. After the completion of reaction time, the product was collected by centrifugation and washed with ethanol and deionized water thrice, and dried in vacuum. Based on the molar ratio of the precursors used in the synthesis, the ratio of CdS to ZnO is expected to be 1:1 in this binary nanocomposite.
### Synthesis of CdS--ZnO--RGO ternary nanocomposite
For the synthesis of the CdS--ZnO--RGO ternary nanocomposite, about 0.01 g of as prepared GO was dispersed in ethanol by ultrasonication. Then, 0.2 g of previously prepared CdS--ZnO was added to the GO solution under vigorous stirring for 2 h to obtain homogenous suspension. Once homogenization is achieved, the suspension is transferred to teflon-lined stainless steel autoclave and kept at 120 °C for 24 h. The desired product was obtained after washing with water and dried at 60 °C. Based on the molar ratio of the precursors used in the synthesis, the ratio of CdS to ZnO is expected to be 1:1 while having 1 wt % of RGO in this ternary nanocomposite.
Photocatalytic activity
-----------------------
Photocatalytic activities of as prepared photocatalysts were evaluated by monitoring the decomposition of MO dye at room temperature. Typically, 10 mg of photocatalysts (CdS--ZnO or CdS--ZnO--GO) was added to 50 mL of the aqueous solution of MO (10^−5^ M). Initially, the suspension was magnetically stirred in the dark for 30 min to attain adsorption--desorption equilibrium. Subsequently, the suspension was continuously stirred under visible light irradiation. For one set of experiments, a solar simulator (OAI Trisol, AM 1.5, 100 mW·cm^−2^) having a UV cut-off filter (λ \> 420 nm) was used for visible light illumination. Another set of experiments were performed under natural sunlight and the light intensity was measured using a LX-101A digital luxmeter. At periodic time intervals, the photo-reacted suspension (ca. 1 mL) was analyzed by recording the absorbance using a UV--vis spectrophotometer.
Characterizations
-----------------
The UV--visible absorption spectra of the samples were recorded using Shimadzu UV-2450 spectrophotometer in the wavelength range from 200 to 800 nm. Fourier transform infrared (FTIR) spectra were collected using Agilent K8002AA Carry 660 FTIR instrument. Optical properties were analyzed by UV--vis diffuse reflectance spectroscopy (DRS) using a Perkin Elmer UV/VIS/NIR Lambda 750 spectrophotometer, in which polytetrafluroethylene (PTFE) polymer was employed as internal reflectance standard. Morphology of the samples was characterized using a field-emission scanning electron microscope (FESEM) JFEI Nova Nano SEM-450 and a high resolution transmission electron microscope (HRTEM) FEI Tecnai G2 20 S-twin microscope operating at 200 kV. Energy dispersive X-ray spectroscopy (EDAX) was obtained using the same HRTEM instrument. X-ray diffraction (XRD) measurements were done using the Agilent Supernova X-ray diffractometer using Ni-filtered Cu Kα irradiation (λ = 0.1542 nm) at 45 kV and 40 mA in 2θ ranging from 5 to 80° with a scan rate of 2°·min^−1^.
Results and Discussion
======================
Powder X-ray diffraction analysis
---------------------------------
In order to investigate the crystalline phase of as prepared nanocomposites, powder X-ray diffraction (XRD) analysis was performed. [Fig. 1](#F1){ref-type="fig"} presents the XRD patterns of graphite, GO, ZnO NR, CdS NP, CdS--ZnO and CdS--ZnO--RGO nanocomposite. Graphite powder shows a very strong peak at 2θ = 26.5°, which could be assigned to (002) reflection plane corresponding to the interlayer distance of about 0.34 nm. In addition to this peak there is another (004) reflection peak at 2θ = 54.3° corresponding to the interlayer distance of 0.17 nm \[[@R40]\]. After the oxidation of graphite, the interlayer distance increases mainly due to the introduction of hydroxy, epoxy and carbonyl groups, which is indicated by the characteristic (001) reflection peak of about 2θ = 8.5°, which correspond to an interlayer distance of about 1.08 nm \[[@R41]\]. The disappearance of characteristic (002) reflection peak and appearance of (001) reflection peak confirms the oxidation of graphite and formation of GO with well-defined lamellar structure \[[@R42]--[@R43]\]. This interlayer distance weakens the van der Waal interactions between sheets and makes exfoliation possible \[[@R44]\]. Once GO is reduced to RGO during hydrothermal treatment, the (002) reflection peak of GO disappears. The XRD patterns of ZnO nanorods show peaks at 2θ = 31.67, 34.31, 36.14, 47.40, 56.52, 62.73, 66.28, 67.91 and 69.03°, which could be assigned to the (100), (002), (101), (102), (110), (103), (200), (112) and (201) lattice planes, respectively, indicating the prepared ZnO NR have polycrystalline wurtzite structure (JCPDS no. 36-1451) \[[@R37]\]. The powder XRD pattern of the prepared CdS NP shows three diffraction peaks at 2θ = 26.8, 44.1 and 55.2°, which corresponds to the (111), (220) and (311) planes of hexagonal CdS (JCPDS 42-1411) \[[@R23]\]. Furthermore, in the powder XRD data of nanocomposites, all peaks that are ascribable to ZnO and CdS structures are evident, which demonstrates that the same crystal phases are retained in both binary (CdS--ZnO) and ternary (CdS--ZnO--RGO) nanocomposites. In the ternary composite, the characteristic peak of GO around 2θ = 10° is absent confirming its reduction to RGO. The characteristic diffraction peak around 2θ = 24° due to RGO is also absent in the ternary composite, which indicates that RGO sheets are not stacked due to the CdS--ZnO composite which inhibits the stacking.
{#F1}
UV--vis diffuse reflectance spectroscopy (DRS)
----------------------------------------------
The optical properties of all prepared samples were analyzed by using UV--vis diffuse reflectance spectroscopy (DRS), the results of which are presented in [Fig. 2](#F2){ref-type="fig"}. It is clear from the DRS spectra that ZnO NR have an absorption band edge at 390 nm corresponding to a band gap value of 3.23 eV, which is in agreement with the reported band gap value of ZnO \[[@R45]\]. As-prepared CdS NP have an absorption edge around 580 nm, which corresponds to a band gap value of about 2.12 eV in agreement with the value reported for CdS \[[@R23]\]. Bare GO also shows the excellent light absorption in the range of 200--800 nm. Various studies have confirmed that the band gap of GO changes with the degree of oxidation \[[@R46]--[@R47]\]. The CdS--ZnO binary nanocomposite shows higher absorption in the visible light region compared to that of ZnO NR and exhibits an absorption edge in the range of 400--500 nm indicating the presence of CdS NP in the binary composite. One can observe a slight decrease in the band gap value of ZnO in this binary nanocomposite compared to its pristine form. The addition of GO to form the ternary composite (CdS--ZnO--RGO) results in continuous absorption in the region of 400--800 nm. Similar to the binary nanocomposite, in this case as well, two distinct absorption edges, corresponding to band gap values of 3.01 and 2.11 eV attributable to ZnO NR and CdS NP, respectively, could be evidenced. The enhanced absorption in visible light region can be attributed to chemical bonding between semiconductors and specific sites of carbon in GO resulting in charge delocalization and hence narrowing of the band gap of semiconductors \[[@R48]--[@R49]\]. The DRS spectra of ternary nanocomposite shows a broad elevated background in visible region, which is mainly due to GO because CdS does not show any absorption edge above its fundamental band edge (580 nm) \[[@R23]\]. Thus the presence of GO affects the optical properties of the ternary nanocomposite and is responsible for the red shift in the absorption spectrum, which ultimately results in narrowing of the band gap, which not only enhances light absorption in the visible light region, but also facilities efficient mobility of the charge carriers between the two semiconductors. Plots obtained by the transformation of the Kubelka--Munk function vs the energy of light are presented in [Fig. 3](#F3){ref-type="fig"}, which clearly shows the band gap narrowing in both the semiconductors due to the addition of GO.
{#F2}
{#F3}
SEM analysis
------------
The surface morphology of all pristine materials as well as the binary (CdS--ZnO) and ternary (CdS--ZnO--RGO) nanocomposites was investigated by scanning electron microscopy (SEM). [Fig. 4](#F4){ref-type="fig"},b clearly shows the GO sheets and CdS NP, respectively. GO sheets show a flake-like morphology and the pristine CdS NP are agglomerated. ZnO has a rod-like morphology with lengths in the range of 2 to 3 µm and diameters from 30 to 50 nm ([Fig. 4](#F4){ref-type="fig"},d). In the preparation of nanocomposites, two semiconductors nanostructures (ZnO NR and CdS NP) are coupled first with each other to form a binary nanocomposite and then with GO sheets through hydrothermal method to form the ternary nanocomposite. [Fig. 5](#F5){ref-type="fig"},b indicate the coupling between CdS and ZnO, wherein CdS NP are present on the surface of ZnO NR. This binary nanocomposite (CdS--ZnO) on the surface of RGO can be seen in [Fig. 5](#F5){ref-type="fig"},d to form the ternary nanocomposite.
{#F4}
{#F5}
TEM analysis
------------
Transmission electron microscopy (TEM) analysis was performed to obtain further information on the surface morphology and microstructures of all prepared nanostructures ([Fig. 6](#F6){ref-type="fig"}). The TEM image of GO shows the flake-like shape ([Fig. 6](#F6){ref-type="fig"}). The CdS NP are about 20 nm in size ([Fig. 6](#F6){ref-type="fig"}). The TEM images of the CdS--ZnO binary and CdS--ZnO--RGO ternary nanocomposites are shown in [Fig. 6](#F6){ref-type="fig"} and [Fig. 6](#F6){ref-type="fig"}, respectively. It can be seen that the CdS--ZnO nanocomposite is distinctly coupled with RGO sheets and the original 2D structure of GO sheets is still retained even after hydrothermal treatment, which is in good agreement with literature reports \[[@R48]\]. The existence of all the constituent components in the final binary and ternary nanocomposites has been proved by the presence of corresponding peaks in the energy dispersive X-ray spectra (EDAX), as shown in [Fig. 6](#F6){ref-type="fig"} and [Fig. 6](#F6){ref-type="fig"}. The results obtained from TEM analysis corroborate well with both powder XRD and SEM characterizations.
{#F6}
FTIR analysis
-------------
FTIR spectra for all samples are presented in [Fig. 7](#F7){ref-type="fig"}. The FTIR spectrum of GO shows strong broad absorption peaks at 3309 cm^−1^ and 1404 cm^−1^, which could be attributed to the O--H stretching vibrations and deformation vibrations of intercalated water \[[@R50]\]. Absorption peaks at 1032 cm^−1^, 1229 cm^−1^ and 1725 cm^−1^ are the characteristic stretching vibrations of C--O, epoxy C--O, and C=O of carbonyl groups, respectively, which are present at the edges of the GO sheets \[[@R51]\]. The absorption peak at 1622 cm^−1^ could be attributed to the aromatic C=C stretching vibrations in GO \[[@R51]\]. Similarly in FTIR spectra of ZnO NP, the broad absorption band at 3454 cm^−1^ and 1424 cm^−1^ could be attributed to O--H stretching and deformation of C--OH groups of water molecules. Zn--O bond stretching vibrations appears at 504 cm^−1^ \[[@R52]\]. FTIR spectra of CdS NP also reveal the presence of O--H stretching vibrations of adsorbed water molecules on its surface. The peak at 1550 cm^−1^ is attributed to the C--N stretching vibration of the PVP monomer, which was used as capping agent \[[@R53]\]. The 1404 cm^−1^ peak corresponds to the C--H bond of PVP \[[@R54]\]. CdS--ZnO nanocomposite also shows O--H stretching vibrations in the range of 3000--3500 cm^−1^ range and Zn--O stretching vibrations between 500 and 600 cm^−1^. The peak at 1394 cm^−1^ in the binary composite is assigned to C--H bonds of the capping agent. It is clear from the FTIR spectrum of CdS--ZnO--RGO nanocomposite that characteristic peaks of oxygen containing functional groups particularly at 1725 cm^−1^ are weakened and the O--H stretching peak decreases with some red shift. This is mainly attributed to the loss of oxygen containing functional groups and the reduction of GO to RGO after the hydrothermal treatment \[[@R55]\].
{#F7}
UV--vis spectroscopic analysis
------------------------------
[Fig. 8](#F8){ref-type="fig"} shows the UV--vis spectra of GO, CdS, ZnO, CdS--ZnO and CdS--ZnO--RGO nanostructures. The UV--vis study of GO determined the degree of conjugation by λ~max~ value and shows two absorption peaks with maximum at 227--230 nm due to aromatic π→π\* transition and a small shoulder at 303 nm due to the n→π\* transition of carbonyl groups \[[@R56]\]. The absorbance peak of ZnO appears at 372 nm, which is in good agreement with literature reports \[[@R57]\]. UV--vis spectra for CdS NP shows a clear absorbance band in 400--500 nm range. which corresponds to the visible region of spectrum \[[@R58]\]. The UV--vis spectrum of the binary nanocomposite (CdS--ZnO) shows the absorption both in the UV and the visible region confirming the presence of both CdS and ZnO in the composite. Finally the absorption spectra of ternary nanocomposite (CdS--ZnO--RGO) shows an absorption near 250 nm, which indicates the red shift of the band at 227 nm in GO. This red shift is mainly due to the reduction of GO to RGO during the hydrothermal reaction, indicating an increase of the electronic conjugation \[[@R59]\]. Enhanced absorption of this ternary nanocomposite in visible region affirms the presence of all three components.
{#F8}
Photocatalytic performance
--------------------------
The photocatalytic performance of CdS--ZnO binary and CdS--ZnO--RGO ternary nanocomposites is evaluated by measuring the photodegradation of methyl orange (MO), as model dye, under visible light irradiation from a solar simulator or under natural sun light. Prior to illumination, the suspension was equilibrated in dark for 30 min. During this period adsorption and desorption equilibrium was achieved between the photocatalyst and MO. This is followed by illumination either under solar simulator or under natural sun light. The degradation of MO was studied by measuring the concentration of MO with a UV--vis spectrophotometer at regular intervals of time. The UV--vis spectra are shown in [Fig. 9](#F9){ref-type="fig"}. The corresponding kinetic curves are shown in [Fig. 10](#F10){ref-type="fig"}, which indicates that in all the cases the reaction follows zero-order kinetics similar to other reports in literature \[[@R60]\].
The degradation efficiency of both the photocatalysts was evaluated on the basis of initial and final concentration of the dye by monitoring the main absorption peak (λ = 454 nm) of MO. So degradation rate of the MO can be calculated by applying following equation \[[@R61]\]:
where *A* *~t~* and A~0~ are the absorbance at reaction time *t* and *t* = 0, respectively.
In accordance with the above equation, with CdS--ZnO--RGO ternary photocatalyst about 98% of dye was degraded in 90 min, under visible light illumination from a solar simulator, but with CdS--ZnO binary photocatalyst, only about 70% of dye was degraded. The degradation of the same concentration of MO was also investigated under natural sunlight illumination having an intensity of 9.5 × 10^4^ lux. Substantial improvement in the photocatalytic activity can be observed for both binary and ternary nanocomposites as shown by the histogram in [Fig. 11](#F11){ref-type="fig"}. It is noteworthy that about 98% degradation of the dye could be achieved within 60 min, when the CdS--ZnO--RGO ternary nanocomposite was used as the photocatalyst, which is 30 min shorter than under the simulated solar light. Similarly, within 60 min the CdS--ZnO binary nanocomposite degraded about 70% of the dye, while it took about 90 min under the visible light irradiation from a solar simulator. The enhanced performance of the nanocomposites can be attributed to the fact that the natural sun light has both UV and visible light components in it, so both of the semiconductor materials (CdS NP and ZnO NR) are active and electron--hole pair formation occurs in both. Hence, the generation of higher number of electron--hole pairs, their effective charge separation and charge transfer are the major factors responsible for the better activity of the photocatalysts under natural sunlight compared to the visible light irradiation using a solar simulator.
{#F9}
{#F10}
{#F11}
Mechanism of photocatalytic activity
------------------------------------
The possible mechanisms of the photocatalytic activity of CdS--ZnO--RGO ternary nanocomposite for degradation of MO under visible light irradiation from a solar simulator and natural sun light are illustrated pictorially in [Scheme 1](#C1){ref-type="fig"} and [Scheme 2](#C2){ref-type="fig"}, respectively.
{#C1}
{#C2}
Under visible light illumination, electron--hole pairs are generated in conduction band (CB) and valence band (VB) of CdS. The CB potential of CdS are *E* ~CB~ = −0.66 eV vs NHE, which is more negative than the *E* ~CB~ of ZnO \[[@R62]--[@R63]\]. This relative position of CB of CdS with respect to ZnO leads to the charge transfer from CB of CdS to the CB of ZnO \[[@R22]\]. The charge transfer dynamics in the CdS--ZnO composite material have been investigated before and it has been reported that this transfer occurs very rapidly in less than 18 ps \[[@R64]\]. The work function of ZnO is 5.2--5.3 eV and that of graphene is 4.5 eV \[[@R65]\]. Thus electrons that are transferred to the CB of ZnO are rapidly transferred to the RGO, as the Fermi level of graphene (−0.08 V vs NHE) is more positive than the redox potential of O~2~/O~2~ ^−^ (−0.13 V vs NHE) but more negative than the redox potential of O~2~/H~2~O~2~ (+0.695 V vs NHE) \[[@R66]\]. This demonstrates that electrons from graphene can react with O~2~ and H^+^ ions to produce H~2~O~2,~ which further decomposes in the presence of light to generate hydroxyl radicals (**^•^**OH) \[[@R67]\]. Also ZnO is inactive under visible light, so hole formation takes place only in the VB of CdS. As the VB edge of CdS is more negative than the standard redox potential of ^•^OH/OH^−^ (2.38 eV vs NHE) and ^•^OH/H~2~O (2.72 eV vs NHE), ^•^OH cannot be generated by reacting with H~2~O molecules. The **^•^**OH radicals are responsible for the dye degradation into CO~2~ and H~2~O \[[@R68]--[@R69]\]. The enhanced photocatalytic activity of the RGO-supported CdS--ZnO nanocomposite could be attributed to the presence of RGO, which has a very good dye adsorption and fast electron transport ability \[[@R70]--[@R71]\]. The enhanced activity under natural sun light could be attributed to the combined activity of both CdS and ZnO, wherein CdS absorbs in the visible region and ZnO absorbs in the UV region, and charge carriers are generated in both of these semiconductors. In this case, in addition to the electron transfer from the CB of CdS to the CB of ZnO, simultaneous hole transfers also occur from the VB of ZnO to VB of CdS, as the VB of CdS is more cathodic than the VB of ZnO \[[@R72]\]. Thus the recombination of photogenerated charges is suppressed more effectively under natural sunlight illumination, where both the semiconductor materials are active and more electron--hole pairs are generated, and their effective separation and rapid transport to the reaction site are responsible for the enhanced activity of the photocatalysts.
Overall, the improved photocatalytic activity of the ternary nanocomposites could be mainly attributed to the better adsorption capacity of graphene, rapid charge transfer at the semiconductor interface and then to graphene, which degrade the adsorbed dye on its surface. This whole proposed mechanism can be formulated as \[[@R67],[@R73]--[@R74]\],
Furthermore, [Table 1](#T1){ref-type="table"} presents the comparison of the photocatalytic activities of CdS--ZnO--RGO, CdS--RGO and ZnO--RGO nanocomposites prepared by different routes for the degradation of various pollutants, including MO. [Table 1](#T1){ref-type="table"} also reveals that our reported CdS--ZnO--RGO nanocomposite exhibits an enhanced photocatalytic performance compared to other reports based on similar materials. The reason for better photocatalytic performance has been discussed in detail above in the mechanism section.
######
Comparison of photocatalytic activities of CdS--ZnO--RGO based nanocomposites for degradation of pollutants, including methylene blue (MB), rhodamine B (Rh B) and MO.
-------------------------------------- ---------------------- ---------------------------- ------------------------- ------------------------ ------------
photocatalyst synthesis route irradiation source pollutant concentration *t* ~completion~ (min) ref.
CdS--ZnO core-shell coupled with RGO soft chemical route simulated solar radiations 3.0 × 10^−5^ M (MB) 80 \[[@R75]\]
ZnO--RGO--CdS hydrothermal 11 W UV lamp 1.0 × 10^−5^ M (MB) ca. 240 \[[@R76]\]
ZnO--graphene solvothermal halogen lamp 6.0 × 10^−6^ M (MO) 90 \[[@R32]\]
3D grapene--ZnO NR CVD and hydrothermal UV light 300 W 5.0 × 10^−3^ M (MO) ca. 60 \[[@R77]\]
graphene--ZnO NR film hydrothermal 250 W Hg lamp (UV) 3.0 M (MB) ca. 450 \[[@R78]\]
CdS--RGO composite wet chemical method UV unknown (MO) 120 \[[@R79]\]
RGO--ZnO NR composite hydrothermal 500 W Hg Lamp (UV) 1.0 × 10^−5^ M (Rh B) 90 \[[@R37]\]
CdS--graphene composite hydrothermal 500 W Xe lamp 3.0 × 10^−5^ (MO) 360 \[[@R80]\]
CdS--ZnO--RGO hydrothermal visible light 1.0 × 10^−5^ M (MO) 90 this work
sunlight 1.0 × 10^−5^ M (MO) 60
-------------------------------------- ---------------------- ---------------------------- ------------------------- ------------------------ ------------
Conclusion
==========
In this work, we have prepared and thoroughly characterized CdS--ZnO semiconductor nanostructures both with and without RGO support. Their photocatalytic activity towards the degradation of methyl orange dye, has been investigated both under visible light irradiation from a solar simulator and under natural sunlight. The obtained results show the significant role played by the RGO support and the source of irradiation on the photocatalytic activity of the mixed metal chalcogenide nanocomposites. The RGO-supported CdS--ZnO nanocomposites exhibits considerably better photocatalytic activity compared to its unsupported counterpart, which could be attributed to the enhanced photo-generated charge separation, facile charge transfer and strong adsorption of dye on to RGO. In addition, superior photocatalytic activity was observed for the nanocomposites irradiated under natural sunlight than visible light from solar simulator. This could be ascribed to the higher generation of electron--hole pairs, their effective separation and rapid transport to the reaction site. In this case, both the semiconductors are active, in their respective wavelength domains, as sunlight is comprised of both UV and visible light regions. This work not only demonstrates the role of the RGO support and irradiation source on the activity of photocatalysts, but also paves way for tailoring the photocatalytic activity of semiconductor nanostructures in general.
We are thankful to Advanced Materials Research Centre (AMRC), IIT Mandi for laboratory and the characterization facilities. VK and VS acknowledges the financial support from Department of Science and Technology, India under INSPIRE Faculty Award. SK acknowledges Research Fellowship from UGC, New Delhi, India.
| {
"pile_set_name": "PubMed Central"
} |
See related research by Bøyesen *et al.*, <http://arthritis-research.com/content/11/4/R103>
The elegant study by Bøyesen and colleagues \[[@B1]\] in the previous issue of *Arthritis Research & Therapy*, in which they observed the changes in hand bone mineral density (BMD) using digital x-ray radiogrammetry (DXR) in 163 patients with early rheumatoid arthritis (RA), is most welcome. DXR measures the cortical thickness, width, and porosity of the central parts of metacarpals 2 to 4. After 1, 2, and 5 years of observation, the decreases in hand BMD were -1.7%, -2.8%, and -5.6%, respectively. Elevated levels of both erythrocyte sedimentation rate (ESR) and anti-cyclic citrullinated peptide (anti-CCP) were independent predictors of hand bone loss at all time points, indicating that the largest amounts of bone loss were found in patients with severe inflammation and high levels of auto-antibodies; from earlier data, we know that these patients are prone to develop severe RA with radiological joint damage \[[@B2]\].
RA is a chronic progressive disorder characterised by synovitis, bone and cartilage degradation of the joints, and extra-articular symptoms. Based on the dramatic improvement in treatment options in RA, clinical remission is a realistic target of therapy \[[@B2]\]. In particular, the use of biologicals early in the course of RA is exciting and addresses two of the most important research questions in modern rheumatology: is it possible to predict which RA patients will have a favourable response to conventional (methotrexate, or MTX) therapy? And what are the characteristics of RA patients with an unfavourable prognosis, estimated by radiological joint damage, for which more aggressive therapy is indicated?
Thus, there is an urgent need to develop validated assessment tools for identifying patients who are at risk for a poor prognosis, estimated by early radiological joint damage. Juxta-articular bone loss and subchondral bone oedema due to the replacement of marrow fat by heavily vascularised inflammatory cells are the earliest signs of bone involvement in RA and may even precede the development of radiologically detectable erosions of the joint \[[@B3],[@B4]\].
In the Norwegian study, changes in hip and spine BMD were not mentioned, but in the BeST (Behandel Strategieën) study (also in patients with early RA), it was observed that the hand bone loss was roughly two to five times higher than the generalised bone loss at the spine and hips \[[@B5]\]. This difference presumably reflects the fact that BMD changes at the hands are more sensitive to cytokine stimulation in the adjacent inflamed joints. Güler-Yüksel and colleagues \[[@B5]\] observed that the changes in hand BMD mirror the progression of radiological joint damage according to the Sharp-van der Heijde score (SHS). This inverse relationship suggests that adjacent joint damage and cortical metacarpal bone loss presumably result from the same mechanisms: inflammation-driven upregulation of tumour necrosis factor-alpha, interleukin-1 (IL-1), IL-6, and receptor activator of nuclear factor-kappa B ligand (RANKL), leading to stimulation of osteoclastic bone resorption, in combination with inhibited osteoblastic bone formation, at least partly related to the upregulation of dickkopf-1 \[[@B6]\]. If indeed the mechanisms are similar, does aggressive antirheumatic therapy protect against both local joint damage and cortical metacarpal bone loss? The answer is yes, as shown for combination therapy of MTX with adalimumab, for combination therapy with MTX and prednisone or infliximab, and for denosumab \[[@B7]-[@B9]\].
One limitation of the study is the lack of data on clinical joint inflammation and radiographic joint damage (SHS) for determining the correlation in space and time with cortical metacarpal bone loss. Indeed, periarticular trabecular bone loss, cortical bone loss (which is most pronounced at the endocortical site), and subchondral bone oedema are probably the first detectable signs of subchondral inflammation. This could be explained by direct communication between the joints and bone marrow by upregulated local blood flow or by the presence of radiologically still-undetectable small erosions or by both \[[@B2],[@B3]\]. Unfortunately, DXR does not allow us to specify the degree to which bone loss occurs at the endocortical bone site and at the periosteal site. The tight relationship between SHS and hand bone loss has another consequence: it could be argued that the SHS at baseline is missing in the Norwegian prediction model. Another limitation is the lack of data on bone markers, such as the RANKL/osteoprotegerin ratio and bone and cartilage markers (CTX-1 and -2), which independently predict radiographic joint damage in RA \[[@B10],[@B11]\].
The main disadvantage of the SHS is that it is a time-consuming and difficult scoring method, usually performed only in clinical trials. The use of hand DXR is potentially attractive since the measurements can be performed with standard x-rays of the hands and because of the early finding of local hand bone loss. On the other hand, the devices to measure the metacarpals are not widely available, and sending the digital radiographs to a central laboratory may be costly. Second, a crucial question remains: what is the clinical relevance of measuring local hand bone density? However, the results of the Norwegian study should incite researchers to further analyse the sequence of juxta-articular bone events (bone oedema and periarticular trabecular and cortical bone loss and erosions) in early RA.
In summary, the observation that elevated levels of both ESR and anti-CCP are predictors of hand bone loss measured by DXR seems to be adequate and important. Nevertheless, more data and clinical experience are needed before the use of hand DXR can be successfully introduced in daily practice.
Abbreviations
=============
anti-CCP: anti-cyclic citrullinated peptide; BMD: bone mineral density; DXR: digital x-ray radiogrammetry; ESR: erythrocyte sedimentation rate; IL: interleukin; MTX: methotrexate; RA: rheumatoid arthritis; RANKL: receptor activator of nuclear factor-kappa B ligand; SHS: Sharp-van der Heijde score.
Competing interests
===================
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
A fence is a useful windbreak because of its effect on wind erosion[@b1][@b2][@b3][@b4], sand movement and deposition[@b5][@b6][@b7], microclimate[@b8][@b9] and soil conditions[@b10] ([Fig. 1](#f1){ref-type="fig"}). Many studies have been conducted to study the influence of the shape, height, spacing and porosity of a fence on changes in airflow fields and on the reduction in the leeward wind velocity. Porosity is defined as the ratio of the total pore area to the total fence area. The critical porosity of a fence, below which flow separation and reversal occur, is approximately 0.3 according to previous works[@b11][@b12][@b13], whereas its optimal porosity is determined to be approximately 0.3 to 0.4 by most researchers[@b2][@b14][@b15][@b16]. Airflow regions behind a low-porosity or solid fence can be identified as: outer flow, overflow, an internal boundary layer, a reverse cell and a small vortex[@b13].
Considering the difficulty in capturing the entire flow field around a fence at an actual site, inherent advantages, such as controllability and repeatability, have made wind tunnel experiments one of the most fundamental methods for studying the influence of a fence on airflow and the effect of fence porosity. For detailed investigations, several types of advanced equipment, such as laser Doppler velocimeters, particle tracking velocimeters and particle image velocimeters (PIVs) have been applied by researchers to study the entire airflow field, particularly leeward of fences; such equipment are superior to conventional pitot tube anemometers or hot-wire anemometers[@b4][@b5][@b13][@b16]. The use of non-intrusive methods is currently limited to wind tunnel experiments because such techniques are inapplicable in the field in terms of releasing fine tracer particles and capturing them at a fixed plane using a charge-coupled device camera.
Key issue in this type of research are determining whether the experiment results are directly applicable to the actual conditions and determining the extent of the result\'s application based on the following reasons: (1) the fence model in a wind tunnel is typically reduced to the millimetre or centimetre level, (2) the schematic of a porous fence and the materials used are hardly similar to those of an actual windbreak and (3) the roughness heights and inlet wind profiles between a wind tunnel floor and a natural land surface are different. Moreover, the ratio of the measured reference height of inlet wind in a wind tunnel to the height of the model is generally not proportional to that in the field; thus, making it is difficult to validate the accuracy of the experiment results under an actual tested wind speed condition. The scale problem has been mentioned by several researchers[@b17][@b18][@b19]; however, no specific investigation is found in the literature.
An independent method that can conduct a comparison between an experimental model in a wind tunnel and an actual fence, while keeping other boundary conditions similar to their actual conditions, will determine whether the experiment results can be applied to actual situations. Computational fluid dynamics (CFD) simulation provides a possible solution to this issue. CFD has already been proven as a reliable means to study the wind flow field of a windbreak to discuss its protective effect[@b6][@b17][@b20][@b21][@b22] and other complex geometries[@b23][@b24][@b25][@b26][@b27]. The literature is also rich with validations of CFD studies on windbreak flow using wind tunnel results[@b8][@b28]. However, these previous works were frequently based on same-scale comparisons. In the present study, we used a CFD simulation method to study the scale effect on mean velocity fields behind fences with different porosities. In particular, we compared the airflow fields of a porous fence model with those of its on-site counterpart.
The wind tunnel experiment performed by Dong et al.[@b13] was selected as a reference because (1) it presented a high-precision result via PIV and (2) this experiment had a practical application for the fences along the highway of the Taklimakan Desert in China, as shown in [Figure 1a](#f1){ref-type="fig"}. The fence in this wind tunnel experiment has a height of 20 mm (*H*) and a thickness of 1.2 mm; these values represent a 1 m high and 0.06 m thick reed-bunch fence with a geometric scale of 1:50. A numerical fence model with the same size in the same wind tunnel experiment was developed 14 *H* after the inlet and 26 *H* before the outlet. The height of the simulation domain was set to 12 *H*, which was sufficient for the complete development of the airflow. To understand how the airflow characteristics around the fence change with different scales, the model was magnified 50 times to represent the actual scenario.
An inlet wind profile was provided by the following logarithmic law:where *u*(*z*) is the wind speed value at a *z*(m) height, *k* = 0.41 is the von Karman coefficient, *u*\* is the friction velocity and *z*~0~ is the surface roughness. Four wind speed conditions, i.e., 8 m/s, 10 m/s, 12 m/s, and 14 m/s, were considered at a 30 cm height in the wind tunnel scenario and a 2 m height in the actual scenario. A value of *z*~0~ = 0.004 mm was used in the wind tunnel simulation based on the measured wind profiles in the wind tunnel, whereas *z*~0~ = 0.001 cm, which is a typical value over a blown sand surface, was used in the actual scenario.
We developed nine porosity scenarios (*η* = 0, 0.05, 0.1, 0.15, 0.2, 0.3, 0.4, 0.5 and 0.6). Conditions in which the porosity exceeds 0.6 were not considered because previous works show that the porosity effect is weak above these values. A total of 72 simulation cases are required for the four wind speed conditions and the two scale levels for each fence. The fence can be regarded as a thin porous boundary that acts as a momentum sink and causes a pressure drop *ΔP* in the flow passing through it:where *v* is the normal velocity toward the fence, *k~r~* is the pressure loss coefficient related to the porosity of the fence and *ρ* is the air density.
The porosity values were defined by changing the *k~r~* of the fence instead of the solid model itself. To estimate the pressure loss coefficient at various fence porosity conditions, two common empirical relationships are available. One of these relationships was from Reynolds[@b29], i.e.,whereas the other relationship was from Hoerner[@b30], i.e.,
The estimation results from the two relationships have the same change patterns given that the *k~r~* values of both are related to 1/*η*^2^. The results from the Reynolds equation are smaller when *η* \< 0.06, and vice versa, with limited absolute differences. However, the relative differences are significant at large porosity conditions, reaching approximately 40% when *η* = 0.6. To reduce errors from the estimation, the average results from the two equations were adopted in our simulations. Moreover, the permeability of the fences was set to a high value of 10^10^ m^2^ to cancel out viscous effects.
The current study may be the first to simulate flow around porous wind fences at different scale scenarios to test the scale problem. As a standard procedure to validate the numerical method and the applicability of the turbulence model, the results of the current study were first compared with the PIV-measured wind tunnel results. If the simulation results were directly comparable with those of the wind tunnel study, then the scale problem of a reduced numerical or wind tunnel model can be discussed. The scale effect was evaluated by comparing wind speed contours and wind profiles at 13 locations (−2, −1, −0.5, 0.5, 1, 2, 4, 6, 9, 12, 16, 20, and 25 *H*) in the wind tunnel-sized simulations with those from the actual-sized simulations.
Results
=======
Comparison with the previous wind tunnel experiment
---------------------------------------------------
The flow field behind a porous fence can be characterised by the shape and size of the reverse cell, the location of the flow separation zone and the flow speed at several specific locations. The flow convergence distance on the leeward side of the fence, which is where the forward flow and reversed backward flow meet to form a zero-speed contour, was sensitive to changes in the fence porosity. The comparison between the flow convergence length from the wind tunnel and the CFD simulation at the two fence scales is shown in [Fig. 2](#f2){ref-type="fig"}. The experiment results are consistent with those of the scaled simulation. A general trend in which the convergence length increased with the wind porosity was observed for all wind speed conditions, except for the wind tunnel result with a wind speed of 8 m/s and *η* = 0.2. This abnormal value may have resulted from an experimental error.
Similar to many previous studies, the CFD-simulated flow field characteristics, including streamline patterns, velocity profiles and flow regions, are consistent with the results of the wind tunnel experiments; thus, the applicability of the numerical method is proven. In addition, Dong et al.[@b13] noted that the convergence length does not increase when the porosity is lower than 0.05. The simulation at the wind tunnel model scale exhibited the same features at all wind speed conditions, further proving that the CFD method could represent the flow field with high precision and reliability. However, the simulation with an actual-sized model resulted in an obviously low relative length when the porosity was less than 0.1 and a high value when the porosity was larger than 0.15. This finding was affected by the different relative intensities of a small vortex that occurred immediately behind the fence at low porosity conditions.
A small vortex behind a solid or low-porosity fence was observed in the wind tunnel[@b13] and in both numerical simulations in the present study. A direct comparison of the vortex in the two scale simulations reveals the difference caused by the scale effect ([Fig. 3](#f3){ref-type="fig"}). Under a 10 m/s wind condition, the vortex behind the solid wind tunnel model fence was well developed to approximately a 0.6 *H* height and 1 *H* distance, whereas the vortex behind the actual size solid fence had a height and distance of only approximately 0.15 *H* and 0.25 *H*, respectively; the relatively small vortex in the latter led to the short convergence length of the actual-sized simulation in [Fig. 2](#f2){ref-type="fig"}. If the measured wind tunnel and CFD-simulated data were extrapolated 50 times to their actual size, then the height and length of the vortex would be 4 times larger than those of the vortex in the actual-sized model. This scale-affected difference indicates that although the general flow characteristics can be captured using scale models in a wind tunnel or same-scale CFD simulations, as tested by many previous works, the results of these methods will not be proportional to the model size and can be deformed if applied directly to a unique area, particularly near obstacles.
Comparison of flow fields at two scaled fences
----------------------------------------------
In the current study, we focused on changes in flow velocity that were caused by different model scales. First, the result shows that the wind speed had a limited effect on the flow pattern, as proven by many previous studies. For each scale and porosity scenario at 13 locations, the wind profiles under four referential wind speeds were proportional and highly correlated with one another (*r* = 1). The percentage differences in the wind velocity of the two scale simulations were calculated by averaging the difference at each measured location under the four wind speed conditions. The average percentage difference at each porosity scenario was then determined by obtaining the average of the differences at all the measured locations.
A direct comparison between the flow fields behind the two scaled fences revised several general similarities and differences caused by the scale effect. The lift streamline that appoached the fence, the flow compression above the fence, the bleed flow immediately after the fence, the high-velocity region above the fence, the flow reattachment distance and the shape of the separation cell for the two scale simulations are consistent, as shown in [Fig. 4](#f4){ref-type="fig"} for a porosity of 0.3 and an inlet wind speed of 10 m/s.
However, a significant difference was observed, i.e., the overall wind speed in the actual-sized simulation was apparently higher than that in the wind tunnel-sized simulation ([Fig. 4](#f4){ref-type="fig"}). This finding is attributed to the referential wind speed in the wind tunnel that was monitored as 0.3 m (15 *H*) above the surface, whereas in the actual-sized simulation, the speed was 2 m high, which corresponds to 2 *H* as the typical field condition. Another difference was observed in the varying relative strengths of the bleed flow behind the porous fence. In the actual-sized simulation, the space for the bleed flow development was larger, and lead to a longer convergence length and a more intensive leeward deformation of the separation cell. In contrast, the space behind the 20 mm-high wind tunnel fence was limited; thus, the intensity of the bleed flow was smaller behind the fence.
A quantitative comparison of the wind profiles at various locations emphasised that the difference in the outer-flow and overflow zones was caused by the varying ratios of the referential wind speed height to the fence models and by the displacement of small vortices or reverse cells leeward ([Fig. 5](#f5){ref-type="fig"}). At the 13 compared locations, the difference in the total wind speed of the 9 porous conditions was 21.1%, which was close to the average wind speed difference of 22.9% from the inlet boundaries. The data are stable at the outer-flow and overflow regions, as well as behind the separation cells. However, the data fluctuate, with large standard deviations, behind the fence that was affected by high-intensity turbulence, particularly near or within a small vortex and before the separation cell of low porosity fences because the absolute wind speed is low and the two flow patterns differ in these areas; thus, a small difference in values can result in a highly variable ratio. These differences indicate that the results from the wind tunnel simulation at a certain referential wind speed or for particular characteristics in a scale model are not directly applicable to the actual condition, even when the same inlet wind speed is assumed. The results also reiterate that the porous effect ceases when *η* is larger than 0.5.
Discussion
==========
This study presents the first numerical analysis of the scale problem of a windbreak in wind tunnel experiments and reduced-scale simulations. The flow field difference that was influenced by the scale effect between the two fence models was numerically measured. The result of the comparison between the wind tunnel experiment and the CFD simulation with an equal size is encouraging, and emphasizes the reliability of the CFD method. However, two major differences between the flow fields of the scale model and the actual-sized fence were determined.
In the outer-flow and overflow regions where the fence had limited influence, the difference was mainly caused by the varying ratios of the referential wind speed heights in the model and of the actual-sized fences. This problem is common because a wind tunnel model is typically scaled down tens or several hundred times from the actual size, which can be generally solved by considering referential wind speed height and model size, for example, by multiplying by approximately 122% in the current study.
Moreover, the small vortex and the separation cell behind a low-porosity fence are not proportional to the size of the model; thus, a significant discrepancy occurs between the wind tunnel and the actual-sized simulations immediately behind the fence and within the separation cell. For example, the small vortex behind a low-porosity fence is relatively four times larger in the 1:50 scale model. This problem can only be addressed by comparing two numerical models at the required scales and then extrapolating the results of a wind tunnel experiment to the actual condition.
Methods
=======
ANSYS Fluent code (version 12.0), which was frequently used in previous studies, was also used in our simulations. Fluid (air) is considered incompressible and Newtonian. The shear stress transport (SST) *k-ω* turbulence model was used to provide closure. The SST model automatically changes from the standard turbulence/frequency-based *k-ω* model in the inner region of the boundary to a higher Reynolds number *k-ε* model in the bulk flow. The accuracy and reliability of this model can be increased for a broader class of flows than the standard *k-ε* and *k-ω* models[@b31].
Boundary conditions
-------------------
The wind profile given in [equation (1)](#m1){ref-type="disp-formula"} was imposed as the inlet boundary condition with a changing *u*\* to alter the referential wind speed, whereas for the top boundary the inlet velocity at its height was assigned. The non-slip boundary condition was applied to the floor with *z*~0~, as described previously. The zero-pressure boundary condition was adopted for the outlet boundary. The porous condition of the fence was applied according to the *k~r~* calculated from [equations (3)](#m3){ref-type="disp-formula"} and (4) and the width of the fence. The solid fence with a porosity of 0 was modelled as a wall.
Discretisation scheme and specifications of the turbulence model
----------------------------------------------------------------
The fully developed turbulent wind flow over the fence was solved by the Reynolds-averaged Navier-Stokes equations that use the finite volume method. The default "standard wall function" applied the wall boundary condition to all variables of the *k-ω* model that were consistent with the logarithmic wind profile in [equation (1)](#m1){ref-type="disp-formula"}. The SIMPLEC algorithm was used to solve pressure-velocity coupling. The pressure interpolation scheme was PREssure Staggering Option (PRESTO!). Second-order discretisation schemes were used for the convection and viscous terms of the governing equations. The least-squares cell-based gradient reconstruction scheme was used because of its accuracy and speed. A structured quadrilateral grid method was used in the upstream and downstream regions, whereas an unstructured triangular grid was built near the fence model to bridge fine cells at the bottom of the domain and coarse cells at the top. The grid resolution was determined via grid-sensitivity analyses; thus, the results were independent of the grid size. The total number of cells was 5,100.
The initial conditions for pressure and velocity were set to zero for the entire domain, except at the left inlet boundary wherein the logarithmic wind profile was imposed. The transport equations for the SST *k-ω* model were then solved iteratively until the convergence criteria were satisfied. The convergence criteria were defined in terms of residuals, i.e., the degree to which the conservation equations were satisfied throughout the flow field. We assumed that convergence was achieved when all normalised residuals for the velocity components *k* and *ω* fall below 10^−5^. The execution time of each run was approximately 15 minutes on a 3.40 GHz Intel i7-4770 processor with 8 GB of RAM.
Author Contributions
====================
B.L. and J.Q. designed the study and wrote the manuscript. W.Z. and L.T. participated in the numerical simulations. Y.G. analysed the data and reviewed the manuscript.
We thank Dr. Tianli Bo for the informative discussions on wind profile inhomogeneity in a wind tunnel. This work was supported by the nonprofit industry special research funds from the Chinese Ministry of Water Resources (201201047, 201201048), China Postdoctoral Science Foundation (2014M550518), the Opening Fund of the Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions (LPCC201301) and the West Light program for Talent Cultivation of Chinese Academy of Sciences.
{#f1}
{#f2}
{#f3}
{#f4}
{#f5}
| {
"pile_set_name": "PubMed Central"
} |
Published: April 3, 2012
Introduction {#sec1}
============
The three-dimensional structures revealed by X-ray analysis and other molecular imaging techniques provide a vast amount of information on the structural, functional, and evolutionary relationships among proteins. Discovery and comprehension of the often delicate and intricate relationships are greatly assisted by appropriate structure-matching techniques. To capture the whole extent of structural correlations, it is frequently necessary to compare oligomers and molecular complexes rather than individual chains. There is a steady supply of new and updated methods for the pairwise comparison of polypeptide chains (for reviews, see, e.g., [@bib42; @bib18; @bib43; @bib35]), but only a few methods are known that accept multiple-chain proteins as input (e.g., [@bib25; @bib28; @bib29]).
The detection of structural matches among protein complexes is hampered by a number of problems and intricacies. The correlations generally extend over many protein chains, where individual chains may contribute only partially to structurally similar regions. Structure-matching techniques have to cope with the flexibility of protein chains, domain movements, subunit rearrangements, and structural models derived from poorly resolved electron densities. Proteins frequently contain symmetries and repetitions that result in independent multiple matches that cannot be superimposed simultaneously. To capture all these relationships, it is necessary to report and comprehend the whole spectrum of structural matches ([@bib37]). Beyond this, the largest molecular complexes known today, like virus capsids ([@bib47; @bib30]), ribosomes ([@bib44; @bib4]), RNA-polymerase ([@bib9]), chloride channel ([@bib11]), cell-puncturing devices ([@bib21]), or clathrin-coated vesicles ([@bib16]) may contain well over 10,000 amino acid residues. The mere size of these structures poses a considerable computational challenge.
The structure-matching technique presented here, called TopMatch, works across the whole bandwidth of protein architectures and molecular dimensions, ranging from peptide chains to large molecular complexes. TopMatch integrates established structure-matching techniques ([@bib33; @bib34; @bib35; @bib36; @bib14; @bib38; @bib39; @bib37; @bib40]) with appropriate techniques for the processing of large molecular complexes. The remarkable aspects introduced here are the construction of composite alignments from precise local alignments and the definition of an alignment score that combines alignment length with the spatial deviation of superimposed structures. The score is used as a measure of significance of structural correlations and provides a convenient numeric criterion for the ranking of alignments. The construction of composite alignments is a strategy that successfully copes with many of the problems and intricacies encountered in the comparative analysis of large molecular complexes.
Detailed quantitative accounts of structural correlations among large molecular complexes are rare. Hence, a particular focus of this communication is to provide a collection of various types of relationships that are biologically relevant. The cases are chosen to highlight many of the peculiarities encountered in the comparison of complex structures. The problems discussed below are all challenging in the sense that they involve distant relationships among large protein chains, oligomers, and molecular aggregates. Conventional structure-matching techniques are generally unable to tackle these problems successfully, since multiple chains are involved. These examples, as well as any other structure-matching problem, can be solved interactively using the TopMatch web service.
The presentation is organized in two parts. The first is a description of the properties of structural matches and alignments. This information is required for the subsequent discussion of structural correlations. In the second part, we investigate a number of structural correlations in some detail.
Results {#sec2}
=======
Basic and Composite Alignments {#sec2.1}
------------------------------
The primary result obtained from a pairwise comparison of a query structure, *Q*, and a target structure, *T*, is a set of precise basic alignments. The construction of basic alignments has been described ([@bib33; @bib14; @bib38; @bib37]) and benchmarked (e.g., [@bib38]) previously. This set of basic alignments captures the whole spectrum of structural similarities among *Q* and *T*. In the case of large proteins and protein complexes, there are generally several basic alignments that are independent in the sense that they do not overlap with respect to the amino acid sequence of *Q* and *T*. Frequently, such alignments can be combined to form a single composite alignment, provided larger spatial deviations are tolerated than in the construction of basic alignments.
The construction of composite alignments involves a combinatorial problem, since proper subsets have to be found among the set of basic alignments. The associated problem of combinatorial explosion can be largely circumvented when the basic alignments are ranked by a proper measure of significance. Below we review the basic properties of alignments, define and discuss a measure of significance for alignment ranking, and describe the construction of composite alignments from basic building blocks.
Properties of Structure Alignments {#sec2.2}
----------------------------------
In the following, the structures of proteins are represented by C*^α^* atoms. This level of approximation is sufficiently precise for most applications. The residues within a protein are numbered consecutively from the N to the C terminus. Multiple chains are joined to a single chain where the order is immaterial. Hence, in terms of the inner workings of TopMatch there is no difference between single- and multiple-chain proteins.
An alignment, *A*, consists of a collection of *n* blocks, $A = \left\{ {B_{1},B_{2},\ldots,B_{n}} \right\}$. A block, *B~p~*, corresponds to a gapless alignment of a fragment in *Q* and a second fragment of the same length in *T*. To be precise, each block *B~p~* is defined by three numbers, *a~p~*, the start of the block in *Q*, *b~p~*, the start of the block in *T*, and *s~p~*, the length of the block. Hence, *B~p~* contains the residue pairs $\left( {a_{p} + i,b_{p} + i} \right)$, for $0 \leq i < s_{p}$, aligning fragment $\left( {a_{p},\ldots,a_{p} + s_{p} - 1} \right)$ of *Q* to fragment $\left( {b_{p},\ldots,b_{p} + s_{p} - 1} \right)$ of *T*. The *n* blocks of an alignment, *A*, are disjoint, i.e., they do not have any residues in common.
A particular structure alignment, *A*, relates two molecules in a certain way. The relationship is rather complex and we need several parameters to describe it. In particular, the length of an alignment, the corresponding spatial deviation of C*^α^* atoms, the similarity of the aligned sequences, and the number of permutations are essential aspects of structure alignments.
The length, *L*, of *A* is obtained as$$L = \sum\limits_{p = 1}^{n}s_{p},$$where *n* is the number of blocks in *A* and *S~p~* is the length of block *B~P~*. Relating *L* to the number of residues, *Q~L~*, of *Q*, we obtain the query cover,$$Q_{c} = 100 \times \frac{L}{Q_{L}},$$and, similarly, the target cover,$$T_{c} = 100 \times \frac{L}{T_{L}},$$where *T~L~* is the size of the target. These numbers measure the extent of similarity in terms of the number of structurally equivalent residue pairs ([@bib34]).
The spatial deviation of two sets of coordinates *Q* and *T* is commonly reported in terms of the residual root-mean-square error *E~r~* calculated from the optimal superposition of *Q* and *T*,$$E_{r} = \sqrt{\frac{1}{L}\ \sum r_{i}^{2}},$$where$$r_{i}^{2} = \left( {\mathbf{x}_{i} - \mathbf{y}_{i}} \right)^{2},$$is the squared distance between the equivalent atoms **x***~i~* (query) and **y***~i~* (target). Here, *L* is the alignment length and $i = 1,\ldots,L$ labels the pairs of equivalent C*^α^* atoms as defined by the respective alignment. The procedure to compute the optimal superposition is described in the Computational Methods section.
It is clear that the similarity between *Q* and *T* increases with increasing alignment length, *L*, and decreases with increasing residual error, *E~r~*. Both aspects are conveniently combined by Gaussian functions,$$S = \sum\limits_{i}^{L}e^{{- r_{i}^{2}}/\sigma^{2}},$$where the *r~i~* are computed from the optimal superposition of *Q* and *T* as defined above. We call *S* = *S*(*A*) the similarity of *Q* and *T* with respect to alignment *A*. For a perfect match, the *r~i~* are zero and the sum *S* evaluates to the alignment length, *L*. Generally, $0 \leq S \leq L$, where the actual value is controlled by the scaling factor *σ*. Dividing *S* by *L* yields the similarity on a per-residue basis,$$s = \frac{S}{L} = \frac{1}{L}\ \sum\limits_{i = 1}^{L}e^{- r_{i}^{2}/\sigma^{2}}.$$
The typical distance error is then obtained from *s* as$$S_{r} = \sigma\ \sqrt{- \text{ln}\ s}.$$
From the construction of *S~r~*, it follows that there is a strong correlation between *S~r~* and *E~r~* (e.g., [Table 1](#tbl1){ref-type="table"}).
In the computation of the similarity *S*, the scale parameter *σ* determines the rate of reduction of alignment length *L* as a function of increasing distance error, *S~r~* (and hence *E~r~*). Therefore, an appropriate value for *σ* can be found from$$\sigma = \frac{S_{r}}{\sqrt{- \text{ln}\ s}}.$$
For example, if we require that an average distance error of *S~r~* = 3.0 Å corresponds to a relative similarity of *s* = 80%, then the appropriate value of *σ* is,$$\sigma = \frac{3.0}{\sqrt{- \text{log}\left( 0.8 \right)}} = 6.35\ \text{Å}.$$
All alignments described here are ranked using *σ* = 7 Å. Although this definition of *σ* involves the average quantity *s*, it is clear that in the computation of the similarity *S* the distances *r~i~* are weighted individually by the exponential $\text{exp}\left( {- r_{i}^{2}/\sigma^{2}} \right)$. A distance *r~i~* = 0 always has weight 1 (independent of *σ*). For *σ* = 7, a distance error of *r~i~* = 3 scales to 0.83, whereas *r~i~* = 6 scales to 0.48; i.e., with increasing error the weight drops in a nonlinear fashion, as determined by the Gaussian function.
Clearly an alignment *A* also establishes a relationship among the sequences of *Q* and *T*. The similarity among the sequences is expressed as the percentage of pairs of identical residues,$$I_{s} = 100 \times \frac{I_{p}}{L},$$with *I~p~* the absolute number of such pairs and *L* the alignment length.
By default, TopMatch generates alignments that may contain permutations of the sequence of *T* relative to *Q*. To detect permutations, the blocks of *A* are sorted in ascending order relative to the sequence of *Q*, i.e., $a_{p} < a_{p + 1}$. In a linear alignment, we always have $b_{p} < b_{p + 1}$ for sequence *T*. Hence, a discontinuity, $b_{p} > b_{p + 1}$, corresponds to a permutation of sequence *T* relative to *Q*. The number of permutations, *P*, is the total number of such discontinuities found among the blocks of *A*. Within a single chain, permutations may point to substantial rearrangements in the respective genes or they may reveal distinct topologies of a particular protein fold. However, in the alignment of multiple chains, the relative order of monomers is arbitrary, and permutations of blocks that involve distinct chains of *T* are not counted as permutations. In the construction of alignments by TopMatch, permutations can be suppressed but then the alignments of multichain complexes may be incomplete.
Construction of Composite Alignments {#sec2.3}
------------------------------------
As already noted, in a pairwise comparison of two structures, *Q* and *T*, TopMatch generally yields several basic alignments ([@bib37]). For example, the comparison of the viral capsid proteins 3izx-A and 3iyl-W ([Figure 1](#fig1){ref-type="fig"}) produces 23 distinct basic alignments, and for large proteins and complexes the number may rise to several hundred. In terms of the residual error, these alignments are quite precise, with *Er* values generally below 3.5 Å. In larger proteins and protein complexes, individual alignments may cover distinct regions of the molecules. They can be combined to larger composite alignments provided the respective basic alignments are compatible in sequence and in structure.
To ensure the compatibility of the sequences of two alignments, *A*~1~ and *A*~2~, we remove all blocks of *A*~2~ that collide with blocks in *A*~1~. Simply, a block in *A*~2~ collides with a block in *A*~1~ whenever the blocks share residues with respect to *Q* or *T*. On the other hand the structures of *A*~1~ and *A*~2~ are compatible when the superposition of *Q* and *T* according to alignment *A*~1~ also superimposes those parts of the structure that are covered by alignment *A*~2~, at least approximately. Therefore, we superimpose *Q* and *T* according to *A*~1~ but evaluate the spatial deviation with respect to *A*~2~ only. Then the respective similarity, *S*, and distance errors, *S~r~* and *E~r~*, are proper estimates for the compatibility of *A*~1~ and *A*~2~. In particular we may control structural compatibility of *A*~1~ and *A*~2~ by a proper bound on the distance error. The particular bound used here is $S_{r} < 6\ \text{Å}$.
With the combination of two alignments defined in this way we are ready to construct composite alignments. First, the basic alignments are sorted by similarity, *S*, so that $S\left( A_{i} \right) > S\left( A_{i + 1} \right)$. Then *A*~1~ is the alignment of maximum similarity *S*. To construct composite alignment, *C*~1~ is set as *C*~1~ = *A*~1~. Then, for each *A~i~ ≠ A*~1~, remove colliding blocks and evaluate *S~r~*(*A~i~*) with respect to the transformation obtained from *C*~1~. If *S~r~*(*A~i~*) is acceptable, set *C*~1~ = *C*~1~ +*A~i~* and continue with *A~i~*~+1~. Once all the *A~i~* are consumed, *C*~1~ corresponds to a composite alignment or is equal to *A*~1~. Finally the parameters *L*, *S*, and so on are computed for the complete alignment *C*~1~. The procedure is repeated with *C*~1~ = *A*~2~, and so on until the basic alignments are exhausted. Finally the *C~i~* are sorted by *S*. The TopMatch web service reports the top five alignments with the parameters defined above together with the alignment type (*b* for basic and *c* for composite alignments) and the query and target structures are superimposed and displayed in a graphical window, as described in the Computational Methods section.
In general, the extent of similarity among two structures *Q* and *T* cannot be captured by a single optimal alignment even if basic alignments are combined to composite alignments. There are several reasons for this. Proteins, in particular assemblies of protein chains, frequently have internal symmetries that translate into multiple structural matches. The corresponding alignments are usually indistinguishable in terms of alignment length, but they match distinct regions of the molecules. In fact, the number of alignments that are connected by symmetry operations is directly related to the symmetries within and between two molecules. In a similar way, repetitions in protein structures result in a multitude of alignments, although in these cases the alignments are generally not related by symmetry operations.
Proteins exhibit various types of structural variation. A particular molecule may have distinct arrangements of chains and structural domains in different crystal structures depending on the specific molecular environment. Similarly, in the evolution of proteins, mutations may cause large-scale movements of one structure relative to another. In all such cases, a rigid body superposition can only match parts of the molecule, and hence, several alignments are required to capture the whole extent of similarity. The structural relationships discussed in the following sections exemplify several of these relationships and the concepts introduced here. They are chosen to illustrate specific issues and subtleties encountered in the comparison of structures, but they also highlight several intricate and perhaps astonishing connections among protein molecules. The respective figures highlight the structural correlations, whereas the associated alignment parameters are assembled in [Table 1](#tbl1){ref-type="table"}.
Distortions and Domain Movements in Polypeptide Chains and Protein Complexes {#sec2.4}
----------------------------------------------------------------------------
The size and flexibility of large protein complexes combined with the limited resolution of the molecular models obtained provide particular challenges for structure-matching techniques. For example, the capsid proteins found in the shells of the cytoplasmatic polyhedrosis virus (Bombyx mori cyprovirus, 3izx, resolution 3.1 Å \[[@bib46]\]) and the subvirion particle of aquareovirus (grass carp reovirus, 3iyl, resolution 3.3 Å \[[@bib47]\]) were both determined by cryo-electron microscopy. The structures of chain A of 3izx (1,057 amino acid residues) and chain W of 3iyl (1,284 amino acid residues) share a number of similarities revealed by a series of structural matches with corresponding alignment lengths around 200 residues or less ([Figures 1](#fig1){ref-type="fig"}A and 1B; [Table 1](#tbl1){ref-type="table"}). These basic alignments can be combined to composite alignments covering large regions of the molecules with *L* = 634 residues and *E~r~* = 5.6 Å ([Figure 1](#fig1){ref-type="fig"}C). The strong correlations among the structures of these capsid proteins point to a common origin, although the percentage of identical residues, which is below 15% in all the alignments obtained, indicates that there is no correlation on the sequence level.
Regarding the construction and interpretation of structural alignments, this example is instructive in several ways. The many dislocations and contortions frequently observed among the structures of large molecules and at low resolution promote mix-ups of strands of *β*-sheets, helix bundles, and other repetitive structures that are close in space, resulting in alignments that are contaminated by mismatched structural elements. Forcing the construction of alignments to proceed from precise basic alignments to composite alignments completely avoids these pitfalls. Whereas composite alignments reveal the global extent of similarity, basic alignments put the focus on those regions where two molecules match quite precisely. Such regions frequently correspond to functional sites, or they may point to structural regions that are important for folding and stability of molecular structures.
The effect of large-scale movements in related oligomeric proteins is illustrated using the structures of the human γδ T cell antigen receptor (1hxm, chains A and B, \[[@bib1])\]) and the Fab fragment of a humanized monoclonal antibody (3qcv, chains H and L, \[[@bib15] \]). Each chain of these homodimers consists of two immunoglobulin domains. Since all domains (four in each dimer) have similar folds, there are a multitude of structural matches between the two dimers, resulting in more than 100 basic alignments. The most significant of these alignments (*L* = 223, *E~r~* = 2.2 Å; [Figure 2](#fig2){ref-type="fig"}A; [Table 1](#tbl1){ref-type="table"}) matches the N-terminal domains of 1hxm (residues 1--126 of chains A and B) with the N-terminal domains of 3qcv (residues 1--109 of chains H and L). The result immediately reveals that the mode of dimerization of these N-terminal domains is the same in both molecules. At the same time, the orientation of the C-terminal domains relative to the N-terminal domains is quite different. However, a second alignment matches 179 residue pairs of the C-terminal domains ([Figure 2](#fig2){ref-type="fig"}B; [Table 1](#tbl1){ref-type="table"}). Again, due to their conserved mode of dimerization, the two domains of each dimer superimpose with a small residual error of *E~r~* = 2.6 Å, and taken together, the similarity among the N- and C-terminal domains covers the complete dimers. However, the respective basic alignments are not joined to a composite alignment, since the spatial deviation obtained from their simultaneous superposition exceeds 6 Å.
Structural Correlations among Dehydrogenases {#sec2.5}
--------------------------------------------
The dehydrogenase family is a widespread group of enzymes participating in many essential metabolic pathways. Many of these enzymes contain domains (Rossman-fold), which are used for the binding of the cofactors NAD and NADH. The molecules frequently assemble to homodimers and higher-order oligomers. The comparative analysis of these molecules reveals a number of surprising large-scale structural transitions that must have occurred in the evolution of these enzymes ([Figure 3](#fig3){ref-type="fig"}; [Table 1](#tbl1){ref-type="table"}). The biologically active unit of human hydroxyisobutyrate dehydrogenase ([2i9p](pdb:2i9p) \[[@bib22]\]) seems to be a homotetramer consisting of 1,175 amino acid residues. The resulting protein complex has several symmetry elements, reflecting the construction of the molecule from four identical chains ([Figure 3](#fig3){ref-type="fig"}A). The structure of the tetramer segregates into five domains. The large central-helical domain is formed by the C-terminal parts of the four chains, which are heavily entangled, whereas the four NAD-binding domains, each corresponding to the N-terminal part of an individual chain, stick out from the complex.
The enzyme 6-phosphogluconate dehydrogenase from E. coli ([3fwn](pdb:3fwn) \[[@bib8]\]), forms a homodimer composed of 934 amino acid residues ([Figure 3](#fig3){ref-type="fig"}B). The NAD binding domains of [2i9p](pdb:2i9p) and [3fwn](pdb:3fwn) are quite similar, but beyond that, the structures of the individual chains appear rather different (not shown). However, the superposition of the complete oligomers reveals that the [3fwn](pdb:3fwn) dimer is completely covered by the [2i9p](pdb:2i9p) tetramer ([Figure 3](#fig3){ref-type="fig"}AB). In particular, the position and orientation of the two NAD-binding domains match quite precisely. From the symmetry of [2i9p](pdb:2i9p), it is clear that there is a second equivalent match with [3fwn](pdb:3fwn) involving the two NAD-binding domains on the opposite side of the molecule (blue domains in [Figure 3](#fig3){ref-type="fig"}AB).
Another dehydrogenase-related enzyme is 2-dehydropantoate 2-reductase from the proteobacterium Methylococcus capsulatus ([3i83](pdb:3i83) \[[@bib6]\]). The active molecule ([Figure 3](#fig3){ref-type="fig"}C) is a homodimer of 587 amino acid residues. Again the structure of the [3i83](pdb:3i83) dimer is largely covered by the [2i9p](pdb:2i9p) tetramer, but in a rather different way compared to [3fwn](pdb:3fwn), since the [3i83](pdb:3i83) dimer crosses the [2i9p](pdb:2i9p) tetramer diagonally ([Figure 3](#fig3){ref-type="fig"}AC). Again the two matching NAD-binding domains are in a very similar relative position and orientation in spite of the large distance between them. And again due to the symmetry of [2i9p](pdb:2i9p), there is a second equivalent superposition along the second diagonal of [2i9p](pdb:2i9p) (not shown). Given the similarity of [2i9p](pdb:2i9p) to [3fwn](pdb:3fwn) and [3i83](pdb:3i83), it follows that there must be a match between the latter two molecules. This is indeed the case ([Figure 3](#fig3){ref-type="fig"}BC), but the match covers only one NAD-binding domain and parts of the central-helical domain. The pairwise mutual sequence similarity among all three molecules is below 20%, but their common structural features are quite extensive.
Given the strong structural correlations among these molecules, the question arises how these structures have evolved. Each of the three complexes is built from the information contained in a single gene. Therefore, the structural transitions that lead from one complex to another should be reflected by recombination events on the gene level. In fact, it seems rather obvious that [3fwn](pdb:3fwn) is the result of a partial internal duplication of a gene of the [2i9p](pdb:2i9p)-type. The mutation quite precisely duplicates the C-terminal part of the [2i9p](pdb:2i9p) chain that is used for the assembly of the central domain. However, whereas in [2i9p](pdb:2i9p) four chains congregate, in [3fwn](pdb:3fwn), the central domain can only accommodate two of the elongated chains. Hence, the partial duplication of the [2i9p](pdb:2i9p) gene yields an elongated gene of the [3fwn](pdb:3fwn) type, but in terms of structure, the mutation effectively splits off two NAD-binding domains from the complex. Structures of the [2i9p](pdb:2i9p) and [3fwn](pdb:3fwn) type are common in eukaryotes and eubacteria, suggesting that the internal duplication leading from [2i9p](pdb:2i9p) to [3fwn](pdb:3fwn) occurred before the divergence of these kingdoms several billion years ago. Besides these examples, the dehydrogenase family contains many correlations that point to intriguing structural transitions in the evolution of these enzymes (e.g., [@bib35]). It is clear that such relationships are comprehensible only if the biologically active oligomers are taken into account.
Pores and Needles {#sec2.6}
-----------------
Bacteriophage T4 uses an efficient cell-puncturing device ([1k28](pdb:1k28) \[[@bib21; @bib26]\]) for infecting cells. The device, consisting of a baseplate, a tail-tube with integrated lysozymes, and a membrane-puncturing needle, has a three-fold symmetry axis. A complex molecule like this is expected to have many structural matches with other proteins of variable size and function. Here, we explore two quite distant relationships.
A clear match is obtained between the baseplate ring of [1k28](pdb:1k28) and the protein secretion apparatus ([1y12](pdb:1y12) \[[@bib27]\]) from *Pseudomonas aeruginosa* ([Figure 4](#fig4){ref-type="fig"}). The similarity is restricted to the inner wall of the ring structure. The six-fold symmetry axis of [1y12](pdb:1y12), which superimposes quite well with the three-fold symmetry of the cell-puncturing device, results in six equivalent rigid-body superpositions. The accuracy of these matches is quite amazing. In each case, 339 pairs of equivalent residues can be superimposed to a residual error of *E~r~* = 2.9 Å where the fraction of identical residue pairs is as low as 8%. The various views of the superimposed structures reveal that the architecture and dimensions of the inner rings, including the inclination of the *β* strands is the same in both molecules ([Figure 4](#fig4){ref-type="fig"}). Another intriguing detail is the precise match of the three helices at the periphery of the baseplate ring of [1k28](pdb:1k28) with three of the six helices of the secretion apparatus, [1y12](pdb:1y12) ([Figure 4](#fig4){ref-type="fig"}C).
A quite exceptional match is obtained between the needle of the cell-puncturing device ([1k28](pdb:1k28)) and the structure of pertactin (1dab \[[@bib12]\]), a Bordetella pertussis virulence factor ([Figure 5](#fig5){ref-type="fig"}). The match is rather unexpected since the [1k28](pdb:1k28) needle is a homotrimer, whereas the pertactin needle is built from a single chain that folds into a *β* helix. The triangular cross-section ([Figure 5](#fig5){ref-type="fig"}) shows that the [1k28](pdb:1k28) trimer has a perfect three-fold symmetry, whereas the 1dab monomer is rather distorted, with many loops protruding predominantly from one of the three sides of the molecule. Given the distinct genetic architectures of these molecules it seems quite impossible that they could have evolved from a common ancestor. Since the overall shape of the molecules is astonishingly similar (321 equivalent residues, *E~r~* = 3.9 Å, *I~s~* = 10%), and since both structures perform quite similar functions (adhesion, docking, penetration), this is an example where a particular molecular device is realized by grossly distinct architectures.
DNA Clamps {#sec2.7}
----------
DNA sliding clamps are ring-shaped proteins of approximate hexagonal symmetry. Their ring structures are highly conserved throughout all kingdoms of life. In particular, the trimeric rings of eukaryotes and archaea are entirely equivalent to the dimeric clamps found in bacteria ([Figure 6](#fig6){ref-type="fig"}). For example, the structures of [2pol](pdb:2pol) ([@bib23]), the dimeric *β*-subunit of the E. coli polymerase III, and [1plq](pdb:1plq) ([@bib24]), the trimeric S. cerevisiae DNA polymerase processivity factor PCNA, match precisely ([Figure 6](#fig6){ref-type="fig"}A; [Table 1](#tbl1){ref-type="table"}). The monomers of DNA clamps ([Figure 6](#fig6){ref-type="fig"}B) contain two (eukaryotes and archaea) and three domains (bacteria). From the approximate hexagonal symmetry, it is clear that the individual domains within the monomers have similar structures. For example, domains D1 and D2 of [1plq](pdb:1plq) are equivalent ([Figure 6](#fig6){ref-type="fig"}C; [Table 1](#tbl1){ref-type="table"}), although their sequences are uncorrelated (*I~s~* = 14%).
At present, proteins corresponding to a single DNA clamp domain are unknown, but there is a striking similarity to the monomeric chains of [2gia](pdb:2gia) ([@bib32]), a mitochondrial RNA-binding complex ([Figure 6](#fig6){ref-type="fig"}D; [Table 1](#tbl1){ref-type="table"}). In terms of the spatial arrangement of structural elements, the two folds are entirely equivalent. However, the alignment contains five permutations, and hence, the topologies of the DNA clamp domain and the RNA-binding proteins are grossly distinct ([Figures 6](#fig6){ref-type="fig"}E and 6F). Also, the DNA- and RNA-binding sites of these molecules are in different locations. Sliding clamps interact with DNA through helices H2 and H7 ([Figure 6](#fig6){ref-type="fig"}E), whereas in the RNA-binding complex, the interaction is mediated by strands S6--S9 on the opposite side of the fold ([Figure 6](#fig6){ref-type="fig"}F). Hence, the structural similarity between these proteins is a curiosity demonstrating that the formation of a particular protein fold can be achieved in entirely different ways.
Large Complexes {#sec2.8}
---------------
Ferritins are iron-storage proteins that form spherical particles consisting of 24 identical amino acid chains. Despite their low sequence similarity (*I~s~* = 17%), the structures of the individual chains of bacterioferritin from Mycobacterium smegmatis ([3bkn](pdb:3bkn), resolution 2.72 Å, 161 amino acids \[[@bib19]\]) and horse apoferritin ([2za6](pdb:2za6), resolution 1.75 Å, 171 amino acids \[[@bib45]\]) superimpose rather precisely with an alignment length of *L* = 152 residue pairs and a corresponding residual error of *E~r~* = 2.4 Å ([Figure 7](#fig7){ref-type="fig"}A). The superposition of the spherical particles (3,864 and 4,104 amino acids, respectively) matches 3,649 residue pairs (*E~r~* = 2.6 Å; [Figure 7](#fig7){ref-type="fig"}B) and hence, the structure of this complex is strongly conserved over large evolutionary distances. Due to the symmetry of the molecules, there are 24 equivalent alignments of identical length and residual error, but the individual alignments link distinct chains. This result implies that all 24 chains are in indistinguishable structural and chemical environments. The degree of structural conservation among these complexes is quite amazing, since the geometry of the interfaces between subunits, as well as the conformations of the subunits themselves, has remained largely invariant in spite of the extensive divergence of their sequences, which resulted in the replacement of more than 80% of amino acid types. Conversely, it is equally surprising that the geometric constraints imposed by the architecture of these molecules can be satisfied by grossly distinct sequences.
Among the largest structures known today are the shells of several viruses. For example, satellite tobacco necrosis virus, a single-stranded RNA virus, forms a capsid containing a total of 11,040 amino acid residues ([2buk](pdb:2buk), resolution 2.45 Å \[[@bib20]\]). The capsid of Sesbania mosaic virus, another single stranded plant RNA virus, has been studied in various modified forms. The CP-NΔ36 deletion mutant ([1vb4](pdb:1vb4) \[[@bib30]\]) generates a capsid containing 11,760 amino acid residues. The capsids of [2buk](pdb:2buk) and [1vb4](pdb:1vb4) are both constructed from 60 identical subunits that assemble to the *T* = 1 variant of the virus particles ([@bib7; @bib13]). The subunits are single polypeptide chains whose folds contain a central *β*-sandwich. Roughly two thirds of the monomeric chains can be superimposed (*L* = 115, *E~r~* = 3.3 Å \[[Figure 8](#fig8){ref-type="fig"}A\]). The structures of proteins containing *β*-sheets always match to a certain extent, even if there is no phylogenetic relationship among them. Hence, the match does not necessarily imply a common origin of these viruses, and the sequences are in fact unrelated (*I~s~* = 4%).
On the other hand, both capsids are envelopes of single-stranded plant RNA viruses, both contain 60 identical subunits, and both have icosahedral symmetry. In both cases, the individual chains form pentamers corresponding to the 12 faces of the icosahedral capsids. These coincidences evoke a possible phylogenetic relation that might exist between these structures. In fact, the pentamers of [2buk](pdb:2buk) and [1vb4](pdb:1vb4) can be superimposed as a whole where the *β*-sandwiches in the pentamers attain similar location and orientation ([Figure 8](#fig8){ref-type="fig"}B). However, the mode of association between two adjacent pentamers differs considerably among the two capsids, so that pairs of pentamers cannot be superimposed simultaneously ([Figures 8](#fig8){ref-type="fig"}C and 8D).
The observed structural correlations are perhaps not strong enough to verify the hypothesis of a common phylogenetic origin of these plant viruses. Rather, the important point here is that the capsids do have remarkable correlations that can be detected, exposed, and immediately visualized by structure matching techniques, despite the size and complexity of these molecular aggregates.
Discussion {#sec3}
==========
The examples discussed in the previous sections cover a wide range of protein architectures, and they include some of the largest protein complexes known today. One obvious goal in choosing these examples is to provide an overview of the broad range of structure-matching problems that can be successfully solved with appropriate computational tools.
It is clear however, that, although structure-matching tools provide the means for discovery, the important part is the analysis and comprehension of the detected correlations (e.g., [@bib41; @bib2]). In the case of the dehydrogenase structures ([Figure 3](#fig3){ref-type="fig"}), the connection between the [2i9p](pdb:2i9p) and [3fwn](pdb:3fwn) oligomers can be traced to a duplication event on the gene level. However, the relationship of these structures to [3i83](pdb:3i83) is obscure. To understand the evolution of these enzymes, it is necessary to study the family as a whole. At present, this is a rather demanding exercise (e.g., Sippl. 2009), where the outcome largely depends on the available structures. But in any case, given the relationships among these molecules ([Figure 3](#fig3){ref-type="fig"}), it is clear that the comparative analysis of protein structures requires that the complete active molecules are taken into account. Important correlations are often not detectable on the level of individual chains.
The number of known structures will soon exceed 100,000, and beyond that, the determination of structures is likely to continue with ever increasing speed. The structures represent an enormous body of information that is exponentiated by pairwise relationships among them. A necessary step in digesting and organizing this information is the identification of mutual similarities among the structures. Hence, structure-matching tools should be routinely accessible to a broad community of structural and molecular biologists. This in turn requires that the tools are efficient, reliable, and easy to use. These goals have been in focus throughout the development of TopMatch and the associated web-based service (see [Experimental Procedures](#sec4){ref-type="sec"}). The computations leading to the structural matches discussed here, as well as any other structure-matching experiment, can be executed and visualized using this service. The response is immediate, except for very large structures, where the response time is in the order of seconds.
Experimental Procedures {#sec4}
=======================
Superposition, Similarity, and Deviation {#sec4.1}
----------------------------------------
The optimal superposition of two structures joined by an alignment, as defined in the main text, is obtained by minimizing the root-mean-square error of two coordinate sets **x***~i~* of the query, *Q*, and **y***~i~* of the target, *T*, where $i = 1,\ldots,L$, and *L* is the length of the alignment. The following recipe computes this transformation most efficiently ([@bib3; @bib36]). We choose to keep the query *Q* fixed in space. Then the transformation that needs to be applied to the target coordinates, **y***~i~*, for the optimal fit with the coordinates of **x***~i~* of *Q* is$$\mathbf{z}_{i} = \mathbf{R}\ \left( {\mathbf{y}_{i} - \mathbf{y}} \right) + \mathbf{x},$$where $\mathbf{x} = L^{- 1}\ \sum\mathbf{x}_{i}$ and $\mathbf{y} = L^{- 1}\ \sum\mathbf{y}_{i}$ are the centroids of *Q* and *T*, respectively, and R is a rotation matrix. The latter is obtained from the singular value decomposition of the matrix **T**,$$\mathbf{V}\mathbf{S}\mathbf{W}^{T} = \mathbf{T} = \sum\limits_{i}^{L}{\left( {\mathbf{x}_{i} - \mathbf{x}} \right)\ \left( {\mathbf{y}_{i} - \mathbf{y}} \right)^{T}},$$i.e., **T** is a sum over the outer products of vectors. Then **R** = **WV***^T^* is the desired rotation provided the determinant of **R** is +1. If the determinant is -1---a rather frequent result, particularly for short alignments---then **R** involves a reflection. In this case, to obtain a proper rotation, it is necessary to multiply the column of **V** associated with the smallest singular value, i.e., with the smallest element of the diagonal matrix **S**, by -1. This subtlety is easily missed, since reversing the sign of any column of **V** yields a proper rotation, but the result may be suboptimal. Then the distances *r~i~* between equivalent pairs of C*^α^* atoms are obtained from **r***~i~* = **x***~i~* - **z***~i~*, from which the root-mean-square error, *E~r~*, the similarity, *S*, and the associated average deviation of distances, *S~r~*, are computed as defined above.
Visualization {#sec4.2}
-------------
In the application of structure-matching techniques, the appropriate and efficient visualization of aligned structures in 3-D, as well as their projection in 2-D, is a most critical step. Given a specific alignment, the TopMatch implementation generates a file in PDB format ([@bib5]) containing the complete query and the transformed target structure, which are then immediately channeled into the molecular graphics programs PyMOl ([@bib10]) or Rasmol ([@bib31]), or, using the web-service (see below), into Jmol ([@bib17]). The standardized coloring scheme for pairwise alignments is blue for the query, *Q*, and green for the target, *T*. The aligned regions, i.e., the ungapped blocks of an alignment, are shown in orange (*Q*) and red (*T*). For the simultaneous display of more than two structures, the color scheme may be extended (e.g., [Figure 3](#fig3){ref-type="fig"}). Highlighting the blocks of an alignment in this way puts the focus on the aligned regions. The figures shown here are all prepared using the PyMOl program.
Program Access {#sec4.3}
--------------
The Center of Applied Molecular Engineering (CAME) has maintained a web service for structure comparison for several years. The remarkable implementation described here (TopMatch 7.3) replaces previous versions and can be accessed via the previous link (<http://topmatch.services.came.sbg.ac.at>). The web service provides instructions for program usage, including visualization of aligned structures in 3-D (Jmol) and facilities to upload structures. However, since all structures in PDB are accessible through the server it is generally not necessary to upload any files. A stand-alone version of the TopMatch program can be downloaded from the CAME web site (<http://www.came.sbg.ac.at>).
We are most grateful to Markus Gruber who repeatedly checked the performance and stability of TopMatch during program development. Alwyn Jones and Lars Liljas clarified several questions regarding the virus capsid structures. M.S. is most thankful for their instant reply and the information they provided. This work was supported by FWF Austria, grant number P21294-B12.
, resolution 3.1 Å) and the subvirion particle of aquareovirus (grass carp reovirus, [3iyl](pdb:3iyl), resolution 3.3 Å) contain a number of local structural correlations that can be combined to cover a large fraction of both molecules.\
(A) Basic alignment of residues 16A--356A of [3izx](pdb:3izx) and residues 24W--377W of [3iyl](pdb:3iyl).\
(B) Basic alignment of residues 554A--696A of [3izx](pdb:3izx) and residues 512W--661W of [3iyl](pdb:3iyl).\
(C) Composite alignment assembled from the complete set of basic alignments obtained for [3izx](pdb:3izx) chain A and [3iyl](pdb:3iyl) chain W.](gr1){#fig1}
, chains A and B, blue) and the Fab fragment of a humanized monoclonal antibody ([3qcv](pdb:3qcv), chains H and L, green) have multiple correlations. The structural matches are colored orange ([1hxm](pdb:1hxm)) and red ([3qcv](pdb:3qcv)).\
(A) Structural match between the N-terminal domains A1 and B1 of [1hxm](pdb:1hxm) and the N-terminal domains H1 and L1 of [3qcv](pdb:3qcv) (A1 matched with H1 and B1 with L1, respectively).\
(B) Structural match between the C-terminal domains A2 and B2 of 1hxm and the C-terminal domains H2 and L2 of [3qcv](pdb:3qcv) (A2 matched with H2 and B2 with L2, respectively).](gr2){#fig2}
, chains A, B, C, and D, blue).\
(B) Homo-dimer of 6-phosphogluconate dehydrogenase from E.coli ([3fwn](pdb:3fwn), chains A and B, green).\
(C) Homo-dimer of 2-dehydropantoate 2-reductase from the proteobacterium Methylococcus capsulatus ([3i83](pdb:3i83), chains A and B, yellow).\
(AB) Superposition of (A) and (B), with aligned regions in orange ([2i9p](pdb:2i9p)) and red ([3fwn](pdb:3fwn)).\
(AC) Superposition of (A) and (C), with aligned regions in orange ([2i9p](pdb:2i9p)) and red ([3i38](pdb:3i38)).\
(BC) Superposition of (B) and (C), with aligned regions in orange ([3fwn](pdb:3fwn)) and red ([3i38](pdb:3i38)).](gr3){#fig3}
, blue).\
(B) Hexameric protein secretion apparatus from *Pseudomonas aeruginosa* ([1y12](pdb:1y12), green).\
(C) Superposition of [1k28](pdb:1k28) and [1y12](pdb:1y12). The matching parts are in orange ([1k28](pdb:1k28)) and red ([1y12](pdb:1y12)). See also [Figure 2](#fig2){ref-type="fig"} in [@bib26].](gr4){#fig4}
, blue) and Bordetella pertussis virulence factor pertactin (1dab, green) superimposed with the matching *β* helix parts in orange and red, respectively.\
(B) The monomeric chain of 1 dab (green).\
(C) Single chain of the [1k28](pdb:1k28) trimer (blue).\
(D) The three chains of the [1k28](pdb:1k28) trimer colored blue, red, and yellow, respectively.\
(E) Cross-section of the [1k28](pdb:1k28) trimer.\
(F) Cross-section of the superimposed [1k28](pdb:1k28) trimer (blue) and 1 dab monomer (green), with the matching parts in orange and red, respectively.\
(G) Cross-section of the 1 dab monomer (green).](gr5){#fig5}
, blue) and the S. cerevisiae DNA polymerase processivity factor PCNA ([1plq](pdb:1plq), green), with the matching parts in orange and red, respectively.\
(B) Domain composition of DNA clamps in bacteria and eukaryotes. The bacterial monomer ([2pol](pdb:2pol), chain A, blue) and the eukaryotic monomer ([1plq](pdb:1plq), chain A, green). The individual domains are labeled from N to C terminus.\
(C) Superposition of domains D1 (residues A1--A127, blue and orange) and D2 (residues A128--A258, green and red) of [1plq](pdb:1plq).\
(D) Superposition of domain D1 of [1plq](pdb:1plq) (blue and orange) and the MRP1 chain of the guide-RNA binding complex [2gia](pdb:2gia) (residues A72--A217, green and red).\
(E) Topology of [1plq](pdb:1plq)-D1 with the helices and strands numbered from N to C terminus. The spectrum of colors progresses from blue (N terminus) to red (C terminus).\
(F) Topology of [2gia](pdb:2gia) with number and color schemes as in (E).](gr6){#fig6}
, chain A, blue) and horse apoferritin ([2za6](pdb:2za6), chain A, green), with the matching parts in orange and red, respectively.\
(B) Superposition of the complete ferritin cages of [3bkn](pdb:3bkn) and [2za6](pdb:2za6). The color scheme is the same as in (A).](gr7){#fig7}
) and Sesbania Mosaic Virus ([1vb4](pdb:1vb4)) Have Structural Correlations on the Monomer and Pentamer Level\
(A) Superposition of the monomers of [2buk](pdb:2buk) (chain A, blue) and [1vb4](pdb:1vb4) (chain A, green), with the matching parts in orange and red, respectively.\
(B) Superposition of [2buk](pdb:2buk) and [1vb4](pdb:1vb4) pentamers (coloring as in (A)). The superimposed pentamers are shown as cartoons and the remaining 55 chains as ribbons.\
(C) The virus capsid of [2buk](pdb:2buk) as in (B), with an additional adjacent pentamer shown as cartoons.\
(D) The virus capsid of [1vb4](pdb:1vb4) as in (B) with an additional adjacent pentamer shown as a cartoon.](gr8){#fig8}
######
Parameters of Structure Alignments
Figure Query Target *T* *L* *Q~C~* *T~C~* *S* *S~r~* *E~r~* *I~s~* *P*
------------------------------ --------------------------- ----------------------------- ----- ------ -------- -------- ------ -------- -------- -------- -----
[1](#fig1){ref-type="fig"}A [3izx](pdb:3izx),A [3iyl](pdb:3iyl),W b 216 17 20 180 2.97 3.09 14 0
[1](#fig1){ref-type="fig"}B [3izx](pdb:3izx),A [3iyl](pdb:3iyl),W b 139 11 13 118 2.84 2.94 14 0
[1](#fig1){ref-type="fig"}C [3izx](pdb:3izx),A [3iyl](pdb:3iyl),W c 634 49 60 380 5.01 5.59 12 0
[2](#fig2){ref-type="fig"}A [1hxm](pdb:1hxm),AB [3qcv](pdb:3qcv),HL b 223 51 51 203 2.12 2.21 25 0
[2](#fig2){ref-type="fig"}B [1hxm](pdb:1hxm),AB [3qcv](pdb:3qcv),HL b 179 41 41 158 2.47 2.56 25 0
[3](#fig3){ref-type="fig"}AB [2i9p](pdb:2i9p),ABCD [3fwn](pdb:3fwn),AB c 641 55 69 497 3.53 3.83 19 1
[3](#fig3){ref-type="fig"}AC [2i9p](pdb:2i9p),ABCD [3i83](pdb:3i83),AB c 406 35 68 294 3.97 4.28 12 1
[3](#fig3){ref-type="fig"}BC [3fwn](pdb:3fwn),AB [3i83](pdb:3i83),AB c 260 28 44 178 4.31 4.87 13 2
[4](#fig4){ref-type="fig"}C [1k28](pdb:1k28)\@1 [1y12](pdb:1y12)\@1 b 339 12 35 289 2.79 2.92 8 0
[5](#fig5){ref-type="fig"}A [1k28](pdb:1k28)\@1 1dab,A c 321 12 60 243 3.69 3.95 10 0
[6](#fig6){ref-type="fig"}A [2pol](pdb:2pol)\@1 [1plq](pdb:1plq)\@1 c 645 83 88 549 2.81 2.94 10 0
[6](#fig6){ref-type="fig"}C [1plq](pdb:1plq),A(1:127) [1plq](pdb:1plq),A(128:258) b 114 87 90 103 2.21 2.30 14 0
[6](#fig6){ref-type="fig"}D [1plq](pdb:1plq),A(1:127) [2gia](pdb:2gia),A(72:217) b 97 72 76 85 2.60 2.70 11 5
[7](#fig7){ref-type="fig"}A [3bkn](pdb:3bkn),A [2za6](pdb:2za6),A b 152 89 94 136 2.33 2.42 16 0
[7](#fig7){ref-type="fig"}B [3bkn](pdb:3bkn)\@1 [2za6](pdb:2za6)\@1 c 3649 89 94 3202 2.53 2.61 17 0
[8](#fig8){ref-type="fig"}A [2buk](pdb:2buk),A [1vb4](pdb:1vb4),A c 115 59 63 94 3.13 3.27 4 0
[8](#fig8){ref-type="fig"}B [2buk](pdb:2buk)\@1,Aaxyz [1vb4](pdb:1vb4)\@1,Aaxyz c 354 36 38 252 4.09 4.37 7 0
Figure, the respective figure and panel; Query, name of the query structure consisting of the PDB code and chain identifier(s) with optional residue numbers in parentheses (from N to C terminus), where the @ sign and the associated number refer to the biological unit as defined in the respective PDB file; Target, name of the target structure (same encoding rules as for the query); *T*, alignment type (basic, b, or composite, c); *L*, alignment length; *Q~c~*, query cover; *T~c~*, target cover; *S*, similarity; *Sr*, average distance error; *E~r~*, root-mean-square error; *I~s~*, fraction of pairs of identical residues; *P*, number of permutations. The numeric parameters are defined in the main text.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .290}
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Numerous NGS pipelines and tools have been developed in recent years that are valuable to users in the field, but also create confusion in selecting the desired tool. Some of the commercial NGS pipelines are CLC genomics workbench, DNASTAR, CASAVA, Genious, Genomatix Solutions, GenoMiner, Partek Genomics Suite and so on. Most of the commercial NGS pipeline tools are targeted to biologists as end-users highlighting easy and user friendly interface. Often, these commercial tools become difficult to customize for speed when processing large number of samples. Alternatively, commercial vendors provide the facility to process and ship the complete variants sets along with the sequencing of samples. Non-commercial open source NGS pipelines such as GATK \[[@CR1], [@CR2]\], SAMtools \[[@CR3]\], SOAP \[[@CR4], [@CR5]\], SNPAAMapper \[[@CR6]\], WEP \[[@CR7]\], Atlas2 \[[@CR8]\] are also being used extensively in academia and many organizations. These open source NGS pipelines are highly customizable but require expertise to set up optimally. Many studies have been done to evaluate NGS data analysis pipelines and tools. Bao S. et al. \[[@CR9]\] have evaluated various mapping and assembly software. Pabinger et al. \[[@CR10]\] have surveyed around 205 variants of NGS tools at different analytical steps like quality assessment, alignment, variant identification, variant annotation and visualization. Nielsen et al. \[[@CR11]\] have evaluated various SNP and genotype calling algorithms. Although these studies have helped tremendously in determining which tools and pipelines to use, they do not answer the concrete question of whether to use data provided from a commercial vendor or to put in extra efforts to run additional well-known open source pipelines. Also, in situations where we fail to identify a causative variant in the data set provided by commercial vendors, we may doubt the pipeline's ability to find the variants. Thus, it becomes important to compare the variant sets provided by commercial vendors with variants obtained through one of the well-reputed tools. Several studies have confirmed the GATK pipeline's excellent performance in detecting variants. The GATK pipeline is being used in large projects, such as the 1000 Genomes Project and The Cancer Genome Atlas \[[@CR1], [@CR12]\]. However, smaller labs and institutes often rely fully on commercial vendors for complete sequencing and analysis services. Illumina Inc. is a leader in providing NGS services. Illumina uses the CASAVA and ISSAC pipelines for variant detection. Illumina has reported comparison among ISAAC, CASAVA and GATK pipelines; mostly for the speed of completing the pipeline \[[@CR13]\]. However, an independent detailed comparison between the Illumina and GATK pipeline using multi-sample calling algorithm in larger cohorts is necessary. Here we compared variant sets supplied by the Illumina CASAVA pipeline and the well-known GATK pipelines in great detail on concrete study cases and discuss the differences from a user's perspective. In general, genotype calling errors by the variant callers are associated with Mendelian violation when the caller is unaware of family structure \[[@CR14]\]. In this study, both GATK and CASAVA are unaware of family pedigree and therefore Mendelian inheritance is examined in familial samples for the genotypes of discordant variants by the pipelines. As an additional independent quality control we use genotyping array data from the Illumina OMNI 2.5 array. We present an evaluation of the CASAVA and the GATK pipelines for three different data sets: 108 unrelated Qatari genomes, 19 trios from studies on obesity and diabetes, and 2 larger families with suspected rare genetic diseases.
Methods {#Sec2}
=======
CASSAVA SNP calling {#Sec3}
-------------------
Illumina SNP calls were based on the CASAVA -1.9.0a1_110909 pipeline. SNPs and the genotype from the CASAVA pipeline were called for each sample individually. We created a pass quality subset of these variants by keeping the variant for which Filter column in VCF file has value \"PASS\" and removing all other variants. Thus, the first set without any quality filter will be called CASAVA ALL and the quality filtered set will be called CASAVA PASS in this paper. In many cases, we have compared the pipelines for a group of samples. In these cases, we merged these SNPs from the CASAVA pipeline using vcftools \[[@CR15]\]. Similarly, we created merged VCF for quality filtered (PASS quality) from the CASAVA pipeline by merging all the PASS quality SNPs based on quality column annotation (Genotype quality \>20) in all single sample VCF files.
GATK best practice pipeline {#Sec4}
---------------------------
In our in-house pipeline, we used Bowtie2 \[[@CR16]\] to align the sequencing reads against the human reference genome build 37. We also used other necessary tools like SAMtools \[[@CR3]\], Novosort and Picard \[[@CR17]\] to process and format alignment files before processing them with GATK. We implemented the best practices of GATK pipeline to call SNPs and Indels. We have used GATK 2.4 version and GATK-UnifiedGenotyper as SNP caller in this study. We have used multi-sample variant calling by GATK-UnifiedGenotyper. The reason of using multi-sample calling is to distinguish non-variant genotypes between homozygous reference genotype and missing genotype in cohort analysis. With single sample calling genotype called only for variants we can't be sure if the non-variants have missing genotype or same as reference. Also, big projects like 1000 genomes have preferred multi-sample calling over single sample calling \[[@CR18]\]. We used GATK-UnifiedGenotyper instead of GATK-HaplotypeCaller, a similar or better variant caller by GATK, in this study because of similar accuracy in calling SNPs and computational feasibility to run for large number of samples. For more than 100 samples, according to GATK website, GATK-UnifiedGenotyper is advised over GATK-HaplotypeCaller. The real advantage of Haplotypecaller over UnifiedGenotyper is in calling Indels but in this paper we are focusing on SNPs only. Next, similar to the CASAVA pipeline, we created two variant sets, GATK ALL (without any quality filter) and GATK PASS (by keeping the variant for which Filter column in VCF file has value \"PASS\" and removing all other variants) from our in-house GATK pipeline. The variants found by GATK pipeline were recalibrated using GATK walker VariantScoreRecalibrater. The input true sites in creating the model were SNPs from dbSNP Human build 132 \[[@CR19]\], genotyping OMNI array calls of 1000 genomes project and Hapmap SNP calls for estimating the probability that SNPs are true genetic variants rather than a sequencing or data processing artifact. The call sets were partitioned into quality trenches and are shown in the plot below. We took the variants until we found 99% of known variants (truth sensitivity) in the GATK PASS variant set.
Genotyping Omni array {#Sec5}
---------------------
Human genotyping array data is from Illumina HumanOmni2.5-8 platform. This array has about 2.37 million tag SNPs from 1000 genomes pilot project with MAF ≥2.5%. Illumina Inc. supplied genotypes for all the samples from HumanOmni2.5-8 by performing Illumina Infinium LCG assay and thereupon calling the genotypes using their propriety software called GenomeStudio. They provide genotype for each of these probes with GenCall scores. Illumina recommends a GenCall score cut-off of 0.15 for their infinium assay based products \[[@CR20]\]. This recommended GenCall score cut-off of 0.15 was used to test the concordance with the GATK and CASAVA pipelines.
Evaluation data sets {#Sec6}
--------------------
108 unrelated individuals from QatarGenotypes from HumanOmni2.5-8 array and Whole Genome Sequencing (WGS) data analysis from CASAVA and GATK pipeline are compared for these 108 unrelated individuals from Qatar. Whole Genome Sequencing was done by Illumina Hiseq 2500 sequencer with the average coverage of 37.99 (see Additional file [1](#MOESM1){ref-type="media"}). Phenotypes of these individuals are either diabetic or normal individuals. Illumina CASAVA pipeline called variants using a single sample (genome) at a time while we used GATK multi-sample calling.19 trios (Father, Mother, Offspring)These 19 trios are selected from another cohort of 64 individuals with 11 different families. Whole Genome Sequencing was done by Illumina Hiseq 2500 sequencer with the average coverage of 39.50 (Additional file [1](#MOESM1){ref-type="media"}). Variants from the GATK pipeline were called for all the 64 individuals together by multi-sample calling using the UnfiedGenoytper and variant sets for trios were filtered using SelectVariants walker. The pedigree structures for families from which trios are selected are shown in Figure [1](#Fig1){ref-type="fig"}.Two clinical case studies of monogenic disordersIn the first family (Figure [2](#Fig2){ref-type="fig"}A), both affected children are present with hypoplasia of cerebellum as a disease phenotype. The other three members (Father, mother and aunt) are unaffected. The second family (Figure [2](#Fig2){ref-type="fig"}B) comprises two affected children with abnormal pain sensation and two unaffected children. Whole Genome Sequencing was done by Illumina Hiseq 2500 sequencer with the average coverage of 43.20 for first family and 42.95 for second family (Additional file [1](#MOESM1){ref-type="media"}). In the pipeline comparison result for these two families, GATK calls were made by multi-sample calling of all members of the family whereas CASAVA calls were from single sample calling.
Figure 1**19 trios in family pedigree.** Rectangular boxes drawn in the family pedigree indicate the trios taken for pipelines comparison analysis. Individual in black are obese. Star marked individuals were not sequenced. Individuals greyed in the pedigrees had unknown phenotype. Individual with no color are non-obese individuals.Figure 2**Pedigrees of the families with children affected with monogenic homozygous recessive disease.** Star marked Individuals were not sequenced. Individuals in grey in the pedigrees had unknown phenotype. Individuals with no color are unaffected. Individuals in black color are affected with hypoplasia of cerebellum in **A**, and with abnormal pain sensation in **B**.
For all three evaluation data sets, although Illumina supplied annotated VCF files, we annotated both Illumina and GATK VCF files using SnpEff \[[@CR21]\] and AnnTools \[[@CR22]\] to provide a uniform annotation for comparison between pipelines.
Results {#Sec7}
=======
The summarized comparison results between the CASAVA and GATK pipeline are presented in Table [1](#Tab1){ref-type="table"}. Both CASAVA and GATK have very high similarity to OmniArray genotypes. However, while comparing all variants from NGS, GATK identifies a higher number of variants than CASAVA. The robustness of these additional variants are analyzed and discussed below in the results presented for comparison between the pipelines for various data sets.Table 1**Summary of GATK and CASAVA comparison**Data setsCASAVAGATKCommonVariant countTsTv ratioVariant countTsTv ratioVariant countTsTv ratio**Omni Array Genotyping Data**108 unrelated (per sample)AllSNP by pipelines708,089 ± 4,5163.56 ± 0.003715,033 ± 4,5513.53 ± 0.003706,378 ± 45023.57 ± 0.003GT Matched SNP705,749 ± 44683.58 ± 0.003703,608 ± 44793.59 ± 0.003698,910 ± 44373.60 ± 0.003Ref or missing by pipeline1661347 ± 45381654613 ± 45731649566 ± 4567False positive988 ± 201.46 ± 0.0206320 ± 1121.61 ± 0.039489 ± 131.92 ± 0.031False negative23283 ± 9022076 ± 5820550 ± 59PASSSNP by pipelines707,128 ± 45063.57 ± 0.003696,960 ± 44803.60 ± 0.003695,589 ± 44563.60 ± 0.003GT Matched SNP705,106 ± 44593.58 ± 0.003693,135 ± 44343.63 ± 0.003689,330 ± 43943.63 ± 0.003Ref or missing by pipeline1662325 ± 45281672894 ± 45011657868 ± 4533False positive810 ± 171.33 ± 0.019277 ± 202.23 ± 0.054229 ± 102.12 ± 0.054False negative24067 ± 13433985 ± 9622735 ± 65**NGS Data Set**108 unrelated (per sample)All4,025,625 ± 44,1022.02 ± 0.0014,331,336 ± 45,8961.86 ± 0.0023,792,293 ± 43,1222.07 ± 0.000PASS3,894,810 ± 43,3882.04 ± 0.0013,438,203 ± 412052.13 ± 0.0013,401,091 ± 403172.15 ± 0.00119 trios ( per trio)ALL5,235,184 ± 47,7902.01 ± 0.0017,003,439 ± 56,4881.88 ± 0.0034,945,042 ± 46,6502.06 ± 0.002PASS4,786,871 ± 47,5642.07 ± 0.0025,125,002 ± 48,7172.13 ± 0.0014,320,414 ± 43,0622.14 ± 0.001Family 1ALL6,082,6242.006,337,1081.885,635,1832.04PASS5,438,3932.075,004,0482.124,898,1262.13Family 2ALL5,192,8911.995,459,7251.844,752,1932.03PASS4,526,2912.074,205,9952.124,104,3432.13Comparison between the pipelines have been done for unfiltered sets (CASAVA ALL, GATK ALL), and for quality filtered sets (CASAVA PASS, GATK PASS).
Comparison of NGS pipelines with genotyping array {#Sec8}
-------------------------------------------------
The Illumina Omni 2.5 platform can detect genotypes at 2.37 million SNP loci in the human genome. In every single individual about 30% of these 2.37 million SNPs were present either in a heterozygous or a homozygous for the non-reference variant state. Illumina only reports genotypes for such variants in the VCF files. Reference allele homozygous calls are not differentiated from non-call. We therefore compare the pipeline only on SNPs that are reported in the VCF files. Both pipelines have very high concordance (\~99%) with genotyping array data (Table [1](#Tab1){ref-type="table"}). GATK pipeline has a higher number of non-reference SNPs compared to CASAVA, but CASAVA has slightly higher genotyping matches (99.67%) compared to GATK (98.33%). For quality passed variants (CASAVA PASS, GATK PASS) both pipelines have approximately the same concordance with Illumina Genotyping OmniArray data (Table [1](#Tab1){ref-type="table"} and Additional file [2](#MOESM2){ref-type="media"}). False positives and false negatives in Table [1](#Tab1){ref-type="table"} are calculated assuming Illumina OMNI 2.5 genotype data to be correct. GATK has lot more false positive compared to CASAVA before PASS filter and the opposite after PASS filter. To our surprise, TsTv ratios of these false positives are not very far from ideal TsTv ratio of 2.0-2.1. Furthermore, TsTv ratio of false positive by GATK is better, closer to 2, than the TsTv ratio of false positive by CASAVA in both before and PASS filter. Moreover, the TsTv ratio of common false positive is near to 2 suggesting these small numbers of common false positive by both pipelines could be false negative in OMNI 2.5 genotype array data.
Pipeline comparison in unrelated individuals {#Sec9}
--------------------------------------------
Venn diagram in Figure [3](#Fig3){ref-type="fig"} shows the comparison between CASAVA and GATK pipeline for the combined variants of all 108 unrelated individuals. For the unfiltered variants set in Figure [3](#Fig3){ref-type="fig"}A, GATK ALL and CASAVA ALL have an approximately equal number of SNPs (24.01 million for GATK and 23.99 million for CASAVA) and an equal number of unique SNPs (2.4 million for GATK and 2.39 million in CASAVA). However, if we look at the individual sample from GATK and CASAVA in Figure [4](#Fig4){ref-type="fig"}E, we find GATK has many more SNP calls than CASAVA (4.33 million by GATK and 4.02 million by CASAVA). This discrepancy, similar number of variants by pipelines at population level but different at sample level, can be explained by exploring shared and unique variants across the samples (Figure [5](#Fig5){ref-type="fig"}). The number of shared variants among 108 individuals identified by GATK is lot more than in CASAVA for both with and without PASS filter (Figure [5](#Fig5){ref-type="fig"}a, Figure [5](#Fig5){ref-type="fig"}b). The distribution of number of unique variants among 108 individuals identified by GATK and CASAVA are overlapping in great extant and thus are very similar (Figure [5](#Fig5){ref-type="fig"}c and Figure [5](#Fig5){ref-type="fig"}d). Also, we can explain the discrepancy by pipelines at population and sample level by looking at the pipeline specific calls (GATK ONLY and CASAVA ONLY calls). Theoretically, CASAVA ONLY calls should be very different across the 108 samples and GATK only calls should be similar across 108 samples to justify the observed discrepancy. When we checked the GATK ONLY 2.4 million SNPs of combined variants set (Figure [3](#Fig3){ref-type="fig"}A), we found that around 56.6% (1.29 million) were present in more than 5 out of 108 samples. In contrast, in CASAVA ONLY 2.39 million combined variants (Figure [3](#Fig3){ref-type="fig"}A), only 18.8% (0.45 million) were present in more than 5 out of 108 samples. The higher percentage of consensus call across the sample in GATK ONLY SNPs compared to CASAVA ONLY SNPs indicates the effects of multi-sampling calling using the GATK pipeline. We hypothesize that this effect is desired since the samples are from the same population. In other words, in order to have confidence in the SNPs that are non-agreeing across the pipelines, the variant calls should have agreement across the samples, provided that the samples originate from same population. However, the variants identified by only one pipeline (GATK ONLY SNPs or CASAVA ONLY SNPs) have lower TsTv ratio compared to variants that are common to both pipeline (Figure [4](#Fig4){ref-type="fig"}A and Figure [4](#Fig4){ref-type="fig"}B). TsTv ratio for GATK ONLY SNPs before pass filter in Figure [4](#Fig4){ref-type="fig"}A is very low (1.096 ± 0.003). Similarly, TsTv ratio of CASVA ONL SNPs in Figure [4](#Fig4){ref-type="fig"}B is low (1.485 ± 0.001). The lower TsTv ratio of pipeline specific variants indicates the presence of false positives. Furthermore, Het/Hom ratio of GATK ONLY subset after GATK PASS filter is very high, as shown in Figure [4](#Fig4){ref-type="fig"}D, indicates that GATK calls more false positive heterozygous calls than homozygous false positive calls. In general, the explanation of lower TsTv for both before and after PASS filter should be similar. The more number of pipeline specific variants has more false positives. In addition to the pipeline specific variant count, the lower quality variants could be reason of of very low Tstv ratio for GATK ONLY in Figure [4](#Fig4){ref-type="fig"}A compared to TsTv ratio of GATK ONLY subset in Figure [4](#Fig4){ref-type="fig"}B. However, before pass filter the number of combined set of variants for GATK ONLY (2.4 million) is similar to CASAVA ONLY (2.39 million) and, therefore, should not have drastically different TsTv in data set. Moreover, GATK ONLY subset has more number of shared variants among 108 samples compared to CASAVA ONLY and intuitively we would be expecting better TsTv for GATK ONLY compared to CASAVA ONLY. The opposite behavior of TsTv can thus be attributed to GATK multi-sample calling which might be placing doubtful SNP in samples at particular locus if it one or more samples have confirmed SNP at that locus. This suggests that multi-sample calling has the advantage of calling more variants but at the cost of more false positives. The other possible explanation of lower TsTv ratio of pipeline specific variants could be non-universal nature of TsTv ratio \[[@CR23]\]. However, we tested this by random sampling the 2.4 million variants 10 times and computed TsTv ratio. We found TsTv ratio of these randomly sampled variant to be 2.051 ± 0.001. This excludes non-universal nature of TsTv ratio as possible explanation. Thus, lower TsTv for pipeline specific (GATK ONLY and CASAVA ONLY) subset is indication of false positives. The non-agreeing SNPs between the pipelines can also be analyzed in a family structure to see the Mendelian violation, which we did by looking at 19 trios (Father, Mother, and Offspring) and 2 families having homozygous recessive diseases.Figure 3**Venn diagrams comparing combined variants of All108 Qatari genomes by GATK and CASAVA pipelines. A)** Between GATK ALL and CASAVA ALL variants sets showing intersection and unique variants sets. **B)** Between GATK PASS and CASAVA PASS variants sets showing intersection and unique variants sets.Figure 4**Individual genome comparison between GATK and CASAVA pipeline in 108 unrelated Qatari individuals. A**, **C,** and **E** show comparison for unfiltered sets (CASAVA ALL, GATK ALL). **B**, **D**, and **F** show comparison for quality filtered sets (CASAVA PASS, GATK PASS). **A** and **B** show boxplots of 108 transition-transversion (TsTv) ratios for pipeline's variants sets (CASAVA, GATK) along with pipeline specific (CASAVA ONLY, GATK ONLY). C and D show boxplots of 108 het-hom ratios for pipeline's variants sets (CASAVA, GATK) along with pipeline specific (CASAVA ONLY, GATK ONLY). E and F show barplot of average variant counts for 108 individuals with error at the top of each bar.Figure 5**Shared and unique variants by GATK and CASAVA in 108 unrelated Qatari individuals. A** shows density plot of unique variants of 108 unrelated Qatari individuals in GATK ALL, GATK PASS, CASAVA ALL, and CASAVA PASS variant sets. **B** shows density plot of shared variants of 108 unrelated Qatari individuals in GATK ALL, GATK PASS, CASAVA ALL, and CASAVA PASS variant sets.
Pipeline differences after PASS filter at per sample level (Figure [4](#Fig4){ref-type="fig"}F) is apposite to before PASS filter (Figure [4](#Fig4){ref-type="fig"}E) i.e. the number of SNPs per sample in GATK call set is lower than in CASAVA. However, at population level GATK called more SNPs in both before and after PASS filter (Figure [3](#Fig3){ref-type="fig"}A and Figure [3](#Fig3){ref-type="fig"}B). It is important to see how PASS filter changed the allele frequency distribution in GATK and CASAVA. Minor Allele Frequency (MAF) distribution plot shown in Additional file [3](#MOESM3){ref-type="media"} and variants frequency distribution shown in Additional file [4](#MOESM4){ref-type="media"} to see the effect of PASS filter for both GATK and CASAVA. In Additional file [3](#MOESM3){ref-type="media"}, we can see that PASS filter removes low frequency with high MAF and, therefore, we see higher frequency for low MAF. In Additional file [4](#MOESM4){ref-type="media"}, we can see the distributions of GATK before and PASS filtering is far apart while the distribution of CASAVA before and PASS filtering has some overlap. This shows that there are many low quality variants from each of the 108 unrelated samples identified by GATK. This also explains the reason of higher false positives and lower TsTv ratio for of GATK compared to CASAVA before PASS filter.
Pipelines comparison in trios {#Sec10}
-----------------------------
The CASAVA and GATK pipelines were compared for 19 trios from the Qatari population by taking combined variants sets of each trio separately (Figure [6](#Fig6){ref-type="fig"} and Additional file [5](#MOESM5){ref-type="media"}). On average GATK ALL have 7 million variants in any trio compared to 5.25 million variants in CASAVA ALL (Figure [6](#Fig6){ref-type="fig"}). The large difference between the GATK ALL and CASAVA ALL variant sets in any trio can be attributed to GATK multi-sample calling, but this gives rise to the question about the qualities of these extra variants. Both pipelines have approximately equal percentage of variants having Mendelian violation (3.40% for CASAVA ALL and 3.47% for GATK ALL (Figure [6](#Fig6){ref-type="fig"}C). Assuming Mendelian violation as a criterion to judge confidence in variants, CASAVA pipeline missed those extra 1.75 million variants present in GATK ALL, which were comparable in quality. However, the lower TsTv ratio of 1.01 for Mendelian violated GATK ALL variants compared to TsTv ratio of 1.47 for Mendelian violated CASAVA ALL variants (Figure [6](#Fig6){ref-type="fig"}A) creates doubt about these extra 1.75 million variants of GATK ALL. The difference between number of variants, Mendelian violation, and TsTv ratio in the GATK and CASAVA pipelines is diminished for quality-filtered sets (CASAVA PASS and GATK PASS). It's important to estimate false positive rate for the decision of applying the PASS filter or not. Since we do not have genotyping array data for trios, we are confined to assess the pipelines performance based on Mendelian violation and TsTv ratio. We can assume the variant set confirmed by both pipelines to be robust to provide us the rough estimate of the Fraction of Mendelian violation (F~MV~) in each trio by the pipeline. Using this fraction we thus computed the Expected number of Mendelian violation in variants subset (E~MV~) which didn't pass the PASS filter (NOT PASS) in the pipelines. We then found the actual number of variants with Mendelian violation (O~MV~) in all the variants in NOT PASS subset (T~NP~). We then calculated the false positive fraction in the NOT PASS subset by (O~MV~ -- E~MV~)/T~NP~. This way we found mean false positive percent of 11.15% for GATK NOT PASS subset and 22.90% for CASAVA NOT PASS subset in 19 trios. The detailed numbers are shown in Additional file [6](#MOESM6){ref-type="media"}. The false positive percent in CASAVA NOT PASS is higher than GATK NOT PASS in trios that is in contrast to false positive percent by pipelines in 108 unrelated individuals when computed using Genotyping OMNI array data. It suggests that GATK multi-sample calling algorithm works better in related individuals compared to unrelated individuals. In order to check the real difference at a functional level, we evaluated the pipeline performance in real case of finding the causative homozygous recessive variant in two diseased families.Figure 6**GATK and CASAVA comparison in 19 trios. A** and **C** show comparison for unfiltered sets (CASAVA ALL, GATK ALL). **B** and **D** show comparison for quality filtered sets (CASAVA PASS, GATK PASS). A and B shows boxplots of 19 transition-transversion ratios for pipeline's variants sets (Total_TsTv) along with TsTv of varaints with Mendelian violations. C and D shows beanplots of 19 trios total and Mendelian violated variant sets.
Pipelines comparison for calling variants in monogenic homozygous recessive diseased families {#Sec11}
---------------------------------------------------------------------------------------------
We analyzed two different families with affected children. In Family 1, affected children were diagnosed with the phenotype of hypoplasia of cerebellum which is monogenic homozygous recessive disease \[[@CR24]--[@CR28]\]. In Family 2, affected children were diagnosed with abnormal pain sensation, which is also monogenic homozygous recessive disease \[[@CR29]--[@CR33]\]. The number of variants between pipelines and between quality filter sets follows the similar pattern of what we saw above in the comparison between the pipelines in trios. However, in these cases, the difference in Mendelian violation between the pipelines is strongly pronounced. The difference is more between CASAVA ALL and GATK ALL variants set and the detail of this shown in Table [2](#Tab2){ref-type="table"} and Table [3](#Tab3){ref-type="table"} for Diseased Family 1 and Diseased Family 2 correspondingly. Since, CASAVA PASS and GATK PASS variants sets have lot of similarity; the details of number of variants for various categories are shown in Additional file [7](#MOESM7){ref-type="media"} and Additional file [8](#MOESM8){ref-type="media"} for Diseased Family 1 and Disease Family 2 correspondingly.Table 2**GATK and CASAVA comparison in diseased Family 1**PipelinesCASAVA allGATK allHRCHRCTotalTotalHRC by CASAVA but not by GATKTotalTotalHRC by GATK but not by CASAVATotal number of variant6082624614974096337108637741349TsTv Ratio2.002.091.881.882.031.07Mendelian Violation2909850018792000In dbSNP556230860769268560911662277643In interGenic368213835416257379084936884846In CDS657856502644526382In '3 UTR77701849168221185311In '5 UTR185522400200932464In Intronic411321047220218441292848805818In Non_coding_intronic214239182626217104183940In Exonic578425801567625793In Non_coding_exonic163561100151961060In Putative Promoter Region127801350155171411Non-Synonymous Coding190801591182991612Common Variant5395888603892055905677628301335Common Homozygougs Minor Variant in 1000genome477717556874104479292057620194Common Homozygougs Minor Variant in Q1084727231576431424915274601691293Common Het (\>5%) Variant in 1000genome44796095640799449448357151189Common Het(\>5) Variant in Q108489516859296635422841620271332Comparison between the pipelines have been done for unfiltered sets (CASAVA ALL, GATK ALL), and subset of variants fulfilling the criteria of Homozygous Recessive Conditions (HRC).Table 3**GATK and CASAVA comparison in diseased Family 2**PipelinesCASAVA allGATK allHRCHRCHRC by CASAVA but not by GATKHRC by GATK but not by CASAVATotalTotalTotalGT mismatch by GATKAbsent in GATKTotalTotalTotalGT mismatch by CASAVAAbsent in CASAVATotal number of variant51928912952291278113154597253065320431499544TsTv Ratio1.992.041.30By CASAVA1.212.051.842.001.29By CASAVA1.371.09By GATK1.23By GATK1.37Mendelian Violation41691531By CASAVA0126540288By CASAVA10105By GATK244By GATK3Mendelian Violation in affected27933600By CASAVA0018904300By CASAVA9290By GATK211By GATK0Mendelian Violation in unaffected26736731By CASAVA0118265988By CASAVA2575By GATK82By GATK3In dbSNP4717672284097386568247546872912914581243215In interGenic315290518780602516863266794195271349999350In CDS5476633963353607359261115In '3 UTR64961476440693954831183In '5 UTR1508781000166359211101In Intronic348473719668487427603797577205621381976405In Non_coding_intronic18042193138299182875972795722In Exonic48608284633475772921477In Non_coding_exonic13960828801252582871In Putative Promoter Region1093256000138016610100Non-Synonymous Coding162999743115429100734Common Variant (1000genome + Q108)4693812280347847285652248462918619361418518Homozygougs Minor Variant in 1000genome4089468261623963643240983012656279672670Homozygougs Minor Variant in Q1084283672264796636283544694852767718611357504Het (\>5%) Variant in 1000genome3888895258243863543238974162622578771770Het(\>5) Variant in Q1084355312276166866741249005232885219221408514Non-Synonymous rare Pain related5100051000Pain genes mapped to Non-Synonymous rare variants4100041000Comparison between the pipelines have been done for unfiltered sets (CASAVA ALL, GATK ALL), and subset of variants fulfilling the criteria of Homozygous Recessive Conditions (HRC).
### Diseased family 1 {#Sec12}
The CASAVA pipeline has 4.78% (290985 out of 6082624) of variants with Mendelian violation in the CASAVA ALL set and 1.23% (66903 out of 5438393) of variants with Mendelian violation in the CASAVA PASS set. In contrast, GATK has only 2.96% (187920 out of 6337108) of variants with Mendelian violation in the GATK ALL set and 0.14% (7122 out of 5004048) of variants with Mendelian violation in the GATK PASS set (Table [2](#Tab2){ref-type="table"} and Additional file [7](#MOESM7){ref-type="media"}). Because both children are affected by the hypoplasia of cerebellum, and the parents and aunt are unaffected, the causative variant should be a homozygous variant \[[@CR34]\]. We further investigated the pipeline performance to find the homozygous recessive variants. In this paper, we use the term Homozygous Recessive Condition (HRC) for any particular variant position in a family when all three of the following conditions are met: 1) all affected off-springs are homozygous, 2) all affected off-springs have the same genotype and their genotype is different than normal individuals in the family, and 3) all affected off-springs follow Mendelian inheritance (e.g. Father GT = A/C, Mother GT = A/C, Affected Child 1 GT = C/C, Affected Child 2 GT = C/C). Both CASAVA and GATK pipelines have approximately a similar number of HRC variants (Table [2](#Tab2){ref-type="table"}). They also have a similar number of region specific or known variants like exonic, CDS, 3'UTR, 5'UTR, intronic, non-synonymous coding, 1000genome and so on. Furthermore, the pipelines have a similar number of commonly known variants such as those in 1000 genomes, and Q108 (108 unrelated individuals from Qatar). After filtering the known variants, we tried to map these variants to known genes for the phenotype in the literature. We could not map the set of possible causative variants to known genes in this case. Therefore, we tried another real case of homozygous recessive disease with a pair of normal and affected siblings.
### Diseased family 2 {#Sec13}
This family is different in structure because of the presence of unaffected siblings (Figure [2](#Fig2){ref-type="fig"}B), which gives extra power to evaluate the pipeline because of the inherent validation about the variants, e.g. evaluating homozygous recessive variants identified by both the pipelines but mismatched in genotypes according to Mendelian inheritance in affected and unaffected separately. We have presented a detailed comparison of the pipeline performances for this family in Table [3](#Tab3){ref-type="table"}. The additional parameters to judge the pipelines in Table [3](#Tab3){ref-type="table"}, as compared to the previous case in Table [2](#Tab2){ref-type="table"}, are due to the additional two normal siblings in this case. Exclusively determined HRC variants are divided into two sets of variants for analyzing pipeline performance: 1) HRC variant by Pipeline1 and not by Pipeline2 and having mismatch in genotype calls between the pipelines, and 2) HRC variant by Pipeline1 and none from Pipeline2 for all five individuals.
In the first set of variants (Table [3](#Tab3){ref-type="table"}, Column \"GT mismatch by GATK\" and \"GT mismatch by CASAVA\"), in which only one pipeline meets HRC and the pipelines have mismatch in genotype calls, the pipeline not meeting HRC can either have Mendelian inheritance or Mendelian violation. The cases, where both the pipelines have Mendelian inheritance and only one pipeline meets HRC, are difficult to evaluate in terms of pipeline performance. Example variant position genotypes in this family are as follows:"*Pipeline 1 Genotypes (Mendelian Inheritance and HRC):* Father GT = A/C, Mother GT = A/C, Unaffected Child1 GT = A/C, Unaffected Child2 GT = A/C, Affected Child1 GT = C/C, Affected Child1 GT = C/C;*Pipeline2 Genotypes (Mendelian Inheritance but no HRC):* Father GT = A/A, Mother GT = A/C, Unaffected Child1 GT = A/A, Unaffected Child2 GT = A/A, Affected Child1 GT = A/C, Affected Child2 GT = A/C."
The cases where one pipeline has both HRC and Mendelian inheritance and other pipeline has neither could be a strong indication that the second pipeline calls are wrong in these variants. Example variant position genotypes in this family are as follows:"*Pipeline 1 Genotypes (Mendelian Inheritance and HRC):* Father GT = A/C, Mother GT = A/C, Unaffected Child1 GT = A/C, Unaffected Child2 GT = A/C, Affected Child1 GT = C/C, Affected Child1 GT = C/C;*Pipeline2 Genotypes (No HRC due to Mendelian violation in affected off-springs):* Father GT = A/A, Mother GT = A/A, Unaffected Child1 GT = A/A, Unaffected Child2 GT = A/A, Affected Child1 GT = C/C, Affected Child1 GT = C/C."
In Table [2](#Tab2){ref-type="table"}, we can see in the CASAVA ALL and GATK ALL sets that out of 1499 exclusively determined HRC variants by GATK, 929(62%) had both Mendelian violations and different genotypes by the CASAVA pipeline. In contrast, out of 781 exclusively determined HRC variants by CASAVA, only 244 (31%) have both Mendelian violations and different genotypes. Therefore, we can say that for exclusively determined HRC where there is mismatch between the genotype calls between the pipelines, the GATK pipeline is more robust than the CASAVA pipeline, if we compare all the variants without any quality filter.
We also examined Mendelian violation in another set of exclusively determined HRC variants by one pipeline where there were no variants in any member of the family by the second pipeline (Table [3](#Tab3){ref-type="table"}, Column \"Absent in GATK\" and \"Absent in CASAVA\"). Both CASAVA and GATK have almost no Mendelian violation in these cases.
Table [3](#Tab3){ref-type="table"} also shows many categories to compare CASAVA and GATK. CASAVA identifies slightly more number of Non-synonymous variants compared to GATK. However, GATK has higher percentage of Non-synonymous variants as HRC variants compared to CASAVA. About one hundred of these Non-Synonymous variants of both the pipelines are linked to 60 pain related genes by literature identified using SnpEff \[[@CR21]\] and AnnTools \[[@CR22]\]. After excluding the common variants (variants present in homozygous state in either 1000 genomes or 108 unrelated Qatari individuals, and variant present in heterozygous state with MAF \>5%) from these non-synonymous variants, there were 5 variants left by both the pipelines (Non-synonymous pain related rare variant in Table [3](#Tab3){ref-type="table"}). From both pipelines, out of these 5 variants only one was HRC variant and most probably the causative variant.
Discussion {#Sec14}
==========
We found excellent performances of both GATK and CASAVA pipelines in matching the genotype calls when matching with Illumina OmniArray genotype calls. However, we saw differences in the number of variants called by each pipeline in unfiltered variant sets (CASAVA ALL, GATK ALL) and generally GATK identifies more variants because of its multi-sample calling algorithm. Most of these additional variants are of low quality but not bad in terms of Mendelian inheritance. CASAVA pipeline, in most of the cases, have TsTv ratio closer to 2 compared to GATK. Since both CASAVA and GATK pipeline were unaware of the pedigree structure while calling the genotypes, in conflicting or discordant genotypes by the pipelines, Mendelian inheritance is a good criterion to judge the confidence of variants for familial samples. In general, GATK pipeline called less Mendelian violation for all different sets. Notably, PASS filter in GATK pipeline drastically minimizes Mendelian Violation, from 2.4% in GATK ALL to 0.14% in GATK PASS in disease family 1 and from 4.86% in GATK ALL to 0.19% in GATK PASS in disease family 2. However, in CASAVA pipeline PASS filter does not reduce Mendelian violation significantly, from 4.78% in CASAVA ALL to 1.23% in CASAVA PASS in disease family 1 and from 8.03% in CASAVA ALL to 1.87% in CASAVA PASS in disease family 2. By assuming Mendelian violation to be inversely correlated to pipeline performance in cases of genotype mismatch and where the other pipeline satisfies HRC, GATK multi-sample calling performs better than CASAVA single sample calling for these cases. However, we didn't find any significant difference in the ability of these pipelines to identify causative variants in this abnormal pain perception family. We also found extremely low Mendelian violation in exclusively determined homozygous recessive condition for which variants were not called in any family member by the other pipeline, which suggests robustness of both GATK and CASAVA pipelines in finding the functional variants. This broad level agreement between the pipelines suggests that normally we can avoid calling variants again using more sophisticated algorithm except for specific scientific goals. One of such specific scientific goals could be finding de novo mutation in samples where comprehensiveness of variants are desired and can be obtained by taking combining the variant sets from the pipelines with tolerated false positives. Also, if the cohort sample size is large and scientific goal is based on the phase SNPs, it is desirable to use more sophisticated SNP calling platform such as GATK multiple-sample calling.
On other note, the results presented here should hold for newer version of GATK as well. In furtherance, we did the sensitivity analysis (see Additional file [9](#MOESM9){ref-type="media"}) for 10 different versions of GATK released in last one and half year for our diseased family 2 data set. The relative standard deviation of variant counts of different versions of GATK for before and PASS filter sets are only 0.89% and 2.02% respectively while the difference between GATK and CASAVA presented in this paper using GATK v2.4 are around 4.9% and 7.6% before and PASS filter set respectively. Similarly, the relative standard deviation of TsTv of different versions of GATK for before and PASS filter sets are and only 0.58% and 0.59% respectively while the difference between GATK and CASAVA presented in this paper using GATK v2.4 are around 8.2% and 2.4% before and PASS filter set respectively. Thus, the different version of GATK have very little effect on the number of variants identified and thus doesn't change the results and conclusion drawn in this paper using GATK v2.4.
We have used 3 different type of data set (108 unrelated, 19 trios, and 2 diseased families) to cover some of the various possible data sets. We have found difference in results for related and unrelated individuals. In general, the pipeline comparison results should hold for most of the possible data set with some limitations. We have only tested for sequences coming from Illumina platform that helps in fair comparison of the pipeline but the result might deviate for sequence reads from some other platform. Also, we have not tested for complex diseases like cancer where somatic mutation is frequent.
Conclusion {#Sec15}
==========
High quality SNP calls delivered by commercial NGS sequencing projects in general show concordance with array genotypes and Mendelian inheritance. Application of more sophisticated SNP calling platforms, i.e. using GATK multiple-sample calling, may be helpful in validating and expanding the number of possible candidates, especially in related individuals, but may not provide additional candidates for monogenic disorders. In general, it is futile effort of calling variants again using open source alternative when commercial vendors had already supplied variants sets. However, in cases of related individuals where commercial providers lack the information of relatedness because of confidentially involved, one should try multiple-sample calling to expand variants conforming Mendelian inheritance.
Electronic supplementary material
=================================
{#Sec16}
######
Additional file 1: **Coverage of samples.** (XLSX 34 KB)
######
Additional file 2: **Comparison of GATK and CASAVA pipeline with OmniArray.** (TIFF 657 KB)
######
Additional file 3: **Effect of PASS filter on Minor Allele Frequency distribution in 108 unrelated Qatari individuals.** (PDF 106 KB)
######
Additional file 4: **Effect of PASS filter on variant count distribution in 108 unrelated Qatari individuals.** (PDF 106 KB)
######
Additional file 5: **Het-hom of individual Father, Mother, and Off-spring in 19 trios with GATK and CASVA comparison.** (TIFF 685 KB)
######
Additional file 6: **Pipelines performances in 19 trios using Mendelian violation.** (XLSX 14 KB)
######
Additional file 7: **GATK and CASAVA comparison for quality filtered set in diseased Family 1.** (XLSX 12 KB)
######
Additional file 8: **GATK and CASAVA comparison for quality filtered set in diseased Family 2.** (XLSX 14 KB)
######
Additional file 9: **GATK version sensitivity analysis.** (XLSX 11 KB)
**Competing interests**
The authors declare that they have no competing interests.
**Authors' contributions**
PK and KS designed the study and wrote the manuscript. PK performed the data analysis. MA, WM, NC, ME, AA collected patient samples. MA and WM contributed to trios data analysis. NC, ME, AA contributed to monogenic disease cases data analysis. All authors read and approved the final manuscript.
This work is supported by the Biomedical Research Program funds at Weill Cornell Medical College in Qatar, a program funded by the Qatar Foundation. The statements made herein are solely the responsibility of the authors.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION {#sec1}
============
Tobacco use is a leading preventable cause of death worldwide; the situation is more morbid in developing nations like India^[@cit0001]^. In India, nearly 1 in 10 adolescents in the age group 13--15 years have ever smoked cigarettes and almost half of these report initiating tobacco use before 10 years of age^[@cit0002]^. Tobacco is easily accessible and legally available. It contributes significantly to premature death and long-term suffering. It is a risk factor for cardiovascular diseases, chronic obstructive pulmonary diseases, cancers and oral diseases^[@cit0003]^.
Tobacco is used in various forms, which include smoking and smokeless tobacco^[@cit0003]^. Bidi is the most popular form in rural areas while cigarettes are popular in urban areas. Chillum, hookah, cigars and pipes are other forms. Smokeless tobacco is consumed predominantly by chewing in the form of pan, pan-masala or gutkha, and mishri^[@cit0003]^. Areca nut is also commonly used together with tobacco. Most people are unaware of the side effects of areca nut. Areca nut is known to be mutagenic and has genotoxic effects on the tissues of the body, eventually leading to various neoplastic and paraneoplastic lesions^[@cit0004]^. Areca nut is the fourth most commonly used psychoactive substance in the world after nicotine, alcohol and caffeine^[@cit0005]^.
According to the Global Adult Tobacco Survey-India (GATS-India) 2009, more than one-third (34.6%) of Indians use tobacco in some form: smoking, chewing, application to teeth and gums, or sniffing. Among the tobacco users, 20.6% consumed only smokeless tobacco, 8.7% smoking tobacco only, and 5.3% used both smoking as well as smokeless tobacco^[@cit0003]^. Tobacco use usually starts in the adolescent age group and continues in adulthood. According to WHO, adolescents are most vulnerable to risk-taking behavior that can lead to substance abuse. They are still going through critical periods of growth and development making them vulnerable to nicotine and its harmful effects^[@cit0006]^. Symptoms of serious addiction can appear within weeks or even days after smoking begins, which can lead to years of tobacco use and dependence^[@cit0007]^.
To reduce its prevalence, awareness should be increased, especially among school children because this is the most susceptible and easily modifiable age group. A study by Chadda and Sengupta^[@cit0006]^ concluded that there is a need to collect nationwide data on the use of different forms of tobacco by children and adolescents, and the factors leading to initiation of such harmful habits^[@cit0006]^. Schools play a powerful role in reducing tobacco and areca-nut use by children. The National Tobacco Control Programme (NTCP) includes training of school teachers and various school programs at the district level^[@cit0008]^. Despite such programs, a large number of adolescents is still addicted to tobacco in any form. Hence, this study was conducted, using a questionnaire, to determine the level of awareness about the risk associated with tobacco use among students of grades 7--12.
METHODS {#sec2}
=======
This is a cross-sectional study conducted across 9 randomly selected schools (3 municipal and 6 government-aided) in Ahmedabad, India ([Table 1](#t0001){ref-type="table"}).
######
Distribution of the students
*Municipal n (%)* *Semi-Government n (%)* *Total n*
--------------- ------------------- ------------------------- -----------
**Boys** 61 (4%) 1520 (96%) 1581
**Girls** 31 (2.5%) 1225 (97.5%) 1256
**Users** 24 (6.5%) 347 (93.5%) 371
Tobacco 0 (0%) 20 (100%) 20
Smoking 2 (10%) 18 (90%) 20
Areca-nut 22 (7%) 309 (93%) 331
**Non-users** 77 (3%) 2607 (97%) 2684
The principals of the 9 schools were approached and written informed consent was obtained from all of them. Students from grades 7--12 were selected for the study and consent was also obtained from them. Students were informed about the purpose of the study and assurance was given about confidentiality. An anonymous self-administered close-ended questionnaire was designed for the study. They were asked to fill the questionnaire after they were given a brief explanation about it. The students were included in the study only after receipt of a written informed consent from the principal of the school and oral consent by the student. Any student who was unwilling to participate in the survey or did not understand the questionnaire (shown in the Supplementary File), even after explanation, was excluded.
The collected information was converted into a computer-based spreadsheet and analysed in Microsoft Excel 15.0. Data from 3055 students were analysed. Prevalence was calculated for each question answered. The chi-squared test was applied to analyze any differences between the responses of users and non-users.
RESULTS {#sec3}
=======
[Table 2](#t0002){ref-type="table"} presents the association between tobacco use and participants characteristics, awareness and knowledge. The Cramer's V test (with measure φ~c~) was applied to determine the strength of association between the users and non-users for the knowledge whether passive smoking was harmful or not. There was a weak association between the variables (φ~c~=0.08). Cramer's V demonstrates that the variables have a weak association for awareness that the habit is harmful and watching anti-smoking, chewing tobacco, sopari and alcohol drinking media messages on TV, etc (φ~c~=0.13 and φ~c~=0.0848, respectively). The test revealed a significant difference between the tobacco users and non-users but a weak association for belief that the habit caused cancer, breathing problems, heart problems, paralysis or it did not cause any problems or caused multiple problems (p\<0.001, φ~c~=0.198). Chi-square tests revealed a significant difference between tobacco users and non-users and a weak association for whether they had seen any actors smoking or chewing tobacco or sopari or drinking alcohol on screen (p\<0.001, φ~c~=0.1) or promotional advertisements (p=0.032, φ~c~=0.067). Chi-squared tests also showed statistically significant differences for whether their teacher had talked about the dangers of such habits (p\<0.001) and for whether there had been any discussion about the ill-effects of smoking in class (p\<0.001). Cramer's V showed a weak association for the same (φ~c~=0.098 and φ~c~=0.301, respectively).
######
Association between tobacco use and subject characteristics, awareness and knowledge
*Users n (%)* *Non-users n (%)* *Total n* *p*
--------------------------------------------------------------- ----------------------- --------------- ------------------- ------------ ---------
Passive smoking - harmful No 39 (11%) 153 (6%) 192 \<0.001
Yes 197 (53%) 1876 (70%) 2073
Habit - harmful No 32 (9%) 64 (3%) 96 \<0.001
Yes 241 (65%) 2325 (87%) 2566
Anti-smoking media Many 253 (68%) 2145 (80%) 2398 \<0.001
Few 69 (19%) 390 (14%) 459
None 31 (8%) 111 (4%) 142
Problems caused by the habits Cancer 226 (61%) 1748 (65%) 1974 \<0.001
Breathing 23 (6%) 185 (7%) 208
Heart 10 (3%) 90 (3%) 100
Paralysis 0 (0%) 10 (0.3%) 10
None 61 (17%) 73 (2.7%) 134
Multiple 51 (13%) 440 (17%) 552
Actors on TV/Movies/Video consuming tobacco/areca-nut/alcohol Never Watch TV/Movies 18 (5%) 79 (3%) 97 (3.2%) \<0.001
Frequently 213 (57%) 1549 (58%) 1762 (57.8%)
Sometimes 102 (27.5%) 929 (35%) 1031 (33.7%)
Never 17 (4.5%) 76 (3%) 93 (3%)
Missing 21 (6%) 51 (1%) 72 (2.3%)
Ads/Promotions Many 181 (49%) 1279 (48%) 1460 (48%) 0.0032
Few 107 (29%) 865 (32%) 972 (32%)
None 63 (17%) 480 (18%) 543 (17.7%)
Missing 20 (5%) 60 (2%) 80 (2.3%)
Discussion by the teacher Yes 219 (59%) 1916 (71%) 2135 (53%) \<0.001
No 78 (21%) 358 (13%) 436 (23%)
Not sure 56 (15%) 347 (13%) 403 (20%)
Missing 18 (5%) 63 (3%) 81 (4%)
Discussion in the class Yes 178 (48%) 1451 (54%) 1629 (53%) \<0.001
No 103 (21%) 606 (23%) 709 (23%)
DISCUSSION {#sec4}
==========
This study determined the awareness about the use of tobacco and areca nut among school children in India. The majority of children were aware of the health effects, but 3--6% of the children may not be aware of the ill-effects of such use. The results of the present study are in accordance with the results found by Taran et al.^[@cit0009]^ in 2016 and Tiwari et al.^[@cit0010]^ in 2014. The authors found that 50--90% of the children were aware of the ill-effects of tobacco use. Also, 23% did not recall any discussion in the class or any talks by the teacher regarding the ill-effects of tobacco or areca-nut usage. There is good awareness about the ill-effects of the usage, but measures need to be taken to combat tobacco use in India. The benefits of early educational programs have been well reported for school children^[@cit0011]^. Out of the tobacco users in our sample, the majority knew about the ill-effects caused by areca-nut and tobacco use but still chose to use it, whereas a small percentage was unaware of the risks. These results are contradictory to the findings of Khandelwal et al.^[@cit0012]^ who found that 70% of the users were uninformed of the harmful effects of areca-nut use. Such programs should not only focus on the harms caused by cigarette smoking but also on those caused by other forms of tobacco use, like smoking hooka and bidis, and by the smokeless forms like gutka and areca nut.
Almost half of the students had seen actors consuming tobacco, areca nut or alcohol on TV, in movies or videos, as well as advertisements or promotions for tobacco use. Such actions lead the youth to become more attracted to such products. Tobacco or areca-nut advertising and promotion effectively target young people with images of smokers as trendy, sporty and successful. Characters in the movies or television serials often demonstrate cigarette smoking or alcohol consumption as a routine of daily life. These scenes often stimulate the impressionable mind of the adolescent to adopt similar behaviour^[@cit0006]^.
The results showed a statistically significant difference between tobacco users and non-users, but there was only a weak association. Increasing awareness may lead to reduced consumption in the adolescents as they may have lower awareness levels. Hence, we need to propose new ideas and techniques for spreading awareness and reducing the consumption of tobacco or areca nut. But since the association is only weak, it cannot be said that an increased awareness may lead to reduced consumption. Special attention needs to be given to promoting cessation among current users, as helping them would, in turn, be beneficial for the other students who are tobacco non-users and may be influenced by them.
The current study focused on a small sample. Future studies with larger samples need to be performed. Furthermore, the present study was limited to urban areas of Ahmedabad, hence studies in rural areas and other parts of the country are warranted.
CONCLUSIONS {#sec5}
===========
There is awareness about the ill-effects of tobacco and areca-nut use among the school children of Ahmedabad, but there is still a need to develop programs to combat usage of these products as some of the children still tend to use them despite being aware of the associated risks. There should be better provisions developed for those who are consumers to aid cessation efforts. Schools need to focus more on such areas, and there should be better legislation at the regional level for controlling the promotion and advertising of tobacco.
Supplementary Material
======================
######
Click here for additional data file.
The authors would like to acknowledge all the students who participated in the study and the principals of all the schools.
CONFLICTS OF INTEREST
=====================
Authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none was reported.
FUNDING
=======
There was no source of funding for this research.
PROVENANCE AND PEER REVIEW
==========================
Not commissioned; externally peer reviewed.
| {
"pile_set_name": "PubMed Central"
} |
In the time span of 4 months, everything has changed in the life. The pandemic of coronavirus disease 2019 (COVID-19) has laid waste to the daily routines we practiced automatically. The most basic assumptions of how we plan our day, and how we practice medicine are gone, and in their place we socially isolate, and we split shifts to decrease exposure. The global impact of this invisible virus is terrifying, and we as physicians and healthcare providers are on the front lines of this war, confronting our own mortality as we continually strategize to protect our patients. We have partnered with our colleagues in infectious disease, respiratory medicine, public health, and administrators to learn quickly, adapt our personal and institutional practice, and adopt policies to allow a new best practice, conserve personal protective equipment (PPE), and save our patients and ourselves. One of the main fields that is affected promptly by this virus is the surgery with its all fields and specialties as most policies within the outbreak recommends decreasing the surgical activity by cease the elective operation and maintain only the emergency surgical cases.
Many authors have shared their experience regarding the surgical practice during COVID-19. One study from Italy investigated the impact of COVID-19 on the surgical practice \[[@bib1], [@bib2], [@bib3], [@bib4], [@bib5], [@bib6], [@bib7]\]. They report that during the month preceding the quarantine, 82 patients underwent surgical emergency operations: 19 appendectomies, 17 colorectal resections, 17 small bowel surgeries, 11 cholecystectomies, 5 thoracic procedures for spontaneous pneumothorax and strangulated diaphragmatic hernia, 2 gastric resections, and 11 minor procedures \[[@bib1]\]. During the month after the quarantine, the emergency surgery volume dropped to the number of 12 cases: 7 appendectomies, one foot amputation, one colostomy, 2 small bowel resections, and one cholecystectomy. They observed a 86% decrease of cases of emergency surgery compared to the month before the quarantine \[[@bib1]\].
In Jordan, King Abdullah University Hospital is a tertiary educational center serves the health and care for all northern in Jordan. The quarantine was initiated in Jordan in 18/3/2020. The impact of COVID-19 on the surgical operation is investigated for a period of one month (from 18/3/2020 to 18/4/2020) and compared to surgical record of our center one month before the quarantine. During this month, a total of 183 operations were performed; 121 of them were females. The mean age for the patients 37.1 years; the youngest patient was a 1-day female neonate for myelomeningocele repair and the oldest was an 88-year old female for lower limb embolectomy. 102 of the cases were managed as a mandatory elective case such as the elective cesarean section or oncological cases and 81 were emergency cases. Obstetrics and Gynecology services serve the most frequent operations with 76 case. Also, following operations were performed: 19 for general surgery (including the oncological operations), 16 for neurosurgery, 18 for urology, 16 for vascular surgery, 16 for orthopedics, 7 for ophthalmology, 7 for pediatrics surgery, 3 for maxillo-facial surgery, 2 for thoracic surgery, 2 for otolaryngology and one cardiac surgery. Among the 81 emergency operations, 13 operation were done due to trauma. Cesarean section was the most frequent performed operation with 65 case. Regarding the type of anesthesia, the aim was to utilize the spinal or local anesthesia as much as possible. The spinal and local anesthesia were conducted in 63 and 15 case, respectively. Only one intraoperative complication was reported (bladder injury in cesarean section). Postoperatively, 7 patients developed inpatient complications mostly (6 cases) in form of multi-organ failure and death. [Table 1](#tbl1){ref-type="table"} summarizes the number of operations.Table 1Demographic distribution surgical operation at King Abdullah University Hospital from 18/3/2020 to 18/4/2020.Table 1VariablesNumberPercent (%)Mean ± SD**Sex** Male6233.9 Female12166.1**Age (y)**37.1 ± 20.7**Type of operation** Elective mandatory10255.7 Emergency8144.3**Specialty** General Surgery1910.4 Obstetrics and Gynecology7641.5 Neurosurgery168.7 Thoracic surgery21.1 Otolaryngology21.1 Urology189.8 Pediatrics Surgery73.8 Vascular Surgery168.7 Maxillo-facial Surgery31.6 Orthopedics168.7 Ophthalmology73.8 Cardiac surgery10.5**Trauma**137.1**Intraoperative complications**10.5**Postoperative complications**73.8**Surgical patients with COVID-19**21.1**Type of anesthesia** General anesthesia10557.4 Spinal anesthesia6334.4 Local anesthesia158.2
In contrast and during the month preceding the lockdown, a total of 1622 operations were carried out. Among them, 1417 operations were elective and 205 was emergency. There is a huge difference in the number of operations in all specialty. This is summarized in [Fig. 1](#fig1){ref-type="fig"}.Fig. 1Chart compares the number of operations for many specialties.Fig. 1
Great efforts were done to postpone any operation that can be delayed. Nevertheless, in the near future, this situation could lead to a large amount of operation and pressure on the surgical services. In addition, this situation may develop many complicated cases in the near future due to the postpone of the operation.
Provenance and peer review {#sec1}
==========================
Not commissioned, externally peer reviewed.
Funding {#sec2}
=======
No fund was received.
Ethical approval {#sec3}
================
Not required.
Consent {#sec4}
=======
Not applicabe
Author contribution {#sec5}
===================
All authors contributed significantly and in agreement with the content of the article. All authors presented substantial contributions to the article and participated of correction and final approval of the version to be submitted.
Registration of research studies {#sec6}
================================
Thank you so much.
This is a letter to editor that describes the surgical practice in Jordan during COVID-19 pandemic.
Guarantor {#sec7}
=========
Dr Tagleb Mazahreh.
Declaration of competing interest
=================================
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1}
============
Esophageal cancer is one of the common aggressive malignant tumors, with a high ratio of tumor recurrence and mortality \[[@B1]\], esophageal squamous cell carcinoma (ESCC) occupies major portion \[[@B2]\]. Adenocarcinoma of the esophagogastric junction (AEG), is a representative malignancy located between the esophagus and stomach, and was originally characterized by Siewert \[[@B3]\]. It was well-known to have unique clinicopathological features and biological behavior. In recent decades, the incidence rate of AEG gradually rose globally, particularly in the western countries \[[@B4]\]. Although the therapeutic methods have improved, such as surgery, surgery with adjuvant chemotherapy, radiotherapy, or a combination of these treatments, the prognosis of ESCC and AEG patients is poor \[[@B5]\]. Therefore, it is important to search suitable clinical prognostic factors to supply more accurate and precise evaluates of survival, extremely important in high-fatality malignancies. This can both enhance outcomes and decrease costs by better choosing patients for eligible treatment \[[@B6]\].
Cancer-related systemic inflammatory response plays an important role in the progression and outcome of tumors \[[@B7],[@B8]\]. Several common inflammation-based prognostic scoring systems, such as the neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR), have been reported to have prognostic value in various cancers, and the high score of NLR and PLR were considered bad prognostic \[[@B9],[@B10]\]. In addition, the hemostatic also plays a key role in cancer progression and metastasis \[[@B11],[@B12]\]. Liver-produced fibrinogen is a key factor in the hemostatic cascade. Recent studies have confirmed that high fibrinogen levels correlate with cancer progression, poor response to chemotherapy and adverse prognostic outcome in various malignancies \[[@B13],[@B14]\]. Recently, several studies analyzed a new scoring system, that is, combining preoperative fibrinogen and the NLR (F-NLR). F-NLR has been demonstrated to be a significant prognostic marker in several types of cancers, such as non-small cell lung cancer and gastric cancer \[[@B15],[@B16]\]. Therefore, the current study aimed to evaluate the prognostic value of F-NLR in patients with ESCC and AEG.
Materials and methods {#sec2}
=====================
Patients {#sec2-1}
--------
We performed a retrospective clinical database of 367 patients with ESCC or AEG who underwent curative surgery at Fujian Provincial Hospital, Provincial Clinical College of Fujian Medical University between January 2007 and December 2016. The entire 367 patients were pathologically confirmed to ESCC or AEG, and patients with other tumor types were precluded from this research. Due to majority patients who received neoadjuvant chemotherapy and/or radiotherapy that could have influenced the blood results, 11 patients who had undergone neoadjuvant chemotherapy and/or radiotherapy were excluded. Finally, 356 patients with AEG or ESCC eligible for analysis. Clinicopathological parameters and laboratory inspections of the patients were acquired from the medical records, including sex, age, tumor size, tumor location, histologic differentiation, surgical procedure, TNM stage and blood cell count. The TNM stage was applied according to the 8th TNM classification of American Joint Committee on Cancer (AJCC) staging manual. The work has been reported in line with AMSTAR (Assessing the Methodological Quality of Systematic Reviews) Guidelines.
F-NLR evaluation {#sec2-2}
----------------
Hematological laboratory measurements including neutrophil count, lymphocyte count, and fibrinogen concentrations, were extracted from the daily blood test administered in the week before surgery. The NLR was defined as dividing the neutrophil count by the lymphocyte count. According to the Youden index by Receiver operating characteristic (ROC) curve, the most appropriate cut-off threshold was found as 3.09 g/l for plasma fibrinogen and 1.89 for NLR; for these values, an area under the curve (AUC) as 0.628 and 0.585, respectively. Based on these cut-off values, the F-NLR score was classified as follows: F-NLR score of 2 \[both a hyperfibrinogenemia (≥3.09 g/dl) and high NLR (≥1.89)\], 1 \[either hyperfibrinogenemia (≥3.09 g/l) or high NLR (≥1.89)\], 0 \[neither hyperfibrinogenemia nor high NLR\].
Statistical analysis {#sec2-3}
--------------------
Statistical analysis was done using SPSS software version 22 (IBM, Armonk, New York, U.S.A.). A two-tailed chi-squared test and Spearman-rho test was used to evaluate differences in categorical variables. Differences between the overall survival (OS) generated by the Kaplan--Meier curves were decided using the log-rank test. OS was defined as the time in months between the date of surgery and the date of death or last follow-up. Univariate and multivariate analyses were carried out by Cox regression models to clarify the independent prognostic factors. All *P*-values were quoted two-sided, and a *P*-value of \<0.05 was considered to represent statistical significance.
The work has been reported in line with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) Guidelines.
Results {#sec3}
=======
Patient characteristics {#sec3-1}
-----------------------
A total of 356 patients who were pathologically diagnosed as ESCC or AEG were included in this retrospective analysis ([Table 1](#T1){ref-type="table"}). All patients underwent curative surgery resection. The present study included 280 (78.7%) male and 76 (21.3%) female, the median age was 62 (range 32--76) years. Patients were categorized into three independent groups as follows: F-NLR = 0, 95 (26.7%) patients; F-NLR = 1, 145 (40.7%) patients; and F-NLR = 2, 116 (32.6%) patients. All 49, 114, and 193 patients presented with pathological TNM stages I, II, and III, respectively. According to tumor location, 129 and 227 patients were classified as having AEG and ESCC, respectively. The median follow-up duration was 48.4 months.
###### Characteristics of the recruited patients
Characteristics Median (25th--75th percentile) or number (%)
------------------- ----------------------------------------------
Gender
Male 76 (21.3)
Female 280 (78.7)
Age (years)
\<60 145 (40.7)
≥60 211 (59.3)
TNM stage
I--II 138 (43.4)
III--IV 180 (56.6)
5-year survival
Yes 134 (42.1)
No 184 (57.9)
Tumor size (cm)
\<5 162 (45.5)
≥5 194 (54.5)
NLR 2.87 (1.93--5.89)
PLR 164.92 (114.17--241.15)
Fibrinogen (g/dl) 3.32 (0.66--8.06)
Neutrophil 4.21 (3.08--6.28)
Platelet 1.40 (0.93--1.76)
Lymphocyte 222.15 (175.9--273.3)
Albumin (g/l) 43 (39.55--45.58)
Hemoglobin (g/l) 121 (102--133)
Prognostic analysis based on plasma fibrinogen or NLR {#sec3-2}
-----------------------------------------------------
Patients with hyperfibrinogenemia had a much worse 5-year OS than those with low fibrinogen (31.4 *vs.* 63.3%, *P*\<0.001; [Figure 1](#F1){ref-type="fig"}). Patients with increased NLR had a poorer 5-year OS than those with low NLR (40.4 *vs*. 50.3%, *P*=0.003; [Figure 2](#F2){ref-type="fig"}).
{#F1}
{#F2}
Correlation between F-NLR and clinicopathological factors {#sec3-3}
---------------------------------------------------------
The association between the F-NLR score and clinicopathologic features of patients with AEG and UGC is shown in [Table 2](#T2){ref-type="table"}. There was significant correlation of F-NLR with tumor size (*P*\<0.001) and pathological stage (*P*=0.010).
###### Univariate and Multivariate analyses of factors for prediction of OS
Univariate analysis Multivariate analysis
-------------------------------------------- -------- --------------------- -------- ----------------------- ---------
Gender (male/female) 1.189 0.276
Age (\<60 years/≥60 years) 1.662 0.197
NLR 6.521 0.000\* 0.008 0.985 (0.699, 1.388) 0.930
Tumor location (ESCC/AEG) 0.860 0.354
TNM stage (I, II / III) 109.07 0.000\* 49.835 2.373 (1.867, 3.017) 0.000\*
Differentiation grade (well/moderate/poor) 0.369 0.544
Albumin (\<42 g/l/≥42 g/l) 0.124 0.725
Tumor size (\<5 cm/≥5 cm) 20.56 0.000\* 6.480 0.724 (0.564, 0.928) 0.011\*
PLR 19.43 0.000 1.684 1.396 (0.843, 2.311) 0.194
F-NLR 44.26 0.000 7.657 1.730 (1.173, 2.551) 0.006\*
Fibrinogen (g/dl) 4.965 0.026 0.125 1.083 (0.696, 1.684) 0.724
Hemoglobin (\<120 g/l/≥120 g/l) 2.784 0.095
\* was considered to be statistically significant.
Survival analysis of F-NLR {#sec3-4}
--------------------------
We conducted the Kaplan--Meier analysis and log-rank test to determine the survival differences between the three groups categorized by F-NLR score. For all the patients, the 5-year OS rates were 69.1, 42.6, and 31.9% for F-NLR = 0, F-NLR = 1, and F-NLR = 2, respectively (*P*=0.001, [Figure 3](#F3){ref-type="fig"}).
{#F3}
F-NLR as prognostic factor {#sec3-5}
--------------------------
To identify the independent prognostic indexes for OS, we carried out univariate and multivariate analyses with a Cox proportional hazard model. In a univariate survival analysis, The tumor size (*P*\<0.001), surgical procedure (*P*=0.043), pathological stage (*P*\<0.001), and F-NLR (*P*\<0.001) were associated with OS. Multivariate analysis demonstrated that the tumor size (*P*\<0.001), pathological stage (*P*\<0.001), and F-NLR (*P*\<0.001) were considered independent prognostic factors for OS ([Table 3](#T3){ref-type="table"}).
###### Relationship between F-NLR and clinicopathologic characteristics
Characteristics F-NLR *P*-value
----------------------- ------- ----------- ---- ---------
Gender 0.581
Female 22 33 21
Male 73 112 95
Age (years) 0.056
\<60 48 57 40
≥60 47 88 76
Tumor size \<0.001
\<5 cm 58 65 39
≥5 cm 37 80 77
TNM stage 0.010
I-II 55 54 44
III-IV 40 81 72
Differentiation grade 0.107
High 30 36 22
Moderate-Low 65 109 94
Tumor location 0.158
ESCC 53 98 76
AEG 42 47 40
Discussion {#sec4}
==========
Although surgical techniques and adjuvant treatments has improved, the median survival of ESCC and AEG malignancies remains unsatisfactory \[[@B17]\]. Early diagnosis and treatment are key to increase the OS time of patients. Recent studies revealed that the development of cancer is related to chronic inflammation and hemostatic system \[[@B18]\]. In our current retrospective study, we investigated the prognostic value of F-NLR score and the relationship between F-NLR and clinicopathological features in the patients with AEG and EDCC.
Inflammation and immune cells are essential components of the tumor microenvironments. The systemic inflammatory responses play important role in tumor growth, progression, and metastasis by creating a favorable microenvironment and inhibiting anti-tumor immunity \[[@B21],[@B22]\]. The systemic inflammatory responses disrupt the balance of circulating white blood cell components \[[@B23]\]. Thus, it affects the numbers of neutrophils and lymphocytes in leukocyte during cancer progression. The NLR has been recognized as a representative prognostic indicator in various malignancies \[[@B24]\].
In addition, more and more studies have demonstrated the association between hemostatic system and cancer progression in recent years. Increasing evidence have suggested that the activation of the hemostatic cascade plays a crucial pathophysiological role in tumor aggressiveness \[[@B27]\]. Fibrinogen is a main acute-phase protein and as an important component of the hemostatic system has been shown to be a necessary regulator of the systemic inflammatory state and malignancy progression \[[@B28]\]. It may mediate the original adhesion of white blood cells to endothelial cells and the release of pro-inflammatory cytokines, thus induce cancer cell proliferation and progression \[[@B29]\]. Hyperfibrinogenemia has been confirmed to be a significant prognostic predictor with tumor progression and poor response to chemotherapy in various malignancies \[[@B30]\].
Therefore, the combination serum fibrinogen and NLR (F-NLR) provides a good prognostic marker for cancer patients. Fibrinogen alone or NLR may have a limited effect on tumor progression. F-NLR increases the adverse effects of F-NLR, ultimately increases the predictive significance of cancer patients. In the current study, we demonstrated that univariate and multivariate analyses revealed that preoperative F-NLR was significantly associated with OS, as well as tumor size and pathological stage, which was consistent with previous study. Based on the Kaplan--Meier method, our study divided the patients into three different risk groups according to the preoperative F-NLR level, and F-NLR score 0 group had a longer survival time. The results suggested that F-NLR might be a reliable prognostic marker. The fact that F-NLR score can be obtained from the routine blood sample makes it practical and inexpensive. Thus, F-NLR may be suitable as a more universally hematological marker than other tumor markers.
The present study had several shortages. First of all, the present study was a single institute, retrospective analysis with a small number of patients. Second, although we restricted some possible mixed factors, the hematologic cell counts can be influenced by several factors. Finally, we were short of the follow-up information for disease-free survival, and our conclusions may be reinforced by using other methods of survival. In the future, we will further improve our study to supply more accurate and precise evaluates of survival.
Conclusion {#sec5}
==========
The preoperative F-NLR score is an independent predictor of survival in patients who underwent curative surgery for AEG and ESCC. As it is objectively measured and daily available, which may be a useful clinical biomarker for identifying patients at high prognostic risk and planning individualized treatment strategies for patients with AEG and ESCC.
Availability of data and materials {#sec6}
==================================
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Ethics {#sec7}
======
The present study was discussed and approved by the ethics committee of our hospital. Written informed consent was obtained from each patient.
Consent for publication {#sec8}
=======================
The authors declare consent for publication.
Author Contribution {#sec9}
===================
Guo Tianxing organized the acquisition of data, performed the statistical analyses and wrote a draft of the manuscript. He was involved in the revision process and gave final approval of the version to be submitted for publications. Zhu Lihuan agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Pan Xiaojie made substantial contributions to conception and design of the study and critically revised the manuscript. She gave final approval of the version to be submitted and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Huang Yangyun made contributions to conception and design of the study, supported the acquisition of data and the statistical analyses. She was involved in drafting the manuscript and the revision process. All authors read and approved the final manuscript.
Competing Interests {#sec10}
===================
The authors declare that there are no competing interests associated with the manuscript.
Funding {#sec11}
=======
This work was supported by the Guiding Projects of Fujian Province Science and Technology Plan, China \[grant number 2017Y0018\].
AEG
: adenocarcinoma of the esophagogastric junction
ESCC
: esophageal squamous cell carcinoma
F-NLR
: fibrinogen and the neutrophil to lymphocyte ratio
NLR
: neutrophil to lymphocyte ratio
OS
: overall survival
PLR
: platelet to lymphocyte ratio
TNM
: Tumor Node Metastasis
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
With rapid advances in the field of sports medicine, many pathologies hitherto inaccessible to arthroscopic surgery have now been rendered to the realm of the arthroscopist. Rotator cuff tears have been seen in this purview ever since Wolf \[[@B1]-[@B4]\] did the first all arthroscopic repair. Many techniques have been enumerated \[[@B3],[@B5],[@B6],[@B8],[@B9]\] and they require a long learning curve and numerous expensive instruments. The objective in all methods is to have a stable suture-cuff and suture-bone interface that permits early and adequate rehabilitation which would hasten the recovery and return to work and sport. The commonly used methods in the U- or L-shaped tears employ either the 2-step or the 1-step techniques \[[@B1]\]. In the former, the sutures are passed through each cuff margin one at a time and in the latter, through both margins simultaneously. The first method requires specialized instrumentation and takes more time while the second may result in incomplete repair since it requires the margins to be properly aligned prior to repair. The need is to have a technique combining the advantages of both that would be technically easier, use less instruments and time and give good fixation. We have tried to address these issues in our technique. The advantages that our technique offers are:
1\. It is a simple, cost and time saving percutaneous technique using spinal needles employing the best possible site of insertion for a particular morphology of tear without using additional portals which may add to the operating time and morbidity.
2\. The needle can be inserted perpendicular to the tear margin which gives the best possible strength of the stitch to withstand tear-out and maintains reduction of the repair.
3\. Only commonly used spinal needles and suture material is employed in repair.
Method
======
We perform the procedure (Figure [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}) under general anesthesia in the lateral decubitus position. The lateral position is preferred because it provides a stable positioning of the patient and a pronounced glenohumeral and subacromial opening through traction on the upper limb. First, routine diagnostic arthroscopy using the standard anterior and posterior portals (Figure [3](#F3){ref-type="fig"}) is performed to examine the glenohumeral joint and confirm the presence of a rotator cuff tear. We also employ the lateral portal as needed. The tear margins are debrided with a shaver and any other joint pathology is noted and addressed to, if possible. Once a tear is identified and considered appropriate for repair by this method, the arthroscope is introduced via the lateral portal and subacromial decompression and acromioplasty are done. Cannulas are placed in the anterior and posterior portals for using instruments, suturing and minimizing extravasation of irrigation fluid to the surrounding soft tissues (Figure [4](#F4){ref-type="fig"}). A spinal needle loaded with a No 0 PDS folded on itself to form a loop is percutaneously inserted into the subacromial area posteriorly and under arthroscopic guidance, the tip of the spinal needle with the suture loop is guided to the posterior cuff margin and penetrated across it near the apex of the tear (Figure [5](#F5){ref-type="fig"}). Then another spinal needle loaded with No 0 PDS till the tip is inserted either percutaneously or through the anterior portal, whichever is more convenient depending upon the tear morphology and the healthy tissue of the anterior margin is penetrated and the suture advanced through the needle lumen. Next, this suture is maneuvered through the loop of suture from the posterior needle and by gently pulling on this loop and needle, the suture is extricated posteriorly. Now we have a PDS suture that has gone through the full thickness of the torn cuff and out of the joint. Then, using a suture retriever, the two limbs of the suture are retrieved through one of the two cannulas and a SMC knot (Figure [6](#F6){ref-type="fig"}) applied and tied down to the cuff to achieve margin convergence. This procedure is then followed by applying additional sutures to the tear in a similar fashion. In cases where the torn edge of the cuff located far medially under the acromion, the tear edge location may prevent accurate passage of the needle. We either pass a single loop through the tear edge separately and use the free ends for traction till adequate reduction is achieved or do not employ the spinal needle technique if it is too difficult. Subsequently, the region of the greater tuberosity of the humerus is abraded with a full-radius shaver/burr to create a bed of bleeding bone to promote healing of the cuff to the tuberosity, one or two suture anchors are placed on the decorticated greater tuberosity and their sutures are also retrieved utilizing the suture loop and needle (Figure [7](#F7){ref-type="fig"}). The sutures are then tied preferably with one suture proximal to the horizontal suture placed earlier which then acts as a check rein for the suture anchor. Thus the cuff is compressed onto the tuberosity as well as having multiple side-to-side sutures in the tear (Figure [8](#F8){ref-type="fig"}).
{#F1}
{#F2}
{#F3}
{#F4}
{#F5}
{#F6}
{#F7}
{#F8}
Discussion
==========
Arthroscopic rotator cuff repair is gaining favor with surgeons due to lower morbidity to the surrounding soft-tissue envelope, no deltoid detachment, better visualization of the pathology, better rehabilitation and improved results. However, repairs are technically demanding and need adequate visualization with additional portals (like the rear viewing portal in massive cuff tears) and special instruments to get correct orientation of the sutures which usually take several steps\[[@B3],[@B5],[@B6],[@B8],[@B9]\]. Millet \[[@B10]\] et al advocated a double anchor footprint with a mattress suture technique. Castagna \[[@B5]\] reported a method using a triple loaded suture anchor in the Alex stitch. Burkhart et al proposed placement of multiple simple sutures for convergence to distribute the load evenly over multiple fixation points and thus reduce the chances of stress failure and tear propagation \[[@B3]\]. Burkhart termed \"margin convergence\" to describe the observation that during side-to-side repair the surgeon can visualize the free margin of the tear converging toward the greater tuberosity with each suture being placed and that using margin convergence in the repair of U-shaped tears decreases the amount of strain at the tendon bone interface of the repair and therefore should be protective to the tendon bone interface of the repair \[[@B1]\]. He described a classification for rotator cuff tears as being either crescent-shaped or U-shaped tears \[[@B4]\]. According to Burkhart, the crescent-shaped tear is a disruption of the tendinous insertion from the greater tuberosity of the humerus without a large element of retraction. The U-shaped tear usually appears on initial inspection to be a large retracted tear often medial to the level of the glenoid. Surgical treatment of full-thickness rotator cuff defects has primarily focused on recreating the anatomy of the intact rotator cuff with reinsertion and fixation of the tendon to the greater tuberosity of the humerus using different types of instrumentation \[[@B11]\]. This is to be done with minimal soft tissue trauma so that the envelope integrity is maintained as much as possible. The creation of flaps or \"dog ears\" indicates a noncongruent repair and tension mismatch due to too much squeeze on the cuff margins that will probably fail under cyclic loading \[[@B12]\]. The strength of the fixation of the tendon to its insertion is of paramount importance for eventual success of the repair hence the need for proper fixation of the suture anchors \[[@B10]\]. We employ the routine anterior, posterior and lateral portals for the procedure and use the needles as per the convenience and expertise of the surgeon, obviating the need for additional portals or using more instrumentation via these portals which may cause additional trauma without getting concentric reduction. Also, revision of the suturing if needed (as in the case of \"dog earing\", which should not be accepted) is easy as the technique can be repeated till the surgeon is completely satisfied with his work since the tissue trauma is minimal. However, this technique has pitfalls and is not a panacea for all tears and we believe that tears which do not allow for adequate manipulation of the needles, for example, far medial and subacromial should preferably not be repaired with needles as suture placement will not be accurate. There is also a possibility of suture breakage when the needle with PDS loop is passed through the hard tendon cuff. This is more liable if repeated attempts are made so if repetitious attempts are employed, the suture loop should be changed. Any broken suture should be promptly retrieved with a grasper.
Conclusions
===========
Our all-arthroscopic technique employs simple spinal needles instead of specialized instrumentation for closure of the tears. Also, the numerous steps involved in the repair are lessened so there is a reduction in operating time. All this translates into cost benefits for the hospital and the patient. The needle being under adequate control of the operating surgeon, the cuff margins can be penetrated at a right angle which gives better grip on the tissue and as the needle can be introduced percutaneously, the need for additional portals is obviated and any suture material can be used. Also, with the passage of each suture, the result of the suture on the entire cuff can be arthroscopically assessed. We use the SMC knot \[[@B7]\] which has excellent knot-holding characteristics with least chances of slipping. This technique combines the advantages of side-to-side suturing of the cuff margins with the tendon-to-bone fixation in lesser time and at a lower cost. The technique is simple, easy to understand and replicate.
Competing interests
===================
Each author certifies that he has no commercial associations (e.g. consultancies, stock ownership, equity interests, patent/licensing arrangements, etc.) that might pose a conflict of interest in connection with the submitted article.
Authors\' contributions
=======================
All the authors contributed towards the clinical and surgical endeavour and in drafting of the manuscript and have given final approval of the version to be published.
Acknowledgements
================
Investigation performed at the Department of Orthopedic Surgery, Seoul Veterans Hospital, Seoul, South Korea.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S0001}
============
Obesity is the most common chronic disease in highly developed countries and is one of the most important health risks of our time \[[@CIT0001]\]. Its prevalence has been constantly growing worldwide since the mid 1970s, including among children and youth \[[@CIT0002], [@CIT0003]\]. This rapid increase implicates environmental rather than genetic factors. Several studies report increased risk of obesity in children of obese parents. A relationship between adolescent obesity and decreased physical activity and/or increased "screen" time has been found in many studies. Low family socioeconomic status and low parents' education, together with unhealthy eating habits, may also play a role in promoting obesity in children.
Obesity is associated with an increased risk of impaired glucose metabolism and diabetes, dyslipidaemia, elevated blood pressure, cardiovascular disease, kidney disease and cancer. Paediatricians are often involved in the initial evaluation of obesity in children. Since the recognized methods of obesity treatment are hardly effective, obesity prevention through educational efforts promoting a healthy lifestyle, physical exercise and a balanced diet seems to be particularly essential. To be effective they should be introduced in young children. The study aim was to find out which of the family and social risk factors contribute to the prevalence of obesity in school children aged 7 to 9 years.
Material and methods {#S0002}
====================
The study was conducted from April to July 2001. It was based on the protocol proposed by the European Childhood Obesity Group \[[@CIT0004]\] and conducted in a randomly selected group of 7- to 9-year-old children. This age group was selected for medical and practical reasons. Seven to 9-year-old children are easy to recruit from schools and at this age they are prepubertal (puberty would induce a significant variation in fat content between genders). According to the protocol the country was divided into geographical areas and among them eight areas were randomly selected. For every area, a medical team composed of paediatricians and an anthropologist was ascribed. All of the teams were centrally trained before data collection to standardize the measurement techniques. A sample of primary schools in each area was randomly selected, and the number of pupils of the age group of interest on January 1, 2001 was identified for each school. Each school was split into clusters, each with fewer than 50 pupils, and the randomisation was performed according to the cluster technique described elsewhere \[[@CIT0004]\]. In each school medical team members collected children\'s date of birth and performed anthropometric measurements. Each team was supplied with centrally purchased electronic scales (Radwag 150 OC, Poland) on which weight was measured with precision to 0.1 kg using a standardized procedure (children with underwear only, without shoes). Height was estimated on the wall-mounted measuring tapes available at schools, on a child standing up without shoes and with head, back and buttocks on the vertical plane of the height gauge. Information on parents' obesity (obesity in parents was defined on the basis of self-assessment) and social status and children\'s health status, lifestyle (including physical activity preferences) and eating patterns was also obtained from the questionnaires completed by the parents.
Statistical analysis {#S20003}
--------------------
Statistical analyses were performed on children with complete information on age, gender, height and weight. Children with chronic diseases and \< 7.0 and ≥ 10 years of age were excluded. Body mass index was calculated in all children by dividing weight in kilograms by height in metres squared. International references \[[@CIT0005]\] published by the International Obesity Task Force (IOTF) with cut-off values based on percentiles passing through 30 kg/m^2^ for obesity at the age of 18 years were used for classification of children as obese. Three classes of age were defined as follows: 7 (7.0 to 7.9 years), 8 (8.0 to 8.9) and 9 (9.0 to 9.9) years old. The cut-offs selected for each reference corresponded to the midyear value. They were 21.01, 22.18, and 23.46 for obese 7-, 8-, and 9-year old girls, respectively, and 21.09, 22.17, and 23.39 for obese 7-, 8-, and 9-year-old boys, respectively. Frequencies of obesity are given in percentages.
The odds ratio (OR) was estimated for each factor separately to examine its influence on the prevalence of obesity. A *p* value = 0.05 was considered statistically significant, with the confidence interval (CI) 95%. Odds ratio was estimated using a logistic regression analysis \[[@CIT0006]\]. Data were analysed using the Statistica software package version 5.0.
Statement of ethics {#S20004}
-------------------
Informed consent for participation in the study was obtained from parents and children. The study was approved by the Local Ethics Committee and supported by a grant from the Polish Ministry of Science.
Results {#S0003}
=======
After excluding 57 pupils not willing to participate, a total of 3370 children (1636 girls and 1596 boys) were examined. Seven hundred and one were not included because of incomplete data or chronic disease that could impair their height or weight. Forty-one were eliminated because they were not in the age range for the protocol. The final analysis was performed in 2571 children (1268 girls and 1303 boys). Obesity was diagnosed in 47 (3.7%) girls and 47 (3.6%) boys.
The analysis of family and social obesity risk factors in girls and boys is presented in [Tables I](#T0001){ref-type="table"} and [II](#T0002){ref-type="table"}. There was a statistically significant association between the prevalence of obesity in girls and the mother\'s obesity: OR = 5.06 (1.96--13.05), *p* \< 0.001, father\'s obesity: OR = 5.19 (1.96--13.69), *p* \< 0.001, and both parents' obesity: OR = 5.43 (1.39--21.29), *p* = 0.01.
######
Family and social risk factors of obesity in girls (*n* = 1268)
Risk factors *N* OR 95% CI Value of *p*
----------------------------------------- ------------- ------ ------------- --------------
Obesity in family:
Obese mother vs. nobody obese 175 vs. 677 5.06 1.96--13.05 \< 0.001
Obese father vs. nobody obese 154 vs. 677 5.19 1.96--13.69 \< 0.001
Both parents obese vs. nobody obese 49 vs. 677 5.43 1.39--21.29 0.01
Mother\'s education:
Primary vs. vocational 465 vs. 129 4.85 0.63--36.99 NS
Primary vs. secondary 465 vs. 515 2.21 0.80--6.09 NS
Primary vs. university 465 vs. 159 1.48 0.71--3.11 NS
Father\'s education:
Primary vs. vocational 101 vs. 692 1.37 0.40--4.64 NS
Primary vs. secondary 101 vs. 375 0.94 0.49--1.82 NS
Primary vs. university 101 vs. 100 -- -- --
Income:
Sufficient vs. restrictions required 543 vs. 586 1.15 0.59--2.21 NS
Sufficient vs. insufficient to live on 543 vs. 139 0.82 0.44--1.53 NS
OR -- odds ratio, CI -- confidence interval, NS -- not statistically significant
######
Family and social risk factors of obesity in boys (*n* = 1303)
Risk factors *N* OR 95% CI Value of *p*
----------------------------------------- ------------- ------ ------------- --------------
Obesity in family:
Obese mother vs. nobody obese 166 vs. 746 5.6 2.6--12.02 \< 0.001
Obese father vs. nobody obese 151 vs. 746 6.21 2.89--13.37 \< 0.001
Both parents obese vs. nobody obese 44 vs. 746 7.22 2.44--21.33 \< 0.001
Mother\'s education:
Primary vs. vocational 126 vs. 533 1.12 0.37--3.37 NS
Primary vs. secondary 126 vs. 514 1.08 0.62--1.87 NS
Primary vs. university 126 vs. 130 1.31 0.87--1.96 NS
Father\'s education:
Primary vs. vocational 121 vs. 723 1.04 0.34--3.20 NS
Primary vs. secondary 121 vs. 372 1.18 0.67--2.06 NS
Primary vs. university 121 vs. 87 1.21 0.77--1.90 NS
Income:
Sufficient vs. restrictions required 530 vs. 593 0.60 0.32--1.12 NS
Sufficient vs. insufficient to live on 530 vs. 180 0.88 0.57--1.35 NS
Risk for obesity in boys was also significantly dependent on mother\'s obesity: OR = 5.6 (2.6--12.02), *p* \< 0.001, father\'s obesity: OR = 6.21 (2.89--13.37), *p* \< 0.001, and both parents' obesity: OR = 7.22 (2.44--31.33), *p* \< 0.001.
There was no significant association between the prevalence of obesity in children and the education of the mother, father, or both parents. Socioeconomic status of the family did not affect the development of obesity either.
The analysis of risk factors for obesity associated with lifestyle in girls is presented in [Table III](#T0003){ref-type="table"}, and in boys in [Table IV](#T0004){ref-type="table"}. TV-watching time did not constitute a risk factor for obesity. Physical activity level was not significantly associated with obesity in both genders, but a tendency for significance was noted in boys. In those active 2-3 times per week as compared to those who were active every day: OR = 1.94 (0.92--4.1), *p* = 0.08, and in boys active once a week as compared to those who were active every day: OR = 2.37 (1.25--4.51), *p* = 0.07. There was no significant association between eating out and prevalence of obesity in both genders.
######
Lifestyle-related risk factors of obesity in girls
Risk factors *N* OR 95% CI Value of *p*
----------------------------- -------------- ------ ------------ --------------
Watching television:
\< 2 h vs. 2-4 h per day 755 vs. 479 1.60 0.85--3.01 NS
\< 2 h vs. \> 4 h per day 755 vs. 33 1.07 0.38--3.02 NS
Physical activity:
Daily vs. 2-3 times a week 1079 vs. 168 1.33 0.58--3.06 NS
Daily vs. once a week 1079 vs. 19 -- -- --
Daily vs. none 1079 vs. 1 -- -- --
Extra eating out:
None vs. once a week 831 vs. 357 1.03 0.49--2.18 NS
None vs. 2-3 times a week 831 vs. 59 1.02 0.20--5.06 NS
None vs. \> 3 times a week 831 vs. 21 -- -- --
Having breakfast:
Always vs. usually 823 vs. 277 2.71 1.33--5.51 0.005
Always vs. never 823 vs. 168 1.63 1.08--2.47 0.01
######
Lifestyle-related risk factors of obesity in boys
Risk factors *N* OR 95% CI Value of *p*
----------------------------- -------------- ------ ------------ --------------
Watching television:
\< 2 h vs. 2-4 h per day 681 vs. 577 1.18 0.67--2.10 NS
\< 2 h vs. \> 4 h per day 681 vs. 45 0.77 0.28--2.12 NS
Physical activity:
Daily vs. 2-3 times a week 1143 vs. 141 1.94 0.92--4.10 0.08 (NS)
Daily vs. once a week 1143 vs. 17 2.37 1.25--4.51 0.07 (NS)
Daily vs. none 1143 vs. 2 -- -- --
Extra eating out:
None vs. once a week 845 vs. 374 0.95 0.50--1.79 NS
None vs. 2-3 times a week 845 vs. 51 0.70 0.25--1.91 NS
None vs. \> 3 times a week 845 vs. 32 1.36 0.90--2.06 NS
Having breakfast:
Always vs. usually 868 vs. 290 1.35 0.69--2.64 NS
Always vs. never 868 vs. 145 1.38 0.94--2.03 0.09 (NS)
Risk for obesity was greater in girls who eat breakfast irregularly as compared to those who always have breakfast: OR = 2.71 (1.33--5.51), *p* = 0.005. Odds ratio for girls who never eat breakfast as compared to those who always eat breakfast was statistically significant: OR = 1.63 (1.08--2.47), *p* = 0.01.
There was no statistically significant association between eating breakfast and prevalence of obesity in boys, but a tendency to significance in boys who never eat breakfast as compared to those who always eat breakfast (*p* = 0.09) was noted.
Discussion {#S0004}
==========
Parental obesity is considered to be the most significant risk factor contributing to the development of obesity in children \[[@CIT0007]\]. Our study confirmed this fact as we found a statistically significant association between the prevalence of obesity in girls and boys and the obesity of the mother, father, or both parents. This risk was higher in boys than girls, particularly in those with both obese parents (OR = 7.22). Among other authors Strauss and Knight \[[@CIT0008]\] demonstrated that children develop obesity when one or both parents are obese, the mother\'s obesity being the most contributing factor in this study. Similar results were obtained by Lake *et al*. \[[@CIT0009]\], although the obesity of both parents was most critical. Perez-Pastor *et al*. reported recently that obesity in offspring seems to be confined to those whose same-sex parents are obese \[[@CIT0010], [@CIT0011]\]. In our study maternal and paternal obesity constituted the same risk of their sons' or daughters' obesity but we did not follow our subjects longitudinally.
We did not find any relationship between the prevalence of obesity in children and the education of the mother, father, or both parents. The impact of parents' education on the development of obesity in children was recently investigated by Van Lenthe *et al*. in the GLOBE research project \[[@CIT0012]\]. Significantly higher BMIs among individuals with lower education were observed in that study. However, de Vito *et al*. \[[@CIT0013]\] found no statistically significant relationship between the mother\'s education and the prevalence of obesity in children. A more detailed assessment is certainly required to understand the discrepancy of the results achieved by various authors. To some extent it can be explained by the financial situation of the family, which may be another risk factor contributing to the development of obesity. Although such a correlation has been proven by de Spiegelaere *et al*. \[[@CIT0014]\], we did not find the family\'s financial situation to have any substantial impact on the development of obesity in both genders.
No statistically significant association between eating out and the development of obesity in boys and girls was found. It is, however, questionable whether the parents who were aware of the reasons for their child\'s excessive body weight and their own contribution to his or her health condition gave true answers. Consumption of fast food among children in the US seems to have an adverse effect on dietary quality and increases the risk of obesity \[[@CIT0015]\]. However, dietary factors were not associated with BMI in the study conducted by Jago *et al*. \[[@CIT0016]\]. Some authors suggest that dietary fat consumption is associated with increased BMI in children \[[@CIT0017], [@CIT0018]\], but other studies reported no relationship between dietary fat intake and preschool children\'s BMI \[[@CIT0019]\].
Analysis of eating habits showed that avoiding breakfast was a substantial risk factor contributing to the development of obesity in girls (*p* = 0.01). In boys, this factor had no statistical significance, but a trend towards significance could be observed (*p* = 0.09). The results achieved for a group of girls were in agreement with those of Rampersand *et al*. \[[@CIT0020]\], who have also demonstrated that avoiding breakfast is a risk factor of obesity. Interestingly, in girls who most often had breakfast the odds ratio for the risk of developing obesity was statistically significantly higher than in those who always had breakfast. It implies that the lack of established eating habits may contribute to the development of obesity.
Television watching time did not constitute in our study a risk factor that might contribute to the development of obesity in either of the sexes. Our findings are different from the results of some other authors \[[@CIT0016], [@CIT0021], [@CIT0022]\]. Jago *et al*. \[[@CIT0016]\] found that physical activity and TV viewing were the only significant predictors of BMI of children aged 3 to 6 years. Other authors, in concordance with our study results, have not found such a relationship \[[@CIT0023], [@CIT0024]\]. The discrepancy of the results obtained by various research teams is probably due to the use of different study protocols, as well as to different age of the children examined.
Interesting results were also provided by an analysis of how much of their time children spend doing exercise. Physical exercises did not affect the development of obesity among girls, which, in turn, would suggest a more prominent role of dietary factors in this gender. It was impossible to calculate the odds ratio for girls engaging in physical exercise once a week as compared to those doing so on a daily basis; nor was it possible for those who did not exercise at all, because there were no obese girls in such groups. In a group of boys, however, a tendency to significance (OR = 1.94, *p* = 0.08) was identified for boys taking exercise only 2--3 times a week as compared to those who did so on a daily basis. It was impossible to calculate OR for boys who did not engage in physical exercise at all, as there were no obese boys in that group.
The present study is the first attempt to analyse the family and socioeconomic risk factors of simple obesity conducted on a large randomly selected group representative of the Polish population of school children aged 7 to 9 years. It is a part of the research on prevalence and risk factors of obesity in children aged 7 to 9 years conducted according to the protocol developed by ECOG \[[@CIT0004]\]. Its first results were published in 2005 \[[@CIT0025]\] and 2007 \[[@CIT0026]\]. A comparison of the results compiled by Polish researchers with those obtained for the same age population of French children and using the same research methodology \[[@CIT0027]\] showed that the prevalence of obesity in children of this age group is similar. The conclusions arising from our study indicate that the risk of obesity in children of both genders aged 7 to 9 years was significantly higher if the mother, father, or both parents were also obese. The risk of obesity in girls increased if they did not eat breakfast. Physical activity did not affect the development of obesity in either boys or girls.
It was rather surprising that low physical activity did not contribute to obesity development in these young children. It could be due to the study limitation, which was the reporting of activity by the parents and not recording through a direct technique. However, it may also reflect the fact that small children are naturally quite active, regardless of being overweight, and diet may play a more important role than activity in obesity promotion. Our findings are to some extent supported by the study of Collins *et al*. \[[@CIT0028]\]. They have recently demonstrated a greater reduction in BMI *z*-score in overweight prepubertal children by treatment with a dietary programme compared to a physical activity programme. They are also in accordance with recently published results of the Early Bird 45 study \[[@CIT0029]\] investigating the factors that lead to obesity in 7 to 10 year old children. Based on the longitudinal observations, its authors conclude that physical inactivity is the result rather than the cause of fatness in children.
In conclusion, the implementation of focused preventive strategies for parents and children could stop further increase in the number of obese children and adults. Extensive efforts encouraging healthy living and eating are the most effective way of preventing obesity and adverse health conditions. The findings of our study may help identify families that are most vulnerable to obesity and develop health promotion programmes for them.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
*Allium* plants (Allieae), such as garlic, onion, and chives, have been cultivated as not only foodstuffs but also medicinal plants in the worldwide from ancient. For example, the extract of *Allium* plants, such as garlic, has shown anticancer, antidiabetic, and antibacterial effects. In addition, the National Cancer Institute in the United States had focused on *Allium* species as expecting cancer prevention (Theisen, [@B21]). *Allium* plants are well-known to have various cysteine sulfoxide derivatives such as alliin, methiin, and propiin (Rose et al., [@B20]). The type and contents of cysteine sulfoxides were also known to be different among *Allium* species (Fritsch and Keusgen, [@B9]). The cysteine sulfoxides change to thiosulfinates, such as allicin (Cavallito and Bailey, [@B5]; Cavallito et al., [@B6]), by the reaction with enzyme called alliinase (Ellmore and Feldberg, [@B8]) when the tissues of *Allium* plants are broken. Allicin has been reported to have several biological effects (Gebhardt et al., [@B13]; Briggs et al., [@B3]; Cañizares et al., [@B4]; Oommen et al., [@B19]; Arditti et al., [@B1]). However, unstable thiosulfinates including allicin are changed to organosulfur compounds, such as ajoene (Block et al., [@B2]), methyl 1-(methylthio)ethyl disulfane, and 5,7-diethyl-1,2,3,4,6-pentathiepane (Kuo et al., [@B15]). These compounds are also comparatively unstable and volatile. Although ajoene was known to have significantly anticancer effect, the application as medicines is difficult. On the other hand, several cyclic organosulfur compounds with anticancer effects isolated from the bulbs of the *Allium sativum* (garlic) have been reported by Nohara et al. ([@B18], [@B16], [@B17]). Thus, cyclic organosulfur compounds are important on the development of medicines including anticancer effect. On the basis of this background, we have isolated several comparatively stable organosulfur compounds from *Allium fistulosum* (green onion and welsh onion) (Fukaya et al., [@B12], [@B10]) and *Allium schoenoprasum* var. *foliosum* (Japanese chive) (Fukaya et al., [@B11]). In the course of our ongoing research program for discovery of bioactive organosulfur compounds, the constituents from the leaves of *Allium sativum* were examined. In this article, we discuss the isolation and the structure elucidation of cyclic organosulfur compounds, foliogarlic disulfanes A~1~ (**1**), A~2~ (**2**), and A~3~ (**3**) and foliogarlic trisulfane A~1~(**4**) and A~2~ (**5**), from the leaves of *A. sativum* and the biosynthetic pathways.
Results and Discussions {#s2}
=======================
The fresh leaves of *A. sativum* (15.0 kg) were mixed with water. Then, acetone was added into the mixture to be 80% acetone solution. The solution was concentrated after standing for 4 days (96 h) at room temperature. The acetone extract was portioned between ethyl acetate (EtOAc) and water. The organic fraction was evaporated *in vacuo* and obtained EtOAc fraction as syrup (41.98 g, 0.27% from the plant). The EtOAc fraction was also subjected with the normal and reversed-phase column chromatography and high-performance liquid chromatography (HPLC) to give foliogarlic disulfanes A~1~ (**1**, 0.00013%), A~2~ (**2**, 0.00021%), and A~3~ (**3**, 0.00009%) and foliogarlic trisulfanes A~1~ (**4**, 0.00015%) and A~2~ (**5**, 0.00008%) ([Figure 1](#F1){ref-type="fig"}).
{#F1}
Foliogarlic disulfanes A~1~ (**1**) was obtained as yellow oil and showed positive optical rotation (+160.9). In the Electrospray Ionization MS (ESIMS) measurement of **1**, a pseudomolecular ion peak \[M + Na\]^+^ was observed at *m/z* 343.0282, and the molecular formula was determined as C~12~H~16~O~6~S~2~ on the basis of the High Resolution ESIMS (HRESIMS) peak and the ^13^C NMR data. The ^13^C NMR spectra of **1** showed signals corresponding to a secondary methyl group at δ~C~ 9.1 (6-CH~3~), a methine at δ~C~ 48.1 (C-6), a diastereotopic oxygen-bearing methylene at δ~C~ 74.98 (C-3); an oxygen-bearing methine at δ~C~ 75.03 (C-2); two methines neighboring the electron-withdrawing atom at δ~C~ 90.4 (C-3a) and δ~C~ 103.2 (C-7); an oxygen-bearing quaternary carbon at δ~C~ 79.6 (C-5a); a two-oxygen--bearing quaternary carbon at δ~C~ 119.1 (C-8a); and a lactone carbonyl carbon at δ~C~ 175.4 (C-5) ([Table 1](#T1){ref-type="table"}, [Figure 2](#F2){ref-type="fig"}, and [Supplementary Material](#SM1){ref-type="supplementary-material"}). The correlations of COSY double-quantum filter (DQF COSY) NMR spectroscopy were observed between 6-CH~3~, H-6, and H-7 and between H-2, H-3, and H-3a ([Figure 2](#F2){ref-type="fig"}). The heteronuclear multiple-bond correlation (HMBC) spectrum of **1** is shown in [Figure 2](#F2){ref-type="fig"}. Namely, the correlation of H-2 to C-8a, H-7 to C-5a and H-3a to C-5a indicates acetal structure, the correlation of H-3a to C-5 and C-5a indicates a lactone, and the correlations of H-6 to C-5a and C-7, H-7 to C-5 and 6-CH~3~, 6-CH~3~ to C-5a, C-6, and C-7 indicate a secondary methyl moiety. These evidences indicate that compound **1** had a tetrahydro-2*H*-difuro\[3,2-b:2′,3′-c\]furan-5(5a*H*)-one skeleton. In addition, 1-propenyl disulfane structure at the side chain was confirmed by High Resolution MS (HRMS) and NMR (Nuclear Magnetic Resonance) data. Next, the NOESY spectrum of **1** showed key correlations between H-3a and H-6; and H-6 and H-7 ([Figure 2](#F2){ref-type="fig"}). The results prove that the relative configurations among H-3, H-6, and H-7 were of the same orientation, respectively. Furthermore, the ^1^H and ^13^C NMR signals of **1** assigned to tetrahydro-2*H*-difuro\[3,2-b:2′,3′-c\]furan-5(5a*H*)-one skeleton were superimposable on those of known compound, kujounin A~3~, except for 1-propenyl disulfane moiety (Fukaya et al., [@B10]). All the evidences support that the chemical structure of **1** was (3*S*^\*^,3a*R*^\*^,5a*S*^\*^,6*R*^\*^,7*R*^\*^,8a*R*^\*^)-3,5a-dihydroxy-6-methyl-7-(allyldisulfanyl)tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one.
######
^1^H NMR and ^13^C NMR data of **1** and **2**.
**Position** **1** **2** **2**
-------------- --------------------- ------- --------------------------- ------- ------------------------------ -------
2 4.05 (m) 75.03 4.00 (dd, *J* = 5.5, 7.5) 75.0 α 4.11 (dd, *J* = 4.8, 10.3) 74.2
4.01 (dd, *J* = 5.5, 9.5) β 4.23 (dd, *J* = 3.4, 10.3)
3 4.30 (m) 74.98 4.30 (m) 75.5 4.44 (m) 74.4
3a 4.62 (d-like) 90.4 4.60 (d, *J* = 2.5) 90.4 4.64 (d, *J* = 1.3) 88.6
5 175.4 175.0 171.4
5a 79.6 82.3 81.6
6 2.91 (m) 48.1 2.63 (m) 49.8 2.75 (m) 49.7
7 5.55 (d, 7.0) 103.2 4.78 (d, *J* = 7.0) 94.5 5.00 (d, *J* = 3.5) 97.5
8a 119.1 117.0 117.0
3′ 3.45 (m) 43.5 3.47 (d, *J* = 7.5) 43.8 3.48 (m) 42.7
4′ 5.88 (m) 134.6 5.86 (m) 134.4 5.88 (m) 133.0
5′ 5.09 (d like, 10.0) 118.9 5.12 (d-like, *J* = 9.5) 119.1 5.15 (d-like, *J* = 11.6) 120.0
5.19 (d like, 17.0) 5.18 (d-like, *J* = 16.0) 5.20 (d-like, *J* = 16.4)
6-CH~3~ 1.18 (d, 7.0) 9.1 1.13 (d, *J* = 7.5) 12.6 1.21 (d, *J* = 7.6) 14.0
*^1^H NMR, ^13^C NMR (CD~3~OD, 500 MHz)*.
*^1^H NMR, ^13^C NMR (CDCl~3~, 600 NMR)*.
{#F2}
Foliogarlic disulfanes A~2~ (**2**) and A~3~ (**3**) were isolated as yellow oil with positive specific rotations (**2**: $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ + 139.0° in MeOH) and negative specific rotations (**3**: $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ -- 213.6° in MeOH). In the ESIMS spectra of **2** and **3**, the same quasi-molecular ion peaks (**2** and **3**: \[M+Na\]^+^) were observed at *m*/*z* 343. The molecular formulas (**2** and **3**: C~12~H~16~O~6~S~2~) were determined on the basis of HRESIMS peaks at \[**2**: *m/z* 343.0277, **3**: *m/z* 343.0282 (calcd. 343.0281)\] and the ^13^C NMR data. The ^1^H and ^13^C NMR spectrum of **2** and **3** showed signals corresponding to a secondary methyl group, a methine, a diastereotopic oxygen-bearing methylene, and an oxygen-bearing methine ([Tables 1](#T1){ref-type="table"}, [2](#T2){ref-type="table"} and [Figure 2](#F2){ref-type="fig"}). On the basis of this evidence and detailed examination of DQF COSY and HMBC experiments, the planner structures of **2** and **3** were found to be the same as that of **1**. Next, the relative configurations of **2** and **3** were characterized by the detailed NOESY experiments. The NOESY spectrum of **2** showed key correlations between H-3a and 6-CH~3~; and H-7 and 6-CH~3~ ([Figure 2](#F2){ref-type="fig"}). The NOESY spectrum of **3** showed key correlations between H-3a and H-6; and H-7 and 6-CH~3~ ([Figure 2](#F2){ref-type="fig"}). In addition, the ^1^H and ^13^C NMR signals of **2** and **3** were superimposable on those of known compounds, kujounin A~1~ and A~2~, respectively, except for 1-propenyl disulfane structure (Fukaya et al., [@B12]). Consequently, the chemical structures of foliogarlic disulfanes A~2~ (**2**) and A~3~ (**3**) were determined as (3*S*^\*^,3a*R*^\*^,5a*S*^\*^,6*S*^\*^,7*R*^\*^,8a*R*^\*^)-3,5a-dihydroxy-6-methyl-7-(allyldisulfanyl)tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one and (3*S*^\*^,3a*R*^\*^,5a*S*^\*^,6*R*^\*^,7*S*^\*^,8a*R*^\*^)-3,5a-dihydroxy-6-methyl-7-(allyldisulfanyl)tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one, respectively.
######
^1^H NMR and ^13^C NMR data of **3**.
**Position** **3**
-------------- ---------------------------- ------- ------------------------------ -------
2 4.04 (dd, *J* = 4.5, 10.0) 75.9 α 4.11 (dd, *J* = 4.1, 10.3) 75.4
4.07 (dd, *J* = 3.0, 10.0) β 4.27 (dd, *J* = 1.4, 10.3)
3 4.29 (m) 74.7 4.48 (m) 73.9
3a 4.75 (s-like) 89.9 4.85 (s-like) 87.5
5 174.9 172.2
5a 80.0 78.4
6 2.69 (m) 47.9 2.71 (m) 46.5
7 5.11 (d, *J* = 10.0) 96.3 5.16 (d, *J* = 9.6) 95.5
8a 119.1 117.2
3′ 3.46 (d, *J* = 7.5) 44.1 3.44 (d, *J* = 7.6) 43.2
4′ 5.85 (m) 134.3 5.85 (m) 132.5
5′ 5.14 (d like, *J* = 10.0) 119.4 5.19 (d like, *J* = 10.3) 119.5
5.20 (d like, *J* = 16.5) 5.22 (d like, *J* = 15.8)
6-CH~3~ 1.09 (d, *J* = 6.5) 8.4 1.19 (d, *J* = 6.8) 7.9
*^1^H NMR, ^13^C NMR (CD~3~OD, 500 MHz)*.
*^1^H NMR, ^13^C NMR (CDCl~3~, 600 NMR)*.
Foliogarlic trisulfanes A~1~ (**4**) and A~2~ (**5**) were isolated as yellow oil with positive specific rotations (**4**: $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ +124.6° in MeOH) and negative specific rotations (**5**: $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ −119.8° in MeOH). In the ESIMS spectra of **4** and **5**, the same quasi-molecular ion peaks (**4** and **5**: \[M+Na\]^+^) were observed at *m*/*z* 375. The molecular formulas (**4** and **5**: C~12~H~16~O~6~S~3~) were determined on the basis of HRESIMS peaks at \[**4**: *m/z* 374.9998, **3**: *m/z* 374.0003 (calcd. 374.0001)\], and the ^13^C NMR data. On the basis of the detailed analysis of the ^1^H and ^13^C NMR, 2D-NMR (DQF COSY, HMBC, NOESY) spectrum of **4** and **5**, the relative structures of tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one skeleton on **4** and **5** were found to be the same as those of **1** and **3**, respectively ([Tables 3](#T3){ref-type="table"}, [4](#T4){ref-type="table"} and [Figure 3](#F3){ref-type="fig"}). Next, the ^1^H and ^13^C NMR spectrum at the side chain showed signals corresponding to an allyl group, as well as those of compounds **1**--**3**. The determination of the sulfur linkage was confirmed by the HRMS spectrum. Namely, the pseudomolecular formula was established as C~12~H~16~O~6~S~3~Na. Therefore, compounds **4** and **5** were found to have a trisulfane bridge. Finally, the relative configurations of **4** and **5** were characterized by the comparison of ^13^C NMR data with **1** and **3** and the NOESY experiments. The ^13^C NMR signals of **4** and **5** were superimposable on those of **1** and **3**. All the evidences supported that the chemical structures of **4** and **5** were (3*S*^\*^,3a*R*^\*^,5a*S*^\*^,6*R*^\*^,7*R*^\*^,8a*R*^\*^)-3,5a-dihydroxy-6-methyl-7-(allyltrisulfanyl)tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one and (3*S*^\*^,3a*R*^\*^,5a*S*^\*^,6*R*^\*^,7*S*^\*^,8a*R*^\*^)-3,5a-dihydroxy-6-methyl-7-(allyltrisulfanyl)tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one, respectively.
######
^1^H NMR and ^13^C NMR data of **4**.
**Position** **4**
-------------- ---------------------------- ------- -------------------------- -------
2 4.07 (m) 75.7 4.28 (m) 75.0
3 4.26 (m) 75.0 4.42 (m) 73.9
3a 4.66 (d, *J* = 1.5) 90.6 4.68 (s-like) 88.5
5 175.1 170.2
5a 79.5 78.0
6 2.97 (m) 48.0 2.83 (m) 47.0
7 5.68 (d, *J* = 7.0) 102.3 5.71 (d, *J* = 6.9) 99.5
8a 119.1 119.0
3′ 3.59 (dd, *J* = 6.5, 13.0) 42.3 3.49 (m) 42.6
3.63 (dd, *J* = 7.5, 13.0)
4′ 5.86 (m) 134.3 5.86 (m) 132.5
5′ 5.17 (d-like, *J* = 10.0) 119.5 5.16 (d-like, *J* = 9.6) 119.6
5.22 (d-like, *J* = 17.0) 5.21 (*J* = 16.5)
6-CH~3~ 1.16 (d, *J* = 7.0) 9.2 1.28 (d, *J* = 6.9) 8.6
*^1^H NMR, ^13^C NMR (CD~3~OD, 500 MHz)*.
*^1^H NMR, ^13^C NMR (CDCl~3~, 600 NMR)*.
######
^1^H NMR and ^13^C NMR data of **5**.
**Position** **5**
-------------- --------------------------- -------
2 4.02 (m) 75.6
3 4.25 (m) 75.1
3a 4.71 (d, *J* = 1.5) 90.2
5 175.0
5a 80.1
6 2.63 (m) 48.2
7 5.23 (d, *J* = 9.5) 96.6
8a 118.8
3′ 3.56 (m) 42.7
4′ 5.82 (m) 134.0
5′ 5.15 (d like, *J* = 10.0) 119.9
5.21 (d like, *J* = 16.5)
6-CH~3~ 1.11 (d, *J* = 6.5) 8.7
*^1^H NMR, ^13^C NMR (CD~3~OD, 500 MHz)*.
{#F3}
The biological synthetic pathways for compounds **1**--**5** are presumed. At first, allicin is generated from alliin by alliinase when plant tissues of *A. sativum* are broken. Next, allicin is decomposed into intermediates (a), (b), and (c) by hydrolysis and is reconstructed to disulfane (d) and trisulfane (e) (Jacob, [@B14]). Finally, the structure of tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one skeleton is formed from semidehydroascorbate by cyclization and sulfane formation with the intermediates d and e. Consequently, compounds **1**--**5** were presumed to be obtained ([Figure 4](#F4){ref-type="fig"}).
{#F4}
Conclusion {#s3}
==========
Five new organosulfur compounds, foliogarlic disulfanes **1**--**3** and foliogarlic trisulfanes **4** and **5**, were isolated from the leaves of *A. sativum*. These compounds **1**--**5** have a tetrahydro-2*H*-difuro\[3,2-*b*:2′,3′-*c*\]furan-5(5a*H*)-one skeleton with methyl group at 6-position and 2-propenyl disulfane or 2-propenyl trisulfane group at 7-position. Particularly, foliogarlic trisulfanes **4** and **5** with a trisulfane moiety are a rare compound derived from medicinal plants. The biological effects of these cyclic organosulfur compounds should be studied further.
Experimental {#s4}
============
General
-------
The following instruments were used to obtain physical data: specific rotations, a Horiba (Kyoto, Japan) SEPA-300 digital polarimeter (*l* = 5 cm); IR spectra, JASCO (Tokyo, Japan) FT/IR-4600 Fourier Transform Infrared Spectrometer; ESIMS, Agilent Technologies (CA, US) Quadrupole LC/MS 6130; HRESIMS, SHIMADZU LCMS-IT-TOF; ^1^H NMR spectra, JEOL (Tokyo, Japan) JNM-LA 500 (500 MHz) spectrometer; ^13^C-NMR spectra, JEOL JNM-LA 500 (125 MHz) spectrometer; NOESY spectra, JNM-ECA 600 (600 MHz) spectrometer; HPLC, a Shimadzu (Kyoto, Japan) SPD-20AVP UV-VIS detector. YMC-triart C18 (250 × 4.6 mm i.d. and 250 × 10 mm i.d.) and YMC-triart PFP (250 × 4.6 mm i.d. and 250 × 10 mm i.d.) columns were used for analytical and preparative purposes. The following experimental materials were used for chromatography: normal-phase silica gel column chromatography, silica gel BW-200 (Fuji Silysia Chemical, Ltd. (Aichi, Japan, 150--350 mesh); reversed-phase silica gel column chromatography, Cosmosil 140C~18~-OPN \[Nacalai Tesque (Kyoto, Japan)\], TLC, precoated TLC plates with silica gel 60F~254~ \[Merck (NJ, US), 0.25 mm\] (ordinary phase), and silica gel RP-18 F~254S~ (Merck, 0.25 mm) (reversed phase); reversed-phase HPTLC, precoated TLC plates with silica gel RP-18 WF~254S~. Detection was achieved by spraying with 1% Ce (SO~4~) 2--10% aqueous H~2~SO~4~ followed by heating.
Plant Material
--------------
Fresh leaves of *A. sativum* cultivated in Kochi prefecture, Japan, were obtained as commercial products purchased from Japan Agricultural Cooperatives (JA) farmers\' market (Kochi, Japan) in April 2017. The plants were identified by the authors (H.M. and S.N.).
Extraction and Isolation
------------------------
The fresh leaves of *A. sativum* (15.2 kg) were chopped and mixed with water, and then acetone was added to the mixture to be 80% acetone solution. The mixture was soaked for 4 days (96 h) at room temperature. Evaporation of the filtrate under reduced pressure provided acetone extract (1,500.37 g, 9.87%). The extract was partitioned between EtOAc and H~2~O (1:1, vol/vol) to obtain EtOAc fraction (41.98 g, 0.27%) and aqueous phase. The EtOAc-soluble fraction (41.98 g) was subjected to normal phase silica gel column chromatography \[1,260 g, CHCl~3~-MeOH (1:0 → 100:1 → 50:1 → 30:1 → 10:1 → 0:1, vol/vol)\] to give nine fractions {Fr.1 (1,471.6 mg), Fr.2 (715.5 mg), Fr.3 (7,193.2 mg), Fr.4 (8,339.2 mg), Fr.5 (4,085.8 mg), Fr.6 (1,334.9 mg), Fr.7 (4,367.0 mg), Fr.8 (617.7 mg), Fr.9 (5,841.2 mg)}. r. 5 (4,085.8 mg) was further separated by reversed-phase silica gel column chromatography \[200 g, MeOH-H~2~O (2:8 → 4:6 → 6:4 → 8:2 → 1:0, vol/vol)\] to give 13 fractions {Fr.5-1 (52.3 mg), Fr.5-2 (21.0 mg), Fr.5-3 (31.0 mg), Fr.5-4 (19.1 mg), Fr.5-5 (84.4 mg), Fr.5-6 (281.7 mg), Fr.5-7 (43.5 mg), Fr.5-8 (26.3 mg), Fr.5-9 (67.9 mg), Fr.5-10 (482.9 mg), Fr.5-11 (2,637.7 mg), Fr.5-12 (178.7 mg), Fr.5-13 (16.9 mg)}. Fr.5-5 (84.4 mg) was purified by HPLC {mobile phase: MeOH-H~2~O (35:65, vol/vol) \[YMC-triart PFP (250 × 10 mm i.d.)\]} to give **1** (6.0 mg) and **2** (16.3 mg). Fr.5-6 (281.7 mg) was purified by HPLC {mobile phase: MeOH-H~2~O (50:50, vol/vol) \[YMC-triart C18 (250 × 10 mm i.d.)\]} to give **1** (14.6 mg), **2** (16.0 mg), **4** (24.2 mg), and **5** (12.3 mg). Fr.5-6-5 (37.1 mg) was purified by HPLC {mobile phase: MeOH-H~2~O (45:55, vol/vol) \[YMC-triart C18 (250 × 10 mm i.d.)\]} to give **3** (14.7 mg) and **4** (4.5 mg).
Foliogarlic Disulfane A~1~ (1)
------------------------------
Yellow oil; $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ +160.9 (MeOH); HRESIMS: calcd for C~12~H~16~O~6~S~2~Na (M+Na)^+^: 343.0281, found: 343.0282; IR(ATR): 3,400, 2,975, 1,782 cm^−1^; ^1^H NMR (CD~3~OD), ^13^C NMR (CD~3~OD, 500 MHz): given in [Table 1](#T1){ref-type="table"}.
Foliogarlic Disulfane A~2~ (2)
------------------------------
Yellow oil; $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ +139.0 (MeOH); HRESIMS: calcd for C~12~H~16~O~6~S~2~Na (M+Na)^+^: 343.0281, found: 343.0277; IR(ATR): 3,400, 2,970, 1,785 cm^−1^; ^1^H NMR (CD~3~OD, CDCl~3~), ^13^C NMR (CD~3~OD, CDCl~3~): given in [Table 1](#T1){ref-type="table"}.
Foliogarlic Disulfane A~3~ (3)
------------------------------
Yellow oil; $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25} -$213.6 (MeOH); HRESIMS: calcd for C~12~H~16~O~6~S~2~Na (M+Na)^+^: 343.0281, found: 343.0282; IR(ATR): 3,400, 2,981, 1,785 cm^−1^; ^1^H NMR (CD~3~OD, CDCl~3~), ^13^C NMR (CD~3~OD, CDCl~3~): given in [Table 2](#T2){ref-type="table"}.
Foliogarlic Trisulfane A~1~ (4)
-------------------------------
Yellow oil; $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25}$ +124.6 (MeOH); HRESIMS: calcd for C~12~H~16~O~6~S~3~Na (M+Na)^+^: 375.0001, found: 374.9998; IR(ATR): 3,400, 2,975, 1,780 cm^−1^; ^1^H NMR (CD~3~OD, CDCl~3~), ^13^C NMR (CD~3~OD, CDCl~3~): given in [Table 3](#T3){ref-type="table"}.
Foliogarlic Trisulfane A~2~ (5)
-------------------------------
Yellow oil; $\left\lbrack \alpha \right\rbrack_{\text{D}}^{25} -$119.8 (MeOH); HRESIMS: calcd for C~12~H~16~O~6~S~3~Na (M+Na)^+^: 375.0001, found: 375.0003; IR(ATR): 3,400, 2,931, 1,789 cm^−1^; ^1^H NMR (CD~3~OD), ^13^C NMR (CD~3~OD): given in [Table 4](#T4){ref-type="table"}.
Data Availability Statement {#s5}
===========================
All datasets generated for this study are included in the article/[Supplementary Material](#s8){ref-type="sec"}.
Author Contributions {#s6}
====================
MF: isolation of constituents and Structure elucidation of new compounds. SeiN: Structure elucidation of new compounds and overall supervision in this study. HH and DN: isolation of constituents. SouN: Preparation of plant extracts and overall supervision in this study. TY: Structure elucidation of new compounds. HM: overall supervision in this study.
Conflict of Interest {#s7}
====================
The 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.
**Funding.** This research was supported in part by a Ministry of Education, Culture, Sports, Science and Technology (MEXT)-Supported Program for the Strategic Research Foundation at Private Universities 2015--2019. This work was supported by JSPS KAKENHI Grant Number 17K08354 (SeiN).
Supplementary Material {#s8}
======================
The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fchem.2020.00282/full#supplementary-material>
######
Click here for additional data file.
[^1]: Edited by: Tao Wang, Tianjin University of Traditional Chinese Medicine, China
[^2]: Reviewed by: Florenci Vicent González, University of Jaume I, Spain; Hitendra M. Patel, Sardar Patel University, India
[^3]: This article was submitted to Organic Chemistry, a section of the journal Frontiers in Chemistry
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
*Yersinia enterocolitica* is widely distributed in the natural world; and being one of the few intestinal bacteria that can grow at low temperatures, the animal hosts of *Y. enterocolitica* are widely distributed, including livestock, poultry, rodents, reptiles, and aquatic animals. It causes gastrointestinal symptoms and systemic disease, e.g., liver and spleen abscess (Thoerner et al., [@B14]), reactive arthritis, erythema nodosum, and anemic goiter (Heyma et al., [@B7]; Stuart and Woodward, [@B13]; Gaede and Heesemann, [@B3]; Wang et al., [@B16]); and can cause septicemia and death (Bottone, [@B2]). Biotyping and serotyping are common and valuable methods for *Y. enterocolitica* identification where at present *Y. enterocolitica* is divided into six biotypes (1A, 1B, 2, 3, 4, and 5) and 60 serotypes. The highly pathogenic *Y. enterocolitica* bioserotype 1B/O:8 strain was distributed worldwide (Wang et al., [@B19], [@B15], [@B16]; Liang et al., [@B9]); while the other bio-serotypes were the lower or non-pathogenic strains and distributed all over the world (Wang et al., [@B18]). From our previous investigation, 58 serotypes of *Y. enterocolitica* had been found and no highly 1B/O: 8 strain has been isolated in China (Wang et al., [@B19]; Gu et al., [@B5]). The bio-serotype 3/O: 3 and 2/O: 9 strains were the primary types found in China.
Since the 1990s, the isolation numbers of bio-serotype 2/O: 9 strains have decreased, while the 3/O: 3 strains became the dominant bio-serotype in China. From 2009 to 2011 in a national survey for *Y. enterocolitica*, 2/O: 9 almost disappeared where only six O: 9 strains were identified from 862 *Y. enterocolitica* (Liang et al., [@B9]). The mechanisms of this change were not known, and therefore to determine the reason, we performed comparative growth of the two strains *in vitro* and *in vivo*, and compared the bacteria produced cytokine changes of the two strains in infected BALB/C mice.
Materials and methods {#s2}
=====================
Bacterial strains
-----------------
NX1998-SA98-837 (bio-serotype 3/O: 3, strain A) and NX1998-SA98-835 (bio-serotype 2/O: 9, strain B) were selected; both strains were isolated from Ningxia Province in 1998. We chose additional two pathogenic strains BJ2009-3-248 (3/O: 3, strain C) and BJ2009-3-247 (2/O: 9, strain D) isolated from Beijing in 2009. All four strains carried *ail, ystA, virF*, and *yadA* virulence genes, indicated that the strains were pathogenic ones.
*In vitro* cultures
-------------------
Two colonies of strain A and B were selected either as a single culture, or the strains were mixed as mixed cultures. All the cultures were inoculated in BHI broth and incubated at 25°C for 48 h, and shaken at 120 rpm. Colony counts were performed every 3 h, and each colony in the mixed bacterial culture was identified using monoclonal antibodies grown in our laboratory; and finally comparative growth curves were drawn. The experimental procedure was performed twice for each strain; and strains C and D were also included.
*In vivo* cultures
------------------
Six week-old healthy female BALB/C mice were purchased from the Chinese Academy of Food and Drug Testing Laboratory Animal Resource Center. Experiments were performed using the four strains as mentioned above. For each experiment, O: 3 and O: 9 strains were mixed to infect animals using equal number of each bacterium in the experimental groups; O: 3 and O: 9 strains singly infected mice were the two control groups (O: 3 control group and O: 9 control group), and healthy mice were selected as a blank group. Approximately 10^7^ cfu/ml of the bacteria were used for intraperitoneal injection, three mice were removed randomly at 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, and 48 h after infection and sub-ocular sinus blood was separated and the serum was stored at −70°C. Mice were sacrificed by cervical dislocation; the spleens were removed after immersion disinfection. Three spleens were mixed with 3 ml PBS to form a suspension, and diluted 10-fold to 10^−3^; and the appropriate concentration to coat plates at 3 h was made. The plates were cultured at 25°C for 24 h.
Cytokines
---------
The content of cytokines from each group of mice was measured and the serum levels of IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-17α, eotaxin, G-CSF, GM-CSF, IFN-r, KC, MCP -1, MIP-1α, MIP-1β, and RANTES were determined using a Bio-Plex mouse 23 cytokine kit purchased from the BioRad Company (USA). TNF-α was measured using an ELISA kit purchased from the R&D Company (USA). After processing the data, a comparison of the cytokine levels in the three groups of mice was performed.
Statistical analysis
--------------------
The amount of bacteria at each time point was bacterial counts per milliliter and each cytokine measured value were expressed as mean ± standard deviation. The statistical differences between each group were analyzed using analysis of variance or the *T*-test where differences had significance when *P* ≤ 0.05.
Ethics statement
----------------
The animals were handled according to the national criterion for animal investigation of China (Ethics Review Committee \[Institutional Review Board (IRB)\] of National Institute for Communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention). The experimental and detection protocols were carried out in accordance with relevant guidelines and regulations. The ethics of this study was approved by Ethics Review Committee of National Institute for Communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention. The License numbers were ICDC-2015001 and ICDC-2015002.
Results {#s3}
=======
The initial bacteria concentration of NX1998-SA98-837 was 1.03 × 10^8^ cfu/ml and NX1998-SA98-835 was 0.89 × 10^8^ cfu/ml; and BJ2009-3-248 was 0.65 × 10^8^ cfu/ml and BJ2009-3-247 was 0.87 × 10^8^ cfu/ml.
The growth of different bio-serotype strains *in vitro* or *in vivo* was different. Comparing the growth curves of the four strains *in vitro*, we found that 2/O: 9 strains had statistically significant higher growth than 3/O: 3 strains (Figures [1A,B](#F1){ref-type="fig"}), either in single culture or mixed cultures. However, the growth curves of four strains *in vivo* were inverse, 3/O: 3 grew statistically significantly higher than the 2/O: 9 (Figures [1C,D](#F1){ref-type="fig"}), either in single infection or mixed infection.
{#F1}
The cytokine changes in mouse sera infected with the two bio-serotype strains were similar where only IL-6, IL-8, IL-13, G-CSF, KC, MCP-1, MIP-1α, RANTES, and TNF-α were different statistically compared to the control group (Figure [2](#F2){ref-type="fig"}). For IL-8, IL-13, MCP-1, MIP-1α, RANTES, and TNF-α, the cytokine values using the mixed infection with two strains were higher than single infections; and there was no difference with 3/O: 3 or 2/O: 9 independently infected. For IL-6 and KC, the values from 3/O: 3 infections were higher than the 2/O: 9 infections and mixed infections; and for G-CSF, the values of 2/O: 9 infections were higher than 3/O: 3 and mixed infections. The experiments were repeated twice, and the all the results were consist.
{#F2}
Discussion {#s4}
==========
In our study, the competitive growth of the two bio-serotype strains were totally different, the 3/O: 3 grew greater than 2/O: 9 *in vivo*; conversely, the 2/O: 9 grew greater than 3/O: 3 *in vitro*. This suggested the 2/O: 9 grew better without a host; however, the 3/O: 3 had stronger resistance to the defense mechanisms of the hosts. Therefore, the clearance of the 3/O: 3 was slower than the 2/O: 9 in BALB/C mice; and might reflected the different pathogenic abilities. This showed the higher resistant ability of 3/O: 3 to host clearance than the 2/O: 9, and might be a possible explanation for distribution of the two strains in human hosts.
Cytokines participate in cellular immunity, humoral immunity, hematogenesis regulation, cell proliferation, and injury repair; however, excess cytokine secretion leads to immunopathogenic effects. Some research showed excess cytokines induced host death after infections (Hancock et al., [@B6]; Segura et al., [@B11]; Smith et al., [@B12]). Currently, the studies on different serotypes of *Y. enterocolitica* that cause different secretions of cytokines were shown (McNally et al., [@B10]). In our study here, the *Yersinia* highly susceptible BALB/C mice were used instead of cells (Autenrieth and Heesemann, [@B1]). Cytokines involved in this study, IL-6, IL-8, MCP-1, MIP-1α, and RANTES in a relatively short period of time reach their peaks in both the two control groups and the experimental group. These cytokines were inflammatory cytokines. This was consistent with the reported literature: *Yersinia* infection of epithelium (HeLa) cells can significantly enhance the levels of transcription and secretion of inflammatory cytokines MCP-1, GM-CSF, and TNF-α. One hour after infection transcript levels of these cytokines began to increase and within 3--4 h reached their peak and then gradually decline (Kampik et al., [@B8]). These cytokines played important roles in the hosts\' immune response against *Y. enterocolitica* and the cytokine changes for *Y. enterocolitica* infection were similar using cell infected models or animal infected models.
Previous studies showed the cytokine values of highly pathogenic *Y. enterocolitica* were higher than the lower pathogenic strains (Gu et al., [@B4]). Brucella and 2/O: 9 bio-serotype *Y. enterocolitica* caused different cytokine changes, involving their different pathogenic ability and life cycles (Wang et al., [@B17]). However, in our study lower pathogenic *Y. enterocolitica* had similar cytokine changes compared with 3/O: 3 and 2/O: 9 strains. Therefore, the similar cytokine changes were caused by analogical pathogenic ability of the *Y. enterocolitica* strains. The different pathogenic abilities or life cycles were important for cytokine changes in *Y. enterocolitica* infections. The cytokines induced by two kinds of strains were similar, while the growth conditions were different; maybe this reflected the different pathogenic ability or defense ability to hosts, and it may as one of the reason for bacteria shift.
Author contributions {#s5}
====================
Design of the work: HJ and XW. Do the experiments: HY, HQ, JL, KL, YX, and RD. Analysis the data: HY and WG. Drafting the work: WG. Revising it critically for important intellectual content: GS.
Funding {#s6}
=======
This work was supported by the National Sci-Tech Key Project (2012ZX10004-201, 2012ZX10004-212, and 2013ZX10004203-002).
Conflict of interest statement
------------------------------
The 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.
We thank Liuying Tang and Jim Nelson for critical reading of and helpful comments on our manuscript.
[^1]: Edited by: Dongsheng Zhou, Beijing Institute of Microbiology and Epidemiology, China
[^2]: Reviewed by: Xue-jie Yu, University of Texas Medical Branch, USA; Shiwen Wang, National Institute for Viral Disease and Prevention, China
[^3]: †These authors have contributed equally to this work.
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper and its Supporting Information files.
Introduction {#sec001}
============
The critical power test is a reliable tool to understand fatigue processes, evaluate human performance \[[@pone.0192552.ref001]\] and prescribe training as it is able to provide aerobic and anaerobic parameters using a non-invasive method \[[@pone.0192552.ref002]\]. The produced power is mathematically plotted against the respective exhaustion times to obtain the intensity of "fatigue threshold" \[[@pone.0192552.ref001]\], or critical power (CP), above which maximal oxygen uptake is evoked and the amount of work that can be performed above CP (W') \[[@pone.0192552.ref002]\]. The evaluation needs different sessions of exhaustive effort (three or more intensities) \[[@pone.0192552.ref003]\]. This method was originally applied in three different models: hyperbolic--power versus time limit (Hyp), linear--power versus 1/time (P vs 1/t) and linear--work versus time limit (Ԏ vs t) \[[@pone.0192552.ref003]\]. Time limit represents the maximum time that the individual maintains a certain exercise intensity \[[@pone.0192552.ref001]\].
More recently, other researchers proposed a different methodology reasoning in critical power linear model theory (power versus 1/time), observing the mechanical power component over a maximal 180-s single effort \[[@pone.0192552.ref004]\]. The 3-minute all-out test (3MT) results can obtain the same parameters as the conventional CP test but using only one day of maximum exercise. In this model, the aerobic capacity is called EP (end power), and the W' is called WEP. The original 3MT application \[[@pone.0192552.ref004]\] was performed on a cycle ergometer, and its reproducibility was confirmed by several authors on different ergometers \[[@pone.0192552.ref005], [@pone.0192552.ref006], [@pone.0192552.ref007]\]. The 3MT has already been applied as a critical velocity predictor in free running \[[@pone.0192552.ref002], [@pone.0192552.ref008]\], and the reliability of this test using measurements of tethered running power output on a non-motorized treadmill (NMT) has been recently demonstrated \[[@pone.0192552.ref007]\].
Mechanical power-based physical assessments in the specific movement of running are important in sports because they provide mechanical performance data that enables to determine the work (Ԏ) executed and allows physiological measures of energy expenditure during exercise \[[@pone.0192552.ref009], [@pone.0192552.ref010]\]. Furthermore, power output measure can be an essential tool to determine the individual potential of runners since it reflects a complete mechanical efficiency parameter--athletes who run faster using less force save energy outlay \[[@pone.0192552.ref007], [@pone.0192552.ref009]\]. In most sports that include running exercises, power output parameters are required to prescribe training, since their performance is a result of the muscle's ability to respond as fast as it can, showing that running power plays a decisive role in several competitions, both in individual and team sports. Therefore, evaluations that determine aerobic and anaerobic parameters in running, measured by mechanical power output, are a reliable tool to identify powerful runners, the best training prescription, and performance prediction at different running intensities \[[@pone.0192552.ref007], [@pone.0192552.ref009]\].
Despite the power output's importance for running performance \[[@pone.0192552.ref009]\], it is difficult to determine this variable maintaining the movement's specificity when obtaining physiological and biomechanical information. Most protocols developed for obtaining physiological measures in runners are performed in the field or on motorized treadmill (MT), and aim to determine such intensities based on the velocity achieved by the individual \[[@pone.0192552.ref002], [@pone.0192552.ref008], [@pone.0192552.ref011]\]. Collecting information on the athlete's power output and velocity capabilities provides a useful running efficiency index \[[@pone.0192552.ref007], [@pone.0192552.ref009], [@pone.0192552.ref012]\]. With this evaluative tool, which measures power output, in addition to physiological measures of aerobic and anaerobic capacity, it is possible to individually investigate the success or failure of trainings aimed at improving performance \[[@pone.0192552.ref009], [@pone.0192552.ref007]\]. The advantage of the 3MT test on NMT is that it is based on mechanical power output, which is the same power output parameter used in the traditional and original application of the CP concept \[[@pone.0192552.ref004]\]. In order to find an assessment method for runners that conserves movement specificity and is able to provide physiological parameters based on power output running in a single day, the objective of this study was to investigate if 3MT is valid for obtaining CP and W' on NMT in tethered running. Similarly to the results found in cycle ergometer \[[@pone.0192552.ref004]\], our hypothesis is that EP and WEP from 3-min all-out test are valid for determining critical power and anaerobic work capacity, respectively, in tethered running.
Materials and methods {#sec002}
=====================
Subjects {#sec003}
--------
Eight physically active men participated in this study (24 ± 3 years; 78.3 ± 8.7 kg; 179 ± 5 cm; 9.0 ± 2.5% body fat), performing the proposed tests after signing a written informed consent form approved by the Human Research Ethics Committee of the University of Campinas (protocol CAAE: 07716512.1.0000.5404), in accordance with the Declaration of Helsinki. The participants answered the International Physical Activity Questionnaire (IPAQ), in which the minimum score to classify them as "physically active" was used as inclusion criterion \[[@pone.0192552.ref013]\].
Experimental procedures {#sec004}
-----------------------
The protocol consisted of seven days of visit to the laboratory. The subjects did not perform 3MT trials before the tests. They performed the tethered running familiarization two days prior to 3MT and CP sessions, consisting of ten sprints of ten seconds in the NMT with the aim of minimizing the learning effect during the acceleration phase. They then followed the protocol for determining anthropometric assessment, as previously described \[[@pone.0192552.ref012]\], and CP in four days. Subsequently, they carried out the 3MT test on the seventh day. Minimum 48-hour and maximum 72-hour intervals between tests were adopted \[[@pone.0192552.ref014]\]. Before all tests, the participants warmed up on a motorized treadmill running for five minutes at a 7.0 km/h speed. The warm-up, NMT ergometer familiarization, laboratory's temperature and humidity conditions, and signal system to determine the mechanical parameters of force (N), velocity (m·s^−1^) and power output (W) followed the protocol established by previous studies in the same ergometer \[[@pone.0192552.ref007], [@pone.0192552.ref012]\] (for additional information see the Details of the methods in [S1 File](#pone.0192552.s001){ref-type="supplementary-material"}).
Critical power test {#sec005}
-------------------
In order to determine CP through the NMT-adapted protocol, the participants performed four efforts until exhaustion against different resistances in a randomized order. They ran tethered by a steel cable attached to an elastic cord ([Fig 1](#pone.0192552.g001){ref-type="fig"}). Resistance was increased between efforts by adding elastic cords (brand new elastics were used). The lowest resistance used three elastics, increasing one by one until the heaviest resistance (six elastics). The resistive force was measured by a signal acquisition system. Predictive efforts lasted between 2 and 15 minutes, and the participants maintained a constant intensity until exhaustion. An exhaustion apparatus was developed by the authors to keep runners at a targeted area, thus avoiding imposed load variation caused by change in position on the treadmill. When the runner was no longer able to sustain the established position on the treadmill, the exhaustion apparatus beeped, and the runner had five seconds to return to the established position. When the runner failed to return to his position due to exhaustion, the test was terminated \[[@pone.0192552.ref012]\].
{#pone.0192552.g001}
Determining the CP standard protocol's aerobic and anaerobic parameters {#sec006}
-----------------------------------------------------------------------
In order to establish the intensity of each CP test predictive load, we calculated all graphs of horizontal power obtained in each of the four test sessions referred to as predictive load 1 (predictive 3 elastics), predictive load 2 (predictive 4 elastics), predictive load 3 (predictive 5 elastics) and predictive load 4 (predictive 6 elastics). The mechanical power values obtained were subsequently plotted with their corresponding time to exhaustion (limit time) for application of three mathematical models: power hyperbolic versus time limit (Hyp), linear power versus 1/time (P vs 1/t) and linear work versus time limit (Ԏ vs t). Through graphical analysis, it was possible to obtain the aerobic (CP) and anaerobic (W') capacity parameters originating from the Hyp model (CP Hyp and W' Hyp), the Ԏ vs t model (CP Ԏ vs T and W' Ԏ vs t) and the P vs 1/t model (CP P vs 1/t and W' P vs 1/t).
3MT test {#sec007}
--------
In the 3MT test, the participants ran on a NMT ([Fig 1](#pone.0192552.g001){ref-type="fig"}) tethered to an adjustable-height pole by a steel cable attached to a load cell (CSL/ZL-250, MK Control and Instrumentation) \[[@pone.0192552.ref007]\]. The test session was only successful when the volunteer ran for the suggested 3 minutes non-stop. The participants were constantly encouraged, but they did not receive any information about the time during the test.
### Determining 3MT aerobic and anaerobic parameters {#sec008}
The mechanical power output generated by each subject during the 3MT test was recorded and analyzed against time to find the aerobic capacity (EP) and anaerobic work capacity (WEP) values of the power output graph \[[@pone.0192552.ref007]\].
Physiological analysis {#sec009}
----------------------
Blood samples were collected from the ear lobe twice while at rest (LAC REP), and during 3MT and CP tests after 5 minutes post-exercise (LAC P5). The LAC P5 concentration was considered the peak value \[[@pone.0192552.ref012]\].
In the 3MT and predictive runs, the blood lactate concentrations were determined using the enzymatic method by spectrophotometry \[[@pone.0192552.ref007], [@pone.0192552.ref012]\]. Lactate concentration was determined at 340 nm against a calibration curve with 5, 10, 15 and 30 mM standards \[[@pone.0192552.ref015]\].
The participants' heart rate (HR) was recorded (beat to beat) using Polar heart monitors (RS800CX). Data was recorded at rest (HR R) and immediately after the end of all tests (HR PT).
Statistical analysis {#sec010}
--------------------
The statistical analyses were performed using Statistica 7.0 (StatSoft, USA) software. Results are expressed as mean values, standard deviation (± SD) and coefficients of variation (CV%). CV% was obtained as SD normalized by the mean of a given variable. Initially, Shapiro-Wilk normality and Levene homogeneity tests were applied to identify the data's characteristics. Based on the normality and homogeneity shown by data, the statistical methods suggested were parametric and adopted in all procedures.
Pearson correlations and intraclass correlation (ICC--absolute agreement) were subsequently applied. The parameters analyzed were the aerobic capacity values of CP Hyp, CP Ԏ vs 1/t and CP P vs 1/t and EP. The anaerobic data was W' Hyp, W' Ԏ vs t and W' P vs 1/t from conventional CP, and WEP from 3MT. Repeated measures ANOVA and post hoc Newman-Keuls were used to compare the variables of aerobic and anaerobic capacity, mechanical data and acute physiological responses of heart rate (HR F) and peak blood lactate (LAC P) of all the assessment methods used. A significance level of p ≤ 0.05 was adopted in all hypothesis tests performed. The power was calculated using GPower software 3.1, and the effect size (ES) obtained in the statistical analysis was interpreted as suggested by Hopkins et al. \[[@pone.0192552.ref016]\]: ES \< 0.2 considered trivial, 0.2--0.5 considered small, 0.6--1.1 considered moderate, 1.2--1.9 considered large, and \> 2.0 considered very large. The Bland-Altman analysis (mean of differences ± 1.96\*SD) was used to identify the existence or not of differences between aerobic and anaerobic parameters of critical power in 3MT \[[@pone.0192552.ref017]\]. The percentage error associated with the prediction of EP and WEP was calculated for the three CP and W' models, respectively.
Results {#sec011}
=======
Mean values, standard deviation and CV% of aerobic and anaerobic data, effect size (ES), confidence interval (IC95%), as well as R^2^ of conventional critical power and 3MT applications, are shown in Tables [1](#pone.0192552.t001){ref-type="table"} and [2](#pone.0192552.t002){ref-type="table"}. The 3MT test's power output peak occurred in the first 7.3 ± 1.8 s, and this variable was stabilized at around 90 and 120 s of application.
10.1371/journal.pone.0192552.t001
###### Aerobic capacity parameters derived from the application of the 3MT (EP) and conventional CP tests (critical power hyperbolic model--CP Hyp; critical power work model versus time--CP Ԏ vs t; critical power model versus 1/time--CP P vs 1/t), R^2^ of mathematical equations, coefficient of variation (CV), confidence interval (IC95%), the error % associated with the prediction of EP (Error %) and effect size (ES) (n = 8).
{#pone.0192552.t001g}
---------------------------------------- -------- ------------ --------------- -----------------
**AEROBIC CAPACITY--POWER OUTPUT (W)**
**EP** **CP Hyp** **CP Ԏ vs t** **CP P vs 1/t**
**Mean** 181.7 178.2 191.4 188.3
**SD** 52 61 55 55
**IC 95%** 43.4 50.9 46 46
**R**^**2**^ 0.78 0.98 0.95 0.95
**CV%** 30.4 35.8 31.3 32.5
**Error %** \- 24.1 24.2 21.3
**EFFECT SIZE**
**EP** **CP Hyp** **CP Ԏ vs t** **CP P vs 1/t**
**EP** \- 0.06 0.18 0.12
**CP Hyp** 0.06 \- 0.22 0.17
**CP Ԏ vs t** 0.88 0.22 \- 0.23
**CP P vs 1/t** 0.12 0.17 0.23 \-
---------------------------------------- -------- ------------ --------------- -----------------
\*The significance criterion adopted was p ≤ 0.05.
10.1371/journal.pone.0192552.t002
###### Anaerobic capacity parameters derived from the application of the 3MT (WEP) and conventional CP tests (W' Hyp; anaerobic work capacity work model versus time--W' Ԏ vs t and anaerobic work capacity model versus 1/time--W' P vs 1/t), R^2^ of mathematical equations, coefficient of variation (CV), confidence interval (IC95%), the error % associated with the prediction of WEP (%Er-WEP) and effect size (ES) (n = 8).
{#pone.0192552.t002g}
---------------------------------- --------- ------------------------------------------- ------------------------------------------- -------------------------------------------
**ANAEROBIC WORK CAPACITY (kJ)**
**WEP** **W' Hyp** **W' Ԏ vs t** **W' P vs 1/t**
**Mean** 17.9 50.2[\*](#t002fn001){ref-type="table-fn"} 44.8[\*](#t002fn001){ref-type="table-fn"} 45.5[\*](#t002fn001){ref-type="table-fn"}
**SD** 4.8 15.3 8.7 8.4
**IC 95%** 4 12.8 7.2 7
**R**^**2**^ 0.78 0.98 0.95 0.95
**CV%** 27.8 33.3 19.9 18.8
**Error %** 245.1 204.5 182.5
**EFFECT SIZE**
**WEP** **W' Hyp** **W' Ԏ vs t** **W' P vs 1/t**
**WEP** \- 3.07 3.92 4.18
**W' Hyp** 3.07 \- 0.45 0.43
**W' Ԏ vs t** 3.92 0.45 \- 0.08
**W' P vs 1/t** 4.18 0.43 0.08 \-
---------------------------------- --------- ------------------------------------------- ------------------------------------------- -------------------------------------------
\* Significant difference in relation to WEP'
For aerobic capacity data, repeated measures ANOVA did not reveal statistical differences among CP Hyp, CP Ԏ vs t, CP P vs 1/t and EP. However, in the anaerobic capacity data, the statistical test indicated significant differences ([Table 2](#pone.0192552.t002){ref-type="table"}), with the same results to W' Hyp, W' Ԏ vs t and W' P vs 1/t, but higher if compared to the 3MT test's WEP.
In response to the increase in resistance, the hyperbolic relationship between strength or power output against time in conventional critical power test was observed, but this condition did not occur in the velocity variable ([Table 3](#pone.0192552.t003){ref-type="table"}) (for additional information see the Raw data in [S2 File](#pone.0192552.s002){ref-type="supplementary-material"}).
10.1371/journal.pone.0192552.t003
###### Mean and standard deviation of the power output, force, velocity and time limit variables of each predictive load from the conventional CP test (n = 8).
{#pone.0192552.t003g}
PREDICTIVE 6 ELASTICS PREDICTIVE 5 ELASTICS PREDICTIVE 4 ELASTICS PREDICTIVE 3 ELASTICS
--------------------- ------------------------------------------------ ------------------------------------------------ ------------------------------------------------ ------------------------------------------------
**Power Output(W)** 527.4±117[\*](#t003fn001){ref-type="table-fn"} 418.7±97[\*](#t003fn001){ref-type="table-fn"} 293.5±42[\*](#t003fn001){ref-type="table-fn"} 253.1±43[\*](#t003fn001){ref-type="table-fn"}
**Time Limit (s)** 164±68[\*](#t003fn001){ref-type="table-fn"} 242.5±102[\*](#t003fn001){ref-type="table-fn"} 453.5±137[\*](#t003fn001){ref-type="table-fn"} 628.6±138[\*](#t003fn001){ref-type="table-fn"}
**Velocity (m/s)** 2.57±0.3 2.46±0.3 2.06±0.2[^●^](#t003fn002){ref-type="table-fn"} 2.0±0.2[^●^](#t003fn002){ref-type="table-fn"}
**Force (N)** 191.3±27[\*](#t003fn001){ref-type="table-fn"} 162.3±27[\*](#t003fn001){ref-type="table-fn"} 136.5±18[\*](#t003fn001){ref-type="table-fn"} 119.1±11[\*](#t003fn001){ref-type="table-fn"}
\*Statistic difference from the other intensities (p ≤ 0.05)
^●^Statistic difference from 3 and 4 predictive elastics (p ≤ 0.05)
The physiological parameters (blood lactate and HR) of the applications of the 3MT test and CP predictive loads did not show significant differences ([Table 4](#pone.0192552.t004){ref-type="table"}). The sample power (1-β err prob) were 0.98 and 0.99 for aerobic and anaerobic parameters, respectively.
10.1371/journal.pone.0192552.t004
###### Mean ± standard deviation of physiological parameters of lactate peak (LAC P) and heart rate (HR PT) derived from the application of 3MT and predictive load tests to obtain conventional CP (predictive 3 elastics, predictive 4 elastics, predictive 5 elastics, predictive 6 elastics) (n = 8).
{#pone.0192552.t004g}
PHYSIOLOGICAL PARAMETERS
-------------------------------- ---------- ------ ------ ------ ------ ------
**LAC P (mmol·L**^**−1**^**)** **Mean** 11.2 10.5 10.6 13.2 11.7
**SD** 3.3 4.1 3.4 3 3.2
**HR PT (bpm)** **Mean** 177 180 179 179 179
**SD** 7 11 10 9 8
\*The significance criterion adopted was p ≤ 0.05.
[Table 5](#pone.0192552.t005){ref-type="table"} shows positive correlations (Pearson and absolute agreement ICC) for aerobic parameters, whereas positive correlation for anaerobic capacity was found only for those parameters obtained by the conventional CP test. The Bland Altman analyses of aerobic and anaerobic data from conventional critical power and 3MT applications are shown in Figs [2](#pone.0192552.g002){ref-type="fig"} and [3](#pone.0192552.g003){ref-type="fig"}, respectively. Linear regression among aerobic parameters of CP Hyp, CP Ԏ vs t, CP P vs 1/t and EP, and anaerobic parameters of W' Hyp, W' Ԏ vs t and W' P vs 1/t, compared to WEP are shown in [Fig 4](#pone.0192552.g004){ref-type="fig"}.
![Limits of agreement among aerobic parameters of CP Hyp, CP Ԏ vs t, CP P vs 1/t and EP through the Bland and Altman analysis \[[@pone.0192552.ref017]\].\
W, Watts. Diff, difference among values of aerobic parameters. A (EP---CP Hyp), B (EP- CP P vs 1/t), C (EP---CP Ԏ vs t), D (CP Hyp---CP P vs 1/t), E (CP Hyp---CP Work vs t) and F (CP Work vs t---CP P vs 1/t).](pone.0192552.g002){#pone.0192552.g002}
![Limits of agreement among aerobic parameters of W' Hyp, W' Ԏ vs t and W' P vs 1/t, compared to WEP through the Bland and Altman analysis \[[@pone.0192552.ref017]\].\
kJ, Kilojoule. Diff, difference among values of anaerobic parameters. G (WEP---W' Hyp), H (WEP--W' P vs 1/t), I (WEP--W' Work vs t), J (W' Hyp--W' P vs 1/t), K (W' Hyp--W' Work vs t) and L (W' work vs t--W' P vs 1/t).](pone.0192552.g003){#pone.0192552.g003}
{#pone.0192552.g004}
10.1371/journal.pone.0192552.t005
###### Results of Pearson correlation and absolute agreement ICC between aerobic and anaerobic parameters from the application of the 3MT (EP--WEP) and conventional CP tests (CP Hyp, CP Ԏ vs t and CP P vs 1/t---W' Hyp, W' Ԏ vs t and CP P vs 1/t) (n = 8).
{#pone.0192552.t005g}
----------------- ------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- -------------------------------------------
**AEROBIC PARAMETERS (W)**
**EP** **CP Hyp** **CP Ԏ vs T** **CP P vs 1/t**
**Pearson** **ICC** **Pearson** **ICC** **Pearson** **ICC** **Pearson** **ICC**
**EP** **R** -- -- 0.81[\*](#t005fn001){ref-type="table-fn"} 0.96[\*](#t005fn001){ref-type="table-fn"} 0.79[\*](#t005fn001){ref-type="table-fn"} 0.90[\*](#t005fn001){ref-type="table-fn"} 0.81[\*](#t005fn001){ref-type="table-fn"} 0.88[\*](#t005fn001){ref-type="table-fn"}
**CP Hyp** **R** 0.81[\*](#t005fn001){ref-type="table-fn"} 0.96[\*](#t005fn001){ref-type="table-fn"} -- -- 0.92[\*](#t005fn001){ref-type="table-fn"} 0.99[\*](#t005fn001){ref-type="table-fn"} 0.96[\*](#t005fn001){ref-type="table-fn"} 0.90[\*](#t005fn001){ref-type="table-fn"}
**CP Ԏ vs t** **R** 0.79[\*](#t005fn001){ref-type="table-fn"} 0.90[\*](#t005fn001){ref-type="table-fn"} 0.92[\*](#t005fn001){ref-type="table-fn"} 0.99[\*](#t005fn001){ref-type="table-fn"} -- -- 0.98[\*](#t005fn001){ref-type="table-fn"} 0.97[\*](#t005fn001){ref-type="table-fn"}
**CP P vs 1/t** **R** 0.81[\*](#t005fn001){ref-type="table-fn"} 0.88[\*](#t005fn001){ref-type="table-fn"} 0.96[\*](#t005fn001){ref-type="table-fn"} 0.90[\*](#t005fn001){ref-type="table-fn"} 0.98[\*](#t005fn001){ref-type="table-fn"} 0.97[\*](#t005fn001){ref-type="table-fn"} -- --
**ANAEROBIC PARAMETERS (kJ)**
**WEP** **W' Hyp** **W' Ԏ vs T** **W' P vs 1/t**
**Pearson** **ICC** **Pearson** **ICC** **Pearson** **ICC** **Pearson** **ICC**
**WEP** **R** -- -- 0.37 0.00 0.52 0.00 0.51 0.00
**W' Hyp** **R** 0.37 0.00 -- -- 0.76[\*](#t005fn001){ref-type="table-fn"} 0.00 0.90[\*](#t005fn001){ref-type="table-fn"} 0.68
**W' Ԏ vs t** **R** 0.52 0.00 0.76[\*](#t005fn001){ref-type="table-fn"} 0.00 -- -- 0.93[\*](#t005fn001){ref-type="table-fn"} 0.10
**W' P vs 1/t** **R** 0.51 0.00 0.90[\*](#t005fn001){ref-type="table-fn"} 0.68 0.93[\*](#t005fn001){ref-type="table-fn"} 0.10 -- --
----------------- ------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- ------------------------------------------- -------------------------------------------
\*The significance criterion adopted for the correlations was p ≤ 0.05.
Discussion {#sec012}
==========
The objective of this study was to investigate if 3MT is valid for obtaining CP and WEP on NMT in tethered running. The main finding of this research suggests that 3MT is a valid test for estimating CP aerobic parameters but not W'. Besides being performed in a single session, this research protocol maintains movement specificity while providing the instruments to acquire dynamic strength, speed and power data in runners, which are essential for running.
The feasibility of CP application was confirmed by the R^2^ values presented ([Table 1](#pone.0192552.t001){ref-type="table"}) \[[@pone.0192552.ref018], [@pone.0192552.ref019]\] and by the stipulated duration restriction of 2 to 15 minutes ([Table 3](#pone.0192552.t003){ref-type="table"}), as proposed by other authors \[[@pone.0192552.ref020], [@pone.0192552.ref021]\]. Regarding 3MT application, its reproducibility and validity were verified in other ergometers \[[@pone.0192552.ref004], [@pone.0192552.ref022], [@pone.0192552.ref023]\] and shown to be reliable for mechanical power output parameters in NMT \[[@pone.0192552.ref007]\]. The mean time for achieving a power output peak and stabilization values (90--120 s) of this research also corroborated with other studies that used 3MT in different ergometers \[[@pone.0192552.ref022], [@pone.0192552.ref023], [@pone.0192552.ref024], [@pone.0192552.ref025], [@pone.0192552.ref026], [@pone.0192552.ref027]\].
When comparing both applications, the responses of physiological markers of LAC P and HR PT ([Table 4](#pone.0192552.t004){ref-type="table"}) showed no significant difference between methodologies. The post-test lactate concentration indicates participation of anaerobic lactic metabolism \[[@pone.0192552.ref003], [@pone.0192552.ref028]\], which strengthens anaerobic measures, despite not assuring maximal effort.
The fact that the aerobic capacities of the 3MT and conventional CP on NMT were not statistically different ([Table 1](#pone.0192552.t001){ref-type="table"}) and showed a high-correlation value and mutual reliability ICC ([Table 5](#pone.0192552.t005){ref-type="table"}) ensures consistency of both applications and strengthens the ergometer model. Furthermore, the Bland Altman analyses corroborated with this assumption, with mean difference (MD) values close to zero ([Fig 2](#pone.0192552.g002){ref-type="fig"}). Additionally the fit presented in the linear regressions of PE in relation to the three CP models were acceptable ([Fig 4](#pone.0192552.g004){ref-type="fig"}). In this context, 3MT has the advantage of requiring only one test session in comparison to conventional CP. Additionally, in mathematical terms, this application minimizes conventional CP errors due to linear or non-linear mathematical treatment, since the variables are directly obtained only by recording the produced mechanical power \[[@pone.0192552.ref004]\]. Although the mean Error---EP is moderate (\~23.2%), this should not be a problem since they are different methodologies, since EP its directly derived from the mechanical power of the final 30s of the test and the values from the conventional CP are mathematical models. The other statistical results support the use of EP to estimate CP because they present consistency, reliability and correlation among each other. The Error---WEP (\~210.7%) value does not support the estimation of W' by such methodology, corroborating with the other statistical results (Figs [3G, 3H, 3I](#pone.0192552.g003){ref-type="fig"}, [4P, 4Q and 4R](#pone.0192552.g004){ref-type="fig"}). Since it is important to ensure spontaneity and specificity during this kind of evaluation \[[@pone.0192552.ref029], [@pone.0192552.ref030]\] and both applications on NMT tethered running are able to provide it, the use of 3MT seems to be more applicable due to the advantages mentioned above.
Until now, no study has analyzed physiological data derived from mechanical power output units through the application of CP or 3MT tests in tethered running. Furthermore, the anaerobic data used for this modality is typically quantified in units of anaerobic distance of the critical velocity test application \[[@pone.0192552.ref002], [@pone.0192552.ref008],[@pone.0192552.ref011]\]. However, recent studies have shown a positive correlation between runner power output and anaerobic capacity measured in NMT using an alternative Maximal accumulated oxygen deficit (MAOD) method, which is a test that has a consolidated physiological robustness \[[@pone.0192552.ref029]\]. Therefore, the power output measure seems to be appropriate for the anaerobic parameter because it considers the force application. Despite these benefits, WEP and W' showed poor agreement in the current investigation.
The CP concept refers to the tolerable duration of severe-intensity exercise and W' refers to the curvature constant of hyperbolic model \[[@pone.0192552.ref001]\]. Theoretically, W' and WEP reflect the amount of work that the individual is able to perform with ATP-CP and glycogen finite energy reserves \[[@pone.0192552.ref031], [@pone.0192552.ref032]\], and recent studies suggest that changes in CP influence W', since both may be affected by hypoxia and hyperoxia conditions \[[@pone.0192552.ref001], [@pone.0192552.ref033], [@pone.0192552.ref034]\]. Furthermore, load profile may affect WEP magnitude \[[@pone.0192552.ref033]\].
Previous studies discussed that high-level and homogenous individuals adopt a strategy that saves energy to use at the end of the test when performing submaximal tests \[[@pone.0192552.ref023], [@pone.0192552.ref035], [@pone.0192552.ref036]\], this is not the case in our sample, which consisted of active males with CV% in each test of around 30%. These characteristics may prevent the adoption of a strategy in this study, not allowing the EP to be overestimated and the WEP to be underestimated. Furthermore, an inspection of the power output's kinetics and other physiological variables, such as blood lactate concentration, was conducted as recommended in order to reduce this problem when using 3MT.
A factor that could influence the anaerobic component of critical power models is ecological validity. Galbraith et al. \[[@pone.0192552.ref008]\] and Triska et al. \[[@pone.0192552.ref037]\] discussed that differences between anaerobic parameters of critical power obtained in laboratory and in field are caused by maintenance of a fixed speed or cadence in the laboratory in contrast to spontaneous variation in the field. Furthermore, our study was performed in a NMT where runners could vary force and velocity at their own will. Possibly, 180 s of test is not sufficient to deplete anaerobic metabolism reserves \[[@pone.0192552.ref004]\] in 3MT, and tests involving maximum sprints usually disregard the initial acceleration phenomenon. A study using measurements of oxygen consumption and critical velocity applied on a treadmill reported that faster outputs accelerate the use of oxidative metabolism and seem to spare the runner's anaerobic energy \[[@pone.0192552.ref038]\].
In the original study of 3MT application versus linear models of CP in cycle ergometer, WEP was not statistically different from W' \[[@pone.0192552.ref004]\], but in six of ten participants, capacity data of the conventional model was considerably higher than that of the 3MT. The reference authors suggest that this actually happened due to methodological characteristics resulting from uncontrolled acceleration of 3MT on the cycle ergometer. They also indicate that there was complete exhaustion of anaerobic metabolism in 3MT, since the stabilization of oxygen consumption was similar in EP and on the application of a ramp test. Another study demonstrated that the W\' value does not change with prior exercise and effort at intensities above or below EP \[[@pone.0192552.ref026]\]. However, no similar results were found when using different cadences in isokinetic cycle ergometer, presenting dependence protocol relation \[[@pone.0192552.ref033]\]. Additionally, Bertram et al. \[[@pone.0192552.ref035]\] also found disagreement between both protocols' parameters while using six high-level cyclists that performed six maximum bouts in two days. Factors such as proper cadence and rest intervals between bouts may be important to the quality of 3MT parameters, but this is yet to be studied. Validity of anaerobic parameters of both tests seems debatable since the anaerobic reserves theory does not seem to be an independent fixed amount. Factors such as the initial acceleration, activity and amount of muscle fibers recruited seem to affect this amount of energy \[[@pone.0192552.ref004], [@pone.0192552.ref023], [@pone.0192552.ref033], [@pone.0192552.ref039]\].
A possible study limitation is the light frontal tilt of the runner's body as a result of treadmill friction, causing a difference in position of the body's center of mass. Additionally, the NMT used here does not account for the vertical force component. On the other hand, the ergometer allows spontaneous development of velocity and power output, which is not possible on a motorized treadmill.
It is important to clarify that the 3MT test was performed after accomplanished the conventional CP test. Considering that the prediction times for the protocol varied between 164--628 seconds, certainly the participants were quite familiar with non-motorized treadmill running exercise. Also, Vanhatalo et al. \[[@pone.0192552.ref007]\] suggests that the initial phase of acceleration during 3MT may interfere with WEP values. The choice of 10s sprints was aimed at avoiding the learning effect during the acceleration phase and adapting the subjects to the 3MT protocol. In fact, in the original article Vanhatalo et al. \[[@pone.0192552.ref007]\] performed the adaptation to the 3MT test, however, by the considerations above, it must not to be consided the lack of specific familiarization to the whole 3-minute test a limitation to the present study. Additionally, in a previous study of our group, the reliability of the 3MT test in NMT tethered running was demonstrated \[[@pone.0192552.ref007]\].
Conclusion {#sec013}
==========
In conclusion, the main finding of this research suggests that 3MT is a valid test estimating aerobic parameters of CP in a single application day but did not the anaerobic parameter of W'. Furthermore, we showed that the conventional application of CP and 3MT, when performed in tethered running on NMT, can be a viable methodology for obtaining mechanical power units in a specific ergometer for runners. The 3MT test has unquestionable advantages such as conclusion in a single 3-minute session and being able to provide aerobic and anaerobic physiologic parameters.
Supporting information {#sec014}
======================
###### This is the S1 Supplementary file.
The supplementary file provides details of the methods.
Details of the methods in S1 File. The details of the methods provide additional information about the ergometer and the mechanical and blood tests.
References in S1 File. The references present the essential manuscripts of methodological details.
(DOCX)
######
Click here for additional data file.
###### This is the S2 Raw data.
The raw data presents the raw results from 3MT and conventional critical power (n = 8).
Table A in S2 file. Mean ± standard deviation of physiological parameters of heart rate (HR PT) derived from the application of 3MT and predictive load tests to obtain conventional CP (predictive 3 elastics, predictive 4 elastics, predictive 5 elastics, predictive 6 elastics) (n = 8).
Table B in S2 File. Mean ± standard deviation of physiological parameters of lactate peak (LAC P) derived from the application of 3MT and predictive load tests to obtain conventional CP (predictive 3 elastics, predictive 4 elastics, predictive 5 elastics, predictive 6 elastics) (n = 8).
Table C in S2 File. Mean and standard deviation of the velocity of each predictive load from the conventional CP test (n = 8).
Table D in S2 File. Aerobic capacity parameters derived from the application of the 3MT (EP) and conventional CP tests (critical power hyperbolic model--CP Hyp; critical power work model versus time--CP Ԏ vs t; critical power model versus 1/time--CP P vs 1/t) (n = 8).
Table E in S2 File. Mean and standard deviation of the power output of each predictive load from the conventional CP test (n = 8).
Table F in S2 File. Mean and standard deviation of the force of each predictive load from the conventional CP test (n = 8).
Table G in S2 File. Mean and standard deviation of the time limit of each predictive load from the conventional CP test (n = 8).
Table H in S2 File. Anaerobic capacity parameters derived from the application of the 3MT (WEP) and conventional CP tests (W' Hyp; anaerobic work capacity work model versus time--W' Ԏ vs t and anaerobic work capacity model versus 1/time--W' P vs 1/t) (n = 8).
(XLSX)
######
Click here for additional data file.
This study was funded by the Coordination for Improvement of Higher Education Personnel (CAPES), process number 01P04517/2013, and State of São Paulo Research Foundation (FAPESP), process number 2009/08535-5. The authors thank Espaço da Escrita--Coordenadoria Geral da Universidade--UNICAMP--for the language services provided.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
###### Article summary
Article focus
=============
- The European Prospective Investigation of Cancer-Norfolk Eye Study is part of a European population-based cohort study, with participants now aged 48--92 years old.
- This paper describes the prevalence of small eyes, proportion with visual impairment, and associated demographic and biometric factors.
Key messages
============
- Ninety-six participants of 8033 (1.20%, 95% CI 0.98% to 1.46%) had an eye with axial length less than 21 mm, of which 74 (77%) were women.
- People with small eyes appear more likely to be blind or have unilateral visual impairment. Presence of a small eye is associated with shorter height, lower body mass index, higher systolic blood pressure and lower intraocular pressure.
- There are no standardised definitions for microphthalmos or nanophthalmos.
Strengths and limitations of this study
=======================================
- Large population-based study sample.
- The included population sample may have healthy volunteer bias.
- The identified associations are cross-sectional rather than longitudinal.
Introduction {#s1}
============
The small eye phenotype ranges from anophthalmos to nanophthalmos and microphthalmos. The latter two conditions are typically considered to be synonymous[@R1] and are subdivided into simplex[@R2] and complex[@R3] depending on the presence of other associated ocular or systemic abnormalities. There is minimal adult data on the prevalence of this phenotype with estimated birth prevalences for microphthalmos being 0.002--0.017%;[@R4] and 0.009% for microphthalmos in China from mass screening programmes.[@R5] Data from a hospital cohort suggest that patients with simple microphthalmos comprise between 0.05% and 0.11% ophthalmic patients.[@R6] There is great heterogeneity in the definition of nanophthalmos and microphthalmos which complicates interpretation of previous studies,[@R2] [@R7; @R8; @R9; @R10; @R11; @R12] with a definition by axial length (AL) \<21 mm being the most inclusive.[@R10; @R11; @R12] Nanophthalmos/microphthalmos is associated with angle closure glaucoma;[@R1] [@R13] and also with significant visual morbidity. In a recent series of nanophthalmic individuals from a Melanese population almost half had either unilateral or bilateral visual impairment.[@R11] There is a paucity of data for comparison. In view of this, we report data on the prevalence and characteristics of small eyes in British adults in the European Prospective Investigation of Cancer (EPIC)-Norfolk Eye Study, and review the definitions used for microphthalmos and nanophthalmos.
Method {#s2}
======
EPIC is a pan-European study that started in 1989 with the primary aim of investigating the relationship between diet and cancer risk.[@R14] The aims of the EPIC-Norfolk cohort were subsequently broadened to include additional endpoints and exposures such as lifestyle and other environmental factors.[@R15] The EPIC-Norfolk cohort was recruited between 1993 and 1997 and comprised 25 639 predominantly white European participants aged 40--79 years. The third health examination was carried out between 2006 and 2011 with the objective of investigating various physical, cognitive and ocular characteristics of participants then aged 48--92 years.[@R16] The third health examination was reviewed and approved by the East Norfolk and Waverney NHS Research Governance Committee (2005EC07L) and the Norfolk Research Ethics Committee (05/Q0101/191) and was performed in accordance with the principles of the Declaration of Helsinki. All participants gave written, informed consent.
All EPIC-Norfolk Eye Study participants underwent a detailed health examination performed by trained nurses following standard operating procedures. Ocular biometry was measured by non-contact partial coherence interferometry using the Zeiss IOLMaster Optical Biometer (IOLMaster, Carl Zeiss Meditech Ltd, Welwyn Garden City, UK). Five measurements of AL as well as anterior chamber depth (ACD, defined as corneal epithelium to anterior crystalline lens surface) and three measurements of central keratometry were made to allow calculation of mean values. Refractive error was measured using an autorefractor (Model 500, Humphrey Instruments, San Leandro, California, USA). Three intraocular pressure (IOP) measurements were made for each participant using the non-contact Ocular Response Analyzer (ORA, Reichert Inc, Depew, New York, USA) and the mean Goldmann correlated IOP (IOPg) calculated. Visual acuity was measured under standardised conditions at 4 m using participants' normal method of distance vision correction and recorded on the LogMAR scale. Fundal photographs were taken of both eyes using a TRC-NW65 non-mydriatic retinal camera (Topcon Corporation, Tokyo, Japan) with Nikon D80 camera (Nikon Corporation, Tokyo, Japan). A masked, expert grader from the Moorfields Grading Centre measured vertical cup--disc ratio (VCDR). Systolic and diastolic blood pressures (BPs) were taken from the right arm with the participant seated for 5 min. A stadiometer was used to record participant height to the nearest 0.1 cm and weight was measured to the nearest 0.1 kg using a body composition analyser (Tanita model TBF 300 s, Chasmors Ltd, London, UK). Self-reported data on education, occupation, alcohol intake and smoking status were recorded by questionnaire.
A small eye was defined by an AL of \<21 mm in at least one eye in keeping with the broadest previously accepted definition for microphthalmos/nanophtahlamos[@R10; @R11; @R12] and being equivalent to 2SD below the population mean value.[@R17] All investigations were performed on both eyes of each participant and the data from the eye with lower AL used for analyses at the participant level, with the exception of visual impairment classification where data from both were used. Visual impairment was defined by the presenting vision in accordance with the International Classification of Diseases Update and Revision 2006[@R18] and the WHO, which formally comprises categories 1--5 with categories 3--5 being blindness. To allow comparison with previous publications we defined blindness as a presenting visual acuity ≥1.3 logMAR in the better eye and low vision as a presenting visual acuity of \>0.48 in the better eye (ie, combination of moderate and severe visual impairment categories). Unilateral visual impairment was defined by using the eye with worse presenting visual acuity.
Statistical analysis was performed using SPSS V.20. Testing of normality was performed by the Kolmogorow-Smirnov method. Comparisons between participants with and without previous lens extraction were performed using the independent samples t test or Mann-Whitney U test. Logistic regression was used to identify factors associated with presence of a small eye and Fisher\'s exact test to compare presence of visual impairment with the presence of a small eye.
Results {#s3}
=======
Partial coherence interferometry data were available on 15 881 eyes of 8033 participants, of which 4442 participants were women (55.3%). Case numbers and overall prevalence values for small eyes stratified by AL value are shown in [table 1](#BMJOPEN2013003280TB1){ref-type="table"} and [figure 1](#BMJOPEN2013003280F1){ref-type="fig"}. Of the 8033 participants with AL data, visual acuity measurements were available on 8016 (99.8%).
######
Number of participants/eyes and overall prevalence values (with 95% CIs) by axial length (mm)
Axial length (mm) Analysis by participant Analysis by eyes
------------------- ------------------------- ------------------------- ----- -------------------------
\<21.00 96 1.195% (0.980 to 1.457) 132 0.831% (0.702 to 0.985)
\<20.50 47 0.585% (0.441 to 0.777) 57 0.359% (0.277 to 0.465
\<20.00 22 0.274% (0.182 to 0.414) 24 0.151% (0.102 to 0.225)
\<19.00 14 0.174% (0.105 to 0.292) 14 0.088% (0.053 to 0.148)
\<18.00 11 0.137% (0.077 to 0.245) 11 0.069% (0.039 to 0.124)
\<17.00 4 0.050% (0.020 to 0.127) 4 0.025% (0.010 to 0.065)
\<16.00 1 0.012% (0.003 to 0.069) 1 0.006% (0.002 to 0.035)
\<15.00 1 0.012% (0.003 to 0.069) 1 0.006% (0.002 to 0.035)
{#BMJOPEN2013003280F1}
Of the 96 participants, 20 were pseudophakic in both eyes, 6 were pseudophakic in one eye, 1 was aphakic in both eyes (congenital cataracts and nystagmus) and 1 aphakic in one eye and pseudophakic in the other. Defined by smallest eye, 26 participants had undergone previous lens extraction. Fourteen participants (15%) had a history of amblyopia or previous squint surgery. Seven participants (7%) had a history of previous laser iridotomy or surgical iridectomy. [Table 2](#BMJOPEN2013003280TB2){ref-type="table"} shows the demographic and biometric characteristics of those with AL \<21 mm.
######
Demographic and biometric data presented as mean values with (standard deviation and range min: max value (range for all participants only)), with \[median values, IQR\] shown for AL & ACD only
-----------------------------------------------------------------------------------------------------------------------
All, axial length \<21 mm Phakic Previous lens extraction p Value
-------------------------------- ------------------------------- ----------------- -------------------------- ---------
Number 96 70 26 --
Age (years) 69.0 (8.8, 50.9 to 89.2) 66.3 (7.5) 76.5 (7.6) \<0.001
Sex 22M/74F 13M/57F 9M/17F 0.11
AL (mm) 20.05 (1.26, 14.27 to 20.98)\ 20.45 (0.85)\ 18.96 (1.55)\ \<0.001
\[20.53, 0.80\] \[20.61, 0.48\] \[18.91, 2.85\]
ACD (mm) 2.94 (0.69)\ 2.67 (0.44)\ 3.75 (0.71)\ \<0.001
\[2.75, 0.78\] \[2.62, 0.44\] \[3.98, 0.92\]
Mean K (D) 45.24 (1.62, 41.71 to 51.19) 45.45 (1.65) 44.64 (1.41) 0.044
SE (D) +3.63 (2.94, −5.50 to +8.38) +5.04 (1.84) −0.15 (1.71) \<0.001
Anisometropia, (D) 1.13 (1.23, 0.00 to 6.76) 1.20 (1.27) 0.94 (1.09) 0.37
V/A (logMAR) 0.31 (0.47, −0.20 to 1.68) 0.37 (0.53) 0.16 (0.24) 0.061
LogMAR difference between eyes 0.31 (0.44, 0.00 to 1.82) 0.38 (0.49) 0.12 (0.19) 0.012
IOP (mm Hg) 15.7 (3.8) 15.6 (3.9) 16.0 (3.3) 0.63
-----------------------------------------------------------------------------------------------------------------------
Comparisons with p values are between phakic and those with previous lens extraction.
ACD, anterior chamber depth; AL, axial length; IOP, intraocular pressure.
Analysis of the difference in AL between eyes showed a bimodal distribution ([figure 2](#BMJOPEN2013003280F2){ref-type="fig"}) with 19 participants (20%) comprising the second peak with a mean AL difference of 5.63 mm (SD 0.97) compared with 77 participants in the first peak with mean AL difference of 0.45 mm (SD 0.39).
{#BMJOPEN2013003280F2}
Both univariable and multiple variable regression analyses investigating ocular biometric parameters in phakic eyes showed small eyes were associated with shallower ACD, steeper corneal keratometry and higher spherical equivalent (all p\<0.001, [table 3](#BMJOPEN2013003280TB3){ref-type="table"}). Separate analyses were performed for other, non-ocular biometric parameters. For these, univariable logistic regression analyses showed female sex (OR 2.75, p\<0.001), height (per 10 cm, OR 0.46, p\<0.001), weight (per 10 kg, OR 0.60, p\<0.001, body mass index (BMI, OR 0.68, p=0.005) and systolic BP (per 10 mm Hg, OR 1.11, p=0.029) were associated with the presence of a small eye. Multiple variable logistic regression models showed shorter height, lower BMI, higher systolic BP and lower IOP to be independent predictors of a small eye ([table 3](#BMJOPEN2013003280TB3){ref-type="table"}).
######
Univariable and multiple variable logistic regression analyses of factors associated with small eyes
OR 95% CI p Value
-------------------------------------------------------- ------ -------------- ---------
*(A) Ocular biometric parameters: phakic participants*
Univariable regression
Anterior chamber depth (per 1 mm) 0.06 0.03 to 0.12 \<0.001
Mean keratometry (per 1D) 2.16 1.82 to 2.57 \<0.001
Spherical equivalent (per 1D) 2.67 2.35 to 3.03 \<0.001
Multiple variable regression
Anterior chamber depth (per 1 mm) 0.02 0.01 to 0.08 \<0.001
Mean keratometry (per 1D) 5.97 3.98 to 8.98 \<0.001
Spherical equivalent (per 1D) 5.89 4.16 to 8.31 \<0.001
*(B) Other parameters: all participants*
Univariable regression
Age (per decade) 1.06 0.82 to 1.36 0.67
Female sex 2.75 1.70 to 4.43 \<0.001
Height (per 10 cm) 0.46 0.36 to 0.58 \<0.001
Weight (per 10 kg) 0.60 0.51 to 0.72 \<0.001
BMI (per 5 kg/m^2^) 0.68 0.52 to 0.89 0.005
Social class
Professional Ref
Managerial/technical 0.81 0.39 to 1.69 0.57
Skilled non-manual 0.91 0.40 to 2.09 0.82
Skilled manual 0.95 0.43 to 2.10 0.90
Partly-skilled 1.06 0.44 to 2.53 0.90
Unskilled 1.76 0.53 to 5.77 0.35
Education level
Less than O level Ref
O level 1.31 0.69 to 2.50 0.41
A level 0.94 0.57 to 1.55 0.81
Degree 0.93 0.50 to 1.76 0.83
Systolic blood pressure (per 10 mm Hg) 1.11 1.01 to 1.23 0.029
Diastolic blood pressure (per 10 mm Hg) 0.97 0.78 to 1.20 0.78
Self-reported alcohol intake
No intake Ref
\<7 units/week 0.81 0.48 to 1.37 0.43
≥7\<14 units/week 0.61 0.33 to 1.13 0.12
≥14\<21 units/week 0.61 0.28 to 1.33 0.22
≥21 units/week 0.70 0.38 to 1.28 0.25
Smoking status
Never Ref
Ever 0.85 0.56 to 1.27 0.41
Intraocular pressure (mm Hg) 0.95 0.90 to 1.01 0.09
Multiple variable regression
Age (per decade) 0.89 0.68 to 1.17 0.40
Female sex 0.91 0.47 to 1.77 0.77
Height (per 10 cm) 0.42 0.29 to 0.59 \<0.001
BMI (per 5 kg/m^2^) 0.69 0.53 to 0.90 0.006
Systolic blood pressure (per 10 mm Hg) 1.11 1.01 to 1.22 0.030
Intraocular pressure (mm Hg) 0.93 0.88 to 0.99 0.030
Ref: reference category. For the multiple variable regression models (either A or B), only parameters reaching statistical significance in the respective univariable analysis were included, and only those in the final model shown.
BMI, body mass index.
Optic disc grading was possible on both eyes of 61/96 (64%) participants and at least one eye of 82/96 (85%) participants (right eyes: 12 missing, 9 ungradable; left eyes: 14 missing, 14 ungradable). Three participants (3/61, 4.9%) had VCDR asymmetry of 0.2 or more, and one additional participant had an optic disc consistent with glaucoma (localised absence of neural rim, one eye only), giving an overall prevalence of 4/61 (6.6%, 95% CI 2.6% to 15.7%) for glaucomatous optic neuropathy. No eye had a VCDR of ≥0.6. Five of 96 (5.2%) participants gave a diagnosis of 'glaucoma' in their medical history, of these only one had a diagnosis consistent with their optic disc photographs. Three participants had one optic disc with disc drusen. There were no cases of macular hypoplasia, macular schisis, coloboma or any other retinal abnormality associated with nanophthalmos.
Visual acuity data were available for all 96 participants and values are shown in [table 2](#BMJOPEN2013003280TB2){ref-type="table"}. One participant (1%) was classified as blind by the WHO definition (visual acuity of less than 1.3 logMAR) and 2/96 (2.1%) had any degree of visual impairment. Using a definition of visual impairment of \>0.30 logMAR in the better eye to allow comparison with previous visual impairment studies, the prevalence was 5/96 (5.2%). The prevalence of blindness was significantly higher in EPIC-Norfolk participants with at least one eye of AL \<21 mm compared with those without, while the overall prevalence of low vision was similar ([table 4](#BMJOPEN2013003280TB4){ref-type="table"}). Unilateral visual impairment by all definitions was more common in EPIC-Norfolk participants with at least one small eye compared with those without (p≤0.001, [table 4](#BMJOPEN2013003280TB4){ref-type="table"}).
######
Percentages of bilateral and unilateral visual impairment in participants with one or both eyes with axial length \<21 mm (n=96) compared with all EPIC-Norfolk participants with no eye of axial length \<21 mm (n=7920) by Fisher\'s exact test
LogMAR Snellen equivalent Classification EPIC-Norfolk participants without small eyes (n=7920 total) EPIC-Norfolk participants with small eyes (n=96 total) p Value
------------------- -------------------- --------------------------------------------------------------- ------------------------------------------------------------- -------------------------------------------------------- --------- -------------------- ---------
\>1.30 better eye \<3/60; 20/400 WHO blindness 2 0.03% (0.00, 0.06) 1 1.0% (0.0, 3.1) 0.036
\>0.48 better eye \<6/18; 20/60 Blindness and visual impairment; 'low vision' 45 0.6% (0.4, 0.7) 2 2.1% (0.0, 5.0) 0.11
\>0.22 better eye \<6/10; 20/32 UK driving standard 422 5.3% (4.8, 5.8) 7 7.29% (2.0, 12.56) 0.36
\>0.30 better eye \<6/12; 20/40 Previous visual impairment studies, American driving standard 259 3.3% (2.9, 3.7) 5 5.2% (0.7, 9.7) 0.25
\>1.0 worse eye \<6/60; 20/200 Unilateral visual impairment 120 1.5% (1.3, 1.8) 11 11.5% (5.0, 18.0) \<0.001
\>0.48 worse eye \<6/18; 20/60 Unilateral visual impairment 470 5.9% (5.4, 6.5) 24 25.0% (16.2, 33.8) \<0.001
\>0.30 worse eye \<6/12; 20/40 Unilateral visual impairment 1341 16.9% (16.1, 17.8) 29 30.2% (20.9, 39.6) 0.001
Bilateral visual impairment is defined as both eyes with a visual acuity less than the respective value and unilateral visual impairment as one eye with a visual acuity less than the respective value.
EPIC, European Prospective Investigation of Cancer.
Discussion {#s4}
==========
There are minimal data describing the prevalence or characteristics of small eyes. In the EPIC-Norfolk Eye Study, the prevalence of a participant with an eye of AL of \<21 mm was 1.20%, and 0.27% for those with an eye of AL \<20 mm. Relative to existing data with estimated birth prevalences for microphthalmos between 0.002% and 0.017%;[@R4] [@R5] and the prevalence of simple microphthalmos in hospital ophthalmic patients being between 0.05% and 0.11%;[@R6] small eyes are more common than previously reported.
Nanophthalmos is traditionally associated with a high prevalence of angle closure glaucoma[@R1] [@R13] and both nanophthalmos and primary angle closure glaucoma have similar ocular phenotypes including a short AL, shallow anterior chamber, hyperopia, small radius of corneal curvature and a thick crystalline lens. In this study, the prevalence of glaucomatous optic neuropathy was estimated to be 6.6% in small eyes based on CDR, CDR asymmetry or rim abnormalities consistent with glaucoma. Othman *et al*[@R19] reported that 12/22 (55%) nanophthalmic individuals had occludable anterior chamber angles or glaucoma. In the case series by Tay *et al*[@R11] of 17 individuals with nanophthalmos, no data on glaucoma prevalence are reported. For comparison, the prevalence of glaucoma (open angle and closed angle combined) has been estimated to be 2.4% in European populations.[@R20] We did not calculate VCDR percentiles for the overall EPIC cohort as per the International Society Geographical & Epidemiological Ophthalmology definition of glaucoma;[@R21] however, no small eye in our series had a VCDR of ≥0.60. Crowston *et al*[@R22] reported optic disc size adjusted VCDR percentiles from the Blue Mountains Eye Study and showed that for small discs (1.2 mm vertical diameter) in non-glaucomatous eyes, the 97.5th centile for VCDR was 0.60 (99th centile: 0.62) while corresponding values for large optic discs (1.9 mm diameter) were 0.75 and 0.83, respectively. Thus our prevalence of 6.6% should be considered as a minimum prevalence estimate for glaucoma in small eyes, and is likely to include those at highest risk for visual impairment over their lifetime.[@R21]
Review of recent studies reporting on nanophthalmos/microphthalmos shows great heterogeneity in the definitions used, with cases defined primarily by short AL with for example, values of \<21,[@R10; @R11; @R12] \<20.9,[@R2] \<20.5,[@R7] \<20,[@R8] \<18,[@R2] or \<17 mm.[@R9] The original description of nanophthalmos (or pure microphthalmos) by Duke-Elder[@R1] is an eye 'reduced in volume without the presence of other gross congenital abnormalities,' 'typical dimensions are 16--18.5 mm sagittal,' 'hyperopia is the rule' and 'the anterior chamber is typically shallow.' The partial relaxation of the definition to its currently accepted form (of at least an AL of \<21 mm) is likely due to the rarity of the condition. If an abnormally short eye is defined based on the lower 2SD and 3SD limits of mean population ALs, then the calculated limits are approximately 21 and 20 mm, respectively.[@R16] In a previous study by our group investigating complications in small eyes (\<21 mm) undergoing phacoemulsification and lens implantation, only AL and the presence of abnormal IOP remained significant predicators of any complication in multiple variable regression analysis.[@R12] Complications were 15 times more likely in cases with AL of \<20 mm (compared with those 20--21 mm, p≤0.001). The differential complication rate supports the previous recommendation by Weiss *et al*[@R2] that microphthalmos and nanophthalmos should be considered as two separate phenotypes, based on AL. Based on the above it would appear reasonable to classify small eyes into microphthalmos (\<21 mm) and nanophthalmos (\<20 mm), respectively.
We found 1% of participants with small eyes were blind and 2% had low vision. When compared with all EPIC-Norfolk participants, blindness appears to be more common in those with a small eye (p=0.036, although case numbers were very low), but low vision was not (p=0.36). Unilateral visual impairment (defined by the worse seeing eye) was more common by all definitions (p≤0.001). When compared with data from population studies, the prevalence of visual impairment in EPIC-Norfolk participants is low overall, and values in those with small eyes are again low or similar. In the Blue Mountains Eye Study,[@R23] 4.6% had a visual acuity of 6/12 (20/40) or less in the better eye and 14.4% had a visual acuity of 6/12 or less in their worse eye, whereas in our cohort of small eyes the equivalent percentages were 5.2% and 30.2%. In the Salisbury Eye Study, 9% of participants aged 75--84 years old had a visual acuity of \<6/12 in their better eye;[@R24] while in the MRC study in Britain this value was 15% for those 75--84 years old.[@R25] There is minimal data on visual impairment in nanophthalmic individuals, with a recent study in a Melanesia population[@R11] (definition: AL usually \<21 mm in at least one eye) reporting 5/17 (29%) had bilateral visual impairment and 9/17 (53%) had unilateral visual impairment (defined as \<6/12 (20/40) Snellen in the better eye).
We found a marked bimodal distribution in AL difference between eyes in individuals with small eyes, with 20% individuals having \>3.5 mm AL asymmetry. A bimodal distribution in AL difference has not previously been described, with Weiss *et al*[@R2] reporting a difference of only 0.4 mm or less in a series of 21 patients with simple microphthalmos.
Our study has a number of limitations; these being primarily the absence of lens thickness and scleral thickness data to further characterise participants with small eyes. Additionally participants were not examined on a slit-lamp, for example, gonioscopy to determine the presence of an occludable anterior chamber angle (and therefore to determine if the glaucomatous optic neuropathy in our five cases were in the presence of an open or closed anterior chamber angle). Our prevalence value for glaucoma was based on glaucomatous optic neuropathy only rather than glaucomatous optic neuropathy and visual field defect.[@R21] Comparisons of visual acuities were only performed in EPIC-Norfolk participants in whom ALs were measurable, and consequently this may have excluded those with visual impairment or blindness where AL could not have been measured optically (ie, underestimating prevalence values). Additionally, those with visual impairment may have self-selected not to participate in the EPIC-Norfolk Eye Study, thus again underestimating case numbers.
In summary, the small eye phenotype was more common than previously reported, and our study provides prevalence values in British adults. There are no standardised definitions for microphthalmos or nanophthalmos; however, based on current evidence, subdivision by AL of \<21 mm for microphthalmos and \<20 mm for nanophthalmos appears reasonable. People with small eyes appear more likely to be blind or have unilateral visual impairment. The estimated prevalence of glaucomatous optic neuropathy in our cohort appeared to be lower than expected and warrants further investigation.
Supplementary Material
======================
###### Author\'s manuscript
###### Reviewer comments
**Contributors:** ACD performed the data analysis and drafted the manuscript. APK and TP contributed to the data analysis. SH and DCB contributed to the conception and design of the study. RL contributed to the design of the study and to the data acquisition and management. K-TK and PJF contributed to the conception and design of the study, and to the data interpretation. All authors read and critically revised the manuscript and approved the final manuscript.
**Funding:** Supported by grant G0401527 from the Medical Research Council, UK and grant 262 from Research into Ageing, UK. APK is funded by a Wellcome Trust Clinical Research Fellowship. PJF was also supported by the Richard Desmond Charitable Trust (via Fight for Sight, grant 1956). ACD, TP & PJF were supported by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
**Competing interests:** None.
**Ethics approval:** Norwich Local Research Ethics Committee (05/Q0101/191) and East Norfolk & Waveney NHS Research Governance Committee (2005EC07L).
**Provenance and peer review:** Not commissioned; externally peer reviewed.
**Data sharing statement:** The data sharing and preservation strategy in EPIC-Norfolk is in accordance with the Wellcome Trust data management and sharing policy. Full details about the study including contact information are on the website <http://www.epic-norfolk.org.uk>. Investigators wishing to work with EPIC data contact the EPIC management group through the website, letter, phone or fax and proposals have to fulfil a number of criteria including that the work is within the bounds of consent given by participants and has been ethically reviewed and approved; there is no serious risk to the viability of continuing the cohort study, for example, through offence to the participants from use of the data supplied; the science of the proposal has been satisfactorily peer reviewed and the proposal does not duplicate work already being done. Access to data for collaborators is provided through password protected website access. The large numbers of collaborators EPIC-Norfolk has locally, nationally and internationally (\>300), as evidenced by collaborative publications, demonstrate the commitment to maximising the value of the study.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Parotidectomy is a commonly employed surgical procedure to remove parotid gland tumors of certain types. The surgical approach can be anterograde or retrograde in type. The former aims to initially identify the facial nerve main trunk, and then trace it forward to dissect the tumor out, whereas the latter aims at finding the peripheral branches of the facial nerve first, and then trace them backward to remove the tumor.
Landmarks for retrograde parotidectomy vary depending on the chosen facial nerve peripheral branches. The retromandibular vein is a commonly used landmark for the marginal mandibular branch of the facial nerve. The angle of the mandible has also been suggested as a landmark for the marginal mandibular branch in some literature, due to its proximity to the retromandibular vein. The buccal branch is another facial nerve branch used by surgeons. However, anatomical studies in this area are still relatively poor.
Early intra-operative identification and preservation of the nerve is of paramount importance in this surgical practice, because facial nerve injury is an important cause of morbidity.
2. Materials and methods {#sec2}
========================
This study specifically compares the distances of chosen anatomical landmarks to corresponding facial nerve branches between: male and female specimens; left and right sides of the specimens; sides with occlusal jaws and sides with non-occlusal jaws. If more than one premolar or one molar tooth from the maxilla are lined up with the corresponding teeth from the mandible, the specimen is defined as having an occlusal maxilla and mandible.
The following distances are chosen and specified in [Table 1](#tbl1){ref-type="table"} and [Fig. 1](#fig1){ref-type="fig"}:
Data were collected using prosected specimens from the anatomy labs of both the University of New South Wales (UNSW) and the University of Sydney, where all landmarks can be vividly appreciated. In addition, the number and presence of teeth were recorded to determine whether the premolars or molars of the maxillae and mandibles came into contact to form occlusal surfaces. Approval was obtained from the UNSW Human Research and Ethics Committee (HREC09372).
Measurement was taken three times for each specimen using the same digital micrometer (Mitutoyo Digimatic Caliper; Mitutoyo Corp., Kawasaki, Japan), and an average was calculated. Each half head was extended to 120°. This angle is defined as the obtuse angle between the red and blue lines ([Fig. 2](#fig2){ref-type="fig"}). The red line is defined from the lateral angle of the eye to the tragus cartilage, and the blue line is down the transverse processes of the cervical vertebrae. The mouths of specimens were closed during measurement. A grand total of 41 superficial dissected half-head specimens were collected for part two. The ethnicity of all specimens is Australian of European descent, and the mean age of the specimens is 81 years old.
Statistical analyses were undertaken using SPSS (20.0) (IBM Corp., Armonk, NY, USA). To determine the differences between the different groups aforementioned, particular tests were applied ([Table 2](#tbl2){ref-type="table"}).
3. Result {#sec3}
=========
There is no statistically significant difference found in all three distances when comparing respectively: genders; left/right; amongst different edentulous groups; sides with opposing upper and lower jaws and sides with non-opposing upper and lower jaws. ([Table 3](#tbl3){ref-type="table"}, [Table 4](#tbl4){ref-type="table"}, [Table 5](#tbl5){ref-type="table"}).
4. Discussion {#sec4}
=============
For retrograde parotidectomy, the targets are the peripheral branches of the facial nerve. There are two types of studies concerning the variations of the landmarks to the facial nerve peripheral branches. The first type focused on describing anatomical relationships according to cases. For instance, Hanazawa et al. found 2 cases of anomalous facial nerve relationships after examining 105 surgical specimens of parotid gland tumor in 1988 [@bib1]. In one case the buccal branch and the zygomatic branch of the facial nerve coursed beneath the retromandibular vein; and in the other case the marginal mandibular branch ran beneath the retromandibular vein [@bib1]. Similarly, the study by Kawakami, Tsukada, and Taniguchi in 1994 exposed 26 facial nerves and discovered one anomalous relationship, such that the temporal branch of the facial nerve lay beneath the retromandibular vein [@bib2]. However, the frequency of this type of study has greatly reduced over the past decade, as the possibility of discovering new cases with anomalous relationships has significantly reduced.
The second type of study investigated and analysed anatomical variations by categorizing them. For instance, Laing and McKerrow illustrated their findings by categorizing the patterns of the relationship between the facial nerve branches and the retromandibular vein into 5 different types, and then calculating the percentage of each type [@bib3]. However, this type of study is of more academic anatomical interest, rather than being of practical surgical relevance, on which future research should focus.
Retrograde parotidectomy approaches the facial nerve main trunk retrogradely through one of the branches of the facial nerve that may be found at predictable sites, for instance, the marginal mandibular nerve of the facial nerve. While current surgical practice for parotidectomy is dominated by the anterograde approach, the retrograde surgical technique is employed in situations where the position of the tumor may obscure both the facial nerve main trunk and the various landmarks used in anterograde parotidectomy are not easily accessible intraoperatively [@bib4]. Furthermore, in circumstances where reoperation is required, the usual surgical landmarks for the facial nerve main trunk become useless, because scar tissue, inflammation, or even recurrent disease may have distorted the surgical field [@bib5]. In addition, retrograde parotidectomy reduces the amount of normal parotid gland that needs to be removed, and may permit a more conservative approach than standard parotidectomy, while having similar complication rates and surgical effectiveness [@bib6]. Hence, studying the variation of landmarks used in retrograde parotidectomy is as important as studying those used in anterograde parotidectomy.
However, from a surgical point of view data are very poor in the literature concerning the variations of these landmarks. Most analyses are of purely anatomical interest with no clinical application [@bib3].
In the present study, three anatomical relationships have been examined but no statistically significant result was found, indicating that these relationships (the distance of the buccal branch of the facial nerve to the distal end of the zygomatic arch; the distance of the marginal mandibular branch of the facial nerve to the retromandibular vein; the distance of the marginal mandibular branch of the facial nerve to the angle of the mandible) are relatively consistent. More importantly, since three landmarks of different positions have been studied in the present study, the surgeons now have more options for use in surgery, which is particularly useful in cases where one landmark has been obscured by the tumor mass.
In addition, the shortest distance is from the angle of the mandible to the marginal mandibular branch of the facial nerve, followed by the retromandibular vein to the marginal mandibular branch, and the distal end of the zygomatic arch to the buccal branch ([Table 3](#tbl3){ref-type="table"}, [Table 4](#tbl4){ref-type="table"}, [Table 5](#tbl5){ref-type="table"}). Therefore, it can be concluded that the angle of the mandible is the preferred landmark in retrograde parotidectomy, given that this osseous structure can be vividly appreciated by light palpation on the face.
5. Conclusion {#sec5}
=============
This cadaveric study has reviewed various anatomical landmarks for retrograde parotidectomy. It added the results of the first quantitative research into this field, which has given surgeons a more accurate reference. Surgeons therefore should be mindful about the variations of some landmarks found this study, not only the variations of them between genders, sides or different levels of dentitions, but also the variations of them at different stages of the surgery.
Ethical approval {#sec6}
================
Yes. Approval was obtained from the UNSW Human Research and Ethics Committee (HREC09372).
Sources of funding {#sec7}
==================
There is no source of funding.
Author contribution {#sec8}
===================
Wenjie Zhong is the main person who conducted this research, including literature review, data collection, data analysis as well as manuscript completion. Ken Ashwell improvised research topic and assisted in final editing of the manuscript.
Conflict of interest {#sec9}
====================
The authors have no conflict of interest to declare.
Trial registry number {#sec10}
=====================
n/a.
Guarantor {#sec11}
=========
Wenjie Zhong.
Research Registration Unique Identifying Number (UIN) {#sec12}
=====================================================
researchregistry1000.
We would like to thank Dr. Priti Pandey and Dr. Nalini Pather for their advice during this study. We would also like to thank the staff from the gross anatomy laboratories of both the University of New South Wales and the University of Sydney, in particular Mr. Vincent Strack and Mr. Marcus Robinson.
{#fig1}
{#fig2}
######
The exact descriptions of the landmarks and the distances measured in this study.
Distance Description Reason
---------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------
1 The point at the distal end of the zygomatic arch just before it becomes the frontal process of the zygoma to the more superior buccal branch just after its emergence from the parotid gland. This is easily palpated without much surgical exploration.
2 The most superior point of the retromandibular vein just after its emergence below the parotid gland to the marginal mandibular branch of the facial nerve just after its emergence from the parotid gland. This is the most commonly used landmarks and it is easily found with superficial dissection.
3 The point at the angle of the mandible just before it becomes the ramus of the mandible to the marginal mandibular branch of the facial nerve just after its emergence from the parotid gland. This is commonly used by surgeons and can be easily palpated.
######
Statistics tests applied for data analysis in this study.
Statistics analysis To test if there is no significant difference in the distance of the chosen anatomical landmarks to the corresponding facial nerve branches between:
--------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------
Two sample T test •male and female specimens•left and right specimens•sides with occlusal jaws and sides with non-occlusal jaws
######
Results from two samples T test comparing the means of these distances (in mm) between genders. M = Male; F = Female.
Gender No. Mean Std. deviation Max. Min. Mean difference P value
------------ -------- ------ ------ ---------------- ------ ------ ----------------- ---------
Distance 1 M 23 35.8 6.2 54.8 26.9 0.5 0.849
F 7 35.3 4.2 40.0 27.2
Distance 2 M 23 10.2 5.6 19.3 1.9 −2.2 0.412
F 7 12.4 7.1 24.8 1.5
Distance 3 M 23 8.9 3.0 17.6 4.1 2.1 0.129
F 7 6.8 3.5 14.3 3.2
######
Results from two samples T tests comparing the means of these distances (in mm) between left and right sides.
Half No. Mean Std. deviation Max. Min. Mean difference P value
------------ ------ ------ ------ ---------------- ------ ------ ----------------- ---------
Distance 1 L 20 35.1 6.5 54.8 26.9 0.3 0.892
R 19 34.8 5.3 45.9 26.9
Distance 2 L 20 11.1 5.5 24.8 14.3 −0.2 0.902
R 20 11.3 6.5 21.6 1.5
Distance 3 L 20 7.3 2.8 14.3 3.9 −1.5 0.177
R 20 8.8 3.9 18.2 4.3
######
Results of comparing the means of these distances (in mm) between two groups: sides with occlusal upper and lower jaws and sides with non-occlusal upper and lower jaws. O = occlusal, N = non-occlusal.
Group No. Mean Std. deviation Min Max Mean difference P value
------------ ------- ------ ------ ---------------- ------ ------ ----------------- ---------
Distance 1 O 11 34.8 6.0 26.9 45.7 −0.3 0.891
N 29 35.1 5.9 26.9 54.8
Distance 2 O 12 11.7 6.1 3.4 21.4 0.9 0.687
N 29 10.8 6.0 1.5 24.8
Distance 3 O 12 8.6 3.5 4.1 18.2 0.8 0.501
N 29 7.8 3.4 3.9 17.6
| {
"pile_set_name": "PubMed Central"
} |
1.. Introduction {#sec1}
==================
X-ray diffraction (XRD) is a well established technique to analyse the crystallographic structure of bulk materials and layers. (100)-oriented Si is the most frequently used substrate in microelectronics. To a lesser extent, (111)-oriented Si also finds application. One standard technique to detect and to study crystalline layers on such substrates is specular X-ray diffraction by ω--2Θ scans using commercially available diffractometers, usually equipped with a Cu anode. Typically, the sample alignment is done using the Si 400 or 111 reflection in such a way that during the measurement the diffraction vector is always parallel to the Si \[100\] or \[111\] direction, respectively.
Such measurements very often show a peak of the Si 200 reflection at about 2Θ = 33° for Si(100) substrates and always the Si 222 reflection at about 2Θ = 58.9° for Si(111) substrates, although both reflections are known as basis-forbidden reflections. Even though the reason for the occurrence of both reflections has been known for many decades, especially concerning the Si 200 reflection daring explanations can still be found in recent publications: for example that it occurs only when the Si lattice is distorted (Zhao *et al.*, 2005[@bb14]) or that it stems from the Si 400 Bragg reflection of half the wavelength of the Cu *K*α radiation (Hesse *et al.*, 2005[@bb3]). The origin of the appearance of forbidden reflections was discovered and theoretically described by Renninger in 1937 (Renninger, 1937[@bb7]). For a well defined in-plane orientation of the sample (typical Φ rotation on modern diffractometers), multiple diffraction may occur, which makes a forbidden reflection visible or increases or decreases the intensity of allowed reflections depending on Φ. For the theoretical description of this phenomenon, also known under the German word *Umweganregung*, see for example Rossmanith (2000[@bb11]) and references therein.
In particular, synchrotron radiation in combination with the forbidden Si or Ge 200 reflection is nowadays used to analyse problems such as the core structure of defects in Si (Richard *et al.*, 2007[@bb9]), the defect type in patterned Ge/Si islands (Richard, Schülli & Renaud, 2011[@bb10]; Richard, Malachias *et al.* 2011[@bb8]) and nanostructures (Kozlowski *et al.*, 2012[@bb4]), and the atomic ordering in Ge/Si islands (Malachias *et al.*, 2010[@bb6]), to mention only a few examples.
Nevertheless, it seems to be worth discussing the behaviour of the Si 200 reflection for a conventional diffractometer in more detail. Since the in-plane orientation of a sample for specular ω--2Θ measurements is usually not of interest, nobody cares in this sense about obtaining an exact and repeatable sample orientation. As a result of this, the appearance of the 200 peak differs significantly from measurement to measurement. As well as the fact that the intensity of the rather sharp peak may vary over orders of magnitude down to the background level, sometimes there are broad shoulders or even subpeaks visible on both sides. Fig. 1[▶](#fig1){ref-type="fig"} demonstrates this on one example. In this case, four Si(100) wafers with Ni layers deposited under different conditions were investigated with a standard X-ray diffractometer. The samples were placed on the stage considering the wafer notch such that they were roughly (±2.5°) oriented in a \[110\] direction. For the subject of this paper, only the range near the Si 200 peak is of interest. Broad side peaks and varying intensity of the Si 200 peak are observed and marked by arrows. While the varying intensity can be explained by the Φ dependence of multiple-diffraction phenomena, the question arises as to the reason for the broad shoulder or side peaks. Ni itself has no diffraction peak in this angular range and thus cannot be responsible for the modified Si 200 reflection, and another material should not be on the Si surface.
The most reasonable explanation is that these additional structures are also results of multiple diffraction in the Si substrate. If this is true, then the question arises, how do we distinguish between substrate-related and real layer peaks? In the case of CoSi~2~ on Si(100) (Londergan *et al.*, 2001[@bb5]) the detection of a clear CoSi~2~ 400 reflection confirms the existence of a (100)-oriented CoSi~2~ layer, while the identification of a peak at about 2Θ = 33° for Si 200 and CoSi~2~ 200 alone would not be sufficient. More questionable is the situation in a Gd~2~O~3~ growth study by Chaudhuri *et al.* (2014[@bb2]), where a broad peak at about 2Θ = 32.7° is marked as Gd~2~O~3~ 400 without any confirmation by a higher diffraction order.
In this paper, the intensity distribution in a limited angular range around the Si 200 and Si 222 reflections will be studied in detail as a function of the in-plane orientation Φ. It will be shown that the divergence of the used diffractometer perpendicular to the diffraction plane and the wavelength distribution in the incident beam have a major influence. As a conclusion of this study, a recommendation will be given for how to prove the existence of a layer peak in the vicinity of the Si 200 reflection.
2.. Experimental details {#sec2}
==========================
Standard industrial Si wafers of 100 mm diameter with (100) and (111) orientation, respectively, were used to demonstrate the behaviour of the Si 200 and Si 222 diffraction on 'perfect' samples.
XRD measurements were performed in parallel beam configuration on two different SmartLab diffractometers from Rigaku. The first, representing the low-resolution case, was a conventional SmartLab with a 9 kW rotating Cu anode, line focus, mirror (acting in the diffraction plane only), and 5° Soller slits on the source and detector side that define the beam divergence perpendicular to the diffraction plane. Besides the Bragg--Brentano geometry, this diffractometer configuration is typical for specular XRD measurements to analyse thin layers on Si substrates by ω--2Θ scans. The second configuration, representing the high-resolution case, was a SmartLab µHR with a 0.8 kW rotating Cu anode, micro focus, confocal Max-Flux optics and no additional Soller slits. The two mirrors of the confocal Max-Flux optics generate a beam divergence of about 0.04° not only in the diffraction plane but also perpendicular to it. With this diffractometer, supplemental measurements were performed with a Ge(400) × 2 crystal collimator added in the beam path to reduce the wavelength distribution of the incident beam nearly to the Cu *K*α~1~ line only.
Samples were aligned on an *Rx*--*Ry* stage by using the 400 or 111 reflection, respectively, so that either the normal of the (100) netplane or the normal of the (111) netplane is exactly parallel to the Φ rotation axis of the diffractometer. A conventional scintillation counter is used for all measurements.
3.. Results and discussion {#sec3}
============================
3.1.. Si 200 reflection {#sec3.1}
-------------------------
The easiest way to investigate the Si 200 reflection is to select after a sample alignment on the 400 reflection its ω--2Θ position at 2Θ = 32.98° and to perform Φ scans. Such measurements are shown in Fig. 2[▶](#fig2){ref-type="fig"} for both diffractometer configurations. The curve obtained in high-resolution mode (Fig. 2[▶](#fig2){ref-type="fig"} *a*) indicates the occurrence of very sharp peaks reaching intensities of 10^5^--10^6^ counts per second (c.p.s.), starting from a background level of about 10 c.p.s. For comparison, the intensity reached for the 400 reflection in the same configuration is of the order of 10^7^ c.p.s. The observed background level was confirmed by ω--2Θ scans between 28 and 38° to be the general background arising from the sample under the given experimental conditions and is not especially related to the angular position of the Si 200 reflection. This indicates that Si 200 is really a forbidden reflection that shows no detectable intensity except for certain Φ orientations, where multiple diffraction (*Umweganregung*) occurs. The observed peaks have a full width at half-maximum of the order of 0.03--0.04°, which corresponds to the beam divergence perpendicular to the diffraction plane of the used diffractometer. The pattern of peaks shows a perfect mirror symmetry to the \[011\] and \[001\] directions, as expected for the lattice structure of Si.
The measurement in low-resolution mode (Fig. 2[▶](#fig2){ref-type="fig"} *b*) shows a similar peak pattern but strongly smeared out so that details get completely lost and the intensity between some adjacent peaks does not reach the background level. This explains why the intensity of the Si 200 peak varies from sample to sample as long as the in-plane orientation of the sample is not exactly reproduced. Under the given low-resolution mode representing typical diffractometer conditions, the intensity of the Si 200 reflection may vary between the background level and more than 10^4^ c.p.s. If the sample is approximately aligned in the \[011\] direction as in the case demonstrated in Fig. 1[▶](#fig1){ref-type="fig"}, the Si 200 reflection is always detected within a Φ range of ±3°, and a variation in Φ by 1° is sufficient to modify the peak intensity by more than one order of magnitude.
The Φ scans in Fig. 2[▶](#fig2){ref-type="fig"} deliver furthermore a clear argument against the explanation of the Si 200 peak as a higher harmonic, *e.g.* Si 400, reflection with half the Cu *K*α wavelength from the continuous spectrum. If this were true then a constant intensity should be visible, independent of the in-plane sample orientation. Furthermore, as long as no X-ray optical element is used in the beam pass in front of the sample (as typical for the Bragg--Brentano setting), the Si 400 reflection would contribute to the background level at any Bragg angle around the Si 200 peak position owing to the corresponding wavelength from the continuous spectrum. The background level and the contribution of the Si 400 reflection in the given Φ scans in Fig. 2[▶](#fig2){ref-type="fig"}, as well as in ω--2Θ scans near the 200 peak, are clearly below 10 counts per second, which is orders of magnitude lower than the observed peaks.
Fig. 3[▶](#fig3){ref-type="fig"} shows a fraction of the high-resolution curve of Fig. 2[▶](#fig2){ref-type="fig"}(*a*) around the \[011\] direction with an indication of the associated first and second reflecting planes of the Si 200 multiple diffraction as calculated by Hwang (2001[@bb1]). The calculated and measured peak positions agree perfectly, with errors less than the measuring step width of 0.012°.
The more complex diffraction behaviour becomes visible in a 2Θ--Φ mapping, as shown in Fig. 4[▶](#fig4){ref-type="fig"}(*a*) for the high-resolution case and the same Φ range as in Fig. 3[▶](#fig3){ref-type="fig"}. Not only are there intensity peaks visible at the exact Bragg position of the Si 200 reflection, but inclined intensity streaks reach 2Θ values ±1.5° away from the peak position. It is necessary to note that the dotted structure of some streaks is caused by finite step widths of 0.02 and 0.1° for 2Θ and Φ, respectively. In agreement with calculations by Rossmanith *et al.* (2001[@bb12]), these streaks are caused by the participation of a certain wavelength interval in the diffraction, since in the used parallel beam configuration there was no crystal collimator involved to suppress the Cu *K*α~2~ line and other adjacent wavelengths from the continuous spectrum. This becomes clear in Fig. 5[▶](#fig5){ref-type="fig"}(*a*), where a 2Θ--Φ mapping of a smaller area near the \[011\] in-plane direction is repeated with lower step width in Φ. The expected angles of the Si 200 reflection for Cu *K*α~1~ and *K*α~2~ radiation are marked. Corresponding intensity maxima can clearly be identified, lying exactly on the observed streaks. Fig. 5[▶](#fig5){ref-type="fig"}(*b*) shows the same mapping for conditions where the incident wavelength distribution is limited by the use of a crystal collimator to the Cu *K*α~1~ line only. As expected, the observed intensities are much lower, but intensity peaks can now be found at the angular position for *K*α~1~ only, and all wavelength-related streaks have disappeared. The weak, nearly perpendicular streak visible at the peak near ΔΦ = 1.6° that is also visible for the parallel beam case is most likely related to the slit-limited angular acceptance of the detector.
Coming back to the more intense parallel beam configuration without collimator crystal, the consequence for ω--2Θ scans on Si with (100) orientation under these high-resolution conditions is that depending on the Φ position of the sample not only one sharp peak at the exact Bragg position of the Si 200 reflection may occur, but two or even three sharp peaks distributed over a wider angular range are possible.
The 2Θ--Φ mapping under low-resolution conditions shown in Fig. 4[▶](#fig4){ref-type="fig"}(*b*) indicates a very diffuse intensity distribution. Nevertheless, the now smeared out streaks of multiple diffraction cause a characteristic intensity pattern, and it becomes clear that the Si 200 peak not only may appear or disappear but also may have shoulders or even broad subpeaks on both sides depending on the in-plane orientation of the sample. The bar between ΔΦ = ±2.5° marks approximately the Φ range within which the measurements shown in Fig. 1[▶](#fig1){ref-type="fig"} were performed. The broad intensity shoulder on the low-angle side of the 200 reflection marked in Fig. 1[▶](#fig1){ref-type="fig"} is clearly caused by the Si substrate itself, and it is not the reflection of any kind of surface layer. Likewise, the variation of the height of the 200 reflection can be easily explained by a slight modification of the sample orientation.
3.2.. Si 222 reflection {#sec3.2}
-------------------------
The situation for the Si 222 reflection is rather different from Si 200. Fig. 6[▶](#fig6){ref-type="fig"}(*a*) shows a Φ scan measured at the Bragg angle of the Si 222 reflection (2Θ = 58.86°) in high-resolution mode. Here a constant intensity of about 2.5 × 10^4^ c.p.s. exists independent of Φ, which is interrupted by sharp peaks generated by multiple diffraction (*Umweganregung*) that are in most cases associated with a dip to lower intensity.
The 2Θ--Φ mapping near the in-plane direction shown in Fig. 6[▶](#fig6){ref-type="fig"}(*b*) confirms the constant intensity band of the Si 222 reflection clearly split into *K*α~1~ and *K*α~2~ lines. Some streaks and even a bow caused by multiple diffraction are visible, but they seem to be less pronounced because of the generally higher level of background intensity.
The fact that most basis-forbidden reflections do not have precisely zero intensity can be explained (Tischler *et al.*, 1988[@bb13]) by noncentrosymmetric parts in the atomic charge distribution, which produces a nonzero structure factor for the 222, 442, 622 and other basis-forbidden reflections. But, for the 200 reflection, where the atomic charge distribution is consistent with the atomic site geometry, the structure factor is exactly zero.
4.. Summary and conclusions {#sec4}
=============================
It was demonstrated that the occurrence of the basis-forbidden Si 200 reflection in ω--2Θ scans is caused by multiple diffraction (*Umweganregung*), and the intensity and peak shape depend on three parameters: the in-plane orientation of the sample Φ, the divergence of the used diffractometer perpendicular to the diffraction plane and the wavelength distribution in the incident beam. The intensity can vary between really zero and values close to non-forbidden reflections. For diffractometers with low divergence perpendicular to the diffraction plane the 200 reflection may be represented by no or up to three relatively sharp peaks that show up in a 2Θ range between about 31 and 35°. For more commonly used diffractometers with a divergence perpendicular to the diffraction plane of the order of some degrees, no matter whether they are used in parallel beam or Bragg--Brentano geometry, the 200 reflection may exhibit shoulders or subpeaks in the same angular range as mentioned above, or in extreme cases a broad peak might even occur somewhere in this angular range, only without any sharp peak at the exact 2Θ value.
This behaviour of the Si substrate itself has significant consequences for the analysis of layer materials on Si(100) substrates that show diffraction peaks in the 2Θ range between 31 and 35°, and the following recommendation can be given. For diffractometers with low divergence perpendicular to the diffraction plane the problem is minor, since a broad diffraction peak from a thin layer can easily be distinguished from the rather sharp Si 200 peak (or multiple peaks). The situation becomes more critical for standard X-ray diffractometers with relatively large divergence perpendicular to the diffraction plane. Although the divergence can be reduced by the use of Soller slits with low angular acceptance, this is usually not done since it is always associated with a significant intensity reduction. To avoid under these conditions any confusion between substrate-related diffraction phenomena and real layer diffraction it is urgently necessary to check the behaviour of an observed peak by rotation in Φ. Only if the peak intensity does not change (or at least does not vanish) depending on Φ can one be sure that this peak is really related to a layer material. A second alternative is the use of a crystal collimator, but this is in many cases not acceptable owing to the strong reduction of intensity associated with that approach.
Taking the results of these investigations into account, the proof of epitaxial Gd~2~O~3~ growth with (100) orientation on Si(100) substrates as reported by Chaudhuri *et al.* (2014[@bb2]), based on ω--2Θ scans on three samples without any detailed information about the exact experimental conditions, cannot be accepted. The observed shoulder on the low-angle side of the Si 200 reflection peak, which was indexed as the Gd~2~O~3~ 400 reflection, is very similar to that shown in Fig. 1[▶](#fig1){ref-type="fig"}. Only a Φ scan at this 2Θ position is able to give final evidence for the growth of (100)-oriented Gd~2~O~3~.
This article will form part of a virtual special issue of the journal, presenting some highlights of the 12th Biennial Conference on High-Resolution X-ray Diffraction and Imaging (XTOP2014).
{#fig1}
{#fig2}
![High-resolution Φ scan close to the \[011\] in-plane direction with indication of the associated first and second reflecting planes of Si 200 multiple diffraction (*Umweganregung*).](j-48-00528-fig3){#fig3}
![2Θ--Φ mapping of the Si 200 reflection near the \[011\] in-plane direction measured (*a*) in high-resolution and (*b*) in low-resolution mode. The dotted structure of some streaks is caused by the finite step widths of 0.02 and 0.1° for 2Θ and Φ, respectively. The bar between ΔΦ = ±2.5° marks approximately the Φ range within which the measurements shown in Fig. 1[▶](#fig1){ref-type="fig"} were performed.](j-48-00528-fig4){#fig4}
![2Θ--Φ mapping of the Si 200 reflection near the \[011\] in-plane direction measured in high-resolution mode and (*a*) parallel beam configuration and (*b*) with an additional Ge(400) × 2 collimator crystal and step widths of 0.02 and 0.02° for 2Θ and Φ, respectively.](j-48-00528-fig5){#fig5}
{#fig6}
| {
"pile_set_name": "PubMed Central"
} |
A 54-year-old man with acrofacial vitiligo of 15 years duration presented with gradual onset of progressively increasing depigmented, translucent, skin-colored raised lesions strictly restricted to the vitiliginous patches on the dorsum of the hands and feet since the past 1 year. Clinical examination revealed discrete, closely-spaced, translucent, shiny, and dome-shaped papules of varying size between 1 to 3 mm \[[Figure 1](#F1){ref-type="fig"}\]. When nicked with a scalpel, a papule extruded clear, gelatinous substance. No lesions were observed on other exposed areas or covered areas of the body. The patient had a history of chronic exposure to UV radiation (20 years duration) and regular contact with lubricating petroleum products (14 years duration) as part of his outdoor occupation \[[Figure 2](#F2){ref-type="fig"}\].
{#F1}
{#F2}
The patient was otherwise healthy, and there was no history of similar lesions in his family. Other routine laboratory investigations were within normal limits.
Histopathological findings and special staining by Van Gieson are shown in Figures [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"}.
{#F3}
{#F4}
Question {#sec1-1}
========
What is your diagnosis?
| {
"pile_set_name": "PubMed Central"
} |
'Ring-enhancing brain lesions with clinical presentation of focal seizures and inconclusive diagnostic features on anatomical neuroimaging modalities' is an area that continues to challenge attending neurologists with regard to the definitive etiologic diagnosis; therapeutic approaches are clearly different and an accurate diagnosis is of great importance in each of these settings; in addition, some of the conditions are potentially treatable. The common differentials in the general population include neurocysticercosis, tuberculomas, brain abscesses, and metastatic lesions. In AIDS patients, the major differential diagnosis is between central nervous system (CNS) lymphoma and CNS toxoplasmosis. How can one envisage the role of ^18^F-FDG-PET/CT in this scenario as a valuable adjunct to further a definitive diagnosis? For the purpose of systematic discussion, we would approach the problem by considering each of the differential diagnoses separately in this clinical setting.
In a recent study, Jolepalem and Wong [@R1] have observed a relatively low-grade/absent ^18^F-FDG uptake in lesions of neurocysticercosis. This finding is relatively consistent with that in a few published case reports by other authors [@R2] and we have also made a similar observation in our practice ([Fig. 1](#F1){ref-type="fig"}). This low uptake of ^18^F-FDG-PET in neurocysticercosis is noteworthy, especially as there is now substantial evidence in the literature on infective--inflammatory lesions demonstrating avid ^18^F-FDG uptake commensurate with their inflammatory activity.
{#F1}
Although the low ^18^F-FDG uptake in neurocysticercosis is uniform in all the published reports, there has been conflicting results and an overlap of ^18^F-FDG uptake reported with intracerebral tuberculomas. One of the earlier studies in this domain documented low-grade ^18^F-FDG uptake in CNS tuberculomas and toxoplasmosis [@R3]. In this study [@R3], the standardized uptake value ratio (lesion to contralateral brain area) in both of these conditions was significantly lower (range: 0.3--0.7) than that noted in patients with lymphoma (range: 1.7--3.1) with virtually no overlap of the uptake values. Contrary to this [@R3], in a recently reported case series studying intracranial tuberculomas [@R4], a relatively high lesion-to-normal gray matter uptake ratio was observed on both ^11^C-methionine (1.8±0.38) and ^18^F-FDG scans (1.64±0.26) in all newly diagnosed cases. The investigators concluded that ^11^C-methionine, similar to ^18^F-FDG, has limited specificity in distinguishing tuberculoma from neoplastic lesions [@R4].
Brain abscess is another differential diagnosis in which there is typically high uptake in the periphery of the lesion with reduced metabolic activity in the center. A similar pattern has been noted in a comparative study with both fluoroethyl tyrosine and ^18^F-FDG, which showed false-positive results in abscesses and in a case of acute demyelinating lesion [@R5]. Such an uptake pattern in brain abscesses has also been documented in another comparative study with ^11^C-methionine, and the possible role of PET with both tracers in assessing response to antibiotic therapy was highlighted [@R6]. A high uptake of ^18^F-FDG has also been documented in a solitary case of tuberculous brain abscess on ^18^F-FDG-PET/CT [@R7].
Thus, from an analysis of the published literature [@R1; @R2; @R3; @R4; @R5; @R6; @R7], one can infer that the positive predictive value of ^18^F-FDG-PET is relatively limited (a biopsy is indicated), but the negative predictive value of ^18^F-FDG-PET could prove beneficial in this setting especially if appropriately correlated with MRI interpretation to the positive predictive value. We have observed that malignant/metastatic lesions could be excluded with a high degree of certainty when there is absent/minimal ^18^F-FDG uptake in the CT/MRI-depicted lesion sites ([Fig. 1](#F1){ref-type="fig"}). In contrast, in the presence of intense ^18^F-FDG uptake in the described lesions, it is imperative to undertake a whole-body PET/CT study in the same sitting. Quite often, this would also detect the site of primary tumor (usually in the lung) and aid in whole-body disease staging in the same diagnostic examination ([Fig. 2](#F2){ref-type="fig"}). This will have important therapeutic implications from the point of view of patient management. Primary CNS lymphoma (an important differential in patients with HIV infection) demonstrates substantially increased metabolic activity on ^18^F-FDG-PET.
{#F2}
In conclusion, in the setting of ring-enhancing lesions of uncertain etiology on conventional anatomical imaging, a positive ^18^F-FDG-PET/CT has limited value in differentiating between the various causes (especially when the differential lies between malignancy and abscess/tuberculoma) and would warrant a biopsy; however, in the absence of discernible uptake in the lesions, the probability of malignancy (including lymphoma) is substantially low. In the presence of intense ^18^F-FDG uptake indicating a malignant/metastatic lesion, a whole-body imaging is of potential help in detecting the site of primary tumor, thus providing an opportunity for whole-body disease staging and thereby influencing the subsequent clinical management of the patient. It needs to be determined whether the newly developed combined simultaneous PET/MRI has additional advantage and whether this modality can be translated into a 'one-stop shop' procedure in this setting.
Conflicts of interest
=====================
There are no conflicts of interest.
Correspondence to Sandip Basu, MBBS(Hons), DRM, DNB, MNAMS, Radiation Medicine Centre, Bhabha Atomic Research Centre, Tata Memorial Centre Annexe, Jerbai Wadia Road, Parel, Mumbai 400012, India Tel: +91 22 24146059/+91 22 24135232/+91 22 24149428; fax: +91 22 24157098; e-mail: <drsanb@yahoo.com>
| {
"pile_set_name": "PubMed Central"
} |
Citation
========
Gutt J, Piepenburg D, Voß J (2014) Asteroids, ophiuroids and holothurians from the southeastern Weddell Sea (Southern Ocean). ZooKeys 434: 1--15. doi: [10.3897/zookeys.434.7622](10.3897/zookeys.434.7622)
Introduction
============
The southeastern Weddell Sea in the Atlantic Sector of the Southern Ocean is a typical high-latitude Antarctic region. It is located in the circumpolar permanent pack-ice zone ([@B19]), characterized by summerly polynyas (areas of open water surrounded by sea ice). Due to its remoteness and persistent sea-ice cover, it was not accessible for extensive scientific surveys before the availability and support of modern research platforms that are capable to operate independently in sea-ice covered waters.
The first multidisciplinary marine research was carried out in this area in the 1980s during the first Antarctic cruises of the German icebreaking research and supply vessel \"Polarstern". Embedded in a broad ecological research programme, addressing a range of evolutionary, systematic, zoogeographical and ecological issues, first comprehensive faunistic inventories of the asteroid (starfish), ophiuroid (brittle star) and holothurian (sea cucumber) bottom fauna were conducted, based on field sampling efforts ([Fig. 1](#F1){ref-type="fig"}) during "Polarstern" cruises ANT-I/2 (PS01), ANT-II/04 (PS04), and ANT-III/3 (PS06) (for cruise reports see [@B18], [@B6], [@B21], and [@B20], respectively).
{#F1}
The major objective of this collection work was to provide material for subsequent zoogeographical and ecological studies on the asterozoan (asteroid and ophiuroids together; [@B25], [@B22]) and holothurian fauna ([@B9], [@B12]), as well as for analyses of entire macrozoobenthos communities ([@B14]; for a compilation: [@B15]). Here, we publish the complete original dataset of asteroid, ophiuroid and holothurian abundances, including zero-abundances (= absences in the catches) that formed the basis of the scientific findings published in the research papers mentioned above, to allow for the general accessibility to such data associated with starfishes, brittle stars and sea cucumbers from the southeastern Weddell Sea.
In addition to making data and metadata available in the public database ANTABIF ([www.biodiversity.aq](http://www.biodiversity.aq)), a robust community and diversity analysis for holothurians and asterozoans combined was carried out to synthesize results already published for holothurians and asterozoans separately ([@B25], [@B12], [@B22]).
Study area, material and methods
================================
Asteroid, ophiuroid, and holothurian specimens were sampled at a total of 59 sites distributed across the southeastern Weddell Sea at water depths between 160 and 1,180 m ([Fig. 1](#F1){ref-type="fig"}; for more detailed information see section on "Geographic coverage" below). In general, this region is characterized by a relatively homogenous physical environment, especially in terms of water masses, poorly sorted sediments, persistent sea-ice cover and hardly predictable occurrence of coastal polynyas. As such, it is representative for the entire high-latitude Antarctic habitat. Some drivers of faunistic heterogeneity, in addition to biological interactions and unknown unpredictable factors, are briefly summarized in the section on "Project Data -- Study area description" below.
The field samples were mainly taken by means of an Agassiz trawl, but also with a commercial bottom trawl and, in one case, a smaller dredge. During the cruises, GPS positions were available approximately each six hours. Between the GPS fixes, the ship\'s positions were death reckoned (ship\'s track calculated by means of ship\'s speed and course starting from a satellite fix). Swept areas were estimated for each haul as described in [@B25] and [@B9]. Water depths were measured by a DWD echo sounder. For more detailed information see section on "Sampling methods -- Sampling description" below.
Specimens were collected from either total catches or, in some cases subsamples, counted and preserved on board. Using the swept-area estimates, individual counts were standardized to abundance values (ind m^-2^). After the cruises, the preserved specimens were identified to species in the lab. Some holothurian species, which were assumed to be new to science, were formally described ([@B10], [@B11]). Some of these new descriptions were later revised. The specimens were not integrated into a museum\'s collection, and original data were never published at that time when electronic data bases did not yet exist.
The quality of the data and metadata published here was enhanced prior to publication following the best practices suggested in the literature during the digitalization and geo-referencing processes. Moreover, the current accurate spelling of scientific names -- except for the ophiuroid *Theodoria conveniens* (\"nomen dubium\") -- was reviewed based on the World Register of Marine Species ([www.marinespecies.org/](http://www.marinespecies.org/)). For more information see "Sampling methods -- Quality control" below.
For 26 stations, at which both holothurians and asterozoans were sampled from Agassiz trawl catches, a simple multivariate community analysis of combined holothurian and asterozoan data were carried out, using the PRIMER 6.1.6 software ([@B3], [@B2]). Abundance values (ind m^-2^) were standardized to percentages per catch, to eliminate bias possibly introduced by differences between-haul catchability. Between-station resemblances were quantified by means of the Bray-Curtis similarity coefficient. The overall pattern of taxonomic resemblances was investigated using cluster analysis (average linkage) and Multidimensional Scaling (MDS).
Results
=======
The dataset comprises a total of 4,509 records of abundances of 35 asteroid species (with a total of 2,089 specimens) and 38 opiuroid species (with a total of 18,484 specimens) from 34 stations, as well as of 66 holothurian species (with a total of 20,918 specimens) from 59 stations including zero-abundances.
Asteroid, ophiuroid and holothurian species were present at all sites investigated in the study area but composition and abundance of the asterozoan and holothurian assemblages varied considerably. The synthesizing community analysis shows four holothurian-asterozoan clusters. Since the cluster \"Overdeepened Basins II\" shows an affinity to \"Overdeepened Basins I\" rather than to \"Eastern Shelf\" ([Fig. 2a](#F2){ref-type="fig"}), these two clusters were merged for analyzing the family-level composition. The \"Eastern Shelf\" assemblage was richest in species, \"Overdeepened Basins II\" was poorest, and the others were similar to each other, with intermediate species numbers ([Fig. 2b](#F2){ref-type="fig"}). The class-level relative abundances were similar in all clusters, with highest values for ophiuroids followed by holothurians and lowest for asteroids ([Fig. 3](#F3){ref-type="fig"}). However, on a family level, major differences became visible between the assemblages, whilst in the \"Overdeepened Basins I & II\" deposit-feeding holothurians (Synallactidae) were dominant. Filter feeders (Psolidae and Cucumariidae) were most abundant in \"Eastern Shelf\" and \"Southern Shelf\". Among asteroids, *Hymenaster* spp., of the family Pterasteridae, a typical deep-sea form, were most abundant at basin sites at water depths down to almost 1,200 m. The genus *Odontaster*, being generally abundant in the Antarctic, was dominant in the assemblage \"Southern Shelf\"; apart from that the evenness among the asteroids was more obvious than among holothurians and ophiuroids. A major difference between ophiuroid assemblages was the second dominance of Ophiolepididae in the \"Eastern Shelf\" assemblage, with *Ophioceres incipiens* being generally rare but occurring locally in relatively high abundances. It is a diatom feeder ([@B4]) and the smallest species in the entire area. The obvious high abundance of Ophiuridae across all clusters is due to the dominance of various species: the omnivorous large-sized *Ophionotus victoriae* in the assemblage \"Overdeepened basins II\", the crustacean feeder *Ophioplinus gelida* in \"Southern Shelf\" and the shallower shelf prefering *Ophioplinthus martensi* and the deeper *Ophioplinthus brevirima* in \"Eastern Shelf\". Since *Ophiacantha antarctica* was by far the most abundant ophiuroid in \"Southern Shelf\" Ophiacanthidae was the dominant family in this assemblage.
{#F2}
{#F3}
In the case of the separate analysis of asterozoan patterns ([@B25], [@B22]), water depth and latitude seemed to be the most important drivers of assemblage distribution and composition. At shallow shelf sites rare asteroid and ophiuroid predators, such as, e.g., *Acodontaster conspicuus* and *Odontaster validus*, respectively, as well as epibiotic ophiuroids, e.g., *Astrotoma agassizii*, occurred in addition to the common and widely distributed opportunistic feeders. In the case of the separate analysis of holothurians ([@B9], [@B12]), one assemblage co-occurred with the rich macrozoobenthic community dominated by dense epifauna consisting of, e.g., sponges and gorgonians, which are often used as substratum by epibiotic filter feeders. Others live mainly as vagrant deposit feeders on the predominantly non-colonised substratum, such as typical deep-sea species, e.g., *Elpidia glacialis* and *Protelpidia murrayi* ([@B13]). In addition, a mixed holothurian assemblage was identified.
General significance
====================
Virtually nothing was known about echinoderms in the southeastern Weddell Sea before the field sampling work, in the course of which the data published here was recorded. Also from other Antarctic regions only sporadic information on the three echinoderm classes, especially holothurians, was available at that time, mainly found published in the taxonomic literature. However, the application of a whole-assemblage approach was novel, and comparable surveys are even nowadays rare.
The unique dataset encompasses some of the first observations of asteroids, ophiuroids and holothurians in this area and represents a significant contribution of primary data about Antarctic benthos assemblages. Moreover, it provides unique baseline data for future faunistic, ecological and conservation studies to evaluate the effects of climate change and possible future fishing activities in this area. At present and in the future these faunistic data can gain further importance in the context of a number of further applications:
\(1\) More comprehensive circumpolar analyses and comparative studies with other large taxa become possible if these data are merged with similar datasets from other regions, see e.g., <http://ipt.biodiversity.aq/resource.do?r=asteroidea_zoogeography> and <http://ipt.biodiversity.aq/resource.do?r=biopearl_asteroida>. Data on these three echinoderm classes can also be compared within the same region with other taxa or environmental parameters, as compiled, e.g., by [@B5].
\(2\) At the time of sampling in the 1980s, the study area was almost pristine and hardly affected by any anthropogenic activities. In the meantime, exploratory fishing started on the deeper shelf, and the knowledge on the fauna before the onset of these activities can serve as a valuable baseline for an assessment of the impact of further fishing.
\(3\) The same holds true for the study of the effects of climate change. The area is so far climatologically relatively stable but an increase of bottom-water temperatures is expected during the 21st century ([@B17]).
\(4\) The data can also be used for nature conservation initiatives ([@B24]). They are especially suited for such applied ecological comparative studies, since all specimens -- with few well-defined exceptions -- were sorted from the catches, meaning that not only the presence of the species were registered but also very valuable absence data and the even more informative abundances. In general, the latter data are known to be more sensitive indicators of environmental change than binary presence-absence data.
\(5\) The community analysis of the combined holothurian-asterozoan data largely confirms the findings of the previous separate analyses of holothurians and asterozoans ([@B25], [@B12], [@B22]). There are basically three assemblages, the composition of which are quite similar on the level of classes but strongly differing on the level of families and species, as well as with regard to trophic guilds. These resemblance patterns can be attributed to geographic and bathymetric conditions.
General description
===================
**Purpose:** The publication of the complete dataset of asteroid, ophiuroid and holothurian abundances (and absences) in the southeastern Weddell Sea, which formed the basis of the scientific findings already published in a number of original research papers, shall allow for the general accessibility to such data associated with starfishes, brittle stars and sea cucumbers from this high-Antarctic region. The unique dataset encompasses some of the first observations of asteroids, ophiuroids and holothurians in the study area and represents a significant contribution of primary data about Antarctic benthos assemblages. Moreover, it provides unique baseline data for future faunistic, ecological and conservation studies to evaluate the effects of climate change and possible future fishing activities in this area.
Project details
===============
**Project title:** Asteroids, ophiuroids and holothurians from the southeastern Weddell Sea (Southern Ocean)
**Funding:** The sampling of all asterozoan and holothurian specimens in the course of the cruises ANT-I/2 (PS01), ANT-II/04 (PS04), ANT-III/3 (PS06) of the German R/V \"Polarstern\" and the subsequent analysis of asteroids and ophiuroids was financed by the Alfred Wegener Institute Helmholtz, Centre for Polar and Marine Research, Bremerhaven, Germany. The study on holothurians was also supported by a grant of the Deutsche Forschungsgemeinschaft (He 89/49).
**Study area descriptions/descriptor:** The study area included regions characterized by almost permanent pack-ice cover in the southernmost Weddell Sea as well as regions featuring coastal polynyas in the eastern Weddell Sea ([@B14]). The shelf figau is mostly rather narrow, only a few 10 km wide, in some areas even \"disappearing\" beneath the floating ice shelf, but can also be much broader in the southern Weddell Sea ([@B1]). Consequently, near-coast habitats can be affected by larger and smaller floating ice shelves or by a glaciated coast. As there is no \"true\" (i.e., non-glaciated) coast, shallow littoral habitats (\<50 m water depth) are not known from this area.
The shelf seabed is usually rather level, and especially habitats on banks and their flanks are disturbed with varying intensity by grounding or scouring icebergs ([@B16]). In addition to the shelf below 160 m water depth sampling included the upper slope and stations in the Filcher depression, an overdeepened trough with water depths of up to 1,180 m.
The hydrography of the study area is characterized by a southwestward flowing coastal current, which is part of the large Weddell Gyre ([@B7]) and flows with a velocity of up to 0.14 m/s above the shelf edge. Different water masses dominated by the \"Eastern Shelf Water\" close to the sea floor are mainly characterized by low temperatures close to the freezing point and high salinities. However, occasionally upwelling \"Warm Deep Water\", with an average temperature of 0.4 °C, can be found on the deeper shelf ([@B7], [@B23]). In the South, the current regime shows water flow from beneath and under the Filchner-Ronne Ice Shelf, with both northward and southward directions ([@B8]). This difference is potentially of high relevance for the food supply to the benthos and also shapes species compositions along the eastern coast with smaller ice shelves.
Surface sediments are generally poorly sorted. However, clear differences in the sand-silt proportion exist, with mainly soft sediments in the deep areas and coarser sediments on the shelf. In some areas, biogenic particles, such as bryozoan debris and sponge spicules, are important components of the sediments, sometimes forming dense mat-like structures ([@B25]).
**Design description:** Asteroid, ophiuroid, and holothurian specimens were sampled during the cruises ANT-I/2 (PS01), ANT-II/04 (PS04), ANT-III/3 (PS06) of the German R/V \"Polarstern\" at a total of 59 sites distributed across the southeastern Weddell Sea at water depths between 160 and 1,180 m (for more detailed information see section on "Geographic coverage"). The field samples were mainly taken by means of an Agassiz trawl, but also with a commercial bottom trawl and, in one case, a smaller dredge (for more detailed information see section on "Sampling methods - Sampling description"). During the cruises, GPS positions were recorded approximately every six hours. Between the GPS fixes, the ship\'s positions were death reckoned. Water depths were measured by a DWD echo sounder. Specimens were collected from either total catches or, in some cases subsamples, counted and preserved on board. They were later identified in the lab. Some holothurian species, which were assumed to be new to science, were formally described. Some of these new descriptions were later revised. The specimens were not integrated into a museum\'s collection, and original data were never published at that time when electronic data bases did not yet exist. The quality of the data and metadata published here was enhanced prior to publication following the best practices suggested in the literature during the digitalization and geo-referencing processes.
Taxonomic coverage
==================
**General taxonomic coverage description:** All asteroids, ophiuroids, and holothurians caught by the gear mentioned in the section on "Sampling Methods\" were considered in this study, with the exception of the very rare species *Amphiura deficiens* Koehler, 1992 and *Amphiura atlantica* Ljungman, 1867. The taxonomic and morphological range even covers two holothurian species, which are assumed or known to be able to swim occasionally, *Rhipidothuria racovitzai* and *Peniagone vignioni*. Due to the mesh size used, not only adult but also juvenile specimens of all three classes are included in the collections. However, their abundance values are likely more biased than those of the adults. The trawls predominantly caught epifaunal species in a semi-quantitative way, the Agassiz trawl obviously with a higher catchability of macro-epibenthic invertebrates than the bottom trawl. Therefore, the swept-area approach is most useful for within-gear comparisons and only with less precision between-gear. The presence of typical infaunal species in the catches, such as the holothurian *Molpadia* and the asteroid *Hymenaster*, suggests that endobenthic species were also sampled to a considerable degree. Overall, organisms from a broad variety of ecological guilds among all three classes, such as deposit, sediment and filter feeders, infaunal, epifaunal and epibiotic (symbiotic) species, predators spezialized on various prey items, and scavengers, are present in the samples.
Taxonomic ranks
===============
Class: Asteroidea; species: *Bathybiaster loripes*, *Macroptychaster accrescens*, *Leptychaster flexuosus*, *Psilaster charcoti*, *Cheiraster (Luidiaster) gerlachei*, *Acodontaster capitatus*, *Acodontaster conspicuus*, *Acodontaster hodgsoni* f. *hodgsoni*, *Acodontaster hodgsoni* f. *stellatus*, *Acodontaster marginatus*, *Odontaster meridionalis*, *Chitonaster johannae*, *Notioceramus anomalus*, *Cycethra verrucosa*, *Perknaster aurorae*, *Perknaster sladeni*, *Porania (Porania) antarctica*, *Kampylaster incurvatus*, *Pteraster affinis*, *Pteraster stellifer*, *Hymenaster* spp., *Peribolaster macleani*, *Remaster gourdoni*, *Solaster regularis*, *Lophaster densus*, *Lophaster gaini*, *Lophaster tenuis*, *Paralophaster antarcticus*, *Paralophaster godfroyi*, *Paralophaster* sp., *Cuenotaster involutus*, *Henricia parva*, *Henricia smilax*, *Rhopiella hirsuta*, *Diplasterias brucei*, *Kenrickaster pedicellaris*, *Lysasterias digitata*, *Lysasterias perrieri*, *Notasterias armata*, *Notasterias bongraini*, *Notasterias haswelli*, *Notasterias stolophora*, *Pedicellaster hypernotius*, *Psalidaster mordax*.
Class: Ophiuroidea; species: *Astrotoma agassizii*, *Astrochlamys bruneus*, *Ophiacantha antarctica*, *Ophiacantha vivipara*, *Ophiacantha pentactis*, *Ophiocamax drygalskii*, *Ophiomitrella ingrata*, *Ophiomitrella* sp., *Ophiosparte gigas*, *Ophiolimna antarctica*, *Amphiura belgicae*, *Amphiura joubini*, *Amphiura proposita*, *Ophioleuce regulare*, *Ophioceres incipiens*, *Ophiocten dubium*, *Ophiocten doederleini*, *Ophiocten megaloplax*, *Ophionotus victoriae*, *Ophioperla koehleri*, *Ophioplinthus brucei*, *Ophiosteira debitor*, *Ophiosteira echinulata*, *Ophiosteira rotundata*, *Ophiura lymani*, *Ophiura (Ophiuroglypha) carinifera*, *Ophioplinthus brevirima*, *Ophioplinthus gelida*, *Ophioplinthus martensi*, *Ophioplinthus tumescens*, *Ophiogona doederleini*, *Ophiura flexibilis*, *Ophiura (Ophiuroglypha) irrorata*, *Ophiura rouchi*, *Theodoria conveniens*, *Ophioplinthus relegata*, *Anophiura* sp., *Amphiophiura* sp.
Class: Holothuroidea; species: *Cucumaria georgiana* s.l., *Psolidiella mollis*, *Cucamba psolidiformis*, *Microchoerus splendidus*, *Trachythyone parva*, *Trachythyone bouvetensis*, *Staurocucumis liouvillei*, *Staurocucumis turqueti*, *Heterocucumis steineni*, *Heterocucumis denticulata*, *Paracucumis turricata*, *Crucella scotiae*, *Crucella hystrix*, *Psolus dubiosus*, *Psolus charcoti*, *Psolus antarcticus*, *Psolicrux coatsi*, *Psolidium gaini*, *Psolidium poriferum*, *Echinopsolus acanthocola*, *Bathyplotes moseleyi* s.l., *Bathyplotes gourdoni*, *Bathyplotes bongraini*, *Pseudostichopus mollis*, *Molpadiodemas villosus*, *Mesothuria (Zygothuria) lactea*, *Laetmogone wyvillethomsoni*, *Rhipidothuria racovitzai*, *Peniagone vignoni*, *Protelpidia murrayi*, *Sigmodota contorta*, *Paradota weddellensis*, *Molpadia musculus*.
**Common names:** Starfish, Brittle stars, Sea cucumbers.
Spatial coverage
================
**General spatial coverage:** The study area extends northward to 70°27\'S, a latitude which is typical for the northern shelf off East Antarctica, with the exception of the large embayments of the Weddell and Ross Seas, and for the more southerly situated West Antarctic shelf regions, with the exception of the Antarctic Peninsula area. The southernmost station in the study area was located at 77°44\'S. With regard to its longitudinal extent, the study area ranged from 008°01\'W in the eastern Weddell Sea to 061°08\'W at the basis of the Antarctic Peninsula at 061°08\'W.
With regard to water depth, the samples were taken along a gradient ranging from 160 m at the shelf to a maximum of 1,180 m at the upper slope, encompassing stations at relatively shallow banks as well as those in overdeepened basins, such as the Filchner Trough.
**Coordinates:** 78°0\'0\"S and 70°0\'0\"S Latitude; 62°0\'0\"W and 8°0\'0\"W Longitude.
**Temporal coverage:** February 4, 1983 -- February 24, 1985.
Methods
=======
**Method step description:** See \"Sampling description\".
**Study extent description:** See \"Study area description\".
**Sampling description:** Three sampling gears were used: An Agassiz trawl that was deployed most frequently consisted of a metal sled, with an opening of 3 m width and 1 m height, to which a net, which was 5 m long and had a mesh size of 20 mm in the front part and 10 mm in the cod end, was attached. In front of the opening a tickler chain was fixed to the rig of the sled. The average trawling speed was 0.5 to 0.7 knots (nm/h), and the haul duration was approx. 20-30 min, resulting in swept areas of approx. 1,700 to 3,000 m^2^. This strategy was a compromise to gain comparable semi-quantitative information about both relatively poor Antarctic communities, based on catches that were, nevertheless, large enough to be representative, and rich communities without clocking the net.
The bottom trawl used was a 140 feet commercial otter trawl with a 22.5 m wide and approximately 3 m high opening. The mesh size of the net was 10 cm in the front part and 15 mm in the cod end. The bottom trawl was towed over ground at an average speed of 3 knots (nm/h), mostly for 0.5 hours, as it is standard for research fishing of demersal fish.
The dredge used has an opening of 100 × 30 cm and a mesh size of 10 mm. The swept areas were calculated according to [@B25] for the asterozoan study and according to [@B9] for the holothurian study.
Due to technical constraints, only one position and time was provided for the catches and only in some cases information on the depths at the start and end of the hauls was available. Metadata (position and time) are available at [www.pangaea.de](http://www.pangaea.de) for the expeditions PS01 and PS06. For cruise PS04, metadata were published by [@B25], with the exception of stn 490 ([@B9]). Metadata are based on the station lists published in the cruise reports ([@B18], [@B6], [@B21]).
**Quality control description:** Identifications were made by Joachim Voss under supervision of Ilse Bartsch for ophiuroids and by Julian Gutt for holothurians partly under supervision of David Pawson (comparison with material at the Smithsonian Institution, Washington) and Bent Hansen (Elasipodida caught during the Galathea expedition). In addition, identifications were based on the taxonomic references cited by [@B25] and [@B9]. All species names in the dataset are in accordance with the World Register of Marine Species ([www.marinespecies.org/](http://www.marinespecies.org/)), with the sole exception of the ophiuroid *Theodoria conveniens* (\"nomen dubium\"). Consequently, some species names have changed in comparison to the ones used in past publications, as these are synonyms that are not valid anymore.
Datasets
========
Dataset description
-------------------
**Object name:** Darwin Core Archive Asteroids, ophiuroids and holothurians from the southeastern Weddell Sea (Southern Ocean)
**Character encoding:** UTF-8
**Format name:** Darwin Core Archive format
**Format version:** 1.0
**Distribution:** <http://ipt.biodiversity.aq/archive.do?r=asteroids_and_ophiuroids_from_the_southeastern_weddell_sea>
**Publication date of data:** 2014-02-20
**Language:** English
**Licenses of use:** This work is licensed under a Creative Commons CCZero 1.0 License <http://creativecommons.org/publicdomain/zero/1.0/legalcode>
**Metadata language:** English
**Date of metadata creation:** 2014-01-09
**Hierarchy level:** Dataset
**Metadata language:** English
**Date of metadata creation:** 2014-01-09
**Hierarchy level:** Dataset
[^1]: Academic editor: Lyubomir Penev
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Concerns regarding blood transfusion-related risks exist in most countries, especially those of Europe and North America^([@r01]-[@r07])^. Likewise, in the international literature, there are numerous observations of the importance of hemovigilance in transfusions^([@r01]-[@r05],[@r08]-[@r11])^. However, in Brazil, there are few studies on this subject in particular about the impact of the implementation of transfusion committees (TCs) in relation to hemovigilance and transfusion safety^([@r12],[@r13])^.
The monitoring, detection, screening and treatment of transfusion reactions, as well as preventive measures, used as strategies to minimize the occurrence of these adverse events, have been mandatory in Brazil since 2004.
In Brazil, better control measures in the selection of blood donors were established after 1980 with the advent of AIDS and its social impact. In addition, remunerated blood donations were prohibited in order to maximize the safety of donor and recipient^([@r03],[@r10])^. Despite major progress in Brazilian transfusion therapy, particularly in respect to measures to prevent the transmission of infectious diseases through donated blood, strategies for the detection and prevention of transfusion reactions are still needed.
Blood transfusion policies started to be effectively discussed, designed and implemented in Brazil only after 2000 when the establishment of multidisciplinary TCs was proposed for health centers that used transfusion therapy^([@r14],[@r15])^. The objective of these TCs was to detect, investigate, treat and prevent transfusion reactions, i.e., to monitor blood transfusion practices in order to minimize transfusion-related risks, as well as to rationalize the use of blood products^([@r15])^. These measures became mandatory in 2004 with the publication of Board Resolution No. 153 (RDC 153) of the Brazilian national health surveillance agency, ANVISA, which determines that \"Health centers where a blood transfusion service is available need to establish a multidisciplinary TC in which one representative of the therapy service must participate\"^([@r16])^. In view of this scenario, the objectives of the present study were to determine the existence of TCs at blood transfusion services linked to the public blood bank network of Minas Gerais, to evaluate their impact on aspects related to transfusion safety, and to provide data that contribute to the elaboration of strategies aimed at improving blood transfusion surveillance in the state and in Brazil.
Methods
=======
A cross-sectional, observational study was conducted. This study was part of the project \"Transfusion safety: evaluation of the physical and operational infrastructure of transfusion services of the public blood bank network of the State of Minas Gerais\" developed by Hemominas. Data were collected between July 2007 and August 2008.
All blood transfusion services and contractors providing blood products to Hemominas were investigated using specific questionnaires based on the current legislation and applied *in loco*. Using an evidence-based approach, the instrument was tested and validated in a pilot study, showing high reproducibility and overall agreement.
Systematic cluster sampling was used for the collection of data regarding registries, requests, production and release of blood and blood products, and records of the occurrence of adverse reactions.
Transfusion reactions can be defined as any adverse event resulting from blood transfusion. These reactions are traditionally classified according to etiology as infectious or non-infectious and immunological or non-immunological, and according to the time of occurrence as immediate or late^([@r17])^.
Evidence of the existence of a TC included the existence of meeting protocols, administrative actions of the committee, and records of events, classes, lectures and standardization. First a descriptive analysis was performed to characterize the profile of transfusion services with and without a TC. Differences between the two groups were evaluated using Pearson\'s or Fisher\'s chi-square test. A technical criterion was adopted to select the variables to be analyzed according to current legislation^([@r16])^. Variables that are not formally described in the legislation but, when present, are considered to be \"measures of good practice\" were also included.
The EpiData (version 3.1) program was used for data entry and SPSS Statistics 17.0 and SPSS Decision Trees for statistical analysis. The \"Decision Tree\" methodology consists of a multivariate analysis based on charts that illustrate decision rules^([@r18])^. These trees start with a node that contains all observations of the sample. Next, the data branch into mutually exclusive subsets that become increasingly more homogenous in relation to the response variable. The existence of a TC was defined as the response variable and factors related to its performance were defined as explanatory variables. The CHAID algorithm (Chi-square Automatic Interaction Detector) was used and a p-value ≤ 0.05 was defined as the stopping criterion. The risk estimate of incorrect classification that compares the value observed for the sample and the value predicted by the model was used to evaluate the goodness of fit of the model.
Results
=======
A total of 226 transfusion services and those that contracted the services of Hemominas, located in 178 municipalities, which performed 33,283 transfusions in the month prior to the visit and application of the questionnaires, were studied.
Evidence of a TC was obtained in 144 (63.6%) transfusion services, which performed 29,197 transfusions, whereas no TC was present in 82 (36.3%) performing 4,086 transfusions. A total of 557 transfusion incidents had been registered over the previous 12 months representing approximately 0.15% notifications/year. These adverse reactions occurred in 53 (36.8%) transfusion services with a TC and in eight (9.6%) without a TC (p \< 0.001), with 543(97.5%) and 14 (2.5%) notifications, respectively. Regarding the investigation of the causes of the transfusion incidents, 40 (75.5%) of the transfusion services with a TC that notified these incidents, but only two (25%) of the other services, investigated their causes. The number of notifications of transfusion reactions was 422, including 419 (99.3%) by transfusion services with a TC and three (0.7%) by services without at TC. Blood component bags were screened in cases of transfusion reactions at 24 (10.6%) transfusion services, 20 of which (83.3%) had a TC. Deletions on transfusion request forms were identified in 64 (44.4%) and 27 (32.9%) transfusion services with and without a TC, respectively (p = 0.160), incorrect identification of the recipient sample in 86 (59.7%) and 54 (65.9% p = 0.361), and the transfusion record book signed by the responsible person in 71 (49.3%) and 29 (35.4%), respectively (p = 0.042).
Individual transfusion charts of the recipients were kept at 38 (26.4%) of the transfusion services with a TC and at nine (11%) services without a TC (p = 0.006). These charts were updated after each transfusion at 37 (97.4%) and seven (77.8%) of these services, respectively (p = 0.031).
A laboratory technician who performed pre-transfusion tests was present in 34 (30.1%) transfusions services with a TC and in 18 (30%) without a TC (p = 0.99).
The results of this study are summarized in [Table 1](#t01){ref-type="table"}. [Figure 1](#f01){ref-type="fig"} shows the predominance of compliance with mandatory technical standards and good practices for transfusion services that possess a TC, including the investigation of the causes of transfusion reactions, existence of forms for the notification of transfusion reactions, signature of the person responsible in the transfusion book, presence of an opening statement in the transfusion book and correct identification of the recipient samples.
######
Comparison of transfusion services of the public blood bank network of Minas Gerais with an established transfusion committee and those without in respect to current legislation and good technical practices
**Parameter** **Transfusion Services with evidence of a transfusion committee (n = 144)** **Transfusion Services without evidence of a transfusion committee (n = 82)** **p-value**
-------------------------- ----------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------- ------- ------ ----------
Current legislation
Existence of forms for the notification of transfusion reactions 133 92.4 62 75.6 \< 0.001
Notification of transfusion reactions 53 36.8 8 09.6 \< 0.001
Investigation of the causes of transfusion reactions 40 27.8 2 02.4 \< 0.001
Incorrect identification of the recipient sample 86 59.7 54 65.9 0.361
Presence of individual transfusion charts 38 26.4 9 11.0 0.006
Update of the transfusion chart 37^a^ 97.4 7^b^ 77.8 0.031
Good technical practices
Technician\'s signature in the transfusion book 71 49.3 29 35.4 0.042
Deletions in transfusion requests 64 44.4 27 32.9 0.16
Clinical pathology technician performing pre-transfusion tests 34^c^ 30.1 18^d^ 30.0 0.99
^a^ n = 38, ^b^ n = 9, ^c^ n = 113, ^d^ n = 60
{#f01}
Discussion
==========
In the present study, we observed that, despite legal and contractual requirements, a significant proportion of the transfusion services linked to Hemominas do not possess a TC. In addition, the rate of non-compliance with current legislation and/or good practices was high even for those services in which a TC was implemented, suggesting ineffective implementation. Corroborating this assertion, the authors suggest that hospital TCs should have a major role in hemovigilance, allowing transfusion incidents related to the transfusion of blood components to be notified, their consequences assessed and appropriate actions in relation to their prevention to be taken. In addition, the indications of blood products should always be reviewed by the TC, thereby optimizing usage^([@r12])^. Within the context of hemovigilance, the TC is a major contributor to promote measures to provide greater safety to transfusion recipients^([@r12],[@r19])^. The highest rate of compliance was observed regarding the existence of forms for the notification of transfusion reactions, with compliance being significantly higher in transfusion services with a TC compared to those without a TC. However, despite the better performance of transfusion services with a TC, the number of notifications of transfusion reactions and the investigation of their causes were very low considering the number of transfusions performed in the previous year. The results of the present study are in contrast to the international literature which reports a frequency of transfusion reactions higher than 2%^([@r20])^ suggesting that a significant number of reactions are not notified even by transfusion services that possess a TC. Thus, the true rates of transfusion reactions in Brazil are unknown and it is believed that most episodes are not identified^([@r15],[@r21])^. The ANVISA Bulletin of Hemovigilance in 2010 estimated, based on the rate of transfusion reactions of France, that about 69.5% of transfusion incidents were not reported in 2009 (ranging from 97.2% in the Midwest to 60.5% in the Southeast),while acknowledging an improvement with the implementation of the System of Health Surveillance Notifications (NOTIVISA) as a tool for hemovigilance. Evidence of this improvement is the observed reduction in the rate of underreporting over the last three years as evaluated by the Bulletin (81.6% in 2007 versus 69.5% in 2009). Moreover, the Bulletin corroborates the findings of this study, showing that in 2009 the rate of underreporting in Minas Gerais was 92.2%. It is important to consider that the monitoring, detection, tracking and management of transfusion reactions as well as prevention measures have been mandatory in Brazil since 2004^([@r16])^.
Another aspect of non-compliance with the current legislation was the incorrect identification of samples. Although transfusion services with a TC showed a slightly better performance than those without a TC, incorrect identification was observed in approximately two-thirds of the transfusion services studied. Several investigations have shown that errors in sample identification, especially those that occur outside the transfusion service, are one of the commonest causes of switching blood products and, consequently, of severe hemolytic reactions^([@r13],[@r20])^. Therefore, the failures documented here may compromise the safety of blood transfusions, especially because of the risk of incorrect identification of the patient during the infusion of blood products.
Two other aspects of non-compliance with the legislation were also identified: the absence of an individual transfusion chart and the failure to update it. Despite a significantly better performance of transfusion services with a TC, the use of these charts is low even in these services.
Failure to comply with good technical practices, which are required by Hemominas, was also identified. The signing of the transfusion book by the person responsible, which was significantly lower in transfusion services without a TC, was observed in less than half the services with a TC. High rates of non-compliance of the requirements of Hemominas were also observed in terms of deletions in transfusion requests and the presence of a clinical pathology technician to perform pre-transfusion tests.
Decision tree analysis demonstrated a correlation between the presence of a TC and compliance with current technical standards, in particular with aspects of blood transfusion surveillance such as the notification and investigation of the causes of transfusion reactions and the presence and update of individual transfusion charts, suggesting greater concern of these services with transfusion monitoring. However, despite some advances, the performance of TCs is still incipient and the action of these committees should be better organized and more effective in order to promote a better control of the transfusion process and, consequently, to minimize risks.
The present study analyzed aspects of transfusion practice that are regulated by current legislation. Compliance with these aspects is mandatory (correct identification of recipient samples, existence of an individual transfusion chart for each recipient, a correctly and completely filled out transfusion request) and is the responsibility of the technician of the transfusion center. The person responsible should guarantee compliance with these requirements since these measures, complemented by the TC, will contribute to minimize transfusion reactions and to facilitate their screening and identification. The present data will enable state health surveillance agencies and Hemominas, as well as other government and public organs, to adopt more effective strategies to inspect services and implement actions to improve transfusion safety. These measures should not only address the implementation of TCs at all hospitals that use blood products, but also a more effective participation of these committees in the blood transfusion practices of health services.
Conflict-of-interest disclosure: The authors declare no competing financial interest
| {
"pile_set_name": "PubMed Central"
} |
###### Key messages
What is already known about this subject?
=========================================
- Caffeine is a competitive inhibitor of adenosine and can potentially impact on the fractional flow reserve (FFR) results. Patients undergoing adenosine perfusion scintigraphy studies are required to abstain from caffeine for at least 12 hours prior to the test. However, no such guidelines exist for FFR measurements.
What does this study add?
=========================
- Based on our findings it may be hypothesized that sensitivity of caffeine and its inhibitory effect on adenosine can vary between the individuals and has potential to alter the FFR results. Further trials are required to confirm these findings.
How might this impact on clinical practice?
===========================================
- A caffeine history may be considered in people undergoing FFR measurements, particularly in context of an unexpected FFR result.
Introduction {#s1}
============
Fractional flow reserve (FFR)-guided percutaneous intervention (PCI) improves outcomes in patients with stable and unstable angina.[@R1] [@R2] Maximal hyperaemia is paramount for optimal FFR measurements. Intravenous adenosine is the most widely used agent to minimise the coronary micro-resistance and thus induce maximal hyperaemia along with intracoronary nitroglycerine.[@R3]
Caffeine is a methylxanthine commonly found in tea, coffee and chocolate. Caffeine is a competitive inhibitor of adenosine A2a receptor at the cellular level.[@R4] Patients undergoing adenosine perfusion scintigraphy studies are required to abstain from caffeine for at least 12 h prior to the test.[@R5] However, no such guidelines exist for FFR measurements. We hypothesise that caffeine, being a competitive inhibitor of adenosine, would influence adenosine-mediated FFR readings.
Methods {#s2}
=======
Study population {#s2a}
----------------
Consecutive patients undergoing standard of care angiogram and FFR measurements to investigate intermediate coronary artery lesion were recruited to participate. Patients with any contraindications to intravenous adenosine or caffeine were excluded, as well as patients with normal or near normal FFR (\>0.9; [figure 1](#OPENHRT2014000060F1){ref-type="fig"}). Participants were required to abstain from caffeine (tea, coffee and chocolate) for 24 h prior to FFR measurements.
{#OPENHRT2014000060F1}
Angiography and baseline FFR {#s2b}
----------------------------
Angiography was performed via right radial or right femoral approach. After guide catheter placement, intracoronary nitroglycerine (200 μg) was administered and repeat angiographic images were obtained to confirm lesion severity. FFR was measured using a 0.014-inch pressure guidewire (St Jude Medical, Minneapolis, Minnesota, USA) after a 3 min peripheral intravenous infusion of adenosine (140 μg/kg/min). The lowest FFR reading over 3 min of adenosine infusion was recorded. All FFR tracings were analysed to ensure absence of artefacts.
Intravenous caffeine and repeat FFR measurements {#s2c}
------------------------------------------------
A period of 3 min was prescribed to allow postadenosine FFR to return to baseline levels. Intravenous caffeine, roughly equivalent of 3--4 cups of average coffee (4 mg/kg),[@R6] was injected over 1 min after obtaining blood samples for baseline caffeine level. Peak caffeine level blood samples were collected after 5 min. A repeat FFR measurement was then performed utilising the same protocol within 12--15 min of initial FFR measurement ([figure 1](#OPENHRT2014000060F1){ref-type="fig"}).
Statistical analysis {#s2d}
--------------------
All data were analysed using SPSS Statistics Software (IBM, Windows, V.19). Data are presented as values and percentages or mean value±SD. Paired t test was used to compare continuous data. A two-tailed p value of \<0.05 was considered statistically significant.
Ethical consideration {#s2e}
---------------------
Therapeutic Goods Administration (TGA, Australia) exemption was sought and obtained to use intravenous caffeine for non-approved trial indication. Informed written consent was obtained from all patients.
Results {#s3}
=======
Two patients were excluded as their baseline adenosine FFR was \>0.9. A total of 10 patients were enrolled. Baseline characteristics of these patients are described in [table 1](#OPENHRT2014000060TB1){ref-type="table"}. Mean coffee consumption was 2.85±1.02 cups a day. Individual FFR results are summarised in the [table 2](#OPENHRT2014000060TB2){ref-type="table"}. Baseline caffeine levels were undetectable (\<5 μg/mL) in all patients and increased significantly postintravenous caffeine administration (16.4±5.5 μg/mL). Mean caffeine dosage administered was 348±41.3 mg. All patients reported flushing symptoms with initial adenosine infusion which were markedly less with repeat adenosine infusion postintravenous caffeine on subjective assessment.
######
Baseline characteristics
----------------------------------------------------- -----------
Age 59.9±9.4
Males 80%
Weight 87.5±15.6
Hypertension 70%
Diabetes 30%
Smoking 60%
Average coffee consumption 2.85±1.02
ACE inhibitors/angiotensin receptor blocker therapy 60%
β-blocker therapy 40%
Aspirin therapy 100%
Clopidogrel 60%
Ticagrelor 40%
Nitrates 20%
----------------------------------------------------- -----------
Values expressed as mean±SD or percentage (%).
######
Individual results
N Sex Age Vessel Peak caffeine levels. (μg/mL) Baseline FFR (preadenosine) Repeat baseline FFR (postcaffeine, preadenosine) Baseline FFR (postadenosine) Postcaffeine FFR (postadenosine)
---- ----- ----- ------------ ------------------------------- ----------------------------- -------------------------------------------------- ------------------------------ ----------------------------------
1 M 57 Mid RCA 26.4 0.92 0.92 0.80 0.91
2 M 73 Mid LAD 11.8 0.93 0.93 0.77 0.93
3 M 58 Prox LAD 15.5 0.95 0.95 0.85 0.87
4 M 60 Mid RCA 12.6 0.90 0.81 0.74 0.75
5 M 49 Prox LAD 18.2 0.80 0.81 0.69 0.61
6 M 58 Distal RCA 12.4 0.80 0.81 0.69 0.67
7 M 45 Mid LAD 15.3 0.93 0.92 0.86 0.87
8 F 76 Mid RCA 15.9 0.97 0.97 0.81 0.81
9 F 59 Mid LAD 10.5 0.97 0.97 0.87 0.95
10 M 64 Prox RCA 25.4 0.97 0.97 0.82 0.87
FFR, fractional flow reserve.
Baseline preadenosine FFR values were similar before and after caffeine administration (0.91±0.06 vs 0.91±0.07; p=0.41). Postadenosine FFR readings were 0.79±0.07, which non-significantly increased to 0.82±0.11 postcaffeine (p=0.15). Interestingly, significant FFR values (≤0.8)[@R2] for two patients changed to non-significant after caffeine administration (0.77--0.93 and 0.8--0.91; [figure 2](#OPENHRT2014000060F2){ref-type="fig"}). Graphical representation of individual results before and after caffeine is shown in [figure 3](#OPENHRT2014000060F3){ref-type="fig"}. There was no relation noted between average baseline coffee consumption (2.85±1.02 cups) and affect to intravenous caffeine on FFR.
{#OPENHRT2014000060F2}
{#OPENHRT2014000060F3}
Discussion {#s4}
==========
Our study demonstrates higher average FFR readings after caffeine administration in 7 (70%) patients using intravenous adenosine when compared with baseline readings. Although this difference did not reach statistical significance over all, it potentially can be clinically relevant in selected individuals, as two patients' results changed from positive[@R2] (≤0.8) to negative (\>0.8). Baseline (preadenosine) readings for both FFR measurements remained unchanged suggestive of caffeine being the causative factor for the change. FFR readings of two patients were decreased after caffeine, the reason of which is unclear. These two patients had significantly positive FFR readings at baseline, which despite the change remained positive after caffeine.
A previous study to elucidate the effect of caffeine on FFR did not show a difference using *intracoronary* adenosine as a hyperaemic agent.[@R4] Although there have been a multitude of studies comparing intravenous and intracoronary adenosine, intravenous adenosine is the most widely used method for inducing maximal hyperaemia.[@R3]
The competitive inhibition of adenosine A2a receptors with caffeine has been studied in a number of adenosine nuclear perfusion studies with mixed outcomes.[@R7] [@R8] Higher doses of adenosine (210 μg/kg/min) have been used by few to overcome the effect of caffeine on adenosine nuclear perfusion studies.[@R8] Though our study did not include higher dose adenosine, it is plausible that similar effect may be seen in FFR readings. Further studies are needed to evaluate that.
Our study is limited by power to achieve any definitive conclusive results. Furthermore, orally consumed coffee or tea has other metabolites such as theophylline and theobromine that can have vasodilatory effects in addition to their adenosine inhibition similar to caffeine. Our study used intravenous caffeine administered over a short duration of time and may not have accounted for any additional effects of these metabolites on FFR. Another weakness of the study is that only 4 out of 10 patients had an FFR value ≤0.8. It does however raise questions which require larger trials to clarify. The present practice across majority of cardiac catheterisation laboratories for elective FFR or PCI is though variable, mostly involves a fasting period of \>4 h, and caffeine history is not routinely obtained. Based on our findings, it may be hypothesised that sensitivity of caffeine and its inhibitory effect on adenosine can vary between the individuals and has potential to alter the FFR results. A caffeine history may be considered particularly in context of an unexpected FFR result.
Conclusion {#s5}
==========
Caffeine may affect FFR results in some patients. A larger scale trial is needed to clarify the extent and magnitude of caffeine/adenosine interaction particularly due to ubiquitous nature of caffeine and increasing importance of FFR in clinical practice.
**Funding:** This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
**Competing interests:** None.
**Ethics approval:** The study complied with the declaration of Helsinki and was approved by institutional ethics committee.
**Provenance and peer review:** Not commissioned; externally peer reviewed.
**Data sharing statement:** No additional data are available.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION {#sec1-1}
============
Tooth caries is a chronic disease possibly afflicting the erupted teeth of preschool-age children, which can cause pain and even death.\[[@ref1]\] *Streptococcus mutans* is a cariogenic bacterium and a predominant etiological factor in the incidence of tooth caries, which is found in large quantities in biofilm environments.\[[@ref2]\] Causality between the incidence of caries and the substantial number of *S. mutans* has been demonstrated by several studies confirming caries development to have been preceded by *S. mutans* colonization,\[[@ref3]\] while others suggested that salivary levels of *S. mutans* are an indication of caries progressivity, as confirmed by Luthfi *et al*. who showed that *S. mutans* level in caries-free children is lower than those in caries children with decayed, extracted, filled-teeth (def-t) index \>6.\[[@ref4]\]
In saliva, neutrophil is the prominent first-line defense of immune cells against pathogen microbes, which is proven by the identification of several severe diseases in human due to the dysfunction of neutrophil or due to the disturbance of its level as found in neutropenia cases.\[[@ref5]\]
Neutrophil, as a defense against the inflammation mediator secreted by the microbes, performs chemotaxis and phagocytosis and eliminates the microbes by releasing human antimicrobial peptides (AMPs) (HNP 1--3) as nonoxidative antibacterial mechanism.\[[@ref6]\]
Neutrophil consists of three main granules and azurophilic being one of those which contains HNP 1--3 secreted in saliva from gingival crevicular fluid as antimicrobial defensin.\[[@ref7]\] HNP 1--3 has a protective role against oral pathogen microbes by controlling the biofilm formation, which effectively prevents the occurrence of oral infection and defending the tooth structure against caries.\[[@ref8]\]
HNP 1--3 secretion occurs as the consequence of neutrophil activation that leads to the degranulation of azurophilic granules which can be detected in CD63 markers on the surface of neutrophil. CD63 is a membrane protein largely found in azurophilic granules which plays a role in controlling the quality of antimicrobial protein.\[[@ref9]\] To gain additional information from this perspective, the correlation between CD63 expressions on the surface of salivary neutrophil with HNP 1--3 secretion in early childhood caries (ECC) was analyzed.
MATERIALS AND METHODS {#sec1-2}
=====================
Experimental design {#sec2-1}
-------------------
This research represented a cross-sectional, analytical observational study conducted on healthy, preschool-age children both male and female aged between 4 and 6 years with or without caries. The exclusion criteria of this research were children with or without caries below 4 years and above 6 years, children with blood and saliva disorders, children under antibiotic and corticosteroid medication which can suppress immune response, and children with erupted permanent teeth. This research was approved by the Ethical Committee of Universitas Airlangga with Laic Ethic No. 01/KKEPK/VII/2013. Informed parental consent was acquired before the sampling being conducted.
This study incorporated the use of random sampling of 4--6-year-old students from four Surabaya-based kindergartens. This age group was selected considering their caries-susceptible condition as the immune response functions have not been optimal yet in eliminating pathogen microorganism, resulting in the high prevalence of caries. Assessments of the dental caries were conducted in forty students by one dentist using disposable mouth mirrors for indirect vision and torchlight and performed by measuring the def-t. Dental explorer was used to confirm visual evidence of caries on the teeth surfaces. During the examination, the children were seated on a chair, across the examiner and two assistants, who sit close to the examiner to record the data and prepare the equipment. The participants were subsequently divided into two groups based on the def-t index: a caries-free group (def-t = 0) and an ECC group (def-t ≥ 6).
Isolation of salivary neutrophils {#sec2-2}
---------------------------------
Salivary neutrophils were obtained from participants by rinsing their oral cavities with 10 mL of sterile and 1.5% NaCl solution while they gargled without swallowing for 30 s before expectorating the resulting fluid into a sterile glass -- a procedure repeated four times. The collected solution was centrifuged (15 min at 450 g) at 4°C and the pellets then mixed with 2 mL of Roswell Park Memorial Institute medium. An EasySep™ Human Neutrophil Enrichment Kit (STEMCELL Singapore PTE Ltd., Singapore) was used to isolate salivary neutrophils. Cell suspension with a concentration of 5 × 10^7^ cells/mL was placed in 5 mL (12 mm × 75 mm) polystyrene round-bottom tube with 50 μL/ml of EasySep^®^ neutrophil cell cocktail subsequently being added. The resulting mixture was incubated at 4°C for 10 min. Three Mix EasySep^®^ nanoparticles were then used by pipetting up and down more than five times. Thereafter, 2 mL of cells was added to 200 μL of nanoparticles, mixed, and incubated at 4°C for 10 min. 2.5 mL of the cell suspension was then placed in the tube (without a cap) with a magnet, mixed, and incubated for 5 min at room temperature. At the next stage, the EasySep^®^ magnet was removed in one continuous motion. The tube remained inverted for 2--3 s before being returned to the upright position. The empty tube was subsequently removed from the EasySep^®^ magnet, replaced with a new tube containing supernatant fraction, placed on the magnet, and left to stand for 5 min. Isolated cells in the new tube were, at that point, ready for use.
Determination of human neutrophil peptide 1--3 secretion {#sec2-3}
--------------------------------------------------------
Saliva samples were analyzed using an ELISA kit (Hycult Biotech, Uden, Netherlands) to determine the extent of HNP 1--3 secretion. Samples were diluted at least five times with sample dilution buffer in polypropylene tubes. About 100 μL of duplicate standard and samples were added to the well, covered, and incubated for 1 h at room temperature. The plates were then washed four times with a wash buffer using a plate washer. About 100 μL of diluted tracer was added to each well, covered, incubated for 1 h at room temperature, and washed four times. About 100 μL of diluted streptavidin--peroxidase was added to each well, with the tray being subsequently covered, incubated for 1 h at room temperature, and washed four times. About 100 μL of TMB substrate was added to each well, with the tray being subsequently covered, incubated for 30 min at room temperature, and covered again with aluminum foil. About 100 μl of stop solution was added to stop the reaction with the solutions in the wells being thoroughly mixed by gently swirling the plate. The plate was read within 30 min after the addition of stop solution at 450 nm using a plate reader.
Flow cytometry examination to detect CD63 expression on salivary neutrophil {#sec2-4}
---------------------------------------------------------------------------
Isolated cell suspension from the NaCL 1.5% rinse was inserted into the microtube containing 500 μl of phosphate buffered saline (PBS), before being centrifuged at 2500 rpm for 5 min at 4°C. Acquired pellets were then stained using 50 μl of conjugated BioLegend antihuman α-CD63FITC, and BioLegend α-PIPE (BioLegend, San Diego, USA) conjugated antibodies at a ratio of 1:200 to PBS.
Stained cells were stored at a temperature of 4°C for 30 min before the suspension was added to 1 ml of PBS and recentrifuged for 5 min. About 100 μl of BioLegend Cytofix Cytoperm (BioLegend, San Diego, USA) was added to the pellets and homogenized. After incubation in a darkened environment at 4°C for 20 min, 1 ml BioLegend Washperm 1X (BioLegend, San Diego, USA) was added and the suspension was recentrifuged at 2500 rpm for 5 min. With the addition of conjugated BD antihuman α-CD64PerCP (Becton Dickinson, New Jersey, USA) as an intracellular antibody in the pellets, the suspension was placed into a flow cytometry cuvette with 300 μl PBS which was then attached to the nozzle to be analyzed by means of an FACSCalibur flow cytometer BD (BioSciences, San Jose, USA).
The neutrophil gate was identified by density and size using side-angle light scatter followed by forward-angle light scatter. This compensation was reached by means of fluorescein isothiocyanate (FITC) and phycoerythrin (PE) labeled with individual antibodies. The results were presented as mean fluorescence intensity and analyzed using FACSCalibur (Becton Dickinson, New Jersey, USA) with a program, CellQuest software.
Statistical analysis {#sec2-5}
--------------------
The results were listed as the mean ± standard deviation. All statistical analyses were performed using SPSS 20 (IBM, New York, USA). The statistical difference was analyzed by *t*-test. Correlation analyses were performed using Pearson correlation with *P* \< 0.05 being considered to be significant.
RESULTS {#sec1-3}
=======
CD63 expression was detected using flow cytometry \[[Figure 1](#F1){ref-type="fig"}\]. Both salivary HNP 1--3 level and activated salivary neutrophil (CD64) which expressed CD63 were analyzed using *t*-test and showed a significant difference in HNP 1--3 level between the caries-free group and the ECC group as shown in Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}.
{#F1}
######
Mean and standard deviation human neutrophil peptide 1-3 level in the saliva of preschool-age children who were caries free and with early childhood caries (pg/ml)
**Group** ***n*** **Mean±SD** **95% CI** ***P***
------------- --------- -------------- --------------- ----------
Free caries 20 140.39±31.91 125.45-155.45 0.009
ECC 20 172.6±41.64 153.27-192.25 (*P*\<α)
SD: Standard deviation; ECC: Early childhood caries; CI: Confidence interval
######
Mean and standard deviation activated salivary neutrophil (CD64) which expressed CD63 in saliva in preschool-age children who were caries free and with early childhood caries (%)
**Group** ***n*** **Mean±SD** **95% CI** ***P***
------------- --------- ------------- ------------ ----------------
Free caries 20 2.67±0.46 2.37-2.96 0.009 (*P*\<α)
ECC 20 2.32±0.57 1.96-2.68
SD: Standard deviation; ECC: Early childhood caries; CI: Confidence interval
The correlation between CD63 expression in the salivary neutrophil surface and HNP 1--3 secretion in the saliva of caries-free children is found to be stronger than their correlation in ECC children as shown in [Table 3](#T3){ref-type="table"}.
######
Correlation test of CD63 expression and human neutrophil peptide 1-3 secretion in saliva in preschool-age children who were caries free and those with early childhood caries
**Independent variable** **Dependent variable** **Group** ***R*^2^** ***B*** ***P***
-------------------------- ------------------------ ------------- ------------ --------- ----------
HNP 1-3 CD63 Free caries 0.919 50.918 \<0.0001
ECC 0.882 69.808 \<0.0001
HNP 1-3: Human neutrophil peptide 1-3; ECC: Early childhood caries
DISCUSSION {#sec1-4}
==========
Saliva is an important element in the prevention of caries. Salivary proteins have a potential role in inhibiting caries development as a result of the AMPs, such as HNP 1--3, human β-defensins 1--3, cathelicidin (LL-37), and histatin, which it contains. These peptides demonstrate broad-spectrum antimicrobial activity and are derived from several sources, including salivary gland acini and ducts, oral mucosa epithelial, and neutrophils. In the oral cavity, defensin and LL-37 have considerable potential to inhibit various strains of bacteria, while histatin acts as an antifungal agent.\[[@ref10]\]
Microbial infection in oral mucosa induces an extensive number of AMPs, including defensin, which does not merely act as antibiotics but also carries out immunomodulatory activity ranging from the innate to the adaptive.\[[@ref11]\] Innate immunity is the primary defense system against the infection by detecting the component of the microbes, such as LTA, which is found in the cell wall *S. mutans*, through toll-like receptors. This leads to the activation of complex cell signaling network and results in the production of various effector molecules, including HNP 1--3, which regulates the host immune response.\[[@ref12]\] HNP 1--3 can increase the innate immunity by stimulating pro-inflammatory response through increasing the mobilization of neutrophil to the infection area, generating reactive oxygen species, facilitating the phagocytosis process, and inducing pro-inflammatory cytokine.\[[@ref13]\]
This research confirmed that HNP 1--3 levels in ECC were higher than those in caries-free situations. This finding challenged the study conducted by Tao *et al*. which stated that salivary HNP 1--3 levels in children with active caries are proven to be lower than those in their caries-free counterparts.\[[@ref14]\] This difference was possibly due to the age range of the participants observed by Tao *et al*. being approximately 11--15 years since different age groups would present contrasting immune response maturity levels. Age variation plays an important role in immune response regulation. Children and the elderly are more prone to infection than other age groups.\[[@ref15]\] Immune cells in adults showed a stable state in protein serum.\[[@ref16]\] As this research was conducted in Indonesia while that of Tao *et al*. was carried out in the USA, it was undoubtedly that a different genetic and demography environment would affect the immune response maturity of each individual. Genetic factors known to be predisposing factors of infection, for instance, immunodeficiency manifestation are possibly caused by an heredity-specific immune system.\[[@ref17]\] Environmental factors also play a role in affecting the development and regulation of the immune system.\[[@ref18]\]
The high level of HNP 1--3 in ECC demonstrated that it could not protect the host from *S. mutans* virulence, probably because the cariogenicity of the *S. mutans* strain contained in the saliva of ECC was more pronounced than in the caries-free research participants. To induce an infection, bacteria must demonstrate resistance to AMP, as found in *S. mutans*. *S. mutans* is Gram-positive bacteria which could inhibit the intrusion of HNP 1--3 as its cell wall consists of a thick peptidoglycan crosslinking with LTA polymer, to prevent the AMP activation.\[[@ref19]\]
Neutrophil consists of three main granules which act as nonoxidative antimicrobials: primary granules (azurophilic granules), secondary granules (specific granules), and cytoplasmic antimicrobial protein. Antimicrobial substances found in primary granules include: defensin (known as HNP 1--3), elastase, collagenase, proteinase, myeloperoxidase, lysozyme, bacterial permeability factor, and cathepsins. Secondary granule proteins include lysozyme, lactoferrin, and collagenase.\[[@ref20]\] These granules interact with phagocytic vesicles resulting in the release of its contents into phagosomes with engulfed microbes.\[[@ref21]\]
Neutrophil is the first innate immunity cell to be found in the infection site as a response to the pathogen microbes. This cell was recruited to opsonize the bacteria through Fc-gamma receptors and C-type, which results in phagosome formation and the release of various antimicrobial proteins and intracellular enzymes.\[[@ref22]\]
Based on the results of this study, the expression level of CD63 in ECC children was found to be lower than that in caries-free children. This could happen because *S. mutans* is internalized by neutrophils through a phagocytosis process mediated by FcαR (CD89) or CR1 (CD35) which facilitates the development of three defense mechanism strategies to avoid being phagocytosed. These include escaping from host phagosome, blocking the phagosome-lysosome fusion, and promoting some form of mechanism which allows itself to live in the phagolysosomes.\[[@ref23]\]
Lysosome-associated membrane protein-2 has an important role as innate immunity to protect the oral cavity by preventing biofilm formation.\[[@ref24]\] LAMP-2 may be involved in lysosome biogenesis as the final process of phagocytosis and required in phagosome maturation.\[[@ref25]\] A study by Beertsen *et al*. showed that LAMP-2 deficiency leads to insufficient maturation and antimicrobial activity of phagosome that increases susceptibility to oral cavity infection.\[[@ref24]\] The low expression of CD6 in ECC-affected children possibly occurred because of LAMP-2 deficiency as this condition was found in mice with LAMP-2 deficiency-induced periodontitis.\[[@ref24][@ref26]\]
The higher level of salivary HNP 1--3 found in ECC children compared to those who are caries free showed a contrasting result to that of CD63 expression, considering that CD63 is an azurophilic granule marker which produces HNP 1--3. One possible explanation is that HNP 1--3 secretion contained in saliva is not only produced by salivary neutrophil but also by oral mucosa epithelial cells, specifically mandibular salivary duct epithelial cells.\[[@ref14]\] During the caries infection caused by *S. mutans*, the host will increase HNP 1--3 production by epithelial cells and salivary neutrophil cells, resulting in high levels of HNP 1--3 in the saliva of ECC children compared to that of caries-free children. High levels of HNP 1--3 have been proven incapable of killing potentially mutated and HNP 1--3-resistant *S. mutans* pathogens.\[[@ref27]\] It is also possible that the virulent *S. mutans* possesses PhoP-PhoQ homolog, as found in Gram-negative bacteria, which enables *S. mutants* to survive high levels of HNP 1--3.\[[@ref28]\] *S. mutans* also a Gram-positive bacterium which is able to change and modify its cell wall consists of teichoic acids to be more virulent and resistant to antimicrobial component and immune response activation.\[[@ref29][@ref30]\]
CONCLUSION {#sec1-5}
==========
There is a reverse correlation between HNP 1--3 secretion in saliva and CD63 expression on the surface of salivary neutrophil of ECC children proven by the increase of HNP 1--3 secretion, while the expression of CD63 was found to decrease. This finding could be used to develop a potential early detection assessment of caries risk in children to advance the caries prevention strategies in the Southeast Asia population.
Financial support and sponsorship {#sec2-6}
---------------------------------
The authors would like to thank Directorate of Research and Community Services of Directorate General of Research and Development Strengthening from Ministry of Research, Technology and Higher Education of the Republic of Indonesia for the grant funding provided for this research.
Conflicts of interest {#sec2-7}
---------------------
The authors of this manuscript declare that they have no conflicts of interest, real or perceived, financial or non-financial in this article.
The authors would like to thank Prof. Muhaimin Rifa\'i, PhD. Med. Sc for the help in conducting this research.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-diagnostics-10-00484}
===============
A ganglion cyst is a common soft tissue mass, usually forming over a joint or tendon, and commonly developing in the wrist on either the dorsal or volar side \[[@B1-diagnostics-10-00484]\]. It contains a thick and gel-like fluid with a connective tissue capsule, and is known to usually originate from the joint capsule or tendon sheath \[[@B1-diagnostics-10-00484]\]. It seems that the synovial fluid within the joint or tendon leaks out, is collected, and forms a cyst. In clinical practice, ganglion cysts attached to the joint or tendon are frequently encountered. However, an intramuscular ganglion cyst without connection to the joint or tendon has rarely been reported \[[@B2-diagnostics-10-00484]\]. In addition, its aetiology has not been clearly elucidated.
In the present study, we describe a case of an intramuscular ganglion cyst in the flexor hallucis brevis muscle (FHB) that arose secondary to a muscle tear. Moreover, through this study, we suggest a possible aetiology for the development of intramuscular ganglion cysts.
2. Case Presentation {#sec2-diagnostics-10-00484}
====================
A 50-year-old woman visited the department of physical medicine and rehabilitation at a university hospital because of acute pain and swelling over the right mid-plantar area for 2 days, which occurred after prolonged kneeling for scrubbing floors. She mentioned that she kneeled with hyperextended toes on her right side for about 3 h, with sudden sharp pain and progressive swelling of the foot developing after work. On physical examination, a focally swollen and mildly erythematous right medial mid-plantar area was noted without ecchymosis ([Figure 1](#diagnostics-10-00484-f001){ref-type="fig"}).
Palpation revealed heat and tenderness along the right medial mid-plantar area near the first metatarsal shaft, but no palpable mass. There was mild weakness of right big toe flexion. Ultrasonography examination was arranged to evaluate the plantar area. Three days after her first visit to our hospital, ultrasonography (12-MHz linear probe, Toshiba, Tokyo, Japan, Aplio 500) of the right medial mid-plantar area revealed a focal disruption of the fibrillary structure of the FHB with an anechoic defect in the mid-portion of the muscle ([Figure 2](#diagnostics-10-00484-f002){ref-type="fig"} and [Video S1 in the Supplementary Materials](#app1-diagnostics-10-00484){ref-type="app"}).
The focal defect was confirmed by its compressibility and visibly retracted torn fibres during compression. On the basis of the finding from the ultrasonography examination, a diagnosis of partial tear of the right FHB was made.
At 24 days after the onset of the symptoms, on magnetic resonance imaging (MRI), a smooth, well-circumscribed, unilocular and homogeneously T2-hyperintense lesion (size, 10.0 × 20.9 × 9.1 cm) was found in the mid-portion of the FHB ([Figure 3](#diagnostics-10-00484-f003){ref-type="fig"}). The MRI finding consisted of a ganglion cyst \[[@B2-diagnostics-10-00484]\].
In addition, on the follow-up ultrasonography examination 54 days after the onset of the patient's symptoms, at the same location where a tear was seen on the initial ultrasonography examination, an anechoic defect in the mid-portion of the FHB was observed, filled with a well-defined, unilocular, avascular, anechoic ovoid mass (size, 15.5 × 22.8 × 9.3 cm) with posterior acoustic enhancement, which is the typical appearance of a ganglion cyst ([Figure 4](#diagnostics-10-00484-f004){ref-type="fig"} and [Video S1 in the Supplementary Materials](#app1-diagnostics-10-00484){ref-type="app"}) \[[@B3-diagnostics-10-00484]\]. The patient's ganglion cyst was not connected to any tendon or joint. Her pain was significantly reduced to the endurable level; thus, we did not perform an aspiration of the ganglion cyst.
3. Discussion {#sec3-diagnostics-10-00484}
=============
In this study, we presented a rare case of disruption of the flexor hallucis brevis muscle with subsequent development of an intramuscular ganglion cyst, with follow-up evaluation with MRI and ultrasonography.
MRI has been shown to have a high sensitivity and specificity, 94.7% and 94.4%, respectively, in accurately diagnosing ganglion cysts \[[@B4-diagnostics-10-00484]\]. The data regarding the accuracy of ultrasonography in diagnosing ganglion cysts based on post-surgical pathology analysis is scarce, but a retrospective analysis of 106 patients revealed ultrasonography had a 69% sensitivity and 100% specificity \[[@B5-diagnostics-10-00484]\]. However, no ganglion cysts were reported on the plantar aspect of the foot in a retrospective analysis of 101 patients surgically treated for foot masses \[[@B6-diagnostics-10-00484]\]. Therefore, even though some authors have reported US-guided aspiration of ganglion cysts as a potential alternative to surgery \[[@B7-diagnostics-10-00484]\], clinicians should be very cautious in performing it on lesions in atypical locations.
In most cases, the ganglion cyst can develop from the joint or tendon sheath \[[@B1-diagnostics-10-00484]\]. However, rarely, a ganglion cyst originates from the muscle without connecting to the adjacent joint capsule or tendon sheath. To the best of our knowledge, seven previous studies reported ganglion cysts confined to a muscle portion with no connection to other adjacent structures \[[@B1-diagnostics-10-00484],[@B8-diagnostics-10-00484],[@B9-diagnostics-10-00484],[@B10-diagnostics-10-00484],[@B11-diagnostics-10-00484],[@B12-diagnostics-10-00484],[@B13-diagnostics-10-00484]\]. In these studies, intramuscular ganglion cysts were developed in the quadriceps, gastrocnemius, extensor digitorum and biceps brachii. However, no previous study described the primary cause or possible mechanism of the occurrence of the intramuscular ganglion cyst. In our study, with a follow-up examination, we found that the intramuscular ganglion could develop after muscle injury.
The exact mechanism underlying the development of the intramuscular ganglion cyst after muscle tear is unclear. However, on the basis of previous studies, we can suggest a possible mechanism. Some studies reported that the ganglion cyst is thought to arise from the myxoid degeneration of connective tissue that developed from defects of the joint capsule or tendon sheaths \[[@B14-diagnostics-10-00484]\]. Likewise, we believe that the pieces of torn muscle tissues turn into myxoid fluid and form an intramuscular ganglion cyst in the empty space created after the tearing of the muscle.
In conclusion, we report the case of an intramuscular ganglion cyst in the FHB. We demonstrated that the ganglion cyst could potentially develop after partial muscle tear by follow-up examination with MRI and ultrasonography. In addition, we suggest the myxoid degeneration of fragments of torn muscle as a possible mechanism of the development of the intramuscular ganglion cyst after muscle tear. However, because this study only involves a single case, further studies including a larger number of cases are necessary.
We gratefully acknowledge the patient for his participation and consent. This study was approved by the Institutional Review Board of Yeungnam university hospital (number 2020-07-031).
The following are available online at <https://www.mdpi.com/2075-4418/10/7/484/s1>, Video S1.
######
Click here for additional data file.
M.-Y.H. was the doctor in charge of the patient and carried the ultrasonographic examinations; M.C.C., M.B.-R., and M.-Y.H. participated in the design of the manuscript; M.B.-R., M.-Y.H. and M.C.C. conceived the figures and the video. All authors participated in the revision of the manuscript and figure. All authors have read and agreed to the published version of the manuscript.
The present study was supported by a National Research Foundation of Korea grant funded by the Korean government (grant no. NRF-2019M3E5D1A02068106).
The authors declare no conflict of interest.
{#diagnostics-10-00484-f001}
{#diagnostics-10-00484-f002}
{#diagnostics-10-00484-f003}
{#diagnostics-10-00484-f004}
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Non-communicable diseases (NCDs), such as cardiovascular diseases, diabetes, chronic obstructive pulmonary diseases and cancer, are the main focus of the Sustainable Development Goals (SDGs) ^[@ref-1]^. Target 3.4 of the SDGs intends to decrease premature death through NCDs by one third up to 2030. The four main risk factors mentioned are smoking, 'unhealthy' diets, physical inactivity and 'harmful' use of alcohol. Governments and public health authorities are encouraged to enforce so called [16 'best buy'](https://www.who.int/ncds/management/WHO_Appendix_BestBuys_LS.pdf) strategies to reach the target. While it is well accepted that tobacco smoking and alcohol abuse are harmful for health and there is general agreement not to smoke and at least to refrain from too much or excessive alcohol consumption, campaigning against 'unhealthy' diets is more problematic. This is because of the complexity of how nutrition relates to health. The 'global obesity pandemic' ^[@ref-2]^ not only is caused by a surplus of energy in the diet, but it's link to diabetes ^[@ref-3]^ and cancer ^[@ref-4],\ [@ref-5]^ and other diseases is driven by complex metabolic pathways ^[@ref-6]--\ [@ref-9]^. Not only does overnutrition play a role in the development of cardiovascular diseases, cancer, diabetes and influences aging, but the risk for obesity is also related to genetic factors ^[@ref-10],\ [@ref-11]^. Through epigenetic pathways under- and overnutrition of pregnant females might result in diabetes, cancer and cardiovascular diseases in their adult children ^[@ref-12]--\ [@ref-15]^. Within the field of nutritional sciences matters are further complicated by recent developments of contradictory nutritional conceptions. There seems to be disagreement about what a healthy and appropriate diet should be and controversial opinions are justified by supportive investigations into molecular factors from both sides when either a 'low fat high carbohydrate-' or a 'low-carbohydrate, high fat diet' are promoted ^[@ref-16]^.
The role of molecular epidemiology in public health actions against overnutrition
=================================================================================
Governmental authorities and public health institution, overseeing the wellbeing of the population, cannot capitulate and stop promoting 'healthy' nutrition, in view of the constraints. So far major public health tools for working against NCDs, in regard to nutrition, are encouragement of [physical activity](https://www.who.int/news-room/fact-sheets/detail/physical-activity), trying to influence behavior and practice through sophisticated methods of social sciences ^[@ref-17]^, and to increase taxes on harmful products such as sugar and soft drinks with the intention to reduce consumption ^[@ref-18]^. Benefit and drawbacks of these methods are not questioned here, but it has been argued by Slattery (2002) that 'while working at the population level of exploration, molecular epidemiology must incorporate knowledge from many disciplines to obtain an understanding of the organism, the system and the cell. Translating complex disease pathways into relevant public health messages should be the goal and the result of the art of epidemiology' ^[@ref-19]^. Far too often interesting results within the field of genetics, insight into metabolic pathways and molecular components, such as enzymes and cytokines, escape the intention of those working in public health. This is plausible, as investigations using laboratory models, worms and rodents, are by no means research tools for public health. Studies exploring epigenetic effects, such as DNA methylation, as recently conducted in Estonia ^[@ref-20]^ are most probably not feasible for large scale epidemiological studies in low- but also middle- and high middle income countries. However, low- and middle-income countries need to make use of advances in public health. In particular low-, but especially high middle-, income countries should have the means to follow up developments in molecular epidemiology in order to have a deeper understanding of the nature of NCDs, and should conduct as much population-based research as possible, including the use of promising biomarkers to give an insight into genetic and metabolic pathways.
In fact, besides anthropometry measurements, a number of clinical laboratory methods and biomarkers are already in common use and in future, epigenetic and molecular epidemiology could be additional suitable aspirants for population-based studies ^[@ref-19],\ [@ref-21],\ [@ref-22]^. However, multiple constraints in the use of biomarkers, as outlined for metabolic syndrome (MetS) should be considered.
Pros and cons of biomarkers -- the examples of the components of the metabolic syndrome
=======================================================================================
In assessing the nutritional status using the body mass index (BMI) and other measurements as independent variables, MetS is frequently used as dependent variable since it incorporates a number of factors related directly or indirectly to NCDs. Variables associated with the syndrome include elevated blood glucose, dyslipidemia, abdominal obesity and high blood pressure. There are five different definitions of MetS, with different thresholds of its components. Because of this ,and other controversial arguments, attempts have failed to agree on either version ^[@ref-23]^. As a compromise the use of the so called 'harmonized version' of MetS has been recommended ^[@ref-24]^. It is now considered that study participants categorized as exposed to MetS should be selected if they display three or more of the five criteria. However, this results in arbitrary groupings of individuals belonging to the 'MetS group', and individuals are integrated with one or two factors of MetS but less than three into the 'non-MetS' group. An example of using the 'harmonized version' MetS version in grouping study participants into the MetS- and the non-MetS group is given in Table 2 of a recent publication ^[@ref-25]^. To apply the 'harmonized version' weakens the validity of MetS as a dependent variable. The 'nature' of MetS as a 'syndrome' is also questioned. The term 'syndrome' should be used in case 'the whole is greater than the parts' ^[@ref-26]^, but this is doubtful ^[@ref-27]^ and before selecting MetS as independent variable it should be considered that a number of factors influence MetS such as age, sex and ethnicity.
Biomarkers representing key factor for the metabolism
=====================================================
As mentioned above, the use of MetS seems to be problematic and the recommendation of 'waist circumference' as a strong indicator of obesity and 'insulin resistance' as one of the metabolic key factors, could be a worthwhile alternatives for MetS ^[@ref-24]^. Waste circumference is a good measurement for overnutrition, because energy intake in excess is stored in the abdominal fat tissue. The adipokines of the fat tissue, by excreting inflammatory molecules, increase the risk to develop diabetes and cancer. Insulin resistance is 'the intersection' either for the way to health or to metabolic disturbances. The absence of a general standard for waist circumference, however, is a disadvantage, and waist circumference needs to be standardized for different population groups ^[@ref-28]^, but with the homeostatic model assessment (HOMA) on hand, a method is available for estimating insulin resistance in epidemiological studies ^[@ref-29]^.
Trying to find and test biomarkers mirroring key metabolic steps for health and disease should be one major objective for molecular epidemiology. It is equally important to gain insight into the interaction of anabolism with catabolism. A candidate for the latter aspect is α-2-macroglobulin (α2M). The importance of α2M has been mentioned by Ohlsson 1972, stating that 'α2M' may have a key role in the body's protection against autodigestion (cited by Schelp FP *et al.* ^[@ref-30]^), which implies that a complete deficiency will not be compatible with life. Within human plasma α2M is the largest non-immunoglobulin, and an almost omnipotent inhibitor of endopeptidases with a unique way to deactivate proteinases ^[@ref-31]^. The inhibitor is found in all mammals, and the biological significance for growth and differentiation can be judged by its presence during embryogenesis, pregnancy, childhood and in aging ^[@ref-32],\ [@ref-33]^. A comprehensive overview about the molecular structure of α2M, mechanism of action, function and pathophysiology is given in the review article from Rehman *et al.* (2013) ^[@ref-34]^.
α-2-Macroglobulin in health and disease
=======================================
Besides the attention molecular biology has given to α2M, in clinical settings the inhibitor was found to be related to the development of Alzheimer's ^[@ref-35]^. Low α2M levels were observed in some patients with lung diseases ^[@ref-36]^, and in advanced prostate cancer ^[@ref-37]^, while in breast- ^[@ref-38]^ and bladder cancer ^[@ref-39]^ elevated levels of α2M were observed. It has been hypothesized that induced increase of endogenous proteinase inhibitors is protective against cancer ^[@ref-40]^. The assumption, among others, was based on the 'fat-related-cancer' hypothesis ^[@ref-41],\ [@ref-42]^, and the low risk of vegetarians for cardiovascular diseases and cancer ^[@ref-43]^, as well as the finding that α2M concentrations and other proteinase inhibitors were higher in Thai vegetarians compared with omnivores ^[@ref-44]^. It was argued that the balance between proteinases and their inhibitors are regulators of tumor growth. The role of proteinase inhibitors in connection with cancer protection considered a number of different inhibitors, and the specific role of α2M remains vague since α2M is incorporated in normal but not in tumor cells. In laboratory mice alpha macroglobulin is active through 'biomediators' but not through its inhibitory capacity ^[@ref-45]^. So far it was concluded that α2M has a role in controlling normal but not malignant growth ^[@ref-46],\ [@ref-47]^.
The role of α-2-macroglobulin in maintaining homeostasis in 'dietary restriction'
=================================================================================
Obesity, as outlined above, is detrimental for health but 'dietary restriction', defined as 'reduced food intake by avoiding malnutrition', extends life span in animals and humans ^[@ref-48]--\ [@ref-50]^. It has been demonstrated recently, by using a mouse model, that epigenetic modification in dietary restriction delayed aging and changed the gene expressions of the lipid profile ^[@ref-51]^. 'Subclinical undernutrition' in preschool children could reflect 'dietary restriction without malnutrition'. The two forms of the condition are 'wasting', a deficit in weight for height and 'stunting', a deficit in height for age, adjusted to a standard ^[@ref-52]^. Children categorized as wasting and stunting are apparently healthy without clinical signs of undernutrition. A number of investigation in Bangkok and villages in rural Thailand disclosed elevated α2M levels and low 3-methylhistidine (3MH) urine excretion in healthy, age matched, village children in comparison to their Bangkok counterparts ^[@ref-53]^. 3MH is supposed to reflect muscle breakdown ^[@ref-54],\ [@ref-55]^. A similar result was obtained when comparing normal nourished preschool children with those deficient in weight for height. α2M serum concentration in the marginal nourished children increased over their well-nourished village counterparts ^[@ref-56]^. In an animal experiment with laboratory growing rats under a marginal diet, with altered protein and energy content, serum proteinase inhibitors increased and 3HM decreased ^[@ref-57]^. The results of the investigations support the hypothesis that in the situation of 'dietary restriction' proteinase inhibitors, including α2M, decrease muscle catabolism, and in the case of the village children kept them healthy though 'an optimal relationship between catabolism and synthesis, thus resulting in stunting' ^[@ref-58]^. In case 'marginal nutritional intake', results in elevated α2M serum concentrations, as expected, overweight and obese Thai adults in Bangkok, showed lower α2M serum levels compared with normal individuals. The proteinase inhibitor selected as dependent variables in a multiple regression, resulted in a model including age, sex HDL cholesterol and BMI ^[@ref-59]^. A similar result was obtained studying hard working male construction laborers, in that a negative correlation were found for the variables age, weight, height, BMI, arm- and midarm circumference, triceps skinfold and HDL with α2M as the dependent variable ^[@ref-60]^. α2M of female construction workers did relate to any of the variables investigated. A dietary survey conducted with apparently health Thai farmers found a statistically significant negative correlation of α2M with energy, protein, fat and carbohydrate intake ^[@ref-61]^. All the results obtained from the variety of different studies seem to be in accordance with the assumption that α2M supports homeostasis in situations of a 'challenged' nutritional status.
α-2-Macroglobulin in protein-energy-malnutrition
================================================
All the results obtained from the variety of different studies, as reviewed here, seem to be in accordance with the notion that α2M supports homeostasis in situations of a 'challenged' nutritional status and suggest that proteinase inhibitors play a key role in maintaining the metabolism in balance. This also was assumed in observing patients suffering from clinical protein-energy malnutrition (PEM). The situation in PEM children is different from subclinical malnutrition. While increased proteinase inhibitors in wasting and stunting children somehow delay catabolism to maintain homeostasis, proteinase inhibitors increasing in seriously malnourished children, interrupt the mobilization of endogenous proteins which are needed to maintain homeostasis, for instance by providing essential amino acids. The increase of proteinase inhibitors aimed to counteract the proteases released in the course of infection, turn marasmus patients to develop clinical symptoms of kwashiorkor ^[@ref-30]^. Comparing marasmus with kwashiorkor, the latter is the more serious condition.
Different levels of homeostasis
===============================
'Homeostasis', as a metabolic condition depends on age, sex, genetic and environmental factors, and is maintained on different levels. This has been pointed out by Pontzer (2015) using the example of energy expenditure and energy balance ^[@ref-62],\ [@ref-63]^. This message from an evolutionary anthropologist is particularly important for public health in relation to physical activity. It is known that to lose weight by exercising has its limits. Increasing physical activity will lead to weight loss, however, excessive physical activity will not increase weight loss with increased energy expenditure; instead energy is acquired by using available resources used for basic functions of the organism under normal conditions. The consequence of this can be deadly in untrained individuals partaking in extreme physical activity e.g. a long run over 40 km. Winners of international marathon events are usually of African descent, whose 'homeostasis' allows them to cover the marathon distance in about two hours, while the rest of the field, using the rest of the day to finish the run. Homeostasis obviously can be achieved on different levels. The metabolism of the marathon winner allows them to economize energy expenditure much more efficiently as compared to other participants. In the regulation of the delicate balance of efficient energy expenditure, α2M might play a key role and might help to better understand the mechanisms regulating total energy expenditure. Recently a genetic hint towards regulating endurance and fatigue of muscles has been described in a rodent model ^[@ref-64]^. Research in this direction might be a further step to allow a better understanding of important metabolic pathways related to energy expenditure and endurance.
Other important issues for public health are also waiting for further exploration, such as how to distinguish biological- from chronological age ^[@ref-65],\ [@ref-66]^ and whether 'dietary restriction' is 'beneficial' for health and if so, are there drawbacks to be observed under certain circumstances and different ages. So for instance higher α2M levels in connection with some biological substances called 'metallothioneins' are beneficial in young adults but might have a harmful role in aging ^[@ref-32]^.
Conclusion
==========
In view of the challenge of non-communicable diseases, the aim of those caring for the health of the population should support the 'intrinsic motivation' of individuals to remain healthy. This might not only be achieved by encouraging physical activity, and restrain from smoking, alcohol and overeating. Motivation to change 'bad' behavior and maintaining 'good behavior', requires understanding what 'bad' behavior is meant to be, and what 'homeostasis' means and how to maintain it. The nature of NCDs are not yet well understood. This hampers the formulation of clear recommendations for 'healthy' behavior and limits the trust of the general public in health messages. Public health authorities at least should try to follow up developments and assess the significance of what is increasingly becoming known about our metabolism. Last but not least, public health should contribute to translate findings in life science, by conducting research applying molecular epidemiology, in such a way that these findings are relevant for human population groups, and may even validate α2M as a meaningful biomarker.
Data availability
=================
Underlying data
---------------
No data are associated with this article.
10.5256/f1000research.21700.r53054
Reviewer response for version 1
Chupeerach
Chaowanee
1
Referee
Institute of Nutrition, Mahidol University, Nakhon Pathom, Thailand
**Competing interests:**No competing interests were disclosed.
21
11
2019
Copyright: © 2019 Chupeerach C
2019
This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Version 1
The manuscript is a good explanation of the need of molecular epidemiology in the public health area. This article reported the role of alpha 2 macroglobulin in the difference of metabolic state and it might be associated to homeostasis and health status. Therefore it could be a public health concern in the future.
However, the abstract should re-arranged to be more representative of the manuscript summary.
For the review body, to interpret and follow the role of alpha 2 macroglobulin, the reader might need more information about this protein in terms of molecular detail.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
10.5256/f1000research.21700.r51079
Reviewer response for version 1
Schweigert
Florian J.
1
Referee
Department of Physiology and Pathophysiology of Nutrition, University of Potsdam, Potsdam, Germany
**Competing interests:**No competing interests were disclosed.
3
9
2019
Copyright: © 2019 Schweigert FJ
2019
This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Version 1
The submitted manuscript is a very well written review on the possible use of a2-macroglobulin as a marker of metabolism. The manuscript describes the general need of valid and meaningful biomarkers in the context of NCDs in public health research, especially with regard to overnutrition and obesity. The manuscript covers the implication of overnutrition and obesity for public health and the need of biomarkers and suggests a2-macroglobulin as such. The main part of the manuscript describes in very much detail the current knowledge of a2M in metabolic function as an omnipotent inhibitor of endopeptidases. With regard to the topic of the manuscript the authors concentrate on the knowledge regarding a2M in energy homeostasis especially with regard to energy-protein malnutrition.
Comments: In general the manuscript itself is well written and balanced. However, the abstract is by no means reflecting the manuscript. It lacks condensed information on a2M and only addresses the general aspect of public health and the need of a biomarker.It is challenging for the reader to cite a citation with refers to another citation. It cannot be retrieved without greater efforts. See second section, page 4 of Ohlsson (1972), as cited by Schelp *et al.*
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
[^1]: No competing interests were disclosed.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Human skin is permanently in contact with microorganisms. Among these microorganisms are many beneficial members of the microbiota. In addition, the skin has to deal with potential pathogenic microorganisms causing partially severe skin infections. There is increasing evidence that antimicrobial peptides and proteins (AMP) control the growth of microorganisms on all epithelial tissues including skin. AMP may play an important role to shape and control the preferred microbiota as well as to protect the skin from pathogenic microorganisms. Human skin is able to produce different types of AMP that exhibit individual activity profiles. One major AMP expressed by keratinocytes and released on the skin surface is the ribonuclease RNase 7. RNase 7 belongs to the human RNase A superfamily which consists of 13 genes located on chromosome 14 and contains eight different members with RNase activities (RNase 1--8) ([@B1]). The RNase A superfamily is named after its founding member, bovine pancreatic ribonuclease (RNase A). The members of this family contain an RNase catalytic domain that is composed of one lysine and two histidine residues. The protein structure is characterized by eight cysteine residues which are connected via four disulfide bridges ([@B1]). A huge repertoire of diverse physiological functions of these RNases has been reported during the last decades. Examples are angiogenic, neurotoxic, and diverse immunomodulatory activities ([@B2]). Several members of the human RNase A superfamily are also characterized by antimicrobial activities and there is increasing evidence that they participate in innate immunity. In particular, RNase 2 (also termed eosinophil-derived neurotoxin, EDN) and RNase 3 (also termed eosinophil cationic protein, ECP) exhibit diverse antiviral and antibacterial activities as well as immunomodulatory activities suggesting an important role in innate host defense ([@B3], [@B4]). Moreover, antimicrobial activities have also been reported for RNase 5 (also named angiogenin due to its angiogenic potency) ([@B4]--[@B6]), RNase 6 ([@B7]), RNase 7 ([@B8], [@B9]), and RNase 8 ([@B10]). RNase 5 and RNase 7 are both expressed by keratinocytes and thus may contribute to cutaneous innate defense ([@B6], [@B11]). Indeed, there is increasing evidence that RNase 7 may play an important role to protect skin from infection ([@B8]). This review aims to summarize the current knowledge about the physiological activities of RNase 7 and to highlight its role in cutaneous host defense.
Expression and Induction of RNase 7 in Skin {#s2}
===========================================
Site of RNase 7 Expression
--------------------------
The RNase 7 protein has been originally isolated from stratum corneum extracts during an attempt to identify and characterize the antimicrobial factors produced by human healthy skin ([@B12]). As the name implies, it is the seventh member of the human RNase A superfamily and exhibits ribonuclease activity. It has a molecular mass of 14.5 kDa and is highly cationic due to the presence of many arginine and lysine residues ([@B12]). Isolation of the protein from human skin and subsequent cloning of the corresponding cDNA enabled the identification of RNase 7 gene expression in human skin-derived keratinocytes ([@B12]). Of note, RNase 7 is not only expressed in human skin. Gene expression analyses revealed expression in various tissues such as heart, liver, kidney, pharynx, and tonsil ([@B12], [@B13]). In addition, the group of Spencer et al. identified a major role of RNase 7 in kidney and bladder host defense ([@B14], [@B15]).
Immunostaining of skin biopsies revealed RNase 7 expression in the keratinocytes throughout all epidermal layers with an increased staining in the more differentiated uppermost layers and an accumulation in the stratum corneum ([@B12], [@B16]). These data are in concordance with expression analyses of cultured keratinocytes demonstrating a higher RNase 7 gene expression in differentiated cultures as compared with proliferating keratinocytes ([@B6]). RNase 7 immunoreactivity was also present in sebaceous glands and hair follicles ([@B12], [@B17]). Analysis of skin rinsing fluids revealed the presence of RNase 7 throughout various body areas confirming the constitutive expression and release of RNase 7 by keratinocytes ([@B16], [@B18]). Interestingly, RNase 7 is also expressed in fetal skin suggesting that RNase 7 may already control cutaneous microbial growth of the fetus and newborn ([@B19]). In addition, RNase 7 may have other, as yet unknown functions in fetal skin such as a regulatory role in development. However, this hypothesis remains to be proven.
Induction of RNase 7 Expression
-------------------------------
In addition to the constitutive expression, RNase 7 expression in keratinocytes can be induced by various stimuli and different signal transduction pathways (summarized in [Figure 1](#F1){ref-type="fig"}). Proinflammatory cytokines such as IL-17A, interferon-gamma and IL-1 are able to induce the expression of RNase 7 in keratinocytes ([@B12], [@B17], [@B20], [@B21]). Especially the synergistic action of proinflammatory cytokines such as IL-17A and interferon-gamma markedly induced the expression of RNase 7 ([@B20]). The synergistic induction of RNase 7 by IL-17A/interferon-gamma was mediated by the signal transducer and activator of transcription 3 (STAT3) ([@B20]). Mohammed et al. reported that the induction of RNase 7 by IL-1beta in corneal epithelial cells was mediated by TGFbeta-activated kinase-1 (TAK-1). This in turn led to activation of the mitogen-activated protein kinase (MAPK) pathway resulting in activation of the transcription factors c-Jun and activating transcription factor 2 (ATF2). In contrast, the NF-kappaB pathway had no major influence on the IL-1beta-mediated induction of RNase 7 in corneal epithelial cells ([@B21]). Induction of RNase 7 by pro-inflammatory cytokines was also detected in the keratinocytes cell line HaCaT ([@B22]). Overall, the induction of RNase 7 by proinflammatory cytokines suggests that the local environment of inflammatory skin diseases like psoriasis trigger the increased expression of RNase 7 ([@B23], [@B24]).
{#F1}
Since RNase 7 exhibits potent antimicrobial activity (see below), it would be of biological significance that keratinocytes respond with an increased RNase 7 expression to the presence of microorganisms, in particular potential pathogens. Indeed, the expression of RNase 7 in keratinocytes can be induced by *Staphylococcus (S.) aureus* ([@B25]--[@B28]), *Pseudomonas (P.) aeruginosa* ([@B29]), *Enterococcus (E.) faecium* ([@B16]), and the dermatophyte *Trichophyton (T.) rubrum* ([@B30]). Of note, also skin commensals such as *Staphylococcus (S.) epidermidis* and *Corynebacterium (C.) amycolatum* induce RNase 7 expression in keratinocytes indicating that the presence of commensals leads to an increased RNase 7 production thereby strengthening cutaneous defense ([@B22], [@B27], [@B31]). *S. epidermidis* also enhanced the *S. aureus*-induced expression of RNase 7 in keratinocytes suggesting that the presence of commensals amplifies the defense response of human keratinocytes toward pathogens such as *S. aureus* ([@B27]). Interestingly, the induction of RNase 7 by the skin commensals *S. epidermidis* ([@B27]) and *C. amycolatum* ([@B31]), and by *P. aeruginosa* ([@B29]) and *T. rubrum* ([@B30]) depends on the involvement of the epidermal growth factor receptor (EGFR). This suggests that the EGFR plays an important role in cutaneous defense by its crucial role to mediate the expression of AMP such as RNase 7 and may offer an explanation for the increased susceptibility for skin infections of cancer patients receiving anti-EGFR therapy ([@B32]). Wanke et al. reported that---in addition to the EGFR---also Toll-like receptor-2 (TLR-2) and the transcription factor NF-kappaB are involved in the induction of RNase 7 in keratinocytes simulated with culture supernatants of *S. epidermidis*. This is in contrast to *S. aureus* which has been reported to activate the MAPK and phosphatidylinositol 3-kinase/AKT signaling pathways to induce RNase 7 expression ([@B27]). Thus, it seems that commensal and pathogenic bacteria activate different signal transduction pathways to induce RNase 7 expression in keratinocytes. RNase 7 expression in keratinocytes can also be induced by *Borrelia burgdorferi*, a bacterium that causes Lyme borreliosis and is transmitted by *Ixodes* ticks. Interestingly, tick saliva proteins have been shown to inhibit the *Borrelia burgdorferi-*mediated induction of RNase 7 and of other antimicrobial peptides and chemokines. This suggests that components of the tick saliva inhibit cutaneous innate defense reactions against *Borrelia burgdorferi* and thereby help the bacterium to evade local host defense and to disseminate into the body ([@B33]).
The role of cutaneous RNase 7 expression in the context of viral infection is less explored. It has been reported that keratinocytes infected with dengue virus showed an increased RNase 7 expression. If this is a direct induction or indirectly mediated by the release of inflammatory mediators is not clear. In addition, if RNase 7 influences the infectivity by dengue virus has not been reported ([@B34]). UV-B radiation also induces the expression of RNase 7 and other AMP in keratinocytes. This induction may contribute to the UV-mediated strengthening of the innate immune response ([@B35]). Niacinamide has also been identified as an inducer of the expression of RNase 7 and other antimicrobial peptides in human keratinocytes. Niacinamide is a well-known cosmetic ingredient and its antimicrobial peptides-inducing capacity may account for its observed beneficial effects on the skin barrier ([@B36]).
Antimicrobial and Ribonuclease Activity {#s3}
=======================================
Antimicrobial Spectrum of RNase 7
---------------------------------
A key feature of RNase 7 is its high antimicrobial activity against a wide spectrum of microorganisms. *In vitro* studies showed that RNase 7 is highly antimicrobial effective in low micromolar concentrations against Gram-positive and Gram-negative bacteria like *S. aureus, P. aeruginosa, C. amycolatum, E. faecium, Mycobacterium vaccae*, the yeast *Candida (C.) albicans* and *Pichia pastoris* and the dermatophyte *T. rubrum* ([@B12], [@B16], [@B30], [@B37]--[@B41]). The known microorganisms susceptible to RNase 7 are listed in [Table 1](#T1){ref-type="table"} ([@B6], [@B7], [@B12], [@B14]--[@B16], [@B25], [@B29]--[@B31], [@B37]--[@B50]). The functional relevance of the antimicrobial activity of RNase 7 and its contribution to the antimicrobial capacity of stratum corneum was demonstrated by the use of antibodies that neutralized the antimicrobial activity of RNase 7. Using such antibodies, we could show that inactivation of RNase 7 in human stratum corneum extracts led to an increased outgrowth of *P. aeruginosa, S. aureus, C. amycolatum*, and *E. faecium*. These results demonstrated that RNase 7 is part of the growth control mechanism in the stratum corneum ([@B16], [@B25], [@B29], [@B31]).
######
Overview of the microorganism susceptible to RNase 7.
**Microorganism** **References**
-------------------------------- -----------------------------------------------------------------
**GRAM-POSITIVE BACTERIA**
*Staphylococcus aureus* ([@B12], [@B25], [@B37], [@B38], [@B42]--[@B46])
*Staphylococcus saprophyticus* ([@B7], [@B47])
*Enterococcus faecium* ([@B6], [@B12], [@B16], [@B44])
*Enterococcus faecalis* ([@B6], [@B7], [@B48])
*Propionibacterium acnes* ([@B12])
*Micrococcus luteus* ([@B44])
*Corynebacterium amycolatum* ([@B31])
*Corynebacterium xerosis* ([@B31])
**GRAM-NEGATIVE BACTERIA**
*Pseudomonas aeruginosa* ([@B12], [@B14], [@B29], [@B37], [@B44], [@B48], [@B49])
*Escherichia coli* ([@B7], [@B12], [@B14], [@B15], [@B37], [@B44], [@B46]--[@B48])
*Proteus mirabilis* ([@B13], [@B22])
*Acinetobacter baumannii* ([@B44])
*Klebsiella pneumoniae* ([@B14])
**MYCOBACTERIA**
*Mycobacterium tuberculosis* ([@B50])
*Mycobacterium vaccae* ([@B40])
**YEAST**
*Candida albicans* ([@B12], [@B37], [@B41])
*Pichia pastoris* ([@B37])
**DERMATOPHYTES**
*Trichophyton rubrum* ([@B30], [@B39])
*Trichophyton mentagrophytes* ([@B39])
*Microsporum canis* ([@B39])
*Epidermophyton floccosum* ([@B39])
The Gram-positive bacterium *S. aureus* is a principal skin pathogen causing many infectious cutaneous diseases. Our work with *ex vivo* skin explants infected with *S. aureus* showed an increased release of RNase 7 expression. Inhibition of the antimicrobial activity of RNase 7 by specific antibodies resulted in considerable outgrowth of *S. aureus* on the skin surface. These results accentuate the functional relevance of RNase 7 in cutaneous defense against hazardous skin pathogens ([@B25]). This goes in line with a study from Zanger et al. who analyzed the RNase 7 expression levels of unaffected skin from healthy control persons and subjects with a *S. aureus* skin infection after a journey to a tropic or subtropic destination. They detected a 64% higher RNase 7 expression level in skin of the control group than in the unaffected skin of the infected persons. These data suggest that high RNase 7 baseline levels in healthy skin provide an increased protection against *S. aureus* infection ([@B51]).
Antimicrobial Mechanisms of RNase 7 and Influence of the Ribonuclease Inhibitor
-------------------------------------------------------------------------------
As described above, RNase 7 is a member of the RNase A superfamily and a potent ribonuclease that is able to degrade RNA. This raises the question of whether the ribonuclease activity is essential for the antimicrobial activity of RNase 7. Huang et al. designed RNase 7 mutants (H15A, K38A, and H123A) lacking ribonuclease activity and showed that these mutants are still able to kill *P. aeruginosa*. In addition, they identified three clusters of cationic residues on the surface of RNase 7. Substitution experiments of cationic residues identified the first N-terminal cluster with the lysine residues K1, K3, K111, and K112 as essential for antimicrobial activity. The other two clusters located on rigid secondary structures did not influence antimicrobial activity of RNase 7 ([@B37]). In general, the N-terminal cluster of RNases from the RNase A superfamily is the conserved region responsible for their antimicrobial activity ([@B44]). In line with the ribonuclease-independent killing activity against *P. aeruginosa*, we constructed a recombinant RNase 7 mutant without ribonuclease activity and showed that this mutant was still able to kill *E. faecium* ([@B16]). Ribonuclease-independent antibacterial activity has also been described for RNase 3/ECP ([@B52]).
The ribonuclease activity of RNases from the RNase A superfamily can be inhibited by binding to the 50 kDa, cytosolic, horseshoe-shaped Ribonuclease Inhibitor (RI) in a 1:1 molar ratio. The RI forms tight complexes with the RNases which is known as one of the tightest formation in biological systems ([@B53]). It is an obvious question if binding of RI influences the antimicrobial activity of RNase 7. Addition of RI to RNase 7 suspensions and subsequent testing of the killing capacity of RNase 7 toward *C. albicans* and *E. faecium* resulted in a reduced antimicrobial activity indicating that the RI blocks the antimicrobial activity of RNase 7 ([@B6]). This inhibition may be due to steric interaction and not due to specific inhibition of the ribonuclease activity because the ribonuclease activity is not essential for antibacterial killing as described above. In concordance with these results, Spencer et al. also reported that the RI bound to RNase 7 and inhibited its antimicrobial activity by blocking its ability to bind the cell wall of uropathogenic bacteria ([@B54]).
Immunohistochemical analysis revealed that the RI is located mainly in the suprabasal epidermal layers but is absent in the stratum corneum. In addition, incubation of the RI with stratum corneum extracts led to the degradation of the RI suggesting that proteolytic activity of the stratum corneum degrades the RI ([@B6]). In contrast to the RI, RNase 7 is abundant in the stratum corneum. This led to the hypothesis that in the epidermis RNase 7 is complexed with the RI and thereby inactivated. Degradation of the RI in the stratum corneum hinders RI-mediated inactivation of RNase 7 thereby liberating RNase 7 to function as antimicrobial factor ([@B6], [@B55]).
The exact mechanisms underlying the antimicrobial activity of RNase 7 are not fully understood. Huang and coworkers performed binding experiments with DNA-binding SYTOX^®^ Green dye and showed that RNase 7 is able to bind to the negatively charged membrane of *P. aeruginosa* and permeabilize the bacterial membrane ([@B37]). Torrent et al. demonstrated by means of liposome leakage assays that RNase 7 provokes membrane disruption by binding to bacterial membranes. Further experiments showed that RNase 7 did not interact with uncharged liposomes. Because of this fact the authors deduced that this membrane interaction is electrostatically driven ([@B56]). Further bacterial studies by this group showed a high leakage affinity of RNase 7 for *S. aureus* and *E. coli* membranes. RNase 7 depolarized the membrane by binding to lipopolysaccharide (LPS) and peptidoglycan, both major components of bacterial cell membranes from Gram-positive and Gram-negative bacteria, respectively ([@B46]). Another study from Lin et al. identified OprI (outer membrane protein I), an outer membrane lipoprotein from *P. aeruginosa*, as the initial binding site for RNase 7 instead of LPS. Addition of exogenous OprI or an anti-OprI antibody inhibited the antimicrobial activity of RNase 7 against *P. aeruginosa*. The authors conclude that upon RNase 7 binding, OprI internalizes along with RNase 7 into the cell leaving the cell membrane permeable to metabolites ([@B49]). These studies suggest that the interaction of RNase 7 with bacteria could be based on a target specific accumulation to different bacterial lipoproteins. This hypothesis needs to be proven in further studies with other bacteria.
In addition to its antibacterial activity, RNase 7 exhibits also activity against the yeast *C. albicans*. Salazar et al. described a dual mode of action for RNase 3 and RNase 7 as antifungal proteins against *C. albicans*. They designed mutants of both RNases by depleting the active catalytic site at His15 which is the complement to His12 and performed depolarization and permeabilization membrane assays to assess the antimicrobial action against *C. albicans*. They could show that both RNases act first as membrane lytic proteins followed by enzymatic cellular RNA degradation ([@B41]).
Immunomodulatory Activities {#s4}
===========================
Besides its antimicrobial activity, some studies provide evidence for an additional immunomodulatory function of RNase 7. Several members of the RNase A superfamily are well-known for their immunomodulatory activities ([@B2]). Although the role of RNase 7 as an immunomodulatory mediator is still emerging, there is first evidence of immunomodulatory functions associated with RNase 7. Resident and transiently migrating cells of the immune system are present in healthy human skin. With regard to antimicrobial defense, plasmacytoid dendritic cells (pDCs) are considered to play a key role within the types of the dermal immune system. pDCs usually circulate in the blood stream and are present in lymph nodes. Under inflammatory conditions they are able to infiltrate the skin ([@B57]). Due to their ability to secrete a several times stronger interferon-alpha (IFNα) response than any other cell types ([@B58], [@B59]), pDCs are crucial for cellular antimicrobial defense. A recent study from Kopfnagel et al. showed that RNase 7 in combination with human self-DNA activates a potent IFNα response in pDCs ([@B60]). This finding is in accordance with the known IFNα-inducing ability of LL-37 ([@B61]) and hBD-2/-3 ([@B62]) in complex with self-DNA. In addition, comparative experiments revealed a markedly stronger IFNα production in response to RNase 7 mixed with self-DNA as compared to stimulation with LL-37 or hBD-2 in mixture with self-DNA ([@B60]). Of note, the amount of the RNase 7-induced IFNα release was high enough to efficiently protect human keratinocytes from herpes simplex type I (*HSV-I*) infection. This was of special significance because RNase 7 itself exhibits no direct antiviral activity against *HSV-I*. In summary, this study revealed the ability of RNase 7 to convert usually not immunogenic self-DNA into a danger signal which enables a strong immunomodulatory response ([@B60]). It remains to be shown if these characteristics of RNase 7 may be linked with auto-inflammatory diseases.
T cells are present in human skin under healthy and inflammatory conditions. In healthy human skin they are of importance in wound healing, stimulate production of AMP by keratinocytes and influence keratinocyte development ([@B63]). Under inflammatory conditions T cells present in the blood can rapidly infiltrate the skin and produce pro-inflammatory cytokines, like IL-4 or interferon-gamma ([@B64]). A study of Kopfnagel et al. investigated the influence of RNase 7 on Th2 cytokine production by human CD4+T cells and Th2 cells. They reported that RNase 7 stimulation lead to a significantly reduced Th2 cytokine release of IL-4, IL-5, and IL-13. This downregulation of Th2 cytokines was found to be mediated by a reduced activation of the transcription factor GATA3. Interestingly, the ribonuclease activity of RNase 7 was dispensable for this effect. Due to this specific regulation the authors assumed a yet unknown receptor-mediated process which needs to be elucidated in further studies ([@B65]).
A novel study documents that RNase 7 mediates recognition of self-DNA by human keratinocytes. Activation of keratinocytes by a DNA/RNase 7 complex resulted in a markedly increased release of the chemokine IP-10 (CXCL10), a process that was mediated by type I interferons. In addition, the stimulation of keratinocytes with RNase 7 and DNA induced an interferon-beta (IFNß) dependent antiviral response which was sufficient to counteract an infection of the keratinocytes with herpes-simplex virus 1 (HSV-1). Taken together, there is increasing evidence that RNase 7 is able to bind self-DNA immediately after its release and induces a rapid DNA-mediated activation of keratinocytes ([@B66]) and pDCs ([@B60]). Thus, RNase 7 may serve as an alarmin which detects a disruption of the skin barrier by converting released self-DNA into a danger signal ([@B66]).
Various immunomodulatory activities of other RNases of the human RNase A superfamily have been described. Examples are processing and clearance of RNA, activation of immune cells, chemotactic activities, angiogenesis and neo-vascularization, tissue remodeling and repair, wound healing activity, and induction of apoptosis \[reviewed in ([@B67])\]. Thus, it is likely that RNase 7 exhibits additional, as yet unknown, immunomodulatory functions. The known immunomodulatory activities of RNase 7 are summarized in [Table 2](#T2){ref-type="table"} ([@B60], [@B65], [@B66]). Since the expression of RNase 7 is induced in inflammatory skin diseases such as psoriasis and atopic dermatitis (see below) it is likely that the immunomodulatory activities of RNase 7 may play a role in these skin diseases. Furthermore, it is also of interest to evaluate if the ribonuclease activity of RNase 7 may play a role in degrading host RNA released from damaged cells thereby controlling RNA-mediated inflammation during skin injury.
######
Overview of the main findings regarding immunoregulatory activity of RNase 7.
**Findings** **References**
--------------------------------------------------------------------------------------------------------------------------- ----------------
Up-regulation of IFNα by human pDCs stimulated with RNase 7 in mixture with self-DNA ([@B60])
Down-regulation of Th2 cytokines (IL-4, IL-13, and IL-5) by activated human RNase 7-stimulated Th2 cells and CD4+ T cells ([@B65])
RNase 7 stimulation of activated T cells results in a reduced activity of the transcription factor GATA3 ([@B65])
RNase 7 mediates sensing of self-DNA in human keratinocytes leading to an antiviral immune response ([@B66])
Role of RNase 7 in Skin Diseases {#s5}
================================
Atopic Dermatitis
-----------------
It has long been assumed that in the chronic inflammatory skin disease atopic dermatitis (AD) the expression of AMP including RNase 7 is impaired. This is based on the observation that a Th2 cytokine predominance has been shown to negatively regulate AMP expression ([@B68], [@B69]). In addition, comparative studies showed a reduced expression of AMP in AD skin as compared to psoriasis skin ([@B70]). However, especially for RNase 7, growing evidence reveals an increased expression in AD skin as compared to healthy skin. By investigating skin biopsies using quantitative real-time PCR, Gambichler et al. showed a significantly higher expression of RNase 7 mRNA as compared to expression in healthy skin ([@B71]). In addition, Harder et al. investigated the protein expression of RNase 7 in skin biopsies using immunostaining and ELISA of skin-derived washing fluids. Both methods revealed an enhanced RNase 7 expression and secretion in lesional skin of AD patients as compared to healthy controls ([@B23]). In another study, Clausen et al. investigated RNase 7 protein expression in the uppermost skin layers using tape stripping followed by ELISA measurements. In line with the above mentioned studies they also found a higher expression of RNase 7 in AD skin as compared to healthy skin ([@B72]). Taken together, there is consistent evidence that expression of RNase 7 is induced in the skin of atopic dermatitis patients. The consequence of this induction on diseases progression remains to be analyzed in future studies.
As mentioned above, RNase 7 is able to downregulate Th2 cytokine production in CD4+ T cells. This effect was reduced in CD4+ T cells derived from AD patients ([@B65]). These data suggest that RNase 7 plays an important role to negatively regulate the expression of Th2 cytokines, a function that may be disturbed in AD thereby promoting a Th2 cytokine environment.
Filaggrin is an essential component of the skin barrier. A deficiency of this protein can lead to an impaired skin barrier function and this is assumed to be a major predisposing factor in the development of AD ([@B73]). van Drongelen et al. used 3D epidermal models with filaggrin knockdown and infected these with methicillin-resistant *S. aureus* (MRSA) bacteria. They showed that IL-31 favored epidermal *S. aureus* colonization by preventing the *S. aureus*-induced expression of RNase 7 and other AMP ([@B26]). IL-31 is a pruritus-causing cytokine with increased presence in AD ([@B74]). Thus, enhanced levels of IL-31 in AD may locally inhibit an adequate induction of RNase 7 in *S. aureus*-infected areas.
Psoriasis
---------
A high-performance liquid chromatography (HPLC)-based analysis detected high amounts of RNase 7 and other AMP in psoriatic scale extracts ([@B24]). In line with these data, analyses of RNase 7 protein expression by immunohistochemistry and ELISA revealed increased expression of RNase 7 in the lesional skin of psoriasis patients ([@B23]). The increased amount of RNase 7 and other AMP in psoriasis may offer an explanation as to why psoriasis patients do not often suffer from skin infections despite the disturbed skin barrier ([@B75]).
One characteristic of psoriasis is the enrichment of pDCs in lesional skin and their key role in driving the development of the disease by the release of IFNα ([@B76]). As mentioned above, Kopfnagel et al. showed that RNase 7 in complex with self-DNA is a potent trigger of pDC-derived IFNα. Thus, an enhanced production of RNase 7 may trigger inflammation in psoriasis through increased production of IFNα ([@B60]). Similarly, the ability of RNase 7 to activate an inflammatory and antiviral response in keratinocytes in the presence of self-DNA may have implications in skin diseases such as psoriasis and atopic dermatitis where self-DNA derived from injured cells may be present in increased amounts ([@B66]).
Dermatomycoses
--------------
Major fungal cutaneous infections are superficial tinea and pityriasis versicolor which are caused by dermatophytes and *Malassezia* spp., respectively. An immunohistochemistry study from Brasch et al. investigated epidermal AMP expression in skin biopsies from infected and healthy persons. RNase 7 staining was significantly more often positive in the stratum granulosum of tinea than in the stratum granulosum of pityriasis versicolor and normal skin ([@B77]). In concordance with these findings, Firat et al. detected a high upregulation of RNase 7 expression in keratinocytes infected with *T. rubrum* ([@B30]). By blocking the EGFR, the *T. rubrum*-mediated RNase 7 induction in keratinocytes was significantly reduced. Interestingly, patients with anti-EGFR therapy have a high prevalence of cutaneous infections including those caused by dermatophytes such as *T. rubrum* ([@B78]). One may speculate that the anti-EGFR therapy may impair antifungal defense of the skin by inhibiting production of AMP such as RNase 7.
*S. aureus* Skin Infections
---------------------------
As already described above, RNase 7 exhibits potent *in vitro* anti-staphylococcal activity and *ex vivo* as well as *in vivo* studies highlight an important role of RNase 7 to control the growth of *S. aureus*. This gives rise to the speculation that RNase 7 may make up a fundamental part of innate cutaneous defense to fight off *S. aureus* infections and that a dysregulation of RNase 7 in general enhances the susceptibility toward *S. aureus* infections. However, this intriguing hypothesis has to be verified in further studies.
Chronic, recurrent, and persistent infections caused by *S. aureus* are often associated with the formation of the small-colony variant (SCV) phenotype. SCV have a longer generation time leading to small colonies on agar plates. It is believed that a switch from the wild-type phenotype to a SCV phenotype makes it possible for *S. aureus* to escape from host defense and to spread the infection ([@B79]). To proof the hypothesis that SCV are less susceptible to the cutaneous AMP-mediated killing activity, Gläser et al. used clinical *S. aureus* SCV strains and exposed them to either AMP directly or to stratum corneum extracts from healthy donors. In both setups the killing activity toward *S. aureus* isolates displaying the SCV phenotype was markedly reduced ([@B43]). These experiments demonstrated a less susceptibility of SCV to the antimicrobial activity of RNase 7 and other human skin-derived AMP. Thus, switching into the SCV phenotype may help *S. aureus* to subvert cutaneous innate defense through a higher resistance toward AMP such as RNase 7.
An increasing threat is the spread of MRSA. MRSA acquired resistance mechanisms against many antibiotics commonly used to treat *S. aureus* skin infections. In human skin equivalents infected with MRSA the expression of RNase 7 was induced, in particular in wounded skin ([@B28]). Of note, RNase 7 is able to dampen the growth of MRSA *in vitro* ([@B12]) and own unpublished results. These data indicate an important role of RNase 7 as cutaneous defense factor to control the growth of MRSA. It is an interesting hypothesis that patients with an impaired RNase 7-based cutaneous defense are more susceptible to the spread of virulent *S. aureus* including MRSA. Notably, the highly virulent strain USA300, a highly pathogenic MRSA causing severe skin and soft tissue infections, displays a comparatively low susceptibility toward the anti-staphylococcal activity of stratum corneum extracts ([@B42]). If a decreased activity of RNase 7 against USA300 contributes to this effect is speculative but not unlikely. This is supported by the fact that RNase 7 is a major component of the anti-staphylococcal activity of stratum corneum skin extracts ([@B25]).
As discussed above the presence of cationic amino acid residues endows RNase 7 with a positively charged N-terminal cluster that is required for antimicrobial activity. It is known that *S. aureus* has the capability to reduce its susceptibility toward the action of cationic AMP through the reduction of its negative surface charge. This can be achieved by the incorporation of D-alanine in its teichoic acids, a mechanism that lowered the susceptibility of *S. aureus* toward several skin-derived AMP including RNase 7 ([@B42]). It remains to be determined if the use of such strategies to lower the sensitivity toward the bactericidal activity of RNase 7 are associated with a higher pathogenic potential of distinct *S. aureus* strains.
Wounds
------
An intact skin barrier is essential for the protection against potential pathogenic microorganisms. In case of wounding, a rapid and potent defense is pivotal. In this regard, it has been shown that RNase 7 is rapidly released on the skin surface after experimental superficial barrier disruption ([@B23]). In contrast, Dressel et al. detected no increase of RNase 7 expression in the margins of chronic wounds, whereas the expression of the AMP hBD-2 and psoriasin was strongly induced. This led to the hypothesis that an insufficient expression of RNase 7 in chronic wounds may contribute to disturbed wound healing ([@B80]). Zanger et al. reported on a 50% lower RNase 7 gene expression in skin 3 days after sterile wounding ([@B81]). This in turn supports the idea that RNase 7 acts primarily as a component of the early and primary cutaneous defense response after injury ([@B81]). Clearly, more studies are needed to decipher the antimicrobial and immunomodulatory functions of RNase 7 in wounded skin.
Mycobacterial Infections
------------------------
It has been reported that RNase 7 exhibits antimicrobial activity against *Mycobacterium vaccae* at low micromolar concentrations ([@B40]). Furthermore, infection of airway epithelial cells with *Mycobacterium tuberculosis* led to induction of RNase 7 expression and an intracellular association of RNase 7 with *Mycobacterium tuberculosis* ([@B50]). This may suggest a direct antimicrobial effect of RNase 7 on *Mycobacterium tuberculosis*, but this has to be confirmed in further studies. Nevertheless, these initial studies give rise to the hypothesis that RNase 7 may play a role in infections caused by mycobacteria. Thus, it remains to be shown whether RNase 7 may also be involved in cutaneous mycobacterial defense ([@B82]).
Outlook {#s6}
=======
The role of RNase 7 in the skin is still emerging. However, as outlined in this review there is increasing evidence that RNase 7 plays an important role in innate cutaneous defense. This is mediated by the antimicrobial and immunomodulatory characteristics of RNase 7 ([Figure 2](#F2){ref-type="fig"} summarizes the role of RNase 7 in skin defense). It is likely that future studies will reveal novel immunomodulatory functions of RNase 7 and shed light on the as yet poorly understood importance of the enzymatic activity of RNase 7 in a physiological context. Moreover, it is of importance to further elucidate the potential link between specific infectious and inflammatory diseases and an impaired expression and/or function of RNase 7.
{#F2}
As mentioned in this review, functional studies as well as *in vivo* observations identified RNase 7 as a crucial factor to control the growth of the major cutaneous pathogen *S. aureus*. There is also evidence that other pathogens such as *P. aeruginosa* are affected by RNase 7. Thus, it is an interesting hypothesis that RNase 7 or optimized derivatives based on RNase 7 may have a high potential as anti-invectives. In this regard, the *in vitro* activity of RNase 7 against multi-resistant bacteria ([@B12]) may offer a promising alternative to fight off bacteria that are hard to kill by common antibiotics. Moreover, the targeted design of chimeric constructs combining selected parts of different molecules with high antimicrobial and ribonuclease activity, as recently reported for an RNase 3/1 hybrid construct ([@B83]), is a promising strategy. However, when considering RNase 7 for application as a therapeutic drug it is of great importance to evaluate the possibility that an excessive medical use of RNase 7 may induce the emergence of bacteria with an acquired resistance toward RNase 7. Such scenario could threaten our own innate host defense. In addition, the influence of immunomodulatory activities as well as the interaction of RNase 7 with the microbiota have to be considered when using RNase 7 in the treatment or prophylaxis of infections. In the case of atopic dermatitis, where *S. aureus* and elevated Th2 cytokines are major drivers of the disease, the capability of RNase 7 to kill *S. aureus* together with its influence on T cells to dampen secretion of Th2 cytokines may qualify RNase 7 as a beneficial drug to treat atopic dermatitis. Finally, the inducibility of RNase 7 gives rise to the speculation that a targeted induction of RNase 7 may offer a useful treatment or prophylactic option. Clearly, more studies are needed to unravel the physiological role of RNase 7 and to define its role in diseases and as a potential drug.
Author Contributions {#s7}
====================
FR, SD, and JH wrote the original draft and edited versions. VK, RG, and TW edited versions. FR and JH prepared the figures.
Conflict of Interest
--------------------
The 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.
**Funding.** Research work was supported by grants of the German research foundation (DFG) HA 3386/8-1 and WE1289/10-1. The work of TW has been funded by the German research foundation (DFG) under Germany\'s Excellence Strategy---EXC 2155 RESIST---Project ID 39087428.
[^1]: Edited by: Ester Boix, Autonomous University of Barcelona, Spain
[^2]: Reviewed by: John David Spencer, Nationwide Children\'s Hospital, United States; David Hains, Riley Hospital for Children, United States; Bruno R-s, Unidad de Investigación Biomédica de Zacatecas (IMSS), Mexico
[^3]: This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology
[^4]: †These authors share first authorship
| {
"pile_set_name": "PubMed Central"
} |
Background & Summary
====================
In retinal degenerative diseases, light-sensitive photoreceptors often die due to mutations in genes specifically expressed in these cells. This photoreceptor-specific defect will lead to a series of cellular changes, which would ultimately affect the well-being of the whole retina. For example, in a zebrafish mutant *pde6c*^*w59*^ (*pde6c*), an A\>G point mutation was identified in the *pde6c (phosphodiesterase 6C, cGMP-specific, cone, alpha prime)* gene^[@b1]^. This mutation was predicted to cause a frameshift in the coding sequence and result either in a truncated PDE6C or degradation of *pde6c* mRNA through nonsense-mediated decay. This mutation ultimately affects both cone and rod photoreceptors.
In wild-type zebrafish retina, photoreceptor progenitors withdraw from the cell cycle at 48 h post-fertilization (hpf) to form the photoreceptor precursors in the outer nuclear layer (ONL)^[@b2]^. These precursors will begin to differentiate into rods and cones at 50 hpf in a ventral part of the retina^[@b3]^. Then, cones gradually distribute evenly throughout retina; whereas rods distribute sporadically in the retina, with higher density in the ventral region. By 3 dpf, these photoreceptors are mature enough that the fish larvae will display the first visually-evoked startle response^[@b4]^. In *pde6c*, cones seem to develop following the normal course. They form outer segment by 3 dpf, but they degenerate starting at 4 dpf. *Pde6c* rods die subsequently as bystanders^[@b1]^, even though they do not express the *pde6c* gene. They also develop first in the retina, but they appear abnormal with more pronounced outer segment at 4 dpf^[@b1]^, when cones first degenerate. Nonetheless, rod number is comparable to WT at least up to 6 dpf^[@b5]^, and only begins to reduce in the central retina by 8 dpf^[@b1]^. The degeneration of these photoreceptors will also cause reactive gliosis as early as 7 dpf^[@b6]^, and abnormal morphology in bipolar cells with displaced nuclei and axonal processes at 8 dpf^[@b1]^. These tissue-level consequences are all initiated by a cone-specific *pde6c* mutation. The molecular basis of these retinal defects can often be analysed by profiling gene expression of the whole retina^[@b7]^, which can detect expression change even in a few cells^[@b11]^. Such expression profiles will facilitate the study of gene-regulatory mechanisms for these pathological events.
In this study, we measured the gene expression in the *pde6c* mutant retina and WT retina by RNA sequencing (RNA-Seq). We conducted RNA-Seq at 5 dpf/ 120 hpf. At this stage, cones have substantially degenerated, whereas the bystander rods display abnormal morphology^[@b1],[@b5]^. The Müller cells also display phenotype that is consistent with reactive gliosis (unpublished observation). Therefore, measuring retinal gene expression at 5 dpf/ 120 hpf would give us a glimpse of the global retinal defects at the initial stage of retinal degeneration. We analysed the sequencing results by various exploratory analyses, which indicate our RNA-Seq dataset was of high quality. We also evaluated the quality of our RNA-Seq results by reverse-transcription quantitative polymerase chain reaction (RT-qPCR), using a few key phototransduction genes. We observed a high correlation of fold changes of these genes as measured by both techniques. This observation strongly implicates that our RNA-Seq dataset effectively measured the expression changes in the *pde6c* retina. This dataset will therefore facilitate downstream characterization of the molecular problems in the *pde6c* retina.
Methods
=======
Fish maintenance and embryo collection
--------------------------------------
The *pde6c*^*w59*^ (*pde6c*) mutant line^[@b1]^ was purchased from the Zebrafish International Resource Center (ZIRC; <http://zebrafish.org/>) and maintained according to standard procedures^[@b12]^. The embryos were collected at 15-min intervals and raised at 28 °C in E3 medium^[@b13]^. The breeding of parents was also staggered so that the embryonic retinas could all be dissected and collected at 120 hpf. The homozygous mutant embryos (i.e., *pde6c*^*w59/w59*^) were obtained from the cross of the *pde6c* heterozygous parents, whereas the WT-control embryos were collected from the cross of genotyped WT siblings of the *pde6c* heterozygous parents. All protocols were approved by the Purdue Animal Care and Use Committee.
Optokinetic response (OKR)
--------------------------
An OKR apparatus was constructed based on the specifications as described previously^[@b14]^. This machine was used to measure the OKR of the zebrafish larvae. During the assay, the larvae were partially immobilized in 3% methylcellulose in a 35-mm Petri dish. The dish was placed in the center of a circular drum with 20° black and white vertical stripes attached on the inner surface. These stripes were illuminated by a Fiber Lite M1--150 illuminator (Dolan-Jenner Industries, Boxborough, MA). The illuminance was approximately 20,000 Lux at the level of the Petri dish, as measured by a LX1010B light meter (Mastech, Taipei, Taiwan). During the OKR measurement, the rotation speed of the drum was set at 8 revolutions per minute. In response to stripe rotation, normal larval eyes would display a characteristic smooth pursuit of rotation, and a rapid, corrective movement called saccade.
Genotyping
----------
The *pde6c*^*w59*^ mutation was genotyped as previously described^[@b5]^. In short, DNA was extracted from adult tail fin or larval body after retinal dissection. Then, the extracted DNA was used to amplify a 157-bps fragment from the *pde6c* gene with a specific pair of primers ([Table 1](#t1){ref-type="table"}). During amplification, a restriction site BsaXI was created only in the mutated *pde6c*^*w59*^ allele but not in the WT allele. This difference was discriminated by restriction analysis of these DNA fragments by BsaXI. On the resulting agarose gel, a WT allele would give a 157-bps band, whereas the *pde6c*^*w59*^ allele would give a 122-bps and a 35-bps band.
Micro-dissection of zebrafish retina
------------------------------------
WT retinas and *pde6c*^*w59/w59*^ retinas were microdissected from 5-dpf larvae using a previously published protocol^[@b7],[@b15]^. These larvae were first screened by the OKR assay. This non-invasive screening streamlined the identification of *pde6c*^*w59/w59*^ larvae from the cross of heterozygous parents, as only ¼ of the clutch would be homozygotes and they looked phenotypically normal. All larvae were dark-adapted for at least 2 h before dissection. Then, three retinas were dissected from three independent larvae and combined as one biological replicate. The remaining larval trunk was further genotyped to confirm its genetic identity. Three biological replicates were finally collected for each genotype ([Table 2](#t2){ref-type="table"}).
Total RNA extraction, characterization and amplification
--------------------------------------------------------
Total RNAs was extracted from the biological replicates by an optimized procedure^[@b7],[@b8]^ that combined TRIzol (Life Technologies, Grand Island, NY) and RNeasy Micro kit (Qiagen, Valencia, CA). The quality of the extracted total RNAs were evaluated by Bioanalyzer electrophoresis (Agilent Technologies, Santa Clara, CA). The RIN values were 9.1, 9.1 and 8.9 for the three replicates of WT retinas, and were 9.5, 9.1 and 9.3 for the three replicates of *pde6c* retinas ([Table 2](#t2){ref-type="table"}). Then, one nanogram of the total RNA from each group was amplified by Ovation RNA-Seq System V2 (NuGEN Technologies, San Carlos, CA). The yield of the amplified cDNA products was evaluated by Nanodrop spectrophotometry (Thermo Scientific).
RNA sequencing and analysis
---------------------------
For each replicate, one microgram of the amplified cDNA was paired-end sequenced by Illumina HiSeq 2,000 (Illumina, Inc., San Diego, CA). The six samples---three WT retinas (WR1, WR2, and WR3 in [Table 2](#t2){ref-type="table"}) and three *pde6c* retinas (PR1, PR2, and PR3 in [Table 2](#t2){ref-type="table"})---were combined and run on three different lanes. Running all samples in one lane eliminated lane-to-lane variation^[@b16]^, whereas running the samples on multiple lanes increased coverage. It should be noted that another six unrelated samples were included in these three lanes during RNA sequencing. This reduced the theoretical coverage of each sample by half.
Sequencing quality check and read alignment
-------------------------------------------
The raw fastq reads were trimmed for adapters and preprocessed to remove low-quality reads using Trimmomatic version 0.3^[@b17]^ with default parameter setting for Illumina paired-end reads. The reads were aligned to primers and/or adaptors with no more than two mismatches. Reads shorter than 36 bases were removed. After adapter removal, the quality of each paired-end sequence file was evaluated using FastQC analysis <http://www.bioinformatics.babraham.ac.uk/projects/fastqc/> (see Code Availability 1) based on the following parameters: 1) Distribution of quality score (Phred score) per base, 2) Distribution of quality scores of the raw sequences, 3) Distribution of duplicated reads, and 4) GC content (%) distribution of the raw sequences. The analyses indicate that our sequencing was of high quality, as illustrated by a typical sample in [Fig. 1](#f1){ref-type="fig"}. First, we evaluated quality of sequences by plotting the boxplots for Phred scores per base ([Fig. 1a](#f1){ref-type="fig"}) and distribution of average Phred scores of all sequences ([Fig. 1b](#f1){ref-type="fig"}). Both the Phred score per base and average Phred score of all sequences are generally above 28, the default threshold for a high-quality base call. Second, we confirmed the diversity of library sequences by plotting a duplication plot ([Fig. 1c](#f1){ref-type="fig"}), which shows the proportion of library sequences with different levels of duplication. The blue line shows the full sequence set and how its duplication levels are distributed. The red line shows the sequences after removal of duplicated reads, i.e., de-duplicated set. In both cases, most sequences fall into the far left of the plot, indicating that the library was diverse and most sequences occurred only a small number of times in the final set. The blue lines are relatively flat compared to red line, thus do not indicate significant contamination or any severe technical duplication. The duplicated reads constitute about 30% of all reads ([Fig. 1c](#f1){ref-type="fig"}). Third, we compared the distribution of sequence GC content (%) to the theoretical normal distribution ([Fig. 1d](#f1){ref-type="fig"}). The two curves substantially overlap with each other, indicating that the adaptors removal was successful and there is no obvious contamination.
These reads were then aligned to the zebrafish genome, release 89 \<ftp://ftp.ensembl.org/pub/release-89/\> using STAR v2.5.3a^[@b18]^ (see Code Availability 1). The aligned reads were then sorted and indexed using SAMTools v1.4.1^[@b19]^. The key outputs of alignment are shown in [Table 3](#t3){ref-type="table"}. These include the number of input reads, average input read length, uniquely mapped read, and mismatch rate per base. The uniquely mapped reads rate is around 70% and the mismatch rate per base is only about 1%. These observations indicate that the quality of all sequencing results was satisfactory.
Read counts, normalization, and differential gene expression
------------------------------------------------------------
To determine transcript abundance, we used the aligned reads to calculate the Fragments Per Kilobase of transcript per Million mapped reads (FPKM) by Cufflinks v2.2.1^[@b20]^(see Code Availability 1). The percentage of genes with nonzero FPKM in all samples are around 65% and shown in [Table 4](#t4){ref-type="table"}. We compared the FPKMs of the three biological replicates of the same genotype in pairwise scatterplots ([Fig. 2](#f2){ref-type="fig"}). Most values on the scatterplots fall along the y=x line, supporting the consistency among the biological replicates.
We also summarized read counts aligned to protein-coding genes in the gene transfer format (gtf) file from Ensemble Genome release 89 by featureCounts v1.5.2^[@b21]^ (see Code Availability 1). The output of featureCounts was fed to the package DESeq2 v1.16.1^[@b22]^ in R statistical environment (see Code Availability 1) to identify differentially expressed genes. This package normalized the count data against library size, log2-transformed the normalized data, calculated fold change between samples, inferred significance using a model with negative binomial distribution, and adjusted for multiple hypothesis testing. The resulting log2 fold changes of all samples were then plotted in an MA plot ([Fig. 3a](#f3){ref-type="fig"}), which shows the fold-change values of genes against the mean of normalized counts of all samples. These fold changes show a typical pattern for genome-wide experiments: the majority of the genes did not show a significant change in expression (log2 fold change about 0); whereas only a few genes showed a significant differential expression (Wald significance test^[@b22]^; adjusted *P*-value less than 0.1). These differentially-expressed genes are highlighted as red points in the plot. If the genes have a log2 fold change either greater than 2 or less than −2, they are plotted as open triangles.
Then, we further validate the reproducibility across the biological replicates by hierarchical clustering (HC) ([Fig. 3b](#f3){ref-type="fig"}) and principle component analysis (PCA) ([Fig. 3c](#f3){ref-type="fig"}), using the replicates' log2-transformed counts after library-size normalization and variance stabilization (see Code Availability 1). In the HC heatmap ([Fig. 3b](#f3){ref-type="fig"}), the 3 WT samples (WR1, WR2, WR3) are clustered together and the 3 *pde6c* samples (PR1, PR2, PR3) are clustered together. In the PCA plot ([Fig. 3c](#f3){ref-type="fig"}), these WT and *pde6c* samples are well separated from each other by their difference in the first PC, which explained 76% of the variance in the expression variables. Together, these sequencing analyses suggest that our RNA-seq had consistently measured the WR and PR samples, and effectively captured their biological difference.
Code availability
-----------------
1\. Code used for quality assessment and data analysis in this study is available at: <https://gist.github.com/coralzhang/fc4e51609ff316486c1682feed6404a9/471afb15b7f7b230a38e4eedaadcf5f679412a07>.
Data Records
============
Raw FASTQ files for the RNA-seq libraries were deposited to the NCBI *Gene Expression Omnibus (GEO)* with accession number GSE101544 and have been assigned BioProject accession PRJNA394760 (Data Citation 1) and SRA number SRP112616 (Data Citation 2). Key outputs of the analysis were deposited to the *GEO* with the same accession number (Data Citation 1) providing access to all relevant data files.
Technical Validation
====================
To validate the RNA-Seq results, we collected independent samples and analyzed the expression difference of selected genes by RT-qPCR. First, 30 eyes were microdissected from 15 larvae in both WT and *pde6c* groups. Then, total RNAs were extracted from these eye samples and reverse-transcribed into cDNAs as described^[@b7]^. These cDNAs were used for RT-qPCR reactions using SYBR Green PCR Master Mix (Life Technologies, Grand Island, NY). The RT-qPCR primers were designed with the RealTime PCR program (Integrated DNA Technologies, Coralville, IA) ([Table 1](#t1){ref-type="table"}). These RT-qPCR reactions were run on an Applied Biosystems 7,300 Real-Time PCR System (Life Technologies, Grand Island, NY) according to manufacturer's instructions.
In the RT-qPCR analysis, we measured the expression of seven phototransduction genes: *rhodopsin* (*rho*), four cone *opsins* (*red, green, blue,* and *uv)*, and two *transducins* (*gnat1* and *gnat2*). β-actin was used as the normalization control. We collected gene expressions in duplicates from independently-dissected eyes, and calculated a fold change between *pde6c* and WT samples for each duplicate using the ΔΔCt method^[@b23]^. Then, we averaged the fold changes of the duplicates. The resulting fold changes highly correlated to those fold changes obtained by RNA-Seq ([Fig. 4](#f4){ref-type="fig"}; r=0.91, *t*-test *P* value=0.0039). This high correlation strongly suggests that the RNA-Seq data effectively captured the gene-expression change of the *pde6c* retina. This RNA-Seq dataset will likely facilitate the characterization of the molecular defects in different cell types of the *pde6c* retina.
Additional information
======================
**How to cite this article:** Zhang, L. *et al.* Expression profiling of the retina of pde6c, a zebrafish model of retinal degeneration. *Sci. Data* 4:170182 doi: 10.1038/sdata.2017.182 (2017).
**Publisher's note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Material {#S1}
======================
We thank Purdue Genomics Core Facility for their technical assistance on RNA sequencing. Liyun Zhang was partially supported by a Charles Kelman M.D. Postdoctoral Fellowship from the International Retinal Research Foundation. Xinlian Zhang and Wenxuan Zhong was partially supported by National Science Foundation (Grant No. DMS-1440037 DMS-1438957) and NIH (R01GM113242, R01GM122080). Chi Pui Pang was partially supported by a Direct grant (Grant No. 2041771) from the Medical Panel, The Chinese University of Hong Kong, and a General Research Fund (Grant No. 2140880) from the Research Grants Council of Hong Kong. Mingzhi Zhang was partially supported by the National Natural Science Foundation of China (Grant No. 81470636).
The authors declare no competing financial interests.
{#f1}
{#f2}
{#f3}
{#f4}
###### Primers used for genotyping and validation of the RNA-Seq dataset.
**Gene** **Forward Primer** **Reverse Primer** **Remark**
------------------------- -------------------------- ------------------------ ----------------
*pde6c* TTGGCCTCTGGAATACTGGCTCTC GTTTGACCAGAACCCGGAAG For genotyping
*gnat1* CGTCAAGTTTGTGTTCGATGC GAGGAAACGAGCTACAAGGAG For RT-qPCR
*gnat2* CAAACCTGACTACCTTCCCAC TCTTCCTCTCGGACCTCTG For RT-qPCR
*opn1lw2 (red opsin)* CCAACAGCAATAACACAAGGG GCGACAACCACAAAGAACATC For RT-qPCR
*opn1mw1 (green opsin)* GGCTGTGTAATGGAGGGATTC ATGGTTTGCGGAGAATTTGAAG For RT-qPCR
*opn1sw2 (blue opsin)* GGTTCCTTTCAGCACCATTG AGAAGCCGAACACCATTACC For RT-qPCR
*opn1sw1 (uv opsin)* TCATTTTCTCCTACTCACAGCTC CACAAAAGAGCCAACCATCAC For RT-qPCR
*rhodopsin* AGTCCTGCCCAGACATCTAG GTACTGTGGGTATTCGTATGGG For RT-qPCR
*β-actin* TGCTGTTTTCCCCTCCATTG GTCCCATGCCAACCATCACT For RT-qPCR
###### Samples used in this study.
**Source** **Protocol 1** **Protocol 2** **Samples** **Protocol 3** **Protocol 4** **Data Citation**
------------------ -------------------- --------------------- ------------------------------ --------------------------- --------------------- ----------------------------
Wild-type retina Retinal dissection TRIzol+RNeasy Micro WR1 (Biological Replicate 1) Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705944
Wild-type retina Retinal dissection TRIzol+RNeasy Micro WR2 (Biological Replicate 2) Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705945
Wild-type retina Retinal dissection TRIzol+RNeasy Micro WR3 (Biological Replicate 3) Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705946
*pde6c* retina Retinal dissection TRIzol+RNeasy Micro PR1 (Biological Replicate 1} Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705947
*pde6c* retina Retinal dissection TRIzol+RNeasy Micro PR2 (Biological Replicate 2) Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705948
*pde6c* retina Retinal dissection TRIzol+RNeasy Micro PR3 (Biological Replicate 3) Ovation RNA-Seq System V2 Illumina HiSeq 2000 Data Citation 1 GSM2705949
###### Details of key QC metrics of RNA-seq library after alignment with STAR v2.5.3a.
**Sample** **WR1** **WR2** **WR3** **PR1** **PR2** **PR3**
------------------------ ---------- ---------- ---------- ---------- ---------- ----------
Number of input reads 48960995 46550508 47008978 51108456 42986393 49621925
Ave. input read length 197 197 197 197 197 196
Uniquely mapped reads 70.16% 71.35% 70.14% 69.34% 70.07% 71.04%
Mismatch rate per base 1.42% 1.45% 1.54% 1.48% 1.57% 1.52%
###### Percentage of genes with nonzero FPKM (output from Cufflinks v2.2.1).
**Sample** **WR1** **WR2** **WR3** **PR1** **PR2** **PR3**
------------ --------- --------- --------- --------- --------- ---------
Percentage 65.85% 61.21% 61.29% 65.25% 60.33% 63.03%
[^1]: Present address: The Wilmer Eye Institute, Johns Hopkins Hospital, 600 N. Wolfe St, Baltimore, MD 21287, USA.
[^2]: These authors contributed equally to this work.
[^3]: L.Z. and Y.F.L. designed the study; L.Z. conducted the experiments; L.Z., X.Z., G.Z., C.P.P., Y.F.L., M.Z., and W.Z. analysed the data; L.Z., X.Z., Y.F.L., and W.Z. wrote the paper.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Emotion regulation is defined as "processes by which individuals influence which emotions they have, when they have them, and how they experience and express these emotions" (p. 275).[@b1-prbm-10-277] Emotion regulation deficits have been demonstrated to be a significant pathomechanism in one of the most prevalent mental disorders -- major depressive disorder (MDD) -- which is, beside other symptoms, characterized by sustained negative affect (NA) and diminished positive affect (PA).[@b2-prbm-10-277],[@b3-prbm-10-277] Authors of modern cognitive theories assume that cognitive biases in attention, memory, or interpreting emotion eliciting, ambiguous events underlie these difficulties in emotion regulation in depressive people.[@b2-prbm-10-277],[@b4-prbm-10-277]--[@b6-prbm-10-277] Moreover, it is proposed that deficits in cognitive control and executive functioning impact negatively the ability to cope with the mentioned cognitive biases and result in a more frequent use of maladaptive emotion regulatory strategies.[@b2-prbm-10-277]
Results of a longitudinal study showed that applying adaptive emotion regulation strategies predicts lower depressive symptom severity over a 5-year follow-up in patients with MDD.[@b7-prbm-10-277] Physical exercise has been demonstrated to have positive acute and long-term emotion regulatory effects and could therefore be considered as a potential successful emotion regulation strategy in depressive individuals.[@b8-prbm-10-277]--[@b10-prbm-10-277] In particular, rock climbing is a type of sport that combines several features which can probably have an impact on cognitive and emotion regulation deficits in depressive patients. It requires high concentration and high levels of coordination, for example. Studies demonstrated that physical activity connected to high levels of coordination have a positive effect, especially on cognitive control.[@b11-prbm-10-277] Moreover, rock climbing can activate intense positive emotions.[@b12-prbm-10-277] Since the difficulty level can be flexibly varied according to individual fitness, climbing is particularly suitable to prompt a sense of goal achievement and self-efficacy.[@b12-prbm-10-277] Moreover, rock climbing also requires and trains the cooperation between individuals within small groups, as the climbing person always has to rely on safety measures provided by others with switching roles throughout the climbing session.
To date, acute changes in emotions associated with rock climbing in depressive patients have not been researched explicitly. Some case reports and observational studies and one controlled pilot trial demonstrated positive general therapeutic effects of rock climbing on depressive and anxiety symptoms and self-efficacy in depressive patients.[@b13-prbm-10-277]--[@b18-prbm-10-277] None of the mentioned studies, however, examined the immediate changes after a single rock climbing experience in different kinds of negative emotional states in depressive individuals. Accordingly, central aims of this nonrandomized, controlled trial were to examine if negative emotional states decrease and positive emotional states or coping emotions (CE; eg, pride, self-confidence) increase in persons with MDD during a 2.5-hour rock climbing session in comparison to an active control condition (relaxation training).
Patients and methods
====================
Participants and study procedure
--------------------------------
This pilot trial included a total of 40 adult inpatients receiving a psychologic and psychiatric inpatient treatment in the Department of Psychiatry and Psychotherapy of the University of Tuebingen (Germany). Eligible participants had to fulfil the criteria of an MDD or a bipolar disorder, with most recent episode of depression, according to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Patients suffering from a psychotic disorder were excluded. Presence of other comorbid mental disorders such as anxiety disorders, substance use disorders, or posttraumatic stress disorder was, however, not an exclusion criterion.
Eligible participants were not randomized, but had free choice of joining either the experimental intervention -- climbing therapy (climbing group \[CG\]; n=20) -- or the control condition (relaxation group \[RG\]; n=20) -- progressive muscle relaxation (PMR) by Jacobson.[@b17-prbm-10-277] The study procedure was the same for both interventions. Participants were asked to fill in the self-report questionnaire on depression and different emotional states, 30 minutes before the intervention started (T1) and immediately after the end of the intervention (T2). All subjects participated in the intervention for the first time during their hospitalization. A written informed consent was obtained from each participant at T1. The study was approved by the ethics commission of the Medical Faculty of the Eberhard Karls University and the University Hospital Tübingen. Data were collected between March 2013 and September 2013.
Materials
=========
In order to assess PA and NA, the Positive and Negative Affect Scale (PANAS) was administered.[@b19-prbm-10-277] The PANAS consists of a list of words describing different emotions and feelings (eg, upset). Participants were instructed to rate on a 5-point Likert scale if they had experienced these emotional states over the last week. The PANAS contains two subscales, PA and NA. The internal consistency of the PANAS subscales (PA: Cronbach's *α*=0.89; NA: Cronbach's *α*=0.85) has been proven and was sufficient in the current sample at both pre- and post-assessment (PA: Cronbach's *α*=0.90--0.95; NA: Cronbach's *α*=0.83--0.91).[@b19-prbm-10-277] In order to assess depressiveness and CE, the authors of this article created further items which were rated on the same answering scale (5-point Likert scale: 1=very slightly or not at all, 5=extremely) and had the same instruction ("Indicate the extent you have felt this way over the past week") as the PANAS items. Depressiveness was assessed with the following three emotion words: worthless, depressed, and hopeless. CE were measured with the following 11 items: confident, optimistic, brave, strong, determined, proud, interested, valuable, safe, secure, and grateful. For both scales -- depressiveness and CE -- the internal consistency was analyzed at the pre- and post-assessment in order to check if the items could be summarized in a scale mean score. Cronbach's *α* values appeared to be satisfactory (depressiveness: *α*=0.77--0.88; CE: *α*=0.90--0.95).
Interventions
-------------
The climbing session took place once a week in a private climbing hall and lasted 2.5 hours. It included 8--12 participants who were guided by at least two specialized nurses trained in climbing therapy. Every session followed the same agenda: After warming up exercises, the climbing equipment was presented. Games were played at the climbing wall in jump height to familiarize the patients with certain kinds of movement. The patients were taught the safety knots before they started climbing with a nurse as belay partner. At the end, the patients and the therapists shared their thoughts and feelings concerning the intervention. The control condition was based on PMR. The intervention took place in the hospital and was implemented by a psychologist. After explaining the principles of PMR to the participants, the relaxation exercise was conducted, lasting between 25 and 35 minutes. After the relaxation exercise, patients were invited to share their feelings about the relaxation exercise and their experiences.
Statistical analyses
--------------------
Data analyses were conducted using SPSS 21.0. Independent *t*-test and chi-square test were applied in order to check for between-group differences in demographic and outcome variables at baseline. Due to mainly significant moderate and large intercorrelations between the outcomes at pre-assessment (0.45≤*r*≤0.88) and at post-assessment (0.38≤*r*≤0.93), we decided for a mixed 2 (intervention: CG vs RG)×2 (time: pre- vs post-intervention) multivariate analysis of variance. In order to break up the interaction effects on depressiveness, CE, NA, and PA, Bonferroni-corrected post hoc univariate mixed 2×2 analyses of covariance (ANCOVAs; according to Bonferroni correction and number of analyses=4, α error level was 0.0125) and pairwise comparisons of cell means were conducted. Significant between-group differences in sociodemographic variables or outcomes were entered as covariates in the univariate ANCOVAs. Within-subject and between-group effect sizes were estimated by calculating Cohen's *d* and Hedges' *g*.
Results
=======
Participants' characteristics
-----------------------------
The sample of 40 patients was on average 40.56 years (SD=12.82) old and included 19 women (47.5%). In the total sample, 50.0% had a secondary school degree, 20.0% had a high school degree, 22.5% had a university degree, and 7.5% did not specify their educational level. The average body mass index of the total sample was 25.54 (SD=5.42). There were no differences between the CG and the RG regarding age, *t* (37)=−1.35, *p*=0.187; gender, *χ*^2^(1, N=40)=0.10, *p*\>0.999; educational level, *χ*^2^(5, n=37)=6.87, *p*=0.231; and body mass index, *t* (36)=0.17, *p*=0.865. In the CG, 85% (n=17) suffered from a major depression, whereas 15% (n=3) had a bipolar disorder. The distribution of the main diagnosis of an affective disorder was similar in the RG (major depression: n=18, 90%; bipolar disorder: n=2, 10%). The following comorbid disorders were found in the CG: substance use disorder (n=2, 10%), anxiety disorder (n=1, 5%), adjustment disorder (n=1, 5%), personality disorder (n=1, 5%), and posttraumatic disorder (n=1, 5%). In the RG, anxiety disorders were the only comorbid disorders (n=1, 5%). Regarding the Patient Health Questionnaire-9 as a measure of the severity of depressive symptoms, participants of the CG reached a sum score of 15.10 (SD=5.42) and participants of the RG reached a sum score of 13.53 (SD=5.68) on average.[@b20-prbm-10-277] Both groups did not differ in regard to the Patient Health Questionnaire-9 score, *t* (37)=0.89, *p*=0.382. There were also no significant differences between the CG (n=5, 25%) and the RG (n=9, 45%) regarding a comorbid medical condition, *χ*^2^ (*df*=1, N=40)=1.26, *p*=2.62. Furthermore, there were no significant baseline between-group differences regarding the scales' depressiveness, *t* (38)=1.15, *p*=0.256; CE, *t* (38)=1.48, *p*=0.149; and NA, *t* (32)=0.56, *p*=0.578. Regarding the scale PA, the two groups showed a significant difference, *t* (38)=2.31, *p*=0.026, at baseline, which indicated a higher PA in the CG.
Effects of climbing vs relaxation on affect and CE
--------------------------------------------------
The mixed 2×2 multivariate analysis of variance revealed a significant effect of time, *F*(4,35)=12.09, *p*\<001; intervention, *F*(4,35)=5.80, *p*=0.001; and a significant interaction time × intervention, *F*(4,35)=5.18, *p*=0.002. Post hoc univariate mixed 2×2 univariate analysis of variance (ANOVA) were calculated for CE, depressiveness, and NA. Regarding PA, a univariate mixed 2×2 analysis of covariance (ANCOVA) with PA values at baseline as covariate was conducted. The covariate reached significance, *F*(1,37)=219.44, *p*\<0.001. After Bonferroni correction, the post hoc mixed 2×2 ANOVAs/ANCOVA demonstrated significant effects of time and significant time × intervention interaction effects for all dependent variables. For PA and CE, but not for NA and depressiveness, even the effect of group reached significance in the 2×2 ANOVAs/ANCOVA. Bonferroni-corrected post hoc pairwise comparisons of mean cells were conducted in order to break up these mean and interaction effects. They revealed a significant decrease of NA (*p*\<0.001) and depressiveness (*p*\<0.001), and a significant increase of PA (*p*\<0.001) and CE (*p*\<0.001) in the CG, but not in the RG (0.093\<*p*\<0.574). In regard to all dependent variables, post hoc pairwise comparisons revealed no significant between-group differences before the intervention (0.149\<*p*\<0.578), but depressiveness (*p*\<0.010) was significantly lower and PA (*p*\<0.001) and CE (*p*\<0.001) were significantly higher in the CG in comparison to the RG. In regard to NA, the difference between CG and RG failed to reach significance (*p*=0.079).
Moderate to large between-group effect sizes in favor of the climbing intervention were found for each dependent variable ([Table 1](#t1-prbm-10-277){ref-type="table"}). In regard to the difference between pre- and post-assessment in the CG and RG, small to very small effect sizes were identified ([Table 2](#t2-prbm-10-277){ref-type="table"}).
Discussion
==========
The aim of this study was to examine if a single rock climbing session is associated with acute changes in PA and NA states in a group of inpatients with MDD. Results showed that rock climbing was significantly more related to a positive emotion regulatory effect regarding depressiveness, NA and PA, and CE, in contrast to a relaxation intervention. These findings are in accordance with general mood-improving effects of physical activity as well as the results of a pilot trial which explicitly examined the therapeutic effect of a rock climbing intervention on depressive symptomatology.[@b8-prbm-10-277]--[@b10-prbm-10-277],[@b18-prbm-10-277],[@b21-prbm-10-277],[@b22-prbm-10-277]
Different mechanisms through which rock climbing could probably act on emotional states in depressive patients are discussed subsequently. First, high levels of concentration and coordination which are required by this type of sport can counteract the cognitive deficits and biases in depressive patients, for example, by helping them to overcome difficulties in disengaging from negative stimuli.[@b5-prbm-10-277] Second, in this trial, the rock climbing intervention was supervised and took place in a group. Thus, it provided a setting of interpersonal experiences. According to a model of the role of social support in depression by Marroquin, interpersonal influences can function as emotion regulation strategy by effects on attentional deployment (eg, encouraging reorientation to positive features of a situation) and on cognitive changes (eg, supplying information which is inconsistent to cognitive schema).[@b23-prbm-10-277] Third, Beck defines in his cognitive theory of depression the cognitive triad comprising a negative view on the self, the world, and the future.[@b24-prbm-10-277] Since rock climbing can foster experiences of self-efficacy and goal achievement, it could act as emotion regulatory strategy by targeting negative cognitive schemas about the self.[@b12-prbm-10-277]
To our knowledge, this is the first controlled study that examined acute changes of emotions after rock climbing in patients with MDD. Results, however, have to be interpreted cautiously due to several limitations such as a nonrandom assignment of patients to study groups, significant between-group difference in PA at baseline, a small sample size, a missing no-intervention control group, lack of control for participants' previous climbing experiences, lack of an objective measure of physical fitness, missing follow-up assessments, the use of scales that were not sufficiently validated, and differences in the duration between the interventions.
However, in summary, our findings demonstrate that types of sport requiring high levels of concentration and coordination and being associated with an early experience of goal achievement are probably associated with short-term emotion regulatory effects. Our results should be replicated with a randomized study design which overcomes all limitations of the current pilot trial. In the context of a randomized allocation to the study conditions, it should be especially examined if the effects we found can be attributed to the climbing activity and how long the effects endure. Furthermore, more attention should be paid to potential mechanisms of action of rock climbing in regard to emotion regulation.
The authors thank the rock climbing team of the Department of General Psychiatry and Psychotherapy of the University Hospital of the Eberhard Karls University, Tübingen, Germany, in particular, Alfred Mollenhauer, for their enthusiastic work.
**Disclosure**
The authors report they did not receive any funding for this work and have no conflicts of interest in this work.
######
Test statistics of main and interaction effects of univariate mixed 2×2 analyses of variance/covariance for all dependent variables and effect sizes with 95% confidence intervals
Outcome Effect *F* *df*~num~/*df*~den~ *p*-value Hedges' g (95% CI)[a](#tfn2-prbm-10-277){ref-type="table-fn"}
------------------------------------------------------------ -------------- --------------------------------------------------- --------------------- ----------- ---------------------------------------------------------------
Depressiveness Group × time 8.99[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 0.005
Group 5.41 1/38 0.025 1.03 (0.62; 1.98)
Time 24.47[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 \<0.001
Positive affect[b](#tfn3-prbm-10-277){ref-type="table-fn"} Group × time 26.55[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/37 \<0.001
Group 26.55[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/37 \<0.001 0.74 (0.10; 1.38)
Time 25.33[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/37 \<0.001
Negative affect Group × time 6.90[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 0.012
Group 0.49 1/38 0.487 0.57 (−0.07; 1.20)
Time 25.66[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 \<0.001
Coping emotions Group × time 18.39[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 \<0.001
Group 8.63[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 0.006 1.30 (0.62; 1.90)
Time 36.80[\*](#tfn4-prbm-10-277){ref-type="table-fn"} 1/38 \<0.001
**Notes:** *df*~num~/*df*~den~= degrees of freedom of numerator/denominator.
Positive values indicate effects in favor for the climbing intervention.
For positive affect, a mixed 2×2 analysis of covariance with baseline values of positive affect as the covariate was conducted.
Statistical significance after Bonferroni correction (*p*=0.0125).
**Abbreviation:** CI, confidence interval.
######
Mean values, SDs, and Cohen's *d* separately for each dependent variable and study group
Climbing group, N=40 Relaxation group, N=40
----------------- ---------------------- ------------------------ ------------------- ------------- ------------- --------------------
Depressiveness 1.52 (0.98) 0.63 (0.72) 1.04 (0.34; 1.73) 1.90 (1.11) 1.68 (1.25) 0.19 (−0.44; 0.81)
Negative affect 1.40 (0.94) 0.53 (0.66) 1.07 (0.37; 1.77) 1.25 (0.61) 0.98 (0.91) 0.36 (−0.27; 0.99)
Positive affect 1.94 (0.75) 2.91 (0.74) 1.30 (0.56; 2.04) 1.36 (0.84) 1.52 (0.83) 0.19 (−0.43; 0.81)
Coping emotions 1.82 (0.72) 2.64 (0.69) 1.16 (0.45; 1.88) 1.45 (0.87) 1.59 (0.91) 0.16 (−0.46; 0.78)
**Abbreviations:** CI, confidence interval; SD, standard deviation.
| {
"pile_set_name": "PubMed Central"
} |
Optical microresonators[@b1] enable or enhance a diverse set of functions in integrated optical systems, spanning filtering and routing[@b2], switching and modulation[@b3], control of light emission[@b4], photodetection, and sensing. Low-volume optical modes are advantageous for a variety of reasons in all of these applications; examples include smaller footprint for dense integration, greater sensitivity to local index perturbations for sensors, greater energy efficiency in tuning for modulation, lower capacitance devices, stronger nonlinear interactions, and coupling to light emitters such as quantum dots and atoms. As a result, due to their ability to support very high-quality-factor (*Q*) resonant optical modes with mode volumes on the order of (*λ*/*n*)[@b3] or less, PhCs have attracted interest from a broad range of fields in recent years[@b5][@b6][@b7][@b8][@b9]. Much progress has been made in the optimization and fabrication of such structures, but due to the \~100-nm feature sizes required for structures with photonic bandgaps in the optical or near-infrared, fabrication typically relies on electron-beam lithography. This has limited the practical use of PhC structures, since it precludes practical integration with large-scale photolithographically defined systems, and also makes challenging the creation of complex structures requiring multiple masks[@b9].
Although the possibility for photolithographically defined quasi-2D PhCs has been explored in the past[@b10][@b11], high-*Q* resonators proved difficult to achieve. Passive optical cavities with measured quality factors (and extracted intrinsic *Q*s) of approximately 3,000 (10,000)[@b12] and 2,200 (on the order of 100,000)[@b13] have been designed and fabricated in SOI CMOS processes, but the potential for reliably creating active electro-optical devices in electrically-contacted PhC cavities capable of *Q*s exceeding 10^5^ in a scalable process has remained unexplored.
In this work, we demonstrate PhC microresonators with radiation-loss limited *Q*s well in excess of 10^5^ and measured material loss-limited *Q*s up to 58,000, fabricated photolithographically within a scaled CMOS process in the same patternable polycrystalline silicon (pSi) layer as used for transistor gates. A 1.2 *μ*m thick, patterned oxide beneath optical devices eliminates the need for undercuts or any post-fabrication processing. We introduce also a laterally electrically-contacted design for a quasi-1D resonator utilizing a patterned, partially etched silicon layer, and demonstrate electro-optic functionality in the form of defect-state photodetection in a resonant p-i-n structure which allows efficient photodetection at 1550 nm wavelengths in an all-silicon CMOS device. These results indicate the potential for high-*Q* PC resonators to achieve both passive and active optical functionalities in scaled photonic systems, in a way that can be straightforwardly integrated with sophisticated electronics[@b14]. The work indicates the practicality of silicon PhC microcavity devices for application in the near term for silicon photonic interconnects[@b15][@b16][@b17] and more speculatively in the long term for integrated quantum photonic systems[@b18].
Results
=======
A schematic chip cross section around the pSi layer in the process used is shown in [Fig. 1(a)](#f1){ref-type="fig"}, showing p- and n-mos transistors and doping profiles in the crystalline silicon substrate. Passive optical structures can be formed within the undoped pSi used as the transistor gates, and losses down to 6 dB/cm at 1550 have been previously reported in waveguides formed in this layer[@b19]. A partial etch step allows thin wings to sit adjacent to the waveguide cores, which can be doped and electrically contacted with metal vias which access three patternable copper wiring layers. A conformal silicon-nitride (*n* ≈ 2.0) layer encloses the pSi, and fills the holes in these designs which are otherwise SiO~2~-clad, resulting in a lower index contrast than the Si/air structures which have achieved the highest *Q*/*V* ratios.
A 193 nm photolithography process with 0.68 numerical aperture was used to define features in the pSi. 100-nm features transferred to a photoresist using such illumination can suffer significant optical proximity effects[@b10], and here an optical proximity correction (OPC) algorithm partially compensated for such effects in definition of the circular holes; [Fig. 1(d)](#f1){ref-type="fig"} shows the OPC-generated mask pattern for a triangular lattice of circular holes, and an example SEM image of a 2D lattice fabricated, with 375 nm lattice constant. From a set of SEM images on devices with a variety of lattice constants (*a*) and hole radii, an approximately linear relation between specified and achieved hole radii was found that was independent of lattice constant, and was used to compensate for uniform lithographic bias in future designs in the same process. Holes with radii down to about 55 nm formed reliably.
A variety of previously studied PC microcavity designs would be suitable for integration here. Minimization of radiative loss into the cladding motivates choice of cavities with smooth, ideally Gaussian envelopes[@b5][@b20][@b21], and quasi-1D structures, relying on photonic bandgap confinement in only one direction, are particularly robust in moderate index-contrast environments[@b22]. The bulk of our devices relied on such cavities.
Quasi-1D resonators for operation near 1550 nm using a linear array of holes in a 225 nm-height pSi waveguide of width ≈ 450 nm were designed with a lattice constant *a* near 330 nm, based on the hole tapering scheme developed in[@b23]. Around the cavity center to either side, hole radii decrease linearly from *r* = 0.33*a* to *r* = 0.28*a*; a set of "mirror" holes, all with *r* = 0.28*a* are added to either side, the number of which controls the strength of coupling to the feeding bus waveguides[@b24]. In the dielectric environment here, these resonators achieve intrinsic (that is, with no coupling to the feeding waveguides) radiative *Q*s of over 2 million, as calculated with finite-difference time-domain (FDTD) simulations[@b25] without material loss.
To allow efficient electrical contact to the resonator without introducing large optical loss, thin ≈ 50 nm-thick pSi "wings" were placed adjacent to the cavity. However, continuous slabs, in conjunction with the SiN liner layers, had effective indices large enough to couple to some Fourier components of the resonant mode, resulting in optical leakage, which could be reduced by patterning the wings with a 2D lattice of triangular low index holes; this had the effect also of reducing the resonant mode field\'s evanescent decay into the wings, allowing doped regions to be brought closer to the cavity (see [Supplementary Information](#s1){ref-type="supplementary-material"}). This patterning was done with a lattice of *r* = 0.3*a* holes here. An SEM image of a resulting structure along with the FDTD-calculated resonant electric field profile is shown in [Fig. 2(b)](#f2){ref-type="fig"}. The intrinsic radiative *Q* of the contacted structure was calculated to be ≈ 300,000, and the mode volume 0.8(*λ*/*n*)^3^.
Light is coupled to independent devices via grating couplers from fibers oriented nearly normal to the chip surface; adiabatic tapers couple to the ≈ 500 nm-wide waveguides, as shown in the optical micrograph of [Fig. 2(a)](#f2){ref-type="fig"}. Transmission spectra through cavities with resonant wavelengths *λ*~0~ = 1510 nm and *λ*~0~ = 1540 nm, with three different waveguide coupling strengths (8, 12, and 18 pairs of mirror holes) are shown in [Fig. 2(c) and (d)](#f2){ref-type="fig"}, along with theoretical fits to Fano profiles[@b26] which allowed determination of each cavity\'s loaded quality factor. The peak transmissions of each cavity, measured and normalized to the grating coupler transmission loss calibrated separately, closely follow the theoretical prediction of a standard coupled mode analysis *T*~pk~ = (*Q*~tot~/*Q*~WG~)^2^ [@b27]. Here the total measured quality factor is related to the waveguide coupling strength via the relation , where *Q*~0~ is the cavity\'s intrinsic loss rate (due to material and radiative loss) and *Q*~WG~ is the quality factor associated only with the cavity loss rate into the bus WG modes. As shown in [Fig. 2(e)](#f2){ref-type="fig"}, the peak transmissions and measured total quality factors are closely fit by the theoretical prediction for intrinsic quality factors *Q*~0~ = 58,000 and 51,000.
These loss rates are thus clearly limited not by the device design, but predominantly from material loss in the pSi and fabrication imperfections. Characterization of propagation loss in rectangular waveguides on the same wafer gives insight into the role of bulk material loss as well as sidewall roughness. The intensity loss coefficient of the bulk material in these samples is measured to be approximately *α* ≈ 4 dB/cm ≈ 0.92 cm^−1^, corresponding to a temporal photon loss rate of *γ* = *αc*/*n*~gr~ ≈ 8 GHz, where *n*~gr~ is the waveguide group index. This results in a material loss-associated *Q* of *Q*~mat~ = *ω*/*γ* ≈ 154,000. However, sidewall roughness makes a significant contribution to loss in these devices as well. Rectangular waveguides of 400 nm width, in which the mode profile has significant overlap with the sidewalls, exhibited propagation losses of about 8.5 dB/cm at the wavelengths of interest. Although the overlap with the sidewalls in the PhC cavities can be expected to be only roughly that of the fundamental mode in the 400 nm rectangular waveguide, this measured loss would correspond to a total *Q*~mat\ +\ roughness~ ≈ 77, 000.
Variance in fabricated hole sizes could play a role as well; scattering loss has been resulted to scale approximately as , where *σ~r~* represents the RMS radius deviation[@b28][@b29] and *A* represents a design and index-contrast-dependent parameter. Scanning electron microscope characterization of fabricated samples indicates *σ~r~* ≈ 2.0 nm, and FDTD simulations incorporating random disorder in the hole sizes with Gaussian distribution indicate *A* ≈ 0.1 (see [supplementary information](#s1){ref-type="supplementary-material"}), resulting in a *Q*~dis~ ≈ 280,000. Together, these loss rates are consistent with the observed intrinsic *Q*s. Use of higher NA photolithography is known to allow lower sidewall roughness, and will likely improve disorder in the PhC features as well, which presents a clear path to reducing the impact of fabrication imperfections on these loss rates.
A significant fraction of the material loss (the dominant loss mechanism in these devices as discussed above) results from absorption events involving pSi grain boundary defect electronic states at energies within the bandgap[@b19][@b30]. Since such absorption events generate free carriers, a significant photocurrent may be expected to flow in the contacted diode structures, which should be observable under reverse bias. Such photodiode behavior has indeed been previously observed in pSi ridge waveguide ring resonator structures[@b31]; in crystalline silicon, such detection has been observed in samples with defects introduced by ion implantation with high responsivities of 0.7 A/W[@b32] and even with avalanche gain allowing photodiode operation with 10 A/W operation and \>35 GHz bandwidths[@b33]. Photodetection has been reported also in defects inherent at the Si/SiO~2~ interfaces[@b34], and defect photogeneration mechanisms have been applied to resonant photodetectors as well[@b35][@b36].
A fabricated 1D PhC structure with lateral electrical contacts is shown in [Fig. 3(a)](#f3){ref-type="fig"}, with p+ and n+ doped regions shaded in red and blue. Variants were tested with doped region spacings of *w~i~* = 1.6 *μ*m (structure and results shown in [Fig. 3](#f3){ref-type="fig"}), as well as 1.0 and 2.2 *μ*m. The cavity is coupled to a feeding waveguide from only one side, and ideally the loaded *Q* including coupling into the bus waveguide mode would be half that of the intrinsic *Q*, to ensure that no light is reflected into the bus, i.e. that the cavity is critically coupled[@b27]. In these devices, material loss in the pSi was higher due to additional processing steps, and devices had measured intrinsic *Q*s of approximately 22,000 with no p+/n+ regions nearby; the loaded *Q* of the resonator devices with *w~i~* = 1.0, 1.6, and 2.2 *μ*m were approximately 4,300, 8,000, and 9,300, respectively, indicating closeness to critical coupling for the larger intrinsic region widths and the effect of free-carrier absorption for the device with lowest *w~i~*.
Dark current in the *w~i~* = 1.6 *μ*m structure is below 500 pA at reverse biases up to 30 V (as shown in [Fig. 3(b)](#f3){ref-type="fig"}), with large separation from the photocurrent when illuminated on resonance (near 1538 nm in this device) over a large voltage range. Photocurrent spectra obtained at a reverse bias of 22 V for input powers between 1 and 100 *μ*W are plotted in [Fig. 3(c)](#f3){ref-type="fig"}; the carrier photogeneration rate is proportional to stored cavity energy, and the spectra are thus Lorentzian at low input powers and exhibit bistability resulting from the thermo-optic effect[@b37] at higher powers. The peak current relates directly to the onresonance quantum efficiency (QE), plotted over a range of input powers and biases in [Fig. 3(d)](#f3){ref-type="fig"}; QE here is defined as number of charge carriers extracted per photon incident on the cavity, calculated from *I*~pk~*hν*/*P*~in~*q*, where *I*~pk~ is the peak resonant photocurrent and *P*~in~ is the optical power input after the grating coupler. A strong voltage dependence is observed, and large voltages of at least 15 V are required to achieve QEs above 10%, likely due to inefficient extraction under low fields due to recombination in the 1.6 *μ*m-wide intrinsic region. The power dependence of the peak QE appears similar to that previously reported in resonant pSi photodetectors[@b31], and could be due to density-dependent recombination rates; this effect must be stronger than any contribution from two-photon absorption at the higher powers in the range of powers studied here, since no increase in QE at high optical powers is observed. We applied a maximum of 30 V bias to the device, and observed a peak QE of 28% (0.35 A/W). At voltages in this range, material properties of the pSi appear to have been affected in a way which resulted in permanent (stable at least for many days) changes in the device response; the data presented here was verified to be stable, and we briefly discuss some of the changes observed in the [Supplementary Information](#s1){ref-type="supplementary-material"}.
Geometry also strongly influenced the modulation response of the photodetectors. Devices with narrower intrinsic regions had significantly faster 3-dB bandwidths; a modulation response curve with an empirical fit is shown in [Fig. 4(a)](#f4){ref-type="fig"} for a device with a 1 *μ*m-wide intrinsic region with -30 V applied (a DC QE of 25% and 30 nA dark current were measured in this device and bias), indicating a 2.9 GHz 3-dB roll-off. Oscillations around the fit are repeatable and due primarily to impedance mismatch between the device\'s large series resistance and the 50 Ω load. The measured bandwidth increases strongly with increasing reverse bias, and is lower in devices with larger intrinsic region widths (the three curves in [Fig. 4(b)](#f4){ref-type="fig"}). The device\'s *RC* limit is significantly higher than the measured roll-offs owing to its extremely small capacitance (), as is the *RC* limit owing to the load 50 Ω resistance and the contact pad capacitance; the measured frequency responses are, however, plausibly explained in terms of the limit imposed by the transit time through the intrinsic region, which increases with intrinsic region width and varies inversely with mobility. Both electron and hole mobilities in pSi thin films are known to decrease with film thickness[@b38], consistent with the strong dependence on the width of the 50-nm partial etched wings the carriers have to traverse in these devices.
Discussion
==========
Simple improvements to the electrical design of the photodetectors presented above should allow faster response, along with higher quantum efficiencies at lower applied biases. Incorporating mid-level doped regions in addition to the p+ and n+ used in these structures so as to form a p+/p/i/n/n+ structure with p and n regions extending closer to the device center would allow lower *w~i~*s without introducing excessive optical loss, and therefore higher transit time-limited bandwidths (inversely proportional to *w~i~*); thicker contact wings would also increase bandwidth by increasing mobility in the wings. Both of these modifications would also be expected to reduce the applied voltages required for carrier extraction by reducing recombination in the intrinsic region.
The photodetector results presented above constitute a first realization of electrically active PhC devices in a CMOS environment, but together with the possibility for increased bandwidth and reduced operating voltage also indicate promise generally for resonant subbandgap photodetection with sub-nA dark currents and low capacitance in this environment. Use of higher NA photolithography in more modern processes should also allow for lower roughness and disorder-induced loss, and higher intrinsic *Q*s in PhC devices generally.
Our work demonstrates that pSi PhC microresonators with quality factors of up to 60,000 in passive devices can be designed and fabricated within a scaled CMOS foundry in the same layer as is used for transistor gates, with potential for significant improvement with process and material quality. The contacted designs and the photodetector device presented here also indicate the potential for active wavelength-scale electro-optic devices. In the context of integrated Si photonic interconnects, we expect these results could impact design considerations primarily for low-capacitance resonant photodetectors (both linear and nonlinear) and high-efficiency modulators. Quantum photonic applications relying on such resonators may eventually benefit as well from the ability to produce them in scaled processes, integrated tightly with control electronics and tuning mechanisms.
Methods
=======
Cavity simulation and design
----------------------------
Design proceeded by first calculating band structures of 1D PhC waveguides[@b39] to approximately locate parameters for the desired resonant frequency, after which FDTD simulation in MEEP[@b25] of the full design allowed for quality factor calculation and optimization. Band structures of the 2D contact slab, as shown in the [supplementary information](#s1){ref-type="supplementary-material"}, were calculated with the same method as for the 1D cavity waveguides. Designed devices were laid out within Cadence Design Systems\' Virtuoso, a common VLSI layout tool[@b40].
Device characterization
-----------------------
An HP 8164A lightwave measurement system was used for spectral measurements; the laser linewidth (and hence spectral resolution) is \< 1 MHz. An Agilent 4156C semiconductor parameter analyzer was used for DC biasing and current measurement. For modulation response measurements, the output of a microwave synthesizer was modulated onto the input light with a lithium niobate modulator, and the RF power in the device photocurrent measured with a microwave spectrum analyzer. Consideration of the RF power measured in the PhC device relative to that under the same measurement performed on a commercial photodiode with a frequency response known to be flat (Discovery Semiconductor DSC20S) for frequencies up to 20 GHz allows for elimination of any frequency response from components in the system other than the PhC device.
Author Contributions
====================
K.K.M. designed, laid out mask designs for and performed experiments on devices; J.S.O. established CAD layout infrastructure and coordinated full chip tapeout; O.T.-Z., Z.S., R.B. and R.M. handled optical proximity correction, CMOS process control, and fabrication; K.K.M. and R.J.R. prepared the manuscript, and all authors reviewed it; and R.J.R. supervised the research.
Supplementary Material {#s1}
======================
###### Supplementary Information
High-Q CMOS-integrated photonic crystal microcavity devices: supplementary information
This work was carried out under the DARPA POEM program, managed by Dr. Jagdeep Shah. K. Mehta acknowledges support from a DOE Science Graduate Fellowship. The views expressed are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. Approved for Public Release, Distribution Unlimited.
{#f1}
{#f2}
{#f3}
{ref-type="fig"}) as a function of applied reverse voltage. Lines are guides to the eye.](srep04077-f4){#f4}
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Sequential human herpes virus (HHV) reactivation is well-known in drug reaction with eosinophilia and systemic symptoms (DRESS), but a comparison study to investigate the association of HHV reactivation with other types of cutaneous adverse drug reactions (cADRs) is still lacking. Besides, the pathomechanisms or mediators of viral reactivations are largely unknown. In this study, we aimed to investigate the cytokine/chemokine profiles before or concurrent with HHV reactivations.
Method
======
We conducted a prospective study to evaluate HHV-6, Epstein\--Barr virus (EBV) and cytomegalovirus (CMV) reactivation rates in various cADRs. Dynamic changes of cytokines and chemokines were also determined by sequential blood tests during acute stages. Interleukin-1£\] (IL-1£\]), IL-1 receptor antagonist (IL-1Ra), IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12, IL-13, IL-15, IL-17, basic fibroblast growth factors, eotaxin, granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating factor, interferon-£\^ (IFN-£\^), interferon £\^-induced protein-10 (IP-10), monocyte chemoattractant protein-1, macrophage inflammatory protein-1£\\ (MIP-1£\\), MIP-1£\], platelet-derived growth factor, chemokine (C-C motif) ligand 5 (CCL5), tumor necrosis factor-£\\ (TNF-£\\) and vascular endothelial growth factor were measured using the Bio-Plex Human Cytokine 27-Plex panel.
Results
=======
A total of 62 patients were enrolled in this prospective study including 23 DRESS, 17 SJS/TEN, 13 MPE, 5 GBFDE and 4 EMM. HHV-6 reactivation was observed in ten DRESS patients (43.5%) but in none of the other cADR patients. In contrast, EBV reactivation was detected in 35 patients distributed in every cADR groups. EBV reactivation rates were 73.9% for DRESS (17 in 23 patients), 29.4% for SJS/TEN (5 in 17 patients), 53.8% for MPE (7 in 13 patients), 80% for GBFDE (4 in 5 patients), and 50% for EMM (2 in 4 patients). CMV reactivation rates were 43.5% in DRESS (10 in 23 patients) and 11.8% in SJS/TEN (2 in 17 patients) respectively, but none of the other cADRs. HHV-6 reactivation in DRESS patients was associated with decreased proinflammatory cytokines and chemokines (IL-1£\], IL-1Ra, IL-2, TNF-£\\, IFN-£\^, and MIP-1£\\) before viral reactivations. Only one chemokine, IP-10, was expressed higher in the HHV-6 reactivation group (P = 0.018).
Conclusions
===========
HHV-6 reactivation in DRESS was associated with dysregulation of microenvironments mediated by several cytokines and chemokines.
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .125}
{#sp2 .126}
{#sp3 .127}
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Cortical autologous iliac crest bone grafts (ICBG) are widely used in orthopedic surgical procedures\[[@CR1]\], such as open wedge osteotomies\[[@CR2]--[@CR4]\], interposition arthrodeses\[[@CR5], [@CR6]\] and bone defects. Autologous bone grafts display excellent biological properties -- osteoconductivity\[[@CR7], [@CR8]\], osteoinductivity\[[@CR9]\], and potentially osteogenicity\[[@CR10]\],- plus favorable biomechanical characteristics\[[@CR11]\], and are considered the gold standard despite the abundance of allogenic and synthetic substitutes available\[[@CR12]\]. However, these favorable characteristics have to be juxtaposed with the risk of associated donor site morbidity, which may be clinically relevant in approximately 5% of patients, with values up to 30% in some reports\[[@CR13]--[@CR18]\].
Patients undergoing ICBG harvest often experience significant post-surgical bleeding\[[@CR14]\] and pain\[[@CR17]\] at the harvest site and may rarely sustain more severe complications including fractures\[[@CR16]\], arteriovenous fistula\[[@CR14]\], nerve injuries\[[@CR15]\] and abdominal wall hernias\[[@CR19]\]. The administration of hemostatics in combination with local anesthetics has been proposed to manage pain and bleeding. However, local anesthetics may clear rapidly from the harvest site, potentially causing systemic toxicity and reducing duration and effectiveness of postoperative analgesia\[[@CR20]\].
These considerations prompted the incorporation of lidocaine into a biodegradable hemostatic putty (Orthostat™ currently marketed as Hemasorb™) to locally provide anesthetic to the surgery site (Orthostat-L™). The Orthostat™ putty is composed of a polyvalent salt of a high molecular weight carboxylic acid, an alkyl benzopyranol ester and Vitamin E which enables a continuous release of lidocaine. Orthostat-L™ additionally comprises lidocaine at a concentration of 16% of its dry weight. In a previous in-vitro and animal study, lidocaine was released from the putty in a exponential fashion with a half-time of approximately 6--8 hours.\[[@CR21]\].
With the addition of Lidocain to Orthostat, pain control may be provided by two mechanisms. The hemostatic component may decrease the formation of irritating subperiostal hematomas, while the continuous release of lidocaine may decrease pain perception at the periosteal level.
In animal studies, the application of Orthostat-L™ resulted in a dose dependent, sustained analgesia in rats over a duration of two to three days with no evidence of systemic toxicity\[[@CR21]--[@CR23]\]. However, to our knowledge no studies have been performed to confirm this in humans.
It was the primary objective of this study to compare the analgesic effectiveness of a lidocaine loaded hemostatic putty (Orthostat-L™) with a lidocaine deficient putty (Orthostat™), defined as visual analog (VAS) and Wong Baker FACES scale outcomes, in a random sample of patients undergoing ICBG harvest for foot and ankle procedures. We hypothesized that the addition of lidocaine to the putty would provide significantly better pain control in the early postoperative period. For the purpose of this study, we defined "early postoperative" as an observation period of 72 hours during which we searched for differences in reported pain.
As a second objective, we aimed to test if, after the administration of the lidocaine loaded hemostatic putty, peripheral blood licocaine levels would stay below the threshold for liver toxicity of 6 mg/l.
Methods {#Sec2}
=======
Study design {#Sec3}
------------
This study was designed as a prospective, comparative, active-controlled, double-blind trial of the analgesic effectiveness of Orthostat-L™ (experimental group) with Orthostat™ (active control) in patients undergoing iliac crest bone graft (ICBG) harvest. The main outcome measurement was pain measured on a VAS and Wong Baker scale. We aimed at being able to detect an at least 10% ± 5% difference between two independent groups in the cumulative VAS scores within the early postoperative time period with a power of no less than 90% and a two-tailed alpha of 5% . This is consistent with an effect size of 2, thus requiring a total sample size of 14, or 7 per group, to reach a power of 90%. Such a sample size would still allow detecting a sample size of 1.5 without less than at least 80% power.
Study subjects {#Sec4}
--------------
From May to July 2008, we enrolled 14 patients undergoing foot and ankle surgical procedures with the use of a structural iliac crest bone graft. Patients who were between 18 and 71 years old and capable of completing a patient administered analgesia (PCA) device and who had given their written informed consent were included in the study. Patients with previous ICBG harvest, serious medical conditions including liver and heart failure, bleeding diathesis, hypersensitivity to lidocaine or to components of the hemostatic putty or any mental (dementia/psychiatric disorder) impeding their cooperation in the postoperative evaluations were excluded. Upon inclusion, patients were randomly allocated to the Orthostat™ or Orthostat-L™ group. There were seven patients in each group. Patients as well as all medical staff were blinded to which product the patient received.
As a result of the random allocation, there were seven patients in each group. Table [1](#Tab1){ref-type="table"} lists the demographic characteristics of patients allocated to the Orthostat™ and Orthostat™-L group.Table 1**Patient demographics and intraoperative data (Data presented as mean ± SD)**Orthostat-L™ (n = 7)Orthostat™ (n = 7)P-valueGender4 Females, 3 Males3 Females,4 MalesMean age (years)42.6 ± 10.551.0 ± 16.90.248Mean duration of Surgery (minutes)169.4 ± 53.3140.9 ± 38.5.0.273Graft size (cm^3^)4.0 ± 2.93.4 ± 1.90.681Putty amount (grams)4.3 ± 1.24.3 ± 0.80.999Mean time to hemostasis after putty administration (seconds)24 ± 1137 ± 370.392Intraoperative blood loss at the ICBG harvest site (ml)15 ± 815 ± 50.999
The study was conducted in accordance with the current version of the Declaration of Helsinki and under the laws and regulations enforced by the local ethics committees. This trial was registered in ClinicalTrials.gov (registry number: ClinicalTrials.gov [NCT01504035](http://www.clinicaltrials.gov/NCT01504035), registered January 2nd 2012) and approved by the local ethical committee (EKBB registry number 14/08).
Application of a continuous popliteal sciatic nerve block {#Sec5}
---------------------------------------------------------
Prior to induction of anesthesia, a board certified anaesthetist (RZ) administered a peripheral nerve block in the popliteal fossa proximal to the surgical site. Blocks were performed with 20--30 ml of mepivacaine 1.5% and bupivacaine 0.5%. After injection a 20G catheter was inserted for continuous postoperative sciatic nerve block with ropivacaine 0.2%, using an elastomer pump with a flow rate of 6-10 ml/h.
Iliac crest bone graft (ICBG) harvest and application of Orthostat™/Orthostat-L™ {#Sec6}
--------------------------------------------------------------------------------
Upon exposure and preparation of the primary surgical site at the foot and ankle, the same surgeon (VV) obtained an iliac crest bone graft in the following manner. A skin incision was made over the anterolateral aspect of the iliac crest. The aponeurosis of the external oblique was exposed and incised. The iliac crest was exposed subperiosteally, and a tricortical ICBG was obtained with an oscillating saw. The size of the graft was adapted to the structural defect at the foot and ankle to be bridged but was not larger than 3 cm × 1 cm × 2 cm (6 cm^3^). The graft dimensions were measured and the donor site was dried with sterile gauzes. Then either Orthostat™ or Orthostat-L™ was administered to the exposed bone surfaces under firm digital pressure until the level of material filling ICBG harvest site was flush with the bone surface. The amount of Orthostat™/Orthostat-L™ was determined by the number of 2 g portions applied to the harvest site and by weighing the leftover of the last portion used. The time from the end of putty application to complete hemostasis at the donor site was recorded and judged to be complete when there was no active bleeding noted by the surgeon. In addition the aspirated blood volume from the iliac crest was measured and recorded for blood loss. The wound was then closed in layers. No blood drain was inserted at the iliac crest harvest site.
Outcome assessment - donor site pain {#Sec7}
------------------------------------
Patients were closely followed every four hours within the first 72 hours of Orthostat™/Orthostat-L™ administration by blinded assessors. The pain level at the pelvic harvest site was rated according to the VAS score and Wong-Baker FACES Pain Rating Scale. VAS scores were measured on a continuous scale ranging from 0--100 mm. The Wong Baker scale ranged from 0 to 5 points, with the verbal anchors "no pain" to "worst pain". The VAS and Wong-Baker pain scores for each patient were plotted over time. The area under the curve (AUC) was calculated to represent the *cumulative* pain experienced within a specific time period. Since the surgical site pain from the lower extremity was addressed by a continuous sciatic nerve block, any pain experienced by the patient was related to iliac crest donor site pain. In case of any dysfunction of the sciatic block, the level of surgical site pain had to be recorded in the protocol.
Additional postoperative pain control was provided for all patients using patient controlled analgesia pumps (PCA) loaded with morphine. The cumulative dose of morphine used within the first 72 h after surgery was quantified.
Outcome assessment - serum Lidocaine levels and adverse events {#Sec8}
--------------------------------------------------------------
Serum lidocaine levels were measured after the administration of the peripheral nerve bloc (baseline) and at 2, 4, 8 12, 48, and 72 hours after the application of the putty using high performance liquid chromatography. Blood samples for serum lidocaine measurements were obtained from the basilic vein of the ipsi-contralateral arm and analysed immediately*.* The formation of a local hematoma at the iliac crest harvest site was assessed by ultrasound performed after 48--72 hours. After discharge from the hospital, patients were followed up at 14 and 30 days after surgery. During these follow-up visits, any adverse event such as prolonged pain, hematoma, herniation etc. at the pelvic harvest site was recorded.
Statistical analysis {#Sec9}
--------------------
Two-tailed t-tests for independent samples were used to compare cumulative PCA, bleeding time, total blood loss, and graft size across treatment groups. Continuous variables including VAS pain scores at different time points as well as the area under the curve for VAS and Wong Baker scores were tested using repeated measure ANOVA. For the 16 hours' time point, more than 20% of VAS and Wong Baker FACES measurements were missing and thus the time point had to be dropped. Discontinuous variables including Wong - Baker scores were analysed using Fisher's exact test with Bonferroni adjustments for multiple measurements. Results are given as mean ± SD. An alpha of 5% was considered significant. All analyses were done using intercooled STATA 10 (Stata Corp LP, College Station, TX).
Results {#Sec10}
=======
Intra-operative data {#Sec11}
--------------------
As shown in Table [1](#Tab1){ref-type="table"}, the duration of surgery, intraoperative blood loss and time to hemostasis after putty administration, as well as graft sizes and amount of putty used were not significantly different between Orthostat-L™ and Orthostat ™ (control) group.
Analgesic effectiveness of Orthostat-L™ - donor site pain {#Sec12}
---------------------------------------------------------
There were no cases of sciatic block dysfunction. All pain recorded was related to the iliac crest donor site. Subjects treated with Orthostat-L™ showed a significantly improved area under the curve (AUC) of the VAS score (AUC~vas~) as compared to Orthostat™ from 1 to 12 hours (mean AUC~vas~ 110 ± 115 vs. 291 ± 270; p = 0.036) (Figure [1](#Fig1){ref-type="fig"}). From 12 to 24 hours the control scores decreased towards baseline and differences between Orthostat-L™ and Orthostat™ were no longer statistically significant (AUC VAS 64 ± 62 vs 87 ± 79, p = 0.281). In the repeated measure ANOVA of all time points individually, there was no significant difference across groups VAS score (p = 0.3644) over the total follow-up.Figure 1**Postoperative VAS pain scores recorded at periodical time intervals after putty administratioon.**
For the Wong-Baker FACES Pain Rating Scale there was a significantly better outcome for Orthostat-L™ compared to Orthostat™ for the AUC for the first 12 hours (AUC ~Wong-Baker~ 6 ± 6 vs. 15 ± 15, p = 0.0024 (Figure [2](#Fig2){ref-type="fig"})) suggesting a beneficial early effect. Between 12 and 24 hours postoperatively as well as beyond 24 hours, as the control scores returned to baseline, the AUC obtained from Wong-Baker FACES Pain Rating Scale was not significantly different across groups (p = 0.999 and p = 0.1772, respectively), suggesting that the early effect equilibrates by the third postoperative day. There was no significant difference across groups regarding Wong-Baker FACES scores (p = 0.3544) at individual time points. There was no significant difference in cumulative PCA delivered morphine dose at 12 h (p = 0.267), 24 h (p = 0.717), or 72 hours postoperatively (p = 0.109) between the two groups.Figure 2**Postoperative Wong-Baker FACES pain scores.**
Safety of Orthostat-L™ - Lidocaine levels and adverse effects {#Sec13}
-------------------------------------------------------------
In the Orthostat-L™ group, the serum levels of lidocaine peaked at 8 hours remained elevated until 40--48 hours. At all time points, lidocaine levels stayed well below toxic levels of 6 mg/l. The mean lidocaine levels in the peripheral blood was significantly higher (p = 0.022) for the Orthostat-L™ group than the control group (Figure [3](#Fig3){ref-type="fig"}).Figure 3**Postoperative Serum lidocaine levels.**
There were no serious adverse effects experienced by subjects in the study. There was one case of seroma at the graft site in the intervention group that became evident prior to discharge on the fifth day after surgery. It required manual expression of the serous fluid to resolve. There were no complications in the control group (p = 0.2994).
Discussion {#Sec14}
==========
Iliac crest bone graft harvesting is associated with a donor site morbidity of up to 30%\[[@CR13], [@CR14], [@CR16]\]. Fortunately, major complications occur only rarely. However, patients frequently experience significant pain at the harvest site after ICBG harvest\[[@CR17]\]. Standard pain management includes use of opioid analgesics, which may cause respiratory depression\[[@CR24]\], nausea, vomiting or decreased gastrointestinal motility\[[@CR25]\], and thus may temporarily worsen the patient's condition and result in unfavorable outcomes and extended hospital stays with increased costs. Hence, reduction of postoperative pain at the harvest site will reduce or eliminate the comorbidities associated with use of systemically applied analgesics such as opioids. In this study we evaluated the effectiveness of a lidocaine loaded hemostatic putty (Orthostat-L™) as compared to a lidocaine deficient putty to lower iliac crest donor site pain.
Our study has potential limitations. Our sample was small, aiming for a between group difference of 10% ± 5%, and a larger study would be capable of detecting a smaller difference than that with statistical significance. However, the statistical significance of such a smaller difference must be put into relation with its clinical meaning, which might be questionable. Also, experiencing postoperative pain and response to analgesic medication has considerable inter-individual variability. Another potential shortcoming of a small sample size is a potentially higher risk of bias due to differential distribution. However, the enrolled patients were not significantly different in demographics and duration of procedure undergone, and systemic disease and lidocaine-related issues were exclusion criteria in our study. Hence it is not likely that our data are affected by substantial bias.
In our study, Orthostat-L™ significantly reduced cumulative VAS and Wong Baker FACES pain scores at the harvest site within the first 12 postoperative hours. Thereafter, pain scores in the control group began to return to baseline values. Nonetheless, pain scores in the Orthostat-L™-group continued to be lower compared to the Orthostat™-group, however, without statistical significance. The limited duration of significant pain relief provided by Orthostat-L™ may be explained by the fact that the control subjects returned to low pain levels just after 12 hours and by 20 hours after putty administration. Thus, there was no relevant pain for Orthostat-L to mitigate after this time period. In addition, the effects of lidocaine could have been also been altered by a local inflammatory response due to the surgical intervention and, additionally, to the putty. It has been postulated that inflammatory cell released peroxynitrite could decrease the effect of local anesthetic effects\[[@CR26], [@CR27]\]*.* Our data do not suggest that the duration of pain relief was restricted by limited lidocaine release from the putty. Lidocaine levels remained elevated in the Orthostat-L™ group even beyond 36 hours which was in fact in line with previous in vitro results\[[@CR21]\]. Thus, Orthostat-L™ showed a much more favorable drug elution profile as compared to single administrations of local anesthetics. After single infiltration of the iliac crest with local anesthetics, plasma levels peak within 20--100 minutes and rapidly decrease thereafter\[[@CR20]\].
Throughout the study, we could not detect a significant difference regarding cumulative PCA use between the two study groups. This result may be interpreted in two ways: First, our study was primarily powered to detect significant differences regarding cumulative VAS scores and *not* narcotic use. Second, Ip et al.\[[@CR28]\] showed in a systematic review that there are determinants of postoperative pain medication use which are independent from VAS rated pain. These parameters include anxiety and behavioral abnormalities which were not assessed in our study. We can therefore not exclude any confounding by these psychological factors.
Our data do not suggest that the duration of pain relief was restricted by limited lidocaine release from the putty. Lidocaine levels remained elevated in the Orthostat-L™ group even beyond 36 hours which was in fact in line with previous in vitro results\[[@CR21]\]. Thus, Orthostat-L™ showed a much more favorable drug elution profile as compared to single administrations of local anesthetics. After single infiltration of the iliac crest with local anesthetics, plasma levels peak within 20--100 minutes and rapidly decrease thereafter\[[@CR20]\].
In terms of safety, our study was able to demonstrate that serum lidocaine levels remained well below the level of toxicity (6 mg/l). The maximum serum lidocaine level noted in our study was 2.1 mg/l and was registered after 4 hours. We noted the occurrence of one seroma in the Orthostat-L™ group. This complication may be rather interpreted as a minor donor site complication and not as an adverse event directly related to the local drug application. Such minor complications may occur with a frequency of 10-15%\[[@CR14], [@CR16]\].
Previous studies have already suggested the installation of local anesthetics to reduce donor site pain at the iliac crest\[[@CR20], [@CR29]--[@CR35]\]. The proposed techniques vary from local infiltration at the time of surgery\[[@CR20], [@CR32], [@CR35]\] to intermittent\[[@CR30]\] or continuous\[[@CR31], [@CR33], [@CR34]\] administration using indwelling catheters. Similar to our study, single or intermittent administration of local anesthetics at the harvest site significantly reduced donor site pain 12--24 hours postoperatively\[[@CR20], [@CR30], [@CR35]\]. Serum concentrations also stayed below the toxic level\[[@CR20]\]. Continuous administration could increase the duration of pain relief to up to 48 hours in one study\[[@CR34]\] while another study could not demonstrate this effect\[[@CR31]\]. Study results were also very conflicting regarding the effect on local anesthetics on the intake of narcotics in the postoperative period. Some studies showed that the application of local anesthetics at the iliac crest could significantly decrease the postoperative opioid demand\[[@CR30], [@CR34]\], while others could not reproduce these results\[[@CR20], [@CR31]\]. However, the aforementioned studies contained heterogeneous sample groups and showed variable confounders or biases. Most importantly, these studies did not eliminate the surgical site pain. Pain perception at one side increases anxiety and thus alters the pain perception across other body sites\[[@CR36]\]. Accordingly, Morgan et al.\[[@CR31]\] could demonstrate in their study a significant correlation between recipient and donor site pain. However, only few studies incorporated the level of pain at the surgical site in the analysis and interpretation of their data\[[@CR20], [@CR31], [@CR34]\]. Our study could avoid these confounding effects since surgical site pain was continuously eliminated by the popliteal blocks. Thus, pain relieve at the donor site can more directly be attributed to the effectiveness of Orthostat-L™.
When comparing Orthostat-L™ with local anesthetics, it must be considered that that Orthostat-L™ is also a hemostatic agent. In contrast to local anesthetics, Orthostat-L™ may therefore additionally help to prevent irritating subperiosteal hematomas, which further contribute to donor site pain. Nevertheless, these speculations on the superior effectiveness of Orthostat-L™ to reduce donor site pain needs clarification in future randomized controlled trials.
Conclusions {#Sec15}
===========
A bioresorbable hemostatic putty with lidocaine, as Orthostat-L™ has significant effectiveness in controlling pain experienced within 12 hours after iliac crest bone graft harvesting compared to a putty without lidocaine, as Orthostat™. We found no evidence for systemic toxicity or an increase in risk of adverse effect using a lidocaine based putty.
Authors' information {#Sec16}
====================
Marc Andreas Müller and Arne Mehrkens first authorship.
Authors' original submitted files for images {#Sec17}
============================================
Below are the links to the authors' original submitted files for images.Authors' original file for figure 1Authors' original file for figure 2Authors' original file for figure 3
Marc Andreas Müller, Arne Mehrkens contributed equally to this work.
**Competing interests**
This study was financially supported by Orthocon (ORTHOCON, Inc. 1 Bridge Street Suite 121, Irvington, NY 10533, USA) which also provided the hemostatic putties either loaded or unloaded with lidocaine. However, the authors had full access to all the data at any time during the study and had final responsibility for the decision to submit for publication. The authors did also not receive any benefits for the preparation of this manuscript. None of the authors has received any salary or royalties from Orthocon. In addition, no author does have any financial relationship with this company by means of employment, consultancy, stock ownership or other significant equity. No additional funding other than from Orthocon was received for this study.
**Authors' contributions**
Designed the study: VV, MAM, RZ. Gathered the data: MAM, RZ, VV, MEA. Analyzed the data: PV, MAM, MEA, VV. Written the initial draft: MAM, MEA, PV, VV. Ensured the accuracy of the data and analysis: PV, MAM, VV, MEA. All authors read and approved the final manuscript.
We wish to thank the clinical trial unit of the University Hospital of Basel, Switzerland, for having supported the conceptualization, design and monitoring of this study. We also thank Orthocon Inc. for financially supporting the planning, design and monitoring of this study and for having provided the hemostatic putties either loaded or unloaded with lidocaine.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Advances in the treatment and diagnosis of cancer and cardiovascular diseases have resulted in a great improvement in patient survival, but the growing population of long-term survivors has been compromised by the occurrence of comorbidities and complications throughout their lifespan, which can substantially affect their quality of life \[[@B1]\]. These chronic disorders include secondary malignant neoplasms, endocrine disorders, cardiopulmonary dysfunction, cardiovascular complications, and cognitive and psychosocial problems \[[@B2], [@B3]\], with cardiac injury being one of the most difficult complications to manage in the clinic \[[@B4]\]. For instance, patients receiving radiation therapy are often at a great risk for developing cardiomyopathy, which increases the risk for myocardial infarction and heart failure \[[@B5]--[@B7]\]. Moreover, the majority of cardiac catheterizations for congenital heart disease and some regular follow-up examinations, including chest X-rays, angiography, scintigraphy, and computed tomography scans, are often performed in children and adolescents who may experience intensive radiation exposure and are at a high risk for long-term adverse effects due to radiation \[[@B8]--[@B10]\]. Meanwhile, the use of medical ionizing radiation has been increasing dramatically in recent years, and cardiologists are also exposed to radiation from various medical sources, such as percutaneous coronary intervention, cardiac radiofrequency ablation, multidetector coronary angiography, and myocardial perfusion imaging scintigraphy \[[@B11]\].
Medical radiation sources are known to be cardiotoxic, but the underlying mechanism by which they induce cardiac damage is not fully understood \[[@B12]\]. Studies have demonstrated that radiation may generate oxygen free radicals, resulting in the generation of excessive reactive oxygen species that induce oxidative stress in cells and tissues \[[@B13]\]. The sustained generation of reactive oxygen species is a primary inducer of apoptosis \[[@B14]\]. Studies in mice have demonstrated that radiation causes acute injury in cardiomyocytes, leading to apoptosis \[[@B15]\]. The loss of myocytes causes thinning of the myocardium with a resulting increase in wall stress, which may also be linked to defective regenerative and adaptive pathways in the remaining myocardial cells \[[@B16]\]. These pathological changes ultimately result in congestive heart failure \[[@B17]\]. Accordingly, reducing oxidative stress is of great importance in preventing or suppressing apoptosis in order to protect the heart against radiation-induced injury \[[@B18]\]. The current treatment for radiation-induced heart disease mainly utilizes symptomatic and supportive approaches or surgery (cardiac transplantation); however, all these methods are largely ineffective. Therefore, the development of a new therapeutic approach is an urgent unmet medical need.
There is an increasing interest in the discovery and development of novel natural product-based approaches for the prevention and treatment of radiation-induced heart diseases. For instance, recent studies \[[@B19], [@B20]\] have revealed that sodium tanshinone IIA sulfonate (STS), a derivative of tanshinone IIA, elicits protective effects on cardiomyocytes against oxidative stress-mediated apoptosis. This plant is a well-known traditional Chinese medicine used for the treatment of various cardiovascular diseases, whose effects on the cardiovascular system are mediated by improving microcirculatory, vasodilatory, anticoagulant, antithrombotic, anti-inflammatory, free radical scavenging, and mitochondria-protective effects \[[@B21]\].
Tanshinone IIA is extracted from the roots of*Salvia miltiorrhiza*, and, as an active diterpenoid component, it can prevent or slow the progression of cardiovascular diseases \[[@B22]\]. STS also has been used clinically as an approved drug by the State Food and Drug Administration of China \[[@B23], [@B24]\]. Its cardioprotective effects might be mediated by free radical scavenging properties and antioxidant activities \[[@B25]\]. There is increasing evidence supporting that STS relies on such an action to attenuate radiation-induced pulmonary and renal diseases \[[@B26]\]. Therefore, we hypothesized that STS could have preventive effects on radiation-induced heart disease. To this end, in the present study, we characterized the effects of STS against radiation-induced oxidative stress and the possible underlying mechanisms using H9c2 cardiomyocytes as an in vitro cell-based model. It was hoped that the results from the present study would provide a basis for the future development of STS as an anti-radiation-induced heart disease reagent in preclinical and clinical settings.
2. Materials and Methods {#sec2}
========================
2.1. Chemicals {#sec2.1}
--------------
STS was obtained from Shanghai First Chemical Company (Shanghai, China). Dulbecco\'s modified Eagle medium (DMEM)/F12 medium was purchased from Gibco (Grand Island, NY, USA). Fetal bovine serum was from Invitrogen (Carlsbad, CA, USA). Hoechst 33258 was obtained from Invitrogen (Eugene, OR, USA). 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl tetrazolium bromide (MTT), dimethyl sulfoxide (DMSO), and Giemsa were from Sigma (Sigma-Aldrich, St. Louis, MO, USA). The lactate dehydrogenase (LDH), superoxide dismutase (SOD), and malondialdehyde (MDA) assay kits were obtained from Nanjing Jiancheng Bioengineering Institute (Nanjing, China). Rabbit polyclonal antibodies for p38, phosphorylated p38, ERK, and phosphorylated ERK were purchased from Cell Signaling Technology (Beverly, MA, USA).
2.2. Cell Culture {#sec2.2}
-----------------
H9c2 cardiomyocytes, a subclone of the original cell line derived from embryonic BD1X rat heart tissue, were obtained from the Chinese Academy of Sciences Cell Bank (Shanghai, China) and cultured in DMEM/F12 medium containing 10% (v/v) fetal bovine serum, penicillin (100 U/mL), and streptomycin (100 *μ*g/mL). The cells were maintained at 37°C in a humidified atmosphere with 5% CO~2~. The medium was replaced every 2-3 days, and the cells were passaged as they grew to 80% confluence; they were subcultured or subjected to subsequent experiments. The purity of STS used in the present study was \>95%, and STS was dissolved in DMEM/F12 medium. The cells were divided into the following groups: 0 Gy (control) group; 2-Gy radiation (2 Gy) group; 4-Gy radiation (4 Gy) group; 6-Gy radiation (6 Gy) group; 8-Gy radiation (8 Gy) group; 10 *μ*g/mL STS (STS) group; 2 Gy + 10 *μ*g/mL STS (2STS) group; 4 Gy + 10 *μ*g/mL STS (4STS) group; 6 Gy + 10 *μ*g/mL STS (6STS) group; and 8 Gy + 10 *μ*g/mL STS (8STS) group.
2.3. Irradiation Procedure {#sec2.3}
--------------------------
The cells were seeded with fresh culture medium and cultured for 24 h prior to the radiation treatment. A single dose of 2, 4, 6, or 8 Gy of X-rays was administered to the cells using a Siemens PRIMUS high energy linear accelerator (Siemens AG, Erlangen, Germany), operating at a dose-pulse rate of 6 MV/min with a source-to-H9c2 cell distance of 100 cm. After radiation, the cells were maintained at 37°C and 5% CO~2~, and the culture medium was replaced every 24 h.
2.4. 3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl Tetrazolium Bromide (MTT) Assay {#sec2.4}
------------------------------------------------------------------------------
To determine the cytotoxicity of STS, H9c2 cells were seeded at 10^5^ cells/mL in 96-well plates. The experiments were carried out after cell exposure to serum-free medium for at least 16 h, and the cells reached more than 80% confluence on the plate bottom. After discarding the old medium, the cells were pretreated with varying concentrations of STS for 24 h, 48 h, and 72 h, respectively. MTT was dissolved in phosphate-buffered saline (PBS) and added to each well at a final concentration of 5 mg/mL. The cells were then incubated at 37°C for 4 h. The medium was discarded, and 150 *μ*L of dimethyl sulfoxide was added to each well. The absorbance of each well at 570 nm was quantified using a spectrophotometer (Bio-Rad, Philadelphia, PA, USA).
Cell viability was determined by the MTT assay. In brief, the cells were plated onto 96-well tissue culture plates (cell density, 4 × 10^3^ cells/well). STS was added to the medium at the doses indicated for 24 h, followed by exposure to X-ray radiation, and then cultured for another 24 h. Cell viability was then monitored. Each treatment and control was assayed in at least six wells, and the cell viability was expressed as a percentage of that of the control cells.
2.5. LDH Release Assay {#sec2.5}
----------------------
Following pretreatment with STS for 24 h, the cardiomyocytes were irradiated with 2, 4, 6, or 8 Gy of X-rays. The culture medium was collected at 24 h after X-ray radiation, and the LDH activity was measured by means of a colorimetric assay using a commercial LDH assay kit, according to the manufacturer\'s instructions.
2.6. Antioxidant and Lipid Peroxide Assay {#sec2.6}
-----------------------------------------
The MDA content was measured as an end product of lipid peroxidation. The defense systems against free radical attack were assessed by the measurement of the activities of SOD. The H9c2 cardiomyocytes were pretreated with or without STS for 24 h before cell exposure to different doses (2, 4, 6, and 8 Gy) of X-ray radiation, and they were cultured for another 24 h. Then, they were washed once with ice-cold PBS and lysed in ice-cold RIPA lysis buffer containing 1 mM phenylmethylsulfonyl fluoride and phosphatase inhibitor for 30 min. After centrifuging the lysates at 12,000 rpm and 4°C for 10 min, the supernatants were collected for measurements of MDA production and SOD activity. The activities of SOD and catalase as well as the contents of MDA were analyzed in H9c2 cardiomyocytes with or without STS pretreatment and exposure to X-ray radiation, according to the manufacturer\'s protocol (Nanjing Jiancheng Bioengineering Institute).
2.7. Cell Morphology {#sec2.7}
--------------------
The cells were seeded on coverslips, exposed to radiation, and incubated in the presence or absence of STS for 48 h under the same conditions described above. The culture medium was removed, Hoechst 33258 was added to each well, and the cells were incubated for an additional 10 min. The coverslips were then washed three times with PBS, placed onto glass slides, and covered with mounting medium. Next, the cells were harvested at the indicated times and washed twice with cold PBS. The nuclear fluorescence was visualized under a Zeiss Axioplan microscope (Zeiss, Germany).
2.8. Clonogenic Assay {#sec2.8}
---------------------
A clonogenic assay was performed to determine the effects of STS treatment on the survival and division of the radiation-treated H9c2 cells. Briefly, the cells were seeded into 60-mm dishes at a cell density of 1 × 10^3^ cells/well in triplicate, pretreated with STS for 24 h, and then irradiated at different doses (0--8 Gy) using an X-ray irradiator operating at a dose-pulse rate of 6 MV/min. At 24 h after the radiation treatment, the culture medium was replaced with fresh complete medium containing STS, and the cells were incubated at 37°C for 8--12 days to allow for colony formation. The colonies were fixed in methanol for 20 min and stained with 0.5% Giemsa for 30 min. The number of colonies (≥50 cells) was scored under a microscope. The surviving fraction was calculated as the ratio of the plating efficiency of the treated cells to that of the control cells.
2.9. Apoptosis Assay {#sec2.9}
--------------------
The effects of STS on apoptosis were determined using the Annexin V/PI double-staining method. After exposure to X-ray radiation for 24 h, the H9c2 cardiomyocytes pretreated with STS or control were centrifuged to remove the medium. Next, the cell pellet was washed with PBS and stained with Annexin V in binding buffer (10 mM HEPES, 140 mM NaCl, and 2.5 mM CaCl~2~) for 20 min. At 10 min before the end of incubation, PI was added to this cell suspension in order to stain necrotic cells, according to the manufacturer\'s instructions (BioVision, Inc., Palo Alto, CA, USA). The cells were analyzed with a FACSCAN flow cytometer (Becton Dickenson Biosciences, San Jose, CA, USA), and the stained cells in the FL1-H and FL2-H channels were analyzed.
2.10. Cell Cycle Detection {#sec2.10}
--------------------------
The H9c2 cells in the logarithmic growth phase were harvested and seeded in DMEM containing 10% FBS at a density of 2 × 10^6^ cells per flask. After allowing the cells to adhere, the culture medium was removed and replaced with medium containing STS or control, and the cells were cultured for 24 h before radiation. After X-ray radiation at different doses, the cells were cultured for 24 h, harvested, washed once with cold PBS, resuspended in 1 mL of PBS, and then fixed with 2 mL of dehydrated ethanol for 30 min. The cells were collected, washed once with PBS, and then stained with 50 g/mL PI at room temperature in the dark for 30 min. The cells were then analyzed by flow cytometry (Coulter XL, Beckman Coulter, Inc., Fullerton, CA, USA) to determine the cell cycle distribution.
2.11. Western Blotting Analysis {#sec2.11}
-------------------------------
After treatment with STS or control, the cells at 80% confluence were collected and washed three times with PBS. Radioimmunoprecipitation assay buffer was added to extract the total protein. Equal amounts of protein from the samples and controls were loaded onto 12% sodium dodecyl sulfate-polyacrylamide gels for protein separation. The proteins were then transferred to nitrocellulose membranes, and the membranes were blocked with 5% nonfat dried milk for 3 h. The membranes were then incubated with primary antibodies (p38 at 1 : 1000, p-p38 at 1 : 1000, Bcl-2 at 1 : 1000, Bax at 1 : 1000, and cleaved caspase 3 or caspase-3 at 1 : 1000) overnight at 4°C. After being washed with Tris-buffered saline (pH 7.2) containing 0.05% Tween 20 (TBST), the membranes were incubated with the secondary antibody at room temperature for 1.5 h. Finally, the membranes were washed with TBST and incubated with the enhanced chemiluminescence reagent to detect the proteins of interest. The levels of GADPH were used as loading controls.
2.12. Statistical Analysis {#sec2.12}
--------------------------
Statistical analyses of the results from the present study were performed using SPSS17.0 (version 12.0; IBM, USA). The results were expressed as means ± standard deviation (SD). Comparisons between the various levels of radiation with or without STS treatment were conducted by two-way analysis of variance (ANOVA); a paired *t*-test was used to evaluate differences between the same levels of radiation versus without treatment; comparisons between groups were performed using one-way ANOVA. A level of *p* \< 0.05 was considered statistically significant.
3. Results {#sec3}
==========
3.1. Radiation Induces Cell Growth Inhibition and Apoptosis in H9c2 Cells in a Dose-Dependent Manner {#sec3.1}
----------------------------------------------------------------------------------------------------
The H9c2 cells were exposed to different doses of radiation (0, 2, 4, 6, and 8 Gy), and the MTT assay was performed to evaluate the cell viability. Radiation decreased the cell viability in a dose-dependent manner ([Figure 1(a)](#fig1){ref-type="fig"}). The cell viability in the X-ray-treated groups was significantly lower than that of the control (*p* \< 0.05). Exposure to 6-Gy and 8-Gy radiation was associated with a greater loss of cell viability, compared with the 2-Gy and 4-Gy groups. LDH, which leaks from cells after plasma membrane disruption, can also be used as an indicator of cell injury. As shown in [Figure 1(b)](#fig1){ref-type="fig"}, a marked increase in LDH activity was observed at 24 h after X-ray exposure.
In order to determine the mechanisms of the decreased cell viability after radiation, we analyzed the number of apoptotic cells, cell division, and cell cycle distribution after radiation. The observed cytotoxic effects were also confirmed with Hoechst 33258 staining, a clonogenic survival assay, and a flow cytometric assay. The cells exposed to various doses (2, 4, 6, and 8 Gy) of radiation showed a significantly reduced number of colonies ([Figure 2(b)](#fig2){ref-type="fig"}). The number of colonies in the 2-Gy and 4-Gy groups was markedly higher than that in the 6-Gy and 8-Gy groups. The results from the assays demonstrated that X-ray radiation significantly suppressed the growth of H9c2 cells in a dose-dependent manner, compared with the control. In addition, the higher X-ray radiation doses increased cell necrosis and apoptosis. As shown in [Figure 3](#fig3){ref-type="fig"}, normal morphology and nuclei were observed in the control group. Chromatin condensation, indicative of apoptosis, was shown by Hoechst staining in the radiation groups. Typical apoptotic cells with fragmented chromatin, chromatin condensation, or apoptotic bodies were observed in the 8-Gy group ([Figure 3](#fig3){ref-type="fig"}). As shown in [Figure 4](#fig4){ref-type="fig"}, the percentages of apoptotic cells in the radiation groups were significantly higher than that of the control group.
3.2. STS Ameliorates Radiation-Induced Apoptosis in H9c2 Cells {#sec3.2}
--------------------------------------------------------------
To observe the effect of STS on cell viability and to select the proper STS concentration for subsequent assays, the H9c2 cells were exposed to various concentrations of STS (1.25--20 *μ*g/mL). The results of the MTT assays showed an increase in cell viability as the STS concentration increased from 1.25 to 10 *μ*g/mL. The proliferation rate was significantly increased when the cells were treated with 10 *μ*g/mL STS for 48 h ([Figure 5](#fig5){ref-type="fig"}). However, an inhibitory effect on cell proliferation was observed when the concentration of STS was greater than 20 *μ*g/mL. Although the mechanism for the high-dose-STS-induced cell growth inhibition was not clear, the STS concentration of 10 *μ*g/mL was selected for subsequent experiments.
The H9c2 cells were pretreated with STS (10 *μ*g/mL) for 24 h prior to exposure to different doses (2, 4, 6, and 8 Gy) of radiation. There was a significant difference in cell viability between the radiation alone group and the pretreatment with STS group. Compared with the same level of radiation group, the cell viability in the STS group was enhanced ([Figure 1(a)](#fig1){ref-type="fig"}). In addition, pretreatment with STS significantly decreased the LDH levels ([Figure 1(b)](#fig1){ref-type="fig"}). Next, clonogenic assays were performed to determine if STS treatment had any impact on radiation-induced H9c2 cell survival and division. Compared with the radiation groups, the number of colonies in the 2STS, 4STS, 6STS, and 8STS groups were 99 ± 4.9%, 55 ± 4.1%, 41 ± 4.8%, and 27 ± 3.7%, respectively, indicating that STS reduced the radiation-induced growth inhibition of H9c2 cells ([Figure 2](#fig2){ref-type="fig"}). STS treatment inhibited the apoptosis of the cultured cardiomyocytes. Hoechst 33258 staining demonstrated that the chromatin condensation and number of apoptotic bodies were decreased in the cells pretreated with STS ([Figure 3](#fig3){ref-type="fig"}). Similarly, STS treatment significantly decreased the percentages of apoptotic cells compared with the corresponding (2, 4, 6, and 8 Gy) radiation groups ([Figure 4](#fig4){ref-type="fig"}).
3.3. STS Exerts Antioxidant and Lipid Peroxidation Activities {#sec3.3}
-------------------------------------------------------------
During radiation, severe oxidative stress leads to lipid peroxidation \[[@B27]\]. To explore the mechanism underlying the antioxidant activity of STS, we evaluated the activities of MDA and SOD; the activities of SOD in the radiation alone groups were significantly decreased, compared with the control group ([Figure 6(a)](#fig6){ref-type="fig"}), and the contents of MDA in the radiation alone groups were 2.11-, 2.64-, 3.41-, and 4-fold higher than that in the control group ([Figure 6(b)](#fig6){ref-type="fig"}). Pretreatment with STS remarkably reversed these alterations ([Figure 6](#fig6){ref-type="fig"}). STS treatment significantly downregulated the contents of MDA and enhanced the levels of SOD in cells exposed to X-ray radiation.
3.4. STS Reverses the Effects of Radiation on the Cell Cycle {#sec3.4}
------------------------------------------------------------
The flow cytometric assay showed that, compared with the control, an increase in the population of cells at the G0/G1 phase in the radiation groups was seen after X-ray radiation in a dose-dependent manner, ranging from 60.9% to 84.14%, which was reversed by STS pretreatment ([Figure 7](#fig7){ref-type="fig"}). STS reduced G0/G1 cell cycle arrest.
Compared with the radiation groups, the percentages of cells at the G0/G1 phase were decreased (60.2% ± 2.8% versus 64.6% ± 3.4%, 68.12% ± 3.65%, 77% ± 4.12%, and 79.59% ± 4.1%, resp., *p* \< 0.05).
3.5. STS Modulates the Levels of Apoptosis-Related Proteins {#sec3.5}
-----------------------------------------------------------
Compared with the control, the protein expression levels of the antiapoptotic gene Bcl-2 were markedly decreased, and the proapoptotic gene Bax was markedly increased; therefore, the Bax/Bcl-2 ratio was increased accordingly in the radiation groups; the expression levels of caspase-3 were also increased ([Figure 8(e)](#fig8){ref-type="fig"}) in the radiation group. These effects were prevented by pretreatment with STS: the expression level of Bcl-2 was increased; the expression levels of Bax as well as caspase-3 were decreased, and the Bax/Bcl-2 ratio was decreased ([Figure 8](#fig8){ref-type="fig"}).
To further assess the signaling pathways potentially involved in the process of cardiomyocyte apoptosis induced by radiation, we examined the key components of the MAPK pathways in the radiation alone groups. As shown in [Figure 8](#fig8){ref-type="fig"}, there was a marked decrease in p-ERK and p38-MAPK in the radiation groups, compared with the control group. STS pretreatment upregulated the expression levels of p-ERK and p38-MAPK.
4. Discussion {#sec4}
=============
Radiation exposure from various medical sources, such as cardiac catheterizations for congenital heart disease and radiation therapy for cancer, is almost unavoidable and often causes irreversible cardiac injury \[[@B28]\] that is accompanied by increased oxidative stress. The overproduction of radicals leads to increased myocardial apoptosis, ultimately resulting in cardiac failure \[[@B29]\]. The severity of radiation-induced heart disease, in terms of the mortality and morbidity, has been reported previously \[[@B30]\]. However, the detailed mechanism underlying radiation-induced heart disease and its treatment and prevention strategies remain elusive \[[@B31]\].
In the present study, we first determined the cytotoxicity of radiation in H9c2 cardiomyocytes by assessing the cell viability, LDH leakage, clonogenicity, and apoptosis. X-ray radiation reduced the cell viability in a dose-dependent manner. The results obtained from the cell viability assay were consistent with the findings from the LDH assay. LDH leakage, as a marker of necrotic cellular death, was measured in the cell culture medium. There were significant differences in the LDH levels between the control group and the radiation alone groups. X-ray radiation-induced LDH release was dose-dependent. Compared with the control group, the X-ray-treated cells underwent apoptosis, as shown by Hoechst 33258 staining. Similar results were also obtained with a clonogenic assay, demonstrating that X-ray radiation promoted cardiomyocyte apoptosis and inhibited cell proliferation.
Given the central role of oxidative stress in the pathogenesis of radiation-induced heart disease, ameliorating oxidative stress through treatment with antioxidants might be an effective prevention strategy. STS has been utilized by traditional Chinese medicine practitioners for its antioxidant effects and has been shown to reduce the risk factors for several cardiovascular diseases \[[@B32]\]. The beneficial role of STS in improving radiation-induced heart disease, however, has not yet been demonstrated. In this study, we used a cell-based radiation-injury model to investigate the antioxidant effects of STS in order to demonstrate the possibility of using this agent for the prevention of radiation-induced cardiomyopathy. The MDA level indicates the membrane lipid peroxidation level during oxidative damage \[[@B33]\] and is indicative of cardiomyocyte oxidative damage, while the endogenous antioxidant SOD protects the cells against oxidative stress. In the present study, the treatment of cells with radiation increased the intracellular MDA levels and decreased the SOD levels, compared with the control cells, which were reversed by STS. STS treatment decreased the radiation-induced cell death, as measured by a clonogenic assay, and restored the alterations of H9c2 morphology and nuclear condensation to nearly normal levels. These results suggested that the antioxidant activity of STS might help protect the cells from radiation-induced cytotoxicity.
In the present study, the flow cytometric analysis results revealed that X-ray radiation alone significantly increased the percentages of necrotic and apoptotic cells, as compared with the control group; moreover, STS treatment significantly decreased the rate of apoptosis in cells exposed to radiation in a dose-dependent manner. STS significantly attenuated cell injury induced by radiation in H9c2 cardiomyocytes, suggesting that STS might be capable of protecting H9c2 cardiomyocytes against radiation-induced injury. The cardioprotective effect of STS was related to its ability to reduce myocardial cell apoptosis and damage induced by oxidative stress.
In the present study, cell cycle analysis indicated that radiation-induced damage increased the proportion of cells in the G0/G1 phase and reduced the proportion of cells in the S phase, demonstrating that X-ray radiation reduces cell survival by arresting cell cycle progression as well as inducing cell apoptosis. STS pretreatment increased the proportion of cells in the S phase. The number of S phase cells could reflect the ability of cells to proliferate, indicating that STS may inhibit radiation-induced cell cycle arrest in H9c2 cells.
Apoptosis occurs with Bax translocation to the mitochondria, which releases apoptotic factors during the apoptosis process \[[@B34]\]. Bcl-2 inhibits Bax activation, and caspase-3 is the effector molecule that is responsible for DNA fragmentation during the terminal events preceding cell death \[[@B35], [@B36]\]. Our results showed that X-ray exposure resulted in a decreased expression of Bcl-2, an increased expression of Bax and caspase-3, and an increased Bcl-2/Bax ratio. Compared with the radiation alone groups, the levels of Bcl-2 and Bcl-2/Bax increased while Bax and caspase-3 decreased markedly in the radiation + STS group, indicating that STS reduced myocardial cell apoptosis through modulating the expression of Bcl-2 family proteins.
p38-MAPK, the major member found in cardiac tissues, plays a key role in diverse biological processes with varied outcomes, either promoting cell death or ensuring survival. p38-MAPK is activated simultaneously with JNKs, which promote survival as in the case of cardiac myocytes exposed to oxidative stress \[[@B37]\]. The beneficial effects of p38 activation include an antihypertrophic action and enhanced myocardial adaptation to stress, perhaps through myocardial angiogenesis \[[@B38]\]. As far as ERKs are concerned, they are involved in regulation of the cell cycle, cell proliferation and differentiation, and cell migration as well as in the stress response. ERKs appear to function principally in favor of cell survival, exerting an antiapoptotic role \[[@B39]\]. In the present study, we also demonstrated that STS treatment activated the p38-MAPK pathway, which may be associated with its protective effect against radiation-induced injury in H9c2 cells. The expression of p-p38 was increased markedly, indicating that STS enhanced the survival of cells exposed to radiation, at least in part by promoting the p38-MAPK signaling pathways.
5. Conclusions {#sec5}
==============
Our results from the present study demonstrated that STS protected myocytes from X-ray radiation-induced injury via suppressing apoptosis-related gene activation and preventing inappropriate apoptosis. These findings suggest that STS may be a promising agent for treating radiation-induced cardiomyocyte injury.
This work was financially supported by the Cardiovascular Institute of Gansu Provincial Hospital and the School of Stomatology of Lanzhou University.
Competing Interests
===================
The authors declare that they have no competing interests.
Authors\' Contributions
=======================
Wenjing Zhang and Yi Li contributed equally to this work.
{#fig1}
{#fig2}
{#fig3}
{#fig4}
{#fig5}
{#fig6}
{#fig7}
{#fig8}
[^1]: Academic Editor: Raffaele Capasso
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Bacteriophage PR772 is a double-stranded DNA (dsDNA) virus with a 70 nm diameter icosahedral protein capsid encapsulating the internal lipid bilayer along with numerous proteins and a 15 kbp long linear genome. It belongs to the *Tectiviridae* family and infects gram negative hosts like *Escherichia coli*, *Salmonella typhimurium* and other bacteria, carrying a R772 plasmid-encoded receptor complex through which DNA can be transported during bacterial conjugation ([@bib15]; [@bib16]; [@bib27]; [@bib32]). Much of the functional knowledge of the viral proteins is inferred from previous studies on PRD1, a close relative of PR772 with a genome sequence identity of 97.2% ([@bib26]; [@bib36]). The most striking features of phages from those members of the *Tectiviridae* family that infect Gram-negative bacteria, is the presence of an inner lipid membrane and lack of a tail in the dormant viral particle. During the process of infection, these viruses produce a membranous tube derived from the inner membrane of the viral particle, lined from the inside by proteins P7, P11 and P32. This tube is used to inject the viral dsDNA into the host ([@bib30]; [@bib36]).
Genome analysis of PR772 identified 32 open reading frames (ORFs) containing at least 40 codons ([@bib26]). Twenty-eight annotated proteins are known to be expressed from the genome, of which three do not make it into the final assembly ([@bib10]; [@bib26]). In previous studies on PRD1, it was shown that the capsid is formed by proteins P3 (major capsid protein), P30 (capsid associated protein) and the vertex complex. The vertex complex includes the penton protein (P31), the host-recognition protein (P5) and the receptor-binding protein (P2) along with the infectivity protein (P16) that acts as a cementing protein, holding the vertex complex together and also stabilizing the pseudo-icosahedral capsid of the virus ([@bib36]). Proteins P6, P9, P20 and P22 are involved in DNA packaging and proteins P7, P14, P11, P18, P32 and P34 are responsible for DNA delivery to the host ([@bib20]; [@bib26]). The *Tectiviridae* family of viruses are also known to have structural similarities to adenovirus ([@bib36]).
In previous studies on bacteriophage PRD1, many models have been proposed to explain the architecture of the penton base and the vertex complex. Most of these models were speculations based on the observations from gene mutation/knock-out and in-vitro studies of proteins P31, P5 and P2 from PRD1. The in-vitro protein expression studies showed that P31 forms pentamers and P5 form multimers of trimers (i.e, (P5~3~)~1~, (P5~3~)~2~, (P5~3~)~3~) ([@bib38]). The gene knock-out studies showed that P31^-^ mutants produced incomplete PRD1 particles that lacked P31, P5 and P2 functions. They failed to form the vertex complex. A P5^-^ mutant produced intact viral particle but lacked both P5 and P2 functions ([@bib34]). These observations led to the hypothesis that the vertex complex appears to be a single spike formed by a pentameric P31 base binds the trimeric P5 spike protein to which P2 is bound ([@bib21]; [@bib11]; [@bib34]; [@bib38]). Later, a combination of SAXS modeling of the P5 protein and a low resolution cryoem study of the vertex region, showed that P2 and P5 form two spikes, not one, as previously described. Based on the low-resolution SAXS model of P5, it was speculated that the N-terminal base of P5 could have a size similar to P31 due to sequence similarity ([@bib21]). With the available experimental data, the generally accepted composition of the vertex complex of *Tectiviridae* is that P31 is a homopentamer and forms the penton base. P5 and P2 are attached to the P31 penton base but their arrangement is not known. ([@bib10]; [@bib21]). The high resolution x-ray crystallographic structure of the P2 protein is known but the location and orientation with respect to the penton base in its functional form is currently not known ([@bib41]). Putatively, it is suggested that the beta-propeller motif of P2 might be involved in binding to the host receptor.
Here we present the high-resolution structure of bacteriophage PR772 using electron cryo-microscopy (CryoEM). The high-resolution map helped us to resolve subtle variations in the protein conformations and their influence on the formation of a viable viral particle. The N-terminal region of the P3 subunits have three conformations, not two as previously described ([@bib1]). The new N-terminal P3 conformation plays an important role in accommodating the P30 protein during particle assembly. The C-terminal region of P3 not only shows the formation of a β-sheet with P30 but also helps in locking the adjacent trisymmetrons through a hinge mechanism, thus facilitating the formation of icosahedral particles and regulating their size. Localized asymmetric reconstruction of the vertex region of PR772 revealed a P5-P31 heteropentameric base and the binding of P2 to P5 in the complex. A combination of high-resolution icosahedral symmetrized single-particle reconstruction, localized asymmetric reconstruction and focused classification has enabled us to answer some of the intriguing questions about the particle architecture, composition of the penton base and arrangement of the vertex complex.
Results {#s2}
=======
High resolution capsid map at 2.3 Å {#s2-1}
-----------------------------------
The structure of bacteriophage PR772 was determined by electron cryo-microscopy. The overall resolution determined by Fourier shell correlation (FSC) \@0.143 was 2.75 Å ([Figure 1---figure supplement 1](#fig1s1){ref-type="fig"}). The local resolution estimated using the two unfiltered final half maps with ResMap ([@bib24]) showed that most of the capsid was resolved to 2.3 Å ([Figure 1A and B](#fig1){ref-type="fig"}). The resolution of the regions that interact with the inner lipid bilayer was lower at about 3.2 Å ([Figure 1C](#fig1){ref-type="fig"}). The areas around the icosahedral five-fold axes had resolutions varying between 2.3--3.0 Å ([Figure 1D--F](#fig1){ref-type="fig"}). At a root mean square deviation (RMSD; deviation away from noise as visualized in Coot, where noise is 0) of 4.2, the side chains of the amino acid residues were visible for most of the capsid region. The inner membrane layers of lipid, protein and dsDNA were smeared due to averaging and symmetry mismatch. The resolution in most of the regions was high enough (3.2 Å -- 2.3 Å) to build a de-novo model of the asymmetric unit comprising of P3 ([Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}), P30 ([Figure 2---figure supplement 2](#fig2s2){ref-type="fig"}), P5 ([Figure 4---figure supplement 3](#fig4s3){ref-type="fig"}), P31 ([Figure 4---figure supplement 3](#fig4s3){ref-type="fig"}) and P16 ([Figure 6---figure supplement 1](#fig6s1){ref-type="fig"}) into the icosahedrally averaged CryoEM map ([Figure 1H](#fig1){ref-type="fig"} and [Table 1](#table1){ref-type="table"}) ([Supplementary file 1](#supp1){ref-type="supplementary-material"}).
{#fig1}
10.7554/eLife.48496.003
###### Data collection, Processing and Model refinement parameters.
Parameter Value
------------------------------------- ----------------------
**Data collection**
Voltage (kV) 300
Magnification (x) 130000
Å/pix 1.06
Energy Filter with Slit (eV) 20
Frames per Micrograph 40
Total Dose (e^-^ /Å^2^) 40
Micrographs 3220
Defocus Range (μm) 0.8--2.6
**Data Processing**
Micrographs 3200
Frames used 4--40
Å/pix 1.06
Particles (Total) 46348 (56275)
Symmetry Applied I4
Overall Resolution\@FSC~0.143~ (Å) 2.75^\*^
B-factor (Å^2^) −104.93^\*^
**Model Refinement**
Composition
Chains 19
Atoms 41731 (Hydrogens: 0)
Refinement
CC~mask~ 0.7961
CC~volume~ 0.7941
ADP (B-factors)
Iso/Aniso 41731/0
Mean 101.95
RMS deviations
Bonds (Å) 0.006
Angles (^O^) 0.806
EMRinger Score 5.40
MolProbity validation
Clash score, all atoms 3.35
MolProbity score 1.57
Rotamer Outliers (%) 0.16
Ramachandran
Favoured (%) 93.04
Allowed (%) 6.83
Outliers (%) 0.13
^\*^ Calculated by RELION.
The CryoEM 3D reconstruction of PR772 shows that the viral particle follows a pseudo T = 25 lattice architecture with a (h,k) of (0,5) resulting in an icosahedral structure with 20 large trisymmetrons and 12 penta-symmetrons ([@bib13]; [@bib40]). However, the analysis revealed that the penta-symmetrons were hetero-pentamers. Each of these trisymmetrons have 36 copies of P3, the major capsid protein (MCP), arranged as 12 trimers, each of which structurally appears to be hexagonal in shape. At the fivefold vertices, typical penta-symmetrons are replaced by vertex complexes to complete the icosahedral shell. With a pseudo T = 25 architecture, PR772 is one of the larger wild type viruses resolved to a resolution below 3 Å.
Major Capsid Protein and its conformations {#s2-2}
------------------------------------------
P3 is the major capsid protein, which builds up the trisymmetrons of the capsid, and it is the most abundant protein found in bacteriophage PR772 ([Figure 2A](#fig2){ref-type="fig"}). The P3 monomers (subunits a, b and c) are interlocked to form a trimer that appears as a hexon ([Figure 2B--K](#fig2){ref-type="fig"}). They exhibit the double-barrel trimer arrangement, as previously seen in viruses of the adenovirus linage ([@bib9]; [@bib31]). The Leu130 - Ala150 loops from subunits a, b and c of P3 interact with each other in a cyclic manner at the centre of the trimer complex to stabilize it ([Figure 2](#fig2){ref-type="fig"}). The monomers have a similar structure in bulk, with minor differences to accommodate the more significant variations in the C and N-terminal conformations. The asymmetric unit has four such unique trimers along with P30, P16 and a penton protein. [Figure 2B--K](#fig2){ref-type="fig"} show the four unique trimers and their subunit arrangement.
{#fig2}
The N-terminal of the P3 monomers have three different conformations; a helix turn helix, a long helix and a long helix with a kink ([Figure 2---figure supplement 1](#fig2s1){ref-type="fig"}). [Figure 2D](#fig2){ref-type="fig"} is a good schematic to visualize all N-terminal conformations in a single trimer (subunit a shows the helix-turn helix, subunit b shows the long helix and subunit c shows the helix with a kink). When the N-terminal adopts a helix turn helix, the shorter helix close to the N-terminal is bent away from the lipid membrane and interacts with the adjacent subunit of the trimer. The long helix with a kink behaves similar to a helix turn helix where the kink twists the helix away from the membrane but it is not embedded deeply into the adjacent subunit of the trimer. To accommodate the helix turn helix or the helix with a kink in the P3 subunits, the loop formed by residue Tyr351 -- Val358 in these subunits is flipped. In case of the long helical conformation, the N-terminal residues Met1 -- Gln6 anchor the P3 subunit to the lipid membrane ([Figure 3A](#fig3){ref-type="fig"}).
{#fig3}
The N-terminal region of subunit a is more flexible as compared to the same region of subunits b or c in the four unique trimers. The N-terminal of subunit a can form either a helix turn helix bent away from the membrane ([Figure 2D--E](#fig2){ref-type="fig"}) or a long helix that anchors the subunit to the membrane ([Figure 2C,F](#fig2){ref-type="fig"}). The N-terminal region of subunit c either forms a helix turn helix ([Figure 2C,E,F](#fig2){ref-type="fig"}) or a long helix with a kink ([Figure 2D](#fig2){ref-type="fig"} and [Figure 3B](#fig3){ref-type="fig"}). The N-terminal of the b subunits always forms a long helix that anchors the subunit to the membrane ([Figure 2C--F](#fig2){ref-type="fig"} and [Figure 3A](#fig3){ref-type="fig"}). Subunit b of trimer 1, which is present close to the penton has the Tyr351 -- Val358 loop flipped even though there is no helix turn helix or a long helix with a kink in this region that must be accommodated. As it will be shown later, this loop accommodates P30 ([Figure 3B](#fig3){ref-type="fig"} and [Figure 2---figure supplement 3](#fig2s3){ref-type="fig"}).
The C-termini are more variable as compared to the N-termini of the trimers. They adopt four different conformations; one conformation is a long strand extending towards the lipid membrane and found in subunit b of all trimers ([Figure 2C--F,H--K](#fig2){ref-type="fig"}). The second conformation extends away from the membrane into the peripheral space between the trimers and it is also the most common C-terminal conformation. Two of the four instances in subunit a ([Figure 2D,E,I,J](#fig2){ref-type="fig"}) and two of the four instances in subunit c ([Figure 2E,F,J,K](#fig2){ref-type="fig"}) adopt this conformation. The third conformation is seen in subunit a of trimer 1, where the C-terminal runs parallel to the long N-terminal helix of the same subunit and it is embedded into the adjacent trimer 1 of the neighboring trisymmetron ([Figure 2C,H](#fig2){ref-type="fig"}). The fourth type of C-terminal conformation is seen in 2 instances of subunit c when they reside close to P30 ([Figure 2C,D,H,I](#fig2){ref-type="fig"}). Here, the C-terminal is elongated and runs towards the lipid membrane, grazing it.
Penton base is a heteropentamer of P5 and P31 {#s2-3}
---------------------------------------------
The penton region of the icosahedrally symmetrized CryoEM map of PR772 was sectioned from the whole viral map using UCSF Chimera ([@bib18]). To generate an initial model for de-novo model building of the penton region, PHENIX ([@bib2]): Find Helices and Strands was used with both RESOLVE and PULCHAR options enabled. Due to the high sequence similarity between P31 and the N-terminal domain of P5, we could not rule out that any of these two proteins or if a mixture of these two proteins could potentially form the penton base. Therefore, two initial models were generated. One model used the P31 protein sequence and the other used the P5 protein sequence as part of the input. In the initial observation of all the predicted segments of the two models, the P31 protein segment with residues 112--126 and the P5 protein segment with residues 107--121 occupied the same region of the sectioned map ([Figure 1E,F](#fig1){ref-type="fig"}). On closer inspection of the side chains from the two models and their fit into the CryoEM map densities in this region, the P5 protein side chains fit clearly at 3.2 RMSD and the P31 side chains could only be fitted at RMSD values lower than 2.6. As the CryoEM map was generated by an icosahedrally symmetrized reconstruction, we assumed that the densities of P5 and P31 were averaged. This hinted that the penton base could be a heteropentamer formed by both P5 and P31.
With the assumption that the penton base could be a heteropentamer, the P5 N-terminal domain (residues 1--124) and P31 (residues 1--126) were modeled using the icosahedrally averaged map at \~3.0 and\~2.2 RMSD respectively in *Coot* ([@bib17]) ([Figure 4E](#fig4){ref-type="fig"} and [Figure 4---figure supplement 3](#fig4s3){ref-type="fig"}). The prominence of the amino acid side chain densities varied based on the protein sequence conservation between P5 and P31 ([Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}). All identities and most of the conserved substitutions in the sequence alignment resulted in clearer well-defined amino acid side chain densities and the non-conserved substitutions resulted in amino acid side chain densities that were averaged between the respective side chain densities ([Figure 1D--F](#fig1){ref-type="fig"} and [Figure 4---figure supplement 2](#fig4s2){ref-type="fig"}).
{#fig4}
Another unexpected feature was observed at a lower RMSD of about 1.0, a branching of the density close to the N-terminal region of either P31 or P5. One prominent branch, which was earlier used for modeling and another branch that is only visible at lower RMSD. This suggested that the N-terminal region of either the P31 or P5 could have an alternate conformation. On inspecting the residues from both the P31 model and the N-terminal domain model of P5, that were close to the branch region, P31 was ruled out as a potential candidate. P31 has Val11-Thr10-Met9 residues, which could not be fitted into the branched density without severely distorting the Cα backbone. P5 in the same region has Ser9-Gly8-Gly7. The 2 Glycine residues provided the needed backbone (Cα chain) flexibility that could facilitate this 'special case' of N-terminal conformation ([Figure 4G](#fig4){ref-type="fig"} and [Figure 4---figure supplement 5](#fig4s5){ref-type="fig"}). In a typical arrangement, the N-terminal ends of both P5 and P31 would hug its neighboring subunit counter-clockwise and stabilize the penton complex ([Figure 4---figure supplement 7](#fig4s7){ref-type="fig"}). In the special case as described above, the N-terminal end of the P5 protein wedges itself between two adjacent P3 subunits ([Figure 4H](#fig4){ref-type="fig"}).
In the icosahedrally symmetrized reconstruction, at lower contour levels (0.065 in chimera), the map showed smeared densities above the 5-fold vertices of the viral particle ([Figure 4A--B](#fig4){ref-type="fig"}). This could be due to a symmetry mismatch of the proteins present in the region. Accordingly, the smeared region over the five-fold vertex was isolated and resolved by capsid signal subtraction followed by localized asymmetric (C1) reconstruction (see Materials and methods). All the classes generated by 3D classification showed a single protruding density except one of the classes, which revealed two significant densities; one poorly resolved knob-like density and another more well resolved density closely interacting with one of the monomers of the penton ([Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}). On closer inspection of every 3D class generated during the process of localized asymmetric reconstruction, we noticed that three of the subunits of the penton base had a stem-like protrusion close to the 5-fold axis, extending outwards and interacting with each other forming a thick stalk ([Figure 4C--D](#fig4){ref-type="fig"}). The other two subunits lacked the stem-like protrusion. The class that revealed the two significant densities also showed that one of the densities interacted with the thick stalk ([Figure 5A--B](#fig5){ref-type="fig"}).
{#fig5}
The number and arrangement of P5 and P31 forming the penton were confirmed by two independent methods, a localized asymmetric reconstruction and a focused classification to a resolution of 4.41 Å and 4.25 Å, respectively ([Figure 4---figure supplement 6](#fig4s6){ref-type="fig"}). All the classes from the 3D classification by localized reconstruction showed that only three subunits formed the stem-like protrusion that interacted with one another to form a thicker stalk. P31 terminates close to the fivefold and thus cannot form the stem-like protrusion whereas P5 residues (121-124) continue up and outward and these residues have the potential to form the stem-like protrusion ([Figure 4E](#fig4){ref-type="fig"} and [Figure 4---figure supplement 4](#fig4s4){ref-type="fig"}). Three copies of the model of the P5 N-terminal base were fitted into the penton density from the localized asymmetric reconstruction. These models did indeed fit the density ([Figure 4F](#fig4){ref-type="fig"}). The overall map:model correlation reduced when the models of P31 and P5 N-terminal domain were swapped with each other in the localized reconstruction and the focused classification density maps. The orientation and alignment of P5 residues (121-124) also confirm previous predictions, which showed the formation of a triple helix with a collagen-like motif (residues 124--140) ([@bib21]; [@bib11]). The poorly resolved knob-like density represents the trimerized C-terminal domain of P5 ([@bib11]; [@bib29]). By this it can be concluded that 3 copies of P5 and 2 copies of P31 form the penton base in PR772 ([Figure 4F](#fig4){ref-type="fig"}).
P2 monomer is bound to P5 and stabilized by the P5 stalk {#s2-4}
--------------------------------------------------------
With the localized asymmetric reconstruction and focused classification methods, the structure and composition of the penton base was established. It is now known that the stalk-like density observed, emanating from the penton base, is built with three copies of the P5 protein, forming a collagen-like motif and eventually the unresolved knob domain.
As described previously, one of the 3D classes, with 467,631 sub-particles from a total of 3,340,224 sub-particles, also revealed an extra rigid density closely interacting with the penton base and the protruding stalk ([Figure 5A--C](#fig5){ref-type="fig"} and [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}). In previous studies, it was shown that P2 and P5 could be interacting at the vertex ([@bib6]). The crystal structure of P2 (PDB:1N7U) from PRD1 (98% protein sequence identity to PR772) was fitted into the density. The fit was good and the density that was resolved by localized reconstruction was sufficient to fit the domains I and II, forming the head and the domain III, forming the tail of P2. The off-centred arrangement of the tail with respect to the head allowed us to specifically assign the map density to the respective regions of the P2 protein ([Figure 5A--C](#fig5){ref-type="fig"}).
The interaction of P5 with P2 can be compared to a ball and socket joint, for instance as seen in a human shoulder, where the N-terminal base of P5 is the ball with domain I and domain II of P2 acting as a socket ([Figure 5---video 1](#fig5video1){ref-type="video"}). The estimation of the electrostatic surface potential of both P2 and P5 shows good charge complementarity in and around the regions of interaction. These regions also have similar hydrophobicity (Kyte-Doolittle scale) ([Figure 5---figure supplement 2](#fig5s2){ref-type="fig"} and [Figure 5---video 1](#fig5video1){ref-type="video"}). P2 also interacts with the collagen-like motif of the P5 stalk, thus introducing some rigidity to the C-terminal of the P5 trimer. The presence of P2 nudges the P5 stalk by \~36^O^ from its usual position ([Figure 5D--E](#fig5){ref-type="fig"}). The P5 stalk acts as a linchpin that locks the P2 molecule and restricts it from swivelling around the 5-fold vertex, thus stabilizing the complex.
The occupancy of P2 on the 5-fold vertices appears to be significantly lower in bacteriophage PR772 than what was observed from other members of the *Tectiviridae* family. The 3D classification of the signal subtracted and isolated vertices revealed that only 467,631 sub-particles from a total of 3,340,224 extracted sub-particles, showed the presence of a density representing P2. This accounts for 16% of all the sub-particles used or about two vertices in an intact viral particle, assuming an equal distribution among the viral particles.
Overall architecture of PR772 {#s2-5}
-----------------------------
P30 adopts an extended conformation in an intact viral particle and it is found wedged in between the trisymmetrons ([Figure 2A](#fig2){ref-type="fig"}, [Figure 6---figure supplement 2](#fig6s2){ref-type="fig"}). Two copies of P30, interlocked at the N-terminal hook, span between the adjacent vertices. In an intact viral particle, P30 forms a cage-like structure, which stabilizes the trisymmetrons themselves and also interact with the neighboring trisymmetrons to form a closed network ([Figure 2A](#fig2){ref-type="fig"} and [Figure 6---figure supplement 2](#fig6s2){ref-type="fig"}). The residues Tyr62 -- Ile64 and Val32 -- Arg35 of P30 form beta sheets with residues Thr384 -- Leu386 from subunit c of P3 trimer one and residues Thr384 -- Asn388 from subunit c of P3 trimer two respectively ([Figure 2A](#fig2){ref-type="fig"} highlighted regions). These C-terminal regions of P3 are sandwiched between P30 and the lipid membrane ([Figure 2A](#fig2){ref-type="fig"} highlighted regions). P30 is also sandwiched between two adjacent P3 subunits of the trimer from a neighboring trisymmetron ([Figures 2A](#fig2){ref-type="fig"} and [6B](#fig6){ref-type="fig"} and [Figure 6---video 1](#fig6video1){ref-type="video"}).
{#fig6}
Five copies of P16 bridge the penton to the five trisymmetrons that meet at the vertex complex. The structure of P16 was only partially resolved. The long C-terminal trans-membrane helix of P16 (Leu7 - Ala28) was poorly resolved and barely visible at 0.6 RMSD. The high-resolution structure of the disorganized region (Tyr58 - Ile96) of P16 also evades us. It was poorly resolved with no visible side chains and part of the density map representing the Cα backbone was missing even at very low RMSD. Due to poor signal to noise ratios, these regions were not modeled, but the difference density map of the vertex region and the modeled penton showed that the C-alpha backbone continued into the empty pocket beneath the penton and also interacts with the penton close to the 5-fold axes ([Figure 6---figure supplement 3](#fig6s3){ref-type="fig"}). However, the partial model shows that P16 forms a clamp-like complex using the loop Asn44 -Val55 and helix Asn101 - Ala115, which attaches onto the loop Val242 - Tyr247 of one of the P3 subunits of trimer one and locks it with the adjacent P3 subunit of trimer one from the neighboring trisymmetrons ([Figure 6B](#fig6){ref-type="fig"}). This further locks the P3-P30-P3 sandwich and makes it stable. As was mentioned earlier, to accommodate the P30 protein in the P3-P30-P16 complex ([Figure 6](#fig6){ref-type="fig"} and [Figure 6---video 1](#fig6video1){ref-type="video"}), the loop formed by residues Tyr351-Val358 of the P3 subunit in trimer one is flipped compared to the orientation seen in this loop of the P3 subunit with a N-terminal long helix ([Figure 2---figure supplement 3D--F](#fig2s3){ref-type="fig"}).
At 2.0 RMSD, we noticed that the density for the C-terminal Gly84 of P30 was pointing towards the penton base and away from the inner lipid membrane. This is different from what is seen in bacteriophage PRD1 ([@bib1]). In case of P5, the C-terminal Gly84 of P30 seems to interact with the Met19 residue of a P5 subunit by a hydrophobic interaction ([Figure 6A](#fig6){ref-type="fig"}). With P31, the hydrophobic interaction is not evident.
Three copies of P5 and two copies of P31 are held together by their N-terminal residues. Met9 -- Val14 of P31 and Ser9 -- Tyr13 of P5 inter-digit with the neighboring subunit of the pentamer as β-sheets. Residues Gln8 -- Asn2 and Gly8 -- Met1 of P31 and P5 respectively continue further and hug the neighboring subunits ([Figure 4---figure supplement 7](#fig4s7){ref-type="fig"}). As was mentioned earlier, residues Gly8 - Met1 of P5 can in special cases have an alternate conformation where they are wedged in between the adjacent P3 subunits ([Figure 4H](#fig4){ref-type="fig"}). There is no obvious direct interaction between the penton and the P3 subunits of the capsid except for the special cases mentioned above. The disorganised region of P16 seems to interact with the penton base and helps in binding the vertex complex to the trisymmetrons ([Figure 6---figure supplement 3](#fig6s3){ref-type="fig"}).
Discussion {#s3}
==========
Each trisymmetron of PR772 is composed of 12 trimers of the P3 protein that are in turn bound by P30, an overall arrangement similar to that of the close relative PRD1 ([@bib1]). The high-resolution structure of PR772 confirms the previously shown variability of the N and C-terminals of the P3 protein depending on the locations within the trisymmetron. The high-resolution also allows us to extend the analysis of these conformations and show that the N-terminal of P3 can adopt three different conformations, not two as previously described. ([@bib1]; [@bib10]) ([Figures 2A](#fig2){ref-type="fig"} and [3B](#fig3){ref-type="fig"}). The newly discovered conformation of the N-terminal region of P3, a long helix with a kink, is structurally critical to accommodate the interlocking region formed by the N-terminal hooks of P30 during the initiation of particle assembly. The N-terminal region of P3 shows plasticity and play an important role in both stabilizing the trimer and anchoring the trimer complex to the membrane. The C-terminal region of P3 also shows differences in its conformation as compared to PRD1. In trimer 1 ([Figure 2A,C and H](#fig2){ref-type="fig"}) subunit a and subunit c have elongated C-terminal regions that interact with the adjacent P3 subunits of the neighboring trisymmetrons and in-turn locking on to P30 ([Figure 2](#fig2){ref-type="fig"} right hand side insert; and [Figure 6](#fig6){ref-type="fig"}). A similar arrangement can be seen in the C-terminal region from subunit c of trimer two and subunit a of trimer 4 of the neighboring trisymmetrons ([Figure 2](#fig2){ref-type="fig"} left hand side insert). The presence of P16 close to the penton region makes the interaction of all the neighboring trimer one more stable ([Figure 6](#fig6){ref-type="fig"}) and also anchors the whole vertex complex to the membrane ([Figure 7](#fig7){ref-type="fig"}), emphasizing the role of P30 and P16 in maintaining the size and structure of the icosahedral viral capsid.
{#fig7}
Our studies show that unlike what was predicted for PRD1, the penton base of PR772 is an asymmetric heteropentamer consisting of three copies of P5 and two copies of P31 ([Figure 4F](#fig4){ref-type="fig"}). P31 has high sequence similarity with the N-terminal domain of P5. Previously, it was shown that P31 can replace P5 to form the penton and produce an intact viral particle ([@bib6]). Even though the viral particles were intact, they were non-infectious due to the lack of the viral receptor-binding protein, P2, that is bound to the host recognition protein, P5 ([@bib6]; [@bib19]). In another study with the PRD1 sus525 mutant that lacks P31, it was also shown that these particles lacked the vertex complex ([@bib34]). With the current model, P31 would promote the formation of an intact viral particle by avoiding the steric-hinderances that could occur during the formation of the vertex complex, if the P31 subunits were replaced by P5. A penton with 3 copies of the P5 subunits will support (i) the formation of a stable collagen-like motif ([Figure 7](#fig7){ref-type="fig"}) and (ii) the formation of the C-terminal host recognition domain of P5. A heteropentameric base of the vertex complex with three copies of P5 and two copies of P31 address both these issues.
The members of the *Tectiviridae* family have been shown to have structural similarities with adenoviruses despite infecting different hosts. In adenoviruses, the symmetry mismatch seen in the vertex region is solved by the interaction of the tail region of the trimeric fibre with three of the five grooves formed by the subunits of the penton base ([@bib12]). Unlike in adenoviruses, a similar problem in PR772 is solved by domain swapping. The trimeric P5 protein responsible for the formation of the spike, swaps the N-terminal base domain with three copies of the P31 protein ([Figure 4F](#fig4){ref-type="fig"}).
The localized asymmetric reconstruction shows that P2 is bound to P5 ([Figure 5](#fig5){ref-type="fig"}). In the current CryoEM data, the orientation of P2 with respect to the penton base is reversed when compared to what was described for PRD1 ([@bib21]; [@bib41]). Here, the beta-propeller motif of P2 with domain I and II, interacts with the N-terminal base of P5 and also with the stalk. This was also speculated on in a previous study, where it was noted that members of the *Tectiviridae* family with large sequence variation in the beta-propeller motif of P2 also showed variations in the P5 subunit as a compensatory effect ([@bib36]). P2 interacts with both the N-terminal base and the stalk region of P5 but does not interact directly with any other structural protein in PR772. The interaction of P2 with the P5 stalk stabilizes the C-terminal region of P5 ([Figure 5](#fig5){ref-type="fig"}). In the case of PRD1, P2 seems to occupy all vertices except for the unique packaging vertex ([@bib10]; [@bib11]; [@bib36]). In this scenario, the special conformation in which the N-terminal residues of P5 are wedged in between the neighboring P3 subunits was found to be prominent ([@bib1]). In the current model of PR772, P2 seems to occupy only 2 vertices of the icosahedral shell. The lower quantities of P2 in PR772 were also indicated in a previous study, comparing the abundance of different viral proteins from various members of the *Tectiviridae* family using western blots ([@bib36]). In PRD1, it was shown that the loss of P2 from a packed viral particle, resulted in spontaneous release of genomic DNA, leading to loss of infectivity and the formation of empty particles ([@bib19]). In a previous study of PR772, it was shown that after purification, the concentration estimates of the particle by plaque assay and nano-particle tracking analysis were similar ([@bib32]). PR772 particles remained intact and infectious after purification. The intact PR772 particles can also be observed in the CryoEM micrographs ([Supplementary file 3](#supp3){ref-type="supplementary-material"}). So, it is unlike that P2 was lost during the purification.
In the icosahedrally averaged model of PR772, we see that the special conformation of P5 where the N-terminal residues are wedged between the P3 subunits of the adjacent trimer one is not prominent, but these P5 densities are visible at lower RMSD values. The presence of P2 in the vertex complex may coincide with the N-terminal region of P5 adopting the special conformation. It was reported that, compared to the CryoEM map of a wild type PRD1 virion, the CryoEM map of *sus690*, a mutant PRD1 virion that lacks P2 and P5, showed absence of density in the region where we see the N-terminal of P5 wedged between two neighboring P3 subunits. The lack of this density was attributed to the conformational changes in P31 due to the binding of P5 or the presence of P5 in between P31 and the P3 trimer ([@bib21]). In light of the current findings in PR772, P5 is known to be part of the heteropentameric base forming the penton and the absence of the above-mentioned density confirms that P5 adopts the special N-terminal conformation and not P31.
The above mentioned observations and the quantitative analysis of different proteins from PRD1 and PR772 using western blots ([@bib36]) could explain the discrepancy in P2 packing in PR772. As mentioned earlier, the presence of P2 in the vertex complex may coincide with the N-terminal region of P5 adopting the special conformation. This special conformation of N-terminal region of P5 could be linked to the low copy number of the membrane protein P14 (or higher copy number of P7) observed in PR772 when compared to PRD1 ([@bib36]). In PRD1, gene VII codes for P7 and the 3' region of the same gene also codes for P14. P14 is smaller and it consists only of the membrane anchor domain and lacks the trans-glycosylase domain that is seen in P7. These proteins, P7 and P14, are proposed to forms a hetero-multimeric complex located under the vertices of the PRD1 particle ([@bib28]; [@bib35]). In an intact PRD1 particle, the copy numbers of P7 and P14 are similar but in case of PR772, the copy number of P14 is significantly lower and the copy number of P7 seems higher ([@bib28]; [@bib36]). In PR772, the copy number of P14, the occurrence of the special conformation of the N-terminal region of P5 protein and the number of P2 proteins packed in an intact viral particle seems to be related. In PR772, the presence of the larger P7 protein in higher numbers at the vertex (along with other proteins in the disordered region) might hinder the formation of the special conformation of the N-terminal region of P5, which could reduce the number of P2 proteins packed in the particle.
The penton has no obvious direct interaction with the neighboring P3 subunits of the capsid, except in the rarely occurring special conformation of P5 where the N-terminus is wedged in between the P3 subunits of trimer one as mentioned above ([Figure 4G--H](#fig4){ref-type="fig"}). The difference density map of the vertex region and the modeled penton shows that the disordered region of P16 interacts with the penton ([Figure 6---figure supplement 3](#fig6s3){ref-type="fig"}). These interactions of P16 with P3, P5, P31 along with the membrane seem to play a central role in the formation of the vertex complex and anchoring it to the membrane. In earlier Raman spectroscopy studies on PRD1, it was shown that in the initial stages of viral assembly, P3 and P5 formed a precursor shell assisted by assembly factors and other membrane-associated proteins such as P16 ([@bib4]; [@bib5]). Similarly, in PR772 the P5 trimers could be accommodated during the formation of the procapsid and two copies of P31 are later added to the penton region, forming a heteropentameric penton.
From the localized asymmetric reconstruction of the vertex complex, one can notice that there is a variation in the map density below the penton. The map density below the P5 subunits is significantly different when compared to the same region below the P31 subunits. This suggests that, along with the interactions of P16 with P5 and P31, there could also be poorly resolved proteins like P11, P7, P14 or P18, etc; in the region ([@bib25]; [@bib28]; [@bib35]). The variation in the map density in this region is also compounded by other interactions, like that of the C-terminal Gly84 of P30 with the Met19 of P5 subunits and lack thereof with respect to the P31 subunits.
Preliminary results for the whole particle asymmetric reconstruction using symmetry relaxation in EMAN2 of the wild type PR772 do not show the presence of a unique packaging portal in the dormant particle in contrast to PRD1. All the vertices of PR772 show the heteropentameric arrangement of the penton with three subunits showing the stalk and the other two without the stalk ([Figure 4C--D](#fig4){ref-type="fig"}, [Supplementary file 2](#supp2){ref-type="supplementary-material"}).
[Figure 7](#fig7){ref-type="fig"} shows our model of the vertex complex in PR772. Using this model, we propose a mechanism for the initiation of infection. In analogy to studies on PRD1, it is known that P5 is needed for host recognition, but the binding of P5 to the host is transient ([@bib19]). The host binding is stabilized by the high-affinity interaction of P2 and its receptor, locking the viral particle to the host ([@bib19]). The relative changes between P5 and P2 upon binding to the receptor on the host could trigger the disruption of the vertex complex by pulling the N-terminal region of the P5 trimers that are wedged between the P3 subunits. This will in turn disrupt the interaction between the P5 proteins and P30 as seen in [Figure 6](#fig6){ref-type="fig"}. This disruption cascades further and disturbs the interaction between P30 and P16, rendering the whole vertex complex unstable. P30 dimers could also transduce these effects to the neighboring vertices. The disruption of the vertex complex exposes the viral membrane and membranous proteins like P18, P11/P7, P32 etc; to the host surface to facilitate the formation of the membranous tube for DNA transport. P16, that holds the vertex complex, along with other membranous proteins, could act as a protein tether that would help in moving the lipid membrane closer to the host ([Figure 6---figure supplement 4](#fig6s4){ref-type="fig"}).
Materials and methods {#s4}
=====================
Preparation and purification of PR772 {#s4-1}
-------------------------------------
Bacteriophage PR772 (ATCC BAA-769-B1) was propagated on *Escherichia coli* K12 J53-1. It was purified as previously described ([@bib32]) and further concentrated to facilitate testing various concentrations of the viral sample during grid optimization for CryoEM. This method yielded about 2--4 mL of viral particles with a concentration of 1 mg/mL using 10--20 agar plates. The sample was further concentrated to 20 mg/mL by using an ultracentrifuge. The sample was added into an ultracentrifuge tube and then a solution of Caesium Chloride in buffer (HEPES 20 mM, NaCl 100 mM, MgSO~4~1 mM, EDTA 1 mM, pH 8.0) at a density of 1.34 g/mL (g/cm^−3^) was gently layered on top. The mixture was centrifuged at 100,000 × g for 30 mins. The intact viral particles migrated to the top as a fine band and the broken particles along with any free DNA that was released from the broken particles stayed at the bottom of the tube. The top band with the intact viral particles was extracted using a needle and syringe. The concentrated sample was dialyzed over-night with the above-mentioned buffer to remove caesium chloride.
CryoEM grid preparation and data collection {#s4-2}
-------------------------------------------
The condition for CryoEM grid preparation was optimized for collecting a large number of particle images. For vitrification of the viral sample by plunge freezing into liquid ethane, we used a Vitrobot Mark IV (ThermoFisher). The best grid condition with uniform sample distribution was obtained by applying 3 μL of 7 mg/mL concentrated viral sample solution on a glow-discharged C-Flat grid CF-2/2--2C under 100% humidity at room temperature.
The data were collected on a Titan KRIOS (ThermoFisher) equipped with a K2 Summit (Gatan) direct electron detector and a GIF Quantum LS (Gatan) energy filter. All the data were collected at a magnification of 130 k in EFTEM mode with a pixel size of 1.06 Å. The slit width of the energy filter was 20 eV. The dose rate was 4.4 e^-^ per Å^2^ per second with a total exposure of 9 s resulting in a total dose of \~40 e^-^/Å^2^. The total dose was distributed over 40 frames in each movie. 3220 movies were collected.
Image processing {#s4-3}
----------------
The movie frames were corrected for beam induced sample motion and aligned using MotionCor2 ([@bib42]). The first 3 frames of the movies were skipped and the rest were aligned. These aligned frames were averaged with and without dose weighting. The non-dose weighted image stacks were used to estimate defocus and correct CTF using CTFFIND4 ([@bib33]). All estimated fits of defocus and CTF were visually inspected. All images with significant astigmatism or a prominent ring due to crystalline ice around 3--4 Å were discarded.
Whole particle reconstruction {#s4-4}
-----------------------------
A total of \~3200 images were used to auto pick 56275 particles using template matching in RELION ([@bib37]) (version 2.1 beta 1) ([@bib23]). The 2D classes generated by 2D classification of 710 manually picked particles were used as templates for auto-picking. The auto-picked particles were binned 2 × during the extraction (box size of 429 × 429 and 2.12 Å/pix). Extensive reference free 2D classification was performed to remove any particle images with ethane contaminants or broken/empty viral particles. The classes with good 2D averages were selected and the particles from these classes were extracted. This resulted in 51893 particles that were used for 3D classification. RELION: 3D initial model tool, which is based on stochastic gradient descent, was used to generate an ab-initio reference map for 3D classification. The 2 × binned particle images were used to generate a low-resolution icosahedrally averaged map. This low-resolution map was used as a reference for 3D classification. The 3D classification was performed with icosahedral symmetry (I4) applied. The most dominant class, with 46348 particles, was selected and the particles were extracted for final refinement. Icosahedral symmetry (I4) was also applied during the final refinement. The refinement with the 2 × binned particles reached Nyquist sampling (\~4.24 Å here). The particles from the final iteration step of the refinement were re-extracted without binning and further refined. The reference map was also scaled to match the new box and pixel size (858 × 858 and 1.06 Å/pix respectively) using e2proc3d.py from EMAN2.1 package ([@bib8]) ([Supplementary file 3](#supp3){ref-type="supplementary-material"}).
After refinement, the maps were corrected for Ewald sphere effects using RELION 3.0 beta 2. These maps were post-processed. A soft binary mask was generated using the 15 Å low pass filtered map extended by 10 pixels and with a soft edge of 15 pixels. An initial binarization threshold of 0.001 was used to include all the map features (i.e, internal membrane, etc) in the mask. This was used as a solvent mask during post processing. The map was corrected for the detector's Modulation Transfer Function (MTF) and sharpened with an inverse B-factor. Overall gold standard FSC\@0.143 was estimated using two independently processed half maps. ResMap ([@bib24]) was used to estimate the variation of resolution across the two unfiltered half maps.
Localized asymmetric reconstruction of the five-fold vertex {#s4-5}
-----------------------------------------------------------
The vertex complex was reconstructed using 2x binned particles by localized reconstruction ([@bib22]). Particle images from one half map were symmetry expanded (icosahedral to C1) and signal subtracted to remove the signal of the viral capsid and the genome from the particle images. Sub-particles representing each of the vertex region, with a box size of 100 × 100 pixels were extracted from the signal subtracted particles. This produced about 3,340,224 sub-particles. These particles were 3D classified without imposing symmetry and disabling image re-alignment. An initial reference map was produced by back projecting the extracted sub-particles using the previously calculated Euler angles. This initial reference map was low pass filtered to 45 Å and used as a reference map during 3D classification. The resolution of the expectation step was limited to 10 Å and a spherical mask of diameter 200 Å was added using the mask diameter and flatten solvent options provided by RELION to reduce the effects of systematic noise due to signal subtraction.
The classes generated from the 3D classification were analyzed and the particles from the class showing two protruding densities were selected and the 3D classification was now further refined with parameters similar to what was mentioned above but with the local image re-alignment enabled. The part of the map density that we were interested in was partially clipped due to the spherical mask that was used. So, the sub-particles were re-extracted with a larger box size of 140 × 140 pixels. These newly extracted particles with a larger box were back projected using the Euler angles calculated during the previous 3D classification.
To improve the resolution of the penton base region, another set of 3D classification was performed without limiting the e-step to 10 Å using the same set of sub-particles but applying a soft-mask around the penton.
Focused classification of the Penton base {#s4-6}
-----------------------------------------
To validate the penton map generated from the localized reconstruction, we also applied a different method, focused classification using the 2x binned particles. The particle representing the final converged map were symmetry expanded using *relion_particle_symmetry_expand* command. Similar to the localized reconstruction, signal subtraction was performed to remove the signal of the viral capsid and the genome from the particle images using *relion_project* command. We used the final converged map (without post-processing) from the icosahedrally averaged reconstruction to generate projections for signal subtraction. 3D classification with the signal subtracted particles was performed without imposing symmetry and disabling image re-alignment. The penton region extracted from the icosahedrally averaged map was low pass filtered to 45 Å and used as a reference map. A soft mask, around the penton region, was generated by extending the binary map of the penton region extracted from the icosahedrally averaged map by three pixels with a soft edge of 4 pixels. This was used as a focus mask to constrain the 3D classification to this region.
Model building and refinement {#s4-7}
-----------------------------
Modeling the whole particle was difficult owing to the large size of the particle and the limited RAM available. The whole viral map was sectioned using the sub-region selection option in Chimera. The sectioned maps were optimized by local sharpening using PHENIX: Autosharpen ([@bib39]). These sectioned maps were used to generate a crude model using PHENIX: find helix and loops ([@bib2]). The crude model was further used for de-novo modeling of the proteins in *Coot* ([@bib17]). These models were further refined using PHENIX: Real space refinement to improve the model. All the models were individually refined and put together to form the asymmetric model. The model of the asymmetric unit was further refined against a new map encasing the asymmetric unit using PHENIX-Real space refinement. The model of the asymmetric unit was validated with *MolProbity* ([@bib14]), Mtriage ([@bib3]) ([Supplementary file 4](#supp4){ref-type="supplementary-material"}) and EMRinger ([@bib7]).
Data availability {#s4-8}
-----------------
CryoEM Density maps and atomic models that support the findings of this study have been deposited in the Electron Microscopy Database and the Protein Databank with the accession codes EMD-4461 (Whole particle reconstruction), EMD-4462 (Vertex Complex), EMD-10237 (Localized reconstruction of the penton region), EMD-10238 (Focused Classification of the penton region) and PDB ID 6Q5U (Atomic model of the asymmetric unit).
Funding Information
===================
This paper was supported by the following grants:
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 828-2012-108 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 628-2008-1109 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 822-2010-6157 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 822-2012-5260 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004063Knut och Alice Wallenbergs Stiftelse KAW-2011.081 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100000781European Research Council ERC-291602 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 349-2011-6488 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100004359Vetenskapsrådet 2015-06107 to Janos Hajdu.
- http://dx.doi.org/10.13039/501100000780European Commission CZ.02.1.01/0.0/0.0/15_003/0000447 to Janos Hajdu.
We would like to thank Kazuyoshi Murata and Naoyuki Miyazaki from National Institute for Physiological Sciences, Japan; for providing us the sample cryoem data set to test the feasibility of the project. We would like to thank Sjors Scheres from MRC Laboratory of Molecular Biology, UK and Björn Forsberg from Stockholm University, Sweden for their valuable feedback in handling large CryoEM maps in RELION.
Additional information {#s5}
======================
No competing interests declared.
Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing---original draft, Writing---review and editing.
Resources, Data curation, Validation, Writing---review and editing.
Resources, Supervision, Funding acquisition, Project administration, Writing---review and editing.
Conceptualization, Resources, Supervision, Validation, Investigation, Project administration, Writing---review and editing.
Additional files {#s6}
================
10.7554/eLife.48496.030
###### Protein traces in the PR772 model and comparison of RMSD of different protein subunits from PR772 and PRD1, from a similar region of the map using Chimera and SuperPose.
10.7554/eLife.48496.031
###### Preliminary results for the whole particle asymmetric reconstruction using symmetry relaxation in EMAN2 of the wild type PR772.
10.7554/eLife.48496.032
###### Flowchart of the 3D reconstruction of the icosahadrally averaged PR772 map.
10.7554/eLife.48496.033
###### Mtriage summary of the map quality analysis.
10.7554/eLife.48496.034
Data availability {#s7}
=================
CryoEM Density maps and atomic models that support the findings of this study have been deposited in the Electron Microscopy Database and the Protein Databank with the accession codes EMD-4461 (Whole particle reconstruction), EMD-4462 (Vertex Complex), EMD-10237 (Localized reconstruction of the penton region), EMD-10238 (Focused Classification of the penton region) and PDB ID 6Q5U (Atomic model of the asymmetric unit).
The following datasets were generated:
HemanthKN ReddyMartaCarroniJanosHajduMartinSvenda2019Vertex Complex of Bacteriophage PR772Electron Microscopy Data BankEMD-4462
HemanthKN ReddyMartaCarroniJanosHajduMartinSvenda2019High resolution electron cryo-microscopy structure of the bacteriophage PR772Protein Data Bank6Q5U
HemanthKN ReddyMartaCarroniJanosHajduMartinSvenda2019High resolution electron cryo-microscopy structure of the bacteriophage PR772Electron Microscopy Data BankEMD-4461
HemanthKN ReddyMartaCarroniJanosHajduMartinSvenda2019Focused Classification of the vertex region of Bacteriophage PR772 showing the heteropentameric pentonElectron Microscopy Data BankEMD-10238
HemanthKN ReddyMartaCarroniJanosHajduMartinSvenda2019Localized Reconstruction of the vertex region of Bacteriophage PR772 showing the heteropentameric pentonElectron Microscopy Data BankEMD-10237
10.7554/eLife.48496.046
Decision letter
Scheres
Sjors HW
Reviewing Editor
MRC Laboratory of Molecular Biology
United Kingdom
Scheres
Sjors HW
Reviewer
MRC Laboratory of Molecular Biology
United Kingdom
San Martin
Carmen
Reviewer
Centro Nacional de Biotecnologia - CSIC
Spain
Hogue
Brenda
Reviewer
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article \"CryoEM of coliphage PR772 reveals the composition and structure of the elusive vertex complex and the capsid architecture.\" for consideration by *eLife*. Your article has been reviewed by three peer reviewers, including Sjors HW Scheres as the Reviewing Editor and Reviewer \#1, and the evaluation has been overseen by Olga Boudker as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Carmen San Martin (Reviewer \#2); Brenda Hogue (Reviewer \#3).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Summary:
In this manuscript, the authors present the cryoEM structure of PR772 -- a large, tailless, membrane containing bacteriophage. PR772 belongs to the same family as PRD1 (*Tectiviridae*), whose structure was solved at \~4 Å resolution by crystallography in 2004. Here the resolution is much higher (2.75 Å on average), allowing a more detailed analysis of the interactions in the capsid. The main findings are the heteropentameric organization of the penton capsomer, formed by the small single jelly roll protein P31 and its homolog P5 N-terminal domain (P5N), which is prolonged into a trimeric spike; a low resolution map of the penton organization together with its double spike, formed by the P5C trimer and the monomeric P2 protein; and a detailed description of the alternate conformations of N- and C-termini in the different copies of the major coat protein P3, as well as in the two penton proteins P31 and P5N
The resolution is impressive, particularly for such a large specimen, and addresses important issues in complex virus assembly, such as the use of flexible regions to achieve a range of quasi-equivalent interactions in the capsid, and the long standing puzzle on the composition and organization of the heterooligomeric vertex capsomer and its double spike. Therefore, the subject is interesting and the quality of the work is high. However, the manuscript is poorly written, and the most interesting results of a P31/P5 hetero-pentamer and a monomeric copy of P2 binding to P5 are based on extremely poor reconstructions, which are possibly caused by incorrect processing. Therefore, this manuscript would need considerable modification before being suitable for publication.
Essential revisions:
1\) The asymmetric reconstructions of the hetero-pentamer (Figure 4F) and the monomeric P2 are very low-resolution compared to the icosahedral reconstruction. This is probably the result of incorrect image processing. The icosahedrally-averaged pentamer has resolutions extending well beyond 3 Å. Just by lowering the number of asymmetric units by 5-fold, the drop in resolution of the asymmetric reconstructions is too large. By keeping the orientations the same as in the original icosahedrally refined structure (i.e. by doing focussed classifications on the symmetry-expanded particles without re-alignments), resolutions around 3 Å should still be achievable in the asymmetric reconstructions. A good control is to perform a reconstruction of the symmetry expanded (partial signal subtracted) images without any classification/realignment, which should give a map that is almost identical to the icosahedrally averaged reconstruction. The authors should fix this mistake in a revised version, as it would yield a much higher-resolution, and thus more interesting, view on the asymmetry in the hetero-pentamer. In addition, the authors should present detailed views of the improved density, including side-chains, for regions where the sequence between P5 and P31 differ, which would then replace the ambiguous tracing of both proteins in the icosahedrally averaged map at two different thresholds in the current manuscript.
To some extent, the same argument also holds for the P2 monomer reconstruction: doing purely focussed classifications (without re-alignments) of symmetry-expanded images should also improve this part of the map. The only caveat (and cause for lower resolution) in this case would be different orientations of the P2 monomer relative to the capsid. However, even this may be captured better by focussed classifications without alignment than by the currently used focussed refinements.
2\) The paper would benefit greatly from extensive editing of the text, while carefully checking for grammar and spelling. In addition, the manuscript organization needs to be improved:
2.1) Previous knowledge hinting at the possible penton heteropentamer/double spike composition and organization is cited, but a clear account of the subject should appear in the Introduction. For instance: \"many models have been proposed\"... why? The authors should explain that the basis for these previous models is the sequence similarity between proteins P31 and P5 (see for example, Merckel et al., 2005; Huiskonen et al., 2007; Abrescia et al., 2004).
2.2) Also lacking is the information on how similar PRD1 and PR772 are at the genome level (Introduction, second paragraph, repeated in the third paragraph: \"a close relative\", how close? Should we expect large differences?).
2.3) The two alternative conformations of the P5 N-terminal tail are described twice: first in the third paragraph of the subsection "Penton base is a heteropentamer of P5 and P31", then in the fourth paragraph of the subsection "Overall Architecture of PR772".
2.4) The description of the different conformations of the P3 termini in the subsection "Major Capsid Protein and its conformations" is difficult to follow. For example, \"the helix... is not embedded deeply into the adjacent subunit of the trimer...\": what is meant here? And \"the loop formed by Tyr351-Val358 is flipped...\": flipped how? Perhaps a superposition of the twelve P3 monomers in the asymmetric units, zooming on the N- and C-termini, would help to better illustrate the mobility of these regions. In general, adding labels to the figures would help the reader to follow the structure description. For example, for some of the residues cited in the text; or for at least one occurrence of each of the P3 N- and C-terminal conformations in Figure 2.
2.5) Including the length of proteins and of their traced part in the Results section would also help the reader (without having to resort to the PDB validation report).
2.6) The start of the Discussion section reads like a repetition of (parts of) the Results section.
2.7) Some sentences seem to be repeated, or at least very similar, e.g. in the paragraph just before Figure 3 and just after Figure 3. Other pieces of text read more like an early draft than like a polished manuscript, e.g. the second paragraph of the subsection "Overall Architecture of PR772".
10.7554/eLife.48496.047
Author response
> Essential revisions:
>
> 1\) The asymmetric reconstructions of the hetero-pentamer (Figure 4F) and the monomeric P2 are very low-resolution compared to the icosahedral reconstruction. This is probably the result of incorrect image processing. The icosahedrally-averaged pentamer has resolutions extending well beyond 3 Å. Just by lowering the number of asymmetric units by 5-fold, the drop in resolution of the asymmetric reconstructions is too large. By keeping the orientations the same as in the original icosahedrally refined structure (i.e. by doing focussed classifications on the symmetry-expanded particles without re-alignments), resolutions around 3 Å should still be achievable in the asymmetric reconstructions. A good control is to perform a reconstruction of the symmetry expanded (partial signal subtracted) images without any classification/realignment, which should give a map that is almost identical to the icosahedrally averaged reconstruction. The authors should fix this mistake in a revised version, as it would yield a much higher-resolution, and thus more interesting, view on the asymmetry in the hetero-pentamer. In addition, the authors should present detailed views of the improved density, including side-chains, for regions where the sequence between P5 and P31 differ, which would then replace the ambiguous tracing of both proteins in the icosahedrally averaged map at two different thresholds in the current manuscript.
>
> To some extent, the same argument also holds for the P2 monomer reconstruction: doing purely focussed classifications (without re-alignments) of symmetry-expanded images should also improve this part of the map. The only caveat (and cause for lower resolution) in this case would be different orientations of the P2 monomer relative to the capsid. However, even this may be captured better by focussed classifications without alignment than by the currently used focussed refinements.
The lower resolution map from the localized asymmetric reconstruction was due to the following reasons: we used 2x binned particles for the localized asymmetric reconstruction. The 2x binned particles have an apix of 2.12 Å, resulting in a Nyquist frequency of 0.2358 (i.e., a resolution of 4.24 Å). So, the reconstructions with these 2x binned particles is limited to a resolution of 4.24 Å. We had also limited the e-step to 10 Å during the classification as recommended by the authors of the localized reconstruction method. We avoided using a mask to remove any bias that could be introduced due to the applied masked. Due to the limitations of storage, we were also limited to using sub-particles from one half map.
We have currently improved the resolution of the penton to 4.41 Å using a soft mask around the penton region and not limiting the e-step during the localized asymmetric reconstruction.
Using the focused classification method as suggested, we were able to reproduce the penton map to a resolution of 4.25 Å, a resolution similar to what was obtained by the localized reconstruction method with 2x binned particles. From the localized reconstruction, we knew that all the classes showed the heteropentameric penton. This suggested that the heteropentameric penton was a common feature. So, we chose a random vertex to perform a C1 focused classification on, without re-alignment. The focused classification of the penton region as well as the localized reconstruction method, show a stalk-like protrusion from 3 of the 5 subunits that form the penton. We have added a new Figure 4---figure supplement 6, showing the maps of the penton region from both methods.
The unmasked localized reconstruction using the sub-particles of the vertex region was performed to determine the orientation of P2 with respect to the viral capsid and avoid any bias by masking. From the localized reconstruction, we now know that the P2 protein is on average present only on 2 vertices of the viral particle. We can also notice that all the vertices display the presence of P5 that would also contribute to map densities (i.e., the highly flexible C-terminal regions with the collagen-like triple helix and the knob domain that are not resolved in the icosahedrally averaged map used to generate the projection for signal subtraction) during the focused classification. These issues along with the caveat mentioned by the reviewers will significantly reduce the resolution. By applying a focus mask around the P2 protein to address some of the issues, we would essentially bias the very orientation that we plan to determine. Hence, we have avoided using whole particles for focused classification of P2. Instead, we provide the P2 map from the masked classification without re-alignment of sub-particles whose orientation was previously determined without using a mask from localized reconstruction. This ensures that the features observed are not due to masking. We have provided two images, one with the map generated by re-aligned sub-particles (previously submitted) and another with a masked map without realignment (submitted here as part of the supporting files). We believe that the Figure 5, currently in the manuscript, illustrates the structure of the vertex complex, best.
We do agree that having a resolution beyond 3 Å for an asymmetric reconstruction would be better to access the heteropentameric nature of the penton. The resolution of the maps from the localized reconstruction and focused classification should be sufficient to assess the secondary structure and domain fit for the proteins P31, P5 N-terminal domain and P2.
> 2\) The paper would benefit greatly from extensive editing of the text, while carefully checking for grammar and spelling. In addition, the manuscript organization needs to be improved:
>
> 2.1) Previous knowledge hinting at the possible penton heteropentamer/double spike composition and organization is cited, but a clear account of the subject should appear in the Introduction. For instance: \"many models have been proposed\"... why? The authors should explain that the basis for these previous models is the sequence similarity between proteins P31 and P5 (see for example, Merckel et al., 2005; Huiskonen et al., 2007; Abrescia et al., 2004).
Most of these models were speculated on based on the observations from gene mutation/knock-out and in vitro studies of proteins P31, P5 and P2 from PRD1. The in vitro protein expression studies showed that P31 forms pentamers and the P5 forms multimers of trimers (i.e., (P5~3)1~, (P5~3)2~, (P5~3)3~) (Sokolova et al., 2001). The gene knock-out studies showed that P31^-^ mutants produced incomplete PRD1 particles that lacked P31, P5 and P2 functions. They failed to form the vertex complexes. A P5^-^ mutant produced intact viral particles but lacked both the P5 and P2 functions (Rydman et al., 1999). These observations led to the hypothesis that the vertex complex appears to be a single spike formed by a pentameric P31 base binds the trimeric P5 spike protein to which P2 is bound (Huiskonen et al., 2007; Caldentey et al., 1999; Sokolova et al., 2001). Later, a combination of SAXS modelling of the P5 protein and a low resolution cryoem study of the vertex region, showed that P2 and P5 form two spikes, not one, as previously described. Based on the low-resolution SAXS model of P5, it was speculated that the N-terminal base of P5 could have a similar size and fold to P31 due to sequence similarity (Huiskonen et al., 2007).
We have added the above explanation to the manuscript to clarify why many models were proposed to explain the architecture of the vertex complex.
> 2.2) Also lacking is the information on how similar PRD1 and PR772 are at the genome level (Introduction, second paragraph, repeated in the third paragraph: \"a close relative\", how close? Should we expect large differences?).
We have now included the DNA sequence identity between PRD1 and PR772. We have also included the reference (Saren et al., 2005), which describes the genome level similarity between different phages in the *Tectiviridae* family.
"Much of the functional knowledge of the viral proteins is inferred from previous studies on PRD1, a close relative of PR772 with a genome sequence identity of 97.2% (Lute et al., 2004; Saren et al., 2005)."
> 2.3) The two alternative conformations of the P5 N-terminal tail are described twice: first in the third paragraph of the subsection "Penton base is a heteropentamer of P5 and P31", then in the fourth paragraph of the subsection "Overall Architecture of PR772".
In the subsection "Penton base is a heteropentamer of P5 and P31", we are establishing that it is the P5 N-terminal that is responsible for the special conformation. in the subsection "Overall Architecture of PR772", we are comparing the N-terminal regions of P5 with P31 and their influence on the overall architecture. We only mention the special conformation of the N-terminal P5 region in one sentence to remind the reader that it exists. This is relevant for the hypothesis we put forward in the discussion regarding the potential infection process.
> 2.4) The description of the different conformations of the P3 termini in the subsection "Major Capsid Protein and its conformations" is difficult to follow. For example, \"the helix... is not embedded deeply into the adjacent subunit of the trimer...\": what is meant here? And \"the loop formed by Tyr351-Val358 is flipped...\": flipped how? Perhaps a superposition of the twelve P3 monomers in the asymmetric units, zooming on the N- and C-termini, would help to better illustrate the mobility of these regions. In general, adding labels to the figures would help the reader to follow the structure description. For example, for some of the residues cited in the text; or for at least one occurrence of each of the P3 N- and C-terminal conformations in Figure 2.
We agree that only text makes it a bit hard to follow the reasoning. As suggested, we have added new figures and have tried to refer to figures everywhere possible to make it easier to understand what is being described in the text. We have added Figure 2---figure supplement 3, that shows the superposition of all the P3 monomers of the asymmetric unit. This figure also includes an image showing the flipped Tyr351-Val358 loop.
> 2.5) Including the length of proteins and of their traced part in the Results section would also help the reader (without having to resort to the PDB validation report)
We have added the details about the proteins, chain length, length of the chain traced and region, where we compare the current model of PR772 with model of PRD1.
> 2.6) The start of the Discussion section reads like a repetition of (parts of) the Results section.
We have rephrased the start of the Discussion to only include differences compared to the PRD1 structure.
> 2.7) Some sentences seem to be repeated, or at least very similar, e.g. in the paragraph just before Figure 3 and just after Figure 3. Other pieces of text read more like an early draft than like a polished manuscript, e.g. the second paragraph of the subsection "Overall Architecture of PR772".
In these two instances (in the paragraphs before and after Figure 3), we are describing different things.
In the paragraph before Figure 3, we describe that the N-terminal of P3 can adopt three different conformations not two as previously described for PRD1. The new conformation as compared to PRD1 can only be observed due to the higher resolution model/reconstruction and has relevance for the detailed molecular interplay between proteins building up the capsid.
In the paragraph after Figure 3 and onwards, we describe how the N and C-terminals of the "same" P3 subunit that form four different trimers have to adopt different conformations depending on where the trimer resides in the trisymmetron and what contacts with other proteins they make at that specific location.
We have polished the text. It now reads:
"Five copies of P16 bridge the penton to the five trisymmetrons that meet at the vertex complex. \[...\] As was mentioned earlier, to accommodate the P30 protein in the P3-P30-P16 complex (Figure 6 and Figure 6---video 1), the loop formed by residues Tyr351-Val358 of the P3 subunit in trimer 1 is flipped compared to the orientation seen in this loop of the P3 subunit with a N-terminal long helix (Figure 2---figure supplement 3D-F)."
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
The human placenta has both maternal and fetal blood circulations, which exchange gases, nutrients and waste through the surface of the villous tree without mixture \[[@R1]\]: the maternal blood from the uterine arteries flows slowly into the intervillous space and returns to the uterine veins while the fetal blood from the umbilical arteries flows into arteries within the chorionic plate, circulates through blood vessels in the villous tree, and returns to the umbilical veins. The influence of the maternal and fetal blood flows on the transport of substances has been reported previously \[[@R2], [@R3]\].
The villous tree is composed of stem villi, intermediate villi, terminal villi and mesenchymal villi. Stem villi are the main support of the villous tree between the two bounding surfaces of the placenta. The stem villi have the contractile cells, which surround the arterial and venous vessels and run along the longitudinal axis of the branch \[[@R4]-[@R11]\]. The contraction of the stem villi has been observed *in vitro* \[[@R5], [@R12]-[@R14]\], which is expected to help the circulations of the maternal and fetal bloods. The maximum velocity of the contraction was much smaller than that of uterus, and the peak isometric tension was 1.39 kPa for electrical tetanus and 1.32 kPa for KCl exposure on average \[[@R14]\]. The contraction directly assists the fetal blood flow in the vessels of the stem villi because the contractile cells surround the vessels axially. However, the fetal blood in the capillary of the villous tree and the maternal blood in the intervillous space are not surrounded by the contractile cells directly, and similarity in the directions between the blood flow and the contractile cells has not been clear. In the meantime, the mechanical properties of the human placenta were evaluated by tensile, compression and shear \[[@R15]\]: The elastic moduli measured by shear were much smaller than those by tensile and compression. The shear stress was less than 1 kPa when the strain was 1 (strain velocity \< 0.04 /s). Considering that the blood vessels were aligned with the direction of the force \[[@R15]\], the shear elastic moduli in the surroundings of the blood vessels would be much less than 1 kPa. Comparing the tension of the contractile cell with the shear elastic moduli, the displacement and its propagation in the placenta would occur although neither direct measurement nor prediction for the displacement is possible.
The conditions of the placenta, concerning the blood flow, are non-invasively observed by ultrasound and magnetic resonance (MR). The velocity of the blood flow in the umbilical artery is measured by ultrasound Doppler velocimetry and described as flow velocity waveforms (FVWs). The histological characteristics in the villous tree, including the distribution of the villous types and vascularization, can be estimated by FVW at end diastolic: positive, absent or reverse \[[@R16]-[@R18]\]. The oxygen environment, which is evaluated by oxygen-enhanced MRI and BOLD MRI \[[@R19]\], influenced the bifurcation pattern of the villous tree: hypoxia enhances the bifurcation \[[@R20], [@R21]\]. The magnitude of the perfusion in the placenta can be expressed as relative values by three-dimensional (3D) power Doppler \[[@R22]\] and contrast-enhanced MR images \[[@R23]\], but the direction is hardly measured.
In general, the normal mature placenta shows legions in some degree, such as infarctions, which can prevent the blood circulations in the placenta. 3D power Doppler indicated that the perfusion in the normal placenta was kept through the gestational ages, from 15 to 40 weeks \[[@R24]\]. In the meantime, the diameter, branching patterns and generation of the stem villi were previously reported \[[@R1], [@R25]\] so that a computational model of the villous tree with active contractions can be developed. If the displacement caused by the contraction is corresponding to the perfusion evaluated by 3D power Doppler or MRI, the direction of the blood flow can be estimated by changing the distribution of the shear elastic moduli as the pattern of the displacement in the placenta agrees with that of the perfusion. Assuming that the contraction force and shear elastic moduli of the region surrounding the stem villi are representative of the placental conditions, the results of the computation can be translated properly.
In this study, a computational model of the villous tree in the human placenta with active contractions was developed for estimating the blood flow condition. The shape of the stem villi was based on the previous reports \[[@R1], [@R25]\], and the surroundings of the stem villi were assumed as one continuum, where the displacement caused by the contraction propagated because the branching pattern of the villi around the stem villi, and the shape of the intervillous space, which is the space surrounded by the villi and the maternal blood passes through, are complicated. By using this model, it was examined whether or not the contraction could assist the fetal blood flow in the capillary and the maternal blood flow in the intervillous space. Moreover, the influence of the mechanical properties in the villous tree on the pattern of the displacement was also examined.
COMPUTATIONAL MODEL
===================
Stem Villi
----------
The stem villi are categorized into the three groups: truncus chorii, next to the chorionic plate; rami chorii, next to the truncus chorii; ramuli chorii, between the rami chorii and the basal plate \[[@R1], [@R25]\]. The diameter gradually becomes smaller from the chorionic plate to the basal plate, and branches are found in all the groups except the truncus chorii. The branching pattern of the ramuli chorii is not equally dichotomous. The generations of branches in the rami chorii and ramuli chorii are up to 4, and from 1 to 30, respectively. Table **[1](#T1){ref-type="table"}** shows the size and branching pattern in the stem villi model. In this model, the chorionic and basal plates, and the boundaries between the categories (truncus chorii, rami chorii and ramuli chorii) were parallel to each other. Firstly, how to use the centripetal and centrifugal orders at the branch to describe the branching pattern in Table **[1](#T1){ref-type="table"}** is explained. Figs. (**[1a](#F1){ref-type="fig"}** and **[1b](#F1){ref-type="fig"}**) show a simple branching pattern designated by the centripetal order and centrifugal order: the tip and trunk are designated as 1 and maximum, respectively, at the centripetal order, and vice versa at the centrifugal order. The centripetal order is used to evaluate dichotomy \[[@R26], [@R27]\]. The bifurcation ratio, *R~b~*, is defined as the following equation:
R
b
=
N
u
N
u
\+
1
where *N~u~* is the number of branches at the centripetal order *u*. If *R~b~* is 2, the branching is equally dichotomous. Assuming that *R~b~* is constant, *N~u~* is a geometrical series given as:
N
u
=
R
b
(
u
max
−
u
)
where *u~max~* is the maximum centripetal order. Hence,
ln
N
u
=
(
u
max
−
u
)
ln
R
b
*R~b~* of the branches was calculated by the method of least squares. Table **[1](#T1){ref-type="table"}** shows that *R~b~* in the rami chorii was 2, but that in the ramuli chorii was not equal to 2. These values indicate that the branching patterns in the rami chorii and ramuli chorii are equally and unequally dichotomous, respectively. The centrifugal order at the tip, *C~f~*, in Table (**[1](#T1){ref-type="table"}**), corresponds to the generation of the branches. While *C~f~* in the rami chorii was constant, that in the ramuli chorii was varied. That is because the branches in the ramuli chorii were not symmetric.
The diameter range in each category was as follows: truncus chorii, 900 -- 3000 μm; rami chorii, 300 -- 1000 μm; ramuli chorii, 50 -- 500 μm \[[@R25]\]. Table **[1](#T1){ref-type="table"}** shows that the diameter ranges agree with the aforementioned one. The change of the diameter at the truncus chorii was much larger than those at the rami chorii and ramuli chorii. For the smooth connection between the truncus chorii and rami chorii, the derivative of the radius with respect to the *z* coordinate should be zero at the boundary. Hence, the following equation was used to determine the radius of the branch at the truncus chorii:
r
=
r
max
−
(
r
max
−
r
min
)
z
tr
{
z
tr
2
−
(
z
−
z
tr
)
2
}
where *r~max~* and *r~min~* are the maximum and minimum radii in the truncus chorii, and *z~tr~* is the boundary between the truncus chorii and rami chorii (*z~tr~* = 2.9 mm). In the rami chorii, the radius was decreased as the distance along the axis was longer. The radius of the branch in the ramuli chorii became larger as the centripetal order increased. The radii of the branches at the connecting point were modulated as all the branches showed the same radius.
The branches in the rami chorii were equally dichotomous as well as symmetric, and 16 branches were connected to those in the ramuli chorii at the boundary. The branches in the ramuli chorii should be unequally dichotomous. For making unequally dichotomous branches, the 2D diffusion-limited aggregation (DLA) models \[[@R28]\] were made by the free-software, dla-nd, which was developed by the Dr. Mark J. Stock (<http://markjstock.org/dla-nd/>). Because the path between the branching points was not smooth in this model, the line between the branching points was set as an axis of a branch.
As Table (**[1](#T1){ref-type="table"}**) shows, the longest distance from the chorionic plate in the model was 24.5 mm. According to the previous reports, the thickness of the human placenta and chorioamniotic membrane, the surface of the placenta, were 25 mm \[[@R1]\] and 243 μm \[[@R29]\] on average, respectively. It is calculated that the distance between the chorionic and basal plates is 24.8 mm. The longest distance from the chorionic plate in the model was close to the calculated distance between the chorionic and basal plates, based on the previous reports \[[@R1], [@R29]\]. In the meantime, the cross section of the model, parallel to the chorionic plate, had the bounding rectangle, whose size was 23.8 mm × 22.6 mm (width × height). The previous report \[[@R1]\] indicated that the placenta at term, whose diameter was 220 mm, had 60 -- 70 villous stems. The area based on this diameter is 3.80 × 10^4^ mm^2^ so that the average cross section of the villous trees is 5.42 × 10^2^ - 6.33 × 10^2^ mm^2^, whose corresponding diameter is 26 -- 28 mm. The size of the cross section in the model was close to that based on the previous report \[[@R1]\]. Figs. (**[1c](#F1){ref-type="fig"}**-**[1e](#F1){ref-type="fig"}**) show the villous tree model developed in this research. The size was 34.8 × 34.8 × 24.5 \[mm\] (1200 × 1200 × 847 \[pixels\], 29 μm/pixel).The Cartesian coordinate system, whose *z* axis was perpendicular to the chorionic plate, was used to describe the position in the model. Its origin was also on the chorionic plate.
Contraction Direction
---------------------
The contractile cells run along the longitudinal axis of the branch \[[@R4]-[@R11]\]. Each point at the boundary surface between the stem villi and the surroundings has two tangential directions as Fig. (**[2](#F2){ref-type="fig"}**) shows. The tangential direction, closer to the axis of the branch than the other, was decided as the contraction direction.
At the truncus chorii, the axis of the branch was parallel to the *z* axis and its radius was largely changed as Equation (4) shows. The angle between the *z* axis and tangential direction (*φ~o~*) and the differentiation of Equation (4) by *z* are as follows:
φ
o
=
atan
(
dr
dz
)
dr
dz
=
r
max
−
r
min
z
tr
z
−
z
tr
z
tr
2
−
(
z
−
z
tr
)
2
Considering that the direction of the contraction at the axis of the branch was (0, 0, -1),
φ
o
=
π
\+
atan
(
dr
dz
)
When the angle between the point at the surface and the *x* axis in the *xy* plane is *θ~o~*, the tangential direction is (*sinφ~o~cosθ~o~*, *sinφ~o~sinθ~o~*, *cosφ~o~*). Comparing the rami chorii with the truncus chorii at Table (**[1](#T1){ref-type="table"}**), the diameter change at the rami chorii was 25% of that at the truncus chorii. In addition, the rami chorii showed the range of the *z* coordinate, which was about 3.7 times larger than that at the truncus chorii. The change of the diameter at the rami chorii was much smaller than that at the truncus chorii. Hence, the tangential direction at the surface was parallel to the axis of the branch at the rami chorii. Because the change of the diameter was also small at the ramuli chorii, the tangential direction was determined in the same way.
Displacement
------------
As Figs. (**[1c](#F1){ref-type="fig"}**-**[1e](#F1){ref-type="fig"}**) show, assuming that the surroundings of the stem villi were one continuum in this model, the propagation of the displacement in the placenta was evaluated by the model. A wave equation is generally described as below:
ρ
∂
2
u
∂
t
2
=
μ
∇
2
u
\+
(
λ
l
\+
μ
)
∇
(
∇
⋅
u
)
where ***u*** is displacement vector, *ρ* is density, *λ~l~* and *μ* are Lamé's constant. *μ* also shows a shear elastic modulus. Generally, biological tissue is incompressible, so that the second term in Equation (8) is zero. Hence,
ρ
∂
2
u
∂
t
2
=
μ
∇
2
u
Hence, the shear wave (transverse wave), whose propagation is normal to the vibration and carried out in solid, was evaluated in this computation. The displacement caused by the shear wave is described as follows \[[@R31]\]:
u
=
ξ
o
cos
(
kr
−
𝜛t
)
where *ξ~o~* is the amplitude, *k* is the wave number (*k* = 2π/*λ*, *λ* is wave length), *r* is the distance from the surface of the stem villi, *t* is time, and *ω* is angular frequency. *ξ~o~* was 0.1 μm in all the computational conditions. The shear elastic modulus, *μ*, is described as follows \[[@R32]\]:
μ
=
𝜛
2
k
2
ρ
Considering that *k* = 2π/*λ* and *ω*=2π*ν* (*ν*, frequency),
μ
=
ρ
λ
2
v
2
*ρ* was 1.0 × 10^3^ kg/m^3^ because of biological tissues \[[@R32]\]. ν was 1.0 Hz, and *λ* was 0.29, 0.58 or 1.45 mm (10, 20 or 50 pixels) in the computation so that the shear moduli were 8.41 × 10^-5^, 3.36 × 10^-4^ and 2.10 × 10^-3^ Pa. The displacement is attenuated by viscoelastic properties so that the maximum distance for the propagation was 1.45, 2.9 and 4.35 mm (50, 100 and 150 pixels). In order to simplify the problem, *t* was set for zero. That is, the computation did not consider the time effect on the displacement. The maximum distance for the propagation was not dependent on the position. *λ* was kept constant in the surroundings, or became longer as the distance from the surface of the stem villi was longer. In the latter case, *λ* was increased from 0.29 mm to 1.45 mm every one-third of the maximum distance from the surface. Hence, there were 12 conditions in this computation.
RESULTS
=======
Characteristic Positions and Visualization
------------------------------------------
The displacement of the surroundings of the stem villi is described by the polar coordinate system (magnitude, *φ* (0° ≤ φ ≤ 180°) and *θ* (-180° ≤ θ \< 180°)) because the coordinate system was useful to separate the displacement into its magnitude and direction. Fig. (**[3](#F3){ref-type="fig"}**) shows the results when *λ* and the maximum distance for the propagation were 1.45 mm and 4.35 mm, respectively. Fig. (**[3a](#F3){ref-type="fig"}**) shows that the displaced area was gradually increased as the *z* coordinate became larger. The long and steep slope was observed from the truncus chorii to the rami chorii. The same features were observed in all the computational conditions. Figs. (**[3b](#F3){ref-type="fig"}**, **[3c](#F3){ref-type="fig"}** and **[3d](#F3){ref-type="fig"}**) show the mean and standard deviation (SD) of each parameter: magnitude, *φ* and *θ*. The mean and SD were calculated for all the points in the displaced area, whose magnitude was more than zero. When the displacement is perpendicular to the chorionic plate, the value of *θ* cannot be determined. Hence, such a critical point was not included for the calculation of the mean and SD in *θ*. Fig. (**[3b](#F3){ref-type="fig"}**) shows that the SD normalized by the mean was calculated in order to evaluate the magnitude range of the displacement in each *z* coordinate. The peak of the normalized SD was observed around the boundary between the truncus chorii and rami chorii, which is named as *z~d~*. The mean and SD about the direction of the displacement in each *z* coordinate are shown in Figs. (**[4c](#F4){ref-type="fig"}** and **[4d](#F4){ref-type="fig"}**). As Fig. (**[4c](#F4){ref-type="fig"}**) shows, the mean of *φ* was kept around 90 degrees, but the SD of *φ* indicated two peaks, whose positions are named as *z~φ1~* and *z~φ2~*, respectively. Fig. (**[3d](#F3){ref-type="fig"}**) shows that the mean of *θ* slightly decreased around the boundary between the rami chorii and ramuli chorii, which is named as *z~θ~*, while the SD of *θ* was kept around 90°. These characteristic *z* coordinates, *z~d~*, *z~φ1~*, *z~φ2~*, and *z~θ~*, were observed at all the computations. Fig. (**[4](#F4){ref-type="fig"}**) shows the images which visualizes the magnitude, *φ*, and *θ* in each characteristic *z* coordinate and the middle *z* coordinates of the truncus chorii (*z~t~*), rami chorii (*z~r~*), and ramuli chorii (*z~rl~*) under the same computational condition as Fig. (**[3](#F3){ref-type="fig"}**) shows. The displaced area became larger as the *z* coordinate was increased. The magnitude of the displacement was almost kept constant at every *z* coordinate although the magnitude in the limited area at *z~d~* was high. The distributions of *φ* and *θ* were largely changed along the *z* coordinate as shown in Figs. (**[3c](#F3){ref-type="fig"}** and **[3d](#F3){ref-type="fig"}**) The visualization of the displacement and direction like Fig. (**[4](#F4){ref-type="fig"}**) is useful to find out critical points and important points for analysis.
Magnitude of the Displacement
-----------------------------
As Fig. (**[4](#F4){ref-type="fig"}**) shows, the magnitude of the displacement was kept almost constant except the magnitude was high near the stem villi at *z~d~*. 90% of the displaced area showed that the magnitude relative to the maximum one was less than 0.17 for the maximum distance for the propagation = 1.45 mm and λ increasing as the distance from the surface of the stem villi, and 0.06 for other results. The SD normalized by the mean was large at *z~d~*, but the high magnitude was in the limited area. Hence, the maternal and fetal blood circulations in the villous tree and intervillous space would be influenced by the constant distribution of the displacement in the placenta. Because the shape of the stem villi directly influences the displacement pattern, the model whose shape near the high magnitude is changed will be developed and used for the computation to examine whether or not such a high magnitude is inevitable.
Direction of the Displacement (*φ* and *θ*)
-------------------------------------------
Figs. (**[5a](#F5){ref-type="fig"}**-**[5e](#F5){ref-type="fig"}**) show the distribution of *φ* at the characteristic *z* coordinates (*z~φ1~*, *z~φ2~*) and the middle *z* coordinate in each category (*z~t~*, *z~r~*, *z~rl~*) for the same condition as shown in Fig. (**[3](#F3){ref-type="fig"}**). The area fraction, which is the area for each *φ* normalized by all the displaced area, was used to evaluate the distribution of *φ*. The area fraction around *φ* = 90° (*φ* = 45° - 135°) was largest at *z~φ1~* (Fig. **[5b](#F5){ref-type="fig"}**) and smallest at *z~φ2~* (Fig. **[5d](#F5){ref-type="fig"}**). The same characteristics were observed in all the computations. For all the computational conditions, the mean and SD of the area fraction at *φ* = 45° - 135° were calculated at each *z* coordinate. As the average values in Fig. (**[5f](#F5){ref-type="fig"}**) show, more than 90% of the displaced area showed *φ* from 45° to 135° at *z~φ1~*. The area with the same range of *φ* was around 10% at *z~φ2~*. Considering that the *φ* = 90° means the direction parallel to the *xy* plane, most of the displacement was parallel to the *xy* plane at *z~φ1~*, and parallel to the *z* axis at *z~φ2~*. Fig. (**[5f](#F5){ref-type="fig"}**) shows that the mean of the area fraction at the other positions was around 0.3. Hence, *φ* did not have a preferred direction there. The displacement could help the maternal blood to go to the chorionic plate at *z~φ2~*, spread parallel to the chorionic plate and the basal plate at *z~φ1~*, and go toward the basal plate at *z~φ2~*. In the meantime, the branches in the stem villi around *z~φ1~* and *z~φ2~* had the axes largely different from the directions of the displacement there: the axis of the branch in the truncus chorii is perpendicular to the chorionic plate, and that in the rami chorii was largely changed because of its curvature. The displacements at *z~φ1~* and *z~φ2~* could assist the fetal blood to pass through the vessels in the stem villi. Moreover, the maternal and fetal bloods could be homogenized by the displacement at *z~t~*, *z~r~* and *z~rl~*. The SD was largest at *z~t~*, and smallest at *z~φ1~*. The SD at *z~rl~* was larger than that at *z~φ1~*, but much smaller than those at the other characteristic positions. Considering that *λ* and the maximum distance for the propagation are corresponding to the mechanical properties of the surroundings of the stem villi, *φ* at *z~φ1~* and *z~rl~* would be hardly influenced by the mechanical property of the surroundings, but *φ* at the truncus chorii would be vulnerable to it.
Figs. (**[6a](#F6){ref-type="fig"}**-**[6d](#F6){ref-type="fig"}**) show the distribution of *θ* at the characteristic *z* coordinates (*z~θ~*) and the middle *z* coordinate in each category (*z~t~*, *z~r~*, *z~rl~*) for the same condition as Fig. (**[3](#F3){ref-type="fig"}**) shows. The area fraction, the area for each *θ* normalized by all the displaced area, was used for describing the distribution. Figs. (**[6a](#F6){ref-type="fig"}** and **[6b](#F6){ref-type="fig"}**) show that the similar distribution pattern was observed every 90° at *z~t~* and *z~r~*. The result agreed with the mean of *θ* at *z~t~* and *z~r~*, around zero. Fig. (**[6c](#F6){ref-type="fig"}**) shows the distribution of *θ* at *z~θ~*, where the area fraction at *θ* = -180° - 90° (the third quadrant) and *θ* = 0° - 90° (the first quadrant) was larger than that at *θ* = -90° - 0° (the fourth quadrant) and *θ* = 90° - 180° (the second quadrant). This result agreed with the decrease of the mean at *z~θ~* in Fig. (**[3d](#F3){ref-type="fig"}**). As Table (**[1](#T1){ref-type="table"}**) shows, the branches in the ramuli chorii were unequally dichotomous as well as asymmetric while those in the rami chorii were equally dichotomous as well as symmetric. Because the rami chorii had 16 branches connecting to those in the ramuli chorii, 16 types of the branching pattern were used in the ramuli chorii. *z~θ~* was located around the boundary between the rami chorii and ramuli chorii. The branches of the ramuli chorii at *z~θ~* were almost parallel to the *z* axis in the second and fourth quadrant, but those in the first and third quadrants were not. The angle of the branch at *z~θ~* would cause the characteristic distribution of *θ* at *z~θ~*. The similar distribution pattern was observed at *z~rl~*, but each peak was much smaller than that at *z~θ~* as Fig. (**[6d](#F6){ref-type="fig"}**) shows. Because *z~rl~* was placed on the middle of the ramuli chorii, the feature caused at *z~θ~* would be weakened by the branches which showed various angles. All the computational results showed the same characteristics as Figs. (**[6a](#F6){ref-type="fig"}**-**[6d](#F6){ref-type="fig"}**) show. The SD of the area fraction in each interval (each value of *θ*) (*SD~in~*) was calculated in order to evaluate the uniformity of the distribution in *θ* for all the computations. Fig. (**[6e](#F6){ref-type="fig"}**) shows that the average value of *SD~in~* at *z~θ~* was largest and that at *z~rl~* was smallest, among all the positions. The distribution of *θ* was most uniform at *z~rl~*, and most fluctuated at *z~θ~* in all the positions. The distribution of *θ* at *z~θ~* would be influenced by the angle of the branches in the ramuli chorii. Because *θ* at *z~rl~* did not have a preferred value, the maternal and fetal bloods could be homogenized. The SD values of *SD~in~* at *z~t~* and *z~θ~* was much larger than those of other two positions, and that at *z~rl~* was smallest in all the positions. The results show that the mechanical properties of the villus tree would strongly influence *θ* at *z~t~* and *z~θ~*, but hardly influence that at *z~rl~*. Even if the mechanical properties of the villous tree are changed, the homogenization of the bloods around the ramuli chorii would be kept.
DISCUSSION
==========
In this study, the computational villous tree model was developed, and used to evaluate the displacement in the human placenta caused by the contraction of the stem villi. The magnitude of the displacement was almost homogeneous, and the direction was useful for the fetal and maternal blood circulations. This tendency was maintained even if the mechanical properties of the placenta were changed. The experimental results such as MRI and 3D power Doppler angiography are described by scalar values so that the magnitude of the displacement could be directly compared with them. The resolution in the computation of this model was much higher than that in MRI or 3D power Doppler. Hence, representative values such as mean and SD will be necessary if the comparison between the experimental data and the computational result is carried out.
In the computation, every point in the stem villi contracted at the same time. Hence, the result in this study indicated that the displacement caused by the contraction would be helpful for the blood circulations when each contractile cell in the stem villi contracts at the same time. How to control the timing of the contraction for the effective blood circulation in the placenta can be investigated by this villous tree model. This investigation will be carried out in the future. The effect of the displacement on the maternal and fetal blood flows would be influenced by the directions of these blood flows. A computational model and method are necessary to investigate this effect. Developing them, with considering the usage of FWVs, will be the important topic to evaluate the blood circulation in the placenta.
CONCLUSION
==========
In this study, the computational model of the villous tree with active contractions was developed. The results based on this model show that the contraction could assist the maternal and fetal blood circulations in the placenta, and its effect would maintain even if the mechanical circumstances are changed. The combination between this computational model and non-invasive measurements will be useful to evaluate the condition of the placenta.
Declared none.
CONFLICT OF INTEREST
====================
The author confirms that this article content has no conflict of interest.
![Villous tree model developed in this study. The branching pattern in the stem villi of the model was indicated by the centripetal and centrifugal orders at the branches, whose example is indicated by **(a)** and **(b)**, respectively. **(c-e)** show the side views of the model: the stem villi, white; the villi except the stem villi, blue; the intervillous space, which the maternal blood passes through, blue. As **(d)** and **(e)** show, this model is between the chorionic and basal plates. The *z* axis of the Cartesian coordinate system is perpendicular to the chorionic plate. The size of this model is 34.8 × 34.8 × 24.5 \[mm\] (1200 × 1200 × 847 \[pixels\], 29 μm/pixel).](TOBEJ-11-36_F1){#F1}
{#F2}
{#F3}
{#F4}
{#F5}
{#F6}
######
Size and branching pattern in the stem villi model.
Parameter Truncus chorii Rami chorii Ramuli chorii
-------------------- ---------------- ---------------------- ------------------- -----------
**Branch** *d* \[mm\] 1 - 3 0.5 - 1 0.3 - 0.5
*z* \[mm\] 0 - 2.9 2.9 - 13.5 13.5 - 24.5
**Axis** *L~c~* \[mm\] 0 (*z* = 2.9) 8.99 (*z* = 13.5) \-
*L~b~*\[mm\] \- 0 (*z* =2.9) \*
0.29 (*z* = 3.9)
1.74 (*z* = 4.7)
4.64 (*z* = 5.8)
*r* \[mm\] \- (*L~b~*\< 1.74) 1.74 \-
(*L~b~*≥1.74) 8.99
*R~b~* \- 2 2.22 - 6.02
*C~f~* \- 4 4 - 20
The stem villi model was set in the space, whose size was 34.8 × 34.8 × 24.5 \[mm\] (*x* × *y* × *z*): the *z* coordinate shows the distance from the chorionic plate; *d*, the diameter of the branch; *L~c~*, the distance from the center of the space in each *z* coordinate (17.4 mm, 17.4 mm, *z*) to the axis of the branch at the boundary between the categories; *L~b~*, the distance from the center of the space in each *z* coordinate to the bifurcation; \*, each branch bifurcated differently; *r*, the radius of curvature *R~b~*, bifurcation ratio; *C~f~*, the centrifugal order at the tip.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Lithium secondary batteries have been studied for large scale energy devices, such as electric vehicles (EVs) and energy storage systems (ESSs), requiring high energy density and superior rate performance. The development of anode materials has been investigated due to importance of the charge rate and good reversibility for lithium secondary batteries. Commercial anode materials for batteries, such as graphite, have high capacities (370mAh/g). On the other hand, the active material has some problems, such as irreversible capacity loss, due to solid electrolyte interface (SEI) layer and lithium dendrite formation due to the low working voltage window at 0.8 V^[@CR1]^. In particular, lithium dendrite formation leads to the safety hazard of lithium secondary batteries and the unsuitability of the active materials for batteries^[@CR1]--[@CR3]^. The Si based materials such as SiO~2~ showed also high capacity but could not be used to high volume expansion^[@CR4]^. In contrast, titanium-based anode materials allow lithium batteries to avoid SEI and lithium dendrite formation due to their safe working voltage area using the Ti^4+^/Ti^3+^ redox reaction (\~1.5 V vs. Li/Li^+^)^[@CR2]^. Typically, Li~4~Ti~5~O~12~ has been studied because of its working voltage area and zero strain properties, resulting in good rate performance due to its strong Ti-O covalent bond^[@CR3]^. Despite this, the material has a low theoretical capacity (175mAh/g) and is unsuitable for large-scale devices.
Recently, titanium niobium oxide (TNO) materials, such as TiNb~2~O~7~ and Ti~2~Nb~10~O~29~, have been introduced as promising titanium-based anode materials owing to their nontoxic, good rate performance, low volume change, stable working voltage window (1--2.5 V), and high theoretical capacity (387\~390mAh/g). The capacities of TNO materials are influenced by many redox reactions, such as one Ti (Ti^3+^/Ti^4+^) and two Nb reactions (Nb^3+^/Nb^4+^ and Nb^4+^/Nb^5+^)^[@CR5],[@CR6]^. On the other hand, they have lower capacity and reversibility than their theoretical capacities due to the low electric conductivity and lithium diffusion properties into the structure called Wesley-Roth 2D structure^[@CR7]--[@CR9]^. To solve these problems, many studies have been conducted to achieve TNO materials with high reversible capacity and improved rate performance, such as doping with other metals (Ru, Mo, etc.) to achieve high ionic conductivity and electrical conductivity and controlling the particle shape and size^[@CR1]--[@CR3],[@CR6]--[@CR12]^. Chunfu Lin *et al*. examined TiNb~6~O~17~, which is a new TNO material. The material is composed a large number of Nb ions and has a higher theoretical capacity (397 mA/g) than TiNb~2~O~7~ and Ti~2~Nb~10~O~29~. Moreover, the material has the same Wisely-Roth structure (monoclinic) but larger lattice parameters and unit cell volume than TiNb~2~O~7~ and Ti~2~Nb~10~O~29~ (1122.541 Å vs. 803.21 Å, 1118.512 Å) due to the larger number of Nb^5+^ ions with a larger size (0.64 Å) than that of Ti^4+^ ions (0.605 Å)^[@CR13]--[@CR15]^. This causes a more open lithium insertion/insertion site and improved rate performance; the schema of this theory is listed in Fig. [1](#Fig1){ref-type="fig"}. The material showed a higher discharge capacity and better lithium diffusion coefficients by charge-discharge, rate capability, and slow scan cyclic voltammetry (SSCV) than Ti~2~Nb~10~O~29~ in Chunfu's study^[@CR13]^.Figure 1Schematic diagram of the kinetic mechanism of lithium diffusion in Li secondary batteries and phenomena about the unit cell size.
Therefore, this study examined the accurate lithium diffusion kinetics and electrochemical performance of TiNb~6~O~17~ compared to TiNb~2~O~7~ which has the smallest unit cell volume among the TNO materials and can clearly be compared with TiNb~6~O~17~. The materials were synthesized using a solid-state method. For electrochemical analysis, the charge-discharge curves and rate capability tests were conducted to determine their electrochemical performance. To examine the lithium diffusion kinetics, SSCV, electrochemical impedance spectroscopy (EIS), and a galvanostatic intermittent titration technique (GITT) were used. As a result, TiNb~6~O~17~ showed higher discharge capacity (284mAh/g vs. 264mAh/g) and better rate performance than TiNb~2~O~7~ (82mAh/g vs. 20mAh/g at 30 C). In addition, TiNb~6~O~17~ showed higher lithium diffusion coefficients than TiNb~2~O~7~ (mean value 10^--12^ S^2^/m vs. 10^--13^ S^2^/m).
Experimental {#Sec2}
============
Synthesis of the active materials and characterization {#Sec3}
------------------------------------------------------
TiNb~2~O~7~ and TiNb~6~O~17~ were synthesized by a solid-state reaction method using TiO~2~ (99.9%, Rare Metallic) and Nb~2~O~5~ (99.99%, Sigma-Aldrich) powders as the starting materials. TiO~2~ and Nb~2~O~5~ were mixed by ball milling at a stoichiometric molar ratio for 4 h at 300 rpm. The mixed powder was pressed into pellets and calcined in air 1300 °C for 12 h (5 °C/min). The morphology and Ti and Nb content in the two TNO materials were observed by field-emission scanning electron microscopy (FE-SEM, Jeol JSM6500F) and energy dispersion spectroscopy (EDS) attached to FE-SEM. The crystalline structures of the materials were analyzed by X-ray powder diffraction (XRD, Rigaku, Ultima4) was conducted using Ka1 radiation at 45KV/40 mA in the range, 10--100° (2θ). Fourier-transform infrared spectroscopy (FT-IR, Shimadzu IR AFFInity-1S) and X-ray photoelectron spectroscopy (XPS, ThermoFisher K-alpha) were used to examine the chemical bonding and oxidation state of the TNO materials, respectively.
Coin cell assembly and electrochemical analysis {#Sec4}
-----------------------------------------------
The composition of the TNO anodes was a mixture of active material (TiNb~2~O~7~ or TiNb~6~O~17~, 70 wt. %), conducting agent (Super-P, 20 wt. %), and polyvinylidene fluoride binder (PVdF 5130, 10 wt. %). The materials were mixed by ball-milling in 1-methyl-2-pyrrolidinone (NMP) until a viscous slurry formed and cast on Cu foil. The electrochemical properties were tested in CR2032-type coin cells. The cells were assembled with a TNO electrode as the working electrode and lithium metal as the counter electrode separated by a membrane with polypropylene in an Ar-filled glove box. The electrolyte was 1 M LiPF~6~ dissolved in a mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) with a volume ratio 1:2. Cyclic voltammetry (CV) was conducted using a battery cycler (Won A tech, WBCS3000) at a scan rate of 0.1mVs^−1^ and ranging from 0.05--0.3 mVs^−1^ from 3.0 to 1.0 V (versus Li/Li^+^). Galvanostatic charge-discharge tests were performed using the battery cycles at 0.1 C (38.7mAg^−1^ of TiNb~2~O~7~ and 39.7mAg^−1^ of TiNb~6~O~17~) from 3.0 to 1.0 V. The rate capabilities were conducted over the voltage range of 3.0--1.0 V with a current density range 1.0 C to 30 C at room temperature. EIS was carried out by applying an AC signal of 5 mV amplitude over the frequency range from 100KHz to 10mHz using an electrochemical analyzer (NeoSience, SP-300). GITT was tested at a current density of 0.1 C over the voltage range of 3.0--1.0 V using the electrochemical analyzer. The procedure of GITT consisted of galvanostatic charge pulses for each duration time (15 min), followed by a relaxation time (30 min).
Results and Discussion {#Sec5}
======================
Characterization {#Sec6}
----------------
Figure [2(a),(b)](#Fig2){ref-type="fig"} shows SEM images (magnification ×20,000) of TiNb~2~O~7~ and TiNb~6~O~17~. A comparison of the particle size and morphology was not accurate due to irregular particle formation by solid state synthesis. On the other hand, the morphologies of the two materials were similar in principle. The mean particle size of the two samples was approximately 1--3 μm. Figure [2(c)\~(f) and (g)\~(j)](#Fig2){ref-type="fig"} present SEM images of (c) TiNb~2~O~7~ and (g) TiNb~6~O~17~ (magnification ×5,000) and EDS mapping images of (d) oxygen, (e) titanium, and (f) niobium in TiNb~2~O~7~ and (h) oxygen, (i) titanium, and (j) niobium in TiN~6~O~17~. The calculated atomic percentages of the two materials by EDS are presented in Fig. [2(k)](#Fig2){ref-type="fig"} TiNb~2~O~7~ and (l) TiNb~6~O~17~. The mapping images of Ti and Nb of the two materials exhibited similar dispersion. On the other hand, the images of Ti in TiNb~2~O~7~ and TiNb~6~O~17~ showed different dispersion and brightness. Ti in TiNb~6~O~17~ was darker than that of TiNb~2~O~7~. The brightness means that the Ti content in TiNb~6~O~17~ is lower than that of TiNb~2~O~7~. These results correspond to atomic percentages of Ti and Nb in the two materials. The atomic percentage ratio of Nb and Ti in TiNb~2~O~7~ was 1:2 (Ti:Nb = 9.06:18.46), whereas the Nb: Ti ratio in TiNb~6~O~17~ was approximately 1:7 (Ti:Nb = 3.48:25.02). These results show that the molar ratio of Nb and Ti is different in the two TNO materials.Figure 2(**a**) SEM images of TiNb~2~O~7~ and (**b**) TiNb~6~O~17~ (magnification ×20,000), (**c**) SEM images of TiNb~2~O~7~ (magnification ×5,000), (**d**)\~(**f**) EDS mapping images of oxygen (yellow), titanium (green), and niobium (red), (**g**) SEM images of TiNb~6~O~17~ (magnification ×5,000), and (**h**)\~(**i**) EDS mapping images of oxygen, titanium, and niobium, and (**k**) the results of EDS analysis of TiNb~2~O~7~ and (l) TiNb~6~O17.
Figure [3](#Fig3){ref-type="fig"} (a) TiNb~2~O~7~ and (b) TiNb~6~O~17~ present XRD patterns of the two TNO materials. The pattern of TiNb~2~O~7~ was well indexed to the calculated patterns according to the monoclinic symmetry of TiNb~2~O~7~ with the monoclinic ReO~3~ shear structure with the space group C2/m (JCPDS card No. 70--2009);^[@CR6]^ however, there have been no studies of TiNb~6~O~17~. Therefore, there is no calculated structural data for TiNb~6~O~17~. On the other hand, the XRD patterns of Ti~2~Nb~10~O~29~ have been reported in many studies, which is similar to that of TiNb~6~O~17~ ^[@CR13]^. Therefore, TiNb~6~O~17~ has a similar crystal structure to Ti~2~Nb~2~O~29~, which is a Wadsley-Roth shear structure with an A2/m space group. Compared to the XRD patterns of calculated Ti~2~Nb~10~O~29~ and TiNb~6~O~17~ synthesized in this study, most peak positions and intensities were in good agreement except for two main peak intensities, which coincides with the XRD patterns reported by Chunfu Lin. The powder XRD patterns of TiNb6O17 was refined with the fullprof software and the rietveld parameters are a = 15.48089 Å, b = 3.81501 Å, c = 20.62921 Å, α&γ = 90°, β = 113.106°, and *V* *=* 1218.356 Å^3^. The calculated rietveld refinement parameters of TiNb~6~O~17~ is well matched the with the crystalline parameters of Ti~2~Nb~10~O~29~ ^[@CR13]^.Figure 3XRD patterns of (**a**) TiNb~2~O~7~ and (**b**) TiNb~6~O~17~.
FT-IR spectroscopy was conducted to characterize the Ti-O and Nb-O bond of TiNb~2~O~7~ and TiNb~6~O~17~. Figure [4(a)](#Fig4){ref-type="fig"} presents the FT-IR spectra of two samples. The peaks at 924 cm^−1^ and 520 cm^−1^ correspond to the stretching vibrations of the Nb-O bonds and Nb-O-Nb bridging bonds and the stretching vibration of at 694 cm^−1^ and 839 cm^−1^ are Ti-O-Ti bonds^[@CR12]^. The BET specific surface area and volume of the TNO materials were studied by nitrogen adsorption techniques; Fig. [4(b)](#Fig4){ref-type="fig"} shows the corresponding isotherm. The specific surface area of TiNb~2~O~7~ and TiNb~6~O~17~ is 2.66 m^2^/g and 2.36 m2/g; the mean pore volume of the materials is 0.11 cm^3^/g and 0.10 cm^3^/g respectively. As the measurement was conducted by using standard multi point BET, the specific surface area of two materials is almost same. The results are corresponded to the SEM images showing similar particle size of two materials. Therefore, the surface area of the electrodes made by two TNO materials is also same and have not an effect on the electrochemical analysis such as lithium diffusion analysis.Figure 4FT-IR spectra in Fig. (**a**) and nitrogen adsorption-desorption isotherm in Fig. (**b**) of two TNO materials.
XPS was used to analyze the chemical oxidation state of Ti and Nb in the samples, as shown in Fig. [5](#Fig5){ref-type="fig"}. Figure [5(a)](#Fig5){ref-type="fig"} TiNb~2~O~7~ and (c) TiNb~6~O~17~ showed Ti 2p~1/2~ and 2p~3/2~ peaks at 464.18 eV & 458.38 eV (TiNb~2~O~7~), and 464.18 eV & 458.18 eV (TiNb~6~O~17~), respectively. These binding energies were similar and corresponded to the binding energies of Ti^4+^ in TiO~2~ ^[@CR3],[@CR5],[@CR6],[@CR8]^. The noise of the Ti spectra was attributed to the smaller content than Nb. In particular, the spectra of Ti in TiNb~6~O~17~ showed more noise than that of TiNb~6~O~17~. This may be because TiNb~6~O~17~ is composed of a lower Ti content than TiNb~2~O~7~. These results match the results of EDS analysis and the mapping images. Figure [5(b)](#Fig5){ref-type="fig"} TiNb~2~O~7~ and (d) TiNb~6~O~17~ present the spectra of Nb^5+^ in Nb~2~O~5~.The Nb 3d~3/2~ and Nb 3d~5/2~ peaks were located at (b) 209.88 & 207.18 and (c) 209.68 & 206.98. These values agree with the binding energies of Nb^5+^ in Nb~2~O~5~ ^[@CR3],[@CR5],[@CR10]^. Therefore, FT-IR spectroscopy and XPS shows that the two TNO materials are composed with Ti^4+^ in TiO~2~ and Nb^5+^ in Nb~2~O~5~.Figure 5XPS spectra of (**a**) Ti and (**b**) Nb element in TiNb~2~O~7~, (**c**) Ti and (**d**) Nb element in TiNb~6~O~17~.
Electrochemical analysis {#Sec7}
------------------------
Figure [6 (a),(b)](#Fig6){ref-type="fig"} presents the charge and discharge curves of TiNb~2~O~7~ and TiNb~6~O~17~ at a current density of 0.1 C (38.7 mAg^−1^ and 39.7 mAg^−1^) over the voltage range of 3.0--1.0 V. The curves of the two TNO anodes showed three plateau regions. The regions 1 and 3 are the solid-solution region^[@CR6],[@CR9]^. These regions mean the redox reaction of Ti^4+^ ↔ Ti^5+^ and Nb^3+^ ↔ Nb^4+^, respectively. Region 2 is a two-phase reaction, which means the reaction of Nb^4+^ ↔ Nb^5+^ ^[@CR3],[@CR5]--[@CR13]^. Compared to the initial discharge capacities, TiNb~6~O~17~ exhibited a larger discharge capacity (284 mAhg^−1^) than that of TiNb~2~O~7~ (264 mAhg^−1^). In addition, the irreversibility of TiNb~6~O~17~ was smaller than TiNb~2~O~7~ particularly from the 1^st^ to 2^nd^ cycles. CV of TiNb~2~O~7~ and TiNb~6~O~17~ was conducted at a scan rate of 0.1mVs^−1^ from 3.0 V to 1.0 V and from 3.0 and 1.0 V for 10 cycles. As shown in Fig. [6(c)](#Fig6){ref-type="fig"} TiNb~2~O~7~ and (d) TiNb~6~O~17~, both curves Fig. [6(a)](#Fig6){ref-type="fig"} TiNb~2~O~7~ and curve (b) TiNb~6~O~17~ showed three current peaks at the oxidation and reduction state, respectively. Each peak is expressed in the curves (C~p~ and A~p~ mean the cathodic peaks and anodic peaks). Although the reduction peaks were C~p1~ (Ti^4+^ → Ti^3+^), C~p2~ (Nb^5+^ → Nb^4+^), and C~p3~ (Nb^4+^ → Nb^3+^), A~p1~, A~p2~, and A~p3~ mean the oxidation reaction of Nb^3+^ → Nb^4+^, Nb^4+^ → Nb^5+^, and Ti^3+^ → Ti^4+^ ^[@CR10],[@CR15]^. These potential regions of the current peaks were matched with the plateau regions in charge and discharge curves. These results show that the reaction mechanisms of the two TNO materials are the same. In addition, the reaction of Nb^4+^ ↔ Nb^5+^, which is corresponded to two-phase regions in the charge and discharge curves, showed the highest current peak area and is regarded as the main reaction. Compared to the CV curves of TiNb~2~O~7~ and TiNb~6~O~17~, TiNb~6~O~17~ exhibits higher reactivity and reversibility from the peak area at all cycles. In addition, the decrease in the peak intensity during the cycle, particularly A~p2~ and C~p1~, suggests that the reversibility of TiNb~6~O~17~ is better than TiNb~2~O~7~. This is in agreement with the results of the charge and discharge tests.Figure 6(**a**) Charge/discharge curves of TiNb~2~O~7~ and (**b**) TiNb~6~O~17~anodes at 0.1 C, (**c**) cyclic voltammetry of TiNb~2~O~7~ and (**d**) TIiNb~6~O~17~anodes in the potential window of 1.0--3.0 V at scan rate of 0.1 mVs^−1^.
To understand the electrochemical performance of the lithium diffusion properties of TiNb~2~O~7~ and TiNb~6~O~17~, the rate capabilities were performed at various C-rates from 1 C to 30 C (discharge rate was fixed at 1 C). Figure [7](#Fig7){ref-type="fig"} presents the rate performance of the two TNO materials. A comparison with the average capacities for the 5^th^cycle at each C-rate revealed TiNb~6~O~17~ to have charge capacities of 252, 230, 206, 187, 107, and 80 mAhg^−1^ at 1 C, 2 C, 5 C, 10 C, 20 C, and 30 C, respectively. These values are larger than that of TiNb~2~O~7~ (234, 210, 174, 152, 52, and 19 mAhg^−1^). In particular, the difference in the charge capacities at a high rate (20 C and 30 C) was distinct. When calculating the ratio of the average charge capacity, 30 C/1 C, the ratio was 8.12% for TiNb~2~O~7~ and 31.7% for TiNb~6~O~17~, which suggests that TiNb~6~O~17~ has better rate properties than TiNb~2~O~7~ ^[@CR13]^. In addition, a comparison of the cycling retention at 5 C to 30 C revealed TiNb~6~O~17~ to have better cycling properties, whereas TiNb~2~O~7~ exhibited a rapid decrease in capacity. This means the better electrochemical reversibility of the TiNb~6~O~17~. These studies including the results of the charge and discharge tests and CV indicated that lithium ion transport of TiNb~6~O~17~ is faster than the rate of TiNb~2~O~7~ due to the larger theoretical capacity and better lithium diffusion kinetics by larger lithium site.Figure 7Capacity retention of TiNb~2~O~7~ and TiNb~6~O~17~ anodes at various scan rates (1 C, 2 C, 5 C, 10 C, 20 C, and 30 C); the discharge rate was fixed at 1 C.
Figure [8](#Fig8){ref-type="fig"} presents the CV data of (a) TiNb~2~O~7~ and (b) TiNb~6~O~17~ at various scan rates in the range, 0.05--0.3 mVs^−1^. CV at various scan rates is usually used to study the oxidation and reduction properties in electrochemical reactions and obtain the apparent chemical diffusion coefficient of Li-ions^[@CR16]--[@CR20]^. With increasing scan rate, the anodic peaks move to a low potential and the cathodic peaks move to a high potential due to the increasing polarization. In addition, the peak intensities of anodic and cathodic reaction increase with increasing scan rate. The peak current density (*I* ~p~) revealed a linear relationship with the square root of the scan rate (*v* ^−0.5^), which is expected for a diffusion-controlled process in Fig. [8(c)](#Fig8){ref-type="fig"} TiNb~2~O~7~ and (d) TiNb~6~O~17~ ^[@CR20]--[@CR22]^. Each color means the linearity of three anodic and cathodic peaks (Black: A~p1~ and C~p1~, Pink: A~p2~ and C~p2~, and Purple: A~p3~ and C~p3~). The relationship and chemical diffusion coefficient can be determined from the Randles-Sevcik equation (Eq. [1](#Equ1){ref-type=""})^[@CR16],[@CR17],[@CR23]^:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$I{\rm{p}}=0.4463{{\rm{n}}}^{3/2}{{\rm{F}}}^{3/2}{{\rm{C}}}_{{\rm{Li}}}+{{\rm{SR}}}^{-1}{{\rm{T}}}^{-1}{{{\rm{D}}}_{({{\rm{Li}}}^{+})}}^{1/2}{v}^{1/2}$$\end{document}$$where n is the charge transfer number; F is Faraday's constant; C~Li~ ^+^ is the Li-ion concentration in TiNb~2~O~7~ and TiNb~6~O~17~; S is the surface area per weight of active materials; R is the gas constant; and T is the absolute temperature (K). $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ is the Li-ion diffusion coefficient, and *v* is the scan rate. In this study, $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ around three anodic and three cathodic peaks in Fig. [6(c) and (d)](#Fig6){ref-type="fig"} was calculated using the above equation. Table [1](#Tab1){ref-type="table"} lists the calculated $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$. As the results, TiNb~2~O~7~ showed the $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ value 10^−14^ cm^2^/s which is similar to the diffusion coefficient in the previous study (for phase transition region)^[@CR24]^. Compared to $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ at the anodic peaks, $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ of TiNb~6~O~17~ was 20 times (A~p1~), 12 times (A~p2~), and 38 times (A~p3~) higher than that of TiNb~2~O~7~. $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ of the peaks A~p1~ (Nb^4+^→ Nb^5+^) and A~p3~ (Ti^3+^→ Ti^4+^) of TiNb~6~O~17~ was particularly high. Although the gap of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ at the A~p2~ (Nb^4+^→ Nb^5+^) between TiNb~2~O~7~ and TiNb~6~O~17~ was smaller than those of A~p1~ and A~p3~, the difference was apparent. In the case of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ at cathodic peaks, the values of TiNb~6~O~17~ were 5 times (C~p1~, Nb^4+^→ Nb^3+^), 15 times (C~p2~, Nb^5+^→ Nb^4+^), and 14 times (C~p3~, Ti^4+^→ Ti^3+^) higher than those of TiNb~2~O~7~. A comparison of the gap of D between TiNb~6~O~17~ and TiNb~2~O~7~ at the anodic peaks revealed the difference in the D~Li~ ^+^ values at the cathodic peaks to be low except for C~p2~. On the other hand, the D~Li~ ^+^ of TiNb~6~O~17~ at A~p2~ and C~p2~ meaning two phase transition in TNO materials were clearly higher than that of TiNb~2~O~7~(12 and 15 times). In addition, the anodic and cathodic reaction of the TNO anodes means the de-lithiation and lithiation process during oxidation and reduction, respectively. Therefore, the lithium diffusion properties of TiNb~6~O~17~ were better than those of TiNb~2~O~7~. The reason is that TiNb~6~O~17~ has a larger unit cell volume and more open Li-ion sites than TiNb~2~O~7~. The advanced crystal structure of TiNb~6~O~17~ leads to a larger size and number of Li-ion transport paths in the crystal structure, facilitating Li-ion transport during the de-lithiation and lithiation processes^[@CR10],[@CR12],[@CR18]^.Figure 8Cyclic voltammetry with various scan rate of (**a**) TiNb~2~O~7~ and (**b**) TiNb~6~O~17~, linear relationship between the peak current of the cathodic/anodic reaction and the square root of the sweep rate (**c**) TiNb~2~O~7~ and (**d**) TiNb~6~O~17~ (●: anodic, ■: cathodic and linear: linear fitting). Table 1Calculated D~Li~ ^+^ values of (a) TiNb~2~O~7~ and (b) TiNb~6~O~17~ anodes from the CV results.TiNb~2~O~7~Anodic peakCathodic peak**D** ~**Li**~ ^**+**^ **(cm** ^**2**^ **s** ^**−1**^ **)**ABCABC5.69 × 10^−15^3.01 × 10^−14^1.60 × 10^−15^1.16 × 10^−14^2.33 × 10^−14^1.08 × 10^−15^TiNb~6~O~17~Anodic peakCathodic peak**D** ~**Li**~ ^**+**^ **(cm** ^**2**^ **s** ^**−1**^ **)**ABCABC1.12 × 10^−13^3.72 × 10^−13^6.13 × 10^−14^5.35 × 10^−14^3.43 × 10^−13^1.56 × 10^−14^
Figure [9](#Fig9){ref-type="fig"} presents the Nyquist plots of TiNb~2~O~7~ and TiNb~6~O~17~ by EIS. EIS has been used to examine electrode materials because it can reveal the relationship between the crystal lattice with the electrochemical properties^[@CR24]--[@CR29]^. This technique provides kinetic information that can be related to a specific state-of-charge or discharge (SOC, SOD), because the measurement is run by applying a low amplitude signal around an equilibrium state^[@CR26]--[@CR29]^. Figure [9(a)](#Fig9){ref-type="fig"} shows the Nyquist plot of TiNb~2~O~7~ and TiNb~6~O~17~ at the open circuit voltage (OCV) and an equivalent circuit (insert image). Each Nyquist plot was composed of a high-frequency semicircle and Warburg tail region followed by a steep sloping line in the low-frequency region^[@CR27]^. The R~1~ and C~dl~ are the ohmic resistance between the electrolyte and surface of the electrode and double layer capacitance. The high-frequency semicircle means the charge-transfer resistance (R~ct~) relevant to the interfacial Li-ion transfer. The Z~w~ is the Warburg impedance, which is related to Li-ion diffusion to the structure of the active materials and corresponds to the tail at a low frequency. Compared to R~ct~, the TiNb~6~O~17~ anode shows a smaller R~ct~ (58Ω) than that of the TiNb~2~O~7~cell (85 Ω). This means that the TiNb~6~O~17~ anode has a faster Li insertion process in the surface area than TiNb~2~O~7~. Figure [9(b)](#Fig9){ref-type="fig"} presents a plot of the real part resistance with the inverse square root of the angular speed in the low-frequency range of TiNb~2~O~7~ and TiNb~6~O~17~ anodes at the OCV state. The Warburg factor ($\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\rm{\sigma }}$$\end{document}$) is determined from the slope, and is substituted using equation (Eq. [2](#Equ2){ref-type=""} and [3](#Equ3){ref-type=""}):$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$Z^{\prime} ={R}_{{1}}+{R}_{ct}+\sigma {\omega }^{(-1/2)}$$\end{document}$$ $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${D}_{Li}=\frac{{R}^{2}{T}^{2}}{2{A}^{2}{n}^{2}{F}^{4}{C}^{2}{\sigma }^{2}}$$\end{document}$$where Z' is the real part resistance; $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\omega $$\end{document}$ is the angular frequency; R is the gas constant; T is the absolute temperature; A is the surface area of the electrode; F is the Faraday constant; and C is the molar concentration of Li ion in an active material. Equations ([2](#Equ2){ref-type=""}) and ([3](#Equ3){ref-type=""}) were used to calculate the Warburg factor and lithium diffusion coefficient, respectively. Table [2](#Tab2){ref-type="table"} lists the calculated $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ from the obtained $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\rm{\sigma }}$$\end{document}$. The $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ value of TiNb~2~O~7~ and TiNb~6~O~17~ anodes at the OCV state was 4.57 × 10^--14^ cm^2^s^−1^ and 1.27 × 10^--13^ cm^2^s^−1^, respectively. As a result, the TiNb~6~O~17~ anode has a better lithium diffusion process than the TiNb~2~O~7~ anode. Although it is not the charge/discharge state, the results revealed the improved lithium diffusion kinetics of TiNb~6~O~17~ at the equilibrium state because the more open Li ion insertion site of TiNb~6~O~17~ than that of TiNb~2~O~7~ affects the barrier energy and electrostatic interaction regarding the Li^+^ insertion mechanism.Figure 9(**a**) Nyquist plots of TiNb~2~O~7~ and TiNb~6~O~17~ anodes at OCV state, (**b**) relationship between imaginary resistance (Z') and inverse square root of angular speed (at $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\rm{\omega }}$$\end{document}$ ^−0.5^) low frequency region, (**c**) Nyquist plots of TiNb~2~O~7~ and (**d**) TiNb~6~O~17~ (Inert images: relationship between Z' and $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\rm{\omega }}$$\end{document}$ ^−0.5^). Table 2Calculated D~LI~ ^+^ values of (a) TiNb~2~O~7~ and (b) TiNb~6~O~17~ anodes from the EIS results.D~Li~ ^+^ (cm^2^s^−1^) (OCV)D~Li~ ^+^ (cm^2^s^−1^) (1.36 V)D~Li~ ^+^ (cm^2^s^−1^) (1.68 V)D~Li~ ^+^ (cm^2^s^−1^) (1.98 V)**TiNb** ~**2**~ **O** ~**7**~4.57 × 10^−14^6.64 × 10^−14^7.91 × 10^−13^1.85 × 10^−12^**TiNb** ~**6**~ **O** ~**17**~1.27 × 10^−13^2.94 × 10^−13^1.12 × 10^−11^4.57 × 10^−11^
To investigate the Li-ion diffusion properties of two TNO materials at the charge state, *ex-situ* EIS experiments were performed on TiNb~2~O~7~ and TiNb~6~O~17~ anodes at three oxidation reaction potentials of Nb^3+^→ Nb^4+^(1.36 V), Nb^4+^→ Nb^5+^(1.68 V), and Ti^3+^→ Ti^4+^(1.98 V) in Fig. [6 (c),(d)](#Fig6){ref-type="fig"}. Before the EIS experiments, the discharge and charge were processed during 1 cycle and the discharge was then conducted to the cut off potential of 1.0 V. Figure [9(c)](#Fig9){ref-type="fig"} TiNb~2~O~7~ and (d) TiNb~6~O~17~ present Nyquist plots of the two anodes from EIS (Inert images: plot of the real part resistance with the inverse square root of angular speed in the low-frequency range at three oxidation potential). The calculated $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ value is listed in Table [2](#Tab2){ref-type="table"} with a value at the OCV state. Compared to $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ of two TNO anodes from EIS, TiNb~6~O~17~ showed higher $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ values of 2.94 × 10^−13^ cm^2^s^−1^, 1.12 × 10^−11^ cm^2^s^−1^, and 1.85 × 10^−12^ cm^2^s^−1^ at 1.36 V, 1.68 V, and 1.98 V, respectively, than those of TiNb~2~O~7~ (6.64 × 10^−14^ cm^2^s^−1^, 1.12 × 10^−11^ cm^2^s^−1^, and 4.57 × 10^−11^cm^2^s^−1^). The $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ values of TiNb~6~O~17~ were 4.4 times (1.36 V), 14 times (1.68 V), and 25 times (1.98 V) higher than those of TiNb~2~O~7~. Compared to the SSCV results, the $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ values of TiNb~6~O~17~ showed different multiples except for the value at 1.68 V (20, 12, and 38 times at A~p1~-1.36 V, A~p2~-1.68 V, and A~p3~-1.98 V, respectively, from SSCV) but exhibited similar tendency showing higher $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ at all redox potentials than TiNb~2~O~7~. In particular, the gap of $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{\rm{Li}}}^{+}}$$\end{document}$ at 1.68 V (A~p2~ peak at CV) meaning that the two phase regions coincide well with the results of SSCV (14 and 12 fold, respectively.) Therefore, TiNb~6~O~17~ has better lithium diffusion properties than TiNb~2~O~7~ due to its structure inducing a larger open lithium site and a number of Li-ion transport paths during the charge processes.
GITT was conducted to determine the Li^+^ chemical diffusion coefficient and analyze the phase transition of the two TNO materials. The techniques developed by Weppner and Huggins assumed one-dimensional diffusion in a solid solution electrode and a uniform current distribution throughout the electrode and estimated the electrochemically active area from the structure of the active material particles not for the diffusion reaction between the electrode surface and electrolyte^[@CR31]--[@CR37]^. At the transitional GITT, a small constant current was applied to an electrode during a short time and the electrode was left to stand after reaching the OCV state^[@CR31]--[@CR35]^. In this study, GITT was performed on the TNO materials to determine the $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${{\rm{D}}}_{{{Li}}^{+}}$$\end{document}$ at a single phase and two phase region as a function of the voltage for the cut off range of the charge/discharge cycle, 1.0--3.0 V. Figure [10](#Fig10){ref-type="fig"} shows the GITT curves of (a) TiNb~2~O~7~ and (b) TiNb~6~O~17~ during the second cycle. The cells were first discharged at a constant current (0.1 C) for a duration time of 15 min and a rest time of 30 min. The curves showed a similar shape and exhibited three plateau regions meaning the solid-solution regions (Ti^4+^ ↔ Ti^5+^ and Nb^3+^ ↔ Nb^4+^) and two-phase reaction (Nb^4+^ ↔ Nb^5+^) with the charge-discharge curves. These regions also showed the cyclic voltammetry peaks of C~p1~ (Ti^4+^ → Ti^3+^), C~p2~ (Nb^5+^ → Nb^4+^), and C~p3~ (Nb^4+^ → Nb^3+^); A~p1~, A~p2~, and A~p3~ mean the oxidation reaction of Nb^3+^ → Nb^4+^, Nb^4+^ → Nb^5+^, and Ti^3+^ → Ti^4+^in Fig. [6(c),(d)](#Fig6){ref-type="fig"}, Fig. [10 (c)](#Fig10){ref-type="fig"} TiNb~2~O~7~ and (d) TiNb~6~O~17~ present the single steps of GITT. The steps are the results measured at the 3^th^ step during the charge state for the same duration and rest time. In Fig. [10(c) and (d)](#Fig10){ref-type="fig"}, ΔE~τ~ and ΔE~s~ shows the change in the cell voltage during the duration time of 15 min from $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\tau }_{0}$$\end{document}$ to $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${\tau }_{0+t}$$\end{document}$ and the variation of the cell voltage during the rest time of 30 min. The voltage changes from the steps are recorded as a function of time and the lithium diffusion coefficient were calculated using the following equation based on Fick's second law^[@CR32]^.$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${D}_{L{i}^{+}}=\frac{4}{\pi }{(\frac{{m}_{B}{V}_{M}}{{M}_{B}A})}^{2}{(\frac{{\rm{\Delta }}{E}_{S}}{\tau (d{E}_{\tau }/d\sqrt{\tau }})}^{2}({\rm{\tau }}\ll \frac{{L}^{2}}{{D}_{L{i}^{+}}})$$\end{document}$$where V~M~ is the molar volume of the active material; M~B~ is molecular weight of the materials; m~B~ is the mass of the active materials in an electrode; L is the lithium diffusion distance (thickness of the electrode); A is the electrode area; and $\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\tau $$\end{document}$ is the duration time. When the change in cell voltage with duration time exhibited a linear relationship on plotting against τ^1/2^, equation ([4](#Equ4){ref-type=""}) can be changed to the following simple equation^[@CR32]^ $$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${D}_{L{i}^{+}}=\frac{4}{\pi \tau }{(\frac{{m}_{B}{V}_{M}}{{M}_{B}A})}^{2}{(\frac{{\rm{\Delta }}{E}_{S}}{{\rm{\Delta }}{E}_{S}})}^{2}({\rm{\tau }}\ll \frac{{L}^{2}}{{D}_{L{i}^{+}}})$$\end{document}$$ Figure 10Charge/discharge GITT curves of (**a**) TiNb~2~O~7~ and (**b**) TiNb~6~O~17~, single step of the relationship of single steps for (**c**) and (**d**) (V vs. τ^1/2^), (**e**) and (**f**) lithium diffusion coefficients calculated from GITT for TiNb~2~O~7~ and TiNb~6~O~17~ as a function of the SOC at the charge process.
This equation assumes that the molar volume (V~M~) is stable with the change in Li content in an active material. In this study, the Li^+^ diffusion coefficient of the two TNO materials could be calculated, as shown in Fig. [10. (c)--(e)](#Fig10){ref-type="fig"}. Figure [10.(e)](#Fig10){ref-type="fig"} shows the linear relationship between the single steps in Fig. [10. (c),(d)](#Fig10){ref-type="fig"}. The Li^+^ diffusion coefficients of the two TNO materials from the GITT results are presented as a function of SOC (%) vs. Log ($\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${D}_{{{Li}}^{+}})$$\end{document}$ during the charge state in Fig. [10. (f)](#Fig10){ref-type="fig"}. The coefficients were calculated at all steps during the GITT measurements except for the 1^st^ step and the end two steps of the end due to the large voltage variations. The two cells showed three minimum Li^+^ diffusion coefficient points in Fig. [10.(f)](#Fig10){ref-type="fig"} and the voltages representing the points are shown. These minimum diffusion coefficients suggest a phase transition for strong attractive interactions between the intercalation species and the host matrix or some order-disorder transition during cycling^[@CR24],[@CR32]^. Compared to the SSCV and EIS results, the voltages are the three redox potentials of TNO materials, in which the cell voltages of TiNb~2~O~7~ and TiNb~6~O~17~ are 1.42 V\~1.38 V (Nb^3+^ → Nb^4+^), 1.71\~1.75 V (Nb^4+^ → Nb^5+^), and 2.01\~2.03 V (Ti^3+^ → Ti^4+^) vs. 1.36 V, 1.68 V, and 1.98 V, respectively, from the SSCV and EIS measurements. This explains why the plot from GITT has an electrochemical reaction mechanism of two TNO materials with SSCV and EIS. The Li^+^ diffusion coefficients of TiNb~2~O~7~ and TiNb~6~O~17~ from three points were calculated to be 1.11 × 10^−11^ and 6.70 × 10^−11^ cm^2^s^−1^ (Nb^3+^ → Nb^4+^), 2.74 × 10^−11^ and 2.23 × 10^−10^ cm^2^s^−1^ (Nb^4+^ → Nb^5+^), and 1.03 × 10^−10^ and 7.47 × 10^−10^ cm^2^s^−1^ (Ti^3+^ → Ti^4+^). The coefficients of TiNb~6~O~17~ showed higher values than that of TiNb~2~O~7~ at all positions with the other Li^+^ diffusion measurements, which indicates that TiNb~6~O~17~ has superior Li^+^ diffusion kinetics than TiNb~2~O~7~ owing to its larger unit cell volume. Compared to the diffusion coefficients of each transition region, the values increased from the (Nb^3+^ → Nb^4+^) reaction to the (Ti^3+^ → Ti^4+^) reaction, which correspond to the EIS results in Table [2](#Tab2){ref-type="table"}. Among the three diffusion values, the diffusion coefficients of the (Nb^4+^ → Nb^5+^) reaction showed the largest increase from TiNb~2~O~7~ to TiNb~6~O~17~ and also corresponds to the EIS results. These trends suggest that the oxidation reaction is a two phase transition region of TNO materials with the charge-discharge curves and cyclic voltammetry results (the most reaction region). In the event of SSCV, the measurements showed a different tendency with EIS and GITT. The coefficients of the (Nb^4+^ → Nb^5+^) reaction (A~p2~) showed the largest values and the diffusion coefficients of the (Nb^3+^ → Nb^4+^) reaction showed the largest increase from TiNb~2~O~7~ to TiNb~6~O~17.~ This may be due to the inaccuracy of the SSCV measurements in this study. Compared to A~p2~, both A~p1~ and A~p3~ showed a small peak current and a broad shape. Therefore, the diffusion coefficients of the two peaks may be not precise values.
Conclusions {#Sec8}
===========
Galvanostatic charge-discharge, cyclic voltammetry, and rate capability test were conducted to analyze the electrochemical performance and properties of TiNb~6~O~17~ and TiNb~2~O~7~. From the results, two TNO materials showed three similar plateau regions and three redox peaks corresponding to two Nb redox and one Ti redox reaction. TiNb~6~O~17~ showed higher capacities of 284mAh/g than that of TiNb~2~O~7~ 264mAh/g. In the rate capability test, TiNb~6~O~17~ exhibited improved rate capacity of 80mAh/g at 30 C than 19mAh/g for TiNb~2~O~7~. SSCV, EIS, and GITT measurement were taken to investigate the performance and lithium diffusion properties related to the unit cell volume of the two TNO materials. The anodic and cathodic Li^+^ diffusion coefficients from SSCV were in the range of 10^−14^ to 10^−15^ cm^2^s^−1^ for TiNb~2~O~7~ and 10^−13^ to 10^−14^ cm^2^s^−1^ for TiNb~6~O~17~. The anodic diffusion coefficients of TiNb~6~O~17~ were 5 times (Nb^3+^ → Nb^4+^), 15 times (Nb^4+^ → Nb^5+^), and 14 times (Ti^3+^ → Ti^4+^). From the EIS measurement, the coefficients were in the range of 10^−12^ to 10^−14^cm^2^s^−1^ of TiNb~2~O~7~ and 10^−11^ to 10^−13^cm^2^s^−1^ of TiNb~6~O~17~ at the OCV state and three oxidation potential region of the two TNO materials during the charging process. The three minimum diffusion coefficients points were determined from the GITT measurement. The diffusion coefficients of the two phase transition region (Nb^4+^ → Nb^5+^) were improved 10 fold compared to that of TiNb~2~O~7~. CV, EIS, and GITT indicated that TiNb~6~O~17~ has better lithium diffusion kinetics and electrochemical performance than TiNb~2~O~7~ because of its large unit cell volume and more open Li^+^ insertion site.
**Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This study was supported by the Priority Research Centers Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Science and Technology (MEST) of the Korean government (2009--0093818).
Kwang-Sun Ryu decided the concept of the experiments, discussed the results and reviewed the manuscript. Yong-Seok Lee performed all experiments, analyzed the data and wrote the manuscript.
Competing Interests {#FPar1}
===================
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-polymers-11-01320}
===============
Thermoplastic polyurethane (TPU) is widely used in daily life and industrial fields due to its exceptional processability and mechanical properties amid rubbers and plastics. It principally consists of a long-chain polyester or polyether with relative high molecular weight as soft segments, short-chain extenders and urethane and/or urea groups as hard segments. Conventionally, tuning the mechanical properties, thermal resistance, and other special features of TPU lies in altering the structure of the soft and hard segments, because its elasticity is essentially originated from the physical network constructed by micro-domains derived from hydrogen bonded hard segments \[[@B1-polymers-11-01320],[@B2-polymers-11-01320],[@B3-polymers-11-01320],[@B4-polymers-11-01320]\].To date, it is still a great challenge to design perfect structure and condensed state for improving tensile strength, elongation, and toughness at the same time, since soft and hard segments have mutually competitive dependence on chain mobility.
Many efforts have been devoted to strengthening and toughening TPU via incorporation of fillers \[[@B5-polymers-11-01320],[@B6-polymers-11-01320],[@B7-polymers-11-01320],[@B8-polymers-11-01320],[@B9-polymers-11-01320]\] and in-situ synthetic methods \[[@B10-polymers-11-01320]\]. Synthesizing TPU with high performance refers to molecular structure and condensed state design that consequentially controls network, soft and hard segments. Recently, sacrificial bonds, defined as covalent or noncovalent bonds which perform rupture of their binding prior to the cleavage of major structural linkage upon applied force \[[@B11-polymers-11-01320],[@B12-polymers-11-01320]\], are prevailing in designing materials with self-healing ability and enhanced mechanical properties. Hydrogen bonds in TPU, induced by either urethane or urea groups, could actually act as sacrificial bonds; however, the weak binding energy somehow limits the enhancing ability. For instance, the presence of numerous hydrogen bonds in TPU generates improved stiffness and strength, and decreased elongation, while a low amount of hydrogen bonds is not favorable for improving strength. Metal-ligand interaction is basically stronger in binding energy than hydrogen bonding \[[@B13-polymers-11-01320],[@B14-polymers-11-01320]\], and thus some researchers have incorporated it into elastomers for constructing dynamic noncovalent network in order to improve both strength and toughness or confer self-healing ability \[[@B15-polymers-11-01320],[@B16-polymers-11-01320],[@B17-polymers-11-01320],[@B18-polymers-11-01320],[@B19-polymers-11-01320],[@B20-polymers-11-01320],[@B21-polymers-11-01320]\]. These approaches arranged ligands as side groups of polyolefin elastomers or thermosetting polyurethane (PU) elastomers at chain ends. In these cases, the metal-ligand interaction either imposes a negligible effect on backbone mobility or is accompanied by the effect of the physical or chemical network. Some researchers have recently developed TPU with ligands aligning in backbone, which showed effective improvement of mechanical properties. For instance, Hong et al. \[[@B22-polymers-11-01320]\] and Zhang et al. \[[@B23-polymers-11-01320]\] utilized triazol-pyridine derivative and curcumin as both chain-extender and ligand. The former authors found that complete complexation of ligands resulted in optimal mechanical properties, but, unfortunately, less interpretation of structure-properties was mentioned. The latter also found that the robustness of TPU could be effectively enhanced with a specific metal/ligand ratio, but the influence of metal-ligand interaction on structure merely referred to distance of hard domains. Therefore, how the metal/ligand ratio and coordinative bonds affect hard segments and formation of noncovalent cross-linkages still requires investigation.
Herein, we present a routine approach to synthesize poly(propylene glycol)-based poly(urethane urea) (PUU) by using 2,6-diaminopyridine as chain extender and ligand. Poly(propylene glycol) is selected as a soft segment due to its noncrystallinity and high chain flexibility. 2,6-Diaminopyridine is unique because it could serve as a backbone ligand and its adjacent urea groups might be notably affected by coordinative bonds. Although a recent paper reported that 2,6-diaminopyridine and poly(tetramethylene ether glycol)-based TPU showed improved robustness and self-healing ability \[[@B24-polymers-11-01320]\], our work focuses on revealing the role of Zn^2+^/pyridine ratio and its metal-ligand interaction in affecting hard segments and formation of noncovalent cross-linkages. It was found that the optimal Zn^2+^/pyridine ratio for effectively improving robustness does not necessarily belong to complexation of entire ligands, and taking advantages of interfering with hard segments by metal-ligand interaction is the key for tuning strength, elongation, and toughness at the same time.
2. Materials and Methods {#sec2-polymers-11-01320}
========================
2.1. Materials and Synthesis {#sec2dot1-polymers-11-01320}
----------------------------
Poly(propylene glycol) (PPG) (Mn = 2000 g/mol) was purchased from Bluestar (Zibo, China) and used without further purification. 4,4'-Diphenylmethane diisocyanate (MDI) was purchased from Wanhua (Yantai, China) and used without further purification. 2,6-Diaminopyridine, 1,6-hexanediamine (HMD), and zinc trifluoromethanesulfonate (Zn(OTf)~2~) were purchased from Adamas (Shanghai, China) and used without further purification. *p*-Chlorophenol and dibutyltin dilaurate (DBTDL) were purchased from Aladdin (Shanghai, China) and used without further purification. *N*,*N*-dimethylformamide (DMF), acetonitrile, and tetrahydrofuran (THF) were purchased from Kermel (Tianjin, China) and used after redistillation for removing residual moisture.
Typical synthesis of PPG and 2,6-diaminopyridine-based poly(urethane urea) (PUU-Py) is presented in [Scheme 1](#polymers-11-01320-sch001){ref-type="scheme"}. PPG (12.00 g, 6 mmol) in a dried glass vessel equipped with a mercury thermometer and a mechanical agitation was heated to 120 °C under vacuum for 40 min to remove residual moisture in raw material, and then cooled down to 80°C. MDI (3.153 g, 12.6 mmol) was added into glass container for 1 h at 80 °C. DBTDL (0.7584 mg, 0.00125 mmol) was then added into glass container for another 1 h, resulting in pre-polymer. Then 2,6-diaminopyridine (0.6842 g, 6.27 mmol) as chain extender dissolved in anhydrous DMF (10 mL) was added into the container for 1 h. *p*-Chlorophenol (0.0848 g, 0.66 mmol) was added into the container as blocking agent. The reaction was continued until the isocyanate group (NCO) peak disappeared from the Fourier transform infrared spectroscopy (FTIR) spectrum, which required 30 min. Zn^2+^/pyridine ratio is defined as molar ratio of Zn^2+^ to pyridine moiety. To introduce a ratio of 1:4, Zn(OTf)~2~ (0.5698 g, 1.568 mmol) dissolved in acetonitrile (10 mL) was added into the container for 10 min to ensure homogeneous mixture, and the as-prepared elastomer is denoted as PUU-Py1/4. PUU-Py1/6, PUU-Py1/3 and PUU-Py0/1 elastomers with various Zn^2+^/pyridine ratios were prepared by an identical method with adjusting content of Zn(OTf)~2~. Synthesis of PUU-HMD was slightly different, i.e., the pre-polymer was preliminarily end-capped with *p*-chlorophenol prior to chain-extending with HMD. All reaction procedures were carried out in argon atmosphere and mechanical agitation.
2.2. Methods {#sec2dot2-polymers-11-01320}
------------
### 2.2.1. Preparation of Films {#sec2dot2dot1-polymers-11-01320}
As-synthesized product was poured into a rectangle Teflon mold with dimensions of 120 mm × 90 mm × 15 mm. The mold was placed in an oven and heated at 80 °C for 48 h, then put the mold into vacuum, drying at 80 °C for 24 h to remove residual solvent.
### 2.2.2. Mechanical Properties Tests {#sec2dot2dot2-polymers-11-01320}
Dumbbell specimens were cut for tensile strength tests and cyclic tensile tests, which were both performed using a tensile tester (TOPHUNG, TH-8203A, Suzhou, China) loaded with a 500 N load cell. The tensile strength tests were conducted at a constant speed of 100 mm/min at room temperature. The cyclic tensile properties were measured by stretching specimens to 400% strain at a constant speed of 50 mm/min. Two cycles were carried at an interval of 0 min.
### 2.2.3. Stress-Relaxation Tests {#sec2dot2dot3-polymers-11-01320}
The test specimens were bar-shaped films with approximate dimensions of 15 mm × 6 mm × 1.0 mm. The stress-relaxation was tested by TA Q800 Instruments (TA instruments, New Castle, DE, USA). Samples were stretched to 100% and the constant strain amplitude was maintained to measure the relaxation of the stress for 60 min.
### 2.2.4. Fourier Transform Infrared Spectroscopy (FTIR) {#sec2dot2dot4-polymers-11-01320}
Bulk samples were brought to perform Fourier transform infrared spectroscopy (FTIR) in attenuated total reflectance (ATR) mode. Solution samples of various PUU-Py elastomers in DMF (2 g/mL) were brought to carry out FTIR in transmitted mode at 25 °C from Bruker (VERTEX 70v, Karlsrule, Baden-Wurttemberg, Germany).
### 2.2.5. UV-vis Spectroscopy {#sec2dot2dot5-polymers-11-01320}
UV-vis spectroscopy was carried out using Hewlett Packard 8453 UV-vis Spectrophotometer (G1103A), where 17 μL of Zn(OTf)~2~ solution in methanol was each time added to 2 mL of PUU-Py solution in THF (0.25 mg/mL). UV-vis spectroscopy of PUU-HMD was carried out by its solution in DMF because of relative good solubility compared to in THF.
3. Results and Discussion {#sec3-polymers-11-01320}
=========================
Recent literature reported \[[@B15-polymers-11-01320],[@B16-polymers-11-01320],[@B20-polymers-11-01320],[@B25-polymers-11-01320]\] that dangling ligand moieties along the backbone was an effective means of enhancing mechanical properties. Our strategy to prepare robust and durable PUU elastomer lies in the incorporation of ligand moiety into backbone in order to form metal-ligand interaction in backbone. Another important factor is that the pyridine moiety might impose interference with hydrogen bonding of the adjacent urea groups, which would compromise micro-phase separation of hard segments and sequentially decrease stiffness and confer large elongation.
3.1. Mechanical Properties {#sec3dot1-polymers-11-01320}
--------------------------
Tensile tests were performed to reflect effect of metal-ligand interaction on robustness of as-prepared elastomers. As shown in [Figure 1](#polymers-11-01320-f001){ref-type="fig"}a, tensile strength of PUU-Py elastomers underwent drastic changes upon introduction of Zn^2+^. Comparison of tensile strength and toughness is shown in [Figure 1](#polymers-11-01320-f001){ref-type="fig"}b,c. Specifically, tensile strength, elongation, and toughness of PUU-Py0/1 was measured to be 4.9 MPa, 995%, and 31.3 MJ/m^3^, respectively. By contrast, PUU-Py1/2, PUU-Py1/3, and PUU-Py1/6 have apparent improved robustness, because the individual tensile strength increased to 6.5 MPa, 8.8 MPa, and 6.7 MPa with paralleled elongation while the corresponding toughness was calculated to be 33.1 MJ/m^3^, 50.5 MJ/m^3^, and 47.5 MJ/m^3^, respectively. It is surprising that PUU-Py1/4 has the highest tensile strength, elongation, and toughness, i.e., 16.0 MPa, 1286%, and 89.3 MJ/m^3^ with 226%, 29%, and 185% increments in comparison with PUU-Py0/1.
Introduction of a chemically covalent network of TPU usually favors in improving merely tensile strength, but is quite difficult to improve tensile strength and elongation at the same time because the extensibility of curled chains is severely constrained by covalent cross-linkages \[[@B26-polymers-11-01320]\]. Analogous to chemical network, physical network may also restrict complete extension of curled chains. To elucidate it clearly, curled chains of soft segment may undergo thorough extension or even chain cleavage, before large hard domains perform complete deformation and chain extension. Therefore, tuning hard domains plays a pivotal role in sufficient exertion of strengthening and toughening.
In the present work, incorporation of Zn^2+^ seems to successfully induce coordinative bonds with backbone pyridine moieties, because the mechanical properties were remarkably influenced. Such metal-ligand interaction also indicates that hard domains were compromised due to increased elongation at break for PUU-Py elastomers complexed with Zn^2+^. It can be observed that PUU-Py1/4 exhibited an upturn of slope around 800% instead of fracture. This result indicates that dislocation of chains was restricted when curled chains were forced to stretch to the largest extent. This phenomenon implies that Zn^2+^ played a critical role in influencing mechanical properties and the optimal ratio for enhancing robustness did not necessarily correspond to high ratio of Zn^2+^/pyridine. Considering that Zn^2+^/pyridine ratio of 1:4 resulted in the most exceptional mechanical properties, it is reasonable to assert that metal-ligand interaction successfully induced noncovalent cross-linkages in our system.
3.2. FTIR Spectroscopic Analysis {#sec3dot2-polymers-11-01320}
--------------------------------
FTIR was performed to validate metal-ligand interaction, as displayed in [Figure 2](#polymers-11-01320-f002){ref-type="fig"}. In the case of PUU-Py0/1, 1601 cm^−1^ and 1727 cm^−1^ was individually assigned to aromatic skeleton vibration and stretching of urethane carbonyl (C=O) ([Figure 2](#polymers-11-01320-f002){ref-type="fig"}a), while the bending of free pyridine moiety at 1590 cm^−1^ and 1572 cm^−1^ was camouflaged by aromatic skeleton vibration compared to neat 2,6-diaminopyridine ([Figure 2](#polymers-11-01320-f002){ref-type="fig"}b). The peak at 1540 cm^−1^ was assigned to be amide II, i.e., combination of N--H bending and C--N stretching \[[@B27-polymers-11-01320]\]. Peaks at 1694 cm^−1^ and 1645 cm^−1^ were stretching of free and ordered hydrogen-bonded urea carbonyl (C=O) \[[@B4-polymers-11-01320],[@B27-polymers-11-01320],[@B28-polymers-11-01320],[@B29-polymers-11-01320],[@B30-polymers-11-01320],[@B31-polymers-11-01320]\]. By contrast, a new peak emerging at 1621 cm^−1^ was assigned to the shifted bending of pyridine moiety due to its coordination with Zn^2+^ \[[@B32-polymers-11-01320]\]. Moreover, an explicit peak could be observed at 1672 cm^−1^, accompanied by disappearance of peaks at 1694 cm^−1^ and 1645 cm^−1^. This is also indicative of metal-ligand interaction because both free and ordered hydrogen-bonded urea carbonyl groups adjacent to pyridine moieties were significantly affected \[[@B27-polymers-11-01320],[@B31-polymers-11-01320],[@B33-polymers-11-01320],[@B34-polymers-11-01320]\]. Interestingly, urethane carbonyl at 1727 cm^−1^ remained unchanged, indicating that Zn^2+^ ions did not coordinate with urethane groups. [Figure 2](#polymers-11-01320-f002){ref-type="fig"}c presents FTIR spectra of PUU-Py elastomers in solvent of DMF. Peaks at 1694 cm^−1^, 1658 cm^−1^ and 1633 cm^−1^ were ascribed to free, disordered and ordered hydrogen-bonded urea carbonyl \[[@B4-polymers-11-01320],[@B27-polymers-11-01320],[@B31-polymers-11-01320]\]. Since PUU-Py chains in solution were free to perform chain motion, hydrogen bonding could be easily fulfilled so that the peak of free urea carbonyl was the weakest, the disordered hydrogen-bonded was the strongest, and the ordered hydrogen-bonded was medium compared to that of bulk samples. It is observable that the relative intensity of peak at 1633 cm^−1^ weakened as the content of Zn^2+^ increased, which also confirms the interference of hydrogen bonding by metal-ligand interaction.
3.3. UV-vis Spectroscopic Analysis {#sec3dot3-polymers-11-01320}
----------------------------------
Coordination behavior of ligand in PUU-Py with Zn^2+^ was investigated by UV-vis spectra. As shown in [Figure 3](#polymers-11-01320-f003){ref-type="fig"}a, the feature peak of free pyridine moiety at λ = 308 nm diminished as Zn^2+^/pyridine ratio increased, while a new peak emerged at λ = 315 nm, which was associated with ligand-to-metal charge transfer \[[@B35-polymers-11-01320],[@B36-polymers-11-01320]\]. The variation of the two featured peaks versus the Zn^2+^/pyridine ratio illustrated in [Figure 3](#polymers-11-01320-f003){ref-type="fig"}b reveals that complete complexation required a Zn^2+^/pyridine ratio of \~0.5, because the absorbance gap between λ = 308 and 315 nm reached to a nearly constant value and further increasing ratio merely resulted in synchronous change of absorbance. These UV-vis spectra results indicate that coordination took place through pyridine-N atom and Zn^2+^. Furthermore, the adjacent amide-N atom and oxygen atom of urea carbonyl might have participated in coordination.
Absorbance of urea and urethane carbonyl at λ = 290 nm underwent a decrease, instead of shift, as the Zn^2+^/pyridine ratio increased ([Figure 3](#polymers-11-01320-f003){ref-type="fig"}a,b), indicating that the carbonyl groups were affected by coordination. PUU-HMD, a control to PUU-Py using 1,6-hexanediamine as chain-extender, also displayed a decrease of absorbance at feature peaks of urea and urethane carbonyl (λ = 290 nm) and aromatic ring (λ = 269 nm) as shown in [Figure 4](#polymers-11-01320-f004){ref-type="fig"}. The former implies urea and urethane carbonyl probably took part in coordination, which was analogous to coordination of curcumin with Eu^3+^ \[[@B23-polymers-11-01320]\]. The latter was assigned to be B band of aromatic ring of MDI, which could have been affected in absorbance by its possibly coordinated neighboring urea-N atom. With reference to the work of Li et al. that metal ions could coordinate with either oxygen atom or nitrogen atom of amide \[[@B37-polymers-11-01320]\], the coordination configuration remains unclear from UV-vis spectra as shown in [Figure 3](#polymers-11-01320-f003){ref-type="fig"} and [Figure 4](#polymers-11-01320-f004){ref-type="fig"}. However, with regard to FTIR spectra shown in [Figure 2](#polymers-11-01320-f002){ref-type="fig"}a, such uncertainty can be solved. Firstly, stretching of the urethane carbonyl peak at 1727 cm^−1^ and the combination of N--H bending and C--N stretching at 1540 cm^−1^ (amide II) exhibited no shift upon introduction of Zn^2+^. Secondly, both free and ordered hydrogen-bonded urea carbonyl underwent remarkable shifts. This result explicitly indicates that only the oxygen atom of the urea carbonyl participated in coordination with Zn^2+^ according to related analysis of the literature \[[@B32-polymers-11-01320]\]. Thus, the coordination configuration could be illustrated as in [Scheme 2](#polymers-11-01320-sch002){ref-type="scheme"}.
3.4. Analysis of Cyclic Tensile Tests {#sec3dot4-polymers-11-01320}
-------------------------------------
Cyclic tensile tests were performed to further reveal how metal-ligand interaction affects mechanical properties. [Figure 5](#polymers-11-01320-f005){ref-type="fig"}a exhibits curves of the first loading-unloading cycle to compare hysteresis loop. PUU-Py0/1 has apparently the largest area of hysteresis loop (mechanical hysteresis) than the other elastomers. Mechanical hysteresis related to delayed recovery of chain configuration and conformation caused by inter-chain friction. Hydrogen bonding induced by the urea and urethane groups was the major inter-chain friction that constrained re-configuring and re-conforming chains for PUU-Py0/1. For PUU-Py elastomers incorporated with Zn^2+^, coordinative bonds significantly interfered with the hydrogen bonding of urea, as validated by FTIR in [Figure 2](#polymers-11-01320-f002){ref-type="fig"}, thus recovery of the chain configuration and conformation was favored, generating a smaller hysteresis loop. Once Zn^2+^ was introduced into PUU-Py, hysteresis energy sharply decreased to a relatively steady level, regardless of the Zn^2+^/pyridine ratio, as shown in [Figure 5](#polymers-11-01320-f005){ref-type="fig"}b. In the meantime, comparison of residual strain also reflected metal-ligand interaction affecting hydrogen bonding. Firstly, re-formation of coordinative bonds happened in a slower way than hydrogen bonding, according to principle of 'strong means slow' \[[@B13-polymers-11-01320]\]. Secondly, metal-ligand interaction compromised total amount of hydrogen bonds. Hence, the residual strain increased with the introduction of more Zn^2+^ ions ([Figure 5](#polymers-11-01320-f005){ref-type="fig"}b). Therefore, this result confirms coordinative bonds, as shown in [Scheme 2](#polymers-11-01320-sch002){ref-type="scheme"}, in backbone could impose great intervention upon hydrogen bonding. If urea groups locate far away from pyridine moieties, its hydrogen bonding might take place without intervention, then mechanical hysteresis and residual strain would not be notably affected.
3.5. Analysis of Stress-Strain Behavior {#sec3dot5-polymers-11-01320}
---------------------------------------
It seems that hydrogen bonds and coordinative bonds have mutually competing roles in tuning mechanical properties, because they actually have great influence on formation of hard domains. The hydrogen bonded hard segments remained intact, so that micro-phase separation resulted in larger hard domains, while coordinative bonds in PUU-Py1/4 notably interfered with hydrogen bonding of urea groups, which reduced the size of hard domains. Thus, PUU-Py0/1 exhibited higher Young's modulus than PUU-Py1/4, as shown in [Figure 6](#polymers-11-01320-f006){ref-type="fig"}a. The stress-strain curve of PUU-Py1/4 is unique with regard to its large elongation and obvious upturn of slope at \~800%. It can be divided into three parts, as shown in [Figure 6](#polymers-11-01320-f006){ref-type="fig"}b: Regime I occurs at small strain (\<\~55%); Regime II contains a constant slope in a wide range of strain from \~55% to \~540%; and Regime III has a greater constant slope in the range of strain from \~1064% to 1286%. Regime I includes elastic deformation and destruction of hard domains. Regime II presents a linear relationship between stress and strain in a wide strain range, which means the force for extending curled chain and cleaving hydrogen bonds increases in a balanced way. As strain exceeds \~540%, it has to not only cleave hydrogen bonds but has also enforced chain disentanglement. An extrapolational crossing point of slope at 836% is a symbol that coordinative bonds, i.e., noncovalent cross-linkages, start to dissociate. Due to higher binding energy of metal-ligand interaction than hydrogen bonding \[[@B13-polymers-11-01320],[@B14-polymers-11-01320]\], the stress inducing deformation in Regime III remarkably increases with a greater slope. In other words, coordinative bonds significantly restrict extended chains from undergoing mutual sliding. It is worthy of noting that rupture and re-formation of hydrogen bonds and coordinative bonds probably coexist in Regime III because both of them have the capacity of dynamic binding.
3.6. Analysis of Stress Relaxation {#sec3dot6-polymers-11-01320}
----------------------------------
Stress relaxation investigation was performed to unveil roles of hydrogen bonding and metal-ligand interaction, as shown in [Figure 7](#polymers-11-01320-f007){ref-type="fig"}. Referring to [Figure 1](#polymers-11-01320-f001){ref-type="fig"}a and [Figure 6](#polymers-11-01320-f006){ref-type="fig"}b, 100% of strain induced destruction of hard domains by cleaving hydrogen bonding and extending of curled chains of soft segments. Obviously, it needs larger external force to induce deformation and destruction of hard domains than to extend curled chains of soft segments. For PUU-HMD, an extremely slow relaxation process was observed, which indicates that the extending of curled chains of soft segments was insignificant, while hard domains gave rise to elastic deformation as well as destruction. This could be explained by the relatively larger size of the hard domains, because the flexible chain-extender favors in micro-phase separation. For PUU-Py0/1, the pyridine moiety had higher steric hindrance than HMD, so that it had smaller sizes of hard domains. Consequently, the extending of curled chains of soft segments and disorderedly-packed hard segments bore more external force. Once the applied force was removed, recovery of configuration and conformation happened faster than PUU-HMD, especially for soft segments. With incorporating Zn^2+^ into PUU-Py1/6, PUU-Py1/4, PUU-Py1/3, and PUU-Py1/2, the size of the hard domains was further reduced due to the intervention of hydrogen bonding of urea by adjacent metal-ligand interaction, and then curled chains of soft segments and disorderedly-packed hard segments dominantly bore the applied force. Therefore, even though reformation of coordinative bonds might take a longer time \[[@B13-polymers-11-01320]\], recovering configuration and conformation could implement in a short time; for instance, the applied force releases 51%, 57%, 66%, and 76% for PUU-Py1/6, PUU-Py1/4, PUU-Py1/3, and PUU-Py1/2 at 20 min, respectively.
3.7. Mechanism of Toughening {#sec3dot7-polymers-11-01320}
----------------------------
In order to better understand the excellent engineering effect on PUU-Py elastomers, we propose a strengthening and toughening mechanism from the perspective of metal-ligand interaction. Pristine PUU-Py elastomer contains urea doublet hydrogen bonding and urethane singlet hydrogen bonding, and both induce micro-phase separation, leading to closely-packed hard segment domains that act as physical cross-linking junctions to improve tensile strength and robustness \[[@B2-polymers-11-01320],[@B4-polymers-11-01320],[@B27-polymers-11-01320]\]. PUU-Py elastomers with Zn^2+^ introduced display improved mechanical properties, since the binding energy of metal-ligand interaction is several orders higher than that of hydrogen bonding \[[@B14-polymers-11-01320],[@B36-polymers-11-01320]\]. Coordinative bonds not only interfere with urea doublet hydrogen bonding, validated by FITR shift of urea carbonyl ([Figure 2](#polymers-11-01320-f002){ref-type="fig"}), but also form noncovalent cross-linkages between neighboring chains. This is different from metal-ligand interaction that arranges dangling ligand groups along the backbone, because the rigid pyridine moiety in the backbone reduces chain flexibility \[[@B18-polymers-11-01320],[@B38-polymers-11-01320]\].
A molar ratio of 1:6 is not enough to enable adequate pyridine moieties to participate in complexation, so the density of noncovalent cross-linkages is too low, resulting in limited improvement of tensile strength and toughness ([Figure 1](#polymers-11-01320-f001){ref-type="fig"}). PUU-Py1/4 has the best excellent tensile strength and toughness, while PUU-Py1/3 and PUU-Py1/2, Zn^2+^/pyridine ratio exceeding 1:4 does not yield further enhancement of the mechanical properties any more. This result demonstrates that the optimal ratio does not necessarily correspond to the complexation of every ligand, which implies that the presence of vacant ligands facilitates the reformation of noncovalent cross-linkages during stretching and is the key for improving robustness. In the case of pristine PUU-Py, factors that constrain dislocation of chains are mainly chain entanglement and hydrogen bonds induced by the urea and urethane groups. It is easy for external forces to destruct hard segments derived from hydrogen bonding interactions, due to their relatively weak binding energy, accompanied by chain extending, disentanglement, and cleavage in sequence. In the situation of PUU-Py1/4, half of the ligands coordinate with Zn^2+^, generating coordinative bonds between neighboring chains, serving as dynamic cross-linkages ([Scheme 3](#polymers-11-01320-sch003){ref-type="scheme"}). By contrast, an overdose of Zn^2+^ leaves fewer or no vacant pyridine moieties for re-complexation. These noncovalent cross-linkages impose significant restriction on segmental chain motion and dislocation. The curled chains between cross-linkages are able to stretch to an extreme extent upon external force, before complete rupture of coordinative bonds. Upon rupture of coordination, Zn^2+^ is able to rebind with the remaining half of the vacant ligands. Therefore, tensile strength and elongation at break could be significantly improved, simultaneously, resulting in excellent durability and robustness.
4. Conclusions {#sec4-polymers-11-01320}
==============
In summary, the incorporation of pyridine moieties into the backbone of polypropylene glycol-based PUU is able to remarkably improve the robustness of this material, upon being complexed with 1:4 equivalents of Zn^2+^. Specifically, tensile strength, elongation, and toughness of PUU-Py with a Zn^2+^/pyridine ratio of 1:4 could be simultaneously increased to 16.0 MPa, 1286%, and 89.3 MJ/m^3^ with 226%, 29%, and 185% increments in comparison with uncomplexed PUU-Py, respectively. The nitrogen of pyridine moiety and adjacent oxygen of urea participates, in coordination with Zn^2+^, results in dynamic noncovalent cross-linkages. Coordinative bonds impose intervention upon urea hydrogen bonding to compromise micro-phase separation of hard segments, and the mutual sliding of extended chains could be restricted in the process of stretching. Coordinating half of the ligands is found to provide the most effective robustness enhancement, because the remaining half of the ligands are vacant for re-complexation during stretching. This work would provide an effective way of engineering TPU elastomer through molecular design and constructing metal-ligand interaction, which is fundamentally meaningful for industrial applications.
Conceptualization, Y.L.; investigation, A.S. and W.G.; methodology, J.Z. and H.Z.; resources, W.L. and X.L.; writing-original draft preparation, A.S.; writing-review and editing, Y.L. and L.W.
This research was funded by China's Post-doctoral Science Foundation and the Major SpecialFund for Synergistic Innovation Research in Zhengzhou city, grant numbers 2019M652572 and 18XTCX12001, respectively.
The authors declare no conflict of interest.
Figures and Schemes
===================
{#polymers-11-01320-sch001}
{#polymers-11-01320-f001}
{#polymers-11-01320-f002}
{#polymers-11-01320-f003}
{#polymers-11-01320-f004}
{#polymers-11-01320-sch002}
{#polymers-11-01320-f005}
{#polymers-11-01320-f006}
{#polymers-11-01320-f007}
{#polymers-11-01320-sch003}
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .274}
| {
"pile_set_name": "PubMed Central"
} |
{#sp1 .403}
{#sp2 .404}
| {
"pile_set_name": "PubMed Central"
} |
Background
==========
Currently, the prevalence of diabetes mellitus is increasing worldwide, and it is now considered to be a pandemic non-communicable disease. Up to 95% of patients suffer from type 2 diabetes mellitus (T2DM), which is characterized by insulin resistance and β-cell dysfunction \[[@b1-medscimonit-24-3293]\]. Currently, several hypoglycemic agents have been developed for the treatment of T2DM, including insulin analogs, insulin-sensitizing agents, dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium glucose cotransporter-2 (SLG2) inhibitors, and glucagon-like peptide-1 (GLP-1) receptor agonists \[[@b2-medscimonit-24-3293]\]. Previous studies on the treatment of diabetes mellitus have shown that long-acting GLP-1 receptor agonists protect pancreatic β-cell function \[[@b2-medscimonit-24-3293],[@b3-medscimonit-24-3293]\]. Furthermore, GLP-1 receptor agonists and insulin analogs have shown hypoglycemic effects \[[@b3-medscimonit-24-3293]\]. Short-term intensive insulin therapy can improve and recover β-cell function in newly diagnosed T2DM \[[@b4-medscimonit-24-3293],[@b5-medscimonit-24-3293]\]. Currently, both GLP-1 receptor agonists and insulin analogs have been widely used to control blood glucose.
GLP-1 is an incretin hormone secreted by intestinal cells in response to orally ingested food, which reduces glucagon secretion, delays gastric emptying, reduces appetite, and promotes pancreatic β-cell proliferation and differentiation. The spectrum of effects in diabetes mellitus gives GLP-1 a unique physiological and pharmacological profile that has been the basis for the development of a hypoglycemic agent for T2DM. However, GLP-1, which is a substrate of dipeptidyl peptidase IV (DPP-IV), is rapidly degraded by DPP-IV and cleared from the plasma, giving the biologically active peptide a half-life of less than two minutes after intravenous administration, and of between one and two hours after subcutaneous administration. Therefore, native GLP-1 is not optimal for therapeutic use because of its pharmacokinetic profile, but the development of GLP-1 receptor agonists provide a new approach to the treatment of T2DM \[[@b6-medscimonit-24-3293]\].
Previous studies have shown that GLP-1 receptor agonists have pleiotropic effects that may enhance their therapeutic effect in patients with T2DM, including glycemic control, improving lipid metabolism, lowering blood pressure, reducing cardiovascular risk factors, and reducing the levels of inflammatory mediators \[[@b7-medscimonit-24-3293],[@b8-medscimonit-24-3293]\]. Liraglutide is a long-acting GLP-1 receptor agonist, which permits 24-hour glycemic control with a once-daily injection \[[@b3-medscimonit-24-3293],[@b9-medscimonit-24-3293]\]. Although the effects of the GLP-1 receptor agonist, liraglutide, and the insulin analog, insulin glargine, on glycemic control have been studied clinically, it remains unclear whether there are differences between the insulin analog, insulin glargine and the GLP-1 receptor agonist, liraglutide, regarding the effects on glucose control and preservation of β-cell function *in vivo*.
The aim of this study was to compare the effects of liraglutide, a long-acting GLP-1 receptor agonist, and insulin glargine, a long-acting insulin analog, on glycemic control and pancreatic β-cell function in db/db mice.
Material and Methods
====================
Treatment of animals
--------------------
Seven-week-old male db/db mice (between 35--40 gm) were obtained from the National Resource Center for Mutant Mice (Nanjing, China, SCXK-2010-0001). To minimize the stress due to handling, all animals were accustomed to blood sampling and dosing procedures for one week before the start of the experiments. Liraglutide was obtained from Novo Nordisk, Denmark.
Eight-week-old male db/db mice (n=40) were divided into five groups: the vehicle-treated group (VG), treated with phosphate-buffered saline (PBS) injection (n=8); the insulin glargine-treated group (GG) (dose, 450 mg/kg) (n=8), the low-dose liraglutide-treated group (LLG) (dose, 75 μg/kg) (n=8), the mid-dose liraglutide-treated group (MLG) (150 μg/kg) (n=8), and the high-dose liraglutide-treated group (HLG) (300 μg/kg) (n=8). Liraglutide and insulin glargine were given once daily via subcutaneous injection to mice between 8--16 weeks-of-age. All animals were housed in an environmentally controlled room at 25°C with a 12-hour light and 12-hour darkness cycle, and fed with a normal diet and with free access to water throughout the experimental period. All experimental animal procedures were approved by the Animal Care and Use Committee of the Second Affiliated Hospital of Nanchang University and were conducted in compliance with Guide of the Care and Use of Laboratory Animals, according to the National Institutes of Health (NIH) publication (No. 86-23, revised 1996).
Assays for metabolic components
-------------------------------
Random blood glucose levels were measured using the Accu-Check Advantage glucose meter (Roche Diagnostics Ltd) with blood obtained from a tail vein. Serum triacylglycerol was measured using a mouse triacylglycerol enzyme-linked immunosorbent assay kit (Biovision, USA) according to the manufacturer's protocol. Body weight, blood glucose, and serum triacylglycerol levels were measured on days 1, 5, 9, 11, 15, 22, 29, 36 and 43 during the six-week study period.
Intraperitoneal glucose tolerance test (IPGTT) and serum C-peptide levels
-------------------------------------------------------------------------
Intraperitoneal glucose tolerance test (IPGTT) was performed at the end of the treatment period. Mice were injected intraperitoneally with glucose (0.5 gm/kg body weight) after an overnight fast, and blood samples were taken from tail veins at 0, 30, 60, and 120 minutes for measurement of blood glucose. The area under the curve (AUC) of IPGTT was calculated by the trapezoid formula. Serum C-peptide levels, at 120 minutes following glucose injection, were measured using a mouse C-peptide enzyme-linked immunosorbent assay (ELISA) (ALPCO, Salem, NH, USA).
Pancreas to body weight coefficient
-----------------------------------
At the end of the treatment period, the mice were sacrificed by decapitation during CO~2~ anesthesia, and the pancreas was removed and weighed for each mouse. An index of the pancreas to body weight was calculated.
RNA extraction and real-time polymerase chain reaction (RT-PCR)
---------------------------------------------------------------
Total RNA was extracted from pancreas using RNeasy Mini Kit (Qiagen, Valencia, CA, USA). Real-time polymerase chain reaction (RT-PCR) was performed using a Reverse Transcription System Kit (Invitrogen, USA) in a LightCycler 480 high-throughput PCR platform (Roche). The primers were designed using Applied Biosystems (Foster City, CA, USA) Primer Express design software. The sequences of the primers are listed in [Table 1](#t1-medscimonit-24-3293){ref-type="table"}. RT-PCR was performed using the LightCycler 480 System (Roche) using SYBR Green Supermix (Takara). Data were normalized to the housekeeping gene, β-Actin.
Statistical analysis
--------------------
All data were presented as the mean ± standard error of the mean (SEM). Statistical analysis was performed with one-way analysis of variance (ANOVA) with Bonferroni pairwise corrections and the Student's t-test. P-values \<0.05 was considered statistically significant. Statistical analysis was performed using SPSS version 11.0 software (SPSS, Inc, Chicago, IL, USA).
Results
=======
Effect of liraglutide and insulin glargine on body weight and blood glucose in db/db mice
-----------------------------------------------------------------------------------------
The db/db mice (n=40) were divided into five groups: the vehicle-treated group (VG) (n=8); the insulin glargine-treated group (GG) (n=8), the low-dose liraglutide-treated group (LLG) (n=8), the mid-dose liraglutide-treated group (MLG) (n=8), and the high-dose liraglutide-treated group (HLG) (n=8), treated from 8--14 weeks-of-age. The body weight of the db/db mice was significantly reduced following different doses of liraglutide injection, with a maximal effect in the HLG compared with the VG at day 36 (P\<0.05) and at day 43 (P\<0.01) ([Figure 1A](#f1-medscimonit-24-3293){ref-type="fig"}). Also, the body weight of the HLG was significantly lower than the GG at day 36 and day 43 (P\<0.005 and P\<0.001, respectively).
The blood glucose levels of the mice in the GG, the LLG, the MLG, and the HLG were significantly reduced when compared with the VG (P\<0.05). Also, blood glucose levels of the mice in the HLG were significantly reduced compared with the GG at days 22--43 (P\<0.05) ([Figure 1B](#f1-medscimonit-24-3293){ref-type="fig"}). However, no statistical differences were observed in serum triacylglycerol levels between the treatment groups ([Figure 1C](#f1-medscimonit-24-3293){ref-type="fig"}). These results showed that liraglutide and insulin glargine treatment effectively controlled glucose levels rather than lipid metabolism in db/db mice.
High-dose liraglutide treatment had a more effective anti-diabetic effect than insulin glargine in db/db mice
-------------------------------------------------------------------------------------------------------------
Glucose tolerance was significantly improved in the GG, the LLG, the MLG and the HLG compared with the VG at the end of the six-week treatment period in db/db mice (P\<0.001) ([Figure 2A](#f2-medscimonit-24-3293){ref-type="fig"}). The reduction of blood glucose levels after intraperitoneal glucose tolerance test (IPGTT) was significantly greater in the HLG compared with the GG (P\<0.001) ([Figure 2A](#f2-medscimonit-24-3293){ref-type="fig"}). The HLG showed receiver operating characteristic (ROC) maximum area under the curve (AUC) of glucose tolerance (P\<0.001) ([Figure 2B](#f2-medscimonit-24-3293){ref-type="fig"}). Also, the serum C-peptide of the db/db mice at the end of the IPGTT, performed to evaluate pancreatic β-cell function after glucose stimulation, showed that the HLG had a significantly increased serum C-peptide compared with the GG, but the MLG and the LLG showed no difference when compared with the GG (P\<0.05) ([Figure 2C](#f2-medscimonit-24-3293){ref-type="fig"}). Following the IPGTT, mice were sacrificed and the pancreas was weighed for each db/db mouse, and compared with the VG; the HLG showed a significant increase in the pancreas to body coefficient (P\<0.05 and P\<0.01, respectively), and this effect was not observed in the GG ([Figure 2D](#f2-medscimonit-24-3293){ref-type="fig"}). These findings may indicate that the high serum C-peptide and increase in pancreas weight might account for the improved glucose tolerance in the HLG.
High-dose liraglutide treatment upregulated pancreatic β-cell functional genes more than insulin glargine treatment in db/db mice
---------------------------------------------------------------------------------------------------------------------------------
The expression levels of the insulin gene, *INS1*, in the pancreas of db/db mice measured by reverse transcription polymerase chain reaction (RT-PCR) was greatest in the HLG than MLG, the LLG, and the GG. Therefore, an improvement in insulin and C-peptide secretion may be a consequence of changes in insulin mRNA expression levels. Regulation of the expression of the insulin gene,*INS1*, was likely to have been due to the expression of a set of transcriptional factors that are typical of mature β-cells including pancreatic and duodenal homeobox 1 (Pdx1), paired box 4 (Pax4) and paired box 6 (Pax6). The expression of mRNA of these transcription factors was increased in a dose-dependent manner in all groups of mice treated with liraglutide, with a significant increase in expression in the HLG compared with the GG (P\<0.05) ([Figure 3](#f3-medscimonit-24-3293){ref-type="fig"}).
Discussion
==========
Since 2005, the long-acting glucagon-like peptide-1 (GLP-1) receptor agonist, liraglutide, has been approved for the treatment of patients with type 2 diabetes mellitus (T2DM). Epidemiological studies and animal experiments have shown that GLP-1 receptor agonists have pleiotropic effects that may enhance their therapeutic effect in patients with T2DM, including the promotion of insulin secretion, the inhibition of gastric emptying, reduction in body weight, the increase in pancreatic β-cell mass, and the reduction in glucagon production \[[@b3-medscimonit-24-3293],[@b11-medscimonit-24-3293]\]. Insulin glargine, a long-acting insulin analog, has been widely used in the treatment of T2DM \[[@b12-medscimonit-24-3293],[@b13-medscimonit-24-3293]\]. Comparisons between the effects of GLP-1 receptor agonists with long-acting insulin analogs on the glucose control can assist in treatment decisions for patients with T2DM in clinical practice \[[@b14-medscimonit-24-3293],[@b15-medscimonit-24-3293]\]. However, the effects on glucose control and protection of pancreatic β-cell functions by different GLP-1 receptor agonists and insulin glargine have not been well studied in diabetic db/db mice, which is widely used as an animal model for T2DM.
In the present study, the db/db mouse model was used in which the development of progressive diabetes is characterized by a reduced proliferation and increased apoptosis in pancreatic β-cells \[[@b16-medscimonit-24-3293]\]. In this comparative study of treatment of the db/db mouse, the insulin glargine-treated group (GG) (dose, 450 μg/kg), was compared with the low-dose liraglutide-treated group (LLG) (dose, 75 μg/kg), the mid-dose liraglutide-treated group (MLG) (150 μg/kg), and the high-dose liraglutide-treated group (HLG) (300 μg/kg) \[[@b17-medscimonit-24-3293],[@b18-medscimonit-24-3293]\]. The findings were that at the end of the six-week treatment period of the study, liraglutide improved glycemic control in a dose-dependent manner in db/db mice, which was consistent with previously published studies \[[@b19-medscimonit-24-3293],[@b20-medscimonit-24-3293]\]. Also, increased levels of serum C-peptide, a marker of insulin production by the pancreas, was associated with insulin glargine treatment (the GG) and was dose-dependent on liraglutide treatment (the HGL), compared with the vehicle treatment, with these changes reflected by the expression of the *INS1* gene. The findings of this study indicated that liraglutide and insulin glargine treatment could enhance pancreatic β-cell function, and the effects on the β-cell function of high-dose liraglutide were significantly greater than insulin glargine and were dose-dependent.
Previously published studies have shown that insulin therapy could significantly increase the body weight of db/db mice, possibly leading to insulin resistance \[[@b21-medscimonit-24-3293]\]. In the present study, insulin glargine treatment reduced the body weight of db/db mice, possibly due to the short period of insulin glargine treatment. However, from the findings of this study, it was not possible to determine whether improving glycemic control by high-dose liraglutide and insulin glargine in db/db mice provided pancreatic β-cell rest resulting in the stimulation of insulin secretion. Therefore, future functional experiments are suggested to address these questions.
Pancreatic and duodenal homeobox 1 (Pdx1) and paired box 6 (Pax6) are crucial for pancreatic β-cell function through transcriptional control of key genes that are involved in insulin biosynthesis and secretion \[[@b22-medscimonit-24-3293]--[@b24-medscimonit-24-3293]\]. In previous studies, overexpression of pancreas-specific transcription factor, PDX1 (also known as insulin promoter factor 1), has been shown to induce the differentiation of non-endocrine pancreatic cells into β-cells \[[@b25-medscimonit-24-3293]\]. Another key transcription factor, paired box 4 (Pax4) is essential in the generation of insulin-producing pancreatic β-cells, and overexpression of Pax4 in adult islets stimulates β-cell proliferation and increases their resistance to apoptosis \[[@b26-medscimonit-24-3293],[@b27-medscimonit-24-3293]\]. In the present study, reverse transcription polymerase chain reaction (RT-PCR) was used to determine mRNA expression levels of key transcriptional factors and showed that Pax4 and also pancreatic and duodenal homeobox 1 (Pdx1) and paired box 6 (Pax6) were significantly upregulated in a dose-dependent manner in liraglutide-treated db/db mice. The expression levels of these three transcription factors were significantly increased in the HLG compared with the GG, indicating that high-dose liraglutide regulated glucose metabolism and protected β-cell function by increasing the mRNA expression levels of Pax4, Pax6, and Pdx1, inducing insulin secretion by promoting pancreatic β-cell proliferation or differentiation. These findings are supported by previous studies that have described a stimulatory effect on the pancreatic β-cell mass of liraglutide in diabetic db/db mice \[[@b28-medscimonit-24-3293],[@b29-medscimonit-24-3293]\]. A limitation of the present study was the lack of the morphologic evaluation of the mouse pancreatic islets and the evaluation of pancreatic β-cell proliferation and differentiation in db/db mice. Future functional and morphological studies in this mouse model of diabetes are recommended to include morphological studies on the pancreatic β-cells.
Conclusions
===========
The findings of this study were that liraglutide acted in a dose-dependent manner on glycemic control of db/db mice, and was more effective than insulin glargine when administered at a high dose. Compared with insulin glargine treatment, and low-dose liraglutide, high-dose liraglutide treatment improved glucose tolerance and β-cell function of db/db mice. The findings in this mouse model of type 2 diabetes mellitus (T2DM) supports the view that treatment with a long-acting glucagon-like peptide-1 (GLP-1) receptor agonist may be a therapeutic approach to preserve pancreatic β-cell function in T2DM.
**Conflicts of interest**
None.
**Source of support:** National Natural Science Foundation of China (No. 81260132), Young Scientist Training Target Program of Jiangxi Province (No. 20142BCB23026) and Young Scientist Research Project of the Second Affiliated Hospital of Guangzhou Medical University (No. 2016A05)
{#f1-medscimonit-24-3293}
{#f2-medscimonit-24-3293}
{#f3-medscimonit-24-3293}
######
Oligonucleotide sequences for quantitative reverse transcription- polymerase chain reaction (qRT-PCR) analysis.
Gene symbol Forward primer Reverse primer
------------- ------------------------ -----------------------
β-actin GCAGAAGGAGATTACTGCTCT GCTGATCCACATCTGCTGGAA
Pax4 GCCTATCTCCAACCCTACTGG GCCAGGCAAATTCCACATA
Pax6 TACCAGTGTCTACCAGCCAAT TGCACGAGTATGAGGAGGTCT
Ins1 GCTTCTTCTACACACCCATGTC AGCACTGATCTACAATGCCAC
Pdx1 CCCCAGTTTACAAGCTCGCT CTCGGTTCCATTCGGGAAAGG
Pax4 -- paired box gene 4; Pax6 -- paired box gene 6; Pdx1 -- pancreatic and duodenal homeobox 1; Ins1 -- Insulin1.
[^1]: Study Design
[^2]: Data Collection
[^3]: Statistical Analysis
[^4]: Data Interpretation
[^5]: Manuscript Preparation
[^6]: Literature Search
[^7]: Funds Collection
[^8]: These authors contributed equally to this work
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Plasmonics is a promising counterpart of nanophotonics which has witnessed major progresses in the control and manipulation of electromagnetic (EM) fields at extreme-subwavelength scales \[[@B1]--[@B3]\]. Central to all studies is the surface plasmon\'s unique capability to harvest, process, and concentrate light and convert it into energetic near-fields, thermal power, and hot carriers \[[@B4]--[@B7]\]. All these remarkable applications and developments of subsequent optical devices have been realized using resonant plasmonic nanostructures across the ultraviolet (UV) to the far-infrared spectra. Resonances in nanophotonics and nanoplasmonics are the fundamental phenomena that play a critical role in defining the operating mechanism, quality, and performance of the tailored tools. In recent years, several types of radiating and nonradiating resonances and spectral effects have been successfully excited and introduced by scientists such as Fano resonances \[[@B8]--[@B10]\], electromagnetically induced transparency (EIT) \[[@B11]--[@B13]\], Bormann and Kerker effects \[[@B14]--[@B20]\], toroidal multipoles \[[@B21]--[@B24]\], anapoles \[[@B23], [@B25]--[@B27]\], and charge transfer plasmons (CTPs) \[[@B28]--[@B30]\]. Excluding the later instance (CTP), other resonances can be induced based on robust coupling of optically driven modes between proximal metallic nanoparticles (NPs) with sharp protrusions in the near-field regime. Conversely, for the CTP, the charge transfer in particle plasmon resonances has been reported in (1) conductively linked NPs \[[@B31]--[@B35]\], (2) reversible electrochemical compounds \[[@B36]\], and (3) metallic systems with subnanometer atomic openings (Fowler-Nordheim (FN) and quantum tunneling principles) \[[@B37]--[@B40]\].
There have been ongoing theoretical and experimental advancements to understand the possibility of direct dynamic charge transport between plasmonic objects in the absence of capacitive coupling between plasmonic particles. Relatively, researchers have shown the possibility of plasmon-driven charge transfer process in DNA-mediated metallic NP dimers \[[@B31], [@B34]\]. In this context, one can fill the gap between metallic NPs with DNA (as a scaffold for the growth of the NPs within the gap), and by increasing the metallic NP concentration, the transition from capacitive to conductive coupling can be formed. Besides, for the conductively bridged NPs, it is confirmed that due to the shuttling of photoexcited electrons across the junction, the CTP spectral feature appears in the lower energies far from the superradiant dipole moment \[[@B41]\]. On the other hand, as an alternative route to this method, it is well-accepted that particles with atomic gaps in between (below \~0.5 nm) are able to sustain CTPs through molecular quantum tunneling principle \[[@B40], [@B42]\]. Theoretically, the electron tunneling across an atomic-scale gap at optical frequencies cannot be explained using classical electrodynamic theory and requires quantum mechanical description.
In spite of possessing interesting properties, CTPs are inherently suffering from lack of spectral tunability. Recent demonstrations that the functionality of CTP feature can be optimized by integrating bulk metallic systems with thermally and electronically controllable materials show why tunable CTP spectral features are important for implementing novel and advanced plasmonic tools \[[@B36], [@B43]--[@B46]\]. More precisely, this shortcoming has successfully been addressed by combining plasmonic structures with, for example, phase-change materials (PCMs) \[[@B43], [@B44], [@B47]\], graphene \[[@B45], [@B46]\], and thermally tunable substances (i.e., InSb).
In this review, we summarize recently introduced mechanisms that underpin the electron transition between plasmonic particles and trending applications of functional CTPs. One of the goals of this article is to present a comprehensive depiction of various modalities developed for the excitation of tunable CTPs. Besides, we explain the state-of-the-art approaches that have been proposed for enhancing the tunability of CTPs and the use of this spectral phenomenon in designing practical photonic tools such as modulators, switches, and nonlinear harmonic signal generators. Next, we illustrate how a careful integration of subwavelength plasmonic resonators with optically, thermally, and electronically functional materials leads to the emergence of multifunctional nanooptoelectronic and terahertz (THz) devices.
2. Charge Transfer Plasmons in Conductively Bridged Particles and Resonators {#sec2}
============================================================================
Direct manipulation of charges *via* redistribution and transfer has been introduced as a simple approach to excite CTP spectral features and studied both theoretically and experimentally in nanoscale plasmonic systems \[[@B31]--[@B36]\]. Possessing a different principle compared to capacitive resonance interference, the direct charge transfer between plasmonic resonators and particles enables tailoring advanced plasmonic tools. Additionally, this direct transition of charges between the metallic structures delivers exceptional abilities to actively control the charge shuttling by varying the intensity of the incoming beam, which could be used for the development of near-infrared and THz plasmonic devices.
A remarkable example for the excitation of CTP mode is observed in a two-member dimer consisting of a pair of proximal metallic NPs that are connected with a conductive junction (insets in [Figure 1(a)](#fig1){ref-type="fig"}) \[[@B32]\]. In the method developed by Wen et al. \[[@B32]\], a dimer antenna with a conductive junction has been considered as the intermediate case between a conventional metallic dimer with capacitive opening and a nanorod. This allows to understand the formation of new spectral features due to the presence of a metallic bridge between NPs. [Figure 1(a)](#fig1){ref-type="fig"} demonstrates the numerically obtained scattering efficiencies of three different structures with judiciously defined geometries that are stated in the figure caption. Under *p*-polarized beam illumination, in the dimer limit, the capacitive coupling between the two nanodisks leads to strong hybridization of plasmons and excitation of a superradiant mode around ∼1.95 eV. On the other hand, bridging the dimer with a conductive nanowire gives rise to the formation of two peaks in the spectrum correlating with the dipolar mode (at ∼2.1 eV) and a narrow CTP mode at lower energies far from the classical dipole (around ∼0.96 eV). Lastly, for the plasmonic nanorod, compared to the dimer structure, the dipolar resonance is red-shifted to ∼1.3 eV. As can be seen in the charge plots in [Figure 1(b)](#fig1){ref-type="fig"}, the fundamental difference between the CTP extreme and dipolar mode is the distinguished oscillation of the electric current density in the junction of the conductive nanowire-bridged plasmonic dimer, while there is no such a feature in both dimer and nanorod structures. Moreover, the charge distribution at the energy of CTP explicitly represents the splitting of charges across the structure (see [Figure 1(b)](#fig1){ref-type="fig"}, (2)).
To further study the role of the conductive junction on the excitation of CTP, Wen and teammates have shown that the junction geometry and conductance strictly determine the position and intensity of the CTP feature. To this end, the researchers employed CTP-resonant nanostructures based on aluminum and gold substances. Considering a nanowire with the frequency-dependent AC conductivity of *σ*(*ω*), and specific length (*l*), width (*w*), and thickness (*t*), the corresponding conductance can be written as $$\begin{matrix}
{G = \sigma\left( \omega \right)\frac{wt}{l}.} \\
\end{matrix}$$
As a critical parameter, in [Figure 1(c)](#fig1){ref-type="fig"}, the junction conductance as a function of nanowire width is plotted, numerically and experimentally. Obviously, by increasing the width of the junction, the relative conductance increases slightly for gold structure, while this value sharply increases for the aluminum nanowire. Here, for structures with the equivalent geometries of the bridging nanowire, aluminum structures hold a higher junction conductance than gold structures. Further analysis for the junction properties and plotting the CTP position as a function of junction conductance reveal that the CTP mode exhibits a dramatic sensitivity to the junction conductance in the small conductance limit ([Figure 1(d)](#fig1){ref-type="fig"}). However, this sensitivity becomes strikingly less as the conductance increases beyond a specific value (\~50) for all the studied cases.
Recent studies have also shown that CTP mode can be viewed as a spectral feature at much lower energies (e.g., far infrared, THz) when the coupling between neighbor resonators is in the weak regime. As a leading work in the THz regime, Ahmadivand et al. \[[@B33]\] have represented the transition from capacitive coupling to direct charge transfer using a symmetric cluster of V-shaped metallic microblocks. The insets in Figures [1(e)](#fig1){ref-type="fig"}--[1(g)](#fig1){ref-type="fig"} illustrate the scanning electron microscopy (SEM) images of the unit cell, where the diameter of the central disk gradually increases. As obvious in the normalized transmission amplitude (NTA) profile ([Figure 1(e)](#fig1){ref-type="fig"}), the capacitive coupling plays a fundamental role in the emergence of two pronounced minima in the spectral response. The deeper minimum around 1.22 THz correlates with the dipolar resonant mode, while the quadrupole mode appears as a weaker dip around 1.85 THz. Here, the middle disk enhances the strength of the induced multipolar modes via the dipole-dipole interaction \[[@B48]\]. To continue, increasing the diameter of the disk monotonically and providing a touching regime between the disk and V-shaped assemblies lead to the elimination of the quadrupolar moment and emerging of the CTP dip around \~0.5 THz ([Figure 1(f)](#fig1){ref-type="fig"}). Due to the formation of insignificant touching spots between the blocks and the central disk, one can expect the direct flow of charges across the unit cell similar to the optical systems and quantum transitions. Further increases in the diameter of the central disk to realize the overlapping regime give rise to a damping in the energy of both dipolar and CTP modes projected from the spectral characteristics of the entire unit cell ([Figure 1(g)](#fig1){ref-type="fig"}). The cross-sectional electric-field (E-field) intensity along the unit cell axis helps to perceive the role of the conductive disk in the excitation of CTP feature. As plotted in Figures [1(h)](#fig1){ref-type="fig"}--[1(j)](#fig1){ref-type="fig"}, by inserting and increasing the size of the disk, the E-field confinement reduces dramatically owing to losing the capacitive coupling between the V-shaped blocks. Additionally, the overlapping areas between the V-shape and central resonators allow to induce sharper and boosted plasmonic dipolar and CTP modes in the THz limit. The geometry of the merging regions enables the shuttling of charges between the linked V-shaped pixels of a standalone assembly. This can be better understood by defining the resistance (*R*) of the disk as a function of the junction geometry \[[@B49]\]: $$\begin{matrix}
{R = \frac{2}{\pi t\sigma\left( \omega \right)}\ln\left( \frac{2d}{\delta} \right),} \\
\end{matrix}$$where *σ*(*ω*) is the frequency-dependent AC conductivity, *d* is the length between junctions, and *δ* is the contact width at the junctions due to overlapping. Moreover, according to Ohm\'s law, the conductance is defined by \[[@B50]\] $$\begin{matrix}
{\sigma = \frac{n_{m}e^{2}\tau}{m_{e}},} \\
\end{matrix}$$where *n*~*m*~ is the electron density of metal, *e* is the elementary charge, *τ* is the mean-time between collisions, and *m*~*e*~ is the mass of electron. Enlarging the diameter of the overlapping disk lengthens the distance between junctions, and the corresponding resistance increases accordingly ([Figure 1(m)](#fig1){ref-type="fig"}). Although the dynamic charges would be able to transit across the trail, the dissipative losses can be significant because of the inherent disk resistance. When the incident THz beam is resonant with the spectral line shape, the fairly strong confinement of charges across the lossy metallic resonators leads to a drastic electron decay. Thus, by extending the size of the middle disk, the induced dipolar and CTP modes shift towards the lower energies due to the enhanced dissipative absorption losses \[[@B51], [@B52]\].
3. Charge Transfer Plasmons in Nonlocal and Quantum Regimes {#sec3}
===========================================================
As discussed [in Introduction](#sec1){ref-type="sec"}, the general idea to model the light-matter interactions and the resulting electromagnetic field distribution is mainly based on solving coherent charge density oscillations using classical electromagnetic theory by applying Maxwell\'s equations \[[@B4]\]. By shrinking the interparticle distance down to nanoscales, localization and intensity of the incident electric field can be enhanced \[[@B53]--[@B55]\]. Nevertheless, when the system requires to operate beyond the nanometer gap regime (i.e., subnanometer openings) or contains a quantum topology (e.g., quantum dots, molecules, or atoms), nonlocal screening effects (due to quantum nature of electron) modify the plasmonic response of structure \[[@B56]\]. In this limit, quantum mechanical effects (e.g., collective quantum electron tunneling, nonlocal transport) occur and the whole system must be described under nonlocal conditions \[[@B57]--[@B59]\], by implementing advanced theoretical approaches \[[@B60]--[@B62]\]. This enables the calculation of the confinement of the induced surface charges appropriately \[[@B63]\]. Such computations in quantum systems have successfully been done using a quantum-corrected model (QCM) \[[@B42]\]. Basically, this concept is an updated version of the time-dependent density-functional theory (TDDFT), which combines the classical electromagnetic framework with the collective quantum electron tunneling \[[@B38]\]. As declared in Savage et al. \[[@B64]\], when the interparticle distance is *d* ≥ 0.4 nm, one can treat the system using classical Maxwell\'s descriptions to understand the plasmonic interactions. For *d* \< 0.4 nm, classical formulation starts to diverge as a result of the increased rate of critical electron tunneling between proximally located NP surfaces (\~ 0.3 nm), where the plasmon interactions start to get into the quantum regime. For *d* \< 0.3 nm, the quantum tunneling dominates the spectral response of the entire system. By considering a rectangular barrier, this phenomenon can be predicted as the following: ${\overset{\sim}{\, d}}_{QR} = \ln\left( {3q\alpha\lambda/2\pi} \right)/2q$, where ${\overset{\sim}{d}}_{QR}$ is the critical gap size, *q* is the semiclassical electron tunneling wavenumber, *λ* is the optical plasmon wavelength, and *α* is the structural constant. More realistically, the addition of "coherent quantum transport" parameter into the formula given above boosts the tunneling rate, which gives rise to the quantum-tunneling CTPs (more detailed information about the calculations can be found in Ref. \[[@B64]\]). In the upcoming part, we briefly review the recent advances in the quantum-mechanical charge shuttling process in nearly touching metallic NPs.
As a pioneer work, Wu et al. \[[@B39]\] demonstrated the first theoretical explanation of plasmon resonances between extremely close NPs due to FN tunneling. In spite of possessing certain conditions to satisfy the direct tunneling (e.g., subnanometer gap), this approach provides a novel outlook to induce CTP modes for a different set of gap sizes by applying high-intensity illumination. From the active nanophotonic device perspective, the proposed concept is significant owing to its ability to control the charge shuttling process by changing the intensity of the incident light. As discussed previously, approaching two NPs to each other (nearly touching regime) and the presence of a metallic path between them lead to transition of oscillating electric charges across the nanoplatforms, in addition to the dipolar mode of the dimer system. As the conductive path becomes narrower, the excited CTP resonance shifts to the longer wavelength and several higher-order modes (e.g., hybridized dipolar and quadrupolar modes) appear \[[@B65]--[@B67]\]. Nevertheless, when the conductive bridge turns out to a \~0.4-0.5 nm gap, the CTP mode disappears, meaning that a quantum mechanical framework should be taking into account to model possible electron tunneling effects. To this end, in [Figure 2(a)](#fig2){ref-type="fig"}, the authors demonstrated possible tunneling mechanisms as direct tunneling (effective for small interparticle distance) and FN tunneling (dominant when a strong electric field is applied) in a simple schematic. For the direct tunneling, electrons tunnel through the square barrier (from A to C); while for the FN tunneling, electrons first tunnel through the triangular barrier (from A to B), then transport to C. In general, the electrons exist in B are called "space charges." Hence, this process can also be termed as space charge-based charge transfer. Moreover, the researchers here utilized the microscopic version of Ohm\'s law (see Equation ([2](#EEq1){ref-type="disp-formula"})) to specify the tunneling electrons in the gap \[[@B68]\]. It is important to note that for the calculation of tunneling electron density in the gap, Wu and colleagues applied the Kohn-Sham density functional theory, to model exchange correlation and Coulomb interactions between electrons \[[@B69], [@B70]\]. Further, they successfully explained the resulting CTP mode in light of the obtained tunneling electron density (*n*~gap~) and gap conductivity (*σ*~gap~), which are determined based on a numerical methodology developed by the authors \[[@B39]\]. To briefly discuss, *ε*(*ω*) = 1 + *i*(*σ*~gap~/*ωε*~0~) is utilized as a part of QCM to derive the optical response of the system. As demonstrated in [Figure 2(b)](#fig2){ref-type="fig"}, when *σ*~gap~ is larger than 1.145 × 10^5^ S/m, a CTP mode is emerged. For the "tunneling" part indicated in [Figure 2(b)](#fig2){ref-type="fig"}, a large plasmon-enhanced electric field (10^10^ V/m) (or "gap field") is required to bring enough electron in the space charge area to induce the CTP resonant mode. In the following, in [Figure 2(c)](#fig2){ref-type="fig"}, to better understand the physical mechanism behind the process, *σ*~gap~ and *n*~gap~ are compared for the gap sizes of 0.4 nm, 0.6 nm, and 0.8 nm in three different gap field regimes. In the direct tunneling limit, the tunneling barrier can be controlled through the charge potential when the gap field is small, which means *σ*~gap~ primarily rely on the gap length. As expected, when the gap is 0.4 nm, electrons can straightly tunnel along the narrower barrier (see [Figure 2(d)](#fig2){ref-type="fig"}, (i)), with the increased tunneling probability of electrons, but for larger gaps (e.g., 0.8 nm), the charge flow process could not be sustained because of the wider tunneling barrier. To address this issue, the employed electric field intensity can be amplified (\~10^10^ V/m) to start the FN tunneling procedure, in which the tunneling barrier is reshaped as plotted in [Figure 2(d)](#fig2){ref-type="fig"}, (ii). In this regime, *σ*~gap~ and *n*~gap~ are very responsive to the intensity of the external illumination, meaning that these parameters can be easily modified depending on the applied electric field intensity. For instance, when the field intensity is altered from 10^9^ V/m to 10^11^ V/m, *σ*~gap~ is increased from 1 to 10^7^ S/m. Any further increase in the field intensity would lower the level of barrier where all tunneling electrons can stay over the barrier (see [Figure 2(d)](#fig2){ref-type="fig"}, (iii)). In this regime, *σ*~gap~ is not dependent on the applied field intensity and gap size, as illustrated in [Figure 2(c)](#fig2){ref-type="fig"}. This results in perfect transmission of electrons through the gap and extremely large *σ*~gap~ to retain the CTP mode for the studied gap dimensions. Lastly, in Figures [2(e)](#fig2){ref-type="fig"}--[2(g)](#fig2){ref-type="fig"}, simulated extinction spectra for different gap size (*d*), *σ*~gap~, and the gap field are presented to prove the tunability of the CTP peak. With the gap size of 0.4 nm, the CTP mode always arise either *via* direct tunneling ([Figure 2(g)](#fig2){ref-type="fig"}) or *via* FN tunneling ([Figure 2(f)](#fig2){ref-type="fig"}). For larger gap sizes (e.g., 0.6 nm and 0.8 nm), high intensity external irradiation (\~10^10^ or 10^11^ V/m) is necessary to excite the CTP resonant mode (see Figures [2(f)](#fig2){ref-type="fig"} and [2(g)](#fig2){ref-type="fig"}).
In another recent study, Kulkarni and Manjavacas \[[@B71]\] investigated the quantum effects related to the charge transfer process by examining the spectral response of gold dimer with a two-level system (TLS) (e.g., an atom or a molecule). In this understanding, fully quantum-based computations showed that CTPs are only recognizable if one of the energy levels of the studied two-level system is in resonance with the Fermi level of the dimer system, by enabling the electron transition along the junction. At resonance, the absorption spectrum of the system and the conductance of the junction are calculated, and the outcomes indicated that the conductance of the junction is equivalent to one quantum of conductance, which is *G*~0~ = 2*e*^2^/*h*. In [Figure 3(a)](#fig3){ref-type="fig"}, a schematic for the studied subwavelength system is illustrated. This system consists of two identical gold spheres of 32 au in diameter and a small sphere with a diameter of 6 au to model the TLS. In all calculations, the whole platform is assumed to be in a vacuum. To explore the capabilities of the system, TDDFT in the adiabatic local density approximation \[[@B72], [@B73]\] is utilized. To this end, the authors only take into account the conduction electrons of the metal and the gold NPs are modelled based on jellium approximation, where the ionic background charge has a uniform charge density (*n*~0~). Here, *n*~0~ is judiciously selected to correlate with the density of gold, which is equal to a Wigner-Seitz radius of 3 au. In a similar fashion, TLS is formed by having a Wigner-Seitz radius to make sure that TLS only adds a single electron to the system. Without applying any external field, the one-electron potential of the platform is defined as *V*~eff~(**r**) = *V*~0~(**r**) + *V*~H~\[*n*(**r**) − *n*~0~(**r**)\] + *V*~xc~\[*n*(**r**)\], where *V*~H~ is the Hartree potential, *n* is the electronic density, *V*~xc~ is the exchange-correlation potential (also known as the Perdew-Zunger functional as a part of local density approximation \[[@B74]\]), *n*~0~ is the background ionic charge density, and *V*~0~ is the uniform background potential which incarcerate the electrons within the NPs and the TLs. For the gold NPs, this potential is fixed at -4.6 eV (5.2 eV lower than the vacuum level) to be able to have a proper electron spill-out. As plotted in [Figure 3(b)](#fig3){ref-type="fig"}, the energy levels of the TLS are tuned according to the Fermi level of the NPs. Additionally, the background and the resulting equilibrium one-electron potentials are represented as gray and red colors, respectively, for *V*~TLS~ = −4.6 eV. It is evident from [Figure 3(b)](#fig3){ref-type="fig"} that the potential barrier of the junction (indicated as dashed black line) is reduced by virtue of the TLS. In the following, the associated equilibrium electronic density is demonstrated in [Figure 3(c)](#fig3){ref-type="fig"}, which clearly indicates Friedel oscillations due to the discretization of the electronic levels \[[@B74]\]. Besides, the electronic density in the junction is a proof of the single energy level of the TLS. Although the TLS decreases the potential barrier, the electronic density in the gaps between the NPs does not reach to zero. In [Figure 3(d)](#fig3){ref-type="fig"}, the authors plotted the absorption spectrum for two different cases (by considering, the incident field is polarized along the *z*-axis): (a) the bare metallic dimer (gray line) and (b) the dimer with the TLS (red line). For the latter case, the background potential of the TLS is selected as equal to the NPs. In both cases, a bonding dipolar mode (BDP) \[[@B75], [@B76]\] appears around 5 eV. For the low-energy part of the spectrum, three new peaks are formed, due to the addition of TLS into the dimer system. To comprehensively evaluate the origin of these modes, the corresponding charge distributions are studied on the surface of the NPs (see [Figure 3(e)](#fig3){ref-type="fig"}). The results clearly show that the mode around 5 eV has a dipolar nature, and the other three modes indicate a monopolar distribution pattern, which is the solid evidence of the charge transfer through the junction. What is more, the modes at 1.1 eV and 1.55 eV display an oscillation of charges inside the NPs owing to the finite size effects. Based on these results, one can state that the only way to induce a CTP in this system is contingent upon the use of the levels of the TLS as the conductive paths (see the schematic in [Figure 3(f)](#fig3){ref-type="fig"}). Thus, the authors analyze the effect of background potential of the TLS on the response of the system. As demonstrated in [Figure 3(g)](#fig3){ref-type="fig"}, *V*~TLS~ is varied to manipulate the position of the levels of the TLS. In this plot, each dot represents an energy level of the whole platform. Depending on the localization rate of the TLS levels, the size of the dot can be bigger, which is beneficial to differentiate localized levels in the TLS. A group of states are localized close to the Fermi level of the system, shown in black dashed line, for small numbers of *V*~TLS~. These states become more localized as the background potential is getting deeper and the states move to lower parts in the energy diagram. However, because of the strong localization, the interplay between the NPs is diminished. Particularly, when *V*~TLS~ is lower than -16.2 eV, an extra level starts to localize on the TLS. Lastly, in Figures [3(h)](#fig3){ref-type="fig"} and [3(i)](#fig3){ref-type="fig"}, the absorption spectrum of the system (at a low-energy part) is presented for the *V*~TLS~ values considered in [Figure 3(g)](#fig3){ref-type="fig"}. The results explicitly verify the requirement of having a localized state in the TLS, whose energy is near the Fermi level of the system, to be able to generate a CTP mode.
4. Functional Charge Transfer Plasmons {#sec4}
======================================
As discussed in the previous section, in the quantum tunneling of energetic optically driven electrons, possessing an active control over the CTP spectral feature is limited to modifying the incident field intensity and/or morphological variations \[[@B42], [@B45]\]. On the other hand, the induced CTPs *via* the direct transition of optically driven electrons across the bulk metallic paths between NPs suffer from the inherent lack of tunability. Recently, these challenges have effectively been addressed by using optoelectronically and optothermally tunable components in the purpose of the excitation of functional CTPs \[[@B43]--[@B47], [@B77]\]. Such an active tunability allows for the exploration of several integrated plasmonic instruments and applications owing to its great potential for the next-generation multifunctional technology. Beyond the fundamental studies, now research in CTP devices has been focused on the experimental attempts to efficiently transform its capabilities into the real-world applications, where two fundamental issues must be addressed: functionality and scalable fabrication. While the former has been successfully realized by employing optoelectronically and thermally tunable compounds, in considering the later concern for industrial applications, one of the major challenges is scalable cost-effective fabrication of the functional metastructures. This can be done, for instance, by developing robust techniques based on nanolithography-free techniques. In this section and following subsections, we briefly review the recent techniques that have been utilized to optimize the tunability of CTPs. We will demonstrate how the integration of the plasmonic nanostructures with optoelectronically and optothermally controllable components improves the tunability of CTP spectral features.
4.1. Graphene-Enhanced Functional CTP Devices {#sec4.1}
---------------------------------------------
A one-atom thick layer of *sp*^2^-hybridized carbon atoms, known as graphene, has received copious interest due to its significantly high electron mobility, mechanical flexibility, and exquisite optical properties \[[@B78]--[@B82]\]. Graphene-enhanced fundamental applications include but not limited to the light harvesting \[[@B82]--[@B84]\], ultrafast optics \[[@B85]--[@B87]\], nonlinear photonics \[[@B88]--[@B90]\], and quantum effects \[[@B91], [@B92]\]. One of the most interesting properties of 2D carbon sheet is the possibility of controlling the photoconductivity of this monolayer *via* modifying the generated carrier density \[[@B93], [@B94]\]. This exquisite advantage has effectively been accomplished by modeling the electronic properties of graphene in terms of massless Dirac fermions \[[@B93]--[@B95]\]. This feature enables graphene to display strong infrared plasmons and made it as a promising component in implementing advanced nanophotonic devices \[[@B96]--[@B100]\]. The tunable AC photoconductivity of graphene allows to provide semimetallic behavior with an optical conductivity as a function of quantum conductance as \[[@B93], [@B94]\] $$\begin{matrix}
{\sigma = \frac{\pi e^{2}}{2h},} \\
\end{matrix}$$where *h* is Planck\'s constant. Analogous to the plasmonic components, the spectral properties and plasmonic response of graphene sheet can be estimated by the Drude absorption model for a wide range of carrier densities \[[@B93], [@B94], [@B101]\]. To describe the free carrier photoconductivity with parabolic dispersion in a 2D sheet, one can demonstrate the temperature-independent model as \[[@B68], [@B102]\] $$\begin{matrix}
{\sigma\left( \omega \right) = \frac{{ne}^{2}}{m\left( {\Gamma - i\omega} \right)},} \\
\end{matrix}$$in which *m* is the electron mass and Γ is the transport scattering rate. The unprecedented levels of beam confinement and EM field enhancement by graphene allow for having electrostatic control over the plasmonic response in the absorption spectra \[[@B96]\]. The successful example of such quantum effect was provided to explain the quantum effects in the plasmonic response of graphene nanostructures linked by a thin molecular junction. Thongrattanasiri and colleagues \[[@B77]\] demonstrated that the intrinsic characteristics of graphene enable to tune the absorption spectra of a dimer structure *via* adding small number of atoms. Using first-principle analyses, the plasmonic response of the entirely graphene-based dimer was studied for the intermediate junction with varying atomic row widths (4-8). The designed triangular structures forming the bowties are oriented with respect to the graphene lattice by assuming having armchair edges. This prevents the possible losses that can be naturally observed in zigzag-edge nanostructures, due to the presence of zero-energy electronic edge states \[[@B103]\].
To begin with, Thongrattanasiri et al. \[[@B77]\] used the random-phase approximation (RPA) theory and finite-element method (FEM) to analyze the plasmonic response of the graphene-based bowties. In addition, by fixing the fix side length of the triangles to 8 nm in all cases (almost ∼10^3^ atoms in each triangle), the researchers utilized the following settings: Fermi energy of the structures was to *E*~*F*~ = 0.4 eV, the intrinsic damping as *ħτ*^−1^ = 1.6 meV, all corresponding to a DC mobility of 10^4^ cm^2^/V·s \[[@B104], [@B105]\]. [Figure 4](#fig4){ref-type="fig"}(a) demonstrates the details of the bowties and junction area for three different characteristic values of the bridge width based on the number of carbon-atom zigzag rows (*m* = 0, 4, 8). It should be underlined that when *m* = 0, this means there is no atomically connection between proximal nanotriangles. Figures [4(b)](#fig4){ref-type="fig"}--[4(d)](#fig4){ref-type="fig"} represent the spectral response of the structures for varying junction length (*n*). As shown in [Figure 4(a)](#fig4){ref-type="fig"}, *n* resembles the number of carbon hexagons that are required to join the graphene triangles for *m* = 2. As can be explicitly seen in the extinction cross-section, by increasing the junction width, the spectral features show a trend from higher to lower energies. [Figure 4](#fig4){ref-type="fig"}(e) shows this effect in details, in which the plasmonic features are arranged as a function of energy and junction width for different values of the junction length *n*. The size of the circles is prepared proportional to the intensity of the plasmon mode defined as the area under the corresponding plasmon peaks in the extinction profile. In the narrower junction limit, the spectra point out pronounced plasmonic features placed around ∼0.47 eV. On the other hand, when the junction becomes wider, the spectra are dramatically dominated by lower-energy plasmonic features around ∼0.22 eV. Finally, for the junction width intermediate geometries, there is a complex transition between the mentioned two regimes, with intermediate-energy features around ∼0.35 eV. As depicted in [Figure 4(f)](#fig4){ref-type="fig"}, the length (*n*) of the junction does not have a significant influence on the spectral changes. It is significant to note that the three distinguished behaviors of the plasmon energies declared before for all narrow, intermediate, and wide bridges perform for all sizes of the junction length.
Further analyses help to understand the nature and properties of the plasmons in the graphene-based structures. In Figures [4(g)](#fig4){ref-type="fig"}--[4(l)](#fig4){ref-type="fig"}, the induced charge distribution profiles are plotted from narrow to wide junctions, while the length (*n*) of the junction was fixed. Relatively, [Figure 4(g)](#fig4){ref-type="fig"} illustrates the polarization profile for the high energy plasmons (∼0.47 eV) in the capacitive coupling regime (nontouching condition, *m* = 0), validating a dipole-dipole interference that is not altered when a narrow junction (*m* = 2) is inserted between the two triangular resonators of the structure (see [Figure 4(h)](#fig4){ref-type="fig"}). Conversely, when the width of the molecular bridge increases towards the intermediate regime (*m* = 6, with the energy of plasmons around \~0.35 eV), one can see a distinct dipolar polarization pattern across the junction (see [Figure 4(i)](#fig4){ref-type="fig"}). In this limit, the junction plasmons become the dominant feature, arising from the local electronic properties of the junction. Lastly, for the wider bridge (*m* = 8), similar to the fully metallic structures studied previously, the CTP becomes the dominant spectral response (see [Figure 4(j)](#fig4){ref-type="fig"}). The signs on the density plots of charge distribution maps qualitatively correspond to the distribution of the plasmon-induced charge as a function distance to the dimer center, which is exhibited in [Figure 4(k)](#fig4){ref-type="fig"}. Interestingly, increasing the length (*n*) of the junction between neighbor graphene-based nanotriangles leads to effects that are consistent with the conclusions for the width of the bridge. Strictly speaking, Thongrattanasiri and teammates verified that the classical description of the investigated graphene nanostructures fails to regenerate the plasmonic behavior derived from first principles. This is obvious, for instance, for the junction width with intermediate sizes, the conventional computations oversight the intermediate-energy junction plasmons that have been observed in the quantum calculations. In addition, another disagreement between classical and quantum modes has been observed, in which the traditional approach predicted a smooth fadeout of the dipole-dipole mode exhibited by nonoverlapping triangles during insertion of the bridge, while the CTP shows a singular behavior, as it migrates towards zero energy in the limit of vanishing junction width.
In [Figure 4(l)](#fig4){ref-type="fig"}, it is illustrated that the electronic states contributed in the plasmon of separated graphene nanotriangles are almost equal to the plasmons from an individual nanopixels. However, this involves a minor amount of Coulombic interaction and hybridization of plasmons. By insertion of a junction with an intermediate width, the strength of the hybridized modes increases significantly, leading to the emergence of novel electronic junction states (see [Figure 4(m)](#fig4){ref-type="fig"}). The noteworthy point here is the observation of two new junction plasmons around the zero energy (Dirac point). This was anticipated by researchers, because of the presence of carbon zigzag edges in the molecular bridge \[[@B106]--[@B108]\]. Therefore, in the junction plasmon regime, the excitation of plasmons contains electron or holes in the corresponding electronic junction states. [Figure 4](#fig4){ref-type="fig"}(n) demonstrates the strength of these electron-hole pair dipole transitions in the graphene-mediated structure as a function of initial and final energies (the area of circles demonstrates the strength of the electron-hole pairs). As can be explicitly seen in this panel, the plasmon energies (shown by solid curves) do not overlap with the dominant electron-hole transports. This energy mismatch reveals that the optical transitions are not single-particle excitations. This strongly supports the claim of collective plasmonic nature of the bowtie optical excitations. The advantage of this technique is the possibility of tuning the doping level of molecular graphene nanobridge, which enables possessing an active control over the charge transition. This results in the excitation of tunable spectral features such as functional CTPs.
To continue, we briefly consider the recent advances in enhancing the functionality of CTP resonances based on graphene-mediated metallic metastructures. Newly, Ahmadivand and colleagues have developed an approach to induce CTPs in gate-controlled graphene monolayer-integrated particle clusters in the THz spectra towards tailoring multifunctional metamodulators (see the schematic in [Figure 4(o)](#fig4){ref-type="fig"}) \[[@B46]\]. Using the exquisite AC photoconductivity of graphene island \[[@B94]\], the researchers demonstrated that the dynamic frequency-dependent conductivity of graphene has a direct relation to the extinction in the transmitted wave, which is defined by \[[@B109], [@B110]\] 1 − *T*/*T*~0~ = 1 − 1/\|1 + *Z*~0~*σ*~AC~(*ω*)/(1 + *n*~*s*~)\|^2^, where *Z*~0~ is the vacuum impedance, *n*~s~ is the refractive index of the thin substrate, and the optical conductivity of graphene (*σ*~AC~(*ω*)) can be taken by using the RPA principle. This leads to the strong capacitive coupling of plasmons at the *charge neutrality point* of graphene islands (resembling high resistance regime). This results in the excitation of a typical electric dipole. On the other hand, when the back-gate voltage is applied, the low resistance of graphene islands gives rise to the direct charge transition between the metallic resonators via the conductive atomic bridge and the excitation of THz-CTPs. [Figure 4](#fig4){ref-type="fig"}(p) exhibits the SEM image of the fabricated assemblies in periodic arrays with the presence of aligned graphene islands at the middle spot between the aluminum V-shaped blocks, which was implemented on a multilayer SiO~2~/ITO substrate. Here, the ITO sublayer acts as a conductive surface for the applied voltages *via* the gate to control the AC photoconductivity of graphene islands. The resistance variations for graphene area in the presence of metallic objects are measured as a function of back-gate bias (*V*~bg~), shown in [Figure 4(q)](#fig4){ref-type="fig"}. The charge neutrality point of graphene islands was determined by neglecting the inherent contact resistance of the electrodes (*V*~bg~^cnp^ = 9.5 V). Moreover, the maximum resistance was measured around \~3.52 k*Ω* corresponding to the dielectric regime, while the lowest resistance was monitored for the graphene islands in the *n*-type phase correlating with the conductive regime. The geometrical parameters of the proposed unit cell are superimposed in the inset of this panel.
To demonstrate the spectral properties of the device, firstly, the researchers computed and measured the spectral response under longitudinal polarized THz beam exposure at room temperature (300 K), as depicted in [Figure 4(r)](#fig4){ref-type="fig"}. In these sets of analyses, the researchers judiciously changed the applied bias to the gate in order to control the transition of charges across the graphene-mediated unit cell. As plotted in both panels, a distinct dipolar mode is excited around \~3.5 THz due to the capacitive coupling between the neighboring metallic resonators. The dipolar mode in all different regimes is unchanged because of its intrinsic independency from the charge transfer between the blocks. Additionally, for graphene islands in the *n*-type doping limit, the atomically thin junction between metallic V-shape resonators turns to a conductive component and enables the transition of photoinduced electrons across between the blocks. This results in the excitation of a pronounced CTP spectral feature around \~1.95 THz. For high bias regime, the central graphene islands exhibit high-resistance and capacitive coupling becomes dominant, giving rise to the elimination of CTP mode due to blocking of the charge transfer. [Figure 4](#fig4){ref-type="fig"}(s) illustrates the charge distribution map for both dipolar and CTP modes, obtained by the FEM method. Theoretically, such a unique functionality was understood and achieved by implementing the intraband AC conductivity of carbon monolayer \[[@B102], [@B111]\]. The intraband AC conductivity of graphene is containing both Drude-like and nonzero conductivities at the charge neutrality point. Hence, careful tuning of this parameter would be possible by adjusting the Fermi energy level "*E*~*F*~" based on applying back-gate bias, given by \[[@B28]\] *E*~*F*~ = *ħν*~*F*~ (*πC*~*A*~*V*~bg~), where *ħ* is the reduced Planck\'s constant, *v*~*F*~ is the Fermi velocity (10^6^ m/s), and *C*~*A*~ is the capacitance per unit area per charge of the multilayer substrate under the graphene islands. This enables active tuning of the frequency-dependent intraband photoconductivity of graphene by modifying the applied back-gate bias. Consequently, one can directly tune the conductance, resistance, and reactance of the graphene-enhanced bridge.
One other area that deserves special attention is that of optical modulation. Optical and optoelectronic metamodulators have previously been designed based on conventionally resonant structures such as Fano and EIT resonant metamolecules \[[@B112]--[@B117]\]. The dramatic dissipative and inherent losses correlating with the classical resonant nanostructures have triggered researchers to substitute these metasurfaces with a new type of ones that are capable to provide much faster and efficient modulation properties such as toroidal resonances \[[@B24], [@B115], [@B118], [@B119]\] and CTP spectral features \[[@B43]--[@B46]\]. In relation to the latest research by Ahmadivand et al. \[[@B46]\], THz plasmonic metamodulators are strategic optical components that have faced fundamental restrictions such as low efficiency, slow operating speed, and lack of tunability to manipulate THz waves. So far, various methods have been used to address these shortcomings towards designing of high-responsive, efficient, and fast plasmonic modulators. Although some of these techniques were effective, the tailored devices do not provide ultrafast switching and high modulation depth. In the research by Ahmadivand and colleagues, the devised gated graphene-mediated plasmonic device exploits the remarkable electrical and optical features of both graphene and metallic unit cell to enhance the intensity and tunability of the induced resonant spectral feature. In [Figure 4(t)](#fig4){ref-type="fig"}, the recorded THz signal amplitude from the photodetector under applied bias variations is shown. This was accomplished by sweeping the signal as a function of modulation frequency up to 10^5^ Hz, which led to a remarkable modulation depth up to 72% and fast operation speed with the rising and falling duration around 19 *μ*s and 21 *μ*s, respectively. From the modulation bandwidth principle, [Figure 4(u)](#fig4){ref-type="fig"} graph exhibits the normalized modulation magnitude, confirming a 3 dB operation bandwidth of \~19.5 kHz and \~22 kHz for numerical simulations and experimental measurements, respectively.
4.2. Phase-Change Material-Enhanced Functional CTP Devices {#sec4.2}
----------------------------------------------------------
As elaborated in Sections [2](#sec2){ref-type="sec"} and [3](#sec3){ref-type="sec"}, inserting a conductive layer underneath a plasmonic nanosystem or reducing the interparticle distance between metallic NPs down to a subnanometer scale leads to the excitation of CTPs at lower energies in the spectrum. Recent efforts have also denoted the use of metallic nanowires between adjacent NPs to make charge flow feasible \[[@B32], [@B120]\]. For the last two examples, the conductance and the geometry of the conductive junction between plasmonic elements are extremely important to tune the spectral response and the local field distribution of the system. Although the mentioned concepts have provided remarkable outcomes for the future of CTP-based real-world applications, they suffer from the lack of active tunability, which requires solid changes in the optical and electrical characteristics of a given design. One possible way to overcome this issue is the use of chalcogenide phase-change materials (PCMs) as optoelectronically controllable structures. Over the past years, these semiconductor alloys, especially Ge~2~Sb~2~Te~5~ (GST), have attained a copious interest for tailoring plasmonic and all-dielectric platforms from visible to infrared for various purposes, such as switching/modulation \[[@B115], [@B121], [@B122]\], sensing \[[@B123], [@B124]\], and beam steering \[[@B125], [@B126]\], due to possessing quick phase-changing capability (with a crystallization temperature lower than 477°C), high cyclability, thermal stability, and versatile nonvolatile optoelectronic features between opposite states \[[@B127]--[@B130]\]. In the following, we concisely present the use of GST, as a phase-change glass, to develop dynamically tunable and all-optical near-infrared devices to switch between dipolar and CTP resonances based on the phase of GST.
In [Figure 5(a)](#fig5){ref-type="fig"}, for the first time, Ahmadivand et al. \[[@B43]\] demonstrated a numerical and theoretical study of a metallodielectric dimer platform as NIR all-optical switch using a PCM. By introducing a GST nanowire into the center of the conductive bridge that links the gold dimer, the researchers obtained remarkable changes in the position and origin of the induced plasmonic modes. To implement this active switching mechanism, Ahmadivand and teammates utilized substantial alterations in both resistivity and permittivity values of GST, owing to optically stimulated phase transition process. The proposed device in this understanding is illustrated in both Figures [5(a)](#fig5){ref-type="fig"} and [5(b)](#fig5){ref-type="fig"}, superimposing the geometrical parameters. As mentioned in the earlier parts of this review, the conductivity of the bridging path between NPs is extremely important to generate CTPs. To this end, successful control of the conductivity of the metallodielectric link was achieved by applying the required energy using an additional light source to initiate the phase toggling. Here, the wavelength-dependent conductivity of the GST section was defined as \[[@B131]\] *σ*~GST~(*λ*) = (*c*/2*λ*)(1 − *ε*~eff~(*λ*)), where *ε*~eff~(*λ*) is the effective permittivity of GST portion and *c* is the velocity of light in vacuum. Besides, depending on the crystallization level of the GST, the Lorentz-Lorenz effective-medium description can be performed using \[[@B132]--[@B134]\] (*ε*~eff~(*λ*) − 1)/(*ε*~eff~(*λ*) + 2) = *f*~*c*~((*ε*~*c*~(*λ*) − 1)/(*ε*~*c*~(*λ*) + 2)) + *f*~*a*~((*ε*~*a*~(*λ*) − 1)/(*ε*~*a*~(*λ*) + 2)), where *f*~*i*~ is the volume function of the *i*th phase as 0 ≤ *f*~*i*~ = (*n*~*i*~/∑*jn*~*j*~) ≤ 1, where *n*~*j*~ is the density of the *j*th phase. It is worth mentioning that to model the photothermal heating process, we utilized the multicapacitive cascading method \[[@B135], [@B136]\]. In this approach, the absorbed photothermal heat energy (*E*~H~) in the platform can be defined as \[[@B137]\] *E*~H~ = *AQ*~abs~*F*(*r*), where *Q*~abs~ is the numerical absorption coefficient, *F*(*r*) is the optical fluence of the impinging pulse, and *A* indicates the area of the dimer antenna (more comprehensive information can be obtained from the Supplementary Information file of the article). Furthermore, the normalized extinction spectra of the dimer system are presented for the following four different configurations of the bridge: full gold, air, amorphous GST (a-GST), and crystalline GST (c-GST) (see [Figure 5(c)](#fig5){ref-type="fig"}). For completely gold bridged-dimer system, a bright dipolar mode (as a shoulder resonance) is excited around 0.73 *μ*m because of the capacitive coupling between the satellite gold nanodisks. Additionally, the charge shuttling across the conductive junction gives rise to a CTP peak at 2.4 *μ*m. When a small air gap (10 nm in length) is introduced to the monolithic gold nanowire, the charge transfer between the opposite sides of the system is hindered. Similar to the full gold case, a dipolar mode is appeared at higher energies, while a pronounced dipolar extreme is observed around 1.7 *μ*m, owing to the strong capacitive coupling between gold nanorods. Moreover, when the gap area is filled up with GST (initially, considered as a-GST), the dipolar peak around 1.7 *μ*m is red-shifted to 2.2 *μ*m with slightly increased intensity, due to negligible extinction coefficient (*k* \~ 0) of the a-GST in this regime \[[@B138]\]. Once the phase of the GST become fully crystallized (c-GST), its corresponding resistivity value decreases \~6 orders of magnitude and the charge flow starts to dominate the spectral response of the system. As a result, a CTP extreme is formed at 2.3 *μ*m and its position is considerably red-shifted (*δ* \~ 100 nm) in comparison to the dipolar mode of a-GST, because of its absorptive behavior at low energies. It is significant to note that for all the studied conditions, the position and the amplitude of the induced dipolar mode shoulder are not affected by the changes made on the gold nanobridge. To verify this, the extinction profile of a bare gold dimer is plotted in [Figure 5(c)](#fig5){ref-type="fig"} as the inset. Next, in Figures [5(d)](#fig5){ref-type="fig"}, (i, ii) and [5(e)](#fig5){ref-type="fig"}, (i, ii), the differences in the *E*-field distributions are clearly demonstrated the presence of the CTP mode for the full gold structure. In the case of a-GST, opposite charges are accumulated around the dielectric region as well as in the surrounding gold nanodisks (see [Figure 5(f)](#fig5){ref-type="fig"}, (i, ii)). On the contrary, in the c-GST limit, the intensity of the *E*-field is reduced and the excited charges can still shuttle across the bridge, owing to the minimized capacitive coupling around the GST region in the bridge (see [Figure 5(g)](#fig5){ref-type="fig"}). Lastly, the sufficiency of the proposed metallodielectric device for all-optical NIR switching has been analyzed. The corresponding analysis in [Figure 5(h)](#fig5){ref-type="fig"} shows that the nanodevice with the length of 100 nm GST portion is the best choice for quick and effective switching purposes with the following features: (a) switching from amorphous to fully crystalline phase in a few ns and toggling back to amorphous case in hundreds of femtoseconds (fs), (b) *δ* \~ 300 nm shift in terms of the resonance point, and (c) 88% of modulation depth at 1.55 *μ*m, which is also known as the telecommunication band.
In another recent study by Nooshnab and Ahmadivand \[[@B44]\], a novel CTP-resonant and optothermally controllable metamodulator is demonstrated. As indicated in [Figure 5(i)](#fig5){ref-type="fig"}, a six-member gold hexamer nanocluster is placed on top of a ring-shaped GST layer to generate actively tunable plasmonic modes. With this approach, manipulation of the charge transfer process is realized by changing the phase of the GST sublayer. To be able to operate at the telecommunication band, the geometrical parameters given in [Figure 5(j)](#fig5){ref-type="fig"} are judiciously selected. In the c-GST limit, a prominent CTP peak is emerged in the vicinity of 1550 nm, owing to the metallic nanocluster and conductive sublayer underneath. Additionally, a dipolar shoulder is formed at higher energies. When the phase of the GST subsurface is reversed to a-GST through optical heating, it acts analogous to a dielectric material and bonding dipolar mode becomes dominant in the spectral response of the system (see [Figure 5(k)](#fig5){ref-type="fig"}). In principle, under cylindrical and vortex beam excitations, a plasmonic hexamer can be tailored to sustain pronounced Fano resonances \[[@B139], [@B140]\]. Nevertheless, in this particular case, the gold hexamer does not support a dark mode due to the plane wave excitation. Furthermore, similar to the previously mentioned work, the authors utilized the Lorentz-Lorenz effective medium theory \[[@B132]--[@B134]\] to model *ε*~eff~(*λ*) of the GST sublayer for these calculations. Moreover, optically generated thermal power was quantified using the following equation: *T*(*r*, *t*) = (*AQ*~*a*~*ϕ*/1.77*ντ*)/exp(−((*t* − *t*~0~)^2^/*τ*^2^)) \[[@B137]\], where *τ* is the time constant of the incident light, *ϕ* is the beam fluence, *r* is the distance from the source, and *t*~0~ is the time delay. To verify the origin of the induced modes, the corresponding charge distribution profiles for the proposed device are investigated in [Figure 5(l)](#fig5){ref-type="fig"}. In the a-GST limit, the excited charges oscillate in the same direction, which is a characteristic signature of the dipolar mode. On the other hand, in the CTP regime, the charges are almost equally separated along the whole structure in connection with the polarization of the impinging light. Next, the modulation performance of the platform is examined. To this end, the reflection and transmission modulation characteristics of the c-GST sublayer based hexamer are plotted in [Figure 5(m)](#fig5){ref-type="fig"}. Based on the obtained results, one can state that the transition from dielectric phase to conductive phase gives rise to the formation of a new spectral feature around the telecommunication band. Besides, the proposed metallodielectric platform provides a prominent modulation depth (up to \~98%) at the NIR bandwidth (see [Figure 5(n)](#fig5){ref-type="fig"}). As a final consideration, the authors pointed out the lossy behavior of the proposed plasmonic system. In light of the calculated insertion loss values, they minimized the possible dissipative losses with the help of the direct charge transfer feature of the c-GST sublayer based hexamer.
5. Outlook {#sec5}
==========
In recent years, the field of plasmonics has experienced rapid progresses in understanding the dynamics of conductively bridged and nearly touching metallic NPs. As we described in this focused review, many intriguing physical effects have been predicted in the systems supporting distinguished CTP excitations. So far, various theoretical and experimental investigations have been performed to realize oscillating electric current along the conductive junction, as a key principle of the CTP excitation, depending on the type of the nanosystem. However, related investigations have been limited to the excitation approaches and verifying this principle by standard measurements. Now that the CTP spectral feature has been reached in a broad range of platforms in modern nanophotonics, we believe that it is time to numerically and experimentally explore the modalities towards enhancing the functionality of this spectral phenomenon, eventually, to find its useful and practical applications. We also speculate that CTP states possess a strong potential to make a profound impact in the development of coming generation multifunctional nanophotonic instruments. More precisely, the transition between hybridized dipolar and distinct CTP modes have been explicitly demonstrated under capacitive and conductive coupling regimes, respectively. Highly attractive features of CTP resonances have opened the door for novel active devices and applications, including but not limited to surface-enhanced IR absorption (SEIRA), surface-enhanced Raman scattering (SERS), optical switching, modulation, and waveguiding; however, active manipulation of these resonances using functional materials is overlooked.
6. Conclusions {#sec6}
==============
In this review, we briefly presented the recent accomplishments in the use of versatile materials towards actively tunable CTP-based nanoscale tools. As indicated in the considered studies, having an active control on the CTP mode is possible using functional materials like graphene and GST, rather than applying morphological variations. The exceptional optical and electrical characteristics of these materials have allowed researchers to design active CTP-resonant devices. We envisage that this review will provide detailed understanding for the evolution from passive to active CTP-based platforms and pave the road for developing next-generation nanophotonic devices to reach new functionalities.
Conflicts of Interest
=====================
The authors declare no conflicts of interest.
Authors\' Contributions
=======================
Burak Gerislioglu and Arash Ahmadivand contributed equally to this work.
![(a) Numerically computed scattering efficiencies of a single dimer (blue), a conductive nanowire-linked dimer (red), and a nanorod (black). The insets show the schematics of the studied nanostructures. The diameter of the disk is 95 nm, the width and the length of junction wire in the junction are 15 and 30 nm, and the thickness of all structures is 35 nm. (b) Charge plots at the position of scattering peaks for the structures in (a): a dipolar plasmon for the nanorod, capacitively coupled superradiant dipolar resonance (1) and CTP resonance (2) for the nanowire-bridged dimer, and a CTP resonance for the dimer. (c) Experimentally and numerically obtained junction conductances of nanowire-bridged dimers at CTP resonances as a function of nanowire width for aluminum and gold nanostructures. (d) CTP resonance as a function of the junction conductance for nanowire-bridged dimers with varying junction substances \[[@B32]\]. Copyright 2016, American Chemical Society. Characterized and simulated normalized transmission spectra for the metallic assembly for the presence of (e) a nanodisk between gaps, (f) a touching disk to the V-shaped resonators, and (g) an overlapping disk. Insets are the corresponding SEM graphs for disk diameter variations. (h--j) Cross-sectional *E*-field concentration (∣*E*∣) diagrams for the presence of a nontouching disk, presence of a touching disk, and presence of an overlapping disk in the middle of the unit cell, respectively. (k) Junction resistance variations as a function of the intermediate disk diameter. Inset is the CTP position as a function of conductive disk diameter \[[@B33]\]. Copyright 2016, Optical Society of America.](RESEARCH2020-9468692.001){#fig1}
![(a) Schematics of considered regions as conductive gold (the charge transfer track consists of free electrons and gold atoms), conductive gap (the charge transfer path is dominated by either direct or FN tunneling electrons), and insulating gap (the charge transfer route is empty). (b) The simulated extinction spectra of these three regions for a possible range of gap conductivities. (c) The calculated gap conductivity and tunneling electron density as a function of the intensity of the applied electric field for three different gap sizes: 0.8 nm, 0.6 nm, and 0.4 nm. (d) The tunneling barrier profiles within the gap for three possible regimes: (i) direct tunneling, (ii) FN tunneling, and (iii) saturation. In all regimes, the shaded areas indicate the energy levels of the tunneling electrons. (e--g) The simulated extinction spectra for diverse compositions of gap lengths and fields. For the weak field condition in (e), 0.6 nm and 0.8 nm gap sizes could not support the CTP mode, but they can underpin hybridized dipolar (*D*) and quadrupolar (*Q*) modes \[[@B39]\]. Copyright 2012, American Chemical Society.](RESEARCH2020-9468692.002){#fig2}
![(a) An artistic sketch of the studied two-level system. (b) One-electron potential (red) and background potentials in equilibrium. The corresponding Fermi level of the nanoplatform is represented by a dashed black line. (c) Equilibrium electronic density of the given system. (d) The calculated absorption spectrum for both bare (gray) and TLS added (red) dimer. Inset: zoomed-in version of the low-energy area. (e) Corresponding charge plots of the induced modes at (i) 5.05 eV, (ii) 0.65 eV, (iii) 1.10 eV, and (iv) 1.55 eV. (f) A schematic to denote the physical mechanism behind the generation of CTP mode. (g) The electronic structure of the system as a function of background potential of TLS. The black dashed line displays the Fermi level of the platform. (h, i) Two different versions of the absorption spectra of the system as a function of *V*~TLS~ \[[@B71]\]. Copyright 2015, American Chemical Society.](RESEARCH2020-9468692.003){#fig3}
![(a) Details of the junction region in the graphene structures, with definitions of junction length *n* and width *m*. (b--d) Computed spectra for particular bridge widths *m* = 0, 4, and 8 and several other lengths as indicated by different colors. The insets show the complete graphene structures for *n* = 2. (e) Exhibited plasmon resonances from the graphene structure as a function of bridge width *m*. The color code for different lengths *n* is given in the upper inset. The area of the circles is proportional to the area under the extinction peak for each plasmon feature. Conventional plasmon energies are indicated using dashed curves for *n* = 4 (red) and *n* = 8 (blue), bordered by shaded areas, representing the strength of the modes. (f) Plasmon resonances as a function of junction length *n*. The graphene is doped to a Fermi energy *E*~*F*~ = 0.4 eV and has a mobility *μ* = 10,000 cm^2^/V·s, and the length of the bowties is 8 nm. (g, h, i, j) Charge density maps with the color of each atom indicating its induced charge for different junction widths *m* = 0, 2, 6, and 8 and the same length *n* = 4. (k) Induced charge integrated along the horizontal direction and averaged over four nearest carbon-atom neighbors. (l, m) Electron density distribution of electronic states in *n* = 4 bowtie structures for *m* = 0 ((l), separated triangles) and *m* = 6 ((m), linked triangles). The energies of these states are shown by black lines under the density plots. (n) Dipole matrix elements between electronic states of the same bowtie as in (m). The area of the circles is proportional to the dipole strength \[[@B77]\]. Copyright 2013, WILEY-VCH. (o) Schematic of the graphene island-mediated THz cluster. (p) SEM image of the fabricated device. The scale bar is 10 *μ*m. (q) Resistance variations of the graphene monolayer, obtained numerically from the source-drain current with *V*~SD~ = 25 mV. The inset is the magnified SEM image for the fabricated sample to introduce the geometrical parts as follows: *a*/*b*/*c*/*d* = 3.5/10/7.5/4 *μ*m. The thickness of the metallic blocks is set to 200 nm, and the scale bar is 5 *μ*m. (r) Normalized transmission amplitude of the graphene-plasmonic structure under applied gate voltage. (s) The current density maps at the dipole and CTP resonance frequencies for the graphene monolayer in dielectric (*V*~bg~ = 11 V) and conductive (*V*~bg~ = 0 V) regimes. (t) The responding optical signal of the tunable device under fast on/off THz radiation modulation. (u) Experimentally (circles) and numerically (solid line) determined normalized modulation magnitude (dB) \[[@B46]\]. Copyright 2019, Royal Society of Chemistry.](RESEARCH2020-9468692.004){#fig4}
![(a) A 3D schematic of the proposed metallodielectric dimer platform. (b) A top-view image of the dimer configuration including the geometrical parameters. (c) Normalized extinction spectra of the bridged dimer for the following conditions: air, a-GST, c-GST, and full gold. (d, e) (i) top-view and (ii) cross-sectional view of the *E*-field maps along the bridged dimer for the dipole and CTP modes in full gold limit, respectively. (f, g) (i) top-view and (ii) cross-sectional view of the *E*-field distributions corresponding to the dipolar (a-GST) and CTP (c-GST) modes. (iii) *E*-field intensity diagrams at the position of each CTP and dipolar modes for all the studied cases. (h) The transmission ratio of the proposed metallodielectric switch in on (c-GST) and off (a-GST) conditions \[[@B43]\]. Copyright 2017, Nature Publishing Group. (i) An artistic sketch and (j) top-view of the studied metallodielectric cluster. (k) Normalized extinction plot of the GST-sublayer mediated hexamer configuration for a-GST and c-GST. (l) Surface charge density plots of the dipolar and CTP resonances for a-GST and c-GST cases, respectively. (m) The transmission and reflection modulation spectra as a function of wavelength. (n) The modulation depth plot of the proposed switch mechanism \[[@B44]\]. Copyright 2017, IEEE.](RESEARCH2020-9468692.005){#fig5}
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
From a clinical perspective, it is often difficult to distinguish amyotrophic lateral sclerosis (ALS) from more treatable motor neuropathies early in the course of the illness, particularly in patients with pure lower motor neuron (LMN) involvement. [@pone.0027041-Kiernan1] For instance, patients with multifocal motor neuropathy (MMN) also present with lower motor neuron (LMN) syndromes, typically with asymmetrical weakness of the distal upper limbs. Weakness and wasting develop in the absence of objective sensory or upper motor neuron (UMN) dysfunction. The demonstration of focal conduction block (CB) on motor nerve conduction studies remains the key neurophysiological hallmark of MMN, and although anti-ganglioside antibodies (GM1 antibodies) may be detectable in a proportion of patients, such antibodies may also be expressed in ALS. [@pone.0027041-Adams1]
Although often difficult in clinical practice, the distinction of ALS and other degenerative lower motor neuron diseases from MMN remains crucial as therapy with IVIg is likely to benefit patients with MMN. Specifically, although MMN is rare, up to 78% of patients will improve with intravenous immunoglobulin (IVIg) therapy, whereas patients with ALS will continue to deteriorate. [@pone.0027041-Lin1], [@pone.0027041-vanSchaik1] IVIg therapy is expensive and prescription is often restricted by regulatory authorities. In addition to common and mild side effects such as headache, fever, and malaise, therapy with IVIg may occasionally be complicated by nephrotoxicity, [@pone.0027041-Levy1] anaphylaxis, myocardial infarction, stroke or even death [@pone.0027041-Hamrock1] further supporting the general view that IVIg therapy should be reserved for patients likely to benefit. Without treatment, patients with MMN develop progressive weakness and functional disability, and in such a context may be misdiagnosed as ALS. In addition, patients with an MMN-like presentation, but without CB, may also be initially diagnosed as ALS, although a therapeutic treatment trial may show benefit from IVIg. [@pone.0027041-Delmont1]
The current consensus criteria for the diagnosis of MMN rely on the demonstration of CB in two or more motor nerve segments. [@pone.0027041-Olney1] The criteria were designed for research use rather than clinical practice and inevitably exclude treatable patients from the diagnosis of MMN. Consequently, the present study was prompted by the recognition of a group of patients who presented with an ultimately treatment responsive LMN syndrome, but did not meet the diagnostic criteria. The aim of the present study was to identify clinical and neurophysiological characteristics using a 'real-life' practical approach, that may prove useful to predict IVIg response amongst patients, to further dissect patients with a pure LMN syndrome in routine clinical practice.
Methods {#s2}
=======
Patients with clinically isolated LMN syndromes were identified from three specialised ALS clinics. Ethical approval for the study was obtained from the South Eastern Sydney and Illawarra Area Health Service Human Research Ethics Committee.
Patients were included in the study if they presented with an undifferentiated isolated LMN syndrome, that did not meet the accepted criteria for either a degenerative motor neuron disease or an inflammatory motor neuropathy (eg MMN or chronic inflammatory demyelinating polyneuropathy) and had received induction treatment with induction IVIg treatment (0.4 g/kg per day for 5 consecutive days) followed by at least three monthly maintenance treatments (0.4 g/kg by single infusion per month). Patients were studied consecutively and data recorded prospectively. Demographic data, symptom duration, presence and pattern of weakness (i.e. distal, proximal or mixed), the presence of wasting, and pattern of onset (unilateral or bilateral, upper limb or lower limb) were all recorded. Therapy was continued until the response to treatment had become clear and, in practice, this often entailed months of maintenance treatment. Treatment was ceased, at the discretion of the treating physician, in patients who deteriorated despite ongoing IVIg treatment, usually due to the development of more typical features of ALS such as UMN signs or bulbar involvement.
The study exclusion criteria were objective sensory deficits or abnormalities on standard sensory nerve conduction studies, marked UMN signs such as pathological hyper-reflexia (defined as exaggerated reflexes elicited with minimal stimulus or spread of reflexes) or an alternative diagnosis. Specifically, patients with chronic inflammatory demyelinating polyneuropathy, or benign focal amyotrophy were excluded from the study. [@pone.0027041-Vucic1], [@pone.0027041-Burrell1] Patients with suspected Hirayama\'s disease underwent cervical MRI in neck flexion, [@pone.0027041-Jakhere1] and if the diagnosis was confirmed, were excluded.
Standard clinical investigations and GM~1~ antibodies were recorded in each patient. Neurophysiological data such as distal compound motor action potential (CMAP) amplitudes and the detection of CB was recorded. Standard neurophysiological investigations were undertaken using an Oxford Teca Synergy machine (Oxford Instruments, Old Woking, Manor Way, UK). Patients had bilateral studies of upper and lower limb nerves including the median, ulnar (with above and below elbow stimulations), common peroneal and tibial motor nerves, with results compared to laboratory normal and published control values. [@pone.0027041-Burke1] CB was defined in accordance with the consensus criteria for the diagnosis of MMN, [@pone.0027041-Olney1] such that definite CB was indicated by a reduction in CMAP amplitude of \>50% in distal median, distal ulnar or proximal peroneal nerve segments, or \>60% in a distal peroneal or tibial nerve segments. CB across common sites of entrapment were not included in the analysis. Probable CB was noted when an amplitude reduction of 40--49% in median and ulnar nerve segments or 50--59% in distal peroneal or tibial nerve segments was detected. In addition, abnormal amplitude reduction (AAR), defined as 30--40% amplitude reduction in median, ulnar and radial nerve segments, or 40--50% amplitude reduction in distal common peroneal and tibial nerve segments was recorded. The detection of electromyographic discharges, fibrillation and fasciculation potentials was also noted. By convention, the frequency of fibrillation potentials recorded in each muscle were graded on a scale from 0 to 4 (0 -- None; 1 -- persistent fibrillation potentials in at least two areas; 2 -- moderate numbers of persistent fibrillation potentials; 3 -- large numbers of persistent fibrillation potentials; 4 -- profuse, widespread, persistent fibrillation potentials which fill the baseline). [@pone.0027041-Daube1] The individual muscle grades were summed and divided by the number of muscles studies to determine a novel fibrillation score per muscle for each patient.
Response to treatment was determined clinically through a combination of clinical history and examination findings on follow-up, the latter with reference to any improvement, stabilisation or deterioration in motor power as graded by the medical research council (MRC) grading scales [@pone.0027041-Brain1] after treatment. Patients who improved or stabilised with IVIg treatment were classified as responders and those who deteriorated in terms of power testing as non-responders. As mentioned, the development of upper motor neuron signs or bulbar dysfunction, or clinical progression suggestive of ALS, was noted and indicated non-response to treatment. Although all patients received a minimum of three months of IVIg therapy, treatment was continued until such time as the clinical outcome had become clear.
Statistical analysis, with P\<0.05 considered significant, was performed by application of the chi-square, paired t and Mann-Whitney tests as required (Statistical Package for Social Sciences 17.0, SPSS Inc; Chicago, IL, USA). In order to compare categorical data (for example proximal/mixed weakness compared to distal weakness), 2×2 tables were constructed and the Chi-square test applied. A receiver operator curve (ROC) was constructed using the fibrillation score to plot the true positive rate (y axis) and the false positive rate (x axis) of a candidate investigation. This process was utilised to establish a threshold at which the both the sensitivity and specificity of the investigation were maximal.
Results {#s3}
=======
In total, 42 patients were eligible for the study (30 males, 12 females, mean age 48.4 +/- 13.9 years, range 18 to 83 years), and patients were included over a period of 10 years. Mean follow-up duration was 35 +/- 32 months (range 3 to 136 months). After treatment with IVIg, 31 patients were classified as responders; 11 patients as non-responders and later fulfilled the criteria for a diagnosis of ALS. During the study period, two non-responders died of ALS-related complications. All responders continued to receive monthly maintenance IVIg infusions and no significant complications of IVIg treatment were encountered.
Clinical phenotype {#s3a}
------------------
The demographic and clinical characteristics of responders and non-responders are presented in [Table 1](#pone-0027041-t001){ref-type="table"}. The mean age of responders was significantly less (45.8 +/- 13.4 years) than non-responders (56.0 +/- 13.1 years, P\<0.05). Median symptom duration prior to diagnosis was longer in responders (18 months) than non-responders (12 months, NS), as was mean symptom duration at first assessment (responders 46.8 +/- 72.3; non-responders 18.2 +/- 20.2, NS). On average, responders had symptoms for 46.8 months prior to treatment, and several responders had symptoms for years before receiving treatment with IVIg.
10.1371/journal.pone.0027041.t001
###### Demographic and clinical characteristics.
{#pone-0027041-t001-1}
Responders Non-responders P value
--------------------------------------------------------------- ---------------- ---------------- ---------
**Number of Patients** 31 11
**Average Age (years +/- SD)** 45.8 +/-13.4 56.0 +/-13.1 \<0.05
**Male Gender (% pts)** 22 (71%) 8 (73%) NS
**Median Symptom Duration (months)** 18 12
**Mean Symptom Duration at First Assessment (months +/- SD)** 46.8 +/- 72.3 18.2 +/- 20.2 NS
**Follow-up Duration (months +/- SD)** 41.6 (+/-34.9) 16.9 (+/-9.0) \<0.05
**Number of involved limbs (+/- SD)** 1.5 (+/-0.6) 2.1 (+/-1.3) NS
**Degree of Wasting**
None or mild 19 (61.3%) 6 (54.5%) NS
Marked 12 (38.7%) 5 (45.5%)
**Pattern of Weakness**
Upper limb 26 (83.9%) 7 (63.6%) NS
Unilateral 25 (80.6%) 5 (45.5%) \<0.05
Distal 17 (54.8%) 1 (9.1%) \<0.05
The demographic and clinical features of 42 patients who presented with an isolated LMN syndrome. Responders were younger than non-responders, and typically had distal, asymmetrical, upper limb weakness.
Weakness was the most prominent presenting feature in both groups, although a minority of patients presented with wasting, muscle cramps or pain. Although severity of muscle wasting did not differ between responders and non-responders, the two patient groups had different patterns of weakness. For example, 54.8% of responders had isolated distal upper limb weakness rather than proximal or mixed weakness, compared to on 9.1% of non-responders (P\<0.05). Unilateral onset was more common in responders than non-responders (P\<0.05) ([Figure 1](#pone-0027041-g001){ref-type="fig"}). Responders also tended to have upper rather than lower limb symptom onset, and had fewer limbs involved at the time of presentation when compared to non-responders. There was no correlation between the degree of limb wasting and treatment outcome.
{#pone-0027041-g001}
Clinical investigations {#s3b}
-----------------------
Results of clinical investigations are summarised in [Table 2](#pone-0027041-t002){ref-type="table"}. GM~1~ (IgM class) antibodies were identified in 12.9% of responders and definite CB (not reaching diagnostic criteria for MMN) was identified in 22.6% of responders, but more than 50% of responders had no evidence of GM~1~ antibodies or CB. When detected, CB was identified in the ulnar nerve (3 patients), the median nerve (2 patients), the common peroneal nerve (1 patient) and the tibial nerve (1 patient). Neither GM~1~ antibodies nor CB was identified in non-responders. Of the 77.4% of responders without definite CB, 6.5% had probable CB and a further 12.9% had ARR, which was also detected in 36.4% of non-responders. In total, 41.9% of responders had CB or ARR compared to 36.4% non-responders, but the difference was not significant. By combining GM~1~ antibodies with CB and ARR to identify responders sensitivity improved (41.9%), but the specificity deteriorated (36.4%).
10.1371/journal.pone.0027041.t002
###### Neurophysiological Characteristics.
{#pone-0027041-t002-2}
Responders Non-responders P value
-------------------------------------------------------- ------------ ---------------- ---------
**Number of Patients** 31 11
**GM~1~ antibodies IgM + (% pts)** 4 (12.9%) 0 NS
**Distal CMAP at initial assessment (mean mV +/- SD)**
Upper limb 7.3 +/-2.7 8.2 +/-2.3 NS
Lower limb 6.6 +/-3.9 4.3 +/-2.7 NS
Overall 7.0 +/-2.4 6.4 +/-2.2 NS
**Definite CB (% pts)** 22.6% 0.0% NS
**Definite or Probable CB (% pts)** 29.0% 0.0% NS
**Definite CB, Probable CB or ARR (% pts)** 41.9% 36.4% NS
**Electromyography (% pts)**
Fibrillations 61.3% 81.8% NS
Complex repetitive discharges 22.6% 36.4% NS
Fasciculations 45.2% 36.4% NS
The laboratory and clinical neurophysiological characteristics of 42 patients who presented with an isolated LMN syndrome. GM~1~ antibodies and conduction block (CB), although only identified in responders, were detected in less than half of all responders.
The distal CMAP amplitudes at the initial assessment did not differ significantly between responders and non-responders overall ([Table 2](#pone-0027041-t002){ref-type="table"}) and initial distal CMAP amplitudes did not predict treatment outcome. However, non-responders demonstrated progressive decline in CMAP amplitudes on progress nerve conduction studies, suggestive of secondary axonal loss. Using electromyography, fasciculation potentials were common in both groups, but fibrillation potentials, positive sharp waves and complex repetitive discharges were more frequent in non-responders than responders, perhaps indicating more aggressive denervation (i.e. likely ALS). An ROC was constructed to determine the fibrillation score threshold which identified non-responders at an optimal sensitivity and specificity ([Figure 2](#pone-0027041-g002){ref-type="fig"}). Using this approach, a fibrillation score of \>0.4 was determined to identify non-responders with moderate sensitivity (64%) and specificity (62%).
{#pone-0027041-g002}
Discussion {#s4}
==========
The present cohort of forty-two patients who presented with an isolated LMN syndrome has identified that the most reliable predictor of a positive IVIg treatment response, and key distinguishing feature from ALS, was the recognition of the typical clinical phenotype of MMN, namely an upper limb, unilateral, and distal onset pattern of weakness. Diagnostic criteria for MMN remain insensitive and would have excluded the majority of responders in this series from a therapeutic trial of therapy. Apart from CB and GM~1~ antibodies, detected in a minority of responders, no neurophysiological or laboratory characteristic reliably distinguished responders from non-responders, with the latter progressing to a diagnosis of ALS. The present study supports the view that patients who present with isolated LMN syndromes should be given an empirical trial of IVIg therapy early in the course of their illness to determine treatment responsiveness.
Due to the lack of demonstrable CB, more than half of the responders in the present series did not satisfy the consensus criteria for the diagnosis of MMN. Nonetheless, responders exhibited a similar clinical phenotype to published cohorts of patients with MMN. [@pone.0027041-Bouche1]--[@pone.0027041-Vucic2] As such, the diagnostic criteria for MMN may be too strict and thus exclude patients with MMN based on an absence of detectable CB. Weakness among responders progressed insidiously, as reflected by symptom durations prior to first assessment, and was predominately upper limb, unilateral, and distal in onset. Responders were younger than non-responders, and as others have observed [@pone.0027041-Slee1], [@pone.0027041-StriglPill1], [@pone.0027041-VandenBergVos1], many responders in the present series had symptoms for several years prior to treatment, often after review by several neurologists. Although not useful in selecting patients for IVIg treatment early in the course of their illness, slow progression of disease over many months or years may allow the distinction of ALS from other conditions. [@pone.0027041-Chi1] Early intervention in such cases might lead to improved treatment outcomes and potentially reduced long term disability due to secondary axonal loss. If, after early initiation of therapy, patients develop typical features of ALS, such as UMN signs or bulbar involvement, withdrawal of IVIg would be appropriate. Such an approach is becoming the de facto standard of care in many major ALS centres.
The presence of detectable GM~1~ antibodies was highly specific for a positive treatment response to IVIg, but only 12.9% of responders in this series were positive for GM~1~ antibodies. This rate is similar to that documented for patients with LMN syndromes treated with IVIg [@pone.0027041-StriglPill1], but lower than in cohorts of MMN. [@pone.0027041-Slee1], [@pone.0027041-Taylor1] Although the reported sensitivity of GM1 antibodies remains highly variable, GM1 titres have recently been correlated with the severity of weakness in MMN. [@pone.0027041-Cats1]
Definite or probable CB on standard motor nerve conduction studies was highly specific, but poorly sensitive, for response to IVIg treatment. The relatively low rate of detectable CB in the present series is similar to cohorts of MMN patients. [@pone.0027041-Slee1], [@pone.0027041-Vucic2] When smaller CMAP amplitude reductions (i.e. ARR) were included in the analysis, the sensitivity improved marginally, but specificity dropped dramatically. Even after including CB, ARR and GM1 antibodies only 45.2% of responders were identified.
Undetected proximal CB among responders in the present series cannot be excluded, as cervical root stimulation was not universally performed. However, cervical root stimulation techniques are technically demanding, and their use in previous studies has yielded variable results. [@pone.0027041-Vucic2], [@pone.0027041-Ghosh1] As such, the role of these techniques has not been definitively established. Other techniques, for example utilising transcranial magnetic stimulation may detect proximal CB [@pone.0027041-Deroide1], or conversely, may detect sub-clinical UMN dysfunction in ALS patients. [@pone.0027041-Vucic3], [@pone.0027041-Vucic4] Given the difficulty in detecting proximal CB using neurophysiological techniques, T2 weighted magnetic resonance imaging is recommended by the European Federation Neurological Societies / Peripheral Nerve Society guidelines to establish proximal involvement -- such as proximal demyelination or nerve root hypertrophy -- in MMN or other immune mediated neuropathies. [@pone.0027041-VanDenBergVos1], [@pone.0027041-VanDenBergh1]
In summary, the presence of CB or GM~1~ antibodies are specific but insensitive predictors of response to IVIg in patients that present with isolated LMN syndromes that do not meet diagnostic criteria for degenerative motor neuron diseases or inflammatory motor neuropathy. Responders may have slowly progressive symptoms for many years prior to starting treatment, that may detrimentally affect the therapeutic outcome. Given that the present cohort may be considered relatively small, a larger randomised placebo control trial would be required to definitively establish the role of IVIg in patients with pure LMN syndromes that do not fulfil diagnostic criteria. Nonetheless, the present study supports the view that patients with pure LMN syndromes should be given an early therapeutic trial of IVIg, even in the absence of GM~1~ antibodies and CB.
**Competing Interests:**The authors have declared that no competing interests exist.
**Funding:**The authors gratefully acknowledge the support of the National Health and Medical Research Council of Australia and the Motor Neurone Disease Research Institute of Australia. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
[^1]: Conceived and designed the experiments: JRB CY DR MCK. Performed the experiments: CY DR MCK. Analyzed the data: JRB. Contributed reagents/materials/analysis tools: JRB CY DR MCK. Wrote the paper: JRB CY DR MCK.
| {
"pile_set_name": "PubMed Central"
} |
Related literature {#sec1}
==================
For background to the properties and uses of Schiff bases, see: Barton & Ollis (1979[@bb1]); Layer (1963[@bb8]); Ingold (1969[@bb7]); Cohen *et al.* (1964[@bb4]); Taggi *et al.* (2002[@bb14]). For hydrogen-bond motifs, see: Bernstein *et al.* (1995[@bb2]). For comparative bond lengths, see: Şahin *et al.* (2009[@bb10]). For related structures, see: Özdemir *et al.* (2010[@bb9]); Tanak *et al.* (2009[@bb15]).
Experimental {#sec2}
============
{#sec2.1}
### Crystal data {#sec2.1.1}
C~14~H~12~INO*M* *~r~* = 337.15Orthorhombic,*a* = 4.6773 (4) Å*b* = 11.6092 (12) Å*c* = 23.6751 (4) Å*V* = 1285.55 (17) Å^3^*Z* = 4Mo *K*α radiationμ = 2.47 mm^−1^*T* = 293 K0.48 × 0.24 × 0.09 mm
### Data collection {#sec2.1.2}
Stoe IPDS II diffractometerAbsorption correction: numerical (*X-AREA*; Stoe & Cie, 2002[@bb13]) *T* ~min~ = 0.520, *T* ~max~ = 0.7697548 measured reflections2267 independent reflections1541 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.086
### Refinement {#sec2.1.3}
*R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.037*wR*(*F* ^2^) = 0.078*S* = 0.862267 reflections156 parametersH-atom parameters constrainedΔρ~max~ = 0.65 e Å^−3^Δρ~min~ = −0.29 e Å^−3^Absolute structure: Flack (1983[@bb6]), 901 Friedel pairsFlack parameter: 0.10 (5)
{#d5e435}
Data collection: *X-AREA* (Stoe & Cie, 2002[@bb13]); cell refinement: *X-AREA*; data reduction: *X-AREA*; program(s) used to solve structure: *SHELXS97* (Sheldrick, 2008[@bb11]); program(s) used to refine structure: *SHELXL97* (Sheldrick, 2008[@bb11]); molecular graphics: *ORTEPIII* (Burnett & Johnson, 1996[@bb3]), *ORTEP-3 for Windows* (Farrugia, 1997[@bb5]) and *PLATON* (Spek, 2009[@bb12]); software used to prepare material for publication: *SHELXL97*.
Supplementary Material
======================
Crystal structure: contains datablocks I, global. DOI: [10.1107/S160053681001826X/dn2563sup1.cif](http://dx.doi.org/10.1107/S160053681001826X/dn2563sup1.cif)
Structure factors: contains datablocks I. DOI: [10.1107/S160053681001826X/dn2563Isup2.hkl](http://dx.doi.org/10.1107/S160053681001826X/dn2563Isup2.hkl)
Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?dn2563&file=dn2563sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?dn2563sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?dn2563&checkcif=yes)
Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [DN2563](http://scripts.iucr.org/cgi-bin/sendsup?dn2563)).
The authors acknowledge the Faculty of Arts and Sciences, Ondokuz Mayıs University, Turkey, for the use of the Stoe IPDS II diffractometer (purchased under grant No. F279 of the University Research Fund).
Comment
=======
Schiff bases are used as starting materials in the synthesis of important drugs, such as antibiotics and antiallergic, antiphlogistic, and antitumor substances (Barton *et al.*, 1979; Layer, 1963; Ingold 1969). On the industrial scale, they have a wide range of applications, such as dyes and pigments (Taggi *et al.*, 2002). There are two characteristic properties of Schiff bases, viz. Photochromism and thermochromism (Cohen *et al.*, 1964). In general, Schiff bases display two possible tautomeric forms, the phenol-imine (OH) and the keto-amine (NH) forms. Depending on the tautomers, two types of intramolecular hydrogen bonds are observed in Schiff bases: O---H···N in phenol-imine (Şahin *et al.*, 2009) and N---H···O in keto-amine tautomers (Tanak *et al.*, 2009). Another form of the Schiff base compounds is also known as zwitterion having an ionic intramolecular hydrogen bond (N^+^---H···O^-^) and this form is rarely seen in the solid state (Özdemir *et al.*, 2010).
The molecular structure of the title compound, C~14~H~17~O~1~N~1~I~1,~shows that the molecule exists in the phenol-imine form (Fig. 1). The C1=N1 \[1.269 (8) Å\] and C9=N1 \[1.397 (7) Å\] bond distances are of double-bond character, whereas, C7---O1 \[1.332 (8) Å\] distance is single bond. These distances are similar to that reported in the literature \[1.277 (3) Å\] and \[1.402 (3) Å\] for C=N and \[1.347 (3) Å\] for C---O respectively (Şahin *et al.*, 2009).
The molecule of title compound is non-planar (Fig. 1), the two phenyl rings are twisted by a dihedral angle of 20.6 (3)°. This conformation is stabilized by intramolecular N-H···O hydrogen bond (Table 1, Fig. 1) forming S(6) ring (Bernstein *et al.*, 1995). weak intermolecular C-H···O hydrogen bonds link the molecules forming a zig-zag chain parallel to the b axis (Table 1, Fig. 2). The I atom is slightly out of the C9-C14 ring by 0.18 (1)Å.
Experimental {#experimental}
============
The compound (*E*)-2-\[(4-Iodophenylimino)methyl\]-6-methylphenol was prepared by reflux a mixture of a solution containing 3-methylsalicylaldehyde (0.1 ml 0.82 mmol) in 20 ml e thanol and solution containing 4-Iodoaniline (0.179 g 0.82 mmol) in 20 ml e thanol.The reaction mixture was stirred for 1 hunder reflux. The crystals of (*E*)-2-\[(4-Iodophenylimino)methyl\]-6-methylphenol suitable for x-ray analysis were obtained from ethylalcohol by slow evaporation (yield 51%; m.p.350-353 K).
Refinement {#refinement}
==========
The position of the H1 atom was obtained from a difference map of the electron density in the unit-cell and was refined freely. Other H atoms were positioned geometrically and treated using a riding model, fixing the bond lengths at 0.93 Å for aromatic CH and at 0.96 Å for CH~3~. The displacement parameters of the H atoms were constrained as U~iso~(H)= 1.2U~eq~(1.5U~eq~ for methyl) of the parent atom.
Figures
=======
{#Fap1}
{#Fap2}
Crystal data {#tablewrapcrystaldatalong}
============
------------------------------- ---------------------------------------
C~14~H~12~INO *F*(000) = 656
*M~r~* = 337.15 *D*~x~ = 1.742 Mg m^−3^
Orthorhombic, *P*2~1~2~1~2~1~ Mo *K*α radiation, λ = 0.71073 Å
Hall symbol: P 2ac 2ab Cell parameters from 7912 reflections
*a* = 4.6773 (4) Å θ = 1.7--27.8°
*b* = 11.6092 (12) Å µ = 2.47 mm^−1^
*c* = 23.6751 (4) Å *T* = 293 K
*V* = 1285.55 (17) Å^3^ Prism, yellow
*Z* = 4 0.48 × 0.24 × 0.09 mm
------------------------------- ---------------------------------------
Data collection {#tablewrapdatacollectionlong}
===============
--------------------------------------------------------------- --------------------------------------
Stoe IPDS II diffractometer 2267 independent reflections
Radiation source: fine-focus sealed tube 1541 reflections with *I* \> 2σ(*I*)
graphite *R*~int~ = 0.086
Detector resolution: 6.67 pixels mm^-1^ θ~max~ = 25.0°, θ~min~ = 1.7°
rotation method scans *h* = −5→5
Absorption correction: numerical (*X-AREA*; Stoe & Cie, 2002) *k* = −13→13
*T*~min~ = 0.520, *T*~max~ = 0.769 *l* = −28→28
7548 measured reflections
--------------------------------------------------------------- --------------------------------------
Refinement {#tablewraprefinementdatalong}
==========
---------------------------------------------------------------- -------------------------------------------------------------------------------------
Refinement on *F*^2^ Secondary atom site location: difference Fourier map
Least-squares matrix: full Hydrogen site location: inferred from neighbouring sites
*R*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.037 H-atom parameters constrained
*wR*(*F*^2^) = 0.078 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0321*P*)^2^\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3
*S* = 0.86 (Δ/σ)~max~ = 0.001
2267 reflections Δρ~max~ = 0.65 e Å^−3^
156 parameters Δρ~min~ = −0.29 e Å^−3^
0 restraints Absolute structure: Flack (1983), 901 Friedel pairs
Primary atom site location: structure-invariant direct methods Flack parameter: 0.10 (5)
---------------------------------------------------------------- -------------------------------------------------------------------------------------
Special details {#specialdetails}
===============
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.
Refinement. Refinement of *F*^2^ against ALL reflections. The weighted *R*-factor wR and goodness of fit *S* are based on *F*^2^, conventional *R*-factors *R* are based on *F*, with *F* set to zero for negative *F*^2^. The threshold expression of *F*^2^ \> σ(*F*^2^) is used only for calculating *R*-factors(gt) etc. and is not relevant to the choice of reflections for refinement. *R*-factors based on *F*^2^ are statistically about twice as large as those based on *F*, and *R*- factors based on ALL data will be even larger.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å^2^) {#tablewrapcoords}
==================================================================================================
----- -------------- ------------- --------------- -------------------- --
*x* *y* *z* *U*~iso~\*/*U*~eq~
I1 0.81565 (9) 0.49212 (4) 0.907466 (15) 0.07878 (18)
O1 −0.3098 (17) 0.6170 (3) 0.6324 (2) 0.0759 (15)
H1 −0.2028 0.6014 0.6588 0.114\*
N1 0.0187 (10) 0.4895 (5) 0.69451 (16) 0.0621 (12)
C1 −0.0685 (14) 0.4026 (6) 0.6669 (2) 0.0596 (16)
H15 0.0064 0.3306 0.6757 0.071\*
C2 −0.2788 (18) 0.4106 (5) 0.6224 (3) 0.0553 (19)
C3 −0.3704 (13) 0.3134 (5) 0.5941 (3) 0.0639 (16)
H3 −0.2933 0.2423 0.6037 0.077\*
C4 −0.5711 (17) 0.3190 (6) 0.5522 (3) 0.071 (2)
H4 −0.6369 0.2523 0.5348 0.085\*
C5 −0.6755 (18) 0.4266 (6) 0.5361 (3) 0.0680 (18)
H5 −0.8058 0.4307 0.5065 0.082\*
C6 −0.5925 (14) 0.5272 (6) 0.5624 (3) 0.0660 (19)
C7 −0.3895 (13) 0.5201 (5) 0.6065 (2) 0.0568 (15)
C8 −0.699 (2) 0.6422 (6) 0.5449 (3) 0.091 (3)
H8A −0.8417 0.6332 0.5161 0.137\*
H8B −0.5434 0.6872 0.5305 0.137\*
H8C −0.7818 0.6806 0.5769 0.137\*
C9 0.2092 (13) 0.4822 (5) 0.7398 (2) 0.0582 (13)
C10 0.3469 (18) 0.5806 (5) 0.7557 (3) 0.068 (2)
H10 0.3162 0.6479 0.7353 0.082\*
C11 0.5354 (18) 0.5827 (6) 0.8024 (3) 0.069 (2)
H11 0.6340 0.6496 0.8118 0.083\*
C12 0.5693 (13) 0.4850 (7) 0.8333 (2) 0.0636 (15)
C13 0.4354 (17) 0.3826 (6) 0.8184 (3) 0.0658 (19)
H13 0.4652 0.3158 0.8392 0.079\*
C14 0.2599 (16) 0.3824 (5) 0.7726 (3) 0.068 (2)
H14 0.1697 0.3141 0.7625 0.081\*
----- -------------- ------------- --------------- -------------------- --
Atomic displacement parameters (Å^2^) {#tablewrapadps}
=====================================
----- ------------ ------------ ------------ ------------- ------------- -------------
*U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^
I1 0.0755 (3) 0.0930 (3) 0.0679 (2) −0.0021 (4) −0.0059 (2) −0.0085 (3)
O1 0.099 (5) 0.053 (2) 0.076 (3) −0.001 (3) 0.001 (3) −0.001 (2)
N1 0.060 (3) 0.068 (3) 0.058 (2) −0.011 (4) 0.002 (2) −0.007 (3)
C1 0.058 (4) 0.056 (4) 0.064 (4) 0.008 (3) 0.009 (3) 0.002 (3)
C2 0.050 (5) 0.060 (4) 0.057 (3) 0.004 (3) 0.001 (3) −0.001 (3)
C3 0.068 (5) 0.060 (3) 0.064 (3) 0.012 (3) 0.003 (4) −0.003 (3)
C4 0.071 (5) 0.076 (5) 0.067 (4) −0.004 (4) 0.000 (4) −0.010 (3)
C5 0.060 (4) 0.086 (5) 0.058 (4) 0.005 (5) 0.006 (4) 0.010 (3)
C6 0.066 (4) 0.072 (5) 0.060 (3) −0.007 (4) 0.011 (3) 0.012 (3)
C7 0.059 (4) 0.061 (4) 0.051 (3) 0.002 (4) 0.012 (2) 0.007 (3)
C8 0.109 (9) 0.080 (5) 0.085 (5) 0.012 (6) −0.006 (5) 0.016 (4)
C9 0.058 (3) 0.052 (3) 0.066 (3) 0.003 (4) 0.003 (3) −0.006 (3)
C10 0.078 (6) 0.054 (4) 0.073 (4) 0.005 (4) 0.007 (4) 0.014 (3)
C11 0.078 (5) 0.064 (4) 0.067 (4) −0.017 (4) 0.001 (4) −0.005 (3)
C12 0.063 (3) 0.076 (5) 0.052 (3) 0.000 (4) 0.004 (2) 0.001 (4)
C13 0.075 (5) 0.060 (4) 0.063 (4) −0.001 (4) −0.004 (4) −0.002 (3)
C14 0.074 (7) 0.058 (4) 0.071 (4) −0.006 (4) −0.010 (4) −0.007 (3)
----- ------------ ------------ ------------ ------------- ------------- -------------
Geometric parameters (Å, °) {#tablewrapgeomlong}
===========================
------------------- ------------ ----------------------- ------------
I1---C12 2.102 (5) C6---C7 1.415 (8)
O1---C7 1.334 (7) C6---C8 1.484 (9)
O1---H1 0.8200 C8---H8A 0.9600
N1---C1 1.269 (8) C8---H8B 0.9600
N1---C9 1.397 (7) C8---H8C 0.9600
C1---C2 1.444 (10) C9---C10 1.364 (9)
C1---H15 0.9300 C9---C14 1.414 (9)
C2---C3 1.381 (8) C10---C11 1.415 (10)
C2---C7 1.423 (9) C10---H10 0.9300
C3---C4 1.367 (9) C11---C12 1.358 (9)
C3---H3 0.9300 C11---H11 0.9300
C4---C5 1.394 (10) C12---C13 1.389 (10)
C4---H4 0.9300 C13---C14 1.359 (9)
C5---C6 1.379 (9) C13---H13 0.9300
C5---H5 0.9300 C14---H14 0.9300
C7---O1---H1 109.5 C6---C8---H8B 109.5
C1---N1---C9 123.5 (6) H8A---C8---H8B 109.5
N1---C1---C2 122.9 (6) C6---C8---H8C 109.5
N1---C1---H15 118.5 H8A---C8---H8C 109.5
C2---C1---H15 118.5 H8B---C8---H8C 109.5
C3---C2---C7 119.2 (6) C10---C9---N1 117.5 (6)
C3---C2---C1 120.9 (6) C10---C9---C14 117.1 (5)
C7---C2---C1 119.9 (6) N1---C9---C14 125.3 (6)
C4---C3---C2 121.8 (6) C9---C10---C11 121.6 (6)
C4---C3---H3 119.1 C9---C10---H10 119.2
C2---C3---H3 119.1 C11---C10---H10 119.2
C3---C4---C5 118.8 (6) C12---C11---C10 118.6 (6)
C3---C4---H4 120.6 C12---C11---H11 120.7
C5---C4---H4 120.6 C10---C11---H11 120.7
C6---C5---C4 122.4 (6) C11---C12---C13 121.7 (5)
C6---C5---H5 118.8 C11---C12---I1 118.7 (5)
C4---C5---H5 118.8 C13---C12---I1 119.5 (5)
C5---C6---C7 118.3 (6) C14---C13---C12 118.4 (6)
C5---C6---C8 122.8 (6) C14---C13---H13 120.8
C7---C6---C8 118.9 (7) C12---C13---H13 120.8
O1---C7---C6 118.6 (6) C13---C14---C9 122.5 (6)
O1---C7---C2 122.0 (5) C13---C14---H14 118.7
C6---C7---C2 119.4 (6) C9---C14---H14 118.7
C6---C8---H8A 109.5
C9---N1---C1---C2 176.1 (5) C3---C2---C7---C6 0.5 (9)
N1---C1---C2---C3 −179.1 (6) C1---C2---C7---C6 179.1 (5)
N1---C1---C2---C7 2.3 (9) C1---N1---C9---C10 162.2 (6)
C7---C2---C3---C4 −2.0 (10) C1---N1---C9---C14 −21.4 (9)
C1---C2---C3---C4 179.4 (6) N1---C9---C10---C11 177.6 (6)
C2---C3---C4---C5 3.1 (10) C14---C9---C10---C11 0.8 (10)
C3---C4---C5---C6 −2.8 (11) C9---C10---C11---C12 −2.7 (11)
C4---C5---C6---C7 1.3 (10) C10---C11---C12---C13 3.3 (11)
C4---C5---C6---C8 179.1 (7) C10---C11---C12---I1 −173.8 (5)
C5---C6---C7---O1 −179.6 (6) C11---C12---C13---C14 −2.0 (11)
C8---C6---C7---O1 2.6 (9) I1---C12---C13---C14 175.1 (5)
C5---C6---C7---C2 −0.2 (9) C12---C13---C14---C9 0.1 (11)
C8---C6---C7---C2 −178.0 (6) C10---C9---C14---C13 0.5 (10)
C3---C2---C7---O1 179.9 (7) N1---C9---C14---C13 −176.0 (6)
C1---C2---C7---O1 −1.5 (9) C2---C1---N1---C9 176.1 (5)
------------------- ------------ ----------------------- ------------
Hydrogen-bond geometry (Å, °) {#tablewraphbondslong}
=============================
------------------- --------- --------- ----------- ---------------
*D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A*
O1---H1···N1 0.82 1.86 2.591 (8) 147
C13---H13···O1^i^ 0.93 2.51 3.348 (8) 150
------------------- --------- --------- ----------- ---------------
Symmetry codes: (i) −*x*, *y*−1/2, −*z*+3/2.
###### Hydrogen-bond geometry (Å, °)
*D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A*
----------------- --------- ------- ----------- -------------
O1---H1⋯N1 0.82 1.86 2.591 (8) 147
C13---H13⋯O1^i^ 0.93 2.51 3.348 (8) 150
Symmetry code: (i) .
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Acute pancreatitis (AP) is a potentially life-threatening condition that is associated with local and systemic complications \[[@gou088-B1], [@gou088-B2]\]. Pancreatic fluid collections (PFCs) are an important and well recognized local complication of acute pancreatitis \[[@gou088-B1], [@gou088-B2]\]. Widespread availability of cross-sectional imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI), have made possible better characterization and understanding of different types of PFCs. Recently, the Acute Pancreatitis Classification Working Group proposed a revision of the Atlanta Classification for PFCs following episodes of acute pancreatitis. An important criterion used in this classification was the content of PFCs, *viz.* liquid alone or solid component admixed with a varying amount of liquid content \[[@gou088-B1]\]. The pancreatic pseudocyst was defined as an encapsulated collection containing essentially nil or minimal solid material, whereas walled-off pancreatic necrosis (WOPN) was defined as an encapsulated collection of solid necrotic material with varying amount of liquid content.
It is important to differentiate between an acute pseudocyst and WOPN by accurate identification and characterization of the solid necrotic debris, as this has implications in management. Patients with WOPN usually require more aggressive endoscopic drainage, in contrast to simple transmural drainage, which may be adequate for treatment of pseudocysts \[[@gou088-B3]\]. Also, in our previously published paper, we showed that the morphological features of WOPNs determine the therapeutic strategy, since collections with greater amounts of solid debris require more aggressive therapeutic interventions---such as direct endoscopic necrosectomy (DEN)---as well as more endoscopic procedures, for successful clinical outcome \[[@gou088-B9]\]. WOPNs contain both liquid and solid necrotic debris and it is usually difficult to distinguish them from pure liquid collections on contrast-enhanced CT (CECT) \[[@gou088-B10]\]. MRI and endoscopic ultrasound (EUS) have been shown to be more sensitive than CECT in detecting the solid necrotic debris \[[@gou088-B10]\]. Although EUS has been shown to be an accurate investigation for the evaluation of solid debris in WOPN, it is invasive and causes patient discomfort. Trans-abdominal ultrasound (USG) is a cheap, non-invasive and widely available investigation but has not been evaluated in patients with WOPN to establish its ability to detect solid necrotic debris. We performed a study to evaluate and compare the diagnostic performances of USG, EUS and MRI in identifying solid necrotic debris in patients with WOPN.
Methods
=======
This was a prospective study, in which we enrolled the study subjects from among patients that had been referred to our endoscopy unit for endoscopic drainage between April 2013 and July 2014. All the enrolled patients had earlier been diagnosed with acute necrotizing pancreatitis and were now symptomatic, with a documented pancreatic fluid collection on CECT. Exclusion criteria were pregnancy, age less than 18 years, congestive cardiac failure, compromised pulmonary status or any contraindication to MRI (presence of metal implants incompatible with MRI and claustrophobia). The study was approved by the Institutional Ethics Committee and an informed written consent was obtained from the patients prior to enrolment in the study. Following inclusion in the study protocol, all subjects underwent USG, MRI and EUS within two days. The trans-abdominal USG was performed by gastroenterologists (SSR, VS or PC) who have extensive experience in abdominal ultrasound, and the images were recorded. The pancreatic MR imaging was done at 1.5 Tesla and these images were also recorded. Similarly, EUS was performed by experienced endosonographers (SSR or DKB) using a radial echoendoscope (Pentax EG-3670 URK radial echoendoscope, Pentax Corp., Japan, or GF-UE 160 radial echoendoscope, Olympus Corp., Japan) or linear echoendoscope (Pentax EG 3870 UTK, Pentax Corp., Japan, or Olympus GF-UCT180 linear echoendoscope, Olympus Corp., Japan) at a frequency of 7.5 MHz. EUS was performed with the patient in the left lateral decubitus position under conscious sedation with intravenous midazolam (dose ranging from 2.5 mg to 5 mg).
On EUS the size, as well as the detailed morphology of the PFCs, was studied with special emphasis on the presence---as well as the amount---of solid necrotic debris. The echogenic material present in the PFCs was suggestive of solid debris. Two endosonographers (SSR and DKB) separately reviewed the EUS images to quantify the solid debris in the PFCs as \<10%, 10--40% and \>40%. This sub-grouping of WOPN, based on the amount of solid debris, has been previously described in a separate study by our group \[[@gou088-B9], [@gou088-B11]\]. Briefly, the quantification of the solid debris was an approximate visual judgment by the endoscopist, based on evaluation of multiple images. Two experienced endosonographers (SSR and DKB) separately reviewed the EUS images to quantify the solid debris in the PFCs and the mean of their findings was taken as the final measure of solid debris in each PFC. In the event of disagreement between the two endosonographers, the images were jointly reviewed by both and their consensus opinion was recorded.
The USG and MRI images were independently interpreted by two gastroenterologists (DKB and SSR) who were blinded to the results obtained with the other imaging modalities. The echogenic material seen in the collection on abdominal ultrasound was considered as necrotic debris whereas, on MRI, the hypo-intense areas inside the collection on T2-weighted images was interpreted as solid debris. The solid material noted within the PFCs was quantified as described above. Also an attempt was made to detect the venous collaterals around the collection on the three imaging modalities.
Statistical analysis
--------------------
The descriptive analysis was performed and the results were presented as percentages for categorical variables and mean ± standard deviation (SD) for continuous variables. The number of patients with \<10%, 10--40% and \>40% solid necrotic debris detected by USG, MRI and EUS were compared using the Chi-squared test. A *P*-value of \<0.05 was considered as significant.
Results
=======
A total of fifty-two patients with WOPN were included in our study of which 37 were males with a mean age of 38.9 ± 12.6 years. All the patients had been earlier diagnosed with acute necrotizing pancreatitis and the etiology of acute pancreatitis was attributable to alcohol in 33 (63%), gall stones in 15 (29%) and idiopathic in 4 (8%) patients. The imaging (EUS, MRI and USG) was done at a mean of 11.7 ± 5.5 weeks after onset of acute pancreatitis ([Figures 1](#gou088-F1){ref-type="fig"} and [2](#gou088-F2){ref-type="fig"}). All these patients had undergone CT at the referring centre and review of the CT images revealed heterogeneous attenuation of collection suggesting solid debris in only 9 (17.3%) patients. Fourteen patients had multiple collections and the largest peri-pancreatic collection was assessed by all three imaging modalities. The mean size of WOPN was 9.3 ± 2.4 cm. The collections were detected by EUS and MRI in all patients whereas USG could not detect WOPN in four (7.6%) patients (3 patients had a predominantly solid WOPN whereas one patient had air foci in WOPN). There were no complications of EUS. Figure 1.Imaging in patient with WOPN having 10--40% solid necrotic debris. a) Abdominal ultrasound; b) Magnetic resonance imaging; c) Endoscopic ultrasound. Figure 2.Imaging in patient with WOPN having \<10 % solid necrotic debris. a) Abdominal ultrasound; b) Magnetic resonance imaging; c) Endoscopic ultrasound.
On EUS collections were well visualized and the contents could be well characterized in 51 (98%) patients. The collection in one patient could not be well assessed on EUS because of presence of air in the collection. In the remaining 51 patients, solid necrotic material appearing as echogenic material could be well visualized in all the patients. On EUS, 10 patients (20%) had \<10% solid content, 33 patients (64%) had solid content varying between10--40% and 8 patients (16%) had \>40% solid content. Twelve (23%) patients had venous collaterals around the WOPN because of thrombosis of the spleno-portal axis.
On MRI, collections could be well visualized in all the patients with hypo-intense areas inside the collection on T2 weighted images suggestive of solid debris. The air in the collection that was noted on CT in one patient could not be appreciated on MRI. Fourteen (27%) patients had \<10% solid content, 30 (58%) patients had solid content varying between10--40% and 8 patients (15%) had \>40% solid content. No collaterals around the collection could be visualized on MRI in any of the patients.
On abdominal ultrasound, the WOPN could be visualized in 48 (92%) patients. Three of these patients with undetectable WOPN on USG had a predominantly solid WOPN and one patient had air within the collection. The patients with undetectable WOPN on ultrasound were imaged within 6 weeks of the onset of symptoms. On USG, 10 patients (20%) had \<10% solid content, 33 patients (69%) had solid content varying between10--40% and 5 patients (11%) had \>40% solid content. No collaterals around the collection could be visualized on USG in any of the patients. The number of patients with \<10%, 10--40% and \>40% solid necrotic debris detected by USG, MRI and EUS were comparable (*P \> 0.05*; [Table 1](#gou088-T1){ref-type="table"}). Table 1.Solid content in walled-off pancreatic necrosis on different imaging modalities (*n* = 52)Imaging modalitiesSolid content in WOPN, *n* (%)\<10%10--40%\>40%Endoscopic ultrasound[^a^](#gou088-TF1){ref-type="table-fn"}10 (19.6%)33 (64.7%)8 (15.7%)Magnetic resonance imaging14 (26.9%)30 (57.7%)8 (15.4%)Abdominal ultrasound[^b^](#gou088-TF2){ref-type="table-fn"}10 (20.8%)33 (68.8%)5 (10.4%)[^1][^2][^3]
Discussion
==========
Accurate differentiation of acute pseudocyst from WOPN is crucial in managing patients of acute pancreatitis complicated by fluid collections. The pancreatic pseudocyst contains essentially nil or minimal solid necrotic material, whereas WOPN has a varying amount of solid necrotic material. In this prospective study of 52 patients, we compared the diagnostic capabilities of USG, EUS and MRI in identifying solid necrotic debris in patients with WOPN detected on CT.
In our study, CT images revealed heterogeneous attenuation of the collection, suggesting solid debris in only 9 (17.3%) patients and this observation is similar to the results of previous studies that have shown that CT has a poor accuracy in detecting solid debris in acute peri-pancreatic collections \[[@gou088-B10]\]. On MRI, we could detect solid debris in all 52 patients and this observation is in accord with results of the earlier studies that have shown that MR imaging depicts solid debris more frequently than CT in patients with WOPN \[[@gou088-B10]\]. Recently, EUS has been shown to be the most accurate imaging modality for characterizing peri-pancreatic collections \[[@gou088-B10]\], and our results support this as well. Moreover, the contents of the collections could be well characterized in all the patients except one, who had air in the collection.
Importantly, we also found that USG, a cheap and widely available imaging modality, could also detect the WOPN in 92% of the study subjects. It could not detect WOPN in patients who had a collection that was predominantly solid or contained air. Also, these patients who had a predominantly solid WOPN underwent imaging within 6 weeks of the onset of symptoms.
USG, MRI and EUS could all quantify, as well as detect, the solid debris in all the patients who had detectable WOPN. There was no significant difference in the number of patients when classified by the percentage of solid necrotic debris detected by these imaging modalities. However, in patients with \<40% solid debris, MRI underestimated the amount of necrosis when compared with EUS and USG. EUS and USG detected \<10% solid content in 10 patients, whereas MRI detected \<10% solid content in 14 patients. EUS could also diagnose venous collaterals around the collection, which could not be identified by MRI and USG. Detection of venous collaterals around the collection is important as, during drainage of these collections, inadvertent puncture of the collaterals could lead to bleeding.
USG is not meant to replace cross-sectional imaging for the diagnosis of PFCs, but can help to better evaluate the morphology, as well as contents, of the PFCs. Patients with PFCs usually undergo EUS or MRI to detect the solid necrotic debris, since standard endoscopic drainage in the presence of solid necrotic debris produces poor results. We have reported in a separate study that patients with \<10% necrotic debris needed a single session of endoscopic drainage; patients with 10--40% necrotic debris needed multiple sessions of drainage for successful outcome, while patients with \>40% necrotic debris required DEN or surgical necrosectomy \[[@gou088-B9], [@gou088-B13]\].
In conclusion, USG can help in the characterization of the majority of WOPNs, with accuracy comparable to that of EUS and MRI. However, it suffers from certain limitations, including inability to detect collaterals around WOPN and inability to characterize collections with high solid content or air. EUS and MRI are comparable for characterization of WOPN but EUS is more accurate for detecting peri-WOPN collaterals.
Disclaimers
-----------
This study was presented at Digestive Disease Week 2014, Chicago, USA
Author contributions
--------------------
Surinder S Rana performed the data analysis and wrote the manuscript. Vinita Chaudhary, Ravi K Sharma, Puneet Chhabra and Vishal Sharma contributed to the data collection and analysis. Deepak K Bhasin contributed to the writing of the manuscript and the revision and editing of the article.
*Conflict of interest statement:* none declared.
[^1]: ^a^One case could not be well assessed on EUS because of the presence of air in the collection
[^2]: ^b^WOPN could not be visualized in four patients, among whom three had a high content of solid debris on EUS and one had air foci within the collection.
[^3]: EUS = endoscopic ultrasound; WOPN = walled-off pancreatic necrosis
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec0005}
===============
The inter-relationship between nutritional status and immune function continues to be the focus of research and debate [@bib0005; @bib0010]. It is well documented that acute and chronic deficiency of both macro- and micro-nutrients results in an impairment to a number of components of the immune system [@bib0015] and supplementation with individual micronutrients has proven efficacious as therapy for certain infectious morbidities; for instance vitamin A and measles infection [@bib0020], and zinc and diarrhoeal disease [@bib0025]. More recent research also suggests that supplementation with specific micronutrients may have non-specific deleterious effects on immune function, with iron [@bib0030] and vitamin A [@bib0035] specifically implicated. Further work to understand the mechanisms of these effects is required.
In addition to the effects of contemporaneous nutritional status on human immune function, recent evidence from our group and others suggests that nutritional status during fetal life and early infancy may be critical for immune development, with effects persisting into adulthood. Using antibody response to vaccination as a functional indicator of immunity, we have previously shown that adults born of a lower birth weight have a reduced antibody response to a polysaccharide vaccine (Typhim Vi) [@bib0040]. This deficit persisted following a second 'booster' dose of the vaccine [@bib0045] but no such association with size at birth was observed with either a protein (rabies) vaccine [@bib0040] or a polysaccharide-conjugate (Hib) vaccine [@bib0045]. This differential response suggests an early-life programming effect on the generation of antibodies during a B-cell-dependent immune response.
Much of the programming literature has focused on poor maternal nutrition as the most likely candidate for these early-life effects, and uses low birth weight as a proxy indicator for poor nutrition *in utero*. However, low birth weight may also be predictive of a number of post-natal factors that could also be implicated in defining later disease risk. Recent attention has focused on the association between an infant\'s rate of growth during early-infancy and later disease risk, with faster rates of post-natal 'catch-up' growth implicated as a possible causative factor for certain chronic disease outcomes [@bib0050]. The current study was therefore designed to investigate in more detail the relationship between nutritional status early in life and response to vaccination in young adults. Here, we investigate antibody response to two polysaccharide vaccines in a cohort of Gambian adults with detailed anthropometric data available from birth and from early infancy.
2. Materials and methods {#sec0010}
========================
2.1. Study population {#sec0015}
---------------------
Since 1949, the UK Medical Research Council (MRC) has been collecting health and demographic data on the populations of three villages (Keneba, Kantong Kunda and Manduar) in the rural West Kiang region of The Gambia. From 1976, and with the establishment of a permanent field station in Keneba, this data collection has incorporated detailed information on maternal and infant health, including birth anthropometry and infant growth. In the current study, our recruitment pool consisted of all adults, born in the three study villages since 1976 and who were aged 18 years or older on 1st January 2006. Subjects were excluded if they could not be traced or were not accessible for follow up, if they were already enrolled in another MRC study or if they were known to be pregnant at the time of recruitment. Ethical approval for the study was given by the Ethics Committee at the London School of Hygiene and Tropical Medicine and by the joint Gambian Government/MRC The Gambia Ethics Committee. Informed written consent was obtained from each individual participant.
The study took place between February and May 2006. Subjects were seen on two occasions, 14 days apart. At visit 1 (Day 0) weight, height, waist and hip circumferences were measured using standard equipment. A single sample of fasted venous blood was collected for measurement of plasma leptin and serum neopterin: leptin was measured as a proxy marker of adiposity and neopterin as a marker of immune activation. This blood sample was additionally used for the assessment of pre-vaccination serum antibody titres and for the preparation of a thick film for detection of malaria parasites by microscopy. Following blood collection, each subject was given a single dose of a Vi polysaccharide vaccine for *Salmonella typhi* (Sanofi-Pasteur, Lyon, France) and a single dose of a 23-valent capsular polysaccharide vaccine (Pneumo, Sanofi-Pasteur, Lyon, France). Fourteen days later (Visit 2), a further venous blood sample was collected for post-vaccination serum antibody titres.
2.2. Laboratory methods {#sec0020}
-----------------------
Plasma leptin and serum neopterin were measured at MRC Human Nutrition Research, Cambridge UK. Leptin was measured by ELISA (R&D Systems, Abingdon, UK) and neopterin by a competitive enzyme immunoassay principle (BRAHMS Atiengesellschaft, Berlin, Germany). Both analytes were measured in duplicate and following manufacturers' guidelines.
Anti-Vi immunoglobulin G (IgG) analysis was conducted at the Laboratory of Developmental and Molecular Immunity, National Institutes of Child Health and Human Development, Bethesda, USA. Briefly, microtitre plates were coated with Vi (0.2 μg/well) purified from *Citrobactor freundii* and goat anti-human IgG (Jackson Immuno Research Laboratories Inc., West Grove, PA) conjugated to alkaline phosphatase were used for ELISA. The anti-Vi IgG standard was a plasma sample from an adult vaccinated with Vi polysaccharide typhoid vaccine (provided by Wendy Keitel, Baylor University, Houston, TX). The Vi antibody content of this serum was also assayed by a radioimmunoassay (RIA) by Pasteur Merieux Connaught. The antibody levels were expressed in ELISA units (EU) and the reference sera were assigned a value of 75 EU. All samples were run in duplicate. Antibody levels were calculated using Program ELISA, version 12 (Center for Disease Control and Prevention, Atlanta, GA). The lowest detectable level of the assay for anti-Vi IgG was 0.1 EU. Prior to analysis, all data were log transformed, and results are presented as geometric means. For anti-Vi antibody levels, data are expressed as ELISA units (EU).
Pneumococcal capsular polysaccharide specific IgG levels were measured at the WHO Pneumococcal Serology Reference Lab at the UCL Institute of Child Health, London, UK. Standard enzyme linked immunosorbent assay methods [@bib0055] were used to quantify anticapsular IgG antibodies to four specific pneumococcal serotypes (1, 5, 14 and 23F). These serotypes were selected on the basis of frequency of carriage within this population setting, 14 and 23F being amongst the most common [@bib0060], and their importance in causing invasive disease (1 and 5 account for \>40% in a recent series of pneumococci causing bacteraemia [@bib0065]).
2.3. Statistical analyses {#sec0025}
-------------------------
Comparisons amongst group means were made using two-sample *t*-tests. Vaccine data are presented as geometric means and 95% confidence intervals (CIs). Sex specific *z*-scores were calculated using UK reference data [@bib0070]. Associations between contemporary measures and antibody response to vaccination were compared by linear (for continuous variables) or logistic (for binary variables) regression analysis. Response to vaccination was assessed in relation to six early-life exposures (separate models); birth weight, low birth weight (\<2.5 kg) *vs*. normal birth weight (as a binary variable), small for gestational age *vs*. appropriate for gestational age (as a binary variable), rate of infant growth from birth to three months of age, infant weight at 12 months of age and season of birth (harvest/wet season January--June; hungry/dry season July--December). Rate of change in weight from birth to three months was calculated as the difference between sex-specific birth weight standard deviation score and sex-specific weight at three months standard deviation score. We also looked at weight for age standard deviation differences between three and six months of age and six and 12 months of age. Associations between these early-life exposures and antibody responses were tested by multiple linear regression analysis. Probability values \<0.05 were considered to be statistically significant for all tests. All statistical analyses were performed using DataDesk, version 6 for Windows, Data Description Inc., Ithaca, NY.
3. Results {#sec0030}
==========
3.1. Subject characteristics {#sec0035}
----------------------------
A total of 858 individuals met the criteria for recruitment into the current study. Of these, 78 were known to have died prior to follow up, leaving a cohort of 781 to be traced. Of this number, 145 were excluded on the basis they were currently participating in another ongoing study and three because they were confirmed to be pregnant by an MRC midwife prior to the start of the study. Of the remaining 633 individuals who were eligible to participate, 241 were not available \[dead (4), self-confirmed as pregnant (45), overseas (24), outside designated study area (58), not traceable (50), traceable but unavailable for study (60)\] and 72 did not consent to participate. A total of 320 subjects (41% of 781 followed up) consented and participated in the current study. Compared to non-participants, participants were younger (22.2 y *vs*. 23.0 y; *p* \< 0.0001) and there were significantly more males than females (51.9% *vs*. 45.3%). No differences were observed between the participants and non-participants in available early-life information (data not presented). [Table 1](#tbl0005){ref-type="table"} details the early-life characteristics of the subjects recruited. A total of 41 (12.8%) of subjects were born of a low birth weight (\<2.5 kg), and a higher proportion of these were female. Of these, 13 were born pre-term (\<37 weeks gestation), although 9 had a missing gestational age. A total of 267 (83%) of the cohort had gestational age assessments available. Using the William\'s reference data [@bib0075], 51 (19%) of these infants would be considered small for gestational age (SGA). Male subjects were significantly heavier at three months and at 12 months of age, but the rate of early growth, expressed as the sex-specific change in *z*-score between birth and three months of age, three to six months, or six to twelve months did not differ between males and females.
Characteristics of the study participants at follow up are detailed in [Table 2](#tbl0010){ref-type="table"}. Male participants were significantly taller and heavier than females, but had a lower mean Body Mass Index (BMI) and plasma leptin level. Significantly more of the males lived in urban areas of The Gambia compared to females, and the distribution of month of study differed between the males and females recruited. No differences were observed in age, waist:hip ratio, or serum neopterin levels between the male and female subjects.
3.2. Antibody response to vaccination {#sec0040}
-------------------------------------
Pre- and post-vaccination geometric mean (95% CI) data for both the pneumococcal and Vi vaccine are detailed in [Table 3](#tbl0015){ref-type="table"}. A total of 112 subjects (37.2%) did not achieve antibody titres \>3.52 EU following Vi vaccination, the estimated level for 90% protection. Using a post-vaccination anti-pneumococcal IgG titre of \>0.35 μg/mL, the level considered indicative of putative protection, all subjects achieved an adequate response to all serotypes.
Simple univariate regression analysis was used to test for unadjusted associations between antibody response to vaccination and the contemporary variables measured at the time of vaccination; sex, age, location (rural *vs*. urban), weight, height, BMI, plasma leptin, month of study (February, March, April, May), malaria parasitaemia (+ve *vs*. −ve), and serum neopterin levels ([Table 4](#tbl0020){ref-type="table"}). Pre-vaccination antibody titres were also included as a potential confounder in all of the models. Variables showing significant associations with antibody response to vaccination were then fitted into a multivariate model; those variables that remained significant are as detailed in [Table 4](#tbl0020){ref-type="table"}. Only those variables that remained significant predictors of antibody response were then added to the models looking at early-life influences on response to vaccination.
We did not predict, *a priori*, that pre-vaccination antibody levels would have such a strong influence on post-vaccination antibody responses. However, and as pre-vaccination levels could themselves be predicted by early life exposures (through immune responses to infection), we repeated the analysis (a) looking at predictors of pre-vaccination levels *per se*, and (b) removing pre-vaccination levels from the final model of predictors of post-vaccination levels. Following adjustment for contemporary factors shown to be associated with pre-vaccination levels, the only significant association observed was between infant weight at 12 months of age and pre-vaccination levels to pneumococcal serotypes 5 and 23 (*p* = 0.028 and 0.016 respectively; analyses not presented). The results of the regression analysis excluding pre-vaccination levels are included in [Table 5](#tbl0025){ref-type="table"}.
Associations between early-life exposures and antibody responses to vaccination were tested by multiple linear regression analysis, adjusting for the contemporary variables identified as predictive of antibody responses. [Table 5](#tbl0025){ref-type="table"} highlights the unadjusted and adjusted results of multiple linear regression analysis using birth weight, low birth weight (\<2.5 kg) *vs*. normal birth weight, rate of infant growth from birth to three months of age, infant weight at 12 months of age, and season of birth (hungry *vs*. harvest) as dependent variables (separate models employed for each variable). No significant associations were observed between the early-life data and antibody response to vaccination with either a Vi polysaccharide vaccine or with serotypes 1, 5 and 23f of the pneumococcal polysaccharide vaccine. For serotype 14, no associations were observed with birth weight or low birth weight, but a trend towards significance was observed for infant growth from birth to three months of age (negative trend), infant weight at 12 months of age (negative trend) and season of birth (higher in hungry season births). The analyses were also performed using change in weight-for-age standard deviation scores between three and six, and six and twelve months of age. No significant associations were observed, with the exception of a marginally significant relationship between rate of growth between six and twelve months of age and antibody response to serotype 14, when adjusted for pre-vaccination antibody levels (*β* = −0.116, *p* = 0.043; other data not presented).
4. Discussion {#sec0045}
=============
Recent research has highlighted a possible association between nutritional status in early-life and development of the human immune system, with long-term programming effects on immune function inferred [@bib0080]. Studies in Gambian [@bib0085] and Bangladeshi [@bib0090] infants have shown correlations between pre- and post-natal nutritional and environmental exposures and development of the thymus during early infancy. In The Gambia, these alterations in thymic size were reflected by changes in both lymphocyte subpopulation counts [@bib0095] and in levels of signal-joint T-cell receptor rearrangement circles (sjTREC), an indirect marker of thymic output, suggesting an effect on thymic function [@bib0100]. Of importance, this early-life effect appears to persist beyond infancy. Results from studies in adolescents from the Philippines [@bib0105] and in adults from Pakistan [@bib0040; @bib0045] indicate a positive association between birth weight and antibody response to a Vi polysaccharide vaccine for *S. typhi*. In the study in Pakistan, no association however was observed in antibody response to either a rabies (protein) vaccine [@bib0040] or polysaccharide conjugate (conjugated *H. influenzae* type b (Hib) vaccine) vaccine [@bib0045]. These contrasting effects suggest that antibody generation to polysaccharide antigens, which have greater B-cell involvement, may be compromised by fetal growth retardation.
The current study was specifically designed to explore the relationship between markers of both pre-and post-natal nutritional status and antibody response to polysaccharide antigen vaccines in adults born in rural Gambia. In this cohort of 320 young Gambian adults, no associations were observed between birth weight, low birth weight (\<2.5 kg), small for gestational age, rate of growth from birth to three months of age, infant weight at 12 months of age, or season of birth with antibody response to the Vi polysaccharide vaccine or serotypes 1, 5 and 23f of the pneumococcal vaccine. Antibody responses to serotype 14 of the vaccine however were higher amongst infants who were smaller at 12 months of age and showed slower growth between 3 and 12 months of age. In addition, infants born during July to December (the 'hungry' season) had higher antibody titres to serotype 14. The data from this study offer only limited support an early-life programming effect of early nutrition on antibody response to vaccination in adulthood within this environment.
The observed associations between early life exposures and response to serotype 14 of the pneumococcal vaccine only are rather difficult to explain. Globally, serotype 14 is the most important serotype causing disease worldwide, although carriage rates vary between populations [@bib0060; @bib0110; @bib0115]. Of the 4 serotypes assessed in the current study (1, 5, 14 and 23f), exposure to 23F and 14 are most likely similar and so early exposures during infancy are unlikely to explain the difference. Technically, type 14 is the 'purest' serotype to assay, with little cross-reaction with other serotypes when measured in ELISA (D Goldblatt, personnel communication), but it is unlikely that this alone explains the observed differences. Selection of serotypes was primarily based on carriage rates amongst infants in The Gambia. However, and since it is known that pneumococcal carriage is not equally distributed between adults and children in this population, and is also variable by age (for infants) and season [@bib0120], knowledge of precise carriage rates (through nasopharyngeal swabs) at the time of vaccination may have been informative. Inclusion of additional serotypes, such as those known to elicit a 'weak' response may help explain this observation. Indeed, previous research has identified serotype 6B as being sensitive to modulation by infant feeding status[@bib0125], following vaccination with a conjugated vaccine. Such serotypes may, therefore, be more sensitive to nutritional exposures early in life.
In interpreting the results presented here, consideration should be given to the limitations of the current study. Much of the programming literature in based on the follow up of cohorts of adults for whom records from early-life are available. In The Gambia, the UK Medical Research Council (MRC) has been maintaining demographic and health-related records for three rural villages since 1949 [@bib0130]. From 1976, these records have included detailed information on maternal and infant health, allowing the study of early-life predictors of current health within this population. However, as with many studies within this field [@bib0135], the current study suffered with considerable loss to follow up. A total of 78 (9%) of the 858 subjects born during the study period were known to have died prior to the start of fieldwork. In addition, we were only able to recruit 41% of the remaining 781 subjects available for follow up. Whilst no differences were observed between the early-life data for those subjects recruited and those available but not included, these subjects only represent the surviving cohort. A major limitation therefore is that the subjects recruited do not provide a true representation of the original cohort; indeed, birth weights amongst subjects who were known to have died prior to follow up were significantly lower than those listed as available for follow up (2.58 kg *vs*. 2.97 kg; ≤0.0001), perhaps indicating that the more vulnerable subjects had already been lost from the cohort. A further limitation of this study design is the lack of any direct measure of early-life nutritional exposures in these subjects, including the assessment of breast feeding practices. Whilst it might be assumed, based on the literature from this population [@bib0140; @bib0145], that all subjects would have been initially exclusively breast fed, followed by a period of extensive breast feeding, given the literature on the association of early breast feeding practices and later antibody response to vaccination *e.g.* [@bib0150], the lack of any detailed information must be viewed as a limitation. Indeed, a strong criticism of much of the programming field is the lack of direct data assessing the impact of nutritional exposures on health outcomes and the reliance on observational data. Future work could usefully focus on cohorts for whom direct measures of early-life nutritional exposures are available, such as the follow-up of randomized control trials of pre- or post-natal nutritional supplementation, and also incorporate more detailed measures of cellular immunity, to help interpret vaccine response data.
To understand the differential results between this study in The Gambia and our previous observations from Pakistan, differences in study design and cohort characteristics need consideration. Firstly, the Gambian adults were significantly younger than the adults in Lahore (mean age 22.3 y *vs*. 29.4 y; *p* ≤ 0.0001) and so it is possible that their relative immaturity contributed to these findings. This, however, seems unlikely since a further study in adolescents from the Philippines (mean age 14.6 y) also observed a positive association between birth weight and antibody response to the same Vi vaccine [@bib0105]. In the current study, the geometric mean (GM) post-vaccination anti-Vi antibody titre was 7.1 EU whilst in Pakistan the GM was 5.9 EU (unadjusted difference between means *p* = 0.1383): in both countries, post-vaccination levels were measured 14 days following vaccination. Although this difference in GMs is not statistically significant, it is possible that it may contribute to the lack of an association in the current study, perhaps by suggesting these young Gambian adult were able to mount an overall improved response to vaccination, diminishing the potential impact of the early-life environment. The most consistent predictor of antibody response to vaccination in the current study was pre-vaccination antibody levels. However, pre-vaccination levels were very similar between the two studies (Gambia GM = 0.56 EU, Pakistan GM = 0.53 EU; *p* = 0.8327) and so unlikely to explain the lack of association with birth weight observed in the current study. Relative differences in relation to the pneumococcal vaccine cannot be compared since this vaccine was not used in the study in Pakistan.
In the current study we observed an interesting effect of a number of contemporaneous measures and antibody response to both vaccines. When combined in multiple regression analyses, the measures shown to have the most significant effects were serum neopterin and plasma leptin levels, and pre-vaccination antibody titres. Neopterin is a macrophage-derived protein commonly used as a marker of immune activation, and elevated levels of peripheral blood neopterin indicate an unregulated cellular immune response. In the current study, serum levels of neopterin independently and positively predicted antibody response to serotypes 1 and 5 of the pneumococcal vaccine, but not to serotypes 14 and 23F or the response to the Vi vaccine. Although it is difficult to explain why individuals with elevated immune activation responded more effectively to these two serotypes only, we speculate that an enhanced vaccine response in subjects could be the result of a co-stimulatory effect of an already elevated state of immune activation. Whether such an effect has any longer term implication on antibody titres, remains to be determined.
Leptin, a primarily adipocyte-derived hormone, was positively correlated with serotype 14 of the pneumococcal vaccine but not with the response to any other serotypes or the Vi vaccine. Leptin levels correlate with body fat mass and leptin has more recently been implicated as a central mediator connecting nutrition to immunity [@bib0010]. Data from animal models have suggested that leptin may mediate the effects of malnutrition on T cell function [@bib0155; @bib0160], although little data currently exists to suggest that these effects translate into compromised specific immune responses in malnourished humans (*e.g.* [@bib0165]). Further work may be warranted to help understand the specific relationship between plasma leptin levels and antibody response to serotype 14 of the pneumococcal vaccine.
With the exception of antibody response to serotype 23F of the pneumococcal vaccine, a highly significant effect of pre-vaccination antibody levels on post-vaccination titres was observed for both vaccines. Pre-vaccination antibody titres are a consequence of previous exposure to the vaccine antigens; for pneumococcal serotypes this is mainly *via* exposure to the same or similar serotypes encountered during nasopharyngeal carriage. A longitudinal study of households in the UK showed strong immune response to the carriage serotype, supporting the assumption that natural immunity to *Streptococcus pneumoniae* is induced by exposure to *S. pneumoniae* [@bib0170].
In conclusion, we have shown that antibody response to 2 polysaccharide vaccines is not influenced by the birth weight, size in infancy or season of birth in rural Gambia, suggesting that young Gambian adults are able to mount an adequate response to both vaccines, irrespective of their early life environment. In addition to the predictive capacity of pre-vaccination antibody levels, these data suggest a role of immune activation and plasma leptin in antibody response to vaccination, but these observations were not consistent between vaccines.
We are grateful to all the subjects who participated in this research project. We also thank the field staff from MRC Keneba for their assistance with this study. We acknowledge the role of the Nutritional Biochemistry Laboratory, MRC Human Nutrition Research, Cambridge in running the leptin and neopterin assays. This study was financed by the UK Medical Research Council. The vaccines were kindly donated by Sanofi-Pasteur, Lyon, France.
######
Early-life variables.
Males (*n* = 166) Females (*n* = 154)
---------------------------------------- ------------------- -------------------------------------------------
Birth weight (kg) 3.10 (0.42) 2.87 (0.41)[a](#tblfn0005){ref-type="table-fn"}
Low birth weight (%) 8.43 17.5[b](#tblfn0010){ref-type="table-fn"}
Gestational age (wks) 38.9 (1.43) 38.6 (1.56)
Small for gestational age (%) 16.9 21.4
Hungry season births (%) 51.8 53.9
Wt at 3 months (kg) 5.92 (0.84) 5.35 (0.79)[a](#tblfn0005){ref-type="table-fn"}
Change in *z*-score, birth to 3 months 0.43 (0.99) 0.36 (0.99)
Change in *z*-score, 3--6 months −0.56 (0.72) −0.55 (0.67)
Change in *z*-score, 6--12 months −0.84 (0.80) −0.72 (0.88)
Wt at 12 months (kg) 8.28 (1.04) 7.65 (1.03)[a](#tblfn0005){ref-type="table-fn"}
All data are means (±SD) or percentages of total. Hungry season = July--December inclusive.
Gestational age (and small for gestational age) data only available for *n* = 267 (83%) of total cohort.
Significantly different from males ≤0.0001.
Significantly different from males ≤0.05.
######
Subject characteristics.
Males (*n* = 166) Females (*n* = 154)
----------------------------------------------------------- -------------------- ---------------------------------------------------------
Age (y) 22.1 (2.77) 22.6 (3.02)
Weight (kg) 60.8 (8.67) 55.7 (8.69)[a](#tblfn0015){ref-type="table-fn"}
Height (cm) 173.3 (722) 160.3 (6.77)[a](#tblfn0015){ref-type="table-fn"}
BMI (kg/m^2^) 20.2 (2.19) 21.7 (3.01)[a](#tblfn0015){ref-type="table-fn"}
Waist:hip ratio 0.77 (0.04) 0.77 (0.05)
Plasma leptin (ng/mL)[b](#tblfn0020){ref-type="table-fn"} 0.73 (0.64--0.84) 9.68 (8.51--11.05)[a](#tblfn0015){ref-type="table-fn"}
Rural living (%) 15.1% 31.8%[b](#tblfn0020){ref-type="table-fn"}
Month of study F/M/A/M (%) 3.0/21.7/24.7/50.6 18.8/19.5/33.8/27.9[a](#tblfn0015){ref-type="table-fn"}
Malaria parasite positive 2/166 0/154
Neopterin (nmol/L) 7.37 (7.10--7.64) 7.51 (7.25--7.79)
All data are means (±SD) or percentages of total. Leptin and neopterin data are reported as geometric means (95% CI). Month of Study data are report as percent of total per month.
Significantly different from males ≤0.0001.
Significantly different from males ≤0.001.
######
Pre-and post-vaccination antibody concentrations.
IgG anti Vi antibody concentrations (EU) IgG anti-pneumococcal antibody concentrations (μg/mL)
--------------- ------------------------------------------ ------------------------------------------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------
*n* 316 301 315 302 316 302 316 302 316 302
GM 0.56 7.19 1.28 10.17 1.92 6.45 8.49 45.5 1.62 7.01
95% CI 0.49--0.63 6.11--8.46 1.18--1.40 9.05--11.4 1.76--2.09 5.83--7.13 7.46--9.67 40.9--50.5 1.45--1.81 6.23--7.88
Non responder 37.2% 0% 0% 0% 0%
GM -- geometric mean; EU -- Elisa Units.
Non-responders: response to Vi vaccination, post-vaccination titres \< 3.52 EU considered as a non-responder. For pneumococcal vaccination, post-vaccination titres \< 0.35 μg/mL considered as a non-response.
######
Associations between contemporary variables and post vaccination antibody concentrations.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Variable Vi Type 1 Type 5 Type 14 Type 23F
--------------------------------- -------------- -------------- --------------- -------------- -------------- --------------- ------------------------- --------------- -------------------------- --------------
Sex *r* = 0.125\ *r* = 0.170 *p* = 0.003 *r* = 0.166 *p* = 0.0037
*p* = 0.030
Age *r* = 0.133\ *β* = −0.041\
*p* = 0.021 *p* = 0.006
Location
Weight *r* = −0.160\
*p* = 0.005
Height *r* = −0.188\ *β* = −0.002\ *r* = −0.141\
*p* ≤ 0.001 *p* = 0.038 *p* = 0.012
BMI *r* = 0.171\
*p* = 0.0005
Leptin *r* = 0.164\ *r* = 0.112\
*p* = 0.0089 *p* = 0.052
Month of study
Malaria
Neopterin *r* = 0.153\ *β* = 0.500\ *r* = 0.125\ *β* = 0.476\
*p* = 0.0083 *p* = 0.025 *p* = 0.032 *p* = 0.017
Pre-vaccination antibody levels *r* = 0.664\ *β* = 0.696\ *r* = 0.566\ *β* = 0.696\ *r* = 0.495\ *β* = 0.544\ *r* = 0.405\ *β* = 0.302\ *r* = 0.559\ *β* = 0.568\
*p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001 *p* ≤ 0.0001
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Data in columns labelled \[1\] represent univariate comparisons between variable and antibody response to vaccination. Data in columns labelled \[2\] represents output from multivariate analysis where significant variables from \[1\] were fitted within the same model. Blank cells represent non-significant associations.
######
Association between early-life variables and post-vaccination antibody concentrations.
Vi Type 1 Type 5 Type 14 Type 23F
---------------------------- ---------- --------- ---------- --------- ---------- --------- ---------- --------- ---------- ---------
Birth weight −*0.246* *0.210* −*0.019* *0.166* −*0.056* *0.645* *0.015* *0.905* −*0.120* *0.393*
−0.222 0.124 −0.131 0.262 −0.008 0.938 0.165 0.185 −0.011 0.927
−0.246 0.210 −0.156 0.266 0.021 0.413 0.015 0.905 −0.120 0.393
Low birth weight −*0.200* *0.415* −*0.156* *0.375* −*0.159* *0.292* −*0.180* *0.252* *0.025* *0.889*
−0.008 0.966 −0.051 0.730 −0.139 0.292 −0.272 0.065 −0.090 0.543
−0.200 0.415 −0.183 0.297 −0.191 0.223 −0.180 0.252 0.025 0.889
Small for gestational age *0.056* *0.814* −*0.032* *0.849* −*0.203* *0.182* −*0.164* *0.280* *0.242* *0.146*
0.098 0.568 −0.113 0.416 −0.217 0.100 −0.179 0.187 −0.080 0.584
0.056 0.814 −0.034 0.837 −0.167 0.274 −0.234 0.117 0.242 0.146
Early catch-up growth −*0.011* *0.892* −*0.025* *0.674* −*0.059* *0.272* −*0.140* *0.009* −*0.006* *0.920*
0.027 0.662 0.024 0.630 −0.042 0.391 −0.121 0.015 0.026 0.645
−0.011 0.892 0.006 0.926 −0.035 0.521 −0.118 0.026 −0.006 0.920
Infant weight at 12 months −*0.045* *0.567* −*0.086* *0.132* −*0.071* *0.151* −*0.134* *0.008* −*0.073* *0.204*
−0.026 0.655 −0.019 0.697 −0.025 0.566 −0.102 0.054 0.601 0.876
−0.045 0.567 −0.058 0.316 −0.090 0.078 −0.134 0.008 −0.073 0.204
Season of birth *0.075* *0.652* *0.147* *0.218* *0.048* *0.642* *0.234* *0.028* −*0.058* *0.630*
−0.031 0.802 0.092 0.350 −0.003 0.975 0.223 0.021 −0.076 0.444
0.075 0.652 0.135 0.254 0.031 0.771 0.234 0.028 −0.058 0.630
Data in italics in first row represents unadjusted results from multiple linear regression analysis, data in second row represents analysis adjusting for contemporary variables correlated with antibody response to vaccination. Data in third row represents analysis adjusting for contemporary variables correlated with antibody response to vaccination but *excluding* pre-vaccination antibody titres.
Low birth weight--binary variable of subjects with a birthweight of \<2.5 kg *vs*. subjects with a birth weight of ≥2.5 kg.
Early-catch up growth is defined as the change in sex-specific standard deviation scores between birth and 3 months of age.
Season of birth: hungry (July--December) *vs*. harvest (January--June).
[^1]: Current address: Department of Social Medicine, University of Bristol, Bristol, UK.
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper and its Supporting Information files.
Introduction {#sec001}
============
Detailed reaction-diffusion models to describe human atrial electrophysiology were first developed in the late 1990s \[[@pone.0190448.ref001]--[@pone.0190448.ref004]\] and are further developed until now. Important steps forward have been made to include specific ionic currents \[[@pone.0190448.ref005]--[@pone.0190448.ref010]\], which in particular allow one to investigate specific effects of pharmaceuticals in treatments of atrial fibrillation and other heart failures. Complementary to these detailed models, Bueno-Orovio, Cherry and Fenton introduced in 2008 a minimal reaction-diffusion model (BOCF model) for action potentials (AP) in ventricular electrophysiology, where the large number of ionic currents through cell membranes is reduced to three net currents \[[@pone.0190448.ref011]\]. This model has four state variables, one describing the transmembrane voltage (TMV), and the other three describing the gating of ionic currents. The TMV, as in detailed reaction models, satisfies a partial differential equation of diffusion type with the currents acting as source terms, and the time evolution of the gating variables is described by three ordinary differential equations coupled to the TMV. By fitting the action potential duration (APD), the effective refractory period and the conduction velocity to the detailed model of Courtemanche, Ramirez and Nattel \[[@pone.0190448.ref001]\] (CRN model), the BOCF model was recently adapted to atrial electrophysiology (BOCF model) \[[@pone.0190448.ref012]\].
In this work we develop a method to model specific AP based on the BOCF model as it is aimed in the clinical context in connection with improved and extended possibilities of diagnosis \[[@pone.0190448.ref013]\]. Compared to the detailed models, the BOCF model has the advantage that it is better amenable to some analytical treatment. This allows us to identify a small set of relevant model parameters for capturing the main features of a specific AP. Our methodology is sketched in [Fig 1](#pone.0190448.g001){ref-type="fig"} and can be summarized as follows. We start by labeling each given AP with its amplitude APA and with four APD, namely at 90%, 50%, 40% and 20% repolarization, denoted as APD~90~, APD~50~, APD~40~, and APD~20~ respectively. These APD~*n*~ (*n* = 20, 40, 50, 90) together with the amplitude APA are suitable to catch a typical shape of a specific AP, see [Fig 2](#pone.0190448.g002){ref-type="fig"}.
{#pone.0190448.g001}
{#pone.0190448.g002}
The APD~*n*~ taken for a specific patient are given to a parameter convertor that retrieves specific parameter values of the BOCF model. As relevant parameters, we adjust three time scales governing the closing and opening of the ionic channels. The parameter convertor consists of an optimization algorithm that searches for the best set of parameter values consistent with the measured AP properties.
The paper is organized as follows. In Section "BOCF model for atrial physiology" we shortly summarize the BOCF model and discuss the role of the three fit parameters that we selected to model specific AP. In Section "Parameter dependence of BOCF action potentials" we show how these parameters can be adjusted to obtain a a faithful representation of the AP properties APA, APD~*n*~, and in Section "Modeling of patient-specific action potentials with the BOCF model" we demonstrate the specific AP modeling for surrogate data generated with the CRN model \[[@pone.0190448.ref001]\]. A summary of our main findings and discussion of their relevance is given in Section "Conclusions". In the supporting information, we provide analytical calculations for the BOCF model that motivated our choice of fit parameters for the AP modeling. We also analyze the robustness of the optimization procedure with the activation frequency.
BOCF model for atrial physiology {#sec002}
================================
The BOCF model has four state variables, which are the scaled TMV *u*, and three variables *v*, *w* and *s* describing the gating of (effective) net currents through the cell membrane. The TMV *V* is obtained from *u* via the linear relation *V* = *V*~*R*~(1 + *αu*), where for atrial tissue we set *V*~*R*~ = −84.1 mV for the resting potential and *α* = 1.02 \[[@pone.0190448.ref012]\]. The time-evolution of *u* is given by the single-cell action potential model, here defined as $$\begin{array}{r}
{\partial_{t}u = J\left( u,v,w,s \right) + J_{\text{stim}}\,,} \\
\end{array}$$ where *J* = *J*(*u*, *v*, *w*, *s*) is the total ionic current and *J*~stim~ an external stimulus current. The total ionic current decomposes into three net currents, a fast inward sodium current *J*~fi~ = *J*~fi~(*u*, *v*), a slow inward calcium current *J*~si~(*u*, *w*, *s*), and a slow outward potassium current *J*~so~ = *J*~so~(*u*), $$\begin{array}{r}
{J\left( u,v,w,s \right) = J_{\text{fi}}\left( u,v \right) + J_{\text{si}}\left( u,w,s \right) + J_{\text{so}}\left( u \right)\,.} \\
\end{array}$$ These currents are controlled by the gating variables, which evolve according to $$\begin{array}{r}
{\partial_{t}\left( v,w,s \right) = \left( E\left( u,v \right),F\left( u,w \right),G\left( u,s \right) \right)\,,} \\
\end{array}$$ where the nonlinear functions *F*, *G* and *H*, are specified in [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"}. There we show that the four differential Eqs ([1](#pone.0190448.e001){ref-type="disp-formula"}) and ([3](#pone.0190448.e003){ref-type="disp-formula"}) can be reduced to a system of two differential equations. This reduction shows that the three characteristic times *τ*~fi~, *τ*~si~ and *τ*~so1~, which fix the typical duration of the respective currents, govern the shape of the AP \[cf. [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"}\]. We take these three time scales as parameters for fitting a specific AP and keep all other parameters fixed. For the values of the fixed parameters we here consider the set determined for the electrically remodeled tissue due to atrial fibrillation \[[@pone.0190448.ref012], [@pone.0190448.ref014]\].
Parameter dependence of BOCF action potentials {#sec003}
==============================================
In this section we show that in the BOCF model the amplitude APA can be expressed by a quadratic polynomial of the times *τ*~fi~, and the APD~*n*~ by cubic polynomials of *τ*~si~ and *τ*~so1~.
The dependence of APA and the APD~*n*~ on the characteristic times, was determined from generated AP in single-cell simulations of the BOCF model by applying periodically, with a frequency *f* = 3 Hz, a square stimulus current of 40 pA, corresponding to an amplitude of 4.76 s^−1^ for the current *J*~stim~ in [Eq (1)](#pone.0190448.e001){ref-type="disp-formula"}, for a time period of 3.5 ms. The resulting time evolution of the TMV in response to this stimulus was calculated by integrating Eqs ([1](#pone.0190448.e001){ref-type="disp-formula"}) and ([3](#pone.0190448.e003){ref-type="disp-formula"}) for the initial conditions *u*~0~ = 0, *v*~0~ = 1, *w*~0~ = 1 and *s*~0~ = 0. This was done for (*τ*~fi~, *τ*~si~, *τ*~so1~)∈\[0.002, 0.210\] × \[5.9, 22.4\] × \[40, 110\] (in ms) with a resolution Δ*τ*~fi~ = 0.0021 ms (100 values), Δ*τ*~si~ = 0.3 ms (56 values) and Δ*τ*~so1~ = 1 ms (71 values). The AP was recorded after a transient time of 10 s.
As shown for a few representative pairs of fixed values of *τ*~fi~ and *τ*~so1~ in [Fig 3(a) and 3(b)](#pone.0190448.g003){ref-type="fig"}, the APA depends only very weakly on *τ*~si~ and *τ*~so1~. Neglecting these weak dependencies, on *τ*~si~ and *τ*~so1~, we find the APA to increase monotonically with *τ*~fi~ in the range \[85, 110\] mV relevant for human atria. In [Fig 3(c)](#pone.0190448.g003){ref-type="fig"} we show that the parameter *τ*~fi~ can be well described by the quadratic polynomial $$\begin{array}{r}
{\tau_{\text{fi}} = c_{0}\text{APA}^{2} + c_{1}\text{APA} + c_{2}\,,} \\
\end{array}$$ where the coefficients *c*~*i*~ and the coefficient of determination *R*^2^ of the fit are given in [Table 1](#pone.0190448.t001){ref-type="table"}.
{#pone.0190448.g003}
10.1371/journal.pone.0190448.t001
###### Polynomial coefficients and *R*^2^ values of the fits of APA to [Eq (4)](#pone.0190448.e004){ref-type="disp-formula"} and of the surfaces APD~*n*(*τ*~si~,\ *τ*~so1~)~ to [Eq (5)](#pone.0190448.e026){ref-type="disp-formula"}.
The values of coefficients $c_{mk}^{(n)}$ are given in units of mV/(ms)^*m*+*k*^.
{#pone.0190448.t001g}
Coeffs. [Eq (4)](#pone.0190448.e004){ref-type="disp-formula"} APA Coeffs. [Eq (5)](#pone.0190448.e026){ref-type="disp-formula"} APD~90~ APD~50~ APD~40~ APD~20~
--------------------------------------------------------------- ----------------------- --------------------------------------------------------------- -------------------- -------------------- -------------------- --------------------
*c*~0~ ± Δ*c*~0~ 0.000235 ± 6 × 10^−6^ $c_{00}^{(n)}$ $\pm \Delta c_{00}^{(n)}$ 98 ± 10 85 ± 10 84 ± 10 82 ± 10
*c*~1~ ± Δ*c*~1~ -0.038 ± 0.001 $c_{10}^{(n)}$ $\pm \Delta c_{10}^{(n)}$ 5.4 ± 0.3 5.0 ± 0.3 4.7 ± 0.4 3.8 ± 0.3
*c*~2~ ± Δ*c*~2~ 1.52 ± 0.05 $c_{01}^{(n)}$ $\pm \Delta c_{01}^{(n)}$ −33 ± 1 −33 ± 1 −33 ± 1 −32 ± 1
**R**^**2**^ **0.9996** $c_{20}^{(n)}$ $\pm \Delta c_{20}^{(n)}$ 0.0001 ± 0.004 −0.0010 ± 0.004 0.0001 ± 0.004 0.003 ± 0.004
$c_{11}^{(n)}$ $\pm \Delta c_{11}^{(n)}$ −0.40 ± 0.01 −0.41 ± 0.01 −0.41 ± 0.01 −0.43 ± 0.01
$c_{02}^{(n)}$ $\pm \Delta c_{02}^{(n)}$ 2.47 ± 0.06 2.56 ± 0.06 2.61 ± 0.07 2.85 ± 0.06
$c_{30}^{(n)}$ $\pm \Delta c_{30}^{(n)}$ −0.00007 ± 0.00002 −0.00005 ± 0.00002 −0.00004 ± 0.00002 −0.00002 ± 0.00002
$c_{21}^{(n)}$ $\pm \Delta c_{21}^{(n)}$ 0.00125 ± 0.00007 0.00096 ± 0.00007 0.00079 ± 0.00007 0.00018 ± 0.00007
$c_{12}^{(n)}$ $\pm \Delta c_{12}^{(n)}$ 0.0027 ± 0.0003 0.0045 ± 0.0003 0.0057 ± 0.0003 0.0103 ± 0.0003
$c_{03}^{(n)}$ $\pm \Delta c_{03}^{(n)}$ −0.045 ± 0.001 −0.050 ± 0.001 −0.053 ± 0.001 −0.069 ± 0.001
**R**^**2**^ **0.9956** **0.9938** **0.9926** **0.9866**
Likewise, as demonstrated in [Fig 4(a)](#pone.0190448.g004){ref-type="fig"} for one fixed pair of values of *τ*~si~ and *τ*~so1~, the APD~*n*~ are almost independent of *τ*~fi~. Their dependence on *τ*~si~ and *τ*~so1~, shown in [Fig 4(b)--4(e)](#pone.0190448.g004){ref-type="fig"}, can be well fitted by the polynomials $$\begin{array}{r}
{\text{APD}_{n}\left( \tau_{\text{si}},\tau_{\text{so}1} \right) = \sum\limits_{m = 0}^{3}\sum\limits_{k = 0}^{3 - m}c_{mk}^{(n)}\tau_{\text{si}}^{m}\tau_{\text{so}1}^{k}\,,} \\
\end{array}$$ where the coefficients $c_{mk}^{(n)}$ are listed in [Table 1](#pone.0190448.t001){ref-type="table"} together with the *R*^2^ values of the fits. [Fig 4(f)--4(i)](#pone.0190448.g004){ref-type="fig"} display contour plots of the APD surfaces, shown in [Fig 4(b)--4(e)](#pone.0190448.g004){ref-type="fig"}.
{ref-type="disp-formula"}. All quantities are given in ms. Plots (f)-(i) show contour plots of the APD surfaces in (b)-(e), respectively.](pone.0190448.g004){#pone.0190448.g004}
The APD~*n*~ of the single cell BOCF model depend on the activation frequency *f* or basic cycle length BCL = 1/*f*. Corresponding restitution curves are shown in [Fig 5](#pone.0190448.g005){ref-type="fig"} for the remodelled tissue. These curves resemble the restitution curves known for others atrial models, see Ref. \[[@pone.0190448.ref015]\]. With higher frequency (shorter BCL) the APD~*n*~ become smaller. This decrease is more pronounced for frequencies above 6 Hz. As a consequence, the optimization procedure becomes less robust for $f \gtrsim 6$ Hz, a feature that is discussed in more detail below in Section "Robustness with respect to the activation frequency".
{ref-type="fig"} and [4](#pone.0190448.g004){ref-type="fig"}, and [Table 1](#pone.0190448.t001){ref-type="table"}). The AP were taken after 10^4^ ms, beyond the time needed for achieving the stationary state.](pone.0190448.g005){#pone.0190448.g005}
Modeling of patient-specific action potentials with the BOCF model {#sec004}
==================================================================
Let us denote by $\mathcal{V}$ the APA and by $\mathcal{D}_{n}$ the values of the APD~*n*~ of a specific patient. To model the corresponding AP with the BOCF model, we determine *τ*~fi~ by inserting $\text{APA} = \mathcal{V}$ in [Eq (4)](#pone.0190448.e004){ref-type="disp-formula"} and (*τ*~si~, *τ*~so1~) by minimizing the sum of the squared deviations between the the APD~*n*~, i. e. the function $$\begin{array}{r}
{\mathcal{F}\left( \tau_{\text{si}},\tau_{\text{so}1} \right) = \sum\limits_{n}\left\lbrack \text{APD}_{n}\left( \tau_{\text{si}},\tau_{\text{so}1} \right) - \mathcal{D}_{n} \right\rbrack^{2}\,.} \\
\end{array}$$ For the numerical procedure we used the Levenberg-Marquardt algorithm \[[@pone.0190448.ref016]\]. As one sees from [Fig 4(b)--4(e)](#pone.0190448.g004){ref-type="fig"}, the APD vary monotonically with the time scales in the ranges fixed above. We checked that the Hessian is positive definite in the corresponding region, implying unique solutions when minimizing $\mathcal{F}$.
To demonstrate the adaptation procedure, we generated surrogate AP with the CRN model \[[@pone.0190448.ref001]\] for electrically remodeled tissue due to atrial fibrillation \[[@pone.0190448.ref014]\]. Specifically, we consider the maximal conductances, *g*~Ca~ and *g*~Na~ of the calcium and sodium currents to vary, while keeping all other parameters fixed to the values corresponding to the electrically remodeled tissue. The conductance *g*~Ca~ affects both the AP plateau and the repolarization phase and the *g*~Na~ controls mainly the amplitude of the AP \[[@pone.0190448.ref001]\].
[Fig 6](#pone.0190448.g006){ref-type="fig"} shows five examples of AP generated with the CRN model, which cover a wide range of APA and APD. In [Fig 6](#pone.0190448.g006){ref-type="fig"} we allow *g*~Na~ and *g*~Ca~ to differ by factors between 40% and 300% from their values *γ*~Na~ = 7.8 nS/pF and *γ*~Ca~ = 0.0433 nS/pF for the electrically remodelled tissue \[[@pone.0190448.ref014]\]. The corresponding AP modeled with the BOCF model, i. e. for *τ*~fi~ from [Eq (4)](#pone.0190448.e004){ref-type="disp-formula"}, and *τ*~si~ and *τ*~so1~ obtained from the minimization of $\mathcal{F}\left( \tau_{\text{si}},\tau_{\text{so}1} \right)$ in [Eq (6)](#pone.0190448.e032){ref-type="disp-formula"}, are shown as dashed lines in the figures. In all cases these reproduce well the AP shapes generated with the CRN model.
{#pone.0190448.g006}
To quantify the difference between the AP, we denote by $\mathcal{A}_{\text{CRN}}\left( t \right)$ and $\mathcal{A}_{\text{BOCF}}\left( t \right)$ their time course, and compute their relative deviation based on the *L*~2~-norm, $$\begin{array}{r}
{\Delta\mathcal{A} = \frac{{||}\mathcal{A}_{\text{BOCF}}\left( t \right) - \mathcal{A}_{\text{CRN}}\left( t \right){||}_{L_{2}}}{{||}\mathcal{A}_{\text{CRN}}\left( t \right){||}_{L_{2}}}\,,} \\
\end{array}$$ where $$\begin{array}{r}
{\left| \middle| \mathcal{A}\left( t \right) \middle| \right|_{L_{2}} \equiv \left( \int_{t_{i}}^{t_{f}}\mathcal{A}^{2}\left( t \right)dt \right)^{1/2}\,.} \\
\end{array}$$ The initial time *t*~*i*~ and final time *t*~*f*~ are defined as the times for which *u*(*t*~*i*~) = *u*(*t*~*f*~) = *θ*~0~ with *θ*~0~ = 0.015473 (see [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"}), with opposite signs of the corresponding time derivatives, i.e. $\frac{du}{dt}|_{t_{i}} > 0$ and $\frac{du}{dt}|_{t_{f}} < 0$.
[Fig 7(a)](#pone.0190448.g007){ref-type="fig"} shows that, when keeping *g*~Na~ = *γ*~Na~ fixed, $\Delta\mathcal{A}$ is below 5% for values of *g*~Ca~ between 10-400% of the reference value *γ*~Ca~. For $g_{\text{Ca}}/\gamma_{\text{Ca}} \gtrsim 4$, $\Delta\mathcal{A}$ starts to increase. Likewise, as show in [Fig 7(b)](#pone.0190448.g007){ref-type="fig"}, $\Delta\mathcal{A}$ does not exceed 9% when varying *g*~Na~ between 10--400% of *γ*~Na~ for fixed *g*~Ca~ = *γ*~Ca~.
{#pone.0190448.g007}
Additionally to the relative deviation between AP, one can compute the relative deviations between the APA and APD~*n*~ retrieved from the BOCF fit, $$\begin{array}{r}
{\Delta = \frac{|X_{\text{BOCF}} - X_{\text{CRN}}|}{X_{\text{CRN}}}\,.} \\
\end{array}$$ Here *X* represents either $\mathcal{V}$, giving Δ~APA~ or $\mathcal{D}_{n}$, giving Δ~APD~*n*~~.
[Fig 7(c) and 7(d)](#pone.0190448.g007){ref-type="fig"} show Δ*X*~APA~ as a function of *g*~Ca~/*γ*~Ca~ and *g*~Na~/*γ*~Na~, again for fixed *g*~Na~ = *γ*~Na~ and *g*~Ca~ = *γ*~Ca~, respectively. Corresponding plots of the Δ*X*~APD~*n*~~ are shown in [Fig 7(e) and 7(f)](#pone.0190448.g007){ref-type="fig"}. [Fig 7(c)](#pone.0190448.g007){ref-type="fig"} shows that Δ*X*~APA~ is always very small, even for large deviations of *g*~Ca~ from the reference value *γ*~Ca~. By contrast, Δ*X*~APA~ is quite sensitive to variations of *g*~Na~. The deviation becomes larger than 5% for $g_{\text{Na}}/\gamma_{\text{Na}} \gtrsim 2$.
As for the Δ*X*~APD~*n*~~ they are typically below 12% except in the case of APD~20~. The APD~20~ refers to the TMV level closest to the maximum and exhibits larger deviations up to about 20% for even small shape deviations.
All in all, [Fig 7](#pone.0190448.g007){ref-type="fig"} shows that the optimization procedure retrieves acceptable fits of single-cell AP in a wide range of calcium and sodium conductances.
Robustness with respect to the activation frequency {#sec005}
===================================================
The optimization procedure described in this paper was illustrated using one single activation frequency, namely *f* = 3 Hz. An important issue is the robustness of the optimization framework for other activation frequencies, which we address in Figs [8](#pone.0190448.g008){ref-type="fig"} and [9](#pone.0190448.g009){ref-type="fig"}.
{ref-type="disp-formula"} on the activation frequency *f*; (d) the *R*^2^ values of the corresponding fits.](pone.0190448.g008){#pone.0190448.g008}
{ref-type="disp-formula"} on the activation frequency *f*, where the symbol assignment refers to the different APD~*n*~ as given in the inset of the last graph, which shows the *R*^2^ values of the corresponding fits \[[Eq (5)](#pone.0190448.e026){ref-type="disp-formula"}\].](pone.0190448.g009){#pone.0190448.g009}
[Fig 8(a)--8(c)](#pone.0190448.g008){ref-type="fig"} show the variation of the three coefficients in [Eq (4)](#pone.0190448.e004){ref-type="disp-formula"} to fit the functional dependence of the APA on the parameter *τ*~fi~. As one sees, all three coefficients are approximately constant for activiations below 7 Hz. In that range of values one also observes a coefficient of determination $R^{2} \gtrsim 0.999$, as shown in [Fig 8(d)](#pone.0190448.g008){ref-type="fig"}. For higher frequencies *f*, the coefficients start to vary and the *R*^2^ values of the fits become smaller, indicating the need of higher order polynomials to describe the relation between *τ*~fi~ and APA.
Similar results are obtained for the coefficients used to fit the APD surfaces as functions of the parameters *τ*~si~ and *τ*~so1~. These are shown in [Fig 9](#pone.0190448.g009){ref-type="fig"} and demonstrate that the optmization procedure can be applied faithfully in the range 1 -- 6 Hz. Outside this range, polynomials of higher order would be needed for better matches.
All in all, this section provides evidence that our optimization procedure derived for an activation frequency of 3 Hz, may also be applicable for frequencies ranging at least between 1 and 6 Hz. [Fig 10](#pone.0190448.g010){ref-type="fig"} shows two illustrative examples of real AP and the respective fit with the optimization procedure. For details about the real data see Ref. \[[@pone.0190448.ref017]\].
{#pone.0190448.g010}
Conclusions {#sec006}
===========
In this work we showed how to model patient-specific action potentials by adjusting three characteristic time scales, which are associated with the net sodium, calcium and potassium ionic currents. The framework explores the possibilities of parameter adjustment of an atrial physiology model, namely the BOCF model \[[@pone.0190448.ref011]\], to reproduce AP shapes with a given amplitude, width and duration. The BOCF model is defined through a reaction-diffusion equation, coupled to three equations for gating variables that describe the opening and closing of ionic channels. It is simple enough to guarantee low computational costs for even extensive simulations of spatio-temporal dynamics \[[@pone.0190448.ref018]\]. Through a semi-analytical approach given in the [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"} we showed why the three ionic currents suffice to derive the main features of empirical AP.
The high flexibility for case-specific applications can be used for clinical purposes. By adjusting a simulation to specific patient conditions one may also analyze numerically the effect of drug therapy under specific conditions. Using the optimization procedure for AP shape adjustment, the three characteristic times are retrieved, which are directly connected to the ion-type specific net currents. AP shapes showing pathological features will be reflected in the values of one (or more) times outside acceptable ranges. Accordingly, one can associate a corresponding net current and therefore identify the class of membrane currents, where pathologies should be present. In this sense the clinical diagnosis can be supported by the modeling. A future application could be to take the retrieved parameters values as a basis for spatially extended simulation by including the diffusion term in [Eq (1)](#pone.0190448.e001){ref-type="disp-formula"} \[[@pone.0190448.ref011]\]. For this, one would need access to conduction properties which then would enable one to model spatio-temporal AP evoluion.
Though our framework is applicable in a quite wide range of values of sodium and calcium conductances, for conductances beyond a few times the reference values for electrically remodelled tissue the matching of AP shapes becomes less accurate. As for changes of the activation frequency *f*, the analysis in the Supporting Information limits the applicability of the AP modeling based on Eqs ([4](#pone.0190448.e004){ref-type="disp-formula"}) and ([5](#pone.0190448.e026){ref-type="disp-formula"}) to the range *f* = 1 -- 6 Hz.
Furthermore, in case information is obtained about AP shapes from different places of the atria, e. g. by using a lasso catheter, a corresponding AP shape modeling would allow one to construct a patient-specific model with spatial heterogeneities. Based on this, it could become possible to generate spatio-temporal activation pattern and to identify possible pathologies associated in the dynamics of the action potential propagation.
Supporting information {#sec007}
======================
###### Time evolution of one AP together with each ionic current.
\(a\) AP variable *u* with the stimulus current *J*~stim~, with (b) a close-up for a time interval of 3.5 ms. Vertical dashed lines intersect the AP at one specific dotted line, thus bounding the time intervals corresponding to each region of *u*-values (see text). The ionic currents correspond to (c) the Na channel (*J*~fi~), (d) the Ca channel (*J*~si~), and (e) the K channel (*J*~so~), see [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"}. All currents are given in (ms)^−1^.
(EPS)
######
Click here for additional data file.
###### Ionic currents as function of the respective gating variables.
\(a\) *J*~fi~ and (b) *J*~si~(*u*, *w*). The red circles indicate the path corresponding to Eqs ([1](#pone.0190448.e001){ref-type="disp-formula"}) and ([3](#pone.0190448.e003){ref-type="disp-formula"}) and sketched in [S1 Fig](#pone.0190448.s001){ref-type="supplementary-material"} as a function of time. (c) Dependence of ionic current *J*~so~ on variable *u*.
(EPS)
######
Click here for additional data file.
###### Time evolution of the four variables of the BOCF model.
**(a)** AP variable *u* and the three gating variables **(b)** *v*, **(c)** *w* and **(d)** *s*. The horizontal dotted lines in (a) indicate the ranges of *u*-values, where the evolution of the set of variables changes discontinuously. Vertical dashed lines intersect the AP at one specific dotted line, thus bounding the time intervals corresponding to each region of *u*-values. In several of such time intervals, some of the variables decay exponentially and independently from the other variables, which simplifies the model considerably. In the regions where no exponential evolution is indicated the model follows the reduced system of equations derived in [S1 Appendix](#pone.0190448.s004){ref-type="supplementary-material"}.
(EPS)
######
Click here for additional data file.
###### Dynamical features of the BOCF model: A semi-analytical approach.
(TEX)
######
Click here for additional data file.
The authors thank C. Lenk and G. Seemann for helpful discussions and the Deutsche Forschungsgemeinschaft for financial support (Grant no. MA1636/8-1). They also thank U. Ravens and G. Seemann for providing AP data.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
All Excel data files are available from the Dryad database (doi:[10.5061/dryad.h66cn](https://doi.org/10.5061/dryad.h66cn)).
Introduction {#sec001}
============
Chronic wasting disease (CWD) is a fatal transmissible spongiform encephalopathy affecting white-tailed deer (*Odocoileus virginianus*), mule deer (*Odocoileus hemionus*), Rocky Mountain elk (*Cervus elaphus nelsoni*), and moose (*Alces alces shirasi)* in North America \[[@pone.0186512.ref001]--[@pone.0186512.ref005]\]. All transmissible spongiform encephalopathies are caused by unconventional infectious agents composed of the proteinase-resistant pathologic isoform (PrP^res^) of the normal cellular prion protein (PrP^C^) \[[@pone.0186512.ref006]--[@pone.0186512.ref008]\]. Chronic wasting disease naturally occurs in free-ranging cervid populations in 21 U.S. states and two Canadian provinces \[[@pone.0186512.ref009]\], but limited information exists regarding the population-level impacts of CWD in the wild. In captivity, annual CWD incidence may exceed 50% in mule deer and white-tailed deer \[[@pone.0186512.ref010]\] and epidemics often end in the depopulation of deer at research facilities. Declines in free-ranging mule deer in Table Mesa, Colorado, elk in Rocky Mountain National Park in Colorado, and white-tailed deer in southeastern Wyoming were attributed to CWD prevalence greater than 13% \[[@pone.0186512.ref011]--[@pone.0186512.ref013]\]. From 2001--2009, the Wyoming Game and Fish Department (WGFD) recorded an average CWD prevalence of 31% from hunter harvested mule deer in southern Converse County, Wyoming \[[@pone.0186512.ref014]\]. Concurrently, WGFD estimated a \>50% reduction in the South Converse Mule Deer Herd \[SCMDH; 14\]. High annual CWD prevalence and declining population trends in this mule deer herd warranted investigation of the influence of CWD-associated declines in vital rates (i.e. survival, pregnancy, and fawn recruitment) and population growth rate (*λ*).
We hypothesized that CWD negatively impacted adult survival, pregnancy, and recruitment of fawns. The effect of CWD on population growth was measured by estimating *λ* for the CWD test-postive and test-negative portions of the population. Prior research revealed mule deer possessing at least one phenylalanine (F) at codon 225 of the prion protein gene (*Prnp*) were less susceptible to CWD infection compared to homozygous serine (S) genotyped deer \[[@pone.0186512.ref015]\]. Therefore, we evaluated the influence of *Prnp* on CWD incidence and compared *λ* estimates of the phenylalanine genotype (225SF and 225FF deer grouped and hereafter referred to as 225\*F deer) and homozygous serine genotype (225SS) segments of the population \[[@pone.0186512.ref015]\]. Other studies suggested CWD-positive deer are more likely to be killed by mountain lions (*Puma concolor*) \[[@pone.0186512.ref016]\] and that mountain lions may selectively prey on prion-infected deer \[[@pone.0186512.ref017]\]. Mule deer also may be more vulnerable to vehicle collisions, especially during the later stages of infection \[[@pone.0186512.ref018]\]. Sympatric CWD-positive white-tailed deer were more likely to be harvested by hunters \[[@pone.0186512.ref013]\]. Thus, we evaluated if CWD-positive deer were more susceptible to specific causes of mortality.
Material and methods {#sec002}
====================
Study area and population {#sec003}
-------------------------
We studied mule deer from the SCMDH that wintered primarily within the LaPrele Valley that surrounds the LaPrele Reservoir in southern Converse County, Wyoming from 2010--2014. The aggregate home range of all marked mule deer occupied an area \~2,576 km^2^. Deer wintered at elevations of \~1,500 m and a portion of the population migrated to summer ranges at \~2,700 m. Our study area was predominantly comprised of private native rangelands with some cultivated meadows along with some small tracts of public land. Some mule deer seasonally migrated to higher elevations where larger tracts of national forest occurs. True mountain mahogany (*Cercocarpus montanus*), antelope bitterbrush (*Purshia tridentata*), and big sagebrush (*Artemisia tridentata*) dominated the foothills while sagebrush and irrigated hayfields dominated the lowland areas. In 2010, the WGFD estimated the SCMDH at \~6,100 deer and by the conclusion of the study in 2014, the herd was estimated at \~5,100 deer based on post-harvest population estimates and different modeling techniques for 2010 (POP-II, Fossil Creek Software) and 2011--2014 (spreadsheet model) \[[@pone.0186512.ref019]\]. The hunting of does and fawns was largely eliminated in 2009 in response to poor population performance \[[@pone.0186512.ref014],[@pone.0186512.ref019]\]. Throughout the course of this study, a seven-day general antlered mule deer season occurred in this population with approximately 300 males harvested each year within the herd unit \[[@pone.0186512.ref019]\]. Annual CWD prevalence of sympatric male and female white-tailed deer and elk harvested during the study averaged 13.32% (n = 42) and 5.92% (n = 529), respectively.
Captures and field data collection {#sec004}
----------------------------------
We aerial net-gunned adult mule deer during winter (February/March) \[[@pone.0186512.ref020]\], focusing primarily on females and capturing at least 40 females each year. Males were captured from 2011--2014 to evaluate sex-associated CWD prevalence and survival. Captured deer were chemically immobilized with an intramuscular (IM) injection of either 0.03 mg/kg of carfentanil and 0.7 mg/kg of xylazine or 0.5 mg/kg of butorphanol, 0.35 mg/kg of azaperone, and 0.22 mg/kg of medetomidine (BAM; \[[@pone.0186512.ref021],[@pone.0186512.ref022]\]). We collected blood by jugular venipuncture and used it for *Prnp* determination using restriction fragment length polymorphism with confirmation by sequencing of PCR fragments of random samples \[[@pone.0186512.ref015]\]. Serum separated from blood samples was used for pregnancy analysis by pregnancy-specific protein B (PSPB) concentration (BioTracking LLC, Moscow, Idaho, USA). Approximate age at capture was determined using tooth eruption and wear \[[@pone.0186512.ref023]\]. Incisors from recovered carcasses were aged using cementum annuli analysis \[[@pone.0186512.ref024]\]. We assessed body condition by assigning a subjective score based on palpation of fat and muscle. Tonsil biopsies were performed to test deer for CWD by immunohistochemistry (IHC) and surgical equipment was cleaned using methods previously published to prevent iatrogenic transmission \[[@pone.0186512.ref025]\]. Deer were administered subcutaneous procaine/benzathine penicillin G combination (25,000 units/kg based on benzathine fraction, Bimeda, Le Sueur, Minnesota, USA) and 1.5 mg/kg of Banamine IM (Intervet Inc., Merck Animal Health, Summit, New Jersey, USA). Deer were given naltrexone (100 mg/mg of carfentanil) and tolazoline (2mg/kg) or naltrexone (50--100 mg), tolazoline (200--300 mg), and atipamezole (15--25 mg) to reverse anesthetic effects of carfentanil/xylazine or BAM, respectively \[[@pone.0186512.ref021],[@pone.0186512.ref022]\]. Deer were fitted with either a store-onboard global positioning system (GPS) radio-collar (Lotek Wireless Inc., Newmarket, Ontario, Canada) or a very-high-frequency (VHF) radio-collar (Advanced Telemetry Systems, Inc., Isanti, Minnesota, USA) equipped with mortality signal that was activated after 4 hours of inactivity. Deer were tagged with a large cattle ear-tag with an identification number and contact information if harvested for postmortem CWD testing and a metal WGFD identification ear-tag. Surviving deer were recaptured annually and processed as described with the exception of known CWD-positive deer, which no longer required biopsy. The study was completed in 2014 with the removal of radio-collars and final release of all surviving deer. All procedures involving deer were performed under the approval of the University of Wyoming Institutional Animal Care and Use Committee (No. A-3216-01) and the Wyoming Game and Fish Department (Permit No. 33--751).
Radio-collared mule deer were monitored at least twice weekly and mortalities were recovered to determine cause of death. Mortalities were investigated immediately after detection to recover carcasses prior to scavenging and autolysis. Necropsies were performed either in the field or at the Wyoming State Veterinary Laboratory. Postmortem CWD tests were performed when feasible. Retropharyngeal lymph nodes (RLNs) were collected and one was tested for PrP^res^ using the enzyme-linked immunosorbent assay (ELISA; \[[@pone.0186512.ref026]\]). Tonsil, obex region of the medulla oblongata, and the other RLN were tested for PrP^res^ by IHC \[[@pone.0186512.ref027]\]. We determined cause of death as clinical CWD if a deer tested positive for CWD postmortem and presented with no other signs of disease and trauma. While CWD is a fatal disease, not all CWD-positive deer died due to clinical disease. The proximate cause of death was recorded for each mortality case regardless of CWD status at the time of necropsy.
Fawn recruitment (number of fawns per marked doe) was documented during November ground surveys from 2011--2013. Females that were pregnant during captures were located via telemetry. If fawns were not seen with marked females during the initial observation, females were displaced to observe fawns fleeing the area. If no fawns were observed during the first attempt, subsequent attempts were made until the end of November.
Kaplan-Meier survival and incidence analysis {#sec005}
--------------------------------------------
Annual survival and incidence were estimated using Kaplan-Meier known-fates estimation \[[@pone.0186512.ref028],[@pone.0186512.ref029]\]. Survival was estimated from previous capture event (*t*-1) to current capture event (*t*) and all deer entered the study analysis at *t* = 0 regardless of initial capture year. Survival and incidence were determined based on biological year (June 1^st^--May 31^st^), which formed the basis of stage-structured Lefkovitch matrix models \[[@pone.0186512.ref030]\]. Daily survival and weekly CWD incidence were calculated using the *survival* package v.2.37--7 \[[@pone.0186512.ref031]\] and *survfit* function in statistical program R v.3.1.0 \[[@pone.0186512.ref032]\]. Mortality dates were determined as the first mortality event recorded by the GPS radio-collar following initial capture. Mortality date for deer tagged with VHF radio-collars was determined either by the condition of the carcass or as the midpoint date between a live observation and a dead observation. Deer were right censored if they were lost to follow-up due to relocation failure, died from unnatural causes such as poaching or capture-related mortality, or survived to the end of the study. Several deer started the study as CWD-negative and subsequently tested positive, thus their survival time was split into two datasets, in which they were right censored as a CWD-negative deer at their CWD-positive test date. Survival estimates of CWD-positive deer used in our analyses were potentially biased low because deer were considered CWD-positive the day they tested positive and we included deer that were initially captured as test-positive animals. Post hoc survival analysis revealed that our survival estimate that included all CWD-positive deer fell within the 95% CI (0.28, 0.69) of 26 deer that experienced a CWD incident event during the study (we left censored 17 deer without known fates because they were test-positive post-mortem or test-positive during the last capture when monitoring ceased). Survival was determined separately based on sex, CWD status, age-class, and *Prnp* genotype. Deer were left censored when calculating incidence if they were initially CWD-positive. An incident event occurred when a CWD-negative deer first tested CWD-positive and right censored if lost to follow-up, CWD status was not determined on subsequent captures, or study ended with a final CWD-negative test. Incidence was calculated separately based on sex, age-class, and *Prnp* genotype. Log-rank tests \[[@pone.0186512.ref029],[@pone.0186512.ref033]\] were performed in R using the function *survdiff* to compare all Kaplan-Meier curves \[[@pone.0186512.ref031]\].
Extended Cox proportional hazards model analysis {#sec006}
------------------------------------------------
We examined the effects of sex, age at capture, CWD status, and *Prnp* genotype on weekly survival probability. An extended Cox proportional hazards model was used to determine which variables had the most influence on annual survival of deer \[[@pone.0186512.ref029],[@pone.0186512.ref034]\]. The analyses were performed using the *coxph* function in R \[[@pone.0186512.ref031]\]. Time-dependent variables were created for CWD status and age as both changed through time for individual deer during the study \[[@pone.0186512.ref029]\]. Proportional hazards assumption was tested using the *cox*.*zph* function, which evaluates correlation between the Schoenfeld residuals and survival time \[[@pone.0186512.ref031]\]. Covariates failed proportionality when their p-value ≤ 0.05 \[[@pone.0186512.ref029]\]. Stepwise forward and backward selection of models were performed using the function *stepAIC* in the package *MASS* v.7.3--31 \[[@pone.0186512.ref035]\]. Models were ranked based on Akaike's Information Criteria (AIC) values \[[@pone.0186512.ref036]\]. Model AIC values within 2 AICs of the best model (Δ AIC) were considered good predictors of survival and individual covariate p-values were evaluated for final model selection \[[@pone.0186512.ref036]\].
Pregnancy and recruitment mixed model analysis {#sec007}
----------------------------------------------
We used generalized linear mixed models to determine the effects of age, CWD status, winter body condition, *Prnp* genotype, and observation year on annual proportion pregnant deer and fawn recruitment. A repeated measures analysis was performed and data grouped by unique deer identification was modeled using the *glmer* function in program *lme4* v.1.1--7 \[[@pone.0186512.ref037]\]. Pregnancy and recruitment indices were calculated separately based on CWD status and observation year.
Population growth rate estimation {#sec008}
---------------------------------
An age- and CWD-structured, female-only Lefkovitch matrix model was used to estimate *λ* in MATLAB^®^ (The MathWorks, Inc., Natick, MA, USA; \[[@pone.0186512.ref038]\]. We used a pre-breeding census, in which deer were counted prior to the birth-pulse in June, thus the first age-class in our model was yearling. Our matrix ([Fig 1](#pone.0186512.g001){ref-type="fig"}) represented both CWD-negative and CWD-positive deer of age *x*~*i*~, where a deer could survive to age *x*~*i*~ + 1 at a probability of ${\hat{\theta}}_{i -}\left( 1 - {\hat{\rho}}_{i} \right)$, where ${\hat{\theta}}_{i -}$ was the probability of a CWD-negative deer surviving and $\left( 1 - {\hat{\rho}}_{i} \right)$ was the transition probability of remaining CWD-negative. Deer that were CWD-negative survived and became CWD-positive at a probability of ${\hat{\theta}}_{i -}^{\frac{1}{2}}{\hat{\theta}}_{i +}^{\frac{1}{2}}\left( {\hat{\rho}}_{i} \right)$, which represents continuous disease transmission from time t to t+1, and CWD-positive deer survived at a probability of ${\hat{\theta}}_{i +}$.
{#pone.0186512.g001}
The vital rates included in the matrix model were pregnancy ($\hat{\beta}$), fawn recruitment ($\hat{\delta}$), overwinter fawn survival (${\hat{\theta}}_{0}$), adult survival (≥ 1 year old) of CWD-negative deer (${\hat{\theta}}_{-}$), adult survival of CWD-positive deer (${\hat{\theta}}_{+}$), and CWD incidence ($\hat{\rho}$). Fawns were not captured in our study; therefore, overwinter fawn survival from mid-December to mid-June was estimated from comparable areas in Colorado, from 1997 to 2008 \[[@pone.0186512.ref039]\]. Due to small sample size and non-significant differences in survival and fecundity among ≥ two-year-olds, we built our matrix model to include two age-classes, yearlings and adults. We calculated the 95% confidence interval for *λ* using previously described methods of parametric bootstrapping \[[@pone.0186512.ref040],[@pone.0186512.ref041]\], modified using our vital rates and matrix configuration. Overwinter fawn survival was bootstrapped using the standard deviation published for the point estimate \[[@pone.0186512.ref039]\]. Sensitivity and elasticity analyses were performed to evaluate how sensitive *λ* was to changes in individual vital rates using the method of Morris and Doak \[[@pone.0186512.ref041]\].
We initially ignored the influence of *Prnp*-genotype on CWD incidence and used the overall female-only CWD incidence during the biological year in our matrix. However, to understand the effect of CWD incidence on *λ*~,~ we varied annual incidence in the matrix from 0% to 100% and calculated the change in *λ*. Additionally, we calculated *λ* using genotype-specific incidence rates to examine estimated growth rates for the 225SS and 225\*F segments of the population. The approach used to model *λ* for all scenarios assumes constant vital rates, thus density dependence was not represented in model results.
Results {#sec009}
=======
Annual CWD prevalence and incidence {#sec010}
-----------------------------------
During the study, 143 mule deer were captured (118 female, 25 male) and *Prnp* genotypic frequencies were 78% 225SS deer and 22% 225\*F deer. Average annual CWD prevalence was 24% (95% CI = 22%--27%). Male CWD prevalence was higher throughout the study (average = 43%) compared to female CWD prevalence (average = 18%). Seventy-seven deer tested positive for CWD during the study, of which 43 were deer that transitioned from test negative to positive. Annual CWD incidence did not differ among observation years (*χ*^*2*^ = 3.2, df = 3, p = 0.36) and did not increase suggesting iatrogenic transmission likely did not occur. Also, annual CWD incidence was not different among age-classes (*χ*^*2*^ = 8.5, df = 7, p = 0.29) and between sex for years 2011 (*χ*^*2*^ = 0.3, df = 1, p = 0.56) and 2012 (*χ*^*2*^ = 2.7, df = 1, p = 0.10). In 2013, annual CWD incidence was significantly higher in males than females (*χ*^*2*^ = 6.1, df = 1, p = 0.01). Annual CWD incidence differed among genotypes (*χ*^*2*^ = 34.5, df = 2, p \< 0.01), with 225SS deer more likely to become CWD-positive compared to 225\*F deer. Average annual female CWD incidence was 0.26 (SE = 0.04) and genotype-specific incidence used in our matrix models were 0.49 (SE = 0.05) for 225SS deer and 0.02 (SE = 0.06) for 225\*F deer.
Cause-specific mortality {#sec011}
------------------------
We documented 97 mortalities of radio-collared deer. Mule deer that were CWD-positive were more susceptible to mountain lion predation (n = 20; *χ*^*2*^ = 6.36, df = 1, p = 0.01), hunter harvest (n = 4; *χ*^*2*^ = 7.98, df = 1, p \< 0.01), and illegal harvest (n = 2; *χ*^*2*^ = 3.99, df = 1, p = 0.05). Mountain lion predation was the number one cause of mortality followed by clinical CWD (n = 14). Other natural causes of mortality of radio-collared mule deer included vehicle collision (n = 3), coyote predation (n = 1), fence entanglement (n = 1), drowning (n = 1), and winter kill (n = 1). Thirteen deer died due to injuries sustained during captures and we were unable to determine the cause of death in 37 cases due to severe autolysis and scavenging.
Extended Cox proportional hazards models {#sec012}
----------------------------------------
Stepwise selection based on AIC values of our extended Cox proportional hazards models resulted in four competing models. The model that incorporated sex and CWD was selected as the best model for predicting survival based on individual covariate p-values ([Table 1](#pone.0186512.t001){ref-type="table"}). Male mule deer were twice as likely (Hazard Ratio = 2.08, p = 0.01, 95% Confidence Interval (CI) = 1.18--3.68) to experience a mortality event compared to females and CWD-positive deer were over three times more likely (Hazard Ratio = 3.30, p \< 0.0001, 95% CI = 1.98--5.49) to die during our study compared to CWD-negative deer. Genotype was included in our models initially; however, models did not converge because they lacked full representation of the *Prnp* genotypes in both CWD categories (CWD-negative and CWD-positive). Only one 225SF deer tested CWD-positive out of 29 deer, but was censored after 154 weeks and neither of the two 225FF deer captured tested CWD-positive during the study. Therefore, we removed *Prnp* genotype from model analysis and we were not able to determine the influence of genotype on CWD-positive survival probability.
10.1371/journal.pone.0186512.t001
###### Extended Cox proportional hazards models with *a priori* variable selection of parameters that potentially influenced mule deer survival in southern Converse County, WY from 2010--2014.
{#pone.0186512.t001g}
Model Model parameters *K* AIC Δ AIC
------- -------------------------------------------------------------------------------------- ----- -------- -------
1 Sex[^a^](#t001fn004){ref-type="table-fn"}, CWD[^b^](#t001fn005){ref-type="table-fn"} 2 557.51 0
2 Sex, CWD, Age\*t 3 557.78 0.27
3 Sex, CWD, Age\*t, CWD\*Age\*t 4 557.98 0.47
4 Sex, CWD, Age\*t, Sex\*CWD 4 559.03 1.52
5 CWD, Age\*t 2 561.18 3.67
6 Sex, Age\*t 2 577.98 20.47
*K*, number of parameters; AIC, Akaike information criterion; ΔAIC, difference with best model AIC value; CWD, chronic wasting disease; t, time
Age\*t, time-dependent covariate of age
\*, interaction.
^a^ Hazard Ratio = 2.08, 95% Lower Confidence Interval (LCI) = 1.18, 95% Upper Confidence Interval (UCI) = 3.68, *P* = 0.01
^b^Hazard Ratio = 3.30, 95% LCI = 1.98, 95% UCI = 5.49, *P* \< 0.01
Annual survival estimates {#sec013}
-------------------------
Kaplan-Meier annual survival was significantly different between CWD-negative and CWD-positive deer (*χ*^*2*^ = 40.10, df = 2, p \< 0.01), CWD-negative and CWD-positive females (*χ*^*2*^ = 38.30, df = 2, p \< 0.01), and CWD-negative females and CWD-negative males (*χ*^*2*^ = 9.00, df = 2, p = 0.002; [Table 2](#pone.0186512.t002){ref-type="table"}). Estimated annual survival of CWD-negative deer (0.76, SE = 0.04) was considerably higher than CWD-positive deer (0.32, SE = 0.06; [Table 2](#pone.0186512.t002){ref-type="table"}). Female deer survived at a rate of 0.79 (SE = 0.04) annually compared to 0.50 (SE = 0.16) annual survival of male deer; however, this sex-associated difference was not observed for CWD-positive deer ([Table 2](#pone.0186512.t002){ref-type="table"}). When comparing female and male survival curves, both declined at similar rates until a dramatic decline in male survival around 250 days that corresponded to the short hunting season ([Fig 2](#pone.0186512.g002){ref-type="fig"}). This accelerated rate of decline in survival curves at about 250 days was prominent when comparing CWD-negative and CWD-positive males ([Fig 2](#pone.0186512.g002){ref-type="fig"}). A similar pattern of accelerated decline was observed between CWD-negative and CWD-positive females starting around day 275 ([Fig 2](#pone.0186512.g002){ref-type="fig"}).
{#pone.0186512.g002}
10.1371/journal.pone.0186512.t002
###### Kaplan-Meier survival rates and log-rank test results by sex, age, and chronic wasting disease (CWD) status of mule deer in southern Converse County, WY from 2010--2014.
{#pone.0186512.t002g}
--------------------------------------------------------------------------------------------------------
Category Results Overall 1.5 2.5 3.5 4.5 5.5+
------------------------------- ----------------------- --------- ------ ------- ------- ------- -------
**CWD (-) vs. (+) deer** **Survival: CWD (-)** 0.76 0.63 0.67 0.91 0.70 1.00
**Survival: CWD (+)** 0.32 0.00 0.15 0.60 0.28 0.51
**χ**^**2**^ 40.10 1.70 14.70 19.20 4.20 2.50
**P-value** 0.00 0.19 0.00 0.00 0.04 0.12
**CWD (-) vs. (+) females**\ **Survival: CWD (-)** 0.79 0.67 0.70 0.97 0.67 1.00
**Survival: CWD (+)** 0.37 0.00 0.20 0.61 0.34 0.44
**χ**^**2**^ 38.30 1.10 14.50 23.20 2.20 2.90
**P-value** 0.00 0.31 0.00 0.00 0.14 0.09
**CWD (-) vs. (+) males** **Survival: CWD (-)** 0.50 0.50 0.33 0.50 n = 1 n = 0
**Survival: CWD (+)** 0.19 0.00 0.00 0.50 0 n = 1
**χ**^**2**^ 1.10 0.20 0.00 0.10
**P-value** 0.29 0.70 0.83 0.78
**CWD (-) females vs. males** **Survival: females** 0.79 0.67 0.70 0.97 0.67 1.00
**Survival: males** 0.50 0.50 0.33 0.50 n = 1 n = 0
**χ**^**2**^ 9.00 0.20 1.60 2.40
**P-value** 0.00 0.62 0.21 0.12
**CWD (+) females vs. males** **Survival: females** 0.37 0.00 0.20 0.61 0.34 0.44
**Survival: males** 0.19 0.00 0.00 0.50 0.00 n = 1
**χ**^**2**^ 2.60 1.00 0.00 0.20 4.80
**P-value** 0.11 0.32 0.99 0.64 0.03
--------------------------------------------------------------------------------------------------------
Annual survival was not significantly different among age-classes for either CWD-negative (*χ*^*2*^ = 7.00, df = 5, p = 0.22) or CWD-positive deer (*χ*^*2*^ = 0.80, df = 4, p = 0.936). Therefore, when combining adult age-classes (≥ 2 years old) for our matrix model and survival from June 1^st^--May 31^st^ of CWD-negative females and CWD-positive females, survival was 0.85 (SE = 0.13) and 0.38 (SE = 0.34), respectively. Survival of CWD-negative females among genotypes was marginally significant (*χ*^*2*^ = 5.8, df = 2, p = 0.05) with higher survival of 225\*F deer compared to 225SS deer.
Annual pregnancy and recruitment estimates {#sec014}
------------------------------------------
Mean annual pregnancy of CWD-negative and CWD-positive females was 0.99 (SD = 0.11, 95% CI = 0.97--1.00) and 0.94 (SD = 0.24, 95% CI = 0.88--1.00), respectively ([Table 3](#pone.0186512.t003){ref-type="table"}). Fawn recruitment from birth to November was similar between CWD-negative (average = 0.48, SD = 0.65, 95% CI = 0.33--0.63) and CWD-positive deer (average = 0.56, SD = 0.65, 95% CI = 0.30--0.82; [Table 4](#pone.0186512.t004){ref-type="table"}). Age, winter body condition, CWD status, *Prnp* genotype, and observation year did not influence pregnancy and recruitment of fawns (p \> 0.05).
10.1371/journal.pone.0186512.t003
###### Proportion of mule deer that were pregnant at approximately 75 days bred in southern Converse County, WY.
{#pone.0186512.t003g}
CWD-Negative CWD-Positive
------------- ------------------- -------------------
**2010** 1.00 (1.00, 1.00) 0.88 (0.64, 1.00)
**2011** 0.97 (0.91, 1.00) 0.91 (0.73, 1.00)
**2012** 1.00 (1.00, 1.00) 0.94 (0.83, 1.00)
**2013** 1.00 (1.00, 1.00) 0.95 (0.85, 1.00)
**2014** 0.95 (0.86, 1.00) 1.00 (1.00, 1.00)
**Average** 0.99 (0.97, 1.00) 0.94 (0.88, 1.00)
LCI, 95% lower confidence interval; UCI, 95% upper confidence interval.
10.1371/journal.pone.0186512.t004
###### Proportion of fawns at heel during November recruitment surveys of radio-collared female mule deer that were either CWD-test negative or positive during winter captures in southern Converse County, WY.
{#pone.0186512.t004g}
CWD-Negative CWD-Positive
------------- ------------------- -------------------
**2011** 0.48 (0.24, 0.72) 0.29 (0.00, 0.65)
**2012** 0.40 (0.17, 0.63) 0.56 (0.08, 1.03)
**2013** 0.56 (0.26, 0.86) 0.78 (0.34, 1.21)
**Average** 0.48 (0.33, 0.63) 0.56 (0.30, 0.82)
LCI, 95% lower confidence interval; UCI, 95% upper confidence interval.
Population models {#sec015}
-----------------
Our matrix model estimated *λ* = 0.79 (0.72, 0.87) that corresponded to a 21% annual decrease in the population with a population half-life of 4 years. The models that assumed 100% CWD prevalence and 0% CWD prevalence estimated *λ* = 0.51 and *λ* = 1.00, respectively ([Fig 3](#pone.0186512.g003){ref-type="fig"}). Using the estimated CWD incidence for 225SS deer, we estimated *λ* = 0.64 and for 225\*F, *λ* = 0.98. The matrix model was most sensitive to changes in survival of CWD-negative deer (${\hat{\theta}}_{-}$) and CWD incidence (${\hat{P}}_{i}$; [Table 5](#pone.0186512.t005){ref-type="table"}). However, when the sensitivities of vital rates were rescaled to account for proportional changes (elasticity), only changes in CWD-negative survival had largest effect on *λ* ([Table 5](#pone.0186512.t005){ref-type="table"}).
{#pone.0186512.g003}
10.1371/journal.pone.0186512.t005
###### Sensitivities and elasticities of vital rates included in our Lefkovitch matrix model representing a mule deer population in southern Converse County, WY from 2010--2014.
{#pone.0186512.t005g}
Vital rate Symbol Estimated value Sensitivity Elasticity
---------------------------------------- ---------------------- ----------------- ------------- ------------
**Pregnancy rate** $\hat{\beta}$ 0.97 0.1461 0.1753
**November recruitment** $\hat{\delta}$ 0.51 0.2779 0.1753
**Over-winter survival rate of fawns** ${\hat{\theta}}_{0}$ 0.72 0.1968 0.1753
**Survival rate of CWD negative deer** ${\hat{\theta}}_{-}$ 0.85 0.7126 0.7491
**Survival rate of CWD positive deer** ${\hat{\theta}}_{+}$ 0.38 0.1609 0.0756
**CWD incidence rate** ${\hat{\rho}}_{i}$ 0.26 -0.6939 -0.2231
Discussion {#sec016}
==========
Our findings support CWD as a population-limiting disease of mule deer with the potential to cause dramatic declines that resemble local population extinction. Other studies have found a negative association between CWD prevalence and *λ* \[[@pone.0186512.ref011],[@pone.0186512.ref012],[@pone.0186512.ref040],[@pone.0186512.ref042]\], but none have documented *λ* estimates resulting from endemic CWD as low as those reported here. The only scenario in which population growth rate was stable (*λ* = 1) was in the absence of CWD. Even without CWD mortality, we predicted that this population would not grow under current conditions. This finding was unremarkable considering mule deer populations throughout North America are underperforming in the absence of CWD \[[@pone.0186512.ref043]\]. Chronic wasting disease may exacerbate population declines in herds that are currently considered CWD-free. From 2010--2014, we predicted the southern Converse County herd would decline by \>50% using our estimated *λ* = 0.79 and a starting population size of \~6,100 deer. Earlier models of CWD epidemics in mule deer using prevalence observed in our study herd forecasted similar dramatic outcomes \[[@pone.0186512.ref044],[@pone.0186512.ref045]\]. This population has experienced population declines of approximately 50% based on WGFD population estimates prior to the start of our study \[[@pone.0186512.ref019]\]. However, this population did not appear to decline as dramatically during the study as our estimate of *λ* would suggest based on WGFD population estimates (approximately a 4% decline from 2010 to 2014) \[[@pone.0186512.ref019]\]. While the 2010 and 2014 population size estimates were not strikingly different, the general trend over time suggests a declining population. From 2011 to 2012, WGFD estimated a 19% decline in mule deer numbers and a 15% decline the following year \[[@pone.0186512.ref019]\]. These declines observed during our study fall within our 95% CI for *λ* (0.72, 0.87). In 2013, greater spring precipitation ended a year-long drought and moderate winter conditions resulted in a 5% increase of the population estimate in 2014 \[[@pone.0186512.ref019]\]. Therefore, while the population experienced productive years and deer numbers increased; these increases were marginal compared to the larger declines observed over multiple years.
We did not find disease-associated declines in reproduction for mule deer, nor have they been observed in sympatric white-tailed deer \[[@pone.0186512.ref040]\]. Females were pregnant regardless of CWD status during captures when they were approximately 75 days bred. Despite evidence that suggests CWD-positive mule deer recruit fewer fawns than CWD-negative deer \[[@pone.0186512.ref011]\], we did not detect a difference in fawn recruitment based on CWD status. Even with a reduction in fawn recruitment of CWD-positive mule deer in Colorado, inclusion of this vital rate in models did not significantly influence *λ* \[[@pone.0186512.ref011]\]. While CWD did not have a detectable impact on annual pregnancy and recruitment, lifetime reproduction of prime-aged females was likely reduced due to increased annual mortality of CWD-infected individuals. Prion-infected Table Mesa mule deer in Colorado survived an additional 1.6 years on average compared to 5.2 years for uninfected deer \[[@pone.0186512.ref016]\]. Furthermore, fawns produced by CWD-negative deer, which more likely possessed the more resistant genotype compared to CWD-positive deer in our study, potentially contributed to the increase of the F allele in the population.
Prion protein genotype was important in determining CWD infection and influenced *λ* for *Prnp*-specific segments of the population. As was expected, mule deer that possessed the 225SS *Prnp* genotype were more likely to be CWD-positive compared to 225SF and 225FF deer in our study. We only detected one 225SF CWD-positive deer even though 225\*F deer comprised 22% of the study population. Two radio-collared 225FF deer were captured in 2013 and survived to study termination in 2014 with negative tonsil biopsy IHC results. However, evidence suggests current IHC techniques may have lower sensitivity in detecting CWD-positive tissues of 225FF mule deer \[[@pone.0186512.ref046]\]. Both 225FF deer were estimated to be 3.5 years old during their initial capture, both were pregnant in 2013 and 2014, and during 2013 recruitment surveys, one had a single fawn at heel. The other 225FF deer was not observed during 2013 fawn recruitment surveys. During 2014 captures, ultrasound revealed that one 225FF deer was pregnant with twins and the other was pregnant with a single fetus. Based on a small sample size, free-ranging 225FF mule deer appeared to be as ecologically fit as 225SS deer. The few 225FF mule deer observed in captivity were characterized as atypical in behavior, body condition, and reproductive performance \[[@pone.0186512.ref046]\]. Formal investigations looking at the effects of *Prnp* genotype on fitness are necessary to determine how populations with greater numbers of 225FF mule deer will persist despite their reduced susceptibility to CWD.
Estimates of *λ* for 225SS and 225\*F segments of the population were mediated by varying CWD incidence rates. Using 225SS CWD incidence in our matrix model, we estimated an annual population decline of 33% of 225SS deer. A model incorporating 225\*F CWD incidence estimated an annual population decline of 1%. These results suggest the 225\*F segment of the population was nearly stable while the 225SS segment of the population was declining rapidly. Using previously published data of mule deer genotyped in the early 2000s from the same geographic area \[[@pone.0186512.ref015]\], we estimated a 10% population increase in the F allele frequency in less than 10 years \[[@pone.0186512.ref047]\]. Other factors were not identified that may potentially increase F allele frequency in the absence of CWD as it was outside of the scope of our study. Adaptation to CWD has previously been demonstrated in elk \[[@pone.0186512.ref048]\] and white-tailed deer \[[@pone.0186512.ref049]\] using empirical data and statistical models.
Natural selection in favor of less susceptible *Prnp* genotypes may be assisted with selective predation by mountain lions and harvest by hunters of prion-infected deer. While CWD-positive deer were more likely to be killed by mountain lions compared to uninfected deer, it is not clear if this source of mortality regulated or influenced the observed CWD epidemic. Selective predation of CWD-positive deer in Table Mesa, Colorado did not appear to control CWD transmission \[[@pone.0186512.ref016]\] and it also did not appear to curtail CWD prevalence in the current study herd. Theoretic modeling incorporating 15% predation rate and four times greater risk of predation of prion-infected deer resulted in the eradication of CWD in a closed population \[[@pone.0186512.ref050]\]. While we observed one year of 15% predation of marked deer in 2010, 3--4% predation rate was typical for most years of the study and it never exceeded 15%. While direct mortality could decrease the subset of infected animals in a population, predators may also act as mechanical vectors that spread prions across the landscape. Infectious prions were demonstrated to pass through the digestive system of coyotes (*Canis latrans*) three days post ingestion suggesting the potential role of carnivores in prion transport and spread \[[@pone.0186512.ref051]\]. At this time, empirical evidence that supports a predator influence on CWD epidemics does not exist. However, with the expected spread of CWD into areas such as the Greater Yellowstone Area that is occupied by several large predators (i.e. wolves (*Canis lupus*), grizzly bears (*Ursus arctos*), and mountain lions), the role of predators in prion transmission dynamics may soon become more relevant \[[@pone.0186512.ref050]\]. A multi-predator system may have a greater impact on an emerging CWD epidemic, especially before significant prion contamination occurs in the environment.
Hunting mortality was minimal in our study, although it appeared that sympatric CWD-positive mule deer and white-tailed deer were selectively harvested \[[@pone.0186512.ref013]\]. It is unclear why others have found no difference in hunting risk between infected and uninfected deer \[[@pone.0186512.ref052]\], but it is logical that CWD-positive individuals are more vulnerable to harvest due to behavioral changes associated with the disease. The precipitous decline in survival of CWD-positive males increased predictably during the short hunting season around day 250. However, unpredictably there was an observed accelerated decline in the survival curve of CWD-positive females after day 275. Multiple factors may have contributed to greater mortality of CWD-positive females on winter range including increased risk of predation and stressors associated with the rut, hunting season, recruitment of fawns, and winter conditions. Regardless of the cause, CWD-positive deer were more likely to die on winter ranges. This has important implications for the spread and translocation of CWD across the landscape. Congregating deer on winter range may act as a source for CWD-infection to disparate populations when deer migrate in the spring to different summer ranges. These temporal behaviors could explain some of the spatial heterogeneity of CWD prevalence across the landscape \[[@pone.0186512.ref053]\].
Without an effective CWD vaccine or treatment, management of this disease is limited to focusing on those individuals that are not yet prion-infected. According to our sensitivity analysis, changes in CWD-negative adult survival would cause the greatest changes in *λ*. Improving survival of uninfected mule deer may partially mitigate the impact of CWD. However, to achieve close to stable population growth rates required an unrealistic scenario of 100% survival of CWD-negative deer under high CWD prevalence conditions. We observed low fawn recruitment (0.51 fawns) during the study regardless of disease status compared to an adjacent herd (0.68 fawns) located north of SCMDH \[[@pone.0186512.ref054]\] and populations throughout the species range (\> 0.75 fawns) \[[@pone.0186512.ref011],[@pone.0186512.ref055]\]. Management strategies that focus on improving both adult survival of CWD-negative deer and fawn recruitment may increase *λ*. Mule deer populations that currently experience low adult and fawn survival should be closely monitored for CWD because our models predicted less than ideal outcomes once CWD was established.
Lastly, we predicted stable population growth only when CWD prevalence was reduced to 0% in our model. Eradication of CWD is an improbable goal in endemic areas, especially where CWD has been detected for over a decade and potentially present for over 50 years \[[@pone.0186512.ref045]\]. However, these findings highlight the importance of preventing or slowing the spread of CWD to naïve populations. Mule deer populations currently undergoing declines in the absence of CWD, such as in Nevada and South-central British Columbia \[[@pone.0186512.ref056],[@pone.0186512.ref057]\] and in western Wyoming, should be routinely surveyed for detection of CWD. Intensive surveillance that could detect the first few positive CWD cases and rapid removal of prion-infected individuals may be the difference between an established epidemic and local CWD eradication as apparently accomplished in New York and Minnesota \[[@pone.0186512.ref058],[@pone.0186512.ref059]\]. While most state agencies focus efforts on collecting hunter harvested and road-killed deer for CWD testing, we recommend incorporating predator-killed deer to the repertoire based on our finding of greater susceptibility of CWD-positive deer to predation \[[@pone.0186512.ref016],[@pone.0186512.ref017]\]. Many other non-disease-associated factors contribute to declining mule deer populations and CWD could be the fatal consequence for many herds. Due to the lack of effective management tools to eliminate CWD once established, we suggest management focus efforts and research on how to slow or potentially prevent the movement of CWD across the landscape into uninfected populations.
Conclusions {#sec017}
===========
With this study, we have demonstrated the long-term consequences of endemic CWD on a free-ranging mule deer population. Chronic wasting disease caused significant declines in the study mule deer herd as well as in sympatric white-tailed deer \[[@pone.0186512.ref013]\]. Unlike sympatric white-tailed deer, where removal of female harvest may permit *λ* to increase to stable levels based on model estimates \[[@pone.0186512.ref013]\], elimination of the mule deer doe/fawn hunting season prior to the onset of our study did not result in *λ* ≥ 1. A limited antlered-only harvest in this herd provides a reliable source for monitoring short-term CWD prevalence trends \[[@pone.0186512.ref053]\]. Additionally, improving and conserving critical mule deer habitats may diminish the negative impacts of CWD, but will not completely mitigate the undesirable population effect of CWD based on our model outcomes. Lastly, without the use of effective vaccines, treatments, and sustainable techniques to reduce CWD incidence, management can currently only focus on slowing the spread of CWD to CWD-free populations.
We thank the personnel of the University of Wyoming and the Wyoming Game and Fish Department for technical assistance in the field and laboratory, including G. Andrews, K. Andrews, K. Bardsley, J. Berg, A. Boerger-Fields, B. Bonner, G. Boyd, R. Burton, J. Cavender, W. Cook, A. Dadelahi, S. Dwinnell, S. Edberg, H. Edwards, K. Fambrough, R. Grogan, C. Hansen, C. Havlik, J. Henningsen, B. Howell, M. Huizenga, N. Hull, M. Jaeger, P. Jaeger, J. Jennings-Gaines, T. Jones, M. Kauffman, A. Kerr, A. Kesterson, H. Killion, K. Latta, R. Lebert, D. Lutz, D. Manore, M. Miller, K. Monteith, M. Nelson, H. O'Brien, B. Saeed, K. Speiser, W. Steen, M. Stuart, F. Sunshine Packard, S. Talbott, T. Tobiasson, M. Vasquez, A. Williams, B. Wise, and M. Wood. We are very grateful to many private landowners that granted us access to their properties for this research. This manuscript was greatly improved by the comments of P. Cross, three reviewers, and our Associate Editor, and figure editing by P. Whelan. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: Current address: Washington Department of Fish and Wildlife, Spokane Valley, Washington, United States of America
[^3]: ¶ Retired, Bovey, Minnesota, United States of America.
| {
"pile_set_name": "PubMed Central"
} |
Exposure to antibiotics in childhood is thought to alter the intestinal microbiome \[[@CIT0001]\]. Observational studies in high-income settings have suggested that there are alterations in the intestinal microbiome in children receiving antibiotics relative to those who have not received antibiotics \[[@CIT0003]\]. Antibiotics disrupt the composition of microbiota as they have activity against both commensal and pathogenic bacteria \[[@CIT0002]\]. Dysbiosis of the intestinal microbiome has been linked to multiple adverse health outcomes, including undernutrition and obesity, asthma, diabetes, and some forms of cancer \[[@CIT0006]\].
A recent randomized controlled trial evaluated the effect of a single dose of azithromycin, as is used for mass distribution in trachoma control programs \[[@CIT0015]\], on the composition of the intestinal microbiome among preschool children in Niger \[[@CIT0001]\]. This study found a significant decrease in the diversity of the intestinal microbiome in a population of relatively antibiotic-naïve children. In many regions of sub-Saharan Africa, macrolides are used less frequently than penicillins or sulfonamides for the treatment of common childhood illnesses \[[@CIT0018]\]. Here, we assess the effect of a short course of 3 commonly used pediatric antibiotics on the intestinal microbiome. We hypothesized that all 3 antibiotics would lead to decreased diversity in the intestinal microbiome compared with placebo.
METHODS {#s6}
=======
Study Setting {#s7}
-------------
This study took place in 2 rural communities of the Health and Demographic Surveillance Site (HDSS) in Nouna District, Burkina Faso \[[@CIT0021], [@CIT0022]\]. A triannual census is performed in the HDSS by the Centre de Recherche en Santé de Nouna (CSRN). Inhabitants of the study communities are primarily subsistence farmers and cattle keepers. The study occurred in July 2017, just before the rainy season \[[@CIT0023]\].
Participants and Procedures {#s8}
---------------------------
The overall objective of the study was to investigate the direct and indirect effects of antibiotic usage on the intestinal microbiome of preschool-aged children. We recruited households in the 2 study communities with 2 or 3 children between the ages of 6 and 59 months based on the most recent HDSS census. In households with 3 children, all children received a study drug but only 2 randomly selected children were monitored as part of the study. Children were eligible for participation in the study if they were between 6 and 59 months of age and with parental consent. We did not exclude children on the basis of preexisting morbidity or recent antibiotic use. Children were assessed before randomization (baseline) and then again 5 days after the last antibiotic dose (post-treatment).
Randomization {#s9}
-------------
We employed a 2-stage randomization procedure. First, each household was randomized in a 1:1:1:1 fashion to 1 of the 4 study arms: 1) amoxicillin, 2) azithromycin, 3) co-trimoxazole, or 4) placebo. Second, each child in the household was randomly assigned to either treatment or placebo. In households with 2 children, 1 was randomized to treatment and the other to placebo. In households with 3 children, 2 were randomized to treatment and the other to placebo. Note that in the placebo households, children assigned to treatment and placebo received the same drug (ie, placebo). The present report is concerned with the direct effects of antibiotics and therefore only includes the 1 child per household randomized to treatment. The randomization sequence was generated by TCP in R, version 3.3.1 (The R Foundation for Statistical Computing), using a masked seed value \[[@CIT0024]\]. The randomization sequence was implemented in the field by preloading syringes with the child's randomized treatment that were labeled with each child's name (described below).
Interventions {#s10}
-------------
All study medications were prepared as pediatric oral suspension. Study medications were procured at local pharmacies in the study area, with the exception of azithromycin, which was procured in Ouagadougou. Weight measurements were collected during the baseline visit as part of anthropometry assessments. These measurements were used to calculate the appropriate dose of study medication for each child. There was no standard dosing for study medication as children were not being treated for an established infection. We therefore used the lower end of approved dosing regimens for each antibiotic to reduce the risk of adverse events. All antibiotics were administered for 5 days. We used the lower end of approved dosing for amoxicillin for children under 12 years of age (25 mg/kg/d in twice-daily doses). Azithromycin dosing was based on the lower end of the range for standard pediatric dosing for mild to moderate infection (a single 10-mg/kg dose on day 1 and then 5 mg/kg once daily for 4 days). Cotrimoxazole dosing was based on prophylactic dosing for children living with HIV (240 mg once daily) \[[@CIT0025]\]. Placebo was prepared by study staff as a mixture of powdered milk, sugar, and bottled water. All study medications were prepared daily in orange opaque syringes and labeled uniquely for each child. Treatment was administered from a central point in each study community. A community mobilizer visited the homes of the children participating in the study and instructed the caregivers to bring the children to the central point for examination and treatment visits. Treatment teams recorded whether each child received the study treatment and reasons for a missed treatment.
Masking {#s11}
-------
Treatment teams were not told the identity of the study medication, and the orange tinting of the syringes helped to conceal the identity of the study drug. However, due to differences in taste and appearance, the treatment team was not masked. In contrast, examination teams were masked to treatment assignment, and laboratory personnel were masked to treatment assignment and time point.
Baseline Questionnaire {#s12}
----------------------
At baseline, the caregiver of each child completed a short questionnaire related to the child and the child's household of residence. Health status questions included if the child had recently visited a health facility and, if so, if the child had been treated with antibiotics after the visit. Caregivers were asked if the child was currently breastfeeding.
Outcome Assessment {#s13}
------------------
Rectal samples were collected in the field at baseline and 5 days after the last antibiotic treatment. Examiners inserted a swab 1--3 cm into the anus and rotated 360 degrees. Examiners changed gloves between each participant. Swabs were immediately placed in a Stool Nucleic Acid Collection and Transport Tube containing Norgen Stool Preservative (Norgen, Ontario, Canada). The transport media preserves DNA and RNA in the sample and prevents growth of organisms. Samples were placed at ambient temperature in the field, and then stored at the CRSN laboratory at --80°C until they were shipped to the United States. Samples were shipped on ice and then stored at --80°C until processing. Samples were de-identified in the field and then placed in a random order for library preparation and sequencing. DNA was extracted from the fecal samples using the Norgen stool DNA isolation kit (Norgen, Ontario, Canada), per the manufacturer's instructions. Concentration of DNA was quantified using the Qubit dsDNA HS Assay Kit (ThermoFisher Scientific, Waltham, MA) and adjusted to 15 ng/uL. The gut bacterial community was assessed by deep sequencing the V3-V4 hypervariable regions of the 16S rDNA gene. Library preparation was performed by SeqMatic (Fremont, CA) per the Illumina16S metagenomic sequencing library preparation protocol. Demultiplexed raw sequences were processed in QIMME, version 1.9, which utilizes the Ribosomal Database Project Classifier and the full GreenGenes 13_8 reference database to assign taxonomy to each sequencing read.
Sample Size Determination {#s14}
-------------------------
The sample size calculation was based on the primary outcome, Simpson's α-diversity. A sample size of 30 children per arm was estimated to provide at least 80% power to detect a 1.5-unit difference in Simpson's α-diversity based on a previous study in Niger \[[@CIT0001]\].
Statistical Methods {#s15}
-------------------
The primary outcome of the study was prespecified as α-diversity (inverse Simpson's) at the genus level, expressed in effective number. Shannon's α-diversity was calculated as a secondary outcome \[[@CIT0029]\]. The primary prespecified analysis assessed mean post-treatment diversities, compared across all arms with an analysis of variance and pairwise comparisons performed with a *t* test. As a sensitivity analysis, the post-treatment diversities were compared between arms in a linear regression model adjusted for baseline diversity. As an additional sensitivity analysis, post-treatment diversities were compared between arms with a term for both the child's age and the baseline diversity measurement in a linear regression model. We used permutational multivariate analysis of variance (PERMANOVA) to assess differences in microbial composition between arms using Manhattan and Euclidean distances. A principal coordinates analysis (PCoA) was used to visually depict the centroids of the groups. All *P* values were calculated using a Monte Carlo permutation test with 10 000 replications, and *P* values \<.05 were considered statistically significant for all tests. All analyses were conducted in R, version 3.4.3 (The R Foundation for Statistical Computing). Diversity measures and Manhattan and Euclidean distances were calculated in the R package "vegan."
Ethical Considerations {#s16}
----------------------
The study was reviewed and approved by the Comité Institutionnel d'Ethique at the Centre de Recherche en Santé de Nouna in Nouna, Burkina Faso, and the Institutional Review Board at the University of California in San Francisco. Written informed consent was obtained from the caregiver of each participant. The trial was registered at clinicaltrials.gov (NCT03187834).
RESULTS {#s17}
=======
In July 2017, 248 children in 124 households were enrolled and randomized to 1 of the 3 antibiotic regimens or placebo ([Figure 1](#F1){ref-type="fig"}). Of these, 124 children were randomized within their household to receive treatment and are included in this analysis. Of these children, 120 (96.8%) had a rectal swab collected 5 days after their last antibiotic dose. The median age (interquartile range) was 36 (21--51) months, and 54.0% of the children were female ([Table 1](#T1){ref-type="table"}). One-quarter (27.4%) were currently breastfeeding, and 7.3% had received an antibiotic from a health facility in the last month. Baseline characteristics were well balanced across the 4 study arms. Adherence to study medication and adverse events have been previously reported \[[@CIT0030]\]. More than 90% of children received their allocated study medication at most time points. Adverse events were uncommon, and no diarrhea was noted in any of the antibiotic arms. 16S rRNA gene analysis of the fecal samples identified 429 unique genera. At both baseline and post-treatment, the most common genus was *Prevotella* spp. ([Supplementary Figure 1](#sup1){ref-type="supplementary-material"}).
{#F1}
######
Baseline Characteristics by Study Arm
---------------------------------------------------------------------------------------------------------
Amoxicillin\ Azithromycin\ Cotrimoxazole\ Placebo\
(n = 31) (n = 31) (n = 31) (n = 31)
------------------------------------------- -------------- --------------- ---------------- -------------
Age, median (IQR), mo 32 (23--48) 29 (21--51) 37 (23--48) 38 (23--51)
Female sex, No. (%) 15 (48.4) 18 (58.1) 20 (64.5) 14 (45.2)
Recently visited health facility, No. (%) 6 (19.4) 4 (12.9) 6 (19.4) 5 (16.1)
Recent antibiotic use, No. (%) 2 (6.5) 2 (6.5) 3 (9.7) 2 (6.5)
Breastfeeding, No. (%) 6 (19.4) 11 (35.5) 9 (29.0) 8 (25.8)
---------------------------------------------------------------------------------------------------------
Abbreviation: IQR, interquartile range.
There was no difference in Simpson's or Shannon's α-diversity across the 4 study arms at baseline (*P* = .55 and *P* = .59, respectively) ([Table 2](#T2){ref-type="table"}). Five days after the last antibiotic dose, there was an overall difference in α-diversity across all arms, as measured by Simpson's and Shannon's diversities (*P* = .003 and *P* = .0001, respectively) ([Table 2](#T2){ref-type="table"}). Results were robust to adjustment for the child's age (*P* = .005 for Simpson's and *P* = .002 for Shannon's diversity). The post-treatment mean Simpson's α-diversity effective number was 6.6 (95% confidence interval \[CI\], 5.5--7.8) in the azithromycin arm, 8.3 (95% CI, 7.0--9.6) in the amoxicillin arm, 8.3 (95% CI, 6.9--9.7) in the cotrimoxazole arm, and 9.8 (95% CI, 8.7--10.9) in the placebo arm ([Figure 2](#F2){ref-type="fig"}; [Table 2](#T2){ref-type="table"}). Simpson's diversity was significantly reduced in the azithromycin arm compared with placebo (*P* = .0001 by *t* test, *P* = .0001 by linear regression). Simpson's diversity index was nonsignificantly lower in children receiving amoxicillin (*P* = .09 by *t* test, *P* = .03 by linear regression) and co-trimoxazole (*P* = .08 by *t* test, *P* = .048 by linear regression) compared with placebo ([Table 2](#T2){ref-type="table"}). Differences in Shannon's diversity index were similar to the primary outcome (azithromycin vs placebo: *P* = .0002; amoxicillin vs placebo: *P* = .09; cotrimoxazole vs placebo: *P* = .048; all *t* test) ([Figure 2](#F2){ref-type="fig"}). Similarly, PERMANOVA analysis found a significant difference in L1 norm (equivalent to Shannon's diversity, *P* = .01) and L2 norm (equivalent to Simpson's diversity, *P* = .0001) across the study arms ([Figure 3](#F3){ref-type="fig"}).
######
Simpson's and Shannon's Alpha Diversity at Baseline and Post-treatment
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Amoxicillin Azithromycin Cotrimoxazole Placebo
------------------------------------------------------- -------------- -------------- --------------- ------------- -------------- -------------- -------------- --------------
Simpson (inverse) effective number, mean (95% CI) 10.2\ 8.3\ 8.8\ 6.6\ 9.7\ 8.3\ 9.6\ 9.8\
(8.8--11.5) (7.0--9.6) (7.5--10.1) (5.5--7.8) (8.2--11.2) (6.9--9.7) (8.6--10.7) (8.7--10.9)
Shannon (exponential) effective number, mean (95% CI) 16.6\ 13.9\ 14.6\ 11.0\ 15.6\ 13.5\ 15.4\ 16.0\
(14.5--18.7) (12.1--15.8) (13.0--16.2) (9.3--12.7) (13.4--17.8) (11.6--15.4) (14.1--16.7) (14.3--17.8)
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Abbreviation: CI, confidence interval.
{#F2}
{#F3}
DISCUSSION {#s18}
==========
A course of azithromycin significantly reduced intestinal bacterial diversity and composition in children in Burkina Faso, consistent with a previous study in Niger \[[@CIT0001]\]. In the previous study, children received a single dose of azithromycin or placebo, with the same dosing used by trachoma control programs and a recent trial investigating the effect of azithromycin on child survival \[[@CIT0015], [@CIT0016], [@CIT0031], [@CIT0032]\]. Here, we used a 5-day course with dosing equivalent to a Z-pack, a common course for treating mild infections in children such as otitis media. Taken together, the results of these studies suggest a definitive decrease in intestinal bacterial diversity following an azithromycin course in children.
The effect of amoxicillin or co-trimoxazole on intestinal bacterial diversity was less clear. Although there was a decrease in diversity between baseline and post-treatment for both antibiotics, the prespecified primary analysis was not statistically significant for either antibiotic. A sensitivity analysis including the baseline measurement for diversity yielded borderline significant results for both antibiotics, suggesting that there may be an effect of both antibiotics on the microbiome but that the present study may have been underpowered to detect smaller differences. A previous observational study among Finnish children showed a decrease in diversity with macrolides but not penicillins and suggested that macrolides had a greater effect on dysbiosis than pencillins \[[@CIT0004]\]. In the current study, because antibiotics were not being given for established infection, amoxicillin dosing was chosen to be the lower end of the approved range to minimize side effects. Cotrimoxazole was dosed following World Health Organization guidelines for cotrimoxazole prophylaxis for children living with HIV \[[@CIT0025], [@CIT0026], [@CIT0028]\]. Higher doses of cotrimoxazole or amoxicillin may yield different effects on the microbiome. Further, the condition in which the medications were stored may have altered their efficacy. Both cotrimoxazole and amoxicillin were purchased in Nouna, a rural area in Burkina Faso, where temperature monitoring of the medications was not observed. Therefore, higher doses of quality amoxicillin or cotrimoxazole may lead to greater changes in the composition of the intestinal microbiota. However, the use of locally sourced antibiotics simulates the conditions under which children in similar settings are treated, and thus these results are likely representative of intestinal microbiome changes following the use of similar antibiotics in west Africa.
Previous studies in high-income settings have indicated that dysbiosis of the intestinal microbiome may be associated with morbidity in children, including asthma \[[@CIT0009]\], food allergy \[[@CIT0033]\], and obesity \[[@CIT0034]\]. In low-income settings, differences in intestinal diversity have been noted in children with different forms of severe acute malnutrition \[[@CIT0012]\] and in twin pairs discordant for kwashiorkor \[[@CIT0008]\]. Antibiotics have been shown to lead to weight gain in children in randomized controlled trials \[[@CIT0030], [@CIT0035]\]. Alteration to the intestinal microbiome induced by antibiotic use may affect nutrient absorption or energy metabolism that can affect weight gain in children \[[@CIT0036], [@CIT0037]\], or weight gain after antibiotic use may be mediated by reduction in the burden of enteropathogens. However, the clinical implications of a single course of antibiotics remain unclear.
Several limitations should be noted. The duration of follow-up in the present study was short, and thus evaluated short-term changes in the intestinal microbiome following a course of antibiotics. Some studies have indicated that there may be longer-term changes in the microbiome following antibiotic use in children \[[@CIT0004]\]. It is possible that the changes observed in the present study are transient, or they may persist for several months. Recent antibiotic use was uncommon in the study population, but contamination of treatment assignment could have occurred if children used antibiotics other than the study medication during the course of the study. This study was conducted in a rural area of Burkina Faso that is characterized by high infection burden and high mortality. The results of this study may not be generalizable to children living in different settings, as the microbiome in children differs substantially geographically \[[@CIT0038]\]. However, the results of this study support previous observational and randomized studies that have indicated effects on intestinal microbial diversity following antibiotic use \[[@CIT0001], [@CIT0004], [@CIT0005], [@CIT0039]\].
In this randomized controlled trial of 3 commonly used antibiotics, we demonstrated that a short course of azithromycin led to a significant decrease in bacterial diversity of the intestinal microbiome in preschool children. Amoxicillin and cotrimoxazole dosing consistent with that used for prophylaxis in children living with HIV did not lead to a significant difference in bacterial diversity. Although the clinical implications of a single course of antibiotics are unclear, the results of this study indicate that the intestinal microbiome in young children is sensitive to antibiotics.
Supplementary Data {#s19}
==================
Supplementary materials are available at *Open Forum Infectious Diseases* online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
######
Click here for additional data file.
***Financial support*. **This work was supported in part by the Research to Prevent Blindness Career Development Award (to C.E.O. and T.D.) and the National Eye Institute of the National Institutes of Health under Award Number K08EY026986 (to T.D.). The funders did not play a role in study design, analysis, interpretation of results, or decision to publish.
***Potential conflicts of interest. ***All authors: no reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
| {
"pile_set_name": "PubMed Central"
} |
**Suggested citation:** EFSA CEF Panel (EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids) , Silano V, Bolognesi C, Castle L, Chipman K, Cravedi J‐P, Engel K‐H, Fowler P, Grob K, Gürtler R, Husøy T, Kärenlampi S, Mennes W, Pfaff K, Rivière G, Srinivasan J, Tavares Poças MF, Tlustos C, Wölfle D, Zorn H, Dudler V, Gontard N, Lampi E, Nerin C, Papaspyrides C, Volk K and Milana MR, 2018 Scientific Opinion on the safety assessment of the process 'Envases Ureña', based on Starlinger Decon technology, used to recycle post‐consumer PET into food contact materials. EFSA Journal 2018;16(1):5118, 13 pp. 10.2903/j.efsa.2018.5118
**Requestor:** Bundesamt für Verbraucherschutz und Lebensmittelsicherheit
**Question number:** EFSA‐Q‐2017‐00244
**Panel members:** Claudia Bolognesi, Laurence Castle, Kevin Chipman, Jean‐Pierre Cravedi, Karl‐Heinz Engel, Paul Fowler, Roland Franz, Konrad Grob, Rainer Gürtler, Trine Husøy, Sirpa Kärenlampi, Wim Mennes, Maria Rosaria Milana, Karla Pfaff, Gilles Rivière, Vittorio Silano, Jannavi Srinivasan, Maria de Fátima Tavares Poças, Christina Tlustos, Detlef Wölfle and Holger Zorn.
**Competing interests:** In line with EFSA\'s policy on declarations of interest, Roland Franz did not participate in the development and adoption of this scientific output.
Adopted: 29 November 2017
1. Introduction {#efs25118-sec-0002}
===============
1.1. Background and Terms of Reference as provided by the requestor {#efs25118-sec-0003}
-------------------------------------------------------------------
Recycled plastic materials and articles shall only be placed on the market if they contain recycled plastic obtained from an authorised recycling process. Before a recycling process is authorised, European Food Safety Authority (EFSA)′s opinion on its safety is required. This procedure has been established in Article 5 of the Regulation (EC) No 282/2008[1](#efs25118-note-1006){ref-type="fn"} of the Commission of 27 March 2008 on recycled plastic materials intended to come into contact with foods and Articles 8 and 9 of Regulation (EC) No 1935/2004[2](#efs25118-note-1007){ref-type="fn"} of the European Parliament and of the Council of 27 October 2004 on materials and articles intended to come into contact with food.
According to this procedure, the industry submits applications to the Member States Competent Authorities which transmit the applications to EFSA for evaluation.
In this case, EFSA received, from the Bundesamt für Verbraucherschutz und Lebensmittelsicherheit, Germany, an application for evaluation of the recycling process Envases Ureña, European Union (EU) register No RECYC0147. The request has been registered in EFSA\'s register of received questions under the number EFSA‐Q‐2017‐00244. The dossier was submitted on behalf of Envases Ureña S.A., Spain.
According to Article 5 of Regulation (EC) No 282/2008 of the Commission of 27 March 2008 on recycled plastic materials intended to come into contact with foods, EFSA is required to carry out risk assessments on the risks originating from the migration of substances from recycled food contact plastic materials and articles into food and deliver a scientific opinion on the recycling process examined.
According to Article 4 of Regulation (EC) No 282/2008, EFSA will evaluate whether it has been demonstrated in a challenge test, or by other appropriate scientific evidence, that the recycling process Envases Ureña is able to reduce any contamination of the plastic input to a concentration that does not pose a risk to human health. The poly(ethylene terephthalate) (PET) materials and articles used as input of the process as well as the conditions of use of the recycled PET make part of this evaluation.
2. Data and methodologies {#efs25118-sec-0004}
=========================
2.1. Data {#efs25118-sec-0005}
---------
The applicant has submitted a dossier following the 'EFSA guidelines for the submission of an application for the safety evaluation of a recycling process to produce recycled plastics intended to be used for the manufacture of materials and articles in contact with food, prior to its authorization' (EFSA, [2008](#efs25118-bib-0001){ref-type="ref"}). Applications shall be submitted in accordance with Article 5 of the Regulation (EC) No 282/2008.
The following information on the recycling process was provided by the applicant and used for the evaluation: General information:--general description--existing authorisationsSpecific information:--recycling process--characterisation of the input--determination of the decontamination efficiency of the recycling process--characterisation of the recycled plastic--intended application in contact with food--compliance with the relevant provisions on food contact materials and articles--process analysis and evaluation--operating parameters
2.2. Methodologies {#efs25118-sec-0006}
------------------
The principles followed for the evaluation are described here. The risks associated to the use of recycled plastic materials and articles in contact with food come from the possible migration of chemicals into the food in amounts that would endanger human health. The quality of the input, the efficiency of the recycling process to remove contaminants as well as the intended use of the recycled plastic are crucial points for the risk assessment (see guidelines on recycling plastics; EFSA, [2008](#efs25118-bib-0001){ref-type="ref"}).
The criteria for the safety evaluation of a mechanical recycling process to produce recycled PET intended to be used for the manufacture of materials and articles in contact with food are described in the scientific opinion developed by the EFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing Aids (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}). The principle of the evaluation is to apply the decontamination efficiency of a recycling technology or process, obtained from a challenge test with surrogate contaminants, to a reference contamination level for post‐consumer PET, conservatively set at 3 mg/kg PET for contaminants resulting from possible misuse. The resulting residual concentration of each surrogate contaminant in recycled PET (C~res~) is compared with a modelled concentration of the surrogate contaminants in PET (C~mod~). This C~mod~ is calculated using generally recognised conservative migration models so that the related migration does not give rise to a dietary exposure exceeding 0.0025 μg/kg bodyweight (bw) per day (i.e. the human exposure threshold value for chemicals with structural alerts for genotoxicity), below which the risk to human health would be negligible. If the C~res~ is not higher than the C~mod~, the recycled PET manufactured by such recycling process is not considered of safety concern for the defined conditions of use (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}).
The assessment was conducted in line with the principles described in the EFSA Guidance on transparency in the scientific aspects of risk assessment (EFSA, [2009](#efs25118-bib-0002){ref-type="ref"}) and considering the relevant existing guidance from the EFSA Scientific Committee.
3. Assessment {#efs25118-sec-0007}
=============
3.1. General information {#efs25118-sec-0008}
------------------------
According to the applicant, the recycling process Envases Ureña is intended to recycle food grade PET containers to produce recycled PET flakes using the Starlinger Decon technology. The recycled flakes are intended to be used up to 100% for the manufacture of recycled materials and articles. These final materials and articles are intended to be used in direct contact with all kinds of foodstuffs for long‐term storage at room temperature, with or without hotfill.
3.2. Description of the process {#efs25118-sec-0009}
-------------------------------
### 3.2.1. General description {#efs25118-sec-0010}
The recycling process Envases Ureña produces recycled PET flakes from PET containers, mainly bottles, coming from post‐consumer collection systems (kerbside and deposit systems). The recycling process comprises of the three steps below. The first step may be performed by a third party or by the applicant.
[Input]{.ul} In Step 1, post‐consumer PET containers, mainly bottles and trays, are processed into hot caustic washed and dried flakes, which are used as input of the process.
[Decontamination and production of recycled PET material]{.ul} In Step 2, the flakes are preheated in one or several batch reactors with a flow of hot gas.In Step 3, the preheated flakes are submitted to solid‐state polycondensation (SSP) in a continuous reactor (one single reactor or several reactors) at high temperature using a combination of vacuum and gas flow.
The operating conditions of the process have been provided to EFSA.
Recycled flakes, the final product of the process, are checked against technical requirements on intrinsic viscosity, colour, black spots, etc. Recycled flakes are intended to be converted by the applicant or by other companies into recycled articles used for hotfill and/or long‐term storage at room temperature, such as bottles for mineral water, soft drinks and beer. The recycled flakes may also be used for sheets which are thermoformed to make food trays. Trays made of this PET are not intended to be used in microwave and conventional ovens.
### 3.2.2. Characterisation of the input {#efs25118-sec-0011}
According to the applicant, the input material for the recycling process Envases Ureña is hot caustic washed and dried flakes obtained from PET containers, mainly bottles, previously used for food packaging, from post‐consumer collection systems (kerbside and deposit systems). A small fraction may originate from non‐food applications. According to the applicant, the amount of this non‐food container fraction depends on the re‐collection system. On the basis of market share data, the applicant indicated that this fraction is below 5%.
Technical data for the washed and dried flakes are provided such as information on residual content of moisture, poly(vinyl chloride) (PVC), glue, wood/paper, metals, other plastics and physical properties (see Appendix [A](#efs25118-sec-0020){ref-type="sec"}).
3.3. Starlinger Decon technology {#efs25118-sec-0012}
--------------------------------
### 3.3.1. Description of the main steps {#efs25118-sec-0013}
To decontaminate post‐consumer PET, the recycling process Envases Ureña uses the Starlinger Decon technology as described below and for which the general scheme, provided by the applicant, is reported in Figure [1](#efs25118-fig-0001){ref-type="fig"}. In step 1, not reported in the scheme, post‐consumer PET containers, mainly bottles, are processed into hot caustic washed and dried flakes from third parties or by the applicant. [Preheating (step 2):]{.ul} The flakes are preheated in one or several batch reactors with a flow of hot gas up to the temperature of the next step.[Solid‐state polycondensation (SSP) (step 3):]{.ul} The flakes from the batch preheater are fed into the SSP reactor. More than one SSP reactor can be run in parallel continuously under high temperature and using a combination of gas flow and vacuum for a predefined residence time. In particular, the SSP reactors remain under vacuum with values varying depending on the feeding mode, while gas flow is applied periodically to support the removal of the contaminants from the flakes. This step increases the intrinsic viscosity of the material and further decontaminates the PET flakes.
{#efs25118-fig-0001}
The process is operated under defined operating parameters of temperature, pressure, gas flow and residence time.[3](#efs25118-note-1008){ref-type="fn"}
### 3.3.2. Decontamination efficiency of the recycling process {#efs25118-sec-0014}
To demonstrate the decontamination efficiency of the recycling process Envases Ureña, a challenge test on the Starlinger Decon technology was submitted to the EFSA.
PET flakes were contaminated with toluene, chlorobenzene, phenylcyclohexane, chloroform, methyl salicylate, benzophenone and methylstearate, selected as surrogate contaminants. The surrogates were chosen in agreement with EFSA guidelines and in accordance with the recommendations of the US Food and Drug Administration. The surrogates include different molecular weights and polarities to cover possible chemical classes of contaminants of concern and were demonstrated to be suitable to monitor the behaviour of PET during recycling (EFSA, [2008](#efs25118-bib-0001){ref-type="ref"}).
For the preparation of the contaminated PET flakes, conventionally recycled[4](#efs25118-note-1009){ref-type="fn"} post‐consumer PET flakes were soaked in a mixture of surrogates and stored for 7 days at 50°C with daily agitation. The contaminated PET flakes were washed with hot water and detergents then air dried. The concentrations of surrogates in this material were determined.
The Starlinger Decon technology was challenged at a pilot plant. The preheater reactor was filled with washed and dried contaminated flakes only (step 2). Preheated flakes were then fed into the SSP reactor (step 3). The flakes were analysed after each step for their residual concentrations of the applied surrogates. In both batch and continuous modes of operation the surrogates diffuse through the flakes to the surface and they are constantly eliminated by the gas flow applied. Therefore, in this case, continuous working processes will result in equivalent cleaning efficiencies as batch processes as long as the same temperature, pressure conditions, gas flow and residence time are applied.
The decontamination efficiency of the process was calculated taking into account the amount of the surrogates detected in washed contaminated flakes before the preheating (before step 2) and after SSP (step 3). The results are summarised below in Table [1](#efs25118-tbl-0001){ref-type="table"}.
######
Efficiency of the decontamination of the Starlinger Decon technology in the challenge test
Surrogates Concentration of surrogates before step 2 (mg/kg PET) Concentration of surrogates after step 3 (mg/kg PET) Decontamination Efficiency (%)
------------------- ------------------------------------------------------- ------------------------------------------------------ --------------------------------
Toluene 206.9 1.1 99.5
Chlorobenzene 393.1 2.1 99.5
Chloroform 120.2 3.4 97.2
Methyl salicylate 369 4.1 98.9
Phenylcyclohexane 404 6.9 98.3
Benzophenone 594.4 22.1 96.3
Methyl stearate 743.4 27.1 96.4
PET: poly(ethylene terephthalate).
John Wiley & Sons, Ltd
As shown in Table [1](#efs25118-tbl-0001){ref-type="table"}, the decontamination efficiency ranged from 96.3% for benzophenone to 99.5% for toluene and chlorobenzene.
3.4. Discussion {#efs25118-sec-0015}
---------------
Considering the high temperatures used during the process, the possibility of contamination by microorganisms can be discounted. Therefore, this evaluation focuses on the chemical safety of the final product.
Technical data such as information on residual content of moisture, poly(vinyl chloride) (PVC), glue, wood/paper, metals, other plastics and physical properties are provided for the input materials (washed and dried flakes (step 1)), for the submitted recycling process. The input materials are produced from PET containers, mainly bottles, previously used for food packaging collected through post‐consumer collection systems. However, a small fraction of the input may originate from non‐food applications. According to the applicant, the amount of this non‐food container fraction depends on the collection system, and on the basis of market share data it is below 5%, as recommended by the EFSA CEF Panel in its 'Scientific opinion on the criteria to be used for safety evaluation of a mechanical recycling process to produce recycled PET intended to be used for manufacture of materials and articles in contact with food' (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}).
The process is well described. The washing and drying of flakes from collected PET containers (step 1) is conducted in different ways depending on the plant but, according to the applicant, this step is under control. The following steps are those of the Starlinger Decon technology used to recycle the PET flakes into decontaminated PET flakes: batch preheating (step 2) and continuous SSP (step 3). The operating parameters of temperature, residence time, pressure and gas flow for both steps have been provided to EFSA.
A challenge test was conducted at a pilot plant scale on process steps 2 and 3 (preheating and SSP reactor) to measure the decontamination efficiency. The Panel considered that the challenge test was performed correctly according to the recommendations in the EFSA guidelines (EFSA, [2008](#efs25118-bib-0001){ref-type="ref"}) and that the steps 2 and 3 are the critical steps for the decontamination efficiency of the process. Consequently, temperature, residence time, pressure and gas flow parameters of steps 2 and 3 of the process should be controlled to guarantee the performance of the decontamination. These parameters have been provided to EFSA.
The decontamination efficiencies obtained from for each surrogate contaminant from the challenge test performed on steps 2 and 3, ranging from 96.3% to 99.5%, have been used to calculate the residual concentrations of potential unknown contaminants in PET (C~res~) according to the evaluation procedure described in the 'Scientific opinion on the criteria to be used for safety evaluation of a mechanical recycling process to produce recycled PET' (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}; Appendix [B](#efs25118-sec-0021){ref-type="sec"}). By applying the decontamination efficiency percentage to the reference contamination level of 3 mg/kg PET, the C~res~ for the different surrogates is obtained (Table [2](#efs25118-tbl-0002){ref-type="table"}).
According to the evaluation principles (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}), the C~res~ value should not be higher than a modelled concentration in PET (C~mod~) corresponding to a migration, after 1 year at 25°C, which cannot give rise to a dietary exposure exceeding 0.0025 μg/kg bw per day, the exposure threshold below which the risk to human health would be negligible.[5](#efs25118-note-1010){ref-type="fn"} Because the recycled PET is intended for general use for the manufacturing of articles containing up to 100% recycled PET, the most conservative default scenario for infants has been applied. Therefore, the migration of 0.1 μg/kg into food has been used to calculate C~mod~ (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}). The results of these calculations are shown in Table [2](#efs25118-tbl-0002){ref-type="table"}. The relationship between the key parameters for the evaluation scheme is reported in Appendix [B](#efs25118-sec-0021){ref-type="sec"}.
######
Decontamination efficiency from challenge test, residual concentration of surrogate contaminants in recycled PET (C~res~) and calculated concentration of surrogate contaminants in PET (C~mod~) corresponding to a modelled migration of 0.1 μg/kg food after 1 year at 25°C
Surrogates Decontamination efficiency (%) C~res~ (mg/kg PET) C~mod~ (mg/kg PET)
------------------- -------------------------------- -------------------- --------------------
Toluene 99.5 0.02 0.09
Chlorobenzene 99.5 0.02 0.10
Chloroform 97.2 0.08 0.10
Methyl salicylate 98.9 0.03 0.13
Phenylcyclohexane 98.3 0.05 0.14
Benzophenone 96.3 0.11 0.16
Methyl stearate 96.4 0.11 0.32
PET: poly(ethylene terephthalate); C~res~: residual concentration in PET; C~mod~: modelled concentration in PET.
John Wiley & Sons, Ltd
The residual concentrations of all surrogates in PET after decontamination (C~res~) are lower than the corresponding modelled concentrations in PET (C~mod~). Therefore, the Panel considered that the recycling process under evaluation using the Starlinger Decon technology is able to ensure that the level of migration of unknown contaminants from the recycled PET into food is below the conservatively modelled migration of 0.1 μg/kg food at which the risk to human health would be negligible.
4. Conclusions {#efs25118-sec-0016}
==============
The Panel considered that the process Envases Ureña is well characterised and the main steps used to recycle the PET flakes into decontaminated PET flakes have been identified. Having examined the challenge test provided, the Panel concluded that the preheating (step 2) and the decontamination in the continuous SSP reactor/s (step 3) are the critical steps for the decontamination efficiency of the process. The operating parameters to control its performance are temperature, residence time, pressure and gas flow. Therefore, the Panel considered that the recycling process Envases Ureña is able to reduce any foreseeable accidental contamination of the post‐consumer food contact PET to a concentration that does not give rise to concern for a risk to human health if: it is operated under conditions that are at least as severe as those obtained from the challenge test used to measure the decontamination efficiency of the process;the input of the process is washed and dried post‐consumer PET flakes originating from materials and articles that have been manufactured in accordance with the EU legislation on food contact materials containing no more than 5% of PET from non‐food consumer applications.
Therefore, the recycled PET obtained from the process Envases Ureña intended to be used up to 100% for the manufacture of materials and articles for contact with all types of foodstuffs for long‐term storage at room temperature, with or without hotfill, is not considered of safety concern. Trays made of this recycled PET are not intended to be used and should not be used in microwave and conventional ovens.
5. Recommendations {#efs25118-sec-0017}
==================
The Panel recommended periodic verification that the input to be recycled originates from materials and articles that have been manufactured in accordance with the EU legislation on food contact materials and that the proportion of PET from non‐food consumer applications is no more than 5%. This adheres to good manufacturing practice and Regulation (EC) No 282/2008, Art. 4b. Critical steps in recycling should be monitored and kept under control. In addition, supporting documentation should be available on how it is ensured that the critical steps are operated under conditions at least as severe as those in the challenge test used to measure the decontamination efficiency of the process.
Documentation provided to EFSA {#efs25118-sec-0018}
==============================
Dossier 'Envases Ureña'. March 2017. Submitted on behalf of Envases Ureña S.A., Spain.Additional data 'Envases Ureña'. November 2017. Submitted on behalf of Envases Ureña S.A., Spain.
Abbreviations {#efs25118-sec-0019}
=============
bwbody weightCEF PanelEFSA Panel on Food Contact Materials, Enzymes, Flavourings and Processing AidsC~mod~modelled concentration in PETC~res~residual concentration in PETECEuropean CommissioniVintrinsic viscosityPETpoly(ethylene terephthalate)PVCpoly(vinyl chloride)SSPsolid‐state polycondensation
Appendix A -- Technical data of the washed flakes as provided by the applicant {#efs25118-sec-0020}
==============================================================================
ParameterValueMoisture\< 2%PVC content\< 100 ppmOther plastics than PET (multilayer (Polyamide))\< 750 ppmFlakes with glue content\< 4000 ppmMetal content (aluminium, ferrous, others)\< 500 ppmWood, paper content\< 100 ppmiV[a](#efs25118-note-1011){ref-type="fn"}0.65--1.2 dL/gTotal contamination other than water\< 2000 ppmBulk density250--750 kg/m^3^Flakes size1--15 mmFlakes thickness50--1200 μm[^1]
Appendix B -- Relationship between the key parameters for the evaluation scheme (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}) {#efs25118-sec-0021}
============================================================================================================================================
\*Default scenario (infant). For adults and toddlers, the migration criterion will be 0.75 and 0.15 μg/kg food, respectively.
Appendix C -- Table of Operational Parameters (Confidential Information) {#efs25118-sec-0022}
========================================================================
Regulation (EC) No 282/2008 of the European parliament and of the council of 27 March 2008 on recycled plastic materials and articles intended to come into contact with foods and amending Regulation (EC) No 2023/2006. OJ L 86, 28.3.2008, p. 9--18.
Regulation (EC) No 1935/2004 of the European parliament and of the council of 27 October 2004 on materials and articles intended to come into contact with food and repealing Directives 80/590/EEC and 89/109/EEC. OJ L 338, 13.11.2004, p. 4--17.
In accordance with Art. 9 and 20 of Regulation (EC) No 1935/2004 the parameters were provided to EFSA and made available to the applicant, the Member States and European Commission (see Appendix [C](#efs25118-sec-0022){ref-type="sec"}).
Conventional recycling includes commonly sorting, grinding, washing and drying steps and produces washed and dried flakes.
0.0025 μg/kg bw per day is the human exposure threshold value for chemicals with structural alerts raising concern for potential genotoxicity, below which the risk to human health would be negligible (EFSA CEF Panel, [2011](#efs25118-bib-0003){ref-type="ref"}).
[^1]: Value refers to food grade PET flakes, after the recycling process.
| {
"pile_set_name": "PubMed Central"
} |
The data underlying the results presented in the study are available from <http://janus.northwestern.edu/janus2/index.php>.
Introduction {#sec001}
============
Ionizing radiation is an unavoidable risk in daily life and understanding its biological impacts is important for setting radiation protection standards. Approximately half of humankind's cumulative annual radiation exposure comes from natural sources, such as cosmic radiation and soil; the other half is derived from human-made sources including medical procedures and nuclear medicine \[[@pone.0231510.ref001]\]. Most of the general population receives low dose chronic ionizing radiation exposures, accumulating to a few hundred mSv over a lifetime \[[@pone.0231510.ref002]\].
Calculating the risks associated with these chronic exposures is challenging because the overall effect of these lower dose/dose-rate exposures is small compared to the baseline risk of the same diseases. There are several helpful data sources that researchers have utilized to help quantify these risks including radiation therapy studies, atomic bomb survivor data, and other epidemiological studies from workers in the field and nuclear disasters. The radiation doses given to patients for radiation therapy are much larger than standard exposure and are only used on a small segment of a patient, not whole-body exposures. The major source of data on whole-body human exposures to gamma radiation is the Life Span Study (LSS) cohort that includes over 120,000 survivors of the atomic bombing in 1945 \[[@pone.0231510.ref003]--[@pone.0231510.ref006]\]. While these data have been a remarkable resource for epidemiological studies determining risks associated with acute exposures \[[@pone.0231510.ref005], [@pone.0231510.ref007]--[@pone.0231510.ref013]\], extrapolation of health risks to humans exposed intermittently to lower doses of radiation remains uncertain. Different mathematical modelling approaches have been used over the past 50 years to extrapolate health risks but they were met with variable enthusiasm from the scientific community \[[@pone.0231510.ref001], [@pone.0231510.ref014]--[@pone.0231510.ref018]\]. Ultimately, epidemiological studies are affected by confounding factors and uncertainties making well-controlled animal studies a valuable resource to supplement conclusions from human studies \[[@pone.0231510.ref019]\].
We utilized the Northwestern University Radiation Archive for animals (NURA), a source of irradiated animal data documenting findings from Janus studies conducted between 1972 and 1989 at Argonne National Laboratory (ANL). Ten large volume experiments with B6CF1 mice were designed to determine the effects of acute and fractionated whole-body radiation on survival and causes of death \[[@pone.0231510.ref020]--[@pone.0231510.ref022]\]. Over 50,000 male and female mice were exposed to acute or fractionated neutrons or gamma rays, at ages between 90--200 days or more than 500 days. Moribund animals were sacrificed and necropsy results were recorded. This experiment was one of the of the largest ever conducted in the USA; at the conclusion of these studies the Janus irradiator and other irradiation facilities at ANL were dismantled making it unlikely that experiments of this scope will be repeated. Numerous studies used the NURA (also known as Janus) database. In most cases, different Janus experiments were used separately \[[@pone.0231510.ref017], [@pone.0231510.ref020], [@pone.0231510.ref023]--[@pone.0231510.ref028]\] or else combined all together into a single dataset \[[@pone.0231510.ref014]\]. In this study, however, many but not all Janus experiments were combined into a dataset. The selection process for inclusion was based on comparability of control animal datasets from sham irradiation conditions in different Janus experiments.
Similar studies on other strains of mice were conducted in Europe \[[@pone.0231510.ref029]\] and more recent work conducted at the Institute of Environmental Sciences (IES) in Japan explored chronic exposures in a closely related animal strain. Tanaka and others compared findings on 4,000 B6C3F1 mice of both genders that were irradiated for 22h daily for approximately 400 days using low dose rate gamma rays with accumulating total doses of 0, 20, 400, or 8000 mGy \[[@pone.0231510.ref030]\]. As the experiments performed at IES are similar to Janus experiments with regard to number of animals and total doses, we performed a side by side comparison of cancer incidence to determine biological similarities of these findings.
We examined whether fractionation, age at which a mouse was first irradiated, and gender modulated the overall death hazard and frequency for specific causes of death in gamma irradiated mice. Our approach included use of general Cox proportional hazards models, cumulative incidence function models, and cause specific hazards models \[[@pone.0231510.ref031]--[@pone.0231510.ref033]\]. We found that the two approaches to represent and evaluate competing risks from the same data complement each other and improve insight into effects of gamma ray fractionation. While this work cannot be directly translated into recommendations for radiation protection policies, it brings to our attention the fact that it should be possible to standardize comparisons between different types of fractionated exposures and perhaps fractionated versus chronic radiation exposures across a large span of total doses. A single mathematical formula cannot be used universally for conversion between any two possible radiation exposure scenarios, but it is possible that the growth of machine learning and artificial intelligence techniques will permit us to craft realistic approaches to predict changes in health complication spectra from one irradiation exposure to another, possibly even among different species. As we prepare for this future, it is necessary to ensure that we preserve radiation data archives with as much granularity as possible. This study is a prime example of utilizing archives by analyzing data in a new light to further augment our understanding of radiation biology.
Methods {#sec002}
=======
Data selection---NURA {#sec003}
---------------------
Argonne National Laboratory conducted a series of 10 large scale ionizing radiation lifespan studies on rodents between 1972 and 1989. These studies are now part of the NURA archive housed by the Woloschak laboratory and posted on the web, allowing access to all who are interested in this dataset \[[@pone.0231510.ref014], [@pone.0231510.ref021], [@pone.0231510.ref022], [@pone.0231510.ref034]\]. Records list individual mouse information with the type of radiation, total dose, dose rate, fractionation schedule, age first irradiated, age at death, cause of death and, in many cases, detailed pathology analyses. All animals received whole body external beam ionizing radiation from cobalt 60 gamma rays or neutrons \[[@pone.0231510.ref020]--[@pone.0231510.ref022], [@pone.0231510.ref035]\]. Most control mice were sham irradiated--transported from their housing location to the room with the irradiator turned off. Background radiation levels in animal housing rooms were closely monitored. Mice listed in S1 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} were censored due to early exit from the study because of causes unrelated to the experimental plan.
To ensure that any significant changes among different groups of mice were due to modulations in radiation exposure and not due to changes in baseline survival, we filtered out groups of mice that showed statistically significant survival probability differences. S2 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} details groups of mice that exhibited sufficient survival variation in control animals to warrant removal of the specific data from our analysis. For this analysis, we focused on gamma irradiated mice. Neutron irradiated mice were studied in a separate analysis. Breeder mice were not used as controls for any of these analyses because of their unique housing conditions. Only *Mus musculus* species B6CF1 strain mice were used for this work; different species such as *Peromyscus leucopus* (white-footed deer mouse) were excluded from this study because of the species to species differences between the controls and in response to radiation \[[@pone.0231510.ref027]\]. Similarly, mice treated with radioprotectors \[[@pone.0231510.ref028]\] were also removed from this study. As a result of data refinement, only two of the ten experiments were completely removed from this study.
The predicted model output graphs from Cox Proportional Hazards (PH) analyses of sham irradiated control mice are shown in S1 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} along with parameter estimates and p-values. Each overall model was significant due to sex, but no other covariates were significant in their respective models. Additionally, Kaplan Meier (KM) curves \[[@pone.0231510.ref036]\] in S1e--S1h Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} validate the proportional hazards assumption for our model. To further validate our model, we used robustness tests, making small modifications to each variable in our models, shown in S3 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} and described in more detail in the Supplementary methods in [S1 File](#pone.0231510.s001){ref-type="supplementary-material"}.
Survival analysis {#sec004}
-----------------
Kaplan-Meier (KM) curves were used for categorical univariate survival analysis using the "survfit" function in the survival package in R \[[@pone.0231510.ref037], [@pone.0231510.ref038]\]. Cox proportional hazard (PH) models were used to analyze survival over time with multivariate models that included a mixture of categorical and quantitative predictor variables and interactions between variables \[[@pone.0231510.ref031]\]. The main models used for Cox PH with sham irradiated mice and shown in S1 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} are as follows: $$\lambda\left( t \right) = \lambda_{0}\left( t \right)e^{({\beta_{1}sex + \beta_{2}experiment})}\mspace{72mu}\left( A \right)$$ $$\lambda\left( t \right) = \lambda_{0}\left( t \right)e^{({\beta_{1}sex + \beta_{2}fractions})}\mspace{72mu}\left( B \right)$$ $$\lambda\left( t \right) = \lambda_{0}\left( t \right)e^{({\beta_{1}sex + \beta_{2}first\ irrad})}\mspace{72mu}\left( C \right)$$
Our main Cox PH model for gamma irradiated mice is as follows: $$\lambda\left( t \right) = \lambda_{0}\left( t \right)e^{({\beta_{1}sex + \beta_{2}first\ irrad + \beta_{3}total\ dose + \beta_{4}fractions + \beta_{5}total\ dose:fractions})},$$ where *λ*(*t*) is the hazard function based on our set of covariates including sex, age first irradiated, total dose, number of fractions, and the interaction between total dose and fractions; β is a vector of their corresponding coefficients, and *λ*~0~(*t*) is the baseline hazard. All Cox PH models were performed using the coxph function from the survival package in R \[[@pone.0231510.ref038]\].
Competing risks analysis {#sec005}
------------------------
A competing risk is anything that decreases the likelihood of an outcome of interest. When looking at specific causes of death, all other causes of death fall into this category. For the competing risks analysis, we examined crude incidences, cause-specific hazard models, and cumulative incidence function (CIF) regression models \[[@pone.0231510.ref039]\]. In the absence of competing risks, the cumulative incidence of events over time can be measured using one minus the Kaplan-Meier estimate of the survival function. In the presence of competing risks, the KM method results in upward biases for the CIF \[[@pone.0231510.ref032]\]. We used the "cuminc" function from the "cmprsk" package in R to investigate crude, nonparametric incidences in the presence of competing risks \[[@pone.0231510.ref040]\].
For multi-variable regression analyses in the presence of competing risks, we used both cause specific hazards and CIF models. The cause specific hazards were estimated using the "coxph" function in R \[[@pone.0231510.ref038]\]. All causes of death, excluding the event of interest, were censored. Concretely: $\lambda_{k}\left( t \right) = \lambda_{0k}\left( t \right)e^{({\beta_{1}sex + \beta_{2}first\ irrad + \beta_{3}total\ dose + \beta_{4}fractions + \beta_{5}total\ dose:fractions})}$, subset data for age first irradiated \< 500 days.
where *λ~k~*(*t*) is the hazard function for the k^th^ cause of death. The cause specific hazards method is used to determine the effect that covariates have on all event free subjects. The cumulative incidence function describes the overall probability of a particular outcome and does not depend on a subject being event free \[[@pone.0231510.ref032], [@pone.0231510.ref033], [@pone.0231510.ref041], [@pone.0231510.ref042]\]. Concretely: $\lambda_{k}^{*}\left( t \right) = \lambda_{0k}\left( t \right)e^{({\beta_{1}sex + \beta_{2}firsr\ irrad + \beta_{3}total\ dose + \beta_{4}fractions + \beta_{5}total\ dose:fractions})}$, subset data for age first irradiated \< 500 days.
where $\lambda_{k}^{*}\left( t \right)$ is the subdistribution hazard function for the k^th^ cause of death. Cumulative incidence hazards were estimated using the "crr" function in the "cmprsk" package in R \[[@pone.0231510.ref040]\].
Cause of death groupings {#sec006}
------------------------
We used data downloaded from the Janus website listed as "Grouped Macros," which includes all pathologies found in animals at the time of death and categorizes them as lethal (L), contributory (C), or non-contributory (N). For the purposes of our investigation, we only examined lethal diseases. To make the data more robust for analyses, we grouped causes of death (CODs) into lymphomas, tumors other than lymphomas--referred to as tumors (sometimes separating them into lung tumors and tumors or no lung tumors), non-tumors, or causes of death unknown (CDU). Specific analyses of diseases affecting the liver, lung, kidney, and vascular system for a subset of these data were previously conducted \[[@pone.0231510.ref043]\].
Data reformatting for comparisons with IES data {#sec007}
-----------------------------------------------
Studies at IES involved low dose rate gamma irradiations of specific-pathogen free (SPF) B6C3F1 mice, F1 progeny of C57BL/6J females and C3H/HeJ males. The B6CF1 mice, F1 progeny of C57BL/6J females and BALB/cJ males, were used during the Janus experiments. Both strains are F1 hybrids that share the same maternal strain C57BL/6. The differences in disease incidence between the control animals point out that only some of these disease "endpoints" are appropriate for direct comparisons between strains when different "test conditions" are being evaluated.
During the IES studies, (SPF) B6C3F1 mice were irradiated with low dose rate ^137^Cs gamma rays for 22 hours a day, beginning the irradiations with acclimated 8-week old animals in a sterile environment. Chronic exposures of 0.05, 1.1 or 21 mGy/day continued for 400 days leading to total doses of 2, 40, or 800 cGy. Similar to the Janus experiments, many mice from the IES experiments were allowed to live out their entire lifespan, and each mouse was assigned a single cause of death \[[@pone.0231510.ref030], [@pone.0231510.ref044]\]. These data were included in this study because of their similarity to the experiments carried out on B6CF1 mice during the NURA experiments.
To compare the Janus data included in this study with the results from the IES studies \[[@pone.0231510.ref030]\], we grouped Janus CODs to match IES CODs as closely as possible (S4 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). The total doses used for Janus experiments spanned a larger range than those used for IES experiments. To make the comparisons more meaningful, we limited the Janus data used for this particular analysis to a subset of conditions that closely matched the IES dataset conditions. S5 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"} provides information about the Janus data included in this comparison.
Tools and scripts {#sec008}
-----------------
Files stored on github: <https://github.com/aliazander>
Results {#sec009}
=======
Mice sham irradiated with 300 fractions had decreased survival {#sec010}
--------------------------------------------------------------
Control mice that received 300 fractions (5 fractions/week) of sham irradiation died significantly earlier than control animals that received fewer than 300 fractions of sham irradiation. This was evident in Cox PH models using sex and fractions as independent variables. This result held true with fractions treated as a continuous variable ([Fig 1A and 1B](#pone.0231510.g001){ref-type="fig"}, p-value \<0.001) and as a categorical variable (S2A and S2B Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p-value \<0.001). KM curves showed a very similar trend to the predicted outcomes from the Cox PH models and validated the proportional hazards assumption of the Cox PH model (S2C Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). Because general stress is the only probable cause for the increased death hazard observed in this group of animals and irradiated animals exposed to 300 fractions most likely experienced the same stress, we excluded mice exposed to 300 fractions from our main analysis. These mice were included as part our robustness testing.
{#pone.0231510.g001}
Mice sham irradiated with 300 fractions had significant changes in causes of death compared to other control mice {#sec011}
-----------------------------------------------------------------------------------------------------------------
The decreased survival in mice that received 300 fractions compared to fewer than 300 fractions during their sham irradiations led us to investigate the specific causes of death for these two groups of mice. All animals sham irradiated with 300 fractions were male. Using non-parametric CIF, we found significant increases in lymphoma and CDU incidences and a significant decrease in lung cancer incidences in mice that received 300 sham irradiation fractions compared to all other sham irradiated male mice ([Fig 1](#pone.0231510.g001){ref-type="fig"}). Additionally, we examined how the number of fractions impacted survival probability over time through KM curves for each COD (S2D--S2H Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}) and closer examination of the CIF curves in [Fig 1C](#pone.0231510.g001){ref-type="fig"} by individually plotting each COD ([Fig 1D--1H](#pone.0231510.g001){ref-type="fig"}). The initial onset for lung tumor deaths ([Fig 1E](#pone.0231510.g001){ref-type="fig"}; S2E Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p = 0.04), CDU deaths ([Fig 1H](#pone.0231510.g001){ref-type="fig"}; S2H Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p \<0.001), and lymphomas ([Fig 1F](#pone.0231510.g001){ref-type="fig"}; S2E Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p = 0.02) was earlier when mice received sham irradiations in 300 fractions, but there did not appear to be a difference for tumors (excluding lung tumors) ([Fig 1E](#pone.0231510.g001){ref-type="fig"}; S2C Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p = 0.35) or non-tumors deaths ([Fig 1G](#pone.0231510.g001){ref-type="fig"}; S2G Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}, p = 0.22). The KM curves also supported these findings (S2C--S2H Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}).
Aged mice were excluded from our analysis because of uneven experimental conditions {#sec012}
-----------------------------------------------------------------------------------
All control and gamma irradiated mice selected for this study (S2 Table in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}) are represented in a box and whisker plot of age at death versus total dose with colors indicating the number of fractions used. Gamma irradiated mice received 21.57 to 4901 cGy ([Fig 2A](#pone.0231510.g002){ref-type="fig"}), with the maximal dose for acute exposures limited to 546 cGy. Because the LD~50/30~ for B6CF1 mice at 110 days of age is approximately 7 Gy \[[@pone.0231510.ref045]\], this maximum acute dose ensured that animals did not die from acute radiation syndromes. [Fig 2A](#pone.0231510.g002){ref-type="fig"} shows that fractionation had a larger impact on age at death as total doses increased. According to the Janus documentation, \[[@pone.0231510.ref021]\] the age first irradiated for mice was intended to be 100 days +/- 15 days, with a small subset of mice acutely irradiated at 500 days in order to investigate how age first irradiated impacted survival. Plotting the frequency of each age first irradiated, we found that the majority of mice were first irradiated within the expected range and there was a small group of mice that were first irradiated over 500 days old ([Fig 2C](#pone.0231510.g002){ref-type="fig"}). [Fig 2B](#pone.0231510.g002){ref-type="fig"} represents age at death against total dose with dark purple bars representing the aged mice and the light purple bars representing mice irradiated closer to 100 days of age. Due to the low sample size for aged mice, the large amount of leverage they would have on the overall model, the absence of sham irradiated aged mice, and the lack of direct dose comparisons with younger mice, we excluded these 560 mice from further analysis. This resulted in 11,618 total mice for the analysis on gamma irradiated mice.
{ref-type="table"} that were controls or gamma irradiated.\
**(A)** Box plot of age at death in days versus total dose in Gy. Colors indicate the number of fractions. **(B)** Histogram of the total number of animals versus age first irradiated in days. **(C)** Age at death in days versus total dose in Gy. Colors indicate whether a mouse was first irradiated before or after 500 days**. (D)** Representative graphs from Cox PH model output with age at death as the time scale and sex (p \< 0.001), age first irradiated (p = 0.14), total dose (p \<0.001), fractions (p = 0.001), and the interaction between total dose and fractions (p \<0.001) as independent variables. The predicted outcomes shown are for female mice first irradiated at 120 days.](pone.0231510.g002){#pone.0231510.g002}
Fractionation increased the overall survival probability in mice exposed to gamma rays {#sec013}
--------------------------------------------------------------------------------------
To determine how fractionation impacts survival, we used a Cox PH model with age at death as the time scale and sex, age first irradiated, total dose, fractions, and the interaction between total dose and fractions as independent variables. All independent variables were significant in the model, except for age first irradiated ([Table 1](#pone.0231510.t001){ref-type="table"}). Given the small range of ages first irradiated included in our sample, this was an expected result (p = 0.14). The main effect of fractions resulted in a positive coefficient from our model output, which corresponds to an increase in hazard (p = 0.001). This result is the outcome of the difference in total doses of exposure for acute and fractionated radiation regimens. The maximum acute exposure was 5.4Gy, but the total doses for fractionated exposures reached as much as 49Gy, causing the interaction between total dose and the number of fractions to be the most relevant for determining the role of fractionation. The interaction term between fractions and total dose was highly significant and its interpretation is best understood by graphical representation of the model's predicted outcome ([Fig 2D](#pone.0231510.g002){ref-type="fig"}, p \<0.001). As the total dose increased, the beneficial effect of fractionation became more pronounced. This result was consistent throughout a series of robustness tests (S3 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). Notably, gamma irradiated mice that received their total doses in 300 fractions had a decrease in the death hazard compared to mice that received acute exposures, even with the added stress that caused control mice to die significantly earlier (S3I--S3L Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). When adding a new interaction term between sex and total dose, we found that males as the total dose increased, the decreased death hazard in males was more pronounced (S3M and S3N Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). We used KM curves to validate the proportional hazards assumption in our model and found parallel survival curves between groups based on sex, number of fractions, age first irradiated, and total dose (S4 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}).
10.1371/journal.pone.0231510.t001
###### Parameter estimates, hazard ratios with 95% confidence interval, and p-values for main Cox proportional hazards model in [Fig 2D](#pone.0231510.g002){ref-type="fig"}.
{#pone.0231510.t001g}
Variable Estimate Hazard Ratio (95% CI) P-value
---------------------- ---------- ----------------------- -------------
sexM -0.17 0.84 (0.80, 0.89) **\<0.001**
Fractions 0.002 1.00 (1.001, 1.003) **0.001**
Total dose 16.61 1.64E7 (8.2E6, 3.3E7) **\<0.001**
First irrad -0.002 0.998 (0.996 1.00) 0.137
Fractions:Total dose -0.109 0.89664 (0.88, 0.91) **\<0.001**
Fractionation significantly decreased the death hazard for mice dying from lymphomas, tumors, non-tumor, and causes of death unknown in gamma irradiated mice {#sec014}
-------------------------------------------------------------------------------------------------------------------------------------------------------------
When analyzing specific causes of death, it is important to consider the effects of competing risks. Cause specific hazards models are one type of competing risk model and their parameter estimates can be interpreted as the hazards for the specific event of interest. The cause specific hazards models for tumors, lymphomas, non-tumors, and CDUs all showed a similar trend ([Fig 3A--3D](#pone.0231510.g003){ref-type="fig"})---as the dose increased, there was an increased rescue effect from fractionation, and more fractions corresponded to less hazard. Estimated model parameters showed that the interaction term between total dose and number of fractions was significant for all four categories of causes of death ([Table 2](#pone.0231510.t002){ref-type="table"}, tumor p = 0.001, lymphoma p\<0.001, non-tumor p \<0.001, CDU p \<0.001). Sex was also significant in all models. Males had a higher death hazard for tumors, while females were at greater risk for all other causes of death, including lymphomas. Additionally, we examined the top two causes of death specifically---lung tumors and generalized non-thymic lymphomas (S5 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}). Males were only at a greater risk of lung tumor death, while females had a higher hazard ratio for tumors excluding lung tumors and generalized non-thymic lymphomas.
{ref-type="table"}. All four models have a significant interaction term between total dose and the number of fractions (tumor p = 0.001, lymphoma p\<0.001, non-tumor p \<0.001, CDU p \<0.001). The graphs represent predicted outcomes for female mice first irradiated at 120 days.](pone.0231510.g003){#pone.0231510.g003}
10.1371/journal.pone.0231510.t002
###### Competing risks model output for cause specific hazards and subdistribution hazards.
{#pone.0231510.t002g}
Cause Specific Hazards Subdistribution Hazards
-------------------------------- --------- ------------------------- ------------------------- ------------- ---------------------- ---------------------- -------------
Tumors sexM 0.195 1.21 (1.11, 1.33) **\<0.001** 0.453 1.573 (1.446, 1.711) **\<0.001**
Fractions -0.001 0.9994 (0.998, 1.001) 0.535 -0.003 0.997 (0.995, 0.999) **0.002**
Total dose 0.123 1.13 (1.11, 1.15) **\<0.001** -0.063 0.939 (0.93, 0.948) **\<0.001**
First irrad -0.0001 0.9992 (0.996, 1.003) 0.859 0 1 (1, 1) 0.71
Fractions: Total dose -0.001 0.9994 (0.9991, 0.9998) **0.001** 0.001 1.001 (1, 1.001) **\<0.001**
Lymphoma sexM -0.37 0.69 (0.64, 0.75) **\<0.001** -0.259 0.772 (0.72, 0.83) **\<0.001**
Fractions 0.005 1.005 (1.004, 1.007) **\<0.001** 0.004 1.004 (1.002, 1.005) **\<0.001**
Total dose 0.166 1.18 (1.17, 1.19) **\<0.001** 0.012 1.012 (1.004, 1.021) **0.007**
First irrad -0.003 0.997 (0.993, 1.0001) 0.058 0 1 (0.999, 1) **.014**
Fractions: Total dose -0.001 0.999 (0.998, 0.999) **\<0.001** 0 1 (1, 1) .12
Non-tumors sexM -0.50 0.61 (0.53, 0.69) **\<0.001** -0.265 0.767 (0.676, 0.87) **\<0.001**
Fractions -0.005 0.994 (0.991, 0.998) **0.001** -0.007 0.993 (0.99, 0.997) **\<0.001**
Total dose 0.191 1.21 (1.19, 1.23) **\<0.001** 0.07 1.073 (1.06, 1.086) **\<0.001**
First irrad -0.006 0.994, (0.988, 0.9999) **0.009** 0.001 1.001 (1.000, 1.001) **\<0.001**
Fractions: Total dose -0.001 0.999 (0.9990, 0.9995) **\<0.001** 0 1 (1, 1) .84
CDU sexM -0.14 0.87 (0.71, 1.07) 0.182 -0.046 0.955 (0.79, 1.16) 0.64
Fractions -0.001 0.999 (0.995, 1.003) 0.667 -0.002 0.998 (0.994, 1.002) 0.34
Total dose 0.154 1.166 (1.14, 1.20) **\<0.001** 0.002 1.002 (0.985, 1.02) 0.82
First irrad 0.001 1.001 (0.993, 1.01) 0.782 3.96E-04 1.00 (1.00, 1.00) 0.37
Fractions: Total dose -0.001 0.999 (0.999, 0.999) **0.011** 0 1 (1, 1.001) 0.54
Lung tumors sexM 0.89 2.44 (2.12, 2.79) **\<0.001** 1.19 3.30 (2.9, 3.8) **\<0.001**
Fractions -0 0.9999 (0.997, 1.002) 0.944 -2.90E-03 0.997 (0.995, 0.999) **0.0097**
Total dose 0.11 1.12 (1.09, 1.14) **\<0.001** -0.066 0.936 (0.925, 0.948) **\<0.001**
First irrad -0.004 0.996 (0.991, 1.001) 0.130 0.001 1.001 (1, 1.001) **.029**
Fractions: Total dose 0 0.9996 (0.999, 1.000) 0.079 0.001 1.001 (1, 1.001) **\<0.001**
Tumors (excluding lung tumors) sexM -0.469 0.63 (0.55, 0.71) **\<0.001** -0.38 0.684 (0.606, 0.773) **\<0.001**
Fractions -0.001 0.999 (0.996, 1.001) 0.348 -0.003 0.997 (0.995, 1) **0.021**
Total dose 0.136 1.15 (1.12, 1.17) **\<0.001** -0.057 0.944 (0.931, 0.958) **\<0.001**
First irrad 0.003 1.003 (0.998, 1.009) 0.222 -0.001 0.999 (0.998, 1) **9.9E-03**
Fractions: Total dose -0.001 0.999 (0.9988, 0.9997) **0.003** 0.001 1.001 (1, 1.001) **\<0.001**
Cumulative incidence rates in deaths from lymphomas, tumors, non-tumors, and causes of death unknown in gamma irradiated mice varied greatly for each COD based on total dose and fractionation status {#sec015}
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
[Fig 4A](#pone.0231510.g004){ref-type="fig"} shows the non-parametric cumulative incidence of death for each of the main causes of death. Lymphomas were the most prevalent COD, followed by lung tumors, tumors (excluding lung tumors), non-tumors, and CDU. When we divided the data into control mice ([Fig 4B](#pone.0231510.g004){ref-type="fig"}) and gamma irradiated mice groups ([Fig 4C](#pone.0231510.g004){ref-type="fig"}), the first instances of death were observed around 750 days in control mice and around 250 days in gamma irradiated animals. These graphs were also subdivided by gender. For control and gamma irradiated mice, males had a lower incidence of lymphoma, tumors (excluding lung tumors), and non-tumors COD, while females had a much lower incidence of lung tumors. The differences between males and females were significant for both controls and gamma irradiated mice for all causes of death, except CDU ([Table 2](#pone.0231510.t002){ref-type="table"}).
{ref-type="table"}**.](pone.0231510.g004){#pone.0231510.g004}
In addition to calculating cause specific hazards for competing risks, it is important to examine subdistribution hazards, also known as cumulative incidence functions, using the Fine and Grey method \[[@pone.0231510.ref032], [@pone.0231510.ref033], [@pone.0231510.ref041], [@pone.0231510.ref042]\]. The parameter estimates from subdistribution hazards have a less direct interpretation, but instead elucidate the overall probability of a particular outcome. We used the Fine and Grey method controlling for sex, age first irradiated, number of fractions, total dose, and the interaction between fractions and total dose, with competing risk groups as lymphoma, lung tumors, tumors (excluding lung tumors), non-tumors, and CDU. For tumors (excluding lung tumors) females were more susceptible but fractionation and increased dose both decreased tumor incidence ([Table 2](#pone.0231510.t002){ref-type="table"}). By graphing predicted outcomes under varying conditions, we discovered that when the difference between high and low total doses is small (10cGy vs 100cGy), fractionation is the biggest determinate for tumor incidence, with acute exposures resulting in the most tumors ([Fig 4D](#pone.0231510.g004){ref-type="fig"}). Conversely, when the high dose was increased to 1000cGy, total dose became the dominant factor and low dose conditions resulted in the most tumor incidences ([Fig 4E](#pone.0231510.g004){ref-type="fig"}). Under all conditions, low dose acute exposures resulted in the greatest tumor incidence. Fractionation and dose had the same impact on lung tumor incidence as in all other tumor incidence ([Fig 4F and 4G](#pone.0231510.g004){ref-type="fig"}). However, males were more likely than females to die of lung tumors specifically, which matches the cause specific hazards results ([Table 2](#pone.0231510.t002){ref-type="table"}).
Examining lymphoma deaths, females were at a greater risk of death than males and increasing the total dose and number of fractions both increased the risk of death ([Table 2](#pone.0231510.t002){ref-type="table"}). Predicted outcomes showed that with a 10-fold difference between high and low total doses, fractionation was the main determinate for lymphoma incidence and fractionated exposures resulted in the most lymphoma cases ([Fig 4H](#pone.0231510.g004){ref-type="fig"}). Predicted outcomes with a 100-fold difference between high and low total doses resulted in total dose becoming the dominant factor and high dose conditions produced the most lymphoma incidences ([Fig 4I](#pone.0231510.g004){ref-type="fig"}). For all conditions, high dose fractionated exposures resulted in the greatest lymphoma incidence, which is the exact opposite from the trend observed for tumor deaths.
Non-tumor deaths were more prevalent in female mice compared to male mice, total dose increased the probability of non-tumor deaths, and fractionation decreased the probability of a non-tumor death ([Table 2](#pone.0231510.t002){ref-type="table"}). When we examined the predicted outcome using a 10-fold difference between high and low total doses, fractionation had the greatest impact on lymphoma incidence, with acute exposures resulting in the most non-tumor cases ([Fig 4J](#pone.0231510.g004){ref-type="fig"}). We observed that dose became the dominant factor when we assessed a 100-fold difference between high and low total doses, and high dose conditions resulted in the most non-tumor deaths ([Fig 4K](#pone.0231510.g004){ref-type="fig"}). High dose acute exposures resulted in the greatest non-tumor incidence consistently for all conditions we analyzed.
Lymphoma and non-tumors both had higher incidence rates with high doses at earlier times than previously anticipated {#sec016}
--------------------------------------------------------------------------------------------------------------------
The CIF regression models for death by lymphomas and non-tumors both demonstrated a shoulder along the CIF curve around 250--500 days ([Fig 4H--4K](#pone.0231510.g004){ref-type="fig"}). When we filtered out total doses above 6Gy (S6 Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}), this shoulder disappeared for lymphoma deaths and non-tumor deaths. Because tumors (excluding lung tumors) and lung tumors ([Fig 4D--4G](#pone.0231510.g004){ref-type="fig"}) did not exhibit the same shoulder, these results act as an internal negative control. After filtering out mice exposed to total doses over 6Gy for tumors and lung tumors, the shape of the CIF curves did not change (S6B and S6D Fig in [S2 File](#pone.0231510.s002){ref-type="supplementary-material"}).
Janus datasets showed similar results to IES data in male versus female comparisons and dose response trends for causes of death {#sec017}
--------------------------------------------------------------------------------------------------------------------------------
Several large-scale chronic exposure studies were done at IES using both genders of B6C3F1 mice. Work by Tanaka and others \[[@pone.0231510.ref030]\] at IES was focused on specific causes of death in response to gamma irradiation and we compared these data with Janus data. As mentioned in the methods, the B6C3F1 mice strain used at IES was genetically similar to the B6C3F1 mice used in the Janus experiments. These F1 mice came from crosses of the same female strain C57BL/6J and two different strains of male mice: C3H/HeJ for IES vs. BALB/cJ for Janus experiments. When examining cause of death between groups ([Table 3](#pone.0231510.t003){ref-type="table"}), we found that hematopoietic system diseases are the most common cause of death for both sets of mice. Additionally, females died of hematopoietic diseases more than males in both datasets. The next most common cause of death for Janus mice was respiratory disease, a result driven by the high incidence of lung cancer in B6CF1 mice from Janus studies. B6C3F1 mice died much less frequently of respiratory disease compared to the B6CF1 mice. However, respiratory diseases were more common in male mice in both datasets. Finally, in male B6C3F1 mice, digestive diseases were much more frequent than in animals used in Janus experiments.
10.1371/journal.pone.0231510.t003
###### The percentage of deaths due to each individual cause of death listed for a comparison between B6CF1 Janus mice and B6C3F1 IES mice.
{#pone.0231510.t003g}
---------------------------- -------- --------------------- ---------------------- ---------------- ---------------- ----------------- --
Males **B6C3F1 IES mice** **B6CF1 Janus mice**
0Gy 0.4Gy 8Gy 0Gy 9.2Gy 9.6Gy
400 fractions 400 fractions Sham fractions 24 fractions 120 fractions
22h/day 22h/day 45min/fraction 45 min/fraction
1.1mGy/day 21mGy/day 0.85cGy/min 0.006 cGy/min
COD
Circulatory System 8.20% 9.20% 13.40% 5.70% 8.40% 6.90%
Digestive System 24.70% 27.80% 19.40% 3.40% 2.10% 0.00%
Endocrine System 0.60% 0.40% 0.20% 0.30% 0.50% 0.00%
Hematopoietic System 40.60% 40.40% 41.50% 38.80% 29.50% 40.30%
Male Reproductive System 0.00% 0.40% 0.00% 0.40% 1.10% 0.00%
Nervous System 0.40% 0.00% 0.20% 0.10% 0.00% 0.00%
Nonneoplastic 11.40% 8.80% 9.20% 9.10% 22.60% 16.70%
Respiratory System 7.20% 6.40% 6.80% 34.10% 21.60% 26.40%
Skeletal System 0.60% 0.60% 0.00% 0.10% 0.50% 0.00%
Skin 0.60% 0.00% 0.40% 0.00% 0.00% 0.00%
Soft Tissue 5.00% 4.20% 5.80% 1.10% 1.60% 0.00%
Special Sense Organs 0.20% 0.40% 1.20% 0.40% 0.00% 0.00%
Unknown 0.00% 0.60% 1.00% 6.20% 6.80% 6.90%
Urinary 0.00% 0.40% 0.40% 0.20% 2.60% 2.80%
Mesothelium/Other Tumor 0.40% 0.40% 0.40% 0.10% 0.00% 0.00%
Females COD
Circulatory System 3.20% 3.60% 5.60% 4.10% 5.10% 7.30%
Digestive System 2.80% 2.40% 3.80% 1.40% 2.20% 2.00%
Endocrine sytem 5.00% 4.20% 2.00% 1.10% 2.20% 1.30%
Female Reproductive System 4.00% 2.80% 6.40% 4.90% 6.40% 7.90%
Hematopoietic System 63.60% 59.20% 57.80% 49.80% 39.80% 31.80%
Nervous System 0.00% 0.20% 0.20% 0.00% 0.30% 0.30%
Nonneoplastic 9.20% 12.50% 7.60% 14.30% 17.80% 20.20%
Respiratory System 1.20% 1.60% 3.00% 12.70% 10.80% 12.30%
Skeletal System 0.00% 1.40% 1.40% 0.50% 1.00% 0.30%
Skin 0.00% 0.20% 0.20% 0.00% 0.00% 0.00%
Soft Tissue 8.60% 10.30% 9.20% 4.10% 4.50% 3.30%
Special Sense Organs 1.00% 0.80% 1.40% 0.20% 0.30% 1.30%
Unknown 1.40% 0.80% 1.20% 6.20% 9.20% 9.60%
Urinary 0.00% 0.00% 0.20% 0.40% 0.30% 2.00%
Mesothelium/Other Tumor 0.00% 0.00% 0.00% 0.20% 0.00% 0.30%
---------------------------- -------- --------------------- ---------------------- ---------------- ---------------- ----------------- --
Discussion {#sec018}
==========
Janus experiments were analyzed in many different ways over the years \[[@pone.0231510.ref014], [@pone.0231510.ref017], [@pone.0231510.ref020], [@pone.0231510.ref023], [@pone.0231510.ref024], [@pone.0231510.ref026]--[@pone.0231510.ref028], [@pone.0231510.ref043], [@pone.0231510.ref046]--[@pone.0231510.ref048]\] and each new approach for analysis of these data brought novel information about the effects of dose fractionation. Common to all these studies is the fact that they either considered each Janus experiment individually, or combined all of them into a single dataset. This is the first study where individual Janus experiments were combined based on control animal datasets compatibility. By analyzing control mice in a way that allowed us to pool Janus experiments together, we gained statistical power to run tests on the importance of fractionation for specific causes of death. The Janus experiments were originally designed with this in mind and taking advantage of the consistency between experiments for a large-scale study was extremely effective. We were able to determine under which circumstances fractionation had a rescuing effect and track changes in risk based on gender for specific causes of death. Our method for pooling data together can be used for future analysis on the Janus dataset using different modeling techniques and answering novel biological questions. Moreover, it is conceivable that a similar approach could be applied to other types of datasets. For example, one can imagine a scenario where animal studies conducted in different laboratories where control animals have similar distribution of cause of death diseases could be combined for a complex combined evaluation of different test conditions.
One of the most interesting findings from the control mice analysis was that mice sham irradiated with 300 fractions died significantly earlier than animals exposed to fewer fractions during their sham irradiations. Mice given 300 sham irradiation fractions died earlier due to tumors and CDU. The simplest and most likely explanation for this phenomenon is general stress caused by frequent exposure to unfamiliar circumstances. Sham exposures involved transporting the mice from the room where they were housed to the room with the irradiator. The irradiator was turned off and there was no excess radiation in the room. It is known that transporting mice induces a stress response \[[@pone.0231510.ref049], [@pone.0231510.ref050]\]. The mice that received 300 sham fractions also had an increase in CDU incidences compared to mice that received fewer fractions. It is possible that the observed decrease in lung tumor and non-tumor deaths was due to misclassification of those deaths as CDU. Further investigation of this mouse cohort may provide us with new insights into the stress response. Analyzing available tissue samples from these mice could enable us to explore cytological or molecular indicators of stress. These results would not only be beneficial for animal studies in radiation biology, but for any investigators utilizing animals in their research.
In all comparisons between acute and fractionated exposures, fractionation significantly decreased the death hazard in gamma irradiated mice. Moreover, fractionation was equally protective for all four pooled categories of diseases (lymphoma, tumors, non-tumors, and CDU) and specific diseases such as non-thymic lymphoma. Lung tumors were the only causes of death that were not significantly affected by fractionation. B6CF1 males were at a significantly higher risk for lung tumors than females. Interestingly, when we excluded lung tumors and examined all other types of tumors, males were at a lower risk than females for all other causes of death. Heidenreich et al. investigated lung tumors from the Janus datasets using Kaplan-Meier plots and the two-step clonal expansion (TSCE) model \[[@pone.0231510.ref025]\]. They concluded that males had a higher lung cancer risk than females in control and gamma irradiated mice. They also found that more fractions administered over a longer duration resulted in less lung tumor risk, again agreeing with the results we found using Cox PH.
Examining CIFs to determine the probabilities for death due to distinct diseases under varying conditions produced many intriguing results. Increasing the number of fractions and increasing the total dose both decreased the incidences of tumors and lung tumors specifically. This finding is not surprising because tumors in mice develop more slowly than lymphomas and even more slowly than non-tumors such as radiation induced pneumonitis. Therefore, a mouse exposed to a high dose of gamma rays would most likely die of non-cancer cause of death before a tumor has time to fully advance.
Mice exposed to fractionated irradiation died from lymphoma more frequently when compared to mice that received acute exposures, while mice exposed to acute exposures developed more non-tumors. Non-tumors were the most common cause of death in response to higher doses and acute exposures. Considering that non-irradiated B6CF1 mice begin to develop lymphomatous spleens by 600 to 700 days of age and that even few spontaneous lymphoma cells have the immunosuppressive effect in spleen \[[@pone.0231510.ref051]\], it is possible that other causes of death may also be partially dependent on pre-symptomatic lymphoma development. Overall, B6CF1 mice are a robust hybrid mouse strain, immunocompetent and long lived (853 ± 10 days on average \[[@pone.0231510.ref051]\]), and almost as radiation resistant as its more radiation resistant parent strain C57/BL mice (LD~50/30~ of about 6.6 Gy for mice exposed at 120 days of age) \[[@pone.0231510.ref045]\].
Lymphomas are easily induced in response to ionizing radiation in rodents. It is typically considered a risk associated with lower doses of ionizing radiation compared to non-tumors. However, our results showed an early increase in lymphoma incidences when the dose delivered to animals was above 6 Gy. A 6Gy cutoff was chosen because all of the animals exposed to doses above 6 Gy received fractionated irradiation. This was done because the LD~50/30~ dose for B6CF1 mice is 6.54 or 6.75 Gy for males and females respectively \[[@pone.0231510.ref025]\] and acute exposures above 6Gy would result in an animal death rate incompatible with robust experimental data. In mice that received total doses over 6Gy (all fractionated exposures), lymphoma deaths began as early as 300 days. The early death shoulder observed in CIF curves for lymphomas is no longer present when excluding data for total doses above 6Gy. Non-tumor deaths demonstrated the same shoulder when the full range of total doses were included to fit the model. Again, the shoulder disappears after removing the data for total doses over 6Gy.
We compared the Janus dataset on fractionated radiation with the IES datasets on chronic irradiation. This comparison between B6CF1 mice used in Janus experiments and B6C3F1 mice used by Tanaka and others \[[@pone.0231510.ref030]\] showed a high degree of similarity despite different radiation delivery approaches and genetic differences between the two strains. The most pronounced differences between B6CF1 and B6C3F1 mice were associated with male mice, which could be attributed to these two hybrid strains differing by the paternally contributing mice. While respiratory system complications affected both F1 mouse hybrids discussed here, not all mice have the same association between gender and radiation associated respiratory diseases. For example, RFM mice exposed to x-rays males were at less risk than females for lung tumors, indicating that differential gender susceptibilities to lung tumors are strain specific \[[@pone.0231510.ref052]--[@pone.0231510.ref054]\]. IES results also showed more death due to digestive diseases, with the effect being most obvious in male mice. Janus mice were not kept in a sterile environment nor fed sterile food. The bacteria present in the guts of Janus mice would have increased the local immune response and may explain the lower percentage of deaths due to digestive diseases. These noticeable differences in digestive and respiratory disease proportions could be explained by several differences in housing. Not only were IES mice specific pathogen free, while Janus mice were not, but it is also likely that standard housing conditions changed between 1972 and 2004. Beyond standard conditions over time, conditions are likely variable across universities and countries, as well. In humans, females have been shown to be at a greater risk of lung cancer than males \[[@pone.0231510.ref055], [@pone.0231510.ref056]\]. Determining the cause for changes in lung tumor sensitivity in response to ionizing radiation between male and female mice could lead to a better understanding of the radiation induction of lung tumors. Given our current, limited amount of information, RFM mice appear to be a better model system for simulating gender differences of humans for lung tumor risk. While radiation doses associated with the LD~50/30~, for example, vary significantly between rodents and humans \[[@pone.0231510.ref006], [@pone.0231510.ref045]\], most interspecies comparisons focus on proportional life shortening \[[@pone.0231510.ref057]\].
In conclusion, we propose to continue to evaluate the NURA database using different models and applying them to different subsets of data in order to outline the finer nuances of consequences of radiation exposures. The differences we described in radiation exposure outcomes that change with alterations in radiation delivery highlight that biological responses to whole body irradiation most likely cannot be described by a single factor that could be applied for the entire spectrum of possible fractionation scenarios.
Supporting information {#sec019}
======================
######
(DOCX)
######
Click here for additional data file.
######
(PDF)
######
Click here for additional data file.
The authors would like to thank Edward Malthouse for his input on the statistical methods used and Carissa Ritner and Benjamin Haley for their constant support and thoughtful discussions.
10.1371/journal.pone.0231510.r001
Decision Letter 0
Li
Jian Jian
Academic Editor
© 2020 Jian Jian Li
2020
Jian Jian Li
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
4 May 2020
PONE-D-20-07144
Analyses of cancer incidence and other morbidities in gamma irradiated B6CF1 mice
PLOS ONE
Dear Dr. Woloschak,
Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE's publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.
We would appreciate receiving your revised manuscript by Jun 18 2020 11:59PM. When you are ready to submit your revision, log on to <https://www.editorialmanager.com/pone/> and select the \'Submissions Needing Revision\' folder to locate your manuscript file.
If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.
To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: <http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols>
Please include the following items when submitting your revised manuscript:
A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled \'Response to Reviewers\'.A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled \'Revised Manuscript with Track Changes\'.An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled \'Manuscript\'.
Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.
We look forward to receiving your revised manuscript.
Kind regards,
Jian Jian Li, M.D., Ph.D.
Academic Editor
PLOS ONE
Journal Requirements:
When submitting your revision, we need you to address these additional requirements.
1\. Please ensure that your manuscript meets PLOS ONE\'s style requirements, including those for file naming. The PLOS ONE style templates can be found at
<http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf> and <http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf>
2\. Thank you for stating the following in the Competing Interests Section of the submission form:\"The authors have declared that no competing interests exist.\"
Please update your Competing Interests Statement to confirm that author Gayle E. Woloschak is a section editor for PLOS ONE.
Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: \"This does not alter our adherence to PLOS ONE policies on sharing data and materials." (as detailed online in our guide for authors <http://journals.plos.org/plosone/s/competing-interests>). If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.
Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.
Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: <http://journals.plos.org/plosone/s/competing-interests>
Additional Editor Comments (if provided):
This work contains some important information on radiation-associated carcinogenic risk. Due to the huge number of animals involved, it is suggested that to recheck the statistical approaches used for validating the data in this study.
\[Note: HTML markup is below. Please do not edit.\]
Reviewers\' comments:
Reviewer\'s Responses to Questions
**Comments to the Author**
1\. Is the manuscript technically sound, and do the data support the conclusions?
The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
2\. Has the statistical analysis been performed appropriately and rigorously?
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
3\. Have the authors made all data underlying the findings in their manuscript fully available?
The [PLOS Data policy](http://www.plosone.org/static/policies.action#sharing) requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data---e.g. participant privacy or use of data from a third party---those must be specified.
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
4\. Is the manuscript presented in an intelligible fashion and written in standard English?
PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.
Reviewer \#1: Yes
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
5\. Review Comments to the Author
Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)
Reviewer \#1: This article studied the risks associated with ionizing radiation by analyzing the Northwestern University Radiation Archive from a series of 10 individual neutron and gamma irradiation experiments conducted on over 50,000 mice. They used rigorous statistical testing on control mice from all Janus experiments to select studies that could be compared to differences among the controls as well as experimental animals. Experiments are conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions are drawn appropriately based on the data presented. . The manuscript was presented in an intelligible fashion and written in standard English.
There are some comments:
1.Janus experiments were analyzed in many different ways over the years, what are the difference of these methods, and what is the meaning of the new approach for analysis used in this article?
2.The purpose and of this article should be stated in the discussion.
3.How radiation doses are converted between humans and mice?
Reviewer \#2: This paper employs the NURA database of mice receiving irradiation treatment to analyze the impact of irradiation dose, fractions, sex on the survival and incidence of death due to different causes, including lymphoma, tumors except lung cancer, lung cancer and CDU, which is designed and conducted logically and seriously. It shows some information for researchers and radiation oncologists, especially the difference of fraction and total dose in the influcence of tumor incidence.
1\. Mice sham irradiated with 300 fractions die significantly ealier than those mice with fewer fractions. Does the truly irradiated group have a similar finding?
2\. The genetic and immune background of B6CF1 mice should be discussed in the discussion.
3\. Fig1,2 and 3, the statistic significance should be caculated and labeled both in the results part and figures.
4\. In Fig 1, It seems there was a sinificant difference between two groups in tumors(excluding lung), CDU and lymphoma. And possibilities that CDU may affect the death caused by tumors should be discussed.
5\. Table 1 legend, Parameter estimates, hazard ratios with 95% confidence interval, and p-values for main
Cox Proportional Hazards model in Fig 1D. It should be Fig1D ? or Fig 2D?
6\. since the sex difference is shared in common by control and irradiation, the sex role in death affected by irradiation should be adjusted.
7\. In Line 392, the B6C3F1 mice strain used at IES was genetically similar to the B6C3F1 mice used in the Janus experiments. It should be B6CF1 or B6C3F1 in the Janus?
8\. The indicated clinical impact and potential use of the findings in this study are recommended to be discussed.
9\. Its not clear whether neutron irradiation treatment is excluded or not.
10\. all figures in the integrated muanscript is too vague to see. The orignal figures in tiff format are good.
\*\*\*\*\*\*\*\*\*\*
6\. PLOS authors have the option to publish the peer review history of their article ([what does this mean?](https://journals.plos.org/plosone/s/editorial-and-peer-review-process#loc-peer-review-history)). If published, this will include your full peer review and any attached files.
If you choose "no", your identity will remain anonymous but your review may still be made public.
**Do you want your identity to be public for this peer review?** For information about this choice, including consent withdrawal, please see our [Privacy Policy](https://www.plos.org/privacy-policy).
Reviewer \#1: No
Reviewer \#2: No
\[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link \"View Attachments\". If this link does not appear, there are no attachment files to be viewed.\]
While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, <https://pacev2.apexcovantage.com/>. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at <figures@plos.org>. Please note that Supporting Information files do not need this step.
10.1371/journal.pone.0231510.r002
Author response to Decision Letter 0
17 Jun 2020
Dear Reviews,
Thank you for taking the time to critically read through our manuscript and give helpful feedback. We responded your comments/suggestions to the best of our ability and details can be found below; we indicated added text by underlining it:
Reviewer 1:
1\. Janus experiments were analyzed in many different ways over the years, what are the difference of these methods, and what is the meaning of the new approach for analysis used in this article?
Recently, we have used Janus archive to re-evaluate dose and dose rate effectiveness factor for life shortening modulation due to fractionation (Haley et al, 2015) and in this work we recreated formalism developed in BEIR VII as R script that could be shared and verified by the community. Having done so, we felt that it would be equally valuable to use R scripts developed for clinical studies for a re-evaluation of animal specific causes of death, both to explore the new approach to analysis (use and exchange of R scripts in github is a recent development) and to evaluate how suitable these tools would be if we wished to look into disease-specific DDREF evaluation. While we continue to work in this direction, this manuscript covers a segment of our effort that has to do with optimization of combining different experiments for joint study. We have now tried to explain this better:
1\) in the abstract: "This study systematically cross-compared outcomes of different modes of fractionation evaluated across different Janus experiments and a wide span of total doses."
2\) in the introduction: "Numerous studies used the NURA (also known as Janus) database. In most cases, different Janus experiments were used separately (17, 20, 23-28) or else combined all together into a single dataset (14). In this study, however, many but not all Janus experiments were combined into a dataset -- process of selection was based on comparability of control animal datasets from sham irradiation conditions in different Janus experiments."
3\) in the methods: "...; different species such as Peromyscus leucopus (white-footed deer mouse) were excluded from this study because of the species to species differences between the controls and in response to radiation"
and later, in the methods subsection discussing IES vs. Janus data:
"The differences in disease incidence between the control animals point out that only some of these disease "endpoints" are appropriate for direct comparisons between strains when different "test conditions" are being evaluated. "
4\) in the discussion: "Common to all these studies is the fact that they either considered each Janus experiment individually, or combined all of them into a single dataset. This is the first study where individual Janus experiments were combined based on control animal datasets compatibility...... Moreover, it is conceivable that a similar approach could be applied for other types of datasets. For example, one can imagine a scenario where animal studies conducted in different laboratories where control animals have similar distribution of cause of death diseases could be combined for a complex combined evaluation of different test conditions."
2\. The purpose and of this article should be stated in the discussion.
Please see the response to question 1 from reviewer 1.
3\. How radiation doses are converted between humans and mice?
We have now added the following to the discussion:
"In addition, it should be noted that radiation doses associated with LD50/30, for example, very significantly in rodents and humans (6, 45), and most interspecies comparisons focus on proportional life shortening (56)."
Reviewer 2:
1\. Mice sham irradiated with 300 fractions die significantly ealier than those mice with fewer fractions. Does the truly irradiated group have a similar finding?
Mice that received radiation in 300 fractions showed a decrease in death hazard compared to acutely exposed mice. We excluded them from the main study because the sham 300-fractions controls died significantly earlier indicating that the data on mice with 300 fractions could no longer be pooled together. To make this a bit clearer we have included:
1\) in the abstract: "For controls, mice sham irradiated with 300 fractions died significantly earlier than those with fewer sham fractions and were excluded from pooled control dataset."
2\) in the results: Supp Fig 3 I and J show the main cox PH model including the mice irradiated in 300 fractions with fractions treated as a continuous variable. Supp Fig 3 K and L show the main cox PH model including the mice irradiated 300 fractions treated as a categorical variable. For both models, the interaction term shows that mice have a lower death hazard if they receive their total dose in 300 fractions compared to acute exposures. The specific text is: "Notably, gamma irradiated mice that received their total doses in 300 fractions had a decrease in the death hazard compared to mice that received acute exposures, even with the added stress that caused control mice to die significantly earlier (S3 Fig I-L)."
3\) in the discussion: "The mice that received 300 sham fractions also had an increase in CDU incidences compared to mice that received fewer fractions. It is possible that the observed decrease in lung tumor and non-tumor deaths was due to misclassification of those deaths as CDU."
2\. The genetic and immune background of B6CF1 mice should be discussed in the discussion.
Thank you for this suggestion. We added more to the discussion: "Considering that non-irradiated B6CF1 mice begin to develop lymphomatous spleens by 600 to 700 days of age and that even few spontaneous lymphoma cells have the immunosuppressive effect in spleen (51), it is possible that other causes of death may also be partially dependent on pre-symptomatic lymphoma development. Overall, B6CF1 mice are a robust hybrid mouse strain, immunocompetent and long lived (853 ± 10 days on average (51)), and almost as radiation resistant as its more radiation resistant parent strain C57/BL mice (LD50/30 of about 6.6 Gy for mice exposed at 120 days of age) (45)."
3\. Fig1,2 and 3, the statistic significance should be caculated and labeled both in the results part and figures.
Figure 1 -- the statistical significance in Fig 1 A was previously listed in Supp Table 6, but we moved it to be Fig 1B, next to the figure 1A and added the p-value to the figure legend and results section. We added statistical significance for the rest of the figures to the results section as well.
Figure 2 -- the best statistical test for these values come from the coxph models. We added the significance output from the cox ph model to the text in the results section and figure legend.
Figure 3 -- we added the p-values for the interaction term between fractions and total dose to the text in the results section and the figure legend.
4\. In Fig 1, It seems there was a sinificant difference between two groups in tumors(excluding lung), CDU and lymphoma. And possibilities that CDU may affect the death caused by tumors should be discussed.
This is an excellent point and we added text to the discussion: "The mice that received 300 sham fractions also had an increase in CDU incidences compared to mice that received fewer fractions. It is possible that the observed decrease in lung tumor and non-tumor deaths was due to misclassification of those deaths as CDU." (also please see reviewer2:1)
5\. Table 1 legend, Parameter estimates, hazard ratios with 95% confidence interval, and p-values for main Cox Proportional Hazards model in Fig 1D. It should be Fig1D ? or Fig 2D?
Thank you for noticing this. We updated the table to say Fig 2D.
6\. since the sex difference is shared in common by control and irradiation, the sex role in death affected by irradiation should be adjusted.
Thank you for pointing out that because baseline differences exist between sexes, we cannot tell which group is more sensitive to radiation treatment. We added an analysis that includes the interaction term between sex and total dose in an effort to see how each gender responds to increased ionizing radiation exposure. We found that the baseline differences become significantly more dramatic as total dose increases. These new results (graph and table with model output) are in supplemental figure 3 M-N with other robustness tests.
We have added text to results:
"When adding a new interaction term between sex and total dose, we found that as the total dose increased, the decreased death hazard in males was more pronounced (S3 Fig M-N)."
7\. In Line 392, the B6C3F1 mice strain used at IES was genetically similar to the B6C3F1 mice used in the Janus experiments. It should be B6CF1 or B6C3F1 in the Janus?
The mice used in the Janus experiments were B6CF1 mice, while the mice from IES were B6C3F1 mice. We have modified and/or added text to clarify:
1\) in methods: "Studies at IES involved chronic low dose rate gamma irradiations of specific-pathogen free (SPF) B6C3F1 mice, F1 progeny of C57BL/6J females (B6) and C3H/HeJ males. The B6CF1 mice, F1 progeny of C57BL/6J females (B6) and BALB/cJ males, were used during the Janus experiments. Both strains are F1 hybrids that share the same maternal strain C57BL/6."
2\) in results: "These F1 mice came from crosses of the same female strain C57BL/6J and two different strains of male mice: C3H/HeJ for IES vs. BALB/cJ for Janus experiments."
8\. The indicated clinical impact and potential use of the findings in this study are recommended to be discussed.
In response to reviewer1:1 & 2 and this comment, we have added the following sentences to the discussion:
"Common to all these studies is the fact that they either considered each Janus experiment individually, or combined all of them into a single dataset. This is the first study where individual Janus experiments were combined based on control animal datasets compatibility...... Moreover, it is conceivable that a similar approach could be applied for other types of datasets. For example, one can imagine a scenario where animal studies conducted in different laboratories where control animals have similar distribution of cause of death diseases could be combined for a complex combined evaluation of different test conditions."
9\. Its not clear whether neutron irradiation treatment is excluded or not.
Neutron irradiated mice were not included in this analysis, but are instead included in another publication that is under review at PLOS ONE. To make this more obvious to readers we added new text in the introduction: "We examined whether fractionation, age at which a mouse was first irradiated, and gender modulated the overall death hazard and frequency for specific causes of death in gamma irradiated mice."
in the methods section: "For this analysis, we focused on gamma irradiated mice. Neutron irradiated mice were studied in a separate analysis."
10\. all figures in the integrated muanscript is too vague to see. The orignal figures in tiff format are good.
We anticipate that the journal will handle this issue or help us if there is need for more work from our end. Thank you for drawing this to everyone's attention.
######
Submitted filename: Response to Reviewers.docx
######
Click here for additional data file.
10.1371/journal.pone.0231510.r003
Decision Letter 1
Amendola
Roberto
Academic Editor
© 2020 Roberto Amendola
2020
Roberto Amendola
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
22 Jul 2020
Analyses of cancer incidence and other morbidities in gamma irradiated B6CF1 mice
PONE-D-20-07144R1
Dear Dr. Gayle E. Woloschak,
We're pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.
Within one week, you'll receive an e-mail detailing the required amendments. When these have been addressed, you'll receive a formal acceptance letter and your manuscript will be scheduled for publication.
An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at <http://www.editorialmanager.com/pone/>, click the \'Update My Information\' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at <authorbilling@plos.org>.
If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they'll be preparing press materials, please inform our press team as soon as possible \-- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact <onepress@plos.org>.
Kind regards,
Roberto Amendola, Ph.D
Academic Editor
PLOS ONE
Additional Editor Comments (optional):
Reviewers\' comments:
Reviewer\'s Responses to Questions
**Comments to the Author**
1\. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the "Comments to the Author" section, enter your conflict of interest statement in the "Confidential to Editor" section, and submit your \"Accept\" recommendation.
Reviewer \#2: All comments have been addressed
\*\*\*\*\*\*\*\*\*\*
2\. Is the manuscript technically sound, and do the data support the conclusions?
The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
3\. Has the statistical analysis been performed appropriately and rigorously?
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
4\. Have the authors made all data underlying the findings in their manuscript fully available?
The [PLOS Data policy](http://www.plosone.org/static/policies.action#sharing) requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data---e.g. participant privacy or use of data from a third party---those must be specified.
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
5\. Is the manuscript presented in an intelligible fashion and written in standard English?
PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.
Reviewer \#2: Yes
\*\*\*\*\*\*\*\*\*\*
6\. Review Comments to the Author
Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)
Reviewer \#2: The authors clearly answered the questions and supporting experiments were supplemented. I believe this paper is ready to be published.
\*\*\*\*\*\*\*\*\*\*
7\. PLOS authors have the option to publish the peer review history of their article ([what does this mean?](https://journals.plos.org/plosone/s/editorial-and-peer-review-process#loc-peer-review-history)). If published, this will include your full peer review and any attached files.
If you choose "no", your identity will remain anonymous but your review may still be made public.
**Do you want your identity to be public for this peer review?** For information about this choice, including consent withdrawal, please see our [Privacy Policy](https://www.plos.org/privacy-policy).
Reviewer \#2: No
10.1371/journal.pone.0231510.r004
Acceptance letter
Amendola
Roberto
Academic Editor
© 2020 Roberto Amendola
2020
Roberto Amendola
This is an open access article distributed under the terms of the
Creative Commons Attribution License
, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
30 Jul 2020
PONE-D-20-07144R1
Analyses of cancer incidence and other morbidities in gamma irradiated B6CF1 mice
Dear Dr. Woloschak:
I\'m pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.
If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they\'ll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact <onepress@plos.org>.
If we can help with anything else, please email us at <plosone@plos.org>.
Thank you for submitting your work to PLOS ONE and supporting open access.
Kind regards,
PLOS ONE Editorial Office Staff
on behalf of
Dr. Roberto Amendola
Academic Editor
PLOS ONE
[^1]: **Competing Interests:**The corresponding author of this paper, Gayle E. Woloschak, is a section editor for PLOS ONE. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#S1}
============
Klotho is an anti-ageing protein predominantly produced in the kidney and several other tissues including parathyroid glands and epithelial cells of the choroids plexus^[@R1]^. Mice homozygous for a hypomorphic *Klotho* allele (*kl*/*kl*) manifest multiple ageing-related phenotypes including skin and muscle atrophy, hyperphosphatemia, osteoporosis, and vascular calcification, and die prematurely at around 2--3 months of age. The full-length Klotho protein is a type-1 membrane protein with a large extracellular domain of 952 amino acids in human, a membrane-spanning segment, and a short 11 amino acids intracellular carboxyl terminus^[@R1]^. Membranous Klotho associates with fibroblast growth factor (FGF) receptors to form co-receptors for the ligand FGF23, a bone-derived circulating hormone that lowers serum phosphate levels by increasing renal phosphate excretion, suppressing 1,25-dihyroxyvitamin D synthesis, and decreasing gastrointestinal phosphate absorption^[@R2]--[@R5]^. Klotho-deficient mice have severe hyperphosphatemia due to defects in the Klotho-FGF23-vitamin D regulatory axis^[@R5]--[@R7]^. This phosphate retention is pivotal for growth retardation and premature death of Klotho-deficient mice. Dietary phosphate restriction rescues growth defects and premature death of the mice^[@R5]--[@R7]^. The notion that FGF receptor and Klotho form obligatory coreceptors for FGF23 is supported by the demonstration that systemic injection of bioactive FGF23 decreases serum levels of phosphate in wild type mice, but not in Klotho-deficient mice^[@R8]^.
The extracellular domain of Klotho is composed of two internal repeats, KL1 and KL2, each sharing amino acid sequence homology to family 1 glycosidases^[@R1]^. The extracellular domain of Klotho is shed into the systemic circulation, urine and cerebrospinal fluid^[@R9]^. In urine, soluble Klotho regulates several ion transporters in the apical membrane of kidney tubules^[@R10]--[@R12]^. The physiological function of soluble Klotho present in the systemic circulation is mostly unknown.
The heart responds to injury and stress signals by pathological growth and remodeling that often progresses to heart failure and sudden death^[@R13]^. One key regulatory step in the development of pathological cardiac growth and remodeling is activation of calmodulin-dependent serine-threonine protein phosphatase calcineurin by abnormal calcium signaling^[@R14]^. Once activated by increases in intracellular calcium, calcineurin dephosphorylates and causes nuclear translocation of nuclear factor of activated T cells (NFAT) transcription factors, which bind the regulator regions of cardiac genes and in conjunction with other transcription factors induce gene expression and promote hypertrophic growth and remodeling.
Extracellular stimuli increase intracellular Ca^2+^ levels by either promoting its release from intracellular organelles or its entry across the plasma membrane. The TRPC family channels are Ca^2+^-permeable cation channels expressed in the plasma membrane of many tissues including the heart^[@R15]^. The TRPC family includes 7 members, and is divided into two groups based on structural and functional similarities: TRPC1/4/5, which are not sensitive to diacylglycerol (DAG), and TRPC3/6/7, which are activated by DAG. TRPC2 is not expressed in humans. Evidence indicates that Ca^2+^ influx through cardiac TRPC channels- including TRPC1, 3, 4, 5, and 6 - is important in calcineurin signaling and hypertrophic growth of hearts^[@R16]--[@R22]^. The expression of TRPC1, 3, 4, 5 and/or 6 is increased in hypertrophic hearts stimulated by various types or forms of stresses and their downregulation protects against cardiac hypertrophy. Some members of TRPC family channels, such as TRPC6, contain NFAT-responsive elements in their promoters which play a pivotal role in amplifying and sustaining gene expression through a feed-forward circuit^[@R16]^. Thus, TRPC6 is an important modulator of cardiac hypertrophy and a potential target for treatment. However, physiological function of TRPC6 in the heart and its regulation remain poorly understood, limiting therapeutic strategies for targeting the pathway. Here, we show that soluble Klotho inhibits cardiac TRPC6 channels and protects the heart against stress-induced pathological hypertrophy and remodeling.
Results {#S2}
=======
Klotho deficiency aggravates pathological heart growth {#S3}
------------------------------------------------------
Klotho expression is decreased in aging, a condition associated with increased risk for heart failure^[@R23],[@R24]^. We examined the role of Klotho in protecting the heart using *Klotho*-hypomorphic mice rescued by dietary phosphate restriction. To avoid potential variations caused by strain and gender differences, we studied male mice congenic for the 129/SvJ background by backcrossing for \>6 generations. As reported previously^[@R5]--[@R7]^, dietary phosphate restriction lowered serum phosphate levels and rescued growth defects and premature death of Klotho-deficient mice ([Supplementary Fig. S1, S2a](#SD1){ref-type="supplementary-material"}). Serum levels of sodium, potassium, chloride, calcium, magnesium, and urea nitrogen were not different between WT and *kl*/*kl* mice on a phosphate-restricted diet ([Supplementary Table S1](#SD1){ref-type="supplementary-material"}). Phosphate restriction did not affect the growth of wild-type mice ([Supplementary Fig. S1](#SD1){ref-type="supplementary-material"}). *Klotho*-hypomorphic mice on low phosphate diet remained markedly Klotho-deficient ([Supplementary Fig. S2b](#SD1){ref-type="supplementary-material"}).
To investigate the potential cardioprotective effect of Klotho, we measured heart weight indices (heart weight normalized to body weight or tibia length) as well as the overall heart size in wild-type and Klotho-deficient mice. Heart weight indices ([Fig. 1a, b](#F1){ref-type="fig"}) and the overall heart size measured using magnetic resonance imaging ([Fig. 1c](#F1){ref-type="fig"}) were not different between wild-type and Klotho-deficient mice at baseline. Overstimulation by isoproterenol (ISO) induced pathological hypertrophy in wild-type (WT) mice as reflected by increases in heart weight indices and the overall heart size, and these ISO-induced changes were aggravated in Klotho-deficient mice ([Fig. 1, a--c](#F1){ref-type="fig"}). ISO overstimulation is a well accepted experimental model of stress-induced cardiac hypertrophy^[@R25],[@R26]^. Phosphate restriction itself did not alter cardiac responses to stress, as baseline and ISO-induced increases in heart mass were not different between wild-type mice fed normal and phosphate-restricted diets ([Supplementary Fig. S2c](#SD1){ref-type="supplementary-material"}).
Pathological cardiac hypertrophy and remodeling are also characterized by increased (re)expression of fetal genes that are normally quiescent in adult hearts, including brain natriuretic peptide (*BNP*), atrial natriuretic peptide (*ANP*), and β-myosin heavy chain (β-*MHC*)^[@R13],[@R14]^. Consistent with the notion that Klotho-deficiency accelerates ISO-induced pathological cardiac remodeling, the expression of cardiac fetal genes was increased by ISO in wild-type mice, and such increase in gene expression was augmented in Klotho-deficient mice ([Fig.1, d--f](#F1){ref-type="fig"}). Increased expression of these cardiac fetal genes is mediated by activation of the calcineurin-NFAT pathway^[@R14]^. The *Trpc6* gene contains NFAT-responsive elements in the promoter and its expression is upregulated in several human and rodent models of heart failure^[@R16],[@R17],[@R20]^. We therefore measured the expression of *Trpc6* in ISO-treated wild-type and Klotho-deficient hearts. *Trpc6* mRNA levels were increased in wild-type hearts after ISO treatment ([Supplementary Fig. S3a](#SD1){ref-type="supplementary-material"}). For comparison, ISO treatment did not alter the expression of *Trpc6* in other tissues including blood vessels, lung, kidney, and liver. As was observed for cardiac fetal genes, ISO-induced increases in *Trpc6* mRNA were enhanced in Klotho-deficient relative to wild-type mice ([Supplementary Fig. S3b](#SD1){ref-type="supplementary-material"}). Interstitial fibrosis is another consequence of pathological cardiac hypertrophy and remodeling^[@R16]^. Trichrome staining of heart sections revealed fibrosis in wild-type hearts after ISO treatment, and Klotho-deficiency worsened ISO-induced cardiac fibrosis ([Fig. 1g](#F1){ref-type="fig"}).
In support of these results from morphometric and gene expression studies, functional analysis of hearts using magnetic resonance imaging showed that ISO treatment decreased the ejection fraction of wild-type hearts, and Klotho-deficiency markedly aggravated the ISO-induced decline in the ejection fraction ([Fig. 1h](#F1){ref-type="fig"}). Left ventricular end-systolic and end-diastolic volumes were markedly increased, and stroke volumes were decreased in Klotho-deficient mice after ISO treatment ([Supplementary Fig. S4a, b](#SD1){ref-type="supplementary-material"}), indicating chamber dilatation as well as impaired contractility of the left ventricle. Severe heart failure with lung edema developed in some Klotho-deficient mice after ISO treatment ([Supplementary Fig. S4c, d](#SD1){ref-type="supplementary-material"}). Thus, Klotho-deficiency does not cause baseline cardiac abnormalities but renders the heart more susceptible to stress-induced pathological cardiac remodeling.
Klotho attenuates stress-induced cardiac hypertrophy {#S4}
----------------------------------------------------
To further corroborate the above experimental data indicating that Klotho protects the heart against stress-induced cardiac remodeling, we examined ISO-induced cardiac changes in transgenic mice that overexpress Klotho (KL-Tg). These mice live \~20--30% longer than wild-type littermates, and the circulating level of soluble Klotho in transgenic mice is \~100% higher than WT (\~200 pM in transgenic mice *vs* \~100 pM in WT mice)^[@R27]^. Klotho overexpression in mice did not cause detectable changes in heart mass index and the heart size at baseline ([Fig. 2a, b](#F2){ref-type="fig"}), and nor did it alter the systemic blood pressure (systolic BP: 103 ± 7 mmHg and 103 ± 4 mmHg, WT vs KL-Tg, *n* = 4 each). Klotho overexpression yet blunted the ISO-induced cardiac hypertrophic responses ([Fig. 2a, b](#F2){ref-type="fig"}). Consistent with the notion that Klotho protects against stress-induced cardiac remodeling, Klotho overexpression did not alter *BNP* and *Trpc6* mRNA levels at baseline, but attenuated ISO-induced increases in *BNP* and *Trpc6* mRNA expression ([Fig. 2c, d](#F2){ref-type="fig"}). It has been reported that elevated serum FGF23 promotes cardiac hypertrophy^[@R28]^. Because Klotho and FGF23 work in the same pathway to regulate phosphate metabolism, we measured serum phosphate and FGF23 levels. Klotho overexpression in mice did not alter serum phosphate or FGF23 levels ([Fig. 2e, f](#F2){ref-type="fig"}), indicating that the cardioprotective effect of Klotho was not mediated by serum FGF23. As Klotho overexpression mice and control wild-type littermates were fed normal phosphate diets, these studies also exclude the role of dietary phosphate restriction in cardioprotection by Klotho.
Klotho protects the heart by downregulation of TRPC6 {#S5}
----------------------------------------------------
Next, we investigated the mechanism by which Klotho protects against stress-induced cardiac hypertrophy and remodeling. We have observed that Klotho-deficiency aggravated ISO-induced increases in *Trpc6* expression, and conversely Klotho overexpression attenuated the ISO-induced increases in cardiac *Trpc6* mRNA expression ([Supplementary Fig. S3b](#SD1){ref-type="supplementary-material"} and [Fig. 2d](#F2){ref-type="fig"}). Inhibition of cardiac TRPC6 by gene silencing or by dominant-negative expression of mutant channels confers cardioprotection^[@R17],[@R20]^. We examined whether Klotho may protect the heart by inhibiting TRPC6 by crossing Klotho-deficient mice with global *Trpc6*-knockout mice^[@R29]^. Mice with global deletion of *Trpc6* grow normally and have no apparent defects in major organ systems at baseline^[@R30]^. Consistently, we found that the baseline heart mass index was not different between mice with global deletion of *Trpc6* and control wild-type littermates ([Fig. 3a](#F3){ref-type="fig"}). Deletion of *Trpc6* partially protected against ISO-induced cardiac remodeling, but completely prevented the exaggerated ISO-induced cardiac hypertrophy in Klotho-deficient mice. Consistent with these results, deletion of *Trpc6* attenuated ISO-induced increases in *ANP* and *BNP* mRNA, and abolished the exaggerated ISO-induced increases in the mRNA in Klotho-deficient mice ([Fig. 3b, c](#F3){ref-type="fig"}).
We further investigated cardioprotection by Klotho using transgenic mice that overexpress TRPC6 in hearts. Mice with cardiac-specific overexpression of TRPC6 develop spontaneous cardiac hypertrophy and also have heightened sensitivity to stress-induced hypertrophy^[@R16]^. We crossed mice with cardiac TRPC6 overexpression with transgenic mice overexpressing Klotho to create double transgenic mice, and compared the survival rate, heart mass index, and cardiac fetal gene expression of double transgenic mice with those of having cardiac overexpression of TRPC6 and with wild-type littermates. Compared to wild-type littermates at 24 months of age, cardiac TRPC6-overexpressing mice had decreased survival and increased heart mass index and cardiac fetal gene expression without ISO treatment ([Fig. 3d--f](#F3){ref-type="fig"}). Klotho overexpression tended to improve the survival of cardiac TRPC6-overexpressing mice, prevented the increase in heart mass, and markedly diminished the increase in fetal gene expression induced by overexpression of TRPC6 in hearts. These studies also support the notion that Klotho protects the heart by inhibiting cardiac TRPC6.
Soluble Klotho inhibits TRPC6 in isolated cardiac myocytes {#S6}
----------------------------------------------------------
Klotho is not expressed in the heart. We tested the hypothesis that soluble Klotho present in the systemic circulation mediates the inhibition of cardiac TRPC6. TRPC6 is activated by diacylglycerol (DAG)^[@R15]^. We examined TRPC6 channel activity in freshly isolated ventricular myocytes by whole-cell patch-clamp recording using stimulation by endothelin-1 (ET1) to release DAG ([Fig. 4a](#F4){ref-type="fig"}). Ventricular myocytes from wild-type mice without ISO treatment showed baseline endothelin-1-activated TRPC-like currents presumably mediated by non-TRPC6 channels, as currents were not different between *Trpc6*-knockout mice (C6^−/−^) and wild-type littermates ([Fig. 4b](#F4){ref-type="fig"}). ISO treatment increased TRPC6-mediated currents in wild-type hearts ([Fig. 4a, b](#F4){ref-type="fig"}); the role of TRPC6 is supported by the facts that the increase was eliminated in *Trpc6*-knockout mice ([Fig. 4b](#F4){ref-type="fig"}), and identical currents were seen in TRPC6-overexpressing hearts ([Fig. 4a](#F4){ref-type="fig"} and [Supplementary Fig. S5](#SD1){ref-type="supplementary-material"}). Cell capacitance (a measurement of surface area of cells) of ventricular myocytes was increased in wild-type hearts after ISO and in TRPC6-overexpressing hearts ([Fig. 4c](#F4){ref-type="fig"}), supporting the conclusion that myocyte hypertrophy occurred under these conditions. Klotho overexpression in mice prevented the ISO-induced increases in currents ([Fig. 4b](#F4){ref-type="fig"}), and acute addition of soluble Klotho to culture media decreased TRPC6-mediated currents in myocytes isolated from wild-type mice after ISO treatment ([Fig. 4d](#F4){ref-type="fig"}). Similarly, direct addition of soluble Klotho inhibited the currents in myocytes isolated from TRPC6-overexpressing mice ([Fig. 4e](#F4){ref-type="fig"}). It is theoretically possible that Klotho inhibits TRPC6 by decreasing the production of DAG. However, soluble Klotho decreased TRPC6 currents in cardiomyocytes directly activated by membrane-permeant DAG ([Fig. 4f](#F4){ref-type="fig"}), indicating that it inhibits cardiac TRPC6 channel function acting downstream of DAG.
Klotho blocks IGF and PI3K-dependent exocytosis of TRPC6 {#S7}
--------------------------------------------------------
Because isolated mouse cardiomyocytes cannot be cultured continuously and the low abundance of endogenous TRPC6 in hearts, we further investigated the mechanism of regulation by soluble Klotho using HEK cells expressing recombinant TRPC6 as well as isolated cardiac myocytes. As in cardiomyocytes, soluble Klotho decreased DAG-activated TRPC6 channels in HEK cells ([Fig. 5a](#F5){ref-type="fig"}). Soluble Klotho treatment decreased cell-surface abundance of TRPC6 measured by biotinylation assays ([Fig. 5b](#F5){ref-type="fig"}). Soluble Klotho exhibits sialidase activity and increases cell-surface abundance of TRPV5 channels by cleaving sialic acids in the N-glycans of channels^[@R11]^. The sialidase activity of Klotho is not responsible for the regulation of TRPC6, since purified sialidase had no effect on TRPC6 whereas it stimulated TRPV5 ([Fig. 5c, d](#F5){ref-type="fig"}). The above results also indicate that soluble Klotho decreases cell surface expression of TRPC6 via a mechanism not restricted to cardiomyocytes.
The decrease in cell surface abundance of TRPC6 by soluble Klotho may be caused by decreased exocytosis and/or increased endocytosis of the channel. Blocking exocytosis by v-SNARE inhibitor tetanus toxin^[@R31]^ decreased TRPC6 currents, and prevented further inhibition by soluble Klotho ([Fig. 5e](#F5){ref-type="fig"}). Because tetanus toxin completely prevented the effect by Klotho, the major (if not the sole) action of Klotho on TRPC6 is by blocking exocytosis. Consistent with this notion, we found that blocking endocytosis using a dominant-negative dynamin did not affect the ability of Klotho to inhibit TRPC6: Coexpression with dominant-negative dynamin increased the basal (i.,e., without KL) TRPC6 currents indicating inhibition of endocytosis of channels, but soluble Klotho decreased TRPC6 currents similarly in cells expressing dominant-negative dynamin and in cell expressing the control wild-type dynamin ([Fig. 5f](#F5){ref-type="fig"}).
Phosphoinositide-3-kinase (PI3K)-activating growth factors increase cell surface abundance of TRPC channels by stimulating exocytosis^[@R32]^. Soluble Klotho inhibits PI3K signaling by insulin and insulin-like growth factors (IGF), which contributes to the anti-aging and tumor-suppression effects of Klotho^[@R27],[@R33]--[@R35]^. We therefore tested the hypothesis that soluble Klotho inhibits TRPC6 by interfering with IGF1 activation of PI3K to promote exocytosis of channels. To allow for studying the effect of IGF1, we first examined the effect of serum deprivation. Serum deprivation lowered TRPC6 cell-surface abundance and activities, and prevented the inhibition by soluble Klotho ([Fig. 6a, b](#F6){ref-type="fig"}). The role of IGF1 is demonstrated by findings showing that physiological concentrations of IGF1 (10 nM) reproduced the effect of serum to promote TRPC6 currents and that soluble Klotho inhibited IGF1-stimulated TRPC6 currents ([Fig. 6c](#F6){ref-type="fig"}). Moreover, PI3K inhibitor wortmannin decreased TRPC6 currents stimulated by IGF1, and prevented a further decrease of TRPC6 by soluble Klotho ([Fig. 6d](#F6){ref-type="fig"}). These results support the notion that soluble Klotho inhibits IGF1 and PI3K-dependent exocytosis of TRPC6. Because wortmannin abrogates its effect, soluble Klotho acts upstream of PI3K.
Finally, we examined whether soluble Klotho regulates TRPC6 in hearts via the same mechanism. Wortmannin decreased TRPC6 currents and prevented the inhibition by soluble Klotho in isolated cardiomyocytes ([Fig. 6e](#F6){ref-type="fig"}). Furthermore, the effect of wortmannin to decrease cardiac TRPC6 currents and to prevent further inhibition by soluble Klotho was reproduced by tetanus toxin, and the effects of wortmannin and tetanus toxin were not additive ([Fig. 6f](#F6){ref-type="fig"}). Thus, tetanus toxin and wortmannin inhibit cardiac TRPC6 via the same mechanism; i.e., by blocking exocytosis of channels. Collectively, these data strongly support the hypothesis that soluble Klotho inhibits TRPC6 by blocking PI3K-dependent exocytosis of channels ([Fig. 7](#F7){ref-type="fig"}).
Discussion {#S8}
==========
The data presented in this study provide compelling evidence indicating that soluble Klotho protects the heart against stress-induced cardiac hypertrophy and remodeling. Klotho expression is decreased in aging^[@R23]^, thus decline in circulating soluble Klotho may contribute to age-related cardiomyopathy in humans. One consequence of cardiac aging is increased sensitivity to stress-induced heart failure^[@R36]^ similar to the changes in Klotho-deficient mice we observed here. Many Klotho-mediated aging phenotypes, such as vascular calcification, growth defects and premature death, are attributed to defects in the function of membrane Klotho as co-receptors for FGF23 and phosphate retention^[@R3]--[@R8]^. Our results show that the cardioprotective effect of soluble Klotho is independent of FGF23 and phosphate metabolism. First, Klotho overexpression in mice confers cardioprotection without altering serum phosphate and FGF23 levels. Second, dietary phosphate restriction normalizes serum phosphate levels of Klotho-deficient mice to the level of wild-type mice, excluding hyperphosphatemia as the culprit of cardiac dysfunction in these mice. Moreover, deletion of *Trpc6* completely prevents exaggerated stress-induced cardiac hypertrophy and remodeling in Klotho-deficient mice, indicating that cardioprotection by Klotho is mediated by down regulation of TRPC6.
TRPC6 is broadly expressed in tissues^[@R15],[@R37]^. While it is possible that the effect of Klotho on TRPC6 in other tissues also contributes to cardioprotection, the following results indicate that Klotho inhibition of cardiac TRPC6 plays a critical role in the process. First, Klotho ameliorates cardiac hypertrophy and remodeling induced by heart-specific overexpression of TRPC6, and Klotho inhibits TRPC6 in isolated cardiomyocytes. Second, Klotho-deficient mice have no cardiac dysfunction at baseline, but develop exaggerated cardiomyopathy in response to ISO treatment. ISO treatment causes upregulation of *Trpc6* mRNA in the heart, but not in other tissues that influence cardiac function, such as blood vessels, lung, and the kidney. The effect of *Trpc6* deletion to prevent ISO-induced exaggerated cardiomyopathy in Klotho-deficient mice is therefore most likely due to abolition of ISO-induced increases of cardiac TRPC6.
Our findings also have important implications in chronic kidney disease (CKD), a disease that affects approximately 10% of the general population^[@R38],[@R39]^. The prevalence of cardiac hypertrophy in patients of advanced stages of CKD is estimated as high as 90%, and cardiac dysfunction is the main cause of death for the patients^[@R39]--[@R41]^. Klotho is predominantly produced in the kidney, and circulating levels of soluble Klotho are reportedly decreased in CKD^[@R42],[@R43]^. Our study supports that decreased levels of soluble Klotho contribute to the pathogenesis of cardiac hypertrophy in CKD. Recently, Faul et al reported that FGF23 stimulates cardiomyocyte growth and increased serum FGF23 contributes to cardiac hypertrophy in CKD^[@R28]^. Interestingly, FGF23 appears to induce cardiac hypertrophy independently of stress factors, whereas Klotho deficiency predisposes the heart to stress-induced pathological hypertrophy. Thus, increased FGF23 and Klotho deficiency may synergistically contribute to cardiac hypertrophy in CKD by participating at different stages of pathogenesis.
The physiological role of TRPC6 in hearts is elusive. Its function appears to be dispensable, as mice with deletion of *Trpc6* have no apparent cardiac dysfunction. Consistent with this observation, we found that TRPC6 channel activity is undetectable in hearts at baseline. Overstimulation by isoproterenol leads to increase in *Trpc6* mRNA levels and functional TRPC6 currents in mouse hearts. Increased expression of cardiac TRPC6 has also been reported in mouse models of cardiac hypertrophy induced by calcineurin gene overexpression, by overstimulation by neuroendocrine hormones including endothelin-1, phenylephrine, and angiotensin II, by thoracic aortic banding pressure overload, and in human failing hearts^[@R16],[@R17],[@R20]^. Thus, soluble Klotho protects the heart by acting on a molecule that is normally quiescent but activated during stresses.
Mechanistically, we propose that insulin-like growth factors such as IGF1 provide a tonic stimulation for exocytosis of TRPC6 via PI3K, and soluble Klotho exerts a tonic inhibition to the system (model in [Fig. 7](#F7){ref-type="fig"}). Cardiac stresses increase the intracellular Ca^2+^ concentration from multiple mechanisms^[@R44]^. The abnormal intracellular calcium signaling in the heart activates calcineurin and NFAT to initiate fetal gene expression and pathological cardiac hypertrophy and remodeling. TRPC6 contains NFAT-responsive elements in its promoter and is also upregulated by stress. The increased Ca^2+^ influx through TRPC6 causes a feed-forward cycle and further amplifies and sustains the process. By placing a brake on the system, Klotho protects the heart. Conversely, Klotho deficiency accelerates stress-induced cardiac remodeling. Without stress signals to upregulate TRPC6, neither Klotho deficiency nor overexpression in mice affects cardiac function at baseline.
Multiple studies have reported that soluble Klotho inhibits intracellular signaling by insulin and IGF1^[@R27],[@R33]--[@R35]^. Kurosu et al first reported that soluble Klotho inhibits insulin and IGF-mediated activation of PI3K pathway by inhibiting activation of receptors and repressing activated receptors^[@R27]^. This anti-insulin/IGF effect contributes to aging-suppression by Klotho in mice. Wolf et al further found that soluble Klotho suppresses the growth of human breast and pancreatic cancer cells^[@R33],[@R34]^. They also found that soluble Klotho coimmunoprecipitated with IGF1 receptors, and suggested that soluble Klotho inhibits the intracellular signaling by IGF1 at least partly by direct interactions with receptors^[@R33]^. It was also reported that Klotho prolongs life span and stress resistance in *C. elegans* by blocking insulin and IGF-like signaling in worms^[@R35]^. Our results support these previous reports and extend the anti-insulin/IGF role of Klotho to cardioprotection. Cardioprotection by Klotho may contribute to the anti-aging effect of Klotho in mice. Activation of PI3K and downstream Akt signaling cascade in the heart is important for physiological cardiac growth, but it can also lead to pathological cardiac hypertrophy^[@R13],[@R45]^. Inhibition of TRPC6 by soluble Klotho may be a mechanism for preventing PI3K to cause pathological cardiac hypertrophy in the normal heart. Interestingly, shedding of soluble Klotho from membranous Klotho is mediated by metalloproteinases ADAM 10 and 17, and insulin stimulates the shedding through the PI3K pathway^[@R46]^. Whether the regulation of shedding of soluble Klotho by insulin/IGF1 plays any roles in the control of cardiac functions awaits future investigation.
Other TRPC channels including TRPC1, 3, 4, and 5 are also present in the heart. Increased expression of these channels is also associated with cardiac hypertrophy induced by pathological stimuli and downregulation confers the protection^[@R17]--[@R20]^. Cardiac TRPC channels are likely heteromultimers of different TRPC members^[@R15],[@R22],[@R23]^, which may partly explain why inhibition of different TRPC channels can confer cardioprotection. The exact molecular composition and stoichiometry of TRPC channels that form heteromultimers with TRPC6 in the heart, however, remains unknown. In this study, deletion of *Trpc6* totally abolishes exaggerated ISO-induced cardiac hypertrophy and remodeling in Klotho-deficient mice, indicating that inhibition of TRPC6 alone is sufficient for cardioprotection. Our study yet does not exclude the possibility that soluble Klotho also exerts inhibition on other TRPC channels that form multimers with TRPC6. Of note, *Trpc6* deletion only partially blunts isoproterenol-induced hypertrophy, and other factors besides TRPC6 (such as other TRPC channels) are also involved in the hypertrophic response to ISO treatment.
Pharmacological TRPC antagonism is in development as a potential treatment of cardiac hypertrophy^[@R21],[@R22],[@R47]^. As an endogenous hormone that may extend human lifespan, soluble Klotho or its analogs or activators^[@R48]^ may prove to be important therapeutic agents. The involvement of TRPC6 in multiple models of cardiac hypertrophy and heart failure^[@R16],[@R17],[@R20]^ and the ability of Klotho to protect against cardiac hypertrophy induced by heart-specific overexpression of TRPC6 suggest that Klotho-based therapeutic strategies may be applicable to diverse cardiac diseases. TRPC6 is also expressed in the kidney, and systemic and pulmonary vasculature, and increased TRPC6 function in these tissues leads to pathologies^[@R15],[@R37],[@R49],[@R50]^. Klotho-based therapeutics may also be valuable in treating TRPC6-related pathologies in other organs.
One major unanswered question in the pathogenesis of cardiac hypertrophy and remodeling is how the heart distinguishes between overwhelming intracellular calcium transients during each normal cardiac cycle and the abnormal calcium signaling induced by stress signals^[@R44]^. Mice with Klotho deficiency or overexpression have no apparent cardiac abnormalities at baseline. The selective targeting to the stress-induced abnormal calcium signaling by Klotho may provide clues to answer this question in the future.
Methods {#S9}
=======
General experimental procedures of mice {#S10}
---------------------------------------
*Klotho*-hypomorphic, KL-Tg, *Trpc6*-knockout, and TRPC6-Tg (line L16) mice have been described^[@R1],[@R16],[@R27],[@R29]^. Each mouse line was backcrossed to 129/SvJ mice for \>6 generations to achieve congenic background. For dietary phosphate restriction, mice were fed a purified diet containing with 0.2% (wt/wt) inorganic phosphate (TD-09073, Harlan Teklad, Madison, WI) from weaning at \~3 weeks of age. Normal phosphate diets contain 0.35% inorganic phosphate. All mice subjected to experiments were males at \~3 months of age unless otherwise specified. Blood pressure was measured in wild type, *Klotho*-hypomorphic, and KL-Tg mice using a tail-cuff sphygmomanometer as previously described^[@R51]^.
For induction of cardiac hypertrophy, isoproterenol (2 mg/kg/day diluted in PBS) was injected subcutaneously to mice once per day for 10 consecutive days^[@R52]^. Control mice received PBS injection. At day 11, mice were euthanized and hearts were isolated. After measurement of weight, a portion of the hearts was snap-frozen in liquid N2 and saved for RNA isolation, and the remainder fixed and stored for histology. All animal protocols were approved by the University of Texas Southwestern Institutional Animal Care and Use Committee.
Real-time quantitative RT-PCR analysis of mRNA {#S11}
----------------------------------------------
RNA was extracted from heart samples with trizol (Invitrogen), reverse-transcribed into cDNA (Taqman reverse transcription reagents, Applied Biosystems-Roche), and mRNA abundance was analyzed by real-time PCR with SYBR-green (iTaq or iQ SYBR-green Supermix, BioRad). Primers: *GAPDH*, 5′-tgcaccaccaactgcttagc, 5′-ggcatggactgtggtcatgag; *ANP*, 5′-gccatattggagcaaatcct, 5′-gcaggttcttgaaatccatca; *BNP*, 5′-ccaaggcctcacaaaagaac, 5′-agacccaggcagagtcagaa; β*-MHC*, 5′-ttggatgagcgactcaaaaa; 5′-gctccttgagcttcttctgc; *Trpc6*, 5′-cgctgccaccgtatgg; 5′-ccgccggtgagtcagt.
Histological analysis {#S12}
---------------------
Dissected hearts were rinsed in PBS and incubated in Krebs-Henseleit solution lacking Ca^2+^ for 30 min, and were then fixed in 4% paraformaldehyde overnight at room temperature. Samples were dehydrated and stored in 50% ethanol, mounted in paraffin, and sectioned. Sections were then stained with H&E or with Masson's Trichrome stain.
Serum collection and measurement {#S13}
--------------------------------
Blood was drawn from mice using retro-orbital bleeding method. Samples were immediately centrifuged, and supernatant collected and stored. Serum phosphate and FGF23 levels were measured using a phosphate assay kit (Stanbio labs, San Antonio, TX) and FGF23 ELISA kit (Kainos lab, Japan), respectively.
Cardiac magnetic resonance imaging {#S14}
----------------------------------
Cardiac MRI of mice was performed in the Mouse MRI Core Facility of UT Southwestern Medical Center as previously described^[@R53]^. To determine the left ventricle (LV) volume, multiple parallel slices of 1-mm thickness perpendicular to the long heart axis were imaged. The area of LV of each slice at both end-diastolic and end-systolic phases was measured using ImageJ software. LV volume was calculated as the sum of area of all slices at either phase. Stroke volume is the difference between end-diastolic and end-systolic volumes. Ejection fraction is the percentage of stroke volume over end-diastolic volume.
Isolation of cardiac ventricular myocyte {#S15}
----------------------------------------
Isolation of mouse ventricular myocytes was performed per established procedure^[@R54]^. Briefly, mice received heparin (100 U/mouse) and anesthesia. Hearts were quickly removed and perfused retrograde via the aorta with a solution containing (in mM) 113 NaCl, 4.7 KCl, 1.2 MgSO~4~, 0.6 KH~2~PO~4~, 0.6 Na~2~HPO~4~, 10 NaHCO~3~, 30 taurine, 5.5 glucose, 10 2,3-butanedione monoxime (BDM), 10 HEPES (at pH 7.4) and followed by a solution containing in addition 1 mg/ml type 2 collagenase (Worthinton) and 0.1 mg/ml protease XIV (Sigma). The perfusion solution was maintained at 37 °C and equilibrated with 100 % O~2~. Thereafter, the ventricle was removed, chopped into small pieces, and further digested in the enzyme solution. After stopping enzyme digestion by adding 2.5% BSA and 0.1 mM CaCl~2~, the tissue-cell suspension was filtered through a sterilized-gauze sponge, centrifuged using a tabletop centrifuge at 50 *g* for 1 min. The resulting cell pellet was resuspended in the stopping buffer and \[Ca^2+^\] titrated to 0.5 mM by addition of 100 mM CaCl~2~ stock solution in four steps over 20 min. Isolated myocytes were stored at room temperature until use.
Whole-cell recording {#S16}
--------------------
For recording of isolated myocytes, cells were transferred into a perfusion chamber mounted on an inverted microscope and continually perfused at the rate of 1 ml/min with bath solution. Whole-cell currents were recorded under voltage-clamp using an Axopatch 200B patch-clamp amplifier (Axon instruments Inc., Foster City, CA, USA)^[@R55]^. Voltage protocol consists of holding at −40 mV and repetitive descending ramp pulses from +120 mV to −120 mV for 500 ms applied every 10 sec. The pipette solution contained (in mM) 9.4 NaCl, 120 CsCl, 1 MgCl~2~, 3.5 CaCl~2~, 10 BAPTA, 10 HEPES, 0.2 NaGTP (pH 7.2) (calculated ionized \[Ca^2+^\] 80 nM) and the bath solution contained 140 NaCl, 5 CsCl, 1 MgCl~2~, 1.2 CaCl~2~, 10 glucose, 10 HEPES (pH 7.4). Bath solution also contained 1 μM of nifedipine and 3 mM of NiCl~2~ to block current flow through L-type Ca^2+^ channel and Na^+^/Ca^2+^ exchanger, respectively. The pipette resistance was \~2--3 MΩ when filled with the pipette solution. Whole-cell access resistance was \<10 MΩ. Endothelin-1 (20 nM) was administrated using focal application method. The distance between the tip of the applicator and myocyte was \<50 μm. Currents were low-pass filtered at 2 kHz and sampled every 0.1 ms. Data acquisition was performed using pClamp9.2 program (Axon Instrument, Inc.) and analysis using Prism (V3.0) software (GraphPad Software, San Diego, CA, USA). For whole-cell recording of recombinant TRPC6 channels in HEK cells, the pipette and bath solution contained (in mM) 120 Cs-aspartate (Cs-Asp), 10 CsCl, 1 MgCl~2~, 2 MgATP, 5 EGTA, 1.5 CaCl~2~ (free \[Ca^2+^\] = 70 nM) and 10 CsHEPES (pH 7.2) and 140 NaCl, 5 KCl, 0.5 EGTA and 10 NaHEPES (pH 7.4), respectively.
Surface biotinylation assay {#S17}
---------------------------
HEK cells expressing hemagglutinn (HA)-tagged TRPC6 (in 35 mm culture dish) were incubated with or without soluble Klotho, washed with 1 ml of ice-cold PBS three times, and incubated with 1 ml of PBS containing 1.5 mg/ml EZ-link-NHS-SS-biotin (Thermo Scientific) for 2 h at 4 °C. After quenching with glycine-containing PBS for 20 min, cells were lysed in a buffer (150 mm NaCl, 50 mm Tris-HCl, 5 mm EDTA, 1% Triton X-100, 0.5% deoxycholate, and 0.1% SDS) containing protease inhibitor mixture for 30 min. For detection of biotinylated proteins, lysates were precipitated by streptavidin-agarose beads (Thermo Scientific) for 2 h at 4 °C. Beads were subsequently washed 3 times with TBS containing 1% Triton X-100. Biotin-labeled proteins were eluted in sample buffer, separated by SDS-PAGE, and transferred to nitrocellulose membranes for Western blotting using mouse monoclonal anti-HA antibody (Sigma-Aldrich; 1:250 dilution) or anti-α-tubulin antibody (Sigma-Aldrich; 1:500 dilution).
Statistical analysis {#S18}
--------------------
Statistical comparison was made between control and experimental groups conducted during the same time period. Each experiment was repeated at least once at separate times and with similar results. Data are presented as means ± s.e.m. Statistical comparison between two groups of data were made using two-tailed unpaired Student's *t*-test. Multiple comparisons were determined using one-way analysis of variance followed by Tukey's multiple comparison tests. Statistical comparison of Kaplan-Meier cumulative survival curves was made using "log-rank" analysis (<http://bioinf.wehi.edu.au/software/russell/logrank/>).
Supplementary Material {#S19}
======================
We thank Eric Olson for TRPC6-Tg mice, Jyothsna Gattineni for assistance with measurements of serum FGF23 and phosphate, Masaya Takahashi and Kim Kangasniemi for cardiac MRI, and Peter Igarashi, Orson Moe, and Aylin Rodan for discussions and comments. This work was supported by NIH (DK59530, DK85726, DK79328, DK91392), the Intramural Research Program of the NIH (ZO1-ES-101684) and by a GRIP grant from Genzyme, Inc. CLH holds the Jacob Lemann Professorship in Calcium Transport of University of Texas Southwestern Medical Center.
**Competing financial interests**
The authors declare no competing financial interests.
**Author contributions**
J.X., S.-K. C. and S.-W. A. designed the study, conducted the experiments, analyzed the data, and participated in writing the paper; M.K. contributed Klotho-deficient and overexpression mice, and analysis of Klotho expression in Klotho-deficient mice; L. B. contributed Trpc6-deleted mice. C.-L. H. supervised the entire project and wrote the final paper. All authors read, commented and approved the paper.
{ref-type="supplementary-material"} for measurements). \* *P* \< 0.01 vs no ISO; \# *P* \< 0.01 between indicated groups.](nihms425662f1){#F1}
{#F2}
{#F3}
{#F4}
![Soluble Klotho decreases cell surface abundance of TRPC6\
(**a**) Soluble Klotho (sKlotho) inhibits recombinant TRPC6 channels. HEK cells expressing hemagglutinin (HA)-tagged TRPC6 were incubated with or without purified recombinant soluble Klotho^[@R11],[@R27]^ (KL 200 pM for 2h) before ruptured whole-cell recording. TRPC6 currents were activated by membrane-permeant DAG. Shown is inward current density at −100 mV (*n* = 7 for each). \* *P* \< 0.01 vs. control (no KL). In control cells without expression of recombinant TRPC6, inward current density after application of DAG was 21 ± 5 pA/pF (at −100 mV; *n* = 9). Note that I-V curves for recombinant TRPC6 currents shown here are more strongly double-rectifying than for native currents shown in [Figure 4a](#F4){ref-type="fig"}. The differences may be partly due to formation of heteromultimers of TRPC6 with other TRPC members in native hearts. (**b**) Soluble Klotho decreases TRPC6 cell surface (labeled as "Surf") abundance. Specific biotinylation of membrane TRPC6 was supported by lack of detection of α-tubulin in the membrane fraction. The abundance of α-tubulin (labeled as "α-Tub") in lysates served as a loading control. Bar graph shows mean ± s.e.m. of four separate experiments. TRPC6 bands (detected by anti-HA antibody) were quantified by densitometry. \* *P* \< 0.05 vs. control (no KL). (**c**) Soluble Klotho, but not purified sialidase^[@R11]^ (Sial, 0.3 U/ml), inhibits TRPC6. *n* = 6 for each. \* *P* \< 0.01 vs. control (no KL). (**d**) Purified sialidase increases TRPV5 currents. Whole-cell TRPV5 currents were recorded as described^[@R11]^. *n* = 6 each. (**e**) Tetanus toxin decreases TRPC6 currents, and prevent the inhibition by soluble Klotho. TRPC6-expressing cells were preincubated with tetanus toxin (50 nM for 3h) before soluble Klotho. *n* = 6--7 for each. \* *P* \< 0.01 vs. control (no KL). ns, not significant between indicated groups. (**f**) Role of endocytosis in inhibition of TRPC6 by Klotho. TRPC6 was coexpressed with a dominant-negative (K44A; lysine-44 to alanine mutation; DN Dyn II) or wild-type dynamin II (Wt DN II). *n* = 7--8 for each. \* *P* \< 0.01 vs. control (no KL). \#, *P* \< 0.02 between indicated groups. All data are expressed as mean ± s.e.m.](nihms425662f5){#F5}
{#F6}
{#F7}
[^1]: These authors contributed equally to the work
[^2]: Present address:Department of Physiology and Institute of Life Style Medicine, Yonsei University Wonju College of Medicine, Ilsan-Dong 162, Wonju, Kangwondo, 220-701, Republic of Korea
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#sec1_1}
============
Approximately 47 million people worldwide are estimated to live with dementia and the incidence per year is estimated to be 9.9 million cases \[[@B1]\].
Dementia presents a wide range of symptoms, such as cognitive impairment and behavioural and motoric changes. Communication difficulties are among the earliest symptoms of dementia \[[@B2]\], especially verbal communication difficulties, which are seen as a significant problem among those with severe dementia \[[@B3], [@B4]\]. However, the urge to communicate and the need to be part of the society will remain, regardless of the degree of dementia \[[@B5]\]. The lack of verbal communication can make people with dementia appear unreachable and lead to fewer attempts by carers to communicate with the person \[[@B6]\], and even result in "social death" \[[@B7]\].
People with dementia at a severe stage are less able to be active and therefore tend to be isolated and bored. The use of psychosocial interventions in dementia care is highly recommended since antipsychotic treatment may have serious side effects in the management of behavioural and psychological symptoms of dementia \[[@B8], [@B9]\]. Psychosocial interventions are also referred to as a way of promoting person-centred care for individuals with demen tia \[[@B10]\].
Lubinski \[[@B11]\] emphasizes the importance of cognitive stimulation through the senses for those with dementia. This is supported by Vozzella \[[@B12]\] who claims that as the person regresses, the need for stimulation increases. Bakshi \[[@B13]\] found that individuals with moderate to severe dementia were given less antipsychotic medication after having received stimulation. However, despite the evidence that communication difficulties are problematic for persons with dementia, a review conducted by Strøm et al. \[[@B14]\] of available sensory stimulation strategies identified only one quantitative study that measured the effect on communication. This study, conducted by Hutson et al. \[[@B15]\], found no effect of the Sonas programme on communication.
The majority of available sensory stimulation interventions include stimulation of only 1 of the senses, such as music, light therapy, acupressure/reflexology, massage/aromatherapy and doll therapy/pet therapy/toy therapy \[[@B14]\]. However, the Sonas programme, which is a multi-sensory stimulation programme developed by Sr. Mary Threadgold in 1990, involves cognitive, sensory, and social stimulation, including all 5 senses: touch, smell, taste, hearing, and sight. The programme is a therapeutic activity for people who have significant communication impairment, primarily as a result of dementia. The aims of the Sonas programme are: (1) to activate whatever potential for communication has been retained by an older person with communication impairment, (2) to encourage the creation of an environment which will facilitate communication, and (3) to have activation of the potential for communication recognized and accepted as an essential part of care planning for older people. However, research investigating the effectiveness of the Sonas programme is limited with only few studies published, one of which assessing the effect on communication as a secondary outcome measure.
Of the relevant studies on this topic, the first was an observation study carried out by Brown \[[@B16]\] who reported some positive changes in well-being, self-confidence, self-esteem, and trust as well as improved alertness, happiness, quality of life (QoL), and relaxation after attending the Sonas sessions. No benefit was found on agitation and aggression. However, this study has several limitations, with an unknown number of participants and undisclosed length of the intervention period as well as outcome measures, and no control group. Parrish et al. \[[@B17]\] used Dementia Care Mapping to observe 51 participants with dementia in day hospitals and care wards in institutions attending Sonas sessions. Although no benefit was reported on reduced agitation or aggression, a significant number of participants showed a positive change in well-being after attending the Sonas sessions as well as being more animated and initiating conversation and singing.
The first randomized controlled trial (RCT) measuring the effect of the Sonas programme was conducted in 2003 by Jackson et al. \[[@B18]\]. All 75 participants were randomly selected to either attend a weekly Sonas session for 8 weeks or be part of the control group where the participants received standard care. No significant differences were found between the groups in relation to aggression, agitation, depression or cognitive impairment. A recent pilot RCT \[[@B15]\] included 36 residents with moderate to severe dementia, of whom 20 participated in Sonas sessions twice weekly over a period of 7 weeks; no statistically significant differences in depression, anxiety, communication, and QoL were reported. However, there was some improvement for both groups in relation to depression and anxiety, where the control group demonstrated greater improvement than the Sonas group. Further, the Sonas group demonstrated an improvement in communication ability and QoL, as well as a greater reduction in behavioural and mood disturbance than the control group.
The need for high-quality research investigating psychosocial interventions for older people with dementia has been emphasized \[[@B19]\]. In addition, only few studies have investigated the effectiveness of techniques to improve communication for people with dementia, which is of importance as communication is essential for social life of people regardless of cognitive function \[[@B20]\]. With just few published studies examining the effect of the Sonas programme, some showing methodological limitations, short duration and small sample sizes, the overall aim of this study was to examine the effect of the Sonas programme on communication ability for people with moderate to severe dementia living in a nursing home. The secondary aim was to explore if any effect could be related to cognitive function and different aspects of communication.
Method {#sec1_2}
======
The study is reported according to the CONSORT statement 2010 \[[@B21]\] and the extended version of the CONSORT statement for non-pharmacological treatment \[[@B22]\].
Study Design {#sec2_1}
------------
The study is an RCT where the participants were assigned to 1 of the 3 groups for 24 weeks: (1) the Sonas programme, (2) the reading group, and (3) standard care. The reading group was set up in order to rule out the possibility that an effect of attending the Sonas sessions could be caused by the attention given by staff leading the group, rather than the actual Sonas programme. The study took place from April to October 2014. Data collection was carried out at baseline (T0) before randomization, and after 12 weeks (T1) and 24 weeks (T2), respectively.
Setting {#sec2_2}
-------
In all 168 nursing homes in Dublin, Ireland, and the 5 surrounding counties, all registered with the independent authority, i.e., "Health Information and Quality Authority" (HIQA), were considered eligible. The goal of HIQA is to perform continuous improvement in Ireland\'s health and social care services \[[@B23]\]. Only long-term facilities with at least 40 beds and which had not used the Sonas programme earlier were initially chosen to take part in the study. Of the 168 nursing homes, 47 have less than 40 beds, 18 are hospitals, and 78 had already used the Sonas programme according to the last inspection report carried out by HIQA. The remaining 25 nursing homes received an e-mail about the trial. The first author surveyed the 25 nursing homes to determine whether there were adequate numbers of potential participants with dementia. Eight nursing homes agreed to participate; however, 1 withdrew. Two more counties were included and 1 of the 9 eligible nursing homes in these counties accepted to participate. Another nursing home withdrew and 1 did not have enough participants for 2 groups, which left us with 6 nursing homes for this RCT (Fig. [1](#F1){ref-type="fig"}).
Participants {#sec2_3}
------------
The participants were recruited between January and March 2014 and the Director of Nursing in each nursing home identified potential male and female residents. A total of 147 persons were considered eligible for the study. Informed consent was obtained from the next-of-kin since the persons with dementia were unable to sign the informed consent. Persons were considered suitable for participation when they were ˃65 years of age, were diagnosed with dementia, had moderate to severe cognitive functioning as classified by pre-trial Mini-Mental State Examination (MMSE) scores of 0--20, spoke English and were living in the nursing home. Persons at a palliative stage, those with major depression, current or partial remission, severe pain, or those having been exposed to previous Sonas sessions were not included. Three of the nursing homes had enough eligible participants for 3 groups of 8, while the other 3 only had enough eligible participants for 2 groups. This left us with 120 participants, while 27 were left out of the study.
Randomization Process {#sec2_4}
---------------------
Due to ethical considerations, it was decided beforehand that each nursing home would have 1 Sonas group in order to give them the possibility to attend a new programme. The Director of Nursing at each site and this article\'s first author placed the name of the participants in a container and study participants were then drawn: 24 participants from 3 nursing homes and 16 from the other 3 nursing homes. After baseline data had been collected, the randomization was carried out by the same persons.
The participants were randomly assigned to either the Sonas group (*n* = 48), the reading group (*n* = 32) or the control group (*n* = 40), by drawing lots. For the nursing homes with 24 eligible participants, the participants were randomly allocated to 1 of the 3 groups. For the nursing homes with 16 eligible participants, 8 were first randomly allocated to the Sonas group, whereas the remaining 8 participants were randomly allocated to either a reading or control group, by drawing lots.
Intervention {#sec2_5}
------------
The Sonas programme was carried out twice a week in groups of 8 residents over a period of 24 weeks. Each session lasted 45 min and was led by a person trained in the Sonas programme (Sonas licensed practitioner) and an assistant. The programme consists of 13 elements and follows the same structure each time, believing that repetition is a way of helping the individual to remember \[[@B24]\]. The participants are seated in a semi-circle and the session begins with a welcome song where each one is individually greeted by name (Fig. [1](#F1){ref-type="fig"}). The reading group received 45 min of reading from a newspaper twice a week for 24 weeks. The participants of the control group did not participate either in the Sonas sessions or the reading groups, but continued with their daily routines, for example, attending the activity programme set up by the nursing home. The staff (Sonas licensed practitioners and care assistants) involved in the study, either received (1) a 3 days' training consisting of training and support in the Sonas programme or (2) preparation to lead the reading group by this article\'s first author. The other staff did not receive any extra training.
Main Outcome and Assessments {#sec2_6}
----------------------------
The main outcome measure was the Holden Communication Scale (HCS) which measures communication ability. The HCS includes 12 items assessing conversation, awareness, humour, and responsiveness. The score range was from 0 to 48, and a higher score denotes more difficulties with communication \[[@B25]\]. The HCS has been evaluated to be a reliable and valid instrument to measure communication ability in persons with dementia \[[@B26]\].
The degree of cognitive functioning was assessed by the MMSE. The questionnaire comprises 20 questions that cover orientation, memory, reading, calculation, recall, and language. Each question is scored, and the sum score can vary between 0 and 30. A higher score denotes better cognition \[[@B27]\]. The MMSE has been found to have a satisfactory reliability and construct validity \[[@B28]\]. Residents' characteristics were obtained from their records at baseline (Table [1](#T1){ref-type="table"}).
Participants were assessed at baseline (T0), after 12 weeks (T1), and after 24 weeks (T2) by nurses who had received training in the use of the instruments and were not involved in either the Sonas sessions or reading groups. They were blinded at baseline data collection, but this was not possible at follow-ups.
Data Analysis {#sec2_7}
-------------
### Power Calculation {#sec3_1}
We aimed at comparing the HCS score between the Sonas and the reading group as the primary analysis, and assumed Cohen\'s *d* = 0.80 as our effect size. With the significance level of 5% and power of 80%, we estimated a minimum sample size of 24 in each group. A drop-out rate of 30% was assumed. The sample size required was therefore adjusted to 32 persons per group.
### Analysis {#sec3_2}
Each participant\'s characteristics at baseline were summarized as frequencies and percentages for the categorical variables and as means and standard deviations for the continuous variables. The normality of continuous variables was assessed by inspecting histograms.
None of the participants changed the group they originally had been assigned to; hence, there was no need for an intention-to-treat approach.
Differences in HCS total scores between the intervention group, the reading group, and the control group were assessed by estimating a linear mixed model with fixed effects for the time component up to second-order and group variable. Interaction between the two was included as well. A significant interaction term would imply differences between the groups in development of the HCS score through the follow-up period. Random effects for participants nested within nursing homes were included in the model. Such a model correctly accounts for intra-participant correlations due to repeated measurements for each participant as well as cluster effect within each nursing home. The linear mixed model handles unbalanced data by including all available information, also from drop-outs. An exploratory analysis was performed by estimating the same model for each component of the HCS.
A similar model as above was estimated in order to assess changes among those with moderate and with severe cognitive decline. For this purpose, an extra fixed effect for cognitive decline level was included together with the interaction between the level and time and the level and the group with moderate and severe cognitive decline. Such an analysis was performed in order to explore which of these 2 groups benefits most from the Sonas programme compared with the reading and control group.
The statistical analyses were conducted using SPSS version 22 and SAS version 9.4. Results with *p* values below 0.05 were considered statistically significant.
Ethical Considerations {#sec2_8}
----------------------
The study was approved by the Regional Ethics Committee of Norway (REK) under the registration number IRB 0000 1870. No ethical approval was needed in Ireland, except from the board of management in each nursing home.
Results {#sec1_3}
=======
Participant Characteristics {#sec2_9}
---------------------------
The mean age of the participants was 84.8 ± 7.0 years (range 67--100) and the majority were women (77.5%). The mean MMSE score was 9.0 ± 7.1. A moderate degree of communication difficulties was seen, with a mean HCS score of 22.0 ± 11.9 (Table [1](#T1){ref-type="table"}).
Drop-Outs and Attendance {#sec2_10}
------------------------
Of the 120 assessed at baseline, a total of 105 residents completed the study. In the Sonas group, 3 dropped out (2.5%; 2 left the group and 1 withdrew because of serious illness), there were 5 drop-outs in the reading group (4.1%; 5 died) and 7 in the control group (5.8%; 5 died and 2 were transferred to another nursing home), i.e., a total drop-out rate of 12.4% (Fig. [2](#F2){ref-type="fig"}).
The participants were reported to attend the Sonas programme at an average of 39.8 ± 11.8 sessions (range 1--48). In the reading group, the average attendance rate was 42.1 ± 6.9 (range 1--48).
Effect of Intervention {#sec2_11}
----------------------
The linear mixed model analysis through 3 time periods (T0, T1, and T2) did not show an overall communication effect of the Sonas programme. However, we found a significant difference in the level of change in communication ability through the whole study period between the Sonas and the reading group (*p* = 0.019) and between the Sonas and the control group (*p* = 0.001). As illustrated in Figure [3](#F3){ref-type="fig"}, the mean HCS score in the Sonas group decreased from 24.7 at T0 to 23.6 at T1 (*p* = 0.014) and 24.7 at T0 to 22.3 at T2 (*p* = 0.002), indicating a significant improvement in communication abilities. In contrast, the control group had an increase in HCS mean score from 21.9 at T0 to 23.5 at T2. An almost unchanged mean score of HCS was reported for the reading group (18.2 at T0 to 18.8 at T2). The number of sessions attended did not affect the outcome of the HCS.
An overall effect was demonstrated regarding *interest in past events, humour* and *having general knowledge,* whereas no significant differences were found in aspects which would be expected to require a higher level of cognitive function such as: *remembering names*, *speech*, *having success in communication* and *showing interest in and response to objects*. For further details, see Table [2](#T2){ref-type="table"}.
When stratifying on MMSE (moderate and severe; cut-off of 10 points), a significant difference was found between the Sonas group and the reading group among those with an MMSE score of 0--10 at T0 (*p* = 0.011) and at T1 (*p* = 0.044), while among those with an MMSE score of 11--20, there was a significant difference between the Sonas and the reading group at baseline (*p* = 0.026). No difference was found between the Sonas and the control group. There was a significant reduction in mean HCS score from T0 to T1 (*p* = 0.015 in the group with an MMSE score of 11--20 and *p* = 0.017 in the group with an MMSE score of 0--10) and from T0 to T2 (*p* = 0.009 in the group with an MMSE score of 11--20 and *p* = 0.008 in the group with an MMSE score of 0--10) in the Sonas group. No significant changes were found in the other 2 groups (Table [3](#T3){ref-type="table"}).
Discussion {#sec1_4}
==========
No overall significant effect of the Sonas programme was found. However, there was an intervention effect between the Sonas group and the reading group and between the Sonas group and the control group from T0 to T1 and T2, as well as a significant improvement in communication in the Sonas group. The reading group demonstrated almost unchanged communication ability, while the control group showed a decrease in communication ability. When exploring the subscores of the HCS, a significant improvement in communication ability was found in some of the subscores. Further, a significant effect between the Sonas and the reading group was found for participants with severe dementia after 3 months, but this effect ceased after 6 months. These findings will be discussed in the following.
Effect of the Sonas Programme on Communication {#sec2_12}
----------------------------------------------
The lack of an overall significant effect on communication ability in our study is consistent with a previous study \[[@B15]\] and could have different explanations. First of all, there is the challenge of measuring effect in people who are expected to have an increase in cognitive decline as part of the dementia progress, which also affects their communication ability \[[@B3]\]. Another aspect is the importance of tailoring the intervention to the resident\'s background and preferences \[[@B29]\], choosing music according to the person\'s cultural context \[[@B30]\] as well as likes and dislikes \[[@B31]\]. Even though our study was carried out in the culture it was developed for, using Irish poetry and music, it was not based on individual preferences, which might also explain the lack of effect. However, the interaction effect found indicates that attending the Sonas sessions could improve communication ability more than by attending the reading group or being part of the control group. It is possible that this could be explained by the multi-sensory approach of the Sonas programme, or by the fact that receiving attention by being part of a reading group is not sufficient to improve communication ability. The significant improvement in communication ability in the Sonas group and an almost unchanged mean score on the HCS in the reading group during the 24 weeks could support these findings.
The Impact on Aspects of Communication {#sec2_13}
--------------------------------------
The overall effect on *interest in past events, humour* and *having general knowledge*, and lack of effect in *remembering names, speech, interest in and response to objects* and *success in communication* could be explained by the fact that sensory stimulation does not have an effect on aspects of communication which require the ability to give a verbal response. Based on this, including persons with severe dementia, who are rarely able to express themselves verbally, demonstrate a challenge and could affect the results. However, we did not expect the Sonas programme to have an effect on *general knowledge* since the ability to communicate verbally decreased as part of the dementia process \[[@B3]\]. It was also interesting to see a significant effect on *humour*, which has been described to be altered in persons with dementia \[[@B32]\] and that the ability to understand humour depends on cognitive function \[[@B33]\]. However, the outcome would depend on the staff\'s interpretation of humour, where they might see smiling as an expression of humour although this has been described as pleasure \[[@B25]\]. It is therefore interesting to observe the significant improvement in *pleasure* in the Sonas group after 24 weeks, while the reading group remained stable and the control group decreased. An explanation for this could be that pleasure, which is a positive feeling or sensation \[[@B34]\], could be an expression of well-being, which has been reported to be preserved in people with severe dementia \[[@B35]\]. Optimizing well-being is, as described by Threadgold \[[@B24]\], one of the purposes performing the Sonas programme.
Effect on Communication When Stratified on Cognitive Function {#sec2_14}
-------------------------------------------------------------
The significant difference in communication ability between the Sonas group compared with the reading group after 12 weeks, when stratified on cognitive function, could have several explanations. First of all implementing something new to a group of people who have difficulties attending other activities offered in the nursing home could explain the effect. At the same time, there is the probability of a greater improvement in this group than for those with moderate dementia since they might be more involved in other activities.
Additionally, lack of communication improvement might be explained by the ceiling effect due to lack of reserve capacity and the fact that dementia is progressing.
Although no previous study of the Sonas programme has measured the effect by stratifying on cognitive function, the findings from our study are consistent with a previous study \[[@B36]\] where they measured the effect of robot-assisted activity on quality of life, reporting a significant effect among people with severe dementia.
The ceased significant effect for the Sonas group with severe cognitive decline, seen after 24 weeks, could be due to progression in dementia or due to the fact that the participants have become bored by attending the same programme twice a week. However, it is important to note that this group was the only group which showed a significant increase in communication ability during the whole study period whereas the other 2 groups showed a decrease.
As well as giving an indication of what aspects of communication are retained in persons with severe dementia, these findings demonstrate that the Sonas programme has a potential to improve communication ability in persons with dementia, emphasizing the importance of individually tailored interventions where the persons cognitive level is taken into account.
Strengths and Limitations {#sec2_15}
-------------------------
To our knowledge, this is the first RCT of the Sonas programme carried out in Ireland. To overcome some challenges in previous research, we included a reading group in order to rule out a possible effect of social contact by attending the Sonas group. Further, we included a larger sample based on a power calculation. The study was also conducted over a longer period of time compared to a previous study with the same main outcome \[[@B15]\], with a low drop-out rate and high attendance rate over 24 weeks. Another strength was that we focused on communication outcome, which is the overall aim of the Sonas programme. Even though we used a psychometrically tested instrument to assess communication, the HCS is assessing communication deficit and such a negative approach is not in line with the philosophy of the Sonas programme, where the importance is to focus on communication abilities. Further, the subscores in the HCS focus mainly on verbal communication and aspects of communication which require verbal response from the participants that might give a lower outcome on the total score for persons with severe dementia. In order to overcome some of these challenges, we explored the subscores as well as stratified the total score on cognitive function.
Even though all nursing homes were registered with HIQA and therefore had a certain standard, there will always be differences. A limitation is that we did not have control over the participant\'s daily routine, other activities or stimuli or the use of psychotropic medication, which can affect the outcome. All 6 centres offered some activities, ranging from a singing group once a week to daily activity programme. We assume that the qualification of staff differed among the nursing homes as well as the involvement from the managers. Although all Sonas licensed practitioners had undergone a 3-day formal training, there is no guarantee that the Sonas sessions were conducted in the same way. They can, for example, have different involvement and suitability, or even the way they were prepared for the sessions. Further, the room where the sessions were carried out, the time of the day and if they were interrupted during the sessions would all influence the result.
Although the assessors where blinded at baseline, this was not possible at the 12 and 24 weeks' assessment. However, in order to minimize the risk of contamination, this article\'s first author collected data after each assessment in order to prevent the assessors from seeing what they scored the previous time. Another limitation was that we were not able to get the same number of groups in all 6 nursing homes, which made the randomization process more difficult. Another consequence of using 6 different nursing homes is the use of 12 different assessors. However, in order to overcome some of the challenges, they underwent a 2-h training about how to carry out the assessment beforehand.
An attention placebo control group with just human presence and no other stimuli might have given other results. However, we found offering human presence without any form of additional stimuli to be unethical and therefore a reading group was chosen as an attention placebo control group.
As the intervention was implemented for people with moderate to severe dementia, the result indicates that this group could benefit from the programme. However, it is not possible to generalize the findings.
Conclusion {#sec1_5}
==========
To our knowledge, this is the largest clinical trial examining the effect of the Sonas programme, and the first performed in Ireland. Even though this study failed to document an overall effect of the Sonas programme on communication, the findings indicate that the Sonas programme has a significant effect on those with severe dementia, as well as some aspects of communication.
Further studies are needed to look at the immediate effect during and after sessions, which could be done by the use of Dementia Care Mapping, where the different aspects of communication could be further explored.
Disclosure Statement {#sec1_6}
====================
The authors declare that they have no conflicts of interest.
We extend our sincere thanks to the participants and the Sonas licensed practitioners in each nursing home. We would also like to thank Sister Morag Collins SJC for her detailed proof reading.
{#F1}
{#F2}
{#F3}
######
Characteristics of participants at baseline
Characteristics Total (*n* = 120) Sonas group (*n* = 48) Reading group (*n* = 32) Control group (*n* = 40)
------------------------------- ------------------- ------------------------ -------------------------- --------------------------
Gender
Women 93 (77.5) 40 (83.3) 25 (78.1) 28 (70)
Men 27 (22.5) 8 (16.7) 7 (21.9) 12 (30)
Marital status
Single 36 (30) 12 (25.0) 11 (34.4) 13 (32.5)
Married 22 (18.3) 8 (16.7) 6 (18.8) 8 (20.0)
Widowed 61 (50.8) 27 (56.3) 15 (46.9) 19 (47.5)
Divorced 1 (0.8) 1 (2.1) 0 (0) 0 (0)
Degree of cognitive function
Severe (MMSE score 0--10) 63 (52.5) 29 (60.4) 15 (46.9) 19 (47.5)
Moderate (MMSE score 11--20) 57 (47.5) 19 (39.6) 17 (53.1) 21 (52.5)
Age, years 84.8±7.0 86.1±6.6 84.9±7.2 83.3±7.1
MMSE score 9.0±7.1 7.7±6.7 10.0±7.2 9.7±7.5
HCS score 22.0±11.9 24.7±12.1 18.2±12.0 21.9±10.9
Values are mean ± standard deviation or *n* (%). MMSE, Mini-Mental State Examination; HCS, Holden Communication Scale.
######
HCS subtype means and standard deviations
Sonas group (*n* = 48) Reading group (*n* = 32) Control group (*n* = 40) Sonas vs. reading, *p* value Sonas vs. control, *p* value
---------------------------------- ------------------------ -------------------------- -------------------------- ------------------------------ ------------------------------
Response
Baseline 1.8 (1.4) 1.5 (1.3) 1.7 (1.4) 0.678 **0.032**
12 weeks 1.7 (1.3) 1.4 (1.2) 1.8 (1.4)
24 weeks 1.7 (1.3) 1.4 (1.3) 1.8 (1.5)
Interest in past events
Baseline **2.4 (1.1)** 1.8 (1.2) 2.1 (1.1) **0.026** **0.010**
12 weeks **2.3 (1.2)** 1.8 (1.1) 2.0 (1.1)
24 weeks **2.2 (1.2)** 1.9 (1.2) 3.0 (1.1)
Pleasure
Baseline **1.5 (0.9)** 1.2 (1.0) 1.5 (1.0) **0.035** 0.232
12 weeks 1.5 (1.0) 1.2 (1.1) 1.5 (0.9)
24 weeks **1.3 (0.8)** 1.2 (0.9) 1.2 (0.9)
Humour
Baseline **2.0 (1.2)** 1.4 (1.1) 1.8 (1.3) **0.018** **0.013**
12 weeks **2.0 (1.2)** 1.3 (1.1) 1.8 (1.2)
24 weeks **1.9 (1.2)** 1.6 (1.2) 1.8 (1.2)
Remembering names
Baseline 2.4 (1.3) 1.8 (1.1) 1.9 (1.2) 0.486 0.413
12 weeks 2.4 (1.1) 2.0 (1.4) 1.9 (1.0)
24 weeks 2.3 (1.1) 1.8 (1.2) 1.9 (1.0)
General orientation
Baseline 3.0 (1.2) **2.3 (1.3)** 2.9 (1.1) **0.047** 0.131
12 weeks 2.1 (1.0) **2.6 (1.7)** 3.0 (1.0)
24 weeks 3.1 (1.0) **2.7 (1.2)** 3.0 (1.0)
General knowledge
Baseline **2.8 (1.0)** 2.2 (1.0) 2.6 (1.1) **0.019** **0.022**
12 weeks **2.8 (0.9)** 2.2 (1.0) 2.6 (1.1)
24 weeks **2.6 (1.0)** 2.3 (1.1) 2.6 (1.0)
Ability to join in games
Baseline 2.1 (1.3) 2.0 (1.6) 2.2 (1.3) 0.893 **0.015**
12 weeks 2.0 (1.2) 1.8 (1.2) 1.5 (1.4)
24 weeks 2.0 (1.3) 1.6 (1.2) 2.3 (1.3)
Speech
Baseline **1.7 (1.5)** 1.1 (1.6) 1.1 (1.4) 0.319 0.063
12 weeks **1.6 (1.4)** 0.9 (1.4) 1.2 (1.5)
24 weeks **1.6 (1.3)** 0.9 (1.4) 1.0 (1.4)
Attempts at communication
Baseline **1.5 (1.5)** 0.8 (1.3) 1.2 (1.4) **0.024** 0.086
12 weeks **1.4 (1.5)** 0.8 (1.3) 1.0 (1.3)
24 weeks **1.4 (1.4)** 0.9 (1.4) 1.0 (1.4)
Interest and response to objects
Baseline **1.7 (1.3)** 1.1 (1.2) 1.7 (1.1) 0.075 0.079
12 weeks **1.5 (1.2)** 1.1 (1.1) 1.7 (1.2)
24 weeks **1.7 (1.4)** 1.1 (1.2) 1.4 (1.1)
Success in communication
Baseline 1.7 (1.4) 1.0 (1.1) 1.1 (1.4) 0.641 0.252
12 weeks 1.8 (1.5) 0.8 (1.3) 0.9 (1.3)
24 weeks 1.6 (1.5) 0.8 (1.3) 0.8 (1.2)
Figures are means with standard deviations in parentheses. Figures in bold indicate a significant effect.
######
Ratings of the ability to communicate across time by groups stratified on the level of cognition (MMSE)
Time Sonas group Reading group Control group Sonas vs. reading, *p* value Sonas vs. control, *p* value
------------------- ------------------- ------------------- ------------------- ------------------------------ ------------------------------
MMSE score 11--20
Baseline 23.9 (20.6; 27.3) 18.0 (14.0; 22.0) 20.4 (16.8; 24.1) **0.026** 0.380
12 weeks 22.7 (19.3; 26.1) 18.1 (14.0; 22.2) 20.9 (17.1; 24.6) 0.092 0.325
24 weeks 21.5 (17.9; 25.1) 18.3 (13.9; 22.8) 21.4 (17.4; 25.4) 0.267 0.301
MMSE score 0--10
Baseline 25.2 (22.0; 28.5) 18.6 (14.5; 22.6) 23.3 (19.7; 27.0) **0.011** 0.081
12 weeks 24.1 (20.8; 27.4) 18.8 (14.8; 22.8) 23.9 (20.3; 27.5) **0.044** 0.061
24 weeks 23.0 (19.5; 26.5) 19.1 (14.8; 23.3) 24.6 (20.7; 28.4) 0.156 0.057
MMSE score 11--20
0 vs. 12 weeks **0.015** 0.867 0.411
0 vs. 24 weeks **0.009** 0.792 0.319
12 vs. 24 weeks **0.015** 0.867 0.411
MMSE score 0--10
0 vs. 12 weeks **0.017** 0.698 0.283
0 vs. 24 weeks **0.008** 0.608 0.194
12 vs. 24 weeks **0.017** 0.698 0.283
Figures are means with 95% CI in parentheses. Figures in bold indicate a significant effect.
| {
"pile_set_name": "PubMed Central"
} |
Related literature {#sec1}
====================
For the pharmacological activity of sulfonamide derivatives, see: Bouissane *et al.* (2006[@bb3]); Mustafa *et al.* (2012[@bb7]); Lopez *et al.* (2010[@bb6]). For similar compounds, see: Abbassi *et al.* (2012[@bb1], 2013[@bb2]).
Experimental {#sec2}
==============
{#sec2.1}
### Crystal data {#sec2.1.1}
C~15~H~15~N~3~O~2~S*M* *~r~* = 301.36Monoclinic,*a* = 8.0026 (3) Å*b* = 12.8195 (4) Å*c* = 14.1321 (4) Åβ = 91.602 (2)°*V* = 1449.24 (8) Å^3^*Z* = 4Mo *K*α radiationμ = 0.23 mm^−1^*T* = 296 K0.43 × 0.36 × 0.28 mm
### Data collection {#sec2.1.2}
Bruker X8 APEX diffractometerAbsorption correction: multi-scan (*SADABS*; Bruker, 2009[@bb4]) *T* ~min~ = 0.960, *T* ~max~ = 0.99217896 measured reflections4048 independent reflections2703 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.047
### Refinement {#sec2.1.3}
*R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.046*wR*(*F* ^2^) = 0.134*S* = 1.024048 reflections190 parametersH-atom parameters constrainedΔρ~max~ = 0.25 e Å^−3^Δρ~min~ = −0.32 e Å^−3^
{#d5e448}
Data collection: *APEX2* (Bruker, 2009[@bb4]); cell refinement: *SAINT* (Bruker, 2009[@bb4]); data reduction: *SAINT*; program(s) used to solve structure: *SHELXS97* (Sheldrick, 2008[@bb8]); program(s) used to refine structure: *SHELXL97* (Sheldrick, 2008[@bb8]); molecular graphics: *ORTEP-3 for Windows* (Farrugia, 2012[@bb5]); software used to prepare material for publication: *PLATON* (Spek, 2009[@bb9]) and *publCIF* (Westrip, 2010[@bb10]).
Supplementary Material
======================
Crystal structure: contains datablock(s) I. DOI: [10.1107/S1600536813023398/bh2482sup1.cif](http://dx.doi.org/10.1107/S1600536813023398/bh2482sup1.cif)
Structure factors: contains datablock(s) I. DOI: [10.1107/S1600536813023398/bh2482Isup2.hkl](http://dx.doi.org/10.1107/S1600536813023398/bh2482Isup2.hkl)
######
Click here for additional data file.
Supplementary material file. DOI: [10.1107/S1600536813023398/bh2482Isup3.cml](http://dx.doi.org/10.1107/S1600536813023398/bh2482Isup3.cml)
Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?bh2482&file=bh2482sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?bh2482sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?bh2482&checkcif=yes)
Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [BH2482](http://scripts.iucr.org/cgi-bin/sendsup?bh2482)).
The authors thank the Unit of Support for Technical and Scientific Research (UATRS, CNRST) for the X-ray measurements.
1. Comment
==========
Sulfonamide derivatives are well known pharmaceutical agents since this group has been the main functional part of many drug structures, due to stability and tolerance in human beings. These compounds exhibit a wide range of biological activities, such as anticancer, anti-inflammatory, and antiviral functions (Bouissane *et al.*, 2006; Mustafa *et al.*, 2012; Lopez *et al.*, 2010). The present work is a continuation of the investigation on sulfonamide derivatives published recently by our team (Abbassi *et al.*, 2012, 2013).
The molecule of 4-methyl-*N*-(1-methyl-1*H*-indazol-5-yl)-benzenesulfonamide is built up from two fused five- and six-membered rings (N2/N3/C1 to C7) linked to the benzenesulfonamide group, as shown in Fig. 1. The fused rings system is almost planar, with the maximum deviation of 0.030 (2) Å arising from atom C1. Moreover, the dihedral angle between the indazole system and the plan through the atoms forming the benzene ring (C9 to C14) is 48.84 (9)°.
In the crystal, the molecules are interconnected through C3---H3···O2*^ii^* weak contacts and N1---H1···N2*^i^* hydrogen bonds, forming a two-dimensional network (Fig. 2 and Table 2; symmetry codes: (*i*) -*x* + 1, *y* + 1/2, -*z* + 1/2; (*ii*) *x* + 1, *y*, *z*).
2. Experimental {#experimental}
===============
A mixture of 1-methyl-5-nitroindazole (1.22 mmol) and anhydrous SnCl~2~ (1.1 g, 6.1 mmol) in 25 ml of absolute ethanol was heated at 333 K for 6 h. After reduction, the starting material disappeared, and the solution was allowed to cool down. The pH was made slightly basic (pH 7--8) by addition of 5% aqueous potassium bicarbonate before extraction with ethyl acetate. The organic phase was washed with brine and dried over magnesium sulfate. The solvent was removed to afford the amine, which was immediately dissolved in pyridine (5 ml) and then reacted with 4-methylbenzenesulfonyl chloride (1.25 mmol) at room temperature for 24 h. After the reaction mixture was concentrated *in vacuo*, the resulting residue was purified by flash chromatography (eluted with ethyl acetate/hexane 1:9). The title compound was recrystallized from acetone.
3. Refinement {#refinement}
=============
H atoms were located in a difference map, but C-bound H atoms were placed in idealized positions and treated as riding, with C---H = 0.96 Å, and C---H = 0.93 Å for methyl and aromatic CH, respectively. Atom H1 was first refined freely, and then fixed (N1---H1 = 0.8759 Å). All H atoms were refined with isotropic displacement parameters fixed as *U*~iso~(H) = 1.2*U*~eq~(C-aromatic, NH) or *U*~iso~(H) = 1.5*U*~eq~(*C*-methyl).
Figures
=======
{#Fap1}
{#Fap2}
Crystal data {#tablewrapcrystaldatalong}
============
------------------------- ---------------------------------------
C~15~H~15~N~3~O~2~S *F*(000) = 632
*M~r~* = 301.36 *D*~x~ = 1.381 Mg m^−3^
Monoclinic, *P*2~1~/*c* Mo *K*α radiation, λ = 0.71073 Å
Hall symbol: -P 2ybc Cell parameters from 4048 reflections
*a* = 8.0026 (3) Å θ = 2.9--29.6°
*b* = 12.8195 (4) Å µ = 0.23 mm^−1^
*c* = 14.1321 (4) Å *T* = 296 K
β = 91.602 (2)° Block, colourless
*V* = 1449.24 (8) Å^3^ 0.43 × 0.36 × 0.28 mm
*Z* = 4
------------------------- ---------------------------------------
Data collection {#tablewrapdatacollectionlong}
===============
------------------------------------------------------------ --------------------------------------
Bruker X8 APEX diffractometer 4048 independent reflections
Radiation source: fine-focus sealed tube 2703 reflections with *I* \> 2σ(*I*)
Graphite monochromator *R*~int~ = 0.047
φ and ω scans θ~max~ = 29.6°, θ~min~ = 2.9°
Absorption correction: multi-scan (*SADABS*; Bruker, 2009) *h* = −11→11
*T*~min~ = 0.960, *T*~max~ = 0.992 *k* = −17→17
17896 measured reflections *l* = −19→19
------------------------------------------------------------ --------------------------------------
Refinement {#tablewraprefinementdatalong}
==========
------------------------------------- -------------------------------------------------------------------------------------------------
Refinement on *F*^2^ Primary atom site location: structure-invariant direct methods
Least-squares matrix: full Secondary atom site location: difference Fourier map
*R*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.046 Hydrogen site location: inferred from neighbouring sites
*wR*(*F*^2^) = 0.134 H-atom parameters constrained
*S* = 1.02 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0633*P*)^2^ + 0.2552*P*\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3
4048 reflections (Δ/σ)~max~ \< 0.001
190 parameters Δρ~max~ = 0.25 e Å^−3^
0 restraints Δρ~min~ = −0.32 e Å^−3^
0 constraints
------------------------------------- -------------------------------------------------------------------------------------------------
Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å^2^) {#tablewrapcoords}
==================================================================================================
------ --------------- --------------- --------------- -------------------- --
*x* *y* *z* *U*~iso~\*/*U*~eq~
C1 0.31920 (19) 0.13913 (14) 0.21183 (12) 0.0348 (4)
C2 0.4588 (2) 0.18535 (14) 0.16945 (13) 0.0386 (4)
H2 0.4693 0.2576 0.1706 0.046\*
C3 0.5789 (2) 0.12762 (14) 0.12688 (13) 0.0381 (4)
H3 0.6709 0.1587 0.0997 0.046\*
C4 0.55641 (19) 0.01961 (13) 0.12629 (12) 0.0336 (4)
C5 0.4161 (2) −0.02812 (13) 0.16553 (12) 0.0352 (4)
C6 0.2952 (2) 0.03317 (13) 0.20985 (13) 0.0361 (4)
H6 0.2022 0.0029 0.2368 0.043\*
C7 0.4424 (2) −0.13606 (15) 0.15117 (14) 0.0448 (4)
H7 0.3685 −0.1880 0.1690 0.054\*
C8 0.8127 (2) −0.04980 (17) 0.04713 (14) 0.0486 (5)
H8A 0.8953 −0.0912 0.0803 0.073\*
H8B 0.8477 0.0218 0.0470 0.073\*
H8C 0.8000 −0.0743 −0.0169 0.073\*
C9 0.1242 (2) 0.36208 (14) 0.13664 (13) 0.0392 (4)
C10 0.1538 (3) 0.35014 (17) 0.04145 (16) 0.0561 (5)
H10 0.1315 0.2868 0.0116 0.067\*
C11 0.2163 (3) 0.43217 (19) −0.00880 (16) 0.0616 (6)
H11 0.2356 0.4234 −0.0729 0.074\*
C12 0.2515 (3) 0.52718 (17) 0.03268 (16) 0.0523 (5)
C13 0.2195 (3) 0.53815 (18) 0.12835 (17) 0.0617 (6)
H13 0.2415 0.6016 0.1580 0.074\*
C14 0.1560 (3) 0.45732 (16) 0.17998 (15) 0.0518 (5)
H14 0.1345 0.4664 0.2438 0.062\*
C15 0.3195 (3) 0.6169 (2) −0.0239 (2) 0.0802 (8)
H15A 0.3339 0.5951 −0.0881 0.120\*
H15B 0.4253 0.6383 0.0032 0.120\*
H15C 0.2425 0.6742 −0.0227 0.120\*
N1 0.20248 (17) 0.20314 (12) 0.26172 (11) 0.0393 (3)
H1 0.2468 0.2458 0.3038 0.047\*
N2 0.5857 (2) −0.15374 (12) 0.10917 (12) 0.0471 (4)
N3 0.65480 (18) −0.05848 (12) 0.09380 (11) 0.0391 (3)
O1 −0.05599 (17) 0.30214 (12) 0.27498 (12) 0.0624 (4)
O2 −0.02085 (16) 0.18164 (11) 0.13971 (11) 0.0572 (4)
S1 0.04597 (5) 0.25800 (4) 0.20396 (4) 0.04273 (16)
------ --------------- --------------- --------------- -------------------- --
Atomic displacement parameters (Å^2^) {#tablewrapadps}
=====================================
----- ------------- ------------- ------------- -------------- -------------- --------------
*U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^
C1 0.0324 (7) 0.0340 (9) 0.0381 (9) 0.0007 (6) 0.0028 (6) 0.0028 (7)
C2 0.0387 (8) 0.0284 (9) 0.0491 (11) −0.0039 (7) 0.0045 (7) 0.0023 (8)
C3 0.0336 (8) 0.0340 (10) 0.0469 (10) −0.0041 (7) 0.0073 (7) 0.0040 (8)
C4 0.0314 (7) 0.0335 (9) 0.0358 (9) 0.0026 (6) −0.0003 (6) 0.0017 (7)
C5 0.0345 (8) 0.0307 (9) 0.0404 (10) −0.0014 (6) −0.0010 (7) 0.0040 (7)
C6 0.0312 (7) 0.0330 (9) 0.0441 (10) −0.0039 (6) 0.0041 (7) 0.0060 (8)
C7 0.0462 (10) 0.0309 (10) 0.0574 (12) −0.0021 (7) 0.0041 (8) 0.0041 (9)
C8 0.0448 (10) 0.0532 (13) 0.0483 (11) 0.0088 (8) 0.0113 (8) −0.0031 (10)
C9 0.0359 (8) 0.0353 (10) 0.0461 (10) 0.0035 (7) −0.0006 (7) −0.0016 (8)
C10 0.0714 (13) 0.0429 (12) 0.0544 (13) −0.0043 (10) 0.0093 (10) −0.0120 (10)
C11 0.0802 (16) 0.0564 (15) 0.0488 (13) −0.0019 (12) 0.0119 (11) 0.0001 (11)
C12 0.0515 (11) 0.0466 (13) 0.0586 (14) −0.0009 (9) −0.0033 (9) 0.0106 (10)
C13 0.0813 (16) 0.0415 (13) 0.0619 (15) −0.0138 (11) −0.0080 (12) −0.0026 (11)
C14 0.0674 (13) 0.0420 (12) 0.0456 (11) −0.0050 (9) −0.0026 (9) −0.0027 (9)
C15 0.0803 (17) 0.0709 (18) 0.0893 (19) −0.0147 (14) −0.0032 (14) 0.0316 (16)
N1 0.0382 (7) 0.0352 (8) 0.0448 (9) 0.0009 (6) 0.0081 (6) −0.0017 (7)
N2 0.0510 (9) 0.0323 (9) 0.0582 (10) 0.0030 (7) 0.0029 (7) −0.0006 (8)
N3 0.0386 (7) 0.0352 (9) 0.0436 (9) 0.0038 (6) 0.0033 (6) −0.0012 (7)
O1 0.0494 (8) 0.0546 (10) 0.0848 (11) 0.0102 (6) 0.0318 (8) 0.0029 (8)
O2 0.0403 (7) 0.0446 (8) 0.0862 (11) −0.0080 (6) −0.0079 (7) −0.0064 (8)
S1 0.0311 (2) 0.0361 (3) 0.0614 (3) 0.00089 (16) 0.00821 (18) −0.0010 (2)
----- ------------- ------------- ------------- -------------- -------------- --------------
Geometric parameters (Å, º) {#tablewrapgeomlong}
===========================
----------------------- -------------- ---------------------- --------------
C1---C6 1.372 (2) C9---C14 1.387 (3)
C1---C2 1.412 (2) C9---S1 1.7635 (19)
C1---N1 1.442 (2) C10---C11 1.371 (3)
C2---C3 1.366 (2) C10---H10 0.9300
C2---H2 0.9300 C11---C12 1.377 (3)
C3---C4 1.396 (2) C11---H11 0.9300
C3---H3 0.9300 C12---C13 1.390 (3)
C4---N3 1.361 (2) C12---C15 1.510 (3)
C4---C5 1.406 (2) C13---C14 1.373 (3)
C5---C6 1.407 (2) C13---H13 0.9300
C5---C7 1.415 (3) C14---H14 0.9300
C6---H6 0.9300 C15---H15A 0.9600
C7---N2 1.325 (2) C15---H15B 0.9600
C7---H7 0.9300 C15---H15C 0.9600
C8---N3 1.446 (2) N1---S1 1.6348 (15)
C8---H8A 0.9600 N1---H1 0.8759
C8---H8B 0.9600 N2---N3 1.361 (2)
C8---H8C 0.9600 O1---S1 1.4280 (14)
C9---C10 1.381 (3) O2---S1 1.4288 (15)
C6---C1---C2 121.25 (15) C9---C10---H10 120.2
C6---C1---N1 118.75 (14) C10---C11---C12 122.1 (2)
C2---C1---N1 119.96 (15) C10---C11---H11 119.0
C3---C2---C1 122.27 (16) C12---C11---H11 119.0
C3---C2---H2 118.9 C11---C12---C13 117.5 (2)
C1---C2---H2 118.9 C11---C12---C15 121.4 (2)
C2---C3---C4 116.61 (15) C13---C12---C15 121.1 (2)
C2---C3---H3 121.7 C14---C13---C12 121.4 (2)
C4---C3---H3 121.7 C14---C13---H13 119.3
N3---C4---C3 130.97 (16) C12---C13---H13 119.3
N3---C4---C5 106.78 (15) C13---C14---C9 119.7 (2)
C3---C4---C5 122.23 (15) C13---C14---H14 120.1
C4---C5---C6 119.96 (16) C9---C14---H14 120.1
C4---C5---C7 104.22 (15) C12---C15---H15A 109.5
C6---C5---C7 135.76 (16) C12---C15---H15B 109.5
C1---C6---C5 117.65 (15) H15A---C15---H15B 109.5
C1---C6---H6 121.2 C12---C15---H15C 109.5
C5---C6---H6 121.2 H15A---C15---H15C 109.5
N2---C7---C5 111.41 (16) H15B---C15---H15C 109.5
N2---C7---H7 124.3 C1---N1---S1 119.89 (12)
C5---C7---H7 124.3 C1---N1---H1 115.6
N3---C8---H8A 109.5 S1---N1---H1 111.2
N3---C8---H8B 109.5 C7---N2---N3 106.19 (15)
H8A---C8---H8B 109.5 N2---N3---C4 111.38 (14)
N3---C8---H8C 109.5 N2---N3---C8 120.41 (15)
H8A---C8---H8C 109.5 C4---N3---C8 128.20 (16)
H8B---C8---H8C 109.5 O1---S1---O2 120.46 (9)
C10---C9---C14 119.56 (19) O1---S1---N1 105.33 (9)
C10---C9---S1 120.99 (15) O2---S1---N1 106.88 (8)
C14---C9---S1 119.44 (15) O1---S1---C9 107.32 (9)
C11---C10---C9 119.6 (2) O2---S1---C9 107.92 (9)
C11---C10---H10 120.2 N1---S1---C9 108.45 (8)
C6---C1---C2---C3 −1.9 (3) C12---C13---C14---C9 0.6 (4)
N1---C1---C2---C3 175.55 (16) C10---C9---C14---C13 −1.2 (3)
C1---C2---C3---C4 0.6 (3) S1---C9---C14---C13 178.93 (17)
C2---C3---C4---N3 −176.89 (18) C6---C1---N1---S1 −95.37 (17)
C2---C3---C4---C5 1.3 (3) C2---C1---N1---S1 87.17 (18)
N3---C4---C5---C6 176.56 (15) C5---C7---N2---N3 −0.9 (2)
C3---C4---C5---C6 −2.0 (3) C7---N2---N3---C4 0.3 (2)
N3---C4---C5---C7 −0.94 (19) C7---N2---N3---C8 −179.02 (17)
C3---C4---C5---C7 −179.50 (16) C3---C4---N3---N2 178.82 (18)
C2---C1---C6---C5 1.1 (3) C5---C4---N3---N2 0.4 (2)
N1---C1---C6---C5 −176.33 (15) C3---C4---N3---C8 −1.9 (3)
C4---C5---C6---C1 0.7 (3) C5---C4---N3---C8 179.70 (17)
C7---C5---C6---C1 177.3 (2) C1---N1---S1---O1 171.99 (13)
C4---C5---C7---N2 1.2 (2) C1---N1---S1---O2 42.75 (15)
C6---C5---C7---N2 −175.7 (2) C1---N1---S1---C9 −73.37 (14)
C14---C9---C10---C11 0.8 (3) C10---C9---S1---O1 −147.78 (17)
S1---C9---C10---C11 −179.30 (18) C14---C9---S1---O1 32.12 (18)
C9---C10---C11---C12 0.2 (4) C10---C9---S1---O2 −16.55 (18)
C10---C11---C12---C13 −0.8 (4) C14---C9---S1---O2 163.35 (15)
C10---C11---C12---C15 −179.7 (2) C10---C9---S1---N1 98.89 (17)
C11---C12---C13---C14 0.4 (4) C14---C9---S1---N1 −81.21 (17)
C15---C12---C13---C14 179.3 (2)
----------------------- -------------- ---------------------- --------------
Hydrogen-bond geometry (Å, º) {#tablewraphbondslong}
=============================
------------------ --------- --------- ----------- ---------------
*D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A*
N1---H1···N2^i^ 0.88 2.21 3.065 (2) 166
C3---H3···O2^ii^ 0.93 2.53 3.277 (2) 137
------------------ --------- --------- ----------- ---------------
Symmetry codes: (i) −*x*+1, *y*+1/2, −*z*+1/2; (ii) *x*+1, *y*, *z*.
###### Hydrogen-bond geometry (Å, °)
*D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A*
---------------- --------- ------- ----------- -------------
N1---H1⋯N2^i^ 0.88 2.21 3.065 (2) 166
C3---H3⋯O2^ii^ 0.93 2.53 3.277 (2) 137
Symmetry codes: (i) ; (ii) .
| {
"pile_set_name": "PubMed Central"
} |
{.299-b1}
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction
===============
According to World Health Organization data, migraine has become the third most common disease in the 21st century.^\[[@R1]\]^ Meanwhile, migraine is one of the top 10 causes of disability in the world.^\[[@R2],[@R3]\]^ Migraine has generally increased in incidence worldwide in recent years, especially in developing countries, possibly with adverse lifestyle changes brought about by rapid urbanization in these regions.^\[[@R4]--[@R6]\]^ Migraine is characterized by moderate to severe headache with fatigue, depression, hyperactivity, nausea, sensitivity to light or sound and other neurological symptoms.^\[[@R7],[@R8]\]^ The international classification of headaches divides migraine into 2 main types: migraine without aura and migraine with aura.^\[[@R9]\]^ There is no consensus on the pathogenesis of migraine in the medical community, but it is generally accepted that migraine is caused by the combined involvement of nerves and blood vessels.^\[[@R10]--[@R12]\]^ Cortical spreading depression (CSD) has been known to play an important role in the pathogenesis of migraine, which can activate and sensitize the trigeminal vascular system (TGVS), then triggers migraine-associated neurological and vascular responses, and finally induces pain.^\[[@R11],[@R13]\]^ Although migraine is a transient phenomenon of cerebral vasoconstriction, the chronicity of migraine may lead to retinal structural abnormalities.^\[[@R14],[@R15]\]^ There is evidence that ganglion cell death in migraine patients may be secondary to alterations in the microcirculation of the optic nerve head or even in the quality of retinal perfusion.^\[[@R16],[@R17]\]^ Optical coherence tomography (OCT) is a rapid, reproducible, and economical imaging technique for high-resolution quantitative assessment of the retina and retinal nerve fiber layer (RNFL).^\[[@R18]--[@R20]\]^ Similar to intravascular ultrasound, OCT uses near-infrared light to generate cross-sectional vascular images and it has been shown to be reliable and reproducible. Since the official commercialization of OCT technology in 2002, an alarming number of literatures have used OCT technology to study optic neuropathy. OCT have been developed to help diagnose neuro-ophthalmic diseases and detect disease development.^\[[@R21],[@R22]\]^ OCT technology continues to evolve, and spectral-domain OCT (SD-OCT) has replaced time-domain OCT (TD-OCT) as the first choice for ophthalmic OCT instruments.^\[[@R23]\]^ In contrast to TD-OCT, SD-OCT uses ultra-fast frequency scanning light source, with faster scanning speed, higher sensitivity, and superior high resolution, especially for more accurate segmentation of the retinal layer.^\[[@R24]--[@R26]\]^
In recent years, many studies have evaluated the changes of RNFL thickness in patients with migraine compared with healthy subjects. Most studies reported a decrease of RNFL thickness in migraines,^\[[@R27],[@R28]\]^ while only a few articles yielded an increase of RNFL thickness.^\[[@R29],[@R30]\]^ A previous meta-analysis evaluated the relationship between migraine and OCT-measured RNFL thickness, but we also found some issues worthy of further exploration.^\[[@R31]\]^ First, the number of included studies was 6 and the latest one included in their meta-analysis was published in 2014. Since then, more eligible studies regarding this topic have been published. Second, the influence of OCT instruments on RNFL thickness measurements was not analyzed. As mentioned earlier, SD-OCT is able to segment the retinal layer more accurately than TD-OCT. RNFL measurement data are affected by the OCT instrument.^\[[@R32]\]^ Third, due to the small number of studies, regression analysis was not performed to investigate the potential effects of age, gender, disease duration, attack frequency, pain intensity, and intraocular pressure on RNFL. Therefore, we will perform an update meta-analysis, and add new subgroup analyses and regression analyses to re-evaluate the relationship between RNFL thickness and migraine.
2. Methods
==========
The protocol of this study was registered on the International Platform of Registered Systematic Review and Meta-Analysis Protocols (INPLASY) and the registration number is INPLASY202060033 (URL = <https://inplasy.com/inplasy-2020-6-0033/>). The preferred reporting items for systematic reviews and meta-analysis protocols (PRISMA-P) statement was the guideline during the design of this study.^\[[@R33]\]^
2.1. Eligibility criteria for study selection
---------------------------------------------
### 2.1.1. Types of studies
We will only select case-control studies using OCT to measure the RNFL thickness in migraine and healthy controls. The language included in the literature is limited to English. Animal studies, abstracts, letters, reviews and case-studies will be excluded.
### 2.1.2. Types of participants
We will include articles on patients older than 18 years with normal visual fields who have been diagnosed with migraine according to the International Headache Disease Classification. Patients will be excluded if they meet the following exclusion criteria:
1. incorporate any form of glaucoma, optic nerve disease, or intraocular surgical intervention;
2. diabetes with evidence of retinopathy such as hemorrhage, hard and/or soft exudate, macular edema;
3. neurological disorders that may affect RNFL thickness;
4. children (under 18 years of age).
### 2.1.3. Types of interventions
We will include all studies that use optical coherence tomography (OCT) to evaluate the effects of migraine on the thickness of the mean and segmental RNFL. The control group will use healthy people who do not suffer from migraine.
### 2.1.4. Types of outcomes
The main outcome of this review is the difference in mean RNFL thickness and segmental RNFL thickness between migraine patients and health controls.
2.2. Search method and strategy
-------------------------------
We will search the electronic databases PubMed, EMBASE and Web of Science. The primary search strategies are: ("optical coherence tomography" OR "retina nerve fiber layer" OR "RNFL") AND "migraine". The date of the last search is set at 18 March, 2020. Additionally, the reference lists of relevant reviews and the articles selected for inclusion will be manually searched.
2.3. Data collection and analysis
---------------------------------
### 2.3.1. Study selection
Data screening and extraction will be performed using Endnote X9 and Excel 2016. Two authors will check the title and abstract of the initially retrieved article to exclude duplicate and irrelevant research. Then the full text of the remaining articles will be read to further screen out the documents that meet the predetermined eligibility criteria. If there is a dispute between the 2 authors, a third researcher will join the discussion until consensus is reached. The process of study selection is fully provided in the following PRISMA flow diagram in Figure [1](#F1){ref-type="fig"}.
{#F1}
### 2.3.2. Data extraction
Two reviewers will independently extract and fill out outcome measures for eligible studies in the excel data extraction form. The information extracted from the studies selected will include: study setting (first author, publication year), participants' characteristics (age, gender, frequency of attacks, migraine disability assessment scores (MIDAS), duration of disease, and OCT model), and main outcomes.
2.4. Risk of bias assessment
----------------------------
The risk of bias for each eligible study will be evaluated by 2 reviewers using the Newcastle-Ottawa Scale (NOS).^\[[@R34]\]^ The tool contains eight assessment indicators, which are divided into 3 aspects: selection (4 items), comparability (1 item), and outcome (3 items). The scoring results will be presented on a table and we will assess the risk of bias in eligible studies.
2.5. Assessment of heterogeneity
--------------------------------
The heterogeneity of data in each literature will be assessed by I^2^ test. If I^2^ ≤ 50%, the fixed effect model will be applied to synthesize the data, while I^2^ \> 50% will be considered as large heterogeneity of the trial. The random effect model will be used, and the source of heterogeneity will be further analyzed by analysis, sensitivity analysis and regression analysis.
2.6. Statistical analysis
-------------------------
### 2.6.1. Subgroup analysis
When the heterogeneity is high and there is sufficient data, we will conduct subgroup analysis based on patient characteristics, instrument type, and test results to obtain an objective conclusion, such as migraine with and without aura, different segmental RNFL, SD-OCT and TD-OCT.
### 2.6.2. Sensitivity analysis
If necessary, we will perform sensitivity analysis to determine the robustness of the results, and to detect whether there are trials with high risk of bias accounting for a large proportion of the heterogeneity. If high-risk studies are deleted, a meta-analysis will be performed again, and the results will be discussed accordingly.
### 2.6.3. Meta-regression analysis
Meta-regression analysis will be used to assess the impact of a series of influencing factors such as age, gender, disease duration, attack frequency, pain intensity, and intraocular pressure on the outcomes.
2.7. Publication bias
---------------------
When more than 10 eligible trials are available for analysis, we will perform an Egger test using Stata 15 software to analyze potential publication bias.
2.8. Ethics and dissemination
-----------------------------
No ethical review is required, as the data used in this study is extracted from published studies, which does not involve the personal data of the participants.
3. Discussion
=============
Migraine, as the most common episodic neurological disorder, has become a global public health problem.^\[[@R35]\]^ OCT is a common non-invasive imaging technique for measuring the retina and RNFL.^\[[@R36]\]^ Altered RNFL thickness investigated by OCT in migraine patients has been reported by many studies. In 2015, a meta-analysis has synthesized the relevant literatures and concluded that the RNFL thickness of migraine patients with optical coherence tomography is lower than that of healthy controls.^\[[@R31]\]^ Since then, there have been many publications eligible for inclusion, which may change the conclusion of the previous meta-analysis. Therefore, we will re-screen and update the relevant articles, hoping to provide some directions for the early diagnosis of migraine by monitoring the RNFL changes in the future.
Overall, we will quantitatively analyze RNFL thickness differences between migraine patients and healthy controls. Assess the quadrant most affected by migraine and analyze the cause. We speculate that thinning of RNFL thickness may be more affected by migraine with aura. Occipital hemispheric vasospasm and subsequent reduction in blood flow in migraine with aura and hypoperfusion around the optic nerve head may lead to retinal ganglion cell death.^\[[@R37]\]^ In addition, RNFL measured by SD-OCT may be more relevant to migraine than TD-OCT., for the SD-OCT can detect more subtle damage to RNFL.^\[[@R38]\]^
Author contributions
====================
**Conceptualization:** XueLing Zhang, HongJie Zhang
**Data curation:** XueLing Zhang, ZhongQuan Yi
**Formal analysis:** ZhongQuan Yi, QinQin Liu
**Investigation:** QinQin Liu, XiaoGuang Lin
**Methodology:** QinQin Liu, ZhongQuan Yi
**Project administration:** PingLei Pan
**Software:** HongJie Zhang, ChaoChun Chen, RuYuan Cai
**Supervision:** QinQin Liu
**Writing -- original draft:** XiaoGuang Lin, XueLing Zhang
**Writing -- review & editing:** PanWen Zhao, PingLei Pan
Abbreviations: CSD = cortical spreading depression, INPLASY = International platform of registered systematic review and meta-analysis protocols, MIDAS = migraine disability assessment scores., NOS = Newcastle-Ottawa Scale, OCT= optical coherence tomography, PRISMA = preferred reporting items for systematic reviews and meta-analyses, RNFL = retinal nerve fiber layer, PRISMA-P = preferred reporting items for systematic reviews and meta-analyses protocols, SD-OCT = spectral-domain optical coherence tomography, TD-OCT = time-domain optical coherence tomography, TGVS = trigeminal vascular system.
How to cite this article: Lin X, Yi Z, Zhang X, Liu Q, Cai R, Chen C, Zhang H, Zhao P, Pan P. Retinal nerve fiber layer changes in migraine: a protocol for systematic review and meta-analysis. *Medicine*. 2020;99:33(e21680).
HJZ, PWZ, and PLP contributed equally to this work.
This work was supported by the National Natural Science Foundation of China (Grant No. 81601161).
The authors have no conflicts of interest to disclose.
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
| {
"pile_set_name": "PubMed Central"
} |
**Core tip:** To study the effect of naringenin (NAR) on carbon tetrachloride (CCl~4~)-induced chronic liver fibrosis, male Wistar rats were administered 400 mg of CCl~4~/kg body weight and 100 mg of NAR/kg body weight for 8 wk. NAR prevented necrosis, cholestasis, oxidative stress, collagen accumulation and the increase in MMP activity caused by CCl~4~ administration. NAR completely prevented the increase in TGF-β, α-SMA, CTGF, Col-1, MMP-13, NF-κB, IL-1 and IL-10 protein levels caused by CCl~4~ administration and pSmad3 and pJNK activation. In conclusion, NAR prevents CCl~4~ induced liver fibrosis due to its antioxidant capacity and by inhibiting the TGF-β-Smad3, JNK-Smad3 and NF-κB pathways.
INTRODUCTION
============
It is known that various agents, including infections, ethanol ingestion, drug intoxication or malnutrition, result in liver damage. When the damage is prolonged, liver fibrosis occurs. Fibrosis is the result of excessive accumulation of extracellular matrix (ECM) proteins after chronic liver damage. Advanced liver fibrosis leads to cirrhosis, which is characterized by scar tissue, loss of parenchymal architecture and organ failure\[[@B1],[@B2]\].
The most important profibrogenic pathways involve transforming growth factor (TGF)-β and plateled-derived growth factor (PDGF), which induce the transdifferentiation of hepatic stellate cells (HSCs) from a non-proliferating cell type into a proliferating activated phenotype. This activation results in α-smooth-muscle actin (α-SMA) and collagen expression, which in turn lead to fibrosis\[[@B3],[@B4]\].
Traditionally, TGF-β has been considered to exert profibrogenic actions by binding to its receptor, which in turn phosphorylates transcription factors called R-Smads (Smad2 and Smad3) at its C-terminal. This is the canonical pathway, generally described as the main fibrogenic pathway, especially when Smad3 is phosphorylated at its C-terminus (pSmad3C)\[[@B5]-[@B8]\]. There is also a non-canonical pathway, in which PDGF activates JNK, which in turn phosphorylates Smad3 in the linker domain to generate pSmad3L, which rapidly translocates to the nucleus where it stimulates HSC proliferation\[[@B9],[@B10]\]. These pathways are critical for the induction of ECM deposition and the development of fibrosis and cirrhosis. Therefore, the canonical and non-canonical pathways constitute valuable targets in the development new and effective drugs to prevent liver fibrosis.
On the other hand, naringenin (NAR), 4',5,7-trihydroxy flavanone (Figure [1](#F1){ref-type="fig"}), is a flavonoid that is widely distributed in cherries, cocoa, grapes, tangelos, blood oranges, lemons, grapefruit, tangerines and tomatoes\[[@B11]-[@B13]\]; it has been demonstrated to prevent acute liver damage induced by alcohol, carbon tetrachloride (CCl~4~), lipopolysaccharide or heavy metals\[[@B14]-[@B17]\]. In addition, there is some experimental evidence on the anticancer properties of NAR\[[@B18],[@B19]\].
{#F1}
Some reports have shown evidence of the antifibrotic properties of NAR. Lee et al\[[@B19]\] showed some histological evidence that NAR prevents collagen deposition in dimethyl nitrosamine-induced liver damage in rats, but no mechanism was evaluated. Du et al\[[@B20]\] reported an antifibrotic effect of NAR in lung tissue that was associated with the downregulation of TGF-β activity. In a study with cultured rat HSCs, NAR prevented ECM deposition induced by TGF-β through the downregulation of Smad3 protein levels\[[@B8]\]. However, to the best of our knowledge, there are no studies on the antifibrotic effect of NAR on liver cirrhosis. Therefore, the above information prompted us to investigate whether NAR was able to prevent CCl~4~- induced liver cirrhosis in the rat and to elucidate if the mechanism of action of the flavonoid was associated with an inhibition of the canonical and/or non-canonical TGF-β pathways and/or the inflammatory pathway (NF-κB) and/or through antioxidant mechanisms.
In fact, we demonstrate for the first time that NAR is able to prevent CCl~4~ -cirrhosis by inhibiting the canonical and non-canonical pathways, including the prevention of pSmad3L phosphorylation by JNK, by exerting anti-inflammatory properties that block the NF-κB inflammatory cascade, and *via* an antioxidant mechanism. These effects lead to the inhibition of HSC transdifferentiation, the downregulation of ECM synthesis, and the prevention of necrosis, cholestasis and distortion of the hepatic architecture. As a result, experimental cirrhosis was prevented.
MATERIALS AND METHODS
=====================
Ethics statement and animal treatment
-------------------------------------
Wistar male rats were used and maintained on a standard diet of rat chow with free access to drinking water. Four or five animals were housed in each polycarbonate cage under controlled conditions (22 ± 2 °C, 50%-60% relative humidity and 12-h light-dark cycles). The study complies with the Institution's guidelines and the Mexican official regulation (NOM-062-ZOO-1999) regarding technical specifications for production, care and handling of laboratory animals.
Study design
------------
Wistar rats initially weighing 120-150 g were used. Cirrhosis was produced by intraperitoneal (i.p.) administration of CCl~4~ (400 mg/kg body weight) dissolved in mineral oil three times per week for 8 wk. To determine the capacity of NAR to prevent liver fibrosis, four groups were formed and treated for 8 wk. Group 1 (*n* = 8) consisted of control animals receiving carboxy methyl-cellulose p.o., the NAR vehicle; group 2 (*n* = 8) was administered CCl~4~; group 3 (*n* = 8) was administered CCl~4~ plus NAR (100 mg/kg body weight, daily p.o.); group 4 (*n* = 8) was administered NAR only. All animals were sacrificed after 8 wk under ketamine (100 mg/kg of body weight) and xylazine (8 mg/kg of body weight) anesthesia. Blood was collected by cardiac puncture, and the liver was rapidly removed and properly stored for further analysis.
Serum enzyme activities
-----------------------
The determination of liver damage was performed by measuring the activities of alanine aminotransferase (ALT)\[[@B21]\], alkaline phosphatase (AP)\[[@B22]\], and γ-glutamyl transpeptidase (γ-GTP)\[[@B23]\] in serum samples of blood that were centrifuged at 1300 rpm for 15 min.
Glycogen determination
----------------------
Small pieces of liver were separated for glycogen determination using the anthrone reagent\[[@B24]\]. Briefly, fresh liver samples (0.5 g) were boiled in 1.5 mL of 30% KOH for 30 min. After cooling, samples were diluted in a volumetric flask of 25 mL. Then, 40 μL (control and NAR groups) or 160 μL (CCl~4~ or CCl~4~ + NAR groups) was added to 960 μL and 840 μL of deionized water, respectively. Then, 2 mL of 0.2% anthrone (dissolved in concentrated sulfuric acid) was added. Samples were boiled for 15 min and the absorbance was read at 620 nm. Appropriate glucose standards were prepared.
Assessment of lipid peroxidation
--------------------------------
The extent of lipid peroxidation (LPO) was evaluated in liver homogenates *via* the measurement of malondialdehyde (MDA) formation with the thiobarbituric acid method\[[@B25]\]. Fresh liver samples (0.5 g) were homogenized in 5 mL of deionized water on ice with a polytron homogenizer. Later, 300 μL of 10% liver homogenate with 700 μL of 150 mmol/L Tris-HCl (pH 7.4) and 2 mL of 0.375% thiobarbituric acid (dissolved in 15% trichloroacetic acid) were boiled for 45 min and centrifuged at 3000 rpm for 10 min. The supernatant absorbance was read at 532 nm. The protein concentration was determined *via* the Bradford method using bovine serum albumin as a standard\[[@B26]\].
Reduced glutathione in liver and blood samples
----------------------------------------------
Fresh liver (0.3 g) and blood (0.3 mL) samples were homogenized in 1.2 mL of precipitating solution (5 mmol/L EDTA in 5% trichloroacetic acid) and centrifuged for 20 min at 12000 rpm. Then, 0.1 mL of the supernatant with 2.1 mL of phosphate solution (0.3 mol/L sodium phosphate dibasic) and 0.25 mL of Ellman's reagent (40 mg of 5,5'-dithiobis nitrobenzoic acid dissolved in 100 mL of 1% sodium citrate) were mixed. The absorbance was read at 412 nm\[[@B27]\].
Glutathione peroxidase activity in the liver
--------------------------------------------
The method of Lawrence and Burk\[[@B28]\] was used to determine glutathione peroxidase (GPx) activity with cumene hydroperoxide as a substrate. An aliquot of 1.5 mL of the 10% liver homogenate with 75 mmol/L potassium phosphate buffer (pH 7.0) was filtered through muslin cloth and centrifuged at 3000 rpm for 5 min at 4 °C. The reaction mixture contained 200 mL of the homogenate supernatant, 2.0 mL of 75 mmol/L potassium phosphate buffer (pH 7.0), 50 mL of 60 mmol/L glutathione, 0.1 mL of 30 U/mL glutathione reductase, 0.1 mL of 15 mmol/L EDTA, 0.1 mL of 3 mmol/L β-nicotinamide adenine dinucleotide phosphate (NADPH) and 0.3 mL of water. The reaction was started by the addition of 0.1 mL of 45 mmol/L cumene hydroperoxide. Oxidation of NADPH was recorded at 340 nm for 4 min, and the enzymatic activity was calculated as nmol of NADPH oxidized min^-1^mg^-1^ of protein using a molar extinction coefficient of 6.22 × 10^6^ M^-1^ cm^-1^.
Collagen quantification
-----------------------
Fresh liver samples (100 mg) were placed in ampoules, 2 mL of 6 N HCl was added, and then the samples were sealed and hydrolyzed at 100 °C for 48 h. Next, the samples were evaporated at 50 °C for 24 h and resuspended in 3 mL of sodium acetate-citric buffer, pH 6.0; 0.5 g of activated charcoal was added, and the mixture was stirred vigorously and then centrifuged at 3000 rpm for 15 min. Then, 1 mL of chloramine T was added to 1 mL of the supernatant. The mixture was kept for 20 min at room temperature, and the reaction was stopped by the addition of 0.5 mL of 2 mol/L sodium thiosulfate and 1 mL of 1 N sodium hydroxide. The aqueous layer was transferred into test tubes. The oxidation product from hydroxyproline was converted to a pyrrole by boiling the samples. The pyrrole-containing samples were incubated with Ehrlich's reagent for 30 min, and the absorbance was read at 560 nm. The recovery of known amounts of standards was carried out on similar liver samples for quantification\[[@B29]\].
Histology
---------
Liver samples were taken from all animals and fixed with 10% formaldehyde in phosphate-buffered saline for 24 h. Tissue samples were then washed with tap water, dehydrated in alcohol and embedded in paraffin. Five-micrometer-thick sections were mounted on silane covered glass slides. Staining was performed with hematoxylin and eosin (H&E) and Masson's trichrome stain.
Zymography
----------
Proteolytic activity was assayed with gelatin-substrate gels. Liver tissue (0.25 g) was homogenized with 1.7 mL of 1 × PBS. Then, the samples were homogenized on ice with a polytron homogenizer, sonicated and centrifuged at 13000 rpm for 10 min. The supernatant was collected and proteins were quantified *via* the bicinchoninic acid method (Pierce BCA Protein Assay Prod \# 23223 Thermo Scientific). Volumes equivalent to 50 μg of non-heated proteins were mixed with sample buffer (2.5% SDS, 1% sucrose and 4 mg/mL phenol red) without reducing agent and applied to 8% acrylamide gels copolymerized with 1 mg/mL gelatin. After electrophoresis at 72 V for 2 h, the gels were rinsed twice in 2.5% Triton X-100 to remove SDS and then incubated in 50 mmol/L Tris-HCl at pH 7.4 and 5 mmol/L CaCl~2~ assay buffer at 37 °C for 48 h. The gels were then fixed and stained with 0.25% Coomassie Brilliant Blue G-250 in 10% acetic acid and 30% methanol. Proteolytic activity was detected as clear bands against the background stain of undigested substrate in the gel at the expected location according to the molecular weight of metalloproteinase (MMP)-9 and MMP-2. Images were digitized and then analyzed densitometrically with ImageJ software.
Western blot assays
-------------------
To carry out western blot assays, lysis buffer (1 mol/L Tris-HCl pH 8, 5 mol/L NaCl, NP40, Triton, 0.5 mol/L EDTA pH 8, 0.1 mol/L PMSF, 0.1 mol/L Na~3~VO~4~, 0.1 mol/L NaF) with protease and phosphatase inhibitors (Pierce BCA Protein Assay Prod \# 23223 Thermo Scientific), each at a ratio of 1:100, was used to isolate total protein from liver tissue samples. Liver tissue (50 mg) was homogenized with 500 μL of lysis buffer. Then the samples were sonicated and centrifuged at 12000 rpm for 2 min. The supernatant was collected and proteins were quantified *via* the bicinchoninic acid method.
Volumes equivalent to 50 and 250 μg of protein were transferred onto 15, 12 and 10% polyacrylamide and electrophoresis was carried out. Separated proteins were transferred onto an Inmuno-Blot^TM^ PVDF membrane (BIO-RAD, Hercules, CA, United States). Next, blots were blocked with 7% skim milk and 0.05% Tween-20 for 2 h at room temperature and independently incubated overnight at 4 °C with specific primary antibodies (Table [1](#T1){ref-type="table"}). The following day, membranes were washed with TBS-tween-20 and then exposed to a secondary peroxidase-labeled antibody α-mouse (62-6520 Invitrogen) or α-rabbit (31460 Thermo Fisher Scientific) in the blocking solution for 2 h at room temperature. Blots were then washed with TBS-tween-20, and protein development was performed with the western lightning^TM^ Plus-ECL Enhanced Chemiluminescence detection system (NEN Life Sciences Products, Elmer LAS Inc., Boston, MA, United States). β-actin was used as a control to normalize cytokine protein expression levels. Images were digitized and then analyzed densitometrically with ImageJ software.
######
Antibodies employed for this research
**Protein** **Company** **Catalog number** **Dilution**
------------- -------------------------- -------------------- --------------
TGF-β Millipore MAB1032 1:500
α-SMA Sigma Aldrich A5691 1:500
CTGF Santa Cruz Biotechnology SC-14939 1:500
Col-1 Sigma Aldrich C-2456 1:500
MMP-13 Millipore MAB13426 1:500
NF-κB (p65) Millipore MAB3026 1:500
IL-1 Millipore AB1832P 1:500
IL-10 Invitrogen ARC9102 1:500
Smad3 Abcam Ab65847 1:500
pSmad3L Abcam Ab63403 1:250
Smad7 Abcam Ab90086 1:500
JNK Cell Signaling 9252 1:500
pJNK Abcam Ab131499 1:500
Statistical analysis
--------------------
All data are expressed as the mean values ± SE. Comparisons were carried out *via* the analysis of one way variance followed by Tukey's test, as appropriate, using the Graph Pad Prism software. Differences were considered statistically significant when *P* was \< 0.05.
RESULTS
=======
NAR prevented necrosis and cholestasis and improved liver biosynthetic capacity in CCl~4~ -treated rats
-------------------------------------------------------------------------------------------------------
As shown in Figure [2](#F2){ref-type="fig"}, chronic administration of CCl~4~ significantly increased the serum activity of ALT, a hepatocyte necrosis indicator\[[@B30]\]. After CCl~4~ administration, AP and γ-GTP (two markers of cholestasis\[[@B30]\]) serum activity significantly increased compared to the levels in the control group. NAR administration completely prevented the increase in ALT, AP and GTP activity, suggesting that NAR is able to prevent necrosis and cholestasis caused by CCl~4~ administration.
{#F2}
Glycogen measurement is used to determine the biosynthetic capacity and proper function of the liver\[[@B30]\]. Cirrhotic livers had significantly lower glycogen levels; however, NAR completely prevented the depletion of hepatic glycogen (Figure [3](#F3){ref-type="fig"}).
{#F3}
NAR prevented the oxidative stress caused by chronic liver damage
-----------------------------------------------------------------
One of the main products of LPO is MDA, which is utilized to measure oxidative stress in tissues\[[@B31]\]. As expected, the induction of cirrhosis triggered LPO, since MDA levels were significantly elevated compared to the control group. Interestingly, NAR prevented the increase in MDA levels (Figure [4A](#F4){ref-type="fig"}).
{#F4}
Reduced glutathione (GSH) is one of the most important endogenous antioxidants\[[@B32]\]. CCl~4~-treated rats exhibited low liver and blood GSH levels compared with the control group, but NAR completely prevented this decrease and even raised GSH level above that of the control group (Figure [4B](#F4){ref-type="fig"} and C).
One of the most important antioxidant enzymes is GPx, which utilizes GSH to detoxify H~2~O~2~\[[@B33]\]. GPx activity was significantly decreased by the chronic administration of CCl~4,~ while NAR coadministration partially prevented this effect (Figure [4D](#F4){ref-type="fig"}).
Together, these results show that NAR prevents oxidative stress during experimental liver cirrhosis at several levels.
Inflammation during liver injury was prevented by NAR administration
--------------------------------------------------------------------
NF-κB is a key protein in inflammation, since it regulates interleukin expression, including IL-1 and IL-10\[[@B34],[@B35]\]. Baseline values of NF-κB, IL-1 and IL-10 were increased after CCl~4~ treatment, and the expression of NF-κB increased almost 3-fold compared to the control group. The levels of IL-1 and IL-10 increased by 3.8- and 1.8-fold, respectively, compared to the control group. NAR prevented inflammation and necrosis in CCl~4~-treated rats by maintaining normal NF-κB, IL-1 and IL-10 levels (Figure [5](#F5){ref-type="fig"}).
{#F5}
Collagen accumulation was prevented by NAR in CCl~4~-induced experimental cirrhosis
-----------------------------------------------------------------------------------
Collagen quantification provides information about the balance between ECM synthesis and degradation\[[@B3],[@B4]\]. Rats treated with CCl~4~ showed a nearly 4-fold increment in collagen content compared to the control group. This effect was completely prevented by NAR (Figure [6](#F6){ref-type="fig"}).
{#F6}
The general appearance of the livers at the macroscopic and microscopic level can be seen in Figure [7](#F7){ref-type="fig"}. Figure [7A](#F7){ref-type="fig"} shows a normal liver from the control group. CCl~4~ treatment produced macro nodular fibrosis (Figure [7B](#F7){ref-type="fig"}) that was prevented by NAR administration (Figure [7C](#F7){ref-type="fig"}). NAR-treated rats had livers that were macroscopically similar to those of control animals (Figure [7D](#F7){ref-type="fig"}).
{#F7}
H&E staining is shown in Figure [7E](#F7){ref-type="fig"}-H. Figure [7E](#F7){ref-type="fig"}, the control group, shows no alterations of the hepatic parenchyma. Figure [7F](#F7){ref-type="fig"} corresponds to a representative liver section of chronic CCl~4~-induced liver injury; in this case, the tissue shows liver parenchymal disruption, steatosis, hyperchromatic nuclear hepatocytes, and atypical pleomorphic nuclei. Fibrosis was decreased by the administration of NAR (Figure [7G](#F7){ref-type="fig"}). NAR treatment of control rats produced no effect on liver histology (Figure [7H](#F7){ref-type="fig"}).
Masson's stained liver slices are shown in Figure [7I](#F7){ref-type="fig"}-L. Figure [7J](#F7){ref-type="fig"} presents a sample of a liver from a cirrhotic rat; a large amounts of collagen around fibrotic nodules was detected. The distortion of the parenchyma is evident when compared with a normal control liver (Figure [7I](#F7){ref-type="fig"}). NAR treatment prevented collagen accumulation and the formation of regenerative nodules (Figure [7K](#F7){ref-type="fig"}), and NAR alone produced no effect (Figure [7L](#F7){ref-type="fig"}).
NAR preserved the normal activity of MMP-9 and MMP-2 in experimental liver cirrhosis
------------------------------------------------------------------------------------
MMPs are enzymes that are responsible for ECM degradation, and among the most important are MMP-9 and MMP-2\[[@B36]\]. As seen in Figure [8](#F8){ref-type="fig"}, during normal conditions, MMP-9 and MMP-2 exhibit basal activity. However, after CCl~4~ administration, MMP-9 and MMP-2 activities increased 3.7- and 5.4-fold compared to the control group, respectively. NAR treatment effectively maintained basal MMP activity levels in animals with experimental cirrhosis induced by CCl~4~.
{#F8}
NAR blocked HSC transdifferentiation and Col-1 synthesis by inhibiting profibrogenic proteins
---------------------------------------------------------------------------------------------
It is well known that TGF-β induces HSC transdifferentiation, as well as α-SMA, Col-1 and connective tissue growth factor (CTGF) expression in HSCs\[[@B1],[@B37]\]. In this study, under normal conditions, basal TGF-β, α-SMA, CTGF and Col-1 protein levels were observed; however, experimental fibrosis induced by CCl~4~ administration increased the expression of such proteins several-fold compared to the control group. NAR administration completely prevented the elevation of TGF-β, α-SMA, CTGF and Col-1 levels during CCl~4~ administration. Unexpectedly, NAR alone reduced α-SMA and Col-1 levels to below the control group (Figure [9A](#F9){ref-type="fig"}-D).
{#F9}
MMP-13 is involved in the migration and proliferation of HSCs and the activation of TGF-β\[[@B38]\]. As seen in Figure [9E](#F9){ref-type="fig"}, during normal conditions MMP-13 depicts basal protein levels; however, due to CCl~4~ administration, the MMP-13 level increased 2.3 fold above the control group. NAR treatment effectively maintained basal MMP-13 protein levels in animals with experimental cirrhosis induced by CCl~4~.
Since Smad7 inhibits the TGF-β signaling pathway by TGF-β receptor ubiquitination\[[@B39]\], we evaluated Smad7 to determine its participation in the mechanism of action of NAR. Smad7 levels were significantly decreased by CCl~4~ administration, but this effect was prevented by NAR treatment. Interestingly, NAR administration by itself increased Smad7 protein expression above control values (Figure [9F](#F9){ref-type="fig"}).
NAR inhibited HSC proliferation by blocking the JNK-pSmad3L pathway
-------------------------------------------------------------------
Smad3 is normally activated by TGF-β, but JNK, *via* the linker phosphorylation pathway, also enables Smad3 to induce HSC proliferation\[[@B9],[@B10]\]. Our results confirmed that animals treated with CCl~4~ over 8 wk showed a significant increase in the expression of pJNK, pSmad3L and Smad3 compared with basal conditions. Interestingly, NAR administration prevented JNK activation, the elevation of Smad3 protein levels and phosphorylation in the linker region (Figure [10](#F10){ref-type="fig"}).
{#F10}
DISCUSSION
==========
The aim of this study was to evaluate the hepatoprotective effect of NAR in a model of CCl~4~-induced chronic damage and to investigate whether the beneficial effects of NAR are associated with its antioxidant properties, and/or the disruption of NF-κB, and/or the canonical or non-canonical TGF-β pathways. The results show that the flavonoid is able to prevent necrosis, cholestasis, glycogen stores depletion and oxidative stress induced by chronic CCl~4~ administration. NAR preserved liver function, normal collagen levels and MMP basal activity during CCl~4~ administration. In addition, NAR prevented HSC transdifferentiation and Col-1 synthesis by inhibiting profibrogenic proteins such as TGF-β and CTGF. NAR also preserved Smad7 protein levels, which decreased during liver injury. Regarding the non-canonical Smad pathway, the flavonoid prevented the activation of JNK and Smad3 phosphorylation in the linker region. Finally, NAR showed anti-inflammatory properties because it maintained normal levels of NF-κB, IL-1 and IL-10 during liver injury. In summary, NAR prevented CCl~4~ -cirrhosis by means of its antioxidant, anti-inflammatory, immunomodulatory and antifibrotic properties.
NAR protects the liver via an antioxidant mechanism
---------------------------------------------------
The antinecrotic and anticholestatic effects of NAR may be associated with the ability of the flavonoid to prevent membrane damage due to its antioxidant properties. The toxic effect of CCl~4~ is dependent on its degraded metabolites, trichloromethyl and trichloromethyl peroxyl radicals, which are formed by the liver microsomal enzyme CYP2E1. These molecules are unstable radicals and exhibit strong binding affinity to protein and lipids of cell membranes by abstracting a hydrogen atom from an unsaturated lipid, thereby triggering LPO and thus causing liver damage\[[@B40],[@B41]\].
NAR was able to prevent LPO due to its molecular structure; perhaps its hydroxyl groups facilitate its adherence to lipid bilayer polar groups, and its nonpolar nucleus may interact with hydrophobic tails of phospholipids, thereby reducing the deleterious effects of free radicals on membranes\[[@B12],[@B42],[@B43]\].
On the other hand, GSH is one of the main components of the endogenous antioxidant system; it scavenges hydroxyl and peroxynitrite radicals and is a GPx cofactor, among other important activities\[[@B32]\]. Chronic CCl~4~ administration decreased GSH levels; however, NAR was could preserve normal GSH levels in the liver and blood. In addition to its antioxidant properties as a free radical scavenger, NAR may also act though Nrf2 activation and induce endogenous antioxidant enzymes. In fact, NAR upregulates the expression of the enzyme glutamate-cysteine ligase, the rate limiting enzyme involved in de novo GSH synthesis\[[@B44]-[@B48]\], as well as glutathione reductase, which catalyzes the reduction of oxidized glutathione to the reduced form\[[@B15],[@B43]\]. GPx detoxifies H~2~O~2~ by reducing it to water and oxygen, utilizing two molecules of GSH to form GSSG in the reduction of a molecule of H~2~O~2~\[[@B49]\]. In agreement with previous reports\[[@B13],[@B15],[@B50]-[@B53]\], CCl~4~ decreased GPx activity in this study. NAR cotreatment partially prevented the decrement in GPx activity, probably by up regulating its expression through Nrf2 modulation\[[@B15],[@B40],[@B43],[@B44]-[@B47],[@B50]\], providing another mechanism to fight oxidative stress induced by CCl~4~.
NAR prevents hepatic necrosis by blocking the NF-κB pathway
-----------------------------------------------------------
Rats treated with CCl~4~ showed increased NF-κB, IL-1 and IL-10 protein levels, but concomitant NAR administration with CCl~4~ prevented this increase. NAR inhibits NF-κB *via* the downregulation of TLR4 and TLR2 mRNA and protein levels and the decreased translocation and DNA binding of NF-κB\[[@B54]-[@B56]\]. This leads to the inhibition of the expression of NF-κB dependent interleukins, such as IL-1 and IL-10, and thus, necrosis is prevented.
NAR preserves MMP-9 and MMP-2 activity in CCl~4~-treated rats
-------------------------------------------------------------
During fibrosis, activated HSCs and Kupffer cells express MMP-9 and MMP-2; these enzymes lead to TGF-β activation by cleaving TGF-β from its reservoir on ECM, thus enhancing HSC invasive activity\[[@B38],[@B53],[@B57],[@B58]\]. CCl~4~ chronic administration increased MMP-9 and MMP-2, while concomitant NAR administration maintained the normal activity of MMPs. In agreement with these findings, several reports indicate that NAR reduces both protein and mRNA levels of MMP-9 and MMP-2, thereby reducing the activity of these MMPs\[[@B59]-[@B61]\]. Therefore, it seems likely that the antifibrotic effect of NAR may be due, in part, to the downregulation of MMP-9 and MMP-2.
NAR inhibits the TGF-β-Smad3 pathway, leading to the downregulation of α-SMA, CTGF and Col-I
--------------------------------------------------------------------------------------------
Through pSmad3C, TGF-β induces the expression of α-SMA, CTGF and Col-1 in activated HSCs; α-SMA is a highly specific transdifferentiation marker that is closely related to HSC contractile and migration capacities. CTGF amplifies the profibrogenic action of TGF-β, and Col-I is one of the main types of ECM collagen\[[@B4],[@B37],[@B62],[@B63]\]. In this study, CCl~4~ administration increased TGF-β, α-SMA, CTGF and Col-I protein levels; importantly, NAR preserved the normal levels of these proteins.
The possible mechanism by which NAR inhibits the TGF-β-Smad3 pathway are (1) a reduction in tissue TGF-β levels; (2) a decrease in Smad3 mRNA and protein levels, with consequent reduction in Smad3 phosphorylation, thus preventing the nuclear translocation of pSmad3C; and (3) a reduction in the binding of TGF-β to its specific receptor, TβRII, leading to the inhibition of Smad3 phosphorylation\[[@B8],[@B20],[@B59],[@B64]-[@B66]\]. In agreement with the results that were obtained, it has been reported that the inhibition of the TGF-β-Smad3 pathway by NAR results in decreased α-SMA, CTGF and Col-I mRNA and protein levels\[[@B8],[@B19],[@B37],[@B67]\], therefore providing a suitable mechanism to further explain the antifibrotic properties of NAR.
NAR downregulates the profibrogenic TGF-β pathway by preserving Smad7 protein levels
------------------------------------------------------------------------------------
Contrary to Smad3, Smad7 exerts an inhibitory effect on the TGF-β pathway by activating TβRI degradation\[[@B39]\]. As reported by others\[[@B68],[@B69]\], CCl~4~ administration reduced Smad7 protein levels, but NAR was able to prevent this event. Lou et al\[[@B59]\], reported that NAR is able to preserve Smad7 mRNA levels in pancreatic cells treated with TGF-β. The prevention of Smad7 diminution by NAR constitutes another valuable mechanism by which the TGF-β pathway is downregulated to prevent fibrosis.
NAR blocks the profibrogenic action of TGF-β by downregulating MMP-13
---------------------------------------------------------------------
MMP-13 is expressed by HSCs, Kupffer cells, and perisinusoidal cells; it releases ECM-bound cytokines such as TGF-β, leading to HSC proliferation and migration\[[@B38]\]. CCl~4~-induced liver damage increased protein levels of MMP-13; however, NAR maintained normal MMP-13 levels during liver damage. MMP-13 can be upregulated by IL-1, NF-κB and JNK\[[@B38],[@B70]\]; however, these proteins were also downregulated by NAR. By downregulating MMP-13, NAR may prevent ECM deposition in rats treated with CCl~4~.
NAR also prevents fibrosis by blocking the non-canonical TGF-β pathway
----------------------------------------------------------------------
After CCl~4~ administration, the activation of JNK was elevated, as were Smad3 protein levels and phosphorylation in the linker region of this protein. During CCl~4~ administration, pSmad3L has been found in the nucleus of HSCs. The phosphorylation of Smad3 in the linker region is catalyzed by pJNK, resulting in a rapid translocation of pSmad3L to the nucleus and the stimulation of expression of c-myc, an important HSC proliferation inducer\[[@B9],[@B71],[@B72]\], which leads to increased ECM production. NAR was capable of preventing the increase in Smad3 protein levels and JNK phosphorylation, and therefore, pSmad3L formation was decreased, leading to the inhibition of ECM deposition.
Figure [11](#F11){ref-type="fig"} summarizes the mechanisms by which NAR acts to prevent liver oxidative stress, necrosis and fibrosis induced by chronic CCl~4~ intoxication; (1) during fibrosis, TGF-β is a major HSC transdifferentiation inductor, so the levels are increased. NAR administration maintained basal levels of TGF-β; (2) normally, latent TGF-β is anchored to the ECM, but during liver fibrogenesis, the activation of metalloproteinases such as MMP-2, MMP-9 and MMP-13 leads to TGF-β activation by cleaving it from its reservoir. NAR effectively preserved basal MMP activity levels and prevented TGF-β activation; (3) TGF-β binds to the TGF-β type II receptor (TβRII), which recruits and phosphorylates the TGF-β type I receptor (TβRI), which in turn phosphorylates Smad3 in its C-terminal region, leading to pSmad3C formation. Smad7 inhibits the TGF-β signaling pathway *via* TβRI ubiquitination and posterior degradation in the proteasome. NAR increased Smad7 expression, therefore downregulating the TGF-β pathway; (4) it is known that PDGF activates JNK, and in turn, this kinase phosphorylates Smad3 in the linker domain to generate pSmad3L. NAR administration prevented JNK activation and Smad3 phosphorylation; (5) Smad3 is indispensable for the canonical and non-canonical pathways and in liver samples from cirrhotic rats Smad3 levels are increased; however, NAR maintained basal Smad3 levels; (6) NF-κB regulates the expression of interleukins such as IL-1 and IL-10, while NAR prevented inflammation by downregulating the expression of NF-κB; and (7) one of the most important antioxidant enzymes is GPx, which utilizes GSH to detoxify H~2~O~2~; during cirrhosis, this activity decreased, and NAR partially prevented this effect and blocked LPO.
{#F11}
In conclusion, our results demonstrate, for the first time, that NAR completely prevents CCl~4~-induced liver fibrosis in rats, not only because of its antioxidant properties but also *via* its effects as an immunomodulator and a downregulator of several profibrogenic pathways. The present results not only provide important information about the mechanism of the antifibrotic actions of NAR but also suggest that this flavonoid may be utilized in patients with fibrosis previous clinical and toxicological evaluation.
ACKNOWLEDGMENTS
===============
The authors thank Ramón Hernández Guadarrama, Benjamín Salinas Hernández, Laura Dayana Buendia Montaño, Paula Vergara, Ma. Teresa García Camacho, Silvia Galindo, Angélica Silva Olivares, Rafael Leyva, Benjamín E. Chávez, and Ricardo Gaxiola for excellent technical assistance. The authors also acknowledge the Animal Lab Facility, UPEAL-Cinvestav.
COMMENTS
========
Background
----------
Liver fibrosis results from chronic liver damage and is characterized by extracellular matrix (ECM) protein deposition. The mechanisms by which the ECM is induced by hepatic stellate cells includes oxidative stress and involves canonical and non-canonical TGF-β pathways and increased metalloproteinases. Naringenin (NAR) has shown some hepatoprotective properties; however, the capacity of NAR to prevent liver fibrosis has not yet been evaluated.
Research frontiers
------------------
NAR is a flavonoid that is widely distributed in nature and possesses antioxidant, anti-inflammatory and immunomodulatory properties that may be very useful to prevent hepatic fibrosis.
Innovations and breakthroughs
-----------------------------
Oxidative stress, NF-κB activity, TGF-β-Smad3 and metalloprotease activity lead to necrosis and fibrosis. NAR may inhibit hepatic necrosis and fibrosis by blocking free radicals and inhibiting these proinflammatory and profibrotic pathways.
Applications
------------
The current results indicate that the antinecrotic, anti-inflammatory and antifibrotic effects of NAR may be due to its ability to directly and indirectly fight free radicals and to inhibit NF-κB, IL-1, IL-10, and TGF-β-Smad3 and metalloproteases activity.
Terminology
-----------
NAR is the natural flavonoid aglycone of naringin; it is a flavanone with a stereogenic center at C2. It has two enantiomers, namely, (R)-NAR and (S)-NAR, and both NAR enantiomers are present in natural sources.
Peer-review
-----------
The paper by Hernández-Aquino et al investigated the hepato-protective and anti-fibrotic effects of NAR using a CCl~4~-induced liver fibrosis model. The authors found NAR protects liver functions in the CCl~4~-treated livers, and reduces levels of oxidative stress, fibrosis, and inflammation. This is an interesting paper which has implications for possible mechanisms underlying hepato-protective and anti-fibrotic effects of NAR. The manuscript was well written and the data support their claims.
Manuscript source: Invited manuscript
Specialty type: Gastroenterology and hepatology
Country of origin: Mexico
Peer-review report classification
Grade A (Excellent): 0
Grade B (Very good): B, B
Grade C (Good): 0
Grade D (Fair): 0
Grade E (Poor): 0
Institutional animal care and use committee statement: The study complies with the Institution's guidelines and the Mexican official regulation (NOM-062-ZOO-1999).
Conflict-of-interest statement: The authors declare no conflicts of interest.
Data sharing statement: No additional data are available.
Peer-review started: November 18, 2016
First decision: March 3, 2017
Article in press: June 1, 2017
P- Reviewer: Dang SS, Ro SW S- Editor: Gong ZM L- Editor: A E- Editor: Wang CH
[^1]: Author contributions: Hernández-Aquino E, Zarco N, Casas-Grajales S, Ramos-Tovar E, Flores-Beltrán RE, Arauz J, Favari L and Segovia J performed the biochemical, molecular and zymography determinations; Shibayama M and Tsutsumi V performed the histological stains and their interpretation; and Muriel P designed the research and wrote the paper together with Hernández-Aquino E.
Supported by National Council of Science and Technology (Conacyt) of Mexico, No. 253037 to Muriel P, and No. 239516 to Segovia J; Fellowship No. 358378 Hernández-Aquino E to from Conacyt; This work was also partially supported by a grant of PRODEP (UABC-PTC-464) Mexico.
Correspondence to: Pablo Muriel, PhD, Laboratory of Experimental Hepatology, Department of Pharmacology, Cinvestav-IPN, Av. Instituto Politécnico Nacional 2508, Apartado Postal 14-740, 07000, Mexico City, Mexico. <pmuriel@cinvestav.mx>
Telephone: +52-55-57473303 Fax: +52-55-57473394
| {
"pile_set_name": "PubMed Central"
} |
The homicide rate in Scotland has been increasing since 1980.[@b1] This increase has accompanied rising death rates due to suicide (among men), chronic liver disease and mental and behavioural disorders attributed to the use of drugs and alcohol.[@b2] [@b3] The changes in mortality from these causes have been most noticeable among younger adults, particularly young men, for whom the effect has been so strong that all-cause death rates at these ages have begun to rise.[@b4]
Much of the previous research on the social patterning of homicide has been conducted in the USA; this has shown the victims of homicide to be more likely to be drawn from certain social groups: generally, homicide rates are higher among black,[@b5; @b6; @b7; @b8; @b9] low income,[@b5] [@b9] poorly educated,[@b5] [@b6] [@b9] unemployed[@b5] [@b6] [@b9] and manual workers.[@b8] [@b10] Research from Finland has also noted elevated rates among unemployed and manual workers,[@b11] and in Sweden there was an association with early life socioeconomic position.[@b12] Area characteristics with noted associations with homicide rates include degree of urbanisation (USA),[@b7] income inequality (USA and internationally),[@b7] [@b9] [@b10] [@b13] per capita income (USA)[@b6] and deprivation or poverty (USA, The Netherlands and Great Britain).[@b6] [@b9] [@b14] [@b15]
The effect of homicide on social inequalities in mortality may be disproportionate. The (area-based) socioeconomic gradient for homicide was shown to be steeper than for many other causes of death in The Hague and, although contributing just 1.3% and 0.7% of all male and female deaths under 65, contributed 6.4% and 3.9% of the excess mortality in the quartile of regions with the highest deprivation scores compared with the quartile with the lowest scores.[@b14] In New York City, inequalities patterned by neighbourhood poverty (as measured by the relative index of inequality), for men and women combined, were 1.7 for all causes and 9.0 for homicide.[@b16] Neither study reported the significance of these results. In Sweden, men with manual social class in early life had a hazard ratio of death from homicide of 2.11 compared with those with non-manual social class; for all-cause mortality, the ratio was 1.31. For women, the ratios were 1.40 and 1.18, respectively.[@b12] This difference was significant for men but not for women. In Estonia, the rate ratio for homicide mortality comparing the lowest educated group to those with university education (2.21) was about the same as that for all-cause mortality (2.38) among men, while the homicide rate ratio for women (5.46) was higher than that for all causes (2.23), although the difference was not significant.[@b17]
Earlier research has shown the increasing importance of homicides involving the use of knives or other sharp weapons among all homicides in Scotland.[@b1] While the male homicide rate increased by 83% between 1982 and 2002, the rate involving knives or similar increased by 164% and more than trebled for those aged 15--34. Between 1983 and 1998, inequalities in homicide in Great Britain increased as homicide rates decreased in the more affluent areas while increasing in the more deprived areas.[@b15] Over this period, the rate ratio in the poorest decile compared with the richest increased from 4.50 to 5.68. Murders involving a knife or a broken glass/bottle showed a social gradient,[@b15] such a method constituting 52% of all homicide deaths in the poorest decile but only about 35% in the least poor decile. Given the high proportion of deprived areas that lie within Scotland, and the West of Scotland in particular, and the high rate of homicides involving knives in Scotland (47% in Scotland between 1981 and 2003[@b1] compared with 38% in England and Wales between 1993 and 1997[@b18]), it is possible that the social gradient reported reflects in part this high Scottish rate. The importance of the use of knife in assaults and homicides has led to policy changes or proposals targeting knives, including the licensing of the sale of non-domestic knives[@b19] and a ban on the sale of all long pointed kitchen knives.[@b20]
It is against this background that we investigate the social patterning of homicide in Scotland between 1980 and 2005, both by individual socioeconomic status and by area-based measures of deprivation. We explored the extent to which inequalities in homicide have increased, the extent to which this may be attributable to homicide involving the use of knives or other sharp weapons and the contribution of homicide to all-cause mortality and to inequalities in mortality.
Methods
=======
Data
----
The analyses presented in this paper are based on death records for Scotland for the years 1980--1982, 1990--1992 and 2000--2005, obtained from the General Register Office for Scotland (GROS). Information on the death certificates includes the age and sex of the deceased as well as the cause of death and a measure of individual socioeconomic status. They also contain the postcode of residence of the deceased, enabling linkage to area-based deprivation scores. The estimation of rates for population groups defined by area of residence requires denominator populations in addition to the events (deaths); for this reason, small area population estimates for these years by age and sex were also obtained from GROS.
Homicides were defined as deaths where the underlying cause was homicide and injury purposely inflicted by other persons (ICD9 E960-E969) or assault or sequelae of assault (ICD10 X85-Y09 and Y87.1). We conducted separate analyses for deaths where the cause was assault by a cutting or piercing instrument (ICD9 E966) or assault by sharp object (ICD10 X99).
The measure of individual socioeconomic status that we have used is the National Statistics Socioeconomic Classification (NS-SEC).[@b21] NS-SEC is based on occupation and employment status but has rules to provide coverage of the whole adult population. It is constructed to measure employment relations and conditions of employment. There are several versions of the classification, and the one used here has eight classes (detailed in [table 1](#tbl1){ref-type="table"}, together with those who could not be classified).[@b22] The proportion of the population assigned an NS-SEC from age 60 is low and so becomes unreliable to use. In addition, the completeness of NS-SEC for ages 16--19 years is also low because of delayed entry into the labour market.[@b2] For these reasons, analyses using NS-SEC focused on the age group 20--59. Earlier results have cast doubt on the reliability of NS-SEC coding for the analysis of female mortality,[@b2] and we therefore restricted the analysis to men. Population estimates by NS-SEC were only available from the 2001 population census (earlier censuses used the alternative social class based on occupation); we used these estimates for men aged 20--59 to standardise the male death rate by NS-SEC.
######
Standardised death rates per 1 000 000 population due to assault, assault involving sharp weapons and other assaults, by NS-SEC category (men, Scotland, 2000--2002)
NS-SEC N (% of population) Assault Sharp weapons Other assault
------------------------------------------------ --------------------- --------- --------------- --------------- ----------- ----------- ---------- ----------- ----------- ---------- ---------- ---------
Higher managerial and professional occupations 77 487 87 446 164 933 11 11 11 0 7 4 11 4 7
(11.2) (13.1) (12.2) (0--33) (0--23) (0--24) (0--17) (0--8) (0--33) (0--11) (0--19)
Lower managerial and professional occupations 132 985 140 994 273 979 15 12 14 13 5 9 2 7 4
(19.3) (21.2) (20.2) (2--29) (1--23) (5--22) (1--26) (0--12) (2--17) (0--6) (0--15) (0--9)
Intermediate occupations 53 658 32 121 85 779 6 10 8 0 10 5 6 0 3
(7.8) (4.8) (6.3) (0--17) (0--30) (0--19) (0--30) (0--15) (0--17) (0--9)
Small employers and own account workers 48 268 82 402 13 0670 23 12 18 19 4 12 4 8 6
(7.0) (12.4) (9.6) (3--43) (0--26) (5--30) (0--37) (0--12) (1--22) (0--12) (0--20) (0--13)
Lower supervisory and technical occupations 96 842 87 407 184 249 28 27 27 25 20 22 3 7 5
(14.0) (13.1) (13.6) (8--48) (7--47) (13--41) (6--44) (2--37) (10--35) (0--9) (0--17) (0--11)
Semiroutine occupations 82 616 62 761 145 377 65 48 57 46 16 31 19 33 26
(12.0) (9.4) (10.7) (34--97) (17--80) (35--79) (20--73) (0--34) (15--48) (2--36) (6--59) (10--41)
Routine occupations 103 228 90 245 193 473 150 103 127 122 43 83 29 59 44
(14.9) (13.5) (14.3) (107--194) (65--141) (98--156) (83--160) (19--68) (60--106) (10--48) (30--89) (27--61)
p Value \<0.001 \<0.001 \<0.001 \<0.001 0.004 \<0.001 0.019 \<0.001 \<0.001
Never worked and long-term unemployed 31 116 22 592 53 708 21 31 26 21 16 19 0 15 7
(4.5) (3.4) (4.0) (0--51) (0--74) (0--52) (0--51) (0--46) (0--40) (0--45) (0--22)
Not classified 64 320 60 086 124 406 181 75 129 103 39 72 78 36 58
(9.3) (9.0) (9.2) (103--260) (34--117) (84--174) (44--161) (8--70) (38--105) (26--131) (8--64) (28--88)
All 595 084 583 376 1 178 460 55 36 46 41 17 29 14 19 17
(100.0) (100.0) (100.0) (45--65) (28--44) (39--52) (32--50) (11--22) (24--34) (9--20) (13--25) (13--21)
p Values show significance of tests for difference between groups.
We used two different measures of area deprivation. The Carstairs score is a census-based deprivation score derived from levels of male unemployment, social class of the head of household, overcrowding and (lack of) car ownership, and has been calculated by adding together the standardised values of the variables following the 1981, 1991 and 2001 censuses.[@b23; @b24; @b25] The continuous score was used to divide areas into (population-weighted) quintiles. The income domain of the Scottish Index of Multiple Deprivation (SIMD) is based on the proportion of the population living in households in receipt of mean-tested benefits and was calculated based on the numbers claiming different allowances or benefits in 2001 and 2004.[@b26] [@b27] Areas were ranked, and these ranks were used to divide areas into (population-weighted) quintiles.
In 2001, the population of Scotland was 5 062 011. The Carstairs score is calculated for postcode sectors; in 2001, there were 1010 postcode sectors in Scotland (including part postcode sectors when their boundaries crossed local council areas) with a mean population of 5012 (range, 51--20 512). The SIMD uses a geographical area with lower mean population size and less variability in population size, the data zone; in 2001, there were 6505 data zones in Scotland with a mean population of 778 (range, 476--2813). To match populations to these deprivation measures, we used census populations for 1981, 1991 and 2001 by age, sex and postcode sector; census populations for 2001 by age, sex and data zone; and GROS population estimates for 2004 by age, sex and data zone. We confined the area-based analyses around the 1991 census to 1991--1992 because of difficulties matching addresses on death records and the census following changes to some postcode sectors in 1990.
Statistical methods
-------------------
All death rates and 95% confidence intervals (CIs) were calculated by direct standardisation to the European standard population for men and women by age group for the selected years.[@b28] Standardised rates were also calculated for population groups defined by NS-SEC group or quintile of area deprivation. All rates reported were per million population per year. P values indicating the significance of trends over time or across deprivation quintiles were calculated using Poisson regression. Tests for trends were not appropriate for the analysis by NS-SEC group since this classification is not hierarchical or ordered[@b21]; therefore, tests of the differences between groups were conducted instead. Rate ratios and 95% CIs were calculated to show the extent of inequality at each time by making comparisons between the populations living in each of the most and least deprived quintiles and the population living in the middle three quintiles.[@b29] P values indicating the significance of differences in the trends over time (from 1981 to 2001) in deaths in the most or least deprived quintile (assessed using the Carstairs score), relative to the population in the middle three quintiles, were calculated using Poisson regression. Finally, we calculated the slope index of inequality across quintiles of the income domain of the SIMD by age and sex for 2000--2002, for all causes and for homicides, and used this to assess the proportion of the inequality in mortality that was attributable to homicide.[@b30] We chose to examine the relative contribution of homicides to inequalities in all-cause mortality for the period 2000--2002 because this was the most recent period for which we were able to use populations derived from the census rather than from estimates.
Results
=======
Based on the analysis of 1109 deaths due to assault in the 12 years covered, [table 2](#tbl2){ref-type="table"} presents death rates by age and sex, with deaths due to assault further broken down into those due to the use of knives and other sharp objects and other assaults. Also shown are the total number of deaths by cause and sex in each 3-year period and p values indicating the significance of trends over time. Death rates under the age of 1 remained high, supporting an analysis of infanticide that is separate from other forms of homicide.[@b11] Male rates tended to be higher between the ages of 15 and 44 than at other ages, and at these ages, the rate more than doubled between 1980--1982 and 2003--2005. Most of these deaths and most of the increase can be seen to be attributable to assaults involving sharp objects; however, the increase in all age death rate associated with other assaults also showed a significant increase. Female death rates were much lower than male rates and were not dominated by sharp objects as male rates were. All age death rate showed a significant decrease over time due to decreases at older ages.
######
Male and female standardised death rates per 1 000 000 population due to assault, assault involving sharp weapons and other assaults, Scotland, 1980--1982, 1990--1992, 2000--2002 and 2003--2005
Age group Assault Sharp weapon Other assault
----------- ----------- -------------- --------------- ----- ---------- ---------- ---------- ---------- ----- --------- ----------- ---------- ---------- ----- ---------
Male
\<1 40 70 59 37 0.973 0 0 0 0 40 70 59 37 0.973
(1--80) (18--122) (7--110) (0--79) (1--80) (18--122) (7--110) (0--79)
1--4 13 15 9 3 0.209 3 0 0 0 10 15 9 3 0.346
(2--24) (3--27) (0--19) (0--9) (0--7) (0--20) (3--27) (0--19) (0--9)
5--14 3 4 2 2 0.473 1 1 0 0 0.324 2 3 2 2 0.765
(0--6) (0--8) (0--5) (0--5) (0--2) (0--3) (0--5) (0--7) (0--5) (0--5)
15--24 32 55 58 59 0.002 21 37 44 45 \<0.001 11 18 14 14 0.574
(22--42) (41--69) (42--73) (44--74) (13--29) (26--48) (31--57) (32--58) (5--16) (10--26) (6--21) (7--21)
25--34 25 48 44 55 0.003 16 37 34 40 0.004 9 12 10 15 0.401
(15--34) (36--61) (31--57) (40--70) (8--23) (26--47) (23--46) (27--52) (3--15) (6--18) (4--16) (7--23)
35--44 25 41 52 58 \<0.001 12 19 34 34 \<0.001 13 22 18 24 0.197
(15--36) (29--53) (39--65) (44--72) (5--20) (10--27) (23--45) (23--45) (6--21) (13--31) (10--25) (15--33)
45--59 22 27 32 41 0.008 9 13 12 16 0.160 13 14 20 24 0.024
(14--30) (18--36) (23--41) (31--51) (4--14) (7--19) (6--17) (10--22) (7--20) (8--21) (13--27) (17--33)
60+ 16 16 14 11 0.294 4 7 3 1 0.175 12 9 10 10 0.695
(8--24) (9--23) (7--20) (5--17) (1--8) (2--12) (0--6) (0--3) (5--19) (4--15) (5--16) (4--15)
15--44 27 48 51 57 \<0.001 16 31 38 40 \<0.001 11 18 14 18 0.115
(22--33) (41--56) (43--59) (49--66) (12--21) (25--37) (31--44) (32--47) (7--15) (13--22) (10--18) (13--23)
All ages 20 31 32 35 \<0.001 10 17 19 20 \<0.001 11 14 13 15 0.057
(17--24) (27--35) (28--36) (31--39) (7--12) (14--20) (15--22) (17--23) (8--13) (11--17) (11--16) (12--18)
Deaths 147 226 228 256 74 129 137 147 73 97 91 109
Female
\<1 43 11 37 39 0.967 0 0 0 0 43 11 37 39 0.967
(1--84) (0--31) (0--79) (0--83) (1--84) (0--31) (0--79) (0--83)
1--4 3 5 12 3 0.379 0 0 0 3 3 5 12 0 0.608
(0--8) (0--13) (0--25) (0-10) (0--10) (0--8) (0--13) (0--25)
5--14 3 2 2 3 0.803 2 0 0 1 0.416 2 2 2 2 0.792
(0--6) (0--5) (0--5) (0--7) (0--4) (0--3) (0--4) (0--5) (0--5) (0--5)
15--24 12 11 8 8 0.347 3 4 3 4 0.766 9 7 5 4 0.168
(6--18) (5--17) (3--14) (3--14) (0--6) (0--8) (0--7) (0--8) (4--14) (2--11) (1--10) (0--8)
25--34 7 8 11 7 0.652 6 4 5 5 0.988 2 3 6 2 0.483
(2--13) (3--12) (5--17) (2--13) (1--10) (1--8) (1--10) (1--10) (0--4) (0--7) (1--10) (0--5)
35--44 17 12 13 6 0.056 5 2 2 6 0.959 12 11 11 1 0.024
(9--26) (6--19) (7--20) (2--11) (1--10) (0--5) (0--5) (1--10) (5--19) (4--17) (5--17) (0--2)
45--59 14 8 8 5 0.018 2 3 4 0 0.583 12 5 4 5 0.016
(8--20) (3--13) (3--13) (2--9) (0--5) (0--6) (1--7) (6--18) (1--9) (1--7) (2--9)
60+ 11 9 3 3 0.002 2 1 2 1 0.515 9 8 1 2 0.002
(6--16) (4--13) (1--6) (1--6) (0--4) (0--2) (0--4) (0--2) (4--13) (4--13) (0--3) (0--5)
15--44 12 10 11 7 0.130 5 3 4 5 0.886 7 7 7 2 0.038
(8--16) (7--14) (7--15) (4--10) (2--7) (1--5) (2--6) (3--8) (4--10) (4--10) (4--10) (1--4)
All ages 11 8 8 6 \<0.001 3 2 3 3 0.708 8 6 6 3 \<0.001
(9--13) (6--10) (6--11) (4--8) (2--4) (1--3) (1--4) (1--4) (6--10) (4--8) (4--8) (2--5)
Deaths 83 65 61 43 24 16 21 20 59 49 40 23
p Values show significance of tests for trends over time.
[Table 1](#tbl1){ref-type="table"} shows death rates by age and NS-SEC for men in Scotland in 2000--2002. The table also shows the number and proportion of the population in each NS-SEC category. Steep social gradients are evident, with the rate among those in routine occupations (127 per 1 000 000 population) being nearly 12 times that of those in higher managerial and professional occupations (11). The gradient was steeper for those aged 20--39 than at ages 40--59 and was also steeper for assaults involving sharp objects. [Table 1](#tbl1){ref-type="table"} highlights the fact that the 193 473 men in routine occupations form a particularly vulnerable group, with rates being more than double those for men in semiroutine occupations. Differences between the NS-SEC groups were significant for both age groups and for assaults involving sharp objects and other assaults. High rates are also evident for the 9% of men whose occupations could not be classified (particularly at ages 20--39), but the rather low rate for the 4% who had never worked or were long-term unemployed illustrates the extent of our concerns about the coding to NS-SEC differing between death certificates and the census. The information collected on death records in Scotland is not able to distinguish those who have never worked or are long-term unemployed with any accuracy; similar findings have been reported for England and Wales.[@b31]
[Table 3](#tbl3){ref-type="table"} examines inequalities in male and female deaths according to area-based deprivation and how these have changed over time. The measures used are rate ratios comparing the rates in each of the most deprived quintile (quintile 5) and the least deprived quintile (quintile 1) with the rates in the middle three quintiles. Changes in the rate ratios for all assaults between 1980--1982 and 2000--2002, based on the Carstairs score for postcode sectors, showed that the relative experience of those living in the most deprived quintile little changed with male rates typically 3--4 times those in the middle three quintiles and female rates 2--2.5 times. In contrast, the rate ratio for men of all ages in the least deprived quintile fell from 0.86 in 1980--1982 to 0.28 in 2000--2002 (p=0.022). Relative to the middle three quintiles, the rate ratio in the most deprived quintile for assaults involving sharp objects decreased over time (p=0.035), although in 2000--2002, the rate ratio was still 3.38 (95% CI 2.27 to 5.03). No other trends were significant. When area deprivation was assessed for data zones using the SIMD income domain in 2000--2002, there was a notable tendency for inequalities to be wider than for postcode sectors assessed using the Carstairs score in the same period, with the rate ratios assessed using the SIMD for the most deprived areas mostly higher and those for the least deprived areas tending to be lower. This is likely to be a result of the smaller area employed by the SIMD with its consequent ability to identify pockets of deprivation set in (relatively) more affluent surroundings and the consequent homogeneity afforded by a classification based on data zones.
######
Age-specific death rate ratios due to assault, assault involving sharp weapons and other assaults, for most and least deprived quintiles relative to the middle three quintiles (men and women, Scotland, 1980--1982, 1991--1992 and 2000--2002 based on Carstairs scores; 2000--2002 and 2003--2005 based on SIMD income domain)
Assault Sharp weapons Other assault
-------------- -------------- --------------- --------------- --------------- --------------- --------------- --------------- --------------- --------------- -------------- --------------- --------------- --------------- ---------------- ---------
Male
15--39
Dep quin 1 0.52 0.37 0.14 0.17 0.20 0.68 0.55 0.15 0.17 0.29 0.40 0.14 0.13 0.17 0.00
(0.20--1.35) (0.16--0.87) (0.04--0.46) (0.05--0.53) (0.06--0.64) (0.19--2.37) (0.21--1.43) (0.04--0.60) (0.04--0.69) (0.09--0.95) (0.09--1.75) (0.02--1.06) (0.02--1.00) (0.02--1.27)
Dep quin 5 4.82 2.96 3.02 5.48 5.17 7.64 4.24 3.38 6.21 5.10 2.61 1.34 2.16 3.86 5.33
(3.04--7.65) (2.00--4.39) (2.14--4.25) (3.83--7.84) (3.66--7.30) (4.03--14.48) (2.61--6.89) (2.27--5.03) (4.07--9.49) (3.35--7.76) (1.27--5.34) (0.65--2.79) (1.08--4.31) (1.94--7.65) (2.90--9.77)
p Value \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.002 0.005 \<0.001 \<0.001 \<0.001
p~1~ 0.090 0.076 0.481
p~5~ 0.122 0.035 0.727
40--59
Dep quin 1 0.29 0.11 0.43 0.09 0.59 0.00 0.00 0.52 0.00 0.49 0.51 0.20 0.39 0.17 0.68
(0.07--1.26) (0.01--0.81) (0.17--1.12) (0.01--0.67) (0.26--1.33) (0.11--2.38) (0.14--1.70) (0.11--2.33) (0.03--1.54) (0.12--1.32) (0.02--1.32) (0.23--2.03)
Dep quin 5 2.61 2.47 3.88 4.89 5.93 3.23 2.66 6.78 4.78 6.12 2.14 2.31 2.45 4.99 5.74
(1.36--5.03) (1.26--4.82) (2.39--6.30) (3.05--7.85) (3.84--9.16) (1.25--8.38) (0.99--7.14) (3.19--14.40) (2.42--9.47) (3.33--11.27) (0.86--5.33) (0.93--5.74) (1.25--4.79) (2.59--9.62) (3.09--10.66)
p Value \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.003 \<0.001 \<0.001 \<0.001 0.104 0.006 \<0.001 \<0.001 \<0.001
p~1~ 0.525 0.980 0.829
p~5~ 0.274 0.175 0.811
All ages
Dep quin 1 0.86 0.25 0.28 0.27 0.39 0.46 0.35 0.21 0.11 0.34 1.12 0.16 0.35 0.47 0.47
(0.49--1.49) (0.12--0.51) (0.15--0.52) (0.13--0.56) (0.21--0.71) (0.16--1.33) (0.14--0.89) (0.08--0.58) (0.03--0.46) (0.15--0.79) (0.58--2.16) (0.05--0.51) (0.16--0.77) (0.20--1.11) (0.20--1.11)
Dep quin 5 3.71 2.31 3.24 5.19 5.50 5.36 3.47 4.07 5.68 5.23 2.59 1.38 2.34 4.57 5.89
(2.64--5.21) (1.69--3.15) (2.49--4.22) (3.97--6.80) (4.24--7.14) (3.27--8.77) (2.28--5.28) (2.88--5.75) (4.00--8.09) (3.72--7.37) (1.59--4.22) (0.84--2.26) (1.54--3.56) (3.00--6.94) (3.93--8.80)
p Value \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.001 \<0.001 \<0.001 \<0.001 \<0.001
p~1~ 0.022 0.250 0.064
p~5~ 0.680 0.376 0.962
Female
15--39
Dep quin 1 0.15 0.24 0.37 0.21 0.61 0.00 0.00 0.00 0.00 1.08 0.29 0.29 0.62 0.29 0.00
(0.02--1.11) (0.03--1.88) (0.09--1.64) (0.03--1.60) (0.13--2.79) (0.22--5.37) (0.04--2.31) (0.04--2.30) (0.13--2.87) (0.04--2.24)
Dep quin 5 2.32 1.35 2.51 3.09 2.28 1.52 4.51 2.89 6.22 2.22 3.10 0.72 2.27 1.95 2.37
(1.15--4.67) (0.51--3.60) (1.20--5.28) (1.49--6.40) (0.93--5.62) (0.47--4.92) (0.75--27.01) (0.93--8.96) (1.87--20.65) (0.68--7.27) (1.26--7.63) (0.20--2.62) (0.84--6.09) (0.74--5.13) (0.59--9.46)
p Value \<0.001 0.027 0.003 \<0.001 0.049 0.077 0.155 0.004 0.005 0.253 0.001 0.085 0.151 0.011 0.091
p~1~ 0.426 1.000 0.537
p~5~ 0.943 0.314 0.477
40--59
Dep quin 1 0.44 0.54 0.58 0.00 0.90 0.00 0.00 0.66 0.00 0.00 0.51 0.88 0.51 0.00 2.39
(0.10--1.93) (0.06--4.60) (0.13--2.69) (0.09--8.62) (0.07--5.94) (0.11--2.26) (0.09--8.47) (0.06--4.39) (0.15--38.27)
Dep quin 5 2.02 2.56 2.48 5.17 7.90 5.08 1.64 0.85 3.48 1.56 1.54 3.14 3.83 6.45 18.50
(0.88--4.68) (0.69--9.53) (0.93--6.67) (2.00--13.34) (2.04--30.56) (0.85--30.40) (0.15--18.12) (0.10--7.63) (0.70--17.24) (0.14--17.18) (0.58--4.11) (0.63--15.58) (1.17--12.54) (1.94--21.42) (2.23--153.67)
p Value 0.011 0.142 0.020 \<0.001 0.007 0.046 0.638 0.460 0.074 0.222 0.095 0.150 0.022 0.003 0.016
p~1~ 0.706 0.460 0.866
p~5~ 0.648 0.199 0.171
All ages
Dep quin 1 0.37 0.53 0.55 0.26 0.57 0.00 0.00 0.40 0.31 0.86 0.49 0.66 0.63 0.24 0.29
(0.16--0.86) (0.20--1.39) (0.24--1.26) (0.08--0.84) (0.17--1.93) (0.09--1.81) (0.04--2.53) (0.18--4.03) (0.21--1.17) (0.25--1.77) (0.23--1.67) (0.06--1.02) (0.04--2.28)
Dep quin 5 2.14 2.39 2.17 3.78 4.63 2.81 3.50 2.07 5.73 3.26 1.90 2.12 2.22 3.15 5.96
(1.37--3.33) (1.28--4.48) (1.29--3.66) (2.28--6.26) (2.54--8.44) (1.26--6.28) (0.94--13.04) (0.83--5.16) (2.29--14.37) (1.29--8.25) (1.12--3.24) (1.03--4.36) (1.17--4.19) (1.71--5.82) (2.70--13.13)
p Value \<0.001 0.001 \<0.001 \<0.001 \<0.001 \<0.001 0.076 0.006 \<0.001 0.011 \<0.001 0.007 0.003 \<0.001 \<0.001
p~1~ 0.269 0.581 0.514
p~5~ 0.582 0.771 0.408
p Values show significance of tests for trends over deprivation quintiles (p) and significance of tests of differences in trends over time, from 1981 to 2001, in the least (p~1~) and most (p~5~) deprived quintile (assessed using the Carstairs score), relative to the population in the middle three quintiles.
[Figure 1](#fig1){ref-type="fig"} shows how deaths that are due to assault vary as a proportion of all deaths by age for men in 2000--2002. Although representing just 0.3% of male deaths at all ages, assault accounts for 1.0% of deaths under the age of 65 and 3.2% between the ages 15 and 44. Also shown in [figure 1](#fig1){ref-type="fig"} is the percentage of inequalities in mortality at each age that is due to assault. The fact that the contribution to inequalities is greater than the proportion of deaths reflects the steeper social gradient that is seen for deaths due to assault than for other causes. Assault accounts for 1.1% of inequalities in male mortality at all ages, rising to 2.4% under 65 and 6.4% between the ages of 15 and 44.
{#fig1}
Discussion
==========
Death rates in Scotland due to assault are high; all age rates in [table 2](#tbl2){ref-type="table"} of 32 and 8 per 1 000 000 men and women respectively in 2000--2002 compare very unfavourably with rates of 14 and 6 per 1 000 000 for high income countries in the European region in 2000.[@b32] The steep social gradient is a notable feature of mortality due to assault in Scotland as in other countries. The extremely high death rate seen among men of working age in routine occupations (127 per 1 000 000) is comparable to the Scottish death rate for stroke among men of this age and higher than the death rate for colorectal cancer.[@b2] Although it is not possible to make direct comparisons between [tables 1](#tbl1){ref-type="table"} and [3](#tbl3){ref-type="table"}, inequalities (as measured by the ratio) between the most and least deprived quintiles as measured by SIMD appear stronger than those between routine occupations and higher managerial and professional occupations. For all assaults in 2000--2002, the lowest rate ratio between the most and least deprived quintile by age group was that seen for all ages (RR=5.19/0.27=19.2); by contrast, the highest rate ratio between the extreme groups of NS-SEC was 13.5 (=150/11) for those aged 20--39. While it is possible that the results would differ if other individual and area socioeconomic factors were employed, we believe that this work suggests that contextual influences of the neighbourhood of residence might be more important than individual characteristics in determining the victims of assault. However, we have previously noted the mismatch between the coding of NS-SEC from death certificates and that from the census[@b2] and, for this reason, are cautious about placing too much emphasis on the interpretation of the results by individual socioeconomic status. Furthermore, formal separation of the contextual and compositional effects can only be conducted in a multilevel framework.
Death rates due to assault in the most deprived areas have not deteriorated significantly over time relative to those in the middle three quintiles. The only significant increase in inequalities has been the widening of the gap between the least deprived areas and the rest of Scotland. However, the magnitude of the inequalities seen---patterned by the deprivation level of the area of residence---is extreme. Between 2000 and 2002, a man under 65 living in the most deprived quintile was 31.9 (95% CI 13.1 to 77.9) times more likely to die due to assault than one living in the most affluent quintile; for women, this ratio was 35.0 (4.8 to 256.2). The equivalent rate ratios for all-cause mortality were 3.5 (3.3 to 3.6) and 2.6 (2.5 to 2.8) for men and women, respectively. Not only do these inequalities for assault exceed those for other causes of death in Scotland, but also they far exceed the ratio reported for homicide in Great Britain[@b15] (5.7 comparing the top and bottom deciles, 1996--2000) and that for emergency hospital admissions for assault in England[@b33] (6.3 comparing the top and bottom quintiles in 2005--2006). The reasons for the different findings are unclear; the measures reported in these studies have some differences (eg, both refer to men and women combined and to all ages), but these could not account for the stark inequalities observed in Scotland. Earlier research has suggested that inequalities in all-cause mortality are wider in Scotland than in the rest of Great Britain; the Scottish excess mortality in 2001 ranged from 2% in the least deprived decile to 17% in the most.[@b34] It is possible that the greater inequalities in mortality due to assault in Scotland reflect greater social inequalities and more acute deprivation. The prioritisation of the harm caused by violence in the Scottish Government\'s report of the ministerial task force on health inequalities[@b35] is recognition that the small numbers of deaths have a marked impact on inequalities, particularly among young men.
Maxwell *et al* showed that while in a single year from April 2004 to March 2005, 4891 patients were admitted to hospital in England with an assault-related stabbing injury, there were only 139 deaths in 2004 due to assault from a sharp object.[@b36] This figure of 35 victims requiring hospital treatment for every homicide fits in with an estimated range of 20--40 for youth violence internationally.[@b37] The suggestion that there might be so many hospital admissions for every death means that the analysis presented in this paper probably represents no more than the tip of the iceberg in terms of the number of events. Throughout Scotland, it has been estimated that "Violence costs the Scottish economy around £3 billion each year in healthcare, law enforcement and lost productivity".[@b38]
What are the known correlates of violent behaviour? Alcohol and drug use have been shown to contribute to homicides and to violence more generally. Shaw *et al* estimated that alcohol or drugs (taken by the perpetrator) contributed to 45% of a national sample of homicides in England and Wales collected over 3 years between 1996 and 1999,[@b39] although they believed this figure to be conservative. A similar level of alcohol use was reported in several international studies of homicide.[@b40] More generally, alcohol use has been shown to contribute to violent behaviour among adolescents,[@b41] while in Scotland school children who used drugs were more likely to carry weapons.[@b42] Social correlates have also been promoted; Heath[@b43] suggested that carrying a knife is another dimension of poverty. The existence of inequalities between social groups has been recognised as an important risk factor for violence,[@b44] and an ecological cross-national study showed a negative correlation between gross national product per capita and a country\'s homicide rate.[@b45]
So what can be done about the high homicide rate and the associated inequalities? Brookman and Maguire[@b40] made a range of policy recommendations for the reduction of homicides in the categories of domestic homicide, the killing of infants, alcohol-related homicide and homicide with guns or knives. The Scottish Government is already seeking ways to reduce alcohol-related problems including violence.[@b46] As a means of reducing knife crime, Hern *et al*[@b20] advocated a ban on the sale of long pointed knives, something that could be considered in Scotland given the considerable role that knives play. For interventions to reduce youth violence to succeed, Kellermann *et al*[@b47] recommended that they should take place before the age of 6 years and must either address multiple risk factors simultaneously or else should be tailored for specific target groups. Addressing the gang culture is an example of such a targeted approach.[@b48] However, the extent of the inequalities seen for assault in Scotland, coupled with high death rates for causes associated with alcohol and drug use and mental wellbeing among the most disadvantaged groups, emphasises the need to understand and address the multiplicity of problems associated with deprivation and poverty.
Our study has demonstrated the extent of inequalities in deaths due to assault in Scotland and the role played by knives and other sharp weapons. Although we have no information on the perpetrators of the assaults from death records, we have offered a comprehensive insight into the inequalities that occur and the extent to which they are patterned by socioeconomic circumstances.
###### What is already known on this subject
- Scotland has a high death rate due to assault compared to other countries.
- Mortality from assault is known to be higher among more disadvantaged populations.
###### What this study adds
- Social inequalities in mortality due to assault in Scotland are of far greater magnitude than those reported elsewhere.
- The scale of these inequalities means that any attempt to address inequalities in mortality in Scotland---particularly among young men---must tackle the problem of assault, particularly that involving knives and other sharp weapons.
We are grateful to the General Register Office (Scotland) for the provision of data and to the referees for their helpful comments.
**Funding:** The Social and Public Health Sciences Unit is jointly funded by the Medical Research Council (MRC) and the Chief Scientist Office (CSO) of the Scottish Government Health Directorates. This work was funded by the Chief Scientist Office as part of the "Measuring health, variations in health and the determinants of health" programme, wbs U.1300.00.001. Neither the MRC nor the CSO had any role in study design, analysis or the interpretation of the results.
**Competing interests:** None.
**Contributors:** Both authors contributed to the study design and analysis. AHL wrote the first draft of the paper. Both authors contributed to subsequent drafts and approved the final version. AHL is guarantor for the paper.
**Ethical approval:** This study was based on the analysis of anonymised mortality data and as such did not require ethical approval.
**Provenance and peer review:** Not commissioned; externally peer reviewed.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
The majority of healthcare in the United States is delivered in a fee for service (FFS) model and this delivery model has likely contributed to the uncontrollable rise in healthcare costs in the United States \[[@CR1]--[@CR3]\]. Under this system, healthcare providers may be financially incentivized to provide a greater volume of services without regard for improved outcomes, provider performance or quality. In fact, utilization and medical costs for spine conditions have increased under this system but these increases in cost did not correlate with improvements in health \[[@CR4]\]. Therefore there is a need to identify and prioritize management strategies that demonstrate high "value proposition" \[[@CR5], [@CR6]\] by being effective for the patient, cost efficient for the payer and provided at the optimal time during an episode of pain for enhanced benefit. Indeed, estimating the value proposition of services provided by physical therapists for common musculoskeletal pain conditions has been highlighted as a priority by the profession \[[@CR5]\] and in draft documents for the National Pain Report \[[@CR7]\].
Recent research supports that early physical therapist management for low back pain is recommended to improve outcomes in comparison to other management pathways such as advanced imaging, prescription medication or advanced care \[[@CR8], [@CR9]\], and can lead to lower downstream healthcare utilization and costs \[[@CR10], [@CR11]\]. This recommendation differs from previous recommendations where management is delayed to account for potential spontaneous recovery and advised only for patients that are resistant to recovery \[[@CR12]--[@CR14]\]. Despite neck pain being the second most common musculoskeletal disorder after low back pain \[[@CR15], [@CR16]\] and physical therapists being the most frequently visited healthcare provider for neck pain \[[@CR17]\], the effect of early physical therapist management on outcomes has not been investigated in patients with neck pain.
Neck pain is one of the most common reasons for entry into the healthcare system \[[@CR18]\] and disability related to neck pain has an enormous impact on individuals and their families, communities and healthcare systems \[[@CR19]--[@CR22]\]. Therefore there is a need to prioritize management strategies to that improve outcomes and decrease the financial burden of neck pain. To examine the value proposition of early physical therapist management of neck pain, the metrics of value and efficiency for physical therapist management need to be explored. Porter (2010) defines value as the health outcomes achieved per dollar \[[@CR23]\]. If value improves, patients, payers, and providers can all benefit while the economic sustainability of the health care system increases \[[@CR23]\]. Moreover, the efficiency of delivery of care is important; that is, how quickly does a patient improve? Greater efficiency in managing neck pain may lead to a decrease in indirect costs and allow the provider manage the patient more effectively with less utilization of resources.
The purpose of the study was, therefore, to provide foundational information for determining the value proposition of early physical therapist management of neck pain. To address this purpose, we compared the odds of achieving a meaningful reduction in disability and pain and we compared the metrics of value and efficiency between patients receiving early physical therapist management and delayed physical therapist management to determine the impact of timing of physical therapist management. We hypothesized that patients who received physical therapist management earlier during an episode of neck pain would have the best odds of clinical improvement and the greatest value and efficiency for decreasing disability and pain.
Methods {#Sec2}
=======
Database {#Sec3}
--------
Data for this retrospective cohort analysis was extracted from the Rehabilitation Outcomes Management System (ROMS). This rehabilitation database contains demographics, patient characteristics, clinical outcomes and rehabilitation utilization data that is maintained by Intermountain Healthcare, a private, non-profit integrated healthcare system. The study protocol was approved by the Institutional Review Board of Intermountain Healthcare.
Patients {#Sec4}
--------
Patients included in this study received physical therapist management for the primary complaint of neck pain between dates January 1, 2008 to December 31, 2012 from 13 outpatient physical therapy clinics located in Salt Lake City, Utah and surrounding regions. Inclusion criteria for these analyses were: non-surgical patient, Neck Disability Index (NDI) score of 10 or greater and Numerical Pain Rating Scale (NPRS) score of 2 or greater at initial evaluation, 2 or more visits, duration of physical therapist management of less than 180 days and have a self-reported primary complaint of non-specific neck pain. The inclusion criteria were implemented to permit the evaluation of clinical outcomes \[[@CR24], [@CR25]\] that exceed measurement error \[[@CR26], [@CR27]\]. See Fig. [1](#Fig1){ref-type="fig"} for derivation of sample.Fig. 1Derivation of the sample
Symptom duration {#Sec5}
----------------
At initial evaluation, patients reported their primary complaint and neck pain symptom duration. This information was recorded by the physical therapist and entered into the ROMS database. Patients utilized in this analysis were identified in the ROMS database if they self-reported non-specific neck pain as their primary complaint. The patients were categorized as receiving "early physical therapist management" of neck pain if they reported non-specific neck pain with duration less than 4 weeks and "delayed physical therapist management" if they reported non-specific neck pain duration greater than 4 weeks. This categorization is supported in the low back pain literature where early physical therapist management has been defined as management within 2 weeks \[[@CR10], [@CR11]\] or within 4 weeks \[[@CR28]\] of either self-report of duration of symptoms or the time from primary care provider consultation to physical therapy consultation. Therefore in this study the authors utilized a modest estimate of a cut-off of duration of symptoms of 4 weeks to define early management, to be inclusive of patient symptoms consistent with acute and early sub-acute pain stages \[[@CR29]\].
Clinical outcomes {#Sec6}
-----------------
The primary clinical outcome measures used in these analyses were the Neck Disability Index (NDI) and the Numerical Pain Rating Scale (NPRS). The NDI is a condition specific outcome measure that is a commonly used outcome measure for people with neck pain and is the most studied and well-established of all the outcome measures for neck pain \[[@CR30]\] and it is found to be reliable and valid \[[@CR30]--[@CR34]\]. The NDI is comprised of 10 items; seven items related to activities of daily living, two items related to pain, and one item related to concentration \[[@CR35]\]. Each item is scored from 0 to 5. The total score is expressed as a percentage and is reflective of a level of disability related to neck pain where high percentages are related to higher disability. In addition to the NDI, the NPRS is also a commonly used outcome measure for patients with neck pain \[[@CR36]\]. The NPRS exhibits fair to moderate test-retest reliability in patients with mechanical neck pain and shows adequate responsiveness in this patient population \[[@CR26], [@CR36]\]. The NPRS is an 11 point scale, anchored with 0 rated as "no pain" and 10 rated as "worst pain" imaginable. Patients are asked to rate their current pain using this scale.
In order to determine whether patients experienced a meaningful improvement the Minimal Clinically Important Difference (MCID) was calculated for the NDI and NPRS. MCID values assist in determining the minimal amount of change that represents a clinically important difference for the patient. The most rigorous estimate of MCID for the NDI for patients with neck pain has been reported as a 19 percentage point change (9.5 raw score). Other estimates of MCID for patients with neck pain include a 10 percentage point change (5 raw score) \[[@CR37]\] and 15 percentage point change (7.5 raw points) \[[@CR35]\]. The authors used the MCID cut point of a 19 percentage point change because it is the most rigorous estimate and it exceeds the measurement of error for the NDI (10 percentage points) \[[@CR33]\]. The MCID for the NPRS in patients with neck pain is 1.3 points \[[@CR36]\] and this value was the cut point utilized for the MCID on the NPRS in this study.
Next, we wanted to determine the value proposition of early physical therapist management of neck pain by examining the metrics of value and efficiency. Value is defined as the change in outcome per dollar. In this study, we calculated two value metrics using the changes in NDI and on the NPRS during an episode of physical therapy. The metric of value for disability was calculated by dividing the change score on the NDI (Initial NDI score - NDI score at the last visit = change in NDI score) during an episode of care by the total charges for physical therapy multiplied by 100 during an episode of care ((Change in NDI/Charges for PT) x100)). This calculation allows for the interpretation of change in disability per 100 dollars. The metric of value for pain intensity was calculated by dividing the change score on the NPRS (Initial NPRS score - NPRS score at the last visit = change in NPRS score) during an episode of care by the total charges for physical therapy multiplied by 100 during an episode of care ((Change in NPRS/Charges for PT) x100)). Smaller numbers indicate lower value and larger numbers higher value.
A variable for efficiency of physical therapist management of neck pain for decreasing disability was also calculated and has been previously reported for musculoskeletal conditions \[[@CR38]\]. The authors also wanted to calculate the efficiency of physical therapist management of neck pain for decreasing pain. Efficiency in decreasing disability is calculated by dividing the change in NDI score by the total number of visits during an episode of care (Change in NDI/\# of visits to PT) \[[@CR38]\]. Efficiency in decreasing pain is calculated by dividing the change in NPRS score by the total number of visits during an episode of care (Change in NPRS/\# of visits to PT). Smaller values represent lower efficiency or less improvement per visit and larger values higher efficiency or more improvement in disability or pain per visit.
Data analysis {#Sec7}
=============
Descriptive statistics and clinical characteristics were calculated for all patients in this sample. Means and standard deviations were reported for normally distributed continuous variables, median and interquartile range for number of visits and percentages were reported for categorical variables. Baseline patient characteristics, clinical outcomes and utilization of physical therapy were compared among timing groups using chi-squared analyses for categorical variables and independent t-tests for normally distributed continuous variables or Mann Whitney U tests for non-normally distributed continuous variables.
The main outcomes of interest in this study were the odds of patients achieving MCID on the NDI and NPRS and the metrics of value and efficiency. Model selection for all analyses were performed using an initial set of potentially confounding variables based on previous literature on prognostic factors for neck pain such as age and gender \[[@CR39]\] and conceptually driven variables (baseline scores for pain and disability, duration of treatment and number of visits) \[[@CR40]\]. Variables were removed if they did not improve model fit as measured by the explained variance in the exploratory models.
Both unadjusted and adjusted analyses were performed with separate binary logistic regression models to compare the odds of achieving MCID on the NDI and the NPRS between patients receiving early physical therapist management and delayed physical therapist management. Covariates in the initial adjusted models included age, gender, and duration of treatment, number of visits and baseline NDI and NPRS scores. All covariates were retained in the final model with the exception of baseline scores on the NDI in the model estimating MCID on the NPRS and baseline scores on the NPRS in the model estimating MCID on the NDI. Odds ratios with 95 % confidence intervals were reported.
Separate generalized linear regression models, normal distribution with identity link function, were estimated to compare the value for disability and value for pain between patients receiving early physical therapist management and delayed physical therapist management. Covariates in the initial models included age, gender, number of visits, duration of treatment and baseline scores for the NDI and NPRS. All covariates were retained in the final model when estimating value for disability and when estimating value for pain, with the exception of baseline scores on the NDI, all covariates were retained in the final model. Separate generalized linear regression models, normal distribution with identity link function, were estimated to compare the efficiency of physical therapist management for both change in disability and change in pain per visit between patients with receiving early physical therapist management and delayed physical therapist management. Covariates in the initial models included age, gender, duration of treatment and baseline scores for NDI and NPRS. In the final models estimating efficiency for change in pain and change in NDI the covariates included age, gender, admission NPRS score and duration of treatment and baseline score of outcome of interest. Estimates of means and 95 % confidence intervals between groups were reported. Data analysis was performed using SPSS statistical software (version 21.0). Significance level for all analyses was set at 0.05.
Results {#Sec8}
=======
One thousand five hundred thirty-one patients with non-specific neck pain were identified from the database after exclusion criteria were met. See Fig. [1](#Fig1){ref-type="fig"}. Of these patients, 451 patients received early physical therapist management and 1080 patients received delayed physical therapist management. Patients receiving early physical therapist management overall were younger, had slightly more males and had greater percentages of patients achieving MCID on both the NPRS and the NDI. These patients also had slightly higher average charges per visit and for an episode of care but these were not significant differences compared to delayed physical therapist management. There was no difference in the median number of visits between groups. See Table [1](#Tab1){ref-type="table"}.Table 1Descriptive and clinical characteristics of the sampleEarly physical therapist management (Duration of symptoms \<4 Weeks)Delayed physical therapist management (Duration of symptoms \>4 Weeks)*Pn* = 451*n* = 1080Age46.23 (15.41)52.41 (16.77)0.07Gender (% Female)64.1 %70.5 %0.01Initial NDI Score40.18 (17.54)34.31 (15.22)\<0.001Discharge NDI Score23.86 (17.34)25.63 (16.31)0.06Initial NPRS Score5.96 (2.23)5.28 (2.14)\<0.001Discharge NPRS Score3.20 (2.50)3.57 (2.37)0.01Percentage of Patients Achieving MCID on the NDI38.8 %23.5 %\<0.001Percentage of Patients Achieving MCID on the NPRS68.1 %64.2 %\<0.001Number of Physical Therapy Visits (Median, IQR)5 (5)5 (4)0.82Total Charges for an Episode of Physical Therapy (\$USD)\$918.42 (\$667.47)\$882.73 (\$621.23)0.29Average Charges per Physical Therapy Visit\$162.56 (33.83)\$158.88 (40.07)0.08Abbreviations: *MCID* Minimal Clinically Important Difference, *NDI* Neck Disability Index, *NPRS* Numerical Pain Rating Scale, *IQR* Interquartile rangeAll values represent Mean (standard deviation) or percentages
The odds of achieving MCID or the amount change on an outcome measure that needs to occur for a patient to identify as important is reported in Table [2](#Tab2){ref-type="table"}. As hypothesized, patients who received early physical therapist management demonstrated increased odds of achieving MCID on the NDI (aOR = 2.01, 95 % CI 1.57, 2.56) and MCID on the NPRS (aOR = 1.82, 95 % CI 1.42, 2.38) compared to patients who received delayed physical therapist management.Table 2Unadjusted and adjusted estimates of odds of achieving MCIDUnadjusted Odds Ratio (OR)Adjusted Odds Ratio (aOR)Early physical therapist management (Duration of symptoms \<4 Weeks)Delayed physical therapist management (Duration of symptoms \>4 Weeks)Early physical therapist management (Duration of symptoms \<4 Weeks)Delayed physical therapist management (Duration of symptoms \>4 Weeks)Odds of achieving MCID NDI2.06 (1.63, 2.61)REF2.01 (1.57, 2.56)REFOdds of achieving MCID NPRS2.05 (1.63,2.59)REF1.82 (1.42, 2.38)REFAbbreviations: *MCID* Minimal Clinically Important Difference, *NDI* Neck Disability Index, *NPRS* Numerical Pain Rating ScaleValues represent odds ratio with 95 % Confidence Intervals
Patients who received early physical therapist management demonstrated better value in decreasing disability and pain compared to patients who received delayed physical therapist management. When controlling for covariates, patients receiving early physical therapist management demonstrated a 2.27 percentage point change (95 % CI 2.03, 2.51) in disability score per 100 dollars compared to patients who received delayed physical therapist management who demonstrated a 1.22 percentage point change in NDI per 100 dollars spent (95 % CI 1.06,1.38). Patients who received early physical therapist management also demonstrated a greater change in pain per 100 dollars (0.38 points, 95 % 0.34, 0.41) compared to patients who received delayed physical therapist management (0.28 points, 95 % CI 0.25, 0.30).
Patients who received early physical therapist management also experienced more efficiency in reducing disability and pain with physical therapist management of neck pain. The patients who received early physical therapist management achieved a 3.44 percentage point change in NDI score per visit (95 % CI 3.11, 3.78) and 0.57 point change (95 % CI 0.52, 0.62) in pain compared to a 1.81 percentage point change (95 % CI 1.58, 2.03) in NDI score and 0.42 point change (95 % CI 0.38, 0.45) in pain per visit when patients received delayed physical therapist management. Both adjusted and unadjusted mean estimates and 95 % CI are reported in Table [3](#Tab3){ref-type="table"}.Table 3Unadjusted and Adjusted estimates of mean value and efficiencyUnadjusted estimated eeanAdjusted estimated meanEarly physical therapist management (Duration of symptoms \<4 Weeks)Delayed physical therapist management (Duration of symptoms \>4 Weeks)Early physical therapist management (Duration of symptoms \<4 Weeks)Delayed physical therapist management (Duration of symptoms \>4 Weeks)Value-Disability (Change in NDI/100 dollars)2.42 (2.16,2.68)1.27 (1.12,1.41)2.27 (2.03,2.51)1.22 (1.06,1.38)Value-Pain (Change in NPRS/100 dollars)0.42 (0.38,0.46)0.26 (0.24,0.28)0.38 (0.34,0.41)0.28 (0.25,0.30)Efficiency-Disability (Change in NDI per Visit)3.65 (3.27,4.03)1.84 (1.64,2.05)3.44 (3.11,3.78)1.81 (1.58,2.03)Efficiency-Pain (Change in Pain per visit)0.64 (0.57,0.70)0.39 (0.35,0.42)0.57 (0.52,0.62)0.42 (0.38,0.45)Abbreviations: *NDI* Neck Disability Index, *NPRS* Numerical Pain Rating ScaleValues represent mean estimates with 95 % Confidence Intervals
Discussion {#Sec9}
==========
Improving value in musculoskeletal healthcare has emerged as an important objective and the Institute of Medicine (IOM) has made the recommendation that the public at large and people with pain, in particular, would benefit from a better understanding of pain and its treatment, in order to encourage timely care and improve medical management \[[@CR41]\]. Moreover the US healthcare system, which currently operates in FFS model, has contributed to an uncontrollable rise in healthcare costs \[[@CR1]--[@CR3]\]. This threatens the long-term performance of the healthcare delivery system and the rise in healthcare costs is widely viewed to be unsustainable economically \[[@CR6]\]. Payment models are evolving from FFS models to other payment strategies that include incentives for quality, outcomes, improved patient experience, and reduced costs; essentially paying for value, not volume \[[@CR42]\]. Therefore, there has been a call to action to determine the value proposition of care provided by healthcare providers, such as physical therapists, for management of musculoskeletal pain \[[@CR6]\]. Value proposition can be operationalized as demonstrating a reduction in disability, pain and improvement in health status of individuals through more cost-effective management of neck pain \[[@CR6]\]. The results of our study describe part of the value proposition for early physical therapist management of neck pain. In this cohort we determined that when neck pain is managed early, within the first four weeks of symptoms, the patient is more likely to have a meaningful reduction in disability and pain, will demonstrate greater value per dollar for physical therapy and experience more efficient care when compared to patients who received delayed management or when symptoms were present for greater than four weeks. Consistent with the key components of a value proposition, these results can be interpreted from the perspectives of three primary stakeholders; the patient, the payer and the provider.
The indirect costs for patients with neck pain can potentially supersede the direct costs to the healthcare system. A study in the Netherlands found that direct costs, such as healthcare, amounted to just 23 % of this figure while indirect costs, such as work absenteeism and disability, amounted to 77 % of the total costs \[[@CR43]\]. In this study, when patients received early physical therapist management they were twice as likely to actually experience a meaningful change in disability and were 1.8 times as likely to achieve a meaningful reduction in pain in comparison if they waited longer than 4 weeks to receive care. This finding supports that early management of neck pain has the potential to decrease the indirect costs associated with neck pain and improve the patient experience by decreasing disability, which can be significant \[[@CR44]\], in patients with neck pain. But further research is needed to examine the relationship between early physical therapist management and indirect costs associated with neck pain through formal economic evaluation.
When investigating the value of early physical therapist management of neck pain from the payer perspective, our findings are consistent with the literature which reports that earlier physical therapist management of spine pain can lead to less resource utilization \[[@CR10], [@CR39], [@CR45]\]. If examining charges alone in this cohort, although not significant, the patients receiving early physical therapist management incurred slightly higher charges for an episode of PT. But the value is realized when considering the decrease in disability and pain per dollars spent. Research supports that the level of disability at discharge is associated with both the short and long term prognosis for neck pain \[[@CR45], [@CR46]\]. Therefore it is cost beneficial for the payer to have a patient demonstrate a greater decrease in disability and pain in the short term with a few more dollars spent, then to increase the likelihood of recurrence \[[@CR36]\] with a potential for increased downstream utilization of healthcare resources long term.
From the provider perspective, this study supports that the most efficient care can be provided when a patient is seen within the first four weeks of symptoms in order to experience the greatest decreases in their disability and pain. Bridging these findings with the patient perspective, this means it takes approximately half the number of visits when patients receive early physical therapist management to achieve a meaningful reduction (ie MCID) on the NDI and NPRS compared to patients who receive delayed management to achieve the same decrease in disability on the NDI and NPRS. Some may argue that patients receiving early physical therapist management of neck pain, may have different care seeking behavior due to experiencing more severe symptoms or simply this group experienced a regression towards the mean that accounted for the greater change in score compared to delayed physical therapist management. Although patients who received early physical therapist management had higher baseline NDI scores, the difference in the NDI scores between groups did not exceed error \[[@CR36]\] nor were the groups categorized in different functional categories by their scores, i.e both categorized as having "moderate disability" on the NDI \[[@CR47]\]. Early physical therapist management of neck pain, if considered as a value driven management approach has the potential to decrease the number of visits needed for physical therapists to effectively manage neck pain, but can also lead to patients experiencing meaningful reductions in pain and disability over a shorter period of time. Further research is needed to weigh the potential benefit of early physical therapist management against over treatment in patients with neck pain.
There are strengths in the current study. The concept of value proposition of early physical therapist management of neck pain was discussed from three perspectives: the patient, payer and provider. The benefit of early physical therapist management was potentially realized from all stakeholders. The policy implications of these findings provide support for early physical therapist management of neck pain. An additional strength of the study was the introduction of the metric of value specifically related to physical therapist management of neck pain. The metric of value calculated in this study is consistent with the definition of value in healthcare services, where value is defined the health outcomes achieved per dollar \[[@CR23]\]. We examined the value of physical therapist management of neck pain utilizing the clinical outcomes of disability and pain. Interpreting these findings provides a meaningful way to communicate to stakeholders what the "return investment" for early physical therapist management of neck pain may be for systems that provide pathways for early management.
There are also limitations in the current study. The authors discuss the value of early physical therapist management of neck pain, not value of early physical therapist management of neck pain and its effect on downstream healthcare costs and utilization. Although this is a limitation, recent research in low back pain supports that providing early physical therapist management after consultation with a primary care provider was not associated with an increase in costs or utilization of specific services \[[@CR48]\]. Therefore the authors feel that the potential benefit of early physical therapist management exceeds the risk of unnecessary or inefficient use of healthcare services. Moreover, this study only examined the effect of timing on outcomes and did not dissect the differences in actual treatments received by patients in each group. The authors were also limited to using charges for physical therapy rather than costs which are known to differ. Though there is the potential that the costs may be lower, the interpretation of the findings remains the same, which is the ratio of change in outcome measure over dollars. Lastly, this study was a retrospective cohort analysis where there is the potential for bias due to study design. Research suggests that are factors such as recurrence \[[@CR49]\], severity \[[@CR44]\], treatment preferences \[[@CR50], [@CR51]\] and psychosocial factors such as depressive symptoms \[[@CR52]\] catastrophizing \[[@CR53]\] and fear avoidance beliefs \[[@CR54]\] that can adversely affect outcomes in patients with spine pain. These factors, which were unmeasured in this study could have introduced bias in patient grouping and difference in outcomes between groups. The authors acknowledge this potential for bias and attempted to minimize this bias with the inclusion of known prognostic factors and baseline pain and disability status where appropriate from the available variables.
Further research in needed to examine the effect of early physical therapist management of neck pain relative to downstream costs and utilizations and to make specific recommendations on interventional strategies based on timing of physical therapist management of neck pain.
Conclusions {#Sec10}
===========
Policy makers and payers contribute to the value equation by designing health policies that promote access and use of timely, appropriate healthcare services \[[@CR6]\]. In this study we provided evidentiary support for early management of neck pain where the patient, payer and provider may benefit. The implication of these findings suggest that healthcare systems that provide pathways for patients to receive early physical therapist management of neck pain may realize improved patient outcomes, increased efficiency in delivery of care and greater value. Further research is needed to determine the overall impact of promoting early physical therapist management of neck pain on the entire healthcare experience to elucidate the benefit relative to other management pathways.
Abbreviations {#Sec11}
=============
aOR, Adjusted odds ratio; FFS, Fee for service; IOM, Institute of Medicine; MCID, Minimal Clinically Important Difference; NDI, Neck disability index; NPRS, Numerical Pain Rating Scale; NSNP, Non-specific neck pain; OR, Odds Ratio; PT, Physical Therapy; USD, United States Dollar.
We would like to acknowledge Matt Speckman and Diane Tracy for their contribution in data retrieval for this study. MH is partially supported by the New Investigator Fellowship in Training Initiative from the Foundation for Physical Therapy.
Funding {#FPar1}
=======
No internal or external funding was provided for this study.
Availability of data and materials {#FPar2}
==================================
Data used in this study will not be shared outside Intermountain in accordance with protections provided by the HIPAA privacy information policy.
Author's contributions {#FPar3}
======================
MH and GB contributed to the data retrieval. MH, GB, JH, SG, and MB participated in planning the data analysis. MH performed the data analysis and drafted the manuscript. All authors contributed to the final draft of the manuscript. All authors read and approved the final manuscript.
Competing interests {#FPar4}
===================
The authors declare that they have no competing interests.
Consent for publication {#FPar5}
=======================
Not applicable
Ethics approval and consent to participate {#FPar6}
==========================================
The study protocol was approved by the Institutional Review Board of Intermountain Healthcare.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Abdominal aortic aneurysm (AAA), an abnormal focal dilation of the abdominal aorta, is a common and potentially life-threatening condition \[[@CR1]\]. Only a half of patient with ruptured AAA can reach the hospital alive with an additional high operative mortality of 49 % \[[@CR2]\]. In asymptomatic patients with AAA, elective repair of AAA is the most effective management to prevent the rupture of AAA with a low procedure mortality \[[@CR3], [@CR4]\]. Thus, early identification and elective repair of AAA is clinically important. Most of the patients with rupture have an undiscovered AAA \[[@CR5]\], because AAA is mostly asymptomatic and occult on physical examination. Abdominal ultrasound, a fast and safe screening method to detect AAA with high sensitivity and specificity, is recommended for screening the patients who are at high risk of AAA \[[@CR6]\]. Although screening for AAA during transthoracic echocardiography (TTE) has been reported to be clinically useful in Western populations \[[@CR7]--[@CR16]\], little is known about the important TTE indices associated with the presence of AAA \[[@CR12], [@CR14], [@CR16]\]. Accordingly, the aim of this study was to evaluate the clinical utility of TTE for screening AAA and to identify important TTE indices associated with AAA in a Japanese population.
Methods {#Sec2}
=======
Study population {#Sec3}
----------------
We prospectively examined 1912 consecutive patients who were referred for routine TTE from June 2013 to May 2014 at Kochi Medical School Hospital. Clinical data were obtained by chart reviews. The study was approved by the Ethics Committee on Human Research of the Kochi Medical School and all the patients gave written informed consent before the study. Patients with prior AAA repair or aortic dissection were excluded from the study.
Echocardiography {#Sec4}
----------------
Echocardiographic evaluation of the aorta was performed by 3 registered sonographers (M.W., D.H., and Y.Y.) in our echocardiographic laboratory. Routine transthoracic 2-dimensional, M-mode, and Doppler echocardiography were performed as recommended by the American Society of Echocardiography \[[@CR17]\]. The aortic root size was measured at the level of the sinus of Valsalva from the parasternal long-axis view at the onset of the QRS complex. We used 4 echocardiographic equipments as follows: 1) Vivid7 (General Electric, Horten, Norway) with a 3S transducer (1.5 to 3.6 MHz variable-frequency), 2) iE33 (Philips, Andover, MA, USA) with an S5-1 transducer (1 to 5 MHz variable-frequency), 3) Vivid Q(General Electric, Horten, Norway) with a M4S transducer (1.5 to 3.6 MHz variable-frequency), 4) Prosound α 7 (Hitachi-Aloka Medical, Tokyo, Japan) with a UST-52105 transducer (3.5 to 5.0 MHz variable-frequency transducer).
In succession to routine TTE, the abdominal aorta was visualized in the supine position using the same cardiac transducer within a short period of time. First, longitudinal image of the abdominal aorta was visualized. Secondly, transverse image of the abdominal aorta was visualized. The abdominal aorta was scanned from the subcostal position and then traced distally as far as possible. Size of the abdominal aorta was measured using still-frame images and 'on-line' video calipers in diastole at the onset of the QRS complex at the longitudinal and transverse images. The maximal size of the abdominal aorta ≥ 30 mm in either anteroposterior or lateral size was defined as AAA. Clinical information was collected using the TTE requisition slip and the electrical recording system. In randomly selected 15 patients, we measured the additional examination time for screening AAA.
Statistical analysis {#Sec5}
--------------------
Categorical variables were presented as total number and % of patients, and continuous variables were presented as means ± standard deviation. Clinical and TTE indices between the 2 groups were compared with chi-square test for categorical variables and Wilcoxon rank-sum test for continuous variables. Correlation between the aortic root and the abdominal aorta sizes was evaluated using linear regression analysis. Receiver operating characteristic curve analysis was used to determine the discriminating cutoff value for predicting AAA. Multiple logistic regression analysis was used to determine statistically significant variables associated with AAA. A *p* value \< 0.05 was considered statistically significant. Data were statistically analyzed using the JMP version 11.0 software (SAS Institute Inc., Cary, NC, USA).
Results {#Sec6}
=======
Feasibility of AAA screening during TTE {#Sec7}
---------------------------------------
Among 1912 patients, the abdominal aorta was visualized in 95.1 % (1818/1912) of the patients. The additional examination time for abdominal screening was less than a minute; the average time for abdominal aorta screening was 31.4 ± 12.3 (ranged from 13.0 to 53.0) seconds in randomly selected 15 patients. TTE was performed with Vivid7 in 36 %, with iE33 in 34 %, with Vivid Q in 17 %, and with Prosound α 7 in 13 % of the patients.
Patient characteristics {#Sec8}
-----------------------
We analyzed 1818 patients in whom adequate visualization of the abdominal aorta were obtained. Forty-one percent (746/1818) were inpatients and 49 % (897/1818) were men with a mean age of 67.4 ± 15.8 years. The clinical and echocardiographic diagnosis of the study patients were: ischemic heart disease (24 %), valvular disease (23 %), myocardial disease (22 %), arrhythmia (7 %), congenital heart disease (3 %), ventricular hypertrophy (3 %), pulmonary hypertension (2 %), cardiac mass (1 %), normal echocardiogram (12 %) and others (3 %). The mean size of the abdominal aorta was 17.2 ± 5.5 mm.
Prevalence of AAA {#Sec9}
-----------------
AAA was identified in 2.6 % (47/1818). The prevalence of AAA was 3.9 % (35/897) in men and 1.3 % (12/921) in women (*p* \< 0.001). The mean size of AAA was 43.6 ± 10.7 (30.0 to 90.0) mm, with 34 patients (72.3 %) having an aortic diameter ≥ 40 mm. Of the 47 patients with AAA, 29.7 % (14/47) underwent surgical repair or endovascular aortic stent graft insertion during 12 months following the TTE. AAA was previously known in 30 patients and unknown in 17 patients. Thirty-four of the 47 patients with AAA (72 %) had a prior or current history of smoking, 33 (70 %) had hypertension, and 26 (55 %) had ischemic heart disease.
Clinical and TTE indices associated with the presence of AAA {#Sec10}
------------------------------------------------------------
Clinical and TTE indices in patients with and without AAA are shown in Table [1](#Tab1){ref-type="table"}. The patients with AAA were significantly older than those without (77 ± 9 vs. 67 ± 16 years, *p* \< 0.001). AAA was not found before the age of 54 years. AAA was significantly more frequent in male patients than female patients. The prevalence of AAA in male patients aged ≥ 55 years was 4.6 % (35/765).Table 1Comparison of clinical and echocardiographic indicesVariablesAAAAAA*p* value(+)(-)Age (years)77 ± 967 ± 16\<0.001Men, n (%)35(75 %)862(49 %)0.001Aortic root (mm)36.0 ± 4.131.7 ± 4.2\<0.001Left atrium (mm)37.3 ± 6.938.1 ± 7.70.664Interventricular septal thickness (mm)9.7 ± 1.49.6 ± 2.00.323LV posterior wall thickness (mm)9.7 ± 1.19.3 ± 1.50.018LV end-diastolic dimension (mm)47.8 ± 4.346.2 ± 6.10.010LV end-systolic dimension (mm)31.0 ± 5.528.8 ± 7.00.001LV fractional shortening (%)34.9 ± 7.837.8 ± 8.50.001LV mass (g)168.1 ± 41.2156.0 ± 51.90.016E (cm/sec)59.6 ± 18.170.6 ± 23.80.001A (cm/sec)85.3 ± 25.279.6 ± 24.60.192E/A0.75 ± 0.370.93 ± 0.470.001E deceleration time (msec)247.1 ± 61.5232.8 ± 65.80.109E' (cm/sec)5.3 ± 1.86.6 ± 2.50.001A' (cm/sec)9.5 ± 2.39.3 ± 2.30.960S' (cm/sec)6.4 ± 1.87.0 ± 1.90.022E/E'12.2 ± 5.911.8 ± 5.60.568Data are presented as mean ± SD or n (%). LV, left ventricular. E, Early diastolic LV inflow velocity; A, Late diastolic LV inflow velocity; E', early diastolic septal mitral annular velocity; A', late diastolic septal mitral annular velocity; S', systolic septal mitral annular velocity; TTE = Transthoracic echocardiography; AAA = abdominal aortic aneurysm
There were significant differences in the aortic root size, left ventricular (LV) posterior wall thickness, LV end-diastolic dimension, LV end-systolic dimension, LV fractional shortening, LV mass, early diastolic LV inflow velocity, early diastolic LV inflow velocity/ late diastolic LV inflow velocity, early diastolic septal mitral annular velocity, and systolic septal mitral annular velocity between patients with and without AAA.
The aortic root size measured by TTE was significantly larger in patients with AAA than those without (36.0 ± 4.1 vs. 31.7 ± 4.2 mm, *p* \< 0.001) (Fig. [1](#Fig1){ref-type="fig"}). All patients with AAA had aortic root size of ≥ 28 mm. The aortic root size correlated significantly with the abdominal aortic size (r = 0.31, *p* \< 0.001) (Fig. [2](#Fig2){ref-type="fig"}). The aortic root size of ≥ 34 mm was predictive of AAA by receiver operating characteristic curve analysis (area under the curve = 0.78, sensitivity: 70 %, specificity: 70 %, *p* \< 0.001). Multiple logistic regression analysis using 12 variables revealed that aortic root size (Hazard ratio 1.23, *p* \< 0.001) and age (Hazard ratio 1.05, *p* = 0.013) were the independent predictors of AAA (Table [2](#Tab2){ref-type="table"}).Fig. 1The aortic root size in patients with and without AAA. The aortic root size was significantly larger in patients with AAA than those without (36.0 ± 4.1 vs. 31.7 ± 4.2 mm, p \< 0.001). All patients with AAA had the aortic root size of ≥ 28 mm. AAA = abdominal aortic aneurysm Fig. 2Correlation between the aortic root and the abdominal aorta sizes. The aortic root size correlated significantly with the abdominal aortic size (r = 0.31, *p* \< 0.001) Table 2Multiple logistic regression analysis for predicting AAAVariablesHazard ratio95 % CI*p* valueAge (years)1.051.01-1.090.013Men, n (%)1.270.57-2.900.562Aortic root (mm)1.231.13-1.33\<0.001LV posterior wall thickness (mm)1.550.93-2.670.10LV end-diastolic dimension (mm)1.230.98-1.530.067LV end-systolic dimension (mm)0.900.75-1.140.352LV fractional shortening (%)0.940.86-1.060.242LV mass (g)0.980.95-0.990.054E (cm/sec)1.000.97-1.020.741E/A0.680.17-1.700.489E' (cm/sec)0.920.69-1.180.558S' (cm/sec)0.860.66-1.100.235Data are presented as mean ± SD or n (%). LV, left ventricular. E, Early diastolic LV inflow velocity; A, Late diastolic LV inflow velocity; E', early diastolic septal mitral annular velocity; A', late diastolic septal mitral annular velocity; S', systolic septal mitral annular velocity; AAA = abdominal aortic aneurysm; CI, confidence interval
Discussion {#Sec11}
==========
The present study had 3 major findings. First, the feasibility of the abdominal aortic visualization during TTE was excellent (95.1 %). Second, the prevalence of AAA during TTE was 2.6 %. Third, the aortic root size measured by TTE was the independent predictor of AAA.
Reynolds T et al. firstly reported in 1990 that evaluation of the abdominal aorta should be carried out during a routine TTE examination \[[@CR18]\]. Afterwards, other authors have reported usefulness of TTE for screening AAA \[[@CR7]--[@CR16]\], because of excellent feasibility and relatively high prevalence of AAA.
The success rate of abdominal aortic imaging during TTE has been reported to range from 79 to 96 % with an additional time range from 0.5 to 8 min for examination \[[@CR7]--[@CR16]\]. In the present study, we used the same cardiac transducer, so that no additional equipment was required. Moreover, an additional examination time for abdominal visualization was less than a minute in randomly selected 15 patients. These results suggest that screening of the abdominal aorta during TTE is feasible with minimal additional time and cost compared to separate abdominal ultrasound examination.
The prevalence of AAA during TTE has been reported to range from 2.2 % to 6.5 % in Western countries \[[@CR7]--[@CR16]\], whereas the prevalence of AAA using abdominal ultrasound ranged from 3.0 % to 8.0 % from population screening surveys in Western countries \[[@CR19]\]. Therefore, previous studies have demonstrated that screening for AAA during TTE is clinically useful \[[@CR7]--[@CR16]\]. However, Seelig MH et al. demonstrated that TTE performed in a highly selected cardiac patient group in a tertiary referral center is not a useful tool for screening clinically unsuspected AAA \[[@CR20]\], because the prevalence of AAA was low (0.8 %) and only 16 % of the patients with AAA underwent surgical repair following TTE. In Korean population \[[@CR21]\], detection rate of AAA was 0.5 % during TTE indicating that the prevalence of AAA in Asia is lower than that in Western countries. Despite the low prevalence of AAA, a routine examination of the abdominal aorta during TTE appeared to be an effective preventive strategy for life-threatening but asymptomatic AAA, because 26 % of the patients with AAA underwent surgical repair or endovascular aortic stent graft insertion following the TTE. In the present study, the prevalence of AAA during TTE was 2.6 %. Moreover, 30 % of the patients of the present study with AAA underwent aortic repair following the TTE. These results suggest that screening for AAA during TTE also appeared to be clinically useful in a Japanese population.
Among the TTE indices associated with the presence of AAA, LV hypertrophy, LV dilatation, or low LV fractional shortening are likely relate to hypertension; 70 % of the patients with AAA had hypertension. Although Bekkers SC et al. reported that patients with AAA had dilated ascending aorta \[[@CR12]\], little is known about the clinical importance of the aortic root size in association with AAA \[[@CR14], [@CR22]\]. Recently, Aboyans et al. have reported that the ascending aorta is larger in patients with AAA than those without by univariate analysis \[[@CR16]\], though it is unknown about the exact position of measurement of the ascending aorta. Agricola E et al. have also demonstrated a high prevalence of dilatation/aneurysm of the ascending aorta and the aortic arch in patients with AAA evaluated by TTE \[[@CR23]\]. These studies suggest a significant correlation between dilatation of the ascending aorta and the arch and AAA. However, discriminating cutoff value of the ascending aortic size for predicting AAA was not determined. Multivariate analysis was also not used to determine the significance of the ascending aortic size in association with AAA. Thus, it still remains unknown whether the ascending aortic size is important TTE indices for screening AAA.
In contrast, we measured the aortic root size at the level of the sinus of Valsalva. The aortic root size of ≥ 34 mm was predictive of AAA by receiver operating characteristic curve analysis. The aortic root size was the independent TTE index associated with the presence of AAA. As the common risk factors for AAA and atherosclerosis may lead to the aortic root dilatation, considering the fact that aortic root size and age were independent predictors of AAA, our study indicate that screening for AAA during TTE can be useful especially in the older patients with a large (≥34 mm) aortic root size. Although screening for AAA during TTE has been reported to be useful \[[@CR7]--[@CR16]\], the screening limiting to the older patients with a large (≥34 mm) aortic root size can be clinically efficient.
The present study has several limitations. First, the incidence of AAA depends on patient selection. However, the prevalence of AAA of 2.6 % in the present study was not significantly different of that reported in Western countries (2.2 % to 6.5 %). Second, AAA was incidentally discovered before the present study at least in 18 patients (60 %) as the result of other abdominal imaging studies obtained to evaluate an unrelated condition such as spine magnetic resonance imaging, computed tomography, or abdominal ultrasound, though the reasons for diagnosis of AAA were not available in the other 12 patients. The increasing use of imaging modalities has revealed asymptomatic and previously undiagnosed AAA. Therefore, we did not exclude patients with previously known AAA. Third, the atherosclerotic risk factors in patients with AAA were only available concerning a prior or current history of smoking, hypertension, and ischemic heart disease in the present study. Moreover, the atherosclerotic risk factors in patients without AAA were not available. The lack of information on the prevalence of traditional risk factors for atherosclerosis in patients with and without AAA does not allow stratifying the patients more likely to have AAA, according to atherosclerotic risk profile. Fourth, early detection of small AAA during TTE could cause needless disease labeling and anxiety resulting in psychological harm. Further studies are required to verify the clinical usefulness, cost-effectiveness, and psychological harm of AAA screening during TTE in different healthcare settings.
Conclusions {#Sec12}
===========
Visualization of the abdominal aorta during TTE is feasible with little additional time and cost. The aortic root size was the important TTE index associated with the presence of AAA. Screening for AAA during TTE was especially useful in the older patients with a large (≥34 mm) aortic root size.
AAA
: Abdominal aortic aneurysm
TTE
: Transthoracic echocardiography
LV
: Left ventricular
**Competing interests**
The authors declare that they have no competing interests.
**Authors' contributions**
YM was contributed to analysis and interpretation of data and designed the study and drafted the manuscript. HK, KO and TS were contributed to conception and design of the study and to acquisition of data and were involved in revising the manuscript. MW, DH, and YY were contributed to acquisition and interpretation of data. NM and HT were involved in drafting the manuscript and revising it. All authors read and approved the final manuscript.
| {
"pile_set_name": "PubMed Central"
} |
Related literature {#sec1}
==================
For bond-length data, see: Allen *et al.* (1987[@bb1]). For background to styryl pyridinium quaternary ammonium compounds, see: Browning *et al.* (1922[@bb3], 1923[@bb2]); Chanawanno *et al.* (2010[@bb5]); Wainwright & Kristiansen (2003[@bb11]). For related structures, see: Chanawanno *et al.* (2008[@bb6]); Fun *et al.* (2009[@bb8]). For the stability of the temperature controller used in the data collection, see: Cosier & Glazer (1986[@bb7]).
Experimental {#sec2}
============
{#sec2.1}
### Crystal data {#sec2.1.1}
C~18~H~23~N~2~ ^+^·I^−^*M* *~r~* = 394.28Monoclinic,*a* = 7.7099 (1) Å*b* = 20.2780 (4) Å*c* = 10.9375 (2) Åβ = 92.527 (1)°*V* = 1708.32 (5) Å^3^*Z* = 4Mo *K*α radiationμ = 1.87 mm^−1^*T* = 100 K0.34 × 0.30 × 0.21 mm
### Data collection {#sec2.1.2}
Bruker APEXII CCD area detector diffractometerAbsorption correction: multi-scan (*SADABS*; Bruker, 2005[@bb4]) *T* ~min~ = 0.570, *T* ~max~ = 0.69123707 measured reflections6198 independent reflections5765 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.026
### Refinement {#sec2.1.3}
*R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.022*wR*(*F* ^2^) = 0.070*S* = 1.106198 reflections282 parametersAll H-atom parameters refinedΔρ~max~ = 0.51 e Å^−3^Δρ~min~ = −0.63 e Å^−3^
{#d5e475}
Data collection: *APEX2* (Bruker, 2005[@bb4]); cell refinement: *SAINT* (Bruker, 2005[@bb4]); data reduction: *SAINT*; program(s) used to solve structure: *SHELXTL* (Sheldrick, 2008[@bb9]); program(s) used to refine structure: *SHELXTL*; molecular graphics: *SHELXTL*; software used to prepare material for publication: *SHELXTL* and *PLATON* (Spek, 2009[@bb10]).
Supplementary Material
======================
Crystal structure: contains datablocks global, I. DOI: [10.1107/S1600536810037505/is2599sup1.cif](http://dx.doi.org/10.1107/S1600536810037505/is2599sup1.cif)
Structure factors: contains datablocks I. DOI: [10.1107/S1600536810037505/is2599Isup2.hkl](http://dx.doi.org/10.1107/S1600536810037505/is2599Isup2.hkl)
Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?is2599&file=is2599sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?is2599sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?is2599&checkcif=yes)
Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [IS2599](http://scripts.iucr.org/cgi-bin/sendsup?is2599)).
The authors thank the Prince of Songkla University for financial support. The authors also thank Universiti Sains Malaysia for the Research University Grant No. 1001/PFIZIK/811160.
Comment
=======
For a long time, styryl pyridinium quaternary ammonium compounds were known to exhibit antiseptic properties (Browning *et al.*, 1922, 1923). However medicinal researchers have long neglected to further develop the styryl pyridinium chromophore compounds for use as antibacterial agents due to the superior properties of penicillin until the incoming of the penicillin-resistant bacteria phenomenon, for example, methicillin-resistant *Staphylococcus aureus*, MRSA. The most interesting feature of styryl pyridinium quaternary ammonium compounds is their very specific activity to MRSA which is a vital drug-resistant bacteria (Wainwright & Kristiansen, 2003; Chanawanno *et al.*, 2010). From this significant reason, our research group has synthesized and characterized several styryl pyridinium derivatives including the title compound (I) in order to search for new potent antibacterial agents. Herein we report the crystal structure of (I).
Figure 1 shows the asymmetric unit of (I), which consists of a C~18~H~23~N~2~^+^ cation and an I^-^ anion. The cation exists in the *E* configuration with respect to the C6═C7 double bond \[1.350 (2) Å\] with the torsion angle C5--C6--C7--C8 = -179.29 (16)°. The pyridinium and benzene rings are nearly coplanar with the ethenyl bridge with the dihedral angle between the pyridinium and benzene rings being 4.63 (7)°. The two ethyl groups of the diethylamino substituent pointed towards the opposite directions with respect to the plane of benzene ring. The conformation of the diethylamino can be indicated by the torsion angles C11--N2--C14--C15 = 84.7 (2)° and C11--N2--C16--C17 = 79.0 (2)°. The bond lengths of cation in (I) are in normal ranges (Allen *et al.*, 1987) and comparable to those in related structures (Chanawanno *et al.*, 2008; Fun *et al.*, 2009).
In the crystal packing (Fig. 2), the cations are arranged in a zig-zag manner along the *b* axis with the iodide ions located in the interstitials of the cations and linked to the cations by a C---H···I weak interaction (Table 1). The cations stacked approximately along the *a* axis in an antiparallel manner by π--π interaction with the *Cg*1···*Cg*2^iii^ distance of 3.5262 (9) Å \[symmetry code: (iii) 1-x, 1-y, 1-z\]; *Cg*1 and *Cg*2 are centroids of N1/C1--C5 and C8--C13 rings, respectively. The crystal structure is further stabilized by C---H···π interactions (Table 1).
Experimental {#experimental}
============
The title compound (I) was prepared by mixing 1:1:1 molar ratio solutions of 1,2-dimethylpyridinium iodide (2 g, 8.5 mmol), 4-diethylaminobenzaldehyde (1.52 ml, 8.5 mmol) and piperidine (0.84 ml, 8.5 mmol) in methanol (40 ml). The resulting solution was refluxed for 6 hours under a nitrogen atmosphere. The orange solid which formed was filtered and washed with diethylether. Orange block-shaped single crystals of (I) suitable for *x*-ray structure determination were recrystallized from methanol by slow evaporation at room temperature over a few weeks (m.p. 527-529 K).
Refinement {#refinement}
==========
All H atoms were located in a difference map and refined isotropically. The highest residual electron density peak is located at 1.57 Å from I1 and the deepest hole is located at 0.48 Å from I1.
Figures
=======
{#Fap1}
{#Fap2}
Crystal data {#tablewrapcrystaldatalong}
============
------------------------- ---------------------------------------
C~18~H~23~N~2~^+^·I^−^ *F*(000) = 792
*M~r~* = 394.28 *D*~x~ = 1.533 Mg m^−3^
Monoclinic, *P*2~1~/*c* Melting point = 527--529 K
Hall symbol: -P 2ybc Mo *K*α radiation, λ = 0.71073 Å
*a* = 7.7099 (1) Å Cell parameters from 6198 reflections
*b* = 20.2780 (4) Å θ = 2.0--32.6°
*c* = 10.9375 (2) Å µ = 1.87 mm^−1^
β = 92.527 (1)° *T* = 100 K
*V* = 1708.32 (5) Å^3^ Block, orange
*Z* = 4 0.34 × 0.30 × 0.21 mm
------------------------- ---------------------------------------
Data collection {#tablewrapdatacollectionlong}
===============
------------------------------------------------------------ --------------------------------------
Bruker APEXII CCD area detector diffractometer 6198 independent reflections
Radiation source: sealed tube 5765 reflections with *I* \> 2σ(*I*)
graphite *R*~int~ = 0.026
φ and ω scans θ~max~ = 32.6°, θ~min~ = 2.0°
Absorption correction: multi-scan (*SADABS*; Bruker, 2005) *h* = −8→11
*T*~min~ = 0.570, *T*~max~ = 0.691 *k* = −29→30
23707 measured reflections *l* = −16→16
------------------------------------------------------------ --------------------------------------
Refinement {#tablewraprefinementdatalong}
==========
------------------------------------- -------------------------------------------------------------------------------------------------
Refinement on *F*^2^ Primary atom site location: structure-invariant direct methods
Least-squares matrix: full Secondary atom site location: difference Fourier map
*R*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.022 Hydrogen site location: inferred from neighbouring sites
*wR*(*F*^2^) = 0.070 All H-atom parameters refined
*S* = 1.10 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0312*P*)^2^ + 1.8072*P*\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3
6198 reflections (Δ/σ)~max~ = 0.001
282 parameters Δρ~max~ = 0.51 e Å^−3^
0 restraints Δρ~min~ = −0.63 e Å^−3^
------------------------------------- -------------------------------------------------------------------------------------------------
Special details {#specialdetails}
===============
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Experimental. The crystal was placed in the cold stream of an Oxford Cryosystems Cobra open-flow nitrogen cryostat (Cosier & Glazer, 1986) operating at 100.0 (1) K.
Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.
Refinement. Refinement of F^2^ against ALL reflections. The weighted R-factor wR and goodness of fit S are based on F^2^, conventional R-factors R are based on F, with F set to zero for negative F^2^. The threshold expression of F^2^ \> 2sigma(F^2^) is used only for calculating R-factors(gt) etc. and is not relevant to the choice of reflections for refinement. R-factors based on F^2^ are statistically about twice as large as those based on F, and R- factors based on ALL data will be even larger.
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å^2^) {#tablewrapcoords}
==================================================================================================
------ --------------- -------------- --------------- -------------------- --
*x* *y* *z* *U*~iso~\*/*U*~eq~
I1 0.385998 (14) 0.649927 (5) 0.254076 (9) 0.01744 (4)
N1 0.27082 (18) 0.63425 (7) 0.69740 (12) 0.0128 (2)
N2 0.1456 (2) 0.33502 (7) 0.09996 (13) 0.0165 (3)
C1 0.3117 (2) 0.66702 (9) 0.80338 (15) 0.0156 (3)
C2 0.3971 (2) 0.63637 (9) 0.89989 (15) 0.0169 (3)
C3 0.4439 (2) 0.57021 (9) 0.88598 (15) 0.0181 (3)
C4 0.4023 (2) 0.53711 (8) 0.77893 (15) 0.0159 (3)
C5 0.3120 (2) 0.56923 (8) 0.68124 (14) 0.0130 (3)
C6 0.2616 (2) 0.53758 (8) 0.56679 (15) 0.0149 (3)
C7 0.3027 (2) 0.47469 (8) 0.53980 (14) 0.0141 (3)
C8 0.2575 (2) 0.44086 (8) 0.42673 (14) 0.0136 (3)
C9 0.1595 (2) 0.46906 (8) 0.32844 (15) 0.0144 (3)
C10 0.1223 (2) 0.43480 (8) 0.22166 (15) 0.0150 (3)
C11 0.1815 (2) 0.36902 (8) 0.20590 (14) 0.0135 (3)
C12 0.2802 (2) 0.34044 (8) 0.30475 (15) 0.0131 (3)
C13 0.3169 (2) 0.37578 (8) 0.41093 (14) 0.0133 (3)
C14 0.0388 (2) 0.36290 (9) −0.00116 (15) 0.0173 (3)
C15 0.1390 (3) 0.40606 (10) −0.08716 (17) 0.0216 (3)
C16 0.2067 (2) 0.26717 (8) 0.08518 (15) 0.0164 (3)
C17 0.0948 (3) 0.21691 (9) 0.14881 (17) 0.0201 (3)
C18 0.1817 (2) 0.67224 (9) 0.59821 (16) 0.0182 (3)
H1A 0.285 (3) 0.7107 (13) 0.805 (2) 0.020 (6)\*
H2A 0.426 (4) 0.6591 (13) 0.971 (3) 0.029 (7)\*
H3A 0.512 (4) 0.5499 (14) 0.954 (3) 0.031 (7)\*
H4A 0.434 (3) 0.4942 (13) 0.774 (2) 0.019 (6)\*
H6A 0.198 (3) 0.5628 (12) 0.507 (2) 0.014 (5)\*
H7A 0.369 (3) 0.4489 (13) 0.598 (2) 0.020 (6)\*
H9A 0.120 (3) 0.5160 (13) 0.332 (2) 0.021 (6)\*
H10A 0.055 (4) 0.4541 (14) 0.158 (3) 0.027 (7)\*
H12A 0.325 (3) 0.2966 (13) 0.298 (2) 0.020 (6)\*
H14A −0.057 (3) 0.3869 (13) 0.030 (2) 0.017 (6)\*
H13A 0.384 (3) 0.3571 (11) 0.476 (2) 0.015 (6)\*
H14B −0.013 (3) 0.3265 (13) −0.047 (2) 0.021 (6)\*
H15A 0.061 (4) 0.4195 (15) −0.156 (3) 0.037 (8)\*
H15B 0.238 (4) 0.3824 (14) −0.117 (2) 0.026 (7)\*
H15C 0.180 (4) 0.4425 (15) −0.047 (3) 0.028 (7)\*
H16A 0.323 (3) 0.2637 (13) 0.115 (2) 0.022 (6)\*
H16B 0.202 (3) 0.2567 (13) 0.002 (2) 0.020 (6)\*
H17A 0.139 (4) 0.1730 (15) 0.135 (3) 0.029 (7)\*
H17B 0.098 (3) 0.2228 (14) 0.236 (2) 0.021 (6)\*
H17C −0.022 (4) 0.2207 (14) 0.121 (2) 0.027 (7)\*
H18A 0.176 (4) 0.7175 (14) 0.623 (2) 0.027 (7)\*
H18B 0.069 (4) 0.6552 (14) 0.586 (3) 0.032 (8)\*
H18C 0.244 (4) 0.6682 (14) 0.525 (3) 0.023 (6)\*
------ --------------- -------------- --------------- -------------------- --
Atomic displacement parameters (Å^2^) {#tablewrapadps}
=====================================
----- ------------- ------------- ------------- ------------- ------------- -------------
*U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^
I1 0.02091 (6) 0.01714 (6) 0.01425 (6) 0.00473 (4) 0.00049 (4) 0.00028 (3)
N1 0.0135 (6) 0.0121 (6) 0.0130 (5) 0.0000 (4) 0.0022 (4) 0.0001 (4)
N2 0.0239 (7) 0.0128 (6) 0.0126 (6) 0.0009 (5) −0.0015 (5) −0.0015 (5)
C1 0.0179 (7) 0.0143 (7) 0.0147 (6) −0.0003 (5) 0.0036 (5) −0.0022 (5)
C2 0.0183 (7) 0.0200 (7) 0.0126 (6) −0.0052 (6) 0.0026 (5) −0.0017 (5)
C3 0.0205 (7) 0.0187 (7) 0.0149 (6) −0.0030 (6) −0.0018 (5) 0.0035 (6)
C4 0.0196 (7) 0.0125 (7) 0.0154 (6) −0.0013 (5) −0.0016 (5) 0.0020 (5)
C5 0.0121 (6) 0.0129 (6) 0.0140 (6) −0.0011 (5) 0.0019 (5) −0.0003 (5)
C6 0.0161 (7) 0.0143 (7) 0.0142 (6) 0.0000 (5) −0.0013 (5) −0.0013 (5)
C7 0.0142 (6) 0.0140 (7) 0.0141 (6) −0.0010 (5) 0.0006 (5) −0.0007 (5)
C8 0.0144 (6) 0.0128 (6) 0.0136 (6) 0.0001 (5) 0.0013 (5) −0.0011 (5)
C9 0.0143 (6) 0.0128 (6) 0.0160 (6) 0.0015 (5) 0.0014 (5) −0.0003 (5)
C10 0.0168 (7) 0.0137 (7) 0.0146 (6) 0.0021 (5) −0.0005 (5) 0.0014 (5)
C11 0.0150 (6) 0.0128 (6) 0.0127 (6) −0.0012 (5) 0.0011 (5) −0.0010 (5)
C12 0.0134 (6) 0.0122 (6) 0.0140 (6) −0.0014 (5) 0.0016 (5) 0.0012 (5)
C13 0.0136 (6) 0.0126 (6) 0.0138 (6) 0.0009 (5) 0.0011 (5) 0.0003 (5)
C14 0.0184 (7) 0.0195 (7) 0.0137 (6) −0.0006 (6) −0.0029 (5) −0.0005 (6)
C15 0.0229 (8) 0.0255 (9) 0.0166 (7) 0.0000 (7) 0.0018 (6) 0.0029 (6)
C16 0.0208 (7) 0.0134 (7) 0.0150 (6) 0.0005 (6) 0.0019 (5) −0.0033 (5)
C17 0.0229 (8) 0.0172 (8) 0.0205 (7) −0.0029 (6) 0.0025 (6) −0.0042 (6)
C18 0.0213 (8) 0.0171 (7) 0.0161 (7) 0.0037 (6) −0.0008 (6) 0.0014 (6)
----- ------------- ------------- ------------- ------------- ------------- -------------
Geometric parameters (Å, °) {#tablewrapgeomlong}
===========================
--------------------- -------------- ----------------------- --------------
N1---C1 1.361 (2) C9---H9A 1.00 (3)
N1---C5 1.369 (2) C10---C11 1.423 (2)
N1---C18 1.475 (2) C10---H10A 0.93 (3)
N2---C11 1.366 (2) C11---C12 1.418 (2)
N2---C14 1.463 (2) C12---C13 1.383 (2)
N2---C16 1.466 (2) C12---H12A 0.96 (3)
C1---C2 1.368 (2) C13---H13A 0.94 (2)
C1---H1A 0.91 (3) C14---C15 1.521 (3)
C2---C3 1.399 (3) C14---H14A 0.96 (3)
C2---H2A 0.92 (3) C14---H14B 0.97 (3)
C3---C4 1.375 (2) C15---H15A 0.99 (3)
C3---H3A 0.98 (3) C15---H15B 0.97 (3)
C4---C5 1.409 (2) C15---H15C 0.91 (3)
C4---H4A 0.91 (3) C16---C17 1.523 (3)
C5---C6 1.445 (2) C16---H16A 0.94 (3)
C6---C7 1.350 (2) C16---H16B 0.94 (2)
C6---H6A 0.95 (2) C17---H17A 0.97 (3)
C7---C8 1.443 (2) C17---H17B 0.96 (2)
C7---H7A 0.95 (3) C17---H17C 0.94 (3)
C8---C9 1.408 (2) C18---H18A 0.96 (3)
C8---C13 1.410 (2) C18---H18B 0.94 (3)
C9---C10 1.378 (2) C18---H18C 0.95 (3)
C1---N1---C5 122.25 (14) C12---C11---C10 117.06 (14)
C1---N1---C18 117.02 (14) C13---C12---C11 120.75 (15)
C5---N1---C18 120.72 (14) C13---C12---H12A 119.0 (15)
C11---N2---C14 122.16 (15) C11---C12---H12A 120.2 (15)
C11---N2---C16 120.88 (14) C12---C13---C8 122.16 (14)
C14---N2---C16 116.90 (13) C12---C13---H13A 120.7 (15)
N1---C1---C2 121.39 (16) C8---C13---H13A 117.1 (15)
N1---C1---H1A 116.3 (16) N2---C14---C15 113.93 (15)
C2---C1---H1A 122.2 (16) N2---C14---H14A 110.1 (15)
C1---C2---C3 117.98 (15) C15---C14---H14A 110.4 (15)
C1---C2---H2A 120.7 (18) N2---C14---H14B 107.8 (16)
C3---C2---H2A 121.3 (18) C15---C14---H14B 109.0 (16)
C4---C3---C2 120.60 (15) H14A---C14---H14B 105 (2)
C4---C3---H3A 122.5 (17) C14---C15---H15A 108.9 (18)
C2---C3---H3A 116.9 (17) C14---C15---H15B 110.5 (17)
C3---C4---C5 120.59 (16) H15A---C15---H15B 110 (2)
C3---C4---H4A 118.0 (16) C14---C15---H15C 110.0 (18)
C5---C4---H4A 121.4 (16) H15A---C15---H15C 109 (2)
N1---C5---C4 117.18 (14) H15B---C15---H15C 108 (2)
N1---C5---C6 118.99 (14) N2---C16---C17 112.73 (15)
C4---C5---C6 123.83 (15) N2---C16---H16A 109.7 (16)
C7---C6---C5 123.55 (15) C17---C16---H16A 109.9 (16)
C7---C6---H6A 118.6 (15) N2---C16---H16B 108.7 (16)
C5---C6---H6A 117.8 (15) C17---C16---H16B 107.1 (16)
C6---C7---C8 125.91 (15) H16A---C16---H16B 109 (2)
C6---C7---H7A 119.8 (15) C16---C17---H17A 109.6 (17)
C8---C7---H7A 114.3 (15) C16---C17---H17B 112.3 (16)
C9---C8---C13 116.92 (14) H17A---C17---H17B 106 (2)
C9---C8---C7 124.21 (15) C16---C17---H17C 110.5 (17)
C13---C8---C7 118.86 (14) H17A---C17---H17C 111 (2)
C10---C9---C8 121.80 (15) H17B---C17---H17C 107 (2)
C10---C9---H9A 117.3 (14) N1---C18---H18A 108.7 (16)
C8---C9---H9A 120.8 (14) N1---C18---H18B 107.8 (18)
C9---C10---C11 121.30 (15) H18A---C18---H18B 110 (2)
C9---C10---H10A 120.4 (17) N1---C18---H18C 109.5 (17)
C11---C10---H10A 118.3 (17) H18A---C18---H18C 110 (2)
N2---C11---C12 121.47 (15) H18B---C18---H18C 111 (3)
N2---C11---C10 121.47 (15)
C5---N1---C1---C2 0.1 (2) C7---C8---C9---C10 179.02 (16)
C18---N1---C1---C2 −178.88 (16) C8---C9---C10---C11 0.0 (3)
N1---C1---C2---C3 1.0 (3) C14---N2---C11---C12 177.77 (15)
C1---C2---C3---C4 −1.1 (3) C16---N2---C11---C12 0.6 (2)
C2---C3---C4---C5 0.1 (3) C14---N2---C11---C10 −2.6 (3)
C1---N1---C5---C4 −1.1 (2) C16---N2---C11---C10 −179.73 (15)
C18---N1---C5---C4 177.87 (15) C9---C10---C11---N2 −179.66 (16)
C1---N1---C5---C6 178.99 (15) C9---C10---C11---C12 0.0 (2)
C18---N1---C5---C6 −2.1 (2) N2---C11---C12---C13 179.33 (15)
C3---C4---C5---N1 1.0 (2) C10---C11---C12---C13 −0.3 (2)
C3---C4---C5---C6 −179.10 (16) C11---C12---C13---C8 0.6 (2)
N1---C5---C6---C7 176.94 (16) C9---C8---C13---C12 −0.6 (2)
C4---C5---C6---C7 −3.0 (3) C7---C8---C13---C12 −179.43 (15)
C5---C6---C7---C8 −179.29 (16) C11---N2---C14---C15 84.7 (2)
C6---C7---C8---C9 −1.4 (3) C16---N2---C14---C15 −98.03 (19)
C6---C7---C8---C13 177.33 (16) C11---N2---C16---C17 79.0 (2)
C13---C8---C9---C10 0.3 (2) C14---N2---C16---C17 −98.24 (18)
--------------------- -------------- ----------------------- --------------
Hydrogen-bond geometry (Å, °) {#tablewraphbondslong}
=============================
--------------------------------------------
*Cg*2 is the centroid of the C8--C13 ring.
--------------------------------------------
---------------------- ---------- ---------- ------------- ---------------
*D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A*
C1---H1A···I1^i^ 0.91 (3) 2.99 (3) 3.7980 (18) 148.8 (19)
C18---H18B···Cg2^ii^ 0.94 (3) 2.79 (3) 3.6270 (17) 149 (3)
---------------------- ---------- ---------- ------------- ---------------
Symmetry codes: (i) *x*, −*y*+3/2, *z*+1/2; (ii) −*x*, −*y*+1, −*z*+1.
###### Hydrogen-bond geometry (Å, °)
*Cg*2 is the centroid of the C8--C13 ring.
*D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A*
------------------------ ---------- ---------- ------------- -------------
C1---H1*A*⋯I1^i^ 0.91 (3) 2.99 (3) 3.7980 (18) 148.8 (19)
C18---H18*B*⋯*Cg*2^ii^ 0.94 (3) 2.79 (3) 3.6270 (17) 149 (3)
Symmetry codes: (i) ; (ii) .
[^1]: Thomson Reuters ResearcherID: A-5085-2009.
[^2]: Additional correspondence author, email: hkfun\@usm.my. Thomson Reuters ResearcherID: A-3561-2009.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
During the living donor liver transplantation (LDLT), the hemodynamic status changes dramatically. Sudden blood loss, clamping of the great vessels and reperfusion syndrome may cause hemodynamic instability. Moreover, the clinical features of cirrhotic patients include high cardiac output (CO), low systolic vascular resistance (SVR) and tachycardia \[[@CR1]\]. Thus, perioperative management becomes extraordinarily challenging for anesthesiologists. Standard intraoperative monitoring includes CO monitoring, which allows the anesthesiologist to make prompt and accurate decisions when needed. The gold standard of CO measurement during liver transplantation is the thermodilution technique using a pulmonary artery catheter (PAC) \[[@CR2]\]. However, complications have been reported regarding the placement of the PAC, such as pneumothorax, air embolus, arrhythmia, right bundle branch block, catheter knotting, thrombosis \[[@CR3]\], right ventricular rupture \[[@CR4]\] or pulmonary artery rupture \[[@CR5]\]. Whether there is an additional benefit to decision making provided by a PAC over standard care in elderly, high-risk surgical patients requiring intensive care \[[@CR6]--[@CR9]\] is questionable. That said, the importance of intraoperative CO monitoring has led to the development of non-invasive or less-invasive cardiac output monitors, which appear promising as replacements for the thermodilution technique in the intensive care unit; however, none of them can provide adequate precision and accuracy during LDLT \[[@CR10]--[@CR12]\].
Electrical velocimetry (EV), a type of non-invasive cardiac output monitoring based on thoracic electrical impedance, during cardiac ejection only uses four standard electrocardiographic electrodes to measure CO. Two electrodes are placed at the base of the neck on the left side and the other two on the left inferior part of the thorax at the level of the xiphoid process. EV is measured based on the changes in thoracic electrical impedance as the ohmic equivalent of the mean aortic blood flow acceleration. The stroke volume (SV (mL)) and CO (L/min) can be derived by the following equations:$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ SV= VEPT\times v\times LVET $$\end{document}$$$$\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$ CO= SV\times HR/1000 $$\end{document}$$
In these equations, VEPT (mL) is the volume of electrically participating tissue calculated from the body mass and body height, ν (/s) is mean aortic blood flow velocity during left ventricular ejection, LVET (s) is the left ventricular ejection time and HR (beat/min) is the heart rate. The Aesculon™ bioimpedance electrical cardiometry monitor (Osypka Medical GmbH, Berlin, Germany) is based on the EV formula.
Studies have been conducted to assess the accuracy of CO measurement using EV compared to thermodilution, Fick equation and transthoracic doppler echocardiography \[[@CR13]--[@CR16]\]. In post-cardiac surgical patients, the accuracy and interchangeability of electrical velocimetry with the thermodilution method has been demonstrated \[[@CR17]--[@CR20]\]. Rajput RS et al. applied the EV device during cardiac surgery and showed that the percentage error ranged from 22 to 32% \[[@CR21]\].
However, limited data are available to evaluate the accuracy and precision of EV during LDLT. The aim of this study is to assess whether non-invasive EV can replace the continuous thermodilution technique (by PAC) during LDLT.
Materials and methods {#Sec2}
=====================
This study was a prospective observational study. The study was approved by the Institutional Review Board of Chang Gung Medical Foundation in Taiwan (registration number: 201600264B0). Informed consent forms were obtained from all the participants of the study. Twenty-three patients undergoing LDLT in Chang-Gung Memorial Hospital between July 2016 and March 2017 were enrolled in this study. The exclusion criteria were preoperative atrial fibrillation, significant valvular pathology, intracardiac shunt, severe pulmonary hypertension and refusal to provide consent.
General anesthesia was started with propofol, 1 to 2 mg/kg; fentanyl, 1 to 2 mg/kg; and cisatracurium, 0.2 mg/kg or rocuronium 1.2 mg/kg. For the cardiac output measurement, the PAC was placed through the right internal jugular vein and a triple lumen central venous catheter was inserted through the left internal jugular vein. The tip position of the PAC was confirmed by the waveforms of pulmonary artery pressure. The PAC was connected to a Vigilance II Monitor (Edwards Lifesciences, USA). The continuous thermodilution machine measures CO every 30 s. CO measured by EV was obtained by the Aesculon™ monitor. After the PAC was inserted, four surface electrodes were applied to the patient according to the Aesculon™ protocol. As the triple lumen catheter was fixed on the left side of neck, two electrodes were applied on the left side of the cheek and neck. The other two electrodes were applied on the lower thorax. Only signals with adequate data quality were included in the analysis.
After the sensors were attached to the skin, the data were checked to ensure proper functioning of the sensors. The clock in the operation room, on the Aesculon™ and on the continuous thermodilution machine were synchronized. CO~Ev~ and CO~PAC~ were documented at nine time points: (1) immediately after PAC placement and calibration, T1; (2) 60--90 min after skin incision, T2; (3) 120--150 min after skin incision, T3; (4) immediately after removing the liver, T4; (5) 15 min after removing the liver, T5; (6) 30 min after removing the liver, T6; (7) immediately after releasing the inferior vena cava clamp, T7; (8) 15 min after releasing the inferior vena cava clamp, T8; (9) 30 min after releasing the inferior vena cava clamp, T9. The data were classified into three phases: dissection phase (T1-T3), anhepatic phase (T4-T6), and reperfusion phase (T7-T9).
During the LDLT, the partial clamp on the inferior vena cava (piggyback technique) was used in our hospital. Piggyback technique was proved to be a safer approach to venous outflow tract reconstruction from the hemodynamic point of view \[[@CR22]\].
Statistical analysis {#Sec3}
--------------------
Statistical analysis was performed using SPSS 22.0 (SPSS Inc., Chicago, IL, US) and R version 3.0.4 (Vienna, Austria). The correlation coefficient and Bland-Altman analysis were used to assess the agreement between two methods \[[@CR23]\]. A correlation coefficient between 0.9 and 1.0 indicates a strong correlation; a correlation coefficient \< 0.5 indicates a weak correlation. Bias, limitation of agreement and percentage error were calculated. The percentage error (1.96\* standard deviation/average of CO~PAC~ and CO~Ev~) was calculated according to Critchley and Critchley \[[@CR24]\]. The clinically acceptable percentage error is less than 30%. The trending ability was assessed using a four-quadrant plot \[[@CR25]\]. The central exclusion zone of the four-quadrant plot was ±1.0 L/min for small changes in CO \[[@CR26]\]. The concordance rate was defined as the percentage of the total number of plots in the first and third quadrants of the four-quadrant plot. The concordance rate was considered good and clinically acceptable if the rate exceeded 92%. Pearson's correlation was also performed to assess the association between the two methods. Statistical significance was set at *p* \< 0.05.
Results {#Sec4}
=======
Twenty-five patients who underwent planned LDLT between July 2016 and March 2017 were initially enrolled in this study. However, two patients were subsequently excluded, leaving data from 23 patients for analysis. One patient was excluded because of a poor electrical signal, so the CO could not be calculated by the EV machine. The operation was canceled for the other patient due to high pulmonary arterial pressure. The baseline demographic data are shown in Table [1](#Tab1){ref-type="table"}.Table 1Patient characteristics of liver transplantation recipientsCharacteristicDescriptive statisticsAge (years)56 ± 7(41--68)Gender Male18 Female5Body mass index(kg/m2)24.5 ± 3.0(20.4--31.9)Ascites amount (ml)2265 ± 3744(0--12,400)Indication for LDLT HBV related HCC11 HCV related Cirrhosis4 HBV with acute liver failure3 Alcoholic related cirrhosis2 Drug1 HBV related cirrhosis2MELD SCORE18 ± 11(6--41) \< 105 10--1910 20--294 \> =304Data are described as mean ± standard deviation (range) or number*LDLT* Living donor liver transplantation, *HBV* Hepatitis B virus, *HCC* hepatocellular carcinoma, *HCV* Hepatitis C virus, *MELD* Model for End-Stage liver disease
A total of 207 paired data sets were recorded. CO data were in the range of 2.8--12.7 L/min measured by PAC and 3.4--14.9 L/min derived from the EV machine, as revealed in Table [2](#Tab2){ref-type="table"}. The highest intragroup correction coefficient was 0.646 at the time point immediately after removing the inferior vena cava partial clamp. The overall mean CO~PAC~ was 7.1 ± 2.2 L/min and the mean CO~Ev~ was 8.4 ± 2.2 L/min. The average CO~Ev~ was slightly higher than the average CO~PAC~. The correction coefficient between CO~PAC~ and CO~Ev~ was 0.415 with *p* \< 0.01. The correction coefficients were 0.553 in the dissection phase, 0.276 in the anhepatic phase and 0.376 in the reperfusion phase (Table [3](#Tab3){ref-type="table"}).Table 2Correction coefficient, bias and 95% limitation of agreement of all measurements and time pointTimeIntragroup correction coefficientBias (L/MIN)Limits of agreement(L/MIN)CO~PAC~ (L/MIN)CO~Ev~ (L/MIN)SVR by PAC dyne\*cm^−5^T1\*\*\*0.592−0.78− 4.38 to 2.826.70 ± 2.027.49 ± 2.05908 ± 460T2\*\*\*0.524−0.65− 4.92 to 3.626.88 ± 2.187.54 ± 2.28897 ± 553T3\*\*\*0.555−1.03−4.87 to 2.817.00 ± 2.248.03 ± 1.85862 ± 410T4\*\*\*0.466−1.20− 5.75 to 3.356.99 ± 2.348.19 ± 2.13794 ± 662T50.334−1.59−6.67 to 3.496.83 ± 2.148.41 ± 2.37829 ± 344T60.012−2.29−7.74 to 3.166.61 ± 2.148.90 ± 1.80896 ± 750T7\*\*\*0.646−2.43−6.25 to 1.396.81 ± 2.299.23 ± 2.33863 ± 364T80.276−0.62−5.27 to 4.038.06 ± 1.968.68 ± 1.99616 ± 230T90.304−0.72−6.35 to 4.918.40 ± 1.999.13 ± 2.77665 ± 233Abbreviations: *CO*~*PAC*~ cardiac output derived from continuous transpulmonary thermodilution method, *CO*~*Ev*~ cardiac output derived from Electrical velocimetry methodData are shown as mean ± standard deviation\*\*\* the *P*-value \< 0.05Table 3Correction coefficient, bias and percentage error between CO~PAC~ and CO~Ev~ measurement in each surgical phasePhasecorrection coefficientBias (L/MIN)Limits of agreement(L/MIN)Percentage error (%)Total\*\*\*0.415−1.26 ± 2.39− 5.94 to 3.4260.0%Dissection\*\*\*0.553−0.82 ± 1.97− 4.68 to 3.0453.1%Anhepatic\*\*\*0.276−1.69 ± 2.57−6.73 to 3.3465.7%Reperfusion\*\*\*0.376−1.26 ± 2.53− 6.22 to 3.7059.1%Data are shown as mean ± standard deviation\*\*\* the *P*-value \< 0.05
The Bland-Altman analysis with a mean bias between CO~PAC~ and CO~Ev~ was − 1.26 L/min and the 95% limitation agreement was − 5.9 to 3.4 L/min (Fig. [1](#Fig1){ref-type="fig"}). The overall percentage error was 60%, which failed to meet the criterion of interchangeability (\< 30%). Figure [2](#Fig2){ref-type="fig"} shows the Bland-Altman plot in three phases. The percentage errors were 53% in the dissection phase, 65.7% in the anhepatic phase and 59.1% in the reperfusion phase.Fig. 1Bland-Altman plot for CO~Ev~ and CO~PAC~. Bias and limits of agreement (±1.96SD) are shown in the plot. Abbreviations: CO~PAC~ cardiac output derived from continuous transpulmonary thermodilution method; CO~Ev~ cardiac output derived from Electrical velocimetry method; SD standard deviationFig. 2Four-quadrant plot for comparing changes in CO~Ev~ and CO~PAC~. Data points within the ±1.0 L/min exclusion zone (box in central) are considered statistical noise and excluded. The dot line represents the line x = y. The N means the number of plots out of the exclusion zone. Abbreviations: CO~PAC~ cardiac output derived from continuous transpulmonary thermodilution method; CO~Ev~ cardiac output derived from Electrical velocimetry method
The SVR data calculated from the Vigilance II Monitor were also collected. The overall mean of SVR was 815 ± 383 dyne\*cm^− 5^ (mean ± SD). The mean of SVR dropped from 863 dyne\*cm^− 5^ to 615 dyne\*cm^− 5^ at T8 or 15 min after liver reperfusion. There was a negative correction between the bias of CO~PAC~ from CO~Ev~ and SVR (*r* = − 0.317 *p* \< 0.01). The highest intragroup correction coefficient was − 0.67 at T6, or 30 min after removing the liver.
Four-quadrant plots were drawn to evaluate the trending ability, as shown in Fig. [2](#Fig2){ref-type="fig"}. The central exclusion zone was set at ±1 L/min. The CO~PAC~ changes and CO~Ev~ changes were compared. In total, 184 paired data sets were placed on the four-quadrant plots. There were 60 sets within the central exclusion zone. After excluding the central zone data, the concordance rate between the two methods was 56.5%. The concordance rates were 48.3% in the dissection phase, 55.8% in the anhepatic phase and 61.5% in the reperfusion phase.
Discussion {#Sec5}
==========
In our study, the continuous thermodilution method was utilized to measure CO instead of the gold standard using the intermittent thermodilution technique, in which the accuracy can be affected by the timing of the injection within the respiratory cycle, the change of pulmonary artery blood temperature, the injectate, the speed of the injection and the placement of the catheter \[[@CR13]\]. Under hemodynamically stable condition, a good correction, accuracy and precision between continuous and intermittent cardiac output measurement during has been shown in the literature \[[@CR27]--[@CR29]\]. The limitation agreement between the continuous thermodilution method and the intermittent thermodilution technique is clinically acceptable. Bottiger BW et al. \[[@CR30]\] has demonstrated significant correlation between intermittent and continuous CO measurements (*r* = 0.87, *p* \< 0.0001), accompanied with a bias of 0.240 L/min during orthotopic liver transplantation. They also revealed that the changes in the pulmonary artery blood temperature would influence the CO measurements more by intermittent thermodilution than by continuous thermodilution during reperfusion. The continuous CO monitor was therefore used to determine the reliability of EV-based CO measurements. We found that CO~Ev~ showed limited accuracy when compared to continuous thermodilution CO assessment.
This study demonstrates that the Aesculon™ system using the EV formula is not interchangeable with the established automatic thermodilution method using a PAC in patients undergoing LDLT. The percentage error was 60% and was not clinically acceptable. In our study, CO~Ev~ was generally higher than CO~PAC~. The mean bias between CO~PAC~ and CO~Ev~ was − 1.26 L/min. Several factors could have contributed to such poor interchangeability between the Aesculon™ system and thermodilution using a PAC during LDLT. First, in the equivalent of CO~Ev~, the VEPT (mL) is related to the patient's body weight \[[@CR31]\]. VEPT could be miscalculated in the presence of a large volume of ascites. As ascites is one of the clinical manifestations in patients with liver cirrhosis requiring LDLT, the presence of ascites may cause an overestimation of the patient's body weight, leading to an overestimation of CO by the Aesculon™ monitor. In our study, the amount of ascites varied greatly among the patients, ranging from 0 ml to 12,000 ml. In the extreme situation, the ascites could make up to 20% of the body weight overestimation, which in turn may give a greater value CO~Ev~ than CO~PAC~. Secondly, the massive ascites could displace the upward the diaphragm upwards, causing geometric changes that could alter the conductivity of the thorax and thus affect the accuracy of the EV method. Thirdly, surgical manipulation of the upper abdomen could have an impact on the CO~Ev~. Surgical interventions to the upper abdomen could cause a shift in the bioimpedance cardiac output index readings by \> 1 L/min/m^2^, and the direction of the shift was unpredictable \[[@CR32]\]. The use of surgical retractor in addition to abdominal wall compression may additionally contribute to a change in conductivity. A poor correlation between those two methods was noted in this study. The correction coefficient between CO~PAC~ and CO~Ev~ was 0.415 with *p* \< 0.01. The subgroup analysis showed that correlation coefficients were 0.553 in the dissection phase, 0.276 in the anhepatic phase and 0.376 in the reperfusion phase. The relatively stable hemodynamics during dissection may be the reason why this phase was shown to have a better correlation coefficient. Manipulation of the great vessels during the anhepatic phase could cause a decrease in the preload, and renal vein congestion may cause hemodynamic instability \[[@CR33]\]. Reperfusion syndrome may also lead to low blood pressure during the reperfusion phase \[[@CR34], [@CR35]\]. The reason why hemodynamic stability affected the correction coefficient was due to the measurement interval of those two techniques. The CO~Ev~ measurement was based on a change in conductivity and measured CO in seconds. However, the continuous CO~PAC~ measurement needs several minutes \[[@CR36]\]. Even though the readings on the CO~PAC~ monitor refreshed CO every 30 s, the CO displayed was the average of CO over the preceding 3--6 min \[[@CR30]\]. When the CO changes, the CO~PAC~ machine needed more time to determine the real CO. With hemodynamic disturbances, CO~Ev~ could be assumed to be closer to real-time CO than CO~PAC~, which could be minutes behind. The best intragroup correction coefficient was achieved immediately after the inferior vena cava partial clamp was released, possibly as a result of the relatively hemodynamic stability and minimal interruption of the electrical signal during vessel anastomosis. It takes minutes to impact the hemodynamic after off-clamp. Due to the relatively slow response of the continuous thermodilution measurement, the correction coefficients were much worse during hemodynamic instability.
The concordance rate was 56.5% when the central exclusion zone was ±1 L/min, much lower than the clinically acceptable concordance rate of 92%. This may also be attributed by the lag in CO measurement using the continuous thermodilution technique.
This study has several limitations. First, the EV used four electrodes to detect the signal and calculate the CO. However, electro-coagulation and cutting could interfere with these signals during surgery. With every interruption, the EV machine required approximately at least 30 s to reacquire the conduction signal. The artefacts overlaying the recorded signal may in fact cause an over- or underestimation of CO, in particularly during the dissection phase. In extreme situations, the EV machine could only acquire one data point per hour due to interference from electro-coagulation. EV interference could also be due to mechanical compression of the thoracic electrodes by the surgeon. Second, the small sample size warrants further studies with a larger population size. Third, the trending ability of the EV monitor was only partially surveyed. During LDLT, fluid challenge, inotropic agent usage and great vessel clamping could cause hemodynamic changes.
Conclusions {#Sec6}
===========
In conclusion, the Aesculon™ monitor exhibited limited accuracy, precision and trending ability when compared to continuous thermodilution CO monitoring during LDLT. The Aesculon™ system is not yet interchangeable with continuous thermodilution cardiac output monitoring during LDLT.
CO
: Cardiac output
CO~Ev~
: Cardiac output measured by electrical velocimetry
CO~PAC~
: Cardiac output measured by the pulmonary thermodilution
PAC
: Pulmonary artery catheter
LDLT
: Living donor liver transplantation
EV
: Electrical velocimetry
RBCs
: Red blood cells
SV
: Stroke volume
VEPT
: Volume of electrically participating tissue
ν
: Mean aortic blood flow velocity during left ventricular ejection
LVET
: Left ventricular ejection time
HR
: Heart rate
SD
: Standard deviation
This work was supported by the Chang Gung Medical Research Project (CORPG3G0591, CORPG3G0601, CORPG3G0611, CORPG3G0621).
Funding {#FPar1}
=======
This study was funded by Chang Gung Medical Research Project (CORPG3G0591, CORPG3G0601, CORPG3G0611, CORPG3G0621).
Availability of data and materials {#FPar2}
==================================
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.
DJW: author is responsible for search and review of the relevant literatures, collect data, analyzed the data, write the article. ISL: author is responsible for search and review of the relevant literatures, collect data, analyzed the data, write the article. AHC: Author is responsible for assistant in literature search, help to collect data. CYC: Author is responsible for assistant in literature search, help to collect data. PCT: Author is responsible for assistant in literature search, help to collect data. YHT: Author is responsible for assistant in literature search, help to collect data. JRL: Review of the statistical analysis. HIT: Correspondance is responsible for research design, drafting of the work or revising it critically for important intellectual content; and final approval of the version to be published. All authors read and approved the final manuscript.
Ethics approval and consent to participate {#FPar3}
==========================================
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
The study was approved by the Institutional Review Board of Chang Gung Medical Foundation in Taiwan (registration number: 201600264B0).
Informed consent: "Informed consent was obtained from all individual participants included in the study.
Consent for publication {#FPar4}
=======================
Not applicable.
Competing interests {#FPar5}
===================
The authors declare that they have no competing interests.
Publisher's Note {#FPar6}
================
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
| {
"pile_set_name": "PubMed Central"
} |
La genética clínica es una nueva especialidad biomédica en España. El 6 de agosto de 2014 se publicó el Real Decreto de Troncalidad en el BOE[@bib0005], y también con él se creó la especialidad de genética clínica para médicos, biólogos, farmacéuticos y químicos. Pero, de momento está en desarrollo. Somos de los últimos países en Europa que tienen dicha especialidad, por lo que estamos en el pelotón de cola. Como consecuencia de lo anterior en el currículum formativo del médico de familia español NO se encuentran sus competencias en genética clínica. Esas competencias ya se recogen en varios países de nuestro entorno[@bib0010].
Desde el Grupo de Trabajo SEMFYC sobre «Genética Clínica y Enfermedades Raras» estamos trabajando para que las cosas cambien. Sabemos que en el currículum formativo del médico de familia español NO se contemplan las competencias en genética clínica. Por eso, en noviembre de 2015 escribimos a la Comisión Nacional de Medicina Familiar y Comunitaria del Consejo Nacional de Especialidades en Ciencias de la Salud, para informarles de las competencias y habilidades en esta materia que se deberían acometer dentro del periodo formativo del MIR de medicina de familia.
Las habilidades que todo médico de familia español debe conocer para desarrollar correctamente su rol dentro de su área de conocimientos de la genética clínica son: *1)* Saber construir un árbol genealógico detallado, y de 3 generaciones según la nomenclatura estandarizada. Genetistas y médicos de familia comparten la visión de que los pacientes se perciben en familias y no como entes aislados; *2)* Conocer los diversos patrones de herencia: autosómico dominante, autosómico recesivo, ligado al cromosoma X, mitocondrial,..., y saberlos identificar en el árbol genealógico de las familias atendidas; *3)* Conocer los criterios de derivación, para derivar todo tipo de enfermedades genético-hereditarias, incluidas las agregaciones familiares de cáncer, desde la atención primaria a los centros clínicos de referencia; *4)* Conocer las enfermedades genéticas más frecuentes en atención primaria, y también las 10-15 enfermedades raras de los pacientes del propio cupo de médico de familia, con el uso del Protocolo DICE-APER ([http://dice-aper.semfyc.es](http://dice-aper.semfyc.es/){#intr0005}); *5)* Saber comunicar al paciente los posibles problemas genético-hereditarios a los que se enfrenta, con un lenguaje accesible al entendimiento de la población general. Hay que ser empático poniéndose en la situación de cada paciente/familia. Al médico de familia ha de quedarle bien claro que todo lo hereditario es genético, pero no todo lo genético es hereditario. Este concepto es fundamental, y *6)* Conocer los límites éticos y legales que implica conocer el estatus genético de una determinada persona de una familia. Ha de quedar bien claro que el médico de familia NO debe ponerse en contacto con los familiares del sujeto afecto. Debe ser el sujeto afecto el que se ponga en contacto con sus familiares y les informe. Esto es así por la ley de autonomía del paciente y la ley de investigación biomédica. El médico de familia debe respetar la confidencialidad de la información genética. En definitiva, el médico de familia ha de tener unos conocimientos y habilidades básicos en genética clínica.
Consideramos que la genética clínica y el cáncer hereditario son una apuesta de futuro en la formación de los médicos de familia. Dada la actual configuración del programa docente de la especialidad, *un mes de rotación* en total entre los servicios de genética clínica y las unidades de cáncer hereditario aportaría al médico de familia la inquietud necesaria para la formación en ella, sin hipotecar otros contenidos formativos, también importantes.
Las competencias que todo médico de familia español debe llevar a cabo dentro de su área de conocimientos de la genética clínica son[@bib0015]: *1)* Identificación de individuos a riesgo para una enfermedad genética; *2)* Asesoramiento preconcepcional desde atención primaria; *3)* Conocimiento de las técnicas de diagnóstico prenatal*; 4)* Teratología: conocer los recursos telefónicos y bibliográficos; *5)* Seguimiento clínico de los pacientes con una enfermedad genética (p. ej., síndrome de Down); *6)* Identificación de problemas psicosociales en las enfermedades raras; *7)* Conocer los distintos tipos de test genéticos; *8)* Conocer los recursos de genética clínica en Internet; *9)* Conocer los servicios de genética clínica y las unidades de cáncer hereditario de referencia, y saber cómo derivar a sus pacientes, y *10)* Conocer las propias limitaciones. No se trata de ser el «genetista de atención primaria».
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#ihz061s5}
============
Acute coronary syndrome (ACS), colloquially known as the 'heart attack', is the leading cause of death globally.^[@ihz061C1]^ Although ACS has historically been considered to be rare in sub-Saharan Africa (SSA), the burden of disease is expected to grow as the region proceeds through the epidemiologic transition.^[@ihz061C2]^ As cardiovascular risk factors like hypertension become increasingly prevalent across SSA,^[@ihz061C3],[@ihz061C4]^ cardiovascular-related death will likely continue to rise. In Tanzania, for example, the Global Burden of Disease study estimated that ACS was already the fourth leading cause of mortality in 2016, with a 47% increase over the prior decade.^[@ihz061C5]^
Despite such projections, there has been little study of ACS in SSA and diagnosed cases of ACS remain rare.^[@ihz061C6]--[@ihz061C8]^ The perplexing scarcity of reported ACS cases in SSA has been the subject of much speculation,^[@ihz061C9]--[@ihz061C12]^ with some suggesting that physician practices may be contributing to rampant underdiagnosis.^[@ihz061C9]^ In Tanzania, there is evidence that physicians may be contributing to ACS underdetection. In an observational study of patients presenting to the emergency department with chest pain or shortness of breath, providers rarely pursued diagnostic workups for ACS even for patients with multiple risk factors (J. Hertz et al., manuscript submitted). This led to few patients being diagnosed with ACS and even fewer receiving appropriate treatment for ACS.
As ACS is a life-threatening emergency with a high mortality rate when not treated appropriately,^[@ihz061C13]^ understanding barriers to accurate diagnosis and proper care of ACS in SSA is an urgent public health priority. In a qualitative study at a referral hospital in Kenya where cardiac catheterization is available, barriers to ACS care included insufficient diagnostic equipment and lack of medications.^[@ihz061C14]^ Beyond this study, however, barriers to ACS care in SSA remain largely unexplored, particularly in settings where cardiac catheterization is unavailable. Understanding physician perspectives on barriers to care is an essential first step to developing effective interventions to improve ACS outcomes across SSA. To address this knowledge gap, we conducted a qualitative study among healthcare providers in northern Tanzania.
Materials and methods {#ihz061s6}
=====================
Setting {#ihz061s6a}
-------
This study was conducted in the Kilimanjaro region of northern Tanzania in 2018. The local prevalence of hypertension and diabetes among adults in Kilimanjaro is estimated to be 22 and 6%, respectively.^[@ihz061C15],[@ihz061C16]^ Despite the prevalence of these risk factors, a retrospective review of medical records at a local referral hospital in 2018 found that only 0.3% of adult admissions were for ACS (J. Hertz et al., manuscript submitted). At the time of the study, cardiac catheterization was not available at any facility in Kilimanjaro.
Sampling {#ihz061s6b}
--------
The study included 11 providers, who were eligible if they were medical doctors (MDs) or clinical officers (COs) working in either an outpatient department or an emergency department in Kilimanjaro. Based on existing data about where adults in northern Tanzania would present for acute chest pain and shortness of breath,^[@ihz061C17]^ we aimed to recruit providers who were most likely to encounter patients with ACS. Thus, providers were recruited from health centers, community hospitals and a referral hospital across the Kilimanjaro region. A purposive sampling technique was used to include a broad range of ages, medical training and work environments. Participants were recruited until thematic saturation was reached.
Procedures {#ihz061s6c}
----------
Participants were approached for recruitment via in-person conversation. Individual in-depth interviews were conducted in a private location of the participant's choosing and lasted approximately 1 h. Interviews followed a semistructured guide exploring barriers to ACS diagnosis and care. The guide was developed by an interdisciplinary committee consisting of emergency physicians, cardiologists, internists and COs from Tanzania and the USA. The interview guide was independently forward- and back-translated from English to Swahili to ensure content fidelity and was piloted prior to use. All interviews were conducted by an emergency physician from the USA (JTH) who is fluent in both English and Swahili. Participants were given the option of speaking in English or Swahili, and all participants chose to speak in a mix of both languages.
All interviews were audiorecorded in their entirety, and the recordings were transcribed and translated into English by the researcher who conducted the interviews. To ensure accurate translation, a native Swahili speaker who is fluent in English subsequently reviewed all audiorecordings and transcripts. Any disagreements regarding optimal translation were resolved by consensus.
Data analysis {#ihz061s6d}
-------------
Thematic analysis was conducted in an inductive manner, applying the principles of practice theory.^[@ihz061C18]^ After reviewing several transcripts, a codebook of themes was developed by an interdisciplinary committee of physicians and COs from the USA and Tanzania. The codebook was iteratively refined throughout the interview and coding process. The final codebook consisted of 25 subthemes grouped into nine dominant themes across three domains: provider-related, system-related and patient-related barriers. All transcripts were independently coded in NVivo 12 (QSR International, Melbourne, Australia) by two researchers: a Tanzanian CO (GLK) and an emergency physician from the USA (JTH). The two coders met regularly to refine the codebook and resolve coding discrepancies by consensus. Representative quotes were reviewed throughout the analysis procedure to capture the breadth and depth of provider perspectives on each theme.
Ethics {#ihz061s6e}
------
This study had ethical approval from the Duke Health Institutional Review Board, the Kilimanjaro Christian Medical Centre Research Ethics Committee and the Tanzania National Institutes for Medical Research Ethics Coordinating Committee. All participants provided verbal informed consent prior to enrollment.
Results {#ihz061s7}
=======
The characteristics of participants are summarized in Table [1](#ihz061TB1){ref-type="table"}. Participants represented a diverse range of medical training, years of practice and practice settings. The data revealed a range of barriers to ACS care across the levels of the provider, system and patient. Table [2](#ihz061TB2){ref-type="table"} summarizes the barriers across the three domains.
######
Characteristics of in-depth interview participants, Kilimanjaro region, 2018 (n=11)
Characteristic n
--------------------- ---
Gender
Female 4
Male 7
Medical degree
Medical doctor 6
Clinical officer 5
Years of practice
1--5 y 3
6--10 y 5
\>10 y 3
Practice setting
Health center 3
Community hospital 3
Referral hospital 5
######
Dominant themes regarding barriers to diagnosis and care of acute coronary syndrome among providers (n=11)
Domain Barrier n
--------------------------- ----------------------------------------------------------------------------------------------------------------------------- ----
Provider-related barriers Inadequate provider training- Inadequate ACS knowledge- Inability to interpret ECGs- Insufficient experience treating ACS 11
Poor application of knowledge- Failure to consider diagnosis of ACS- Misdiagnosis of ACS cases 11
System-related barriers Ill-equipped facilities- Lack of diagnostic equipment- Lack of necessary treatment- Lack of specialists 11
Cost of care 11
Lack of data and guidelines- Lack of disease burden data- Lack of locally relevant guidelines 11
Referral system challenges- Referral system delays- Transportation difficulties 10
Patient-related barriers Inadequate patient knowledge- Lack of ACS knowledge- Community misperceptions 9
Patient healthcare-seeking behavior- Delayed healthcare-seeking- Use of traditional and faith healers 8
Non-adherence- Medication non-adherence- Follow-up non-adherence- Testing non-adherence- Non-adherence to lifestyle changes 7
ACS: acute coronary syndrome; ECG: electrocardiogram.
Provider-related barriers {#ihz061s7a}
-------------------------
### Inadequate provider training {#ihz061s7a_1}
When participants were asked to identify barriers to ACS care, inadequate provider training was typically the first response offered. The providers felt their medical education emphasized communicable diseases over non-communicable diseases, leading to limited knowledge of ACS. Respondents felt that most Tanzanian providers had insufficient knowledge of ACS presentation, diagnosis and management, and nearly all viewed themselves as lacking education in these areas. Participants expressed a desire for more training in ACS care and suggested that continuing medical education be used to increase competence."*For my education, it wasn't enough to say that I can treat ACS completely. I mean, it wasn't really sufficient, it wasn't enough. I need more education, I need more knowledge. So I feel uncomfortable, I can't say that I can manage the disease from A to Z.* (Participant 6, CO, health center.)"
Participants emphasized widespread inability to interpret electrocardiograms (ECGs) as a significant barrier to ACS diagnosis. Both medical doctors and clinical officers did not feel comfortable interpreting ECGs, and all participants cited a need for improved ECG training."*ECGs are difficult for me. Yeah, I need support. I didn't get any training to read ECGs. Even to put the ECG leads, those electrodes or whatever, on the body, I wasn't taught.* (Participant 4, MD, community hospital.)"
Respondents also cited a general lack of experience managing ACS, which created an additional barrier to care. Because participants had little experience caring for ACS patients, they were not comfortable managing such cases. Many respondents reported only caring for a handful of ACS patients in their entire careers and four participants reported that they had never diagnosed or treated such a patient. This provider emphasized the way in which such lack of experience limits one's ability to diagnose ACS: "*But if you've never seen a case before, it's hard to suspect the diagnosis. I mean, you see what your brain knows. If your brain doesn't know the diagnosis, you can't see it.* (Participant 9, MD, referral hospital.)"
### Poor application of provider knowledge {#ihz061s7a_2}
Beyond insufficient training, respondents felt that providers often failed to properly apply their medical knowledge to the clinical setting. Participants worried that even when doctors had textbook-based knowledge of ACS, they did not routinely consider the diagnosis of ACS when caring for patients. This failure to consider ACS was identified by all but one participant as an important driver of ACS underdetection in Tanzania. This provider spoke about the pervasiveness of this failure to consider ACS: "*The patients with those symptoms like chest pain, some doctors interpret it as pneumonia, other doctors think that the chest pain is due to fatigue or whatever. But only a few doctors think to examine the heart.* (Participant 10, MD, community hospital.)"
Participants felt that this failure to consider the diagnosis resulted in frequent misdiagnosis of ACS. Participants believed that patients with ACS in Tanzania are often given inaccurate diagnoses, such as peptic ulcer disease or pneumonia. They worried that such misdiagnoses resulted in poor patient outcomes and reinforced misperceptions that ACS is rare."*For many others, they are misdiagnosed. For those symptoms, the doctor says 'This is something else.' So he won't do any testing, he'll just say it's something else and he'll give him other medicines.* (Participant 7, CO, community hospital.)"
Systems-related barriers {#ihz061s7b}
------------------------
### Ill-equipped facilities {#ihz061s7b_1}
Providers emphasized a lack of diagnostic equipment as a major barrier to ACS diagnosis in Tanzania. They reported that many facilities, even some referral hospitals, do not have ECG machines or troponin assays, which substantially limits clinicians' ability to diagnose ACS. Even in facilities where testing is available, participants cited frequent equipment failures and long delays in result times as common obstacles to prompt ACS diagnosis."*You might suspect ACS is the diagnosis but you don't have an ECG machine to at least confirm or you don't even have that troponin test. So you'll just say, this is just acid reflux and you'll treat the patient for that.* (Participant 8, MD, referral hospital.)"
Participants also cited the unavailability of treatments necessary for ACS management as an important barrier to care. They reported that many facilities did not have access to some mainstays of ACS treatment, such as clopidogrel, heparin and oxygen. Many respondents also noted the lack of a local center capable of performing cardiac catheterizations, the gold standard intervention for most ACS cases. They reported that the nearest cardiac catheterization center was a 10-h drive from Kilimanjaro, making emergent catheterization impossible and creating substantial financial and logistical obstacles for patients seeking this treatment."*It's difficult when we discover that someone has ACS, because you know the drug that the patient should be given, but unfortunately we usually don't have the medicine that they need.* (Participant 1, CO, health center.)"
A dearth of specialists was also frequently cited as a barrier to ACS care. None of the respondents reported having a cardiologist at their facility, including those working at a referral hospital. Respondents expressed a desire for a wider distribution of cardiologists to facilitate ACS care and offer expertise to providers across the country."*I'd also like to say that we need more cardiologists in Tanzania, especially in our hospital. We don't have a cardiologist. If we had even just one, he could help us a lot to diagnose heart diseases.* (Participant 11, MD, referral hospital.)"
### Cost of care {#ihz061s7b_2}
Providers reported that the cost of ACS care was sometimes prohibitive for some patients. Respondents felt that ECGs, cardiac biomarker testing, ACS medications and cardiac catheterization are too expensive for many Tanzanians. Participants emphasized the importance of health insurance and some shared stories of uninsured patients declining ACS testing or treatment because of the cost."*It's difficult. For the patient to open a hospital chart? Cash. To do testing? Cash. To get treatment? Cash. Because we don't do any testing until you pay for it.* (Participant 4, MD, community hospital.)"
### Lack of local data and evidence-based guidelines {#ihz061s7b_3}
Another challenge mentioned by all but one participant was lack of data regarding the local burden of ACS. Respondents were uncertain whether ACS was rare in Tanzania or whether it was common but misdiagnosed. They felt that this lack of local prevalence data caused providers to fail to consider the diagnosis in clinical care. Participants described a cycle of inadequate data leading to underdiagnosis, which in turn resulted in a persistent lack of data."*So it's possible that the disease is very common but it could also be rare. The challenge is there has been no research so we can know exactly the magnitude of the problem.* (Participant 8, MD, referral hospital.)"
Participants also wished for more locally relevant ACS guidelines to assist them with clinical care. Although the Tanzanian Ministry of Health does publish standard treatment guidelines for ACS,^[@ihz061C19]^ the majority of respondents were unaware of them. Providers also desired hospital-specific protocols based on locally acquired evidence to reduce ACS underdiagnosis and aid clinicians who lack experience managing the disease."*When I did orientation I wasn't given any guidelines. I mean, it's different than common diseases that we think are common like diarrhea or malaria that they give us guidelines for. So there are guidelines for those diseases but heart attacks? No.* (Participant 2, CO, health center.)"
### Referral system challenges {#ihz061s7b_4}
Because the resources needed to manage ACS are often found only in referral centers, providers felt the referral system itself could be a barrier to efficient diagnosis and care. Participants noted that patients with ACS typically had to visit multiple facilities over several days before reaching a facility that could care for them. They cited inefficiencies and delays in the referral system as obstacles to prompt, emergency ACS care. These providers described the series of delays that patients face in attempting to access appropriate care for ACS: "*The challenge the patient would face would be that delay---let's say the disjointed referral of his care. Because, first, it will be difficult for the doctors there in his village to know what is going on with him. \[\...\] And then maybe he'll be sent to another hospital that's a higher level of care. Because it's not like he'll be sent directly here, no. Maybe he'll be sent to a district hospital. At the district hospital, they'll say maybe, take him to \[the regional hospital\]. And all that time the patient has that same pain there in his chest and sometimes they die on the way. (*Participant 9, MD, referral hospital.)""*The normal amount of time it takes to confirm a diagnosis is very long. You have to send him to another hospital for the testing, then he has to bring back the results. You do another test there. And all that time the thrombus it continues to grow, and eventually the patient dies.* (Participant 3, CO, referral hospital.)"
Patient-related barriers {#ihz061s7c}
------------------------
### Inadequate patient knowledge {#ihz061s7c_1}
Providers felt that there was a pervasive lack of ACS knowledge among patients, which had negative effects on ACS diagnosis and outcomes. Participants believed that many patients do not know ACS symptoms or causes and that some patients were more likely to attribute symptoms like chest pain to witchcraft than heart disease. Providers reported that this lack of ACS awareness led to failure to present to hospitals and non-adherence to treatment, resulting in underdetection of ACS and poor outcomes. All but one participant stated that community educational programming is essential to improving ACS care in Tanzania."*Only a few people know about heart diseases, so many people don't really know what are the causes, what are the risks. So you can be treating the patient, but she really doesn't know why she got there, why she got that problem. And, most of them, because of that, they end up getting worse in their disease.* (Participant 6, CO, health center.)"
### Inappropriate healthcare-seeking {#ihz061s7c_2}
Participants felt that this inadequate patient education led to care-seeking behavior that negatively affected ACS detection and outcomes. Providers reported that many patients either do not seek care at all for ACS symptoms, seek care on a very delayed basis or seek care in inappropriate places, such as traditional healers. Participants believed that educational interventions were needed to encourage patients to report promptly to a hospital for possible ACS symptoms in order to improve clinical outcomes and to allow for higher rates of disease detection."*Some will look for treatment with local herbs, others will go to faith healings. So each person looks for treatment that they think will help. And other people come to the hospital. So you can see that there are many people with heart attacks but we only end up seeing a few.* (Participant 11, MD, referral hospital.)"
Many respondents felt that the popularity of traditional healers, faith healers and herbalists was a particularly vexing barrier to care. They reported that many patients seek the advice of these kinds of healers before coming to a hospital. Several participants shared stories of patients stopping their medications for diabetes or hypertension and deteriorating based on the advice of a healer. Providers were pessimistic that the popularity of such healers could be easily diminished."*They go to the preachers and they tell them, 'Stop taking those medicines, throw them away, Jesus will heal you, Allah will heal you. You don't have to take these medicines.' And so they stop taking them.* (Participant 6, CO, health center.)"
### Non-adherence {#ihz061s7c_3}
Closely related to low levels of patient education is the issue of non-adherence. Participants referenced patient non-adherence to medications, lifestyle changes, follow-up appointment attendance and recommended diagnostic testing. Providers felt that such non-adherence was often due to lack of awareness of the causes and risks of ACS, although they also cited cost as a barrier to adherence. Participants worried that such non-adherence resulted in worse patient outcomes."*They normally only decide to come back when the pain worsens. We have the same problem with hypertension. We normally tell patients with hypertension to come back in one week, but patients usually say, 'I forgot to come back and I feel fine, so I'll stop taking my medication.'* (Participant 1, CO, health center.)"
Discussion {#ihz061s8}
==========
This study identified multiple actionable barriers to ACS care at the levels of patients, providers and the healthcare system in Tanzania. Participants described providers who had inadequate knowledge and frequently missed cases of ACS, a healthcare system characterized by insufficient supplies and inefficient referrals, and patients who lacked education about proper care-seeking behavior. The findings presented here suggest that a multifaceted strategy is needed to improve ACS care in Tanzania, and that such a strategy should include improved provider training, expansion of access to diagnostic equipment and treatment, research to quantify the burden of disease, improvements in referral system efficiency, development of locally relevant guidelines and patient education. In 2016, the Tanzanian Ministry of Health released a strategic action plan to combat non-communicable diseases that emphasized health system strengthening.^[@ihz061C20]^ The results of this study highlight the importance of this mission to improving care for ACS and non-communicable diseases in general.
Prior research in Tanzania demonstrated that referral system delays led to increased mortality for patients with infectious diseases,^[@ihz061C21]^ suggesting that efforts to streamline the referral system may improve outcomes for both communicable and non-communicable diseases. Beyond health system strengthening, however, our findings identified a need for provider and patient educational interventions to improve ACS care in Tanzania. In high-income settings, both health system and educational interventions have been shown to improve ACS outcomes,^[@ihz061C22]--[@ihz061C24]^ and implementation science research is needed to determine whether a package of evidence-based interventions can be implemented in SSA to impact ACS outcomes.
Delayed or inappropriate patient healthcare-seeking was identified as an important barrier to ACS care in our study. This finding is consistent with the results of a recent community survey conducted in northern Tanzania, which found that most adults would not present to a hospital for ACS symptoms like chest pain, were not aware that chest pain might be a symptom of cardiovascular disease and did not perceive themselves to be at risk of ACS.^[@ihz061C17],[@ihz061C25]^ As ACS is a life-threatening emergency requiring time-sensitive treatments,^[@ihz061C26]^ our findings highlight the urgent need for community education regarding ACS in Tanzania.
Cost of care was another significant barrier identified in our study, particularly for those without health insurance. In 2015, the Tanzanian government released the Fourth National Health Sector Strategic Plan, which included financing reforms to help the country achieve universal health coverage (UHC).^[@ihz061C27]^ Although the country has made considerable progress towards UHC in recent years,^[@ihz061C28]^ substantial gaps in coverage remain: less than one-third of Tanzanians have health insurance and catastrophic health expenditures are still common among the uninsured.^[@ihz061C28]--[@ihz061C30]^ Indeed, although many of the medications needed to manage ACS appear on the national essential medicines list and are partially subsidized,^[@ihz061C19]^ our findings suggest that the cost of ACS testing and treatment remain prohibitive for many patients. As Tanzania makes progress towards achieving the Sustainable Development Goals, additional efforts to expand health coverage and reduce costs for care of non-communicable diseases are needed.
The findings of this study must be considered in light of its limitations. First, social desirability bias may have made participants feel social pressure to appear competent and knowledgeable when speaking to a non-Tanzanian physician. This may have resulted in minimizing concerns about their own inadequacies in knowledge or quality of care. Participants did speak with candor, however, about general clinician incompetence related to ACS. Second, this study included only physicians and COs; additional interviews with patients, caregivers, nurses and administrators may identify additional important barriers to ACS care in Tanzania.
In conclusion, healthcare providers in northern Tanzania identified multiple barriers to ACS care at the patient, provider and healthcare system levels. Locally relevant interventions in all three areas are needed to improve ACS care and outcomes. As ACS shares common risk factors with a wide range of other cardiovascular diseases, the findings presented here may also have implications for control strategies for other non-communicable diseases. As the burden of cardiovascular disease across SSA continues to rise, it will be increasingly imperative to implement strategies to address barriers to ACS care.
JTH and FMS conceived the study; JTH, PM and FMS developed the interview guide; GLK and FMS recruited participants; JTH conducted the interviews; JTH and GLK performed the coding; JTH, PM and MHW conducted the thematic analysis; JTH drafted the manuscript; GLK, PM, MHW and FMS critically revised the manuscript for intellectual content. All authors read and approved the final manuscript. JTH and FMS are guarantors of the paper.
**Acknowledgements:** None.
**Funding:** This study received support from the Fogarty International Center (grant number D43TW009337).
**Competing interests:** None declared.
**Ethical approval:** This study was conducted in accordance with the ethical standards of the Helsinki Declaration. This study received ethical approval from the Duke Health Institutional Review Board, the Kilimanjaro Christian Medical Centre Research Ethics Committee and the Tanzania National Institutes for Medical Research Ethics Coordinating Committee. All participants provided informed consent prior to enrollment.
| {
"pile_set_name": "PubMed Central"
} |
INTRODUCTION
============
The hair follicle (HF) is a unique mini organ in the skin of mammals that produces hair. Hair functions to thermoregulate and protect the skin, helps detect senses, and a marker for sexual development. HFs undergo a continuous cycle consisting of growth (anagen), regression (catagen), and relative quiescence (telogen) phases. Growth of new hair shafts and regression of old hair shafts are repeated through this cycle[@B1]. Multiple signaling pathways and genes are involved in regulating the hair cycle and HF morphogenesis[@B2]-[@B4].
Wnts are a family of secreted glycoproteins that play a critical role in the embryonic development and maintenance of homeostasis in mature tissues by regulating proliferation, differentiation, migration, and apoptosis of cells[@B5]-[@B8]. The Wnt pathway regulates the regeneration and differentiation of HFs[@B9]-[@B11]. Wnt signaling pathway activity is regulated by secreted inhibitors, which are divided into two functional classes. One class includes the secreted frizzled-related protein (*Sfrp*) family, Wnt inhibitory factor 1 (*Wif1*), and Cerberus which binds directly to Wnt, thus, sequestering Wnt from its receptors. The Dickkopf (*Dkk*) class binds to the LRP5/LRP6 component, thereby inhibiting binding of Wnt to its receptor[@B12].
The other class, sFRP, a family of glycoproteins, has five members (Sfrp1\~Sfrp5) that participate in the developmental and disease processes in various cells and tissues by controlling the Wnt pathway[@B13],[@B14]. The sFRP family possesses structural similarities to the cysteine-rich domain of frizzled receptors; thus, they interact with Wnt and form the Wnt-Sfrp complex. Therefore, the Wnt protein is sequestered from the frizzled receptor[@B15].
*Sfrp2* has been studied in different cell types and organs to determine its role[@B16]-[@B18]. Although *Sfrp2* is expressed in the upper and lower matrix, outer root sheath, and dermal papilla[@B19],[@B20], its role in HFs has not been investigated.
Thus, we investigated *Sfrp2* expression at various HF stages in mouse dorsal skin and the effect of SFRP2 on keratinocytes to identify the role of SFRP2 in HFs. We found that *Sfrp2* expression peaked at the catagen phase, and that this expression pattern was inversely related to that of Wnt target genes. We also observed inhibited mouse keratinocyte proliferation by SFRP2. These results suggest that *Sfrp2* may play an important role during the catagen phase by inhibiting keratinocyte proliferation in HFs.
MATERIALS AND METHODS
=====================
Animal
------
BALB/C mice were purchased from Orient Bio (Seongnam, Korea) and maintained in the barrier system with regulated light (700 to 1,900 h), temperature (23±1℃), humidity (50%±5%), and ventilation (10 to 12 times/h). All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Catholic University of Korea and were carried out in accordance with the Guidelines for Animal Experimentation.
Histological study
------------------
Mouse dorsal skins of BALB/C mice at postnatal days 10 (P10), P14, P17, P21 and P28 were gathered as previously described[@B21]. All skin samples were harvested from the same region (2×2 cm) of the upper back skin. Six-micrometer-thick paraffin sections were prepared. Hematoxylin and eosin (H&E) staining was carried out using the standard method, and the stained sections were observed with an optical microscope (Olympus, Tokyo, Japan).
RNA isolation and reverse transcription
---------------------------------------
Total RNA was extracted from the dorsal skins of BALB/C mice at P10, P12, P14, P17, P21, P28, and P35 using TRIZOL following the manufacturer\'s instructions (Invitrogen, Camarillo, CA, USA). Two micrograms of RNA were utilized to synthesize single stranded cDNAs using the PrimeScript 1st strand cDNA Synthesis kit following the manufacturer\'s instructions (TaKaRa Bio Inc., Shiga, Japan).
Reverse transcription-polymerase chain reaction and Real-time polymerase chain reaction
---------------------------------------------------------------------------------------
Polymerase chain reaction (PCR) was performed using Thermal Cycler-100 (MJ Research Inc., Waltham, MA, USA). PCR conditions were as follows: initial denaturation for 2 minutes at 95℃ followed by 28\~30 cycles of 15 seconds at 94℃, 15 seconds at 62℃, and 15 seconds at 72℃. The final extension was performed for 10 minutes at 72℃. Real-time PCR was performed with the cDNAs prepared from the total RNAs of the skin at various time points (See the above section) using SYBR Premix Ex Taq (TaKaRa Bio Inc.) in an Mx3000P (Stratagene, La Jolla, CA, USA). Specific primer sequences for each gene are listed in [Table 1](#T1){ref-type="table"}. The cycling conditions were as follows; initial heating for 2 minutes at 95℃ followed by 45 cycles of 15 seconds at 94℃, 15 seconds at 62℃, and 15 seconds at 72℃, and the final extension was performed for 10 minutes at 72℃. Gene expression levels were determined by the comparative ΔΔCt method using the expression level of *Gapdh* as control[@B22]. Relative mRNA expression levels were determined based on Real-time PCR performed in duplicate using various numbers of independent samples for each point (number of sample for each point; P10=3, P12=4, P14=7, P17=4, P21=5, P28=5, P35=4).
Plasmids
--------
The 888 bp of *Sfrp2* coding sequence (CDS) were amplified by PCR using Expand High Fidelity enzyme (Roche Diagnostics, Basel, Switzerland) from skin cDNAs of BALB/C mice. Forward and reverse primer sequences are listed in [Table 1](#T1){ref-type="table"}. PCR products were subcloned into pcDNA 3.1 using EcoRI cloning sites (Invitrogen).
Cell culture and transfection
-----------------------------
Mouse keratinocyte cells (PAM212 cell line) were cultured in DMEM (Invirogen) containing 10% FBS with 5% CO~2~ in a 37℃ incubator. Transfection experiments were performed using polyethyleneimine (Sigma-Aldrich, St. Louis, MO, USA) according to the manufacturer\'s instructions. A total of 8×10^5^ cells were plated in 60 mm dishes in triplicate. Following this, either 500 ng of pcDNA 3.1/ *Sfrp2* CDS or only pcDNA 3.1 plasmid was transfected into the cells with 0.4 µg of pCMV3.1/β-gal. Transfection experiments were normalized against transfection efficiency determined by β-galactosidase activity. To ensure that *Sfrp2* overexpression was well induced, we performed Real-time PCR and found a 27118-fold increase in the *Sfrp2* mRNA level compared with that of pcDNA 3.1 transfection. After incubation for 48 hours at 37℃, the transfected cells were observed using a microscope (Olympus) or harvested for extraction of total RNAs.
SFRP2 protein treatment
-----------------------
SFRP2 protein was purchased from R&D system and reconstituted in sterilized phosphate buffered saline (PBS). A total of 5×10^5^ PAM212 cells were plated on a 100 mm culture dish in DMEM media and were treated with SFRP2 the following day by replacing media with the media containing SFRP2 (50 ng/ml or 100 ng/ml). The treatment was continued for 3 days by changing the media every day.
Immunocytochemistry
-------------------
For immunocytochemistry, PAM212 cells either treated with SFRP2 or transfected with pcDNA 3.1/*Sfrp2* were washed three times with cold PBS and fixed using 4% paraformaldehyde for 10 minutes at room temperature. The cells were then treated with 0.5% Triton X-100 for 10 minutes and washed three times with PBS. After blocking with 3% bovine serum albumin for 1 hour, the cells were incubated with antibodies against Ki67 (Thermo Scientific, Hudson, NH, USA; 1:200) or *Involucrin* (Covance, Berkeley, CA, USA; 1:200) overnight. After washing, Alexa Fluor 488 goat anti-rabbit secondary antibody was applied for 3hr (Invitrogen, 1:500), and 4 g/ml of Hoechst 33342 (Sigma-Aldrich) was also applied for 1 minute to stain the nuclei. Fluorescence signal was observed with a Fluorescent microscope (Olympus).
Statistical analysis
--------------------
*p*-values were calculated using Student\'s t-test. *p*\<0.05 was regarded as statistically significant. Analyses were performed using SPSS for Windows version 12.00 (SPSS Inc., Chicago, IL, USA).
RESULTS
=======
*Sfrp2* mRNA expression during the hair cycle
---------------------------------------------
To determine the role of *Sfrp2* in HFs, first we investigated the relative expression level of *Sfrp2* compared to those of other Wnt inhibitors using reverse transcription-polymerase chain reaction (RT-PCR) and Real-time PCR. Among the Wnt inhibitors with relatively high expression at P10, *Sfrp2* was the only one whose expression was dramatically increased at P17. Furthermore, the expression level of *Sfrp2* was 5.9-fold higher than that of the second most expressed Wnt inhibitor, *wif1* ([Fig. 1](#F1){ref-type="fig"}). These results suggest that *Sfrp2* is the most abundantly expressed Wnt inhibitor in HFs at the catagen phase.
Relative expression level of *Sfrp2* mRNA was determined at different hair cycle stages by RT-PCR as well as Real-time PCR analyses. We found that *Sfrp2* was highly expressed during catagen, which was maintained until the early telogen phase ([Fig. 2A](#F2){ref-type="fig"}). During anagen, *Sfrp2* expression was weaker than those of other stages. After this, the expression level increased significantly by 2.4\~3.6 fold as the stages progressed into the catagen and early telogen phases. As the next anagen phase began, *Sfrp2* expression started to decline and then decreased again to a low level that is similar to that of the first anagen phase by P35 ([Fig. 2B](#F2){ref-type="fig"}). These *Sfrp2* mRNA expression patterns suggest that *Sfrp2* may function during the catagen phase.
Effect of SFRP2 on keratinocyte proliferation
---------------------------------------------
During the catagen phase, HFs undergo degeneration through cessation of proliferation and subsequent differentiation of keratinocytes. Because *Sfrp2* expression increased at the catagen phase as shown in [Fig. 2](#F2){ref-type="fig"}, we investigated whether *Sfrp2* is involved in the regulation of proliferation and/or differentiation using the PAM212 mouse keratinocyte cell line. Cells were treated with SFRP2 at 50 and 100 ng/ml for 48 hours, and the number of live cells was counted using the trypan blue exclusion method. We found that SFRP2 significantly reduced the number of cells by 20% compared to the mock-treated control at both 50 and 100 ng/ml ([Fig. 3A, B](#F3){ref-type="fig"}).
To investigate whether this decrease was caused by inhibition of cell proliferation, we used immunofluorescent staining for the Ki67 proliferation marker[@B23]. Immunocytochemistry showed decreased Ki67 expression in cells treated with SFRP2 and in cells overexpressing SFRP2 ([Fig. 3C](#F3){ref-type="fig"}), indicating that SFRP2 inhibits keratinocyte proliferation.
Effect of SFRP2 on *Involucrin* expression in keratinocytes
-----------------------------------------------------------
To investigate whether SFRP2 treatment would affect the differentiation status of keratinocytes, we performed immunocytochemistry for *Involucrin* expression, a keratinocyte differentiation marker[@B24], in SFRP2 treated or *Sfrp2*-overexpressed PAM212 cells. Unlike Ki67, we found no difference in *Involucrin* expression between control and SFRP2-treated cells ([Fig. 4A](#F4){ref-type="fig"}).
RT-PCR and Real-time PCR analyses also revealed no significant difference in the *Involucrin* mRNA level between control and SFRP2-overexpressed cells ([Fig. 4B, C](#F4){ref-type="fig"}). These results indicate that *Sfrp2* does not affect *Involucrin* expression and suggest that *Sfrp2* is not involved in keratinocyte differentiation.
Effect of SFRP2 on the Wnt pathway during the hair cycle
--------------------------------------------------------
SFRP2 function in the Wnt pathway is controversial. SFRP2 was reported to be inhibitory in the Wnt pathway[@B12], whereas it was also shown to activate Wnt signaling through the canonical activation of β-catenin in intestinal epithelial cells and canine mammary glands[@B25]-[@B28]. To determine SFRP2 function in the Wnt pathway of HFs, we investigated the expression of the wnt target genes *Ccnd1* and *C-myc* on the back skin of mice. Real-time PCR analysis revealed an inverse relationship between the *Ccnd1* and *C-myc* expression pattern and that of *Sfrp2* throughout the HF cycle ([Fig. 5](#F5){ref-type="fig"}), suggesting that *Sfrp2* may function as a Wnt inhibitor in HFs.
DISCUSSION
==========
Many signaling pathways crosstalk with each other to control HF development and cycling[@B29]. Among these signaling pathways, the Wnt pathway has been widely accepted to function in hair morphogenesis and HF cycling[@B30]-[@B33]. Excess induction of Wnt signaling causes abnormal hair cycles and abnormal formation of HFs[@B34]. Several mice with Wnt inhibitor mutations have been reported. For example, a *Dkk1* or *Dkk2* mutation in mice causes complete hair loss through failure to initiate development and mis-spacing of HFs[@B6],[@B9], showing that *Dkk1* and *Dkk2* are essential in normal HF formation and control of the hair cycle. Although mice with the *Sfrp2* mutation have been reported and studied[@B35], its function in the normal hair cycle has not been documented.
In this study, we found that *Sfrp2* is mainly expressed during the catagen phase and inhibited mouse keratinocyte proliferation. This inhibition was also observed in human HaCaT keratinocytes (data not shown). These results suggest that *Sfrp2* may function in the catagen phase by inhibiting keratinocyte proliferation. The inhibition rate by SFRP2 in keratinocytes was 20%, suggesting that SFRP2 may not be the main catagen regulator. Nevertheless, SFRP2 must be considered a contributor in the catagen phase of HFs.
Several investigators have shown that the action of *Sfrp2* in the Wnt pathway differs in different cell types. These studies suggest that *Sfrp2* acts not as a Wnt inhibitor but as an activator[@B25],[@B26]. However, in our study, we showed that *Sfrp2* expression is inversely related to Wnt target gene expression ([Fig. 5](#F5){ref-type="fig"}), suggesting that *Sfrp2* is a Wnt inhibitor in HFs. In addition to *Sfrp2* expression, other Wnt inhibitors are also expressed in HFs ([Fig. 1](#F1){ref-type="fig"}). Thus, Wnt target gene expression in HFs must be the sum of Wnt inhibitor expression. Therefore, although *Sfrp2* acts as a Wnt inhibitor in the normal hair cycle, a further study is required to understand its precise function in HFs.
SFRP2 is also known to regulate the differentiation of myoblasts and osteoblasts[@B18],[@B36]. Our study found that SFRP2 treatment on keratinocytes did not affect the expression of Involucrin, a differentiation marker of keratinocytes. Furthermore, there was no significant difference between the control and SFRP2 treated human HaCaT keratinocytes (data not shown). These results suggest that SFRP2 may not play a role in the differentiation of keratinocytes.
In conclusion, our results suggest that SFRP2 plays a role in the catagen phase by inhibiting keratinocyte proliferation. A further study with microdissection of the epidermis is needed for a better understanding of SFRP2 function in hair cycle regulation.
This research was supported by the Basic Science Research Program of the National Research Foundation of Korea (NRF) and funded by the Ministry of Education, Science and Technology (313-2008-2-E00397).
{#F1}
{#F2}
{#F3}
{#F4}
{#F5}
######
List of gene specific primers
RT: real-time, CDS: coding sequence.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1-molecules-25-01488}
===============
Bacterial diseases of rice plants will lead to the reduction of rice yield and hence serious decreases in crop quality and insufficient food supply \[[@B1-molecules-25-01488],[@B2-molecules-25-01488],[@B3-molecules-25-01488]\]. Bacterial leaf blight of rice infected by *Xanthomonas oryzae* pv. *oryzae (Xoo)* will reduce rice yield by affecting rice growth \[[@B4-molecules-25-01488],[@B5-molecules-25-01488]\]. Citrus canker, the devastating citrus disease caused by *Xanthomonas axonopodis* pv. *citri (Xac)*, can severely affect citrus production \[[@B6-molecules-25-01488],[@B7-molecules-25-01488]\]. Bismerthiazol (BT) and thiodiazole copper (TC) are traditional systemic fungicides, which are commonly used to treat rice bacterial leaf blight and citrus canker \[[@B8-molecules-25-01488],[@B9-molecules-25-01488]\]. However, the long-term frequent application of them has led to bactericide-resistant, therefore the phenomenon that rice bacterial leaf blight and citrus canker cannot be effectively controlled has emerged \[[@B10-molecules-25-01488]\]. So, it is urgent to develop efficient new chemical pesticides to deal with this problem.
We previously found that 1,3,4-oxadiazole derivatives have a variety of biological effects, including antibacterial \[[@B11-molecules-25-01488],[@B12-molecules-25-01488],[@B13-molecules-25-01488]\], antifungal \[[@B14-molecules-25-01488],[@B15-molecules-25-01488]\], antiviral \[[@B16-molecules-25-01488]\], nematocidal \[[@B17-molecules-25-01488]\] and insecticidal \[[@B18-molecules-25-01488],[@B19-molecules-25-01488]\] activity. 1,3,4-oxadiazole has the ideal heterocyclic structure to be developed into efficient pesticides. Meantime, sulfonate or carboxylate derivatives have broad-spectrum biological activity in agriculture, such as insecticidal \[[@B20-molecules-25-01488]\], antibacterial \[[@B21-molecules-25-01488]\], antiviral \[[@B22-molecules-25-01488]\] and antifungal \[[@B23-molecules-25-01488]\] activity. For example, pyraoxystrobin \[[@B24-molecules-25-01488]\], chlorfenson \[[@B25-molecules-25-01488]\] and nimrod \[[@B26-molecules-25-01488]\] ([Figure 1](#molecules-25-01488-f001){ref-type="fig"}) containing sulfonate or carboxylate respectively have been widely used in agriculture \[[@B27-molecules-25-01488],[@B28-molecules-25-01488],[@B29-molecules-25-01488]\].
In addition, we reported the splicing of oxymethyl and 1,3,4-oxadiazole sulfone derivatives could provide excellent antibacterial activity \[[@B10-molecules-25-01488],[@B30-molecules-25-01488]\]. Based on those prior works, we hypothesized that sulfonate/carboxylate moiety functionalized 1,3,4-oxadiazole derivatives might also show promising antibacterial activity. Hence in the present work, a series of novel compounds were synthesized by introducing sulfonate or carboxylate moiety to 1,3,4-oxadiazole to discover new structures with potential antibacterial activity. The design and synthesis of the targets are depicted in [Figure 1](#molecules-25-01488-f001){ref-type="fig"} and [Scheme 1](#molecules-25-01488-sch001){ref-type="scheme"} respectively.
2. Results and Discussion {#sec2-molecules-25-01488}
=========================
2.1. Chemistry {#sec2dot1-molecules-25-01488}
--------------
As described in [Scheme 1](#molecules-25-01488-sch001){ref-type="scheme"}, starting from ethyl glycolate, the key intermediate (5-mercapto-1,3,4-oxadiazol-2-yl) methanol **2** was synthesized in two steps involving acylation and cyclization. Subsequently, intermediate **2** was converted into its corresponding thioether derivative **3** by thioetherification with R~1~I. Finally, the target compounds **4a/5a** was obtained by esterification with R~2~SOOCl/R~3~COCl. The structures of all target compounds were confirmed by nuclear magnetic resonance spectra including ^1^H NMR, ^13^C NMR and electrospray ionization high-resolution mass spectrometry (ESI-HRMS). Fluorine nuclear magnetic resonance (^19^F NMR) was involved for some fluoride structures.
2.2. In Vitro Antibacterial Activity {#sec2dot2-molecules-25-01488}
------------------------------------
The antibacterial activity of all the compounds was evaluated in vitro against *Xanthomonas oryzae* pv. *oryzae (Xoo)* and *Xanthomonas axonopodis* pv. *(Xac)* via the turbidimeter test \[[@B10-molecules-25-01488]\]. Bismerthiazol and thiodiazole copper served as positive controls to compare the bactericidal potency of the tested compounds.
As shown in [Table 1](#molecules-25-01488-t001){ref-type="table"}, most of the compounds **4a** exhibited higher antibacterial activity than either bismerthiazol or thiodiazole copper against the tested plant bacteria. Among them, inhibitory rates for *Xoo* of compounds **4a-1--4a-5** and **4a-11-4a-16** at 100 µg/mL as well as **4a-1**, **4a-2**, **4a-14** at 50 µg/mL were all above 90%. Inhibitory rates for *Xac* of compounds **4a-1**--**4a-3**, 93%--97% at 200 µg/mL and 69%--82% at 100 µg/mL were also superior to those of positive controls. At the same time, the present tests were parallelly conducted on compounds **5a**. Actually, no similar tendency was observed on **5a** against tested bacteria. It was confirmed that compounds **4a** bearing sulfonate moiety were more potent in combating *Xoo* and *Xac* and presented remarkable higher activity as compared to compounds **5a** and positive controls.
Further, compounds acting better than positive controls bismerthiazol or thiodiazole copper ([Table 1](#molecules-25-01488-t001){ref-type="table"}) were performed for their EC~50~ values ([Table 2](#molecules-25-01488-t002){ref-type="table"}). Compounds **4a-1-4a-4** and **4a-11-4a-16** revealed outstanding activity against *Xoo* with EC~50~ values around 50.1--112.5 µM, which was lower than bismerthiazol (253.5 µM) and thiodiazole copper (467.4 µM). Meanwhile, EC~50~ (95.8--155.2 µM) values of **4a-1-4a-4** against *Xac* were also lower than 274.3 µM displayed by bismerthiazol and 406.3 µM by thiodiazole copper.
In particular, **4a-2** bearing 4-F substituted benzenesulfonate, performed the best on *Xoo* and *Xac* with EC~50~ values of 50.1 and 95.8 µM respectively, which were quite better than two commercial positive controls. So, compound **4a-2** appeared to be promising antibacterial agents against plant bacterial diseases.
2.3. In Vivo Antibacterial Activity {#sec2dot3-molecules-25-01488}
-----------------------------------
With outstanding bactericidal activity of compounds **4a-1, 4a-2, 4a-3** in vitro, they were further explored for their antibacterial potency in vivo against rice bacterial leaf blight via the leaf-cutting method \[[@B10-molecules-25-01488]\]. Bismerthiazol and thiodiazole copper served as positive controls for this investigation. All inoculated plants in 14 days exhibited blight symptoms with 100% morbidity.
At the concentration of 200 µg/mL, as shown in [Figure 2](#molecules-25-01488-f002){ref-type="fig"} and [Table 3](#molecules-25-01488-t003){ref-type="table"}, the control efficiency of the protection activity of compounds **4a-2** and **4a-3** were 68.6% and 62.3%, which were superior to Bismerthiazol (49.6%) and Thiodiazole copper (42.2%). As shown in [Figure 3](#molecules-25-01488-f003){ref-type="fig"} and [Table 4](#molecules-25-01488-t004){ref-type="table"}, the control efficiency of the curative activity of compound **4a-1**, **4a-2** and **4a-3** were 44.6%, 62.3% and 56.0%, which were superior to bismerthiazol (42.9%) and thiodiazole copper (36.1%).
2.4. Scanning Electron Microscopy Studies {#sec2dot4-molecules-25-01488}
-----------------------------------------
Scanning electron microscopy offers the ability to observe the bacterial cell surface \[[@B30-molecules-25-01488]\]. Based on the analysis of antibacterial against *Xoo* results in vitro and in vivo, the antibacterial mechanism of compound **4a-2** was studied by SEM. As shown in the [Figure 4](#molecules-25-01488-f004){ref-type="fig"}, when the compound **4a-2** was at a concentration of 25 µg/mL, the bacterial cell was deformed, and part of the bacterial cell wall was slightly ruptured. When the compound **4a-2** concentration was increased to 50 µg/mL, most cell membrane were wrinkled and ruptured. Then observing the control group (A), these bacterial cells were round and smooth, without any breakage. Scanning electron microscopy images had further demonstrated that the compounds **4a-2** have antibacterial activity against *Xoo*.
2.5. Structure-Activity Relationship (SAR) Analyses {#sec2dot5-molecules-25-01488}
---------------------------------------------------
Based on the activity values shown in [Table 1](#molecules-25-01488-t001){ref-type="table"} and [Table 2](#molecules-25-01488-t002){ref-type="table"}, a preliminary conclusion could be drawn about the structure-activity relationship. First, according to the antibacterial research of those **4a** and **5a** derivatives, it had shown that compound **5a** derivatives containing the sulfonate structure was significantly higher efficient than the corresponding **5a** containing the carboxylate structure. Obviously, the existence of the sulfonate structure was very important to improve inhibitory effect.
Further antibacterial evaluation on *Xoo* and *Xac* showed that 4-substituted halogenated phenyl sulfonate derivatives expressed significant antibacterial activity. Three (**4a-2**, **4a-3**, **4a-4**) of them worked well on both *Xoo* and *Xac*, which appeared an obvious decreasing potency (50.1, 87.2, 99.4 µM) with increasing halogen size in **4a-2**(R~2~ = F), **4a-3**(R~2~ = Cl), **4a-4**(R~2~ = Br) respectively. In this regard, it was consistent with previous reports \[[@B10-molecules-25-01488]\]. The other six compounds **4a-11-4a-16** (R~1~ = C~2~H~5~) also showed extensive potency on the *Xoo*. However, their EC~50~ are slightly decreased like 98, 95, 86.4 µM and not necessarily following the tendency **4a-11**(F) \>**4a-12**(Cl) \>**4a-13**(Br). It can be refereed that R~1~ in thioether side chain also make difference in the activity of the structure. So in particular, when R~1~ = CH~3~, R~2~ = F, compound **4a-2** would be the most promising compound both in vitro and in vivo against the tested plant bacteria.
3. Experimental {#sec3-molecules-25-01488}
===============
3.1. Chemicals and Instruments {#sec3dot1-molecules-25-01488}
------------------------------
All reagent products from the Chinese Chemical Reagent Company were analytical or chemical pure. Thin-layer chromatography (TLC) of a GF254 silica gel pre-coated plate (Qingdao Haiyang Chemical Co., Ltd., Qingdao China) was used to evaluate the progress of the reaction and the purity of the compounds. Melting points were determined using an XT-4 digital melting-point apparatus (Beijing Tech. Instrument Co., Beijing, China) and reading was uncorrected. ^1^H NMR, ^13^C NMR and ^19^F NMR spectra were recorded on a 400 MHz spectrometer (Swiss Bruker, Swiss, Germany) with CDCl~3~ or (CD~3~)~2~CO-*d~6~* as the solvent. The antibacterial mechanism was studied by scanning electron microscopy (FEI, Hillsboro Oregon, America). Single crystal structure was collected by single crystal diffractometer (Gemini E, Oxford, United Kingdom). High-resolution mass spectral (HRMS) data were performed with Thermo Scientific Q Exactive (Thermo, Waltham, MA, USA).
3.2. General Synthetic Procedure for the Target Compounds {#sec3dot2-molecules-25-01488}
---------------------------------------------------------
### 3.2.1. Preparation of Intermediate 1 {#sec3dot2dot1-molecules-25-01488}
Ethyl glycolate (0.05 mol) was dissolved with 100 mL ethanol in a round bottom flask. Then, 80% of hydrazine hydrate (0.1 mol) was slowly added to the round bottom flask at room temperature. After a day of reaction, white product **1** will precipitate out in 85%--90% yields.
### 3.2.2. Preparation of Intermediate 2 {#sec3dot2dot2-molecules-25-01488}
To a three-necked round bottom flask was added intermediate **1** (0.01 mol), KOH (0.012 mol) and 100 mL of ethanol in this order. Then, carbon disulfide (0.012mol) was slowly added under a stirred condition. The mixture was reacted at room temperature for 1--2 h and then heated to 78 °C for refluxing of six hours. The solution was removed under reduced pressure on a rotary evaporator, and product **2** was obtained in 65%--70% yields.
### 3.2.3. Preparation of Intermediate 3 {#sec3dot2dot3-molecules-25-01488}
Tetrahydrofuran was used for dissolving intermediate **2** (0.01 mmol), then added KOH (0.012 mmol) and R~1~I (0.012 mmol). The reaction was judged complete by TLC, the solvent was removed under reduced pressure, and the residue was purified by silica gel column chromatography to obtain Intermediate **3** in yield of 70%--80%.
### 3.2.4. Preparation of Target Compound 4a/5a {#sec3dot2dot4-molecules-25-01488}
At room temperature, added intermediate **3** (0.001 mol), tetrahydrofuran (10 mL), and sodium hydride (0.001 mol) to the round-bottomed flask in order. After stirring for 30 min, R~2~SOOCl/R~2~COCl (0.001 mol) was slowly added, and the reaction was followed by TLC and filtered to get **4a/5a**.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl benzenesulfonate* (**4a-1)**. White solid; m.p.: 54--55 °C; yield, 80.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.93 (dd, *J* = 8.4, 1.2 Hz, 2H, Ar-H), 7.70 (t, *J* = 7.5 Hz, 1H, Ar-H), 7.58 (t, *J* = 7.8 Hz, 2H, Ar-H), 5.23 (s, 2H, -CH~2~-), 2.69 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.60, 160.69, 135.10, 134.50, 129.49, 128.13, 59.75, 14.51. HRMS calculated for C~10~H~11~O~4~N~2~S~2~ \[M + H\]^+^ 287.01547, found 287.01529.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-fluorobenzenesulfonate* (**4a-2)**. White solid; m.p.: 64--65 °C; yield, 86.5%; ^1^H NMR (400 MHz, Acetone) δ 8.04 (dd, *J* = 9.0, 5.0 Hz, 2H, Ar-H), 7.48 (t, *J* = 8.8 Hz, 2H, Ar-H), 5.44 (s, 2H, -CH~2~-), 2.71 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, Acetone) δ 166.81, 166.10 (d, *J* = 254.9 Hz), 161.21, 131.72 (d, *J* = 3.5 Hz), 131.30 (d, *J* = 10.2 Hz), 116.91 (d, *J* = 23.3 Hz), 60.57, 13.67. ^19^F NMR (376 MHz, Acetone) δ -104.45. HRMS calculated for C~10~H~10~FO~4~N~2~S~2~ \[M + H\]^+^ 305.00605, found 305.00592.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-chlorobenzenesulfonate* (**4a-3**). White solid; m.p.: 85--86 °C; yield, 86.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.78 (d, *J* = 8.8 Hz, 2H, Ar-H), 7.47 (d, *J* = 8.8 Hz, 2H, Ar-H), 5.18 (s, 2H, -CH~2~-), 2.63 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.73, 160.51, 141.34, 133.69, 129.80, 129.53, 59.92, 14.52. HRMS calculated for C~10~H~10~O~4~N~2~ClS~2~ \[M + H\]^+^ 320.97650, found 320.97629.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-bromobenzenesulfonate* (**4a-4**). White solid; m.p.: 79--80 °C; yield, 76.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.69 (d, *J* = 8.7 Hz, 2H, Ar-H), 7.64 (d, *J* = 8.7 Hz, 2H, Ar-H), 5.18 (s, 2H, -CH~2~-), 2.63 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.76, 160.48, 134.18, 132.80, 129.93, 129.55, 59.97, 14.55. HRMS calculated for C~10~H~10~O~4~N~2~BrS~2~ \[M + H\]^+^ 364.92599, found 364.92548.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-methoxybenzenesulfonate* (**4a-5**). White solid; m.p.: 63--64 °C; yield, 66.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.85 (d, *J* = 9.0 Hz, 2H, Ar-H), 7.02 (d, *J* = 8.9 Hz, 2H, Ar-H), 5.19 (s, 2H, -CH~2~-), 3.90 (s, 3H, -CH~3~), 2.69 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.50, 164.34, 160.87, 130.46, 126.20, 114.67, 59.51, 55.83, 14.50. HRMS calculated for C~11~H~13~O~5~N~2~S~2~ \[M + H\]^+^ 317.02604, found 317.02472.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-nitrobenzenesulfonate* (**4a-6**). White solid; m.p.: 86--87 °C; yield, 74.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 8.42 (d, *J* = 9.0 Hz, 2H, Ar-H), 8.12 (d, *J* = 9.0 Hz, 2H, Ar-H), 5.36 (s, 2H, -CH~2~-), 2.70 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.88, 160.22, 151.07, 140.95, 129.52, 124.59, 60.42, 14.50. HRMS calculated for C~10~H~10~O~6~N~3~S~2~ \[M + H\]^+^ 332.00055, found 332.00040.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-(trifluoromethyl)benzenesulfonate* (**4a-7**). White solid; m.p.: 60--61 °C; yield, 84.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.98 (d, *J* = 8.2 Hz, 2H, Ar-H), 7.77 (d, *J* = 8.3 Hz, 2H, Ar-H), 5.23 (s, 2H, -CH~2~-), 2.61 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 166.76, 159.29, 137.84, 135.19, 134.86, 127.66, 125.55 (q, *J* = 3.6 Hz), 123.24, 120.53, 59.16, 13.39. ^19^F NMR (376 MHz, CDCl~3~) δ -63.36. HRMS calculated for C~11~H~10~O~4~N~2~F~3~S~2~ \[M + H\]^+^ 355.00286, found 355.00241.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 2-fluorobenzenesulfonate* (**4a-8**). White solid; m.p.: 51--52 °C; yield, 82.5%; ^1^H NMR (400 MHz, Acetone) δ 7.97--7.87 (m, 2H, Ar-H), 7.52--7.46 (m, 2H, Ar-H), 5.52 (s, 2H, -CH~2~-), 2.71 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, Acetone) δ: 166.87, 161.12, 159.11 (d, *J* = 257.6 Hz), 137.54 (d, *J* = 8.8 Hz), 130.94, 125.17 (d, *J* = 3.9 Hz), 123.50 (d, *J* = 13.9 Hz), 117.59 (d, *J* = 20.7 Hz), 61.03, 13.68. ^19^F NMR (376 MHz, Acetone) δ -109.11. HRMS calculated for C~10~H~10~FO~4~N~2~S~2~ \[M + H\]^+^ 305.00605, found 305.00592.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 3-fluorobenzenesulfonate* (**4a-9**). White liquid; yield, 72.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.66 (ddd, *J* = 7.9, 1.6, 1.0 Hz, 1H, Ar-H), 7.59--7.54 (m, 1H, Ar-H), 7.51 (td, *J* = 8.1, 5.2 Hz, 1H, Ar-H), 7.34 (tdd, *J* = 8.3, 2.5, 0.9 Hz, 1H, Ar-H), 5.19 (s, 2H, -CH~2~-), 2.63 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.75, 162.34 (d, *J* = 253.3 Hz), 160.43, 136.89 (d, *J* = 7.2 Hz), 131.38 (d, *J* = 7.8 Hz), 123.94 (d, *J* = 3.5 Hz), 121.87 (d, *J* = 21.1 Hz), 115.51 (d, *J* = 24.9 Hz), 60.01, 14.50. ^19^F NMR (376 MHz, CDCl~3~) δ -108.20. HRMS calculated for C~10~H~10~FO~4~N~2~S~2~ \[M + H\]^+^ 305.00605, found 305.00571.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 3-chlorobenzenesulfonate* (**4a-10**). White liquid; yield, 81.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.90 (t, *J* = 1.9 Hz, 1H, Ar-H), 7.84--7.79 (m, 1H, Ar-H), 7.67 (ddd, *J* = 8.1, 2.0, 1.0 Hz, 1H, Ar-H), 7.53 (t, *J* = 8.0 Hz, 1H, Ar-H), 5.27 (s, 2H, -CH~2~-), 2.71 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.77, 160.41, 136.71, 135.75, 134.65, 130.76, 128.07, 126.18, 60.02, 14.51. HRMS calculated for C~10~H~10~ClO~4~N~2~S~2~ \[M + H\]^+^ 320.97650, found 320.97635.
*(5-(ethylthio)-1,3,4-oxadiazol-2-yl)methyl 4-fluorobenzenesulfonate* (**4a-11**). White liquid; yield, 80.5%; ^1^H NMR (400 MHz, Acetone-*d~6~*) δ 8.04 (dd, *J* = 9.0, 5.0 Hz, 2H, Ar-H), 7.48 (t, *J* = 8.8 Hz, 2H, Ar-H), 5.45 (s, 2H, -CH~2~-), 3.26 (q, *J* = 7.3 Hz, 2H, -CH~2~-), 1.42 (t, *J* = 7.3 Hz, 3H, -CH~3~). ^13^C NMR (100 MHz, Acetone-*d~6~*) δ 166.09 (d, *J* = 254.8 Hz), 166.04, 161.15, 131.73 (d, *J* = 3.4 Hz), 131.29 (d, *J* = 10.2 Hz), 116.92 (d, *J* = 23.3 Hz), 60.58, 26.54, 14.19. ^19^F NMR (376 MHz, Acetone-*d~6~*) δ -104.41. HRMS calculated for C~11~H~12~O~4~N~2~ClS~2~ \[M + H\]^+^ 319.02170, found 319.02142.
*(5-(ethylthio)-1,3,4-oxadiazol-2-yl)methyl 4-chlorobenzenesulfonate* (**4a-12**). White liquid; yield, 78.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.80--7.75 (m, 2H, Ar-H), 7.50--7.45 (m, 2H, Ar-H), 5.19 (s, 2H, -CH~2~-), 3.17 (q, *J* = 7.4 Hz, 2H, -CH~2~-), 1.40 (t, *J* = 7.4 Hz, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.14, 160.31, 141.34, 133.66, 129.82, 129.53, 59.96, 26.97, 14.55. HRMS calculated for C~11~H~12~O~4~N~2~ClS~2~ \[M + H\]^+^ 334.99215, found 334.99191.
*(5-(ethylthio)-1,3,4-oxadiazol-2-yl)methyl 4-bromobenzenesulfonate* (**4a-13**). White liquid; yield, 72.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.69 (d, *J* = 8.8 Hz, 2H, Ar-H), 7.63 (d, *J* = 8.8 Hz, 2H, Ar-H), 5.19 (s, 2H, -CH~2~-), 3.17 (q, *J* = 7.4 Hz, 2H, -CH~2~-), 1.40 (t, *J* = 7.4 Hz, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.13, 160.30, 134.22, 132.80, 129.92, 129.54, 59.98, 26.99, 14.55. HRMS calculated for C~11~H~12~O~4~N~2~BrS~2~ \[M + H\]^+^ 378.94164, found 378.94110.
*(5-((2-fluoroethyl)thio)-1,3,4-oxadiazol-2-yl)methyl 4-fluorobenzenesulfonate* (**4a-14**). White liquid; yield, 62.5%; ^1^H NMR (400 MHz, Acetone-*d~6~*) δ 8.05 (dd, *J* = 9.0, 5.0 Hz, 2H, Ar-H), 7.48 (t, *J* = 8.8 Hz, 2H, Ar-H), 5.45 (s, 2H, -CH~2~-), 4.81 (t, *J* = 5.8 Hz, 1H, -CH-), 4.69 (t, *J* = 5.8 Hz, 1H, -CH-), 3.64 (t, *J* = 5.8 Hz, 1H, -CH-), 3.59 (t, *J* = 5.8 Hz, 1H, -CH-).^13^C NMR (100 MHz, Acetone-*d~6~*) δ 166.12 (d, *J* = 254.9 Hz), 165.41, 161.48, 131.69 (d, *J* = 3.2 Hz), 131.32 (d, *J* = 10.1 Hz), 116.93 (d, *J* = 23.2 Hz), 80.99 (d, *J* = 169.1 Hz), 60.50, 32.06 (d, *J* = 22.0 Hz). ^19^F NMR (376 MHz, Acetone-*d~6~*) δ -104.33, -216.98. HRMS calculated for C~11~H~11~O~4~N~2~F~2~S~2~ \[M + H\]^+^ 337.01228, found 337.01169.
*(5-((2-fluoroethyl)thio)-1,3,4-oxadiazol-2-yl)methyl 4-chlorobenzenesulfonate* (**4a-15**). White liquid; yield, 80.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.81--7.75 (m, 2H, Ar-H), 7.50--7.45 (m, 2H, Ar-H), 5.19 (s, 2H, -CH~2~-), 4.72 (t, *J* = 5.7 Hz, 1H, -CH-), 4.60 (t, *J* = 5.7 Hz, 1H, -CH-), 3.49 (t, *J* = 5.7 Hz, 1H, -CH-), 3.43 (t, *J* = 5.7 Hz, 1H, -CH-). ^13^C NMR (100 MHz, CDCl~3~) δ 166.17, 160.81, 141.37, 133.62, 129.81, 129.51, 80.62 (d, *J* = 172.1 Hz), 59.74, 32.33 (d, *J* = 22.2 Hz). ^19^F NMR (376 MHz, CDCl~3~) δ -215.71. HRMS calculated for C~11~H~11~O~4~N~2~ClFS~2~ \[M + H\]^+^ 352.98273, found 352.98209.
*(5-((2-fluoroethyl)thio)-1,3,4-oxadiazol-2-yl)methyl 4-bromobenzenesulfonate* (**4a-16**). White solid; m.p.: 90--91 °C; yield, 80.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.73--7.68 (m, 2H, Ar-H), 7.68--7.61 (m, 2H, Ar-H), 5.19 (s, 2H, -CH~2~-), 4.72 (t, *J* = 5.7 Hz, 1H, -CH-), 4.61 (t, *J* = 5.7 Hz, 1H, -CH-), 3.49 (t, *J* = 5.7 Hz, 1H, -CH-), 3.44 (t, *J* = 5.7 Hz, 1H, -CH-); ^13^C NMR (100 MHz, CDCl~3~) δ 166.21, 160.80, 134.20, 132.83, 129.98, 129.55, 80.66 (d, *J* = 172.3 Hz), 59.78, 32.37 (d, *J* = 22.2 Hz). ^19^F NMR (376 MHz, CDCl~3~) δ -215.68. HRMS calculated for C~11~H~11~O~4~N~2~BrFS~2~ \[M + H\]^+^ 396.93222, found 396.93161.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl dimethylsulfamate* (**4a-17**). White liquid; yield, 80.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 5.27 (s, 2H, -CH~2~-), 2.92 (s, 6H, -N(CH~3~)~2~), 2.76 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.55, 161.45, 59.55, 38.46, 14.56. HRMS calculated for C~6~H~12~O~4~N~3~S~2~ \[M + H\]^+^ 254.02637, found 254.02617.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 2-(trifluoromethyl)benzenesulfonate* (**4a-18**). White liquid; yield, 64.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 8.20 (d, *J* = 7.5 Hz, 1H, Ar-H), 7.88 (d, *J* = 7.5 Hz, 1H, Ar-H), 7.73 (ddd, *J* = 14.0, 11.1, 6.7 Hz, 2H, Ar-H), 5.26 (s, 2H, -CH~2~-), 2.63 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.77, 160.48, 134.50, 133.84, 132.58, 132.35, 128.84 (d, *J* = 6.1 Hz), 123.47, 120.75, 60.04, 14.48. ^19^F NMR (376 MHz, CDCl~3~) δ -58.49. HRMS calculated for C~11~H~10~O~4~N~2~F~3~S~2~ \[M + H\]^+^ 355.00286, found 355.00241.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl benzoate* (**5a-1**). White solid; m.p.: 30--31 °C; yield, 77.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 8.09 (d, *J* = 7.1 Hz, 2H, Ar-H), 7.62 (t, *J* = 7.4 Hz, 1H, Ar-H), 7.48 (t, *J* = 7.7 Hz, 2H, Ar-H), 5.52 (s, 2H, -CH~2~-), 2.76 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 166.92, 165.49, 162.81, 133.74, 130.00, 128.70, 128.58, 55.53, 14.60. HRMS calculated for C~11~H~11~O~3~N~2~S \[M + H\]^+^ 251.04849, found 251.04831.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-fluorobenzoate* (**5a-2**). White solid; m.p.: 53--54 °C; yield, 75.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 8.02 (dd, *J* = 9.0, 5.4 Hz, 2H, Ar-H), 7.06 (t, *J* = 8.7 Hz, 2H, Ar-H), 5.42 (s, 2H, -CH~2~-), 2.67 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 166.99, 166.22 (d, *J* = 255.3 Hz), 164.52, 162.68, 132.66 (d, *J* = 9.5 Hz), 124.94 (d, *J* = 3.0 Hz), 115.85 (d, *J* = 22.1 Hz), 55.62, 14.61. ^19^F NMR (376 MHz, CDCl~3~) δ -104.05. HRMS calculated for C~11~H~10~O~3~N~2~FS \[M + H\]^+^ 269.03907, found 269.03897.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-chlorobenzoate* (**5a-3**). White solid; m.p.: 53--54 °C; yield, 74.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.95--7.90 (m, 2H, Ar-H), 7.39--7.34 (m, 2H, Ar-H), 5.42 (s, 2H, -CH~2~-), 2.67 (s, 3H, -CH~3~), ^13^C NMR (100 MHz, CDCl~3~) δ 166.98, 164.64, 162.60, 140.32, 131.37, 128.97, 127.15, 55.70, 14.61. HRMS calculated for C~11~H~10~O~3~N~2~ClS \[M + H\]^+^ 285.00952, found 285.00958.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-bromobenzoate* (**5a-4**). White solid; m.p.: 82--83 °C; yield, 70.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 7.86 (d, *J* = 8.7 Hz, 2H, Ar-H), 7.54 (d, *J* = 8.7 Hz, 2H, Ar-H), 5.42 (s, 2H, -CH~2~-), 2.68 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 167.01, 164.81, 162.57, 131.98, 131.48, 129.06, 127.59, 55.70, 14.61. HRMS calculated for C~11~H~10~O~3~N~2~BrS \[M + H\]^+^ 328.95900, found 328.95895.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl 4-methoxybenzoate* (**5a-5**). White solid; m.p.: 35--36 °C; yield, 79.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 8.04 (d, *J* = 9.0 Hz, 2H, Ar-H), 6.95 (d, *J* = 9.0 Hz, 2H, Ar-H), 5.49 (s, 2H, -CH~2~-), 3.89 (s, 3H, -CH~3~), 2.76 (s, 3H, -CH~3~).^13^C NMR (100 MHz, CDCl~3~) δ 166.83, 165.18, 163.97, 163.02, 132.13, 121.01, 113.84, 55.52, 55.28, 14.60. HRMS calculated for C~12~H~13~O~4~N~2~S \[M + H\]^+^ 281.05905, found 281.05884.
*(5-(methylthio)-1,3,4-oxadiazol-2-yl)methyl dimethylcarbamate* (**5a-6**). White liquid; yield, 81.5%; ^1^H NMR (400 MHz, CDCl~3~) δ 5.28 (s, 2H, -CH~2~-), 2.97 (s, 3H, -CH~3~), 2.95 (s, 3H, -CH~3~), 2.75 (s, 3H, -CH~3~). ^13^C NMR (100 MHz, CDCl~3~) δ 166.51, 163.49, 155.10, 56.13, 36.79, 36.05, 14.60. HRMS calculated for C~7~H~12~O~3~N~3~S \[M + H\]^+^ 218.05939, found 218.05922.
3.3. X-ray Diffraction Analysis {#sec3dot3-molecules-25-01488}
-------------------------------
All target compounds had been confirmed by ^1^H NMR, ^13^C NMR and high-resolution mass spectrometry (HRMS). After a preliminary in vitro and in vivo bactericidal analysis, Compound **4a-2** had the best bactericidal activity. The structural composition of compound **4a-2** was determined by single crystal X-ray analysis.
Crystal structure of compound **4a-2** (C~10~H~9~FO~4~N~2~S~2~) is shown in [Figure 5](#molecules-25-01488-f005){ref-type="fig"}. Colorless crystal of compound **4a-2** (0.4 × 0.28 × 0.2 mm) is monoclinic system and space group C 2/C. Cell parameters: a = 26.431(2), b = 5.1560(5), c = 21.7311(18), alpha = 90, beta = 121.147(4), gamma = 90, V = 2534.5(4), Z = 8. Cell dimensions and intensities were measured at 298 K on Bruker SMART diffractometer with MoK\\a radiation (λ = 0.71073 Å). A total of 2215 reflections were measured, of which 1662 were unique in the range of 3.10 \<θ\< 25.02^°^ (h, -18 to 31; k, -6 to 6; l, -25 to 24), The structure was solved by direct method with the SHELXL-2014 program. All of the non-H atoms were refined anisotropically by full-matrix least-squares to give the final R = 0.0409 and WR2 = 0.1075. All hydrogen atoms were computed and refined using a riding model. The completeness of the crystal data is 99.4%. The atomic coordinates for **4a-2** have been deposited at the Cambridge Crystallographic Data Centre. CCDC 1,975,227 contains the [supplementary crystallographic data](#app1-molecules-25-01488){ref-type="app"} for this paper.
3.4. Antibacterial Bioassay by Scanning Electron Microscopy {#sec3dot4-molecules-25-01488}
-----------------------------------------------------------
The sample preparation method was as follows \[[@B30-molecules-25-01488]\]. A certain quantity of microcentrifuge tube (2 ml) was prepared and added bacteria solution of *Xoo* (1.5 mL). Then microcentrifuge tubes were washed with PBS buffer and centrifuged 3 times, in order to discard supernatant, microcentrifuge tubes were centrifuge at 7000 rpm for one minute. Compound **4a-2** was added into the centrifuge tube to prepare 0 µg/mL, 25 µg/mL and 50 µg/mL solution, three parallel groups for each concentration. 2.5% glutaraldehyde fixing solution was added to each microcentrifuge tube for 12 h and removed. Next, microcentrifuge tubes were washed by 30%, 50%, 70%, 90% and absolute ethanol in this order. At last, the samples were flattened and sprayed gold (45s) for observing by SEM.
3.5. Antibacterial Bioassay In Vitro {#sec3dot5-molecules-25-01488}
------------------------------------
The inhibitory efficiency of target compounds on two bacteria in vitro was tested by the turbidimeter method at different concentrations \[[@B10-molecules-25-01488]\]. For initial screening of all 24 compounds, the solution concentration was set at 200 and 100 μg/mL which was incubated with bacterial solution and then procedurally measured for the OD value. A solution with no compound was set as a negative check and bismerthiazol and thiodiazole copper served as the positive control. Compounds that were active at this concentration were further tested at five lower gradient concentrations to get EC~50~. Data were collected in triplicate for each compound concentration. Based on the OD value, the inhibitory effect of the compound on bacteria was calculated. I (%) = (CK-T)/T × 100%. I (%) meant inhibition rate. CK meant the OD value of non-drug control group. T meant the OD value of drug group.
3.6. Antibacterial Activity Bioassay In Vivo {#sec3dot6-molecules-25-01488}
--------------------------------------------
Compounds **4a**-**1**, **4a**-**2** and **4a**-**3** were tested for the protective and curative activity in vivo against rice bacterial leaf blight by leaf-cutting method \[[@B10-molecules-25-01488],[@B30-molecules-25-01488]\] at 200 µg/mL, with comparing to bismerthiazol and thiodiazole copper. A negative control check (CK) was set up identically with absence of the test compound. Data were collected in triplicate treatment. Then the control efficiency could be calculated by analyzing plant disease index. Control efficiency (%) = (C − T)/C × 100%, where C represented the plant disease index of the negative control CK; T represented the disease index of plant with the compound treatment.
4. Conclusions {#sec4-molecules-25-01488}
==============
In summary, 24 novel sulfonate/carboxylate functionalized 1,3,4-oxadiazole derivatives were synthesized and evaluated for antibacterial activity on both bacteria *Xanthomonas oryzae* pv. *oryzae* and *Xanthomonas axonopodis* pv. *citri*. Among them, ten compounds (**4a-1 to 4a-4 and 4a-11 to 4a-16**) showed extensive potency on the *Xoo* in vitro. Four (**4a-1 to 4a-4)** of them also performed well on *Xac* in vitro. In particular, compound **4a-2** with the best antibacterial activity in vitro indicated excellent protective and curative activity against rice bacterial leaf blight in vivo. Furtherly, scanning electron microscope analysis on **4a-2** verified its antibacterial action mechanism. Structure-activity relationship illustrated that sulfonate structure (**4a**), rather than carboxylate moiety(**5a**), play important role for inhibitory effect of target compounds. In conclusion, as expected, 1,3,4-Oxadiazole derivatives containing sulfonate moiety showed promise antibacterial activity and might provide potent plant bactericide.
**Sample Availability:** Samples of the compounds **4a/5a** are available from the authors.
######
Click here for additional data file.
The following are available online. ^1^H, ^13^C and ^19^F NMR spectra of all the compounds are presented as Supporting Information; crystallographic data of compound **4a-2** (CCDC 1975227) for this paper could be obtained free of charge from The Cambridge Crystallographic Data Centre via [www.ccdc.cam.ac.uk](www.ccdc.cam.ac.uk).
L.J. conceived the project; L.W. performed most of the experimental work while H.L. and X.M. implemented the biological test protocols; X.Z. supervised the project and wrote the manuscript. All authors analyzed the data and contributed to manuscript preparation. All authors have read and agreed to the published version of the manuscript.
This work was supported by the National Nature Science Foundation of China NSFC 21967006, and by The Excellent Young Science and Technology Talent Cultivation Plan 201122, and by International Science and Technology Cooperation Program of Guizhou province 2009700112.
The authors declare no conflict of interest.
Figures, Scheme and Tables
==========================
{#molecules-25-01488-f001}
{#molecules-25-01488-sch001}
{#molecules-25-01488-f002}
{#molecules-25-01488-f003}
{#molecules-25-01488-f004}
{#molecules-25-01488-f005}
molecules-25-01488-t001_Table 1
######
Inhibition rate (%) of target compounds against *Xanthomonas oryzae* pv. *oryzae* and *Xanthomonas axonopodis* pv. *citri* ^a^.
Compd. *Xanthomonas Oryzae* Pv. *Oryzae* *Xanthomonas Axonopodis* Pv. *Citri*
-------------------------------- ----------------------------------- -------------------------------------- ------------ ------------
**4a-1** 98.3 ± 1.2 90.4 ± 1.4 94.2 ± 0.2 79.4 ± 1.3
**4a-2** 99.5 ± 1.8 94.1 ± 2.1 97.1 ± 0.3 82.1 ± 2.0
**4a-3** 94.0 ± 1.1 85.4 ± 0.9 93.8 ± 0.4 69.3 ± 2.7
**4a-4** 91.2 ± 2.8 78.7 ± 2.2 80.4 ± 0.2 45.8 ± 4.6
**4a-5** 91.8 ± 2.6 41.5 ± 2.3 62.4 ± 3.6 48.1 ± 3.0
**4a-6** 20.0 ± 1.9 5.0 ± 2.2 20.2 ± 2.7 15.9 ± 2.7
**4a-7** 10.1 ± 1.2 72.7 ± 1.1 22.9 ± 4.1 19.1 ± 4.7
**4a-8** 76.8 ± 1.5 39.8 ± 2.1 92.3 ± 4.6 25.5 ± 2.7
**4a-9** 60.7 ± 1.5 20.2 ± 2.1 18.8 ± 1.5 10.5 ± 2.1
**4a-10** 48.9 ± 2.6 18.5 ± 1.1 17.3 ± 2.3 15.2 ± 1.8
**4a-11** 97.5 ± 1.2 82.2 ± 1.4 47.3 ± 1.8 33.2 ± 1.2
**4a-12** 96.4 ± 2.2 78.4 ± 2.1 45.5 ± 2.1 32.4 ± 1.8
**4a-13** 94.9 ± 1.7 81.2 ± 2.4 41.9 ± 0.8 30.0 ± 2.8
**4a-14** 99.1 ± 1.5 94.0 ± 1.9 41.2 ± 4.8 33.0 ± 3.8
**4a-15** 97.0 ± 0.9 72.5 ± 1.0 38.7 ± 3.6 32.1 ± 1.8
**4a-16** 94.0 ± 1.2 48.0 ± 1.5 36.9 ± 2.8 20.0 ± 1.9
**4a-17** 30.2 ± 1.8 18.1 ± 0.9 25.6 ± 1.5 11.4 ± 1.2
**4a-18** 10.4 ± 2.3 6.2 ± 1.5 48.3 ± 3.7 24.2 ± 5.3
**5a-1** 35.1 ± 1.2 18.2 ± 1.5 47.9 ± 6.0 23.8 ± 4.0
**5a-2** 96.4 ± 2.5 11.5 ± 1.5 55.1 ± 1.5 35.5 ± 3.9
**5a-3** 15.2 ± 5.3 7.3 ± 4.9 68.5 ± 1.4 38.5 ± 1.2
**5a-4** 12.5 ± 2.1 7.0 ± 3.2 32.4 ± 1.7 12.0 ± 2.2
**5a-5** 11.0 ± 1.1 2.3 ± 2.1 45.1 ± 1.8 29.8 ± 2.7
**5a-6** 7.2 ± 2.8 3.5 ± 1.7 32.2 ± 2.5 18.8 ± 1.5
**Bismerthiazol** ^**b**^ 60.2 ± 3.1 28.3 ± 2.8 72.5 ± 2.8 58.2 ± 2.1
**Thiodiazole Copper** ^**b**^ 57.3 ± 1.9 30.2 ± 2.1 79.1 ± 1.8 53.1 ± 1.1
^a^ Average of three replicates; ^b^ The commercial agricultural antibacterial agents bismerthiazol and thiodiazole copper were used as positive control.
molecules-25-01488-t002_Table 2
######
EC~50~ (µM) of some target compounds against *Xanthomonas oryzae* pv. *oryzae* and *Xanthomonas axonopodis* pv. *citri* ^a^.
Compd. *Xanthomonas Oryzae* Pv. *Oryzae* *Xanthomonas Axonopodis* Pv. *Citri*
------------------------ ----------------------------------- --------------------------------------
**4a-1** 63.4 ± 3.8 114.0 ± 6.6
**4a-2** 50.1 ± 4.2 95.8 ± 4.6
**4a-3** 87.2 ± 4.7 132.5 ± 7.5
**4a-4** 99.4 ± 4.7 155.2 ± 5.8
**4a-11** 98.0 ± 6.6 /
**4a-12** 95.3 ± 3.9 /
**4a-13** 86.4 ± 4.8 /
**4a-14** 69.0 ± 4.4 /
**4a-15** 83.4 ± 6.0 /
**4a-16** 112.5 ± 6.0 /
Bismerthiazol ^b^ 253.5 ± 7.6 274.3 ± 8.6
Thiodiazole copper ^b^ 467.4 ± 15.5 406.3 ± 13.0
^a^ Statistical analysis was conducted by ANOVA method at the condition of equal variances assumed (*p* \> 0.05) and equal variances not assumed (*p* \< 0.05); ^b^ Commercial agricultural antibacterial agents bismerthiazol, and thiodiazole copper were used as positive control. ^c^ Corresponding regression equations and r values for this EC~50~ were provided in [supplementary data](#app1-molecules-25-01488){ref-type="app"}.
molecules-25-01488-t003_Table 3
######
Protective effect of compounds **4a-1**, **4a-2** and **4a-3** against *Xanthomonas oryzae* pv. *oryzae*.
Treatment 14 Days after Spraying
----------------------- ------------------------ ------ ------------
**4a-1** 100 34.6 48.1 ± 2.5
**4a-2** 100 16.7 68.6 ± 3.5
**4a-3** 100 22.2 62.3 ± 4.3
Bismerthiazol 100 33.3 49.6 ± 3.1
Thiodiazole copper 100 39.9 42.2 ± 3.0
CK (negative control) 100 87.6 /
^a^ Average of three replicates. Statistical analysis was conducted via the ANOVA method at a condition of equal variances assumed (*p* \> 0.05) and equal variances not assumed (*p* \< 0.05).
molecules-25-01488-t004_Table 4
######
Curative effect of compounds **4a-1**, **4a-2** and **4a-3** against *Xanthomonas oryzae* pv. *oryzae*.
Treatment 14 Days after Spraying
----------------------- ------------------------ ------ ------------
**4a-1** 100 37.8 44.6 ± 2.9
**4a-2** 100 22.2 62.3 ± 3.8
**4a-3** 100 27.8 56.0 ± 3.5
Bismerthiazol 100 39.2 42.9 ± 2.4
Thiodiazole copper 100 45.2 36.1 ± 2.5
CK (negative control) 100 87.6 /
^a^ Average of three replicates. Statistical analysis was conducted via the ANOVA method at a condition of equal variances assumed (*p* \> 0.05) and equal variances not assumed (*p* \< 0.05).
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Because movement promotes energy flow across habitat boundaries [@pone.0045958-Depczynski1], [@pone.0045958-Gaines1], ecological and evolutionary processes are inherently linked to movement, including ecosystem function and biodiversity [@pone.0045958-Abecasis1]. Predicting organismal movement is central to establishing effective management and conservation strategies, such as restoring degraded habitats, reducing exploitation rates, preventing spread of invasive species, and protecting wildlife (i.e. "movement ecology" [@pone.0045958-Nathan1]). A key aspect of movement ecology is interactions among species, especially predators and prey. Dynamics between predators and prey are often complex when considered across relevant spatial and temporal scales [@pone.0045958-Nathan1], [@pone.0045958-Benhamou1]. However, growing evidence reveals that predators can regulate ecosystem structure and function via trophic cascades arising through both consumption and predator-induced modifications in prey behavior [@pone.0045958-Creel1]. Therefore, studies of predator movement patterns are becoming increasingly important for predicting the ecosystem consequences of their declines, especially for marine species that are experiencing significant population declines due to overfishing [@pone.0045958-Myers1]--[@pone.0045958-Ferretti1]. Consequently, further studies of marine predator movements and habitat use are needed to identify and prioritize areas for protection (e.g. feeding and natal grounds) as well as generate sufficient data for modeling how changes in their habitat use can affect sustainability and potentially alter community dynamics [@pone.0045958-Heithaus1]--[@pone.0045958-Cosner1].
{#pone-0045958-g001}
Bull sharks (*Carcharhinus leucas* Müller & Henle, 1839) are apex predators in tropical and subtropical seas [@pone.0045958-Tuma1]--[@pone.0045958-Carlson1]. In the western Atlantic, the species grows to a relatively large size \[\>340 cm, \>230 kg; 34\] and occurs from northeastern United States to Brazil. Within this geographic range, bull sharks are common to coastal, estuarine, lagoon and fresh waters, especially certain large lakes and rivers [@pone.0045958-Curtis1], [@pone.0045958-Martin1]. Bull sharks are unique among elasmobranchs for their ability to inhabit brackish or freshwater systems for relatively prolonged periods due to unique physiological adaptations that permit osmoregulation in low salinity environments [@pone.0045958-Martin1]--[@pone.0045958-Hammerschlag1]. Studies from Florida and the Gulf of Mexico have found that young of the year and juvenile bull sharks regularly occupy inshore rivers as nursery habitats [@pone.0045958-Heupel1]--[@pone.0045958-Simpfendorfer2], but transition out of these areas once they reach about 160--180 cm TL [@pone.0045958-Curtis1]. Gravid females likely return to pup [@pone.0045958-Curtis1]. Only two papers [@pone.0045958-Carlson1], [@pone.0045958-Brunnschweiler1] have reported on movement patterns of large (\>150 cm TL) bull sharks using archival satellite tags, the latter being the first to describe movements of adult bull sharks in the Gulf of Mexico and southeastern United States. These studies found that adult bull sharks exhibit high site fidelity and primarily utilize shallow coastal zones [@pone.0045958-Curtis1], [@pone.0045958-Brunnschweiler1].
Atlantic tarpon (*Megalops atlanticus* Valenciennes, 1847) are highly mobile mesopredators and very popular sportfish [@pone.0045958-Ault1], [@pone.0045958-Mill1]. Satellite tagging of Atlantic tarpon in the southeastern United States, Gulf of Mexico and Florida Keys has revealed that, similar to bull sharks, tarpon also tend to utilize inshore coastal, estuarine and freshwater areas where they co-occur [@pone.0045958-Luo1]--[@pone.0045958-Ault3]. Bull sharks are commonly observed preying upon tarpon at popular fishing locations in the Florida Keys, southern Florida and Gulf of Mexico during recreational catch and release angling [@pone.0045958-Ault2]. Examination of bull shark stomachs from the aforementioned region has shown that in addition to tarpon, sharks feed on mullet *Mugilcephalus*, menhaden *Brevoortiapatronis* and ladyfish *Elopssaurus*, all favored food items of the Atlantic tarpon [@pone.0045958-Ault3]. Given similarities in spatial and trophic niches, tarpon may be susceptible to bull shark predation while foraging.
Here we conducted a joint tagging study of bull sharks and Atlantic tarpon in southern Florida to describe their spatial distribution, habitat use and movement patterns relative to one another. Our first goal was to describe seasonal abundance and general movement patterns of bull sharks and tarpon. Our second goal was to identify core areas of bull shark activity and then examine the movement patterns and swimming behaviors (speed, tortuoisty) of tarpon relative to these core areas of bull shark habitat use. We used these data to address four general questions. First, how do the seasonal abundance patterns of bull sharks and tarpon compare? Second, how do the movement patterns of bull sharks and tarpon compare, and what proportion of time do their primary ranges overlap? Third, do tarpon movement patterns (e.g., straight versus convoluted paths) and/or their rate of movement (ROM) differ in areas of low versus high bull shark abundance? Finally, given the potential for predator-prey interactions, can any general conclusions be reached concerning whether tarpon may mitigate risk of predation by sharks when they are in areas of high bull shark abundance?
10.1371/journal.pone.0045958.t001
###### Summary data for 16 bull sharks tracked with SPOT5 tags. TL = Total Length.
{#pone-0045958-t001-1}
Shark ID TL (cm) Sex Tagging Location Date Tagged Last Detection Days at Large
---------- --------- ----- ------------------ ------------- ---------------- --------------- -----
33919 176 F 25.00644 −80.99969 11/7/2010 5/21/2011 194
33937 221 F 24.69740 −80.85227 6/5/2010 10/8/2010 123
33991 210 F 25.00644 −80.99969 3/26/2010 6/20/2010 84
34208 173 M 25.00644 −80.99969 11/7/2010 1/14/2011 67
55492 189 F 26.36898 −81.97914 10/29/2010 1/17/2011 78
55493 216 M 24.69740 −80.85227 8/19/2010 10/21/2010 62
55496 200 M 25.00644 −80.99969 11/6/2010 2/13/2011 97
60695 234 M 25.00644 −80.99969 8/20/2010 9/7/2010 17
60696 170 M 26.36898 −81.97914 10/29/2010 6/11/2011 222
60697 154 M 26.36898 −81.97914 8/10/2010 4/6/2011 236
60698 176 F 25.00644 −80.99969 11/6/2010 1/25/2011 79
60699 195 F 25.00644 −80.99969 10/7/2010 6/11/2011 244
68479 160 M 25.00644 −80.99969 1/28/2011 6/10/2011 132
68483 194 F 25.00644 −80.99969 12/4/2010 2/12/2011 68
105596 245 F 24.81300 −80.90960 2/27/2011 5/9/2011 72
68467 245 M 26.86000 −79.03833 2/19/2011 5/20/2011 91
10.1371/journal.pone.0045958.t002
###### Total positions received and corresponding accuracy for each tagged animal.
{#pone-0045958-t002-2}
Location Class
--------------- ---------------- -- ---- -- ---- -- ---- -- ---- -- ----- -- --- -- -----
**Shark ID**
33919 7 8 16 30 13 62 1 137
33937 0 1 0 1 1 28 0 31
33991 3 6 4 4 5 45 1 68
34208 0 1 1 2 1 13 0 18
55492 0 0 0 0 0 0 0 0
55493 0 0 0 0 1 0 1 2
55496 2 3 5 14 6 40 0 70
60695 0 0 0 0 2 1 0 3
60696 1 2 2 5 1 36 2 49
60697 2 0 0 2 4 24 0 32
60698 5 6 11 19 12 71 0 124
60699 13 12 18 25 10 100 0 178
68479 1 0 1 0 1 4 0 7
68483 1 1 2 2 10 60 3 79
105596 7 1 1 2 0 10 0 21
68467 0 2 1 3 1 7 0 14
**Tarpon ID**
T-176 93 13 6 0 5 42 0 159
T-177 12 7 1 0 4 21 0 45
T-178 13 11 10 1 8 27 0 70
T-179 114 48 11 3 55 203 0 434
T-180 20 5 1 0 1 8 0 35
T-181 1 0 0 0 2 81 0 84
T-182 10 6 3 0 3 6 0 28
T-184 104 54 43 6 22 238 0 467
T-186 17 10 6 2 18 124 0 177
T-187 16 5 1 6 7 96 0 131
T-188 7 7 2 0 5 17 0 38
T-196 37 17 12 1 15 217 0 299
Accuracies are indicated by a location class (LC), ranging in accuracy with the following radius of error: LC 3\<250 m, 250 m \< LC 2\<500 m, 500 m \< LC 1\<1500 m; median error for LC 0, A and B ranges from 1 to 3 km. Positions with LC Z were excluded from data analyses (See text).
Methods {#s2}
=======
Bull Sharks {#s2a}
-----------
Between October 2009 and May 2012, standardized surveys were conducted to capture and tag sharks as part of an ongoing shark abundance and movement study in the Florida Keys (Biscayne Bay, Key Largo, Islamorada, Dry Tortugas) and southeastern Gulf of Mexico (Florida Bay, Everglades National Park, Fort Myers). Sharks were captured using baited circle-hook drumlines as described by Hammerschlag et al [@pone.0045958-Hammerschlag2]. Briefly, sets of 5 drumlines were deployed and left to soak for 1.0 hour before being checked for shark presence. Upon capture, shark sex was recorded, total length (TL) in cm was measured and thereafter, sharks were marked with an identification tag and then released back into the water. Catch per unit effort (CPUE) of drumlines (for all years combined, averaged by season) was used to determine if there were seasonal changes in occurrence and size (TL). CPUE was expressed as the number of bull sharks caught per set and average size of bull sharks caught per set within each season (Winter: Dec, Jan, Feb; Spring; Mar, Apr, May; Summer; Jun, Jul, Aug; Fall: Sep, Oct, Nov).
If a large bull shark (\>150 cm TL) was captured during a survey, a satellite telemetry tag was affixed to the sharks' first dorsal fin. We used Smart Position and Temperature Transmitting (SPOT) tags (SPOT5, Wildlife Computers; [www.wildlifecomputers.com](http://www.wildlifecomputers.com)) because they provided relatively detailed horizontal movements that could be analyzed at a much higher resolution than light-based position data derived from pop-up archival satellite tags [@pone.0045958-Sims1]. SPOT tags were coated with Propspeed, a non-toxic, nonmetallic anti-fouling agent, to minimize biofouling [@pone.0045958-Hammerschlag3], [@pone.0045958-Hammerschlag4]. Transmitters were attached using titanium bolts, neoprene and steel washers, and high carbon steel nuts to prevent any metallic corrosion from contacting the fin as well as to ensure that the steel nuts corroded, resulting in eventual tag detachment [@pone.0045958-Hanson1].
{#pone-0045958-g002}
{#pone-0045958-g003}
Tarpon {#s2b}
------
Data on seasonal abundance patterns for tarpon were obtained from two sources: (1) the Cooperative Tagging Center (CTC) based at NOAA's National Marine Fisheries Service, Southeast Fisheries Science Center, Miami, Florida [@pone.0045958-Mather1], [@pone.0045958-Scott1]; and, (2) creel surveys of professional fishing guides in Everglades National Park acquired from the National Park Service, Homestead, Florida [@pone.0045958-CassCalay1]. Release locations of conventionally tagged tarpon from 1962 to 2004 derived from the CTC database were plotted with ArcGIS by season (Winter: Dec, Jan, Feb; Spring; Mar, Apr, May; Summer; Jun, Jul, Aug; Fall: Sep, Oct, Nov) and by size (weight in kg). In addition, numbers of tarpon caught by recreational fishers in ENP by month, averaged for the period 1980--2006, were extracted from the creel survey database.
Between March 2011 and June 2011, tarpon were captured for satellite tagging using standard hook-and-line gears on chartered recreational fishing boats in the southern Florida Keys (Islamorada, Bahia Honda), Biscayne National Park (Broad Key), Everglades National Park (Whitewater Bay), Boca Grande and southeastern Gulf of Mexico. Upon capture, tarpon fork length (FL) and girth (G) were measured in cm and weight in kg was computed with the algorithm of [@pone.0045958-Ault4]; thereafter, a SPOT tag was attached to the tarpon's body via a 40 cm long stainless steel wire tether to a titanium anchor dart. The anchor dart was inserted into the flank of the tarpon about 15--20 cm anterior to the dorsal fin and roughly 5--10 cm above the lateral line.
{#pone-0045958-g004}
Movement Data Analysis {#s2c}
----------------------
The geographic location of satellite-tagged sharks and tarpon were determined by Doppler-shift calculations made by the Argos Data Collection and Location Service, [www.argos-system.org](http://www.argos-system.org)) whenever a passing satellite received signals from the tag at the surface. To improve location accuracy, we processed all Doppler derived data using Kalman filtering (KF). Argos provides the following radius of error for each KF-derived location class (LC): LC 3\<250 m, 250 m \< LC 2\<500 m, 500 m \< LC 1\<1500 m; Argos states that the median error for LC 0, A and B ranges from 1 to 3 km [@pone.0045958-Bernard1]. Class Z indicates that the location process failed and estimates of position are highly inaccurate. All transmitted locations were filtered to remove positions with LC Z, those on land, and those exceeding a speed of 2 m/s (following Weng et al [@pone.0045958-Weng1]). Argos-derived locations were plotted using ESRI ArcGIS 9.3.
We performed utilization distribution analyses on position data using fixed kernel density metrics. Kernel density estimates quantify the core regions of occupancy within an animal's home range or activity space [@pone.0045958-Heupel1], [@pone.0045958-Domeier1]. Kernel density values are cumulated from the highest to lowest density areas to create kernel density contours. Thus, the 25% contours represent areas of the top highest observed densities, while the 95% contours represent up to 95% density areas. These metrics were calculated according to the equations provided by Worton [@pone.0045958-Worton1] and plotted using Interactive Data Languages (IDL, [www.ittvis.com](http://www.ittvis.com)) software. Following Domeier and Nasby-Lucas [@pone.0045958-Domeier1] and Weng et al [@pone.0045958-Weng2], kernel density estimates were calculated for all sharks grouped as species-specific habitat utilization instead of the individual's home range.
{#pone-0045958-g005}
![Average total numbers of tarpon caught by month by recreational anglers in Everglades National Park from 1980 to 2006.\
Data based on National Park Service Creel Surveys of professional fishing guides in Everglades National Park [@pone.0045958-CassCalay1].](pone.0045958.g006){#pone-0045958-g006}
Kernel estimates cannot be conducted on SPOT-derived raw data because of the irregular sampling intervals at which data are acquired, gaps in data, and autocorrelations due to successive locations [@pone.0045958-Katajisto1]. To account for these biases, filtered tracks were regularized to a frequency of 12 hour intervals (midnight and noon), using piecewise Bézier interpolation methods similar to Tremblay et al [@pone.0045958-Tremblay1], but modified with the algorithm by Lars Jenson (<http://ljensen.com/bezier/>). We employed the modified algorithm to eliminate unnatural loops in the tracks that occur with Bezier method used in [@pone.0045958-Tremblay1]. Interpolating track sections with large temporal gaps increases uncertainty (reduces confidence) in data. To explicitly deal with this, we did not interpolate gaps in the data that exceeded three days following the methods of [@pone.0045958-Weng2].
To describe potential interactions between sharks and tarpon, ROM and tortuosity of tarpon movements were compared relative to bull shark core areas of occupancy (i.e., shark kernel densities) by applying generalized linear models [@pone.0045958-Fox1]. ROM was calculated as the linear distance traveled in 12 hours. We used the VFractal d [@pone.0045958-Nams1] as a metric of movement tortuosity. VFractal d values were calculated as a function the turning angle for each pair of consecutive movements described in detail by Nams [@pone.0045958-Nams1]. ROM and VFractal d were calculated based on the filtered interpolated positions. VFractal d is different, but similar to, fractal d (each point versus each track) which can be estimated by calculating the mean of VFractal d values of all location points for each tarpon movement track [@pone.0045958-Nams1]. In this study, we used only the VFractal d values, not the fractal d.
{#pone-0045958-g007}
{#pone-0045958-g008}
{#pone-0045958-g009}
10.1371/journal.pone.0045958.t003
###### Summary data for 10 Atlantic tarpon tracked with SPOT5 tags.
{#pone-0045958-t003-3}
Tarpon ID FL (cm) G (cm) W (kg) Latitude Longitude Tag Date Days at Large
----------- --------- -------- -------- ---------- ----------- ------------ ---------------
T-176 199 99 78.8 24.8469 −80.7508 9/4/2011 20
T-177 155 92 46.5 24.8411 −80.7503 2/4/2011 6
T-178 175 94 59.0 25.2908 −81.0161 03/19/2011 8
T-179 186 100 70.9 25.3524 −80.2595 04/27/2011 60
T-180 172 75 43.4 25.3524 −80.2595 04/28/2011 12
T-181 180 90 58.8 26.7500 −82.1580 06/15/2011 39
T-182 150 78 35.8 26.6322 −82.2366 05/23/2011 5
T-184 172 90 54.3 24.6576 −81.2874 05/17/2011 62
T-186 150 76 34.7 26.6310 −82.2369 05/23/2011 30
T-188 155 78 37.9 25.7580 −80.1293 8/6/2011 6
FL = Fork Length; G = Girth.
Results {#s3}
=======
During shark surveys, we deployed 1,382 standardized sets, in which 3,699 individual drumlines were deployed. During these sets, 815 sharks were caught, of which 56 were bull sharks, ranging in size from 142--269 cm TL (average 200 cm TL). Bull shark catch rates and average size by season are plotted in [Figure 1](#pone-0045958-g001){ref-type="fig"}. Bull sharks were caught year round, but catch rates were highest in December and January; although the largest sharks (\>230 cm TL) were caught between April and July.
Between March 2010 and February 2011, 18 bull sharks were SPOT tagged off southern Florida. Of these, a total of 16 (8 male and 8 female) transmitted valid geolocations that permitted movement tracks to be evaluated. Sharks ranged in size from 154--245 cm TL (average 197.4 cm TL, [Table 1](#pone-0045958-t001){ref-type="table"}). Accuracies of spatial locations ranged from \<250 m to \<3 km ([Table 2](#pone-0045958-t002){ref-type="table"}). No sharks within the dataset moved into inshore rivers.
Bull sharks exhibited high site fidelity, primarily restricting movements to shallow inshore areas where they were tagged ([Fig. 2](#pone-0045958-g002){ref-type="fig"}, [3](#pone-0045958-g003){ref-type="fig"}). Only one shark (194 cm female, \# 68483, [Table 1](#pone-0045958-t001){ref-type="table"}) made a relatively long-distance migration. Initially tagged in Everglades National Park (17 miles west of Islamorada), this shark traveled northwest into the Gulf of Mexico over the course of 10 days and after approximately one month, it returned to the Florida Keys. Over the next month, the shark moved northward along the Florida Keys crossing the Straits of Florida to the Bahamas, swimming to the vicinity of Bimini. The shark then traveled southeast, again crossing the Straits of Florida before entering Biscayne Bay when transmissions ceased a month later. The minimum straight line distance of this 68 day trip was approximately 1,200 km.
The fixed kernel results for tagged bull sharks, displayed as volume contours, showed that a core area of 670 km^2^ (25% kernel contour) centered at the northwestern region of Florida Bay ([Fig. 4](#pone-0045958-g004){ref-type="fig"}). The 50% kernel contour (2,260 km^2^) indicates the areas of moderate use extended out to most of Florida Bay, Florida Keys and the Biscayne Bay ([Fig. 4](#pone-0045958-g004){ref-type="fig"}). The 95% kernel contour (18,042 km^2^) shows the areas of 95% habitat utilization by our tagged bull sharks. We consider areas where bull shark kernel densities exceeded 50% to be "high density" zones, whereas areas where kernel densities were less than 50% were "low-density" zones.
Tarpon were captured year-round by recreational anglers in southern Florida waters; however, strong seasonal differences in catch rates and sizes of animals caught were found ([Fig. 5](#pone-0045958-g005){ref-type="fig"}, [6](#pone-0045958-g006){ref-type="fig"}). Large mature fish (\>45.4 kg) appear to be virtually absent from the region in winter (early December-late March). The bulk of the migratory front arrives in late spring (mid- to late-April) and departs the area (going northward) by early-summer (late June) ([Fig. 5](#pone-0045958-g005){ref-type="fig"}). There is a secondary surge of catch rates in fall as tarpon travel southward through the area during the October to mid-November period ([Fig. 5](#pone-0045958-g005){ref-type="fig"}). Other tarpon caught during the year are largely immature fish that tend to use the local rivers and estuaries. Creel data derived from surveys of anglers fishing in Everglades National Park showed the same bi-modal pattern in catches and catch rates ([Fig. 6](#pone-0045958-g006){ref-type="fig"}). Tarpon catches were lowest from November through February; highest from April to a peak in June, declining from July through September, and then with a secondary peak again in October (although to a much lesser extent than in early summer).
Tarpon that were satellite-tagged ranged in size from 150--199 cm FL (average 169.4 cm FL, [Table 3](#pone-0045958-t003){ref-type="table"}). Accuracies of spatial locations were similar to those for sharks ([Table 2](#pone-0045958-t002){ref-type="table"}). Of the 10 tarpon tracks, three were on the east coast of Florida, two in the Florida Keys, and five along the west coast of Florida ([Fig. 4](#pone-0045958-g004){ref-type="fig"}, [7](#pone-0045958-g007){ref-type="fig"}). The first three tracks (T-176, T-177, T-178) were relatively short due to apparent tag failures. The other three short tracks (T-180, T-182, T-188) were most likely a result of shark attack ([Fig. 4](#pone-0045958-g004){ref-type="fig"}, see [discussion](#s4){ref-type="sec"} for more details). Relatively few tarpon tracks, in relation to bull shark tracks, were distributed over open or deep waters ([Fig. 4](#pone-0045958-g004){ref-type="fig"}, [7](#pone-0045958-g007){ref-type="fig"}). In contrast, tracks of tarpon, relative to bull sharks, were clustered around shallow Keys and passes. Moreover, tarpon tracks were also concentrated up rivers, where tracked bull sharks were absent ([Fig. 4](#pone-0045958-g004){ref-type="fig"}, [7](#pone-0045958-g007){ref-type="fig"}). Tarpon ROM were highest (\>1 m/s) where bull shark kernel densities were highest (\<50% kernel contour) and ROM were slowest (\<0.5 m/s) where shark kernel densities were lowest (\>50% kernel contour, [Fig. 4](#pone-0045958-g004){ref-type="fig"}). Tagged tarpon spent most of their time (\>90%) swimming at relatively low ROM (\<0.5 m/s, [Fig. 4](#pone-0045958-g004){ref-type="fig"}, [8](#pone-0045958-g008){ref-type="fig"}), coinciding with areas where shark kernel densities were lowest (kernel contours \>50%, i.e., "low-density" zones). In contrast, tarpon spent little time (\<4%) swimming at high ROM (\>1 m/s), coinciding with areas where shark kernel densities were high (kernel contours \<50%, i.e., "high-density" zones, [Fig. 4](#pone-0045958-g004){ref-type="fig"}, [8](#pone-0045958-g008){ref-type="fig"}). This is statistically supported by the positive correlation from the regression model of tarpon ROM dependent on bull shark kernel density ([Fig. 9](#pone-0045958-g009){ref-type="fig"} a). To inspect the data at different levels, two statistical analyses were conducted: (1) with all tarpon data (black and red dots in [Fig. 9](#pone-0045958-g009){ref-type="fig"} b) overlain on bull shark distribution; and, (2) with bottom 25% of ROM values (red dots) for each 0.1 bin of bull shark kernel density. In these analyses, bull shark kernel densities were rescaled from 0 to 1.0 for low to high density and ROMs were transformed by log10. In both analyses, correlations from the regressions were statistically significant: for all data correlation coefficient (r) was 0.1035 (P\<0.005, intercept (b~0~) = −1.239, slope (b~1~) = 0.6869), and for the bottom 25% data r = 0.5806 (P\<0.0001, b~0~ = −2.1233, b~1~ = 2.1499).
The tortuosities along tarpon movement tracks were negatively correlated with bull shark kernel density ([Fig. 9](#pone-0045958-g009){ref-type="fig"} b). Similar to the ROM analysis, two additional analyses were conducted for VFractal d data: one with all tarpon data (black and red dots in [Fig. 9](#pone-0045958-g009){ref-type="fig"} b) overlapped with bull shark distribution range; the other with top 25% of VFractal d values (red dots) for each 0.1 bin of bull shark kernel density. In both analyses, these correlations were statistically significant: for all data r = −0.093 (P\<0.005, b~0~ = 1.3021, b~1~ = −0.2917); and, for the top 25% data r = −0.5887 (P\<0.0001, b~0~ = 1.7762, b~1~ = −0.8009). These results indicated that tarpon generally used low tortuous (i.e., straight-line) movement patterns in shark high-density zones, and used high tortuous movement patterns in shark low density zones.
{#pone-0045958-g010}
Discussion {#s4}
==========
Our study had several key findings. First, bull sharks were present in the ecosystem year round; but, abundance was generally higher in the winter. In contrast, tarpon catches were highest in early summer with a secondary peak in the early fall. However, presence of the largest bull sharks (\>230 cm) coincided with peak tarpon abundance. Second, bull sharks and tarpon generally occupied different aquatic habitats despite similar trophic niches. Bull sharks preferred shallow marine habitats close to the coast of Florida; while tarpon preferred estuarine and riverine regions, with only occasional forays into deeper marine waters where bull shark abundance was greatest. Third, the locomotor behavior and ROMs of tarpon also differed notably between inland riverine habitats and the more open coastal marine habitats. Specifically, tarpon approximately doubled their average ROM in marine coastal regions where bull sharks appeared to concentrate. Finally, tarpon also had straighter and more direct paths in areas of high bull shark patch use and more convoluted paths in areas of low bull shark use. We propose several hypotheses relating to optimal foraging strategies of both tarpon and bull sharks to explain these observed patterns.
At a regional scale, tarpon migration is likely driven principally by water temperatures and prey abundance [@pone.0045958-Luo1], [@pone.0045958-Ault2]. Tarpon migrate characteristically with the 26°C isotherm, for example, which passes northward through southern Florida waters during the period of mid-April to late May each year. The timing of large mature tarpon movement into Florida Bay and the Florida Keys is coincident with the spawning event (i.e., specifically the process of building the gonad just before spawning, and ensuring survivorship of the fertilized eggs to larvae via biophysical factors) and feeding (building the soma for survivorship, and preparing the long northward migrations ahead) [@pone.0045958-Luo1], [@pone.0045958-Ault2].
The core area of bull shark activity found within northwestern area of Florida Bay is likely driven by the high abundance of teleost prey concentrated there. By conducting shark and fish surveys throughout Florida Bay, Torres et al [@pone.0045958-Torres1] found that the abundance of seven species of sharks (including bulls) in the northwestern area of Florida Bay was highly correlated with the abundance of 45 teleost species. Given tarpon feeding habits, we would have similarly expected tarpon habitat use to have also been relatively high within the northwestern area of Florida Bay. However, tarpon movements were suggestive of avoiding this area (low residence and high rate of movement in directed lines). In contrast, tarpon exhibited highly tortuous movements over relatively long time periods along the outskirts of Florida Bay as well as in adjacent rivers, which is indicative of foraging, although prey abundance patterns are relatively low in these areas compared to the northwestern area of the Bay.
Productive habitats that contain the greatest food resources are often inherently dangerous for prey, thus creating the need for prey to modify their locomotor behavior and habitat use in response to the threat of predation [@pone.0045958-Brown1]--[@pone.0045958-Brown2].The observed movements by tarpon in Florida Bay are suggestive of a food-risk trade-off. For example, studies with lizards and rodents [@pone.0045958-Kenagy1] have each shown that they tend to use a bimodal distribution of locomotor speeds, with slower speeds in more protected, safer, habitats and faster speeds in more open, risky, habitats. Desert lizards (*Uma scoparia*) move slowly along convoluted paths underneath vegetation when undisturbed, which likely shields them from both overheating and from predators, but they then move rapidly in direct lines in open areas [@pone.0045958-Jayne1]. Given that prey can elude predators by escaping into a refuge, moving through exposed habitats results in dramatically increased locomotor effort [@pone.0045958-Lima1]--[@pone.0045958-Peckarsky1]. This pattern is consistent with the alterations in speed by tarpon in areas of high and low bull shark density observed in this study.
We suggest that another trade-off may be associated with the additional metabolic costs incurred by tarpon that occupy brackish or freshwater zones where bull shark density is low. Generally, the energetic costs of osmoregulation in teleost fish are higher in freshwater than seawater (e.g., Febry and Lutz [@pone.0045958-Febry1]). The energetic expense occurs because of the need to maintain fluid volume balance by excreting the extra water, while at the same time, trying to conserve internal ionic balance, a biological process which is energetically expensive ([@pone.0045958-Febry1]; G. Anderson, personal communication). The fact that tarpon spend relatively little time in what would appear to be more optimal coastal marine habitats (from both a food and osmotic perspective), and move so quickly through them, further suggests that these habitats may be risky for them.
It is worth noting that our own anecdotal observations indicate threat of predation mortality to tarpon in areas of high bull shark use. For example, Tarpon T-182 was tagged and released on May 23, 2011, in an area of low bull shark density. The tarpon moved southward through Florida Bay and into a bull shark high density area, at which point it was likely attacked and consumed by a shark on May 28th ([Fig. 4](#pone-0045958-g004){ref-type="fig"}, [10](#pone-0045958-g010){ref-type="fig"}). This presumption is based on two factors. First, the depth and light-level data derived from the recovered tag is indicative of being ingested ([Fig. 10](#pone-0045958-g010){ref-type="fig"} a,b). Additionally the recovered tag displayed scratch marks that appear to have been inflicted by a shark based on tooth spacing and serration ([Fig. 10](#pone-0045958-g010){ref-type="fig"} c,d). Although we cannot identify the species of shark by the bite marks on the tag, we believe only tiger (*Galeocerdo cuvier)*, hammerhead (*Sphyrna sp.*) and bull sharks are likely candidates for attacking a large tarpon (and severing the tag's stainless steel tether). However, it seems plausible that a bull shark was responsible given that the former two species are relatively rare in the region, whereas the attack site represents the location of highest bull shark density in the area.
Critical examination of bull shark diet from the region is limited [@pone.0045958-Ault3] and although tarpon have been found in bull shark stomach contents, there exists little evidence of bull sharks routinely targeting tarpon as preferred prey. In contrast, bull sharks are commonly observed preying upon tarpon in the region during recreational catch and release angling [@pone.0045958-Ault2]. Therefore, we hypothesize that a behaviorally mediated indirect interaction (BMII; reviewed by [@pone.0045958-Dill1]) may be occurring between sharks and tarpon. Specifically, we speculate that higher shark abundance in the northwestern area of Florida Bay is largely driven by relatively high teleost abundance (preferred prey) there [@pone.0045958-Torres1], which in turn, indirectly causes tarpon to reduce their use of this productive area when foraging to minimize their risk of potential mortality by sharks. A similar BMII has been described in Shark Bay, Western Australia, among tiger sharks, duogongs (*Dugong dugon*), dolphins (*Tursiops aduncus*), turtles (*Chelonia mydas*) and cormorants (*Phalacrocoraxv arius*) [@pone.0045958-Heithaus2]. Here, seasonal presence of dugongs (preferred prey of tiger sharks) in shallow waters during summer results in peak tiger shark abundance in these habitats. This, in turn, causes dolphin, turtles and cormorants (species not routinely attacked by sharks) to reduce their use of these productive habitats during summer to minimize risk of potential predation [@pone.0045958-Heithaus2]. That said, our hypotheses outlined above require significant investigation by increasing tracking efforts and gathering further ecological data for sharks, tarpon and their potential prey. For example, greater confidence in our hypotheses would be achieved if changes in the spatial and/or temporal movements of sharks corresponded with compensatory adjustments in tarpon swimming behavior and distribution in areas previously occupied by sharks [@pone.0045958-Heithaus3]. Because movement patterns in animals are complex and can be influenced by many different variables, our study cannot directly reveal whether the movements of tarpon or bull sharks influence one another *per se*. Tarpon seasonal migrations are likely cued to the changes in water temperature in combination with the movement and distribution of prey [@pone.0045958-Luo1]. Therefore, the observed tarpon swimming behavior could also be driven by other factors or the combination of them such as environmental preferences (temperature and salinity), feeding needs, and reproductive behaviors [@pone.0045958-Luo1].
Although use of SPOT tags provided spatial data at higher resolution than archival tags, the major limitation of using Argos-derived data from SPOT tags is the need for animals to surface for long enough to allow successive transmissions for obtaining accurate positions and, therefore, estimating fine scale measurements of speed and fractal values. This is problematic because sharks and tarpon surface irregularly and thus can generate gaps in data acquisition and autocorrelation due to consecutive positions [@pone.0045958-Katajisto1]. To overcome this issue, we used filtered tracks that were regularized to a frequency of 12 hour intervals using interpolation. Ideally, it would be better to use higher resolution temporal data (i.e. \<12 hrs) if sharks and tarpon transmitted frequently; however, we found that a 12 hr intervals was optimal in this study based on the frequency of transmissions received. Further, given the limitations in estimating tarpon versus bull shark density, results were strongly influenced by several high shark density values; however, these data were not outliers, but the analysis (and its interpretation) would benefit from a larger data set. We are aware that it would have been ideal to analyze potential overlap in kernel densities between tarpon and sharks. However, since tarpon were concentrated up inland rivers, kernel density estimates calculated would have indicated primary activity space over land, therefore negating such a comparison. Additionally, kernel density estimates for bull sharks could have been biased to the site of tagging, and although this cannot be ruled out, we believe it is unlikely since sharks were tagged throughout the middle keys on both the Atlantic and Gulf coasts (where they also transmitted). Further, restricting focus on data derived from Florida Bay, where shallow water depths likely favored transmission, would not impact the general conclusions drawn from this work. Another potential shortcoming of this study worthy of consideration is that tracking period and duration for tarpon was shorter than for sharks, making our discussion on predator-prey interactions somewhat speculative. Also, positional data used varied in accuracy from less than 250 m up to 3 km. However, we believe that this error scale, when compared to the scale of shark and tarpon movements, was sufficient to describe the spatial habitat use patterns observed.
Investigating the movements and fine scale foraging behaviors of marine predators presents several formidable biological and logistical challenges. Future investigations of this kind in marine systems will benefit from employing multiple types of animal-borne instrumentation and sensors (e.g. video, accelerometers, satellite and acoustic telemetry, etc.) to better understand and quantify dynamic interactions among marine predators and between highly mobile fishes and their prey [@pone.0045958-Papastamatiou1]. Given their relatively high site fidelity in shallow near shore waters, both bull sharks and tarpon may be disproportionately vulnerable to coastal fishing and other anthropogenic impacts including reduced water quality, pollution, reductions in their prey, and habitat modifications. Accordingly, increasing studies of these and other marine predator movement patterns are needed to identify and prioritize areas for protection as well as for predicting how anthropogenic-driven changes in their habitat use may impact ecosystem dynamics and vice versa.
We thank Michael Heithaus, Tobey Curtis, Martin Grosell and Gary Anderson for useful insights that helped with data interpretation. Technical and logistical support was provided by Capt. Curt Slonim, Dominique Lazarre, Piper Wallingford, Julia Wester, Kyra Hartog, Capt. Bruce Ungar, Capt. Rich Smith and Capt. Rick Murphy. We thank Adam Stow and the anonymous reviewers whose comments helped significantly strengthen this manuscript. This work was conducted under permits from the National Marine Fisheries Service Highly Migratory Species Division (SHK-EFP-10-01), Florida Keys National Marine Sanctuary (FKNMS-2010-006), Florida Fish and Wildlife (SAL-957), Everglades National Park (EVER-2011-SCI-0012), and the University of Miami Institutional Animal Care and Use Committee (Protocol \# 09--187).
[^1]: **Competing Interests:**The authors have received funding from the following commercial funder: Wells Fargo. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.
[^2]: Conceived and designed the experiments: NH JL JA. Performed the experiments: NH JL JA. Analyzed the data: NH JL JA. Contributed reagents/materials/analysis tools: NH JL JA. Wrote the paper: NH JL JA DI.
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Benign prostatic hyperplasia (BPH) often causes bladder-outlet obstruction (BOO) and commonly results in lower urinary tract symptoms (LUTS). LUTS diminish quality of life by interfering with daily activities and decreasing psychological well-being. In men, LUTS are typically treated initially with agents that target the prostate, such as an alpha-blocker. However, alpha-blockers may have limited efficacy in relieving overactive bladder (OAB) symptoms.[@b1-cia-9-1021]
Storage symptoms often independently occur and persist in many men after pharmacologic treatment for BOO.[@b2-cia-9-1021] Therefore, treatments that target the prostate often fail to alleviate OAB symptoms and may not be the most appropriate therapy for men with storage LUTS.[@b3-cia-9-1021] Treatment with an alpha-blocker and an anticholinergic agent improves LUTS compared with an alpha-blocker alone.[@b4-cia-9-1021]--[@b6-cia-9-1021] LUTS are one of the major causes of nocturia in older men and are associated with a decreased bladder capacity due to detrusor overactivity or high post-voiding residual volume (PVR), which results in decreased voiding volume and increased micturition frequency.[@b7-cia-9-1021] Because BOO is a potential risk factor for nocturia, alpha-blockers are considered a potential treatment for nocturia. However, the relief of BOO is not sufficient to correct nocturia.[@b8-cia-9-1021] While the pathophysiology of nocturia is considered multifactorial, nocturnal polyuria (NP) and decreased nocturnal bladder capacity (NBC) are the main mechanisms.[@b9-cia-9-1021]
The efficacy and safety of desmopressin in the treatment of adults with nocturia have been demonstrated in randomized trials.[@b10-cia-9-1021]--[@b12-cia-9-1021] However, the effects of an anticholinergic agent or an antidiuretic agent as an add-on therapy to an alpha-blocker in patients with LUTS remain unknown. In the study reported here, we analyzed the efficacy of anticholinergic agent and antidiuretic agent add-on therapy for refractory nocturia in men previously treated with an alpha-blocker for LUTS, according to voiding disorders (nocturnal polyuria, decreased nocturnal bladder capacity \[NBC\], or nocturia by both causes subgroups).
Patients and methods
====================
Patient selection and data collection
-------------------------------------
We obtained approval for the study from the institutional review board at our hospital. Between July 2010 and April 2013, men ≥50 years of age diagnosed with LUTS due to BOO, with a maximum urinary flow rate (Q~max~) ≤15 mL/second, nocturia (≥1 void/night), and a total International Prostate Symptom Score (IPSS) ≥14 (voiding sub-score ≥8 and storage sub-score ≥6) were treated with an alpha-blocker for at least 4 weeks.[@b13-cia-9-1021]
Four hundred and thirty five patients with persistent nocturia were prospectively/retrospectively enrolled after providing written informed consent. All patients had a 4-week drug washout period before enrollment. Patients were excluded if they had neurogenic bladder dysfunction, hyponatremia, uncontrolled hypertension, congestive heart failure, history of prostate surgery, interstitial cystitis, elevated prostate specific antigen, or had been previously treated with anticholinergic drugs or diuretics. Patients were randomly divided into two groups at the time of enrollment. Group A had an alpha-blocker (tamsulosin 0.2 mg) at bedtime more than 4 weeks of an alpha-blocker plus an anticholinergic agent (solifenacin 5 mg) orally at bedtime, then, finally, 4 weeks of an alpha-blocker plus an antidiuretic agent (desmopressin 0.2 mg). Group B had an alpha-blocker for 4 weeks orally at bedtime followed by 4 weeks of an alpha-blocker plus an antidiuretic agent (desmopressin 0.2 mg) orally at bedtime, then, finally, 4 weeks of an alpha-blocker plus an anticholinergic agent (solifenacin 5 mg). Patients were required to visit the outpatient clinic at the start of the study for screening and after 4, 8, and 12 weeks ([Figure 1](#f1-cia-9-1021){ref-type="fig"}). If adverse events occurred, patients were excluded during the study without dose titration or drug resting time.
Assessment tools
----------------
All patients recorded a 3-day VD at baseline and at 4, 8, and 12 weeks. Patients in both groups were subdivided into three subgroups: NP (voided urine volume during sleep \>33% of the 24-hour output), decreased NBC (NBC index is greater than 0; NBC index corresponds to the actual number of voids subtracted by the predicted number of voids; predict number of voids, nocturia index --1; nocturia index is the nocturnal urine volume divided by functional bladder capacity; and functional bladder capacity is the single largest volume voided and recorded in VD), and nocturia due to both causes. Total IPSS and IPSS sub-scores (voiding and storage) and Overactive Bladder Symptom Score (OABSS) were analyzed at baseline and at 4, 8, and 12 weeks.[@b14-cia-9-1021] Q~max~ and PVR were assessed using flowmetry and ultrasound (9032A0072, Medtronic, Skovlunde and TYPE 2202; B&K Medical, Herlev, Denmark), respectively. Safety was also evaluated from reported adverse events and laboratory data.
Statistical analysis
--------------------
All data are reported as the mean ± standard deviation. Data were analyzed using SPSS software (v 15.0; IBM Corporation, Armonk, NY, USA). A *P*-value of \<0.05 was considered significant. One-way analysis of variance tests were used to compare the baseline data between the groups. Mean changes from baseline to 4 weeks, 4--8 weeks, and 4--12 weeks in total IPSS score, IPSS sub-scores, OABSS, Q~max~, and PVR were analyzed using paired Student's *t*-test.
Results
=======
Among the 435 patients screened for the study, eight withdrew due to screening failure. A total of 427 patients were prospectively enrolled and 405 patients completed the study ([Figure 1](#f1-cia-9-1021){ref-type="fig"}). Twenty-two (5.4%) patients withdrew due to adverse events. Of those, nine withdrew from worse voiding by the addition of anticholinergic agent, seven experienced dry mouth, and one experienced constipation. Four withdrew due to dizziness and hyponatremia when taking an antidiuretic agent as add-on therapy. One patient withdrew due to uncontrolled hypertension during follow-up.
The baseline characteristics of IPSS and OABSS, uroflowmetry parameters, and 3-day VD details of Groups A and B are summarized in [Table 1](#t1-cia-9-1021){ref-type="table"}. In Groups A and B, the NP subgroup displayed significantly higher nocturnal urine volume and OABSS, while the number of episodes of urgency in 3 days and NBC index were lower in this subgroup than in the others (*P*\<0.05; [Tables 2](#t2-cia-9-1021){ref-type="table"} and [3](#t3-cia-9-1021){ref-type="table"}). In the decreased NBC subgroup, the IPSS storage sub-score was significantly higher than in the other subgroups in Groups A and B (*P*\<0.05; [Tables 2](#t2-cia-9-1021){ref-type="table"} and [3](#t3-cia-9-1021){ref-type="table"}). During the treatment, the changes from baseline in total IPSS significantly decreased at 4 weeks (*P*\<0.05; [Tables 4](#t4-cia-9-1021){ref-type="table"} and [5](#t5-cia-9-1021){ref-type="table"}, [Figures 2](#f2-cia-9-1021){ref-type="fig"} and [3](#f3-cia-9-1021){ref-type="fig"}). The IPSS storage sub-score was not significantly altered in any subgroup at 4 weeks compared with at baseline. The IPSS voiding sub-score was significantly decreased in the NP subgroup and decreased in the NBC subgroup but not in the nocturia due to both causes subgroups ([Tables 4](#t4-cia-9-1021){ref-type="table"} and [5](#t5-cia-9-1021){ref-type="table"}, [Figures 2](#f2-cia-9-1021){ref-type="fig"} and [3](#f3-cia-9-1021){ref-type="fig"}). Q~max~ and PVR were not changed at 12 weeks.
In the NP subgroup of Group A, the number of episodes of nocturia in 3 days, nocturnal urine volume, and nocturnal index were significantly decreased at 12 weeks compared with at 4 weeks (*P*\<0.05; [Table 4](#t4-cia-9-1021){ref-type="table"} and [Figure 4](#f4-cia-9-1021){ref-type="fig"}). In Group B, nocturnal urine volume and NP index were also decreased significantly at 8 weeks compared with at 4 weeks (*P*\<0.05; [Table 5](#t5-cia-9-1021){ref-type="table"} and [Figure 5](#f5-cia-9-1021){ref-type="fig"}).
In the decreased NBC subgroup of Group A, IPSS storage sub-score, OABSS, number of episodes of nocturia in 3 days, number of episodes of urgency in 3 days, and NBC index were significantly decreased at 8 weeks compared with at 4 weeks (*P*\<0.05; [Table 4](#t4-cia-9-1021){ref-type="table"} and [Figures 2](#f2-cia-9-1021){ref-type="fig"} and [4](#f4-cia-9-1021){ref-type="fig"}). Also in Group B, IPSS storage sub-score, OABSS, number of episodes of nocturia in 3 days, number of episodes of urgency in 3 days, and NBC index were decreased significantly at 12 weeks compared with at 4 weeks (*P*\<0.05; [Table 5](#t5-cia-9-1021){ref-type="table"} and [Figures 3](#f3-cia-9-1021){ref-type="fig"} and [5](#f5-cia-9-1021){ref-type="fig"}).
In the nocturia due to both causes subgroup of Group A, IPSS storage sub-score, OABSS, and number of episodes of urgency in 3 days were significantly decreased at 8 weeks compared with at 4 and 12 weeks, while nocturnal urine volume and OABSS were significantly decreased at 12 weeks compared with at 4 weeks (*P*\<0.05; [Table 4](#t4-cia-9-1021){ref-type="table"} and [Figures 2](#f2-cia-9-1021){ref-type="fig"} and [4](#f4-cia-9-1021){ref-type="fig"}). In Group B, IPSS storage sub-score, OABSS, and number of episodes of urgency in 3 days were decreased significantly at 12 weeks compared with at 4 weeks, and nocturnal urine volume and OABSS were significantly decreased at 8 weeks compared with at 4 weeks (*P*\<0.05; [Table 5](#t5-cia-9-1021){ref-type="table"} and [Figures 3](#f3-cia-9-1021){ref-type="fig"} and [5](#f5-cia-9-1021){ref-type="fig"}).
Discussion
==========
LUTS include OAB symptoms like frequency, urgency, and incontinence. OAB symptoms are generally more bothersome and represent an important target in the management of BPH. Incontinence frequency, urgency, and urge have been attributed to detrusor overactivity, which reportedly occurs in 40%--70% of patients with BOO.[@b15-cia-9-1021] Alpha-blockers, which promote relaxation of the bladder neck and prostate smooth muscle, have had limited success in the treatment of OAB-related symptoms.[@b16-cia-9-1021] Kaplan et al demonstrated that relative to subjects receiving placebo, men receiving extended-release tolterodine and tamsulosin had significant improvements in the total IPSS and quality-of-life scores, number of urgency episodes and urge incontinence episodes.[@b6-cia-9-1021] Lee et al reported benefits for men with OAB symptoms from an initial combination treatment of an alpha-blocker with anticholinergics, without increased risk of voiding difficulty and acute urinary retention.[@b17-cia-9-1021]
In our study, the combination treatment of an alpha-blocker plus an anticholinergic agent proved beneficial for the decreased NBC subgroup of Groups A and B. The IPSS storage sub-score, OABSS, number of episodes of nocturia in 3 days, number of episodes of urgency in 3 days, and NBC index were significantly decreased in these subjects.
Nocturia cannot be explained based on BPH alone, but can be consecutive to polyuria, diabetes mellitus, neurogenic bladder, cardiac failure, polydipsia, reduced bladder capacity, insomnia, or psychiatric problems. To reduce nocturia, actions can be directed at BOO, bladder sensitivity by anticholinergics, sleepiness by hypnotic drugs, or urinary volume by antidiuretics.[@b18-cia-9-1021] The serum concentration of the antidiuretic agent vasopressin increases during the night, which decreases urine secretion. Secretion of this hormone diminishes with age, when the renal response to antidiuretic agents decreases and the reduction of the total amount of nephrons limit the renal response to the hormone.[@b18-cia-9-1021] Antidiuretic agents reduce nocturnal diuresis and the number of nocturnal voids, and increase the time between going to bed and the first nocturnal void. Several clinical trials investigating the benefits of desmopressin therapy for patients with nocturia have been completed.[@b10-cia-9-1021]--[@b12-cia-9-1021] In our study, the combination treatment of an alpha-blocker with an antidiuretic agent provided benefits for the NP subgroup of Groups A and B. The number of episodes of nocturia in 3 days, nocturnal urine volume, and nocturnal index were significantly decreased.
Objective measurement of subjective LUTS has proven to be a clinical challenge. A patient-completed 3-day VD is commonly used in clinical trials as a primary tool for measuring these symptoms. Frequency-volume chart records the volumes voided as well as the time of each micturition, day and night, for at least 24 hours, while a VD records the times of micturitions and voided volumes, incontinence episodes, pad usage, and other information, such as the degree of urgency and degree of incontinence.[@b19-cia-9-1021] These measures are noninvasive, inexpensive, and accurate. As the 3-day VD is usually recorded during normal daily activities, it can also provide important information on a patient's voiding problem.[@b20-cia-9-1021] The 3-day duration seems to be the most commonly used form to ensure the accuracy of the VD, as well as minimizing the burden of recording.[@b21-cia-9-1021] Combined with other clinical findings and diagnostic tools, the 3-day VD might provide additional information.
Our study did not show a benefit with the medication order of anticholinergic agent or antidiuretic agent as add-on therapy for nocturia in men previously treated with an alpha-blocker for LUTS. An anticholinergic agent or an antidiuretic agent added to alpha-blocker therapy, according to voiding disorder (nocturnal polyuria, decreased NBC, or nocturia by both causes subgroups), was found to provide benefits to refractory nocturia in men previously treated with an alpha-blocker for LUTS.
There are some limitations to our study that should be considered when interpreting the results. First, 5-alpha reductase inhibitors were not considered as a medication for LUTS. A 5-alpha reductase inhibitor along with an anticholinergic agent or antidiuretic agent added to an alpha-blocker could be helpful to patients with refractory nocturia. Second, the period of treatment for each sequence was limited to 4 weeks. A longer period for each sequence could have provided more information and should be assessed. Third, the collection of information was based on a self-reported VD, so reporting bias was inevitable. A monitoring system should be considered in subsequent studies to bolster the results.
Notwithstanding these limitations, as far as we are aware, the present study is the first to report that adding an anticholinergic or antidiuretic agent to an alpha-blocker provides better results in terms of improving LUTS, especially nocturia, than an alpha-blocker alone.
This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare, Republic of Korea (HI13C0104). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
**Author contributions**
Study concept and design: Jong Kwan Park and Yu Seob Shin; organization and data management: Yu Seob Shin, Chen Zhao, Li Tao Zhang, Young Gon Kim, and Jong Kwan Park; intervention design: Jong Kwan Park, Yu Seob Shin; study physicians: Jong Kwan Park and Yu Seob Shin; statistical analysis: Jong Kwan Park and Yu Seob Shin; analysis and interpretation of data: Jong Kwan Park and Yu Seob Shin, Chen Zhao, and Li Tao Zhang. All authors drafted or revised, commented on, and approved the final version of the manuscript for publication.
**Disclosure**
The authors declare no conflicts of interest in this work.
{#f1-cia-9-1021}
{#f2-cia-9-1021}
{#f3-cia-9-1021}
{#f4-cia-9-1021}
{#f5-cia-9-1021}
######
Baseline characteristics in International Prostate Symptom Score (IPSS) and uroflowmetry parameters, 3-day voiding diary (VD) of Groups A[a](#tfn2-cia-9-1021){ref-type="table-fn"} and B[b](#tfn3-cia-9-1021){ref-type="table-fn"}
Characteristic Group A Group B *P*-value
--------------------------------- ------------- ------------- -----------
Patients
N 209 196
Age, years 66.6±5.4 64.6±4.4 0.56
BMI, kg/m^2^ 23.5±2.5 22.8±1.9 0.72
IPSS
Total 18.0±5.3 16.5±4.5 0.38
Voiding symptoms 9.0±3.2 8.4±3.5 0.64
Storage symptoms 9.0±2.7 8.1±2.5 0.73
Nocturia 2.9±0.5 2.8±1.1 0.86
OABSS 11.4±5.2 12.8±5.7 0.23
Uroflowmetry
Q~max~, mL/second 13.5±4.7 14.9±5.9 0.17
PVR, mL 33.6±22.7 32.9±20.2 0.69
3-day VD
Nocturia episodes in 3 days, n 8.7±3.2 8.6±2.9 0.93
Nocturnal urine volume, mL 666.0±132.9 710.0±152.1 0.14
Nocturnal index 1.4±0.3 1.6±0.7 0.37
Urgency episodes in 3 days, n 5.6±2.9 5.9±1.9 0.54
NBC index 0.5±0.1 0.5±0.2 0.93
**Notes:** Values are mean ± standard deviation.
Alpha-blocker for 4 weeks, followed by alpha-blocker plus an anticholinergic agent for 4 weeks, then, lastly, an alpha-blocker plus an antidiuretic agent for 4 weeks
alpha-blocker for 4 weeks, followed by alpha-blocker plus an antidiuretic agent for 4 weeks, then, lastly, an alpha-blocker plus an anticholinergic agent for 4 weeks.
**Abbreviations:** BMI, body-mass index; NBC, nocturnal bladder capacity; OABSS, Overactive Bladder Symptom Score; PVR, post-void residual urine; Q~max~, maximum urinary flow rate; n, number.
######
Baseline characteristics of International Prostate Symptom Score (IPSS) and uroflowmetry parameters, 3-day voiding diary (VD) of Group A[a](#tfn7-cia-9-1021){ref-type="table-fn"}
Characteristic NP Decreased NBC Nocturia due to both causes *P*-value
--------------------------------- -------------------------------------------------------- ------------------------------------------------------ ----------------------------- -----------
Patients
N 83 51 75
Age, years 62.3±5.7 63.7±5.1 65.1±7.5 NS
BMI, kg/m^2^ 23.8±2.3 25.3±1.5 22.1±3.5 NS
IPSS
Total 16.7±6.5 20.2±5.7 18.2±7.4 NS
Voiding symptoms 9.3±3.7 8.3±3.8 9.3±4.9 NS
Storage symptoms 7.4±2.5 11.9±2.5[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 8.9±3.2 \<0.05
Nocturia 2.7±0.6 3.0±1.3 3.2±1.3 NS
OABSS 6.8±3.5[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 15.5±7.2 13.8±5.6 \<0.05
Uroflowmetry
Q~max~, mL/second 13.4±5.1 9.9±2.9 16.2±8.7 NS
PVR, mL 35.2±24.5 42.5±13.2[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 25.9±20.4 \<0.05
3-day VD
Nocturia episodes in 3 days, n 8.6±3.1 9.7±4.4 8.2±3.7 NS
Nocturnal urine volume (mL) 768.0±126.5[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 539.0±104.9 642.0±96.2 \<0.05
Nocturnal index 1.5±0.4 1.2±0.6[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 1.7±0.4 \<0.05
Urgency episodes in 3 days, n 1.2±0.3[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 9.8±3.5 7.7±4.2 \<0.05
NBC index −0.3±0.1[\*](#tfn6-cia-9-1021){ref-type="table-fn"} 1.4±0.5 0.8±0.3 \<0.05
**Notes:** Values are mean ± standard deviation
*P*\<0.05.
Alpha-blocker for 4 weeks, followed by alpha-blocker plus an anticholinergic agent for 4 weeks, then, lastly, an alpha-blocker plus an antidiuretic agent for 4 weeks.
**Abbreviations:** BMI, body-mass index; NBC, nocturnal bladder capacity; NP, nocturnal polyuria; NS, nonsignificant; OABSS, Overactive Bladder Symptom Score; PVR, post-void residual urine; Q~max~, maximum urinary flow rate.
######
Baseline characteristics of International Prostate Symptom Score (IPSS) and uroflowmetry parameters, 3-day voiding diary (VD) of Group B[a](#tfn11-cia-9-1021){ref-type="table-fn"}
Characteristic NP Decreased NBC Nocturia due to both causes *P*-value
--------------------------------- --------------------------------------------------------- ------------------------------------------------------ ----------------------------- -----------
Patients
N 76 42 78
Age, years 65.2±4.3 62.5±3.2 65.3±6.7 NS
BMI, kg/m^2^ 24.5±1.3 22.4±1.7 21.4±2.4 NS
IPSS
Total 14.5±6.5 18.2±5.7 17.7±7.4 NS
Voiding symptoms 8.2±2.6 7.5±2.8 9.1±4.5 NS
Storage symptoms 6.3±2.5 10.7±3.3[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 8.6±2.8 \<0.05
Nocturia 2.5±0.7 3.2±1.1 3.1±1.6 NS
OABSS 7.2±4.1[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 17.2±6.3 16.1±6.3 \<0.05
Uroflowmetry
Q~max~, mL/second 15.3±5.4 9.2±2.2[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 17.8±10.8 \<0.05
PVR, mL 33.8±22.3 37.2±11.1 29.8±22.5 NS
3-day VD
Nocturia episodes in 3 days, n 8.3±3.5 9.2±5.1 8.6±2.9 NS
Nocturnal urine volume (mL) 853.0±150.3[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 562.0±93.7 651.0±95.8 \<0.05
Nocturnal index 1.6±0.2 1.3±0.5[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 1.8±0.7 \<0.05
Urgency episodes in 3 days, n 1.4±0.5[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 9.6±4.2 8.3±3.4 \<0.05
NBC index −0.2±0.1[\*](#tfn10-cia-9-1021){ref-type="table-fn"} 1.3±0.5 0.9±0.2 \<0.05
**Notes:** Values are mean ± standard deviation
*P*\<0.05.
Alpha-blocker for 4 weeks, followed by alpha-blocker plus an antidiuretic agent for 4 weeks, then, lastly, an alpha-blocker plus an anticholinergic agent for 4 weeks.
**Abbreviations:** BMI, body-mass index; NBC, nocturnal bladder capacity; NP, nocturnal polyuria; NS, nonsignificant; OABSS, Overactive Bladder Symptom Score; PVR, post-void residual urine; Q~max~, maximum urinary flow rate.
######
Changes in the International Prostate Symptom Score (IPSS) and uroflowmetry parameters, 3-day voiding diary (VD) after treatment in Group A[a](#tfn16-cia-9-1021){ref-type="table-fn"}
Characteristic/parameter/score Baseline 4 weeks (alpha-blocker) 8 weeks (alpha-blocker + anticholinergic agent) 12 weeks (alpha-blocker + antidiuretic agent)
------------------------------------ ------------- ------------------------------------------------------ ----------------------------------------------------- -------------------------------------------------------
NP (n=83)
IPSS
Total 16.7±6.5 14.2±3.2[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 13.9±4.3 11.2±3.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Voiding symptoms 9.3±3.7 7.7±4.2[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 7.8±3.1 6.4±2.9
Storage symptoms 7.4±2.5 6.5±2.5 6.1±3.3 4.8±1.5[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Nocturia 2.7±0.6 2.3±0.2[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 2.4±0.6 1.8±0.1[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
OABSS 6.8±3.5 6.1±2.7 5.5±2.9 5.7±2.2
Uroflowmetry
Q~max~, mL/second 13.4±5.1 14.7±7.2 14.9±6.7 15.3±4.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
PVR, mL 35.2±24.5 32.1±22.3 33.5±21.7 29.1±15.3
3-day VD
Nocturia episodes in 3 days, n 8.6±3.1 7.2±2.3[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 8.0±3.7 6.2±2.3[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Nocturnal urine volume (mL) 768.0±126.5 732.0±103.1 751.0±117.2 526.0±96.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Nocturnal index 1.5±0.4 1.4±0.2 1.4±0.5 1.1±0.1[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Urgency episodes in 3 days, n 1.2±0.3 1.0±0.4 1.1±0.3 1.2±0.2
NBC index −0.3±0.1 −0.3±0.2 −0.2±0.1 −0.3±0.1
Decreased NBC (n=51)
IPSS
Total 20.2±5.7 18.4±7.3[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 16.2±5.1[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 18.1±9.5
Voiding symptoms 8.3±3.8 7.6±1.8[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 7.6±1.7 7.9±3.3
Storage symptoms 11.9±2.5 10.8±2.5 8.6±1.5[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 10.2±2.5
Nocturia 3.0±1.3 2.9±1.1 2.2±0.3[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 2.8±1.2
OABSS 15.5±7.2 14.7±6.3 12.9±6.7[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 14.2±5.7
Uroflowmetry
Q~max~, mL/second 9.9±2.9 12.4±3.5 15.3±2.9 12.2±1.7
PVR, mL 42.5±13.2 36.2±11.1 38.5±10.1 40.4±11.8
3-day VD
Nocturia episodes in 3 days, n 9.7±4.4 8.2±2.9[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 6.3±2.7[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 8.1±3.4
Nocturnal urine volume (mL) 539.0±104.9 568.0±86.2 553.0±97.9 543.0±85.1
Nocturnal index 1.2±0.6 1.2±0.2 1.3±0.5 1.2±0.1
Urgency episodes in 3 days, n 9.8±3.5 9.7±4.1 7.2±2.6[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 8.8±3.4
NBC index 1.4±0.5 1.3±0.3 0.9±0.3[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 1.4±0.2
Nocturia due to both causes (n=75)
IPSS
Total 18.2±7.4 16.6±6.1[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 17.1±7.4 16.2±3.6
Voiding symptoms 9.3±4.9 8.9±4.9 9.9±4.9 10.3±4.3
Storage symptoms 8.9±3.2 7.7±3.2 5.9±3.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 7.2±3.2
Nocturia 3.2±1.3 2.5±1.1[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 2.2±0.9[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 2.2±1.4
OABSS 13.8±5.6 13.9±5.2 9.7±3.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 10.2±5.3[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Uroflowmetry
Q~max~, mL/second 16.2±8.7 18.8±12.7 19.4±12.3 22.7±40.8
PVR, mL 25.9±20.4 22.9±21.3 23.5±18.1 18.5±16.2
3-day VD
Nocturia episodes in 3 days, n 8.2±3.7 7.2±2.9[\*](#tfn14-cia-9-1021){ref-type="table-fn"} 6.3±2.7[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 8.1±3.4
Nocturnal urine volume (mL) 642.0±96.2 595.0±78.8 601.0±94.2 543.0±85.1[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Nocturnal index 1.7±0.4 1.6±0.3 1.4±0.7 1.3±0.2[‡](#tfn15-cia-9-1021){ref-type="table-fn"}
Urgency episodes in 3 days, n 7.7±4.2 7.2±3.9 5.4±2.3[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 7.5±3.3
NBC index 0.8±0.3 0.8±0.3 0.5±0.1[‡](#tfn15-cia-9-1021){ref-type="table-fn"} 0.9±0.2
**Notes:** Values are mean ± standard deviation
*P*\<0.05 changes from baseline
*P*\<0.05 changes from 4 weeks.
Alpha-blocker for 4 weeks, followed by alpha-blocker plus an anticholinergic agent for 4 weeks, then, lastly, an alpha-blocker plus an antidiuretic agent for 4 weeks.
**Abbreviations:** NBC, nocturnal bladder capacity; NP, nocturnal polyuria; OABSS, Overactive Bladder Symptom Score; PVR, post-void residual urine; Q~max~, maximum urinary flow rate; VD, voiding diary.
######
Changes in the International Prostate Symptom Score (IPSS) and uroflowmetry parameters, 3-day VD after treatment in Group B[a](#tfn21-cia-9-1021){ref-type="table-fn"}
Characteristic/parameter/score Baseline 4 weeks (alpha-blocker) 8 weeks (alpha-blocker + antidiuretic agent) 12 weeks (alpha-blocker + anticholinergic agent)
------------------------------------ ------------- ------------------------------------------------------ ------------------------------------------------------- -----------------------------------------------------
NP (n=76)
IPSS
Total 14.5±6.5 12.1±3.2[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 11.5±2.7[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 13.7±5.2
Voiding symptoms 8.2±2.6 6.7±4.4[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 7.2±3.3 7.9±4.1
Storage symptoms 6.3±2.5 6.4±3.1 4.3±1.5[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 5.8±2.6
Nocturia 2.5±0.7 2.1±0.4 1.7±0.2[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 2.5±0.7
OABSS 7.2±4.1 6.8±2.7 6.5±2.4 7.0±3.5
Uroflowmetry
Q~max~, mL/second 15.3±5.4 15.9±5.2 14.6±5.1 12.5±4.2
PVR, mL 33.8±22.3 31.5±15.2 29.7±12.5 33.5±21.7
3-day VD
Nocturia episodes in 3 days, n 8.3±3.5 6.4±1.7[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 5.8±2.6[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 7.0±4.5
Nocturnal urine volume (mL) 853.0±150.3 798.0±92.4 559.0±93.1[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 815.0±106.5
Nocturnal index 1.6±0.2 1.5±0.4 1.0±0.2[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 1.4±0.7
Urgency episodes in 3 days, n 1.4±0.5 1.3±0.8 1.1±0.2 1.4±0.4
NBC index −0.2±0.1 −0.2±0.2 −0.2±0.1 −0.1±0.1
Decreased NBC (n=42)
IPSS
Total 18.2±5.7 16.2±5.4[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 16.1±9.5 13.2±5.1[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Voiding symptoms 7.5±2.8 6.5±1.8[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 6.4±3.3 6.9±1.7
Storage symptoms 10.7±3.3 9.7±2.5 9.7±2.8 6.3±1.2[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Nocturia 3.2±1.1 2.8±1.3 2.7±0.9 2.0±0.6[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
OABSS 17.2±6.3 15.8±7.1 15.5±6.7 13.1±7.4[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Uroflowmetry
Q~max~, mL/second 9.2±2.2 10.8±2.5 15.3±2.7 12.6±2.2
PVR, mL 37.2±11.1 30.3±13.2 23.1±10.7 32.1±13.2
3-day VD
Nocturia episodes in 3 days, n 9.2±5.1 7.3±2.5[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 7.4±2.4 6.1±2.5[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Nocturnal urine volume (mL) 562.0±93.7 571.0±89.1 533.0±82.1 623.0±121.9
Nocturnal index 1.3±0.5 1.3±0.2 1.1±0.1 1.2±0.3
Urgency episodes in 3 days, n 9.6±4.2 9.7±3.5 9.3±3.7 7.1±5.3[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
NBC index 1.3±0.5 1.2±0.6 1.3±0.3 0.5±0.1[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Nocturia due to both causes (n=78)
IPSS
Total 17.7±7.4 14.3±5.9[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 13.0±3.1 12.4±7.4[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Voiding symptoms 9.1±4.5 6.9±2.5 7.1±3.9 6.2±4.9
Storage symptoms 8.6±2.8 7.4±3.2 5.9±2.5 5.2±3.2[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Nocturia 3.1±1.6 2.3±1.2[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 2.2±1.7 2.3±0.5
OABSS 16.1±6.3 14.8±7.5 13.2±6.1[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 12.4±4.8[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
Uroflowmetry
Q~max~ mL/sec 17.8±10.8 20.1±13.4 20.5±10.6 18.2±10.9
PVR, mL 29.8±22.5 25.7±20.7 21.5±18.3 25.5±17.3
3-day VD
Nocturia episodes in 3 days, n 8.6±2.9 7.0±2.1[\*](#tfn19-cia-9-1021){ref-type="table-fn"} 6.8±3.3 6.9±2.5
Nocturnal urine volume (mL) 651.0±95.8 623.0±89.8 521.0±64.7[‡](#tfn20-cia-9-1021){ref-type="table-fn"} 631.0±92.3
Nocturnal index 1.8±0.7 1.5±0.6 1.4±0.3 1.4±0.8
Urgency episodes in 3 days, n 8.3±3.4 7.9±2.8 8.1±2.9 6.5±3.2[‡](#tfn20-cia-9-1021){ref-type="table-fn"}
NBC index 0.9±0.2 0.9±0.5 0.8±0.5 0.8±0.1
**Notes:** Values are mean ± standard deviation;
*P*\<0.05 changes from baseline.
*P*\<0.05 changes from 4-weeks.
Alpha-blocker for 4 weeks, followed by alpha-blocker plus an antidiuretic agent for 4 weeks, then, lastly, an alpha-blocker plus an anticholinergic agent for 4 weeks.
**Abbreviations:** NBC, nocturnal bladder capacity; NP, nocturnal polyuria; OABSS, Overactive Bladder Symptom Score; PVR, post-void residual urine; Q~max~, maximum urinary flow rate; VD, voiding diary.
| {
"pile_set_name": "PubMed Central"
} |
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus.
Introduction
============
Acute complications of traumatic brain injury (TBI) should be a major concern for neurosurgeons worldwide, as the delayed onset of an epidural hematoma, although rare, can occur, resulting in the rapid decline of a patient's condition \[[@REF1]\]. A large-population case study conducted by Irie et al. in 2011 revealed that individuals aged 10 to 24 typically acquired epidural hematomas as a result of motor vehicle accidents. Additionally, 81% of the total epidural hematomas result from accidental injury. In addition, 75% of the study population also had a varying severity of skull fracture \[[@REF2]\]. When evaluating TBI patients, the possibility of delayed onset epidural hematoma (EDH) should be floating in the midst of a neurosurgeon's mind. Factors such as age, the severity of the TBI, and neurological status are key diagnostic parameters that impact the risk of developing a delayed-onset epidural hematoma. Routine CT radiological imaging every four to six hours is essential for detecting delayed EDH onset, while quite easy to do so, overlooking the parameters we discussed could position the patient down an undesirable route that may lead to morbidity and/or mortality \[[@REF3]\].
Case presentation
=================
A Hispanic male, 18 years of age, presented to the emergency department (ED) after a head-on collision with a vehicle traveling at 30 miles per hours (mph) while riding his bicycle. Our patient initially presented in a promising condition with a Glasgow Coma Score (GCS) of 13, however, this rapidly deteriorated to a GCS of 5 within a four-hour window. Immediate CT studies (Figure [1](#FIG1){ref-type="fig"}) revealed the presence of a large, right-sided, acute EDH with a 6-mm midline shift, indicating the need for immediate surgical intervention.
{#FIG1}
An emergent right craniotomy for acute EDH evacuation was performed, where, intraoperatively, a nondisplaced frontotemporal skull fracture was noted with bleeding from the branches of the middle meningeal artery, which was immediately stopped. Postoperatively, the patient's condition improved from a pre-operative GCS of 5 to a postoperative GCS of 15. Our patient was hospitalized for 72 hours and released on postoperative day four with a follow-up scheduled for two weeks later. The patient was recommended to complete a four-week course of physical and occupational therapy to regain full pre-traumatic injury quality of life.
Discussion
==========
The ideal management of traumatic brain injury (TBI) patients has been a neglected subset of neurological surgery and emergency medicine for much of allosteric medicinal practices since its existence. Recent advancements in medicinal technology, specifically high-resolution radiological imaging, has opened the door for novel pathology discovery techniques. Patients presenting with any trauma to the head should be subjected to a thorough neurological examination by a neurosurgeon or neurologist to ensure the intactness of the nervous system. However, the management of TBI patients is not a static task; constant monitoring of the patient's condition and re-evaluation is of key importance. In our case, an 18-year-old male initially presented with a GCS of 13 but within a mere four hours, there was a rapid deterioration to a GCS of 5, a state of coma and unconsciousness. The rapid deterioration of a patient's condition following a TBI serves as a highly reliable diagnostic tool for the need for immediate radiological studies and, often, surgical intervention. Upon our patient\'s rapid deterioration, a stat CT scan was ordered where a large, right-sided, acute EDH, with a 6-mm midline shift was identified. Patients with delayed onset epidural hematomas are also more likely to have an overlying skull fracture. Through stat emergency department imaging studies, no skull fracture was visualized, however, intraoperatively, a nondisplaced frontotemporal skull fracture was noted with bleeding. This highlights a key concept that radiological imaging is not a concrete diagnostic tool and serves as an aid to our professional opinion. Upon the completion of craniotomy, the patient made a full, expected recovery and was discharged on postoperative day four. Patients with traumatic brain injury are unique and, often, high-demand patients who require constant monitoring and advanced care. Ignoring the symptoms of any complexity may become deadly in a short period of time and greatly reduce the chances of a complete postoperative outcome. Throughout the years, EDH recovery rates have improved greatly and, today, only bear a two to five percent mortality rate, thanks to the improved care of TBI patients from the emergency department, neurosurgery, and allied healthcare team members.
Conclusions
===========
The imperative findings of this report highlight how essential the rapid detection of delayed onset EDH is in saving a life. Letting the sun fully rise may lead to not so bright times. Serial radiological imaging is crucial for TBI patients in identifying developing neurologically relevant pathology; paying close attention to the condition of the patient is a key diagnostic tool. Any enlarging EDH with mass-effect on the brain constitutes the need for immediate surgical intervention to maximize post-traumatic outcomes.
The authors have declared that no competing interests exist.
Consent was obtained by all participants in this study
| {
"pile_set_name": "PubMed Central"
} |
Introduction
============
Hepatocellular carcinoma (HCC) is a common malignancy globally, and its mortality rate ranked fourth in 2015 ([@b1-ol-0-0-7419],[@b2-ol-0-0-7419]). In total, \~60% of patients with HCC are diagnosed with unresectable HCC, which represents an incurable disease ([@b3-ol-0-0-7419],[@b4-ol-0-0-7419]). Barcelona Clinic Liver Cancer (BCLC) guidelines recommend patients with intermediate-stage HCC to receive trans-arterial chemoembolization (TACE) as a standard treatment. Two randomized controlled clinical trials demonstrated the benefits of TACE treatment for patients with HCC ([@b5-ol-0-0-7419],[@b6-ol-0-0-7419]). According to BCLC guidelines, patients with HCC who present an asymptomatic HCC, lack of portal vein thrombosis or extra-hepatic spread, compensated cirrhosis, and patients classified with Child-Pugh scores \<8 and a performance status (PS) of 0, are advised to receive TACE treatment ([@b4-ol-0-0-7419]). In total, \~20% of patients with HCC are at an intermediate stage. These patients form a heterogeneous group, owing to discrepancies in tumor burden, serum biomarker, liver function, performance status, etiology, etc. However, BCLC guidelines do not consider all these parameters. The reported survival rates of patients with intermediate-stage HCC varies between 11 and 45 months ([@b7-ol-0-0-7419]).
Staging systems, including model to estimate survival rates for hepatocellular carcinoma (MESH), hepatoma arterial embolization prognostic score (HAP), modified HAP (mHAP), performance status combined Japan Integrated Staging system (PSJIS) and tumor-node-metastasis (TNM), have been developed to determine the optimal treatment for the patients. However, it remains unknown which system is optimal for the prediction of patient survival rates.
In the present study, the performance of five staging systems, i.e., MESH, HAP, mHAP, PSJIS and TNM, was compared in predicting 3-month survival, 6-month survival, 1-year survival and OS survival rates of patients with HBV-associated HCC undergoing TACE. The aim of the present study was to determine the optimal staging system for patients with HCC.
Patients and methods
====================
### Patients
A total of 220 sequential patients with HCC treated with TACE were retrospectively reviewed at the Third Affiliated Hospital of Sun Yat-Sen University (Guangzhou, China) between July 2009 and June 2012. Patient characteristics are provided in [Table I](#tI-ol-0-0-7419){ref-type="table"}. The diagnosis of HCC was confirmed using pathology or magnetic resonance imaging (MRI)/computed tomography (CT) according to the American Association for the Study of Liver Diseases guidelines ([@b3-ol-0-0-7419]). All participants studied were patients with HBV-associated HCC and received TACE following multidisciplinary team discussion. Patients with advanced HCC classified as Child-Pugh grade A or B and with a performance status (PS) of 0 to 2, platelet count ≥30×10^9^ cells/l and hemoglobin level ≥60 g/l, were eligible for enrollment. However, patients were excluded if a second type of cancer and/or intractable comorbid medical illness existed. Patients classified as Child-Pugh grade C were also excluded.
### Data collection
The Institutional Review Board of the Third Affiliated Hospital of Sun Yat-Sen University reviewed and approved the present study. Prior to enrollment, all participants provided written informed consent for data sharing.
A range of demographic data were collected including risk factors, blood results, imaging and therapy data. Collected data also included sex, age, date of diagnosis, date of mortality or last follow-up date. The clinical records of the patients were retrospectively assessed. Tumor characteristics, including tumor size and extension, vascular invasion and lymph node metastases, were assessed using CT or MRI. Routine blood tests, liver function and coagulation tests were also conducted.
### Staging
Baseline data were collected to classify patients according to MESH, HAP, mHAP, PSJIS and TNM systems. All eligible patients were classified by MESH, HAP, mHAP, PSJIS, TNM and BCLC in the first diagnosis, and 91.8% of patients with HCC were classified according to BCLC-C. Patients with a PS of 2 were also the classified according to BCLC-C at first diagnosis. A baseline evaluation that included laboratory studies, imaging studies (CT or MRI) and clinical examination was performed. Data were collected at the time the patients were diagnosed with advanced HBV-associated HCC. Survival times were defined as the time from first TACE treatment until mortality or last follow-up. Patients who lacked the required data or who were lost to follow-up within 3 months of diagnosis were excluded from the present study.
### Statistical analysis
The primary endpoints of the present study were 3-month survival and OS. The secondary endpoints of the study were 6-month and 1-year survival. Kaplan-Meier estimator survival curves and log rank tests were used to evaluate the OS rate. Likelihood ratio tests (LRTs) were used to compare different staging systems. The degree of freedom was set at 1, so that different prognostic systems with different numbers of stages could be compared. Bias correction of Akaike information criterion (AIC) was applied. Lower AIC and higher likelihood ratio values indicate improved prognosis capacity of a staging system. Statistically significant prognostic variables in univariate analyses were identified by multivariate analysis using Cox\'s proportional hazards model. Receiver operating characteristic (ROC) curve analysis was performed for each staging system\'s predictive value for predicting 3-month, 6-month and 1-year mortality. Higher area under curve (AUC) values of the ROC curves indicate better predictive ability. Data were analyzed using SPSS (version 22.0; IBM Corp., Armonk, NY, USA) and SAS (version 9.0; SAS Institute, Inc., Cary, NC, USA). P\<0.05 was considered to indicate a statistically significant difference.
Results
=======
### Patient characteristics
A total of 220 patients were classified using the MESH, HAP, mHAP, TNM and PSJIS systems. The baseline characteristics of all patients studied are presented in [Table I](#tI-ol-0-0-7419){ref-type="table"}.
### Staging system comparison
Analysis of the prognostic performance of the staging systems to predict 3-month, 6-month and 1-year survival rates was performed. The AUC values for predicting 3-month survival rates for MESH, HAP, mHAP, PSJIS and TNM systems were 0.858, 0.728, 0.690, 0.688 and 0.699, respectively. Additionally, the AUC values of MESH, HAP, mHAP, PSJIS and TNM for predicting 6-month survival rates were 0.822, 0.747, 0.720, 0.722 and 0.715, whereas the respective values for predicting 1-year survival rates were 0.725, 0.664, 0.672, 0.645 and 0.654.
Pairwise comparison of the AUC to predict 3-month, 6-month and 1-year survival rates revealed that the MESH staging system performed optimally in predicting 3- and 6-month survival rates ([Tables II](#tII-ol-0-0-7419){ref-type="table"} and [III](#tIII-ol-0-0-7419){ref-type="table"})([Figs 1](#f1-ol-0-0-7419){ref-type="fig"} and [2](#f2-ol-0-0-7419){ref-type="fig"}). A statistical trend was only observed when MESH was compared with mHAP in predicting 1-year survival rate (P=0.0797; [Table IV](#tIV-ol-0-0-7419){ref-type="table"}; [Fig. 3](#f3-ol-0-0-7419){ref-type="fig"}). MESH exhibited an improved performance compared with HAP, PSJIS and TNM in predicting 1-year survival rate (P\<0.05), while mHAP performed equally well compared with PSJIS and TNM in predicting 1-year survival rate (P\>0.05; [Table IV](#tIV-ol-0-0-7419){ref-type="table"}).
Additionally, the staging systems, including MESH, HAP, mHAP, TNM and PSJIS were analyzed separately, using Kaplan-Meier estimator curves ([Figs. 4](#f4-ol-0-0-7419){ref-type="fig"}--[8](#f8-ol-0-0-7419){ref-type="fig"}, respectively). For the analysis of the MESH system, patients were assigned to risk groups (MESH score 0--1, low risk; score 2--4, intermediate risk; score 5--6, high risk). The median survival was 3 months for the high-risk group, 27 months for the intermediate-risk group and 41 months for the low-risk group, indicating that high-risk patients had a poor survival rate ([@b8-ol-0-0-7419]). For the analysis of the PSJIS system, patients with a score of 0--2 represented a low-risk group, whereas patients in the high-risk group had a score of 5--7 ([@b9-ol-0-0-7419]). The Kaplan-Meier estimator curves exhibited different prognostic strata for MESH, HAP, mHAP, PSJIS and TNM, which was statistically different (log-rank P\<0.05 in all cases). Subsequently, Kaplan-Meier estimator analysis of survival rate revealed that the MESH staging system exhibited an excellent stratified prognostic capacity.
Following comparison of the LRT χ^2^ and AIC values of the five staging systems, MESH demonstrated the highest χ^2^ and lowest AIC value, thus suggesting an improved predictive performance compared with that of the HAP, mHAP, PSJIS and TNM systems ([Table V](#tV-ol-0-0-7419){ref-type="table"}).
### Prognostic factors of survival in patients with HCC
Independent prognostic factors including tumor size, portal vein invasion (segmental), antiviral therapy and bilirubin for OS were revealed by univariate and multivariate analyses ([Table VI](#tVI-ol-0-0-7419){ref-type="table"}).
Discussion
==========
Many patients with HCC are diagnosed at an advanced stage and TACE is a major therapeutic approach. BCLC guidelines recommend patients with intermediate-stage HCC to receive TACE treatment for first-line therapy. Evidence comes from two randomized controlled clinical trials ([@b5-ol-0-0-7419],[@b6-ol-0-0-7419]). Llovet *et al* ([@b5-ol-0-0-7419]) conducted a randomized controlled trial and analyzed 112 patients. It was identified that patients with HCC undergoing chemoembolization exhibited a longer survival time compared with patients undergoing conservative treatment. Chemoembolization decreased mortality by 53%. It was concluded that under careful selection, patients with unresectable HCC received a survival benefit from chemoembolization ([@b5-ol-0-0-7419]). Another trial was conducted by Lo *et al* ([@b6-ol-0-0-7419]), in which 80 patients with HCC were examined. The survival rate of the chemoembolization group was significantly increased compared with that of the control group. Chemoembolization decreased mortality by 51%. This study verified the previous results that TACE treatment significantly prolonged the survival of Asian patients with HCC at an unresectable stage.
However, it is difficult to predict which group of patients would benefit most from TACE treatment. Precise prognosis for patients with HCC under TACE treatment is needed. First, many patients may not respond to TACE although they fulfill the eligibility criteria. Furthermore, patients receiving their first course of TACE may develop liver failure and become unsuitable to receive the second embolization ([@b10-ol-0-0-7419]). Lastly, the development of TACE techniques has broadened the use of TACE beyond the initial eligibility criteria, which widens the heterogeneity of the treatment group survival.
Several staging systems, including MESH, HAP, mHAP and PSJIS, have been developed for more precise prognosis for patients with HCC undergoing TACE. Liu *et al* ([@b8-ol-0-0-7419]) proposed the MESH staging system. This model was derived from the analysis of 3,182 patients with HCC from Taiwan ([@b8-ol-0-0-7419]), where multiple factors, including vascular invasion or metastasis, tumor size, serum α-fetoprotein (AFP) and alkaline phosphatase (ALP) levels, were employed. MESH scores range between 0 and 6. The authors identified that MESH improved prognostic accuracy and refined treatment strategies for patients with HCC when compared with other staging systems including BCLC, Taipei Integrated Scoring (TIS), Cancer of the Liver Italian Program (CLIP) and Hong Kong Liver Cancer (HKLC) ([@b8-ol-0-0-7419]). The second system, known as HAP, was established by Kadalayil *et al* ([@b11-ol-0-0-7419]) following examination of 281 patients with HCC (114 in the training set; 167 in the validation set) undergoing TACE/trans-arterial embolization (TAE). The authors employed independent prognostic factors analyzed using Cox\'s regression ([@b11-ol-0-0-7419]). Those parameters included albumin levels, tumor size, AFP levels and bilirubin levels (albumin, \<36 g/dl; maximum tumor diameter, \>7 cm; AFP, \>400 ng/ml; and bilirubin, \>17 µmol/l). Patients were categorized into HAP groups A-D. The median survival rates for HAP A, B, C and D groups were 27.6, 18.5, 9.0 and 3.6 months, respectively. Patients in the HAP C and D groups were recommended not to receive TACE owing to the poor survival rate. The HAP scoring system was further validated by Pinato *et al* ([@b12-ol-0-0-7419]) who examined 923 patients with HCC from Asia and Europe. The authors proposed a modified version of the HAP score (mHAP), based on the tumor size, albumin levels and AFP levels, but not bilirubin levels. This mHAP system was identified to offer an improved prediction of overall survival (OS) rate compared with HAP ([@b12-ol-0-0-7419]). Another system, the Japan Integrated Score (JIS), was established based on analysis of 722 Japanese patients with HCC ([@b13-ol-0-0-7419]). In addition to this system, Nishikawa *et al* ([@b9-ol-0-0-7419]) proposed the PSJIS system, which is a combination of PS with JIS and derived from 1,170 patients with HCC and with liver cirrhosis. PSJIS was identified to be an improvement over the original JIS system and the BCLC, TNM and CLIP scoring systems in predicting 1-, 3- and 5-year survival rates in patients with transcatheter arterial therapies ([@b9-ol-0-0-7419]).
HAP and mHAP staging systems were established based on the prognostic analysis of TACE/TAE-treated patients with HCC ([@b11-ol-0-0-7419],[@b12-ol-0-0-7419]). Patients with poor prognosis may not benefit from TACE. In the HAP staging system, patients were classified into HAP groups A-D. Median survival rates of patients classified as HAP A, B, C and D was 27.6, 18.5, 9.0 and 3.6 months, respectively ([@b11-ol-0-0-7419]). Patients in the HAP C and D group were advised not to receive TACE because of poor survival. The MESH score includes six common clinical variables including Child-Pugh score, vascular invasion or metastasis presence, tumor number and tumor size, PS, AFP and ALP. The MESH score considers tumor burden, serum biomarker, liver function and PS. It was demonstrated that for BCLC stage B-D patients with HCC, patients may be classified into different prognostic groups based on MESH score ([@b8-ol-0-0-7419]). The MESH score provided an improvement over TIS, HKLC ([@b14-ol-0-0-7419]) and CLIP ([@b15-ol-0-0-7419]).
To the best of our knowledge, MESH has not been studied in geographical areas other than Taiwan, therefore the present study is the first to compare MESH and other staging systems, including HAP, mHAP, PJIS and TNM in patients with HBV-associated HCC under TACE therapy. According to the results of the present study, the MESH score exhibited the highest AUC value when predicting 3-month, 6-month and 1-year survival rates. Life expectancy \>3 months is a common inclusion criterion of TACE clinical trials. Routinely, TACE is repeated every 2--3 months ([@b5-ol-0-0-7419]). As for OS, MESH exhibited the highest χ^2^ value and the lowest AIC value, suggesting that MESH exhibited the optimum performance in terms of discriminatory ability, homogeneity and monotonicity. The MESH score is user-friendly and precise. The median survival of patients with HCC with a MESH score of 5--6 in the present study was 3 months. They would not benefit from TACE due to their poor survival rates.
Sub-classification of the intermediate stage of BCLC was proposed by Bolondi ([@b16-ol-0-0-7419]). Patients with intermediate-stage HCC were classified into four sub-classes (B1-B4) based on Child-Pugh score, tumor burden (assessed by the Milan criteria), PS and portal vein thrombosis. In the present study, the majority of patients with HCC were classified as BCLC-C, therefore Bolondi\'s sub-classification was not evaluated ([@b16-ol-0-0-7419]).
Antiviral therapy was identified as an independent prognostic factor by multivariate analyses. In the present study, all patients with HCC were associated with HBV. In total \~50% of the patients received antiviral therapy. HBV reactivation and hepatic decompensation are major risks in patients with HBV-associated HCC undergoing TACE. Previous studies have also confirmed that patients with HCC and with HBV should be considered for antiviral therapy for preventing hepatic decompensation and HCC development ([@b17-ol-0-0-7419]--[@b19-ol-0-0-7419]).
The present study has certain limitations. First, the patients included were restricted to a single center and the number of patients was limited. Additionally, the patients had HBV-associated advanced HCC. Whether the results of the present study are applicable to patients with HCC not associated with HBV is uncertain. Additional etiologies of HCC, including HCV and alcohol, require further study. Therefore, standard investigations and large-scale prospective studies are required to validate the results of the present study.
In conclusion, MESH score was identified as the most accurate score system for predicting 3-month survival, 6-month survival, 1-year survival and OS rates among the five systems analyzed in the patients with HCC who received TACE treatment in the present study.
The present study was supported by National Natural Science Foundation (grant nos. 31600710 and 81372374), Natural Science Foundation of Guangdong (grant nos. 2014A030313146 and 2016A030313302) and Project on the Integration of Industry, Education and Research of Guangdong Province (grant no. 2012B091100460). The authors thank Mr. Teddy Huang (Bruker Nano Surfaces, Goleta, CA, USA) for his help in language editing.
{#f1-ol-0-0-7419}
{#f2-ol-0-0-7419}
{#f3-ol-0-0-7419}
{#f4-ol-0-0-7419}
{#f5-ol-0-0-7419}
{#f6-ol-0-0-7419}
{#f7-ol-0-0-7419}
{#f8-ol-0-0-7419}
######
Baseline demographic and clinical characteristics of patients.
Characteristic Patients
------------------------------------------- ---------------------
Total patients, n (%) 220 (100)
Sex, n (%)
Male 200 (90.90)
Female 20 (9.10)
Median age, years (range) 52.5 (11--84)
Etiology, n (%)
HBsAg 220 (100)
Tumor size, n (%)
\<2 cm 20 (9.1)
2-5 cm 62 (28.2)
\>5 cm 138 (62.7)
Ascites, n (%) 55 (25.0)
Portal vein invasion (segmental), n (%) 102 (46.4)
Extrahepatic spread 21 (9.5)
White blood cell count, ×10^9^ cells/l 5.90 (1.54--20.6)
(range)
α-fetoprotein, ng/ml (range) 503.47 (1--1210)
Albumin, g/l (range) 38.06 (22.0--53.3)
Creatinine, µmol/l (range) 73.67 (41.0--160.1)
Alkaline phosphatase, U/l (range) 139.90 (44--1048)
Platelets, ×10^9^ cells/l (range) 161.31 (31--520)
Hemoglobin, g/l (range) 130.72 (60--190)
Fibrinogen, g/l (range) 3.48 (1.26--9.39)
Total bilirubin, µmol/l (range) 21.28 (4.7--109.8)
AST, IU/l (range) 82.28 (12--931)
γ-glutamyltransferase, U/l (range) 166.50 (17--1136)
Blood urea nitrogen, mmol/l (range) 5.24 (2.08--12.85)
PT, sec (range) 14.18 (11.2--24.3)
INR (range) 1.11 (0.84--2.21)
90-day survival rate, n (%) 194 (88.18)
6-month survival rate, n (%) 166 (75.45)
1-year survival rate, n (%) 132 (60.00)
2-year survival rate, n (%) 108 (49.09)
5-year survival rate, n (%) 18 (8.18)
TNM 7th edition, n (%)
I 14 (6.4)
II 52 (23.6)
III 38 (17.3)
IV 116 (52.7)
Child-Pugh class, n (%)
A 153 (69.5)
B 67 (30.5)
BCLC, n (%)
A 6 (2.7)
B 12 (5.5)
C 202 (91.8)
MESH, n (%)
0 5 (2.3)
1 27 (12.3)
2 43 (19.5)
3 60 (27.3)
4 53 (24.1)
5 31 (14.1)
6 1 (0.5)
HAP, n (%)
A 26 (11.8)
B 59 (26.8)
C 76 (34.5)
D 59 (26.8)
mHAP, n (%)
A 50 (22.7)
B 74 (33.6)
C 74 (33.6)
D 22 (10.0)
JIS, n (%)
0 6 (2.7)
1 52 (23.6)
2 81 (36.8)
3 56 (25.5)
4 22 (10.0)
5 3 (1.4)
PSJIS, n (%)
0 1 (0.5)
1 12 (5.5)
2 44 (20.0)
3 76 (34.5)
4 46 (20.9)
5 27 (12.3)
6 12 (5.5)
7 2 (0.9)
AST, aspartate aminotransferase; BCLC, Barcelona Clinic Liver Cancer; HAP, hepatoma arterial embolization prognostic score; HBsAg, hepatitis virus B surface antigen; INR, international normalized ratio; MESH, model to estimate survival for hepatocellular carcinoma; mHAP, modified HAP; PSJIS, performance status combined Japan Integrated Staging system; PT, prothrombin time; TNM, tumor-node-metastasis.
######
Pairwise comparison of receiver operating characteristic curves predicting 3-month survival rates.
System 1 System 2 Difference P-value
---------- ---------- ------------ ----------
MESH HAP 0.130 0.0008
MESH mHAP 0.168 0.0001
MESH PSJIS 0.170 0.0002
MESH TNM 0.159 \<0.0001
HAP mHAP 0.0373 0.1839
HAP PSJIS 0.0395 0.4660
HAP TNM 0.0283 0.5530
mHAP PSJIS 0.00218 0.9706
mHAP TNM 0.00902 0.8639
PSJIS TNM 0.0112 0.8479
HAP, hepatoma arterial embolization prognostic score; MESH, model to estimate survival for hepatocellular carcinoma; mHAP, modified HAP; PSJIS, performance status combined Japan Integrated Staging system; TNM, tumor-node-metastasis.
######
Pairwise comparison of receiver operating characteristic curves predicting 6-month survival rates.
System 1 System 2 Difference P-value
---------- ---------- ------------ ---------
MESH HAP 0.0752 0.0076
MESH mHAP 0.102 0.0016
MESH PSJIS 0.0999 0.0032
MESH TNM 0.107 0.0002
HAP mHAP 0.0266 0.2054
HAP PSJIS 0.0247 0.5311
HAP TNM 0.0320 0.3498
mHAP PSJIS 0.00184 0.9639
mHAP TNM 0.00541 0.8813
PSJIS TNM 0.00725 0.8446
HAP, hepatoma arterial embolization prognostic score; MESH, model to estimate survival for hepatocellular carcinoma; mHAP, modified HAP; PSJIS, performance status combined Japan Integrated Staging system; TNM, tumor-node-metastasis.
######
Pairwise comparison of receiver operating characteristic curves predicting 1-year survival rates.
System 1 System 2 Difference P-value
---------- ---------- ------------ ---------
MESH HAP 0.0610 0.0318
MESH mHAP 0.0525 0.0797
MESH PSJIS 0.0796 0.00116
MESH TNM 0.0704 0.0189
HAP mHAP 0.00848 0.6508
HAP PSJIS 0.0186 0.6167
HAP TNM 0.00943 0.7866
mHAP PSJIS 0.0271 0.4793
mHAP TNM 0.0179 0.6189
PSJIS TNM 0.00917 0.7800
HAP, hepatoma arterial embolization prognostic score; MESH, model to estimate survival for hepatocellular carcinoma; mHAP, modified HAP; PSJIS, performance status combined Japan Integrated Staging system; TNM, tumor-node-metastasis.
######
Homogeneity LRT χ^2^ test and AIC of different staging systems.
Staging system Homogeneity LRT χ^2^ test AIC P-value
---------------- --------------------------- ------ ---------
MESH 31 1339 \<0.01
PSJIS 19 1354 \<0.01
TNM 16 1354 \<0.01
mHAP 14 1358 \<0.01
HAP 11 1361 \<0.01
AIC, Akaike information criterion; HAP, hepatoma arterial embolization prognostic score; LRT, likelihood ratio test; MESH, model to estimate survival for hepatocellular carcinoma; mHAP, modified HAP; PSJIS, performance status combined Japan Integrated Staging system; TNM, tumor-node-metastasis.
######
Univariate and multivariate analysis of prognostic factors for overall survival in 220 patients with hepatocellular carcinoma undergoing trans-arterial chemoembolization.
Univariate analysis Multivariate analysis
------------------------------------------- --------------------- ----------------------- -------- ------ -------------- ------
Sex (male/female) 0.77 0.41 to 1.46 0.43
Age, years (\>53/≤53) 0.95 0.68 to 1.32 0.75
Tumor size (\>50% of liver/≤50% of liver) 2.07 1.48 to 2.90 \<0.01 1.64 1.11 to 2.42 0.01
Node status (N0/N1) 1.99 1.27 to 3.11 0.01 1.24 0.75 to 2.04 0.40
Extrahepatic spread (yes/no) 1.87 1.11 to 3.14 0.02 1.20 0.67 to 2.14 0.54
Portal vein invasion (segmental; yes/no) 2.12 1.52 to 2.97 \<0.01 1.64 1.13 to 2.37 0.01
AFP (\>400/≤400 ng/ml) 1.41 1.01 to 1.97 0.04 1.03 0.73 to 1.48 0.85
Child-Pugh grade (A/B) 1.32 0.93 to 1.88 0.12
Antiviral therapy (yes/no) 0.64 0.45 to 0.89 \<0.01 0.70 0.49 to 1.01 0.05
AST (\>40/≤40 U/l) 0.58 0.39 to 0.86 \<0.01 0.75 0.50 to 1.13 0.18
Bilirubin (\>51.3/≤51.3 µmol/l) 2.22 1.09 to 4.54 0.03 2.54 1.21 to 5.36 0.01
AFP, α-fetoprotein; AST, aspartate transaminase; HR, hazard ratio; CI, confidence interval.
[^1]: Contributed equally
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction {#sec1}
===============
Today\'s technological designs combine innovation with sustainability, and focus in the overall transportation industry not just on alternative fuel sources but also on lightweight construction. The idea of weight reduction of structural components describes lower fuel consumption and therefore, lower CO~2~ emissions. One of the most promising materials for lightweight designs are carbon fiber reinforced plastics. Carbon fibers show impressive properties in terms of low density and high strength \[[@bib1], [@bib2], [@bib3], [@bib4]\] and are therefore, a popular choice for lightweight construction in industries such as aerospace \[[@bib5], [@bib6]\], automotive \[[@bib7], [@bib8], [@bib9]\], construction \[[@bib10], [@bib11], [@bib12]\] and sports \[[@bib13], [@bib14], [@bib15]\].
The task distribution within carbon fiber reinforced plastics (CFRP) is defined clearly: the carbon fibers absorb mechanical loads and the plastic forms the composite\'s shape and secures the fibers\' position within the hybrid structure. CFRP composites combine two rigid and brittle materials to a hybrid structure, which is seemingly less rigid and brittle than its single structures. This reinforcement effect is caused by the structure of the composite and its influence on crack behavior and load distribution \[[@bib1]\]. A crack can grow unhindered in a pure material, but in a fiber reinforced material, the crack propagation stops at every single fiber and the mechanical load is then being redistributed to several neighboring fibers. More specifically, a crack does not break a fiber immediately, but has a notch effect on it. Depending on the severity of the induced stress on the fiber, the crack propagation stops or the fiber breaks and the crack continues growing until it reaches the next fiber. This failure behavior has two advantages: first, even if a few fibers break, it does not mean, that the entire composite fails; and second, the crack propagation is slowed down, so that even brittle matrices do not fail abruptly \[[@bib16]\].
The visible component separation between CFRP laminates is called delamination. The detachment of fibers and matrix can be caused by brittleness and rigidity of the single structures, low adhesive forces between fiber and matrix, external high-impact loads and high internal stress in the composite. The fiber-matrix interface has direct influence on the energy transfer between both components: the stronger the adhesion between them, the more energy is necessary for their separation. This means, strong adhesive forces between fiber and matrix lower the speed of crack propagation and lower the possibility of fiber-matrix separation and delamination.
The improvement of interfacial fiber-matrix adhesion is promised by sizing of carbon fibers. This process describes a coating of the carbon fiber with a polymer. The sizing protects the carbon fibers from environmental influences and stress due to transportation and processing. However, the use of polymers as sizing agents grants the incorporation of the carbon fiber surface into the polymeric network of the final CFRP composite. Meaning that, the new created interphase can be tailored precisely to the adhesive requirements of fiber and plastic. Typical sizing agents are epoxy resins \[[@bib17], [@bib18], [@bib19]\], polyurethane \[[@bib19], [@bib20], [@bib21]\], vinyl ester resins \[[@bib22], [@bib23]\], polyamide \[[@bib19], [@bib21]\], polyimide \[[@bib21]\], acrylic acid \[[@bib24], [@bib25]\], polymethylmethacrylate \[[@bib26]\] and polystyrene \[[@bib27]\]. Next to polymer coatings, particle sizing is possible and introduces materials like nickel particles \[[@bib28]\], carbon nanotubes \[[@bib29], [@bib30], [@bib31], [@bib32], [@bib33], [@bib34]\] and graphene oxide \[[@bib35], [@bib36]\]. Interesting approaches combine different sizings, e.g. carbon nanotubes with silane coating and zinc oxide nanorods \[[@bib37]\], incorporate sensors and damage detection or self-healing mechanisms \[[@bib38], [@bib39], [@bib40]\].
In this study, β-cyclodextrin was covalently bonded on carbon fibers and its influence through host-guest complex formation was analyzed. Cyclodextrins are cyclic oligosaccharides and can form complexes with aromatic compounds ([Figure 1](#fig1){ref-type="fig"}). They have a hydrophilic exterior and a hydrophobic interior, which makes them soluble in water and able to incorporate hydrophobic compounds into their cavity \[[@bib41], [@bib42]\]. Since they can act as host molecules for aromatic compounds like the monomers of epoxy resins, polymerizations can be performed with location specifity. The complex formation was performed between the cyclodextrin, which was bonded covalently on the carbon fibers, and the epoxy resin system.Figure 1(a) Molecular structure of β-cyclodextrin. (b) Schematic illustration of cyclodextrin conus and (c) inclusion complex of aromatic compound.Figure 1
2. Material and methods {#sec2}
=======================
2.1. Materials {#sec2.1}
--------------
β-Cyclodextrin (Sigma Aldrich, 98 %) was grafted onto PAN-based, high strength, unsized carbon fibers (Sigrafil C30 T050 Uns SGL Carbon) and its reactivity was improved by p-toluenesulfonyl chloride (ABCR, 98 %). The carbon fiber surface was functionalized beforehand by t etraethylenepentamine (Merck, ≥ 95 %).
For the preparation of the pull-out tests, a solvent-free, liquid epoxy resin BECKOPOX EP140 and EH637 by Allnex was used. The samples were cured for two weeks at room temperature.
2.2. Oxidation and coating procedure {#sec2.2}
------------------------------------
Coating of carbon fibers with cyclodextrin requires a four-step synthesis. First, the carbon fiber surface had to be activated through oxidation with nitric acid, following a treatment with tetraethylenepentamine for the introduction of amine groups. Since β-cyclodextrin has a low reactivity, it was tosylated with p-toluenesulfonyl chloride. This way, a good leaving group is available instead of the less reactive hydroxyl group. The coating process describes the reaction between aminated carbon fibers and tosylated β-cyclodextrin.
The oxidation of unsized carbon fibers was performed in concentrated nitric acid at 120 °C for 3 h (based on Pittman\'s instructions \[[@bib43], [@bib44]\]). Deactivation of nitrous gases was performed with a gas washing bottle filled with a saturated solution of iron sulfate. After the reaction was completed, the acid solution was poured onto ice and the oxidized fibers were washed with distilled water until a neutral pH was reached. For the amination, the oxidized carbon fibers were mixed into tetraethylenepentamine and heated to 200 °C for 4 h (based on Pittman\'s instructions \[[@bib43], [@bib44]\]). After the mixture cooled down, the fibers were washed with distilled water. The resulting fibers contain reactive amine groups on the fiber surface and can be used directly for the coating with tosylated β-cyclodextrin.
For every 0.1 g aminated carbon fibers, 1.14 g (1 mmol) β-cyclodextrin were tosylated (based on Fujita\'s instructions \[[@bib45]\]). β-cyclodextrin was dissolved in 50 ml water and 0.2 g (1 mmol) p-toluenesulfonyl chloride was added. The mixture was stirred at room temperature for 2 h and the resulting gas of hydrochloric acid had to be removed constantly. After the tosylation was completed, the mixture was poured into dichloromethane. The excess solvent was removed from the precipitated solid and the product was dissolved in dimethylformamide.
The coating process was performed in DMF at temperatures between 80 and 120 °C for 4 h. The coated fibers were then washed three times with 50 ml DMF and dried at 80 °C. Two additional samples were then prepared for the beforehand complex formation between the coated β-cyclodextrin and the aromatic parts of the resin system. One of the samples describes a beforehand complex formation between the coated β-cyclodextrin and the epoxy resin binder and for the other sample, the amine adduct was used for the complex formation. The beforehand complex formation was performed in water, where β-cyclodextrin coated carbon fibers and either the epoxy binder or the amine curing agent were added. The mixture was stirred at room temperature for 24 h and later on dried at 80 °C.
2.3. Characterization: scanning electron microscopy {#sec2.3}
---------------------------------------------------
The topography and elemental composition of the carbon fiber surface were analyzed with scanning electron microscopy (SEM) with a Zeiss Neon 40 including energy-dispersive X-ray spectroscopy (EDX). The carbon fibers were glued onto aluminium stubs with an electrically conductive adhesive. To obtain electric conductivity, the coated fibers were sputtered with a film thickness of 3 nm. SEM was performed at a working distances of 7 mm and an accelerating voltage of 2 kV. The EDX measurements were performed at an acceleration voltage of 5 kV and a working distance of 5 mm.
2.4. Contact angle analysis {#sec2.4}
---------------------------
A Krüss Tensiometer K100SF was used to perform contact angle measurements of single carbon fibers. Using an adapted Wilhelmy method, the measurements provide information about surface energy, polarity and wettability.
2.5. Characterization: pull-out tests {#sec2.5}
-------------------------------------
A test machine MTS Criterion C45.105 with a HBM load cell S2M (10 N) was used to perform the micromechanical analysis. Single carbon fibers were embedded in epoxy resin droplets and pulled out after curing of the resin. The measurements were performed at room temperature, with speeds of 0.05 mm/min and free fiber lengths of 10 mm. The detailed sample preparation and test procedure were published previously \[[@bib46]\]. Pull-out tests are used to determine the bond strength between fibers and resin and deliver quantitative results concerning chemical and physical interactions in the fiber-resin interface. The IFSS is defined as force per interface area ([Eq. (1)](#fd1){ref-type="disp-formula"}) and describes the adhesion between carbon fiber and matrix.$$IFSS = \frac{F_{max}}{\pi \cdot l \cdot d}$$•IFSS interfacial shear strength•F~max~ recorded maximal force•l embedded length of the fiber•d diameter of the fiber
3. Results and discussion {#sec3}
=========================
The β-cyclodextrin coating on carbon fibers was visually analyzed via scanning electron microscopy, including energy-dispersive x-ray spectroscopy. The coating can be observed via SEM, as seen in [Figure 2](#fig2){ref-type="fig"}. Even though the coating is not homogenous, the covered area is high.Figure 2Exemplary SEM images of (a) unsized carbon fibers, and (b) & (c) covalently bonded β-cyclodextrin on carbon fibers.Figure 2
The chemical composition was analyzed with energy-dispersive x-ray spectroscopy and is listed in [Table 1](#tbl1){ref-type="table"}. The Kα lines for carbon is at 0.277 keV, for nitrogen at 0.392 keV, for oxygen at 0.525 keV and for sulfur at 2.307 keV. The reactivity of the carbon fiber surface was improved by a treatment with tetraethylenepentamine; therefore, an increase in nitrogen content was detected. The following coating with β-cyclodextrin increased the oxygen content. P-toluenesulfonyl chloride was used as a coupling agent, since it describes a good leaving group and the primary hydroxyl groups of β-cyclodextrin are low in reactivity. Sulfur was detected through EDX measurements and leads to the conclusion, that even though the reaction was set up stoichiometric, some β-cyclodextrin molecules got tosylated multiple times. The maximum amount of tosylations of β-cyclodextrin is seven times.Table 1Element composition determined by EDX measurements. \[∗ - calculated without hydrogen and chloride content\].Table 1sampleC \[at%\]N \[at%\]O \[at%\]S \[at%\]carbon fiber85.06.98.1cyclodextrin coated carbon fiber68.011.318.02.7cyclodextrin ∗54.545.5tetraethylenepentamine ∗61.538.5p-toluenesulfonyl chloride ∗70.020.010.0
Furthermore, contact angle measurement of single β-cyclodextrin coated fibers were performed. Coating the fiber surface changes the chemical composition of the surface and leads to changes in polarity. [Table 2](#tbl2){ref-type="table"} summarizes the advancing contact angles. Unsized carbon fibers contain mainly carbon, but also have a few oxygen and nitrogen containing functions on the surface. The expected wettability \[[@bib47], [@bib48], [@bib49]\] matches the measured contact angle of 89°. The coating process of the carbon fiber surface with β-cyclodextrin increases the hydrophilicity and therefore, decreases the contact angle, due to the hydrophilic exterior of β-cyclodextrin. The complex formation with both, epoxy binder and amine curing agent, was detected with contact angle measurements, since just the aromatic parts are part of the inclusion complex and the remaining parts of the molecules stick out and increase the hydrophobicity.Table 2Comparison of contact angles of carbon fibers and β-cyclodextrin coated carbon fibers.Table 2sampleθ \[°\]carbon fiber88.8 ± 7.3β-cyclodextrin coated carbon fiber54.5 ± 3.1β-cyclodextrin complex with epoxy resin coated carbon fiber74.2 ± 4.8β-cyclodextrin complex with amine curing agent coated carbon fiber71.6 ± 16.8
The coating with β-cyclodextrin increases the interfacial shear strength between carbon fibers and epoxy resin drastically ([Figure 3](#fig3){ref-type="fig"}). Unsized carbon fibers have interfacial shear strengths of about 12 ± 0.6 MPa, while β-cyclodextrin coated carbon fibers reach IFSS values of 37.6 ± 1.8 MPa. Using prior complex formation with the epoxy binder or the amine coupling agent, increases the IFSS even further, to 41.0 ± 1.9 MPa. Meaning, the beforehand complex formation has a positive influence on the increase of the interfacial shear strength. These high values in adhesion cannot be explained by just functional groups, since oxidized carbon fibers (compared to the epoxy binder complex) and aminated carbon fibers (compared to the amine coupling agent complex) show IFSS values of 16.2 ± 1.1 MPa and 21.3 ± 2.8 MPa. Meaning, the bonding mechanism and complex formation have a significant impact on the adhesion between fiber and resin.Figure 3Pull-out tests of β-cyclodextrin coated carbon fibers. (a) Summary of the interfacial shear strength of fibers without and with a beforehand complex formation. (b) Summary of the measurement series of β-cyclodextrin coated carbon fibers without beforehand complex formation. (c) Summary of the measurement series of β-cyclodextrin coated carbon fibers with beforehand complex formation with epoxy resin and (d) summary of the measurement series of β-cyclodextrin coated carbon fibers with beforehand complex formation with amine coupling agent.Figure 3
4. Summary {#sec4}
==========
The interface of CFRPs was designed with host-guest complex formation by the successful use of β-cyclodextrin. The complex formation was done with the inclusion of either the epoxy binder or the amine curing agent, since both components are partly aromatic.
The ability of β-cyclodextrin to include aromatic guest molecules in its cavity has a significant impact on the interfacial shear strength of carbon fiber reinforced plastics. The complex formation itself lead to an increase of the fiber-resin adhesion from IFSS values of 12 MPa--38 MPa. It was possible to increase the interfacial shear strength even further, to 41 MPa, through a prior complex formation with one of the plastics components.
It can be stated, that the complex formation of β-cyclodextrin and the plastic components in the interface of CFRPs has a huge impact on the improvement of the interfacial shear strength. At this, an emergence can be observed, since the increase in adhesion cannot be explained by the covalent use of the single components.
Declarations {#sec5}
============
Author contribution statement {#sec5.1}
-----------------------------
Anna Becker-Staines: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.
Wolfgang Bremser, Thomas Tröster: Conceived and designed the experiments; Contributed reagents, materials, analysis tools or data.
Funding statement {#sec5.2}
-----------------
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interest statement {#sec5.3}
----------------------------
The authors declare no conflict of interest.
Additional information {#sec5.4}
----------------------
No additional information is available for this paper.
| {
"pile_set_name": "PubMed Central"
} |
Background {#Sec1}
==========
Antarctica is a unique geographic region that experiences extreme environmental conditions and ongoing threats from a changing climate, with some of the fastest warming waters globally (e.g. \[[@CR1], [@CR2]\]). Historically, repeated glacial cycles and fragmentation of populations in the region have promoted speciation in many groups (e.g. \[[@CR3], [@CR4]\]) and as a result Antarctica and the surrounding sub-Antarctic islands boast a rich diversity of marine life (e.g. \[[@CR5], [@CR6]\]). In fact, recent work has identified the Antarctic Peninsula and sub-Antarctic islands as centres of marine diversity for notothenoid fishes \[[@CR7]\] as well as marine invertebrates, including pycnogonids \[[@CR8]\] and sponges \[[@CR9]\]. Some benthic marine invertebrate groups are more well-represented than others in Antarctica, including echinoderms, which are abundant and diverse on the continental shelf (e.g. \[[@CR10], [@CR11]\]). Echinoderms are also important as hosts for parasites and commensal organisms, but although these echinoderm-parasite interactions have been studied in temperate and tropical systems, they have been rarely documented and studied in Antarctica (but see \[[@CR12], [@CR13]\]).
*Asterophila* \[[@CR14]\], a genus of eulimid gastropod, is exclusively endoparasitic in asteroid sea stars. There are currently three described members in this genus, *A. japonica* \[[@CR14]\] from the NorthWest Pacific, *A. perknasteri* \[[@CR15]\] from Antarctica, *A. rathbunasteri* \[[@CR15]\] from California, as well as a fourth undescribed species from the Kermadec Trench \[[@CR15]\]. The level of host specificity varies between congenerics, but may reflect a lack of knowledge of host associations in this genus. For instance, *A. japonica* has been found in eight host species from four families and three orders \[[@CR16]\] while *A. perknasteri* has been found from three different *Perknaster* species \[[@CR15]\] and *A. rathbunasteri* has been found only from a single host (*Rathbunaster californicus*) \[[@CR15]\]. Because these parasites display a highly simplified body plan for their endoparasitic lifestyle, only a few morphological characters have been shown to be informative between species \[[@CR16]\]. Additionally, no prior study has employed molecular data for exploring phylogenetic relationships in this genus, thus much remains unknown. In fact, little is known about Antarctic eulimids in general, despite a potentially high diversity of species in the region \[[@CR12]\].
Parasites comprise a large portion of global biodiversity and they may be important in driving allopatric diversification in their hosts \[[@CR17], [@CR18]\]. Additionally, when speciation between host and parasite is tightly coupled, a loss of the host lineage may result in a loss of associated species (co-extinction) \[[@CR19]\], thereby having significant impacts on total biodiversity loss \[[@CR20]\]. The close association between endoparasitic *Asterophila* and asteroid hosts lends itself to exploring co-phylogenetic patterns and determining whether these groups are evolving in synchrony and which coevolutionary events are promoting diversification in this system. Early studies of coevolutionary biology found evidence for strict cospeciation between host and parasite (e.g. gophers and lice \[[@CR21]\]) and it is thought that cospeciation plays a particularly important role in endoparasitic taxa. However, recent work has challenged this idea by showing that host-switching and duplication events are more prevalent than cospeciation in other host-parasite systems (e.g. myzostomid worms and crinoids \[[@CR22]\], nematodes and stick insects \[[@CR23]\]), and some endoparasites show host promiscuity (e.g. copepods and nudibranchs \[[@CR24]\]). Although echinoderm-parasitizing eulimid gastropods are diverse and exhibit a variety of parasitic lifestyles \[[@CR25], [@CR26]\], most prior work has focused only on the morphology and ecology of this clade and little is known about the evolution of parasitism in the group. Moreover, recent work has shown that parasitism and symbiosis are more common in Antarctica than once thought (e.g. \[[@CR12], [@CR13]\]).
Recent sampling in Antarctica uncovered a range of seastar hosts from different taxonomic groups that were parasitized by *Asterophila,* prompting us to explore diversity in this genus and investigate cophylogenetic patterns between parasites and hosts. This study employs molecular and morphological data for phylogenetic analysis and species delimitation. These phylogenetic methods shed light on an apparent radiation of *Asterophila* in Antarctica and aid in determining whether speciation in asteroid hosts has been paralleled by *Asterophila*. This work extends our knowledge of diversity in *Asterophila* and provides insight into the coevolutionary events promoting diversification in this system.
Results {#Sec2}
=======
*Asterophila* occurrence and distribution {#Sec3}
-----------------------------------------
A total of 61 *Asterophila* parasites were recovered from 38 hosts forming large and noticeable cysts on the aboral side of the host in the arms or near the central disc. *Asterophila* was found between the epidermis and coelomic lining, with no affiliation to specific organs or body systems. This description is congruent with Sasaki et al. \[[@CR16]\] who describe the anatomy of *Asterophila japonica* in detail. Roughly 1500 potential hosts were collected in trawls on each cruise, and therefore an occurrence of only 38 infected hosts suggests the prevalence of parasitism in this group is low, though it may have been that only obvious infections were noted. Infected hosts and parasites were only recovered from near the Antarctic Peninsula, at Shetland Islands, Elephant Island, Shag Rocks and South Georgia (Fig. [1](#Fig1){ref-type="fig"}), despite sampling at various locations around the continent. An additional host with parasite (BIC-SIO E5030) was discovered and imaged from South Orkney in 2011, but the specimen was not included in these analyses. Fig. 1*Asterophila* distribution in Antarctica. **a** Host specimens infected by *Asterophila* (left: P00407, right: P00340)- white circles indicate cysts on the asteroid arm. **b** *Asterophila* specimens retrieved from hosts (left: M13618, right: M12882). **c** Localities marked with a square denote locations where *Asterophila* was present, and localities marked with a circle represent locations where hosts were sampled but *Asterophila* was absent (Wikimedia Commons contributors). **d** Detailed sampling map showing co-distribution of *Asterophila* species at South Shetland Islands created with the R package *ggplot2* \[[@CR27]\] and with bathymetry data from the National Oceanic and Atmospheric Administration (NOAA) in the R package *marmap* \[[@CR28]\]
Phylogenetic reconstruction and species delimitation {#Sec4}
----------------------------------------------------
A total of 58 COI, 41 16S, 49 H3, 38 28S, and 14 ANTsequences were recovered from 61 *Asterophila* specimens and concatenated for phylogenetic analysis with a final alignment of 4150 bp (COI: 658 bp; 16S: 455 bp; H3: 328 bp; ANT: 499 bp; 28S: 2210 bp) and 477 variable sites. The COI, 16S, and ANT markers were more variable among species than H3 and 28S. Despite some instances of low bootstrap support at interior nodes, there was still strong support for nine putative species-level entities within *Asterophila* (BS 93--100), forming the PSH (Additional file [1](#MOESM1){ref-type="media"}: Figure S1). Each of these nine putative species form highly supported least-inclusive monophyletic clades in the phylogeny and were found in different host families, further supporting their status as separate species. The ML results of the bPTP analysis partitioned *Asterophila* into 11 species, while the initial partition of ABGD delimited 8 species, suggesting over-splitting in the former and merging in the later (Fig. [2](#Fig2){ref-type="fig"}). However, the two methods were congruent regarding all but two lineages (sp. 4 and 5). Because each of the clades identified in the PSH were included in one of the species delimitation algorithms, the SSH (Fig. [2](#Fig2){ref-type="fig"}) remains the same as the PSH (Additional file [2](#MOESM2){ref-type="media"}: Figure S2). One species was identified as *A. perknasteri*, currently known from Antarctica, and the remaining eight were assigned interim names (i.e. *A.* sp. 1- *A.* sp. 8)(Fig. [2](#Fig2){ref-type="fig"}). Some sister-species relationships within *Asterophila* were resolved, including *A. perknasteri* and *A.* sp. 8 from the asteroid order Valvatida and *A.* sp. 3, *A.* sp. 4 and *A.* sp. 5 from Forcipulatida, although the position of *A.* sp. 7, also from Forcipulatida, was not well supported (Fig. [2](#Fig2){ref-type="fig"}). The positions of *A.* sp. 1 and *A.* sp. 2 were also not well supported in this analysis. Interspecific distances for COI ranged from a minimum of 1.9% between *A.* sp. 4 and *A.* sp. 5 and a maximum of 18.3% between *A.* sp. 1 and *A.* sp. 2 (mean = 13.9%), while intraspecific distances ranged from 0 to 0.9%. An ancestral state reconstruction suggests that a Forcipulatida asteroid is the ancestral host for *Asterophila* (likelihood 68), with one transition to each of Paxillosida, Spinulosida, Valvatida and Velatida (Additional file [2](#MOESM2){ref-type="media"}: Figure S2), although a fully resolved phylogeny with additional genes would aid in clarifying these patterns. Fig. 2Secondary species hypothesis. ML phylogeny (COI + 16S + H3 + 28S + ANT) of *Asterophila* representing the secondary species hypothesis, with host order-level taxonomy provided in red. Nodes with less than 50% bootstrap support have been collapsed, and triangles represent single clades. Hash marks denote that the branch has been truncated to one quarter of its original length. For species-delimitation analysis (ABGD and bPTP) black boxes represent congruent clades, red boxes represent splits within a clade, and blue boxes represent merged clades
This study presents COI sequences for 35 out of 38 individual asteroid hosts with a final alignment of 700 bp. Three hosts missing molecular data were excluded from cophylogenetic analyses, including two *Labidiaster annulatus* and one *Perknaster.* Because there are currently seven species of *Perknaster* known from Antarctic waters (<http://www.scarmarbin.be>) the species-level identity of this host could not be validated. In all, we report *Asterophila* from five orders, seven families, ten genera, and 15 species of asteroid sea star.
Veliger larvae were recovered from individuals of three *Asterophila* species (*A. perknasteri*, *A.* sp. 3, *A.* sp. 5) and a total of nine larvae were examined and compared to larval morphologies in the existing literature (Fig. [3](#Fig3){ref-type="fig"}) (e.g. \[[@CR15], [@CR16]\]). From initial observation of SEM images, the larval shells of *A.* sp. 3 and *A.* sp. 5 appear more depressed than the globose shape of *A. perknasteri*. The size of the larval shell was significantly different among species (*P* = 0.002) and measured 690--720 μm in *A. perknasteri*, 650--690 μm in *A.* sp. 3, and 540--600 μm in *A.* sp. 5. The shell width reported for *A. perknasteri* in this study corresponds to the size range reported in the original description of this species (650--760 μm) \[[@CR15]\]. Moreover, the umbilicus of *A. perknasteri* appears slightly deeper than both *A.* sp. 3 and *A.* sp. 5, although this character is not mentioned in the existing *Asterophila* descriptions. The larval shells of all species examined in this study appear smooth, with no sculpturing. Fig. 3*Asterophila* larvae. Larval morphologies of **a** *Asterophila japonica*, *A. rathbunasteri* and *A. perknasteri* from the literature (adapted from \[[@CR15], [@CR16]\] with permission) and **b** *A.* sp. 3, *A.* sp. 5 and *A. perknasteri* from this study, with scale bar
Host specificity and partitioning {#Sec5}
---------------------------------
*Asterophila* was discovered in 38 hosts from a wide taxonomic breadth. Phylogenetic conservatism in host-use and host-specificity was observed in this system, with eight putative *Asterophila* species each discovered from a different host genus and one species discovered from two host genera (Fig. [4](#Fig4){ref-type="fig"}a; Additional file [2](#MOESM2){ref-type="media"}: Figure S2). Four species of *Asterophila* (*A.* sp. 3, *A.* sp. 6, *A.* sp. 8, *A. perknasteri*) were found on multiple hosts, but these hosts were always closely related congeners (i.e. *Lysasterias heteractis*, *Lysasterias perrieri*) or confamilials (i.e. *Lophaster gaini*, *Paralophaster antarcticus*) (Fig. [4](#Fig4){ref-type="fig"}a), confirmed by significantly lower cophenetic distances between these host pairs than between all other hosts (*P* \< 0.0001). Those species found parasitizing multiple hosts also had some of the largest sample sizes (*N* = 7 for *A.* sp. 3, *N* = 7 for *A.* sp. 6, *N* = 4 for *A.* sp. 8, *N* = 20 for *A. perknasteri*). The haplotype networks show some level of host partitioning for *A.* sp. 6 and *A. perknasteri,* demonstrated by a lack of haplotype sharing between individuals from different hosts and some clustering of haplotypes based on host species (Fig. [4](#Fig4){ref-type="fig"}b). Individual hosts were only ever parasitized by a single *Asterophila* species. Fig. 4Host use by *Asterophila*. **a** ML phylogeny (COI + 16S + H3 + 28S + ANT) of *Asterophila* with host species designations. Nodes with less than 50% bootstrap support have been collapsed, and triangles represent single clades. Hash marks denote that the branch has been truncated to one quarter of its original length. *Asterophila* species in bold have been recovered from multiple host species. **b** Corresponding TCS haplotype networks (COI) for *Asterophila* species recovered from multiple hosts. Hash marks on the haplotype network correspond to mutational steps and circle size represents the number of sequences per haplotype. Colours in the haplotype network correspond to different host species
Cophylogenetic analysis {#Sec6}
-----------------------
A total of 57 *Asterophila* and 35 asteroid hosts were included in the cophylogenetic analysis, with the resulting cophylogenetic plot showing 57 individual associations (Fig. [5](#Fig5){ref-type="fig"}). Three specimens of *A.* sp. 5 from *Labidiaster annulatus* and one specimen of *A. perknasteri* from *Perknaster* were excluded from the cophylogenetic analysis because sequence data was unavailable for the host. The event-based analysis in Jane was conducted with four different cost regimes (Table [1](#Tab1){ref-type="table"}), and in all cases the number of non-cospeciation events was greater than the number of cospeciation events, with the number of host-switching events particularly high in the equal costs regime and when duplication and host switches were assigned values of 0. The total cost using 50 random parasite trees was significantly greater than the costs recovered for two of the regimes using the real dataset (*P* \< 0.0001) (Table [1](#Tab1){ref-type="table"}), suggesting a strong signal of coevolution in this system. There was no variation in costs using the random parasite trees or the real dataset for the equal costs regime and when duplications and host switches were assigned values of 0. The distance-based analysis in ParaFit revealed that a global test of coevolution was strongly significant (*P* = 0.0001), rejecting the null hypothesis of random association (i.e. that hosts and parasites are evolving independently). A test of coevolution for the 57 associations with a Bonferroni correction revealed a total of 44 significant links (*P* \< 0.05) (Table [2](#Tab2){ref-type="table"}). Fig. 5Host-parasite associations. Cophylo plot of 57 host-parasite relationships between *Asterophila* and asteroid hosts. Branches and associations are coloured based on host order. Host colours match ancestral state reconstruction in Additional file [2](#MOESM2){ref-type="media"}: Figure S2 Table 1Results of event-based reconciliation analysis in JaneRegimeCospeciationDuplicationHost-switch & duplicationLossFailure*p*-value111110134300n/a0121118731100\< 0.0001112116183210\< 0.0001100110134300n/aThe total number and cost is provided for each coevolutionary event. The total number of cospeciation and non-cospeciation events is also provided. *P*-values were obtained by comparing total cost of the host/parasite phylogenies to the total cost with a random parasite tree Table 2Results of distance-based analysis in ParaFitHost SpeciesParasite SpeciesF1 statistic*p*-value*Psilaster charcoti* P00408*Asterophila* sp. 1 P004080.00020.161*Rhopiella hirsuta* P00029*Asterophila* sp. 3 M13022− 0.00080.489*Rhopiella hirsuta* P00409*Asterophila* sp. 3 M13018− 0.00080.489*Rhopiella* sp. 1 P00376*Asterophila* sp. 3 M13021b−0.00080.488*Rhopiella* sp. 1 P00376*Asterophila* sp. 3 M13021c−0.00080.487*Rhopiella* sp. 1 P00376*Asterophila* sp. 3 M13021d−0.00080.488*Rhopiella* sp. 1 P00376*Asterophila* sp. 3 M13021e−0.00080.489*Rhopiella* sp. 1 P00376*Asterophila* sp. 3 M13021a−0.00080.489***Lophaster gaini*** **P00387*Asterophila*** **sp. 8 M13043b0.00100.022***Lophaster gaini* P00387*Asterophila* sp. 8 M13043a−0.00130.500*Lophaster gaini* P00417*Asterophila* sp. 8 M13028−0.00150.500*Paralophaster antarcticus* P00422*Asterophila* sp. 8 M13027−0.00140.500***Perknaster*** **sp. 1 P00423*Asterophila perknasteri*** **M130300.00360.0002*Perknaster antarcticus*** **P00406*Asterophila perknasteri*** **M13039b0.00290.0003*Perknaster antarcticus*** **P00406*Asterophila perknasteri*** **M13039a0.00290.0002*Perknaster antarcticus*** **P00406*Asterophila perknasteri*** **M13039d0.00290.0002*Perknaster antarcticus*** **P00406*Asterophila perknasteri*** **M13039c0.00290.0003*Perknaster aurorae*** **P00407*Asterophila perknasteri*** **M130310.00300.0004*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019f0.00070.011*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019e0.00070.011*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019 g0.00180.007*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019b0.00160.001*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019a0.00160.001*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019d0.00150.001*Perknaster*** **cf.** ***densus*** **P00384*Asterophila perknasteri*** **M13019c0.00150.001*Perknaster*** **cf.** ***densus*** **P00472*Asterophila perknasteri*** **P004720.00210.0004*Perknaster*** **cf.** ***densus*** **P00365*Asterophila perknasteri*** **M130240.00210.0003*Perknaster*** **cf.** ***densus*** **P00364*Asterophila perknasteri*** **M130200.00110.043*Perknaster*** **cf.** ***densus*** **P00363*Asterophila perknasteri*** **M13033b0.00080.034*Perknaster*** **cf.** ***densus*** **P00363*Asterophila perknasteri*** **M13033a0.00090.029*Perknaster*** **cf.** ***densus*** **P00373*Asterophila perknasteri*** **M130230.00090.028*Pteraster*** **sp. 1 P00419*Asterophila*** **sp. 2 M130370.00060.045*Pteraster*** **sp. 1 P00420*Asterophila*** **sp. 2 M130350.00070.021*Lysasterias perrieri*** **P00354*Asterophila*** **sp. 6 M13048b0.00060.041*Lysasterias perrieri*** **P00354*Asterophila*** **sp. 6 M13048a0.00060.042*Lysasterias perrieri*** **P00347*Asterophila*** **sp. 6 M130460.00060.040*Lysasterias heteractis*** **P00362*Asterophila*** **sp. 6 M13047b0.00070.039*Lysasterias heteractis*** **P00362*Asterophila*** **sp. 6 M13047a0.00070.038*Lysasterias heteractis*** **P00361*Asterophila*** **sp. 6 M13038b0.00070.038*Lysasterias heteractis*** **P00361*Asterophila*** **sp. 6 M13038a0.00070.038*Diplasterias*** **sp. 1 P00527*Asterophila*** **sp. 4 M13042a0.00070.037*Diplasterias*** **sp. 1 P00527*Asterophila*** **sp. 4 M13042b0.00060.036*Notasterias*** **sp. 1 P00473*Asterophila*** **sp. 7 P004730.00100.009*Notasterias*** **sp. 1 P00393*Asterophila*** **sp. 7 M130450.00090.008*Notasterias*** **sp. 1 P00418*Asterophila*** **sp. 7 M130250.00100.007*Notasterias*** **sp. 1 P00003*Asterophila*** **sp. 7 M130320.00100.007*Notasterias*** **sp. 1 P00528*Asterophila*** **sp. 7 M130440.00100.008*Notasterias*** **sp. 1 P00526*Asterophila*** **sp. 7 M130360.00090.009*Notasterias*** **sp. 1 P00350*Asterophila*** **sp. 7 M130410.00100.008*Notasterias*** **sp. 1 P00348*Asterophila*** **sp. 7 M130400.00090.008*Notasterias*** **sp. 1 P00388*Asterophila*** **sp. 7 P003880.00090.011*Notasterias*** **sp. 1 P00355*Asterophila*** **sp. 7 M13054c0.00120.003*Notasterias*** **sp. 1 P00355*Asterophila*** **sp. 7 M13054d0.00120.003*Notasterias*** **sp. 1 P00355*Asterophila*** **sp. 7 M13054a0.00140.0003*Notasterias*** **sp. 1 P00355*Asterophila*** **sp. 7 M13054b0.00140.0003***Notasterias* sp. 1 E7112*Asterophila* sp. 7 M130340.00020.154*Labidiaster annulatus* E711*Asterophila* sp. 5 M136180.00020.156Tests of individual congruence between host and parasite and the corresponding *p*-value. Associations in bold are significant (*P* \< 0.05). Global congruence analysis in ParaFit was significant (ParaFitGlobal = 0.03; *P* = 0.0001)
Discussion {#Sec7}
==========
Phylogeny and diversification of *Asterophila* {#Sec8}
-----------------------------------------------
This study uncovers a radiation of *Asterophila* in Antarctica and increases the diversity of this group nearly tenfold, from one previously known species in the region (*A. perknasteri*) to a putative nine species. Although each of these putative nine species-level entities was strongly supported in the ML phylogeny, deeper relationships remain poorly supported. A lack of support in the *Asterophila* phylogeny, despite the use of both mitochondrial and nuclear markers that should be informative at both shallow and deep divergences, suggests that additional taxonomic and genomic sampling is required. Nevertheless, molecular data has been crucial for delimiting species boundaries in this genus. Each putative *Asterophila* species was recovered from a different host family, suggesting that host identity also serves as an informative character for species delimitation. The COI divergence between *A.* sp. 4 and *A.* sp. 5 is relatively low (1.9%) compared to interspecific distances for other molluscs (e.g. \[[@CR29], [@CR30]\]), but because these species were separated in the bPTP analysis and were recovered from different host families they are retained as different lineages in this study. Additional specimens suitable for molecular and morphological analysis would aid in clarifying these species boundaries. In all, this study reveals a novel diversification of endoparasites in Antarctica and extends our knowledge of diversity in *Asterophila*.
Previous work by Sasaki et al. \[[@CR16]\] used larval shell characters to diagnose three known species of *Asterophila*. These characters include an apical depression on the protoconch of *A. japonica*, rugose sculpture and a less depressed protoconch in *A. rathbunasteri*, and a large, smooth larval shell in *A. perknasteri*. The larval shells of two undescribed species examined in this study (*A.* sp. 3, *A.* sp. 5) appeared more similar in general shape to *A. perknasteri* than to either *A. japonica* or *A. rathbunasteri* but had a shallower umbilicus and appeared more depressed. Although larval shell size appeared to differ significantly between three Antarctic *Asterophila* species examined in this study, this variation in size may be related to differences in developmental stage and age of the larvae when they were retrieved from adult females. As such, there exists a deficit of informative morphological characters in this endoparasitic genus. Recent studies have also highlighted difficulties in using morphological characters for species delimitation in highly modified endoparasites (e.g. splanchnotrophid copepods \[[@CR24], [@CR31]\]), and this work further demonstrates that molecular data is critical for resolving species boundaries in endoparasites that often lack informative external characters.
Some benthic marine invertebrate groups have radiated extensively in Antarctica, including isopods \[[@CR32]\], dorid nudibranchs \[[@CR4], [@CR33]\], octopods \[[@CR34]\], and pycnogonids \[[@CR35]\], and repeated glacial cycles and the subsequent fragmentation of populations has been important in promoting speciation in these marine taxa \[[@CR3]\]. Glacial and interglacial cycles have presumably had an impact on the diversity and distribution of *Asterophila* in the region. The retreat of populations into small, closely distributed glacial refugia around Antarctica would have facilitated host-switching events by bringing multiple potential hosts into proximity, as well as by driving some hosts to extinction \[[@CR13]\]. It is also likely that the close coupling of endoparasite and host has been important in driving the diversification of this group.
Patterns of coevolution between *Asterophila* and asteroid hosts {#Sec9}
----------------------------------------------------------------
Coevolution and cospeciation are often used synonymously in the literature, but the former simply refers to reciprocal evolution between two or more species and can involve several processes, mainly cospeciation, duplication, host-switching, loss of a lineage, or failure of a parasite to diverge \[[@CR36]\]. Although a distance-based analysis in ParaFit sheds light on a signal of cospeciation in the system, the event-based analysis in Jane reconciles the specific events explaining the host-parasite associations. Both the event-based and distance-based analysis showed a global signal of coevolution, thereby rejecting the null hypothesis of independent evolution between host and parasite. A total of 30 links were significant in the ParaFit analysis (*P* \< 0.05), with an increase to 44 following a Bonferroni correction. These results suggest a strong signal of coevolution, likely driven by host-switching speciation events that have produced congruent phylogenies, but the parasite phylogeny was poorly supported and the host phylogeny was based on a single mitochondrial marker. The different cost regimes in Jane produced different numbers of events, most notably ranging from zero cospeciation events in the first regime to 18 cospeciation events in Jane's default regime, suggesting that some regimes under-detect cospeciation events and that choosing a regime a priori may prove problematic. In any case, most events were explained by duplication and host-switching, suggesting that non-cospeciation events are primarily driving the coevolution of *Asterophila* and their hosts.
Recent work has highlighted the importance of non-cospeciation events in the diversification of endoparasites (e.g. endoparasitic nematodes and stick insect hosts \[[@CR23]\]) and in highly specialized symbionts (e.g. feather mites and birds \[[@CR37], [@CR38]\]), demonstrating that these obligate host-parasite systems do not always follow strict cospeciation and synchronous evolution \[[@CR39]\]. Host-switching events are thought to be of central importance in driving parasite diversity (e.g. \[[@CR40], [@CR41]\]), and several studies have shown that host switches are particularly important in the diversification of Antarctic parasites, suggesting that repeated glacial cycles in the region fragmented populations and differentially removed potential hosts \[[@CR12], [@CR13]\]. Alternatively, the high level of host specificity and host switching observed in this system, and the sympatric distribution of most parasites and hosts, could indicate that parasites are able to diversify and adapt to new hosts in sympatry, rather than through allopatric speciation (e.g. \[[@CR24]\]).
Several additional factors may also have played a role in promoting host-switching in endoparasitic *Asterophila*, including the presence of a planktonic larval phase, which results in greater dispersal potential and the ability to encounter new hosts \[[@CR40]\], and the sympatric distribution of host species allowing for more host-switching opportunities (e.g. \[[@CR42]\]), with multiple hosts of *Asterophila* co-occurring in Antarctica. The mechanism by which *Asterophila* larvae locate and occupy a host is currently unknown, but chemical cues from hosts induce settlement of planktonic larvae in other molluscs (e.g. coral-associated nudibranchs \[[@CR43]\]) and are likely integral to this system. Furthermore, the high diversity of *Asterophila* species uncovered in this study may also be linked to host-switching, as the isolation of parasites on new hosts can promote their diversification (e.g. \[[@CR38], [@CR39], [@CR41]\]). In all, non-cospeciation events explained most of the associations between *Asterophila* and their hosts, highlighting the importance of these events even in endoparasitic lineages.
Most *Asterophila* species in this study were discovered from a single host genus, and always from a single host family, which differs from other endoparasites, including a species of splanchnotrophid copepod (*Lomanoticola brevipes*) that was recovered from five different nudibranch families \[[@CR24]\]. However, incongruence between the parasite and host phylogenies in this study implies that this specificity varies among species \[[@CR42]\]. For instance*, A.* sp. 1, 2, 4, 5, and 7 were all recovered from a single host species, while *A.* sp. 3, *A.* sp. 6, and *A. perknasteri* were recovered from two to four congeners, and *A.* sp. 8 was recovered from two genera in Solasteridae. The haplotype networks for *A.* sp. 6 and *A. perknasteri* show a complete lack of haplotype sharing between individuals from different hosts, indicating potential host partitioning and incipient host-shift speciation, but additional samples are needed to validate these patterns. In any case, multihost *Asterophila* species were recovered from phylogenetically-related hosts, as demonstrated by lower cophenetic distances between host pairs than between all other hosts, and these hosts often had the highest sample size, illustrating that the recovery of hosts, and ultimately parasite richness, may be linked to sampling effort (e.g. \[[@CR44], [@CR45]\]).
Poulin \[[@CR42]\] suggests that a pattern of host generalism is often shown in parasites that are recovered from i) species-rich host groups and ii) hosts with unstable or fluctuating populations, both of which are true of the Antarctic echinoderm fauna and may explain multihost use in *Asterophila*. Both the ML phylogeny and transformation suggest phylogenetic conservatism of host use, although further resolution of the *Asterophila* phylogeny, possibly through the use of a reduced representation library, will be crucial to better understanding these patterns. Phylogenetic conservatism of host use has been shown in other marine parasites, including in myzostomid annelids and their echinoderm hosts \[[@CR22]\]. An ancestral state reconstruction also suggests that a member of Forcipulatida was the ancestral host to *Asterophila*, but with only moderate support \[[@CR46]\]. Continued sampling will undoubtedly enable the recovery of additional hosts, and other *Asterophila* species, which will aid in clarifying their coevolutionary relationships. Future work should also look to sample from greater depths as *Asterophila* has yet to be reported from two orders of deep-sea asteroids. Lastly, previous studies have suggested that polyparasitism, where hosts are infected by multiple parasite species, is prevalent in natural populations \[[@CR47]--[@CR49]\], but this study recovered only a single *Asterophila* species from each host individual.
Antarctic distribution of *Asterophila* {#Sec10}
----------------------------------------
Combining this study and the original description of *A. perknasteri* \[[@CR15]\], *Asterophila* has now been reported from the South Shetland Islands, Elephant Islands, South Orkney Islands, South Georgia, and Shag Rocks, comprising the tip of the Antarctic Peninsula and parts of the Scotia Arc. Eight putative *Asterophila* species in this study were found co-distributed along the South Shetland Islands and Elephant Island, and a ninth putative species was found at Shag Rocks and South Georgia, but the absence of a single *Asterophila* species at both locations suggests their ranges are restricted. The octopus *Pareledone charcoti* also appears restricted to the South Shetland Islands and although this distribution may be due to deep-water and currents restricting gene flow in the area, it is also a consequence of a lack of a planktonic larval phase \[[@CR50]\], which *Asterophila* does possess. A single host with parasitic cyst was imaged from South Orkney but this specimen was not recovered from the collection for analysis. This host appears to be *Diplasterias* however, suggesting that the accompanying parasite could be *Asterophila* sp. 4. If so, this would extend the distribution of *A.* sp. 4 from the northern Shetland Islands and Elephant Island into South Orkney Islands, although additional samples and molecular analysis are needed to confirm this result. Moreover, some marine taxa show distinct genetic differences between populations at Shag Rocks and South Georgia (e.g. mackerel icefish \[[@CR51]\], octopus *Pareledone turqueti* \[[@CR52]\]), despite a short distance between them, but only four specimens of *A.* sp. 5 were retrieved from Shag Rocks and South Georgia in this study and therefore additional sampling is needed before elucidating any phylogeographic patterns. Shag Rocks has also been declared a biodiversity hotspot for benthic marine invertebrates \[[@CR53]\], warranting further investigation in this area.
The distribution of parasite depends on the distribution of its host \[[@CR42]\], but a plethora of potential hosts are distributed around Antarctica and the subantarctic islands and thus the absence of *Asterophila* at many locations outside the Scotia Arc is surprising. Salmen et al. \[[@CR54]\] uncovered a similar pattern in endoparasitic copepods, where parasites were only recovered from Sulawesi, Indonesia despite nudibranch hosts being broadly distributed across the tropical Indo-Pacific. The restricted co-distribution of *Asterophila* may also indicate that host-specific parasites are able to adapt to new hosts in sympatry, as suggested by Anton et al. \[[@CR24]\]. However, the reported distribution of *Asterophila* in this study could also be an artefact of concentrated sampling around the Antarctic Peninsula and Scotia Arc, combined with a low prevalence of parasitism in this group. For instance, Sasaki et al. \[[@CR16]\] found the rate of *A. japonica* infection ranged from just 2.2 to 17.5% in Japan, and Schiaparelli et al. \[[@CR12]\] found similarly low rates of infection by the eulimid *Bathycrinola tumidula* on the crinoid *Notocrinus virilis* in Antarctica. Given that hundreds of asteroids were examined in this study, and parasites were only discovered from 38 hosts, one could assume that the prevalence of parasitism in this system is comparably low and thus intensive sampling will be required for a more complete understanding of the distribution of this genus in Antarctica. Sampling effort at present has been focused around the Antarctic Peninsula and Scotia Arc, likely skewing the reported distribution of *Asterophila* in this study. In fact, specialist host-parasite systems rarely exhibit a broad distribution \[[@CR46]\], so it is likely that continued sampling around Antarctica, particularly in east Antarctica, will uncover a new suite of *Asterophila* species. Alternatively, the use of different trawling equipment among research cruises may have negatively impacted host and parasite recovery in this study, ultimately skewing rates of parasitism. Lastly, although adult *Asterophila* form large, noticeable cysts in asteroid seastars, it is likely that recently settled juveniles would be less detectable in hosts. In any case, the marine fauna of the Antarctic Peninsula and surrounding subantarctic islands has been shown to be incredibly diverse (e.g. \[[@CR7]--[@CR9]\]), and this study reveals this area may also be a 'coevolutionary hotspot', although these results may also point to a general lack of knowledge and study of parasite diversity on a global scale. As annual shelf temperatures in this region continue to warm (e.g. \[[@CR55]\]) the conditions are expected to promote range expansions in marine taxa \[[@CR56]\], potentially affecting the distribution of *Asterophila* through expansions in host range. Moreover, given the threat of coextinction to the ongoing biodiversity crisis (e.g. \[[@CR57]\]), coupled with accelerated rates of climate change in the region, it is crucial that we continue to document and understand host-parasite diversity in Antarctica.
Conclusions {#Sec11}
===========
This is the first study addressing Antarctic *Asterophila* since the initial description in 1994, uncovering eight putative new species, adding several new host genera, and expanding its distribution from the tip of the Antarctic Peninsula to Shag Rocks and South Georgia. This work also explores cophylogenetic patterns between *Asterophila* and their hosts, demonstrating that non-cospeciation events are primarily driving diversification in this system. Four of nine *Asterophila* species were found on multiple hosts, but these hosts were always phylogenetically related, being from one or two genera from the same family. Relationships between *Asterophila* species remain poorly resolved and obtaining molecular data for other congeners will be essential to better understand interspecific relationships in this genus. Additionally, a more robust understanding of the life history of these animals is needed, particularly regarding larval development and dispersal potential as well as the biochemical pathway responsible for host recognition. The distribution of *Asterophila* reported in this study is likely not reflective of its true range, due to both under-sampling in most of Antarctica and a potentially low rate of infection. Intensive sampling is needed in east Antarctica and the deep sea, the latter of which is particularly interesting as *Asterophila* has never been recorded from two orders of deep-sea asteroids. Furthermore, future work should continue to investigate cophylogenetic patterns in marine invertebrates and their parasitic counterparts, as these relationships are currently underrepresented in the literature.
Methods {#Sec12}
=======
Specimen collection and sequence acquisition {#Sec13}
--------------------------------------------
A total of 61 *Asterophila* specimens from 38 asteroid hosts were collected on three cruises (RVIB Nathaniel B. Palmer 11--05 in 2011, FS Polarstern ANT-XXVIII/4 in 2012, RVIB Nathaniel B. Palmer 13--03 in 2013) in Antarctica (Fig. [1](#Fig1){ref-type="fig"}; Additional file [3](#MOESM3){ref-type="media"}: Table S1). Additional hosts were also collected on the NOAA Antarctic Marine Living Resources (AMLR) cruise in 2009 and the Antarctic Circumnavigation Expedition (ACE) in 2016/2017 (Fig. [1](#Fig1){ref-type="fig"}). All *Asterophila* and corresponding hosts were collected with a benthic trawl at depths of 123 to 423 m on the continental shelf. Most specimens were preserved in 90--100% ethanol but some vouchers and veliger larva were fixed in 8% formalin and later transferred to 50--70% ethanol for morphological analysis. Some veliger larva are permanently mounted on SEM stubs at the Western Australian Museum (WAM) and remaining specimens and tissue samples are housed at both the Benthic Invertebrate Collection at Scripps Institution of Oceanography (BIC-SIO) and the Western Australian Museum (WAM). Both host and parasite specimens were identified in the field and these identifications were later confirmed and corrected by comparing cytochrome *c* oxidase subunit I (COI) sequences against the Barcode of Life Data System (BOLD) and NCBI BLAST databases.
Preserved tissue from the pseudopalium of *Asterophila* and tube feet from asteroid hosts were used for DNA extraction with DNeasy spin columns (Qiagen). Some molecular methods employed in this study have been previously described by \[[@CR58]\]. Primers used for PCR and sequencing are listed in Table [3](#Tab3){ref-type="table"}. For samples that were difficult to amplify, DNA dilutions (1:100) were used as template in PCR. Each PCR reaction included: 16.8 μl molecular grade water, 5 μl 5x MyTaq PCR buffer (Bioline), 0.5 μl forward and reverse primers (10 μM), and 0.2 μl platinum Taq polymerase (Bioline). The thermocycling regime for COI was: 3 min at 95 °C, 4 cycles of 40 s at 95 °C, 40 s at 45 °C, and 50 s at 72 °C, followed by 35 cycles of 40 s at 95 °C, 40 s at 51 °C, and 50 s at 72 °C, with a final extension for 10 min at 72 °C. Some asteroids were run with the following conditions for COI: 3 min at 95 °C, 8 cycles of 30 s at 95 °C, 30 s at 50 °C, and 45 s at 72 °C, followed by 38 cycles of 30 s at 95 °C, 30 s at 48 °C, and 45 s at 72 °C, with a final extension for 5 min at 72 °C. The thermocycling regime for 16S ribosomal RNA (16S), histone 3 (H3), and 28S ribosomal RNA (28S) was: 2 min at 94 °C, 35 cycles of 30 s at 94 °C, 40 s at 50 °C, and 60 s at 72 °C, with a final extension for 10 min at 72 °C. For some 28S reactions that failed in the first round of amplification, the following touch-down thermocycling regime was used in the second round; 2 min at 94 °C, 10 cycles of 30 s at 60 °C and decreasing by 0.5 °C every cycle until a final annealing temperature of 55.5 °C, 20 cycles of 30 s at 94 °C, 40 s at 55 °C, and 60 s at 72 °C, with a final extension for 10 min at 72 °C. The adenine nucleotide translocase (ANT) R1 primer was used for initial PCR and 2 μl of diluted (1:100) PCR template from this reaction was used for a second nested PCR with the ANT R2 primer. The thermocycling regime for ANT was: 3 min at 95 °C, 34 cycles of 40 s at 95 °C, 45 s at 51 °C, and 60 s at 68 °C, with a final extension for 5 min at 68 °C. Amplicons were screened on E-gels (Invitrogen) and all positive reactions were bidirectionally sequenced using the chain termination method and Big Dye Terminator kit. Sequencing was outsourced to the Australian Genome Research Facility (Perth). All sequences were edited in Geneious v.8.0.5 and aligned with MAFFT using default settings. Table 3Primers used for PCR and sequencing in this studyPrimer Name (F/R)MarkerF/R Sequence (5′ to 3′)ReferenceLCO1490/ HCO1490COI parasitesGGTCAACAAATCATAAAGATATTGG/TAAACTTCAGGGTGACCAAAAAATCA\[[@CR59]\]COIceF/ COIceR or jgLCO1490/ jgHCO2198COI hostsACTGCCCACGCCCTAGTAATGATATTTTTTATGGTN-ATGCC/ TCGTGTGTCTACGTCCATTCCTACTGTRAA-CATRTG TITCIACIAAYCAYAARGAYATTGG/ TAIACYTCIGGRTGICCRAARAAYCA\[[@CR60], [@CR61]\]16SarL/ 16SbrH16SCGCCTGTTTATCAAAAACAT/ CCGGTCTGAACTCAGATCACGT\[[@CR62]\]H3MF/H3MRH3ATGGCTCGTACCAAGCAGACTGC/ TGGATGTCCTTGGGCATGATTGTTAC\[[@CR63]\]ANTF/ANTR1/ ANTR2ANTCCATTYTGGMGIGGWAACWTGGC/ TTCATCAAIGACATRAAICCYTC/ CCCTTGTAYTTRACASCYTCACC\[[@CR64]\]1100F (forward) na2 (reverse) with28SGGACCCGAAAGATGGTGAACTATGC/ AGCCAATCCTTATCCCGAAG\[[@CR25], [@CR65]\]FL (forward)\*AAGTGGAGAAGGGTTCCATGT\[[@CR25]\]LSU5/ LSU1600R with ECD25 (reverse)\*TAGGTCGACCCGCTGAAYTTAAGCA/ AGCGCCATCCATTTTCAGG CTTGGTCCGTGTTTCAAGACGG\[[@CR66], [@CR67]\]\[[@CR68]\]Multiple primer sets were used to amplify the entire fragment of 28S, with an asterisk denoting sequencing primers
Phylogenetic reconstruction and species delimitation {#Sec14}
----------------------------------------------------
*Niso matsumotoi* was selected as an outgroup for *Asterophila* phylogenetic analysis based on a larger phylogeny of Eulimidae (Layton et al. in preparation), and these sequences were obtained from GenBank (AB930469, AB930413, AB930440, AB930335). The phylogenetic methods employed in this study have been previously described by \[[@CR58]\]. A dataset comprised of sequences from two mitochondrial genes (COI, 16S) and three nuclear genes (H3, 28S, ANT) was concatenated for subsequent phylogenetic analysis. A maximum likelihood (ML) tree was constructed in RAxML \[[@CR69]\] implemented in the raxmlGUI v1.3 \[[@CR70]\] using a GTR + G model with partitions set for each gene and 1000 bootstrap replicates calculated with joint partition support. Species were recognized as highly supported least-inclusive clades of terminals in the phylogenetic analysis, forming the primary species hypothesis (PSH) (Additional file [1](#MOESM1){ref-type="media"}: Figure S1). A secondary species hypothesis (SSH) was reassessed based on results from two species-delimitation analyses. The tree file was imported into RStudio and the ape package \[[@CR71]\] was used to collapse nodes in the SSH with low support (\< 50%). Pairwise distances were calculated in MEGA v6.0 using the COI dataset and a maximum composite likelihood model with 100 bootstrap replicates and pairwise deletion. The Bayesian Poisson tree processes (bPTP) \[[@CR72]\] and Automated Barcode Gap Discovery (ABGD) \[[@CR73]\] algorithms were used to partition the concatenated dataset into unique genetic clusters. The parameters employed for the former were 100,000 MCMC generations, a thinning value of 100, and 10% burn-in, and for the latter were Pmin = 0.001, Pmax = 0.10, 10 steps, X = 1.5, Nb bins of 20, and a Jukes Cantor (JC69) model. The final ML tree was used as an input file for the bPTP analysis and the fasta file for this ML tree was employed for the ABGD analysis. Finally, host type, defined as asteroid order, was mapped onto the uncollapsed ML phylogeny with branch-lengths retained and ancestral state reconstruction was conducted with an Mk1 model in Mesquite 3.04 \[[@CR74]\].
Cophylogenetic analysis {#Sec15}
-----------------------
Both distance-based and event-based cophylogenetic analyses were conducted on host and parasite phylogenies. A host phylogeny was generated using COI sequences from asteroids that were found with parasites. The ML tree was generated in RAxML \[[@CR69]\] implemented in the raxmlGUI v1.3 \[[@CR70]\] using a GTR + G model and 1000 bootstrap replicates. An ophiuroid was used as an outgroup and the topology was constrained to reflect the most current asteroid phylogeny \[[@CR75]\]. A corresponding parasite ML phylogeny was generated using only those individuals for which host sequence data was available. The cophylo function in the package Phytools \[[@CR76]\] was used in RStudio to visualize host-parasite associations. All individual hosts and parasites were included in subsequent cophylogenetic analysis, but outgroups were excluded. The host and parasite phylogenies were used to construct a tanglegram in TreeMap3.0b \[[@CR77]\], which was imported into Jane 4 \[[@CR78]\] to run event-based reconciliation analysis with 99,999 permutations, a generation size of 200, and a population size of 100. Random parasite trees were also constructed in Jane with a population size of 50, and a Welch's two-sample t-test was used to determine whether the costs of each regime using these 50 random parasite trees was different to the costs obtained using real parasite trees. If the cost generated from the random parasite trees is significantly greater than the observed cost then a global signal of coevolution is present \[[@CR79]\]. Four different cost regimes were employed in Jane; i) equal costs (11111), ii) assigning cospeciation events a value of 0 (01211), iii) assigning host-switching events a value of 2 (11211), and iv) assigning duplication and host-switching events a value of 0 (10011). Distance-based analysis was run in ParaFit \[[@CR80]\] using the ape package \[[@CR71]\] in RStudio, which tests for both global and individual congruence. This analysis tests the null hypothesis that hosts and parasites are evolving independently. For ParaFit, cophenetic distances (branch length pairwise differences) were calculated from host and parasite phylogenies using the cophenetic function in the ape package \[[@CR71]\] in RStudio, along with an association matrix, 99,999 permutations, and a Cailliez correction for negative eigenvalues. A Bonferroni correction was applied to *p*-values from the ParaFit analysis as two ParaFitLink tests were conducted for each comparison. The Jane and ParaFit analyses were chosen as they allow for uneven numbers of parasites and hosts, including multihost parasites. Lastly, for *Asterophila* species that were discovered from multiple hosts, TCS haplotype networks \[[@CR81]\] with COI data and 5000 iterations were generated in PopART \[[@CR82]\] for exploring genetic structure and potential host partitioning. To determine whether *Asterophila* species were parasitizing phylogenetically related hosts, cophenetic distances between host pairs (hosts parasitized by the same *Asterophila* species) were compared to cophenetic distances between all other hosts using a Wilcoxon 2-sample rank sum test.
Larval morphology {#Sec16}
-----------------
Published descriptions of known *Asterophila* species highlight only a few diagnostic morphological characters in this group, some of which are based on larval shell characters \[[@CR15]\]. As such, larval shells from four preserved specimens per species were examined using a Hitachi TS3030Plus tabletop scanning electron microscope (SEM) at the Western Australian Museum. Larval shells were transferred directly from 100% ethanol to the SEM stub. The general shape, sculpture and umbilicus were assessed by eye, and larval shell length was measured across the widest point of the shell. A one-way ANOVA was employed in the stats v3.6.0 package \[[@CR83]\] in RStudio for comparing mean shell size between species.
Additional files
================
{#Sec17}
Additional file 1:**Figure S1.** Primary species hypothesis. ML phylogeny (COI + 16S + H3 + 28S + ANT) of *Asterophila* with highly supported least inclusive clades (and singletons) representing the PSH. Hash marks denote that the branch has been truncated to one half of its original length. (TIF 474 kb) Additional file 2:**Figure S2.** Host type**.** Ancestral state reconstruction (ML) for host type using asteroid order and an Mk1 likelihood model. Asterisks mark nodes with a likelihood of \> 99%, with values less than this provided at nodes. (TIF 1268 kb) Additional file 3:**Table S1.** Specimen details. *Asterophila* specimens collected and analysed in this study, along with corresponding host details, locality information, and GenBank accession numbers. Hosts marked with an asterisk lack sequence data. (DOCX 37 kb)
16S
: 16S ribosomal RNA
28S
: 28S ribosomal RNA
ABGD
: Automated Barcode Gap Discovery
ACE
: Antarctic Circumnavigation Expedition
AMLR
: NOAA Antarctic Marine Living Resources
ANT
: Adenine nucleotide translocase
BIC-SIO
: Benthic invertebrate Collection at Scripps Institution of Oceanography
BOLD
: Barcode of Life Data System
bPTP
: Bayesian Poisson tree processes
BS
: Bootstrap support
COI
: Cytochrome *c* oxidase subunit I
H3
: Histone 3
JC69
: Jukes Cantor
ML
: Maximum likelihood
PSH
: Primary species hypothesis
SEM
: Scanning electron microscope
SSH
: Secondary species hypothesis
WAM
: Western Australian Museum
**Publisher's Note**
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We are grateful to all voyage participants for assistance with sampling, as well as the benthic team on the Antarctic Circumnavigation Expedition. Christopher Jones and the US NOAA AMLR benthic team are thanked for inviting NGW to participate on cruises, and for sampling assistance, especially Susie Lockhart and Eric Lazo-Wasem. We are grateful to Diana Prada and Mia Hillyer at the Western Australian Museum for molecular advice.
KKSL and NGW conceived and designed the study. KKSL, GWR, and NGW collected specimens and KKSL and GWR generated molecular data. KKSL analysed the data. All authors contributed to writing the manuscript and have read and approved the final manuscript.
Funding for this research comes from the Antarctic Circumnavigation Expedition, The University of Western Australia, and the National Science Foundation (USA) ANT-1043749. KKSL was supported by a University Postgraduate Award for International Students (UPAIS) and an RTP International Fees Offset scholarship (RTPFI) administered by The University of Western Australia, as well as a postgraduate doctoral scholarship from the Natural Sciences and Engineering Research Council of Canada (NSERC). These funding bodies did not have a role in the design of the study, the collection, analysis, and interpretation of data, or in writing the manuscript.
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request and GenBank accessions appear in the manuscript.
No ethics approvals were required for work on invertebrates. Sampling permissions were granted by the Government of South Georgia and the South Sandwich Islands (084/2011, 096/2012).
Not applicable.
The authors declare that they have no competing interests.
| {
"pile_set_name": "PubMed Central"
} |
1. Introduction
===============
Dehydrocorydaline (DHC), a quaternary protoberberine-type alkaloid ([Figure 1](#molecules-19-16312-f001){ref-type="fig"}), is an important bioactive component of a very well-known Chinese herbal medicine, Yan-Hu-Suo, which is the tuber of *Corydalis yanhusuo*. The herb has been widely used as an analgesic agent for treating spastic, abdominal, and menstrual pains, and for pain due to injury \[[@B1-molecules-19-16312]\]. It has also been widely used to promote blood circulation and treat coronary heart disease \[[@B2-molecules-19-16312],[@B3-molecules-19-16312],[@B4-molecules-19-16312],[@B5-molecules-19-16312]\]. Previous studies have identified that quaternary and tertiary alkaloids are the active secondary metabolites of the plant \[[@B6-molecules-19-16312],[@B7-molecules-19-16312]\]. The pharmacological effects of the two different types of alkaloid focus on different aspects. Tertiary alkaloids have been found to be effective at alleviating pain (e.g., tetrahydropalmatine, corydaline, and protopine, in order of decreasing efficiency \[[@B1-molecules-19-16312]\]). In contrast, some quaternary alkaloids, such as DHC, have been found to be more active than tertiary alkaloids in increasing the tolerance of mice to monobasic and hypobaric hypoxia. They also have greater efficacy than tetrahydropalmatine (a tertiary alkaloid) in protecting cardiomyocytes against ischemia in the myocardium *in vitro*. As well as two other quaternary alkaloids, berberine and palmatine, it appears that these are the main active ingredients in *Corydalis yanhusuo* for the treatment of coronary heart disease \[[@B8-molecules-19-16312],[@B9-molecules-19-16312],[@B10-molecules-19-16312]\].
{#molecules-19-16312-f001}
Through qualitative and quantitative research on the effective fraction of *Corydalis yanhusuo* (EFY) \[[@B11-molecules-19-16312],[@B12-molecules-19-16312],[@B13-molecules-19-16312],[@B14-molecules-19-16312],[@B15-molecules-19-16312],[@B16-molecules-19-16312]\], we have found that quaternary alkaloids are the main bioactive components ([Figure 2](#molecules-19-16312-f002){ref-type="fig"}). In addition, our Caco-2 cell experiment \[[@B17-molecules-19-16312]\] suggested that synergy between the alkaloids in the fraction might promote their oral absorption across the intestinal epithelium. On the other hand, another study showed that the plasma concentration--time curve for DHC can only be obtained when it is given orally in high doses after a total alkaloid extraction from *Rhizoma Corydalis*, indicating that DHC is absorbed poorly \[[@B18-molecules-19-16312]\]. There are, at present, large differences in the studies on the absorption of quaternary protoberberine-type alkaloids.
{#molecules-19-16312-f002}
On the one hand, quaternary protoberberine-type alkaloids are considered to be poorly absorbed because, possibly, of the action of P-glycoprotein (P-gp) inhibitors resulting in trace plasma concentrations after oral administration. On the other hand, it was found that the absorbed amounts of berberine, a quaternary protoberberine-type alkaloid, accounted for 33.6% of the given dose 1 h after oral administration in an *in situ* assay of the intestinal loop \[[@B19-molecules-19-16312]\]. Berberine may be quickly excreted from the blood in rats through bile \[[@B20-molecules-19-16312],[@B21-molecules-19-16312]\]. This indicates that DHC can be absorbed well but has a low dosage in systemic circulation. This might be because there may be a very large "first pass" effect in the liver after oral administration. Therefore, data on the cumulative excretion of unchanged quaternary protoberberine-type alkaloids in the bile after oral administration would be very important for clarifying the absorption extent.
Up to now, only two papers investigated the pharmacokinetics of DHC \[[@B18-molecules-19-16312],[@B22-molecules-19-16312]\]. In this paper, a highly sensitive liquid chromatography--tandem mass spectrometry (LC-MS/MS) method is developed for DHC determination. Comparing with the published reports, our method with isocratic elution in R.T. (retention time) is much better, the LLOQ is 0.625 ng/mL. With SPE, we got high recovery and less matrix effect. It is subsequently successfully applied to a pharmacokinetic study of DHC when given to rats in two different dosage forms: one, an effective fraction of *Corydalis yanhusuo*, and the other, the pure compound. After oral administration, the results were analyzed to see if there were any synergies between the compounds in EFY and to further assess whether the influence of other compounds in EFY affected the pharmacokinetics and blood concentration of DHC. In addition, excretion of DHC in the bile and urine of the rats was investigated to clarify the pharmacokinetic properties of DHC.
2. Results and Discussion
=========================
2.1. Method Validation of DHC in Rat Plasma
-------------------------------------------
In order to investigate the comparative pharmacokinetics of DHC in rats, a sensitive and reliable analytical method was developed and validated. Assay selectivity was evaluated by analyzing blank plasma samples from six rats. Positive-ion electrospray mass spectra for DHC and IS are shown in [Figure 3](#molecules-19-16312-f003){ref-type="fig"}. The peaks \[M\]^+^ = 366 and 348 for DHC and IS, respectively, were chosen for quantification purposes due to their high stability and intensity. [Figure 4](#molecules-19-16312-f004){ref-type="fig"} shows representative MRM chromatograms of blank plasma (A), plasma spiked with 2.5 ng/mL DHC and 800 ng/mL IS (B), plasma spiked with 100 ng/mL DHC and 800 ng/mL IS (C), and a plasma sample 1 h after oral administration of 97.5 mg/kg DHC to a rat. No interfering peaks were observed in the blank plasma under assay conditions.
{#molecules-19-16312-f003}
{#molecules-19-16312-f004}
A calibration curve was constructed using unweighted linear regression of the DHC/IS peak-area ratio (*Y*) against the corresponding spiked plasma DHC concentration in ng/mL (*X*) over the range 0.625--250 ng/mL. The regression equation had the form *Y* = 0.00916*X* + 0.000813 (r = 0.9976, n = 8). The concentrations in unknown samples were subsequently calculated using this calibration expression. The LLOQ of the assay was 0.625 ng/mL for DHC. The LLOQ, as the lowest concentration, was determined with both accurately and precisely. The precision (RSD, %) was 8.45 and the accuracy (RE, %) was 6.24. The results showed LLOQ of DHC can achieve 0.625 ng/mL for established method ([Table 1](#molecules-19-16312-t001){ref-type="table"}).
molecules-19-16312-t001_Table 1
######
Accuracy and precision results for analysis of DHC in rat plasma (3 days, six replicates per day).
C~nom~ (ng/mL) C~det~ (ng/mL) Precision (RSD%) Accuracy (RE%)
---------------- ---------------- ------------------ ---------------- -------
0.625 0.664 8.45 \- 6.24
1.25 1.2 3.29 13.68 −6.89
25 23.7 4.06 7.07 −5.29
250 258.1 3.27 7.39 3.22
The precision and accuracy data for the assays is shown in [Table 1](#molecules-19-16312-t001){ref-type="table"}. The results indicate that the LC-MS/MS method developed has good reproducibility with precision less than 13.68% and excellent accuracy ranging from −6.89% at low (1.25 ng/mL) to 3.22% at high (250 ng/mL) concentrations. Extraction recovery of the DHC was found to be satisfactory with average values ranging from 92.08% to 107.00% at the three QC concentrations ([Table 2](#molecules-19-16312-t002){ref-type="table"}).
molecules-19-16312-t002_Table 2
######
The extraction recovery results for DHC in rat plasma (n = 6).
Concentration (ng/mL) Recovery (%) (mean ± SD)
----------------------- -------------------------- --------------
1.25 107 ± 2.31
25 92.08 ± 3.82 74.56 ± 4.05
250 99.65 ± 4.07
The stability study showed that the variation in the concentration after three cycles of freezing and thawing was within ±15% of the nominal concentration, indicating no significant loss of substance during freezing and thawing. When the processed samples were stored in an auto-sampler at 4 °C, the DHC also showed good stability. This is evidenced by the fact that the responses varied no more than ±10% over 24 h at the QC concentrations. After storage at ambient temperature for 12 h, the concentration of DHC in the plasma deviated less than ±13% from those in freshly-spiked plasma ([Table 3](#molecules-19-16312-t003){ref-type="table"}).
molecules-19-16312-t003_Table 3
######
Stability results for analysis of DHC in rat plasma (n = 6).
Storage C~nom~ (ng/mL) C~dec~ (ng/mL) (mean ± S.D.) Relative Error (%) RSD (%)
---------------------------------------------- ---------------- ------------------------------ -------------------- ---------
**Postpreparative stability (24 h)** 1.25 1.18 ± 0.06 −5.47 5.26
25 26.2 ± 0.86 4.8 8.13
250 270.33 ± 9.42 8.14 3.48
**Stability after three freeze-thaw cycles** 1.25 1.25 ± 0.1 −0.27 9.86
25 27.5 ± 0.87 10 3.17
250 272.67 ± 8.14 9.07 2.99
**Short-term temperature stability (12 h)** 1.25 1.2 ± 0.1 −3.87 8.14
25 21.97 ± 0.51 −12.13 2.33
250 230.67 ± 15.96 −7.73 6.92
As shown in [Table 4](#molecules-19-16312-t004){ref-type="table"}, no significant matrix effect was observed for the six blank plasma lots, indicating that the extracts were "clean" with no co-eluting of "unseen" compounds that could influence the ionization of DHC.
molecules-19-16312-t004_Table 4
######
Matrix effects data for DHC at 25.0 ng/mL and IS 200 ng/mL in six different sources of rat plasma.
Matrix Effect (%) (mean ± S.D.) RSD (%)
----- --------------------------------- ---------
DHC 89.40 ± 0.03 3.1
IS 92.3 ± 0.04 4.1
2.2. Method Validation of DHC in Bile and Urine Samples
-------------------------------------------------------
For biliary and urinary excretion study, the calibration curves were constructed using unweighted linear regression of the DHC/IS peak-area ratio (*Y*) against the corresponding spiked plasma DHC concentration in ng/mL (*X*), the ranges are 1.25--500 ng/mL for biliary and 0.625--250 ng/mL for urinary. The regression equations had the form *Y* = 0.000437*X* + 0.0149 (r = 0.9971, n = 8) and *Y* = 0.00675*X* + 0.00279 (r = 0.9984, n = 8). The concentrations in unknown samples were subsequently calculated using this calibration expression. The LLOQ of the assay was 1.25 ng/mL in bile and 0.625 ng/mL in urine. Assay selectivity was evaluated by analyzing blank samples from four rats. Results showed that no interfering peaks were observed in both blank bile and urine under assay conditions.
2.3. Application
----------------
### 2.3.1. Plasma Pharmacokinetics
DHC is the most abundant quaternary protoberberine-type alkaloid in EFY and has clear cardiovascular pharmacological activity. A fully validated LC-ESI-MS/MS method was developed and was successfully applied to a pharmacokinetic study of DHC when given in the two different dosage forms (EFY and pure compound) via oral administration in rats. The doses of DHC and EFY were 97.5 mg/kg and 483 mg/kg (equivalent to 97.5 mg/kg DHC), respectively. The mean plasma concentration-time profiles of DHC after the two doses are illustrated in [Figure 5](#molecules-19-16312-f005){ref-type="fig"} and the estimated pharmacokinetic parameters are presented in [Table 5](#molecules-19-16312-t005){ref-type="table"}.
{#molecules-19-16312-f005}
Except for the AUC~0--*t*~ parameter of DHC, all the other pharmacokinetic parameters were statistically significantly different (*p* \< 0.05) between the two dosage forms. The *C*~max~ and AUC~0--∞~ of DHC after being given in EFY form were about 3.1- and 1.9-fold higher than those when given as the pure compound, respectively. The *t*~1/2~ of the DHC was also changed, being about 2.7-fold longer in the EFY form than in the pure form. The *T*~max~ of the DHC was decreased 0.31-fold shorter than when given as pure compound. These results indicate that, when given in mixture form (*i.e.*, EFY), DHC seems to be absorbed faster and eliminated more slowly from the body compared to administration of the pure substance.
molecules-19-16312-t005_Table 5
######
Pharamcokinetic parameters of DHC in rats following oral administration, either in an EFY form or in pure form, at an equivalent dose of 97.5 mg/kg.
Parameter Dosage Forms *p* Value
-------------------- ---------------- --------------- -------
C~max~ (ng/mL) 28.7 ± 6.92 9.40 ± 4.18 0.001
T~max~ (h) 0.31 ± 0.13 1.00 ± 0.00 0.000
AUC~0-t~ (ng∙h/mL) 71.92 ± 14.79 52.39 ± 12.82 0.056
AUC~0-∞~ (ng∙h/mL) 115.12 ± 34.12 59.08 ± 11.53 0.008
T~1/2~ (h) 21.71 ± 12.35 7.93 ± 1.34 0.038
K~e~ (1/h) 0.05 ± 0.03 0.09 ± 0.02 0.023
The differences in the pharmacokinetic parameters of DHC after oral administration of pure DHC and EFY demonstrate that other components co-existing in the EFY appear to affect the absorption and elimination of DHC. However, the mechanisms for this interaction are not yet clear. Nevertheless, there are two basic conclusions that are clear from earlier studies.
First, berberine is a substrate of P-glycoprotein since P-gp is involved in the efflux activity of berberine in Caco-2 cells and the liver \[[@B23-molecules-19-16312],[@B24-molecules-19-16312]\]---it has an obscure function to increase or inhibit P-gp function. Although there are not any reports that DHC is a substrate of P-gp, DHC has structure similar to berberine, it is also presumably a substrate of P-pg. Thus, we can infer that other alkaloids in EFY, which have structures similar to berberine's, are most likely to be substrates of P-gp. They will compete with DHC for the P-gp involved in the efflux activity and so indirectly decrease the amount of DHC effluxed by P-gp. This will increase the DHC blood concentration and prolong elimination time.
Secondly, the process of drug combining with plasma protein is usually reversible. When the process is inhibited by combined drug and other reasons, the concentration of free drugs will greatly increase. Therefore, the mutual competition of several drugs will lead to drug concentration and other parameters changing rapidly while they are sharing limited binding site. It seems from former studies that berberine might be a weak cytochrome P450 3A4 inhibitor \[[@B25-molecules-19-16312]\]. As a result, it is reasonable to reach the inference that other alkaloids in EFY with similar structures to berberine are likely to enter into the competition for biding sites with DHC for enzymes involved in metabolism in the liver (and other organs). Therefore, the metabolic rate of DHC is decreased and the half-life of DHC becomes longer as a result. Up to now, there are no reports about binding values of DHC. *In vivo* metabolism of DHC was investigated, and several metabolites of DHC were detected and characterized by LC-MS. The research results will be published elsewhere.
Overall, compared with rats given pure DHC, the reason for the significant changes occurring in DHC's pharmacokinetic parameters after oral administration of EFY may be that other components co-existing in EFY inhibit P-gp function. The other components co-existing in EFY may compete with DHC for P-gp transportation, metabolization by P450, and binding to plasma proteins. The mechanisms for these interactions should be investigated further.
### 2.3.2. Excretion
The cumulative excretion of DHC in bile and urine after a single i.g. dose of DHC (97.5 mg/kg) was determined. Results ([Figure 6](#molecules-19-16312-f006){ref-type="fig"}) showed that there are wide individual differences in the biliary cumulative excretion of unchanged DHC. After a single oral dose of DHC, the rate of total biliary excretion was 60.04% ± 24.66% of the dose up to 48 h. However, the urinary cumulative excretion of DHC was little. The mean accumulated amount of DHC in urine over 96 h was 0.17% ± 0.13% of dose. Above results showed that DHC was mainly excreted in bile after oral administration of DHC. The mean accumulated amount of DHC in bile in just 2 h was 12.2% ± 7.38% of dose, indicating that DHC can be absorbed quickly following oral administration. There might be a huge "first pass" effect in liver after oral administration and account for the trace plasma concentration.
{#molecules-19-16312-f006}
3. Experimental Section
=======================
3.1. Chemicals and Reagents
---------------------------
The nitidine chloride used as internal standard (IS) was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). DHC was isolated from the dried tuber of *Corydalis yanhusuo*. Its purity was found to be 99.6% using HPLC with photodiode array detection. The effective fraction of *Corydalis yanhusuo* (containing DHC 20.2%) was prepared in our laboratory. The methanol and acetonitrile used were of HPLC grade and were obtained from Fisher Co. Ltd. (Emerson, IA, USA). The formic acid and other reagents used were of analytical grade and purchased from Beijing Chemical Reagent Company (Beijing, China). Milli-Q water (Milford, MA, USA) was used throughout the study.
3.2. Animals and Doses
----------------------
Specific pathogen free male Sprague-Dawley rats (weight: 220 ± 20 g) were obtained from the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences, Beijing, China. They were kept for 7 days in a controlled environment (22--24 °C, 60% relative humidity) under a 12 h light-dark cycle with free access to a soy-free custom diet and tap water. All the rats were fasted for 12--16 h before the experiment. All animal procedures were performed in compliance with the guidelines approved by the Animal Ethnics Committee of the Chinese Academy of Medical Sciences.
For pharmacokinetics study, the rats were divided into two groups (n = 5 per group). Doses of 97.5 mg/kg of DHC or 483 mg/kg of EFY (equivalent to 97.5 mg/kg DHC, 20.19% DHC in EFY) were all gavage administrated. Blood samples (500 μL) were withdrawn from the oculi chorioideae vein into heparinized tubes before dosing and subsequently at 15, 30 and 45 min and then at 1, 2, 4, 6, 8, 12 and 24 h following initial administration. After 1 h and 8 h, 2 mL of normal saline was injected into the caudal vein to compensate for blood loss. After centrifuging at 4000 rpm for 10 min, plasma samples were obtained and frozen at −20 °C until analysis.
For biliary excretion study, four rats were light anesthetized by ether and a PE-10 cannula was implanted into the bile duct to collect bile. A single dose of DHC was administered as described above. Bile samples were collected at 0--2, 2−4, 4−6, 6−8, 8−10, 10−24 and 24−48 h post-dosing and kept at −20 °C after the volume of each collection was recorded.
For urinary excretion study, four rats were used. DHC was administered as described above. The animals were housed in stainless steel metabolic cages with free access to tap water. Urine was collected at 0--4, 4--8, 8--12, 12--24, 24--36, 36--60, 60--84 and 84--96 h post-dosing and kept at −20 °C after the volume of each collection was recorded.
3.3. Preparation of Calibration Standards and Quality Control Samples
---------------------------------------------------------------------
Stock solutions of DHC and IS were both prepared in methanol at a final concentration of 1 mg/mL and stored at −80 °C. A series of working standard solutions of DHC ranging from 2.5 to 1000 ng/mL and an IS solution at 800 ng/mL were prepared by diluting their stock solutions with 10% acetonitrile (stored for less than one week at −20 °C for the assay). Calibration standards were prepared using blank rat plasma (200 μL) spiked with 50 μL of DHC working solutions, to yield solutions with concentrations of 0.625, 1.25, 2.5, 6.25, 25, 62.5, 125 and 250 ng/mL. Quality control (QC) samples were prepared in the same way as the calibration samples at 1.25, 25.0 and 250.0 ng/mL, representing low, medium, and high concentrations of DHC in the plasma, respectively. The linearity for DHC in bile and urine established separately to be within 1.25--500 ng/mL and 0.625--250 ng/mL range.
3.4. Sample Preparation
-----------------------
A 200 μL aliquot of plasma was vortex mixed for 3 min with 200 μL of water and 50 μL of IS (800 ng/mL). After centrifugation at 9000 rpm for 10 min, all the supernatant was loaded onto an Oasis HLB cartridge (Waters, Milford, MA, USA), which was pre-conditioned with 1 mL of methanol followed by 1 mL of deionized water. The loaded-cartridge was sequentially washed with 1 mL of water and 1 mL of 20% methanol. The analytes were then eluted with 1 mL of methanol in water containing 2% formic acid. The collected eluate was evaporated to dryness at 40 °C under a gentle stream of nitrogen. The residue was reconstituted in 150 μL of 10% acetonitrile, and 30 μL of the resulting sample was injected into the LC-MS/MS system for assay. Bile and urine samples were all processed in a similar manner as the plasma samples.
3.5. LC-MS/MS Condition
-----------------------
An Agilent 1100 system (Palo Alto, CA, USA) equipped with a vacuum degasser, a quaternary pump, and an auto-sampler were used. Chromatographic separation was achieved on an Atlantis T3 column (100 mm × 2.1 mm, 5 μm; Waters, Milford, MA, USA), which was eluted with a mobile phase of acetonitrile and water (containing 0.8% formic acid and 10 mM ammonium acetate) (28:72, v/v). The mobile phase was delivered at a flow-rate of 0.25 mL/min and 40% of the eluate was introduced into an API 3000 triple-quadrupole mass spectrometer (Applied Biosystems MDS SCIEX, Concord, ON, Canada) equipped with a TurboIonSpray electrospray ionization (ESI) source. Detection was performed using positive-ion electrospray ionization in multiple reaction monitoring (MRM) modes. The transitions from molecular ion to dominant product ion *m/z* 366→351\[M−CH~3~\] and *m/z* 348→332\[M−CH~3~−H\] \[[@B26-molecules-19-16312],[@B27-molecules-19-16312],[@B28-molecules-19-16312]\] were monitored for DHC and IS, respectively. The optimized working parameters for mass detection were as follows. The nebulizer, curtain, and collision-activated dissociation gases were set at 11, 7 and 11 (instrument units), respectively. The TurboIonSpray voltage and temperature were set at 4.5 kV and 400 °C, respectively. All data acquisition was performed using Analyst software (v1.4, AB MDS SCIEX).
3.6. Method Validation
----------------------
### 3.6.1. Calibration Linearity and Low Quantification Limits
A calibration curve for DHC was constructed by plotting the DHC/IS peak-area ratios *vs.* concentration of DHC in the plasma. Linearity was determined using linear least-squares regression. The lower limit of quantification (LLOQ) of the assay was defined as the lowest concentration on the standard curve that can be quantified with accuracy within ±15% bias of the nominal concentration and precision not exceeding 15% CV for DHC.
### 3.6.2. Precision, Accuracy and Extraction Recovery
The precision (relative standard deviation) and accuracy (%) of the assay were determined from replicate analyses (n = 6). Three QC analyses were made on the same day ("intra-day") and the others on 3 consecutive days ("inter-day"). The accuracy was calculated from the nominal concentration (*C*~nom~) and the mean value of the observed concentration (*C*~obs~) as follows:
The extraction recovery for DHC and IS was measured by comparing the peak areas of the extracted (pre-spiked) QC samples with those of the unextracted biological samples at an equivalent concentration. The recovery of DHC was determined at three levels (1.25, 25 and 250 ng/mL), while IS at a single concentration of 200 ng/mL. The results were expressed as mean ± standard deviation (SD).
### 3.6.3. Stability
The stability of DHC in rat plasma was evaluated under conditions that mimicked those likely to be encountered during sample storage and analytical processing. Six replicates of QC samples were analyzed for DHC. The QC samples were frozen and stored at −20 °C for a week. The concentration variation of the DHC in the prepared QC samples was detected after (i) three cycles of freezing and thawing, (ii) placement in an auto-sampler at 4 °C for 24 h, and (iii) storing at ambient temperature for 12 h.
### 3.6.4. Matrix Effects
Matrix effects (co-eluting undetected endogenous matrix compounds that may influence ionization of the DHC) were examined by comparing the DHC and IS peak-areas between two different sets of samples. In Set 1, standard DHC was dissolved in the reconstitution solvent and analyzed at DHC concentrations of 1.25, 25 and 250 ng/mL and an IS concentration of 200 ng/mL. These analyses were repeated six times at each concentration. In Set 2, blank plasma samples obtained from six rats were extracted and then spiked with the same concentration of DHC and IS dissolved in the reconstitution solvent. Deviation between the mean peak-areas of Set 2 *vs.* Set 1 indicates the possibility of ionization suppression or enhancement for DHC and IS. This is called an "absolute" matrix effect \[[@B29-molecules-19-16312]\].
3.7. Pharmacokinetic Study
--------------------------
Pharmacokinetic parameters were established from the plasma concentration--time data using non-compartmental analysis. The terminal elimination rate constant (*K~e~*) was determined by linear regression of the terminal portion of the plasma concentration-time data. The elimination half-life (*t*~1/2~) was calculated using the expression: The area under the curve (AUC) for the plasma concentration-time trace from zero to the last plasma drug concentration (AUC~0--*t*~) was calculated using the trapezoidal rule. Extrapolation to infinite time (AUC~0--∞~) was calculated using the expression: where *C~t~* is the last measurable plasma concentration. The value of the maximal plasma concentration (*C*~max~) and time to maximal concentration (*T*~max~) were obtained directly from the plasma concentration-time curve.
The statistical significance of the pharmacokinetic parameters obtained from the two different forms (mixture/pure) was estimated using analysis of variance (ANOVA). SPSS 10.0 one-way ANOVA tests were used to compare the pharmacokinetic parameters from DHC and the mixture. A *p* value of less than 0.05 was considered to be significantly different. All results were expressed as arithmetic mean ± standard deviation.
4. Conclusions
==============
For the first time, a sensitive and reliable LC-MS/MS method has been developed for the determination of DHC in rat plasma using solid-phase extraction as sample clean-up procedure. Full validation indicated that the established method was excellent sensitivity, linearity, precision, and accuracy. The method was successfully applied to the pharmacokinetic study of DHC in rat plasma after oral administration of pure DHC and EFY. Significant differences of *T*max and *C*max showed that DHC from EFY reached the peak concentrations more rapidly with higher concentrations, which indicated that the influence of the herb-herb interactions on DHC and other components co-existing in the EFY should be considered. Although the mechanism was still ambiguous and complex, the knowledge obtained in this study might help to explain the pharmacological mechanism and evaluate the impact of the differences in the efficacy and safety in clinical applications of EFY. Further research is in progress.
This work was supported by grants from the NSFC (No. 30772710) and National Mega-project for Innovative Drugs (Nos. 2009ZX09103-382 and 2012ZX09301002-001-27) from the Ministry of Science and Technology of the People's Republic of China.
*Sample Availability*: Samples of Dehydrocorydaline (DHC) and the effective fraction of *Corydalis yanhusuo* (EFY) are available from the authors.
Qiu-Yue Li made substantial contributions to conception and design, acquisition of data, analysis and interpretation of data. Kai-Tong Li, Hong Sun, Wen Jin and Jia-Wen Shi gave suggestions for writing. Qiu-Yue Li and Yue Shi participated in drafting the article and revised it critically for important intellectual content; and gave final approval of the version to be submitted and any revised version.
The authors declare no conflict of interest.
| {
"pile_set_name": "PubMed Central"
} |
Community event--based surveillance (CEBS) systems have been used for case finding during outbreaks and to increase sensitivity for detection of diseases targeted for eradication ([@R1]*--*[@R5]), but this type of surveillance has not been implemented rapidly on a national scale during a large health emergency. In October 2014, near the peak of the Ebola virus disease (EVD) epidemic in West Africa, the International Rescue Committee, Sierra Leone's Bo District Health Management Team, and the US Centers for Disease Control and Prevention developed CEBS in Sierra Leone to serve as a village-level active surveillance system for reporting possible EVD cases ([@R6]). At that time, many infected persons were not detected until after they had died by the national surveillance system, which consisted of contact tracing, healthcare facility surveillance, and a telephone hotline for reporting events; thus, opportunities for virus transmission were prolonged ([@R6]). CEBS was designed to supplement the national surveillance system by training community members to identify, within their own communities, unsafe burials and persons with signs and symptoms compatible with EVD infection. Through its community presence, CEBS was positioned to detect EVD cases that were not epidemiologically linked to other confirmed cases at the time of detection; identification of such cases could provide early warning of new chains of transmission.
A brief pilot study in Bo District during November and December 2014 demonstrated that community leadership accepted CEBS and that CEBS could identify possible EVD cases. Thus, the Ebola Response Consortium, led by the International Rescue Committee and consisting of 15 humanitarian organizations committed to stopping the Ebola virus epidemic, worked with the Sierra Leone Ministry of Health and Sanitation to implement CEBS in 9 of Sierra Leone's 14 districts ([Figure 1](#F1){ref-type="fig"}). CEBS began operations on February 27, 2015, when the surveillance system recorded its first alert.
{#F1}
During February 27--September 30, 2015, we evaluated the ability of CEBS to detect possible EVD cases and unsafe burials in the 9 districts. We also conducted a subanalysis of the system during its first 6 weeks of operation in Kambia District to assess the sensitivity, positive predictive value (PPV), and timeliness of case detection for persons with no epidemiologic links at the time of detection. Among the districts in which CEBS was operational, Kambia District was the only one that experienced ongoing active virus transmission during the subanalysis period.
Methods
=======
Data Collection, Data Flow, and Reporting
-----------------------------------------
The CEBS system included community health monitors, community surveillance supervisors, and community health officers, each of whom received job-specific training in the month before beginning operations. Formal evaluations of staff knowledge were not conducted due to the rapid nature of the system deployment, but some districts provided informal refresher training when possible. Community health monitors were volunteers or existing community health workers who were trained to detect 6 trigger events suggestive of Ebola virus transmission: 1) [\>]{.ul}2 sick or dead members in a household, 2) a sick or dead person after an unsafe burial or corpse washing, 3) a sick or dead health worker or traditional healer, 4) a sick or dead traveler, 5) a sick or dead contact of an EVD patient, or 6) an unsafe burial or corpse washing. A seventh category, "other," was included so that community health monitors could report and describe other unusual events that did not fall under any of the 6 defined trigger events. Community health monitors reported events to their community surveillance supervisors via mobile telephone calls; the supervisors then conducted preliminary investigations. The community health officers, who were trained professionals within the national public health system, often assisted surveillance supervisors with preliminary investigations, but some delegated that responsibility to the supervisors and only assisted when needed. The surveillance supervisors or community health officers then reported events that remained suspect to their local District Ebola Response Center for follow-up.
Community health monitors were responsible for their own village and sometimes, to help ensure adequate coverage, a few small villages within walking distance of where they lived. Surveillance supervisors were assigned a larger area but were provided with motorcycles to facilitate investigations. All CEBS staff were given a mobile telephone with monthly phone credit or a subscription in a prepaid, closed user-group network. Community health monitors were expected to immediately contact their surveillance supervisors to report alerts and to contact them weekly to confirm the absence of reportable events.
When a surveillance supervisor received an alert from a community health monitor, the supervisor recorded the alert information on a standardized form, a weekly alert log. The following data were captured: date; time; trigger event involved; type of alert (sickness, death, unsafe burial, or other); and age, sex, and location of the person(s) concerned. The surveillance supervisor also recorded what, if any, actions were taken to respond to the alert; whether the alert was raised to the District Ebola Response Center; and whether local social mobilization teams were notified to provide health education activities.
Surveillance supervisors submitted alert logs to the CEBS district lead at the end of each week. On a weekly basis, the district lead entered the data into a standardized spreadsheet, checked for duplicate reporting, and submitted the document to the CEBS coordination unit in Freetown, Sierra Leone. The district lead also cross-checked each CEBS alert against those in the District Ebola Response Center alert records and confirmed the final alert status as 1) the identified illness or death did not meet the suspected case definition, 2) the alert involved a suspected or probable case-patient who tested negative, or 3) the alert identified a confirmed case. In Sierra Leone, a suspected case-patient was defined as 1) a person with a fever (temperature \>38°C) who was a known contact of a suspected, probable, or confirmed EVD clinical case-patient; 2) a person with [\>]{.ul}3 EVD-compatible symptoms (e.g., headache, vomiting, and diarrhea) and who had had contact with a clinical case-patient; 3) a person with fever and [\>]{.ul}3 EVD-compatible symptoms; 4) a person with inexplicable bleeding or miscarriage; or 5) a deceased person with an unexplained death. A probable EVD case-patient was defined as a person who was determined likely to have EVD based on clinical or epidemiologic factors. A confirmed case-patient was defined as a person who tested positive for Ebola virus RNA by quantitative reverse transcription PCR or a similar diagnostic test ([@R7]).
By September 30, 2015, CEBS had been implemented by 7,416 community health monitors and 137 surveillance supervisors across 9 districts in Sierra Leone ([Figure 1](#F1){ref-type="fig"}). Implementation was undertaken by the International Rescue Committee (in Bo, Kenema, Kono, and Tonkolili Districts); Save the Children International (in Kailahun and Pujehun Districts); CARE International (in Bombali District); Action contre la Faim (in Kambia and Moyamba Districts); and ABC Development (in Kambia District). The Ebola Response Consortium provided technical assistance for system implementation and operations.
As a new surveillance system, CEBS had no trained staff, field equipment (e.g., telephones and motorbikes), or reporting infrastructure. The startup costs were estimated at US\$1.3 million. Once the system was operational, the monthly costs were ≈US\$129,000, which covered training, telephones, motorbikes, fuel, and incentives.
Methods of Evaluation
---------------------
We described the alerts by type (illness, death, unsafe burial, and other) and by trigger event used. We calculated alert rates and death rates per 100,000 persons per day by district, using 2004 district population estimates ([@R8]). The sensitivity of CEBS for case detection was assessed using the Ministry of Health and Sanitation's surveillance data ([@R9]). Sensitivity was evaluated by dividing the number of CEBS-detected confirmed cases by the total number of confirmed cases detected by the overall surveillance system. PPV of confirmed case detection was determined by dividing the confirmed cases detected by CEBS by the suspected, probable, and confirmed cases detected by CEBS.
During April 13--May 30, 2015 (i.e., from the date CEBS first became operational in Kambia to the end date of our team's field investigations), we evaluated cases in persons in Kambia District who had no identified epidemiologic links at the time of detection. We used the Ministry of Health and Sanitation's Epi Info database to identify all confirmed cases in Kambia during the evaluation period ([@R10]). The database served as the line list of suspected, probable, and confirmed cases in Sierra Leone and integrated information from the epidemiologic investigation, including date of symptom onset and potential risk factors. By interviewing frontline public health and CEBS staff, we were able to determine whether and how CEBS was involved in case detection. We considered case-patients to have no identified epidemiologic links at the time of detection if they were not on the contact list used by the district contact tracers and, therefore, were not being actively monitored. The timeliness of detection of these cases was determined by calculating the interval in days between the date of symptom onset and the date of detection.
This assessment was a part of a nonresearch public health response activity and thus did not undergo institutional review board review. In addition, we used only information that had already been collected for public health surveillance purposes, so informed consent was not obtained.
Results
=======
During February 27--September 30, 2015, a total of 12,126 alerts were reported through CEBS in 9 Sierra Leone districts ([Figure 2](#F2){ref-type="fig"}). Tonkolili was the first district to report on a consistent basis, beginning on February 27, followed by Moyamba (March 6), Pujehun (March 14), Kenema (March 31), Kambia (April 13), Bo (April 18), Kono (April 23), Bombali (April 27), and Kailahun (June 14). From June 14 onward, the districts were collectively reporting an average of 79 alerts per day.
{#F2}
Of the 12,126 alerts reported, 86% (10,421) were for deaths, 14% (1,646) for illnesses, and \<1% for unsafe burials (7) or other (52) ([Table 1](#T1){ref-type="table"}). The alert rate per 100,000 persons per day differed by district and alert type. Of note, Moyamba and Kambia generated the highest rates of death alerts (3.81 and 2.19, respectively), and Kailahun and Bombali reported the highest rates of sick alerts (0.67 and 0.32, respectively) ([Table 2](#T2){ref-type="table"}). The CEBS death reporting rates were substantially lower than the expected crude death rate of 4.66 deaths/100,000 persons/day used by the World Health Organization ([@R11]).
###### CEBS for Ebola virus disease, Sierra Leone, 2015\*
Surveillance variable No. (%)
----------------------- -------------
Alerts, n = 12,126
Death 10,421 (86)
Sickness 1,646 (14)
Unsafe burial 7 (\<1)
Other† 52 (\<1)
Trigger events
---------------------------------------------- -------------
[\>]{.ul}2 persons sick or dead in household 205 (2)
Sickness or death after unsafe burial 59 (\<1)
Sickness or death in HCW 70 (\<1)
Sickness or death in traveler 191 (2)
Sickness or death in contact of case-patient 36 (\<1)
Unsafe burial or washing of corpse 7 (\<1)
Other‡ 11,558 (95)
Cases
----------------------------------- -------------
Suspected, probable, or confirmed 287 (2)
Tested and ruled negative 271
Confirmed 16
Did not meet case definition 10,173 (84)
Not escalated as an alert 774 (6)
Lost to follow-up§ 892 (7)
\*This analysis was conducted during February 27--September 30, 2015. CEBS, community event--based surveillance; HCW, healthcare worker.
†Alerts for unusual events that did not fall under any of the first 6 events listed in the trigger events section of the table.
‡A total of 10,042 (86.9%) of these events were deaths in the community.
§No follow-up or missing information on follow-up.
###### CEBS alert rates by type and district, Sierra Leone, February 27--September 30, 2015\*
------------------ ---------------------- ------------------------ -------------- ------ ------------- ----- ------
District Population estimate† Days of CEBS operation Death alerts Sick alerts
Total no. alerts Rate‡ Total no. alerts Rate‡
Moyamba 278,119 208 2,203 3.81 74 0.13
Bombali 494,139 156 1,137 1.47 250 0.32
Kambia 341,690 170 1,273 2.19 148 0.25
Bo 654,142 165 1,775 1.64 238 0.22
Tonkolili 434,937 215 1,343 1.44 192 0.21
Kenema 653,013 183 1,327 1.11 308 0.26
Kono 325,003 160 573 1.10 48 0.09
Pujehun 335,574 200 432 0.64 108 0.16
------------------ ---------------------- ------------------------ -------------- ------ ------------- ----- ------
Kailahun
465,048
108
358
0.71
339
0.67
\*CEBS (community event--based surveillance) was conducted in 9 of the country's 14 districts.
†Estimates from the 2004 Population and Housing Census: Analytical Report on Population Projection for Sierra Leone (*8*).
‡No. alerts/100,000 persons/d.
In terms of the 6 defined trigger events, the most commonly cited was [\>]{.ul}2 sick or dead household members (n = 205, 2%). Sickness or death of a traveler was the second most cited (n = 191, 2%). In total, the 6 defined trigger events accounted for \<5% of the alerts ([Table 1](#T1){ref-type="table"}). The seventh trigger event category (i.e., other) accounted for the most alerts (n = 11,558, 95%); a total of 10,042 (87%) of the alerts categorized as other were for deaths in the community ([Table 1](#T1){ref-type="table"}). Surveillance supervisors and community health officers escalated 93% of the alerts to the District Ebola Response Centers for follow-up.
A total of 287 (2%) of all persons who triggered alerts met the suspected, probable, or confirmed case definitions. Of these 287 persons, 215 (75%) were detected after death, and 271 (94%) tested negative for EVD. During the study period, 16 confirmed EVD cases were reported by CEBS in Kambia (n = 13) and Tonkolili (n = 3); half of the infected persons were detected while alive. During this same period, the Ministry of Health and Sanitation documented 53 confirmed cases in the CEBS districts. Overall, the sensitivity for confirmed case detection by CEBS was 30% (16/53 confirmed cases), and the PPV was 6% (16/287 suspected, probable, or confirmed cases). Sensitivity was 27% (13/49 confirmed cases) in Kambia and 75% (3/4 confirmed cases) in Tonkolili; PPV was 7% (13/175 suspected, probable, or confirmed cases) in Kambia and 9% (3/33 suspected, probable, or confirmed cases) in Tonkolili.
During the 6-week subanalysis in Kambia, the Ministry of Health and Sanitation database identified 13 confirmed EVD cases in the district. CEBS staff reported 8 of these patients, of whom 7 were alive at the time of the alert. Upon further investigation, we found that 3 of the cases were reported by community health monitors who also served as contact tracers through the contact tracing reporting system. Therefore, CEBS was not the main reporting source for these 3 cases, and the cases were not recorded in the CEBS database or counted as cases detected by CEBS. For the remaining 5 cases, 3 were classified as other trigger events, 1 was in a sick contact of a confirmed case-patient, and 1 was a sick member of a household.
Six of the 13 confirmed case-patients in Kambia had no epidemiologic links when they were detected; CEBS staff identified 4 of these case-patients. The time from symptom onset to detection ranged from 1 to 3 days for the 4 cases identified by CEBS and was 5 and 7 days, respectively, for the 2 cases detected by other components of the national surveillance system ([Table 3](#T3){ref-type="table"}). For the latter 2 cases, 1 of the ill persons lived in a village that was not covered by a community health monitor at the time of detection, and the case was identified after the person had died; the other person resided in a CEBS-covered village, but the case was not detected until the ill person was admitted to a local hospital.
###### Timeliness of identification of confirmed Ebola virus disease cases with no known epidemiologic links to other confirmed cases at detection, Kambia District, Sierra Leone, 2015\*
Detected by CEBS Patient age, y/sex Symptom onset date Detection date Days from onset to detection
------------------ -------------------- -------------------- ---------------- ------------------------------
Yes 52/M Apr 17 Apr 20 3
No 45/M Apr 17 Apr 22 5
Yes 23/F Apr 23 Apr 25 2
Yes 25/F Apr 24 Apr 27 3
No 56/F Apr 23 Apr 30 7
Yes 29/M May 28 May 29 1
\*The data are for April 13--May 30, 2015. CEBS, community event--based surveillance.
During the subanalysis in Kambia, surveillance staff from other districts indicated that the CEBS network had detected additional outbreaks that were not caused by Ebola. Community health monitors identified 2 measles clusters in Kono and 1 measles cluster in Bombali, leading to the initiation of investigations and implementation of control measures, including isolation of ill persons and vaccination of susceptible children. Community health monitors in Bombali and Kono also reported suspected chickenpox clusters.
Discussion
==========
Our evaluation indicates that, during its period of operation, CEBS effectively generated alerts for and detected nearly one third of all EVD cases found in its districts. Although this would rightly be considered a low sensitivity for an independent surveillance system, CEBS was designed to supplement a larger, established system. The low PPV of CEBS also was expected because of the tendency of event-based systems to provide higher sensitivity while generating a large number of false alerts ([@R12]) and because there were few true EVD cases. Ruling out EVD in times of low transmission requires investigation of all alerts and isolation and testing of all suspected case-patients ([@R13]), most of whom will be determined to be uninfected.
Our data from the subanalysis in Kambia are too few to draw a meaningful conclusion, but they suggest that CEBS may be capable of quickly finding cases with no identified epidemiologic links. If this is true, the system could be used to detect the early stages of new infectious disease outbreaks or to rapidly identify the spread of disease to new geographic areas during ongoing outbreaks or epidemics. Nevertheless, even within this small sample of cases, CEBS failed to detect 2 cases with no known epidemiologic links, which highlights the need for adequate coverage of villages by community health monitors, development of stronger links between communities and health monitors, and vigilance by the monitors.
One unexpected finding was that CEBS detected a large number of deaths in the community. Although not intended to serve as a reporting system for community deaths, CEBS did contribute to death reporting, which was a major initiative of the national government and social mobilization programs. By detecting dead bodies that were then tested and found to be negative for Ebola infection, CEBS helped to confirm the lack of virus transmission, thereby providing some evidence that the epidemic had, in fact, ended in a given district. However, death reports are a late indicator of infection and, thus, do not enable isolation of patients early in the disease course, a control measure that could result in reduced transmission ([@R14]*--*[@R16]). The death reporting rates also were considerably lower in most CEBS districts than would be expected based on estimated death rates ([@R11]); consequently, CEBS reporting rates were not a substitute for death surveillance or registration.
Another unexpected finding was the detection of 3 measles outbreaks. Given extensive undervaccination and undertreatment of other communicable diseases during the epidemic, it was expected that a large number of disease outbreaks would go undetected ([@R17]*,*[@R18]), but we did not anticipate that CEBS would detect a few of them. CEBS staff might have detected these clusters because some of the trigger events, such as [\>]{.ul}2 sick or dead persons in a household, were not specific to EVD. The staff also might have identified the clusters because they were looking for signs of illness in their communities, irrespective of the cause or signs or symptoms. This finding provides some indication that community-based surveillance could be used to provide early warning of a variety of diseases of public health concern.
Our evaluation also revealed several critical weaknesses in CEBS, some of which may be due to the rapid implementation of the system. First, community health monitors primarily relied on the trigger category other to classify community deaths and alerts, rather than the defined trigger events that they were taught to seek out. This lack of use of defined trigger events could imply that some of the triggers were not sensitive enough to capture Ebola virus transmission. It is also possible that the staff miscategorized the alerts and that many alerts did, in fact, fit a trigger event category. However, most alerts categorized as other were not reported with sufficient information to assess whether they fit a defined trigger event category. Before future systems are widely implemented, the validity of triggers should be more rigorously tested, and refresher training of staff should be regularly provided to reinforce trigger event recall. Rapid implementation at scale is difficult to achieve while also providing comprehensive training and developing strong links between the community and the surveillance team. Ideally, community-based surveillance should be developed and implemented when a large outbreak is not underway. Such a system would then be in place and available for adaptation if a public health crisis arises.
Another weakness of CEBS is that community health monitors reported relatively few illnesses, which is concerning for a system that aimed to detect illness quickly to reduce opportunities for virus transmission. The low proportion of illness alerts may indicate that the intended meaning of illness was unclear or, more likely, that community health monitors were concerned about negative consequences from the community for reporting an event, particularly if the affected person was not infected with Ebola virus.
A final weakness of note is that CEBS detected few unsafe burials. This lack of reporting could reflect the general challenge faced by EVD surveillance in exposing a cultural tradition that communities intentionally guard closely. However, by February 2015, when awareness of the Ebola virus transmission risks of traditional burials were more fully understood, community members may have bypassed their community health monitors to relay information about unsafe burials directly to the preexisting burial management alert system ([@R19]).
Our evaluation has several limitations. First, we conducted the evaluation during a period of low Ebola virus transmission; therefore, we cannot draw conclusions about how CEBS would perform in a high-transmission environment. Second, the subanalysis in Kambia lasted only 6 weeks and involved only a few cases. Consequently, the results regarding the ability of CEBS to find cases with no identified epidemiologic links at the time of detection cannot be considered conclusive. Additional implementation and evaluation of CEBS in future EVD outbreaks would provide data to assess the relative merits of this approach. Third, we could not analyze the sensitivity of CEBS to detect other disease outbreaks because no reference standard exists to inform the denominator, and no reporting mechanism exists within CEBS to inform the numerator. Fourth, CEBS was implemented primarily in rural settings, so we do not know how the system would perform in a densely populated urban setting, such as a capital city. Last, an anthropologic understanding of the lack of illness and burial reporting would inform a more comprehensive interpretation of these results.
The Sierra Leone Ministry of Health and Sanitation plans to use community-based surveillance as part of its Integrated Disease Surveillance and Response system, which is tasked with detecting and responding to several priority diseases, conditions, and events ([@R20]). Community-based surveillance could extend disease surveillance beyond district health clinics to the village level and provide an early warning function. This would contribute to meeting the core capacity requirements of the International Health Regulations to detect and report disease at the community level to facilitate the immediate implementation of control measures before an outbreak expands further ([@R21]*,*[@R22]). Nonetheless, our evaluation reveals several challenges that should be addressed. Detailed assessments should be undertaken to determine how community health monitors recognize and categorize symptomatic illness and the barriers to their ability and willingness to report illness. The assessment results should then be applied to refine trigger definitions and processes. Given that valid disease measures are the basis of an effective surveillance system, these issues are the most pressing ones that need to be studied and addressed to strengthen future iterations of community-based surveillance. In addition, trigger definitions should remain simple to ensure that community health monitors can understand and correctly apply them, which may mean that a few salient, event-based triggers would be more effective than several case-based, specific triggers. Alternatively, more extensive, regularly repeated training of community health monitors might be needed to ensure adequate recall and reporting of more complicated triggers. Furthermore, to sustain efficacy and performance, community-based surveillance must be fully integrated into the overall surveillance system and adequately supported to ensure response capacity.
*Suggested citation for this article*: Ratnayake R, Crowe SJ, Jasperse J, Privette G, Stone E, Miller L, et al. Assessment of community event--based surveillance for Ebola virus disease, Sierra Leone, 2015. Emerg Infect Dis. 2016 Aug \[*date cited*\]. <http://dx.doi.org/10.3201/eid2208.160205>
These first authors contributed equally to this article.
We thank the Ebola Response Consortium partners who implemented CEBS: ABC Development, Action contre la Faim, CARE International, International Rescue Committee, and Save the Children International.
CEBS was supported by Department for International Development and the US Agency for International Development's Office of Foreign Disaster Assistance.
Mr. Ratnayake is an epidemiologist with the International Rescue Committee. His research interests include epidemiologic methods, epidemic control, humanitarian emergencies, and surveillance among vulnerable populations.
| {
"pile_set_name": "PubMed Central"
} |
Gorham\'s syndrome or vanishing bone disease is a rare disorder with massive osteolysis and idiopathic fractures, which may result in neurological and respiratory complications. For anesthetist, Gorham\'s syndrome presents various challenges such as long duration, blood loss, air embolism, positioning, and respiratory complications. We present a case of an 18-year-old male, who was diagnosed with the disease 2 years back, following a fall. He developed numbness and tingling sensation over chest wall with motor weakness in both lower limbs, with a gradual change in shape of the chest with increasing scoliosis of spine over the next few months. He was operated for decompression of dorsal vertebrae and grafted with mother\'s fibula. Postoperatively, excisional specimen was sent to histopathology confirming the diagnosis of Gorham\'s syndrome.
The patient suffered a fall again and came with power grade 4/5 in lower limbs, which progressed to 1/5 within 1 week of admission. Airway examination revealed a normal mouth opening and neck movements. On palpation, there was a retraction of right upper hemithorax and the upper ribs could not be palpated, and scoliosis toward right side was noted. Chest X-ray revealed right 4-8^th^ rib\'s destruction on anterior, posterior aspect \[Figures [1](#F1){ref-type="fig"}\]. Magnetic resonance imaging spine: Severe scoliosis of the dorsal spine with convexity to the right. D2-8 vertebra appears osteoporotic with bony erosions. Pulmonary function tests show restrictive lung disease with total lung capacity 38% of normal forced vital capacity, and forced expiratory volume in 1 s 38% and 41% of normal. Arterial blood gas showed normal values.
{#F1}
Inside theatre, anesthesia was induced with midazolam 1.5 mg, fentanyl 150 mcg and propofol 120 mg, and rocuronium 50 mg. He was intubated with 8.5 mm endotracheal tube with manual inline stabilization. The right internal jugular vein was cannulated. The patient was given lateral position for the surgery. Maintenance was with O~2~:N~2~O 50:50 as per institutional protocol, isoflurane 1 minimum alveolar concentration, dexmedetomidine infusion at 0.7 mic/kg/h and buprenorphine 120 mcg. Volume control mode was used with a tidal volume of 450 ml, respiratory rate of 16/min, and PEEP of 5 cm. Maximum peak airway pressures intraoperatively were 17 cm of H~2~O. There was no intraoperative episode of hypotension, desaturation, or fall in end-tidal CO~2~.
Intraoperative blood loss was 1000 ml, which was replaced with five pints of ringer lactate and 500 ml of colloid. No blood products were used as preoperative, hemoglobin was 15.4 and maximum allowable blood loss was calculated to be 2343 ml. Duration of surgery was 4 h. The patient was extubated uneventfully.
Gorham\'s syndrome is a disease of unknown etiology, most commonly in the second and third decades with equal sex distribution. The pathologic process shows replacement of normal bone by an aggressively expanding but non neoplastic vascular tissue similar to hemangioma or lymphangioma.\[[@ref1]\] Patients present with pain in the involved region with pathological fractures. Diagnosis is by clinical examination, X-ray and histopathological findings. The treatment includes radiation therapy, bisphosphonates, and alpha-2b interferon.\[[@ref2]\] Definitive treatment is by surgery.
For anesthetist, Gorham\'s syndrome presents various challenges. Complications specific to surgery such as long duration, blood loss, air embolism, and positioning and those related to the disease such as respiratory complications, chylothorax, and effusions. If there is restrictive lung disease mainly due to the disease process and accompanying scoliosis, ventilation using low tidal volume and high respiratory rate, pressure-controlled ventilation seems more appropriate. Postoperative ventilatory problem leading to re-intubation and prolonged ventilation may be seen. Extubation has to be planned carefully with preparation for prolonged Intensive Care Unit management.\[[@ref3]\]
Thus to conclude, Gorham\'s disease is a challenge for both the surgeon and anesthetist, and a team effort is needed to manage these cases successfully.
Financial support and sponsorship {#sec2-1}
=================================
Nil.
Conflicts of interest {#sec2-2}
=====================
There are no conflicts of interest.
| {
"pile_set_name": "PubMed Central"
} |
Would a flower by any other name always smell as sweet? Maybe not, if air pollution has anything to do with it. Researchers at the University of Virginia in Charlottesville report that three common constituents of smog destroy floral scents released by flowers to attract bees and other pollinators. In fact, flower scents traveled four times farther in the 1840s, when European scientists first began documenting ozone pollution, than they do under today's air conditions, according to modeled simulations run by the researchers. Because pollinating insects rely partly on scents to find flowers, the loss of fragrant plumes could make it harder for insects to locate pollen sources, jeopardizing pollinators and crops alike. The researchers are beginning field trials this month to see if real-world tracking of airborne floral scents match the results predicted by the model.
The vitality of plants depends on pollination, according to principal investigator Jose Fuentes, a meteorologist. Additionally, if pollinators are forced to spend more time foraging for food yet collect less pollen to feed their young, insect colonies may suffer nutritionally. Both problems could impact our food supply. "We need to preserve pollinators because they provide useful services," Fuentes says.
It had already been established that when fragrance molecules wafting downwind meet up with air pollutants, chemical reactions alter the floral scents and contribute to production of compounds such as acetone, formaldehyde, and carbon monoxide. In the current study, Fuentes and colleagues Quinn McFrederick and James Kathilankal estimated the fate of three common volatile hydrocarbons emitted by flowers as they encountered increasing levels of ozone, hydroxyl radicals, and nitrate radicals. In the summer of 2002, the researchers measured temperature, wind turbulence, and other factors that determine the rate of floral scent emission and movement at an experimental farm in Virginia. They observed how these factors affected the release of scent from a plot of snapdragons that grew wild on the farm.
The researchers plugged these data into a model to test different air pollution scenarios ranging from conditions that prevailed during the 1840s to current summertime conditions in large eastern U.S. cities, where ozone levels can exceed 120 ppb by volume. Under 1840s conditions, only 20% of scents were altered by chemical reactions within a 1,000-meter radius downwind of the floral source. However, even slight elevations in pollutants---comparable to air quality today in rural areas with little industrial emissions---altered more than 40% of the scents within a 500-meter radius of the floral source. In the most polluted scenario, only 25% of the scents survived 300 meters downwind. Fuentes and colleagues reported the findings in volume 42, issue 10 (March 2008) of *Atmospheric Environment*.
Do compounds generated when floral scents are chemically altered actually worsen air pollution? "We haven't thought about this in \[terms of\] air quality," Fuentes says. He says the findings also raise special concerns about the fate of nighttime pollinators such as moths, which rely largely on scent to find flowers (in contrast to bees, which use both color and scent during daytime).
Fuentes cautions against applying the new findings to colony collapse disorder (CCD), an unexplained phenomenon that has decimated bee colonies in the past two years. He says some of the best evidence to date suggests CCD is more directly related to infectious agents and pesticides. However, he says, any effects from air pollution "are probably an added stress that bees have to cope with."
Other insect behaviors that orient by chemical scents, such as beetles mating, also could be disturbed by air pollution. On the other hand, some plants may benefit from disrupted airborne signals: "If insects can't smell plants, they can't come to eat them," says Jay Evans, a research entomologist at the U.S. Department of Agriculture, who calls the Fuentes study "a beautiful mix of good ecology and chemistry."
Pollinators aren't the only species dealing with the olfactory effects of air pollution. In the January 2006 issue of *Chemical Senses*, Robyn Hudson and colleagues reported that residents of Mexico City, which has some of the world's worst air pollution, were significantly less able to detect and distinguish between food odors than were residents of Tlaxcala, a geographically similar area with much lower air pollution. The difference was observed in multiple age groups, even when smokers were removed from the analysis, strongly suggesting a link with pollutants that may damage the olfactory epithelium.
{#f1-ehp0116-a00334}
| {
"pile_set_name": "PubMed Central"
} |
All relevant data are within the paper and its Supporting Information files.
Introduction {#sec001}
============
The impact of the increase in temperature due to climate change has been a worldwide concern for several research areas \[[@pone.0191273.ref001]--[@pone.0191273.ref007]\]. Among the various potential impacts of climate change, those of special concern are on resources essential to human wellbeing, like water availability and agriculture, with direct impacts on economic activities and food security \[[@pone.0191273.ref008]\].
Agriculture is among the human activities most vulnerable to climate change \[[@pone.0191273.ref009], [@pone.0191273.ref010]\]. More specifically, the impacts are related to the shortening of the growth and flowering season, along with reduction of the number and size of grains, as well as the total yield \[[@pone.0191273.ref011], [@pone.0191273.ref012]\]. Consequently, in many countries, the agricultural productivity of crops is impacted in some way by the increase in temperature \[[@pone.0191273.ref013], [@pone.0191273.ref014]\], and many models have predicted that the yield of many crops will continue to decrease in future climatic scenarios \[[@pone.0191273.ref015], [@pone.0191273.ref016]\]. Thus, understanding how climate change can affect agriculture has now become a key step in guaranteeing global food security.
The negative impact of climate change on crop yields is expected to be more severe in low latitudes, affecting mainly developing countries \[[@pone.0191273.ref017]\]. Countries in these areas, which are highly dependent on agricultural activity, such as Brazil, will suffer great economic damage due to the effect of climate change on crops \[[@pone.0191273.ref018],[@pone.0191273.ref019]\]. However, the impact of climate change at low latitudes will probably have consequences for food supply at a global scale, as some countries in the low latitudes are important producers of food grains.
Many studies have investigated the impact of global climate change on the biology and productivity of agricultural plant species \[[@pone.0191273.ref010], [@pone.0191273.ref014], [@pone.0191273.ref017]--[@pone.0191273.ref021]\]. Moreover, the availability of environmentally suitable areas is another important agricultural factor that can be affected by climate change. Although contemporary agriculture benefits from investments in high-tech plant production, climate is still critical to the productivity and distribution of the species \[[@pone.0191273.ref022]\]. Thus, investigating the relative effect of climate and technology on agricultural productivity can help in understanding the impact of global climate change. Despite the fact that the impact of climate change on the environmental suitability of a species has been known for a long time, only recently studies have investigated the impact of climate change on the geographic distribution of agricultural plant species \[[@pone.0191273.ref010], [@pone.0191273.ref023]--[@pone.0191273.ref025]\].
On large spatial scales, species density and distribution are mainly constrained by the climatic conditions in which species can survive \[[@pone.0191273.ref026],[@pone.0191273.ref027]\]. Considering that species occur in all climatically suitable places, and are absent from all unsuitable places (i.e., species are in equilibrium with the actual climate; \[[@pone.0191273.ref028],[@pone.0191273.ref029]\]), changes in climatic variables will change the suitability of the areas for the species, leading to changes in species density and distribution. As environmental suitability and species density have been shown to be positively correlated \[[@pone.0191273.ref030]\], future climate change can alter the geographical pattern of distribution and the production of many crop species \[[@pone.0191273.ref031]--[@pone.0191273.ref033]\].
Ecological niche models (ENMs hereafter) have been one of the most widely used statistical and computational tools to evaluate the environmental suitability of species throughout their geographic range and to generate hypotheses about shifts in their distribution \[[@pone.0191273.ref034]--[@pone.0191273.ref036]\]. These models relate the occurrence of species with environmental variables, generating response curves that can be used to predict environmental suitability in other areas or over time \[[@pone.0191273.ref035]\]. ENMs have been usefully applied for a variety of ecological purposes, including assessing the impacts of climate change on species' environmental suitability and distribution \[[@pone.0191273.ref037],[@pone.0191273.ref038]\]. Due to the potential of ENMs to provide information about the spatial structure of species' environmental suitability over different climates and time, these models have been used to model responses in agricultural species in order to evaluate the impacts of climate change on the future yield.
Agricultural species have been more frequently evaluated using more specific and "mechanistic" models (see Nabout et al. \[[@pone.0191273.ref023]\] and Estes et al. \[[@pone.0191273.ref039]\], for comparisons). However, because of their generality and simplicity, correlative ENMs can still be quite useful for quick and empirical predictions. Moreover, studies that associate environmental suitability (obtained by ENM) with demographic aspects of the species (e.g., productivity, density, and frequently others) have used correlation or linear regression models, with results varying on the support or not of this relationship (e.g., Nabout et al. \[[@pone.0191273.ref040], [@pone.0191273.ref041]\], Torres et al. \[[@pone.0191273.ref042]\] Weber et al. \[[@pone.0191273.ref043]\], Filz et al. \[[@pone.0191273.ref044]\], Jiménez-Valverde et al. \[[@pone.0191273.ref045]\]). However, the relationship of species suitability and its yield may present non-stationary geographical patterns, i.e., can systematically vary from one region to another \[[@pone.0191273.ref046]\]. For example, it is possible that, in areas with similar suitability values, different agricultural practices can lead to differences in yield. Consequently, linear and regression models that assume a constant relationship over space cannot account for this spatial variability. Thus, geographically weighted regression (GWR) may be a more appropriate approach \[[@pone.0191273.ref047]\] to evaluate the relationship between species suitability and productivity. The GWR is a statistical technique that captures spatial differences in the relationship of variables by allowing the modelling of processes to vary over the study area. This analysis is frequently used in broad-scale ecological studies (e.g., Terribile and Diniz-Filho \[[@pone.0191273.ref048]\], Eme et al. \[[@pone.0191273.ref049]\]), and thus its combination with ENM may be useful to evaluate the changes in environmental suitability over the space \[[@pone.0191273.ref046]\].
In economic and social terms, soybeans (*Glycine max* (L.) Merrill) are a globally important crop. Soybean production is one of the greatest among the oil crops and is a relevant source of income for producers \[[@pone.0191273.ref050]\]. In addition, soybeans represents a significant source of protein and calories in human nutrition \[[@pone.0191273.ref051]\].
Soybeans arrived in Brazil via the United States in 1882, and the first soybean crop in Brazil was recorded in 1901 in Rio Grande do Sul state (in the extreme south), where there were suitable conditions to develop and expand soybean cultivation. Thus, it is likely that the climate conditions prevailing in southernmost Brazil are similar to those of the original distribution of soybeans in the southern US \[[@pone.0191273.ref052]\]. Additionally, apart from favourable environmental conditions, in the mid-1950s, an official programme encouraged national wheat cultivation in the southern region \[[@pone.0191273.ref052]\]. It also provided an incentive for soybean cultivation, which was regarded as the best summer alternative to succeed the wheat crop planted in the winter \[[@pone.0191273.ref053]\]. In the 1960s, soybeans were established as an economically important crop in Brazil, and, in the following decade, soybeans became the main crop of Brazilian agribusiness \[[@pone.0191273.ref052], [@pone.0191273.ref054]\]. Currently, Brazil is the second largest worldwide soybean producer, and the largest in Latin America \[[@pone.0191273.ref055]\], producing 96,228 thousand tons in an area of \~32,000 thousand hectares (56% of the cultivated areas in Brazil) \[[@pone.0191273.ref056]\]. Thus, considering the importance of soybean agricultural activity for the global economy, and climate change possibly shifting the present agricultural scenario, here, we aim to evaluate the relative effect of climate (as indicated by ENMs) and agricultural technology on actual soybean productivity in Brazilian municipalities. Moreover, we estimate the future geographic distribution of soybeans, coupling ENM with agricultural technology variables and, using spatial analyses, we are able to address complex local and non-stationary patterns (the Geographically Weighted Regression--GWR).
Materials and methods {#sec002}
=====================
Productivity data and areas of soybean presence {#sec003}
-----------------------------------------------
Soybean productivity for each Brazilian municipality was calculated by the ratio of the production (in tons) to harvested area (in hectares). Productivity values are given in tons per hectare (ton ha^-1^) for the period of 1994--2010. Production and harvested area values were obtained using the IBGE Automatic Recovery System ("Sistema IBGE de Recuperação Automática"---SIDRA; IBGE 2016 \[[@pone.0191273.ref057]\]) data bank. We identified a total of 2,304 Brazilian municipalities producers of soybean in this period. Average soybean productivity values (i.e., 1994--2010) are higher in the mid-southern region of Brazil ([Fig 1A](#pone.0191273.g001){ref-type="fig"}). The average soybean productivity distribution, as compared to the number of soybean producing municipalities ([Fig 1B](#pone.0191273.g001){ref-type="fig"}), has an asymmetric distribution, with few municipalities having the highest production. The productivity of 597 municipalities (\~25% of total) ranged from 2.01 to 2.50 ton ha^-1^. Seven municipalities located in the southern region have the highest average productivity values (3.01 to 3.50 ton ha^-1^). Additionally, 174 of the 329 municipalities with average productivity, ranging from 2.51 to 3.00 ton ha^-1^, are found in Paraná (see [S1 Table](#pone.0191273.s001){ref-type="supplementary-material"} for productivity by municipalities).
{#pone.0191273.g001}
The geographic coordinates (i.e., longitude and latitude) of the central town of each one of the Brazilian and other Latin American municipalities with average soybean productivity greater than zero ([S1 Table](#pone.0191273.s001){ref-type="supplementary-material"}) were used in ENMs. Thus, a total of 2,493 occurrence records distributed in Argentina, Paraguay, Bolivia, Ecuador, Venezuela and Brazil were obtained (see Figure A in [S1 File](#pone.0191273.s002){ref-type="supplementary-material"}). These records were overlaid on a spatial grid with 2,842 cells with a resolution of 0.5° x 0.5° covering the Neotropical region (see below). The other spatial analysis (eg., GWR, see below) were restricted to Brazil because they require records of soybean production for the Brazilian municipalities.
The efforts like genetic improvement and choice of cultivars, among other factors, can increase productivity and reduce climatic effects on productivity. From Brazilian soybean data, it is not possible to determine which variety was planted in each municipality. Additionally, the productivity data are temporal, but the cultivars may have changed over the years. In present paper we used occurrence records of soybean distributed along the Latin America to better describe the overall ecological niche of the species by ENMs (see below), therefore we contemplated all varieties used in each municipalities. Theses different varieties and their adjustments of local climatic conditions were used in ENMs. Therefore, the ENMs generated can be used to understand the current Soybean's productivity of municipalities, however, for future climate scenarios, the ENMs should be evaluated with caution, since new varieties may emerge.
Agricultural technology data for each municipality {#sec004}
--------------------------------------------------
Technological attributes can influence agricultural productivity, therefore, information on the technological efforts of each municipality can be used as surrogates of technological investment (equipment, specialized cultivars, among others) that would affect soybean productivity.
We selected 11 variables that indicate agricultural technological advances to be used in our models: 1) Individuals working in agricultural establishments; 2) Number of agricultural machinery equipment and implements existing in family agriculture establishments; 3) Number of agricultural machinery equipment and implements existing in non-family agricultural establishments; 4) Number of agricultural establishments that use electric energy; 5) Area of agricultural establishments (hectares); 6) Number of agricultural establishments with pest control and/or plant diseases; 7) Number of agricultural establishments with water resources; 8) Number of agricultural establishments using fertilizers; 9) Number of agricultural establishments using agrochemicals; 10) Area of agricultural establishments with irrigation use (hectares); 11) Number of agricultural establishments with irrigation use. The data were obtained from IBGE for the year 2006 (most updated census in Brazil).
Ecological niche models {#sec005}
-----------------------
For modelling soybean suitability, we used climatic data scaled for the same grid used for the records. We obtained the bioclimatic variables for the pre-industrial time period by using simulations for the middle of the eighteenth century stabilized across a 200-year time period to represent current climatic conditions. Future climatic conditions for 2080--2100, a 20-year average for the end of the century, were based on the RCP4.5 emission scenarios, which were derived from four coupled Atmosphere-Ocean General Circulation Models (AOGCM). These were the Community Climate System Model (CCSM4), Centre National de Recherches Météorologiques (CNMR), Marine-Earth Science and Technology-National Institute for Environmental Studies (MIROC-ESM) and the Meteorological Research Institute (MRI-CGCM3). Climate data used in ENMs were obtained from EcoClimate database (ecoclimate.org \[[@pone.0191273.ref058]\]).
To generate ENMs, we selected five bioclimatic variables from 19 available variables (see Hijmans et al. \[[@pone.0191273.ref059]\]). This was based on a factor analysis that considered the correlation among the variables in a manner that minimized their collinearity (see Terribile et al. \[[@pone.0191273.ref060]\]). The climatic variables used in the ENMs were average annual temperature, extent of annual temperature, precipitation in the wettest month, precipitation in the driest month and precipitation in the warmest quarter. Additionally, we used soil pH for the depth of 30--100 cm (data obtained from the Harmonized World Soil Database--version 1.1, FAO/ IIASA/ISRIC/ISS-CAS/JRC 2009). Soil acidity is one of the major constraints of soybean production in Brazilian soils, as the acidity of these soils can affect plant growth and development by interfering with the availability of many nutrients required by plants, such as nitrogen, phosphorus and potassium \[[@pone.0191273.ref061]\]. In addition, it has been previously demonstrated that adding the pH variable improves ENMs for plant species \[[@pone.0191273.ref060],[@pone.0191273.ref062],[@pone.0191273.ref063]\]. Soil pH was treated as a continuous variable, used as a constraint variable and was kept constant over the time periods. The mean and amplitude values of the variables used can be found in Table A in [S1 File](#pone.0191273.s002){ref-type="supplementary-material"}.
We used ensemble methodologies \[[@pone.0191273.ref064]\] to determine the environmental suitability for soybeans in the present time period and to project it to the future. We followed the ensemble protocol proposed by Diniz-Filho et al. \[[@pone.0191273.ref065]\] and used by several recent papers dealing with ensembled ENM \[[@pone.0191273.ref063], [@pone.0191273.ref066], [@pone.0191273.ref067], [@pone.0191273.ref068], [@pone.0191273.ref069]\]. Twelve different ENMs were used, including six presence-only methods (i.e., BIOCLIM, Euclidian, Gower, Mahalanobis distances, Genetic Algorithm for Rule Set Production---GARP, and Maximum Entropy---MAXENT) and six presence-absence methods (i.e., Generalized Linear Models---GLM, Random Forest, Generalized Additive Models---GAM, Flexible Discriminant Analysis---FDA, Ecological Niche Factor Analysis---ENFA, and Neural Network). Franklin \[[@pone.0191273.ref070]\] and Peterson \[[@pone.0191273.ref035]\] provide general descriptions of the methods.
For model comparison, in both types of ENMs, i.e., presence-only and presence-absence, we used the same pseudo-absence data, but in the presence-only ENMs, pseudo-absences were used as background \[[@pone.0191273.ref069], [@pone.0191273.ref071]\]. We randomly divided the presence of each species and their pseudo-absences, which were randomly selected for a background region with the same proportion of species records (i.e., with a prevalence of 0.5), into 75% for calibration and 25% for evaluation and repeated this process 50 times. Based on thresholds determined by the ROC curve, the 2,400 resulting models (i.e., 50 cross-validation x 12 ENMs x four AOGCMs) were used to generate consensual occurrence maps. The ROC curve was based on the soybean frequency of occurrence in each neotropical grid cell, which was obtained from each ENM for every AOGCM (i.e., resulting in 48 frequency maps, from 12 ENMs x four AOGCMs) (for methodological details, see \[[@pone.0191273.ref063], [@pone.0191273.ref067], [@pone.0191273.ref069]\]. These 48 frequency maps were ensembled by their average and weighted by the True Skill Statistic (TSS, \[[@pone.0191273.ref072]\]; Table B in [S1 File](#pone.0191273.s002){ref-type="supplementary-material"}) into a single frequency map. This map was used as a measure of environmental suitability for soybeans across the neotropical region, ranging from 0, with no environmental suitability (i.e., the cell has no occurrence in any of the 48 models from the 50 randomizations) to 1, with maximum environmental suitability (i.e., the cell occurred in all models). The analyses were performed in the computational platform BioEnsembles \[[@pone.0191273.ref063], [@pone.0191273.ref065], [@pone.0191273.ref066], [@pone.0191273.ref067], [@pone.0191273.ref069], [@pone.0191273.ref071]\].
To inspect the main changes in suitability from the current to the future climate, we calculate the differences between the current and future suitability. Negative values indicate that municipalities will lose environmental suitability for soybean cultivation in the future scenario, whereas positive values denote the opposite. Zero values suggest that municipalities will not change their climate suitability values from the current scenario relative to the future.
In addition, we assessed the environmental similarity between calibration data and future climatic scenarios using extrapolation detection (EXDet) software \[[@pone.0191273.ref073]\]. The analysis detects two types of non-analogue conditions: type 1- points that fall outside the range covariates and type 2- points that are within the univariate range of climatic conditions but that constitute novel combinations of covariates \[[@pone.0191273.ref073]\]. This analysis identifies areas where models have extrapolated when making predictions and can help to more carefully interpret SDM results \[[@pone.0191273.ref074]\].
Assessing the relative effect of environmental suitability and technology on soybean productivity {#sec006}
-------------------------------------------------------------------------------------------------
Soybean productivity, environmental suitability and all eleven technological variables were Z-transformed to facilitate the calculation of standardized local regression coefficients in a geographically weighted regression (GWR). Moreover, we used forward-selection for selected technological variables, so 7 out of the 11 initial variables were used (the variables 1, 2, 3, 4, 9, 10 and 11, described above).
We used GWR to investigate the relative influence of climate (suitability of ENM) and agricultural technology (seven variables) in explaining the spatial variation of soybean productivity. The GWR generates a regression model for each municipality, by using the entire dataset, and weighting the regression model by a geographic function, based on the spatial proximity of all other municipalities (see more details in Fotheringham et al. \[[@pone.0191273.ref047]\]). Thus, for each municipality, it was possible to determine all regression parameters (intercept, slope and R^2^). We performed three types of GWR: i) Using all predictor variables (GWRall); ii) Using only climate variables (GWRsuit); iii) Using only technological variables (GWRtec). In this way, we used the variation partitioning technique (see Legendre & Legendre 2012 \[[@pone.0191273.ref075]\]) based on the adjusted R^2^ to determine the relative influence of each predictor. The difference between GWRall and GWRtec indicates the pure effect of the climate variables, and the difference between GWRall and GWRsuit indicates the pure effect of the technology variables. The shared effect was determined as the difference between GWRtec and the pure effect of Technology (or GWRsuit and the pure effect of climate). The adjusted R^2^ of each pure effect and shared variables could thus be mapped and used to investigate how the relationship between the variables change across geographic space. To fit the GWR, we used a Bisquare spatial weight function with a fixed bandwidth of 5°. We quantified the improvement of the GWR with respect to an Ordinary Least Square (OLS) model using the adjusted Akaike Information Criterion (AICc). The GWR analyses were performed using Spatial Analysis in Macroecology (SAM), freely available at <http://www.ecoevo.ufg.br/sam> \[[@pone.0191273.ref076]\]. All shapefiles used in our figures are freely available at Ministério do Meio Ambiente (<http://mapas.mma.gov.br/i3geo/datadownload.htm>).
Results {#sec007}
=======
Ecological niche modelling {#sec008}
--------------------------
In the current scenario, the environmental suitability for soybean cultivation ([Fig 2A](#pone.0191273.g002){ref-type="fig"}) reached its highest values in the central and southeast regions of Brazil, with values above 0.9 being considered favourable for soybean cultivation. On the other hand, northern and northeastern Brazilian municipalities had the lowest environmental suitability values, ranging from 0.0 to 0.2.
{#pone.0191273.g002}
Climatically suitable areas in the future scenario ([Fig 2](#pone.0191273.g002){ref-type="fig"}) tend to reach their highest values in the south and southeast of Brazil, with values near 0.98. Actually, relative to the current scenario, over 72% of the Brazilian territory may lose environmental suitability for soybean cultivation in the future climate change scenario ([Fig 2](#pone.0191273.g002){ref-type="fig"}).
All future scenarios showed non-analogue climates to the current climate data of soybean areas (Figure B in [S1 File](#pone.0191273.s002){ref-type="supplementary-material"}). The type 1 novelty (at least one variable falling outside the training range) occurred in all scenarios. The MIROC climate model had the most regions with non-analogue climates. Type 2 novelty (new combinations of variables values) was rare, occurring in only 3 cells of the GISS and MIROC models.
Geographically weighted regression {#sec009}
----------------------------------
The actual environmental suitability and technology explained 30% of soybean productivity (GWRs, adjusted R^2^ = 0.3). Moreover, when comparing the GWR and OLS models (adjusted R^2^ OLS = 0.2), the AICc results showed that the GWR local model was significantly more appropriate than the OLS global model (F = 28.6 and P \< 0.001, for the improvement of GWR over OLS). Using variation partitioning of the adjusted R^2^ obtained by GWR, we found that pure environmental suitability explained 5% (R^2^ = 0.05 P\<0.05), and pure agricultural technology explained 11% (adjusted R^2^ = 0.11; P\<0,05) of soybean productivity. The shared effect was 14% (adjusted R^2^ = 0.14). Moreover, partitioning for each municipality showed that the largest R^2^ occurred in a municipality in the southern and northern region of Brazil, with values widespread between the overall range of 0 to 0.1 ([Fig 3A](#pone.0191273.g003){ref-type="fig"}) On the other hand, the pure technology effect ranged from 0.1 to 0.44, with most of the highest values occurring in the north of the country ([Fig 3B](#pone.0191273.g003){ref-type="fig"}). The shared effect (climate and technology) showed the highest values in northern Brazil ([Fig 3C](#pone.0191273.g003){ref-type="fig"}). Thus, the GWR demonstrated that the influence of climate and agricultural technology on soybean productivity varies throughout Brazil. Nonetheless, technology was shown to be the most important factor to explain productivity values. Moreover, the regions with low environmental suitability (northern regions) showed the highest importance of agricultural technology.
{#pone.0191273.g003}
Discussion {#sec010}
==========
Here, we showed that global climate change will impact the geographic distribution of soybeans in Brazil, mainly in the central and northern regions. Moreover, the productivity of municipalities was affected by climate and agricultural technology, though in different proportions for each predictor.
The use of ENMs to predict productivity using a linear model has been commonly used in many previous analyses (see Nabout et al. \[[@pone.0191273.ref040]\], for example, using only climate variables). However, ENMs can be biased because they assume a stationary spatial process. In other words, the relationship between predictors and species presence remains constant across the entire area under study. However, this is usually not the case, and non-stationary patterns are commonly found in geographical analyses \[[@pone.0191273.ref077]\]. Thus, techniques like GWR can be more useful than common regression to better evaluate the geographical variation of productivity. Moreover, the variation partitioning of the R^2^ of GWR can show the unique and shared effect of each predictor across the geographic space. In our analyses, climate was more important in explaining the productivity of soybeans in southern Brazil. This was where soybean cultivation was introduced to Brazil \[[@pone.0191273.ref052], [@pone.0191273.ref053]\], and thus, it is expected that this region would be more affected by climate variations. Nevertheless, this geographic area of Brazil (southern) has strong technological advances, which helps explaining the high soybean productivity in this region. This fact actually tends to support the overall use of niche models to predict the future distribution and cultivation of soybeans, coupled with information about the advances in technology applied to agriculture.
The investment in agricultural technology has been important in increasing the productivity of agriculture by reducing the negative effects of diseases, pests and climate change \[[@pone.0191273.ref078]\]. Moreover, the risks of climate change have intensified the importance of technology in agriculture \[[@pone.0191273.ref079]\]. In the present paper, technology was the main variable explaining soybean productivity in Brazil, mainly in northern Brazil. The introduction of soybeans in this region is more recent \[[@pone.0191273.ref052], [@pone.0191273.ref053]\], and technology has allowed productivity in a region with low environmental suitability. Despite the technological advances, climate change can lead to large socioeconomic changes (e.g. \[[@pone.0191273.ref080]\]), mainly because the technological resources demand financial expenses and the small farmer can have financial difficulty to have access to the technological resources \[[@pone.0191273.ref081]\].
In the present paper, the ecological niche models indicated vulnerable areas as well as new areas favourable to the cultivation of soybeans in Brazil. However, considering the impacts of climate change on ecosystems and biodiversity, we do not recommend new areas for the cultivation of soybeans. The conflict between the needs of food production and biodiversity conservation is normally more impactful on biodiversity and increases conflict around conservation (see \[[@pone.0191273.ref082]\]). For example, Bradley et al. \[[@pone.0191273.ref083]\] found that 40% of the areas surrounding protected areas in South Africa can be affected by changes in crop suitability, increasing threats for conservation and generating social conflicts. Our results show that the higher values of future environmental suitability for soybeans in Brazil would be located in the Atlantic forest biome, considered a biodiversity hotspot \[[@pone.0191273.ref084]\]. It would be, at the very least, questionable to consider expanding soybean cultivation to these areas as an option due to the priority given to Atlantic forest conservation. In addition, the Atlantic forest is the most populous region of Brazil, sheltering approximately 72% of Brazilian population \[[@pone.0191273.ref085]\], which means that, in this area, less space at much higher economic cost would be available for cultivation.
A solution we advocate to overcome the impacts of climatic change on crop production is the adoption of technological strategies to adapt to climate impacts in a way that increases the production of actual cultivation areas or, at least, minimizes productivity losses before the negative impacts become too severe and expensive to be reversed \[[@pone.0191273.ref016]\]. Examples would be altering the sowing of the species in order to avoid water stress during the initial growing phases \[[@pone.0191273.ref086], [@pone.0191273.ref087]\], the development of new cultivars and hybrids tolerant to local abiotic stresses (e.g., droughts and extreme temperatures) \[[@pone.0191273.ref012], [@pone.0191273.ref088]\], intercropping with different crops \[[@pone.0191273.ref021]\], the use of cover crops to diminish soil warming, and changing the spacing and seed rate \[[@pone.0191273.ref014]\]. More recently, the use of Ecosystem-based Adaptation (EbA) practices can potentially assist farmers in reducing the impacts of climate change on agricultural production (see \[[@pone.0191273.ref089], [@pone.0191273.ref090]\]). EbA practices are based on conservation, management of biodiversity and ecosystem services in a way that reduces the effects of climate change and promotes social, economic and environmental improvement \[[@pone.0191273.ref091]\]. Studies about how and which climate variables will change in the future, and how soybean varieties are impacted by them, are necessary to define which of the abovementioned strategies will be more effective for soybean production in Brazil.
Caveats for the use of ecological niche models {#sec011}
----------------------------------------------
It is important to note that ENMs---as many other predictive models---are known to have uncertainties inserted in all stages of the modelling processes, that can, in turn, affect their results \[[@pone.0191273.ref068], [@pone.0191273.ref092]\]. For example, the input data can be climatic biased or inaccurate \[[@pone.0191273.ref093], [@pone.0191273.ref094]\]. For crop species, absence of cultivation or productivity in some regions can be due to socioeconomic aspects \[[@pone.0191273.ref095], [@pone.0191273.ref096]\], rather than the lack of climatic favorable conditions. Thus, the occurrences/productivity could be biased for some regions and consequently, for some climatic conditions, which could generate erroneous response curve of the relationship of species occurrence and the climatic variables. Here we consider that due to the economic importance of soybean its currently distribution represents the overall climatic conditions that permits its cultivation, thus that our samples are not biased related to soybean requirements. However, other factors unrelated with climatic variables not included in our models as interaction with other species, soil types, irrigation or other historical processes affects the distribution and productivity of cultivars and can also be a source of variation \[[@pone.0191273.ref010], [@pone.0191273.ref097]\]. Furthermore, ENMs predictions for future scenarios should be interpreted with caution, as climate change can lead to novel combinations of climatic variables, that are different from those used for model calibration, which associated with uncertainty about the future atmospheric CO~2~ and its fertilization effects on plants productivity, increase uncertainty of ENMs results \[[@pone.0191273.ref098], [@pone.0191273.ref099], [@pone.0191273.ref039]\]. Although it is important to be aware about these limitations when interpreting ENMs results, they should not prevent the use of ENMs, as even with such drawbacks when compared with more complex and detailed models ENMs had led to similar conclusions, using simpler and easy to get data \[[@pone.0191273.ref010]\].
Conclusions {#sec012}
===========
Agricultural technology and climate are important factors explaining soybean productivity in Brazilian municipalities. Moreover, some regions can be more affected by climate and, consequently, climate change. Although the environmental suitability of some areas would increase, there was an overall decrease in environmental suitability, indicating that soybean cultivation in Brazil could be highly threatened in the future. Considering the importance of Brazil soybean cultivation for the global supply, our results highlight the imminent need to develop strategies to mitigate the impacts of climate change and maintain high productivity in the actual cultivated areas. Thus, for areas that may loose productivity, we recommend programmes such as Ecosystem-based Adaptation (EbA; see \[[@pone.0191273.ref088]\]). For areas where suitability will increase, since the environmental conditions are more favourable, we suggest studies that focus on the development of strategies to increase productivity and reduce costs (social and environmental). We do not suggest the exploration of new areas in any of these cases, as this can increase environmental and food degradation and consequently affect overall human wellbeing.
Supporting information {#sec013}
======================
###### Data about the municipalities used in present papers.
(XLSX)
######
Click here for additional data file.
###### Information on methods and results of ecological niche model of soybean.
(DOCX)
######
Click here for additional data file.
We thank the anonymous reviewers for criticisms that improved the manuscript. We thank Thiago F. Rangel for providing access to the BIOENSEMBLES computational platform.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#Sec1}
============
Bretschneider (histidine-tryptophan-ketoglutarate, HTK) solution is routinely administered for cardioplegic arrest in many countries (Careaga et al. [@CR4]). In clinical studies as well as in experimental models the Bretschneider solution has been shown to achieve a better myocardial protection during ischemia as compared to pure electrolyte-containing crystalloid cardioplegic solutions without histidine or to blood cardioplegia (Beyersdorf et al. [@CR3]; Careaga et al. [@CR4]; Kober et al. [@CR12]; Korun et al. [@CR13]; Sakata et al. [@CR22]; Scrascia et al. [@CR24]). The effectiveness of the solution becomes apparent in an increased cardiac output, fewer arrhythmias, more frequent spontaneous defibrillation as well as decreasing length of stay in the intensive care unit (Careaga et al. [@CR4]; Sakata et al. [@CR22]). The included histidine (198 mM) is associated with enhanced efficiency of anaerobic glycolysis due to a high buffer capacity, whereas α-ketoglutarate (1 mM), an intermediary of the Krebs cycle, is assumed to serve energy production. Tryptophan (2 mM) and mannitol (30 mM) are proposed to function as a stabilizer of cell membranes and to reduce cellular edema formation, respectively (Careaga et al. [@CR4]). Usually, if Bretschneider solution is employed, almost all of the administered volume enters the systemic circulation. Thus, due to the large volume of the Bretschneider solution applied for induction of cardioplegic arrest, there is a massive incorporation of histidine.
To date, only very few data exist on plasma amino acid concentrations originating from histidine metabolism related to Bretschneider cardioplegia (Doetsch et al. [@CR5]; Schayani-Mühlschlegel [@CR23]). Therefore, we will here analyze concentrations of especially histidine but also further amino acids as well as urea and ammonia in plasma derived from intraoperatively taken blood gas samples as well as urine samples obtained at the beginning and the end of the operation, thus trying to establish an overall balance of amino acid and nitrogen metabolism under these conditions.
Materials and methods {#Sec2}
=====================
Study design and patient population {#Sec3}
-----------------------------------
Between 07/2014 and 10/2014, a total of 29 consecutive patients scheduled for elective isolated coronary artery bypass grafting (CABG) with cardiopulmonary bypass (CPB) were enrolled in the prospective observational designed study at the Department of Thoracic and Cardiovascular Surgery, University Hospital Essen. The study was approved by the Medical Ethics Committee of the University Hospital Essen and confirms to the principles of the Declaration of Helsinki. All individuals gave written informed consent. In short, myocardial protection was achieved using antegrade cold crystalloid Bretschneider cardioplegia (Custodiol, Dr. Franz Koehler Chemie, Bensheim, Germany), employing 1.6 ± 0.2 L on average supplemented by topical cooling, and single aortic cross-clamping for all distal anastomoses. After weaning from the heart--lung machine, patients received 56 mL Inzolen (Dr. Franz Koehler Chemie, Bensheim, Germany) on average. For further details see (Teloh et al. [@CR900]).
Patient Characteristics {#Sec4}
-----------------------
Of all patients, 76 % were male gender. The median values for age, height, weight, cardiopulmonary bypass time, and cross-clamp-time were 71 years, 173 cm, 84 kg, 86 min and 53 min, respectively. On average, patients received three grafts each.
Data collection {#Sec5}
---------------
Blood gas samples were routinely taken during operative procedures (initially, after beginning of CPB, before cessation of CPB, after cessation of CPB, before the end of operative procedures), and centrifuged at 3000*g* for 10 min at room temperature. Subsequently, the plasma was taken off and stored at −80 °C until analysis.
Immediately after catheterization of the patient's urinary bladder, a urine sample was obtained in order to represent baseline conditions. At the end of the operative procedures, a second sample was gathered from the volume that had been collected during the operation.
Measurements {#Sec6}
------------
For ammonium quantification in urine, capillary electrophoresis (P/ACE MDQ, Beckmann Coulter, Krefeld, Germany) was used. For this purpose, a fused silica capillary was employed with an effective length of 50 cm, an I.D. of 75 µm and an O.D. of 375 µm. Samples of initial urine were diluted with ultrapure water 1:50. Analysis was performed using a cation analysis kit (ABSciex, Fullerton, USA) and pressure injection. The subsequent separation proceeded using a voltage of 30 kV and normal polarity of the capillary. Indirect detection was performed employing a photo diode array at a wavelength of 200 nm. Due to the low sensitivity of the employed capillary electrophoresis, ammonium quantification in plasma was performed with an enzymatic method in the central laboratory of the University Hospital Essen.
Urine and plasma were analyzed for urea. Plasma samples were diluted 1:4, whereas urine was diluted 1:10 with 0.9 % NaCl. Urea was determined with the help of a fully automated clinical chemistry analyzer (Respons 920, DiaSys Diagnostics, Holzheim, Germany) using a commercially available reagent (DiaSys Diagnostics, Holzheim, Germany) for the enzymatic reactions of urease and glutamate dehydrogenase and subsequent detection of NADH decrease at 340 nm.
Urine and plasma samples were also analyzed for amino acids. For deproteinization, 300 µL sample were added to 75 µL sulfosalicylic acid (10 % in water) and thoroughly mixed. This mixture was centrifuged at 12,000 U/min for 5 min. Subsequently, 200 µL of the supernatant were diluted with reagent buffer at the ratio of 1:1. Of this formulation, 50 µL were injected into the liquid chromatograph (biochrom 30+, biochrom, Cambridge, UK). With the help of the employed cation exchanger, separation of amino acids took place at a rate of 0.25 mL/min due to varying pH and molarity of used running buffers containing citrate. Following this, the amino group reacted with ninhydrin, forming a colored complex which was detected at 570 nm (primary amino acids) and 440 nm (secondary amino acids), respectively. Urine as well as plasma samples were treated in the same way. Due to technical limitations in association with the high histidine concentration in plasma subsequent to Bretschneider application, tryptophan could not be exactly quantified.
Statistical analysis {#Sec7}
--------------------
All data are expressed as mean values ± standard deviation (SD) unless otherwise stated. Comparisons among different time points were performed using one-way independent analysis of variance (ANOVA) followed by the Fisher (LSD) post hoc analysis. A *P* value \<0.05 was considered significant.
Results {#Sec8}
=======
After start of CPB with concomitant induction of cardioplegic arrest, plasma histidine concentration sharply increased from an initial value of 71 to 21000 µM (Fig. [1](#Fig1){ref-type="fig"}a). Subsequently, during the course of the operation, it steadily decreased, reaching a concentration of 8000 µM at the end. The aspartate's concentration in plasma rose from 5 µM before to 1600 µM at the end of operation (Fig. [1](#Fig1){ref-type="fig"}b). Within the same interval, plasma glutamate concentration increased from 23 to 360 µM (Fig. [1](#Fig1){ref-type="fig"}c). Plasma concentrations of glutamine, asparagine, glycine, alanine and serine rose modestly from initial values of 596, 38, 214, 318, and 108 µM, respectively, to 862, 65, 325, 807, and 174 µM, respectively, at the end of the operation (Figs. [1](#Fig1){ref-type="fig"}d, [2](#Fig2){ref-type="fig"}a--d). Plasma concentrations of arginine, leucine, lysine, methionine, ornithine, phenylalanine, proline, tyrosine, threonine as well as valine varied little, and stayed within the respective reference ranges (Table [1](#Tab1){ref-type="table"}).Fig. 1Intraoperative plasma concentrations of **a** histidine, **b** aspartate, **c** glutamate, **d** glutamine. Patients received 1.6 L cardioplegic solution at the onset of cardiopulmonary bypass, which contained 198 mM histidine for induction of cardioplegia. Values are shown as mean ± SD. *Asterisk* \<0.05 compared with the initial value. *Double asterisk* \<0.01 compared with the initial value. *Triple asterisk* \<0.001 compared with the initial valueFig. 2Intraoperative plasma concentrations of **a** asparagine, **b** glycine, **c** alanine, **d** serine. Patients received 1.6 L cardioplegic solution at the onset of cardiopulmonary bypass, which contained 198 mM histidine for induction of cardioplegia. Values are shown as mean ± SD. *Triple asterisk* \<0.001 compared with the initial valueTable 1Intraoperative plasma concentrations of different amino acids. From (Duran [@CR200])Amino acidInitial1. BGA during CPBLast BGA during CPB1. BGA past CPBEnd of operationReference rangeArginine (µM)71 ± 1870 ± 890 ± 2081 ± 2171 ± 1915--190Leucine (µM)135 ± 27174 ± 30196 ± 36173 ± 34159 ± 3570--200Lysine (µM)179 ± 21192 ± 28222 ± 39207 ± 37194 ± 31115--300Methionine (µM)22 ± 424 ± 527 ± 624 ± 525 ± 610--40Ornithin (µM)54 ± 1549 ± 1657 ± 1555 ± 1752 ± 2250--200Phenylalanine (µM)55 ± 753 ± 944 ± 1342 ± 1237 ± 1035--85Proline (µM)149 ± 28151 ± 32185 ± 22182 ± 30201 ± 3997--330Tyrosine (µM)56 ± 1158 ± 1154 ± 1251 ± 1147 ± 935--115Threonine (µM)117 ± 31131 ± 31176 ± 41173 ± 36170 ± 3660--225Valine (µM)237 ± 28254 ± 37271 ± 39263 ± 43258 ± 38120--340Values are given as mean value ± SD
Histidine excretion largely increased from 47 µmol/mmol creatinine at basal conditions to 6760 µmol/mmol creatinine at the end of operation (Fig. [3](#Fig3){ref-type="fig"}a). Taking the intraoperatively excreted urine volume into account, it amounted to 19.5 mmol, i.e. just under 7 % of the incorporated amount of histidine (300 mmol). In the same interval, glutamate excretion rose from 1 µmol/mmol creatinine to 126 µmol/mmol creatinine and glutamine excretion from 33 µmol/mmol creatinine to 150 µmol/mmol creatinine (Fig. [3](#Fig3){ref-type="fig"}c, d). Excretion of aspartate, asparagine, glycine, alanine and serine slightly increased as well but stayed within the reference range (Figs. [3](#Fig3){ref-type="fig"}b, [4](#Fig4){ref-type="fig"}a--d). Amounts of carnosine, 1-methylhistidine and 3-methylhistidine in urine were subjected to only minimal changes (data not shown).Fig. 3Excreted amounts of **a** histidine, **b** aspartate, **c** glutamate, **d** glutamine during the intraoperative phase. Values are shown as mean ± SD. *Double asterisk* \<0.01 compared with the initial value. *Triple asterisk* \<0.001 compared with the initial valueFig. 4Excreted amounts of **a** asparagine, **b** glycine, **c** alanine, **d** serine during the intraoperative phase. Values are shown as mean ± SD
Urea and ammonium excretion per hour increased during the time of operative procedures from 9.5 and 1.1 mmol/h, respectively, at basal conditions to 10.1 and 2.6 mmol/h, respectively (Fig. [5](#Fig5){ref-type="fig"}). This increase in excretion amounts to 3 mmol urea and 8 mmol ammonium (both median) in the intraoperative interval compared to basal excretion. The magnitude of increase in urea excretion differed among patients, thus leading to fluctuating values in the range of 2--30 mmol/h. In plasma, over the course of the operation, the median of urea concentration was about 30 mg/dL (5 mM). Plasma ammonium concentration represented about 90 µg/dL (50 µM).Fig. 5Urea as well as ammonium excretion per hour from every patient at basal conditions and in the intraoperative interval, respectively. *Lines* represent median
Discussion {#Sec9}
==========
In the human organism, histidine is degraded by two major pathways (Bender [@CR1]; Doetsch et al. [@CR5]). First, it can be deaminated to glutamate via urocanic acid, which is supposed to be the dominant one. Second, it can be decarboxylated to finally yield aspartate. As soon as these two pathways that ensure a specific degradation are exhausted due to a high histidine concentration in plasma, other pathways are activated in addition, yielding those amino acids with short unbranched side chains while maintaining the histidine's α-amino-carboxylic acid group (Doetsch et al. [@CR5]). That way, glycine, alanine or serine may be obtained, depending on the length of the hydrocarbon chain, and a possible hydroxylation.
Patients received 1.6 L Bretschneider solution on average for induction of cardioplegic arrest. Based on the applied volume and its histidine concentration of 198 mM, a total of about 300 mmol histidine had been incorporated (Fig. [6](#Fig6){ref-type="fig"}). Assuming an estimated blood volume of approximately 5.2 L \[calculation on the basis of the formula of Nadler (Nadler et al. [@CR17])\], a plasma concentration of about 60 mM would have to be expected. However, only about a third of this calculated concentration, i.e. 20 mM, was detected in plasma in accordance with two former studies (Doetsch et al. [@CR5]; Schayani-Mühlschlegel [@CR23]). This strongly suggests the participation of the interstitial space for distribution. Actually, inclusion of the entire extracellular volume for calculation \[intravascular plus interstitial, approximately 2.9-fold the blood volume (Grocott et al. [@CR8])\] would result in a histidine concentration of approximately 20 mM. Thus, obviously, soon after incorporation (first sample during CPB), an equilibrium between blood and the interstitial space had already been achieved.Fig. 6Survey of application, metabolization and renal excretion of several amino acids during cardiopulmoanry bypass (CPB) in coronary artery bypass grafting patients. Patients received Bretschneider solution (main component histidine) for induction of cardioplegic arrest and Inzolen solution (main component potassium aspartate) after weaning from CPB to correct a shortage of potassium. In the course of intracellular metabolization, several amino acids emerge that show up in plasma, some of them are renally excreted. ↑↑↑, strong concentration increase; ↑↑, indicates moderate concentration increase; ↑, indicates slight concentration increase; ↔, indicates no change or changes staying within the reference range
The physiologic histidine concentration in plasma accounts for approximately 100 µM (Table [2](#Tab2){ref-type="table"}). Under these conditions, 5 % of the filtered histidine are renally excreted due to major reabsorption in the proximal tubule (Lingard et al. [@CR14]; Silbernagl and Volkl [@CR27]), i.e. about 100 µmol/mmol creatinine (Table [2](#Tab2){ref-type="table"}). In the present study, attributable to the highly elevated plasma histidine level (20 mM), the excreted amount accounted for 6800 µmol/mmol creatinine during the intraoperative interval. This value (equivalent to an absolute amount of 19.5 mmol histidine) corresponds to 7 % of the incorporated histidine (300 mmol). For comparison, Doetsch et al. reported an amount of histidine excreted within the first 72 postoperative hours representing 20 % of the administered dose (Doetsch et al. [@CR5]). During the intraoperative phase, plasma histidine concentration decreased from 20 mM to 8 mM. Renal excretion contributes only to a minor extent to this decrease (19.5 mmol/extracellular space of 15.1 L = 1.3 mM). Thus, uptake into cells and metabolic degradation (see below) should mainly be responsible for the rapid decline.Table 2Reference ranges of different amino acids, urea and ammonia in plasma as well as amino acids in urineReference rangePlasma Histidine70--125 µM Aspartate0--25 µM Glutamate10--130 µM Glutamine200--760 µM Asparagine35--75 µM Glycine150--490 µM Alanine175--580 µM Serine60--180 µM Urea17--43 mg/dL Ammonia27--90 µg/dLUrine (µmol/mmol creatinine) Histidine52--162 Aspartate5--27 Glutamate5--37 Glutamine22--58 Asparagine11--53 Glycine83--475 Alanine27--76 Serine27--76From Duran ([@CR200]), Thomas ([@CR202]) and Waters et al. ([@CR203])
In accordance with the histidine's main degradation pathways (see above), plasma concentrations of both aspartate and glutamate increased during the operation (Fig. [1](#Fig1){ref-type="fig"}b, c). Interestingly, however, the increase in plasma aspartate concentration was clearly more pronounced than the increase in plasma glutamate concentration (about 1600 µM at the end of the operation, 63-times above the upper limit of the reference range vs. 360 µM which is 2.8-times above the upper limit of the reference range, Table [2](#Tab2){ref-type="table"}), although the degradation pathway yielding glutamate is supposed to constitute the major route (Ghadimi [@CR6]; Mehler and Tabor [@CR15]). A plausible explanation for this unexpected behavior in plasma aspartate concentration is the application of Inzolen (Fig. [6](#Fig6){ref-type="fig"}). In conjunction with cardiac surgery, Inzolen solution, consisting mainly of racemic potassium aspartate and further trace elements, is usually applied to correct a shortage of potassium. In the present study, due to the solution's composition, patients received approximately 24 mmol aspartate after weaning from CPB this way. Despite the aspartate's highly elevated plasma level, urine analysis revealed only minor excretion (Fig. [3](#Fig3){ref-type="fig"}b), but enhanced excretion of glutamate (126 µmol/mmol creatinine at the end of operative procedures, 3-times the upper limit of the reference range, Fig. [3](#Fig3){ref-type="fig"}c; Table [2](#Tab2){ref-type="table"}). Because aspartate and glutamate possess the same net charge at physiological pH, they are reabsorbed by the same carrier in the proximal tubule \[in the case of aspartate independent of the isomer (Silbernagl [@CR26]; Silbernagl and Volkl [@CR28])\]. Obviously, reabsorption of aspartate is preferred which is in line with the higher affinity of the carrier for aspartate (K~M~ 0.10 mM for aspartate vs. 0.17 mM to 0.50 mM for glutamate, both determined in the rat) (Silbernagl [@CR25], [@CR26]).
Those amino acids having been formed by side chain conversion while maintaining the histidine's original α-amino-carboxylic acid group, i.e. glycine, alanine and serine, also increased in plasma in the course of the operation, although variable in magnitude, with glycine and serine staying within the reference interval (Fig. [2](#Fig2){ref-type="fig"}b--d; Table [2](#Tab2){ref-type="table"}). The increase of alanine (300 µM to 800 µM at the end of the operation, 1.4-times above the upper limit of the reference range, Table [2](#Tab2){ref-type="table"}) was plainest among those three amino acids, perhaps either due to the transamination reaction with glutamate or with histidine itself. The latter reaction occurs rarely under physiologic conditions but becomes more important in diseases associated with histidinemia (Bender [@CR1]), characterized by histidine plasma levels up to 1.8 mM (Ghadimi [@CR6]; Virmani and Widhalm [@CR31]). In accordance with the elevated concentrations in plasma, urinary excretion of glycine, alanine and serine increased slightly during the intraoperative interval, but stayed within the reference interval (Fig. [4](#Fig4){ref-type="fig"}b--d; Table [2](#Tab2){ref-type="table"}). Glutamine as well as asparagine possess the capability to accommodate an additional amino group, therefore representing the possibility to store further nitrogen. However, the plasma concentrations of glutamine as well as asparagine increased only moderately during the operation (Figs. [1](#Fig1){ref-type="fig"}d, [2](#Fig2){ref-type="fig"}a). For glutamine, this probably results from continuous catabolism for the purpose of ammonium synthesis and an increased excretion (150 µmol/mmol creatinine, 2.6-times the upper limit of the reference range; Fig. [3](#Fig3){ref-type="fig"}d; Table [2](#Tab2){ref-type="table"}). Increased glutamine excretion might arise from end product inhibition of glutamine and glutamate catabolism by alpha-ketoglutarate (Yao et al. [@CR34]). Carnosine (β-alanyl-histidine, \<5 µM), 1-methylhistidine as well as 3-methylhistidine (both \<3 µM) as further degradation products of histidine (Bender [@CR1]) stayed below the limit of quantification in plasma and were comparable to baseline conditions at the end of operative procedures in urine (data not shown). Thus, these pathways of histidine metabolism remained unused.
Alterations in glutamate, glutamine, alanine, asparagine and aspartate may arise from histidine metabolism. In addition, aspartate is applicated with the Inzolen solution. All are, directly or indirectly, glucoplastic amino acids (Bender [@CR1], [@CR2]). Therefore, they can be used for gluconeogenesis that is *per se* energy consuming. Energy consumption of asparagine, aspartate, glutamine and glutamate is smaller during this process compared to that of alanine, since pyruvate deriving from alanine degradation must first be carboxylated to oxaloacetate, which costs additional two molecules ATP per mol glucose. In contrast, metabolism of the aforementioned amino acids directly yields oxaloacetate or alpha-ketoglutarate that is converted to oxaloacetate via the citric acid cycle. Alpha-ketoglutarate deficiency in cardiac tissue occurs rapidly during ischemia (Peuhkurinen et al. [@CR18]). Provision of alpha-ketoglutarate (as an additive in blood cardioplegia) has been shown to attenuate myocardial ischemic injury in patients undergoing coronary revascularization (Kjellman et al. [@CR11]). Hence, exogenous supply might preserve myocardial oxidative capacity. In addition, it may help to minimize postoperative muscle catabolism (see below) (Wernerman et al. [@CR33]). Thus, the addition of 1 mM alpha-ketoglutarate to the cardioplegic solution seems beneficial, despite the possible conversion of amino acids arising from histidine metabolism into alpha-ketoglutarate in the later course. The cardioplegic solution also contains 2 mM tryptophan. In the course of tryptophan metabolism, the antioxidant melatonin might be formed but also nicotinamide adenine dinucleotide might even increase which has also been reported to act as an antioxidant (Kirsch and de Groot [@CR10]). Due to the melatonin's inherent antioxidant function (Tan et al. [@CR29]) but also the induction of antioxidant enzymes, it can exert cardio-protective effects against amongst others ischemia/reperfusion injury (Giacomo and Antonio [@CR7]; Reiter and Tan [@CR20]). Since during temporary cardioplegia the myocardium becomes ischemic very easily, this might have an appreciable contribution to myocardial protection during this phase as well.
The entire amount of histidine having been metabolized during the intraoperative period should be reflected by the differences in amino acid as well as urea and ammonium concentrations/amounts before (baseline values) and at the end of the operation in plasma as well as in urine, averaging 30 mmol in total. In this regard, the amount of further amino acids in urine except for histidine is negligible in terms of quantity. Of these 30 mmol, two-thirds account for the rise in several amino acids in plasma (19 mmol altogether) and one-third for elevation of ammonium (8 mmol) plus urea (3 mmol) excretion. Plasma urea and ammonia concentrations were in accordance with physiological values in the literature (Table [2](#Tab2){ref-type="table"}), as expected for substances that are obligatory for excretion by urine. Hence, metabolism in this intraoperative phase is small, but principally, it is supposed to continue in the postoperative phase. In the special case of aspartate, this amino acid was left out of consideration. Due to Inzolen application beginning after weaning from CPB, the amount of aspartate rose continuously towards the end of the intraoperative period. For this reason, a considerable share originating from histidine metabolization is unlikely.
Apart from the α-amino group, every histidine possesses two additional nitrogen atoms located in the imidazole ring. In the course of metabolism, this nitrogen should be excreted either as ammonium ions or as urea in the long term. Nitrogen excretion *per se* is intimately linked to systemic acid--base status, since generation of urea is bicarbonate consuming (Han [@CR9]; Meijer [@CR16]; Pitts [@CR19]), whereas ammonium is mainly obtained by deamination reactions from glutamine, glutamate or histidine (Han [@CR9]; Pitts [@CR19]; Weiner et al. [@CR32]). As it is common for amino acid metabolism, the histidine's α-amino group is converted to urea together with bicarbonate derived from the degradation of the remaining α-keto acid, thus being neutral as regards systemic acid--base homeostasis. The excretion of a surplus of 3 mmol urea (see above) should result in an additional base deficit of only −1 mEq/L. Thus, in relation to acute metabolic acidosis originating from massive dilution of endogenous bicarbonate based on administration of both the priming and the cardioplegic solution (Teloh et al. [@CR900]), the impact of nitrogen metabolism on acid base status in the intraoperative phase is minor. Due to persistent metabolism in the postoperative phase, however, the acidifying effect caused by enhanced urea excretion might get more pronounced over the course of time. In contrast, metabolization of infused aspartate (24 mmol) as part of the Inzolen solution should have an alkalizing effect in general based on its additional carboxyl group.
Subsequent to severe trauma or operations, amongst others, a stress-induced increase in sympathetic nervous activity is observed, resulting in a hypermetabolic state, the so-called postaggression syndrome (Sachs et al. [@CR21]). One of its characteristics is an increased gluconeogenesis from glucoplastic amino acids originating from enhanced protein degradation in skeletal muscle. Therefore, in such a condition, nitrogen balance is principally negative. Some evidence exist, that postaggression syndrome might be mitigated by amino acid administration. A benefit resulting from thereof is supported by a study of Umenai et al. that was able to show positive effects of perioperative amino acid infusion in patients undergoing off-pump CABG, resulting in a significantly shorter duration of postoperative mechanical ventilation as well as intensive care unit stay (Umenai et al. [@CR30]). Therefore, intraoperative provision of several amino acids arising from histidine metabolism (19 mmol in the present study) might achieve a similar result, especially since many of them represent glucoplastic amino acids (see above). Endogenous protein sources like skeletal muscle might be spared from degradation this way. Because the provision of several amino acids takes place before the onset of postaggression syndrome, this application's effect might be particularly beneficial.
In conclusion, in patients undergoing CABG, receiving approximately 1.6 L Bretschneider solution, a very substantial elevation of plasma histidine concentration was observed a few minutes after onset of CPB. Of the incorporated amount of 300 mmol histidine approximately 7 % were excreted without prior metabolization. In addition, about 10 % were metabolized, mostly being converted into other amino acids. Therefore, the influence on acid--base homeostasis originating from nitrogen metabolism is minor. Moreover, altered amino acid levels in plasma may have beneficial effects on postaggression syndrome in the postoperative phase.
We are greatly indebted to Markus Mallek for the excellent conduction of amino acid measurements.
Funding {#FPar1}
=======
No external funding was received.
Conflict of interest {#FPar2}
====================
The authors declare that they have no conflict of interest.
Ethical approval {#FPar3}
================
All procedures performed in studies involving human participants were in accordance with the ethical standards of the national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent was obtained from all individual participants included in the study.
This article does not contain any studies with animals performed by any of the authors.
[^1]: Handling Editor: C.-A. A. Hu.
| {
"pile_set_name": "PubMed Central"
} |
Introduction {#s1}
============
Emerging evidence has revealed that human thought draws from one\'s embodiment, which refers both to actual bodily states and to simulations of embodied experiences in the brain\'s modality-specific systems for perception, action, and introspection [@pone.0089768-Niedenthal1]. For example, Borghi, Glenberg, and Kaschak showed that verifying whether a given object (e.g., a car) had a certain part (e.g., a roof) involved perceptual simulation processes such that when participants\' reaction movement (upward or downward) corresponded with the position of the part relative to the object (e.g., roof vs. upward), the reaction was quicker [@pone.0089768-Borghi1]. In another study, after watching an action cartoon, participants were asked to describe the cartoon to a listener when the cartoon was no longer present. The result showed that participants who were prevented from gesturing (keeping stationary) processed the cartoon\'s description significantly slower than control participants [@pone.0089768-Rauscher1]. Similarly, Tucker and Ellis found representations of grapes and hammers can be activated through simulations of motor processes involved in precision and power grips, respectively [@pone.0089768-Tucker1].
Such findings can be explained through a perceptual symbol systems account, which undergirds such "embodied cognition" effects [@pone.0089768-Niedenthal1]. According to perceptual symbol systems theory, conceptual representations are tied to their perceptual basis and conceptual processing involves the partial simulation of those perceptual experiences that initially accompanied category exemplars. Perceptual symbols are raw materials that make up the variable constructions (i.e., simulations) and they draw from all senses, including proprioception, introspection, and motor programs [@pone.0089768-Schubert1], and they are derived from multiple sources of direct experience [@pone.0089768-Barsalou1].
Many studies have shown that processing abstract and concrete concepts activates modality-specific simulation of physical space [@pone.0089768-Schubert1]. In one study, two words (e.g., root and branch) were presented above each other, and their order either followed the canonical arrangement (i.e., branch above root) or a contradictory arrangement (i.e., root above branch). Participants were asked to judge whether the two were related or not. The results showed that reactions were quicker when the arrangement of the words followed the canonical arrangement of the objects [@pone.0089768-Zwaan1]. Some studies demonstrated that judging valence involves perceptual simulation of the vertical spatial dimension, on which good is up and bad is down [@pone.0089768-Meier1], [@pone.0089768-Meier2], [@pone.0089768-Crawford1], [@pone.0089768-Wapner1]. Other studies showed that the representation *TIME* involves the perceptual simulation of both horizontal space dimension [@pone.0089768-Boroditsky1], [@pone.0089768-Boroditsky2], [@pone.0089768-Boroditsky3], [@pone.0089768-Casasanto1], [@pone.0089768-Gentner1] and spatial size dimensions [@pone.0089768-Bruner1], [@pone.0089768-Bruner2]. Along these lines, additional research has revealed that the representation and processing of *SOCIAL POWER* also recruits vertical spatial perceptual simulation such that power = up and powerless = down [@pone.0089768-Schubert1], [@pone.0089768-Giessner1], [@pone.0089768-MoellerS1], [@pone.0089768-Robinson1]. Additional findings indicate that power was also represented in terms of size cues in which power = big and powerless = small [@pone.0089768-Schubert2].
It has been argued that gender is the most frequently utilized domain in human categorization [@pone.0089768-Haslam1], yet there is a sparse amount of research investigating its embodiment. Some research has shown that gender category (male and female) also be grounded in sensorimotor metaphors [@pone.0089768-Slepian1]. For example, Slepian and colleagues found that *MALE* was associated with the proprioceptive experience of "tough" and *FEMALE* was associated with the proprioceptive experience of "tender" [@pone.0089768-Slepian1], and the authors explained the results by the perspective that the largest trait difference between the male and female was the difference in tenderness [@pone.0089768-Feingold1]. In addition to traits like "tender" and "tough", it can be argued that stature is another important and salient physical difference between males and females. On average, men are taller and larger than women. According to the perceptual symbols account, the representation of gender may involve the perceptual simulation of experiences rooted in physical dimensions of vertical height and spatial size. The purpose of the present research was to demonstrate that the processes involved with gender categorization activate perceptual simulations involved with spatial dimensions. Based on the above analysis, we hypothesized that the representation of gender (male and female) will also involve the processing of perceptional simulation such that *MALE* categories will be processed faster along lines of greater size and higher verticality and vice versa for *FEMALES*.
Study 1 {#s2}
=======
We conducted study 1 to determine if the representation and processing of gender involves the perceptual simulation of vertical dimension. We predicted that the male faces would be judged faster if they were presented at the top of the computer screen, but the female faces would be judged faster if they were presented at the bottom of the computer screen.
Methods {#s2a}
-------
### Ethics Statement {#s2a1}
This study was reviewed and approved by the committee for the protection of subjects at Central China Normal University, School of Psychology Ethics Committee. Written consent was also obtained from each participant before the experiment according to the established guidelines of the committee. This procedure was followed in studies 2 and 3 as well.
### Participants and design {#s2a2}
Participants were 43 Chinese undergraduates (23 females) each whom were offered a notebook as compensation. The study had a 2 (facial gender: male or female)×2 (position: top or bottom) repeated-measures design.
### Stimulus, materials, and procedure {#s2a3}
Eighty-eight white and black photographs of faces (44 males and 44 females) were selected on the basis of earlier prior testing. All picture files were standardized in size to 100×120 pixels, and all the faces displayed neutral expressions. Participants arrived to the laboratory individually and were greeted by a male experimenter. They were seated facing the computer screen and told that the study was investigating aspects of face perception. The procedure was the same as the one used by [@pone.0089768-Meier1]. In each trial, a fixation cross was firstly presented at the center of the screen for 300 ms. Following this central cue, a subsequent fixation cross was flashed for 300 ms either at the 40% position (from top to down, above the central cue) or at the 60% position (from top to down, below the central cue) of the screen. A third fixation cross was flashed for 300 ms either at the 30% position (from top to down, above the central cue) or at the 70% position (from top to down, below the central cue) of the screen (in the same vertical direction as the second cross). The face pictures then appeared either at the 25% position (from top to down, above the central cue) or at the 75% (from top to down, below the central cue) position of the screen for 2000 ms (in the same vertical direction as the third cross), or disappeared when the participants made a response. The spatial cues (cross) were intended to direct attention to the spot of the picture\'s appearance, and thereby reduce random spatial exploration and additional error variance [@pone.0089768-Meier1]. All of the fixations and pictures appeared centered horizontally on the screen. Participants were required to report, by means of a key press, whether each face depicted a male or female target as quickly and accurately as possible, and the response keys were counterbalanced across the sample. If the response was inaccurate, the word "incorrect" appeared in a red font for 1500 ms. Accurate trials were separated by a blank screen for 500 ms. On completion of the experiment, participants were debriefed and dismissed.
Results and Discussion {#s2b}
----------------------
Mean categorization latencies served as the dependent measure of interest. Given the presence of extreme responses in the data set, response times that were slower than 2.5 standard deviations were excluded from the analysis, as were trials on which errors were committed [@pone.0089768-Ratcliff1]. This resulted in 0.09% (350) of the data being excluded from the statistical analysis. Latencies were then log-transformed to normalize their distribution [@pone.0089768-Cloutier1]. For ease of interpretation, however, the untransformed means are reported in [Figure 1](#pone-0089768-g001){ref-type="fig"}. The transformed latencies were submitted to a 2 (facial gender: male vs. female) ×2 (position: top vs. bottom) ANOVA. The analysis revealed no main effects of position \[*F*(1, 42) = 0.005, *p* = 0.94,\<0.001\] and facial gender \[*F*(1,42) = 2.690, *p* = 0.108, = 0.060\]. As we expected, the facial gender × position interaction was significant \[*F*(1, 42) = 9.028, *p* = 0.004, = 0.177; see [fig. 1](#pone-0089768-g001){ref-type="fig"}\]. Simple effects analysis demonstrated that responses to male faces in the up position (*M* = 548.78, *SD* = 90.30) were significantly faster than in the down position (*M* = 566.85, *SD* = 108.48), \[*F*(1, 42) = 4.58, *p* = 0.038, = 0.098\]. Alternatively, responses to female faces in the up position (*M* = 552.58, *SD* = 99.71) were significantly slower than in the down position (*M* = 539.79, *SD* = 90.98), \[*F*(1, 42) = 5.30, *p* = 0.026, = 0.112\].
{#pone-0089768-g001}
The results of study 1 indicated that the representation of gender involved the perceptual simulation of vertical space. If the gender-linked spatial height (e.g., male face was presented on the top of screen) was consistent with the actual physical feature of gender, gender categorization was facilitated. If gender and spatial height were incongruent, gender categorization was degraded.
Study 2 {#s3}
=======
In addition to vertical height, size is also a salient physical difference between males and females. We conducted study 2 to determine if the representation and processing of gender involves the perceptual simulation of spatial size. We predicted that male names would be judged faster if they were presented in bigger font size, whereas female names would be judged faster if they were presented in smaller font size.
Methods {#s3a}
-------
### Participants and design {#s3a1}
Participants were 48 Chinese undergraduates (23 males) who were offered a notebook as compensation. The experiment had a 2 (name type: male names or female names) ×2 (name font size: big or small) repeated-measures design.
### Stimulus, materials, and procedure {#s3a2}
Eighty names (40 male names and 40 female names) were selected on the basis of earlier pilot testing. One hundred and twenty-two common names (66 male names, 66 female names, not including the names of famous people) were chosen from the internet, which 28 participants (14 males) rated on 7-point scales (extremely suited to female, "1" to extremely suited to male, "7"), "the extent the name can be used for male or female". We then chose the highest scored 40 male names (the lowest score is 6.07, *M* = 6.44, *SD* = 0.219), and we chose the lowest scored 40 female names (the highest score is 1.36, *M* = 1.36, *SD* = 0.73). The scores of the male name were significantly higher than the female name, *t*(39) = 219.44,*p*\<0.001. Finally, male names and female names were matched with the same 40 common family name.
Participants arrived at the laboratory individually and were greeted by a male experimenter. They were seated facing the computer screen and told that the study was investigated people\'s ability to classify names by gender. In each trial, a fixation cross was firstly presented at the center of the screen for 800 ms, at which point a name appeared at the center of the screen for 2000 ms or disappeared when the participants made a response. Twenty male (female) names were presented at the screen in large font size (70 point), and twenty different male (female) names were presented on the screen in small font size (25 point). The inter-trial interval was 250 ms. Participants were required to report, by means of a key press, whether each name was a male name or female name as quickly and accurately as possible, and the response keys were counterbalanced across the sample. If the response was inaccurate (slower than 2000 ms), the word "incorrect" ("please be quicker") appeared in a red font for 1000 ms. On completion of the experiment, participants were debriefed and dismissed.
Results and Discussion {#s3b}
----------------------
Mean categorization latencies served as the dependent measure of interest. Given the presence of extreme responses in the data set, response times that were slower than 2.5 standard deviations were excluded from the analysis, as were trials on which errors were committed [@pone.0089768-Ratcliff1]. This resulted in 0.05% (191) of the data being excluded from the statistical analysis. Latencies were then log-transformed to normalize their distribution [@pone.0089768-Cloutier1]. For ease of interpretation, however, the untransformed means are reported in [Figure 2](#pone-0089768-g002){ref-type="fig"}. The transformed latencies were submitted to a 2 (name type: male names or female names) ×2 (name font size: big or small) ANOVA. The analysis revealed no main effects of name gender \[*F*(1, 47) = 1.69, *p* = 0.20, = 0.035\]. The main effect of name size was significant \[*F*(1, 47) = 4.48, *p* = 0.04, = 0.087\], such that participants were faster to categorize big font size names (*M* = 589.20, *SD* = 78.90) than small font size names (*M* = 598.16, *SD* = 86.62). Most importantly, as predicted, the name type × name size interaction was significant \[*F*(1, 47) = 78.26, *p*\<0.001, = 0.625; see [fig. 2](#pone-0089768-g002){ref-type="fig"}\]. Simple effects analysis demonstrated that responses to big sized male names (*M* = 575.70, *SD* = 69.33) were significantly faster than small sized male names (*M* = 618.12, *SD* = 85.34), \[*F*(1, 47) = 57.30, *p*\<0.001, = 0.549\]. Contrarily, responses to the big sized female names (*M* = 602.70, *SD* = 88.47) were significantly slower than small sized female names (*M* = 578.20, *SD* = 87.91), \[*F*(1,47) = 22.44, *p*\<0.001, = 0.323\].
{#pone-0089768-g002}
The results of study 2 indicated that representation of gender also involved the perceptual simulation of spatial size such that male was associated with larger spatial size, and female was associated with smaller spatial size.
Study 3 {#s4}
=======
The results of studies1 and 2 suggest that representations of gender involved perceptual simulation processes of spatial height and size dimensions. In study 1, participants showed a congruency effect in categorizing male and female faces, high and low, respectively. In study 2, participants showed a congruency effect in categorizing male and female names, in large and small font, respectively. But these effects can also be driven by the metaphors of social power. According to the Chinese tradition culture, the power and social status of men were higher than women, and the men were thought to make greater contributions to society than women. Even now, men occupy more senior official careers than women in China. Until January 1, 2010, there were only three women in the 26 ministers of Chinese government departments, accounting for 11.5% (Institute of Women, the All China Women\'s Federation) [@pone.0089768-Institute1]. Previous research has demonstrated that vertical positions were important metaphors of power, and the powerful was associated with up and the powerless was associated with down [@pone.0089768-Schubert1], [@pone.0089768-Giessner1], [@pone.0089768-MoellerS1], [@pone.0089768-Robinson1], and mental representation of power was also associated with size cues [@pone.0089768-Schubert2]. Participants reacted faster when powerful groups appeared on top and powerless groups appeared at the bottom [@pone.0089768-Schubert1], and participants reacted faster when the names of powerful groups appeared in big size (compared to small size) and vice versa for powerless groups [@pone.0089768-Schubert2]. Based on the above analysis, the effect found in studies 1 and 2 may also result from the metaphors of power. To examine the possibility that these results are outcomes of metaphor endorsement, rather than perceptual symbols, in study 3 we recruited a group of participants who do not endorse the power-verticality metaphor for gender.
Methods {#s4a}
-------
### Participants and design {#s4a1}
Two hundred and two undergraduates (184 females, 18 males; age ranged from 18 to 24) were chosen randomly to complete Attitudes Towards Women Scale (AWS), a 25-item questionnaire [@pone.0089768-Spence1]. The AWS measures traditional and conservative attitudes of women\'s place, including separate factors of rights, position relative to men, freedom, family role, and legal rights for college-aged participants [@pone.0089768-Byrne1]. The 25 items are measured on a scale ranging from 1 (strongly disagree) to 7 (strongly agree), where lower scores indicate more traditional, antifeminist views and higher scores indicate more positive and pro-feminist attitudes [@pone.0089768-Spence2]. The AWS has been used in Korea, Taiwan, and China and was found to have good validity and reliability in these samples [@pone.0089768-Chia1], [@pone.0089768-Daugherty1], [@pone.0089768-Smith1]. We then chose the 38 highest scoring undergraduates (37 females, 1 male, age ranged from 18 to 22) to complete the experiment. Their scores ranged from 141 to 173 (*M* = 148.86, *SD* = 5.99). The experiment had a 2 (name type: male names or female names) ×2 (name font size: big or small) repeated-measures design.
### Stimulus, materials, and procedure {#s4a2}
The procedure was the same as study 2.
Results and Discussion {#s4b}
----------------------
Mean categorization latencies served as the dependent measure of interest. Given the presence of extreme responses in the data set, response times that were slower than 2.5 standard deviations were excluded from the analysis, as were trials on which errors were committed [@pone.0089768-Ratcliff1]. This resulted in 0.05% (139) of the data being excluded from the statistical analysis. Latencies were then log-transformed to normalize their distribution [@pone.0089768-Cloutier1]. For ease of interpretation, however, the untransformed means are reported in [Figure 3](#pone-0089768-g003){ref-type="fig"}. The transformed latencies were submitted to a 2 (name type: male names or female names) ×2 (name font size: big or small) ANOVA. The analysis revealed no main effect of name font size \[*F*(1, 37) = 2.88, *p* = 0.098, = 0.072\]. The main effect of name type was significant \[*F*(1, 37) = 4.48, *p* = 0.04, = 0.108\], such that participants were faster to categorize female names (*M* = 590.69, *SD* = 74.15) than male names (*M* = 603.16, *SD* = 74.23). As in study 2, the name type × name size interaction was significant \[*F*(1, 37) = 42.19, *p*\<0.001, = 0.53; see [fig. 3](#pone-0089768-g003){ref-type="fig"}\]. Simple effects analyses demonstrated that responses to big sized male names (*M* = 584.33, *SD* = 71.87) were significantly faster than smaller sized male names (*M* = 621.99, *SD* = 76.59), \[*F*(1, 37) = 28.73, *p*\<0.001, = 0.44\]. Conversely, responses to big sized female names (*M* = 601.70, *SD* = 79.42) were significantly slower than small sized female names (*M* = 579.67, *SD* = 68.88), \[*F*(1, 37) = 12.20, *p* = 0.001, = 0.25\]. Including AWS scores as a covariate, we repeated the above analysis as an ANCOVA. The analysis revealed no main effect of AWS scores \[*F*(1, 36) = 2.29, *p* = 0.67, = 0.005\]. The interactions between AWS scores and other factors were also not significant, *ps*\>0.28.
{#pone-0089768-g003}
The results of study 3 indicated that feminists, who do not endorse the power-verticality metaphor for gender, represent and process gender in similar spatial and size dimensions, thus confirmed that the effects reported in studies 1 and 2 are not a result of "social power" metaphors.
General Discussion {#s5}
==================
A wide array of research now supports the view that human cognition is grounded in and shaped by sensorimotor experiences and that our conceptual representations include sensory, motor, and introspective activations that are recruited into partial simulations, which reenact various embodied states [@pone.0089768-Barsalou2]. While previous research has revealed that gender-category representations include sensorimotor information related to handling hard and soft objects and proprioceptive experience (toughness) [@pone.0089768-Slepian1], no research to date has explored how salient gender differences in size affect the processing of gender representations. In the present research, we have provided evidence that the representation and processing of gender activates simulations involved with vertical height and spatial size dimensions, with *MALE* being associated with up and big and *FEMALE* being associated with down and small. Importantly, by recruiting feminist participants in a separate study and replicating the observed effects, we ruled out the alternative explanation that concepts involving "social power" might be driving these effects. These results are consistent with perceptual symbol systems theory by spotlighting how processing concepts can be affected by their perceptual basis [@pone.0089768-Barsalou1], which we have extended to include the social categorization of gender in terms of vertical height and spatial size.
While the present findings could potentially be explained by either conceptual metaphor or perceptual symbol accounts, we proposed the latter approach appears to have the most explanatory power. First, it has been argued elsewhere that some metaphors (such as vertical position) are more basic than others [@pone.0089768-Schnall1]. One implication would be that basic metaphors (vertical position) are so prevalent because they draw from similar embodied experiences and thus use perceptual simulation in order to ground their meaning. That is perceptual simulation may be the basic of some basic metaphors [@pone.0089768-Schnall1]. Second, the results from the feminist participants suggest that metaphors are not being used, as they would naturally recruit different metaphors regarding size and gender. Third, metaphor serves to only make abstract relationships more concrete [@pone.0089768-Landau1]. In actuality, males are on average taller than females. Additionally, on average, men are larger than women. Yet men being larger and taller than women are not abstract in this sense, and it can be seen directly and this correlation then, is based in concrete sensory experience. Finally, a perceptual account is more parsimonious and is better aligned with one\'s general empirical experiences in that men and women typically correspond to specific physical sizes.
Our data provide further evidence that early perceptual processes contribute to social-categorical thinking. Whereas prior research demonstrated that facial cues and visual acuity can affect the speed of social categorization [@pone.0089768-Cloutier1], the studies presented here demonstrated vertical height and spatial size can also contribute to social-categorical thinking. Just as visual cues are present on every human face and are therefore likely to have a ubiquitous influence on categorical thinking about other people, spatial position and size are normally present for males and females and are therefore likely to have a ubiquitous influence on gender categorization. In particular, a person\'s position (e.g., sitting on a high or low stool) or stature (e.g., taking up more or less physical space) might influence how they are judged and potentially how they judge others in the social world.
Based on present studies, further research should further explore whether and how the present findings influence formation of specific gender stereotypes. The previous studies showed that the high position was associated with good and the low position was associated with bad [@pone.0089768-Meier1]. We can infer that because processing the concept of male (female) involved perceptual stimulation associating with high (low) position which is relevant with positive (negative) valance respectively, the perceptual simulation involving gender categorization may play an important role in the formation of negative stereotypes towards women. For example, a lot of studies have shown that people had the stereotype that woman can not reach high achievements in mathematics [@pone.0089768-Eccles1], [@pone.0089768-Fennema1], [@pone.0089768-Swim1]. Future research should explore the formation of this negative female stereotype by the perspective of representation of concept of gender (male and female) based on the present study. Further, it would be interesting to test whether inconsistent perceptual-gender properties (e.g., very tall women and very small men) influence social judgments in stereotype inconsistent ways. Specifically, additional research could explore if the activation of female stereotypes to (such as females are timid and so on) who are in high positions may be inhibited compared to females in low positions, and whether the activation of male stereotypes (such as males are rude and so on) who are in low positions may be inhibited compared to males in high positions.
The authors thank the two anonymous referees for their helpful comments on an earlier version of this article.
[^1]: **Competing Interests:**The authors have declared that no competing interests exist.
[^2]: Conceived and designed the experiments: XBZ QL BZ. Performed the experiments: XBZ QL. Analyzed the data: XBZ QL. Contributed reagents/materials/analysis tools: XBZ QL. Wrote the paper: XBZ QL KJE BZ.
| {
"pile_set_name": "PubMed Central"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.