Datasets:

suolyer

/

pile_pubmed-central

like 1

Introduction {#sec1-1} ============ Type 2 diabetes mellitus is a major global health problem. The World Health Organization estimates that diabetes, primarily type 2, affects 5.2% of the world\'s adult population.\[[@ref1]\] India now has the largest population with type 2 diabetes mellitus internationally, with a 10.4% prevalence and totaling about 19 million. India is also expected to show the highest increase in the prevalence internationally by 2045 (84% increase).\[[@ref2]\] This increase is attributed to increase in life expectancy and changes in lifestyle such as fewer healthy diets and physical inactivity. Countries need to strengthen prevention to prevent increase in the prevalence of type 2 diabetes mellitus. Several factors such as low awareness, ignorance about complications, poor adherence to medications and lifestyle interventions, bad health behaviors such as smoking and alcohol, physical inactivity, and obesity contribute to the increasing prevalence of this problem.\[[@ref3]\] Results from the Diabetes Prevention Program showed that a 7% weight loss in the 1^st^ year through increased physical activity (150 min of brisk walking per week) reduced the 4-year incidence of type 2 diabetes by 58% in men and women with impaired glucose tolerance and was twice as effective as metformin therapy.\[[@ref4]\] Moreover, improvement in fasting blood glucose (FBG) is directly related to the relative amount of weight lost.\[[@ref5]\] Moderate weight loss may not improve glycemic control in all obese patients who have diabetes;\[[@ref6]\] however, it is possible that patients with long-standing disease or severe pancreatic β-cell dysfunction are not as responsive to weight loss as those with less extensive disease. Guidelines and recommendations suggest 150 min of aerobic moderate exercise (50%--70% of maximum heart rate) or 75 min of vigorous-intensity exercise or a combination of both over at least 3 days in a week for glycemic control.\[[@ref7][@ref8]\] Several components in naturopathic intervention such as raw diets, calorie restriction (fasting), massage, yoga, physical exercise, and hot baths are independently known to confer weight reduction and bring glycemic control.\[[@ref9][@ref10][@ref11][@ref12][@ref13]\] Several studies have shown acute effects of these interventions using raw vegetable juices with or without yoga on glycemic control.\[[@ref14][@ref15]\] Our earlier study using naturopathy interventions showed good glycemic control in patients with type 2 diabetes mellitus.\[[@ref16]\] However, there is a paucity of controlled studies with structured naturopathy intervention in patients with type 2 diabetes mellitus. In India, several people flock to residential naturopathy centers for weight loss, cardioprotective effects, and glycemic control. These interventions have been shown beneficial in the short term.\[[@ref9][@ref10][@ref11][@ref12][@ref13][@ref14][@ref15]\] However, the long-term effects of such acute calorie restriction and weight loss following a residential naturopathic intervention have not been studied. In this open-label parallel cohort study, we aim to study the effect of an intensive residential naturopathic intervention versus usual care in type 2 diabetes mellitus patients immediately following intervention and long-term effects at 6 and 12 months. Methods {#sec1-2} ======= Subjects {#sec2-1} -------- Patients with type 2 diabetes mellitus attending outpatient clinics in endocrinology department of a tertiary medical teaching hospital were invited to participate in a residential naturopathy and yoga-based lifestyle intervention over a 3-month period in a comprehensive Naturopathy Centre, Manthena Satyanarayana Raju Arogyalayam in Vijayawada, India. A convenience sampling approach was used, Patients were recruited into the study if they satisfied selection criteria and gave written inform consent to participate. Patients who agreed to undergo intensive residential naturopathy and yoga-based life style intervention were recruited as cases. Those who agreed to participate in the study but were not undergo residential naturopathy and yoga intervention were recruited as controls. The participants were required to give blood samples and fill in questionnaires and undergo periodic clinical assessments over 1 year. The intervention was conducted in a residential naturopathy and yoga center near to the hospital. The study was supervised by endocrinologists and physicians of the referral medical teaching hospital. The study was cleared by the institutional review board of both the institutions. The participants were recruited only if they satisfied the selection criteria and gave written consent. Selection criteria {#sec2-2} ------------------ ### Inclusion criteria {#sec3-1} Patients age range between 18 and 60 of both genders, with a history of type 2 diabetes mellitus for the past 1 year or more along with glycated hemoglobin (HbA1c) \>7% levels; those who are depending on oral or parenteral hypoglycemic agents; those with Zubrod\'s performance status 0--2; and willing to give consent to participate in the study were included in the study. ### Exclusion criteria {#sec3-2} Patients with secondary complications of type 2 diabetes mellitus such as Grade III nephropathy, end-stage renal disease, and diabetic ulcers. Patients with a history of recent myocardial infarction or transient ischemic attacks, hot water epilepsy, exercise-induced asthma, uncontrolled blood pressure (systolic blood pressure \>160 mmHg and diastolic blood pressure \>110 mmHg), anemia (Hb \<10 g%), hyponatremia (sodium \<136 mg/dl), neutropenia, pancytopenia or thrombocytopenia, major depressive disorders, or psychiatric or neurological illness. Those who participated in regular exercise, yoga, nutrition, or lifestyle modification program in the preceding six months were excluded. Patients with active infections or fever, those with New York Heart Association Class III cardiac failure, or chronic obstructive pulmonary disease resulting in dyspnea or orthopnea were excluded from the study. Sample size {#sec2-3} ----------- The sample size was calculated using Open source Epi Info Software (version 3.1, CDC, USA) by Fleiss method with continuity correction with 95% confidence interval (CI) and 80% power, for a 15-point difference in an outcome between the control (unexposed, *n* = 5) and intervention (exposed, *n* = 20) groups. The total sample size was 176 with allocation ratio of 1:1, i.e., 88 patients in each arm. Considering 20% attrition, we recruited 106 patients in each arm. Study procedure {#sec2-4} --------------- A total of 102 patients in naturopathy group and 109 participants in the control group provided data at the baseline. Those in naturopathy group underwent residential naturopathy and yoga treatment in a naturopathy center. Written and signed informed consent was obtained at the time of enrolment at baseline. Briefly, sociodemographic characteristics of the patients, their personal and medical history, and details of various medications taken by the patient were abstracted on the case report form. Blood draws were carried out under aseptic conditions by a trained phlebotomist and 10 ml of blood was collected in sterile vacationers. FBG and serum insulin 2-h postprandial blood glucose (PPBG), HbA1c and lipid profile, liver function tests, and renal function tests were carried out. Samples were collected between 7 am and 9 am on all days. Follow-up assessments were done at 3, 6, and 12 months from the start date. Medication score was calculated by assigning a score of 0 or 1 (No/Yes) for common diabetes oral medications and a score of 0 or 2 (No/Yes) for insulin, a more demanding regimen which has been used in the Diabetes Medication Satisfaction Tool earlier.\[[@ref17]\] At 3 months, 102 and 92 participants provided data in naturopathy and control group. Thereafter, dropouts in naturopathy group were due to failure to adhere to diet and physical activity regimen (*n* = 12), defaulted on tests or assessments (*n* = 5), not available for follow-up (*n* = 14). In control group, dropouts were mainly because of not available for follow-up (*n* = 20), increase in blood glucose levels, uncontrolled diabetes (*n* = 3), defaulted on tests or assessments (*n* = 4), and death (*n* = 1) due postoperative renal failure following lower limb amputation for gangrene with uncontrolled diabetes \[see [Figure 1](#F1){ref-type="fig"}: Trial profile\]. {#F1} Outcome measures {#sec2-5} ---------------- ### Primary outcomes {#sec3-3} HbA1c was measured using high-performance liquid chromatography (HPLC) with Biorad D10 Analyzer and Biorad reagents were used. The level of HbA1c is proportional to both the average glucose concentration and the life span of the red blood cell in the circulation. The measurement of HbA1c has therefore been accepted for the clinical management of type 2 diabetes mellitus through routine monitoring. Methods for the determination of HbA1c include electrophoresis, immunoassays, and chromatography. HbA1c determination with the D-10 Dual Program has been optimized to eliminate interferences from hemoglobin variants, labile A1c, and carbamylated hemoglobin. The D-10 Dual Program is based on chromatographic separation of the analytes by ion-exchange HPLC. Secondary outcomes {#sec2-6} ------------------ Blood glucose (fasting/postprandial) was measured by spectrophotometry (Hexokinase method) using an auto-analyzer (Beckman Unicel DxC 800 Chemistry Analyzer) with Beckman reagentsMedication score -- Both oral hypoglycemic agents and parenteral insulin were calculated by using the Diabetes Medication Satisfaction Tool. Intervention {#sec2-7} ------------ ### Naturopathy and yoga-based lifestyle intervention {#sec3-4} In this group, the participants were enrolled in a 3-month residential naturopathy intervention program comprising diet, yoga, hydriatic treatments, massage, and didactic and interactive lectures on lifestyle modification and type 2 diabetes mellitus self-management. Patients underwent a structured routine which involved physical activity, yoga program with asanas, pranayama, meditation and relaxation, calorie restriction, and salt-restricted diet. The diet prescribed was a low glycemic index, low-salt high-fiber plant-based diet containing whole grains, legumes, vegetables, and fruits with no added oil, sugar, or salt. They also underwent short intermittent juice fasting with calorie restriction over 3--4 days a month. They had to undergo a structured routine of hydriatic treatments such as hip bath, immersion bath, jets, sprays, douche, and mud and steam bath apart from partial and full-body Swedish massages. The goal of treatment was weight reduction if overweight, stress reduction, and dietary intervention to manage glycemic index. The patients were in this facility for a period of 3 months and were supervised by doctors and diabetologists over the course of their intervention. The oral antidiabetic medication was constantly monitored and tapered based on their blood glucose and HbA1c levels. ### Control intervention {#sec3-5} The control group received conventional antidiabetic treatment as per the standard guidelines. The pharmacological management of type 2 diabetes mellitus patients included both oral hypoglycemic agents, parenteral insulin, and other medicines to manage their comorbid conditions and supportive care. The control group also received a diabetes self-management program by a diabetes educator during their hospital visits. They were also counseled on the diet and physical activity by a clinical nutritionist. Both HbA1c levels, blood glucose, and other biochemical tests were routinely conducted and monitored. Dose escalation or tapering was done based on these levels by a diabetologist. Data analysis {#sec2-8} ------------- Data were analyzed using Statistical Package for the Social Sciences (spss) software (version 18, IBM, USA) for Windows. Data were found to be normally distributed. A repeated-measures ANOVA was done with assessments being time and group being independent variables. The group by time interaction effects was computed using *post hoc* Bonferroni correction for four time points and two group measures. Therefore, 95% CI of the adjusted *P* value was fixed at *P* ≤ 0.008 for *post hoc* correction. Both within- and between-group effects were analyzed using intention-to-treat analysis. The missing values were imputed using the mean of the respective group for that assessment interval. Changes from the baseline between groups were assessed using independent samples *t*-test. Results {#sec1-3} ======= The mean age of the study population was 51 years in naturopathy group and 48.8 years in control group. Participants in the naturopathy group had a long-standing type 2 diabetes mellitus, mean 10.27 years compared to control group mean 5.83 years. Gender distribution, comorbid illness, and other sociodemographic variables were similar across groups \[[Table 1](#T1){ref-type="table"}\]. There was a significant weight loss in the naturopathy group compared to the conventional group at 3 (*P* \< 0.01) and 6 (*P* \< 0.01) months, but not in 12^th^ month, whereas there was significant weight loss within the group at 3 (*P* \< 0.001) and 6 (*P* \< 0.001) months when compared with baseline. ###### Sociodemographic and clinical characteristics of the study population Particulars Mean±SD -------------------------------------- ----------- ----------- Age, mean±SD 51.0±8.1 48.8±8.1 Duration of type 2 diabetes mellitus 10.27±6.6 5.83±5.7 Gender, *n* (%)  Male 43 (42.2) 50 (45.9)  Female 59 (57.8) 59 (54.1) No comorbidity 35 (34.3) 31 (28.4) Comorbidity 1 48 (47.1) 65 (59.6) Comorbidity 2 18 (17.6) 12 (11.0) Comorbidity 3 1 (1.0) 1 (0.9) Comorbidity obesity 28 (27.5) 37 (33.9) Comorbidity hypertension 46 (45.1) 47 (43.1) Comorbidity hypothyroidism 3 (2.9) 5 (4.6) Comorbidity CAD 3 (2.9) 0 (0.0) Comorbidity nephropathy Grade 1 7 (6.9) 3 (2.8) Comorbidity others 36 (35.3) 16 (14.7) Occupation, *n* (%)  Government employee 1 (1.0) 2 (1.8)  Nongovernment employee/private 9 (8.8) 10 (9.2)  Homemaker 56 (54.9) 40 (36.7)  Retired employee 11 (10.8) 5 (4.6)  unemployed 5 (4.9) 2 (1.8)  Business 3 (2.9) 2 (1.8)  Others 17 (16.7) 48 (44.0) Education, *n* (%)  Illiterate 5 (4.9) 31 (28.4)  Less than primary 4 (3.9) 8 (7.3)  Primary 24 (23.5) 24 (22.0)  High school 31 (30.4) 30 (27.5)  Intermediate 12 (11.8) 8 (7.3)  Graduate 22 (21.6) 5 (4.6)  Postgraduate 4 (3.9) 3 (2.8) SD=Standard deviation, CAD=Coronary artery disease Fasting blood glucose and postprandial blood glucose levels {#sec2-9} ----------------------------------------------------------- Multivariate analysis showed that there is no significant group by time interaction effects for fasting blood sugar between naturopathy and conventional treatments. However, the between-subject\'s effects showed a significant difference between groups *F* (1, 208) = 4.62, *P* = 0.03. *Post hoc* Bonferroni correction showed significant decrease in fasting blood sugar within the naturopathy group between baseline and postintervention at 3 months (*P* \< 0.001), 6 months (*P* \< 0.001), and 12 months (*P* = 0.006). There was no significant decrease in fasting blood sugar in naturopathy group compared to the conventional group postintervention at 3, 6, and 12 months \[[Table 2](#T2){ref-type="table"}\]. ###### Comparison of changes in fasting and postprandial blood sugar over time following naturopathy and conventional treatment using repeated-measures ANOVA Group FBG (mg/dl), mean±SD FBG change score (mg/dl), mean±SD -------------- -------------------------------------- --------------------------------------------------- ------------------ ---------------- -------------- -------------- ---------------- Naturopathy 190.66±65.51 136.10±44.93\*\* 144.13±53.40\*\* 167.09±58.45\* 54.56±65.51 46.53±72.48 23.57±75.70 Conventional 148.54±46.62 136.82±38.71 150.00±57.48 157.33±62.60 10.35±49.64 −0.35±49.64 −0.35±49.640 **Group** **Postprandial blood sugar (mg/dl)** **Postprandial blood sugar change score (mg/dl)** Naturopathy 250.69±104.13 172.31±56.15\*\* 184.10±63.90\*\* 273.13±81.69 73.47±114.02 61.67±108.99 −1.67±108.99\* Conventional 212.40±77.04 183.73±55.14\* 195.64±63.49 202.19±87.50 26.72±81.04 14.81±87.08 8.96±96.88 *P*\<0.008, *P*\<0.01 for between groups; \**P*\<0.008, \*\**P*\<0.001 for within group effects using *post hoc* Bonferroni correction. BL=Baseline, 3 M=3 months, 6 M=6 months, 12 M=12 months, FBG=Fasting blood sugar, SD=Standard deviation Multivariate analysis showed that there is a significant group by time interaction effect for PPBG between naturopathy versus conventional treatment *F* (3, 204) = 21.78, *P* \< 0.001. The between-subject\'s effects showed a significant difference between groups *F* (1, 206) = 9.4, *P* = 0.003. *Post hoc* Bonferroni correction showed significant decrease in postprandial blood sugar within the naturopathy group between baseline and postintervention that is at 3 months (*P* \< 0.001), 6 months (*P* \< 0.001), but not in 12 months. There was no significant decrease in postprandial blood sugar in naturopathy group compared to the conventional group postintervention at 3 and 6 months. Whereas, there is a significant increase postprandial blood sugar in naturopathy group compared with conventional group at 12 months (*P* \< 0.001) \[[Table 2](#T2){ref-type="table"}\]. Glycemic control at 3, 6, and 12 months from baseline {#sec2-10} ----------------------------------------------------- There was a significant decrease in FBG at 3 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 5.0, *P* \< 0.001, 95% CI = 26.8--61.6), at 6 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 5.1, *P* \< 0.001, 95% CI = 30.4--68.3), and at 12 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 3.4, *P* = 0.001, 95% CI = 14.2--53.1). There was a significant decrease in PPBG at 3 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 3.4, *P* = 0.001, 95% CI = 20.0--73.4), at 6 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 3.4, *P* = 0.001, 95% CI = 20.1--73.5), and at 12 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = −2.3, *P* = 0.021, 95% CI = −65.7--−5.4) \[[Table 2](#T2){ref-type="table"}\]. Glycated hemoglobin levels {#sec2-11} -------------------------- Multivariate analysis showed that there is a significant group by time interaction effect for HbA1c between naturopathy versus conventional medicine *F* (3, 207) = 11.83, *P* \< 0.001. However, the between-subject\'s effects showed that there was no significant difference between groups. *Post hoc* Bonferroni correction showed significant decrease in HbA1c within the naturopathy group between baseline and postintervention, that is, at 3 (*P* \< 0.001), 6 (*P* \< 0.001), and 12 months (*P* \< 0.001). *Post hoc* Bonferroni correction showed significant decrease in HbA1c within the conventional group between baseline and postintervention, that is, at 3 (*P* \< 0.001), 6 (*P* = 0.006), and 12 months (*P* = 0.019). There was significant decrease in HbA1c in naturopathy group compared to the conventional group postintervention at 3 (*P* \< 0.001) and 6 months (*P* = 0.035) but not at 12 months \[[Table 3](#T3){ref-type="table"}\]. ###### Comparison of changes in glycated hemoglobin over time following naturopathy and conventional treatment using repeated-measures ANOVA with *post hoc* Bonferroni correction HbA1c (%), mean±SD HbA1c change score (%), mean±SD -------------------- -------------------- --------------------------------- ----------------------- ----------------------- ------------------- ------------------- ------------------- Naturopathy group 9.6 (81)±1.8 (19) 7.5 (58)±0.8 (8)\*\* 7.9 (63)±1.5 (16)\*\* 8.5 (69)±1.7 (18)\*\* 2.0 (21)±1.6 (17) 1.6 (17)±2.0 (21) 1.1 (12)±2.0 (21) Conventional group 9.0 (75)±1.7 (18) 8.2 (66)±1.2 (13)\*\* 8.4 (68)±1.6 (17)\* 8.4 (68)±1.7 (18) 0.7 (7)±1.7 (18) 0.6 (6)±1.8 (19) 0.5 (5)±1.8 (19) *P*\<0.008, *P*\<0.001, for between groups; \**P*\<0.008, \*\**P*\<0.01 for within group effects using *post hoc* Bonferroni correction, HbA1c values in mmol/mol is in parentheses. SD=Standard deviation, BL=Baseline, 3 M=3 months, 6 M=6 months, 12 M=12 months, HbA1c=Glycated hemoglobin There was a significant decrease in HbA1c at 3 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 5.7, *P* \< 0.001, 95% CI = 0.86--1.76), at 6 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 3.9, *P* \< 0.001, 95% CI = 0.5--1.6), and at 12 months from baseline in naturopathy group compared to conventional group on independent samples *t*-test (*t* = 2.2, *P* = 0.028, 95% CI = 0.06--1.11) \[[Table 3](#T3){ref-type="table"}\]. Medication score {#sec2-12} ---------------- There was a significant decrease in number medication score using Diabetes Medication Satisfaction Tool in naturopathy group compared to conventional group at 6 months (*P* \< 0.008) only but not at 3 and 12 months using Mann--Whitney U-test for between-group comparison \[[Table 4](#T4){ref-type="table"}\]. ###### Comparison of medication score of the study population at baseline, 3, 6, and 12 months using Diabetes Medication Satisfaction Tool Total score, mean±SD -------------------- ---------------------- ----------- ------------ ----------- Naturopathy group 1.62±0.90 1.16±0.95 0.79±0.90ł 0.75±0.93 Conventional group 1.55±0.83 1.24±1.02 1.18±1.08 0.82±1.06 *P*\<0.008 for between groups using Mann--Whitney U-test. BL=Baseline, 3 M=3 months, 6 M=6 months, 12 M=12 months, SD=Standard deviation ### Parenteral insulin {#sec3-6} There was a significant increase in insulin usage at 6 months in conventional group compared to naturopathy group on Pearson Chi-square test for intergroup comparison (Chi-square = 4.15, *P* = 0.04) but not at 3 and 12 months \[[Table 5](#T5){ref-type="table"}\]. ###### Parenteral medication (insulin) score at baseline, 3, 6, and 12 months Parenteral insulin BL, *n* (%) 3 M,*n* (%) 6 M, *n* (%) 12 M, *n* (%) -------------------- ------------- ------------- -------------- --------------- ----------- ----------- ----------- ----------- Naturopathy group 36 (35.5) 66 (64.7) 23 (22.5) 79 (77.5) 18 (17.6) 84 (82.4) 15 (14.7) 87 (85.3) Conventional group 36 (33.3) 72 (66.7) 34 (31.5) 74 (68.5) 32 (29.6) 76 (70.4) 21 (19.4) 87 (80.6) *P* 0.76 0.14 0.04 0.36 *P*\<0.05, *P*\<0.01, *P*\<0.001 for between groups using Chi-square test of proportions. BL=Baseline, 3 M=3 months, 6 M=6 months, 12 M=12 months Discussion {#sec1-4} ========== The results suggest that the 3-month residential naturopathy and yoga intervention significantly reduced blood glucose levels, glycated hemoglobin levels, and reduced need for antidiabetic medications in the intervention group compared to controls. The effects were more profound immediately following intervention and lasted up to 6 months from the start of the study. However, the effects were not sustained over a longer period of 1 year from the study start in the intervention. The results from our study are similar to earlier studies that have shown reductions in blood glucose levels and HbA1C levels with structured physical activity and dietary interventions.\[[@ref9][@ref10][@ref11][@ref12][@ref15][@ref16][@ref18][@ref19]\] While the earlier studies have shown benefit finding with physical activity, dietary intervention, and naturopathy intervention in the short term,\[[@ref11][@ref14][@ref15][@ref19][@ref20]\] our study showed that the effects can be sustained over a 6 months period. The effect size of the intervention was small for change in PPBG levels from baseline (Effective size (ES) = −0.32 \[95% CI − 0.12-- −0.51\]), moderate for FBG levels (ES = −0.52 \[95% CI − 0.33--−0.71\]), and large for HbA1C levels (ES = −1.0 \[95% CI − 0.12--−0.51\]) at 3 months following intervention. While the effects were maintained at 6 months, they tended to decrease at the end of 1 year. The patients in the study underwent intermittent fasting ranging from 3 to 5 days and were on a naturopathy based plant diet rich in fruits, vegetables, sprouts, nuts, salads, and juices. Basically, intervention group received a high-fiber, high-protein, low-carbohydrate, and nonketogenic diet. This is different from earlier studies, wherein ketogenic diets have been used to improve insulin sensitivity and reduce resistance in overweight and obese populations. Fasting, in addition to lowering insulin levels, also improves insulin resistance significantly.\[[@ref21]\] A low-salt diet devoid of added salt can further potentiate weight loss and bring better glycemic control. Moreover, hot baths, yoga, fruit and vegetable diets, and active lifestyle also exert anti-inflammatory effects that also lead to glycemic control. Patients with type 2 diabetes mellitus have two--fourfold increased risk for cardiovascular disease than nondiabetic patients.\[[@ref22][@ref23][@ref24]\] Therefore, using interventions that reduce weight and bring glycemic control is needed to mitigate this risk. Naturopathy and yoga-based intervention promises better glycemic control that could also reduce risk for cardiovascular diseases.\[[@ref11][@ref16][@ref18][@ref25][@ref26][@ref27]\] Diets rich in whole grains, fruits, vegetables, nuts, with a moderate alcohol intake, a lower intake of red meat, processed refined foods, sweets, dairy products with high fat, and soft drinks have been correlated with a reduced risk of type 2 diabetes mellitus, better glycemic control, and lipid profile in patients with type 2 diabetes mellitus.\[[@ref15][@ref28][@ref29][@ref30][@ref31][@ref32][@ref33][@ref34]\] Our diet is a salt-restricted plant-based vegan diet shown to have better glycemic control than seen with the earlier studies. The residential program was an intensive self-management intervention in type 2 diabetes mellitus populations the effects of which led to better control of type 2 diabetes mellitus in the long-term than type 2 diabetes mellitus patients undergoing standard of care. The effects are similar to structured self-management programs that have improved glycemic control in type 2 diabetes mellitus.\[[@ref35]\] One of the major findings of the study was the follow-up and long-term effects conferred by naturopathy intervention in type 2 diabetes mellitus patients unlike earlier studies that have only looked at acute short-term effects of these interventions. The study highlights the potential issues with adherence to these interventions in a field setting. Furthermore, the results also suggest that the effects of this intensive intervention are sustained for a longer duration (3 months after the end of intervention). The biochemistry parameters were carried out in external accredited laboratory and laboratory personnel were blinded to the study allocation. Unlike earlier studies, the pharmacological management of the study population was closely followed up. One of the major limitations of the study was that this was not a randomized controlled study. Second, it was also not possible to have a waitlist design as the intervention and follow-up was for over a year. Third, since patients received multicomponent interventions such as low-salt diet, plant-based diet, yoga, and physical activity, it is difficult to delineate the effects to any single intervention in our study. This is a prospective cohort study, and the findings from this study are hypotheses generating and cannot be deemed confirmatory. Overall, the residential naturopathy and yoga intervention shows promising effects in bringing glycemic control in type 2 diabetes mellitus patients. The short-term effects of these interventions are profound and last for a longer time. However, the effects wane with time as seen with 1-year follow-up of these patients. More randomized controlled studies with active control arm are needed to validate the effects of low-salt diet in these populations. Financial support and sponsorship {#sec2-13} --------------------------------- Nil. Conflicts of interest {#sec2-14} --------------------- There are no conflicts of interest.

1. Introduction {#sec1} =============== Anti-neutrophil cytoplasmic antibody- (ANCA-) associated vasculitis (AAV) is a group of heterogeneous diseases, including microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA), eosinophilic granulomatosis with polyangiitis (EGPA), and renal-limited AAV \[[@B1]\]. Diffuse alveolar haemorrhage (DAH), one of the major lung involvements in MPA patients, induces acute respiratory failure and is, therefore, associated with a poor prognosis in these patients \[[@B2], [@B3]\]. In general, most DAH patients have various respiratory symptoms, such as breathlessness, cough, haemoptysis, dyspnoea, and chest pain \[[@B4]\]. Herein, we report a case of chronic and asymptomatic DAH in a patient with MPA. In addition, we reviewed the reported cases of chronic progressive and asymptomatic DAH in patients with AAV. 2. Case Presentation {#sec2} ==================== A 78-year-old Japanese woman was found to have anaemia and reticular shadows on her chest X-ray during a health examination, for which she came to our hospital. She did not have any symptoms, including fever, cough, and purulent sputum, at that time. Her dietary habits were normal and she had not experienced any weight loss. Her haemogram and serum biochemistry revealed the following: haemoglobin: 7.8 (11.3--15.2) g/dL, MCV: 78.5 (79--100) fL, MCHC: 29.5 (30.7--36.6)%, serum iron: 62 (43--172) *μ*g/dL, ferritin: 67.2 (4.6--204.0) ng/dL, unsaturated iron-binding capacity: 188 (137--327) *μ*g/dL, vitamin B12: 470 (180--914) pg/mL, folic acid: 3.8 (3.1--9.7) ng/mL, blood urea nitrogen: 20.9 (8--22) mg/dL, creatinine: 1.13 (0.4--0.7) mg/dL, sodium: 145 (138--146) mEq/L, potassium: 3.5 (3.6--4.9) mEq/L, and chlorides: 109 (99--109) mEq/L. Her faecal occult blood test was negative. We diagnosed the cause of anaemia as defective iron utilization, and we followed up her laboratory data and chest X-ray findings to clarify the cause of defective iron utilization. Three months later, she was admitted to the hospital with renal failure, chest radiograph abnormality, and severe anaemia. She had a past medical history of hypertension but had never smoked and had no history of tuberculosis (TB). She had not been exposed to any fine particles. She did not have fever or any respiratory symptoms, such as breathlessness, cough, haemoptysis, dyspnoea, and chest pain. Physical examination showed a body temperature of 36.3°C and no crackles were audible on auscultation. Her chest X-ray showed persistence of the reticular shadows in the right middle and lower lung fields. In addition, a calcified lesion was observed in the left middle part ([Figure 1(a)](#fig1){ref-type="fig"}). Chest computed tomography (CT) showed a ground glass shadow and consolidation in the right upper and lower lobes with left sided pleural thickening and calcification (Figures [1(b)](#fig1){ref-type="fig"}--[1(e)](#fig1){ref-type="fig"}). Her laboratory findings at this time were haemoglobin 6.9 g/dL, white blood cell count 6,600 (3,500--9,000)/*μ*L, blood urea nitrogen 61.5 mg/dL, creatinine 2.56 mg/dL, C-reactive protein (CRP) 0.0 (\<0.3) mg/dL, erythrocyte sedimentation rate (ESR) 90 (2--15) mm/h, myeloperoxidase (MPO)-ANCA 54 (\<10) EU, proteinase 3 (PR3)-ANCA \<10 (\<10) EU, and anti-glomerular basement membrane (GBM) antibody \<10 (\<10) EU. More than 100 counts per high power field of red blood cell (RBC) casts were found on urinalysis. Renal biopsy was performed and pathological evaluation of the specimen revealed cellular crescent formations and lobulation in more than half the glomeruli ([Figure 2](#fig2){ref-type="fig"}). Evaluation of bronchoalveolar lavage fluid (BALF) revealed increased RBC counts, and cytology indicated that 90% of the detected macrophages were haemosiderin-laden. These findings indicated the presence of alveolar haemorrhage and, hence, she was diagnosed as MPA. She did not consent to undergo surgical lung biopsy at this time. We initially planned to treat her with corticosteroids and intravenous cyclophosphamide (IVCY), and, therefore, we first started treatment with corticosteroid therapy (prednisolone 50 mg/day) ([Figure 3](#fig3){ref-type="fig"}). After prednisolone therapy at this dose for 14 days, we started tapering it by 10 mg every 2 weeks, and once the dose of 20 mg was reached we tapered it by 5 mg every 4 weeks. This resulted in improvement in the renal failure, anaemia and lung abnormalities and, hence, IVCY was not required. Once the corticosteroid dose was reduced to 5 mg, it was combined with azathioprine. Currently, the patient is on maintenance therapy with corticosteroids (prednisolone 3 mg/day) and azathioprine (5 mg/day) without relapse of lung and renal lesions. She has been carefully followed up for development of TB by examination of chest X-rays once a month and sputum culture every three months, but with no TB prophylaxis. 3. Discussion {#sec3} ============= MPA is pathologically characterized by inflammation and fibrinoid necrosis of small vessel walls, leading to multiple organ involvement, including of the kidneys, lung, skin, digestive system, and nervous system. The aetiology of AAV is considered to involve genetic and environmental factors, such as silica exposure, infections, and drugs \[[@B5]\]. The pathogenesis of AAV is known to be related to both innate and adaptive immune systems. The complement system and neutrophils play major roles in the effector phase of the pathogenic immune response in AAV. Specifically, T helper (Th) cells have been noted to be involved and recently, Th17 cells were reported to contribute to production of proinflammatory cytokines (IL-1*β* and TNF-*α*), which results in the priming of neutrophils \[[@B5]\]. These mechanisms suggest that the pathogenesis and activity of AAV is strongly related to inflammation. DAH is a common and clinically important complication in MPA patients. Typical chest radiological patterns of DAH show focal or diffuse areas of ground glass opacification and/or consolidation \[[@B6]\]. However, these findings are nonspecific, and other aetiologies, such as lung infections, and interstitial pneumonia, show similar radiographic abnormalities. Fibreoptic bronchoscopy, particularly with BALF evaluation, is the best method to specifically diagnose DAH and for excluding other diseases \[[@B7]\]. Some studies of DAH patients with AAV showed that almost all patients have respiratory symptoms, such as dyspnoea and haemoptysis \[[@B3], [@B8], [@B9]\]. Additionally, DAH progresses rapidly and causes acute respiratory failure, such that the patients sometimes require admission to the Intensive Care Unit (ICU). Cartin-Ceba et al. reported that 34 of 73 patients with DAH secondary to AAV (47%) underwent mechanical ventilation (MV), and 41 patients (56%) were admitted to the ICU \[[@B8]\]. Moreover, DAH was reported to be an independent prognostic factor of MPA and the mortality rate of MPA patients with DAH is 8.65 times greater than that of MPA patients without DAH \[[@B10]\]. These data suggest that DAH induces various respiratory symptoms and that the development of acute respiratory failure in MPA patients is associated with high disease activity. Although our patient did not have any respiratory symptoms, we believe that the alveolar haemorrhage in the present case might have existed for more than 3 months before diagnosis, because the shadow in her right lung had gradually progressed over 3 months without any features of pneumonia or interstitial lung disease. However, the DAH was finally diagnosed by performing fibreoptic bronchoscopy and was successfully treated with corticosteroids. Three similar cases of chronic and asymptomatic DAH with MPA have been previously reported \[[@B11], [@B12], [@B13]\]. A review of the four cases, including the present case, is presented in [Table 1](#tab1){ref-type="table"}. In these cases, MPO-ANCA was the most detected antibody and anti-glomerular basement membrane (GBM) antibody was detected in one case. All the cases had severe but unidentified anaemia, suggesting chronic progression of anaemia. All three previous cases, but not our case, had a prolonged interval between DAH onset and diagnosis. Our patient, on the other hand, was diagnosed at an early stage, since we promptly performed fibreoptic bronchoscopy to identify the aetiology of the lung abnormalities, and we were, thus, able to make a definitive diagnosis from among DAH, lung infections, and interstitial pneumonia. Furthermore, detection of the DAH also led to a definite diagnosis of MPA in our case. Our experience suggests that physicians should aggressively perform fibreoptic bronchoscopy to identify the aetiology of lung disorders, even if patients do not have any respiratory symptoms or acute progressive lung findings. Furthermore, we believe that demonstrable anaemia is one of the clinical findings of chronic and asymptomatic DAH. Interestingly, almost all four cases did not have inflammatory reactions, such as elevation of fever and serum CRP. However, the present case and the patient described in \[[@B11]\] had elevated ESR levels. ESR tends to show false positive results in patients with anaemia and renal failure. A previous large scale study indicated that CRP levels are a better biomarker of inflammation than ESR and ESR is frequently a misleading biomarker which shows false positive \[[@B14]\]. The present case and the patient in \[[@B11]\] both had severe anaemia and renal failure, which could have resulted in false positive ESR levels. Regarding the relationship between the disease activity of AAV and biomarkers, Kronbichler et al. reported that CRP titre was one of the important factors of ANCA-associated disease activity \[[@B15]\]. Moreover, DAH patients with MPA without acute respiratory failure often demonstrate significantly lower CRP levels than patients with acute respiratory failure \[[@B8]\]. These data suggest that there may be rare phenotypes of DAH with MPA, which manifest with less inflammation and lower disease activity. Although the present case did not have an obvious medical history of TB, her X-ray showed pleural calcification, indicating the possibility of resolved TB. Immunosuppressive therapy and abnormal chest X-ray findings are reported to be associated with a high risk for development of active TB \[[@B16]\]. We did not administer therapy for TB to our patient in consideration of its adverse effects. Instead, we followed her up carefully for the development of TB by frequent chest X-ray and sputum culture examinations. In conclusion, we report a case of DAH with MPA that progressed chronically with no respiratory symptoms. Physicians should perform fibreoptic bronchoscopy in MPA patients with lung abnormalities and unidentified anaemia as soon as they can, even if the patients have no respiratory symptoms or elevation of CRP. Competing Interests =================== All the authors declare that there are no competing interests regarding the publication of this paper. {#fig1} {#fig2} {#fig3} ###### Clinical characteristics of cases of chronically progressive DAH with MPA reported in the literature, including the present case. ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Ref. number Age Gender (M/F) Detected antibody Duration at the onset of DAH Respiratory symptoms Complications inflammatory reaction such as fever, CRP titre -------------- ----- -------------- ------------------------- ------------------------------ ---------------------- ------------------------- ------------------------------------------------ \[[@B10]\] 61 F Anti-GBM antibody, ANCA 2 years Breathless Renal failure, anaemia\ N.A. (Hb 6.0 g/dL) \[[@B11]\] 23 F MPO-ANCA 4 years No Renal failure, anaemia\ no fever, normal CRP titre, 80 mm/h of ESR (Hb 6.3 g/dL) \[[@B12]\] 11 F MPO-ANCA 1 year No anaemia\ no fever (Hb 5.9 g/dL) Present case 78 F MPO-ANCA More than\ No Renal failure, anaemia\ no fever, normal CRP titre 3 months (Hb 6.9 g/dL) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- GBM: glomerular basement, ANCA: anti-neutrophil cytoplasmic antibody, MPO-ANCA: myeloperoxidase anti-neutrophil antibody, Hb: haemoglobin, CRP: C-reactive protein, and ESR: erythrocyte sedimentation rate. [^1]: Academic Editor: Tsai-Ching Hsu

Abbreviations ============= SCCOHT : small cell carcinoma of the ovary, hypercalcemic type IHC : immunohistochemistry LOH : loss of heterozygosity SWI/SNF : SWItch/Sucrose NonFermentable AT/RT : atypical teratoid/rhabdoid tumor MRT : malignant rhabdoid tumor MRTK : malignant rhabdoid tumor of the kidney SCCOHT: A Rare, Lethal, and Complex Cancer {#s0001} ========================================== Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT), is a rare and deadly ovarian cancer that predominantly affects young women. Fewer than 500 cases have been described in the literature since the disease was first characterized and it accounts for less than 1% of all ovarian cancer diagnoses.[@cit0005] While the average age of diagnosis for most ovarian cancers is 63 years, SCCOHT primarily arises in young women at an average age of 24 years[@cit0005]^,^[@cit0009] Histologically, SCCOHT is characterized by sheets of small rounded cells with hyperchromatic nuclei, single nucleoli, minimal cytoplasm and brisk mitotic activity. Roughly half of tumors contain variable numbers of larger cells with a luteinized or rhabdoid appearance. Though the presence of follicle-like spaces is diagnostically informative, other nonspecific morphologic and immunohistochemical features render the diagnosis challenging to establish.[@cit0005] Indeed, until the discoveries outlined in this addendum, no specific immunohistochemical markers existed. Although SCCOHT is often diagnosed at an early stage, little evidence exists to support treatment selection and the prognosis is dismal with 2-year survival being less than 35%.[@cit0005] SCCOHT\'s early age of onset and aggressive clinical course clearly establish a pressing need for innovations in management of this disease. Inactivating SMARCA4 Mutations in SCCOHT {#s0002} ======================================== Prior to the discoveries described below, SCCOHT\'s molecular etiology was understood primarily according to its clinical pathology and expression profile. No mutations had been identified by targeted sequencing of candidate genes such as KRAS, BRAF, BRCA1, BRCA2, and TP53 and the genome was seen to be predominantly diploid by comparative genomic hybridization.[@cit0011] Now, next-generation sequencing analyses from independent laboratories have reframed our biological understanding of SCCOHT by revealing that nearly all tumors harbor inactivating, often bi-allelic, mutations in the chromatin-remodeling tumor suppressor gene SMARCA4.[@cit0001] We previously sequenced tumor or germline DNA from 12 SCCOHT cases and the patient-derived BIN-67 cell line, identifying inactivating mutations in 9 of these samples.[@cit0001] We also found 15 of 18 cases with loss of SMARCA4 protein expression by immunohistochemistry. We now report the SMARCA4 status of an additional 12 SCCOHT tumors, 3 matched germlines, and the cell line SCCOHT-1[@cit0014], bringing the total number of cases analyzed in our hands to 24. This analysis was performed by PCR amplification of all coding exons of the SMARCA4 gene using DNA extracted from formalin-fixed paraffin-embedded (FFPE) blocks followed by Sanger sequencing in addition to immunohistochemistry against SMARCA4 and SMARCB1 as previously described.[@cit0001] In total, we have now identified 19 of 24 sequenced tumors with SMARCA4 mutations and 16 of 19 stained tumors with loss of SMARCA4 protein ([**Table 1**](#t0001){ref-type="table"}). Table 1.SMARCA4 mutations identified in DNA from SCCOHT patients and cell linesSample IDPublicationAge at diagnosis (years)SMARCA4 mutationsIHC   GermlineTumorSMARCA4SMARCB1SCCO-001New case22N/Ap.Ala161ValNegativePositive    p.Ala532fs  SCCO-004New case32Nonep.Val204fsNegativePositiveSCCO-005New case18Nonep.Asp1299fsN/AN/ASCCO-006New case32Nonep.Trp764fsNegativePositive    p.Gly836\*  SCCO-007New case25N/Ap.Gln331\*NegativePositive    p.Ile542fs  SCCO-009New case27N/Ap.Tyr1050fsNegativePositiveSCCO-011New case30N/AHomozygous p.Arg1189\*NegativePositiveSCCO-016New case12N/AHomozygous p.Arg1329fsNegativePositiveSCCO-018New case5N/ANonePositiveNegative+SCCO-019New case27N/Ap.Phe844fsNegativePositive+SCCOHT-1New caseTumor cell lineN/Ap.Pro1180fs p.Arg1077\*Negative+N/ASCCO-002Ramos et al.26NoneNoneNegativePositiveSCCO-008Ramos et al.9p.Arg979\*N/AN/AN/ASCCO-010Ramos et al.6NoneNone^2+^PositiveNegativeSCCO-012Ramos et al.21N/ANoneNegativePositiveSCCO-014Ramos et al.33N/Ap.Glu667fsN/AN/A    p.Leu1161fs  SCCO-015Ramos et al.27N/Ap.Arg1189\*N/AN/ASCCO-017Ramos et al.10p.Gly241fsHomozygous p.Gly241fs^2+^NegativePositiveDAH23Ramos et al.30N/Ac.2438+1_2438+2insTGANegativeN/ADAH456Ramos et al.39NoneNonePositivePositiveDAH457Ramos et al.23N/Ap.Arg1093\*N/APositiveDG1006Ramos et al.34Nonep.Glu952fsNegativeN/A    p.Ser1591fs  DG1219Ramos et al.37Nonec.3168+1\>ANegativeN/ABIN-67Ramos et al.Tumor cell lineN/Ac.2438+1G\>ANegativePositive    c.2439--2A\>T  [^2][^3] Across all published studies to date and including the new data reported here, nearly 100 mutations have been identified in SMARCA4 ([**Fig. 1**](#f0001){ref-type="fig"}) in 64 of 69 SCCOHT cases including 2 cell lines ([**Table 1**](#t0001){ref-type="table"} and **Supplementary Table 1**).[@cit0001] With the exception of 3 missense mutations, all other SMARCA4 mutations identified in SCCOHT are truncating, frameshift, deletion, or splice-site mutations. Two of the 3 missense mutations were found in SMARCA4-negative tumors bearing a second inactivating SMARCA4 mutation, while in one case the tumor harbored the missense mutation p.Gly1080Asp and loss of heterozygosity (LOH) alongside SMARCA4 protein retention.[@cit0003] Bi-allelic inactivation of SMARCA4 in SCCOHT is common either through the presence of 2 mutations or a single mutation and LOH at the SMARCA4 locus.[@cit0003] In keeping with these findings, immunohistochemistry has revealed loss of SMARCA4 protein in 54 of 61 SCCOHT tumors and cell lines presumably due to nonsense-mediated decay as has been shown in several cases.[@cit0001] However, a number of SMARCA4 negative cases carry heterozygous nonsense mutations and 2 cases have been shown to lack SMARCA4 protein with no identified sequence, copy number, or methylation alterations in the SMARCA4 gene ([**Table 1**](#t0001){ref-type="table"} and **Supplementary Table 1**).[@cit0001] Mechanisms leading to gene inactivation in SMARCA4-negative tumors with heterozygous or unidentified gene alterations remain to be elucidated. Figure 1.Schematic of SMARCA4 mutations in SCCOHT. SMARCA4 mutations identified in germline and tumor DNA from 62 SCCOHT patients, and in 2 SCCOHT cell lines (Case 103 from Jelinic et al. with exon deletion is not shown).[@cit0001] QLQ, Gln, Leu, Gln motif; HSA, helicase/SANT-associated domain; BRK, brahma and kismet domain; DEXDc, DEAD-like helicase superfamily domain; HELICc, helicase superfamily C-terminal domain; Bromo, bromodomain. Conversely, all but 4 SMARCA4-mutant SCCOHTs for which IHC has been performed also lack expression of SMARCA4 protein.[@cit0001] These 4 SMARCA4-mutant, positive-staining tumors harbored either splice site or missense mutations or, in one case, an in-frame homozygous deletion of exons 25 and 26 that resulted in expression of an inactive protein product.[@cit0003] In our cohort, we also found 3 tumors that had no SMARCA4 mutations and showed retention of protein expression by IHC. Two of these cases, both associated with paraneoplastic hypercalcemia, lacked the protein SMARCB1, a SWI/SNF-associated tumor suppressor gene known to be frequently mutated in rhabdoid tumors (**Supplementary Fig. 1**). Targeted Sanger sequencing of the coding exons of SMARCB1 in these tumors revealed a novel homozygous frameshift mutation, p.Asn34fs, resulting from the deletion of 14 base pairs in exon 2 of SMARCB1 in SCCO-010, a large cell variant SCCOHT. This finding supports the hypothesis that SCCOHT may share an etiological link with rhabdoid tumors and that SMARCB1 inactivation can also promote the development of SCCOHT.[@cit0001] The third case retained both SMARCA4 and SMARCB1 protein expression and may bear an as-yet unidentified SCCOHT driver gene mutation or may simply be a misdiagnosis.[@cit0001] Overall, SMARCA4 is clearly a tumor suppressor inactivated by 2 hits in the majority of SCCOHTs, but several of the above exceptional cases provide clues to a more complex disease etiology. Further supporting the prominence of SMARCA4\'s tumor suppressor role in SCCOHT, germline mutations have been identified in 17 SCCOHT cases, predominantly in younger patients.[@cit0001] Such mutations have been found to segregate in 4 families in which all affected members whose tumors could be tested developed either a second inactivating mutation or LOH in the remaining wild-type allele.[@cit0003] Alongside previous clinical descriptions of SCCOHT families, these mutations elucidate a heritable component to the disease and suggest that the broad age distribution of SCCOHT could reflect inherited versus acquired SMARCA4 mutations.[@cit0011]^,^[@cit0016] SMARCA4 mutation also occurs in the absence of recurrent secondary genomic alterations and amidst relative karyotypic stability and, therefore, appears to be the primary driving event in SCCOHT tumorigenesis. The total number of somatic non-silent mutations detected by paired exome or whole-genome sequencing analysis in SCCOHT tumors and matched normal DNAs ranges from 2 to 12, reflecting a low mutation rate, similar to other pediatric tumors and tumors of non-self-renewing tissues.[@cit0001] Among paired tumor and normal samples evaluated by exome, whole-genome, or panel-based sequencing, few secondary mutations in cancer genes were discovered and each such mutation (those in ASXL1, JAK3, NOTCH2, and WT1) occurred in only a single case.[@cit0001] Overall, the low SCCOHT mutation rate, the nearly universal presence of inactivating SMARCA4 mutations in SCCOHT, the presence of these mutations in patient germlines and families, and the lack of recurrent secondary alterations in these tumors strongly suggest that loss of SMARCA4 is sufficient for SCCOHT initiation. Biological, Diagnostic, and Therapeutic Implications of SMARCA4 Loss in SCCOHT {#s0003} ============================================================================== SMARCA4 is one of 2 mutually-exclusive ATPases of the SWItch/Sucrose NonFermentable (SWI/SNF) chromatin-remodeling complex which was originally discovered to modulate mating type-switching and sucrose fermentation in yeast.[@cit0023] This complex uses the energy of ATP hydrolysis to destabilize histone-DNA interactions and move, eject, or restructure nucleosomes, thereby regulating access to DNA of transcription, replication, and repair machinery.[@cit0023] SWI/SNF subunits such as SMARCA4 have also been shown to interact with tumor suppressors such as p15^INK4b^, p16^INK4a^, p21^CIP/WAF1^, and hypophosphorylated RB to modulate cell cycle progression.[@cit0027] Broadly, mutations in epigenetic regulators such as SWI/SNF members are enriched in many cancer types, particularly pediatric cancers in which as many as 30% of brain tumors and leukemias and 17% of solid tumors bear such mutations.[@cit0029] This enrichment may be due to the pleiotropic effects these mutations exert on gene expression and cell differentiation programs. Many SWI/SNF subunits such as SMARCA4 have also been shown to interact with, or regulate, tumor suppressors with approximately 20% of cancers bearing mutations in these genes.[@cit0030] SMARCA4 is one of the most commonly mutated subunits across cancer types, occurring at a frequency of about 4% in all cancers and arising regularly in non-small cell lung cancer, Burkitt\'s lymphoma, and medulloblastoma while also occurring occasionally in melanoma, pancreatic adenocarcinoma, ovarian clear cell carcinoma, and other tumor types.[@cit0031] Loss of SMARCA4 is thought to lead to dependence on SMARCA2-bearing SWI/SNF complexes that induce gene expression changes driving oncogenic pro-survival and/or anti-apoptotic signaling.[@cit0032] Elucidation of the impact of such mutations on SWI/SNF composition and downstream effects on expression programs and pathway regulation will shape future study of SCCOHT tumorigenesis and therapeutic vulnerability. Given SCCOHT\'s complex histological appearance and the absence of known precursor lesions, the cellular origin of SCCOHT and its relationship to other tumor types remains unclear. SCCOHTs are characterized by poorly differentiated small tumor cells with scant cytoplasm and hyperchromatic nuclei, and the presence of follicle-like structures contained within sheets of cells.[@cit0005] Despite SCCOHT\'s name, about half of tumors have populations of large cells with rhabdoid features.[@cit0005] Indeed, there are many similarities between SCCOHT and atypical teratoid/rhabdoid tumors of the brain (AT/RTs) and malignant rhabdoid tumors (MRTs) of the kidney (MRTK). All 3 tumor types are linked to mutations in the SWI/SNF genes SMARCB1 (AT/RT, MRT, and now SCCOHT) or SMARCA4 (SCCOHT and AT/RT), all have diploid genomes and all occur in young or pediatric patients.[@cit0005] Shared morphology and mutational spectra make a compelling case that SCCOHT may be a type of MRT.[@cit0015] The strikingly similar morphology and genetics of rhabdoid tumors in 3 very different organs suggests either a common cell of origin or convergent morphologic evolution upon SMARCA4 or SMARCB1 loss (or both) although no MRT cell of origin has yet been identified.[@cit0002] On the other hand, there is some histological evidence for a germ cell etiology for SCCOHT. In particular, a recent report identified immature teratoma in 2 SCCOHTs, one of which also contained foci of yolk-sac tumor.[@cit0002] This finding agrees with Ulbright et al. who, in one of the earliest publications on SCCOHT in 1987,[@cit0016] also suggested that SCCOHTs might be related to yolk-sac tumors based on presence of shared histopathological and ultrastructural features. Unfortunately, no clear origin has been indicated by mouse models of SMARCA4 mutation. The homozygous Smarca4 null genotype is embryonic lethal and, while 10% of heterozygotes spontaneously develop mammary tumors at 1 year, these tumors are molecularly heterogeneous, genomically unstable, and lack LOH at the Smarca4 locus and therefore do not contain a genomic landscape resembling that of SCCOHTs.[@cit0036] Engineered SMARCA4 knockouts in putative precursor cells in vitro and in vivo are needed to shed light on SCCOHT histogenesis. Among ovarian tumors, the loss of SMARCA4 protein appears to be highly specific for SCCOHT. Our assessment of 485 primary ovarian epithelial, sex cord-stromal, and germ cell tumors showed only 2 tumors (0.4%), both clear cell carcinomas, with negative SMARCA4 staining.[@cit0001] Other ovarian tumors in the differential diagnosis of SCCOHT -- undifferentiated carcinomas, adult and juvenile granulosa cell tumor, and germ cell tumors -- all expressed SMARCA4 protein or were wild-type for the SMARCA4 gene.[@cit0001] The expression status of SMARCA4 remains to be determined in several other primary and metastatic ovarian tumors in the differential diagnosis of SCCOHT including endometrioid stromal sarcoma, desmoplastic small round cell tumor, primitive neuroectodermal tumor, neuroblastoma, and others. However to date, the absence of SMARCA4 protein is highly sensitive and specific for SCCOHT and can be used to distinguish it from other ovarian tumors with similar histology to facilitate diagnosis. SCCOHTs are extremely aggressive and refractory to treatment that most commonly includes surgical debulking followed by high-dose chemotherapy and/or radiation.[@cit0005] Some evidence suggests that chemotherapeutic combinations including cisplatin or carboplatin, etoposide and vinca alkaloids may be associated with improved survival, yet patient outcomes are abysmal in most cases with a 65% recurrence rate and 2-year survival less than 35%.[@cit0005] SCCOHT rarity limits the implementation of prospective clinical trials to guide effective treatment and its infrequency has also limited the study of its pathogenesis to uncover potential therapeutic vulnerabilities. Our finding that the majority of SCCOHTs contain SMARCA4 mutations amidst otherwise simple genomic backgrounds provides an opportunity to empirically develop effective treatment strategies with a high probability of impact for many of these patients. Given that this disease derives in virtually all cases from the loss of a tumor suppressor, the path to an effective small molecule may hinge on identification of a synthetic lethal target. To this end, a synthetic lethal dependence of SMARCA4-deficient cancers cells on SMARCA2 has recently been described in non-small cell lung cancer, ovarian and liver cancer cell lines.[@cit0032] This dependence is likely due to SMARCA2\'s status as the only known alternative ATPase subunit of the SWI/SNF complex. However, preliminary SMARCA2 staining in 2 SCCOHT cell lines showed lack of protein in both cases (data not shown), suggesting that SCCOHT may lack the expression of both SMARCA2 and SMARCA4 and that investigation of other synthetic lethal partners is therefore warranted. Although it has been shown in other cancers such as non-small cell lung adenocarcinoma cell lines that the SWI/SNF core complex still forms in the absence of both SMARCA4 and SMARCA2,[@cit0040] it remains to be determined whether this complex retains chromatin remodeling activity and whether targeting the residual complex can selectively kill SCCOHT cells. Of further importance will be identification of the mechanism inactivating the second SMARCA4 allele in cases in which only monoallelic mutations have been identified. Epigenetic lesions may present compelling targets for re-expression of SMARCA4 and/or SMARCA2. Ultimately, future progress in SCCOHT treatment will depend on expansion of the currently limited number of in vitro and in vivo model systems. The BIN-67 and SCCOHT-1 cell lines are the only such models to have yet been described and they have been implemented in few studies in vivo.[@cit0013] The breakthrough identification of inactivating SMARCA4 mutations in almost all cases of SCCOHT is the first significant insight into the pathogenesis of the disease and offers the opportunity for genetic testing of family members at risk. The loss of the SMARCA4 protein is a highly sensitive and specific marker of the disease, highlighting its potential role as a diagnostic marker. Studies are currently in progress at our institutions to elucidate the cell of origin in hopes of better understanding the pathogenesis of this disease and to identify therapeutic vulnerabilities guiding clinical trials to further advance treatment options for patients with SCCOHT. Materials and Methods {#s0004} ===================== Samples {#s0004-0001} ------- At TGen, all patients and their relatives signed consent forms according to IRB-approved and Health Insurance Portability and Accountability Act--compliant protocols. At the University of British Columbia, biospecimens were obtained from the Ovarian Cancer Research Program (OvCaRe) tissue bank in Vancouver, British Columbia, Canada; the University of Toronto in Toronto, Ontario, Canada; the Children\'s Oncology Group at Nationwide Children\'s Hospital in Columbus, Ohio, USA; and the Hospital de la Santa Creu i Sant Pau at the Autonomous University of Barcelona in Barcelona, Spain, using an IRB-approved protocol. All of the specimens were SCCOHT, with 4 cases (SCCO-009, SCCO-010, SCCO-017 and SCCO-019) classified as large cell variants of SCCOHT in their pathology reports. Cases of small cell carcinoma of pulmonary type were excluded from the study. DNA extraction {#s0004-0002} -------------- FFPE DNA was extracted using Qiagen\'s All Prep DNA/RNA FFPE kit (Qiagen; Valencia, CA). Blood leukocytes (buffy coat) were isolated from whole blood by centrifugation at room temperature and resuspended in Buffer RLT plus. Samples were then processed for DNA isolation using the AllPrep kit (Qiagen). DNA was quantified using the Nanodrop spectrophotometer (Nanodrop; Wilmington, DE) on the basis of 260 nm/280 nm and 260 nm/230 nm absorbance ratios. PCR amplification and Sanger sequencing analysis {#s0004-0003} ------------------------------------------------ PCR amplification of SMARCA4 was performed using previously published primers[@cit0003] targeting 34 coding exons (the alternative exon 29 was not sequenced). Amplification of all SMARCB1 coding exons was performed using the following primers, some of which have been previously published[@cit0041]: Primer NameForward SequenceReverse SequenceExon 1CTTCCGGCTTCGGTTTCCCTGATGAATGGAGACGCGCGCTExon 2GTTGCTTGATGCAGTCTGCGTTCATGACATAAGCGAGTGGExon 3GATGTGCTGCATCCACTTGGTTCAGAAAAGACCCCACAGGExon 4TTAGTTGATTCCTGGTGGGCGAACTAAGGCGGAATCAGCAExon 5TGTGCAGAGAGAGAGGCTGACACGTAACACACAGGGGTTGExon 6CAATCTCTTGGCATCCCTTCCAGTGCTCCATGATGACACCExon 7TGGGCTGCAAAAGCTCTAACAGTTTTGCAGGGAGATGAGGExon 8GGCCAAAGCTTTCTGAGGATCATGGGAGACTGGGAAAGGCExon 9CCCTGTAGAGCCTTGGGAAGGTCCTTGCCAGAAGATGGAG Universal M13 tails were added to all primers. Each primer pair was mixed with 10 ng of genomic DNA and subjected to the following cycling parameters: 94°C for 2 min., 3 cycles at each temperature: 30 sec. at 94°C, 30 sec. at 60--57°C, 45 sec. at 72°C; 25 cycles: 30 sec. at 94°C, 30 sec. at 62°C, 45 sec. at 72°C; final extension of 5 min. at 72°C. All amplification reactions were performed using Platinum Taq DNA Polymerase \#10966--034 (Life Technologies; Carlsbad, CA). PCR amplicons were sequenced using M13 forward and reverse primers at the Arizona State University\'s DNA Laboratory (Tempe, AZ). Immunohistochemistry {#s0004-0004} -------------------- Whole slide sections were prepared from paraffin blocks of formalin-fixed SCCOHT tumor cases and SCCOHT-1 cells. Unstained slides were processed using the Ventana Discovery Ultra system (Ventana Medical Systems), using a rabbit monoclonal antibody to SMARCA4 (BRG1; Abcam, ab110641; 1:25 dilution) and mouse monoclonal antibody to SMARCB1 (INI1; BD Transduction Laboratories, 612110; 1:50 dilution). The antibody to SMARCB1 was used to confirm the antigenic reactivity of the tumor cells and cell lines that were negative for SMARCA4 expression. Tumors were scored positive if any tumor cell nuclei showed moderate to strong (definite) positive nuclear staining. Tumors were scored negative when tumor cells showed no nuclear staining only if there was adequate nuclear staining of an internal positive control (endothelial cells, fibroblasts or lymphocytes). No cytoplasmic staining was seen for SMARCA4. Disclosure of Potential Conflicts of Interest {#s0004-0005} ============================================= No conflicts of interest were disclosed. Acknowledgments =============== We particularly thank the families and patients who joined this IRB-approved study for their critical contributions. We further thank Scottsdale Healthcare for institutional support and clinical leadership in addition to that provided by many other physicians: Drs. Jaime Prat, Emanuela D'Angelo, Blaise Clark, Joseph Pressey, and Richard Roden. We also thank the faculty and staff at TGen of the Macromolecular Analysis & Processing Center (Drs. Hostetter and LoBello), and the Office of Research Compliance & Quality Management (Lora Nordstrom, Stephanie Althoff, and Stephanie Buchholtz); Children\'s Oncology Group for sample collection and Drs. Ralf Haas and Barbara Vanderhyden for establishing and sharing the SCCOHT-1 and BIN-67 cells, respectively. Funding ======= This study was supported by grants from The Marsha Rivkin Center for Ovarian Cancer Research, The Anne Rita Monahan Foundation, The Ovarian Cancer Alliance of Arizona, The Small Cell Ovarian Cancer Foundation, and Foster and Lynn Friess. Further support was provided to Yemin Wang by the Michael Smith Foundation for Health Research and to Anthony N. Karnezis and David G. Huntsman by the Terry Fox Research Initiative New Frontiers Program in Cancer. Supplemental Material {#s0006} ===================== Supplemental data for this article can be accessed on the [publisher\'s website](http://dx.doi.org/10.4161/2167549X.2014.967148). ###### 967148_Supplementary_Materials.zip [^1]: These authors contributed equally to this work. [^2]: ^+^This tumor was not previously stained for this marker [^3]: ^2+^The tumor for this case was not previously sequenced

Background {#Sec1} ========== Renal transplantation remains an effective treatment for end-stage renal dysfunction \[[@CR1]\], facilitating a return to normal health and prolonging life. However, antibody-mediated rejection is a major issue that affects long-term renal allograft survival. Despite the rapid development of new immunosuppressive drugs for attenuating acute rejection, improving the long-term survival of grafts is still a challenge mainly because of chronic allograft rejection. The features of chronic renal allograft rejection are hypertension, proteinuria, progressive deterioration of graft function, peritubular capillary C4d deposition, presence of donor-specific antibodies (DSA) and morphological changes with transplant vasculopathy, glomerulopathy, fibrosis and lymphocyte infiltration. However, the causes leading to chronic rejection are complex and not well understood yet \[[@CR2]\]. Allograft rejection is characterized by an increase in activated CD4+ T-lymphocytes, especially regulatory and cytotoxic T cells, leading to an imbalance of immune responses in the transplant recipients \[[@CR3], [@CR4]\]. Functionally, CD4+ T helper cells that interact with antigen-specific B cells are required for the production of alloantibodies \[[@CR5]\]. Among them, Tfh cells, a recently defined subset of CD4+ T cells, play a particular role in mediating B cell-driven allogeneic responses. Tfh cells can migrate into germinal centers and promote B-cell activation and differentiation into immunoglobulin-producing plasmablasts or plasma cells \[[@CR5]\]. They can express PD-1, CXCR5, ICOS, IL-21 and the transcription factor B-cell lymphoma 6 (Bcl-6) \[[@CR6], [@CR7]\], thereby displaying their regulatory functions. Circulating Tfh cells, peripheral counterparts of conventional Tfh cells, express PD-1, CXCR5, ICOS and IL-21, but not Bcl-6 \[[@CR5]--[@CR7]\]. They play an important role in human humoral immunity through these functional molecules. Their abnormal activities are critically involved in the onset of several human diseases such as autoimmune disorders, cancer and infective diseases \[[@CR7]--[@CR10]\]. Therefore, an alteration in circulating Tfh cells may be correlated with disease conditions and might be used as a biomarker of certain diseases \[[@CR11]--[@CR13]\]. Moreover, recent clinical studies have shown that peripheral Tfh cells in the kidney transplant recipients with acute rejection can regulate B-cell alloreactivity and the number of these Tfh cells alters the immunization status and DSA levels \[[@CR14]\]. However, their function and relevance to chronic renal allograft rejection are not known yet. This study was conducted to explore the potential association between circulating Tfh cells and chronic rejection in kidney transplant recipients. The outcome results may provide a useful hint for clinical prediction of renal status after transplantation and for a potential new therapy for chronic allograft rejection. Methods {#Sec2} ======= This study was approved by the Institutional Ethics Committee of Third Affiliated Hospital of Soochow University, Jiangsu Province, China. Written-informed consent was obtained from all participants of the study. Subjects {#Sec3} -------- The patients with primary renal transplantation for 1--3 years were enrolled from October 2013 to December 2014. Totally 42 recipients were studied in this work, including 24 patients with chronic rejection (CR group) and 18 patients with normal renal function as the normal control (NC group). All of them received the treatment with cyclosporine A, methylprednisolone and mycophenolate mofetil or azathioprine after the renal transplantation. The diagnosis for chronic allograft rejection was confirmed by renal biopsy, biochemical measurements and immunological assays, including DSA as described in other studies \[[@CR15]--[@CR18]\]. In specific,the diagnostic criteria included 1) clinical evidence of slowly deteriorating graft function; 2) biopsy evidence and diffuse deposition of C4d; and 3) the presence of circulating DSA at the time of biopsy. Their peripheral blood samples were collected in a standard way by the clinical laboratory of the hospital. All the subjects had no infective disease when sampling blood. Surface staining and flow cytometry analysis {#Sec4} -------------------------------------------- The whole blood was subjected to flow cytometry detection by a BD bioscience FACSCantoII cytometer with FACSDiva software for measuring the frequency of circulating Tfh cells and B cells as well as the expression of their surface markers. The following conjugated monoclonal antibodies were used to stain the cells: CD4-FITC, CXCR5-APC, ICOS-PE, PD-1-PE, CD19-PE-Cy5.5, ICOSL-PE and PDL-1-APC. The cells were incubated with the antibodies for 30 min at room temperature in the dark. Totally 50,000 lymphocytes were acquired in each sample. Data were analyzed using Flow Jo software 7.6.1. Enzyme-linked immunosorbent assay (ELISA) {#Sec5} ----------------------------------------- The levels of serum IL-21 were quantified by using the human IL-21 ELISA kit (eBioscience) according to the manufacturer's instructions. The concentration in each individual sample was calculated according to the standard curve. Statistical analyses {#Sec6} -------------------- All experimental data were analyzed by Graph Prism version 5.0. The results were expressed as mean ± SD and subjected to *t* test for statistical comparisons between the NC and CR groups. If a *p*-value was found to be less than 0.05, the result would be considered statistically significant. Results {#Sec7} ======= Demographics {#Sec8} ------------ The general information of the renal transplant recipients was summarized in Table [1](#Tab1){ref-type="table"}. Gender and age were similar between CR and NC groups. The mean serum creatinine (sCr) and blood urea nitrogen (BUN) were almost three-fold higher in the CR patients than those of NC group (*p* \< 0.001).Table 1General information of the renal transplant patientsGroupGenderAgeSCrBUN(male/female)(μmol/L)(mmol/L)NC16/842 ± 795.4 ± 18.65.3 ± 1.6CR12/649 ± 9277.0 ± 124.3\*\*\*15.2 ± 7.8\*\*\**NC*, normal control; *CR*, chronic rejection. *P* \< 0.001 \*\*\* vs. NC Increased circulating Tfh cells in the CR patients {#Sec9} -------------------------------------------------- To determine if chronic rejection was associated with an alteration in circulating Tfh cells in the renal transplantation recipients, we first evaluated the frequency of CD4 + CXCR5+ Tfh cells through flow cytometry. As shown in Fig. [1](#Fig1){ref-type="fig"}, the percentage of CD4 + CXCR5+ Tfh cells among total CD4+ T cells was significantly increased in the CR group as compared to that of the NC group (35.3 ± 8.5 % vs.19.0 ± 5.0, *P* \< 0.001).Fig. 1Frequency of Tfh cells in the patients with renal allograft. **a**, representative contour plots of the ratio of Tfh cells in the NC and CR groups, **b**, mean values of the frequency of Tfh cells in the two groups. \*\*\*, *P* \< 0.001. Note a significant increase in Tfh cells in the CR group compared to that of the control group Differential changes in PD-1, CXCR5, ICOS, and IL-21 of Tfh cells in the CR patients {#Sec10} ------------------------------------------------------------------------------------ We further detected the changes of PD-1, CXCR5 and ICOS in the high frequency of Tfh cells and found that PD-1 was significantly down-regulated in these Tfh cells (Fig. [2a](#Fig2){ref-type="fig"}). In sharp contrast, there were no significant changes in CXCR5 and ICOS in the CR group as compared to the NC group (Fig. [2b](#Fig2){ref-type="fig"} and [c](#Fig2){ref-type="fig"}).Fig. 2Comparative changes in PD-1, CXCR5 and ICOS expression on Tfh-cell surfaces. **a**,PD-1. **b**, CXCR5. **c**, ICOS. \* *P* \< 0.05. Note a significant decrease in PD-1 expression with no appreciable changes in CXCR5 and ICOS expression in the CR patients as compared to the NC group IL-21 is a biological hallmark of Tfh cells because this immune cytokine is produced by Tfh cells and is involved in mediating Tfh-B-cell interaction \[[@CR6]\]. Therefore, we further measured the concentration of IL-21 in the serum of the CR patients (Fig. [3](#Fig3){ref-type="fig"}). Similarly as the changes in CXCR5 and ICOS, serum IL-21 did not show any significant change in the CR patients as compared to that of the control group (406.9 ± 123.9 pg/ml vs. 449.1 ± 101.7 pg/ml, *P* \> 0.05).Fig. 3The level of serum IL-21 in renal allograft patients. Note that there was no statistic difference in the IL-21 level between the CR and NC groups No changes in circulating B cells, PDL-1 and ICOSL in the CR patients {#Sec11} --------------------------------------------------------------------- Since B cells are key players in the graft rejection \[[@CR19]\], we further determined the frequency of total B cells and compared their change with that of Tfh cells. The ratio of B lymphocytes was not significantly different between CR and NC groups (Fig. [4](#Fig4){ref-type="fig"}), unlike Tfh cells. Because PDL-1-expressing B cells interact with PD-1+ Tfh cells to regulate the maturation and survival of B cells \[[@CR20]\], we next detected the expression of PDL-1 in B cells. Unlike the change in the PD-1 of Tfh cells, PDL-1 did not decrease in B cells at all in the CR group (Fig. [5a](#Fig5){ref-type="fig"}). Also, ICOSL had no significant change in the CR patients (Fig. [5b](#Fig5){ref-type="fig"}).Fig. 4The frequency of B cells. Note that there was no significant difference between two groupsFig. 5The expression of PDL-1 and ICOSL in B cells. Note that there were no appreciable differences in both PDL-1 (**a**) and ICOSL (**b**) between two groups All these results suggest that the increase in circulating Tfh cells with PD-1 down-regulation is a specific and characteristic change in the CR patients. Discussion {#Sec12} ========== We have made a novel finding in this work, i.e., a major increase in circulating Tfh cells with a significant decrease in PD-1 in the patients with chronic renal allograft rejection. In sharp contrast, B cells and the alloimmune-regulating molecules such as CXCR5, ICOS, ICOSL, PDL-1 and IL-21 did not show any appreciable change in parallel. Tfh cells display multiple features for their helper functions in secondary lymphoid organs or tissues with inflammation \[[@CR21]\]. They migrate into B cell follicles of germinal centers \[[@CR22]\] and thereby help B cells generate antibodies for humoral immunity \[[@CR3], [@CR6], [@CR23]\]. In fact, Tfh cells, as an immune regulator, are critically involved in the pathological processes of many immune diseases \[[@CR8]\]. In renal allograft rejection, the germinal center reactions are dependent on Tfh, while B cells are indispensable for the immune attack to the newly transplanted kidney \[[@CR24]\]. Tfh cells express PD-1 \[[@CR6]\], while B cells produce PDL-1, an endogenous ligand of PD-1 \[[@CR19]\]. The PD-1/PDL-1 signaling has been shown to play an important role in regulating immune functions and affecting the activation of regulatory T cells, cytotoxic T cells and Dendritic cells \[[@CR25]\]. Such signaling also influences the generation and differentiation of Tfh and B cells themselves \[[@CR26]\]. Recent evidence suggests that blocking the PD-1 signaling induces an up-regulation of Tfh generation and differentiation, which may directly lead to autoimmune encephalomyelitis \[[@CR27]\]. In contrast, stimulating this pathway can prolong the survival of the patients after cardiac allograft transplantation \[[@CR28]\]. More recently, PD-1 ligands are found to protect the kidneys from ischemia reperfusion injury \[[@CR29]\]. In fact, PDL-1, the endogenous ligand of PD-1, has been demonstrated as a required factor for peripheral transplantation tolerance and protection aganist chronic allograft rejection \[[@CR30]\]. Taken together, PD-1 signaling is a key regulator for attenuating the Tfh cells and down-regulating the overreaction of humoral immunity against the transplanted kidneys. Therefore, the novel finding of the present study strongly suggests that the deficiency of PD-1 expression causes the increase in Tfh cells, thereby leading to an overreaction of humoral immunity against the allergenic organ, which may be a major reason for chronic allograft rejection. Tfh and B cells also express many other immune-regulating molecules such as CXCR5, ICOS, ICOSL, and IL-21 \[[@CR6], [@CR19], [@CR26]\]. All these molecules are actively involved in the regulation of immune function \[[@CR26]\]. For example, an increase in the expression of CXCR5 can enable Tfh cells to migrate into germinal centers \[[@CR21]\]. On the other hand, ICOS, another surface receptor like PD-1, also mediate the generation, development and function of Tfh cells by activating ICOS/ICOSL signaling \[[@CR31], [@CR32]\]. Moreover, IL-21,a pro-inflammatory cytokine secreted by Tfh cells,has an important role in Tfh cell differentiation, B cell proliferation \[[@CR33]\], and the expression of PD-1 \[[@CR34]\] and CXCR5 \[[@CR26]\]. However, all of these immune regulators did not shown any change in the patients with chronic allograft rejection. We are therefore confident that the deficient PD-1 expression with increased circulating Tfh cells is a specific and characteristic change in chronic allograft rejection. In the lymph nodes, CD4+ CXCR5+ Tfh cells are more effective in helping B cells than their peripheral counterparts \[[@CR7]\]. Humoral response in the lymph nodes can be suppressed by anti-CD40 mAb via regulating Tfh cells \[[@CR23]\]. In the transplanted kidneys with acute rejection, infiltrated Tfh cells have been found to participate in the antibody-mediated rejection \[[@CR14]\]. However, little is known about the role of these special Tfh cells in the transplanted kidneys with chronic rejection. We speculate that the increase in circulating Tfh cells with a decrease in PD-1 expression might, at least partially, contributes to the genesis of the renal chronic rejection by migrating and infiltrating into germinal centers of renal allografts and lymphoid organs. We will further clarify this issue in our future work. In addition, pre-existent DSA storing before renal transplantation and *de-novo* DSA developing after renal transplantation are associated with antibody-mediated rejection and allograft failure \[[@CR35]\]. However, recent studies have shown that despite the numbers of circulating Tfh cells were higher in the patients with pre-existent DSA than those without pre-existent DSA, the levels of circulating Tfh cells were not different among the patients with or without *de-novo* DSA \[[@CR35], [@CR36]\]. Therefore, the relationship between the frequency of circulating Tfh cells and the level of DSA in renal allograft rejection is not clear yet and needs more investigations. Conclusions {#Sec13} =========== Our first data show that decreased PD-1 expression may contribute to the increase in circulating Tfh cells in the patients with chronic renal allograft rejection. This finding provides a potential hint for a new target for the treatment of chronic rejection. Moreover, a dynamic change in the expression of PD-1 and the number of circulating Tfh cells may be used as an index for monitoring chronic allograft rejection after kidney transplantation as. Tfh : Follicular helper T cells CXCR5 : Chemokine receptor type 5 ICOS : Inducible T cell co-stimulator PD-1 : Programmed death-1 ICOSL : Inducible T cell co-stimulator ligand PDL-1 : Programmed death-1 ligand IL-21 : Interleukin-21 Bcl-6 : Transcription factor B-cell lymphoma 6 CR : Chronic rejection NC : Normal control ELISA : Enzyme-linked immunosorbent assay sCr : Serum creatinine BUN : Blood urea nitrogen DSA : Donor-specific antibodies. **Competing interests** The authors declare that they have no competing interests. **Authors' contributions** JS, XX, XH and YX conceived and designed the experiments. JS, FL and QS performed the experiments. JS analyzed the data. XH provided research reagents. JS and YX wrote the paper. All authors read and approved the final manuscript. This work was supported by the National Natural Science Foundation of China (Grant No.81273267) . YX was partially supported by NIH (AT-004422).

Introduction {#s0005} ============ Globally, approximately 30% of women who have ever been in an intimate relationship have reported physical or sexual violence by an intimate partner ([@bib39]). There is a growing body of evidence that intimate partner violence (IPV), and attitudes accepting of IPV, are socially clustered, supported by community and family social practices, and transmitted through families ([@bib31], [@bib33]). For example, people who have witnessed IPV as children are more likely to experience or perpetrate it as adults ([@bib29]), and female victims of IPV are more likely to report attitudes accepting of IPV ([@bib16]). While previous research has inferred social clustering of IPV acceptance through individual-level questions or aggregated measures at the level of states or other area units, few studies have used social network data to investigate these behavioral and attitudinal clusters. Conceptual model {#s0010} ---------------- The pattern of social ties in which a person is embedded, and the normative beliefs and practices of those to whom s/he is connected, may clearly affect an individual's beliefs and practices. In social norms theory, *reference groups* are those to whom an individual turns for information on the expected ways of behaving within group-specific contexts ([@bib2], [@bib30]). *Descriptive norms* are the behaviors commonly practiced in a group, and are supported through observation of what the majority of others are doing (or not doing) ([@bib8]). *Injunctive norms,* on the other hand, reflect the expectations of the community, and are often reinforced through perceived social or individual consequences in the form of sanctions. Because of the threat of sanctions, the proscribed behavior may rarely be observed, making it difficult to ascertain whether the behavior is simply uncommon or is actually against an underlying social norm ([@bib4]). Sanctions can be positive for compliance, and can include social rewards such as approval or inclusion in social groups. They can also be negative for non-compliance, and may be as overt as stoning or as subtle as quiet disapproval, or may simply consist of the withholding of social rewards ([@bib3]). Social network analysis is a powerful tool for investigating social norms among specific groups because it can identify the people to whom individuals are most closely connected and these people's salient characteristics. For example, previous work on latrine adoption in India demonstrated the social clustering of behaviors through social network analysis as well as the positive relationship between injunctive norms and the level of connection within a community ([@bib30]). There may be multiple reference groups for any given behavior, and social network analysis can be used to at least partially identify those groups and the levels and directions of their influence. The ability of social network analysis to identify these groups, however, will depend upon the questions used to elicit the social networks, the utility of those questions in capturing the relevant relationships, and the scale of the network study ([@bib31], [@bib32]). While a community may be opposed to IPV, families within that community may accept it ([@bib31]). If there are injunctive norms at the community level against perpetrating IPV, but IPV is occurring within families, then family-level norms may contribute to its continuation ([@bib31]). The family may be at least one of the reference groups to which individuals (subconsciously or consciously) turn for information on behavioral expectations regarding IPV. IPV is as an example of a practice that, because it often takes place in the privacy of the home, is generally less detectable than practices such as child marriage. With such practices, the varying influences of different reference groups is particularly important. Previous research has in fact pointed towards the possibility of an "inner norm" within the family that is supportive of IPV, versus an "outer norm" within the community that opposes it ([@bib31]). Thus, family-level characteristics, as opposed to those within the greater community, can offer important insights about factors that contribute to IPV in different contexts. However, families cannot easily change their views or practices if community norms are against it, as in the case of female genital cutting ([@bib38]). Research on violence has shown that family cohesion, defined as emotional support and positive communication, can be protective against violence among youth ([@bib12]), possibly by providing protection against social stress, an environment of security, strong parental monitoring, and positive family communication ([@bib20]). Consistent with the findings on family cohesion, research on community-level violence has also identified community social cohesion as an important factor in violence prevention ([@bib19]). While cohesive communities may protect against violence by providing a warm and nurturing environment, it is also hypothesized that more socially cohesive communities are able to maintain social control through the creation and maintenance of injunctive norms which can be used effectively to discourage violence ([@bib19]). This sort of control, however, can also be effective at encouraging violence in contexts in which violence is acceptable and normative. Literature on social networks has demonstrated that cohesion can reinforce norms, whether positive or negative ([@bib7], [@bib22]). Thus, understanding the association between cohesion and acceptance of violence, and the possible mechanisms by which this association occurs, is an important question in research on family-level violence prevention. Despite the evidence on the relationship between violence and social cohesion, few studies have considered family cohesion in the context of IPV ([@bib27]). Furthermore, the majority of cohesion research has operationalized cohesion using survey questions that ask respondents to report on the quality of their interactions within their families or communities, which can be subject to response bias depending upon who within the community or the family is being asked the questions. For instance parents are more likely to report positive parent-child interactions than are children ([@bib36]). Full social network data in developing countries is rare ([@bib28]). Here, we combine two very uncommon social network datasets, one from rural Honduras and one from rural Uganda, to investigate the household-level social network correlates of IPV acceptance. Although Honduras and Uganda represent distinct geographic settings, both countries represent low-income, low-resource societies with strong patriarchal cultural traditions ([@bib14], [@bib17]). In addition, both countries have a history of societal violence ([@bib18], [@bib5]) and exhibit strongly unequal gender norms ([@bib26], [@bib11], [@bib23]). Past studies have shown some of these factors and others (alcohol consumption, limited social support, gender inequality, witnessing violence as a child) to be associated with IPV ([@bib10], [@bib21]). We investigate the extent to which social network factors at the individual and household level are associated with individual attitudes accepting of IPV. We hypothesize that individuals from more cohesive households will be less likely to accept IPV; distinctly, we hypothesize that individuals in households in which a greater proportion of household members accept IPV will be more likely to accept IPV. We will also consider the interaction between these two: is household cohesion more strongly associated with IPV acceptance in households in which IPV acceptance is higher overall? Finally, given that IPV acceptance can differ according to education, gender, and marital status ([@bib31]), we consider the proportion of the household that is male, the proportion of the household that is married, and the mean level of household education separately, as possible factors associated with individual IPV attitudes accepting of IPV. Methods {#s0015} ======= Data collection {#s0020} --------------- In 2014, we collected full sociocentric network data from individuals aged 13+ in two villages in *La Unión, Lempira*, Honduras, as part of a larger ongoing study ([@bib34]). Sociocentric studies attempt to ascertain all of the social relationships within a defined population ([@bib24]). Although adolescents 13--17 years of age were surveyed, we eliminated their observations to maintain consistency with the Uganda sample. In each Honduran village, we took a complete census of all households, which included mapping each household in the village and enumerating all of the residents within them. We later returned to each household to gather data about individual health indicators, attitudes and beliefs, demographics, and social network connections. In total, our Honduras household census revealed a population of 1307 eligible individuals, and we were able to collect survey and network data on 837 (64%) individuals (691 adults after excluding adolescents). In each of eight villages within one parish in rural southwest Uganda, a data collection procedure similar to that used in Honduras was implemented based on a complete census of all adults aged 18 years and above. A total of 1669 people (out of 1747 residents or 96%) were interviewed in 2011 and 2012 across 716 households. In both Uganda and Honduras, households in which only one individual was surveyed were excluded from the analyses (Honduras N = 43, Uganda N = 172). Thus, the total sample size from Honduras was 691 and the total sample size from Uganda was 1392. Social ties {#s0025} ----------- A "name generator" is a question asked of a respondent to elicit important social connections ([@bib32]). In both Uganda and Honduras, participants were asked a series of questions to elicit the names of important social connections along a variety of domains. For this analysis, we were interested in ascertaining the networks that most effectively captured affective support. The challenge was that each dataset included different questions, specific to the population under consideration. Our previous research has suggested that the use of one name generator can bias the network, while too many can generate networks irrelevant to the question at hand ([@bib32]). To determine the name generators most appropriate for our analyses, we ran factor analyses on each dataset, looking for groupings of name generators that would form one coherent cluster of questions. Factor analyses identified coherent factors for name generators in both Honduras and Uganda. Name generators in Honduras measured emotional support by asking participants to whom they go to discuss important matters, and who they trust to discuss something personal and private: 1. What is the name of a person with whom you discuss important matters? 2. What is the name of a person that you could trust to talk about something personal or private? Respondents were told that answer choices could include friends, family, people you work with, people who work for you, neighbors, etc. For each name generator, respondents were asked to nominate up to five individuals. In Uganda, respondents were asked separately: With whom the respondent discusses financial matters, discusses health issues, and goes to for emotional support. Respondents were told to name adults who lived within the parish (in any of the eight villages) and that nominations could be repeated across questions. 1. Over the last 12 months, with whom in this parish have you usually talked about any kind of financial issues? This may include conversations about school fees, employment, giving, receiving, or paying loans, starting businesses, financing for big events, or other issues. 2. Over the last 12 months, with whom in this parish have you usually talked about any kind of health issues? This may include topics like your child's health, family planning, nutrition, HIV, mental health, immunizations, sanitation methods, alcohol abuse or other issues. 3. Over the last 12 months, whom in this parish have you gone to for emotional support? This may include talking about both positive and negative topics such as deaths, marriages, births, loss of job, or other topics of emotional importance for you. Questions that were not used for the analysis in Honduras included questions about who the person seeks help with a medical emergency, who they borrow and lend money to, and who they sit with at church. The three questions that were not used in Uganda were to whom they would give a honey stick, with whom they exchange food, and with whom they socialize. Measures {#s0030} -------- We used four questions from the DHS to assess the conditions under which a person believes that it is acceptable for a man to perpetrate physical violence against his wife or partner ([@bib35]). The questions asked: "*In your opinion, is a husband/companion justified in hitting or beating his wife/companion in the following situations:* (a) *If she leaves the house without telling him?* (b) *Neglects the children?* (c) *Argues with him?* (d) *Burns the food?* Answer choices were either yes or no. We coded a person as positive on IPV acceptance if they answered positively to any of the four questions. Cronbach's alpha on the full scale was 0.82 for the Honduras sample and 0.71 for the Uganda sample ([@bib31]). In this study, we calculated a measure of density for each household. Density is a measure that captures the cohesiveness of a network, and is calculated by dividing the total number of observed ties by the total number of possible ties. To do so, we identified all of the respondents within each household included in the study, and the nominated social ties amongst those individuals across all name generators. Higher density households had many individuals within the household who nominated other individuals within the household as social contacts; lower density households had few household members nominate other individuals within the household as social contacts. Household-level measures also included the proportion of people in the individual's household who accepted IPV, the proportion of people in the individual's household that were male, mean household education, the proportion of the people in the individual's household that were married or in a civil union, and the number of individuals living within each household. Household-level proportions were calculated to exclude the respondents' own values on each measure. Finally, we measured several individual-level factors including age, gender, education, income, religion, marital status, and, in Uganda, ethnic group. Our measure of respondents' education was a continuous measure based on 9 categories of schooling, including early primary, primary, 4 levels of secondary, tertiary, university and postgraduate. We measured respondents' income by asking "How much income did you personally earn from all economic activities in the past month (include farm, wage, and business work)?" Both income and education were included in the models as continuous variables. To maintain consistency across datasets, we created standardized mean-centered measures for proportion of household that reports accepting IPV, household density, individual's proportion of network ties that are same household, proportion of household that is male, proportion of household that is married, and household mean level of education, plus individual education and income. Mean centering was done at the level of the country dataset. As ethnic group was a relevant covariate for the Uganda dataset but was not measurable in Honduras, we included a "Honduras" ethnic group for all Honduras observations. All other measures were comparable across datasets and so left as they were. Finally, we combined the two datasets, adding in a separate measure for country. After removing observations with missing data, we had 829 respondents in Honduras and 1395 respondents in Uganda. Statistical analyses {#s0035} -------------------- We used logistic regression to estimate the relationship between individual and household characteristics and the probability of expressing acceptance of IPV, including village level fixed effects to account for village level clustering. Results {#s0040} ======= Summary statistics {#s0045} ------------------ [Table 1](#t0005){ref-type="table"} shows the summary statistics for both study populations. Average age was 34 years in Honduras and 38 years in Uganda. Slightly less than half of each population was male. The average number of people interviewed per household was 3.9 in Honduras and 3.6 in Uganda. In Uganda, 30% of respondents accepted IPV; in Honduras, 24% of respondents accepted IPV. Honduran households had a lower density (0.22 (SD 0.18)) than those in Uganda (0.26 (SD 0.27)).Table 1Descriptive statistics Uganda and Honduras.Table 1Uganda N = 1392Honduras N = 691Age mean (SD)37.05 (17.55)Age mean (SD)37.92 (15.23)Education mean 1--93.22 (1.7)Education mean 1--50.68 (0.72)Gender (male)47%Gender (male)47%HH Assets quintile mean 1--53.21 (1.37)HH income security mean 1--42.34 (0.79)Marital statusMarital status Married60% Married35% Civil unionNA Civil union41% Widowed8% Widowed2% Separated4% Separated3% Single28% Single19%ReligionReligion Protestant71% ProtestantNA Catholic24% Catholic75% EvangelicalNA Evangelical17% No religionNA No religion8% Other5% OtherNAEthnic group Banyankore92%NA Bakiga4%NA Baganda3%NA Other1%NAHousehold number3.65 (1.87)Household number3.86 (1.95)Respondent accepts IPV29%Respondent accepts IPV24%Household density0.26 (0.27)Household density0.23 (0.19)Mean proportion of ties same HH0.21 (0.18)Mean proportion of ties same HH0.48 (0.44)Mean proportion HH accept IPV0.29 (0.38)Mean proportion HH accept IPV0.23 (0.34)Mean proportion of HH married or civil union0.56 (0.38)Mean proportion of HH married or civil union0.62 (0.32)Mean proportion of HH male0.51 (0.37)Mean proportion of HH male0.49 (0.32) Individual characteristics {#s0050} -------------------------- [Table 2](#t0010){ref-type="table"} shows the association between demographic factors and IPV acceptance in the combined sample. Men were significantly less likely to accept IPV than women ([@bib15]). People of lower education and lower income were more likely to report IPV acceptance, while people who were older were less likely to report IPV acceptance. Although marital status, religion, ethnic group, and village were included in the model, they did not exhibit statistically significant associations with IPV acceptance at the p \< 0.05 level. To test country-level differences, we removed ethnic group and village from the model (as both contained categories exclusive to one or more country). Consistent with our descriptive statistics, we found that people in Uganda were significantly more likely to accept IPV than people in Honduras.Table 2Demographic predictors of IPV acceptance in Uganda and Honduras, combined multivariate models.Table 2Combined Uganda and HondurasOR95% CIPGender male0.60(0.48, 0.74)0.00Education0.77(0.68, 0.87)0.00Income0.82(0.74, 0.92)0.00Age0.99(0.98, 1.00)0.00[^1] Household level characteristics {#s0055} ------------------------------- For our main statistical analyses, we first tested the bivariate associations between our primary predictors and reported acceptability of IPV ([Table 3](#t0015){ref-type="table"}). We found that household density, proportion of household that accepts IPV, mean household education, number of people in the household, and proportion of household that are male were all associated with IPV acceptance at the bivariate level. Factors with a p value of less than 0.10 in the bivariate models were incorporated into a set of multivariate models.Table 3Bivariate associations between individual and household predictors and individual IPV acceptance.Table 3Combined Uganda and HondurasOR95% CIPProportion ties same HH0.84(0.76, 0.92)0.00Household density0.88(0.80, 0.98)0.01Number of HH members0.95(0.90, 1.00)0.04Proportion of HH that accepts IPV1.49(1.34, 1.64)0.00Proportion of HH that is male1.18(1.08, 1.30)0.00Proportion of HH that is married or in union1.06(0.96, 1.17)0.27Mean HH education0.80(0.72, 0.88)0.00 [Table 4](#t0020){ref-type="table"} shows the results of a multivariate analysis of the association between our predictors of interest and individual IPV acceptance. This model excluded proportion of the household which accepts IPV, which had the strongest association with IPV acceptance in the bivariate models and which, as a proxy measure for household level norms, could affect the relationship between our cohesion variables and reported acceptance of IPV by individual subjects. We found that household density and mean education of the household retained significance in the multivariate models. The higher the density of the household, the less likely it was that an individual respondent would report acceptance of IPV: for each 1 standard deviation increase in household density the chance of any individual accepting IPV decreased by 15% (95% CI 5--24%). [Fig. 1](#f0005){ref-type="fig"} depicts households in both Honduras and Uganda, with the larger darker colored red nodes in households with the highest density. [Fig. 2](#f0010){ref-type="fig"} is a bar graph that depicts the difference in mean household density stratified by those who accept IPV and those who do not.,Similar to the density results, individuals who lived in households with higher levels of education were also less likely to accept IPV regardless of their own educational level. In fully-adjusted Model 2, we found that the proportion of the household that reported acceptance of IPV was associated with individual IPV acceptance. With each one standard deviation increase in proportion of household that accept IPV, the probability of any individual accepting IPV increases by 46% (95% CI 32%-51%). This model also showed that the associations between individual IPV acceptance and mean household education and household density, separately, were slightly attenuated after including proportion of household accepting IPV.Fig. 1Depicts a random set of households in Honduras (left), and a random set of households in Uganda (right). Nodes (e.g. circles) represent individuals, lines represent relationships between the individuals within the same household, hence each cluster of nodes is a distinct household. Circular nodes do not accept IPV while square nodes do. Node color and node size are proportionate to household density: the large darkest red nodes are in high-density households while the small white nodes are in the low-density households. Note square nodes tend to be white, illustrating lower household cohesion for those individuals. Note also how the majority of those who accept IPV live in households in which at least one other household member also accepts IPV.Fig. 1Fig. 2Illustrates the difference in mean household density for individuals who accept IPV versus those who do not. In both Honduras and Uganda, it is clear that household density is negatively associated with IPV acceptance.Fig. 2Table 4Multivariate logistic regression showing the association between alter's ipv acceptance and ego\'s ipv acceptance, combined Uganda and Honduras dataset.Table 4Model 1Model 2OR95% CIPOR95% CIPProportion of HH that accepts IPV1.46(1.32, 1.61)0.00Proportion ties same HHHousehold density0.85(0.76, 0.95)0.010.88(0.79, .99)0.03Number of HH members0.98(0.92, 1.04)0.550.98(0.92, 1.05)0.63Proportion of HH that is male1.04(0.92, 1.18)0.491.12(0.99, 1.27)0.08Mean HH education0.89(0.79, 1.00)0.000.92(0.81, 1.04)0.17Sex0.62(0.47, 0.80)0.000.62(0.47, 081)0.00Education0.80(0.71, 0.91)0.000.82(0.72, 0.93)0.00Income0.85(0.76, 0.96)0.010.86(0.77, 0.97)0.01Age0.99(0.98, 1.00)0.000.99(0.98, 1.00)0.00AIC23852331[^2] As an exploratory analysis, for each individual, we also calculated the proportion of that individual's nominated ties who lived in the same household, as an alternative measure to household density, which is a household-level measure. [Table 5](#t0025){ref-type="table"} shows the results of the same set of analyses but using the proportion of alters that are same household instead of household density in the models. The results are similar.Table 5Multivariate logistic regression showing the association between alter's ipv acceptance and ego's ipv acceptance, combined Uganda and Honduras dataset, testing proportion of household that accepts IPV.Table 5Model 1Model 2OR95% CIPOR95% CIPProportion of HH that accepts IPV1.471.33 1.620.00Proportion ties same HH0.85(0.76, 0.95)0.000.860.77 0.970.01Household densityNumber of HH members1.03(0.97, 1.10)0.301.030.97 1.090.39Proportion of HH that is male1.04(0.92, 1.17)0.531.120.99 1.270.09Mean HH education0.89(0.79, 1.00)0.050.920.82 1.040.19Sex0.60(0.46, 0.78)0.000.610.46 0.790.00Education0.80(0.71, 0.91)0.000.820.72 0.930.00Income0.85(0.76, 0.96)0.010.860.77 0.970.01Age0.99(0.98, 1.00)0.000.990.98 1.000.01AIC23852330 Cohesive social networks can reinforce norms (whether protective or harmful), therefore it is likely that, while cohesion in general may decrease the likelihood that an individual accepts IPV, as we found in our results, this may change according to the overall level of IPV acceptance within the household. To understand this more deeply, we stratified our analysis to look at the association of household cohesion on IPV acceptance for those in households above or below the median proportion of household IPV acceptance within the sample. [Table 6](#t0030){ref-type="table"} shows the results of this analysis using household density as the primary predictor. In Model 1, for households that were below or equal to the median proportion that accept IPV, household density was significantly protective against individual IPV acceptance, as was living in a household with more members, and a household with a higher level of education. At higher levels of IPV acceptance (Model 2), however, we see that those associations are not observed. Statistical interaction tests (not shown) suggest that this difference is more pronounced in Honduras then in Uganda. [Fig. 3](#f0015){ref-type="fig"} shows the relationship between household density and the probability that an individual accepts IPV for those in households above the median level of IPV acceptance, and for those below. There appears to be no relationship between household density and IPV acceptance for those in households with higher acceptance of IPV, while there is a clear relationship for those in households with lower acceptance. Further research is warranted to investigate this dynamic.Fig. 3Shows the relationship between household density and IPV acceptance, stratified by those individuals in households with high acceptance of IPV compared to those in household with low acceptance of IPV.Fig. 3Table 6Multivariate logistic regression showing the association between alter's ipv acceptance and ego's ipv acceptance, combined Uganda and Honduras dataset, stratified by household proportion of household that accepts IPV.Table 6Model 1 Lower proportion accepts IPVModel 2 Higher proportion accepts IPVOR95% CIPOR95% CIPHousehold density0.830.720.960.010.970.801.170.76Number of HH members0.850.750.960.010.970.891.050.43Proportion of HH that is male1.160.971.380.111.040.851.260.72Mean HH education0.750.630.880.001.241.021.510.03Sex0.650.430.990.040.590.410.840.00Education0.770.630.930.010.880.731.060.18Income0.870.731.040.130.860.731.020.08Age0.990.981.000.040.990.981.000.03 To further understand these results, we investigated the household level factors associated with household density. We first calculated the mean household density for those households above the median proportion of household acceptance of IPV and for those below it. Households below the median proportion of household IPV acceptance had a mean-centered household density of 0.14 (95% CI 0.06--0.21) while households above had a mean-centered density of − 0.003 (95% CI − 0.06 to 0.05). [Table 7](#t0035){ref-type="table"} shows the results from a household-level linear regression estimating the relationship between household density and proportion of household that accepts IPV, proportion of household that is male, proportion of household that is married, and mean household education along with tribe and village. Here we see that household density is significantly and inversely related to the proportion of household that accepts IPV at the household level, inversely related to the proportion of household that is male, positively related to proportion of household that is married, and positively associated with mean household education level.Table 7Linear regressions analysis of household level predictors on household density.Table 7Combined Uganda and HondurasBetaSePProportion of HH that accepts IPV− 0.080.020.00Number in household− 0.170.010.00Proportion of HH that is male− 0.040.020.04Proportion of HH that is married0.140.020.00Mean HH education0.090.020.00 We tested interactions by country (not shown) and found no significant differences between the association of household density and IPV acceptance by country but found a significant association with the proportion of household that accepts IPV and IPV acceptance. While proportion of household that accepts IPV was strongly associated with IPV acceptance in both countries, it was more strongly associated in Honduras compared to Uganda. Discussion {#s0060} ========== We describe the relationship between household cohesion, as measured through social ties within household members, and individual IPV acceptance in Honduras and Uganda. The more closely connected people are within a household, the less likely it is for an individual in that household to accept IPV, controlling for the overall acceptance of IPV within the household. Those in households more accepting of IPV were more likely to personally accept IPV. In stratified analyses, when household IPV acceptance was especially high, the benefit of household cohesion with respect to IPV was potentially attenuated, although further research is needed to investigate that preliminary evidence. The household cohesion results provide possible evidence of the protective mechanism of strong family ties. More cohesive families might be more likely to enforce any existing familial norms against IPV in which case we would see a strong positive relationship between density and household acceptance of IPV in the households with higher overall levels of IPV acceptance. Alternately, families that are more cohesive may have better communication, more loving relationships, and an overall safer environment than less cohesive families ([@bib9]). Given previous work on the positive relationship between density and the maintenance of social norms ([@bib30], [@bib13]) it seems less likely that the relationship between cohesion and acceptance of IPV is simply one of social control, as the social control hypothesis would predict a positive association between density and IPV acceptance in household with high acceptance of IPV. Densely connected networks of any sort are likely to reinforce positive or negative norms ([@bib30], [@bib13]). In this study, however, there was a lower likelihood that a person would accept IPV as the density of their household increased though this association potentially depended on the prevalence of IPV acceptance within the household. Notably, the protective association was possibly attenuated in households with high acceptance of IPV and the inclusion of household acceptance of IPV in the model diminished the significance of household density as a predictor. This dynamic is suggestive of counter-balancing mechanism by which the positive impact of family cohesion is in essence neutralized by the negative impact of high levels of acceptance of IPV. Also, our final analysis found that households with higher reported acceptance of IPV were associated with lower overall density. The fact that households with higher levels of IPV acceptance tend to also be those that are less densely connected suggests that density may reflect positive relationships within the family, and therein directly protect against approval of IPV through protective norms, rather than density simply reinforcing norms. Alternatively, the causal order could be reversed, in that higher prior levels of IPV among two members of a family may lead to lower density throughout the household. Finally, some other unmeasured factor in the household might have contributed to both IPV and decreased social ties. Although household density was negatively associated with individual acceptance of IPV, the strongest predictor of individual acceptance was household acceptance of IPV. This finding is consistent with previous social network research in which the attitudes and behaviors of socially connected others are similar. In this case we are restricting those relationships to people who live within the same household. Our previous research in Honduras demonstrated that same household ties were the strongest predictor of shared attitudes towards IPV acceptance, suggesting that household members may be an important reference group for norms around IPV ([@bib31]). In this study, we find that household level factors are important in Uganda as well, although the magnitude of the effect was slightly smaller. This may be the because overall IPV acceptance is higher overall in Uganda than Honduras, making it slightly less sensitive to household-level factors. It is important to note that, in this study, we did not have direct measures of either descriptive norms or injunctive norms, but in fact are inferring possible normative dynamics through measured attitudes. The fact that higher levels of IPV acceptance within a household is associated with individual IPV acceptance is a clue towards descriptive norms -- the perception of something as prevalent and acceptable within a reference group increases the likelihood that any individual within that group will follow suit. That household density did not seem to reinforce higher levels of household IPV acceptance is a clue that acceptance of IPV is not necessarily normatively reinforced. However, the negative relationship between household density and individual IPV acceptance at lower levels of household IPV acceptance may suggest that when IPV is generally not acceptable, there is normative reinforcement to *not* accept IPV. Further research, with more robust measurement of social norms, is necessary to more clearly identify these dynamics. Our study has limitations. First, the data are restricted to two villages in Honduras, and eight villages in Uganda. Thus, results might not be generalizable to other contexts. However, the consistency of our results across two international contexts suggests that interpretation of this information may be applicable outside of these specific settings. Second, analyses from both countries are cross-sectional. Therefore, we cannot infer causality, and the time-dependent process by which we could track changes in IPV attitudes with a change in household density is not possible. Finally, reports of IPV attitudes are self-report and can be subject to response bias, although we have used the standard DHS measures and so our results should be comparable to previous research conducted in this area. Moreover, attitudes are less prone to bias than behaviors. Finally, because of resource constraints, we were only able to survey 64% of the population in the Honduras sample. Conclusion {#s0065} ========== Here, we demonstrate the association between household characteristics and individual acceptance of IPV in 2 distinct geographic and cultural contexts. We find that not only does the proportion of an individual's household that accepts IPV increase the likelihood that the individual accepts IPV, but individuals in more cohesive households are less likely to accept IPV. Future research on this topic should include measures of IPV perpetration, and longitudinal data so that changes in these dynamics can be tracked across time. Public health implications {#s0070} ========================== Prevention of IPV is an important public health priority. While the best methods for IPV prevention are still being explored ([@bib6]), some prior work has suggested that promotion of gender equality can reduce rates of IPV ([@bib25], [@bib1], [@bib37]). While this may be the case, it is important to take into consideration the target audience for these sorts of interventions. Our results provide evidence to suggest that gender equity and other preventive interventions may work best at the household level, where norms and attitudes on IPV are often enforced and reinforced. Furthermore, such intervention efforts may be most successful when they focus on fostering strong, healthy relationships within families, while promoting familial norms that serve to prevent violence against women in those families. Conflict of interests {#s0075} ===================== Authors have no conflict of interests to disclose. IRB approval for this study was provided by the University of California San Diego project number 151733 and Yale University project number 1506016012. IRB approvals for this study were also obtained from Harvard University, Partners, Mbarara University of Science and Technology, and Uganda National Council of Science and Technology. Work was funded by the Bill and Melinda Gates Foundation Grant OPP1098684, Friends of a Healthy Uganda and by a Roybal Center grant through U.S. National Institutes of Health (NIH) P30AG034420NICHDK01HD087551‐01. The authors also acknowledge salary support from NIH K23MH096620 and NICHD K01HD087551-01. [^1]: Models also included marital status, religion, tribe, and village (not shown). [^2]: Models include marital status, religion, tribe and village (not shown).

Introduction {#Sec1} ============ Enteric duplication cysts (EDCs) are rare congenital anomalies found anywhere along the gastrointestinal tract (GT) from the mouth to the rectum; most commonly in the ileum (33%), followed by the oesophagus (20%), colon (13%), jejunum (10%), stomach (7%) and duodenum (5%) \[[@CR1]--[@CR4]\]. The incidence is 1:4,500 births, found in 0.2% of all children, with a slight male predominance \[[@CR3], [@CR5]--[@CR7]\]. EDCs are believed to occur between the 4th and 8th weeks of embryonic development. Their aetiology is still unknown; several theories have been proposed to explain their pathophysiology, but no single hypothesis can justify all duplications, locations and associated anomalies. Split notochord theory is often postulated \[[@CR8]\]. The luminal recanalisation theory explains duplications in those portions of the GT that have a solid stage, including the oesophagus, small bowel and colon; nevertheless, it does not explain duplications at other levels. Incomplete or partial twinning theory could explain the colorectal duplications that are associated with duplication of genital and urinary structures. Persistent embryonic diverticula theory suggests that small diverticula, usually transient along the antimesenteric border of the intestinal wall, persist and develop intestinal duplications, although most ECDs are in the mesenteric border. The intrauterine vascular accident theory suggests that gastrointestinal duplications arise from an intrauterine vascular accident during early fetal development and may be a valid explanation for isolated duplication. These different theories lead to think that the origin of EDCs can be multifactorial \[[@CR1], [@CR2], [@CR4], [@CR9], [@CR10]\]. Associated anomalies such as spinal defects, cardiac or urinary malformations, are reported with an incidence rate of 16--26%. Other digestive anomalies are present in about 10% of cases. Therefore, once an EDC is found, a search for other anomalies is needed \[[@CR6], [@CR10]--[@CR12]\]. EDCs must have three characteristics: an epithelial lining containing the mucosa of the alimentary tract, an envelope of smooth muscle, and the cyst must be closely attached to the GT by sharing a common wall (Fig. [1](#Fig1){ref-type="fig"}). The mucosal lining does not always correlate with the adjacent gastrointestinal tissue, but the duplications are named according to the part of the GT to which these are intimately attached. Ectopic gastric mucosa is found in 20--30% of these cysts, more frequently in oesophageal and small bowel duplications \[[@CR13], [@CR14]\]. Prominent gastric mucosa can also be seen as a polypoid mass covering the base of the cyst, being transmural (Fig. [2](#Fig2){ref-type="fig"}) \[[@CR6]\]. Ectopic pancreatic mucosa is most common in gastric duplications \[[@CR2]\].Fig. 1An 11-month-old boy with abdominal pain is studied. **a** US view showing the typical US features of an EDC: an inner hyperechoic epithelial lining containing the mucosa of the alimentary tract (*wide arrow*) and the outer hypoechoic layer of smooth muscle (*white long arrow*), closely attached to the gastrointestinal tract by sharing a common wall. *RK* right kidney. **b** Surgical findings: typical ileal EDC. **c** Detailed picture of the EDC after resection from the ileal wallFig. 2A 4-year-old boy in a routinary US control of a horseshoe kidney. **a** Abdominal ultrasound view of an EDC (*black arrow*) with a peripheral eccentric hypoechoic cap (*white arrows*). **b** Surgical findings: the gastric mucosa was visible as a polypoid mass (*white arrows*) arising from the external surface of the EDC (*black arrow*) Structurally, EDCs can be either cystic or tubular. Spherical cysts are the most common duplications (80%) and typically do not communicate with the adjacent lumen. Tubular duplication cysts (20%) run parallel to the GT, being communicated with it (Fig. [3](#Fig3){ref-type="fig"}) \[[@CR4], [@CR12], [@CR15], [@CR16]\]. Then, when a duplication cyst is tubular, the connection with GT must be demonstrated for surgical planning \[[@CR17]\].Fig. 3Types of duplication cysts seen in the abdominal US. **a** Spherical EDC in gastric antrum (*arrows*). *S* stomach. **b** Tubular EDC (*arrows*) next to the descending colon (*DC*) Multiple duplication cysts are rather uncommon (1--7%) \[[@CR4], [@CR16]\]. These include multiple EDCs within one segment of the GT or less frequently in two or more segments (Fig. [4](#Fig4){ref-type="fig"}) \[[@CR11], [@CR16], [@CR18], [@CR19]\].Fig. 4A 2-year-old girl with a splenic lesion (not shown) is studied with abdominal MR. **a** FSE T2 sagittal MRI: two similar cystic lesions are found (*arrows*). *S* stomach, *LK* left kidney, *SP* spleen. **b, c** FSE T2 axial MRI: a gastric and a jejunal duplication cyst are shown (*arrows*) in both images. Surgical findings: multiple EDCs Atypical EDC is a non-communicating isolated duplication cyst completely separated from the bowel with no communication or shared wall. A vascular insult could have led to the isolation. They are extremely rare \[[@CR19], [@CR20]\], especially multiple isolated EDCs, which are even rarer \[[@CR19]\]. The size, location, type, mucosal pattern and presence of complications produce different clinical presentations and several imaging findings of the EDCs. Ultrasonography (US) is the most used imaging method for diagnosis of abdominal EDCs. Magnetic resonance (MR) and computed tomography (CT) are utilised for oesophageal EDCs and for helping in difficult surgical approaches. We review the different forms of presentation of EDCs, showing both typical and atypical image findings with the different imaging techniques. We correlate the imaging findings with the surgical results and the final pathological features. Clinical presentations {#Sec2} ====================== The intrauterine presentation is increasing, mostly due to the improvement in prenatal screening US, routine second-trimester screening and improved imaging resolution. However, prenatal diagnosis of EDCs is often difficult, and US identifies only 20--30% of them, and sometimes they are discovered by chance (Fig. [5](#Fig5){ref-type="fig"}) \[[@CR21], [@CR22]\].Fig. 5A 21-week-old fetus with polyhydramnios and absent normal gastric bubble in the US is studied. **a** Sagittal FIESTA fetal MRI showing a mediastinal cyst (thick arrow). Oesophageal atresia without fistula is suspected. *B* bladder. Trachea (*arrows*). **b** Coronal HASTE fetal MRI: detailed view of the mediastinal cyst (*thick arrow*). **c** Postnatal thoraco-abdominal radiograph: the gastric line tip is seen (*black arrow*) confirming the oesophageal atresia. The absence of air in the abdomen indicates a type-I or -II oesophageal atresia (without fistula). Venous umbilical catheter (*white arrow*) The natural history of EDCs is quite variable. The clinical presentation or onset symptoms of these malformations range from infancy and early childhood to adulthood. Almost 70% of EDCs present symptoms within the first year of life and 85% in the second one \[[@CR3], [@CR10], [@CR14]\]. The signs and symptoms depend on the type and location of the duplication. Oral and oesophageal cysts may cause respiratory distress or dysphagia. Retrosternal pain, haemoptysis and infection can occur in case of large cysts with rapid growth (Fig. [6](#Fig6){ref-type="fig"}) \[[@CR6], [@CR23]\].Fig. 6A 10-month-old boy with a congenital cardiopathy presents respiratory distress. **a** Chest X-ray: a left cervicothoracic mass is suspected displacing the trachea to the right (*arrows*). **b** Chest US: cystic mass (*M*) with slightly echogenic content inside is seen next to the thymus (*T*). Its origin is unclear. **c** Coronal view SSFSE T2 MRI: a well-delineated and hyperintense lesion (*star*) is seen. **d** Transversal view of the lesion in a gadolinium-enhanced VIBE MRI confirms the cystic nature of the mass (*star*) next to the anterior oesophageal wall (*arrow*). **e** Thoracoscopic findings: a 6 × 4-cm lesion with close contact with the trachea originated from the muscular wall of the oesophagus Gastric and intestinal duplications may produce nausea, vomiting, abdominal distention or palpable abdominal mass (Fig. [7](#Fig7){ref-type="fig"}). Recurrent abdominal pain is one of the most frequent forms of presentation and is usually attributed to high pressure inside the duplication cyst because of the accumulation of secretions. Intussusception is another complication in which the cyst serves as a head point and pain, obstruction or bleeding are possible forms of manifestation. Extrinsic compression of the adjacent bowel is also possible, which causes obstruction. However, the most serious complications are produced if gastric mucosa is present within the cyst, leading to inflammation, bleeding, ulceration and even perforation \[[@CR6], [@CR9], [@CR13], [@CR14], [@CR24], [@CR25]\].Fig. 7An 18-month-old girl with an abdominal mass in the physical exam is studied. **a** Abdominal X-ray: round, dense mass is discovered (*arrows*) in the left upper quadrant. **b** Transversal US view of the lesion: a cyst with the "double-wall" sign: the mucosa is hyperechoic (*arrow*) and the muscular layer is hypoechoic (*dashed-line arrow*). A gastric duplication cyst was suspected. *L* liver, *P* pancreas, *AO* aorta. **c** Laparotomy: antral duplication cyst was found (*arrow*). *S* stomach Nonetheless, EDCs can sometimes be detected incidentally (Fig. [8](#Fig8){ref-type="fig"}).Fig. 8A 12-year-old girl with a gynaecological malformation and haematometrocolpos. **a** A hypointense lesion is seen (*white arrow*) next to the rectum in axial T1 weighted pelvic MR. **b** On T2-weighted pelvic MR, the cyst next to the left wall of the rectum is seen. **c** Sagittal plane of the T2 MRI showing the posterior location of the cyst (*arrow*). *VG* vagina, *BL* bladder, *U* uterus. Imaging findings of a rectal duplication cyst. Note the haemorrhagic content of the uterus and vagina because of haematometrocolpos Imaging findings {#Sec3} ================ US is the imaging method of choice in the diagnosis of EDCs; only limited in the evaluation of oesophageal EDCs. Transoesophageal ultrasound might be useful but is not routinely used \[[@CR26], [@CR27]\]. Classical findings of uncomplicated EDC include: presence of a cyst in relation to the gut with double-wall or muscular rim sign (gut signature sign), which is caused by inner hyperechoic mucosa and outer hypoechoic smooth muscle layer (muscularis propia) (Figs. [1](#Fig1){ref-type="fig"} and [7](#Fig7){ref-type="fig"}). However, the double-wall sign in other cystic lesions (mesenteric cyst, Meckel's diverticulum or torsioned ovarian cyst) may be seen \[[@CR1], [@CR2], [@CR11], [@CR24], [@CR28]--[@CR30]\]. New US signs are described according to the characteristics of the EDCs:As an EDC contains the same multi-layered wall architecture as the normal GT, the sign "five-layered cyst wall" is proposed. It corresponds to the innermost hyperechoic mucosa, hypoechoic muscularis mucosa, hyperechoic submucosa, hypoechoic muscularis propia and the outermost hyperechoic serosa. Identification of all five layers in a cyst is pathognomonic of EDC. However, this sign is difficult to demonstrate and needs expertise and high-resolution US (12--18 MHz) \[[@CR27], [@CR29], [@CR30]\]. For this reason, the use of US linear probe is recommendable when the GT is examined.An EDC shares wall with the adjacent GT. Therefore, the diagnosis is carried out if it is possible to demonstrate the "Y" sonographic configuration of the muscle layer caused by the splitting of the shared muscularis propria between the cyst and the adjacent loop. This sign is not described for other abdominal cysts and reflects one of the histological characteristics of the EDCs (Fig. [9](#Fig9){ref-type="fig"}) \[[@CR29], [@CR31]\].Fig. 9A 3-month-old boy with vomiting is admitted to the emergency room. **a** US shows a cystic round-shape lesion with the "five-layers sign" (between *arrows*). **b** The "Y sign" is seen (*long arrow*). *Star* ileum, *L* liver. Laparoscopic findings: a non-complicated ileal duplication cyst US is a dynamic study and allows to visualise the peristalsis of the cyst wall. It appears as a transient change of the shape and contour of the cyst because of a concentric contraction of the cyst wall (Fig. [10](#Fig10){ref-type="fig"}) when the transductor stays still on the cyst for a while \[[@CR2], [@CR27]\].Fig. 10An 8-month-old boy with abdominal pain. **a** Longitudinal grey-scale US image showing a cystic lesion with an incomplete septum inside (*small white arrows*), next to the terminal ileum (*big arrow*). The "Y" sign is shown (*dashed-line arrow*). **b** US image obtained a few seconds later: peristalsis of the cyst causes small angulation of the contour and changes shape (*black star*). **c** Surgical findings: ileal duplication cyst Although EDCs are frequently anechoic or hypoechoic, mucinous material or septations can be present without being complicated (Fig. [11](#Fig11){ref-type="fig"}) \[[@CR27]\].Fig. 11An 8-month-old boy with abdominal pain and abdominal mass on physical exam. **a** Longitudinal US view of a multiseptated cystic mass in the right flank with the "Y" sign (*white arrows*). *I* ileum, *L* liver, *Ps* psoas, *RK* right kidney. **b** Transversal US of the same mass demonstrates the relation with the ileal walls. **c** Surgical findings: ileal duplication cyst Complicated EDCs rarely present the classic five layers or double-wall sign. Ectopic rest of pancreatic tissue can produce enzymatic destruction of the mucosa with inflammation, as well as loss of the wall layers showing a hyperaemic thick wall. In such cases, the "Y configuration" sign helps in establishing the correct diagnosis of EDC (Fig. [12](#Fig12){ref-type="fig"}) \[[@CR3], [@CR29], [@CR30]\]. If ectopic gastric mucosa produces haemorrhage and bleeding, fluid levels or echogenic debris can be seen. When infection occurs, ulceration of mucosa can appear, and internal debris may be seen. The transmural extension can produce important inflammatory changes in the surrounding mesentery fat (Figs. [13](#Fig13){ref-type="fig"} and [14](#Fig14){ref-type="fig"}) \[[@CR3], [@CR7], [@CR24]\]. Ileal EDC, near the ileocecal valve, can act as intussusception head, showing on US a cystic mass inside the intussusception requiring emergency treatment (Fig. [15](#Fig15){ref-type="fig"}) \[[@CR2]\].Fig. 12A 3-week-old term newborn with abdominal distention and gastric intolerance. **a** Longitudinal US view of the right low quadrant: thickened wall (*stars*) cystic lesion next to a bowel loop (*dashed-line arrow*). The "Y" sign between the bowel and the lesion (*arrow*). **b** Power Doppler US demonstrates the significant vascularisation in the cyst wall. **c** Surgical findings: a cystic tumour next to the ileocecal valve. Pathological findings: ileal duplication cyst with heterotopic pancreatic tissueFig. 13A 3-year-old boy with fever and abdominal pain is studied. **a** US shows a cystic mass (*star*) with internal debris and next to an ileal loop (*L*). **b** The lesion (*star*) is surrounded by echogenic mesenteric fat (\*) as an inflammatory sign. Surgical findings: a 5-cm ileal complicated duplication cyst was found with gastric mucosa with haemorrhagic and ulcerated wallsFig. 14An 8-month-old boy baby with continuous crying is taken to the emergency department. Because of high suspicion of intussusception, a US exam is required. **a** Abdominal pear-shaped cystic lesion (*star*) in the left flank was found in a coronal US view. It shows a typical outer hypoechoic wall and inner hyperechoic layer, with hypoechoic content inside. **b** Doppler US showing light hyperaemia of the lesion wall that was also thickened. Surgical findings: complicated duplication cyst in the colonic splenic flexureFig. 15A 14-month-old girl with acute gastroenteritis and continuous crying. **a** Abdominal US shows an intestinal intussusception with a cyst (*C*) as the leading cause. *L* liver, *RK* right kidney. **b** A detailed US view: the intussuscipiens (*arrows*) and intussuscepted bowel (*arrowheads*) with the cyst inside (*C*) and the hyperechoic and thickened walls. Surgical findings: ileal duplication cyst as the cause of the intussusception In case of atypical or isolated EDC, the pseudokidney sign is described when there is a complete loss of typical wall layers because of severe congestion, thus producing a thick hypoechoic rim with a hyperechoic central layer \[[@CR32]\]. US prenatal diagnosis of EDCs includes the same signs as postnatal cyst: the double--wall sign and the presence of peristalsis. However, on the prenatal US, the "double wall" is not always seen or can be partial \[[@CR10], [@CR33], [@CR34]\], and it may require the differential diagnosis with other cystic lesions such as mesenteric, omental, ovarian and choledochal cysts. If it is possible to demonstrate the presence of peristalsis in the cyst wall, an intestinal origin is probed. MR imaging is suggested to have a supplemental value in the assessment of fetal abdominal cysts (Fig. [16](#Fig16){ref-type="fig"}) \[[@CR10], [@CR35], [@CR36]\].Fig. 16A 22-week-old fetus with an abdominal cyst seen on ultrasound is studied with MR. **a** Coronal fetal FIESTA MR: a cystic lesion is seen next to the stomach (*S*). Oesophageal lumen is seen (*arrows*). **b** FIESTA transversal MR: the hypointense wall of the suspected gastric duplication cyst (*arrow*). *L* liver, *S* stomach. **c** Postnatal abdominal US view: the lesion (*star*) imprinting the gastric wall (*arrow*). The content of the stomach is seen (*S*). *SP* spleen. Surgical findings: duplication cyst of the oesophageal-gastric transition CT is not typically performed for evaluation of EDCs due to radiation. CT may depict the location and extension of the cyst, as well as complications, the associated anomalies and anatomical relationship with surrounding structures. At CT, an EDC manifests as a cystic mass with a thin and slightly enhancing wall adjacent to the gastrointestinal wall. A high attenuation inside the cyst may be seen due to haemorrhage or proteinaceous material. A thick enhancing wall, air bubbles inside and cyst-surrounding inflammation may indicate an EDC complicated by infection (Fig. [17](#Fig17){ref-type="fig"}) \[[@CR1], [@CR11], [@CR12], [@CR15]\].Fig. 17A 22-month-old boy with fever and abdominal pain. **a** Transversal US view of the pelvis shows a cystic mass between the bladder (*BL*) and the sacrum (*S*). The lesion presents anterior and left wall thickening (*arrow*) and contains a fluid-fluid level suggesting the presence of a complicated rectal mass. **b** Contrast-enhanced CT confirmed the presence of a complicated rectal cyst with remarkable inflammatory findings (*white arrow*) and gas (*dashed-line arrow*). Surgical findings: rectal duplication cyst with mucous and purulent content. Rectum muscular wall was oedematous and it was shared with the duplication cyst Like CT, MR is not routinely used as a diagnostic method for EDCs, especially due to sedation requirement. On MR imaging, most duplications have low signal intensity on T1-weighed images and very high intensity on T2-weighted images (Fig. [18](#Fig18){ref-type="fig"}). Both CT and MR play a major role prior to surgery in establishing the relationship between the cyst and its adjacent structures \[[@CR12]\], and in locations where US presents a limited use, particularly in oesophageal and rectal duplications \[[@CR1], [@CR2], [@CR6], [@CR12]\].Fig. 18A 2-year-old girl with fever and abdominal pain. **a** Abdominal US view of a complex cystic lesion (*white arrows*) in the duodenal area, next to the liver (*L*). *GB* gallbladder. **b** Axial gadolinium-enhanced GRE T1 demonstrating the presence of a multilocular cystic mass (*white arrows*) with wall thickening. **c** Axial T2 MRI: the mass (*white arrow*) shows three cystic cavities inside, close to the anterior wall of the duodenum (*dashed-line arrow*) and lateral to the gallbladder (*star*). Surgical findings: multiple duodenal duplication cyst, without connection with the lumen and with ectopic gastric mucosa Management {#Sec4} ========== The main considerations in the management of EDCs are: the condition of the patient, the location of the cyst, whether it involves one or more anatomic locations, whether its structure is cystic or tubular, and if it is communicated with the true intestinal lumen. With the widespread availability of antenatal diagnosis, EDCs are often diagnosed prenatally. The optimal time to perform the resection in children with antenatal diagnosis is not defined. These patients should undergo early investigation, followed by early resection even within the first 6 months of life \[[@CR3], [@CR37], [@CR38]\]. Treatment of asymptomatic EDCs remains controversial. The clinical behaviour of EDCs is unpredictable. EDCs tend to increase in size gradually and can cause symptoms and important complications that might be fatal, such as obstruction, massive bleeding or even a potential risk for malignant transformation in the adulthood \[[@CR13], [@CR14], [@CR17], [@CR39]\]. Early excision is associated with less morbidity and a shorter length of stay compared to excision in symptomatic patients. There are significant post-operative morbidities after resection of complicated EDCs, compared with its elective surgery. Cyst excision alone could be considered, but if there is a communication, sometimes a resection of the adjacent bowel is necessary. It is important to ensure that the cyst is entirely resected because recurrence or malignant changes may occur \[[@CR40]\]. Currently, minimally invasive surgery is becoming the elective approach, and most of the cysts can be resected successfully, either thoracoscopically or laparoscopically, as long as an exhaustive imaging diagnosis is available \[[@CR41]\]. Conclusions {#Sec5} =========== EDCs are uncommon congenital abnormalities arising anywhere along the GT. Their clinical presentations vary according to the site of duplication; ileum appears as the most commonly involved. Nowadays, antenatal diagnosis is becoming more frequent. US is the method of choice to diagnose gastrointestinal EDCs. Although double-wall US sign in a cyst is the most typical for diagnosis of EDCs, the findings of the five layers sign or the "Y configuration" of the muscular layer are more specific features. Complicated cysts present atypical imaging findings. CT and MR imaging can be required in oesophageal or rectal EDCs for planning complicated surgical approach. Surgery is necessary because of the severe complications they can develop. The diagnosis is confirmed by histological examination. EDC : Enteric duplication cyst GT : Gastrointestinal tract **Publisher's Note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

1.. Introduction {#sec1} ================ In an effort to extend the structural coverage of proteins for which the biological function is unknown and cannot be deduced by homology (domains of unknown function; DUFs), targets were selected from Pfam protein family PF09410 (DUF2006). Here, we report the crystal structure of NE1406, the first structural representation of this family, which was determined using the semiautomated high-throughput pipeline of the Joint Center for Structural Genomics (JCSG; Lesley *et al.*, 2002[@bb22]) as part of the NIGMS Protein Structure Initiative (PSI). The *NE1406* gene of *Nitrosomonas europaea*, an obligate chemolithoautotroph, encodes a protein with a molecular weight of 40.1 kDa (residues 1--356) and a calculated isoelectric point of 5.0. 2.. Materials and methods {#sec2} ========================= 2.1.. Protein production and crystallization {#sec2.1} -------------------------------------------- Clones were generated using the polymerase incomplete primer extension (PIPE) cloning method (Klock *et al.*, 2008[@bb18]). The gene encoding NE1406 (GenBank NP_841447, gi\|30249377, Swiss-Prot Q82US3) was amplified by polymerase chain reaction (PCR) from *N. europaea* strain ATCC 19718 genomic DNA using *PfuTurbo* DNA polymerase (Stratagene) and I-PIPE (Insert) primers (forward primer 5′-ctgtacttccagggcATGCGTTACTTATGGATACTGTTG-3′, reverse primer 5′-aattaagtcgcgttaCATCGATAACGGACGTACG-3′; target sequence in upper case) that included sequences for the predicted 5′ and 3′ ends. The expression vector pSpeedET, which encodes an amino-terminal tobacco etch virus (TEV) protease-cleavable expression and purification tag (MGSDKIHHHHHHENLYFQ/G), was PCR-amplified with V-PIPE (Vector) primers. V-PIPE and I-PIPE PCR products were mixed to anneal the amplified DNA fragments together. *Escherichia coli* GeneHogs (Invitrogen) com­petent cells were transformed with the V-PIPE/I-PIPE mixture and dispensed onto selective LB-agar plates. The cloning junctions were confirmed by DNA sequencing. Using the PIPE method, the part of the gene encoding residues Met1--Pro22 was deleted. Expression was performed in a selenomethionine-containing medium with suppression of normal methionine synthesis. At the end of fermentation, lysozyme was added to the culture to a final concentration of 250 µg ml^−1^ and the cells were harvested and frozen. After one freeze--thaw cycle, the cells were sonicated in lysis buffer \[50 m*M* HEPES pH 8.0, 50 m*M* NaCl, 10 m*M* imidazole, 1 m*M* tris(2-­carboxyethyl)phosphine--HCl (TCEP)\] and the lysate was clarified by centrifugation at 32 500*g* for 30 min. The soluble fraction was passed over nickel-chelating resin (GE Healthcare) pre-equilibrated with lysis buffer, the resin was washed with wash buffer \[50 m*M* HEPES pH 8.0, 300 m*M* NaCl, 40 m*M* imidazole, 10%(*v*/*v*) glycerol, 1 m*M* TCEP\] and the protein was eluted with elution buffer \[20 m*M* HEPES pH 8.0, 300 m*M* imidazole, 10%(*v*/*v*) glycerol, 1 m*M* TCEP\]. The eluate was buffer-exchanged with TEV buffer (20 m*M* HEPES pH 8.0, 200 m*M* NaCl, 40 m*M* imidazole, 1 m*M* TCEP) using a PD-10 column (GE Healthcare) and incubated with 1 mg TEV protease per 15 mg of eluted protein. The protease-treated eluate was run over nickel-chelating resin (GE Healthcare) pre-equilibrated with HEPES crystallization buffer (20 m*M* HEPES pH 8.0, 200 m*M* NaCl, 40 m*M* imidazole, 1 m*M* TCEP) and the resin was washed with the same buffer. The flowthrough and wash fractions were combined and concentrated by centrifugal ultrafiltration (Millipore) to 19.4 mg ml^−1^ for crystallization trials. NE1406 was crystallized using the nanodroplet vapor-diffusion method (Santarsiero *et al.*, 2002[@bb31]) with standard JCSG crystallization protocols (Lesley *et al.*, 2002[@bb22]). Sitting drops composed of 200 nl protein mixed with 200 nl crystallization solution were equilibrated against a 50 µl reservoir at 293 K for 50 d prior to harvest. The crystallization reagent consisted of 1.4 *M* ammonium sulfate and 0.1 *M* CHES \[2-(*N*-cyclohexylamino)ethanesulfonic acid\] pH 9.0. Glycerol was added to the crystal to a final concentration of 10%(*v*/*v*) as a cryoprotectant. Initial screening for diffraction was carried out using the Stanford Automated Mounting system (SAM; <http://smb.slac.stanford.edu/facilities/hardware/SAM/UserInfo>; Cohen *et al.*, 2002[@bb6]) at the Stanford Synchrotron Radiation Lightsource (SSRL; Menlo Park, California, USA). Diffraction data from a plate-shaped crystal with approximate dimensions 0.2 × 0.1 × 0.05 mm mounted in a nylon loop were indexed in the orthorhombic space group *P*2~1~2~1~2~1~ (Table 1[▶](#table1){ref-type="table"}). The oligomeric state of NE1406 was determined to be a monomer using a 0.8 × 30 cm Shodex Protein KW-803 column (Thomson Instruments) pre-calibrated with gel-filtration standards (Bio-Rad). Protein concentrations were determined using the Coomassie Plus assay (Pierce). 2.2.. Data collection, structure solution and refinement {#sec2.2} -------------------------------------------------------- Multiple-wavelength anomalous diffraction (MAD) data were collected at the APS on beamline 23-ID-D at wavelengths corresponding to the inflection (λ~1~), high-energy remote (λ~2~) and peak (λ~3~) points of the Se *K* absorption spectrum. The data sets were collected at 100 K using a MAR Mosaic300 CCD detector (Rayonix). The MAD data were integrated and reduced using *MOSFLM* (Leslie, 1992[@bb23]) and scaled with the program *SCALA* (Collaborative Computational Project, Number 4, 1994[@bb8]). Phasing was performed with *SOLVE* (Terwilliger & Berendzen, 1999[@bb38]), with a mean figure of merit of 0.28 with eight selenium sites (no selenium site was found for the disordered C-terminal SeMet356 for either chain). Density modification with *RESOLVE* (Terwilliger, 2002[@bb37]) was followed by automated model building with *ARP*/*wARP* (Cohen *et al.*, 2004[@bb7]). Model completion and refinement were carried out with *Coot* (Emsley & Cowtan, 2004[@bb13]) and *REFMAC* 5.2 (Winn *et al.*, 2003[@bb40]) using data set λ~1~. Refinement included experimental phase restraints in the form of Hendrickson--Lattman coefficients from *SOLVE*, NCS restraints (positional weights of 0.5 and 5.0 and thermal weights of 2.0 and 10.0 for the main-chain and side-chain atoms, respectively) and TLS refinement with one group per chain. NCS restraints were applied as two sets: to the N-terminal residues 24--74 and the C-terminal residues 83--351. Data-collection and refinement statistics are summarized in Table 1[▶](#table1){ref-type="table"}. 2.3.. Validation and deposition {#sec2.3} ------------------------------- Analysis of the stereochemical quality of the model was accomplished using *AutoDepInputTool* (Yang *et al.*, 2004[@bb41]), *MolProbity* (Davis *et al.*, 2007[@bb10]), *SFCHECK* 4.0 (Collaborative Computational Project, Number 4, 1994[@bb8]) and *WHAT IF* 5.0 (Vriend, 1990[@bb39]). Protein quaternary structure was analyzed using the *PISA* server (Krissinel & Henrick, 2007[@bb20]). Fig. 1[▶](#fig1){ref-type="fig"}(*b*) was adapted from an analysis using *PDBsum* (Laskowski *et al.*, 2005[@bb21]) and all other figures were prepared with *PyMOL* (DeLano Scientific). Atomic coordinates and experimental structure factors for NE1406 at 2.0 Å resolution have been deposited in the PDB with code [2ich](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2ich). 3.. Results and discussion {#sec3} ========================== 3.1.. Overall structure {#sec3.1} ----------------------- The crystal structure of a truncated version of NE1406 (Fig. 1[▶](#fig1){ref-type="fig"} *a*) was determined to 2.0 Å resolution using the MAD phasing technique. Data-collection, model and refinement statistics are summarized in Table 1[▶](#table1){ref-type="table"}. The final model includes 643 residues in two protein molecules (*A* and *B*), two CHES molecules, three glycerol molecules, one sulfate ion and 394 water molecules in the asymmetric unit. No electron density was observed for Gly0 (from the purification tag), Val23 (the first residue after Gly0), Thr75--Pro82 and Arg352--SeMet356 in chain *A* or for Thr75--Asp80 and Pro353--SeMet356 in chain *B*. The side-chain atoms of Leu24, Arg144, Glu169, Gln200, Asp222 from chain *A* and Leu24, Gln89 and Arg352 from chain *B* were omitted owing to poor electron density. The two chains are nearly identical, with an r.m.s.d. of 0.30 Å over 320 C^α^ atoms (0.60 Å over all 2524 equivalent atoms). The Matthews co­efficient (*V* ~M~; Matthews, 1968[@bb25]) is 2.35 Å^3^ Da^−1^ and the estimated solvent content is 47.3%. The Ramachandran plot produced by *MolProbity* (Davis *et al.*, 2007[@bb10]) shows that 98 and 100% of the residues are in favored and allowed regions, respectively. SCOP classifies NE1406 as an all-β protein with an AttH-like fold characterized by two flattened, orthogonally packed, β-barrels of lipocalin-like topology (<http://scop.mrc-lmb.cam.ac.uk/scop/data/scop.b.c.bai.b.b.b.html>). Lipocalins (PF00061) are an increasingly diverse family of predominantly small, single-domain, secreted proteins exhibiting high affinity and selectivity for hydrophobic molecules. Structurally, lipocalins form a subset of the calycin superfamily, which additionally includes avidins and fatty-acid binding proteins (FABPs) (Flower *et al.*, 1993[@bb14]; Pfam clan CL0116). Calycins are an example of a superfamily with members sharing structural similarities that cannot be detected at the sequence level. The calycin core fold comprises an eight-stranded calyx-shaped antiparallel β-barrel which opens toward one end, where the binding site is located. In the case of lipocalins and avidins, the core fold is maintained and differences are observed in the loop lengths and compactness of the barrel. In FABPs, the core calycin fold is supplemented by two additional β-strands and two short helices that pack on top of the lipid-binding cavity. In all cases, a short 3~10~-helix caps the barrel at one end, which is also latched by a conserved cation--π interaction involving a tryptophan from the first β-strand and a lysine or arginine residue from the final β-strand of the barrel. Both of these residues additionally form hydrogen bonds to main-chain atoms in the 3~10~-helix (Flower *et al.*, 2000[@bb15]). The N-terminal domain of NE1406 (residues 24--220) comprises 13 β-strands arranged in the form of a flattened barrel with a 3~10~-helix (H1 in Fig. 1[▶](#fig1){ref-type="fig"}) capping the barrel at one end (Fig. 1[▶](#fig1){ref-type="fig"} *a*). The C-terminal domain (residues 221--352) is arranged perpendicular to the long axis of the N-terminal barrel and comprises ten β-strands. It can be superimposed on the N-terminal domain with a C^α^ r.m.s.d. of 2.4 Å over 105 residues (Fig. 2[▶](#fig2){ref-type="fig"} *a*), suggesting gene duplication, although the sequence identity of only 9% is nonsignificant (Fig. 2[▶](#fig2){ref-type="fig"} *b*). Strands β5--β6 are absent from the C-terminal domain, while β11 is replaced by another 3~10~-helix (H3 in Fig. 2[▶](#fig2){ref-type="fig"} *b*). The 3~10~-helix cap of the N-terminal barrel is replaced by two longer strands, β18--β19 (in the C-terminal domain), that extend over one end of the barrel (Figs. 1[▶](#fig1){ref-type="fig"} *a* and 2[▶](#fig2){ref-type="fig"}). 3.2.. Detection of the calycin superfamily signature {#sec3.2} ---------------------------------------------------- A search with *FATCAT* (Ye & Godzik, 2004[@bb42]) using the entire NE1406 structure gave no significant hits. Individually, the N- and C-­terminal domains both showed structural similarity to a variety of β-barrel proteins, including outer membrane proteins (PDB codes [2erv](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2erv), [2jmm](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2jmm), [1k24](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1k24) and [1p4t](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1p4t)), avidin-related and streptavidin-related proteins (PDB codes [1avd](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1avd), [1wbi](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1wbi), [1y52](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1y52), [2ciq](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2ciq), [2uyw](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2uyw) and [1stp](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1stp)), fatty-acid binding proteins (PDB codes [1g5w](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1g5w) and [2q9s](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2q9s)), nitrophorin (PDB codes [1d2u](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1d2u) and [1u17](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1u17)) and a retinoic acid-binding protein (PDB code [1blr](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1blr)). The best score was for the outer membrane protein PagL from *Pseudomonas aeruginosa* (PDB code [2erv](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2erv)), which gave a C^α^ r.m.s.d. of 3.4 Å over 198 residues with a sequence identity of only 3%. This calycin-family signature in NE1406 (Fig. 3[▶](#fig3){ref-type="fig"} *b*) is conserved in the DUF2006 family. In the N-terminal domain of NE1406, the Arg214 side chain from β13 interacts with main-chain residues in both β1 and the N-terminal 3~10~-helix, whereas hydrogen bonding of the Trp50 indole to the 3~10~-helix is mediated *via* a glycerol molecule (Fig. 3[▶](#fig3){ref-type="fig"} *b*). Although the calycin signature is absent from the NE1406 C-terminal domain (Fig. 2[▶](#fig2){ref-type="fig"}), its presence in the N-terminal domain served to direct our analysis towards calycin-superfamily members. Analysis of the structural superposition of NE1406 with members of the calycin superfamily revealed a number of systematic differences (Figs. 3[▶](#fig3){ref-type="fig"} *c* and 3[▶](#fig3){ref-type="fig"} *d*). The β-sheets forming the NE1406 β-barrel are both longer and flatter than those in lipocalins, resulting in a narrower opening at the bottom of the barrel where the lipocalin-binding site would reside. The difference is even more pronounced when NE1406 is compared with avidins (PF01382; Fig. 3[▶](#fig3){ref-type="fig"} *d*), which have barrels that are more circular and compact than in lipocalins. In this respect, NE1406 resembles FABPs, which also exhibit a barrel that is flatter and more elliptical than in lipocalins. However, NE1406 lacks two additional helices at the top of the barrel that are a characteristic of FABPs. Secondary-structure elements, such as the long C-terminal α-­helix characteristic of most lipocalin-like calycins, *e.g.* nitrophorin (PF02087; Flower *et al.*, 2000[@bb15]; Skerra, 2000[@bb33]), are also absent from NE1406. Finally, the calycin signature residues are in different conformations to those typically described for calycins, with Trp50 adopting a different rotamer in NE1406 than in calycins and Arg214 not adopting a fully extended conformation. 3.3.. Similarities and differences with lipocalins {#sec3.3} -------------------------------------------------- NE1406 is likely to provide the first structural template for two other protein families. A search with *HHpred* (Soding *et al.*, 2005[@bb35]) against Pfam gave *E* values of 1.0 × 10^−15^ and 1.5 × 10^−7^ for protein families PF07143 and PF08622, respectively. PF07143 is a prokaryotic family of hydroxyneurosporene synthases that are implicated in carotene metabolism, while PF08622 is a family of fungal proteins that inhibit the generation of reactive oxygen species and promote survival during oxidative stress. The role of isoprenoids in photoprotection in plants (Penuelas & Munne-Bosch, 2005[@bb27]) and antioxidant defence in other eukaryotes (Tapiero *et al.*, 2004[@bb36]; Rao & Rao, 2007[@bb28]) has been well documented. A number of lipocalins, such as apolipoprotein D (ApoD; Sanchez *et al.*, 2006[@bb30]; Charron *et al.*, 2008[@bb5]; Eichinger *et al.*, 2007[@bb12]), neutrophil gelatinase-associated lipocalin (Roudkenar *et al.*, 2008[@bb29]; Goetz *et al.*, 2002[@bb16]) and α~1~-microglobulin (Olsson *et al.*, 2008[@bb26]; Schonfeld & Wojtczak, 2008[@bb32]), provide protection against oxidative stress by means of isoprenoids such as carotene. Other members of the calycin superfamily, such as avidins (PF01382), are not involved in this response. We therefore searched for other indications that NE1406 might be related to the lipocalin/cytosolic fatty-acid binding protein family (PF00061). Lipocalins have been likened to antibodies because of the high degree of structural plasticity that their binding sites exhibit, with numerous examples in which structural consolidation occurs upon binding (for a review, see Skerra, 2008[@bb34]). As a result, the lipocalin fold has been employed in a number of protein-engineering studies (Beste *et al.*, 1999[@bb1]; Korndorfer *et al.*, 2003[@bb19]). In the NE1406 crystal structure, the two lipocalin-like barrels lack the large internal cavity that is typical of lipocalins and also the long structurally flexible loops at the open end of the β-barrel (Skerra, 2000[@bb33]). In fact, only one of the β-­barrel domains of NE1406 harbors a small glycerol molecule from the crystallization solution as a ligand. However, the complete internalization of the glycerol molecule in the NE1406 structure suggests that the N-terminal lipocalin-like barrel might adopt different conformations in the presence of a natural ligand. We therefore propose that this region, which encompasses the calycin signature, acts as a ligand-binding site, the shape and accessibility of which may change with natural ligands. The ability to form dimers is another feature of the lipocalin family, with ligand presence influencing oligomerization (Grzyb *et al.*, 2006[@bb17]). Analytical size-exclusion chromatography shows that NE1406 forms a monomer in solution, whereas crystal-packing analysis suggests a dimer with a total buried surface area of 1290 Å^2^ per monomer. While it is possible that dimerization of NE1406 is modulated by ligand binding, the relative orientation of the two protein domains within the polypeptide chain could also be subject to regulation by a second ligand. The two barrels are stabilized in a perpendicular orientation with respect to each other. The mainly aromatic and hydrophobic residues implicated in the interaction with CHES are highly or strictly conserved among DUF2006 homologs, suggesting that the domain interface plays a functional role. As with the glycerol molecule bound within the N-terminal barrel, the CHES molecule is also fully enclosed within NE1406 with no exposure to solvent, suggesting some flexibility at the interdomain interface to accommodate ligands. Ligand binding at the domain interface might act to regulate the shape of the binding cavity within one or both of the β-barrels in a similar manner to the regulation by dimerization observed in lipocalins. Finally, some lipocalins, such as the bacterial lipocalin (Blc), ApoD and lazarillo, are known to be peripherally anchored to biological membranes, where they are thought to play a role in membrane biogenesis and repair (Bishop, 2000[@bb3]; Eichinger *et al.*, 2007[@bb12]). Expressed under conditions known to exert stress on the bacterial envelope, Blc from *E. coli* has a high affinity for lysophospholipids (LPLs), which may also be bound inside the β-barrel and are thought to be involved in cell-envelope LPL transport (Campanacci *et al.*, 2006[@bb4]). Although the exact mechanisms of transperiplasmic movement of lipids between inner and outer membranes are largely unknown, ATP-binding cassette transporters are involved in this process (Doerrler *et al.*, 2004[@bb11]). As expected, a search with *PROFtmb* (Bigelow *et al.*, 2004[@bb2]) shows that NE1406 is not predicted to be a transmembrane β-barrel (*Z* score 2.9). However, calculations with the program *PPM* (Lomize *et al.*, 2006[@bb24]) suggest weak peripheral association of the protein with membrane. The ligand-binding cavity of the β-barrel opens towards the membrane surface in the predicted orientation (Supplementary Fig. 1[1](#fn1){ref-type="fn"}), similar to ApoD (Eichinger *et al.*, 2007[@bb12]). The membrane-interacting residues of the protein include the exposed hydrophobic Phe85 and a large patch of basic residues (Arg46, Arg113, Lys249, Arg284, Arg287, Arg319 and Arg352). 3.4.. Genome-context analysis {#sec3.4} ----------------------------- The genome context (<http://string.embl.de>) of NE1406 shows a predicted functional association with the lipoprotein-releasing system ATP-binding protein LolD (*lolD*) and co-occurrence with an ATP-binding protein ABC transporter (NE1404). A high degree of con­fidence is predicted for the functional association of many DUF2006 homologs with ATP-dependent ABC transporters, as well as with other transmembrane proteins including Na^+^/H^+^ antiporters, sensor histidine kinases and lipoproteins (*e.g.* LprI precursor in *Mycobacterium tuberculosis*). The systematic presence of ATP-dependent cassettes and lipoproteins is compatible with a role for the DUF2006 family in lipid transport, while the presence of numerous signal transduction genes might indicate expression under specific conditions, such as environmental stress. Further experiments will be required in order to functionally characterize NE1406 and to determine whether it associates with lipids *in vitro* or *in vivo* and whether its transcription is subject to environmental regulation. The DUF2006 protein family contains over 400 homologs distributed among trypanosomata, fungi, mycobacteria, bacteroidetes, rhizobia, *Vibrio*, spirochaetes, firmicutes and archaea. Given the wide phylogenetic presence of the DUF2006 family, if an experimental connection to lipocalins is determined, this finding would present the first evidence of a lipocalin-related protein in the Archaea domain and would settle the question of whether or not this protein family may have arisen *via* horizontal transfer to eukaryotic cells from the endosymbiotic α-proteobacterial ancestor of the mitochondrion (Bishop, 2000[@bb3]). The availability of more DUF2006 sequences and structures might shed light on the evolutionary history of this intriguing protein family. The information presented here, in combination with further bio­chemical and biophysical studies, should yield valuable insights into the functional role of NE1406. Models of NE1406 homologs can be accessed at <http://www1.jcsg.org/cgi-bin/models/get_mor.pl?key=2ichA>. Additional information about the protein described in this study is available from *TOPSAN* (Krishna *et al.*, 2010[@bb43]) at <http://www.topsan.org/explore?PDBid=2ich>. 4.. Conclusions {#sec4} =============== NE1406 adopts a lipocalin-like fold with domain duplication. Analysis based on the calycin-superfamily signature present in the N-­terminal domain reveals a potential binding site, while remote sequence homology and the genome context suggest involvement in isoprenoid metabolism and survival under oxidative stress. Supplementary Material ====================== PDB reference: [NE1406 from *N. europaea*, 2ich, r2ichsf](2ich) Supplementary material file. DOI: [10.1107/S1744309109037749/wd5116sup1.pdf](http://dx.doi.org/10.1107/S1744309109037749/wd5116sup1.pdf) Supplementary material has been deposited in the IUCr electronic archive (Reference: [WD5116](http://scripts.iucr.org/cgi-bin/sendsup?wd5116sup1.pdf)). This work was supported by National Institutes of General Medical Sciences Protein Structure Initiative grant No. U54 GM074898. Portions of this research were carried out at the Advanced Photon Source (APS) and SSRL. GM/CA CAT has been funded in whole or in part by Federal funds from the National Cancer Institute (Y1-CO-1020) and the National Institute of General Medical Science (Y1-GM-1104). Use of the Advanced Photon Source was supported by the US Department of Energy, Basic Energy Sciences, Office of Science under contract No. DE-AC02-06CH11357. The SSRL is a national user facility operated by Stanford University on behalf of the US Department of Energy, Office of Basic Energy Sciences. The SSRL Structural Molecular Biology Program is supported by the Department of Energy, Office of Biological and Environmental Research and by the National Institutes of Health (National Center for Research Resources, Biomedical Technology Program and the National Institute of General Medical Sciences). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of General Medical Sciences or the National Institutes of Health. Genomic DNA from *N. europaea* strain ATCC 19718 (ATCC \#19718D) was obtained from the American Type Culture Collection (ATCC). {#fig1} {#fig2} ![Similarities and differences between NE1406 and the calycin superfamily. (*a*) Stereo ribbon diagram of the binding sites for the two buffer molecules 2-(*N*-cyclohexylamino)ethanesulfonic acid (CHES) and glycerol (GOL). Conserved residues are indicated. (*b*) NE1406 exhibits the calycin-superfamily structural signature. Stereo ribbon diagram of the N-terminal domain of NE1406 showing the stacked arginine and tryptophan residues characteristic of the calycin fold (Flower *et al.*, 2000[@bb15]). Hydrogen bonds are indicated by dashed lines. A glycerol molecule (cyan) mediates bonding of Trp50 to the 3~10~-helix. (*c*) Ribbon diagrams depicting the front and back view of NE1406 (PDB code [2ich](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2ich), residues 24--220; gray) superposed with nitrophorin 4 from *Rhodnius prolixus* (PDB code [1d2u](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1d2u), residues 22--205; red. The heme ligand for nitrophorin 4 is colored cyan. (*d*) Ribbon diagrams depicting the front and back view of NE1406 (PDB code [2ich](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2ich), residues 24--220; grey) superposed with avidin from *Gallus gallus* (PDB code [1avd](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=1avd), residues 3--125; pink). The Trp-Arg signatures are represented as sticks. The biotin ligand for avidin is shown in cyan.](f-66-01153-fig3){#fig3} ###### Summary of crystal parameters, data-collection and refinement statistics for NE1406 (PDB code [2ich](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=pdb&pdbId=2ich)) Values in parentheses are for the highest resolution shell.   λ~1~ MADSe λ~2~ MADSe λ~3~ MADSe --------------------------------------------------------------------- ---------------------------------------- -------------------------- -------------------------- Data collection  Space group *P*2~1~2~1~2~1~  Unit-cell parameters (Å) *a* = 63.27, *b* = 95.57, *c* = 121.75  Wavelength (Å) 0.9794 0.9493 0.9792  Resolution range (Å) 29.20--2.00 (2.05--2.00) 29.20--2.00 (2.05--2.00) 29.10--2.00 (2.05--2.00)  No. of observations 178048 177082 176130  No. of unique reflections 49800 49531 49656  Completeness (%) 98.4 (95.9) 97.9 (95.0) 98.4 (95.6)  Mean *I*/σ(*I*) 9.3 (2.1) 9.8 (2.3) 8.8 (2.0)   *R*~merge~ on *I*[†](#tfn1){ref-type="table-fn"} 0.117 (0.599) 0.109 (0.535) 0.121 (0.602) Model and refinement statistics  Resolution range (Å) 29.2--2.00      No. of reflections (total) 49646[‡](#tfn2){ref-type="table-fn"}      No. of reflections (test) 2528      Completeness (%) 98.0      Data set used in refinement λ~1~      Cutoff criterion \|*F*\| \> 0       *R*~cryst~[§](#tfn3){ref-type="table-fn"} 0.182       *R*~free~[¶](#tfn4){ref-type="table-fn"} 0.232     Stereochemical parameters  Restraints (r.m.s.d. observed)         Bond angles (°) 1.65       Bond lengths (Å) 0.018      Average isotropic *B* value (Å^2^) 27.9[††](#tfn5){ref-type="table-fn"}      ESU[‡‡](#tfn6){ref-type="table-fn"} based on *R*~free~ (Å) 0.16      Protein residues/atoms 643/5142      Water molecules/ions/other solvent[§§](#tfn7){ref-type="table-fn"} 394/1/5     *R* ~merge~ = . Typically, the number of unique reflections used in refinement is slightly less that the total number that were integrated and scaled. Reflections are excluded owing to systematic absences, negative intensities and rounding errors in the resolution limits and unit-cell parameters. *R* ~cryst~ = , where *F* ~calc~ and *F* ~obs~ are the calculated and observed structure-factor amplitudes, respectively. *R* ~free~ is the same as *R* ~cryst~ but for 5.1% of the total reflections chosen at random and omitted from refinement. This value represents the total *B* that includes TLS and residual *B* components. Estimated overall coordinate error (Collaborative Computational Project, Number 4, 1994[@bb8]; Cruickshank, 1999[@bb9]). Two CHES and three glycerol molecules.

{.137}

Sir, A 58-year-old woman with poorly controlled diabetes mellitus presented with a 2-week history of high grade fever with rapidly progressive shortness of breath 1 day prior to admission. She was a known case of dilated cardiomyopathy and diabetic nephropathy. Systemic examination revealed tachypnea, decreased breath sounds bilaterally with basal rales, decreased intensity of heart sounds and audible S3, tender hepatomegaly with fluid thrill. The complete blood cell count showed a hematocrit of 30% and a leukocyte count of 32,000/mm^3^. The serum glucose level was 245 mg/dl, serum urea nitrogen level was 168 mg/dl, serum creatinine was 4.6 mg/dl and urinalysis revealed 20-30 pus cells with proteinuria. Serum procalcitonin was 6.77 ng/ml. Arterial blood gas value showed metabolic acidosis (anion gap-26 mmol/l) with negative blood and urine ketones. The chest radiograph demonstrated cardiomegaly and blunting of both cardiophrenic angles \[[Figure 1](#F1){ref-type="fig"}\]. Ultrasonography of chest and abdomen showed bilateral moderate pleural effusion and small ascites. Transthoracic echocardiography revealed ejection fraction of 25% with global hypokinesia and moderate pericardial effusion. {#F1} Clinical symptomatology suggested a provisional diagnosis of urinary tract infection with septic shock and cardiac failure. The patient was started on broad spectrum antibiotics and vasopressors. Urine output remained between 500 and 800 ml/day. By the fifth day, she was afebrile, weaned off inotropes and non-invasive ventilator support but required alternate day hemofiltration. Blood and urine cultures sent at admission were sterile. It was on ninth hospital day that she again became febrile and her chest X-ray revealed left upper lobe collapse \[[Figure 2](#F2){ref-type="fig"}\]. Her oxygen requirement increased, she was intubated and put on mechanical ventilator. Computed tomography (CT) of the chest revealed heterogeneous mass measuring 5.3 × 3.2 cm in the left hilum, completely obliterating the left upper lobe \[[Figure 3](#F3){ref-type="fig"}\]. Fiberoptic bronchoscopy revealed reddish polypoidal growth causing almost complete obstruction of left upper lobe bronchus. Histopathology examination of the bronchial biopsy showed irregular broad aseptate hyphae branching at right angles, consistent with mucormycosis. \[[Figure 4](#F4){ref-type="fig"}\]. The patient was started on liposomal amphoterecin and surgical resection of upper lobe was planned, but patient\'s condition deteriorated. She developed multi-organ failure and had massive hemoptysis followed by cardiac arrest and succumbed to treatment on the 24^th^ day of hospitalization. {#F2} {#F3} {#F4} Mucormycosis is an opportunistic infection causing deep tissue infection. It is a potentially fatal opportunistic fungal infection caused by certain fungi of the order Mucorales. It occurs commonly in immunocompromised hosts but has also been demonstrated in patients with normal immune status. Five predominant forms of mucormycosis seen clinically are: Rhinocerebral, pulmonary, disseminated, cutaneous (particularly burn wounds), and gastrointestinal mucormycosis. Pulmonary mucormycosis is commonly seen in patients with hematological malignancies. Common pulmonary manifestations include consolidation (66%) or cavitation (40%), less commonly as a solitary pulmonary nodule and multiple mycotic pulmonary artery aneurysms.\[[@ref1]\] Presentation as an endobronchial mass lesion is extremely rare. The production of spores, which become airborne, leads to the primary route of inoculation in the respiratory tract. Risk factors for mucormycosis are neutropenia, lymphopenia, hyperglycemia, pre-existing renal failure and prolonged steroid use. Our patient had uncontrolled diabetes with renal failure, a combination that made her an ideal host for mucormycosis. The clinical manifestations of pulmonary mucormycosis cannot be easily distinguished from those of bacterial infection. The most prominent physical finding is bilateral fixed rales or pleural rubs. Rapidly progressive pneumonia is seen in those patients with underlying hematological malignancies. Few case reports in the literature have mentioned a distinct clinical syndrome of mucormycosis in diabetics, as endobronchial polypoidal lesion.\[[@ref2]\] Nearly, 78% of patients of pulmonary mucormycosis can have an acute presentation and our patient had a very rapidly progressive clinical course as well . Definite diagnosis is made by a biopsy and histopathology. The characteristic histological feature is tissue invasion by aseptate, broad, right angled branching hyphae with a propensity to invade blood vessels. Culture is considered as the gold standard for disease diagnosis and species identification. However, unfortunately the recovery of fungi from culture is less sensitive due to hyphal damage during processing of the specimen. Hence, it has been agreed that microscopic identification of characteristic fungi invading affected tissues should be considered significant \[[Figure 4](#F4){ref-type="fig"}\].\[[@ref3]\] Therapy for pulmonary mucormycosis should be aggressive as the fungus produces a locally invasive infection with a high fatality rate. According to IDSA Guidelines, amphoterecin B deoxycholate (0.7-1.0 mg/kg/d) or its lipid formulations (5 mg/kg/d) remain the primary antimicrobial agent for mucormycosis.\[[@ref4]\] Mucormycosis has an extremely high mortality rate ranging from 25% to 80%. Prognosis of pulmonary mucormycosis depends on the underlying immune-compromised state. Survival rate is better in diabetics (45-60%), organ transplant recipients than patients with renal insufficiency.\[[@ref3]\] Our patient had the two most important predisposing factors that made the chances of her survival worst. The common causes of death are fungal sepsis (42%), respiratory insufficiency (27%) and hemoptysis (13%). Massive hemoptysis may be due to vascular invasion, multiple pulmonary artery pseudoaneurysms and can lead to death due to asphyxiation.\[[@ref5]\] Patients on medical therapy alone have a mortality of 55% compared to 27% in patients who get surgical treatment with or without medical treatment.\[[@ref3]\] Surgical therapy is highly recommended and must be performed without delay because of the aggressive nature of the disease with propensity for vascular invasion. We managed our patient conservatively and planned for lobectomy, but she developed multi-organ failure and did not give time for surgery. In conclusion, pulmonary mucormycosis is a rare, rapidly spreading disease and fatal disease, which can masquerades many common conditions including lung cancer. A high level of index should be kept in diabetics and immune-compromised hosts who fail to respond to antibacterial therapy. Early recognition and aggressive management requires systemic antifungal therapy, surgical resection, coupled with control of underlying disease.

Introduction ============ There are numerous factors that can make the work of healthcare providers stressful. For instance, caregivers sometimes have to face the death of patients, and there can be conflicts with the family of the patient. They may also lack the necessary equipment to do a good job, but still be taxed with excessive responsibility for negative outcomes. The work load may simply be too high, and the cooperation with team members may be ineffective and unpleasant. Health care workers often feel untrained at the beginning of their career and not respected at the end. They often fail to grasp patients' expectations about the intensity and quality of the care; this is also true for how satisfying the given care is experienced by the patients. The administrative mechanisms might be complex, and health care workers may not get the appreciation they deserve from their superiors. Moreover, many healthcare providers suffer a reduced quality and quantity of sleep which makes coping with the challenges listed even harder (for reviews see [@B44]; [@B23]; [@B26]; [@B16]). Given these difficult work conditions, one wonders what kind of interventions might help healthcare professionals to reduce their stress. One intervention that is found to be effective in all kinds of work settings is teaching people to redirect attention away from incomplete work goals during their leisure time ([@B38]); another successful approach focuses on teaching people what determinants facilitate recovery from different types of job stress (e.g., control during off job-time) thus heightening respective self-efficacy feelings ([@B12]). Further effective interventions aim at enhancing physical activity thus increasing sleep quality and quantity, as well as reducing lower back pain (e.g., [@B22]). Interventions that specifically help healthcare providers to distance themselves from upsetting events and negative experiences (i.e., to adopt a watch tower perspective) were found to buffer even highly anxious caregivers from short and longer-term emotional distress ([@B31]). In all of these interventions, healthcare professionals were taught about specific behaviors that have been found to reduce stress (e.g., redirecting attention away from incomplete goals, control during off-job time, increasing physical activity, distancing from upsetting events) through individual coaching or in workshops with a group of participants. The relevant information was provided through presentations, in the form of brochures, or via the internet. Some of these interventions also stimulated the participants to think about when and where in everyday life they wanted to act on what they had been taught (e.g., [@B38]). The intervention we tested in the present study is different to these approaches in many respects. In contrast to the previously described approaches, which taught participants to establish specific stress reduction behaviors (e.g., increased physical activity) and encouraged participants to use them, we left it up to the participants to identify their own personal idiosyncratic wish of less stress at the workplace. In addition, we asked them to identify and imagine the best outcome of wish fulfillment, and identify and imagine the most important personal obstacle that they anticipated with regard to wish fulfillment. Once participants detected their own personal obstacle they had to vividly imagine its occurrence. Thereafter they had to name an effective idiosyncratic behavior to surmount the obstacle, before making an if-then plan in the form of "if...obstacle, then I will... behavior to surmount obstacle." Such a plan links the idiosyncratic obstacle (if-part), to the idiosyncratic behavior to overcome the obstacle (the then-part of the plan). In sum, whereas in the stress reduction interventions used so far people are informed about an established way of coping with stress and encouraged to use it in order to reduce stress, the self-regulation strategy introduced in the present research focuses on heightening the autonomous control of stress reducing behavior through mental imagery. The imagery procedure is based on the integration of two established self-regulation strategies: mental contrasting and forming implementation intentions (e.g., [@B29]). Self-Regulation By Mental Contrasting (MC) ========================================== Mental contrasting triggers goal pursuit by juxtaposing positive future fantasies with obstacles of present reality. When people contrast their wishes of a positive future with the obstacles standing in its way, the energy needed to overcome these obstacles is activated, given that this obstacle can potentially be overcome. For example, a person might identify a wish related to stress reduction at the work place: "I would love to stay calm even when blamed by patients!" After this first step of wish identification, she would identify the best outcome of staying calm and imagine that outcome, such as herself responding to the patient in a comforting manner. She would then identify and imagine her obstacle: "my urge to tell patients' off when they become unreasonable." Mental contrasting fosters behavior change across life domains: academic and professional achievement, interpersonal relationships, and health and wellbeing (reviews by [@B27], [@B28]). Mental contrasting enabled students to learn foreign language vocabulary, improve in math, study abroad, and complete a vocational training. Mental contrasting also helped with finding integrative (win--win) solutions in negotiations and with good decision making in everyday life. In the social realm, mental contrasting has been found to strengthen interpersonal relations and lead to effective reconciliation. It also heightened tolerance, encouraged taking responsibility for members of disadvantaged groups, promoted help seeking in college students and help giving in emergency care nurses. It was even effective in enhancing physical activity in stressed out students who just started their college education ([@B32]). Extensive experimental research on the underlying mechanisms of mental contrasting effects has demonstrated that mental contrasting is a conscious imagery strategy that affects non-conscious cognitive processes, motivation, and responses to feedback, which in turn facilitate wish fulfillment. Three cognitive processes taking place outside of awareness have been delineated as such mediators. First, mental contrasting induces people to interpret the present reality as an obstacle to wish fulfillment (e.g., a party is no longer a fun event but an obstacle to meeting one' wish of getting a good grade in an upcoming exam; e.g., [@B20]). At the same time, mental contrasting strengthens the implicit associative links between the desired future and the obstacle as well as between the obstacle and the instrumental behavior to overcome the obstacle (e.g., [@B19]; [@B17]). With respect to motivation, a heightened energization level, as measured by systolic blood pressure and by subjective reports of energy, was found to mediate the effects of mental contrasting on a person's efforts to realize her wishes (see e.g., [@B36]). And mental contrasting facilitates dealing with setbacks in a way that fosters resilience: It fosters the processing of information contained in setbacks, and it protects against a loss of subjective competence ([@B18]). Self-Regulation by Implementation Intentions (II) ================================================= Implementation intentions are if-then plans (review by [@B9]) in the following format: "If the critical situation X is encountered, then I will perform the goal-directed response Y!" These implementation intentions are to be differentiated from mere goal intentions. The latter merely specify desired end states ("I want to achieve goal X!" or "I want to exert behavior X!"). In implementation intentions, on the other hand, the if-component of an implementation intention specifies a future critical event or point in time, and the then-component specifies how one will respond once these situational cues are encountered. The person described above who fails to control her urge to tell patients off when they approach her with unreasonable complaints could make the following if-then plan: "And if I feel too weak to decisively put a halt to my urge, then I'll tell myself: Just take a watchtower perspective and aim at being calm and constructive!" Evidence that forming if-then plans enhances the rate of attaining desired outcomes and execute the respective instrumental responses have been obtained in many studies regarding achievement, health, sports, and social relationships (meta-analysis by [@B10]). Implementation intentions have been shown to be an effective strategy to overcome external distractions (e.g., an exciting video) or internal hindrances such as self-doubts ([@B43]) in the service of persistent action. Implementation intentions' self-regulatory benefits for action control extend into emotion regulation as well (e.g., fear, disgust, and anger; summary by [@B45]). Finally, with respect to the regulation of cognitive responses research demonstrated that if-then plans help people to switch from reflexive to reflective thinking thus improving decision making (e.g., [@B2]). But how do implementation intentions work? Experimental research found that the mental representation of the selected situation in the if-part becomes highly activated and hence more accessible (e.g., [@B46]). Moreover, linking the if-part to the then-part produces automaticity ([@B8]) in the sense that encountering the specified situation triggers the specified response in an automatic fashion: this response is now performed immediately, efficiently, and no further conscious intent is needed. Not surprisingly, then, various studies could show that people who form implementation intentions are in a good position to break unwanted habitual responses (e.g., [@B1]). Mental Contrasting With Implementation Intentions (MCII) ======================================================== Sometimes people face obstacles that are surmountable but are particularly hard to deal with (e.g., impulsive behavior, strong emotions, and ingrained habits). While mental contrasting builds non-conscious associative links between the obstacle and the behavior instrumental to overcoming the obstacle thereby fostering goal pursuit, it might be useful to add a strategy that strengthens these associative links even further. Forming implementation intentions qualifies as such a strategy. Accordingly, MC and II were combined into one strategy called mental contrasting with implementation intentions (MCII; [@B29]). As described above, mental contrasting instigates goal pursuit (goal commitment and goal striving), and goal commitment is a prerequisite for the beneficial effects of implementation intentions ([@B37]). Mental contrasting also helps to identify the critical situation for the if-part (obstacle) of implementation intentions and the instrumental action for the then-part of the plan (overcoming the obstacle). Numerous studies have demonstrated the effectiveness of MCII as an intervention, many of them in the health domain. Participants who employed MCII to eat more healthily engaged in more physical exercise over a period of 4 months ([@B41]), consumed more fruits and vegetables over a period of 2 years ([@B42]), and ate less red meat over 5 weeks ([@B24]). MCII helped increase physical exercise and weight reduction in patients who had a stroke over a period of 1 year ([@B25]), and increased physical capacity in patients with chronic back pain over 3 months ([@B3]). An intervention study by [@B34] observed that MCII even helped patients with schizophrenia in autonomy-focused clinical hospital settings to translate their exercising intentions into action. When applied to the domain of interpersonal relationships, MCII increased commitment to the relationship and decreased insecurity-related behaviors such as avoiding sensitive topics ([@B15]). MCII also helps people to manage their time. For example, it supported both working mothers from low-income backgrounds who were enrolled in vocational education and medical residents who were serving in intensive care units to find the time to study for their exams ([@B30]; [@B33]). Regarding MCII interventions taught online, [@B21] found in two studies that MCII delivered as an 8-min online intervention in the context of massive open online courses (MOOC) substantially increased completion rates of the online courses. The Present Research ==================== We invited nurses from all over Germany to participate in our online study to insure that institutions with different kinds of organizational and hierarchical structures were covered. Based on the results of past research on behavior change via MCII described above, we hypothesized that MCII should help nurses to fulfill their wishes regarding reducing their stress, whether the specific content of the wish pertains to health, interpersonal relations, or achievement. We tested the effectiveness of engaging in MCII (to be performed over 3 weeks on a daily basis) compared to a no-treatment control group that only was asked to explore their wishes regarding achieving less stress. We also added a further intervention group (i.e., IIMCII); here participants were asked to furnish the goal to engage in daily MCII exercises, assigned by the experimenter, with an implementation intention that specified when and where they planned to execute these MCII exercises. We included this third condition because we were worried that healthcare professionals might not find the time to perform MCII on a regular basis, and therefore only a moderate stress reduction might be observed in the mere MCII group. Based on previous research showing beneficial effects of implementation intentions on goal attainment ([@B9]), we hypothesized that participants in the IIMCII condition would benefit from using if-then plans that specify the situation in which they wanted to engage in MCII, thus showing even more stress reduction and even more work engagement than participants in the mere MCII condition. Participants' stress level was assessed prior to the intervention and 3 weeks later with established self-report questionnaires pertaining to perceived stress, stress-related physical symptoms, and work engagement. Materials and Methods {#s1} ===================== Design ------ The study used a randomized 3 (Intervention, between: control vs. MCII vs. IIMCII) × 2 (Time, within: baseline-measurement vs. post-treatment measurement) factorial design. As dependent variables we assessed (a) perceived stress, (b) current physical symptoms, and (c) work engagement, once at the beginning of the interventions and then again 3 weeks later. Participants ------------ The administrations of health care institutions all over Germany were contacted via regular mail, emails, and phone calls over a time period of 3 months (October to December). The administrators were asked to pass on a prepared email message to their nurses, asking them whether they would be interested in participating in a study on stress reduction. This message explained that participation would potentially help reduce one's stress level, and that there is a chance (albeit small) of winning a gift certificate of 100 Euro. The message also contained the email address of the experimenter whom the nurses should contact if they wanted to register for the study. Those who registered (*N* = 251 nurses) were contacted in return by the experimenter (again via email) and given access to the study website that had been created by using the *soscisurvey.de* data collection service. Participants who entered the website (*N* = 129) were randomly assigned to the three conditions of the study (MCII = 41, and IIMCII = 41, Control = 47) (see **Figure [1](#F1){ref-type="fig"}**). {#F1} On the study website, all participants were first asked to complete three questionnaires assessing their baseline stress level. Thereafter, they were guided to the instructions of the respective experimental condition. Three weeks later, the experimenter contacted the participants via email encouraging them to re-enter the study website at *soscisurvey.de.* This time they were only asked to fill out the three stress-related questionnaires a second time. Thirty-three participants of the MCII group, 34 participants of the IIMCII group, and 38 participants of the control group answered the final stress questionnaires: a return rate of 80.5, 82.9, and 80.9%, respectively. These 105 participants (82% female) had a mean age of 40.22 years (*SD* = 10.18), and an average length of work experience of 17.60 years with a minimum of 0 years (in training) and a maximum of 40 years. Eighty-one of the participants worked in a hospital, 5 in residential care, 5 in a nursing home, 5 in psychiatric institutions, 3 in day hospitals, 2 in rehabilitation centers, and 4 in not specified institutions. Also, 70.5% had full-time positions; 54.3% worked in shifts, 40.0% in night shifts, and 48.6% claimed to irregularly work extra hours. Materials and Procedure ----------------------- ### Baseline Measurement At Time 1, all participants had to first fill out three established questionnaires. Two of them focused on assessing participants' overall stress level, one of them by targeting participants' perceived stress and the other by obtaining information on physical symptoms of stress. Perceived stress and physical symptoms in response to a stressor both pertain to people's overall stress level ([@B7]). In addition, we used a questionnaire pertaining to a potential consequence of having a lot of stress: participants' level of work engagement. The order of the questionnaires was as follows: perceived stress, work engagement, and physical symptoms of stress. We placed the work engagement questionnaire in between the two stress questionnaires so that participants' answers to the perceived stress questionnaire would not carry over to reporting on physical symptoms of stress. ### Perceived Stress Perceived stress was measured with an adapted version of the Perceived Stress Questionnaire-20 (PSQ-20; [@B5], [@B6]). The 20-item questionnaire asks participants to base their responses on the last 3 weeks: "Please specify how often the following statements applied to you in the last 3 weeks." Sample items: *You felt rested* (reverse coded), *You had problems relaxing*, *You felt safe and protected* (reverse coded), and *You were under time pressure*. Responses ranged from 1 = *almost never* to 4 = *usually*. This questionnaire covers the subjective experience of stress independent of specific stressors on four subscales with five items each. The subscale *worries* targets current worries, anxiety about the future and emotions of frustration; the subscale *tension* pertains to current fatigue, imbalance, and the lack of physical relaxation; the subscale *joy* covers positive feelings; and the subscale *demands* measures current lack of time, time pressure, or burden of tasks. The first three scales target the internal stress reactions of an individual, while the scale *demands* assesses the perception of external stressors. The PSQ has been demonstrated to be reliable and valid. [@B5], [@B6]) report that the questionnaire has high internal consistency (Cronbach's α = 0.85). In our study, the baseline perceived stress questionnaire had a good internal consistency as well (Cronbach's α = 0.93); composite reliability (CR) was 0.76. #### Physical symptoms Symptoms were measured with the physical symptoms subscale of the Burnout Screening Scales II inventory (BOSS II; [@B11]). This subscale targets the physical symptoms of individuals, especially as they would typically occur in people with a burnout syndrome or with chronic stress. It contains ten items that cover various physical disabilities, pains, and somatic ailments. The emphasis of this scale is placed on cardiovascular complaints. Further, functional constraints of the respiratory, digestion, and immune systems as well as general parameters of states of stress and sleeping quality were assessed. We instructed participants by stating: "Please specify whether the following statements applied to you in the last 3 weeks." Examples were: *I feel pressure in my chest*, *I am plagued by serious headaches*, and *I have respiratory difficulties*. Responses ranged from 1 = *does not apply* to 6 = *strongly applies*. The Burnout Screening Scales (BOSS II) inventory has been demonstrated to be reliable and valid ([@B11]). In our sample the internal consistency at the baseline assessment was also high, Cronbach's α = 0.83; composite reliability (CR) was 0.76. #### Work engagement We assessed work engagement as an indirect indicator of stress assuming that stressed out nurses show reduced work engagement. Work engagement was assessed using a condensed version of the Utrecht Work Engagement Scale (UWES-9; [@B35]). The UWES-9 questionnaire measures work engagement with three subscales: *vigor* in the sense of energy, strength and perseverance at work, *dedication* in the sense of interest, inspiration, pride, and challenge, and *absorption* in the sense of concentration at work, loss of a sense of time, and felt importance of work. The instructions for this questionnaire were: "Please choose the respective answer that best applies to you regarding the last 3 weeks." Sample items are: *My work inspires me*, *My work is fulfilling*, and *I am proud of my wor*k. The answer scale ranged from 1 = *never*, 2 = *almost never*, 3 = *occasionally*, 4 = *regularly*, 5 = often, 6 = *very often*, to 7 = *always*. [@B35] found a good internal consistency of the questionnaire (Cronbach's α = 0.80). In our study, the UWES-9 assessed at baseline had a high internal consistency as well (Cronbach's α = 0.94); composite reliability (CR) was 0.95. ### Instructions to Use Stress-Reduction Strategies After having filled out the three questionnaires, participants of all conditions read instructions asking them to first think about the topic of stress at work and to answer three general questions about stress (see **Figures [2](#F2){ref-type="fig"}**, **[3](#F3){ref-type="fig"}**). Participants read: 1. Less Stress at Work -- A Dream? Make this dream become a reality! As a start, please think about stress. Please answer the following questions (take as much time as you need). 2. What does stress mean for me? 3. What does less stress mean for me? 4. Why do I want less stress? 5. My wish for less stress: ... 6. Please write down your most important personal wish for less stress at your work place in the next 3 weeks in one or two sentences. What exactly do you want? {#F2} {#F3} #### MCII group Thereafter, MCII-participants were asked to go through the following mental exercise at their own pace. First, they had to note the best possible outcome of their wish for less stress. Afterward, they were asked to imagine the events and experiences associated with this best outcome. They subsequently had to write down their thoughts. The exact instructions read: Please think about the best outcome of fulfilling your wish for less stress and take note of your thoughts by writing down one or two sentences. 1. The best outcome of fulfilling my wish in the next 3 weeks is: ... 2. Now imagine the best outcome in your thoughts. Imagine the events and experiences you associate with the best outcome. Please take as much time as you need to imagine it as fully as you can. 3. Now please write down all the thoughts and images you had regarding the best outcome: ... Then, the main obstacle had to be identified and elaborated. The participants were asked to consider what speaks against the fulfillment of the wish in the next 3 weeks. They had to imagine the events and experiences related to this main obstacle separating them from reaching their goal. Then, they had to take note of this main obstacle and vividly imagine it. The instructions stated: Sometimes, our wishes are not fulfilled. Think about what speaks against your wish being fulfilled in the next 3 weeks. 1. What is your main obstacle? Please write it down in one or two sentences. 2. My main obstacle to fulfilling my wish in the next 3 weeks is:... 3. Now please imagine this main obstacle. What are the events and experiences that could hold you back? Please take as much time as you need to imagine it as fully as you can. 4. Now please write down all the thoughts and images you had regarding your main obstacle:... Finally, the participants were introduced to the forming of an if-then plan. The participants had to make their own if-then plan by transferring their obstacle to the if-part and the wish-fulfilling action to the then-part of the plan. Afterward, they were instructed to repeat the if-then plan several times. Instructions read: 1. Please think about how you could act to overcome or prevent your obstacle in the next weeks. How can you act to fulfill your wish? Please briefly describe this wish-fulfilling action. My action to fulfill my wish in the next weeks by overcoming my obstacle is:... 2. An important tool for your stress management strategy is the forming of if-then plans. If-then plans link your obstacle with an action that is instrumental to realizing your wish. They have the following format: 3. ***If*** *the obstacle arises*, ***then*** *I will execute the following wish-fulfilling action!* 4. Please formulate your own if-then plan now by linking your previously mentioned obstacle with your chosen wish-fulfilling action: I f \... \[ Please enter your previously mentioned obstacle here \] o c c u r s , then I will \... \[ Please enter the previously mentioned wish-fulfilling acyion here \] ! Now imagine this if-then plan and go through it in your mind! The participants were then instructed to perform this four step strategy in their mind on a daily basis at work for the following 3 weeks. To facilitate this, the MCII strategy was summarized in the form of four questions: 1. What is the best possible outcome today of my wish to have less stress? 2. What is the main obstacle today to fulfilling this wish? 3. How can I act to overcome this obstacle? 4. What is my if-then plan today? Finally, participants were asked to think about a calm situation and quiet moment in their daily work that would be ideal for them to go through this exercise in their mind. The participants were then asked to set a personal goal for the following 3 weeks: "I will use the stress management strategy once a day!" To summarize how MCII is done, participants received a practice sheet listing the four steps of the MCII exercise. This practice sheet was sent to them again via email by the experimenter twice, once after 1 week and then after 2 weeks. #### IIMCII group The instructions given to the IIMCII participants was identical to those received by the MCII participants with the following difference: The participants in the IIMCII condition were not only assigned to set a goal of using MCII every day for the next 3 weeks (i.e., "I will use the stress management strategy once a day!") but in addition to make an if-then plan in their mind that specified when and where the daily exercise should take place. The instructions for making this plan read: Please write down a calm situation and time most suitable to performing this exercise: \[ Please enter the chosen situation and time \] Now make an if-then plan in which you specify that you will perform this exercise once a day in this calm situation and time over the next 3 weeks: I f \[ Please enter the chosen situation and time \] occurs, then I will do this exercise! Please repeat this if-then plan several times in your mind. As with the MCII participants, the IIMCII participants then also received a practice sheet listing the four steps of MCII. This practice sheet was also sent to IIMCII participants a week as well as 2 weeks later. #### Control group The participants of the control group were only requested to answer the following three questions: What does stress mean for me? What does less stress mean for me? Why do I want less stress? In addition, they had to specify a personal wish with respect to achieving less stress at their work place in the next 3 weeks, and what exactly that wish implied. ### Assessment of the Dependent Variables Three weeks later, participants of all three conditions received an email from the experimenter with a link to the follow-up website at *soscisurvey.de.* When opening this website, participants received instructions to fill out the three questionnaires used at baseline within the next couple of days. The questionnaires were presented in the order used at baseline: perceived stress, work engagement, and then physical symptoms. Again, the internal consistency for each of the three questionnaires was high; Cronbach's α of 0.95, 0.95, and 0.85, respectively, were observed; composite reliability (CR) was 0.86, 0.96, and 0.78. Results ======= Descriptive Analyses -------------------- As perceived stress and physical symptoms both speak directly to the participants' stress level, we *z*-transformed and combined the 20 items of the PSQ-20 and the 10 items of the BOSS II to an overall stress index. Internal consistency was high at both T1, Cronbach's α = 0.93, and T2, Cronbach's α = 0.95; composite reliability (CR) was 0.78 at T1 and 0.83 at T2. At Time 1, overall stress and work engagement did not differ across conditions, *p*s \> 0.14 (see **Table [1](#T1){ref-type="table"}** for means and standard deviations of all dependent variables). ###### Means (standard deviations) of the dependent variables. Overall stress Work engagement --------- ---------------- ----------------- -------------- ------------- ------------- Overall 0.02 (0.58) 0.00 (0.64) 4.23 (1.20) 4.24 (1.26) Control 0.16 (0.65) 0.22 (0.73) 4.06 (1.23) 4.03 (1.40) IIMCII -0.04 (0.41) -0.05 (0.46) 4.22 (1.18) 4.11 (1.01) MCII -0.09 (0.61) -0.20 (0.63) 4.43 (1.21) 4.63 (1.27) Change of Overall Stress ------------------------ ### Statistical Analysis An analysis of covariance was used to assess whether participants of the MCII group or the IIMCII group show less stress after 3 weeks than participants of the control group. Change of overall stress was analyzed with a univariate ANCOVA, adjusting for overall stress at Time 1. We followed up this ANCOVA with two-tailed *post hoc* pairwise comparisons testing for differences between the MCII group and the control group as well as the IIMCII group, and between the IIMCII and the control group. Due to the dropout rate we then performed an intention to treat analysis with all 129 participants who answered the baseline questionnaire. The missing values of 24 participants at Time 2 were replaced by their respective baseline value. ### Statistical Results We observed an almost significant difference between conditions with respect to the change in overall stress, *F*(2,101) = 2.85, *p* = 0.062, $\eta_{p}^{2}$ = 0.053. Pairwise comparisons revealed less stress in the MCII group as compared to the control group, *t*(101) = 2.39, *p* = 0.019, $\eta_{p}^{2}$ = 0.053, 95% CI \[-0.348, -0.032\]. We neither found a difference between the MCII group and the IIMCII group, *t*(101) = 1.30, *p* = 0.198, $\eta_{p}^{2}$ = 0.016, 95% CI \[-0.264, 0.055\] nor between the IIMCII and the control group, *t*(101) = 1.09, *p* = 0.277, $\eta_{p}^{2}$ = 0.012, 95% CI \[-0.242, 0.070\] (see **Table [2](#T2){ref-type="table"}** and **Figure [4](#F4){ref-type="fig"}**). Bonferroni-Holm adjustment of the *p*-values yielded the following results: *p* = 0.057, *p* = 0.396, and *p* = 0.396, respectively. {#F4} ###### Comparing the three conditions regarding overall stress and work engagement. *F* *p* $\eta_{p}^{2}$ *t* *p* $\eta_{p}^{2}$ 95% CI ----------------- ------ ------- ---------------- -------------------- ------ ------- ---------------- -------- -------- Overall stress 2.85 0.062 0.053 MCII vs. Control 2.39 0.019 0.053 -0.348 -0.032 MCII vs. IIMCII 1.30 0.198 0.016 -0.264 0.055 IIMCII vs. Control 1.09 0.277 0.012 -0.242 0.070 Work engagement 2.98 0.055 0.056 MCII vs. Control 2.02 0.046 0.039 0.006 0.564 MCII vs. IIMCII 2.22 0.029 0.047 0.034 0.601 IIMCII vs. Control 0.23 0.816 0.001 -0.309 0.244 In our intention to treat analysis, we observed the same pattern of results between conditions with respect to change in overall stress, *F*(2,125) = 2.87, *p* = 0.060, $\eta_{p}^{2}$ = 0.044. Again, pairwise comparisons revealed less stress in the MCII group as compared to the control group, *t*(125) = 2.39, *p* = 0.018, $\eta_{p}^{2}$ = 0.044, 95% CI \[-0.285, -0.027\]. Again, we neither found a difference between the MCII group and the IIMCII group, *t*(125) = 1.32, *p* = 0.189, $\eta_{p}^{2}$ = 0.014, 95% CI \[-0.219, 0.044\] nor between the IIMCII and the control group, *t*(125) = 1.06, *p* = 0.292, $\eta_{\mathit{p}}^{2}$ = 0.009, 95% CI \[-0.196, 0.059\]. Bonferroni--Holm adjustment of the *p*-values yielded the following results: *p* = 0.054, *p* = 0.378, and *p* = 0.378, respectively. Change of Work Engagement ------------------------- ### Statistical Analysis Work engagement is often influenced by working conditions. To adjust for different working conditions, we coded reported working condition with 0 = *no shift work*, 1 = *either shift work or night shifts*, and 2 = *shift work as well as night shifts*. Working condition correlated significantly with change of work engagement, *r* = 0.20, *p* \< 0.05. To control for the influences of unfavorable working conditions on work engagement ([@B40]), working condition was incorporated in the analysis as a covariate. Change of work engagement was analyzed with a univariate ANCOVA, adjusting for work engagement at Time 1 and working condition. As we have done in our analysis regarding change in overall stress, we followed up this ANCOVA with two-tailed *post hoc* pairwise comparisons testing for differences between the MCII group and the control group as well as the IIMCII group, and between the IIMCII and the control group. Finally, we conducted the intention to treat analysis just like we have done it with respect to change in overall stress. ### Statistical Results We found an almost significant difference in work engagement, *F*(2,100) = 2.98, *p* = 0.055, $\eta_{p}^{2}$ = 0.056 between conditions. Pairwise comparisons revealed (1) more work engagement in the MCII group than in the IIMCII group, *t*(100) = 2.22, *p* = 0.029, $\eta_{p}^{2}$ = 0.047, 95% CI \[0.034, 0.601\], as well as (2) in the control group, *t*(100) = 2.02, *p* = 0.046, $\eta_{p}^{2}$ = 0.039, 95% CI \[0.006, 0.564\]. As with overall stress, we found no difference between the IIMCII group and the control group, *t*(100) = 0.23, *p* = 0.816, $\eta_{p}^{2}$ = 0.001, 95% CI \[-0.309, 0.244\] (see **Table [2](#T2){ref-type="table"}** and **Figure [5](#F5){ref-type="fig"}**). Bonferroni--Holm adjustment of the *p*-values yielded the following results: *p* = 0.087, *p* = 0.092, and *p* = 0.816, respectively. ![Work engagement after 3 weeks assessed by a questionnaire from [@B35] in the three groups of the intervention study, controlling for baseline scores.](fpsyg-09-00838-g005){#F5} The intention to treat analysis showed the same difference between conditions with respect to change in work engagement, *F*(2,124) = 3.24, *p* = 0.042, $\eta_{p}^{2}$ = 0.050. Pairwise comparisons revealed more work engagement in the MCII group as compared to the IIMCII group, *t*(124) = 2.33, *p* = 0.021, $\eta_{p}^{2}$ = 0.042, 95% CI \[0.041, 0.504\] and to the control group, *t*(124) = 2.09, *p* = 0.039, $\eta_{p}^{2}$ = 0.034, 95% CI \[0.012, 0.466\]. We did not find a difference between the IIMCII and the control group, *t*(124) = 0.29, *p* = 0.770, $\eta_{p}^{2}$ = 0.001, 95% CI \[-0.258, 0.191\]. Bonferroni--Holm adjustment of the *p*-values yielded the following results: *p* = 0.063, *p* = 0.078, and *p* = 0.770, respectively. Discussion and Outlook ====================== The MCII intervention reduced the participating nurses' overall stress and improved work engagement as compared to a no-treatment control group. Even though we measured stress reduction and work engagement after a relatively short period of time (i.e., already after 3 weeks), we observed improvements on both of these variables measured reliably with established questionnaires covering the different aspects of perceived stress (i.e., worries, tension, reduced joy, heightened demands as well as typical physical symptoms such as headaches, back pain) and work engagement (i.e., vigor, dedication, and absorption). We did not give the nurses hints on what they can do to reduce their worries, tensions, demands, and headaches, and what can be done to increase vigor, dedication, and absorption with their work. Rather, we left it up to the nurses to detect their stress-reducing wish and best outcome, and to identify what hinders them to realize their wishes for less stress at the work place, and what they want to do to overcome these obstacles. Also, we did not send daily reminders to the nurses to perform the MCII exercise. We just told them at the end of teaching the exercise to perform MCII on a daily basis. That is, the whole intervention encompassed one session at the outset in which the participants had to go through the MCII exercise in writing. From then on, the participants were on their own, engaging in the MCII exercise in their mind over the subsequent 3 weeks. In order to ensure high fidelity, our instructions in the three conditions were adapted from past MCII intervention research (e.g., summary by [@B29]; [@B27], [@B28]). These intervention studies pertained to a variety of samples reaching from children at risk for ADHD to healthy adults and to patients recovering from stroke ([@B25]). In addition, our participants received three practice sheets that depicted the steps of the intervention in the respective condition (**Figures [2](#F2){ref-type="fig"}**, **[3](#F3){ref-type="fig"}**). The first practice sheet was provided right after the intervention, the other two 1 and 2 weeks later. This way we tried to ensure that participants closely followed the instructions they had received in the intervention session. The only additional information they received were two emails from the experimenter, 1 and 2 weeks after the training session, containing a practice sheet depicting the four steps of MCII. The IIMCII group did not show the hypothesized pattern of results: It did not show the expected heightened reduction in stress level and it did not show the enhanced work engagement as compared to the MCII group; not even differences to the control group emerged. We established the IIMCII group as we were worried that healthcare providers simply may be too overburdened in their daily work to act on the assigned goal to engage in MCII on a regular basis. Simply thinking of a quiet time and place in which the MCII exercise could be performed may not suffice. Rather, it might need an additional implementation intention that specifies this critical time and place in the if-part of the plan. After all, implementation intentions have been found to help people remember to act on their goals, in particular when they suffer from high cognitive load (e.g., [@B4]). So why did we not find the expected effects on stress reduction and heightened work engagement in the IIMCII group? Three answers come to mind: First, asking IIMCII participants right after the training session has ended to make an if-then plan at which time and place they want to engage in the daily MCII exercise may have limited IIMCII participants to this very time and place for the rest of the 3 weeks. In contrast, the MCII group only specified the goal to use MCII on a daily basis at a quiet time and place. This should allow for more flexibility in case unexpected or even better opportunities for performing MCII open up (i.e., more appropriate quiet times and places). Second, action control by implementation intentions is known to be characterized by features of automaticity (i.e., it is fast, runs off outside of awareness, and is effortless). Given that mental contrasting is an effortful cognitive procedure that requires slowness and imagination, the mindset associated with acting on the implementation intention to use a pre-specified time and place may undermine the effortful cognitive procedure of mental contrasting. Third, and most importantly, we cannot assume that participants had a high commitment to use the MCII exercise on a daily basis. They have heard about it for the first time just before -- when they had started the experiment -- and thus they might have not yet fully trusted its effects. Having no strong commitment to use MCII on a daily basis should attenuate the effectiveness of forming implementation intentions created in the service of using MCII. As described before, the prerequisite of implementation intentions is the commitment for the overarching goal ([@B37]). Strengths and Weaknesses ------------------------ This procedure is more parsimonious than classic stress interventions. The latter focus on intensively training participants in performing specific stress reduction strategies (e.g., emotion control strategies such as distancing oneself from an overly arousing event, conflict resolution strategies, and relaxation techniques), and this is commonly done via close supervision by the interventionist. Accordingly, the present study could have added a further control group that used a typical classic stress intervention (e.g., directing attention away from work problems). It would have been interesting to see whether the time and cost effective MCII exercise lives up to the success of classic intervention programs. Also, from a methodological point of view the present study would have benefited from assessing data on participants' adherence to the MCII instructions and the frequency and context of participants using MCII. How often did the participants in the MCII and IIMCII perform the MCII exercise, and did the frequency of performing it correlate positively with stress reduction? Future research will have to make the necessary changes to the design (add a further control group) and assessment (add a measure of degree of adherence). Future Research --------------- Future research may address the question of how MCII effects can be strengthened even further. One way of doing so pertains to combining MCII with other types of interventions. For instance, it might be worthwhile to add a self-affirmation exercise ([@B13]) or a mindfulness exercise ([@B39]) prior to having participants perform MCII. Both of these exercises are known to reduce self-defensiveness and thus it seems possible that participants, in the aftermath of these exercises, find it easier to detect truly personal wishes, outcomes, and obstacles when engaging in MCII, which in turn should make MCII more effective. Second, future research may address the question of the content area for which MCII is taught and practiced. Our MCII exercise solely focused on wishes reducing stress at the work place (though we do not know whether participants had used MCII for other wishes as well). Our teaching did not explicitly include wishes on how to recover from work stress at home and during leisure time. It is, however, a person's work-life balance that is increasingly recognized as an important factor for living a physically healthy life ([@B40]). Future research may ask participants to perform MCII on integrative wishes regarding work-life balance. Third, we did not provide information on effective stress-reduction strategies as discovered in stress research (e.g., effective emotion regulation strategies at work or relaxation techniques at home). Such a combination of providing relevant information first before participants engage in the MCII exercise has been used effectively in changing people's eating behavior and making people more physically active (e.g., [@B41], [@B42]; [@B25]). Future research may extend the focus of MCII interventions regarding the work place to relaxation after work, and by adding an information session on how to best reduce stress at work and at home prior to asking participants to engage in the MCII exercise. Fourth, it might be valuable to learn more about the nurses' stress-relevant personal attributes before the MCII instructions are given. For instance, if one nurse is suffering from emotion control problems whereas another is overly impulsive, knowing about this by handing out relevant personality questionnaires at the outset of the study would allow gearing their wishes to more effectively cope with these shortcomings. Recent research shows that MCII can indeed be targeted toward affective aspects of one's outcome and obstacle ([@B32]) or toward enhancing reflection over impulsivity ([@B2]). And finally, there remains the question of how one can make MCII effects long-lasting. One approach pertains to going beyond helping the nurses to individually cope with their stressful situations but encouraging them to get involved with efforts to change the team or organization they are working for into a less stressful one. This may often be not possible, but if successful it heightens the chances that the reduced level of personal stress the nurses have achieved via improved self-regulation will stay stable over time ([@B14]). A more practical and feasible route to achieving long-lasting stress reduction, however, might be to turn the use of MCII into a habit. Anecdotally, we find that using MCII by facilitating behavior change is rewarding. If so, open access to guidelines of how to use MCII during daily life might be a first step to establish the habitual use of MCII. Indeed, the guidelines of how to use MCII -- which colloquially has been given the acronym of WOOP for Wish, Outcome, Obstacle, Plan -- is available to the public on a website^[1](#fn01){ref-type="fn"}^ and in an app called WOOP. Ethics Statement ================ This study was carried out in accordance with the recommendations of the ethics committee of the University of Konstanz. All subjects gave written informed consent in accordance with the Declaration of Helsinki. Author Contributions ==================== PG, GO, and CF jointly designed the study and CF conducted it. DM analyzed the data and prepared the presentation of the results. All authors worked on preparing a first draft and on revising the manuscript into its final form. 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. **Funding.** This work was supported by the German Research Foundation \[DFG, GO 387/ 14-3\]. [www.woopmylife.org](http://www.woopmylife.org) [^1]: Edited by: Jesus de la Fuente, University of Almería, Spain [^2]: Reviewed by: Agnes Von Wyl, Zurich University of Applied Sciences, Switzerland; Jesús Nicasio García Sánchez, Universidad de León, Spain [^3]: This article was submitted to Educational Psychology, a section of the journal Frontiers in Psychology

Supramolecular self-assembly on surfaces is an active area of research aimed toward the realization of a range of (functional) two-dimensional (2D) crystals.^[@ref1]−[@ref4]^ Control over the network morphology and defect density is a necessity for precise engineering of these materials. Formation of high-quality interfaces is important for engineering organic thin-film devices with π-conjugated molecules, where the relative orientation of molecules is crucial to enable efficient charge transport across electrodes.^[@ref5],[@ref6]^ Great efforts are taken to improve the quality of 2D crystals,^[@ref7]−[@ref9]^ by avoiding molecular defects and limiting domain boundaries. A host of variables are available to achieve this, either intrinsically (molecular design and symmetry)^[@ref10]^ or extrinsically (temperature,^[@ref11],[@ref12]^ solute concentration,^[@ref13]^ type of solvent,^[@ref14],[@ref15]^ capillary flow,^[@ref16]^ and substrate).^[@ref17]^ Despite the tunability of these parameters, the time scales and complexity of molecular recognition and assembly processes have thus far limited a complete comprehension and control of the 2D crystal formation. As such, studies targeting the fundamental understanding of kinetic and thermodynamic parameters within these systems are rare.^[@ref11],[@ref18]−[@ref20]^ An ability to manipulate nucleation and growth processes would afford greater control over network formation and possibly a more detailed molecular level understanding of assembly principles. One approach to study the elementary aspects of self-assembly consists of compartmentalizing the assembling molecules into laterally confined areas. Beebe *etal*. demonstrated that by heating highly oriented pyrolytic graphite (HOPG) in the presence of oxygen, thermally oxidized pits can be etched into the surface.^[@ref21]^ These pits were proven to be useful tools for molecular self-assembly confinement studies; however their preparation methods limit control over the shape, dimensionality, or orientation with respect to the symmetry axes of the underlying substrate lattice.^[@ref22],[@ref23]^ Other studies within confined spaces have been reported by de Oteyza *etal*., who showed that under ultra-high-vacuum conditions, self-assembly of diindenoperylene on Cu (111) step edges leads to the formation of a long-range ordered structure with co-directionally oriented molecules.^[@ref24]^ Alignment is observed along the confined step edges of the Ag (877) vicinal surface for terminal alkyne-functionalized polyphenylene building blocks.^[@ref25]^ Studies targeting confinement at very small length scales^[@ref500]^ (a few nanometers) have demonstrated the role of electronic surface states using well-organized adsorbates.^[@ref26]−[@ref28]^ Nevertheless, alternative approaches that yield well-defined corrals for studying molecular self-assembly under nanoconfinement may afford greater insight into the fundamentals of recognition and growth processes. In this work, we create confined spaces (nanocorrals) with geometric size, shape, and orientational control on covalently modified HOPG surfaces. Corral production involves two separate steps. First, HOPG is covalently modified by aryl radicals that are electrochemically (EC) generated from an aryl diazonium precursor species ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a).^[@ref29]−[@ref600]^ In the second step, the tip of a scanning tunneling microscope (STM) is used to precisely remove the covalently bound species from the surface, a process that is referred to as "nanoshaving" ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}b). Subsequent self-assembly investigations ([Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}c) within these well-defined nanocorrals are carried out at the liquid--solid interface using a saturated solution of 10,12-pentacosadiynoic acid (PCDA) in 1-phenyloctane (1-PO). The PCDA molecule contains typical structural and functional features such as alkyl chains and a carboxylic acid headgroup that are known to foster supramolecular self-assembly at the liquid--solid interface on graphite.^[@ref34]−[@ref37]^ Previous work involving the STM tip-mediated removal of a physisorbed self-assembled network containing porphyrin derivatives on HOPG has been reported by Hipps *etal*.^[@ref33]^ Local areas of bare graphite were obtained, however diffusion is believed to result in poorly defined structures. In contrast, the covalent linkage of aryls to the graphite surface as presented here creates a rigid platform for the formation of nanocorral patterns. Due to the localized nature of the probe and the rigidity of the grafted molecules on the surface, this nanoshaving process has the potential to create dimensionally controlled, stable corrals down to a 10 × 10 nm^2^ scale. {#fig1} {#fig2} Here, nanoshaving and subsequent liquid--solid PCDA self-assembly experiments are performed both *in situ* (corral creation as well as PCDA self-assembly occurs in a single step at the liquid--solid interface) and *ex situ* (corral creation under dry conditions followed by PCDA self-assembly at the liquid--solid interface). Corrals formed *ex situ* show negligible impact on the PCDA domain size and shape. However, the gradual exposure of the graphite surface from *in situ* nanoshaving is shown to have a pronounced effect on 2D crystal nucleation and growth processes through the observation of fewer yet much larger domains. Using different *in situ* corrals, we demonstrate how nucleation and growth events involved in self-assembly can be placed under geometric and kinetic constraints to preferentially bias resultant PCDA lamellae orientation. This knowledge is used to design and experimentally demonstrate "seeding and growth" under nanoconfinement conditions. Results and Discussion {#sec2} ====================== Supramolecular Self-Assembly of PCDA on Unconfined HOPG {#sec2.1} ------------------------------------------------------- First, supramolecular self-assembly of PCDA on freshly cleaved HOPG was evaluated as an internal benchmark for comparison studies between the self-assemblies of PCDA^[@ref38]−[@ref40]^ on open terraces and laterally confined terraces. As shown in [Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}, PCDA readily self-assembles at the 1-PO/HOPG interface. Individual molecules are clearly distinguished in the STM images. Self-assembly of the molecules is largely promoted by two types of intermolecular interactions: directional hydrogen bonding between the carboxylic acid moieties and van der Waals interactions between neighboring alkyl chains. The supramolecular packing consists of rows of parallel stacked molecules, called lamellae. The brightest features in the STM image provided in [Figure [2](#fig2){ref-type="fig"}](#fig2){ref-type="fig"}b correspond to the diacetylene units, which are tilted with respect to alkyls. The darkest rows correspond to the meeting of the methyl ends of the molecules, whereas the rows of medium contrast are assigned to the regions of hydrogen bonding. The substrate epitaxy is reflected in a selective orientation of the alkyl chains along the high-symmetry axes of graphite. The angle, θ, between the lamellae direction and the graphite axis with commensurate alkyls was measured to be 86.7 ± 0.3°. This results in two equivalent twinned domains for each of the three high-symmetry axes of graphite: six domain orientations in total ([Figure S2](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). {#fig3} Nanocorral Creation {#sec2.2} ------------------- To create the nanocorrals, the HOPG surface was first covalently modified with a dense monolayer of aryl species using cyclic voltammetry. The electrochemical modification of HOPG is carried out in aqueous solutions containing diazonium cations generated *in situ* from stable aniline precursors.^[@ref41]^ Incorporating sterically hindering substituents on the aniline precursor limits the grafting to monolayer species.^[@ref29],[@ref31]^ When concentrations greater than 2 mM are used, a high density of monolayer aryl species can be covalently bound or "grafted" to the surface. After diazotization, the mechanism of covalent attachment involves reduction of the diazonium cation, expulsion of N~2~, and formation of aryl radicals. These unstable aryl radicals subsequently react with the HOPG surface (see [Figure [1](#fig1){ref-type="fig"}](#fig1){ref-type="fig"}a) through a radical addition reaction. The result is a newly formed sp^3^-hybridized carbon--carbon bond that covalently links the aryl species to the HOPG surface. Further experimental details involving the preparation of the covalently grafted HOPG are provided in the [Methods](#sec4){ref-type="other"} section and [Supporting Information](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf). Nanocorrals inside the covalently modified HOPG surface are created with detailed control over the nanoshaving process using the Keysight PicoLITH 2.1 software package. Different shapes for the creation of nanocorrals are first designed within the PicoLITH software ([Figure S3](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). The software then rasters the STM tip in the desired areas to nanoshave the corrals. The fast nanoshaving direction moves *across* the corral to define the width, and the slow direction moves *downward* to define the height of the corral. During this nanoshaving process, the STM is operating in high current (typically 200 pA) and low sample bias (−1.0 mV). These scan parameters bring the tip in close proximity to the graphite substrate, such that the covalently bound aryls are degrafted and removed from the surface, *i*.*e*., nanoshaving. While the mechanism for this degrafting process is still poorly understood, the HOPG surface carbon atoms have previously been demonstrated to revert back to their original sp^2^ hybridization.^[@ref29]^ Detailed investigations targeting the mechanism of nanoshaving and its dependence on nanoshaving parameters (voltage, tunneling current, and tip speed) are currently ongoing. In most cases, the tip speed for nanoshaving was set to 400 nm/s. For all of the corrals in this work, the distance covered along the slow nanoshaving direction by each raster sweep varies only slightly (between 4.0 and 8.0 Å), depending on the corral size. Thus, the time required to complete the nanoshaving depends on the area of the corral. Assembly attempts on HOPG surfaces modified with a high density of grafted species showed no self-assembly of PCDA on top or in between the grafted features. Rather, the only observation made by STM imaging is the grafted surface itself ([Figure S4](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). In contrast, we have recently shown that when the grafting density is lower, perturbations within the molecular self-assembly are observed. In these cases, the grafted species act as barriers and impede self-assembly.^[@ref42]^ *Ex Situ vs in Situ* Creation of Nanocorrals and Its Impact on the Assembly Process {#sec2.3} ----------------------------------------------------------------------------------- To initially establish an understanding of the influence of lateral confinement on the self-assembly of PCDA, experiments were performed on bare (unconfined) HOPG, *ex situ* corrals, and *in situ* corrals ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}). *Ex situ* corrals were created by nanoshaving the grafted (dry) surface, then exposing the corrals to a 1-phenyloctane solution of PCDA. Within *ex situ* corrals, the entire nanocorral surface is exposed and self-assembly can occur in all areas. The self-assembly of PCDA inside an *ex situ* nanocorral (180 × 180 nm^2^) was found to be nearly identical to that observed on the open terraces of unmodified HOPG. In both cases, domains of varying size and orientation are observed ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}a,b). This is a strong indication that multiple nucleation events occurred. Despite the fluid environment above the corral, ripening into single domains was not observed. Similar to observations made by Beebe *etal*., the corral boundaries do not appear to favor nucleation, as unstable/no self-assembly is observed at the edges ([Figure S5](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)).^[@ref21]^ Empty surface regions are also found in the assembly of PCDA on bare HOPG ([Figure S5](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Occasionally, features indicative of PCDA multilayer lamellae are observed inside the *ex situ* nanocorrals. STM topography measurements show the suspected multilayer structure is ∼0.5 Å above the monolayer lamellae ([Figure S6](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)).^[@ref43]^ Similar multilayer structures from PCDA are also observed on the bare HOPG ([Figure S6](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Additional STM images of the *ex situ* corrals and of PCDA self-assembly within *ex situ* nanocorrals can be found in [Figures S7 and S8](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf), respectively. From our observations it appears that assembly within relatively large *ex situ* corrals proceeds similar to that observed on bare HOPG. *In situ* corrals were created by carrying out nanoshaving directly in the presence of a 1-phenyloctane solution containing PCDA. The self-assembly of PCDA was found to be drastically different in corrals that are created *in situ* when compared to bare HOPG or *ex situ* corrals. *In situ* corrals almost always show a single large domain of PCDA lamellae. Importantly, the *in situ* created nanocorrals always show the presence of PCDA networks inside them. Thus, the self-assembly of PCDA within the *in situ* generated corral occurs within the time needed for the nanoshaving and the subsequent imaging scan (∼2 min). In comparison, the time lapse between the *ex situ* generation of corrals and the subsequent imaging of the PCDA network was on the order of 1 h. Thus, the assembly behavior of PCDA in the *in situ* nanocorrals cannot be the result of a ripening effect occurring over time. Rather, the act of nanoshaving inside the PCDA solution must govern the observed molecular alignment within the nanocorrals. *In Situ* Nanocorral Effects: Nanoshaving Orientation with Respect to Graphite {#sec2.4} ------------------------------------------------------------------------------ The impact of lateral confinement on PCDA self-assembly within the *in situ* nanocorrals was further probed by rotating the nanoshaving orientation with respect to the underlying hexagonal graphite lattice. Importantly, the slow nanoshaving direction for the corral in [Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}c is approximately *orthogonal* to a major axis of graphite, colored red. This creates a situation where three possible general orientations of PCDA lamellae can occur ([Figure S9](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Parallel, +60° diagonal, and −60° diagonal domain orientations of the PCDA lamellae can occur within the corral as defined by the angle between the direction of slow nanoshaving and the lamellae direction. The assembly within the corral appearing in [Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}c is thus characterized as a parallel PCDA domain. In this case, the long axis of the molecule is parallel with the fast nanoshaving direction. When the slow nanoshaving direction is rotated so that it runs *along* a major graphite axis, a different result is observed ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}d). In this case, the lamellae align diagonally in the nanocorral. Assemblies that demonstrate this behavior are referred to as diagonal PCDA domains. Epitaxial matching and assembly constraints in the corral require the molecules to arrange in either a diagonal (+30° or −30°) or a perpendicular manner; a parallel alignment is substrate registry forbidden ([Figure S9](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). The assembly of PCDA displays a domain twinning effect as a result of the 2D chirality ([Figure S2](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). It was not possible to selectively induce a specific 2D chirality within any particular nanocorral. The angle separating the equivalent mirror structures is reasoned to be too acute for such a selection. Therefore, the twinned domains are treated equally, creating only three possible categories for PCDA lamellae alignment for any specific nanoshaving direction ([Figure S9](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). A detailed explanation of the nanoshaving process, the registry forbidden structures, and the assignment of domain behavior is further described in the [Supporting Information](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf). For a statistical understanding of the dependence of the PCDA lamellae orientation on the nanoshaving orientation with respect to graphite, more than 50 *in situ* corrals of each type (orthogonal to and along a graphite major symmetry axis) were created. For faster analysis, the size of the nanocorral was reduced to ∼50 × 50 nm^2^. When the square corrals are created with the slow nanoshaving direction orthogonal to a major graphite axis, the PCDA lamellae overwhelmingly (90.4%) assemble into parallel PCDA domains. Compare this to the statistical outcome of 33.3%. When a diagonal domain exists, they were typically accompanied by a separate parallel domain ([Figure S9](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Alternatively, when the slow nanoshaving direction was parallel to a major graphite axis, the reverse was observed. Diagonal PCDA lamellae are slightly preferred with a nearly equal population of + and -- domains at 38.3% and 37.9%, respectively. Perpendicular PCDA domains occupy only 23.8% ([Figure S9](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). These results demonstrate that the orientation of slow nanoshaving direction with respect to the symmetry axes of graphite can be used to influence the orientation of PCDA lamellae within *in situ* corrals. *In Situ* Nanocorral Effects: Corral Size Impact on PCDA Alignment {#sec2.5} ------------------------------------------------------------------ After establishing the difference between *ex situ* and *in situ* corrals, as well as orientational effects relative to HOPG symmetry axes, we went on to investigate the impact of the *in situ* nanocorral size on PCDA self-assembly. Corrals of varying sizes were created to evaluate how the nanocorral size affects the PCDA lamellae directionality bias observed from *in situ* nanoshaving. The nanocorrals were created by orienting the slow nanoshaving direction orthogonal to the graphite lattice, allowing the PCDA lamellae to preferentially align parallel. The results of this size-dependent study are shown in [Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}a. In the four larger corrals (average size: 67 × 67 nm^2^, 46 × 46 nm^2^, 37 × 37 nm^2^, 28 × 28 nm^2^) the results are consistent with those presented in [Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}c, where the lamellae align along the slow nanoshaving direction. For the two smaller corrals (20 × 20 nm^2^ and 11 × 11 nm^2^) the PCDA lamellae are not aligned along the slow nanoshaving direction (inset [Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}a). Instead, the lamellae are aligned diagonally with respect to the slow nanoshaving direction. A statistical representation of the dependence of lamellae orientation on the corral size is presented with a red-colored trend in [Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}b. Statistics were acquired by counting the number of domains present in each corral for each particular orientation. In general, the alignment of PCDA lamellae in larger corrals is consistent until the corral size approaches ∼30 nm. Below this size, the tendency of the lamellae to preferentially orient parallel decreases sharply and eventually it becomes close to the statistically unbiased outcome within corrals of ∼10 nm lateral size ([Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}b). Thus, the orientational bias of the PCDA lamellae within *in situ* generated corrals is dependent on the size of the corral. {#fig4} The corral size also influences the probability of self-assembly when significant confinement restraints exist ([Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}b, black). Self-assembly is frequently observed inside square corrals of 20 × 20 nm^2^ or larger. However, the propensity of the PCDA molecules to self-assemble inside corrals of 10 × 10 nm^2^ is reduced to 65% ([Figure [4](#fig4){ref-type="fig"}](#fig4){ref-type="fig"}b, black). A representative image of these empty corrals is shown in [Figure S10](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf). This observation is consistent with previous studies, where decreasing the size of the confined space results in prolonged molecular ordering time scales.^[@ref21]^ It is important to note that only the image immediately after nanocorral creation was used for the analysis of the assembly statistics. Time-dependent studies under these strict nanoconfinement conditions are expected to produce other valuable insights. Nevertheless, the smallest corral created was suitable for assembly and STM imaging of an ordered PCDA domain. Corrals below 10 nm could not reliably be created with size and shape control. At such a small scale, the corral size begins to approach the dimension of the covalently bound aryls that confine the assembly. *In Situ* Nanocorral Effects: Geometric Shape Impact on PCDA Alignment {#sec2.6} ---------------------------------------------------------------------- To further investigate the impact of *in situ* nanocorral formation on the resultant PCDA assembly, the geometric shape of the corral was changed. Again, the slow nanoshaving direction was carefully chosen to promote parallel PCDA lamellae alignment. Nanocorrals in the shape of a square, a circle, a downward facing triangle, and an upward facing triangle are shown in [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}. The lamellae of PCDA align with preference along the slow nanoshaving direction for the square, the circle, and the downward triangle. A statistical analysis with more than 30 corrals of each particular shape is shown comparing the relative number of domains with a particular orientation ([Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}). With 86% preference, the square fosters the formation of lamellae parallel to the slow nanoshaving direction. Similarly, the PCDA lamellae in circular corrals and downward triangles show a pronounced tendency to align (80% and 83%, respectively) along the slow nanoshaving direction. The upward triangle, on the other hand, dramatically reduces this alignment trend with only 66% alignment along the slow nanoshaving direction. A (+) diagonally aligned domain inside the upright triangle, occupying 22% of the total population, is shown in [Figure [5](#fig5){ref-type="fig"}](#fig5){ref-type="fig"}d. The triangle study demonstrates how the method (top-to-bottom or bottom-to-top) of nanoshaving a particular shape can alter the tendency of PCDA lamellae to preferentially align. More importantly, this study demonstrates that the initial stages of the corral formation must play an important role in directing the PCDA self-assembly. Hence, the seeding and growth of the assembly of PCDA must take place as the HOPG surface is being gradually exposed. {#fig5} Origin of Preferential PCDA Lamellae Alignment {#sec2.7} ---------------------------------------------- The question thus arises: how does the nanoshaving of *in situ* nanocorrals give rise to preferential parallel PCDA lamellae alignment? Standard self-assembly of molecules at the liquid--solid interface proceeds in three distinct stages: nucleation, free growth, and ripening.^[@ref44]^ When molecules are deposited on freshly cleaved HOPG, all of these processes contribute to the final observed structure. We propose, however, that the gradual revelation of the surface from *in situ* nanoshaving at the liquid--solid interface gives rise to a combination of geometric and kinetic constraints that affect the standard processes by which self-assembly typically occurs. To demonstrate the concept of geometric constraints, we focus on the variations observed in the corral size study. Importantly, the nanoshaving of each individual corral was completed, starting with the largest corral, before moving on to the next corral. The shaving rate is consistently held at 400 nm/s. Thus, the time required for nanoshaving each particular corral decreases proportionally with decreasing size of the corral. This also creates a situation where the geometric limitations placed on the system at the initial stages of corral formation are very different for each particular corral. At early stages (\<2 s) of nanoshaving, the largest corrals exist as high aspect ratio rectangles where assembly can occur. On the other hand, the aspect ratio for the smaller nanocorrals is drastically lower for the same nanoshaving time ([Figure S11](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). While the total nanoshaving area between the corrals is the same at these early times, the geometric constraints are very different ([Figure S11](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Ultimately, these different constraint conditions are believed to bias the nucleation and growth processes to impact the observed size-dependent preferential alignment. The behavior observed in the upright and downward triangles provides strong evidence that kinetic factors created by the nanoshaving process are also influential in determining the available PCDA orientations for adsorption. The downward facing triangle is first nanoshaved along the base, whereas the upright triangle is nanoshaved from the apex first. This places the assembly of PCDA within both corrals under different kinetic constraints that limit the possible orientations for adsorption of PCDA ([Figure S12](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). During the initial stages of corral formation, nanoshaving from the triangular base restricts the options for PCDA assembly to one particular orientation with respect to the graphite lattice ([Figure S13](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). However, nanoshaving from the apex of the triangle opens the diagonal axes of the graphite surface at the same rate as the parallel orientation ([Figure S13](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)). Thus, the "method" of nanoshaving the triangle creates kinetic limitations that can restrict the possible PCDA assembly orientations. It is important to note that the same geometric constraints that exist within the varying sized corrals also play an important role in the triangular assembly as well. The corral size series is also influenced by similar kinetic factors that restrict assembly to one particular orientation for a period of time. Given the same nanoshaving rate, the time scales of these kinetic restrictions decrease proportionally with the size of the corral. Through these kinetic and geometric constraints, *in situ* nanoshaving afforded control over nucleation and growth processes to preferentially select specific orientations of the PCDA lamellae. Finally, the presence of the potentially large STM tip in close proximity to the newly created corrals places an additional kinetic constraint on molecular adsorption events. Steric blocking effects from the tip are expected to hinder molecular diffusion in the vicinity of the tip. During this time, the PCDA molecules may experience limited access to the exposed surface. This effect is particularly impactful when the dimensionality of the corral is similar to that of the tip. Such restrictions to PCDA self-assembly place additional kinetic constraints on the system that likely impact the observed orientation of the PCDA lamellae within small corrals (below 30 × 30 nm^2^). Similarly, for the upward facing triangle steric hindrance of the tip might delay PCDA self-assembly within the initial degrafted area. Assembly alignment observed in previous work from Beebe *etal*. inside *ex situ* fabricated corrals likely occurs *via* different pathways.^[@ref21]^ In their work, formation of a film using a pure liquid compound resulted in assembly inside etched corrals, as well as on the open terraces of HOPG. Ripening effects and registry impacts from the assembly on the open terraces and elevated layers are believed to contribute to the observed alignment.^[@ref21]^ In our work, corrals of similar size to the work from Beebe *etal*. fabricated under dry conditions (*ex situ*) and later topped with a PCDA solution in 1-phenyloctane yielded multiple domain orientations. This suggests that multiple nucleation events are responsible for the final observed structure, which is not surprising given the relatively large size of the corral (∼180 × 180 nm^2^) when compared to the average domain size on bare HOPG ([Figure [3](#fig3){ref-type="fig"}](#fig3){ref-type="fig"}a). Since no assembly is observed on top of the covalently modified graphite, our assemblies cannot experience the same registry/ripening effects from open terrace assemblies. Related to Beebe's work, we observe poor ordering at the corral boundaries, also suggesting an assembly inhibition from corral perimeters. We also never observed full molecular disassembly and reorientation within the corrals, but dynamics were occasionally observed. Beebe's methods allow for confinement studies on elevated terraces. With our corrals, such an investigation is not possible. Size, shape, and orientation control on the graphite surface can, however, be achieved. This control motivated us to probe nucleation and growth events with this highly precise method for confining molecules. *In Situ* Nanocorral Effects: Seeding and Growth Demonstration {#sec2.8} -------------------------------------------------------------- Using the constraints inherent to *in situ* nanocorrals, we designed an experiment to directly demonstrate the basic crystallization concept^[@ref45],[@ref46]^ of "seeding and growth" under nanoconfinement conditions. The experiment involves the creation of a circular seed corral followed by two other rectangular arm corrals made in the presence of PCDA to allow additional growth. First, the expected alignment is shown by creating the individual corrals separately ([Figure [6](#fig6){ref-type="fig"}](#fig6){ref-type="fig"}a). In this case, the alignment of the domain of PCDA lamellae inside the circular seed is parallel. Similarly, the alignment of PCDA inside the rectangular arm is also classified as parallel (given the slow nanoshaving direction), but the orientation with respect to the circular seed is clearly different. This result is expected given the nanoshaving orientation of the corrals with respect to the major graphite axes. {#fig6} In a separate experiment, the rectangular arms are overlapped with the circular seed when the nanoshaving is performed. Importantly, the circular seed is nanoshaved first followed by the rectangular arms. In this case, the lamellae within the rectangular corrals are in registry with the lamellae inside the circular seed corral ([Figure [6](#fig6){ref-type="fig"}](#fig6){ref-type="fig"}b). The PCDA lamellae are observed to run continuously from the seed into the arms. Hence, once the seed domain was formed, the PCDA growth upon exposure of additional free graphite (rectangular arm) is observed in the same direction as the lamellae observed in the seed. Occasionally, new nucleation events occur inside the arms and create domain boundaries ([Figure [6](#fig6){ref-type="fig"}](#fig6){ref-type="fig"}b, top rectangle). The propensity for these new nucleation events within the rectangular corrals is expected to be dependent on the seed (and arm) size, shape, orientation, and degrafting parameters. These parameters along with the rate of surface exposure and concentration impact on 2D crystal nucleation and growth are currently under more detailed investigation. Conclusions {#sec3} =========== In conclusion, we have revealed several general effects involved in laterally confined molecular self-assembly occurring inside nanocorrals at a liquid--solid interface. Corrals of various size, shape, and orientation are fabricated by *in situ* STM nanoshaving on covalently modified HOPG surfaces. Statistical analysis of these results demonstrates the importance of geometric and kinetic constraints on the resultant assembly orientation. Confinement effects during the initial stages of nanocorral formation are shown to strongly affect nucleation and growth processes. Furthermore, the confinement constraints placed on the assembly were employed for control over "seeding and growth" processes under nanoconfinement conditions. In the future coupling these corral methods with fast scanning probe microscopy techniques may allow direct observation of molecular assembly events.^[@ref20]^ Confinement impacts on switching, chirality, multicomponent assembly, phase transitions, and stimuli-responsive networks are expected to emerge.^[@ref1]^ The concept of nanoshaving can possibly be extended to surfaces other than HOPG, given that diazonium modification on both conducting metals and semiconducting silicon surfaces is well documented.^[@ref32],[@ref47]−[@ref53]^ The nanocorral approach and confinement principles described herein are expected to be invaluable tools for future studies on the thermodynamic and kinetic parameters involved in crystallization. Methods {#sec4} ======= STM Experiments {#sec4.1} --------------- All experiments were performed at room temperature (20--22 °C) using a PicoLE (Keysight) or Molecular Imaging STM system operating in constant-current mode at the 1-phenyloctane/HOPG interface. STM tips were prepared by mechanical cutting from Pt/Ir wire (80%/20%, diameter 0.25 mm). For self-assembly imaging, a saturated solution of 10,12-pentacosadiynoic acid (≥97%) in 1-phenyloctane (98%) was drop-casted on the surface of a freshly cleaved (or covalently modified) HOPG substrate (grade ZYB, Advanced Ceramics Inc., Cleveland, OH, USA). Both PCDA and 1-phenyloctane were purchased from Sigma-Aldrich and used without further purification. The reported unit cell parameters of PCDA on HOPG are averaged values deduced from examination of six images that have been corrected for drift using recorded graphite images under the same conditions except for *V*~s~ = −1 mV and *I*~t~ = 200 pA. Nanoshaving was performed using the PicoLITH v.2.1 software. All images were processed using the Scanning Probe Imaging Processor (SPIP) software (Image Metrology ApS). Imaging parameters are indicated in the figure captions and are denoted by *V*~s~ for the sample bias and *I*~t~ for the tunneling current. Covalent Modification of HOPG {#sec4.2} ----------------------------- Electrochemical grafting of 3,5-bis-*tert*-butylbenzenediazonium (3,5-TBD) was performed using cyclic voltammetry in aqueous solutions. Due to the low stability of 3,5-TBD, it was generated *in situ* from the corresponding aniline compound. To this end, ≥3 mg of 3,5-bis-*tert*-butylaniline (3,5-TBA) was dissolved in 5 mL of aqueous hydrochloric acid (50 mM), and 100 μL of aqueous NaNO~2~ (0.1 M) was added for activation of the diazotization reaction. The solution was gently shaken for 1.5 min before injection into the electrochemical cell. A lab-built single-compartment three-electrode cell, with a working electrode area of 50.3 mm^2^, Pt wire counter, and Ag/AgCl/3.0 M NaCl reference electrode was used to carry out the cyclic voltammetry. Prior to each experiment, the HOPG working electrode was freshly cleaved using Scotch tape. A typical current--voltage diagram obtained during cyclic voltammetry is shown in [Figure S1](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf). Every experiment consisted of three voltage sweeps. After modification, the HOPG samples were rinsed with Milli-Q water to remove any physisorbed material from the surface. 3,5-Bis-*tert*-butylaniline (98%) and analytical grade hydrochloric acid were purchased from Sigma-Aldrich and used without further purification. High-purity water (Milli-Q, Millipore, 18.2 MΩ cm, TOC \< 3 ppb) was used for preparation of the aqueous solutions. All electrochemical measurements were performed using an Autolab PGSTAT101 potentiostat (Metrohm_Autolab BV, The Netherlands). The Supporting Information is available free of charge on the [ACS Publications website](http://pubs.acs.org) at DOI: [10.1021/acsnano.6b05954](http://pubs.acs.org/doi/abs/10.1021/acsnano.6b05954).Extra STM and cyclic voltammetry results; schematics illustrating geometric and kinetic constraints during *in situ* self-assembly ([PDF](http://pubs.acs.org/doi/suppl/10.1021/acsnano.6b05954/suppl_file/nn6b05954_si_001.pdf)) Supplementary Material ====================== ###### nn6b05954_si_001.pdf ^†^ School of Physics & Astronomy, University of Nottingham, Nottingham, NG7 2RD, United Kingdom. The authors declare no competing financial interest. This work has received funding from the Fund of Scientific Research--Flanders (FWO), KU Leuven Internal Funds, Belgian Federal Science Policy Office (IAP-7/05), and the European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013)/ERC Grant Agreement No. 340324. L.V. thanks the FWO for the fellowship received. B.H. acknowledges the Belgian American Educational Foundation and the FWO.

{#sp1 .479}

Most neurologic research on emotion dysregulation focuses on pseudobulbar affect, which occurs in a broad range of disorders with diffuse or poorly characterized pathology^[@R1][@R2][@R5]^ often implicating the brainstem and cerebellum.^[@R1],[@R6][@R7][@R8]^ The fact that pseudobulbar affect has not been associated with focal limbic damage is consistent with its being understood as "a disorder of emotional expression rather than a primary disturbance of feelings."^[@R7]^ Autoimmune limbic encephalitis (aLE) is associated with the subacute onset of amnesia and seizures and high T2-signal (acute MRI) in the limbic system, especially the hippocampus. Patients often respond satisfactorily to immunosuppressive therapy,^[@R9]^ although many develop hippocampal atrophy and residual cognitive impairment.^[@R10],[@R11]^ While behavioral/psychiatric symptoms may occur acutely,^[@R12]^ persisting problems with readily provoked tearfulness are only mentioned in passing,^[@R13][@R14][@R15]^ and we have encountered complaints of such symptoms among many of our patients. We aimed to determine the so far unexplored nature and neural correlates of pathologic tearfulness following aLE in a uniquely large cohort of patients (n = 38). We investigated its relationships with demographic and clinical data, self-reported measures of emotion regulation, and performance on neuropsychological tests. We hypothesized that it is associated with abnormalities in the hippocampus, the amygdala, hippocampal-diencephalic-cingulate networks, and cerebro-ponto-cerebellar loops: aLE results in relatively focal hippocampal atrophy,^[@R11],[@R16]^ and the limbic system is involved in emotion processing.^[@R17][@R18][@R19]^ Amygdala abnormalities are sometimes observed^[@R20]^ and have been associated with abnormal autonomic arousal.^[@R21]^ Furthermore, the hippocampus is embedded within broader hippocampal-diencephalic-cingulate networks supporting emotion regulation.^[@R22]^ We have recently shown abnormalities in this extended circuitry in aLE.^[@R23],[@R24]^ Finally, in a prominent pathophysiologic account, emotion dysregulation in pseudobulbar affect was caused by disruption to cerebro-ponto-cerebellar pathways,^[@R7]^ with which the hippocampus communicates.^[@R25]^ Methods {#s1} ======= Standard protocol approvals, registrations, and patient consents {#s1-1} ---------------------------------------------------------------- Ethical approval was received from the South Central Oxford Research Ethics Committee (REC no. 08/H0606/133). All participants provided written informed consent according to the Declaration of Helsinki. Participants {#s1-2} ------------ We report data relating to pathologic tearfulness in 38 patients with aLE (26 male, 12 female; median age at research MRI 63.06 years; interquartile range \[IQR\] 16.16 years)^[@R24]^ after the acute stage of the disease (median 5.41; IQR 5.36 years since symptom onset). All patients were fluent in English (37 native speakers; 1 non-native speaker) and had undergone MRI at the time of initial clinical presentation as well as neuropsychological assessment at the Russell Cairns Unit, Oxford, UK (2013--2018). All patients had been diagnosed with aLE according to established diagnostic criteria^[@R26]^: (a) subacute symptom onset suggesting involvement of the limbic system; (b) bilateral abnormalities restricted within the medial temporal lobes (MTLs) on T2-weighted MRI; (c) CSF pleocytosis (white blood cells \>5/mm^3^) or slow-wave/epileptic activity involving the temporal cortex (EEG); (d) exclusion of alternative causes (e.g., CNS infections/drug toxicity/stroke/Creutzfeldt-Jakob disease/Kleine-Levin syndrome, mitochondrial/neoplastic/epileptic/rheumatologic disorders, septic/metabolic encephalopathy); (e) antibodies against cell-surface/synaptic/onconeural proteins. Criteria (a--d) are required for a diagnosis of definite limbic encephalitis, unless, in the absence of one of (a--c), criterion (e) is satisfied.^[@R26]^ A total of 34 of 38 patients satisfied the criteria for a diagnosis of definite aLE; the remaining 4/38 had been diagnosed with aLE, meeting criteria (a, b, d), but not (e). No data could be recovered regarding (c). In 28/38 patients, an aLE-associated autoantibody was identified. A total of 10/38 patients demonstrated the clinical profile of aLE with no identified antibody; such cases are well-recognized^[@R27]^ and are generally thought to involve antibodies not detected in clinical practice at the time of screening. No patient presented with positive PCR testing for herpes simplex virus or with anti-NMDAR encephalitis.^[@R28]^ Two of 38 patients had neoplastic lesions, thought to be the triggers for their autoimmune disorder, which were treated and were in full remission at the time of study participation. A total of 31/38 patients had been treated acutely with immunotherapy (e.g., plasma exchange, IV or oral prednisolone). A total of 34/38 patients had shown abnormal hippocampal signal, volume, or diffusion on clinical MRI conducted acutely. Six of 38 patients showed amygdala abnormalities, 1 in the parahippocampal cortex, 1 in the entorhinal cortex, 4 patients had mild microangiopathic changes in keeping with their age, and 1 patient showed extra-MTL abnormalities (bright caudate). No acute abnormalities were detected in 4/38 patients, who nonetheless demonstrated clinical features characteristic of aLE; 35/38 patients had presented acutely with seizures. Moreover, patients had no history of previous neurologic or psychiatric disorder that could have resulted in cognitive impairment. They were assessed by a single neurologist (CRB) prior to study inclusion. Their (acute) clinical and (postacute) neuropsychological details have been presented previously.^[@R24]^ Healthy controls (HCs) were recruited through the Oxford Project to Investigate Memory and Ageing (OPTIMA) and through local advertisement. Neuropsychological profile {#s1-3} -------------------------- Postacutely, all patients and 57 HCs (38 men, 19 women; age at assessment: median, 61.50; IQR 17.26 years; HCs vs patients: male:female ratio: χ^2^ = 0.032, *p* = 0.858; age at assessment: *U* = 933.50, *p* = 0.258) underwent neuropsychological assessment. Patients showed preserved executive function, above-average premorbid intelligence, and spared motor, executive, and visuospatial function, but impaired episodic memory.^[@R24]^ Review of medical records {#s1-4} ------------------------- Details were extracted from medical records and interviews with the patients and caregivers using a standard proforma regarding clinical history, acute aLE presentation, and subsequent clinical course of each patient (age at symptom onset; presenting symptoms; premorbid and acute phase depression, anxiety, agitation, obsessionality, or hallucinations; seizure occurrence/recency; delay between symptom onset and start of treatment; autoantibody type; past/present immunotherapy, antiepileptics, and antidepressants). Emotion regulation assessment {#s1-5} ----------------------------- ### Questionnaires {#s1-5-1} In order to assess patients\' pathologic tearfulness, we administered the Center for Neurologic Study--Lability Scale (CNS-LS),^[@R29]^ a 7-item questionnaire comprising 2 subscales ("labile crying" and "labile laughter"). A series of additional questionnaires were administered to examine the relationship of patients\' pathologic tearfulness with (1) anxiety and depression (Hospital Anxiety and Depression Scale \[HADS\]),^[@R30]^ (2) impulsivity (Barratt Impulsiveness Scale \[BIS\]),^[@R31]^ (3) irritability (Irritability Questionnaire \[IRQ\]),^[@R32]^ and (4) empathy (Cambridge Behaviour Scale \[CBS\]^[@R33]^; [docs.autismresearchcentre.com/tests/EQ40_ScoringKey.doc](http://docs.autismresearchcentre.com/tests/EQ40_ScoringKey.doc)). A total of 25/38 patients and 29/57 HCs completed and returned those self-administered questionnaires by post. Patients filled out the questionnaires together with their next of kin or family members. Patients who completed the emotion regulation questionnaires did not differ from those who did not in the following: (1) neuropsychological tests in which patients showed preserved group-level performance^[@R24]^ (all *p*s, *p*~corr~ ≥0.340); (2) tests in which patients showed group-level impairment^[@R24]^ (all *p*s, *p*~corr~ ≥0.304); (3) clinical/demographic variables (see previous section; all *p*s, *p*~corr~ ≥0.999); (4) volumes of manually delineated MTL structures and automatically delineated subcortical structures in which there was no group-level atrophy^[@R24]^ (all *p*s, *p*~corr~ ≥0.209); and (5) structural/functional brain abnormalities identified at group level^[@R24]^ (all *p*s, *p*~corr~ ≥0.260). We also assessed the relationship of patients\' emotion regulation with their memory by conducting bivariate correlation analyses between memory test scores and scores on questionnaires of emotion regulation in which patients showed impairment compared with HCs. ### Self-report (clinical interview) {#s1-5-2} In a complementary approach, and since the CNS-LS may not be sensitive to the symptoms described by our patients, we dichotomized the cohort according to clinical complaint at interview. The interviewer was blind to patients\' responses in the above questionnaires. Patients and their family members were asked whether there had been instances of "labile laughter" or labile crying, and to provide examples from their daily life. ### Relationship with demographic, clinical, and neuropsychological profiles {#s1-5-3} We conducted (1) bivariate correlations of CNS-LS scores with continuous variables and independent-samples comparisons on CNS-LS scores for binary variables across patients; and (2) comparisons among HCs, patients with, and patients without pathologic tearfulness (independent-samples comparisons for continuous variables, χ^2^ tests for binary variables). Brain imaging {#s1-6} ------------- ### Structural MRI {#s1-6-1} We acquired 3D T1-weighted images using a magnetization-prepared rapid gradient echo sequence (echo time 4.7 ms, repetition time 2,040 ms, 8° flip angle, field of view 192 mm, voxel size 1 × 1 × 1 mm). All 38 patients (26 male, 12 female; age at imaging: median 63.06; IQR 16.06 years) underwent structural brain imaging, along with 67 HCs (35 recruited by the Memory and Amnesia Project \[MAP\]; 32 datasets were made available through OPTIMA; 40 male, 27 female; age at imaging: median 64.70; IQR 19.87 years; HCs vs patients: M:F ratio: χ^2^ = 0.79, *p* = 0.374; age at imaging: *U* = 1,239.5; *p* = 0.825) (methods also in [@R24]). ### Volumetry {#s1-6-2} MTL structures (left/right hippocampus, amygdala, temporopolar, entorhinal, perirhinal, and parahippocampal cortices) were manually delineated in native space (protocol: [ndcn.ox.ac.uk/files/research/segmentation_protocol_medial_temporal_lobes.pdf](https://www.ndcn.ox.ac.uk/files/research/segmentation_protocol_medial_temporal_lobes.pdf)).^[@R23],[@R24]^ Subcortical structures (brainstem, left/right thalamus, caudate nucleus, putamen, pallidum, nucleus accumbens) were automatically delineated using FSL-FIRST (v.6.0; <https://fsl.fmrib.ox.ac.uk/fsl/fslwiki>).^[@R34]^ ### Whole-brain voxel-based morphometry (VBM) {#s1-6-3} In order to identify gray matter (GM) volume reduction in our patient group at a whole-brain level, the T1-weighted MRIs were analyzed with VBM, conducted using Statistical Parametric Mapping software (SPM12; [fil.ion.ucl.ac.uk/spm/software/spm12](http://www.fil.ion.ucl.ac.uk/spm/software/spm12)) in MATLAB R2017b. Images were examined for scanner artefacts and reoriented to have the same point of origin (anterior commissure) and spatial orientation. They were then bias-corrected to remove intensity nonuniformities, and segmented into GM, white matter (WM), and CSF with the unified segmentation procedure. The diffeomorphic anatomical registration through the exponentiated lie algebra (DARTEL) toolbox was applied to participants\' GM, WM, and CSF to refine intersubject registration, and study-specific GM templates were generated.^[@R35]^ After affine registration of the GM DARTEL templates to the tissue probability maps in Montreal Neurologic Institute (MNI) space, nonlinear warping of GM images was performed to this template in MNI space. Voxel values in the tissue maps were modulated by the Jacobian determinant (calculated during spatial normalization), with modulated GM images reflecting tissue volume. These images (voxel size: 1 mm^3^ isotropic) were smoothed using a Gaussian filter of 8 mm full width at half maximum (FWHM). We compared GM volume between groups (HCs \> patients; between-subject covariates: age, sex, total intracranial volume \[TIV\], study \[MAP, OPTIMA\]). We report clusters surviving family-wise error (FWE) correction (*p* \< 0.05) at peak voxel level over *p* \< 0.001 (uncorrected), as well as clusters surviving correction for nonstationary smoothness^[@R36]^ and FWE correction for cluster size (*p* \< 0.05). Volumes (calculated from manual/automated segmentation, or the volume reflected by each VBM cluster) that showed reduction in patients at whole-group level were residualized against age, sex, TIV, and study and entered in bivariate correlation analyses with scores in questionnaires of emotion regulation. We also contrasted patients with pathologic tearfulness with those without and HCs across all volumes delineated as well as across the whole brain (VBM). ### Resting-state fMRI (rsfMRI) {#s1-6-4} Whole-brain blood oxygenation level--dependent (BOLD)--weighted fMRI data were acquired (gradient echo echoplanar imaging (EPI) sequence; 180 volumes; slice thickness 3.5 mm, echo time 28 ms, repetition time 2,410 ms, 89° flip angle, field of view 192 mm, voxel size 3 × 3 × 3.5 mm). Participants were instructed to lie still, not to fall asleep, to keep their eyes open, and to watch a fixation cross presented on the in-scanner projector. A total of 35 of 38 patients (3 datasets discarded due to acquisition errors or movement; 24 men, 11 women; median age at imaging, 61.45; IQR 15.85 years) underwent rsfMRI, along with 32 HCs (3 datasets discarded due to movement or acquisition errors; only structural MRIs were available for the HCs that were made available through OPTIMA; 23 men, 9 women; median age 55.71; IQR 17.18 years; HCs vs patients: male:female ratio: χ^2^ = 0.087; *p* = 0.768; age at imaging: *U* = 425.00; *p* = 0.091). ### Preprocessing {#s1-6-5} EPIs were spatially realigned and slice time--corrected. Structural MRIs were coregistered to the EPIs, segmented and normalized along with EPIs in MNI space, followed by motion outlier detection (artifact detection tools--based scrubbing). Denoising, including the anatomical component-based correction method (CompCor), was employed to remove sources of noise in the BOLD time series data, deriving principal components from WM and CSF. WM, CSF, and the 6 movement measures were included as first-level nuisance covariates. A temporal bandpass filter (0.01--0.1 Hz) was applied to this residual BOLD signal, in order to remove motion artefacts and physiologic and other artefactual effects. Images were smoothed using a Gaussian filter (8 mm FWHM). Resting-state amplitude of low frequency fluctuations and functional connectivity {#s1-7} --------------------------------------------------------------------------------- We further examined whether resting-state abnormalities in the local amplitude of low-frequency fluctuations (rsALFF) and hippocampal functional connectivity (rsFC) were associated with pathologic tearfulness. Preprocessing, rsALFF, and rsFC analyses were conducted using the CONN toolbox v. 18.a ([nitrc.org/projects/conn](https://www.nitrc.org/projects/conn)).^[@R37]^ ### rsFC: connectome--multivariate pattern analysis (MVPA) {#s1-7-1} In order to identify seed regions for post hoc seed-to-voxel connectivity analyses in a data-driven fashion, we used MVPA as implemented in the connectome-MVPA CONN toolbox. MVPA assesses the multivariate pattern of pairwise connections between voxels across the entire brain by means of a principal component analysis (PCA) separately for each voxel that characterizes its rsFC with the rest of the brain. In the first PCA step, separately for each participant, a default number of 64 PCA components were retained while characterizing each participant\'s voxel-to-voxel correlation structure. The resulting component scores were stored as first-level voxel-to-voxel covariance matrices for each participant. In the second PCA step, separately for each voxel and jointly across participants, the 7 strongest components were retained from a PCA decomposition of the between-subjects variability in seed-to-voxel connectivity maps between this voxel and the rest of the brain, according to a conventionally employed ratio of 1:10 between the number of components extracted and the number of participants (n = 67). Second-level analyses were then conducted in order to test for group differences in whole-brain connectivity (*F* test across all MVPA components), comparing for each voxel the component scores between the 2 groups (HCs \< \> patients; between-subjects covariates: age, sex). The results for each voxel reflected between-group differences in rsFC between this voxel and the rest of the brain. ### rsFC: seed-to-voxel connectivity analysis {#s1-7-2} We followed up the MVPA with post hoc analyses to determine specific connectivity patterns. We thus conducted a whole-brain seed-to-voxel analysis, seeding from the regions identified from the MVPA contrast (HCs \< \> patients), in order to assess connectivity between those regions and the rest of the brain. ### Resting-state hemodynamic activity: rsALFF {#s1-7-3} Along with rsFC, we also examined local abnormalities in the intensity of slow spontaneous fluctuations of hemodynamic activity at rest across the whole brain, using an analysis of rsALFF, that is, the total power within the frequency range between 0.01 and 0.1 Hz, indexing the strength of low-frequency oscillations. All the rsfMRI analyses involved age and sex as between-subjects covariates. Statistical parametrical connectivity maps were thresholded at a voxel level of *p* \< 0.001 and FWE-corrected (*p* \< 0.05) at cluster or peak level. The mean values in clusters of reduced rsALFF or rsFC in patients at whole-group level, as compared with HCs, were residualized against age and sex across participants and then entered in bivariate correlations with scores in questionnaires of emotion regulation. We also contrasted patients with pathologic tearfulness against the rest of the patients and HCs across the whole brain. Statistical analysis {#s1-8} -------------------- Statistical (nonimaging) analyses were conducted using SPSS (v. 25.0, SPSS Inc., Chicago, IL). Significance values were corrected for multiple testing with the Holm-Bonferroni sequential correction method (*p*~corr~). We used the Levene test to assess variance homogeneity and the Shapiro-Wilk test to assess normal distribution. When normal distribution was violated (and log-transformation did not suffice), nonparametric tests were employed. Parametric (Student or Welch *t* tests) and nonparametric tests (Mann-Whitney *U*) were used appropriately for independent-samples comparisons. For comparisons among 3 groups, univariate analyses of variance (ANOVAs) or Kruskal-Wallis *H* tests were used appropriately, and post hoc comparisons between groups were Bonferroni-corrected. Pearson *r* and Spearman ρ were used appropriately to examine correlations between questionnaire scores and other measures of interest. We used multiple stepwise linear regression analysis (default α level of 0.05 for entry to model and 0.1 for removal) to assess the proportion of the variance of patients\' scores (questionnaires on emotion regulation) that could be explained by brain abnormalities. Data availability {#s1-9} ----------------- The deidentified data will be available and shared by request for purposes of replicating procedures and results. Results {#s2} ======= Emotion regulation assessment {#s2-1} ----------------------------- ### Questionnaires: patients vs HCs {#s2-1-1} Patients scored higher than HCs for labile crying (CNS-LS) (*t* = −2.79, *p*~corr~ = 0.049) but not for laughter (*t* = 0.44, *p*~corr~ \>0.999; 2-way mixed-effects ANOVA: group: *F* = 2.49, *p* = 0.12; emotion: *F* = 1.81, *p* = 0.19; group × emotion: *F* = 5.73, *p* = 0.02). They did not differ from HCs in their empathy quotient (CBS) (*t* = 0.79, *p*~corr~ \>0.999), in irritability (IRQ) (frequency: *U* = 235, *p*~corr~ = 0.450; intensity: *U* = 240.5, *p*~corr~ = 0.450), or anxiety (HADS: *U* = 517, *p*~corr~ = 0.090). They scored higher in the planning (*t* = −4.97, *p*~corr~ \<0.0005) and attention facets (*t* = −3.90, *p*~corr~ = 0.002), but not in the motor facet for impulsiveness (BIS) (*t* = 0.38, *p*~corr~ \>0.999). They also scored higher for depression (HADS) (*U* = 357.5, *p*~corr~ \< 0.0005), although no patient scored within the severe range (also noted in [@R24]). Scores for labile crying did not correlate across patients with impulsiveness (attention and planning facets: ρ = 0.12, *p* = 0.60), depression (ρ = 0.24, *p* = 0.28), or any memory score in which patients had shown impaired performance as compared with HCs (all *p*s, *p*~corr~ ≥0.240), and were not associated with any demographic or clinical variables examined (all *p*s, *p*~corr~ ≥0.440). ### Self-report: patients with vs patients without pathologic tearfulness and HCs {#s2-1-2} In a research-oriented clinical interview, 19 of 38 patients were identified as presenting with pathologic tearfulness. In particular, they reported being moved to tears easily by relatively minor stimuli in a manner at odds with their premorbid state ([table 1](#T1){ref-type="table"}). The other 19 reported never having experienced such instances. No patient reported experiencing episodes of labile laughter. ###### Patients\' self-reports  The majority of patients and their family members reported specific triggers of such reactions, including sad stories on the news and witnessing other people crying ([table 2](#T2){ref-type="table"}). ###### Triggers of pathologic tearfulness as identified by patients and their family members  Patients with pathologic tearfulness did not differ from the rest of the participants in any demographic or clinical details or in episodic memory impairment, depression, or impulsiveness. Moreover, they did not differ from the rest of the patients or HCs in premorbid intelligence, vocabulary, semantic knowledge, visuomotor function or executive function, anxiety, empathy, or irritability. Among all the tests and questionnaires administered, the only one in which they scored differently from both the rest of the patients and HCs was CNS-LS ([table 3](#T3){ref-type="table"}). ###### Neuropsychological, clinical, and demographic profile of patients with vs patients without pathologic tearfulness vs healthy controls (HCs)     Structure/function--Behavior relationships {#s2-2} ------------------------------------------ ### Questionnaires: correlations with CNS-LS and BIS scores {#s2-2-1} In our previous study,^[@R24]^ we identified a series of brain abnormalities (n = 13) that patients showed at group level: volume reduction in the left and right hippocampus, captured by both VBM and manual delineation; volume reduction in the anterior-mediodorsal thalamus and right dorsolateral thalamus (VBM) and the left thalamus (automated delineation), as well as the right entorhinal cortex (manual delineation); reduced right hippocampal rsFC with left hippocampus, ventral-posterior posteromedial cortex (posterior cingulate, retrosplenial cortex, and precuneus; Brodmann area \[BA\] 23, 31), and medial prefrontal cortex (BA 10, 32, 24); and reduced rsALFF in the posterior cingulate and the precuneus (BA 23, 31). We entered the mean values of the clusters that reflected these abnormalities (residualized against age and sex for functional abnormalities, as well as TIV and study \[MAP, OPTIMA\] for volumes) in bivariate correlations with CNS-LS scores for labile crying. Patients\' scores correlated strongly with their reduced right hippocampal rsFC with the ventral-posterior posteromedial cortex (*r* = −0.61, *p*~corr~ = 0.030; rest of *p*s, *p*~corr~ ≥0.190; [figure 1A](#F1){ref-type="fig"}). No such correlations were identified with impulsivity (BIS attention and planning facets) or depression (HADS) scores, even at uncorrected levels (\|*r*\| \<0.29, *p* \> 0.18). {#F1} Moreover, given our a priori hypotheses on the role of the hippocampus in emotion dysregulation, we examined, at uncorrected levels, correlations with CNS-LS scores. Right anterior hippocampal volume correlated negatively across patients with scores for labile crying (*r* = −0.52, *p* = 0.01; left anterior, right/left posterior hippocampus: *p* \> 0.07; [figure 1B](#F1){ref-type="fig"}), but not with right hippocampal rsFC with the posteromedial cortex (*r* = 0.33, *p* = 0.05). When these 2 factors were entered as independent variables in a multiple stepwise linear regression, the analysis was terminated in 2 steps, with the right hippocampal--posteromedial cortical rsFC included in the first model as a predictor of patients\' scores of labile crying (*F* = 12.50, *p* = 0.002; *R*^2^ = 0.37), and with the volume of the right anterior hippocampus entered in the model in the second step (*F* = 9.55, *p* = 0.001; *R*^2^ = 0.49). No volumetric correlation of any hippocampal segment was identified with impulsivity or depression scores (\|*r*\| \<0.25, *p* \> 0.19). Self-report: patients with vs without pathologic tearfulness and HCs {#s2-3} -------------------------------------------------------------------- ### Structural abnormalities {#s2-3-1} Patients with pathologic tearfulness did not differ from the rest of the patients in any MTL or subcortical volumes (*p*~corr~ ≥ 0.350). Nevertheless, a whole-brain VBM analysis disclosed lower volume for these patients relative to the other 2 groups in the right anterior hippocampus, the right cerebellar hemispheric HVI/HVIIa Crus I, and the left fusiform gyrus (BA 37; [figure 2](#F2){ref-type="fig"}). ![Structural abnormalities in patients with pathologic tearfulness\ Results of whole-brain voxel-based morphometry (VBM) on modulated gray matter (GM) (reflecting GM volume). Contrast: healthy controls (HCs) and patients without pathologic tearfulness \> patients with pathologic tearfulness; between-subjects nuisance regressors: age, sex, total intracranial volume (TIV), and study (Memory and Amnesia Project \[MAP\], Oxford Project To Investigate Memory and Ageing \[OPTIMA\]). (A) Right anterior hippocampus: kE = 19, *p* family-wise error-corrected (FWE) = 0.037; peak voxel: *t* = 4.79; x = 34, y = −12, z = −17. (B) Left fusiform gyrus/posterior portion of inferior temporal gyrus: kE = 17; *p* FWE = 0.038; peak voxel: *t* = 4.79; x = −44, y = −62, z = −5. (C) Right cerebellar hemispheric lobules VI/VIIa Crus I: kE = 23; *p* FWE = 0.042; peak voxel: *t* = 4.76; x = 24, y = −75, z = −18; clusters are displayed here at *p* \< 0.001 (unc) for display purposes, and survive FWE correction (*p* \< 0.05) at peak-voxel level over *p* \< 0.001 (unc) (minimum cluster volume: kE \> 10). The cerebellar cluster also survived correction for nonstationary smoothness and cluster size (p-FWE \< 0.05). Clusters are overlaid here on a diffeomorphic anatomical registration through exponentiated lie algebra GM template in Montreal Neurological Institute space (sagittal sections presented); heat bar represents *t* values; bar graphs display the average GM volume of each of those 3 clusters for the 3 different groups; error bars represent +1/−1 SEM. aHPC = anterior hippocampus; FG = fusiform gyrus; ITG = inferior temporal gyrus; kE = cluster size (number of voxels); R, L = right, left (hemisphere); z-res = mean values residualized against age, sex, study (MAP, OPTIMA), and TIV across participants.](NEUROLOGY2019010967FF2){#F2} ### Functional abnormalities {#s2-3-2} A connectome--MVPA analysis on rsFC across the whole brain identified a cluster in the right hippocampus as a region in which patients with pathologic tearfulness differed from the other 2 groups. We thus seeded from the right hippocampus in native space (unsmoothed timeseries), in order to identify regions with which these patients showed abnormal right hippocampal rsFC: they showed aberrantly increased right hippocampal rsFC with the right middle frontal gyrus (BA 9) and reduced rsFC with a region in the right posterior cingulate extending to the precuneus and lingual gyrus (BA 23, 18). Patients with pathologic tearfulness also showed aberrantly increased rsALFF in the left fusiform gyrus (BA 37) and the ventral pons, as well as reduced rsALFF in the right inferior parietal lobule (BA 39; [figure 3](#F3){ref-type="fig"}). {#F3} Discussion {#s3} ========== Our study is the first to investigate the nature and neural foundations of emotion dysregulation in a uniquely large, homogeneous cohort of patients after aLE, a nondegenerative neurologic syndrome characterized by primary limbic pathology. Clinical features and correlates of emotion dysregulation {#s3-1} --------------------------------------------------------- In particular, we describe a novel disorder of emotion regulation following aLE that is characterized by residual pathologic tearfulness. In our cohort, this was reported by 50% of patients. This symptom may be misdiagnosed as a manifestation of depression; for example, an indirect consequence of reduced quality of life due to memory impairment. If present alongside disinhibition and impulsiveness, it may otherwise be interpreted as a sign of a broader dysexecutive syndrome, continuous with that sometimes present in the acute stage of aLE.^[@R10]^ However, we showed that pathologic tearfulness was not associated with depression or impulsiveness, and occurred in the face of preserved executive function, and at normal levels of anxiety and irritability. Notably, no clinical or behavioral difference was detected between the patients with pathologic tearfulness and the equally sized subset with no such symptoms, apart from their scores on labile crying (CNS-LS). To our knowledge, this symptom has only been mentioned in passing in case or case series studies of aLE^[@R13][@R14][@R15]^ or in studies of larger yet less homogeneous cohorts of autoimmune encephalitis or epilepsy,^[@R38],[@R39]^ as "emotional lability," "mood lability," or "uncharacteristic tearfulness," with no further discussion of its clinical features and correlates. Direct comparisons with other patient groups such as temporal lobe epilepsy will be needed in future studies. Moreover, the profile of pathologic tearfulness observed in our patients with aLE is strikingly different from the syndrome of pseudobulbar affect seen in other neurologic conditions (e.g., amyotrophic lateral sclerosis, stroke, multiple sclerosis, Parkinson disease, Alzheimer disease, and traumatic brain injury^[@R1][@R2][@R5]^), where dramatic and debilitating bouts of laughing or crying occur often without any appropriately valanced trigger^[@R7]^ or congruence between the experience and expression of emotion.^[@R40]^ For instance, none of our patients presented with pathologic laughing. Furthermore, most patients who presented with pathologic tearfulness readily identified specific triggers that were congruent with their albeit exaggerated emotional responses. Many of these triggers pertained to situations that evoked empathic concern (e.g., children or animals in distress or acting affectionately). This may suggest that aberrantly increased empathy underlies the patients\' symptoms. There are, indeed, strong links between empathy and proneness to crying in the healthy population, which may suggest that increased empathy is associated with increased likelihood to experience distress, resulting in a higher crying proneness (see [@R41]). Whereas the CBS did not disclose abnormality in patients with pathologic tearfulness, it may lack sensitivity in capturing increased, rather than decreased, empathy. Likewise, while the CNS-LS represents the most broadly employed self-report measure of affective lability,^[@R29]^ more targeted instruments need to be employed, examining autonomic responses within the context of finer-grained behavioral tasks. This might lead to identification of similar symptoms in other neurologic disorders, such as temporal lobe epilepsy, where suggestive evidence has already been presented.^[@R42]^ Structural and functional correlates {#s3-2} ------------------------------------ In line with our hypotheses, we found correlates of pathologic tearfulness in the anterior hippocampus, the posterior cingulate cortex, the ventral pons, and the neocerebellum. Whether these abnormalities result directly from the acute, primary pathology of aLE or occur subsequently as a form of functional diaschisis or as a consequence of Wallerian degeneration^[@R23]^ remains to be determined (see also discussion in reference [@R12]). [Figure 4](#F4){ref-type="fig"} summarizes the insight our study provides on the impairments underlying pathologic tearfulness in aLE, based on the model in [@R7]. {#F4} Anterior hippocampal volume {#s3-3} --------------------------- Right anterior hippocampal atrophy was associated with pathologic tearfulness. The right anterior hippocampal volume correlated across patients with scores for labile crying (CNS-LS), and patients with pathologic tearfulness showed less volume in the right anterior hippocampus in a voxel-wise whole-brain analysis. That hippocampal lesions should be associated with pathologic tearfulness is consistent with the involvement of limbic circuitry in emotion processing,^[@R17]^ especially with the relationship between recurrent stress and hippocampal damage in nonhuman primates,^[@R19]^ as well as with hippocampal pathology in psychiatric disorders.^[@R43]^ In particular, the primate anterior hippocampus is the homologue of the rodent ventral hippocampus, which plays a role in negative affect, by virtue of its connectivity with the amygdala and the hypothalamus.^[@R18]^ However, manually delineated hippocampal volumes did not differ between patients with and those without pathologic tearfulness, suggesting that atrophy may be confined to specific regions within the anterior hippocampus, a possibility that could be explored using subfield volumetry in future studies. Hippocampal dysconnectivity with the posteromedial cortex {#s3-4} --------------------------------------------------------- Scores for labile crying (CNS-LS) strongly correlated with patients\' reduced right hippocampal rsFC with the ventral posteromedial cortex (posterior cingulate, retrosplenial cortex, and precuneus). Evidence from functional neuroimaging of healthy adults supports a role of this region in empathic concern for emotional suffering and admiring virtue.^[@R44]^ Aberrant perspective taking and empathy has also been reported in hippocampal patients.^[@R45],[@R46]^ Pontocerebellar abnormalities {#s3-5} ----------------------------- Volume reduction was also noted for patients with pathologic tearfulness in posterior portions of the right hemispheric cerebellar lobules VI/VIIa Crus I. These regions are embedded within the default mode network, which is fundamental for self-referential cognition.^[@R47]^ The cerebellum receives input from the basilar pons, and disruption of cortico-ponto-cerebellar circuits may lower the threshold for emotional expression.^[@R7]^ Moreover, anatomical and electrophysiologic work has recently disclosed evidence for cerebellar (lobules VI/VIIa Crus I)--hippocampal interactions, possibly via the pons.^[@R25]^ Consistent with these accounts, patients with pathologic tearfulness demonstrated aberrantly increased rsALFF in the ventral pons. The ventral pons relays input to the cerebellum from cortical regions including the parietal association cortices,^[@R48]^ where these patients showed reduced right hippocampal rsFC. Interestingly, pontine hemodynamic hyperactivity has been reported previously in a single case study of pathologic laughing,^[@R6]^ consistent with earlier reports of pathologic crying in cases of pontine myelinosis.^[@R8]^ Abnormalities in the inferior parietal lobule, fusiform, and middle frontal gyri {#s3-6} -------------------------------------------------------------------------------- Beyond the relationships that we had hypothesized, we also observed a series of unpredicted abnormalities associated with pathologic tearfulness: GM volume reduction and aberrantly increased rsALFF in the left fusiform gyrus, reduced rsALFF in the right inferior parietal lobule, and reduced rsFC between the right hippocampus and the right middle frontal gyrus. While activations in all of these regions have been repeatedly shown in self-face processing,^[@R49],[@R50]^ the aberrantly increased rsFC and rsALFF in patients with pathologic tearfulness require further investigation, as they may reflect compensatory or maladaptive mechanisms. Our study describes a novel disorder of emotion regulation following aLE that is characterized by residual pathologic tearfulness, is not related to low mood or cognitive impairment, and is associated with specific abnormalities within networks supporting emotion regulation. Clinicians need to be aware of the potential for such symptoms to develop after aLE and of the distress they can cause. Furthermore, pathologic tearfulness offers a useful neuropsychological model for exploring the neural mechanisms of emotion regulation and may provide insight into the breakdown of these mechanisms across a wide range of neurologic conditions. This will inform the development and refinement of behavioral and pharmaceutical interventions. Podcast: [NPub.org/965c4u](http://NPub.org/965c4u) CME Course: [NPub.org/cmelist](http://NPub.org/cmelist) The authors thank the participants of this study. Study funding ============= C.R.B. is supported by a Medical Research Council Clinician Scientist Fellowship (MR/K010395/1). S.R.I. is supported by the Wellcome Trust (104079/Z/14/Z), the UCB--Oxford University Alliance, BMA Research Grants--Vera Down grant (2013) and Margaret Temple (2017), Epilepsy Research UK (P1201), and the Fulbright UK--US commission (MS Society Research Award). The research was funded/supported by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. Disclosure ========== G. Argyropoulos, L. Moore, C. Loane, A. Roca-Fernandez, C. Lage-Martinez, and O. Gurau report no disclosures relevant to the manuscript. S. Irani is a coapplicant and receives royalties on patent application WO/2010/046716 (UK patent no. PCT/GB2009/051441) titled "Neurologic autoimmune disorders." The patent has been licensed to Euroimmun AG for the development of assays for LGI1 and other VGKC-complex antibodies. A. Zeman and C. Butler report no disclosures relevant to the manuscript. Go to [Neurology.org/N](https://n.neurology.org/lookup/doi/10.1212/WNL.0000000000008934) for full disclosures. aLE : autoimmune limbic encephalitis ANOVA : analysis of variance BA : Brodmann area BIS : Barratt Impulsiveness Scale BOLD : blood oxygenation level--dependent CBS : Cambridge Behaviour Scale CNS-LS : Center for Neurologic Study--Lability Scale DARTEL : diffeomorphic anatomical registration through the exponentiated lie algebra EPI : echoplanar imaging FWE : family-wise error FWHM : full width at half maximum GM : gray matter HADS : Hospital Anxiety and Depression Scale HC : healthy control IQR : interquartile range IRQ : Irritability Questionnaire MAP : Memory and Amnesia Project MNI : Montreal Neurologic Institute MTL : medial temporal lobe MVPA : multivariate pattern analysis OPTIMA : Oxford Project to Investigate Memory and Ageing PCA : principal component analysis rsALFF : resting-state abnormalities in the local amplitude of low-frequency fluctuations rsFC : resting-state functional connectivity rsfMRI : resting-state fMRI TIV : total intracranial volume VBM : voxel-based morphometry WM : white matter [^1]: Go to [Neurology.org/N](https://n.neurology.org/lookup/doi/10.1212/WNL.0000000000008934) for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. [^2]: The Article Processing Charge was funded by Medical Research Council.

Background ========== Surgery is the first choice for the treatment of most solid tumors without metastasis. However, surgery releases tumor cells into the circulation \[[@b1-medscimonit-22-2379]\]. Many studies have demonstrated that surgery can promote cancer metastasis by suppressing immunity and natural killer (NK) cell activity, both *in vivo* and *in vitro* \[[@b2-medscimonit-22-2379],[@b3-medscimonit-22-2379]\] Morphine- or fentanyl-based general anesthesia elicits an immunosuppressive effect by decreasing NK cell activity \[[@b4-medscimonit-22-2379]\]. A stress response is an unavoidable adverse effect of general anesthesia \[[@b5-medscimonit-22-2379],[@b6-medscimonit-22-2379]\]. Epidural anesthesia decreases the stress response after surgery, which can promote tumor metastasis \[[@b7-medscimonit-22-2379]--[@b9-medscimonit-22-2379]\]. Administration of opioids and inhaled anesthetics also contributes to increased NK cell activation during neural blockade. Studies have shown that epidural-supplemented anesthesia reduces cancer recurrence and postoperative mortality \[[@b8-medscimonit-22-2379],[@b9-medscimonit-22-2379]\]. However, other studies found no association between anesthesia (epidural and non-epidural experimental groups) and survival or the recurrence of cancer after cancer surgery \[[@b10-medscimonit-22-2379],[@b11-medscimonit-22-2379]\]. In another study, epidural anesthesia was associated with a lower recurrence of cancer after cancer surgery in patients older than 64 years. In a large cohort study, epidural anesthesia was associated with prolonged survival in cancer patients without metastasis; however, recurrence in these patients was unaffected \[[@b12-medscimonit-22-2379]\]. Therefore, due to the conflicting results, we retrospectively analyzed 273 patients who underwent gastric cancer surgery. The endpoint was death. We sought to determine whether epidural-supplemented anesthesia is associated with prolonged survival. Material and Methods ==================== After we received approval from the Review Board and Ethics Committee of ZheJiang Cancer Hospital, we collected the clinical records of 273 patients who underwent primary gastric cancer resection between August 2006 and December 2010. The follow-up ended at death, and the longest follow-up period lasted nearly 8 years (September 2006 to September 2014). In total, 116 patients were in the non-epidural group, and 157 patients were in the epidural-supplemented group. Epidural anesthesia was administrated in gastric cancer patients except for those with low blood volume, shock, local infection/surgery/trauma/malformation, bacteremia that may cause epidural infection, low coagulation status or insufficient stopping time anticoagulant, children, mental illness, and patients who refused the epidural. General anesthesia was induced with 0.5 μg/Kg sufentanil (Yichang Humanwell Pharmaceutical Co. Ltd. Yichang, Hubei, China) 1.3--3.0 mg/Kg propofol (AstraZeneca, UK), and 30--50 μg/Kg midazolam in patients of both groups. Anesthesia was maintained with 4--12 mg/Kg propofol and 0.5--1.0 μg/Kg remifentanil in the non-epidural group. Remifentanil (Yichang Humanwell Pharmaceutical Co. Ltd. Yichang, Hubei, China) was replaced with 0.25% levobupivacaine (Rundu, Zhuhai, China) or ropivacaine (AstraZeneca, UK) via an epidural catheter in the epidural group. We identified factors that may influence survival after surgery. These factors included age, sex, ASA class, tumor stage, anesthetic technique, surgical method, hemorrhage volume during the operation, and radiotherapy with or without chemotherapy, as well as a history of hypertension, diabetes, coronary heart disease, chronic obstructive pulmonary disease, liver disease, or any other disease. The TNM (topography, lymph node, and metastasis) staging was also obtained, and the tumors were divided into stages 0--IV based on the pathologic staging system that was in use when the surgery was originally performed. The data were recorded in Excel, and SPSS statistical analysis software was the main tool used for data analyses. Normally distributed continuous variables were compared using independent-samples *t* tests. Non-normally distributed variables were compared using the Mann-Whitney U test. Categorical variables were compared using the chi-square or Fisher's exact tests. We used the Kaplan-Meier log-rank test for univariate analyses and the Cox proportional hazards regression model for multivariate analyses of the survival time after cancer surgery. All factors that affect survival time after cancer surgery were selected as the main effects in the multivariate Cox model. The significance level for each hypothesis was 0.05. We used SPSS Statistics version 22.0.0.0 (IBM Corp. USA). Results ======= All baseline factors, including perioperative and postoperative risk factors, were compared between the groups ([Table 1](#t1-medscimonit-22-2379){ref-type="table"}). There were more men in the epidural group. The median survival times for cancer patients were 13.15 (7.1--21.95) and 13.2 (7.8--23.0) months for patients in the non-epidural and epidural-supplemented groups, respectively. No difference was found between the groups (p=0.541). The Kaplan-Meier analysis indicated no association between epidural-supplemented anesthesia and prolonged survival (p=0.680 log-rank), as shown in [Figure 1](#f1-medscimonit-22-2379){ref-type="fig"}. However, epidural anesthesia increased the long-term survival in patients younger than 65 years (p=0.035 log-rank test) ([Figure 2](#f2-medscimonit-22-2379){ref-type="fig"}). The multivariate Cox regression model indicated no significant association between epidural use and prolonged survival after cancer surgery (p=0.522), and the adjusted estimated hazard ratio was 0.919 (95% CI 0.71--1.19) ([Table 2](#t2-medscimonit-22-2379){ref-type="table"}). [Table 3](#t3-medscimonit-22-2379){ref-type="table"} shows the main effects for another multivariate Cox model. ASA status, tumor clinical stage, and chemoradiotherapy still affected prolonged survival after gastric cancer surgery. Discussion ========== Currently, whether epidural anesthesia prolongs the survival of cancer patients after surgery is controversial. Biki \[[@b8-medscimonit-22-2379]\], and Hiller \[[@b9-medscimonit-22-2379]\] demonstrated that epidural use is associated with lengthened survival in patients receiving surgery for prostatic carcinoma, pancreatic adenocarcinoma, and gastro-esophageal cancer. In contrast, studies by Christopherson \[[@b13-medscimonit-22-2379]\] and Myles \[[@b14-medscimonit-22-2379]\] showed no association between epidural anesthesia and prolonged survival in patients receiving surgery for gastric, colorectal, gastro-esophageal, non-small cell lung, and abdominal cancers. A Cochrane Review addressing this topic showed no survival benefit for the 2 groups by using a fixed-effects model with an HR of 1.03 (95% CI 0.86--1.24) in the pooled results \[[@b15-medscimonit-22-2379]\]. In recent decades, scientists have focused on the effects of perioperative factors and interventions on cancer recurrence and overall survival. Many factors are involved in cancer recurrence and overall survival, including tumor type, tumor stage and size, surgical skill and techniques, anesthetic technique, radiotherapy with or without chemotherapy, blood loss, and transfusions during the perioperative period. Other factors include comorbid diseases, such as hypertension, immunodeficiency, diabetes, or chronic obstructive pulmonary disease. The mixed results of the present study are difficult to interpret. Further investigation is needed before a reliable conclusion can be drawn. The stress of surgery negatively affects the adaptive immune system \[[@b16-medscimonit-22-2379]\]. Because immune surveillance is the first indicator for preventing cancer metastasis, immune suppression or disturbance may decrease the defensive barrier against intruders such as tumor cells, as well as inflammation. Immune suppression may then facilitate cancer cell migration and lead to cancer recurrence at either a local or distant site. The suppression of NK cell activity due to the stress of surgery promotes tumor movement \[[@b17-medscimonit-22-2379]--[@b20-medscimonit-22-2379]\]. Some anesthetics such as opioids and inhaled anesthetics may inhibit immune function, including NK cell activity, lymphocyte proliferative responses to mitogens, and phagocytic activity \[[@b8-medscimonit-22-2379],[@b18-medscimonit-22-2379],[@b19-medscimonit-22-2379]\]. Morphine and fentanyl have been well studied for their ability to suppress NK cell cytotoxicity \[[@b4-medscimonit-22-2379]\]. Local anesthesia is traditionally used for surgery. As previously described, the limited use of opioids during local anesthesia may attenuate the suppression of NK cell cytotoxicity. Studies also support this hypothesis \[[@b9-medscimonit-22-2379]--[@b12-medscimonit-22-2379]\]. Additionally, regional anesthesia inhibits the surgical stress response and is linked to earlier recovery times by reducing the incidence of thromboembolic, pulmonary, and gastrointestinal complications, as well as inflammation after major surgery \[[@b20-medscimonit-22-2379],[@b21-medscimonit-22-2379]\]. The conclusion of this study is consistent with that reached by Cakmakkaya \[[@b15-medscimonit-22-2379]\] that epidural anesthesia is not associated with the overall survival of gastric cancer patients after surgery. However, there were some differences between this study and previous studies. Hiller \[[@b9-medscimonit-22-2379]\] found that effective epidural analgesia improved the estimated median time to death among participants who received surgery for gastro-esophageal cancer. This study focused on gastric cancer patients who underwent surgery at the same hospital, and the study end-point was death. Interestingly, we found that epidurals improved the long-term survival of younger patients (aged 65 years or younger). Because of the complex effects involved, we still have not reached a conclusion concerning the direct interaction between epidural use and prolonged survival after gastric cancer surgery. The beneficial effect observed in younger patients may be attributed to their education level, better health, and better immunity, or the administration of postoperative chemotherapy with or without radiotherapy. The majority of gastric cancer surgeries were performed in less than 2 hours in our hospital. Fluid infusion during the surgery was continuous with a crystal: colloid ratio of 2: 1 during a period of 2 hours. The surgical and anesthesia techniques were performed according to standard procedures. The outcomes for cancer patients can be influenced by many factors (described above). The present study has certain limitations. Retrospective studies have inherent weaknesses, including susceptibility to bias, selection bias, and confounding variables. Moreover, the classification of gastric cancer has been changed, so "misclassification" in the past may have biased and nullified the result today. Also, the non-epidural group may have received other analgesics, which would increase the heterogeneity of the group. This retrospective study was unable to demonstrate any significant benefit for survival after the administration of epidural anesthesia during gastric cancer surgery. More results from randomized controlled trials, such as studies sponsored by The Cleveland Clinic \[[@b22-medscimonit-22-2379]\], are needed to draw a reliable conclusion. Our group is preparing a randomized controlled study on this topic. Conclusions =========== According to the results of the Mann-Whitney U test, the median survival after surgery for patients receiving epidural anesthesia was 17.1 months (range, 10.0--52.8 months). Epidural anesthesia combined with general anesthesia during gastric cancer surgery had no effect on the long-term survival according to the Cox proportional hazards regression model. Younger patients who received epidurals were more likely to have longer survival after surgery. **Source of support:** Departmental sources **Conflicts of interest** The authors declare no financial support or conflict of interest. {#f1-medscimonit-22-2379} {#f2-medscimonit-22-2379} ###### Patient characteristics in two types of anesthesia techniques. Factor Level Non Epidural (N=116) Epidural (N=157) P Value ----------------------------------------------------------------- ------- ---------------------- ------------------ ----------------------------------------------------------- **Baseline** Age Yr. 70 (63, 78) 67 (59, 76) 0.033 Sex Male 86 (74) 115 (73) 0.869[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} ASA I 52 (45) 82 (52) 0.432[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} II 50 (43) 58 (37) III 13 (11) 17 (11) IV 1 (0) 0 (0) ST[\*\*](#tfn7-medscimonit-22-2379){ref-type="table-fn"} 1 92 (79) 119 (76) 0.536[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} 2 18 (16) 32 (20) 3 6 (5) 6 (4) Stage[\*\*\*](#tfn3-medscimonit-22-2379){ref-type="table-fn"} I 5 (4) 6 (4) 0.599[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} II 1 (0) 10 (6) III 42 (36) 67 (43) IV 68 (59) 74 (47) T[\*\*\*\*](#tfn4-medscimonit-22-2379){ref-type="table-fn"} T1 2 (2) 3 (2) 0.601[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} T2 8 (7) 9 (6) T3 27 (23) 48 (31) T4 79 (68) 97 (62) N[\*\*\*\*\*](#tfn5-medscimonit-22-2379){ref-type="table-fn"} N0 9 (8) 15 (10) 0.627[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} N1 10 (9) 15 (10) N2 14 (12) 26 (17) N3 83 (72) 101 (64) M[\*\*\*\*\*\*](#tfn6-medscimonit-22-2379){ref-type="table-fn"} M0 13 (11) 10 (6) 0.155[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} M1 103 (89) 147 (94) EBL (ml) 300 (212, 400) 300 (200, 400) 0.337 RC Yes 53 (46) 72 (46) 0.978[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} HBP Yes 18 (16) 15 (10) 0.135[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} DIA Yes 4 (3) 5 (3) 0.904[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} CHD Yes 3 (3) 2 (1) 0.424[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} HEP Yes 15 (13) 18 (11) 0.713[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} COPD Yes 2 (2) 3 (2) 0.909[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} OTH Yes 30 (26) 47 (30) 0.46[\*](#tfn2-medscimonit-22-2379){ref-type="table-fn"} Statistics are median (Q1, Q3), or N (%), as appropriate. Wilcoxon Rank Sum Test unless specified. Pearson chi-square test. Tumor stage, stage 0 -- Tis; N0, M0; stage I -- T1, N0, M0/T2, N0, M0; stage II -- T1 N1 M0/T2 N1 M0/T3, N0, M0/T1, N3, M0/T3 N1 M0/T2 N2 M0/T4a N0 M0; stage III -- T2--4a, N3, M0/T3--4b, N2, M0/T4 N0--1 M0; stage IV -- any T, any N, M1. T0, no evidence of primary tumor; T1, invasion *via* submucosa into lamina; T2, invasion into the muscularis propria; T3, invasion through the subserosa; T4, invasion of surrounding structures. N0, no lymph nodes involved; N1, one to two nodes involved; N2, three or seven nodes involved. N3, seven or more nodes involved. M0, no metastasis; M1, metastasis. Surgical Technique, 1, Distal gastrectomy with Billroth I or Billroth II reconstruction; 2, Total gastrectomy with Roux-en-Y Gastric Bypass surgery. 3, Other surgical techniques not specified. CHD -- coronary heart disease; HEP -- hepatopathy; COPD -- chronic obstructive pulmonary disease; OTH -- other disease not specified. ASA -- American Society of Anesthesiologists; ST -- surgical technique; T -- topograph; N -- lymph node; M -- metastasis; EBL -- estimated blood loss; RC -- radiotherapy and/or chemotherapy; HBP -- hypertension; DIA -- diabetes; OTH -- other disease not specified. ###### COX regression results for multivariable of baseline factors. Baseline factors P value Hazard ratio 95.0% CI ------------------------------------------------------------------------------- --------- -------------- ------------------------------------------------------- ------- Sex .336 .863 .639 1.165 Age .386 .994 .979 1.008 Anesthetic technique .522 .919 .710 1.190 ASA Status (I) .002 [\*](#tfn10-medscimonit-22-2379){ref-type="table-fn"} ASA Status (II) .472 .471 .061 3.662 ASA Status (III) .914 .894 .118 6.775 ASA Status (IV) .867 .839 .108 6.549 Stage (I) [\*\*](#tfn11-medscimonit-22-2379){ref-type="table-fn"} .000 .494 .373 .653 Stage (II) .022 .437 .216 .886 Stage (III) .001 .321 .166 .621 Stage (IV) .000 .494 .373 .653 Surgical technique 1[\*\*\*](#tfn12-medscimonit-22-2379){ref-type="table-fn"} .836 [\*](#tfn10-medscimonit-22-2379){ref-type="table-fn"} Surgical technique 2 .954 1.018 .557 1.861 Surgical technique 3 .712 1.134 .582 2.207 Radiotherapy and/or chemotherapy .000 1.968 1.472 2.630 Estimate blood loss .809 1.000 .999 1.001 Transfusion .508 1.000 .999 1.001 Hypertension .419 1.199 .772 1.862 Diabetes .072 2.166 .933 5.029 Chronic heart disease .532 .725 .264 1.988 Hepatopathy .704 1.086 .710 1.660 Chronic obstructive pulmonary disease .830 .890 .306 2.588 Other disease .695 1.061 .790 1.423 Estimated overall survival function of each factor in multivariable COX regression model. No Estimated of hazard ratio (95% confidence interval \[CI\]) provided be- cause of very large proportion (96%) of "yes." Tumor stage, stage 0 -- Tis; N0, M0; stage I -- T1, N0, M0/T2, N0, M0; stage II -- T1 N1 M0/T2 N1 M0/T3, N0, M0/T1, N3, M0/T3 N1 M0/T2 N2 M0/T4a N0 M0; stage III -- T2--4a, N3, M0/T3--4b, N2, M0/T4 N0--1 M0; stage IV -- any T, any N, M1. Surgical technique, 1, Distal gastrectomy with Billroth I or Billroth II reconstruction; 2, Total gastrectomy with Roux-en-Y Gastric Bypass surgery. 3, Other surgical techniques not specified. ###### COX multivariable regression results Factors P value Hazard ratio 95.0% CI ----------------------------------------------------------------- --------- -------------- ---------- ------- Age .961 1.000 .987 1.013 Anesthetic technique .422 .901 .699 1.161 ASA Status (I) .005 ASA Status (II) .564 .557 .076 4.062 ASA Status (III) .946 .933 .129 6.778 ASA Status (IV) .741 .712 .095 5.325 Stage (I) [\*](#tfn14-medscimonit-22-2379){ref-type="table-fn"} .000 Stage (II) .032 .490 .255 .941 Stage (III) .001 .347 .184 .655 Stag (IV) .000 .522 .401 .679 Radiotherapy and/or chemotherapy .000 1.873 1.424 2.464 for main effectors. Estimated overall survival function of main factors in multivariable COX regression model. Tumor stage, stage 0 -- Tis; N0, M0; stage I -- T1, N0, M0/T2, N0, M0; stage II -- T1 N1 M0/T2 N1 M0/T3, N0, M0/T1, N3, M0/T3 N1 M0/T2 N2 M0/T4a N0 M0; stage III -- T2--4a, N3, M0/T3--4b, N2, M0/T4 N0--1 M0; stage IV -- any T, any N, M1. [^1]: Study Design [^2]: Data Collection [^3]: Statistical Analysis [^4]: Data Interpretation [^5]: Manuscript Preparation [^6]: Literature Search [^7]: Funds Collection

Introduction {#S1} ============ Microbial eukaryotes are versatile components of aquatic environments, covering multiple functional roles -- from autotrophy to heterotrophy (predators, decomposers, parasites) and mixotrophy -- and contributing to biogeochemical cycling ([@B107]; [@B47]). Unveiling the diversity of microeukaryote communities contributes remarkably to our understanding of microbial food web structure in aquatic ecosystems. For a long time, this pivotal component of the marine environment could only be explored morphologically, with resolution levels that left most small and featureless forms largely unexplored. Recent technological developments of molecular microbial ecology have expanded our capacity to describe and investigate the community diversity and structure and the biogeography of the single-celled eukaryotes, informally called protists ([@B19]; [@B52]). These molecular approaches have been widely applied in protist studies in different aquatic ecosystems, including inland, oceanic and coastal waters, and extreme environments (e.g., [@B25]; [@B2]; [@B109]; [@B17]). Community composition of planktonic protists may differ among offshore (e.g., [@B25]; [@B59]), coastal regions (e.g., [@B16]; [@B63]) and water depths (e.g., [@B71]); further the community may shift over seasons as well as over smaller (days to weeks) time scales (e.g., [@B8]; [@B68]). In ecological studies of protistan communities, one key goal is understanding the patterns of community biodiversity and structure, at both the spatial ([@B63]; [@B55]; [@B23]) and temporal scales ([@B35]; [@B8]), which in turn requires investigation of the factors that shape the communities ([@B94]; [@B105]; [@B67]; [@B108]). Combining molecular data of marine microeukaryote communities with tools from macroecology can be particularly useful to investigate the main factors affecting the community structure and patterns. In coupled ecosystems, such as transitional waters and the adjacent marine environment, habitat heterogeneity, and connection coexist as intrinsic ecological features. Transitional water ecosystems ([@B29]; [@B28]), such as shallow coastal lagoons, are fairly heterogeneous, mainly due to the geomorphology and catchment geology, the close benthic--pelagic coupling, the freshwater inputs, and the marine water exchanges -- through tides and currents -- with the adjacent marine ecosystems ([@B5]; [@B64]; [@B103]). Through the connection between transitional and marine waters, complex coupled ecosystems are established, where the regional and local factors are at play in shaping the planktonic community structure. In the northern Adriatic Sea, the Lagoon of Venice (LoV), and the Gulf of Venice (GoV) are examples of such transitional/marine coupled ecosystems. Both are research sites belonging to the Long Term Ecological Research (LTER) network LTER-Italy^[1](#footnote1){ref-type="fn"}^. LTER is an essential component of worldwide efforts to improve our knowledge of the structure and functions of ecosystems and of their long-term response to environmental, societal and economic drivers ([@B66]). LTER is organized in distributed networks of research sites at the global (ILTER, LTER-International), regional (LTER-Europe) and national levels. LTER-Italy, a formal component of ILTER and LTER Europe, consists of 79 research sites, which include terrestrial, freshwater, transitional and marine ecosystems ([@B9]). The marine component of LTER-Italy is mainly represented by transitional and coastal ecosystems ([@B81]). Therefore, consistent knowledge exists on the LoV and the GoV, in particular on phytoplankton communities and related abiotic factors ([@B1]; [@B12], [@B13], [@B11]; [@B30]), which may provide a useful background for molecular investigation on the whole protist community. In the area, a High Throughput Sequencing metabarcoding approach has so far been applied to the bacterial communities ([@B82]), while a few protist samples collected in June 2016 have been analyzed in the frame of the Ocean Sampling Day (OSD) project (e.g., [@B72]; [@B101]; [@B100]). In this study we analyze for the first time in the area the entire microeukaryote community through a metabarcoding approach, by sequencing the V4 region of the 18S rRNA gene. We analyse protist communities sampled at different temporal (seasonal) and spatial (lagoon and sea) scales, aiming at contributing to the identification of the relative importance of regional processes and local characteristics in structuring the protist community in aquatic ecosystems. Our study focuses on the differences between the lagoon and sea communities, with insights into their functional diversity revealed by their trophic habits and on the affinity of individual taxa for each of the two environments investigated. Materials and Methods {#S2} ===================== Study Area and Sampling Strategy {#S2.SS1} -------------------------------- The LoV, the largest (550 km^2^) lagoon of Italy and one of the largest in the Mediterranean basin, is a microtidal, polyhaline lagoon, located in the north-western part of the Adriatic Sea in a densely inhabited and industrial area, affected by high numbers of tourists and hosting ports, shipyards, marinas, fisheries, and aquaculture ([@B104]; [@B92]). The LoV has an average depth of 1 m with large shallow areas and a network of deeper (5--10 m) navigation channels. Twelve main tributaries discharge an annual average of about 35 m^3^ s^--1^ of freshwater into the lagoon, with seasonal peaks in spring and autumn ([@B92]). The LoV is connected to the coastal waters of the GoV through three inlets (Lido, Malamocco, and Chioggia), with water exchanges mainly governed by a microtidal (average amplitude 100 cm) regime ([@B92]). Water residence time ranges from a few days (close to the inlets) to 1 month (in landward areas), depending on the interplay of tide, wind, and topography ([@B102]; [@B36]). The GoV is a shallow marine system (maximum depth: 45 m) that receives the freshwater and nutrient inputs of several rivers, of which the Po (the main Italian river) is the major contributor, while in its eastern part it is influenced by the more saline and oligotrophic waters from the southern Adriatic basin. As a consequence, the trophic conditions are remarkably variable at both the spatial and temporal scales, ranging from the permanently meso-eutrophic western coastal area to the highly dynamic transition zone with saline oligotrophic waters offshore ([@B31]; [@B91]). The area considered in this study comprises the northern and central part of the LoV, and the coastal area from the Lido inlet up to 15 km offshore ([Figure 1](#F1){ref-type="fig"}). Four sampling campaigns, in April, July, November 2016, and February 2017 were conducted at seven stations. Four stations (St1, St2, St3, and St5) were located in the LoV and three in the coastal areas of the GoV (St4, PELL, and PTF); six of them (St1, St2, St3, St4, St5, and PTF) are LTER stations. The stations in the LoV were chosen more than 20 years ago as representative of the natural and anthropogenic environmental variability of the northern and central basins, being influenced by a complex interplay of freshwater and marine inputs ([@B93]; [@B22]) and by different human impacts ([@B15]). Stations St4 is located in one of the three inlets, while PELL and PTF are in the coastal and offshore GoV, respectively, with PTF close to the oceanographic tower "Acqua Alta"^[2](#footnote2){ref-type="fn"}^. {#F1} Environmental Parameters {#S2.SS2} ------------------------ The four sampling campaigns were performed separately for each environment in two consecutive days, always at neap tide, sampling only the near-surface water layer. At each station, temperature and salinity were measured with a CTD SBE 911 and water samples were collected with a Niskin bottle. For chlorophyll *a* (Chl *a*), 2 L of seawater from the GoV or 500 mL from the LoV were immediately filtered through Whatman GF/F fiberglass filters (nominal porosity = 0.7 μm); filters were stored frozen and subsequently analyzed according to [@B44]. Dissolved Inorganic Nitrogen (DIN: sum of ammonia, nitrite, and nitrate), orthophosphates (P-PO~4~) and orthosilicates (Si-SiO~4~) in filtered seawater were analyzed with a Flow-Solution III autoanalyzer (OI-Analytical) Systea-Alliance auto-analyser, according to [@B97] and [@B42]. Filtration, DNA Extraction, and Sequencing {#S2.SS3} ------------------------------------------ At each of the seven stations on the four sampling dates, water samples (3 L) were collected in duplicate in the near-surface layer, prefiltered on a 200 μm mesh-size net and then filtered onto cellulose ester 1.2 μm pore size filters (47 mm Ø, Whatman) using a peristaltic pump. A total of 56 filters were obtained and stored at −80°C until molecular analysis. Total DNA from each filter was extracted using the DNeasy 96 Plant Kit (QIAGEN) according to the manufacturer's instructions. DNA concentrations were determined by Qubit dsDNA HS kit (Thermofisher) and the DNA samples were stored at --80°C until PCR. The hypervariable V4 region of eukaryote SSU rRNA gene was amplified using the primers TAReuk454FWD1 and TAReukREV3 ([@B96]) modified as in [@B80] (5′ CCAGCASCYGCGGTAATTCC-3′ 5′A CTTTCGTTCTTGATYRATGA-3′). Finally, the hypervariable V4 region was sequenced (2 × 250 bp sequencing) on the Illumina MiSeq platform as described in [@B80]. Raw sequences were deposited in the Sequence Read Archive (SRA)^[3](#footnote3){ref-type="fn"}^ under the BioProject [PRJNA576330](PRJNA576330)^[4](#footnote4){ref-type="fn"}^. Sequence Analyses {#S2.SS4} ----------------- Paired-end reads were processed using Mothur v.1.33.0 ([@B89]). Contigs between read pairs were assembled and differences in base calls in the overlapping region were solved using ΔQ parameter as described in [@B50]. Primer sequences were removed (pdiffs = 3), and no ambiguous bases were allowed; the maximum homopolymer size was 8 bp. The remaining sequences were dereplicated and screened for chimeras using UCHIME in *de novo* mode ([@B26]). Taxonomic assignment was initially performed using a naïve Bayesian classifier ([@B106]) trained using the PR2 database (v.4.10.0^[5](#footnote5){ref-type="fn"}^; [@B40]), with an 80% bootstrap confidence threshold, in order to detect non-protist groups (including Bacteria, Archaea, Metazoa macroalgae and Fungi), which were excluded from further analyses. Sequences were clustered into operational taxonomic units (OTUs) at 97% of similarity using vsearch ([@B86]) clustering (method = dgc) through Mothur. OTUs containing only one read (singleton) were removed from downstream analyses. Taxonomic assignment was performed on a single representative sequence from each OTU (the most abundant) using BLASTN ([@B3]) against the PR2 database \[v.4.10.0 (see footnote 5); [@B40]\], discarding the assignments with similarity ≤ 90% and query coverage ≤ 70% of the sequence length. Based on these criteria, 1.5% of the sequences were excluded from further analyses. Functional Diversity {#S2.SS5} -------------------- To gain insight into different trophic levels of protists, taxa identified as described above were classified into four trophic functional groups: autotrophs, heterotrophs, mixotrophs, and parasites, based on information from the literature ([@B98]; [@B60]; [@B69]; [@B49], and other specific papers in [@B84]). The four selected trophic categories were sufficient for the identification of similarities and differences between the two environments investigated, while a functional investigation based on more detailed annotations of nutritional modes is beyond the aim of this study. Assignment to different categories was made at the group level in some cases, but the individual species, genera (or families) were considered whenever possible. We conventionally attributed only Bacillariophyta and Mamiellophyceae to autotrophs, whereas all other chloroplast-bearing taxa were considered as mixotrophs. Some groups as Labyrinthulea and Oomycota, which can be parasites, commensalists and/or saprophytes, were all included in parasites, as any other separate categories would consist of a low number of OTUs. Some unresolved supraspecific taxa (e.g., *Gymnodinium* spp.) or unnamed species (e.g., Dinophyceae_XXX) were not attributed to any specific trophic group in case they could include both autotrophs and heterotrophs. These taxa were annotated as "information not available (NA)." Statistical Analyses {#S2.SS6} -------------------- Hierarchical clustering of the 56 samples based on Bray--Curtis distance matrix showed duplicates for each sampling event to be paired in almost all cases (data not shown). The duplicates were hence pooled, resulting in a dataset of 28 samples. All the statistical analyses and plots were generated using several R packages (R version 3.5.2; [@B83]). Rarefaction curves of observed OTUs and α-diversity estimators (Shannon index, H) were calculated on this total dataset. For multivariate and comparative analyses, the abundance dataset was normalized with a random subsampling to the second lowest number of sequences (*n* = 21,181) with the "rrarefy" function, R package *vegan* version 2.5-5 ([@B70]). Non-metric multiDimensional Scaling (nMDS) was performed using the "metaMDS" function based on a Bray--Curtis dissimilarity matrix, using an OTU table modified following the Hellinger transformation. The adonis approach was applied to identify the explanatory environmental variables, i.e., those showing significant correlations with the community dissimilarity (Bray--Curtis index) matrix. The selected variables were used to perform Canonical Correspondence Analysis (CCA). Venn diagrams were calculated and plotted using *venn* version 1.7 R package. To assess the species' affinities for different habitats, OTUs were classified as "generalists" and "specialists" using the "clamtest" function in the *vegan* package. Based on the abundance in two habitats and a specialization threshold (*K*), the clam multinomial model classifies taxa into one of four groups: (1) "generalist"; (2) "habitat A \[lagoon\] specialist"; (3) "habitat B \[sea\] specialist"; and (4) "too rare to classify" with confidence. As *K* we applied a conservative threshold based on the super-majority rule (*K* = 2/3, *P* = 0.005). This approach permits a robust statistical evaluation of habitat specialization for a large number of species and does not rely on measurements of individual performance or exclude rare species *a priori* ([@B20]). The test was performed using a sub-dataset from two marine (PELL and PTF) and at two lagoon stations (St2 and St5), selected as representative of the natural (PTF and St5) and anthropogenic (PELL and St2) variability of the two environments ([@B12], [@B11]). For each habitat the original data from eight samples were pooled and the two resulting datasets were normalized to the lowest number (1,382,776 reads) ([Supplementary Material 3](#SD1){ref-type="supplementary-material"}, clam dataset). For network analysis, statistical relationships between taxa were calculated as described in [@B45] and [@B82], based on the MIC-MINE (Maximal Information Coefficient -- Maximal Information-based Non-parametric Exploration) algorithm ([@B85]). For this analysis we used the whole dataset (28 samples) normalized as described above for multivariate and comparative analyses. In order to satisfy the computational requirements of MIC-MINE, the OTUs with total abundance less than 50 reads were excluded from the analysis. After calculation of MIC MINE on the OTU table, we selected only those relationships that resulted statistically significant (*p* \< 0.01) upon comparison with *P*-value Tables available online^[6](#footnote6){ref-type="fn"}^. The MIC value is a statistical measure that indicates the strength of the interaction between OTUs but does not provide information on their sign (positive or negative). Connections were assigned as positive or negative using the MIC-p2 (non-linearity) parameter, provided in the MIC output, as described in [@B43]. MIC-p2 \> 0.5, indicating non-linear relationships, were considered negative connections. Significant values were selected and analyzed in Cytoscape 3.6.0 ([@B90]). The original network was compared against 100 randomized versions generated with Network Randomizer 1.1.2 ([@B99]). The NetworkAnalyzer tool within Cytoscape was used to calculate network topological parameters including the degree as a proxy for node importance (number of edges/connections arriving to or leaving from a node). The most connected nodes (high degree, range values 39--20) were considered as hub nodes. We interpreted the topology and hub nodes of the network using the classification resulting from clam test. The final visualization of the network was performed using the Force Atlas algorithm in Gephi software ([@B6]). Results {#S3} ======= Environmental Variables {#S3.SS1} ----------------------- The environmental conditions widely varied across the study area and the seasons ([Supplementary Table 1](#TS1){ref-type="supplementary-material"}). Water temperature reached the minimum in February (9.0°C in the LoV and 12.9°C in the GoV) and peaked in July (24.3°C in the LoV, and 25.7°C in the GoV). Salinity was lower in the LoV (28.8--29.7) than in the GoV (33.5--34.8). Inorganic nutrient concentrations were generally higher in the LoV (DIN: 18.7--43.7 μM; P-PO~4~: 0.5--1.1 μM; Si-SiO~4~: 17.8--38.7 μM) than in the GoV (DIN: 5.4--21.8 μM; P-PO~4~: 0.2--0.3 μM; Si-SiO~4~: 5.5--14.6 μM), with peaks in November in both sites. Chl *a* fluctuated between 0.5 and 4.8 μg L^--1^ in the LoV, with peak values in July, and from 1.1 and 4.0 μg L^--1^ in the GoV, with peak values in April. Differences between the two environments were only significant for salinity and P-PO~4~ (*p* \< 0.01) and for DIN and Si-SiO~4~ (*p* \< 0.05). The range values recorded in the study period are quite typical for the two areas and in line with the seasonal values recorded during 5 years of time series for each area ([Supplementary Table 1](#TS1){ref-type="supplementary-material"}). Protist Community {#S3.SS2} ----------------- ### Protist Diversity and Seasonality {#S3.SS2.SSS1} The 28 samples consisted of 12,268,180 raw reads. Filtering procedure generated a final curated dataset including 5,883,770 V4-18S protist reads which were clustered into OTUs at 97% similarity. After the removal of singleton, 7,336 OTUs remained. Most of the reads (88%, 1651 OTUs) were assigned to references of named species with similarity \> 99%, while for only 9% (3748 OTUs) and 3% (1937 OTUs) of the reads the similarity to a reference was in the range 98--95 and 94--90%, respectively ([Supplementary Material 1](#SD1){ref-type="supplementary-material"}, taxonomic assignations summary). Rarefaction curves on the whole dataset showed that the sequencing effort applied was sufficient to describe the protist diversity in the two areas, with an overall higher number of reads in the lagoon than in the marine samples, as well as at individual stations ([Supplementary Figure 1](#SD1){ref-type="supplementary-material"}). The normalized dataset used for most analyses resulted in an overall number of 570,936 reads, corresponding to 4,506 OTUs that belonged to 34 high-level taxonomic groups ([Supplementary Figure 2](#SD1){ref-type="supplementary-material"} and [Figure 2A](#F2){ref-type="fig"}). Bacillariophyta and Dinophyta were the most abundant groups in the entire dataset in terms of number of reads (23.9 and 21.2%, respectively), followed by Ciliophora (13.2%), Syndiniales (8.4%), and Cryptophyceae (7.7%). Chrysophyceae, Dictyochophyceae, Mamiellophyceae, Trebouxiophyceae, MAST (nanoheterotrophic marine stramenopiles), Picozoa, Rhodophyta, and Chlorophyceae ranged between 3 and 1%. The remaining 21 taxonomic groups showed a read percentage \< 1%. In terms of OTUs ([Figure 2B](#F2){ref-type="fig"}), Dinophyta showed the highest number (1673), followed by Ciliophora (426), Syndiniales (419), and Bacillariophyta (297). Cercozoa, Labyrinthulea, Chrysophyceae, Bicoecea, MAST, and Oomycota were present with 100--200 OTUs, and the remaining groups with less than 100 OTUs. {#F2} Most groups showed different distributions and composition across sites and seasons ([Figures 2C--E](#F2){ref-type="fig"}, [3A,B](#F3){ref-type="fig"} and [Supplementary Figure 3](#SD1){ref-type="supplementary-material"}). Diatoms were most abundant in spring and summer, when they dominated the LoV assemblages (51% in July, [Figure 3A](#F3){ref-type="fig"}), with the centric taxa *Cyclotella* sp. and *Thalassiosira concaviuscula* in April, *Minutocellus polymorphus* and *Chaetoceros tenuissimus* in July, and benthic pennate diatoms (*Undatella* sp., *Cymbella* sp., *Achnanthes* sp.) in February. In the sea, diatoms were less abundant than in the LoV and were dominated both in April and July by *C. tenuissimus*, accompanied by *Thalassiosira* spp. and other single cells *Chaetoceros* spp. Dinoflagellates were the most abundant group in the sea (39% in April and November, [Figure 3B](#F3){ref-type="fig"}) where they were mainly represented by chloroplast-bearing species such as *Heterocapsa pygmaea*, *Gymnodinium dorsalisulcum*, and *Tripos furca* across the year, *Alexandrium margalefii* and the heterotroph *Noctiluca scintillans* in April and *Alexandrium pseudogonyaulax* in July. The contribution of dinoflagellates was lower in the lagoon (\<13% in November, [Figure 3A](#F3){ref-type="fig"}), with *H. pygmaea* and *G. dorsalisulcum* dominating the group across the seasons, *Gymnodinium* sp. in November and *Protoperidinium tricingulatum* in February. {#F3} Ciliophora peaked in November at both sites with various Strobilidiidae in the LoV and *Strombidium* sp., *Pelagostrombidium neptunii* and Tintinnidae in the GoV. Ciliophora were abundant in the LoV also in April (e.g., *Strombidium* spp. and *Tintinnopsis* sp.) and in July (e.g., *Parastrombidinopsis minima* and *Parastrombidinopsis shimi*). Differences among seasons and between the two environments were also common in other groups ([Figures 3A--B](#F3){ref-type="fig"}). For example, Cryptophyceae (mainly *Teleaulax acuta*) were abundant in November at both sites, but *Rhodomonas* sp. only in the lagoon in April, while Syndiniales (unnamed taxa) peaked from April to November in the sea and in November-February in the lagoon. Some of the minor groups attained higher abundances in individual seasons and sites: Mamiellophyceae, mainly with *Ostreococcus mediterraneus* and *Micromonas bravo*, in the LoV in April (7.1%) and *M. commoda*, *M. bravo* and several other species in the GoV in February (5%); Trebouxiophyceae (*Picochlorum* sp.) in the LoV and in the GoV in July (16.1 and 8.1%, respectively); Dictyochophyceae (Pedinellales) in the LoV November (7.7%) and February (16.8%); Chrysophyceae in the LoV in November (8.4%). In spite of the wide fluctuations in community composition, the diversity within the different groups (i.e., OTU numbers) remained relatively stable over time, especially for diatoms ([Supplementary Figure 4](#SD1){ref-type="supplementary-material"}). Differences in group-specific OTU numbers between the two habitats were also less marked compared to abundance data (i.e., reads) and mainly consisted of a higher diversity of dinoflagellates in the GoV. Besides Syndiniales, a high OTUs diversity in individual samples was shown by parasitic and heterotrophic groups such as Cercozoa, Labyrinthulea, Oomycota, and Bicoecea, which covered up to 18.8% of the total OTU numbers in the LoV. Among the chloroplast-bearing taxa, the autotrophs (diatoms and Mamiellophyceae) prevailed in April--July while the mixotrophs (dinoflagellates and other groups) were more abundant in November--February ([Figure 4](#F4){ref-type="fig"}), the difference across the year being more evident in the LoV. Heterotrophs (mainly ciliates, Picozoa, *Paraphysomonas* spp. and MAST) attained the highest relative abundance (42.2% in the LoV) in November. Parasites showed an opposite trend in the two environments, with the highest levels in April (19.2%, Syndiniales and Oomycota) and July (17.2%, Syndiniales) in the GoV and in February in the LoV (17.8%, Syndiniales and Labyrinthulea) ([Supplementary Material 2](#SD1){ref-type="supplementary-material"}, trophic level normalized dataset). {#F4} The α-diversity (richness and H index) was rather similar in the two sites, as both overall and monthly values ([Supplementary Table 2](#SD1){ref-type="supplementary-material"}), with a peak in April (GoV = 1362; LoV = 1376, H \> 3.70). LoV samples in July had the lowest number of OTUs (826) and the lowest H (2.92), while in the GoV the minimum richness in February (757) was accompanied by a high H (4.14). All groups were present in both the LoV and GoV, but some minor groups showed more than 90% of the reads in the lagoon (Bicoecea, Perkinsea, and Pirsonia) or in the sea (Radiozoa) ([Figure 2C](#F2){ref-type="fig"}). The two environments shared, 96% of the total reads of the normalized dataset which however constituted only 26% of the total OTUs (1.168). Indeed, a high number of OTUs were exclusive of the LoV or the GoV (1,691 and 1,675, respectively) but they included only 1.6--1.8% of the reads of the normalized dataset ([Supplementary Figure 5](#SD1){ref-type="supplementary-material"}). In multivariate analyses, nMDS showed a clear separation of the community structure between the two ecosystems ([Supplementary Figure 6](#SD1){ref-type="supplementary-material"}), while CCA identified temperature and salinity as statistically significant variables (*P* \< 0.001), followed by Chl *a*, Si-SiO~4~, and DIN (*P* \< 0.05) ([Supplementary Figure 7](#SD1){ref-type="supplementary-material"}). For the whole area, the first two canonical axes explained 17% of the total variance ([Supplementary Figure 7](#SD1){ref-type="supplementary-material"}). The marine and lagoon samples were separated along the salinity gradient while all samples were spread along the temperature gradient. In the LoV ([Figure 5A](#F5){ref-type="fig"}), the first two canonical axes explained 21% of the total variance, to which only temperature made significant contributions (*P* \< 0.001). In the GoV ([Figure 5B](#F5){ref-type="fig"}) the first two canonical axes explained 35% of the total variance. Temperature and salinity made significant contributions to the variance (*P* \< 0.001), followed by DIN, P-PO~4~, and Chl *a* (*P* \< 0.01). {#F5} ### Community Composition: Generalists vs. Specialists {#S3.SS2.SSS2} The CLAM test conducted on 2 LoV and 2 GoV stations identified 3657 OTUs (0.6% of total reads) that could not be classified with statistical confidence ([Supplementary Material 3](#SD1){ref-type="supplementary-material"}, clam dataset). Of the classified OTUs, almost 80% were "specialists," i.e., taxa with a preference for one of the two environments ([Figure 6](#F6){ref-type="fig"}). LoV specialists (541 OTUs) and GoV specialists (851 OTUs), however, were not always exclusive of one of the two environments. A lower number of OTUs (355) were "generalists," i.e., with no clear preference for any of the two environments ([Supplementary Material 3](#SD1){ref-type="supplementary-material"}, clam dataset). {ref-type="supplementary-material"}.](fmicb-10-02736-g006){#F6} The taxonomic composition of the LoV and GoV specialists reflected that of the total community of the samples selected for this analysis ([Figure 6](#F6){ref-type="fig"} and [Supplementary Figure 8](#SD1){ref-type="supplementary-material"}), yet with an even higher contribution of Dinophyceae to the GoV specialists (412 OTUs, 56.5% of the reads) compared with their minor contribution (67 OTUs, 3.1%) to the LoV specialists. *Heterocapsa minima* represented more than a half (56.07%) of the GoV-specialist dinophycean reads, whereas a few other taxa (*Noctiluca scintillans*, *Alexandrium margalefii*, and *Tripos furca*, along with an unknown species) hardly attained more than 1% of the total ([Supplementary Material 3](#SD1){ref-type="supplementary-material"}, clam dataset). A long list of still less abundant taxa represented the remnant 408 GoV-specialist Dinophyceae OTUs. Bacillariophyta (34 OTUs) were only 14.27% of the GoV specialist reads, and were represented mainly by *Chaetoceros tenuissimus* (8.78 of the total LOV specialist reads), *Thalassiosira profunda* (2.65%), *C. muelleri*, and *C. socialis*, the latter two with less than 1% of the GoV specialist reads. An unknown Syndiniales was also quite abundant (5.45%) among the GoV specialists. In the LoV, the most abundant and diversified specialists were Bacillariophyta (62 OTUs, 21.17% of the LoV specialist reads), of which most were *Cyclotella* sp. (43.69%) and *Minutocellus polymorphus* (35.86%), followed by *Thalassiosira* spp. (12.08%) and by a quite rich list of ca. 30 benthic araphid and raphid diatoms. A wide variety and relatively high abundance of Ciliophora (85 OTUs, 16.03% of reads) was another distinctive feature of the LoV compared to the GoV specialists. Other abundant groups of the LoV specialists were Syndiniales (157 OTUs, 15.48% of the total LoV specialists reads), followed by Cryptophyceae (11.34%, mainly *Teleaulax acuta* and *Rhodomonas* sp.) and Trebouxiophyceae (7.50%, mainly *Picochlorum* spp.). The most abundant generalists were Dinophyceae (80% represented by *Gymnodinium dorsalisulcum*), several species of Ciliophora, Cryptophyceae, Mamiellophyceae (*Micromonas* and *Ostreococcus* spp.) and MAST. Differences in composition were also reflected in differences in the trophic structure of the two environments, with mixotrophs more widely represented in the GoV specialists and autotrophs and heterotrophs more abundant in the LoV specialists ([Figure 6](#F6){ref-type="fig"}). Parasites were more represented in specialists than in generalists, whereas the opposite occurred for heterotrophs. ### Network Analysis {#S3.SS2.SSS3} The MIC-MINE analysis produced a network with 283 nodes and 1071 edges. The clustering coefficient (C) and the characteristic path length (L) in the original network (C = 0.353, L = 4.061) were higher than the average from the random networks (Cr = 0.025, Lr = 2.97). The annotation of nodes with the generalist/specialist classification identified a clear separation between LoV and GoV, along with a lower degree of connections among GoV specialists compared to LoV ones ([Figure 7](#F7){ref-type="fig"}). A total of 22 nodes showed degree values between 39 and 20 (hubs); among them. 16 were annotated as LoV specialists, two as GoV specialists and four as generalists ([Supplementary Material 4](#SD1){ref-type="supplementary-material"}, network annotation). The 16 LoV specialist hubs included taxa such as Ciliophora (*Tintinnopsis* sp. and *Cyclotrichia* sp.), *Telonemia*, Dictyochophyceae (including *Apedinella radians* and undetermined species), Chrysophyceae (including the terrestrial species *Pedospumella encystans* and undetermined species), Cryptophyceae and Bacillariophyta (*Thalassiosira* spp.). The two GoV specialist hubs comprised Bacillariophyta (*Chaetoceros muellerii*) and one Centroheliozoa (Pterocystida) OTU. The four generalist hubs were represented by taxa belonging to the Mamiellophyceae, Chrysophyceae, Dictyochophyceae and MAST. {#F7} In addition to the higher number of hubs. the number of connections was higher in the lagoon than in the sea (571 vs. 263 connections), with a quite high number of negative interactions (more than 60%) which were more densely found between the LoV nodes. Discussion {#S4} ========== The analysis of the V4 hypervariable region of the 18S rRNA gene in the Lagoon and Gulf of Venice on four dates provides the first detailed overview of planktonic protists in an area where morphology-based studies on these communities date back to XIX century. During the last decades, ca 300 microalgal ([@B13], [@B11], [@B14]) and a few microzooplankton taxa ([@B4]) were identified in light microscopy, which is a much lower number compared to the more than 7,000 OTUs recorded in the present study. Considering that our dataset only included four sampling dates, the actual diversity of planktonic protists is likely even greater than the one observed in this study. This result is common to other metabarcoding studies in LTER areas (e.g., [@B80]) and is not unexpected. Traditionally whole groups, mainly heterotrophs, parasites, and picoeukaryotes, have received scarce attention if any, while other groups are difficult to identify with light microscopy or even unseen in fixed material (e.g., small-sized flagellates). Results of this study clearly show that all these groups are well represented with several distinct species in both the LoV and the GoV. For example, several green picoplanktonic coccoids or flagellates, not reported before from neither the LoV nor the GoV (e.g., *Picochlorum*, *Micromonas*, *Ostreococcus*, and *Bathycoccus*), were abundant in the metabarcoding dataset. Also in the case of diatoms, which instead have been studied based on morphology ([@B12], [@B13], [@B11], [@B14]; [@B18]; [@B62]), the present study reveals many more taxa than those known for the area, likely because of the capability of the molecular approach to detect rare species and to resolve cryptic diversity, which is widespread also in this taxonomic group ([@B87]; [@B58]; [@B73]). Benthic, tychoplanktonic species (e.g., *Nitzschia* spp., *Navicula* spp., and *Amphora* spp.), reported as typical lagoon inhabitants ([@B11]), were quite diversified in the metabarcoding results but with low read numbers, probably due to the limited time span or because they were not identified due to the well-known lack of reference sequences for benthic diatoms ([@B79]). Among dinoflagellates, particularly interesting was the detection of several potentially toxic species of the genera *Alexandrium* and *Azadinium* and of *Gambierdiscus australes*, as well as, of rare taxa such as the dinoflagellate *Posoniella tricarinelloides*, a species only found in fossil records until its recent recovery from cyst germination ([@B38]). Yet the attribution of reads to a named species should be taken with caution even when the environmental reads fully match the reference sequences, because distinct species may share the same V4 sequence (e.g., in several diatoms and dinophycean species), and therefore their sequence-based identification is not certain ([@B57]). A common feature in the two environments was the temporal variability signal, with the dominance of autotrophic forms, mainly diatoms, especially in spring and summer and the alternation of different taxa across the year. Similarly, signals of seasonality and shift of protist community composition along the year emerged clearly from several previous studies from marine ecosystem in the Mediterranean Sea ([@B80]; [@B37]), and English Channel ([@B34]; [@B51]). However, the relative abundance of groups and species and their temporal variability were clearly different between the lagoon and the sea. The difference of salinity between the LoV and the external coastal Adriatic waters is low and should not constitute a barrier for marine microeukaryotes. Hence it is not surprising that marine species are predominant in the whole dataset and that most species thriving in the lagoon also inhabit the external coastal waters, and viceversa, like in the case of other mixohaline lagoons of the Italian coasts (e.g., [@B88]). Some of these shared species can also attain comparably high abundance in both sites (i.e., the generalists in the CLAM analysis). However, high proportion of specialists with a differential distribution in the two areas indicates specific ecological conditions that structure the local communities in different ways. Among LoV specialists are several species commonly found in lagoons or estuarine environments, such as *Teleaulax acuta*, *Picochlorum* spp., *Cyclotella* sp. and *Minutocellus polymorphus* ([@B88]; [@B7]; [@B65]; [@B41]), while the majority of GoV specialists are Dinophyceae that have a considerably lower importance in the LoV. In addition to the dominant specialists, also less abundant taxa were distributed preferentially or even exclusively in the LoV or in the GoV. The latter was the case of a large part of the OTUs that included only a minor part of the total reads. This large diversity allocated in the rare OTUs (Avg: 6 reads per OTU, ca. 0.001% of the total dataset) points at a common structure in protist communities where a limited number of species play the main functional roles and the rare ones constitute the biodiversity reservoir ([@B46]). The finding of a high number of rare OTUs exclusively in the LoV or in the GoV could depend on their abundance at the boundary of the detection limit ([@B53]), which makes their differential distribution not completely sound, but could also indicate a specific composition of the biodiversity reservoir of the two ecosystems, in line with the biogeographic signal found in the rare component of metabarcoding datasets at a larger, European scale ([@B54], [@B53]). The spatial differences in plankton communities emerging from the present study confirm what was known from previous years and studies ([@B1]; [@B12], [@B11]) and can therefore be considered quite typical of the area. Interestingly bacterial communities analyzed with the same approach also showed both spatial and temporal differences in the GoV and LoV, along with a prevalent role of seasonality in shaping prokaryotic assemblages ([@B82]). In the same study, prokaryotic diversity was much higher in the lagoon due to the contribution of microbial communities from both the watershed and coastal waters in that site. This pattern was less evident in our data, where various diversity indices do not show significant differences between the two environments. The relationship between the LoV and GoV is further confirmed by the ecological network, where specialists of each ecosystem identified two distinct but connected sub-networks, with a high number of connections of which many were negative. High amount of negative interactions is generally interpreted as the result of functional heterogeneity, direct competition for limiting resources or interactions like predator-prey relations or allelopathy ([@B56]). However, positive or negative ecological interactions may simply reflect co-occurrence or non-coexistence patterns among populations ([@B43]). Particularly in this study negative interactions may just derive from the spatial and temporal ecological niche partitioning, which matches all other results that point to different communities characterizing the two ecosystems in the different seasons. Positive interactions, indicating groups of organisms having similar, complementary or cooperative functions or activities, but also common preferred environmental conditions ([@B33]; [@B27]), were less evident among the communities of the area. In the interpretation of network properties, the ecological role of hub species is still unclear. In several studies, hubs are often proposed to be critical or keystone components for network ([@B78]; [@B21]; [@B43]) but a recent study has also demonstrated that known keystone species do not necessarily result in detectable signals in co-occurrence networks ([@B32]). In our study, most hubs were not dominant in abundance, suggesting that scarcely abundant but highly connected OTUs could play an important role in network dynamic and stability not only in marine communities ([@B108]) but also in the lagoon habitat. The higher number of hubs ascribed to LoV specialists and the higher number of connections suggest a higher complexity and species inter-dependence among planktonic protist communities in the lagoon ecosystem. Interactions among species would frequently be enhanced in lagoons as well as in harbors and other semi-enclosed areas, possibly due to the proximity to the bottom and to the benthic vegetation, or to any other effects deriving from being confined environments ([@B61]; [@B39]; [@B75], [@B76]; [@B88]; [@B48]) and to the degree of connectivity with the sea ([@B36]; [@B74]). Different drivers are probably involved in shaping protist communities in the GoV, where the number of connections among specialist hubs is considerably lower. Considering also the higher variance explained by environmental variables, a higher influence of abiotic factors seems to be at stake in the GoV. This picture would also be reflected in the annual phytoplankton cycle of the GoV, which is more irregular than in the lagoon, with minor peaks alternating in spring and summer, due to the combination of nutrient depletion and sporadic nutrient inputs ([@B10], [@B13]; [@B91]). Conclusion {#S5} ========== Long Term Ecological Research sites constitute inspiring places, where long term observations stimulate a wide range of specific research activities ([@B77]) that provide, in turn, precious tools for interpreting long term data, thus increasing their informative value ([@B110]). This is the case for metabarcoding studies which are particularly valuable at LTER sites, where the existing background ecological knowledge allows optimizing both the arrangement of the molecular research and the interpretation of its results ([@B24]; [@B95]). At the LTER-Italy sites of the Lagoon and Gulf of Venice, two strictly coupled transitional and marine ecosystems, long-term studies conducted with light microscopy have so far provided the basis for the current understanding of the spatial and temporal variability of plankton in the area. This investigation of the whole protistan community based on the V4-18S rRNA, which is the first in the area, has largely increased the knowledge about protist diversity not only for groups that have traditionally been neglected (mainly heterotrophs, parasites, and picoeukaryotes), but also for the main phytoplankton taxa studied in the long term with morphology-based approaches (e.g., diatoms and dinoflagellates). The molecular database obtained in this study will be a reference for future studies and foster further taxonomic explorations, which will result in an improvement of the quality of the long term dataset. Further, metabarcoding results clearly indicate spatial differences in the structure of the protistan communities and their changes over the time in an area where habitat heterogeneity and connection coexist. Despite the presence of the most abundant taxa at both sites, their relative contribution and temporal variability are indeed different between the two environments, while the rare taxa are also exclusive of one of the two environments in many cases. Indeed, specific features of the lagoon and the sea communities clearly emerge from all our results. In these so close ecosystems, environmental heterogeneity appears strong enough to allow for ecological segregation, despite no clear barrier to dispersal processes among local protist communities. Finally, a higher degree of species inter-dependence among planktonic protists has emerged in the lagoon compared to the adjacent coastal waters, suggesting that different drivers are at play in shaping communities in the two ecosystems, with a prevalence of biotic interactions in the lagoon and a higher influence of abiotic factors in the sea. Overall, these results provide a starting point and a sound motivation for extending the analysis to multiple years also including other components of the community, e.g., bacteria and metazoans, in order to deepen the knowledge of the seasonal patterns and the biotic interactions that is needed for monitoring and managing changes in these LTER sites. Data Availability Statement {#S6} =========================== All datasets generated/analyzed for this study are included in the article/[Supplementary Material](#SD1){ref-type="supplementary-material"}. Author Contributions {#S7} ==================== SA, GQ, FB, AP, and AZ designed the experiment. SA, SF, FB, and MB provided samples and environmental data. SA, RP, GQ, and SF performed the DNA extraction. GQ and RP performed the bioinformatic and statistical analyses. SA, GQ, RP, SF, FB, AP, and AZ contributed to the interpretation of the results. SA and RP wrote the first draft of the manuscript. SA, RP, GQ, FB, SF, AZ, and AP finished the manuscript. All authors contributed to the revised version, and read and approved the final version of the manuscript. Conflict of Interest {#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.** RP was supported by the Italian MIUR Flagship Project RITMARE and the European Union's Horizon 2020 Research and Innovation Programme EMBRIC (GA 654008). SA was supported by LifeWatch-Italy, the Italian component of the EU LifeWatch-ERIC. This work was possible thanks to the partial financial support by LifeWatch-Italy, the Italian component of LifeWatch-ERIC. We thank Dr. Francesco Acri (CNR-ISMAR, Venice) for providing abiotic analyses. [www.lteritalia.it](http://www.lteritalia.it) <http://www.ismar.cnr.it/infrastructures/piattaforma-acqua-alta> <https://www.ncbi.nlm.nih.gov/sra> <https://www.ncbi.nlm.nih.gov/bioproject/PRJNA576330> <https://github.com/pr2database/pr2database> <http://www.exploredata.net> Supplementary Material {#S10} ====================== The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fmicb.2019.02736/full#supplementary-material> ###### Click here for additional data file. ###### Click here for additional data file. [^1]: Edited by: Eva Ortega-Retuerta, UMR 7621, Laboratoire d'Océanographie Microbienne (LOMIC), France [^2]: Reviewed by: Jean-François Mangot, Spanish National Research Council (CSIC), Spain; Savvas Genitsaris, International Hellenic University, Greece; André M. Comeau, Dalhousie University, Canada [^3]: ^†^These authors have contributed equally to this work and share first authorship [^4]: This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology

Background {#Sec1} ========== The research of chalcogenide glassy (ChG) materials formed a general understanding of electronic phenomena in disordered structures \[[@CR1], [@CR2]\]. The numerous investigations of their fundamental physical and chemical properties have been already performed \[[@CR3]--[@CR6]\]. Unique structural, electronic, and optical properties determined their various applications. The high infrared (IR) transparency of fibers on the basis of the ChG allows transmitting high-power IR light. The large refractive indices and third-order optical nonlinearities of the chalcogenide glasses make them the best candidates for the photonic devices for ultrafast all-optical switching and data processing \[[@CR7]\]. Various applications have been proposed on the basis of the light sensitivity of non-crystalline chalcogenides, especially in amorphous thin film form \[[@CR8]--[@CR10]\]. Thus, photosensitivity is the main feature of chalcogenide glasses for phase-change memory, direct waveguides, and grating patterning. The high-quality optical elements are required for the development of all-optical signal processing systems. Possibility of high-level integration of these elements in optical chips implies improved fabrication technology in order to achieve low optical losses at the near surface layers and the high level of laser damage threshold at femtosecond laser pulses. Also the large IR transparency or high optical nonlinearity of amorphous As--S binary systems make them a prospective optical media for the future ultrafast photonic systems. Our previous Raman studies of non-crystalline As--S binary system reveal the differences between the structures of As~2~S~3~ films and bulk glass at nano-scale dimension \[[@CR11]\]. The analysis shows that it is caused mainly by the phase separation, i.e., contribution of As~4~S~4~ cage-like molecules in the vapor during As~2~S~3~ thermal evaporation. In the further studies of the structure of amorphous As~2~S~3~ glasses and films using photon-energy-dependent Raman spectroscopy, the effect of laser-induced transformation of As~4~S~4~ molecules was observed \[[@CR12]\]. Therefore, the As~4~S~4~ molecules can be classified as light absorption centers in As~2~S~3~ structure and leading to increasing the optical losses of an optical media. The structure and properties of As~45~S~55~ glassy material and thin film were investigated earlier \[[@CR13], [@CR14]\]. Using macro FT-Raman spectroscopy, energy-dependent micro-Raman spectroscopy and first principle calculations established that the light-induced structural transformations in As~45~S~55~ glass take place mainly from alterations of As~4~S~4~ molecules in glass network. An impact of near-bandgap laser illumination transforms α(β)-As~4~S~4~ molecules to pararealgar-like p-As~4~S~4~ \[[@CR13]\]. The extended X-Ray absorption fine structure (EXAFS) study of photoinduced structural changes in amorphous As~*x*~S~1-*x*~ thin films showed that effect of the near-bandgap light illumination to the evaporated a-As~42~S~58~ and a-As~45~S~55~ films results in more disordered state and photostructural transformations are related to changes in the amorphous As-S network \[[@CR14]\]. Raman spectroscopy of As-rich As~50~S~50~ thin films revealed pararealgar structure of χ-As~4~S~4~, β-As~4~S~4~ molecules, clusters of amorphous arsenic, S~2~As-AsS~2~ and As~4~S~5~ structural units (s.u.), some part of α-As~4~S~4~, As~4~S~3~ molecules, and AsS~3~ pyramids \[[@CR15]\]. The aim of the present work is a complex structural investigation of thin film surface nanolayers prepared from As~40~S~60~, As~45~S~55~, and As~50~S~50~ chalcogenide glasses using X-ray photoelectron (XPS) and Raman spectroscopy, near-bandgap laser light's influence on structural and compositional changes, and their electronic structure. In addition, the changes of surface morphology induced by laser light illumination were investigated using surface profilometry method. Methods {#Sec2} ======= High-quality optical glasses were used as the source materials for sample deposition in order to avoid the contamination in the volume of the films. The bulk As~*x*~S~100-*x*~ (*x* = 40, 45, 50) samples were prepared by conventional melt-quenching route in evacuated quartz ampoules from a mixture of high purity 99.999% As and S precursors. Nanolayers were prepared by thermal vacuum evaporation of appropriate bulk glass powders onto silicon and glass substrates. The thicknesses of obtained films were \~0.7 μm. Green diode laser operating at λ = 532 nm wavelength (photon energy of \~2.4 eV) with power *p* = 25 mW was used to investigate the influence of the near-bandgap light irradiation on the samples of As~40~S~60~, As~45~S~55~, and As~50~S~50~ (E~g~ \~2.4 eV) nanolayers. Optical irradiation was carried out with 280 mW/cm^2^ intensity at ambient conditions. Laser intensity was chosen based on our previous studies of As--S glasses by means of Raman spectroscopy, mentioned above. To determine exposition of the laser illumination following experiments was done. The absorption edge of the As~*x*~S~100-*x*~ (*x* = 40, 45, 50) thin films and its shift under in situ illumination by green laser light was investigated using millisecond CCD spectrometer ThorLabs CCS200. Sample illumination with in situ optical spectra recording were done until saturation of the shift of the absorption edge. Photoemission experiments were conducted by using Al k-α anode (E = 1486 eV) as a source of X-ray. Spectra were recorded using hemispherical energy analyzer series Phoibos 100. An As 3d and S 2p core levels and valence bands (VB) were measured at normal emission geometry. Apart from this, the C 1s and O 1s core level spectra were recorded in order to normalize the positions of all spectra to a position of the graphitic peak (at 284.5 eV, \[[@CR16]\]). C 1s core level spectra were fitted by C--C and C--O components only, and this agrees with the O--C components founded in O 1s core level spectra. Due to that C 1s and O 1s core levels would not be included in the further consideration. The CASA XPS program was used to fit core level spectra. For core level fitting, the Voigt profile components were used and Shirley background was subtracted. Raman spectra were measured with using Renishaw system 1000 Raman spectrometer, equipped with a CCD detector. The diode laser operating at 785 nm was used as the excitation source. The measurements were made in micro-Raman configuration with using back-scattering geometry. In order to avoid stimulated by this laser photoinduced changes in the structure of the samples, the output power of the excitation source was limited by optical filters \[[@CR17]\]. Results {#Sec3} ======= Absorption Edges of As~*x*~S~100-*x*~ Films Under External Influence {#Sec4} -------------------------------------------------------------------- Optical spectra of absorption edges of the As~40~S~60~, As~45~S~55~, and As~50~S~50~ films are shown in Fig. [1](#Fig1){ref-type="fig"}. As can be seen, the absorption edge of As~40~S~60~ sample shifts towards longer wavelengths when film exposed to green laser light illumination (λ = 532 nm) with photon energy of \~2.4 eV, which is very close to E~g~ of As~40~S~60~ material (Fig. [1](#Fig1){ref-type="fig"}, left). Typical red shift of the absorption edge of As--S films during the near bandgap illumination was observed earlier \[[@CR5], [@CR18]\]. After 45 min of laser illumination of As~40~S~60~ film, the shift becomes less and at exposure time of \~150 min, the changes almost disappeared.Fig. 1A shift of absorption edge of As~40~S~60~, As~45~S~55~, and As~50~S~50~ thin films under green (λ = 532 nm) laser light illumination A similar phenomenon was observed for the absorption edge of illuminated As~50~S~50~ film (Fig. [1](#Fig1){ref-type="fig"}, right). However, the results of the same investigation of As~45~S~55~ film demonstrate the opposite effect. For this composition, the blue shift of the absorption edge is observed under the green laser illumination and 90 min was enough to reach the saturation. The structural interpretation of this phenomenon will be provided in the next paragraph using the XPS analysis and Raman spectroscopy data. The determined particular exposure times sufficient to saturate the changes of the absorption edges of As~40~S~60~ (t~exp.~ = 150 min), As~45~S~55~ (t~exp.~ = 90 min), and As~50~S~50~ (t~exp.~ = 150 min) films were used for further experiments. XPS Spectroscopy and Valence Band Spectra of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) Film Surfaces {#Sec5} ----------------------------------------------------------------------------------------------- X-ray photoelectron spectroscopy can be a useful technique to investigate the surface (i.e*.* few topmost layers) of the materials at short-range order scale and to determine the structural units which form the investigated material. This method was successfully used to characterize the top surface nanolayer structure of the amorphous materials \[[@CR19]--[@CR21]\]. The results of XPS investigation of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) thin film surface nanolayers are summarized in Fig. [2](#Fig2){ref-type="fig"}. As can be seen, all S 2p core level spectra can be fitted by two components. The energy position of component 1 (and their spin-orbit split 1') allows assigning them to S--As~2~ s.u. \[[@CR22], [@CR23]\]. It is expected that this component is a characteristic s.u. in crystalline As~2~S~3~ and is a main component for both stoichiometric, As~40~S~60~, and As-rich As--S glasses and films. The component 2 (and 2') can be assigned to S-rich S--SAs s.u., and it is in a good agreement with our earlier investigations and theoretical estimations \[[@CR19]\]. The As 3d core level spectra of As~40~S~60~ nanolayers (both as-deposited and illuminated by green laser light) are fitted using three components: arsenic bonded to three sulfur atoms As--S~3~ (1, 1'), arsenic bonded to two sulfur, and one arsenic atoms As--S~2~As (2, 2') and finally, arsenic bonded to one sulfur and two arsenic atoms As--SAs~2~ (3, 3'). These assignments are based on our previous study \[[@CR19]\] and are in excellent agreement with the published data \[[@CR16], [@CR24]\]. It should be noted that for the best fit of the As 3d core level spectra of As~45~S~55~ nanolayers, the fourth component is needed. The position of this component allows to interpret it as arsenic bonded to three arsenic atoms \[[@CR16]\]. All peak component parameters are listed in Table [1](#Tab1){ref-type="table"}.Fig. 2S 2p and As 3d fitted core level spectra of as-deposited (*top*) and illuminated by green (λ = 532 nm) laser (*bottom*) As~40~S~60~, As~45~S~55~, and As~50~S~50~ thin films top nanolayers. For S 2p: *1*, *2* denote 2p~3/2~ and *1´*, *2´* denote 2p~1/2~ peaks of S--As~2~ (1, 1') and S-SAs (2, 2') components; for As 3d: *1*, *2*, *3*, *4* denote 3d~5/2~ and *1´*, *2´*, *3´*, *4'* denote 3d~3/2~ peaks of As--S~3~ (1, 1'), As--S~2~As (2, 2'), As--SAs~2~ (3, 3'), and As--As~3~ (4, 4') components Table 1Binding energies (BE, ±0.1 eV) and full width at half maximum (FWHM) (±0.05 eV) data of individual components determined from curve fitting of S 2p and As 3d XPS spectra of as-deposited and illuminated by green (λ = 532 nm) laser As~40~S~60~, As~45~S~55~, and As~50~S~50~ nanolayersCore level/componentAs~40~S~60~As~45~S~55~As~50~S~50~As receivedIlluminatedAs receivedIlluminatedAs receivedIlluminatedBEFWHMBEFWHMBEFWHMBEFWHMBEFWHMBEFWHM**S 2p:**S--As~2~162.11.2162.11.2162.11.2162.21.1162.11.1162.11.1S--SAs163.21.3163.21.1163.31.3163.20.9163.11.1163.11.3**As 3d:**As--S~3~43.01.343.01.343.01.342.91.242.91.542.91.3As--S~2~As42.51.342.51.342.41.242.51.242.51.042.51.3As--SAs~2~42.01.342.01.342.01.342.11.242.01.542.01.3As--As~3~ -- -- -- --41.51.341.51.2 -- -- -- -- The valence band spectra of as-deposited and illuminated by laser As~*x*~S~100-*x*~ (*x* = 40, 45, 50) thin film surface nanolayers were also measured and shown in Fig. [2](#Fig2){ref-type="fig"}. The general view of these spectra for all As--S compositions is similar and correlates well with the valence band spectra of As~40~S~60~ films \[[@CR25], [@CR26]\]. For a better understanding of the structural changes caused by laser light illumination, differential VB spectra were constructed (Fig. [3](#Fig3){ref-type="fig"}, bottom part).Fig. 3Valence band spectra of as-deposited (*black lines*) and illuminated by green (λ = 532 nm) laser light (*green lines*) As~40~S~60~, As~45~S~55~, and As~50~S~50~ thin film nanolayers (*top part*). Differential (illuminated minus as received) original (*blue spiked lines*) and smoothed (*red thick lines*) spectra (*bottom part*) Raman Spectroscopy of As~40~S~60~, As~45~S~55~, and As~50~S~50~ Glasses and Films {#Sec6} --------------------------------------------------------------------------------- Results of XPS measurements give a possibility to analyze the structure of investigated materials at micro-level (short-range order) and to determine structural units which form the substance. For a better understanding of the nature of processes and stimulated structural changes, it is necessary to investigate the structure of the samples at the extended scale range (i.e., medium range order) in order to determine the macrostructure of materials. The cage-like molecules, rings, chain-like and bigger clusters in the structure of the As--S glasses and films can clearly be detected and identified from the Raman spectra \[[@CR27]\]. Also, the photon energy-dependent micro-Raman spectroscopy can successfully be used for monitoring the photoinduced molecular transformations \[[@CR7], [@CR12]\]. Therefore, this technique was used for complex investigation of the structure and induced transformations in As~40~S~60~, As~45~S~55~, and As~50~S~50~ films. The Raman spectra of source As~40~S~60~, As~45~S~55~, and As~50~S~50~ glasses and corresponding As--S thin films are shown in Fig. [4](#Fig4){ref-type="fig"}. As can be seen, the Raman spectra of g-As~40~S~60~ demonstrate a broad band with the maximum at 340 cm^−1^ and shoulders at 310 and 380 cm^−1^. The main band centered at 340 cm^−1^ is a characteristic band of symmetric As--S vibrations in AsS~3~ pyramids. The shoulders at 310 and \~380 cm^−1^ are connected with the assymetric As--S vibrations in AsS~3~ pyramids and As--S--As vibrations of "water-like" molecule, respectively. The Raman band at 130 cm^−1^ can be connected with deformational vibrations of --As--S--As-- and --S--As--S-- structures. In addition to these bands, the very weak features at 143, 165, 186, 220, 230, and 360 cm^−1^, connected with homopolar As--As bonds and realgar As~4~S~4~ inclusions, and small intensive band at 490 cm^−1^ associated with S--S bonds are detectable in the Raman spectra of As~40~S~60~ glass. In contrast with the Raman spectra of stoichiometric glass, the broad band in the region of As--S valence vibrations (\~300--400 cm^−1^) in the Raman spectra of corresponding As~40~S~60~ films clearly show the double-peak structure. The simultaneous increases in intensities of 360 cm^−1^ Raman band and bands in the region of molecular and As--As valence band vibrations (100--300 cm^−1^) can indicate the increasing of the concentration of cage-like As~4~S~4~ molecules in As~40~S~60~ films in comparison with those found in the structure of corresponding target glass. As can be seen from intensities of 490 cm^−1^ bands (Fig. [4](#Fig4){ref-type="fig"}, curve 1), the concentration of homopolar S--S bonds in the structure of As~40~S~60~ films is larger than in As~40~S~60~ glass. At the same time, the new band at \~270 cm^−1^ is detected in the Raman spectra of As~40~S~60~ films. This band is assigned to the vibrations in As-rich As~4~S~3~ cage-like molecules \[[@CR28]\].Fig. 4Raman spectra of As~40~S~60~ (*1*), As~45~S~55~ (*2*), and As~50~S~50~ (*3*) target glasses (*dotted line*) and corresponding thin films (*solid line*) The Raman spectra of As~45~S~55~ and As~50~S~50~ glasses are very similar (Fig. [4](#Fig4){ref-type="fig"}, curves 2 and 3). The main contributions in the Raman spectra of both glasses originate from As~4~S~4~ cage-like molecules. Weak band at 270 cm^−1^ characteristic of As~4~S~3~ molecules was detected, and no S--S bonds (490 cm^−1^ Raman mode) were found for both glass compositions. The difference in the Raman spectra of As~45~S~55~ and As~50~S~50~ glasses is connected with redistribution of 340 and 360 cm^−1^ band intensities only. In contrast with the glasses, the Raman spectra of As~45~S~55~ and As~50~S~50~ thin films are different. The main differences are connected with the intensity of Raman band at 270 and 360 cm^−1^. These bands are more intensive in the Raman spectra of As~45~S~55~ films indicating the drastic separations of cage-like As~4~S~3~ and As~4~S~4~ molecules from pyramidal network. Also, the very weak band at \~490 cm^−1^ (S--S bonds) is detected in the Raman spectra of both As~45~S~55~ and As~50~S~50~ films. Discussion {#Sec7} ========== Atomic Stoichiometry of As--S films {#Sec8} ----------------------------------- From the core level spectra of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) thin film surface nanolayers which are shown in Fig. [3](#Fig3){ref-type="fig"}, the atomic concentrations and As to S ratios of as-deposited and illuminated by green laser light samples were calculated. The appropriate values are given in Table [2](#Tab2){ref-type="table"}.Table 2Atomic concentrations, As/S ratio of As~40~S~60~, As~45~S~55~, and As~50~S~50~ nanolayers calculated from XPS data (the values of As/S ratio for the bulk glasses are given in parentheses for comparison) and contribution (area, ±5%) to the core level of each doublet of individual components determined from curve fitting of S 2p and As 3d XPS spectraElement/Core level/ComponentAs~40~S~60~As~45~S~55~As~50~S~50~As-depositedIlluminatedAs-depositedIlluminatedAs-depositedIlluminatedAs, %42.943.048.951.045.445.8S, %57.157.051.148.954.654.2As/S0.75 (0.67)0.750.96 (0.82)1.040.83 (1)0.84**S 2p:**S--As~2~, %84.890.493.094.386.489.2S--SAs, %15.29.67.05.713.69.8**As 3d:**As--S~3~, %90.292.144.516.557.575.6As--S~2~As, %5.74.143,845.835.621.4As--SAs~2~, %4.13.86.627.66.93.0As--As~3~, %----5.110.1---- As it can be seen, the thermal evaporation of the bulk chalcogenide glass of As~40~S~60~ composition causes the As~42.9~S~57.1~ composition of deposited thin film. Laser light illumination with near-bandgap photon energy leads to further slight arsenic enrichment. More As-rich thin film in comparisons with target composition is obtained when the As~45~S~55~ glass is evaporated (see Table [2](#Tab2){ref-type="table"}). Further arsenic content increment from 48.9% in the as-deposited sample to 51.0% in the sample illuminated by a green laser light during 90 min takes place. Correspondingly, the appropriate As to S ratio is changed from 0.96 to 1.04. Unexpectedly, the thermal evaporation of As~50~S~50~ glass leads to deposition of thin film with the As/S ratio which is less than for the bulk glass (As~45.4~S~54.6~ composition) (Table [2](#Tab2){ref-type="table"}). The laser treatment of this sample causes small arsenic enrichment, but the ratio between As and S remains far from the appropriate value in the bulk glass. Such deviations of the thin film stoichiometry from the bulk glasses and further changes to them under the external (laser) influence with photon energy close to the band gap of investigated materials can be understood and explained from a detailed component analysis \[[@CR29], [@CR30]\]. Component Analysis of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) Thin Films Under External Influence {#Sec9} ---------------------------------------------------------------------------------------------- As mentioned above, the fit of core level spectra of all films (before and after treatment) demonstrates the presence of structural units with homopolar bonds (see Fig. [2](#Fig2){ref-type="fig"}). Because of sulfur is twofold coordinated and arsenic is threefold coordinated in As-S system, the As~40~S~60~ films in ideal composition should demonstrate the water-like S--As~2~ and pyramidal As--S~3~ components only in their S 2p and As 3d core level spectra, respectively. However, the homopolar As--As bonds were detected in the As 3d core level spectra of all As~*x*~S~100-*x*~ (*x* = 40, 45, 50) nanolayers which is expected from As-enrichment (*x* \> 40 at. % As) of their top surface. In accordance with this for the As~40~S~60~ composition, it was found of 5.7% s.u. which are assigned to arsenic bonded to two sulfur and one arsenic atoms, and of 4.1% s.u. which mean the presence of the arsenic bonded to one sulfur and two arsenic atoms (peaks 2, 2' and 3, 3', respectively). The contributions of these two As-rich components are much significant in As 3d core level spectra of As~45~S~55~ film surface (see Table [2](#Tab2){ref-type="table"}). Moreover, the fourth component, As--As~3~ is appeared with 5.1% contribution, which is reasonable for calculated composition (As~51.1~S~48.9~). Finally, the thin film obtained by thermal evaporation of the As~50~S~50~ target glass contains 35.6% of As--S~2~As s.u. and 6.9% of As--SAs~2~ s.u. apart from the pyramidal one. Despite to arsenic enrichment of all three as-deposited As-S samples in comparison with the stoichiometric composition, the S 2p core level spectra contain a component with the homopolar S--S bond (Fig. [2](#Fig2){ref-type="fig"}). There is a strong correlation between the As to S ratio of as-deposited samples and the percentage of S--SAs s.u. in the appropriate S 2p core levels. However, the further explanation of the existence of S--S bonds in As-rich structures is needed. The properties and micro structure of vapor-deposited films depend on the deposition methods and conditions. Therefore, the resulting film structure can be different from the structure of the corresponding bulk glasses as established earlier \[[@CR31]\]. The As--As bond formation in the as-deposited As~2~S~3~ film was detected with using X-ray diffraction technique. On the basis of the arsenic enrichment of the film and detected by mass spectroscopy fragmentation into S~2~ and As~4~S~4~ during the evaporation of the bulk As~2~S~3~ glass, the formation of a sheet-like open structure of the film is supposed \[[@CR31]\]. Also, it is pointed out that the As--As bonds may be incorporated as S~2~As--AsS~2~ units, as in As~4~S~4~ molecules as determined using extended X-ray absorption edge fine structure and Raman and IR spectroscopy. Apart from this, the dominance of As~4~S~4~ and sulfur molecules in as-deposited films were shown by neutron diffraction study \[[@CR31]\]. The mass spectrometry study shows the presence of S~2~ and different AsS particles in the gas phase of As--S system \[[@CR32]\]. Therefore, the presence of S--S s.u. (S 2p spectra) in the structure of all As--S films and the appearance of As--S~2~As s.u. in the As 3d core level spectra of even stoichiometric As~40~S~60~ composition can be understood. Moreover, apart from the composition of the target glass, the type of molecules in vapor plays a significant role in the formation of the film composition and structure. This way, the differences in compositions of the films and corresponding source materials can be explained. Additional support of mentioned As-enrichments of top surface As--S nanolayers can be confirmed by Raman spectroscopy. In particular, the biggest arsenic enrichment is found for As--S film deposited from As~45~S~55~ glass where the most significant contribution of As-rich As~4~S~3~ molecules is detected (Fig. [4](#Fig4){ref-type="fig"}, curve 2). Near-bandgap laser light illumination of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) samples causes decreasing of the contribution of components with the homopolar S--S bonds in all the samples (Table [2](#Tab2){ref-type="table"}). This phenomenon was observed in our previous investigations \[[@CR19]\] and was explained by the processes of the structural ordering under the laser light illumination. In addition, the decreasing of components with homopolar As--As bonds in the structure of As~40~S~60~ and As~50~S~50~ nanolayers under the near-bandgap laser light illumination was observed (Fig. [2](#Fig2){ref-type="fig"}, Table [2](#Tab2){ref-type="table"}). This is in accordance with the results of our in situ under-bandgap laser light illumination of As~2~S~3~ nanolayers \[[@CR19]\]. The decreasing of concentration of homopolar As--As bonds in the structure of As~40~S~60~ and As~50~S~50~ films under laser illumination correlates well with the partial disappearance of the S--S bonds and creation of new As--S bonds. However, the different behavior in compositional and structural changes under laser illumination was observed for As~45~S~55~ films. In contrast with As~40~S~60~ and As~50~S~50~ nanolayers, the increasing of concentration of components with As--As bonds was detected in As~45~S~55~ films as a result of the near-bandgap laser light illumination (Fig. [2](#Fig2){ref-type="fig"}, Table [2](#Tab2){ref-type="table"}). Similar increasing was also detected in As~2~S~3~ nanolayers when over-bandgap laser light illumination was applied \[[@CR19]\]. This effect was explained by atomic movement of As from deeper to top layers under the laser treatment, leading to As-enrichment of the sample surface. Such movement appears due to a creation of electric field gradient which is driving force on dipoles and charged defects resulting in mass transport. The larger magnitude of laser-induced changes in As--S system was found for As-rich compositions with As~4~S~4~ inclusions \[[@CR7]\]. It should be noted here that XPS spectroscopy reveals the most As-enriched composition of As--S film prepared from As~45~S~55~ glass among studied As~*x*~S~100-*x*~ (*x* = 40, 45, 50) films. Moreover, the significant As enrichment of the As~45~S~55~ sample was confirmed by Raman spectroscopy (Fig. [4](#Fig4){ref-type="fig"}, curve 2) where the 270 cm^−1^ Raman band characteristic of As~4~S~3~ molecules show maximal intensity. In this manner, the specific behavior and the structural rearrangement of the As~45~S~55~ nanolayers are conditioned by a considerable number of As-rich s.u., particularly As~4~S~3~. This can stimulate laser-induced mass transport effect resulting in further arsenic enrichment of the sample surface. Valence Band Spectra of As--S films {#Sec10} ----------------------------------- In general, the valence band can be determined as the highest range of electron energies which can be occupied at absolute zero temperature \[[@CR33]\]. According to Mott and Devis model, the valence band of amorphous materials contains the states formed by defect centers \[[@CR34]\]. For the As--S system, the top of the VB is formed by lone-pair 3p electrons of sulfur (at \~3 eV), as 4p and S 3p levels (bonding electrons) are situated at \~5 and \~7 eV, respectively. Next energy band is located lower than 10 eV and formed by the S 3s and As 4s electrons. These data were confirmed by DFT electronic structure calculations of As- and S-centered s.u. \[[@CR19]\]. A qualitative comparison of the VB spectra of As~2~S~3~ nanolayers investigated in situ \[[@CR19]\] with the VB spectra of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) films show the differences connected with the presence of additional states at energies ranged from −1.7 to 0.6 eV (Fig. [3](#Fig3){ref-type="fig"}). According to the calculated data, the formation of homopolar As--As bonds leads to the appearance of the energy levels in the band gap of the As--S structures \[[@CR35], [@CR36]\]. The concentration of As--As bonds in As~*x*~S~100-*x*~ (*x* = 40, 45, 50) films was found to increase in order: As~40~S~60~, As~50~S~50~, and As~45~S~55~ (Table [2](#Tab2){ref-type="table"}). Taking into account, the changes of the concentrations of s.u. with homopolar As--As bonds induced by laser treatment (decreasing for As~40~S~60~, As~50~S~50~ composition and increasing for As~45~S~55~ structure) (Table [2](#Tab2){ref-type="table"}) and intensities of electronic states in the VB spectra of As--S films (from −1.7 to 0.6 eV) (Fig. [3](#Fig3){ref-type="fig"}) can be assumed that they are formed by structural units with As--As bonds. The main changes in the electronic structure of As--S samples induced by laser light illumination can be selected (highlighted regions in Fig. [3](#Fig3){ref-type="fig"}). The changes in these regions (denoted as A, B, and C) can clearly be seen from the differential valence band spectra (Fig. [3](#Fig3){ref-type="fig"}. bottom). The right highlighted region (A) in the differential VB spectra of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) films points out to the changes of the states in the band gap of the structures. Middle highlighted region (B) indicates that the band gap decreases in the As--S samples of the As~40~S~60~ and As~50~S~50~ compositions and increases in the As~45~S~55~ structure under the near-bandgap laser light illumination. It should be noted that these results correlate with the shift of absorption edge measurements. And finally, left marked region (C) demonstrates common decreasing of the concentration of S--S homopolar bonds in all As--S films under laser treatment (see Figs. [2](#Fig2){ref-type="fig"} and [3](#Fig3){ref-type="fig"}, Table [2](#Tab2){ref-type="table"}). Profilometry Analysis of Laser Induced Relief Formation in As~*x*~S~100-*x*~ (*x* = 40, 45, 50) Films {#Sec11} ----------------------------------------------------------------------------------------------------- As it was mentioned above, the increasing of the components with the As--As bonds under the near-bandgap laser light illumination takes place due to a creation of electric field gradient resulting in mass transport. In order to examine of this effect, the additional experiments were performed. The as-deposited films of all three compositions were illuminated by green laser light through the copper mash with grating period of 60 μm during time sufficient for saturation of shift of absorption edge, measured previously (see previous chapter). Then, the surface of irradiated As--S samples was examined by AMBIOS XP-1 type profilometer with 10 nm vertical resolution (stylus tip radius---2.5 μm). For the profilometer measurements, a 0.5 mg load was applied. This load was small enough to make the accurate profiling without destroying the surface morphology. Results are shown in Fig. [5](#Fig5){ref-type="fig"}.Fig. 5Profiles of the surface of As~40~S~60~, As~45~S~55~, and As~50~S~50~ thin films illuminated by green (λ = 532 nm) laser light through the copper grid As can be seen, the laser light illumination does not change the shapes of the surfaces of both As~40~S~60~ and As~50~S~50~ thin films. Thus, it can be concluded that laser illumination was not influenced on morphology of the surface of these samples. However, the different behavior in laser-induced transformations and surface morphology changes was discovered for As~45~S~55~ thin film (see Fig. [5](#Fig5){ref-type="fig"}, curve 2). It is clearly seen that under the laser light illumination the "wave-like" relief on the surface of As~45~S~55~ film is formed. The parameters of the induced grating can be seen in the insert of Fig. [5](#Fig5){ref-type="fig"}. This relief corresponds to the period of the mash grating. Taking into account that the As~45~S~55~ composition demonstrates peculiarities and opposite induced phenomena (shift of the absorption edge, stoichiometry and local structure changes, untypical VB shift, and shape changes) in comparison with the As~40~S~60~ and As~50~S~50~ thin films and contains the largest concentration of As~4~S~3~ cage-like molecules, it can be concluded that the presence of polar As~4~S~3~ molecules (which are sensitive to the electric field generated by the laser) is responsible for observed laser-induced mass transport effect. The drift and re-arrangements of this molecules results in photoexpansion of illuminated areas. The observed phenomena can be used for optical grating formation, controlled laser surface modification, laser induced surface activation, etc. Conclusions {#Sec12} =========== The local and molecular structures of As~*x*~S~100-*x*~ (*x* = 40, 45, 50) thin film surface and their transformations induced by coherent near-bandgap laser illumination have been investigated using XPS and Raman spectroscopy. The optical properties and induced transformation of surface morphology of As--S nanolayers have been also studied by means of absorption edge spectroscopy and 2D profilometry. A significant difference in surface stoichiometry between amorphous As--S films and composition of corresponding target glasses was established, and it was found to be related with the peculiarities in molecular constituent of gas phase during the deposition process, indicating that the type of molecules in vapor plays a crucial role in resulting film composition. Near-bandgap laser illumination decreases the concentration of the homopolar S--S bonds in the structure of all As~*x*~S~100-*x*~ (*x* = 40, 45, 50) nanolayers. However, the decreasing of the concentration of homopolar As--As bonds upon laser illumination was observed in the structure of As~40~S~60~ and As~50~S~50~ films only. In contrast with As~40~S~60~ and As~50~S~50~ films, the contribution of As--S~2~As and As--SAs~2~ components and appearance of a new As-rich As--As~3~ s.u. in the structure of As~45~S~55~ thin film during laser illumination were detected. Moreover, this particular film (As~45~S~55~) demonstrates peculiarities in laser-induced shift of the absorption edge, in Raman spectra, and finally, in effect of induced surface morphology transformation. The results of Raman investigation of As--S films indicate the presence of As-rich As~4~S~3~ molecules in the structure of As~45~S~55~ nanolayers in largest concentration among studied samples. Therefore, the As~4~S~3~ molecules were found to be responsible for drastic difference in behavior of absorption edge spectra and surface morphology transformation of As~45~S~55~ nanolayers during near-bandgap laser illumination. The presence of these As~4~S~3~ structures in the structure of As~45~S~55~ nanolayers results in laser-induced mass transport effect observed for this material and can be useful for optical grating formation and related external nanofabrication technologies. BE : Binding energy ChG : Chalcogenide glass FWHM : Full width at half maximum VB : Valence bands XPS : X-ray photoelectron spectroscopy O.K. and R.H. gratefully acknowledge support from the Hungarian Academy of Sciences within the Domus Hungarica Scientiarum et Artium. The work/publication is supported by the GINOP-2.3.2-15-2016-00041 project. The project is co-financed by the European Union and the European Regional Development Fund. Authors' Contributions {#FPar1} ====================== All authors (OK, RH, ACh, VT, MV, and VM) equally contributed in developing the general idea and methodological aspects of performed investigation. OK, RH, and VM prepared the source glasses and synthesized As~*x*~S~100-*x*~ thin films of different (*x* = 40, 45, 50) compositions. OK, VT, and ACh performed the XPS, absorption edge spectroscopy, and profilometer measurements to characterize the samples. RH and MV performed the Raman spectroscopy measurements and spectral interpretation. VM carried out the general control of the processing and analysis of the results. All authors read and approved the final manuscript. Competing Interests {#FPar2} =================== The authors declare that they have no competing interests.

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi:10.1111/ajt.16252 [^1]: Ulrik Stervbo and Timm H. Westhoff share senior authorship and contributed equally

Introduction ============ Juvenile Systemic Lupus erythematosus (JSLE) is a common rheumatologic disease with a global incidence rate of 0.28-0.9 per 100,000 population and prevalence rate of 500 cases in 1 million people^\[[@B1],[@B2]\]^. Incidence rate of JSLE differs in various racial and ethnical populations and is more prevalent in Asian populations^\[[@B3]\]^. This disease is able to involve numerous organ systems such as cardiovascular, renal, hematological, musculoskeletal, neurological and mucocutaneous systems^\[[@B2],[@B4],[@B5]\]^. Although SLE mostly occurs in the ages of 20-40 years, however it can occur in all age groups including childhood and its prognosis will be better with increasing age^\[[@B6]\]^. Prognosis in children in comparison with adults is poorer, probably related to more frequent renal and neurological involvement in JSLE^\[[@B7]\]^. Several studies have shown that severity and tissue damage in children is higher than in adults with SLE and majority of these complications occur within the first years after manifestation of the disease^\[[@B5]-0\]^. SLE is a chronic disease with a various severity from a mild prolonged to an acute life threatening disease. In the past years pediatric lupus was considered as a fatal disease, but nowadays its clinical course and prognosis has become better in children; so that, 5-year survival is now more than 90%^\[[@B9]\]^. Since clinical manifestations and outcomes are different in various racial and ethnical groups, we designed this study to determine mortality and morbidity rates in Iranian children with systemic lupus erythematosus. Subjects and Methods ==================== In this cross sectional study, we extracted data from our pediatric rheumatology registry. All JSLE patients who had been diagnosed between 2004- 2010 were selected for this study. Our center is a tertiary pediatric medical center in Iran and it is one of the two pediatric rheumatology centers in Iran. All patients met diagnostic criteria of lupus according to the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus^\[[@B10]\]^. Patients were divided into 3 groups according to age at the time of study: less than 7 years, between 7-14 years and more than 14 years. Moreover, demographic data, first clinical manifestation, follow-up period, morbidity and mortality of the patients and family history of rheumatologic diseases were evaluated. All patients were treated with high dose steroid, hydroxychloroquine and/or immunosuppressives based on organ disorder. New cases and patients who had been followed up less than six months were excluded from the study. In this study morbidity was considered as a permanent symptom or complication of the disease or treatment and remission was defined as a state or period during which the clinical and laboratory of the disease are abated. This study was approved by ethical committee of Tehran University of Medical Sciences. Data was analyzed by SPSS software ver17. We used Fisher exact test and Chi-2 test for statistical analysis. *P*-value less than 0.05 was considered as significance level. Findings ======== Totally, 120 patients with lupus referred to our center during this study. Female to male ratio was 3.3:1 (92 females and 28 males). Out of these 120 patients, 4 (3.3%) cases were younger than 7, 30 (25%) were 7-14 years and 86 (71.7%) older than 14 years. Mean age at the time of manifestation of disease was 10.34± 2.9 years. Mean follow-up period was 56±32 (range: 12-120) months. All patients have had constitutional symptoms. Joint, mucocutaneous and hematological manifestations are respectively the most common organ involvement. Frequencies of organ involvement have been summarized in [Table 1](#T1){ref-type="table"}. Joint and bone involvement were more common in patients aged above 7 years at the time of diagnosis and the time of study (*P*=0.01). Indeed, oral ulcer and ophthalmic involvement in boys were significantly higher than in girls (*P*\<0.05). ###### Frequency of different types of organ involvement in the 120 children with JSLE **Morbidity** **Frequency** **Morbidity** **Frequency** ------------------------------- --------------- -------------------------- --------------- **Joint involvement** 91% Bone involvement 52% **Mucocutaneous involvement** 90% Cardiac involvement 51% **Hematological involvement** 86% Pulmonary involvement 41% **Liver dysfunction** 40% Neurological involvement 39% **Endocrine involvement** 70% Ophthalmic involvement 28% **Renal involvement** 60% Vascular involvement 24% JSLE: Juvenile systemic lupus erythematosus ###### Frequency of different types of morbidity in 120 patients with JSLE according to disease remission **Morbidity** **Remission** ***P.*** **value** ------------------------------------------------------- --------------- -------------------- ------- **Bone and joint involvement** 26.1% 29.2% 0.7 **Gastrointestinal and increase liver transaminases** 26.3% 10% 0.1 **Cardiovascular involvement** 32.7% 7.7% 0.005 **Renal involvement** 37.3% 10.5% 0.001 **Nervous system involvement** 23.3% 22.5% 0.9 **Ophthalmologic involvement** 24% 25% 0.9 **Hematological involvement** 25.5% 18.2% 0.5 **Endocrine involvement** 23.1% 11.8 0.3 JSLE: Juvenile systemic lupus erythematosus Generally, 24% of the patients experienced remission. Frequency of different types of morbidity in our patients according to experience of remission is summarized in [Table 2](#T2){ref-type="table"}. Frequency of cardiovascular and renal involvement was significantly higher in the patients with active diseases (*P*=0.005 and *P*=0.001, retrospectively). Mortality rate was 10% (12 cases). [Table 3](#T3){ref-type="table"} shows the effect of different variables on mortality rate. According to this table, there is a significant association between mortality rate and cardiac \[*P=*0.02, OR=4.90 (1.24-19.63)\], pulmonary \[*P*=0.01, OR=10.17 (1.24-19.63)\] and increased liver transaminases \[*P*=0.04, OR=1.19 (1.24-19.63)\] involvement. Survival rate was 97% (SD ±0.02) after 1 year and 89% (SD ±0/04) after 5 years. There was no significant difference between morbidity and mortality rates in the groups with and without positive family history. Discussion ========== Systemic lupus erythematosus is a rheumatologic disease that may involve children and adults^\[[@B2]\]^. Several studies have evaluated mortality and morbidity rate in the children with SLE worldwide^\[[@B1],[@B7],[@B11]-[@B15]\]^. Some authors have found that prognosis in children is poorer than in adults^\[[@B6],[@B7],[@B11]\]^. Investigations in different areas in the world show some differences in clinical features and also in morbidity and mortality rates^\[[@B1],[@B4],[@B6],[@B12],[@B15],[@B16]\]^. We evaluated morbidity and mortality in Iranian children with JSLE. Mortality rate in our series was 10%. This is approximately similar to Gulay\'s^\[[@B12]\]^ and Lumina\'s results^\[[@B16]\]^, pointing to a mortality rate of 11.5% and 11.8% respectively. In our study no significant difference was found in male and female paticnts. In addition, our results show no significant difference in mortality rate between various age groups, similar to Vachvanichsanong\'s study^\[[@B14]\]^. ###### The effect of different variables on mortality rate in our patients ----------------------------------------------- --------------------------------------- ---------------- -------------------- **Variables** **Frequency** ***P*** **-value** **Gender** Male 5 (17.9%) 0.1 Female 7 (7.6%) **Age at the time of study** \<7 years 25% 0.6 7-14 years 10% \>14 years 9.3% **Age at initiation of the disease** \<7 years 9.1% 0.5 \> 7 years 10.2% **Family history for rheumatologic diseases** No 11 cases (10%) 1 Yes 1 case (10%) **Organ involvement** Liver (increased liver transaminases) 100% 0.04 Renal 83% 0.09 Neurological 82% 0.15 Cardiovascular 75% 0.02 Pulmonary 75% 0.01 ----------------------------------------------- --------------------------------------- ---------------- -------------------- JSLE: Juvenile systemic lupus erythematosus The most frequent organ involvement in the patients who died due to JSLE complications included liver (increased liver transaminases) in 100%, kidney in 83%, nervous system in 82% and heart and lungs in 75% of patients. We found a significant association between mortality rate and liver, cardiac and pulmonary involvement. Similarly, in a study in Mexico, mortality rate was significantly higher in the JSLE patients with cardiac and pulmonary complications^\[[@B15]\]^. Unlike previous reports, in our study liver dysfunction was more common than hepatomegaly (40% vs 13% of the cases). There are few reports on liver involvement in JSLE. Hepatomegaly and liver dysfunction were reported as common findings in active phase of lupus in about 25% and 50% of patients^\[[@B17]\]^. It can be a clue for disease activity or due to lupoid hepatitis, infection, drug reaction or toxicities and complication of the disease course^\[[@B18],[@B19]\]^. In adult patients with SLE, liver dysfunction has been reported in 33-60% of the patients^\[[@B19],[@B20]\]^. In our patients 1-year and 5-year survival rate was 97% and 89%, respectively. With early diagnosis and progress in treatment regimens of pediatric lupus with new immunosuppressive drugs (such as Miclofenolate-Mefotile and Cyclosporine) and decrease in prevalence of infections, survival of patients with JSLE has become better during last 3 decades^\[[@B21]\]^. Similar to other studies 5-year survival in children with JSLE was between 82% to 92%^\[[@B22]-[@B4]\]^. Reports on frequency of various morbidities have been different^\[[@B5],[@B13],[@B15],[@B25]-[@B27]\]^. The most common morbidity in our study was joint involvement (91%) that was similar to Marini\'s study^\[[@B13]\]^. In our patients joint, skin and hematological involvement were the most common morbidities alike Hui-Yuen\'s series, in which cutaneous manifestations, arthritis and hematological abnormalities the most common morbidities at the time of SLE diagnosis^\[[@B28]\]^. Although these findings are similar to those of other researchers, the next most common morbidity in our study was gastrointestinal involvement which is different from Hui-Yuen\'s study that introduced renal involvement as the next common morbidity. Joint involvement in SLE may be from minor joint pain to severe arthritis. Severe joint deformity or Jaccoud's arthropathy is a rare finding in JSLE^\[[@B29]\]^. Overlap of JSLE and juvenile idiopathic arthritis is a rare syndrome with erosive joint involvement which has been named RHUPUS syndrome. Three cases of our patients had RHUPUS syndrome and we reported them in previous report^\[[@B30]\]^. Based on age at the time of diagnosis, only joint involvement showed significant difference between various age groups, being more frequent in the group of 7-year olds and older. Alike Descloux\'^\[[@B31]\]^, prevalence of neurological disorders in our study was 38%. In Muscal\'s patients headache and mood disorders were the most common neurological disorders^\[[@B25]\]^. In the present series prevalence of headache was 44%. In our previous report on 55 cases, 43% of patients with SLE had lupus headache^\[[@B32]\]^. In the present study majority of complications occurred more frequently in male gender, only frequency of oral ulcer and ophthalmic complications show statistically significant difference in males and females. Similar to Niaudet^\[[@B26]\]^ and Vachvanichsanong^\[[@B14]\]^ findings, we observed that lupus in male gender is associated with higher mortality, frequent hospitalization and poor outcome. In this study, having a family with rheumatologic disease had no effect on morbidity or mortality. In our patients, frequency of cardiovascular involvement and renal involvement was significantly lower in remission group. On the other hand, one fifth of the patients with cardiac or renal involvement experienced remission compared to the patients without involvement. The sample size of this study was higher than that reported previously. As a limitation of this study, we did not study the causes of death related to heart, lung and kidney. These manifestations may be more related to infection and/or multiple organ failure due to lupus activity and vasculitis. However, physicians should be alert about these organ involvements as a predictive factor for mortality regardless of cause of involvement in JSLE. Conclusion ========== Mortality rate in Iranian JSLE patients was 10%. Liver, cardiac and pulmonary involvement is associated with higher mortality. Follow up with closer intervals is recommended in the cases with liver, heart and pulmonary involvement. This study was part of a post graduate dissertation of Dr F. Tavangar-Rad and was supported by Vice‐Chancellor for Research of School of Medicine, Tehran University of Medical Sciences. ***Conflict of Interest:*** The authors declare that there is no conflict of interests. Authors' Contribution ===================== V. Ziaee: Concept / design, critical revision of the manuscript F. Tavangar-Rad: Acquisition of data, manuscript preparation M.H. Moradinejad: Concept / design, data interpretation F. Tahghighi: Acquisition of data, data interpretation All authors approved final version of the paper.

IMMUNOHISTOCHEMISTRY AS A LABORATORY TEST {#s0010} ========================================= Although they do not publicize it, pathologists have long recognized their fallibility.[@bib1] As a result, more objective means of validating morphologic judgments have been sought. Stains using histochemical methods are of value in accentuating morphologic features but do not provide objective evidence of the lineage or biologic potential of a cell. The objective of immunohistochemistry is to use antibodies to identify antigens, increasing the specificity of the stain for the tissue with which it reacts. In doing so, immunohistology has transformed surgical pathology from a highly subjective discipline into a much more objective science, while still taking advantage of the light microscope and standard morphologic practices. *Immunohistochemistry*, as the name implies, is the combination of histology and immunology. The resulting technique is a powerful tool that not only enables pathologists to detect whether particular antigens are present within a given cell but also allows the identification of the microanatomic (cellular) location of the antigen. These abilities permit the lineage of cell populations to be identified, an important consideration when confronted with a poorly differentiated neoplasm of undetermined origin. The technique is also useful in defining distinct populations of cells within the same lineage and defining functional differences. In addition, this technique preserves the histologic architecture and enables the pathologist to confirm that the positive cells are the cells in question. This confirmation is not possible with molecular methods, such as reverse transcriptase polymerase chain reaction or standard flow cytometry methods. Immunohistochemistry is used by a variety of disciplines to study a wide range of questions. This chapter discusses the application of this technology in surgical pathology, in which immunohistochemistry has had a profound and fundamental impact on the practice of pathology.[@bib1], [@bib2] Technical Considerations {#s0020} ------------------------ Immunohistochemistry has the potential to transform surgical pathology from a subjective art to an objective science, based on the recognition of cells by microscopic methods. Although this potential has resulted in its almost universal use, immunohistochemistry has not produced uniformly high standards of practice.[@bib3], [@bib4] Therefore, certain technical considerations must be borne in mind to ensure the accuracy of results. It is self-evident that the quality of an immunohistochemical stain depends on the integrity of an antibody-antigen interaction and on the extent to which the relevant antigen has been preserved during tissue fixation and processing.[@bib1] There is a high degree of variability in the way tissues are initially prepared. These variations include differences in the fixative used; the amount, age, and pH of the fixative; how long the tissue sat unfixed before being placed in fixative; the thickness of the tissue when first placed in fixative; and the time the tissue is left in fixative. All these variables, which may not alter the results of routine hematoxylin-eosin (H&E) staining to a significant degree, can lead to widely discrepant results when it comes to immunohistochemistry. These variables cannot always be predicted or remedied, but they can be mitigated to a great extent by the advent of successful and relatively simple antigen retrieval methods (discussed later).[@bib5], [@bib6] In addition to the variables in tissue fixation, as in all other laboratory tests, the reagents and techniques used must be optimized and thoroughly validated to ensure consistent, reliable, and clinically meaningful results. When developing an immunohistochemical protocol, it is important for each laboratory performing a test to validate every reagent used. This validation includes a determination of the specificity and the optimal working dilution of each primary antibody, secondary antibody, linking antibody, labeling reagent, and substrate. Repeat validation is required for each new lot of reagents because of variations in origin, composition, concentration, and specificity that can occur among different lots even when supplied by the same company.[@bib1] Also, reputable manufacturers should be used. The higher standards and more rigorous quality control of products from the better manufacturers have been counterbalanced by the concurrent proliferation of smaller manufacturers that are able to produce or otherwise acquire and market large numbers of different antibodies through monoclonal antibody technology, molecular engineering, and the like. In addition to the need for high-quality antibodies and reagents, proper incubation times and ideal temperatures for each antibody must be determined. The optimal buffering agent must be determined, as well as the need for any predigestion techniques or antigen enhancement procedures. ### Premanufactured Kits {#s0030} Premanufactured, all-inclusive kits have been marketed in an attempt to simplify the performance of immunohistochemistry; for example, these kits obviate the need for each individual clinical laboratory to validate each reagent because preoptimized working dilutions and recommended working protocols are provided. Nevertheless, there are some pitfalls that should be kept in mind when working with premanufactured kits. The protocols and reagents have been formulated to work on the prototype tissue used by the manufacturer and may not be as effective on the actual tissues tested because of laboratories' different fixation and processing protocols, all of which may have adverse effects on the results. Adjustments in the recommended protocol are necessary to optimize the kits in each individual laboratory, which effectively means that the kits must be customized, and any changes made in the manufacturer\'s protocol require that the *entire* staining procedure be revalidated by the performing laboratory.[@bib1] This procedure may be complicated by the fact that many of the working dilutions of the reagents supplied are already at the critical level of sensitivity. ### Automated Staining {#s0040} Another method that can potentially enhance consistency and reproducibility is automation. A variety of automated immunostaining systems are now commercially available. The theoretical advantages of these systems over manual staining techniques include improved reproducibility, facilitation of interlaboratory comparisons, reduced reagent expenses, and increased technician productivity. Automation does not mitigate the need to thoroughly validate each step of the staining procedure or the need to evaluate every reagent used to ensure high-quality, consistent results. The same quality control issues that apply to manual staining apply to automated systems. As with manual staining, it is important that a complete reevaluation be performed if there is any departure from the validated protocol.[@bib1] Automation does not guarantee an optimal result. Finally, automation cannot replace the pathologist, who must choose the appropriate antibodies and then interpret the final result. ### Automated Image Analysis {#s0050} There is a growing need to be able to quantify immunohistochemical staining results, which is probably best accomplished by automated image analysis. Current systems, including the Automated Cellular Imaging System II (ACIS II Clarient, Aliso Viejo, Calif), have the ability to assess marker positivity in terms of both percentage positive and intensity of staining. The ability to quantify markers more precisely is especially important in the identification of targets of therapy. One example of this is Her-2/*neu*. When Her-2/*neu* protein expression by immunohistochemistry was compared with gene amplification by fluorescence in situ hybridization using routine manual methods and the assistance of a digital microscope, both accuracy and reliability were improved when the digital microscope was used.[@bib7], [@bib9000] Precise quantification of hormone receptors may also be important because there is evidence that in patients with high levels of hormone receptors, the addition of cytotoxic chemotherapy has a deleterious effect on outcome.[@bib8], [@bib9] Another exciting development in the field of automated cellular imaging is spectral imaging. This technology allows multiple markers to be assessed on the same slide---even on a single cell. Computer software can isolate a single chromogen from other chromogens present based on its emission spectrum. Automated cellular imaging provides greater objectivity and reproducibility and thus minimizes interobserver discrepancies. ### Positive and Negative Controls {#s0060} Immunohistochemical tests performed and interpreted in the absence of the appropriate controls are valueless and even dangerous. Minimal controls should include a tissue known to express the particular antigen of interest, processed in a manner analogous to that of the unknown tissue (the positive control), and a second section of the test specimen in which the primary antibody is replaced either by diluent or, better, by an irrelevant antibody of the same isotype, from the same species, and at the same concentration (the negative control). In the positive control, only cells expected to express the antigen should show positivity; all other cells and structural elements should be negative. In the negative control, there should be no specific staining. The "sausage" technique, in which samples of multiple tissues are gathered into a single tissue block, is a useful control method.[@bib10] Controls are performed for a variety of purposes; in addition to indicating whether a reaction occurred (or not), they are essential for judging the nature of the reaction. A vast array of immunohistochemical tests are judged not by a *positive* or *negative* result but by the intensity and localization of the result (a good example is Her-2/*neu* and hormone receptor analysis in breast cancer). Immunohistochemistry results should never be interpreted in the absence of the known positive results because the assessment of *quality* and *quantity* of the reaction is essential. ### Results and Reporting {#s0070} Interpretation of the results of immunohistochemical stains is the province of the surgical pathologist and is best accomplished by pathologists who have the appropriate level of experience not only in the morphologic aspects of diagnosis but also with regard to immunohistochemical findings. As in any other area of pathology, experience matters: A pathologist with little experience with immunohistochemistry, who runs a few different tests each week or month, will obtain very different results from a pathologist who performs and interprets immunohistochemical tests on a daily basis. As the impact of immunohistochemistry on surgical pathology increases, these differences will become more profound. As described previously, many factors influence the results. All these factors must be considered by the pathologist in interpreting the findings. Negative, weak, or uninterpretable results should lead to a repetition of the test after the use of antigen retrieval.[@bib11], [@bib12] One measure of antigen preservation is to test for expression of the intermediate filament vimentin, a fixation-sensitive protein that is typically expressed by vascular or connective components; this technique often serves as an internal indicator of the conservation or loss of antigenicity.[@bib11] Test results may also be affected by technical artifacts and by the nature of the tissue under study. For instance, if tumor cells are crushed, false-positive or nonspecific staining may be encountered. Nonviable areas of tissue from a necrotic tumor may also be a source of false-positive results, attributable in part to leakage of serum proteins (e.g., immunoglobulins). The subcellular distribution of immunoreactivity is critical to the interpretation of immunohistochemical results. For example, Her-2/*neu* shows membranous staining, whereas antibodies to estrogen and progesterone receptors produce nuclear staining. When unexpected staining patterns are observed with an antibody, the results should be discounted.[@bib1] To interpret the results effectively, the pathologist must have extensive knowledge of the staining patterns of the primary antibodies under consideration, including a detailed knowledge of tissue specificity and subcellular localization of the antigen, and an awareness of technical variables. Each laboratory performing immunohistochemical staining should have established written criteria for determining and reporting positive and negative findings for each immunohistochemical stain, with particular reference to stains that are expected to produce cell surface membrane, cytoplasmic nuclear, or extracellular staining. Although it seems obvious, it often is overlooked that staining should be recorded as *positive* only if it occurs in the expected cellular or tissue location. ### Validation and Proficiency Testing {#s0080} The Food and Drug Administration\'s increased attention to the reagents used in immunohistochemistry has undoubtedly contributed to an improvement in their quality.[@bib13] It is highly recommended that all laboratories performing diagnostic immunohistochemistry participate in the College of American Pathologists' certification program, which includes a checklist of the essential elements required for a successful immunostaining program.[@bib9000] With regard to staff qualifications, the National Society of Histotechnologists has focused its efforts on continuing education and certification programs for technologists performing immunohistochemical staining. Federal law requires a high degree of testing and validation. In the United States, laboratories performing immunohistochemistry are required under the Clinical Laboratory Improvement Amendments of 1988 to validate the performance of their test reagents for accuracy, specificity, sensitivity, and precision.[@bib14] First, the testing procedure is optimized (as described earlier), and performance expectations are established. During the validation process of each analyte, multiple slides with known pathology (generally 20 representative cases) are evaluated with the optimized procedure to assess the accuracy of diagnostic staining, sensitivity of signal, and reproducibility. Validations that meet specifications must be signed by qualified individuals, and the documents are maintained in the laboratory. Quality control and proficiency testing must be performed to monitor performance. Limitations {#s0090} ----------- Although immunohistochemistry is an extremely valuable technique in experienced hands, its limitations must be recognized for it to be used to its maximum potential. ### Experience {#s0100} Although immunohistochemistry is more objective than routine morphologic examination, the experience of the pathologist assessing the slides is critical. A firm understanding of the principles of immunohistochemical staining is necessary because the reporting pathologist must be equipped to deal with the unexpected and conflicting results that inevitably occur. To evaluate the immunohistochemical slides properly, the pathologist must have a firm understanding of the limitations of antibodies in terms of their technical aspects as well as their inherent specificity, sensitivity, and expected subcellular location. ### Availability of Antibodies {#s0110} The advent and refinement of the hybridoma technique for the production of monoclonal antibodies have produced a large number of available antibodies. Often a newly developed antibody is hailed as exquisitely specific. In time, however, most are found to be considerably less specific than initially hoped, generally because the antigen the antibody detects has a wider distribution than expected. This fact does not negate the usefulness of the antibody in question, but it may mean that panels of antibodies must be used in conjunction with standard morphologic features and clinical history. ### Loss of Antigenicity in Stored Cut Paraffin Sections {#s0120} Many studies have shown that a loss of antigenicity can occur on cut paraffin sections that have been stored for varying lengths of time.[@bib15], [@bib16], [@bib17] Among the antibodies studied, those most adversely affected by storage include p53,[@bib15], [@bib17] MIB1,[@bib16], [@bib17] factor VIII--related antigen,[@bib15] estrogen receptor,[@bib15] bcl-2,[@bib15] p27kip1,[@bib16] CD-44s,[@bib16] and androgen receptor.[@bib16] In many cases, the use of carefully selected and tested antigen retrieval techniques can compensate for this loss.[@bib17] Antigen Retrieval {#s0130} ----------------- Formalin is the most widely used fixative in surgical pathology. Cross-linking of proteins is the essential feature of formalin fixation. This cross-linking interferes with the antigen\'s ability to react with the primary antibody. In 1991 the antigen retrieval technique was developed.[@bib12], [@bib18] This technique is a heat-induced modification of the protein conformation that allows the antigen to be accessible again for chemical reactions, in this case, antibody binding. Hydrolysis of cross-linking resulting from formalin fixation probably plays a major role in this modification process.[@bib19], [@bib20], [@bib21], [@bib22] The application of antigen retrieval to sections derived from formalin-fixed, paraffin-embedded blocks produces consistent results of acceptable quality,[@bib6] although a few antigens remain undetectable even after antigen retrieval has been performed. Antigen retrieval methods have revolutionized immunohistochemistry and have become a standard part of diagnostic immunohistochemistry in surgical pathology. These methods result in higher sensitivity and more consistent antibody reactivity. Antigen retrieval technology has led to a proliferation of protocols that may produce different results in different laboratories. The successful application of these methods allows the detection of some antigens that were previously undetectable in paraffin sections, rendering much of the early literature (prior to 1993) obsolete. This fact continues to escape the notice of some practicing pathologists, leading to errors of interpretation ([Fig. 5-1](#f0010){ref-type="fig"} ).Figure 5--1 ▪Section of lymphoid tissue stained with antibody against lambda light chain without antigen retrieval, showing no positive cells (**A**), and with antigen retrieval, showing scattered positive cells (**B**). It should be recognized that two major factors influence the effectiveness of antigen retrieval: the conditions under which heating takes place, and the pH value of the buffer solution used during the heating process.[@bib5], [@bib11], [@bib23] The most critical factor is the combination of the temperature and the duration of heating, which have a reverse correlation. Based on these two factors (heating conditions and buffer pH), a test battery approach has been developed to establish optimal antigen retrieval protocols for immunostaining on archival paraffin-embedded tissue sections.[@bib1], [@bib6], [@bib23] A typical test battery consists of nine serial sections of a specimen known to express the antigen under study. The sections are evaluated with buffer at three different pH values (e.g., pH 1 to 2, 7 to 8, and 10 to 11) and three heating conditions (e.g., 90°C, 100°C, and 120°C) for various lengths of time (or some other comparable heating versus time schedule). The best result is selected as the optimal retrieval condition for that antigen. In the event that a satisfactory result is not obtained, other variations may be explored, including different buffer solutions and more or less vigorous heating methods.[@bib19], [@bib23] Protocols for antigen retrieval differ in their effectiveness for retrieving certain antigens, and a single universally effective retrieval method does not exist. Many laboratories use more than one method for different antibody and antigen combinations. Overall, citrate buffer at pH 6.0 has the broadest applicability for the widest range of antigens, although several studies have demonstrated that the use of higher pH retrieval solutions yields satisfactory results.[@bib21], [@bib24] Retrieval solutions with lower pH values, TRIS (tromethamine) buffer at pH 8.0, and EDTA (ethylenediaminetetraacetic acid)-NaOH solution (pH 8.0), are effective in certain special situations.[@bib25], [@bib26], [@bib27] The selection of heating method (water bath, steamer, microwave, pressure cooker, or autoclave) is influenced by custom and availability. CURRENT APPLICATIONS OF IMMUNOHISTOCHEMISTRY {#s0140} ============================================ Diagnostic Tool for Tumors of Unknown Origin {#s0150} -------------------------------------------- Immunohistochemistry has become an integral and essential part of surgical pathology. It is applied to define tumor origin, establish prognosis, and determine treatment response. In this textbook, the role of immunohistochemistry in defining the origin, prognosis, and treatment response of tumors is discussed in the chapters devoted to the specific organ systems; therefore, a full discussion is not provided here. Because the evaluation of tumors of unknown origin does not fall under any particular organ system, it is discussed in this chapter. Tumors are classified most often by their tissue of origin (e.g., breast, colon, prostate) or histogenetically (e.g., tissue of epithelial, mesenchymal, or neural origin). A tumor cannot be staged accurately, and proper therapy cannot be administered, without such classification. Although accepted and fairly reproducible criteria exist for the morphologic diagnosis of most tumors, there is inherent subjectivity in any morphologic evaluation. It is well recognized that morphologic features often overlap among different entities and that one disease can present with myriad histologic pictures. Most tumors can be classified correctly by routine histologic techniques when the clinical situation is clear (e.g., a breast mass); however, an important subset of tumors defies morphologic interpretation. The magnitude of this problem is substantial. The diagnosis of "metastatic cancer of unknown primary site" is the eighth most common cancer diagnosis and may represent up to 15% of cancers at large hospitals.[@bib28] Much more common is the diagnosis of "tumor of uncertain origin." This occurs when (1) the tumor is first identified in a metastatic site, and the primary site is not apparent; (2) the tumor is so poorly differentiated that no specific morphologic features can be identified; (3) the morphologic appearance of the tumor is compatible with more than one distinct tissue (e.g., epithelial versus lymphoid origin); and (4) the histogenesis of a tumor is clear (e.g., adenocarcinoma), but the primary site is in question. This distinction has important consequences to the patient. ### Test Selection {#s0160} Immunohistochemical tests should be performed with a defined objective in mind. The results of a single immunohistochemical procedure can be misleading not only because of variables in the staining procedure but also because of unanticipated patterns of reactivity of certain antibodies.[@bib1] Although myriad antibodies are available, the choice in a particular case should be judicious and designed to address the diagnostic possibilities. The use of too few antibodies rarely provides sufficient information to support a specific diagnosis and can produce misleading information. Antibodies should be selected on the basis of their ability to affirm or exclude considerations in the differential diagnosis. This so-called problem-oriented approach is based on the selection of appropriate panels of antibodies. When selecting antibodies, factors that should be considered include the clinical history, morphologic features of the tumor, and results of other tests that may have been performed, including serologic and radiographic tests. Pathologists can find guidance in the literature and in a few specialized textbooks that address the use of immunohistochemistry,[@bib2], [@bib29] but this is a rapidly evolving field. The limited antibody panels of a few years ago are inadequate to deal with tumors of unknown or uncertain primary sites today. With this in mind, the panels presented here must be considered elementary guides. ### Panel Approach: Basic Principles {#s0170} When evaluating tumors of uncertain origin by immunohistochemistry, certain basic guidelines of interpretation must be kept in mind. A positive staining reaction is generally more helpful than a negative one because a lack of immunoreactivity may represent a technical problem with the tissue or the way it was fixed, as discussed earlier. The more poorly differentiated a tumor is, the less likely it is to express tissue differentiation antigens. There is often staining variation within a tumor; by extension, variations in staining patterns may be seen between the primary tumor and the metastatic focus. Most important, the final diagnosis should never depend on immunohistochemistry alone; it must be made using all the clinical, serologic, radiographic, morphologic, and epidemiologic data available. Other techniques, such as the assessment of specific DNA alterations, are becoming increasingly important adjuncts to the pathologic evaluation. Although immunohistochemical evaluation is essential, it is only one of many tools that must be used in the evaluation of pathologic processes. The application of a primary panel of antibodies to characterize tumor histogenesis (epithelial, mesenchymal, neural, or hematopoietic) is often the first step. When this panel has been established, additional antibodies can be used to identify the tumor type more specifically. Included in the first tier would be antibodies against pan-keratin, vimentin, S-100 protein, neuron-specific enolase (NSE) and CD45 (common leukocyte antigen) to differentiate epithelial, mesenchymal, melanomatous, neural, and lymphoid malignancies ([Table 5-1](#t0010){ref-type="table"} ; [Fig. 5-2](#f0020){ref-type="fig"} ).TABLE 5--1Screening Immunophenotypes of Undifferentiated NeoplasmsAE1/AE3VimentinCD45S-100Tumor Type+--/+--RCarcinomaR+----/+Sarcoma----+--Lymphoma--+--+Melanoma[^1]Figure 5--2 ▪**A,** Colon carcinoma showing keratin positivity, which is typically seen in carcinomas. **B,** Malignant fibrous histiocytoma showing vimentin positivity, which is typically seen in sarcomas. Note the fine reticular pattern of the intermediate filaments in the giant cells. **C,** S-100--positive melanoma. Note the nuclear reactivity, characteristic of melanoma. **D,** CD45 (common leukocyte antigen)--positive lymphoma. ### Intermediate Filaments {#s0180} The expression of intermediate filament proteins, which function as the supporting cytoskeleton in normal and neoplastic cells, is extremely useful in the initial assessment of tumors of unknown primary origin.[@bib1] There are five major classes of intermediate filaments, based on protein composition and cellular distribution: cytokeratin, vimentin, desmin, neurofilament, and glial fibrillary acidic protein (GFAP).[@bib30] Most neoplasms show the predominant expression of one or more of these intermediate filaments. Carcinomas usually express cytokeratin; sarcomas, melanomas, and lymphomas are generally vimentin positive; myogenic tumors are characteristically positive for desmin or muscle actins and vimentin; and glial tumors are predominantly positive for GFAP.[@bib1] Some tumors characteristically coexpress more than one intermediate filament (e.g., renal and thyroid carcinomas contain keratin and often vimentin), whereas others show aberrant or no intermediate filament expression. Immunohistochemical markers for intermediate filaments on tumors of uncertain origin are one of the most useful and productive ways to begin classifying the lesion. Anaplastic neoplasms can be characterized as keratin positive (carcinomas, mesotheliomas), vimentin positive (sarcomas, lymphomas, melanomas), or neurofilament and GFAP positive (neuroendocrine, neural, and astrocytic tumors).[@bib1] ### Keratin-Positive Tumors {#s0190} Cytokeratins are present in almost all epithelial cells and are highly sensitive markers for carcinomas. In a generic sense, malignant cells expressing keratin positivity indicate an epithelial origin. Antibodies against keratin are also extremely useful as markers for occult metastases (micrometastases) in the peripheral blood, bone marrow, and lymph nodes (discussed later). There are more than 20 different subtypes of cytokeratin found in human epithelial cells. These subtypes are distinguishable by their molecular weight and isoelectric pH.[@bib31] Monoclonal antibodies specific for many of these subtypes have been developed. Carcinomas of different types tend to express characteristic keratin profiles.[@bib32] There is a general correlation between the complexity of the epithelium from which the tumor is derived and the complexity of the keratin subunits expressed. Low-molecular-weight or nonsquamous keratins appear early in development and predominate in tumors derived from simple, nonstratified epithelium (e.g., ductal carcinoma of the breast, gastrointestinal adenocarcinoma). High-molecular-weight or squamous keratins appear in more complex stratified epithelium and predominate in tumors derived from stratified epithelium (e.g., squamous cell carcinoma). Some tumors, such as those derived from pseudostratified columnar epithelium, contain a mixture of high- and low-molecular-weight keratins, with a predominance of the latter. In some instances, especially in extremely poorly differentiated tumors, as few as 5% of tumor cells may express keratin reactivity.[@bib1] When a tumor of uncertain primary site has been defined as epithelial by either immunohistochemistry or morphology, it is important to attempt to define its specific origin. This presents a problem when, for example, a patient with a prior history of breast carcinoma presents with a lung mass that, on biopsy, is adenocarcinoma. Determining whether the lung mass represents a primary pulmonary tumor or metastasis from the breast has enormous consequences in terms of patient outcome and choice of specific treatment. Although the immunohistochemical evaluation of primary epithelial tumors is problematic, advances have been made. Monoclonal antibodies against keratin subtypes may help determine the origin of certain poorly differentiated neoplasms. Hepatocellular carcinoma (positive for AE3 and CAM 5.2 but negative for AE1) can be distinguished from bile duct carcinoma and adenocarcinoma metastatic to the liver (positive for AE1).[@bib33] In particular, the differential expression of cytokeratins 7 and 20 (CK7, CK20) is extremely useful in the characterization of epithelial neoplasms ([Fig. 5-3](#f0030){ref-type="fig"} ).[@bib32], [@bib34], [@bib35], [@bib36] These patterns are not absolute, but they can be useful guides in establishing origin.Figure 5--3 ▪Algorithm for carcinoma of unknown primary site. Cytokeratin 5/6 (CK5/6) has received considerable attention recently. CK5/6 has been found to be positive in the majority of squamous cell carcinomas, basal cell carcinomas, thymomas, salivary gland tumors, and biphasic malignant mesotheliomas and in a subset of endometrial adenocarcinomas, transitional cell carcinomas, and pancreatic adenocarcinomas.[@bib37] CK5/6 is rarely positive in adenocarcinoma of the lung and has therefore been used to distinguish malignant mesothelioma from pulmonary adenocarcinoma.[@bib37], [@bib38] In addition, p63 is frequently seen in squamous cell carcinoma and transitional cell carcinoma, whereas mesothelioma is uniformly negative for p63. Therefore, positive immunostaining for both p63 and CK5/6 is highly predictive of a primary tumor of squamous epithelial origin.[@bib39], [@bib40] Because p63 is also known to immunoreact on the basal cell nuclei in benign prostate glands, this marker can be used to distinguish prostate cancer from benign mimics.[@bib41] There is an increasing array of *tissue-specific* markers, such as prostate-specific antigen (PSA) and thyroglobulin, as well as *tissue-associated* markers, such as GCDFP-15 and mammaglobin for breast epithelium,[@bib42], [@bib43] OC-125 for ovary,[@bib44] uroplakins for urothelium,[@bib45], [@bib46] and synaptophysin for neuroendocrine lesions ([Fig. 5-4](#f0040){ref-type="fig"} ).[@bib47] [Table 5-2](#t0020){ref-type="table"} summarizes many of these tissue-associated antibodies. It is important to keep in mind patterns of antigenic coexpression, which can lead to erroneous assessments. Also, as mentioned earlier, the more poorly differentiated the neoplasm, the less likely it is to express tissue-specific or -associated antigens. Nevertheless, the specific origin of an epithelial neoplasm of uncertain primary site can be elucidated in an increasing proportion of cases through careful evaluation of morphology, clinical data, and antigen expression.Figure 5-4 ▪**A,** Carcinoma of the colon showing CEA positivity. **B,** Ovarian carcinoma showing OC-125 positivity.TABLE 5--2Antibodies Useful in Determining the Origin of Undifferentiated Tumors and Tumors of Uncertain Primary SitePanelAntibodiesUndifferentiated tumorsPan-keratinCD 45 (CLA)S-100VimentinCarcinoma panelPan-keratinCK5/6AE1AE3CAM5.2MAK6Squamous keratin (HMW)Nonsquamous keratin (LMW)Sarcoma panelPan-keratinVimentinS-100DesminCD45 (CLA)Actin (muscle/HHF-35-MSA)Actin (smooth muscle specific---SMA)Myoglobin (skeletal muscle)LN-5 (histiocytes)Lysozyme (histiocytes)LN-6 (nonlymphoid vimentin)Factor VIII antigen (endothelial cells)CD34 (vascular antigen)CD31 (vascular antigen)Ulex (vascular antigen)O13 (Ewing\'s sarcoma/PNET)Breast panelBRST-2 (GCDFP)MammaglobinCu-18 (breast-related antigen)LactalbuminPrognosis (breast carcinoma)Estrogen receptor (monoclonal)Progesterone receptor (monoclonal)Her-2/*neu* (c-*erb* B-2)p52 (luminal epithelial antigen)p53Factor VIII antigenLung panelCK7CK20TTF-1Prostate panelProstate-specific antigen (PSA)Prostatic acid phosphatase (PAP)Androgen receptor34BE12 (SK)Gastrointestinal panelCEACOTACDX-2CK7CK20Kidney/bladder panelRenal antigenCEAp53Ovary panelCOTAOC-125 (CA-125)Estrogen receptorProgesterone receptorLiver panelAlpha fetoproteinα~1~-Antitrypsinα~1~-AntichymotrypsinNonsquamous keratinHepatitis B surface antigenHepatitis B core antigenMesothelioma panelSquamous keratinNonsquamous keratinCEA-negativeEP4 (epithelial antigen)-negativeCD15 (Leu M1)-negativeEpithelial membrane antigenB72.3Secretory componentVimentinOC-125Melanoma panelS-100HMB-45Melan AVimentinPan-keratinCentral nervous system/neural panelGlial fibrillary acidic protein (GFAP)NeurofilamentS-100NSEVimentinPan-keratinSynaptophysinNeuroendocrine panelNSEChromograninSerotoninNeuron-endocrineSynaptophysinNonsquamous keratinCD57 (Leu 7, HMK 1)Vasointestinal peptidePituitary hormone panelAdrenocorticotropic hormone (ACTH)Follicle-stimulating hormone (FSH)Growth hormoneLuteinizing hormoneProlactinThyroid-stimulating hormone (TSH)α-SubunitPIT-1Pancreatic panelAmylaseInsulinGlucagonGastrinSomatostatinCDX-2Urothelial panelp63Uroplakin[^2] ### Keratin-Negative Tumors {#s0200} #### LYMPHOMAS {#s0210} Lymphomas often present morphologically as undifferentiated malignant neoplasms. A large cell lymphoma may be difficult to differentiate from carcinoma or melanoma. Anaplastic large cell lymphomas occasionally react with antibodies against keratin or epithelial membrane antigen. Similarly, small cell lymphomas often resemble other tumors, such as small cell undifferentiated carcinoma. Immunohistochemistry can be invaluable in classifying a tumor as lymphoid in origin. Immunohistochemistry is also used widely to aid in the subclassification of non-Hodgkin\'s lymphoma (discussed in detail in Chapter 41). Another important application of immunohistochemistry is the phenotyping of a lesion to determine the immunoglobulin light-chain expression. One can distinguish between malignant and benign lymphoid proliferation by demonstration of light-chain (or heavy-chain) restriction. CD45 is an excellent screening marker to determine whether a tumor is lymphoid in origin. Staining is characteristically membranous. Some neoplasms of lymphoid origin may not express CD45, and rare nonlymphoid neoplasms may show cytoplasmic staining for CD45. More specific lymphoid markers such as CD3, which marks T cells, and CD20, which marks B cells, may further delineate a lesion. Immunohistochemistry can also be used in cases of Hodgkin\'s disease, in which markers useful in identifying Reed-Sternberg cells include CD15, CD30, and BLA-36.[@bib48], [@bib49] [Figure 5-5](#f0050){ref-type="fig"} illustrates an algorithmic approach to the immunohistochemical diagnosis of malignant lymphomas.Figure 5-5 ▪Diagnostic algorithm for hematolymphoid neoplasms. #### MELANOMAS {#s0220} Melanomas are typically (but not always) negative for cytokeratin[@bib50] and positive for vimentin. S-100 protein is a sensitive marker for melanoma and occurs, in almost all cases, in a nuclear pattern (see [Fig. 5-2C](#f0020){ref-type="fig"}). S-100 is not specific for melanoma, however, and is seen in a variety of lesions, including Langerhans histiocytes, many sarcomas, and certain carcinomas.[@bib51], [@bib52] Positive immunoreactivity for HMB-45, Melan A, or tyrosinase, which are much more specific markers for melanoma and melanocytes, can confirm the diagnosis.[@bib53] This topic is discussed in detail in Chapter 49. #### SARCOMAS AND SOFT TISSUE TUMORS {#s0230} A diagnosis of sarcoma is worth considering when a spindle cell neoplasm expresses vimentin but not keratin, CD45, or HMB-45. Immunohistochemical analysis can delineate a specific type of sarcoma such as myogenic and fibrohistiocytic tumors, tumors with neural differentiation, and vascular sarcomas. The differentiation of soft tissue tumors by immunohistochemistry is discussed in Chapter 46. The major subtypes are described briefly here. Myogenic sarcomas react with antibodies to muscle-specific actin and desmin. Antibodies against myoD-1 smooth muscle actin show preferential reactivity with leiomyosarcomas.[@bib54] Alveolar soft part sarcomas typically coexpress vimentin and desmin.[@bib55] Tumors derived from skeletal muscle often contain myoglobin; positive immunoreactivity for muscle-specific actin, desmin, and myoglobin indicates that the tumor in question may be a rhabdomyosarcoma. Fibrohistiocytic tumors, including malignant fibrous histiocytoma, are the most common form of soft tissue sarcoma in adults. Malignant fibrous histiocytoma reacts with vimentin and lacks significant reactivity for keratin. These lesions may also coexpress α~1~-antitrypsin, α~1~-antichymotrypsin, HAM-56, and CD68. α~1~-Antichymotrypsin is not specific for malignant fibrous histiocytoma and can be seen in other sarcomas, some carcinomas, and melanomas.[@bib56], [@bib57], [@bib58] Neurogenic tumors, including malignant peripheral nerve sheath tumors and schwannomas, are positive for antibody against S-100 protein, myelin basic protein, and Leu-7.[@bib59], [@bib60] Although most benign neurogenic tumors (schwannomas and neurofibromas) contain S-100, less than half of malignant peripheral nerve sheath tumors contain detectable S-100 protein.[@bib61] Normal and neoplastic vessels can be identified with endothelial markers such as factor VIII---related antigen and *Ulex europaeus* lectin. Antibodies to factor VIII--related antigen react only with endothelial cells and megakaryocytes and are more specific than *Ulex europaeus*. Other markers of endothelial cells include CD34 and CD31; these are more sensitive but less specific than factor VIII. Other tumors to be considered in the differential diagnosis of anaplastic spindle cell tumors include liposarcomas, chondrosarcomas, osteogenic sarcomas, fibrosarcomas, and synovial sarcomas. Although certain variants of liposarcoma may be diagnosed by histologic criteria alone, the diagnosis of pleomorphic liposarcoma is aided by immunohistochemistry. Most liposarcomas react with vimentin and S-100 but are nonreactive for HMB-45, in contrast to most melanomas. The same pattern is seen for chondrosarcomas.[@bib62] Osteosarcomas react with vimentin. Antibodies to the so-called osteonectin or osteosarcoma antigens also may assist in the diagnosis.[@bib62] Fibrosarcomas are rare neoplasms that express only vimentin. [Figure 5-6](#f0060){ref-type="fig"} provides the immunostaining patterns characteristic of anaplastic spindle cell tumors.Figure 5-6 ▪Diagnostic algorithm for spindle cell tumors. Cytogenetic profiling of sarcomas of unknown origin is becoming common. Tumors with known abnormalities include endometrial sarcoma, myeloid sarcoma, synovial sarcoma, Ewing\'s sarcoma, and visceral clear cell sarcoma, among others.[@bib63], [@bib64], [@bib65], [@bib66], [@bib67], [@bib68], [@bib69], [@bib70] This topic is discussed in more detail in Chapter 46. #### NEURAL AND NEUROENDOCRINE TUMORS {#s0240} Neural and neuroendocrine tumors may be classified as neural tumors, neuroepithelial tumors, or neural neoplasms of mesenchymal origin. These three categories are based on the different predominant intermediate filaments found in the cytoplasm of these lesions. Neural tumors usually express neurofilament, NSE, chromogranin, and synaptophysin.[@bib71], [@bib72], [@bib73] Examples include neuroblastomas, paragangliomas, and pheochromocytomas. Neuroepithelial tumors coexpress keratin and neuroendocrine markers. These tumors include carcinoids, Merkel cell carcinomas, and small cell carcinomas. Neural neoplasms of mesenchymal origin, which consist of primitive neuroectodermal tumors, Ewing\'s sarcomas, and medulloblastomas, are positive for vimentin and may express NSE and Leu-7.[@bib74] A good marker for small cell tumors such as Ewing\'s sarcoma, primitive neuroectodermal tumor, and peripheral neuroepithelial tumor is O13, which identifies the CD99 (p30/32, mic2, HBA71) antigen.[@bib75] NSE, although a sensitive marker for neuroendocrine tumors, lacks specificity and can be seen in a wide variety of tumor types. Chromogranin and synaptophysin are more specific than NSE but lack sensitivity. Chromogranin tends to be positive in better-differentiated neuroendocrine tumors but is less often positive in the more poorly differentiated tumors, such as small cell carcinoma.[@bib72], [@bib76], [@bib77], [@bib78] [Table 5-3](#t0030){ref-type="table"} provides immunostaining patterns of endocrine tumors, and [Figure 5-7](#f0070){ref-type="fig"} illustrates an algorithm used to distinguish small round cell tumors.TABLE 5--3Dominant Immunophenotypes of Endocrine NeoplasmsCAM 5.2CEAChromograninSerotoninSynaptophysinTTF-1VimentinAdditional MarkersTumor Type+--+/----/++----Calcitonin --/+;\ Somatostatin --/+\ Gastrin --/+Carcinoid+--/+--/+--+--/+--/+Neuroendocrine+----/+--+----Insulin +Insulinoma+--+--+----Glucagon +Glucagonoma+----/+--+----Somatostatin +Somatostatinoma+--+--+----Gastrin +Gastrinoma+++--+++Calcitonin +Medullary, thyroid+--+--+----PTH +Parathyroid+--+/----+----CK20 +Merkel cell--/+--+/----+--+S-100 +/--Paraganglioma, pheochromocytoma--/+------+--+Melan A +; inhibin +Adrenal cortical[^3]Figure 5-7 ▪Diagnostic algorithm for small round cell tumors. #### GFAP-POSITIVE TUMORS {#s0250} GFAP is expressed by glial cells and is seen in astrocytomas, ependymomas, medulloblastomas, some oligodendrogliomas, and choroid plexus tumors.[@bib79], [@bib80] It has also been reported in a few extracerebral tumors, including pleomorphic adenomas of the salivary gland,[@bib81] neurofibromas, and schwannomas.[@bib82] Intracerebral tumors in which GFAP is not expected to be positive include meningiomas, lymphomas, and metastatic carcinomas. GFAP is discussed in more detail in Chapter 51. ### Molecular and Genetic Markers of Tumor Origin {#s0260} Cancer results from defects in gene structure, expression, or both. Alterations in chromosome and DNA structure---including cytogenetic changes, point mutations, deletions, amplifications, translocations, and DNA methylation---are being identified at an increasing rate. Characterization of these defects is becoming an important component of tumor evaluation, particularly in terms of prognosis and response to treatment. In addition, genetic defects highly characteristic of specific tumor types are being identified. The identification of DNA alterations is also becoming an increasingly important component in the evaluation of tumors of uncertain primary site. Molecular and genetic evaluation is particularly useful in the evaluation of hematopoietic and soft tissue tumors.[@bib83], [@bib84] Recently, a highly sophisticated evaluation of hundreds of genes using so-called microarray chip technology was used to identify gene expression differences in several tumor types[@bib84], [@bib85], [@bib86] and will no doubt become an important component of tumor evaluation in the future. #### TRANSCRIPTION FACTORS {#s0270} Transcription factors are proteins involved in the regulation of gene expression that bind to the promoter elements upstream of genes and either facilitate or inhibit transcription. They may be tissue specific, or they may be present in more than one tissue type. Even the so-called tissue-specific transcription factors are, however, usually not restricted to a single tumor type. Examples include thyroid transcription factor-1 (TTF-1), which is found in the thyroid and lung, and the pituitary transcription factor Pit-1, which is found in the placenta as well as the pituitary gland. Nevertheless, they can be useful in determining the primary site of tumors of unknown origin (see [Table 5-2](#t0020){ref-type="table"}). ##### TTF-1 {#s0280} TTF-1 belongs to a family of homeodomain transcription factors and plays a role in regulating genes expressed within the thyroid, lung, and diencephalons. TTF-1 is considered a reliable marker for distinguishing primary tumors of the lung, including adenocarcinoma (75%), non--small cell carcinoma (63%), neuroendocrine and small cell carcinoma (\>90%), and squamous cell carcinoma (10%),[@bib87], [@bib88], [@bib89], [@bib90] and thyroid from other tumor types. However, primary adenocarcinoma of the colon is positive in some cases.[@bib36], [@bib91], [@bib92] The pattern of reactivity is nuclear. Hepatocytes and hepatocellular carcinoma reportedly show cytoplasmic positivity.[@bib93] ##### CDX-2 {#s0290} *CDX-2* is a homeobox gene that encodes a transcription factor involved in the development of intestinal epithelium. It is expressed in normal colonic epithelium and in most colorectal adenocarcinomas and is a useful marker to identify colorectal metastases.[@bib94], [@bib95], [@bib96] CDX-2 is also useful in extramammary Paget\'s disease; the endodermal subtype is positive for CDX-2, whereas the cutaneous subtype is negative.[@bib97] ##### ATF3 {#s0300} Activating transcription factor 3 (ATF3) is a member of the basic leucine zipper/cyclic adenosine monophosphate--responsive element binding protein family of transcription factors. Studies indicate that *ATF3* is an androgen-regulated gene that stimulates cell proliferation. ATF3 protein detected by immunohistochemistry is present in prostate tumors; it has increased expression in high-Gleason-score disease and in tumors refractive to therapy.[@bib98] Prognostic Markers in Cancer {#s0310} ---------------------------- One of the outstanding achievements of modern medicine is the ability to predict the behavior of tumors based on specific clinical and pathologic criteria. Tumor stage and grade provide only general estimates of outcome for a particular patient, however. Current clinical and pathologic staging parameters cannot identify those patients who are destined to experience relapse or those whose disease will be cured by local therapy alone. These considerations have obvious consequences for the patient and enormous economic implications. Efforts are under way to identify enzymes, oncogenes, or tumor suppressor genes whose presence or absence may predict more accurately the biologic behavior of tumors. Such studies represent a fundamental shift in the means by which tumor behavior is defined---a change from an outcome-based empirical analysis (i.e., prediction of what a tumor will do based on what it has done) to one focused on tumor biology (i.e., predictions of behavior based on specific genetic alterations). The immunohistochemical analysis of tumors is also undergoing a profound shift in emphasis. Although initial studies focused on defining tumor histogenesis, the goal of much current research is to reveal the biologic potential of tumors, providing a more scientific basis for patient management. The use of advanced technologies to define tumor prognosis is described in detail throughout this book. Presented here are some general principles that pertain to the prognostic evaluation of all tumors. ### Occult Metastases {#s0320} The most important factor affecting the outcome of patients with invasive cancers is whether the tumor has spread either regionally (to regional lymph nodes) or systemically. A proportion of patients with no evidence of systemic dissemination as evaluated by routine methods (careful pathologic, clinical, biochemical, and radiologic evaluation) develop recurrent disease. In addition, the success of adjuvant therapy is assumed to stem from its ability to eradicate occult metastases before they become clinically evident.[@bib99] Immunohistochemistry is commonly used to identify occult metastatic cancer cells in the bone marrow, peripheral blood, and lymph nodes of patients with cancer. Although many of the initial studies focused on breast cancer,[@bib100], [@bib101], [@bib102] tumors from other organs such as the stomach,[@bib103], [@bib104] colon,[@bib105], [@bib106] prostate,[@bib107], [@bib108] lung,[@bib109], [@bib110] nervous system,[@bib111] and skin[@bib112] have been investigated. Immunohistochemical methods are based on the ability of monoclonal antibodies to distinguish between cells of different histogenesis (e.g., epithelial cancer cells versus the hematopoietic and stromal cells of the bone marrow and lymph nodes). The results indicate that it is possible to identify occult metastatic cancer cells in lymph nodes and bone marrow before their detection by other methods and that the presence of these cells may be an important risk factor for disease recurrence ([Fig. 5-8](#f0080){ref-type="fig"} ).Figure 5--8 ▪**A,** Bone marrow aspirate with a single keratin-positive cell from a patient with lung cancer. **B,** Section of lymph node shows a small focus of early metastatic breast cancer. The most widely used monoclonal antibodies to detect occult metastatic carcinoma cells are directed toward epithelium-specific antigens. These antibodies do not react with normal hematopoietic or stromal cells present in the bone marrow or lymph nodes. None of the antibodies used in any study is specific for cancer; all react with normal and malignant epithelial cells. They are useful because they can identify an extrinsic population of epithelial cells in bone marrow or lymph nodes, where there are normally no epithelial elements. The reported sensitivity of immunohistochemistry ranges from the detection of 1 epithelial cell in 10,000 to 2 to 5 epithelial cells in 1 million hematopoietic cells.[@bib100], [@bib113] A potentially more sensitive approach for the detection of occult metastasis is the reverse transcriptase polymerase chain reaction (RT-PCR) technique, which has been applied to several malignancies using a variety of marker transcripts as targets. Since the first study by Smith and colleagues in 1991,[@bib114] many authors have reported molecular diagnoses in the lymph nodes, blood, and bone marrow in cancer patients.[@bib42], [@bib107], [@bib115], [@bib116], [@bib117], [@bib118], [@bib119] Application of RT-PCR in regional and sentinel lymph nodes has been described for a number of cancers, including melanoma, colorectal cancer, and cancers of the prostate, breast, and lung.[@bib118], [@bib120], [@bib121], [@bib122], [@bib123], [@bib124] Many of these compare immunocytochemical-based detection with RT-PCR for sensitivity and conclude that RT-PCR may achieve enhanced detection, provided the target markers are sufficiently specific. Various formats of RT-PCR assays[@bib125], [@bib126], [@bib127] have also been used to detect disseminated tumor cells in the bone marrow of patients with cancers of the breast, colon, and lung, among others. With the exception of some organ-specific markers such as maspin or mammaglobin for breast cancer[@bib116], [@bib128] or uroplakins for urothelial tumors,[@bib129] most of the molecular targets used in these RT-PCR assays lack the requisite specificity owing to illegitimate expression in nontarget hematopoietic cells.[@bib130], [@bib131], [@bib132] Unlike immunohistochemistry, morphologic confirmation of the cells in question to verify tumor origin is not possible with RT-PCR. RT-PCR has also been used to enhance the sensitivity of the detection of tumor cells in the peripheral blood in a variety of cancers, including prostate, breast, gastrointestinal tract, colorectal, and head and neck cancers and melanoma.[@bib117], [@bib133], [@bib134], [@bib135] Concerns about nonspecificity owing to illegitimate transcription of target genes in the nontarget hematopoietic cells also apply to the blood, which has hampered the use of these assays in routine clinical diagnosis. ### Bone Marrow and Peripheral Blood {#s0330} In breast cancer, the bone marrow is the single most common site of metastasis, and 80% of patients with recurrent tumors develop bone marrow metastases at some point during the evolution of their disease.[@bib136] Immunohistochemistry can show the presence of occult metastases in the bone marrow in approximately 10% to 45% of patients with low-stage disease.[@bib102], [@bib127], [@bib137], [@bib138], [@bib139], [@bib140], [@bib141], [@bib142], [@bib143] Several studies have addressed the clinical significance of these early metastatic cells in the bone marrow, including a pooled analysis of more than 4700 patients.[@bib143] They have found that the presence of such cells is an independent prognostic indicator of disease-free survival and overall survival.[@bib144], [@bib145], [@bib146] Occult metastases in the bone marrow are also prognostically important in other malignancies, including primary non--small cell lung cancer,[@bib109], [@bib110], [@bib147], [@bib148] esophageal and gastric cancers,[@bib103] colorectal cancer,[@bib149] and neuroblastoma.[@bib111] The finding of positive cells in the bone marrow of patients with colorectal cancer---a tumor that rarely shows overt metastasis to the bone---indicates that this may be a general indicator of tumor dissemination. In addition, the prognostic significance of occult metastatic cells in the blood is under investigation. Peripheral blood has the advantage of being easier to access than bone marrow. However, detection rates are considerably lower than with bone marrow, a fact that has hampered studies to date. In the detection of tumor cells in the bone marrow and blood, epithelial cell adhesion molecule in conjunction with immunomagnetic enrichment has been used to detect circulating tumor cells in breast cancer patients. Circulating tumor cells in the blood have been used to monitor response to therapy in patients with metastatic cancer.[@bib150], [@bib151], [@bib9001] The presence of epithelial adhesion molecule--positive cells before and after the initiation of therapy was found to be an independent prognostic factor. Other markers used in patients with breast cancer include mammaglobin, epidermal growth factor receptor, and carcinoembryonic antigen.[@bib116], [@bib152], [@bib153] Although none of these markers is entirely specific for the detection of metastatic breast cancer, and although the sensitivity of peripheral blood is less than that of bone marrow, there is growing evidence that the detection of occult metastatic cells in the peripheral blood has a negative impact on prognosis.[@bib154] A recent study found that the presence of five or more tumor cells in the peripheral blood from patients with breast cancer examined upon the initiation of therapy was important in predicting outcome.[@bib151] Peripheral blood from patients with colorectal,[@bib106] stomach,[@bib104] prostate,[@bib155] and skin [@bib155], [@bib156] cancer have also been studied. Markers that have been studied in colorectal cancer include cytokeratins,[@bib106] carcinoembryonic antigen,[@bib157], [@bib158] apolipoprotein,[@bib159] and CD44v6.[@bib160] PSA messenger RNA (mRNA) is the most commonly used marker in patients with prostate cancer.[@bib155], [@bib161] Tyrosinase mRNA is the marker of choice for detecting circulating tumor cells in patients with melanoma.[@bib161], [@bib162] ### Lymph Nodes {#s0340} Studies undertaken to detect occult lymph node metastases by routine histologic methods have generally been performed by cutting serial sections from all paraffin blocks containing lymph nodes, followed by routine staining and microscopic review.[@bib163] Several studies simply reviewed the original histologic slides. Newer studies involve cytokeratin immunohistochemistry on one or more lymph node sections. PSA immunohistochemistry has also been used to confirm the prostatic origin of cytokeratin-positive cells in the lymph nodes of patients with prostate cancer.[@bib108] All these studies have shown that deposits of tumor can be detected using these methods. In previously determined node-negative cases of breast cancer, 7% to 33% convert to node-positive status after review. Neville and colleagues[@bib164] found the mean conversion rate to be approximately 13%. Although virtually all studies have shown that lymph node metastases can be overlooked, there has been surprising disagreement about the prognostic importance of these occult tumor deposits.[@bib165], [@bib166], [@bib167] However, it is now widely accepted that the detection of occult lymph node metastases is an important predictor of outcome in patients with histologically node-negative cancer.[@bib168], [@bib169], [@bib170] In a key study (Ludwig Trial V), occult breast cancer metastases were detected by immunohistochemistry in 20% of patients and were associated with significantly poor disease-free and overall survival in postmenopausal patients but not in premenopausal patients.[@bib168] Additional studies in patients with breast cancer found occult lymph node metastases to be predictive of a poorer outcome.[@bib169], [@bib170] Studies in patients with lung,[@bib171], [@bib172] prostate,[@bib107], [@bib108] and colorectal [@bib173], [@bib174] cancer suggest that occult metastases in the lymph nodes in these patients may also predict a worse prognosis. The finding that occult bone marrow and lymph node metastases are prognostically important has motivated several major clinical trials, notably by the American College of Surgeons Oncology Group, in breast cancer (Z0010) and lung cancer (Z0040). The advent of the use of sentinel lymph node biopsy in tumor surgery (for breast cancer and melanoma) has caused physicians to examine these lymph nodes by more sensitive techniques, owing to the limited material available for histologic review.[@bib175] It is likely that the detection of occult metastases will soon be the general standard of care; this is true at many institutions that treat large numbers of patients with cancer. ### Oncogenes, Growth Factors, and Receptors {#s0350} #### HER-2/*NEU* {#s0360} Her-2/*neu* (or c-*erb* B-2) is a proto-oncogene. The gene encodes for a protein (185 kD) that shows homology with epidermal growth factor and displays tyrosine activity. Amplification of the gene coding for Her-2/*neu* has been described in breast, ovarian, prostate, gastric, salivary gland, lung, colon, and squamous cell carcinoma.[@bib176], [@bib177], [@bib178], [@bib179], [@bib180], [@bib181], [@bib182], [@bib183], [@bib184], [@bib185], [@bib186] When overexpressed, the protein accumulates at the cell membrane and is seen as a crisp membrane stain; a cytoplasmic staining pattern is not associated with protein or gene overexpression.[@bib176], [@bib183] Although Her-2/*neu* overexpression and amplification have been described in several tumor systems, it has been studied most extensively in the breast. Her-2/*neu* overexpression occurs in 10% to 34% of primary breast carcinomas[@bib187] and is restricted to cancer cells. There is an inverse association between Her-2/*neu* amplification and the expression of estrogen and progesterone receptors. Her-2/*neu* overexpression is also associated with high-grade tumors[@bib176], [@bib188] and is considered an adverse prognostic indictor in patients with breast cancer.[@bib187] The presence of Her-2/*neu* overexpression is associated with resistance to tamoxifen therapy[@bib189], [@bib190], [@bib191] and to CMF (cyclophosphamide, methotrexate, 5-fluorouracil) adjuvant chemotherapy but is associated with an increased response to regimens that use high-dose doxorubicin.[@bib192], [@bib193], [@bib194], [@bib195] Recent studies have linked amplification of the Her-2/*neu* and topoisomerase IIα genes to the effects of anthracyclines. Preliminary data suggest that coamplification of these two genes may identify a subgroup of high-risk breast cancer patients who might benefit from individually tailored and dose-escalated adjuvant anthracyclines.[@bib196], [@bib197] Her-2*/neu* can also be assessed through amplification of the gene by fluorescence in situ hybridization. #### EPIDERMAL GROWTH FACTOR RECEPTOR {#s0370} Epidermal growth factor receptor (EGFR) belongs to a family of growth factor receptors involved in normal growth. The gene is located on chromosome 7p12. It is the receptor for epidermal growth factor and is a member of the receptor tyrosine kinase family. It is closely related to Her-2/*neu*, Her-3, and Her-4. EGFR is known to be involved in carcinogenic processes such as cell proliferation, apoptosis, angiogenesis, cell motility, and metastasis. The expression of EGFR has been examined in a wide variety of tissues, and in many cases, increased expression of EGFR is predictive of tumor progression (e.g., cancer of the breast, esophagus, adrenals, lung, bladder, thyroid, and gastrointestinal tract and glioblastoma multiforme).[@bib198], [@bib199], [@bib200], [@bib201], [@bib202], [@bib203], [@bib204], [@bib205] In addition to immunohistochemical methods, fluorescence in situ hybridization has been used successfully to identify EGFR mutation or deletions on formalin-fixed, paraffin-embedded tissue.[@bib206] EGFR is also showing promise as a therapeutic target. Studies are under way in lung and colorectal cancer to determine the usefulness of targeting EGFR for anticancer therapy.[@bib207], [@bib208] ### Tumor Suppressor Genes and Gene Products {#s0380} The primary characteristics of tumor suppressor genes are that they encode normal cellular products involved in growth control, and both alleles must be inactivated for loss of function (i.e., loss of tumor suppression) to occur. The most well known are retinoblastoma (Rb) protein, p53, p27, p21, and p16. The two best characterized are the *Rb* and *p53* genes. Both are thought to be involved in growth control through the regulation of transcription. #### RETINOBLASTOMA GENE {#s0390} The *Rb* gene is located on chromosome 13q14 and is dysfunctional in a number of types of cancer. Its normal function is to prevent the replication of damaged DNA; it does so by preventing cell replication by binding and inhibiting the transcription factor E2F.[@bib209], [@bib210] The retinoblastoma protein (pRb) is activated when it is dephosphorylated and inactivated when it is phosphorylated. Alterations in this gene have been described in many human tumors, including retinoblastoma, osteosarcoma, other sarcomas, leukemias, lymphomas, and certain carcinomas, including breast, lung, prostate, bladder, kidney, and testicular carcinoma.[@bib29], [@bib211], [@bib212] Gene alterations are associated with advanced tumor grade and stage in a variety of tumors.[@bib211], [@bib213] Alterations in the *Rb* gene correlate with loss of expression of pRb as determined by immunohistochemistry.[@bib214] Assessment of *Rb* gene loss by immunohistochemistry is based on the loss of detectable nuclear staining for pRb. There is growing evidence that gene alterations may identify tumors that have a higher risk of developing metastases.[@bib215] Loss of heterozygosity, mutations, or deletions of the *Rb* gene usually result in the loss of pRb expression, which has been regarded as an indicator of loss of pRb function in human tumors. In addition to loss of pRb expression, aberrantly high pRb expression indicates a loss of pRb function in bladder tumors compared with moderate pRb expression.[@bib210], [@bib215] It has been shown that tumors with pRb overexpression demonstrate pRb hyperphosphorylation, mediated in part by the loss of p16 expression or overexpression of cyclin D1.[@bib210] #### P53 {#s0400} The *p53* gene is located on chromosome 17p13.1. The p53 protein is expressed by all normal cells, but the half-life of the normal protein is so short (6 to 30 minutes) that it does not accumulate in levels high enough to be detected by standard immunohistochemical techniques. Mutant p53 protein, by contrast, has an extended half-life, accumulates, and is readily detectable in the cell nucleus; mutation is indicated by positive staining. Alterations of the *p53* gene are extremely common in human cancer and have been described in bladder, colon, lung, breast, and other carcinomas; astrocytomas; leukemias; sarcomas; and mesotheliomas.[@bib1], [@bib29], [@bib213], [@bib216], [@bib217] Because of the importance of *p53* alterations in human cancer and the ease of detecting *p53* mutations by molecular or immunohistochemical methods, *p53* alterations have been the focus of intense examination. As with *Rb* alterations, *p53* alterations are associated with tumors of high histologic grade and a high proliferative index. There is growing evidence that, at least for some types of tumors, *p53* alterations identify patients with shorter disease-free intervals and poorer overall survival.[@bib217], [@bib218] #### CYCLIN-DEPENDENT KINASE INHIBITORS {#s0410} The cyclin-dependent kinase inhibitors are a family of cell cycle regulators. Their primary function seems to be the formation of stable complexes with cyclin-dependent kinase proteins and the subsequent inhibition of the cell cycle. These complexes inactivate the catalytically operative units. Among the most well known and clinically relevant are p21, p27, and p16. ##### p21 {#s0420} A member of the WAF/CIP/KIP family of cyclin-dependent kinase inhibitors, p21 is probably the best characterized. It acts as a regulator of epithelial carcinogenesis and differentiation and is thought to play an important role in tumor suppression by regulating cell cycle progression, DNA replication, and DNA repair.[@bib29], [@bib219] The protein expression of p21 has been studied in a variety of tumor types, including breast,[@bib220] gastric,[@bib221] ovary,[@bib222] colorectal,[@bib223] and bladder[@bib213], [@bib224] carcinomas. The alteration of protein expression assessed by immunohistochemical methods has been associated with higher tumor grade and worse prognosis in patients with bladder cancer.[@bib123], [@bib213] ##### p27 {#s0430} The p27 inhibitor is involved in the regulation of the cell cycle at the G~1~-S transition, ultimately through the inhibition of pRb phosphorylation.[@bib225] Mutations in the human *p27* gene appear to be rare.[@bib226] Loss of p27 expression is associated with colon, breast, prostate, and gastric cancer progression.[@bib227], [@bib228], [@bib229], [@bib230], [@bib231] ##### p16 {#s0440} Also known as p16^INK4^ and CDKN2A, p16 is a tumor suppressor protein encoded on the INK4a/ARF locus of chromosome 9p21, which is one of the most frequent sites of genetic loss in human cancer.[@bib232] Numerous studies have found abnormal p16 protein in a variety of tumor types, including melanomas; gliomas; esophageal, pancreatic, lung, and bladder carcinomas; and certain types of lymphomas.[@bib232], [@bib233], [@bib234], [@bib235], [@bib236], [@bib237], [@bib238], [@bib239], [@bib240] In addition, p16 is known to regulate *Rb*, and immunohistochemical expression of pRb and p16 is inversely correlated in a variety of tumors.[@bib241], [@bib242] #### COMBINED EFFECTS OF P53, P21, AND PRB {#s0450} It is known that, individually, p53, p21, and pRb are independent predictors of time to recurrence and overall survival in patients with bladder cancer.[@bib215], [@bib217], [@bib224] Efforts have therefore been made to examine these determinants in combination.[@bib123], [@bib213] In one study, patients were analyzed according to whether none, one, two, or all three markers were positive. The 5-year survival rates were 70%, 58%, 33%, and 8%, respectively. These data suggest that alterations in p53, p21, and pRb act in cooperative or synergistic ways to promote bladder cancer progression. #### TUMOR ONCOGENE CYCLIN D1 {#s0460} Cyclin D1 plays a key role in the regulation of the G~1~-S transition phase of the cell cycle. It has been linked to a number of different cancers, including colorectal, esophageal, gastric, lung, head and neck, and pancreatic cancer.[@bib243] Predicting Response to Therapy {#s0470} ------------------------------ Although a major purpose of the molecular assessment of cancer is to better understand the risk for disease progression, advanced technologies are also being used to understand the specific patterns of response and resistance to therapeutic regimens. The traditional means of determining appropriate systemic treatment generally involved histogenic classification. It has long been recognized, however, that response to hormonal therapy can be predicted specifically by molecular determinants (e.g., the expression of estrogen and progesterone receptors in breast and other cancers of reproductive organs).[@bib244] Tumors arising from the breast, prostate, endometrium, and ovary are known to be regulated by steroid sex hormones (estrogens, androgens). It was discovered that removing the source of hormones that control tumor growth (by oophorectomy, orchiectomy, or chemical methods) sometimes resulted in dramatic tumor remission.[@bib244] Growth regulation was found to be associated with the amount of specific hormone receptors: Tumors that expressed high levels of these receptors tended to respond well to hormone ablation, whereas those with few or no receptors tended not to respond to this type of therapy. Accurate methods for determining the presence or absence of hormone receptors are essential for determining the best method of treatment. The availability of monoclonal antibodies to estrogen, progesterone, and androgen receptors has made immunohistochemical detection of hormone receptor status the current method of choice. These immunohistochemical methods can be performed on formalin-fixed, paraffin-embedded tissue and on cytology specimens. Immunohistochemical antireceptor assays allow one to predict breast cancer\'s response to hormonal treatment.[@bib245], [@bib246] Tumors that do not express estrogen or progesterone receptors have a low probability of responding to hormonal manipulation, whereas estrogen receptor-- and progesterone receptor--positive tumors have a high probability of responding to such treatment. Many practitioners believe that the only relevant result for hormone receptors in breast cancer is "positive" or "negative." However, some investigators have shown that the level of hormone receptor is important as well.[@bib8], [@bib9] Although a proportion of patients with low levels of hormone receptor will respond to hormone therapy, most benefit from the addition of systemic cytotoxic chemotherapy. In contrast, in patients with high levels of hormone receptor, the addition of cytotoxic chemotherapy has a deleterious effect on outcome.[@bib8], [@bib9] Recently, attention has focused on expression of the estrogen receptor subtypes α and β and on various isoforms of the β subtype. It has been found that estrogen receptor α--negative tumors express significant levels of estrogen receptor β1 and β5 and that their expression levels are no different from levels in estrogen receptor α--positive tumors.[@bib246] Therefore, these two estrogen receptor isoforms may be potential molecular targets for designing chemopreventive drugs to treat estrogen receptor α--negative breast cancers. P-glycoprotein is a transmembrane protein of 170 kD that has been associated with intrinsic and acquired resistance to certain chemotherapeutic agents, particularly anthracyclines and vinca alkaloids. P-glycoprotein also may play a role in tumor progression and has been associated with blood vessel invasion and lymph node metastases.[@bib2] Some tumors inherently express P-glycoprotein, whereas other tumors acquire expression only after exposure to certain chemotherapeutic agents.[@bib2] Overexpression is associated with failure of chemotherapy.[@bib2], [@bib247] Other predictors of response to specific forms of chemotherapy are being explored. The prevailing view has been that *p53* alterations should result in a chemoresistant phenotype. This view is based on a body of evidence showing that wild-type *p53* is required for entrance into the apoptotic pathway at the G~1~- to S-phase transition.[@bib248], [@bib249] Because chemotherapy works through the induction of apoptosis, *p53* alterations may result in resistance to such agents. We are conducting a clinical trial concerning the role of *p53* in predicting progression and response in patients with bladder cancer.[@bib250] It is also possible that *p53* may promote chemoresistance by other mechanisms, such as through induction of the multidrug resistance (*MDR-1*) gene.[@bib249], [@bib251] In tumors in which *p53* alterations confer increased (selective) chemosensitivity, combining agents that have different actions (e.g., DNA damage versus inhibition of the G~2~M checkpoint) may have synergistic effects on tumor cell killing, a finding that has important implications in the design of new combination chemotherapy regimens.[@bib252] The expression of thymidylate synthase in colorectal tumors predicts resistance to the most common type of systemic chemotherapy used in that disease, 5-fluorouracil.[@bib253], [@bib254] As mentioned earlier, Her-2/*neu* overexpression in breast cancer predicts resistance to hormone therapy in estrogen receptor--positive tumors[@bib189], [@bib190], [@bib255] and resistance to some types of chemotherapy, but increased sensitivity to doxorubicin-based regimens.[@bib193], [@bib194] Her-2/*neu* and EGFR are specific targets of antibody-directed therapy. In the case of Her 2/*neu*, it seems that only those tumors that overexpress the target are likely to respond to such therapies.[@bib256], [@bib257], [@bib258] In the case of colorectal cancer, therapy directed against EGFR appears to work only in tumors with wild-type KRAS; tumors with mutant KRAS do not respond to EGFR therapy.[@bib9002] Similar findings are seen in lung cancer.[@bib9003] The ability to predict the specific response of individual tumors to chemotherapeutic agents can have a profound effect on treatment decisions for patients with cancer. It is not difficult to envision the day when drug selection is based on the resistance patterns of individual tumors to specific agents. Treatment decisions will become less organ based and will better reflect the biology of the tumors. Infections {#s0480} ---------- Traditionally, the stains available to surgical pathologists to identify infectious organisms in tissue sections consisted of Gram stain, variations of the acid-fast stain, periodic acid--Schiff, and silver stains. There is now a wide range of immunohistochemical or in situ hybridization techniques available for the detection of specific types of organisms within fixed paraffin sections. Although culturing techniques remain the most important method for diagnosing most infections, immunohistochemical methods are as effective as, or even superior to, culture and routine H&E methods for the detection of certain infectious organisms such as cytomegalovirus ([Fig. 5-9](#f0090){ref-type="fig"} ), mycobacteria, *Toxoplasma*, *Pneumocystis carinii*, *Histoplasma capsulatum*, *Helicobacter pylori,* and human papillomavirus.[@bib259], [@bib260], [@bib261], [@bib262], [@bib263], [@bib264], [@bib265], [@bib266], [@bib267], [@bib268], [@bib269] [Table 5-4](#t0040){ref-type="table"} summarizes some of the infectious agents that can be identified by immunohistochemistry.Figure 5--9 ▪Placenta showing infection with cytomegalovirus by immunohistochemistry.TABLE 5--4Infectious Agents for Which Antibodies Are Available for Use on Paraffin SectionsAdenovirus*Aspergillus*Baboon endogenous virus*BlastomycesBorrelia burgdorferi*Buffalo pox virus*Campylobacter coliCampylobacter jejuniCampylobacter* spp.*CandidaChylamydiaCoccidioides*Coronavirus*Cryptococccus neoformansCryptosporidium*CytomegalovirusDistemper virus*Entamoeba histolytica*Epstein-Barr virus*Escherichia coliFasciola hepatica*Friend\'s virus*GiardiaHelicobacter pylori*Hepatitis A virusHepatitis B core antigenHepatitis B surface antigenHepatitis C virusHerpes simplex virus 1 and 2*Histoplasma capsulatum*Human immunodeficiency virus (HIV-1)Human papillomavirusInfluenza*KlebsiellaLegionellaLeishmania*Lymphocytic choriomeningitis virusMeasles antigenMoloney virusMouse mammary tumor virus antigenMycobacteria*MycoplasmaParainfluenzaPneumoncystis carinii*PolioPolyomavirus*Pseudomonas aeruginosa*Rabies virusRespiratory syncytial virusRotavirusRubella*SalmonellaShigella*Shope\'s fibroma virus*StaphylococcusStreptococcus*SV40 virus*Toxoplasma gondiiTreponema pallidumTrichophyton*Varicella-zoster virus*Yersinia* [^1]: +, always positive; --, negative; --/+, mostly negative; R, rare positive cells. [^2]: CEA, carcinoembryonic antigen; CLA, common leukocyte antigen; COTA, colonic ovarian tumor antigen; HMW, high molecular weight; LMW, low molecular weight; NSE, neuron-specific enolase; PNET, primitive neuroectodermal tumor. [^3]: CEA, carcinoembryonic antigen; PTH, parathyroid hormone.

Introduction ============ Breast cancer diagnosis and prognosis is currently based on histological grade, disease stage (including lymph node status) and expression of hormone receptors (oestrogen receptor (ER), progesterone receptor (PgR) and *ERBB2* (HER2)) ([@ddw177-B1]). As breast cancer is a very heterogeneous disease and clinical outcomes can be highly variable, the predictive power of these markers is sometimes limited. The majority of ER+ patients receive tamoxifen or other forms of endocrine therapies, although many display or develop resistance, with a 33% recurrence rate after 5-years of tamoxifen treatment ([@ddw177-B2]). Such endocrine resistance is common but difficult to predict (reviewed in Osborne and Schiff, 2011 ([@ddw177-B3])). Patients with HER2 positive tumours may be treated with Trastuzumab (Herceptin), however relapse in the form of metastasis is a regular occurrence ([@ddw177-B4]). Thus, there is a clear and on-going need to identify additional prognostic and predictive molecular biomarkers that will augment traditional indicators and ultimately improve patient management and better predict response to therapies in these two types of breast cancer ([@ddw177-B5]). Gene transcription is regulated by a complex interplay of proximal and distal *cis*-acting regulatory elements, including promoters and enhancers, and *trans*-acting factors such as protein transcription factors (TFs) and non-coding RNAs. Whilst promoter elements lie proximal to the transcriptional start site (TSS) and are responsible for basal gene transcription, enhancers can be kilobases or megabases from the gene they regulate and are associated with cell- and tissue-specific expression ([@ddw177-B6],[@ddw177-B7]). Disruption of enhancers through epigenetic or genomic aberrations can therefore impact cell identity, a key feature of tumourigenesis (reviewed in Kron *et al*., 2014 ([@ddw177-B8]) and Herz *et al*., 2014 ([@ddw177-B9])) and so it is no surprise that these defects have been associated with cancer initiation and progression. For example, several breast cancer-associated SNPs map to enhancer elements bound by oncogenic transcription factors which regulate the expression of breast cancer susceptibility genes ([@ddw177-B10],[@ddw177-B11]). Defects in the factors that mediate enhancer function have also been associated with cancer, including CTCF in breast and bladder cancer, and MED12 in prostate cancer. Enhancer elements and the factors controlling them therefore have significant potential to predict breast cancer progression ([@ddw177-B8],[@ddw177-B12],[@ddw177-B13]). Non-coding RNA genes are regulated at both the transcriptional and posttranscriptional level. Whole genome chromatin immunoprecipitation (ChIP) sequence analysis has shown that most miRNA promoters are over 1kb in length and map to distances up to 10kb from the miRNA sequence ([@ddw177-B14]). The regulation of long non-coding RNA (lncRNA) genes is less understood, however it is now known that histone modification marks are associated with the regulatory elements controlling lncRNA gene expression ([@ddw177-B18]). In addition, this analysis has revealed a distinctive methylation pattern around the transcription start site (reviewed in Venkatesh and Workman 2012 ([@ddw177-B19])), strong evolutionary conservation and binding sites for transcription factors that play a key role in cell proliferation and differentiation ([@ddw177-B20]). Enhancers have emerged as crucial regulators of lncRNAs and their status is associated with the regulation of critical processes including the determination of cell fate and identity ([@ddw177-B7],[@ddw177-B21]). These enhancer elements can bind a plethora of transcription factors and physically interact with gene promoters through chromatin looping to alter transcription (reviewed in Fraser 2006 ([@ddw177-B22])). Hox antisense intergenic RNA (*HOTAIR*) is a lncRNA that has been implicated in chromatin remodeling and transcription. *HOTAIR* RNA associates with the polycomb repressive protein complex 2 (PRC2) ([@ddw177-B23]), which is part of a vital poly-protein structure used to condense chromatin and effectively control transcription ([@ddw177-B24]). *HOTAIR* regulates a wide range of genes, including *HOXD* genes, genes associated with epithelial-to-mesenchymal transition (EMT) and cell cycle regulation ([@ddw177-B25]). *HOTAIR* is a marker of metastasis and poor prognosis in a range of cancers, including breast cancer ([@ddw177-B27]), where it is overexpressed in a third of metastatic breast tumours ([@ddw177-B26]). The ability of *HOTAIR* or its regulators to predict therapeutic response and outcome in ER and HER2 positive breast cancer has yet to be determined. Several elements and factors controlling the transcriptional and post-transcriptional regulation of *HOTAIR* have been described and implicated in cancer (reviewed in Hajjari and Salavaty, 2015 ([@ddw177-B28])). At the transcriptional level, a proximal *HOTAIR* promoter has been identified approximately 1kb upstream of the *HOTAIR* TSS and several transcription factors have been shown to bind this region and regulate *HOTAIR* expression, including c-Myc ([@ddw177-B29]), IRF1 (30), ERα ([@ddw177-B31]) and MLL proteins. A susceptibility single nucleotide polymorphism (SNP) for esophageal squamous cell carcinoma (ESCC) and breast cancer has been found located within an enhancer in intron 2 of *HOTAIR* ([@ddw177-B32],[@ddw177-B33]). In ESCC, the TT allele is associated with increased expression through higher enhancer activity. Methylation of proximal CpG islands mapping approximately 1kb downstream of *HOTAIR* have been associated with clinical and pathological features of breast cancer ([@ddw177-B34]). The role of long-range regulation in the control of *HOTAIR* has not been reported and is the subject of this manuscript. In this paper, we have elucidated the role of long-range regulation in the control of HOTAIR expression and explored the potential of long-range elements and factors to be valuable biomarkers for therapeutic response and outcome in breast cancer. Here, we report that FOX proteins can regulate *HOTAIR*, and can enhance its ability to stratify both HER2-enriched and ER+ breast cancer. Results ======= Identification of a distal enhancer downstream of the *HOTAIR* gene ------------------------------------------------------------------- To identify transcription control elements and factors that regulate *HOTAIR* expression in breast cancer cells, a chromatin state map for the *HOXC* locus housing the *HOTAIR* gene was generated using publically available data on evolutionary conservation, histone modification, ChIP-Seq and RNA-Seq. This analysis revealed a series of highly conserved regions that exhibit histone modifications and TF-binding densities suggestive of *cis*-regulatory activity ([Figure 1A](#ddw177-F1){ref-type="fig"}, higher resolution of the HOXC locus: [Supplementary Material, Fig. S1](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Figure 1.Identification of a *HOXC* putative *cis*-regulatory element by investigating chromatin modifications and long-range interactions. **(A)** Histone modification and transcription factor binding across the *HOXC* locus, sourced from publically available ChIP-Seq data (Hg18), suggesting a putative enhancer ∼150 kb downstream of the *HOTAIR* gene. **(B)** ChIA-PET interactions for ESR1, RNA Polymerase II and CTCF with data sourced from GEO, GSE39495. Plots representing each interaction library include a wiggle track showing the binding of each transcription factor and a curved line that connects interacting genomic fragments indicating relative interaction frequency between fragments (right Y axis). Of particular interest was a region located in a gene desert approximately 150 kb downstream of the *HOTAIR* TSS, marked by open chromatin (FAIRE-Seq), lysine mono-methylation (H3K4me1), RNA Polymerase II binding, as well as ER and PR binding, in multiple different breast cancer cell lines ([Figure 1A](#ddw177-F1){ref-type="fig"}). These marks are consistent with enhancer activity, and we therefore denoted this region as a putative distal enhancer, which we have termed H[OXC D]{.ul}istal [E]{.ul}nhancer (HDE). The enrichment of ERα and PgR binding at the HDE was associated with binding of FOXA1, a critical mediator of ER DNA binding in breast cancer cells. By interrogating ChIA-PET (Chromatin Interaction Analysis by Paired-End Tag Sequencing) data, we also found that the HDE engages in long-range interactions with the *HOXC* locus involving ERα, RNA Polymerase II and CTCF TFs ([Figure 1B](#ddw177-F1){ref-type="fig"}). Mapping of enhancer-promoter interactions reveals an alternative promoter for *HOTAIR* -------------------------------------------------------------------------------------- To investigate the interaction between the HDE and the *HOTAIR* promoter, chromosome conformation capture (3C) was performed. 3C libraries were generated for both ER+ (MCF7, ZR751) and ER- (MDA-MB-453 and MDA-MB-468) cell lines. Using the *Hind*III fragment containing the HDE as bait, 3C-qPCR relative interaction graphs were constructed for all fragments across the *HOXC* locus. In MCF7, ZR-751 and MDA-MB-453 cells, the HDE showed significant interactions with the *HOTAIR* start site ([Figure 2A](#ddw177-F2){ref-type="fig"}). Digestion with *Bgl*II effectively increased the resolution and revealed that the HDE interacts with two distinct regions within and upstream of the *HOTAIR* gene ([Figure 2B](#ddw177-F2){ref-type="fig"}). The first region corresponds to the promoter (P1) and the first intron of the canonical *HOTAIR* gene, and the second region, approximately 5kb upstream, corresponds to an alternate *HOTAIR* TSS, within intron-1 of *HOXC11* ([Figure 2C](#ddw177-F2){ref-type="fig"}, alternative promoter cloned regions 1, 2 and 3, (ALT1, ALT2 and ALT3)). The alternate TSS is enriched for the histone mark H3K4me3 and RNA Polymerase II, suggesting *bona fide* promoter activity. The canonical *HOTAIR* promoter exhibits low levels of histone modifications that are suggestive of an active regulatory element (H3K4me3 and H3K27ac), perhaps suggesting a poised or inactive promoter in MCF7 cells. RT-PCR analysis demonstrated that an alternate longer form of the *HOTAIR* transcript is expressed in MCF7 cells originating from the alternative TSS (promoters ALT1-3) ([Supplementary Material, Fig. S2](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Figure 2.Mapping of enhancer-promoter interaction reveals alternative promoter for *HOTAIR*. **(A)** 3C analysis using libraries generated by *Hind*III in MCF7, ZR-75-1, MDA-MB-453 and MDA-MB-468. *Hind*III fragment length and position are shown in black corresponds to the center of each fragment within the graphs. **(B)** 3C fine mapping analysis using *Bgl*II in MCF7 breast cancer cells, position and fragment length relative to *HOXC* genes shown in black. X-axis is the relative genomic position corresponding to the start of *Bgl*II fragments. 3C Y-axis represents the relative interaction frequency between the *Hind*III/*Bgl*II fragments and the HDE. Error bars represent standard deviation. **(C)** *HOTAIR* promoter (P1) and alternative promoters (ALT1-3). Histone marks and RNA Pol II binding are characteristic of gene promoters. Black bars highlight the regions used for each pGL3 construct. **(D)** Schematic for luciferase reporter constructs used in E. **(E)** Luciferase reporter results for activity of *HOTAIR* alternative promoters in MCF7 cells. X-axis represents the construct used while the Y-axis is the relative light units (luciferase normalized to pRL-TK Renilla and pGL3-Basic (V)). Statistical significance was found via multiple two-tailed *t-*Tests between indicated columns, P-values are \<0.05 (\*), \<0.01 (\*\*), \<0.001 (\*\*\*) and \<0.0001 (\*\*\*\*). The putative enhancer augments *HOTAIR* promoter activity --------------------------------------------------------- To determine whether the HDE can enhance the transcription of genes at the *HOXC* locus, we generated luciferase reporter constructs by sub-cloning the HDE upstream of the *HOXC10*, *HOXC11* or *HOTAIR* promoters ([Supplementary Material, Figures S3 and S4A](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Luciferase assays were then performed in several breast cancer cell lines. We were unable to detect transcriptional activity from the *HOXC11* promoter in these breast cancer cells and therefore excluded this construct from further analysis. Of the promoters tested, the HDE had the greatest effect on the canonical *HOTAIR* promoter, inducing a five-fold increase in luciferase expression compared to the *HOTAIR* promoter alone ([Supplementary Material, Fig. S4B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). To investigate the promiscuity of the HDE in affecting HOXC promoter activity we cloned it upstream of the SV40 promoter ([Supplementary Material, Fig. S4C](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Use of the SV40 promoter demonstrates a 12-fold increase in transcriptional activity when in combination with the HDE, suggesting that the HDE is non-specific in its activity on the *HOTAIR* and *HOXC10* promoters ([Supplementary Material, Fig. S4D](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Three regions were cloned to test promoter activity at the alternative TSS, upstream of the TSS (ALT1), the centre of the histone markers and RNA Pol II binding (ALT2) and a larger fragment encompassing the alternative TSS and peaks of RNA Pol II binding. Luciferase reporter assays demonstrated promoter activities for P1, ALT1 and ALT3 that were significantly augmented by the HDE ([Figures 2D and E](#ddw177-F2){ref-type="fig"}). These luciferase data support the RNA Pol II ChIA-PET and 3C interactions in that the HDE specifically interacts with and augments *HOTAIR* promoters ([Figure 1B](#ddw177-F1){ref-type="fig"}). In support of these findings, analysis of MCF7 RNA-Seq data clearly demonstrates RNA produced from the longer alternative HOTAIR transcripts ([Supplementary Material, Fig. S5](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). To identify the active sequences of the HDE, we mutagenized the *HOTAIR*-HDE canonical promoter (P1) reporter construct ([Figure 3A](#ddw177-F3){ref-type="fig"}), and localized *HOTAIR* enhancer activity to a 114 base-pair region in MCF7 cells (between mutants 4 and 5; [Figure 3B](#ddw177-F3){ref-type="fig"}). Interestingly, this region contains several highly conserved, predicted transcription factor binding sites, including consensus binding sequences for forkhead box proteins (FOX Site 1 and 2) and AP1, which are associated with oestrogen-mediated transcriptional activity ([Figure 3B](#ddw177-F3){ref-type="fig"}). Consistent with this, analysis of ChIP-Seq data demonstrated that both c-Fos (part of the AP1 complex) and FOXM1 bind to the HDE ([Figure 3C](#ddw177-F3){ref-type="fig"}). Interestingly, in MCF7 cells c-Fos binding appears to be dependent on the presence of oestrogen. Binding of FOXM1, also in MCF7 cells, is decreased following the addition of the antibiotic thiostrepton (ThSt), which is known to prevent binding of FOXM1 to regulatory elements and chromatin. Furthermore, mutagenizing the sites in the context of the P1-HDE construct significantly reduced the enhancer activity of the HDE ([Figures 2D](#ddw177-F2){ref-type="fig"} and [3B](#ddw177-F3){ref-type="fig"}). Figure 3.The *HOTAIR* distal enhancer is dependent on FOX and AP1 binding. **(A)** Functional HDE sequence top and bottom the HDE mutant constructs for deletion and transcription factor motifs. Deletion mutant positions were chosen based on conservation with lines indicating remaining DNA. Up to three base pairs were edited within the TF motif sites for three separate constructs, one per motif listed. **(B)** Luciferase reporter assay measuring the transcriptional activity of the *HOTAIR* enhancer and the separate mutants, Y-axis as above. All mutants were generated using the P1 + HDE reporter vector as of [Figure 2D](#ddw177-F2){ref-type="fig"}. **(C)** Re-analyzed ChIP-Seq data (\[GSE26831\] ([@ddw177-B80]), \[GSE40767\] ([@ddw177-B44])) demonstrating binding of both cFOS and FOXM1 to the HDE in MCF7 cells with the indicated treatment. All statistical tests performed were one-way ANOVAs with Tukey corrected multiple comparisons between (B) P1 + +oE, P-values are \<0.05 (\*), \<0.01 (\*\*), \<0.001 (\*\*\*) and \<0.0001 (\*\*\*\*). *HOTAIR* is transcriptionally upregulated by FOX proteins --------------------------------------------------------- Given the role of FOX proteins in mediating enhancer activity of the HDE, we assessed the effect of FOXA1 and FOXM1 siRNA mediated knockdown on *HOTAIR* expression. FOXA1 and FOXM1 are key components of the ER signalling pathway in breast cancer and play important roles in disease progression, hence were prioritized for this study. This analysis showed a significant reduction in *HOTAIR* expression following knockdown of the FOX proteins ([Figure 4A](#ddw177-F4){ref-type="fig"}), suggesting that FOXM1 and FOXA1 positively regulate *HOTAIR*. In further support of these data, MCF7 cells treated with ThSt display significantly reduced the expression of *HOTAIR* and a known transcriptionally target of ThSt, *FOXM1* ([Figure 4B](#ddw177-F4){ref-type="fig"}). Figure 4.*HOTAIR* is transcriptionally regulated by FOX proteins and repressed by oestrogen. **(A)** Top, western blot for indicated proteins, demonstrated siRNA knockdown in MCF7 cells and below qRT-PCR results for *HOTAIR* expression in each knockdown, normalized to *RPLP0* and the scrambled control. **(B)** Expression of *HOTAIR* and *FOXM1* following treatment of MCF7 cells by DMSO (vehicle) or thiostreptin (ThSt) **(C)** Re-analysis of MCF7 GRO-Seq data following E2 treatment \[GSE27463\] ([@ddw177-B81]). The yellow histograms represent the positive strand and blue, the negative. The X-axis corresponds to the genomic position from Hg18 and the Y-axis the relative reads, *HOTAIR* promoters are shown above gene track. **(D)** RNA-Seq results for *HOTAIR* from E2 treated MCF7 cells over 72 hours (Reads per kilobase per million = RPKM). **(E)** qRT-PCR results of *HOTAIR* expression from a 10 nM E2 time course in MCF7 cells, representative of three separate experiments, data not shown. Y-axis represents relative expression (ΔΔCt of *RPLP0*/DMSO). Statistical analysis was performed via a one-way ANOVAs with Tukey's corrected multiple comparisons between indicated columns or data points for A and E and via a two-tailed *t-*Test for B, P-values are \<0.05 (\*), \<0.01 (\*\*), \<0.001 (\*\*\*) and \<0.0001 (\*\*\*\*). FOXA1 binds extensively throughout the HOXC locus ([Supplementary Material Figure S1](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)) and given the multiple 3C interactions between this locus and the HDE ([Figure 2A](#ddw177-F2){ref-type="fig"}), we assessed gene expression changes of the locus following depletion of FOXA1 in MDA-MB-453 cells. All *HOXC* genes and the miRNAs MIR196A and MIR615 responded with reduction in expression following FOXA1 depletion except for *HOXC4* ([Supplementary Material, Figs. S6A and B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). *HOTAIR* is transcriptionally repressed by oestrogen ---------------------------------------------------- AP1, FOXA1 and FOXM1 facilitate the activity of oestrogen-mediated transcription in breast cancer. Given the role of AP1, FOXA1 and FOXM1 in augmenting the activity of the HDE and that c-Fos binding to the HDE is oestrogen-dependent ([Figure 3C](#ddw177-F3){ref-type="fig"}), we investigated the effects of oestrogen on *HOTAIR* expression. We re-analyzed GRO-Seq data from MCF7 E2 treated cells; a sensitive, genome-wide run-on sequencing assay that maps the position, amount, and orientation of transcriptionally engaged polymerases. This analysis revealed that the amount of polymerase engaged with the *HOTAIR* alternative promoter (negative strand) decreased over time, with minimal engagement by 160 min post-E2 treatment ([Figure 4C](#ddw177-F4){ref-type="fig"}). Importantly, we found that *HOTAIR* (canonical transcript) levels in MCF7 cells decreased in response to E2 treatment, as determined by RNA-Seq and qRT-PCR analysis using two different TaqMan® Probes, that detect both the canonical and alternative isoforms ([Figures 4D and E](#ddw177-F4){ref-type="fig"}). These data suggest that oestrogen negatively regulates *HOTAIR* transcription. *HOTAIR* expression is inversely correlated with oestrogen receptor signalling and directly correlated with *FOXM1* in breast cancer ------------------------------------------------------------------------------------------------------------------------------------ To investigate the relationship between *HOTAIR* and oestrogen signalling in breast cancer, we performed unsupervised hierarchical clustering of publically available (TCGA) breast cancer RNA-Seq data ([Figure 5A](#ddw177-F5){ref-type="fig"}). Given that numerous *HOXC* genes appear to be regulated by FOXA1 and possibly oestrogen, we clustered the expression of *HOXC* protein-coding genes, *HOTAIR* and known ERα cofactors. This analysis revealed a large cohort (Cluster 1 comprising 264 tumours and representing 49% of all cases) in which *HOTAIR* was inversely correlated with *ESR1* and ERα cofactors: *FOXA1*, *GATA3*, *GREB1*, *TLE3* and *FOS* ([Figure 5A](#ddw177-F5){ref-type="fig"} and [Supplementary Material, Table S1](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). The proportion of breast cancer subtypes in Cluster 1 is heterogeneous, but is significantly enriched with triple-negative tumours (ER/PgR/HER2-negative by IHC), as well as those classified as basal-like, normal-like and HER2 by gene expression profiling ([Figure 5A](#ddw177-F5){ref-type="fig"}, [Supplementary Material, Figures S7A and B](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Figure 5.*HOTAIR* associates with *FOXM1* to significantly stratify ER positive breast tumours. **(A)** Manhattan based hierarchical clustering of *HOTAIR*, HOXC genes, ERα and ERα cofactors in TCGA breast tumour data ([@ddw177-B75],[@ddw177-B78]) with hormone receptor IHC status for each tumour displayed above. Pearson correlation coefficients for each cluster on the right. For full table and P-values see [Supplementary Material, Table S1](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1). Arrows highlight rows for *HOTAIR, FOXM1* and *ESR1* (*ERα*) **(B-D)** Kaplan-Meier curves for the stratification of distant metastasis free survival (DMFS) for patients with ER+ node positive (N+) tumours by *HOTAIR* or *FOXM1* alone or their combination. Logrank hazards ratios with their 95% confidence intervals (CI) and P-values are shown and tumour number in the low or high expression groups indicated on right. **(E-G)** As above, except for relapse free survival (RFS) of ER+ tumours in patients that received tamoxifen (Tam). **(H)** *HOTAIR*, *ESR1*, *FOXM1* and *FOXA1* expression across MCF7 and MCF7 derived cell lines (ΔΔCt of RPLP0/MCF7, n = 2). All qRT-PCR error bars represent standard deviation. *HOTAIR* and *FOXM1* in combination predict response to therapy for ER+ tumours ------------------------------------------------------------------------------- Given the association between *HOTAIR* and *FOXM1* expression and regulation of the HDE by FOX proteins, we sought to determine if these factors in combination may have prognostic potential in ER+ breast cancer. *HOTAIR* expression significantly stratifies the survival of patients with ER+ tumours and the combination of *HOTAIR* and *FOXM1* expression increases this stratification ([Supplementary Material, Table S2](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). The more aggressive ER+/N+ tumours were stratified by *HOTAIR* expression based on relapse free survival (RFS), however this was not observed for distant metastasis free survival (DMFS) ([Supplementary Material, Table S3](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1) and [Figure 5B](#ddw177-F5){ref-type="fig"}). *FOXM1* alone is able to significantly stratify these tumours, however, the combination of *HOTAIR* and *FOXM1* increased this stratification through an increase of the hazard ratio ([Figures 5C and D](#ddw177-F5){ref-type="fig"}). *HOTAIR* and *FOXM1* were also assessed for their potential to stratify patients on the basis of response to endocrine treatment and chemotherapy. Utilizing these biomarkers the RFS of patients treated with tamoxifen only (Tam), any endocrine therapy (ET) or the combination of endocrine and chemotherapy (CT) was stratified ([Supplementary Material, Table S4](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1), [Figure 5E](#ddw177-F5){ref-type="fig"}). *HOTAIR* alone could only stratify survival of patients who received CT in combination with ET, where high expression associated with poor survival. *FOXM1* alone is a significant biomarker for response to tamoxifen in these ER+ tumours, however HOTAIR + FOXM1 display the largest hazard ratio where patients with low expression have a 95% chance of survival beyond 10 years ([Figures 5F and G](#ddw177-F5){ref-type="fig"}). Taken together these data indicate that *HOTAIR* in combination with *FOXM1* enhances the value of these genes as biomarkers to predict response to any of the therapeutic options, especially patients with aggressive ER+ tumours. Given the intersection of *HOTAIR* transcriptional regulation and the oestrogen-signalling pathway, we hypothesized that *HOTAIR* plays a role in endocrine resistance. To explore this, *HOTAIR* expression was determined in MCF7-derived *in-vitro* models of oestrogen deprivation or anti-oestrogen resistance: MCF7X, TAMR and FASR. These cell line models were derived following prolonged *in vitro* oestrogen deprivation (MCF7X), tamoxifen treatment (TAMR) or fulvestrant treatment (FASR). *HOTAIR* expression levels were increased in MCF7X and TAMR cells ([Figure 5H](#ddw177-F5){ref-type="fig"}). In contrast, FASR cells had significantly less *HOTAIR* RNA. Together with *HOTAIR*, expression of *ESR1*, *FOXM1* and *FOXA1* were also altered, particularly in TAMR cells, which have decreased expression of *ESR1* while maintaining expression of the FOX genes ([Figure 5H](#ddw177-F5){ref-type="fig"}). These data indicate that *HOTAIR* expression increases in cell line models of endocrine therapy resistance. *HOTAIR* regulators enhance the power of *HOTAIR* as a biomarker in HER2-enriched tumours ----------------------------------------------------------------------------------------- To determine whether the prognostic potential of *HOTAIR* and *FOXM1* coexpression applies to other breast cancer subtypes, *HOTAIR* levels were assessed across all of the intrinsic molecular subtypes of breast cancer. This analysis revealed significant enrichment of expression in the HER2-enriched subtype ([Figure 6A](#ddw177-F6){ref-type="fig"}). Both *FOXA1* and *FOXM1* are also highly expressed in this subtype, whilst *ESR1* expression is depleted ([Figure 6B](#ddw177-F6){ref-type="fig"}). Interestingly, both *HOTAIR* and *FOXM1* (HOTAIR + FOXM1) expression is significantly enriched in the basal-like and HER2-enriched subtype ([Supplementary Material, Figure S8](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). Figure 6.*HOTAIR* regulators improve utility of *HOTAIR* in stratifying HER2-Enriched tumours. **(A and B)** Relative expression of *HOTAIR* and transcription factors across the PAM50 intrinsic breast cancer subtypes. Referring to A, horizontal bars are the mean and error bars standard deviation. Significance was found via one-way ANOVAs with Tukey's corrected multiple comparisons for HER2 tumours against all others. P-values are \<0.05 (\*), \<0.01 (\*\*), \<0.001 (\*\*\*) and \<0.0001 (\*\*\*\*). Tumour numbers are as follow Basal-like (140, Basal), HER2-enriched (67, HER2), Luminal A (420, LumA) Luminal B (194, LumB) and Normal-like (24, N-Like). **(C-E)** Kaplan-Meier curves stratifying the overall survival of patients with HER2-enriched tumours by *HOTAIR* alone, the transcription factors or their combination. Logrank hazards ratios with their 95% confidence intervals (CI) and P-values indicated b elow. Tumour number in the low or high expression groups are indicated in each graph. **(F to G)** As above, except for the relapse free survival (RFS) of patients with HER2-enriched Grade 3 tumours. *HOTAIR* expression significantly stratified the RFS of patients with luminal B, HER2-enriched and ER+ tumours, with high expression associating with poor outcome ([Supplementary Material, Table S5](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). High expression of *HOTAIR*, *FOXA1*, *FOXM1* and *FOS* combined with low expression of *ESR1*, (HOTAIR + TFs), as a functional network, enhanced the stratification overall survival (OS) for patients with HER-enriched tumours ([Figures 6C-E](#ddw177-F6){ref-type="fig"}). This enhanced stratification is most evident in the more aggressive Grade 3 HER2-enriched tumours where HOTAIR + TFs expression significantly stratified the RFS greater than either HOTAIR or the TFs alone ([Figures 6F--H](#ddw177-F6){ref-type="fig"}). This stratification indicated that patients with low expression of HOTAIR + TFs, at the end of the studies, had a 93.75% chance of survival compared to 41.705% for the high expression group. These patients with dramatically worse survival are likely to respond poorly to therapies. Collectively, we have identified several transcription factors that regulate the expression of *HOTAIR* and enhance the capacity of *HOTAIR* to predict the therapeutic response of patients with ER+ breast tumours and survival in breast cancer patients with HER2-enriched tumours. Discussion ========== Transcriptional enhancers enable precise regulation of gene expression, are frequently disrupted in cancer, and are emerging as the next generation of prognostic biomarkers for a range of human diseases. This study describes the long-range regulatory elements and factors controlling the transcription of *HOTAIR* in breast cancer. Further, we show that transcription factors correlated with *HOTAIR* expression in breast tumours are associated with poor prognosis. In this paper, we have identified several novel transcriptional control elements regulating the expression of the lncRNA *HOTAIR*. These include an alternate promoter situated approximately 8 kb upstream of the previously described *HOTAIR* promoter, and a distal enhancer (HDE) located approximately 150 kb downstream of the *HOTAIR* gene. These elements are distinct from the core promoter and CpG elements described previously ([@ddw177-B29],[@ddw177-B34]). We also show that the *HOTAIR* enhancer regulates *HOTAIR* expression by DNA looping and that this enhancer can augment the transcriptional activity of both the canonical and alternative promoter elements. Importantly we demonstrate that HOTAIR and its regulators, as identified through mapping of the HDE, combine to significantly stratify survival of breast cancer patients, supportive of previous work also demonstrating that enhancer activity and transcription factor binding align closely with clinical outcome ([@ddw177-B35]). It would be worth exploring this aspect for the HDE, by utilizing ChIP-Seq, DNA accessibility assay or 3C data from tumours samples, it may be possible to stratify clinical response through activity of this enhancer. Several studies have highlighted the promiscuous nature of promoter-enhancer interactions ([@ddw177-B36],[@ddw177-B37]). To further the research presented in this study, it would be worth exploring the interaction landscape of the HOXC locus through 5C, a technique that looks at all interactions within a large genomic region ([@ddw177-B38]). In addition to these experiments, mutagenesis of the HDE using a CRISPR/Cas9 approach would be valuable for further establishing the role of this enhancer ([@ddw177-B39],[@ddw177-B40]). Such studies may shed more light on the interaction of the HOXC distal enhancer, the *HOTAIR* promoter and potentially other targets of this enhancer. Reporter mutagenesis, ChIP-seq and siRNA experiments show that *HOTAIR* transcription is regulated by a number of breast cancer-associated transcription factors, including FOXA1 and FOXM1. FOXA1 is a crucial pioneer factor for ERα binding ([@ddw177-B41]). FOXM1 has a wide variety of functions including regulating G2/M phase transition, mammary gland luminal progenitor cell maintenance, co-binding with ERα in cell lines, and is strongly implicated in the progression of multiple cancer types including breast cancer ([@ddw177-B42]). *FOXM1* expression has previously been linked with ERα activity and resistance to endocrine therapies, where elevated expression is correlated with poor prognosis ([@ddw177-B44], [@ddw177-B45]). Whilst the tumour expression data for *HOTAIR*, *ESR1* and *FOXM1* was consistent with *in vitro* results, this was not the case for *HOTAIR* and *FOXA1*. The latter may represent differences between the molecular environment in cell lines and tumours. The data presented here support an association between forkhead box proteins and *HOTAIR* expression, likely through direct transcriptional regulation. Re-analyses of published ChIP-Seq data and oestrogen treatment *in vitro* demonstrate that oestrogen represses *HOTAIR* transcription. Our observations are consistent with recent finding by Xue *et al*. ([@ddw177-B46]) indicating a direct association between ER and *HOTAIR* repression following oestrogen treatment in MCF7 cells. In support of our findings, there is a strong inverse correlation between *HOTAIR* and *ESR1* expression in breast tumours, arguing against the notion of oestrogen enhancing *HOTAIR* expression. Our findings, and those of others ([@ddw177-B46]), conflict with a previous report showing induction of *HOTAIR* expression by oestrogen ([@ddw177-B31]). This discrepancy may reflect differences in experimental design, including time of oestrogen treatment and the method of transcript detection, or the use of a different sub-line of MCF7 cells. We hypothesized that the association between oestrogen and *HOTAIR* may play a role in endocrine resistance. To explore this, we used the well-documented tamoxifen resistant, fulvestrant resistant and oestrogen independent sub-lines of MCF7 cells ([@ddw177-B47]). *HOTAIR* expression increased in both the oestrogen-deprived MCF7X, and tamoxifen-resistant TAMR cells, concordant with a repressive role of oestrogen/ER signalling. Importantly, our previous work (Zhuang *et al*. 2014 ([@ddw177-B52])) also demonstrated that *HOTAIR* expression was significantly increased in a cell line model of resistance to induced cell death via TNF, which displays loss of ER and altered oestrogen signalling. These findings may be of clinical significance as they suggest therapies blocking oestrogen signalling may induce *HOTAIR*, which could in turn promotes breast cancer progression. Endocrine therapy (e.g. tamoxifen, aromatase-inhibitors) is a standard of care for patients with HR+ breast tumours, but deciding which patients should be managed more aggressively by the addition of chemotherapy can be difficult ([@ddw177-B53]). These patients are therefore a primary focus of prognostic and predictive biomarker development ([@ddw177-B54]). It is highly significant that *FOXM1* and *HOTAIR* co-expression significantly enhances the prognostic power of either factor individually in predicting relapse-fee survival in patients with ER+ and ER+/N+ breast cancer. There are several pathways in which *HOTAIR* and FOXM1 are likely to collaborate in tumourigenesis, explaining their prognostic power. Although we have shown that FOXM1 regulates the expression of *HOTAIR* through the HDE, we acknowledge that their link in tumourigenesis may be through other means, such as a common functional pathway. FOXM1 is involved in cell cycle regulation ([@ddw177-B55],[@ddw177-B56]) and the cyclin-dependent kinases CDK1 and CDK2 are known to phosphorylate EZH2 during S and G2 phases of the cell cycle ([@ddw177-B57]), actively promoting EZH2-*HOTAIR* interactions ([@ddw177-B57]). *HOTAIR* over-expression has also been implicated in cell cycle progression in pancreatic cancer ([@ddw177-B27]). It is possible that FOXM1 and *HOTAIR* corroborate to regulate the aspects of the cell cycle, further work should aim to understand this, and if this functionality explains the prognostic potential of FOXM1 and *HOTAIR* in breast cancer. FOXM1 has been implicated in chromosome instability (CIN), a process associated with aneuploidy, aggressive tumours and poor response to therapy ([@ddw177-B58]). CIN, together with ER signalling, has recently been described as a powerful predictor of survival in ER+ patients ([@ddw177-B61]). It is tempting to speculate that *HOTAIR* may play a role in the relationship between CIN and ER signalling. Reduced ER signalling (e.g. reduced oestrogen) may drive the expression of *HOTAIR* (via FOXA1 or FOXM1) possibly in association with increased expression of the CIN genes. This could explain the poor prognosis of subsets of ER+ tumours that appear genomically unstable and have reduced response to endocrine therapy (tamoxifen) alone or in combination with chemotherapy. It also raises the possibility that patients with ER+/N+ tumours who are resistant to endocrine therapy may benefit from anti-aneuploidy drugs that target molecules in the CIN module, such as TTK, AURK and PLK1 ([@ddw177-B62]). Indeed, we have previously shown that oestrogen-deprived or deficient cells are more sensitive to CIN inhibitors (Stone *et al*. 2013 ([@ddw177-B50]); Al-Ejeh *et al*. 2014 ([@ddw177-B61])). We demonstrate that the expression of *HOTAIR* in concert with its transcriptional regulators, are informative in predicting outcome in HER2 breast cancer. Our findings support an earlier study indicating that *HOTAIR* is most highly expressed in the HER2-enriched intrinsic molecular subtype of breast cancer ([@ddw177-B63]). The primary treatment for a patient with a HER2+ tumour is chemotherapy and Herceptin, a targeted monoclonal antibody to the HER2 receptor ([@ddw177-B64],[@ddw177-B65]). Herceptin resistance is a significant clinical problem, resulting in lower RFS, and leading to poor disease outcome ([@ddw177-B66]). The combination of *HOTAIR* and its long-range transcriptional regulators can better predict outcome in patients with high-grade HER2-enriched tumours. Use of these biomarkers, could immediately facilitate better targeting of expensive Herceptin treatment and improve patient surveillance. Longer term, inhibitors of this functional network could be investigated as targeted therapies. The results presented describe a novel regulatory pathway controlling the expression of the lncRNA *HOTAIR* and show that the associated elements and factors can illuminate the mechanisms underlying the role of *HOTAIR* in breast cancer. This information can be used to identify novel biomarker combinations to better predict response to therapy and survival of breast cancer patients. Continuation of this work should aim at understanding the molecular mechanisms that explain the prognostication potential of *HOTAIR* and its regulatory elements, with particular relevance to endocrine therapy resistant tumours that have aberrant ER signalling and the relationship this may have to chromosome instability. Materials and Methods ===================== Cell culture ------------ Cell lines; MCF7, T47D, ZR-751, MDA-MB-453 and MDA-MB-468; were cultured as per recommendations from ATCC. MCF7 cells were a kind gift from Chris Ormandy at the Garvan Institute of Medical Research (NSW, Australia) whilst T47D, ZR-751, MDA-MB-453 and MDA-MB-468 cells were sourced from ATCC. Estradiol (E2) treatment experiments followed a time course of 10 or 1 nM E2 across indicated time points within the text. Positive controls for the RNA-Seq and qRT-PCR experiments are shown in Figure S6. MCF7 cells, from which the endocrine resistant sublines were derived, were originally obtained from AstraZeneca. MCF7-derived acquired endocrine resistant sub-line RNA was sourced through A. Stone and J. M. W. Gee see author list for details. Tamoxifen-resistant (TAMR), fulvestrant-resistant (FASR), and oestrogen-deprived (MCF7X) derivatives were cultured as previously described ([@ddw177-B47],[@ddw177-B48],[@ddw177-B51]). All cell lines were authenticated by short-tandem repeat (STR) profiling (Cell Bank, Australia) and cultured for less than 6 months after authentication. Chromosome conformation capture (3C) ------------------------------------ Chromosome conformation capture was performed as previously described by in Tan-Wong *et al*. 2008 ([@ddw177-B67]). Briefly, cells were grown to 60--80% confluence and fixed with 1% formaldehyde. Libraries were generated for each cell line using *Hind*III and or *Bgl*II with control libraries undigested and unligated, representing native DNA without chromosome conformation. Any 3C-qPCR products were excluded if they amplified within the control libraries. GAPDH primers (amplified fragment contains no cut sites for these restriction enzymes) were used to determine the digestion and ligation efficiency for each library by comparing 3C-qPCR values to primers that amplifed a fragment containing a *Hind*III or *Bgl*II cut sites (Table S6). For each 3C-qPCR, primers were designed between 100--250 bp up or downstream of each *Hind*III cut site with the primer across the putative *HOTAIR* distal enhancer used as bait in each 3C-qPCR ([Supplementary Material, Table S6](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1)). *Bgl*II primers were sourced from Ferraiuolo *et al*. ([@ddw177-B70]). 3C-qPCR conditions 50 °C, 2 min, 95 °C 10 min, (95 °C 15 sec and 60 °C 1 min) for 45 cycles. With a final melt starting at 50 °C 90 sec rising to 99 °C in 1 °C increments and 5 sec at each step. Each 3C-qPCR was performed on a pool of at least three independent libraries for each cell line and the qPCR done in duplicate. BAC controls were constructed following the protocol from ([@ddw177-B69]) and primer pairs assayed for appropriate efficiency. Cloning and luciferase reporter assays -------------------------------------- *HOTAIR* enhancers and promoters were cloned into the luciferase reporter plasmid pGL3-Basic; see [Supplementary Material, Table S7](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1) for primers. The canonical *HOTAIR* promoter used in this study incorporated the previously described promoter region and E-box with additional bases either side ([@ddw177-B29]). MCF7, T47D and MDA-MB-453 cells were transfected in antibiotic-free media with 700 ng of modified pGL3 promoter less reporter plasmids, 20 ng of Renilla and with 1μL of Lipofectamine® 2000 (Life Technologies, Grand Island, NY, USA). 48hours post transfection luciferase readings were measured using a DTX-880 luminometer and Dual-Glo® luciferase reporter kit (Promega, Madison, WI, USA), following the manufacturer's recommended protocol. For oestrogen treatment and luciferase reporter levels of *HOTAIR* promoters and the HDE we followed the previously published method by Tan-Wong *et al*. ([@ddw177-B67]). Gene expression analysis ------------------------ RNA for gene expression analysis was extracted using TRIzol® reagent (Life Technologies) and phenol-chloroform purified. RNA was then DNase I (Ambion, ThermoFisher) treated and cDNA generated using SuperScript® III (Invitrogen, ThermoFisher) using random hexamers supplemented with RNaseOUT™ (Invitrogen, ThermoFisher. Gene expression was assayed with TaqMan® probes (Life Technologies, ThermoFisher) following the manufacturer's recommended protocol, probe list in [Supplementary Material, Table S8](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1). RNA-Seq in MCF7 cells was performed as previously described by K. Nephew (see author list) ([@ddw177-B71]). Analysis of *HOTAIR* and *FOXM1* expression in response to MCF7 cells treated with thiostreptin was sourced from GSE40767 ([@ddw177-B44]). The probe sequence for *HOTAIR* expression on the array was as follows: TACACGCCTCTCCAAG ACAC AGTGGCACCGCTTTT CTAACTGGC AGCACA and FOXM1: GCCA CCTCCCCGTGTT TCCAAGTCAGCTTTC CTGCAAGAA GAA A T CCT GG. HOTAIR isoforms were sourced from public repositories for the RefSeq ([@ddw177-B72]) and Broad ([@ddw177-B73]) data sets. For analysis of *HOTAIR* and *FOXM1* in the PAM50 subtypes, TCGA RSEM data was log2 transformed and mean-center normalized for all tumours, tumuor numbers as follows Basal-like (140, Basal), HER2-enriched (67, HER2), Luminal A (420, LumA) Luminal B (194, LumB) and Normal-like (24, N-Like). RNAi ---- Short-interfering RNA-mediated knockdown was performed using Flexitube siRNA (Qiagen). SiRNAs utilized in this study were FOXA1; SI04311888 and SI04217038 and FOXM1; SI04261831 and SI04166309 which were compared to the AllStars Negative Control (Qiagen). MCF7 cells were transfected using 3L of RNAi-Max (Life Technologies, ThermoFisher) and with a final concentration of 10nM for each siRNA. RNA was extracted and conversion into cDNA and subsequent qRT-PCR for gene expression as described above. Bioinformatic analysis of publically available data --------------------------------------------------- All data was sourced from public repositories and analyzed through the IGV browser and the human genome assembly Hg18 and/or Hg19, as indicated. Publically available data were sourced from the Gene Expression Omnibus (GEO). Accession codes for each dataset are listed in [Supplementary Material, Table S9](http://hmg.oxfordjournals.org/lookup/suppl/doi:10.1093/hmg/ddw177/-/DC1). ChIP-Seq data that was not available as processed mapped profiles was done so utilizing standard mapping software. Briefly, data were downloaded in .sra format and converted to .fastq using fastq-dump and subsequently mapped with Bowtie where any multi-mapping reads were disregarded. Output .sam files were then converted to .bam files, using SamTools, .bed files using BedTools and finally converted to .bedGraphs that were tiled by IGV-Tools for IGV browser analysis ([@ddw177-B74]). The Cancer Genome Atlas (TCGA) ([@ddw177-B75]) was accessed and analyzed through the University of California Santa Cruz (UCSC) Cancer Browser ([@ddw177-B75],[@ddw177-B76]). HOTAIR + TFs were determined based on the average expression of *HOTAIR*, *FOXM1*, *FOXA1*, *FOS* together with the inverted average expression of *ESR1*. HOTAIR + FOXM1 is the combined average expression for these two genes. Kaplan-meier analysis --------------------- The online tool, Kaplan-Meier Plotter (<http://kmplot.com/analysis/>) was used to produce survival curves based on gene expression with a maximum follow-up time of 10-years post diagnosis ([@ddw177-B77]). High- and low-expression cutoffs were determined by the 'Autoselect best cutoff,' feature that determines the most significant stratification based on median, tertile or quartile splits. Kaplan-Meier Plotter uses logrank hazard's ratios with corresponding P-values to determine significance. Probe IDs for each gene utilized in this analysis: HOTAIR (239153_at), ESR1 (205225_at), FOXA1 (204667_at), FOXM1 (202580_x\_at) and FOS (209189_at). Tumour clustering and correlations ---------------------------------- Normalized TCGA ([@ddw177-B78]) microarray expression data were used to cluster *HOTAIR* expression with *ERα* and *ERα* cofactors. The heat-map was produced using hierarchical average-linkage Manhattan-based clustering performed in Multiple experiment Viewer (MeV ([@ddw177-B79])) and Pearson correlation coefficients calculated for each comparison. The dendrogram tree cut point is indicated in [Figure 5A](#ddw177-F5){ref-type="fig"}, which creates 5 clusters of tumours, clusters 1-4 were grouped together based on the negative correlation between *HOTAIR* and *ESR1* expression. Western blot analysis --------------------- Total protein from MCF7 cells was extracted using RIPA buffer with 15 μg total per well and analyzed using Novex™ NuPAGE® gels from Life Technologies. Antibodies for FOXA1 and FOXM1 were ab23738 (Abcam, Cambridge, MA, USA) and sc-502 (Santa Cruz, Dallas, TX, USA) respectively. Supplementary Material ====================== Supplementary Material is available at *HMG* online. ###### Supplementary Data We thank Jonathan Harris for the provision of DHT and David Miller for technical assistance and Kyle Upton for his critical review of the manuscript. *Conflict of Interest statement*. None declared. Funding ======= Research was funded by the National Breast Cancer Foundation (NBCF: 2007003445 and CG-12-07) of Australia, the Australian Research Council (ARC: DP0985025), the Cancer Council of Queensland (CCQ: 1026095) and The University of Queensland. DHD was supported by a Collaborative Program Grant from the, National Breast Cancer Foundation \[NBCF; CG-08-03\]. Institute, Integrative Cancer Biology Program U54CA1113001 (KPN). SLE, JDF, AMS and ED are supported by Fellowships from the NBCF \[ID\# ECF-10-05, ECF-12-04, ECF-12-12 and ECF-13-04 respectively\]. FA is supported by Future Fellowship from the Australian Research Council \[ID: FT130101417\]. JMWG is funding by a Scientific Fellowship from Breast Cancer Now. MJM and JAB are supported by an Australian Postgraduate Award (APA). AS was supported from NBCF program grant. SJC supported by NHMRC fellowship. Funding to pay the Open Access publication charges for this article was provided by The University of Queensland.

Introduction ============ A 21-year-old female with a past medical history of irritable bowel syndrome and Lyme disease presented to our office with right knee pain. She was found to have a 9 cm parosteal osteosarcoma of the proximal tibia (Figure [1](#F1){ref-type="fig"}). She underwent a radical resection and reconstruction with an 11 cm modular segmental tumor prosthesis (Link; Hamburg, Germany) along with a rotational medial gastrocnemius muscle flap. A negative-pressure wound care system was used and the patient was also placed in a posterior leg splint. The incisional wound vacuum was prophylactically used at index procedure to promote faster wound healing and aid it by taking tension off of the incision as well as the gastrocnemius muscle flap. There was no communication between the sponge and the patient\'s tumor prosthesis. Patient\'s immediate postoperative course was unremarkable. Thirteen days after surgery the patient\'s dressings were taken down and the incision was healing well. The patient was then placed in a hinged knee brace locked in extension for a total of 6 weeks after surgery to allow her extensor mechanism to heal. Flexion was gradually permitted starting at week 6. At 8 weeks post-operatively the patient had a well-healed surgical incision and radiographs demonstrated all implants were in excellent position (Figure [2](#F2){ref-type="fig"}). The patient was instructed to start more progressive range-of-motion exercises with physical therapy. At 11 weeks post-operatively, the patient presented with a small cutaneous abscess on the inferior aspect of her incision. The incision became erythematous and began to produce a purulent discharge (Figure [3](#F3){ref-type="fig"}). At this time, the patient gave a history about going to the nail salon to have a pedicure in the weeks prior. She underwent irrigation and debridement in the operating room. The infection was superficial and did not go deep down to the implant. Cultures were sent off during surgery for aerobic, anaerobic, acid fast bacilli (AFB), and fungal. In our orthopedic oncology practice, many patients are immunocompromised secondary to chemotherapy and so we routinely send specimens for aerobic, anaerobic, AFB, and fungal cultures when infection is suspected. Initial gram stain, aerobic, and anaerobic cultures were negative and the patient was discharged with a PICC line on empiric daptomycin and piperacillin/tazobactam. Twenty-one days after the debridement procedure, the acid-fast bacillus cultures became positive for *M. fortuitum* (Table [1](#T1){ref-type="table"}). At this time the patient was readmitted to the hospital and underwent repeat irrigation and debridement of the surgical wound due to continued erythema and drainage to also include the deeper tumor prosthesis in an attempt to preserve the initial prosthesis. She was started on intravenous amikacin 750mg daily and imipenim/cilastin 500mg every 6 hours as well as oral ciprofloxacin 750mg twice daily. The patient was seen 4 and 8 weeks after the second debridement procedure with minimal improvement in her inflammatory markers and clinical picture. The patient was subsequently admitted to the hospital and, approximately 6 months after the index procedure, underwent removal of the tumor megaprosthesis and placement of a static antibiotic spacer; the tibial and femoral stems were left in place and the wound and surrounding tissues were copiously scrubbed with chlorhexidine, hydrogen peroxide, and betadine. Post-operatively she was started on intravenous imipenim/cilastin 500mg every 6 hours, oral ciprofloxacin 750mg twice daily, and oral minocycline 100mg twice daily. 8 weeks later, after the patient had completed her antibiotics and her laboratory values had normalized, she underwent exchange of the antibiotic spacer for a new megaprosthesis. Post-operatively, she was re-started on amikacin 750mg IM three days a week, intravenous imipenim/cilastin 500mg every 6 hours, and oral ciprofloxacin 750mg twice daily until all of her intraoperative cultures were finalized. All operating room cultures at the time of exchange, including aerobic, anaerobic, fungal, and AFB were negative. On subsequent post-operative visits the patient demonstrated excellent healing and after completing her initial course of post-operative antibiotics she was transitioned to oral doxycycline 100mg twice daily for one year given her history of multiple surgeries, cancer diagnosis, and her infection. At 24-months post-revision surgery she is doing well with the implants in good position with no evidence of infection. Discussion ========== NTM are an atypical type of mycobacteria that do not cause tuberculosis or leprosy. M. Fortuitum is a member of the rapidly growing Reunion Group IV NTM and is most frequently acquired from water, soil, and nosocomial sources. It is recognized as an opportunistic pathogen, mainly causing infection and morbidity in immunocompromised patients. However, there have also been cases reported of a M. Fortuitum infection in immunocompetent hosts. Other authors have previously presented prosthetic joint infections (PJI) in primary joint arthroplasty patients by NTM, including *M. fortuitum* [@B1], [@B2], [@B3], [@B4], [@B5]*.* In many of the cases reported in the literature on M. Fortuitum infections, the standard of treatment was surgical removal of the entire prosthesis, placement of an antibiotic spacer with concomitant intravenous antibiotics, and then replantation after the infection had cleared [@B1], [@B2], [@B3], [@B4]. However, Eid et al. presented eight patients with nine infections due to rapidly-growing Mycobacteria (seven knees, one hip, one elbow), three of which were *M. fortuitum* [@B5]. Median onset of symptoms was 312 weeks after implantation (range 1-720). Two of these patients, both total knee arthroplasty patients, were treated with debridement and implant retention. The remaining patients all had their implants removed. The two patients with retained prosthesis received treatment with moxifloxacin, trimethoprim-sulfamethoxazole, and azithromycin or levofloxacin and trimethoprim-sulfamethoxazole for long-term suppression of *M. fortuitum* infection. Both were asymptomatic at 24 and 189 weeks after debridement, respectively. In our study, after one superficial and one deep wound debridement with an effort to preserve the initial prosthesis, the infection unfortunately could not be eradicated. The patient was treated with removal of the prosthesis with cemented stem preservation, and insertion of an antibiotic spacer. In addition, the patient was administered IV and oral antibiotics post-operatively. She is currently asymptomatic at 24 months post-operatively. Other authors have looked at factors that may indicate the need to remove prostheses in mycobacterial infections. Harwin et al. found that patients with later onset of symptoms (\> 6-8 weeks), presence of a draining sinus, concomitant bacterial infection, osteolysis or prosthesis instability indicate the need to remove the prosthesis [@B6]. In our patient, the decision for stem retention was made because the stems were firmly fixed and removal could result in significant bone loss, thus compromising an ultimate reconstruction. Thus, retention of certain components with appropriate antibiotic treatment may be an option, particularly in major reconstructive cases. The timing of *M. fortuitum* infections has also been described in the literature. Jitmuang et al. presented a case series of 16 patients with NTM infections, nine of which were *M. fortuitum* [@B7]*.* In their study, they found that of the 11 patients with NTM mycobacterial infection, 9 of the infections occurred within 90 days after the index procedure. This is in contrast with previous literature, which showed more delayed onset of infection [@B5]. In our study, the patient showed the first signs of infection approximately 11 weeks post-operatively. It is also imperative to emphasize that these bacteria may not be readily apparent on gram stains and may require several weeks to grow from AFB cultures. Thus, it is vital to consider an agent such as *M. fortuitum* during any phase in the post-operative period as a potential cause of infection, especially in immunocompromised individuals, patients who have visited a nail salon or been exposed to lake water. The fungal cultures should be followed up for 4-6 weeks after surgery. One interesting aspect of the case was the patient reporting her recent pedicure at a nail salon after her index procedure. It has previously been reported that there is an association between mycobacterial infections and pedicures, particularly the footbaths performed at many salons [@B8], [@B9], [@B10]. Vugia et al. reported an outbreak of mycobacteria in nail salons from 5 California counties. The study team swabbed 30 footbaths in 18 nails salons and found cultures of mycobacteria in 29 (97%) [@B10]. Although it is difficult to draw any direct conclusion from these results, there is still a significant risk factor associated with visiting nail salons, especially for post-operative patients with incisions that may not have fully healed. In addition, while these infections are rare, it is crucial to understand any risk factors for a post-operative patient, regardless of how trivial it may seem. This case was unusual because our patient was immunocompetent, and thus should be less prone to these types of infections. Therefore, when counseling any patient on post-operative care or working up a potential infection, there is a need to know about pedicures as a potential source of infection. Patients undergoing joint replacement surgery should be counseled about the need to refrain from pedicures and manicures, especially in the first 90 days after surgery when the patient is most vulnerable to developing an infection. Surgeons should also be mindful that all post-operative patients are at risk for these rare types of infection, thus every specimen received for pyogenic culture must be processed for AFB examination and culture. Any delay in proper treatment makes it difficult for the surgeon to eradicate this type of bacteria and concomitantly puts the patient at a much higher risk of amputation. Conclusion ========== To our knowledge this is the first case presented of a *M. fortuitum* infection developing post radical resection and reconstruction with a proximal tibia megaprosthesis. This case emphasizes the importance of considering mycobacterium as a source of infection in these patients. Furthermore, it highlights the importance of considering all patient risk factors when determining the appropriate procedure to eradicate the infection. Finally, surgeons may need to consider the risk of post-operative patients receiving a pedicure at a nail salon, however trivial it may seem. Author Contributions ==================== All authors contributed to the conception and design of the study. TN, TH, JP, and JW helped write the manuscript. CM, DG, and KZ helped gather information about the patient\'s case from the hospital database. JW performed all surgeries outlined in this paper. {#F1} {#F2} {#F3} ###### Anti-Mycobacterial Susceptibility Testing und Results  [^1]: Competing Interests: The authors have declared that no competing interest exists.

INTRODUCTION ============ From the perspective of parasitology, increases both in population density and the filth might have led to the soil contamination by parasite eggs in permanent settlements \[[@b1-kjp-54-5-555]--[@b4-kjp-54-5-555]\]. Indeed, parasite infections must have been common among the concentrated populations of more sedentary agricultural societies \[[@b5-kjp-54-5-555]\]. However, corroborative reports to date have been surprisingly rare due to difficulties in historical or scientific studies on this subject. Paleoparasitology, which entails the analysis of ancient samples by modern parasitological techniques, is thus a powerful investigative tool in the study of parasite infections in human history. It is the study of parasites found in ancient archaeological remains and on that basis, the tracing of the history of host-parasite relationships \[[@b6-kjp-54-5-555]--[@b9-kjp-54-5-555]\]. By the morphological or molecular techniques on the samples obtained at various archaeological sites, scientific clues for consolidating tentative concepts about parasite infections in the history of mankind could have been obtained successfully. Fortunately, for the past several years, parasitologists have also been provided with invaluable opportunities to conduct examinations on the samples from ancient archaeological ruins of South Korea. By a series of scientific analyses, the characteristics of parasite infections in ancient major cities began to be revealed for the first time ever in this country. Considering that the scientific knowledge is very rarely available about ancient parasite infections in highly populated areas of the past, the present review can be significant to concerned parasitologists who have tried to obtain information about the parasitic infection patterns prevailing among ancient city-dwellers in so much detail. Actually, we found several parasitic eggs from the strata soil in highly populated ancient cities of Korea and analyzed the results according to population density, weather (flood), and environmental factors. EXAMINATION OF STRATA SOIL SAMPLES TO DETECT PARASITE EGGS ========================================================== Paleoparasitological studies on pre-modern Korean populations have utilized different forms of archaeological evidence, most notably coprolites obtained from Joseon mummies. By comparison of data thus obtained with 20th century National Survey statistics, clearer pictures of parasitic infections in Joseon populations could have been obtained successfully \[[@b9-kjp-54-5-555]\]. In order to draw conclusions on parasitism patterns prevailing in historical settlements; however, parasitological examinations have to proceed in a different way. Actually, it is well recognized that soil contamination of parasite eggs is a preliminary factor central to a coherent understanding of soil-transmitted parasitism in history \[[@b10-kjp-54-5-555]--[@b14-kjp-54-5-555]\]. This is clearly based on the concept that soil-transmitted parasite infections even in the 20th century developing countries are usually coincided by heavy parasite-egg contamination of soils \[[@b15-kjp-54-5-555]--[@b18-kjp-54-5-555]\]. We therefore examined strata soil samples from archaeological sites to deduce the parasite infection status among ancient town- or city-dwellers ([Fig. 1A](#f1-kjp-54-5-555){ref-type="fig"}). The chronology of each geological stratum could be differentiated by archaeologists' date estimation of the accompanying cultural remains or carbon dating results of the stratum. For a particular settlement, together with archaeological estimation of their dates, examinations of strata soil samples can reveal the patterns of parasite-egg soil contamination prevailing there for a specific historical period. Whenever we find ancient parasite eggs in significant quantity of soil samples from a certain stratum ([Fig. 1B--D](#f1-kjp-54-5-555){ref-type="fig"}), the implication is that many city-dwellers there were highly infected by parasites at that particular moment in history. The technique by which soil-sediment samples are processed is by now well established \[[@b19-kjp-54-5-555]--[@b21-kjp-54-5-555]\]. Briefly, soil samples are re-hydrated in 0.5% trisodium phosphate solution for 1 week. They are then filtered through multiple-layered gauze and precipitated for 1 additional day. Once the upper turbid layer is discarded, the precipitates are again dissolved in 10% neutral buffered formalin and finally, dropped onto slides for light-microscopic examinations. As for any parasite eggs subsequently found, their sizes were measured and their numbers per gram (EPG) of soil and per slide (EPS) were estimated. On the basis of the data thus obtained, conjectures can be made on the parasitic infection situation prevailing among a population of a given settlement. POPULATION DENSITY AND PARASITISM ================================= The first insight we obtained from paleoparasitological study on Korean archaeological sites is the close relationship of ancient parasitism to population density. Actually, in our studies of the past several years, we found that the higher the population density within a given area in history, the heavier the contamination of parasitic eggs in strata soil samples by microscopic examinations ([Tables 1](#t1-kjp-54-5-555){ref-type="table"}[](#t2-kjp-54-5-555){ref-type="table"}--[3](#t3-kjp-54-5-555){ref-type="table"}). During our full-fledged survey of paleoparasitology in archaeological sites of Korean peninsula (2010--2012; Project of National Research Institute of Cultural Heritage), numerous ancient parasite eggs were discovered in the samples from ancient major-capital city ruins such as Old Seoul City (Joseon), Buyeo (Baikje), and Gyeongju (Silla). On the other hand, we noted that the presence of ancient parasite eggs was reported very rarely from the samples of rural areas at the same time period ([Tables 1](#t1-kjp-54-5-555){ref-type="table"}[](#t2-kjp-54-5-555){ref-type="table"}--[3](#t3-kjp-54-5-555){ref-type="table"}). The pattern is much clearly seen in a series of our detailed studies on the archaeological ruins of Buyeo city, the capital of the ancient Baekje kingdom (18 BCE to 660 CE), and it's nearby rural areas (Gongju, Yeongi, and Cheonan). Although Buyeo city is now only a small town in the southwestern area of the Korean peninsula, during the period of 538 to 660 CE, it was a big political and cultural center of the Baekje Kingdom, surrounded by vast areas of rural provinces. By the excavations of Buyeo's ancient districts, still buried several meters underground, we got ancient strata soil samples, from which the information on parasite infection patterns among the Buyeo people of Baekje period could be obtained \[[@b22-kjp-54-5-555]\]. Briefly, in microscopic examinations on the Buyeo soil sediment samples from multiple districts (Gatap-ri, Gua-ri, Ssangbuk-ri, and Seokmok-rk) representing Baekje period, we revealed a numbers of *Ascaris, Trichuris, Clonorchis*, and *Pygidiopsis summa* eggs. Meanwhile the strata soil samples from more outlying suburban or rural areas of Baekje kingdom (Gongju, Yeongi, Cheonan) included very few ancient parasite eggs \[[@b22-kjp-54-5-555]\] ([Table 1](#t1-kjp-54-5-555){ref-type="table"}). This suggests that parasite infection prevalence might have been different in terms of the areas between the ancient city and its surrounding provincial areas. Parasite-egg contamination patterns during ancient Baekje period might have been closely correlated with population densities at that time because the higher the number of ancient parasite eggs within a given amount of strata soil samples, the heavier the parasitic infection among the people during the same time period \[[@b22-kjp-54-5-555]\]. Relative to Buyeo city, the soil contamination in the other ancient highly populated areas such as Old Seoul City (Hansung), the capital of the Joseon Dynasty (1392--1910 CE) \[[@b2-kjp-54-5-555],[@b21-kjp-54-5-555],[@b23-kjp-54-5-555]\], Gyeongju, the capital of ancient Silla kingdom, and a mountaintop fortress Hwawangsansung of Three Kingdom Period \[[@b24-kjp-54-5-555]\], was also found to be similarly heavy in our investigations ([Tables 2](#t2-kjp-54-5-555){ref-type="table"},[3](#t3-kjp-54-5-555){ref-type="table"}). In the study of the strata samples of Joseon Hansung city, we found a heavy contamination of ancient *Trichuris, Ascaris*, *Clonorchis*, and *Diphyllobothrium* eggs that were originated from samples of streambed, alley, alley-gutter, Royal palace wall, the Yukjo (main) street, and Royal Arsenal ruins, etc. ([Table 2](#t2-kjp-54-5-555){ref-type="table"}). As ancient parasite eggs were rarely discovered in the samples from the rural area ruins of Korean peninsula, the profuse discovery of ancient parasite eggs in Hansung city samples was very exceptional to us. Like Buyeo city, a higher number of ancient parasite eggs within Hansung strata soil could have been caused by a heavy parasitic infection among the people in the major city of Joseon period. Sampling at an ancient fortress Hwawangsansung built on the top of a mountain also showed interesting patterns to paleoparasitologists. Actually, in Korean history, there were many fortresses constructed on the mountaintop areas. As these ancient facilities were used as refugees in times of war or rebellion; or were established as political centers in the provinces, they also became highly populated areas at a certain period of ancient times, like major cities of the country \[[@b9-kjp-54-5-555],[@b14-kjp-54-5-555]\]. The presence of ancient eggs in fortress soil strata might therefore have been caused by the overpopulation within the area. In fact, our parasitological examinations clearly exhibited *Ascaris*, *Trichuris*, and Taenia eggs in the ancient soil strata samples from the ancient fortress Hwawangsansung area \[[@b24-kjp-54-5-555]\] ([Table 3](#t3-kjp-54-5-555){ref-type="table"}). Taken together, by paleoparasitological investigations, we managed to know the glimpses of the patterns about ancient parasitism in the highly populated areas. Briefly, the great number and ubiquity of parasite eggs in the strata soil samples from ancient major city ruins clearly suggested that the parasite infection must have been the major cause of people's diseases, especially among urban populations in history. Serious parasite egg contamination in ancient urban area soils might have been a common phenomenon shared by any of historical major cities worldwide, based on the paleoparasitological studies previously done in each country \[[@b25-kjp-54-5-555]--[@b29-kjp-54-5-555]\]. Of course, we must admit the limitation of this speculation because there are many possibilities concerning higher soil contamination of parasite eggs in the samples from ancient major city ruins. More studies on ancient samples are still needed for the hypothesis to be placed upon firm scientific basis. PLAUSIBLE MECHANISMS OF PARASITE INFECTION AMONG ANCIENT CITY-DWELLERS ====================================================================== Pioneering studies successfully revealed a serious parasite-egg contamination in the soils from such highly populous areas of the past \[[@b5-kjp-54-5-555]\]. Based upon it, it demonstrated the possibility that parasite infection prevalence might have been far higher among city-dwellers in history. Even so, however, its exact reinfection mechanism has remained elusive until recently. In fact, the scientific data based upon simple parasitic examination is not sufficient for answering to this question. Rather, we need much sophisticated way approaching this subject from quite a different angle. For instance, as to why the several hundred-year-old Joseon strata soils within Old Seoul City were so extensively contaminated by ancient parasite eggs, our research on Joseon documents efficiently raised the possibility that it might have been caused by vegetable growing practices around the highly populated city areas. In order to support the large population of the capital area in the 17th century, the Joseon people must have cultivated almost every corner of the city, maintaining a large-scale vegetable farm within or around Old Seoul City area \[[@b2-kjp-54-5-555]\]. To meet the inevitably urgent demand for enormous amounts of fertilizer for growing vegetables, the farmers of Old Seoul City resorted to the purchase of the human waste from so-called '*night-soil men*' \[[@b2-kjp-54-5-555],[@b30-kjp-54-5-555]\]. Night soil, in the centuries prior to industrialization, became the fertilizer of choice in East Asia, especially for vegetable growing, which require larger amount of fertilization than do other crops \[[@b2-kjp-54-5-555],[@b31-kjp-54-5-555]\]. This use of night-soil fertilizers for vegetable farming might have caused hitherto unexpected health problems for the city-dwellers. It is likely that once human waste had been introduced to nearby farmlands, a vicious cycle of parasite infections might have been repeated continuously in Old Seoul City \[[@b2-kjp-54-5-555]\]. The inevitable results of the operation of this vicious cycle, not only in Old Seoul City, but also in any of the pre-modern East Asian cities, could have been increasing rates of parasitic infections among the city-dwellers \[[@b2-kjp-54-5-555],[@b21-kjp-54-5-555],[@b23-kjp-54-5-555],[@b32-kjp-54-5-555],[@b33-kjp-54-5-555]\]. Although a review of extant Joseon records can be instructive to why Old Seoul City people was infected by soil-transmitted parasites so seriously, this hypothesis is not firmly based on scientific evidences. The limitation must be overcome by future studies of paleoparasitology on more ancient samples from archaeological sites in South Korea. THE FLOODING ============ Our parasitological examination of samples from Old Seoul City ruins uncovered another intriguing fact. That is, parasite eggs were not detected exclusively in specific places, public toilets, for example, that once would have included human waste. Rather, they were discovered much more indiscriminately in multiple districts, irrespective of those areas' or sites/original purposes \[[@b21-kjp-54-5-555],[@b23-kjp-54-5-555]\]. Particularly difficult to understand about this was how parasite eggs could have been so widespread in areas such as the open, high-traffic public space of the busiest main street of the capital city \[[@b21-kjp-54-5-555]\]. Considering that people would have abstained from defecation in areas of that type, the exact cause of the soil's contamination by parasite eggs remained unclear to us \[[@b2-kjp-54-5-555]\]. In this regard, the extant historical records also provided a possible answer once again. We found a number of contemporary descriptions of natural disasters occurring in Old Seoul City during the Joseon period: a repeated rainy-season flooding in most of the areas situated close to streams \[[@b2-kjp-54-5-555],[@b34-kjp-54-5-555]--[@b36-kjp-54-5-555]\]. Concerning the likely cause of that flooding, we noted that each hill in the city had been recklessly stripped bare to procure wood for the purposes of home heating during the harsh winter months \[[@b2-kjp-54-5-555],[@b37-kjp-54-5-555]\]. The land erosion resulting from the hillside deforestation had begun to choke the streams of the capital. The streambeds elevated markedly, making the water-flow sluggish and thereby heightening the risk of flood during the rainy season \[[@b21-kjp-54-5-555],[@b32-kjp-54-5-555]\]. The areas prone to flooding were widely scattered in Old Seoul City, according to the historical record. These were also the locations of the 2 most likely site-sources of human waste: many public toilets in the Old Seoul City area \[[@b2-kjp-54-5-555]\] and the vegetable fields and gardens in and around the city on which human waste was spread as the main ingredient in fertilizer. Therefore, when flooding did occur, it could have conveyed human wastes rapidly and indiscriminately throughout the capital districts \[[@b2-kjp-54-5-555]\] ([Fig. 1E, F](#f1-kjp-54-5-555){ref-type="fig"}). In fact, the widespread parasite-egg contamination of soil due to the failure in preventing or controlling flood was not exclusive to Old Seoul City. As another example, we have seen the archaeological ruins of Buyeo, the capital city (538--660 CE) of the ancient Baekje kingdom. Our parasitological analysis of the area revealed a very wide distribution of ancient parasite eggs, as represented in most of the samples there, irrespective of the original purposes of the given sites \[[@b22-kjp-54-5-555]\]. We thus conjecture that the ancient Buyeo city areas prone to flooding might also have been extensively distributed, just as they were in Old Seoul City. It thus evidently caused the spread of parasite-egg-contaminating silts throughout the Baekje kingdom's capital \[[@b22-kjp-54-5-555]\]. Corroborating evidence was found in the extant historical record of the Baekje period (Samguksagi), specifically an account of flooding that had occurred in Buyeo in its earliest days as the capital city \[[@b22-kjp-54-5-555]\] ([Fig. 2A, B](#f2-kjp-54-5-555){ref-type="fig"}). Overall, based on historical evidences, it seems that in ancient times, repeated flooding might have caused parasite-egg contamination of city environs much more frequently than scholars have expected. WATER FACILITIES ================ Much like modern-day city planners, ancient-city officialdom concerned themselves with the construction of facilities that could deliver basic life improvements or maintain the prosperity or security of the polity \[[@b38-kjp-54-5-555]\]. One of such facility was the moat, basically a water-filled ditch dug around castles, towns, or cities. In East Asia, as in all other civilizations, moats were constructed from earliest days, chiefly as lines of defense against enemy attack. Archaeologists in this region have made many references to moat ruins in the courses of their investigations of ancient archaeological sites. Moats' unquestioned utility aside, their stagnant water must have incurred serious sanitary problems. For instance, parasitological investigations of an ancient Japanese archaeological site in Yayoi village found the evidence of contamination in the moat by human waste that must have been washed into it by periodic heavy rains \[[@b26-kjp-54-5-555]\]. We reported similar findings made in the course of a paleoparasitological investigation of a palace dating to the Silla Dynasty (57 BCE to 935 CE). In the mud-soil strata of a moat there, we found a plenty of ancient *Trichuris* eggs \[[@b39-kjp-54-5-555]\]. Considering that soil-transmitted parasite eggs generally are shed in human or animal waste, the contents of toilets located somewhere nearby the royal palace might have been drained into the moat at certain times. Thus highly contaminated, the moat almost certainly would have become the source of various infectious diseases among the ancient Silla inhabitants \[[@b39-kjp-54-5-555]\]. Moats were not the only parasite-contaminated water facilities in ancient cities. In Korean history, there are multiple cases in which reservoirs were constructed at mountaintop fortresses to provide water to soldiers and civilians resisting invaders \[[@b24-kjp-54-5-555]\] ([Fig. 2C](#f2-kjp-54-5-555){ref-type="fig"}). In our parasitological study on the reservoir ruins of an ancient Silla fortress Hwawangsansung, we found many *Ascaris*, *Trichuris*, and *Taenia* eggs in soil samples \[[@b24-kjp-54-5-555]\] ([Fig. 2D--F](#f2-kjp-54-5-555){ref-type="fig"}). Concerning this abundance, we suspected that human or animal waste scattered in and around the fortress was swept via streams into the reservoir, particularly as it was situated at a relatively lower altitude \[[@b24-kjp-54-5-555]\]. It seems likely that this reservoir, ironically, might have been transformed into a serious sanitary menace to the same patriots it had been constructed to sustain. In a sense, the moats and reservoirs of the ancient cities or fortresses clearly showed the facilities' reverse side of convenience. CONCLUSION ========== With respect to parasite infections in the high-density populations of ancient towns and cities, we have managed to catch certain glimpses of the patterns prevalent therein via paleoparasitological investigations. Our studies on archaeological pre-modern city or fortress ruins in South Korea have revealed various types of ancient parasite eggs, and usually in large numbers, in representative soil-sediment samples. As to the issue of the great number and ubiquity of parasite eggs in samples from ancient city ruins, parasite infections indeed must have been one of the most common diseases among urban populations in history. Practices that are in hindsight questionable (the use of human-waste-based fertilizer, deforestation, and the unintended consequences of stagnant water, among others), as combined with periodic flooding during the rainy season, had had their effect on higher parasite infection prevalence in pre-modern city with high population density. A plausible mechanism by which the people of Old Seoul City and other major East Asian cities might have suffered from high prevalence of parasite infections could also be constructed in this review. This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) grant funded by the Ministry of Education, Science and Technology (MEST) (2016R1A2B4015669). {#f1-kjp-54-5-555} {#f2-kjp-54-5-555} ###### Paleoparasitological Examinations on the Baekje Soil Samples from Buyeo and Its Adjacent Rural Areas Area Estimated Date Results Eggs Per Gram (EPG) -------------------------------------------------------------------------- ---------------------------------------- --------------------- --------------------- Hongsung Bronze Negative ND Yesan Bronze Negative ND Buyeo/Songguk-ri Bronze/Baekje Negative ND Gongju/Tancheon Proto-Three Kingdom/Baekje Negative ND Yeongi/Nasung-ri 3--4 C CE (Proto-Three Kingdom/Baekje) Negative ND Yeongi/Seoksam-ri 3--4 C CE (Proto-Three Kingdom/Baekje) Negative ND Cheonan/Wiryesung 4--5 C CE (Baekje) Negative ND Buyeo/Gatap-ri[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} 5--6 C CE (Baekje) *Ascaris* *Trichuris* *Clonorchis* *Pygidiopsis summa* Buyeo/Gua-ri (319)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} 6--7 C CE (Baekje) *Ascaris* *Trichuris* *Clonorchis* Buyeo/Ssangbuk-ri (314--5)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Baekje *Ascaris* 8.6 *Trichuris* 7.1 Buyeo/Seokmok-ri (143--26)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Baekje *Ascaris* 10.0 *Trichuris* 5.0 Buyeo/Ssangbuk-ri (184--11)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Baekje *Ascaris* 0.3 *Trichuris* 0.4 Buyeo/Gwanbuk-ri (159--1)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Unified Silla Negative ND Buyeo/Dongnam-ri (321--3)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Baekje Negative ND Buyeo/Gugyo-ri[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Joseon *Ascaris* *Trichuris* Buyeo/Gugyo-ri (387--7)[\*](#tfn2-kjp-54-5-555){ref-type="table-fn"} Unified Silla/Goryeo Negative ND Shaded parts represent the archaeological sites of highly populous cities in history; Buyeo, the capital city of Baekje kingdom; ND, Not Determined. ###### Results of paleoparasitological examinations on the samples from archaeological sites of Seoul, Gyeonggi, and Gangwon Provinces Provinces Sites Estimated date Archaeological findings Results Eggs per gram ---------------------------------------------------------------------------------- --------------------------- ----------------------------------------------------------------- ------------------------- ------------- --------------- Seoul Pungnabtosung 4--5 C CE (Baekje) House Negative ND Hansung/Cheongjin-dong[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} 15--17 C CE (Joseon) House/Toilet? *Ascaris* ND *Trichuris* ND Hansung/Jongmyo-Gwangjang[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} Joseon period House/Streambed/Alley/Alley-Side Gutter *Ascaris* 21.3--161.6 *Trichuris* 10.0--135.0 Hansung/Buam-dong[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} (Baiksokdongcheon) Joseon House/Toilet? *Ascaris* ND *Trichuris* ND Hansung/Namsan-dong 1ga[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} Joseon-Modern Toilet? *Ascaris* ND *Trichuris* ND *Diphyllobothrium* ND Hansung/Royal Palace Wall[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} Joseon Soil strata underneath or in the vicinity of Joseon palace wall *Ascaris* 37.6--165.2 *Trichuris* 33.1--90.3 Hansung/Yukjo Street[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} Joseon Soil strata of the main street *Ascaris* 13.3 *Trichuris* 16.5--34.8 *Clonorchis* 17.9 Hansung/Royal Arsenal[a](#tfn4-kjp-54-5-555){ref-type="table-fn"} Joseon Soil samples from stream bed during Joseon period *Ascaris* 23.7--27.1 *Trichuris* 16.7--44.2 *Diphyllobothrium* 19.3 Gyeonggi Pyeongtaek/Segyo-dong Proto-Three Kingdom Period House Negative ND Paju/Hyeeumwon-ji Goryeo Toilet?/Drain Negative ND Gangwon Youngwol/Jeongan Fortress Three Kingdom Period Drain Negative ND Samcheok/Jukseoru Goryeo to Joseon House/Drain Negative ND ND, Not determined. Old Seoul city of Joseon dynasty (Hansung). ###### Results of paleoparasitological examinations on the samples from archaeological sites of Gyeongsang, Jeolla, and Jeju Provinces Provinces Sites Estimated date Archaeological information Results Eggs per gram ------------ ----------------------------------- ----------------------------- ---------------------------- --------------------- --------------- Gyeongsang Changnyeong/Hwawangsansung Silla Reservoir of Fortress *Ascaris* 3.2--7.0 *Trichuris* 2.5--5.6 *Taenia* 0.4--0.5 Gimhae/Bonghwang-dong Silla to Goryeo Shell midden *Ascaris* 0.8--5.0 *Trichuris* 0.8--11.6 *Dicrocoelid* 2.5 Gyeongju/Gyeongju National Museum Silla ND/Storing place/House Negative ND Gyeongju/Nodong-dong 12 Silla to Modern House *Ascaris* 0.4 *Trichuris* 0.4 Gyeongju/Weolseong Palace Silla Moat *Ascaris* 0.4 *Trichuris* 100.25 Pohang/Chogok Bronze-Three Kingdom Period Rice Paddy/Farm *Ascaris* 1.6 Gimhae/Daeseong-dong Gaya Tomb/Shell midden Negative ND Uljin/Jukbyeon Joseon Toilet? Negative ND Jeolla Yeosu/Songhyeon-maeul Bronze Tomb Negative ND Damyang/Hwabang-ri Bronze-Three Kingdom Period Rice Paddy Negative ND Gwangju/Songha-dong Bronze Farm Negative ND Naju/Heojin-ri Goryeo Kiln/House Negative ND Iksan/Songhak-dong Proto Three Kingdom Period ND Negative ND Jangheung/Hyangyang-ri Bronze/Joseon Tomb/Drain Negative (Tomb) ND *Trichuris* (Drain) 0.2 *Ascaris* (Drain) 0.2 Jangheung/Yonggang-ri Iron Age to Goryeo Tomb/House Negative ND Suncheon/Seongsan-ri 4--5C CE House Negative ND Naju/Samyeong-dong Goryeo Farm Negative ND Naju/Hoejin-ri Goryeo House Negative ND Gokseong/Shin-ri Iron Age Pit Negative ND Sunchang/Musu 4--5C CE House Negative ND Gochang/Hwangsan Three Kingdom Period Toilet? Negative ND Jeju Jeju/Yongdam-2dong 3C BCE to 4C CE House Negative ND Jeju/Samyang-2dong 7C BCE Toilet? Negative ND Jeju/Oedo 1--4C CE Pit Negative ND Jeju/Aewol 14--16C CE Cemetery Negative ND Seoguipo/Hwasun-ri 2,000 BP Pit Negative ND Jeju/Gosan-ri Neolithic House Negative ND

INTRODUCTION {#s1} ============ The survival rates of tumor patients are decreasing because of tumor cell resistance to anticancer chemotherapies \[[@R1]\]. This is due to acute alterations in the tissues surrounding the tumor and in the environment as a result of treatment with anticancer therapeutics, which facilitate the drug resistance of tumor cells. As many types of solid tumors adaptively respond to hypoxic (low O~2~ levels) stress, several survival pathways in hypoxic tumor cells are activated to carry out those essential biological processes that differ from the processes of normal cells \[[@R2]\]. Hypoxia-inducible factor 1 (HIF-1) is an oxygen-sensitive transcription factor \[[@R3]\] comprised of α and β subunits that form a heterodimer that binds to the hypoxia response element (HRE) region in DNA \[[@R3]--[@R5]\]. HIF-1α controls the expression of downstream target genes, including glucose transporter 1 (GLUT1), vascular endothelial growth factor (VEGF), insulin growth factor 2 (IGF2), transforming growth factor (TGF), erythropoietin (EPO) and so on. HIF-1α is correlated with vascular density \[[@R6], [@R7]\], tumor metastasis, angiogenesis, poor patient prognosis, as well as anticancer drug resistance \[[@R2]\]. Under normoxic conditions, HIF-1α hydroxylation of proline residues within its oxygen-dependent degradation (ODD) domain allows for HIF-1α association with von Hippel--Lindau (VHL) protein. Then, HIF-1α is degraded by 26S proteasome \[[@R8], [@R9]\]. Additionally, anticancer efficacy is reduced by HIF-1α overexpression or hypoxic conditions, which leads to the induction of drug resistance \[[@R10], [@R11]\]. Tumor hypoxia also leads to resistance to radiotherapy \[[@R12]\]. However, it is unknown whether HIF-1α can induce drug resistance of histone deacetylase (HDAC) inhibitor trichostatin A (TSA). TSA was first obtained from *Streptomyces hygroscopicus* as an antifungal antibiotic that is active against *Trichophyton* species and is used for the specific inhibition of HDACs including the class I and II, but not class III HDACs \[[@R13]\]. Highly acetylated histones are accumulated by TSA \[[@R14]\]. Reduced HDAC activity blocks the cell cycle, cell proliferation, and apoptosis \[[@R15]\]. TSA inhibits the hypoxia-induced accumulation of HIF-1α and VEGF under hypoxic conditions \[[@R16]--[@R19]\]. TSA also decreases HIF-1αlpha protein levels and VEGF expression in multiple cancer cells, including HeLa cells \[[@R20]--[@R23]\]. In contrast, changes in various intracellular molecules play a role in drug resistance. For example, overexpression of multidrug resistance-associated protein 8 (MRP8) \[[@R24]\], glucose-regulated protein 78 kDa (GRP78/BiP) \[[@R25]\], or p21WAF1 \[[@R26]\] leads to resistance to HDAC inhibitor-induced cancer cell apoptosis. However, it is unknown whether drug resistance can be induced by treatment with antitumor therapeutics, such as the HDAC inhibitor TSA, *via* alterations in HIF-1α acetylation under normoxic conditions. We determined whether HIF-1α acetylation by TSA affects tumor cell survival *via* nuclear translocation and binding to the HRE of the VEGF promoter. Our results suggest that the therapeutic effects of anticancer agents such as TSA may be hampered by HIF-1α acetylation under normoxic conditions. RESULTS {#s2} ======= TSA enhanced VEGF-HRE reporter activity and HIF-1α expression {#s2_1} ------------------------------------------------------------- To examine the effects of TSA on cell viability, HeLa cells were treated with TSA for 48 h. TSA treatment decreased cell viability at concentrations ranging from 300 nM to 1,000 nM, as determined by the MTT assay (Figure [1A](#F1){ref-type="fig"}). TSA also increased VEGF-HRE reporter activity (Figure [1B](#F1){ref-type="fig"} and [1C](#F1){ref-type="fig"}). The mRNA expression levels of HIF-1α (Figure [1D](#F1){ref-type="fig"} and [1E](#F1){ref-type="fig"}), total VEGF, and VEGF-A (Figure [1F](#F1){ref-type="fig"} and [1G](#F1){ref-type="fig"}) were enhanced by TSA treatment. No changes were detected in VEGF-B, VEGF-C, or VEGF-D (Figure [1F](#F1){ref-type="fig"}). TSA treatment elevated the protein levels of HIF-1α and VEGF (Figure [1H](#F1){ref-type="fig"}, top). HDAC inhibition by TSA was confirmed by an increase in acetylation at histones 3 and 4 (Figure [1H](#F1){ref-type="fig"}, bottom). Transfection with pEGFP-HIF-1α caused an increased number of TSA-treated cells expressing GFP-HIF-1α (Figure [1I](#F1){ref-type="fig"}, left and middle). HIF-1α expression was also increased by TSA treatment, which was detected by western blot analysis (Figure [1I](#F1){ref-type="fig"}, right). These data suggest that an increase in VEGF-HRE reporter activity by TSA might be associated with the binding of HIF-1α to the HRE following nuclear localization of HIF-1α under normoxic conditions. {#F1} VEGF-HRE reporter activity in various types of cells was augmented by TSA treatment {#s2_2} ----------------------------------------------------------------------------------- We assessed the effects of TSA on VEGF-HRE reporter activity in various types of cells. Our data showed that VEGF-HRE activity was increased by the incubation with TSA in HEK293T human embryonic kidney cells (Figure [2A](#F2){ref-type="fig"}), HCT116 human colorectal carcinoma cells (Figure [2B](#F2){ref-type="fig"}), MCF7 human breast adenocarcinoma cells (Figure [2C](#F2){ref-type="fig"}) and HepG2 human hepatocellular carcinoma cells (Figure [2D](#F2){ref-type="fig"}). TSA-induced increase in VEGF-HRE activity was the highest in HepG2 cells under our experimental condition. Then, we tested the expression of HIF-1α and VEGF in HepG2 cells. As shown in Figure [2E](#F2){ref-type="fig"}, HIF-1α and VEGF expression in HepG2 cells was increased by TSA treatment under normoxic conditions. It suggests that TSA might enhance VEGF-HRE reporter activity via nuclear localization of HIF-1α and the binding of HIF-1α to the HRE in various types of cells including HeLa cells. {#F2} TSA treatment increased the nuclear localization of HIF-1α {#s2_3} ---------------------------------------------------------- The effects of TSA-mediated HIF-1α expression on VEGF-HRE were examined by transfecting cells with pEGFP-C3-HIF-1α plasmid and determining the nuclear localization of HIF-1α. As shown in Figure [3A](#F3){ref-type="fig"}, the nuclear localization of GFP-HIF-1α was increased by TSA treatment. The relative percentage of nuclear localization of GFP was increased in the nucleus 1 h after TSA treatment (Figure [3B](#F3){ref-type="fig"}). No changes were detected in mock-transfected cells (Figure [3B](#F3){ref-type="fig"}). Additionally, when cells were co-transfected with pEGFP-C3-HIF-1α and the HRE reporter plasmid, VEGF-HRE reporter activity increased following TSA treatment (Figure [3C](#F3){ref-type="fig"}). We re-affirmed that TSA enhanced the nuclear localization of HIF-1α by the co-transfection with pMXs-RFP1 and pEGFP-C3-HIF-1α plasmid. HeLa cells were incubated with 300 nM TSA and GFP expression in cytoplasm and/or nucleus with RFP expression in cytoplasm was observed under fluorescence microscope. While GFP fluorescence was shifted from cytoplasm to nucleus, no translocation in RFP was detected (Figure [3D](#F3){ref-type="fig"}). The percentage of cells with GFP-HIF-1α was increased in nucleus but it was reduced in cytoplasm (Figure [3E](#F3){ref-type="fig"}). When cells were separated into cytoplasmic and nuclear fractions after TSA treatment, HIF-1α amount was increased in both cytoplasm and nuclear fractions (Figure [3F](#F3){ref-type="fig"}). It suggests that HIF-1α in cytosol should be primarily increased by TSA treatment and translocated into nucleus, which binds to the VEGF promoter to induce its mRNA transcription. ###### TSA treatment increased nuclear localization of HIF-1α HeLa cells were transfected with pEGFP-C3-HIF-1α plasmid and incubated with 300 nM TSA. GFP expression in cytoplasm and/or nucleus was observed under fluorescence microscope with 400× (**A**) or 1,000× (A, inset) magnification. Three blinded individuals counted the percentage of cells with the nuclear translocation of GFP-HIF-1α under fluorescence microscope (**B**). HeLa cells were co-transfected with VEGF-HRE-pSV40min and pEGFP-HIF-1α plasmids. Then, cells were incubated with 300 nM TSA and VEGF-HRE activity was measured by using luminometer (**C**). HeLa cells were co-transfected with pMXs-RFP1 and pEGFP-C3-HIF-1α plasmid and incubated with 300 nM TSA. GFP expression in cytoplasm and/or nucleus with RFP expression in cytoplasm was observed under fluorescence microscope (**D**). Three blinded individuals counted the percentage of cells with the nuclear translocation of GFP-HIF-1α (**E**). HeLa cells were incubated with 300 nM TSA and nuclear fraction was separated from cytoplasm. HIF-1α was detected with western blot analysis (E, top). Each band was quantified by using IamgeJ 1.34 and the results were represented as fold changes to control (E, bottom). Data are the representative of three experiments. Data in bar graph represent mean ± SD. ^\*^*p* \< 0.05; ^\*\*^*p* \< 0.01, significantly different from TSA-untreated control group transfected with mock or pEGFP-C3-HIF-1α. ^&&^*p* \< 0.01, significantly different from mock-transfectd group in the absence of TSA. ^\#\#^*p* \< 0.01, significantly different from mock-transfectd group in the presence of TSA.   TSA treatment increases HRE binding of HIF-1α {#s2_4} --------------------------------------------- We next examined the binding of HIF-1α to the VEGF promoter by transfecting pSG5-HIF-1α and performing a chromatin immunoprecipitation (ChIP) assay. TSA treatment enhanced HIF-1α binding to the HRE of hVEGF-A (Figure [4A](#F4){ref-type="fig"}). HIF-1α amount in cells transfected with pSG5-HIF-1α was confirmed by immunoprecipitation with anti-GST antibodies (Figure [4B](#F4){ref-type="fig"}). VEGF-HRE reporter activity was also increased by co-transfection with pSG5-HIF-1α (Figure [4C](#F4){ref-type="fig"}). These results suggest that TSA treatment induces the movement of HIF-1α into the nucleus for binding to the HRE of the VEGF promoter. {#F4} TSA treatment increases the acetylation of HIF-1α at lysine 674 {#s2_5} --------------------------------------------------------------- Because TSA is an HDAC inhibitor \[[@R13]\], we determined whether the acetylated lysine (K) is the mechanism underlying the nuclear localization of HIF-1α induced by TSA. When cells were transfected with pSG5-HIF-1α and treated with TSA, acetylation was increased in many proteins (Figure [5A](#F5){ref-type="fig"}). HIF-1α acetylation was detected by immunoprecipitating GST-HIF-1α and immunoblotting acetylated lysine (Figure [5B](#F5){ref-type="fig"}). We also determined whether endogenous HIF-1α might be localized into nucleus via lysine acetylation induced by TSA. As shown in Figure [5C](#F5){ref-type="fig"}, acetylation was increased in many proteins following TSA treatment. HIF-1α acetylation was detected by immunoprecipitating acetylated proteins and immunoblotting HIF-1α (Figure [5D](#F5){ref-type="fig"}). We also detected TSA-mediated acetylation of HIF-1α at K674 (Figure [5E](#F5){ref-type="fig"}). Results implicate that antitumor effect of TSA might be modified by the acetylation of HIF-1α at K674. {#F5} Drug resistance to TSA is dependent on HIF-1α acetylation at lysine 674 {#s2_6} ----------------------------------------------------------------------- According to TSA-mediated acetylation of HIF-1α at lysine (K) 674, we examined the changes in antitumor effect of TSA by the mutation at K674. To compare the effect of wildtype and mutant of HIF-1α on TSA-induced cancer cell death, cells were transfected with the wildtype or mutant (K674R) of HIF-1α (Figure [6A](#F6){ref-type="fig"}). The effect of HIF-1α on HRE was confirmed by the reduction of VEGF-HRE reporter activity in HIF-1α mutant exchanged lysine (K) 674 to arginine (K674R)-transfected group as compared to wildtype HIF-1α-transfected group (Figure [6B](#F6){ref-type="fig"}). Then, we assessed the changes in antitumor effect by the trypan blue exclusion assay. As shown in Figure [6C](#F6){ref-type="fig"}, the percentage of trypan blue-stained dead cells was reduced by overexpression of wildtype HIF-1α, which was rescued by the transfection with HIF-1α mutant (K674R). No significant changes were detected in the effect of vehicle on the changes in trypan blue-stained dead cell number. It suggests that the resistance to TSA could be associated with HIF-1α acetylation at K674. We propose the mechanism on TSA resistance in Figure [6D](#F6){ref-type="fig"}, which might be mediated by the increase in blood vessel formation and oxygen supply of VEGF to tumor cell survival. Additionaly, cell death inhibition by TSA-induced HIF-1α might be associated with various target gene expression such as TGF, EPO and so on to control tumor cell growth and survival. {#F6} DISCUSSION {#s3} ========== The efficacy of anticancer therapeutics is limited by tumor cells with resistance to chemotherapy, thus decreasing the survival rate of tumor patients \[[@R1]\]. Among the many significant factors that induce resistance, hypoxic (low O~2~ levels) stress surrounding many types of solid tumor tissue environments activates essential biological responses that differ from those in normal cells \[[@R2]\]. The anticancer efficacy of paclitaxel is reduced by HIF-1α overexpression or hypoxic conditions \[[@R10], [@R11]\]. The acetylation levels of target proteins may determine the responsiveness or sensitivity of different types of cancer cells to anticancer agents \[[@R19]\]. The expression of GRP78/BiP or MRP8 or p21WAF1 induces drug resistance to HDAC inhibitors, such as TSA \[[@R24]--[@R26]\], which accumulates highly acetylated histones \[[@R14]\]. In acute myeloid leukemia (AML) cells, HDAC inhibitors induce drug resistance *via* hyperacetylation of histone proteins in the promoter regions of MDR1, BCRP, and MRP8 \[[@R24]\], indicating that HIF-1α acetylation may also regulate the effects of anticancer therapeutics. However, little is known regarding TSA resistance induced by changes in the level of HIF-1α via acetylation. In the present study, we investigated whether TSA-mediated HIF-1α acetylation affects tumor cell survival *via* nuclear translocation and binding to the HRE of the VEGF promoter using HeLa cells for the continuation of our previous study \[[@R10], [@R27], [@R44]\]. Our data showed that TSA increased HRE activity of the VEGF promoter and the amount of HIF-1α protein under normoxic conditions (Figure [1](#F1){ref-type="fig"}). We confirmed the effect of TSA on VEGF promoter activation using other types of cells including HEK293T human embryonic kidney cells, HCT116 human colorectal carcinoma cells, MCF7 human breast adenocarcinoma cells and HepG2 human hepatocellular carcinomar cells (Figure [2](#F2){ref-type="fig"}). TSA treatment also increased the nuclear localization of HIF-1α and it\'s binding to the HRE of the VEGF promoter (Figures [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"}). TSA acetylated HIF-1α at K674 (Figure [5](#F5){ref-type="fig"}), which led to the activation of VEGF-HRE activity and the inhibition of tumor cell death (Figure [6](#F6){ref-type="fig"}). These data suggest that drug resistance to TSA may be dependent on HIF-1α acetylation at K674. While the percentage of GFP-HIF-1α-positive cells was increased at 3 and 6 h after TSA treatment (Figure [3B](#F3){ref-type="fig"}), VEGF-HRE activity was increased from 1 to 12 h (Figure [1C](#F1){ref-type="fig"}). VEGF-A expression by RT-PCR was increased from 0.5 to 2 h after TSA treatment (Figure [1F](#F1){ref-type="fig"}), which is not consistent with data increased at 2 h after TSA treatment by Q-PCR (Figure [1G](#F1){ref-type="fig"}). Then, we confirmed nuclear translocation of HIF-1α at common effective time, 2 h after TSA treatment by western blot analysis (Figure [3F](#F3){ref-type="fig"}). It suggests that nuclear translocation of HIF-1α might be gradually increased from 1 to 6h after TSA treatment and TSA-mediated HIF-1α could be also degraded at the same time under normoxic condition. TSA is a natural product that inhibits HDACs of Class I (HDAC1, 2, 3, and 8) and Class II (HDAC4, 5, 6, 7, 9, and 10) \[[@R14], [@R28], [@R29]\]. TSA inhibited proliferation of eight breast carcinoma cell lines with mean ± SD IC~50~ of 124.4 ± 120.4 nM (range, 26.4--308.1 nM) *in vitro*. TSA had also pronounced antitumor activity *in vivo* when administered to rats at a dose of 0.5 mg/kg (≒1,650 nM) by daily injection for 4 weeks compared with control animals. Furthermore, TSA did not cause any measurable toxicity in doses of up to 5 mg/kg (≒16,500 nM) \[[@R30]\]. Due to the direct contact of TSA to tumor cells *in vitro*, we tested the effect of TSA on HeLa cells from 100 nM to 1,000 nM for 48 h-incubation (Figure [1](#F1){ref-type="fig"}). So, we thought that these concentrations are physiologically or clinically reasonable to see whether tumor cell death might be modified by TSA treatment. In the beginning of our study, we tested the effect of TSA as one of anticancer agents on HeLa cell death. As shown in Figure [1A](#F1){ref-type="fig"}, when tumor cells were incubated with TSA for 48 h, we observed that tumor cell death was significantly increased from 300 nM which is the lowest concentration under our experimental condition. TSA-induced tumor cell death was time-dependently increased and little changes were detected by the incubation with 300 nM TSA for less than 12 h \[[@R27]\]. Then, we are interesting to see what is happened in tumor cell death by the treatment with 300 nM or less concentration of TSA for most experiments in our study. Anticancer efficacy is reduced by HIF-1α overexpression or hypoxic conditions, which can lead to drug resistance \[[@R10]--[@R12]\]. It is also possible for drug resistance to be induced by HIF-1α acetylation. HDAC4 regulates HIF-1α acetylation, which can be increased by HDAC4 short hairpin RNA (shRNA) and decreased by HDAC4 overexpression. In contrast, HIF-1α protein stability is inhibited by HDAC4 shRNA, which leads to a decrease in both the HIF-1α transcriptional activity and target gene expression \[[@R17]\]. Whereas no changes were shown in HIF-1α stability by HDAC1 or HDAC3 shRNA \[[@R17]\], HDAC1 and HDAC3 positively regulate HIF-1α stability *via* directly binding to the ODD domain of HIF-1α \[[@R21]\]. HDAC1 also regulates VEGF expression in normal keratinocytes (HaCaT cells) \[[@R31]\]. Endogenous HIF-1α increase by TSA treatment could hamper antitumor effect of TSA, which might be resulted in the reduction of TSA-mediated cell death. Our results showed that TSA increased the number of GFP-HIF-1α-positive cells in Figure [1I](#F1){ref-type="fig"}, which might be associated with the acetylation of K674 in HIF-1α (Figure [5](#F5){ref-type="fig"}). VEGF-HRE activity was much more enhanced by HIF-1α overexpression (Figure [6B](#F6){ref-type="fig"}), which was inhibited by mutant HIF-1α (K674R) overexpression. This is why the percentage of tumor cell death by TSA treatment looks like decreased by HIF-1α overexpression, which was overcome by mutant HIF-1α (K674R) overexpression (Figure [6C](#F6){ref-type="fig"}). So, antitumor effect of TSA might be increased by the inhibition of endogenous HIF-1α level. It implicates that GFP-HIF-1α-positive cells are more resistant to TSA-mediated cell death. These results demonstrate that drug resistance could be regulated by K674 acetylation in HIF-1α by various types of HDAC inhibitors, such as TSA. It has been reported that silent mating type information regulation 2 homolog 1 (SIRT1), NAD-dependent deacetylase, binds to HIF-1α and deacetylates it at K674 \[[@R32]\]. So, it can\'t be ruled out the possibility that TSA might inhibit SIRT1 to increase the acetylation of K674 in HIF-1α. Except the acetylation of K674 in HIF-1α, K532 is another acetylation site. ARD1 protein specifically acetylates K532 in the oxygen-dependent degradation (ODD) domain of HIF-1 residues 401--603, which regulates a HIF-1α stability \[[@R33]\]. K532 acetylation in HIF-1α increases the stability of Von Hippel-Landau (VHL) binding to HIF-1α and leads to the rapid proteasomal degradation of HIF-1α via ubiquitination \[[@R33]\]. Therefore, it will be further studied whether TSA might inhibit ARD1 and prevent VHL binding to and the degradation of HIF-1α through K532 acetylation. VEGF is essential for the blood vessel formation to maintain oxygen supply. HIF-1α increased by TSA-mediated acetylation enhances VEGF expression with HRE, which might be the effort for tumor cells to be survived in the environment with cell death signal. So, the increase in VEGF expression by TSA-induced HIF-1α acetylation could be resulted in promoting blood vessel formation and oxygen supply to increase tumor cell survival. Then, tumor cell death could be decreased by the increase in VEGF expression, which leads to TSA resistance. In the meanwhile, VEGF-HRE activity (\~ 9 fold changes) by HIF-1α (K674R) lacking the acetylation site was decreased roughly 28% as compared to its activity (\~ 12.5 fold changes) by wildtype HIF-1α (Figure [6B](#F6){ref-type="fig"}). This is rather smaller than VEGF-HRE activity (\~ 45 fold changes) in Figure [1](#F1){ref-type="fig"}. It might be due to the overexpression of these HIF-1α genes in the cell (Figure [6B](#F6){ref-type="fig"}). So, the effect of acetylation by TSA on VEGF-HRE activity is much stronger on just endogenous HIF-1α gene. TSA-mediated HIF-1α accumulation is also associated with the activation of Erk1/2 and Akt \[[@R16]\]. Because HIF-1α expression can be induced *via* the production of H~2~O~2~ \[[@R34]\], reactive oxygen or nitrogen species may be required for HIF-1α stabilization \[[@R33], [@R35]--[@R38]\]. However, it is necessary to further define various signaling pathways that are associated with the induction of TSA resistance via HIF-1α. A previous report showed that TSA is associated with the regulation of P-glycoprotein in drug-resistant cells \[[@R39]\]. p21WAF1 acts as a resistance factor to the combination therapy of TSA and interferon alpha (IFNα) \[[@R26]\]. Overexpression of GRP78/BiP, an antiapoptotic chaperone of the endoplasmic reticulum, reduces HDAC inhibitor-induced apoptosis in cancer cells \[[@R25]\]. Antitumor effects are also inhibited by a decrease in microtubule dynamics or MMP \[[@R40]\]. Therefore, it should not be ruled out the possibility that other pathways are involved in TSA resistance. In conclusion, our data demonstrate that tumor cell death induced by TSA is inhibited by HIF-1α acetylation, its nuclear translocation and binding to the HRE of the VEGF promoter. Our results suggest that the induction of drug resistance could be concomitant with the anticancer effects of TSA under normoxic conditions, providing additional evidence that the combined anticancer therapy regimen should be clinically beneficial to prevent drug resistance in populations at high risk of tumor relapse. MATERIALS AND METHODS {#s4} ===================== Reagents {#s4_1} -------- Trichostatin A (TSA) and MTT \[3(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide\] were purchased from the Sigma-Aldrich (St. Louis, MO, USA). Antibodies which are reactive with tubulin (T4026) and actin (A2066) were purchased from the Sigma-Aldrich (St. Louis, MO, USA). Antibodies which are reactive with acetylated-lysine (9441S), histone 4 (2592S) and normal rabbit immunoglobulin G (IgG) control (2729S) were purchased from Cell Signaling Technology (Beverly, MA, USA). Antibodies which are reactive with histone 4 (05--499), acetylated-histone 3 (06--599), acetylated-histone 4 (06--866) and normal mouse immunoglobulin G (IgG) control (12--371) were obtained from Milipore (Billerica, MA, USA). Antibodies which are reactive with hemagglutinin A (HA, MAB060) were purchased from R&D Systems, Inc. (Minneapolis, MN, USA). Antibodies which are reactive with HIF-1α (sc-10790), GFP (sc-9996), VEGF (sc-152), Lamin B (sc-6217) and glutathione-S-transferse (GST) (sc-138) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Antibodies which are specific to acetylated HIF-1α (K674) were kindly provided by professor Jong-Wan Park, College of Medicine, Seoul National University (Seoul, Korea). Luciferase assay system with reporter lysis buffer was purchased from Promega (Madison, WI, USA). Except where indicated, all other materials were obtained from the Sigma-Aldrich (St. Louis, MO, USA). Cell culture {#s4_2} ------------ HeLa human cervix adenocarcinoma cells, MCF7 human breast adenocarcinoma cells, HCT116 human colorectal carcinoma cells, HepG2 human hepatocellular carcinoma cells and HEK293T human embryonic kidney cells were obtained from the Korea Research Institute of Bioscience and Biotechnology (KRIBB) cell bank (Daejeon, Korea). Cells were maintained and cultured in Dulbecco\'s modified Eagle\'s medium (GIBCO BRL, Gaithersburg, MD, USA) supplemented with 10% fetal bovine serum (Hyclone, Kansas City, MO, USA), 2 mM L-glutamine, 100 units/ml penicillin, and 100 units/ml streptomycin (GIBCO BRL, Gaithersburg, MD, USA). No mycoplasma contamination was regularly tested by using mycoplasma detection kit (Thermo Fisher Scientific, Waltham, MA, USA) every other month. Plasmids and transient transfection {#s4_3} ----------------------------------- HRE reporter plasmid 5× VEGF-HRE-pSV40min was kindly provided from Dr. Dong-Soo Im, Korea Research Institute of Bioscience and Biotechnology (KRIBB, Daejeon, Korea). HA-HIF-1α and HA-HIF-1α (K674R) mutant plasmis were kindly provided by professor Jong-Wan Park, College of Medicine, Seoul National University (Seoul, Korea). pEGFP-C1-HIF-1α, pMXs-RFP1 and pSG5 plasmids were kindly provided from professors Mi-Ock Lee, College of Pharmacy, Seoul National University (Seoul, Korea), Jeong-Soo You, School of Medicine, Konkuk University (Seoul, Korea) and Soo-Jong Um, College of Bioscience and Biotechnology, Sejong University (Seoul, Korea), respectively. pSG5-HIF-1α plasmid was generated by subcloning HIF-1α sequences derived from pEGFP-C1-HIF-1α plasmid into the sites of EcoRI and BamHI of pSG5. To express the specific target molecules, the amount of each plasmid in the range of 0.3 \~ 0.5 μg was used for the transfection of each cell line in 35 mm^2^ culture dishes using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). To determine whether GFP-HIF-1α was tanslocated into nucleus by TSA treatment, cells were transfected with pEGFP-C1-HIF-1α plasmids or co-transfected with pEGFP-C1-HIF-1α and pMXs-RFP1 plasmids. Then, over 1,000 single GFP-positive or dual GFP- and RFP-positive cells were respectively counted under fluorescence microscope (Nikon Instruments Korea Co. Ltd, Seoul, Korea) MTT assay {#s4_4} --------- We quantified cell survival using colorimetric assay described for measuring intracellular succinate dehydrogenase content with MTT \[[@R41]\]. Confluent cells were cultured with various concentrations of each reagent for 48 h. Cells were then incubated with 50 μg/ml of MTT at 37°C for 2 h. Formazan formed were dissolved in dimethylsulfoxide (DMSO). Optical density (OD) was read at 595 nm. Hypoxia response element (HRE) reporter assay {#s4_5} --------------------------------------------- Active transcription factor HIF-1α recognizes and binds to the hypoxia-response elements (HRE; 5′-A/GCGTG-3′) in hypoxia-inducible promoters \[[@R42]\]. HRE reporter plasmid 5× VEGF-HRE-pSV40min was generated by cloning five tandem couples of HRE derived from the human VEGF promoter into the BglII site of pGL3 \[[@R43]\]. To measure the activity of VEGF transcription, confluent HeLa cells were transfected with VEGF-HRE-pSV40min plasmid using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) \[[@R44]\]. At the same time, cells were co-transfected with pcDNA-lacZ for monitoring transfection efficiency by β-galactosidase assay \[[@R45]\]. Luciferase activity was measured by using luciferase assay system with reporter lysis buffer (Promega, Madison, WI, USA). Luminescence was measured using luminometer (Berthold Technologies, Oak Ridge, TN, USA). VEGF-HRE activity was normalized to this control β-galactosidase activity. Fold changes in luciferase units of experimental group were also normalized to the control group \[[@R45]--[@R47]\]. Purification of nuclear fraction {#s4_6} -------------------------------- Cells were harvested and suspended with buffer A (25 mM Tris-Cl (pH 8.0) 10 mM KCl 1 mM DTT 0.5 mM PMSF) followed by the incubation on ice for 15 min. Cell suspension was treated with 10% NP-40 to the final concentration of 0.75% NP-40, mixed with vortexing for 30 sec and centrifuged at 1,400 × g for 1 min. Then, pellet was separated from supernatant. Nuclear fraction was obtained by the lysis of pellet in buffer C (50 mM Tris-Cl (pH 8.0) 400 mM NaCl 1 mM DTT 1 mM PMSF), the centrifugation at 15,000 × g for 30 min and the collection of supernatant. Cytoplasmic fraction was obtained by the centrifugation of supernatant at 1,500 × g for 15 min and the collection of its supernatant. Chromatin immunoprecipitation (ChIP) assay {#s4_7} ------------------------------------------ ChIP assay was performed as describied previously \[[@R48], [@R49]\]. Cells were crosslinked with final concentration 1% formaldehyde for 10 min at room temperature. Then 125 mM glycine was added to quench unreacted formaldehyde. Cells were gathered and sonicated to make DNA fragments with a size range of 200 \~ 1000 bp. Cell extracts were immune-precipitated using anti-GST (Santa Cruz, CA, USA) and rabbit IgG control antibodies (Abcam, Cambridge, UK) for all ChIP experiments. PCR analysis were performed by using primer sets spanning the HIF-1α binding site on hVEGF-A gene promoter. Primer sequences are 5′-aac aag ggc ctc tgt ctg c-3′ (--1,037 \~ --1,019 bp) and 5′-ggg gag aag aat ttg gca cc-3′ (--888 \~ --869 bp) including VEGF-HRE. Immunoprecipitation {#s4_8} ------------------- Immunoprecipitation was performed to detect proteins with acetylated lysine including HIF-1α as below. Cell lysates were incubated with rabbit polyclonal anti-acetylated lysine antibodies (Cell Signaling Technology, Beverly, MA, USA) for 2 h. Immunoprecipitates were collected by means of protein A/G PLUS agarose beads (Santa Cruz, CA, USA). Proteins with acetylated lysine were released from the beads by addition of 1× sample buffer and boiling for 5 min. HIF-1α with acetylated lysine was then detected by western blot analysis using an anti-HIF-1α antibody and enhanced chemiluminescence (ECL) (Pierce, Rockford, IL, USA). Reverse transcription polymerase chain reaction (RT-PCR) {#s4_9} -------------------------------------------------------- Total RNA was extracted from HeLa cells using TRIZOL reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized from 1μg of total RNA using oligo-dT~18~ primers and reverse transcriptase in a total volume of 21 μl (Bioneer, Daejeon, Korea). For standard PCR, 1 μL of the first-strand cDNA product as a template and 10 pmol of specific primers were used for PCR amplification with Taq DNA polymerase (Cosmo Genetech, Seoul, Korea). PCR amplification was performed using oligonucleotides specific for target genes, HIF-1α, VEGF, VEGF-A, VEGF-B, VEGF-C, VEGF-D, and β-actin (Table [1](#T1){ref-type="table"}). PCR products were detected by 1.2% agarose gel electrophoresis with a marker of 100 bp ladder. ###### Sequences of oligonucleotides specific for RT-PCR to target genes Target gene Forward primer Reverse primer ------------- ---------------------------------- ---------------------------------- HIF-1α 5′-ctc aaa gtc gga cag cct ca-3′ 5′-gat tgc ccc agc agt cta ca-3′ VEGF 5′-tga cag gga aga gga gga ga-3′ 5′-tgg ttt caa tgg tgt gag ga-3′ VEGF-A 5′-ctt gcc ttg ctg ctc tac ct 5′-gca gta gct gcg ctg ata ga-3′ VEGF-B 5′-tgt ata ctc gcg cta cct gc 5′-atc tgc atc cgg act tgg tg-3′ VEGF-C 5′-gac ctg ccc cac caa tta ca 5′-tcc agc tcc ttg ttt ggt ct VEGF-D 5′-gaa cac cag cac ctc gta ca 5′-aca gac aca ctc gca acg at β-actin 5′-gtc acc aac tgg gac gac at-3′ 5′-gca cag cct gga tag caa cg-3′ Realtime quantitative(Q)-PCR analysis {#s4_10} ------------------------------------- To perform real-time quantitative(Q)-PCR, total cellular RNA (1 μg) was reverse transcribed into cDNA as described in RT-PCR. Real-time Q-PCR was performed using the CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad laboratories, Hercules, CA, USA). The RT reaction product (10 ng) was amplified with Thunderbird SYBR qPCR mix (TOYOBO Co. Ltd., Osaka, Japan) using primers specific for target genes, HIF-1α primers (forward; 5′-gat gat gac ttc cag tta cg-3′ and reverse; 5′-tgg tag tgg tgg cat tag-3′), VEGF-A primers (forward; 5′-gga gtc caa cat cac cat-3′ and reverse; 5′-tgt ctt gct cta tct ttc ttt g-3′) and β-actin primers (forward; 5′-gcc agg tca tca cca ttg-3′, reverse; 5′-gtt gaa ggt agt ttc gtg gat-3′). Samples were heated to 95°C for 1 min and amplified for 40 cycles (95°C for 10 s, 55°C for 10 s and 72°C for 30 s) followed by a final extension step of 72°C for 10 min. β-actin was used as an internal control. Relative quantification of each mRNA was analyzed by the comparative threshold cycle (CT) method and normalized to β-actin expression using Bio-Rad CFX Manager™ Software. Western blot analysis {#s4_11} --------------------- Western blotting was performed using a standard protocol. Cells were lysed in ice-cold lysis buffer containing 0.5% Nonidet P-40 (vol/vol) in 20 mM Tris-HCl (pH 8.3); 150 mM NaCl; protease inhibitors (2 mg/ml aprotinin, pepstatin, and chymostatin; 1 mg/ml leupeptin and pepstatin; 1 mM phenylmethyl sulfonyl fluoride (PMSF); and 1 mM Na~4~VO~3~. Lysates were incubated for 1 h on ice before centrifugation at 13,000 × g for 10 min at 4°C. Protein concentration in each supernatant was measured using a SMART™ BCA protein assay kit (iNtRON Biotech. Inc., Seoul, Republic of Korea). Proteins were denatured by boiling for 5 min in sodium dodecyl sulfate(SDS) sample buffer. Proteins were separated by 12% SDS-polyacrylamide gel electrophoresis(SDS-PAGE), and transferred to nitrocellulose membranes by electro-blotting. Following transfer, equal loading of protein was verified by Ponceau staining. The membranes were blocked with 5% skim milk in Tris-buffered saline with Tween 20 (TBST) (10 mM Tris-HCl, pH 7.6; 150 mM NaCl; 0.5% Tween 20) and incubated with the indicated antibodies. Most antibodies were diluted to 1:000, except anti-actin and anti-tubulin antibodies to 1:5000, anti-histone 3/4 and acetylated histone 3/4 antibodies to 1:3000 with TBST. Bound antibodies were visualized with HRP-conjugated secondary antibodies with the use of enhanced chemiluminescence (ECL) (Pierce, Rockford, IL, USA). Immunoreactive bands were detected using X-ray film. Statistical analyses {#s4_12} -------------------- Experimental differences were examined using ANOVA and Students' *t*-tests, as appropriate. *P* values of \< 0.05 or \< 0.01 were considered to indicate significance. This work was supported by Grants from Mid-career Researcher Program (\#2016-R1A2B400746) and Public Problem-Solving Program (NRF-015M3C8A6A06014500) through the National Research Foundation (NRF) funded by the Ministry of Education, Science and Technology (MEST), Republic of Korea. **CONFLICTS OF INTEREST** None.

Energy Poplar (Enhancing Poplar Traits for Energy Applications) is an EC Seventh Framework Programme project aimed at further improving poplar trees as an energy crop. The work is directed to understand and improve traits such as yield and wood properties coupled to Bioethanol production. The project also addresses environmental and economical sustainability questions. The final goal of ENERGYPOPLAR is to develop poplar as a bioenergy short rotation coppice crop, suitable for large-scale deployment in Europe in areas unlikely to be used for food agricultural production. All will be placed in an environmental framework to ensure environmental sustainability with respect to land use, inputs and soil status World primary energy consumption increased by 2.4% in 2007. With the worlds growing energy consumption, the development and use of renewable, sustainable liquid bio fuels has become a strategic priority for the EU. Bio fuels can minimize energy import dependence, reduce greenhouse gas emissions and assist rural and agricultural development. Bio ethanol can be produced from energy crops that do not compete with food crops for land use and can be directly used by current transportation vehicles. This alcohol can be produced from biomass feedstock and in particular from cellulose, a sugar based polymer present in the cell wall of plants. Such crops are known as 'second generation' bio fuel crops. To develop a new bio ethanol industry competitive with fossil fuels, the quality of the biomass feedstock, the methods to produce ethanol from cellulose and the yield per hectare must be improved. Energy trees must also support an environmentally sustainable agriculture that uses less agrochemicals, develops rural economies and spares natural forests from agricultural expansion. Poplar is an economically and ecologically attractive energy crop because it displays a wide range of growth habits and can grow on marginal lands unsuited for food crops with reduced input costs and optimized land management. Additionally, poplar is a commercial crop and the model organism for hardwood tree genomics and physiology. <http://WWW.Energypoplar.eu> for more information.

Introduction {#Sec1} ============ Chimeric antigen receptor (CAR) T-cell therapy is an immunotherapeutic strategy which genetically modifies T cells expressing a cell-surface antigen to promote T-cell function and persistence. The remarkable efficacy of CD19-CAR T against hematologic malignancies encourages the exploration of CAR T therapies in solid tumors.^[@CR1]--[@CR3]^ A majority of CAR T studies utilize viral vectors to deliver CAR transgenes into T cells, owing to the high efficiency^[@CR4]--[@CR7]^. However, viral vector-based CAR T products require testing for the replication-competent virus to ensure safety, thereby leading to a complicated manufacturing process and high cost. Alternatively, non-viral vector piggyBac (PB) transposon can integrate CAR genes into T cells by electroporation, which is cost-effective, simple to use and no infectious risk^[@CR8],[@CR9]^. To generate high efficiency of CAR T cells, it is crucial to choose appropriate antigens to eliminate tumor cells with minimum toxicity. Mesothelin (MSLN) is becoming a promising antigen, because of its low expression on normal tissues and high expression on various solid tumors^[@CR10]^. MSLN is a cell-surface glycoprotein with normal expression in peritoneum, pleura, and pericardium, but with overexpression in a variety of cancers, including mesothelioma, pancreatic, lung, gastric, and ovarian cancers^[@CR10],[@CR11]^. Aberrant expression of MSLN plays a central role in cancer cell proliferation, invasion and metastasis through activating PI3K, ERK, and MAPK signaling pathways^[@CR12]^. In contrast, the function of MSLN in healthy tissues is dispensable^[@CR13]^, because of which a favorable safety profile had been observed in MSLN-targeted immunotherapies^[@CR14]--[@CR16]^. Thus, MSLN is an ideal cancer antigen for targeted immunotherapy^[@CR17],[@CR18]^. MSLN is a cell-surface glycoprotein, and its extracellular domain comprises of region I (residues 296--390), II (residues 391--486), and III (residues 487--598)^[@CR10],[@CR19]^. The N-terminal region I and C-terminal region III correspond to the membrane-distal region (MDR) and membrane-proximal region (MPR) individually. Most of the current MSLN-based immunotherapies target the MDR of MSLN^[@CR17],[@CR20]^, which also interacts with other functional proteins especially with CA125/MUC16^[@CR21]^. Although the CAR T-cell therapies that target MDR of MSLN have been reported to inhibit cancer growth, the overall efficacy is still relatively low in vivo^[@CR18],[@CR22],[@CR23]^. The low antitumor effect of CAR T cells may be caused by the MDR epitope that interacting with other functional proteins^[@CR24],[@CR25]^. In addition, a recent study suggested that a novel monoclonal antibody targeting the MPR of MSLN was efficacious for solid tumors^[@CR19]^. However, the direct functional comparison between MDR and MPR based immunotherapies has yet been identified. Therefore, we developed meso1 CAR T and meso3 CAR T targeting MDR and MPR of MSLN using the non-viral PB system, and the antitumor effects of the modified T cells were examined both in vitro and in vivo. Results {#Sec2} ======= Generation and characterization of meso1 CAR T and meso3 CAR T cells {#Sec3} -------------------------------------------------------------------- In order to generate the MSLN-CAR T cells, two CAR vectors targeting either region I of MSLN or region III of MSLN were constructed using a modified PB transposon system (Fig. [1a](#Fig1){ref-type="fig"}). Previous studies had demonstrated the PB system was an optimal method to deliver genes into T cells^[@CR26],[@CR27]^. We modified it to improve the efficiency of positive cell enrichment by using MSLN peptide coated plates. Flow cytometry showed that the mean positive ratios of both CARs were above 80% (*n* = 3; 87.85% vs 91.75%, Fig. [1b, c](#Fig1){ref-type="fig"}). The CAR expression was further confirmed by Western blotting (Fig. [1d](#Fig1){ref-type="fig"}). In addition, the CAR genes were integrated into the genome of T cells at an average of 7--8 copies/cell (Fig. [1e](#Fig1){ref-type="fig"}). Collectively, these data display the successful construction of meso1 CAR T and meso3 CAR T cells.Fig. 1Generation and characterization of meso1 CAR T and meso3 CAR T cells.**a** Schematic illustration of the meso1 CAR and meso3 CAR constructs. **b** Positive ratio of meso1 CAR T and meso3 CAR T cells were detected by flow cytometry. **c** Statistic analysis of positive rate of CAR T cells. **d** The protein expression of exogenous CD3ζ was detected by western blotting. **e** The average copy-number of CAR in meso1 CAR T cells and meso3 CAR T cells was detected by RT-PCR. Data are expressed as the mean ± SD; *n* = 3; ns *p* \> 0.05, \*\*\**p* \< 0.001 Contrastive analysis of meso1 CAR T and meso3 CAR T cells in vitro {#Sec4} ------------------------------------------------------------------ To test the immunocompetence of CAR T cells, T-cell markers, and cytokines were measured. Flow cytometry data indicated the mean percentage of CD3^+^CD8^+^ cells was significantly higher in both CAR T cells as compared with that in mock T cells (*n* = 3; 47.79% and 48.23% vs 34.33%, Fig. [2a, c](#Fig2){ref-type="fig"}). Also, the ratios of memory T cells in both CAR T cells were higher than that in control cells (Fig. [2a, c](#Fig2){ref-type="fig"}). In addition, the mean levels of activation marker CD69 were upregulated in both CAR T cells (*n* = 3; 29.11% and 31.80% vs 15.56%, Fig. [2b, c](#Fig2){ref-type="fig"}). Interestingly, although the levels of the lethality marker CD107α in the two CAR T groups were comparatively higher than that in mock T cells, the mean proportion of CD107α in meso3 CAR T cells showed much higher amount than that in meso1 CAR T cells (*n* = 3; 68.60% vs 46.69%, *p* \< 0.01, Fig. [2b, c](#Fig2){ref-type="fig"}). The surface marker PD-1 was also tested and showed that the mean PD-1 expression was higher in meso1 CAR T cells and meso3 CAR T cells than that in mock T cells (27.59% and 31.44% vs 10.46%, *p* \< 0.001, Fig. S[1](#MOESM1){ref-type="media"}).Fig. 2Contrastive analysis of the immunological activity of meso1 CAR T and meso3 CAR T cells.**a** CD3, CD4, CD8, and memory T cells of CAR T cells were assessed. Memory T cells contain central memory T cells (CD45RO^+^CCR7^+^) and effector memory T cells (CD45RO^+^CCR7^−^). **b** The activation markers CD69, and CD107α are detected. **c** Statistic analysis of T-cell markers. **d** Cytokines (IL-2, IL-4, IL-6, IL-10, TNF-α, and IFN-γ) were detected by flow cytometry using the CBA method. Data are expressed as the mean ± SD; *n* = 3; ns *p* \> 0.05, \**p* \< 0.05, \*\**p* \< 0.01, \*\*\**p* \< 0.001 The proliferation of T cells was analyzed by Hoechst 33342/Ki-67 assay, showing that mean T cells at the S/G2/M phase increased in meso1 CAR T and meso3 CAR T groups as compared with that in the control group (*n* = 3; 39.09% and 40.70% vs 24.03%, *p* \< 0.01, Fig. S[2](#MOESM2){ref-type="media"}). These data indicate both CAR T cells exhibit stronger proliferation capacities. The immunologic activities of CAR T cells were also tested by cytokines. The levels of interleukin-2 (IL-2), IL-10, TNF-α, and IFN-γ secreted in both CAR T cells were significantly higher than that in the control T cells. Compared with meso1 CAR T cells, meso3 CAR T cells induced higher levels of IL-2, TNF-α, and IFN-γ (Fig. [2d](#Fig2){ref-type="fig"}). Taken together, compared with meso1 CAR T cells, meso3 CAR T cells express higher levels of CD107α upon activation and produce increased levels of IL-2, TNF-α, and IFN-γ cytokines against MSLN-positive tumor cells. Cytotoxicity of meso1 CAR T and meso3 CAR T cells against MSLN^+^ cancer cell lines in vitro {#Sec5} -------------------------------------------------------------------------------------------- To select suitable cancer cell lines with high expression of MSLN, a panel of cell lines were analyzed by flow cytometry (Fig. [3](#Fig3){ref-type="fig"}). The results indicated that MSLN was highly expressed in H520, HGC-27, SKOV-3, ASPC-1, H292, and BT483 cells; but with low expression in CALU-6 and MDA-MB-231 cells. Therefore, HGC-27 and SKOV-3 cells were selected as targeted cell lines. To further confirmed MSLN expression in tissue levels, tissue chips from gastric cancer and ovarian cancer were measured by IHC (Fig. S[3](#MOESM3){ref-type="media"}--[5](#MOESM5){ref-type="media"}, Table [S1](#MOESM8){ref-type="media"}). MSLN expressed mostly on cytomembrane, with limited expression in cytoplasm. In gastric cancer, MSLN expression in the meso3 group was higher than that in the meso1 group (46.6% vs 25.0%, *p* \< 0.001). In ovarian cancer, a similar positive rate was detected in the two groups (55.8% vs 54.7%). In summary, these results confirm that MSLN is highly expressed in HGC-27 and SKOV-3 cell lines as well as in the corresponding cancer tissues.Fig. 3The expression of MSLN in cancer cell lines was detected by flow cytometry.Meso1 CAR-Fc and meso3 CAR-Fc were used as primary antibodies, whereas goat anti-human IgG Fc-PE was used as secondary antibody. **a** MSLN was expressed in H520. **b** MSLN was expressed in HGC-27. **c** MSLN was expressed in SKOV-3. **d** MSLN was expressed in ASPC-1. **e** MSLN was expressed in H292. **f** MSLN was expressed in BT483. **g** MSLN was expressed in CALU-6. **h** MSLN was expressed in MDA-MB-231 In cytotoxicity assays, the results revealed that meso3 CAR T cells exhibited stronger cytotoxicity than meso1 CAR T cells in a dose-dependent manner (Fig. [4a--d](#Fig4){ref-type="fig"}). At the low ratio of 1:1, the cytotoxicity of meso3 CAR T cells is stronger than that of meso1 CAR T cells (*n* = 3; 68.9% vs 41.2% against gastric cancer, *p* \< 0.001; 62.6% vs 35.6% against ovarian cancer, *p* \< 0.001). Then, we generated MSLN knockdown SKOV-3 cells using shMSLN and control cells with shCtrl. Results showed that meso1 CAR T cells and meso3 CAR T cells had stronger cytotoxicity against SKOV-3 cells transduced with shCtrl, while got weak cytotoxicity for SKOV-3 cells with MSLN knockdown (Fig. S[6](#MOESM6){ref-type="media"}). Furthermore, in the three-dimensional (3D) cancer spheroid model, although both CAR T cells were found to infiltrate into the tumor sphere, meso3 CAR T cells were more lethal to cancer cells than meso1 CAR T cells (Fig. [4e--h](#Fig4){ref-type="fig"}). Collectively, these results above indicate that meso3 CAR T cells exhibit greater cytotoxicity against MSLN-positive cancer cells than meso1 CAR T cells in vitro.Fig. 4Meso3 CAR T cells are more cytotoxic than meso1 CAR T cells in vitro.**a** Cytotoxic activities of the meso1 CAR T cells and the meso3 CAR T cells against HGC-27 cells were measured using RTCA system at the E:T ratio of 2:1. **b** Quantified data on the specific lytic levels of CAR T cells against HGC-27 cells were analyzed at different E:T ratios (mean ± SD; *n* = 3; one way ANOVA test). **c** Cytotoxic activities of the meso1 CAR T cells and the meso3 CAR T cells against SKOV-3 cells were measured using RTCA system at the E:T ratio of 2:1. **d** Quantified data on the specific lytic levels of CAR T cells against SKOV-3 cells were analyzed at different E:T ratios (mean ± SD; *n* = 3; one way ANOVA test). **e** The killing activity of meso1 CAR and meso3 CAR T cells was detected using the 3D cancer spheroid model in gastric cancer. **f** The time effect of death rate of tumor cells was shown by histogram in gastric cancer. **g** The killing activity of meso1 CAR and meso3 CAR T cells was detected using the 3D cancer spheroid model in ovarian cancer. **h** The time effect of death rate of tumor cells was shown by histogram in ovarian cancer Antitumor activities of meso1 CAR T and meso3 CAR T cells against MSLN^+^ gastric cancer in vivo {#Sec6} ------------------------------------------------------------------------------------------------ To determine the efficacy in vivo, the antitumor responses of CAR T cells in HGC-27 xenograft mouse model were examined. It was shown that the average tumor volume in the mock T group was increased to 420.4 mm^3^, whereas that in meso1 CAR T group was 98.8 mm^3^. Interestingly, tumors were almost eliminated in meso3 CAR T group (Fig. [5a--c](#Fig5){ref-type="fig"}), which is consist with the total fluorescence intensity data showing that the fluorescence intensity in meso3 CAR T group was significantly reduced compared with meso1 CAR T group (Fig. [5d](#Fig5){ref-type="fig"}). No differences in body weight among these three groups were observed (Fig. [5e](#Fig5){ref-type="fig"}). The potential organ toxicity was examined using hematoxylin and eosin (HE) staining, which revealed that no organ toxicities were found in normal tissues (Fig. S[7](#MOESM7){ref-type="media"}), indicating that meso3 CAR T cells are of safety with no serious off-target toxicity. Thus, the results indicate that meso3 CAR T cells display a stronger antitumor response than that of meso1 CAR T cells in vivo, and no harm to normal tissue.Fig. 5Therapeutic efficacy of meso1 CAR T cells and meso3 CAR T cells against gastric cancer in vivo.**a** Schematic representation of the procedure in gastric cancer. **b** Imaging scans of mice during the treatment in HGC-27 cancer model. **c** Tumor volume was measured and analyzed. **d** Fluorescence intensity of cancer cells was detected using an in vivo imaging system. **e** Body weight was recorded and analyzed. Data are expressed as the mean ± SD; *n* = 4 per group; \**p* \< 0.05,\*\*\**p* \< 0.001 Antitumor activities of Meso3 CAR T cells against large established ovarian cancer in vivo {#Sec7} ------------------------------------------------------------------------------------------ As meso3 CAR T cells had shown better antitumor activities in the gastric cancer model, it was sought to further explore their efficacy in an ovarian cancer mouse model with large established tumors (Fig. [6a](#Fig6){ref-type="fig"}). It was shown that meso3 CAR T cells eliminated most of the ovarian tumors in the early treatment group. Importantly, in the late treatment group, meso3 CAR T cells showed an efficient inhibition of the growth of large established tumors, and got no effect on body weight (Fig. [6b--e](#Fig6){ref-type="fig"}). Furthermore, the survival time of mice in meso3 CAR T early and late treatment groups was significantly longer as compared to that in the mock T group (Fig. [6f](#Fig6){ref-type="fig"}). Taken together, results above show that meso3 CAR T cells exhibit excellent antitumor response against the early stage of ovarian cancer, and a promising antitumor activity against the large ovarian cancer.Fig. 6Efficacy of meso3 CAR T-cell therapy against large established ovarian cancer in vivo.**a** Schematic representation of the procedure in ovarian cancer. **b** Imaging scans of mice during treatment in the SKOV-3 cancer model. **c** Tumor volume was measured and analyzed. **d** Tumor fluorescence intensity was detected using an in vivo imaging system. **e** Body weight was recorded. **f** The survival time was analyzed. Data are expressed as the mean ± SD; *n* = 5 per group; ns *p* \> 0.05, \**p* \< 0.05, \*\**p* \< 0.01, \*\*\**p* \< 0.001 Discussion {#Sec8} ========== In this study, we generated Meso1 CAR and Meso3 CAR targeting to MDR and MPR epitopes of MSLN, respectively, using the non-viral PB transposon system. Comparing with Meso1 CAR, Meso3 CAR produces higher levels of cytokines as well as eliciting stronger cytotoxicities against MSLN-expressing cancer cells in vitro and in a gastric cancer xenograft mouse model. Therefore, we suggest that MSLN MPR is a better epitope to target than MDR for CAR T design. We further provide evidence that Meso3 CAR T cells efficiently suppresses the large ovarian tumors in an NGS mouse model. Considering MSLN is an attractive solid tumor antigen, our work that identifies the better epitope for MSLN CARs contributes to the future clinical application for MSLN-positive patients. Although MSLN has been targeted in a few CAR T therapies, the antitumor activities showed in these studies are modest. In a PDX model of pancreatic cancer, tumor only had slight regression when treated with MSLN-CAR T cells^[@CR18]^. Another study had two patients with solid malignancies infused with MSLN-CAR T cells multiple times, both of them displayed stable disease^[@CR28]^. They also prove the safety and antitumor activity of MSLN-CAR T cells in pancreatic ductal adenocarcinoma, in which two of six patients had stable disease^[@CR29]^. These modest effects toward solid tumors clearly unmet the clinical request for improving responses. To be noted, all CARs used in the above studies targeted SS1 domain in MSLN region I (296--390), which belongs to the MDR. Here, our study provides an alternative MSLN epitope for CARs, which was proven to have a better antitumor performance in vivo. The MDR of MSLN is the binding domain for some functional proteins such as CA125/MUC16. In this case, antibody-based products targeting the region I have to compete with CA125/MUC16 for the MSLN antigen interaction, which may weaken the binding and function of these therapeutic reagents. For the same consideration, a previous study developed an MSLN region III (MPR)-targeted immunotoxin and showed its potent antitumor activities in vivo^[@CR19]^. Another explanation for the better antitumor effect observed in meso3 CAR could be due to the specific characteristics of the MPR epitope. As the MPR bridges the extracellular domain and transmembrane region of MSLN, it is the region where might have a rigid structure or responsible for a specific function, which is to provoke stronger antitumor response. A similar mechanism has been applied to HIV vaccine design, in which the MPR of gp41 was considered as the key factor to cause stronger immune responses^[@CR30],[@CR31]^. Reportedly, increased levels of the degranulation marker CD107α positively correlated with the function of T cells^[@CR32]^. The present study also shows that CD107α in meso3 CAR T cells was highly expressed as compared with that in mock T and meso1 CAR T cells. Moreover, IFN-γ production was significantly increased in the meso3 CAR T cells accompanied with high level of IL-2 and TNF-α, which exerted a synergistic enhancing effect for CAR T-cell function. This finding is agreed with a previous study, which found that IFN-γ production results in strong cytotoxicity of T cells^[@CR33]^. The primary advantages of the PB transposon system are high safety and easy to manufacture, making it suitable for clinical translation of CAR T-cell therapy^[@CR8],[@CR9]^. Moreover, the PB transposon system is adaptable for large and multiple gene fragments. However, its transfection efficiency is relatively low in T cells, averaging from 20--70%^[@CR34],[@CR35]^. In this study, we optimized the PB transfection protocol and developed an MSLN-coated plate cell enrichment method, by which positive CAR T cells can be above 80%, making it adequate for the clinical application. The treatment of large tumors using CAR T cells is usually challenging. Carpenito et al.^[@CR36]^ successfully treated large tumors (≈ 500 mm^3^) using lentiviral vector engineered MSLN-CAR T cells which contain costimulatory domains of CD28 and CD137 (4--1BB). Another study documented that the growth of large solid tumors can be repressed using PD-1-CAR T cells, owing to increased CAR T-cell infiltration and reduced inhibitory PD-1 signaling^[@CR37]^. The present study shows that meso3 CAR T cells can significantly inhibit large tumor growth (≈ 350 mm^3^), suggesting that targeting MPR epitope of MSLN is also a promising treatment for MSLN-expressing cancers. For CAR design in solid tumors, two of co-stimulating domains, 4--1BB and CD28, are most widely used. Some researchers have adopted 4--1BB as costimulatory domain, which can get persistent therapeutic effect with minor side effects^[@CR4],[@CR38]^, other researchers have utilized CD28, which can get stronger antitumor effect^[@CR39],[@CR40]^. A study has indicated that CD28 or 4--1BB costimulation can lead into distinct signaling in glycometabolism, 4--1BB costimulation enhances mitochondrial oxidative phosphorylation of CAR T cells, whereas CD28 costimulation improves glycolysis to meet their metabolic demands^[@CR41]^. Given the hypoxic microenvironment of solid tumors, CD28 costimulatory domain may benefit better than 4--1BB for CAR T cells to survive within solid tumors. Moreover, a common advantage of non-viral systems relative to viral systems is that the obtained CAR T cells can persist a higher level of memory T cells^[@CR42],[@CR43]^. Taken together, use of the non-viral system mediated by piggyBac transposon and adoption of CD28 costimulatory domain might be an ideal option to improve persistence of CAR T cells both in vitro and in vivo. Although meso3 CAR T cells showed safety and therapeutic advantage in gastric cancer and ovarian cancer, there are still several limitations. First, a subset of MSLN-negative cancer cells limits the cytotoxic effect of meso3 CAR T cells. Second, the tumor microenvironment might reduce the efficacy of CAR T cells. For improving the CAR T cells effect, combined immunotherapies are becoming a new research field. PD-1 blocking antibodies were used in combination with CAR T-cell therapy to improve the antitumor effect^[@CR44],[@CR45]^. Another strategy is that the CAR T cells are engineered to express the checkpoint inhibitors, which also gets promising efficacy for advanced solid tumors^[@CR37],[@CR46]^. In summary, the present study demonstrates meso3 CAR T cells targeting MPR of MSLN exhibit a strong efficacy against MSLN-expressing tumors as compared to that of meso1 CAR T cells targeting MDR. Moreover, meso3 CAR T cells could effectively inhibit the growth of large tumors in vivo, suggesting that MPR of MSLN is a promising epitope to target for solid tumor. Although CAR T-cell therapy has shown its potential in solid cancers, the efficacy will be weakened facing to large tumor burdens, in which cases, combined immunotherapies are expected to have a better performance. Materials and methods {#Sec9} ===================== Cell lines {#Sec10} ---------- Cancer cell lines were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China), including H292, H520, and CALU-6 (lung cancer), BT483 and MDA-MB-231 (breast cancer), HGC-27 (gastric cancer), SKOV-3 (ovarian cancer), and ASPC-1 (pancreatic cancer). These cells were cultured in RPMI-1640 or DMEM medium containing 10% fetal calf serum at 37 °C with 5% CO2. The *ffluc/GFP* fusion gene was transduced into HGC-27 and SKOV-3 cells to establish the HGC-27-luc and SKOV-3-luc cells. Generation of modified CAR T cells {#Sec11} ---------------------------------- The *meso3 CAR* gene was cloned into the PB transposon vector pNB328-EF1α to construct pNB328-meso3 CAR. pNB328-meso1 CAR and empty pNB328 vectors were used as controls. The antibody sequence for Meso3 CAR T was derived from the YP218 antibody, which was originally discovered by the NIH (<https://www.nature.com/articles/srep09928>; US Patent: US9803022: <https://patents.google.com/patent/US9803022>). In addition, the antibody sequence for Meso1 CAR T was derived from the SS1 antibody, which was also originally discovered by the NIH (US Patent: US7081518: <https://patents.google.com/patent/US7081518?oq=patent:7081518>). Fresh blood was collected from healthy volunteers after informed consent under a protocol approved by the Ethics Committee of the Second Military Medical University, China. For the generation of meso3 CAR T cells, peripheral blood mononuclear cells (PBMCs) were isolated, suspension cells were collected after adherent culture, then resuspended in electroporation buffer, and recombinant pNB328-meso3 CAR plasmids were electroporated into T cells according to the manufacturer's instructions (Lonza, Switzerland). Then, the T cells transfected with MSLN-CAR or Mock plasmid were seeded in six-well plates, which had been coated with MSLN antigen (5 μg mL^−1^)/anti-CD28 antibody (5 μg mL^−1^) or anti-CD3 antibody (5 μg mL^−1^)/anti-CD28 (5 μg mL^−1^) antibody, respectively. The T cells were specifically stimulated with the antigens/antibodies for 3 days in medium containing 200 U/mL recombinant human IL-2. Thereafter, the activated cells were cultured in medium containing 100 U/mL IL-2. All modified T cells were maintained in the medium for 10--14 days to proliferate enough quantity of CAR T cells. Flow cytometry {#Sec12} -------------- The expression of MSLN on cancer cells was detected by flow cytometry, using meso1 CAR-Fc and meso3 CAR-Fc as primary antibodies followed by goat anti-human-PE secondary antibody (eBioscience, USA). The expression of CAR on CAR T cells was detected using MSLN-Fc-biotin, followed by staining with PE-streptavidin. The immunophenotypes of T cells were tested using flow cytometry. Antibodies used for analysis include: CD3-PE-CY5, CD4-PE, CD8-FITC, and CD45RO-PE-CY5, CCR7-FITC, CD69-PC5, CD107α-PE-CY5, and PD-1-PE (BD Biosciences, USA). The proliferation of T cells was also assessed by flow cytometry. T cells were fixed using fixation/permeabilization solution kit, then incubated with Ki-67-APC and Hoechst 33342. All the data above were analyzed using the Kaluza analysis software (Beckman Coulter, USA). Immunohistochemistry (IHC) {#Sec13} -------------------------- The paraffin-embedded samples were sliced into 4-µm sections and baked at 70 °C for 2 h, followed by being deparaffinized in xylene and rehydrated in graded ethanol. The endogenous peroxidase was blocked, the antigen was retrieved, and blocked using goat serum. The sections were then probed with primary antibodies (biotinylated meso1 and meso3 antibodies), followed by horseradishperoxidase (HRP)-conjugated anti-biotin antibody. Subsequently, the slides were developed with DAB and counterstained with hematoxylin. Pancreatic cancer tissues served as the positive control for MSLN staining, whereas the pre-immune mouse IgG was used as the negative control. Generation of MSLN knockdown SKOV-3 cells {#Sec14} ----------------------------------------- Knockdown of MSLN in the SKOV-3 cells and the mock vector control cells were generated through shRNA lentiviral vectors with two shMSLN and scrambled shRNA (Genechem, China), respectively, according to the manufacturer's instructions. The lentiviral vectors and polybrene were added into the medium when the cells grew up to 30--40% confluence. 12 h later, fresh medium was replaced; then, transfected cells were selected with puromycin. The knockdown effect was verified by Western blotting and the cells were used for further experiments. Western blot analysis {#Sec15} --------------------- Western blotting was performed as described previously^[@CR47]^. T cells were harvested, lysed, and boiled to prepare the samples, subsequently separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis, and transferred to polyvinylidene difluoride membranes. After blocking, the membranes were probed with antibodies specific for CD3ζ or MSLN (Abcam, UK), followed by incubation with HRP-conjugated secondary antibodies. Protein bands was exposed to ECL (GE Healthcare, USA) followed by autoradiography. The endogenous CD3ζ or GAPDH was served as an internal control. Real-time PCR (RT-PCR) {#Sec16} ---------------------- T cells were collected, and total DNA was extracted using a gDNA extraction kit (Takara, Japan). The copies number of CAR genes was analyzed by RT-PCR. The reaction was carried out using a SYBR Green PCR Master Mix Kit (Toyobo, Japan) according to the manufacturer's instructions. The relative expression level was normalized to that of β-actin and calculated using the 2^−ΔΔCt^ method. Cytokines assays {#Sec17} ---------------- A total of 1 × 10^6^ CAR T cells were cultured in MSLN-coated plates for 24h. Then, cell supernatants were collected and treated using the Cytometric Bead Array (CBA) Human Th1/Th2 Cytokine Kit II according to the manufacturer's instructions (BD, USA). The levels of IL-2, IL-4, IL-6, IL-10, TNF-α, and IFN-γ were measured by flow cytometry (Beckman Coulter, USA). Cytotoxicity assays {#Sec18} ------------------- A real-time cell analyzer system (RTCA) and 3D spheroid cancer cell fluorescence measurements were used to assess the cytotoxicity. In the RTCA system, 1 × 10^4^ tumor cells/well were seeded and cultured for 24 h. Then, T cells were added into the unit at various effector/target cell (E:T) ratios (4:1, 2:1, and 1:1). The impedance signals were recorded at 5 min intervals. The signal-time curves were drawn to display the cytotoxicity. In the 3D spheroid model, 5 × 10^3^ cancer cells were stained using Hoechst stain (Beyotime, China) and plated into each well for 48 h to generate 3D spheroid cancer cells. Then, the T cells were stained with Calcein-AM (Dojindo, Japan) and added to the wells at the E:T ratio of 2:1, together with propidium iodide. The fluorescence values were analyzed using high-intension confocal microscopy (Opera Phenix, PerkinElmer) at 0, 4, 6, and 24 h after co-culturation. In vivo experiments {#Sec19} ------------------- Animal experiments were approved by the Institutional Animal Care and Use Committee of the Second Military Medical University, China. Female NSG mice, aged 4--6 weeks, were obtained from Beijing Vitalstar Biotechnology Co. Ltd (Beijing, China). In the HGC-27-luc xenograft experiments, 5 × 10^6^ HGC-27-luc cells were subcutaneously injected into NSG mice. At 10 days post injection, 12 mice were randomly divided into three groups: mock T, meso1 CAR T, and meso3 CAR T. Then, mice were injected with 1 × 10^7^ T cells intravenously. Every 7 days, the fluorescent images were acquired using Xenogen IVIS imaging system (PerkinElmer, USA), and the tumor volume was calculated using the formula: *V* = ½ (length × width^2^); also, the body weight was analyzed. In the SKOV-3-luc xenograft experiments, the same method was used to assess the antitumor effect. In all, 5 × 10^6^ SKOV-3-luc cells were subcutaneously injected into each mouse. Twenty mice were randomly divided into four groups: PBS, mock T, meso3 CAR T early treatment (a small tumor was established), and meso3 CAR T advanced treatment (a large tumor was established). Mice were intravenously injected with 100 μL PBS or 1 × 10^7^ meso3 CAR T cells/mouse on day 7 or 14. The tumor volume, fluorescent images, body weight, and survival were recorded and analyzed every 3--4 days. HE staining {#Sec20} ----------- The sections obtained from organs (including heart, liver, spleen, lung, kidney, and brain) from each group were paraffin-embedded and sliced into 4-µm sections. Then, the slides were baked at 65 °C for 1 h, deparaffinized in xylene, rehydrated by graded ethanol, and stained with HE successively. Statistical analysis {#Sec21} -------------------- All data are presented as mean ± SD. Statistical significance was analyzed by Student's *t* test, variance, or chi-square test. The survival data were analyzed by Kaplan--Meier curves and log-rank test. *p* \< 0.05 were considered statistically significant. All analyses were performed using GraphPad Prism v7.0 (La Jolla, CA, USA). Supplementary information ========================= {#Sec22} Figure S1 Figure S2 Figure S3 Figure S4 Figure S5 Figure S6 Figure S7 Table S1 Supplementary figure legends Edited by H-U Simon **Publisher's note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Zhiwei Zhang, Duqing Jiang, Huan Yang, Zhou He Supplementary information ========================= **Supplementary Information** accompanies this paper at (10.1038/s41419-019-1711-1). This study was supported by National Science Fund Projects of China (81672997), Shanghai Science and Technology Development Funds (18431903900), and National Key Research and Development Program on Precision Medicine of China (2017YFC0909800). QJQ and HJJ designed the experiment. ZWZ, DQJ, HY and ZH carried out most of the molecular and cellular experiments and drafted the manuscript. XZL, WXQ and LFL did the in vivo experiments. CW and YL carried out the statistical analysis. HL and HX performed the pathological experiments. All authors read and approved the final manuscript. The authors declare that they have no conflict of interest.

Introduction {#sec1-1} ============ Physical restraint is any manual method attached to the patient\'s body that restricts freedom of movement.\[[@ref1]\] In hospital settings, physical restraints were used largely to avoid falls and stop confused patients from harming themselves and others.\[[@ref2]\] Confused patients often remove their therapeutic devices their restraints act as a safety device. This will avoid harming themselves to promote the practice of medical modalities.\[[@ref3]\] However, in today\'s health care settings use of physical restraints is a controversial issue as it has dubious ethical and legal issues concerning the autonomy and dignity of patients.\[[@ref1]\] It was reported that the use of physical restraints causes physical injuries and has many psychological and sociological effects. On the other hand, hospital personnel argue that physical restraints are helpful in older patients for the prevention of falls and for keeping the medical devices applied in place.\[[@ref4]\] Despite the lack of evidence to approve the efficacy and safety of physical restraints and its possible risks, the use of physical restraints continues. The overall incidence of physical restraining has been reported to vary from 6% to 13% in various hospitals in India as well as worldwide.\[[@ref5]\] Nursing ethics bound a nurse to ensure patients\' safety and their fundamental rights. Nurses are the ones who initiate restraint use for ensuring safety of the patients. Nurses\' attitude toward physical restraints for maintaining patient safety can lead to conflicts with patients\' rights, along with their autonomy. Many studies proved that nurses\' knowledge toward physical restraint was average and their attitudes were either negative or ambivalent.\[[@ref6]\] A study by Sujata and Kaur from India reported that majority of (76.66%) nurses were found to have an average level of knowledge, (21.66%) were having poor knowledge, and only 1.66% were having a good knowledge regarding physical restraints.\[[@ref7]\] A study carried out by Sequeira and Halstead mentioned that nurses from a psychiatric ward did not have any emotional reaction, and majority of them verbalized that they had automatic responding during restraint application without any emotion.\[[@ref8]\] Nurses are the key decision-makers in the application of physical restraints to patients. Their major role to select the appropriate type of restraints, applying it safely to the patients, and providing care based on the frequent assessment of patients condition and response.\[[@ref9]\] The maintaining a positive attitude can affect nurses\' practice regarding restraints.\[[@ref10]\] However, many nurses have various doubts and misconceptions about their proper use which was found by a study by Huang *et al*., in 2003.\[[@ref11]\] Physical restraints are often a debatable issue among health care professionals who are involved in the care of patients in critical care and emergency units. Continuous use of restraints can cause lead long-term use of physical restraints can lead to numerous physical, psychological, social and functional problems in patients. Thus, the nurses need to be knowledgeable enough so that they can anticipate and prevent complications related to restraints such as abrasions, functional level, stress, fecal and urine incontinence, suffocation and dehydration.\[[@ref12]\] There is huge literature available that reflects that nurses have little knowledge about the application of restraints, for instance, a recent cross-sectional study conducted by Eskandari *et al*. showed that nurses have modest level of knowledge regarding the use of physical restraints. Most of the nurses were not knowing the exact reason of restraints\[[@ref1]\] Similarly, another study done in Hong Kong revealed that nurses have insufficient knowledge and unfavorable attitude toward restraint.\[[@ref6]\] Furthermore, a study from Turkish hospital reported nine deaths which occurred due to improper use of chest restraints.\[[@ref13]\] In order to decrease the mishappenings related to restraints, nurses need to be educated and properly trained beforehand The educational needs of nurses can be best understood by checking their knowledge and attitude toward restraints.\[[@ref14]\] It is imperative for the nurses to understand the reason and rationale behind the use of physical restraint, consideration of other measures and adverse consequences, polices regarding ethical issues, and document of restraint use.\[[@ref15]\] Since nurses\' knowledge and attitude toward physical restraints is directly linked to their practice, it is crucial to assess nurses\' knowledge, attitude regarding the use of physical restraints to develop standardized guidelines and initiate training programs in health-care settings.\[[@ref16]\] Although the use of physical restraints is well studied in western and many parts of Asian countries this study is rare in Uttarakhand. In addition, there is no relevant nursing guidelines for physical restraints have been published in India. Hence, this study intended to assess the knowledge and attitudes of nurses working in AIIMS, Rishikesh, regarding the use of physical restraints. Aim of the study {#sec2-1} ---------------- The present study aimed to assess nurses\' knowledge and attitude regarding physical restraint use in a tertiary care institute. Key factors that affect them were also studied. Materials and Methods {#sec1-2} ===================== Instrument {#sec2-2} ---------- The present study is a descriptive cross-sectional survey. Pretested, self-administered questionnaire items were used. The questionnaire has three sections. The first section included questions related to the nurses\' sociodemographic characteristics. Section 2 contained 20 items evaluating the extent of the nurses\' knowledge about physical restraints. Section 3 is about 15 items finding the nurses\' attitudes toward physical restraints. In the present study, the questionnaire was used in a tertiary care center and tested for reliability using split-half method with Cronbach\'s alpha coefficient for section 2 and 3 were *r* = 0.86 and *r* = 0.85, respectively. The tools were also validated by five experts. Sample selection {#sec2-3} ---------------- The included participants fulfilled the following criteria The registered nurses who are directly involved in patient careThose with a minimum 1 year of clinical experienceNurses who were full-time employees of the hospital andThose who were willing to participate in the study. Exclusion criteria {#sec2-4} ------------------ The registered nurses who were not willing to participate in the studyThe registered nurses who are not involved in direct patient care like nurses working in OPD and nurses who are working in teaching institute. The required sample size was estimated using  where *n* = Sample size. (1 − *n/N*) = Not taking in consideration as it has very little effect on sample size estimation. *t* = Standard deviation value is 1.96 at 0.05 level of significance. *p* = Assumed prevalence 7.83.\[[@ref17]\] *q* = 1 − p. *d* = Confidence interval 5%=0.05. The calculated sample size is = 110. The study population consists of 110 nurses randomly selected from various wards of the selected hospital who agreed to participate in the study. Nurses working in intensive care units, recovery unit, general wards (psychiatric ward, geriatric ward, medicine ward, neuro ward, and oncology ward) of the selected hospital were involved. Ethical approval and informed consent {#sec2-5} ------------------------------------- Permission was obtained from the Institutional Ethical Committee (IEC) of AIIMS, Rishikesh, to carry out the study. The study protocol was approved by the IEC of AIIMS, Rishikesh (No. 19/IEC/STS/2019). A written consent form was also obtained from all the participants. Data collection and analysis {#sec2-6} ---------------------------- This study was conducted from March 2019 to May 2019 among 110 nurses working in AIIMS, Rishikesh, to investigate the knowledge and attitude regarding the use of physical restraints. Both descriptive (frequency, percentage, mean, mean percentage, and Standard deviation) and inferential statistics (Mann--Whitney U-test and independent *t*-test) were applied to analyze and interpret the raw data. The data were then transferred into SPSS 23. Evaluation Version (IBM corporation, US) and were analyzed using descriptive and inferential statistics. Results {#sec1-3} ======= Majority of the nurses participated in the study were between 21 and 25 years (55%, *n* = 61) old, males (54%, *n* = 59), graduates (87%, *n* = 96), married (69%, *n* = 76), participants had work experience of 1--2 years (64%, *n* = 70), and working in intensive care units (25%, *n* = 27). About 63% (*n* = 69) of them reported that they did not undergo any in-service education regarding the use of physical restraints \[[Table 1](#T1){ref-type="table"}\]. ###### Comparison of sociodemographic factors and the mean scores of nurse's knowledge and attitude on the use of physical restraints (*n*=110) Variable Frequency (%) Knowledge Attitude ------------------------------------ --------------- ----------- ---------- ----------- ------- ------------ -------- --------- Age (years)  21-25 61 (55) 13 51.61 −1.453^a^ 0.147 34.06±4.92 −3.099 0.002\*  \>25 49 (45) 60.35 36.79±4.13 Gender  Male 59 (54) 14 60.16 −1.677^a^ 0.094 35.25±5.12 −0.65 0.948  Female 51 (46) 50.11 35.31±4.37 Qualification  GNM 14 (13) 13 55.11 −0.050^a^ 0.962 36.35±4.04 0.90 0.369  Graduation and above 96 (87) 55.56 35.10±4.86 Work experience (years)  1-2 70 (64) 13 53.65 −0.819^a^ 0.416 34.71±4.69 −1.664 0.099  \>2 40 (46) 58.74 36.27±4.79 Type of working area  ICU's 27 (25) 13 62.20 −1.279^a^ 0.203 34.74±5.05 −0.677 0.500  Non-ICU 83 (75) 53.32 35.45±4.69 Ever restrained a person  Yes 56 (51) 13.5 60.46 −1.883^a^ 0.060 34.98±4.50 −0.681 0.497  No 53 (49) 49.24 35.60±5.05 In-service education on restraints  Yes 40 (37) 14 61.15 −1.572^a^ 0.116 35.22±4.75 −0.094 0.925  No 69 (63) 51.43 35.31±4.81 ^a^Mann-Whitney U-Test, ^b^Independent *t*-test; \*Significant at *P*\<0.05. SD=Standard deviation, ICU=Intensive care units, \*GNM=General Nursing and Midwifery Knowledge regarding restraints {#sec2-7} ------------------------------ [Table 2](#T2){ref-type="table"} highlights that the knowledge scores of 110 participants were ranged from 7 to 20 points (mean 13.7 ± 1.9; mean percentage 46%). Majority 87 (79%) of the nurses had a moderate level of knowledge while 21 (19%) of them have adequate knowledge regarding the use of physical restraints. Only a handful of participants 2 (2%) had a low level of knowledge. ###### Knowledge regarding restraints (*n*=110) Knowledge Frequency (%) Mean (%) Total mean score SD ------------------ --------------- ---------- ------------------ ----- Low (≤7) 2 (2) 46 13.7 1.9 Moderate (8-14) 87 (79) Adequate (15-20) 21 (19) SD=Standard Deviation Attitude regarding restraints {#sec2-8} ----------------------------- Tables [3](#T3){ref-type="table"} and [4](#T4){ref-type="table"} depict that the nurse\'s attitude scores ranged from 28 to 55 points (mean 35.2 ± 4.7; mean percentage 47%). Equal proportion of the nurses has both favorable and unfavorable attitude toward the use of physical restraints 49% and 51%, respectively. ###### Attitude Regarding Restraints (*n*=110) Attitude Frequency (%) Mean (%) Total mean score SD ---------------------- --------------- ---------- ------------------ ----- Favorable attitude 54 (49) 47 35.2 4.7 Unfavorable attitude 56 (51) SD=Standard Deviation ###### Relationship between knowledge and attitude (*n*=110) Variables *r* *P* -------------------------------- ------- ------- Knowledge regarding restraints 0.084 0.381 Attitude toward restraints Relationship between knowledge and attitude {#sec2-9} ------------------------------------------- Data presented in [Table 5](#T5){ref-type="table"} illustrates that there is no relationship between knowledge and attitude of staff nurses regarding the use of physical restraints (*r* = 0.084). ###### Item Wise Mean Scores of Attitude Toward Restraints (*n*=110) Items Mean±SD ---------------------------------------------------------------------------------------------------------------------- ----------- If I were the patient, I would feel that I should have the right to refuse or resist the placing of restraints on me 1.80±0.94 I think that the patient's family members have the right to refuse the use of restraints 1.90±0.94 I think that the nurses have the right to refuse the application of patient restraints 2.92±1.08 I think that communication with patients during restraining is of no value 2.02±1.05 I think that the main cause of restraint application in the hospital is a staff shortage 1.75±1.05 I think that every patient on a ventilator should be restrained 2.27±1.04 It makes me feel terrible if the patient gets more upset after restraints are applied 2.30±0.87 I feel embarrassed when a patient's family members enter the room of a patient who is restrained 2.83±1.09 I feel guilty about placing a patient on restraints 2.76±1.06 I feel that a patient who is noisy deserve physical restraint 2.16±1.00 I think that a patient suffers a loss of dignity when placed on restraints 2.67±1.20 I think that restraints protect elderly from sustaining harm and injury 2.28±1.01 I think that aggressive or homicidal patients can be controlled by physical restraints only 2.91±1.10 I believe that restraints increase the risk of strangulation 2.50±0.91 I believe that restraints lead to a reduction in the number of fall injuries 2.15±0.95 SD=Standard deviation Comparison of knowledge and attitude regarding the use of restraints with selected sociodemographic variables. As the data were nonnormally distributed Mann--Whitney U-test and Kruskal--Wallis H-test was used to find the association between knowledge score and selected demographic variables. However, the findings revealed that there is no significant association found between the knowledge and sociodemographic variables. The attitudes score and the age group of subjects had a significant association (*P* \< 0.05) and analyzed by independent *t*-test. No other variables had a significant association with attitude scores \[[Table 1](#T1){ref-type="table"}\]. Discussion {#sec1-4} ========== This study is performed on 110 nurses working in a tertiary institute of Uttarakhand to investigate the knowledge and attitude regarding the use of physical restraints. Majority of them were younger than 25 years of age, have completed a degree course in nursing, and professional experience of 1--2 years. These findings were similar to other findings of a study conducted among nurses from Turkey and mentioned that young and inexperienced nurses are working in specialized medical units.\[[@ref17]\] In the present study majority 87 (79%) of the nurses had a moderate level of knowledge toward the use of physical restraints. It is surprising to know that the majority of the nurses did not undergo any training concerning the use of physical restraints. This could be one of the reasons in which their knowledge was at a moderate level. The findings of the study were similar with the findings of Chien and Lee which revealed that about two-thirds of nurses had modest level of knowledge of restraint use.\[[@ref18]\] Nuhu *et al*. in his study found that staff nurse knowledge on some aspects of restraints was poor and this may be due to lack of training.\[[@ref19]\] Another study done by Weiner *et al*. found that nurses working in geriatric wards of various nursing homes had more knowledge about the guidelines on restraints.\[[@ref20]\] One of the elements that play on a significant role in turning information into practice is the attitude of the individual.\[[@ref21]\] Attitude is a vibrant and guiding predisposition. Nurses\' attitudes toward physical restraints were assessed in this research so that they are not at the level anticipated. Overall the present study proved the nurses\' attitudes toward physical restraint tended to be neutral. Contradictory results were shown by Chien and Lee who reported that nurses reported a negative attitude toward the use of physical restraints.\[[@ref18]\] These results are comparable to the study results of Eskandari *et al*. where nurses had some negative attitude toward the use of physical restraints.\[[@ref1]\] The current study revealed that there is no significant relationship between knowledge and attitude of nurses regarding the use of physical restraints (*r* = 0.084). These findings are in contradiction with the findings of Eskandari *et al*.,\[[@ref1]\] Azab *et al*.,\[[@ref2]\] where a positive correlation was found between knowledge and attitude scores. Similarly, there was no significant relationship found between knowledge and other sociodemographic variables such as qualification, work experience, and type of working area. In general, nurses with bachelor\'s degree are expected to have higher scores in knowledge in comparison to diploma holders.\[[@ref22]\] Keeping this in mind, continuing nursing education on the use of physical restraints is the need of the hour for nurses working in clinical areas. Health professionals are using restraints in the view that it helps to manage the violent behavior of patients, but it has been seen that they are often using it along with pharmacological agents to control the patients. Hence, there is a need to educate professionals and develop guidelines for the use of physical restraints. Conclusion {#sec1-5} ========== The results of the present study indicated a moderate knowledge and attitude among nurses regarding restraint use. Underpinning the awareness and understanding about proper use of physical restraints among nurses is the key measure to improve the quality of patient care. Once the knowledge gaps are filled, a favorable and positive attitudes toward restraints can be developed in nurses which ultimately will affect their practice, and patient safety can be better ensured. Research emphasizes the development of proper nursing guidelines and training of nursing personnel regarding proper use of physical restraints. Financial support and sponsorship {#sec2-10} --------------------------------- AIIMS, Rishikesh. STS project. Conflicts of interest {#sec2-11} --------------------- There are no conflicts of interest. We are thankful to Ethical committee, AIIMS Rishikesh, for support and approval (approval code-NO.19/IEC/STS/2019) for conducting this study. We would like to thank all the nurses who participated in this study.

Background {#Sec1} ========== Technology advances now allow high-throughput genetic data to be generated with ever-improving speed and affordability. One major bottleneck in utilizing this data is the development of bioinformatics tools that can identify true signals amongst the high level of noise. Since the rise in popularity of the genome-wide association study (GWAS) over a decade ago, thousands of variants have been identified that are associated with complex human traits, including pharmacological outcomes \[[@CR1]\]. However, a large portion of the estimated heritability remains unexplained for many traits \[[@CR2]\]. Machine learning methods are promising candidates to address this issue and are currently used in other scientific fields, including drug design \[[@CR3]\]. One type of machine learning method is Random Forests (RF) \[[@CR4]\]. A caveat to RF is that there is no standard method for selecting a set of variants with low levels of false positives while retaining adequate power. More commonly used parametric analyses, such as linear or logistic regression, produce statistics with generally accepted values for null error rates. However, due to factors such as multiple testing correction and correlated variables, these parametric thresholds may be too stringent and could result in a large number of false negatives. One way to obtain null error rates for machine methods is to generate empirical distributions by running thousands of permutation analyses. This is computationally impractical for genome-wide studies. We propose a more efficient method, which integrates different selection parameters to identify the appropriate threshold between signal and noise called the Relative Recurrency Variable Importance Metric (r2VIM). The ultimate goal of our project is to generate variant sets and prediction models to further our understanding and prediction of complex traits. Methods {#Sec2} ======= Variable selection {#Sec3} ------------------ RF is a machine learning method that grows a collection of decision or regression trees to identify variables (e.g., single-nucleotide polymorphisms) that are associated with an outcome (e.g., blood pressure) while taking into account main and interaction effects. RF output is a ranked list of variables according to an importance score. Importance is calculated as the percent change in mean squared error after variable permutation. There is no standard method of selecting an importance score threshold that separates signal from noise. To this end, we incorporate and extend upon a previously proposed threshold selection method \[[@CR5], [@CR6]\]. Specifically, r2VIM combines 3 different variable selection components, as described below \[[@CR6]\].*Permutation-based importance score:* The raw variable importance metric (VIM) is calculated as the percent change in mean squared error (MSE) before and after random variable permutation.*Estimate of null variance*: If no variables are associated with the trait, the VIMs should be symmetrically and randomly distributed around zero. In practice, the lowest VIM is usually negative. Thus, we use the absolute value of the lowest VIM as an estimate of the null variance \[[@CR5], [@CR6]\]. This estimate can be used as a threshold by selecting only those variables with VIMs greater than the null variance. In previous studies, we observed that this estimate alone may be too liberal for genome-wide data. To address this, we multiply the estimate by factors, or integers, to create more stringent thresholds \[[@CR6]\]. For example, if the lowest VIM was-0.05, the null variance estimate would be 0.05. For more stringent thresholds, we could multiply by factors 2, 3, and 4 to get new thresholds of 0.10, 0.15, and 0.20, respectively.*Recurrency*: Due to the inherent randomness of machines, variables that are deemed important in one run may disappear in a second run with a different random seed. Variables with high importance scores across runs are more likely to be true signals. In r2VIM, we run RF 5 to 10 times and select variables greater than the threshold factor from variable selection component (2) across runs. For this analysis, we ran the above selection algorithm using RF on one of the simulated data sets for the systolic blood pressure (SBP) and Q1 (permuted) traits. We used the parallelizable Random Jungle (RJ) software to allow for the large number of input variables \[[@CR7]\]. We ran RJ with regression trees (numeric inputs and outputs), 60,000 variables sampled at each node (mtry), and 4000 trees in the forest (ntree). These parameters were selected after testing several mtry/ntree combinations and selecting the one with the lowest prediction error. The analysis took a total of 5 h (1 h for each run). Each analysis was run using 64 cores (approximately 63 trees per node). Along with the genetic variants, we included sex, blood pressure medication, smoking status, and the top ten principal components (PCs) in the RJ analysis. Age was not included because of missingness (18/1937 with no age given). We calculated the variable relative importance score (RIS) for each run as *VIM/abs(min. VIM)*. This allows us to compare scores across runs. We combined the RIS values over the 5 runs by selecting the minimum value. This is considered "recurrency-corrected." We analyzed the same data set using linear regression. The model included main effect terms for the variant, blood pressure medication, smoking status, sex, age, and the top 10 PCs. Age was included in the linear regression analysis, as the inability to handle missingness is a weakness of r2VIM. Incorrect results would represent this weakness and be a fair comparison of the 2 methods. A Bonferroni correction on the *p* value for the variant term was used as the selection threshold. We compared the selected variants for the 2 methods based on the simulated disease model for this data. Data set {#Sec4} -------- We used the Genetic Analysis Workshop 19 (GAW19) simulated SBP and Q1 phenotype data for 1937 unrelated individuals \[[@CR8]\]. For both we used data set replicate numbered 100. The genetic data was generated using whole exome sequencing. Singletons and variants with any missingness were removed leaving 353,103 total variants. In the filtered set, 1047 variants were directly simulated to be functional, and 4328 variants are in genes with simulated functional variants. The functional variants had a wide range of effect sizes and minor allele frequencies, as described by the Genetic Analysis Workshop data contributors. This data set has previously undergone several quality filtering steps; however, an independent quality control analysis identified more samples that did not meet certain data quality requirements. A sibpair was identified by identity-by-descent (IBD) estimation analysis. The member of the pair with lower overall coverage and higher missingness rate was dropped. We also dropped 3 samples with 10× coverage of less than 0.7, 1 sample with missingness greater than 0.05, and 1 sample that was a clear outlier for the number of singletons. After these 5 samples were dropped, a PCs analysis was completed. The top 10 PCs were added as covariates into the linear regression model and were included as variables in the r2VIM analysis. The genetic data set is formatted as a standard PLINK binary input with genotypes coded as 0/1/2 indicating the number of alternative alleles present. Results {#Sec5} ======= We ran r2VIM and linear regression on the GAW19 SBP simulated phenotype. Figure [1](#Fig1){ref-type="fig"} shows the results for the linear regression analysis of SBP. Each row represents a different *p* value selection threshold: *p* \<0.05 and *p* \<5 × 10^−7^ (Bonferroni corrected). All of the variants selected at the given thresholds are shown in black (not simulated to be functional) and red (functional). The left column shows the functional variants as red, and the right column shows the variants in functional genes as red. The counts for the selected variants are shown in Table [1](#Tab1){ref-type="table"}. Figure [2](#Fig2){ref-type="fig"} shows the results for the r2VIM analysis. Here we show the selected variants at 3 minimum RIS thresholds: min.RIS \<0, min.RIS \<0.5, and min.RIS \<1. The counts for the total functional versus nonfunctional variants selected are shown in Table [1](#Tab1){ref-type="table"}.Fig. 1Results for the linear regression analysis of the simulated SBP phenotype for 2 *p* value thresholds (*p* \<0.05 and *p* \<5 × 10^−7^). The *x*-axis represents the variant index, which is in order of genome location. The *y*-axis shows the − log~10~(*p* value). The variants in *red* indicate functional variant (*left*) and variants in functional genes (*right*) Table 1Counts for the number of variants selected at different thresholds for linear regression (*p* value) and r2VIM (min.RIS)MethodThresholdAll (\~350 k)Func. vars. (1047)Func. genes (4328)Linear Regression*p* \<0.05773935136*p* \<5 × 10^−7^5969r2VIMmin.RIS \>034069min.RIS \>0.53758min.RIS \>12558The total number selected, the number of variants simulated directly to be functional, and the number of variants in the functional genes are shown Fig. 2Results for the r2VIM analysis of the simulated SBP phenotype for 3 min.RIS thresholds (RIS \<0, RIS \<0.5, and RIS \<1). The *x*-axis represents the variant index, which is in order of genome location. The *y*-axis shows the min.RIS. The variants in *red* indicate functional variant (*left*) and variants in functional genes (*right*) Next, we assessed the correlation between the *p* values and the minimum RIS scores for linear regression and r2VIM, respectively (Fig. [3](#Fig3){ref-type="fig"}). These graphs suggest agreement between the scores. It does not appear that functional variants were identified as having strong signals in one method and not the other. Of note, variants in black are not necessarily nonfunctional, as they could still be in linkage disequilibrium with a functional variant but not in the same gene. This makes it difficult to do a reliable true-positive versus false-positive assessment.Fig. 3Comparison of the min.RIS score from the r2VIM analysis (y-axis) and the − log~10~(*p* value) for the linear regression analysis of the simulated SBP phenotype. The variants in *red* indicate functional variant (*top*) and variants in functional genes (*bottom*) Finally, we ran r2VIM using the Q1 phenotype (Fig. [4](#Fig4){ref-type="fig"}). This phenotype was not simulated to be correlated with any of the variants. Therefore, any positive associations should represent false-positive selection. None of the variants were significant after Bonferroni correction for the linear regression analysis of the Q1 data set (results not shown). Variants simulated to be functional in the SBP phenotype model are shown in red. Certain variants in this model have high min.RIS scores; however, this could be an artifact of the simulation method and further testing needs to be done.Fig. 4Results for the r2VIM analysis of the simulated Q1 phenotype. The *x*-axis represents the variant index, which is in order of genome location. The *y*-axis shows the min.RIS. The variants in *red* indicate functional variant (*left*) and variants in functional genes (*right*) from the SBP phenotype simulated model Discussion {#Sec6} ========== For this GAW19 data set, linear regression and r2VIM had similar true-positive and false-positive selection counts when a Bonferroni corrected *p* value and a minimum RIS score of 0.5 were used as thresholds. Both methods identified the 2 genes with the strongest simulated effects, as shown by the 2 highest peaks in Figs. [1](#Fig1){ref-type="fig"} and [2](#Fig2){ref-type="fig"}. Although a key motivation for r2VIM development is the identification of more complex models (e.g., interactions with small main effects), this is an important proof-of-concept for main-effect-only selection, as these are likely to be present along with interaction effects. Notably, neither method identified a large proportion of the functional variants. Both appeared to have an effect size/allele frequency threshold that resulted in no power for selection. Future r2VIM testing will be done using techniques to boost power for variants, such as binning by functional region. There are several major limitations of RF (and, therefore, r2VIM) including the inability to handle any data missingness. In this analysis, we dropped variants with missing data points. Imputation is another option; however, it is often time-consuming and can result in many variants with unreliable calls. Future work will be done on the best and most efficient way to impute genotypes for r2VIM analysis to balance data gain with data quality. Another current limitation is the amount of computational power required to run RJ on large data sets. This analysis required a large number of high-memory processors. Future work will be done on ways to improve memory consumption. Finally, the best way to select the RIS threshold is still not clear. Here we showed 3 different thresholds and selected the best one according to the simulated model. This threshold selection process would not be possible in a biological data set where the underlying model is unknown. However, the optimal threshold is going to be dependent on the underlying model and will be different for different data sets. To this end, we plan on incorporating a "null distribution" analysis, by permuting the phenotype and running r2VIM a computationally feasible number of times. By comparing this distribution to the alternate, we may be able to determine a more regimented threshold selection process. Conclusions {#Sec7} =========== For this analysis, we were able to show that r2VIM is a promising candidate for variable selection, as it performs as well as the more commonly used linear regression method for the identification of main effects. Power to detect causal variants was low as expected given the simulated model, but false positive rates were similar between linear regression and RF for the SBP trait. Perhaps the most important model to simulate will be one that contains main and interactions effects, as true biology is likely to contain both. The final goal is to generate predictive models that allow for all types of effects, which would require a method robust to more than just main effects. With these models, we will be able to gain deeper insight into the true etiology of complex human disease. This research was supported by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health and the Center for Inherited Disease Research, Johns Hopkins University (contract \# HHSN268201200008I). Declarations {#FPar3} ============ This article has been published as part of *BMC Proceedings* Volume 10 Supplement 7, 2016: Genetic Analysis Workshop 19: Sequence, Blood Pressure and Expression Data. Summary articles. The full contents of the supplement are available online at <http://bmcproc.biomedcentral.com/articles/supplements/volume-10-supplement-7>. Publication of the proceedings of Genetic Analysis Workshop 19 was supported by National Institutes of Health grant R01 GM031575. Authors' contributions {#FPar1} ====================== EH, SS, JM, and JBW are the main developers of r2VIM. EH performed all of the r2VIM analysis in this manuscript. EP, HL, SG, PZ, QL, and CC were involved in the pre-processing steps of the GAW19 data (acquisition, formatting, and quality control procedures). All authors read and approved the final manuscript. Competing interests {#FPar2} =================== The authors declare they have no competing interests.

All relevant data are within the paper. Introduction {#sec005} ============ Pulmonary emphysema and idiopathic pulmonary fibrosis (IPF) are two distinct entities defined by different clinical, functional, radiological, and pathological criteria \[[@pone.0157789.ref001]\]. IPF is the most common of the idiopathic interstitial lung diseases (i-ILDs) and has the histopathologic and/or radiologic pattern of usual interstitial pneumonia (UIP) \[[@pone.0157789.ref002]\], while emphysema is defined as an enlargement of the air spaces distal to the terminal bronchioles due to the destruction of tissues forming their walls \[[@pone.0157789.ref003]\]. These two entities coexist in a condition characterized by upper lobe emphysema and lower lobe pulmonary fibrosis, which is known as combined pulmonary fibrosis and emphysema (CPFE) \[[@pone.0157789.ref001], [@pone.0157789.ref004]\]. The coexistence of the two conditions, which have different pathophysiological and functional characteristics, results in the development of a disease entity, with distinct clinical and functional characteristics and different prognosis compared to its individual components \[[@pone.0157789.ref005]\]. Pulmonary surfactant is a highly surface-active mixture of proteins and lipids that is synthesized and secreted onto the alveoli by type II epithelial cells \[[@pone.0157789.ref006], [@pone.0157789.ref007]\]. The protein part of surfactant constitutes of four types of surfactant proteins (SP), SP-A, SP-B, SP-C and SP-D. SP-A and SP-D are hydrophilic proteins that regulate surfactant metabolism and have immunologic functions, whereas SP-B and SP-C are hydrophobic molecules, which play a direct role in the organization of the surfactant structure in the interphase and in the stabilization of the lipid layers during the respiratory cycle \[[@pone.0157789.ref008], [@pone.0157789.ref009]\]. Different polymorphisms of SP-A \[[@pone.0157789.ref010]\], and SP-B \[[@pone.0157789.ref010]--[@pone.0157789.ref013]\] genes and mutations in SP-C \[[@pone.0157789.ref014]\] genes have been related to COPD. Furthermore, studies have shown that circulating SP-A and SP-D levels were increased in patients with COPD compared to normal smokers and non-smoking controls and correlated to airway obstruction \[[@pone.0157789.ref015]--[@pone.0157789.ref017]\]. On the other hand, there is evidence showing a possible role of surfactant proteins in the development of ILDs. Infants and children with mutations in the genes encoding surfactant proteins develop such a disease early in their life \[[@pone.0157789.ref018]\], while mutations of SP-A and C genes are associated with familial interstitial lung disease \[[@pone.0157789.ref019]--[@pone.0157789.ref023]\]. Serum levels of SP-D have been shown to be higher in patients with IPF compared to control subjects \[[@pone.0157789.ref024], [@pone.0157789.ref025]\]. Furthermore, serum SP-D concentrations are related to the annual rate of deterioration of pulmonary function \[[@pone.0157789.ref026]\], while high serum levels of SP-A and SP-D seem to be predictors of rapid disease deterioration and have been associated with poor survival in patients with IPF \[[@pone.0157789.ref025], [@pone.0157789.ref027], [@pone.0157789.ref028], [@pone.0157789.ref029]\]. Finally, there is evidence than SP-C gene mutations are associated with CPFE \[[@pone.0157789.ref030]--[@pone.0157789.ref033]\], while SP-D has been reported to correlate to pulmonary function in these patients \[[@pone.0157789.ref034]\]. According to the above, we hypothesized that in patients with CPFE, SP would be more disturbed in comparison to patients with only IPF and patients with only emphysema in a clinically significant way. The aim of the present study was to evaluate serum levels of Surfactant Proteins (A, B, C and D) in patients with CPFE, emphysema only, IPF only, and healthy controls and to test their possible associations to pulmonary function, disease severity, and survival. Methods {#sec006} ======= Subjects {#sec007} -------- From February 2013 to December 2014 we enrolled 147 consecutive patients (31 with emphysema, 62 with IPF, and 29 with CPFE) and 25 healthy controls (smokers and non-smokers). All study subjects were in stable condition and did not report any exacerbation or respiratory tract infection during the last 8 weeks. Patients with α~1~-antitrypsin deficiency, connective tissue disease at the time of diagnosis of CPFE, diagnosis of other interstitial lung disease (such as drug induced ILDs, pneumoconiosis, hypersensitivity pneumonitis, sarcoidosis, pulmonary histiocytosis, LAM, and eosinophilic pneumonia), patients unable to perform PFTs, as well as patients with inability or unwillingness to collaborate with the investigators were not included in the study. All subjects were followed up for one year and their vital status has been recorded. Study design {#sec008} ------------ All patients provided a detailed history upon arrival, and underwent physical examination and blood samples collection. Subsequently they were submitted to pulmonary function tests, measurement of body mass index (BMI) and evaluation of exercise capacity using the 6 minutes walking test. Arterial blood gases were also measured. Finally, all patients underwent high resolution computed tomography of the chest (HRCT). The study protocol was approved by the ethics committee of "Attikon" University Hospital, Chaidari Athens Greece and all patients gave written informed consent. Pulmonary Function Tests {#sec009} ------------------------ Pulmonary function tests (PFTs) were performed with commercially available system (Master Screen, Erich Jaeger GmbH, Wuerzburg, Germany) and included post-bronchodilator forced expiratory volume in one second (FEV~1~), FVC, FEV~1~/FVC ratio, total lung capacity (TLC), residual volume (RV), inspiratory capacity (IC) and diffusing capacity for carbon monoxide (DL~CO~). Diffusing capacity for carbon monoxide (DL~CO~) and diffusing capacity for carbon monoxide adjusted for alveolar volume (DL~CO~/V~A~) were assessed by means of the single breath method with the patient in the sitting position. Lung function measurements were expressed as percentages of predicted values. Tests were performed according to the American Thoracic Society guidelines \[[@pone.0157789.ref035]\] by the same technician in order to ensure consistency of the results. The composite physiologic index that is associated to mortality in IPF and CPFE \[[@pone.0157789.ref036]\] was \[calculated according to the following formula: 91.0---(0.65 x percent predicted diffusing capacity for carbon monoxide \[DLCO\])---(0.53 x percent predicted FVC) + (0.34 x percent predicted FEV~1~)\]\[[@pone.0157789.ref037]\] was determined in all subjects \[[@pone.0157789.ref038]\]. In all patients, arterial blood samples were taken for the measurement of PaO~2~ and PaCO~2~ using a commercially available blood gas analyzer (model 1630; Instrumentation Laboratories, Milan Italy) HRCT {#sec010} ---- All patients underwent HRCT using Brilliance CT 64-channel scanner (Philips, Eindhoven, the Netherlands). Scans were performed with 1--1.5mm section thickness and a 1--2 sec scanning time during breath holding at end inspiration. Films were read by a radiologist with expertise in HRCT who was blinded to the rest of the measurements. The degree of emphysema and/or fibrosis was calculated using a visual score as previously described \[[@pone.0157789.ref039], [@pone.0157789.ref040]\], using a five-point scale based on the percentage of lung involved: 0: no emphysema/fibrosis; 1: up to 25% of the lung parenchyma involved; 2: between 26--50% of lung parenchyma involved; 3: between 26--75% of the lung parenchyma involved; and 4 between 76--100% of lung parenchyma involved. Grades of the axial images of each lung were added and divided by the number of images evaluated to yield emphysema and/or fibrosis scores that ranged from 0 to 4 \[[@pone.0157789.ref039], [@pone.0157789.ref040]\]. The percentage of destructed lung area (%DLA) was obtained by summing %emphysema and %fibrosis \[[@pone.0157789.ref041]\]. The presence of emphysematous lesions ≥15% (i.e. score ≥0.6) of the pulmonary parenchyma in CPFE patients was considered as significant \[[@pone.0157789.ref042], [@pone.0157789.ref043]\] Assessment of Dyspnea and Exercise Capacity {#sec011} ------------------------------------------- Dyspnea was assessed with the modified (5-point) Medical Research Council (MRC) dyspnea scale, as previously described \[[@pone.0157789.ref044]\]. Exercise capacity was measured with the 6 minutes walk test in a 100 ft corridor, according to the ATS guidelines \[[@pone.0157789.ref045]\]. The distance walked in 6 minutes, difference in dyspnea according to Borg scale and alterations in oxygen saturation were recorded in each patient. Blood Samples Collection and Measurements of Surfactant Proteins {#sec012} ---------------------------------------------------------------- Blood samples were collected from each subject and were immediately centrifuged at 1500g for 15 min at 4°C and the supernatant (serum) was stored at -80°C until measurements. Serum surfactant proteins were measured using commercially available enzyme-immunosorbent assay (ELISA) kits (SP-A, SP-B and SP- C, USCN Life Science, China and, SP-D, Bioventor, Germany) according to the manufacturer\'s protocol. The lower limits of detection were the following: SP-A: 8.2 pg/mL, SP-B: 0.31 ng/ml, SP-C: 0.112 ng/mL and SP-D: 0.01 ng/mL. Statistical analysis {#sec013} -------------------- Normality of distributions was checked with Kolmogorov-Smirnov test. Comparisons between groups were performed with Kruskal-Wallis tests, with appropriate post-hoc tests (Dunn\'s test). Correlations were assessed using Spearman's rank correlation coefficient. For the avoidance of type I errors during the performance of multiple correlations we have used the correction method described by Benjamini and Hochberg (1995)\[[@pone.0157789.ref046]\] according to which a p-value\<0.023 was considered statistically significant. Group data are expressed as mean±SD or as median (interquartile ranges) for normally distributed and skewed data, respectively. Kaplan Meier Curves were created in order to compare survival in patients with low and those with elevated SP levels and the groups were compared using Log Rank test. The 75^th^ percentile of each SP was used as a cut-off point in order to divide the patients in those with elevated SP levels and those with non-elevated SP levels. Graphical presentations of times to events were performed with Kaplan-Meier curves. Results are presented as hazard ratios (HR) with 95% confidence intervals (CI). With the exception of the multiple correlations, all other p-values \<0.05 were considered statistically significant. Analysis was performed using SPSS 17 statistical package (SPSS, Chicago, IL) and GraphPad Prism 5 (GraphPad Software Inc, La Jolla, CA, USA). Results {#sec014} ======= We evaluated 184 consecutive patients with either emphysema, IPF and CPFE and 122 of them were eligible to be included in the present study. The flow chart of the study participants is shown in [Fig 1](#pone.0157789.g001){ref-type="fig"}. An additional 25 healthy subjects (smokers and non-smokers) were included as healthy controls. The demographic and functional characteristics as well as serum levels of surfactant proteins of the 122 patients and 25 healthy controls included in the study are presented in [Table 1](#pone.0157789.t001){ref-type="table"}. Serum SP-A and SP-D levels significantly differed between groups (p = 0.006 and p\<0.001 for SP-A and SP-D, respectively; [Fig 2A--2D](#pone.0157789.g002){ref-type="fig"}). In post-hoc analysis, SP-A levels differed significantly only between controls and CPFE (p\<0.05) and between CPFE and emphysema (p\<0.05). Similarly, post hoc analysis regarding serum SP-D levels revealed significant differences between controls and fibrosis or CPFE as well as emphysema and fibrosis or CPFE (p\<0.001 for all comparisons). {#pone.0157789.g001} {#pone.0157789.g002} 10.1371/journal.pone.0157789.t001 ###### Demographic and functional characteristics of the study participants. {#pone.0157789.t001g} Controls N = 25 Emphysema only N = 31 IPF only N = 62 CPFE N = 29 p value -------------------------------- -------------------------- -------------------------- --------------------------- -------------------------- ------------- **Age (years)** **64.0(55.5, 67.5)** **66.0 (59.0, 71.0)** **72.0 (66.8, 78.0)** **75.0 (63.0, 77.0)** **\<0.001** **Gender (Female) N(%)** **14 (56.0%)** **7 (22.6%)** **19 (30.6%)** **3 (10.3%)** **0.002** **BMI kg/m**^**2**^ **26.7 (24.5, 29.5)** **24.9 (21.5, 26.8)** **27.2 (24.5, 29.6)** **27.9 (23.6, 29.1)** **0.005** **Smoking (current/ex/never)** **9/5/11** **18/13/0** **5/30/27** **5/22/2** **\<0.001** **Pys** **10.0 (0.0, 35.0)** **70.0 (40.0, 95.0)** **12.5 (0.0, 36.0)** **50.0 (30.0, 75.0)** **\<0.001** **FEV**~**1**~ **(%pred)** **92.0 (87.5, 99.0)** **56.4 (33.7, 79.6)** **76.4 (65.5, 88.1)** **76.3 (69.5, 98.0)** **\<0.001** **FVC (%pred)** **90.0 (82.0, 93.5)** **72.8 (60.1, 93.7)** **68.3 (57.9, 82.1)** **68.5 (60.3, 94.1)** **0.001** **FEV**~**1**~**/FVC** **82.0 (78.5, 85.2)** **54.4 (40.4, 69.0)** **85.7 (81.3, 90.8)** **79.0 (74.4, 86.7)** **\<0.001** **DLCO (%pred)** **85.0 (82.0, 88.0)** **52.4 (36.5, 72.0)** **46.8 (35.2, 62.9)** **32.0 (22.8, 46.4)** **\<0.001** **TLC (%pred)** **85.0 (82.0, 87.5)** **93.0 (81.2, 105.8)** **59.1 (50.2, 73.2)** **63.4 (56.5, 78.5)** **\<0.001** **RV (%pred)** **83.0 (80.5, 88.0)** **119.0 (90.3, 126.5)** **54.9 (43.1, 67.4)** **57.4 (47.9, 80.5)** **\<0.001** **FRC (%pred)** **83.0 (81.0, 88.5)** **100.0 (83.1, 129.8)** **60.7 (52.9, 77.1)** **67.0 (53.5, 80.0)** **\<0.001** **PO**~**2**~**/FiO**~**2**~ **390.5 (380.9, 392.9)** **333.3 (300.0, 371.4)** **352.4 (321.4, 3711.4)** **285.7 (258.6, 350.0)** **\<0.001** **6MWD (m)** **550.0 (497.5, 551.0)** **455.0 (300.0, 515.0)** **459.5 (330.0, 500.8)** **207.0 (100.0, 424.5)** **\<0.001** **DLA (%)** **N/A** **1.68 (0.6, 2.9)** **1.5 (1.14, 2.2)** **2.4 (1.75, 3.3)** **0.001** **CPI** **20.9 (17.7, 24.9)** **36.5 (21.4, 51.4)** **49.4 (36.6, 60.2)** **57.2 (48.3, 63.4)** **\<0.001** **SP-A (ng/ml)** **180.4 (118.9, 285.2)** **193.9 (105.8, 257.6)** **218.0 (148.5, 293.2)** **251.2 (225.9, 314.1)** **0.006** **SP-B (ng/ml)** 14.3 (6.6, 21.3) 17.6 (7.2, 26.3) 17.9 (10.7, 25.9) 18.4 (11.9, 27.3) 0.353 **SP-C (ng/ml)** 1.4 (0.8, 2.2) 1.3 (0.7, 2.0) 1.2 (0.9, 1.8) 1.1 (0.8, 2.1) 0.925 **SP-D (ng/ml)** **17.0 (7.1, 43.6)** **35.2 (19.1, 70.1)** **259.8 (196.7, 333.2)** **250.7 (181.8, 321.0)** **\<0.001** Abbreviations: BMI: Body Mass Index, FEV~1~: Forced expiratory Volume in one Second, FVC: Forced Exhaled Vital Capacity, DLCO: Diffusing Capacity for Carbon Monoxide, TLC: Total Lung Capacity, FRC: Functional Residual Capacity, RV: Residual Volume, 6MWD: 6 Minute Walking Distance, DLA: Destructed Lung Area, CPI: Composite Physiological Index, SP: Surfactant Protein, IPF: Idiopathic Pulmonary Fibrosis, CPFE: Combined Pulmonary Fibrosis and Emphysema. Bold represent statistically significant differences. Correlations of surfactant proteins and pulmonary function tests {#sec015} ---------------------------------------------------------------- Correlations of surfactant proteins and pulmonary function are shown in [Table 2](#pone.0157789.t002){ref-type="table"}. In patients with fibrosis serum SP-B levels correlated significantly to pulmonary function test results, and specifically to FEV~1~, FVC, DLCO, TLC, and FRC. The CPI presented only a weak correlation with SP-A (p = 0.016; [Fig 3A--3D](#pone.0157789.g003){ref-type="fig"}). {#pone.0157789.g003} 10.1371/journal.pone.0157789.t002 {#pone.0157789.t002g} FEV~1~ FVC FEV~1~/FVC DLCO TLC RV FRC HRCT score 6MWD PO~2~/FIO~2~ CPI -------------------- ------------ ------------ ------------ ------------ ------------ ------------ ------------ ------------ -------- -------------- ----------- **SP-A** **Emphysema only** ** p** 0.838 0.992 0.374 0.322 0.616 0.354 0.747 0.155 0.048 0.426 0.428 ** r** -0.039 -0.002 -0.168 -0.227 -0.099 -0.182 -0.065 0.271 -0.363 -0.156 0.150 **IPF** ** p** 0.258 0.119 0.487 **0.025** 0.162 0.167 0.229 0.967 0.225 0.634 **0.016** ** r** -0.147 -0.202 0.091 **-0.290** -0.183 -0.181 -0.158 -0.005 -0.165 -0.062 **0.307** **CPFE** ** p** 0.338 0.385 0.812 0.215 0.466 0.105 0.071 0.575 0.063 0.349 0.136 ** r** -0.209 -0.190 0.052 -0.269 -0.160 -0.347 -0.383 -0.115 -0.393 -0.188 0.320 **SP-B** **Emphysema only** ** p** 0.082 0.232 0.385 0.483 0.127 0.152 0.346 0.317 0.845 0.269 0.553 ** r** 0.322 0.225 0.164 0.138 0.295 0.278 0.189 -0.189 0.037 0.205 -0.113 **IPF** ** p** **0.005** **0.012** 0.429 **0.012** **0.003** 0.042 **0.020** 0.798 0.192 0.030 0.046 ** r** **-0.358** **-0.318** -0.103 **-0.322** **-0.378** -0.263 **-0.300** 0.034 -0.177 -0.276 0.256 **CPFE** ** p** **0.006** **0.003** **0.010** 0.371 0.248 0.587 0.715 0.447 0.814 0.188 0.057 ** r** **0.538** **0.570** **-0.507** 0.187 0.240 -0.114 -0.077 -0.150 -0.050 0.252 -0.385 **SP-C** **Emphysema only** ** p** 0.536 0.580 0.735 0.707 0.078 0.767 0.117 0.718 0.558 0.689 0.693 ** r** 0.127 0.114 0.070 -0.079 0.359 0.062 0.321 0.074 -0.118 -0.081 -0.081 **IPF** ** p** 0.610 0.337 0.047 0.963 0.298 0.685 0.850 0.847 0.987 0.752 0.675 ** r** 0.071 0.133 0.271 -0.007 0.146 0.057 -0.027 -0.074 -0.002 0.044 -0.058 **CPFE** ** p** 0.161 **0.009** 0.376 0.668 0.684 0.458 0.834 0.572 0.328 0.393 0.567 ** r** 0.326 **0.571** -0.209 -0.102 -0.097 -0.176 -0.050 0.124 0.231 0.187 -0.136 **SP-D** **Emphysema only** ** p** 0.260 0.981 0.051 0.097 0.099 0.071 0.044 0.306 0.245 0.966 0.107 ** r** 0.212 -0.005 0.360 -0.320 -0.318 -0.346 -0.391 0.193 -0.215 0.008 0.300 **IPF** ** p** 0.212 0.072 0.088 0.106 0.033 **0.023** 0.047 0.240 0.908 0.910 0.046 ** r** -0.165 -0.236 0.224 -0.214 -0.281 **-0.299** -0.262 0.158 -0.016 -0.015 0.261 **CPFE** ** p** 0.784 0.575 0.508 0.444 0.243 0.798 0.352 0.208 0.900 0.172 0.156 ** r** -0.058 -0.118 0.139 -0.160 -0.242 -0.054 0.194 0.245 0.027 -0.261 0.292 Significance was considered at the level of p\<0.023, Bold represents statistically significant correlations. Abbreviations: FEV~1~: Forced expiratory Volume in one Second, FVC: Forced Exhaled Vital Capacity, DLCO: Diffusing Capacity for Carbon Monoxide, TLC: Total Lung Capacity, FRC: Functional Residual Capacity, RV: Residual Volume, 6MWD: 6 Minute Walking Distance, HRCT: High Resolution Computed Tomography, CPI: Composite Physiological Index, PO~2~: Partial Pressure of Oxygen, FiO~2~: Fraction of inhaled oxygen, SP: Surfactant Protein, IPF: Idiopathic Pulmonary Fibrosis, CPFE: Combined Pulmonary Fibrosis and Emphysema SP levels as predictors of survival {#sec016} ----------------------------------- There were limited deaths during the 1-year follow-up in this study, including 8 patients with IPF and 1 patient with emphysema, whereas no control subjects and no patients with CPFE died during the 1-year follow up. Therefore, the potential predictive role of SP levels on survival was evaluated only in patients with IPF. We have evaluated the role of SP levels as predictors of survival in our patients using the 75^th^ percentiles levels as cut-off points, specifically SP-A ≥280ng/mL, SP-B ≥26ng/mL, SP-C ≥1.9 ng/mL, and SP-D ≥287ng/mL. In patients with IPF, SP-B levels above the 75^th^ percentile of 26 ng/mL were associated with increased mortality (p = 0.05 Log rank test; [Fig 4](#pone.0157789.g004){ref-type="fig"}). The levels of SP-A, C and D were not related to survival in patients with IPF (p = 0.642, p = 0.632 and p = 0.623 for SP-A, SP-C and SP-D respectively). {#pone.0157789.g004} The effect of smoking on SP levels {#sec017} ---------------------------------- Smoking seemed to influence serum SP-A and SP-D levels only in control subjects. Current smokers had higher serum SP-A and SP-D levels compared to current non-smokers (both never smokers and ex-smokers) \[325.7 (152.6, 357.0) vs 156.3 (104.0, 188.2), p = 0.026\] and \[41.0 (14.3, 110.2) vs 10.4 (5.6, 33.3), p = 0.023 for SP-A and SP-D, respectively\]. No differences were observed according to smoking habit regarding SP-B (p = 0.300) and SP-C (p = 0.508) levels ([Fig 5A--5D](#pone.0157789.g005){ref-type="fig"}). Furthermore, no significant differences of serum levels of any of the four SPs have been found according to current smoking habit in patients with emphysema, fibrosis or CPFE (data not shown). {#pone.0157789.g005} The effect of the extend of fibrosis on SP levels {#sec018} ------------------------------------------------- CPFE patients were divided in two subgroups according to the presence of significant extent of emphysematous lesions: 23 patients were considered to have significant emphysematous lesions (i.e. ≥15% of total lung area) whereas 6 patients had emphysematous lesions \<15%. SP-D levels were significantly higher in CPFE patients with significant extent of emphysema compared to those without significant extent of emphysema \[290.7 (195.5, 334.7) vs 191.5 (37.65, 244.5), p = 0.042\] for CPFE patients with and without significant extent of emphysematous lesions respectively ([Fig 6](#pone.0157789.g006){ref-type="fig"}). SP-A, SP-B and SP-D levels did not differ between these two groups (p = 0.569, p = 0.309, and p = 0.969 respectively). {#pone.0157789.g006} Discussion {#sec019} ========== In this study we have shown that the levels of SP-A and SP-D differ between controls, patients with emphysema, IPF and CPFE, with IPF and CPFE demonstrating the highest values. Serum SP-D levels were also significantly higher in CPFE patients with significant emphysematous component. In patients with IPF serum SP-A, levels significantly correlated also to CPI a composite index of estimation of disease severity and extent. Furthermore, serum SP-A and SP-D levels were increased in healthy smokers compared to non-smokers. Finally, although SP-B did not differ between groups its blood levels in patients with IPF correlated to FEV~1,~ FVC, DLCO, FRC and TLC and patients in the upper quartile of SP-B values were associated with worse survival. To our knowledge, this is the first study examining the possible differences of the levels of the four SPs in these four groups of patients. Regarding SP-A and SP-D, our findings confirm previous studies that have shown increased levels in the circulation \[[@pone.0157789.ref024]\] of patients with fibrosis, although these levels were decreased in the BAL fluid \[[@pone.0157789.ref022], [@pone.0157789.ref047]\] but without a significant association between the circulating levels of these proteins and the extent of fibrosis in radiographic studies, a finding that was also confirmed in our study. Our finding that serum SP-A and SP-D levels are higher in patients with fibrosis (IPF or CPFE) compared to those with only emphysema, suggests that the combination of alveolar parenchymal derangement that characterizes fibrosis with the hyperplasia of type II alveolar cells \[[@pone.0157789.ref048]\], may lead to the translocation of greater amounts of these SPs from the lung parenchyma to the peripheral circulation through the pulmonary vasculature in these patients. However, the fact that patients with CPFE and more extensive emphysematous lesions in HRCT also expressed significantly higher levels of both SP-A and SP-D, might suggest that the emphysematous component on top of the fibrotic process, stretched and overdistended by the nearby fibrotic lesions, also contributes through an alveolar damage to the elevation of SP levels in these patients \[[@pone.0157789.ref049]--[@pone.0157789.ref051]\]. Interestingly, in our study circulating levels of SP-A, correlated to CPI in patients with IPF, CPI being a composite score found to be an important indicator of the total respiratory impairment in IPF and related to the total extent of fibrosis in these patients \[[@pone.0157789.ref037]\]. Although in patients with fibrosis, CPI has been acknowledged as a predictor of survival, this was not the case in patients with CPFE \[[@pone.0157789.ref036]\], which probably explains the lack of any significant correlation of SPs with CPI in this group in our study. Furthermore, in our control group, we have observed that healthy current smokers had higher serum levels of SP-A and SP-D compared to current non-smokers (including both never and ex-smokers). This is in accordance to previous studies showing that smokers had decreased levels of SP-A and SP-D in the bronchoalveolar lavage (BAL) fluid \[[@pone.0157789.ref052]\] although they seemed to express higher circulating levels of the former \[[@pone.0157789.ref053]\]. A possible explanation of these findings was that smokers may present an alveolo-capillary leakage of surfactant proteins into the blood \[[@pone.0157789.ref054]\]. The fact that serum levels of SP-A are higher in current smokers compared to ex-smokers \[[@pone.0157789.ref055]\], also observed in our cohort, might also represent the ongoing lung inflammation caused by cigarette smoke which leads to increased vascular permeability and to SP-A and SP-D translocation from the pulmonary compartment to the circulation \[[@pone.0157789.ref056]\]. It is known that smoking cessation seems to rapidly restore the alveolar-capillary barrier integrity \[[@pone.0157789.ref057]\] and this is probably the reason why healthy ex-smokers presented with similar serum levels of SP-A and SP-D. However, this was not the case in patients with emphysema, fibrosis or CPFE in which circulating SP levels did not differ between current smokers and current non-smokers and it seems that the established lung inflammation in both pulmonary emphysema \[[@pone.0157789.ref058]--[@pone.0157789.ref061]\] and fibrosis \[[@pone.0157789.ref062]--[@pone.0157789.ref065]\] continues despite smoking cessation. Increased serum SP-A and SP-D levels in current smokers might be an early indicator of an ongoing parenchymal damage which might lead to the development of lung disease. Regarding SP-B, although its blood levels did not differ among groups, in patients with fibrosis we found a correlation with both individual indices of disease severity and extent such as FVC and DLCO, and in addition a marginal difference in survival, with patients with SP-B levels at the higher quartile presenting lower survival. This finding is in accordance to previous studies showing the relation of SP-B with the development of interstitial lung disease in both animals \[[@pone.0157789.ref066]\] and humans \[[@pone.0157789.ref067]\]. FVC and DLCO are known to be significant predictors of survival in IPF \[[@pone.0157789.ref068], [@pone.0157789.ref069]\] and this might explain the weak association between serum SP-B levels and mortality in this group of patients in our study. The mortality findings related to the increased SP-B levels, however, need to be further validated in larger studies. Finally regarding SP-C there is some evidence that might play a role in the development of CPFE \[[@pone.0157789.ref030], [@pone.0157789.ref031]\] and patients with lung disease comparable with CPFE were found to have mutations in the SP-C gene \[[@pone.0157789.ref032], [@pone.0157789.ref033]\]. It is important to mention that in our group of patients, serum levels of SP-C did not differ in patients with CPFE and were not associated with any disease characteristic including pulmonary function impairment, and the extent of fibrosis. The lack of such an association might relate to fact that SP-C gene mutations possibly result to functional rather than quantitative alterations of SP-C \[[@pone.0157789.ref070]\]. Our study has significant limitations. First, we have used an observational method for the quantification of fibrosis and/or emphysema in HRCT, instead of dedicated CT software. However, this method presents excellent correlation with densitometry quantitation \[[@pone.0157789.ref039], [@pone.0157789.ref040]\] and can be performed in everyday clinical practice. Second, patients were followed up for only one year which is probably the reason for the lack for any significant associations regarding SPs and survival in patients with emphysema and CPFE. Finally, it is a fact that SP-A and SP-D are additionally produced in several extrapulmonary locations, including the brain, the salivary glands, the heart, the kidneys and the reproductive tract \[[@pone.0157789.ref071]\] and for this reason we cannot confirm that the levels that are measured in serum directly originate from the lung. However, since the greater amounts of all four surfactant proteins are synthesized by alveolar type II cells and SP-A, SP-B and SP-D are also produced by different types of airway cells, including Clara cells and submucosal cells, we can conclude that the major source of serum SP levels comes from the pulmonary epithelium. In conclusion, in this study including controls, emphysema, IPF, and CPFE patients we have shown that serum SP- A and SP-D present significantly higher serum levels where fibrosis exists or coexists while serum SP-A levels are also related to fibrosis severity and extent as reflected by the CPI score of estimation of disease severity. The above findings might suggest that serum SP-A and SP-D levels are related to the type of alveolar damage installed on fibrotic lungs and reflect their overproduction due to hyperplasia of type II alveolar cells, or overleakage in the systemic circulation caused by alveolar basement membrane splintering off and hyperpermeability. In IPF also SP-B relates to survival. Further studies are needed to define the role of the above SPs as markers of disease severity. [^1]: **Competing Interests:**The authors have declared that no competing interests exist. [^2]: Contributed reagents/materials/analysis tools: GP AR AM IT IP SL KC. Wrote the paper: AIP KK EM SP. Performed the samples measurement and results interpretation: AS PK. Performed the statistical analysis: AIP MG.

{#sp1 .238} {#sp2 .239}

Introduction {#s1} ============ Ribosome assembly is an essential process that is tightly connected to cellular growth and proliferation ([@bib93]). In the eukaryotic model organism budding yeast, this universal translating machine is built of two subunits: a large subunit (60S) consisting of three different rRNAs (25S, 5.8S, 5S) and 46 ribosomal proteins (r-proteins) and a small subunit (40S) that contains a single rRNA (18S) and 33 r-proteins ([@bib6]; [@bib44]; [@bib73]). Assembly of the eukaryotic ribosome takes place in three distinct cellular territories: the nucleolus, the nucleoplasm and the cytoplasm ([@bib94]; [@bib29]). RNA polymerase I drives production of the 35S pre-rRNA transcript in the nucleolus, which initiates the assembly process. The emerging 35S pre-rRNA transcript undergoes co-transcriptional modification and processing ([@bib66]; [@bib48]), and associates primarily with 40S subunit r-proteins and ∼50 assembly factors to form the earliest pre-ribosome, the 90S ([@bib14]; [@bib30]; [@bib78]). Cleavage of 35S pre-rRNA releases the pre-40S particle, permitting the remaining pre-rRNA to associate with r-proteins of the 60S subunit and ∼200 additional assembly factors to undergo further maturation and pre-rRNA processing ([@bib21]; [@bib30]; [@bib63]). Nuclear maturation of pre-ribosomal particles also requires the release of assembly factors, a process thought to require \>50 energy consuming enzymes ([@bib87]; [@bib50]). Export competent pre-ribosomal particles are separately transported through nuclear pore complexes (NPCs) into the cytoplasm by multiple export factors. In yeast, export factors include the exportin Xpo1, which recognizes nuclear export sequences (NESs) in a RanGTP-dependent manner, and additional factors ([@bib91]). Export factors bind pre-ribosomal particles and interact simultaneously with FG-repeat nucleoporins lining the NPC channel ([@bib26]; [@bib40]; [@bib65]; [@bib8]; [@bib95], [@bib96]; [@bib32]; [@bib1]; [@bib5]; [@bib23]; [@bib64]). Following export, pre-ribosomal particles undergo final maturation prior to initiating translation. This involves the release of shuttling assembly factors, transport factors, incorporation of the remaining r-proteins and final pre-rRNA processing ([@bib68]; [@bib67]). Within the pre-40S particle, immature 20S pre-rRNA is endonucleolytically cleaved into mature 18S rRNA by the nuclease Nob1 rendering the subunit translation competent ([@bib22]; [@bib52]; [@bib70]). Although, Nob1 is recruited to 40S pre-ribosomes in the nucleus, it is activated in the cytoplasm within an 80S-like pre-ribosomal particle formed upon interaction with a mature 60S subunit ([@bib53]; [@bib89]). Additionally, multiple conserved ATPases Prp43, Rio2, Rli1 and Fap7, the Prp43-activator Pfa1, the kinase Rio1, the assembly factor Ltv1 and the r-protein uS11 (yeast Rps14) are implicated in this cleavage step ([@bib28]; [@bib92]; [@bib36]; [@bib31]; [@bib70]; [@bib89]; [@bib34]). Despite the identification of a plethora of factors and their general order of action, how nuclear and cytoplasmic assembly steps are coordinated remains largely unknown. In addition to the tremendous energy required to assemble ribosomes, this process also accounts for the major proportion of the nucleocytoplasmic transport in a growing yeast cell ([@bib76]; [@bib90]). All mRNAs encoding r-proteins must be exported into the cytoplasm, where translation occurs. Nearly all newly synthesized r-proteins are then imported into the nucleus. In yeast, the importin Kap123 has been shown to be an important mediator of r-protein import, but the related importin Pse1 can functionally substitute Kap123 in vivo ([@bib76]; [@bib80]). Unlike other cargos, r-proteins contain large unstructured regions that form intricate interactions with rRNA within the mature ribosome and are prone to non-specific interactions with nucleic acids, aggregation and proteolytic degradation in their non-assembled state ([@bib37]; [@bib38]; [@bib44]; [@bib73]). In contrast to typical protein transport events, nuclear import of r-proteins and subsequent transfer to the ribosome production site pose logistical challenges. In addition to their transport role, importins have been implicated to chaperone basic r-proteins during their transport to the nucleus ([@bib38]). How these intrinsically unstable and aggregation-prone proteins are targeted to assembling pre-ribosomal particles after dissociating from importins remains unclear. Here, we report the discovery of a carrier Tsr2 that coordinates transfer of the eukaryote specific r-protein eS26 (yeast Rps26; [@bib3]) after nuclear import to the assembling 90S pre-ribosome. Tsr2 extracts eS26 from its importins to terminate its import process. Hereby, we reveal an atypical RanGTP-independent mechanism to dissociate an importin:cargo complex. Tsr2 binds and protects the released eS26 from aggregation and proteolysis thereby ensuring its safe transfer to the 90S pre-ribosome. Our data raise the possibility of a yet unidentified fleet of carriers that securely link the nuclear import machinery with the ribosome assembly pathway. Results {#s2} ======= Tsr2 is required for cytoplasmic processing of 20S pre-rRNA to mature 18S rRNA {#s2-1} ------------------------------------------------------------------------------ Previous genome-wide studies revealed a strong accumulation of immature 20S pre-rRNA in a *TSR2* (20 [S]{.ul} r[R]{.ul}NA accumulation [2]{.ul}) deficient yeast strain (*tsr2Δ*) ([@bib69]). Tsr2 is a conserved 23.7 kDa protein ([Figure 1---figure supplement 1A](#fig1s1){ref-type="fig"}) with no identified structural homologues that could provide clues into its role in 20S pre-rRNA processing. To dissect the function of Tsr2, we generated a conditional mutant in which the endogenous *TSR2* was placed under the control of the *GAL1* promoter (P~*GAL1*~-*TSR2*). On repressive glucose media, Tsr2 protein levels were undetectable and the P~*GAL1*~-*TSR2* strain was severely impaired in growth compared to a wild-type (WT) strain between 20--37°C ([Figure 1A](#fig1){ref-type="fig"}).10.7554/eLife.03473.003Figure 1.Tsr2 is required for cytoplasmic processing of 20S pre-rRNA to mature 18S rRNA, and directly binds eS26.(**A**) Tsr2-TAP, Tsr2-GFP and Tsr2-3xGFP cells are not impaired in growth. Left panel: indicated strains were spotted on glucose containing media in 10-fold dilutions and grown at indicated temperatures for 3--7 days. Right panel: Tsr2 protein levels in whole cell extracts derived from the indicated strains were determined by Western analyses using α-Tsr2 antibodies. Protein levels of Arc1 served as loading control. (**B**) Tsr2 localizes predominantly to the nucleus. The Tsr2-TAP and the Tsr2-GFP strain and the P~*GAL1*~-*TSR2* strain containing a centrometric plasmids encoding Tsr2-3xGFP were grown at 30°C to mid-log phase. Localization of Tsr2-TAP was visualized by indirect immunofluorescence microscopy using polyclonal α-TAP antibody (red). Nuclear and mitochondrial DNA was stained with DAPI (blue). Localization of Tsr2-GFP and Tsr2-3xGFP was analyzed by fluorescence microscopy. Scale bar = 5 µm. (**C**) Tsr2-deficient cells accumulate immature 20S pre-rRNA in the cytoplasm. WT and P~*GAL1*~-*TSR2* cells were grown at 30°C in glucose containing media to mid-log phase. Localization of 20S pre-rRNA was analyzed by FISH using a Cy3-labeled oligonucleotide complementary to the 5′ portion of ITS1 (red). Nuclear and mitochondrial DNA was stained with DAPI (blue). Scale bar = 5 µm. (**D**) Tsr2-depleted cells accumulate 80S-like particles. WT and P~*GAL1*~-*TSR2* cells were grown at 30°C in glucose containing media to mid-log phase. Cell extracts were prepared after cycloheximde treatment to preserve polysomes and subjected to sedimentation centrifugation on 7--50% sucrose gradients. Polysome profiles at OD~254nm~ were recorded and the peaks for 40S and 60S subunits, 80S ribosomes and polysomes are indicated (top panels). The gradients were fractionated and the RNA was extracted, separated on a 2% Agarose gel, stained with GelRed (Biotium, middle panels) and subsequently analyzed by Northern Blotting using probes against indicated rRNAs (bottom panels). Exposure times for phosphoimager screens were 20 min for 25S and 18S rRNA, and 3--4 hr for 20S pre-rRNAs. (**E**) Tsr2 does not co-sediment with 40S subunits. WT cells were grown at 30°C to mid-log phase, extracts were prepared and fractionated as described in (**D**). The polysome profile at OD~254nm~ is shown in the upper panel. The peaks for 40S and 60S subunits, 80S ribosomes and polysomes are indicated. The gradient was fractionated, TCA precipitated and the protein content was assessed by Western analyses using the indicated antibodies. (**F**) Tsr2-TAP co-enriches the r-protein eS26. Tsr2-TAP was isolated by tandem affinity purification and the Calmodulin-eluate was separated by 4--12% gradient SDS-PAGE and analyzed by Silver staining. The indicated proteins were identified by mass spectrometry. (**G**) Tsr2 interacts with eS26 in a yeast two-hybrid assay. Plasmids encoding the indicated *GAL4* DNA-binding domain *(BD)* and *GAL4* activation domain *(AD)* fusion proteins were transformed into the yeast reporter strain NMY32. Transformants were spotted in 10-fold serial dilutions onto SDC-Leu-Trp (-Leu-Trp) or SDC-Ade (−Ade) and incubated at 30°C for 4 days. Growth on SDC-Ade indicates a strong two-hybrid interaction. The SV40 Large T antigen served as negative control for these analyses. (**H**) Tsr2 directly binds eS26 in vitro. GST-Tsr2 was immobilized on Glutathione Sepharose before incubation with an *E. coli* lysate containing recombinant eS26. After incubation, bound proteins were eluted by SDS sample buffer, separated by SDS-PAGE and visualized by Coomassie Blue staining. L = input.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.003](10.7554/eLife.03473.003)10.7554/eLife.03473.004Figure 1---figure supplement 1.Tsr2 and eS26 depletion does not impair pre-40S nuclear export.(**A**) Sequence alignment of Tsr2 from the indicated organisms done by ClustalO ([@bib83]; [@bib84]). Conservation at each position is depicted as a gradient from light blue (50% identity) to dark blue (100% identity). (**B**) Tsr2- and eS26-depletion does not impair pre-40S subunit nuclear export. The indicated strains expressing uS5-GFP were grown in repressive glucose containing liquid media to mid-log phase at 30°C. Localization of uS5-GFP was monitored by fluorescence microscopy. Scale bar = 5 µm. (**C**) Human Tsr2 rescues the slow growth of Tsr2-depleted cells. The P~*GAL1*~-*TSR2* cells transformed with indicated plasmids were spotted in 10-fold dilutions on selective glucose containing plates and grown at indicated temperatures for 3--7 days.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.004](10.7554/eLife.03473.004) Next, we localized Tsr2 using an integrated C-terminal -GFP and -TAP tag at the genomic locus. These cell-biological studies revealed that both fusion proteins predominantly localize to the nucleus ([Figure 1B](#fig1){ref-type="fig"}). A similar location for the Tsr2-3xGFP fusion protein (expressed from a CEN plasmid under its natural promoter and terminator regions) was observed in a Tsr2-depleted strain. The strains expressing the various fusion proteins were not impaired in growth ([Figure 1A](#fig1){ref-type="fig"}) suggesting that addition of the -GFP, -TAP, and -3xGFP tags did not affect Tsr2 function. We conclude that Tsr2 mainly localizes to the nucleus. The location of Tsr2 led us to test whether the accumulation of 20S pre-rRNA in *tsr2Δ* cells [@bib69] is due to impaired nuclear export of pre-40S subunits. To this end, we monitored localization of 40S subunits in Tsr2-depleted cells using the established reporter uS5-GFP (yeast Rps2-GFP; [@bib60]). We used the *yrb2Δ* mutant, which is specifically impaired in pre-40S subunit export, as a control ([@bib61]). As expected, the *yrb2Δ* mutant showed a nuclear accumulation of uS5-GFP, in contrast to WT, which displayed cytoplasmic localization of this reporter ([Figure 1---figure supplement 1B](#fig1s1){ref-type="fig"}). Surprisingly, P~*GAL1*~-*TSR2* cells grown on glucose also showed cytoplasmic uS5-GFP localization ([Figure 1---figure supplement 1B](#fig1s1){ref-type="fig"}), indicating no apparent impairment in nuclear export of pre-40S subunits. The data above raised the possibility that cytoplasmic processing of 20S pre-rRNA is impaired in Tsr2-depleted cells. To this end, we monitored the localization of the 5′ portion of the internal transcribed spacer 1 (ITS1) that is present within immature 20S pre-rRNA, but not in mature 18S rRNA, by fluorescence in situ hybridization (FISH). In a WT strain, due to efficient nuclear export of pre-40S subunits, Cy3-ITS1 (red) is detectable only in the nucleolus ([Figure 1C](#fig1){ref-type="fig"}). After nuclear export, ITS1 is cleaved from 20S pre-rRNA by the endonuclease Nob1 and degraded by the exonuclease Xrn1 ([@bib86]; [@bib61]). Tsr2-depleted cells exhibited strong cytoplasmic accumulation of Cy3-ITS1 ([Figure 1C](#fig1){ref-type="fig"}), indicating that cytoplasmic processing is impaired in these cells. Two studies proposed that 20S pre-rRNA processing occurs within an 80S-like particle formed via interaction between a mature 60S subunit and a pre-40S subunit in the cytoplasm ([@bib53]; [@bib89]). One possibility is that formation of this particle is impaired in Tsr2-depleted cells, thereby indirectly interfering with 20S pre-rRNA processing. To test this, we performed polysome analyses. Cell extracts from WT and Tsr2-depleted cells prepared under polysome preserving conditions were analyzed by sucrose gradient centrifugation. In agreement with a role in the 40S biogenesis pathway, the polysome profile of Tsr2-depleted cell extracts revealed strongly reduced levels of free 40S subunits and polysomes ([Figure 1D](#fig1){ref-type="fig"}, top panel). Northern analyses revealed that mature 25S rRNA and immature 20S pre-rRNA co-peak ([Figure 1D](#fig1){ref-type="fig"}, bottom panel), indicating accumulation of 80S-like particles, similar to the one seen upon Fap7-depletion ([@bib31]; [@bib89]). Thus, pre-40S subunits that are exported into the cytoplasm in Tsr2-depleted cells interact with mature 60S subunits, but fail to undergo 20S pre-rRNA processing. We conclude that Tsr2 is required for cytoplasmic maturation of pre-40S subunits. Tsr2 directly binds the eukaryote specific r-protein eS26 {#s2-2} --------------------------------------------------------- Next, we analyzed the sedimentation behavior of Tsr2 on sucrose density gradients. Cell extracts from WT cells were subjected to polysome analyses. The gradient was fractionated and analyzed by Western analyses. Unexpectedly, Tsr2 did not co-sediment with the 40S peak or with heavier fractions, but was found exclusively in lighter fractions at the top of the gradient ([Figure 1E](#fig1){ref-type="fig"}). These data indicate that Tsr2 does not stably associate with pre-ribosomal particles in the 40S biogenesis pathway. To identify interaction partners of Tsr2, we isolated Tsr2-TAP. In agreement with the sedimentation studies above, Tsr2-TAP did not isolate a pre-40S particle. Instead, Tsr2-TAP co-enriched stoichiometric amounts of the eukaryotic specific r-protein eS26 ([Figure 1F](#fig1){ref-type="fig"}; [@bib69]). Further, yeast two-hybrid analysis revealed a strong interaction between Tsr2 and eS26, as determined by growth on stringent adenine deficient media ([Figure 1G](#fig1){ref-type="fig"}). In vitro binding studies using recombinant proteins showed that eS26 and Tsr2 formed a robust complex ([Figure 1H](#fig1){ref-type="fig"}). We conclude that eS26 directly binds Tsr2. eS26 is required for cytoplasmic processing of 20S pre-rRNA {#s2-3} ----------------------------------------------------------- In budding yeast, two non-essential genes, *RPS26A* and *RPS26B*, encode the r-protein eS26. To investigate the phenotypes of *RPS26* deficiency, we created a conditional double mutant in which the endogenous promoter of *RPS26A* in the *rps26bΔ* strain was replaced with the *GAL1* promoter (P~*GAL1*~-*RPS26A*). Consistent with an essential function of eS26 in yeast, the P~*GAL1*~-*RPS26Arps26bΔ* strain did not grow on repressive glucose containing medium ([Figure 2A](#fig2){ref-type="fig"}). Using this strain, we investigated whether eS26 is required for nuclear export of pre-40S subunits and/or cytoplasmic 20S pre-rRNA processing by monitoring the localization of uS5-GFP and Cy3-ITS1. eS26-depletion did not induce nuclear accumulation of uS5-GFP ([Figure 1---figure supplement 1B](#fig1s1){ref-type="fig"}), indicating no apparent impairment in pre-40S subunit nuclear export. However, these cells showed a strong cytoplasmic accumulation of Cy3-ITS1 ([Figure 2B](#fig2){ref-type="fig"}), indicating impairment in final 20S pre-rRNA processing. Further, polysome analyses of eS26-depleted cell extracts revealed strongly reduced levels of free 40S subunits ([Figure 2C](#fig2){ref-type="fig"}, top panel). Northern analyses revealed that mature 25S rRNA and immature 20S pre-rRNA co-peaked ([Figure 2C](#fig2){ref-type="fig"}, bottom panel), indicating an accumulation of 80S-like particles. Thus, as observed in Tsr2-depleted cells, eS26-depleted cells contain pre-40S subunits that fail to process 20S pre-rRNA in the cytoplasm. Based on these data we conclude that eS26 is required for cytoplasmic maturation of pre-40S subunits.10.7554/eLife.03473.005Figure 2.eS26 is required for cytoplasmic processing of immature 20S pre-rRNA to mature 18S rRNA.(**A**) eS26 is essential for viability in yeast. Left panel: WT, *rps26aΔ, rps26bΔ* and the conditional mutant P~*GAL1*~-*RPS26Arps26bΔ* were spotted in 10-fold dilutions on galactose and repressive glucose containing media and grown at 30°C for 2--4 days. Right panel: protein levels of eS26 in whole cell extracts of indicated strains were determined by Western analyses using α-eS26 antibodies. Arc1 protein levels served as loading control. (**B**) eS26-depleted cells accumulate immature 20S pre-rRNA in the cytoplasm. P~*GAL1*~-*RPS26Arps26bΔ* cells transformed with indicated plasmids were grown in glucose containing liquid media at 37°C to mid-log phase. Localization of 20S pre-rRNA was analyzed by FISH using a Cy3-labeled oligonucleotide complementary to the 5′ portion of ITS1 (red). Nuclear and mitochondrial DNA was stained with DAPI (blue). Scale bar = 5 µm. (**C**) eS26-depleted cells accumulate 80S-like particles. The indicated strains were grown in glucose containing liquid media at 30°C to mid-log phase. Cell extracts were prepared after cycloheximide treatment and subjected to sedimentation centrifugation on 7--50% sucrose density gradients. Polysome profiles were recorded at OD~254nm~ (top panels). The peaks for 40S and 60S subunits, 80S ribosomes and polysomes are indicated. Sucrose gradients were fractionated, the RNA was extracted, separated on a 2% Agarose gel, stained with GelRed (Biotium, middle panels) and subsequently analyzed by Northern blotting using probes against the indicated rRNAs (bottom panels). Exposure times for phosphoimager screens were 20 min for 25S and 18S rRNA, and 3--4 hr for 20S pre-rRNAs.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.005](10.7554/eLife.03473.005) eS26 is recruited to the earliest pre-ribosomal particle, the 90S {#s2-4} ----------------------------------------------------------------- The robust interaction between the predominantly nuclear localized Tsr2 and eS26 prompted us to investigate at which stage eS26 is recruited to pre-40S subunits. To address this, we isolated pre-ribosomal particles at different maturation stages along the 40S biogenesis pathway ([@bib30]; [@bib78]). Noc4-TAP purifies the earliest precursor of the pre-40S subunit, the 90S pre-ribosome; Enp1-TAP purifies both the 90S and early pre-40S subunits; Rio2-TAP purifies a late pre-40S subunit containing immature 20S pre-rRNA; and Asc1-TAP purifies a 40S subunit containing mature 18S rRNA and devoid of late assembly factors ([Figure 3A](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1A](#fig3s1){ref-type="fig"}). Co-enrichment of eS26 with pre-ribosomal particles was assessed by (1) Western analyses using antibodies that recognize eS26 and (2) selected reaction monitoring mass spectrometry (SRM-MS). SRM-MS is a reliable tool that overcomes stochastic under sampling of peptides, a critical deficit in shotgun mass spectrometry which complicates the reproducible detection and precise quantitation of proteins in a complex mixture ([@bib71]). SRM relies on the development of specific mass spectrometric-based assays for every target protein and their subsequent application to the relative or absolute quantification within multiple biological samples. We developed a set of SRM assays that enabled us to simultaneously monitor the co-enrichment of eS26 and different r-proteins: uS7 (Rps5), eS28 (Rps28), eS1 (Rps1) and uS11 (Rps14) ([Figure 3B](#fig3){ref-type="fig"}) with multiple pre-ribosomal particles. Both Western and SRM analyses revealed that eS26 co-enriches efficiently with the earliest ribosomal precursor, the 90S, and different pre-ribosomes along the 40S maturation pathway ([Figure 3A,B](#fig3){ref-type="fig"}). The Western signal for eS26 on the 90S pre-ribosome (Noc4-TAP) is specific since no association was detected with the earliest 60S pre-ribosome (Ssf1-TAP) ([Figure 3---figure supplement 1B](#fig3s1){ref-type="fig"}).10.7554/eLife.03473.006Figure 3.eS26 is incorporated into the earliest pre-ribosome, the 90S.(**A**) eS26 co-enriches with pre-ribosomal particles along the 40S maturation pathway. Pre-ribosomal particles in the 40S maturation pathway were purified using the indicated TAP-tagged baits. Calmodulin-eluates were analyzed by Silver staining and Western analyses using the indicated antibodies. The r-protein uS7 served as loading controls for the TAPs. (**B**) SRM-MS reveals co-enrichment of eS26 with pre-ribosomal particles. Upper panel: the relative abundance of different r-proteins was normalized to uS7 levels in the indicated TAP purifications (three independent biological replicates). The error bars show the standard deviation. Lower panel: the intensity of different transitions (listed in the box) of two specific peptides of eS26 was determined by SRM mass spectrometry in the indicated TAP purifications. (**C**) eS26-GFP accumulates in the nucleus in a *yrb2Δ* strain. Left panel: WT, *rps26aΔ* and *RPS26A-GFP* cells were spotted in 10-fold dilutions and grown at indicated temperatures for 3--7 days. Right panel: WT and *yrb2Δ* cells expressing eS26-GFP were grown in glucose containing liquid media to mid-log phase at 20°C. Localization of eS26-GFP was monitored by fluorescence microscopy. Scale bar = 5 µm.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.006](10.7554/eLife.03473.006)10.7554/eLife.03473.007Figure 3---figure supplement 1.Tsr2 and eS26 protein levels in the indicated TAP strains and levels of 20S pre-rRNA and 18S rRNA in the indicated TAP purified particles.(**A**) Noc4-, Enp1- and Rio2-TAP purify pre-40S subunits containing immature 20S pre-rRNA whereas Asc1-TAP purifies a 40S subunit containing mature 18S rRNA. 1 µg of RNA isolated from the indicated pre-40S TAP-eluates was separated on a 2% Agarose gel and probed against indicated rRNAs by Northern blotting. 1 µg of total RNA extracted from WT cells was used as a control. (**B**) eS26 does not co-enrich with the earliest 60S pre-ribosome. Noc4-TAP, the earliest pre-ribosomal particle and Ssf1-TAP, the earliest pre-ribosome in the 60S maturation pathway were isolated. The Calmodulin eluates were visualized by Silver staining and by Western analyses using the indicated antibodies. The CBP signal served as loading controls for the TAPs. (**C**) Tsr2 and eS26 protein levels in indicated TAP strains (also used in [Figure 3A](#fig3){ref-type="fig"}) are equal to levels in WT cells. Whole cell extracts (WCE) were prepared from the indicated strains and analyzed by Western analyses using antibodies against Tsr2 and eS26. The protein Arc1 served as loading control.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.007](10.7554/eLife.03473.007) To support these biochemical data, we performed a complementary cell-biological experiment. If eS26 were targeted to the 90S pre-ribosome, then impairment in pre-40S subunit export should result in its accumulation in the nucleus. To monitor eS26 localization in vivo, we tagged *RPS26A* with GFP at the C-terminus (eS26-GFP) in WT and *yrb2Δ* cells at the genomic locus. Unlike the *rps26aΔ* mutant, the *RPS26A-GFP* strain was not impaired in growth at 20°C and 37°C indicating that addition of GFP does not impair its function on the 40S subunit ([Figure 3C](#fig3){ref-type="fig"}, left panel). As expected, WT cells displayed a strong cytoplasmic localization of eS26-GFP ([Figure 3C](#fig3){ref-type="fig"}, right panel). In contrast, in *yrb2Δ* cells eS26-GFP accumulated in the nucleus ([Figure 3C](#fig3){ref-type="fig"}, right panel). Together all these data suggest that eS26 is transported to the nucleus for loading on the 90S pre-ribosome. Consistent with the sedimentation studies and direct binding to only eS26 ([Figure 1](#fig1){ref-type="fig"}), Tsr2 did not detectably co-enrich with affinity purified pre-ribosomal particles in the 40S maturation pathway ([Figure 3A](#fig3){ref-type="fig"}). This lack of co-enrichment was not due to altered protein levels in the different TAP strains, since Western analyses of whole cell extracts revealed that Tsr2 was expressed at WT levels ([Figure 3---figure supplement 1C](#fig3s1){ref-type="fig"}). Altogether, these results suggest that there are at least two populations of eS26 in vivo, one bound to ribosomes and another bound to Tsr2. eS26 is imported primarily by Kap123 and Kap104 into the nucleus {#s2-5} ---------------------------------------------------------------- We next investigated how eS26 is imported into the nucleus prior to its incorporation into the 90S pre-ribosome. In yeast, the most abundant importin Kap123 transports various r-proteins into the nucleus ([@bib76]; [@bib80]). However, r-proteins also utilize additional importins, including Pse1, Kap104, Sxm1 and Nmd5 ([@bib76]; [@bib90]). We investigated the interaction between eS26 and all yeast importins in vitro. These studies revealed that the importins Kap123, Kap104 and Pse1 efficiently bound eS26 ([Figure 4A](#fig4){ref-type="fig"}). A very weak interaction was observed between Sxm1, Kap95 and Nmd5 and eS26, and no binding was observed with the remaining importins ([Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}). In contrast, none of the importins bound to either Tsr2 or the Tsr2:eS26 complex ([Figure 4A](#fig4){ref-type="fig"}, [Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}), indicating that eS26 alone specifically interacts with importins.10.7554/eLife.03473.008Figure 4.Kap123, Kap104 and Pse1 transport eS26 to the nucleus.(**A**) eS26, but not Tsr2:eS26 or Tsr2, interacts with Kap123, Kap104 and Pse1. Recombinant, GST-Kap123, GST-Kap104, GST-Pse1 and GST alone were immobilized on Glutathione Sepharose and incubated with purified 3.4 µM Tsr2, 4 µM Tsr2:eS26, or *E. coli* lysate containing ∼4 µM eS26^FLAG^ in PBSKMT combined with competing *E. coli* lysates for 1 hr at 4°C. After washing with PBSKMT, bound proteins were eluted in SDS sample buffer and separated by SDS-PAGE. Proteins were visualized by Coomassie Blue staining or Western analyses using indicated antibodies. L = input. GST-tagged importins are indicated with asterisks. (**B**) Nuclear uptake of GFP-eS26 is impaired in *kap123Δ* and *kap104Δ* mutants. Strains expressing GFP-eS26 were grown in synthetic media at 25°C (*ts*-mutants: *pse1-1* and *kap104Δ*) or 30°C to mid-log phase. *Ts*-mutant strains were then shifted to 37°C for 4 hr and localization of GFP-eS26 was analyzed by fluorescence microscopy. Percentage of cells displaying cytoplasmic mislocalization of the GFP-eS26 fusion is indicated. Scale bar = 5 µm. (**C**) Tsr2-3xGFP is targeted to the nucleus by Kap123. Importin mutant strains expressing Tsr2-3xGFP were grown in synthetic media at 25°C (*ts*-mutants: *pse1-1* and *kap104Δ*) or 30°C to mid-log phase. *Pse1-1* and *kap104Δ* cells were then shifted to 37°C for 4 hr. P~*GAL1*~-*RPS26Arps26bΔ* cells containing Tsr2-3xGFP were grown for 15 hr in glucose containing media. Localization of Tsr2-3xGFP was analyzed by fluorescence microscopy. Scale bar = 5 µm. (**D**) RanGTP (His~6~-Gsp1Q71L-GTP) does not efficiently release eS26 from Kap123 and Pse1. GST-importin:eS26^FLAG^ complexes immobilized on Glutathione Sepharose were incubated with either buffer alone or with 1.5 µM RanGTP or 3 nM 3′-end of 18S rRNA for 1 hr at 4°C. Washing, elution, and visualization were performed as in (**A**). GST-tagged importins are indicated with asterisks. (**E**) Tsr2 efficiently dissociates the Kap123:eS26^FLAG^ complex. The GST-Kap123: eS26^FLAG^ complex immobilized on Glutathione Sepharose was incubated with either buffer alone or with 1.5 µM or 375 nM RanGTP or 1.5 µM or 375 nM Tsr2. Samples were withdrawn at the indicated time points. Washing, elution, and visualization were performed as in (**A**). GST-tagged Kap123 is indicated with an asterisk. (**F**) eS26 stably associates with Tsr2 after its release from Kap123. Left panel indicates the experimental setup as flowchart. Immobilized GST-Kap123:eS26^FLAG^ complex was incubated with 1.5 µM His~6~-Tsr2 or buffer alone. As shown in the flowchart, the supernatant was incubated with Ni-NTA Agarose for 1 hr at 4°C (IP-Sup). Washing, elution, and visualization were performed as in (**A**). GST-tagged Kap123 is indicated with an asterisk.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.008](10.7554/eLife.03473.008)10.7554/eLife.03473.009Figure 4---figure supplement 1.eS26, but not Tsr2:eS26 or Tsr2, interacts with importins.Recombinant GST tagged importins, immobilized on Glutathione Sepharose, were incubated with purified 3.4 µM Tsr2, 4 µM Tsr2:eS26 or *E. coli* lysate containing ∼4 µM eS26^FLAG^ in PBSKMT and competing *E. coli* lysates for 1 hr at 4°C. After washing, bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE, and visualized by either Coomassie Blue staining or Western analyses using indicated antibodies. L = input. GST-tagged importins are indicated with asterisk, His~6~-Srp1 is indicated with a rectangle.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.009](10.7554/eLife.03473.009)10.7554/eLife.03473.010Figure 4---figure supplement 2.GFP-eS26 binds to importins and Tsr2 but is not incorporated into pre-ribosomes.(**A**) Location of N- and C-terminus of eS26 within the mature 40S subunit ([@bib73]). eS26 N-terminus (green) is embedded deeply within the 40S subunit whereas the C-terminus (red) projects away from the body of the 40S subunit. Red letters indicate the 20 C-terminal residues that are not visualized in the structure (**B**) GFP-eS26 is not found in heavier fractions on sucrose gradients. WT lysates and lysates containing GFP-eS26 were subjected to sucrose gradient sedimentation as described in [Figure 1D](#fig1){ref-type="fig"}. The peaks for 40S and 60S subunits, 80S ribosomes and polysomes are indicated. The proteins in the gradient were detected by Western analyses using the indicated antibodies. (**C**) GFP-eS26 binds to Kap123, Kap104 and Pse1. Recombinant GST-Kap123, -Kap104, -Pse1 and GST alone were immobilized on Glutathione Sepharose and then incubated with *E. coli* lysate containing GFP-eS26 for 1 hr at 4°C. Bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE and visualized by Coomassie Blue staining and Western analyses using α-GFP antibody. L = input. (**D**) GFP-eS26 is unable to rescue the lethality of the eS26 deficient strain. The P~*GAL1*~-*RPS26Arps26bΔ* strain transformed with different plasmids encoding eS26 or GFP-eS26 were spotted in 10-fold dilutions on selective glucose containing plates and grown at indicated temperatures for 3--7 days. (**E**) GFP-eS26 and GFP-eS26D33N levels are strongly reduced in Tsr2-depleted cells. Whole cell extracts (WCE) prepared from WT and Tsr2-depleted cells were assessed by Western analyses using antibodies against the indicated proteins. Arc1 protein levels served as loading control. (**F**) Upper panel: the Zn^2+^-binding domain of eS26 is required for efficient nuclear uptake. WT cells expressing GFP-eS26 truncations were grown in synthetic media at 30°C to mid-log phase and the localization of GFP-eS26 truncations was analyzed by fluorescence microscopy. Scale bar = 5 µm. Lower panel: Schematic for the eS26 truncations used for fluorescence microscopy. (**G**) GFP-eS26C77W protein levels are strongly reduced in (WCE) extracts. Whole cell extracts were prepared from P~*GAL1*~-*RPS26Arps26bΔ* cells transformed with plasmids encoding for GFP-eS26 WT and mutant proteins. eS26 protein levels were assessed by Western analyses using α-GFP antibodies. Arc1 served as loading control.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.010](10.7554/eLife.03473.010)10.7554/eLife.03473.011Figure 4---figure supplement 3.Tsr2 efficiently releases the conserved eS26 from importins.(**A**) Left panel: sequence alignment of eS26 from the indicated organisms done by ClustalO ([@bib83]; [@bib84]). Conservation at each position is depicted as a gradient from light blue (50% identity) to dark blue (100% identity). Mutated residues linked to DBA are depicted with orange (Asp33) and green (Cys77) dots. Right panel: location of eS26 within the mature 40S subunit ([@bib73]). eS26 clamps the 3′-end of the mature 18S rRNA at the site where the endonuclease Nob1 cleaves the immature 20S pre-rRNA. Inset depicts the 3′-end portion of 18S rRNA (red) in contact with eS26 (blue). The position of amino acids D33 (orange) and C77 (green) that are mutated in DBA or KFS and the coordinated Zn^2+^ ion (black) are depicted. (**B**) RanGTP and the 3′-end of 18S rRNA cannot dissociate the Kap123:eS26 complex. GST-Kap123:eS26a^FLAG^ complexes, immobilized on Glutathione Sepharose, were incubated with buffer alone or with 1.5 µM RanGTP, 1.5 µM Tsr2, 3 nM of the 3′-end of 18S rRNA or the combination of RanGTP and the 3′ end of 18S rRNA for 1 hr at 4°C. Bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE and visualized by Coomassie Blue staining and Western analyses using α-eS26 antibodies. L = input. GST-tagged importins are indicated with asterisks. (**C**) eS26 stably associates with Tsr2 after its release from Pse1. Immobilized GST-Pse1:eS26^FLAG^ complex was treated with 1.5 µM His~6~-Tsr2 or buffer alone. The supernatant was incubated with Ni-NTA Agarose for 1 hr at 4°C (IP-Sup). Washing, elution, and visualization were performed as in [Figure 4E](#fig4){ref-type="fig"}. GST-tagged Pse1 is indicated with an asterisk. (**D**) RanGTP, but not Tsr2 dissociated the Pse1:Slx9 complex in vitro. Pse1:Slx9 complexes were immobilized on Glutathione Sepharose and incubated with buffer alone or with 1.5 µM RanGTP, 1.5 µM Tsr2 or 3 nM 3′-end of 18S rRNA for 1 hr at 4°C and analyzed as in [Figure 4C](#fig4){ref-type="fig"}. GST-tagged importins are indicated with asterisks. (**E**) Tsr2 efficiently dissociates importin:eS26^FLAG^ complexes. GST-Kap104: eS26^FLAG^ and GST-Pse1:eS26^FLAG^ complexes immobilized on Glutathione Sepharose were incubated with either buffer alone or with 1.5 µM or 375 nM RanGTP or 1.5 µM or 375 nM Tsr2. Samples were withdrawn at the indicated time points (1, 2, 4, 8 min). Washing, elution, and visualization were performed as in [Figure 4A](#fig4){ref-type="fig"}. GST-tagged importins are indicated with asterisks.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.011](10.7554/eLife.03473.011)10.7554/eLife.03473.012Figure 4---figure supplement 4.RanGTP and Tsr2 do not release eS31, eS8 and uS14 from Kap123.GST-Kap123 and GST alone were immobilized on Glutathione Sepharose and incubated with *E. coli* lysate containing ∼4 µM eS14^FLAG^, eS31^FLAG^ or eS8^FLAG^ in PBSKMT combined with competing *E. coli* lysates for 1 hr at 4°C.GST-Kap123:eS14^FLAG^, GST-Kap123:eS31^FLAG^, GST-Kap123:eS8^FLAG^ complexes were incubated with either buffer alone or with 1.5 µM RanGTP or 1.5 µM Tsr2 for 1 hr at 4°C. Bound proteins were eluted in SDS sample buffer and separated by SDS-PAGE. Proteins were visualized by Coomassie Blue staining or Western analyses using α-FLAG-antibodies. L = input. GST-Kap123 is indicated with asterisks.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.012](10.7554/eLife.03473.012) To verify our interaction data in vivo, we monitored nuclear uptake of eS26 in WT cells and in different importin mutants. The r-protein eS26 is assembled into the 90S pre-ribosome and is then rapidly transported to the cytoplasm as part of the 40S pre-ribosome. To investigate eS26 nuclear uptake in vivo we uncoupled its import from its export. Structural analyses of the 40S subunit showed that the N-terminus of eS26 is embedded within the rRNA framework ([Figure 4---figure supplement 2A](#fig4s2){ref-type="fig"}; [@bib73]). We fused GFP to the N-terminus of eS26 with the aim to impair its incorporation into the 90S pre-ribosome. Sucrose gradient analyses showed that GFP-eS26 co-sediments only in lighter fractions at the top of the gradient suggesting that it is not incorporated into pre-ribosomes ([Figure 4---figure supplement 2B](#fig4s2){ref-type="fig"}). In vitro binding studies showed that like eS26, GFP-eS26 interacts with Kap123, Kap104 and Pse1 ([Figure 4---figure supplement 2C](#fig4s2){ref-type="fig"}). Thus, the GFP-eS26 fusion protein is functional to recruit the import machinery, although it does not complement the eS26-depleted strain ([Figure 4---figure supplement 2D](#fig4s2){ref-type="fig"}). Further, GFP-eS26 directly binds Tsr2 ([Figure 4---figure supplement 2C](#fig4s2){ref-type="fig"}) and importantly, like eS26, GFP-eS26 is degraded upon Tsr2-depletion ([Figure 4---figure supplement 2E](#fig4s2){ref-type="fig"}). We exploited the GFP-eS26 fusion protein as a tool to monitor the nuclear uptake of eS26 in different importin mutants. Consistent with in vitro binding assays, nuclear uptake of GFP-eS26 was reduced in *kap123Δ* and *kap104Δ* cells ([Figure 4B](#fig4){ref-type="fig"}), indicating that eS26 import requires these importins. Nuclear localization of GFP-eS26 in the *pse1-1 ts* mutant at restrictive temperature remained unaffected ([Figure 4B](#fig4){ref-type="fig"}) indicating that impairment of this importin alone does not inhibit the nuclear import of eS26. The *pse1-1 kap123Δ* mutant showed only a slight increase in cytoplasmic staining of GFP-eS26 ([Figure 4B](#fig4){ref-type="fig"}). Nuclear import of GFP-eS26 was unaffected in the *kap114Δ sxm1Δ* double mutant and *sxm1Δ kap120Δ nmd5Δ* triple mutant ([Figure 4B](#fig4){ref-type="fig"}). Next, we investigated which region of eS26 contributes to its nuclear uptake. For this, we monitored the localization of different truncated versions of eS26 fused to -GFP at the N-terminus. These cell-biological analyses revealed that the Zn^2+^-binding domain is required for efficient nuclear uptake of eS26 ([Figure 4---figure supplement 2F](#fig4s2){ref-type="fig"}). If eS26 were imported into the nucleus in complex with Tsr2, then we reasoned that depletion of eS26 would induce Tsr2 mislocalization to the cytoplasm. However, localization of Tsr2-3xGFP was not affected upon eS26-depletion ([Figure 4C](#fig4){ref-type="fig"}). These studies together with the observation that Tsr2:eS26 complex is unable to recruit importins argue against the idea that eS26 is transported to the nucleus in complex with Tsr2. We conclude that Kap123 and Kap104 target eS26 to the nucleus and that Tsr2 is not a component of this import complex. Kap123 targets Tsr2 to the nucleus {#s2-6} ---------------------------------- Next, we investigated how Tsr2 is targeted to the nucleus in vivo. For this, we monitored the location of Tsr2-3xGFP in different importin mutants. We found that Tsr2-3xGFP mislocalizes to the cytoplasm in the *kap123Δ* mutant, but not in other importin mutants for e.g. *kap104Δ* and *pse1-1* ([Figure 4C](#fig4){ref-type="fig"}). Thus, Kap123 seems to be the major import receptor for Tsr2. However, we did not observe a direct interaction between Tsr2 and Kap123 or any other importin in vitro ([Figure 4A](#fig4){ref-type="fig"}, [Figure 4---figure supplement 1](#fig4s1){ref-type="fig"}). One possibility could be that import of Tsr2 by Kap123 is regulated by post-translational modification. Alternatively, Tsr2 might be transported into the nucleus via a 'piggy bag' mechanism bound to another yet unknown Kap123 cargo. We can exclude the possibility that eS26 serves as an adaptor to import Tsr2 since (1) Tsr2 does not mislocalize to the cytoplasm in a eS26-depleted strain ([Figure 4C](#fig4){ref-type="fig"}) and (2) in vitro binding assays show that the Tsr2:eS26 complex does not interact with Kap123 ([Figure 4A](#fig4){ref-type="fig"}). Tsr2 dissociates importin:eS26 complexes in a RanGTP-independent manner {#s2-7} ----------------------------------------------------------------------- After transport of an importin:cargo complex into the nucleus, RanGTP binds to the N-terminal region of the importin, triggering cargo release and allowing recycling of the importin to participate in subsequent import cycles ([@bib55]; [@bib10]; [@bib45]). To test whether the release of eS26 from the importins is RanGTP-dependent, we performed in vitro dissociation assays. A pre-formed importin:eS26 complex was incubated with 1.5 µM Gsp1Q71L-GTP (equivalent to the human RanQ69L mutant that cannot efficiently hydrolyze GTP, hereafter Gsp1Q71L-GTP is termed RanGTP; [@bib7]; [@bib58]). Although RanGTP was able to dissociate the Kap104:eS26 complex, and partially dissociate the Pse1:eS26 complex, we did not observe dissociation of the Kap123:eS26 complex even after 1 hr incubation ([Figure 4D](#fig4){ref-type="fig"}). It was reported that both RNA and RanGTP are required to release of the mRNA binding proteins Nab2 and Nab4 from Kap104 and the mRNA export factor Npl3 from Mtr10 ([@bib81]; [@bib54]). Because eS26 directly interacts with the 3′-end of the 18S rRNA ([Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}, right panel), we tested if this region of the 18S rRNA is required to release eS26 from Kap123. However, eS26 remained stably bound to Kap123 in the presence of this RNA, either alone or in combination with RanGTP ([Figure 4D](#fig4){ref-type="fig"}, [Figure 4---figure supplement 3B](#fig4s3){ref-type="fig"}). Since the Tsr2:eS26 complex was unable to interact with importins, we tested whether Tsr2 stimulates the release of eS26 from importins. Surprisingly, Tsr2 alone efficiently removed eS26 from Kap123, Pse1 and Kap104 ([Figure 4---figure supplement 3B,C](#fig4s3){ref-type="fig"} and data not shown). This release was specific, since only RanGTP, but not Tsr2, was able to remove the 40S assembly factor Slx9 ([@bib23]) from the Pse1:Slx9 complex under the same conditions ([Figure 4---figure supplement 3D](#fig4s3){ref-type="fig"}). Moreover, Tsr2 specifically releases eS26 from the importin:eS26 complex, since it did not dissociate other tested importin:r-protein complexes (Kap123:uS14, Kap123:eS31 and Kap123:eS8) ([Figure 4---figure supplement 4](#fig4s4){ref-type="fig"}). Since eS26 was inefficiently removed from the Pse1:eS26 complex after 1 hr incubation with RanGTP ([Figure 4D](#fig4){ref-type="fig"}), we investigated the dissociation kinetics of importin:eS26 complexes in the presence of RanGTP or Tsr2. For this, the importin:eS26 complex was incubated with 1.5 µM of either RanGTP or Tsr2 and the release of eS26 from the importin was monitored over time. We found that the amount of eS26 bound to Kap123, Pse1 or Kap104 was only slightly reduced after 8 min, even though RanGTP was efficiently recruited to the different importins:eS26 complexes ([Figure 4E](#fig4){ref-type="fig"}, left panel and [Figure 4---figure supplement 3E](#fig4s3){ref-type="fig"}, left panel). In contrast, Tsr2 completely removed eS26 from these importins within 1 min incubation ([Figure 4E](#fig4){ref-type="fig"}, right panel and [Figure 4---figure supplement 3E](#fig4s3){ref-type="fig"}, right panel). Notably, even at lower concentrations (375 nM) Tsr2 was able to release eS26 from the importin:eS26 complex ([Figure 4E](#fig4){ref-type="fig"}, [Figure 4---figure supplement 3E](#fig4s3){ref-type="fig"}). Moreover, Tsr2 stably associated with the released eS26 ([Figure 4F](#fig4){ref-type="fig"}, [Figure 4---figure supplement 3C](#fig4s3){ref-type="fig"}). Tsr2 shields eS26 from proteolysis and aggregation, and promotes a safe transfer to the 90S pre-ribosome {#s2-8} -------------------------------------------------------------------------------------------------------- The observation that Tsr2 is able to extract eS26 from importins, prompted us to investigate whether Tsr2 plays a role in the transfer of eS26 to the assembling pre-ribosome. To test this, we isolated Enp1-TAP, which purifies both the 90S pre-ribosome and an early pre-40S subunit, from WT and Tsr2-depleted cells and assessed co-enrichment of eS26 by Western analyses. Consistent with a role for Tsr2 in supplying eS26 to the 90S pre-ribosome, we found that eS26 does not efficiently co-enrich with Enp1-TAP in Tsr2-depleted cells ([Figure 5A](#fig5){ref-type="fig"}). This was specific for eS26, since the recruitment of uS7 and uS3 to Enp1-TAP particles was not affected in these cells ([Figure 5A](#fig5){ref-type="fig"}). This lack of enrichment was due to decreased eS26 protein levels, since Western analyses of whole cell extracts derived from Tsr2-depleted cells revealed strongly reduced eS26 protein levels ([Figure 5B](#fig5){ref-type="fig"}). These data led us to test whether eS26 becomes susceptible to proteolysis in Tsr2-depleted cells. To this end, we monitored eS26 protein levels over time in whole cell extracts after switching the P~*GAL1*~-*TSR2* strain to repressive glucose containing media. These analyses revealed that eS26 protein levels decreased over time upon Tsr2-depletion ([Figure 5C](#fig5){ref-type="fig"}).10.7554/eLife.03473.013Figure 5.Tsr2 shields eS26 from proteolysis and aggregation, and promotes safe transfer to the 90S pre-ribosome.(**A**) Efficient recruitment of eS26 to Enp1-TAP requires Tsr2. Enp1-TAP was isolated from WT and Tsr2-depleted cells. After tandem affinity purification, eluates were separated by 4--12% gradient SDS-PAGE and subjected to Western analyses using indicated antibodies. CBP (α-TAP) levels served as loading control. (**B**) eS26 levels are strongly reduced in Tsr2-depleted cells. Whole cell extracts (WCE) prepared from WT and Tsr2-depleted cells were assessed by Western analyses using antibodies against the indicated proteins. Arc1 protein levels served as loading control. (**C**) Tsr2 protects eS26 from proteolysis in vivo. The conditional mutant strain P~*GAL1*~-*TSR2* growing on galactose medium was transferred to repressive glucose containing liquid media at 30°C. Cells were withdrawn at the indicated time points and whole cell extracts were prepared. Western analyses were performed to determine the levels of the indicated proteins. Arc1 served as loading control. (**D**) Tsr2 prevents aggregation of recombinant eS26 in vitro. The aggregation assay was performed in a 384-well plate. In each well 33 µM GST-eS26 and a given concentration of Tsr2 (0 up to 266 µM) in PBSKMT was pre-incubated for 1 hr at 4°C (final volume: 90 µl). 250 nM of PreScission protease was added to initiate aggregation. After 1 hr of incubation, the scattering signal of the aggregated eS26 was monitored by a 384-well plate reader by measuring the intensity at 450 nm (Y-axes). Concentration of Tsr2 used in the assay (X-axes) are expressed as a molar ratio of eS26:Tsr2. Four replicates for each well were measured. The error bars show the standard deviation.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.013](10.7554/eLife.03473.013) We observed that purified recombinant eS26 was highly prone to aggregation. Expressing eS26 as a fusion protein with a highly soluble GST tag suppressed its tendency to aggregate. However, removal of the GST tag after cleavage by PreScission protease resulted in immediate aggregation of free eS26, as determined by a massive increase in the light scattering intensity ([Figure 5D](#fig5){ref-type="fig"}). We tested whether Tsr2 could suppress the aggregating ability of recombinant eS26. We treated GST-eS26 with PreScission protease in absence and presence of Tsr2. A concomitant decrease in the light scattering of the reaction mixture was observed ([Figure 5D](#fig5){ref-type="fig"}), as the Tsr2 concentration in the cleavage buffer was increased, indicating aggregation of free eS26 was suppressed. Altogether, these data indicate that Tsr2 protects eS26, and thereby ensures a safe transfer to the 90S pre-ribosome. An eS26 mutant associated with Klippel-Feil syndrome in Diamond-Blackfan anemia patients is impaired in importin binding {#s2-9} ------------------------------------------------------------------------------------------------------------------------ Mutations in r-proteins have been linked to Diamond-Blackfan anemia (DBA), a rare congenital red blood cell aplasia ([@bib19]; [@bib27]; [@bib62]; [@bib18]; [@bib59]; [@bib17]). Several mutations in the start codon of *RPS26*, including two mutations within eS26, D33N and C77W have been linked to DBA ([@bib13]; [@bib9]). Both residues are highly conserved from yeast to humans ([Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}, left panel). The C77W mutation is additionally linked to Klippel-Feil syndrome (KFS), a skeletal developmental disorder in DBA patients ([@bib9]). In order to analyze the phenotypes induced by the D33N and C77W mutations, we introduced the individual mutations into yeast *RPS26A*. First, plasmids encoding DBA-linked mutants were transformed into the P~*GAL1*~-*RPS26Arps26bΔ* strain and growth was analyzed on glucose containing media. Whereas the D33N mutant partially rescued the lethality of the eS26-conditional mutant, the C77W variant did not allow any growth ([Figure 6A](#fig6){ref-type="fig"}). Further, as in the P~*GAL1*~-*TSR2* strain under repressive conditions, both variants resulted in strongly reduced eS26 protein levels ([Figure 6B](#fig6){ref-type="fig"}). Neither strain displayed defects in the nuclear export of pre-40S subunits ([Figure 1---figure supplement 1B](#fig1s1){ref-type="fig"}). As expected, neither variant was able to rescue the 20S pre-rRNA processing defect of eS26-deficient cells, as determined by the strong cytoplasmic localization of Cy3-ITS1 ([Figure 6C](#fig6){ref-type="fig"}). Thus, eS26 mutants linked to DBA are impaired in cytoplasmic processing of 20S pre-rRNA.10.7554/eLife.03473.014Figure 6.The eS26C77W mutant associated with Klippel-Feil syndrome in Diamond-Blackfan anemia patients is impaired in binding importins.(**A**) The DBA linked eS26D33N and eS26C77W mutants are unable to fully rescue the growth defect of eS26-depleted cells. The P~*GAL1*~-*RPS26Arps26bΔ* strain transformed with different plasmids encoding eS26 mutants were spotted in 10-fold dilutions on selective glucose containing plates and grown at indicated temperatures for 3--7 days. Residues mutated in DBA are depicted in [Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}. (**B**) DBA linked mutations cause strongly reduced eS26 protein levels. Whole cell extracts were prepared from P~*GAL1*~-*RPS26Arps26bΔ* cells transformed with indicated plasmids encoding for eS26 WT and mutant proteins. eS26 protein levels were assessed by Western analyses using α-eS26 antibodies. Arc1 served as loading control. (**C**) eS26 mutants linked to DBA accumulate 20S pre-rRNA in the cytoplasm. P~*GAL1*~-*RPS26Arps26bΔ* cells transformed with plasmids encoding for eS26 WT and mutant proteins were grown at 37°C to mid-log phase in glucose containing medium. Localization of 20S pre-rRNA was analyzed by FISH using a Cy3-labeled oligonucleotide complementary to the 5′ portion of ITS1 (red). Nuclear and mitochondrial DNA was stained with DAPI (blue). Scale bar = 5 µm. (**D**) Tsr2 interacts with eS26 mutants linked to DBA. Recombinant GST-Tsr2 was immobilized on Glutathione Sepharose and then incubated with *E. coli* lysates containing eS26a^FLAG^, eS26D33NFLAG or eS26C77WFLAG lysates for 1 hr at 4°C. Bound proteins were eluted by SDS sample buffer, separated by SDS-PAGE and detected by Coomassie Blue staining. L = input. (**E**) eS26C77W is impaired in binding to Kap123, Kap104 and Pse1. Recombinant GST-Kap123, -Kap104, -Pse1 and GST alone were immobilized on Glutathione Sepharose and then incubated with *E. coli* lysate containing eS26^FLAG^, eS26D33NFLAG or eS26C77WFLAG for 1 hr at 4°C. Bound proteins were eluted in SDS sample buffer, separated by SDS-PAGE and visualized by Coomassie Blue staining and Western analyses using α-eS26 antibody. L = input. (**F**) The GFP-eS26D33N fusion protein is efficiently targeted to the nucleus. WT cells expressing GFP-eS26 and GFP-eS26D33N were grown in synthetic media at 30°C to mid-log phase and the localization of GFP-eS26 was analyzed by fluorescence microscopy. Scale bar = 5 µm.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.014](10.7554/eLife.03473.014) We tested whether the identified eS26 binders could interact with D33N and C77W variants in vitro. Pull-down assays demonstrated that both mutant proteins efficiently bound Tsr2 ([Figure 6D](#fig6){ref-type="fig"}), suggesting these mutations do not contribute to the Tsr2:eS26 interaction surface. The eS26D33N mutant efficiently binds to Kap123, Kap104 and Pse1 ([Figure 6E](#fig6){ref-type="fig"}). In agreement with these interaction studies, nuclear uptake of GFP-eS26D33N was not affected ([Figure 6F](#fig6){ref-type="fig"}) and the levels of GFP-eS26D33N were strongly reduced upon Tsr2-depletion ([Figure 4---figure supplement 2E](#fig4s2){ref-type="fig"}). In contrast, the eS26C77W mutant interacted weakly with these importins ([Figure 6E](#fig6){ref-type="fig"}). We were unable to localize GFP-eS26C77W; whole cells extracts revealed that GFP-eS26C77W protein levels were strongly reduced ([Figure 4---figure supplement 2G](#fig4s2){ref-type="fig"}). Discussion {#s3} ========== A growing yeast cell manufactures ∼200,000 ribosomes during one generation time ([@bib93]). This process requires the import of ∼14 million r-proteins into the nucleus through ∼200 NPCs. Such a process entails rapid transport of importin:r-protein complexes into the nucleus, and necessitates an efficient mechanism to dissociate these complexes to terminate the import process. This permits rapid recycling of importins back to the cytoplasm for subsequent rounds of import. Although it is recognized that r-proteins employ multiple import pathways to reach the nuclear compartment, it remains unclear how these intrinsically unstable and aggregation-prone proteins are targeted to the assembling pre-ribosome. It is assumed that, like a typical import cargo, RanGTP releases the r-protein from the importin and the r-protein somehow finds its way to its cognate rRNA site. Here, we reveal that Tsr2 extracts eS26 from importins and ensures its safe transfer to the 90S pre-ribosome. These data implicate an atypical RanGTP-independent mechanism that terminates the import process, and uncovers an unanticipated link between the nuclear import machinery and the ribosome assembly pathway. eS26 is recruited to the 90S pre-ribosome {#s3-1} ----------------------------------------- Using Western analyses and targeted SRM assays, we found that untagged eS26 is recruited to Noc4-TAP and co-enriches with nuclear pre-40S subunits that contain 20S pre-rRNA ([Figure 3A,B](#fig3){ref-type="fig"}, [Figure 3---figure supplement 1A](#fig3s1){ref-type="fig"}). Moreover, eS26-GFP accumulated in the nucleus of *yrb2Δ* cells that are specifically impaired in 40S pre-ribosome export ([Figure 3C](#fig3){ref-type="fig"}). These data suggest that eS26 can be transported to the 90S pre-ribosome. Our findings contrast a previous report wherein a FLAG-tagged eS26 immunoprecipitated mainly 18S rRNA ([@bib24]). In addition, eS26 was not identified in mass spectrometry studies of a pre-40S subunit, and was suggested to replace the assembly factor Pno1 ([@bib88], [@bib89]). However, we found that Pno1-TAP efficiently co-enriched eS26 ([Figure 3A](#fig3){ref-type="fig"}). Further, Tsr2-depletion impaired recruitment of Pno1 to pre-40S subunits, suggesting that eS26 helps to recruit Pno1 ([Figure 5A](#fig5){ref-type="fig"}). eS26-depletion impaired only 20S pre-rRNA processing in the cytoplasm, suggesting that eS26 does not apparently affect 90S assembly per se, but is specifically required for final maturation ([Figure 2B](#fig2){ref-type="fig"}). Although the precise timing of eS26 recruitment remains unclear, based on our data, we propose that it is a late event during 90S assembly. eS26 clamps the 3′-end of mature 18S rRNA ([Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}, right panel; [@bib73]), precisely at the site where the endonuclease Nob1 cleaves the 20S pre-rRNA. Pre-40S subunits that lack eS26 escape nuclear proofreading and are efficiently transported into the cytoplasm. These incompletely assembled pre-40S subunits recruit the endonuclease Nob1 ([Figure 5A](#fig5){ref-type="fig"}) and form 80S-like particles ([Figures 1D and 2C](#fig1 fig2){ref-type="fig"}). However, they fail to process 20S pre-rRNA ([Figures 1C and 2B](#fig1 fig2){ref-type="fig"}), an essential pre-requisite to form a mature 40S subunit. Thus, 20S pre-rRNA within an 80S-like particle becomes an optimal substrate for Nob1 only when the pre-40S subunit has satisfied a checklist that assesses its potential to translate, including the incorporation of eS26. We propose that the cytoplasmic 20S pre-rRNA cleavage functions as one of the checkpoints that prevent incompletely assembled, pre-40S subunits from entering translation. RanGTP-independent dissociation of importin:eS26 complexes by Tsr2 {#s3-2} ------------------------------------------------------------------ eS26 is targeted to the 90S pre-ribosome and therefore must reach the nucleolus. Unlike the Kap104 adaptor Syo1 that co-imports uL18 (yeast Rpl5) and uL5 (yeast Rpl11) ([@bib49]), Tsr2 does not mediate interactions between eS26 and importins. Instead, our data identified Kap123 and Kap104 as the major importins that directly bind and transport eS26 into the nucleus ([Figure 4](#fig4){ref-type="fig"}). Recruitment of RanGTP did not efficiently trigger the dissociation of importin:eS26 complexes ([Figure 4](#fig4){ref-type="fig"}). One possibility could be that eS26 engages in a novel interaction with the importins, thereby delaying its release. Such a delay may ensure the coordinated handover to the next binding factor, Tsr2. Structural analyses of the importin:eS26 complex should provide clues into why eS26 is inefficiently released from importins by RanGTP. In contrast to RanGTP, Tsr2 efficiently removed eS26 from its importins ([Figure 4](#fig4){ref-type="fig"}, [Figure 4---figure supplement 3](#fig4s3){ref-type="fig"}), identifying an atypical RanGTP-independent mechanism to terminate the import cycle. The observation that Tsr2 prevents proteolysis and aggregation of eS26 ([Figure 5](#fig5){ref-type="fig"}) indicates an additional 'private' chaperone function. Thus our study adds Tsr2:eS26 to the growing list of known chaperones:r-proteins pairs (Sqt1:uL16; Rrb1:uL3; Yar1:uS3) required for ribosome assembly ([@bib15]; [@bib35]; [@bib79]; [@bib46]). Tsr2 may prevent eS26 from undergoing non-specific interactions with nucleic acids during its journey towards the 90S pre-ribosome. How eS26 is transferred from Tsr2 to 90S pre-ribosomes remains unclear. It is tempting to speculate that posttranslational modifications and/or energy consuming enzymes couple the extraction of eS26 from Tsr2 and subsequent incorporation. Based on our data, we propose a model in which eS26 is transported to the nuclear compartment predominantly by importins Kap123 and Kap104 ([Figure 7](#fig7){ref-type="fig"}). Inside the nucleus, eS26 is removed from its importins in a RanGTP-independent mechanism mediated by Tsr2. The released eS26 forms a stable complex with Tsr2. After Tsr2:eS26 complex formation, Tsr2 guarantees a safe transfer of eS26 to the 90S pre-ribosome. Although RanGTP is able to inefficiently release eS26 from its importin, failure to immediately bind Tsr2 results in eS26 degradation. Therefore, in absence of Tsr2, only a smaller fraction of eS26 may reach the 90S pre-ribosome, providing a possible explanation as to why Tsr2-deficient cells are severely impaired in growth but are still viable, although the r-protein eS26 is essential. Notably, human Tsr2 can rescue the severe growth defect of the Tsr2-depleted strain ([Figure 1---figure supplement 1C](#fig1s1){ref-type="fig"}), strongly suggesting an evolutionarily conserved role of Tsr2 in 40S assembly.10.7554/eLife.03473.015Figure 7.A model for the transport of eS26 to the 90S pre-ribosome.Newly synthesized eS26 is transported from the cytoplasm into the nucleus by importins. In the nucleus, Tsr2 alone removes eS26 from importins by a RanGTP-independent mechanism. Subsequently, Tsr2 binds the released eS26, protects it from proteolysis and aggregation, and enables safe transfer to the 90S pre-ribosome. If eS26 is released from the importin by RanGTP it may not immediately encounter Tsr2, resulting in a smaller fraction reaching the 90S pre-ribosome. See 'Discussion' for details of the proposed model.**DOI:** [http://dx.doi.org/10.7554/eLife.03473.015](10.7554/eLife.03473.015) Etiology of eS26 mutants linked to Diamond-Blackfan anemia and Klippel-Feil syndrome {#s3-3} ------------------------------------------------------------------------------------ Similar to the Tsr2-depletion, both DBA mutants (eS26D33N and eS26C77W) accumulate 20S pre-rRNA in the cytoplasm ([Figure 6C](#fig6){ref-type="fig"}). The eS26C77W mutant interacted poorly with its import receptors, suggesting that the inability to interact with importins may cause its degradation. Cysteine 77 is one of four conserved cysteines within eS26 that coordinates a Zn^2+^ ion ([Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}, right panel; [@bib73]). Our data raise an intriguing possibility that the NLS within eS26 becomes available to interact with importins only when the Zn^2+^ ion is correctly coordinated. In addition to their transport role, importins may select correctly folded eS26. Notably, the eS26D33N mutant interacted with Kap123, Kap104, Pse1 and Tsr2 in vitro ([Figure 6D,E](#fig6){ref-type="fig"}). We speculate that the in vivo instability of this variant might be due to a failure to incorporate eS26 into the 90S pre-ribosome. Several mutations in eS26 have been linked to DBA, the majority of which are in the start codon, thereby causing eS26 haploinsufficiency ([@bib13]). Notably, eS26 levels are strongly reduced in Tsr2-depleted cells. Interestingly, about half of DBA cases are due to unidentified mutations. Based on these data, we speculate that the *TSR2* gene may be a potential hotspot for DBA. A family of escortins? {#s3-4} ---------------------- More than 20 years ago, a system was envisioned to efficiently transfer r-proteins from the NPCs towards the nucleolus ([@bib77]). Here, we identify Tsr2 as the first component of this transfer system that connects the nuclear import machinery with the ribosome assembly pathway. We propose the term 'escortin' to describe this 'linker' function. Aggregating r-proteins in the nucleolus aggravate the toxicity of a *Caenorhabditis elegans* Huntington disease model and decrease their lifespan [@bib11], emphasizing the importance to safely transfer r-proteins to the assembling pre-ribosomes. Due to their unstable and aggregation-prone nature [@bib47] we envisage an escortin network to securely connect the nuclear import machinery with the ribosome assembly pathway. Intriguingly, like in the case of Kap123:eS26 complex, RanGTP is unable to efficiently release uL14 (human Rpl23a) from importin 7 (RanBP7) ([@bib37]). Moreover, we found that yeast r-proteins (uS14, eS31a and eS8a) bound to Kap123 were not released upon RanGTP treatment ([Figure 4---figure supplement 4](#fig4s4){ref-type="fig"}) suggesting that these r-proteins may require specific escortins for their release. Affinity purifications coupled to mass spectrometry have identified \>200 non-ribosomal factors that are directly involved in ribosome assembly ([@bib4]; [@bib33]; [@bib14]; [@bib21]; [@bib30]; [@bib63]; [@bib78]). However, escortins, which are not stably bound to pre-ribosomal particles, may have escaped identification. Individual subunits/sub-complexes of other macromolecular complexes involved in genome replication, genomic stability and gene expression must be imported into the nucleus prior to their assembly. The fate of these cargoes after being released from importins in the nucleus remains largely unexplored. Many of these components may rely on escortins that will ensure their transfer to their assembly site. Thus, we expect that the list of escortins for ribosome assembly and other biological pathways will expand in the near future. Materials and methods {#s4} ===================== Yeast strains and plasmids {#s4-1} -------------------------- The *Saccharomyces cerevisiae* strains used in this study are listed in [Supplementary file 1A](#SD1-data){ref-type="supplementary-material"}. Genomic disruptions, C-terminal tagging and promoter switches at genomic loci were performed as described previously ([@bib56]; [@bib72]; [@bib39]). Preparation of media, yeast transformations and genetic manipulations were performed according to established procedures. Plasmids used in this study are listed in [Supplementary file 1B](#SD1-data){ref-type="supplementary-material"}. Details of plasmid construction will be provided upon request. All recombinant DNA techniques were performed according to established procedures using *E. coli* XL1 blue cells for cloning and plasmid propagation. Point mutations in *RPS26A* were generated using the QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA). All cloned DNA fragments and mutagenized plasmids were verified by sequencing. Fluorescence in situ hybridization and microscopy {#s4-2} ------------------------------------------------- Localization of 20S pre-rRNA was analyzed using a Cy3-labeled oligonucleotide probe (5′-Cy3-ATG CTC TTG CCA AAA CAA AAA AAT CCA TTT TCA AAA TTA TTA AAT TTC TT-3′) that is complementary to the 5′ portion of ITS1 as previously described ([@bib23]). Pre-40S subunit export, monitored by localization of uS5-GFP and localization of GFP-eS26 was performed as previously described ([@bib23]; [@bib2]). Indirect immunofluorescence using affinity-purified polyclonal antibodies against the TAP-tag (1:1000; Thermo Scientific; Rockford, IL, USA) and staining of the nuclear and mitochondrial DNA with DAPI was performed as described previously ([@bib80]; [@bib85]). Cells were visualized using DM6000B microscope (Leica, Germany) equipped with HCX PL Fluotar 63 × /1.25 NA oil immersion objective (Leica, Solms, Germany). Images were acquired with a fitted digital camera (ORCA-ER; Hamamatsu Photonics, Hamamatsu, SZK, Japan) and Openlab software (Perkin--Elmer, Waltham, MA, USA). Polysome analyses {#s4-3} ----------------- Sedimentation analysis of yeast lysates by sucrose gradient ultracentrifugation was performed as described previously ([@bib42]; [@bib2]). For Western analyses, 500 μl fractions were precipitated by TCA (trichloroacetic acid), washed in acetone, resuspended in 100 μl of onefold SDS sample buffer and separated by SDS-PAGE. Tsr2, eS26 and uL3 were detected by Western analyses. For rRNA analysis, 500 µl fractions were collected and diluted with an equal volume of lysis buffer. RNA was extracted with Phenol-Chlorofom-Isoamylalcohol and precipitated in isopropanol. RNA pellets were washed with 80% ethanol and resuspended in 20 µl water. rRNAs were then separated on a 1.2% Agarose/formaldehyde gel for 1.5 hr at 200 V. For Northern analysis, rRNAs were blotted onto a Hybond-XL (Amersham, Pittsburg, PA, USA) membrane by capillary transfer and probed for 18S (5′-CATGCATGGCTTAATCTTTGAGAC), 20S (5′-GGTTTTAATTGTCCTATAACAAAAGC) and 25S rRNA (5′-TGCCGCTTCACTCGCCGTTAC) using radioactively labeled probes. rRNAs were detected using phosphoimaging screens (GE Healthcare, Pittsburg, PA, USA). Tandem affinity purifications (TAPs) and Western analyses {#s4-4} --------------------------------------------------------- Whole cell extracts were prepared by alkaline lysis of yeast cells as previously described ([@bib42]). Tandem affinity purifications (TAP) of pre-ribosomal particles were carried out as previously described ([@bib23]; [@bib2]). Calmodulin-eluates were separated on NuPAGE 4--12% Bis-Tris gradient gels (Invitrogen, Carlsbad, CA, USA) and visualized by either Silver staining or Western analyses using indicated antibodies. To analyze RNAs after TAP purification, RNA was extracted with Phenol-Chlorofom-Isoamylalcohol from Calmodulin-eluates and precipitated in isopropanol. RNA pellets were washed with 80% ethanol and finally resuspended in 20 µl water. 1 µg of total RNA was separated on a 1.2% Agarose/formaldehyde gel for 1.5 hr at 200 V. Western analyses were performed as previously described ([@bib42]). The following primary antibodies were used in this study: α-Tsr2/S26 (1:3000; this study), α-Arc1 (1:4000; E Hurt, University of Heidelberg, Heidelberg, Germany), α-uL3 (yeast Rpl3) (1:5000; J Warner, Albert Einstein College of Medicine, Bronx, NY, USA), α-uS7 (yeast Rps5) (1:4000; Proteintech Group Inc., Chicago, IL, USA), α-uS3 (yeast Rps3) (1:3000; M Seedorf, University of Heidelberg, Heidelberg, Germany); α-TAP (CBP) (1:4000; Thermo Scientific, Rockford, IL, USA), α-Pno1 (1:10,000; K Karbstein, Scripps Research Institute, Jupiter, FL, USA), α-Dim1 (1:10,000; K Karbstein, Scripps Research Institute, Jupiter, FL, USA), α-Nob1 (1:500; Proteintech Group Inc., Chicago, IL, USA), α-Tsr1 (1:10,000; K Karbstein, Scripps Research Institute, Jupiter, FL, USA), α-Ltv1 (1:5000; K Karbstein, Scripps Research Institute, Jupiter, FL, USA), α-Rio2 (1:1000; Proteintech Group Inc., Chicago, IL, USA), α-FLAG (1:3000; Sigma-Aldrich, St. Louis, MO, USA). The secondary HRP-conjugated α-rabbit and α-mouse antibodies (Sigma-Aldrich, USA) were used at 1:1000-1:5000 dilutions. Protein signals were visualized using Immun-Star HRP chemiluminescence kit (Bio-Rad Laboratories, Hercules, CA, USA) and captured by Fuji Super RX X-ray films (Fujifilm, Tokyo, Japan). Recombinant protein expression and binding assays {#s4-5} ------------------------------------------------- All recombinant proteins were expressed in *E. coli* BL21 cells by IPTG induction. His~6~-tagged proteins were affinity purified in 50 mM Hepes pH 7.5, 50 mM NaCl, 10% glycerol using Ni-NTA Agarose (GE healthcare), GST fusion proteins were purified in PBSKMT (150 mM NaCl, 25 mM sodium phosphate, 3 mM KCl, 1 mM MgCl2, 0.1% Tween, pH 7.3) using Glutathione Sepharose (GE healthcare). GST-tagged importins, His~6~-taggged importins and RanGTP (His~6~-Gsp1Q71L-GTP) were expressed and purified as previously described ([@bib85]; [@bib58]; [@bib25]). Recombinant GST-Tsr2 was immobilized in PBSKMT on Glutathione Sepharose (GE healthcare), and incubated with *E. coli* lysates containing recombinant eS26, eS26^FLAG^, eS26D33NFLAG, eS26C77WFLAG for 1 hr at 4°C. After incubation, the immobilized GST-proteins were washed three times with PBSKMT 4°C. The bound proteins were eluted with LDS. The in vitro binding studies between recombinant eS26^FLAG^, eS26D33NFLAG, eS26C77WFLAG, Tsr2, Tsr2:eS26 complex and yeast importins as GST-fusion proteins were performed as previously described ([@bib85]). 1/5^th^ of the bound proteins and input (eS26, eS26^FLAG^, eS26D33NFLAG, eS26C77WFLAG) were analyzed on a Coomassie Blue stained gel. 1/10^th^ of the bound proteins and 1/1000^th^ of the input was used for Western analyses. To dissociate the GST-importin:eS26^FLAG^ (Kap123, Pse1 and Kap104) complex or GST-Kap123:eS31^FLAG^, GST-Kap123:eS8^FLAG^, GST-Kap123:eS14^FLAG^ complexes pre-immobilized GST-importin:ribosomal protein complexes were incubated with buffer alone or 3 nM of 3′-end of 18S rRNA (only for eS26^FLAG^), 1.5 µM Tsr2, 1.5 µM His~6~-Tsr2 (only eS26^FLAG^) and/or 1.5 µM RanGTP (His~6~-Gsp1Q71L-GTP) for 1 hr at 4°C (protocol modified from [@bib75]). To show that eS26 stably associated with Tsr2 after release from importins, the supernatant of the samples with buffer alone and His~6~-Tsr2 were incubated with Ni-NTA Agarose for 1 hr at 4°C. For dissociation kinetics, 1.5 µM RanGTP (His~6~-Gsp1Q71L-GTP) or Tsr2 were added to pre-immobilized importin:eS26^FLAG^ complexes and samples were withdrawn at 1, 2, 4 and 8 min. Bound proteins were eluted in twofold LDS/SDS-sample buffer by incubating at 70--95°C and separated by SDS-PAGE. Proteins were visualized by Coomassie Blue staining or by Western analyses using antibodies against Tsr2 and eS26. Aggregation assay {#s4-6} ----------------- The aggregation assay was performed in a 384-well plate (Polystyrene, clear bottom, low volume, Corning, USA). In each well 33 µM GST-eS26 and a given concentration of Tsr2 (0 up to 266 µM) in PBSKMT was pre-incubated for 1 hr at 4°C (final volume: 90 µl). 250 nM of PreScission protease was added to initiate aggregation. Aggregation of free eS26 was measured at 450 nm using a Multiskan GO plate reader (Thermo Scientific, USA). As controls, scattering intensities of individual components used in the aggregation assay such as 33 µM of GST-eS26 alone, 266 µM of Tsr2 alone, PreScission protease and buffer were measured. Four replicates were performed for each sample measured. SRM assay development, quantitation and statistical analysis {#s4-7} ------------------------------------------------------------ ### Sample preparation {#s4-7-1} Affinity purified pre-40S particles were processed for mass spectrometric analysis as described earlier ([@bib1]). Affinity-purified protein samples were denatured and cysteine residues were reduced and alkylated. After tryptic digest the peptides were purified with C18 columns. Before mass spectrometric analysis, 11 retention time calibration peptides (iRT peptides, RT-kit WR, Biognosys) were added to every sample at a ratio of 1:20. SRM assay development {#s4-8} --------------------- To develop SRM assays, peptide samples of the affinity purified pre-40S particles were analyzed on a nanoLC 1Dplus system (Eksigent) connected to a TripleTOF 5600+ mass spectrometer (ABSciex). Peptides were separated by reversed-phase liquid chromatography on a 20-cm fused silica microcapillary (75 µm inner diameter, New Objective) packed in-house with 3 µm C18 beads (Magic C18 AQ, 200 Å pore size; Michrom BioResources, Auburn, CA, USA) with a linear gradient from 98% solvent A (98% acetonitrile, 0.1% formic) and 2% solvent B (98% acetonitrile, 0.1% formic acid) to 35% solvent B over 120 min at a flow rate of 300 nl/min. The mass spectrometer was operated in information-dependent acquisition (IDA) mode. MS1 spectra were recorded in the range of 360--1460 m/z for 500 ms. Up to 20 precursor ions with charge state 2--5 were selected for fragmentation and MS2 spectra were recorded in the range of 50--2000 m/z for 150 ms in high sensitivity mode. Selected precursor ions were excluded for 20 s after one occurrence. Raw data files were centroided and converted to mzML format using the ABSciex Data Converter and then converted to mzXML format using ProteoWizard MSConvert ([@bib43]). MS2 spectra were searched with Sorcerer-SEQUEST (SageN Research) against a *S. cerevisiae* protein database (SGD, May 2013) to which the sequences of the 11 spiked-in iRT peptides and various common contaminants were added. Reversed sequences of all proteins were appended to the protein database to assess the number of false positive peptide-spectrum matches ([@bib16]). Tryptic cleavage was defined to occur after lysine and arginine, unless followed by a proline residue, and peptides were allowed to have up to one non-tryptic end and up to two missed cleavages. Cysteine carbamidomethylation was added as static modification and methionine oxidation as variable modification. Precursor mass tolerance was set to 50 ppm. Resulting peptide-spectrum matches were statistically assessed using PeptideProphet and iProphet as part of the TPP ([@bib41]; [@bib12]; [@bib82]). The iProphet output was processed with MAYU ([@bib74]), which has been modified to work with iProphet probabilities. Peptide-spectrum matches were selected at a false discovery rate (FDR) of 0.07% to obtain a protein FDR of 1%. An in-house written script was used to convert all retention times into iRT values ([@bib20]). SpectraST ([@bib51]) was used to generate a consensus spectral library from which the six most intense fragment ions (b- or y-ions) per peptide precursor were selected in Skyline ([@bib57]). The final SRM assays for target proteins and iRT peptides are given in [Supplementary file 2](#SD2-data){ref-type="supplementary-material"}. ### SRM analysis {#s4-8-1} The SRM data was acquired on a TSQ Vantage triple quadrupole mass spectrometer (Thermo Fisher Scientific) coupled to a nanoLC 1Dplus system (Eksigent). Peptides were separated by reversed-phase liquid chromatography on a 10.5-cm fused silica microcapillary (75 µm inner diameter, New Objective) packed in-house with 5 µm C18 beads (Magic C18 AQ, 200 Å pore size; Michrom BioResources) with a linear gradient from 95% solvent A (98% acetonitrile, 0.1% formic) and 5% solvent B (98% acetonitrile, 0.1% formic acid) to 35% solvent B over 35 min at a flow rate of 300 nl/min. The mass spectrometer was operated in positive mode using electrospray ionisation with a voltage of 1400 V. The capillary temperature was set to 280°C and the collision gas pressure to 1.5 mTorr. All transitions were monitored in scheduled mode with a retention time window of ±600 s, a cycle time of 2 s, and a mass window of 0.7 of half-maximum peak width (unit resolution) in Q1 and Q3. The SRM data was analyzed manually in Skyline ([@bib57]). After removing non-detectable peptides and interfered transitions, peptide intensities (sum of integrated transition peak area) were exported for further processing in Excel. All peptides were normalized on the protein abundance of uS7 (Rps5). Peptides of each protein were ranked by their average intensity over all samples and the three most intense peptides were averaged to obtain an abundance value for every protein. The standard deviation was calculated assuming that the values are a sample of the entire population. The SRM data can be viewed in and downloaded from Panorama: <https://daily.panoramaweb.org/labkey/project/Aebersold/schubert/2014_Schuetz_Ribo-40S/begin.view>. Funding Information =================== This paper was supported by the following grants: - Swiss National Science Foundation [FundRef identification ID:](http://www.crossref.org/fundref/)http://dx.doi.org/10.13039/501100001711 to Vikram G Panse. - Swiss Federal Institute of Technology Zurich to Vikram G Panse. - European Research Council [FundRef identification ID:](http://www.crossref.org/fundref/)http://dx.doi.org/10.13039/501100000781 Starting Grant EURIBIO260676 to Vikram G Panse. - Deutsche Forschungsgemeinschaft [FundRef identification ID:](http://www.crossref.org/fundref/)http://dx.doi.org/10.13039/501100001659 to Gabriel Schlenstedt. - Universität des Saarlandes Homburger Forschungsförderung to Gabriel Schlenstedt. - European Research Council [FundRef identification ID:](http://www.crossref.org/fundref/)http://dx.doi.org/10.13039/501100000781 Advanced Grant 233226 to Olga T Schubert. - European Union PRIME-XS 262067 to Olga T Schubert. We are grateful to M Peter, E Michel, G Braus, K Karbstein, M Seedorf for generously sharing plasmids, strains and antibodies. We thank J Ottl, O Esser, D Barlier and R Brunner from the Novartis Institute of Biomedical research, Basel for protein aggregation measurements. We thank S Guthörl for technical assistance. We thank R Aebersold and J Pfannstiel for mass spectrometry, J Matos, C Azzalin, all members of the Panse laboratory and in particular C Weirich for enthusiastic discussions. Additional information ====================== The authors declare that no competing interests exist. SS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article. UF, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article. MA, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article. PN, Acquisition of data, Analysis and interpretation of data, Contributed unpublished essential data or reagents. MG, Acquisition of data, Analysis and interpretation of data, Contributed unpublished essential data or reagents. SC, Acquisition of data, Analysis and interpretation of data, Contributed unpublished essential data or reagents. CP, Conception and design, Acquisition of data, Analysis and interpretation of data. YC, Conception and design, Acquisition of data, Analysis and interpretation of data. OTS, Conception and design, Acquisition of data, Analysis and interpretation of data. GS, Conception and design, Analysis and interpretation of data. VGP, Conception and design, Analysis and interpretation of data, Drafting or revising the article. Additional files ================ 10.7554/eLife.03473.016 ###### (**A**) Yeast strains used in this study. (**B**) Plasmids used in this study. **DOI:** [http://dx.doi.org/10.7554/eLife.03473.016](10.7554/eLife.03473.016) 10.7554/eLife.03473.017 ###### Peptides used for SRM assays. **DOI:** [http://dx.doi.org/10.7554/eLife.03473.017](10.7554/eLife.03473.017) 10.7554/eLife.03473.018 Decision letter Hegde Ramanujan S Reviewing editor MRC Laboratory of Molecular Biology , United Kingdom eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see [review process](http://elifesciences.org/review-process)). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers. Thank you for sending your work entitled "A RanGTP-independent Mechanism Allows Ribosomal Protein Nuclear Import for Ribosome Assembly" for consideration at *eLife*. Your article has been favorably evaluated by Randy Schekman (Senior editor) and 2 reviewers, one of whom is a member of our Board of Reviewing Editors. The Reviewing editor and the other reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission. This manuscript describes the function of the ribosome assembly factor Tsr2 as a chaperone for the essential ribosomal protein eS26 in yeast. Originally identified in a RNA processing screen as needed for 20S rRNA processing, the function of Tsr2 was poorly understood. Through a series of systematic analyses, the authors trace its role to pre-40S maturation and subsequently to ribosomal protein eS26 incorporation. Biochemical experiments illustrated a direct physical interaction between Tsr2 and eS26 that was mutually exclusive of the eS26-importin interaction. Further analysis showed that Tsr2, and not Ran-GTP, efficiently disassembles the eS26-importin complex in a substrate-specific reaction. These findings, together with the observation that eS26 is both aggregation prone and degradation prone in the absence of Tsr2, lead the authors to an interesting model whereby Tsr2 represents a new functional family of proteins that mediate the exchange of ribosomal proteins from importins to the preribosome. The referees agreed that the data are of very high technical quality, represent a large body of work, and are nicely organized and clearly described in the paper. The primary criticism centers around the issue of solidifying the central conclusion that Tsr2 is localized to and acting in the nucleus after eS26 import. Since earlier studies have suggested that eS26 is loaded onto the preribosome in the cytosol, it is worth more rigorously excluding this possibility. In this alternative interpretation of the authors\' data, Tsr2 would bind to eS26 in the cytoplasm to prevent its interaction with karyopherins and to promote its late assembly, possibly in the cytoplasm. Addressing the following points would help discriminate these models. 1\) Tsr2 has previously been reported to be both nuclear and cytoplasmic. The authors monitored the localization of Tsr2 coupled to 3 copies of GFP and expressed on top of WT protein. The higher expression level due to the extra copies from the CEN vector could lead to a mislocalization artifact. The authors should nail down the localization of functional/endogenous Tsr2 by either using an integrated GFP-tagged functional protein or indirect IF using anti-Tsr2. This would strengthen their model regarding its site of action. 2\) It is unclear in [Figure 4B](#fig4){ref-type="fig"} why GFP-eS26 accumulates in the nucleus in WT cells. It should be incorporated into ribosomes and exported. Its accumulation in the nucleus suggests that it may not load efficiently into ribosomes. This is an important point because if the protein is not functional and not incorporated into ribosomes, its localization may be aberrant. The authors should clarify whether GFP-eS26 is functional and incorporated into ribosomes. 3\) If the argument is that free eS26 is unstable when not bound by TSR2 or Kaps, then why do we apparently see a strong eGFP-eS26 signal in the cytoplasm in the absence of Kap123 (assuming identical imaging conditions for all panels). Presumably this is protein that is not bound by a Kap, hence its cytoplasmic localization, and not bound by Tsr2, which should be restricted to the nucleus. The apparent stability seems counter to the model. Again, the knowing the functionality of eGFP-eS26 would help with interpreting this result. 4\) If the D33 and C77 mutant eS26 proteins cannot interact with Kap123, they should remain in the cytoplasm. This should be investigated to correlate a loss of binding in vitro with an in vivo effect. Minor comments: 1\) The title is not a very good description of what is actually discovered. First, it\'s too general given that for now, the Tsr2-eS26 system is unique. Second, it implies that Tsr2 allows nuclear import, but actually, Tsr2 acts after the import has happened. I therefore suggest something more specific such as "Tsr2 captures a ribosomal protein from its nuclear importin to facilitate incorporation into an assembling pre-ribosome". 2\) The data are all quite clear in general, but I was slightly confused by what I\'m looking at with the stained gels of the GST pulldowns. In most cases, the GST tagged protein is a minor species. Can the authors comment on what the other bands are (*E. coli* lysate proteins, impurities of the GST-purified protein, something else)? 3\) Please indicate what the numbers to the side of the images in [Figure 4B](#fig4){ref-type="fig"} mean. Presumably a quantitative assessment of nuclear localization? 4\) The authors may want to explain in one or two sentences what SRM is for the unfamiliar reader. 5\) It appears to my eye in [Figure 4--figure supplement 2](#fig4s2){ref-type="fig"} that 375 nM RanGTP binds importin and releases substrate better than 1.5 uM. Are the panels reversed, or are these differences not significant? 6\) To rule out effects of Kap123 on Tsr2, the authors should report the localization of Tsr2 in kap123Δ cells. 7\) Is it possible that eS26 interaction with Kaps is nonspecific and electrostatic, such that it can be competed off by Tsr2, a true binding partner? 8\) [Figure 6E](#fig6){ref-type="fig"}. Why do the *E. coli* lysates expressing WT and mutant eS26 look so different? Is the C77 mutant less soluble? 9\) The authors should cite other examples of chaperones -- Yar1, for uS3 and Sqt1 for uL16. 10\) [Figure 1](#fig1){ref-type="fig"}. I don\'t think it can be 100% of input loaded on the gel. 11\) [Figure 3B](#fig3){ref-type="fig"}. The data would be more convincing if one of the proteins had shown differential binding. Possibly uS3. 10.7554/eLife.03473.019 Author response We have performed additional experiments and clarified the issues raised by the reviewers to strengthen the conclusion that Tsr2 localizes to the nucleus and functions in targeting eS26 to the 90S pre-ribosome. *The referees agreed that the data are of very high technical quality, represent a large body of work, and are nicely organized and clearly described in the paper. The primary criticism centers around the issue of solidifying the central conclusion that Tsr2 is localized to and acting in the nucleus after eS26 import. Since earlier studies have suggested that eS26 is loaded onto the preribosome in the cytosol, it is worth more rigorously excluding this possibility. In this alternative interpretation of the authors\' data, Tsr2 would bind to eS26 in the cytoplasm to prevent its interaction with karyopherins and to promote its late assembly, possibly in the cytoplasm. Addressing the following points would help discriminate these models*. A\) Where is endogenous Tsr2 localized? We have localized endogenous Tsr2 using an integrated -TAP and -GFP tag at the genomic locus. These studies show that Tsr2-TAP and Tsr2-GFP predominantly localize to the nucleus ([Figure 1B](#fig1){ref-type="fig"}). B\) Is eS26 targeted to the 90S pre-ribosome? Tsr2-TAP co-enriches stoichiometric amounts of eS26 ([Figure 1F](#fig1){ref-type="fig"}). Further, Western analyses using antibodies directed against eS26 and selected reaction monitoring mass spectrometry (SRM-MS) showed that eS26 co-enriches with multiple nucleolar/nuclear pre-ribosomal particles in the 40S maturation pathway ([Figure 3A](#fig3){ref-type="fig"} and [Figure 3B](#fig3){ref-type="fig"}). These data strongly suggest that eS26 can be targeted to the nucleus for incorporation into the 90S pre-ribosome. To strengthen the biochemical data above, we have performed an additional experiment. The ribosomal protein uS5 is transported to the nucleus and then assembled into the 90S pre-ribosome, the precursor for the 40S pre-ribosome. If nuclear export of the 40S pre-ribosome is impaired, uS5 accumulates in the nucleus: for e.g. in the *yrb2Δ* mutant, uS5-GFP accumulates in the nucleus ([Figure1--figure supplement 1B](#fig1s1){ref-type="fig"}). Likewise, if eS26 were loaded on the 90S pre-ribosome in the nucleus, then it should also accumulate in the nucleus in the *yrb2Δ* strain. To test this, we monitored localization of C-terminally GFP tagged eS26 (eS26-GFP) in WT and *yrb2Δ* cells. In WT cells, eS26-GFP localizes to the cytoplasm. In agreement with the nuclear loading model, the *yrb2Δ* mutant accumulates eS26-GFP in the nucleus ([Figure 3C](#fig3){ref-type="fig"}). These cell-biological data support the idea that eS26-GFP is targeted to the 90S pre-ribosome. We have incorporated these findings in the Results Section and included these cell-biological data in [Figure 3C](#fig3){ref-type="fig"}. *1) Tsr2 has previously been reported to be both nuclear and cytoplasmic. The authors monitored the localization of Tsr2 coupled to 3 copies of GFP and expressed on top of WT protein. The higher expression level due to the extra copies from the CEN vector could lead to a mislocalization artifact. The authors should nail down the localization of functional/endogenous Tsr2 by either using an integrated GFP-tagged functional protein or indirect IF using anti-Tsr2. This would strengthen their model regarding its site of action*. Is the nuclear location of the Tsr2-3xGFP fusion in WT cells an artifact? The following observations argue against this possibility: We have localized endogenous Tsr2 using an integrated -TAP and -GFP tag by immunofluorescence. Tsr2-TAP and Tsr2-GFP localize predominantly to the nucleus as determined by the co-localization of the Alexa Fluor 568 (red) and DAPI (blue) signals and GFP fluorescence ([Figure 1B](#fig1){ref-type="fig"}). Tsr2-TAP and Tsr2-GFP strains are not impaired in growth like the Tsr2-depleted strain suggesting that the fusion proteins are functional ([Figure 1A](#fig1){ref-type="fig"}). The Tsr2-3xGFP produced from a CEN plasmid (under its natural promoter and terminator) also localizes predominantly to the nucleus in Tsr2-depleted cells ([Figure 1B](#fig1){ref-type="fig"}) and rescues the severe growth phenotype of a Tsr2-depleted strain indicating it is functional ([Figure 1A](#fig1){ref-type="fig"}). *2) It is unclear in* [*Figure 4B*](#fig4){ref-type="fig"} *why GFP-eS26 accumulates in the nucleus in WT cells. It should be incorporated into ribosomes and exported. Its accumulation in the nucleus suggests that it may not load efficiently into ribosomes. This is an important point because if the protein is not functional and not incorporated into ribosomes, its localization may be aberrant. The authors should clarify whether GFP-eS26 is functional and incorporated into ribosomes*. Different lines of evidence (stated above) support the idea that eS26 is targeted to the 90S pre-ribosome. *In vitro* binding studies revealed that amongst the 11 yeast importins Kap104, Kap123 and Pse1 directly interacted with eS26. To verify these interaction data *in vivo* we have used the GFP-eS26 fusion protein as a tool to monitor the nuclear targeting of eS26. We apologize to the reviewers for not explaining the use of this fusion protein. It is complicated to monitor the nuclear uptake of eS26 *in vivo* (and other ribosomal proteins) since it finally resides in the cytoplasm. One strategy is to uncouple nuclear import of eS26 from its export. This would mean: to allow eS26 import, but selectively impair eS26 recruitment to the 90S pre-ribosome. We took advantage of the crystal structure of 40S subunit ([@bib73], doi: [10.1126/science.1198308](http://dx.doi.org/10.1126/science.1198308)). In contrast to the C-terminus, the N-terminus of eS26 is deeply embedded within the 18S rRNA framework ([Figure 4--figure supplement 2A](#fig4s2){ref-type="fig"}). We therefore fused GFP to the N-terminus of eS26, with the aim of constructing an eS26 that fails to be incorporated into the 90S pre-ribosome. Sucrose gradient analyses showed that GFP-eS26 co-sediments in lighter fractions at the top of the gradient suggesting that it is not incorporated into pre-ribosomes ([Figure 4--figure supplement 2B](#fig4s2){ref-type="fig"}). Binding studies showed that like eS26, GFP-eS26 interacts with Kap123 and Kap104 ([Figure 4--figure supplement 2C](#fig4s2){ref-type="fig"}). Thus, like eS26, GFP-eS26 is functional to recruit specific components of the import machinery. Consistent with these studies, GFP-eS26 is efficiently targeted to the nucleus in a Kap123 and Kap104 dependent manner. Therefore, we do not think the nuclear location of GFP-eS26 is aberrant. Further, GFP-eS26 directly binds Tsr2 ([Figure 4--figure supplement 2C](#fig4s2){ref-type="fig"}) and, like eS26, the stability of the GFP-eS26 depends on Tsr2 ([Figure 4--figure supplement 2E](#fig4s2){ref-type="fig"}), i.e. GFP-eS26 is degraded upon Tsr2-depletion. Although the GFP-eS26 fusion does not complement the eS26-depleted strain ([Figure 4--figure supplement 2D](#fig4s2){ref-type="fig"}), in combination with the biochemical data it serves as a valuable tool to investigate the nuclear uptake of eS26. We have now explained the use of GFP-eS26 fusion protein to monitor its nuclear import and have included these additional data in [Figure 4--figure supplement 2](#fig4s2){ref-type="fig"}. *3) If the argument is that free eS26 is unstable when not bound by TSR2 or Kaps, then why do we apparently see a strong eGFP-eS26 signal in the cytoplasm in the absence of Kap123 (assuming identical imaging conditions for all panels). Presumably this is protein that is not bound by a Kap, hence its cytoplasmic localization, and not bound by Tsr2, which should be restricted to the nucleus. The apparent stability seems counter to the model. Again, the knowing the functionality of eGFP-eS26 would help with interpreting this result*. We have exploited the GFP-eS26 fusion as a proxy to investigate transport of eS26 to the nucleus. Consistent with our binding data, nuclear uptake of GFP-eS26 is impaired in the *kap123Δ* and *kap104Δ* mutant. Western analyses show that the eS26 is stable in the *kap123Δ* and *kap104Δ* mutant ([Author response image 1](#fig8){ref-type="fig"}).Author response image 1. It is unlikely that TSR2 and importins are the only factors that protect eS26 *in vivo* during its journey towards the 90S pre-ribosome. It appears that newly synthesized eS26 after emerging from the translating ribosome is stabilized by yet unknown cytoplasmic factors before being captured by importins. Intriguingly, yeast deficient for the ribosome associated NAC and SSB-RAC chaperone systems that associate with newly synthesized polypeptides accumulate ribosomal protein aggregates (including eS26) ([@bib47], doi: [10.1083/jcb.200910074](http://dx.doi.org/10.1083/jcb.200910074)). We speculate that the NAC and SSB-RAC chaperone systems hold eS26 in the cytoplasm before transport to the nucleus. *4) If the D33 and C77 mutant eS26 proteins cannot interact with Kap123, they should remain in the cytoplasm. This should be investigated to correlate a loss of binding in vitro with an in vivo effect*. The reviewers appear to have inadvertently mistaken our binding analyses for eS26-mutants linked to DBA. The eS26D33N mutant is not impaired in binding importins and Tsr2 ([Figure 6E](#fig6){ref-type="fig"}). As expected, nuclear uptake of GFP-eS26D33N is not impaired ([Figure 6F](#fig6){ref-type="fig"}). Notably, like GFP-eS26, the *in vivo* stability of the nuclear-targeted GFP-eS26D33N also depends on Tsr2 ([Figure 4--figure supplement 2E](#fig4s2){ref-type="fig"}). In contrast, the eS26C77W mutant weakly interacts with importins *in vitro* ([Figure 6E](#fig6){ref-type="fig"}). However, we found that the GFP-eS26C77W is unstable in WT cells ([Figure 4--figure supplement 2G](#fig4s2){ref-type="fig"}), therefore we were unable to localize GFP-eS26C77W mutant *in vivo*. The inability of eS26 to coordinate the Zn^2+^ ion might trigger cytoplasmic degradation of the GFP-eS26C77W fusion protein. As a complementary experiment, we investigated which region within eS26 contributes to its nuclear uptake. For this, we monitored the location of several truncations of eS26 fused to GFP. These studies suggest that the Zn^2+^-binding domain is required for efficient nuclear uptake of eS26 ([Figure 4--figure supplement 2F](#fig4s2){ref-type="fig"}). Minor comments: *1) The title is not a very good description of what is actually discovered. First, it\'s too general given that for now, the Tsr2-eS26 system is unique. Second, it implies that Tsr2 allows nuclear import, but actually, Tsr2 acts after the import has happened. I therefore suggest something more specific such as "Tsr2 captures a ribosomal protein from its nuclear importin to facilitate incorporation into an assembling pre-ribosome"*. Import cargos are ferried to the nucleus by import receptors. Nuclear import of a cargo is completed only after dissociation from the receptor. For typical import cargos, this irreversible step occurs upon binding of RanGTP to the importin:cargo complex in the nucleus. Here, *in vitro* studies revealed that Tsr2, without RanGTP, is able to dissociate eS26 from an importin:eS26 complex to terminate the import process. Therefore, we suggest Tsr2 "allows" RanGTP-independent nuclear import. While, we agree with the reviewer that the Tsr2-eS26 system is unique for now, the finding that three other Kap123:ribosomal protein complexes (Kap123:uS14, Kap123:eS31 and Kap123:eS8) cannot be dissociated by RanGTP ([Figure 4--figure supplement 4](#fig4s4){ref-type="fig"}), implicate additional escortins that target these ribosomal proteins to the 90S pre-ribosome. Therefore, we feel that the title reflects the overall findings of the paper. *2) The data are all quite clear in general, but I was slightly confused by what I\'m looking at with the stained gels of the GST pulldowns. In most cases, the GST tagged protein is a minor species. Can the authors comment on what the other bands are (*E. coli *lysate proteins, impurities of the GST-purified protein, something else)*? Importins are very large proteins (\> 120 kDa) and are quite sensitive to proteolysis during purification and/or binding assays. The additional bands in the Commassie stained gel are C-terminal truncated forms of importins fused to a N-terminal GST tag as judged by a Western against GST (see [Author response image 2](#fig9){ref-type="fig"}, right).Author response image 2. *3) Please indicate what the numbers to the side of the images in* [*Figure 4B*](#fig4){ref-type="fig"} *mean. Presumably a quantitative assessment of nuclear localization*? Indeed, this is a quantitative assessment of the nuclear localization of GFP-eS26. This is now mentioned in the legend of [Figure 4](#fig4){ref-type="fig"}. *4) The authors may want to explain in one or two sentences what SRM is for the unfamiliar reader*. We have explained SRM and its specific use for detecting and quantitating proteins in the text. *5) It appears to my eye in* [*Figure 4--figure supplement 2*](#fig4s2){ref-type="fig"} *that 375 nM RanGTP binds importin and releases substrate better than 1.5 uM. Are the panels reversed, or are these differences not significant*? We have repeated these experiments. As shown in [Figure 4--figure supplement 3E](#fig4s3){ref-type="fig"}, there are no significant differences in recruitment of RanGTP to importins at the different concentrations. *6) To rule out effects of Kap123 on Tsr2, the authors should report the localization of Tsr2 in kap123Δ cells*. Tsr2-3xGFP mislocalizes in *kap123Δ* cells to the cytoplasm, but not in the *pse1-1* or the *kap104Δ* mutant ([Figure 4C](#fig4){ref-type="fig"}). Thus, Kap123 seems to be an import receptor for Tsr2. However, we did not observe a direct interaction between Tsr2 and Kap123 or any other importin *in vitro* ([Figure 4A](#fig4){ref-type="fig"} and [Figure 4--figure supplement 1](#fig4s1){ref-type="fig"}). One possibility could be that import of Tsr2 by Kap123 is regulated by posttranslational modification. Alternatively, Tsr2 may be transported into the nucleus *via* a "piggy bag" mechanism bound to another yet unknown Kap123 cargo. We can exclude the possibility that eS26 serves as an adaptor to import Tsr2 since (1) Tsr2-3xGFP does not mislocalize to the cytoplasm in a eS26-depleted strain ([Figure 4C](#fig4){ref-type="fig"}) and (2) *in vitro* binding assays show that the Tsr2:eS26 complex does not interact with Kap123 ([Figure 4A](#fig4){ref-type="fig"}). These data have been included in Results Section. *7) Is it possible that eS26 interaction with Kaps is nonspecific and electrostatic, such that it can be competed off by Tsr2, a true binding partner*? We do not think that the interaction between the importin and eS26 is a non-specific electrostatic interaction for the following reasons: First, all the binding assays have been performed in presence of competing *E. coli* lysates. Second, only 3 of the 11 tested yeast importins, efficiently bind eS26. Cell-biological studies support the idea that Kap104 and Kap123 function as the major importins for eS26. Finally, a single amino acid change in eS26 (eS26C77W) impairs binding to the importins. *8)* [*Figure 6E*](#fig6){ref-type="fig"}*. Why do the* E. coli *lysates expressing WT and mutant eS26 look so different? Is the C77 mutant less soluble*? We have used *E. coli* lysates that contain equal amounts of eS26^FLAG^ protein in our binding assays. We have repeatedly observed that the eS26C77W mutant (for unknown reasons) is better expressed as compared to the WT and the eS26D33N mutant. For our binding assays, we need to use much lower amounts of lysate that contains the eS26C77W mutant. This is why the load lysates look different. Please compare Coomassie stained gel containing equal amounts of *E. coli* lysate expressing WT eS26 and mutant proteins (see [Author response image 3](#fig10){ref-type="fig"}).Author response image 3. *9) The authors should cite other examples of chaperones -- Yar1, for uS3 and Sqt1 for uL16*. The chaperones Yar1 and Sqt1 are now cited. *10)* [*Figure 1*](#fig1){ref-type="fig"}*. I don\'t think it can be 100% of input loaded on the gel*. Indeed, it not that we have not loaded the entire (100%) of input. We have now clearly explained the load for our binding assay in the Materials and methods . *11)* [*Figure 3B*](#fig3){ref-type="fig"}*. The data would be more convincing if one of the proteins had shown differential binding. Possibly uS3*. We were unable to develop an SRM assay for uS3, since the Triple-Quadruple mass spectrometer in our proteomics facility is booked out until the middle of September. To directly address the concern regarding differential binding, we have performed Western analyses using antibodies directed against uS3 ([Figure 3A](#fig3){ref-type="fig"}). These studies revealed that uS3 co-enriches with Enp1-TAP but is not found on the earliest 90S pre-ribosome (Noc4-TAP) ([Figure 3A](#fig3){ref-type="fig"}).

Prominent cognitive models of insomnia have emphasized the notion that the disorder is in part maintained by an attentional bias for sleep related "threat" cues which may be internal (i.e., bodily sensations) or external (i.e., environmental) in nature (Harvey, [@B24]; Espie et al., [@B17]). To support this proposition, a growing number of studies have examined the presence of a sleep-related attentional bias for words and images using experimental tasks including the dot-probe, flicker, Posner, emotional Stroop, and eye-tracking paradigms (see Harris et al., [@B23] for a review). Many of these studies have provided encouraging evidence for the presence of such a bias in insomnia. However, the evidence base advocating the presence of such a bias remains mixed with a number of studies yielding no statistically significant effects. While a recent review (Harris et al., [@B23]) cautiously suggests biased attention for sleep-related threat information to be a likely feature of insomnia, the authors highlight the need to understand the specificity of this bias and its relationship with mechanisms believed to underpin the disorder (e.g., sleep preoccupation). Furthermore, whilst it is possible that the mixed evidence may stem from methodological differences relating to the task or population used, the possible moderating influence of these factors on the relationship between attentional bias for sleep-related threat information and insomnia have only recently been examined (e.g., Zheng et al., [@B56]). With this in mind, we propose candidate factors that may play a crucial role in addressing moderating questions such as "when," "for whom" and "under which" conditions are sleep-related attentional biases evident in individuals characterized by insomnia. 5HTTLPR Polymorphism and Brain Reactivity {#s1} ========================================= First, we consider the role of the serotonin transporter polymorphism (5HTTLPR). Some authors have demonstrated the 5HTTLPR short allele to be related to an increased risk for insomnia/poor sleep quality relative to the long allele (Deuschle et al., [@B15]; Huang et al., [@B28]), though findings are inconsistent (e.g., Barclay et al., [@B8]). Others have documented that the association between 5HTTLPR polymorphism and risk for insomnia/poor sleep is moderated by life stress (Brummett et al., [@B13]; Huang et al., [@B28]); and recently the short allele has been associated with increased hypothalamic-pituitary-adrenal (HPA) axis reactivity in response to stress (van Dalfsen and Markus, [@B50]). However, this association is moderated by sleep quality: long-long homozygotes experiencing poor sleep quality exhibit heightened stress reactivity relative to short-short homozygotes (van Dalfsen and Markus, [@B50]). Regardless of the mechanisms involved, it appears that 5HTTLPR genotype, sleep quality and stress reactivity are intricately linked, and one possibility is that 5HTTLPR genotype may differentially alter an individual\'s reactivity to stress under different environmental conditions. In the context of attention bias, it is possible that 5HTTLPR genotype differentially affects behavioral outputs, dependent on sleep quality. In other words, it is possible that 5HTTLPR polymorphisms differentially affect biased attention for "threat" relevant stimuli (possibly due to differences in the activity of the HPA axis) in good vs. poor sleepers. In other psychiatric populations (e.g. anxiety/depression), a relationship between the short allele and an attentional bias for emotionally salient words and images has been evidenced (Beevers et al., [@B12]; Fox et al., [@B19]; Perez-Edgar et al., [@B43]). Specifically, this allelic variation has been linked with difficulty disengaging attention away from response to threat (Beevers et al., [@B12]), whereas the long allele has been associated with threat avoidance and increased attention for positively valanced stimuli (Hariri et al., [@B22]; Munafò et al., [@B40]). It would be worthy to determine whether a similar pattern would occur in insomnia patients, or whether the presence of poor sleep in this population reverses the 5HTTLPR association as would be expected in light of van Dalfsen and Markus\'s ([@B50]) finding. Either way, we hypothesize that allelic variation in the 5HTTLPR polymorphism to also modulate the relationship between insomnia and attentional bias. Given that the amygdala is involved in the processing of emotional information, and because one of the roles of serotonin is regulating mood (Ressler and Nemeroff, [@B46]), researchers have examined whether 5HTTLPR genotype differentially affects neural responses of the amygdala to emotional stimuli. In healthy participants, short allele carriers exhibit greater neural activity in the amygdala, and greater connectivity between the amygdala and ventromedial prefrontal cortex, compared to long-long homozygotes when presented with negatively toned stimuli (Pezawas et al., [@B44]; Heinz et al., [@B26]). Given that individuals with insomnia display increased amygdala reactivity in response to sleep-related stimuli (which may be interpreted as negative; Baglioni et al., [@B7]) it is plausible that 5HTTLPR genotype may underlie this neural response, with genetically vulnerable individuals exhibiting greater sensitivity, and possibly greater attention, to potential threats. Valance of Mood State {#s2} ===================== Second, previous research has statistically controlled for trait-like emotional distress characterized by disposition to experience anxiety or depression (Jansson-Fröjmark et al., [@B30]; Akram et al., [@B4]). However, attentional bias to threat is a transient process (MacLeod et al., [@B37]) and emotional tone has differential effects. For example, dysphoric mood is often associated with negative cognitive activity (e.g., catastrophising, worry, rumination) whereas this can be attenuated with enhanced mood (Baglioni et al., [@B6], [@B7]). As such, current mood state should be taken into account in future studies rather than just trait-like characteristics. Indeed, inducing a negative mood state in people with insomnia influences the extent to which an attentional bias to threatening information is displayed (Zheng et al., [@B56]). Specifically, when assigned to a negative mood inducing (i.e., autobiographical recall of poor sleep) or control (i.e., reading recall) condition prior to completing a dot-probe task comprised of images differing in emotional valence and relatedness to sleep (i.e., general threat, sleep positive, sleep negative), an overall attentional bias to all images emerged only amongst individuals with insomnia who were subject to negative mood state induction Despite an absence of evidence for a sleep-related attentional bias in this study, it remains possible that the overall valance of an individual\'s emotional mood state, as well as its intensity, duration and frequency of occurrence, may act as moderating factors. That said, further research is required to clarify this position, and expand on these novel findings by examining the therapeutic potential of a positive mood state induction. Sleep-Related Worry {#s3} =================== Next, we speculate that attentional bias for threatening information might extend particularly to individuals who attribute worry as the main cause of their insomnia. Worry, by definition, is a cognitive process that involves a chain of emotionally negative thoughts about potential future events (Hirsch and Mathews, [@B27]). Whilst worry may be utilized in a productive manner in terms of solving a specific problem, it appears to be unproductive in the context of sleep, as evidenced by the consistent relationship between worry about sleep and longer sleep onset latency (Harvey, [@B24]; Harvey and Greenall, [@B25]; Weise et al., [@B53]). Indeed, research has consistently evidenced strong relationships between a tendency to worry and high levels of insomnia symptoms (Watts et al., [@B52]; Jansson and Linton, [@B29]; Carney et al., [@B14]; O\'Kearney and Pech, [@B41]). In addition, this population displays a greater propensity to worry during the pre-sleep period, than good sleepers (e.g., Lancee et al., [@B33]). Given that by nature, worry is a negative affect-laden process, it is plausible that individuals with insomnia exhibit greater attentional selectivity of emotionally negative information. Therefore, a factor that may moderate this relationship is the degree to which those with insomnia differ in experiencing increased sleep-related worry prior to initiating sleep (e.g., the consequences of not getting enough sleep on functioning; their perceived lack of ability to fix their sleep problem) which is inherently unproductive. Potentially, a population of those with insomnia who excessively worry about the immediate and longer-term consequences of sleeplessness eventually develop an attentional bias to sleep-related negative information due to an increased personal relevance of this type of information (Espie, [@B16]). In contrast, those with insomnia who do not worry specifically about sleep are less likely to develop an attention bias to sleep-related negative information. If true, empirical studies would observe a sleep-related attentional bias only in individuals for whom sleep is a major concern on nights where this more profoundly manifested (e.g., attention to signs of wakefulness such as racing heart beat and racing thoughts). While prior research has examined whether cognitive activity in general and worry differentially predict insomnia (Wicklow and Espie, [@B54]; Harvey and Greenall, [@B25]), to the best of the author\'s knowledge no study thus far has examined the specific possibility that differences in the content and intensity of worries (i.e., worry about sleep vs. worry about other topics) differentially predict attentional bias in insomnia. However, prior research has obtained suggestive evidence consistent with this latter possibility in the context of catastrophic worry and sleep-related threat (Barclay and Gregory, [@B10]; Barclay and Ellis, [@B9]). In particular, Barclay and Gregory ([@B10]) asked poor and good sleepers to: catastrophise (i.e., iterate negative aspects of a problem: Kendall and Ingram, [@B31]) about their sleep and a personal worry; and to iterate a hypothetical topic in a positive manner. Here, whilst poor sleepers catastrophised more for each topic compared to good sleepers, the frequency of catastrophic worry didn\'t vary by topic (i.e., sleep-related worry, personal worry, or hypothetical) for poor sleepers. Additionally, these outcomes were mediated by anxiety. The authors suggest that poor sleepers\' orientation of catastrophic thoughts may not always be sleep-specific; rather related to a perseverative iterative style, fuelled by anxiety. Next, Barclay and Ellis ([@B9]) observed that poor sleepers were slower to detect sleep-related stimuli compared to non-sleep-related negative stimuli, and this pattern was not observed in good sleepers. The authors speculate that the personal relevance of the "threat" differentially impacts speed of response: sleep-related stimuli hindered performance, whereas conversely non-specific threats facilitated performance. Taken together, these findings suggest that cognitive styles (i.e., worry focused) and personal relevance of the sleep-related stimuli may moderate the attentional bias effect and may account for the inconsistencies in the literature. Finally, it is relevant to note that worry and rumination come under the broader cognitive process of "repetitive negative thinking" or "recurrent negative thinking" which is now recognized to be a transdiagnostic cognitive process for anxiety and depressive disorders (Gustavson et al., [@B21]). First, however we propose examining differential aspects of thought and how this is presented in relation to insomnia and then subsequently determining whether these specific insomnia-relevant constructs moderate the relationship between attentional bias and insomnia. In turn, this area could learn whether the "repetitive negative thinking" phenomenon in anxiety and depressive disorders can be extrapolated to insomnia and other sleep disturbances. Misperception of Sleep and Daytime Impairment {#s4} ============================================= It is possible that individuals with insomnia exhibiting a sleep-related attentional bias may be experiencing increased misperception pertaining to their nocturnal sleep and the extent of their daytime impairment resulting from poor sleep. Specifically, it is theorized that increased attention toward sleep-related cues during sleep and the pre-sleep period may distort the distinction between sleep and wakefulness resulting in a misperception of sleep (Harvey, [@B24]). Indeed, it is well-evidenced that some individuals with insomnia often misperceive attributes of sleep: self-reported sleep onset latency is usually overestimated, and total sleep time underestimated, relative to objectively recorded data (e.g., Wicklow and Espie, [@B54]; Tang and Harvey, [@B49]; Van Den Berg et al., [@B51]). This misperception may also extend to daytime impairment and perception of sleep-deficit. Indeed, whilst individuals with insomnia report their perception of daytime cognitive functioning (i.e., attention, working and episodic memory, problem solving) to be in a manner that confirms the presence of a sleep deficit (i.e., an increased daytime impairment: Fortier-Brochu et al., [@B18]), objective performance on neuropsychological measures of such functioning do not always coincide with their perception (Orff et al., [@B42]; Goldman-Mellor et al., [@B20]). In a similar manner, it has also been evidenced that individuals with insomnia consider their facial appearance to appear more physically tired than they actually are (Akram et al., [@B2]). However, not all those with insomnia misperceive attributes of their sleep and/or the true extent to which daytime functioning is impaired due to poor sleep. This suggests these two forms of sleep misperception (about sleep and daytime consequences) may be functionally underpinned by a third factor. One plausible factor the present commentary has alluded to is worry. It is possible that worry moderates the relationship between sleep misperception and objective measures of sleep, such that increased levels of worry serve to perpetuate misperception of sleep and of the extent to which poor sleep impairs daily functioning. In contrast, a reduced level of worry may result in more accurate perceptions of one\'s sleep compared with objective measures. Furthermore, worry and misperception of sleep may facilitate attention toward cues pertaining to nocturnal sleep and daytime performance as a counterproductive form of self-assessment. Here, once attention is placed on a particular cue (e.g., heart rate), it may then be interpreted in a way that confirms the sleep disturbance (e.g., "why is my heart *still* racing" during sleep onset) consequently feeding back to accentuate sleep-related worry (e.g., "If it doesn\'t slow down soon I won\'t be able to sleep") and misperception of sleep attributes (e.g., "I can\'t function at work because I slept poorly last night") in a cyclical nature (Harvey, [@B24]). Symptom Variation {#s5} ================= Several studies investigating the presence of an attention bias in insomnia have found group differences (e.g., insomnia vs. normal-sleepers) in relation to attentional bias outcomes (Spiegelhalder et al., [@B47]; Jansson-Fröjmark et al., [@B30]; Barclay and Ellis, [@B9]; Beattie et al., [@B11]; Akram et al., [@B1],[@B3],[@B4]; Koranyi et al., [@B32]). However, increased insomnia symptom severity or severity of poor sleep quality do not appear to be related to attentional bias outcomes (Spiegelhalder et al., [@B47]; Jansson-Fröjmark et al., [@B30]; Barclay and Ellis, [@B9]; Beattie et al., [@B11]; Akram et al., [@B1],[@B3],[@B4]; Koranyi et al., [@B32]). With that in mind, little variation in yielded effect sizes relating to sleep-related attentional bias in insomnia are reported between clinically diagnosed patients, opportunistic samples of individuals meeting diagnostic criteria, and poor sleepers (Harris et al., [@B23]). That said, Spiegelhalder et al. ([@B48]) demonstrated positive relationships between attentional bias indices and polysomnographically determined total sleep time, sleep efficiency and duration of slow-wave sleep amongst individuals with insomnia when using the dot-probe task. This pattern of findings did not extend to attentional bias when using the Stroop task. Thus, it is possible that objective measures of severity of sleep disturbance may be predictive of attention bias. Future research should examine the potential role of both subjective and objective variation in sleep continuity as a moderating factor of attention bias. Task and Stimuli {#s6} ================ There is mixed evidence concerning the presence of a sleep-related attentional bias in insomnia and these inconsistencies may stem from variation in the methodological approach used. Indeed, when examining group differences (insomnia/poor-sleeper vs. control) in reaction time tasks, Harris et al. ([@B23]) determined the flicker, dot-probe and Posner tasks to demonstrate moderate to large effects sizes. In contrast, the Stroop task appears less sensitive, with two studies (Spiegelhalder et al., [@B47]; Barclay and Ellis, [@B9]) out of five (Lundh et al., [@B35]; Spiegelhalder et al., [@B47], [@B48]; Zhou et al., [@B57]) conducted to date demonstrating an attentional bias in insomnia. Moving forward from reaction time assessments of attentional bias, which can be considered an indirect measure of attention, a number of recent studies have employed eye-tracking paradigms with the aim to examine selective attention in insomnia (Woods et al., [@B55]; Beattie et al., [@B11]; Akram et al., [@B4]). Here, visual attention can be continuously recorded throughout stimuli presentation, providing a more ecological observation of visual and selective attention relative to reaction time measures (Armstrong and Olatunji, [@B5]; Marks et al., [@B38]). Interestingly, using this methodology, only studies using sleep-related and neutral images, rather than words, as part of a free viewing task evidenced increased attention allocated to the spatial location of insomnia salient stimuli (Beattie et al., [@B11]; Akram et al., [@B4]). A final consideration is whether people with insomnia compared to normal sleepers differ in the extent to which they consider sleep-related stimuli as *threatening*. Moreover, it remains unclear whether this *threat* drives attentional biases in insomnia; or whether perceptions of threat are stimulated by monitoring of the external environment for sleep-related cues. This latter explanation would be consistent with the idea that the attentional bias in insomnia represents a *craving* for sleep rather than interpreting sleep-related stimuli as *threatening*. Application and Summary {#s7} ======================= Recent evidence from the anxiety literature shows attentional bias modification (ABM) to be effective in ameliorating disorder consistent symptoms amongst individuals who elicit an attentional bias (MacLeod and Grafton, [@B36]). In poor sleepers, ABM administered immediately prior to bed improved subjective sleep quality and reduced pre-sleep arousal and sleep onset latency across a single sleep episode, relative to alternative nights where a control task was completed (Milkins et al., [@B39]). Expanding on this research, Lancee et al. ([@B34]) evidenced no therapeutic effect of ABM amongst those meeting diagnostic criteria for insomnia. As such, the applicability of ABM to insomnia remains elusive. Therefore, studies assessing the efficacy of ABM for insomnia should incorporate measures to assess factors that potentially moderate not only the relationship between attentional bias and insomnia, but also any therapeutic effect. To that end, an ongoing randomized controlled trial of ABM for insomnia is concurrently examining the role of sleep-related worry and sleep-associated monitoring in the therapeutic potential of ABM (Akram et al., [@B1],[@B3]). One can postulate that further steps in this line of enquiry would be to (1) determine the mechanism of action of ABM, if successful even in a subset of patients; and (2) identify moderators of response to ABM. Price et al. ([@B45]) have highlighted the inter-relatedness of moderators and mediators in the therapeutic potential of ABM in relation to anxiety. In other words, is the mechanism of anxiety symptom reduction following ABM due to the successful reduction of attention bias (mediator), and is that successful reduction contingent on particular moderating factors? The same questions can be posed in the insomnia arena. It is possible that potential moderators of response to ABM may overlap with those that predict the presence of attention bias, but it is also possible that there may be distinct moderating factors to consider. For example, as suggested in relation to anxiety (Price et al., [@B45]), potential moderators of therapeutic response to ABM may be strength of attention bias (i.e., positive responders to ABM may be those exhibiting high attention bias); age; ABM training setting (lab vs. home); and clinician assessed outcomes. These are worthy considerations for future trials examining efficacy of ABM in insomnia. A recent review tentatively supports the notion of a sleep-related attentional bias in insomnia based on six out of nine studies which confirm group differences in relation to attentional allocation to sleep-related stimuli (Harris et al., [@B23]). However, the number of studies conducted to date still remains limited. Whilst it is possible that a publication bias exists precluding studies demonstrating null effects, to the best of our knowledge, only a further seven studies have been conducted (Woods et al., [@B55]; Beattie et al., [@B11]; Akram et al., [@B1],[@B3],[@B4]; Koranyi et al., [@B32]; Zheng et al., [@B56]; Zhou et al., [@B57]), of which four provide additional support (Beattie et al., [@B11]; Akram et al., [@B1],[@B3],[@B4]; Koranyi et al., [@B32]). Therefore, we suggest that further research is required to clarify the presence of a sleep-related attentional bias in insomnia. Additionally, research should pursue the role of potential factors moderating the sleep-related attentional bias/insomnia relationship. In turn, this may allow a particular sub-set of insomnia patients to benefit therapeutically from ABM through appropriate screening. Author Contributions {#s8} ==================== All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication. 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. The authors would like to thank the reviewers, for providing their expert opinion and insight. [^1]: Edited by: Nicola Cellini, Università degli Studi di Padova, Italy [^2]: Reviewed by: Markus Jansson-Fröjmark, Karolinska Institutet (KI), Sweden; Heather Cleland Woods, University of Glasgow, United Kingdom [^3]: This article was submitted to Cognition, a section of the journal Frontiers in Psychology

1. Introduction {#sec1-ijms-17-01682} =============== Due to the increasing trend of life expectancy and awareness towards lifestyle-related diseases, the nutraceutical industry is gaining prominence and has penetrated into consumers' average daily diet. While the term nutraceutical has not been well defined, it generally refers to any food or supplements that have a beneficial nutritional effect. The current market trend has segmented the nutraceutical industry into two major categories, i.e., functional foods and dietary supplements. The global market size is estimated at USD 140.1 billion in 2010 \[[@B1-ijms-17-01682]\]. Among the major global ingredients, vitamin E has market revenue of USD 83.4 million, alongside with omega-3 fatty acids, amino acids, probiotics and soy proteins \[[@B1-ijms-17-01682]\]. While vitamin E is generally referred to α-tocopherol, the role of tocotrienols in human nutrition is frequently underestimated. Vitamin E is a family of compounds consisting of two categories: tocopherols (TP) and tocotrienols (T3). Structurally, TP and T3 share a similar chromanol head. While tocopherols are attached with a saturated tail at the C2 position, tocotrienols have three double bonds in the side chain. Both TP and T3 have four homologs, namely alpha (α), beta (β), gamma (γ) and delta (δ). Nomenclature of the homologs is dependent on the degree and position of methylation at C5 and C7 position at the chromanol head. Although α-TP is widely known for its function in maintaining cardiovascular health, T3 tend to exhibit various health benefits beyond the antioxidant properties. Among the major sources of T3 are palm oil, annatto and rice bran oil \[[@B2-ijms-17-01682]\]. Global trend of nutraceutical ingredients is moving towards disease or condition specific formulations. Among the highly demanded formulations are those targeted for cardiovascular diseases, weight management, cognitive function, and bone/joint health. In this review, the clinical effects of T3 are summarized according to their functionalities. Supplementation of T3 in various population groups was found to show beneficial effects in cardiovascular health, cancer, immune modulation, neuroprotection and skin protection, as depicted in [Figure 1](#ijms-17-01682-f001){ref-type="fig"} and [Figure 2](#ijms-17-01682-f002){ref-type="fig"}. Details of the clinical studies and research outcome will be discussed further in this review. 2. Cardiovascular Health {#sec2-ijms-17-01682} ======================== Association of reduced cardiovascular disease risk with long term vitamin E supplementation was well established among health practitioners despite contradictory findings on high dose α-TP supplementation \[[@B3-ijms-17-01682]\]. While α-TP was found to reduce low density lipoprotein (LDL) oxidation primarily via anti-oxidant pathways \[[@B4-ijms-17-01682]\], T3 were reported to exert their lipid-lowering effect beyond the anti-oxidant properties. *In vitro* study in liver cells and *in vivo* study in guinea pig showed inhibition of liver 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase after T3 treatments, leading to regulation of cholesterol production \[[@B5-ijms-17-01682],[@B6-ijms-17-01682]\]. Since then, a number of human studies was undertaken to determine the effect of T3 on cardiovascular health. Many of these studies investigated lipid profile as primary outcome, including total cholesterol (TC), LDL, high density lipoprotein (HDL) and triglycerides (TG). In trials involving hypercholesterolemic subjects, supplementation of tocotrienol-rich fractions (TRF) ranging from 200 to 300 mg per day resulted in significant decrease in TC and LDL \[[@B7-ijms-17-01682],[@B8-ijms-17-01682],[@B9-ijms-17-01682],[@B10-ijms-17-01682],[@B11-ijms-17-01682]\]. Nevertheless, a dose dependent study of 60 to 240 mg of T3 per day was not able to demonstrate similar lipid lowering effect in 52 hypercholesterolemic subjects \[[@B12-ijms-17-01682]\]. Another study involving 41 subjects of hypercholesterolemic men also did not find significant difference in the lipid profile after six weeks supplementation of TRF at 40 mg four times daily \[[@B13-ijms-17-01682]\]. A possible explanation for the mixed results is insufficient dose that limits the blood concentration of T3 to exert an apparent effect. For example, when given in combination with lovastatin, a low dose of 50 mg TRF along with AHA Step-1 Diet was sufficient to lower the TC and LDL levels \[[@B9-ijms-17-01682]\]. In this crossover study, combination of lovastatin and TRF showed additive effect in suppressing TC and LDL levels compared to either treatment alone. Besides, an observation reported by Qureshi et al. \[[@B7-ijms-17-01682]\] found that the variability in response is likely to be attributed by inherent variation in serum cholesterol levels. Up to three non-respondents were identified out of the total 15 subjects in the study. When studied in a different population, TRF seemed to have less lipid lowering effect in healthy subjects with normal cholesterol levels of TC \< 6.2 mmol/L. Rasool et al. \[[@B14-ijms-17-01682],[@B15-ijms-17-01682]\] and Chin et al. \[[@B16-ijms-17-01682]\] reported the lack of significant changes in TC and LDL after supplementation of up to six months. However, Rasool et al. \[[@B14-ijms-17-01682],[@B15-ijms-17-01682]\] demonstrated improved antioxidant status and arterial compliance upon supplementation with Tocotrienol-rich vitamin E (TRE). In the former study published by Rasool et al. \[[@B14-ijms-17-01682]\], 80, 160 and 320 mg of TRE were supplemented to healthy volunteers. Meanwhile in the follow-up study \[[@B15-ijms-17-01682]\], 50, 100 or 200 mg of self-emulsifying preparations of TRE were given to 36 healthy males for two months. Similarly, Chin et al. \[[@B16-ijms-17-01682]\] reported a decrease in antioxidant enzymes including superoxide dismutase, catalase, and glutathione peroxidase following daily supplementation of 160 mg TRF for six months, especially in the age group above 50 years old. In a single arm study involving healthy subjects, significant reduction in TC was observed but no changes were detected for LDL and HDL after 30 days of supplementation with about 18, 42 and 240 mg of palm oil vitamin E concentrate (Palmvitee) \[[@B17-ijms-17-01682]\]. Several clinical trials were also initiated to investigate the effect of T3 in diseased populations. In a pilot study where patients with type 2 diabetes mellitus were recruited and given 6 mg/kg/day of TRF, reductions in TC, LDL, and TG in addition to plasma glucose and glycated hemoglobin (HbA1c) levels were detected after 60 days \[[@B18-ijms-17-01682]\]. While patients with end stage renal disease are associated with higher risk of cardiac events and accelerated atherosclerosis, a supplementation trial was conducted in 81 chronic hemodialysis patients \[[@B19-ijms-17-01682]\]. Upon 180 mg/day of TRF supplementation for 16 weeks, significant reduction in plasma TG and apolipoprotein A1 were detected while changes in LDL and inflammatory markers were not apparent \[[@B19-ijms-17-01682]\]. Among the few studies involving patients with previous history of cardiovascular events, Tomeo et al. \[[@B20-ijms-17-01682]\] recruited a cohort of subjects with carotid artery atherosclerosis and supplemented the patients with 160--240 mg of TRF for 18 months. Disease progression determined using duplex carotid ultrasonography showed regression in the TRF group at 24% of patients by one category on the apparent degree of stenosis and 4% of patients by two categories, while 24% patients in the placebo group appeared to have exhibited progression of carotid atherosclerosis \[[@B20-ijms-17-01682]\]. On the other hand, the four homologs of tocotrienols (α, β, γ, δ) were reported to have different potencies depending on the outcome measurement. For example, δ- and γ-T3 were more potent in cancer studies while α-T3 was more efficient in neuroprotection \[[@B21-ijms-17-01682],[@B22-ijms-17-01682]\]. When translated into clinical trials, studies conducted using different T3 fractions showed distinctive results compared to TRF. Mustad et al. \[[@B23-ijms-17-01682]\] conducted a study in hypercholesterolemic subjects supplemented with T3 containing high percentage of α- and γ-T3 with a dose equivalent to 200 mg per day. Results on the lipid profiles showed no difference in all lipid markers between intervention groups despite a marked increase in serum tocotrienol concentrations. In contrast, in a trial where subjects were supplemented with γ- and δ-T3, significant reduction in TG and very low density lipoprotein (VLDL) was observed although TC, LDL and HDL did not change much between groups \[[@B24-ijms-17-01682]\]. In addition, when synthetically modified 250 mg/d of purified α-, γ-, or δ-tocotrienyl acetates was supplemented for eight weeks, only α-tocotrienyl acetate provided significant protection to LDL against oxidation. Conversely, the group supplemented with γ-tocotrienyl acetate showed increased levels of TC and LDL measured post-supplementation \[[@B25-ijms-17-01682]\]. One of the limitations in studies using individual T3 fractions is the discrepancy in composition and purity. In most cases, the purity of the T3 fractions was not reported and the clinical effects from minor components such as desmethyl and didesmethyl T3 were conveniently neglected. 3. Cancer {#sec3-ijms-17-01682} ========= Tocotrienols have gained the attention of clinicians by demonstrating compelling anti-cancer activities in pre-clinical research. The underlying mechanism facilitating this effect has been attributed to the anti-proliferative, anti-angiogenic, pro-apoptotic and immune enhancing nature of T3 \[[@B26-ijms-17-01682]\]. The first clinical trial of tocotrienols in breast cancer patients was reported by Nesaretnam and colleagues \[[@B27-ijms-17-01682]\]. In this five-year placebo controlled and double blinded study, 240 women with either Stage I or II estrogen receptor positive breast cancer were assigned to either placebo plus Tamoxifen or the treatment group which consisted of TRF with Tamoxifen. Women in the treatment arm took daily dose of 400 mg TRF. Upon completion of the study, there were six deaths and 20 local or systemic recurrence reported in the placebo arm while the TRF arm had only two deaths and 16 incidence of recurrence. Meanwhile, the five-year breast cancer specific survival was 98.3% in the TRF group versus 95% in the placebo group. Despite the mortality risk being lowered by 60% in the TRF arm, this finding was statistically insignificant due to the limitation encountered with small sample size being inadequate to detect breast cancer specific death between the two groups. All the women well-tolerated the TRF supplementation without any deviation reflected in their liver and blood parameters \[[@B27-ijms-17-01682]\]. Consequently, in a recently published phase 1 dose escalation study, 25 pre-operative patients with pancreatic ductal neoplasia were given Vitamin E δ-Tocotrienol (VEDT) for two weeks prior to surgery \[[@B28-ijms-17-01682]\]. The dose ranged from 200 to 3200 mg daily. Except for one patient with drug-related grade 1 diarrhea, the rest tolerated the treatment well without any dose limiting toxicity. The biological effective dose was assessed based on induction of apoptosis in neoplastic cells. Increased caspase-3 positive cells in the surgically removed tumor cells denote enhanced apoptotic activity and were associated with good biological effect response. The findings from this study showed that dose levels of 200 to 600 mg induced higher percentage of caspase-3 positive cells, which later reduced in the 800 to 3200 mg cohorts. The daily VEDT dose of 800 mg was chosen as statistically significant highest dose level with the biological effect response rate \[[@B28-ijms-17-01682]\]. 4. Immune Modulation {#sec4-ijms-17-01682} ==================== Modulation of the immune system has become an emerging field of interest in recent years due to its multi-targeted effect. One of the earliest clinical trial to study the immunomodulatory effect of T3 was conducted by Radhakrishnan et al. \[[@B29-ijms-17-01682]\]. Healthy subjects supplemented with 200 mg TRF daily were assessed for levels of immune cells including T-helper, cytotoxic T-lymphocytes, CD4:CD8 ratio, B-lymphocytes and natural killer cells. Although the study failed to detect noticeable changes in the immune cells, a follow up study was designed to improve the outcome measurement. Mahalingam et al. \[[@B30-ijms-17-01682]\] investigated the effect of TRF supplementation on immune response to tetanus toxoid vaccination in healthy volunteers. Subjects were given 400 mg TRF per day for two months and received an intramuscular injection of tetanus toxoid vaccine on day 28. The study found marked increment in anti-tetanus toxoid IgG in the TRF group. When challenged by tetanus toxoid, blood leukocytes from the TRF group showed significant higher production of interferon-gamma and interleukin-4 compared to placebo group. Although the effect of TRF might be specific to tetanus toxoid, this study served as an indication that T3 might be beneficial in enhancing immune response in healthy population upon immunogenic challenge. From a different perspective, Jubri et al. \[[@B31-ijms-17-01682]\] investigated the role of TRF supplementation in the immune functions in cigarette smokers. Supplementation of TRF in smokers did not give rise to noticeable changes in total white cells, total lymphocytes, lymphocyte proliferation, and total T-cells, while an increase in the number of B-cells was observed in TRF supplemented non-smokers. Insufficient dose (200 mg per day) and raised basal values due to smoking were postulated to be the cause of unnoticeable changes in the outcome measurements \[[@B31-ijms-17-01682]\]. Nevertheless, the study provided ample information on the distinctive patterns of immune functions between smokers and non-smokers in the Malaysian population. 5. Neuroprotection and Cognitive Function {#sec5-ijms-17-01682} ========================================= Neuroprotective ability conferred by T3 in pre-clinical research has gained momentum and was further explored as a novel approach in humans. Beyond their antioxidative nature, T3, particularly α-T3, were found to induce neuroprotection at nanomolar concentrations. In fact, the circulating plasma T3 were found to be 20 times more than the required dose for neuroprotection \[[@B32-ijms-17-01682]\]. Thereupon, Gopalan et al. \[[@B33-ijms-17-01682]\] investigated the effect of TRF in white matter protection. Volunteers with white matter lesion (WML) were recruited and randomized into placebo or TRF groups. The volunteers in TRF group took 400 mg of TRF daily for two years and at the end of two years, the TRF group was reported to have statistically significant reduction in mean WML volume change as compared to the placebo group. This attenuation of WML and the safety of TRF as a supplement strengthened the neuroprotective ability of T3 and thus pave a path for future studies in larger population \[[@B33-ijms-17-01682]\]. On the other hand, Attention Deficit/Hyperactive Disorder (ADHD) is a common behavioral disorder observed in school-going children. The medications commonly prescribed to reduce the symptoms are reported to cause adverse effects. Since oxidative stress has been identified as a known cause for neuronal damage that is modestly increased in ADHD, the efficacy of TRF, being a natural anti-oxidant, was investigated by Tan and colleagues in children with ADHD \[[@B34-ijms-17-01682]\]. The children were supplemented with either 200 mg TRF or placebo daily for six months and were evaluated by validated questionnaires NICHQ Vanderbilt ADHD Parent Rating Scale (VAPRS) and NICHQ Vanderbilt ADHD Teacher Rating Scale (VATRS). The children were evaluated at baseline, three and six months. Although VAPRS showed improvement in both groups while VATRS showed improvement only in the TRF group, these results were not statistically significant. The limitation of this study might be attributed to supplementation at later age and the lack of an objective end point \[[@B34-ijms-17-01682]\]. 6. Skin Protection {#sec6-ijms-17-01682} ================== Ultraviolet radiation is a common cause for oxidative damage of the skin by increasing the production of ROS which initializes a series of signaling cascade that aggravates the condition. Being an antioxidant, the photoprotective effect of T3 in combination with TP was investigated in photosensitive volunteers who were subsequently subjected to photoprovocative test in this trial \[[@B35-ijms-17-01682]\]. Upon a single application prior to the provocative test, the topical formulation containing T3 was able to confer effective protection as compared to the vehicle and Vitamin A products. This finding has unveiled a new avenue for T3 as fewer studies have looked into the photoprotective nature of T3. Being a natural product, T3 is safe and well tolerated and formulations containing T3 could be investigated in more clinical trials in future. On the contrary, in another study, application of topical formulation consisting of T3 in post-surgical patients did not show any significant improvement on hypertrophic scar development after 6 weeks of application \[[@B36-ijms-17-01682]\]. The differences in formulation, the composition of the active ingredients and severity of the condition might have caused these different observations. Nevertheless, no adverse event was reported during the study indicating that topical application of T3 was well tolerated \[[@B36-ijms-17-01682]\]. Larger studies with variations in T3 formulations in variable degrees and types of skin diseases may answer the missing information in the topical protection conferred by T3. 7. Other Clinical Effects {#sec7-ijms-17-01682} ========================= In addition to the above functionalities, several exploratory studies found various response of TRF supplementation in different individuals. Heng and colleagues reported the effect of TRF supplementation and modulation of protein expression in young and old individuals \[[@B37-ijms-17-01682]\]. Each group was further separated into placebo and TRF arm and supplemented for six months. In groups supplemented with TRF, blood levels of tocotrienols was increased in both the younger and older population with the younger group having higher concentration of plasma T3. An interesting finding from this study was despite having lower plasma T3 levels, the older group reflected greater changes in the number of proteins expressed. These proteins are mainly involved in cholesterol homeostasis (APOE and APOA1), acute-phase proteins (CRP), immune response (FHR1 and FHR2) and protease inhibitor (CBPN and AMBP). Increase in APOA1 and APOE proteins were associated with lowered risk of developing atherosclerosis and cardiovascular disease. Similarly, lowering of CRP also correlates with better prognosis for cardiovascular related diseases. Augmentation of immune response is one of the characteristics observed with TRF supplementation and, in this study, increase in FHR1 and FHR2 plasma proteins might be the underlying mechanism behind this beneficial effect. In another study, the effect of TRF supplementation on DNA damage was investigated in individuals above and below 50 years old whereby the older group had higher DNA damage at baseline \[[@B38-ijms-17-01682]\]. Supplementation of 160 mg of TRF daily for 6 months were reported to have markedly reduced DNA damage and increased chromosomal stability in older individuals as compared to those who are below 50 years old. Age related DNA damage that was profoundly elevated in the older individuals improved significantly upon T3 supplementation. The researchers postulated that, at cellular level, T3 may have either interfered in prevention of DNA damage or augmented DNA repair rate \[[@B38-ijms-17-01682]\]. Increasing prevalence of Non Alcoholic Fatty Liver Disease (NAFLD) in both Asian and western nations led to initiation of a study to investigate the hepatoprotectivity of T3 in 64 individuals with NAFLD. After supplementation of 400 mg of TRF for a year, patients in the tocotrienol arm showed significant improvement in normalization of hepatic echogenic response in addition to absence of worsening of NAFLD grade \[[@B39-ijms-17-01682]\]. On the other hand, two patients in the placebo arm were reported to show disease progression with worsening grade. Moreover, significant reduction of TC, LDL and TG levels was observed in the TRF arm compared to the baseline. All the patients tolerated the supplementation well without any adverse event \[[@B38-ijms-17-01682]\]. Similar findings were reported in another study aimed at assessing the safety and tolerability of TRF in subjects with metabolic syndrome. Two weeks supplementation of 400 mg TRF daily did not cause any implication to hematological markers, serum liver function markers and liver enzymes. This implies that TRF supplementation does not cause hematoxicity and hepatotoxicity in metabolic syndrome subjects \[[@B40-ijms-17-01682]\]. Mahdy and colleagues carried out a study to assess the association between early supplementation of TRF and the incidence of pregnancy induced hypertension (PIH) in healthy primigravidae recruited in their early second trimester and followed till delivery \[[@B41-ijms-17-01682]\]. These women were supplemented with 100 mg TRF or placebo. The incidence of PIH although not significant but were found to be lower in TRF group with 64% reduction in the incidence of PIH. Remarkably, women in TRF arm were reported to have reduced blood loss in comparison with women on placebo. This finding dismisses the conventional concern that tocotrienols increases risk of bleeding and further asserts its safety \[[@B41-ijms-17-01682]\]. 8. Pharmacokinetics and Biodistribution {#sec8-ijms-17-01682} ======================================= Despite promising findings on the health protective properties of T3, their bioavailability has been debated among health authorities. Several studies have been conducted to assess the fasting and postprandial levels of T3 in plasma. Yap and colleagues reported that food intake played a crucial role in the absorption of T3 \[[@B42-ijms-17-01682]\]. The concentration of α-, γ- and δ-tocotrienols in plasma elevated significantly when taken with food in healthy individuals. A two-fold greater absorption was noted in the fed state based on the analysis of area under the curve (AUC~0--∞~). Twice daily dosing was justified from the findings of this study as the half-life of tocotrienols was relatively much shorter than TP. Therefore, double dosing will ensure the presence of T3 in the circulating plasma for longer duration \[[@B42-ijms-17-01682]\]. Another important factor to be considered in determining the bioavailability of T3 is the formulation which greatly influences the extent of absorption. The following study by the same team of researchers proved that self-emulsifying systems that produces finer droplets of emulsion prominently enhances the absorption of T3 by about two to three folds higher compared to non-emulsified formulation \[[@B43-ijms-17-01682]\]. It was also observed that the lag time required for the secretion of bile salts to emulsify the T3 in order to facilitate their absorption were relatively shorter with the presence of self-emulsifying systems. The findings by Yap et al. \[[@B42-ijms-17-01682]\] were further confirmed by Fairus et al. \[[@B44-ijms-17-01682]\], wherein the absorption of T3 was enhanced with food intake. Seven healthy subjects were administered a single-dose of either 1011 mg TRF or 1074 mg α-TP following a standardized breakfast. The major transporters of T3 were reported as triacylglycerol-rich particles (TRP) and HDL. However, the concentration of T3 was found to be much lower than TP in both circulating plasma and lipoproteins in this study. Likewise, comparable results were obtained in a follow-up study in 10 healthy volunteers supplemented with 526 mg TRF or 537 mg α-TP in similar study settings \[[@B45-ijms-17-01682]\]. A study on the delivery and bioavailability of T3 to vital organs upon supplementation of 400 mg TRF ranging from one to 96 weeks in surgical patients was published by Patel and colleagues in 2012 \[[@B32-ijms-17-01682]\]. Tocotrienol isoforms were found to be concentrated in circulating blood, skin, liver, cardiac muscle, brain and adipose tissue. Low bioavailability of T3 raised many uncertainties about their physiological role in humans. This study asserted that circulating levels of T3 were adequate and was higher than the therapeutic dose required to exert certain biological effects, such as neuroprotection. Another vital finding from this study was half of the individuals in the TRF arm reported reduction in model for end stage liver disease (MELD) score in comparison with only 20% in the TP arm \[[@B32-ijms-17-01682]\]. In light of heightened risk of liver disease, the ability of T3 to delay the progression of the disease suggests a promising platform for future clinical trials to further ascertain the effects. Most pharmacokinetic studies discussed above reported the concentration of individual isoforms upon supplementation with mixed T3 or TRF. In the following studies, newer formulations that consisted of T3 isoforms were investigated. These isoforms, namely gamma and delta tocotrienols were identified as the most potent compounds in pre-clinical anti-cancer research \[[@B46-ijms-17-01682],[@B47-ijms-17-01682]\]. Springett et al. \[[@B28-ijms-17-01682]\] studied (VEDT) in pancreatic cancer patients supplemented from 200 to 3200 mg daily for two weeks prior to surgery. The pharmacokinetic parameters such as Cmax, AUC~0--12~ and AUC~0--∞~ increased upon VEDT supplementation from 200 to 1600 mg and declined at 3200 mg. Despite high level of dosage given, the subjects tolerated the supplementation well. Similar to Patel et al. \[[@B32-ijms-17-01682]\] whereby oral supplementation of 400 mg/day of TRF reported to significantly increase the T3 concentration in blood and major organs as well as reduction in MELD score, evident physiological effect was observed with respect to VEDT supplementation indicating the favorable role exerted by tocotrienols. Increased apoptotic activity was reported with preoperative supplementation of VEDT in preoperative pancreatic cancer patients. In a different study, a formulation consisting of gamma and delta tocotrienols (GDT) were compared with conventional TRF in 12 healthy subjects in terms of bioavailability, safety and tolerability \[[@B48-ijms-17-01682]\]. The gamma isoform in GDT showed superior bioavailability as compared to γ-T3 in TRF. All the volunteers tolerated the supplementation well indicating both the formulations were safe at given dosage. In 2014, Fu et al. \[[@B49-ijms-17-01682]\] reviewed the pharmacokinetics and bioavailability of T3 based on clinical studies conducted in the past two decades. Over the years, many issues that restrict and limit the absorption of T3 have been addressed and enhanced. Available evidence from human trials, suggests that T3 concentration in plasma increases upon supplementation and lower bioavailability is no longer a drawback factor for T3 to exert their therapeutic properties. 9. Conclusions {#sec9-ijms-17-01682} ============== Based on this review, we understand that all clinical trials that had been conducted with T3 notably differ in their study designs, populations, formulations and dosing regimen. Despite the variations in the findings, these studies have facilitated our understanding on the potential physiological effect of T3. One of the major limitations of T3 debated in human trials was the lower bioavailability in plasma. Key findings from several studies have reported substantial concentration of T3 in plasma and vital organs, which are sufficient to exert health protective role in humans. Bleeding was a major concern with tocotrienols being labeled as anti-coagulant. Consistently, almost all the trials discussed have reported that T3 supplementation at different doses was safe and well-tolerated in the population studied. From healthy subjects to pre-surgical cancer patients and even pregnant women have tolerated T3 supplementation for short and long duration without any serious adverse event. With many of our queries and skepticisms on the bioavailability and physiological effect of T3 being answered, more studies are warranted to explore the therapeutic effect of T3 in various clinical settings and larger population to emphasize on the beneficial role of this natural product. The rising cost of drugs and the adverse effects commonly associated with them have led to many individuals and medical community to look for alternate solution. Tocotrienols being a well-tolerable natural compound with many distinct therapeutic properties may provide a possibility for better health care that may not only used to treat but may also play a major role in prevention. In order to materialize this idea into reality, more clinical trials are warranted to convey the convincing virtues of tocotrienols to the scientific community as well as the health care providers. In line with the market trend, there is a shift of market demand towards natural ingredient and new delivery mechanisms \[[@B1-ijms-17-01682]\]. Tocotrienols being a potent natural nutraceutical ingredient is a gift from nature. While more work on exploring new delivery systems will give added benefits to improve the bioavailability of tocotrienols, clinical evidence serve as the scientific fundamentals towards better acceptance among the consumers, clinicians and ultimately, the health authorities. The authors wish to thank the Malaysian Palm Oil Board for the support in writing this paper. Puvaneswari Meganathan and Ju-Yen Fu wrote the paper. All authors have read and approved the final manuscript. The authors declare no conflict of interest. {#ijms-17-01682-f001} {#ijms-17-01682-f002}

The authors confirm that all data underlying the findings are fully available without restriction. Data are included within the Supporting Information files. Introduction {#s1} ============ Glaucoma is one of the leading causes of blindness in the world [@pone.0098525-Quigley1], [@pone.0098525-Congdon1]. Glaucomatous visual field (VF) damage usually initiates in the mid-peripheral VF while the central region tends to be preserved until late on in the disease process. In advanced glaucoma, VF damage is often characterized by large arcuate scotomata, which have connected to form a ring, threatening visual function in the central area of the VF [@pone.0098525-Aulhorn1], [@pone.0098525-Aulhorn2]. The central VF is especially important because cortical magnification in this area is much larger than in the peripheral region [@pone.0098525-Daniel1], [@pone.0098525-Cowey1] and indeed it has been reported that VF damage in the central area results in disability in various daily tasks [@pone.0098525-Sumi1]. Thus, treatments should be intensified when VF damage threatens the patient\'s visual function, particularly in the central region. In glaucoma, VF sensitivities are highly correlated across corresponding regions of the retina [@pone.0098525-Mandava1]--[@pone.0098525-Suzuki1] and several maps have been proposed to describe these correlations as clusters in the central 30 degrees [@pone.0098525-Mandava1], [@pone.0098525-Suzuki1]--[@pone.0098525-GarwayHeath1]. In most of these maps, test points in the central 10 degrees tend to be clustered in small numbers. However, parafoveal defects can occur preferentially in early glaucoma [@pone.0098525-Park1]--[@pone.0098525-Langerhorst1], probably due to a distinctive pathological mechanism [@pone.0098525-Park1], and it has been suggested that patients with paracentral defects cannot be well-monitored unless a central 10° test program is used to densely measure this region [@pone.0098525-Weber1]--[@pone.0098525-Park2]. A recent paper reported that the 24-2 VF is not optimal for detecting early damage of the macula but VF damage in the central area can be detected early in the disease process using the 10-2 VF [@pone.0098525-Traynis1]. Interesting case examples are illustrated in [Figure 1](#pone-0098525-g001){ref-type="fig"}; in patients a and b, much deeper VF scotomata are observed in the 10-2 VFs than one might expect from inspection of their 30-2/24-2 VFs; however, the central damage is supported by findings obtained with optical coherence tomography. On the other hand, in case c, a considerable region of the VF can be seen to be preserved in the 10-2 VF, despite the appearance of almost complete blindness in the 24-2 VF; moreover, it is interesting to note that visual acuities in this patient were maintained: 20/32 in the right eye and 20/25 in the left eye. These examples illustrate the motivation for the current study: cluster test points in glaucoma patients\' 30-2 and 10-2 VFs in order to map the different regions damaged by the disease. {#pone-0098525-g001} Method {#s2} ====== This study was approved by the Research Ethics Committee of the Graduate School of Medicine and Faculty of Medicine at the University of Tokyo. Written consent was given by patients for their information to be stored in the hospital database and used for research. This study was performed according to the tenets of the Declaration of Helsinki. This was a retrospective study that included 128 eyes from 128 patients with a diagnosis of: primary open-angle glaucoma (125 eyes), and secondary open angle glaucoma (pseudoexfoliation; 3 eyes). Patients were followed in the general glaucoma clinic at the University of Tokyo Hospital. Patients who underwent measurements with both the 30-2 and 10-2 VF test patterns (Humphrey Field Analyzer: HFA, Carl Zeiss Meditec, Dublin, CA) in a single day were included. One eye was chosen randomly when both eyes met the criteria; right eye VFs were mirror-imaged to a left eye format. Other criteria for inclusion in the study were visual acuity better than 6/12, no previous ocular surgery (except for cataract extraction and intraocular lens implantation), and no other posterior segment eye disease. All VFs were recorded using the SITA standard strategy with a Goldmann size III target. Reliability criteria applied were fixation losses less than 25% and false-positive responses less than 15%, a false-negative rate was not used to exclude VFs based on results in Bengtsson and Heijl [@pone.0098525-Bengtsson1]. Comparison between 30-2 VF and 10-2 VF test results {#s2a} --------------------------------------------------- Mean deviation (MD) values derived from patients\' 30-2 VFs and 10-2 VFs were compared using Pearson\'s correlation coefficient. In addition, Pearson\'s correlation coefficient was calculated (i) between the mean of the total deviation (TD) values of the four innermost test points in the 30-2 VF (locations \[3,3\], \[3,−3\], \[−3,3\] and \[−3,−3\]; \[x-axis coordinate, y-axis coordinate\]) denoted and the mean of the TD values of all 68 test points in the 10-2 VF (denoted ), and (ii) between the each of the TD values of the four innermost test points in the 30-2 VF (denoted ) and the mean of the TD values of the innermost 17 test points of the 10-2 VF (denoted ) in each of superior-temporal, superior-nasal, inferior-temporal and inferior-nasal quadrants. VF clustering analysis {#s2b} ---------------------- Test points from the 30-2 and 10-2 VFs were clustered using the Hierarchical Ordered Partitioning and Collapsing Hybrid (HOPACH) -- Partitioning Around Medoids (PAM) algorithm, for its ability to hierarchically order and partition clusters into finite groups in an unbiased manner. The HOPACH-PAM algorithm is a hybrid between hierarchical ordered partitioning and collapsing [@pone.0098525-vanderLaan1]--[@pone.0098525-Bioconductor1]; in other words, HOPACH builds a hierarchical tree of clusters by recursively partitioning the VF, while ordering and possibly collapsing clusters at each level to identify finite structures in a dataset. The HOPACH-PAM algorithm uses the Mean/Median Split Silhouette (MSS) criteria which is particularly apt at identifying structures in a dataset [@pone.0098525-Glinca1]. One of strengths of this approach is that the optimum number of VF clusters is mathematically inferred by the algorithm; this is in contrast to many other clustering approaches, such as k-means and hierarchical clustering methods, that arbitrarily decide the number of optimum clusters, which can lead to incorrect results [@pone.0098525-vanderLaan1]. Using the HOPACH-PAM algorithm, the 74 test points from the 30-2 VF and 68 points from the 10-2 VF were clustered according to their TD values; the four test points (locations \[3,3\], \[3,−3\], \[−3,3\] and \[−3,−3\]; \[x-axis coordinate, y-axis coordinate\]) overlapping between the two test patterns were taken from the 30-2 VF (please note there are no test points at the locations of locations \[3,9\], \[9,3\], \[−3,9\], \[−9,3\], \[3,−9\], \[9,−3\] \[−3,−3\] and \[−3,−9\]; \[x-axis coordinate, y-axis coordinate\] in 10-2 VF). Clustering methods are frequently carried out in an exploratory manner and often the patterns found are not translatable to other datasets [@pone.0098525-Hennig1] hence bootstrap analysis (10,000 re-samples) was performed to explore the significance of the clustering results. All statistical analyses were carried out using the statistical programming language R (ver. 2.15.1, The R Foundation for Statistical Computing, Vienna, Austria). The R package "hopach" was used to carry out the analysis of HOPACH-PAM. Results {#s3} ======= [Table 1](#pone-0098525-t001){ref-type="table"} shows the subjects\' demographics. 10.1371/journal.pone.0098525.t001 ###### Subject demographics. {#pone-0098525-t001-1} ------------------------------------ ----------------------------- age (mean ± sd, \[range\]) 58.1±13.2 \[21 to 86\] male/female 63/65 Type of glaucoma NTG 74 POAG 51 PEG 3 MD (30-2 VF, mean ± sd, \[range\]) −14.8±7.5 \[0.5 to −29.2\] MD (10-2 VF, mean ± sd, \[range\]) −15.5±7.7 \[−0.3 to −30.9\] ------------------------------------ ----------------------------- sd: standard deviation, NTG: normal tension glaucoma, POAG: primary open angle glaucoma, PEG: pseudoexfoliation glaucoma, MD: mean deviation. There was a significant and moderate association between MD values derived from patients\' 30-2 VFs and 10-2 VFs (R^2^ = 0.51, p\<0.0001). There was also a significant and strong relationship between and values (R^2^ = 0.80, p\<0.0001, [Figure 2](#pone-0098525-g002){ref-type="fig"}). Finally, there was a significant but relatively weak correlation between and values (R^2^ = 0.34, 0.51, 0.32 and 0.56 (p\<0.001) in the superior-temporal, superior-nasal, inferior-temporal and inferior-nasal quadrant, respectively): see [Figure 3](#pone-0098525-g003){ref-type="fig"}. {#pone-0098525-g002} {#pone-0098525-g003} Results of the HOPACH-PAM clustering are illustrated in [Figure 4](#pone-0098525-g004){ref-type="fig"}; 65 clusters ('sectors') were identified, adjacent VF test points tended to be clustered into the same sector even when test points were derived from different VF tests (30-2 or 10-2 VF grids). Furthermore, clustering appears to follow the average distribution of the retinal nerve fiber layer (RNFL). [Figure 5](#pone-0098525-g005){ref-type="fig"} illustrates the results of bootstrapping; we can see which clusters are most stable (wide bars) and which pairs of clusters are most likely to exchange test points with one another. Most sectors had good stability and this did not appear to be affected by the particular test pattern. {#pone-0098525-g004} {#pone-0098525-g005} Discussion {#s4} ========== In this study the relationship between test locations in the 30-2 VF and 10-2 VF was investigated. Although TD values in the 10-2 VF were highly correlated with the central four TD values of the 30-2 VF on average, there are still some instances of large agreement in individual patients. In particular in early glaucomatous cases where tends to better preserved than (see [Figure 3](#pone-0098525-g003){ref-type="fig"}). In addition, the relationship between and is relatively weak, which indicates that the magnitude of association varies significantly from patient-to-patient. Thus, it appears that the and are insufficient to precisely estimate visual defects in this region of the central VF. Among the 65 sectors identified from the HOPACH-PAM clustering algorithm, 38 sectors were located outside the 10-2 VF whereas 29 sectors fell inside the 10-2 VF. Bootstrapping suggested that clusters from both the 10-2 VF and 30-2 VF were stable. Moreover, these sectors appeared to follow the distribution of retinal nerve fiber bundles. As shown in the [Figure 4](#pone-0098525-g004){ref-type="fig"}, a large number of sectors were identified in the 10-2 VF that were completely independent of points in the 30-2 VF. This clearly suggests that clinicians cannot solely rely on the 30-2 VF when evaluating damage in glaucoma patients; closer examination of the central ten degrees, using a denser test grid than currently employed in the 30-2 VF, is essential. Park et al. have suggested that eyes presenting with parafoveal scotomata have distinctly different risk factors to patients with other patterns of VF damage [@pone.0098525-Park1]; moreover, Park et al. argue that progression of this defect can be detected more effectively with the 10-2 VF test grid compared with the 30-2 VF [@pone.0098525-Park2]. On the other hand, clinicians cannot rely exclusively on the 10-2 VF because typical glaucomatous VF damage such as arcuate scotoma and the nasal step defect [@pone.0098525-Drance1] would not be detected with this test grid. Hence, it is not recommended to carry out 10-2 VFs as an alternative to 30-2 VF testing but neither is it recommended to alternate 30-2 and 10-2 VF tests, because the detection of progression will, in general, be delayed with a reduced number of consistent test grids. Therefore, clinicians are advised to continue to use the same strategy (e.g., always 30--2 or 24-2 VFs) [@pone.0098525-Chauhan2]. Furthermore, VF testing at some hospitals is performed much less than recommended [@pone.0098525-Malik1], hence alternating test patterns will only compound this problem. Consequently, it may be beneficial to develop a new VF test grid that spans the central 30 degrees but includes a higher density of test points in the central ten degrees; the clustering results presented here could help to devise this new test pattern. A possible caveat of the current study is the sample of patients studied; in particular, patients may have been undergone 10-2 *and* 30-2 VF tests on the same day because the clinician suspected a paracentral defect that had not been detected using the 30-2 VF test. Nonetheless, the purpose of our study was to investigate whether the spatial sampling of the 30-2 VF is sufficient to detect paracentral damage, which is greatly important for VRQoL. Several studies have clustered 30-2 or 24-2 VFs in order to inform the relationship between test points [@pone.0098525-Mandava1], [@pone.0098525-Chauhan1], [@pone.0098525-Katz1], the anatomical structure of retinal nerve fiber bundles [@pone.0098525-GarwayHeath1] and the pattern of progression rates [@pone.0098525-NouriMahdavi1]; these studies, like the one presented here, have all identified clusters that to some extent follow the structural distribution of RNFLs. In addition, Koseki et al. have clustered points in the 10-2 VF [@pone.0098525-Koseki1], revealing groups of points that also appear to follow the paths of RNFLs. Our results are in agreement with these studies, with clusters identified in both the 10-2 VF and the 30-2 VF that follow the distribution of the RNFL. Noteworthy in the current results is the large number of 10-2 VF clusters relative to the number of clusters from the 30-2 VF; in addition, the stability of 10-2 VF clusters (as revealed by bootstrapping) was, in general, as stable as clusters in the 30-2 VF. These findings suggest that many 10-2 VF clusters exist independently from clusters in the 30-2 VF. In our results, no clusters spanned the meridian line; this outcome is in good agreement with the anatomical distribution of the RNFL [@pone.0098525-Hood1]. One exception, however, is sector 39, which is located in the temporal area of the VF; we suppose that the sparse distribution of RNFL in this area and low frequency of glaucomatous VF deterioration in this region is one explanation for this finding. It would be interesting to further investigate the separation of the anatomical upper and lower hemifields in the temporal VF area. Glaucomatous RNFL damage predominantly occurs in the supero- and infero-temporal angles around the optic nerve head. Hood et al. [@pone.0098525-Hood1] and Heijl et al. [@pone.0098525-Heijl1] have both reported that glaucomatous VF damage usually commences in the central VF, particularly in the superior hemifield; Hood et al. have suggested that this is because most of the corresponding RNFLs flow into the optic disc at the infero-temporal angle. Supporting this suggestion is the observation that early glaucomatous VF change occurs in this area almost as frequently as the 'classic' arcuate defect and nasal step defect [@pone.0098525-Langerhorst1], [@pone.0098525-Tan1], [@pone.0098525-Schiefer1]. Furthermore, the density of RNFLs is very high in this region and despite the fact that it represents less than 2% of the retinal area it contains more than 30% of ganglion cell [@pone.0098525-Curcio1]. Consequently, it is perhaps not surprising to see so many clusters in the central superior hemifield. On the other hand, the central inferior VF area tends to be preserved until late stage glaucoma. Hood et al. have suggested this is because the RNFLs in this area penetrate the optic disc margin at the temporal angle, which is usually less likely to be affected in early to moderate glaucoma [@pone.0098525-Weber1]. As a consequence, the clusters identified in the inferior central hemifield may be a result of VF damage observed in a subsample of late stage glaucoma patients. Interestingly, Hood et al. have suggested superior VF test points, just above the horizontal line, tend not to be involved in early to moderate glaucomatous VF change; accordingly, Sectors 10, 11 and 12 are probably a consequence of VF change in advanced glaucoma patients, similarly to clusters in the central inferior VF area. As shown in [Figure 5](#pone-0098525-g005){ref-type="fig"}, most of the identified sectors were reproduced in the same area in the bootstrapped samples. However, some sectors, such as 11, 12, 21, 28, 37, 54 and 55, were not stable. As structure-function mapping is influenced by many ocular parameters, such as position of the ONH in relation to the fovea, disc area, axial length, spherical equivalent, disc shape, disc orientation and disc tilt [@pone.0098525-Lamparter1], it is not surprising that no one clustering result will be applicable to all patients. Future studies should be carried out to continue efforts to create patient-customized VF cluster maps that consider these parameters. In conclusion, this study suggests that many areas in the 10-2 VF as well as the 30-2 VF are affected by glaucoma. Thus, it is not sufficient to merely measure a glaucoma patient\'s 30-2 VF; instead it should be recommended to additionally measure a dense test grid, such as the 10-2 VF pattern, when evaluating the status of VF damage. Supporting Information {#s5} ====================== ###### **Visual field data analyzed.** (CSV) ###### Click here for additional data file. [^1]: **Competing Interests:**The author has declared that no competing interests exist. [^2]: Conceived and designed the experiments: RA. Performed the experiments: RA. Analyzed the data: RA. Contributed reagents/materials/analysis tools: RA. Contributed to the writing of the manuscript: RA.

Background ========== The antimitotic anticancer drug paclitaxel (taxol) has a unique mechanism of action: unlike other mitotic spindle poisons (*viz*. vinca alkaloids), taxol binds to the N-terminal end of β-tubulin and promotes microtubule assembly \[[@B1]-[@B3]\]. The effect of taxol on the microtubule dynamics has been studied extensively in cell free systems as well in living cells \[[@B4]-[@B6]\]. Observations thus made indicate taxol-induced suppression of microtubule dynamics inhibits the rate and extent of shortening of microtubules causing aberrant mitotic spindles formation and cell cycle arrest \[[@B4]-[@B6]\]. Overexpression of P-glycoprotein (P-gp; which functions as a drug efflux pump, \[[@B7],[@B8]\]) is the common phenotype observed in taxol-resistant tumor cells. In addition, considering its intracellular target, it is not surprising that resistance to taxol has also been associated with alterations in microtubules resulting in reduced drug-binding affinity or decreased intracellular levels of polymerized tubulin and/or changes in expression of specific tubulin isotypes \[[@B8]-[@B10]\]. Recently, Goncalves *et al*. reported association of increased microtubule dynamics with taxol resistance in lung cancer cells \[[@B11]\]. However, they did not identify the cellular mediators of the increased dynamic behavior of microtubule in the taxol-resistant cells. We have established a series of taxol-resistant human ovarian adenocarcinoma cell lines \[[@B12]\]. The \> 1500-fold taxol-resistant 2008/17/4 cell line displayed the classical multidrug resistance phenotype with overexpression of P-gp and cross-resistance to other natural product drugs \[[@B12]\]. In contrast, the 2008/13/4 cell line, which was 252-fold resistant to taxol relative to the parental cell line (2008), did not exhibit overexpression of the P-gp and/or alterations in the microtubules \[[@B12]\]. The intracellular accumulation of taxol in the 2008/13/4 cells was similar to that observed in 2008 cells. The *in vitro*binding of taxol to microtubules semi-purified from the 2008/13/4 and 2008 cells was also identical \[[@B12]\]. The basal levels of various α- and β-tubulin isotypes (except Bα-1) as well as polymerized tubulin were similar in the taxol-resistant and the parental cells (unpublished observation). This suggests that mechanisms not hitherto identified and not typically associated with development of taxol-resistance were operative in the 2008/13/4 sub-line. Considering the potential utility of taxol in the treatment of primary and cisplatin-resistant ovarian cancers, it was of great clinical importance to elucidate these mechanisms. Employing the mRNA differential display analysis we identified an increased expression of a protein involved in the G-protein coupled signal transduction pathway, named Gαi1, in the taxol-resistant cells compared to the parental cell. The Gα proteins, based on sequence comparisons are subdivided into four classes, Gs, Gi, Gq and G12-13. The Gαi is the regulatory signaling molecule that **inhibits**the adenylate cyclase activity \[[@B13]\]. Increased expression of Gαi1 could potentially inhibit the adenylate cyclase activity and thus the cAMP mediated signal transduction (that act through the PKA). Decrease in the PKA activity leads to activation of the mitogen-activated protein kinase (MAPK) pathway initiated by c-raf-1 \[[@B14]\]. In resting cells, the Gαi forms a heterotrimeric complex (with the β- and γ- subunits of guanine nucleotide binding protein) that provides signal coupling through a series of G-protein coupled receptors \[[@B13]\]. Upon activation by an appropriate signal, the receptor interacts with the plasma membrane bound heterotrimeric G-protein complex and catalyzes the exchange of bound GDP for GTP in the α-subunit. Subsequently, the GTP-bound Gα subunit and the βγ-subunit dissociate from the receptor as well as from each other. The \"active\" α-subunit and the \"free\" βγ-subunit initiate cellular response by altering the activity of intracellular effector molecules \[[@B13]\]. However, recent studies have indicated that a large fraction of the Gαi proteins show perinuclear localization \[[@B15],[@B16]\]. In addition, Roychowdhury *et al*. \[[@B17]\] have demonstrated that Gαi1 specifically binds to the microtubules and activates the GTPase activity of tubulin resulting in increased microtubule dynamic instability and increased depolymerization of the microtubules. We thus hypothesized that increased Gαi1 association with the microtubules in the taxol-resistant cells would be sufficient to inhibit the taxol-induced suppression of microtubule dynamic instability such that it would allow the taxol-bound microtubules to depolymerize. This would negatively influence the cytotoxic effects of taxol (on the formation of stable microtubules leading to mitotic arrest) and provide a mechanism *via*which drug-resistant cells would be able to evade the cytotoxic effects of taxol. Based on our results presented in this study as well as those of Roychowdhury *et al*. \[[@B17]\] and Goncalves *et al*. \[[@B11]\], it is conceivable that increased binding of Gαi1 with the microtubules in the drug-resistant cells could inhibit taxol-induced suppression of microtubule dynamics, thus allowing cells to complete mitosis. Results ======= Two distinct taxol-resistant sublines were utilized in this study. The 2008/17/4 subline displays a classical MDR phenotype; an increased expression of the P-gp and decreased intracellular taxol accumulation \[[@B12]\]. In contrast, the 2008/13/4 subline has negligible P-gp expression and no defect in the intracellular accumulation of taxol \[[@B12]\]. The basal levels of tubulin expression were similar in the parental 2008 and the taxol-resistant 2008/13/4 cells and the *in vitro*binding affinities of taxol to purified microtubules derived from these cells were not significantly different \[[@B12]\]. Differential expression of Gαi1 in the taxol-resistant cells ------------------------------------------------------------ To identify genes whose expression is either induced or reduced in the taxol-resistant cells, we systematically compared mRNA display patterns between parental (2008) and the taxol-resistant (2008/13/4 and 2008/17/4) cells. We choose only those bands that showed a consistent differential expression in the duplicate samples from the 2008/13/4 cells and/or 2008/17/4 cells compared to the 2008 cells. Furthermore, we chose only those bands that exhibited at least 3-fold change in densitometric intensity. We thus identified the upregulation of the alpha-i1 subunit of the guanine nucleotide binding protein mRNA (Gαi1; data not shown) in the 2008/13/4 and 2008/17/4 cells compared to the 2008 cells. The increased expression of the Gαi1 in the taxol resistant cells was confirmed by Northern and Western blotting analysis. As shown in Figure [1](#F1){ref-type="fig"}, a 15-fold and 6-fold higher expression of Gαi1 (protein and mRNA, respectively) was observed in the 2008/13/4 and 2008/17/4 cells, respectively, compared to the 2008 cells. {#F1} The effect of taxol on the cAMP levels and PKA activity in the parental and taxol-resistant cells ------------------------------------------------------------------------------------------------- The Gαi is a regulatory signaling molecule that **inhibits**the adenylate cyclase activity \[[@B13]\]. Increased expression of Gαi1 could potentially inhibit the adenylate cyclase activity and thus the cAMP mediated signal transduction (that act through the PKA). We assessed the basal levels of cAMP as well as the basal PKA activity in the parental and taxol-resistant cells. Effect of taxol exposure on the intracellular levels of cAMP and the PKA activity were also evaluated in these cells. The basal level of cAMP in the 2008, 2008/13/4 and 2008/17/4 cells was measured using a competitive immunoassay kit from Biomol Research Laboratories (Plymouth Meeting, PA). Surprisingly, the basal level of cAMP was 4.5-fold and 9-fold higher in the 2008/13/4 cells compared to the 2008 and 2008/17/4 cells, respectively (Fig. [2](#F2){ref-type="fig"}). These results indicate that increased expression of Gαi1 observed in the 2008/13/4 cells does not inhibit the basal AC activity in these cells. In contrast, treatment (for 16 hr) with increasing concentration of taxol did not alter the cAMP levels in the 2008, 2008/13/4 and 2008/17/4 cells significantly. It is likely that exposure to taxol causes a mild inhibition of adenylate cyclase activity with the resultant insignificant decrease in the intracellular levels of camp observed (Fig. [2](#F2){ref-type="fig"}). {#F2} We then determined the activity of cAMP-dependent Protein kinase A in the parental and taxol-resistant cells. As shown in Fig. [3](#F3){ref-type="fig"}, the PKA activity in the untreated 2008 cells was 8 units/mg protein and it was 4-fold (32 units/mg protein) and 2.5-fold (21 units/mg protein) higher in the 2008/13/4 and the 2008/17/4 cells, respectively. Treatment (for 16 hr) with increasing concentrations of taxol did not significantly affect the PKA activity in any of the cell lines studied (Fig. [3](#F3){ref-type="fig"}). {#F3} Based on these observations we concluded that overexpression of Gαi1 in the 2008/13/4 and 2008/17/4 cells does not inhibit the cAMP-dependent signal transduction pathways, otherwise inhibition of adenylate cyclase (due to increased activity of Gαi1) would have led to a decrease in the intracellular levels of cAMP and a decrease in the activity of the cAMP-dependent PKA. Association of Gαi1 with the microtubules in taxol resistant cells ------------------------------------------------------------------ Results presented thus far indicate that the mode of action of Gαi1 in the taxol-resistant cells maybe distinct from its ability to inactivate AC. Recent studies have demonstrated that Gαi1 specifically binds to the microtubules and activates the GTPase activity of tubulin \[[@B17]\]. Considering that the cellular target of taxol is the microtubule, we wanted to investigate whether alterations in the interaction of Gαi1 with the microtubules accompanied development of taxol resistance in the 2008/13/4 and the 2008/17/4 cells. Utilizing confocal microscopy (Fig. [4](#F4){ref-type="fig"}) and differential extraction (Figs. [5](#F5){ref-type="fig"}, [6](#F6){ref-type="fig"}), we now show an increased association of Gαi1 with the microtubules in the taxol-resistant cells compared to the sensitive cells, both prior to and after treatment with taxol. It is noteworthy here that vincristine, an anticancer drug known to depolymerize microtubules did not increase the association of Gαi1 with the microtubules in either the 2008 or the 2008/13/4 and 2008/17/4 cells (Figs. [4](#F4){ref-type="fig"} and [7](#F7){ref-type="fig"}). {#F4} ![**Taxol dose-dependent increase in the association of Gαi1 protein with the microtubules exclusively in the 2008/13/4 cells**. The 2008 and 2008/13/4 cells were seeded at a density of 1 × 10^6^/ml and incubated under normal growth conditions for 36 hours. Thereafter, the 2008 cells untreated (lane 1) or treated for 24 hr with 25 nM (lane 2) and 500 nM (lane 3) taxol and the 2008/13/4 cells untreated (lane 4) or treated for 24 hr with 500 nM (lane 5) and 5 μM (lane 6) and then washed with chilled PBS (3 ×). The cytoskeletal fraction was isolated essentially as described previously \[22\]. Briefly, the attached cells were incubated in a microtubule-stabilizing buffer (0.1 M PIPES, 1 mM EGTA, 1 mM MgSO~4~, 2 M glycerol, pH 8.0) for 20 min. Thereafter, the cytosolic protein were removed by incubating the cells in the microtubule stabilizing buffer containing 0.1% NP-40 and protease inhibitor cocktail for 20 min. The cytoskeletal fraction (attached to the plastic dishes) was scraped into RIPA buffer (PBS containing 1% NP-40, 0.5% sodium deoxycholate and 0.1% SDS) containing protease inhibitor cocktail. The cytoskeletal fraction (5 μg protein/lane) was then subjected to SDS-PAGE. The separated proteins were transferred to PVDF membranes and Western blotting was performed using the polyclonal antibody against Gαi1.](1750-2187-1-3-5){#F5} {#F6} {#F7} As shown in Fig. [4](#F4){ref-type="fig"}, a distinctly cytoplasmic localization of the Gαi1 protein (green) was observed in all the cell lines studied. The microtubules (red) were also stained with antibodies directed against β-tubulin. Co-localization of Gαi1 with microtubules (assessed by presence of yellow coloration) was not observed in the untreated 2008 cells, but was clearly visible in the taxol-resistant 2008/13/4 and 2008/17/4 cells. Moreover, treatment with taxol induced a time-dependent increase in the association of Gαi1 protein and the microtubule network in the taxol-resistant cells. This phenomenon was completely absent in the parental 2008 cells at 4 and 24 hr post-treatment with taxol. However, after 24 hr exposure to taxol, a few 2008 cells (\<1%) displayed spotty co-localization of the Gαi1 protein with the microtubules. We also evaluated the effects of vincristine on the distribution of cellular Gαi1 protein. As shown in Fig. [4](#F4){ref-type="fig"}, treatment with vincristine decreased the association of the Gαi1 protein with the microtubules in the taxol-resistant cells without affecting its localization in 2008 cells. These results clearly illustrate that Gαi1 is differentially associated with the microtubules in the taxol-resistant cells in response to taxol. In addition to the *in-situ*localization of the Gαi1 with the microtubules we also evaluated the association of Gαi1 with the microtubules and the effect of taxol treatment on this association in the 2008 and the 2008/13/4 cells utilizing a differential extraction procedure. As shown in Fig. [5](#F5){ref-type="fig"}, utilizing 5 μg protein/lane, Gαi1 was not found to be associated with the cytoskeletal fraction in the untreated 2008 cells or the 2008/13/4 cells. However, it should be pointed out that increasing the protein concentration to 30 μg does demonstrate a low basal level of Gαi1 bound to the cytoskeletal fraction in the 2008 and the 2008/13/4 cells (as shown in Fig. [7](#F7){ref-type="fig"}). Treatment of 2008 cells with 25 nM (IC~50~) or 500 nM (20 × IC~50~) concentration of taxol did not induce binding of Gαi1 with the microtubules. In marked contrast, treatment of the 2008/13/4 cells with 500 nM or 5 μM (IC~50~) concentration of taxol demonstrated a dose-dependent increase in the presence of Gαi1 in the cytoskeletal fraction. Next, we evaluated the distribution of Gαi1 in the various cellular fractions (cytoplasmic, cytoskeletal, membrane and nuclear) before and after treatment with taxol in the 2008, 2008/13/4 and 2008/17/4 cells. As shown in Fig. [6](#F6){ref-type="fig"}, expression of Gαi1 was significantly higher in the whole cell lysates of untreated 2008/13/4 and 2008/17/4 cells compared to 2008 cells. Treatment with taxol did not affect the levels of Gαi1 in the 2008/13/4 whole cell lysate, however a significant decrease in Gαi1 levels was observed in the whole cell lysate obtained from 2008 and 2008/17/4 cells treated with taxol for 24 hr. Very low to negligible levels of Gαi1 were observed in the membrane fraction from each cell line (data not shown) and treatment with taxol did not increase the Gαi1 levels in the membrane fraction. Localization of Gαi1 in the nuclear fraction was not observed in any cell lines (data not shown). Similar to results obtained earlier (Fig. [5](#F5){ref-type="fig"}), association of Gαi1 with the microtubules was not observed at any time (before or after taxol treatment) in the 2008 cells. In contrast, a time-dependent increase in the partitioning of Gαi1 to the cytoskeletal fractions was observed in the 2008/13/4 cells. In case of the 2008/17/4 cells, a significant fraction of Gαi1 was already associated with the microtubules in the untreated cells, which was not affected by taxol treatment. When we evaluated the localization of Gαi1 in the cytoplasmic fraction in untreated and taxol- treated cells, we observed that levels of soluble Gαi1 in untreated 2008 and 2008/13/4 cells were similar, while very low levels of soluble Gαi1 were observed in the 2008/17/4 cells. The latter could be due to increased association of Gαi1 and microtubules in the untreated 2008/17/4 cells. Treatment with taxol for 24 hr decreased the levels of soluble Gαi1 in all the cell lines. Based on these observations and those reported in Fig. [4](#F4){ref-type="fig"} (*in-situ*immunolocalization using confocal microscopy) we hypothesize that while decrease in the levels of cytoplasmic Gαi1 in the 2008 cells could be a direct result of increased degradation of the protein, decrease in the soluble Gαi1 in response to taxol in the 2008/13/4 cells and possibly 2008/17/4 cells is due to an increased partitioning of the Gαi1 from the cytoplasmic fraction to the cytoskeletal fraction. We next determined whether the differential partitioning of Gαi1 to the microtubule fraction in the taxol-resistant cells was drug-specific (Figure [7](#F7){ref-type="fig"}). We treated the parental and taxol-resistant (2008/13/4) cells with vincristine and assessed the levels of Gαi1 in the whole cell lysate and in the cytoskeletal fraction. Vincristine was utilized because it is (like taxol) an antimitotic anticancer drug, however unlike taxol, vincristine induces depolymerization of the microtubules. Vincristine (50 nM) decreased the level of Gαi1 in the 2008 whole cell lysate (Figure [7](#F7){ref-type="fig"}, lane 2) without affecting the total levels in the 2008/13/4 cells (Figure [7](#F7){ref-type="fig"}, lane 4). Further, in contrast to the observation made with taxol, treatment of 2008 (50 nM) and 2008/13/4 (5 μM) cells with vincristine decreased the association of Gαi1 with the cytoskeletal fraction (Figure [7](#F7){ref-type="fig"}, lanes 6 and 8, respectively). These results suggest that the differential partitioning of Gαi1 to the microtubule fraction observed in the taxol-treated drug-resistant cells is indeed drug-specific providing further evidence that Gαi1 may be involved in inhibiting the taxol-induced suppression of microtubule dynamics. Discussion ========== Heterotrimeric G proteins transduce signals by virtue of their localization in close juxtaposition with the G-protein coupled receptors on the cytoplasmic side of the plasma membrane \[[@B13]\]. Activated receptors induce exchange of GDP for GTP on the Gα subunit and thus mediate the dissociation of Gα and βγ-subunits \[[@B13]\]. Dissociated Gαi1 decreases cAMP levels (due to inhibition of adenylate cyclase) and thus suppresses the cAMP-dependent PKA pathway. However, in the present study despite increased expression of Gαi1, under basal conditions and in the presence of taxol, intracellular cAMP levels as well as the PKA activity was higher in the taxol-resistant cells (Figs. [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"}) compared to the parental cells and \>95% of the Gαi1 displayed intracellular localization (either in the cytoplasmic and/or in the cytoskeletal fraction; Fig. [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}, [6](#F6){ref-type="fig"}) in the parental and taxol-resistant cells. Furthermore, treatment with taxol induced a differential partitioning of Gαi1 in the cytoskeletal fraction specifically in the 2008/13/4 and 2008/17/4 cells. These results suggest that the biochemical machinery necessary for targeting of Gαi1 is not operative under basal conditions or under conditions of stress associated with exposure to taxol either in the parental or in the taxol-resistant cells. Several studies indicate that covalent modification, including myristoylation and palmitoylation of Gα subunits are important in its targeting to the membrane \[[@B18],[@B19]\]. Whilst constitutively activated mutants of Gα subunit have been identified, their importance in the cellular localization seems to be associated with the lack of affinity of these mutants with the βγ-subunit partner, an interaction necessary for membrane localization \[[@B20]\]. Thus, it is likely that alteration in the conformation of the Gαi1 protein due to posttranslational modifications and/or changes in the levels of expression of partner protein(s) (that aid in targeting it to the microtubules) is responsible for the differential localization of Gαi1 with the microtubules in the taxol-resistant cells. Cell membrane localization of Gαi1 is necessary for its inhibitory effect on the adenylate cyclase pathway. The absence of Gαi1 localization to the cell membrane in the taxol-sensitive and -resistant human ovarian cancer cell lines used in this study, could explain the lack of correlation between the increased intracellular levels of Gαi1 in the taxol-resistant cells with the cellular cAMP content and the activity of PKA in these cells. Gαi1 is expressed in the 2008, albeit at levels lower than in the taxol-resistant cells; thus it is unclear as to why specific targeting of Gαi1 to the microtubules (before and after treatment with taxol) is observed only in the taxol-resistant cells. Whilst it is possible that differential posttranslational modification of the Gαi1 protein could account for its microtubule targeting in the taxol-resistant cells it is also likely that treatment of 2008/13/4 cells with taxol induces (a) the expression of a \"chaperone\" protein (bound to the microtubules and expressed only in the taxol-resistant cells) that could bind the soluble Gαi1 and thereby increase its association with the cytoskeleton or (b) binding of Gαi1 with a \"chaperone\" protein increases the ability of the complex to interact with the microtubules. While a third possibility exist; i.e., alteration in the microtubules themselves such that it increases the affinity of Gαi1 binding, such a change would lead to constitutively increased association of the Gαi1 with the microtubules, a phenomenon not observed in the 2008/13/4 cells. We had originally hypothesized that the observed upregulation of the G-proteins might function *via*the adenylate cyclase -- cAMP -- PKA-mediated signal transduction pathway in the development of taxol resistance. In case of the 2008/13/4 cells, although changes were identified in this pathway these could not be directly attributed as an effect of increased expression of the Gαi1. These observations make a compelling argument that other subtle and uncommon effects of the Gαi1 may be involved in the development of taxol resistance observed in the 2008/13/4 and 2008/17/4 cells. In addition to being membrane bound, several reports \[[@B15],[@B16]\] including the present study have demonstrated that the Gαi1 protein is present in the perinuclear region of the cell and is probably involved in transfer of signals from cell surface receptors to intracellular effector molecules. Wang *et al*. recently demonstrated that the tubulin binds to purified Gαi1 and Gαs protein with high specificity but not to other members of the Gαi family of proteins \[[@B21]\]. Furthermore, Roychowdhury *et al*. demonstrated that binding of Gαi1 to tubulin dimers and/or microtubules resulted in activation of tubulin GTPase activity and modulation of microtubule polymerization dynamics \[[@B17]\]. An elegant model of regulation of microtubule dynamics in vivo by Gαi protein proposed by these authors suggest that binding of the G-protein to the end of a microtubule induces hydrolysis of GTP and subsequent loss of the stabilizing cap, resulting in microtubule depolymerization (a catastrophe event) \[[@B17]\]. In contrast, binding of taxol to microtubule leads to stabilization of microtubule polymer, in other words suppression of dynamic instability of microtubules (4--6). Taken together, it is very attractive to speculate that the increased expression of the Gαi1 protein would be able to induce depolymerization of the taxol-stabilized microtubule polymers in the taxol resistant 2008/13/4 and the 2008/17/4 cells, resulting in completion of mitosis and subsequent cell division. Conclusion ========== We hypothesize that all else being equal, the presence of increased quantity of Gαi1 protein in the vicinity of the taxol-bound microtubules is likely to negate the suppressive effects of the drug and allow microtubules to depolymerize at the end of mitosis and complete the cell cycle. The notion is that Gαi1 with its ability to activate the intrinsic tubulin-GTPase activity will enable taxol-stabilized microtubules to depolymerize. Studies are underway in this laboratory to ascertain whether such a phenomenon occurs in vitro utilizing microtubules purified from the taxol resistant cells. Methods ======= Materials --------- Taxol was obtained from the Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment, National Cancer Institute, National Institutes of Health. It was dissolved in DMSO at a final concentration of 20 mM and the concentration of the solvent never exceeded 0.1% in any experimental protocol. \[α-^32^P\] dCTP (3000 Ci/mmol) was purchased from Dupont-NEN Research Products. The random priming kit from Amersham Biosciences (Piscataway, NJ) was utilized for radioactively labeling the Gαi1 cDNA. The enhanced chemiluminescence reagents were from Pierce Biochemicals (Rockford, IL). The antibodies utilized in this study and their suppliers were; rabbit polyclonal Gαi1 antibody was from Santacruz Biotechnology Inc. (Santa Cruz, CA), mouse monoclonal β-tubulin antibody was from BD Biosciences (San Diego, CA), rhodamine-conjugated anti-mouse antibody and the FITC-conjugated anti-rabbit antibody was from Pierce Biochemicals (Rockford, IL). Cell lines ---------- The parental human ovarian carcinoma cells, 2008 and its taxol-resistant derivatives (2008/13/4 and 2008/17/4) were maintained as described previously \[[@B12]\]. The 2008/13/4 and the 2008/17/4 cells were derived by exposure of the 2008 cells to stepwise increasing concentrations of taxol \[[@B12]\]. The 2008/13/4 cells are cross-resistant to other antimitotic drugs (*viz*., vincristine, vinblastine) and etoposide but are not cross-resistant to adriamycin and cisplatin. The 2008/13/4 cells have no defect in the transport of taxol, express negligible levels of P-glycoprotein and are 252-fold resistant to taxol compared to the 2008 cells. In contrast the \> 1500-fold taxol-resistant 2008/17/4 cells display a classic multidrug resistant phenotype with cross-resistance to all natural product drugs tested, expression of high levels of P-glycoprotein and a significant defect in the intracellular accumulation of taxol. However, the binding of taxol to microtubules is similar in the taxol-resistant and the parental cells \[[@B12]\]. Intracellular cAMP levels ------------------------- The intracellular cAMP levels in the 2008, 2008/13/4 and 2008/17/4 cells were determined using an acetylated version of a cAMP measurement kit from Biomol Research Laboratories (Plymouth Meeting, PA). A standard curve of known concentrations of acetylated cAMP was generated for each experiment. The cAMP level (pmol/mg protein) in the taxol resistant cells was compared with those observed in the 2008 cells considered as 1. cAMP-dependent PKA activity --------------------------- PKA activity was determined using the Pierce Colorimetric Assay Kit (Rockford, IL) that utilizes a fluorescent-labeled Kempeptide (a PKA-specific peptide (LRRASLG) substrate). The parental and taxol-resistant cells were seeded at a density of 2 × 10^6^cells/100 mm plate and incubated for 24 hr. Thereafter, the cells were either left untreated or treated with varying concentrations of taxol for the indicated time periods. Untreated and taxol treated cells were homogenized in a Dounce homogenizer (10 strokes with a tight fitting pestle) in a buffer containing 25 mM Tris-HCl, pH 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 10 mM DTT and protease inhibitor cocktail. The homogenate was centrifuged at 13000 × *g*for 20 min and the PKA activity was measured (in triplicate) in the supernatant fraction as per the instructions of the kit supplier. Values shown are mean ± SD of 3 separate experiments. Protein preparations -------------------- The 2008, 2008/13/4 and 2008/17/4 cells were seeded at a density of 1 × 10^6^/ml and incubated under normal growth conditions for 36 hours. Thereafter, the cells were left untreated or treated with varying concentrations of taxol or vincristine for the indicated time periods. At the end of each time period the cells were washed with chilled PBS (3 ×) and various protein fractions were extracted as outlined below. Whole cell lysate ----------------- Whole cells lysate was prepared from each of the cell line by scraping cells into a buffer containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% (v/v) Triton-X-100, 0.5% (v/v) Nonidet P40, 2.5 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 50 mM sodium fluoride and 1× protease inhibitor cocktail and incubating on ice for 15 minutes. Then, the lysate was centrifuged at 13,000 × *g*for 20 min and the supernatant was transferred to a fresh tube and stored at -80°C until use. Proteins were separated on a SDS-PAG and transferred to a PVDF membrane. Western blotting analysis was performed using rabbit polyclonal antibody against Gαi1 (Santacruz Biotechnology Inc., Santa Cruz, CA) and enhanced chemiluminescence reagents from Pierce Biochemicals (Rockford, IL). Cytoskeletal protein fraction ----------------------------- The cytoskeletal fraction was isolated essentially as described previously \[[@B22]\]. Briefly, the attached cells will be incubated in a microtubule-stabilizing buffer (0.1 M PIPES, 1 mM EGTA, 1 mM MgSO~4~, 2 M glycerol, pH 8.0) for 20 min. Thereafter, the cytosolic protein were removed by incubating the cells in the microtubule stabilizing buffer containing 0.1% NP-40 and protease inhibitor cocktail for 20 min. The cytoskeletal fraction (attached to the plastic dishes) was scraped into RIPA buffer (PBS containing 1% NP-40, 0.5% sodium deoxycholate and 0.1% SDS) containing protease inhibitor cocktail. This selective extraction procedure has been shown to be ideal for preserving the association between the cytoskeleton and other associated regulatory molecules \[[@B22]\]. The cytoskeletal fraction (5 μg protein/lane) was then subjected to SDS-PAGE. The separated proteins were transferred to PVDF membranes and Western blotting was performed using the polyclonal antibody against Gαi1 as described earlier. Subcellular protein fractions ----------------------------- The cells were processed for the extraction of cytoplasmic and nuclear fraction utilizing the protocol supplied by manufacturer of the NE-PER extraction kit (Pierce Biochemicals, Rockford, IL). The membrane fraction was extracted utilizing the single step Mem-PER extraction kit and protocol supplied by the manufacturer (Pierce Biochemicals, Rockford, IL). Protein (5 -- 20 μg/lane) from each fraction were resolved on a 12% (w/v) SDS-PAG and then transferred to a PVDF membrane. Western blotting was performed utilizing the Gαi1 polyclonal antibody (Santacruz Biotech, Inc., Santacruz, CA) as described above. Northern blotting analysis -------------------------- Total RNA (20 μg) extracted from each cell line was fractionated on 1% agarose-formaldehyde gel and then stained with ethidium bromide to ensure equal RNA loading. Thereafter the RNA was transferred to a Nylon membrane and probed with a full-length Gαi1 cDNA labeled with ^32^P-dCTP with a random priming kit from Amersham Biosciences (Piscataway, NJ). Hybridization was performed at 55°C overnight and nonspecific binding was removed by washing the nylon membrane in a high stringency buffer at 55°C for 30 min. Immunolocalization of Gαi1 -------------------------- To immunolocalize the Gαi1 protein, the 2008, 2008/13/4 and 2008/17/4 cells were grown on tissue culture-treated slides. After 48 hr, cells were treated with either taxol or vincristine (50 nM in case of 2008 cells, and 5 μM in case of the 2008/13/4 and 2008/17/4 cells) for 4 hr and 24 hr. At the end of each time period, the cells were rapidly rinsed in chilled PBS and then fixed with 4% (w/v) paraformaldehyde and processed for immunostaining. After blocking nonspecific binding sites by incubating the slides with 5% (v/v) normal goat serum, the slides were incubated with primary antibody directed against Gαi1 (1:100 rabbit polyclonal) and β-tubulin (1:200 mouse monoclonal) for 1 hr. After washing in chilled PBS (3×), the slides were incubated with FITC-conjugated anti-rabbit antibody or rhodamine-conjugated anti-mouse antibody for 30 min (Pierce Biochemicals, Rockford, IL). At the end of the incubation, the slides were washed again in PBS and mounted in media containing anti-fade. Localization of Gαi1 and β-tubulin was accomplished using an Olympus Confocal microscope. Abbreviations ============= Taxol, Paclitaxel; P-gp, P-glycoprotein; Gαi1, guanine nucleotide binding protein, subunit alpha i1; PKA, Protein kinase A; MAPK, Mitogen-activated protein kinase; PBS, phosphate-buffered saline; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; MDR, multidrug-resistance. Competing interests =================== The author(s) declare that they have no competing interests. Acknowledgements ================ Supported by NIH Grant RO1-CA76400 to H.K.P

Introduction {#sec1_1} ============ Dimebon is an over-the-counter drug that has been used as a non-selective oral antihistamine from 1983 and until more selective agents became available. In addition to its major activity as an inhibitor of histamine H1 and H2 receptors \[[@B1]\], dimebon has been reported to inhibit butyrylcholinesterase and acetylcholinesterase, voltage-gated calcium channels, N-methyl-*D*-aspartatic acid, adrenergic receptors and serotonin receptors and mitochondrial permeability transition pore opening, as well as to stabilize glutamate-induced Ca^2+^ signals and enhance hippocampal neurogenesis \[[@B2],[@B3],[@B4],[@B5],[@B6]\]. Consequently, it has been suggested that dimebon may be used as a neuroprotective drug slowing down the progression of diseases characterized by gradual deprivation of cognitive functions. Indeed, dimebon demonstrated cognition and memory-enhancing properties in rats treated with the neurotoxin AF64A, a compound that selectively destroys cholinergic neurons, and it protected cerebellar neurons in cell culture against beta-amyloid fragment-caused neurotoxicity \[[@B7]\]. In clinical trials dimebon improved cognition and memory of Alzheimer\'s disease \[[@B8]\] and Huntington\'s disease \[[@B9]\] patients. The pathogenesis of these two etiologically and clinically diverse neurodegenerative disorders shares a common feature, namely accumulation and deposition of aggregation-prone proteins (beta-amyloid and huntingtin, respectively) and therefore, they belong to a class of diseases known as proteinopathies \[[@B10]\], which also include Parkinson\'s disease (PD). A suggestion that dimebon may interfere with the process of protein aggregation and deposition instigated studies of the effect of this drug in various models of proteinopathies. It has been demonstrated that dimebon is capable of inhibiting the formation of TDP-43 inclusions in cellular models \[[@B11]\] and the formation/clearance of pathological deposits in a mouse model of gamma-synucleinopathy \[[@B12]\]. One of the most prominent features of PD is aggregation of alpha-synuclein with formation of toxic intermediates and characteristic cytoplasmic deposits of final aggregation products \[[@B13]\]. However, the effect of dimebon on alpha-synuclein aggregation in models of this neurodegenerative disorder or on disease progression in PD patients has not been assessed. The aim of the present study was to check whether dimebon is able to ameliorate the pathology caused by accumulation of alpha-synuclein in vivo. Transgenic mice expressing mutated forms of human alpha-synuclein, which cause neuronal dysfunction through a toxic gain-of-function mechanism, are the most adequate model systems for such study. It is beyond doubt that successful neuroprotective therapies would be most effective when used at an early clinical or even preclinical stage of the disease when it is still possible to slow down or reverse its development. Thus, it seemed reasonable to assess the effect of dimebon in a mouse model replicating early stages of PD pathology. Mice expressing the truncated (lacking the C-terminal 20 amino acids) form of human alpha-synuclein under control of tyrosine hydroxylase (TH) promoter on the endogenous alpha-synuclein null background, further referred to as aSyn (1--120) TG, recapitulate features of early-stage PD. Adult mice show no marked loss of dopaminergic neurons and only minor defects in the nigrostriatal system \[[@B14]\]. The late-onset pathology includes rare human alpha-synuclein-positive neuronal inclusions, reduction in striatal dopamine content associated with decreased locomotion and progressive reduction of striatal dopamine release associated with SNARE protein redistribution and synaptic alpha-synuclein accumulation \[[@B15]\]. However, these animals do not develop severe motor deficit characteristic for advanced stages of PD \[[@B13]\]. Olfactory impairment is observed in 70--95% of patients with PD and precedes the severe motor symptoms by up to 4 years \[[@B16],[@B17],[@B18]\]. In aSyn (1--120) TG mice the highest levels of exogenous transgene expression and accumulation were detected in the olfactory bulb (OB) dopaminergic neurons \[[@B14]\]. The aSyn (1--120) TG mice represent a model mirroring the earliest alterations in the dopaminergic systems that precede significant neuronal loss in PD. Here, we use this model to study the effects of chronic dimebon treatment on biochemical and histological markers of pathology induced by expression of mutated human alpha-synuclein. Animals and Methods {#sec1_2} =================== Animals {#sec2_1} ------- The transgenic mouse line aSyn (1--120) TG was created on the alpha-synuclein null background \[[@B18]\] as described previously \[[@B14],[@B15]\]. Only male mice were used in this study. Experimental mice were housed one per cage with a normal light/dark cycle and ad libitum access to food and water. All animal experiments were carried out in accordance with the UK Animals (Scientific Procedures) Act, 1986. The minimal required number of animals was used and all efforts were made to minimize animal suffering. Dimebon Administration {#sec2_2} ---------------------- Dimebon (3,6-dimethyl-9-(2-methyl-pyridyl-5)-ethyl-1,2,3,4-tetrahydro-γ-carboline dihydrochloride) was obtained from OJSC Organika (Novokuznetsk, Russia). A cohort of aSyn (1--120) TG male mice received dimebon with the drinking water (10 μg/ml solution that was changed 3 times a week) from the age of 3 months. Control littermate animals received drinking water without dimebon. At the age of 14 months mice were behaviorally tested and tissues were collected for further analysis. Behavioral Tests {#sec2_3} ---------------- *Home-Like Cage Activity.* Spontaneous locomotor activity was assessed in the home-like cage equipped with a system of infrared beams to monitor horizontal movements. Animals were monitored for 24 h with a normal light/dark cycle and ad libitum access to food and water. Exploratory activity in the novel environment was analyzed for 2 h after placing mice in the home-like cage for the first time. Data (number of beam breaks) were collected at 15-min intervals. *Rotarod Test.* Animals were introduced to the rotarod (UGO Basil 7560, Italy) for 5 min and tested using an accelerating (from 4 to 40 rpm over 300 s) mode later on the same day. Each animal was tested 3 times with a 30-min break between tests. The mean of latency to fall for these three trials was included in final statistics. Results were expressed as mean ± SEM for each experimental group. *Inverted Grid Test.* Balance and coordination were assessed as the ability of animals to stay on the inverted grid. Mice were placed on a 30 × 30 cm square mesh consisting of 5-mm squares of 0.5-mm diameter wire. The grid was slowly rotated to inverted position and held 30 cm above the surface of a thick layer of bedding material. The latency to fall (in seconds) was noted. The maximum test time was 60 s. The best time from three attempts was included in the statistics. Immunohistochemistry {#sec2_4} -------------------- For the quantification of dopaminergic neurons in the substantia nigra pars compacta (SNpc) and ventral tegmental area, brains of control and experimental animals were simultaneously collected, processed, embedded and stained with antibody against TH as described previously \[[@B20],[@B21]\]. Alexa Fluor-conjugated goat anti-rabbit and anti-mouse antibodies (Invitrogen) were used as secondary antibodies in double-immunofluorescence experiments. Neuronal Cell Counts {#sec2_5} -------------------- Dopaminergic neurons in the SNpc and ventral tegmental area of experimental and control mice were stereologically counted using fractionator method and the Axiovision imaging program (Carl Zeiss Vision) as described in our previous publications \[[@B20],[@B21]\]. Antibodies {#sec2_6} ---------- Mouse monoclonal antibody against TH (clone TH-2, Sigma) was used for staining of histological sections in 1:1,000 dilution. Mouse monoclonal antibody Syn-1 (Transduction Laboratories) was used in 1:1,500 dilution for detection of alpha-synuclein on immunoblots and in 1:100 dilution for double-immunofluorescence analysis together with rabbit polyclonal antibody against TH (H-196, Santa Cruz Biotechnology) diluted 1:100. High-Pressure Liquid Chromatography Analysis {#sec2_7} -------------------------------------------- Monoamines were extracted by homogenization of dissected mouse dorsal striata in 0.06 *M* HClO~4~. Concentrations of striatal dopamine and 3,4-dihydroxyphenylacetic acid (DOPAC) were measured by high-pressure liquid chromatography (HPLC) with electrochemical detection using a 4.6 × 150 mm Microsorb C18 reverse-phase column (Varian) and Decade II ECD with a Glassy carbon working electrode (Antec Leyden) set at +0.7 V with respect to a Ag/AgCl reference electrode. The mobile phase consisted of 12% methanol (v/v), 0.1 *M* monosodium phosphate, 2.4 m*M* 1-octane sulphonic acid, 0.68 m*M* EDTA, pH 3.1. Immunoblotting {#sec2_8} -------------- Olfactory bulbs were dissected and homogenized directly in SDS-PAGE loading buffer followed by incubation for 10 min at 100°C. Protein concentration was measured, and equal amounts of total protein were run on 14% SDS-PAGE and transferred to PVDF membrane using iBlot technology according to manufacturer\'s (Invitrogen) instructions. Membrane blocking, incubation with antibodies, washes and enhanced chemiluminescence detection were carried out as described previously \[[@B20],[@B21]\]. RNA Expression Analysis {#sec2_9} ----------------------- Total RNA extraction, first-strand cDNA synthesis and quantitative PCR on a StepOne Real-Time PCR System (Applied Biosystems) were performed as described previously \[[@B22]\]. Primers 5′-GAATCCTGGAAGACATGCCT-3′ and 5′-AGGCTTCAGGCTCATAGTCT-3′ were used for amplification of alpha-synuclein cDNA. For amplification of GAPDH mRNA (normalization standard) primers 5′-CACTGAGCATCTCCCTCACA-3′ and 5′-GTGGGTGCAGCGAACTTTAT-3′ were used. Fold change was determined by 2^−ΔΔCT^ method \[[@B23]\] using StepOne v2.0 software. Statistical Analysis {#sec2_10} -------------------- Allcalculations were performed using STATISTICA 6.0 software. Non-parametric Kruskal-Wallis one-way ANOVA on ranks and Mann-Whitney U test were used to assess differences between experimental groups. Results {#sec1_3} ======= Effect of Chronic Dimebon Treatment on Mouse Locomotion, Exploratory Activity, Balance and Coordination {#sec2_11} ------------------------------------------------------------------------------------------------------- Analysis of dimebon-treated and control mice locomotion in a home-like cage during a 24-hour period revealed no statistically significant difference (p = 0.662) between the groups (fig. [1A](#F1){ref-type="fig"}). This indicates that chronic dimebon treatment does not affect spontaneous locomotor activity of aSyn (1--120) TG mice. However, dimebon-treated transgenic mice were significantly (p = 0.019) more active than control transgenic mice in a novel non-anxiogenic environment during the first 15 min of habituation, when their activity was similar to activity of wild-type mice (fig. [1B](#F1){ref-type="fig"}). Balance and coordination of transgenic mice that did or did not receive dimebon and age-matched wild-type animals were assessed in inverted grid and accelerating rotarod tests (fig. [1C, D](#F1){ref-type="fig"}). No difference in the ability to stay on the inverted grid was observed between the three groups (p \> 0.05). Although 14-month-old aSyn (1--120) TG mice noticeably underperformed in the rotarod test compared to wild-type animals of the same age (p = 0.030), chronic administration of dimebon did not improve their performance (p = 0.806). Dopaminergic Neurons in the SNpc and Dopamine Content in the Dorsal Striatum of Dimebon-Treated Alpha-Synuclein Transgenic Animals {#sec2_12} ---------------------------------------------------------------------------------------------------------------------------------- It has previously been reported that expression of truncated human alpha-synuclein in dopaminergic neurons results in the development of pathological changes in the SNpc of aSyn (1--120) TG mice by the age of 12--14 months \[[@B14]\]. These changes include shrunken perikarya, beaded or dystrophic processes, inclusions and microgliosis, and may lead to progressive dysfunction of dopaminergic neurons. Therefore, we assessed effects of chronic dimebon administration on the number of dopaminergic neurons in the SNpc and dopamine content in the dorsal striatum of aSyn (1--120) TG mice. Stereological counting of TH-positive neurons showed that both dimebon-treated and non-treated transgenic mice have the same number of dopaminergic neurons in the SNpc (fig. [2A](#F2){ref-type="fig"}) as wild-type mice. This is consistent with the earlier observation that transgene expression does not cause detectable loss of TH-positive neurons in the SNpc of aSyn (1--120) TG mice \[[@B24]\] and demonstrates that dimebon has no effect on the survival of SNpc or ventral tegmental area (online supplementary fig. [1](#S1){ref-type="supplementary-material"}, see [www.karger.com/doi/10.1159/000324989](http//www.karger.com/doi/10.1159/000324989)) dopaminergic neurons in adult animals. In contrast to unchanged number of neurons, the tissue content of dopamine and its metabolites is significantly reduced in the dorsal striatum of adult and aging aSyn (1--120) TG mice \[[@B14]\], which reflects the compromised function of the dopaminergic system in these mice \[[@B15]\]. Our HPLC analysis demonstrated the same dopamine and DOPAC content in the dorsal striatum of dimebon-treated and non-treated transgenic mice (fig. [2B, C](#F2){ref-type="fig"}), suggesting that dimebon does not improve dopaminergic neuron function in this animal model. Dimebon Does Not Alter Transgene Expression, Accumulation or Intracellular Localization of Alpha-Synuclein in Dopaminergic Neurons of the Olfactory Bulb {#sec2_13} -------------------------------------------------------------------------------------------------------------------------------------------------------- In aSyn (1--120) TG mice, the highest level of exogenous alpha-synuclein expression driven by the TH promoter was observed in the OB. To assess whether dimebon might influence the expression of transgene in neurons we studied the levels of human alpha-synuclein mRNA by quantitative RT-PCR and protein by immunoblotting in the OB of dimebon-treated and control aSyn (1--120) TG mice. For these experiments OBs of 6 dimebon-treated transgenic mice were combined in one group (D1) and OBs of another 4 similarly treated animals in another group (D2). Similarly, OBs of 4 and 5 non-treated transgenic mice were combined in groups C1 and C2. Total RNA and total proteins form each group of combined tissues were extracted and analyzed. The same level of human alpha-synuclein mRNA was detected in all four groups (fig. [3A](#F3){ref-type="fig"}). On immunoblots, the intensity of bands corresponding to the 12-kDa truncated form of human alpha-synuclein protein was also very similar in four samples (fig. [3B](#F3){ref-type="fig"}). These results clearly demonstrate that dimebon does not affect transgene expression from the TH promoter or accumulation of exogenous alpha-synuclein in dopaminergic neurons of the OB. In contrast to predominantly presynaptic compartmentalization of endogenous full-length alpha-synuclein and its exclusion from the perikaryon of dopaminergic neurons in the glomerular layer of OB, ectopic accumulation of truncated human alpha-synuclein was observed in the cell bodies of these neurons (fig. [3C](#F3){ref-type="fig"}). However, simultaneous immunofluorescent staining of histological sections through the OB of dimebon-treated and control transgenic mice revealed no changes in either localization or the amount of alpha-synuclein accumulated in the cell bodies (fig. [3D](#F3){ref-type="fig"}). Discussion {#sec1_4} ========== A disease-modifying effect of dimebon on the mild to moderate form of Alzheimer\'s disease and Huntington\'s disease has been demonstrated in phase II clinical trials \[[@B8]\]. Together with the recently shown ability of this drug to improve behavioral and histopathological impairments caused by pan-neuronal overexpression of gamma-synuclein \[[@B12],[@B22]\] these clinical data suggested that dimebon might be efficient in the treatment of various types of proteinopathies, including PD. Therefore, the main aim of this study was to evaluate the effects of dimebon in a mouse model recapitulating several characteristics of early-stage PD because results of clinical trials suggested that dimebon might protect against early rather than advanced forms of neurodegeneration. Experimental animals had constant access to drinking water containing 10 μg/ml of dimebon that was changed every 60 h. Because of the stability of this compound in solutions \[[@B25],[@B26],[@B27]\] and the same rate of consumption for water and dimebon solution (average daily water consumption for mice in the control group was 5.2 ml and average daily dimebon solution consumption for mice in the experimental group was 5.4 ml) this protocol permitted to achieve a cumulative daily dose of ∼54 μg/30--35 g mouse or ∼1.5 mg/kg. This dose is close to the dose (1 mg/kg) used in the phase II clinical trials of dimebon for patients with Alzheimer\'s disease \[[@B8]\]. Moreover, recent studies have demonstrated that after oral administration of a single dose, dimebon levels in the plasma and in the brain of rodents are stable for at least 6 h \[[@B28]\] and therefore, our protocol of drug administration with drinking water supplied ad libitum assures uninterrupted exposure of the experimental animal brain to dimebon. Mice of the aSyn (1--120) TG line expressing a truncated form of human alpha-synuclein in catecholaminergic neurons have previously been shown to develop an age-dependent reduction in exploratory activity in a novel environment \[[@B14]\]. Here we confirmed this effect on a separate cohort of 14-month-old aSyn (1--120) TG mice and over a longer period of adaptation to a novel environment (120 min in contrast to 30 min in the previous study), and furthermore revealed that performance of these transgenic mice in the accelerated rotarod test is also markedly compromised when compared to the performance of age-matched wild-type animals. However, aging aSyn (1--120) TG mice performed very similarly to wild-type animals in the inverted grid test, suggesting that expression of the truncated form of human alpha-synuclein in dopaminergic neurons causes very selective changes of animal behavior. In PD patients more than 80% reduction of striatal dopamine is required for the development of motor symptoms, whereas aSyn (1--120) TG mice display only ∼30% reduction \[[@B14]\], suggesting that their compromised rotarod performance could not be explained simply by dopamine deficiency but probably involves other functional changes in dopaminergic neurons or their synapses caused by expression of truncated human alpha-synuclein \[[@B15]\]. Thus, two behavioral parameters that appeared to be compromised in aSyn (1--120) TG mice could be used for assessing the effect of dimebon on alpha-synuclein-induced pathology. Transgenic mice chronically treated with dimebon for 11 months from the age of 3 months showed no improvement in the accelerated rotarod test compared to untreated animals of the same genotype. This correlated with the lack of the effect of dimebon on decreased levels of dopamine and DOPAC in the striatum of aSyn (1--120) TG mice. The general activity of aging aSyn (1--120) TG mice in the home-like cage over a 24-hour period was not affected by dimebon. However, immediately after placing in a novel environment, dimebon-treated animals were significantly more active than untreated transgenic animals, suggesting that dimebon intensifies exploratory activity, which is attenuated in 14-month-old aSyn (1--120) TG mice. As exploratory behavior depends on the coordinated function of many brain networks rather than solely the dopaminergic system, this effect might reflect a general improvement of aging brain function by dimebon as suggested in previous studies \[[@B6],[@B7],[@B8],[@B29]\]. Another prominent feature of the aSyn (1--120) TG model is the ectopic accumulation of transgenically expressed truncated human alpha-synuclein in cell bodies of dopaminergic neurons. This was most obvious in neurons of the glomerular layer of OB and therefore these cells seemed an appropriate system for assessing the effect of dimebon treatment on alpha-synuclein accumulation and intraneuronal localization. However, neither expression analysis nor the comparison of histological sections through the OB revealed differences in the tissue level or in the pattern of human alpha-synuclein immunostaining in TH-positive neurons of dimebon-treated and untreated aSyn (1--120) TG mice. We also did not observe any effect of dimebon on micro- or astrogliosis in the OB or other neural tissues of these mice. In conclusion, we demonstrated a very limited effect of chronic dimebon treatment on the biochemical, histological and behavioral alterations caused by expression of truncated human alpha-synuclein in dopaminergic neurons of transgenic mice. Although these negative results question the value of dimebon as a potential treatment for PD and other alpha-synucleinopathies, it should be noted that aSyn (1--120) TG mice recapitulate only some changes typical of the early stages of these diseases and it is possible that this drug might affect other manifestations, for example, those associated with late disease stages. Further studies are required to reveal whether dopaminergic neurons are less sensitive to this drug than other previously studied types of neurons or whether dimebon does not affect pathways associated with alpha-synuclein toxicity in any other types of cells and at any stage of pathology. Supplementary Material ====================== ###### supplementary fig. ###### Click here for additional data file. This work was supported by Research Grants from the Russian Foundation for Basic Research, RAS 'Fundamental Science for Medicine', Alzheimer\'s Society and The Wellcome Trust (075615/Z/04/z). {#F1} {#F2} {#F3}

Introduction {#section1-2050312119865114} ============ Breast cancer (BC) is one of the most frequently occurring cancers in females and represents a significant public health concern. The American Cancer Society estimates that one in eight females will develop BC at some point in their lives, with the incidence increasing with age.^[@bibr1-2050312119865114]^ Significant geographic and ethnic-specific differences in both incidence and mortality rates are reported. This is in part attributed to sociodemographic factors which influence adherence to recommendations for early screening for BC, as well as the likelihood of seeking appropriate medical advice upon detection of a breast mass.^[@bibr2-2050312119865114],[@bibr3-2050312119865114]^ The aetiology behind BC is equally complex and involves interactions between environmental, lifestyle and genetic factors that collectively determine cancer risk. BC typically arises when cells lose the ability to halt the process of proliferation, coupled to resistance or reduction in the process of cell death by apoptosis. BC cells express high levels of phosphatidylinositol-3-kinase (PI3K)/Akt and mammalian target of rapamycin (mTOR) signalling molecules, which impairs their ability to undergo apoptosis.^[@bibr4-2050312119865114]^ Pathologically, BC is classified either as invasive or non-invasive type. The non-invasive subtypes include ductal and lobular carcinoma in situ, whereas, ductal and lobular carcinomas are considered as invasive subtypes. On average, ductal carcinoma accounts for 80% of reported cases in females, whereas lobular carcinoma accounts for only 5%--10% of the cases.^[@bibr5-2050312119865114]^ Currently, BCs are treated either surgically or via chemoradiotherapy, in addition to the use of Trastuzumab (Herceptin^®^) for HER2^+^ tumours.^[@bibr6-2050312119865114]^ Some cases of BC do not respond well to traditional treatment, particularly in diabetics. Recent studies have shown that metformin, a primary anti-diabetic agent, confers anti-tumorigenic effects on cancer cells and can be considered as a potential adjuvant in the management of BC.^[@bibr7-2050312119865114]^ A number of population-based observational studies had initially suggested that metformin reduces cancer incidence and/or mortality among type 2 diabetic patients, however, no causal relationship can be established from epidemiologic data alone.^[@bibr8-2050312119865114]^ Preclinical research using both BC cell lines and mouse models subsequently showed that metformin represses cancer cell and xenograft growth.^[@bibr9-2050312119865114][@bibr10-2050312119865114]--[@bibr11-2050312119865114]^ These effects are achieved through various mechanisms, including cell cycle arrest, apoptosis, AMP-activated protein kinase (AMPK) activation and mTOR inhibition. In addition, metformin exerts in vitro chemo preventive effects through modulation of cytochrome P4501A1 (CYP1A1)/aryl hydrocarbon receptor (AhR) pathways.^[@bibr12-2050312119865114]^ The anti-tumour effects of oral anti-diabetic therapy are not restricted to metformin, as thiazolidinediones (synthetic ligands of peroxisome proliferator-activated receptors γ -- PPARγ) possess similar properties. Animal studies have shown that pioglitazone inhibits chemical carcinogenesis in rats.^[@bibr13-2050312119865114]^ Human studies demonstrate that rosiglitazone reduces BC risk in females with type 2 diabetes, and that this effect is enhanced by metformin.^[@bibr14-2050312119865114]^ The pleiotropic oncostatic effects of oral anti-diabetic drugs is reinforced by meta-analysis showing that thiazolidinediones are associated with a lower incidence of cancer, particularly colorectal and breast tumours.^[@bibr15-2050312119865114]^ Furthermore, metformin also decreases the development of resistance in BC cells, thereby allowing current chemotherapy agents to work synergistically with metformin.^[@bibr16-2050312119865114]^ It also blocks the two cellular pathways for nicotinamide adenine dinucleotide (NAD^+^) regeneration, which then results in a complete loss of cells' NAD^+^ recycling capacity. The resulting depletion of NAD^+^, in turn, leads to cell death.^[@bibr17-2050312119865114]^ This article aims to provide an overview of the pathomolecular mechanisms in which metformin may elicit its anti-cancerous effects and discuss its potential role as an adjuvant in the management of BC. Pathophysiology of BC {#section2-2050312119865114} ===================== Inflammation and neoplastic transformation in BC {#section3-2050312119865114} ------------------------------------------------ The importance of the immune response in BC development and progression has been well documented. DeNardo and Coussens^[@bibr18-2050312119865114]^ highlight a possible immunological connection between BC and Th2 inflammatory cells that results in the promotion of tumour development and disease progression, whereas acute anti-tumour responses involving cytolytic T lymphocytes appear to protect against tumour development. Physiologically, injured tissues or cells exposed to chemical irritants are eliminated by apoptosis. This is followed by enhanced cell proliferation to facilitate tissue regeneration and re-establish tissue function. Moreover, proliferation and inflammation may persist until the insulting agent is removed, allowing the tissue to heal completely. If the inflammation persists, cells may undergo dysplastic changes, which then increases the risk of neoplasia.^[@bibr19-2050312119865114]^ The role of leukocytes, especially the cytotoxic T lymphocytes in tumorigenesis, has been explored extensively. These cells are believed to assist in the eradication process of neoplastic cells with the help of natural killer (NK) cells.^[@bibr20-2050312119865114]^ Nevertheless, T-cell infiltration in invasive BC has been reported, especially the activated CD4^+^ Th1 polarised cells that secrete several inflammatory cytokines -- including IFNγ, transforming growth factor beta (TGF-β), tumour necrosis factor alpha (TNFα) and interleukin-2 (IL-2). These cytokines then interact with other cytotoxic T-cells and upregulate the MHC class I and II molecules, as well as other antigen display co-factors in neoplastic cells.^[@bibr18-2050312119865114],[@bibr20-2050312119865114]^ This process is an essential part of immune-mediated anti-tumorigenic effects. Conversely, activation of Th2-polarised CD4^+^ T-helper cells results in expression of inflammatory cytokines (IL-4, IL-5, IL-6, IL-10, and IL-13), which then enhances humoral immunity and downregulates cell-mediated anti-tumour immunity; thereby, promoting the pro-tumour humoral response.^[@bibr18-2050312119865114],[@bibr21-2050312119865114][@bibr22-2050312119865114][@bibr23-2050312119865114]--[@bibr24-2050312119865114]^ Neoplastic transformation is a complex multistage event {#section4-2050312119865114} ------------------------------------------------------- BC originates in the undifferentiated lobules type 1, which are composed of three cell types: the highly proliferating cells (ER^−^), non-proliferating cells (ER^+^) and very few ER^+^ cells that proliferate. Endogenous 17 beta-oestradiol (E2), when metabolised by cytochrome P450 enzymes may also act as a carcinogen which ultimately leads to genomic changes and transformation phenotypes observed in spontaneously developing primary BCs. Endogenous E2 is metabolised by P450 cytochromes that also activate benzo\[a\]pyrene B\[a\] a carcinogen present in cigarette smoke.^[@bibr25-2050312119865114]^ The genomic alterations induced by E2 and B\[a\]P in vitro are also observed in ductal hyperplasia DCIS and invasive ductal carcinoma. Transcriptional repressors, such as Polycomb Group Protein (EZH2), which traditionally controls the cellular memory have been linked to cancer cell invasion and BC progression. Kleer et al.^[@bibr26-2050312119865114]^ demonstrated that EZH2 protein levels were strongly associated with BC aggressiveness. Moreover, EZH2 overexpression promoted anchorage-independent growth and cell invasion through the SET domain and histone deacetylase activity. Dysregulated cellular memory, transcriptional repression and neoplastic transformation are interlinked, and EZH2 may be a marker for aggressive BC and neoplastic transformation. The actual neoplastic transformation process involved in BC is more complex than previously thought and warrants more long-term molecular studies to better understand the actual transformation process and ways to halt such process. Type 2 diabetes mellitus and cancer {#section5-2050312119865114} =================================== Type 2 diabetes mellitus (T2DM) is a metabolic disorder which is associated with several cancers. It is characterised by hyperglycaemia, insulin resistance and hyperinsulinemia. These factors interact to promote cell proliferation through the mitogenic effect of the insulin receptor and insulin-like growth factors (IGFs), while hyperglycaemia provides the metabolic substrate for cell proliferation.^[@bibr27-2050312119865114]^ Overexpression of the insulin growth factor receptor-1R (IGF-1R) or insulin receptors leads to mitogenic signalling, which increases activation of phosphoinositide 3-kinase (PI3)-Akt-mTOR signalling pathways. Excess adiposity increases local production of oestrogen via the enhanced activity of aromatase, which augments oestrogen receptor alpha signalling (ER-α) in tumour cells. The inflammatory effect of hyperinsulinemia, in addition to increasing production of local cytokines, may lead to an increased susceptibility to cancer development in diabetes.^[@bibr28-2050312119865114]^ A number of large-scale epidemiological studies and meta-analyses have reported an increase in the incidence of several cancers in T2DM.^[@bibr29-2050312119865114],[@bibr30-2050312119865114]^ A population-based cohort study by Ballotari et al.^[@bibr31-2050312119865114]^ showed a higher cancer incidence in subjects with diabetes. This relationship was only observed in those with T2DM, but not in Type 1 diabetes mellitus (T1DM) and was attributed to obesity. Notably, the risk was higher among insulin users. An increased risk of cancer at several tissues, including liver, pancreas, endometrium, colorectum, breast and bladder has been described in multiple similar studies. Notably, these observations could be either causal -- driven by the metabolic disturbances in diabetes or else due to the confounding effects of the underlying excess adiposity in diabetes. Tsilidis et al.^[@bibr29-2050312119865114]^ show that individual studies are, however, characterised by substantial heterogeneity, small study effects and excess significance, with 28% (135/474) of studies adjusting risk estimates either for age or gender. Despite the evidence from epidemiological studies linking diabetes to cancer incidence, the specific mechanisms driving this association are not fully understood. T2DM and BC {#section6-2050312119865114} =========== Mechanisms behind T2DM and BC {#section7-2050312119865114} ----------------------------- Studies have shown that BC in women with diabetes is often diagnosed at an advanced stage compared with women without diabetes.^[@bibr32-2050312119865114],[@bibr33-2050312119865114]^ Furthermore, the overall mortality following BC diagnosis is 30%--60% higher in women with diabetes compared with women without diabetes after adjusting for tumour stage.^[@bibr34-2050312119865114],[@bibr35-2050312119865114]^ A cross-sectional study by Bronsveld et al.^[@bibr34-2050312119865114]^ also showed no relation between diabetic status and tumour morphology and grade. However, premenopausal diabetic women tended to develop breast tumours that do not express progesterone receptor and HER2, which are typically associated with poor prognosis. No association between insulin therapy and clinicopathological subtypes was noted, even though insulin use in T2DM may induce oestrogen (ER) and progesterone receptors expression.^[@bibr36-2050312119865114]^ Conversely, a systematic review of in vitro, animal and human studies found no evidence of increased BC risk with commercially available insulin analogues and human insulin.^[@bibr37-2050312119865114]^ Conflicting findings were reported by other investigators. Tseng,^[@bibr38-2050312119865114]^ showed that prolonged use of insulin carries a significantly higher BC risk. A recent study by Overbeek et al.^[@bibr39-2050312119865114]^ showed that females with T2DM were at an increased risk of being diagnosed with a more aggressive type of BC than non-T2DM females. Interestingly, exogenous insulin administration was not associated with the development of more advanced BC tumours in this study. These findings suggest that insulin may not be involved directly in the development of BC. Instead, it may promote BC progression by upregulating mitogenic signalling pathways.^[@bibr37-2050312119865114]^ The precise mechanisms linking T2DM to BC progression remains uncertain, but is believed to involve insulin-like growth factor-1 (IGF-1). IGF-1 pathways are activated by a high concentration of insulin, which then goes on to promote cancer development via the insulin/IGF-1 hybrid receptors. These have a higher affinity for IGF-1 than for insulin and are overexpressed in BC tissues of T2DM patients.^[@bibr40-2050312119865114][@bibr41-2050312119865114]--[@bibr42-2050312119865114]^ Nevertheless, due to insufficient evidence on the specific oncogenic pathways connecting hyperinsulinemia to BC, it is difficult to ascertain the role of insulin in the development of BC in premenopausal and postmenopausal diabetic females. Oestrogen, diabetes and BC {#section8-2050312119865114} -------------------------- Epidemiological and clinical studies have shown that T2DM is a risk factor for BC and is consequently associated with poor prognosis.^[@bibr43-2050312119865114]^ Wairagu et al.^[@bibr44-2050312119865114]^ investigated the effects of oestradiol on MCF-7 BC cells primed with and without insulin chronically. The study found that insulin priming was a prerequisite for oestradiol-induced growth in BC cells. The authors demonstrate that oestradiol exposure increases expression of cyclins A and B, which are both involved in cell cycle progression and leads to the activation of genes in the pentose phosphate and serine biosynthesis pathways. Oestradiol also increased anti-apoptotic Bcl-xL expression in the insulin-primed cells. In addition, metformin suppresses oestradiol-induced growth in the insulin-primed cells. Critically, this study showed that insulin priming dramatically sensitises BC growth to 100 pmol of oestradiol. Conversely, other studies have shown that at least 10--100 nM of oestradiol concentration is required before maximum cell growth is attainable in BC cells.^[@bibr45-2050312119865114],[@bibr46-2050312119865114]^ These findings suggest that insulin priming happens readily in diabetics as a result of the chronic hyperinsulinemic state even at physiological levels of oestradiol, thus exposing diabetics to an elevated risk of developing BC. Oestradiol modulates cell cycle and apoptotic processes in insulin-primed cells, which then further promotes cancer cell growth. Wairagu et al.^[@bibr44-2050312119865114]^ also showed that both insulin-primed and unprimed MCF-7 cells exposed to dihydrotestosterone (DHT) exhibit no growth response, which further indicates that there is crosstalk between insulin priming and ER-induced BC cell growth. In the insulin resistant state, suppression of sex hormone binding globulin (SHBG) increases the bioavailability of free oestrogen, leading to elevated levels of serum oestrogen.^[@bibr47-2050312119865114]^ Moreover, IGF-1 is known to increase the production of androgens, which may then subsequently displace oestrogen binding from SHBG.^[@bibr48-2050312119865114],[@bibr49-2050312119865114]^ Furthermore, IGF-1 can interact with 17-beta-oestradiol leading to increased proliferation of BC cells.^[@bibr50-2050312119865114]^ Therefore, altered levels of endogenous oestrogen may contribute to the proliferation of ER-positive BC in T2DM.^[@bibr42-2050312119865114]^ Since the prevalence of obesity is high in T2DM, elevated levels of oestradiol and oestrone can result from increased adipose tissues aromatase activity.^[@bibr51-2050312119865114]^ Also, hyperinsulinemia in T2DM may induce the expression and binding capacity of ER, which can subsequently enhance insulin mitogenic properties by promoting IRS-1 function, and through activation of PI3K and Ras/MAPK signalling.^[@bibr52-2050312119865114]^ The production of inflammatory mediators in T2DM, mainly TNF-α and interleukin-6 (IL-6), which are both associated with insulin resistance in diabetics, secondarily enhances the oestrogen synthesis in normal and BC cells. This further potentiates BC development.^[@bibr53-2050312119865114],[@bibr54-2050312119865114]^ Oxidative stress, diabetes mellitus and BC {#section9-2050312119865114} ------------------------------------------ Hyperglycaemia induces oxidative stress through direct or indirect pathways in BC cells by increasing levels of insulin/IGF-1 and inflammatory cytokines, particularly IL-6 and TNF-α.^[@bibr55-2050312119865114]^ Together, they activate nuclear factor kappa (NFκB), signal transducer activator of transcription 3 (STAT3) and the hypoxia-inducible factor 1-alpha (HIF1α).^[@bibr56-2050312119865114]^ These factors result in increased free radical production, leading to damage to DNA, lipids and further amplification of the inflammatory processes \[27\]. The reactive oxidative species derived may then initiate carcinogenesis by modifying the apoptotic responses, as well as disrupting cell anchoring sites and increasing angiogenesis.^[@bibr57-2050312119865114],[@bibr58-2050312119865114]^ In addition, studies have shown that hyperglycaemia also indirectly activates endothelial growth factor receptor (EGFR) via the Rho family GTPase Rac1 and cell division control protein 42 homolog (Cdc42), which then stimulates the cell proliferation, thus providing another mechanistic link between hyperglycaemia and tumorigenesis.^[@bibr59-2050312119865114]^ Hyperglycaemia leads to the modulation of various pathways that control cell proliferation, migration and invasion.^[@bibr60-2050312119865114]^ Cancer cells demonstrate enhanced glucose uptake and metabolism, a process referred to as the 'Warburg effect'. Hyperglycaemia thus provides the necessary fuel which the cancer cells require, and this then allows cancer cells to proliferate rapidly.^[@bibr61-2050312119865114]^ Hyperglycaemia also stimulates upregulation of protein kinase C (PKC), PPARs and proliferation in MCF-7 BC cell lines.^[@bibr62-2050312119865114]^ Hyperglycaemia also promotes BC cell migration via zinc and its transporters (ZRT/IRT-like protein 6, ZRT/IRT-like protein 10). High serum glucose leads to increased expression of zinc transporters (ZIP6 and ZIP10), which are essential for promoting cell migration and motility in BC cells.^[@bibr63-2050312119865114],[@bibr64-2050312119865114]^ These findings emphasise the importance of stringent control of glucose levels in both T2DM and BC in order to reduce cancer cell proliferation. The pharmacologic management of hyperglycaemia hinges on the use of sulphonylureas, metformin and insulin. Therapy that leads to hyperinsulinemia, such as sulphonylureas and exogenous insulin, are thought to increase the risk of cancer, while treatment that reduces insulin resistance, such as metformin, are thought to reduce the risk of cancer development. A meta-analysis investigating cancer risk associated with metformin and sulphonylureas in T2DM showed that use of metformin was associated with significantly decreased risk of all cancers. However, no evidence that use of metformin is associated with the risk of BC was derived.^[@bibr65-2050312119865114]^ This meta-analysis was characterised by extensive between-study heterogeneity and evidence of publication bias with regard to metformin. Hence long-term randomised double-blinded clinical trials are required to substantiate the benefit and efficacy of using anti-diabetic agents in BC treatment. Metformin and BC {#section10-2050312119865114} ================ Mechanisms of metformin action in normal cells {#section11-2050312119865114} ---------------------------------------------- Metformin (1,1-dimethyl biguanide) is a biguanide which acts by reducing hepatic glucose output and increasing insulin sensitivity. This results in a reduction in serum glucose levels without the risks of either hypoglycaemia or weight gain. Metformin also modulates multiple components of incretin pathways. It increases the plasma levels of glucagon-like peptide 1 (GLP-1) and induces islet incretin receptor gene expression via PPAR-α.^[@bibr66-2050312119865114]^ Metformin is taken up by hepatocytes via the organic cation transporter 1 (OCT1) and inhibits hepatic gluconeogenesis by modulating enzymes and substrate which are involved in the gluconeogenic pathways.^[@bibr67-2050312119865114][@bibr68-2050312119865114][@bibr69-2050312119865114]--[@bibr70-2050312119865114]^ The decrease in hepatic glucose production results in the activation of AMPK, which is a cellular metabolic sensor responsible for protecting cellular functions under low energy conditions.^[@bibr71-2050312119865114],[@bibr72-2050312119865114]^ AMPK is normally activated by an increase in the intracellular AMP: ratio, which results from an imbalance between the ATP production and consumption.^[@bibr71-2050312119865114]^ Upon activation, phosphorylation of AMPK by tumour suppressor serine/threonine kinase 11 (STK11/LKB1) and calcium/calmodulin-dependent protein kinase kinase-2 (caMKK-2) causes AMPK to switch cells from an anabolic to the catabolic state. In doing so, it shuts down the ATP-consuming pathways by inhibiting glucose, lipid, protein synthesis and cellular growth, whereas fatty acid oxidation, as well as glucose uptake, is stimulated to restore the AMP:ATP ratio.^[@bibr71-2050312119865114]^ Metformin primarily acts on the mitochondria by inducing mild and specific inhibition of mitochondrial respiratory-chain complex 1 (MRCC1), which is present in hepatocytes, skeletal muscles, endothelial cells, pancreatic beta-cells and neurons.^[@bibr73-2050312119865114][@bibr74-2050312119865114][@bibr75-2050312119865114][@bibr76-2050312119865114][@bibr77-2050312119865114][@bibr78-2050312119865114]--[@bibr79-2050312119865114]^ In addition, metformin also reduces mitochondrial reactive oxygen species (ROS) production by selectively inhibiting reverse electron flow through MRCC1.^[@bibr80-2050312119865114],[@bibr81-2050312119865114]^ ROS are important mediators of cell and genomic damage and play essential roles in the pathophysiology of both T2DM and BC. Inhibition of ROS generation through metformin may thus have benefits that extend beyond its traditional use as an oral hypoglycaemic agent. In this context, several studies have shown that metformin exhibits anti-cancer effects in BC patients with diabetes. Conversely, the efficacy of metformin and its use in non-diabetic BC patients is not widely studied, with conflicting effects being reported. A summary of the findings from studies involving metformin therapy in BC patients without DM is provided in [Table 1](#table1-2050312119865114){ref-type="table"}. ###### A summary of the salient findings from studies investigating metformin administration in non-diabetic females with breast cancer.  Title Study Study population Study findings Comments ------------------------------------------------------------------------------------------------------------------------------------------------------------------ ---------------------------------------------- ---------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Metformin may protect non-diabetic breast cancer women from metastasis El-Haggar et al.^[@bibr82-2050312119865114]^ Non-diabetics newly diagnosed with BC A total of 102 women newly diagnosed with BC were divided into the control group and metformin group. All women were treated with adjuvant therapy and the metformin-treated group received 850 mg of metformin twice daily. The metformin group demonstrated lower IGF-1, IGF1: IGFBP-3 ratio, insulin, FBG and HOMA-IR. Metformin had potential anti-tumour and anti-metastatic effects that require further exploration Follow-up for 12 months only. and study did not indicate whether metformin use had affected different stages of the BC Metformin anti-proliferative effects on a cohort of non-diabetic breast cancer patients Sadighi et al.^[@bibr83-2050312119865114]^ Non-diabetics with BC on metformin versus non-diabetics with BC not on metformin A prospective randomised controlled study about metformin efficacy in the window time between biopsy and definite surgery were carried out. Primary endpoint was changes in Ki67 expression. The intervention group was prescribed 1500 mg/day metformin from biopsy to night before surgery. A significant reduction in Ki67 was noted. Study concluded that metformin prescription for short period exhibits an inhibitory effect on BC cell growth The population size for the study was small (50 patients) and the time frame between metformin use and the obtained results was relatively short. Further long-term trials are needed Insulin-lowering effects of metformin in women with early breast cancer Goodwin et al.^[@bibr84-2050312119865114]^ Early BC patients 22 non-diabetic females with BC treated with metformin (500 mg tid × 6 months) showed lower insulin levels, and it improves insulin resistance in non-diabetic females with BC. More long-term trials are recommended Evidence for biological effects of metformin in operable breast cancer: a pre-operative, window-of-opportunity, randomised trial Hadad et al.^[@bibr85-2050312119865114]^ Operable invasive BC Trial examined effects of metformin on Ki67 and gene expression in primary BC in non-diabetic females. 47 patients were randomised to metformin 500 mg od for a week and increased to 1 g bd for a further week until surgery. Gene set analysis revealed p53, BRCA1 and cell cycle pathways showed reduced expression following metformin administration Population size was small and not age-matched Evidence for biological effects of metformin in operable breast cancer: biomarker analysis in a pre-operative window of opportunity randomised trial Hadad et al.^[@bibr86-2050312119865114]^ Operable invasive BC in non-diabetics Non-diabetic females with operable invasive BC randomised received pre-operative metformin (500 mg daily for 1 week and then 1 g bd for a further week) and control receiving no metformin. Level of Ki67 and cleaved caspase-3 fell significantly in metformin-treated cohorts but not in control Metformin acts on BC cells via pAMPK upregulation and downregulation of pAkt and insulin suppression through cytostatic mechanisms Dual effect of metformin on breast cancer proliferation in a randomised pre**-**surgical trial Bonanni et al.^[@bibr87-2050312119865114]^ Operable BC A total of 200 non-diabetic females with operable BC administered with 850 mg bd metformin (n = 100) and placebo (n = 100). Metformin prior to surgery did not affect Ki67 levels but did affect the insulin resistance in tumour cells The effect of metformin on apoptosis in a breast cancer presurgical trial Cazzangia et al.^[@bibr88-2050312119865114]^ Operable BC A total of 87 females (45 on metformin, 42 on placebo) assessed for 4 weeks before surgery. No modulation of apoptosis by metformin was found but it did affect the insulin resistance status and the Ki67 levels Differential effects of metformin on breast cancer proliferation according to markers of insulin resistance and tumor subtype in a randomised pre-surgical trial DeCensi et al.^[@bibr89-2050312119865114]^ Operable BC A total of 200 non-diabetic females randomly allocated to metformin (850 mg bd) or placebo for 4 weeks prior to BC surgery. Compared to placebo, metformin reduced Ki67 in women with HOMA-IR \> 2.8 and HER2 positive tumours. No trend was noted in women with HOMA-IR \< 2.8 Study showed that at conventional anti-diabetic dosage, the effect of metformin on Ki67 depends on the tumour characteristics and insulin resistance Presurgical trial of Metformin in overweight and obese patients with newly diagnosed breast cancer Kalinsky et al.^[@bibr90-2050312119865114]^ Invasive BC or DCIS A total of 1500 mg of metformin was administered to 35 non-diabetics with stages 0--III BC and BMI of \>25 kg/m^2^. No reduction in Ki67 was noted after metformin administration compared to controls, but there was significant reduction in BMI, cholesterol and leptin levels Trial registration NCT00930579 Effect of metformin vs placebo on and metabolic factors in NCIC CTG MA.32 Goodwin et al.^[@bibr91-2050312119865114]^ Early BC The NCIC Clinical trial group MA.32 investigated effects of metformin vs placebo on invasive disease-free survival and other outcomes in early BC. 492 females were randomly assigned treatment for 6 months. Study concluded that metformin significantly improved weight, insulin, glucose, leptin and hsCRP at 6 months Metformin in early breast cancer: a prospective window of opportunity neoadjuvant study Niraula et al.^[@bibr92-2050312119865114]^ Operable BC A total of 39 patients with newly diagnosed, untreated, non-diabetic BC patients were treated with 500 mg tid metformin after diagnostic core biopsy until definitive surgery. Study concluded that short-term use of metformin resulted in beneficial clinical and cellular changes (reduced BMI, weight, HOMA, reduced Ki67 staining in invasive tumour) consistent with beneficial anti-cancer effect Changes in insulin receptor signalling underlie neoadjuvant metformin administration in breast cancer: a prospective window of opportunity neoadjuvant study Dowling et al.^[@bibr93-2050312119865114]^ Operable BC in non-diabetics Neoadjuvant, single-arm trial conducted to determine the clinical and biological effects of metformin on patients with BC. Females with untreated BC were administered 500 mg of metformin tid for over 2 weeks from after diagnostic biopsy until surgery. Results showed a reduction in PKB/Akt and ERK 1/2 phosphorylation with decreased insulin and IR levels Clinical trial NCT00897884 Metformin intervention in obese non-diabetic patients with Breast Cancer: phase II randomised, double-blind, placebo-controlled trial Ko et al.^[@bibr94-2050312119865114]^ Obese females with BC The randomised, double-blinded, placebo-controlled trial performed to evaluate the efficacy of metformin for controlling physical and metabolic profile in non-diabetic BC patients. 105 females with BC at 6 months post-mastectomy with BMI of \>25 kg/m^2^ and/or pre-diabetes were assigned into three groups (placebo, metformin 500 mg and metformin 1000 mg) stratified by tamoxifen use. Study found that metformin 1000 mg had favourable effects on controlling glucose and HbA~1C~ levels in obese non-diabetic BC patients AMPK: AMP-activated protein kinase; BC: breast cancer; IGF: insulin-like growth factors; DCIS: ductal carcinoma in situ; BMI: body mass index. Mechanism of metformin action in BC {#section12-2050312119865114} ----------------------------------- Insight into the anti-tumour role of metformin in BC has been provided by Dowling et al.^[@bibr93-2050312119865114]^ in a clinical trial (NCT00897884). Non-diabetic females with untreated BC were trialled on neoadjuvant metformin from biopsy till surgery for BC. Immunohistochemical analysis of tumour specimens showed a significant reduction in expression of insulin receptors, phosphorylation of protein kinase B (PKB)/Akt, AMPK, extracellular signal-regulated kinase1/2 and acetyl coenzyme A carboxylase. These insulin-dependent effects are consistent with the beneficial anti-cancer effects of metformin. Metformin indirectly activates AMPK, leading to the inhibition of protein synthesis and gluconeogenesis. Thus, it may act to limit the availability of nutritional substrates that are mandatory for cancer cell proliferation.^[@bibr95-2050312119865114]^ An overview of these effects is provided in [Figure 1](#fig1-2050312119865114){ref-type="fig"}. Furthermore, AMPK also inhibits mTOR, which is a downstream activator of growth factors in malignant cells and is associated with resistance to anti-cancer drugs.^[@bibr95-2050312119865114]^ The role of metformin is not limited to AMPK pathways. It induces cell cycle arrest, thereby inducing sub-G~1~ populations and activating apoptotic pathways through downregulation of p53 and differentiated embryo chondrocyte 1 (DEC1) proteins.^[@bibr96-2050312119865114]^ Metformin administration also leads to an increase in intracellular ROS by disrupting the mitochondrial electron transport chain and collapsing the mitochondrial membrane potential. Queiroz et al.^[@bibr97-2050312119865114]^ showed that metformin has time- and concentration-dependent anti-proliferative properties on MCF-7 cells. Metformin exhibits pro-apoptotic effects and promotes cell cycle arrest via increased oxidative stress, as well as AMPK and FOXO3a activation. {#fig1-2050312119865114} The proliferation and migration of BC cells is suppressed by metformin via the dysregulation of the matrix metalloproteinases MMP-2 and MMP-9, in addition to downregulation of oncogenic microRNAs miR-21 and miR-155.^[@bibr98-2050312119865114]^ Giles et al.^[@bibr99-2050312119865114]^ demonstrated that metformin can decrease the size of mammary tumours and inhibit tumour formation in ovariectomised rats with 1-methyl-1-nitrosourea (MNU)-induced mammary tumours. Furthermore, metformin promotes a decrease in the number of aromatase-positive, CD68-positive macrophages within the tumour microenvironment, as well as decreased lipid accumulation in the livers of treated rats. This study showed that metformin targets both whole-body metabolism and the tumour microenvironment and that the perimenopausal period may represent a window of opportunity where metformin may be highly effective in women at risk for or with established BC. Other investigators have produced similar findings, demonstrating decreased tumour volumes and reduced proliferation in in vivo animal models of BC.^[@bibr100-2050312119865114],[@bibr101-2050312119865114]^ Recently, Bojkova et al. showed that administration of metformin in a rat model with MNU-induced mammary tumours resulted in an increased proportion of low-grade tumours.^[@bibr102-2050312119865114]^ A significant positive correlation between histological grade and Ki67 expression was noted. However, no differences in tumour incidence and frequency were observed. The improved tumour histopathological profile was accompanied by a reduction in serum IGF-1 levels. Metformin also exhibits cytostatic effects analogous to antifolate chemotherapeutic agents. In vitro metabolomic studies have shown that metformin has mitochondrial-independent AMPK-activating effects that cause defects in de novo purine/pyrimidine biosynthesis and homocysteine accumulation.^[@bibr103-2050312119865114]^ Metformin also exerts anti-inflammatory effects in cell models by inhibiting the NFκB pathways necessary for transformation and cancer stem cell formation. It inhibits nuclear translocation of NFκB and phosphorylation of STAT3 in cancer stem cells compared with non-stem cancer cells in the same population, thus suppressing the early stages of the inflammatory pathway that is associated with cancer.^[@bibr104-2050312119865114][@bibr105-2050312119865114]--[@bibr106-2050312119865114]^ In light of these findings from cell and animal models, it is natural to question whether metformin is a suitable adjuvant and if it should be implemented in current clinical practice guidelines for the treatment of BC. Clinical data extracted from drug trials have shown that metformin does show synergistic apoptotic effects when used with chemotherapeutic agents in BC.^[@bibr50-2050312119865114],[@bibr107-2050312119865114]^ Furthermore, when metformin is used as a single-agent, it may trigger cell cycle arrest in both oestrogen receptor positive (ER^+^) and ER-negative (ER^--^) BCs cells.^[@bibr108-2050312119865114],[@bibr109-2050312119865114]^ Metformin also elicits toxic effects on cancer stem cells, but not on normal stem cells. This property of metformin is valuable since cancer stem cells play a critical role in cancer recurrence.^[@bibr104-2050312119865114],[@bibr110-2050312119865114]^ A number of systematic reviews and meta-analyses highlighting metformin's role in BC and their limitations are summarised in [Table 2](#table2-2050312119865114){ref-type="table"}. ###### Summary of systematic reviews and meta-analysis available on metformin's role in breast cancer.  ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Title Authors and date of publication Summary of the study Comments ------------------------------------------------------------------------------------------------------------------------------------------------------------- ----------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- The effect of metformin on biomarkers and survival for breast cancer -- a systematic review and meta-analysis of randomised clinical trials Zhang et al.^[@bibr111-2050312119865114]^ Ten studies comprising 1520 BC patients. Metformin reduces levels of insulin, HOMA-IR, sex hormones, SHBG, Ki67, caspase-3, p-Akt, obesity, hs-CRP, blood glucose and lipid profile. Overall survival was non-significantly better in metformin arm than control arm. The authors concluded that further large-scale trials are needed The authors suggested that observational studies used for the systematic review and meta-analysis had residual confounding, selection biases and warranted data from more randomised clinical trials. They concluded that current data regarding metformin in BC is lacking and inconsistent Association of Metformin with Breast Cancer Incidence and Mortality in Patients with Type II diabetes: A GRADE-Assessed Systematic Review and Meta-analysis Tang et al.^[@bibr112-2050312119865114]^ Twelve observational studies were included for BC incidence and 11 studies for all-cause mortality. No significant association was found between metformin exposure and incidence of BC. A 45% risk reduction was observed for all-cause mortality. The use of metformin may improve overall survival in patient with T2D and BC Metformin as an adjuvant therapy for cancer: A systematic review and meta-analysis Coyle et al.^[@bibr113-2050312119865114]^ A meta-analysis of 27 observational studies with 24,178 subjects. No significant beneficial outcome was observed with regard to BC in metformin users compared with non-users. Significant benefit was, however, observed for early-stage colorectal and prostate cancer This meta-analysis was not specifically exploring BC incidences, instead it was evaluating the effect of metformin on recurrence-free survival, overall survival and cancer-specific survival. Data were not available to conduct analyses on the impact of metformin dose and duration Diabetes mellitus is associated with breast cancer: a systematic review, meta-analysis and in silico reproduction Zhou et al.^[@bibr114-2050312119865114]^ Twenty studies were used for this meta-analysis to explore the association between DM and BC survival outcome. Pre-existing diabetes was associated with 37% increase in all-cause mortality risk in females with BC Cohorts used by the selected studies were not age-matched, nor was the dose of metformin used in each study specified Adjuvant chemotherapy for early female breast cancer: a systematic review of the evidence for the 2014 Cancer Care Ontario systemic therapy guideline Gandhi et al.^[@bibr115-2050312119865114]^ The Program in Evidence-Based Care (PEBC) of Cancer Care Ontario complied a guideline on the systematic treatment of early BC. The Early Breast Cancer Trialists' Collaborative Group meta-analyses included RCTs. Chemotherapy was divided into three classes: anti-metabolite-based regimens, anthracyclines and taxane-based regimens. In terms of metformin use in early BC, the review specified that this drug is still in the early investigation phase Metformin and cancer risk and mortality: a systematic review and meta-analysis taking into account biases and confounders Gandini et al.^[@bibr116-2050312119865114]^ A meta-analysis was carried out using 71 articles related to metformin and cancer incidence or mortality. Results showed that metformin may reduce cancer incidence and mortality in diabetics, but the reduction is modest and does not affect all populations equally The limitations of the study included heterogeneity of study design and treatment comparators. Approximately 2/3 of the studies were prone to source of bias as they were retrospective.\ The comparator group included treatment with insulin and insulin secretagogues. These drugs increase levels of insulin and have been associated with increased cancer risk. Allocation bias was also reported, with metformin users being at different stage of diabetes than comparators The prognostic value of metformin for cancer patients with concurrent diabetes: a systematic review and meta-analysis Zhang and Li^[@bibr117-2050312119865114]^ Twenty-eight studies were used for this meta-analysis. The authors showed that metformin lowers the risk of all-cause mortality in cancer patients with concurrent diabetes, particularly in BC. The findings from the study supported the notion that metformin improves the survival for cancer patients with concurrent diabetes, particularly breast, colorectal, ovarian and endometrial cancers, but warrants further investigation in relation to the efficacy Limitation of the meta-analysis include the high heterogeneity across the studies. The information on cancer treatment was not described clearly by each of the studies that were used, thus creating a bias effect on the survival rate. Moreover, the dose of metformin and cancer stage were not reported Metformin therapy and risk of cancer in patients with type 2 diabetes: systematic review Franciosi et al.^[@bibr118-2050312119865114]^ Twelve randomised controlled trials (21,595 patients) and 41 observational studies (1,029, 389 patients) were used for this analysis. Metformin use was associated with significantly reduced risk of some cancers (liver, colorectal, pancreas, stomach and oesophagus) and cancer-related mortality. No effect on the risk of BC was reported The authors concluded that the results derived from the observational studies were prone to bias and confounding inherent due to their design. In addition, some studies were retrospective, and suffered from interviewer and recall bias. The data on dose, duration of therapy and variation over time for treatment, as well as full information on risk factors and potential confounders were incomplete Metformin and breast cancer risk: a meta-analysis and critical literature review Col et al.^[@bibr119-2050312119865114]^ Seven independent observational studies were included in this systematic review. The authors showed that metformin has a protective effect on BC risk among postmenopausal females with diabetes Results obtained were limited by the observational nature of the reports and different comparison groups that were not age-, race- or stage-matched Metformin and cancer risk in diabetic patients: a systematic review and meta-analysis Decensi et al.^[@bibr120-2050312119865114]^ Eleven studies reporting 4042 cancer events and 529 cancer deaths were selected for this meta-analysis. A 31% reduction in overall relative risk of pancreatic and hepatocellular cancer in subjects taking metformin compared with other anti-diabetic drugs was reported. No significant findings for breast, colon and prostate cancer were reported ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ BC: breast cancer; SHBG: sex hormone binding globulin; hs-CRP: high-sensitivity C-reactive protein; DM: diabetes mellitus; RCT: randomised controlled trial. Triple-negative BC (TNBC) carries the poorest prognosis of all BC subtypes. In vitro studies have shown that metformin administration enhances the sensitivity of TNBC cell lines to TRAIL receptor agonists.^[@bibr121-2050312119865114]^ TRAIL agonists (TNF-related apoptosis-inducing ligand (TRAIL) are tumour-specific inducers of apoptosis that have strong anti-tumour effects in preclinical models.^[@bibr122-2050312119865114]^ Metformin reduces the levels of XIAP, a negative regulator of TRAIL-induced apoptosis, and provides evidence supporting the combined administration of these drugs. Other in vitro studies have demonstrated that metformin reduces the percentage of TNBC stem cells through mechanisms that downregulate Krüppel-like factor 5 (KLF5) and target its degradation.^[@bibr123-2050312119865114]^ The downregulation of KLF5 is mediated by glycogen synthase kinase-3β (GSK3β) and inhibition of protein kinase A activity in TNBC cells. KLF5 is a crucial stem cell transcription factor in basal-type TNBC cells, and it promotes TNBC cell proliferation, survival, migration, invasiveness and stemness.^[@bibr124-2050312119865114][@bibr125-2050312119865114]--[@bibr126-2050312119865114]^ The reduced TNBC stem cell viability observed in vitro has significant consequences which need to be evaluated further in in vivo animal models. Metformin also has been shown to downregulate fatty acid synthase (FASN) levels via miR-193b in TNBC cells. FASN is an essential component of de novo fatty acid synthesis and is thus necessary for the survival of TNBC cells.^[@bibr127-2050312119865114]^ Despite the beneficial anti-tumour potential of metformin in TNBC, other studies have suggested that this effect is reduced by higher glucose concentrations.^[@bibr128-2050312119865114][@bibr129-2050312119865114]--[@bibr130-2050312119865114]^ Recently, Varghese et al. show that TNBC cell lines exposed to glucose levels in the diabetic range significantly abolished the effect of metformin on cell proliferation, cell death and cell cycle arrest. This study also showed that metformin was most effective and inhibited the mTOR pathway under glucose starvation conditions; suggesting that it should be combined with inhibitors of the glycolytic pathway for more beneficial treatment of TNBC in diabetic patients.^[@bibr129-2050312119865114]^ In view of the mechanistic evidence linking the anti-proliferative effect of metformin to glucose concentration in TNBC, it is natural to advocate stringent glucose level monitoring in BC patients with diabetes, particularly as the hyperglycaemic state may further fuel malignant cell proliferation. The anti-cancer effects of metformin are not limited to triple-negative and ER^+^ BC subtypes. Metformin is also effective against HER2^+^ BC since it confers anti-proliferative effects in females with HER2^+^ BC co-expressed with ER^+^ with ductal carcinoma in situ (DCIS).^[@bibr131-2050312119865114]^ Nonetheless, the molecular mechanisms behind these findings are inadequately explained. Clearly, the use of metformin in the management of BC requires further evaluation. Preliminary population-based studies have shown that metformin does not affect BC staging in older women with long-standing diabetes.^[@bibr132-2050312119865114]^ These findings contrast with both the short-term window of opportunity studies and with functional research highlighted earlier that show an effect of metformin on tumour growth characteristics. Non-specific effects and limitations of metformin in BC {#section13-2050312119865114} ------------------------------------------------------- Like many other chemotherapy agents currently in use, the development of multidrug resistance by cancer cells proves to be a considerable challenge for clinicians. Interestingly, some studies have suggested that metformin may prevent multiple drug resistance (MDR) and may even re-sensitise cancer cells to standard chemotherapy agents to which they were once sensitive. In vitro and in vivo animal studies show that metformin reduces the expression of several proteins that cause MDR.^[@bibr133-2050312119865114]^ Metformin also has MDR reversing properties in BC cell lines and re-sensitises cells to 5-fluorouracil (5-FU), adriamycin and paclitaxel through the activation of AMPL and mTOR pathways.^[@bibr134-2050312119865114],[@bibr135-2050312119865114]^ In addition, it has been shown to modulate the metabolic and miRNA profile of chemoresistant cells, rendering them similar to chemosensitive BC cell populations.^[@bibr136-2050312119865114]^ Other investigators have demonstrated that co-treatment of BC cells with metformin and flavone inhibits cell viability and increases apoptosis of cancer cells more effectively compared with metformin or flavone alone.^[@bibr137-2050312119865114]^ This potentiation of apoptosis is achieved by the modulation of MDMX/p53 proteins through PI3K/AKT pathways. Conversely, chronic exposure to metformin has been shown to lead to the development of resistance in BC cell lines. Acquired resistance to metformin is accompanied by transcriptomic changes that generate a metastatic stem-cell like phenotype.^[@bibr138-2050312119865114]^ In addition, it has been shown that long-term treatment with metformin can lead to the development of cross-resistance to both metformin and tamoxifen in MCF-7 cells.^[@bibr139-2050312119865114]^ Scherbakov et al.^[@bibr117-2050312119865114]^ show that the acquired resistance to both drugs is based on the constitutive activation of Akt/Snail1/E-cadherin signalling pathways. Why metformin confers anti-tumour effects in some, but not all BC cases is not clear. The AMPK-activating ability of metformin is central to its metabolic function in cells. Buac et al.^[@bibr104-2050312119865114]^ show that breast cancer-associated gene 2 (BCA2) is an endogenous inhibitor of AMPK activation in BC cells. BCA2 encodes a RING-finger-containing ubiquitin E3 ligase that is expressed in about 50% of breast tumours. This gene has been associated with both in vitro BC cell proliferation and clinical outcome.^[@bibr140-2050312119865114]^ Inhibition of BCA2 enhances the growth inhibitory effect of metformin in cell models, thus suggesting that metformin co-administration with a BCA2 inhibitor can be a more powerful strategy than metformin therapy in isolation.^[@bibr141-2050312119865114]^ The dose of metformin required to achieve a therapeutic effect is similarly controversial. Several doses have been used in studies with varying clinical effects. In fact, Schexnayder et al.^[@bibr142-2050312119865114]^ showed that metformin at pharmacologically achievable concentrations does not significantly improve the viability of BC cells. Instead, it inhibits inflammatory signalling and metastatic progression of the disease through reduced ICAM1, COX2, PGE2 and ROS levels. Cell cycle arrest and decreased cell viability were only reported at higher concentrations of metformin. The mechanistic findings from preclinical in vitro research are not directly translatable to clinical practice. A recent meta-analysis of observational studies on the effect of metformin on the incidence and all-cause mortality of BC in patients with type 2 diabetes failed to identify a significant association between metformin exposure and incidence of BC, while a 45% risk reduction for all-cause mortality was observed.^[@bibr112-2050312119865114]^ The uncertainty regarding the optimal dosage, duration of therapy and whether additional drugs should be co-administered with metformin to achieve synergistic effect further limits its clinical use. An ongoing phase II randomised clinical trial (NCT01589367) aims to investigate the effect of the aromatase inhibitor letrozole with metformin in postmenopausal patients with ER + BC.^[@bibr143-2050312119865114]^ Further such studies are required in order to formulate guidelines to advise clinicians on the possible therapeutic implementation of metformin in BC. This review on metformin therapy in BC has several limitations. Primarily, it does not aim to provide a systematic review of all mechanistic and epidemiologic evidence on the subject. Several authors have compiled evidence from individual studies in an attempt to resolve discrepancies and inconsistencies between different investigations, and selected key publications are summarised in [Tables 1](#table1-2050312119865114){ref-type="table"} and [2](#table2-2050312119865114){ref-type="table"}. The extent of heterogeneity and discordant findings among individual studies is significant and serves to highlight the complexity of the subject. Second, the link between metformin exposure and BC is unlikely to be a simple cause--effect relation. BC, glucose metabolism and the pharmacodynamics of metformin represent cellular events that are intrinsically heterogeneous and multidimensional and that are not fully elucidated. The interplay between each element of this complex interaction depends on various genetic, epigenetic and lifestyle factors that cannot be fully quantified and might not be faithfully reproduced in invitro preclinical studies. In the era of precision medicine and single-cell tumour biology, it is essential for researchers to acknowledge disease heterogeneity and functional diversity within solid tumours as this can significantly impact on clinical outcomes. Conclusion and future directions {#section14-2050312119865114} ================================ BC is an etiologically complex devastating disease driven by a combination of genetic, reproductive, hormonal and environmental factors. The epidemiologic link between BC and disordered glucose metabolism is mechanistically interesting, given that glucose is an essential cellular metabolic substrate and that insulin signalling has mitogenic effects. The common underlying mechanism uniting T2DM and BC involves hyperinsulinemia, which activates several molecular pathways driving cell proliferation.^[@bibr144-2050312119865114]^ BC has been traditionally treated with a combination of chemotherapy, surgery and targeted hormonal therapy; however, growing interest lies in the use of metformin. Metformin activates AMP-activated protein kinase and inhibits mTOR pathways, thereby decreasing insulin levels in the circulation. In addition, it also inhibits the proliferation and invasion of cancer cells, which could limit metastatic spread. Studies have also demonstrated that metformin may enable cancer cells to overcome the development of resistance to chemotherapy, hormone therapy and trastuzumab treatment. The use of metformin in the management of both T2DM and BC may seem practical considering that metformin is one of the most commonly prescribed oral anti-diabetic agents. It accounts for 40% of all anti-diabetic drugs dispensed in England over the past few decades, but only 7% of the costs.^[@bibr145-2050312119865114]^ The recent ALTTO trial on metformin use in HER2^+^ BC showed that metformin may improve the worse prognosis that is associated with diabetes and insulin treatment in patients with HER2^+^ and hormone receptor positive BC.^[@bibr146-2050312119865114]^ Furthermore, promising in vivo studies suggest that metformin can have beneficial synergistic effects with natural anti-tumour compounds such as curcumin.^[@bibr147-2050312119865114]^ Meta-analysis of large cohorts have demonstrated that metformin use is associated with improved survival and decreased all-cause mortality in diabetic patients with BC.^[@bibr148-2050312119865114],[@bibr149-2050312119865114]^ Conversely, conflicting clinical findings with regard to the efficacy and anti-tumour role of metformin have been reported in the literature, thus strengthening the need for further research.^[@bibr132-2050312119865114],[@bibr150-2050312119865114]^ The potential for therapeutic benefits of metformin in diabetics with BC is rapidly becoming an area of interest in both clinical oncology and endocrinology. However, more long-term double blinded-randomised trials are needed to explore the precise role which metformin may play and its possible use as an adjuvant in clinical practice. Most current studies that examine metformin's use in BC have reported a mixed picture on its efficacy, which could be due to the different doses of metformin as well as varying periods of follow-up used in these studies. It is clear that metformin holds considerable promise with regard to a potential anti-tumour role. **Author contribution:** All the authors contributed equally to this work. **Declaration of conflicting interests:** The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. **Funding:** The author(s) received no financial support for the research, authorship, and/or publication of this article. **ORCID iDs:** Mohsin HK Roshan  <https://orcid.org/0000-0003-4555-8596> Nikolai P Pace  <https://orcid.org/0000-0002-7332-874X>

Introduction ============ Fungal infections in humans are mainly classified according to infection site into two categories, superficial and systemic, and the pathogenic fungi causing superficial infections preferably inhabit keratinized tissues such as stratum corneum of the skin, nail, and hair. The tinea corporis and tinea pedis caused by *Trichophyton* spp., cutaneous candidiasis caused by *Candida* spp., and tinea verisicolor caused by *Malasezzia furfur* are successfully treated by recently developed topical antifungal drugs \[[@B1]\]. In contrast to these superficial fungal infections, tinea unguium or onychomycosis caused by *Trichophyton* spp. is usually resistant to topical antifungal chemotherapy and is therefore treated with oral drugs such as itraconazole, an azole antifungal drug, and terbinafine, an allylamine antifungal drug \[[@B2]--[@B6]\]. Itraconazole capsules, for instance, are given in a pulse-treatment regimen (400 mg/day for 1 week at monthly intervals) that is repeated twice or three times for finger nail onychomycosis and is repeated three or four times for toenail onychomycosis \[[@B4]--[@B6]\]. Because some strains of *Trichophyton* spp., however, are resistant to azole compounds \[[@B7]\] and itraconazole has toxic side effects \[[@B8]\], we want to provide a new tool for the treatment of onychomycosis. Antifungal effects of reactive oxygen species have long been documented \[[@B9]--[@B13]\], and hydroxyl radicals are known to damage proteins, lipids, and nucleic acids \[[@B14]--[@B17]\]. Since hydroxyl radicals are generated when water is exposed to ultrasound \[[@B18], [@B19]\], we evaluated the fungicidal effects seen when suspensions of the dermatophytes that are the major cause of onychomycosis are exposed to ultrasound. Materials and Methods ===================== Reagents -------- 5,5-Dimethyl-1-pyrroline-*N*-oxide (DMPO) of analytical grade was purchased from Labotec Co., Ltd. (Tokyo, Japan), and all the other reagents used were also of analytical grade. Fungus preparation ------------------ Two clinical isolates of *Trichophyton mentagrophytes* and three clinical isolates of *Trichophyton rubrum* were obtained from patients with tinea pedis in Showa University Fujigaoka Hospital (Yokohama, Japan), and for each strain a conidial suspension in sterile physiological saline containing 0.1% (w/v) Tween 80 was prepared from cultures grown on Sabouraud dextrose agar slants at 27°C for 1--4 weeks. Following filtration through sterilized gauze to remove hyphal fragments and pieces of agar, the concentration of each suspension was adjusted to 10^6^ conidia/ml. ESR analyses of hydroxyl radicals generated by ultrasound in water ------------------------------------------------------------------ Our device using 1.6-MHz ultrasound to generate hydroxyl radicals is shown schematically in Fig. [1](#F1){ref-type="fig"}. A glass tube (15 mm in diameter and 85 mm long) containing 1 ml of 89 mM DMPO dissolved in pure water was set into the device and exposed to sonication for 1 min. The reaction mixture obtained after the exposure was immediately transferred to an ESR spectrometry cell, and its ESR spectrum was recorded. The spin concentration of the DMPO-OH (a DMPO and hydroxyl radical adduct) was determined from the double integrals of the DMPO-OH peak by comparing them with those of the peak for the stable nitroxide radical, 4-hydroxy-2,2,6,6-tetramethylpiperidine-*N*-oxyl. The measurement conditions for ESR were as follows: field sweep, 330.50 to 340.50 mT; field modulation frequency, 100 kHz; field modulation width, 0.08 mT; amplitude, 10--100; sweep time, 1 min; time constant, 0.03 s; microwave frequency, 9.420 GHz; microwave power, 4 mW. Measurement of fungicidal activity by ultrasound radiation ---------------------------------------------------------- One ml of each conidial suspension was exposed to ultrasound radiation and then diluted with phosphate-buffered saline. A small aliquot was inoculated on a Sabouraud dextrose agar plate and cultured at 27°C for 5 to 14 days. Fungicidal activity was assessed by counting the colonies grown on the agar plate in order to determine the number of colony-forming units (CFUs) that had survived the ultrasound treatment. Scanning electron microscopy (SEM) ---------------------------------- A sample of each conidial suspension was smeared on a silicon-coated glass plate and fixed with 2.5% (w/v) glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4) at 4°C overnight. After washing with 0.1 M cacodylate buffer (pH 7.4), the sample specimens were post-fixed with 2% (w/v) OsO~4~ in 0.2 M cacodylate buffer (pH 7.4) at 4°C for 2 h and then routinely processed for scanning electron microscopic observation. In brief, the post-fixed specimens were dehydrated through an ethanol series, substituted with isoamyl acetate, and dried with liquid CO~2~ by using a ID-2 critical point dryer (Eiko Ltd., Tokyo, Japan). Each dried specimen was coated with gold by using an SC7640 sputter coater (Quorum Technologies Ltd., Hailsham, UK) and observed by using a Topcon DS701 scanning electron microscope at 10 kV. Transmission electron microscopy (TEM) -------------------------------------- After conidial suspensions of each fungus were washed in physiological saline several times and centrifuged, the resultant pellets were fixed with 2.5% (w/v) glutaraldehyde in 0.2 M cacodylate buffer (pH 7.4) at 4°C overnight. After washing with 0.1 M cacodylate buffer (pH 7.4), the specimens were post-fixed with 2% (w/v) OsO~4~ in 0.2 M cacodylate buffer (pH 7.4) at 4°C for 2 h, and were routinely processed for transmission electron microscopic observation. In brief, the post-fixed specimens were dehydrated through an ethanol series, substituted with propylene oxide, and embedded in Epon812 so that ultra-thin sections could be prepared. The sections were stained with uranyl acetate and lead citrate and were observed by using a Hitachi H7600 transmission electron microscope (Hitachi High-Technologies Corp., Tokyo, Japan) at 75 kV. Results and Discussion ====================== Fig. [2](#F2){ref-type="fig"} shows the representative ESR spectra of DMPO-OH generated by ultrasound in water. The hyperfine coupling constants of the ESR signal were a~N~ = a~H~ = 1.49 mT. The effects of water temperature and exposure time on the amount of ·OH generated when water was exposed to ultrasound are summarized in Figs. [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"}, where one sees that the DMPO-OH concentration increased with time and with increasing temperature. The effects of water temperature and exposure time on the fungicidal activity of water exposed to ultrasound are shown in Fig. [5](#F5){ref-type="fig"}, where one sees that the fungicidal effect of ultrasound radiation was enhanced with time and increasing water temperature and that four of the five strains tested did not survive 10 min in 50°C water exposed to ultrasound. One also sees from the bottom-right bar in Fig. [5](#F5){ref-type="fig"} that 50°C without ultrasound had almost no fungicidal effect. Since hydroxyl radical generation increased with ultrasound duration and water temperature, the fungicidal action shown in Fig. [5](#F5){ref-type="fig"} is likely to be due to attributable to hydroxyl radicals generated by ultrasound in water. A SEM image of *T. mentagrophytes* conidia exposed for 5 min to ultrasound in water at 50°C (Fig. [6](#F6){ref-type="fig"}) shows that the ultrasound caused the conidia to collapse and their surfaces to exfoliate. TEM images of *T. mentagrophytes* conidia exposed to ultrasound radiation at 50°C for different periods of time (Fig. [7](#F7){ref-type="fig"}) indicate that the exposure degraded the cytoplasmic structure in ways that are likely to lead to necrosis \[[@B20]--[@B22]\]. In the references, the term "necrosis" is used to describe irreversible structural degeneration such as crushed, bent, and flattened cells. Ultrasound in water has thus been shown to generate hydroxyl radicals that degrade cytoplasmic and surface structures and thereby lead to necrosis of fungal cells. Since the dermatophytes used in this study are major causes of onychomycosis, ultrasound in hot water is expected to be a useful treatment for onychomycosis. In a future study, to check if ultrasound itself has fungicidal action, we will further examine the effect of specific scavengers of the hydroxyl radical on the fungicidal action of ultrasound in water. {#F1} {#F2} {#F3} {#F4} {#F5} {#F6} {#F7}

Introduction ============ The recombinant fusion protein aflibercept (Eylea^®^; Regeneron, Tarrytown, NY, USA) is composed of vascular endothelial growth factor (VEGF)-binding receptors-1 and -2 fused to the Fc portion of human immunoglobulin G.[@b1-opth-13-373] This agent was first approved in the United States for the treatment of neovascular age-related macular degeneration.[@b2-opth-13-373] Notably, an increased VEGF level as measured in vitreous samples has been identified as the main cause of both diabetic macular edema (DME) and cystoid macular edema (CME) due to retinal vein occlusion (RVO). Aflibercept has been proven effective for CME due to central retinal vein occlusion (CRVO) in COPERNICUS and GALI-LEO studies[@b2-opth-13-373]--[@b4-opth-13-373] and for DME in VIVID, VISTA[@b5-opth-13-373] and DA-VINC[@b6-opth-13-373] studies. It has also been proven effective for CME due to branch retinal vein occlusion (BRVO) in the VIBRANT study.[@b7-opth-13-373] Given these promising characteristics, Japan approved the use of aflibercept for the treatment of DME in November 2014, for CME due to CRVO in November 2013 and for CME due to BRVO in June 2015. This drug has since become the standard of care for treatment of these diseases in Japan. In humans, the pharmacokinetic profile of aflibercept after intravitreal injection has not yet been determined definitively. However, a study found that bevacizumab, another recombinant humanized VEGF-binding monoclonal antibody, can be detected in the anterior chamber of the eye at 1 month after intravitreal injection.[@b8-opth-13-373] However, as VEGF and its receptors are expressed in the corneal endothelium,[@b9-opth-13-373]--[@b11-opth-13-373] which facilitated the progression of severe corneal changes in humans treated with bevacizumab.[@b12-opth-13-373] Therefore, aflibercept may also affect VEGF function in the corneal endothelium following intravitreal injections into the vitreous humor. Aflibercept was also shown to interfere with the physiology of retinal pigment epithelial cells[@b13-opth-13-373] and to negatively affect the proliferation and viability of mesenchymal stem cells derived from the ciliary body and limbus in a concentration-dependent manner.[@b14-opth-13-373] Aflibercept has a predicted longer intravitreal activity than ranibizumab,[@b15-opth-13-373] which suggests a stronger potential for both beneficial and adverse effects. The above findings suggest that it is reasonable to consider whether this drug is potentially cytotoxic to human corneal endothelial cells. Lass et al[@b16-opth-13-373] reported that intravitreal injections of aflibercept did not affect the corneal endothelium in patients with unilateral neovascular age-related macular degeneration. The corneal endothelial cells are known to be susceptible to surgical insults in diabetes. The number of corneal endothelial cells is prone to reduce in diabetic eyes compared to that in nondiabetic eyes after cataract surgery.[@b17-opth-13-373],[@b18-opth-13-373] Therefore, the intravitreal injection of aflibercept may place significant stress on the corneal endothelial cells of diabetic patients. As specular microscopy with morphometric analysis is a sensitive indicator of endothelial cell function,[@b19-opth-13-373],[@b20-opth-13-373] we used this technology to prospectively analyze in vivo corneal endothelial toxicity after intravitreal injection of aflibercept for the treatment of DME and CME due to RVO. Material and methods ==================== Forty-six eyes of 44 consecutive patients with DME and CME due to RVO were recruited for this observational prospective case series study. The study protocol was approved by the Institutional Review Board of the Dokkyo Medical University Saitama Medical Center, and the study was conducted in accordance with the tenets of the Declaration of Helsinki. All participants provided written informed consent. The inclusion criteria were clinical signs and angiographic evidence of DME and CME due to RVO with a best-corrected visual acuity (VA) of 20/30 or worse; an age of ≥50 years; no history of contact lens use; no ocular disease that could alter the corneal endothelium morphology, such as Fuchs' corneal dystrophy or iridocorneal endothelial syndrome; no intraocular surgery within a 6-month period before and after aflibercept injection; no laser treatment within a 2-month period before and 6-month period after the injection; no anti-VEGF antibody injection within a 6-month period before the injection and an endothelial cell count of \>1,500/mm^2^ before the initial aflibercept injection. All patients initially received a single intravitreal injection of aflibercept (2 mg in a total volume of 0.05 mL), followed by pro re nata use according to the study protocol. All patients were followed for 6 months and underwent VA testing, slit-lamp examination, IOP measurement using Goldmann applanation tonometry and indirect ophthalmoscopy, and optical coherence tomography (PLEX Elite 9000; Carl Zeiss Meditec, Inc, Dublin, CA, USA) examinations were performed at baseline and before the administration of aflibercept. The treatment effect was monitored based on improvements in VA and optical coherence tomography measurements. Intravitreal aflibercept injection technique -------------------------------------------- Conjunctival anesthesia was topically induced by instillation of 4% lidocaine (Xylocaine^®^ solution 4%; Aspen Japan, Tokyo, Japan). The eyelids and ocular surface were disinfected with 0.027% iodine (PA⋅IODO^®^ Ophthalmic and Eye washing Solution; Nitten Pharmaceutical Co, Ltd, Nagoya, Japan). In an operating room, aflibercept was injected into the superotemporal quadrant via the pars plana and into the vitreous cavity 3--4 mm posterior to the limbus using a 32-gauge needle. The post-injection light perception was assessed. A topical 0.5% levofloxacin solution (0.5% Cravit^®^ ophthalmic solution; Santen Pharmaceutical, Osaka, Japan) was instilled four times daily for 3 days before and after each intravitreal injection. Endothelial cell analysis ------------------------- Noncontact specular microscopy of the central cornea was performed using a specular microscope (CEM-530; NIDEK, Tokyo, Japan) that could measure corneal endothelial cell density (ECD) in a small area (0.25 × 0.55 mm). Corneal endothelial images before and 1, 3 and 6 months after the initial intravitreal aflibercept injection were obtained by experienced ophthalmic technicians. The specular microscope automatically evaluated the ECD, average cell size (AVG), standard deviation of cell size (SD), coefficient of variation of cell size (CoV), maximum of cell size (MAX), minimum of cell size (MIN) and percentage of the hexagonal cells (Hex%). The specular microscopy also provided measurements of optical pachymetry and evaluated the central corneal thickness. IOP measurement --------------- The first author (TM) measured the IOP before and 1, 3 and 6 months after the initial intravitreal injection of aflibercept. The IOP was measured once at each time point. Statistical analysis -------------------- All data are expressed as mean ± SD. Differences between preinjection and postinjection measurements were evaluated using a one-way ANOVA with the Dunnet's multiple comparison posttest. Differences between the treated and untreated eyes were evaluated using an unpaired *t*-test. Differences between patients with DME and CME due to RVO were evaluated using an unpaired *t*-test. Differences between single and multiple injections were evaluated using an unpaired *t*-test. The statistical analysis was performed using StatMate version V for Macintosh (ATMS, Tokyo, Japan). A *P*-value \<0.05 was considered to indicate significance. Results ======= The study group included 27 men and 17 women with a mean age of 72.1 years (range, 55--88 years). Of the treated eyes, 27 (58.7%) and 19 (41.3%) were phakic and pseudophakic, respectively; the corresponding values were 22 (47.8%) and 24 (52.2%) for the untreated eyes. None of the pseudophakic eyes had a damaged posterior capsule. Eighteen, seven and 19 patients had DME, CME due to CRVO and CME due to BRVO, respectively. Patients with DME and CME due to RVO had mean ages of 67.5 (range, 60--82 years) and 75.3 years (range, 55--88 years), respectively. Five of 26 patients with CME due to RVO had diabetes. Ten eyes in patients with CME due to RVO received two intravitreal injections of aflibercept while 16 eyes received single intravitreal injections. Among patients with DME, six eyes received two intravitreal injections, two eyes received three injections and eleven eyes received a single injection. Overall, the per-patient average number of injections was 1.43±0.58, with an average of 1.53±0.70 and 1.39±0.50 injections in patients with DME and CME due to RVO, respectively. Among eyes that received multiple (\>1) injections, an average of 2.11±0.32 injections were administered. All patients completed the 6-month follow-up. No systemic or ocular adverse events were recorded during this period. ECD and morphology ------------------ The following normal score ranges were set for each parameter: ECD: 2,777--3,410 cells/mm^2^; AVG: 296--367 µm^2^; CoV: 0.26--0.40; Hex: 55%--67%.[@b21-opth-13-373] The measurements obtained before and at 1, 3 and 6 months after the first intravitreal aflibercept injection from patients with DME and CME due to RVO are presented in [Table 1](#t1-opth-13-373){ref-type="table"}. Of all evaluated parameters, only MAX differed significantly from before to after the intravitreal injections at any time point. The photographs of the endothelial cells and optical coherence tomography of a representative eye are presented in [Figure 1A--D](#f1-opth-13-373){ref-type="fig"}. The measurements obtained at all time points were compared between the treated and untreated eyes, and the results are presented in [Tables 2](#t2-opth-13-373){ref-type="table"} and [3](#t3-opth-13-373){ref-type="table"}. Again, of all evaluated parameters, only MAX differed significantly at 6 months after the intravitreal injections. The measurements obtained at all time points were also compared between patients with DME and those with CME due to RVO, and the results are presented in [Tables 4](#t4-opth-13-373){ref-type="table"} and [5](#t5-opth-13-373){ref-type="table"}. No significant differences were observed in any of the evaluated parameters from before to after the intravitreal injections at any time point. The measurements obtained at all time points were compared between eyes subjected to single and multiple intravitreal aflibercept injection, and the results are presented in [Tables 6](#t6-opth-13-373){ref-type="table"} and [7](#t7-opth-13-373){ref-type="table"}. Again, of all evaluated parameters, only MAX differed significantly at 3 months after the intravitreal injections. IOP --- The mean IOPs before injection and at 1, 3 and 6 months after the initial injection were 13.3±3.2, 12.5±2.7, 13.0±3.2 and 12.8±2.9 mmHg, respectively ([Table 1](#t1-opth-13-373){ref-type="table"}). The difference between time points was not statistically significant (*P*=0.55). A comparison of the mean IOPs of treated and untreated eyes measured before and 1, 3 and 6 months after the initial injection yielded the following: 13.3±3.2 vs 13.8±3.4, 12.5±2.7 vs 13.6±2.9, 13.0±3.2 vs 13.3±3.1 and 12.8±2.9 vs 13.6±3.3 mmHg, respectively ([Tables 2](#t2-opth-13-373){ref-type="table"} and [3](#t3-opth-13-373){ref-type="table"}). Only the difference at 1 month was statistically significant. A comparison of the mean IOPs of patients with DME and CME due to RVO obtained before and 1, 3 and 6 months after the initial injection yielded the following: 14.2±3.1 vs 12.8±3.2, 13.3±2.8 vs 12.0±2.4, 12.6±3.0 vs 13.3±3.1 and 13.2±2.9 vs 12.3±3.0 mmHg, respectively ([Tables 4](#t4-opth-13-373){ref-type="table"} and [5](#t5-opth-13-373){ref-type="table"}). These differences were not statistically significant at any time point. Furthermore, a comparison of the mean IOPs measured before injection and 1, 3 and 6 months after the initial injection in eyes receiving single or multiple intravitreal doses of aflibercept yielded the following: 13.4±3.3 vs 13.1±3.2, 12.3±2.6 vs 12.8±2.7, 12.4±2.8 vs 14.1±3.4 and 12.3±2.6 vs 13.3±3.4 mmHg, respectively ([Tables 6](#t6-opth-13-373){ref-type="table"} and [7](#t7-opth-13-373){ref-type="table"}). These differences were not statistically significant at any time point. Discussion ========== Experimental animal models have demonstrated that intravitreal injections yield adequate concentrations of ranibizumab,[@b22-opth-13-373] bevacizumab[@b8-opth-13-373] and aflibercept[@b23-opth-13-373] in the anterior chamber. However, the intracameral injection of ranibizumab and bevacizumab was shown to reduce corneal endothelial cell densities.[@b24-opth-13-373] Taken together, these data underscore the importance of determining the effects of intravitreal aflibercept injection on corneal endothelial cells. In the present study, these injections did not generally adversely affect the corneal endothelial cells, and no significant changes were observed in any parameter except the MAX cell size during the 6-month follow-up. Three inter-group comparisons revealed differences in IOP between treated and untreated eyes at 1 month after injection, in MAX between treated and untreated eyes at 6 months after injection and in MAX between eyes receiving single and multiple injections at 3 months after injection. No significant differences were observed between patients with DME and CME due to RVO at any time point. The corneal endothelium in diabetic patients is vulnerable compared to that of normal patients and susceptible to surgical stress.[@b17-opth-13-373],[@b18-opth-13-373] However, the current study findings suggest that an intravitreal injection of aflibercept might not represent a relevant surgical stress. As we mentioned previously, only MAX differed significantly at 3 months after the injection when eyes receiving single vs multiple intravitreal injections of aflibercept were compared. Therefore, only slight differences between the administration of single and multiple injections were observed. Limitations =========== There are several limitations of our study, such as the small sample size, short follow-up period and low average number of injections (1.43 times/6 months). In general, intravitreal aflibercept injections are required every 2 months after three initial monthly doses to treat wet age-related macular degeneration[@b25-opth-13-373] in Japan. Lass et al[@b16-opth-13-373] assessed the effect of intravitreal aflibercept injection on corneal endothelium every 8 weeks after 3 monthly doses up to 52 weeks; on specular microscopy, no apparent corneal endothelial toxicity was observed because of intravitreal aflibercept injection. Intravitreal aflibercept injections were performed every 4 or 8 weeks after five initial monthly doses for DME in VIVID and VISTA studies;[@b5-opth-13-373] the results of these large clinical trials led to the recommendation of the injections every 4 weeks after five initial monthly doses for DME in Japan. In contrast, intravitreal aflibercept injection pro re nata after a single dose is recommended for CME due to RVO in Japan. Although we initially considered pro re nata administration after five initial monthly intravitreal aflibercept injections for patients with DME, many patients rejected the high number of initial injections because of financial issues. Accordingly, we adopted the protocol of pro re nata administration after single intravitreal injection in our study. Other factors may also affect the observed outcomes of intravitreal aflibercept injections. For example, retinal photocoagulation with an indirect ophthalmoscopy contact lens reportedly decreased the number of corneal endothelial cells.[@b26-opth-13-373] Accordingly, we addressed the effects of retinal photocoagulation on the corneal endothelium as much as possible. We additionally evaluated the mean IOP over the 6-month follow-up period and observed very slight changes between treated and untreated eyes at 1 month. Chhablani et al[@b27-opth-13-373] reported that a single intravitreal aflibercept injection (2 mg) did not affect IOP after 1 month. Furthermore, several case series have reported sterile inflammation after intravitreal aflibercept injections.[@b28-opth-13-373]--[@b31-opth-13-373] Although no cases of sterile inflammation were observed in the current study, the previous reports raised concerns about the potential effects of sterile inflammation on the corneal endothelium. We further note that previous results from experimental animal models were not completely reliable because of the differences between animal and human tissues. Furthermore, although previous studies on intravitreal bevacizumab,[@b32-opth-13-373] ranibizumab[@b33-opth-13-373],[@b34-opth-13-373] and aflibercept[@b16-opth-13-373] injections did not report effects on the human corneal endothelium, the safety of use of aflibercept on human tissues is a very important issue facing clinical practice. However, our findings very slightly differed from previous reports describing the effects of anti-VEGF antibodies.[@b16-opth-13-373],[@b32-opth-13-373]--[@b34-opth-13-373] As intravitreal anti-VEGF antibody injections will be used for the treatment of many retinal diseases in future, considerable attention must be paid to not only the effectiveness but also the potential adverse effects. Conclusion ========== The results of this clinical study indicate that an intravitreal injection of aflibercept (2 mg) might induce corneal endothelial damage or very slight changes in corneal endothelial cells within 6 months of the first injection. Corneal endothelial damage can lead to partial abnormal corneal hydration and, consequently, corneal edema. Our findings warrant further studies regarding the potential side effects of intravitreal aflibercept injection on the human corneal endothelium. **Disclosure** The authors report no conflicts of interest in this work. {#f1-opth-13-373} ###### Morphometric analysis of the corneal endothelium in the treated eye before and after intravitreal injections of aflibercept in patients with DME and CME due to RVO Before injection 1 month AI 3 months AI 6 months AI *P*-value ------------------- ------------------ --------------- --------------- --------------- ----------- ECD (cells/mm^2^) 2,515.0±422.3 2,528.5±441.5 2,500.0±427.1 2,550.5±451.1 0.95 AVG (µm^2^) 410.7±81.4 412.5±96.7 415.5±98.0 409.0±103.8 0.99 SD (µm^2^) 111.2±32.4 111.8±32.9 115.2±33.8 113.1±32.8 0.94 CoV 28.4±4.5 28.7±4.4 29.4±5.6 29.3±4.4 0.70 MAX (µm^2^) 1,070.1±220.4 1,138.9±251.5 1,180.6±283.5 1,041.5±238.7 0.033 MIN (µm^2^) 146.3±28.4 146.1±27.2 145.0±16.2 146.6±19.7 0.99 Hex (%) 67.5±4.9 67.6±5.3 66.9±6.4 66.5±6.0 0.78 CCT (µm) 541.7±27.8 543.2±27.1 541.7±29.1 542.4±28.1 0.99 IOP (mmHg) 13.3±3.2 12.5±2.7 13.0±3.2 12.8±2.9 0.55 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated and untreated eyes before and 1 month after the intravitreal injection of aflibercept in patients with DME and CME due to RVO Before injection 1 month AI ------------------- ------------------ --------------- ------ --------------- --------------- ------- ECD (cells/mm^2^) 2,515.0±422.3 2,495.5±435.9 0.83 2,528.5±441.5 2,524.1±436.8 0.96 AVG (µm^2^) 410.7±81.4 415.8±92.7 0.78 412.5±96.7 413.8±102.8 0.95 SD (µm) 111.2±32.4 114.2±35.9 0.67 111.8±32.9 110.3±30.4 0.82 CoV 28.4±4.5 29.0±5.4 0.53 28.7±4.4 28.3±4.5 0.70 MAX (µm^2^) 1,070.1±220.4 1,115.5±261.7 0.37 1,138.9±251.5 1,132.2±249.4 0.90 MIN (µm^2^) 146.3±28.4 153.1±34.9 0.31 146.1±27.2 146.5±18.4 0.93 Hex (%) 67.5±4.9 69.0±5.8 0.18 67.6±5.3 66.8±5.2 0.48 CCT (µm) 541.7±27.8 546.2±33.3 0.48 543.2±27.1 546.0±28.2 0.64 IOP (mmHg) 13.3±3.2 13.8±3.4 0.52 12.5±2.7 13.6±2.9 0.049 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated and untreated eyes at 3 and 6 months after the intravitreal injection of aflibercept in patients with DME and CME due to RVO 3 months AI 6 months AI ---------------- --------------- --------------- ------ --------------- --------------- ------- ECD (cells/mm) 2,500.0±427.1 2,487.0±422.3 0.88 2,550.5±451.1 2,492.4±447.4 0.54 AVG (µm^2^) 415.5±98.0 418.7±109.2 0.88 409.0±103.8 419.8±115.2 0.64 SD (µm) 115.2±33.8 115.1±35.5 0.98 113.1±32.8 115.4±46.2 0.78 CoV 29.4±5.6 29.2±6.7 0.89 29.3±4.4 28.9±5.7 0.70 MAX (µm) 1,180.6±283.5 1,122.5±254.3 0.30 1,041.5±238.7 1,149.1±272.1 0.047 MIN (µm) 145.0±16.2 147.6±25.4 0.57 146.6±19.7 145.0±24.7 0.74 Hex (%) 66.9±6.4 69.0±5.9 0.12 66.5±6.0 68.6±5.8 0.11 CCT (µm) 541.7±29.1 544.7±30.4 0.63 542.4±28.1 544.9±28.3 0.67 IOP (mmHg) 13.0±3.2 13.3±3.1 0.65 12.8±2.9 13.6±3.3 0.22 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated eye before and 1 month after the intravitreal injection of aflibercept in patients with DME vs CME due to RVO Before injection 1 month AI ------------------- ------------------ --------------- ------ --------------- --------------- ------ ECD (cells/mm^2^) 2,503.1±419.2 2,528.2±440.0 0.85 2,542.3±379.2 2,535.8±488.8 0.96 AVG (µm^2^) 412.0±80.4 409.5±85.2 0.92 405.8±69.3 414.7±114.6 0.75 SD (µm) 113.5±33.9 110.3±32.3 0.75 113.0±27.4 110.7±37.5 0.81 CoV 29.0±5.4 28.2±3.6 0.57 29.6±5.4 28.0±3.4 0.27 MAX (µm^2^) 1,063.1±249.5 1,081.2±203.7 0.80 1,171.6±286.5 1,120.5±229.7 0.53 MIN (µm^2^) 152.2±41.7 142.8±11.8 0.35 148.4±36.4 144.0±19.2 0.64 Hex (%) 67.6±5.6 67.3±4.6 0.86 66.5±6.4 68.5±4.4 0.25 CCT (µm) 546.4±26.7 538.2±29.1 0.34 550.5±24.0 539.3±28.1 0.17 IOP (mmHg) 14.2±3.1 12.8±3.2 0.16 13.3±2.8 12.0±2.4 0.10 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated eye at 3 and 6 months after the intravitreal injection of aflibercept in patients with DME vs CME due to RVO 3 months AI 6 months AI ---------------- --------------- --------------- ------ --------------- --------------- ------- ECD (cells/mm) 2,514.2±463.8 2,487.0±422.3 0.87 2,539.3±367.8 2,542.0±510.9 0.98 AVG (µm) 417.5±116.5 418.7±109.2 0.83 403.0±68.8 416.2±125.3 0.65 SD (µm) 112.6±35.6 115.1±35.5 0.53 114.7±22.2 113.1±39.3 0.87 CoV 28.4±2.8 29.2±6.7 0.22 30.5±5.7 28.5±3.1 0.19 MAX (µm) 1,216.6±272.9 1,122.5±254.3 0.43 1,018.4±164.5 1,066.0±283.9 0.48 MIN (µm) 143.8±13.4 147.6±25.4 0.81 151.6±25.2 143.5±14.3 0.22 Hex (%) 67.4±5.3 69.0±5.9 0.58 64.5±7.0 67.7±4.8 0.091 CCT (µm) 538.3±30.4 544.7±30.4 0.26 549.7±26.1 537.0±29.4 0.14 IOP (mmHg) 12.6±3.0 13.3±3.1 0.13 13.2±2.9 12.3±3.0 0.32 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated eye before and 1 month after a single or multiple intravitreal injections of aflibercept in patients with DME and CME due to RVO Before injection 1 month AI ------------------- ------------------ --------------- ------ --------------- --------------- ------ ECD (cells/mm^2^) 2,513.9±434.9 2,498.7±426.6 0.91 2,498.6±463.9 2,575.0±412.8 0.57 AVG (µm^2^) 411.7±84.9 412.7±79.9 0.97 419.3±111.0 401.9±70.5 0.52 SD (µm) 113.4±35.8 110.4±27.5 0.75 113.8±38.5 108.8±22.5 0.58 CoV 28.7±4.7 28.4±4.1 0.83 28.5±4.8 28.8±3.6 0.81 MAX (µm^2^) 1,060.9±250.5 1,121.2±199.3 0.37 1,113.0±266.5 1,179.2±227.5 0.39 MIN (µm^2^) 141.3±11.9 156.0±41.8 0.88 142.5±15.1 151.7±39.3 0.35 Hex (%) 67.6±5.1 67.7±5.2 0.97 67.5±6.1 67.8±3.8 0.82 CCT (µm) 540.8±26.5 540.3±32.5 0.96 544.3±22.9 541.6±33.1 0.77 IOP (mmHg) 13.4±3.3 13.1±3.2 0.71 12.3±2.6 12.8±2.7 0.47 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion. ###### Morphometric analysis of the corneal endothelium in the treated eye after 3 and 6 months single or multiple intravitreal aflibercept injections in patients with DME and CME due to RVO 3 months AI 6 months AI ---------------- --------------- --------------- ------- --------------- --------------- ------ ECD (cells/mm) 2,478.6±451.5 2,533.6±396.5 0.67 2,499.8±487.1 2,589.1±387.0 0.49 AVG (µm) 422.5±113.1 404.6±69.8 0.51 421.4±120.8 395.7±68.3 0.36 SD (µm) 120.4±39.4 107.2±21.1 0.15 118.9±37.8 105.9±20.4 0.14 CoV 30.2±6.5 28.1±3.6 0.17 30.0±4.5 28.3±4.1 0.21 MAX (µm) 1,108.3±263.9 1,293.2±283.0 0.029 1,024.8±184.9 1,071.7±306.3 0.56 MIN (µm) 146.0±19.3 143.4±9.9 0.55 144.8±17.7 149.7±22.5 0.44 Hex (%) 66.5±7.1 67.7±5.1 0.51 65.9±5.9 67.6±6.4 0.38 CCT (µm) 541.6±25.9 541.8±34.2 0.98 540.0±26.5 543.8±32.6 0.68 IOP (mmHg) 12.4±2.8 14.1±3.4 0.086 12.3±2.6 13.3±3.4 0.31 **Note:** Data are presented as mean ± SD. **Abbreviations:** AI, after injection; AVG, average cell size; CCT, central corneal thickness; CME, cystoid macular edema; CoV, coefficient of variation of cell size; DME, diabetic macular edema; ECD, endothelial cell density; Hex, hexagonal cells; MAX, maximum of cell size; MIN, minimum of cell size; RVO, retinal vein occlusion.

{#sp1 .518}

Introduction ============ Prostate cancer (PCa) is a major urological problem associated with significant morbidity and mortality \[[@B1]\]. Although the majority of PCa cases are localised to the prostate, nearly one-third of newly diagnosed patients have advanced or metastatic disease \[[@B2]\]. Radical prostatectomy and radical radiotherapy are potentially curative treatment options for localised PCa. For those with locally advanced or metastatic disease, the initial systemic therapy is androgen deprivation therapy. Unfortunately, nearly all patients in this group eventually progress to castration resistant PCa (CRPC). For CRPC, docetaxel based chemotherapy is used as first-line. However, this only provides modest survival benefit (2.4 months) and is associated with significant side effects \[[@B3]\]. Insights into the regulation of immune responses in malignancies have facilitated the emergence of novel immune-based strategies. PCa is an attractive target for vaccination due to its slow growth which can allow sufficient time for immune activation \[[@B4]\]. Additionally, the identification of prostate tumour associated antigens (that are recognised by T cells) has created the opportunity to develop novel immune based therapeutic approaches. Several tumour associated antigens have been identified, including prostate specific antigen (PSA), prostatic acid phosphatase (PAP), prostate stem cell antigen (PSCA) and prostate specific membrane antigen (PSMA) \[[@B5]\]. These antigens are largely prostate specific and their expression is strongly upregulated in PCa, both locally and at metastatic sites \[[@B6],[@B7]\]. This makes these antigens a viable target of active immunotherapy and can be used as DNA vaccines. Furthermore, PSA-specific cellular immune responses have been detected in some PCa patients and in normal individuals, suggesting that tolerance towards this antigen may be broken \[[@B8]\]. These observations led to development of a variety of vaccines for PCa in pre-clinical and clinical trials. These vaccines can be broadly classified as whole cell vaccines \[[@B9]\], protein/peptide based vaccines \[[@B10],[@B11]\], viral vector based vaccines \[[@B12]\], dendritic cell vaccines \[[@B13]\] and naked DNA vaccines \[[@B14],[@B15]\]. Like other forms of targeted therapy, cancer vaccines hold the promise of achieving cancer control without inducing overt toxicity. The focus of extensive research for PCa vaccine has led to the approval of the first therapeutic vaccine (sipuleucel-T) by the FDA \[[@B16]\]. The sipuleucel-T is an autologous antigen presenting cell (APC) based and antigen-targeted immunotherapeutic innovation for men with CRPC \[[@B16]\]. The success story of the sipuleucel-T encouraged researchers to explore other agents/strategies to activate immune system against PCa. Gene therapy (including DNA vaccines) is a realistic prospect for the treatment of prostate and other cancers, and involves the delivery of genetic information to the patient to facilitate the production of therapeutic proteins. A DNA vaccine consists of tumour specific/associated antigen and additional immune-stimulatory factors cloned into a bacterial plasmid downstream of an appropriate eukaryotic promoter for strong and stable expression. DNA vaccination can induce effective anti-tumour responses against various malignant cells and provides an attractive strategy for the management of PCa \[[@B17]-[@B19]\]. Why DNA vaccines? ================= DNA immunisation can efficiently stimulate humoural and cellular immune responses to protein antigens. This strategy has been used successfully for infectious diseases \[[@B20]\] and potentially can be applied for malignant conditions. There are several advantages associated with DNA vaccines: Various groups have demonstrated tumour protection using DNA immunisations in several pre-clinical cancer models \[[@B21]-[@B26]\]. Gene sequences can be manipulated easily to provide multiple potential epitopes that stimulate both humoural and cellular immunity \[[@B22],[@B27]-[@B29]\]. DNA vaccines generally skew the immune system toward the desired T-helper1 (Th1) response. This is most likely because plasmid DNA contains unmethylated CpG motifs, which have been shown to be a very potent immunological adjuvant \[[@B30],[@B31]\]. DNA immunisation is safe in humans \[[@B14],[@B32]\] and can induce antigen-specific immune responses \[[@B30],[@B33]\]. DNA vaccines can be produced readily at a large scale \[[@B34]\]. Mechanisms of action ==================== Anti-tumour immune responses ---------------------------- For malignant diseases, immunological therapies fall into two general categories; active and passive immunotherapy. Active immunotherapy attempts to directly elicit tumour-specific host immune responses that control or eradicate tumours. In contrast, passive immunotherapy involves the direct administration of effector molecules, such as cytokines or monoclonal antibodies. These molecules promote the development of anti-tumour responses, directly killing tumour cells, or inhibit cell invasion and angiogenesis \[[@B35]\]. The nature of the host immune response that can control tumour growth has been the focus of many studies. With few exceptions, the most effective anti-tumour immune responses in animal models have depended on the efficient generation of Th1 cell immunity, characterised by strong cytotoxic T lymphocyte (CTL) responses \[[@B27],[@B28],[@B36]\]. This is supported by observations in humans that progressive disease is characterised by an anti-tumour T-cell response \[[@B37]\] as well as by clinical evidence that Th1 T cells can control and eliminate metastatic disease \[[@B28],[@B38]\]. B cell effector functions are another important component of this anti-tumour immune response (see below). Process of DNA vaccination -------------------------- A typical DNA vaccine consists of a transgene that encodes the sequence of a target protein (e.g. PSA, PSMA etc.) under the control of a eukaryotic promoter. Various modalities exist for delivering such DNA to appropriate cells (see below). After uptake of the plasmid, the target protein is produced within the cell and processed into small antigenic peptides by host proteases. The peptides then enter the lumen of the endoplasmic reticulum (E.R.) by membrane-associated transporters. In the E.R., the peptides bind to Major Histocompatability Complex I (MHC I) molecules. These peptides are presented on the cell surface in the context of MHC I. Subsequently CD8^+^ CTL are stimulated resulting in cell-mediated immunity. CTLs cause tumour destruction through both cytolysis of malignant cells and non-cytolysis mechanisms such as cytokines production. The foreign protein on the plasmid can also be presented by MHC II pathway by APC which elicit CD4^+^ helper T cells responses. These CD4^+^ T cells can recognise the peptides generated from the exogenous proteins that were endocytosed or phagocytosed by APC, then degraded to peptide fragments and loaded onto MHC II molecules. Depending on the type of CD4^+^ T cell that binds to the complex, B cells are activated and antibody production is stimulated. This is the same manner in which traditional vaccines work \[[@B39]\]. Vaccine-elicited antibodies can mediate direct effects against tumour cells by fixing complement or facilitating antibody-dependent cellular cytotoxicity. Overall, stimulation of both the T and B cell arms of the immune system mediates synergies and creates a large pool of effectors cells to control tumour growth and induce generation of memory cells (Figure [1](#F1){ref-type="fig"}). {#F1} Methods of DNA delivery ======================= Transfection of the host cell with the plasmid is a limiting step for a successful DNA vaccination. Most commonly, the DNA vaccines are given by intramuscular (i.m.) or intradermal (i.d) injections. To facilitate the gene delivery, various methods have been reported including viral, liposomal, bacterial, ultrasound and electroporation (EP), as well as approaches involving *ex vivo* transfection/transduction of cells (e.g. APC) \[[@B40]\]. To avoid safety issues such as immune response and cytotoxicity associated with viruses and liposomal transfection, physical methods (ultrasound, EP etc.) have been widely used for either *in vivo* or *ex vivo* gene delivery. *In vivo* EP has emerged as a potent method for DNA vaccine delivery and significantly improves transfection efficiency of naked plasmid DNA \[[@B41]\]. EP driven DNA vaccination increases antigen expression by increasing transfection efficiency (Figure [2](#F2){ref-type="fig"}) and is accompanied by local tissue injury and inflammation \[[@B42]\]. Hence, the outcomes of EP mediated vaccination are dramatic enhancement of humoural and cellular immunity \[[@B43],[@B44]\]. {#F2} Preclinical studies =================== Numerous preclinical studies have been carried out to demonstrate the efficacy of DNA vaccines in animal models of PCa. The following vital issues regarding successful vaccines were addressed in these preclinical trials; Induction of immune response/type of immune activation Optimisation of vaccine dose/schedule Mode of delivery Breaking tolerance to self-antigen Systemic responses /control of metastatic disease Need for adjuvant to vaccines Transferable immunity Prostate-specific antigens examined ----------------------------------- Roos *et al.*\[[@B45]\] showed for the first time that a PSA DNA vaccine can induce anti-tumour immunity *in vivo*. They investigated a DNA vaccination strategy to immunise mice by inducing PSA-specific cellular responses. A plasmid expressing PSA, alone or in combination with plasmids coding for granulocyte-macrophage colony-stimulating factor (GM-CSF) and/or IL-2, were used. It was found that the DNA vaccine induced PSA-specific cytotoxic T lymphocytes and when co-injected with GM-CSF and IL-2 it can protect mice against a PSA-expressing tumour challenge. This demonstrated that immunisation with a PSA DNA vaccine can evoke PSA-specific cellular immunity. The ability of the DNA vaccine to stimulate both humoural and cellular immune responses was demonstrated by Kim *et al.*\[[@B24]\]. They observed a strong and persistent antibody response against PSA for at least 180 days following immunisation. Additionally, significant T helper cell proliferation was also detected against PSA. Furthermore, immunisation with PSA plasmid induced CD8^+^ T cell-restricted cytotoxic T cell response against tumour cell expressing PSA. PSMA is a 750 amino acid surface protein expressed primarily in prostate epithelium, Most reported antibodies to PSMA apparently recognise epitopes in the residue 43--570 region of the extra cellular domain, and upon binding are rapidly removed from the cell surface by internalisation. This would potentially limit their ability to mediate Fc-dependent cytotoxicity. Kuratsukuri *et al.*\[[@B46],[@B47]\] used a model system to target a defined region of the extra cellular domain of PSMA. Their results suggested that vaccination with plasmid expressing extra cellular domain of PSMA induced adaptive humoural activity, which was directed against the extracellular region of human PSMA and can significantly inhibit human PCa growth in athymic mice. Viral delivery of vaccine antigens is an active research area. However, a potential difficulty with viral-based immunisations is that immune responses elicited to the viral vector might limit the possibility of multiple immunisations. Johnson *et al.*\[[@B48]\] investigated a DNA vaccine encoding PAP, to elicit antigen-specific CD8^+^ T cell immune responses. In their study, Lewis rats were immunised with either a plasmid DNA-based (pTVG-HP) or Vaccinia-based (VV-HP) vaccine each encoding human PAP (hPAP). They observed Th1-biased immune response (as indicated by proliferating PAP-specific CD4^+^/CD8^+^ cells and IFN- *γ* production) in rats immunised with a DNA vaccine encoding hPAP. Immunisation with Vaccinia virus (encoding hPAP) could not induce PAP-specific response, unless boosted with a heterologous vaccination scheme. Furthermore, they also established that multiple immunisations with a DNA vaccine encoding the rat PAP homologue (pTVG-RP) could overcome peripheral self-tolerance against rat PAP (rPAP) and generate a Th1-biased antigen-specific CD4^+^ and CD8^+^ T cell response. In separate experiments, this DNA vaccine has not shown significant toxicities in terms of animal weights, histopathology, haematological changes, or changes in serum chemistries \[[@B49]\]. The vaccine was found to be effective in eliciting PAP-specific CD4^+^ and CD8^+^ T cells, predominantly Th1 in type, in all immunised animals at all doses and numbers of immunisations. PAP-specific IgG were detected in a dose-dependent fashion, with titres increasing after multiple immunisations. Optimisation of DNA vaccination ------------------------------- In preclinical trials different doses and vaccination schedules are investigated with variable results however, there is lack of consensus on these issues. Ahmad *et al.* have shown that the a DNA vaccine encoding human PSA significantly delayed the appearance of tumours and resulted in prolonged survival of the animals \[[@B50]\]. Additionally, a four-dose vaccination regimen provided optimal immunological effects and co-administration of synthetic CpG enhanced the tumour protective responses \[[@B50]\]. Furthermore, these immune responses were tumour specific and were transferable in adoptive T cell transfer experiments \[[@B50]\]. A DNA vaccination has the potential to break tolerance to self antigen \[[@B34],[@B48],[@B51]\]. A PSCA DNA vaccine when delivered by i.m. EP \[[@B51]\] or orally administered bacteria, \[[@B52]\] resulted in induction of anti-tumour immune responses against PSCA expressing subcutaneous tumours and metastatic deposits. There was activation of Th-1 type immunity against PSCA, indicating the breaking of tolerance to a self-antigen \[[@B51]\]. This immunity was tumour specific and was transferable by adoptive transfer of splenocytes \[[@B51]\]. PSMA is present in both secretary form and in prostate cells. Mincheff *et al.*\[[@B53]\] evaluated two plasmid DNA vaccines, encoding either PSMA products that are retained in the cytosol and degraded in the proteasome (tVacs; hPSMAt), or secreted proteins (sVacs; hPSMAs) for stimulation of cytotoxic cell or antibody responses. They observed that immunisation with both vectors led to generation of cell cytotoxicity, provided GM-CSF was administered with the vaccine. Spleen cells from animals immunised with hPSMAt demonstrated stronger cytotoxicity to the target cells. Interestingly, priming with a vector that encoded a xenogeneic protein (hPSMAt; 'xenogeneic' construct) and boosting with a vector that encoded an autologous protein (rPSMAt; 'autologous' construct) gave the best protection against tumour challenge. Immunisation with tVacs did not lead to formation of antibodies to the target protein, while immunisation with sVacs or with the protein did (mixed Th1-Th2 isotype). But, priming with tVacs and boosting with protein also resulted in the antibodies from the cytotoxic Th1 isotype. Hence, the best strategy to obtain a strong cellular cytotoxic response seems to be gene-based vaccinations with tVacs, priming with the 'xenogeneic' and boosting with the 'autologous' constructs. While DNA vaccine injections, either i.m. or i.d., have been used successfully in many trials, it is still to establish which route is better. To enhance the efficacy of DNA vaccine against PCa, Roos *et al.*\[[@B54]\] have demonstrated that i.d. DNA vaccination, followed by two sets of electrical pulses of different length and voltage, can effectively induce PSA-specific T cells response. Ahmad *et al.* have reported successful DNA vaccination following i.m. injection coupled with EP. \[[@B50],[@B51]\] These studies indicate that EP significantly enhances transfection and hence immune activation. Clinical trials =============== Preclinical studies provided encouraging evidence of enhanced immune responses and tumour protection by DNA vaccine in animal models of PCa. The success achieved in these studies resulted in exploration of application of DNA vaccine in PCa patients. To our knowledge, no naked DNA vaccine has been used in a randomised clinical trial to date \[[@B55]\]. However, DNA vaccines have been used in phase I/II clinical trials including patients with PCa (Table [1](#T1){ref-type="table"}). Some of these clinical trials are discussed below. ###### Summary of prostate cancer DNA vaccination clinical trials **References** **Antigen /+− co stimulatory molecules** **No. of patients/ patient's characteristics** **Type of study** **Route of vaccination** **Immunological responses** **Adverse effects** **PSA response** ---------------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------ ------------------- -------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------- \[[@B14]\] Extracellular human PSMA & CD~86~ into separate expression vectors (PSMA & CD~86~ ), and into a combined plasmid (PSMA/CD~86~) 26 Phase I/II i.d. \- All patients who received initial inoculation with viral vector followed by PSMA plasmid boosts showed immunisation. In contrast, with PSMA and CD~86~ plasmids, only 50% were immunised. \- \- \+ Expression cassette from PSMA plasmid into a replication deficient adenoviral expression vector \- Of the patients who received PSMA & GM-CSF, 67% were immunised. However, PSMA/CD~86~ & GM-CSF vaccination immunised all recipients. \[[@B15]\] Plasmid vector expressing PSA & GM-CSF/IL-2 9 CRPC Phase I i.m, i.d. PSA-specific cellular immune response (measured by IFN- *γ* & anti-PSA IgG levels) were detected in highest dose cohort of patients. \- Systemic effects; running nose, fatigue, myalgia, chills and fever (*n = 6*). \- Drop in PSA (*n=3*). \- At the injection site; erythema, swelling, induration, itching, pain, urticaria (*n = 7*). \- Increase in PSA (*n= 5*). \[[@B56]\] Vaccine encoding a domain of fragment C of tetanus toxin fused to a tumour-derived epitope from PSMA 5 patients / dose level Phase I/II, i.m. or i.m. + EP Delivery of DNA+EP at all five vaccinations resulted in activation of humoral immunity. \- Mild pain at injection site. \- Recurrent PCa \- EP did not add toxicity. \[[@B57]\] Vaccine encoding PAP co-administered with GM-CSF 22 Stage D~0~ PCa Phase I/IIa i.d. \- Three of 22 patients developed PAP-specific IFN-*γ* secreting CD8^+^ T-cells. While 9 (41%) patients developed PAP-specific CD4^+^ and/or CD8^+^ T-cell proliferation. No significant adverse events PSA doubling time increased from a median 6.5 months per treatment to 8.5 months on-treatment & 9.3 months in one year post treatment. \- Antibody responses to PAP were not detected. *PSMA* Prostate Specific Membrane Antigen, *CD*~*86*~ Cluster of Differentiation 86, *i.d* Intradermal, *i.m* Intramuscular, *PSA* Prostate Specific Antigen, *GM-CSF* Granulocyte-macrophage colony-stimulating factor, *IL2* Interleukin 2, *CRPC* Castrate Resistant Prostate Cancer, *IFN-γ* Interferon Gamma, *PCa* Prostate Cancer, *PAP* Prostate Acid Phosphatase. In a phase I trial, Pavlenko *et al.*\[[@B15]\] investigated the feasibility, safety and immunogenicity of a DNA vaccine (pVAX/PSA) in patients with CRPC. Cytokines, GM-CSF and IL-2, were also used as vaccine adjuvants. The results of this trial demonstrated that DNA vaccination with a PSA-coding plasmid vector, given with GM-CSF and IL-2, is safe and can induce cellular and humoural immune responses against PSA. However, a dose--response was observed with regard to induction of a PSA-specific immune response. Interestingly, in this trial, two patients that developed cellular immune response to PSA exhibited stabilisation of disease. While only one of six that did not develop PSA-reactivity showed clinical stabilisation. Mincheff *et al.*\[[@B14]\] performed phase I/II trials to determine the safety of the PSMA vaccine after repeated i.d. injections. Twenty-six patients with PCa were entered into this toxicity-dose escalation study. Immunisations were performed i.d. at weekly intervals. Doses of DNA between 100 and 800 μg and of recombinant virus at 5×10^8^ PFUs per application were used. They observed no immediate or long-term side effects following immunisations. All patients who received initial inoculation with the viral vector followed by PSMA plasmid boosts showed the development of a delayed-type hypersensitivity reaction after the PSMA plasmid injection. In contrast, to the patients who received a PSMA plasmid and CD86 plasmid, only 50% showed signs of successful immunisation. Of the patients who received PSMA plasmid and soluble GM-CSF, 67% were immunised. However, all patients who received the PSMA/CD86 plasmid and sGM-CSF became immunised. The patients who did not immunise during the first round were later successfully immunised after a boost with the viral vector. Further to this study, Todorova *et al.*\[[@B58]\] characterised the humoural immune response against PSMA in PCa patients. They demonstrated that PSMA is a target for humoural immune response induced by gene-based PSMA vaccination. It is also proposed that detection of anti-PSMA antibodies by immunoblot analysis and by indirect immunofluorescence could be used to monitor the vaccination effects. These results were quite encouraging, proving evidence of immune activation with different vaccination regimens. However, the heterogeneity of the medical status and the presence of concomitant hormone therapy do not permit unequivocal interpretation of the data with respect to the effectiveness of the therapy. However, several responders, as evidenced by a change in the local disease, distant metastases, and PSA levels, were identified in this cohort. Low *et al.*\[[@B56]\] evaluated the use of EP to deliver a novel DNA vaccine (p.DOM-PSMA~(27)~). This vaccine encodes a domain of fragment C of tetanus toxin to induce CD4^+^ T cell help, fused to a tumour-derived epitope from PSMA for use in HLA-A2^+^ patients with recurrent PCa. In this open label phase I/II, two-arm, dose escalation trial, the DNA vaccine was delivered either by i.m injection or by i.m. injection followed by EP. Three vaccinations were given at 0, 4, and 8 weeks, with booster doses at 24 and 48 weeks. In 20 patients with first two dose cohorts, EP did not appear to add toxicity to the vaccination apart from brief and acceptable pain at injection site. They also observed highest level of humoural responses with DNA+EP strategy and these responses persisted to 18 months of follow-up. These data favour EP as a potent method for stimulating humoural responses induced by DNA vaccination in humans. McNeel *et al.*\[[@B57]\] conducted a phase I/IIa trial with a DNA vaccine encoding human PAP in patients with stage D~0~ PCa. Twenty-two patients were treated in a dose-escalation trial with 100 μg, 500 μg, or 1,500 μg plasmid DNA, co-administered i.d. with 200 μg GMC-SF as a vaccine adjuvant, six times at 14-day intervals. All of these patients were observed for one year after the treatment. They did not observe any significant adverse events. Three of 22 patients developed PAP-specific IFN-*γ* secreting CD8+ T-cells immediately after the treatment course. While nine (41%) out of 22 patients, developed PAP-specific CD4+ and/or CD8+ T-cell proliferation. However, no humoural response (antibodies against PAP) was detected. Overall, the PSA doubling time was observed to increase from a median 6.5 months pre-treatment to 8.5 months on-treatment (*P* = 0.033), and 9.3 months in the 1-year post-treatment period (*P* = 0.054). This study provided 12 months follow up data and demonstrated that the DNA vaccine is safe, elicits an antigen-specific T-cell response, and may be associated with an increased PSA doubling time. Conclusions and future directions ================================= DNA vaccination for PCa is at a crucial developmental stage. The ultimate goal of any given immunotherapy including DNA vaccination is eradication of each and every cancer cell from the patient. However, this goal may be hard to achieve. DNA vaccination is a step forward in achieving immune eradication of PCa. Induction of tumour-specific T cell activation has been demonstrated with PCa DNA vaccination in both preclinical and clinical trials. However, in clinical settings, limited success has been seen in terms of tumour regression and survival. This vaccine failure may be attributed to several potential tumour escape mechanisms such as defects in antigen presentation, production of immunosuppressive substances, T cell dysfunction, and the presence of regulatory T cells \[[@B59]-[@B61]\]. Additionally, there is paucity in the literature of advanced clinical trials to evaluate the role of DNA vaccination in PCa. Furthermore, in the available trials there is currently a lack of long-term follow up. Ideally, the availability of data from randomised clinical trials featuring robust end points such as biochemical response, progression free and overall survival will provide categorical evidence for DNA vaccination's potential. In order to provoke an immune response, a tumour vaccine should not only maximise antigen-specific signals, but should also provide the necessary "co-stimulatory" environment. One approach is to include lymphokines (GM-CSF, IL-12, IL-15) or include tumour cell expression of membrane bound molecules (CD80, CD86) \[[@B62]\]. Furthermore, doses and schedules need optimisation; however it is clear that the immune responses depend on a primary vaccination followed by booster vaccination(s). Some studies have suggested that the best strategy for achieving an intense immune response can be priming with naked DNA followed by boosting with a viral vector \[[@B14],[@B63]\]. The use of viral vectors can enhance the immunogenicity of the vaccine due to the adjuvant properties of some of the viral products. However, naked DNA immunisation offers several potential advantages over viral mediated transduction. Among these are the inexpensive production and the inherent safety of plasmid vectors, as well as the lack of immune responses against the carrier \[[@B63]\]. Although the current evidence suggest that the DNA vaccines can induce immune activation resulting in PCa control, certain areas still need to be explored such as selecting the ideal antigen, identifying suitable stages of PCa for vaccine therapy, optimum dosage/schedule and adjuvant agents. Evidence suggests that it is unlikely for a single therapy to achieve the goal of curing PCa especially patients with CRPC. However, better understanding of immunobiology of PCa will lead to pathways for development of novel therapies. The perceived difficulty in translating the effectiveness of DNA vaccination from small animals to human subjects appears to have been overcome by use of EP \[[@B42]\]. Current data are from early stage studies and further research in this field is essential to determine the place of DNA vaccination as an adjuvant to surgery, radiotherapy and chemotherapy. Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== SA carried out literature search, interpreted the data and drafted the manuscript. MT, PS and GCOS helped with data interpretation and drafting the manuscript. All authors read and approved the final manuscript. Acknowledgement =============== This work was supported by the Cork Cancer Research Centre.

Dear Editor-in-Chief ==================== WHO goals for oral health 2020 target to reduce the number of edentulous persons and to increase the individuals with functional dentition. The prevalence of edentulismn differs among countries (5% Switzerland, 8% Denmark, 40% Poland, 53% Bulgaria, 78% Bosnia and Herzegovina) ( [@B1] ). In Poland, between 1998 and 2009, the percentage of edentulous patients increased from 34.7% to 43.9% ( [@B2] ). The Oral Health Impact Profile is an instrument including conceptual dimensions of OHRQoL ( [@B3] ) giving greater weight to psychological and behavioral outcomes. Our aim was to assess oral health impacts related to the Oral Health Impact Profile in edentulous patients with complete dentures from Bialystok, Poland. The participants were recruited from two randomly chosen dental clinics in Bialystok, which constituted 11.1% of accessible public dental clinics. The inclusion criteria were: a public dental service, a prosthodontist, and patients with complete dentures. Participants consecutively filled in the questionnaire during their dental checkup. The participation was anonymous, voluntary. Patients unable to fill in the questionnaire were excluded from the survey. The Polish version of the OHIP-14 was used ( [@B4] ). The values of Cronbach's alpha (0.969) were calculated to assess the internal consistency for the whole score and for particular items removed. The statistical analysis was performed using the IBM SPSS Statistics 20.0 software. Statistical hypotheses were verified at *P*\<0.05. The final sample consisted of 100 edentate people. The age ranged from 49 to 97 yr, (mean age 72.65+/−9.25). [Table 1](#T1){ref-type="table"} shows mean values of the total OHIP-14 and its dimensions in relation to different variables. The age was significantly associated with the functional limitation dimension. Level of education, dry mouth, chewing ability and self-rated oral health indicated significant association with total OHIP-14 and its dimensions. In [table 2](#T2){ref-type="table"} chewing problems, dry mouth and self-rated oral health were significantly associated with the OHIP-14. Participants who reported chewing problems had a 6.17 times greater odds of having the OHIP-14 score above the median than those without chewing problems. ###### Mean values (SD) of the total OHIP-14 scores and its dimensions by age, gender, education, dry mouth, chewing ability and self-rated oral health **Variables** **OHIP total** **Functional limitation** **Pain and discomfort** **Psychological impact** **Behavioral impact** ------------------------------------------- ---------------- --------------------------- ------------------------- -------------------------- ----------------------- Age (yr)[\*](#TFN1){ref-type="table-fn"} \<59 15.17(12.98) 1.83(0.40) 2.17(2.48) 6.33(5.53) 4.83(4.99) 60--74 25.83(15.00) 3.75(2.10) 3.42(1.96) 9.91(5.86) 8.75(5.77) 75\< 26.66(14.79) 4.20(2.32) 3.66(2.12) 10.00(5.59) 8.80(5.54) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.74) 8.54(5.66) *P* 0.241 0.022 0.217 0.352 0.280 Gender Male (41) 27.05(14.37) 3.93(1.99) 3.53(1.98) 10.54(5.64) 9.00(5.62) Female 24.47(15.30) 3.73(2.33) 3.34(2.13) 9.17(5.79) 8.22(5.72) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.74) 8.54(5.66) *P* 0.453 0.606 0.530 0.348 0.557 Education[\*](#TFN1){ref-type="table-fn"} Primary (39) 28.10(13.67) 4.28(1.98) 4.00(2.05) 10.49(5.33) 9.33(5.28) Secondary 26.79(15.20) 3.92(2.23) 3.50(1.97) 10.27(5.89) 9.10(5.81) Tertiary (13) 13.15(11.95) 2.08(1.89) 1.54(1.33) 5.46(4.82) 4.08(4.34) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.74) 8.54(5.66) *P* 0.008 0.008 0.001 0.017 0.010 Dry mouth No (13) 19.10(14.03) 3.00(2.04) 2.51(1.94) 7.45(5.51) 6.14(5.24) Yes 32.22(12.79) 4.67(2.02) 4.41(1.74) 12.10(5.00) 11.04(5.00) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.74) 8.54(5.66) *P* 0.000 0.000 0.000 0.000 0.000 Chewing ability No(51) 16.31(12.14) 2.77(1.92) 2.23(1.64) 5.85(4.41) 5.46(4.55) Yes 26.91(14.85) 3.98(2.19) 3.62(2.07) 10.31(5.71) 9.00(5.69) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.743 8.54(5.66) *P* 0.017 0.027 0.019 0.008 0.046 Self rated oral health Bad 30.81(14.70) 4.42(2.13) 4.09(1.95) 11.77(5.72) 10.53(5.66) Good(57) 18.53(12.15) 3.02(2.02) 2.58(1.91) 7.02(4.56) 5.91(4.52) Total 25.53(14.90) 3.82(2.19) 3.44(2.07) 9.73(5.74) 8.54(5.66) *P* 0.000 0.001 0.000 0.000 0.000 Kruskal--Wallis\*; Mann--Whitney's U ###### Odds ratios of achieving OHIP-14 score above median value **Variable** **OR** **OHIP-14 95%CI** ------------------------------------- -------- --------------------------------------------- Chewing problems (yes / no) 6.17 1.29--29.49[\*](#TFN2){ref-type="table-fn"} Dry mouth (yes / no) 4.12 1.79--9.48[\*](#TFN2){ref-type="table-fn"} Self-rated oral health (bad / good) 8.19 3.25--20.60[\*](#TFN2){ref-type="table-fn"} Gender (male / female) 1.47 0.66--3.27 95%Cl does not include 1; Our study is the first in Poland and Eastern Europe conducted on OHRQoL and associated factors in edentulous people using complete dentures. Our findings may be useful in developing preventive strategies and dealing with their impairments. The two most often reported domains were psychological and functional outcomes. They included "uncomfortable to eat foods", "self-conscious" and "unsatisfactory diet". A substantial proportion of participants demonstrated a number of oral health problems which were significantly associated with dry mouth, chewing problems, poor self-rated oral health, lower level of education and age over 75. Our data revealed significant level of impairment in elderly people using complete dentures (median-27; mean 25.53). The retained chewing function is a condition for a varied diet, and nutritional status which is significant at an older age. Our findings are consistent with results of the survey on chewing ability of elderly population conducted in different countries ( [@B5] ). A poor OHRQoL in edentulous persons was clearly related to the occurrence of dry mouth, and this condition had a significant impact on all dimensions of OHRQoL. Level of education has different impact on oral health-related quality of life ( [@B6] ). Age showed no impact on the OHIP total score, but the problems of participants over 75 years old were significantly associated with the difficulty in pronouncing words and a worse sense of taste. Concluding, the Polish version of the OHIP-14 demonstrated good reliability and validity. Dry mouth, chewing problems and poor self-rated oral health, level of education and age over 75 were strongly associated with poor OHRQoL. The authors declare that there is no conflict of interests.

1. Introduction {#sec1} =============== Sarcopenia is defined as decreased skeletal muscle mass and muscle strength with age. Muscle mass and strength gradually decrease after reaching a peak in early adulthood, and the degree of decrease varies among individuals \[[@B1]\]. Elderly people with sarcopenia have a much higher fall risk and lower physical performance than do nonsarcopenic individuals \[[@B2]\]. Decreased muscle strength also reduces functional capacity and is a major cause of disability, mortality, and other adverse health outcomes \[[@B3], [@B4]\]. Because of individual differences, it is important to reduce the rate at which muscle mass declines to avoid premature sarcopenia. Sarcopenia can be evaluated by measuring skeletal muscle mass. It is common practice to examine the cross-sectional area, thickness, and weight of muscles using magnetic resonance imaging (MRI), computed tomography (CT), anthropometry, bioelectrical impedance analysis (BIA), and ultrasound. Muscle mass and strength are reduced in the third decade, and the prevalence of sarcopenia can be increased by the presence of obesity and the amount of physical activity. Therefore, managing the risk factors of sarcopenia through exercise is important in young and healthy adults \[[@B1], [@B5]\]. In addition, low physical performance can be assessed using functional measurements such as gait speed (e.g., 4 minute walking test) and grip strength \[[@B4], [@B6]\]. Muscle quality may be more important than muscle size in estimating the risk of falling, and monitoring muscle activity during daily activities can help in preventing sarcopenia and estimating the degree of frailty \[[@B7], [@B8]\]. It is also important to evaluate muscle quality in healthy elderly people before and after exercise and according to age \[[@B9], [@B10]\]. Muscle activity can be monitored and muscle quality can be evaluated, through surface electromyography (sEMG) \[[@B11], [@B12]\]. However, the sEMG devices developed so far are expensive and difficult to operate, which limits their use by nonspecialists. Therefore, a new sEMG device , that is, simple to use and highly accessible has been developed for people who are not familiar with EMG. The purpose of this study was to establish the validity and reliability of the new device. 2. Materials and Methods {#sec2} ======================== 2.1. Experimental Protocol {#sec2.1} -------------------------- In total, 20 healthy participants (10 males, 10 females) between the ages of 21 and 34 years (males, age 30.3 ± 2.9 years, height 171.9 ± 3.8 cm, weight 74.1 ± 11.3 kg, body mass index (BMI) 25.4 ± 3.33 kg/m^2^; females, age 22.3 ± 2.7 years, height 162.1 ± 5.0 cm, weight 56.4 ± 5.0 kg, BMI 21.5 ± 1.9 kg/m^2^; mean ± SD) were recruited; all participants who provided informed consent prior to the study were recruited. Ethical approval was granted, and the informed consent form was approved by the Ethics Committee of Pusan National University Hospital, Busan, Korea (IRB number: 1703-018-052). Exclusion criteria included musculoskeletal disease, cardiopulmonary disease, and other diseases that could prevent exercise. At each session, participants were first required to perform three maximal voluntary isometric contractions (MVIC) for 5 seconds each, with a 5-minute rest between contractions. Each session consisted of two trials. After three MVIC measurements, 15-second isometric contractions were performed at different intensity levels. In the pretest, it took at least 10--15 seconds to maintain the same intensity isometric contraction through visual feedback. First, 40% MVIC was performed, followed by 80% MVIC after 5 minutes of rest. In the second trial, the placement of the electrodes for the two devices (BTS-FREEEMG1000 and newly developed device) was interchanged, and contractions were again measured by the same method ([Figure 1](#fig1){ref-type="fig"}). During the test, participants received visual feedback about their performance from a monitor, which enabled them to maintain the muscle contraction at the target intensity. The same procedure was employed for the biceps brachii (BB) and rectus femoris (RF) muscles \[[@B13]--[@B15]\]. All tests were performed only with the dominant arm and leg. Participants were tested twice, with a week between sessions. 2.2. Mechanical Recording {#sec2.2} ------------------------- The participants sat on a Biodex System 3 PRO dynamometer (Biodex Medical Systems, Shirley, NY, USA) with a visual torque feedback monitor. Each participant sat in an upright posture and was strapped firmly to the chair with adjustable belts across the arm, trunk, and thigh. To evaluate BB muscle contractions, the participant sat with the dominant arm flexed at 90° and the forearm flexed at 120° relative to the upper arm. To evaluate quadriceps (RF) contractions, the hips were flexed at 90° and the tested knee was flexed at 45°. The axis of the dynamometer was positioned at the center of the tested elbow or knee joint. The lever arm was fixed by the precalibrated force sensor \[[@B16]\]. 2.3. EMG Recording {#sec2.3} ------------------ The EMG signal was recorded simultaneously using two different sEMG devices. The BTS-FREEEMG (BTS-FREEEMG1000; BTS Bioengineering, Milan, Italy) was set to a sampling rate of 1,000 Hz per channel, and the signals were band-pass filtered from 20 to 500 Hz. The newly developed sEMG device (PSL-EMG-Tr1; PhysioLab Co., Ltd., Busan, Korea) was set to a sampling rate of 30,000 Hz, and signals were amplified with a 3--2,000 Hz bandwidth ([Figure 2(a)](#fig2){ref-type="fig"}). Adhesive hydrogel surface electrodes (35 mm teardrop-shaped Kendall™ 200 Foam Electrodes; Medtronic, Minneapolis, MN, USA) were used, and the interelectrode distance, electrode placement procedure, and skin preparation followed standard Surface Electromyography for the Non-Invasive Assessment of Muscles (SENIAM) guidelines \[[@B17]\]. Two pairs of surface electrodes were attached parallel to the muscle fibers at an interelectrode distance of 2.0 cm. The distance between the pairs of electrodes was also 2.0 cm. After the first trial, the second trial was performed by interchanging the positions of the two pairs of electrodes of the each EMG devices ([Figure 2(b)](#fig2){ref-type="fig"}). After the interchange of the electrodes, the average of the values was used to compare the concurrent validity of the two devices. The root mean square (RMS) value was used to analyze and process the recorded electrical signals in the muscles. Based on the square root calculation, the RMS reflects the mean power of the signal and is the preferred recommendation for smoothing. The RMS value can be used as a parameter to reflect the physiological activities of the motor unit during muscle contraction \[[@B18]\]. 2.4. Statistical Analysis {#sec2.4} ------------------------- Sample size was calculated using G^*∗*^Power software (ver. 3.1; Heinrich--Heine Universität, Düsseldorf, Germany). In this study, the number of subjects required for a null-correlation (R0) = 0, alternative correlation (R1) = 0.6, alpha = 5%, power = 80%, and two-tailed test value was 19. Statistical analysis was performed using SPSS software (ver. 18.0; SPSS Inc., Chicago, IL, USA). To determine concurrent validity between the two sEMG machines, Pearson\'s correlation coefficient (*r*) was used for the average of two trials in each device. Interpretation of the correlation coefficients was based on guidelines for Pearson\'s coefficients suggested by Portney and Watkins \[[@B19]\]: *r* \> 0.75, good to excellent correlation; *r* = 0.50--0.75, moderate to good correlation; *r* = 0.25--0.50, fair correlation; and *r* = 0.00--0.25, little to no relationship \[[@B19]\]. The Bland--Altman plot was used to visually compare the mean values of the two trials in each device. Mean differences were calculated by subtracting the % MVIC of the PSL-EMG-Tr1 from the % MVIC of the BTS-FREEEMG1000. Limits of agreements (LOA) were calculated by using 2 standard deviations around the mean difference. The intraclass correlation coefficient (ICC) of two trials performed on each week was used to indicate the relative reliability of the measurements. For the test-retest reliability, ICC using a two-way mixed-effects model and absolute agreement definition is used \[[@B20]\]. Munro\'s descriptors for reliability coefficients were used to index the degree of reliability: very high correlation, 0.90--1.00; high correlation, 0.70--0.89; moderate correlation, 0.50--0.69; low correlation, 0.26--0.49; and little or no correlation, 0.00--0.25 \[[@B21]\]. The paired *t*-test was also conducted comparing the RMS (*μ*V) and torque (N·m) between two trials of each week. Biodex is a device that has proved its reliability and validity. Therefore, Biodex was used only to evaluate the exact intensity during muscle contraction, and validity was compared between two sEMG devices \[[@B22]\]. 3. Results {#sec3} ========== 3.1. Torque Measurements {#sec3.1} ------------------------ Twenty participants completed a total of four trials over 2 weeks. [Table 1](#tab1){ref-type="table"} shows the peak torque (N·m) values at three isometric contractions levels (MVIC, 40% MVIC, and 80% MVIC) for the first and second weeks. Very high relative reliability was found at all three isometric contraction levels for both muscles (ICC: 0.985--0.994 for BB; 0.948--0.981 for RF). 3.2. Amplitude of sEMG {#sec3.2} ---------------------- Recorded sEMG data were processed for RMS analysis. The amplitudes (*μ*V) of the nonnormalized RMS values recorded by the PSL-EMG-Tr1 devices at the three contractions levels for the first and second weeks are shown in [Table 1](#tab1){ref-type="table"}. Moderate to very high relative reliability was found for all three contraction levels in both muscles (ICC: 0.832--0.937 for BB; 0.814--0.957 for RF). Overall, the reliability at various contraction levels was slightly lower for the PSL-EMG-Tr1 than for the Biodex device. This may be because the muscle group generating the torque includes other muscles in addition to the one measured by sEMG; this is discussed further below. To compare EMG activity in the same muscle on different days or different individuals, or to compare EMG activity between muscles, the EMG must be normalized. Normalization of EMG signals (% MVIC) is shown in [Table 2](#tab2){ref-type="table"}. Normalization of EMG signals is performed by dividing the EMG signals during the submaximal isometric contraction by a maximal EMG signal (MVIC). The normalized RMS values showed no to high relative reliability in BB and moderate to high relative reliability in RF (ICC: 0.030--0.831 for BB; 0.547--0.828 for RF). There were no statistical differences of normalized RMS values for two sEMG devices between the first and second trial (*p* \> 0.05; [Table 2](#tab2){ref-type="table"}). 3.3. Validity {#sec3.3} ------------- [Figure 3](#fig3){ref-type="fig"} shows the validity of the two sEMG devices. Pearson\'s *r* values were used to evaluate validity because all of the % MVIC values measured in the BB and RF muscles were normally distributed. The two sEMG devices were compared using averages of all four trials for two weeks of 40% MVIC and 80% MVIC. The 40% MVIC displayed excellent validity for BB (*r* = 0.907) and RF (*r* = 0.965), and the 80% MVIC showed good to excellent validity for BB (*r* = 0.781) and RF (*r* = 0.757). [Figure 4](#fig4){ref-type="fig"} shows Bland--Altman plots which show the dispersion of the % MVIC of the two sEMG devices. The mean difference in % MVIC was small (0.0--1.1) with the majority of the data points within the 95% limits of agreement. 4. Discussion {#sec4} ============= In this study, we confirmed the reliability and validity of the newly developed sEMG device for monitoring muscle activity during exercise and in daily life. The nonnormalized RMS values measured on both devices showed high reliability (ICC: 0.832--0.937 for BB; 0.814--0.957 for RF). The normalized RMS values showed good to excellent validity (*r* = 0.781--0.907 for BB; *r* = 0.757--0.965 for RF) and showed nonsignificant results on the paired *t*-test (*p* \> 0.05). Especially in the BB muscle, there was little reliability because of the low ICC of normalized RMS values. Gaudet et al. reported the intersession reliability of maximal contraction of the elbow flexor group (BB, brachialis, and brachioradialis muscles) using sEMG. In this study, a relatively low ICC (range: 0.57--0.80) was seen in the BB in intersession single measurement. However, the average of repeated sEMG measurement was considered to obtain high reliability rather than single measurement \[[@B23]\]. Kollmitzer et al. evaluated the intersession reliability of measures of the knee extensor group (RF, vastus lateralis, and medialis muscles) and found that the overall reliability was good, especially for the RF muscle \[[@B16]\]. To improve reliability, testing in the lower limb can be a better choice than the upper limb and requires repeated measurements rather than a single measurement. Our study also confirmed higher ICC in RF than in BB. And our study compared the single measurements in each trial, which is one of the reasons for the low ICC. Based on previous studies and our results, it seems that EMG signal reproducibility has a great effect on the selection of certain muscles in upper and lower limbs. In elderly populations, physical activity is reduced, with less than one-fifth of elderly individuals engaging in the recommended level of physical activity \[[@B24]\]. Commercially available computer-based physical activity monitors have been developed in recent years; these devices can enhance motivation in elderly individuals, record physical activity, and support physical exercise \[[@B25], [@B26]\]. In frail elderly persons, single-repetition maximum resistance training (RT) at 30% MVIC or greater can significantly improve muscle strength, muscle power, and functional outcomes. Therefore, supervised and controlled RT can be an effective measure against frailty \[[@B27]\]. RT also has neuromuscular benefits, and changes in muscle quality can be monitored via sEMG \[[@B28]\]. Applying wearable devise to monitor physical activities is good for feasibility and effectiveness and can encourage exercise through self-monitoring and goal setting \[[@B29]\]. Developing sEMG devices with high accessibility will therefore be very useful in establishing therapeutic strategies and evaluating the muscle condition of elderly people. The newly developed sEMG device is small in size with low cost, and therefore has good portability ([Table 3](#tab3){ref-type="table"}). These advantages can also increase the accessibility of muscle monitoring during various physical activities or exercise. To evaluate the accuracy of the new device, we recruited young people free of disease and disability. However, elderly individuals may have chronic diseases that could cause peripheral neuropathy or myopathy due to disuse atrophy, so the quality of surface EMG data may vary. However, such difficulties would not indicate a problem with the accuracy of the device itself. In other words, it is possible to compare muscle activity within one individual, but caution is needed when comparing individuals with one another. Despite these limitations, it is very encouraging that there is a tool that allows easy and objective evaluation of muscle quality in elderly people. For the clinical application of the device, further studies will be necessary for healthy elderly or sacropenic patients. This study had the other limitation that should be addressed: the electrodes were attached according to the SENIAM guidelines, but it is considered that the 2.0 cm interelectrode distance was the main cause of the lower ICC in this study. However, this problem will not affect the clinical application of muscle monitoring over time for people undergoing monitoring or engaging in exercise, as only one device will be used. This is because ICC was also lower in clinically widely used sEMG device (BTS-FREEEMG1000), and good to excellent validity was found in normalized RMS values of the two sEMG devices. The newly developed sEMG device is wired for a single channel, although a wireless model for two or more channels is under development. A high sampling rate was used for the sEMG device in this study. Since most signals from the human muscles have frequency characteristics that are valid at less than 400 Hz, we recommend that the signal analysis sample more than twice the major quard of interest frequency. While we do not need this sampling frequency for existing RMS and MDF analysis, we are interested in signal characteristics that we have not known before by increasing the maximum sampling frequency. This will be used for further study of the characteristics of EMG signals of damaged muscles. Therefore, the newly developed instrument is measuring at a sampling frequency much higher than the frequency of interest. However, since many burdens are expected in the signal processing during commercialization, the sampling rate will be reduced to 2,000 Hz; it will enable fast signal processing and long recording time. In addition, if the EMG signals from the patient\'s muscles are automatically stored and the system is programmed so that this information can be delivered to medical staff located elsewhere, the therapeutic value of the device will increase still further. 5. Conclusion {#sec5} ============= Signals from the BB and RF muscles, recorded by the newly developed PSL-EMG-Tr1 device, showed good to excellent validity and moderate to high ICC values with nonnormalized RMS values. However, low ICC values were seen with the normalized RMS values. Although the sEMG itself may have limitations, it can be overcome somewhat through repeated measurements and appropriate muscle selection. Since it has a high correlation compared to conventional sEMG devices, it can be used as an alternative to conventional sEMG devices. The newly developed device may be effective to evaluate and monitor the condition of individuals\' muscles during repetitive daily activities or exercise, and it has higher accessibility and portability than do conventional sEMG devices. This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation, funded by the Korean government, Ministry of Science, ICT and Future Planning (2016M3A9E8942066). Data Availability ================= The data used to support the findings of this study are available from the corresponding author upon request. Conflicts of Interest ===================== The authors declare that they have no conflicts of interest. {#fig1} {#fig2} {#fig3} {#fig4} ###### Reliability comparison of the Biodex System 3 PRO and BTS-FREEEMG, PSL-EMG-Tr1. Device Variable Trial 1 Trial 2 ICC Difference of means (95% CI) *p* value ----------------------------- ----------------- ----------------- ----------------- ----------------------- ------------------------------ --------------------- ------------ *(a) Biceps brachii muscle*               MVIC Biodex (N·m) Week 1 36.65 ± 14.56 35.98 ± 13.78 0.985 0.68 (−0.9, 2.3) 0.387 Week 2 38.60 ± 14.53 38.59 ± 14.27 0.986 0.01 (−1.6, 1.6) 0.990 BTS-FREEEMG (*μ*V) Week 1 460.50 ± 257.49 395.52 ± 216.69 0.930 64.98 (14.4, 115.5) 0.014^*∗*^ Week 2 470.34 ± 243.38 413.91 ± 208.48 0.848 56.43 (−18.7, 131.5) 0.132 PSL-EMG-Tr1 (*μ*V) Week 1 398.07 ± 231.36 435.05 ± 281.16 0.937 −36.98 (−96.3, 22.3) 0.207 Week 2 392.39 ± 227.36 435.65 ± 268.89 0.875 −43.26 (−120.4, 33.9) 0.255 80% MVIC Biodex (N·m) Week 1 29.11 ± 11.65 28.52 ± 10.66 0.992 0.60 (−0.3, 1.5) 0.197 Week 2 31.10  ± 11.18 30.95 ± 11.52 0.986 0.15 (−1.1, 1.4) 0.807 BTS-FREEEMG (*μ*V) Week 1 364.27 ± 220.15 333.32±183.99 0.934 30.95 (−15.2, 77.1) 0.177 Week 2 387.76 ± 220.15 356.38 ± 192.92 0.859 31.08 (−37.13, 99.3) 0.352 PSL-EMG-Tr1 (*μ*V) Week 1 304.71 ± 198.40 354.03 ± 218.37 0.872 −49.32 (−115.1, 16.4) 0.132 Week 2 324.07 ± 185.61 368.97 ± 233.94 0.916 −44.90 (−97.6, 7.8) 0.090 40% MVIC Biodex (N·m) Week 1 14.41±5.74 14.55 ± 5.09 0.986 −0.14 (−0.8, 0.5) 0.645 Week 2 15.48 ± 5.72 15.75 ± 6.00 0.994 −0.28 (−0.7, 0.2) 0.199 BTS-FREEEMG (*μ*V) Week 1 118.42 ± 81.22 101.81 ± 75.70 0.920 16.61 (−2.5, 35.7) 0.084   Week 2 121.36 ± 75.72 115.91 ± 69.50 0.906 5.45 (−14.8, 25.7) 0.579 PSL-EMG-Tr1 (*μ*V) Week 1 96.69 ± 67.98 94.32 ± 57.30 0.922 2.36 (−14.2, 19.0) 0.768 Week 2 94.98 ± 53.06 116.68 ± 72.75 0.832 −21.81 (−43.1, −0.5) 0.045^*∗*^ *(b) Rectus femoris muscle*               MVIC Biodex (N·m) Week 1 147.80 ± 40.50 154.08 ± 36.38 0.948 −6.28 (−13.9, 1.4) 0.103 Week 2 159.18 ± 46.04 160.05 ± 45.95 0.981 −0.88 (−6.8, 5.1) 0.762 BTS-FREEEMG (*μ*V) Week 1 147.82 ± 65.95 144.25 ± 59.11 0.945 3.58 (−10.0, 17.2) 0.588 Week 2 175.99 ± 59.53 156.78 ± 59.53 0.864 19.21 (−2.7, 41.1) 0.082 PSL-EMG-Tr1 (*μ*V) Week 1 123.96±49.66 124.87 ± 55.45 0.899 −0.91 (−16.1, 14.3) 0.901 Week 2 129.90 ± 52.05 138.30 ± 56.79 0.907 −8.40 (−23.2, 6.4) 0.249 80% MVIC Biodex (N·m) Week 1 115.58 ± 31.82 121.67 ± 30.34 0.964 −6.09 (−10.9, −1.3) 0.015^*∗*^ Week 2 127.06 ± 35.52 126.45 ± 35.75 0.977 0.61 (−4.5, 5.7) 0.805 BTS-FREEEMG (*μ*V) Week 1 118.59 ± 80.28 115.13 ± 51.59 0.910 3.47 (−15.0, 21.9) 0.699 Week 2 136.71 ± 67.52 123.54 ± 48.67 0.898 13.21 (−2.9, 29.3) 0.103 PSL-EMG-Tr1 (*μ*V) Week 1 95.46 ± 57.58 101.13 ± 49.94 0.922 −5.66 (−19.3, 8.0) 0.396 Week 2 100.46 ± 41.52 109.22 ± 45.45 0.943 −8.76 (−17.6, 0.1) 0.051 40% MVIC Biodex (N·m) Week 1 57.70 ± 15.90 61.50 ± 15.37 0.956 −3.80 (−6.3, −1.3) 0.005^*∗*^ Week 2 63.59 ± 18.34 63.81 ± 17.74 0.979 −0.23 (−2.7, 2.2) 0.851 BTS-FREEEMG (*μ*V) Week 1 43.45 ± 17.65 44.53 ± 16.57 0.931 −1.08 (−5.2, 3.0) 0.590 Week 2 50.89 ± 23.43 48.05 ± 21.21 0.883 2.84 (−4.0, 9.6) 0.393 PSL-EMG-Tr1 (*μ*V) Week 1 36.15 ± 14.45 39.43 ± 18.16 0.814 −3.28 (−9.3, 2.8) 0.271 Week 2 38.44 ± 15.68 41.24 ± 18.17 0.957 −2.80 (−5.8, 0.2) 0.070 Values are number or mean ± SD. MVIC, maximum voluntary isometric contraction; ICC, intraclass correlation coefficients; paired *t*-test, ^*∗*^*p* \< 0.05; CI, confidence interval. ###### Normalization of RMS and reliability of BTS-FREEEMG and PSL-EMG-Tr1.   Variable Trial 1 Trial 2 ICC Difference of means (95% CI) *p* value ----------------------------- --------------- --------------- --------------- -------------------- ------------------------------ -------------------- ------- *(a) Biceps brachii muscle*               80% MVIC (%) BTS-FREEEMG Week 1 78.92 ± 13.00 84.84 ± 18.92 0.321 −5.93 (−15.6, 3.7) 0.213 Week 2 81.44 ± 16.97 84.09 ± 19.76 0.185 −2.65 (−14.2, 8.9) 0.637 PSL-EMG-Tr1 Week 1 76.10 ± 14.83 82.08 ± 19.99 0.030 −5.98 (−17.8, 5.9) 0.306 Week 2 83.14 ± 15.51 82.62 ± 17.93 0.246 0.53 (−11.1, 12.2) 0.925 40% MVCI (%) BTS-FREEEMG Week 1 27.52 ± 10.17 26.83 ± 10.89 0.862 0.69 (−2.8, 4.2) 0.683 Week 2 26.86 ± 9.62 28.01 ± 11.22 0.662 −1.14 (−6.1, 3.8) 0.636 PSL-EMG-Tr1 Week 1 26.07 ± 9.39 24.43 ± 10.10 0.831 1.63 (−1.9, 5.2) 0.350 Week 2 26.07 ± 9.49 28.16 ± 11.58 0.307 −2.09 (−8.4, 4.3) 0.500 *(b) Rectus femoris muscle*               80% MVIC (%) BTS-FREEEMG Week 1 76.46 ± 18.41 79.91 ± 12.59 0.509 −3.45 (−11.9, 5.0) 0.406 Week 2 76.70 ± 12.19 78.22 ± 10.13 0.736 −1.53 (−6.4, 3.3) 0.517 PSL-EMG-Tr1 Week 1 74.87 ± 19.24 80.50 ± 11.95 0.547 −5.63 (−13.9, 2.7) 0.171 Week 2 77.56 ± 14.27 78.81 ± 10.54 0.580 −1.25 (−7.7, 5.2) 0.688 40% MVCI (%) BTS-FREEEMG Week 1 30.43 ± 7.32 32.11 ± 7.78 0.870 −1.68 (−4.0, 0.7) 0.152 Week 2 30.07 ± 8.18 31.07 ± 8.06 0.752 −1.01 (−4.4, 2.4) 0.545 PSL-EMG-Tr1 Week 1 29.93 ± 7.93 32.35 ± 8.23 0.765 −2.42 (−5.7, 0.8) 0.132 Week 2 30.69 ± 7.89 30.84 ± 7.93 0.828 −0.15 (−3.0, 2.7) 0.912 Values are number or mean ± SD. MVIC, maximum voluntary isometric contraction; ICC, intraclass correlation coefficients; paired *t*-test, ^*∗*^*p* \< 0.05; CI, confidence interval. ###### Technical specifications of newly developed sEMG device (PSL-EMG-Tr1; PhysioLab Co., Ltd., Busan, Korea) and BTS-FREEEMG (BTS-FREEEMG1000; BTS Bioengineering, Milan, Italy). --------------------------------------------------------------------------------------------------------------------------------------------------   PSL-EMG-tr1 BTS-FREEEMG1000 ------------------------------------ --------------------------------------- ---------------------------------------------------------------------     Price ∼\$500 USD ∼\$25,000 USD Dimensions (mm) 48 L × 93 W × 15.5 H\ EMG probes:\ Lead wire length\ 41.5 L × 24.8 W × 14 H main electrode Ø 16 × 12 satellite electrode 1,100 mm (main 500 mm; branch 600 mm) USB receiver: 82 L × 44 W × 22.5 H Charger: 350 L × 185 W × 20 H Weight (g) EMG device: 47 g\ EMG probes: 10 g Lead wire: 41 g USB receiver: 80 g Charger: 1450 g Channels 1 channel Up to 10 wireless probes Bandwidth (Hz) 3--2,000 25--500 Hz Gain (V/V) 25 2,000 Sampling rate (Hz) 30,000 1,000 Common mode rejection (dB) 73 92 -------------------------------------------------------------------------------------------------------------------------------------------------- [^1]: Academic Editor: Fumiharu Togo

Background ========== Busy clinicians must choose carefully what reports of medical research findings to read, and in searching the literature should choose publications of relevance to practice with the methodological rigour capable of changing practice \[[@B1]\]. Physicians\' perception of research quality \[[@B2]\] and generalisability \[[@B3]-[@B5]\] are likely to affect their willingness to change practice in response to research findings. Little research has been done to determine if the site of research or publication affects the likely impact of research findings on clinical practice. The need to look at this issue is even greater in developing countries where resources to do research are limited \[[@B5],[@B6]\] and physicians have a more acute need to rely on research done in other countries to guide their clinical decisions. The purpose of this study is to determine the likely impact of research location and journal location on physicians\' practice in developing countries. More specifically, the objectives are to answer the following questions: how likely is research published in journals from different regions of the world, including their own, to effect change in physicians\' clinical practice? and, how likely is research done in different regions of the world, including their own, to effect change in physicians\' clinical practice? In order to determine how perception of research quality affects the latter answer, we repeated the latter question with the proviso that research quality is the same in all regions. In addition, we also seek to identify the factors that are likely to explain variation in responses. Methods ======= The International Clinical Epidemiology Network (INCLEN) is a network dedicated to improving the quality of health research in the Developing World through institutional capacity building for evidence based medicine \[[@B7],[@B8]\]. Clinicians who are members of the International Clinical Epidemiology Network (INCLEN) were invited to participate, and six centres agreed: Shanghai and Chengdu in China, Bangkok in Thailand, Nagpur in India, Ismalia in Egypt and Nairobi in Kenya. The study was coordinated in the Centre for Clinical Epidemiology and Biostatistics in Newcastle, Australia. Each centre was asked to identify hospitals and physicians within those hospitals who would be expected to treat patients with pneumonia, in a way that would represent the generality of tertiary and secondary hospital settings in their region. Physicians within the hospitals were chosen from those working in Internal Medicine either at random or to represent a spread of academic/non-academic and seniority levels. A questionnaire was given to each consenting doctor. The sampling procedure varied between centres due to local circumstances. In Bangkok, Chengdu, Shanghai and Nagpur a sample of tertiary care (3 in Bangkok and Nagpur, 5 in Chengdu and Shanghai) and secondary care (4 in Bangkok and 5 in Chengdu, Shanghai and Nagpur) hospitals were selected either at random or to cover the spread of teaching/non-teaching, geography and hospital size. In Ismailia, a random sample was taken from a list of doctors working in all three city hospitals. In Nairobi, a list of all the physicians in the country rather than hospitals was the sampling frame. The questionnaire asked respondents how they rate research in journals from North America, Europe, their region and their country with regard to the likelihood of influencing their clinical practice. They were also asked the same question, but with reference to clinical research from the same regions. They were then asked to re-answer the latter question with the assumption that research quality is the same in all regions. Answers were given on a scale of 1 to 5, (very unlikely, unlikely, neutral, likely, very likely) to influence clinical practice. The study formed part of a larger study which examined variations in stated clinical practice based on a case scenario of a patient with pneumonia \[[@B9]\]. One reminder was sent to non-respondents. The questionnaire was translated into the local language by respective investigators at local centres. Pretesting was done prior to the definitive study, where each investigator gave the questionnaire to a sample of physicians to assess comprehension and feasibility. On the basis of this pretest, the Thai sample excluded the question that assumed equal quality as it was found that this modification did not change the physicians\' perceptions in that centre. Questionnaires or computer discs with coded data were sent to the coordinating centre in Newcastle, where analyses were performed. Statistical analysis -------------------- The data were tabulated and proportions calculated. The differences in impact score (the 5 point Likert scale that assessed likely influence) between current research and research assuming equal quality were assessed using the Wilcoxon ranked sum test. The p values here were adjusted after calculating the relevant design effect induced by the clustered nature of the data. The proportional odds model \[[@B10]\] for ordered categorical data was used to analyse the impact scores; the comparison is presented as a proportional odds ratio of more influence compared with the baseline category. The proportional odds assumption was checked \[[@B11]\] and where appropriate, generalised ordered logistic regression models were fitted instead \[[@B12]\]. The models were fitted to the data using the Huber estimator of variance \[[@B13],[@B14]\]. The models thus took account of the fact that individuals were clustered within hospitals which were stratified by the centres within which the samples were taken. Statistical significance was determined by p values after calculating Wald and F ratio statistics. In some cases, the 5-point scale was collapsed into a 3- or 4-point outcome to reduce problems caused by zero cells. The variables investigated include sex, number of years since graduation, physician specialty, access to a medical library, rural versus urban/suburban location and country of practice. Variables remained in the model if the relevant p value was less than 0.10. Graphs are presented to show the difference between perceived influence of respective research/journals in comparison with local research/journals as the reference. If the difference was -2 or less, then the graph reported \"prefer local\"\' if the difference was -1, 0, or 1, then the graph reported \"little difference\", and if the difference was greater than or equal to 2, the graph reported \"prefer other\". The statistical program Stata release 5.0 \[[@B15]\] was used for all the analyses. All p values are two sided. Results ======= Response rates were high, with one exception. The Chinese and Indian samples had response rates of 100%. The response rates in the Egyptian, Thai and Kenyan samples were 91%, 80% and 48% respectively. Table [1](#T1){ref-type="table"} shows the demographic and practice features of the physicians in the sample. ###### Demographic and practice features of the physicians included in the study SPECIALTY (%) ---------- ---- ---- ---- --------- ------------- ------------- --------------- --------- --------- --------- Chengdu 10 50 25 26 (52) 40 (23--60) 15 (1--37) 14 (28) 36 (72) 0 (0) 49 (98) Shanghai 10 50 25 28 (56) 40 (22--64) 14.5(1--40) 1 (2) 46 (92) 3 (6) 44 (88) Bangkok 7 40 25 10 (25) 28 (24--49) 4.5 (1--24) 25 (63) 15 (38) 0 (0) 37 (93) Nagpur 8 28 18 11 (39) 38 (26--51) 15 (2--30) 17 (61) 4 (14) 5 (18) 24 (86) Ismalia 3 20 7 1 (5) 28 (26--47) 4 (3--23) 3 (16) 16 (84) 0 (0) 5 (75) Nairobi \* 40 \* 10 (25) 40 (34--51) 14 (10--36) 11 (28) 17 (43) 12 (30) 30 (78) \* The physicians in Kenya were not selected by hospital. Factors affecting physician journal preferences ----------------------------------------------- Table [2](#T2){ref-type="table"} and Figure [1](#F1){ref-type="fig"} show the distribution of preferences for publications in journals from different regions. In general, North American journal articles were ranked fairly highly in ability to influence clinical practice with some variation from country to country. Ordinal logistic regression revealed that country was the only factor that statistically significantly affected physicians\' impression of the likely effect of studies published in North American journals on their practice: physicians from Kenya and Egypt reported that these publications were most likely to influence a change in practice and Thai doctors were the least likely to be so influenced (F~3,31~= 6.48, p = 0.0007). {#F1} ###### The likelihood of journals published/ research done in various regions to affect physicians clinical practice according to country of practice. Figures are the number (%) of physicians choosing the highest two of five influence categories. Number (%) of physicians likely to be influenced ------------------- ---------------- -------------------------------------------------- ---------- Country Origin Journals Research China (n = 100) North American 61 (61) 55 (55) European 19 (19) 20 (20) Regional 22 (22) 14 (14) Local 96 (96) 94 (94) Thailand (n = 40) North American 19 (48) 19 (48) European 8 (20) 15 (38) Regional 7 (18) 12 (30) Local 26 (65) 33 (83) India (n = 28) North American 17 (61) 17 (61) European 18 (64) 20 (71) Regional 14 (50) 16 (57) Local 26 (93) 27 (96) Egypt (n = 20) North American 14 (70) 19 (95) European 11 (55) 17 (85) Regional 1 (5) 1 (5) Local 2 (10) 7 (35) Kenya (n = 40) North American 29 (73) 29 (73) European 34 (85) 32 (80) Regional 32 (80) 27 (68) Local 36 (90) 30 (75) Overall (n = 228) North American 140 (61) 139 (61) European 90 (39) 104 (46) Regional 76 (33) 70 (31) Local 186 (82) 191 (84) Egyptian, Indian and Kenyan were more likely than Chinese doctors to be influenced by European journals \[odds ratios 6.5 (95% CI 3.1,13.7); 6.7 (95% CI 2.7, 16.3); and 23.9 (95%CI 8.8, 65.0) respectively, F~3,31~= 12.0, p \< 0.0001\] while the Thai responses were not statistically significantly different from those of their Chinese counterparts. Physicians working in tertiary care hospitals were more likely to be influenced by European journals than those in secondary hospitals \[odds ratio 2.3, (95% CI 1.3,4.0)\] and the other factors studied were not statistically significantly related to likely influence of European journals. In general, the physicians studied were unlikely to change their practice on account of papers in regional medical journals (that is, those from regions surrounding the country of practice), although the Indian and Kenyan physicians were exceptions to this \[odds ratios 3.4 (95% CI 1.1, 10.5) and 10.3 (95% CI 2.7, 39.5) relative to Chinese physicians respectively\]. Subspecialist physicians were less likely to be influenced by regional journals relative to primary care and other doctors \[odds ratio 0.56 (95% CI 0.33, 0.94)\]. Physicians from Kenya, China, India and to a lesser extent Thailand were likely to be influenced to change clinical practice as a result of local publications. This however was not the case with the Egyptian physicians \[odds ratio 0.008 relative to Chinese physicians, (95% CI 0.0004,0.11)\]. Physicians in urban and suburban centres were more likely to be influenced by local journals relative to those from rural centres \[odds ratio 2.2, (95% CI 0.99, 4.9)\]. Subspecialist physicians revealed less tendency to be influenced by local journals relative to primary care physicians \[odds ratio 0.33, (95% CI 0.13, 0.84)\]. Factors affecting likely influence of research done in different regions ------------------------------------------------------------------------ Table [2](#T2){ref-type="table"} and Figure [2](#F2){ref-type="fig"} show the distribution of preferences for research performed in different regions. Research done in North America ranked fairly highly in ability to change clinical practice with significant variation from country to country. The proportional odds assumption was not fulfilled and thus stratum specific estimates were obtained. Kenyan and Egyptian physicians were much more likely to be influenced by North American research than physicians in India, Thailand and China (χ^2^~12~= 35, p = 0.005). Those who had access to a medical library (compared to those without) \[χ^2^~3~= 14.9, p = 0.002\] and those working in tertiary hospitals (compared to those in secondary hospitals) \[χ^2^~3~= 10.4, p = 0.016\] were also more likely to be influenced. In addition, more experienced physicians were more likely to choose the highest two Likert categories of influence than those less experienced (χ^2^~3~= 13.6, p = 0.004). {#F2} Kenyan, Egyptian and Indian physicians appeared more likely to be influenced by European research than their Chinese and Thai colleagues (F~4,31~= 12.7, p \< 0.0001). Physicians working in tertiary hospitals were twice as likely as their colleagues in secondary hospitals to be influenced by European research \[odds ratio 2.0, (95% CI 1.1, 3.6)\]. The other factors studied were not statistically significantly associated with the likely influence of European research. Regional medical research was more likely to influence physicians in India and Kenya compared to those in Egypt and in China (F~4,31~= 13.8, p \< 0.001). There was also a tendency for subspecialist physicians to be less readily influenced by regional research compared to physicians without subspecialty training (F~1,34~= 7.9, p = 0.008). The proportional odds assumption was not fulfilled in the analysis of local research. Local research was judged as very likely to change clinical practice by most physicians studied except those from Egypt. Local research was most likely to change practice in physicians from China and India. (χ^2^~8~= 42.5, p \< 0.001). Other factors positively associated with likely change of practice by local research in the highest Likert category were being primary care physician (compared to those with subspecialty training and other doctors, χ^2^~4~= 21.9, p \< 0.001); access to a medical library (χ^2^~2~= 12.9, p = 0.002); and seniority as measured by years since medical school graduation (χ^2^~2~= 11.6, p = 0.003). Difference in research influence scores if research quality was the same in all regions --------------------------------------------------------------------------------------- Very little change was observed in the influence scores obtained for North American research if research quality became the same from region to region (Wilcoxon signed ranked sum test adjusted for clustering, z = -1.1, p = 0.28). The median change in influence score was 0 (interquartile range 0,0). This was different in Egypt and to a lesser extent in Kenya (χ^2^~6~= 23.9, p \< 0.001) where 50% and 30% of physicians decreased their scores respectively. The findings for European research were similar in that there was no statistically significant overall change in influence scores (Wilcoxon signed ranked sum test adjusted for clustering, z = -0.64, p = 0.53). The median change in influence score was 0 (interquartile range 0,0). Egyptian and Kenyan physicians reported reduced scores here relative to their Indian and Chinese colleagues (χ^2^~6~= 23.0, p \< 0.001). With regard to regional research, there was a statistically significant increase in perceived influence reported by physicians if research quality should become the same in all regions (Wilcoxon signed ranked sum test adjusted for clustering, z = 4.6, p \< 0.0001). The median change in scores was 0 (interquartile range 0, +1). Although statistically significant, there was less country to country variation in this instance (χ^2^~6~= 19.0, p = 0.004). Physicians in the Kenyan sample were more likely than their colleagues in the other countries studied to have an increase in influence scores. Subspecialist physicians were more likely to increase their influence score relative to primary care physicians and other doctors \[χ^2^~2~= 16.6, p \< 0.001\]. There was also a statistically significant increase in influence scores for local research if research quality became the same. (Wilcoxon signed ranked sum test adjusted for clustering, z = 2.15, p = 0.031). Seventeen percent of physicians showed an increase in their influence scores. There was also a significant country effect here (F~2,27~= 4.6, p = 0.01). The proportion of Egyptian and Kenyan physicians showing an increase in influence scores was 35% and 25% respectively. Subspecialist physicians and other doctors were more likely to increase scores relative to primary care physicians (F~2,27~= 6.2, p = 0.006). Physicians with access to a medical library were less likely to increase scores relative to those without (F~1,28~= 6.24, p = 0.019). None of the other factors studied had a statistically significant effect on change in the influence scores should research quality become the same. Discussion ========== With the exception of Egyptian physicians, research findings published in local journals are more likely to result in change in clinical practice relative to journals published in other regions followed closely by research findings published in North American journals. This was demonstrated by the fact that more than 80% of the physicians in this study chose the highest two influence categories indicating a high willingness to change practice in response to findings in local journals. This contrasted with approximately 60% for North American journals and less for European and regional publications. The relative influence of research done in different regions is similar to the pattern for journal information. It is also clear that the physicians\' impressions of the difference in research quality in different regions affect the degree to which they are willing to change their practices. This was evident from the fact that there was a statistically significant increase in influence scores for local and regional research if research quality is considered the same. The changes are most evident among the Kenyan and Egyptian physicians. The changes in impact scores for North American and European research are however not statistically significant. These contrasting findings suggest that developing world physicians think that the quality of medical research in North America and Europe is better than that in their own regions and countries. This study also demonstrated that there is significant variation between countries in the likely influence of journals from and research studies done in different regions. Most physicians are likely to be influenced by North American publications. There is more variation with regard to the likely effects of European publications on physician practice: physicians from Kenya, Egypt and India are more likely to be influenced by European research relative to those sampled in China and in Thailand. This may directly relate to the relative contact between the medical establishment in Europe and those in these countries. Europe has had a longer history of influence in the medical establishments in Kenya, India and in Egypt relative to those in China and in Thailand. Differences in language may also add to this influence. This study also reveals that physicians working in tertiary care hospitals are more likely to be influenced by North American and European publications than physicians from secondary care hospitals. This may relate to the greater exposure these physicians have to publications and research done in North America and Europe. There is also much variation with regard to the physicians\' impressions of the likely impact of regional research and publication on their practice. Kenyan and Indian physicians are more likely to be influenced by their regional publications and research than are physicians from the other countries studied. Egyptian physicians are especially unlikely to be influenced by their regional journals The design of this study involved random sampling of physicians after an initial random sampling of hospitals. This was however not carried out uniformly, and where random sampling was performed the method was left to the individual investigators. There is a strong possibility of selection bias being present in this study, thus limiting the interpretation of between-country differences in the results. It is unlikely, however, that any such selection would be related to the outcome factor examined (the relative importance of the source of the research or publication) and hence internal validity should not be compromised. In some centres all hospitals were used since there were only a few physicians located in each. In Kenya, a national sampling frame was used rather than identifying hospitals first, and the 48% response rate indicates uncertainty about the validity of the results. We report answers to a questionnaire rather than observations on practice, and have not established the validity of the stated responses. It is possible that \'national pride\' may explain the large difference seen between local and regional journals. The understanding of \'region\' may also be difficult, we gave examples in the question of East Africa, Asia and Latin America. In addition, it is possible that the influence and credibility of various information sources may be different for different clinical problems in different settings. The study did not differentiate between type of research study -- a randomized controlled trial would usually be more highly regarded than a descriptive study, wherever it was conducted or published. In order to allow for this issue, we asked the question about change in perceptions of the research if the quality were the same in all regions. This is the first study to assess the differences in likely impact of medical research and medical journals published in different parts of the world on physicians\' practices. The study was carried out in developing countries where few resources are available for doing local medical research and for guiding health policy \[[@B6]\], although the burden of disease is great \[[@B16]\]. Insufficient numbers of clinical trials are performed in sub-Saharan Africa despite the heavy disease burden \[[@B17]\]. Hepatitis B and C \[[@B18]\], the AIDS epidemic \[[@B19]\], the emergence of resistant strains of organisms to antibiotics \[[@B20]\], the need for culture specific and cost-effective methods for child care \[[@B19]\], and appropriate contraceptive methods \[[@B21]\] are only a few of the problems facing developing countries. Given these burdens and that so little financial resources are available for health, it is essential that doctors in developing countries use the most cost-effective methods of health management. Although the respondents to our survey reported high levels of access to medical libraries (Table [2](#T2){ref-type="table"}), and also reported high levels of access to \"up to date\" medical journals, we do not know which journals they are or if they were read. Unfortunately, even in the \'best\' settings worldwide, medical practice is not necessarily driven by peer-reviewed evidence. It is therefore important that we identify how physicians use evidence to guide their practice. This can in turn lead to appropriate education programs to guide developing world physicians on how to use evidence. Evidence-based practice needs to be taught to developing world medical practitioners \[[@B22]\]. Initiatives like the International Clinical Epidemiology Network which build research and education capacity in evidence based medicine \[[@B7],[@B8]\] should therefore be encouraged and supported. In addition, given that physicians are more likely to respond to local research than research from other countries, local researchers need to be given support to improve the quality and quantity of local research output. This obviously makes sense since local research is more likely to be directly applicable to the population involved \[[@B5]\]. However, it is neither sensible nor cost-effective to repeat every study in local settings. It is therefore important that the development of culturally sensitive evidence-based guidelines which guide physicians on how to use the results of research findings from settings other than their own be encouraged. Conclusions =========== Since local research and publications were considered most likely to change clinical practice, the conduct of high quality local research is likely to be an effective way of getting research findings into practice in developing countries. Local research should be encouraged through education and collaboration and supplemented by appropriate education programs to guide physicians on how to use evidence. Competing interests =================== None declared. Authors\' contributions ======================= The study was conceived by RFH and JP and designed by all authors in collaboration. Statistical analysis was performed by JP and LL. The paper was drafted by JP and RFH with contributions from LL and SK, and approved and corrected by all authors. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2458/3/6/prepub> Acknowledgments =============== Funding was provided by the International Clinical Epidemiology Network (INCLEN).

Introduction ============ Presently, low-cost tin-oxide chemical sensors applied to gas detection present some well-known problems (lack of selectivity, drift, etc.) which motivates active research in material science, different measurement strategies and signal-processing algorithms, different measurement strategies including sensors arrays, static and dynamic measurements, etc. Several attempts have been focused on dynamic measurements which include temperature transient or pulsed techniques and temperature modulation through oscillation of heater voltage because they have suggested that temperature modulation of tin-oxide sensor provides more information from a single sensor than static measurement \[[@b1-sensors-04-00095]-[@b19-sensors-04-00095]\]. However, these works always focused on the identification of certain gases such as H~2~S, CO, NO~2~, CO~2~, ethanol, methane, n-butane, ethane, propane, propylene, ammonia, and so on under a given constant heating waveform, frequency and operation temperature. In our previous work, we reported rapid detecting of pesticide residues using temperature modulation by employing only a single sensor rather than an array. We also reported the amplitudes of the higher harmonics of the FFT signal exhibited characteristic changes that depended not only on the species of pesticide gases but also on the concentration of pesticide gases by means of analysis dynamic responses \[[@b20-sensors-04-00095], [@b21-sensors-04-00095]\]. In the present paper, we will report on the advantages of the dynamic measurement and discuss the influencing factors of nonlinear responses such as modulation temperature, frequency, heating shape waveform (rectangular, triangular, saw-tooth, pulse, sinusoidal) throughout the experimental trails. Experimental ============ The sol was prepared using a mixed ethanol solution of SnCl~2~, SbCl~3~, CaCl~2~ and SrCl~2~. The amounts of Sb, Ca and Sr elements were fixed to 2.5 mol% as M/Sn, respectively. As a binder, a given amount of commercial glass powder was added in doped SnO~2~ powder formed by calcinating the powder at 500 °C for half an hour in air. The screen-printing technique was used to prepare SnO~2~ thick films on alumina ceramic substrates with a RuO~2~ layer as a heating element on the back. The thick films were sintered at various temperatures for half an hour in air to obtain doped SnO~2~ TF gas sensors. These elements were aged in working temperature until reproducible steady-state resistances were obtained. A headspace sample (HP-7694) was used to inject sample gases measured into a 2500 ml sensor test chamber, where a single SnO~2~ gas sensor was kept. Butanone, acetone, ethanol, methanol, formaldehyde and cyclohexanone (Analytical standard, provided by Sigma-Aldrich Laborchemikalien Gmbh) were measured. The signal-producing circuit board was made by ECU Electronics Industrial Co., LTD (No. 38 Research Institute of China Electronics & Technology Group Corp., China). The signal frequency and operating temperatures settings were adjusted to obtain good sensitivities and selectivity to the pesticide gases measured. The sensor resistance was monitored, acquired and stored in a PC for further analysis. [Fig.1](#f1-sensors-04-00095){ref-type="fig"} shows the experimental set-up. The measurement process was as follows: dry air at a constant flow rate of 10 ml/sec was used as carrier gas. Data acquisition started from 80 s before the injection of the test gases sample into the air-flow. The sampling rate was set to two points per second and it took several minutes to complete the measurements. The surface temperature of the gas-sensing element was measured with an infrared thermometer (Keyence, IT2-01, Japan). Results and discussion ====================== Static responses of the SnO~2~-based gas sensor ----------------------------------------------- The static responses to 0.5 ppm butanone, acetone, ethanol, methanol and formaldehyde at 300 °C for the SnO~2~ sensor are reported in [Fig. 2](#f2-sensors-04-00095){ref-type="fig"}. As shown in the figure, one could clearly observe that the resistance of the sensing element changes obviously upon exposure to the organic gas, meanwhile the response time could be observed audio-visually by means of the tendency of the static curves. It was also necessary to point out that, however, in addition to the changes in resistance and response time, there was no other information about reaction processes, so it was difficult to analyses the sensing mechanism for butanone, acetone, ethanol and methanol. Except the static response to formaldehyde, the other responses were similar to each other. In fact, there are two types of different functional groups (carbonyl group and hydroxyl group). There is no way to distinguish between the carbonyl and hydroxyl group because of the shortcoming concerning their lack of sufficient phenomenon. In a word, during static measurements, only the resistance changes of the sensing element in initial and final state were observed, as for the other changes during reaction processes no other information was obtained. Dynamic nonlinear responses of SnO~2~-based gas sensor ------------------------------------------------------ [Fig. 3](#f3-sensors-04-00095){ref-type="fig"} reports the dynamic nonlinear responses to 0.5 ppm ethanol, methanol, formaldehyde and cyclohexanone of a single SnO~2~ gas sensor, respectively. Experimental conditions were as follows: applied potential 7 V, modulation frequency 20 mHz, rectangular mode. Firstly, as seen in [Fig. 3](#f3-sensors-04-00095){ref-type="fig"}, it was worth to note that there was sufficient reaction information in the dynamic nonlinear responses, so the sample gases could be easily identified by means of different nonlinear responses. In the four selected species, there are three types of different functional groups (hydroxyl group, aldehyde group and ring ketone); one could easily obtain the distinguishing and connection among the testing species by comparison the static responses. In the case of the dynamic responses to ethanol and methanol, because of their identical hydroxyl group, there is a similar tendency in the curves. Secondly, it was widely accepted that oxygen in air could be chemisorbed and decomposed as O~2~^-^, O^-^ and O^2-^. At a constant temperature, there existed an equilibrium state at the surface of the SnO~2~ sensing material: O~2~↔ O~2~^-^(ad) ↔ O^-^(ad) ↔ O^2-^ (ad). Semiconductor gas sensors monitor changes in the conductance during the interaction of a chemically sensitive material such as SnO~2~ with molecules to be detected in the gas phase; the reaction steps involved the low-temperature surface reactions and secondly, involved the high-temperature bulk reactions between point defects in the SnO~2~ crystal and oxygen (O~2~) in gas phase. The first step was adsorption and catalytic reactions at active sites (the latter involves intrinsic point defects such as oxygen vacancies and/or extrinsic point defects, like segregated metal atoms) and similar reactions at grain boundaries or at three-phase boundaries (e.g., at metallic contacts on surface metallic clusters). All of these reactions involved adsorbed negatively charged molecular (O^2-^) or atomic (O^-^) oxygen species as well as hydroxyl groups (OH) at different surface sites. During the static measurement processes, the adsorbed oxygen results in oxidation of testing gases on the surface and in a decrease of the chemisorbed oxygen concentration, inducing an increase of the conductance. As seen from [Fig. 3](#f3-sensors-04-00095){ref-type="fig"}, it was found that gas identification in a rectangular temperature-modulated mode was related to the different reaction kinetics of the interacting gases on the tin-oxide surface. It is clear that, by temperature modulation, it became possible to provide the surface oxygen species at constant temperatures at which, in equilibrium conditions, they would not exist. In this way, the reaction with the reducing and oxidizing gases was dramatically influenced, e.g. at lower temperatures and at higher temperatures the responses to sample gases exhibited their characteristic wave shape due to the reaction with different oxygen species. So, from the above mentioned, we suggested that the dynamic nonlinear responses are beneficial to analyse the sensing mechanism of the sample gases. Thirdly, it was also found that the operation temperature during the dynamic processes was lower than that of the static processes. For example, the static operation temperature using an applied potential of 7 V reaches 450 °C, whereas the dynamic temperature at operation frequency of 20 mHz was in the range of 300∼210 °C.Therefore, we suggested that a modulated temperature operation mode not only enhances the selectivity of a gas sensor but also reduces the overall power consumption. Effect of the modulation temperature at frequency of 20 mHz ----------------------------------------------------------- To optimize the selectivity of the temperature-modulated sensor, it was necessary to obtain a relationship between a given temperature and its conductance response in the presence of a specific gas. [Fig. 4](#f4-sensors-04-00095){ref-type="fig"} reports the effect of a given temperature on the responses of 0.5 ppm acetone at a constant frequency of 20 mHz in a rectangular mode. As seen in the figure, the different responses at different operation temperatures could be easily observed. This case noted that the sensor exhibited an enhanced selectivity to acetone with increasing temperature. It was suggested that acetone can be identified by means of a relatively complete response between 250 °C and 300 °C. Obviously, there were different surface reaction mechanisms between acetone and chemisorbed oxygen at different given operation temperatures. Effect of the duty ratio at an applied potential of 7 V ------------------------------------------------------- [Fig. 5](#f5-sensors-04-00095){ref-type="fig"} clearly shows the time-dependent change shape of the resistance of the sensor in the presence of butanone in air at different duty ratios by controlling an applied potential of 7 V in a rectangular mode. As seen in the figure, one can easily observe the effects of electrical outage and switch-on on the dynamic responses. In the left part of the figure, the lower half of the response was influenced evidently by the state of switch-on electrical current. With the lengthening of the switch-on time (electrical outage time in thermal cycle), the response was appeared gradually. Inversely, the upper half of the response was influenced obviously by the state of electrical outage; the more the time of the electrical outage, the more evident the upper half of the response. Therefore, the duty ratio was beneficial to investigate the sensing mechanism, and it was in agreement with the analysis in the dynamic nonlinear response. Effect of modulation waveform ----------------------------- In order to improve the detection, the modulation of the wave shape was carried out. The dynamic nonlinear responses to methanol are reported in [Fig. 6](#f6-sensors-04-00095){ref-type="fig"} using different modulation waveforms. Although the object was commentarial, one could see that the nonlinear responses were different to each other. These experimental data are in agreement with A. Ortega\'s report. A. Ortega and his co-workers have reported that CO and CH~4~ could be detected by using pulse and triangular heating waveforms \[[@b7-sensors-04-00095]\]. According to the analysis of the sections above, the change of the heating shape waveform had influence on the sensing behavior of the testing gas by means of the change of the sensing element\'s surface temperature. Conclusions =========== The study carried out over the dynamic and static measurements has allowed to demonstrate the advantage of the former measures in order to obtain the maximization of the information extracted from a single gas sensor. Experimental test showed that the dynamic measurement was beneficial to facilitate the feature extraction algorithm as the most significant output signal to separate the testing gases. Meanwhile, a variety of influencing factors such as modulation temperature, duty ratio, heating shape waveform (rectangular, saw-tooth, pulse, sinusoidal) were investigated. It was found that the characteristic optimum oxidation temperatures of acetone were 250∼300 °C at a frequency of 20 mHz. At a frequency of 20 mHz, the temperature contrast reached a maximum which was beneficial for detecting butanone. Experimental data also showed that different waveforms could improve the detection if the separation of certain gases is not clear. Finally, all data showed that temperature was the most essential influencing factor. This work was financially supported by the National Natural Science Foundation of China (project 60274061) and Anhui Province Natural Science Foundation (project 01041404), which are gratefully acknowledged. {#f1-sensors-04-00095} {#f2-sensors-04-00095} {#f3-sensors-04-00095} {#f4-sensors-04-00095} {#f5-sensors-04-00095} {#f6-sensors-04-00095}

ANNOUNCEMENT {#s1} ============ Characterization of *Mycobacterium orygis* as a subspecies of the Mycobacterium tuberculosis complex (MTBC) that can cause clinical features of tuberculosis in animals and humans was first published by Van Ingen et al. in 2012 ([@B1]). Several studies have since reported the isolation of *M. orygis* from captive wild animals ([@B2]) and from dairy cattle and captured monkeys ([@B3]). A putative case of human-to-cattle transmission of *M. orygis* in New Zealand was reported, the source of which was traced back to India based on epidemiological analysis ([@B4]). *M. orygis* is suggested to be endemic in Southeast Asia, including in India, Pakistan, and Nepal ([@B4], [@B5]). Here, we report the isolation of *M. orygis* from cattle from India. A postmortem examination performed on comparative intradermal test (CIT)-positive cattle from a farm in Chennai, India, revealed a macroscopic appearance consistent with severe tuberculosis of the lungs. Tissue samples from the lungs were homogenized, decontaminated with 5% sulfuric acid in phosphate-buffered saline (PBS), filtered with sterile muslin cloth ([@B6]), and inoculated onto Lowenstein-Jensen (LJ) slants and mycobacterial growth indicator tubes (MGIT). Positive mycobacterial growth on MGIT and LJ slants was confirmed by Ziehl-Neelsen (ZN) staining, and the MTBC was confirmed by immunochromatographic testing (ICT) ([@B7]). Mycobacterial colonies on LJ slants were suspended in Tris-EDTA (TE) buffer, and genomic DNA was isolated by the cetyltrimethylammonium bromide (CTAB)-NaCl method ([@B8]). Spoligotyping was performed as previously described ([@B9]), and the spoligotype pattern was compared against those in the SpolDB4 database ([@B10]). The genomic DNA was checked for quality by measuring the *A*~260/280~ ratio using the NanoDrop method and was quantified using Qubit. The sequencing library was prepared using a TruSeq Nano DNA LT library prep kit as per the manufacturer's protocol. The quality of the library was checked using an Agilent 2200 tape station. Whole-genome paired-end sequencing was carried out on an Illumina HiSeq 2500 instrument, generating 20,510,576 read pairs of 150 bp. Raw reads were filtered using Trimmomatic v0.36 (quality value, \>20; minimum length, \>60 bp) ([@B11]). Filtered reads were aligned to the reference genome of H37Rv (GenBank accession number [NC_000962](https://www.ncbi.nlm.nih.gov/nuccore/NC_000962)) using Burrows-Wheeler Aligner (BWA) v0.7.12 ([@B12]) with default parameters, and alignments were corrected using GATKv3.5 ([@B13]). Variants were called using SAMtools v1.3.1 ([@B14]) and bcftools v1.3.1 with default parameters. Variants with a quality of \>50, a mapping quality of \>30, and a depth of at least 5 with at least one read in either direction were filtered for analysis. RD-Analyzer v1.0 (default parameters) was used to detect regions of difference (RDs; regions of the MTBC genome whose presence or absence is lineage specific and that are widely used as classification markers) ([@B15]). Filtered reads were also assembled with SPAdes v3.11.0 with autocorrection ([@B16]). Comparison of the spoligotype profile with those in the SpolDB4 database classified the isolate as spoligotype 587 (ST587), which belongs to *M. orygis* ([@B1]). RD-Analyzer results confirmed the absence of RD7, RD8, RD9, and RD10 and the presence of RD1 and RD4 ([@B1]). The assembled genome (size, 4,293,394 bp) consisted of 107 contigs with an *N*~50~ value of 99,756 bp and a G+C content of 65.59%. Analysis of identified variants revealed a previously reported T-to-G mutation in the 38th codon of Rv2042c and a C-to-T mutation at the 321st codon in the gene *PPE55* ([@B3]). These genomic features, along with the spoligotyping results, confirmed the strain to be *M. orygis.* This genome sequence will contribute to our understanding of the genomic characteristics of this species and will facilitate further comparative genomic studies. Data availability. {#s1.1} ------------------ This whole-genome shotgun project has been submitted to NCBI under BioProject identifier [PRJNA545406](https://www.ncbi.nlm.nih.gov/bioproject/PRJNA545406). The raw reads and the assembled contigs have been submitted under accession numbers [SRR9157804](https://trace.ncbi.nlm.nih.gov/Traces/sra/?run=SRR9157804) and [VDER00000000](https://www.ncbi.nlm.nih.gov/nuccore/VDER00000000), respectively. The version of the assembled contigs described in this paper is the first version, VDER01000000. A.K.R. acknowledges the N-PDF (PDF/2016/003220) funding from DST-SERB. K.P. acknowledges the intramural funding from ICMR-NIRT. S.J.P. and N.K. acknowledge the funding from the Health Innovation Challenge Fund (grants WT098600 and HICF-T5-342), a parallel funding partnership between the Department of Health and Wellcome. The views expressed in this publication are those of the author(s) and not necessarily those of the Department of Health or of Wellcome. [^1]: **Citation** Refaya AK, Kumar N, Raj D, Veerasamy M, Balaji S, Shanmugam S, Rajendran A, Tripathy SP, Swaminathan S, Peacock SJ, Palaniyandi K. 2019. Whole-genome sequencing of a *Mycobacterium orygis* strain isolated from cattle in Chennai, India. Microbiol Resour Announc 8:e01080-19. <https://doi.org/10.1128/MRA.01080-19>.

Background {#Sec1} ========== Aortic pseudoaneurysms are rare, life-threatening sequelae of cardiac surgery \[[@CR1]\]. They can also occur as a result of infections, genetic disorders, or trauma \[[@CR2]--[@CR5]\]. They are uncommon in children. Several cases have been reported, most of which were secondary to bacterial endocarditis \[[@CR6]--[@CR8]\]. Our case highlights a rare postoperative complication that has not been reported previously in our region. Various management strategies have been suggested, including surgical and catheter-based interventions. Because of the age and size of our patient, and the infected nature of the lesion, we opted for surgical intervention. Timely diagnosis with effective surgical treatment in combination with antibiotic coverage can result in good survival without complications in such life-threatening situations. Case presentation {#Sec2} ================= Our patient is a 2-year-old Middle Eastern boy, weighing 12 kg below fifth centile, height 88 cm at the tenth centile, with bicuspid aortic valve (BAV) coarctation of the aorta, a large inlet ventricular septal defect (VSD) and subaortic membrane with mild left ventricular outflow tract obstruction. He was the product of a consanguineous marriage and there is no history of congenital heart disease in the family. He underwent aortic arch repair and pulmonary artery (PA) banding at 10 days of life. After that he was on regular follow up at our center. He was started on antifailure medication and was referred for complete repair of VSD closure and PA debanding as he was not gaining weight. He was developmentally normal and no other noncardiac comorbidity was identified. The surgery was performed successfully and postoperative transesophageal echocardiography (TEE) showed no residual lesions. On day 5 postoperatively, he developed lethargy, fever, and sternal wound discharge. He had a temperature of 39 °C and tachycardia with heart rate of 140 beats per minute. He showed leukocytosis with a total white blood cell count of 16 × 109/L and raised acute phase reactants. Our patient had methicillin-resistant *Staphylococcus aureus* (MRSA) growth confirmed by two blood and wound cultures which showed sensitivity to vancomycin and meropenem. He was treated with intravenously administered antibiotics (meropenem and vancomycin). The rest of laboratory parameters including renal, liver, and urine analyses were normal. Repeat blood cultures at 3 weeks of treatment were still positive for MRSA. Rifampicin was added. His follow-up echocardiography (Fig. [1](#Fig1){ref-type="fig"}) at 3-weeks postoperatively, while still on antibiotics, revealed a large pseudoaneurysm arising from the anterior wall of his ascending aorta (AsAo). "Smoky" blood flow into the aneurysm cavity was seen, but no thrombi or vegetation was detected. These findings were confirmed on computed tomography (CT) angiogram, which showed a pseudoaneurysm arising from the anterior wall of the AsAo. The lesion was 4.5 × 4.0 × 3.5 cm in size and had a 10--11 mm neck. It extended to the right and superiorly, causing rightward deviation of his superior vena cava and innominate vein (Figs. [2](#Fig2){ref-type="fig"} and [3](#Fig3){ref-type="fig"}). He was immediately sent to the operating theater. Surface cooling was started early while exposing his right femoral vessels. A 3.5 mm Gore-Tex tube was anastomosed to the side of his proximal right common femoral artery using a 7/0 polypropylene stitch. After heparin was given, a 10 Fr arterial cannula was inserted into the Gore-Tex tube, and was fixed. When his core temperature reached 28 °C, the lower part of his sternum was opened and his right atrium was cannulated with a 24 Fr metal-tip cannula. A cardiopulmonary bypass was started with a full flow of 120 ml/kg, and cooling was continued on bypass until a core temperature of 20 °C was reached. His head was additionally cooled with ice packs. The dissection was then continued. When the upper part of his sternum was reached, the sac of the false aneurysm was opened. His heart was fibrillating at this point. Very low flow was used to improve exposure. A 1 × 1 cm hole at the site of the previous aortic cannulation was found. The edges of the hole and the surrounding tissue appeared infected and friable. Cultures were taken. A cross-clamp was then applied to his AsAo distal to the hole. No cardioplegia was used. A bovine pericardial patch was used to repair the ascending aortic wall. Because of tissue friability, the patch was sutured away from the edges using 7/0 polypropylene mattress continuous stitches. This was further reinforced by an outer 6/0 polypropylene stitch. BioGlue was applied over the patch. After rewarming, the cardiopulmonary bypass was weaned off easily. The femoral arterial line was removed and the Gore-Tex tube was cut short and clipped. The cross-clamp time was 28 minutes, and bypass time was 111 minutes.Fig. 1Transthoracic echocardiography, suprasternal view, revealing a huge pseudoaneurysm arising from the anterior wall of the ascending aorta. *AAO* ascending aortaFig. 2Computed tomography angiogram (colored) which shows a pseudoaneurysm arising from the anterior wall of the ascending aorta (arrow). Dimensions were 4.5 × 4.0 × 3.5 cm with a 10--11 mm neck. The aneurysm extended to the right and superiorly causing rightward deviation of the superior vena cava and innominate vein. *AAo* ascending aorta, *LA* left atrium, *RPA* right pulmonary arteryFig. 3Computed tomography angiogram which shows a pseudoaneurysm arising from the anterior wall of the ascending aorta. Dimensions were 4.5 × 4.0 × 3.5 cm with a 10--11 mm neck. The aneurysm extended to the right and superiorly causing rightward deviation of the superior vena cava and innominate vein *AAO* ascending aorta, *LA* left atrium, *RPA* right pulmonary artery Our patient had a smooth postoperative course with no signs of infection and no neurological or ischemic complications. Antibiotics were continued for 6 weeks and he was discharged with negative blood cultures and normal inflammatory markers. We followed our patient 6 weekly for the first 3 months, then every 6 months, then annually in clinic and repeated a cardiac CT scan (Fig. [4](#Fig4){ref-type="fig"}) at 2 months following surgery which showed no aneurysm. At 6-month out-patient follow up he had an unremarkable clinical examination and an echocardiogram showed no signs of aneurysm and had normal function. He gained weight to 15 kg.Fig. 4Computed tomography angiogram after surgery showed no aneurysm (arrow). *AAo* ascending aorta, *LA* left atrium, *RPA* right pulmonary artery Discussion {#Sec3} ========== Acquired aneurysms of the AsAo, although rarely seen, are most commonly found at cannulation sites and suture lines, particularly after repair of coarctation of the aorta \[[@CR9]\]. As reported these aneurysms are diagnosed with a high index of suspicion as symptoms are usually absent or nonspecific. They carry a high risk of mortality if diagnosed late and surgery is performed after the rupture \[[@CR6]\]. In the past, infected aneurysms were seen mostly as a result of infected endocarditis and repaired valvular diseases. These aneurysms can occur as direct invasion of aortic intima by circulating bacteria or through lymphatics in the presence of infection and mediastinitis. An added risk is therapeutic intervention like aortic cannulation \[[@CR6]\]. Similarly, our patient showed signs of infection on postoperative day 5 and his echocardiogram revealed a large aneurysm at the site of his aortic cannulation which was confirmed by CT angiogram. Our patient had MRSA growth confirmed by two blood and wound cultures, which along with other streptococcal species is one of the common organism isolated after mycotic aneurysm \[[@CR10]\]. It is extremely important to promptly diagnose these cases with a high index of suspicion especially after coarctation repair or history of aortic cannulation like in our case. An echocardiogram is a noninvasive, immediately available screening tool; cardiac CT angiogram is more sensitive. In our patient, with a high index of suspicion due to wound infection, an aortic cannulation and echocardiogram were done immediately and, later, cardiac CT was done to confirm the diagnosis. Both immediate and long-term types of pseudoaneurysms have been reported \[[@CR7]\]. Various treatments of pseudoaneurysm are advocated in the literature. There is no well-defined treatment for this lesion in the pediatric population. Both percutaneous and surgical repairs have been advocated in the literature. However, percutaneous closure has been successfully reported in the adult population only \[[@CR11], [@CR12]\]. Surgical repair still remains the gold standard, especially because of the infected nature of the lesion that precludes the use of percutaneous devices. The main challenge for the surgeon is to plan an approach to the lesion without risking hemodynamic compromise, neurological insults that might be caused by air embolism, and hypotension secondary to bleeding or heart manipulation during dissection. Adhesions from previous cardiac surgery increase these risks. Various surgical approaches have been employed, such as femoral or neck cannulation to establish cardiopulmonary bypass, deep hypothermic circulatory arrest, and antegrade cerebral perfusion \[[@CR4]\]. Barth *et al*. in their case report of a similar case used the left common iliac artery for repair of a pseudoaneurysm of the AsAo in a 16-month-old girl \[[@CR7]\]. Our patient had a tear at the site of aortic cannulation that appeared to be caused by bacterial infection with MRSA. Our approach included surface cooling, femoral cannulation mitigated by the use of a Gore-Tex graft for arterial cannulation, right atrial cannulation, and use of deep hypothermia. We found this method to be a safe and reproducible. A similar case was reported by Miyaji *et al*., where the common iliac artery was cannulated. Their case was postoperative mycotic pseudoaneurysm in the AsAo secondary to mediastinitis caused by MRSA after a modified Fontan procedure \[[@CR13]\]. Conclusions {#Sec4} =========== In conclusion, aortic pseudoaneurysms are rare but recognized complications of cardiac surgery that can be life-threatening. Prompt diagnosis and management, as in our case, can result in successful outcomes. Not applicable. Funding {#FPar1} ======= This study did not receive any grants. Availability of data and materials {#FPar2} ================================== All materials are available for use by any other researcher. KH and LF contributed the surgical part of the case report, MA in the initial drafting of the case report, AO contributed in images selection, SM and KN in the case report editing and literature review. All authors read and approved the final manuscript. Ethics approval and consent to participate {#FPar3} ========================================== Not applicable. Consent for publication {#FPar4} ======================= Written informed consent was obtained from the patient's legal guardian for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal. 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.

Introduction {#Sec1} ============ Use of non-invasive ventilation (NIV) in critical and intensive care has expanded greatly in recent decades in response to evidence of its benefits as a means of reducing dependence on invasive (i.e. with tracheal intubation) ventilation and associated complications, and for the management of acute respiratory failure. Adoption and application of NIV are nevertheless still in a phase of evolution \[[@CR1]\]. We examine this subject with specific reference to the use and impact of sedation during NIV. This essay examines a non-exhaustive series of questions about NIV and the possible role of sedation as a co-intervention to NIV. The answers - to the extent that answers are available - emerge from sources that range from evidence-based medicine to expert opinion. This range is itself indicative of the incomplete nature of our current understanding of these matters. Review {#Sec2} ====== Is sedation universal or mandatory during NIV? {#Sec3} ---------------------------------------------- A non-exhaustive inspection of recent reports invites the conclusion that we as physicians are either indifferent to or unconcerned about the general topic of sedation during NIV. The authors of the most recent Cochrane review on the place of NIV as a weaning strategy for intubated adults with respiratory failure reported that only one of the studies they identified used a standardized sedation protocol before or after initiation of NIV (Table [1](#Tab1){ref-type="table"}) and argued that the role of sedation as a co-intervention requires specific investigation in future trials \[[@CR2]--[@CR8]\]. Separately, Scala has identified only eight small clinical studies of sedation in NIV, in which a total of 183 patients were assigned to seven different sedative drug regimens, the effects of which were assessed by multiple different and non-congruent outcome parameters (Tables [2](#Tab2){ref-type="table"} and [3](#Tab3){ref-type="table"}) \[[@CR9]--[@CR17]\].Table 1**Summary of studies making any report of sedation use in the context of NIV in adults with critical illness (compiled from Burns et al.** \[[@CR2]\] **)**StudyNo. of patientsPatient characteristicsExperimental NIV strategySedation statusHill et al. 2000 \[[@CR4]\]21ARFVPAP in ST-A modeSedation protocol reportedly used. Study published as abstract onlyNava et al. 1998 \[[@CR5]\]50Exacerbation of COPD; mechanical ventilation for at least 36--48 hNon-invasive pressure support on conventional ventilator delivered with face maskSedation reportedly used during NIV but apparently not protocolized or definedPrasad et al. 2009 \[[@CR6]\]30COPD; AHRFBilevel NIV (pressure mode) delivered by full face maskPatients received neuromuscular blocking drugs and sedatives in immediately preceding phase of invasive ventilation. Use of sedation during NIV not clearRabie Agmy et al. 2004 \[[@CR7]\]37Exacerbation of COPDProportional-assist NIV in timed mode, delivered by face or nasal maskPatients received neuromuscular blocking drugs and sedatives in immediately preceding phase of invasive ventilation. Use of sedation during NIV not clearVaschetto et al. 2012 \[[@CR8]\]20Hypoxaemic respiratory failure; invasive mechanical ventilation for at least 48 hHelmet NIVSedation reportedly used during NIV but apparently not protocolized or defined. Rates of continuous sedation during NIV reported to be similar in both groups (*a priori* study outcome)In all the studies, the control strategy used was invasive pressure support.ARF, acute respiratory failure; AHRF, acute hypercapnic respiratory failure; ST, spontaneous/timed; VPAP, variable positive airway pressure.Table 2**Studies of sedation in NIV (compiled from Scala** \[[@CR9]\] **)**StudyNo. of patientsIndicationNIV interfaceType of sedativeInitiation of sedationRocker et al. \[[@CR10]\]10ARFFFMMorphineAt start of NIVConstantin et al. \[[@CR11]\]13ARF (n = 10); AHRF (n = 3)FFMRemifentanil, midazolam\*Poor NIV acceptanceRocco et al. \[[@CR12]\]36ARFFFM, helmetRemifentanilPoor NIV acceptanceAkada et al. \[[@CR13]\]10ARFFFMDexmedetomidine^†^Poor NIV acceptanceTakasaki et al. \[[@CR14]\]2SAAFFMDexmedetomidinePoor NIV acceptanceClouzeau et al. \[[@CR15]\]10ARF (n = 7); AHRF (n = 3)FFMPropofolPoor NIV acceptanceSenoglu et al. \[[@CR16]\]40COPDFFMDexmedetomidine (n = 20); midazolam (n = 20)At start of NIVHuang et al. \[[@CR17]\]62ACPOFFM, helmetDexmedetomidine (n = 33); midazolam (n = 29)Poor NIV acceptanceThe last two rows identify randomized controlled trials; other trials were reported not to have been controlled. See also Table [3](#Tab3){ref-type="table"} of this review.ACPO, acute cardiogenic pulmonary oedema; ARF, acute respiratory failure; AHRF, acute hypercapnic respiratory failure; FFM, full face mask; SAA, severe asthma attack.\*Combined with propofol in three cases. ^†^Combined with morphine in one case and with propofol in one case.Table 3**Reported dosages of sedatives administered in studies of sedation in NIV (compiled from Scala** \[[@CR9]\] **)**DrugNo. of patientsDosageSedation target rangeDexmedetomidine411 μg/kg (bolus); 0.2-0.7 μg/kg/h (infusion)RSS 2-3; RASS 2-4; BIS \>85Midazolam410.05 mg/kg (bolus); 0.05-0.1 mg/kg/h (infusion)RSS 2-3; RASS 2-4; BIS \>85Remifentanil380.025-0.1 μg/kg/min (infusion)RSS 2-3Propofol430.4 μg/mL (target serum concentration; step-down to 0.2 μg/mL)OAAS/S 3-4The information in this Table is derived from references 10-17. See also Table [2](#Tab2){ref-type="table"} of this review.BIS, bispectral index; OAAS/S, Observer's Assessment of Alertness/Sedation Scale, RASS, Richmond Agitation and Sedation Scale; RSS, Ramsey Sedation Scale. The drift of these observations appears to be that either sedation has not been perceived as a major problem or opportunity within the wider context of NIV use or that it has not been studied systematically. The first of these positions may have led to the second; alternatively, the lack of studies may reflect complaisance (if not complacency) about this aspect of NIV technique and procedure. In any event, both these positions suggest a lack of attention to the possibilities in this area of NIV. Narrowly considered, therefore, the answer to our own question must be "We do not know". However, the findings of the only survey of practice that we have identified make it clear that in reality the answer is "No", as only ≈ 25% of NIV patients received sedation (rising to ≈ 40% in critical care) \[[@CR18]\]. Whether or not that percentage represents an appropriate level of sedation use among recipients of NIV is just one of many questions to which there is at present no definitive answer. Inspection of Table [2](#Tab2){ref-type="table"} suggests that, in those studies where sedation during NIV was explicitly addressed, the main reason for sedation was "poor NIV acceptance". One general aspect of poor NIV acceptance could be related to the patient-device interface, i.e. the type of mask that is used and also the assisted ventilation pattern. The choice of NIV interface may influence the need for sedation. While it can be argued that "the clinical efficacy of different masks is on average very similar" \[[@CR19]\], our experience is aligned with reports that patient acceptance is greatest with the least constricting interfaces -such as the helmet - and declines as more intrusive forms of mask are used \[[@CR20]\]. As for the breathing modes, bilevel positive airway pressure often produces a need for anxiolysis or sedation whereas spontaneous breathing patterns such as continuous positive airways pressure seldom require such interventions. In summary, sedation is not mandatory for NIV but it may help in specific situations. There are at present no explicit guiding principles or simple formulae to identify those situations. Does NIV affect sedation goals? {#Sec4} ------------------------------- The goals of sedation for a cooperative patient in the intensive care unit (ICU) are to provide analgesia and comfort, preserve day/night cycles (including natural sleep) and avoid nuisances such as ambient light and noise. Additional goals include haemodynamic stability, preservation of metabolic homeostasis, muscular relaxation, preservation of diaphragmatic function and attenuation of the stress/immune response, as well as considerations such as programmed withdrawal from sedation: they should be no different during NIV. Nevertheless, if we regard NIV as a stage in the progression from intermittent mandatory ventilation to spontaneous breathing then there must be a parallel expectation of a progressive reduction in use of sedation. Two other points need to be discussed for sedation during NIV. Firstly, the avoidance of the respiratory depressant effects of different sedatives; secondly, untoward effects of sedative drugs on the upper airway, a topic that has received new impetus with increased awareness of obstructive sleep apnoea \[[@CR21]--[@CR24]\]. Therefore the answer to this question is "Yes", both quantitatively (i.e. less 'deep' sedation) and qualitatively in that sedation during NIV must be performed with no/minimal respiratory depression and no/minimal impairment of the upper airway. Is sedation *per se*a factor contributing to the success or failure of NIV? {#Sec5} --------------------------------------------------------------------------- Patient acceptance and compliance are essential to the success of NIV \[[@CR13]\]. Achieving patient acceptance and compliance is a multifaceted exercise, with staff proficiency and competence being one major influence \[[@CR25]\]. However, acceptance/compliance also relate directly to sedation since neither can be expected from an insensate patient; nor are they likely to be forthcoming if the patient is anxious, agitated or disoriented. The necessity of a sedation regimen that brings the patient to a state of calm, alert cooperation is clearly implied by these considerations. Any decision to resort to sedation must be taken as the last stage in a careful evaluation of the causes of actual or pending failure, as outlined in Figure [1](#Fig1){ref-type="fig"}. It must also take account of the fact that the likely success of NIV in hypoxaemic respiratory failure varies considerably according to the presenting condition \[[@CR26]\] and that there is no robust evidence that sedation will materially affect situations where the response rate to NIV is intrinsically poor. Indeed, adding sedation in these situations may be disadvantageous by obscuring a failure of NIV due to underlying pathology and thus delaying a necessary intubation. Similarly, sedation does not obviate any of the contraindications to NIV \[[@CR26]\].Figure 1**Clinical reasoning pathway for the use of sedation in NIV.** Nava and Ceriana \[[@CR25]\] have partitioned NIV failure into immediate (\< 1 h after commencement), early (1--48 h) and late (\> 48 h) and identified predictors of failure for each time segment. Factors implicated in immediate NIV failure include "intolerance, agitation, and patient--ventilator asynchrony", for which "judicious sedation" is recommended but not characterized in detail. Sleep disruption was identified as a factor in late NIV failure \[[@CR25]\]. The subject of sleep and sleep disruption in ICU patients has attracted much research in recent years (see for example Cabello et al. \[[@CR27]\] and Roche Campo et al. \[[@CR28]\]) and one instinctively inclines to the belief that more normal patterns and quality of sleep are part of the restoration of health. Whether this is indeed the case is harder to demonstrate than might be expected but investigations of the differential effects of sedatives on electrophysiological dimensions of sleep provide food for thought \[[@CR29]\]. This is, of course, an area where environmental alterations (i.e. minimizing noise and disturbance) may also have an important influence and such measures should always precede any use of sedatives. In mechanically ventilated patients, ventilator asynchrony may adversely affect sleep \[[@CR30]\]: the extent to which this disruptive interaction may work in the opposite direction is not clear, but there is evidence that, in NIV patients, sleep is associated with more asynchrony and blood-oxygen desaturations than the awake state \[[@CR31]\]. These data were obtained from a small cohort of patients receiving long-term NIV at home and probably do not fully illuminate the situation of ICU patients receiving both NIV and sedatives. Choice/type of ventilator is also an important influence on asynchrony \[[@CR32], [@CR33]\] and, at least in some cases, is likely to be more significant than use or non-use of sedatives. Similarly, ventilators with specific leak compensation modes and the ability to generate flows ≥ 10 L/min provide better patient compliance than ICU ventilators with little leak adaptive capacity. The answer to the question is thus that we do not know for certain but that it is unlikely that sedation can 'rescue' poorly configured or inappropriate NIV. However, we hypothesize that adequate sedation can improve patient comfort in specific situations (see below). What types of patients/specific situations might benefit from sedation during NIV? {#Sec6} ---------------------------------------------------------------------------------- Reference to English language guidelines for NIV \[[@CR1]\] reveals that the range of indications for which there is compelling or even strongly persuasive (Grade 2B or better) evidence for benefit of NIV is small and may be summarized as follows. *Acute respiratory failure in the forms of*: Acute-on-chronic exacerbation of chronic obstructive pulmonary disease (COPD) with acidotic and hypercapnoeic featuresRespiratory failure secondary to cardiogenic pulmonary oedema not arising from shock or acute coronary syndromeAcute respiratory failure in immunocompromised patients. *As an adjunct to extubation (but only in expert centres) for*: Patients who have COPDPatients considered to be at high risk of recurrent respiratory failure. We would expect most candidates for sedation during NIV to come from these categories, and hence to have certain common presenting features. As noted above, discomfort, anxiety, agitation, pain, dyspnoea and the disappointed expectations of the patient are central in many cases to failure of NIV and hence also to the decision to use sedation in NIV. Delirium may also be a consideration. Use of sedation in NIV is, in this context, firmly within the sphere of the clinical practice guidelines for the management of pain, agitation, and delirium \[[@CR34]\] and many of the provisions of those guidelines are relevant, most especially the obligation to identify and correct (by non-pharmacological means if feasible) the causes of pain, agitation, etc. Agitation and delirium {#Sec7} ---------------------- A first category to consider is patients who are already agitated before tracheal extubation. A systematic investigation of the causes of anxiety should precede any prescription of a sedative drug. If all otherwise correctable causes of anxiety are eliminated, and anxiety *sine materia* or anxiety due to a decrease/change in sedative regimen is diagnosed, a case may be made (albeit on the basis of very slender clinical experience) for dexmedetomidine as an aid to extubation \[[@CR35]\]. The mechanisms underlying this benefit are not clear and require further, systematic investigation. The impact of delirium in patients receiving NIV is urgently in need of attention. The most substantial report we have found identified a high prevalence of delirium in NIV patients (≈37%) and linked that to a marked increase in risk of NIV failure. However, the data on which these findings were based were described as "scarce and of low quality" \[[@CR36]\]. The role of sedation in the promotion or prevention of ICU delirium has attracted much comment in recent years but firm conclusions are still hard to come by. For the moment, we are unable to go beyond the pain, agitation, and delirium guidelines, which offer weak (Grade 2B) endorsement for dexmedetomidine in delirium management. "Prevention*"* of delirium rests substantially on non-pharmacological methods, particularly early mobilization. Where sedation is used at all, an emphasis on early mobilization implies a sedative regimen that facilitates patient participation. (See Nydahl et al. \[[@CR37]\] for a recent perspective on this matter). Dyspnoea {#Sec8} -------- A second category of patients could be those who are dyspnoeic and anxious, dyspnoea being associated with delay of extubation \[[@CR38]\]. The neuro(patho)physiology and clinical aspects of dyspnoea have been examined in detail in a report of the American Thoracic Society that stratifies dyspnoea according to the quality of the dyspnoea experience, the stimuli that evokes it and the afferent neuronal pathways that mediates it \[[@CR39]\]. A full discussion of this matter is beyond the scope of this review but it is worth noting that there is an affective component of dyspnoea that may be differentiated from the sensory dimension and might be amenable to independent manipulation \[[@CR40]--[@CR43]\]. This signals the importance of identifying and appraising the anxiety component of dyspnoea. Self-evidently, the cooperation of the patient is needed for this and any existing sedation regimen must be adapted to that need. The affective dimension of dyspnoea may be investigated using either single-item ratings of severity of distress or unpleasantness or multi-item scales of emotional responses such as anxiety \[[@CR39], [@CR44], [@CR45]\]. However, there are a vast array of dyspnoea rating scales, which address different aspects of the condition. This is not a reason not to measure dyspnoea but it is essential to specify which scale is used for the purpose and to recognize that the nature of any intervention on dyspnoea is likely to be determined (or at least influenced) by the aspects of dyspnoea privileged by the chosen scale. Hence, sedation could be of benefit in situations where NIV is clearly indicated and where careful evaluation identifies anxiety, dyspnoea with a high affective dimension or delirium as barriers to its successful implementation. Are there evidence-based reasons to prefer specific sedative drugs during NIV? {#Sec9} ------------------------------------------------------------------------------ There are no robust data to favour any one drug, class of drugs or protocol over all others. Some of the criteria that may shape the selection of sedatives for NIV are summarized in Table [4](#Tab4){ref-type="table"}. No drug or class fully satisfies all these criteria and the final decision rests for the moment on a sequence of clinical reasoning.Table 4**Properties of sedative drug classes relevant to delivery of sedation in NIV**SedativeHaemodynamic stabilityAnalgesiaAmnesiaAnxiolysisPVDAvoidance of PONVPromotion of natural sleepSuitability for use after extubationDelirium avoidanceTotal**Propofol**22222422120**Midazolam**32422221119**Opioids**44121112120**Dexmedetomidine**32244244328**Ketamine**43214114121Larger numbers indicate a more satisfactory impact on the nominated property. This is primarily a qualitative and relative assessment of the features and benefits of different drugs and drug classes, framed in general terms. Hence, the individual category scores and in particular scores shown in the 'Total' column are crude summaries that should not be over-interpreted and which do not necessarily reflect the net merits or demerits of particular agents in the circumstances of a particular patient.PONV, postoperative nausea and vomiting; PVD, preservation of ventilatory drive. We fully endorse the general presumption against benzodiazepines expressed in the pain, agitation, and delirium guidelines. It is, therefore, a matter of concern that the data of Devlin et al. suggest that benzodiazepines retain a hold in NIV \[[@CR18]\]: however, those data are 7 years old and may not represent current habits. Fresh research into prescribing patterns (and the reasoning underpinning them) is desirable: study of the effects of switching sedatives would also be illuminating. Patterns of sedation when NIV is delivered on standard wards are undocumented as far as we know and this is another area that deserves more attention. Given the pathophysiology of NIV failure, at least three aspects could be influenced by the choice of sedative drugs: the patency of the upper airway, respiratory depression and the affective dimension of dyspnoea. From a pharmacological point of view, dexmedetomidine appears to offer the range of qualities best configured to address these concerns (see for example Hsu et al. \[[@CR46]\]). However, the evidence from controlled trials is not sufficient to give this conclusion the force of a guideline and additional considerations (see Table [4](#Tab4){ref-type="table"}) may shape the final selection of drugs for individual patients. (See also Ho et al. \[[@CR47]\] for a cautionary case report). Effects on airways patency and the timing and drive of respiration have been examined earlier in this review and make a case against benzodiazepines (and perhaps other γ-aminobutyric acid-ergic agents) and possibly in favour of dexmedetomidine. Opioids and benzodiazepines decrease upper airway diameter and are probably deleterious during NIV. Propofol has also been shown to increase the collapsibility of the upper airway in a dose/concentration-dependent manner \[[@CR23]\]. Dexmedetomidine has no direct effects on the patency of the upper airways. When used as adjunct therapy it may reduce requirements for opioids (or other sedatives) and so reduce the likelihood of opioid-induced compromise of the upper airways. Among the sedative drugs, dexmedetomidine has the lowest risk of depression of the respiratory centres. Ketamine does not cause respiratory depression at doses given for analgesia or procedural sedation \[[@CR48]\]. Furthermore, it decreases airway resistance, improves dynamic compliance and preserves functional residual capacity, minute ventilation and tidal volume, while retaining protective pharyngeal and laryngeal reflexes \[[@CR49]\]. Ketamine can produce hypersalivation and emergence reactions \[[@CR49]\]. Because of its effects on the sympathetic nervous system, ketamine should not be used in decompensated heart failure (typically cardiogenic pulmonary oedema in the context of NIV). There is a relatively abundant literature concerning the use of ketamine for procedural sedation \[[@CR50]\], but experience for sedation during NIV is practically non-existent. The answer to this question, therefore, is that sedative drugs used during NIV should have properties that further the goals of sedation during NIV. From a pharmacological point of view, benzodiazepines should be avoided: outcome studies in mechanically ventilated patients are consistent with that view but there are no similar trials in NIV to support definitive guidance \[[@CR51]\]. (Other recent work provides a timely reminder that without a robust experimental structure it is difficult to make meaningful comparisons between sedatives \[[@CR52]\]). Dexmedetomidine and ketamine seem to have the most suitable overall pharmacological profiles. Propofol and opioids (such as remifentanil) are in an intermediate position. Conclusions {#Sec10} =========== Much of the use of sedation in NIV appears to be empirical and is perhaps unstructured. Above all, it appears to be under-researched: we may be doing better than we realize or we may be doing less well than we could. With the continuing growth of NIV as a clinical resource there is a need for sedation practice to be put on a more systematic footing. To that end we propose a three-point plan.Repeat the international survey of 2007 \[[@CR18]\] to ascertain how, if at all, patterns of sedation use in NIV have changed in the intervening years and, if possible, identify the drivers of change.Review best practice frameworks in NIV to ensure that guidance includes indications for sedation use, standardized sedation protocols and clearly expressed criteria and conditions for terminating NIV.Develop at least one (preferably more) prospective multicentre trial on the effects of sedation in NIV. Such a trial needs to enrol substantially more patients than have hitherto been recruited to such studies and needs to incorporate the principles identified in point (2) and a set of unambiguous, informative and reliable endpoints, including a predefined and generally accepted definition of NIV failure. The work of Huang et al. \[[@CR17]\] provides a useful, though not exhaustive, template in this respect. In addition to standardization of sedation protocols, such a trial needs to apply consistent and predefined NIV modalities. Given the need for substantial numbers of patients, we anticipate that such a study would, in the first instance, be confined to patients with acute exacerbation of COPD, as this is the only category likely to yield sufficient patients within an acceptable and practicable period of time. An alternative or additional possibility would be a prospective trial of NIV to prevent acute respiratory failure in patients recovering from cardiac, abdominal or possibly thoracic surgery \[[@CR53], [@CR54]\]. Given that these patients would be emerging from surgical anaesthesia, there might be a larger role for sedation and that would facilitate rapid recruitment. **Competing interests** DL reports honoraria from Orion Pharma for guest lectures at international symposia in anaesthesia and intensive care and editorship of *intensetimes*, an electronic journal supported by Orion Pharma. GC reports honoraria for lectures from Orion Pharma (Italy), Covidien and Sylcomed. The Catholic University of Rome has received an institutional research grant from Orion Pharma. JM reports honoraria from Orion Pharma as a guest lecturer at international symposia in intensive care. AF reports travel and educational grants from Orion Pharma, CSL Behring and Pulsion Medical Systems and research grants from Pulsion Medical Systems and Köhler Chemie. RA has been a paid consultant for Orion Corporation (Espoo, Finland) and Abbott Laboratories (Abbott Park, IL, USA), the original co-developers of dexmedetomidine, and also for Hospira (Lake Forest, IL, USA). Hospira has a license agreement with Orion Corporation concerning dexmedetomidine (Precedex). He is also one of the three original patent holders of "Use of dexmedetomidine for sedative effect in patients in an intensive care unit" (WO/1999/049854). PT reports honoraria for lectures at national and international symposia from Orion Pharma, Abbvie, Baxter, Bard, B. Braun and Ratiopharm. **Authors' contributions** All authors contributed equally to the discussions that gave rise to this article. All authors contributed equally to the development of the final manuscript and approved the final manuscript. The authors thank Alexander Scott (Leeds, UK), Dorte Bulow Keld (Aarhus, Denmark), Alberto Sandiumenge (Tarragona, Spain), Ilona Bubek (Budapest, Hungary) and Per-Olof Joachimson (Uppsala, Sweden) for their contributions to the roundtable debate that provided the intellectual framework of this article. The roundtable was made possible through the financial support of Orion Pharma, Espoo, Finland. Editorial preparation of this manuscript was assisted by Hughes associates, Oxford, UK, on behalf of Orion Pharma.

I[NTRODUCTION]{.smallcaps} {#sec1-1} ========================== Breast carcinoma is one among the most common cancers occurring globally. In India, breast cancer is one among the top three cancers and the incidence of breast cancer in Kolar district is around 6.4%.\[[@ref1]\] Variations in nuclear structure are the morphologic hallmark of cancer diagnosis. There is a gradual shift in the nuclear parameters as the disease progresses from benign to malignant. Nuclear size, shape, chromatin pattern, and nucleoli size and a number have all been reported to change in breast cancer.\[[@ref2]\] These nuclear morphometric features have been shown to predict the prognosis of the breast cancer patients.\[[@ref3]\] Fine needle aspiration cytology (FNAC) has been routinely employed as a screening test for the breast cancer along with mammography and the clinical examination. However, cytological diagnosis is based on the subjective evaluation of nuclear features and thus, may raise a difficulty in establishing the precise diagnosis pre-operatively.\[[@ref4]\] There have been studies on computerized nuclear morphometric analysis of benign and malignant breast aspirates, and these may be supportive in diagnostic decisions.\[[@ref4]\] Studies on nuclear morphometric analysis of breast aspirates in South- Indian population are limited. Hence we have undertaken this study with an aim to compare the nuclear morphometric parameters of benign and the malignant breast aspirates. M[ATERIALS AND]{.smallcaps} M[ETHODS]{.smallcaps} {#sec1-2} ================================================= This was a retrospective study. We collected sixty fine needle aspiration samples from the archives of our department. Cytology was used to categorize aspirates from the breast lumps into 4 groups. Group I-fibroadenomas (10 cases), Group II-fibrocystic disease (10 cases), Group III-hyperplasia (10 cases), and Group IV-carcinoma (30 cases). Only those cases which had confirmed histopathological correlation, were included in the study. We used a microscope with an × 2.5 ocular and an × 40 objective to visually select a field for analysis. A 640 × 400 pixel digital image of the field was produced by a camera on the microscope and frame grabber card in a PC. Around 50 nuclei/case were analyzed using the Image J 1.44C morphometric software for image processing, and analysis (JAVA) developed by the National Institute of Health, USA. The following nuclear features were analyzed: Radius computed by averaging the length of radial line segments from the center of the nuclear mass to each of the points of the nuclear border.Nuclear area was the area within the outlined nuclear perimeter.Perimeter was measured as the distance around the nuclear border.Diameter was the diameter of the circle with the same area as the outlined nucleus.Compactness of the cell nuclei calculated using the formula: Perimeter^2^ /area.Concave points counted the number of points on the nuclear border that lie on an indentation.\[[@ref5]\] The computer calculated the mean, standard deviation, and range for all the nuclear features. Inclusion criteria: Only ductal carcinomas were considered for the study.Exclusion criteria: Lobular, medullary, and metaplastic carcinomas were excluded.Ethical clearance was obtained by the Institutional Ethics Committee. Statistical analysis {#sec2-1} -------------------- The results obtained by the computerized cytomorphometry were compared between the four groups. Data were analyzed to evaluate the most distinctive morphometric features of all the features available. The nuclear parameters between all the 4 groups were compared using Analysis of variance (ANOVA) and between the groups using a *post hoc* test i.e., Bonferroni Multiple Comparisons Test. Statistical analysis was performed using the statistical software Graph Pad Instat. A *P* \< 0.05 was considered as statistically significant. R[ESULTS]{.smallcaps} {#sec1-3} ===================== Nuclear morphometric analysis {#sec2-2} ----------------------------- Our sample size was 60, which was categorized in to 4 groups: Group I-fibroadenomas (10 cases), Group II-fibrocystic disease (10 cases), Group III-hyperplasia (10 cases), and Group IV-carcinoma (30 cases). Cytological features {#sec2-3} -------------------- ### Fibroadenoma {#sec3-1} Benign appearing ductal epithelial cells in sheets, antler horn pattern or honeycomb pattern. Background shows bare nuclei \[[Figure 1](#F1){ref-type="fig"}\]. {#F1} ### Fibrocystic disease {#sec3-2} Benign appearing ductal epithelial cells in sheets with cyst macrophages in the background \[[Figure 2](#F2){ref-type="fig"}\]. {#F2} ### Hyperplasia {#sec3-3} Hypercellular smear showing ductal epithelial cells arranged in sheets showing mild variation in size and shape. Few cells may show nuclear atypia \[[Figure 3](#F3){ref-type="fig"}\]. {#F3} ### Carcinoma {#sec3-4} Loosely arranged clusters of ductal epithelial cells showing nuclear pleomorphism, increased nuclear cytoplasmic ratio \[[Figure 4](#F4){ref-type="fig"}\], nuclear indentations \[[Figure 5](#F5){ref-type="fig"}\], and hyperchromatic nucleus. Mitotic activity may be seen. No bare nuclei. {#F4} {#F5} The age distribution of the cases is shown in [Table 1](#T1){ref-type="table"}. Benign lesions were in the age group ranging from 21 to 40; hyperplasia was seen between 31 and 60 and malignancy between 40 and 70. ###### Age distribution of cases  Nuclear morphometric analysis was carried out using the Image J 1.44C morphometric software for image processing and analysis \[[Figure 6](#F6){ref-type="fig"}\]. The basic results of our study are shown in [Table 2](#T2){ref-type="table"}. {#F6} ###### Nuclear morphometric analysis between the groups  Using one-way ANOVA, the nuclear area, perimeter, diameter, compactness, and concave points were found to be statistically significant (*P* \< 0.05). For comparisons between the individual groups we employed *post hoc* test i.e., Bonferroni Multiple Comparisons Test. There was a significant difference in the nuclear area and diameter between fibroadenoma, fibrocystic disease, and carcinoma with a *P* value of (0.0009) and (0.0007), which is considered to be extremely significant. There was a significant difference in perimeter and compactness between fibroadenoma, fibrocystic disease, hyperplasia, and carcinoma with a *P* \< 0.0001, which is considered to be statistically significant. Unpaired "*t*" test was used to find the significance of concave points between hyperplasia and carcinoma, which was highly significant (*P* \< 0.0001). The mean nuclear area and perimeter were useful in differentiating benign and malignant breast aspirates. The ductal carcinoma cells showed higher values for nuclear area, perimeter, diameter, compactness, and concave points when compared to fibroadenomas, fibrocystic disease, and hyperplasia. D[ISCUSSION]{.smallcaps} {#sec1-4} ======================== The leading cause of cancer mortality in Indian women is breast cancer with an annual diagnosis of 80,000 new cases every year.\[[@ref6]\] Hence, adequate screening of the breast lumps is essential to safeguard the health of women. The progress of normal breast to carcinoma follows a sequence of events. There are several diagnostic modalities starting from the clinical examination, mammography, FNAC, biopsy etc., However, how precise are each one of them in giving accurate diagnosis. Though cytology is able to categorize benign and the malignant breast diseases, there are gray zones is cytology where an inconsistent diagnosis may be offered. The gray zones in cytology are around 8.9% as reported by al-Kaisi.\[[@ref7]\] These included technical limitations (4.5%), inexperience of the cytopathologist (2.4%), and overlap of cytological features of benign vs. malignant (2%).\[[@ref7]\] Our study aimed to explore the possible role of nuclear morphometric analysis to differentiate benign from the malignant lesions. Morphometric analysis of nuclear parameters has been studied by several authors.\[[@ref2][@ref4][@ref8]--[@ref11]\] In the present study, the size related parameters (area, perimeter, diameter, concave points and compactness) of the nucleus were appropriate parameters to differentiate between benign lesions and infiltrative ductal carcinoma of the breast. These parameters showed significant differences between the benign breast lesions and carcinoma (*P* \< 0.05). Some studies have also measured long axis and short axis as nuclear morphometric parameters.\[[@ref3][@ref4][@ref8]\] However, among the nuclear parameters nuclear area and perimeter are important. In our study, there was a gradual increase in the nuclear area and perimeter in carcinomas when compared to benign lesions. Our results were in concordance with that of Fathi *et al*.\[[@ref4]\] with the mean nuclear area being 64-82 um^2^ for benign cases and 72-163 um^2^ for malignant cases. Abdalla *et al*.\[[@ref4]\] also showed that clearly reduced cohesiveness was associated with larger nuclear size. Wittekind and Schulte in their study showed that perimeter was the most powerful feature to differentiate between benign and malignant breast lesions.\[[@ref12]\] In our study, nuclear perimeter and compactness was highly significant in differentiating hyperplasia from carcinoma (*P* \< 0.0001). Concave points represent the number of indentations present on the nuclear border. This parameter was found to be statistically significant (*P* \< 0.0001) in differentiating hyperplasia from carcinoma. Shape is one of the factors to assess nuclear atypicality. Shape factors have been shown to have prognostic value in breast cancer as proved by Yan *et al*.\[[@ref3]\] He reported that the shape factor that includes short nuclear axis and the longest nuclear axis is of value to predict subsequent development of breast cancer among women with benign breast disease.\[[@ref3]\] Nuclear form factor, a measure of the regularity of the nuclear perimeter was shown to have predictive value for discriminating benign and malignant conditions as proved by Mapstone and Zakhour.\[[@ref9]\] However, studies by Abdalla *et al*.\[[@ref4]\] and Kalhan *et al*.\[[@ref8]\] showed that shape factors were not significant in differentiating benign from the malignant lesions. Hence we did not analyse shape factors in our study. Many studies have shown that there is a progressive pattern of nuclear morphometric parameters with gradually increasing values from benign to atypical, to ductal carcinoma in-situ (DCIS), further to invasive carcinoma and carcinoma with the lymph node involvement.\[[@ref8]\] Keunen-Boumeester *et al*. in a prognostic study of breast carcinoma aspirates concluded that the standard deviation of nuclear area along with the presence of axillary metastases was the most important predictor of prognosis.\[[@ref13]\] Similarly, Pienta and Coffey\[[@ref2]\] showed that there was a sharp increase in the nuclear area in patients with the node positive disease when compared to node negative disease. Boon *et al*.\[[@ref14]\] used nuclear/cytoplasmic ratio for characterizing cells of different tumors; however, Abdalla *et al*.\[[@ref4]\] opined that such parameter to be avoid as outlining of cellular margins is difficult due to indistinct cytoplasmic outline than nuclear outline, thus making the analysis less reproducible and more subjective. C[ONCLUSIONS]{.smallcaps} {#sec1-5} ========================= Nuclear morphometry is thus, a useful objective tool in the differentiating benign and the malignant breast lesions. It can be of immense help when diagnostic dilemmas are encountered especially in gray zones.\[[@ref8]\] It can be combined with other ancillary methods such as mammography, DNA cytometry chromatin texture analysis, flow cytometry, and cDNA array analysis for selecting the patients for adjunct therapy.\[[@ref4]\] **Source of Support:** Nil. **Conflict of Interest:** None declared.

1. Introduction {#s0005} =============== *Mikania cordata* (Burm. f.) Robinson belongs to the family Asteraceae is a climbing herb and distributed in tropical Asia, the Philippines, Papua New Guinea and tropical Africa ([@b0005]). Presence of cordate leaves, capitulum inflorescence, whitish flowers, and narrowly oblong cypsela with white pappus distinguishes the species from remaining members of the Asteraceae. In Bangladesh, *M. cordata* is widely distributed throughout the country and used as an ethnomedicinally important plant in the treatment of cuts and wounds. Leaf decoction is applied for treatment of dysentery, indigestion and gastro-intestinal sore ([@b0045]). The plant is also used as leafy vegetable and applied against coughs, eye sores and gastro-intestinal disorders. Different ethnic communities residing in different districts of Bangladesh use leaves of *M. cordata* for treating cuts, wounds, fever and pain. The *Garo* tribal people residing in Netrokona district employ leaves of this species in cuts and wounds ([@b0095]). In order to treat gastric pain fried leaves of this species are eaten by the *Mandi* tribal people in Tangail district ([@b0090]). The *Santals* tribal people residing in Thakurgaon district use the leaf of *M. cordata* for the treatment of cuts and wounds, and as remedy of dengue fever ([@b0100]). *M. cordata* has already been investigated and reported to exhibit different pharmacological activities. [@b0065] showed psycho-pharmacological, neuro-pharmacological, antimicrobial and therapeutic properties of *M. cordata* against pain, inflammation, hyperthermia, ulcer as well as carcinogenesis. The CNS depressant activity with potential antioxidant properties of *M. cordata* was found ([@b0050]), while crude ethanolic extract of *M. cordata* presented analgesic, cytotoxicity, and antibacterial activities ([@b0075]). Very recently, chloroform extract of the aerial parts of *M. cordata* was found very bioactive in nature ([@b0120]). To provide scientific support to traditional and folklore usage of *M. cordata* in Bangladesh for the treatment of different ailment, in this study, we aimed to explore and identify the chemical compound(s) from *M. cordata* and evaluate the role of the compounds behind bioactivities of the species. In this view, we subjected chloroform extract for investigation to isolate bioactive compounds and as a consequence, a pentacyclic triterpenoid, 16-hydroxy betulinic acid (16 HBA) was isolated, identified and characterized for the first time from *M. cordata*. The isolated 16-hydroxy betulinic acid was further studied extensively for its antimicrobial, analgesic, anti-inflammatory and antipyretic activities. 2. Materials and methods {#s0010} ======================== 2.1. Plant material {#s0015} ------------------- *Mikania cordata* was collected from Bhabrasur of Muksudpur upazila under Gopalganj district, Bangladesh in 2010. The aerial parts were plucked from the collected specimens. The habit and inflorescence of the plant are shown in [Fig. 1](#f0005){ref-type="fig"}. The voucher specimen has been prepared and deposited in Bangladesh National Herbarium for further reference (Voucher No. 37894).Fig. 1Habit and inflorescence of *Mikania cordata*; A. Habit showing cordate leaves; B. Capitulum inflorescence. 2.2. Drugs and chemicals {#s0020} ------------------------ Morphine manufactured by Gonoshasthaya Pharmaceuticals Ltd, Dhaka, and Diclofenac sodium manufactured by ACI Pharmaceutical Ltd., Dhaka Bangladesh were used for the experiments. In addition, acetic acid (Merck, Germany) and carrageenan (BDH, UK.) were employed in this study. For other chemicals, only the analytical grade chemicals (Sigma and Merck) were used in the experiments. 2.3. Experimental animal models {#s0025} ------------------------------- Swiss albino mice of 25--30 g and Wister albino rats of 150--200 g were employed in the experiment which were collected from ICDDR, B (International Centre for Diarrhoeal Diseases and Research, Bangladesh). The animals were kept at 25 ± 1 °C with 60--70% humidity and normal day/night cycle (12 h each) in standard polypropylene cages. Standard pellets as basal diet (ICDDR, B formulated) and water *ad libitum* were provided to the animals for one week. After fasting for overnight, the animals were weighted prior to run the experiment. During carrying out the experiments, the guidelines of the National Institute of Health for the Care and Use of Laboratory Animals (NIH Publication revised in 1996) were strictly followed. 2.4. Preparation of chloroform extract {#s0030} -------------------------------------- After air drying, the collected leaves of *M. cordata* were crushed to make leaf powder. Then 200 g of the powder was subjected to extraction by 1000 ml of chloroform at room temperature for 7 days. After filtration, the solvent was evaporated from the solution with the help of rotary evaporator operated at 50 °C which yielded 7 g of extract. 2.5. Isolation of compound from chloroform extract {#s0035} -------------------------------------------------- Chloroform extract (7 g) was subjected to silica gel column chromatography (364 g, Kieselgel 60, 230--400 mesh, Merck) and in this case, the mobile phase was started with 100% n-hexane on fraction 1; Hexane-CHCl~3~ (8.5:1.5) solvent systems on fraction 2; Hexane-CHCl~3~ (7.5:2.5) on fraction 3; Hexane-CHCl~3~ (5:5) solvent systems on fraction 4; Hexane-CHCl~3~ (3:7) solvent systems on fraction 5; CHCl~3~ (100%) solvent systems on fraction 6; CHCl~3~-MeOH (9:1) solvent systems on fraction 7; CHCl~3~-MeOH (8:2) solvent systems on fraction 8; CHCl~3~-MeOH (7:3) solvent systems on fraction 9--10; CHCl~3~-MeOH (5:5) solvent systems on fraction 11--12; CHCl~3~-MeOH (3:7) solvent systems on fraction 13 and MeOH solvent on fraction 14--15. A total of 15 fractions were collected and based on their TLC profiles, some of the fractions were combined with each other. The fraction 3 obtained from first column chromatography was then subjected for further column chromatography using same silica gel for fractionations. The starting solvent system was petroleum ether on fraction 1; Pet.ether-EtOAc (9:1) solvent system on fraction 2; Pet.ether-EtOAc (8:2) solvent system on fraction 3; Pet.ether-EtOAc (7:3) solvent system on fraction 4; Pet.ether-EtOAc (6:4) solvent system on fraction 5; Pet.ether-EtOAc (5:5) solvent system on fraction 6; Pet.ether-EtOAc (4:6) solvent system on fraction 7--8; Pet.ether-EtOAc (3:7) solvent system on fraction 9--10; EtOAc (100%) solvent system on fraction 11; EtOAc-MeOH (9:1) solvent system on fraction 12; EtOAc-MeOH (8:2) solvent system on fraction 13; EtOAc-MeOH (6:4) solvent system on fraction14; EtOAc-MeOH (5:5) solvent system on fraction 15; MeOH (100%) solvent system on fraction 16. Fraction 2 was then subjected to prepare thin layer chromatography using solvent system ethyl acetate: petroleum ether (6:4) as mobile phase on silica gel 60 (mesh 230--400). The compound 16-hydroxy betulinic acid was separated from fraction 2 with the R~f~ value of 0.66. The compound was characterized as 16-hydroxy betulinic acid on the basis of ^1^H-NMR spectra recorded at room temperature on a Bruker Avance 300 instrument operating at a frequency of 300 MHz. The schematic diagram for isolation and purification of 16 HBA is shown in [Fig. 2](#f0010){ref-type="fig"}.Fig. 2Schematic diagram for isolation and purification of 16-hydroxy betulinic acid. 2.6. Antimicrobial activity {#s0040} --------------------------- ### 2.6.1. Disc diffusion assay for antibacterial activity {#s0045} *Bacillus subtilis* IFO 3232, *Sarcina lutea* IAM 1671, *Escherichia coli* IFO 3007, *Klebsiella pneumoniae* ATTC 10031, *Xanthomonas campestris* IAM 1671 and *Pseudomona*s sp. ATCC 13867 were used in this study. For this assay, the method of [@b0070] was followed. 100 μL of standardized inoculums suspension containing 10^7^ CFU/mL of bacteria was used. Dimethylsulfoxide (DMSO) was used in preparing 16-hydroxy betulinic acid sample by dissolving in it and the concentration of this solution was maintained at 10 mg/mL. Then 10 μL from the solution (containing 100 μg) were soaked on discs (6 mm diameter) made of sterilized Whatman No.1 filter paper and applied on the inoculated LB agar medium. DMSO (used in dissolving samples) was taken as negative control. Kanamycin (30 μg/disc) sourced from Sigma-Aldrich Co. (St. Louis, MO, USA) was employed in the experiment as positive control. All the assays were replicated thrice. ### 2.6.2. Minimum inhibitory concentration (MIC) against tested bacteria {#s0050} The MIC values of 16 HBA against the tested bacteria were determined by the method developed by [@b0020]. At first, the 16 HBA sample was mixed with LB medium in a way that different concentrations ranging from 0 to 1000 μg/mL can easily be obtained. The experiment was done by transferring 20 μL inoculums of bacteria strain (10^7^ CFU/mL) to the tube and the final test volume was 2 mL. For control tube, the test was run using only bacteria strain without samples. The incubation of the culture tubes was carried out at 37 °C over 24 h. The growth of the bacteria was observed by microscopic evaluation. The lowest inhibiting concentration (MIC) was expressed as μg/mL. ### 2.6.3. Preparation of spore suspension and test samples {#s0055} The 5--10 days old spore suspension of *Rhizoctonia solani* AG-1 (IB) KACC 40111, *R. solani* AG-2 (IV) KACC 40132, *Pythium graminicola* KACC 40155 *Tricoderma harzianum* KACC 40791 and *Fusarium oxysporum* KACC 40052 were homogenized with sterile distilled water to prepare the suspension of 10^5^ spore/ml. 16 HBA was dissolved in DMSO to prepare the stock solution which was further diluted to prepare test samples. ### 2.6.4. Determination of antifungal activities {#s0060} Petri dishes (9 cm in diameter) containing 20 ml of potato dextrose agar (PDA) medium were used for assessing antifungal activity using disc diffusion assay ([@b0035]). Sterile Whatman paper discs (6 mm in diameter) impregnated with 100 μL (1000 μg/disc) of 16 HBA were placed at equidistant and near the border of the petri dishes. A disc of fungal inoculum of 6 mm in diameter was placed upside down in the centre of the petri dishes. Then the incubation of the petri dishes at 28 ± 2 °C was continued till the growth of control slates unto the corner of the plates. The plates were used in triplicates for each treatment. Growth inhibition was calculated using the following formula:$$\text{Inhibition\ ratio}\mspace{6mu}(\%) = \{ 1\text{-mycelium\ growth\ of\ treatment}\mspace{6mu}(\text{mm})/\text{mycelium\ growth\ of\ control}\mspace{6mu}(\text{mm})\} \times 100$$ ### 2.6.5. Minimum inhibitory concentration (MIC) against tested fungi {#s0065} Two-fold dilution method ([@b0070]) was used in this assay. Different sequential concentrations of 125, 250, 500 and 1000 μg/ml of 16 HBA were obtained by serial dilution with DMSO and subsequent adding to PDB medium. The minimum concentrations (MIC) which completely inhibit fungal growth were determined by inoculation of 10 μl spore suspension in the test tubes containing specified concentrations of test sample, followed by incubation for 5--7 days at 28 ± 2 °C. The control tubes containing PDB medium were inoculated only with fungal suspension. The MIC values were expressed as μg/mL. 2.7. Analgesic activity study {#s0070} ----------------------------- ### 2.7.1. Analgesic activity against peripheral nociceptive pain {#s0075} Evaluation of 16 HBA for analgesic activity against peripheral nociceptive pain was performed using the method of [@b0060]*.* Each group included five randomly selected mice and two different groups were received suspension of 50 and 100 mg/kg b.w. (body weight) of 16 HBA. After 30 min, each treatment group received 3% acetic acid to induce pain and then the writhing count was started from 5 min and continued for 20 min. Diclofenac sodium (40 mg/kg b.w.) was used as standard reference drug, whereas the control group received only 5% Tween 80 in normal saline (10 ml/kg b.w.). The percentage of inhibition was calculated as per following equation:$$\%\mspace{6mu}{Inhibition} = \frac{{Mean}\mspace{6mu}{writhing}\mspace{6mu}{count}\mspace{6mu}({control}\mspace{6mu}{group} - {treated}\mspace{6mu}{group}) \times 100}{{Mean}\mspace{6mu}{writhing}\mspace{6mu}{count}\mspace{6mu}{of}\mspace{6mu}{control}\mspace{6mu}{group}}$$ ### 2.7.2. Analgesic activity against central nociceptive pain {#s0080} For hot plate experiment, the method of [@b0085] with slight modification was followed. The hot plate temperature was kept at 55 ± 1.0 °C and mice were selected based on their response to hot plate thermal stimulation after 4 s. Suspension of 16 HBA of 50 and 100 mg/kg b.w. and morphine (10 mg/kg) as a positive control were given to mice. The negative control group received only 5% Tween 80 in normal saline (10 ml/kg b.w.). Each group of mice was assayed for reaction time at 30, 60, and 90 min after treatment with test samples or morphine. To avoid tissue impairment a latency period of 30 s was considered as complete analgesia cut off time.$$\%\mspace{6mu}{of}\mspace{6mu}{elongation}\mspace{6mu}{of}\mspace{6mu}{reaction}\mspace{6mu}{time} = \frac{\text{Average\ reaction\ time\ of\ test\ group} - \text{Average\ reaction\ time\ of\ control\ group}}{\text{Average\ reaction\ time\ of\ control\ group}} \times 100$$ 2.8. Carrageenan-induced paw edema assay {#s0085} ---------------------------------------- The efficacy of 16 HBA in reducing induced inflammation was evaluated by acute inflammation method in rats as designated by [@b0150]. The basal volume (Co) of the right hind paw of each of the five rats of each group was measured by plethysmometer (model 7150, UgoBasile, Italy). After 30 min of treatment with suspension of 16 HBA (50 and 100 mg/kg b.w.) and phenylbutazone (100 mg/kg) as positive control, each of the rats of all groups received 1% carrageenan (0.1 ml) into sub-plantar surface of right hind paw. The control group received only 5% Tween 80 in normal saline (10 ml/kg b.w.). The volume of the right hind paw edema was measured at 0 h, 1 h, 2 h, 3 h and 4 h. Inhibitory activity was calculated by the following formula:$$\%\mspace{6mu}{Edema}\mspace{6mu}{Inhibition} = \frac{({Ct} - {Co}){Control} - ({Ct} - {Co}){treated}}{({Ct} - {Co}){Control}} \times 100$$ where Ct refers to paw size after a specific time interval in hours after carrageenan injection and Co denotes paw size before carrageenan injection. 2.9. Yeast-induced hyperthermia assay {#s0090} ------------------------------------- After recording initial rectal temperature by digital electric thermometer (EXACON, Denmark), the rats were given 20 ml/kg (20%) brewer's yeast suspension subcutaneously to induce pyrexia ([@b0135]). The rat groups (n = 5) which exhibited an increase in temperature of 0.3--0.5 °C after 18 h, were further treated with 16 HBA (50 and 100 mg/kg b.w.) and paracetamol (100 mg/kg b.w.) as standard drug, whereas the control group received only 0.3 ml of saline. Then the rectal temperature of the rats was measured at 1 h, 2 h, 3 h and 4 h. The average temperature of every group was compared with mean temperature of the group treated with standard drugs. 2.10. Statistical analysis {#s0095} -------------------------- All data were analyzed by means of ANOVA test followed by Student's t-test. Significant levels were determined at P \< 0.05 (95% confidence limit). 3. Results and discussion {#s0100} ========================= 3.1. Structure elucidation of the compound {#s0105} ------------------------------------------ Based on bio-assay, the chloroform leaf extract was selected for the purification of compounds. The leaf extract (7 g) was subjected to column chromatography over silica gel (230--400 mesh ASTM, Merck, Germany) to give 15 fractions. The fraction 3 was then further subjected to column chromatography over silica gel to give 16 fractions. Then the compound was obtained from fraction 2 by PTLC using suitable solvent system and was finally characterized as 16-hydroxy betulinic acid by ^1^H NMR spectroscopic data analysis. 16 HBA was obtained as yellowish oily liquid. It was insoluble in water but highly soluble in chloroform, pyridine and acetic acid. TLC examination showed that it to be a single compound with R~f~ value of 0.66 in EtOAc: Pet. ether (6:4). It was appeared as a dark spot on TLC plate (silica gel 60 GF254) under UV light at 254 nm and after the development of the chromatoplates in closed iodine chamber, it was appeared as yellow spot. The ^1^H NMR (CDCl~3~, 300 MHz) spectrum of the compound ([Fig. 3](#f0015){ref-type="fig"}) revealed the presence of a lupine type carbon skeleton. It displayed signals attributable to an exomethylene group at *δ* 4.39 and 4.41 ppm (1H, each, br.s) which together with an allylic methyl at *δ* 1.77 that indicated an isopropenyl function. The two olefinic protons H-29 (a) and H-29 (b) resonated at *δ* 4.39 and *δ* 4.41 ppm, respectively and their relative assignment could be made due to the vicinity of H-29 (a) to the methyl group at C-30.Fig. 3^1^H NMR (CDCl~3~, 300 MHz) spectrum of 16-hydroxy betulinic acid. The next signal in the proton spectrum at *δ* 3.5 ppm integrated for two protons which form their chemical shift, and were assigned to H-19 and H-3. The signal of methyl group at C-30 attached to a SP^2^ centre could be safely identified from its chemical shift of *δ* 1.77. In addition, the compound showed five singlets at *δ* 0.83 (3H, Me-25), 1.01 (3H, Me-24), 1.12 (3H, Me-27), 1.13 (3H, Me-26) and 1.21 (3H, Me-23), each integrating for three protons which assigned for five methyl groups in the compound. The characteristic signal at *δ* 5.18 was due to highly deshielding suggested the presence of hydroxyl group at C-16. All other peaks in ^1^H NMR data of 16 HBA also matching with literature values ([@b0125]) and are in good agreement with those of 16 HBA. The ^1^H NMR (CDCl~3~, 300 MHz) data of 16 HBA acid is shown in [Table 1](#t0005){ref-type="table"} and its chemical structure is presented in [Fig. 4](#f0020){ref-type="fig"}. Betulinic acid and some of its derivatives have been isolated from some other plant species ([@b0155]). *M. cordata* has already been reported to contain some chemical compounds e.g. vitamin A, B, and C, mikanin, friedelin, efifriedinol, dilactones, etc. ([@b0045]), however, there is no report available so far on isolation of betulinic acids or any of its derivatives from this plant. This is the first report for isolation and purification of a betulinic acid derivative i.e. 16-hydroxyl betulinic acid (a triterpenoid) from the leaves of *M. cordata*.Table 1^1^H NMR spectral data of the 16-hydroxy betulinic acid (300 MHz, *δ*, CDCl~3~) with reference to [@b0125].Position*δ*~H~Position*δ*~H~10.84 (s)1721.24 (m)181.80 (m)33.48 (t)193.52 (m)42050.83 (m)2161.26 (m)221.80 (m)7231.21 (s)8241.01 (s)91.28 (t)250.83 (s)10261.13 (s)111.18 (m)271.12 (s)121.15 (m)28132.46 (m)294.39(br.s)/4.41(br.s)14301.77 (s)151.80 (m)COOH7.53 (s)165.18 (d)OH7.74 (s)Fig. 4Structure of 16-hydroxy betulinic acid isolated from *Mikania cordata*. 3.2. Antimicrobial activity {#s0110} --------------------------- The results of the assays for antibacterial activities of the 16-hydroxy betulinic acid are shown in [Table 2](#t0010){ref-type="table"}. The 16-hydroxy betulinic acid showed potent efficacy against all six bacteria. For all the six bacteria, the zone of growth inhibition ranged from 12.0 to 17.5 mm. The present study revealed that the efficacy shown by 16-hydroxy betulinic acid is somewhat weaker than that of Kanamycin. The blind control showed no activity against the bacteria. From the minimum inhibitory concentration (MIC) assay, it was observed that the two gram positive bacteria, *Sarcina lutea* IAM 1671 and *Bacillus subtilis* IFO 3232 were very much susceptible to 16-hydroxy betulinic acid with lowest MIC 31.5 μg/mL values ([Table 2](#t0010){ref-type="table"}), whereas the gram negative bacteria, *Escherichia coli* IFO 3007, *Klebsiella pneumoniae* ATTC 10031, *Pseudomonas* sp. ATCC 13867 and *Xanthomonas campestris* IAM 1671 were found somewhat less susceptible to the 16-hydroxy betulinic acid with MIC values (62.5--125) μg/mL. Our findings are in agreement with previous reports where some hydroxy derivatives of betulinic acid were found to have antibacterial activities ([@b0110]).Table 2Antibacterial activity and minimum inhibitory concentrations of 16-hydroxy betulinic acid.BacteriaZone of growth inhibition (in mm)[a](#tblfn1){ref-type="table-fn"}Minimum inhibition concentration (μg/ml)16-hydroxy betulinic acidKan[b](#tblfn2){ref-type="table-fn"}*Bacillus subtilis* IFO 323216 ± 0.721.5 ± 0.731.5*Sarcina lutea* IAM 167117.5 ± 0.416 ± 1.131.5*Escherichia coli* IFO 300714 ± 1.320.5 ± 1.262.5*Xanthomonas campestris* IAM 167114 ± 0.924 ± 1.462.5*Klebsiella pneumoniae* ATTC 10,03112 ± 1.720 ± 0.6125*Pseudomonas* sp. ATCC 13,86714.5 ± 0.320 ± 1.2125[^1][^2] The 16-hydroxy betulinic acid exhibited a moderate antifungal activity against all the tested fungi except *Tricoderma harzianum* KACC 40791 and a low inhibition effect was observed against *Fusarium oxysporum* KACC 41083. At the concentration of 10 μl (1000 μg/mL), the 16-hydroxy betulinic acid showed a potent inhibitory effect on the growth of *Rhizoctonia solani* AG-1 (IB) KACC 40111 (45.4%), *R. solani* AG-2 (IV) KACC 40132 (41.9%), *Pythium graminicola* KACC 40155 (54.5%), and *Fusarium oxysporum* KACC 40052 (37.6%) as shown in [Table 3](#t0015){ref-type="table"}.Table 3Antifungal activity and minimum inhibitory concentrations of 16-hydroxy betulinic acid.Fungal strainsRadial growth inhibitionMinimum inhibition concentrationmm[a](#tblfn3){ref-type="table-fn"}(%)[b](#tblfn4){ref-type="table-fn"}(μg/ml)*Rhizoctonia solani* AG-1 (IB) KACC 401112.1 ± 0.4[c](#tblfn5){ref-type="table-fn"}45.4 ± 0.5[c](#tblfn5){ref-type="table-fn"}500*Rhizoctonia solani* AG-2-2 (IV) KACC 401322.7 ± 0.9[c](#tblfn5){ref-type="table-fn"}41.9 ± 0.4[c](#tblfn5){ref-type="table-fn"}500*Pythium graminicola* KACC 401552.5 ± 0.9[c](#tblfn5){ref-type="table-fn"}54.5 ± 0.8[c](#tblfn5){ref-type="table-fn"}250*Tricoderma harzianum* KACC 40791ndndnd*Fusarium oxysporum* KACC 400521.9 ± 0.8[c](#tblfn5){ref-type="table-fn"}37.6 ± 0.9[c](#tblfn5){ref-type="table-fn"}1000[^3][^4][^5] In the MIC assay, 16-hydroxy betulinic acid was found most effective against *P. graminicola* (250 μg/mL) as compared to those of *R. solani* AG-1 and *R. solani* AG-2 (500 μg/mL), whereas the MIC values were found lowest (1000 μg/ml) in *Fusarium oxysporum* KACC 40052 ([Table 3](#t0015){ref-type="table"}). Our result is congruent with some other previous studies where betulinic acid and some other derivatives of betulinic acid were reported to have antifungal effects ([@b0115]). The antimicrobial activities of pentacyclic compounds having lupane skeleton is well reported and the compounds were found to exert their activities by initiating changes in membrane permeability resulting from the leakage of reducing sugars, proteins and reduction of respiratory chain dehydrogenase enzyme's activity ([@b0105]). This mechanism can also be attributed to 16-hydroxy betulinic acid for its antimicrobial activity. 3.3. Analgesic activity study {#s0115} ----------------------------- 16-hydroxy betulinic acid showed analgesic efficacy against peripheral nociceptive pain by reducing acetic acid induced writhing in mice ([Fig. 5](#f0025){ref-type="fig"}). The percentage inhibition of writhing for 16-hydroxy betulinic acid at 50 and 100 mg/kg were 41% and 55.19%, respectively. The percentage inhibitions were determined by comparing the writhing response of animals received test specimen to the animals that received vehicle only. By comparison, 40 mg/kg diclofenac sodium was found to produce highest 52% analgesic effect in this nociception model. In this study, the tested samples exhibited dose dependent analgesia.Fig. 5Effect of the treatment 16-hydroxy betulinic acid (50, 100 mg/kg), and diclofenac sodium as control (40 mg/kg) on the intraperitoneal administration of 10 ml/kg of 3% acetic acid. Each group represents the mean of 5 animals. Values are expressed as mean ± SEM, ^∗^P \< 0.05 and ∗∗P \< 0.01 compared with control group. In the hot plate study, the 16-hydroxy betulinic acid (50 and 100 mg/kg) elongated the reaction time at different time points. All these elongation in the reaction time was determined by comparing to the corresponding control groups ([Table 4](#t0020){ref-type="table"}; [Fig. 6](#f0030){ref-type="fig"}). Morphine significantly prolonged the reaction time of the animals, e.g. 116.99%, 117.52% and 85.55% relative elongation of reaction time at 30 min, 60 min and 90 min, respectively, confirming centrally mediated activity. The compound 16-hydroxy betulinic acid showed central anti-nociceptive activity in dose dependent manner and the highest analgesia was induced at 100 mg/kg b.w. by elongating the reaction time 97.97%, 103.11%, and 89.29% at 30 min, 60 min and 90 min, respectively.Table 4Effects of 16-hydroxy betulinic acid on hot plate test in mice.TreatmentDose (mg/kg)Reaction time (s)30 min60 min90 minControl (Vehicle)--6.24 ± 0.616.28 ± 0.286.92 ± 0.34Morphine1013.54 ± 0.49^\*^13.66 ± 0.39^\*\*^12.84 ± 0.56^\*\*^16-hydroxy betulinic acid509.73 ± 0.19^\*\*^9.67 ± 02110.39 ± 0.15^\*\*^10012.35 ± 0.12^\*\*^12.75 ± 0.09^\*\*^13.1 ± 0.03^\*\*^[^6][^7]Fig. 6Effect of the treatment with 16-hydroxy betulinic acid (50, 100 mg/kg) and morphine as control (10 mg/kg) on the hot plate analgesic test. Each mouse is served as its own control. Before treatment, the time for hind paw licking or jumping on the heated plate of analgesiometer was determined thrice at 1 h interval and mean of these were taken as the initial reaction time (control latency). Mice in each group were tested 30, 60 and 90 min after drug treatment. Values are expressed as mean ± SEM, ^∗^P \< 0.05 and ^∗∗^P \< 0.01 compared with control group. The analgesic activity (both peripheral and central) of the test specimens were investigated by applying two different methods. The most suitable and widely used acetic acid induced writhing method was applied to identify the peripheral effects, whereas hot plate method was employed to detect analgesic activity against central nociceptive pain of the test samples. Our findings clearly indicated that 16-hydroxy betulinic acid possessed both peripheral (writhe reduction) and central (thermal reaction time prolongation) analgesic effects. In the peripheral nociception model, the pain is developed by inducing local inflammatory sensation. Following inflammation, there is biogenesis of prostaglandins from cyclooxygenase pathway and leukotrienes from lipoxygenase pathway ([@b0030]). The released prostaglandins, mainly prostacyclin (PGI2), and to a lesser extent, prostaglandin E, are supposed to induce pain and abnormal contraction of mice body that is termed as 'writhing'. Several lupane triterpenoids found to prevent the biogenesis process of lipooxygenase and/or cyclooxygenase which in turn, affect the formation of PGE2 ([@b0130]). Thus, the lupane triterpenoids under investigation (16-hydroxy betulinic acid) might suppress transformation mechanism in the nociceptor which results in reduced or no pain perception. Therefore, to evaluate the actual analgesic activity, another test model e.g. hot plate thermal stimulation in combination with the acetic acid induced writhing model is generally suggested. Likewise, 16-hydroxy betulinic acid was also found to prolong reaction time in hot plate thermal stimulation test. The effect of 16-hydroxy betulinic acid is comparable with that obtained by morphine. Central nervous system depression properties of morphine resulted in reduced behavioral activities. Previous studies revealed that pentacyclic lupane triterpenoids might be able to produce locomotors depressant, muscle relaxant and sedative potential effect which, in turn, leads to the conclusion that those compounds have some central nervous system depression properties ([@b0130]) and all those observation can help in explaining central analgesic activity of 16-hydroxy betulinic acid in hot plate analgesic model. 3.4. Carrageenan-induced rat paw edema assay {#s0120} -------------------------------------------- Results of the anti-inflammatory effects of 16-hydroxy betulinic acid are presented in [Table 5](#t0025){ref-type="table"} and [Fig. 7](#f0035){ref-type="fig"}. Dose dependent paw edema inhibition effect was observed and the highest inhibition of paw edema (77.08%) was recorded at 100 mg/kg of 16-hydroxy betulinic acid at 4 h.Table 5Anti-inflammatory activities of 16-hydroxy betulinic acid and phenylbutazone (PBZ) on carrageenan-induced edema in the right hind-limb paw of rats.TreatmentDose (mg/kg)Time (h)Average edema formation01234EV (ml)EV (ml)EV (ml)EV (ml)EV (ml)Control (saline) edema formation (mm) size----0.35 ± 0.040.41 ± 0.050.44 ± 0.0500.48 ± 0.060.42Phenylbutazone100--0.24 ± 0.030.16 ± 0.05^\*\*^0.15 ± 0.03^\*^0.06 ± 0.04^\*\*^0.1516-hydroxy betulinic acid50--0.31 ± 0.030.30 ± 0.01^\*^0.28 ± 0.1^\*^0.22 ± 0.07^\*\*^0.28100--0.29 ± 0.0^\*^0.25 ± 0.04^\*\*^0.19 ± 0.3^\*\*^0.11 ± 0.12^\*\*^0.21[^8]Fig. 7Effect of the treatment with 16-hydroxy betulinic acid (50, 100 mg/kg) and Phenylbutazone as control (100 mg/kg) on paw edema induced by intradermal injection of carrageenan (1% 0.1 ml) in rats. Zero (time zero) corresponds to baseline measurement of animals performed 24 h before the test. Each group represents the mean of 5 animals, and the vertical lines indicate the SEM. ^∗^P\< 0.05 and ^∗∗^P \< 0.01 compared with control group. The anti-inflammatory activity of 16-hydroxy betulinic acid can be explained based on previous studies. IL-10, a cytokine with potent anti-inflammatory properties, inhibits TNF-α release ([@b0040]). It has been reported that the betulinic acid (BA) increase in the levels of serum IL-10 ([@b0025]). [@b0145] showed that BA inhibits the expression of cyclooxygenase 2 (COX-2) in cultures of human peripheral blood mononuclear cells, leading to a decrease in PGE2 production. The increase in IL-10 induced by betulinic acid might be related to the inhibition of PGE2 production, another inflammatory mediator, since IL-10 is known to inhibit COX-2 expression ([@b0080]) which play a vital role against inflammation conferred by betulinic acid. However, the fact that betulinic acid modulates IkBα phosphorylation ([@b0145]) suggests that its action on TNF-α inhibition may also be related to alterations in signaling pathways, such as NF-kB activation. Moreover, it has been observed that betulinic acid inhibits the production of inflammatory mediators, such as NO and TNF-α ([@b0025]). All these effects of betulinic acid in suppressing inflammatory mediators and increasing anti-inflammatory cytokines might be applicable for the 16 hydroxy betulinic acid, a derivatives of betulinic acid, to explain its anti-inflammatory effects. 3.5. Yeast-induced hyperthermia assay {#s0125} ------------------------------------- The antipyretic results of 16-hydroxy betulinic acid are presented in [Table 6](#t0030){ref-type="table"}. Administration of brewer's yeast to rats produced a noticeable increase in rectal temperature after 18 h of yeast injection ([Table 6](#t0030){ref-type="table"}). In this study, although 16-hydroxy betulinic acid reduced the rectal temperature right from 1 h onward but the weaker antipyretic effect was observed in comparing to paracetamol (100 mg/kg orally) which significantly reduced the temperature ([Table 6](#t0030){ref-type="table"}).Table 6Antipyretic activities of 16-hydroxy betulinic acid and paracetamol on brewer's yeast-induced pyrexia in rats.TreatmentDose (mg/kg)Rectal temperatureBefore drugAfter drug−18 h0 h1 h2 h3 h4 hControl (Saline)0.3 ml37.22 ± 0.2038.106 ± 0.3038.018 ± 0.2838.00 ± 0.1938.05 ± 0.1938.05 ± 0.2016-hydroxy betulinic acid5036.88 ± 0.3237.92 ± 0.2137.44 ± 0.15^\*^37.04 ± 0.1136.85 ± 0.31^\*\*^36.77 ± 0.17^\*\*^10036.84 ± 0.2238.38 ± 0.15^\*^38.46 ± 0.1237.62 ± 0.31^\*^37.04 ± 0.12^\*\*^36.66 ± 0.19^\*\*^Paracetamol10036.97 ± 0.2838.19 ± 0.3237.76 ± 0.12^\*\*^37.13 ± 0.19^\*^36.98 ± 0.05^\*\*^36.95 ± 0.09^\*\*^[^9] The drugs or medicinal compounds bearing the suppressing effect on biogenesis of prostaglandin generally have antipyretic properties ([@b0140]). In our study, 16-hydroxy betulinic acid was found to inhibit pyrexia considerably developed by yeast administration. As discussed earlier, 16-hydroxy betulinic acid have impacts on central nervous system and likely to inhibit the synthesis of prostaglandin E2 and thus lessen hyperthermia. In this case, 16-hydroxy betulinic acid might exert some inhibitory effect by its action on COX-3 which decrease the amount prostaglandin E2 in the hypothalamus region of brain ([@b0010]). There is another possibility in which the tested compounds might have some positive influence to increase the secretion of vasopressin and arginine like antipyretic agents within the body ([@b0015]). 4. Conclusion {#s0130} ============= The present study confirmed that 16-hydroxy betulinic acid isolated from *M. cordata* has potent antimicrobial, analgesic, anti-inflammatory and antipyretic properties. In the recent past, betulinic acid and its different derivatives have attracted great research interest in the field of drug discovery. Our findings will not only be a support to bolster the research efforts on betulinic acids to emerge new commercial therapeutic agents from betulinic acid and its derivatives but also attract researchers to study structural and functional relationship of 16 hydroxy betulinic acid. Our results can also be seen as scientific support for the traditional and folklore usage of *Mikania cordata* in Bangladesh for the treatment of different ailments and provide opportunities to explore this plant as a source of bioactive compounds for biochemical and pharmaceutical industries. Acknowledgements {#s0140} ================ This work was supported by the Department of Applied Chemistry & Chemical Technology, Islamic University, Kushtia and the Centre for Biomedical Research, University of Dhaka, Bangladesh. The authors are thankful to Prof. Dr. A.S. Shamsur Rouf, Department of Clinical Pharmacy and Pharmacology, University of Dhaka for his cooperation during the course of the study. The authors from KSU extend sincere appreciation to the Deanship of Scientific Research at King Saud University (research group project number RG-195). Peer review under responsibility of King Saud University. [^1]: Diameter of inhibition zones (mm) of the compound around the discs (6 mm) impregnated with 10 μL/disc corresponding to 100 μg/disc. [^2]: Kanamycin (30 μg). Values are given as mean ± S.D of triplicate experiment and considered to be significantly different at P \< 0.05. [^3]: Radial growth of fungal pathogens. [^4]: Percentage of radial growth inhibition. [^5]: Values are given as mean ± S.D. (n = 3), and considered to be significantly different at P \< 0.05. nd: not detected of anti-fungal activity. [^6]: Each value represents the mean (±S.E.M.) of five observations. [^7]: ^∗^P \< 0.05 and ^∗∗^P \< 0.01 compared with control group. [^8]: EV: Edema volume. Values represent the mean ± S.E.M. (n = 5), ^∗^P \< 0.05 and ^∗∗^P \< 0.01 compared with control group. [^9]: Values are mean ± S.E.M. rectal temperature (n = 5), ^∗^P \< 0.05 and ^∗∗^P \< 0.01 compared with control group.

This study was performed at the Graduate School of Veterinary Medicine, Hokkaido University, Hokkaido, Japan. AIC : Akaike\'s information criteria *A* : late diastolic mitral inflow velocity *A*~m~ : late diastolic velocity of the septal mitral annulus AUC : area under the receiver operating characteristic curve BW : body weight CHF : congestive heart failure *E*~m~ : early diastolic velocity of the septal mitral annulus *E* : peak early diastolic mitral inflow velocity FAC : fractional area change FS : fractional shortening LA/Ao : left atrial to aortic root ratio LAA~max~ : LA maximum area LAA~min~ : LA minimum area LAA~p~ : LA area pre‐atrial contraction LA : left atrium LVIDd inc% : percent increase in LVIDd LVIDd : left ventricular diameter in diastole LVIDs inc% : percent increase in LVIDs LVIDs : left ventricular diameter in systole LV : left ventricle MMVD : myxomatous mitral valve disease MR : mitral regurgitation ROC : receiver operating characteristic RV : right ventricle *S*~a~ : strain before atrial contraction *S*~max~ : maximum strain *S*~min~ : minimum strain SR~A~ : second negative peak strain rate during atrial contraction SR~E~ : first negative peak strain rate during early ventricular diastole SR~S~ : positive peak strain rate during ventricular systole SR : strain rate STE : speckle‐tracking echocardiography ε~A~ : left atrial longitudinal strain during atrial contraction ε~E~ : left atrial longitudinal strain during early ventricular diastole ε~S~ : left atrial longitudinal strain during ventricular systole The left atrium (LA) plays an important role in maintaining adequate cardiac performance, interdependently with the left ventricle (LV). The LA modulates LV filling with its reservoir, conduit, and booster pump functions. First, the LA functions as a reservoir, receiving blood from the pulmonary veins during LV systole and isovolumic relaxation. This function is influenced by LV contraction through the descent of the mitral annulus,[1](#jvim14660-bib-0001){ref-type="ref"}, [2](#jvim14660-bib-0002){ref-type="ref"} by right ventricle (RV) contraction through the pulmonary blood flow[3](#jvim14660-bib-0003){ref-type="ref"} and by LA compliance.[4](#jvim14660-bib-0004){ref-type="ref"} Secondly, the LA functions as a conduit. Blood flows passively from the LA into the LV during early LV diastole and diastasis. Conduit function is influenced by LV relaxation and early diastolic pressure. Finally, the LA functions as a booster pump. Blood flows actively from the LA into the LV during LA contraction in late LV diastole. This function is influenced by LA contractility, LV compliance, and LV end‐diastolic pressure. The most accurate and representative index for characterizing LA mechanical function is the evaluation of volume--pressure curves.[5](#jvim14660-bib-0005){ref-type="ref"} However, obtaining these curves requires combined invasive measurements, which limits their use to experimental studies. In contrast, echocardiography is a simple and widely available tool that has been increasingly used for the noninvasive assessment of LA function.[6](#jvim14660-bib-0006){ref-type="ref"} The maximum size of the LA as assessed with 2D echocardiography is one of the best indicators of prognosis and severity in dogs with myxomatous mitral valve disease (MMVD).[7](#jvim14660-bib-0007){ref-type="ref"}, [8](#jvim14660-bib-0008){ref-type="ref"}, [9](#jvim14660-bib-0009){ref-type="ref"} Decreased LA booster pump function as indicated by fractional area change (FAC) is a more precise indicator of severity and prognosis of MMVD than maximum LA size in dogs.[10](#jvim14660-bib-0010){ref-type="ref"}, [11](#jvim14660-bib-0011){ref-type="ref"} LA strain and strain rate (SR) describe the longitudinal shortening and lengthening of the LA wall, allowing the quantification of all 3 LA functions: as reservoir, conduit, and booster pump. Speckle‐tracking echocardiography (STE) is a novel, angle‐independent imaging method that allows assessment of strain and SR using the tracking of acoustic speckle patterns. In humans, assessing LA strain with STE is feasible and reproducible,[12](#jvim14660-bib-0012){ref-type="ref"} and LA strain is a useful severity indicator of mitral regurgitation (MR).[13](#jvim14660-bib-0013){ref-type="ref"}, [14](#jvim14660-bib-0014){ref-type="ref"}, [15](#jvim14660-bib-0015){ref-type="ref"} The clinical feasibility and reproducibility of tissue Doppler imaging[16](#jvim14660-bib-0016){ref-type="ref"} and STE[17](#jvim14660-bib-0017){ref-type="ref"} in evaluating LA strain and SR in normal dogs have been reported. However, the clinical utility of these methods in veterinary medicine remains unclear. The aim of this study was to examine whether STE‐derived indicators of LA strain and SR are associated with MMVD stage in dogs. Materials and Methods {#jvim14660-sec-0008} ===================== Animals {#jvim14660-sec-0009} ------- Client‐owned dogs with MMVD were enrolled prospectively between December 2013 and April 2015 at Hokkaido University Veterinary Teaching Hospital. All of the dogs included in this study underwent physical examination, routine hematology, and blood biochemistry with minimum database consisting of packed blood cell volume, total protein, ALT, blood urea nitrogen, creatinine, sodium, potassium, and chloride concentrations, thoracic radiographs, and echocardiography. Dogs with atrial flutter or fibrillation and those with other concurrent cardiac disease, such as cardiomyopathy, infective endocarditis, or congenital cardiac disease, were excluded. Myxomatous mitral valve disease was classified as stage B1, B2, C, or D in each dog based on clinical signs, echocardiographic examination, and thoracic radiographs, according to the American College of Veterinary Internal Medicine consensus. Stage B1 includes dogs without clinical signs of MMVD, and radiographic and echocardiographic evidence of LA enlargement (left atrial to aortic root ratio \[LA/Ao\] \< 1.6). Stage B2 includes animals with echocardiographic evidence of LA enlargement (LA/Ao \> 1.6), but without clinical signs of MMVD. Stages C and D include dogs with the presence or history of clinical and radiographic signs of left‐sided congestive heart failure (CHF). Stages C and D dogs were united into 1 group (group C/D) because of the small sample size. Echocardiography {#jvim14660-sec-0010} ---------------- Echocardiography was performed in all dogs by an experienced veterinarian (KN) with an ultrasound unit[a](#jvim14660-note-1002){ref-type="fn"} equipped with a 4.4--6.2 MHz sector probe.[b](#jvim14660-note-1003){ref-type="fn"} Dogs were not sedated and were gently restrained in left and right lateral recumbency during the examination. The LA/Ao was obtained from the right parasternal short‐axis 2D view on the first frame after closure of the aortic valve as previously described.[18](#jvim14660-bib-0018){ref-type="ref"}, [19](#jvim14660-bib-0019){ref-type="ref"} The LV diameter in diastole (LVIDd) and LV diameter in systole (LVIDs) were measured on the M‐mode echocardiogram from the right parasternal short‐axis 2D view with concomitant ECG registration. M‐mode values were used to derive the fractional shortening (FS) and the percent increase in LVIDd (LVIDd inc%) and LVIDs (LVIDs inc%), according to the following equation: % increase = 100 × (observed dimension − expected normal dimension)/expected normal dimension. Expected normal dimensions were calculated according to the following equations: expected normal LVIDd = 1.53 × (body weight \[BW\])^0.294^ and expected normal LVIDs = 0.95 × (BW)^0.315^.[20](#jvim14660-bib-0020){ref-type="ref"} From the left apical 4‐chamber view, pulsed wave Doppler was used to measure the peak early (*E*) and late (*A*) diastolic mitral inflow velocity and tissue Doppler was used to measure the early diastolic (*E* ~m~) and late diastolic (*A* ~m~) velocity of the septal mitral annulus. Parameters of LA phasic function as indicated by FAC were determined as previously reported.[21](#jvim14660-bib-0021){ref-type="ref"} LA area was measured with planimetry in the left apical 4‐chamber view by tracing the endocardial border, excluding the confluence of the pulmonary veins. LA area was measured at 3 time points: at ventricular end‐systole as LA maximum area (LAA~max~), at the onset of the P wave on the ECG as LA area pre‐atrial contraction (LAA~p~), and at ventricular end‐diastole as LA minimum area (LAA~min~). Total, passive, and active FAC of the LA were calculated as follows:$$\text{total\ FAC} = 100 \times {(\text{LAA}_{\max} - \text{LAA}_{\min})}/\text{LAA}_{\max}$$ $$\text{passive\ FAC} = 100 \times {(\text{LAA}_{\max} - \text{LAA}_{p})}/\text{LAA}_{\max}$$ $$\text{active\ FAC} = 100 \times {(\text{LAA}_{p} - \text{LAA}_{\min})}/\text{LAA}_{p}$$ Total, passive, and active FAC are indicators of LA reservoir, conduit, and booster pump functions, respectively. LA strain and SR were analyzed with 2DSTE. For each analysis, 2D cine loops from the left apical 4‐chamber view were obtained with simultaneous ECG trace recording (lead II) to be analyzed with offline software[c](#jvim14660-note-1004){ref-type="fn"} originally designed to conduct LV analysis.[22](#jvim14660-bib-0022){ref-type="ref"}, [23](#jvim14660-bib-0023){ref-type="ref"} The depth and sector width were minimized for frame‐rate maximization (between 96 and 164 frame per second). A frame corresponding to the end‐diastolic phase was selected, and the endocardial border of the left atrium was manually defined using a point‐and‐click technique. Epicardial surface tracing was automatically generated by the system, creating a region of interest that was manually adjusted to cover the full thickness of the myocardium. After processing, the LA myocardium was separated into 6 segments. The software automatically produced time‐longitudinal strain and time‐longitudinal SR curves for each segment and their averaged values (global strain and SR) (Fig [1](#jvim14660-fig-0001){ref-type="fig"}). A cine loop preview was used to confirm whether the internal line of the region of interest followed the LA endocardial border throughout the cardiac cycle. If the internal line stopped following the endocardial border, it was manually adjusted. LA strain was measured at 3 time points: minimum strain at a negative peak during the ventricular end‐diastolic phase (*S* ~min~), maximum strain at a peak during the ventricular systolic phase (*S* ~max~), and strain before atrial contraction (*S* ~a~) on the global strain curve (Fig [1](#jvim14660-fig-0001){ref-type="fig"}A). Left atrial longitudinal strain during ventricular systole (ε~S~) was calculated as an indicator of reservoir function, strain during ventricular early diastole (ε~E~) as an indicator of conduit function, and strain during atrial contraction (ε~A~) as an indicator of booster pump function, as follows:$$\varepsilon_{S} = S_{\max} - S_{\min}$$ $$\varepsilon_{E} = S_{\max} - S_{a}$$ $$\varepsilon_{A} = S_{a} - S_{\min}$$ {#jvim14660-fig-0001} Strain rate was measured at a positive peak during ventricular systole (SR~s~), a first negative peak during early ventricular diastole (SR~E~), and a second negative peak during atrial contraction (SR~A~) (Fig [1](#jvim14660-fig-0001){ref-type="fig"}B). The mean of 3 consecutive cardiac cycles was calculated for all variables of LA function. The intra‐ and interobserver variations in strain variables in healthy Beagle dogs obtained by this method showed adequate repeatability and reproducibility (all intraobserver coefficient of variation was under 15% and interobserver coefficient of variation was under 20% in all strain variables with 6 dogs and 2 observers) (data not yet published). Statistical Analysis {#jvim14660-sec-0011} -------------------- Measurements are presented as the median (interquartile range). An overall difference between groups was determined with the Kruskal--Wallis test; posthoc multiple comparisons were made with the Steel--Dwass test for continuous variables. Fisher\'s exact test was used for categorical variables. The relationships between different parameters were assessed with Spearman\'s correlation analysis. To assess the comparative accuracy of different echocardiography variables in differentiating patients with a history of CHF (stage C/D) versus those without a CHF history (stage B1/B2), receiver operating characteristic (ROC) curves and the respective area under the ROC curve (AUC) were calculated. Predictors of heart failure were assessed with binary logistic regression analysis. The effect of retaining or dropping variables from the model was assessed with Akaike\'s information criteria (AIC) scores. AIC scores are statistical criteria that enable logistic regression model comparisons: the smaller the AIC, the better the model. For multivariate analysis, 5 echocardiographic variables without high correlation were selected based on the results of univariate analysis and a stepwise forward selection method with the lowest AIC was performed. All statistical analyses were performed with commercially available statistical software.[d](#jvim14660-note-1005){ref-type="fn"} ^,^ [e](#jvim14660-note-1006){ref-type="fn"} A 2‐sided *P* value \< .05 was considered significant. Results {#jvim14660-sec-0012} ======= A total of 52 dogs with MMVD were enrolled, including 24 dogs (4 Pomeranians, 4 chihuahuas, 3 Miniature Schnauzers, 2 Miniature Dachshund, 2 Shi Tzus, 2 Malteses, 1 Toy Poodle, 1 Yorkshire Terrier, and 5 others) in stage B1, 15 dogs (4 Chihuahuas, 2 Cavalier King Charles Spaniels, 2 Papillons, 2 Miniature Schnauzers, 1 Maltese, 1 Shih Tzu, and 3 others) in stage B2, 10 dogs (2 Shih Tzus, 1 Chihuahua, 1 Maltese, 1 Miniature Schnauzer, 1 Yorkshire Terrier, and 4 others) in stage C, and 3 dogs (1 Shih Tzu, 1 Miniature Dachshund, and 1 Maltese) in stage D. Table [1](#jvim14660-tbl-0001){ref-type="table-wrap"} shows the demographic data, physical examination results, and radiographic and echocardiographic characteristics of the study population. ###### Clinical and echocardiographic characteristics of dogs with MMVD at different stages (ACVIM consensus) B1 (n = 24) B2 (n = 15) C/D (n = 13) -------------------------------------------------------- ------------------------- ------------------------- ----------------------- Age (years) 10^a^ (8--11) 10^b^ (9--13) 12^b^ (11--14) Sex (male/female) 15/9 10/5 10/3 Body weight (kg) 4.5^a^ (3.4--6.5) 4.6^a^ (3.2--7.0) 5.9^a^ (4.0--7.5) Heart rate (bpm) 138^a^ (114--144) 150^a^ (124--167) 139^a^ (115--157) LA/Ao 1.44^a^ (1.27--1.50) 1.92^b^ (1.73--2.15) 2.44^c^ (2.29--2.83) Medication ACE inhibitor[\*](#jvim14660-note-0004){ref-type="fn"} 1 (4.2%) 4 (26.7%) 11 (84.6%) Pimobendan[\*](#jvim14660-note-0004){ref-type="fn"} 1 (4.2%) 1 (6.7%) 11 (84.6%) Diuretics[\*](#jvim14660-note-0004){ref-type="fn"} 1 (4.2%) 2 (13.3%) 12 (92.3%) LVIDd inc% −0.21^a^ (−11.8-- 4.69) 15.5^b^ (5.05--36.3) 40.0^b^ (25.4--48.4) LVIDs inc% −21.0^a^ (−43.3--−6.75) −13.3^ab^ (−17.6--9.96) 9.32^b^ (−0.29--24.2) FS (%) 50.7^a^ (41.8--58.4) 50.8^a^ (42.1--55.4) 46.0^a^ (43.8--51.6) *E* (cm/s) 66.5^a^ (55.5--92.5) 97^b^ (84--118) 158^c^ (126--176) *A* (cm/s) 65^a^ (59.3--83.8) 94^b^ (82--110) 87^ab^ (51.5--121) *E*/*A* 1.01^a^ (0.86--1.30) 1.01^a^ (0.79--1.29) 1.92^b^ (1.37--2.78) *E* ~m~ (cm/s) 6.6^a^ (5.2--8.0) 8.2^ab^ (6.4--9.0) 9.0^b^ (8.4--10.3) *A* ~m~ (cm/s) 8.1^a^ (5.4--9.5) 7.8^a^ (7.1--10.7) 6.2^a^ (5.4--10.2) Sm (cm/s) 8.3^a^ (6.7--10.0) 8.9^a^ (8.2--10.4) 9.1^a^ (7.5--10.5) *E*/*E* ~m~ 11.1^a^ (8.8--13.2) 13.2^ab^ (10.2--16.7) 14.7^b^ (13.4--20.0) Total FAC (%) 59.7^a^ (55.4--65.4) 61.9^a^ (58.4--66.8) 44.6^b^ (35.5--50.9) Passive FAC (%) 27.8^a^ (20.0--35.2) 35.4^a^ (26.9--38.2) 27.7^a^ (25.6--31.6) Active FAC (%) 46.1^a^ (38.6--50.1) 44.5^a^ (37.9--49.8) 18.9^b^ (11.4--26.7) ε~S~ (%) 32.7^a^ (28.9--39.2) 35.6^a^ (31.7--41.9) 23.6^b^ (16.9--26.1) ε~E~ (%) 12.7^a^ (9.9--18.4) 16.8^a^ (13.3--22.7) 13.3^a^ (11.7--17.8) ε~A~ (%) 19.1^a^ (15.3--24.3) 19.6^a^ (14.1--21.4) 6.2^b^ (3.18--11.2) SR~S~ (/s)[†](#jvim14660-note-0005){ref-type="fn"} 1.97^a^ (1.70--2.43) 2.26^a^ (1.98--3.52) 1.72^a^ (1.39--1.91) SR~E~ (/s)[†](#jvim14660-note-0005){ref-type="fn"} 2.17^a^ (1.85--2.77) 2.80^a^ (1.88--3.87) 1.33^a^ (1.17--2.30) SR~A~ (/s)[†](#jvim14660-note-0005){ref-type="fn"} 3.09^a^ (2.14--3.94) 3.92^a^ (2.58--4.22) 1.04^b^ (0.75--1.49) *A*, late diastolic mitral inflow velocity; ACE, angiotensin‐converting enzyme; ACVIM, American College of Veterinary Internal Medicine; *A* ~m~, late diastolic velocity of the septal mitral annulus; *E*, peak early diastolic mitral inflow velocity; *E* ~m~, early diastolic velocity of the septal mitral annulus; ε~A~, left atrial longitudinal strain during atrial contraction; ε~E~, left atrial longitudinal strain during ventricular early diastole; ε~S~, left atrial longitudinal strain during ventricular systole; FAC, fractional area change; FS, fractional shortening; LA/Ao, left atrial to aortic root ratio; LVIDd inc%, percent increase in left ventricular diameter in diastole; LVIDs inc%, percent increase in left ventricular diameter in systole; SR~A~, left atrial longitudinal strain rate during atrial contraction; SR~E~, left atrial longitudinal strain rate during early ventricular diastole; SR~S~, left atrial longitudinal strain rate during ventricular systole; SR, strain rate. Continuous data are expressed as median (interquartile range). Values with different superscripted letters indicate statistically significant differences between groups. \*Indicates significant difference between groups (*P* \< .05). ^†^Evaluations of 8 dogs in the B1, 6 dogs in the B2, and 7 dogs in the C/D. John Wiley & Sons, Ltd The echocardiographic variables of each study group are also shown in Table [1](#jvim14660-tbl-0001){ref-type="table-wrap"}. For the conventional variables, LA/Ao and *E* wave velocity increased significantly with advancing stage. Although there were no significant differences in parameters indicating reservoir function (total FAC and ε~S~) or booster pump function (active FAC, ε~A~, and SR~A~) between B1 and B2 groups, these values were significantly lower in the C/D group (Figs [2](#jvim14660-fig-0002){ref-type="fig"} and [3](#jvim14660-fig-0003){ref-type="fig"}). SR could be measured in 8 of 24 dogs in the B1 group, in 6 of 15 in the B2 group, and 7 of 13 in the C/D group, because the exact peak of each phase could not be determined due to the corrugation of the SR curve in some dogs. {#jvim14660-fig-0002} {#jvim14660-fig-0003} Variables indicating LA strain were significantly correlated with age, heart rate, and some conventional echocardiographic variables (Table [2](#jvim14660-tbl-0002){ref-type="table-wrap"}). There were significant correlations between ε~S~ and total FAC (*R* = 0.748), ε~A~ (*R* = 0.763), and SR~A~ (*R* = 0.824), and between ε~A~ and active FAC (*R* = 0.711). The correlation between LA/Ao and ε~S~ was weak (*R* = −0.433); that between LA/Ao and ε~A~ was moderate (*R* = −0.684). ###### Correlates of echocardiographic variables ε~S~ ε~E~ ε~A~ ------------------------------------------------ -------- -------- --------- -------- -------- -------- Age .001 −0.397 0.041 −0.258 .013 −0.312 Body weight .35 0.86 .15 Heart rate .55 0.11 .024 −0.283 LA/Ao \<.001 −0.433 0.37 \<.001 −0.684 LVIDd inc% .007 −0.337 0.043 0.256 \<.001 −0.681 LVIDs inc% .003 −0.366 0.44 \<.001 −0.583 FS .40 0.38 .72 *E* .009 −0.326 0.093 \<.001 −0.637 *A* .28 0.036 0.265 .54 *E*/*A* \<.001 −0.409 0.94 \<.001 −0.520 *E* ~m~ .091 0.079 \<.001 −0.478 *A* ~m~ .014 0.317 0.45 .010 0.331 Sm .50 0.19 .67 *E*/*E* ~m~ .51 0.12 .95 Total FAC \<.001 0.748 0.030 0.302 \<.001 0.645 Passive FAC .27 \<0.001 0.561 .55 Active FAC \<.001 0.605 0.71 \<.001 0.711 ε~S~ \<0.001 0.517 \<.001 0.763 ε~E~ \<.001 0.517 .51 ε~A~ \<.001 0.763 0.51 SR~S~ [a](#jvim14660-note-0008){ref-type="fn"} .001 0.598 0.007 0.511 .073 SR~E~ [a](#jvim14660-note-0008){ref-type="fn"} \<.001 0.642 \<0.001 0.709 .073 SR~A~ [a](#jvim14660-note-0008){ref-type="fn"} \<.001 0.824 0.37 \<.001 0.632 *A*, late diastolic mitral inflow velocity; *A* ~m~, late diastolic velocity of the septal mitral annulus; *E*, peak early diastolic mitral inflow velocity; *E* ~m~, early diastolic velocity of the septal mitral annulus; ε~A~, left atrial longitudinal strain during atrial contraction; ε~E~, left atrial longitudinal strain during ventricular early diastole; ε~S~, left atrial longitudinal strain during ventricular systole; FAC, fractional area change; FS, fractional shortening; LA/Ao, left atrial to aortic root ratio; LVIDd inc%, percent increase in left ventricular diameter in diastole; LVIDs inc%, percent increase in left ventricular diameter in systole; SR~A~, left atrial longitudinal strain rate during atrial contraction; SR~E~, left atrial longitudinal strain rate during early ventricular diastole; SR~S~, left atrial longitudinal strain rate during ventricular systole; MMVD, myxomatous mitral valve disease; SR, strain rate. This study included 52 client‐owned dogs with MMVD; 24 dogs were classified as stage B1, 15 as stage B2, and 13 as stage C/D. ^a^Evaluations of 8 dogs in the B1, 6 dogs in the B2, and 7 dogs in the C/D. John Wiley & Sons, Ltd The ROC curves and the corresponding AUC were calculated to facilitate comparison of the accuracy of the echocardiographic variables in identifying dogs in stage C/D. As shown in Table [3](#jvim14660-tbl-0003){ref-type="table-wrap"}, the highest accuracy was obtained with ε~A~, followed by *E*, active FAC, and LA/Ao. ###### AUC and optimal diagnostic cutoffs between stages B1/B2 and C/D Variable Cutoff AUC \[95% CI\] Sensitivity Specificity AIC ---------------------------------------------------- -------- ---------------------- ------------- ------------- ------ ε~A~ (%) 11.8 0.990 \[0.94--1.00\] 1.00 0.96 16.3 *E* (cm/s) 114 0.982 \[0.91--1.00\] 1.00 0.90 19.9 Active FAC (%) 26.8 0.967 \[0.89--0.99\] 1.00 0.90 24.9 LA/Ao 2.16 0.963 \[0.83--0.99\] 0.92 0.93 26.8 SR~A~ (/s)[a](#jvim14660-note-0011){ref-type="fn"} 1.49 0.959 \[0.72--1.00\] 0.86 1.00 14.2 ε~S~ (%) 28.4 0.941 \[0.87--0.99\] 0.92 0.84 33.0 Total FAC (%) 46.8 0.934 \[0.84--0.98\] 1.00 0.80 32.5 LVIDd inc% 21.3 0.899 \[0.78--0.96\] 1.00 0.79 40.7 LVIDs inc% −6.12 0.852 \[0.69--0.94\] 0.92 0.69 45.0 *E*/*A* 1.53 0.842 \[0.61--0.95\] 0.77 0.92 39.7 *E* ~m~ (cm/s) 8.3 0.840 \[0.66--0.93\] 0.92 0.70 45.5 *A*, late diastolic mitral inflow velocity; AIC, Akaike\'s information criteria; AUC, area under the receiver operating characteristic curve; *E*, peak early diastolic mitral inflow velocity; *E* ~m~, early diastolic velocity of the septal mitral annulus; ε~A~, left atrial longitudinal strain during atrial contraction; ε~S~, left atrial longitudinal strain during ventricular systole; FAC, fractional area change; LA/Ao, left atrial to aortic root ratio; LVIDd inc%, percent increase in left ventricular diameter in diastole; LVIDs inc%, percent increase in left ventricular diameter in systole; SR~A~, left atrial longitudinal strain rate during atrial contraction; MMVD, myxomatous mitral valve disease; SR, strain rate. This study included 52 client‐owned dogs with MMVD; 24 dogs were classified as stage B1, 15 as stage B2, and 13 as stage C/D. ^a^Evaluations of 8 dogs in the B1, 6 dogs in the B2, and 7 dogs in the C/D. John Wiley & Sons, Ltd Based on univariate logistic regression analysis, 5 variables, including ε~A~, *E*, LA/Ao, LVIDd inc%, and *E*/*A*, were selected for multivariate logistic regression analysis. Subsequently, ε~A~ and *E* were identified as independently correlating with stage C/D (Table [4](#jvim14660-tbl-0004){ref-type="table-wrap"}). ###### Binary logistic regression analysis between stages B1/B2 and C/D Variable Univariate Multivariate ------------ ------------ -------------- --------- ε~A~ (%) 0.24 0.03--0.60 \<.0001 *E* (cm/s) 1.17 1.07--1.40 \<.0001 LA/Ao (%) 1.07 1.04--1.12 -- LVIDd inc% 1.09 1.05--1.16 -- *E*/*A* 1.42 1.18--1.84 -- *A*, late diastolic mitral inflow velocity; *E*, peak early diastolic mitral inflow velocity; ε~A~, left atrial longitudinal strain during atrial contraction; LA/Ao, left atrial to aortic root ratio; LVIDd inc%, percent increase in left ventricular diameter in diastole; MMVD, myxomatous mitral valve disease. This study included 52 client‐owned dogs with MMVD; 24 dogs were classified as stage B1, 15 as stage B2, and 13 as stage C/D. John Wiley & Sons, Ltd Discussion {#jvim14660-sec-0013} ========== The present study demonstrates that LA longitudinal strain during atrial contraction (indicator of booster pump function) and during ventricular systole (indicator of reservoir function) was lower in dogs with CHF due to MMVD than in those without CHF. Although the LA enlarges in accordance with the ACVIM stage, LA strain was maintained until the onset of CHF. Impaired LA booster pump function evaluated with strain imaging is related to the presence of heart failure symptoms in hypertrophic cardiomyopathy[23](#jvim14660-bib-0023){ref-type="ref"} and aortic stenosis[24](#jvim14660-bib-0024){ref-type="ref"} in humans. Whether the reduced LA booster pump function in heart failure patients results from impairment of LA intrinsic contraction or from LA afterload mismatch (elevated LV end‐diastolic pressure) remains unclear. LA booster pump function is determined by LV compliance, LV end‐diastolic pressure, and intrinsic LA contractility. LA contraction augments LV stroke volume by approximately 20--30% in normal human subjects and substantially more in the presence of impaired LV relaxation.[25](#jvim14660-bib-0025){ref-type="ref"} In atrial fibrillation, LA booster pump function is lost, diminishing the LA stroke volume by as much as 20%. Consequently, blood accumulates in the LA, increasing LA pressure and in turn increasing pressure within the pulmonary veins. This process promotes fluid shift, initially from the intravascular to interstitial space and then to the alveoli, culminating in pulmonary congestion and edema.[26](#jvim14660-bib-0026){ref-type="ref"} In dogs with MR, regurgitant jet flow and/or increased LA pressure causes LA dilation and degeneration, including interstitial fibrosis, fatty replacement, and chronic inflammation.[27](#jvim14660-bib-0027){ref-type="ref"}, [28](#jvim14660-bib-0028){ref-type="ref"}, [29](#jvim14660-bib-0029){ref-type="ref"} These pathological changes in the LA may contribute to decreased LA intrinsic booster pump function, leading to elevation of LA pressure and pulmonary edema. LA booster pump dysfunction is attributable more to LA afterload mismatch than to LA intrinsic contraction abnormalities.[30](#jvim14660-bib-0030){ref-type="ref"}, [31](#jvim14660-bib-0031){ref-type="ref"} In chronically decompensated MR, LV stiffness and end‐diastolic pressure rise with LA pressure. This situation contributes to decreased transmitral flow during LA contraction (restrictive pattern). In human heart failure patients, reduced transmitral A wave velocity is recovered after reduction in LV filling pressure with optimal treatment of heart failure.[30](#jvim14660-bib-0030){ref-type="ref"} This reversibility of mitral A flow suggests that LA dysfunction results from LA afterload mismatch rather than intrinsic LA abnormality.[30](#jvim14660-bib-0030){ref-type="ref"} Another study demonstrated that LV filling during LA contraction was significantly reduced in a dog model of ischemic heart disease, while LA systolic shortening during atrial contraction was not changed.[31](#jvim14660-bib-0031){ref-type="ref"} The preserved LA systolic shortening resulted from LA afterload mismatch, which was closely linked to an increase in the volume of blood ejected from the LA backward into the pulmonary veins. These results indicate that the morphological assessment of LA function is less dependent on afterload than transmitral blood flow analysis. Further studies are needed to investigate to what degree the variables of STE‐derived strain imaging are dependent on afterload. Active FAC, a measure of LA booster pump function, was significantly lower in CHF dogs than in those without CHF in the present study, as in our previous study.[11](#jvim14660-bib-0011){ref-type="ref"} There is a significant correlation between ε~A~ and active FAC, but ε~A~ has a higher predictive power for CHF. LA FAC derived from 2D echocardiography is one of the commonly used measures of LA function in dogs.[10](#jvim14660-bib-0010){ref-type="ref"}, [11](#jvim14660-bib-0011){ref-type="ref"}, [21](#jvim14660-bib-0021){ref-type="ref"}, [32](#jvim14660-bib-0032){ref-type="ref"}, [33](#jvim14660-bib-0033){ref-type="ref"} However, the main limitation of LA FAC is the inability to distinguish between the increase in LA function due to a larger volume of blood received/ejected and a real increase in intrinsic LA compliance/contractility.[6](#jvim14660-bib-0006){ref-type="ref"} We previously demonstrated that active FAC is increased during volume loading in normal dogs.[34](#jvim14660-bib-0034){ref-type="ref"} Although the degree of preload dependency of ε~A~ remains unclear, SR~A~ is confirmed to be a relatively preload‐independent parameter.[35](#jvim14660-bib-0035){ref-type="ref"} The difference in preload dependency may contribute to the superiority of ε~A~ over active FAC in the present study. SR~A~ is another measure of booster pump function, and its superiority to ε~A~ as a severity indicator has been demonstrated in human patients with heart failure and preserved left ventricular ejection fraction.[22](#jvim14660-bib-0022){ref-type="ref"} The present study also showed the utility of SR~A~, but SR~A~ could not be obtained in more than half of included dogs; this was because the exact peak could not be determined due to the corrugation of the SR curve in some dogs, especially in early cases. LA strain rate has been less studied because noises in LA strain rate curves make it difficult to obtain consistent data in human patients with various cardiovascular disease.[36](#jvim14660-bib-0036){ref-type="ref"} In contrast, a previous study demonstrated the feasibility of evaluating LA strain and SR with STE in most healthy dogs (83%) with median heart rate 118 beats/min.[17](#jvim14660-bib-0017){ref-type="ref"} This discrepancy might result from differences between the studies in observers, ultrasound machines, software, or dogs included. Including smaller dogs with MMVD with higher heart rates might have lowered the feasibility of strain rate in the present study. Higher frame rate and/or zooming might be needed to get appropriate image. Further studies are needed to examine the exact cause of this failure and to improve the feasibility of this technique. A combined model with ε~A~ and *E* wave was the best predictor of CHF in the present study. *E* wave velocity is an indicator of peak LA pressure, representing LA preload. Increased preload augments LA contraction, according to the Frank--Starling law.[37](#jvim14660-bib-0037){ref-type="ref"} In other words, the lower the preload, the weaker the LA contraction. In the present study, some dogs in stage B1 without *E* wave elevation showed low ε~A~ similar to those in C/D. Therefore, this combined model may diminish the drawback of ε~A~ resulting from its preload dependency. LA reservoir function as assessed with ε~S~ and total FAC was also a useful predictor of CHF in the present study. LA reservoir function is modulated by LV contraction, RV systolic pressure transmitted through the pulmonary circulation, and LA relaxation and stiffness.[6](#jvim14660-bib-0006){ref-type="ref"} Impaired LA reservoir strain in human patients is associated with MR severity and worse prognosis.[14](#jvim14660-bib-0014){ref-type="ref"}, [15](#jvim14660-bib-0015){ref-type="ref"}, [38](#jvim14660-bib-0038){ref-type="ref"} The impairment of LA reservoir function in MR may be caused by LA fibrosis due to severe LA dilation and RV dysfunction due to pulmonary hypertension or LV enlargement. Although the utility of ε~S~ as predictor of CHF was lower than that of ε~A~ in this study, the feasibility of using ε~S~ in dogs with atrial fibrillation that lack booster pump function is a major advantage over ε~A~. Impaired LA reservoir function was demonstrated by SR~S~ in human patients with persistent atrial fibrillation who lacked booster pump function,[39](#jvim14660-bib-0039){ref-type="ref"} and ε~S~ was the only echocardiographic variable associated with CHADS2 score estimating the risk of stroke in atrial fibrillation.[40](#jvim14660-bib-0040){ref-type="ref"} LA conduit function is mainly modulated by LV diastolic properties (LV relaxation and early diastolic pressure).[6](#jvim14660-bib-0006){ref-type="ref"} Both *E* and *E*′ velocity, conventional variables of LV diastolic function, were higher in dogs with CHF in the present study. In MR, the decrease in LV relaxation coexists with the increase in the pressure gradient between LA and LV. The more severe the MR, the greater the LA pressure becomes, leading to the increase in *E* wave velocity. Although *E*′ is considered as a preload‐independent index of LV relaxation,[41](#jvim14660-bib-0041){ref-type="ref"} *E*′ is positively correlated with MR volume in human patients,[42](#jvim14660-bib-0042){ref-type="ref"} and was higher in MR dogs with CHF than in those without CHF.[43](#jvim14660-bib-0043){ref-type="ref"} LA conduit function as assessed with ε~E~ and passive FAC was not a statistically useful predictor of CHF in the present study; however, strain‐derived variables of LA conduit function were decreased with MR severity in humans.[38](#jvim14660-bib-0038){ref-type="ref"}, [44](#jvim14660-bib-0044){ref-type="ref"} These results may indicate that LA strain could be less load‐dependent and more reliable for evaluation of LV diastolic function than conventional Doppler‐derived variables. Some limitations of this study must be considered. First, no invasive assessment of LA mechanical properties or afterload was performed to confirm the determinations of reduced LA function. Such assessments must be the focus of further specific studies. Second, the study lacked a control group. Investigation into the reference value of LA function in stage A with the same ultrasound machine and software used in this study is needed to strengthen the clinical importance of the LA strain in dogs with MMVD. Third, the number of dogs studied was small, especially strain rate, rendering the study underpowered to detect differences between groups. Fourth, dogs in stages C and D were united into 1 group because of the small sample size, making it impossible to analyze differences between refractory and nonrefractory dogs. Fifth, the blood pressure was not evaluated in all dogs. Sixth, strain measurements in the present study were obtained using the software originally designed for LV. Measurements obtained by other ultrasound system or software might be incomparable with those of the present study. Last, it is possible that medication use influenced the echocardiographic indices. In conclusion, the results of the present study indicate that LA longitudinal strain was significantly lower in dogs in the advanced stages of MMVD. LA longitudinal strain during atrial contraction was the best predictor of the presence or history of CHF. LA strain assessed on conventional echocardiography and radiography could add meaningful information to confirm the presence of CHF in the clinical setting. Further studies are needed to determine the clinical implications of these findings for treatment decisions and/or prognosis determination. *Grant support*: This study was partially supported by a Grant‐in‐Aid for Scientific Research (No. 25850203) from the Japanese Society for the Promotion of Science (K.N.). *Conflict of Interest Declaration*: Authors declare no conflict of interest. *Off‐label Antimicrobial Declaration*: Authors declare no off‐label use of antimicrobials. Artida, Toshiba Medical Systems Corp, Tochigi, Japan PST‐50BT, Toshiba Medical Systems Corp UltraExtend V3.10, Toshiba Medical Systems Corp JMP Pro, 11.2.0, SAS Institute Inc, Cary, NC IBM^®^ SPSS^®^ Statistics, version 21; IBM Corp, Chicago, IL

Introduction {#Sec1} ============ A compelling body of evidence, summarized in several reviews, has suggested that women with breast cancer gain weight during adjuvant or neo-adjuvant chemotherapy \[[@CR1]--[@CR7]\]. A meta-analysis from our group showed that women with breast cancer gain on average 2.7 kg body weight during chemotherapy \[[@CR8]\], but also that weight gain was most pronounced in patients receiving cyclophosphamide, methotrexate and 5-fluorouracil (CMF) regimens---regimens that are nowadays less often used for treatment of breast cancer. Weight gain seemed less pronounced in patients treated with more recent types of chemotherapy \[[@CR8]\]. Only few studies stratified their results for menopausal status or BMI \[[@CR8]\], which impeded the ability to assess whether these factors affected any changes in body weight. Changes in body weight during and after chemotherapy have been characterized by an increase in fat mass with a stable lean mass or loss of lean mass. Similar changes in body composition have been reported in body weight stable patients \[[@CR1]--[@CR4], [@CR9]\]. Weight gain during chemotherapy can negatively impact self-perception and quality of life \[[@CR10]\]. Changes in body weight and/or body composition are of potential clinical relevance as it has been suggested that increases in weight, increases in fat mass and/or decreases in lean mass are associated with cancer recurrence and mortality \[[@CR3], [@CR11]\], although data on this are not fully consistent \[[@CR12]--[@CR15]\]. Moreover, recent studies suggest that body composition is importantly associated with toxicity-induced modifications of treatment \[[@CR16]--[@CR18]\], which warrants further research into how body composition changes over time during chemotherapy. Earlier studies that investigated changes in body composition during chemotherapy in breast cancer patients generally included a small study sample of 8 to 76 patients \[[@CR5], [@CR6], [@CR19]--[@CR28]\]. More importantly, most of these studies did not compare the changes in body weight and body composition in breast cancer patients to a comparison group of women without cancer \[[@CR5], [@CR6], [@CR19]--[@CR21], [@CR23]--[@CR29]\]. The only study that did include a comparison group concluded that women with breast cancer did not differentially change in body weight from before chemotherapy to 6 months after chemotherapy as compared with a comparison group of women who were measured over the same interval. Comparisons with regard to body composition were however not made \[[@CR22]\]. Therefore, it is still unclear whether there are differential changes in body composition in breast cancer patients compared with a comparison group of women not diagnosed with cancer. The objective of this study was to describe changes in body weight and body composition over time in breast cancer patients (stage I--IIIB) from start of chemotherapy until 6 months after chemotherapy compared with changes over a similar time frame in an age-matched comparison group of women without cancer. Materials and methods {#Sec2} ===================== Study population {#Sec3} ---------------- The analyses were done using data of the observational, multicentre COBRA study \[[@CR30]\]. The COBRA study recruited women with breast cancer receiving chemotherapy and women without cancer. The study was designed to compare changes in body composition among breast cancer patients with changes in body composition among women without cancer \[[@CR30]\]. Women with newly diagnosed, stage I--IIIB, operable breast cancer, who were scheduled for second- or third-generation adjuvant or neo-adjuvant chemotherapy were eligible for the study and were recruited via the staff of 11 participating hospitals in the Netherlands prior to commencement of chemotherapy. Participants in the comparison group were recruited via patients; patients were asked to distribute envelopes with study information to female friends, acquaintances and colleagues who were of similar age (± 2 years); women in the comparison group could not be family members of the patients. All study participants needed to be at least 18 years old and able to communicate in Dutch. Exclusion criteria for both groups were history of another cancer, history of treatment with chemotherapy, (intended) pregnancy, dementia, or other mental conditions that made it impossible to comply with the study procedures. The study was approved by the Medical Ethical Committee of Wageningen University & Research, Wageningen, the Netherlands. All participants provided written informed consent. Study design {#Sec4} ------------ In all participants, measurements took place at three time points T1, T2, and T3. For the patient group, these time points were as follows: before start of chemotherapy or during the first cycle of chemotherapy (T1), shortly after chemotherapy, which was within 1 to 3 weeks after completion of the last cycle of chemotherapy (T2), and 6 months after chemotherapy (T3). For the comparison group, these time points were baseline (T1), 6 months after baseline (T2), and 12 months after baseline (T3). Body weight, body composition {#Sec5} ----------------------------- At T1, T2, and T3, participants were invited to their own respective clinic for assessment of body composition. Body weight and body composition were assessed using dual-energy X-ray absorptiometry (DEXA) scan at those three time points. DEXA is a commonly used technique to estimate lean mass, fat mass, and bone mineral content of a person and is based on the difference in attenuation of X-rays between those different tissues \[[@CR31], [@CR32]\]. It has been used in various populations including cancer patients \[[@CR5], [@CR21]--[@CR23], [@CR33]\] to detect changes in body composition over time. As there can be differences between different scanners \[[@CR34]\], during the study, participants were always measured in the same clinic using the same scanner by trained technicians using a total body scan protocol. Body weight (kg), total body fat mass (kg), and total lean mass (kg) were obtained from the total body DEXA scan. In addition, lean mass (kg) of the arms, legs, and torso were derived from DEXA. Demographic, personal, and medical information {#Sec6} ---------------------------------------------- At T1, we received an information package with surveys that contained questions about demographic information, body height, age, smoking status, educational level, and menopausal status. Patients who were perimenopausal were categorized as premenopausal. BMI was calculated based on self-reported body height and on body weight as measured from DEXA scan. Information on tumour stage (pTNM), tumour characteristics, and treatment was obtained from reviewing patients' medical records using a standardized form. Data analyses {#Sec7} ------------- Population characteristics were described as median with an interquartile range (IQR) or counts and percentages separately for the patient and comparison group. Differences in body weight and body composition trajectories over time for the patient and comparison group were analysed using linear mixed models, with time, group and their interaction term as fixed factors and subjects as random factors in the model. Linear mixed models estimated marginal means and standard errors, presented as mean ± SE. In all models, a random intercept model was used with an unstructured covariance structure. Using a top-down model fitting procedure, the appropriate covariates were chosen (age, educational level, fat mass, lean mass). In a meta-analysis \[[@CR8]\], we identified menopausal status and baseline BMI as potential effect modifiers influencing changes in body weight and body composition over time. Therefore, we performed exploratory analyses stratifying results by menopausal status (pre- vs post-menopausal) and stratifying by BMI (based on the cutoff point for overweight: 25 kg/m^2^). In 60 participants, we measured body weight using both a calibrated scale and DEXA scan, at all three time points. We compared whether changes in body weight over time differed between these two methods, using the Wilcoxon signed rank test. In these 60 patients, change in body weight over time from T1 to T3 was 0.1 kg (IQR − 1.2, 1.8) when assessed with a scale and was 0.2 kg (IQR − 1.2, 1.9) when assessed by DEXA. These values were not statistically significant different (*p* = 0.57). In all analyses, a two-sided *p* value \< 0.05 was considered statistically significant. Statistical analyses were performed in SAS 9.4 (SAS Institute, Cary, NC). Results {#Sec8} ======= At baseline (T1), the patient and comparison group were similar in age, and menopausal status, while women in the comparison group tended to be higher educated and were less often current smokers, see Table [1](#Tab1){ref-type="table"}. Total time between T1 and T3 was on average 323 days (SD 39 days) for patients and 376 days (SD 26 days) for the comparison group. For the patients, the average time between the first and the second measurement (T1 and T2) was 143 days (SD 31 days), and for the comparison group this was 194 days (SD 27 days). Between T2 and T3, the average time was 179 days (SD 24 days) for the patients, and 188 days (SD 24 days) for the comparison group.Table 1Baseline characteristics of breast cancer patients and of women without cancer, presented as median (IQR) or *n* (%)Patient group (*n* = 181)Comparison group (*n* = 180)DemographicsAge, years (median, IQR)51.8 (46.7; 58.9)53.3 (46.7; 62.3)Menopausal status (*n*, %)\*PremenopausalPostmenopausal103 (57.5)76 (42.5)90 (50.3)89 (49.7)Education level (*n*, %)LowMediumHighMissing15 (8.3)56 (30.9)99 (54.7)11 (6.1)13 (7.2)48 (26.7)116 (64.4)3 (1.7)Smoking status (*n*, %)CurrentFormerNeverMissing27 (14.9)71 (39.2)72 (39.8)11 (6.1)15 (8.3)81 (45)81 (45.0)3 (1.7)Clinical factorsStage of diseaseI45 (24.9)II110 (60.8)III26 (14.4)ChemotherapyAdjuvant treatment117 (64.6)Neo-adjuvant treatment64 (35.4)Type of chemotherapyCombined regime81 (44.8)Sequential regime100 (55.3)Number of cycles of chemotherapy received6 or less128 (70.7)More than 653 (29.3)Hormone receptor statusER+143 (79.0)ER−38 (21.0)PR+121 (66.9)PR−60 (33.2)Her2+36 (19.9_Her2−145 (80.1)Anthropometry, body composition \*Height, cm (median, IQR)168.0 (164.0; 173.0)169.0 (165.0; 172.5)Weight, kg (median, IQR)70.5 (63.7; 81.8)69.9 (62.7; 77.5)BMI, kg/m^2^ (median, IQR)25.5 (22.5; 29.1)24.2 (22.4; 27.2)Total fat mass, kg (median, IQR)25.9 (20.2; 34.4)24.8 (19.9; 30.7)Total fat mass, % (median, IQR)36.6 (31.2; 42.1)35.7 (31.1; 40.7)Arm fat, kg (median, IQR)2.6 (2.0; 3.6)2.3 (1.8; 2.9)Leg fat, kg (median, IQR)9.5 (8.1; 11.9)8.4 (6.8; 11.3)Trunk fat, kg (median IQR)12.0 (8.8; 17.0)11.8 (8.9; 15.0)Total lean mass, kg (median, IQR)43.1 (39.3; 47.1)43.2 (39.8; 46.2)Arm lean mass, kg (median, IQR)4.3 (4.0; 4.7)4.3 (3.9; 4.7)Leg lean mass, kg (median IQR)13.7 (12.6; 14.5813.8 (12.7; 15.1)Trunk lean mass, kg (median, IQR)22.1 (19.4; 24.1)21.6 (20.1; 23.1)\*Data missing for *n* = 2 patients and *n* = 1 women in the comparison group For the patient group, we had data of body weight at T1 and T3 for 163 women: of these, 125 patients (77%) were weight stable between T1 and T3, defined as a change in body weight of no more than 5%. For the comparison group, we had complete data on body weight for 162 women: of those, 142 (88%) were weight-stable. Among patients, 25 women (15%) gained at least 5% in body weight over the period from baseline to 6 months after the end of chemo (T3); among the comparison group, 13 women (8%) gained at least 5% body weight between baseline and T3. Between baseline and T3, 13 patients (8%) lost at least 5 kg, while among the comparison group this was 7 women (4%). Baseline (T1) body weight and body composition were not statistically significantly different between the patient and comparison group according to the mixed model analyses, while baseline BMI was statistically significantly higher for patients (25.6 kg/m^2^ ± 0.23 (SE)) than for the comparison group (25.0 kg/m^2^ ± 0.25), see Fig. [1](#Fig1){ref-type="fig"} and Table [2](#Tab2){ref-type="table"}. In the comparison group, no statistically significant changes in body weight, BMI, or body composition were observed over time.Fig. 1Trajectories of body weight and body composition for breast cancer patients (solid line) receiving chemotherapy and for a comparison group of women not diagnosed with cancer (dashed line). Presented are estimated marginal means ± SE from linear mixed models. Body weight **a** and BMI **b** were adjusted for age, education level, baseline lean mass, and baseline fat mass. Fat mass **c** was adjusted for age and education level, and lean mass in kg at baseline. Lean mass **d** was adjusted for age and education level, and fat mass in kg at baseline. T1 is before chemotherapy for patients and is baseline for comparison group; T2 is shortly after chemotherapy for patients and is 6 months after baseline for comparison group; T3 is half a year after chemotherapy for patients and is 1 year after baseline for comparison group. \* indicates significant differences at this time point between the two groups at *p* \< 0.05. \*\* indicates a significant difference over time in the patient group at *p* \< 0.05. \*\*\* indicates a significant difference over time between the patient group and comparison group at *p* \< 0.05Table 2Estimated marginal means for body weight, BMI, fat mass, and lean mass of breast cancer patients and of women without cancer, at three different time points (T1, T2, T3)Patient GroupComparison GroupBody weight in kg*n*Mean\*SE*n*Mean\*SET117972.10.417971.20.5T216673.30.411171.40.5T316373.00.416271.70.5BMI in kg/m^2^T117925.60.217925.00.2T216626.00.211125.10.3T316325.90.216225.20.2Fat mass in kgT117830.00.917828.91.0T216630.30.911129.21.0T316331.00.916229.41.0Lean mass in kgT117843.10.517843.10.6T216644.00.511143.00.6T316343.10.516243.10.6\*Estimated marginal means and SE of the mixed models as presented in Fig. [1](#Fig1){ref-type="fig"} Changes in body weight over time differed between patients and women without cancer (*p* interaction = 0.03 Fig. [1a](#Fig1){ref-type="fig"}). Breast cancer patients significantly increased in body weight from baseline (T1) to shortly after chemotherapy (T2) (from 72.1 kg ± 0.4 kg at T1, to 73.3 kg ± 0.4 kg at T2). Six months after chemotherapy (T3), body weight slightly decreased to 73.0 kg ± 0.4 kg. Compared with women in the comparison group, patients had a significantly higher body weight shortly after chemotherapy (T2) and at 6 months after chemotherapy (T3). Similar results were found for BMI (Fig. [1b](#Fig1){ref-type="fig"}). Fat mass did not differentially change over time between patients and women without cancer (*p* interaction = 0.19; Fig. [1c](#Fig1){ref-type="fig"}). Change in lean mass over time in patients differed from change in lean mass over time in the comparison group (*p* for interaction = \< 0.01 (Fig. [1d](#Fig1){ref-type="fig"}). In the patient group, lean mass significantly increased from 43.1 kg ± 0.5 kg at baseline (T1) to 44.0 kg ± 0.5 kg shortly after chemotherapy (T2), but returned to 43.1 kg ± 0.5 kg 6 months after chemotherapy (T3). Yet, at the three time points, lean mass did not differ statistically significantly between the patients and the women in the comparison group. In exploratory analyses, we stratified the results for body weight and body composition by menopausal status (Fig. [2](#Fig2){ref-type="fig"} and Table [3](#Tab3){ref-type="table"}). The change in body weight and body composition over time seemed most pronounced in premenopausal patients. Premenopausal patients experienced a modest gradual increase in body weight, while postmenopausal patients were relatively weight stable. In premenopausal patients, fat mass gradually increased over time, while lean mass appeared stable. Postmenopausal patients showed a slight increase in lean mass and decrease in fat mass between T1 and T2, but these returned to baseline values half a year after chemotherapy (T3). We additionally stratified our figures by BMI (above or below BMI of 25 kg/m^2^). There were no apparent differential changes over time in body weight or body composition between patients of lower versus higher BMI (data not shown).Fig. 2Body weight **a**, BMI **b**, fat mass **c**, and lean mass **d** trajectories for breast cancer patients receiving chemotherapy (solid lines) and for a comparison group of women without cancer (dashed lines); results are stratified by menopausal status. Presented are estimated marginal means ± SE from linear mixed models. Body weight **a** and BMI **b** were adjusted for education level, baseline lean mass, and baseline fat mass. Fat mass **c** was adjusted for: education level, and lean mass in kg at baseline. Lean mass **d** was adjusted for: education level, and fat mass in kg at baseline. T1 is before chemotherapy for patients and is baseline for comparison group; T2 is shortly after chemotherapy for patients and is 6 months after baseline for comparison group; T3 is half a year after chemotherapy for patients and is 1 year after baseline for comparison groupTable 3Estimated marginal means for body weight, BMI, fat mass, and lean mass of breast cancer patients and of women without cancer, at three different time points (T1, T2, T3) stratified by menopausal statusPatient groupComparison groupPremenopausalPostmenopausalPremenopausalPostmenopausal*n*Mean\*SE*n*Mean\*SE*n*Mean\*SE*n*Mean\*SEBody weight in kgT110271.80.67572.60.69070.70.68971.80.7T29473.50.67173.10.65671.30.75571.60.7T39473.40.66872.50.67771.50.685720.7BMI in kg/m^2^T110225.20.37526.10.39024.60.38925.40.3T29425.80.37126.30.35624.80.35525.40.4T39425.70.36826.20.37724.80.38525.50.3Fat mass in kgT1101291.27531.21.39029.31.38928.41.3T29429.91.27130.81.35629.71.35528.51.4T39430.61.26831.31.37729.91.38528.71.3Lean massT110144.30.77541.70.89043.80.88942.50.8T29445.10.77142.70.85643.90.85542.20.8T39444.50.76841.60.87743.80.88542.40.8\*Estimated marginal means of the mixed models as presented in Fig. [2](#Fig2){ref-type="fig"} Discussion {#Sec9} ========== In this study, we compared changes in body weight and body composition during and after chemotherapy in breast cancer patients with changes in body weight and body composition over a comparable timeframe in women without cancer. Our results suggest that weight gain of more than 5% over a period of about a year occurred slightly more often in patients with breast cancer undergoing chemotherapy (15% of patients), than among women without cancer (8%). We showed that weight trajectories differed for breast cancer patients and women without cancer, but that differences were very modest. The differential change in body weight between patients and women in the comparison group was merely explained by an increase in lean mass observed in patients shortly after the end of chemotherapy; lean mass returned to pre-chemotherapy values 6 months after chemotherapy. In women without cancer, body weight and body composition remained stable during the study. In our study, we observed an overall increase of 1.2 kg in body weight in the patients during the period of chemotherapy. This is in line with the results of our meta-analysis where we found a mean weight gain of 1.4 kg in women receiving newer chemotherapy regimens \[[@CR8]\]. Newer regimens are regimens containing anthracyclines and/or taxanes, whereas older regimens typically include CMF (cyclophosphamide, methotrexate, and fluorouracil): none of the women in our study received CMF, and all of them received a combination of anthracyclines with or without taxanes. In our study, half a year after chemotherapy (T3), body weight among patients had decreased again, and although body weight at T3 was still higher than at baseline, it was not statistically significantly different from body weight in women in the comparison group. Our findings on body weight are in line with findings of the only other previous study where changes in body weight were compared with a comparison group of women without breast cancer \[[@CR22]\]. Although that study observed a slight increase in body weight among patients after the end of chemotherapy, changes were not different from the comparison group \[[@CR22]\]; therefore, both that and our own study concluded that patients do not show significant changes in weight during the first year of their treatment compared with a comparison group of women without cancer. Our study suggested that lean mass initially increased during chemotherapy, but returned to baseline values in the 6 months after chemotherapy. It is plausible that this initial increase in lean mass did not represent an actual increase in muscle mass, but merely an increase in body fluid. The findings from Pedersen et al. \[[@CR25]\] support this hypothesis, as they showed an increase in total body water, as predicted by bioelectrical impedance analysis, 6 months after start of chemotherapy in breast cancer patients, which returned to baseline values after 12 months \[[@CR25]\]. A possible explanation for fluid retention could be the use of a chemotherapy regime including docetaxel, since docetaxel is known to be associated with fluid retention \[[@CR35]\]. In our exploratory analyses, the increases in body weight and fat mass seemed most pronounced in premenopausal patients. This is consistent with findings from others \[[@CR1], [@CR36], [@CR37]\]. Ovarian failure and alterations in sex hormone production as a result of chemotherapy potentially mimic normal menopausal-related physiological changes, which result in fat accumulation and decreases in lean mass \[[@CR38], [@CR39]\]. In the interpretation of our findings, it is important to consider that we have self-reported information on menopausal status. Especially for women on anti-conceptives, this may not accurately reflect the true menopausal status \[[@CR40]\]. Patients were recruited from 11 hospitals in the Netherlands. We opened the study in 11 centres to enable successful completion of the study in a reasonable timeframe. Not all clinics started at the same time; as a result, recruitment rate varied largely between sites. Body weight and body composition were assessed using a DEXA scan. Although different scanner types were used in different clinics, each participant was always measured on the same scanner in the same hospital. In our study, we used body weight as assessed with the DEXA scan, since it turned out to be logistically impossible to use validated scales to assess body weight in each recruiting centre. In a subgroup of 60 participants, we measured body weight by scale and DEXA scan on all three time points and showed that changes in body weight over time were not different based on scale vs DEXA measurement. The median BMI of patients was in the 'overweight' category. Furthermore, there was a group of patients gaining more than 5% of body weight during and after the period of chemotherapy. Given that being overweight or obese and gain in body weight are associated with higher risk of several diseases and co-morbidities \[[@CR41]\], physicians may want to address this with patients, especially since physicians have a powerful social role in facilitating behaviour change. In conclusion, we observed that over the period from start of chemotherapy until 6 months after the end of chemotherapy mean changes in body weight and body composition are minimal in women with breast cancer, and that these changes do not differ substantially from those in women of similar age without breast cancer. Therefore, this study does not confirm findings from other studies where fat mass and/or lean mass changed substantially during chemotherapy in breast cancer patients. During chemotherapy, we observed slight increases in body weight and lean mass but no change in fat mass; it is plausible that these increases merely reflect an increase in body fluid, as 6 months after the end of chemotherapy differences between patients and women without cancer were no longer present. **Publisher's note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. We thank all participants for their time to participate in the study. Furthermore, we thank the staff of the following hospitals that helped recruiting the participants: Ziekenhuis Gelderse Vallei, Maxima Medisch Centrum, Reinier de Graaf Ziekenhuis, Onze Lieve Vrouwen Gasthuis, Amphia Ziekenhuis, Canisius Wilhelmina Ziekenhuis, Radboud Universitair Medisch Centrum, Alexander Monro Ziekenhuis, St. Antonius Ziekenhuis, St. Anna Ziekenhuis, and Flevoziekenhuis. This study was funded by the Dutch Cancer Society (grant numbers UW2011-4987 and UW2011-5268). The funder had no role in the preparation of this manuscript. The study was approved by the university's institutional review board. Written informed consent was obtained from all participants. The authors declare that they have no conflict of interest.

Background {#Sec1} ========== Grapevine is one of the most important economic fruit crops worldwide. Strategies for improving quality and disease resistance are a key focus of grapevine cultivation. The whole grape berry, including the skin, pulp and seeds, contains melatonin (N-acetyl-5-methoxytryptamine) \[[@CR1]\]. Melatonin is a signaling molecule that plays protective roles against biotic and abiotic stresses. Melatonin participates in the regulation of tolerance to various plant diseases. For example, melatonin induces disease resistance to *Botrytis cinerea* in tomato fruit by regulating JA signaling and H~2~O~2~ levels and improves the disease resistance of *Arabidopsis* against pathogen infection via NO signaling \[[@CR2], [@CR3]\]. The increase in melatonin content activated by MeRAV1 and MERAV2 enhances plant disease resistance against cassava bacterial blight \[[@CR4]\]. Additionally, both exogenous treatment with and endogenous induction of melatonin increase abiotic stress tolerance in many plants. It has been reported that melatonin alleviates salt damage in grapes \[[@CR5]\] and enhances the drought tolerance of apple plants \[[@CR6]\]. Endogenously increasing melatonin via the overexpression of *ASMT1* significantly enhances the drought tolerance of *Arabidopsis* plants \[[@CR7]\]. Melatonin is a multifunctional signaling molecule \[[@CR8]\] that promotes grape berry ripening and quality formation via its interplay with other signaling molecules \[[@CR9]\]. Melatonin has also been reported to promote tomato and banana ripening and thereby affect fruit quality \[[@CR10], [@CR11]\]. Additionally, some other studies have emphasized the role of melatonin in regulating metabolite accumulation in fruits. For example, melatonin treatment enhances the contents of total anthocyanins, phenols and flavonoids in grape berries and wine \[[@CR12], [@CR13]\]. Post-harvest treatment with melatonin increases the contents of total phenols and anthocyanins in tomato fruits \[[@CR10]\]. The effects of melatonin on stress tolerance and fruit development are largely ascribed to the regulation of gene expression. Global gene expression changes caused by melatonin revealed that melatonin induces salt tolerance via ROS scavenging in rice shoots \[[@CR14]\]. Transcriptomic analysis revealed global changes in gene expression related to polyphenol metabolism, carbohydrate metabolism and ethylene biosynthesis and signaling upon melatonin treatment in grape berries \[[@CR12]\]. Proteomics data show that 241 proteins are significantly influenced by melatonin in tomato fruits \[[@CR15]\]. Arnao and Hernández-Ruiz summarized the genes that are up- and downregulated by melatonin under given physiological conditions, including various abiotic stressors and biotic stressors in different plants \[[@CR16]\]. DNA methylation is an important epigenetic modification that plays a very important role in the regulation of multiple biological processes, including stress responses, growth, development and fruit ripening, by modulating gene expression pretranscriptionally \[[@CR17], [@CR18]\]. Methylome analysis showed that most DNA methylation-modified genes are transcriptionally altered under Cd stress in rice \[[@CR19]\], and a global increase in DNA methylation occurs during orange fruit ripening \[[@CR20]\]. Additionally, some specific genes, such as fruit ripening-related RIN and PSY1, show demethylation at their promoters to activate their transcription during fruit ripening \[[@CR21]\]. Notably, an increasing number of studies indicate that melatonin alters global DNA methylation levels \[[@CR22]\] or the DNA methylation of specific genes \[[@CR23]\] in humans. However, the regulation of gene expression by melatonin via the modification of methylation remains unclear in plants. In this study, the DNA methylome and RNA transcriptome were analyzed in control and melatonin-treated grape berries, and the analysis of their interconnection revealed the key role of melatonin in increasing flavonoid biosynthesis and disease resistance. These functions of melatonin were further verified by determining the melatonin-induced changes in flavonoid contents, the berry decay rate and the levels of expression and promoter methylation of selected genes. These findings provide a new perspective for understanding the role of melatonin in regulating gene transcription in grape berries. Results {#Sec2} ======= Exogenous melatonin treatment increases melatonin content in 'merlot' grape berries {#Sec3} ----------------------------------------------------------------------------------- Exogenous melatonin treatment was used to increase the melatonin content of grape berries. Veraison (onset of berry ripening) is the key period for ripening regulation using signaling molecules, and melatonin levels begin to increase at veraison \[[@CR9]\]. The occurrence of veraison was indicated by the accumulation of sugars and anthocyanins and a decline in titratable acidity. In the 'Merlot' grape berries, anthocyanins began to accumulate at 80 days after bloom (DAB), and coloring began at 90 DAB (Fig. [1](#Fig1){ref-type="fig"}a, b). Total soluble solids (TSS) and titratable acid continued to accumulate and decrease, respectively, from 80 DAB onward (Fig. [1](#Fig1){ref-type="fig"}a, b). Therefore, veraison occurred at approximately 80 DAB, and melatonin treatment was performed at this time point. Treatment with 50 μM melatonin significantly increased the melatonin content of the berries, and increments of 1.24-, 4.45- and 9.02-fold were generated at 4, 48 and 144 h after treatment (HAT), respectively, compared with the control berries (Fig. [1](#Fig1){ref-type="fig"}c). Fig. 1Determination of berry veraison (**a**, **b**) and changes in melatonin content (**c**) in 'Merlot' berries under 50 μM melatonin treatment. Berry veraison was indicated by berry coloring, increases in total soluble solids and anthocyanins and decreases in titratable acid. Values represent the means ± SD of three replicates. \* Significant difference, *P* \< 0.05; \*\* highly significant difference, *P* \< 0.01 DNA methylation profiling in control and melatonin-treated 'merlot' grape berries {#Sec4} --------------------------------------------------------------------------------- Control and melatonin-treated berries collected at 48 HAT were used for single-base DNA methylation BS-Seq. A total of 877,654,62 and 889,758,30 clean reads were generated, with high conversion rates (C to G) of 99.54 and 99.47%, respectively. The average read depth of the control and melatonin-treated berries was 12.25 G and 12.42 G, respectively, accompanied by mapping rates of 54.33 and 54.31% (Table S[1](#MOESM1){ref-type="media"}). A total of 1,169,852 and 1,008,894 methylated cytosines (mCs) were identified in the control and melatonin-treated grapes, respectively. In the control/melatonin treatments, 42.10%/40.52% mCs were identified at CG sites, 30.35%/29.47% at CHG sites and 27.53%/29.99% at CHH sites (H = A, T, or C) (Table S[1](#MOESM1){ref-type="media"}). The control and melatonin-treated berries showed very similar mC profiles on 19 chromosomes (Fig. [2](#Fig2){ref-type="fig"}a). The three types of mCs exhibited similar distribution patterns on the same chromosome. In contrast, the different chromosomes showed varying methylation levels and mC distributions. For example, Chrs 16 and 8 possessed the highest and lowest levels of methylation, respectively, among the chromosomes; a high-mC region occurred around the central position of Chrs 4, 6, and 7, while high methylation levels were found on one side of Chrs 18, 1, 8 and 17 (Fig. [2](#Fig2){ref-type="fig"}a). Additionally, a high methylation level was detected in the CG context compared to the CHG and CHH contexts in the control and melatonin-treated berries (Fig. [2](#Fig2){ref-type="fig"}b). The comparison of methylation levels between the control and melatonin-treated berries revealed that a higher percentage of differentially methylated cytosines in exons than in promoters, introns and downstream regions (Fig. [2](#Fig2){ref-type="fig"}c). Additionally, the gene body generally showed a much higher methylation level than its upstream and downstream regions (Fig. [2](#Fig2){ref-type="fig"}b). Moreover, melatonin greatly reduced methylation levels in CHH contexts in the upstream, gene body and downstream regions of genes. In contrast, melatonin widely decreased methylation levels in CHG contexts to a lesser extent in various gene regions and led to a slight decline in CG methylation levels only in gene bodies (Fig. [3](#Fig3){ref-type="fig"}b). Fig. 2Methylation levels of different chromosomes and genomic regions in 'Merlot' berries. **a** The outermost bold lines indicate different chromosomes and their lengths at a 50 kilobase resolution. Red, green, black and blue peak shape diagrams indicate the methylation levels of mC, mCG, mCHG and mCHH, respectively, at different chromosome sites, and the peak height indicates the methylation level. The second to fifth circles are from control berries, and the sixth to ninth circles are from melatonin-treated berries. **b** Changes in the levels of CG, CHG and CHH methylation in melatonin-treated berries compared to the control. TSS, transcription start site. TTS, transcription termination site. Up2K and Down2K represent the 2000 bp upstream of the TTS and downstream of the TTS, respectively. **c** Percentages of differentially methylated cytosines in different genomic regionsFig. 3Analysis of the GO (**a**) and KEGG enrichment (**b**) of DMR-associated genes. *P* values are corrected to --log~10~ (*P* values) ranging from 0 to infinity, and a lower *P* value (i.e., a greater --log~10~ (*P* value)) indicates a higher intensity. The top 10 GO enrichments and the top 20 enriched KEGG pathways are displayed in this paper Gene ontology (GO) and KEGG enrichment of differentially methylated region (DMR)-related genes {#Sec5} ---------------------------------------------------------------------------------------------- A total of 25,125 DMRs were detected (fold change ≥4 or ≤ 0.25, *P* \< 0.05) between the control and melatonin-treated berries. Compared to the control, the methylation levels of 9000 and 16,125 DMRs were increased and decreased, respectively, by melatonin in grape berries. The total DMRs included 6517 DMR-associated genes in upstream, exon, intron and downstream regions (Table S[2](#MOESM2){ref-type="media"}). GO enrichment of the DMR-associated genes showed that the three most significantly altered biological processes were dTDP-rhamnose biosynthetic processes, ethanol metabolic processes and iron chelate transport. Enzyme-directed rRNA 2′-O-methylation and alternative mRNA splicing via the spliceosome were also significantly changed (Fig. [3](#Fig3){ref-type="fig"}a). The molecular functions of the DMR-associated genes primarily included carbohydrate kinase activity, cholesterol binding, triose-phosphate transmembrane transporter activity and cystathionine gamma-lyase activity. Most of the DMR-associated proteins were located in the nucleus (Fig. [3](#Fig3){ref-type="fig"}a). KEGG pathway analysis indicated that the three most significantly changed pathways were pentose and glucoronate interconversion, inositol phosphate metabolism, and fatty acid degradation. Additionally, the metabolism of amino acids, including lysine, linoleic acid and alpha-linolenic acid, and phenolics, including carotenoids and isoflavonoids, was significantly altered (Fig. [3](#Fig3){ref-type="fig"}b). Melatonin treatment alters global gene expression patterns in 'merlot' grape berries {#Sec6} ------------------------------------------------------------------------------------ RNA-Seq was employed to detect the transcript abundance of the control and melatonin-treated berries. A total of 3551 differentially expressed genes (DEGs) (\|log~2~FC\| \> 1, *P* \< 0.05) were identified in melatonin-treated berries compared to the control. Compared to the control, 2479 genes were upregulated, and 1072 genes were repressed (Table S[3](#MOESM3){ref-type="media"}, Fig. [4](#Fig4){ref-type="fig"}a). GO enrichment demonstrated that the DEGs in the biological process category were primarily related to defense responses, including the responses to biotic stimuli, chitin and fungi. Among molecular functions, the DEGs were mainly involved in trihydroxystilbene synthase activity, DNA binding transcription factor activity and isomerase activity. The inferred proteins of the DEGs were primarily located in the apoplast and plasma membrane (Fig. [4](#Fig4){ref-type="fig"}b). KEGG pathway analysis showed that the most significantly altered pathways were flavonoid biosynthesis, circadian rhythm-plant and plant-pathogen interaction pathways (Fig. [4](#Fig4){ref-type="fig"}c). Fig. 4Analysis of GO and KEGG enrichment of differentially expressed genes (DEGs) between the control and melatonin-treated 'Merlot' berries. **a** Differential transcript abundance between control and melatonin-treated berries. Red dots indicate differentially expressed genes. **b** GO clustering analysis of DEGs based on their functional enrichment. The top 10 GO enrichments are displayed. **c** KEGG enrichment analysis of DEGs. The top 10 enriched pathways are displayed. *P* values are corrected to --log~10~ (*P* values) ranging from 0 to infinity, and a lower *P* value (i.e., greater --log~10~ (*P* value)) indicates a higher intensity Interconnection of the DNA methylome and RNA transcriptome {#Sec7} ---------------------------------------------------------- A total of 1626 genes were identified via a cross-analysis of DMR-related genes and DEGs. Among these genes, the expression levels of 886 genes (54.5%) were negatively related to their methylation levels, and 740 genes (45.5%) showed a positive relationship between expression and methylation levels (Table S[4](#MOESM4){ref-type="media"}, Fig. [5](#Fig5){ref-type="fig"}a). Additionally, most of the interconnected genes showed upregulated expression levels in melatonin-treated berries compared to the control (Fig. [5](#Fig5){ref-type="fig"}a). Fig. 5Cross-sectional analysis of differentially methylated genes and DEGs and GO enrichment of the interconnected genes. **a** Cross-analysis was performed between the DMR-related genes and DEGs. Red dots represent negatively related genes, and blue dots represent positively related genes. The y-axis represents the differences in methylation levels (%). The x-axis represents the differences in gene expression. FC, fold change. **b**, **c** Validation of the results obtained from BS-Seq and RNA-Seq via methylation-specific PCR and qRT-PCR analyses of three selected loci. *STS1*, *VIT_216s0100g01100*; *EDS1*, *VIT_217s0000g07370*; *CML41*, *VIT_218s0001g11830*. Functional annotations of the above genes were listed in Table [1](#Tab1){ref-type="table"}. FC (fold change) was calculated using the ratio of the expression in the treated berries to that in the control. **d** GO enrichment of the interconnected genes obtained through the comparison of DMP-related genes and DEGs. The top 10 enriched GO categories are displayed. *P* values are corrected to --log~10~ (*P* values) ranging from 0 to infinity, and a lower *P* value (i.e., greater --log~10~ (*P* value)) indicates a higher intensity It is well established that methylation at a gene promoter region can inhibit DNA transcription \[[@CR24]\]. In this study, we identified 144 differentially methylated promoter (DMP)-related genes exhibiting a negative association between promoter methylation levels and expression levels, among which 133 genes presented upregulated expression and decreased promoter methylation, and 13 genes showed the opposite relationship (Table S[5](#MOESM5){ref-type="media"}). To verify the association between the levels of promoter methylation and gene expression, three genes involved in flavonoid biosynthesis and plant-pathogen interaction (Table [1](#Tab1){ref-type="table"}) were selected for analysis by methylation-specific PCR (MS-PCR) and qRT-PCR. Compared to the control, the melatonin-treated berries exhibited a low level of promoter methylation and a high level of gene expression (Fig. [5](#Fig5){ref-type="fig"}b, c). Almost all of the interconnected genes were related to defense responses, including responses to biotic stimuli, salicylic acid, chitin, bacteria and fungi. Their molecular functions included ferric-chelate reductase, calcium-transporting ATPase, and trihydroxystilbene synthase enzyme activities. The inferred proteins were located in the plasma membrane and outside the plasma membrane (Fig. [5](#Fig5){ref-type="fig"}d). Additionally, two KEGG pathways were significantly altered based on these 144 genes: plant-pathogen interaction (*P* = 7.57 × 10^− 6^) and flavonoid biosynthesis (*P* = 7.47 × 10^− 4^). Table 1The most significantly altered genes in promoter methylation levels (\|log~2~FC\| \> 2 or detected only in control berries) and expression levels (\|log~2~FC\| \> 3 or detected only in melatonin-treated berries)Gene accession noPosition of DNA methylation(log~2~FC)Gene descriptionGene functionMethylationExpressionVIT_216s0039g01320chr16:724,559--726,558− 2.06DTPhenylalanine ammonia lyase 1 (PAL1)Flavonoid biosynthesisVIT_216s0100g01100chr16:16,557,734--16,559,733−2.563.86Stilbene synthase 1 (STS1)VIT_200s0203g00220chrUn:12,072,634--12,074,633DCDTG-type lectin s-receptor-like serine threonine-protein kinaseImmune response^30^VIT_212s0059g00160chr12:5,071,913--5,073,912−6.64DTAnkyrin repeat-containing protein (ACD6)Disease resistance and antioxidation metabolism^33^VIT_217s0000g07370chr17:8,237,664--8,239,663−5.644.86Enhanced disease susceptibility 1 (EDS1)Plant-pathogen interaction^42,43^VIT_218s0001g11830chr18:10,101,409--10,103,408−5.643.58probable calcium-binding protein CML41VIT_204s0008g07140chr4:7,279,720--7,281,719DC3.64Aspartic proteinase CDR1-likeDisease Resistance^36^VIT_216s0050g01150chr16: 18052767--18,054,766DC4.11Heat shock protein 83-likeVIT_204s0044g01420chr4:22,951,153--22,953,152−3.834.70Probable polygalacturonase-likeVIT_210s0003g05450chr10: 10098885--10,100,884−4.323.41Reticuline oxidase-like proteinResponse of plants to pathogenic attack ^38^VIT_219s0014g00470chr19: 478152--480,151−3.063.07Leucine-rich repeat receptor-like serine/threonine-protein kinaseMAMP-triggered innate immunity ^38^VIT_210s0003g01220chr10:2,552,768--2,554,767−4.324.42Heavy metal-associated isoprenylated plant protein 26Plant responses to environmental changes^35^VIT_205s0077g01540chr5: 1240180--1,242,179−2.253.44Pathogenesis-related protein 10Response to biotic and abiotic stresses^37^VIT_204s0008g03530chr4: 2888437--2,890,436DC3.69Ankyrin repeat-containing proteinUnknownVIT_203s0063g00550chr3:4,059,484--4,061,483DC6.01Unnamed protein productUnknownVIT_218s0001g08500chr18:6,946,257--6,948,256−2.403.51Unnamed protein productUnknownVIT_200s0270g00120chrUn:19,991,241--19,993,240−4.324.59Kunitz-type trypsin inhibitorUnknown*DC* detected only in control berries, *DT* detected only in melatonin-treated berries, *FC* fold change In addition, the gene functions of the 17 genes with the largest changes in expression (\|log~2~FC\| \> 3 or detected only in melatonin-treated berries) and methylation levels (\|log~2~FC\| \> 2 or detected only in control berries) in the promoter region are listed. Among these genes, *PAL1* and *STS1* are responsible for flavonoid biosynthesis. The functions of four of the genes are unknown. The remaining genes including *EDS1* and *CML41* are involved in disease resistance and/or abiotic stress responses (Table [1](#Tab1){ref-type="table"}). Melatonin and DNA methylation inhibitor treatments increase disease resistance and flavonoid accumulation in grape berries {#Sec8} -------------------------------------------------------------------------------------------------------------------------- To investigate whether melatonin affected the disease resistance of grape berries, 50 μM melatonin treatment of detached 'Merlot' and 'Shine Muscat' berries was performed. Melatonin clearly decreased mycelial growth on the surface of wounded 'Merlot' berries inoculated with *Botrytis cinerea* (*B. cinerea*) compared with the control berries (Fig. [6](#Fig6){ref-type="fig"}a). Melatonin decreased the decay rate of 'Shine Muscat' berries without inoculation with *B. cinerea* at 14 and 21 DAT (Fig. [6](#Fig6){ref-type="fig"}b, c). Additionally, the commonly used DNA methylation inhibitor 5́-azacytidine (5́-Aza) was applied in this study \[[@CR19]\]. The application of 5́-Aza led to similar effects on mycelial growth of *B. cinerea* and berry decay rate compared with melatonin (Fig. [6](#Fig6){ref-type="fig"}a-c). Therefore, melatonin and 5́-Aza increased disease resistance of grape berries. On the other hand, the effects of melatonin and 5́-Aza on the flavonoid content were determined. Melatonin and 5́-Aza increased the flavonoid content of the 'Merlot' and/or detached 'Shine Muscat' berries at 3 DAT (Fig. [6](#Fig6){ref-type="fig"}d). Fig. 6Effects of melatonin and 5́-Aza on disease resistance, flavonoid accumulation, transcripts and DNA methylation of the genes involved in 'Merlot' and/or 'Shine Muscat' berries. **a** Mycelial growth on *B. cinerea*-inoculated 'Merlot' berries treated with or without melatonin and 5́-Aza. **b** Phenotypes of the control and treated 'Shine Muscat' berries with melatonin and 5́-Aza. Decayed berries were removed at 21 DAT. **c** The average decay rate of 'Shine Muscat' berries determined at 21 DAT. **d** Flavonoid content in 'Merlot' and detached 'Shine Muscat' berries at 3 DAT. ND, not detected. **e**, **f** Expression difference in 'Merlot' (**e**) and detached 'Shine Muscat' berries (**f**) at 3 DAT. FC (fold change) was calculated using the ratio of the expression in the treated berries to that in the control. *PAL1*, *VIT_216s0039g01320*; *STS1*, *VIT_216s0100g01100*; *EDS1*, *VIT_217s0000g07370*; *CML41*, *VIT_218s0001g11830*. Functional annotations of the above genes were listed in Table [1](#Tab1){ref-type="table"}. **g** DNA methylation in 'Shine Muscat' berries at 3 DAT Additionally, melatonin upregulated the expression levels of *PAL1*, *STS1*, *EDS1* and *CML41*, which are related to flavonoid biosynthesis or plant-pathogen interaction (Table [1](#Tab1){ref-type="table"}), in the 'Merlot' berries at 3 DAT (Fig. [6](#Fig6){ref-type="fig"}e). Both melatonin and 5́-Aza increased the expression levels of the above four genes in the detached 'Shine Muscat' berries at 3 DAT (Fig. [6](#Fig6){ref-type="fig"}f). Moreover, both melatonin and 5́-Aza decreased the methylation levels in the promoter regions of *STS1*, *EDS1* and *CML41* (Fig. [6](#Fig6){ref-type="fig"}g). Collectively, melatonin and 5́-Aza increased disease resistance and flavonoid accumulation of grape berries, which might be involved in the upregulation of gene expression caused by the demethylation of the promoter regions. The melatonin-induced increase in the promoter transcription-driving capacity is negatively regulated by DNA (cytosine-5)-methyltransferase 1 (MET1) {#Sec9} ---------------------------------------------------------------------------------------------------------------------------------------------------- As shown in Figs. [5](#Fig5){ref-type="fig"}b, c and [6](#Fig6){ref-type="fig"}e-g, melatonin treatment decreased the promoter methylation levels and increased the expression levels of the three selected genes. Additionally, the expression levels of *MET1*, *MET1B* and *S-adenosylmethionine-dependent methyltransferase* (*SadMET*) were significantly downregulated by melatonin according to the results of RNA-Seq and qRT-PCRs (Fig. [7](#Fig7){ref-type="fig"}a), which was consistent with the melatonin-induced decrease of methylation levels (Fig. [2](#Fig2){ref-type="fig"}b). In contrast, the changes in the expression levels of two *CMT2* genes were not associated with the decrease of methylation levels under melatonin treatment (Figs. [2](#Fig2){ref-type="fig"}b, [7](#Fig7){ref-type="fig"}a). To further elucidate whether melatonin regulated gene expression by modifying promoter methylation, *MET1* (VIT_212s0035g01770) was used to increase the methylation level of the *enhanced disease susceptibility 1* (*EDS1*) (VIT_217s0000g07370) promoter under melatonin treatment. The *Agrobacterium*-mediated transient expression of P~eds~ (800 bp upstream of ATG of *EDS1*, Fig. [7](#Fig7){ref-type="fig"}a)-35S miniGUS activated *GUS* expression in grape calluses (Fig. [7](#Fig7){ref-type="fig"}c). In contrast to the transformation of P~eds~-35S miniGUS alone, the cotransformation of P~eds~-35S miniGUS and 35S::MET1 decreased the extent of GUS staining and GUS activity (Fig. [7](#Fig7){ref-type="fig"}c, d). Therefore, the increased methylation level caused by MET1 reduced the transcription-driving capacity of P~eds~. Additionally, the melatonin-treated calluses expressing P~eds~-35S miniGUS were bluer in color and showed higher GUS activity than control calluses and melatonin-treated calluses cotransformed with P~eds~-35S miniGUS and 35S::MET1 (Fig. [7](#Fig7){ref-type="fig"}c, d). Moreover, the above cotransformants showed an increased methylation level of the *EDS1* promoter compared to the calluses expressing P~eds~-35S miniGUS alone under melatonin treatment (Fig. [7](#Fig7){ref-type="fig"}e). Collectively, melatonin increased the transcription-driving capacity of P~eds~, at least partially by decreasing the methylation level of P~eds~. Fig. 7Transcription capacity of the *EDS1* promoter to drive GUS expression in the presence of melatonin and/or MET1. **a** Expression difference of the five selected genes in 'Merlot' berries at 48 HAT from RNA-Seq and qRT-PCR analyses. *MET1*, *VIT_212s0035g01770*; *MET1B*, *VIT_212s0035g01755*; *SadMET*, *VIT_214s0006g02170*; *CMT2_1*; *VIT_216s0039g02470*; *CMT2_2*, *VIT_216s0039g02460*. **b** Prediction of methylated cytosine in the *EDS1* promoter (<http://www.urogene.org/cgi-bin/methprimer2>). O/E values indicate the ratio between the actual value and expected value of the CpG locus. The 800-bp region with a high CpG level indicated by the black box was used as the promoter of *EDS1* (P~eds~) to produce the construct of P~eds~-35S miniGUS. **c** Histochemical analysis of the transcriptional capacity of P~eds~ to drive *GUS* expression in grape calluses agroinfiltrated with different vector constructs: B1, control calluses; B2 and B4, P~eds~-35S miniGUS; B3, P~eds~-35S miniGUS and 35S::MET1; B5, P~eds~-35S miniGUS and 35S::MET1 with melatonin treatment; B6, P~eds~-35S miniGUS with melatonin treatment. **d** Gus activities of grape calluses infiltrated by *Agrobacterium* containing the B1-B6 constructs. **e** DNA methylation level of the *EPS1* promoter, including endogenous DNA and DNA provided by P~eds~- 35S miniGUS, in calluses infiltrated by *Agrobacterium* containing the B4-B6 constructs Discussion {#Sec10} ========== Methylation occurs predominantly in the CG context, followed by CHG and CHH contexts, in plants including rice (CG: 54.7%, CHG: 37.3%, CHH: 12%) \[[@CR25]\] and *Arabidopsis* (CG: 24%, CHG: 6.7%, CHH: 1.7%) \[[@CR26]\]. A similar pattern was observed in grape berries; however, the methylation rates in the CHG and CHH contexts in grape berries (30.35 and 27.53%, respectively) were higher than those in other species such as rice and *Arabidopsis,* as mentioned above. Additionally, CHH methylation is higher in tomato fruits (13.52--14.20%) than in leaves (8.63%) \[[@CR27]\], suggesting the important role of CHH methylation in fruits. The gene body displayed a higher methylation level than its two flanking regions in grape berries (Fig. [2](#Fig2){ref-type="fig"}c), while contrary results have been found in citrus fruit \[[@CR28]\]. Additionally, it is noteworthy that a global increase in DNA methylation occurs during orange fruit ripening, whereas global demethylation accompanies tomato fruit ripening \[[@CR20], [@CR27]\]. Therefore, the methylation patterns of CG, CHG and CHH contexts and their biological significance vary between different species and even in different tissues of the same species, and CHH methylation might play an important role in fruits. Cytosine methylation is regulated by DNA methyltransferases (MET1, CMT and DRM), demethylases (DME, ROS1, DML) and the chromatin-remodeling factor DDM1 \[[@CR29]\]. In plants, the methyltransferase MET is mainly responsible for DNA methylation at the CG site \[[@CR30]\], however MET1 also influences CHH methylation \[[@CR31], [@CR32]\]. In this study, the significant downregulation of two *MET* genes (VIT_212s0035g01770 and VIT_212s0035g01755) might have contributed to the reduced DNA methylation observed in melatonin-treated berries (Table S3). Methylated DNAs are produced using S-adenosylmethionine as the donor of methyl groups under catalysis by SadMET \[[@CR33]\]. The considerable downregulation of *SadMET* (Vit_214s0006g02170) might decrease DNA methylation levels as well (Table S[3](#MOESM3){ref-type="media"}). The chromoethylase CMT mainly maintains DNA methylation at CHH and CHG sites \[[@CR34]\]. Melatonin treatment resulted in much greater effects on CHH methylation than on CG and CHG methylation; therefore, *CMT*s were expected to be downregulated by melatonin. However, only two upregulated *CMT2*s with very low FPKM values were detected in melatonin-treated berries (Table S[3](#MOESM3){ref-type="media"}), indicating that they might not be the key enzymes responsible for the melatonin-induced decrease in CHH methylation. On the other hand, DNA demethylases such as DML2 are indicated to regulate DNA methylation during tomato and citrus ripening \[[@CR20]\]. Differentially expressed DNA demethylases were not detected in the melatonin-treated berries. Taken together, the present results suggested that the melatonin-induced declines in DNA methylation may be related to *MET* and *S-adenosylmethionine-dependent methyltransferase* genes and other underlying mechanisms need to be explored. Our study showed that melatonin treatment broadly decreased genomic DNA methylation levels and modified gene expression. The correlation between DNA methylation and gene expression is very complex and is influenced by various factors, including tissue type and genomic regions \[[@CR35]\]. Only a subset of DMRs are associated with the expression of nearby genes \[[@CR36]\]. The role of gene body methylation in modifying gene expression is less well characterized \[[@CR37]\]. Limited studies have shown that gene body methylation potentially plays roles in producing new functional gene transcripts by inhibiting RNA splicing \[[@CR38]\]. The reduction in gene body methylation (Fig. [2](#Fig2){ref-type="fig"}b) and increase in RNA splicing (Table S[6](#MOESM6){ref-type="media"}) suggested that melatonin increased RNA splicing by decreasing the methylation level of the gene body. In contrast, methylation in a gene promoter region prevents RNA polymerases and transcription factors from binding the promoter, thereby inhibiting DNA transcription \[[@CR24]\]. In *Arabidopsis*, reduced DNA methylation contributes to the regulation of pathogen-induced gene expression in hypomethylated and hypermethylated mutants \[[@CR39]\]. In this study, the functional annotation of 144 genes exhibiting a negative association between the levels of promoter methylation and gene expression (Table S[5](#MOESM5){ref-type="media"}) suggested that melatonin might regulate gene expression by modifying promoter methylation and, hence, affect disease resistance and flavonoid biosynthesis in grape berries. This inference was also strongly supported by the similar effects of melatonin and 5́-Aza on the disease resistance, total flavonoid content and changes in promoter methylation and gene expression (Fig. [6](#Fig6){ref-type="fig"}). Additionally, the promoter of *EDS1* was used to verify that melatonin increased the transcription-driving capacity of the promotor by decreasing its methylation level (Fig. [7](#Fig7){ref-type="fig"}). Collectively, the results indicated that melatonin increased disease resistance and flavonoid biosynthesis at least partially by decreasing the methylation of the gene promoters involved. Among the DMP-related genes and DEGs showing changes caused by melatonin, two *PAL*s and *STS*s were interconnected in promoter methylation and gene expression (Table [1](#Tab1){ref-type="table"}, Table S[5](#MOESM5){ref-type="media"}). PAL is the first key enzyme in the phenylpropanoid pathway, in which it catalyzes the biosynthesis of cinnamic acid and provides the initial precursor for other phenolic compounds \[[@CR40]\]. Melatonin might broadly affect phenolic metabolism by increasing *PAL* expression. STS is responsible for the biosynthesis of resveratrol \[[@CR41]\], and the upregulation of *STS*s suggests a possible role of melatonin in promoting resveratrol biosynthesis in grape berries. On the other hand, melatonin might affect disease resistance through multiple pathways based on the identified interconnected genes (Table S[5](#MOESM5){ref-type="media"}). For example, EDS1 increases the robustness of the innate immune system by promoting SA biosynthesis \[[@CR42]\]. CML41 is required for a complete defense response against bacterial pathogens by enabling Ca^2+^ signaling specificity, a critical component of the immune response \[[@CR43]\]. Lectin receptor kinase is related to the activation of immune signaling involving the mitogen-activated protein kinase (MAPK) pathway \[[@CR44]\]. Conclusion {#Sec11} ========== The results indicated that melatonin led to a global decrease in the DNA methylation of grape berries, and the decrease occurred primarily at CHH sites, followed by CHG and CG sites. Simultaneously, melatonin broadly modified gene expression in grape berries. Melatonin increased gene expression at least partially by reducing promoter methylation and thereby promoted disease resistance and flavonoid biosynthesis. Methods {#Sec12} ======= Plant materials and experimental treatments {#Sec13} ------------------------------------------- Grape berries were collected from 'Merlot' and 'Shine Muscat' vines that were grown at an experimental vineyard in Tai-An City, Shandong Province, China. The 'Merlot' berries were used to determine the effects of melatonin on the DNA methylome and RNA transcriptome. Each grape cluster on a vine at 80 days after bloom was soaked for 5 s in a 50 μM melatonin solution plus 0.05% (v/v) Triton X-100. Control berries were treated with 0.05% (v/v) Triton X-100. Detached 'Shine Muscat' berries were used to evaluate the effects of melatonin and a methylation inhibitor (5′-Aza) on berry disease resistance and flavonoid accumulation. The control and melatonin treatments were the same as those described above. For 5′-Aza treatment, the detached grape clusters were soaked for 5 s in a 20 μM 5-Aza solution plus 0.05% (v/v) Triton X-100. The treated berries were placed in an environment controlled chamber (20 ± 1 °C, 80% RH, dark) for 21 days. 'Merlot' grape calluses were used for promoter assays. The calluses were subcultured on MS medium supplemented with 2.2 mg/L thidiazuron, 10 mg/L picloram and 0.59 g/L 2-(N-morpholino) ethanesulfonic acid at 25 °C under dark conditions. The berries and calluses were collected, rinsed, frozen in liquid nitrogen, and stored at − 70 °C for the determination of DNA methylation, gene expression and other parameters. Determination of TSS, titratable acid, relative anthocyanin contents, total flavonoids and average decay rate {#Sec14} ------------------------------------------------------------------------------------------------------------- Fresh berry pulp was homogenized and filtered. The filtrate was used for the determination of TSS and titratable acid. The TSS content was determined with a PAL-1 digital-display sugar meter (Atago, Tokyo, Japan). Titratable acid was measured by the titration of the filtrate with 0.1 M NaOH to pH 8.3. The results are expressed as mg tartaric acid per g FW. Anthocyanins were extracted and quantified, and the relative anthocyanin content was calculated as reported in our previous study \[[@CR9]\]. The total flavonoid content was spectrophotometrically measured using rutin as the standard as described by Dewanto et al. \[[@CR45]\]. Berry decay rate was calculated by dividing the number of decayed berries by the total number of berries. The decayed berries were visually evaluated according to the following standard: having slight mildew and moderate shrivel and brown spotting. Measurements were performed with three biological replicates. Each replicate consisted of 10 clusters (approximately 600 berries). Melatonin content determination {#Sec15} ------------------------------- Melatonin was extracted using a C~18~ solid-phase extraction cartridge (ProElut™; DIKMA, China) according to our previous study \[[@CR9]\]. Ten microliter samples were separated using a BEH C~18~ column (Waters, 2.1 mm internal diameter × 50 mm length, and 1.7 μm particle size) in an Acquity UHPLC system (Waters, Milford, MA, USA). Samples were analyzed using a QTof-Micro mass spectrometer (Waters, Milford, MA, USA). The parameters and conditions of the UHPLC-MS analysis were set according to our previous study \[[@CR12]\]. Bisulfite sequencing (BS-Seq) library construction, sequencing and analysis of differentially methylated cytosines {#Sec16} ------------------------------------------------------------------------------------------------------------------ Genomic DNA was extracted from the control and melatonin-treated 'Merlot' grape berries. The extracted DNAs were fragmented by sonication to 200--300 bp. After end repair, the adenylation of 3́ ends, and adaptor ligation, the purified ligation products of 275--350 bp were treated with sodium bisulfite using the EZ DNA Methylation Gold Kit (Zymo Research, USA). The resultant DNAs were subjected to paired-end sequencing on the Illumina HiSeq Xten platform (Illumina, Inc., San Diego, CA, USA). Methylome sequencing and analysis were conducted by OE Biotech Co., Ltd. (Shanghai, China). Clean reads of more than 75 bp were obtained by filtering the raw data. Cytosines in the forward strands of the clean reads and the genome were changed to thymidines, and guanines on the reverse strands were changed to adenosines in silico using Bismark software. Then, the clean reads were aligned to the grape reference genome (<http://genomes.cribi.unipd.it/DATA/GENOME_12X/>). The detection of methylated cytosine sites and the analysis of DMRs were conducted using MethyKit software. Only the cytosine sites covered by at least 5 reads were used. The true methylated cytosine sites were confirmed by the binomial distribution of methylated and unmethylated cytosines and a false discovery rate (FDR) ≤0.05. A sliding-window approach was used to screen DMRs. The methylation levels (%) of different cytosine sites were integrated in each 1000-bp tiling window, and the integrated data for each window were used for DMR analysis. DMRs were analyzed using the logistic regression mode and identified on the basis of changes in methylation levels (fold change ≥4 or ≤ 0.25) and an FDR ≤ 0.05. Additionally, the region 2000 bp upstream of a transcription start site was taken as the promoter region for the analysis of differentially methylated promoters (DMPs). For DMP analysis, the sliding window was set to 2000 bp, and the other analysis methods were the same as those for DMR analysis. Three biological replicates were performed for the control and melatonin treatment. RNA library construction, sequencing and DEG analysis {#Sec17} ----------------------------------------------------- RNA library construction was performed using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. The RNA libraries were sequenced using the Illumina HiSeq Xten platform (Illumina, Inc., San Diego, CA, USA), and 150-bp paired-end reads were generated. The clean reads were mapped to the grape genome (<http://genomes.cribi.unipd.it/DATA/GENOME_12X/>) using HISAT2. Unigene expression levels were quantified according to FPKM values, which were calculated using Cufflinks. DEGs between two samples were screened using an absolute log~2~(fold change) ≥1 and an FDR ≤0.05 as the thresholds. Three biological replicates were generated for the control and melatonin-treated berries. Bisulfite-RT-PCR (BS-PCR) and quantitative RT-PCR (qRT-PCR) {#Sec18} ----------------------------------------------------------- BS-PCR analysis of three genes was performed to validate the quality of BS sequencing. The DNAs extracted from the control and melatonin-treated berries were treated with bisulfate using an EZ DNA Methylation Gold Kit (Zymo Research, USA). The promoter regions of the above three genes were amplified using specific primers designed with Methyl Primer Express v2.0 (<http://www.urogene.org/cgi-bin/methprimer2>; Table S[7](#MOESM7){ref-type="media"}). The BS-PCR products were cloned into pMD19-T (TaKaRa, Dalian, China), and 30 positive clones were sequenced. The sequencing results were used to calculate methylation levels with the online software BiQ Analyzer (<http://biq-analyzer.bioinf.mpi-inf.mpg.de/>). qRT-PCR was performed using SYBR Green Master-Mix (SYBR Premix EX Taq TM, Dalian, China) on a Bio-Rad iQ5 (Hercules, CA, United States) instrument, and the primers are listed in Table S[7](#MOESM7){ref-type="media"}. Transient transformation of the *EDS1* promoter and *MET1* into grape calluses {#Sec19} ------------------------------------------------------------------------------ The promoter sequence of *EDS1*, 800 bp upstream of ATG, was cloned (see the specific primers in Table S[7](#MOESM7){ref-type="media"}) and fused upstream of the 35S minimal promoter of pRI101-GUS (Takara, Dalian, China) to generate the P~eds~::35S miniGUS plasmid. The ORF of *MET1* was cloned (see the specific primers in Table S[7](#MOESM7){ref-type="media"}) and fused downstream of the 35S sequence of pRI101 to generate the 35S::MET1 plasmid. The above two plasmids were introduced into *Agrobacterium* strain GV3101. The *Agrobacterium*-mediated transient transformation of grape calluses was performed according to a previous study \[[@CR5]\]. Grape calluses were immersed in an *Agrobacterium* suspension and gently shaken for 20 min. After blotting dry on sterile filter paper, the calluses were transferred to solid MS medium containing 100 μM acetosyringone. After 3 days of coculture in darkness at 28 °C, the calluses were collected and subjected to GUS staining and activity assays. GUS histochemical staining and activity assays were conducted according to a previously reported method as described by Jefferson et al. \[[@CR46]\]. GUS activity was expressed as nmol of 4-methylumbelliferone per mg protein per minute. Inoculation of single detached berries with *B. cinerea*. The detached 'Merlot' berries were soaked for 5 s in a 50 μM melatonin or 20 μM 5′-Aza solution plus 0.05% (v/v) Triton X-100. The berries were then gently wounded with a razor blade to break the surface of the berries. Wounded berries were inoculated by spraying with a suspension of *B. cinerea* (1 × 10^4^ conidia/ml). After inoculation, berries were placed in an environment controlled chamber (20 ± 1 °C, 90% RH, dark). Statistical analyses {#Sec20} -------------------- Statistical analysis was performed using SPSS (v19.0) software. One-way analysis of variance followed by Duncan's multiple range test was employed. Supplementary information ========================= {#Sec21} **Additional file 1 Table S1** Details of the bisulfite sequencing, mapping and identification of methylated cytosines (mCs). **Additional file 2 Table S2** Differentially methylated regions (DMR, fold change ≥4 or ≤ 0.25, *P* \< 0.05) and DMR-related genes. **Additional file 3 Table S3** Differentially expressed genes (DEGs) between the control and melatonin-treated berries (\|log~2~FC\| \> 1, P \< 0.05). **Additional file 4 Table S4** Cross-analysis of DMR-related genes and DEGs. **Additional file 5 Table S5** Cross-analysis of DMP-related genes and DEGs. **Additional file 6 Table S6** Alternative splicing frequency statistics for the control and melatonin-treated berries. **Additional file 7 Table S7** Primers used in this study. DAB : Days after bloom DAT : Days after treatment DEG : Differentially expressed genes DMP : Differentially methylated promoter DMR : Differentially methylated region EDS1 : Enhanced disease susceptibility 1 FC : Fold change 5́-Aza : 5́-azacytidine GO : Gene ontology mC : methylated cytosine HAT : Hours after treatment MET1 : DNA (cytosine-5)-methyltransferase 1 SadMET : S-adenosylmethionine-dependent methyltransferase TSS : Total soluble solids **Publisher's Note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information ========================= **Supplementary information** accompanies this paper at 10.1186/s12870-020-02445-w. Not applicable. YY and SG conceived and designed the research; SG, WM and XL performed the experiments; SG and XC analyzed the data; and YY wrote the manuscript. All authors read and approved the manuscript. This study was financially supported by National Key R&D Program of China (2018YFD1000200), the Key Research and Development Program of Shandong Province (2019GNC106149), the National Natural Science Foundation of China (31872068), and the Funds of Shandong "Double Tops" Program (SYL2017YSTD10). None of the funding bodies have any role in the design of the study or collection, analysis, and interpretation of data as well as in writing the manuscript. Full BS-Seq and RNA-Seq data were submitted to the sequence read archive (SRA) of NCBI under BioSample accessions PRJNA603630 and PRJNA603632, respectively (<https://www.ncbi.nlm.nih.gov/sra>). Not applicable. Not applicable. The authors declare that they have no competing interests.

1. Introduction =============== Paget\'s disease can arise in the breast (mammary Paget disease \[MPD\]) or in other locations (extramammary Paget disease \[EMPD\]) such as the anogenital skin in females (Paget disease of the vulva \[PDV\]). PDV is an uncommon pathologic condition^\[[@R1]--[@R8]\]^ accounting for from 1% to 2% of all tumors of the female genital area.^\[[@R3],[@R9]\]^ The estimated incidence in Europe is about 0.7 cases per 100,000 women.^\[[@R2],[@R4]\]^ It is thought to originate from adnexal structures, such as apocrine glands, multipotent stem cells in the epidermal basal layer or infundibular stem cells of the hair follicle.^\[[@R2],[@R5]--[@R8]\]^ In 2001, Wilkinson et al subdivided vulvar Paget disease into primary and secondary disease.^\[[@R3],[@R14],[@R15]\]^ Primary (cutaneous) Paget disease is defined as an intraepithelial adenocarcinoma arising within the epidermis and extending into the contiguous epithelium of skin appendages. It is further classified into 3 subtypes: type 1a, intraepithelial Paget disease, representing 75% of cases; type 1b, intraepithelial Paget disease with invasion, representing16% of cases; and type 1c, intraepithelial Paget disease with an underlying adenocarcinoma of a skin appendage or a subcutaneous gland, with 9% of incidence.^\[[@R3],[@R11]\]^ Secondary (noncutaneous) Paget disease is considered as an epidermotropic metastasis originating from an underlying malignancy of the gastrointestinal tract, of the urogenital tract, or of an adenocarcinoma originating elsewhere.^\[[@R3],[@R11]\]^ The presence of Paget cells characterizes EMPD. These cells have prominent pale cytoplasm and a prominent central nucleolus, and are distributed throughout the epithelium, either as single cells or clusters with variable extent. The cytoplasm contains diastase resistant PAS positive material^\[[@R1]\]^ Paget cells can be visualized using hematoxylin and eosin staining. The scattered Paget cells are diagnostic, but they are interspersed within the normal epithelium and it can be difficult to detect them at times.^\[[@R2]\]^ Invasion is characterized by the presence of dyscohesive neoplastic Paget cells infiltrating the underlying dermis or submucosa.^\[[@R2],[@R16]\]^ The degree of invasion beyond the basement membrane is categorized into 3 levels: 1. in situ in the epidermidis; 2. microinvasion to the papillary dermis; 3. deep invasion to the reticular dermis or subcutaneous tissue.^\[[@R3],[@R16]\]^ Immunohistochemistry is mandatory for correct diagnosis of Paget disease. Paget cells express Cytokeratin 7 (CK7), Cytokeratin 5.2, carcinoembryonic antigen, gross cystic disease fluid protein 15, human epidermal growth factor receptor 2, carbohydrate antigen 125, and androgen receptors, but do not express markers of squamous cell differentiation, such as p63 and p40, and melanocyte markers, such as Melan-A, Human Melanoma Black-45, or S100 proteins. p53 protein overexpression in the intraepidermal component is associated with invasion. Uroplakin-III, CK7 and CK20 and GATA3 gene are expressed in Paget disease secondary to urothelial carcinoma.^\[[@R1],[@R2],[@R9]\]^ CK20, Homeobox protein CDX-2, and Mucin 2 positivity (but not CK7) might indicate an underlying anorectal adenocarcinoma.^\[[@R1],[@R2],[@R9]\]^ EMPD is also associated with other malignancies, such as malignancies outside the vulva from 2% to 54% of cases. These may include breast, intestinal, and urologic malignancies. For this reason, in patients with biopsy-confirmed Paget disease, there should be further evaluation of the breast, genitourinary tract, and gastrointestinal tract.^\[[@R2],[@R4],[@R15],[@R17]\]^ There are no pathognomonic symptoms or specific clinical aspects. Irritation, itching, burning, and vulvar pain are the most frequent symptoms. Sometimes vulvar paget disease (VPD) can be asymptomatic.^\[[@R2]--[@R4]\]^ Often VPD looks like a red eczematoid pruritic lesion, sometimes as an erythematous plaque with typical white scaling. The plaque may be ulcerated and crusted with a papillomatous surface.^\[[@R1],[@R2]\]^ The clinical picture is usually aspecific^\[[@R1]--[@R4],[@R17]\]^ and for this reason diagnosis is usually made in very extensive disease. Surgery plays a major role in the treatment of PDV and the occurrence of positive margins is very common.^\[[@R3]--[@R6],[@R8]\]^ However, there is no clear relationship between recurrent disease and the presence of positive margins or invasive disease.^\[[@R3]--[@R6],[@R8]\]^ We retrospectively review our experience with PDV in order to evaluate clinical presentation, surgical treatment, and prognostic factors for recurrence and survival. 2. Materials and methods ======================== A retrospective chart review was conducted on all patients with PDV managed at the Department of Obstetrics and Gynecology, of the University of Bari, and at the Gynecologic Oncology Unit, "Giovanni Paolo II" National Cancer Institute in Bari, Italy between 1998 and 2018. Institutional Review Board approval was obtained with a waiver of informed consent. Medical records were reviewed for demographic information, clinical data, pathologic findings, treatment modalities, and outcomes. At the time of diagnosis, all patients received intensive staging with pelvic examination, transvaginal ultrasound, PAP smear, chest X-ray, mammography, cystoscopy, and rectoscopy to rule out other sites of disease. All patients were operated by a gynecologic oncologist. Pathologic diagnosis of PDV was confirmed by 2 gynecologic pathologists (LR and OP). Invasive Paget\'s disease was defined as disease extending for at least 1 mm beyond the basement membrane. A positive margin was defined as Paget\'s cells within 1 mm of the surgical margin. No secondary surgical procedure or other adjuvant treatment (radiotherapy or medical treatment) were administered in such cases. Inguinal radiation was administered only in cases of nodal metastasis. Physical examination was scheduled every 6 months, with biopsy in case of suspicious of recurrent disease. The follow-up period was defined as the time between initial Paget\'s disease diagnosis and the date of the last contact. Descriptive statistics were used to summarize the patient demographic and clinical characteristics. Fisher exact test was used to compare patients with recurrence to those without recurrence, with respect to types of primary treatment received and margin status among those patients with primary surgery. 3. Results ========== Twenty-four patients were identified. The median age at diagnosis was 69.3 years (range 38--84 years). All were Caucasian. The most common presenting symptoms were itching, burning, and pruritis associated with vulvar lesions in 5 patients, pain associated with pruritus in 3 patients, 1 patient complained of only itching another only vulvar pruritis. In 14 records, the symptoms presented by the patients were not reported (Table [1](#T1){ref-type="table"}). ###### Symptoms.  Median duration of symptoms before diagnosis was 33.6 months (range 12--72 months). Twenty-three patients (96%) were primarily diagnosed and managed by a gynecologist, and 1 by a dermatologist (4%). Two patients had medical treatment with imiquimod and 5 fluorouracil before surgery without any benefit. All patients underwent surgery, including wide local excision 6 (25%), simple vulvectomy 8 (33%), and extended vulvectomy 10 (42%). Lymphadenectomy was performed in 2 cases (8%) and reconstructions with advancement flaps in 7 cases (29%). Wound dehiscence and urethral stenosis were found in 4 cases (17%) and 1 case (4%), respectively and were all managed conservatively (Table [2](#T2){ref-type="table"}). The median operative time was 130 minutes (range 35--230 minutes) and the median length of hospital stay was 7.8 day (range 3--16). No patient had blood transfusion during or after surgery. ###### Surgical data.  The median tumor area was 48.9 cm^2^ (range 2.5--143 cm^2^). At pathologic examination 4 patients (16.67%) were found to have invasive disease and 1 of the 2 patients with invasive disease who had lymphadenectomy was found to have a single inguinal node involvement (4.17%). Margin status was available for all patients and positive margins were found in 12 patients (50%). The presence of positive margins was neither related to the extent of surgery (5 among patients with extended vulvectomies versus 7 among patients with simple vulvectomy) nor to the presence of invasive disease (50% of positive margins in patients with invasive disease and 50% in patients without invasive disease). No patient received adjuvant treatment with radiotherapy after primary surgery. Diagnosis of synchronous cancer was made in 2 cases (vulvar squamous carcinoma, melanoma) and of metachronous cancer in 2 other cases (adenocarcinoma of the ampulla of Vater, 1 patient had endometrioid adenocarcinoma and urothelial carcinoma). Two patients had previously been diagnosed with breast cancer, 1 with breast cancer and ovarian cancer, 1 with bilinear myelodysplastic syndrome, and 1 with endometrioid adenocarcinoma (Table [3](#T3){ref-type="table"}). ###### Synchronous, metachronous cancer and previous diagnosis of other cancer diseases.  Median follow-up was 39 months, ranging from 1 to 240 months. One patient was lost at follow-up. Eight recurrences (33%) were observed, regardless of histological type or positive surgical margins. In fact, in 6 cases of recurrence the margins were not affected by the disease as compared to 12 cases in patients with positive margins (17%). In all but 1 case, recurrence remained noninvasive. Four recurrences (50%) occurred in patients with urothelial carcinoma and metachronous carcinoma of the papilla of Vater, with synchronous melanoma, and with previous breast and ovarian carcinoma. In all cases of recurrence surgery was the only treatment modality applied with complete resection of the lesion in 100% of cases. To date, 13 patients are free of Paget disease. Of these, 1 is affected by squamous cell carcinoma of the vulva and another 1 by bilinear myelodysplastic syndrome. Ten patients at follow-up are dead, 5 due to nononcological diseases. The cause of death of the other 5 is respectively: melanoma, adenocarcinoma of the ampulla of Vater, ovarian cancer, and urothelial carcinoma. Only 1 died because of invasive PDV. 4. Discussion ============= Clinicopathologic information regarding PDV is limited and controversial due to the rarity of the disease.^\[[@R2],[@R5]--[@R8]\]^ Since in the initial phase it does not present characteristic clinical features, there can be delayed diagnosis or misdiagnosis.^\[[@R4]--[@R6]\]^ The time between the first symptoms and the correct diagnosis of PDV can vary from a few to several months.^\[[@R3]--[@R6]\]^ In our patients, the date of onset of the vulvar symptoms precedes from 1 to 6 years the biopsy and the diagnosis of Paget\'s disease. Differential diagnoses may include cutaneous candidiasis, tinea cruris, seborrheic dermatitis, and psoriasis, Bowen disease, or melanoma (pigmented variant of EMPD),^\[[@R5]\]^ and for this reason VPD has to be diagnosed only with vulvar biopsy. In our series, 20 patients (83.33%) were affected by primary Paget disease, similar to the majority of series in the literature.^\[[@R3],[@R4],[@R6],[@R8],[@R15]--[@R17]\]^ The importance of immunohistochemistry for the diagnosis of EMPD has now been reiterated in all studies. It is also important to make possible differential diagnoses with other oncological diseases of the vulva, such as melanoma and squamous carcinoma. Finally, with immunohistochemistry we can highlight secondary forms of Paget\'s disease that underlie urinary and gastrointestinal tract carcinomas. This is also confirmed by our study despite its limitations; in fact, immunohistochemistry in our study is available in only 11 cases out of 24.^\[[@R1],[@R2],[@R5]--[@R9]\]^ In our series, 4 patients were found to have invasive disease and this finding is similar to that reported in other large series.^\[[@R2],[@R4],[@R6],[@R8],[@R15]--[@R17]\]^ Surgery is still considered the gold standard of treatment for patients with PDV,^\[[@R7]\]^ and it is a protective factor for overall survival.^\[[@R8]\]^ In our study all patients underwent surgery. In our series, primary management was medical in the only patient managed by a dermatologist, whereas when diagnosis was made by a gynecologist, surgical treatment was the preferred initial management.^\[[@R7],[@R8]\]^ Due to the extent of disease at the time of diagnosis, 7 patients required plastic reconstructive procedures after surgical resection and this data is higher than that reported in other series. Different reconstructive techniques may be applied, but in our centre, we preferred V-Y advancement flaps. The recurrence rate was high (33%) regardless of histological type or positive surgical margins. In fact, in 6 cases of recurrence the margins were not affected by the disease. In all cases of recurrence, the patients underwent surgery again. In the 2 cases of second and in the only case of third recurrence, surgery was still the therapy carried out. This finding is in line with the mainly surgical treatment of relapses which is reported in the literature.^\[[@R3]--[@R6],[@R8]\]^ Our retrospective study confirms that in case of recurrence the disease rarely become invasive, so once the presence of underlying or co-existent malignancy is excluded, we could opt for the use of Topical 5% imiquimod cream, now widely considered effective for high grade vulvar squamous cell intraepithelial lesions.^\[[@R20],[@R21]\]^ It is interesting to note that 4 recurrences occurred in patients with urothelial carcinoma and metachronous carcinoma of the papilla of Vater, with synchronous melanoma, and with a previous breast and ovarian carcinoma. Five of the 10 dead patients died due to non-oncological reasons, suggesting that Paget disease is not a fatal disease. Only 1 patient had a very aggressive disease leading to sudden system diffusion and death. In fact, the course is unpredictable. In situ forms often do not progress for many years, but in cases of invasion of the dermis the prognosis also becomes rapidly unfavourable.^\[[@R3],[@R9]\]^ In the remaining 4 patients death was due to associated cancer: melanoma, adenocarcinoma of the ampulla of Vater, ovarian cancer, and urothelial carcinoma. This confirms the data according to which the most aggressive EMPD is the secondary form or that PVD is always associated with the presence of other oncological pathologies. Despite the limitations of our study (limited number of patients retrospectively collected) we confirm that the diagnosis of Vulvar Paget\'s disease is often delayed, that recurrences are common independently of the presence of microinvasion and positive margins. Early recognition of secondary aggressive forms may reduce morbidity and mortality. Different alternative treatments to surgery such as imiquimod therapy,^\[[@R18]--[@R20]\]^ as already mentioned, or photodynamic treatment are under investigation.^\[[@R18]--[@R20]\]^ In our case study, unlike others reported in the literature,^\[[@R3],[@R4],[@R21]--[@R23]\]^ there was only 1 case of a patient with a very aggressive disease with rapid spread and death. Furthermore, our study highlights the important role of plastic surgery. Our study has no other novelties compared to other studies^\[[@R3],[@R4],[@R21]--[@R23]\]^ with a greater number of cases, but could serve to emphasize once again the delay with which this pathology is often diagnosed and the need to establish further therapeutic alternatives, considering the large number of relapses with repeated surgical treatments and consequent increase in morbidity due to repeated mutilations. We hope that research will soon be carried out on large series of cases with the implementation of guidelines for this pathology which, although rare, may have a course with vulvar functional impairment, and negative surgical outcomes. Author contributions ==================== **Investigation:** Rosalia Maria Rita Loiacono, Paola Traversi, Maria Assunta De Liso, Giulio Gargano, Anila Kardhashi, Roberta Francescato, Vera Loizzi, Marta Spinelli, Serena Lovascio, Maura Fiorito, Ondina Popescu. **Supervision:** Leonardo Resta, Ettore Cicinelli, Gennaro Cormio. Abbreviations: EMPD = extramammary Paget disease, MPD = mammary Paget disease, PDV = Paget disease of the vulva. How to cite this article: Loiacono RMR, Traversi P, Deliso MA, Gargano G, Kardhashi A, Francescato R, Loizzi V, Spinelli M, Lovascio S, Fiorito M, Resta L, Popescu O, Cicinelli E, Cormio G. Paget disease of the vulva an analysis of 24 cases. *Medicine*. 2019;98:41(e17018). The authors have no conflicts of interest to disclose.

Introduction {#Sec1} ============ Recent advances in genetic research have enabled us to identify individuals at risk for a wide variety of medical conditions due to their genetic makeup (Collins et al. [@CR10]). At the same time, these advances have created the need to educate and guide these individuals (Lerman et al. [@CR29]). Informing them of their hereditary risk and of the options for how to deal with this risk is the primary aim of genetic services (Wang et al. [@CR52]). Genetic services involve both genetic counseling and genetic testing; of these, genetic counseling in particular aims to enable at-risk individuals to accurately identify, understand and adaptively cope with their genetic risk (Biesecker [@CR2]; Pilnick & Dingwall [@CR40]). The National Society of Genetic Counselors' (NSGC) Task Force defines genetic counseling as "the process of helping people understand and adapt to medical, psychological, and familial applications of genetic contributions to disease" (Resta et al. [@CR41], p. 79). As such, genetic counselors are faced with three important tasks: (1) to interpret family and medical histories to enable risk assessment, (2) to educate counselees about issues related to heredity, preventive options (e.g., genetic testing), and personal risk, and (3) to facilitate informed decisions and adaptation to personal risk (cf. Trepanier et al. [@CR50]). The latter task may be considered the "core" (i.e., the desired outcome) of genetic counseling, with the former tasks in service of its fulfillment. Informed decision making and adaptation to personal risk, however, are abstract concepts that cannot easily be assessed. As such, several measures have been developed to assess the efficacy of genetic counseling. Kasparian, Wakefield and Meiser ([@CR26]) summarized 23 available measurement scales which include satisfaction, knowledge, psychological adjustment, and risk perception measures. Although each of these measures significantly contributes to our understanding of the effect of genetic counseling, risk perception measures (and especially risk perception accuracy) may be regarded as one central concept. Indeed, several influential models of health behavior, such as the Health Belief Model (Janz & Becker [@CR23]), the Protection Motivation Theory (Rogers [@CR43]), and the Extended Parallel Process Model (Witte [@CR54]), posit that adequate risk perception acts as a motivator to take (preventive) action and, as such, is a prerequisite of preventive behavior. Moreover, risk perception and risk perception accuracy have been shown to be related to several other important outcomes of genetic counseling, such as coping (Nordin et al. [@CR36]), worry (Hopwood et al. [@CR19]), and anxiety (Meiser et al. [@CR33]). The effect of genetic counseling on risk perception has been heavily examined during the past two decades, from early research into reproductive genetic counseling (e.g., Humphreys & Berkeley [@CR22]) to recent studies into genetic predispositions to cancer (e.g., Bjorvatn et al. [@CR3]). While these studies are valuable in their own right, few have investigated the effect of genetic counseling on risk perception *accuracy*. Indeed, to facilitate informed decision making and adaptation to personal risk, counselees must have *accurate* risk perceptions. In their 2002 meta-analysis, Meiser and Halliday ([@CR35]) identified only six studies that assessed the effects of genetic counseling on risk perception accuracy. Their meta-analysis showed that individuals at risk for breast cancer significantly perceive their own risk more accurately after genetic counseling. In particular, they observed an average increase of 24.3% of the participants who accurately estimated their personal risk after counseling. A systematic review by Butow and colleagues ([@CR6]) 1 year later confirmed the positive impact of genetic counseling in breast cancer risk perception accuracy, although 22--50% continued to overestimate their risk even after counseling. Research thus suggests that genetic counseling may indeed improve risk perception accuracy in some individuals. However, Meiser and Halliday ([@CR35]) and Butow et al. ([@CR6]) only included studies examining breast cancer risk. To date, there is no systematic review or meta-analysis which examines the effect of genetic counseling on perception of genetic risks in general. Thus, the purpose of the present review is twofold: (1) to provide an updated overview of the impact of genetic counseling on risk perception accuracy in papers published between January 2000 and February 2007, and (2) to extend the results of Meiser and Halliday's ([@CR35]) meta-analysis and Butow et al.'s (2003) systematic review to other genetic conditions. Methods {#Sec2} ======= Search Strategy {#Sec3} --------------- We searched the Pubmed, EMBASE, Web of Science, ERIC and PsycINFO databases. We also used the search engine Google Scholar to find papers and grey literature (literature not published in a journal---e.g., in press or under review---but nevertheless available on the internet) on risk perception accuracy and genetic counseling on the internet. To this end, we used the search term "(risk perception OR perceived risk OR perceived susceptibility OR susceptibility estimate OR risk estimate) AND (genetic counsel\* OR genetic risk OR familial risk OR genetic predisposition)." If available in the databases, we used the standardized, subject-related indexing terms of the concepts in the search term. We also searched the following journals manually: *Journal of Genetic Counseling*, *Patient Education and Counseling*, *Genetics in Medicine*, *Community Genetics*, *American Journal of Medical Genetics: Part A, Clinical Genetics*, *American Journal of Human Genetics*, and *Hereditary Cancer in Clinical Practice* from January 2000 until February 2007. Additionally, we performed key-author and reference list searches. Selecting Relevant Studies {#Sec4} -------------------------- The selection procedure was performed independently by two reviewers. The review process then consisted of three phases. During the first phase, papers were reviewed based on title only. In the second phase, the reviewers examined the abstracts of papers that could not be definitively included or excluded based on their title. Papers thought to be relevant to the review based on their abstracts were included; those judged irrelevant were excluded. In the third phase, the reviewers examined the papers included during the previous two phases for content. As recommended by the Cochrane guidelines (Higgins & Green [@CR17]), we erred on the safe side during the whole selection process; if in doubt, we included the paper for more extensive review in the subsequent phase. The following inclusion and exclusion criteria were used to determine whether papers were eligible for the review. Studies should be published after 2000 (i.e., upper limit of the 2002 Meiser and Halliday meta-analysis, since one goal of this review was to provide an update of that analysis); studies published before 2000 were excluded (*n* = 8; e.g., Evans et al. [@CR14]).Studies should focus on genetic risk perception; studies which did not (*n* = 9; e.g., Clementi et al. [@CR8]) or which discussed the effect of genetic mutations, prevalence, incidence, morbidity, or mortality only were excluded (*n* = 0).Studies should examine the effect of genetic counseling on risk perception *accuracy*; that is, should explicitly link perceived risk to objective risk estimates to examine whether they more closely align after (rather than before) counseling. Studies were excluded if they examined changes in risk perception without linking them to some objective risk estimate (*n* = 19; e.g., Burke et al. [@CR5]), if they investigated risk perception as a determinant of genetic counseling participation (*n* = 6; e.g., Collins et al. [@CR11]), or if they focused on the effectiveness of decision aids as compared to standard genetic counseling (*n* = 3; e.g., Warner et al. [@CR53]).To accurately assess whether genetic counseling affected risk perception accuracy, studies should employ either a prospective or a randomized control trial design. Studies using other designs were excluded (*n* = 12; e.g., Cull et al. [@CR12]).Risk perception accuracy should be assessed as a quantitative outcome measure; studies were excluded if they assessed risk perception as a qualitative outcome measure (*n* = 0).Studies should focus on at-risk individuals; those focusing on intermediaries (e.g., genetic counselors, nurses) would be excluded (*n* = 0).Studies should describe original research published in peer-reviewed journal in English. Studies describing secondary data or reviewing other studies, editorials, commentaries, book reviews, bibliographies, resources or policy documents were excluded (*n* = 5; e.g., Palmero et al. [@CR37]) as they provided too little detail. Data Abstraction {#Sec5} ---------------- Risk perception outcomes were abstracted by two authors independently, using standardized extraction forms. In the event of disagreement, the authors discussed the particular paper until they reached consensus. We abstracted the characteristics of the study, the participants and the genetic counseling session, as well as the results and quality of the study (cf. Higgins & Green [@CR17]). Results {#Sec6} ======= Selecting Relevant Studies {#Sec7} -------------------------- Figure [1](#Fig1){ref-type="fig"} presents the flowchart of the study selection process. From the initial sample of 3,798 eligible papers from the database searches and the 62 unique papers from the Google Scholar, journal, reference list and key author searches, a total of 82 papers were eligible for extensive review. Of these, 19 papers were included in the review. Table [1](#Tab1){ref-type="table"} lists the included papers and information about the study design, genetic counseling session content, criteria for risk perception accuracy, measurement time points, and finally the risk perception outcomes. Given the heterogeneity in the studies, we decided against pooling the studies in a meta-analysis. Fig. 1Flowchart of the Study Selection Process. Table 1Results of the reviewed studies*Study*NDesignContent of the genetic counseling sessionCriteria for risk perception accuracyMeasurement momentsRisk perception accuracy outcomesPercentage participants within each categoryDegree of overestimation of objective risk estimateAccurateUnder estimationOver estimationPerceived Estimate (SD)\*Average overestimation (SD)\**I. Studies examining percentage of patients within accuracy category*Bjorvatn et al.(2007)213Prospective- Family historyWithin two categories of counseled risk (on a six point scale)Pre-counseling81910- EmotionImmediately post-counseling8695- Heredity- Personal risk estimatep \< .001- Incidence- Disease development- SurveillanceHopwood et al. ([@CR18])158Prospective- Family historyCorrect counseled risk estimate (odds)Pre-counseling75238- Personal risk estimate3 months68920- Disease development6 months63925- Follow-up letter9 months6392512 months post counseling61925p \< .001Hopwood et al. ([@CR20])256ProspectiveNACorrect counseled risk estimatePre-counseling632791 month7121812 months post counseling73217*p* = nsHuiart et al. ([@CR21])397ProspectiveNACorrect counseled risk estimateLow risk:*p* \< .001Pre-counseling6.3093.71-7 days post counseling23.8076.3High risk:*p* = nsPre-counseling87.712.301-7 days post counseling89.510.50Lidén et al. ([@CR30])86ProspectiveNACorrect counseled risk estimatePre-counseling173647Post-counseling5418281 year follow-up283339p \< .01Lobb et al. ([@CR31])89Prospective RCT-GeneticsWithin category of counseled risk estimatePre-counseling502723- Genetic testingPost-counseling702010- Family history- Personal risk estimate- SurveillanceMeiser et al. ([@CR33])218Prospective- Personal risk estimateCorrect counseled risk estimatePre-counseling541234- Genetic testing12 months541431- Advice on early detection and surgery*p* = nsNordin et al. ([@CR36])63ProspectiveNACorrect counseled risk estimatePre-counseling183844Post-counseling571825Pieterse et al. ([@CR39])51Prospective- Personal risk estimateCorrect counseled risk estimatePre-counseling48NANAPost-counseling51NANA*p* = nsRimes et al. ([@CR42])150Prospective- Personal risk estimateWithin 10% of counseled risk estimatePre-counseling12.63.384.141.2- Genetics and cancer6 months184.0078.034.3- Population rates cancer*p* = nsP \< .0005- Recommendations- SurveillanceRothemund et al. ([@CR44])44Prospective RCT-Psychological counselingWithin 50% of counseled risk estimatePre-counseling4718.8Post-counseling4217.5-Risk assessmentCounselees39048- Information on menopauseControls381448, *p* = ns*p* = ns- Personal risk estimate- Genetic testing*II. Studies examining the degree of overestimation*Bowen et al. ([@CR4])211RCT- Family historyCorrect counseled risk estimatePre-counseling31(24)19- Personal risk estimate6 months postcounseling18(16)6- Incidencep \< .001- Other risk factors- Breast screeningCodori et al. ([@CR9])101Prospective-HeredityCorrect risk estimatePre-counseling52.47(21.3)30- Genetic testingImmediately post-counseling52.17(21.1)30- Meaning of possible test results *p* = ns- NO personal risk estimateGurmankin et al. (2005)108ProspectiveNACorrect counseled risk estimatePre-counseling61%(26)42%1-7 days follow up44% (24)19p \< .001Kaiser et al. ([@CR25])123Prospective-ScreeningCorrect counseled risk estimatePre-counseling15.5 (14.28)14.94-Description conditionPost-counseling8.34 (9.34)7.8- Personal risk estimatep \< .0005- Implications- Decisional supportKelly et al. ([@CR27])99Prospective- Family and personal historyCorrect counseled risk estimatePre-counseling72.6 (29.2)23- Heredity1-2 days post-counseling66.2 (30.7)16.6- Risks /benefits testing p \< .01- Efficacy health- Personal risk estimateKent et al. ([@CR28])90Prospective- Personal risk estimate- Correct counseled risk estimatePre-counseling55- Translation to lay terms3 months52- Coping/Prevention strategies6 months51*p* = nsTercyak et al. ([@CR48])129Prospective- Personal and family historyCorrect counseled risk estimatePre-counseling12.6(18.9)11.5- Genetic testingImmediately post-counseling8.9(16.7)7.8- Patient educationp \< .001- Personal risk estimateVan Dijk et al. (2003)241Prospective- HeredityCorrect counseled risk estimateLow risk:54.93-SurveillancePre-counseling83(21.31)- Genetic testingImmediately post-counseling5643.86- Personal risk estimate(21.03)High risk:Pre-counseling*p* = nsImmediately89Post-counseling89RCT = Randomized controlled trial; NA = not assessed or reported in this study. Percentages may not add up to 100% due to missing values or rounding.\*Means and (standard deviations) Content of the Counseling Session {#Sec8} --------------------------------- Concerning the content and quality of the genetic counseling sessions, four studies mentioned using a genetic counseling protocol (Bjorvatn et al. [@CR3]; Bowen et al. [@CR4]; Kaiser et al. [@CR25]; van Dijk et al. [@CR51]). Two mentioned using a standardized counseling script (Codori et al. [@CR9]; Tercyak et al. [@CR48]). An additional three used audiotapes as a content check of the counseling session (Hopwood et al. [@CR18]; Kelly et al. [@CR27]; Lobb et al. [@CR31]), while the remaining twelve did not mention the use of any protocol, standardized script or audio- or videotapes as a content check. In-depth analyses of the content (see Table [1](#Tab1){ref-type="table"}) revealed that a majority of the studies described counseling sessions with similar content. However, four studies did not provide a description of the counseling session at all (Hopwood et al. [@CR20]; Huiart et al. [@CR21]; Lidén et al. [@CR30]; Nordin et al. [@CR36]). Comparing the descriptions of the counseling sessions of the remaining fifteen studies to the recommendations of the NSGC Task Force, we observed that only six of these mentioned the first task, "interpretation of family and medical histories to enable risk assessment" (Bjorvatn et al. [@CR3]; Bowen et al. [@CR4]; Hopwood et al. [@CR18]; Kelly et al. [@CR27]; Pieterse et al. [@CR39]; Tercyak et al. [@CR48]; van Dijk et al. [@CR51]). Likewise, only five studies explicitly mentioned performing the second task, "educate counselees about issues related to heredity and treatment and preventive options" (Bjorvatn et al. [@CR3]; Codori et al. [@CR9]; Kelly et al. [@CR27]; Meiser et al. [@CR33]; van Dijk et al. [@CR51]). Although judging whether counselors "facilitated decision making an adaptation to personal risk" is difficult, we did observe six studies claiming to advise counselees on surveillance (Bjorvatn et al. [@CR3]; Kaiser et al. [@CR25]; Meiser et al. [@CR33]; Rimes et al. [@CR42]; Rothemund et al. [@CR44]; Tercyak et al. [@CR48]), which may be regarded as facilitating informed decisions. Risk Perception Accuracy {#Sec9} ------------------------ The included studies used two different types of measures to determine the effect of genetic counseling on risk perception accuracy: several studies reported changes in the proportion of individuals who accurately perceive their risk, while others reported the degree of overestimation or underestimation as a measure of risk perception accuracy. Where available, we report both types of measures (see Table [1](#Tab1){ref-type="table"}). Overall, the studies indicate that genetic counseling has a positive impact on risk perception accuracy (cf. Table [1](#Tab1){ref-type="table"}). However, some studies observed no effect on risk perception accuracy at all, or only for low-risk individuals (cf. Table [1](#Tab1){ref-type="table"}). The studies assessing the proportion of individuals who accurately estimated their risk (see Table [1](#Tab1){ref-type="table"}, subsection I) showed an average increase of approximately 25% (range: 2--55%) of counselees who correctly estimated their risk after counseling; from an average of 42% pre-counseling to an average of 58% post-counseling. However, on average 25% (range: 5--76%) continued to overestimate and 19.5% (range: 7--55%) continued to underestimate their risk even after counseling.[1](#Fn1){ref-type="fn"} Other studies which assessed changes in the average overestimation of participants' perceived risk (see Table [1](#Tab1){ref-type="table"}, subsection II) still observed an average overestimation of approximately 18% (range: 6--40%) after counseling, in comparison with 25% (range: 11.5--42%) before counseling. Across the studies, the average decrease in overestimation was approximately 8%.[2](#Fn2){ref-type="fn"} Linking Content to Risk Perception Accuracy {#Sec10} ------------------------------------------- Linking the outcome (i.e., risk perception accuracy) to the content of the counseling session (i.e., whether counselors performed the tasks as recommended by the NSGC Task Force), we observed that the studies in which the counselor gave information about family history and heredity as well as personal risk estimates positively influenced risk perception accuracy (Bjorvatn et al. [@CR3]; Bowen et al. [@CR4]; Hopwood et al. [@CR18]; Kelly et al. [@CR27]; Tercyak et al. [@CR48]), although this improvement was not significant in two studies (Pieterse et al. [@CR39]; van Dijk et al. [@CR51]). In contrast, the studies that did not mention giving counselees this information observed no significant improvement of risk perception accuracy as a result of genetic counseling (Codori et al. [@CR9]; Kent et al. [@CR28]; Meiser et al. [@CR33]; Rothemund et al. [@CR44]), with the exception of one study (Kaiser et al. [@CR25]). The results for the other two tasks were mixed. While some studies that educated counselees about heredity observed a positive impact on risk perception accuracy (Bjorvatn et al. [@CR3]; Kelly et al. [@CR27]; van Dijk et al. [@CR51]), others did not (Codori et al. [@CR9]; Meiser et al. [@CR33]). Similar results were observed for the third task of facilitating informed decision making and adaptation to personal risk. Three out of the six studies identified as performing this task observed a positive impact of genetic counseling on risk perception accuracy (Bjorvatn et al. [@CR3]; Rimes et al. [@CR42]; Tercyak et al. [@CR48]), while the other three did not (Kaiser et al. [@CR25]; Meiser et al. [@CR33]; Rothemund et al. [@CR44]). Discussion {#Sec11} ========== The purposes of this review were (1) to provide an updated overview of the impact of genetic counseling on risk perception accuracy from January 2000 until February 2007, and (2) to extend the Meiser and Halliday ([@CR35]) meta-analysis and the Butow et al. ([@CR6]) systematic review to other genetic conditions. Overall, the studies showed that an increased proportion of individuals correctly perceived their risk after counseling rather than before, and those who did not had smaller deviations from their objective risk than before counseling. These positive effects were sustained even at follow-up 1 year later. Some studies, however, observed no positive effect of genetic counseling, or only for low-risk individuals. These results are in line with those reported in the 2002 Meiser and Halliday meta-analysis and the 2003 systematic review conducted by Butow and colleagues. The research in the present review may shed some light on why some studies observe positive effects of genetic counseling on risk perception accuracy and others do not. First, one study (Codori et al. [@CR9]) that observed no effect explicitly mentioned that personal risk information was not communicated during the relevant counseling session. Second, the provision of information about the role of family history, as recommended by the NSGC Task Force, may provide an appropriate context in which counselees can make sense of the risk information (cf. Codori et al. [@CR9]), resulting in accurate risk perceptions. Third, some counselors may go to great lengths to explain risk information in terms the counselees can understand (cf. Kent et al. [@CR28]). Unfortunately, research has shown that verbal and numerical risk estimates often do not coincide. That is, verbal risk information results in more variability in risk perception than does numerical information (Gurmankin et al. [@CR16]). Bjorvatn et al. ([@CR3]), for example, observed incongruence between numerical and verbal measures of risk perception. Similarly, Hopwood et al. ([@CR18]) observed that counselees included a wide range of numerical risk estimates within the same verbal category. The significance of this is discussed below, where we present the implications of our study for clinical practice. Finally, several studies (Pieterse et al. [@CR39]; Rothemund et al. [@CR44]) that observed no effect of genetic counseling on risk perception accuracy had small sample sizes, and thus may not have observed a significant effect due to power limitations. The present review has several important implications for future research. First, we selected a large number of studies assessing risk perception changes as a result of genetic counseling. However, we had to exclude 19 of these studies because they did not explicitly link risk perception to an objective risk figure. Assuming that researchers are aware of these objective risk figures, future studies should link risk perception changes to objective risk figures to assess changes in risk perception *accuracy*. A second implication concerns the definition of risk perception accuracy, which differs between studies. For instance, in several studies accurate risk perception is defined as falling within a certain category (e.g., Bjorvatn et al. [@CR3]; Kelly et al. [@CR27]; Lidén et al. [@CR30]) or within 50% of the counseled risk (e.g., Pieterse et al. [@CR39]; Rothemund et al. [@CR44]), while the majority define it as the correct counseled risk estimate (e.g., Bowen et al. [@CR4]; Hopwood et al. [@CR18]; Tercyak et al. [@CR48]). Additionally, the reviewed studies based the counseled risk estimate on different methods, such as family history assessment (Huiart et al. [@CR21]), Gail's score (Bowen et al. [@CR4]), or the BRCAPRO procedure (Kelly et al. [@CR27]). These issues reduce our ability to compare the results of the studies, thereby lessening their value. Future researchers should define risk perception accuracy as correct counseled risk, and base their risk estimate on generally accepted and applied methods to allow for better interpretation of the results and comparison between studies. A third, related issue concerns the type of outcome measure used: several studies report changes in the proportion of individuals who correctly perceive their risk, while others report the degree of overestimation or underestimation as a measure of risk perception accuracy. Researchers are advised to include both measures in their studies, as both provide valuable information about the effect of genetic counseling on risk perception accuracy. Further, we observed that the quality of the genetic counseling descriptions (in those descriptions that were present) was poor. Although the counseling sessions were labeled as standardized, they were described in general terms, such as "discussion about the risk" and "information was given about how hereditary factors contribute to disease." These general descriptions leave room for substantial differences between counseling sessions. This is especially problematic given that perceptions of genetic risks before genetic counseling can determine the content of the counseling session (Julian Reynier et al. [@CR24]), which tends to alter patient outcomes (Lobb et al. [@CR31]). Differences in the quality of the counseling session content may well explain the fact that not all studies in the present review observed a positive effect on risk perception accuracy. Future studies should therefore try to link the content of the counseling session to risk perception to determine which feature of the session actually contributes to improved risk perception accuracy (cf. Pieterse et al. [@CR39], or Shiloh et al. [@CR46]). The present review provides some insight into how the content of the counseling session relates to risk perception accuracy. Indeed, the provision of information on the role of family history was observed to positively impact risk perception accuracy, perhaps because it creates a context in which the counselee can understand the information. Additionally, forcing numerical risk estimates to fit lay terms to aid counselees' understanding may lead to inaccurate risk perceptions (Kent et al., [@CR28]). A possible avenue for further research may be to link effectiveness to certain sociodemographic variables. We could then examine the influence of known psychological differences between certain groups, which is a more complex process and should thus occur later in time. By associating these psychological differences to the effectiveness of genetic counseling, we may be able to identify the processes responsible for the positive effect of genetic counseling on risk perception accuracy. Knowledge of such processes will enable us to match the session's content to these processes and thus to increase the session's effectiveness. Finally, we observed a relative lack of diversity in research on genetic counseling and genetic test result disclosure in terms of the genetic disorder under consideration. Although genetic counseling and testing can be effective for a variety of disorders (Biesecker [@CR2]; Lerman et al. [@CR29]; Pilnick & Dingwall [@CR40]), most recent studies focus on their impact on cancer risk perception, particularly breast cancer. Although genetic counseling on cancer has been shown to positively affect risk perception accuracy, this does not guarantee it will do the same for other genetic conditions. Extensive research is needed to assess whether genetic counseling also effectively enhances risk perceptions for other genetic predispositions. Based on the results, we have formulated some implications for practice. First, in accordance with the recommendations of the NSGC Task Force, we again strongly urge genetic counselors to discuss the role of family history and perform a family history assessment. We suggest that this information is an important factor in accurate risk perception because it may provide the necessary context in which counselees can understand the risk information. Indeed, the results seem to suggest that the provision of such information is positively related to risk perception accuracy. While this implication may seem redundant as it repeats the earlier recommendations by the NSGC Task Force, we nonetheless repeat it here since several studies in this review did not mention communicating this information to the counselee (Codori et al. [@CR9]; Meiser et al., [@CR33]; Kaiser et al. [@CR25]; Kent et al. [@CR28]; Rothemund et al. [@CR44]). Second, while explaining risk information in lay terms seems to be a useful strategy to help counselees to better understand their risk (cf. Trepanier et al. [@CR50]), the one study that explicitly mentioned doing so did not observe a significant effect on risk perception accuracy (Kent et al. [@CR28]). Moreover, there appears to be incongruency between verbal and numerical risk estimates (e.g., Bjorvatn et al. [@CR3]; Hopwood et al. [@CR18]). Both types of risk estimates, however, possess qualities that would make them especially suited for counseling. Compared to verbal risk estimates, numerical risk estimates have been shown to increase trust in (Gurmankin et al. [@CR15]) and satisfaction with (Berry et al. [@CR1]) the information. On the other hand, individuals have been shown to more readily use verbal information when describing their risk to others (Erev & Cohen [@CR13]) and when deciding on treatment (Teigen & Brun [@CR47]). We therefore advise genetic counselors to present numerical risk estimates first, as they are accurate, objective information. The patient may then be asked what that risk estimate means to him or her. The patient's verbal response will provide an opportunity for further discussion of the meaning and impact of the risk information. Genetic counselors should, however, be aware of the disadvantages of verbal information in accurately communicating risk information. A third, related implication concerns the presentation of numerical risk information. Research has shown that visual presentation of risk information (e.g., odds or percentages) may be better understood than written presentation formats. Indeed, there seems to be general agreement that graphical formats, in comparison with textual information, are better able to accurately communicate risk information (Schapira et al. [@CR45]; Timmermans et al. [@CR49]) although contradictory evidence has also been published (Parrot et al. [@CR38]). Furthermore, graphical information seems to have a larger impact on risk-avoiding behavior than textual information (Chua et al. [@CR7]). We therefore advise genetic counselors to use visual aids when communicating numerical risk information (cf. Tercyak et al. [@CR48]). Conclusions {#Sec12} =========== Overall, this review suggests that genetic counseling may have a positive impact on risk perception accuracy. It has also resulted in several implications for future research. First, future researchers should link risk perception changes to objective risk estimates to assess the effect of genetic counseling on risk perception *accuracy*. Researchers are advised to define risk perception accuracy as the correct counseled risk estimate instead of falling within a certain percentage of the counseled risk. Additionally, they should report both the proportion of individuals who correctly estimate their risk and the average overestimation of risk. Second, as the descriptions of the counseling sessions were generally poor, future research should include more detailed descriptions of these sessions, and link their content to risk perception outcomes to enable interpretation of the results. Finally, the effect of genetic counseling should be examined for a wider variety of hereditary conditions. Genetic counselors are advised to discuss the role of family history and perform a family history assessment to provide the necessary context in which counselees can understand the risk information. They should also use both verbal and numerical risk estimates to communicate personal risk information, and use visual aids when communicating numerical risk information. Note that the numbers reported here are not consistent; that is, 58% is not a 25% increase from 42%. This inconsistency is due to the fact that all three figures are means of the raw data, which do not have the property of consistency in such calculations. Note that the numbers reported here are not consistent; that is, 18% is not an 8% decrease from 25%. This inconsistency is due to the fact that all three figures are means of the raw data, which do not have the property of consistency in such calculations. This study was financially supported by Maastricht University and performed at the School for Public Health and Primary Care (Caphri). Caphri participates in the Netherlands School of Primary Care Research (CaRe), recognized by the Royal Dutch Academy of Science (KNAW) in 1995. Open Access {#d30e2586} =========== This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Case report {#cesec10} =========== An otherwise asymptomatic 55-year-old man with chronic hepatitis C, status postantiviral treatment with a normal alpha-fetoprotein (AFP), presented for routine followup ultrasound (US) screening to exclude hepatocellular carcinoma (HCC). Previous screening US demonstrated a diffusely fatty liver but no focal liver lesions to suggest HCC. US tests performed at the time of presentation showed multiple small round homogeneous hyperechoic liver masses in both lobes without associated mass effect or posterior acoustic enhancement ([Fig. 1](#fig1){ref-type="fig"}). No internal vascularity appeared on Doppler evaluation. Subsequent magnetic resonance imaging (MRI) showed multiple small, irregular areas of signal loss on the opposed-phase compared to the in-phase sequence ([Figs. 2A-D](#fig2ad){ref-type="fig"}), consistent with multifocal areas of focal fat deposition. A fat-suppressed T2-weighted fast-spin-echo (FSE) image sequence showed homogeneous liver parenchyma without any focal lesions ([Fig. 2E](#fig2e){ref-type="fig"}). Contrast-enhanced axial 3-D fat-suppressed fast-gradient-echo (GRE) images using gadobenate dimeglumine (MultiHance, Bracco) did not show arterial phase enhancement or washout on the portal venous phase ([Fig. 2F, G](#fig2fg){ref-type="fig"}) to suggest HCC. Since accurate characterization was done using MRI, a biopsy was not performed. The patient is currently being followed up using MRI and has not developed a HCC. Discussion {#cesec20} ========== Hepatic steatosis is a common finding with many different imaging characteristics. It is usually diffuse or geographic but can be focal and, rarely, multifocal ([@bib1]). When multifocal, it can mimic metastasis or HCC as seen on US and computed tomography (CT), which are used as screening tools for at-risk populations. If hepatic steatosis is not recognized, unnecessary biopsy can result.  Accurate diagnosis depends on detection of microscopic fat within the lesions, coupled with a lack of associated enhancing soft tissue. Fat seen on US is mostly hyperechoic, although in some areas it may be hypoechoic. Hyperechoic lesions seen on US can be classified based on presence or absence of a peripheral hypoechoic halo. Although some lesions have a variable appearance, the two most common lesions that are hyperechoic without a halo are focal fat and hemangioma ([@bib2], [@bib3], [@bib4]). Focal fatty infiltration often occurs in characteristic locations such as in the gall bladder fossa, adjacent to the falciform ligament, and anterior to the portal vein bifurcation in segment 4 ([@bib4], [@bib5]). Margins of focal fat are geographic, and vessels are seen to course normally through the area of fatty infiltration without displacement or encasement. A typical cavernous hemangioma is uniformly echogenic with sharp margins, without a capsule, and with a hypoechoic halo. Posterior acoustic enhancement may be seen if these are larger than 3 cm in size, but larger lesions can have atypical appearances on US ([@bib6]). On Doppler imaging, there is absent to minimal flow within, but these lesions are never hypervascular. Such an appearance is diagnostic and is seen in approximately 80% of hemangiomas. Both these lesions are benign, but when they are multifocal, further imaging is needed to exclude metastatic disease or multifocal HCC (especially if there is underlying malignancy, hepatitis, cirrhosis, or chronic liver disease). A hypoechoic halo around a liver lesion is suspicious for malignancy and was not present in our case ([@bib4]). Small HCCs have variable US echogenecity ([@bib4], [@bib7], [@bib8]). These can be hypoechoic, hyperechoic, or isoechoic, or can have a nodule-in-nodule appearance, with the hypoechoic pattern being the most common. The hyperechoic pattern is more common in hepatitis C-related cirrhosis and in patients under 69 years, and it has a trend towards lower AFP levels ([@bib9]). On CT, fat is hypodense relative to the liver, with attenuation values ranging from −10 to −100. A hypodense appearance on CT makes differentiation from focal liver tumor or metastatic disease difficult when the lesions are multifocal. A central core of normal-appearing hepatic tissue within focal areas of fatty replacement (reversal of the typical tumor necrosis pattern) suggests hepatic steatosis ([@bib10]). However, the most specific imaging technique to demonstrate the presence of fat is MRI. MRI using fat suppression and chemical shift imaging with opposed-phase sequences can detect even a small amount of fat within lesions. During the in-phase, lipid and water signals are additive, but in the opposed-phase, the signal consists of the water signal minus the lipid signal, allowing detection of small amounts of fat ([@bib11]). The differential for liver lesions with only intracellular fat, without associated soft tissue, is small and consists of benign lesions such as focal steatosis, multifocal nodular steatosis, and some adenomas. Focal nodular hyperplasia (FNH) may have areas of intracellular fat, but has associated soft tissue. Angiomyolipoma (AML) and adenoma can have both microscopic and macroscopic fat. HCCs have an associated soft-tissue component but can have both intracellular and macroscopic fat. In general, hepatic metastases (with the exception of liposarcomas) do not contain fat, though some cases have been reported. These are associated with a soft-tissue component ([@bib12]). Fat is demonstrated in 10% of hepatic adenomas on CT and in 36% to 77% on MRI ([@bib13], [@bib14], [@bib15], [@bib16]). However, they can be heterogeneous from necrosis, old hemorrhage, or calcification and can demonstrate early arterial enhancement with a centripetal pattern due to subcapsular feeding vessels. Enhancement does not persist because of arteriovenous shunting. Lack of uptake in the delayed phase using hepatocyte-specific contrast agents (gadolinium benzyloxypropionictetraacetate, Gd-BOPTA) distinguishes them from FNH ([@bib13]). HCC is the most common malignant primary liver tumor. Fatty change in HCCs is seen in up to 35% of small HCCs ([@bib12], [@bib17]). Smaller HCCs usually have diffuse fatty change, while larger HCCs have patchy fatty change ([@bib5]). Hyperintensity on T1-weighted images may relate to intratumoral fat or hemorrhage, or to copper or zinc content of the surrounding liver parenchyma ([@bib18]). FNH is the second most common benign liver lesion after hemangioma; it is often incidental, and occurs in young asymptomatic women. On US, these are isoechoic and may show the characteristic "spoke-wheel― vascularity on color Doppler US ([@bib19]). Fat within these lesions is rare, usually patchy, rarely peripheral, and may be associated with underlying fatty liver ([@bib20]). On CT and MRI, FNH demonstrates intense homogeneous enhancement with rapid washout. A central scar bright on T2-weighted images with delayed enhancement is characteristic ([@bib12]). Hepatic involvement in Langerhans-cell histiocytosis is uncommon; it is seen in extensive disease and is associated with a high mortality rate. These lesions can look identical to multinodular hepatic steatosis but are characteristically located in the periportal regions with mass effect. This diagnosis should be considered if there are stigmata of the disease such as cutaneous, osseous, or pulmonary lesions with diabetes insipidus or pituitary insufficiency ([@bib12], [@bib21]). Hepatic lipomas are rare, usually asymptomatic, with imaging characteristics typical of lipomas. Fat is present in only 50% of hepatic AMLs ([@bib22]). Adrenal rest tumors are usually subcapsular, and they often contain macroscopic fat with briskly enhancing soft tissue. They can be difficult to distinguish from fat-containing HCCs ([@bib12]). In conclusion, we present the US and MRI findings for multifocal nodular steatosis and review the differential diagnosis of liver lesions containing microscopic fat. In high-risk patients, liver masses seen on screening raise the possibility of HCC. MRI has an advantage over other modalities because of its superior soft-tissue characterization. The use of various MRI sequences allows for a confident diagnosis of multifocal nodular steatosis. Awareness of imaging findings of this entity can avoid further workup and liver biopsy. Published: July xx, 2010 {#fig1} {#fig2ad} {#fig2e} {#fig2fg} [^1]: All the authors are at the University of Washington, Seattle WA, and Drs. Bhargava and Medverd are also associated with the VA Puget Sound Health Care System.

1. Introduction {#sec1-ijerph-16-01821} =============== In the education field, physical education is a subject that can contribute to improving the well-being and health of children and adolescents. The concept of quality physical education, which is understood as an interrelated system of inclusive and active teaching and learning, must be considered a key framework for integral approaches (i.e., education and health) \[[@B1-ijerph-16-01821]\]. It can also be seen as a physically active teaching and learning experience that can positively impact students' psychomotor abilities, cognitive comprehension and social and affective aptitudes \[[@B2-ijerph-16-01821]\]. Moreover, in our view, quality physical education could be grouped in the social and emotional learning category. Quality physical education aims to achieve an integral education commitment \[[@B3-ijerph-16-01821],[@B4-ijerph-16-01821]\] that allows students to be physically literate \[[@B5-ijerph-16-01821],[@B6-ijerph-16-01821]\]. Physical literacy is the pillar of quality physical education, and can be defined as the motivation and cognitive, physical and affective competence necessary to encourage and preserve an active attitude in life, enabling a positive development of the aptitude to achieve, understand and use decisions about one's health efficiently \[[@B7-ijerph-16-01821]\]. Students who are physically literate intrinsically value their own psychomotor capabilities, as well as the contribution of these abilities to well-being and health \[[@B6-ijerph-16-01821]\]. The connection between health and physical activity is widely accepted \[[@B8-ijerph-16-01821],[@B9-ijerph-16-01821]\]. However, the effects of physical activity on health in the educational context should be deepened through experimental studies. Designing teaching and learning processes in this area will support physical, psychological, emotional and social development \[[@B10-ijerph-16-01821]\]. The synergy between the practice of physical-sport activity together with physical and psychological health is a gradually growing interest area for education researchers \[[@B11-ijerph-16-01821],[@B12-ijerph-16-01821],[@B13-ijerph-16-01821],[@B14-ijerph-16-01821]\]. Moreover, different investigations in the education framework of the evolution of quality physical education have emphasised the need for methodological change \[[@B4-ijerph-16-01821],[@B15-ijerph-16-01821]\]. Several pedagogical models share the same features \[[@B16-ijerph-16-01821]\]. This study is based on quality physical education, manifested through a specific sport education model \[[@B17-ijerph-16-01821]\]. Sport education is a pedagogical model that uses essential features of sports (seasons, competitions, membership, data register, culminating event and festivity), and aims to achieve the inclusive goal of all students living real and meaningful sport experiences in physical education. In addition, this model aspires to develop competence, enthusiasm and a physical-sport culture in students \[[@B18-ijerph-16-01821]\]. The pedagogical potential of the sport education model, if correctly implemented \[[@B19-ijerph-16-01821]\], results in benefits at a physical level \[[@B20-ijerph-16-01821],[@B21-ijerph-16-01821],[@B22-ijerph-16-01821]\]. Similarly, it has been shown to have a positive impact on psychological variables in adolescents. These positive benefits include: basic psychological needs \[[@B23-ijerph-16-01821]\]; improvement in competence \[[@B24-ijerph-16-01821]\] and the feeling of belonging to a group \[[@B25-ijerph-16-01821]\]; decrease in attitudes towards violence and improvements in social responsibility and participants' relationships \[[@B25-ijerph-16-01821],[@B26-ijerph-16-01821]\]; more self-determined behaviour \[[@B27-ijerph-16-01821]\]; improvements in friendship and sport goals \[[@B24-ijerph-16-01821]\]; decrease in aggressive behaviour and improvements in friendship relationships \[[@B25-ijerph-16-01821],[@B28-ijerph-16-01821]\]; positive changes in the perception of the social climate \[[@B29-ijerph-16-01821]\]; improvement in social relationships \[[@B30-ijerph-16-01821]\]; improvement in trait emotional intelligence and motivational mediators \[[@B31-ijerph-16-01821]\]; and improvement in sport culture and enthusiasm. However, no benefits in terms of life satisfaction have been found \[[@B32-ijerph-16-01821]\]. By making use of sport, these studies provide evidence of a meaningful and positive impact on the psychological and physical development of the school-age population \[[@B11-ijerph-16-01821],[@B33-ijerph-16-01821]\]. Health is generally determined by several physical--biological, psychological and social indicators \[[@B34-ijerph-16-01821]\]. Therefore, good health is a fundamental dimension in personal and social progress, and an important sphere in quality of life \[[@B35-ijerph-16-01821]\]. Approached from the perspective of positive health, a state of wellness encourages individuals to reach complete social and psychological development \[[@B36-ijerph-16-01821]\]. The World Health Organization (WHO) aims to promote physical and psychological health \[[@B34-ijerph-16-01821]\] that supports a good quality of life \[[@B37-ijerph-16-01821]\]. The construct of subjective well-being is among the factors that affect health. Consequently, research on the influence of subjective well-being in different social and educational contexts has received increasing attention in recent decades \[[@B38-ijerph-16-01821]\]. Subjective well-being comprises two main factors: a cognitive aspect (satisfaction with one's own life) and an affective aspect (positive and negative affect) \[[@B39-ijerph-16-01821],[@B40-ijerph-16-01821]\]. The cognitive side of well-being reflects the assessment of how individuals process information in their lives \[[@B41-ijerph-16-01821]\]. The affective side of well-being implies a hedonistic individual balance; that is, how often individuals experience positive and negative emotions \[[@B39-ijerph-16-01821],[@B40-ijerph-16-01821]\]. Recent research has focused on studying the effects of positive psychological variables on personal and social development \[[@B42-ijerph-16-01821]\]. These studies have been categorised as positive psychology \[[@B43-ijerph-16-01821]\]. Emotional intelligence is a positive variable that currently has broad support because of its close connections to subjective well-being and physical and mental health \[[@B38-ijerph-16-01821],[@B44-ijerph-16-01821],[@B45-ijerph-16-01821]\]. Variables such as social anxiety have a negative effect on subjective well-being \[[@B46-ijerph-16-01821]\]. In this sense, social anxiety can be defined as a person's constant fear of one or more social or performance situations, in which they are exposed to unknown persons or the possible scrutiny of other people \[[@B47-ijerph-16-01821]\]. Social anxiety has a negative impact on subjective well-being in adolescents because of the anguish individuals may feel \[[@B48-ijerph-16-01821]\], which in turn may negatively affect the quality of their interpersonal relationships \[[@B49-ijerph-16-01821],[@B50-ijerph-16-01821]\]. We consider that education should promote social and emotional learning, which the WHO defines as a heterogeneous set of life skills, as this is a potential factor that supports and encourages mental health \[[@B51-ijerph-16-01821]\]. Scholars in favour of this teaching proposal argue that emotional education may also promote public health \[[@B52-ijerph-16-01821],[@B53-ijerph-16-01821]\], because it's ultimate goal is improvement of the general quality of health and well-being in citizens. In the school context, many researchers claim that a key purpose of education is to improve peoples' lives so that they can reach an optimal degree of personal happiness and well-being in adulthood \[[@B54-ijerph-16-01821]\]. This suggests that a healthy pedagogical and psychological school environment may facilitate students' positive adjustment; therefore, such an environment is essential for the development of well-being in children and adolescents \[[@B55-ijerph-16-01821]\]. The theoretical and practical justification for this study was rooted in the work of various authors who developed educational interventions based on the sport education model and recommended that further research should evaluate the impact of this model on the promotion of optimal personal and social development \[[@B25-ijerph-16-01821],[@B26-ijerph-16-01821],[@B56-ijerph-16-01821],[@B57-ijerph-16-01821]\]. In this sense, Metzler's \[[@B16-ijerph-16-01821]\] contributions are very relevant, stating that a sports model teaching program is mainly focused on different domains \[[@B16-ijerph-16-01821]\]: affective, cognitive and motor. In line with this statement, our study focuses on the affective domain. We agree with several previous authors \[[@B39-ijerph-16-01821],[@B40-ijerph-16-01821]\] that subjective well-being is a key variable that influences balanced personal and social development. In addition, the existing relationship between subjective well-being and trait emotional intelligence suggests that it is necessary to further explore this topic. Social anxiety generates inappropriate social relationships in adolescents \[[@B46-ijerph-16-01821],[@B48-ijerph-16-01821]\]. Given the positive effects of the sport education model on social relationships \[[@B30-ijerph-16-01821]\], it is possible that such interventions may reduce social anxiety. This study aimed to evaluate the impact of a pilot programme based on the sport education model on the three variables: subjective well-being, trait emotional intelligence and social anxiety. The hypotheses focused on the assumptions that the programme will result in improvements in our participants' subjective well-being (Hypothesis 1), trait emotional intelligence (Hypothesis 2) and social anxiety (Hypothesis 3). 2. Methods {#sec2-ijerph-16-01821} ========== 2.1. Participants {#sec2dot1-ijerph-16-01821} ----------------- This study used non-probability incidental or accessibility sampling. The sample comprised 113 students in compulsory secondary education aged 12--15 years (mean age (*M*) = 13.82 years, standard deviation (*SD*) = 0.79 years). The research was conducted in a state school with students from five class groups. The control group comprised 44 students (two class groups) and the experimental group included 69 students (three class groups). The experimental and control group assignment was based on a cluster-randomised controlled trial. The gender distribution was 64 (57%) boys and 49 (43%) girls ([Table 1](#ijerph-16-01821-t001){ref-type="table"}). The main inclusion criterion was parental consent. The exclusion criteria were: (a) attending less than 80% of the sessions of the intervention programme (less than 13 sessions); (b) students with special educational needs associated with intellectual disability; and (c) students that were removed from school for disciplinary reasons. 2.2. Procedure {#sec2dot2-ijerph-16-01821} -------------- We requested the collaboration of the educational centre (Spain) in this study. The management board of the participating school was contacted to obtain their approval and authorisation for the study. Permission was also obtained from the families of the participating students, and from the teaching staff and school council. This study respected the relevant ethical values and guaranteed participants' confidentiality and anonymity. In addition, this study was developed in accordance with the Declaration of Helsinki regarding human experimentation. The study procedures were conducted in accordance with the Universidad de Castilla-La Mancha code of ethics. This study used a quasi-experimental design with repeated pre-test and post-test measures and a control group. The study was conducted over three stages. First, before the intervention began, the assessment instruments (pre-test evaluation) were handed out for completion in the first 20 minutes of two sessions, to avoid burdening the students. Next, the programme based on the sport education model was implemented. The programme sessions took place during the second term of the school year. Finally, the assessment instruments were completed a second time (post-test evaluation). The independent variable was the intervention programme, and the dependent variables were subjective well-being, trait emotional intelligence and social anxiety. 2.3. Measures {#sec2dot3-ijerph-16-01821} ------------- Four evaluation instruments were used to assess the variables in this study, under the psychometric parameters of reliability and validity. The Kidscreen-10 Index \[[@B58-ijerph-16-01821]\] was used to assess subjective health-related quality of life and well-being. This 10-item scale was designed for children and adolescents aged 8--18 years. Each item has five response options, ranging from 'Never' to 'Always' or 'Not at all' to 'Extremely'. The 10 items cover: affective symptoms of depressed mood; cognitive symptoms of disturbed concentration; psychovegetative aspects of vitality, energy and feeling well; and psychosocial aspects correlated with mental health, such as the ability to experience fun with friends or getting along well with others at school. The adapted version of the questionnaire used in this study has adequate internal consistency reliability (*Cronbach's alpha* = 0.82) and test-retest stability (*r* = 0.73; *ICC* = 0.72) \[[@B59-ijerph-16-01821]\]. The Positive and Negative Affect Schedule \[[@B60-ijerph-16-01821]\] was used to assess participants' positive and negative affect. The Spanish version of this scale for children and adolescents was validated by Sandín \[[@B61-ijerph-16-01821]\]. This scale comprises 20 items on two dimensions: positive affect and negative affect. Each subscale contains 10 items. The questionnaire is completed by participants based on the way they normally feel and behave. The scale has three response options: 'Never' = 1, 'Sometimes' = 2, and 'Many times' = 3. We used the Trait Emotional Intelligence Questionnaire Adolescents Short Form (TEIQue-ASF) \[[@B62-ijerph-16-01821]\] (adapted into Spanish in its abridged version for teenagers by Ferrando and Serra \[[@B63-ijerph-16-01821]\]) to evaluate trait emotional intelligence based on the theoretical model of Petrides and Furnham \[[@B64-ijerph-16-01821]\]. The 30 items that make up the TEIQue-ASF are scored on a 7-point Likert scale (1 = 'Completely disagree' to 7 = 'Completely agree'). The general emotional intelligence score of the total scale is obtained by summing the 30 items. Finally, participants completed the Social Anxiety Scale for Adolescents (SAS-A) \[[@B65-ijerph-16-01821]\]. The SAS-A comprises 22 items; 18 items are self-descriptive and four are distracting elements that are not taken into account for the score. The SAS-A contains three subscales: (a) fear of negative evaluation (eight items), (b) anxiety and social avoidance before strangers or new social situations (six items) and (c) anxiety and social avoidance in social situations in general (four items). Responses are on a 5-point Likert-type scale from 1 ('Never') to 5 ('Always'). In addition, a global index of social anxiety (SAS-T) is obtained by summing the scores for the items (excluding neutral items). High scores reflect high levels of social anxiety \[[@B65-ijerph-16-01821]\]. The scale was adapted to the Spanish population by Olivares, Ruiz, Hidalgo, García-López, Rosa and Piqueras \[[@B66-ijerph-16-01821]\]. Only the SAS-T score was used in this study. 2.4. Intervention Programme {#sec2dot4-ijerph-16-01821} --------------------------- The physical-sport programme was completed following the sport education model structure \[[@B17-ijerph-16-01821]\]: (1) *season*: lengthy didactic units; (2) *membership*: development of a team spirit and cooperation; (3) *regular competition*: showing technical--tactical abilities; (4) *data register*: giving evidence of and analysing the process that has been followed; and (5) *festivity*: a festive atmosphere. This highlighted other important education aspects such as: cooperative learning; autonomy and personal initiative; positive interdependence; and self-management of responsibility roles in conflict resolution (i.e., referee and coach). This helped to make the sport experience more real and positive, including how students transferred responsibilities by means of organisation roles (i.e., referee and scorer), team roles (i.e., coach and physical trainer) and how sport content was modified when adapted to the students \[[@B17-ijerph-16-01821]\]. Hastie and Casey's guidelines were followed for the design and validation of the programme \[[@B19-ijerph-16-01821]\] (p. 423): (a) thoroughly detailed curricular elements; (b) precise certification of the applied model; and (c) an in-depth explanation of the context of the programme. The intervention programme was implemented in the experimental group following sequencing of content and activities in three stages (initial, intermediate and final) over 16 sessions ([Table 2](#ijerph-16-01821-t002){ref-type="table"}). This pilot programme was developed to reflect the teaching hours of the physical education subject, which covers 16 55 minute sessions (2--3 sessions per week for 6 weeks). The total duration was considered sufficient to analyse the possible effects of the programme on the dependent variables, as indicated by previous research \[[@B67-ijerph-16-01821]\]. The educational intervention applied to the experimental group consisted of a didactic unit that used an alternative sport, called ringo \[[@B68-ijerph-16-01821],[@B69-ijerph-16-01821]\]. Ringo is an alternative, modified and reduced sport of divided court and net. It is played with a hoop (ringo) and a volleyball net. The objective is to score when the ringo falls on the opposite court ([Figure 1](#ijerph-16-01821-f001){ref-type="fig"}). In this pilot programme, the application of an alternative sport that was novel and unknown to students meant that everyone started with the same theoretical and practical sports knowledge, and there were few initial differences in their levels of technical--tactical sports skill. Alternative sports are characterised by being motivating, cooperative, socialising and adapted to participants' characteristics. The selection and organisation of teams (five teams per classroom) was developed by drawing lots. In addition, different responsibility roles were assigned to participating students: player; referee; coach--captain; physical trainer; person responsible for statistics and reports; and member of the discipline and organisation committee. An essential rule in the development of the pilot programme was that all students would actively participate in the programme with the assignment of two roles (player role and another responsibility role). The pilot programme also used various learning and curricular resources (self-designed portfolio, worksheets and reports) that had been used by other authors \[[@B70-ijerph-16-01821]\]. For the control group, a didactic unit of traditional collective sport with a conventional teaching style was developed \[[@B71-ijerph-16-01821]\]. This traditional teaching model aimed to improve students' technical motor skills only. In the teaching--learning process, the teacher assumed a managerial role and the students adopted passive individual roles limited to following the directive instructions of the teacher. This intervention consisted of 12 55 minute sessions (two sessions per week for 6 weeks). The first nine sessions were aimed at learning the technical fundamentals of basketball (pot, dribbling, passing, throwing and receiving) through a task assignment teaching style \[[@B71-ijerph-16-01821]\]. These traditional sports sessions were based on a 10 minute warm-up, 40 minute main session that included explanations and basketball practice and a 5-minute warm down in which stretching was performed. During these sessions, all tasks were directed by the teacher without students' participation. The last three sessions were dedicated to team competition. Two compulsory secondary education teachers, both with advanced degrees in Sports Science participated in this research. The first teacher (with a Master's of Science in Psychology) participated in the design and implementation of the pilot programme. The second teacher developed an intervention based on the traditional model. Both teachers received a 10 hour training course on the specific theoretical and practical aspects of each teaching model. In addition, supervision and tutoring was provided by a researcher expert in the sports education model and a researcher expert in the traditional education model. This tutoring consisted of: (a) session-by-session analysis during the intervention programmes; (b) telephone conversations and emails to resolve doubts, concerns and problems; and (c) weekly visits to the teaching centre. In these visits, the experts visited the centre randomly, without prior notice, with the objectives of: verifying that there were no gaps between what was planned and what was implemented, and checking that the teaching models were applied with all of their characteristics. 2.5. Data Analysis {#sec2dot5-ijerph-16-01821} ------------------ After fulfilment of the requirements of normality and homoscedasticity was verified, we examined the distribution of the data for a univariate normality analysis. The results showed asymmetry and kurtosis values lower than 1.2. Next, we calculated the reliability coefficients (Cronbach's alpha, composite reliability, average variance extracted and McDonald's omega coefficient) to obtain reliability evidence. Then, to determine the impact of the programme, descriptive analyses (mean and SD) and analyses of variance (ANOVA) were performed with the scores collected in the pre-test stage. Subsequently, descriptive analyses and analyses of covariance (ANCOVA) were used with post-test scores to determine the impact of the programme on each of the variables. Bonferroni correction was applied for multiple comparisons. For all analyses, a *p*-value \<0.05 was considered to indicate statistical significance. After application of Bonferroni correction, a *p*-value \<0.012 was considered significant. The effect size (*µ*^2^) of the differences was calculated using partial eta-squared \[[@B72-ijerph-16-01821]\]. The effect size was analysed based on four ranges: 0--0.009, negligible; 0.010--0.089, low-effect size; 0.090--0.249, medium-effect size; and \>0.250, big-effect size \[[@B72-ijerph-16-01821]\]. The data were analysed with SPSS version 24.0 (IBM Corp., Armonk, NY, USA). 3. Results {#sec3-ijerph-16-01821} ========== 3.1. Reliability {#sec3dot1-ijerph-16-01821} ---------------- In this study, we used well-established measures with appropriate psychometric properties ([Table 3](#ijerph-16-01821-t003){ref-type="table"}). 3.2. Effects of the Programme {#sec3dot2-ijerph-16-01821} ----------------------------- The pre-test MANOVA results did not reveal statistically significant differences between the groups prior to the intervention, Wilks' Lambda, *Λ* = 0.571, *F* (5, 108) = 0.739, *p* = 0.333, with a small effect size (*η2* = 0.062, *r* = 0.11). The ANOVA using the pre-test scores ([Table 4](#ijerph-16-01821-t004){ref-type="table"}) revealed no statistically significant differences in any of the dependent variables before the programme began, except for a significantly higher score for trait emotional intelligence in the experimental group compared with the control group. The size of the effect was low for trait emotional intelligence (*µ*^2^ = 0.009). Applying Bonferroni correction showed no significant differences in any of the variables. Results from the pre-test--post-test MANCOVA revealed significant differences between the two conditions, Wilks' Lambda, *Λ* = 0.899, *F* (5, 108) = 5.295, *p* = 0.003, with an average effect size (*η2* = 0.267, *r* = 0.32). Next, we performed ANCOVA for the dependent variables using the post-test scores. To assess the magnitude of these differences, the effect size for each variable was calculated by partial eta-squared ([Table 4](#ijerph-16-01821-t004){ref-type="table"}). ### 3.2.1. Effects on Subjective Well-Being {#sec3dot2dot1-ijerph-16-01821} There was no significant improvement in post-test health-related quality of life in the experimental group ([Table 4](#ijerph-16-01821-t004){ref-type="table"}). The experimental group did not show a significant increase in positive affect scores after testing. We confirmed a significant decrease in post-test negative affect scores in the experimental group ([Table 4](#ijerph-16-01821-t004){ref-type="table"}), with a medium effect size (*µ*^2^ = 0.123; partial eta-squared). ### 3.2.2. Effects on Trait Emotional Intelligence {#sec3dot2dot2-ijerph-16-01821} The analysis revealed significant improvements in trait emotional intelligence in the experimental group after the programme, with a medium effect size (*µ*^2^ = 0.241) ([Table 4](#ijerph-16-01821-t004){ref-type="table"}). ### 3.2.3. Effects on Social Anxiety {#sec3dot2dot3-ijerph-16-01821} There were no significant differences between the experimental and the control groups in SAS-T scores ([Table 4](#ijerph-16-01821-t004){ref-type="table"}). 4. Discussion {#sec4-ijerph-16-01821} ============= This study evaluated the impact of a pilot programme based on the sports education model on compulsory secondary education students' subjective well-being, trait emotional intelligence and social anxiety. The preliminary results obtained in this study revealed significant improvement of a specific indicator of subjective well-being (NA) in the experimental group after the pilot programme. The experimental group did not show a significant improvement in health-related quality of life when compared with the control group. Our results are consistent with the findings reported in other studies \[[@B32-ijerph-16-01821]\], which did not confirm significant benefits for life satisfaction among adolescents following a sport education-based experience. Our findings also indicated that the programme showed a significant decrease in negative affect, improving an indicator of the affective component of subjective well-being (i.e., a decrease in negative emotions) \[[@B39-ijerph-16-01821],[@B40-ijerph-16-01821]\]. These results partially verify Hypothesis 1, and highlight the importance of further research in this context. This is consistent with the findings reported in other studies \[[@B32-ijerph-16-01821]\], as previous studies established a connection between physical activity and subjective well-being \[[@B73-ijerph-16-01821]\]. These findings are also consistent with research that argues that active, inclusive and effective teaching and learning processes applied within a quality physical education framework fosters a motivating school climate in affective and psychological terms \[[@B1-ijerph-16-01821],[@B74-ijerph-16-01821]\]. Pedagogical and methodological aspects highlighted by this intervention pilot programme (e.g., cooperative learning, a feeling of membership to a team, positive interdependence and self-management or autonomy/use of responsibility roles) could have influenced these results. Furthermore, a motivating school context, enabled by the implementation of the sport education model \[[@B31-ijerph-16-01821]\], may also strengthen affective bonding in adolescents \[[@B12-ijerph-16-01821]\]. Significant improvement in trait emotional intelligence was observed in the experimental group after the programme, which confirmed Hypothesis 2. These results are also consistent with those reported by other authors \[[@B31-ijerph-16-01821]\]. The relationship between trait emotional intelligence and subjective well-being \[[@B38-ijerph-16-01821],[@B45-ijerph-16-01821]\], as well as that between trait emotional intelligence and physical and psychological health \[[@B44-ijerph-16-01821]\], may trigger these improvements in adolescents. This indicates that good trait emotional intelligence promotes positive emotional states and a reduction of negative moods, thereby positively impacting well-being and health \[[@B45-ijerph-16-01821]\]. We found no significant improvement in students' social anxiety, meaning that we could not confirm Hypothesis 3. Although this is similar to the results obtained by different authors for social relationships variables \[[@B23-ijerph-16-01821],[@B24-ijerph-16-01821]\], our results contradicted the findings of other studies \[[@B25-ijerph-16-01821],[@B26-ijerph-16-01821],[@B28-ijerph-16-01821],[@B30-ijerph-16-01821]\]. Further research on the effects of the sport education model is therefore necessary, especially given the theoretical specificity of social anxiety and its incidence in social relationships among adolescents. In addition, social anxiety can present opposing consequences. It can have positive effects on social relationships for some individuals, whereas it can have negative effects on others, characterised by anguish and social avoidance \[[@B65-ijerph-16-01821]\]. However, in our opinion this study has been very exhaustive in the evaluation methodology of the intervention program (including the Bonferroni corrections). In this sense, we have not found any study on the effectiveness of the sport education model that uses these statistical corrections. This fact could be influencing the comparison of our results with those obtained in other similar researches on the sport education model as a teaching model. Despite these promising results, this study had some limitations. First, the sampling procedure was chosen for reasons of convenience and not by random procedures. However, allocating students to either the experimental or control group was performed randomly based on the class group to which they belonged. Second, it would be necessary to increase the sample size to minimise potential biases in the results and increase the generalisability. Third, the instruments used were self-reported, and the results might have been influenced by bias related to social desirability in adolescents. It would be necessary to use high-performance tests or hetero-evaluation to minimise such bias. Similarly, differences in the trait emotional intelligence pre-test scores between the experimental and control groups might have had an impact on our results. Finally, it is necessary to highlight the difficulties encountered when following all of the recommendations for the implementation of the sport education model \[[@B19-ijerph-16-01821]\]. Similarly, it would be necessary to include session analysis procedure in order to evaluate if the main principles of the model were followed by the teachers \[[@B75-ijerph-16-01821]\]. Several aspects can be suggested regarding future lines of investigation, such as increasing the number of participants and diversifying their sociocultural background. It may also be worthwhile analysing the impact of the programme on other variables, such as academic performance and social and school adjustment. Similarly, to study the effects on depression with the use of biological correlates (HPA markers, cortisol immunitarian parameters, etc.). In addition, it would be interesting to conduct a follow-up evaluation to assess the sustainability of the effects of the programme. This study presents innovative contributions at both theoretical and practical levels. The theoretical contribution is related to fact that the lack of physical activity can have a harmful effect on individual's health and is currently an important public health concern \[[@B76-ijerph-16-01821]\]. In this respect, United Nations Educational, Scientific and Cultural Organization (UNESCO) \[[@B1-ijerph-16-01821]\] emphasised the importance of fostering and promoting active behaviours \[[@B6-ijerph-16-01821]\] in all contexts, especially at schools. Consequently, it should be noted that there is a positive connection between health and physical activity: sedentarism is a major risk factor for mortality, which gives rise to concern about the prevalence of sedentarism and socio-educative patterns of inactivity, especially in school contexts. At a practical level, our findings may help teaching staff in their tasks at school, as they provide a tool that may be used in teaching practice. In addition, the findings open up interesting fields of research in terms of the application of sport education, especially in terms of its impact on psychological variables. 5. Conclusions {#sec5-ijerph-16-01821} ============== In conclusion, our findings suggest that the pilot programme stimulated some improvement in adolescents' subjective well-being and trait emotional intelligence, but did not impact social anxiety. Therefore, on the basis of quality physical education and the social and emotional learning approach, the implementation of such programmes is recommended given the possible psychological benefits for adolescents in the educational context. A commitment to sports and other physical-sport activity options within a quality physical education framework, efficiently applied by means of relevant pedagogical models (such as sport education), may play an important role in students' integral development \[[@B77-ijerph-16-01821],[@B78-ijerph-16-01821]\]. We thank Edanz Group ([www.edanzediting.com/ac](www.edanzediting.com/ac)) for editing a draft of this manuscript. P.L., J.G. and J.C. conceived and designed the experiments; P.L., J.G. and J.C. performed the experiments; P.L., J.G. and J.C. analysed the data; P.L., J.G. and J.C. contributed reagents/materials/analysis tools; and P.L., J.G. and J.C. wrote the paper. This research received no external funding. The authors declare no conflicts of interest. {#ijerph-16-01821-f001} ijerph-16-01821-t001_Table 1 ###### Sex and age of the sample. *n* *%* --------- ------ ----- ----- **Sex** Male 64 57 Female 49 43 **Age** 12 42 37 13 45 40 14 23 20 15 3 3 ijerph-16-01821-t002_Table 2 ###### Sequencing of stages and activity sessions in the intervention programme. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Stage Session Sport Education Model ------------------------ ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Initial\ 1--2 Introduction and presentation of the Sport Education Model with digital and audio-visual support (ICT). Presentation and distribution of learning resources. Division and organisation of classroom groups in teams (assignment of team names with a didactic and cross curricular theme). Distribution and selection of responsibility roles. (Theoretical Sessions) 3 Explanation for the self-design of learning resources on digital format (ICT). Selection and assignment of anthems, badges, mascots and t-shirts representing a team. Intermediate\ 4--7 Practical implementation of the roles of each member of the teams. Learning of technical-tactical elements and abilities: kicking-off, catching, moving, throwing, defence and attack. Learning game rules. (Practical Sessions) 8--9 Warming-up, training and friendly matches. Meetings for comprehension and reflection with intervention of the responsibility roles. 10--14 Regular stage competition (Round Robin). Final\ 15--16 Inter-class groups final competitions (final matches with class groups), final event, giving awards and diplomas. (Practical Sessions) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ijerph-16-01821-t003_Table 3 ###### Reliability evidence of the instruments used (*n* = 113). Measures *α* *CR* *AVE* *Ω* ------------ ------ ------ ------- ------ KIDSCREEN 0.91 0.89 0.674 0.92 PANASN-PA 0.70 0.77 0.502 0.72 PANASN-NA 0.74 0.76 0.519 0.77 TEIQue-ASF 0.71 0.70 0.503 0.79 SAS-T 0.85 0.80 0.687 0.87 Notes: α = Cronbach's alpha; *CR* = composite reliability, *AVE* = average variance extracted; *Ω* = McDonald's omega index. ijerph-16-01821-t004_Table 4 ###### Mean, standard deviation, analysis of variance, analysis of covariance and effect size for differences in means (partial eta-squared) as a function of the experimental and control groups at pre-test and post-test. Measures PRE-TEST POST-TEST ---------------- -------------- -------------- ------- ------- -------- -------------- -------------- -------- ------- ------- **KIDSCREEN** HRQL 35.18 (5.84) 35.09 (6.01) 1.414 0.697 0.001 36.94 (6.14) 35.04 (5.99) 1.975 0.018 0.072 **PANASN** PA 21.43 (3.90) 20.02 (4.21) 3.293 0.07 0.008 21.95 (2.78) 20.33 (2.88) 5.438 0.017 0.144 NA 11.23 (3.54) 11.58 (3.58) 0.251 0.62 0.004 9.96 (2.95) 11.62 (3.86) 7.044 0.010 0.123 **TEIQUE-ASF** TEI 4.82 (0.60) 4.58 (0.64) 4.368 0.04 0.009 5.02 (0.63) 4.52 (0.53) 16.394 0.000 0.241 **SAS** SAS-T 2.61 (0.67) 2.64 (0.68) 0.040 0.84 −0.001 2.38 (0.63) 2.58 (0.62) 3.419 0.062 0.014 Note: HRQL = health-related quality of life; PA = positive affect; NA = negative affect; TEI = trait emotional intelligence; SAS-T = Total Social Anxiety Scale; SD = standard deviation.

Related literature {#sec1} ================== For medicinal background, see: Al-Madhoun *et al.* (2002[@bb1]); Arner & Eriksson (1995[@bb2]); Bello (1974[@bb3]); Celen *et al.* (2007[@bb6]); Eriksson *et al.* (2002[@bb7]); Wei *et al.* (2005[@bb9]); Welin *et al.* (2004[@bb10]). Experimental {#sec2} ============ {#sec2.1} ### Crystal data {#sec2.1.1} C~14~H~11~N~2~O~2~ ^+^·C~7~H~7~O~3~S^−^*M* *~r~* = 410.43Monoclinic,*a* = 7.6944 (5) Å*b* = 33.626 (2) Å*c* = 7.9426 (5) Åβ = 116.416 (1)°*V* = 1840.5 (2) Å^3^*Z* = 4Mo *K*α radiationμ = 0.21 mm^−1^*T* = 90 (2) K0.30 × 0.25 × 0.20 mm ### Data collection {#sec2.1.2} Bruker APEX CCD diffractometerAbsorption correction: multi-scan (*SADABS*; Sheldrick, 2008[@bb8]) *T* ~min~ = 0.939, *T* ~max~ = 0.95819080 measured reflections4482 independent reflections3753 reflections with *I* \> 2σ(*I*)*R* ~int~ = 0.049 ### Refinement {#sec2.1.3} *R*\[*F* ^2^ \> 2σ(*F* ^2^)\] = 0.054*wR*(*F* ^2^) = 0.124*S* = 1.124482 reflections263 parametersH-atom parameters constrainedΔρ~max~ = 0.61 e Å^−3^Δρ~min~ = −0.38 e Å^−3^ {#d5e472} Data collection: *SMART* (Bruker, 2002[@bb5]); cell refinement: *SAINT* (Bruker, 2002[@bb5]); 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: *DIAMOND* (Brandenburg & Putz (1999[@bb4]); software used to prepare material for publication: *SHELXTL* (Sheldrick, 2008[@bb8]). Supplementary Material ====================== Crystal structure: contains datablocks I, global. DOI: [10.1107/S1600536808038816/hb2840sup1.cif](http://dx.doi.org/10.1107/S1600536808038816/hb2840sup1.cif) Structure factors: contains datablocks I. DOI: [10.1107/S1600536808038816/hb2840Isup2.hkl](http://dx.doi.org/10.1107/S1600536808038816/hb2840Isup2.hkl) Additional supplementary materials: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?hb2840&file=hb2840sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?hb2840sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?hb2840&checkcif=yes) Supplementary data and figures for this paper are available from the IUCr electronic archives (Reference: [HB2840](http://scripts.iucr.org/cgi-bin/sendsup?hb2840)). The authors gratefully acknowledge the support of the National Science Foundation (CHE-0604527) and Molecular Insight Pharmaceuticals Inc. Comment ======= Radiolabeled nucleosides and nucleoside analogs may be good candidates for imaging and therapeutic applications because of their metabolic entrapment in rapidly dividing cells like tumor cells. These radiolabeled nucleoside derivatives may act as substrates for the human cytosolic thymidine kinase (hTK-1), an enzyme of the salvage pathway which catalyzes the phosphorylation of nucleosides to their corresponding 5\'-monophosphates (Welin *et al.*, 2004). The phosphorylation would mainly occur in proliferating tumor cells since hTK-1 shows a dramatically increased activity in tumor cells compared to quiescent cells (Bello, 1974). The phosphorylated nucleosides would be entrapped inside the proliferating cells because of their negatively charged phosphate moiety retarding the cellular efflux (Arner *et al.*, 1995). Thus, a radiolabeled nucleoside analog could be used as probe for tumor cell proliferation since the entrapment results in an accumulation in tissue with elevated hTK-1 activity. The main problem for the development of a suitable nucleoside analog lies in the narrow substrate specifity of hTK-1 (Eriksson *et al.*, 2002). The natural substrates for hTK-1 are thymidine and uridine. Major modifications of the corresponding nucleoside, however, may lead to a highly decreased activity. The literature on the interaction of thymidine derivatives with hTK-1 is not totally unambiguous about the effects of various substitutions. For example, N-3 derivatized thymidine analogs have been reported to be inactive (Celen *et al.*, 2007). On the other hand, N-3 modified carboranylalkyl thymidine analogs show acceptable conversion rates (Al-Madhoun *et al.*, 2002). Therefore, we made a set of several thymidine and uridine analogs modified at different positions of the ribose and the base moiety to get further insight on the effects of various derivatizations. To expand our SAAC concept (single amino acid chelate) for radioimaging and radiotherapeutic purposes on nucleosides, the title phthalimidomethylpyridinium *p*-toluenesulfonate salt, (I), was prepared as part of a series of tosylalkylphthalimide derivatives recently synthesized in our group. This series is used for the attachment of a SAAC chelate at the N-3 and C-5 position of the base moiety of thymidine (Bartholomä *et al.* unpublished results). The SAAC chelate allows hereby the radiolabelling of thymidine and uridine derivatives by the coordination of the \[*M*(CO)~3~\]^+^ core (*M* = ^186/188^Re, ^99^*^m^*Tc) (Wei *et al.*, 2005). The ideal decay properties, low cost and convenient availability of ^99^*^m^*Tc from generator columns make the corresponding nucleoside complexes interesting candidates for imaging purposes while their corresponding rhenium complexes could be used as therapeutic counterparts. Due to the tetrahedral arrangement of the connecting methyl group, the phthalimidomethylpyridinium cation in (I) is not planar. The tosylate anion sits on top of one end of the pyridinium residue showing aromatic π-π interaction. The centroid distance between those two aromatic rings is 3.5783 (2) Å. The other end of the pyridinium moiety shows some interaction with the phthalimide part of neighbored phthalimidomethyl-pyridinium cation. Thus, a T-stacking between the pyridinium residue and the benzyl ring of the phthalimide residue occurs (Table 1). The distances between the interacting C---H of the phthalimide and the centroid of the pyridinium residue are C2--Centroid = 3.5313 (2) Å and H2--Centroid = 2.70Å, respectively. The corresponding angle C2--H2···Centroid is 147°. The phthalimide moiety itself has a planar geometry. All bond length and angles fall in expected ranges. In the crystal, the ionic units are aligned in a zigzag arrangement in direction of the *b* axis. Experimental {#experimental} ============ 2.00 g (11.29 mmol) *N*-(Hydroxymethyl)phthalimide were dissolved in 20 ml anhydrous pyridine under an inert atmosphere followed by a dropwise addition of 3.23 g (16.93 mmol, 1.5 equiv.) *p*-Toluenesulfonyl chloride in 20 ml anhydrous pyridine. After complete addition of the tosylchloride, the reaction mixture was stirred for additional 16 h. About 2 h after the addition was completed, a white precipitate started to form. This white solid was filtered off, washed three times with 100 ml chloroform, and finally dried for several days at h.v.. The product was obtained in good yields as a colourless amorphous powder (3.80 g, quantitative); colourless blocks of (I) suitable for X-ray diffraction were collected directly from the reaction mixture. ^1^H NMR (d~6~-DMSO): *δ* = 2.28 (s, 3 H), 6.41 (s, 2 H), 7.10 (d, J = 7.98 Hz, 2 H), 7.47 (d, J = 7.95 Hz, 2 H), 7.90--7.99 (m, 4 H), 8.20 (t, J = 7.05 Hz, 2 H), 8.67 (t, J = 7.68 Hz, 1 H), 9.09 (d, J = 5.80 Hz, 2 H). p.p.m.. IR: *ν* = 3398 (br), 3124, 3087, 3037, 2980, 2935, 1781, 1729, 1627, 1485, 1404, 1361, 1331, 1300, 1268, 1206, 1168, 1116, 1089, 1067, 1031, 1008, 951, 826, 801, 777, 728, 680, 632, 587, 565, 529 cm^-1^. Refinement {#refinement} ========== The H atoms were placed in calculated positions and refined as riding. Figures ======= {#Fap1} {#Fap2} {#Fap3} Crystal data {#tablewrapcrystaldatalong} ============ ---------------------------------------- --------------------------------------- C~14~H~11~N~2~O~2~^+^·C~7~H~7~O~3~S^−^ *F*(000) = 856 *M~r~* = 410.43 *D*~x~ = 1.481 Mg m^−3^ Monoclinic, *P*2~1~/*c* Mo *K*α radiation, λ = 0.71073 Å Hall symbol: -P 2ybc Cell parameters from 3297 reflections *a* = 7.6944 (5) Å θ = 2.4--26.4° *b* = 33.626 (2) Å µ = 0.21 mm^−1^ *c* = 7.9426 (5) Å *T* = 90 K β = 116.416 (1)° Block, colourless *V* = 1840.5 (2) Å^3^ 0.30 × 0.25 × 0.20 mm *Z* = 4 ---------------------------------------- --------------------------------------- Data collection {#tablewrapdatacollectionlong} =============== --------------------------------------------------------------- -------------------------------------- Bruker APEX CCD diffractometer 4482 independent reflections Radiation source: fine-focus sealed tube 3753 reflections with *I* \> 2σ(*I*) graphite *R*~int~ = 0.049 Detector resolution: 512 pixels mm^-1^ θ~max~ = 28.1°, θ~min~ = 2.4° ω scans *h* = −10→10 Absorption correction: multi-scan (*SADABS*; Sheldrick, 2008) *k* = −42→44 *T*~min~ = 0.939, *T*~max~ = 0.958 *l* = −10→10 19080 measured reflections --------------------------------------------------------------- -------------------------------------- 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*\[*F*^2^ \> 2σ(*F*^2^)\] = 0.054 Hydrogen site location: inferred from neighbouring sites *wR*(*F*^2^) = 0.124 H-atom parameters constrained *S* = 1.12 *w* = 1/\[σ^2^(*F*~o~^2^) + (0.0508*P*)^2^ + 1.1529*P*\] where *P* = (*F*~o~^2^ + 2*F*~c~^2^)/3 4482 reflections (Δ/σ)~max~ \< 0.001 263 parameters Δρ~max~ = 0.61 e Å^−3^ 0 restraints Δρ~min~ = −0.38 e Å^−3^ ------------------------------------- ------------------------------------------------------------------------------------------------- Special details {#specialdetails} =============== ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Geometry. All e.s.d.\'s (except the e.s.d. in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell e.s.d.\'s are taken into account individually in the estimation of e.s.d.\'s in distances, angles and torsion angles; correlations between e.s.d.\'s in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell e.s.d.\'s is used for estimating e.s.d.\'s 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~ S1 0.71561 (7) 0.151380 (15) 0.83629 (7) 0.01495 (13) O1 1.0063 (2) 0.07297 (5) 0.3896 (2) 0.0232 (3) O2 0.4402 (2) 0.01341 (5) 0.3092 (2) 0.0247 (4) O3 0.8408 (2) 0.15516 (5) 0.7439 (2) 0.0209 (3) O4 0.6566 (2) 0.11048 (4) 0.8427 (2) 0.0232 (3) O5 0.7929 (2) 0.17090 (5) 1.0188 (2) 0.0217 (3) N1 0.7117 (2) 0.05127 (5) 0.3668 (2) 0.0173 (4) N2 0.5370 (2) 0.11140 (5) 0.3597 (2) 0.0158 (4) C1 0.8823 (3) 0.04795 (6) 0.3432 (3) 0.0169 (4) C2 0.8696 (3) 0.00828 (6) 0.2554 (3) 0.0159 (4) C3 0.9997 (3) −0.01049 (6) 0.2060 (3) 0.0191 (4) H3 1.1166 0.0020 0.2219 0.023\* C4 0.9513 (3) −0.04884 (7) 0.1314 (3) 0.0228 (5) H4 1.0367 −0.0628 0.0952 0.027\* C5 0.7805 (3) −0.06681 (7) 0.1096 (3) 0.0239 (5) H5 0.7523 −0.0930 0.0598 0.029\* C6 0.6488 (3) −0.04768 (7) 0.1584 (3) 0.0217 (4) H6 0.5315 −0.0601 0.1423 0.026\* C7 0.6972 (3) −0.00982 (6) 0.2314 (3) 0.0170 (4) C8 0.5931 (3) 0.01743 (6) 0.3028 (3) 0.0179 (4) C9 0.6823 (3) 0.08173 (6) 0.4785 (3) 0.0204 (4) H9A 0.8071 0.0953 0.5547 0.025\* H9B 0.6379 0.0694 0.5657 0.025\* C10 0.3571 (3) 0.11045 (6) 0.3473 (3) 0.0177 (4) H10 0.3245 0.0910 0.4150 0.021\* C11 0.2198 (3) 0.13761 (6) 0.2367 (3) 0.0182 (4) H11 0.0917 0.1367 0.2257 0.022\* C12 0.2708 (3) 0.16635 (6) 0.1417 (3) 0.0193 (4) H12 0.1781 0.1855 0.0657 0.023\* C13 0.4584 (3) 0.16703 (6) 0.1581 (3) 0.0192 (4) H13 0.4952 0.1867 0.0942 0.023\* C14 0.5902 (3) 0.13900 (6) 0.2678 (3) 0.0185 (4) H14 0.7184 0.1390 0.2790 0.022\* C15 0.4978 (3) 0.17744 (6) 0.6922 (3) 0.0147 (4) C16 0.3230 (3) 0.16559 (6) 0.6887 (3) 0.0170 (4) H16 0.3196 0.1430 0.7591 0.020\* C17 0.1535 (3) 0.18648 (6) 0.5831 (3) 0.0180 (4) H17 0.0355 0.1783 0.5832 0.022\* C18 0.1543 (3) 0.21947 (6) 0.4764 (3) 0.0169 (4) C19 −0.0316 (3) 0.24117 (7) 0.3577 (3) 0.0221 (5) H19A −0.0026 0.2652 0.3049 0.033\* H19B −0.0957 0.2486 0.4357 0.033\* H19C −0.1176 0.2238 0.2553 0.033\* C20 0.3304 (3) 0.23071 (6) 0.4797 (3) 0.0184 (4) H20 0.3336 0.2529 0.4069 0.022\* C21 0.5012 (3) 0.21025 (6) 0.5872 (3) 0.0167 (4) H21 0.6198 0.2186 0.5889 0.020\* ------ ------------- --------------- ------------- -------------------- -- Atomic displacement parameters (Å^2^) {#tablewrapadps} ===================================== ----- ------------- ------------- ------------- --------------- -------------- --------------- *U*^11^ *U*^22^ *U*^33^ *U*^12^ *U*^13^ *U*^23^ S1 0.0132 (2) 0.0163 (2) 0.0136 (2) −0.00043 (18) 0.00446 (18) −0.00117 (18) O1 0.0220 (8) 0.0201 (8) 0.0253 (8) −0.0060 (6) 0.0086 (7) −0.0012 (6) O2 0.0170 (7) 0.0361 (9) 0.0229 (8) −0.0006 (7) 0.0105 (6) 0.0016 (7) O3 0.0184 (7) 0.0250 (8) 0.0212 (8) 0.0008 (6) 0.0106 (6) −0.0010 (6) O4 0.0203 (8) 0.0182 (8) 0.0288 (8) −0.0001 (6) 0.0087 (7) 0.0030 (6) O5 0.0183 (7) 0.0290 (8) 0.0130 (7) 0.0020 (6) 0.0027 (6) −0.0037 (6) N1 0.0168 (8) 0.0177 (9) 0.0160 (8) 0.0022 (7) 0.0060 (7) 0.0003 (7) N2 0.0154 (8) 0.0182 (8) 0.0110 (8) 0.0043 (7) 0.0035 (7) −0.0007 (6) C1 0.0165 (10) 0.0187 (10) 0.0135 (9) 0.0029 (8) 0.0049 (8) 0.0032 (7) C2 0.0181 (10) 0.0150 (9) 0.0120 (9) 0.0015 (8) 0.0044 (8) 0.0032 (7) C3 0.0207 (10) 0.0221 (11) 0.0144 (10) 0.0022 (8) 0.0077 (8) 0.0034 (8) C4 0.0315 (12) 0.0237 (11) 0.0137 (10) 0.0088 (9) 0.0105 (9) 0.0037 (8) C5 0.0371 (13) 0.0169 (10) 0.0133 (10) −0.0018 (9) 0.0074 (9) 0.0002 (8) C6 0.0262 (11) 0.0208 (11) 0.0149 (10) −0.0056 (9) 0.0063 (9) 0.0012 (8) C7 0.0158 (9) 0.0192 (10) 0.0132 (9) 0.0000 (8) 0.0041 (8) 0.0036 (7) C8 0.0167 (10) 0.0220 (10) 0.0119 (9) 0.0004 (8) 0.0036 (8) 0.0047 (8) C9 0.0212 (10) 0.0241 (11) 0.0129 (9) 0.0076 (8) 0.0047 (8) 0.0020 (8) C10 0.0193 (10) 0.0194 (10) 0.0144 (9) −0.0002 (8) 0.0076 (8) −0.0026 (8) C11 0.0148 (9) 0.0223 (10) 0.0149 (10) 0.0009 (8) 0.0044 (8) −0.0050 (8) C12 0.0198 (10) 0.0186 (10) 0.0142 (10) 0.0044 (8) 0.0026 (8) −0.0026 (8) C13 0.0247 (11) 0.0171 (10) 0.0152 (10) 0.0006 (8) 0.0083 (8) 0.0006 (8) C14 0.0177 (10) 0.0207 (10) 0.0163 (10) −0.0010 (8) 0.0068 (8) −0.0031 (8) C15 0.0139 (9) 0.0166 (9) 0.0102 (9) 0.0000 (7) 0.0022 (7) −0.0026 (7) C16 0.0199 (10) 0.0185 (10) 0.0123 (9) −0.0001 (8) 0.0068 (8) 0.0001 (7) C17 0.0163 (9) 0.0213 (10) 0.0158 (10) 0.0002 (8) 0.0067 (8) −0.0027 (8) C18 0.0205 (10) 0.0159 (9) 0.0107 (9) 0.0013 (8) 0.0038 (8) −0.0042 (7) C19 0.0207 (11) 0.0228 (11) 0.0184 (10) 0.0049 (8) 0.0047 (9) −0.0019 (8) C20 0.0233 (10) 0.0155 (10) 0.0139 (9) −0.0002 (8) 0.0061 (8) −0.0010 (7) C21 0.0162 (9) 0.0188 (10) 0.0134 (9) −0.0037 (8) 0.0050 (8) −0.0035 (7) ----- ------------- ------------- ------------- --------------- -------------- --------------- Geometric parameters (Å, °) {#tablewrapgeomlong} =========================== ------------------- -------------- ----------------------- -------------- S1---O3 1.4531 (15) C9---H9B 0.9900 S1---O5 1.4556 (15) C10---C11 1.377 (3) S1---O4 1.4562 (16) C10---H10 0.9500 S1---C15 1.783 (2) C11---C12 1.386 (3) O1---C1 1.201 (3) C11---H11 0.9500 O2---C8 1.208 (2) C12---C13 1.391 (3) N1---C8 1.405 (3) C12---H12 0.9500 N1---C1 1.410 (3) C13---C14 1.375 (3) N1---C9 1.437 (3) C13---H13 0.9500 N2---C10 1.343 (3) C14---H14 0.9500 N2---C14 1.352 (3) C15---C21 1.390 (3) N2---C9 1.482 (3) C15---C16 1.391 (3) C1---C2 1.488 (3) C16---C17 1.387 (3) C2---C3 1.380 (3) C16---H16 0.9500 C2---C7 1.393 (3) C17---C18 1.397 (3) C3---C4 1.399 (3) C17---H17 0.9500 C3---H3 0.9500 C18---C20 1.396 (3) C4---C5 1.386 (3) C18---C19 1.506 (3) C4---H4 0.9500 C19---H19A 0.9800 C5---C6 1.393 (3) C19---H19B 0.9800 C5---H5 0.9500 C19---H19C 0.9800 C6---C7 1.380 (3) C20---C21 1.390 (3) C6---H6 0.9500 C20---H20 0.9500 C7---C8 1.486 (3) C21---H21 0.9500 C9---H9A 0.9900 O3---S1---O5 113.25 (9) H9A---C9---H9B 108.0 O3---S1---O4 112.78 (9) N2---C10---C11 120.30 (19) O5---S1---O4 112.75 (10) N2---C10---H10 119.8 O3---S1---C15 106.13 (9) C11---C10---H10 119.8 O5---S1---C15 105.58 (9) C10---C11---C12 119.12 (19) O4---S1---C15 105.53 (9) C10---C11---H11 120.4 C8---N1---C1 112.38 (17) C12---C11---H11 120.4 C8---N1---C9 123.06 (18) C11---C12---C13 119.60 (19) C1---N1---C9 123.35 (18) C11---C12---H12 120.2 C10---N2---C14 121.75 (18) C13---C12---H12 120.2 C10---N2---C9 119.46 (18) C14---C13---C12 119.4 (2) C14---N2---C9 118.78 (17) C14---C13---H13 120.3 O1---C1---N1 124.31 (19) C12---C13---H13 120.3 O1---C1---C2 130.5 (2) N2---C14---C13 119.82 (19) N1---C1---C2 105.22 (17) N2---C14---H14 120.1 C3---C2---C7 121.9 (2) C13---C14---H14 120.1 C3---C2---C1 129.71 (19) C21---C15---C16 119.42 (18) C7---C2---C1 108.40 (18) C21---C15---S1 120.78 (15) C2---C3---C4 116.9 (2) C16---C15---S1 119.78 (15) C2---C3---H3 121.6 C17---C16---C15 120.50 (19) C4---C3---H3 121.6 C17---C16---H16 119.8 C5---C4---C3 120.9 (2) C15---C16---H16 119.8 C5---C4---H4 119.5 C16---C17---C18 120.81 (19) C3---C4---H4 119.5 C16---C17---H17 119.6 C4---C5---C6 122.1 (2) C18---C17---H17 119.6 C4---C5---H5 119.0 C20---C18---C17 118.02 (19) C6---C5---H5 119.0 C20---C18---C19 121.63 (19) C7---C6---C5 116.7 (2) C17---C18---C19 120.33 (19) C7---C6---H6 121.7 C18---C19---H19A 109.5 C5---C6---H6 121.7 C18---C19---H19B 109.5 C6---C7---C2 121.6 (2) H19A---C19---H19B 109.5 C6---C7---C8 129.7 (2) C18---C19---H19C 109.5 C2---C7---C8 108.70 (18) H19A---C19---H19C 109.5 O2---C8---N1 124.3 (2) H19B---C19---H19C 109.5 O2---C8---C7 130.4 (2) C21---C20---C18 121.47 (19) N1---C8---C7 105.31 (17) C21---C20---H20 119.3 N1---C9---N2 111.61 (16) C18---C20---H20 119.3 N1---C9---H9A 109.3 C20---C21---C15 119.78 (19) N2---C9---H9A 109.3 C20---C21---H21 120.1 N1---C9---H9B 109.3 C15---C21---H21 120.1 N2---C9---H9B 109.3 C8---N1---C1---O1 −179.12 (19) C1---N1---C9---N2 108.7 (2) C9---N1---C1---O1 −11.4 (3) C10---N2---C9---N1 104.0 (2) C8---N1---C1---C2 0.2 (2) C14---N2---C9---N1 −76.5 (2) C9---N1---C1---C2 167.94 (17) C14---N2---C10---C11 1.0 (3) O1---C1---C2---C3 0.9 (4) C9---N2---C10---C11 −179.58 (18) N1---C1---C2---C3 −178.3 (2) N2---C10---C11---C12 −1.3 (3) O1---C1---C2---C7 179.2 (2) C10---C11---C12---C13 0.6 (3) N1---C1---C2---C7 −0.1 (2) C11---C12---C13---C14 0.4 (3) C7---C2---C3---C4 −0.4 (3) C10---N2---C14---C13 0.0 (3) C1---C2---C3---C4 177.63 (19) C9---N2---C14---C13 −179.44 (18) C2---C3---C4---C5 −0.2 (3) C12---C13---C14---N2 −0.7 (3) C3---C4---C5---C6 0.7 (3) O3---S1---C15---C21 −30.90 (19) C4---C5---C6---C7 −0.5 (3) O5---S1---C15---C21 89.57 (17) C5---C6---C7---C2 −0.2 (3) O4---S1---C15---C21 −150.81 (16) C5---C6---C7---C8 −177.4 (2) O3---S1---C15---C16 150.97 (16) C3---C2---C7---C6 0.6 (3) O5---S1---C15---C16 −88.55 (17) C1---C2---C7---C6 −177.78 (19) O4---S1---C15---C16 31.06 (19) C3---C2---C7---C8 178.36 (18) C21---C15---C16---C17 −0.7 (3) C1---C2---C7---C8 −0.1 (2) S1---C15---C16---C17 177.49 (15) C1---N1---C8---O2 179.69 (19) C15---C16---C17---C18 0.8 (3) C9---N1---C8---O2 11.9 (3) C16---C17---C18---C20 −0.1 (3) C1---N1---C8---C7 −0.2 (2) C16---C17---C18---C19 178.17 (19) C9---N1---C8---C7 −168.02 (17) C17---C18---C20---C21 −0.8 (3) C6---C7---C8---O2 −2.3 (4) C19---C18---C20---C21 −179.02 (19) C2---C7---C8---O2 −179.7 (2) C18---C20---C21---C15 1.0 (3) C6---C7---C8---N1 177.6 (2) C16---C15---C21---C20 −0.2 (3) C2---C7---C8---N1 0.2 (2) S1---C15---C21---C20 −178.34 (15) C8---N1---C9---N2 −84.8 (2) ------------------- -------------- ----------------------- -------------- Hydrogen-bond geometry (Å, °) {#tablewraphbondslong} ============================= ------------------ --------- --------- ----------- --------------- *D*---H···*A* *D*---H H···*A* *D*···*A* *D*---H···*A* C5---H5···Cg1^i^ 0.95 2.70 3.531 (2) 147 ------------------ --------- --------- ----------- --------------- Symmetry codes: (i) −*x*+1, −*y*, −*z*. ###### Hydrogen-bond geometry (Å, °) *D*---H⋯*A* *D*---H H⋯*A* *D*⋯*A* *D*---H⋯*A* ------------------ --------- ------- ----------- ------------- C5---H5⋯*Cg*1^i^ 0.95 2.70 3.531 (2) 147 Symmetry code: (i) . *Cg*1 is the centroid of the N2/C10--C14 ring.

{.114} {.115}

Background ========== Metal speciation is a primary control on metal mobility and bioavailability in the environment, and adsorption reactions can play a significant role in this process (e.g., \[[@B1]\]). Therefore, many researchers have worked to develop predictive models to describe metal adsorption for a wide range of systems. In studies of natural sediments and soils, empirical approaches based on partition coefficients (K~d~) or semi-empirical Langmuir or Freundlich isotherms are often used to describe metal partitioning between solutions and solid substrates. However, because partition coefficients depend on solution and substrate composition, they cannot be extrapolated beyond the conditions for which they are measured (e.g., \[[@B2],[@B3]\]). Furthermore, because partition coefficients do not include any consideration of mass balance, they can result in very misleading predictions regarding metal speciation and mobility \[[@B2]\]. The application of Langmuir or Freundlich isotherms is similarly hindered because these also depend on solution and substrate composition and do not account for the development of electrical charge at mineral surfaces, nor do they consider the structure of adsorbed species \[[@B3]\]. In contrast, thermodynamically-based surface complexation models (SCMs) include explicit descriptions of reaction stoichiometries and the development of electrical charge at the solid surface \[[@B4]\]. These models have a significant advantage over empirical or semi-empirical models because once calibrated, they should allow accurate prediction of metal speciation under varying solution compositions (e.g. in ionic strength, background electrolyte, competing ions, etc.), and thus should be useful in predicting metal speciation in a wide variety of systems. In spite of the significant potential advantages of surface complexation models (SCMs), widespread application of these models, especially to complex sediments and soils, has been lacking for a variety of reasons. Determining the mineralogy of the finest, and therefore highest surface area and presumably most reactive, constituents of soils and sediments is often difficult. Even when the bulk mineralogy is well characterized, deriving reactive surface areas to include in SCMs can be hampered by a lack of information regarding flow paths and the presence of coatings at solid surfaces (e.g., \[[@B3],[@B5]\]). Furthermore, there is a lack of data for adsorption of many metals on substrates that are relevant for natural systems. Lastly, there is little information regarding the applicability of surface models parameterized using pure, single solid systems to more complex systems containing mixtures of solids (e.g. \[[@B6],[@B7]\]). Overcoming these obstacles is crucial if existing surface complexation models are to become widely used and useful for understanding metal speciation in natural systems. Developing models which are better able to accurately predict the speciation of copper in the environment is important because, although ecosystems require trace quantities of copper to maintain physiological functions \[[@B8]\], at higher concentrations copper is toxic to both plants and animals \[[@B9]\]. Furthermore, copper tends to bind strongly to organic and mineral substrates, potentially resulting in mobilization of competing metal ions. Concentrations of trace metals, including copper, have increased dramatically in many ecosystems worldwide due to anthropogenic activities, including dredging of river sediments \[[@B10]-[@B13]\], application of pesticides and fungicides \[[@B14],[@B15]\], and through mining and smelting operations. This has resulted in toxic levels of trace metals in many soils and sediments \[[@B8],[@B9],[@B16]\], and a pressing need to develop accurate predictive models of Cu speciation in the environment. In order to better understand and quantify copper bioavailability and transport, copper adsorption has been extensively studied. However, Cu adsorption has been described using SCMs for a relatively small suite of single, pure minerals (e.g. Cu/goethite: \[[@B17]-[@B20]\]; Cu/hematite: \[[@B21]\]; Cu/gibbsite: \[[@B22]\]; Cu/kaolinite: \[[@B23]-[@B28]\]; Cu/hydrous manganese oxide: \[[@B29]\]; Cu/hydrous ferric oxide: \[[@B4]\]) and in even fewer studies in the presence of mixed mineral assemblages (e.g., \[[@B6]\]) or natural soils or sediments (e.g., \[[@B30],[@B31]\]). Furthermore, many of the existing SCMs cannot be used to model adsorption of copper on mixtures of minerals, because the surface complexation parameters have been derived using different treatments of the electrical double layer (e.g. constant capacitance, double layer, triple layer models). The goal of this study is to develop internally-consistent descriptions of copper adsorption on hydrous ferric oxide and kaolinite at a variety of ionic strength and sorbate/sorbent ratios using a diffuse double layer model (DLM). Models for the individual solid systems are assessed using 95% confidence intervals of a goodness-of-fit parameter, V(Y). The performance of DLMs parameterized using single solid systems are assessed in mixture solid systems by quantitative comparison of measurements and predictions based on V(Y). The double layer model is chosen because of the extensive database of stability constants that has been derived for metal adsorption on hydrous ferric oxide \[[@B4]\] and hydrous manganese oxide \[[@B29]\]. Furthermore, the DLM approach requires fewer fit parameters than other SCMs (e.g. triple layer models) and yields stability constants that, at least in theory, do not vary with ionic strength (unlike those obtained using a constant capacitance model). These features make the DLM approach an attractive option for modeling adsorption on natural sediments and soils. Experimental methods ==================== Materials --------- All reagents used were ACS reagent or trace metal grade. DDI water was purified using a Barnstead E-pure (Model D4641) water system. Powdered kaolinite from Edgar, Florida was purchased from Ward\'s Scientific. X-ray diffraction indicates that the kaolinite is moderately well ordered with quartz and mica or smectite impurities, with perhaps 1--2% mica in the \< 1 μm size fraction (data courtesy of Ray Ferrell, Louisiana State University). Powdered low defect kaolinite from Washington County, GA, USA (KGa-1b) was obtained from the Clay Minerals Society Source Clays Repository. The most significant impurities in the KGa-1b kaolinite are \~1.64% TiO~2~\[[@B32],[@B33]\], 0.21% Fe~2~O~3~\[[@B32]\] and 231 ppm total organic carbon \[[@B32]\]. Kaolinite powder was dried at 90°C for 4 days, and then stored in a refrigerator until usage. No other precleaning was done. Hydrous ferric oxide (HFO) was synthesized based on procedures proposed by Schwertmann and Cornell \[[@B34]\]. Briefly, \~40 g of ferric nitrate was dissolved in \~500 mL of DDI water in a glass beaker. Concentrated trace metal grade NaOH was slowly titrated into the beaker under constant stirring as precipitate formed, until the mixture reached a pH of \~7.0. The mixture was kept at pH 7 for \~72 hours, then poured into a plastic tube, centrifuged, the supernatant decanted, and the remaining precipitate washed with DDI. The centrifuging and washing procedure was repeated \~5--6 times. The final washed precipitate was freeze-dried and the freeze-dried solid ground gently using a mortar and pestle to break up large aggregates. Specific surface areas for HFO and both types of kaolinite were determined at atmospheric pressure using a Quantachrome Nova Surface and Pore Analyzer Model 2200e. Replicate \~2 g samples of each solid were degassed for \~24 hrs and analyzed using 11-point N~2~BET. A degassing temperature of \~80°C was used for the HFO and 25°C was used for both types of kaolinite. Measured specific surface areas were: 220 (HFO), 13.6 (KGa-1b kaolinite), and 25.7 (Ward\'s kaolinite) m^2^/g. Dzombak and Morel \[[@B4]\] argue that due to the presence of significant microporosity N~2~BET underestimates the surface area available for sorbates on HFO, and therefore suggest that a specific surface area of 600 m^2^/g be adopted for modeling. This recommendation has been widely used in DLM descriptions of metal adsorption on HFO. To be consistent with these previous modeling efforts, this value is adopted here for all surface complexation models derived for Cu adsorption on HFO. Experimental approach --------------------- Adsorption experiments were completed using continuously stirred batch reactors (500 mL), at room temperature and open to the atmosphere, containing dissolved Cu(II) and NaNO~3~as the background electrolyte. Batch reactors were typically titrated first to an acidic initial pH (\~2--4.5) using trace metal grade HNO~3~. A 10 mL aliquot of this initial suspension of Cu and NaNO~3~was removed for subsequent analysis of the initial Cu concentration. Next, the HFO, kaolinite or mixture of these solids was added to the well-stirred 500 mL batch reactor. This suspension was typically preequilibrated for 24 hours. The preequilibration procedure may result in some dissolution, especially of kaolinite at low pH. However, Landry et al. \[[@B7]\] demonstrate that preequilibration at acidic compared to circumneutral pH does not significantly influence Co adsorption on kaolinite under conditions similar to those used here. Thus, although some dissolution of the solids may have occurred during preequilibration, this should not significantly influence metal adsorption at the conditions used in this study. After 24 hours, the pH of the preequilibrated suspension was titrated upwards by additions of small volumes of 0.1 M NaOH sufficient to increase the pH by increments of 0.2 to 0.5. Several experiments were also completed in which the base titration was followed by an acid titration using 0.1 M HNO~3~. The acid and base legs of these experiments exhibited no significant hysteresis, i.e. any hysteresis was less than the experimental uncertainty (see also below). After titrant addition and stabilization of the pH to within 0.05 pH log units per minute, which typically occurred in about 10 minutes, a 10 mL aliquot of the mixed suspension was pipetted into an acid-washed 15 mL plastic centrifuge vial. The 15 mL tubes, including the initial mineral-free control sample, were subsequently agitated with a benchtop shaker for 24 hrs, removed from the shaker and the pH measured again. The 24 hr period should be more than sufficient for the adsorption reaction to reach equilibrium (see below). In most experiments, each aliquot was then centrifuged and the supernatant filtered through a 0.2 μm syringe filter. However, several experiments were also completed to compare the effect of syringe-filtering to centrifugation only. No significant difference was observed between samples prepared by filtering and those prepared by centrifugation only. All supernatants were acidified using concentrated trace metal grade HNO~3~, amended to 1000 ppb with an internal indium standard and analyzed for Cu using either a ThermoElectron PQ Excell ICP-MS or a Perkin Elmer Optima 2100DV ICP-OES with matrix-matched calibration standards. The amount of Cu adsorbed was calculated by the difference between Cu concentration in the analyzed supernatants and the initial Cu solution. To determine the adsorption kinetics and the reversibility of Cu adsorption, adsorption and desorption of Cu on kaolinite was tested as a function of time. Using a batch slurry of 2 g/L KGa kaolinite, 10^-5^M Cu and 0.01 M NaNO~3~, adsorption was initiated by titrating the slurry to a pH of 10.5. 98 ± 2% of the initial Cu was adsorbed by the kaolinite within 5 minutes (data not shown). Periodic sampling over the following 72 hours demonstrated that this Cu remained sorbed on the kaolinite surface. To test the reversibility of sorption, after the 72 hour period the slurry was titrated to pH 2.3. Within 10 minutes only 9 ± 4% of the Cu remained sorbed to the kaolinite and after 24 hours 100% of the Cu was recovered from the kaolinite. Under the conditions of the initial sorption experiment (pH 10.5, 10^-5^M Cu, 0.01 M NaNO~3~), tenorite is supersaturated and might precipitate. Although it is not possible to distinguish adsorption from surface precipitation in these macroscopic experiments, the rapid desorption of the Cu suggests that adsorption, rather than precipitation, occurs. Modeling approach ----------------- Surface complexation stability constants for individual adsorption edge experiments were optimized using FITEQL4.0 \[[@B35]\]. Each optimization was completed for a specific reaction stoichiometry (see Table [1](#T1){ref-type="table"} and discussion below) with activity corrections based on the Davies equation (see \[[@B35]\]) and including a stability constant of -7.29 for formation of CuOH^+^~(aq)~from the JCHESS default thermodynamic database, which is based on the EQ3/6 database \[[@B36]\]. Due to their small influence (\< 2.5%) on calculated Cu speciation at the measured conditions, CO~2(g)~and other aqueous Cu species were not included in the FITEQL input files. The optimization procedure was used to obtain best-fit stability constants for each edge obtained in a single experiment. In some cases, replicate experiments were completed. The resulting edges were fit individually, and were not aggregated in the modeling. Sets of edges were obtained on kaolinite and HFO to span a range of ionic strength and sorbate/sorbent ratios (see below). The median stability constant(s) derived for sorption onto each solid was input into the speciation code JCHESS, together with all reaction constituents, including CO~2(g)~. The resulting edges were calculated in JCHESS with activities based on the Debye-Huckel equation and using the default JCHESS thermodynamic database, which includes stability constants for formation of HNO~3(aq)~, HCO~3~^-^~(aq)~, CO~2(aq)~, CuOH^+^~(aq)~, CuO~2~^-2^~(aq)~, NaHCO~3~^-^~(aq)~, CuCO~3(aq)~, NaOH~(aq)~, NaCO~3~^-^~(aq)~, CuCO~3~(OH)~2~^-2^~(aq)~and Cu(CO~3~)~2~^-2^~(aq)~. The calculated Cu adsorbed differed by \< 2.5% from those calculated using the more simplified FITEQL model, even for the highest ionic strength experiments. JCHESS was also used to assess saturation states of minerals and to complete speciation calculations for mixed solid assemblages, which is not possible using the FITEQL software. The default JCHESS thermodynamic database contains data for copper-bearing minerals including tenorite, malachite, azurite, and cuprite. No data is included for Cu(OH)~2(s)~. Calculations using the stability constant for formation of Cu(OH)~2(s)~provided with the JCHESS MINTEQ database (log K = -8.64) indicate that for a given set of experimental conditions, tenorite saturates at lower pH then Cu(OH)~2(s)~. ###### Surface areas, surface site types and site densities used in DLM calculations. **Solid** **Surface Area (m^2^/g)** **Site Types** **Site Density (μmol/m^2^)** -------------------- --------------------------- ---------------- ------------------------------ HFO (2-site model) 600^\[4\]^ \>Fe~(s)~OH 0.094^\[4\]^ \>Fe~(w)~OH 3.74^\[4\]^ HFO (1-site model) 600^\[4\]^ \>FeOH 16.6^\[38\]^ Kaolinite (KGa) 13.6 (this study) \>SOH 16.6^\[38\]^ X 2.2^a^ Kaolinite (Wards) 25.7 (this study) \>SOH 16.6^\[38\]^ X 1.2^a^ a\. From CEC (3.0 meq/100 g) measured by Bordon and Giese \[[@B41]\] for kaolinite KGA-1b Adsorption edge data calculated with JCHESS were compared to the experimental data to assess goodness of fit V(Y) using the model proposed by Heinrich et al. \[[@B37]\]: $$V(Y) = \frac{\sum\limits_{P,Q}\left( \frac{Y}{s_{Y}} \right)^{2}}{n_{p} \times n_{Q} - n_{R}}$$ where Y is the difference between the calculated and measured concentration of metal remaining in solution for each data point, P; s~Y~is the standard deviation; n~P~is the number of data points; n~Q~is the number of components, Q, for which the concentration, C, and the total concentration, t, are known (n~Q~= 1 for all edges in this study); and n~R~is the number or parameters being optimized. The standard deviation, s~Y~, was assumed to be equal to 5% of the experimentally measured copper concentration in solution with a lower limit of 10 ppb (based on ICP OES errors and detection limits). Confidence intervals for V(Y) values were calculated according to the equation proposed by Heinrich et al. \[[@B37]\]: $$\frac{\left( {n_{Q} \times n_{P} - n_{R}} \right) \ast V(Y)}{\chi_{1 - \alpha/2}^{2}},\frac{\left( {n_{Q} \times n_{P} - n_{R}} \right) \ast V(Y)}{\chi_{\alpha/2}^{2}},$$ where *χ*^2^~*p*~is the quantile of the chi-square distribution with (n~Q~xn~P~-n~R~) degrees of freedom with exceedence probability, *p*, and *α*= .05 (95% confidence interval). For a given set of adsorption edges, the simplest model yielding a statistically superior V(Y), i.e. the model with the lowest V(Y) having no overlap with the 95% confidence interval of the next best model, was accepted as the best model. The V(Y), together with the 95% confidence intervals, was also used to compare the fit of models calibrated for the single solids with those obtained for the mixed mineral systems. Results and discussion ====================== Cu adsorption on hydrous ferric oxide (HFO) ------------------------------------------- Cu adsorption on HFO was measured as a function of pH, ionic strength and sorbate/sorbent ratio (Fig. [1](#F1){ref-type="fig"}). For a given sorbate/sorbent ratio, there is little dependence of adsorption on ionic strength. Increasing the sorbate/sorbent ratio by an order of magnitude, from 10^-5^M Cu to 10^-4^M Cu on 2 g/L HFO, increases the pH of 50% adsorption (pH~50~) from \~4.4 to \~4.7. Replicate experiments completed with 0.1 M NaNO~3~and 10^-5^M Cu are in reasonable agreement (Fig. [1](#F1){ref-type="fig"}, blue symbols). JCHESS calculations indicate that the only solid that may become supersaturated in this system is tenorite (CuO). In the absence of adsorption, saturation with respect to tenorite occurs at pH\~6 and 6.5, for the 10^-4^M and 10^-5^M Cu experiments, respectively, which is well above the measured pH edges. ![**Cu adsorption as a function of pH on HFO**. Solid concentration is 2 g/L. Lines indicate fits for (A) Dzombak and Morel \[[@B4]\] 2-site HFO model and (B) Sverjensky and Sahai \[[@B38]\] 1-site HFO model calculated using parameters shown in Tables 1 and 2. Replicate experiments (0.1 M NaNO~3~and 10^-5^M Cu) are distinguished by separate symbols.](1467-4866-9-9-1){#F1} Cu adsorption on HFO has been described by Dzombak and Morel \[[@B4]\] using a 2-site double layer surface complexation model (DLM) with the parameters shown in Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. Amphoteric strong and weak surface hydroxyl sites are included in the model, but Cu adsorption is assumed to occur as a monodentate complex only on the strong site, according to: ###### Reaction stoichiometries and stability constants used in DLM calculations for HFO. **Reaction** **Log Stability Constant** **V(Y) (V(Y)~min~, V(Y)~max~)** --------------------------------------------------------- ---------------------------- --------------------------------- *HFO (2-site model):* \>Fe~(s)~OH + H^+^~(aq)~= \>Fe~(s)~OH~2~^+^ 7.29^\[4\]^ \>Fe~(w)~OH+ H^+^~(aq)~= \>Fe~(w)~OH~2~^+^ 7.29^\[4\]^ \>Fe~(s)~OH = \>Fe~(s)~O^-^+ H^+^~(aq)~ -8.93^\[4\]^ \>Fe~(w)~OH = \>Fe~(w)~O^-^+ H^+^~(aq)~ -8.93^\[4\]^ 14.0 \>Fe~(s)~OH + Cu^+2^~(aq)~= \>Fe~(s)~OCu^+^+ H^+^~(aq)~ 2.89^\[4\]^ (10.9, 18.5) *HFO (1-site model):* \>FeOH + H^+^~(aq)~= \>FeOH~2~^+^ 7.5^\[38\]^ \>FeOH = \>FeO^-^+ H^+^~(aq)~ -10.2^\[38\]^ 12.2 \>FeOH + Cu^+2^~(aq)~= \>FeOCu^+^+ H^+^~(aq)~ 0.98 (this study) (9.5, 16.1) Average goodness-of-fit parameters (V(Y)) and 95% confidence intervals of V(Y) are for the fit of each model to all of the Cu on HFO adsorption edge data (n~p~= 61). The resulting model curves are in very good agreement with the experimental data (Fig. [1A](#F1){ref-type="fig"}; V(Y) = 14.0). The Dzombak and Morel model adequately captures both the lack of ionic strength dependence and the shift in the edges with increased sorbate/sorbent ratio. In experiments with 10^-4^M Cu, the relatively small number of strong sites (1.13·10^-4^M) are nearly, but not entirely, saturated with sorbed Cu at high pH. An internally consistent set of single-site DLM parameters for the protonation and deprotonation of a wide variety of solids, including HFO and kaolinite, has been predicted by Sverjensky and Sahai \[[@B38]\] based on Born solvation theory. A goal of the current study is to develop robust DLM descriptions of cation adsorption on environmentally-relevant solids, while minimizing the number of fit parameters. Adsorption of Cu on kaolinite is described in this study (see discussion below) using a DLM based on the work of Sverjensky and Sahai \[[@B38]\]. A second goal of this study is to develop internally-consistent DLMs for Cu adsorption on both HFO and kaolinite. Therefore, the Cu adsorption edges for HFO were used to derive a best-fit stability constant using the site density, protonation and deprotonation values recommended by Sverjensky and Sahai \[[@B38]\] and assuming monodentate adsorption of Cu on the single site, according to: The V(Y) value for the resulting model (V(Y) = 12.2), while slightly lower than that determined for the 2-site Dzombak and Morel \[[@B4]\] model, is not statistically superior (Table [2](#T2){ref-type="table"}). Furthermore, the fits obtained with the single-site model, although statistically inseparable at the 95% confidence interval from those of the 2-site model, fail to capture the dependence of the adsorption edges on sorbate/sorbent ratio (Fig. [1B](#F1){ref-type="fig"}). In general, we recommend choosing the simplest model, with the least number of fitting parameters, when multiple models produce statistically inseparable results. However, the Dzombak and Morel \[[@B4]\] model has been carefully calibrated with a very large dataset (including many metals besides Cu), and this model is already in widespread use. Therefore, we apply both the Dzombak and Morel 2-site model and the simpler 1-site model developed in this study to predict Cu adsorption for systems containing both HFO and kaolinite (see discussion below). Cu adsorption on kaolinite -------------------------- Cu adsorption was measured on two types of kaolinite (KGa and Wards) as a function of ionic strength (0.001 to 0.1 M NaNO~3~) and sorbate/sorbent ratio (10^-4^to 10^-6^M Cu on 2 or 5 g/L kaolinite). The pH~50~decreases with smaller sorbate/sorbent ratios and typically increases with increasing ionic strength (Fig. [2A, B](#F2){ref-type="fig"}). This is in agreement with the results of prior studies of Cu adsorption on kaolinite (e.g., \[[@B23]-[@B28]\]). Replicate experiments (10^-5^M Cu, 0.1 M NaNO~3~in Fig. [2A](#F2){ref-type="fig"}; 10^-4^M Cu, 0.01 M NaNO~3~in Fig. [2A](#F2){ref-type="fig"}; 10^-5^M Cu, 0.02 M NaNO~3~in Fig. [2B](#F2){ref-type="fig"}) are generally in agreement, although one of the 10^-5^M Cu, 0.02 M NaNO~3~(Fig. [2B](#F2){ref-type="fig"}) does show higher adsorption then expected. Tenorite is predicted to be supersaturated in the absence of adsorption at pH ≥ 6 for 10^-4^M Cu and ≥ 6.5 for 10^-5^M Cu experiments. This is well above the 10^-5^M Cu sorption edges, but could influence a portion of the 10^-4^M Cu edges. However, as discussed above, rapid desorption of 10^-5^M Cu from kaolinite after equilibration for 72 hours at pH 10.5 suggests that adsorption is the primary uptake mechanism. {#F2} A number of surface complexation models have been proposed to describe Cu adsorption on kaolinite. Most of these follow the lead of Schindler et al. \[[@B23]\] and use a 2-site approach. Schindler et al. \[[@B23]\] derived a 2-site constant capacitance model (CCM) with the formation of an innersphere Cu surface complex on a variable charge site together with exchange of Cu^+2^for H^+^or Na^+^on a permanent charge, or ion exchange site. Schindler et al. \[[@B23]\] interpret the variable charge site as an aluminol site and suggest that the exchange site may either be due to isomorphous substitution giving rise to a small permanent structural charge on the kaolinite or may be due to the presence of a small amount of 2:1 interlayer clay impurity in the kaolinite specimen. Subsequent work by Ikhsan et al. \[[@B25]\] used a similar modeling approach except that a bidentate complex was used to describe Cu adsorption on the variable charge site. Likewise, Heidmann et al. \[[@B26]\] chose a 1-pK Stern model to describe Cu adsorption on variable charge edge sites of kaolinite with ion exchange sites used to describe adsorption at lower pH. An alternative approach for modeling Cu adsorption on kaolinite was proposed by Jung et al. \[[@B24]\], who used a triple layer surface complexation model assuming that Cu binds as an innersphere monodentate complex on an amphoteric aluminol site and as an outersphere monodentate complex on a deprotonatable silanol site. Similarly, Hizal and Apak \[[@B28]\] assumed formation of monodentate Cu complexes on two variable charge sites, presumed to correspond to aluminol and silanol sites. Finally, Peacock and Sherman \[[@B27]\] used an extended constant capacitance model to describe Cu adsorption on kaolinite. Using EXAFS data as a guide, Peacock and Sherman \[[@B27]\] proposed a model with Cu sorption occurring at three sites, forming a bidentate, mononuclear innersphere complex with an aluminol edge site; a tridentate, binuclear complex with an aluminol edge site; and binding to an ion exchange site on the basal plane of the kaolinite. The goal of this study is to derive a simple, DLM description of Cu binding on kaolinite consistent with the DLM description of Cu binding on HFO that can be used to make predictions of Cu speciation in mixed solid systems. A variety of models were tested, including monodentate or bidentate binding of Cu to an amphoteric variable charge site in the presence or absence of an ion exchange site (see Table [3](#T3){ref-type="table"}). ###### Reaction stoichiometries and stability constants used in DLM calculations for kaolinite. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- **Reaction** **Log Stability Constant** **V(Y) (V~min~(Y), V~max~(Y))** ---------------------------------------------------------------------------------------------------------------------------------------- ---------------------------- --------------------------------- \>KaoliniteOH + H^+^~(aq)~= \>KaoliniteOH~2~^+^ 2.1^\[38\]^ \>KaoliniteOH = \>KaoliniteO^-^+ H^+^~(aq)~ -8.1^\[38\]^ *Monodentate variable charge site:* 23.3 \>KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteOCu^+^+ H^+^~(aq)~ -1.7 (this study) (20.2, 27.1) *Bidentate variable charge site:* 28.5 2 \> KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteO~2~Cu + 2H^+^~(aq)~ -4.6 (this study) (24.8, 33.2) *Monodentate variable charge + ion exchange site (all log K\'s fit in this study):* \>KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteOCu^+^+ H^+^~(aq)~ -1.9 (this study) X(Na) + H^+^~(aq)~= X(H) + Na^+^~(aq)~ 4.1 (this study) 44.9 2X(Na) + Cu^+2^~(aq)~= X~2~(Cu) + 2Na^+^~(aq)~ 0.72 (this study) (38.9, 52.2) *Monodentate variable charge + ion exchange site model (fixed Na-H exchange stability constant):* \>KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteOCu^+^+ H^+^~(aq)~ -2.3 (this study) X(Na) + H^+^~(aq)~= X(H) + Na^+^~(aq)~ 4.3 (this study) 108 2X(Na) + Cu^+2^~(aq)~= X~2~(Cu) + 2Na^+^~(aq)~ 2.5^\[7\]^ (93.6, 126) *Bidentate variable charge + ion exchange site model (fixed Na-H exchange stability constant):* -5.3 (this study) 2\>KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteO~2~Cu + 2H^+^~(aq)~ 4.6 (this study) 62.6 X(Na) + H^+^~(aq)~= X(H) + Na^+^~(aq)~ 2.5^\[7\]^ (54.4, 72.9) 2X(Na) + Cu^+2^~(aq)~= X~2~(Cu) + 2Na^+^~(aq)~ *Monodentate variable charge + ion exchange site model, no Cu sorption on ion exchange site (fixed Na-H exchange stability constant):* -1.8 (this study)\ 190\ 2.5^\[7\]^ (165, 221) \>KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteOCu^+^+ H^+^~(aq)~ X(Na) + H^+^~(aq)~= X(H) + Na^+^~(aq)~ *Bidentate variable charge + ion exchange site model, no Cu sorption on ion exchange site (fixed Na-H exchange stability constant):* -4.6 (this study)\ 33.8\ 2.5^\[7\]^ (29.4, 39.3) 2 \> KaoliniteOH + Cu^+2^~(aq)~= \>KaoliniteO~2~Cu + 2H^+^~(aq)~ X(Na) + H^+^~(aq)~= X(H) + Na^+^~(aq)~ ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- The same protonation/deprotonation constants for the variable charge site is used in all models. Average goodness-of-fit parameters (V(Y)) and 95% confidence intervals of V(Y) are for the fit of each model to all of the Cu on kaolinite adsorption edge data (n~p~= 360). Variable charge surface site densities can be estimated from crystallographic considerations, although this typically yields a range of values, depending on assumptions made regarding crystal morphology and the definition of a surface site (see discussion in \[[@B39]\]). For example, Koretsky et al. \[[@B39]\] estimated a range of 0 to 21.8 sites/nm^2^for kaolinite, based on crystallographic considerations. Due to this uncertainty, site densities are often treated as additional fit parameters in surface complexation models. Previous work demonstrates that surface complexation stability constants are dependent on the choice of site densities and that typically a wide range of site densities can provide a satisfactory fit to measured data (e.g., \[[@B40]\]). Therefore, in this study the variable charge surface site density is set equal to 10 sites/nm^2^, as recommended by Sverjensky and Sahai \[[@B38]\]. For models with an additional ion exchange site, the ion exchange site density was calculated directly from the measured cation exchange capacity for KGA-1b reported by Bordon and Giese \[[@B41]\] (Table [1](#T1){ref-type="table"}). Protonation and deprotonation stability constants are typically fit using measured potentiometric acid-base titrations for the mineral of interest. However, Sverjensky and Sahai \[[@B38]\] have developed a predictive scheme for estimating protonation and deprotonation stability constants for a wide variety of solids, based on Born solvation theory. There are several advantages to using their predicted constants. First, if their method produces satisfactory results, then development of the SCM is simplified, requiring fewer experimental measurements and fewer fit parameters. Secondly, their estimates are produced from analysis of many experimental datasets and therefore should be both robust and internally consistent. Finally, although perhaps difficult, a goal of this study is to develop a generally-applicable model of Cu adsorption on kaolinite, which is not specific to a particular specimen of kaolinite. Therefore, protonation and deprotonation stability constants for the variable charge site were taken from the predictions of Sverjensky and Sahai \[[@B38]\] (Table [3](#T3){ref-type="table"}) and were not treated as fit parameters, in spite of the fact that doing so might produce a better model fit to the data. Fits were assessed by calculating the goodness-of-fit for the median stability constant(s) when applied to all of the measured edge data (n~p~= 360). The lowest V(Y) is obtained for the simplest model tested: a single variable site model with formation of a monodentate Cu complex (Table [2](#T2){ref-type="table"}; Fig. [2](#F2){ref-type="fig"}) according to: This model produces a reasonable description of the pH edge dependence on ionic strength and sorbate/sorbent for both specimens of kaolinite although, especially for the Wards kaolinite, sorption is somewhat underestimated at low pH (\< 4.5). XAS and macroscopic isotherm data suggest the formation of a bidentate Cu complex on kaolinite \[[@B25],[@B27]\]. Modeling Cu adsorption on kaolinite using a simple single site model with a bidentate Cu complex, i.e., produces a slightly higher V(Y) compared to the simple monodentate model, although the fit is not statistically distinct at the 95% confidence interval (Table [3](#T3){ref-type="table"}). However, visual inspection of the fits produced by the bidentate model suggests that the predicted edges are systematically steeper than the experimental data and also that the description of sorbate/sorbent dependence is poorer than for the monodentate model (Fig. [3](#F3){ref-type="fig"}). {#F3} As described above, 2-site models have been used to describe Cu adsorption on kaolinite. Furthermore, two-site models are necessary to produce adequate descriptions of weaker ion sorption (e.g. Co and Cd) on kaolinite (e.g., \[[@B7],[@B25],[@B42]\]). The eventual goal of the approach in this study is to develop a relatively simple surface complexation model that can be applied in natural systems containing mixtures of multiple solids and cations. Thus, a 2-site approach, even if it does not produce a statistically superior description of the experimental Cu data, may be necessary to describe metal adsorption in natural systems that contain mixtures of weakly and strongly sorbing ions. Therefore, the simple single variable charge site model was expanded to include a permanent charge, or ion exchange, site, which can bind Na^+^or Cu^+2^according to respectively (e.g. \[[@B7],[@B42]\]). Although good fits, correctly describing the slightly elevated sorption at low pH, are obtained for individual edges by fitting for the three stability constants associated with reactions (5), (7) and (8), applying the median values to the full set of data results in a statistically poorer fit to the kaolinite data compared to the 1-site model (Table [3](#T3){ref-type="table"}). Furthermore, for many experiments, FITEQL did not converge when stability constants for reactions (6), (7) and (8) were fit simultaneously. Landry et al. \[[@B7]\] found that a 2-site approach is necessary to adequately describe Co adsorption on kaolinite. Using an amphoteric variable charge site and an ion exchange site as shown above, Landry et al. \[[@B7]\] derived a stability constant of 2.5 for reaction (7). Using this value for the Na-H exchange stability constant and fitting for the stability constants associated with Cu binding on the ion exchange site and on the variable charge site (as either a bidentate or monodentate complex) still yields a statistically poorer fit to the data compared to the 1-site model (Table [3](#T3){ref-type="table"}), producing overestimates of sorption at low ionic strength and sorbate/sorbent ratio and underestimates of sorption at high ionic strength and sorbate/sorbent ratio, particularly on KGa kaolinite (Fig. [4](#F4){ref-type="fig"}). The best model fit with the ion exchange included is obtained when Cu sorption to the ion exchange site (reaction 8) is excluded from the model (i.e. only reaction 7 is allowed) and Cu forms a bidentate site on the variable charge site (V(Y) = 33.8; Table [3](#T3){ref-type="table"}; Fig. [5](#F5){ref-type="fig"}). The goodness-of-fit for this model is not statistically different from the 1-site bidentate model, although it is not as good as the 1-site monodentate model at the 95% confidence interval. Nonetheless, this model has the advantage that it is consistent with the 2-site models required to describe adsorption of weakly binding ions such as Cd^+2^or Co^+2^on kaolinite. It is also consistent with formation of a bidentate Cu species as inferred from isotherm data \[[@B25]\] and XAS studies \[[@B27]\]. {#F4} {#F5} Predictions from the two best models, i.e., the 1-site model (monodentate Cu adsorption on a variable charge site) and the 2-site model (ion exchange site that does not sorb Cu; variable charge site with bidentate adsorption of Cu) were compared to adsorption edge data obtained in previous studies \[[@B24],[@B25],[@B27],[@B28]\]. For calculations with the 2-site model, the electrolyte was assumed to be NaNO~3~, although two studies used KNO~3~\[[@B24],[@B25]\] and one used NaClO~4~\[[@B28]\]. The 1-site monodentate model produced a V(Y) of 634 for all of the compiled data (n~p~= 73), whereas the 2-site bidentate model resulted in a significantly better fit to the complete dataset (V(Y) = 281; Fig. [6](#F6){ref-type="fig"}). The adsorption edge reported by Ikhsan et al. \[[@B25]\] is in reasonable agreement with predictions from the 1-site monodentate model derived independently in this study (Fig. [6A](#F6){ref-type="fig"}). The 2-site bidentate model yields a pH~50~within \~0.3 of the measured pH~50~, but the predicted edge is steeper than the data of Ikhsan et al. \[[@B25]\]. Both models predict saturation of the surface at less than 100% for the conditions reported by Peacock and Sherman (\[[@B27]\]; Fig. [6B](#F6){ref-type="fig"}). The 2-site bidentate model produces a reasonable fit to both datasets reported by Jung et al. \[[@B24]\], while the 1-site monodentate model underestimates the amount of Cu adsorbed (Fig. [6C](#F6){ref-type="fig"}). The 1-site mondentate model is in slightly better agreement with the data of Hizal and Apak \[[@B28]\] at high pH compared to the 2-site bidentate model, which produces a steeper edge, but both models underestimate adsorption compared to the data at lower pH (Fig. [6D](#F6){ref-type="fig"}). ![**Comparisons 1-site and 2-site kaolinite models with previously reported data**. The 1-site model includes a monodentate Cu complex on a variable charge site. The 2-site model includes a bidentate cu complex on a variable charge site and an ion exchange site that does not bid Cu (parameters shown in Table 2). Data is from: (A) Ikhsan et al. \[[@B25]\] for 0.005 M KNO~3~, 10^-4^M Cu, 6.8 g/L kaolinite with 14.73 m^2^/g surface area, (B) Peacock and Sherman \[[@B27]\] for 0.1 M NaNO~3~, 3.93·10^-4^M Cu, 3.33 g/L kaolinite with 12.2 m^2^/g, (C) Jung et al. \[[@B24]\] for 0.01 or 0.1 M KNO~3~, 10^-4^M Cu, 2.0 g/L kaolinite with 7.99 m^2^/g and (D) Hizal and Apak \[[@B28]\] with 0.1 M NaClO~4~, 1.57·10^-4^M Cu, 50 g/L kaolinite with 26.68 m^2^/g (Clay II) or 17.8 m^2^/g (Clay III).](1467-4866-9-9-6){#F6} The general agreement between the predicted edges from the model derived here and data from four independent studies using different ionic strength, background electrolyte, sorbate/sorbent ratio and kaolinite specimens is encouraging. Discrepancies between model predictions and these experiments could be due to differences in experimental conditions, for example in the choice of background electrolyte (e.g., KNO~3~or NaClO~4~rather than NaNO~3~). Another possibility is that the disagreement reflects differences in the purity, solid solution chemistry, defect structure or other characteristics of the kaolinite specimens. Differences in measured N~2~-BET surface areas of the kaolinite specimens may also play a role in the model misfits. For example, Peacock and Sherman \[[@B27]\] report an N~2~BET surface area of 12.2 m^2^/g for the Cornwall kaolinite used in their study. This, combined with the site density of 10 sites/nm^2^chosen here, results in saturation of the surface at \~80%, and underestimates the percentage of Cu adsorbed reported by Peacock and Sherman \[[@B27]\]. Increasing the surface area to 25.7 m^2^/g (as measured for the Wards sample in this study) results in much better agreement with the reported data (Fig. [6B](#F6){ref-type="fig"}). The significant dependence of derived stability constants on measured surface area and choice of site density is well known (e.g. \[[@B40],[@B43]\]), and may prove to be particularly problematic in the application of surface complexation models to natural sediments, where reactive surface area is difficult to assess \[[@B3]\]. Cu adsorption on mixture of kaolinite and hydrous ferric oxide -------------------------------------------------------------- Cu adsorption was measured on mixtures of HFO and kaolinite (both Wards and KGa) at a range of conditions, although most experiments were conducted in 0.01 M NaNO~3~and using 10^-5^M Cu. Total solid concentrations ranged from 4 g/L to 7.5 g/L and ratios of kaolinite to HFO ranged from 1:1 to 500:1. As might be expected, adsorption increases at a given pH, ionic strength and sorbate/sorbent ratio with increasing quantities of HFO (Fig. [7](#F7){ref-type="fig"}). Several experiments were conducted with base titration (closed symbols), followed by acid titration (open symbols), to test for hysteresis. Within the experimental uncertainty, no significant hysteresis was observed. ![**Predicted Cu adsorption as a function of pH on mixtures of kaolinite and HFO**. Lines indicate fits using the Sverjensky and Sahai \[[@B38]\] 1-site HFO model and 1-site kaolinite model with formation of a monodentate Cu complex on a variable charge site. Model fits calculated using parameters shown in Tables 1, 2 and 3. Open and closed symbols indicate experiments with base titration (closed symbols) followed by acid titration (open symbols).](1467-4866-9-9-7){#F7} Goodness-of-fit parameters for the mixed solid systems (n~p~= 255) were assessed for various combinations of the single solid models discussed above (Table [4](#T4){ref-type="table"}). The resulting V(Y) range from 12.6 to 58.2, with values typically intermediate between those obtained for the pure HFO and pure kaolinite systems. In the pure systems, the 1-site HFO model and the 1-site monodentate variable charge kaolinite model produced the lowest V(Y) values (12.2 and 23.3, respectively). Combining these two models yields V(Y) = 14.1 (Fig. [7](#F7){ref-type="fig"}). Although this is not the lowest V(Y) obtained by combining individual solid models, it is not statistically distinguishable from the lowest value of 12.6 (Fig. [8](#F8){ref-type="fig"}), obtained by combining the 1-site HFO model with the 2-site monodentate kaolinite model, without sorption of Cu on the ion exchange site, at the 95% confidence interval. ###### Average goodness-of-fit parameters (V(Y)) and 95% confidence intervals of V(Y) for the fit of each model to all of the Cu on kaolinite + HFO adsorption edge data (n~p~= 255). Exchange site models use reaction (7) stability constant of 2.5 from Landry et al. \[[@B7]\] **Model HFO/Kaolinite** **V(Y) (V~min~(Y), V~max~(Y))** ------------------------- --------------------------------- *DM/MV* 18.0 (15.2, 21.6) *DM/BV* 13.6 (11.5, 16.3) *DM/MVE* 58.2 (49.3, 69.8) *DM/BVE* 27.0 (22.8, 32.3) *DM/MVE(noCu)* 21.6 (18.3, 25.9) ***DM/BVE(noCu)*** **13.6 (11.5, 16.3)** *SS/MV* 14.1 (12.0, 16.9) *SS/BV* 15.9 (13.5, 19.1) *SS/MVE* 19.2 (16.3, 23.1) *SS/BVE* 29.1 (24.6, 34.9) *SS/MVE(noCu)* 12.6 (10.7, 15.1) *SS/BVE(noCu)* 15.9 (13.5, 19.1) DM = 2-site HFO model from \[[@B4]\] SS = 1-site HFO model MV = 1-site monodentate variable charge kaolinite model BV = 1-site bidentate variable charge kaolinite model MVE = 2-site kaolinite model with monodentate variable charge and ion exchange site MVB = 2-site kaolinite model with bidentate variable charge and ion exchange site MVE(noCu) = 2-site kaolinite model with monodentate variable charge and ion exchange site (no Cu Sorption on exchange site) MVBnoCu) = 2-site kaolinite model with bidentate variable charge and ion exchange site (no Cu Sorption on exchange site) ![**Predicted Cu adsorption as a function of pH on mixtures of kaolinite and HFO**. Lines indicate fits using the Sverjensky and Sahai \[[@B38]\] 1-site HFO model and 2-site kaolinite model with formation of a monodentate Cu complex on a variable charge site and no Cu adsorption on the ion exchange site. Model fits calculated using parameters shown in Tables 1, 2 and 3. Open and closed symbols indicate experiments with base titration (closed symbols) followed by acid titration (open symbols).](1467-4866-9-9-8){#F8} Although combining the simplest models producing statistically indistinguishable V(Y) values for the individual systems is conceptually appealing, these models may not represent the best DLM descriptions of the systems. As discussed above, the 2-site HFO model of Dzombak and Morel \[[@B4]\] produces a visually better fit to the HFO data and has been calibrated for a greater breadth of data than the 1-site HFO model presented above. Combining the 1-site monodentate kaolinite model with the 2-site HFO model results in a V(Y) of 18.0 (Fig. [9](#F9){ref-type="fig"}). As can be seen in Figs. [7](#F7){ref-type="fig"} and [9](#F9){ref-type="fig"}, combining either HFO model with the 1-site monodentate kaolinite model produces a reasonable fit to the measured edges, although in both cases sorption is overestimated for experiments with the greatest kaolinite to HFO ratio. Use of the 2-site HFO model produces a better prediction of the dependence of sorption on sorbate/sorbent ratio (Figs. [7A](#F7){ref-type="fig"}, [9A](#F9){ref-type="fig"}; yellow data/lines). ![**Predicted Cu adsorption as a function of pH on mixtures of kaolinite and HFO**. Lines indicate fits using the Dzombak and Morel \[[@B4]\] 2-site HFO model and the 1-site kaolinite model with formation of a monodentate Cu complex on a variable charge site. Model fits calculated using parameters shown in Tables 1, 2 and 3. Open and closed symbols indicate experiments with base titration (closed symbols) followed by acid titration (open symbols).](1467-4866-9-9-9){#F9} As discussed above, the 2-site kaolinite model with inclusion of an ion exchange site that does not significantly bind Cu may be preferable to the simpler 1-site model because of its potential for representing adsorption of weaker ions to kaolinite in mixed metal solutions. Combining the 1-site or 2-site HFO models with the 2-site bidentate kaolinite model produces V(Y) values of 15.9 and 13.6, respectively (Figs. [10](#F10){ref-type="fig"}, [11](#F11){ref-type="fig"}). The resulting edges are again somewhat steeper than the experimental data, but do adequately capture the dependence of the edges on kaolinite to HFO ratios. ![**Predicted Cu adsorption as a function of pH on mixtures of kaolinite and HFO**. Lines indicate fits using the Sverjensky and Sahai \[[@B38]\] 1-site HFO model and the 2-site kaolinite model with formation of a bidentate Cu complex on a variable charge site and an ion exchange site that does not bind Cu. Model fits calculated using parameters shown in Tables 1, 2 and 3. Open and closed symbols indicate experiments with base titration (closed symbols) followed by acid titration (open symbols).](1467-4866-9-9-10){#F10} ![**Predicted Cu adsorption as a function of pH on mixtures of kaolinite and HFO**. Lines indicate fits using the Dzombak and Morel \[[@B4]\] 2-site HFO model and the 2-site kaolinite model with formation of a bidentate Cu complex on a variable charge site and an ion exchange site that does not bind Cu. Model fits calculated using parameters shown in Tables 1, 2 and 3. Open and closed symbols indicate experiments with base titration (closed symbols) followed by acid titration (open symbols).](1467-4866-9-9-11){#F11} In all cases, surface complexation models calibrated for the single mineral systems do an adequate job of describing metal adsorption in the mixed solid systems. Goodness-of-fit parameters are intermediate between those obtained for the pure HFO and pure kaolinite systems. Discrepancies between model predictions and measured data are similar to those observed in the pure mineral systems. For example, edges that are too steep are produced from kaolinite models assuming bidentate adsorption of Cu in both the pure and mixed mineral systems. Similarly, the 2-site HFO model produces better descriptions of the dependence of Cu sorption on sorbate/sorbent ratio in both pure and mixed mineral systems. The relatively good fits (low V(Y)) obtained for predicted Cu sorption in mixed mineral systems suggest that the stability constants obtained for the pure mineral systems are reasonably robust, and even more importantly, that HFO-kaolinite interactions need not be explicitly included in the speciation model. In other words, a simple component additivity approach produces predictions for the mixed mineral systems that are as good as those for the calibrated pure mineral systems. However, it is important to point out that these experiments were conducted over a relatively brief temporal scale (days). It is possible that aging over months or years may produce surface coatings that physically or chemically block ion adsorption sites. This must be tested using further experimental work, together with adsorption studies on natural sediments and soils. Conclusion ========== This study demonstrates that a simple diffuse layer surface complexation model produces reasonable descriptions of Cu adsorption on HFO and kaolinite over a range of pH, ionic strength, and sorbate/sorbent ratios. In particular, Cu adsorption on kaolinite can be adequately modeled using a very simple DLM with formation of a monodentate Cu complex on a single amphoteric variable-charge site. However, if this model approach is to be extended to natural systems, which contain many other cations, a 2-site approach may be required. Thus, a 2-site kaolinite model, including both a permanent charge and a variable charge site, is developed. Cu adsorption on mixtures of HFO and kaolinite is predicted reasonably well using a simple component additivity approach, with the DLM parameters derived for the pure mineral systems. Goodness-of-fit values derived for the predicted fits, together with visual inspection of the predictions, suggests that discrepancies between models and data for the mixed mineral systems are similar to those observed for the pure mineral systems. Thus, interactions between the kaolinite and the HFO, such as blocking of ion exchange or variable charge sites, can be neglected in the speciation calculations. However, it is important to note that all of the experiments in this study were conducted over short temporal scales, and that aging could produce coatings or other physical changes not apparent in the present study. Further experiments must be completed using longer timescales to assess the possible effects of aging on Cu speciation in mixed solid systems. Nonetheless, the results of this study are encouraging, suggesting that relatively simple models with few adjustable parameters may produce useful predictions of metal speciation in natural sediments and soils containing many solid components. However, it is also important to note that testing the component additivity approach for simple systems such as this is only a first step toward application of the component additivity approach to field settings. Even if mineral-mineral interactions can be ignored, methods must still be developed to assess reactive surface areas in field settings if the component additivity approach is to gain widespread use. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= All authors contributed to the design of the experiments and the modeling approach. TJL and MSS carried out all experiments. CMK conceived the study, participated in the design and coordination of all experiments and, together with TJL, drafted the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ This study was supported by a grant from the National Science Foundation CAREER program (NSF-EAR 0348435). Dr. Ray Ferrell at Louisiana State University kindly provided XRD analyses for the kaolinite purchased from Ward\'s Scientific. Johnson Haas provided helpful comments on an initial draft of the paper. TJL, SD and CJL appreciate additional support from the Western Michigan University graduate and undergraduate research and creative activities funds. Helpful comments from the handling editor and two anonymous reviewers are also very much appreciated.

All relevant data are within the paper and its Supporting Information files. Introduction {#sec004} ============ For patients with suspicious lung lesions found on chest x-ray or CT, endobronchial and/or transbronchial biopsy of the lung is the preferred method for obtaining a diagnosis, with transthoracic CT-guided biopsy and surgical biopsy as alternative approaches. The NELSON study found that 62.7% of lesions were located in the periphery of the lung compared to the pleural wall (10%), middle (11.6%), or central (15.3%) airways \[[@pone.0189963.ref001]\]. In the central airways, approximately 4 endobronchial (tools remain in the airways) biopsies are needed to obtain an adequate diagnostic yield which has been reported to be 70--90% \[[@pone.0189963.ref002]\]. For peripheral lesions, fluoroscopy-guided transbronchial (tools traverse the airway wall) biopsy is the standard technique. Reported rates for diagnostic yield in peripheral lesions are 30--78% \[[@pone.0189963.ref002],[@pone.0189963.ref003],[@pone.0189963.ref004],[@pone.0189963.ref005]\] and are especially low for small lesions \<2cm in diameter (30--34%) \[[@pone.0189963.ref004],[@pone.0189963.ref005]\]. Attempts have been made with endobronchial ultrasound (EBUS) probes, electromagnetic (EM) navigation, virtual navigation bronchoscopy, and bronchoscopic transparenchymal nodule access to improve diagnostic yield; however, none of these techniques have gained wide acceptance \[[@pone.0189963.ref006],[@pone.0189963.ref007]\]. Overall \~60% of patients will return for a second procedure (transthoracic or surgical biopsy) if diagnostic yield is insufficient \[[@pone.0189963.ref003],[@pone.0189963.ref006]\]. In the last decade several large population studies have been completed to assess the benefits of low-dose CT (LDCT) screening in the general population which has led to some countries implementing screening programs \[[@pone.0189963.ref008],[@pone.0189963.ref009],[@pone.0189963.ref010],[@pone.0189963.ref011],[@pone.0189963.ref012],[@pone.0189963.ref013],[@pone.0189963.ref014],[@pone.0189963.ref015],[@pone.0189963.ref016],[@pone.0189963.ref017]\]. Given the increasing number of small, peripheral lesions detected by these screening programs, better biopsy tools will be required to achieve good diagnostic yield and follow-up. There are many types of new lung tools that have been reported in the literature and some that are now entering the market. These new tools aim to improve diagnostic yield by providing better navigational guidance to the lesion (example, EM tracking \[[@pone.0189963.ref018],[@pone.0189963.ref019],[@pone.0189963.ref020]\]) or by evaluating the tissue prior to biopsy (examples, EBUS \[[@pone.0189963.ref021]\], radial endobronchial ultrasound \[REBUS\] \[[@pone.0189963.ref021]\], optical coherence tomography \[[@pone.0189963.ref022],[@pone.0189963.ref023],[@pone.0189963.ref024],[@pone.0189963.ref025],[@pone.0189963.ref026],[@pone.0189963.ref027]\], fluorescence spectroscopy \[[@pone.0189963.ref028],[@pone.0189963.ref029]\], diffuse reflectance spectroscopy \[[@pone.0189963.ref028],[@pone.0189963.ref029],[@pone.0189963.ref030]\], Raman spectroscopy \[[@pone.0189963.ref031],[@pone.0189963.ref032],[@pone.0189963.ref033]\], or differential path length spectroscopy \[[@pone.0189963.ref030],[@pone.0189963.ref034],[@pone.0189963.ref035],[@pone.0189963.ref036]\]). These new tools will improve diagnostic yield but only if they can physically reach the lesion. After evaluating the current literature we were unable to find sufficient information on the ideal tool size required to be effective in targeting the majority of lung lesions endobronchially. The goal of this manuscript was to therefore investigate the percentage of lesions that can be reached for various diameter tools if the tools remain inside the airways (i.e. endobronchial biopsy) and the distance a tool must travel "off-road" (or outside of the airways) to reach all lesions. Methods {#sec005} ======= 3D simulations of the human airways and randomly distributed lesions {#sec006} -------------------------------------------------------------------- A 3D model of the human airways, based on a deterministic algorithm that incorporates both duct branching and space division, has been developed by Hiroko Kitaoka et al. \[[@pone.0189963.ref037],[@pone.0189963.ref038]\]. This model was used to generate an airway tree (Model Type 1) with 3311 branches. In the model the region of interest was set to 0 (whole lung), lung capacity at beginning of inspiration to 0.35, lung capacity at end of inspiration to 1.0, inspiratory time to 0.4, and a supine body posture assumed. For this exercise, only two time points in the respiratory cycle were used--when the lung is fully inflated (here forth referred to as 'inspiration') and when fully exhaled ('expiration'). This model also generated a 3D volume of the entire lung. Lung lesions were randomly generated using MATLAB (MathWorks, Natick, MA). Results from the NELSON trial \[[@pone.0189963.ref013]\] showed the distribution of lesions in the x-z (transverse) and x-y (coronal) planes. These distributions were used to define a probability distribution function (PDF) in the x- and y-axes; for the z-axis a uniform PDF was used for simplicity. A PDF was also defined for the left and right lung. The PDF's were contained within the lung volume by manually selecting the minimum and maximum x, y, and z coordinates from the generated lung volume image; this was repeated 3 times to ensure the correct lung volume was used. The x, y, and z coordinates of 1000 points (i.e. lesions) were randomly drawn based on these PDF's and 3 lung volumes. Three sets of lesions was drawn for the inspiration models and another 3 for the expiration models (in total 6 models of airways and lesions were created). A single set couldn't be generated for both inspiration and expiration since the airways are in motion and the lung lesions are defined by fixed x, y, and z coordinates. The majority of lung lesions generated were found in the upper half of the lung or in the periphery, as expected based on the NELSON distributions. [Fig 1](#pone.0189963.g001){ref-type="fig"} shows the expiration airway model with the three sets of 1000 randomly generated lung lesions. {#pone.0189963.g001} The airway model is exported in VTK-format and consists of vertices and triangular faces. MATLAB was used to calculate the diameter of the airways at each face by iteratively stepping through each face, finding the opposite face, and calculating the distance between the two respective planes. When an opposite face could not be determined the diameter at that face was set to NaN (not a number). Next, for each of the 1000 lesions in the 6 models (3 different distributions x 2 respiratory states), a spherical mesh was created to represent the lesion with diameters ranging from 1mm-5cm in 1mm increments. The *intriangulation* function in MATLAB was used to determine if there were any vertices or faces of the airway mesh that were inside the lesion mesh. For the faces and vertices that were found to be within the lesion mesh, the corresponding airway diameter was recorded. This was repeated for all lesion diameters, lesions, and models. To calculate the percentage of reachable lesions we counted the number of lesions that intersected with an airway whose diameter was larger than the tool size and divided by the total number of lesions (i.e. 1000); this percentage was calculated for tool sizes ranging from 1-9mm in 0.5mm increments. In these models, lesions are not always connected to an airway. Therefore, it is important to understand how far a tool must reach beyond the bronchial wall (i.e. going "off-road" in the parenchyma as in the case of transbronchial biopsy) to reach the boundary of a lesion. [Fig 2](#pone.0189963.g002){ref-type="fig"} shows how this off-road distance was calculated. For example, given a lesion with radius r1, a small device (Device1) can be navigated inside the airways until it reaches an airway with diameter D1 which is approximately the same size as the tool itself. A larger device (Device2) can't be navigated as far down the airways and gets stuck when the airway diameter is D2. At this point the tool would have to exit the airway and travel through the parenchyma to reach the lesion. In the case of Device1 this distance is simply the difference between r2 and r1, where r2 is taken as the radius of the first spherical lesion mesh which intersects with an airway greater than or equal to diameter D1. For Device2 the off-road distance would then be the difference between r3 and r1. {#pone.0189963.g002} A few limitations should be noted regarding these simulations. First, the smallest airway diameter that is calculated from the Kitaoka airway tree is 0.1mm. If every branch termination was this small we could make conclusions about tool sizes down to 0.1mm. However, not every branch termination was smaller than 1mm, therefore, we will only draw conclusions based on tool diameters \>1mm. Second, this analysis does not take into account any information about how the airways stretch or deform when a tool is navigated through them. Because the airways have some elasticity a tool may be able to be navigated into airways smaller than the tool diameter. Lastly, expiration and inspiration states were treated as two separate models with different lesion distributions and cannot be compared exactly. CT image analysis {#sec007} ----------------- This manuscript involves a retrospective analysis of CT images from lung cancer patients at the Netherlands Cancer Institute--Antoni van Leeuwenhoek Hospital. All patients underwent CT imaging as part of their standard clinical diagnostic care; no additional imaging examinations were performed for the data in this manuscript. The hospital ethics committee was consulted; under Dutch law the study did not require IRB approval (or patient informed consent). Only the authors associated with the hospital (GCL, FL) had access to the imaging data and performed the image analysis; measurement data was fully anonymized and de-identified. Images were restricted to patients with lung lesions \>1cm and \<4cm without prior radiotherapy. Helical CT Single Breath Hold Scans were reconstructed with 1.0mm slice thickness and 1.0mm slice increment. For lesions that were in contact with an airway, the airway diameter was measured. The distance between the lesion and closest airway of 1, 2, 4 and 6mm diameter was measured. Measurements were performed in the VUE PACS (Carestream Health, Rochester, NY, USA) viewing software using Cross-sectional MPR visualization. Results {#sec008} ======= From the CT scans 21 lesions were included. The lesions were non-small cell lung carcinoma and sarcoma or melanoma metastases, ranging in diameter from 9.7mm to 3.6cm (1.6cm mean, 1.3cm median). [Fig 3](#pone.0189963.g003){ref-type="fig"} shows the lesions projected in a 3D model derived from a CT scan without lesions: *isosurface* function in MATLAB and exported to Blender (Blender Foundation, Amsterdam, The Netherlands) for visualization. The size of the lesion represents the average diameters in the x, y and z direction. The location is the location of the measured lesion to the center of a 3D box enclosing the lungs. Most lesions appear in the right lung or upper lobes, similar to the NELSON trial \[[@pone.0189963.ref013]\] and simulated distributions. {#pone.0189963.g003} In Figs [4](#pone.0189963.g004){ref-type="fig"} and [5](#pone.0189963.g005){ref-type="fig"} the simulated 3D data and the CT data are shown together for comparison. [Fig 4](#pone.0189963.g004){ref-type="fig"} shows the percentage of reachable lesions versus various sized tools for different lesion sizes. Each of the 3 simulated models is shown as a separate color but for the most part these 3 models have very similar results. As would be expected, the smaller the tool size, the more lesions that can be reached. And the larger the lesion, the more likely it can be reached even with a larger tool. Although the number of lesions acquired from the CT scans is limited, there is reasonable agreement for the lesion sizes. In the CT data the majority of the lesions which touched an airway were next to 1 or 2mm airways; only one was next to a 3mm airway and one next to a 5mm airway. {#pone.0189963.g004} {#pone.0189963.g005} [Fig 5](#pone.0189963.g005){ref-type="fig"} shows the simulated average (and standard deviation) distance a tool would have to travel off-road to reach various sized lesions. Only one simulated model is shown since the other two models are nearly identical. For all tool sizes, the larger the lesion size the shorter the off-road distance. For example, a 6.3mm bronchoscope would have to go nearly 2cm to reach the smallest lesions, while a 1mm tool would only need to travel off-road \~1cm. The CT data overlaps reasonably well with the 1mm and 2mm diameter tools, however, diverges for the 6mm diameter tool. Discussion {#sec009} ========== The 3D simulations were used to better understand how small an endobronchial tool must be to reach the majority of lung lesions and how far a tool must travel beyond the bronchial wall through the parenchyma (off-road) if a lesion is not connected to the airways. The simulations show that the percentage of small (\<1cm) lesions that can be reached, even with a 1mm tool, is quite low at only 17.3% (inspiration) and 23.8% (expiration). To provide further insight into this, in one of the inspiration models 989/1000 1mm diameter lesions did not touch an airway; this only slightly decreases to 918/1000 for 5mm lesions. It is possible that these small lesions are touching an airway but they would be \<1mm in diameter as this was the limit of our model. For the small lung lesions that will most likely be found with LDCT screening, this means that in endobronchial biopsy, the biopsy tool has to pass through the bronchiolar wall towards the lesion (transbronchial biopsy). For lesions 1cm in diameter, a 1mm tool would have to travel on average 8.5mm (inspiration) or 6.7mm (expiration) off-road, and farther for larger sized tools. To consider these results in another context, a standard 6.3mm diameter bronchoscope would reach \<10% of 1cm lesions, \<30% of 2cm lesions, and \<50% of 3cm lesions. Most pulmonologists would agree that a bronchoscope this size can only reach central lesions which according to the NELSON study consisted of 15.3% of the overall lesions in the lung \[[@pone.0189963.ref001]\]; thus within the range of the percentages found here. The CT data supports the results derived from the simulations; reasonable overlap is seen in both Figs [4](#pone.0189963.g004){ref-type="fig"} and [5](#pone.0189963.g005){ref-type="fig"}. Compared to the 1000 simulated lesions in each model, the CT data is limited; with additional CT data it is assumed that the distributions in [Fig 4](#pone.0189963.g004){ref-type="fig"} would match the simulated distributions more closely. Regardless of the data size, the CT lesions 0.7--1.3cm in diameter are close to the 1cm simulated curve, while the 1.7--2.3cm CT lesions are close to the 2cm simulated curve with the error between CT and simulation increasing with increasing tool size. On average, the off-road distance measured from the CT scans is close to the average calculated from the simulations for the 1mm and 2mm diameter tools, however, quite different for a 6mm tool. This is likely due to the fact that the airway diameters are slightly different between the simulated and CT data. From the CT scans the trachea measures \~1.8-2cm, the main bronchi \~1--1.4cm, and the lobar bronchi \~6-9mm. In the simulated data the trachea is \~1.4--1.5cm, the main bronchi \~1--1.2cm, and the lobar bronchi \~1--1.1cm; 6mm sizes do not occur until the segmental branches. The diameter difference in the lobar bronchi is likely the reason for the deviation between the CT and simulated data in both Figs [4](#pone.0189963.g004){ref-type="fig"} and [5](#pone.0189963.g005){ref-type="fig"}. Since the lobar bronchi are larger in the simulated data, a large tool (i.e. 6mm bronchoscope) can be navigated further into the airways and hence the off-road distance to distal lesions is shorter compared to the CT data where a large tool will get stuck sooner. To confirm this assumption, we approximated the position of each CT nodule in the airway model and calculated the nearest airway diameter and off-road distance. The diameter of the closest airways were slightly larger when the CT nodules were placed in the airway model compared to the CT scans; and the off-road distance was smaller, especially for the 6mm diameter tool. As was already mentioned in the Methods section there are some limitations with the models used in this analysis which may impact the conclusions that can be drawn from the results. First, our 3D lung model had 3311 branches. The smallest diameter airway in this model was 0.1mm, but this was not at the ends of every branch, therefore conclusions cannot be drawn for tools \<1mm in diameter. Regardless, the ability to make a \<1mm diameter tool with all the required functionality is almost impossible at this time and may therefore not be relevant. The second, and likely more impactful, limitation of these models is that there is no stretching or deformation of the airways that is taken into account in determining the airway to lesion distances. Biological tissues have some elasticity and the airways will stretch when the endobronchial tools are advanced. Therefore, it is likely that in reality more lesions are reachable and the distance that the tools would need to travel is likely to be shorter. It is also assumed that a tool traveling along an optimal path can be accurately navigated and localize any lesion. Obviously there are limitations to this today. However, with new techniques like electro-magnetic navigation bronchoscopy or image-guided bronchoscopy the ability to accurately navigate a tool along an optimal path may become more reasonable. Highly steerable, flexible tools would also be required for this. Looking at the current literature and products on the market there are a variety of tools becoming available to pulmonologists to assess airways, biopsy tissue, and treat abnormalities. Given all of these new tools that could impact patient care it is important to understand how impactful they can be given their current sizes. Additionally for anyone designing new endobronchial tools it is important to understand the design criteria needed to target the majority of lung lesions. Medtronic's superDimension^™^ Navigation system and Veran Medical Technologies' SPiN SYSTEM^™^ are probably the most widely known systems that are on the market to provide better navigational guidance, like a GPS, in the lung to localize lesions. superDimension^™^ consists of a tracking tool enabled with an electro-magnetic (EM) sensor and separate biopsy tools (1.7--1.9mm), like brushes, needles, forceps, etc., which are designed to fit down the working channel of a standard bronchoscope. From our simulations, the \~2mm EM sensor and biopsy tools would reach only 13.8--18.5% of 1cm lesions if the sensor remains in the airways. The benefit of a tracked technology is that the tools can still be located even when they must be used off-road. Overall, the clinical usefulness of EM navigational guidance is still under consideration. Some argue that it is a safe and effective tool for obtaining a diagnosis in high risk individuals that can't undergo invasive procedures \[[@pone.0189963.ref018]\]. While other reports are less optimistic as diagnostic yield can be heavily impacted by a bronchus sign on CT and respiratory motion \[[@pone.0189963.ref019],[@pone.0189963.ref020]\]. Ultrasound techniques, including linear endobronchial ultrasound (EBUS) and radial endobronchial ultrasound (REBUS) are beginning to play a larger role in the diagnosis and staging of lung malignancies by providing additional imaging of the airways \[[@pone.0189963.ref021]\]. Linear EBUS probes are typically quite large (\>6mm diameter) and therefore restricted to the central airways. REBUS probes on the other hand, are much smaller (1.4 mm), potentially reaching more peripheral lesions. From the simulations a 1.4mm REBUS probe would directly reach 16.4--22.0% of 1cm lesions. It should be noted however that REBUS has a penetration depth of a few centimeters and would therefore be able to image lesions beyond the airway wall; thus the percentage of "reachable" lesions would be at least 4x higher than a tool that needs to be in direct contact with the tissue--which is the case for a biopsy. Although REBUS can potentially image more lesions than EBUS, the probe must be exchanged with forceps or a needle to take the actual tissue sample as it's too large for the working channel to accommodate both tools. A variety of optical based techniques can be found in the literature where they are predominately used to image the structure of the airways, like ultrasound, or to provide real-time biochemical and morphological information related to tissue composition. Light-based techniques are an attractive option because they are non-destructive to the tissue and are compact enough to fit through the working channel of a bronchoscope \[[@pone.0189963.ref022]\]. Tissue characterization tools like Raman spectroscopy \[[@pone.0189963.ref031],[@pone.0189963.ref032],[@pone.0189963.ref033]\], diffuse reflectance spectroscopy (DRS) \[[@pone.0189963.ref028],[@pone.0189963.ref029],[@pone.0189963.ref030]\], fluorescence spectroscopy \[[@pone.0189963.ref028],[@pone.0189963.ref029]\], and differential path-length spectroscopy (DPS) \[[@pone.0189963.ref030],[@pone.0189963.ref034],[@pone.0189963.ref035],[@pone.0189963.ref036]\] have been investigated for lung cancer detection, achieving \>92% sensitivity and specificity with Raman \[[@pone.0189963.ref031],[@pone.0189963.ref032]\] and \>80% sensitivity and specificity with DRS and DPS \[[@pone.0189963.ref028],[@pone.0189963.ref029],[@pone.0189963.ref030],[@pone.0189963.ref039]\]. These technologies have not yet been investigated for endobronchial access and it is unknown how small these types of tools could be made. Like ultrasound, they can also measure tissue beyond the airway wall with a penetration depth on the order of millimeters, thus increasing the percentage of reachable lesions 2-fold. Another optical technique, autofluorescence bronchoscopy, is already on the market and shows improved performance in sensitivity but is still limited to the central airways due to its size and specificity is low \[[@pone.0189963.ref031]\]. To our knowledge the only group that has investigated the feasibility of peripheral airway optical sensing is that of Suter et al. \[[@pone.0189963.ref023],[@pone.0189963.ref024],[@pone.0189963.ref026],[@pone.0189963.ref027]\]. They developed an optical coherence tomography (OCT) imaging catheter integrated inside a 21-gauge TBNA needle to target peripheral airways which was tested in *ex-vivo* swine lungs \[[@pone.0189963.ref023]\]. OCT uses light scattering to produce an image of the tissue structure, similar to ultrasound but with much finer resolution and at the expense of imaging depth. The integration of both the imaging and biopsy needle into a single catheter shows real hope that a fully integrated tool could eventually be made. From our simulations, their 21-gauge needle (assuming a 2mm penetration depth) would reach \~28--36% of 1cm lesions. From the 3D simulations and CT analyses, we have shown that for most lung lesions that will be detected with LDCT screening the small size of these lesions will make it challenging to reach, biopsy, and treat them endobronchially. In the majority of patients biopsy tools and/or treatment devices such as ablation needles will need to traverse the bronchial walls and proceed off-road to successfully reach the lesions; or we need tools that are \<1mm in diameter--a challenging task. Typically the tool tip will need to travel 1-2cm off-road to reach and biopsy/treat a lesion. Given this information, any new technology being developed for navigational guidance, tissue confirmation, or treatment needs to be made small enough and should be able to traverse the bronchial wall to have the greatest impact in the pulmonology workflow. Supporting information {#sec010} ====================== ###### Matlab (.mat) file containing the x, y, and z coordinates of the simulated lesion locations with respect to the airway models. Inspiration and expiration are two different variables. Each variable contains 3 cells corresponding to the 3 models. (MAT) ###### Click here for additional data file. ###### Excel file containing the airway diameters, lesion diameters, and distances to airways from the CT scans. (XLSX) ###### Click here for additional data file. We would like to thank Hiroko Kitaoka for creating the tool to generate various airway models and for providing assistance and answering questions when needed. We would also like to thank Vijay Parthasarathy, Jeroen de Jong, Robert Poel, Tobias Klinder, Jarich Spliethoff, and Lisanne de Boer for their contributions on earlier experiments that supported portions of this work. DPS : Differential path-length spectroscopy DRS : Diffuse reflectance spectroscopy EBUS : Endobronchial ultrasound EM : Electromagnetic LDCT : Low dose CT NaN : Not a number OCT : Optical coherence tomography PDF : Probability distribution function REBUS : Radial endobronchial ultrasound TBNA : Transbronchial needle aspiration [^1]: **Competing Interests:**TMB and BHWH are employees of Philips Research in that they receive salaries. Any patents by TMB or BHWH are owned by the employer \[Philips Research\]. BHWH has stock options as part of the bonus salary payment of Philips. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

All relevant data are within the paper and its Supporting Information files. In addition RNAseq data have been deposited at EGA under accession number EGA00001001101. Introduction {#sec001} ============ Rhabdomyosarcoma (RMS) is the third most common soft-tissue sarcoma in children with an annual incidence of five new cases per million. It accounts for 5--8% of all pediatric cancer. RMS belongs to the family of small round blue cell tumors of childhood and exhibits histological features of skeletal muscle. Two major histological subtypes of RMS can be distinguished, embryonal (E-RMS) and alveolar (A-RMS). E-RMS has its highest incidence in infants and young children whereas A-RMS is more frequent in older children and adolescents. A-RMS has a more aggressive clinical behavior with early dissemination, a poor response to chemotherapy, frequent relapses, and a 5-year failure-free survival of 65% after treatment \[[@pgen.1004951.ref001]\]. A-RMS is found predominantly in the extremities (42%), parameningeal (17%), head and neck (11%) and other locations (21%) \[[@pgen.1004951.ref001]\] including the trunk, perirectal and perianal areas \[[@pgen.1004951.ref002], [@pgen.1004951.ref003]\]. Cytogenetically A-RMS is distinguished from E-RMS by one of two recurrent chromosome translocations: t(2;13) or t(1;13), which result in fusion of *PAX3* or *PAX7* to *FOXO1*, respectively \[[@pgen.1004951.ref004]\]. In spite of multiple attempts to identify the cell of origin in which the t(2;13) occurs the question remains unanswered. It was shown previously that transcription occurs at a few hundred discrete nuclear sites called transcription factories \[[@pgen.1004951.ref005]\]. Some genes frequently involved in a recurrent chromosome translocation (*MYC* and *IGH* in B lymphoid progenitors, *TMPRSS2* and *ERG* or *ETV1* in prostate cancer, *RET* and *H4* in in radiation-associated papillary thyroid cancer) co-localize to the same transcription factory \[[@pgen.1004951.ref006]--[@pgen.1004951.ref009]\]. Initial chromosome conformation capture experiments in activated mouse B cells suggested that physical proximity of the *IGH* and *MYC* loci is a minor contributor to the frequency of chromosome translocation \[[@pgen.1004951.ref010]\]. However, combined high resolution Hi-C mapping and genome-wide translocation sequencing in transformed mouse pre-B cells found good coincidence between chromosomal translocation and spatial proximity \[[@pgen.1004951.ref011]\]. A possible driver of double strand DNA breaks might be the co-localization of replication stress-induced early replication fragile sites (ERFSs) with highly expressed gene clusters \[[@pgen.1004951.ref012]\]. Though it was demonstrated that ectopic expression of PAX3-FOXO1/Pax3-Foxo1 can transform mouse mesenchymal stem cells *in vitro* \[[@pgen.1004951.ref013]\] as well as Myf6^+^ myofibers *in vivo* \[[@pgen.1004951.ref014]\] in view of the above these cell types seem unlikely hosts for the chromosome translocation given that they do not express Pax3. In fact, the suggestion that Myf6^+^ myofibers might be the host of the PAX3-FOXO translocation was recently rectified \[[@pgen.1004951.ref015]\]. In contrast, Pax3 is expressed in activated myoblasts upon muscle injury or in growing muscles during normal development \[[@pgen.1004951.ref016]\]. Moreover, PAX3-FKHR, in cooperation with loss of p16^INK4A^ expression, transforms both fetal and postnatal primary human skeletal muscle cell precursors \[[@pgen.1004951.ref017]\]. Together these observations suggest that translocation might occur in a population of activated myoblasts that express PAX3 (PAX3^+^). It has been shown that Pax3 expression differs among different muscles in the mouse \[[@pgen.1004951.ref018], [@pgen.1004951.ref019]\]. There are many more Pax3^+^ cells in fore limb than in hind limb muscles \[[@pgen.1004951.ref019]\]. Muscle satellite cells from the masseter and soleus did not express Pax3 while only 7% of those from the *extensor digitorum longus* (EDL) did. In contrast 49% of satellite cells from the biceps were Pax3^+^. In addition, most ventral trunk muscles were Pax3-positive and 64% of satellite cells from the diaphragm expressed Pax3. Importantly, primary myoblast cultures of Pax3^+^ satellite cells remain Pax3^+^, while Pax3^-^ satellite cells from hind limb remain negative \[[@pgen.1004951.ref019]\]. Studies addressing the relation between spatial chromosome proximity and translocation have been performed in cells of the B-lymphoid lineage or of hormone-responsive lineages mostly using transformed cell lines \[[@pgen.1004951.ref006], [@pgen.1004951.ref007], [@pgen.1004951.ref009]\]. Recently CRISPR-Cas9 nuclease (Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/CRISPR-associated systems) \[[@pgen.1004951.ref010]\] was used to engineer human tumor-associated translocations \[[@pgen.1004951.ref020]\]. To answer the question if locus proximity of *Pax3* and *Foxo1* in low-passage primary mouse myoblasts contributes to the frequency of *Pax3-Foxo1* fusion gene formation we used the CRISPR-Cas9 system to induce double strand DNA breaks (DSBs), which spurred by non-homologous end joining repair (NHEJ) produce chromosome translocations between these two loci. We used synthetic single-guide RNAs (sgRNA) to program Cas9 to induce DNA double-strand breaks (DSBs) in *Pax3* and *Foxo1* \[[@pgen.1004951.ref021]--[@pgen.1004951.ref023]\]. Unlike the human *PAX3/7* and *FOXO1* genes, mouse *Pax3/7* and *Foxo1* are in opposite orientation on their respective chromosomes (1, 4, and 3). Compared with human chromosome 13, *Foxo1* is part of an inverted 4.9Mb syntenic region on mouse chromosome 3. Although a recurrent complex inversion/translocation event involving the oppositely oriented *ETV6* and *c-ABL* genes in humans gives rise to the *ETV6-ABL* fusion gene in some myeloid and lymphoid malignancies, the frequency of this event is extremely low \[[@pgen.1004951.ref024]\]. Therefore, to successfully generate a CRISPR-Cas9-mediated *Pax3-Foxo1* fusion gene we used chromosomal engineering via Cre recombinase-mediated genetic alterations to create a mouse in which the Foxo1 containing 4.9 Mb syntenic region is inverted (Foxo1-inv^+/+^ mice). Previously, Cre recombinase-mediated inversions of large fragments of chromosomes have been used to create balanced chromosomes \[[@pgen.1004951.ref025]--[@pgen.1004951.ref029]\]. We show that myoblasts isolated from fore and hind limb keep their Pax3-expressing identity and co-localization of *Pax3* and *Foxo1* loci strongly correlates with the level of Pax3 expression and generation of a CRISPR-Cas9 induced t(1;3), which is more frequent in fore limb myoblasts. Our Foxo1inv^+/+^ mice will be a valuable tool for studying mechanisms underlying the initial stages of the A-RMS implicated chromosome translocations resulting in development of better animal models for this pediatric cancer and other human diseases caused by chromosome translocations. Results {#sec002} ======= Expression of *Pax3* in primary myoblasts correlates with co-localization of the *Pax3* and *Foxo1* loci {#sec002a} -------------------------------------------------------------------------------------------------------- Since close physical proximity of translocation partners might facilitate chromosome translocation, we determined if *Pax3* and *Foxo1---*the translocation partners in A-RMS---are co-localized in actively proliferating low-passage primary mouse myoblasts. DNA-FISH analyses of the *Pax3* and *Foxo1* loci in interphase nuclei of primary limb myoblasts of newborn pups, after one week in culture showed 13% co-localization, which was significantly higher than in similarly cultured MEFs (2%, the background of this method; [Fig. 1A](#pgen.1004951.g001){ref-type="fig"}). We hypothesized that co-localization of *Pax3* and *Foxo1* loci in myoblasts reflects the percentage of Pax3^+^ cells in the original newborn muscles. To test this hypothesis we isolated myoblasts from hind and fore limbs of newborn pups and compared the frequency of co-localization of *Pax3* and *Foxo1* loci in proliferating myoblasts from these two sources. It was shown previously with a *Pax3* knock-in reporter gene that many more satellite cells in fore limb muscle express *Pax3* than in hind limb muscle \[[@pgen.1004951.ref019]\]. In accordance, the percentage of co-localized *Pax3* and *Foxo1* loci was notably higher in fore limb than in hind limb myoblasts in 8 independent experiments ([Fig. 1A--C](#pgen.1004951.g001){ref-type="fig"}). In addition, Q-RT-PCR of RNA from these myoblasts confirmed that expression of *Pax3* was six-fold higher in fore limb myoblasts ([Fig. 1D](#pgen.1004951.g001){ref-type="fig"}). These results are in agreement with the published observation that satellite cells maintain their Pax3^+^ identity upon activation *in vitro*. Expression of other genes such as *Foxo1* and *Pax7* was similar in the two types of myoblasts ([Fig. 1D](#pgen.1004951.g001){ref-type="fig"}). The results for *Pax3* expression were reproducible given that a number of independent experiments produced similar data ([S1 Fig](#pgen.1004951.s001){ref-type="supplementary-material"}.). Because diaphragm was shown to contain the highest number of Pax3^+^ myoblasts \[[@pgen.1004951.ref019]\], we compared by FISH the co-localization of the *Pax3* and *Foxo1* loci in myoblasts isolated from fore limb, hind limb, and diaphragm of the same adult mouse. Indeed, diaphragm myoblasts showed a higher co-localization of the two loci (20%) than fore limb (11%) or hind limb (9%) myoblasts. {#pgen.1004951.g001} Chromosome engineered Foxo1-inv^+/+^ mice {#sec002b} ----------------------------------------- The mouse *Foxo1* gene is located on chromosome 3 in a 4.9 Mb DNA fragment (ch3:52,059,615--56,995,963) that is syntenic with human chromosome 13 (ch13:41,254,213--34,463,185) but positioned in the opposite orientation ([Fig. 2A](#pgen.1004951.g002){ref-type="fig"}). This places *Foxo1* in the mouse in a reverse transcriptional direction with respect to that of the *Pax3* or *Pax7* genes. To engineer a mouse capable of acquiring productive *Pax3/7-Foxo1* fusion genes via a simple balanced t(1;3) or t(4;3), we performed two consecutive rounds of ES cell targeting in which we introduced two pairs of non-compatible *LoxP* sites at either border of this syntenic region with the goal to create a Cre-recombinase mediated permanent inversion of the 4.9Mb DNA fragment ([S2 Fig](#pgen.1004951.s002){ref-type="supplementary-material"}.). Without inversion there would only be two ways to produce a productive fusion: 1) Via a translocation in which the resulting chromosomes would carry a double centromere and no centromere, respectively, an option likely to be non-viable in primary myoblasts and 2) Via a complex inversion/translocation event as described for the human *ETV6-ABL* fusion gene \[[@pgen.1004951.ref024]\], a rare event, which likely would reduce the frequency of fusion gene formation below detectable levels. {#pgen.1004951.g002} The centromeric border of the mouse/human syntenic region is located 15 kb upstream of the *Foxo1* start codon ([Fig. 2B](#pgen.1004951.g002){ref-type="fig"}). To precisely target this border in ES cells we used recombineering in *E. coli* \[[@pgen.1004951.ref030]\] to modify the RP24--391O12 BAC (bacterial artificial chromosome) clone, so that it carries non-compatible mutant *511-ILoxP* and *wtLoxP* sites \[[@pgen.1004951.ref031]\] flanking the *hph* (hygromycin B resistance) and *tk* (HSV1-thymidine kinase) selectable marker genes (Figs. [2B](#pgen.1004951.g002){ref-type="fig"}, [S2](#pgen.1004951.s002){ref-type="supplementary-material"}). The precise targeting of the border of the syntenic region minimizes the chance of disturbing any potentially important regulatory sequences that might affect *Foxo1* expression ([Fig. 2B](#pgen.1004951.g002){ref-type="fig"}, top). We targeted ES cells with linearized RP24--391O12-LoxP-hygro-TK BAC DNA and counter selected hygromycin B resistant clones carrying random integrations by screening for the presence of vector sequences remaining on either side of the insert. Colonies containing such vector segments were discarded \[[@pgen.1004951.ref032]\]. The remaining clones were subjected to the 'loss-of-native-allele' assay using real-time quantitative PCR \[[@pgen.1004951.ref033]\]. For copy number control of stably integrated target DNA we used the telomeric border of the syntenic region as a reference. In total 273 clones were analyzed, two of which contained a single copy of the wild type locus ([Fig. 2C](#pgen.1004951.g002){ref-type="fig"}). These clones were submitted to FISH analysis and karyotyping which confirmed the presence of only two native signals on chromosome 3 when hybridized with a wild type BAC RP24--391O12 probe ([Fig. 2D](#pgen.1004951.g002){ref-type="fig"}). For consecutive targeting of the telomeric border of the syntenic region we selected clone XIIB3, which had a 100% normal diploid karyotype. For targeting of the telomeric border of the syntenic region we followed the same strategy and engineered a BAC clone carrying the 511-ILoxp-Neo-TK-wtLoxP cassette inserted at the precise syntenic border ([Fig. 2B](#pgen.1004951.g002){ref-type="fig"}, bottom). The recombinant RP23--422I13-LoxP-Neo-TK BAC was linearized in such a way that only very short vector fragments remained at either side of the insert. After targeting in ES cells, analysis with the 'loss-of-native-allele' assay of 48 clones proved sufficient to obtain the desired recombinant. Two clones carrying a single copy of the wild type telomeric locus ([Fig. 2E](#pgen.1004951.g002){ref-type="fig"}) were analyzed by FISH using the RP24--391O12 and RP23--422I13 BAC probes. One of them (13D3) showed two native signals on chromosome 3 with either BAC probe ([Fig. 2F](#pgen.1004951.g002){ref-type="fig"}). This clone had a 90% normal diploid karyotype and we next determined if it carried cis- or trans-targeted borders of the 4.9 Mb syntenic region. To discriminate between these two possibilities we transiently transfected a Cre recombinase plasmid into the double-targeted 13D3 ES cells and DNA isolated from the pool of electroporated cells was analyzed by PCR using only forward or reverse primers from both targeted borders. Both PCRs produced bands indicating that the pool contained cells carrying the 4.9Mb inversion. The same pool of cells was counter selected with FIAU and 23 resistant clones were analyzed by PCR ([Fig. 3A](#pgen.1004951.g003){ref-type="fig"}). Sixteen clones harbored the 4.9Mb inversion and two of these were selected, A6 and C5, which had a 100% and 93% normal karyotype, respectively. Inversion of the 4.9Mb region in these clones was subsequently confirmed by FISH analysis ([Fig. 3B](#pgen.1004951.g003){ref-type="fig"}) using the RP24--391O12 and RP23--422I13 BAC probes. The chromosome containing the inversion showed split hybridization signals while the wild type chromosome produced contiguous signals with these probes. These ES cell clones were used to generate chimeric mice that transmitted the inversion of the Foxo1 syntenic region to heterozygous Foxo1-inv^+/-^ offspring. {#pgen.1004951.g003} Foxo1-inv^+/-^ mice were fertile and produced Foxo1-inv^+/+^ offspring at the expected Mendelian frequency. Foxo1-inv^+/+^ animals did not exhibit any obvious phenotypic abnormalities and showed normal fecundity and life span. Moreover, western blot analysis confirmed that Foxo1-inv^+/+^ primary myoblasts and wild type myoblasts expressed equal amounts of Foxo1 protein ([Fig. 3C](#pgen.1004951.g003){ref-type="fig"}) and co-localization of the *Pax3* and *Foxo1* loci was equal in Foxo1-inv^+/+^ and wild type myoblasts (8%, [S5 Fig](#pgen.1004951.s005){ref-type="supplementary-material"}.). Finally, DNA-FISH analysis of Foxo1-inv^+/+^ fibroblasts with RP24--391O12 and RP23--422I13 BAC probes confirmed that both chromosomes 3 carried the 4.9Mb inversion ([Fig. 3D](#pgen.1004951.g003){ref-type="fig"}). CRISPR-Cas9 induced t(1;3) reciprocal translocation {#sec002c} --------------------------------------------------- Nuclear receptor-induced chromosomal proximity of *TMPRSS2* and *ERG* or *TMPRSS2* and *ETV1* promotes the occurrence of nonrandom ligation sites upon translocation between these partner genes, thereby generating unique breakpoint "hot spots" \[[@pgen.1004951.ref006]\]. It is possible that translocations in A-RMS are non-random and occur predominantly at sites, coming in close proximity during co-regulated expression. We hypothesized that directing DSBs to sites in mouse *Pax3* and *Foxo1* homologous to those in *PAX3* and *FOXO1* in an ARMS cell line carrying a t(2;13) might increase the chance of generating a t(1;3) in proliferating Foxo1-inv^+/+^ myoblasts after Cas9 induced DSBs. We chose to mimic the breakpoints of the widely used ARMS cell line RH30 ([S3](#pgen.1004951.s003){ref-type="supplementary-material"} and [S4](#pgen.1004951.s004){ref-type="supplementary-material"} Figs.). Alignment of human and mouse *Pax3* and *Foxo1* sequences mapped the RH30-like breakpoints at positions 78105273 on mouse chromosome 1 and 52300558 on mouse chromosome 3 ([Fig. 4A](#pgen.1004951.g004){ref-type="fig"}). We chose unique protospacer sequences followed by a 5'-GGT PAM as close as possible to the RH30-like breakpoints in both *Pax3* and *Foxo1* ([Fig. 4B](#pgen.1004951.g004){ref-type="fig"}). Cas9 introduces DSB three nucleotides downstream of the two PAM sequences, which would result in DSBs between nucleotides 78105248 and 78105247 on chromosome 1 in intron 7 of *Pax3* and between nucleotides 52300541 and 52300542 (coordinates in the non-inverted sequence) on chromosome 3 in intron 1 of *Foxo1* ([Fig. 4B](#pgen.1004951.g004){ref-type="fig"}). {#pgen.1004951.g004} For gene delivery to the myoblasts we cloned the human codon optimized Cas9 (hCas9) into the pCL20C \[[@pgen.1004951.ref034]\] lentiviral vector downstream of the MSCV promoter and upstream of an IRES-YFP fluorescent marker ([Fig. 4C](#pgen.1004951.g004){ref-type="fig"}). In order to express two different sgRNAs form a single vector we constructed a second pCL20C dual sgRNA vector in which the *Pax3*-specific sgRNA was driven by the human U6 promoter and the *Foxo1*-specific sgRNA by the mouse U6 promoter ([Fig. 4C,D](#pgen.1004951.g004){ref-type="fig"}). We first determined that with our current batch of serum maximum co-localization of *Pax3* and *Foxo1* occurred at 7--8 days of culture after myoblast isolation. This time point synchronized with Cas9 and sgRNA expression should therefore maximize the probability of introducing DSB in closely positioned *Pax3* and *Foxo1* loci. Hence 24 hours after isolation we transduced primary fore and hind limb myoblasts of Foxo1-inv^+/+^ pups, Foxo1-inv^+/+^ MEFs and fore limb myoblasts from wild type mice with Cas9 lentivirus ([Fig. 4C](#pgen.1004951.g004){ref-type="fig"}). After FACS sorting for YFP, cells were expanded and transduced with lentivirus expressing the RH30-like sgRNAs at day 7 after isolation and with an SV40 large T antigen expressing lentivirus at day 8. The latter was done to prevent senescence of the myoblasts during puromycin selection and allows subsequent expansion of the culture. To detect the *Pax3-Foxo1* fusion DNA fragments from Cas9/sgRNAs expressing myoblasts and MEFs we used the Pax3-RH30F (forward) and Foxo1-RH30R (reverse) primers for PCR analysis, which are positioned downstream and upstream of the putative Cas9-induced *Pax3* and *Foxo1* DSBs ([Fig. 5A](#pgen.1004951.g005){ref-type="fig"}). PCR amplification of DNA from 10^4^ cells produced bands of 250 bp or shorter in Cas9/sgRNAs expressing hind limb and fore limb Foxo1-inv^+/+^ myoblast ([Fig 5B](#pgen.1004951.g005){ref-type="fig"}, lanes 2 and 4). However, no product was detected upon PCR amplification of DNA from 10^4^ hind and fore limb Foxo1-inv^+/+^ myoblasts not treated with Cas9/sgRNAs ([Fig. 5B](#pgen.1004951.g005){ref-type="fig"}, lanes 1 and 3) or from 10^4^ Cas9/sgRNAs expressing Foxo1-inv^+/+^ MEFs or wild type fore limb myoblasts ([Fig. 5B](#pgen.1004951.g005){ref-type="fig"}, lanes 5 and 6). As a control we verified that the difference in translocation frequency between myoblasts and MEFs was not caused by differences in CRISPR-Cas9's accessibility to chromatin, given that *Pax3* is not expressed in MEFs. The CRISPR-Cas9 breakpoint in *Pax3* falls within a *MaeIII* restriction endonuclease site and that in *Foxo1* within a *DdeI* site. Therefore we PCR amplified the Pax3 and Foxo1 fragments spanning the breakpoints and digested them with *MaeIII* or *DdeI*. This showed that 96% (*Pax3*) and 97% (*Foxo1*) of the PCR products of CRISPR-Cas9 treated myoblasts were resistant to *MaeIII* or *DdeI* digestion, whereas in CRISPR-Cas9 treated MEFs these numbers were 72% for both enzymes ([Fig. 5C,D](#pgen.1004951.g005){ref-type="fig"}). Thus, there was no great difference in chromatin accessibility. Moreover, the *Pax3* and *Foxo1* chromatin in MEFs was equally accessible to CRISPR-Cas9, despite the fact that *Foxo1* is and *Pax3* is not expressed in these cells. {#pgen.1004951.g005} Cloning of the CRISPR-Cas9 induced fusion DNAs, followed by sequencing of 45 individual clones of each of the PCR products, produced 39 and 34 translocation breakpoint sequences from fore and hind limb myoblasts, respectively. This identified 6 different breakpoint sequences from fore limb and 3 different breakpoint sequences from the hind limb myoblasts. This represents the minimal number of translocation events per 10^4^ cells ([Fig. 5E](#pgen.1004951.g005){ref-type="fig"}, top and bottom). Taking into account the percentage of locus co-localization ([Fig. 5F](#pgen.1004951.g005){ref-type="fig"}) these numbers translate to a minimal translocation frequency of 1 in 150 in fore limb and 1 in 200 in hind limb myoblasts, respectively. The only sequence in common between the fusion fragments from these two types of myoblasts was the cleanly re-ligated fusion, without missing or added base pairs. The other 7 (5 from fore limb myoblast and 2 from hind limb myoblast) were all unique and carried NHEJ-mediated deletions varying from 6 to 71 bp. Superimposed on the deletion, two of the clones also contained randomly added base pairs. Notably, three additional breakpoint sequences obtained from an independent experiment ([S5 Fig](#pgen.1004951.s005){ref-type="supplementary-material"}.) were different from the 7 shown in [Fig. 5E](#pgen.1004951.g005){ref-type="fig"} and underline the mutation-prone repair of the NHEJ DNA-repair machinery during the translocation event. Together these results show excellent correlation between the frequency of translocation, co-localization, and expression of the *Pax3* and *Foxo1* loci in primary myoblasts. It was highest in fore limb myoblasts, lower in hind limb myoblasts and undetectable in MEFs. Although wild type myoblasts show the same frequency of locus co-localization as Foxo-inv+/+ myoblasts ([S6 Fig](#pgen.1004951.s006){ref-type="supplementary-material"}.), the opposite orientation of Foxo1 prevented the formation of a productive *Pax3-Foxo1* fusion gene. Next we performed RT-PCR on equal amounts of total RNA from fore limb and hind limb myoblasts to detect the Pax3-Foxo1 fusion mRNA. In support of the higher frequency of chromosome translocation in fore limb myoblasts, we were able to RT-PCR amplify the *Pax3-Foxo1* cDNA from these myoblasts ([Fig. 5G](#pgen.1004951.g005){ref-type="fig"}) but not from the hind limb myoblasts using an equal amount of input RNA (not shown). Sequence analysis of the cDNA confirmed the correctly spliced *Pax3* exon 7-*Foxo1* exon 2 fusion ([Fig. 5G](#pgen.1004951.g005){ref-type="fig"}). FISH detection of the t(1;3) reciprocal translocation {#sec002d} ----------------------------------------------------- To further characterize the t(1;3) we repeated the experiment in Foxo1-inv+/+/Ink4a-ARF-/- myoblasts. Due to loss of a functional p53 pathway Ink4a-ARF-/- myoblasts do not senesce during further experimental manipulation. Based on the *Pax3* and *Foxo1* co-localization data at the time of induction of the t(1;3) (11% in fore limb myoblasts and 7% in hind limb myoblasts) we assumed that the frequency of translocation events in these myoblasts should not be lower than in the myoblasts used in [Fig. 5](#pgen.1004951.g005){ref-type="fig"}, i.e. at least 6 independent translocation events per 10^4^ fore limb myoblasts. This frequency is too low for further molecular and functional analyses. To enrich the cell pool for the t(1;3) carrying cells, we evenly distributed 10^4^ cells between the wells of three 96-well plates (on average 30 cells per well). PCR analyses of the DNA of 95 wells from the first plate identified 3 potentially t(1;3)-enriched cell pools ([S7 Fig](#pgen.1004951.s007){ref-type="supplementary-material"}.). Pool 1E10 was lost during the freeze-thawing cycle but FISH analyses detected the reciprocal t(1;3) in 64% of pool 1G3 metaphase cells ([Fig. 6A--C](#pgen.1004951.g006){ref-type="fig"}) and in 4% of pool 1D10 metaphase cells. Both the derivative chromosomes 1 and 3 were detected in all t(1;3) positive cells, confirming that the translocation was reciprocal. ![The reciprocal t(1;3) in myoblasts expresses Pax3-Foxo1 protein, which upregulates target gene expression.\ A-C. FISH analysis of an t(1;3)-enriched pool 1G3 myoblasts showing the reciprocal translocation. B and C show the magnified der(1) and der(3) chromosomes, respectively. D. Lysate of 1H3 (0%) and 1G3 \[64% t(1;3)\] myoblasts was immunoprecipitated with an anti-Pax3 antibody and immunoblotted with anti-Pax3 antibody (Pax3/Pax3) or anti-Foxo1 antibody (Pax3/Foxo1). The IP with anti-Foxo1 antibody was immunoblotted with anti-Pax3 antibody (Foxo1/Pax3) and an anti-Foxo1 antibody (Foxo1/Foxo1). The positions of the Pax3, Pax3-Foxo1 and Foxo1 bands are indicated.](pgen.1004951.g006){#pgen.1004951.g006} Functional analyses of cells harboring the reciprocal t(1;3) {#sec002e} ------------------------------------------------------------ To determine if the t(1;3) resulted in expression of Pax3-Foxo1 protein we immunoprecipitated three cell lysates each of the t(1;3)-negative (1H3) and t(1;3)-positive (1G3) pools with either an anti-Pax3 or an anti-Foxo1 antibody. The Pax3 IPs were then immunoblotted with the anti-Pax3 antibody, showing the Pax3 and Pax3-Foxo1 bands ([Fig. 6D](#pgen.1004951.g006){ref-type="fig"}, Pax3/Pax3 panel), or with anti-Foxo1 antibody showing only the Pax3-Foxo1 band ([Fig. 6D](#pgen.1004951.g006){ref-type="fig"}, Pax3/Foxo1 panel). Similarly, immunoblotting the Foxo1 IPs with anti-Pax3 antibody again showed the Pax3-Foxo1 fusion protein ([Fig. 6D](#pgen.1004951.g006){ref-type="fig"}, Foxo1/Pax3 panel) while immunoblotting with the Foxo1 antibody showed both Foxo1 and the fusion protein ([Fig. 6D](#pgen.1004951.g006){ref-type="fig"}, Foxo1/Foxo1 panel). This confirmed that the engineered t(1;3) expressed the fusion protein, which allowed us to assess if it affected the expression of Pax3-Foxo1's transcriptional targets. We performed RNA-seq analysis comparing the mapped sequence reads of presumed PAX3-FOXO1 target genes \[[@pgen.1004951.ref002]\] in the 1G3 pool (64% Pax3-Foxo1 positive) with those in the 1H3 pool (Pax3-Foxo1 negative) ([S1 Table](#pgen.1004951.s008){ref-type="supplementary-material"}). This showed that roughly half the targets of PAX3-FOXO1 were correctly up or down regulated in the 1G3 pool. The same comparison with a PAX3-FOXO1 expression signature obtained with the ectopic PAX3-FOXO1 expressing ERMS cell line RD \[[@pgen.1004951.ref035]\], also showed coincident regulation of half the targets ([S2 Table](#pgen.1004951.s009){ref-type="supplementary-material"}), suggesting that the t(1;3) generated fusion protein is active. Discussion {#sec003} ========== For the precise modeling of human recurrent chromosome translocations and their impact on disease development in mice, reenactment of the actual translocation would be the closest possible recapitulation of the sequence of events in humans. Until now such reenactment was a daunting task as the translocation would require introduction of *LoxP* \[[@pgen.1004951.ref036], [@pgen.1004951.ref037]\] or *Frt* recombination sites into both translocation partners via homologous recombination in ES cells, followed by expression of Cre or Flp recombinase to create DSBs that would mediate the translocation. As shown by others \[[@pgen.1004951.ref020]\] and here, the availability of the CRISPR-Cas9 system has paved the way to implementing this approach without such major technical or time investment. Given the high homology between mouse and human genes and their regulatory sequences, this approach is likely to include all sequences that are important for the precise regulation of the mouse fusion gene as it occurs in humans. The first and only published model for ARMS \[[@pgen.1004951.ref038]\] in which expression of a conditional *Pax3-Foxo1* knock-in fusion oncogene is induced by a *Myf6* driven Cre had a low incidence and long latency of tumor development, requiring the presence of two *Pax3-Foxo1* alleles on a Trp53-null or Ink4a/Arf-null background. One reason for this might be that the level of expression of the fusion oncogene in this KI model is inadequate for shorter latency tumor development. An argument against this possibility is that a high level of PAX3-FOXO1 expression induces cell death \[[@pgen.1004951.ref039]\], most likely due to transcriptional activation of the *Pax3-Foxo1* pro-apoptotic target gene *Noxa1* \[[@pgen.1004951.ref040]\]. Unlike other studies \[[@pgen.1004951.ref041], [@pgen.1004951.ref042]\], the KI *Pax3-Foxo1* gene contained some *Foxo1* genomic sequences that allowed expression of the fusion gene in adult mice, but despite their presence the construct might lack sequences that mediate human-like regulation of fusion gene expression, which in turn might be crucial for efficient tumor development. In agreement with published data \[[@pgen.1004951.ref019]\] we established that co-localization of *Pax3* and *Foxo1* in our culture system was higher in forelimb than in hind limb myoblasts, which coincided with higher Pax3 expression in forelimbs. Due to experimental variability the percentage of co-localization of the two loci varied in 8 independent experiments, but co-localization in the fore limbs was always higher than in the hind limbs. Therefore our myoblast model represents a graded system to determine if these features contributed to the frequency of chromosome translocation in low passage primary myoblasts upon introduction of targeted DSBs. To perform these experiments and to eventually develop a precise mouse model of ARMS, the transcriptional orientation of *Foxo1* on chromosome 3 needed to be inverted. We followed the Cre-dependent one-way inversion of a DNA fragment in mice as was previously demonstrated by Schnütgen and colleagues \[[@pgen.1004951.ref043]\]. To avoid disturbing the transcriptional regulation of the inverted *Foxo1*, we decided to invert the mouse/human 4.9 Mb syntenic region encompassing *Foxo1*, rather than the gene itself. Although the centromeric border of this region is only 15 kb upstream of *Foxo1*, we reasoned that all important *Foxo1* regulatory sequences should be contained within this region otherwise it would not be syntenic with human *FOXO1* on chromosome 13q14.1. Although we did not analyze the detailed expression of *Foxo1* in Foxo1-inv^+/+^ mice during pre- and postnatal life, the animals did not show any obvious phenotypic abnormalities. In addition, they had a normal lifespan, normal fecundity, and the level of Foxo1 protein expression and co-localization of the *Pax3* and *Foxo1* loci in myoblasts were identical to those of wild type mice. Together these observations made the Foxo1-inv^+/+^ myoblasts suitable for our translocation experiments. To determine if the level of co-localization of *Pax3* and *Foxo1* in primary myoblasts affected the frequency of chromosome translocation between these loci upon induction of targeted DSB, we transduced the cells with Cas9 and dual sgRNA expressing lentiviruses. Combining the three genes into a single lentiviral vector failed to produce viral particles. We targeted the CRISPR-Cas9 DSBs to sequences in *Pax3* and *Foxo1* that mediated the t(2;13) in the A-RMS cell line Rh30. Both breakpoints are present in sequences conserved between the mouse and human genes, suggesting that they occurred in non-redundant sequences that might bind factors with a role in expression regulation of both genes. Currently we do not know if this affects the frequency of translocation, which is a possibility that can be tested in future by choosing sgRNAs targeting non-conserved sequences within the target *Pax3* and *Foxo1* introns. We found excellent positive correlation between the frequency of the t(1;3) and the percentage of locus co-localization using FISH analysis. This also correlated with the level of Pax3 expression, which is much higher in fore limb than hind limb myoblasts and absent in MEFs, while Foxo1 expression is ubiquitous. Given that the frequency of CRISPR-Cas9 induced DSBs in *Pax3* and *Foxo1* is comparable in myoblasts and MEFs, it is the proximity of the loci in these primary cells that facilitates trans-chromosomal ligation producing the two expected derivative chromosomes during NHEJ DNA repair. The derivative chromosome 3 produced correctly spliced *Pax3-Foxo1* mRNA, encoding active Pax3-Foxo1 protein that up/down-regulated expression of approximately half the presumed PAX3-FOXO1 targets in the 64% *Pax3-Foxo1*-positive cell pool ([S1 Table](#pgen.1004951.s008){ref-type="supplementary-material"}). The genes compiled in this table are differentially expressed in ARMS versus ERMS tumors or have been identified by forced expression of PAX3-FOXO1 in different cell lines, including NIH3T3 cells, MEFs, SAOS2 cells and C2C12 cells (\[[@pgen.1004951.ref002]\] and references therein). Because the cell background affects the range of PAX3-FOXO1 target gene expression \[[@pgen.1004951.ref044]\], none of the published scenarios reflect expression of Pax3-Foxo1 in primary p16/Arf^-/-^ mouse myoblasts. Possibly this is the reason for the 45% match of reported PAX3-FOXO1 up or down regulated genes. Comparison with genes up or down regulated in the ERMS cell line RD transduced with *PAX3-FOXO1* retrovirus \[[@pgen.1004951.ref035]\] showed 52% coincident regulation ([S2 Table](#pgen.1004951.s009){ref-type="supplementary-material"}). Clearly, the t(1;3) generated Pax3-Foxo1 protein in mouse myoblasts is active and changes the expression of target genes in an ARMS-like manner. One would expect that the frequency of translocation in myoblast that show co-localization of the two translocation partners would be the same irrespective of the source of myoblasts. We found a frequency of 1/150 and 1/200 in fore and hind limb myoblasts, respectively, which we believe does not represent a difference given the uncertainty of how many translocation events actually took place (we can only count those that give distinguishable fusion products). Our results in mouse myoblasts suggest that human myoblasts can be a cell of origin for the *PAX3-FOXO1* translocation as they would provide a favorable environment for the translocation to occur, i.e. expression of both genes and spatial co-localization. It is curious that A-RMS is more frequent in the lower than in the higher extremities in humans, as reported by Neville and co-workers \[[@pgen.1004951.ref045]\]. This apparent inconsistency with our mouse data might be explained by the possibility that humans may not have a difference in the distribution of PAX3 expression in the upper and lower extremities. In addition, the muscle mass and presumably the number of satellite cells in the lower extremities in humans is much higher than in the upper extremities, hence increasing the number of translocation-competent cells and frequency of translocation. By using CRISPR-Cas9 nuclease we showed that targeted chromosome translocations could be induced with high efficiency. Unlike other approaches that have relied on induction of chromosome translocation using γ−irradiation, DSB-inducing chemicals, or the lymphoid cell-specific gene rearrangement machinery, CRISPR-Cas9 can be employed in any cell type. Due to its specificity the system is suitable for use *in vivo* in cell culture or in mice. Application of this system will greatly facilitate the development of mouse models that precisely recapitulate chromosome translocation-induced human diseases. Materials and Methods {#sec004} ===================== Strains of *E. coli*, BAC clones, PCR Primers and oligonucleotides {#sec004a} ------------------------------------------------------------------ A complete list of *E. coli* strains used for this work can be found in [S1 Protocol](#pgen.1004951.s010){ref-type="supplementary-material"}. BAC clones RP24--391O12 (centromeric border of the 4.9 Mb syntenic region) and RP23--422I13 (telomeric border of the 4.9 Mb syntenic region) were purchased from the BACPAC Resource Center (BPRC), Children's Hospital Oakland Research Institute in Oakland, California, USA (<http://bacpac.chori.org>). The complete list of PCR Primers and oligonucleotides can be found in [S1 Protocol](#pgen.1004951.s010){ref-type="supplementary-material"}. Construction of plasmid and BAC-based targeting vectors {#sec004b} ------------------------------------------------------- A modified pNeoTKLoxP was recombineered into BAC RP23--422I13 (telomeric border of the syntenic region). In pNeoTKLoxP we replaced the wild type (wt) LoxP site downstream of the TK gene with the 511-ILoxP sequence (annealed oligonucleotides TK-511-ILoxP and TK-511-ILoxP-C). Then, via recombineering, we introduced the EM7 promoter upstream of the Neo gene. We therefore transformed electrocompetent DY380 *Ecoli* cells, containing the wtLoxPNeoTK-511-IloxP plasmid with the TK-EM7-Neo fragment (ends of the annealed oligonucleotides TK-EM7 and EM7-NeoC had been filled-in with Klenow DNA polymerase (Invitrogen) following the manufacturer's protocol). For recombineering we followed the protocol posted on the Frederick National Laboratory for Cancer Research web site: <http://ncifrederick.cancer.gov/research/brb/protocol/Protocol1_DY380.pdf>. A short 5'-arm (annealed phosphorylated oligonucleotides 5-tel-s and 5-tel-s-C) was cloned downstream of 511-IloxP and a short 3'-arm was cloned upstream of wtLoxP (annealed phosphorylated oligonucleotides 3-tel and 3-tel-C). A modified pBSLoxPTKhygro plasmid (kind gift from Drs. M. Roussel and F. Zindy, SJCRH) was recombineered into BAC RP24--391O12 (centromeric border of the syntenic region). In this construct we inserted a 511-ILoxP sequence upstream of the TK-promoter-Neo sequence. Since the activity of TK promoter in prokaryotic cells was sufficient to ensure Hygromycin B resistance, we did not introduce the bacterial EM7 promoter in this construct. A short 5'-arm (annealed phosphorylated oligonucleotides 5-cent-s and 5-cent-s-C) was cloned upstream of the 511-IloxP site and a short 3'-arm was cloned downstream of wtLoxP (annealed phosphorylated oligonucleotides 3-cent and 3-cent-C). The pCL20c-MSCV-IRES-YFP vector backbone was generated by replacing GFP of pCL20c-MSCV-GFP \[[@pgen.1004951.ref046]\] with I-YFP from MSCV-I-YFP \[[@pgen.1004951.ref038]\]. hCas9 \[[@pgen.1004951.ref023]\] was then cloned downstream of MSCV into pCL20c-MSCV-IRES-YFP. The mU6 fragment was generated by PCR using pSicoR-GFP (Addgene, Cambridge, MA, USA) and cloned downstream of hU6 in pLKO.1 (Addgene, Cambridge, MA, USA). The cassette containing the human and mouse U6 promoters (hU6 and mU6) followed by AgeI and EcoNI cloning sites was cloned upstream of the β-actin promoter of the modified pCL20c vector, containing the β-actin-puro cassette from pJ6.OMEGA.puro \[[@pgen.1004951.ref047]\]. The spacer sequence of hU6 driven sgRNA starts with GG followed by 18 specific nucleotides from the target sequence, and mU6 driven sgRNA starts with GT followed by 18 specific nucleotides from the second target sequence ([Fig. 4C](#pgen.1004951.g004){ref-type="fig"}). Synthetic ds-DNA fragments, coding Pax3_RH30 sgRNA and Foxo1_RH30 sgRNA were cloned into AgeI and EcoNI sites under control of hU6 and mU6 promoters of pCL20C-hU6-mU6-βact-puro, respectively ([Fig. 4C,D](#pgen.1004951.g004){ref-type="fig"}). pCL20C-MSCV-Luc2--2A-LgT was constructed by replacing IRES-YFP with a Luc2--2A-LgT cassette. Lentivirus production {#sec004c} --------------------- Lentivirus was produced as described in \[[@pgen.1004951.ref046]\]. ES cell targeting, Cre-mediated inversion, and screening of ES cell clones {#sec004d} -------------------------------------------------------------------------- F12 (129SvJ-derived) embryonic stem (ES) cells were electroporated and selected for hygromycin B or G418 resistance using standard procedures. In short, 25--45 μg of linearized BAC DNA was electroporated into 2\*10^7^ ES cells followed by selection with 100 μg/ml Hygromycin B or 200 μg/ml G418. RP24--391O12-LoxP-hygro-TK was linearized with PI-SceI (NEB) and RP23--422I13-LoxP-Neo-TK was linearized with NotI (NEB). Drug resistant clones were picked after 7--9 days of selection. DNA from these clones was used for PCR analysis. Screening of homologously recombined ES cell clones was done by PCR and qPCR. The presence of vector arms remaining on either side of the insert was detected by PCR with primers pTARBAC1--3F and pTARBAC1--3R for 3'-located sequences and pTARBAC1--5F and RP24--5R for 5'-located sequences. The "loss-of-native allele" assay was performed as described in \[[@pgen.1004951.ref033]\] with some minor modification. For quantitative (q)PCR we used SYBR®Green PCR Master Mix (Applied Biosystems). qPCR was performed with the RP24-F and RP24-R primers to determine the copy number of the centromeric locus and the RP23-F and RP23-R primers to determine the copy number of the telomeric locus. Ratios between the copy numbers of the two loci were determined either by a standard dilution curve or by the Δct method. A double targeted ES cell clone was electroporated with a Cre-expressing plasmid (pMC-CRE) using the Amaxa™ Mouse ES Cell Nucleofector™ Kit (Lonza, Germany) according to the manufacturer's protocol. After 5 days of selection with 0.2μM of fialuridine (FIAU) (a kind gift of Bristol Myers) ES cells were collected for DNA isolation as a pool or as single clones. Cre-mediated inversion was detected by standard PCR using the RP24-F/RP23-F2, and RP24-R/RP23-R2 primer pairs. Fluorescence *in situ* hybridization (FISH) {#sec004e} ------------------------------------------- For FISH analyses of *Pax3* and *Foxo1* co-localization and detection of t(1;3) reciprocal translocation we used BAC probes RP23--260F1 and RP24--391O12. For FISH analyses of targeted ES cells and Foxo1-inv^+/+^ fibroblasts we used BAC probes RP24--391O12 (centromeric border of the syntenic region) and RP23--422I13 (telomeric border of the syntenic region). BAC probes were labeled with nick translation using either Green (RP23--260F1and RP24--391O12) or Red (RP23--422I13) dUTP (Abbott Molecular). Probes were hybridized to metaphase and/or interphase cells either separately or as a 1:1 mixture in hybridization solution (50% formamide, 10% dextran sulfate, and 2X SSC). Slides were washed in 2X Saline-Sodium Citrate (SSC) buffer containing 50% formamide at 37°C for 5 minutes. Cells were counterstained with DAPI and analyzed using a Nikon E80i fluorescence microscope (Nikon) with a 100× oil immersion objective. Successfully targeted clones showed 2 native signals for the centromeric or telomeric targeted regions. Inverted chromosomes 3 appeared as two linked pairs of red and green signals on interphase cells, each pair representing one end of the inverted chromosome segment. Normal chromosomes 3 appeared as a single loosely paired red and green signal. One hundred interphase nuclei were scored for the presence of co-localization of Pax3/Pax7 and Foxo1 signals. Only nuclei with discernible red and green signals were scored. Fifty metaphase cells from CRISPR-Cas nuclease treated myoblasts were scored for the presence of cells containing the reciprocal translocation between *Pax3* and *Foxo1a*. Identification of the translocation breakpoints in A-RMS {#sec004f} -------------------------------------------------------- Experimental details are provided in [S1 Protocol](#pgen.1004951.s010){ref-type="supplementary-material"}. Position of primers, used for LD-PCR, gel electrophoresis of LD-PCR products and sequences flanking the breakpoint in the Rh30 cell line are shown in [S3](#pgen.1004951.s003){ref-type="supplementary-material"} and [S4](#pgen.1004951.s004){ref-type="supplementary-material"} Figs. Detection of RH30-like translocation in primary myoblasts {#sec004g} --------------------------------------------------------- Cas9 induced translocation was detected by PCR of chromosomal DNA from 10^4^ cells using Pax3-RH30F and Foxo1-RH30R primers. qRT-PCR and RT-PCR of myoblasts {#sec004h} ------------------------------- For each qRT-PCR reaction we used RNA isolated from either 2.5×10^3^ (data in [Fig. 1](#pgen.1004951.g001){ref-type="fig"}) or 1.6×10^3^ (data in [S1 Fig](#pgen.1004951.s001){ref-type="supplementary-material"}.) cells. For each RT-PCR we used RNA isolated from 6.7×10^3^ cells. For RT we used SuperScript III First-Strand Synthesis SuperMix (Invitrogen) with an equimolar mix of Pax3R primer and random hexonucleotides and performed the reaction following the manufacturer's protocol. For the qPCR step we used TaqMan Gene Expression Master Mix (Applied Biosystems). Ratios between gene expression in different cell lines were determined by a standard dilution curve. Immunoprecipitation and western-blot analyses {#sec004i} --------------------------------------------- Myoblasts (5×10^6^) were lysed in 0.5 ml CHAPS lysis buffer (40 mM HEPES \[pH 7.4\], 1 mM EDTA 120 mM NaCl, 10 mM sodium pyrophosphate, 10 mM β-glycerophosphate, 0.3% CHAPS, 50 mM NaF, 1.5 mM NaVO, 1 mM PMSF, and 1 tablet of EDTA-free protease inhibitors \[Roche\] per 10 mL solution) and freeze-thawed 3 times, followed by centrifugation at 20,000 ×*g* for 10 min at 4°C. After adding 2 μg anti-Pax3 antibody \[[@pgen.1004951.ref048]\] or anti-Foxo1 (C29H4) Rabbit mAb (Cell Signaling) Pax3, Foxo1 and Pax3-Foxo1 were immunoprecipitated overnight at 4°C. Immunoprecipitated material was bound onto 10 μl protein G-coated Dynabeads (Invitrogen) for 90 minutes at 4°C, which were captured using a DYNA-Mag-2 magnet (Invitrogen), washed 4 times with CHAPS buffer, and removed from the beads by heating to 70°C in 1.25xLDS loading buffer (Invitrogen) in CHAPS and separated on pre-cast 4%--12% bis-tris polyacrylamide gels. Western-blotting was performed using the same anti-Foxo1 and anti-Pax3 antibodies. Enrichment of t(1;3) harboring cells {#sec004j} ------------------------------------ To enrich for cells harboring the t(1;3), 10^4^ of the Cas9/sgRNAs expressing fore limb myoblasts were evenly distributed over three 96-well plates (on average 30 cells per well). DNA from each cell pool was isolated and analyzed for the presence of t(1;3) translocation using PCR. RNA sequence analysis {#sec004k} --------------------- Libraries were generated from \~ 500 ng total RNA of the 1H3 (no Pax3-Foxo1) and 1G3 (64% Pax3-FOXO1) cell pools using the Illumina TruSeq Stranded mRNA Sample Preparation Kit. Libraries were sequenced on an Illumina HiSeq 2500 using paired-end 100 bp sequencing chemistry. Paired-end reads from RNA-seq were aligned to the following 4 database files using BWA (0.5.10) aligner: (1) the human GRCh37-lite reference sequence, (2) RefSeq, (3) a sequence file representing all possible combinations of non-sequential pairs in RefSeq exons, (4) AceView database flat file downloaded from UCSC representing transcripts constructed from human ESTs. The mapping results from (2) to (4) were aligned to human reference genome coordinates. In addition, they were aligned using STAR 2.3.0 to the human GRCh37-lite reference sequence without annotations. The final BAM file was constructed by selecting the best alignment among the five map events. We used HTSeq \[[@pgen.1004951.ref049]\] to count the number of fragments that mapped to each gene (Gencode v 15), where each gene is considered as the union of all its exons. Then we normalized the count to FPKM (fragments per kilobase of exons per million fragments mapped) as the expression value of the gene. RNA-seq of both samples produced 55M reads each, with a 20X coverage of 43.561% of the exons in 1H3 and 43.992% of the exons in 1G3. Ethics statement {#sec004l} ---------------- This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The Institutional Animal Care and Use Committee (IACUC) of St. Jude Children's Research Hospital (Animal protocol number 209--100171) approved the protocol. Supporting Information {#sec005} ====================== ###### Q-RT-PCR analysis of *Pax3*, *Foxo1*, *and Pax7* expression in primary mouse fore limb and hind limb myoblasts. Upper 3 panels: Q-RT-PCR of a second fore and hind limb myoblast isolation experiment showing expression of *Pax3*, *Foxo1* and *Pax7* in both types of myoblasts. Lower 3 panels Q-RT-PCR showing the difference in *Pax3* expression in fore and hind limb myoblasts in 3 additional, independent myoblast isolations. (PPT) ###### Click here for additional data file. ###### CRE recombinase-induced irreversible inversion of a 4.9Mb region on mouse chromosome 3 in ES cells. 1°---Schematic representation of the mouse/human 4.9 Mb syntenic region (gradient blue rectangle) on wild type mouse chromosome 3. The red arrow inside the gradient blue rectangle indicates the position and transcriptional direction of *Foxo1*. The arrows above the rectangle indicate the positions of the blue F1 centromeric and the red F2 telomeric forward primers and the blue R1 centromeric and red R2 telomeric reverse primers. 2°---Mouse chromosome 3 after consecutive centromeric and telomeric targeting of the borders of the 4.9 Mb syntenic region. Arrows underneath 511-ILoxP and LoxP---two incompatible loxP sites, indicate their relative orientation; hph---Hygromycin B resistance gene; neo---G418/Neomycin resistance gene; tk---HSV1-thymidine kinase gene. 3°A---First step of Cre-induced inversion of the double-targeted 4.9 Mb region. If the LoxP sites are used, the inversion of the 4.9Mb fragment places one LoxP site, the *tk* genes, and the antibiotic resistance genes between head-to-tail oriented 511-ILoxP sites. 3°B---Alternative first step of Cre-induced inversion of the double-targeted 4.9 Mb region. If the 511-ILoxP sites are used, the inversion of the 4.9Mb fragment places one 511-ILoxP site, the *tk* genes, and the antibiotic resistance genes between tail-to-head oriented LoxP sites. 4°---Cre-induced recombination via the 511-ILoxP (3°A) or the LoxP (3°B) sites results in the irreversible inversion of 4.9M syntenic region and flips the transcriptional orientation of *Foxo1a*. (PPT) ###### Click here for additional data file. ###### Localization of primers used in LD-PCR. The human sequence of *PAX3* intron 7 (A) and *FOXO1* intron 1 (B) was compared to the mouse sequence using the ECR Browser genome analysis tool to show the location of evolutionarily conserved regions within these loci. Sequence comparison was performed with an ECR window of 100bp with a minimum similarity of 70%. The position of the exons (blue boxes) is indicated as well as the 5' to 3' direction of the gene (blue arrows). Blue peaks correspond to coding exons, yellow peaks correspond to 5' or 3' untranslated regions, orange peaks correspond to intronic non-coding conserved sequences and green peaks correspond to repetitive sequences. Primer positions are indicated (black arrows) with their names shown above. The *PAX3* primers *PAXLD1--7* are designated pP1--7. Primers *PAXLD1--6* were designed 3kb apart from each other to span the *PAX3* intron 7. Primer *PAXLD7* is located immediately upstream of *PAX3* exon 7. The *FOXO1* primers *FOXLD1--12* are designated pF1--12. Primers *FOXLD1--11* are located 10kb apart from each other. Additional primers were designed between primers *FOXLD4--11* to reduce the primer interval to 5kb. Primer *FOXLD12* is located immediately downstream of exon 2. (TIF) ###### Click here for additional data file. ###### Amplification of the translocation breakpoint of the derivative chromosome 13 of the ARMS cell line Rh30. \(A\) Gel electrophoresis of LD-PCR products obtained in the amplification across the breakpoint in the Rh30 cell line. Lanes 1--3: reverse primer *FOXLD6* in combination with *PAXLD2*, *PAXLD3* and *PAXLD4* generated fragments of 5.8kb, 8.8kb and 11.8kb, respectively. Lane 4: primers *FOXLDwt* and *FOXLD8* resulting in a 10.1kb long fragment (positive control). Lane 5: 1kb DNA ladder (Invitrogen). (B) Sequences flanking the breakpoint in the Rh30 cell line showing the seamless transition between chromosomes 2 and 13. (TIF) ###### Click here for additional data file. ###### Additional *Pax3-Foxo1* fusion sequences from an independent CRISPR-Cas9 translocation experiment. Three additional fore limb myoblast *Pax3-Foxo1* fusion sequences are shown below the predicted *Pax3-Foxo1* fusion sequence. These were generated in an independent experiment producing sequences that are distinct from those depicted in [Fig. 5E](#pgen.1004951.g005){ref-type="fig"}. Nucleotides in lower case represent Pax3 sequences, capitals represent Foxo1 sequences. Nucleotides in red lower case have been added randomly via NHEJ repair. (PPTX) ###### Click here for additional data file. ###### Co-localization of *Pax3* and *Foxo1* in Foxo1-inv^+/+^ and wild type myoblasts. FISH analysis of Foxo1-inv^+/+^ (top left) and wild type myoblasts (top right) hybridized with BAC probes RP23--260F1 (green, *Pax3*) and RP24--391O12 (red, *Foxo1*). The table underneath shows the frequency of locus co-localization. (TIF) ###### Click here for additional data file. ###### Enrichment of t(1;3) myoblasts from pools of CRISPR-Cas9 treated Foxo1-inv+/+/Ink4a-ARF-/- myoblasts. PCR analysis of pools, originated from \~30 cells of Cas9/RH30-like transduced Foxo1-inv+/+/Ink4a-ARF-/- myoblasts, 96-well plate \#1. Numbers and letters indicate the position of the pool in the 96-well plate; m---1kb+ marker: position H12 corresponds to DNA from the original non-enriched Cas9/RH30-like transduced Foxo1-inv+/+/Ink4a-ARF-/- myoblasts. (TIF) ###### Click here for additional data file. ###### PAX3-FOXO1 target genes in ARMS. List of genes reported to be regulated by PAX3-FOXO by comparing gene expression levels in ARMS versus ERMS tumors and by forced expression of PAX3-FOXO1 in MEFs, NIH3T3, SAOS2 and C2C12 as published in \[[@pgen.1004951.ref002]\] and references therein. The last column lists the up or down regulated genes in 64% t(1;3)-positive mouse myoblasts. These numbers were obtained by dividing the number of mapped RNA-seq reads in 64% *Pax3-Foxo1*-positive (1G3) cell pool by those in the wild type (1H3) cell pool. 45% of the genes (bold script) are regulated in the direction published. ND = no reads detected. (XLSX) ###### Click here for additional data file. ###### PAX3-FOXO1 expression signature in *PAX3-FOXO1* transduced cells. List of genes regulated by PAX3-FOXO1 in the ERMS cell line RD transduced with a *PAX3-FOXO1* retrovirus as reported by \[[@pgen.1004951.ref050]\]. The last column lists the up or down regulated genes in 64% t(1;3)-positive mouse myoblasts. These numbers were obtained by dividing the number of mapped RNA-seq reads in the 64% *Pax3-Foxo1*-positive (1G3) cell pool by those in the wild type (1H3) cell pool. 52% of the genes (bold script) are regulated in the direction published. ND = no reads detected. (XLSX) ###### Click here for additional data file. ###### Strains of *E. coli.* PCR primers and oligonucleotides. Identification of the translocation breakpoints in ARMS. (DOCX) ###### Click here for additional data file. We thank Drs. M. Roussel and F. Zindy (Dept. of Tumor Cell Biology, SJCRH) for DNA constructs. S. T. Ragsdale (Cytogenetics Core Lab, SJCRH), T. K. Walker (Transgenic Core Unit, SJCRH), S. Knoeppel and L. Emmons (Cryopreservation Lab, SJCRH) for technical support. We thank Scott Olsen for RNA sequencing and Yinping Fan for analysis of the RNA-sequence data. We are indebted to Dr. F. Zhang for providing the CRISPR-Cas9 reagents via Addgene (plasmid 42230) that made this work possible. [^1]: The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: IVL VV MBV FP FH BVB JJC GCG. Performed the experiments: IVL VV MBV FP FH BVB JJC. Analyzed the data: IVL MBV JJC GCG. Wrote the paper: IVL JJC GCG. Designed and executed the work and wrote the manuscript: IVL. Performed the FISH co-localization experiments: VV MBV. Performed western blotting: FP. Performed IPs followed by western blotting: FH. Determined the RH30 breakpoints: BVB JJC. Wrote part of the paper: JJC. Designed the experiments and wrote the paper: GCG.

1. Introduction {#sec1-ijerph-15-00919} =============== Geogenic arsenic (As) contamination in groundwater occurs in more than 107 countries worldwide \[[@B1-ijerph-15-00919]\]. About 100 million people of Bangladesh and the state West Bengal of India together \[[@B2-ijerph-15-00919]\], and more than 296 million of the global population in total \[[@B1-ijerph-15-00919]\], are at risk of As poisoning. In As-prone areas, many people suffer from skin disorders, respiratory, nervous, obstetric system disorders, diabetes, cardiovascular diseases, as well as cancers of various organs, such as skin, kidney, liver, bladder, etc. \[[@B2-ijerph-15-00919]\]. Bangladesh is a densely-populated country and to cope with the demand of the burgeoning population, it has intensified rice production especially during the dry season, which provides higher a yield than rice varieties cultivated in the wet season. The dry season rice requires large amounts of irrigation water that reach as high as 11,500 m^3^/ha \[[@B3-ijerph-15-00919]\], mostly extracted from groundwater through shallow tube-wells. Unfortunately, elevated levels of naturally-originated As have been detected in groundwater of Bangladesh \[[@B4-ijerph-15-00919]\] and the As-contaminated groundwater is extensively used for irrigating crops, especially paddy rice. A study estimated that irrigation with As-contaminated groundwater introduced 1360 tons of As into paddy soil each year in Bangladesh \[[@B5-ijerph-15-00919]\], although the mass balance and fate of the introduced As in agricultural soils is yet to be determined. Rice grown in As-contaminated soil and irrigation water increase As concentration in rice grain, shoos, and straw, which are primarily used as human food and cattle feed \[[@B6-ijerph-15-00919]\]. The concentration of As exceeded the World Health Organization (WHO) guideline value (10 µg/L) for drinking water in 42% of the 52,202 tube wells in the contaminated regions \[[@B4-ijerph-15-00919]\]. Concentrations of other elements, such as Mn, Cr, Pb, and Ni also exceeded the WHO health-based drinking water guideline values in the groundwater of Bangladesh \[[@B7-ijerph-15-00919]\]. Mushrooms are heterotrophic eukaryotic organisms classified in the kingdom of fungi. There are about 5.1 million fungal species in the world \[[@B8-ijerph-15-00919]\]. Over 75,000 species of fungi exist in the European continent, of which over 15,000 species are macro fungi (fungi that form fruiting bodies or sporocarps) which are visible to the naked eye \[[@B9-ijerph-15-00919]\]. The fruiting body bears spores is the morphological part of the fungus that is commonly called mushroom. More than 2000 mushroom species prevail in nature, of which only about 22 species are cultivable \[[@B10-ijerph-15-00919]\]. They have been an important part of diet in many countries \[[@B11-ijerph-15-00919]\], especially the cultivated ones, such as *Agaricus* spp., *Pleurotus* spp., *Lentinus edodes*, *Volvariella volvacea*, and *Auricularia* spp. \[[@B12-ijerph-15-00919]\]. Across the world mushroom cultivation is now a multi-billion dollar business \[[@B13-ijerph-15-00919]\]. Commercial mushroom cultivation was initiated by the Bangladesh Agricultural Research Council in Mushroom Culture Centre at Savar, Dhaka in early 1980s \[[@B14-ijerph-15-00919]\] as Bangladesh is one of the most suitable countries for mushroom cultivation due to its tropical monsoon type climate and low production cost. Apart from Savar, mushroom cultivation has been booming near the capital and other districts of Bangladesh. In the last 20 years, mushroom consumption and production have increased at a faster rate than almost any other agricultural food products. Rice straw and sawdust are usually used as media to cultivate mushrooms in Bangladesh; of course, wheat straw, gypsum, oak and beech sawdust or chicken manure are also used as typical substrates for mushroom cultivation in many parts of the world \[[@B15-ijerph-15-00919],[@B16-ijerph-15-00919],[@B17-ijerph-15-00919]\]. The literature demonstrates that any of the substrates, if collected from polluted areas, can facilitate higher accumulation of trace elements, particularly As and Hg \[[@B18-ijerph-15-00919],[@B19-ijerph-15-00919]\]. One study revealed that As concentration in rice straw increased significantly (from 3.9 mg/kg to 91.8 mg/kg) with increasing As concentration in irrigation water (up to 8.0 mg/L of As) \[[@B6-ijerph-15-00919]\]. In As-contaminated areas, rice straw has been found to contain 1.15 mg/kg of As \[[@B20-ijerph-15-00919]\]. Various investigations have dealt with metal concentrations of mushrooms, especially edible ones, and many edible mushroom species are known to accumulate high levels of heavy metals \[[@B21-ijerph-15-00919]\]. However, there have not been any previous studies focused on the concentrations of As and other toxic elements in edible mushrooms grown in Bangladesh. This study reports concentrations of As and other elements in edible mushrooms available near Dhaka City, Bangladesh. We use those data to assess whether mushroom consumption poses any health risks to humans, given that As, Cr, Cd, Pb, and Hg are highly toxic and the exposure route is often through food. 2. Materials and Methods {#sec2-ijerph-15-00919} ======================== 2.1. Mushroom Cultivation and Consumption in Bangladesh {#sec2dot1-ijerph-15-00919} ------------------------------------------------------- There is no information available on daily consumption of mushroom/person in Bangladesh. In Bangladesh, mushroom production and consumption is limited to urban and semi-urban areas. Mushrooms are consumed by 77% of higher-income groups and by 37% of medium-income groups \[[@B22-ijerph-15-00919]\]. Before the 1970s, mushrooms were usually consumed by some of the tribes as per their custom, but now the situation has changed due to mushrooms' agreeable aroma, excellent subtle flavour and, above all, their nutritional and medicinal values. Many exotic food preparations, such as soup, vegetables, pickles, etc., are made from mushrooms, and have become highly popular for about 58% of customers in Bangladesh \[[@B22-ijerph-15-00919]\]. According to the most recent database, about 620--675 tons of fresh mushrooms is produced in Bangladesh per annum \[[@B23-ijerph-15-00919]\], thus, the mushroom consumption rate is estimated as 0.09 g (fresh weight) weekly, considering 675 tons of mushroom production per annum and the total population for that time was 140.6 million \[[@B23-ijerph-15-00919]\]. Data from the wholefood catalog \[[@B24-ijerph-15-00919]\] indicates that the dry matter content of the mushrooms averages 13%, giving an average weekly intake on a dry matter basis of 0.01 g. Generally mushrooms from three genera, namely *Pleurotus*, *Agaricus*, and *Calocybe* are cultivated in Bangladesh at the rate of 97%, 1%, and 1%, respectively \[[@B23-ijerph-15-00919]\] and they are commonly called Oyster, Button, and Milki mushrooms, respectively. The additional 1% of mushrooms are mainly *Volvariella* spp. and *Auricularia* spp. It was also reported that among the *P.* spp., 61% were *P. ostreatus*, followed by 19% *P. sajorcaju*, 12% *P. florida*, and 8% of *P. high king*-*51* \[[@B23-ijerph-15-00919]\]. At the time of collecting fresh edible mushrooms for the present study from the Mushroom Development and Extension Institute, Savar (on the outskirts of the capital, Dhaka) and a commercial farm of Gazipur (about 10 km from the main city), only *P. ostreatus* and *P. high-king* were in the production phase. 2.2. Sampling, Sample Processing, Digestion and Analysis {#sec2dot2-ijerph-15-00919} -------------------------------------------------------- Fresh edible mushrooms (*P. high-king* and *P. ostreatus*; *n* = 2 × 20) with respective growing media (rice straw; *n* = 2 × 20) were collected from the "Mushroom development and Extension Institute" Savar, Dhaka, and the same from a private mushroom farm at Gazipur during 2014. Some commercially-available powdered mushroom samples (*Pleurotus* sp.; *n* = 5) were also purchased from local markets in Dhaka and Gazipur. The substrate debris and stalks of the mushroom samples were removed with a disposable plastic knife. The mushroom samples were then washed in running tap water (three times) to remove the debris followed by washing three times with deionised water. The washed samples were then air dried for approximately 48 h at 20 °C and then placed in an oven for drying at 60 °C. The dried samples were then homogenized and ground. A similar procedure was also employed for the processing of growing media. 2.3. Sample Preparation {#sec2dot3-ijerph-15-00919} ----------------------- A microwave digestion system with 40 rotors (model: MARS 6, CEM Corporation, Matthews, NC, USA) was used for the digestion of the samples. Trace grade concentrated nitric acid (Fisher Chemicals, Hampton, NH, USA) was used for the digestion of mushroom. The As speciation study was conducted after extracting samples with 2 M trifluoroacetic acid (TFA) as per standard procedure \[[@B6-ijerph-15-00919]\]. 2.4. Sample Analysis {#sec2dot4-ijerph-15-00919} -------------------- An Agilent 7500ce (Agilent Technologies, Tokyo, Japan) inductively-coupled plasma mass spectrometer (ICP-MS) was used to determine the amount of As and other elements in mushrooms and rice straw. An Agilent liquid chromatography system (1100 series, Agilent Technologies, Tokyo, Japan) was used for As speciation study. A Hamilton PRPX-100 along with a guard column coupled with ICP-MS was used to measure As species (As(III), As(V)), MMA(V), and DMA(V) in mushroom. The elemental concentration of samples was determined on a dry weight basis. 2.5. Quality Control: Analysis of Standard Reference Material (SRM) {#sec2dot5-ijerph-15-00919} ------------------------------------------------------------------- Standard reference material (SRM) from the National Institute of Standards and Technology (NIST), such as 1573a (tomato leaves), were analysed for As and other elements utilizing the same procedure as that used for mushroom and rice straw samples to check the effectiveness of the digestion and analytical procedure. 2.6. Health Risk Index (HRI) {#sec2dot6-ijerph-15-00919} ---------------------------- The health risk index was measured as the ratio of estimated exposure of mushroom and oral reference dose \[[@B25-ijerph-15-00919]\]. Estimated exposure was obtained by dividing weekly intake of elements by their safe limits or oral reference dose. An index of more than 1.0 was considered unsafe for human health \[[@B26-ijerph-15-00919]\]. The weekly intake of elements (WI~E~) was calculated by using the following equation: where C~E~, W~MI~, and B~AW~ represent the elemental concentrations in mushrooms (mg/kg), weekly intake of mushrooms, and average body weight (45 kg) of adults in Bangladesh \[[@B27-ijerph-15-00919]\]. 2.7. Statistical Analysis {#sec2dot7-ijerph-15-00919} ------------------------- All the statistical analyses were carried out using R software \[[@B28-ijerph-15-00919]\]. The data on arsenic and heavy metals concentration in mushrooms were analysed using analysis of variance (ANOVA) using the 'aov' command. The residual sum of squares was then used for calculating Fisher's protected least significant differences. The accumulation factors were assessed as the ratio of the concentration in the mushroom divided by the concentration in the medium. Correlations between concentration in the medium and concentration in the mushroom were performed using the 'cor' function in R. Principal component analyses were performed using the 'princomp' command in R and were based on the correlation matrix. 3. Results and Discussion {#sec3-ijerph-15-00919} ========================= [Table 1](#ijerph-15-00919-t001){ref-type="table"} shows the concentration of As and other trace elements in mushrooms under investigation. It also demonstrates the percent recovery of analytical results where we noticed about 88--112% recovery using tomato leaves (NIST SRM). [Figure 1](#ijerph-15-00919-f001){ref-type="fig"} demonstrates that As levels were higher (*p* \< 0.001) at Savar (0.62 mg/kg) than at Gazipur (0.38 mg/kg) and at Savar As in *P. high-king* (0.76 mg/kg) was higher (*p* \< 0.05) than the level in *P. ostreatus* (0.49 mg/kg). Cd showed the opposite trend with Cd being higher (*p* \< 0.001) at Gazipur (0.49 mg/kg) than at Savar (0.28 mg/kg) where *P. ostreatus* (0.34 mg/kg) was higher (*p* \< 0.05) than *P. high-king* (0.22 mg/kg). Mercury was higher (*p* \< 0.001) at Savar (0.15 mg/kg) than at Gazipur (0.11 mg/kg). Concentration of Pb in *P. high-king* (0.40 mg/kg) was almost double than that of *P. ostreatus* (0.22 mg/kg) and the difference was significant (*p* \< 0.05). The other elemental differences between the species were not statistically significant. The order of the median concentrations (mg/kg) of elements in mushrooms was found to be as Zn \> Cu \> Mn \> As \> Cd \> Pb \> Cr \> Ni \> Hg \> Co with value of 50.4, 12.5, 12.2, 0.49, 0.30, 0.23, 0.22, 0.22, 0.12, and 0.01, respectively, for *P. high-king*. On the other hand in case of *P. ostreatus* the concentrations (mg/kg) were 54.8, 12.7, 12.6, 0.45, 0.38, 0.31, 0.29, 0.19, 0.12, and 0.01 for those elements that follows the order of Zn \> Mn \> Cu \> As \> Cd \> Cr \> Ni \> Pb \> Hg \> Co, respectively. The elemental concentrations of cultivated mushrooms in this study were compared to research findings on cultivated mushrooms by various researchers around the world ([Table 2](#ijerph-15-00919-t002){ref-type="table"}) though available data on cultivated mushrooms are less \[[@B29-ijerph-15-00919],[@B30-ijerph-15-00919]\] as compared to wild ones. Reported results showed that elemental concentrations in the mushrooms varied widely depending on locations and species. 3.1. Arsenic {#sec3dot1-ijerph-15-00919} ------------ Mean concentration of total As (mg/kg, dw) in *Pleurotus high-king* and *Pleurotus ostreatus* were 0.56 ± 0.25 and 0.45 ± 0.11, respectively ([Table 1](#ijerph-15-00919-t001){ref-type="table"} and [Table 2](#ijerph-15-00919-t002){ref-type="table"}). Noteworthy, however, is *Pleurotus high-king*, with As concentration reaching near 1.0 (0.91 mg/kg, dw) which could be considered low as those were below the normal level (\<1 mg/kg, dw) stated for mushrooms grown in uncontaminated media \[[@B44-ijerph-15-00919],[@B45-ijerph-15-00919]\]. Amongst different cultivated mushrooms in different countries, total As concentration of this study were lower with an exception of *Lentinula edodes* in Spain ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). However, an investigation in Bangladesh found As in mushrooms below the detection level \[[@B31-ijerph-15-00919]\]. For dry mushrooms, the UK guideline value for As is 10.0 mg/kg \[[@B46-ijerph-15-00919]\] and the As concentration in the mushroom samples both from cultivated and market basket were within the safe limit according to the guideline value ([Figure 1](#ijerph-15-00919-f001){ref-type="fig"}). Furthermore, the concentration of As in mushrooms of this study were within the limit of Australia and New Zealand Food Standard Code \[[@B47-ijerph-15-00919]\], and also below the maximum allowable concentrations (1.0 mg/kg) of As recommended by the Ministry of Health of China \[[@B48-ijerph-15-00919]\]. Inorganic As (arsenate and arsenite) is classified as human carcinogen \[[@B49-ijerph-15-00919]\]. Speciation of As (arsenite, arsenate, MMA, and DMA) in selected mushroom samples (*n* = 9) was conducted and found the percentage of inorganic As species ranged from 38 to 72%. This was consistent with study in which the researchers also reported that inorganic species were the major components in mushrooms \[[@B50-ijerph-15-00919]\] and ranged from 0.14 to 0.89 mg/kg of As, similar to this study. In a study in Bangladesh during 2002 scientists \[[@B6-ijerph-15-00919]\] found inorganic As species as the predominant species in rice straw which estimated arsenate, arsenite, and DMA as about 90%, up to 8% and up to 4% of the total, respectively. The inorganic species are the most harmful and potentially pose a health risk to the population if those are consumed on a regular basis through foods. Additionally, DMA is a probable human carcinogen \[[@B51-ijerph-15-00919],[@B52-ijerph-15-00919]\] according to the United States Environmental Protection Agency (USEPA), though less toxic than inorganic As. 3.2. Cadmium {#sec3dot2-ijerph-15-00919} ------------ The Cd concentrations of mushrooms in this study were very low (below 1.0 mg/kg, dw). However, undetectable level of Cd was found in mushroom in a study in Bangladesh \[[@B31-ijerph-15-00919]\]. Our results were in agreement with the study conducted in Mexico \[[@B39-ijerph-15-00919]\], Brazil \[[@B36-ijerph-15-00919]\], and Australia \[[@B40-ijerph-15-00919]\], with the only exception in Mexico \[[@B39-ijerph-15-00919]\] for *P. ostreatus* mushroom ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). For dry mushrooms, the UK guideline \[[@B46-ijerph-15-00919]\] value for Cd is 2.0 (mg/kg, dw) and the Cd concentrations of the mushroom samples in the current study (0.18--0.68 mg/kg) were far below the UK guideline value ([Figure 1](#ijerph-15-00919-f001){ref-type="fig"}). 3.3. Chromium {#sec3dot3-ijerph-15-00919} ------------- The Cr concentrations of mushrooms in this study were much less than those in a previously-published database ([Table 2](#ijerph-15-00919-t002){ref-type="table"}) on cultivated mushrooms except a study in Bangladesh \[[@B31-ijerph-15-00919]\]. The highest concentration of chromium (63.0 mg/kg, dw) was observed in Mexico \[[@B39-ijerph-15-00919]\] in *P. ostreatus* mushroom and the lowest (4.06 mg/kg, dw) was in *A. bisporus* in the same country ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). The same cultivated mushroom (*P. ostreatus*) showed significantly higher content (63.0 mg/kg, dw) in Mexico \[[@B39-ijerph-15-00919]\] than in Bangladesh. 3.4. Cobalt {#sec3dot4-ijerph-15-00919} ----------- The concentrations of Co in this study were similar (0.011 ± 0.004, 0.013 ± 0.005 mg/kg, dw) in both *P. high-king* and *P. ostreatus* which were about ten times lower in *A. bisporus* in Australia \[[@B40-ijerph-15-00919]\] and one thousand times lower than *P. sajor-kaju* in Turkey \[[@B41-ijerph-15-00919]\] ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). 3.5. Copper {#sec3dot5-ijerph-15-00919} ----------- The concentrations of Cu in this study (6.7--24.3 mg/kg, dw) were slightly higher than *P. sajor-kaju*, a cultivated mushroom of Turkey \[[@B41-ijerph-15-00919]\] but highest concentration was observed in Mexico \[[@B39-ijerph-15-00919]\]. However, *P. ostreatus* of another study in Bangladesh obtained higher Cu concentration \[[@B31-ijerph-15-00919]\] than the same one of this study. *A. bisporus* in Australia \[[@B40-ijerph-15-00919]\] and Hungary \[[@B42-ijerph-15-00919]\] contained about four times higher Cu than this investigation ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). 3.6. Lead {#sec3dot6-ijerph-15-00919} --------- Lead concentrations of mushrooms found in this study were 0.40 ± 0.39 and 0.22 ± 0.13 mg/kg in *P. high-king* and *P. ostreatus,* respectively, which were slightly higher than another study conducted in Bangladesh \[[@B31-ijerph-15-00919]\] with four species including *P. ostreatus*. Cultivated *P. sajor-caju* of Turkey \[[@B41-ijerph-15-00919]\] showed about fifty time higher Pb content than our results whereas *A. bisporus* of Australia \[[@B40-ijerph-15-00919]\] demonstrated about ten and twenty times higher than *P. high-king* and *P. ostreatus* in our study, respectively. However, a study carried out in Mexico \[[@B39-ijerph-15-00919]\] found lower levels of Pb in *P. ostreatus* and *A. bisporus*. For dry mushrooms, the UK guideline value for Pb is 3.0 mg/kg \[[@B46-ijerph-15-00919],[@B53-ijerph-15-00919]\]. The Pb concentration in the mushroom samples was in compliance with the UK guideline value ([Figure 1](#ijerph-15-00919-f001){ref-type="fig"}). The Pb concentrations in this study were much below the maximum allowable concentrations (2 mg/kg) of Pb recommended by the Ministry of Health of China \[[@B48-ijerph-15-00919]\]. 3.7. Manganese {#sec3dot7-ijerph-15-00919} -------------- A little variation was observed in Mn concentration in cultivated mushrooms of this study, e.g., 11.2 ± 2.17 and 12.2 ± 2.4 mg/kg in *P. high-king* and *P. ostreatus*, respectively (dry weight basis). The results of this study were much lower than the results in *P. sajor-caju* in Turkey \[[@B41-ijerph-15-00919]\], *P. ostreatus* in Mexico, *P*. *ostreatus, A. bisporus, V. volvacea, G. lucidum* in Bangladesh \[[@B31-ijerph-15-00919]\], and slightly higher than *A. bisporus* in Mexico \[[@B39-ijerph-15-00919]\] and Australia \[[@B40-ijerph-15-00919]\] ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). 3.8. Mercury {#sec3dot8-ijerph-15-00919} ------------ Mercury is highly toxic element of the environment poses health risk depending on its form, concentration, duration, and method of exposure \[[@B54-ijerph-15-00919]\]. The concentration of Hg in the substrate is of much significance for mushroom production as a high concentration in the substrate accelerates its accumulation into the cap and stipe \[[@B18-ijerph-15-00919]\]. The Hg concentrations in mushrooms in this study were 0.126 and 0.124 mg/kg (dw) in *P. high-king* and *P. ostreatus*, respectively ([Table 2](#ijerph-15-00919-t002){ref-type="table"},) which were much lower than that of Turkey \[[@B41-ijerph-15-00919]\] and Mexico \[[@B39-ijerph-15-00919]\] ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). 3.9. Nickel {#sec3dot9-ijerph-15-00919} ----------- The Ni concentrations of mushrooms in this study were much less (0.12--0.95 mg/kg) than those reported in the literature ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). High levels of Ni (31.5 mg/kg, dw) were reported in mushrooms in Turkey \[[@B41-ijerph-15-00919]\] whereas the same species, *P. ostreatus*, in this study demonstrated a maximum Ni concentration of 0.665 mg/kg. 3.10. Zinc {#sec3dot10-ijerph-15-00919} ---------- In this study the highest mean Zn level (56.9 mg/kg) was observed in *P. high-king* and the lowest (50.2 mg/kg, dw) was in *P. ostreatus*. The species *P. ostreatus* showed a higher Zn concentration (59.2 mg/kg, dw) in Ethiopia \[[@B43-ijerph-15-00919]\] and in another study in Bangladesh \[[@B31-ijerph-15-00919]\] as compared to our results. Same scenarios were also observed for *P. sajor-caju* of Ethiopia \[[@B43-ijerph-15-00919]\] and *A. bisporus* in Hungary \[[@B42-ijerph-15-00919]\] where very high differences in Zn concentration were obtained for the same species in different locations ([Table 2](#ijerph-15-00919-t002){ref-type="table"}). 3.11. Concentrations of Elements in Powdered Mushrooms for Sale {#sec3dot11-ijerph-15-00919} --------------------------------------------------------------- Ground mushroom samples (*P. ostreatus and P. high-king*) available in the market were also collected for analysis to investigate whether there is any toxicological hazard prevails for their consumption. The concentrations (range) of As, Cd, Cr, Co, Cu, Pb, Mn, Hg, Ni, and Zn in mushroom samples (*n* = 5) were 0.15--0.36, 0.14--0.44, 0.17--2.4, 0.03--0.24, 9.20--15.6, 0.15--0.61, 12.9--24.9, 0.041--0.10, 0.15--1.04, and 11.1--69.6 mg/kg, respectively, which proved that mushroom powder of the market contained lower concentrations of As, Cd, Cu, Pb, Hg, Zn, and higher concentrations of Cr, Co, Mn, and Ni compared to mushrooms directly collected from the growers in this study. 3.12. Elemental Accumulation {#sec3dot12-ijerph-15-00919} ---------------------------- The average As, Cd, Cr, Co, Cu, Pb, Mn, Hg, Ni, and Zn concentration in the substrates (rice straw) of Savar were 0.563, 2.15, 2.05, 0.789, 6.48, 2.91, 1165, 0.134, 6.09, and 62.1 mg/kg, respectively, while the concentration of those elements in Gazipur were 0.47, 0.18, 1.27, 0.58, 8.64, 2.58, 1079, 0.148, 6.57, and 58.1 mg/kg, respectively. The accumulation efficiency of elements in mushrooms from growing media was calculated using the accumulation factor, which is defined as the ratio of elemental concentration in the mushroom to that in the growth medium. Organisms are termed as metal hyper-accumulators when the accumulation factor remains above 1.0 \[[@B55-ijerph-15-00919]\] or, in another way, the species is termed as a hyper-accumulator when it contains more than 100 times higher concentrations of an element than other species growing on substrates of similar features \[[@B33-ijerph-15-00919]\]. The data in [Table 3](#ijerph-15-00919-t003){ref-type="table"} demonstrated that element accumulation factors were more than 1 for Cu and Cd at both sites, and greater than 1.0 for As and Zn at Savar, but less than 1.0 in other cases (representing exclusion). It has been observed that mushroom species are able to accumulate high amounts of some elements, e.g., As, Cd, Pb, Hg, and Se \[[@B56-ijerph-15-00919]\], particularly when mushrooms are collected from sites adjacent to heavy metal smelters \[[@B57-ijerph-15-00919]\], and landfills of sewage sludge emission areas \[[@B58-ijerph-15-00919]\]. Concentrations of heavy metals were also observed high in proximity to highways with high traffic \[[@B59-ijerph-15-00919],[@B60-ijerph-15-00919],[@B61-ijerph-15-00919]\] and emission areas \[[@B62-ijerph-15-00919]\]. The sampling site at Savar was close to highways, industries, and brick kilns, which might pose high emission and pollution \[[@B63-ijerph-15-00919]\], whereas the Gazipur site was far away from highway traffic and industries and surrounded by large trees and vegetation. Contamination of mushrooms grown at Gazipur might, therefore, be attributed to the substrate rather than the environment. 3.13. Correlations between Elements in Mushrooms {#sec3dot13-ijerph-15-00919} ------------------------------------------------ There were significant statistical correlations between elemental concentrations in mushrooms which are shown in [Table 4](#ijerph-15-00919-t004){ref-type="table"}. Significantly positive correlations were established between As and Hg (*p* ≤ 0.01), Hg and Zn (*p* ≤ 0.01), Cu and Hg (*p* ≤ 0.01), Pb and Cr (*p* ≤ 0.01), Cu and Mn (*p* ≤ 0.01), Zn and Mn (*p* ≤ 0.01), Zn and Cu (*p* ≤ 0.01), As and Cu (*p* ≤ 0.01), As and Zn (*p* ≤ 0.01), Co and Cr (*p* ≤ 0.01), As and Mn (*p* ≤ 0.01), Pb and Co (*p* ≤ 0.01), and Cd and Ni (*p* ≤ 0.01). The other correlations between elements were not significant. Significant negative correlations were found between Cd and As (*p* ≤ 0.01), and Hg and Cd (*p* ≤ 0.05). 3.14. Correlations between Elemental Concentrations in the Medium and Mushroom {#sec3dot14-ijerph-15-00919} ------------------------------------------------------------------------------ It was noted that As concentrations in the substrate for *P. high-king* in Savar were 0.660 mg/kg which was significantly higher (*p* \< 0.05) than that of Gazipur (0.473 mg/kg). There was effectively no correlation between the concentrations of element in the substrate with the corresponding element in the mushroom. This finding agrees with some other studies \[[@B33-ijerph-15-00919],[@B64-ijerph-15-00919],[@B65-ijerph-15-00919]\]. 3.15. Market Basket Study {#sec3dot15-ijerph-15-00919} ------------------------- A comparison of analyte levels in the market basket survey with the commercial products can be obtained from [Figure 1](#ijerph-15-00919-f001){ref-type="fig"}. The As level in the market basket (0.22 mg/kg) was lower (*p* \< 0.01) in the levels from the mushroom farm (0.50 mg/kg). The Hg levels were also lower (*p* \< 0.001) in the market basket than in those obtained from the mushroom farms. The Co levels were higher in the market basket (0.081 mg/kg) than in the values from the mushroom farm (0.012 mg/kg). 3.16. Daily Consumption of As and Other Elements from Mushrooms: Health Risk Index {#sec3dot16-ijerph-15-00919} ---------------------------------------------------------------------------------- Based on using weekly intake rates of mushrooms and the median concentrations (fresh weight) of elements, we have estimated the weekly total consumption of As and other elements from mushrooms for adults ([Table 5](#ijerph-15-00919-t005){ref-type="table"}). The Joint FAO/WHO Expert Committee on Food Additives \[[@B66-ijerph-15-00919]\] recommends that provisional tolerable weekly intakes (PTWI) of As, Cd, Pb, and Hg are 15, 7, 25, and 5 µg/kg/week, respectively. Therefore, the weekly tolerable intakes of As, Cd, Pb, and Hg for a person weighing 45 kg would be 0.675, 0.315, 1.125, and 0.225 mg, respectively \[[@B67-ijerph-15-00919]\]. We note that the PTWI value for As was withdrawn in 2011. The health risk index (HRI) was estimated based on standard methods \[[@B68-ijerph-15-00919]\]. Considering the average body weight (45 kg) of an adult Bangladeshi \[[@B69-ijerph-15-00919]\], fresh mushroom consumption (0.09 g weekly) and median concentration of As, Cd, Cr, Cu, Pb, Mn, Ni, and Zn in mushroom found in this study, then the estimated HRI for As, Cd, Cr, Cu, Pb, Mn, Ni, and Zn are 0.0448, 0.0097, 0.0026, 0.0089, 0.0015, 0.0025, 0.0004, and 0.0050, respectively ([Table 5](#ijerph-15-00919-t005){ref-type="table"}). All the values of HRIs based on median concentration were less than one (closer to zero), suggesting that consumption of the investigated mushrooms do not pose any health risks from As and other elements for Bangladeshi consumers. 4. Conclusions {#sec4-ijerph-15-00919} ============== The concentrations of elements varied between mushrooms and collection sites. Mushrooms alone contributed to the intake of As and other elements which were below the PTWI values due to the very low consumption of mushrooms in the diet of Bangladeshis. Moreover, the estimated daily intakes were based on the best available production and consumption data, which was meagre. Due to the low consumption rate of mushrooms, it is, therefore, important to examine other food sources, such as rice, vegetables, etc., to assess the total exposure load from main Bangladeshi foods for the pollutants. As the tolerable/permissible values of As, Cd, and Pb are available from UK guidelines, future work should focus on formulating the values for other metals, which will provide more clear information for risk assessment. We are grateful to the Crawford Fund, Australia, for the financial support provided to the first author to conduct academic research at CERAR located in the University of South Australia and GCER in The University of Newcastle. We thank the Project Director, Nirod Chandra Sarker and the scientific officer, Ruhul Amin of the Mushroom Development and Extension Institute, Savar, Bangladesh, and also Shafiqul Islam, Assistant Professor, Division of Soil Science, Bangladesh Agricultural University, Mymensingh for their cooperation during sampling. M.H.R. and M.M.R. planned the experimental design, executed the study, drafted the manuscript; R.C. conducted the statistical analysis of the data as well as English correction; R.N. provided the funding for the experiment, checked, proof read the final draft and provided full supervision. The authors declares that there is no conflicts of interest. {#ijerph-15-00919-f001} ijerph-15-00919-t001_Table 1 ###### Concentrations of elements in *Pleurotus high-king* (*n* = 40) and *Pleurotus ostreatus* (*n* = 40) mushroom samples (mg/kg, dw). Element Mushroom Species Mean ± SD Median Range NIST SRM 1573a (Tomato Leaves) ---------------- ------------------ --------------- -------------- -------------- -------------------------------- --------------- ------ As *P. high-king* 0.56 ± 0.25 0.489 0.18--0.91 0.112 ± 0.004 0.122 ± 0.003 109 *P. ostreatus* 0.45 ± 0.11 0.45 0.263--0.62 Cd *P. high-king* 0.35 ± 0.17 0.301 0.18--0.68 1.52 ± 0.04 1.46 ± 0.060 96.1 *P. ostreatus* 0.41 ± 0.10 0.38 0.24--0.599 Cr *P. high-king* 0.27 ± 0.14 0.223 0.096--0.533 1.99 ± 0.06 1.76 ± 0.050 88.4 *P. ostreatus* 0.3 ± 0.16 0.313 0.078--0.697 Co *P. high-king* 0.011 ± 0.004 0.011 0.005--0.022 0.57 ± 0.02 0.61 ± 0.040 107 *P. ostreatus* 0.013 ± 0.005 0.014 0.005--0.024 Cu *P. high-king* 14.2 ± 4.9 12.5 7.30--24.3 4.7 ± 0.14 5.3 ± 0.19 113 *P. ostreatus* 13.2 ± 3.7 12.6 6.70--18.8 Pb *P. high-king* 0.40 ± 0.39 0.23 0.08--1.37 *P. ostreatus* 0.22 ± 0.13 0.194 0.079--0.72 Mn *P. high-king* 11.2 ± 2.17 12.2 6.45--13.4 246 ± 8 232 ± 12 94.3 *P. ostreatus* 12.2 ± 2.4 12.7 5.1--15.8 Hg *P. high-king* 0.126 ± 0.03 0.12 0.11--0.144 0.034 ± 0.004 0.032 ± 0.006 94.1 *P. ostreatus* 0.124 ± 0.03 0.121 0.094--0.154 Ni *P. high-king* 0.263 ± 0.18 0.22 0.12--0.95 1.59 ± 0.07 1.50 ± 0.08 94.3 *P. ostreatus* 0.3 ± 0.12 0.292 0.136--0.665 Zn *P. high-king* 56.9 ± 19.5 50.4 26.3--85.8 30.9 ± 0.7 27.6 ± 0.6 89.3 *P. ostreatus* 50.2 ± 11.3 54.8 27.5--63.2 *P. high-king: Pleurotus high-king, P. ostreatus: Pleurotus ostreatus*. ijerph-15-00919-t002_Table 2 ###### Total As concentration (mg/kg, dw) in cultivated mushrooms in this study and in literature (adapted and modified \[[@B30-ijerph-15-00919]\]). Element Study Location Species Concentration Reference ------------------------- ---------------------------------------- ---------------------------------------- -------------------------- --------------------------------------------------- ------------ As Bangladesh *P. high-king* 0.56 ± 0.25 0.18--0.91 This study *P. ostreatus* 0.45 ± 0.11 0.263--0.62 Bangladesh *P. ostreatus* BDL \[[@B31-ijerph-15-00919]\] *A. bisporus* *V. volvacea* *G. lucidum* Spain *L. edodes* 1.393 \[[@B32-ijerph-15-00919],[@B33-ijerph-15-00919]\] *A. bisporus* 0.185 *P. ostreatus* 0.335 Canada *A. bisporus* 0.14 ± 0.04 \[[@B34-ijerph-15-00919]\] Poland *A. arvensis* BDL \[[@B35-ijerph-15-00919]\] *A. bisporus* 0.08--0.71 *A. Bisporus* 0.15--1.4 Brazil *P. ostreatus* 0.056 ± 0.004 \[[@B36-ijerph-15-00919]\] *P. florida* 0.073 ± 0.018 *P. eryngui* 0.009 ± 0.003 *P. salmoneostramineus* 0.043 ± 0.004 *A.* sp. 0.125 ± 0.014 *A. bisporus* 0.097 ± 0.024 *L. edodes* 0.210 ± 0.009 Ghana *P. ostreatus* 0.04 \[[@B37-ijerph-15-00919]\] *T. clypeatus* 0.1 India *A. bisporus* 0.64 ± 0.16 \[[@B38-ijerph-15-00919]\] Cd Bangladesh *P. high-king* 0.35 ± 0.17 0.18--0.68 This study *P. ostreatus* 0.41 ± 0.10 0.24--0.60 Bangladesh *P. ostreatus* BDL \[[@B31-ijerph-15-00919]\] *A. bisporus* *V. volvacea* *G. lucidum* Mexico *P. ostreatus* 5.39 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 0.54 Brazil *P. ostreatus* 0.074 ± 0.002 \[[@B36-ijerph-15-00919]\] *P. florida* 0.220 ± 0.013 *P. eryngui* 0.011 ± 0.003 *P. salmoneostramineus* 0.229 ± 0.004 *A.* sp. \<LD *A. bisporus* \<LD *L. edodes* 0.190 ± 0.011 Australia *A. bisporus* 0.18 ± 0.02 \[[@B40-ijerph-15-00919]\] Cr Bangladesh *P. high-king* 0.27 ± 0.14 0.096--0.53 This study *P. ostreatus* 0.30 ± 0.16 0.078--0.70 Bangladesh *P. ostreatus* 0.21 ± 0.00--0.30 ± 0.01 \[[@B31-ijerph-15-00919]\] *A. bisporus* 0.23 ± 0.01--0.30 ± 0.01 *V. volvacea* 0.24 ± 0.01 *G. lucidum* 0.21 ± 0.01 Mexico *P. ostreatus* 63.0 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 4.06 Turkey *P. sajor-caju* 8.50 \[[@B41-ijerph-15-00919]\] Co Bangladesh *P. high-king* 0.011 ± 0.004 0.005--0.022 This study *P. ostreatus* 0.013 ± 0.005 0.005--0.024 Turkey *P. sajor-caju* 12.5 \[[@B41-ijerph-15-00919]\] Australia *A. bisporus* 0.126 ± 0.001 \[[@B40-ijerph-15-00919]\] Cu Bangladesh *P. high-king* 14.2 ± 4.9 7.3--24.3 This study *P. ostreatus* 13.2 ± 3.7 6.7--18.8 Bangladesh *P. ostreatus* 39.2 ± 0.88--102.1 ± 2.6 \[[@B31-ijerph-15-00919]\] *A. bisporus* 54.6 ± 0.86--163.4 ± 3.9 *V. volvacea* 101.8 ± 2.3 *G. lucidum* 72.5 ± 1.22 Turkey *P. sajor-caju* 10.5 \[[@B41-ijerph-15-00919]\] Mexico *P. ostreatus* 732 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 352 Australia *A. bisporus* 53.4 ± 0.251 \[[@B40-ijerph-15-00919]\] Hungary *A. bisporus* 58 ± 2--65 ± 1 \[[@B42-ijerph-15-00919]\] Pb Bangladesh *P. high-king* 0.40 ± 0.39 0.08--1.37 This study *P. ostreatus* 0.22 ± 0.13 0.079--0.72 Bangladesh *P. ostreatus* 0.14 ± 0.02--0.59 ± 0.03 \[[@B31-ijerph-15-00919]\] *A. bisporus* 0.15 ± 0.01--0.22 ± 0.02 *V. volvacea* 0.25 ± 0.02 *G. lucidum* 0.13 ± 0.01 Turkey *P. sajor-caju* 27.5 \[[@B41-ijerph-15-00919]\] Mexico *P. ostreatus* 0.91 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 0.41 Australia *A bisporus* 3.9 ± 0.49 \[[@B40-ijerph-15-00919]\] Mn Bangladesh *P. high-king* 11.2 ± 2.17 6.45--13.4 This study *P. ostreatus* 12.2 ± 2.4 5.1--15.8 Bangladesh *P. ostreatus* 52.9 ± 1.04--104.5 ± 1.8 \[[@B31-ijerph-15-00919]\] *A. bisporus* 56.2 ± 1.34--91.1 ± 1.38 *V. volvacea* 78.5 ± 0.97 *G. lucidum* 64.0 ± 0.92 Turkey *P. sajor-caju* 17.5 \[[@B41-ijerph-15-00919]\] Mexico *P. ostreatus* 18.3 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 9.42 Australia *A. bisporus* 10.6 ± 0.01 \[[@B40-ijerph-15-00919]\] Hg Bangladesh *P. high-king* 0.126 ± 0.03 0.11--0.14 This study *P. ostreatus* 0.124 ± 0.03 0.094--0.15 Poland *P. ostreatus* 0.10 ± 0.01 (cap), 0.02 ± 0.01 (stipe) \[[@B18-ijerph-15-00919]\] *A. bisporus* 0.08 ± 0.01 (cap), 0.05 ± 0.01 (stipe) *H. erinaceus* 0.07 ± 0.02 (cap), 0.03 ± 0.01 (stipe) Ni Bangladesh *P. high-king* 0.263 ± 0.18 0.12--0.95 This study *P. ostreatus* 0.3 ± 0.12 0.136--0.665 Turkey *P. sajor-caju* 17.5 \[[@B41-ijerph-15-00919]\] Mexico *P. ostreatus* 31.5 \[[@B39-ijerph-15-00919]\] *A. bisporus (caps)* 9.02 Zn Bangladesh *P. high-king* 56.9 ± 19.5 26.3--85.8 This study *P. ostreatus* 50.2 ± 11.3 27.5--63.2 Bangladesh *P. ostreatus* 30.1 ± 0.19--75.5 ± 0.54 \[[@B31-ijerph-15-00919]\] *A. bisporus* 36.3 ± 0.23--47.6 ± 0.46 *V. volvacea* 36.5 ± 0.43 *G. lucidum* 52.2 ± 0.47 Turkey *P. sajor-caju* 110 \[[@B41-ijerph-15-00919]\] Ethiopia *P. ostreatus* 59.2 ± 0.02 \[[@B43-ijerph-15-00919]\] *P. sajor-caju* 59.2 ± 0.03 Australia *A. bisporus* 43.8 ± 0.20 \[[@B40-ijerph-15-00919]\] Hungary *A. bisporus* 60 ± 0--62 ± 0 \[[@B42-ijerph-15-00919]\] BDL: below detection limit (level 0.01 mg/kg 1 dw), LD: limit of detection, P: PleurotusL: Lentinula, A: Agaricus, T: Termitomyces, V: Volvariella, G: Ganoderma, H: *Hericium.* ijerph-15-00919-t003_Table 3 ###### Accumulation factors of elements from substrate to mushrooms. Location Mushroom As Cd Cr Co Cu Pb Mn Hg Ni Zn ---------------- ---------------- ------ ------- ------- ------- ------- ------- ------- ------- ------- ------ Savar *P. high-king* 1.35 1.30 0.132 0.009 3.28 0.073 0.008 1.63 0.058 1.18 *P. ostreatus* 1.54 3.12 0.249 0.038 2.03 0.090 0.016 0.82 0.047 1.01 Gazipur *P. high-king* 0.79 2.11 0.293 0.032 1.61 0.271 0.012 0.79 0.042 0.76 *P. ostreatus* 0.98 2.19 0.378 0.032 1.42 0.121 0.013 0.58 0.05 0.92 Mean 1.07 2.19 0.263 0.028 2.09 0.139 0.012 0.956 0.049 0.97 SD 0.51 1.19 0.189 0.020 1.13 0.169 0.006 0.754 0.020 0.43 ijerph-15-00919-t004_Table 4 ###### Correlation coefficient matrix of elements for mushroom samples. As Cd Cr Co Cu Pb Mn Hg Ni Zn ---- ------------- ------------ ------------ ------------ ------------ ----------- ------------ ------------ ---------- ------- As 1.000 Cd −0.508 \*\* 1.000 Cr −0.016 ns 0.343 ns 1.000 Co −0.119 ns 0.132 ns 0.504 \*\* 1.000 Cu 0.778 \*\* −0.056 ns 0.170 ns 0.130 ns 1.000 Pb −0.312 ns 0.351 ns 0.679 \*\* 0.522 \*\* −0.288 ns 1.000 Mn 0.552 \*\* 0.177 ns 0.312 ns 0.234 ns 0.645 \*\* −0.004 ns 1.000 Hg 0.891 \*\* −0.470 \* −0.158 ns −0.259 ns 0.666 \*\* −0.383 ns 0.352 ns 1.000 Ni −0.055 ns 0.522 \*\* 0.215 ns 0.259 ns 0.117 ns 0.290 ns 0.154 ns 0.027 ns 1.000 Zn 0.956 \*\* −0.322 ns 0.121 ns −0.044 ns 0.806 \*\* −0.168 ns 0.699 \*\* 0.826 \*\* 0.075 ns 1.000 \*\* and \* denote *p* ≤ 0.01 and *p* ≤ 0.05, respectively; ns = non-significant. ijerph-15-00919-t005_Table 5 ###### Health risks assessment for As and other elements from intake of mushroom for adults. Element Weekly Intake Rate (Fresh wt, g) Median Concentration wt (mg/kg) Weekly Intake (mg) of Elements Weekly Intake of Element mg/kg/bw Oral Reference Dose mg/kg/day Oral Reference Dose mg/kg/week HRI PTWI µg/kg PTWI mg/45 kg Maximum Consumption per Week \* (g) --------- ---------------------------------- --------------------------------- -------------------------------- ----------------------------------- ------------------------------- -------------------------------- -------- ------------ --------------- ------------------------------------- As 0.09 0.047 0.0042 0.00009 0.0003 0.0021 0.0448 15 0.675 14.4 Cd 0.034 0.0030 0.00007 0.001 0.007 0.0097 7 0.315 9.3 Cr 0.027 0.0024 0.00005 0.003 0.021 0.0026 Co 0.001 0.0001 0.00000 Cu 1.250 0.1125 0.00250 0.04 0.28 0.0089 Pb 0.021 0.0019 0.00004 0.004 0.028 0.0015 25 1.125 53.6 Mn 1.240 0.1116 0.00248 0.14 0.98 0.0025 Hg 0.012 0.0011 0.00002 5 0.225 18.8 Ni 0.026 0.0023 0.00005 0.02 0.14 0.0004 Zn 5.260 0.4734 0.01052 0.3 2.1 0.0050 \* Consumption rate of mushroom to reach provisional tolerable intake, g/week.

Yawning is characterized by a powerful gaping of the jaw with inspiration, followed by a brief period of peak muscle contraction and a passive closure of the jaw with shorter expiration ([@zoz060-B3]). While yawning appears to be a stereotyped action pattern across vertebrates ([@zoz060-B27]), recent research has identified distinct variations in the expression of this behavior. For example, some primate species display different yawn types that vary in intensity and morphology, which may correspond with transitions in daily repertoires, activity patterns, and internal states ([@zoz060-B26]; [@zoz060-B32]; [@zoz060-B21]). This work supports research suggesting yawns might be multifunctional across different contexts ([@zoz060-B2]; [@zoz060-B28]; [@zoz060-B9]). Studying differences in the expression of yawning across species could also help elucidate the neurophysiological significance of this highly conserved response. Based on hypotheses stating that the motor action pattern of yawns serves to enhance intracranial circulation ([@zoz060-B33]) and regulate brain temperature ([@zoz060-B10], [@zoz060-B11]; [@zoz060-B29]), the duration or magnitude of this response should correspond to the degree of neurophysiological change. In particular, the brain cooling hypothesis states that the gaping of the jaw and deep inhalation of air during yawning functions to promote thermal homeostasis by altering the rate and temperature of arterial blood traveling to the skull ([@zoz060-B12]). Support for this hypothesis comes from measuring changes in brain temperature before and after yawns ([@zoz060-B31]; [@zoz060-B30]) and naturalistic and experimental studies assessing the relationship between yawning and ambient temperature (e.g., [@zoz060-B16]; [@zoz060-B23]; [@zoz060-B7]). Although the neural structures necessary for yawning appear to be located in the brain stem ([@zoz060-B18]), recent comparative research shows that yawns produce large-scale neurophysiologic effects as evidenced by sizable cooling at the surface of the skull ([@zoz060-B6]; [@zoz060-B14]). Therefore, according to this hypothesis, animals with larger and more complex brains should display longer and more powerful yawns to achieve these outcomes. In the first study to test this prediction, [@zoz060-B13] collected yawn duration data across a diverse sample of mammals and linked these with published data on both brain weight and cortical neuron number. The results of this work revealed average yawn duration to be highly variable across 24 mammalian taxa (range: 0.8--6.5 s). Moreover, consistent with neurophysiologic hypotheses of yawning, interspecific differences in yawn duration were robustly correlated with both brain weight (correlation coefficient = 0.911) and cortical neuron number (correlation coefficient = 0.951) across species. Variation in yawn duration was not tied to particular features of body/jaw size, but was correlated positively with encephalization quotient across species. As a direct follow-up to this work, [@zoz060-B15] investigated whether the relationship between yawn duration and brain size could be identified within a single family of mammal species. Examining this link at a restricted taxonomic scale allowed for the determination of whether variability in yawn duration was related specifically to brain size or to other features pertaining to phylogenetic history and morphological diversity between species. In a sample of 58 animals across 7 wild cat species in the family *Felidae*, a similarly robust association (correlation coefficient = 0.937) was observed between yawn duration and brain weight ([@zoz060-B14]). This study also showed that body size was not a significant predictor of yawn duration. These combined findings support the view that the motor action pattern of yawning serves an important function tied to brain size. Here, we extend upon this line of research by testing whether intraspecific variation in brain weight predicts differences in yawn duration within a single species. Domesticated dogs *Canis lupus familiaris* were chosen due to large differences in brain size across breeds and the availability of open access yawning data on the Internet. A report by [@zoz060-B4] that documented the average adult brain and body weights from over 2 dozen dog breeds was used as a basis for the inclusion of breeds within this study. This report also included a scaling function for estimating brain weight based on the average body weight for a given breed, which was recently validated and applied to a much larger sample of domesticated dogs ([@zoz060-B19]). The same methods used in Gallup et al. ([@zoz060-B13], [@zoz060-B14]) were applied to acquire representative yawn durations from each breed using public videos available online. We hypothesized that dog breeds with larger brains would display longer yawns, and that this relationship would remain significant after controlling for body size. Materials and Methods ===================== Using the list of 26 dog breeds compared in [@zoz060-B4], a researcher was instructed to find up to 12 adult dogs yawning from each breed by searching videos posted on YouTube and related websites. For each video identified, the researcher noted the time(s) at which the yawn(s) on the clip occurred and the duration of all yawning events using the operational definition provided by [@zoz060-B3]. Durations were recorded to the nearest 0.01 s using the stopwatch feature of a smartphone. Data collection occurred during Fall 2016, and yielded a total of 299 yawns from 213 dogs across the 26 breeds. In Fall 2017, an independent rater was provided with the complete list of uniform resource locators (URLs) identified from the original search and instructed to score the duration of each yawn across the sample. In doing so, 4 videos were no longer available and 8 yawns were excluded because they were noted as either being too difficult to distinguish or there was no clear way to determine their beginning or end. In addition, links for 3 dogs were excluded because they were identified by both raters as juveniles. The inter-rater reliability for yawn duration was high for the remaining sample: intraclass correlation = 0.87. The average yawn duration for each breed was then calculated by summing all the respective durations from members of that group and then dividing by the number of yawners. To avoid pseudo-replication in cases where the same individual displayed multiple yawns in a given video, the average duration for this animal was used as a single data point when generating the overall average for that breed. The maximum yawn duration was also recorded from each breed. Following the same criterion set within Gallup et al. ([@zoz060-B13], [@zoz060-B14]), analyses were restricted to breeds with yawns from at least 3 individuals. The final sample included a total of 272 yawns from 198 dogs across 23 breeds. Individual yawning data from the following breeds were included in the sample: Beagle (11), Boston Terrier (12), Boxer (12), Bulldog (11), Chihuahua (9), Cocker Spaniel (3), Collie (5), Dachshund (12), Doberman Pinscher (9), German Shepherd (11), Golden Retriever (12), Great Dane (12), Labrador Retriever (12), Miniature Poodle (8), Miniature Schnauzer (6), Old English Sheepdog (3), Pekingese (6), Pug (11), Standard Poodle (4), Standard Schnauzer (4), Toy Fox Terrier (3), Toy Poodle (12), and Weimaraner (10). The Supplemental Material includes URLs to all videos used in the study. For the average brain and body weights across breeds, [@zoz060-B4] provided data based on an average sample of 31.69 dogs per breed (range: *N* = 5--110). This report also included a scaling function for estimating brain weight based on the average body weight for a given breed. [@zoz060-B19] recently validated this scaling function and applied it to a much larger sample of domesticated dogs (*N* = 7,397 across 74 breeds). Therefore, where applicable in this study (19/23 breeds), these recent and more representative body weight and estimated brain weight data were used (see [Table 1](#zoz060-T1){ref-type="table"}). ###### Descriptive statistics for yawn duration and brain and body weights Breed (alphabetical order) Dog *N* Mean ± SD yawn duration (s) Maximum yawn duration (s) Mean brain weight (g) Mean body weight (kg) --------------------------------------------------------- --------- ----------------------------- --------------------------- ----------------------- ----------------------- Beagle[^a^](#tblfn2){ref-type="table-fn"} 11 2.35 ± 0.49 3.28 79.66 12.79 Boston Terrier[^a^](#tblfn2){ref-type="table-fn"} 12 1.67 ± 0.28 2.09 73.22 9.41 Boxer[^a^](#tblfn2){ref-type="table-fn"} 12 2.20 ± 0.39 3.00 98.95 28.16 Bulldog[^a^](#tblfn2){ref-type="table-fn"} 11 1.84 ± 0.32 2.45 94.78 24.08 Chihuahua^**a**^ 9 1.58 ± 0.44 2.17 55.10 3.35 Cocker Spaniel[^a^](#tblfn2){ref-type="table-fn"} 3 1.88 ± 0.34 2.25 77.76 11.72 Collie[^a^](#tblfn2){ref-type="table-fn"} 5 2.03 ± 0.43 2.56 98.91 28.12 Dachshund[^a^](#tblfn2){ref-type="table-fn"} 12 1.78 ± 0.30 2.24 68.03 7.20 Doberman Pinscher[^a^](#tblfn2){ref-type="table-fn"} 9 2.14 ± 0.74 3.27 104.01 33.77 German Shepherd[^a^](#tblfn2){ref-type="table-fn"} 11 2.59 ± 0.73 3.97 104.63 34.50 Golden Retriever[^a^](#tblfn2){ref-type="table-fn"} 12 2.25 ± 0.65 3.72 101.63 31.04 Great Dane[^a^](#tblfn2){ref-type="table-fn"} 12 2.31 ± 0.43 2.94 120.54 57.75 Labrador Retriever[^a^](#tblfn2){ref-type="table-fn"} 12 2.21 ± 0.46 3.00 102.12 31.59 Miniature Poodle[^a^](#tblfn2){ref-type="table-fn"} 8 1.70 ± 0.46 2.72 67.54 7.02 Miniature Schnauzer[^a^](#tblfn2){ref-type="table-fn"} 6 1.88 ± 0.33 2.48 69.76 7.89 Old English Sheepdog[^b^](#tblfn3){ref-type="table-fn"} 3 2.47 ± 0.86 3.40 104.40 38.60 Pekingese[^b^](#tblfn3){ref-type="table-fn"} 6 1.43 ± 0.39 2.11 53.40 4.90 Pug[^a^](#tblfn2){ref-type="table-fn"} 11 1.72 ± 0.30 2.20 71.85 8.79 Standard Poodle[^a^](#tblfn2){ref-type="table-fn"} 4 2.29 ± 1.09 3.90 92.53 22.06 Standard Schnauzer[^a^](#tblfn2){ref-type="table-fn"} 4 2.20 ± 0.45 2.66 84.09 15.58 Toy Fox Terrier[^b^](#tblfn3){ref-type="table-fn"} 3 1.52 ± 0.72 2.34 52.30 3.40 Toy Poodle[^b^](#tblfn3){ref-type="table-fn"} 12 1.76 ± 0.31 2.40 59.10 3.20 Weimaraner[^a^](#tblfn2){ref-type="table-fn"} 10 2.83 ± 0.64 3.64 101.33 30.71 *Note:* Brain and body weight data obtained from [@zoz060-B19] and [@zoz060-B4]. Based on output from Kolmogorov--Smirnov normality tests, Pearson correlations were used to assess the relationships between measures of yawn duration and brain and body weights across the 23 breeds. Separate partial correlations were then run controlling for body and brain weight, respectively. Despite our a priori hypotheses, all analyses included more conservative 2-tailed tests with the alpha set at 0.05. Moreover, Benjamini and Hochberg corrections were applied for multiple comparisons, and thus adjusted *P*-values are provided. Results ======= The average yawn duration for all dogs in the sample was 2.04 ± 0.59 s, with the breed average ranging from 1.43--2.83 s ([Table 1](#zoz060-T1){ref-type="table"}). Consistent with previous research, there were strong positive correlations between measures of yawn duration and brain weight across breeds (average duration: *r*~23~ = 0.819, *P \< *0.001; maximum duration: *r*~23~ = 0.703, *P \< *0.001; [Figure 1](#zoz060-F1){ref-type="fig"}). Breed brain and body weights were highly correlated across the sample (*r*~23~ = 0.957, *P \< *0.001), and thus yawn duration was also positively correlated with body weight (average duration: *r*~23~ = 0.749, *P \< *0.001; maximum duration: *r*~23~ = 0.649, *P *=* *0.001). Partial correlations showed that average yawn duration remained significantly correlated with brain weight after controlling for body weight (average duration: *r*~20~ = 0.535, *P *=* *0.016; maximum duration: *r*~20~ = 0.374, *P *=* *0.111). However, yawn duration was not correlated with body weight when controlling for brain weight (average duration: *r*~20~ = −0.222, *P *=* *0.361; maximum duration: *r*~20~ = −0.125, *P *=* *0.579). {#zoz060-F1} Discussion ========== Recent research has shown that mammalian brain weight is a robust predictor of yawn duration across different taxonomic scales ([@zoz060-B13], [@zoz060-B14]), supporting past research suggesting that this ubiquitous motor action pattern serves an important neurological function. This study reveals that yawn duration is also tied to intraspecific variation in brain size among domesticated dogs. In particular, we show a wide range in the duration of this action pattern across breeds, with larger brained dogs tending to have longer yawns. Notably, over two-thirds of the variability in the average yawn duration across breeds can be explained by brain weight differences between these populations. Partial correlations show that this association is not due to differences in body size across breeds, but rather differences in yawning appear to be specifically associated with variation in the brain size of these animals. It remains unknown exactly which neurological features are driving this association, but past research suggests that cortical neuron number is the strongest predictor of interspecies differences in yawn duration ([@zoz060-B13]). Although research on neuron numbers in dogs is currently limited, in a comparison of 2 animals [@zoz060-B20] provide some evidence that larger breeds have more neurons (i.e., Golden Retriever: 627 million; unspecified smaller breed: 429 million). While we show here that intraspecific variation in brain size appears to be a robust predictor of yawn duration in domesticated dogs, it is unknown how these measures would correlate within a given breed or in a different wild type species. Due to artificial selection, the variation in brain and body weights across dog breeds exceeds that of any other species, making this study a distinctive and atypical within-species comparison. Based on the findings of recent work uncovering breed differences in neuroanatomical organization ([@zoz060-B17]), one could argue that assessing neurological differences across dog breeds becomes similar to interspecies comparisons. Thus, until further research is conducted, we are hesitant to conclude that yawn duration would be predictive of intraspecific variation in brain size in other animals. One intriguing species to investigate would be humans, since there is sizable variation in the overall brain volume among healthy adults (1.25--1.88 dm^3^; [@zoz060-B22]). Research on humans or another species could also provide a more precise analysis of this relationship, by linking brain volumes acquired through magnetic resonance imaging to electromyography from the masseter and submental muscles to measure yawn duration and intensity (see [@zoz060-B8]). It is important to acknowledge limitations to this study with regard to the methods of data collection. Although visual media can provide a powerful tool for investigating and developing insights into animal behavior ([@zoz060-B25]), and research utilizing YouTube for this purpose has increased within the past decade (see [@zoz060-B24]), one potential issue with human posted videos of this nature is nonrandom sampling ([@zoz060-B5]). While we do not view this as a major concern given the strength of the correlations both for average and maximum yawn duration, naturalistic observations could be conducted to confirm the true variability in spontaneous yawn duration for a given breed. The inability to determine the contextual triggers of yawning from online videos also presents a limitation. While there is consistency in that most videos occurred within the context of close dog--human interaction, other potentially important factors known to alter the morphology of yawning in other species (i.e., [@zoz060-B21]; [@zoz060-B26]; [@zoz060-B32]) could not be evaluated. To date, however, no study has reported differential yawn types for dogs. Overall, this study replicates and extends upon previous results showing a robust link between yawn duration and brain weight, providing further support for predictions derived from the brain cooling hypothesis. However, it is important to acknowledge that the current findings do not necessarily exclude other neurophysiological hypotheses of yawning. We propose that, in addition to examining the connection between yawn duration and brain size in other classes of vertebrates (e.g., avian species), future research could investigate how differences in the duration of yawning events correlate with immediate changes in neurophysiological measures and cognitive processing. Work of this nature could help elucidate the evolutionary significance of this ubiquitous response. Supplementary Material ====================== ###### Click here for additional data file. We are thankful to Jeff Clemishaw for helping to score video data and the College of Arts and Sciences at SUNY Polytechnic Institute for providing support for this research. We are also grateful to Anne Clark for carefully reviewing some of the data presented in [Table 1](#zoz060-T1){ref-type="table"} prior to publication. Authors' contributions ====================== A.C.G. conceived of the study and analyzed the data; L.M. collected the data; and A.C.G., L.M., and J.J.M.M. wrote the article. Supplementary Material ====================== [Supplementary material](#sup1){ref-type="supplementary-material"} can be found at <https://academic.oup.com/cz>.

Introduction {#S1} ============ Schizophrenia is a mental disorder affecting approximately 1% of the population worldwide ([@B44]). The disorder typically manifests in puberty or adolescence and is characterized by so-called positive symptoms such as delusions and/or hallucinations ([@B27]), but patients also exhibit negative symptoms and cognitive deficits. Studies using structural magnetic resonance imaging (sMRI) provide evidence that patients with schizophrenia display subtle volumetric brain aberrations at the time of diagnosis as compared to healthy controls ([@B6]; [@B13]). Moreover, the brain of chronic, medicated patients appears to undergo progressive, structural changes over the course of the illness, with ventricular volume increases, cortical thinning, and basal ganglia enlargement among the most consistent findings ([@B5]; [@B41]; [@B23]; [@B49]; [@B13]). Antipsychotic drugs (APD) are the gold standard for treatment of positive symptoms ([@B27]), but since illness and treatment go hand in hand, separating the effects of medication and disease on brain structure is difficult ([@B16]). In 1976 it was discovered that antipsychotics exert their function by antagonizing the dopamine D~2~ receptors (D~2~R) in striatum, and that drug efficacy is directly proportional to the affinity for the receptor ([@B10]). This led to the dopamine hypothesis of schizophrenia, which suggests that a hyperactive striatal dopamine-system leads to 'aberrant salience', meaning that wrongful interpretations of harmless stimuli can eventually lead to core psychotic symptoms such as hallucinations and delusions ([@B31]). Further studies of striatum have found increased presynaptic dopamine synthesis capacity and -release compared to controls, as well as higher dopamine concentrations in the synaptic cleft ([@B25], [@B26]; [@B7]; [@B44]). All currently marketed antipsychotics antagonize the D~2~R, thereby blocking the down-stream signaling in the post-synaptic neuron ([@B19]; [@B34]; [@B1]). However, most antipsychotics are characterized by broad receptor profiles, and bind to e.g., serotonin 2A-, histaminergic- and cholinergic receptor systems ([@B34]). This complex pharmacology has further limited the investigations of causal mechanisms linking antipsychotic treatment to structural brain changes. Nevertheless, longitudinal studies on antipsychotic-naïve patients as well as meta-analyses studies have reported associations between antipsychotic exposure and volumetric increase in basal ganglia ([@B18]; [@B14]; [@B29]; [@B28]; [@B12]). Studies on rodents have replicated the basal ganglia volume increase in response to antipsychotic treatment ([@B52]), and investigations in dopamine D~2~ or D~3~ receptor knock-out- and wild-type mice provide evidence that this increase is likely to be mediated through D~2~-like receptors ([@B21], [@B22]). In humans, dopamine D~2~-like receptor availability and blockade following antipsychotic treatment can be investigated with single-photon emission computed tomography (SPECT) examinations ([@B44]). The association between antipsychotic treatment, dopamine D~2/3~ receptor occupancy, and basal ganglia enlargement has, however, yet to be established in a longitudinal study of antipsychotic-naïve patients with schizophrenia. To address this gap in our knowledge, we completed a prospective study, wherein we examined a cohort of first-episode, antipsychotic-naïve schizophrenia patients, before and after 6 weeks of treatment with amisulpride, a relatively selective dopamine D~2/3~ receptor antagonist. Baseline- and follow-up examinations included sMRI, SPECT, and Positive and Negative Syndrome Scale (PANSS) examinations. We hypothesized that selective blockade of dopamine D~2/3~R would lead to a dose-dependent striatal volume increase. Further, we explored correlations between symptom severity decrease, striatal volume increase, dose and receptor occupancy. Materials and Methods {#S2} ===================== Participants {#S2.SS1} ------------ We included participants between the ages of 18--45 years from 2008 to 2014. Patients with schizophrenia were first-episode, antipsychotic-naïve, and were recruited from hospitals and psychiatric out-patient clinics in the capital region of Denmark, as a part of the PECANS I (Pan European Collaboration Antipsychotic-naïve Studies, PECANS) cohort. All patients met the International Classification of Diseases (ICD-10) criteria for schizophrenia (F20) verified by the structured diagnostic interview SCAN (Schedule of Clinical Assessment in Neuropsychiatry, version 2.1). Exclusion criteria included previous exposure to antipsychotic medication, methylphenidate, or use of antidepressants less than 1 month prior to baseline examinations. Healthy controls were recruited through advertisement, and matched to patients on age, gender and parental socioeconomic status. Exclusion criteria for the healthy controls were identical to the criteria for patients, but also comprised any former or current psychiatric illnesses, psychiatric diagnoses within first-degree relatives and/or any drug-abuse (classified by ICD-10). For all participants, previous or current medical history of serious head trauma, neurological diseases, developmental disorders or current drug dependency (by ICD-10 classification), and current pregnancy were exclusion criteria. All participants were screened for drug-use with urine samples (Rapid Response, Jepsen HealthCare) prior to SPECT scan. Included participants are a subsample of [@B56], [@B55] from the PECANS I cohort. Wulff and colleagues also reported on binding potentials in their sample, although a different method of binding potential extraction was used. Subcortical volumes have not yet been investigated in this subgroup. Medication {#S2.SS2} ---------- The atypical APD, amisulpride, was chosen as a tool compound because of its relative selectivity toward dopamine D~2/3~ receptors ([@B43]). Amisulpride treatment was initiated after completion of baseline examinations, and dosage was slowly increased and adjusted to the individual patient, according to clinical judgment and patients' reports of adverse effects. Pharmaceutical treatment against adverse effects was not allowed. Follow-up examinations were conducted after six weeks, and treatment dose in mg was recorded. To ensure a steady concentration at examinations, dosage was kept stable in the week prior to follow-up. Compliance was continuously ensured through dialogue with the patient, and measurement of serum-amisulpride (S-amisulpride) levels at follow-up. Benzodiazepines were allowed on an "as-needed basis" to secure sleep and reduce anxiety but were not allowed 12 h prior to SPECT examinations. Healthy control subjects were not treated. Symptom Severity {#S2.SS3} ---------------- Symptom severity was assessed with PANSS ([@B32]) within the same week as MRI and SPECT scan examinations. PANSS total score as well as sub-scores (positive-, negative-, and general sub-scores) was assessed at baseline and at follow-up. To ensure consistency in PANSS ratings between clinicians, ratings were regularly evaluated using systematic video recordings of the interviews. Duration of untreated illness was assessed from the patient history of worsening in functions due to symptoms. Healthy controls did not undergo PANSS examinations. Magnetic Resonance Imaging {#S2.SS4} -------------------------- T1-weighted scans of the whole head (sagittal 3D sequence, TR = 10 ms, TE = 4.6 ms, FA = 8°, voxel size = 0.79 mm × 0.79 mm × 0.80 mm) were acquired with an 8-channel SENSE head coil on a 3T Philips Achieva scanner (Philips Healthcare, Best, Netherlands) at baseline and after 6 weeks. MRI scans were acquired within the same week as SPECT and PANSS. Subcortical segmentation and volume extraction were performed with tools from the FSL, FMRIB software library v5.0.10 ([@B40]). In this study we focused on striatum as our region of interest, estimated as a sum of volumes from the bilateral subregions of caudate nucleus, putamen and nucleus accumbens ([Figure 1](#F1){ref-type="fig"}). Anatomically, striatum is also referred to as a part of basal ganglia ([@B53]). {#F1} Single Photon Emission Computed Tomography {#S2.SS5} ------------------------------------------ Single-photon emission computed tomography acquisition has previously been described ([@B56]). In short, SPECT images were acquired using a Siemens Symbia T2 series SPECT-CT scanner, with the \[^123^I\]-Iodobenzamide (\[^123^I\]-IBZM) as the radioactive ligand, because of its dopamine D~2/3~R selectivity ([@B33]; [@B2]). After 180 min of rest, a CT scout and 2 × 30 min tomography were performed. CT-scout and tomography were performed to optimize positioning in the scanner and for attenuation correction. Patients underwent both baseline and follow-up SPECT scans, whereas controls only underwent baseline SPECT to minimize their exposure to radiation. At follow-up, the individual dose of amisulpride was administered 3 h prior to the scan, and s-amisulpride was measured prior to and at 60, 120, 150, 180, 210, and 240 min after administration. The mean s-amisulpride during SPECT-scan was calculated. Image Processing {#S2.SS6} ---------------- Because SPECT images contain limited anatomical information, it was not possible to automatically extract SPECT counts (counts/s) directly from our regions of interest. First, we co-registered CT- and MR anatomical images using a statistical parametric mapping method (SPM8) to calculate the transformation matrix. This step was visually inspected with an image overlay method, and manually adjusted if needed ([@B54]). Next, the CT-MR transformation matrix was used to co-register the SPECT images to the MRIs, and FSL subcortical region segmentations were resliced to fit the individual SPECT images. Subsequently, SPECT counts were extracted from FSL-MRI defined regions. Lastly, extracted SPECT counts were scatter- and decay corrected. The specific binding potentials were calculated by subtracting non-specific binding from a reference region from total binding in the regions of interest divided by the metabolite corrected plasma counts. Cerebellum as defined in [@B47] was used as reference region for non-specific binding, as in our previous study on binding potentials ([@B56]). Dopamine receptor occupancy was calculated using the following equation: O c c u p a n c y ( \% ) = ( 1 \- S ⁢ p ⁢ e ⁢ c ⁢ i ⁢ f ⁢ i ⁢ c ⁢ b ⁢ i ⁢ n ⁢ d ⁢ i ⁢ n ⁢ g ⁢ p ⁢ o ⁢ t ⁢ e ⁢ n ⁢ t ⁢ i ⁢ a ⁢ l ⁢ ( f ⁢ o ⁢ l ⁢ l ⁢ o ⁢ w ⁢ \- ⁢ u ⁢ p ) S ⁢ p ⁢ e ⁢ c ⁢ i ⁢ f ⁢ i ⁢ c ⁢ b ⁢ i ⁢ n ⁢ d ⁢ i ⁢ n ⁢ g ⁢ p ⁢ o ⁢ t ⁢ e ⁢ n ⁢ t ⁢ i ⁢ a ⁢ l ⁢ ( b ⁢ a ⁢ s ⁢ e ⁢ l ⁢ i ⁢ n ⁢ e ) ) × 100 \% Statistical Analyses {#S2.SS7} -------------------- Statistical analyses were conducted using IBM SPSS version 25. Normal distributions were assessed by Shapiro--Wilk. Equality of variance was assessed by Box's- or Levene's test. For between-groups comparisons, unpaired students t-test was used for normally distributed data and Mann--Whitney for non-normally distributed data (demographics, volumes, and binding potentials). Within-group comparisons were analyzed with paired students t-test and Wilcoxon for non-normally distributed data. Cohens *d* was used to calculate effect sizes, with effect size 0.2 considered low, 0.5 considered medium and 0.8 considered high. Pearson's Chi^2^ was used for nominal data. When correlating data, Pearson's correlation coefficient was used for parametric data, otherwise Spearman's rho was used. Our primary hypothesis was tested in two steps. First, striatal volume changes over time were tested with a repeated measure analysis. Significant group-by-time interactions were further investigated with *post-hoc* t-tests. The repeated measures analysis was initially performed for striatum, and afterward we separately analyzed the striatal subregions, i.e., caudate nucleus, putamen and nucleus accumbens. Second, we applied a multiple regression analysis to investigate the individual predictive effect of a set of variables on striatal volume increase, whilst controlling for the following included variables: amisulpride dose, striatal receptor occupancy, and baseline PANSS positive score. PANSS baseline positive scores were included in the model to control for the disease severity. Assumptions of normal distribution and no multicollinearity (Variance Inflation Factor \<10) were met. If variables were initially non-normally distributed, they were transformed to normal distributions using log10- or square root functions. Finally, we explored Spearman correlations between changes in symptom severity, striatal volumes, amisulpride dose, and D~2/3~R occupancy. Explored correlations were Bonferroni corrected for number of hypotheses tested on the same data, with a threshold of α/m, where α-level was set at 0.05, and m was number of hypotheses tested. For all other analyses, a two-sided *p*-value less than 0.05 was accepted as significant. Results {#S3} ======= Patients Compared to Healthy Controls {#S3.SS1} ------------------------------------- We included 21 patients and 23 controls with full datasets in our analyses ([Supplementary Figure S1](#SM1){ref-type="supplementary-material"}). Patients had higher use of tobacco and fewer years of education compared to controls ([Table 1](#T1){ref-type="table"}) but did not differ in other demographic factors. No difference in mean striatal volumes between patients and controls was found at baseline (*p* = 0.82) or at follow-up (*p* = 0.28). No difference in mean specific binding potentials to dopamine D~2/3~R was found between patients (2.49 ± 0.82) and controls (2.68 ± 0.71) (*p* = 0.25). ###### Demographic and clinical data. Between-groups Group; mean ± SD \[mean\] ------------------------------------------------------------------------------------------ --------------------------- ---------------- -------------------- Demographics Age, years 23.5 ± 4.8 24.1 ± 5.01 0.92^b^ Sex, male:female 10:11 12:11 0.76^c^ Handedness, right:ambidextrous:left 16:3:2 20:2:0 0.31^c,f^ Handedness score,−100:100 59.2 ± 60.7 54.6 ± 68.7 0.78^b^ Parental socioeconomic status, high:moderate:low 4:11:6 5:14:4 0.68^c,f^ Educational level, higher education/self employed, medium education, uneducated, student 0:3:4:9 0:2:0:15 0.06^c,f^ Years of education 11.9 ± 2.0 14.3 ± 2.5 **0.001**^a^ Weight, kg 78.5 ± 20.6 68.5 ± 11.0 0.058^a^ Height, cm 172.8 ± 9.5 175.1 ± 10.3 0.54^a^ Substance use, alcohol, tobacco, cannabis, benzo, opioids, stimulants 16:13:4:0:1:3 20:3:1:0:0:0 **\<0.001**^c,e,f^ Volumes (cm^3^) Baseline 18.31 ± 2.3 18.04 ± 2.5 0.82^b^ Follow-up 18.67 ± 2.3 17.92 ± 2.3 0.28^a^ Specific binding potentials (counts/s) Baseline 2.49 ± 0.82 2.68 ± 0.71 0.25^b^ Follow-up 1.38 ± 0.68 -- -- **Within patients** **Baseline** **Follow-up** PANSS scores^d^ Positive 19.8 ± 4.0 13.4 ± 3.4 **\<0.001** Negative 18.7 ± 7.2 20.3 ± 5.8 0.081 General 40.1 ± 8.5 30.2 ± 7.5 **\<0.001** Total 78.5 ± 16.4 64.0 ± 13.8 **\<0.001** Medication Dose amisulpride (mg/day) -- 233.3 ± 109.9 S-amisulpride (ng/ml) -- 399.7 ± 283.8 Duration of untreated illness (weeks) 80.8 ± 96.2 − Receptor occupancy Striatum − 44.65% ± 18.7% SD, standard deviation. a t-test, b Mann-Whitney U, c Pearsons Chi, d Wilcoxon, e p \< 0.05 only for tobacco use, f Groups have expected counts less than 5. Significant p -values are in bold. Symptom Severity and Receptor Occupancy in Patients After Treatment {#S3.SS2} ------------------------------------------------------------------- After six weeks of treatment, patients' PANSS total-, positive- and general symptom scores were significantly decreased, but negative symptoms were not ([Table 1](#T1){ref-type="table"}). Patients were treated with a mean dose of 233.3 (SD = 109.9) mg amisulpride. Oral dose and s-amisulpride correlated positively (*r*^2^ = 0.76, *p* \< 0.001). Mean receptor occupancy was 44.65% (SD = 18.7%) and correlated positively with oral dose (*r*^2^ = 0.60, *p* = 0.004) and s-amisulpride (*r*^2^ = 0.68, *p* = 0.001). Receptor occupancy is illustrated in [Figure 1](#F1){ref-type="fig"}. Amisulpride dose did not correlate with symptom severity (PANSS total) at baseline (*r*^2^ = 0.292, *p* = 0.199). Striatal Volume Increase Is Predicted by Amisulpride Dose, But Not D~2/3~R Occupancy {#S3.SS3} ------------------------------------------------------------------------------------ The repeated measures analysis revealed no volume difference between groups at either time-point, but instead a significant group-by-time interaction was observed (*p* = 0.01). The *post hoc* analysis revealed that the interaction was driven by a significant volume increase in striatum of 2.1% (95% CI = 0.52--3.68%, *p* = 0.01, Cohens *d* = 0.45) in patients. Sub-regional increases were observed in left and right caudate nucleus (2.6%) and right putamen (2.4%) ([Table 2](#T2){ref-type="table"}). The multiple regression model significantly predicted striatal volume increase (*r*^2^ = 0.411, *p* = 0.026) ([Figure 2](#F2){ref-type="fig"}), with amisulpride oral dose as the only unique, predictive factor (beta = 0.553, *p* = 0.028) ([Supplementary Table S1](#SM1){ref-type="supplementary-material"}). ###### Volumes of regions of interest. Volume (cm^3^) Patients Controls ---------------- ------------- ------------- ------------- ------------- ------------- ---------- Striatum 18.31 ± 2.3 18.67 ± 2.3 **0.01** 18.04 ± 2.5 17.92 ± 2.3 0.121^a^ Caudate 7.68 ± 1.1 7.88 ± 0.1 **\<0.001** 7.45 ± 0.9 7.37 ± 0.8 0.187 Left 3.74 ± 0.5 3.88 ± 0.5 **0.003** 3.68 ± 0.4 3.61 ± 0.4 0.067 Right 3.94 ± 0.6 4.00 ± 0.6 **0.004** 3.77 ± 0.5 3.76 ± 0.4 0.770 Putamen 9.66 ± 1.3 9.82 ± 1.3 **\<0.001** 9.63 ± 1.6 9.66 ± 1.5 0.224^a^ Left 4.87 ± 0.7 4.91 ± 0.6 0.347 4.81 ± 0.9 4.88 ± 0.9 0.670^a^ Right 4.79 ± 0.6 4.91 ± 0.7 **0.007** 4.82 ± 0.8 4.80 ± 0.7 0.212^a^ Accumbens 0.97 ± 0.2 0.98 ± 0.2 0.732 0.96 ± 0.2 0.97 ± 0.2 0.484^a^ Left 0.53 ± 0.1 0.55 ± 0.1 0.627 0.55 ± 0.1 0.55 ± 0.1 0.879^a^ Right 0.43 ± 0.09 0.43 ± 0.08 0.614^a^ 0.41 ± 0.1 0.42 ± 0.1 0.346^a^ Equal variances were tested. p -values marked with a was tested with Wilcoxon. All other volumes were tested with paired students t -test. Significant p -values are in bold. {ref-type="supplementary-material"}.](fnins-14-00484-g002){#F2} Symptom Severity Exploratory Correlations {#S3.SS4} ----------------------------------------- Reduction in positive symptoms correlated significantly with striatal volume increase (*r*^2^ = −0.472, *p* = 0.031) and this correlation was driven by a reduction in hallucinations (*r*^2^ = −0.515, *p* = 0.017). The correlations did not survive Bonferroni correction. Changes in PANSS total- or subscores did not correlate to either amisulpride dose, s-amisulpride or receptor occupancy. Discussion {#S4} ========== Primary Findings {#S4.SS1} ---------------- In line with our hypothesis, we found a significant volume increase in striatum in patients (2.1%) with a medium effect size (Cohens *d* = 0.45) after six weeks of amisulpride treatment. Our predictive model showed that dose was a predictor of volume increase, but positive symptom severity at baseline and D~2/3~R occupancy were not. Our exploratory correlation analyses indicated that striatal volume increase was associated with an improvement in positive symptoms, particularly hallucinations. Results Compared to Previous Findings {#S4.SS2} ------------------------------------- Structural brain differences between patients and healthy controls at the time of diagnosis have previously been reported, but are not consistently replicated, and often no differences are found, indicating that changes are subtle ([@B18]; [@B41]; [@B13]). Consistent with this, we did not find any significant differences in striatal volumes between patients and healthy controls. Specific D~2/3~R binding potentials did not differ between patients and controls prior to treatment, which is a replication of previous findings ([@B25]; [@B44]). In this study, mean amisulpride dose was relatively low (233.3 mg), and approximately half of the dose used in phase one of the OPTiMiSE study (488.0 mg for completers) ([@B30]). However, included patients in the OPTiMiSE study were not all antipsychotic-naïve, and due to a potential compensatory upregulation of D~2~R in response to antipsychotic treatment ([@B39]; [@B58]), this could explain the need for higher treatment doses compared to our antipsychotic-naïve patients. Furthermore, medication against adverse symptoms was not allowed in this study, which made clinicians upregulate dose slowly. Treatment dose and blood levels are inherently linked to occupancy, and it is generally accepted that a striatal dopamine receptor occupancy of 65--80% is necessary for clinical response ([@B48]; [@B57]). However, we found a significant decrease in symptom severity at mean receptor occupancy-levels of 44.65%. Most recent studies investigating occupancy levels in patients use estimations, but a cautious comparison can be made to the CATIE data ([@B37]). Authors found that for patients in stable remission, approximately half did not have continuous dopamine D~2~ blockade of ≥65%. A prospective PET study further found an optimal therapeutic window between 50 and 60% receptor occupancy on clinically stable patients with late-life schizophrenia ([@B20]). Altogether, this indicates that lower doses/D~2~ receptor occupancies in selected patient populations are sufficient, and also reduce risk of adverse effects such as extrapyramidal symptoms. We found no correlation between reduction in positive symptoms and dose or occupancy as previously found for amisulpride ([@B46]) and other antipsychotics ([@B57]), but a negative finding has also been reported ([@B3]). Different patient groups, different antipsychotics and different methodology makes results difficult to compare. Considering that amisulpride in low doses has a higher affinity toward the presynaptic dopamine D~2~ auto receptors rather than post synaptic receptors ([@B43]), the discrepancy may be explained by the relatively low dose used in our study. We did, however, find a correlation between symptom reduction and volume increase. When specific positive symptoms were examined, this correlation was linked to a decrease in hallucinations. Similar results were found in a study by [@B36], in which PANSS decrease correlated to volume increase in putamen. Our results did not survive a Bonferroni correction and should be interpreted with caution. However, this plays well into hypotheses related to altered striatal structure and connectivity linked to symptom severity ([@B45]). Basal Ganglia Volume Increases {#S4.SS3} ------------------------------ Correlation between antipsychotic dose and volume changes in striatum is a subject of much debate ([@B42]; [@B28]) in part because separating the effect of disease and medication is inherently difficult. Vernon and colleagues found proof of concept in healthy rodent models, in which chronic (8 weeks) exposure to antipsychotics, but not other psychotropics (e.g., lithium) using clinically comparable dosing, leads to structural brain changes in naïve rats, including striatal enlargement ([@B52]). In line with these data, we found a dose-dependent volume increase in the striatum after six weeks of treatment with an atypical APD with predominant D~2/3~R blockade. It is still, however, unknown what causes this volume increase. Investigation has been made into the cellular components of the volume increase, but linking structural MR changes to their cellular correlates is challenging, and although antipsychotic exposure has been found to moderate microglial activation, neuronal dendritic spine density and astrocytes ([@B51]; [@B9]; [@B1]), no studies to date have linked any of these changes to striatal volume increase. Another explanation for the volume increase could be augmented blood flow to striatum. This was found in a functional MRI study in healthy males after one dose of APD ([@B24]), as well as in patients treated with a mean of 27 days ([@B8]). Increased blood flow could possibly lead to an "apparent" volume change, but the difference in flow did not seem to have an impact on volume changes or brain structure investigated by [@B24]. Notably, [@B52] reported that striatal volume increases in rodents chronically exposed to haloperidol (8 weeks) was normalized after an equivalent period of drug washout ([@B52]). The same tendencies of volume decrease in putamen after withdrawal of antipsychotics was reported from a small schizophrenia patient cohort ([@B4]). Finally, our previous functional MRI study on a subset of the current cohort showed changes in the task-related blood oxygen level-dependent activation in striatal regions after amisulpride treatment ([@B38]). Collectively, these data suggest that the effects of antipsychotics on brain structure, including the basal ganglia, are dynamic and potentially reversible. It has long been discussed whether striatal volume changes are specific to so-called typical antipsychotics ([@B14]), but our findings together with ([@B29]) and ([@B18]) show that this is not the case. The assumption may have been due to striatal volume decreases seen in patients treated with atypical clozapine ([@B17]; [@B29]) or quetiapine ([@B15]). Both drugs, however, have low affinity toward dopamine D~2~-like receptors, whereas typical antipsychotics have high affinities ([@B10]; [@B34]; [@B29]). Dopamine D~2~-like receptor knock-out mice also show striatal volume increases, mirroring the effects of chronic exposure (9 weeks) to different APDs ([@B21], [@B22]). Notably, when chronically exposed to the same antipsychotics as wild-type mice, no additional basal ganglia volume increases were found. Taken together, these data strongly support that volume increases following antipsychotic exposure are mediated via the dopamine D~2~-like receptor ([@B21], [@B22]), rather than depending on the drug-class ('typical' vs. 'atypical' antipsychotic). Dopamine Receptor Occupancy {#S4.SS4} --------------------------- We expected the volume increase to be predicted by striatal dopamine D~2/3~R occupancy, because occupancy may be considered a more direct measure of effect than oral dose. This was not the case. We speculate that it may be due to several issues regarding SPECT imaging. First, receptor occupancy is calculated as the difference in available dopamine receptors between baseline and follow-up, and is therefore subjective to interfering factors such as changes in endogenous dopamine levels, which in turn affect dopamine receptor availability and occupancy. Second, as previously mentioned, studies also suggest a possible compensatory upregulation of D~2~R in response to treatment ([@B39]; [@B58]), which again may affect occupancy, and a potentially decreased effect of drug dose in the long-term. Third, SPECT measurements are subjective to noise, which may have obscured a potential true correlation. Another issue to consider is that the multiple regression analysis assumes linearity, which might not be the case between volume increase and receptor occupancy. Lastly, a measure of cumulative dose (although likely correlated to the mean dose) might have been a more accurate measure, but unfortunately not possible within our study. To our knowledge, only one study has done a similar investigation, also reporting no association between occupancy and volume increase ([@B12]). However, since amisulpride primarily acts by blocking D~2/3~R, we argue that the observed striatal volumetric increases still can be mediated through occupancy, and we assign the negative association with occupancy to the aforementioned issues. Strengths and Limitations {#S4.SS5} ------------------------- We conducted a clinically challenging prospective study on a cohort of antipsychotic-naïve first-episode schizophrenia patients and matched, healthy controls. Men and women were equally represented and confounding effects of previous exposure to antipsychotics could be ruled out. Amisulpride was chosen for treatment because of its selectivity toward dopamine D~2/3~R, thereby excluding potential involvement of other neuroreceptors. Because of the extensive examination program, the study included a limited sample of patients, and therefore selection bias cannot be ruled out. On the other hand, with a mean baseline PANSS total score of 78.5, the patients in our study may be considered moderately ill ([@B35]). The limited number of patients restricted the degrees of freedom in the multivariate linear regression model, and therefore it was not possible to include- and control for further variables in our analyses. The effect of nicotine on basal ganglia volumes is unresolved ([@B50]; [@B11]). Conclusion {#S5} ========== We found a dose-dependent striatal volume increase in antipsychotic-naïve schizophrenia patients in response to six weeks dopamine D~2/3~ receptor blockade with an atypical antipsychotic compound. Thus, our findings contrast the notion that striatal volume increase is restricted to "typical" antipsychotics. However, the underlying mechanisms warrant further investigation. We found the striatal volume increase to be clinically relevant, since it appears correlated to a reduction in positive symptoms. Data Availability Statement {#S6} =========================== The datasets generated for this study are available on request to the corresponding author. Ethics Statement {#S7} ================ The study was conducted in accordance with the Helsinki declaration II, and approved by the Danish research ethics committee (H-D-2008-088), as well as the Danish Data Committee (RHP-2016-025, I-suite no. 05181). [Clinical Trials.gov](https://clinicaltrials.gov/) Identifier: NCT01154829. The patients/participants provided their written informed consent to participate in this study. Author Contributions {#S8} ==================== BG and BE conceived and designed the study. SW and MN collected the data. JR, CS, SW, ER, LJ, PA, and LP contributed with data processing and analysis tools. HA, JR, LBJ, ER and BE conducted the statistical analyses. HA and BE drafted the manuscript. HA, JR, CS, and SW particularly contributed with method section. AV contributed to the interpretation and discussion. All authors fulfill authorship criteria of the ICMJE by substantial contribution to the conception and design, to acquisition of data, or to the analysis and interpretation of the data contributed to manuscript revision, read and approved the submitted version. Conflict of Interest {#conf1} ==================== BE has received lecture fees and/or is part of Advisory Boards of Bristol-Myers Squibb, Eli Lilly and Company, Janssen-Cilag, Otsuka Pharma Scandinavia AB, Takeda Pharmaceutical Company and Lundbeck Pharma A/S. BG is the leader of a Lundbeck Foundation Centre of Excellence for Clinical Intervention and Neuropsychiatric Schizophrenia Research (CINS), which is partially financed by an independent grant from the Lundbeck Foundation based on international review and partially financed by the Mental Health Services in the Capital Region of Denmark, the University of Copenhagen, and other foundations. Her group has also received a research grant from Lundbeck A/S for another independent investigator-initiated study. All grants are the property of the Mental Health Services in the Capital Region of Denmark and administrated by them. She has no other conflicts to disclose. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. **Funding.** HA was funded by a scholarship from the Lundbeck Foundation and a grant from A. P. Møller Fonden. The Centre for Clinical Intervention and Neuropsychiatric Schizophrenia Research, CINS is funded by Lundbeck Foundation grant number R25-A2701. We would like to thank patients and controls for their participation in this study, and the Lundbeck Foundation for sponsoring the scholar-stipendium to HA. We would also like to thank Anthony Vernon for accepting a visit from HA at the Maurice Wohl Neuroscience Institute. Supplementary Material {#S11} ====================== The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fnins.2020.00484/full#supplementary-material> ###### Click here for additional data file. [^1]: Edited by: Andrew J. Grottick, Beacon Discovery Inc., United States [^2]: Reviewed by: Doris Doudet, University of British Columbia, Canada; Robert Warren Gould, Wake Forest School of Medicine, United States [^3]: This article was submitted to Neuropharmacology, a section of the journal Frontiers in Neuroscience

Introduction ============ Background ---------- Smartphone apps, software designed to run on mobile devices, have integrated into many aspects of daily lives and are increasingly used for disease management. For example, there are more than 2000 consumer apps to choose from to self-manage diabetes \[[@ref1]\]. In contrast to Web-based information accessed from the computer, apps can help the user to conveniently access information with their portability and improved technological capabilities, such as increased human interaction, interoperability with other devices, and easy data collection. Health apps can have a multitude of uses, including offering advice on healthy living, communication with health care providers, and providing decision support through granular biometric data collection (eg, blood glucose) \[[@ref2]-[@ref4]\]. The adoption of health apps for chronic conditions such as diabetes is expected to grow \[[@ref5]\]. There are currently 2.5 billion smartphone users in the world \[[@ref6]\] and more than 300,000 health apps available for consumer download \[[@ref5]\]. One main reason for the proliferation of health apps is the low barrier of entry for app developers to publish apps \[[@ref3]\]. Although the 2 major app stores (ie, Apple and Google Play) review apps before publication ([Multimedia Appendix 1](#app1){ref-type="supplementary-material"}), many apps that do not conform to the prereview checklist fall through the cracks and are published. The regulation of app stores also falls outside the purview of governmental agencies, such as purview of the Food and Drug Administration (FDA) \[[@ref7]\]. Only a small subset of apps that can pose a higher risk and meet the regulatory definition of *device* are regulated by the FDA \[[@ref8]-[@ref10]\]. Therefore, apps with inaccurate content or advertisements may still be published and be available to consumers \[[@ref11]-[@ref13]\]. The lack of transparency regarding an app's source of content may compromise the reliability of the information it disseminates \[[@ref9],[@ref14],[@ref15]\] and can potentially mislead or cause harm to patients who have low health literacy \[[@ref16]\]. Concerns over the credibility and reliability of Web-based health information sources began to surface in the early days of internet usage \[[@ref17],[@ref18]\]. The Health On the Net code of conduct (HONcode), which covers 8 principles (ie, authoritative, complementarity, privacy, attribution, justifiability, transparency, financial disclosure, and advertising policy) for website certification, was developed to help guide and standardize the reliability of health and medical information published on the internet \[[@ref19]\]. Websites that were certified by the HONcode were assessed to be of higher quality and may reduce consumers' burden of searching for good-quality websites \[[@ref20],[@ref21]\]. Objectives ---------- As part of a larger study investigating the medication management features of diabetes apps \[[@ref22]\], this study aimed to assess the transparency and reliability of information disseminated via these apps against 8 criteria adapted from the HONcode principles. Methods ======= Development of App Assessment Criteria -------------------------------------- We adapted the 8 HONcode principles (ie, authoritative, complementarity, privacy, attribution, justifiability, transparency, financial disclosure, and advertising policy) and termed it the App-HONcode criteria to suit the context of health apps and apps assessment. The initial versions of our criteria were piloted with highly downloaded (≥100,000 downloads) diabetes apps to refine and improve the clarity of the assessment criteria. Unclear statements were discussed among the app assessment team members until a consensus was reached. The *not applicable (N/A)* option was included for the assessment of *attribution* and *justifiability* to account for apps that did not provide health information within the app. The final assessment criteria are shown in ([Table 1](#table1){ref-type="table"}). ###### Adapted Health On the Net code of conduct criteria for health app assessment. Attribute HONcode^a^ App-HONcode criteria Options ---------------------- ------------------------------------------------------------------------------------------------ ----------------------------------------------------------------------------------------------------------------------------------------------------- --------------------------- Authoritative The qualifications of the authors are indicated. Does the app indicate the qualifications of specific individuals who developed or contributed to the development of the app? Yes or no Complementarity Information should support, not replace, the doctor-patient relationship. Is there a disclaimer stating or which implies that the information provided and/or app functions do not replace the health care provider's advice? Yes or no Privacy Respect the privacy and confidentiality of personal data submitted to the site by the visitor. Is there a privacy and confidentiality clause in the app? Yes or no Attribution Cite the source(s) and date of published information on medical and health pages. Are information sources in the app cited? Yes, no, or N/A^b^ Justifiability Justifiability: site must back up claims relating to benefits and performance. Are the claims relating to benefits and performance in the app description backed up by evidence? (Answer *N/A* if there are no claims) Yes, no, or N/A Transparency Accessible presentation and accurate email contact. Are the developers contactable by email? Yes or no Financial disclosure Identify funding sources. Does the app indicate any funding sources? (*Yes* if the app is managed by a registered commercial company) Yes or no Advertising policy Clearly distinguish advertising from editorial content. Are advertorials distinguishable from content of the app? Yes, no or no advertising ^a^HONcode: Health On the Net code of conduct. ^b^Not applicable. App Selection and Assessment ---------------------------- As the app assessments were part of a larger study investigating medicines management functionalities of diabetes apps, a more detailed description of app selection and assessment can be found in another paper \[[@ref22]\]. ### App Search Strategy Diabetes terms in the English language were searched in the Google Play and Apple app stores via an app market explorer, 42matters \[[@ref23]\], on June 12, 2018, to identify apps that were marketed for diabetes self-management. The search terms *(Diabetes OR Diabetic OR Diabetics) OR (glucose OR glycaemic OR glycemic OR blood sugar OR HbA~1c~ OR A~1c~) OR insulin* were used to search app descriptions, and a list of app titles with descriptions was produced for screening. ### App Selection A random sample of 100 apps was first screened by 2 researchers (ZH and MLT) to ensure consistency in interpretation of the inclusion and exclusion criteria listed below. Differences in interpretations were resolved via consensus discussion with 2 other team members (EL and GJ). Random samples of apps were rescreened until an interrater agreement of above 80.0% was achieved. ### Inclusion Criteria The inclusion criteria were as follows: Apps with medication self-management features (ie, medication scheduling, reminders, tracking, information provision, and adherence review), apps with any blood glucose logging features, apps in the English language, free apps, and apps requiring payment. ### Exclusion criteria The exclusion criteria were as follows: Patient health portals linking to patients' electronic health records, apps that were not updated after January 1, 2017, intended only for health care professionals, insulin calculators/bolus correctors only, apps with exclusive blood glucose monitoring device tie-in requirement (ie, not permitting manual entry of blood glucose values), apps duplicated on the same platform, apps with regional restrictions, and technical problems (eg, crashes, screen hangs, unable to login, and unable to download). ### App Assessments Included apps were evaluated against the adapted App-HONcode criteria. In total, 5 team members (ZH, EL, GJ, CT, and MLT) underwent a calibration exercise to ensure consistency in criteria interpretation before the app assessments. Apps available on the iOS and Android platforms were treated as unique apps and assessed on both platforms because of possible differences in versions across platforms. The number of *yes* responses was summed up for each app to determine the number of App-HONcode criteria met. *N/A* responses for *attribution* and *justifiability* were treated as *not meeting the criteria* for a more conservative approach, whereas the *no advertising* response for *advertising policy* was treated as a *yes* (conforming to this principle). Statistical Analyses -------------------- Apps were grouped by operating platform (ie, Android and iOS) and profiled according to each App-HONcode criterion using descriptive statistics. Only Android apps were further classified by the number of downloads (ie, ≥100,000 downloads and \<100,000 downloads), as information on the number of downloads was not available for iOS apps. Pearson chi-square test was used for comparisons between groups, and a 2-tailed Fisher exact test was used where the expected count was less than 5 in a group. Statistical significance was set at a *P* value of less than .05. All analyses were performed using SPSS (version 22; IBM Corp). Results ======= App Screening ------------- We identified and downloaded 351 apps (191 Android and 160 iOS) after app title and description screening, of which 143 apps (81 Android and 62 iOS) met the inclusion criteria and were assessed against the app assessment criteria. The detailed app search results can be found in our study \[[@ref22]\]. Characteristics of Included Apps -------------------------------- The number and proportion of the assessed apps meeting the App-HONcode criteria are shown in [Figure 1](#figure1){ref-type="fig"}. Most of the apps on the Android and iOS platforms fulfilled between 2 and 5 criteria; 1 Android app met all 8 criteria, whereas another did not meet any criteria. A higher proportion of apps published on the iOS platform met more App-HONcode criteria compared with apps on the Android platform. For example, 43.6% of iOS apps met at least five App-HONcode criteria compared with 28.3% of Android apps. {#figure1} The profile of app attributes grouped by platform is shown in [Table 2](#table2){ref-type="table"}. Few (20/143, 14.0%) apps mentioned the qualifications of individuals who contributed to app development. Less than half (56/143, 39.2%) of the apps had a disclaimer stating that the information provided/app functions do not replace a health care provider's advice, and approximately two-thirds (93/143, 65.0%) of apps had a privacy and confidentiality clause. Of the apps providing health or medical information or made claims on its efficacy, only one-third cited their information sources (15/42, 36%) and/or backed up the claims relating to benefits and performance in the app by evidence (7/23, 30%). ###### Profile of app attribute grouped by platform. HONcode^a^ principle App-HONcode criteria All apps, n (%) Android, n (%) iOS, n (%) *P* value ----------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------- ----------------- ---------------- ------------- ----------- Authoritative Does the app indicate the qualifications of specific individuals who developed or contributed to the development of the app? 20 (14.0)^b^ 9 (11)^c^ 11 (18)^d^ .33 Complementarity Is there a disclaimer stating or which implies that the information provided and/or app functions do not replace the health care provider's advice? 56 (39.2)^b^ 27 (33)^c^ 29 (47)^d^ .12 Privacy Is there a privacy and confidentiality clause in the app? 93 (65.0)^b^ 48 (59)^c^ 45 (73)^d^ .11 Attribution Are information sources in the app cited? 15 (34)^e,f^ 8 (36)^e,g^ 7 (32)^e,h^ \>.99^i^ Justifiability Are the claims relating to benefits and performance in the app description backed up by evidence? (Answer *not applicable* if there are no claims) 7 (30)^e,j^ 4 (29)^e,k^ 3 (33)^e,l^ \>.99^i^ Transparency Are the developers contactable by email? 131 (91.6)^b^ 79 (98)^c^ 52 (84)^d^ .005^m^ Financial disclosure Does the app indicate any funding sources? (*Yes* if the app is managed by a registered commercial company) 88 (61.5)^b^ 46 (57)^c^ 42 (68)^d^ .23 Advertising policy There are no advertisements in the app 119 (83.2)^b^ 64 (79)^c^ 55 (89)^d^ .18 Are advertorials distinguishable from content of the app? 18 (75)^n,o^ 12 (71)^n,p^ 6 (86)^n,q^ .63 ^a^HONcode: Health On the Net code of conduct. ^b^N=143. ^c^N=81. ^d^N=62. ^e^Not applicable removed before statistical analysis and percentage computation. ^f^N=44. ^g^N=22. ^h^N=22. ^i^Two-tailed *P* value calculated using Fisher exact test, as the expected count is less than 5 in at least one group. ^j^N=23. ^k^N=14. ^l^N=9. ^m^Statistical significance *P*\<.05 in the comparison between Android and iOS app features. ^n^Percentage is computed by dividing the number of apps with distinguishable advertisements with the total number of apps with advertisements. ^o^N=24. ^p^N=17. ^q^N=7. There were no significant differences between the Android and iOS platforms in the proportion of apps fulfilling each criterion except for the principle of *transparency*. Android apps had a significantly higher proportion of apps with the developer's email listed on the Google play store compared with apps listed on the Apple store (Android: 75/81, 98%; and iOS: 52/62, 84%; *P*=.005). More than half (88/143, 61.5%) of the apps disclosed funding sources. Finally, most 119/143, 83.2%) of the assessed apps did not have advertisements; of apps with advertisements, three-fourths (18/24, 75%) were distinguishable from the content of the app. Android Apps by Downloads ------------------------- [Table 3](#table3){ref-type="table"} shows the profile of app attribute grouped by a low (\<100,000 downloads) and high number of downloads (≥100,000 downloads) for Android apps. There were no significant differences between apps with a low and high number of downloads in terms of the *authoritative, complementarity, attribution, justifiability, transparency,* and *advertising policy*. A significantly higher proportion of highly downloaded apps had a privacy and confidentiality clause (high downloads: 15/17, 88%; and low downloads: 33/64, 52%; *P*=.006) and were more likely to discuss their funding sources (high downloads: 15/17, 82%; and low downloads: 32/64, 50%; ###### Profile of Android app attributes grouped by the number of downloads. HONcode^a^ principle App-HONcode criteria All Android apps, n (%) \<100,000 downloads, n (%) ≥100,000 downloads, n (%) *P* value ----------------------------------------------------------- ----------------------------------------------------------------------------------------------------------------------------------------------------- ------------------------- ---------------------------- --------------------------- ----------- Authoritative Does the app indicate the qualifications of specific individuals who developed or contributed to the development of the app? 9 (11)^b^ 8 (13)^c^ 1 (6)^d^ .68^e^ Complementarity Is there a disclaimer stating or which implies that the information provided and/or app functions do not replace the health care provider's advice? 27 (33)^b^ 18 (28)^c^ 9 (53)^d^ .08 Privacy Is there a privacy and confidentiality clause in the app? 48 (59)^b^ 33 (52)^c^ 15 (88)^d^ .006^f^ Attribution Are information sources in the app cited? 8 (36)^g,h^ 7 (37)^g,i^ 1 (33.3)^g,j^ \>.99^e^ Justifiability Are the claims relating to benefits and performance in the app description backed up by evidence? (Answer *not applicable* if there are no claims) 4 (29)^g,k^ 3 (30)^g,l^ 1 (25)^g,m^ \>.99^e^ Transparency Are the developers contactable by email? 79 (98)^b^ 63 (98)^c^ 16 (94)^d^ .38 Financial disclosure Does the app indicate any funding sources? (*Yes* if the app is managed by a registered commercial company) 46 (57)^b^ 32 (50)^c^ 14 (82)^d^ .026^f^ Advertising policy There are no advertisements in the app 64 (79)^b^ 50 (78)^c^ 14 (82)^d^ \>.99^e^ Are advertorials distinguishable from content of the app? 12 (71)^n,o^ 11 (79)^n,p^ 1 (33)^n,q^ .19^n^ ^a^HONcode: Health On the Net code of conduct. ^b^N=81. ^c^N=64. ^d^N=17. ^e^Two-tailed *P* value calculated using Fisher exact test, as the expected count is less than 5 in at least one group. ^f^Statistical significance *P*\<.05 in the comparison between Android and iOS app features. ^g^Not applicable removed before statistical analysis and percentage computation. ^h^N=22. ^i^N=19 ^j^N=3. ^k^N=14. ^l^N=10. ^m^N=4. ^n^Percentage is computed by dividing the number of apps with distinguishable advertisements with the total number of apps with advertisements. ^o^N=17. ^p^N=14. ^q^N=3. Discussion ========== Principal Finding ----------------- We evaluated 143 apps against 8 App-HONcode criteria to assess the transparency and reliability of information disseminated via diabetes apps. Most apps fulfilled between 2 and 5 assessment criteria, and only 1 Android app fulfilled all 8 criteria. Overall, a higher proportion of iOS apps fulfilled more App-HONcode criteria compared with Android apps, although the differences were not significant. Many apps were not transparent in indicating the content source of the app. More than half of the assessed apps did not fulfill important criteria, such as indicating the qualifications of individuals involved in the app development and disclaiming that the app does not replace health care provider's advice. This concurs with a study assessing eczema apps, where only 15% of the app developers indicated their qualifications, and 46% disclaimed that the app does not replace the advice of the health care provider \[[@ref24]\]. Although it may be challenging to indicate the qualifications of all individuals involved in the development of a complex app, the qualifications of the main content contributors and a representation of their collaborators should minimally be quoted to account for the content source of the app. Approximately three-fourths of the assessed apps did not provide any health information. This was not surprising, as disease management apps tend to emphasize management aspects rather than educating the patient \[[@ref1]\], which, in our view, presents a missed opportunity for patient education, which can be incorporated into apps. Of the apps that provided health information, only one-third cited the source of information. Few of the assessed apps had claims relating to the benefits and performance of the app. However, only one-third of these apps backed the claims with evidence. Apps or any consumer products with unsubstantiated claims have the potential to mislead and cause harm to the undiscerning consumer \[[@ref16],[@ref25]\]. Therefore, it is imperative for app stores to check the veracity of claims used in the app description before its publication on the app stores. Most apps had an email of the developer displayed, but the email address does not guarantee access to the app developer. We contacted the developers of apps that had access restrictions, and only 10% responded to our emails (2 emails sent a week apart). This percentage should be higher for apps that are accessible, but there is a possibility of inoperative email addresses being displayed on the app store. App stores should ensure the inclusion of a valid email address for all health apps for consumer inquiries. Privacy breaches can erode consumers' trust in the app. Two-thirds of the apps assessed had a privacy and confidentiality clause. This was an improvement from a study published 6 years ago assessing the availability, scope, and transparency of mobile health (mHealth) app privacy policies on 600 commonly used mHealth iOS and Android apps \[[@ref15]\]. One explanation for this improvement could be the changes made to the app store policies to improve the quality of apps over the years \[[@ref26],[@ref27]\]. However, stricter scrutiny is required on the part of the app stores, given the absence of privacy policies in many of the assessed apps. Although there is an improvement in the presence of privacy policy of the English-language apps we assessed, those that are published on other platforms and in other languages may yield different assessment results. Approximately 40% of the assessed apps did not have their funding sources indicated. The funding source of an app will affect its development, quality, and the services provided. This represents a gap in which app stores can play a role to improve the quality assurance of health apps. Advertisements were not present in 80% of the assessed apps. The proportion of apps with advertisements in our study may be lower, as we assessed the best version of apps requiring in-app payments for feature upgrades. Previous studies have shown that paid apps were not of higher quality compared with free apps \[[@ref28],[@ref29]\]. Approximately one-third of the in-app advertisements were judged as being nondistinguishable from the content of the app. We did not scrutinize the appropriateness of the advertising content, which may present an additional gap in the quality and reliability of information disseminated via these apps. The originators of the HONcode recently published an extension for apps---mHONcode---to cover the certification of health information disseminated on apps \[[@ref30]\]. This was only available after we completed our app assessments, but there are minimal conceptual differences between our App-HONcode and the mHONcode. Our criteria were worded to minimize the subjectivity of assessment by different researchers. Limitations ----------- There were limitations to the study, despite attempts to minimize bias. First, the scope was limited to diabetes management apps because of being part of a larger study investigating the medication management features of diabetes apps. However, our findings are generalizable to other diabetes apps, as the apps were identified using a systematic search and selection strategy. Second, our assessment criteria may have underrated apps that do not provide health information (eg, medication logging apps). Even so, many apps were not transparent in data privacy and in clearly distinguishing the complementarity of the app (ie, not replacing the health care provider's advice). Third, the assessment may not reflect the current state of the apps because of constant app updates. However, we believe that our findings remain unchanged, as app updates were mainly for bug fixes and feature upgrades. Fourth, we neither assessed issues surrounding data management nor the content of privacy and confidentiality clauses, which may not accurately disclose the sharing of some personal information \[[@ref31]\]. Finally, app assessments were subjective to researcher interpretation, although we attempted to reduce researcher bias by piloting the assessment and using a standardized patient profile when interacting with apps. Implications and Future Research -------------------------------- The fulfillment of the 8 App-HONcode criteria are actionable, but not many developers may be aware of the need to indicate background information or to check the content of advertisements, as their main aim is to get the app published. App developers and consumers would benefit from the availability of a standardized checklist to assess the information quality of health apps. It would be challenging for governments to regulate all health apps because of its ubiquity, pace of development, and ambiguity in definition. Therefore, app stores should play a larger role in making the transparency and reliability of information dissemination a basic requirement for app publication. As observed from our app assessments, owing to the higher barriers of entry currently set by the Apple app store (see [Multimedia Appendix 1](#app1){ref-type="supplementary-material"}), it had a larger proportion of apps that fulfilled more App-HONcode criteria. As health apps collect a lot more user data than the internet, the consent-seeking process for data collection and data management policies of these apps should be evaluated in the future. The appropriateness of advertising and clarity of privacy clauses should also be checked, for diabetes management and other chronic disease management apps in other languages and on other platforms, to provide a more complete landscape of the transparency and reliability of information disseminated and collected through these apps. Conclusions ----------- Our systematic app assessments of the transparency and reliability of health information disseminated in diabetes apps discovered gaps in the disclosure of the developer's qualification, funding source, and the complementary role of the app in disease management. App stores should play a larger role in scrutinizing app publication, as higher barriers of app entry will lead to the publication of apps with better disclosure of the app's content source. As the development of the App-HONcode criteria is preliminary, future work can further examine the consent-seeking process for data collection, data management policies, the appropriateness of advertising content, and the clarity of privacy clauses. This study was funded by the Nanyang Technological University, Singapore, through (1) the Centre for Population Health Sciences, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore; (2) the Ageing Research Institute for Society and Education, Nanyang Technological University, Singapore; and (3) the Institute for Health Technologies, Nanyang Technological University, Singapore. The funder did not have direct involvement in the research including manuscript preparation, apart from employing the authors. The research presented in this paper is solely the responsibility of the authors and does not reflect the views of the funder. The authors wish to acknowledge Mr Geronimo Jimenez for his input in piloting and refining the app assessment criteria and Mr Geronimo Jimenez, Ms Christina Tan, and Ms My Linh Thai for their assistance in app assessment. ZH would like to acknowledge funding from the Nanyang Technological University Research Scholarship. Authors\' Contributions: ZH conceptualized and contributed to the development and refinement of the adapted HONcode app assessment criteria, screened apps, assessed apps, cleaned and analyzed data, interpreted data, and drafted and revised the manuscript. EL coconceptualized and provided critical input into the developed app assessment criteria, assessed apps, interpreted data, and revised the manuscript. JC conceptualized the study, obtained the funding, supervised the team, and provided critical input into all stages of the study and critical review of the draft manuscript. All authors reviewed and approved the final version of the manuscript before submission. Conflicts of Interest: None declared. Selected pre-review app publication checklist of Apple and Google Play app stores. FDA : Food and Drug Administration HONcode : Health On the Net code of conduct mHealth : mobile health

1. Introduction =============== The Gram-positive bacterium, *Staphylococcus aureus* (*S. aureus*), is an important pathogen that can cause life-threatening bacterial infections, such as pneumonia, meningitis, osteomyelitis, endocarditis and toxic shock syndrome \[[@B1-molecules-16-06255]\]. The introduction of β-lactam antibiotics has greatly improved the prognosis of patients with severe staphylococcal infections; however, the resistance of *S. aureus* to wide spectrum β-lactam antibiotics, such as penicillin G, methicillin and oxacillin, has emerged in several countries \[[@B2-molecules-16-06255]\]. Methicillin-resistant *Staphylococcus aureus* (MRSA) has become a worldwide public health problem and is a major cause of both nosocomial and community infections. Although the glycopeptide antibiotic, vancomycin, is considered indispensable for the treatment of MRSA infections, the first MRSA strain to acquire resistance to vancomycin was isolated from a Japanese patient in 1996, and this was followed by reports of similar resistant strains from the USA, France, Korea, South Africa, and Brazil \[[@B3-molecules-16-06255]\]. Under these circumstances, the development of new antibacterial agents to control MRSA is urgently needed. Plants are rich in a wide variety of secondary metabolites, including alkaloids, terpenes, flavonoids and tannins, all of which are known to possess antibacterial activity \[[@B4-molecules-16-06255]\]. The ethanolic extract of Thai mango (*Mangifera indica* L. cv. 'Fahlun', Anacardiaceae) seed kernels (MSKE) contains a relatively high phenolic content of pentagalloylglucopyranose (PGG) (61.28%) and relatively smaller amounts of methyl gallate (MG) (0.68%) and gallic acid (GA) (0.44%) \[[@B5-molecules-16-06255]\]. MSKE and its principles have been pharmacologically documented to have antioxidant, anti-tyrosinase, anti--inflammatory, and hepatoprotective activities \[[@B5-molecules-16-06255],[@B6-molecules-16-06255]\] as well as anti-enzymatic activities against snake venom \[[@B7-molecules-16-06255],[@B8-molecules-16-06255]\]. The objectives of this study were to investigate the inhibitory potential of MSKE and its isolated phenolic principles against MRSA and the capacity of these principles to modulate β-lactam resistance in MRSA. In addition, the effect of MSKE and PGG on the bacterial structure was observed using electron microscopy. 2. Results and Discussion ========================= 2.1. Disc Diffusion Method -------------------------- The MSKE displayed antimicrobial activity against both *S. aureus* ATCC 25923 and all of the 19 tested MRSA strains as shown by the presence of inhibition zones in the disc diffusion study in [Table 1](#molecules-16-06255-t001){ref-type="table"}. molecules-16-06255-t001_Table 1 ###### Antibacterial activity of MSKE against *S. aureus* ATCC 25923 and 19 clinical MRSA isolates (nz = no inhibition zone,^a^ mean values ± S.D. of triplicate results, ^b^ mean values ± S.D. from 19 MRSA strains,\*no significant difference from *S. aureus* ATCC 25923 at *p* \> 0.01). Bacterial strains Mean diameter of inhibition zone (mm)^a^ -------------------------------- ------------------------------------------ ---------------- --------------- --------------- --------------- ---------------- ***S. aureus*** **ATCC 25923** nz 11.44 ± 0.59 14.81 ± 0.49 13.94 ± 1.34 17.06 ± 3.23 18.31 ± 0.52 **Clinical MRSA strains** M 01 nz 12.00 ± 0.14 14.08 ± 0.25 14.17 ± 0.23 18.25 ± 1.14 21.00 ± 0.33 M 02 nz 12.03 ± 0.19 14.11 ± 0.46 17.20 ± 2.31 17.92 ± 1.53 20.44 ± 2.50 M 03 nz 11.92 ± 0.68 15.58 ± 0.51 15.53 ± 0.90 20.50 ± 2.28 22.89 ± 2.01 M 04 nz 11.28 ± 0.38 14.08 ± 0.82 14.56 ± 0.55 18.19 ± 1.61 19.67 ± 0.33 M 05 nz 11.58 ± 0.52 13.69 ± 0.35 13.20 ± 0.52 16.03 ± 0.20 19.78 ± 0.19 M 06 nz 11.92 ± 0.82 14.89 ± 0.71 14.50 ± 1.65 18.78 ± 2.67 19.33 ± 0.67 M 07 nz 10.92 ± 0.46 14.28 ± 0.57 14.33 ± 0.71 18.53 ± 0.91 20.89 ± 0.84 M 08 nz 11.31 ± 0.51 14.36 ± 0.27 15.59 ± 0.52 19.34 ± 0.54 19.78 ± 0.69 M 09 nz 10.39 ± 0.46 13.08 ± 0.76 13.36 ± 0.51 17.97 ± 0.20 20.44 ± 1.95 M 10 nz 12.44 ± 0.51 15.72 ± 1.50 14.52 ± 0.35 18.36 ± 0.20 18.33 ± 0.58 M 11 nz 8.17 ± 0.29 10.06 ± 0.10 12.17 ± 0.76 13.56 ± 0.51 18.61 ± 0.54 M 12 nz 10.33 ± 0.44 12.11 ± 0.10 14.11 ± 0.25 15.72 ± 0.10 19.11 ± 1.07 M 13 nz 9.11 ± 0.10 10.72 ± 0.59 12.72 ± 0.69 14.72 ± 0.86 18.33 ± 1.04 M 14 nz 8.56 ± 0.35 10.56 ± 0.35 13.00 ± 0.17 14.61 ± 0.54 21.89 ± 1.95 M 15 nz 10.44 ± 0.67 12.39 ± 0.54 14.22 ± 1.51 16.83 ± 1.09 20.61 ± 1.69 M 16 nz 9.56 ± 0.25 11.00 ± 0.29 13.67 ± 0.33 15.39 ± 0.38 19.67 ± 0.73 M 17 nz 10.28 ± 0.19 11.61 ± 0.42 13.61 ± 0.69 15.39 ± 0.54 19.28 ± 0.54 M 18 nz 9.44 ± 1.08 10.83 ± 0.60 12.61 ± 0.92 15.00 ± 0.76 20.44 ± 1.80 M 19 nz 9.94 ± 1.55 11.28 ± 1.44 12.89 ± 1.55 15.06 ± 1.40 18.72 ± 0.92 **Mean MRSA (n = 19) ^b^** nz 10.61 ± 1.25\* 12.87 ±1.82\* 14.00 ±1.21\* 16.85 ±1.94\* 19.96 ± 1.20\* The mean diameter of the inhibition zone was increased from 11.44 ± 0.59 to 14.81 ± 0.49, 13.94 ± 1.34 and 17.06 ± 3.23 mm for *S. aureus* ATCC 25923 over an increase in concentration of 0.625 to 1.25, 2.50 and 5.00 mg/disc, respectively. The uneven increase trend in the inhibition zones could be a result of uneven drug diffusion from the paper disc to agar plate. At the same concentration range, similar mean inhibition zones (10.61 ± 1.25 to 12.87 ± 1.82, 14.00 ± 1.21 and 16.85 ± 1.94 mm) were found for all 19 clinical MRSA isolates. There was no significant difference between the mean inhibition zone of the reference MSSA strain and the MRSA strains (*p* \> 0.01). The solvent used, 10% DMSO, produced no visible inhibition zone in this study, whereas the positive control, vancomycin at 30 μg/disc, produced mean inhibition zones of 18.31 ± 0.52 mm for the standard MSSA strain and 19.96 ± 1.20 mm for the 19 MRSA strains. 2.2. The Minimum Inhibitory Concentrations (MICs) and Minimum Bactericidal Concentrations (MBCs) ------------------------------------------------------------------------------------------------ [Table 2](#molecules-16-06255-t002){ref-type="table"} shows the MIC and MBC values of MSKE and its phenolic principles (PGG, GA, MG) against *S. aureus* ATCC 25923 and 19 clinical MRSA isolates as compared to the reference antibiotics. The MIC and the MBC values of these phenolic principles against the standard MSSA strain were lower than those generated by MSKE. The order of potency for MICs was PGG (0.13 ± 0.00 mg/mL) \< GA (0.19 ± 0.00 mg/mL) \< MG (0.38 ± 0.00 mg/mL) \< MSKE (0.47 ± 0.00 mg/mL). The order of potency for MBCs was PGG (0.50 ± 0.00 mg/mL) \< GA (0.75 ± 0.00 mg/mL) \< MG (1.00 ± 0.00 mg/mL) \< MSKE (1.88 ± 0.00 mg/mL). For the 19 clinical MRSA isolates, the MICs of MSKE and its phenolic principles against these MRSA strains ranged from 0.13 ± 0.00 to 0.25 ± 0.00 mg/mL for PGG, 0.47 ± 0.00 mg/mL for MSKE, 0.19 ± 0.00 to \>3.00 ± 0.00 mg/mL for GA, and 0.75 ± 0.00 to 2.50 ± 0.87 mg/mL for MG. The MBC values were as follows: PGG 1.00 ± 0.00 to \>1.00 ± 0.00 mg/mL, MG 1.00 ± 0.43 to 2.50 ± 0.87 mg/mL, GA 0.38 ± 0.00 to \>3.00 ± 0.00 mg/mL and MSKE 0.94 ± 0.00 to 3.75 ± 0.00 mg/mL. For the reference antibiotics, vancomycin exhibited both lower MIC (0.78 ± 0.00 µg/mL) and MBC (3.13 ± 0.00 µg/mL) values than penicillin G (MIC and MBC = 41.67 ± 18.04 µg/mL) against the standard MSSA strain. All 19 MRSA isolates were resistant to penicillin G with MIC and MBC values ranging from 125.00 ± 0.00 to 8,000.00 ± 0.00 µg/mL. The 19 MRSA isolates were susceptible to vancomycin with MIC and MBC values ranging from 1.56 ± 0.00 to 6.25 ± 0.00 µg/mL and 3.13 ± 0.00 to 12.50 ± 0.00 µg/mL, respectively. These broth microdilution results agreed with the disc diffusion assays, which demonstrated that the antibacterial activity of MSKE against the standard MSSA strain and all of the clinical MRSA strains were not significantly different (*p* \> 0.01) as shown by their similar MBC mean values (1.88 ± 0.00 and 1.83 ± 0.79 mg/mL, respectively) and equivalent MIC value (0.47 ± 0.00 mg/mL) ([Table 2](#molecules-16-06255-t002){ref-type="table"}). Among the phenolic principles of MSKE, PGG exhibited the most potent antibacterial activity against both the standard MSSA strain and all 19 clinically isolated strains, exhibiting the lowest mean MIC values (0.13 ± 0.00 and 0.16 ± 0.03 mg/mL, respectively) compared to GA (0.19 ± 0.00 and \>1.07 ± 1.19 mg/mL, respectively) and MG (0.38 ± 0.00 and 1.33 ± 0.34 mg/mL, respectively). We observed that the mean MIC of PGG against all tested strains was approximately 3−fold lower than that of MSKE. Similarly, PGG also exhibited lower mean MBC values (\>1.00 ± 0.00 mg/mL) against all 19 MRSA strains compared to GA (\>1.25 ± 1.09 mg/mL) and MG (1.42 ± 0.32 mg/mL). Because PGG is the major principle of MSKE, present at the highest concentration at approximately 65% and GA and MG are present at only trace levels (0.88 and 0.62%, respectively), the data imply that PGG may be the major contributor to the antibacterial activity of MSKE; there may also be other unidentified constituents of MSKE that possess lower antibacterial potencies than PGG that have not yet been isolated. molecules-16-06255-t002_Table 2 ###### The mean MICs and MBCs of MSKE and its phenolic principles against *S. aureus* ATCC 25923 and 19 clinically isolated MRSA strains (^a^ mean values ± S.D. of triplicate results, ^b^ mean values ± S.D. from 19 MRSA strains). Bacterial strains Susceptibility of Bacteria^a^ -------------------------------- ------------------------------- ------------- ------------- --------------- ------------- ------------- --------------- --------------- ------------- -------------- --------------------- --------------------- ***S. aureus*** **ATCC 25923** 0.47 ± 0.00 1.88 ± 0.00 0.13 ± 0.00 0.50 ± 0.00 0.38 ± 0.00 1.00 ± 0.00 0.19 ± 0.00 0.75 ± 0.00 0.78 ± 0.00 3.13 ± 0.00 41.67 ± 18.04 41.67 ± 18.04 **Clinical MRSA strains** M 01 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.25 ± 0.43 1.50 ± 0.00 0.25 ± 0.11 0.75 ± 0.00 1.56 ± 0.00 6.25 ± 0.00 2,000.00 ± 0.00 8,000.00 ± 0.00 M 02 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.00 ± 0.43 1.25 ± 0.43 0.38 ± 0.00 0.63 ± 0.22 1.56 ± 0.00 6.25 ± 0.00 2,000.00 ± 0.00 5,333.33 ± 2,309.40 M 03 0.47 ± 0.00 1.88 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.38 ± 0.00 0.63 ± 0.22 1.56 ± 0.00 6.25 ± 0.00 2,000.00 ± 0.00 4,000.00 ± 0.00 M 04 0.47 ± 0.00 1.88 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.25 ± 0.11 0.50 ± 0.22 1.56 ± 0.00 6.25 ± 0.00 1,333.33 ± 577.35 2,666.67 ± 1,154.70 M 05 0.47 ± 0.00 1.88 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.25 ± 0.43 1.50 ± 0.00 0.38 ± 0.00 1.00 ± 0.43 1.56 ± 0.00 6.25 ± 0.00 2,000.00 ± 0.00 5,333.33 ± 2,309.40 M 06 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.25 ± 0.11 0.38 ± 0.00 1.56 ± 0.00 12.50 ± 0.00 1,666.67 ± 577.35 4,000.00 ± 0.00 M 07 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.25 ± 0.43 1.75 ± 1.15 0.63 ± 0.22 0.75 ± 0.00 1.56 ± 0.00 6.25 ± 0.00 2,000.00 ± 0.00 8,000.00 ± 0.00 M 08 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 \>1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.38 ± 0.00 0.38 ± 0.00 1.56 ± 0.00 6.25 ± 0.00 1,666.67 ± 577.35 4,000.00 ± 0.00 M 09 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.38 ± 0.00 0.38 ± 0.00 1.56 ± 0.00 12.50 ± 0.00 1,666.67 ± 577.35 4,000.00 ± 0.00 M 10 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 \>1.00 ± 0.00 1.25 ± 0.43 1.50 ± 0.00 0.63 ± 0.22 0.75 ± 0.00 1.56 ± 0.00 6.25 ± 0.00 1,666.67 ± 577.35 8,000.00 ± 0.00 M 11 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.50 ± 0.00 1.50 ± 0.00 \>3.00 ± 0.00 \>3.00 ± 0.00 1.56 ± 0.00 12.50 ± 0.00 2,000.00 ± 0.00 8,000.00 ± 0.00 M 12 0.47 ± 0.00 3.75 ± 0.00 0.13 ± 0.00 \>1.00 ± 0.00 2.50 ± 0.87 2.50 ± 0.87 \>3.00 ± 0.00 \>3.00 ± 0.00 1.56 ± 0.00 12.50 ± 0.00 4,000.00 ± 0.00 8,000.00 ± 0.00 M 13 0.47 ± 0.00 3.75 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 0.75 ± 0.00 1.00 ± 0.43 0.25 ± 0.11 0.75 ± 0.00 6.25 ± 0.00 12.50 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 14 0.47 ± 0.00 0.94 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.25 ± 0.43 1.25 ± 0.43 0.63 ± 0.22 0.75 ± 0.00 3.13 ± 0.00 12.50 ± 0.00 4,000.00 ± 0.00 8,000.00 ± 0.00 M 15 0.47 ± 0.00 0.94 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.50 ± 0.00 1.50 ± 0.00 \>3.00 ± 0.00 \>3.00 ± 0.00 3.13 ± 0.00 12.50 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 M 16 0.47 ± 0.00 0.94 ± 0.00 0.25 ± 0.00 1.00 ± 0.00 1.50 ± 0.00 1.50 ± 0.00 \>3.00 ± 0.00 \>3.00 ± 0.00 3.13 ± 0.00 12.50 ± 0.00 4,000.00 ± 0.00 8,000.00 ± 0.00 M 17 0.47 ± 0.00 0.94 ± 0.00 0.13 ± 0.00 1.00 ± 0.00 1.00 ± 0.43 1.00 ± 0.43 0.31 ± 0.11 0.75 ± 0.00 3.13 ± 0.00 12.50 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 M 18 0.47 ± 0.00 0.94 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.50 ± 0.00 1.50 ± 0.00 0.19 ± 0.00 0.38 ± 0.00 3.13 ± 0.00 12.50 ± 0.00 2,000.00± 0.00 8,000.00 ± 0.00 M 19 0.47 ± 0.00 1.88 ± 0.00 0.17 ± 0.07 1.00 ± 0.00 1.50 ± 0.00 1.50 ± 0.00 \>3.00 ± 0.00 \>3.00 ± 0.00 1.56 ± 0.00 3.13 ± 0.00 2,000.00± 0.00 8,000.00 ± 0.00 **Mean MRSA (n = 19) ^b^** 0.47 ± 0.00 1.83 ± 0.79 0.16 ± 0.03 \>1.00 ± 0.00 1.33 ± 0.34 1.42 ± 0.32 \>1.07 ± 1.19 \>1.25 ± 1.09 2.22 ± 1.20 9.38 ± 3.45 2,328.95 ± 1,729.87 5,767.54 ± 2,748.78 2.3. Time--Kill Curves ---------------------- The bactericidal effects of MSKE and its phenolic principles against standard MSSA strain are shown in [Figure 1](#molecules-16-06255-f001){ref-type="fig"}A--[Figure 1](#molecules-16-06255-f001){ref-type="fig"}E. The time−kill curves of MSKE and its principles against the M03 MRSA strain are shown in [Figure 1](#molecules-16-06255-f001){ref-type="fig"}F− [Figure 1](#molecules-16-06255-f001){ref-type="fig"}J. {#molecules-16-06255-f001} Using concentrations of 4 and 8 MICs; we showed that for the treatment of standard MSSA ([Figure 1](#molecules-16-06255-f001){ref-type="fig"}A--[Figure 1](#molecules-16-06255-f001){ref-type="fig"}E), MSKE, PGG and GA induced complete cell death within 8, 6 and 2 h, respectively, whereas vancomycin exhibited complete eradication within 4 h. MG was able to kill MSSA cells completely within 24 and 16 h at concentrations of 4 and 8 MICs, respectively. Treatment of M03 MRSA ([Figure 1](#molecules-16-06255-f001){ref-type="fig"}F--[Figure 1](#molecules-16-06255-f001){ref-type="fig"}J) with either MSKE or PGG at 4 and 8 MICs exerted the most bactericidal activity (≥5 log~10~--fold decrease) within 24 h, whereas treatment with MG using the same concentration range resulted in complete cell death within 16 and 8 h, respectively. GA produced no bactericidal effects on the M03 MRSA strain at any time and exerted only a 1--2 log~10~ reduction within 8 h compared with controls ([Figure 1](#molecules-16-06255-f001){ref-type="fig"}I). In the case of vancomycin, a bacteriostatic effect on M03 MRSA was detected by 24 h after incubation at 4 MICs, whereas a bactericidal effect was observed at 8 MICs within 24 h. The ability of MSKE, PGG and GA to kill the standard MSSA strain at 4 and 8 MICs occurred in a time--dependent manner, whereas the bactericidal activity of MG demonstrated a concentration− and time−dependent pattern. Although the killing rate of GA against the standard MSSA strain at 4 and 8 MICs was time−dependent, GA failed to inhibit the growth of the M03 MRSA strain at either concentration, which is not in agreement with the MBC values. This finding implied that the MIC and MBC values resulting from the broth microdilution study that provided a static view may disagree with the results from the time--kill assays that measured the killing rates of the microorganism in a dynamic manner \[[@B9-molecules-16-06255]\]. This phenomenon may also be explained by the survival of adaptive, resistant forms such as those that have been observed after incubating *Enterococcus faecalis* with the chloroform fraction of rhizomes from *Aristolochia paucinervis* Pomel \[[@B10-molecules-16-06255]\]. 2.4. Electron Microscopy ------------------------ Scanning electron microscopy (SEM) images of the standard MSSA strain are shown in [Figure 2](#molecules-16-06255-f002){ref-type="fig"}. The SEM images revealed that MSKE and PGG induced an alteration in cell morphology. Control cells in the presence of 1% DMSO showed a spherical shape in grapelike clusters ([Figure 2](#molecules-16-06255-f002){ref-type="fig"}A). In contrast, cells treated with either MSKE or PGG at a MIC of 4 displayed clusters of non--separated cells ([Figure 2](#molecules-16-06255-f002){ref-type="fig"}B and [Figure 2](#molecules-16-06255-f002){ref-type="fig"}C). {#molecules-16-06255-f002} The SEM images indicated an impaired cell division of the tested microorganisms after treatment with MSKE and were in agreement with the results obtained from transmission electron microscopy (TEM) images, which demonstrated ultra--structural changes in the tested MSSA cells ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}B--[Figure 3](#molecules-16-06255-f003){ref-type="fig"}D) and M09 MRSA cells ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}F--[Figure 3](#molecules-16-06255-f003){ref-type="fig"}H) after treatment with MSKE. At the MIC of MSKE, the separation of daughter cells was severely inhibited as demonstrated by the incomplete septum formation; this phenomenon is demonstrated by the pseudomulticellular appearance and asymmetrical initiation of septum formation that are visible in groups of non--separated cells ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}B and [Figure 3](#molecules-16-06255-f003){ref-type="fig"}F, arrows). In addition, the thickness of the bacterial cell walls was significantly increased following treatment. These damaging effects and the observed ultra--structural changes appeared to be more prominent with the increase in MSKE concentration from 1 to 2 or 4 MICs. The splitting of cell materials to the outer surface of the disrupted membrane ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}C, [Figure 3](#molecules-16-06255-f003){ref-type="fig"}D, [Figure 3](#molecules-16-06255-f003){ref-type="fig"}G and [Figure 3](#molecules-16-06255-f003){ref-type="fig"}H) and the fibrous matrix extending from the surface of the treated cells ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}G, arrow) are obviously seen. The untreated cells of both standard MSSA and M09 MRSA strains showed symmetrical initiation and completed septum formation ([Figure 3](#molecules-16-06255-f003){ref-type="fig"}A and [Figure 3](#molecules-16-06255-f003){ref-type="fig"}E, arrows). These results were similar to the ultra--structural changes observed in the MRSA strains treated with *Quercus infectoria* extract, tannic acid \[[@B11-molecules-16-06255]\] or green tea extract (*Camellia sinensis*) \[[@B12-molecules-16-06255]\]. Since polyphenols are known to form complexes with proteins and polysaccharides \[[@B13-molecules-16-06255]\], bacterial surfaces have the ability to bind large amounts of polyphenols \[[@B14-molecules-16-06255],[@B15-molecules-16-06255]\]. The inhibitory mechanism of MSKE and its phenolic principles against MSSA and MRSA strains may therefore be due to the damage to the bacterial membrane. This leads to permeability of the outer and inner membranes of treated cells and disruption of membranes, resulting in the release of small cellular molecules. This hypothesis is in accordance with the inhibitory mechanism of tea polyphenols towards *Pseudomonas aeruginosa* suggested by Yi *et al.* \[[@B16-molecules-16-06255]\]. {#molecules-16-06255-f003} 2.5. The Enhancing Effects of MSKE and Its Phenolic Principles on the Antibacterial Activity of Penicillin G ------------------------------------------------------------------------------------------------------------ [Table 3](#molecules-16-06255-t003){ref-type="table"} demonstrates the capabilities of MSKE and its phenolic principles to enhance the antibacterial activity of penicillin G. It was found that the addition of MSKE at 0.25 or 0.5 MIC to penicillin G led to a marked reduction in the MIC of penicillin G against the standard MSSA strain from 41.67 ± 18.04 to 3.91 ± 0.00 and 1.95 ± 0.00 µg/mL, respectively. The MIC of penicillin G was also reduced when it was combined with 0.25 and 0.5 MICs of the phenolic principles, showing reductions from 41.67 ± 18.04 to 7.81 ± 0.00 µg/mL at both concentrations for PGG, to 15.63 ± 0.00 and 7.81 ± 0.00 µg/mL, respectively, for GA and to 31.25 ± 0.00 and 15.63 ± 0.00, respectively, for MG. A similar enhancement effect was also observed in the antibacterial activity of penicillin G against 19 clinical MRSA strains when it was combined with 0.25 and 0.5 MICs of MSKE and its phenolic principles. The mean MIC of penicillin G against the 19 MRSA strains at the two concentrations was reduced at least 5--fold from 2,328.95 ± 1,729.87 µg/mL to 361.84 ± 159.45 and 309.21 ± 145.05 µg/mL, respectively, for MSKE, to 409.95 ± 300.04 and 324.42 ± 222.60 µg/mL, respectively, for PGG, to 250.41 ± 439.73 and 9.16 ± 28.08 µg/mL, respectively, for MG and to 244.04 ± 205.36 and 237.46 ± 207.16 µg/mL, respectively, for GA. It was noted that the addition of MG at 0.5 MIC dramatically reduced the mean MIC of penicillin G against the 19 clinical MRSA strains of approximately 254−fold from 2,328.95 ± 1,729.87 to 9.16 ± 28.08 µg/mL. These results suggest that despite the weak antimicrobial effects of MG against MRSA and MSSA, MG may have the capacity to enhance the bacterial susceptibility to penicillin G similarly to the alkyl gallates, which have been shown to dramatically intensify susceptibility to β--lactams in MSSA and MRSA strains \[[@B17-molecules-16-06255]\]. These results imply that the enhancing ability of MSKE on the antibacterial activity of penicillin G against MRSA strains may be due to the major phenolic principle, PGG, as the capacity of PGG to reduce the mean MIC of penicillin G against MRSA strains was close to the capacity of MSKE at both 0.25 and 0.5 MICs. Although MG and GA alone exhibited lower antibacterial activity than PGG, GA and MG at 0.25 and 0.5 MICs in combination with penicillin G demonstrated higher capacities to reduce the mean MIC of penicillin G against MRSA strains than the addition of PGG at the same concentration ([Table 3](#molecules-16-06255-t003){ref-type="table"}). This finding suggests that the enhancing capacity of MSKE and its phenolic principles may not be related only to their antibacterial activity. Moreover, the combination of 0.5 or 1 MIC of MSKE with penicillin G did not enhance the bactericidal activity of penicillin G against the 19 clinical MRSA isolates. Their mean MBCs were 5,767.54 ± 2,748.78 for penicillin G alone, 5,907.89 ± 2,757.52 when 0.5 MIC of MSKE was added and 5,907.89 ± 2,757.52 µg/mL when 1 MIC of MSKE was added ([Table 4](#molecules-16-06255-t004){ref-type="table"}). For the standard MSSA strain, the mean MBC of penicillin G was reduced from 41.67 ± 18.04 to 31.25 ± 0.00 µg/mL with the addition of either 0.5 or 1 MIC of MSKE. This result implies that MSKE can enhance the antibacterial activity of penicillin G against MRSA with a bacteriostatic effect but not a bactericidal effect. However, the enhancing mechanism of MSKE and its phenolic principles on the antibacterial activity of penicillin G against MRSA and MSSA has not been elucidated, but it could be similar to tellimagrandin I and corilagin, which enhanced the antibacterial activity of β−lactam antibiotics by inactivating the penicillin−binding protein and suppressing the activity of β−lactamase \[[@B18-molecules-16-06255]\]. molecules-16-06255-t003_Table 3 ###### The enhancing effects of MSKE and its phenolic principles on the antibacterial activities of penicillin G against *S. aureus* ATCC 25923 and 19 clinical MRSA isolates (^a^ mean values ± S.D. of triplicate results, ^b^ mean values ± S.D.from 19 MRSA strains). Bacterial strains MIC of Penicillin G (µg/mL)^a^ ---------------------------- -------------------------------- ----------------- ---------------- ----------------- ----------------- ----------------- --------------- ----------------- ----------------- ***S. aureus* ATCC 25923** 41.67 ± 18.04 3.91 ± 0.00 1.95 ± 0.00 7.81 ± 0.00 7.81 ± 0.00 31.25 ± 0.00 15.63 ± 0.00 15.63 ± 0.00 7.81 ± 0.00 **Clinical MRSA strains** M 01 2,000.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 62.50 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 02 2,000.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 M 03 2,000.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 1,000.00 ± 0.00 500.00 ± 0.00 125.00 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 04 1,333.33 ± 577.35 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 05 2,000.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 06 1,666.67 ± 577.35 250.00 ± 0.00 250.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 62.50 ± 0.00 1.95 ± 0.00 3.91 ± 0.00 3.91 ± 0.00 M 07 2,000.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 1.95 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 08 1,666.67 ± 577.35 250.00 ± 0.00 250.00 ± 0.00 1,000.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 09 1,666.67 ± 577.35 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 10 1,666.67 ± 577.35 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 11 2,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 12 4,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 1,000.00 ± 0.00 1,000.00 ± 0.00 2,000.00 ± 0.00 3.91 ± 0.00 1,000.00 ± 0.00 1,000.00 ± 0.00 M 13 8,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 3.91 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 M 14 4,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 15 125.00 ± 0.00 62.50 ± 0.00 62.50 ± 0.00 7.80 ± 0.00 7.80 ± 0.00 3.91 ± 0.00 0.06 ± 0.00 3.91 ± 0.00 3.91 ± 0.00 M 16 4,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 17 125.00 ± 0.00 62.50 ± 0.00 62.50 ± 0.00 31.25 ± 0.00 31.25 ± 0.00 3.91 ± 0.00 0.06 ± 0.00 3.91 ± 0.00 3.91 ± 0.00 M 18 2,000.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 62.50 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 M 19 2,000.00 ± 0.00 500.00 ± 0.00 500.00 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 62.50 ± 0.00 3.91 ± 0.00 250.00 ± 0.00 250.00 ± 0.00 **Mean MRSA (n = 19) ^b^** 2,328.95 ± 1,729.87 361.84 ± 159.45 309.21 ±145.05 409.95 ± 300.04 324.42 ± 222.60 250.41 ± 439.73 9.16 ± 28.08 244.04 ± 205.36 237.46 ± 207.16 molecules-16-06255-t004_Table 4 ###### Effects of MSKE and its phenolic principles on the MBCs of penicillin G against *S. aureus* ATCC 25923 and clinical MRSA strains (^a^ mean values ± S.D. of triplicate results, ^b^ mean values ± S.D. from 19 MRSA strains). Bacterial strains MBC of Penicillin G (µg/mL) ^a^ ---------------------------- --------------------------------- --------------------- --------------------- ***S. aureus* ATCC 25923** 41.67 ± 18.04 31.25 ± 0.00 31.25 ± 0.00 **Clinical MRSA strains** M 01 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 02 5,333.33 ± 2,309.40 4,000.00 ± 0.00 4,000.00 ± 0.00 M 03 4,000.00 ± 0.00 4,000.00 ± 0.00 4,000.00 ± 0.00 M 04 2,666.67 ± 1,154.70 4,000.00 ± 0.00 4,000.00 ± 0.00 M 05 5,333.33 ± 2,309.40 8,000.00 ± 0.00 8,000.00 ± 0.00 M 06 4,000.00 ± 0.00 4,000.00 ± 0.00 4,000.00 ± 0.00 M 07 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 08 4,000.00 ± 0.00 4,000.00 ± 0.00 4,000.00 ± 0.00 M 09 4,000.00 ± 0.00 4,000.00 ± 0.00 4,000.00 ± 0.00 M 10 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 11 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 12 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 13 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 14 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 15 125.00 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 M 16 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 17 125.00 ± 0.00 125.00 ± 0.00 125.00 ± 0.00 M 18 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 M 19 8,000.00 ± 0.00 8,000.00 ± 0.00 8,000.00 ± 0.00 **Mean MRSA (n = 19) ^b^** 5,767.54 ± 2,748.78 5,907.89 ± 2,757.52 5,907.89 ± 2,757.52 3. Experimental =============== 3.1. Materials -------------- ### 3.1.1. Test Materials Gallic acid (GA; ≥98% purity) and methyl gallate (MG; ≥98% purity) were purchased from Fluka (Buchs, Switzerland). Pentagalloylglucopyranose (PGG; \>95% purity) was purchased from Endotherm BmbH (Germany). Penicillin G (1651 U/mg) was USP grade and was purchased from Bio Basic Inc. (Canada). Vancomycin for injection (1,055 µg/mg) was obtained from CJ Cheiljedang Corp (Kyunggi−Do, Korea). Reference discs of vancomycin were produced by Oxoid (Basingstoke, UK). Other chemicals and solvents were of analytical grade and obtained from local distributors. ### 3.1.2. Plant Material and Preparation of Plant Extract Fully grown, unripened Thai mango fruits (*Mangifera indica* L. cv. 'Fahlun', Anacardiaceae) were purchased from a local market. The seed kernels were removed and extracted following the method of Nithitanakool *et al.* \[[@B5-molecules-16-06255]\]. Briefly, the kernels were homogenised in hot ethanol (80 °C) and defatted with hexane. After the solvents were evaporated, the remaining aqueous residue was freeze--dried to generate a crude mango seed kernel extract (MSKE) with a yield of 6.69% w/w (on the basis of wet weight). MSKE and its isolated compounds were dissolved in 10% aqueous dimethyl sulfoxide (DMSO) before use. ### 3.1.3. Standardisation The MSKE was standardised with regard to the content of its three polyphenolic compounds (PGG, GA and MG) using the thin layer chromatographic (TLC)−UV densitometric method \[[@B5-molecules-16-06255]\]. Briefly, an aliquot of MSKE (25 mg/mL) was applied to TLC plates together with serial dilutions of the standard solutions of PGG, GA and MG. The TLC plates were then developed in a pre-saturated TLC tank containing CHCl~3~/ethanol/formic acid (3:5:1, v/v/v) as the mobile phase for PGG and CHCl~3~/methanol/ethyl acetate/ethyl methyl ketone/formic acid (6:1.6:2:2:5, v/v/v/v/drop) for GA and MG. The developed TLC plates were scanned using a TLC densitometer at 286 nm, and the amount of each compound in the MSKE (PGG 65.61 ± 0.95%, GA 0.88 ± 0.16% and MG 0.62 ± 0.09% w/w based on dry weight) was calculated from the calibration curves. ### 3.1.4. Bacterial Strains Tested A total of 20 microbial cultures (both standard and clinical strains) were used in this study. Nineteen strains of clinical MRSA isolates cultured from patient samples were kindly provided by Ramathibodi Hospital and Nakhon Pathom Hospital, Thailand. Pathogen purification and identification were confirmed using a microbial identification system at the Department of Microbiology, Faculty of Pharmacy, Mahidol University, Thailand. The standard strain of *S. aureus*,ATCC 25923 (methicillin−sensitive *Staphylococcus aureus*, MSSA), was obtained from the Thailand National Institutes of Health and was used as a control strain. The microorganisms were maintained in a mixture of tryptic soy broth (TSB; Becton Dickinson & Co., France) and 30% w/v glycerol at −80 °C until use. For experiments, all of the bacterial strains were grown separately on tryptic soy agar (TSA; Becton Dickinson & Co., France) at 37 °C for 18--24 h. The isolated bacterial colonies of actively growing cultures from agar plates were transferred to a test tube with TSB and incubated at 37 °C for 24 h in shaker. The culture turbidity was adjusted spectrophotometrically at 600 nm to obtain an optical density (OD) of 0.2 (approximately 10^6^--10^7^ CFU/mL) before use as an inoculum in the antimicrobial susceptibility test. 3.2. Antimicrobial Susceptibility Test -------------------------------------- ### 3.2.1. Disc Diffusion Method The antimicrobial activity of the MSKE was screened for its inhibitory activity by the disc diffusion test following the procedures recommended by the Clinical and Laboratory Standards Institute (CLSI) with slight modifications \[[@B19-molecules-16-06255]\]. Briefly, the following concentrations of MSKE were prepared by serial dilution in 10% aqueous DMSO: 500, 250, 125 and 62.5 mg/mL. Each prepared inoculum was swabbed on TSA and air−dried at room temperature (25 °C). A 6-mm sterile paper disc was loaded with 10 μL of the prepared MSKE solution, and the disc was placed on the agar plate. The plates were left to dry and then were incubated at 37 °C for 24 h under aerobic conditions. A negative control was prepared using the same solvents employed to dissolve the MSKE, and vancomycin (30 µg/disc) was used as a positive control. All disc diffusion tests were performed in triplicate, and the antibacterial activity was expressed as the mean of the inhibition diameter (mm). ### 3.2.2. Determination of the MICs and MBCs The broth microdilution method was used to determine the MIC of MSKE and its phenolic principles. Serial two--fold dilutions of test compounds were mixed with TSB at a 1:1 ratio (v/v) in 96-well sterile microtitre plates to obtain final concentrations of 0.06--3.75 mg/mL for MSKE, 0.02--1.00 mg/mL for PGG and 0.05--3.00 mg/mL for GA and MG. Then, 50 μL of a 1:5 dilution of the prepared inoculum was added to TSB supplemented with the test compounds to obtain a 100 μL final volume in each well. The microtitre plates were then incubated at 37 °C overnight under aerobic conditions. In each test, the following controls were used: (1) a negative control including the test sample but not the organism; and (2) a positive control without the test sample but containing the organism. Vancomycin and penicillin G were used as reference standards. The MIC was defined as the lowest concentration at which no bacterial growth was observed by the unaided eye. The amount of growth in the wells containing test samples was compared with the amount of growth in the control wells when determining the growth endpoints. To establish the MBC, 20 µL of each culture medium was removed from wells with no visible growth and inoculated on TSA plates. After aerobic incubation at 37 °C overnight, the number of surviving organisms was determined. The MBC was defined as the lowest concentration that produced a complete suppression of visible colony growth. Each sample was tested in triplicate in separate experiments. The enhancing effects of MSKE and its phenolic principles on the antibacterial activity of penicillin G were evaluated by the broth microdilution method. The MIC and MBC values of penicillin G were determined in combination with 0.25, 0.5 or 1 MIC of the test compounds. ### 3.2.3. Time−Kill Assay The bactericidal activities of MSKE and its phenolic principles were determined according to the time--kill assay of Chusri *et al.* \[[@B20-molecules-16-06255]\] with slight modification. The bacterial suspension (100 µL, approximately 10^6^ CFU/mL) was added to TSB (900 µL) containing the test sample at 4 and 8 MICs. After incubation at 37 °C, sample (100 µL) was collected at different time intervals (0, 2, 4, 6, 8, 12, 16 and 24 h) and a ten--fold serial dilution was prepared in sterile saline (0.9% w/v NaCl). Thereafter, 25 µL of each dilution was placed on a TSA plate and incubated at 37 °C for 24 h. A count for viability was performed, and the number of CFU/mL was recorded. A bacterial growth control was included in each assay and consisted of 1% DMSO without the addition of test samples. Vancomycin was also used as the reference antibiotic. All experiments were carried out at least in duplicate. Time--kill curves were constructed by plotting log~10~ CFU/mL against time. Bactericidal activity was defined as a ≥3 log~10~--fold decrease in the number of survivors at each time point compared with the number inoculated at time zero. This activity was equivalent to 99.9% killing of the inoculum \[[@B9-molecules-16-06255]\]. 3.3. Electron Microscopy ------------------------ ### 3.3.1. SEM The prepared inoculum (200 µL) of *S. aureus* ATCC 25923 was transferred into TSB (1.8 mL) containing MSKE or PGG at a MIC of 4. Bacterial growth controls were performed with the addition of 1% DMSO without the test samples. After the suspensions were incubated at 37 °C for 12 h, bacterial cells were collected by centrifugation at 3,000 rev/min for 10 min. Samples were then fixed in 2.5% w/v of glutaraldehyde at 4 °C for at least 2 h. The cells were washed with 0.1 M phosphate buffer solution (PB, pH 7.2) and postfixed in 1% w/v osmium tetroxide in 0.1 M PB for 1--2 h. Cells were dehydrated using serial concentrations of ethanol (35, 50, 70, 95, and 100%). After critical point drying and coating with gold sputter, samples were examined using a Hitachi S501 scanning electron microscope. ### 3.3.2. TEM The prepared inoculum (200 µL) of *S. aureus* ATCC 25923 and the M09 MRSA strain were transferred into TSB (1.8 mL) supplemented with MSKE at 1, 2 or 4 MICs. Bacterial growth controls were performed with the addition of 1% DMSO without the test samples. The suspensions were incubated at 37 °C for 12 h. The bacterial cells were then harvested by centrifugation at 3,000 rev/min for 10 min. The samples were fixed in 2.5% w/v of glutaraldehyde at 4 °C for at least 2 h and postfixed in 1% w/v osmium tetroxide in 0.1 M PB for 1--2 h. The bacterial cells were dehydrated using serial concentrations of ethanol (35, 50, 70, 95, and 100%) and embedded in Spurr's resin. The samples were cut with an ultramicrotome (LKD, Sweden) and stained with uranyl acetate and lead citrate. The ultrathin sections were examined under a transmission electron microscope (JEM−200CX, JEM−2100, JEOL, Japan). 3.4. Statistical Analysis ------------------------- All experimental results were expressed as mean ± standard deviation (S.D.). All statistical analyses were carried out using SPSS (version 16.0 for Windows). Analysis of variance was performed by ANOVA. Significant differences between the means were determined using Tukey's pairwise comparison test at a significance level of *p* \< 0.01. 4. Conclusions ============== MSKE and its phenolic principles exhibited potent inhibitory effects against the standard MSSA strain and clinical MRSA isolates. PGG, the major phenolic principle of MSKE, appeared to be the major contributor to the inhibitory potency of MSKE. Damaging effects on the cell membrane that led to the alteration in cell morphology and interference with bacterial division were possible inhibitory mechanisms. MSKE at 0.25 and 0.5 MICs displayed a remarkable capacity to enhance the antibacterial activity of penicillin G by lowering the MIC of penicillin G by 10−20 fold. Despite the weak antimicrobial effects of MG against MRSA and MSSA, MG demonstrated the capacity to enhance the bacterial susceptibility to penicillin G of approximately 254−fold. Together, these results indicate that MSKE may potentially be useful as an alternative natural therapeutic agent or as an adjunctive therapy along with penicillin G against MRSA infections. Financial supports from the Thailand Research Fund (TRF) through the Royal Golden Jubilee Ph.D. Program and Mahidol University (Grant No. PHD/0313/2550) and the 60th Year Supreme Reign of His Majesty King Bhumibol Adulydej Scholarship of the Faculty of Graduate Studies, Mahidol University are gratefully acknowledged. *Sample Availability:* Samples of the compounds are available from the authors. The authors declare no conflict of interest.

All relevant data are within the paper and its Supporting Information files. Introduction {#sec005} ============ Many patients experienced acute postoperative pain, and approximately 86% had moderate, severe, or extreme pain. Although pain management system has been fully established in most institutions, yet there were still many patients complaining of inadequate pain control after acute pain care \[[@pone.0205959.ref001]\], which would affect quality of life, functional recovery, and increase risks of complications and postoperative chronic pain \[[@pone.0205959.ref002]\]. Sufentanil, one of the opioids with strong analgesic potency, is now widely used for surgery patients during anesthesia, has also been extensively used in postoperative pain control and labor pain relief, specifically in patient controlled epidural analgesia (PCEA) \[[@pone.0205959.ref003]--[@pone.0205959.ref005]\]. Although PCEA is preferred in certain types of surgery than patient controlled intravenous analgesia (PCIA) because of better pain relief and less side effects \[[@pone.0205959.ref006]\], but no long-termed outcome was superior to PCIA \[[@pone.0205959.ref007]\]. Besides to the association to postoperative hypotension and delay of urinary catheter removal, actually, epidural analgesia is technically difficult in certain cases and even contradicted to some cases. Abundant observations on the PCEA using sufentanil combined with local anesthetics have been conducted \[[@pone.0205959.ref008]\], but there are very few reports about PCIA using sufentanil. With appropriate administration regimen, sufentanil PCIA would achieve great satisfaction like other opioids. Inspired by the use of sufentanil in target controlled infusion (TCI) \[[@pone.0205959.ref009]\], we proposed that intravenous PCA with continuous background infusion would have a steady effect-site concentration compared to intravenous PCA with only PCA bolus dose which could cause fluctuation of plasma concentration. Sufentanil TCI consists of a priming dose, a decreasing infusing rate and the constant maintaining rate. Then, a loading dose followed by a constant continuous infusion could mimic the TCI regimen of sufentanil. Moreover, the PCA bolus dose with constant background infusion could only cause slight fluctuation of the effect-site concentration of sufentanil. Therefore, we designed this study to determine if there was a certain dose of sufentanil better than other doses in relieving postoperative pain using PCIA in patients underwent moderate surgeries. The primary endpoint of the study was pain scale VAS. Methods {#sec006} ======= Study design {#sec007} ------------ This is a prospective, single center, randomized study with a three arm parallel group design. No changes were made to methods after trial commencement. Ethical approval {#sec008} ---------------- This trial was registered at ClinicalTrials.gov (NCT02503826). The study was conducted in accordance with the principles of Good Clinical Practice and the Declaration of Helsinki, approved by the Ethics Committee of the Second Affiliated Hospital of Xi'an Jiaotong University, Xi'an, China (NO. 005 2015, Chairperson Prof. Yang Dechang) on February 5, 2015. Written informed consent was obtained from all patients or a legal surrogate. It was a single center, randomized, double-blind, three arm parallel group study. It was conducted at the Second Affiliated Hospital of Xi'an Jiaotong University between February 11, 2015 and December 8, 2016. The authors confirm that all ongoing and related trials for this drug/intervention are registered. The delay in registering this study is due to: a. failed contact to the organization\'s PRS administrator to request a user login to the website for the registration (we requested repeatedly but had no reply), b. we had to finally apply a PRS account in person, c. we were not very familiar with the registration process on the website. Patient population {#sec009} ------------------ Patients diagnosed with gastrointestinal cancer, aging from 20 to 75, BMI 18 to 28 kg/m\^2, American Society of Anesthesiologists (ASA) grade I to II, anticipated surgery duration within 4 hours and agreed to sign the consent paper were scheduled for elective abdominal surgery including gastrectomy, colectomy, and rectectomy. Women participants in the study had to either be menopause or use routine contraceptive method. Patients with known allergy or contradiction to the treatment drugs, or severe respiratory, cardiovascular or neurological diseases, hepatic or renal dysfunction, psychiatric history or unstable mental state were excluded from the study. Patients with history of drug or alcohol abuse, chronic use of opioids and pregnancy or breast-feeding were excluded as well. Conduct of the study {#sec010} -------------------- Patients were enrolled according to the inclusion and exclusion criteria by our nurse. All patients received general anesthesia. No premedication was given. Ten minutes before anesthesia induction a loading dose of dexmedetomidine with 1 μg/kg was infused over 10 minutes. Then anesthesia was induced with midazolam 2 mg, sufentanil 0.5 μg/kg, propofol 1--2 mg/kg. Cisatracurium 0.2 mg/kg was given to facilitate orotracheal intubation with a cuffed tube. Anesthesia was maintained with continuous infusion of propofol 3--7 mg/kg/h, remifentanil 0.1--0.25 μg/kg/min, cisatracurium 0.1 mg/kg/h and dexmedetomidine 0.2--0.5 μg/kg/h with positive pressure ventilation in a circle system. And the bispectral index (BIS) value was maintained between 40 and 60. Mean blood pressure (MAP), Heart rate (HR), pulse oxymetry (SpO~2~), end-tidal CO~2~ (ETCO~2~), blood loss, and transfusion were recorded during anesthesia. Cisatracurium and dexmedetomidine were discontinued until peritoneum closure, while propofol and remifentanil were stopped until the last stitch of skin. Patients were enrolled to our study only when the surgery duration was within 4 hours. The eligible patients were randomly assigned to one of three treatment arms (A, B, C). The randomization was generated by permuted block randomization with block size 6 manually. Briefly, there should be 2 patients assigned to A treatment, 2 assigned to B treatment and 2 assigned to C treatment in each block. Fifteen permutations and combinations were selected by a statistician based on random number table. The allocation sequence of patients was determined by the sequence of surgery. The allocation marks were sealed in 90 sequentially numbered envelopes and could only be opened 30 min before the end of surgery by the nurse who was responsible for the preparation of the pump. After checking up with the pump preparation and allocation mark, the allocation mark was concealed in envelop again by the nurse. And the allocation mark remained sealed till complete of last follow-up and start of data input. No changes were made for blinding. The PCA pump was prepared with different doses of sufentanil and 10 mg tropisetron. The dose per body weight of sufentanil in the intravenous PCA pump was determined by which treatment arm the patient was allocated to, with 1.5 μg/kg sufentanil in arm A, 2.0 μg/kg sufentanil in arm B and 2.5 μg/kg sufentanil in arm C. The total volume of the pump was 100 ml. Immediately after surgery, the PCA pump was attached to the peripheral venous line by the investigator who was unaware of the formula of the PCA pump or the treatment arm. A loading dose of 10 μg sufentanil was administered right at the end of the surgery. And then 5 μg at a time according to the patients' complaint of pain untill the pain relieved or the respiratory rate (RR) was less than 10 breaths per minute. The emergence and extubation time was recorded and emergence time was defined as the time from anesthesia discontinuation to eye opening on command. The background infusion of PCIA was set at 2 ml/h, the bolus volume of each PCA press was 0.5 ml and lockout interval was 10 min. All patients were monitored in the post-anesthesia care unit (PACU) for at least 30 min until the discharge criteria were met. All the patients were followed-up 2, 6, 24 and 48 hours after the surgery. Clinical evaluations including MAP, HR, RR, SpO~2~, total pressing times, total consumption of sufentanil, rescue analgesic requirement, visual analogue scale (VAS), numerical rating scale (NRS) and Ramsay sedation scale (RSS) score were recorded. Side effects including nausea and vomiting, urinary retention, pruritus, respiratory depression were also recorded. Overall satisfaction index of the patients was recorded as well. The overall study duration of each patient was less than 5 days, including the screening period (1 or 2 days before surgery), the treatment period (from surgery day to 48 hours after surgery) and the following-up period (from the end of the surgery to 48 hours after surgery). Patients, anesthesiologist, and data collectors involved in the study were unaware of which treatment arm the patients would be in. The double blind method was secured by putting no tag on the PCIA pump. Rescue analgesic of 5 μg sufentanil was administered intravenously by the investigators when the patient had maximal pressing times per unit time and the pain score was still above 5 at rest for more than 30 min. The efficacy of the PCIA pump was assessed based on the patients' pain level of VAS and NRS at the time points of 2, 6, 24, 48 hours after surgery. VAS: the patient was presented with a horizontal 10-cm line (from 0 = no pain to 10 = worst pain) and was asked to mark the spot which indicated the current pain level. NRS: the patient was asked to describe the current pain level through an 11 points numerical scale from 0 = no pain to 10 = worst possible pain. The safety of the PCIA pump was evaluated based on the incidence, intensity, seriousness, and causality of pump-related adverse events (AEs) and the frequency of clinically significant changes in physical examination, HR, BP, laboratory safety tests (hematology, biochemistry, and urinalysis), and 5-lead ECG. Outcomes {#sec011} -------- The primary endpoint was pain level assessed by VAS up to 48 hours after surgery. Secondary endpoints included NRS score of pain, nausea and vomiting, degree of sedation, hypotension, pruritus, urinary retention, respiratory depression and rescue analgesia. No changes were made to trial outcomes after the trial commenced. Statistics {#sec012} ---------- In order to find out the appropriate dose of sufentanil in PCIA after moderate surgery, the primary endpoint pain level assessed by VAS was analyzed. The expected standard deviation of means was 6.9 mm \[[@pone.0205959.ref010]\], and standard deviation of subjects was 20 mm \[[@pone.0205959.ref011]\]. The significance level was set at 0.05 and the power at 0.8. Then sample size assumption was made through the software program *PASS 11* (NCSS, LLC, Kaysville, Utah, USA) based on one-way ANOVA. And the calculated sample size was 29 in each group and we included 30 patients in each group. Comparisons of VAS or NRS for pain were performed by linear mixed model among groups using IBM SPSS Statistics 19.0. Ranked data of sedation degree and vomiting and nausea scale were analyzed with the Kruskal--Wallis tests. Probability values under 0.05 were considered significant. No interim analysis was planned and conducted. Results {#sec013} ======= The participant flow {#sec014} -------------------- Ninety patients in our hospital were assigned to three treatment arms ([Table 1](#pone.0205959.t001){ref-type="table"}). The enrollment started on 11 February 2015 and completed on 5 December 2016, and the last follow-up was on 8 December 2016. The participants received the intended treatment with the randomized group tag (A, B, or C) by time sequence and were analyzed for the primary outcome ([Fig 1](#pone.0205959.g001){ref-type="fig"}). Patient data was collected in the operation room and in ward. {#pone.0205959.g001} 10.1371/journal.pone.0205959.t001 ###### Treatment arms. {#pone.0205959.t001g} Assignment group Preparations for PCA pump ------------ -------- ------------------------------------------------------------ A SF 1.5 sufentanil 1.5 μg/kg, 10 mg tropesitron, diluted to 100 ml B SF 2.0 sutentanil 2.0 μg/kg, 10 mg tropesitron, diluted to 100 ml C SF 2.5 sufentanil 2.5 μg/kg, 10 mg tropesitron, diluted to 100 ml Patient demographics {#sec015} -------------------- The mean age of all patients was 60 (range 26--75) yrs old. There were 30 females and 60 males. Patient characteristics and baseline data were compared and no difference among three treatment arms ([Table 2](#pone.0205959.t002){ref-type="table"}). Other related factors like anesthesia duration, surgery duration, surgical types, dose of sufentanil for induction, remifentanil and dexmedetomidine consumptions were comparable among arms as well ([Table 2](#pone.0205959.t002){ref-type="table"}). 10.1371/journal.pone.0205959.t002 ###### Patients' baseline characteristics and operational data. {#pone.0205959.t002g} ----------------------------------------------------------------------------------- Group SF 1.5 SF 2.0 SF 2.5 ------------------------------ --------------- --------------- --------------- ---- Overall participants 30 30 30 Age, years 60 (10) 61 (9) 60 (10) Gender, Female/Male 11/19 10/20 9/21 Height, cm 164.4 (7.06) 165.3 (7.40) 165.7 (7.61) Weight, kg 63.2 (7.90) 64.7 (9.30) 62.45 (10.3) Hemoglobin, g/L 125.8 (15.09) 124.4 (14.76) 123.8 (21.17) Creatinine, μmol/l 68.87 (12.39) 65.76 (10.78) 65.49 (11.32) WBC Count, 10\^9 cells/L 6.00 (1.99) 5.53 (1.25) 6.38 (1.98) Alanine Transaminase, u/l 15.0 (8.49) 18.3 (9.56) 18.5 (12.71) Blood Urea Nitrogen, mmol/l 4.87 (1.10) 5.63 (1.59) 5.00 (1.91) Blood glucose, mmol/l 5.0 (0.75) 4.9 (0.66) 4.9 (0.75) Anesthesia duration, minutes 173 (29.6) 167 (39.0) 182 (39.5) Surgery duration, minutes 143 (33.3) 140(37.2) 150 (39.2) Emergence time,minutes 8.6 (3.55) 9.9 (5.56) 10.6 (6.85) Time to extubation, minutes 12.2 (4.00) 14.1 (7.68) 13.9 (7.09) Preoperative SBP, mmHg 145(19) 143 (18) 143 (20) Preoperative HR, beats/min 78 (14) 78 (14) 79 (13) Remifentanil consumption, mg 1.44 (0.46) 1.41 (0.57) 1.50 (0.58) DEX consumption, μg 81 (25) 74 (27) 80 (26) Surgery type, participants stomach 21 17\ 16 colon 5 8 9 rectum 4 5 5 ----------------------------------------------------------------------------------- All data is presented as Mean (Standard Deviation) except for gender and surgery types. WBC, White Blood Cell. SBP, systolic blood pressure. HR, heart rate. DEX, dexmedetomidine. 1\. The postoperative pain assessment The postoperative pain of VAS and NRS was analyzed by mixed linear model. The included factor was group, subjects was set as random effect, repeated effect was time and covariance structure was unstructured. The fixed effect of group for VAS at rest was significant (*P*\<0.001), as well as the time effect (*P* = 0.001) and the interaction effect (*P =* 0.033). VAS scores at rest in both group SF 2.0 and SF 2.5 were decreased compared with group SF 1.5 respectively ([Fig 2](#pone.0205959.g002){ref-type="fig"}, *P*\<0.001). But no difference between group SF 2.0 and group SF 2.5 was detected (*P* = 0.392). The fixed effect of group for VAS at activity was significant (*P*\<0.001), as well as the time effect (*P* = 0.001). But no interaction effect was detected (*P =* 0.114). VAS at activity was decreased in group SF 2.0 compared with group SF 1.5 ([Fig 2](#pone.0205959.g002){ref-type="fig"}, *P*\<0.001). And VAS at activity in group SF 2.5 was decreased as compared with group SF 2.0 ([Fig 2](#pone.0205959.g002){ref-type="fig"}, *P* = 0.038). The fixed effects of group for NRS at rest was significant (*P*\<0.001), as well as the time effect (*P*\<0.001), but not the interaction effect (*P* = 0.138). NRS scores at rest in both group SF 2.0 and SF 2.5 were decreased compared with group SF 1.5 respectively ([Fig 3](#pone.0205959.g003){ref-type="fig"}, *P*\<0.001), while there was no difference between group SF 2.0 and SF 2.5 (*P* = 0.694). The fixed effect of group for NRS at activity was significant (*P*\<0.001) as well as the interaction effect (*P* = 0.028). But no time effect was detected (*P =* 0.980). NRS scores at activity in both group SF 2.0 and SF 2.5 were decreased compared with group SF 1.5 respectively ([Fig 3](#pone.0205959.g003){ref-type="fig"}, *P*\<0.001). But no difference between group SF 2.0 and group SF 2.5 was detected (*P* = 0.214). In order to further clarify the difference between group SF 2.5 and SF 2.0, we analyzed patients with VAS or NRS scores at activity less than 5.0 \[[@pone.0205959.ref012]\], and it turned out that in group SF 2.5 there were more patients with VAS or NRS scores at activity less than 5.0 than that in group SF 2.0 at 24 hours after surgery ([Table 3](#pone.0205959.t003){ref-type="table"}). {#pone.0205959.g002} {#pone.0205959.g003} 10.1371/journal.pone.0205959.t003 ###### Pain score at activity less than 5.0 at 24 hours after surgery. {#pone.0205959.t003g} ****Group**** Pain scale at activity less than 5 at 24 hours, participants --------------- -------------------------------------------------------------- -------- SF 2.0 17/30 18/30 SF 2.5 25/29 25/29 *P* value 0.0078 0.0391 2\. The rescue analgesia requests The rescue analgesia demanding was less in arm B and C when compared with arm A. It was measured by the number of patients who asked for rescue analgesia and the total times of rescue analgesia for each patient ([Table 4](#pone.0205959.t004){ref-type="table"}). 10.1371/journal.pone.0205959.t004 ###### Rescue analgesia among arms. {#pone.0205959.t004g} Group Participants times -------- ------------------------------------------- ------------------------------------------- SF 1.5 13/30 17/60 SF 2.0 5/30[\*](#t004fn001){ref-type="table-fn"} 6/60[\*](#t004fn001){ref-type="table-fn"} SF 2.0 3/30[\*](#t004fn001){ref-type="table-fn"} 4/60[\*](#t004fn001){ref-type="table-fn"} \* *P*\<0.05 compared with SF 1.5. 3\. The opioid related side effects The number of patients with opioid-related sedation after surgery showed no difference among arms ([S1 Table](#pone.0205959.s005){ref-type="supplementary-material"}). Patients with RSS scores less than 5 were mentioned in the data and no patient required rescue medication for sedation. Postoperative nausea and vomiting was documented as well. The number of patients suffering from nausea was no different among arms ([S1 Table](#pone.0205959.s005){ref-type="supplementary-material"}). No patient suffered from postoperative vomiting in this study. No respiratory depression was observed, which was defined as respiratory rate less than 10 or SPO~2~ less than 92% when breathing air ([S1 Table](#pone.0205959.s005){ref-type="supplementary-material"}). Other side effects related to sufentanil including dizziness, pruritus, headache and constipation were not observed. Discussion {#sec016} ========== In this study we reported the use of sufentanil in postoperative intravenous patient controlled analgesia (PCIA) with continuous background infusion for the first time. Meanwhile we explored the optimal dose of sufentanil for PCIA in abdominal surgeries from three incremental doses. As to PCA pump parameters, we set background infusion rate at 2 ml/h, bolus volume at 0.5 ml and the lockout time at 10 min. The dose of sufentanil in PCA pump was calculated by the body weight of the patient and the parameters were set based on the sufentanil TCI system of Gepts \[[@pone.0205959.ref013]\]. After infusion of 10 μg sufentanil, also known as the loading dose, the effect-site concentration would reach to 0.2 ng/l in 6.5 minutes and drop to 0.05 ng/L about 0.5 h later. While the effect-site concentration 0.05 ng/l could be achieved by constantly infusing sufentanil at 0.05 μg/kg/h, which happened to be the background infusion rate in arm C. Each bolus volume of 0.5 ml in arm C would increase the effect-site concentration by 0.01 ng/l within 5 minutes. Background infusion is not recommended when morphine or fentanyl is used in PCA pump because it won't improve pain relief and may increase the risk of delayed respiratory depression. The latter is mostly due to the accumulation of opioids during continuous infusion \[[@pone.0205959.ref014],[@pone.0205959.ref015]\]. And M6G, the active metabolites of morphine, is partly responsible for respiratory depression caused by morphine \[[@pone.0205959.ref016]\]. Sufentanil infusion could be adequately described by a linear three-compartmental mammillary model which makes it eligible for target controlled infusion with stable effect-site concentration \[[@pone.0205959.ref013],[@pone.0205959.ref017]\]. A reliable target controlled infusion regimen contains a priming dose, a decreasing infusing rate and a constant maintaining rate \[[@pone.0205959.ref017]\]. Given the linear pharmacokinetics of sufentanil \[[@pone.0205959.ref013]\], minimal active metabolites with no clinical relevance \[[@pone.0205959.ref018]\] and its use in target controlled infusion (TCI), the background infusion of sufentanil in PCIA will produce steady effect-site concentration which might effectively alleviate pain with less side effects. In this study, only a few side effects were observed and this might attribute to the steady plasma concentration of sufentanil. Moreover, no severe side effects of sufentanil occurred even in geratic patients (20 of 90 participants older than 70 yr, 6/8/6 in each arm respectively, [S2 Table](#pone.0205959.s006){ref-type="supplementary-material"}). No deep sedation with the Ramsay scale score less than 4 occurred, no respiratory depression and no vomiting. Nausea was only observed in a few cases. And the low incidence of postoperative nausea and vomiting in this study might attribute to tropisetron prophylaxis during anesthesia and a large dose of tropisetron in PCIA pump formula \[[@pone.0205959.ref019]\]. It has been reported that the wake-up concentration of sufentanil is around 0.1 ng/ml or higher \[[@pone.0205959.ref020]\]. Using the TCI system of Gepts, we reckoned that the effect-site concentration of sufentanil in arm C was around 0.05--0.11 ng/ml, which could be achieved with a constant infusion rate of 0.05--0.175 μg/kg/h. Other researchers have observed that patients can be well sedated with sufentanil and midazolam in ICU. When a large dose of sufentanil combined with midazolam was administered, the plasma concentration of sufentanil would be at least 0.3 ng/ml at which patients were still arousable \[[@pone.0205959.ref021]\]. And in prolonged TCI for postoperative analgesia after cardiac surgery, the concentration of sufentanil was around 0.08--0.1 ng/ml \[[@pone.0205959.ref022]\]. With lower plasma concentration of 0.05--0.11 ng/ml in arm C in this study, side effects like postoperative sedation might be less than in the studies mentioned above in which the concentration of sufentanil was higher. It might explain why no patients had deep sedation and no sedation needed antagonist in this study. However, this estimation in the study needs to be confirmed by more solid evidence with evaluation of the effect-site concentration. A report about the comparison of continuous infusion and bolus administration of sufentanil during anesthesia for cardiac valve surgery showed that there was no benefit in continuous infusion except for the simplicity in clinical practice. The continuous infusion group consumed 26% more of sufentanil than bolus administration, while both groups had stable hemodynamics \[[@pone.0205959.ref023]\]. In our study, background infusion might consume more sufentanil when compared with bolus only. But this needs to be verified because with very large dose of sufentanil administered after cardiac valve surgery, the difference between continuous infusion and bolus might be more significant. Moreover, other evidence showed the combination of TCI and PCA using hydromorphone could provide satisfied postoperative pain relief \[[@pone.0205959.ref024]\]. With our study we have shown that in simulation of TCI, sufentanil PCIA with background infusion can be regarded as an effective alternative for postoperative pain control. Even though there was no difference in pain relief at rest in arm B and arm C, we considered that dose 2.5 μg/kg would be the preferred optimal dose. On one hand with 2.5 μg/kg sufentanil more patients experienced pain relief when they were at activity (pain intensity score less than 5 when coughing) 24 hours after surgery. On the other hand, with 2.5 μg/kg sufentanil it didn't increase side effects compared to 2 μg/kg sufentanil. Theoretically, a higher dose of sufentanil will lead to more significant pain relief because sufentanil has no ceiling effect, as long as the side effects of sufentanil are acceptable. In our study the patients were encouraged to walk in the ward after abdominal surgery as early as possible, and most patients could manage to walk at about 24 hours postoperatively. Therefore, we only observed the pain intensity at activity 24 hours and 48 hours after surgery. The efficacy of sufentanil PCIA was proved and the safety of its use was preliminarily testified. However, how sufentanil would influence the recovery of patients after 48 hours was not discussed in this study, including physical recovery, the gastrointestinal function and hospital stays. Another question was whether the development of chronic postoperative pain would be influenced by sufentanil PCIA. Finally, in light of sufentanil TCI we found that sufentanil PCIA with continuous background infusion could provide effective pain relief with less side effects. The optimal dose of sufentanil was 2.5 μg/kg in this study, which is more preferred to the other two lower doses. This study has several limitations. For a dose-effect study, 30 subjects each group is enough. But if patient safety is to be evaluated, a study with more cases and more centers might be more convincing. The biases of the study were inevitable. The sample size calculation was conducted by a software program *PASS 11* on basis of one-way ANOVA but not the linear mixed model. This would cause incorrect power. We run the linear mixed model analysis and both the effects of treatment and time were significant except for the interaction. Other biases may come from the education backgrounds, personalities and pain sensitivities of all subjects. Supporting information {#sec017} ====================== ###### CONSORT checklist of the study. (DOC) ###### Click here for additional data file. ###### English version of study protocol. (DOCX) ###### Click here for additional data file. ###### Original version of study protocol. (DOC) ###### Click here for additional data file. ###### Raw data files. (RAR) ###### Click here for additional data file. ###### Side effects related to postoperative analgesia among arms. (DOC) ###### Click here for additional data file. ###### Age distribution of old patients. (DOC) ###### Click here for additional data file. The authors are very grateful to Prof. Zhao Yaqin and Prof. Li Fang in the department for their full clinical supports. [^1]: **Competing Interests:**The authors have declared that no competing interests exist.

Macular atrophy (MA) is a common occurrence during treatment of neovascular age-related macular degeneration (nAMD) with intravitreal anti-vascular endothelial growth factor (VEGF) therapy.^[@i1552-5783-58-14-6038-b01][@i1552-5783-58-14-6038-b02]--[@i1552-5783-58-14-6038-b03]^ Like typical geographic atrophy (GA) occurring in non-nAMD, MA associated with anti-VEGF therapy for nAMD is characterized by the loss of the outer retina, retinal pigment epithelium (RPE), and choriocapillaris (CC).^[@i1552-5783-58-14-6038-b02],[@i1552-5783-58-14-6038-b04][@i1552-5783-58-14-6038-b05]--[@i1552-5783-58-14-6038-b06]^ Some authors have noted that, compared with typical GA, MA occurring in the setting of anti-VEGF therapy is often smaller in size but with a more diffuse distribution.^[@i1552-5783-58-14-6038-b06]^ Zanzottera et al.^[@i1552-5783-58-14-6038-b07],[@i1552-5783-58-14-6038-b08]^ showed histologic differences in RPE morphology, basal laminar deposit (BLamD), and the descent of the external limiting membrane toward Bruch\'s membrane at the atrophy border in eyes with non-neovascular versus those with nAMD. The occurrence of macular atrophy in treated eyes is a common cause of poor long-term visual function following initial short-term visual gains.^[@i1552-5783-58-14-6038-b01],[@i1552-5783-58-14-6038-b09],[@i1552-5783-58-14-6038-b10]^ Development of prevention and treatment strategies for MA will be enabled through a better understanding of its pathophysiology and risk factors. Decreased central choroidal thickness,^[@i1552-5783-58-14-6038-b11],[@i1552-5783-58-14-6038-b12]^ presence of type 3 neovascularization (NV),^[@i1552-5783-58-14-6038-b01],[@i1552-5783-58-14-6038-b13],[@i1552-5783-58-14-6038-b14]^ and pre-existing MA^[@i1552-5783-58-14-6038-b01]^ in the fellow eye are known to herald the appearance of MA during anti-VEGF treatment. Atrophic macular changes have been noted in the eyes of mice with genetically downregulated RPE-derived VEGF^[@i1552-5783-58-14-6038-b15]^ and the deleterious effect of anti-VEGF drugs on VEGF production by RPE has been suspected in humans.^1,\]6^ Subretinal drusenoid deposits (SDD), often referred to by their reticular pattern seen on en face images as reticular pseudodrusen, have been linked to MA.^[@i1552-5783-58-14-6038-b11],[@i1552-5783-58-14-6038-b16],[@i1552-5783-58-14-6038-b17]^ SDD are extracellular conglomerates of unesterified cholesterol and various types of proteins located between RPE and photoreceptors^[@i1552-5783-58-14-6038-b18]^ that are visualized on optical coherence tomography (OCT) and en face imaging modalities in two morphologically distinctive forms, "dots" and "ribbons."^[@i1552-5783-58-14-6038-b19]^ SDD have been described in the settings of outer retinal ischemia and loss of photoreceptors, RPE, and choriocapillaris.^[@i1552-5783-58-14-6038-b20][@i1552-5783-58-14-6038-b21]--[@i1552-5783-58-14-6038-b22]^ The cause and effect of these lesions are yet to be ascertained. While some researchers consider SDD to be a marker of outer retinal ischemia resulting from choroidal hypoxia due to local^[@i1552-5783-58-14-6038-b21]^ or systemic vascular factors,^[@i1552-5783-58-14-6038-b23]^ others propose that SDD themselves are able to disrupt metabolic pathways needed for photoreceptor and RPE survival resulting in their loss and subsequent CC atrophy.^[@i1552-5783-58-14-6038-b20]^ Despite the similarities in conditions under which MA and SDD are observed, relationships between these two entities remain poorly understood. The goal of this study is to explore the relationship between SDD morphology and localization and MA development in patients undergoing intravitreal anti-VEGF therapy for nAMD. Methods {#s2} ======= This study design was approved by Western institutional review board (Olympia, WA, USA). It complied with the Health Insurance Portability and Accountability Act of 1996 and followed the tenets of the Declaration of Helsinki. Participants {#s2a} ------------ Data collection and imaging analysis were described in detail previously.^[@i1552-5783-58-14-6038-b14]^ In brief, data from 82 consecutive patients (88 eyes) were retrospectively collected. Among those, 71 patients (74 eyes) met the inclusion criteria below. All patients were followed by a single physician (KBF) at two offices of the Vitreous Retina Macula Consultants of New York (New York, NY, USA). Inclusion criteria for this study were as follows: (1) 50 years and older, (2) any untreated NV lesion type in one or both eyes at baseline, (3) administration of any anti-VEGF agent (ranibizumab, bevacizumab, aflibercept) on a continuous treat-and-extend regimen (TER) for at least 2 years before this review, and (4) availability of eye-tracked spectral-domain (SD) OCT B-scans, near-infrared reflectance (NIR), color digital fundus photographs (CFP), and fluorescein angiography (FA) at baseline and SD-OCT B-scans and NIR at the last follow-up visit. Exclusion criteria were as follows: (1) any previous treatment for NV (photodynamic therapy, intravitreal steroids or anti-VEGF, thermal laser), (2) presence of GA, MA secondary to NV, subfoveal fibrosis, RPE tears, and NV secondary to maculopathies other than AMD at baseline, and (3) the spherical equivalent exceeding the range of ± 2 diopters (D). Basic demographics (age, sex, and race), as well as information about smoking status (ever or never smoker), and comorbidities (cardiovascular disease, hypertension, diabetes, hypercholesterolemia/hyperlipidemia) were collected as part of routine self-reported medical history at baseline. Imaging Modalities {#s2b} ------------------ Eye-tracked SD-OCT images were obtained with the Spectralis HRA + OCT (Heidelberg Engineering, Inc., Franklin, MA, USA). Macula volumes were centered over the fovea with at least 19 horizontal B-scans across an area of 20° × 15°. NIR (λ = 830 nm) images with 30° field of view centered over the macula were acquired with the confocal scanning laser ophthalmoscope (Spectralis HRA + OCT). These images were analyzed with the AutoRescan function of the Spectralis software that allowed spatial point-to-point correlation between SD-OCT and NIR. CFP with 50° view centered over the macula were taken with the Topcon TRC ×501 fundus camera (Topcon Imagenet, Tokyo, Japan). FA images were obtained with either the confocal scanning laser ophthalmoscope (Spectralis HRA + OCT) or the Topcon TRC ×501 fundus camera. All images were acquired after pupil dilation. Image Processing and Grading {#s2c} ---------------------------- The grading of NV lesion types, MA detection, and the measurement of subfoveal choroidal thickness (SCT) were assessed at baseline and described in detail previously.^[@i1552-5783-58-14-6038-b14]^ In brief, the anatomic classification of NV was defined as type 1 (sub-RPE), type 2 (subretinal), or type 3 (intraretinal/retinal angiomatous proliferation) based on grading of SD-OCT and FA images at baseline.^[@i1552-5783-58-14-6038-b14],[@i1552-5783-58-14-6038-b24]^ When multiple NV types were present in one eye, the lesion was classified as "mixed" type, but with each constituent type specified. MA was considered present when: (1) on NIR there was a hyperreflective area with a sharp border spanning at least 250 μm in the maximum linear dimension and not adjacent to peripapillary chorioretinal atrophy, and (2) on SD-OCT this area corresponded to the degeneration of RPE and outer retina with hypertransmission into the choroid.^[@i1552-5783-58-14-6038-b14],[@i1552-5783-58-14-6038-b25]^ MA grading was performed at both baseline and follow-up visits. Using calipers available in the SD-OCT review software, SCT was measured under the fovea at a single location between the outer border of RPE/Bruch\'s membrane complex and the choroidal-scleral junction on the foveal SD-OCT B-scan at both baseline and follow-up visits.^[@i1552-5783-58-14-6038-b12]^ The type of NV and MA area were assessed by two independent graders blinded to patients\' data followed by an open adjudication with a senior grader (KBF) in cases of disagreement. Separate analyses were performed for three categories of SDD: SDD of any phenotype, ribbon phenotype SDD, and dot phenotype SDD. The grading of SDD and their phenotypes were assessed using SD-OCT, NIR, and CFP based on the multimodal approach described previously.^[@i1552-5783-58-14-6038-b26],[@i1552-5783-58-14-6038-b27]^ In brief, lesions were required to be confirmed by SD-OCT and at least one en face modality or shown by two en face modalities, when lesions were outside the SD-OCT volume. The characteristics of SDD and their two phenotypes were assessed with previously described definitions.^[@i1552-5783-58-14-6038-b19],[@i1552-5783-58-14-6038-b26]^ Briefly, dots form sharp peaks or domes of hyperreflective material in the subretinal space seen on SD-OCT that correspond to discrete yellowish-gray and hyporeflective dots on CFP and NIR, respectively. Ribbons form broad or rounded hyperreflective elevations in subretinal space seen on SD-OCT that correspond to interlocking yellowish-gray and faint hyporeflective ribbons on CFP and NIR, respectively. Five or more lesions of a particular phenotype detected within one of the prespecified locations described below were required for such a phenotype to be considered present in this location.^[@i1552-5783-58-14-6038-b19]^ For SDD of any phenotype category, ≥5 lesions of any phenotype had to be present in one such location.^[@i1552-5783-58-14-6038-b27],[@i1552-5783-58-14-6038-b28]^ For the purposes of this study, we identified locations based on the retinal anatomy: macula or extramacular region.^[@i1552-5783-58-14-6038-b29]^ The macula was defined as a 6000-μm diameter circle centered at the fovea as determined by the central horizontal SD-OCT B-scan. The extramacular region was defined as the area outside of this 6000-μm diameter circle. The macula and extramacular regions were divided into "superior" and "inferior" areas by the foveal horizontal SD-OCT B-scan ([Fig. 1](#i1552-5783-58-14-6038-f01){ref-type="fig"}). Thus, a total of four retinal locations were assessed for lesions by the graders: superior macula, superior extramacular, inferior macula, and inferior extramacular. SDD were considered to be present at the eye-level when SDD of any phenotype were present within ≥1 of the prespecified locations. SDD of the ribbon or dot phenotype were considered to be present at the eye-level if SDD of the respective phenotype were present within ≥1 of the prespecified locations. {#i1552-5783-58-14-6038-f01} SDD grading was performed at baseline by two independent graders (AVZ) and (OG-O), who were blinded to the MA status at follow-up. The senior grader (KBF) evaluated in cases of disagreement after an open adjudication. Statistical Analysis {#s2d} -------------------- Baseline characteristics were compared between those with and without SDD using generalized estimating equations (GEE) to account for the within-person correlation. Logistic regression models using GEE were used to evaluate the association between the presence of SDD and the development of MA, and crude and adjusted odds ratios and their 95% confidence intervals (CI) were calculated. Models were adjusted for age alone, then were further adjusted for neovascularization type and choroidal thickness. The odds of MA development was also compared between presence or absence of each SDD phenotype (ribbons and dots) and presence or absence of SDD in each location (superior and inferior macula, superior and inferior extramacula).^[@i1552-5783-58-14-6038-b30]^ All statistical analyses were performed using SAS, version 9.3 software (SAS, Cary, NC, USA). *P* \< 0.05 were considered statistically significant. Results {#s3} ======= A total of 88 eyes of 82 patients with untreated nAMD at baseline were screened for the current study. Among these eyes, 14 of 88 (16%) had MA at baseline and were excluded from further analysis. The remaining 74 eyes (71 patients) met the inclusion criteria and constituted the study population for the current analysis. Demographic and clinical characteristics of eyes at baseline are summarized in [Table 1](#i1552-5783-58-14-6038-t01){ref-type="table"}. The study group was predominantly white (97%) with female patients comprising 73% of all patients. The mean age of patients was 81 years (range, 52--98 years). Of subjects, 39% were smokers and 23% reported a history of cardiovascular disease. At baseline, most eyes (76%) had type 1 (38%) or type 3 (38%) neovascular lesions, with smaller percentages of type 2 (8%) and mixed (16%) lesions. SDD were present in 63% (46/73 eyes). The mean duration of follow-up was 4.69 ± 1.15 years with a mean of 7.14 ± 2.72 injections per year. ###### Demographic and Clinical Characteristics of Patients With nAMD at Baseline; Treatment Characteristics and Macular Atrophy Status at Follow-up in the Overall Study Group and Groups Stratified by SDD Status  New MA had developed in 51% (38/74) of eyes at the most recent follow-up. Those with SDD at baseline were more likely to develop MA at follow-up compared with those without SDD (63% vs. 30%, *P* = 0.0069). Patients with SDD were more likely to be older (*P* = 0.0199) and have thinner SCT (*P* = 0.0009). Additionally, distribution of NV types varied significantly between eyes with and without SDD (*P* = 0.0455). As prior reports have described an association between thin choroids and MA,^[@i1552-5783-58-14-6038-b11],[@i1552-5783-58-14-6038-b12]^ and between type 3 NV and MA,^[@i1552-5783-58-14-6038-b01],[@i1552-5783-58-14-6038-b13]^ we repeated the analysis using these parameters as binary variables and confirmed their significance (*P* = 0.0212 and *P* = 0.0103, respectively). The lowest SCT quartile was considered pathologic and measured \<118 μm, in line with prior reports defining abnormally thin SCT.^[@i1552-5783-58-14-6038-b31],[@i1552-5783-58-14-6038-b32]^ Sex, race, smoking status, and history of hypertension, cardiovascular disease (CVD), diabetes, and hypercholesterolemia/hyperlipidemia were not statistically different by SDD presence. The comparison of clinical characteristics of eyes at baseline with and without SDD is summarized in [Table 1](#i1552-5783-58-14-6038-t01){ref-type="table"}. We further analyzed the relationship between SDD presence and location with MA development, adjusting for the identified confounders ([Table 2](#i1552-5783-58-14-6038-t02){ref-type="table"}). Eyes with SDD at baseline were 3.0 times more likely to develop MA at follow up when compared with eyes without SDD at baseline (odds ratio \[OR\] 3.0, 95% confidence interval \[CI\] 1.1--8.5, *P* = 0.0343). Eyes with SDD present in the inferior macula and inferior extramacular field at baseline were 3.0 and 6.5 times more likely to develop MA at follow-up compared with eyes without SDD in these locations at baseline (OR 3.0, 95% CI 1.0--8.9, *P* = 0.0461 and OR 6.5, 95% CI 1.3--32.4, *P* = 0.0218, respectively). The association between SDD presence in the superior macula and the superior extramacular field and MA development was not statistically significant (OR 1.9, 95% CI 0.7--5.3, *P* = 0.2193 and OR 2.1, 95% CI 0.7--6.4, *P* = 0.1928, respectively). Distribution of SDD across the four analyzed fields is shown in [Figure 2](#i1552-5783-58-14-6038-f02){ref-type="fig"}. ###### Crude and Adjusted Association Between the Presence of SDD and Their Locations at Baseline and MA Development at Follow-up  {#i1552-5783-58-14-6038-f02} In the overall study group, the presence of SDD with combined ribbon and dot phenotypes was 38% (28/73 eyes). Only 7% (5/73) and 18% (13/73) eyes had isolated ribbon and dot phenotypes, respectively. When analyzing the relationship between SDD phenotype and MA development using the same multivariate regression model ([Table 3](#i1552-5783-58-14-6038-t03){ref-type="table"}), the positive association between the presence of SDD of the ribbon phenotype and MA development was not statistically significant (OR 1.6, 95% CI 0.5--4.5, *P* = 0.4105). Similarly, the positive association between SDD of the dot phenotype and MA was also not statistically significant (OR 1.9, 95% CI 0.7--5.3, *P* = 0.1946). ###### Crude and Adjusted Association Between SDD Phenotypes at Baseline and MA Development at Follow-up  Discussion {#s4} ========== We found that while only a minority (16%) of eyes with treatment-naïve nAMD had MA at baseline, 51% of eyes lacking MA at baseline developed MA while receiving anti-VEGF therapy on a TER over a mean follow-up of 4.7 years. This high incidence of MA and its associated loss of visual function is of great concern to clinicians and their patients^[@i1552-5783-58-14-6038-b06],[@i1552-5783-58-14-6038-b10],[@i1552-5783-58-14-6038-b13],[@i1552-5783-58-14-6038-b33],[@i1552-5783-58-14-6038-b34]^ and has generated significant interest in the ophthalmology research community.^[@i1552-5783-58-14-6038-b15],[@i1552-5783-58-14-6038-b35][@i1552-5783-58-14-6038-b36]--[@i1552-5783-58-14-6038-b37]^ We confirm the previously reported association between MA and SDD,^[@i1552-5783-58-14-6038-b11],[@i1552-5783-58-14-6038-b16],[@i1552-5783-58-14-6038-b17]^ in which the baseline presence of SDD conferred a 3-fold increase of odds of subsequent MA development independent of other factors including age, intraretinal neovascularization, and thin choroid. We ascribe special importance to this association and think that understanding the nature of SDD may shed light on the pathophysiology of MA and explain the link between the two entities. A growing body of histopathologic and observational clinical studies lends support to the RPE origin of SDD. In particular, SDD are thought to result from the derangement of intraretinal lipid metabolism^[@i1552-5783-58-14-6038-b18]^ that to a large extent involves RPE cells^[@i1552-5783-58-14-6038-b38]^ and have been found to be compositionally similar to and frequently accompany the RPE-derived soft drusen, basal linear deposits, and BLamD^[@i1552-5783-58-14-6038-b18],[@i1552-5783-58-14-6038-b39][@i1552-5783-58-14-6038-b40]--[@i1552-5783-58-14-6038-b41]^; SDD are dynamic structures^[@i1552-5783-58-14-6038-b42]^ and their more advanced stages exhibit slower growth rate^[@i1552-5783-58-14-6038-b43]^; SDD fade out over NV and atrophic areas^[@i1552-5783-58-14-6038-b44],[@i1552-5783-58-14-6038-b45]^ characterized by RPE degeneration.^[@i1552-5783-58-14-6038-b46]^ It seems reasonable to suggest that the RPE is the cellular structure that links SDD and MA development. Several theories have been proposed explaining the relationship between these entities, but with only circumstantial evidence available in their support, all are debatable. According to the proponents of the vascular etiology of MA development, VEGF suppression in the settings of nAMD therapy reduces choriocapillaris permeability with subsequent impaired oxygen and nutrient delivery to the RPE.^[@i1552-5783-58-14-6038-b01]^ RPE ischemia may also result from a "steal" phenomenon created by the diversion of blood flow from the choriocapillaris to new pathologic blood vessels^[@i1552-5783-58-14-6038-b04]^ or vascular disease recognized by choroidal thinning that was found to be associated with both MA and SDD presence.^[@i1552-5783-58-14-6038-b11]^ While no specific mechanism has been proposed to explain these observations, RPE ischemia/hypoxia may lead to upregulation of VEGF production in an attempt to improve endothelial cell permeability and rescue RPE cells by paracrine and autocrine mechanisms, respectively.^[@i1552-5783-58-14-6038-b35]^ This, however, may theoretically shift the energy balance away from the phagocytic, lipid, and retinol recycling and other functions of RPE and result in SDD formation in particular. Iatrogenic neutralization of VEGF by anti-VEGF therapy could reduce this rescue mechanism leading to both endothelial and RPE cell loss and subsequent MA. However, in an animal model of RPE ischemia resulting from the sustained systemic VEGF neutralization, direct endothelial toxicity was short-lived and mitigated by increased VEGF production by an otherwise unaffected RPE.^[@i1552-5783-58-14-6038-b35]^ Discovery of complement fragments and other immune system constituents within SDD^[@i1552-5783-58-14-6038-b39]^ led some to believe that inflammation plays a part in their formation.^[@i1552-5783-58-14-6038-b44]^ Indeed, the single nucleotide polymorphism Y402H of the complement factor H gene results in the increased presence of macrophage attack complexes (MAC) at the interface of Bruch\'s membrane and choriocapillaris.^[@i1552-5783-58-14-6038-b47]^ While MAC are thought to aid in clearance of cellular debris under normal conditions, their increased presence is thought to induce "bystander injury" to the surrounding cells and has been associated with choroid thinning,^[@i1552-5783-58-14-6038-b47]^ in which situation SDD may be a byproduct of inflammation generated by complement activation. Also, aging RPE express progressively more immunogenic amyloid precursor protein with subsequent increase in activation of MAC even without high-risk polymorphisms and subsequent sub-RPE drusen formation.^[@i1552-5783-58-14-6038-b48]^ In fact, aging itself has been proposed to be the major determinant of RPE cell function relating to the transport of the cholesterol to and from photoreceptors and choriocapillaris.^[@i1552-5783-58-14-6038-b49]^ SDD may thus be related to the changes in RPE milieu detrimental to its proper functioning. Intravitreal anti-VEGF drugs or *VEGF* gene downregulation may further attenuate the VEGF signal to endothelial cells inducing angiopathic changes^[@i1552-5783-58-14-6038-b50]^ that, in turn, could lead to ischemia and tissue loss manifesting as MA. Although almost any permutation of the above theories may explain the pathogenesis of MA and SDD, RPE appears to play the critical role instigating injury to the involved tissues. Further, we propose that SDD with their specific morphology, easily recognizable on different imaging modalities, may serve as a clinical biomarker of RPE dysfunction that may help predict eyes at greater risk for MA. We found that the presence of SDD in the inferior macula and the inferior extramacular field at baseline conferred a 3.0 and 6.5 increased odds, respectively, of developing MA during the anti-VEGF treatment. Based on the published studies^[@i1552-5783-58-14-6038-b18],[@i1552-5783-58-14-6038-b45],[@i1552-5783-58-14-6038-b51]^ and our observations, SDD are most frequently present in the upper fundus with subsequent downward advancement into the inferior macula and outward advancement into the superior and inferior extramacular fields. Interestingly, this pattern of SDD distribution is reminiscent of the topography of rods,^[@i1552-5783-58-14-6038-b52]^ the age-dependent loss of which is exacerbated by AMD^[@i1552-5783-58-14-6038-b53]^ with characteristic cellular and subcellular RPE changes^[@i1552-5783-58-14-6038-b54]^ and is possibly linked to RPE dysfunction.^[@i1552-5783-58-14-6038-b55],[@i1552-5783-58-14-6038-b56]^ In contrast, RPE architecture and function in healthy aging eyes remains stable throughout the lifespan.^[@i1552-5783-58-14-6038-b57]^ The extent of the SDD spread across the retinal fields is, therefore, likely proportional to the duration and extent of RPE dysfunction and may confer a higher risk of MA development. The significance of SDD pattern of distribution is further supported by the higher risk of late AMD development associated with SDD located outside, but not within the macula.^[@i1552-5783-58-14-6038-b58]^ The association between a specific SDD phenotype and MA development was not statistically significant. This could be explained by the insufficient sample size of our eye cohort and the considerable (∼60%) overlap between the dot and ribbon phenotypes, which could conceal a potential effect of SDD phenotype that was of smaller magnitude relative to SDD localization. Notwithstanding, different SDD phenotypes merit recognition, as at least two studies showed that confluent ribbon phenotype was associated with GA development and dot phenotype was associated with nAMD.^[@i1552-5783-58-14-6038-b30],[@i1552-5783-58-14-6038-b59]^ Similar to some,^[@i1552-5783-58-14-6038-b16],[@i1552-5783-58-14-6038-b26],[@i1552-5783-58-14-6038-b41]^ but not all^[@i1552-5783-58-14-6038-b60][@i1552-5783-58-14-6038-b61]--[@i1552-5783-58-14-6038-b62]^ prior studies, none of the other demographic or clinical characteristics, including female sex, smoking, hypertension, CVD, and diabetes were associated with SDD presence. Our findings, therefore, lend further support to the idea that the pathogenesis of SDD is driven primarily by some local mechanisms and less dependent of systemic factors. Strength and Limitations {#s4a} ------------------------ To the best of our knowledge, our systematic categorization of SDD phenotype and location represents the first attempt to relate these parameters to the development of MA. A multimodal approach and standard definitions were rigorously applied by two independent graders to the evaluation of NV lesion types, presence of MA, SDD, and their phenotypes.^[@i1552-5783-58-14-6038-b19],[@i1552-5783-58-14-6038-b24],[@i1552-5783-58-14-6038-b26],[@i1552-5783-58-14-6038-b27],[@i1552-5783-58-14-6038-b63][@i1552-5783-58-14-6038-b64][@i1552-5783-58-14-6038-b65]--[@i1552-5783-58-14-6038-b66]^ The availability of an extended follow-up period allowed us to accurately capture incident MA.^[@i1552-5783-58-14-6038-b25]^ Our work has limitations typical of a retrospective clinic-based study. With the relatively small number of participants and a predominance of females, it is challenging to ascertain sex differences in SDD prevalence within subgroups. The infrequency of isolated ribbon or dot phenotype of SDD could lead to type I error (i.e., failure to detect the significant association of SDD phenotype and MA). Because SD-OCT scans were largely unavailable outside the macula, we were limited to en face imaging modalities for SDD detection, however, other authors employed similar approach to SDD detection outside the SD-OCT volume.^[@i1552-5783-58-14-6038-b27]^ Although characterization of spatial relationship between SDD and MA would further inform our understanding of the relationship between these two entities, we could not evaluate SDD adjacency to MA areas at the last follow-up visit because not all patients had CFP required for identifying ribbon SDD^[@i1552-5783-58-14-6038-b67]^ which were identified through multimodal imaging. The growing body of knowledge on SDD established their association with clinically relevant outcomes, such as MA development and early through late AMD incidence and progression.^[@i1552-5783-58-14-6038-b11],[@i1552-5783-58-14-6038-b16],[@i1552-5783-58-14-6038-b17],[@i1552-5783-58-14-6038-b30],[@i1552-5783-58-14-6038-b44],[@i1552-5783-58-14-6038-b58][@i1552-5783-58-14-6038-b59]--[@i1552-5783-58-14-6038-b60],[@i1552-5783-58-14-6038-b68][@i1552-5783-58-14-6038-b69]--[@i1552-5783-58-14-6038-b70]^ There is an important need for building consensus on an appropriate SDD classification system that would define the phenotype, distribution, size, and other SDD parameters similar to the way other manifestations of the RPE dysfunction (e.g., soft drusen and pigmentary changes)^[@i1552-5783-58-14-6038-b71]^ are defined and currently used in the Age-Related Eye Disease Study classification system. This would allow the future studies to accurately describe the natural history of SDD and their phenotypes, their influence on the evolution early through late AMD, effects of anti-VEGF drugs on SDD development and regression, and their effects on the evolution of nAMD-related MA. Further experimental models targeting specific RPE functions would be valuable in establishing the causative relationship between RPE dysfunction and retinal diseases. In conclusion, we confirm that MA frequently develops in eyes during anti-VEGF treatment. SDD are independently associated with MA development and potentially share a common pathophysiology stemming from RPE dysfunction. The extension of SDD into the inferior fundus, particularly in the inferior extramacular field, appears to be related to subsequent MA development. Histologic studies and accurate SDD and MA models are needed to further refine the biologic relationship between these two clinically important entities. Supported by grants from the National Institutes of Health (R01AG04212; Bethesda, MD, USA), the Macula Foundation, Inc. (New York, NY, USA), Research to Prevent Blindness (New York, NY, USA), the EyeSight Foundation of Alabama (Birmingham, AL, USA), and the Dorsett Davis Discovery Fund (Birmingham, AL, USA). Disclosure: **A.V. Zarubina**, None; **O. Gal-Or**, None; **C.E. Huisingh**, None; **C. Owsley**, None; **K.B. Freund**, Optovue (C), Optos (C), Heidelberg Engineering (C), Genentech (C), Graybug Vision (C), Genentech/Roche (F)

Introduction {#Sec1} ============ The in vitro micronucleus assay is a robust platform for the assessment of chromosomal damage following the treatment of genotoxic agents. In this assay, a quantitative measure of the induced chromosomal damage (chromosomal breaks and chromosomal loss) is acquired by scoring micronuclei (MN) (Fenech [@CR9]). These events can be detected following mitosis, where the lost or broken chromosome resides in the cytoplasm, and not the nucleus. Traditionally, MN scoring is carried out manually by using bright field or fluorescent microscopy. However, the manual scoring procedure has been scrutinised for its subjectivity and extensive scoring time (Doherty et al. [@CR7]; Seager et al. [@CR17]). To overcome these issues, efforts have been made to automate the MN scoring platform. These include the use of both the semi-automated and the fully automated MN scoring approaches that are compatible with multi-endpoint MN analysis and high-through scoring (Bryce et al. [@CR2]; Varga et al. [@CR20]). Commercially available platforms such as the Litron Laboratories automated flow cytometric platform (MicroFlow^®^) and the semi-automated image analysis platform (Metafer™ and Pathfinder™) are among the most widely used MN scoring procedures. The Metafer™ MN scoring platform is often used in the pharmaceutical industry and in academia to assess the genotoxic potential of various DNA damaging agents, and it shows a good concordance with conventional MN scoring platform (Chapman et al. [@CR5]). The MicroFlow^®^ MN scoring platform is proposed as a viable alternative to the manual scoring to conduct objective, multi-parametric MN scoring, with reduced data acquisition time using flow cytometry. Furthermore, the incorporation of nuclear stains ethidium monoazide (EMA) allows discrimination of apoptotic bodies and necrotic cells from MN which can be difficult to define manually, and re-probing with pan nuclear stain SYTOX green following cell lysis provides precision MN scoring (Avlasevich et al. [@CR1]). Even so, it is likely that chromatin from a certain fraction of early-stage apoptotic cells may not always be excluded from analysis based on EMA staining. Also, cells with multiple MN and multi-nucleated cells with MN would be scored differently from lysed (nuclei) preparations compared with intact cells. We predict that both of these situations would tend to result in somewhat higher flow cytometry-based MN frequencies relative to microscopy. The aim of the present study was to assess the reproducibility of the MN dose responses generated with the MicroFlow^®^ and Metafer™ systems as compared to traditional manual scoring. For this purpose, human lymphoblastoid TK6 cells were treated with a clastogen (MMS), an aneugen (carbendazim) and a DNA damaging agent (ochratoxin A), with the cells scored using the three different approaches. Methods and materials {#Sec2} ===================== Chemicals {#Sec3} --------- Methyl methanesulfonate (CAS no. 12925), carbendazim (CAS no. 10605-21-7) and ochratoxin A (CAS no. 303-479) were purchased from Sigma-Aldrich, UK. Cell lines and treatment {#Sec4} ------------------------ Human lymphoblastoid TK6 cells were obtained from American Type Culture Collection (ATCC), Manassas, VA, USA. TK6 cells were cultured in RPMI 1640 media (Gibco, Paisley, UK), supplemented with 1% pen-strep and 10% heat inactivated horse serum (Gibco, Paisley, UK). Cells were seeded at 2 × 10^5^ cells in 25-cm^2^ flask (Fisherbrand), incubated at 37 °C for either 4 or 30 h (1.5--2 cell cycles) in the presence of MMS, carbendazim and ochratoxin A (OTA). Subsequently, the treatment was removed and the cells were harvested following 0- or 26-h recovery period. Resulting MN was scored in the absence of cyto-B by using the Metafer™ (MetaSystems, Althlussheim, Germany) and the MicroFlow^®^ (Litron laboratories, Rochester, USA) platforms. The manual scoring procedure was used as a validation tool to verify the results between the MicroFlow^®^ and the Metafer™ scoring procedures. Cytotoxicity and cytostasis {#Sec5} --------------------------- Cell counts were determined using a Coulter counter (Beckman Coulter Inc.). Relative population doubling (RPD) was used to estimate the highest cytotoxic concentration. MN scoring was restricted to the concentration that induced 50% cell death and cytostasis. The RPD calculation is described in detail elsewhere (Lorge et al. [@CR15])$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\% {\text{RPD}}\;{ = }\;\frac{{{\text{Number}}\;{\text{of}}\;{\text{population}}\;{\text{doubling}}\;{\text{in}}\;{\text{treated}}\;{\text{cultures}}}}{{{\text{Number}}\;{\text{of}}\;{\text{population}}\;{\text{doubling}}\;{\text{in}}\;{\text{the}}\;{\text{vehicle}}\;{\text{control}}}}\; \times 100$$\end{document}$$ Population doubling (PD) was calculated as follows:$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{PD }} = {\text{ Log }}\left( {{\text{Cell count after treatment}}/{\text{cell count in the control}}} \right)/{ \log }2.$$\end{document}$$ The manual scoring procedure {#Sec6} ---------------------------- Cells were harvested following 4- or 30-h treatment. Briefly, treated cells were transferred to 15-ml centrifuge tubes and were centrifuged at 200×*g* for 10 min. Supernatant was aspirated, and the pellet was re-suspended in 10 ml phosphate-buffered saline (Gibco^®^). Subsequently, the cell suspension was cytospun (Cytospin™ centrifuge) on a polished glass slides, fixed in 90% ice cold methanol for 10 min and were air-dried at room temperature. Air-dried slides were stained in 4% Giemsa solution (VWR International Ltd., Poole, UK) at room temperature. Giemsa stained slides were washed under tap water and air-dried, and a cover slip was mounted on these slides using DPX mounting solution. Mononucleated cells with intact nuclear and cytoplasmic membrane were considered suitable for MN identification. The parameters used for MN scoring were size (between 1/3rd and 1/16th the diameter of nuclei), morphology (circular or oval) and their association with the main nuclei (not linked or overlapping the nuclei) (Fenech et al. [@CR10]). The MN scoring was carried out by using 20× magnifications on a light microscope (Olympus BX 51). The MN frequency was obtained by manually assessing 2000 mononucleated cells per replicate. A total of 6000 mononucleated cells were scored using the manual scoring platform. Metafer™ analysis {#Sec7} ----------------- Cells were harvested post-treatment. At the time of harvest, treated cells were transferred to 15-ml centrifuge tubes (Fisherbrand) and centrifuged at 200×*g* for 10 min. Supernatant was aspirated, and the pellet was re-suspended in hypotonic solution 5% KCl (KCL, 75 Mm; Sigma-Aldrich). The cell suspension was centrifuged, supernatant was removed, and the pellets were fixed in 5 ml of Fix 1 \[methanol/acetic acid/NaCl (5:1:6)\] for 10 min at room temperature. Fix2 (methanol/acetic acid 5:1, Fisher Scientific) was used to re-suspend the pellet following centrifugation. Cells were incubated in Fixative 2 for 10 min at room temperature and centrifuged at 4 °C, 200×*g* for 10 min. These pellets were re-suspended in Fixative 2 and stored overnight at 4 °C. For Metafer™ analysis, 100 μl of cell suspension was dropped on to a polished glass slide. Slides were then air-dried, and 20 µl of 4,6-diamidino-2-phenylindole (Vector Laboratories, Peterborough, UK) was use to label nuclei and MN. A cover slip was mounted, and slides were incubated for 15 min at room temperature. Subsequently, the MN induction was assessed using a semi-automated Metafer™ MN scoring platform (Meta System, Althlussheim, Germany). The Metafer MN scoring platform consists of a motorised slide loading platform, Carl Zeiss Axio Imager fluorescence microscope and a charge-coupled device (CCD) camera. Image acquisition was carried out by using Metafer 4 software (version 3.9.8). Stained slides were loaded on to a motorised slide scanning platform of Metafer system. Slides were scanned; images of nuclei and MN were captured with 10× objective. A 100× objective was used for MN scoring by relocating the cell and MN on the slide form the coordinates displayed in the gallery view. Non-overlapping, DAPI stained circular/oval nuclei with a size between 1/3rd and 1/16th of the main nuclei were scored as MN (Fenech et al. [@CR10]). A total of 18,000 mononucleated cells were assessed to enumerate MN frequency. The MicroFlow^®^ approach {#Sec8} ------------------------- Total 5 × 10^5^ treated cells were transferred to 15-ml centrifuge tube and were centrifuged at 300×*g* for 5 min. The supernatant was aspirated, and the pellets were incubated on ice for 20 min. The cells were stained with ethidium monoazide (EMA) following 30-min photo-activation. During this incubation period, the cells were placed on ice 2 cm below the source of light. This process was used to label cell with compromised cytoplasmic membrane. The fold change in EMA-positive events was used alongside %RPD to estimate increased cytotoxicity and to predict highest test concentration. A greater than fourfold increase in EMA-positive event was used as an indicator of increased apoptosis/necrosis (Bryce et al. [@CR4]). The cytoplasmic membrane and the cellular RNA were digested by using detergents and RNase solution following photo-activation step. Subsequently, the nuclei and MN were labelled with SYTOX Green stain. Stained samples were then incubated overnight prior to flow cytometric analysis. Flow cytometric scoring {#Sec9} ----------------------- Prior to the flow cytometric assessments, the suspension of sequentially stained nuclei and MN was incubated at room temperature for 30 min. Samples were acquired on a flow cytometer (BDFACS Aria, BD Biosciences, USA) equipped with 488-nm laser, and BD FACS Diva software (version 6.1.3) was used for MN scoring. EMA-associated fluorescence collected in the FL3 channel was used to monitor increased levels of apoptotic/necrosis. Scoring of nuclei and MN was limited to the cells that displayed SYTOX-associated fluorescence signals in FL1 channel. With the MicroFlow^®^ approach, the viable mononucleated cells were detected from their SYTOX Green-associated fluorescence, DNA content as determined by side scatter and size based on the forward scatter characteristics. For an event to be classified as MN with the MicroFlow^®^ approach, the MN should not be labelled with EMA, exhibit SYTOX Green fluorescence between 1/10th and 1/100th for the main nuclei and should fall in the side and forward scatter regions (Bryce et al. [@CR2]). A total of 24,000 events that displayed SYTOX intensities were used to enumerate MN frequency. Statistical analysis {#Sec10} -------------------- Shapiro--Wilk normality test, Bartlett test or homogeneity of variance and Bonferroni test for outlier identification were conducted. Data were transformed in order to achieve normally distributed data and homogeneity of within-dose variance. If the raw or transformed data passed these trend tests, then the 1-sided Dunnett's test was used to identify the no-observed and the lowest observed genotoxic effect levels (NOGEL, LOGEL) and if the data failed these trend tests, then the 1-sided Dunnett's test was used (Johnson et al. [@CR14]). Covariate benchmark dose (BMD) analysis was carried out using PROAST (v60.12) to compare dose responses (Slob [@CR18]). This approach relies on constant shape parameters for log-steepness and maximum response being used between each independent dose response, which provides increased precision for each dose response and allows for potency ranking to be carried out (Soeteman-Hernández et al. [@CR19]; Wills et al. [@CR21], [@CR22]). In this instance, it was carried out to observe any trends in equipotency or not between the chemicals and MN scoring approach. Overlapping BMDs show that equipotency cannot be rejected and non-overlapping BMDs show that there is a difference. Furthermore, when there is no response at the concentrations tested, conserved shape information from the other responses is used to fit suitable models to allow for BMDL to be derived but with infinite BMDU. Results {#Sec11} ======= Cytotoxicity and cytostasis {#Sec12} --------------------------- The 50 ± 5% reduction in percentage RPD is a standardised method to estimate highest test concentration for accurate MN enumeration (OECD [@CR16]). The fold change in EMA-positive events alongside percentage RPD was used to monitor apoptosis/necrosis at the highest test concentration. The concentration of 5 μg/ml MMS was selected as the highest test concentration to cause 50 ± 5% cytotoxicity, following a 4- or 30-h treatment (Fig. [1](#Fig1){ref-type="fig"}a, b). At this test concentration, no evidence of increased cytotoxicity and cytostasis was seen from the %RPD and the fold change in EMA-positive events. In response to 5 μg/ml MMS, the %RPD dropped to 66% following 4 h and 56% following 30-h treatment. The fold change in EMA-positive events, a 1.7-fold increase following 4-h treatment and a 2.5-fold increase following a 30-h treatment in response to 5 μg/ml MMS, was well below the cut-off (≥4-fold) change for a dose to be considered overly cytotoxic.Fig. 1Cytotoxic and apoptotic/necrotic effects of MMS (**a**, **b**), carbendazim (**c**, **d**) and OTA (**e**, **f**) in TK6 cells following 4-h (*left-hand panel*) or 30-h (*right-hand panel*) treatment. The mean percentage RPD (*blue solid lines*) and EMA-positive fold change (histograms) were used as parameters to assess cytotoxicity (*n* = 3). Overly cytotoxic concentration (*black box*) as indicated by  %RPD or fold change in EMA-positive events (≥4 fold increase above the control) or both (Bryce et al. [@CR4]) (colour figure online) Carbendazim did not cause any increase in cytotoxicity or apoptosis/necrosis in TK6 cells following 4-h treatment (Fig. [1](#Fig1){ref-type="fig"}c). In contrast, increased apoptosis/necrosis was evident for the fold change values for EMA-positive events following 30-h continuous treatment. Sixfold and 9.5-fold increases in EMA-positive events were observed for 1 and 1.6 μg/ml concentrations. These fold change values for EMA staining were greater than fourfold increase above the control for these concentrations and hence considered overly cytotoxic. Contradictory results were also seen in TK6 cells following 4-h treatment with OTA. The 18 μg/ml concentration of OTA was identified as overly cytotoxic as 41% RPD (59% cytotoxicity) was seen at this dose (Fig. [1](#Fig1){ref-type="fig"}e). In contrast to %RPD, a 2.5-fold increase in EMA-positive fold change was recorded at the same analysed concentration. Following 30-h continuos treatment, 10 μg/ml OTA was identified as overly cytotoxic by both %RPD and fold change in EMA-positive events (See Fig. [1](#Fig1){ref-type="fig"}f). Therefore, MN enumeration was limited to 8 μg/ml concentration of OTA following continuous treatment. Evaluation of MN induction using the automated MN scoring platforms {#Sec13} ------------------------------------------------------------------- In the case of MMS, discrepancies were seen between the MN dose responses when using the Metafer and the MicroFlow approaches, following 4-h treatment (Fig. [2](#Fig2){ref-type="fig"}a). The Metafer scoring platform did not detect any significant increase in the MN induction following 4-h treatment. In contrast, a significant increase in MN frequency was detected at 5 μg/ml MMS when scoring was carried out using the MicroFlow approach. The mean MN responses were comparable between the scoring platforms in TK6 cells treated continuously for 30 h (Fig. [2](#Fig2){ref-type="fig"}b). Both the systems detected a significant (*p* \< 0.05) increase in MN induction in response to 2.5 and 5 μg/ml MMS.Fig. 2Genotoxic effects of MMS, carbendazim and OTA in TK6 cells following 4-h (*left-hand panel*) and 30-h (*right-hand panel*) treatment. The mean MN frequencies derived by the MicroFlow (*black bars*) approach and the Metafer (*grey bars*) scoring platforms. Increased cytotoxicity (*black box*) as indicated by  %RPD and fold change in EMA-positive events (≥4-fold increase above the control). *Asterisk* indicates a significant increase in the MN formation over the control using a (*p* \< 0.05). *Error bars* represent mean ± SD (*n* = 3). The  %RPD values in these graphs are same as those seen in Fig. [1](#Fig1){ref-type="fig"} In the cells treated with carbendazim, no increase in MN frequency was detected following 4-h treatment by either platform. Using the MicroFlow approach, a significant increase in MN was observed at 0.8, 1 and 1.6 μg/ml carbendazim at 30 h. However, increased apoptosis/necrosis was also seen when measuring fold change in the EMA-positive events at 1 and 1.6 μg/ml concentrations. The Metafer MN scoring platform detected a significant (*p* \< 0.05) increase in MN frequencies at carbendazim concentrations of 1 and 1.6 μg/ml. OTA induced a significant (*p* \< 0.05) increase in the MN induction above control at 16 and 18 μg/ml was detected by both the MN scoring platforms following 4-h treatment (Fig. [2](#Fig2){ref-type="fig"}c). In contrast, conflicting results were seen following 30-h continuous treatment with OTA (Fig. [2](#Fig2){ref-type="fig"}d). In this instance, no increase in the MN frequency was detected when using the Metafer platform at the analysed test concentrations, whereas a clear increase in MN induction above the control was seen at 8 μg/ml OTA with the MicroFlow approach. Furthermore, the MN frequencies obtained following 30-h treatment of MMS, carbendazim and OTA were comparable to those obtained in the cytokinesis block micronucleus assay using Metafer platform (please see supplementary data, Fig. 6) The manual scoring approach {#Sec14} --------------------------- Significant differences were seen between the MN responses derived by using the MicroFlow^®^ and the Metafer™ scoring platforms in TK6 cells following a 4-h treatment of MMS and 30-h OTA. To resolve this issue, the manual scoring procedure was used alongside the MicroFlow^®^ and the Metafer™ scoring platforms to assess MN induction at 4 h using MMS and carbendazim. In the case of MMS, MN dose response derived following 4-h MSS treatment by using the manual scoring method was comparable to that of the MicroFlow^®^ approach (Fig. [3](#Fig3){ref-type="fig"}a). Both the MicroFlow^®^ and manual scoring approaches detected a significant (*p* \< 0.05) increase in MN frequency at 5 μg/ml MMS concentration. In contrast, no significant increase in MN induction was seen when the scoring was carried out using the Metafer™ platform.Fig. 3Comparison of the MN responses derived by the MicroFlow (*black bars*), manual scoring (*green*) and the Metafer (*grey bars*) in TK6 cells treated with MMS and carbendazim for 4 h. *Asterisk* indicates a significant increase in the MN formation over the control using a (*p* \< 0.05). *Error bars* represent mean ± SD (*n* = 3) (colour figure online) Surprisingly, the MN response derived in TK6 cells with the manual scoring platform following 4-h treatment of carbendazim was different to those obtained using the MicroFlow^®^ and the Metafer™ scoring platforms (Fig. [3](#Fig3){ref-type="fig"}b). In this instance, a significant increase in MN induction was observed when manual scoring at 0.8 μg/ml and concentrations above it. In contrast, no such increase in the MN formation was seen when using the MicroFlow and the Metafer scoring platforms. Covariate BMD analysis {#Sec15} ---------------------- The order of endpoint sensitivity was deduced from the covariate BMD analysis, where the horizontal lines represent the BMDL~10~-BMDU~10~ metrics, with the lines in the top left being the lowest and most sensitive, and the ones on the bottom right being the highest and therefore least sensitive. Overlapping lines show equipotency between endpoints, and dotted lines represent poor model fits with infinite BMDL~10~ metrics (Fig. [4](#Fig4){ref-type="fig"}). The potency evaluations show that the MicroFlow approach was the most sensitive when compared to other techniques at 30 h. Using Metafer at 30-h chemical exposure was used to accurately characterise MN for all three chemicals, but for OTA it did produce wider BMD confidence intervals than the other approaches. Metafer at 4-h exposure was not suitable for MMS or carbendazim, but was suitable for the assessment of OTA.Fig. 4MN BMD Covariate analysis, potency ranking, from most potent/sensitive *top left* to least potent/sensitive bottom right. X-axis, Log10.dose (μg/ml). Carb, carbendazim; 30, 30-h treatment; 4, 4-h treatments, flow, MicroFlow; met, Metafer; man, manuals scoring For MMS, the BMD confidence intervals were indistinguishable between 30-h MicroFlow, 4-h manual, 4-h MicroFlow and 30-h Metafer with high precision, whereas 4-h Metafer provided the least precision in the BMD estimate. For carbendazim at 30-h treatment, the lowest BMD metrics were provided by MicroFlow followed by Metafer. At 4-h time-point, both Metafer and manual provided equivalent metrics, but the MicroFlow did not provide a good estimate of the BMD. Following 30-h treatment, MicroFlow and Metafer provided equivalent BMD metrics. At 4 h, MicroFlow and Metafer provided equivalent BMD metrics that were non-distinguishable from 30-h Metafer due to the wide confidence intervals, but with higher BMDs than for 30-h MicroFlow. Carbendazim altered the morphology of the micro nucleated cells and induced larger MN {#Sec16} ------------------------------------------------------------------------------------- The greatest discrepancies among scoring platforms were seen for TK6 cells treated with carbendazim. Whilst scoring MN using the manual scoring platform, it was observed that the nucleus of these micronucleated cells was crescent-/kidney-bean-shaped. Thus, it was speculated that these nuclear anomalies alongside large size MN were causing misclassification of micronucleated cells and MN with the Metafer scoring platform. The Metafer uses predefined parameters such as the size, aspect ratio, eccentricity and DAPI staining intensity for the detection of nuclei and MN (Varga et al. [@CR20]). Therefore, any deviations from these parameters for observed nuclei/MN will have a significant effect on MN frequency, and such MN cells will be excluded resulting into lower proportion of micronucleated cells. Studies with spindle poisons have previously shown to induce larger MN in TK6 and NH32 cells (Hashimoto et al. [@CR13]). Hence, it was postulated that carbendazim-induced MN were larger and thus not appropriately identified by Metafer classifier that had been standardised on micronucleated cells induced by clastogens. Hence, further MN scoring in TK6 cells treated with Carbendazim was carried out manually by using florescent microscopy in cells stained with DAPI and chromosomes counter stained with human pan centromeric probes. With this dual staining approach, two parameters such as the morphology of the micronucleated cells and the number of centromeric signals with the MN were evaluated to address the issue of underscoring with the Metafer system. A total of 100 micronucleated cells were assed for the occurrence of larger MN (MN with 2 or more centromeric signals) and morphologically altered MN cells. Carbendazim caused a concentration-dependent increase in the number of micronucleated cells with morphologically abnormal nuclei (Fig. [5](#Fig5){ref-type="fig"}a) and large size MN (Fig. [5](#Fig5){ref-type="fig"}b) in TK6 cells treated for 4 h. These results clearly indicate that the classifier standardised for detecting MN induced by clastogens might not be suitable to detect MN induced by aneugens.Fig. 5FISH to assess the induction of larger MN (**a**) and assessment of morphologically altered MN cells (**b**) in TK6 cells treated with carbendazim Discussion {#Sec17} ========== The reproducibility, sensitivity and transferability of the MicroFlow^®^ and Metafer™ approaches were compared with manual scoring through analyses of the dose response data. Using this approach, the MicroFlow^®^ data were comparable to the Metafer™ data for MMS, carbendazim and OTA, although there was a clear difference in the MN response magnitude. This difference could be due to underscoring by Metafer™ current classifier settings, where cells with novel nuclear morphology are not identified, or where there is misclassification of large MN as nuclei. However, this could be overcome with a visual detection step (Decordier et al. [@CR6]), or an updated classifier. A potential for overestimation of the MN frequencies with the MicroFlow^®^ approach could be due to the cell lysis step, where MN is not always associated with a single mononuclear cell. In both cases, the fold change in EMA-positive events along side %RPD was considered suitable to estimate cytotoxic concentration and to study apoptosis/necrosis. The flow-based MN scoring procedure provides benefits over manual scoring and, to an extent, semi-automated Metafer in terms of high-throughput MN scoring and multiplexing. With the MicroFlow approach, 10,000 events (cells) can be scored within a minute, whereas it takes up to 3 min to visually certify images of MN derived from 3000 thousands cells with Metafer and 15 min to visually inspect 1000 cells with the manual scoring platform. In addition to automated MN scoring, the MicroFlow approach permits assessment of additional cellular parameters such as cell cycle changes and apoptosis/necrosis which are otherwise difficult to assess using the conventional platforms. The BMD covariate analysis showed that for MMS and OTA, the Metafer and MicroFlow approaches produced equivalent BMD metrics, but for the aneugen carbendazim, the MicroFlow provided the most sensitive BMD estimates which were achieved at 30-h treatment. However, at 4 h, neither the Metafer or MicroFlow approaches were suitable for deriving BMD metrics. One major disadvantage of using the MicroFlow MN approach is the inability to differentiate bi-, tri- and multi-nucleated cells with MN and cells with multiple MN. This can lead to elevated MN frequencies compared with analyses conducted with intact cells. This effect was seen clearly in response to MMS and Carbendazim (Fig. [2](#Fig2){ref-type="fig"}). Doherty et al. ([@CR8]) also observed an increase in the frequency of micronucleated bi-nucleated cells with MMS in non-cyto-B treated cells (Doherty et al. [@CR8]). Additionally, the origin of the MN via clastogenic or aneugenic mechanisms cannot be elucidated following cell lysis procedure with the MicroFlow approach, although some cell lines (e.g., CHO-K1) are reported to provide aneugenicity signatures that include hypodiploidy and increased median MN fluorescence intensity (Bryce et al. [@CR3]). There are some issues in using the MicroFlow^®^ approach, such as re-validation of the misleading positive and negative result following cell lysis and flow cytometric analysis and stained samples cannot be store for a long period when compared with Metafer and manual scoring where slide can be store for months (Fenech et al. [@CR11]). However, there are also some disadvantages when using the Metafer system, with the major one being that some MN events are not picked, which leads to the underscoring as shown in the dose responses for all the three chemicals (Fig. [2](#Fig2){ref-type="fig"}). Since the samples were prepared from the same treated culture it was postulated that the Metafer™ classifier settings were incompatible for scoring MN induced by the aneugen Carbendazim. With the Metafer™ platform, the classifier is configured to assess parameters such as shape, circularity, aspect ratio and size to detect nuclei and MN (Reference). Therefore, it is possible that subtle changes in the morphology of nuclei/MN and induction of larger MN could cause underscoring with this system. The FISH and morphological studies provided some evidence on the induction of larger MN and morphologically altered nuclei in TK6 cells exposed to Carbendazim (Fig. [5](#Fig5){ref-type="fig"}a, b), which is in line with previous studies (Hashimoto et al. [@CR12]). However, a larger sample size and increase doses are required to confirm these findings for carbendazim, and this hypothesis also needs to be tested on other aneugens. The advantages and disadvantages of the each of these different scoring methodologies for scoring MN are summarised in Table [1](#Tab1){ref-type="table"}. In order for these approaches to be more widely used for MN scoring and dose response analysis, future ring trials should focus on addressing these considerations as well as assessing interlaboratory reproducibility.Table 1Summary of the advantages and disadvantages of manual, Metafer™ and the MicroFlow^®^ approachesMN scoring approachesScoring platformsAdvantagesDisadvantagesImage analysisManual microscopy (light microscopy)Suitable for dose response and mode of action analysis\ Simple, economical and adaptable\ Suitable for MN scoring in the presence or the absence of cyto-B\ Stained slides can be stored for a long time and can be re-analysed\ Suitable for assessing bi-, tri- and poly-nucleated cellsInteroperational variations can result in subjective MN scoring\ Slow, tedious and time-consuming\ Lack multiplexing abilities\ Total number of cells scored manually is limited which reduces the overall statistical powerMetafer™ (fluorescent microscopy)Semi-automated platform\ High content for higher statistical precision\ Suitable for dose response and mode of action analysis for most substances\ Images of nuclei and MN can be stored for re-validationClassifier settings have to be optimised for different cell lines and chemicals that induce MN via varied mechanisms\ Lack of cytoplasmic staining, detection of small MN and manual validation of the imagesFlow cytometryMicroFlow^®^Fully automated platform to score MN objectively\ Suitable for dose response\ High content and high throughput\ Permits cell cycle analysis\ 10,000 events scored in 1--2 minCell lysis is required prior to MN scoring\ Misleading MN cannot be re-validated from same sample\ Overestimation and underestimation of MN are both possible and require expert analysis\ Lack of MOA analysis with TK6 cells Both the MicroFlow and Metafer scoring approaches are suitable for automated MN scoring. However, in cases of equivocal with chemicals with unknown activity, it may be advisable to additionally process the same treated samples for manual scoring. These manually scored slides can be used to reduce the occurrence of misleading results, assess cytotoxicity or conduct mechanistic studies. Whilst conducting MN scoring on the semi-automated Metafer system, the classifier setting should be adopted to account for chemical or cell line-specific morphological changes and to reduce the occurrence of misleading results (positive and negative). These semi-automated and fully automated platforms can therefore be used for dose response analysis as substantially higher number of cells can be scored with these methods which allows for much statistical power. A test system that combines the high high-throughput, high-content and multiplexing potential of flow cytometry, with the re-validation and data storage benefits for image analysis, would be a major step forward in achieving a truly twenty-first century approach. Electronic supplementary material ================================= {#Sec18} Below is the link to the electronic supplementary material. Supplementary material 1 (PPTX 1078 kb) **Electronic supplementary material** The online version of this article (doi:10.1007/s00204-016-1903-8) contains supplementary material, which is available to authorized users. The authors thank Dr. Steven Bryce from Litron Laboratories (USA) for providing technical assistance and providing the MicroFlow^®^ kit. Special thanks to Dr. Leena Lodha for her financial support and National Centre for the Replacement, Refinement and Reduction of animals in Research (NC3Rs) for partly funding this research work (NC/K500033/1). Conflict of interest {#d29e999} ==================== The authors declare that they have no conflict of interest.

Introduction {#Sec1} ============ Measuring force and strain via telemetric methods is highly desirable in many industries, including civil engineering, geology exploration and biomedical diagnosis^[@CR1],\ [@CR2]^. Electromagnetic (EM)-based sensors that measure applied forces by reading the operating frequency shift of EM devices, which are highly sensitive to mechanical loading, have been proposed to address this challenges^[@CR2]--[@CR4]^. EM metamaterials are some of the first metamaterials studied and their remarkable properties have been the topic of numerous reports^[@CR5]--[@CR9]^. EM metamaterials have potential for application in a wide variety of areas, in particular for sensing^[@CR10]--[@CR15]^. Inspired by the metamaterial approach, researchers have designed various sensors to diagnose strain or force telemetrically^[@CR2]--[@CR4],\ [@CR16]--[@CR19]^. Previously reported strain- or force-measurement devices typically comprise metallic resonators, such as split-ring resonators (SRRs)^[@CR2]--[@CR4]^, or complex metallic nanostructures^[@CR18]--[@CR22]^. However, because SRR-based sensors with substrates such as silicon^[@CR2]^ or Kapton tape^[@CR16]^ have limited sensing and deformation ranges, they are only suitable for large force and small strain measurement, such as detecting the rigidity of buildings or bones. SRR-based sensors are not suitable for measurement of small forces and large strains, which commonly occur in soft materials such as biological tissues, because of the limited sensitivity and large size. Furthermore, the sensors usually need to be adhered to the surface of the tested materials, which can cause measurement errors and lead to difficulties in measuring objects with complicated structures. Recent studies have shown that either linear or nonlinear performance can be achieved by introducing elastomer materials such as polydimethylsiloxane into metal nanoparticle structures^[@CR19],\ [@CR23]--[@CR26]^, which has enabled the development of flexible EM devices that can be mechanically tuned with large strain deformation and have broad applications in biomedical engineering. These nanostructures usually require complex fabrication processes and operate in high-frequency ranges, such as near-infrared wavelengths^[@CR19]^ or in the visible region^[@CR21],\ [@CR22]^. Because lower EM frequencies are more easily transmitted in biological material than high frequencies, it is highly desirable to reduce the operating frequency of the metamaterial sensors. However, decreasing the frequency usually requires the dimensions of the metamaterial sensors to increase, which has a negative influence on the measured system and prevents their application in many industries, in particular in medical diagnosis. To measure force and strain inside complicated biological structures such as soft tissues and muscle, it is necessary to develop small, high-sensitivity, biologically-compatible sensors that are able to operate in the low-frequency region. In this paper, we propose a dielectric Mie resonant sensor that operates in the microwave region and is able to detect small forces and large strains inside biological materials. The sensor has dimensions of 2 mm × 2 mm × 6 mm. Besides, it has deep transmission dips, which permits an accurate measurement of small forces, such as those that occur in joints, muscles and soft tissues. Results {#Sec2} ======= The force-sensitive sensor was designed by combining a dielectric resonator and an elastomer layer. The dielectric resonator comprised two dielectric ceramic cubes, and the elastomer layer was fabricated from silicone rubber. The compressive stress--strain relationship of the silicone rubber was tested using a material testing machine (Zwick/Roell Z020, Germany) and the result is shown in Fig. [1(a)](#Fig1){ref-type="fig"}. Our sensor is designed in a strain range from 0 to 0.33, as marked by the shading area in Fig. [1(a)](#Fig1){ref-type="fig"}. From the figure, we can see that the rubber presents a slight non-linearity at the beginning of loading. To determine the reason for the nonlinear behavior, we performed 10 cycles of loading and unloading tests on the rubber. The results are shown in Fig. [S1](#MOESM1){ref-type="media"} in the supporting information, which present a permanent nonlinearity at the beginning of each loading process. We believe that it can be attributed to two sources: the first is the nonideal boundary conditions that apply during the measurement, which are unavoidable, and the second is the inherent nonlinearity of the rubber material. While we cannot determine which of the two is the main cause of the nonlinearity, we can conclude that the rubber material presented a permanent nonlinearity at the beginning of loading process. The hysteresis between the loading and unloading is derived from the viscoelasticity of the rubber, indicating that the rubber cannot recover to its original shape immediately after unloading. About this point, we will present in detail in the experimental section.Figure 1(**a**) The measured relationship between the compressive stress and strain for the silicone rubber. (**b**) The simulation setup in the CST microwave studio. (**c**) The simulated transmission spectra curves of the dielectric Mie resonant antenna under different compressive force levels. (**d**) The calculated relationship between the frequency shift and the applied force. The inset of (**d**) shows the simulated frequency shift at five loading levels from 0 N to 1 N. For a ceramic particle with fixed dimensions and permittivity, there will be a series of resonance dips in the transmission spectrum^[@CR9]^. We used the transmission spectra in the S~21~ configuration to observe the frequency shift of the first Mie resonance dip because it is sharper than subsequent dips. The deformation of the rubber layer caused a redshift of the first Mie resonance frequency because of the coupling effect between dielectric cubes. Strong magnetic resonance occurs at the first Mie resonance frequency and the dielectric cubes can be considered as magnetic dipoles. When compressive force is applied to the dielectric cubes, the interplay between the magnetic dipoles becomes stronger and leads to an increase of the resonance frequency^[@CR27]--[@CR29]^. To explore the relationship between the applied force and the first Mie resonance frequency shift of the sensor, we performed simulations using the Microwave Studio software package (CST Studio Suite 2015, Germany). The system was modelled as a rectangle having dimensions 22.86 mm × 10.16 mm and surrounded by a perfect conductor, as shown in Fig. [1(b)](#Fig1){ref-type="fig"}. The permittivity of the dielectric cubes (CaTiO~3~--2 wt% ZrO~2~ ceramic) was set to 115 at room temperature and the permittivity of the intermediate layer was set to 2.3, to coincide with the experimental parameters. The dimensions of the dielectric cubes and the elastic layer were 2 mm × 2 mm × 2 mm. A compressive force between 0.0 N and 8.0 N was applied in the direction orthogonal to the dielectric cubes, as shown in the inset of Fig. [1(c)](#Fig1){ref-type="fig"}. Because the Young's modulus of ceramic cubes was much higher than that of the silicone rubber, their deformation under the compressive force was neglected in both the simulation and experiment. The simulated frequency shift of the sensor with different compressive forces applied is shown in Fig. [1(c)](#Fig1){ref-type="fig"}. The first resonance frequency of the dielectric sensor experienced a redshift as the applied load increased. From the simulated curves shown in Fig. [1(c)](#Fig1){ref-type="fig"}, we obtained the relationship between the frequency shift (denoted as Δ*f*) and the applied force, as shown in Fig. [1(d)](#Fig1){ref-type="fig"}. The magnitude of the frequency shift of the sensor increased with the applied force, which means that the applied force can be determined telemetrically by observing the redshift of the operating frequency. From Fig. [1(a)](#Fig1){ref-type="fig"} and Fig. [S1](#MOESM1){ref-type="media"}, we can see that the rubber layer always presents a non-linearity mechanical performance at the beginning of loading. Therefore, we decided to use a nonlinear method to analyze the relationship between the frequency shift and the applied compressive force. We found that the simulated values could be fitted well using a quadratic equation, as shown in Fig. [1(d)](#Fig1){ref-type="fig"}. From the simulated results, we can conclude that the silicone rubber is very sensitive to mechanical loads and experiences a large frequency shift under small loading levels. To demonstrate the performance of the designed force sensor experimentally, we chose ceramic cubes as the dielectric resonators and silicone rubber as the elastic layer, as in the simulation. Figure [2(a)](#Fig2){ref-type="fig"} shows the dielectric cubes with dimensions of 2 mm × 2 mm × 2 mm. The main constituent of the dielectric cubes was calcium titanate (CaTiO~3~), which has been shown to be biologically compatible and has been widely used as a biomedical material^[@CR30],\ [@CR31]^. The dielectric cubes had a permittivity of approximately 115 and a loss tangent of 0.001 at room temperature. This allows the cubes to be used to make high-permittivity and low-loss sub-wavelength resonators with a high quality factor, which are suitable for operation in the microwave bands. The dielectric sensor was fabricated by bonding two dielectric cubes with a silicone rubber intermediate layer with the same dimensions, as shown in Fig. [2(b)](#Fig2){ref-type="fig"}. The total size of the sample was 2 mm × 2 mm × 6 mm.Figure 2(**a**) The dielectric ceramic cubes with dimensions of 2 mm × 2 mm × 2 mm. (**b**) The fabricated force sensor composed of two dielectric cubes and a silicone rubber layer. To measure the transmission performance of the sensor under different mechanical loading levels, we constructed an experimental setup that combined a material testing machine with a vector network analyser (Agilent PNA-LN5230C), as shown in Fig. [3(a)](#Fig3){ref-type="fig"}. A force gauge was used to record the compressive force applied on the sensor. To minimize the measurement error, we used two wood sticks, which had a dielectric constant of 2.6 at room temperature, to apply force on the dielectric sensor, as shown in Fig. [3(c)](#Fig3){ref-type="fig"}. Two WR-90 rectangular waveguides operating in the frequency range from 8.2 GHz to 12.4 GHz were used as the excitation transmitter and receiver. The waveguides were connected to the input and output ports of a network analyser to measure the S~21~ parameter when different compressive forces were applied in the direction orthogonal to the sensor, as shown in Fig. [3(a)](#Fig3){ref-type="fig"}.Figure 3(**a**) An overview of the experiment setup. (**b**) The sample placed on the wood ejector bar. (**c**) The compressive force applied on the sensor orthogonally by the ejector bars. The transmission spectra measured in S~21~ configuration of the dielectric sensor under different applied force magnitudes are shown in Fig. [4(a)](#Fig4){ref-type="fig"}. A frequency shift was observed when the compressive force increased from 0.0 N to 8.0 N. For all of the measured loading levels, the transmission dips were greater than −12.5 dB, which made it straightforward to determine the frequency shift. Comparing the experimental and simulated results (shown in Figs [1(c)](#Fig1){ref-type="fig"} and [4(a)](#Fig4){ref-type="fig"}, respectively), we can see that the measured transmission spectra of the sample agree well with the simulated spectra. Slight discrepancies result from the influence of the wood ejectors and fabrication errors. From the transmission curves shown in Fig. [4(a)](#Fig4){ref-type="fig"}, we obtained the relationship between the resonance frequency shift and the applied compressive force. As in the simulation, we also fitted the measured values by using quadratic equations. The relationship between the frequency shift (Δ*f*) and the applied force (*F*) is shown in Fig. [4(b)](#Fig4){ref-type="fig"} while the relationship between Δ*f* and the strain of the sample (*S*) is shown in Fig. [4(c)](#Fig4){ref-type="fig"}. We also calculated the errors that occurred between the measured values and the corresponding values that were calculated from the quadratic fitted Δ*f*--*F* curve. The sensor errors at the different data points were calculated as percentages and are shown in Fig. [4(d)](#Fig4){ref-type="fig"}. From the figure, we see that the maximum fitting error for the data is less than 17%. It is well known that higher orders of fitted polynomials will produce more accurate fitting results. However, increasing the order of the polynomial will also produce a more complex calculation process. Therefore, we adopt the quadratic fitting procedure here to reach a compromise between the fitting accuracy and the calculation costs.Figure 4(**a**) The measured resonance frequency shift with different compressive force level applied on the sensor. (**b**) The measured relationship between the resonant frequency shift and the applied force. (**c**) The measured relationship between the resonant frequency shift and the strain of the sensor. (**d**) The errors that occurred between the measured values and the corresponding values that were calculated from the quadratic fitted Δ*f*--*F* curve. To ensure that the sensor can be used within soft materials with high stability, we performed force tests by placing the sample inside soft materials. We stuck two pieces of rubber on to the wood ejector, and placed the sensor between these pieces, as shown in Fig. [5(a)](#Fig5){ref-type="fig"}. Then, we performed loading and unloading processes 100 times at loading levels of both 1.8 N and 2.5 N, and recorded the resulting frequency shifts in the resonant frequency. The results are shown in Fig. [5(b)](#Fig5){ref-type="fig"}, which illustrates the relatively good performance stability of the sensor. From the statistics shown in Fig. [5(c) and (d)](#Fig5){ref-type="fig"}, we can conclude that under a compressive load of 2.5 N, the number of measured values of Δ*f* that are fluctuating around the average value within an error of 8% is 97, while the corresponding number for 1.8 N is 84; these results demonstrate the relatively good stability of the sensor. It should be noted here that after each loading and unloading process, we must wait for several seconds after each measurement performed using the sensor before the next test begins. Otherwise, measurement errors will be introduced by the viscoelastic properties of the rubber pieces.Figure 5(**a**) Measurement setup for testing of the stability of the sample between soft materials. (**b**) Measured values of the frequency shift of the sensor for 100 repeated tests at loading levels of 1.8 N and 2.5 N. (**c**) Statistical distribution for the 100 repeated tests under a loading force of 1.8 N. (**d**) Statistical distribution for the 100 repeated tests under a loading force of 2.5 N. Discussion {#Sec3} ========== The high sensitivity, biological compatibility and small size of the proposed dielectric force sensor make it suitable for many applications that require small forces and large strains, such as those that commonly occur inside soft materials like soft tissues and muscle, to be measured. The high sensitivity of the sensor described here is derived from the low elastic modulus of the silicone rubber, which undergoes greater mechanical deformation under the same applied load compared with materials such as silicon or Kapton tape, which have been used as substrates in previously studies^[@CR2],\ [@CR16]^. Of course, we do believe that with the continuing developments in materials engineering, new soft materials with greater flexibility, less viscoelasticity, and better biocompatibility are likely to be synthetized in the near future that would represent suitable replacements for the silicone rubber, and we will consider these materials. Because the resonant frequency depends on the size and permittivity of the ceramic particles^[@CR9]^, by carefully designing the size parameters of the sensor, or replacing the intermediate elastic layer with another elastomer with a different Young's modulus, various dielectric sensors with tunable sensitivity and operating frequency region could be constructed to meet specific application demands. Compared with the previous sensors based on SRRs or metallic nanoparticle structures, this method provides a lower-cost and simpler approach to measuring force in soft biological materials. Conclusion {#Sec4} ========== A dielectric force sensor with high sensitivity and small dimensions was designed by introducing an elastic material with low Young's modulus into dielectric ceramic cubes. Experimental and simulation results show an approximate quadratic relationship between the frequency shift of the sensor and the applied compressive force with a maximum fitting error less than 17%. The sensor can be used over an ultra-large deformation strain range and has a relatively low operating frequency range in the microwave region. It should be noted that our current design cannot measure the shear strain. In our future work, we will extend our efforts to design new sensors that are capable of measuring the shear and normal strains simultaneously. Appropriate selection of the dimensions and permittivity of the dielectric resonators, as well as the Young's modulus of the elastic layer, could enable the design of electromagnetic sensors with different sensitivity and measurement ranges for use in a variety of industrial applications. Methods {#Sec5} ======= Sample preparation {#Sec6} ------------------ The ceramic dielectric was fabricated by solid-state reaction by mixing CaTiO~3~ powders with 2 wt%ZrO~2~. The dielectric cubes were achieved by cutting the dielectric ceramic plate into dimensions of 2 mm × 2 mm × 2 mm. The permittivity of the obtained ceramic was 115 + 0.001*i* at room temperature. Simulation and measurement {#Sec7} -------------------------- The samples were measured using two WR-90 rectangular waveguides with sectional sizes of 22.86 mm × 10.16 mm × 100 mm. The other ends of the two waveguides were connected to the input and output of a vector network analyzer (N5230C, Agilent Technologies, USA). The simulated microwave transmission S~21~ was achieved by using the Microwave Studio software package (CST Studio Suite 2016, Germany). Data Availability {#Sec8} ----------------- The datasets generated during and analyzed during the current study are available from the corresponding author on reasonable request. Electronic supplementary material ================================= {#Sec9} Supporting information **Electronic supplementary material** **Supplementary information** accompanies this paper at doi:10.1038/s41598-017-04911-2 **Publisher\'s note:** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This work was supported by the National Natural Science Foundation of China under Grant Nos 11274198 and 51532004 and the Science and Technology Plan of Shenzhen City under grant JCYJ 20150827165038323. Ji Zhou conceived the idea. Lingling Wu and Xiaoqing Xi designed experiments and carried out numerical calculations. Bo Li revised the full paper. All authors contributed to scientific discussion and critical revision of the article. Competing Interests {#FPar1} =================== The authors declare that they have no competing interests.

Sample analysis of the Human Genome Diversity Project-Centre d'Etude du Polymorphisme Humain (HGDP-CEPH) panel with 377 microsatellites or short tandem repeats (STR) conducted in five Native Amerindian populations (Pima, Maya, Colombian, Karitiana, and Surui) found a high frequency of a small allele (275 basepairs) at the tetra-nucleotide locus D9S1120, which was absent in 47 other worldwide populations ([@R1]). Based on the corresponding number of repeats, this private allele was identified as "9RA" (9 repeats allele). The ubiquitous presence of 9RA in North and South American populations, including the Na-Dene and Aleut-Eskimo, and in related Western Beringian groups suggested that all modern Native American populations originated from the same founding population ([@R2]). A later extended survey of 678 STRs in 29 American populations found high frequencies of 9RA across all American regions (average 0.301 in North America, 0.471 in South America, and 0.364 in the full Native American sample), which was also interpreted as evidence of a single main colonization event ([@R3]). Interestingly, this interpretation based on a single autosomal marker is in agreement with archeological, mitochondrial, and Y-chromosomal data ([@R4]-[@R10]). Finally, the single main colonization hypothesis is supported by the following observations: 1) all the chromosomes with 9RA share the same haplotypic background in the vicinity of D9S1120, suggesting they are identical by descent; 2) the positive selection hypothesis was shown as unlikely; and 3) the range of time estimated for the most recent common ancestor for the 9RA marker is consistent with other recent estimates based on archeological and genetic data concerning the origin of Native American populations ([@R11]). The forensic potential of D9S1120 for detecting Native American ancestry was evaluated in a third study, which typed three native and two admixed populations from Colombia and three non-American populations ([@R12]). For this purpose, a new primer set reducing the amplicon sizes was designed and an allelic ladder was constructed, characterizing 13 alleles ([@R12]). However, the ability of D9S1120 to identify Native American ancestry requires a fuller evaluation in Latin American populations, both for anthropological and forensic genetics purposes. For instance, the greatest part of Mexican population belongs to an ethnic group created by post-Colombian admixture, commonly known as Mestizos (\>90%); in Mestizos, the frequency of the European ancestry component increases toward the northwest and the frequency of the Amerindian ancestry component increases toward the central-southeast ([@R13],[@R14]). The frequency of African ancestry component is low and evenly distributed through the country ([@R13],[@R14]). Mexico has a large number of Amerindian populations, with over 68 ethnic groups representing 9.6% of the total population ([@R15]). We analyzed the D9S1120 STR in five Mestizo and eight Amerindian populations from different regions of Mexico. The genetic diversity and population differentiation based on D9S1120 were compared with those obtained by STRs of the Combined DNA Index System (CODIS-STRs). Materials and methods ===================== D9S1120 genotyping ------------------ DNA was extracted from fresh blood samples by salting-out method ([@R16]). Its quality was evaluated by 1% agarose gel electrophoresis and observed by ethidium bromide staining. The study included 247 and 707 unrelated persons from five Mestizo and eight Amerindian populations from Mexico, respectively ([Table 1](#T1){ref-type="table"}; [Figure 1](#F1){ref-type="fig"}). All participants signed a written informed consent, according to the ethical guidelines of the Helsinki Declaration and the study was approved by the Ethics Research Committee of the CUCiénega, University of Guadalajara. For amplification of D9S1120, we used primers and conditions described by Phillips et al ([@R12]). The polymerase chain reaction products were separated by capillary electrophoresis using the ABI Prism 310 and profiles were analyzed with GeneMapper ID software, version 3.2 (Applied Biosystems, Foster City, CA, USA). D9S1120 alleles were named according to the repeat structure and size (bp) described by Phillips et al ([@R12]), which follows the International Society of Forensic Genetics guidelines for STR analysis recommending thorough sequence analysis of alleles used to construct reference ladders. ###### Description of the Mexican populations analyzed with the short tandem repeat (STR) D9S1120\* Mestizo population Abbreviation Location, state Region Sample size (n) ---------------------- -------------- -------------------------------------- ----------- ----------------- **Chihuahua** Chih Chihuahua, Chihuahua North 51 **Jalisco** Jal Guadalajara, Ocotlán, Jalisco West 52 **Veracruz** Ver Veracruz, Veracruz Center 42 **Chiapas** Chis Tapachula, Chiapas Southeast 51 **Yucatán** Yuc Mérida, Yucatan Southeast 51 **Amerindian group** **Tarahumara** Tar Chihuahua, Chihuahua North 125 **Huichol** Hui San Sebastián Teponamastlán, Jalisco West 61 **Purépecha** Pur Zipiajo and Angahuan, Michoacán West 111 **Mazateco** Maz San Miguel Soyaltepec, Oaxaca South 41 **Lacandón** Lac Lacanjá Chansayab, Ocosingo, Chiapas Southeast 78 **Tzotzil** Tzo San Juan Chamula, Chiapas Southeast 113 **Tojolabal** Toj Las Margaritas, Chiapas Southeast 52 **Mayas** May Yucatan and Quintana Roo Southeast 126 \*For some analyses these populations were grouped to estimate allele and genotype distribution. {#F1} Data analyses ------------- Allele distribution and the following statistical parameters of forensic importance were computed with the PowerStats software ([@R17]): observed heterozygosity (Ho), power of exclusion (PE), power of discrimination (PD), polymorphism informativity content (PIC), and typical paternity index (IP). For each population sample, Hardy-Weinberg expectation was verified by exact tests (95% confidence interval, CI) using the Genetic Data Analysis program (GDA), version 1.1 ([@R18]). Genetic differentiation was evaluated by F~ST~ distances and exact test *P* values including previously published population data ([@R12]) using Arlequin 3.1 software ([@R19]). We compared the genetic diversity (*Het*) and population differentiation (F~ST~) based on D9S1120 polymorphism with those previously estimated with nine CODIS-STRs in Mexican Mestizos from Chihuahua, Jalisco, Yucatan, and Veracruz ([@R13],[@R20],[@R21]), and available data from Maya and Purépecha Amerindian groups ([@R15]). The CODIS-STRs were used for comparison purposes because they were analyzed in all the studied Mexican populations and are included in both Identifiler and Profiler kits (Applied Biosystems). Genetic distances were shown in a multidimensional scaling (MDS) plot to explore the genetic relationships among the populations with SPSS, version 10.0 (SPSS Inc., Chicago, IL, USA). Analysis of molecular variance (AMOVA) was carried out in the total Mexican population sample, and separately in Mestizos and Amerindian groups. Results ======= Allele frequencies and forensic parameters ------------------------------------------ In Mexican populations, we identified nine alleles (alleles 9, 10, and 13-19), but the allele 10 was exclusively observed in the Purépecha group. The increased allele number by population was related to the presence of alleles 13 and 19, which were observed in Mestizos from Chihuahua (north), Jalisco (west), and the Maya group (southeast). The smallest allele number was observed in the Tojolobal (4 alleles) and Lacandon (5 alleles) native groups. The modal alleles in Amerindian groups were 9 (38.2%) and 16 (30.5%), but in the Tojolabal and Maya groups the allele 16 was prevalent over the allele 9 (modal values of 47% and 34%, respectively) ([Tables 2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}). In Mestizos, the modal allele was 16 (39.1%) followed by 9 (21%). ###### Statistical parameters of forensic importance in five Mexican-Mestizo populations for the locus D9S1120\***^†^** Chihuahua Jalisco Veracruz Chiapas Yucatán Global ---------------------------- --------------- --------------- --------------- --------------- --------------- --------------- **9** 0.1373 0.2212 0.2381 0.2255 0.2353 0.2105 **13** 0.0098 0.0192 \- \- \- 0.0061 **14** 0.0392 0.0288 0.0238 0.0294 0.0294 0.0304 **15** 0.1961 0.0865 0.1429 0.1667 0.0980 0.1376 **16** [0.4118]{.ul} [0.4712]{.ul} [0.2976]{.ul} [0.3627]{.ul} [0.3922]{.ul} [0.3907]{.ul} **17** 0.1569 0.125 0.2619 0.1667 0.2157 0.1822 **18** 0.0392 0.0385 0.0357 0.0490 0.0196 0.0364 **19** 0.0098 0.0096 \- \- 0.0098 0.0061 **MAF** 0.0512 0.0460 0.0584 0.0512 0.0512 **Genotype** n = 51 n = 52 n = 42 n = 51 n = 51 n = 247 **9/9** 0.0392 0.0962 0.0952 0.0588 0.0588 0.0691 **9/13** \- 0.0192 \- \- \- 0.0041 **9/14** 0.0196 \- 0.0238 0.0392 0.0196 0.0203 **9/15** 0.0392 \- 0.0714 0.0588 0.0784 0.0488 **9/16** 0.0980 0.1731 0.0476 0.1373 0.1373 0.1219 **9/17** 0.0196 0.0385 [0.1429]{.ul} 0.0784 0.0980 0.0732 **9/18** 0.0196 \- \- 0.0196 0.0196 0.0122 **9/19** \- 0.0192 \- \- \- 0.0041 **13/15** 0.0196 \- \- \- \- 0.0041 **13/17** \- 0.0192 \- \- \- 0.0041 **14/15** \- \- \- 0.0196 \- 0.0041 **14/16** 0.0196 0.0385 0.0238 \- \- 0.0163 **14/17** 0.0392 0.0192 \- \- 0.0196 0.0163 **14/18** \- \- \- \- 0.0196 0.0041 **15/15** 0.0392 0.0192 0.0238 0.0588 \- 0.0285 **15/16** 0.1765 0.0962 0.0952 0.1176 0.0980 0.1138 **15/17** 0.0392 0.0385 0.0714 0.0196 0.0196 0.0366 **15/18** 0.0196 \- \- \- \- 0.0041 **15/19** 0.0196 \- \- \- \- 0.0041 **16/16** 0.1569 [0.2500]{.ul} [0.1429]{.ul} 0.1176 0.1569 [0.1667]{.ul} **16/17** [0.2157]{.ul} 0.0962 0.1190 [0.1765]{.ul} [0.2157]{.ul} [0.1667]{.ul} **16/18** \- 0.0385 0.0238 0.0588 \- 0.0244 **16/19** \- \- \- \- 0.0196 0.0041 **17/17** \- 0.0192 0.0714 0.0196 0.0392 0.0285 **17/18** \- \- 0.0476 0.0196 \- 0.0122 **18/18** 0.0196 0.0192 \- \- \- 0.0081 **Statistical parameters** Average **Ho** 0.7451 0.5962 0.6667 0.7451 0.7451 0.6996 **PIC** 0.7116 0.6671 0.7243 0.7220 0.6913 0.7033 **PD** 0.8774 0.8713 0.9048 0.8989 0.8774 0.8860 **PE** 0.5014 0.2863 0.3786 0.5014 0.5014 0.4338 **Typical PI** 1.9615 1.2381 1.5000 1.9615 1.9615 1.7245 ***P*^‡^** 0.1027 0.1289 0.5024 0.6445 0.5218 \*Abbreviations: MAF -- minimum allele frequency; Ho -- heterozygosity observed; PIC -- polymorphism informativity content; PD -- power of discrimination; PE -- power of exclusion; PI -- paternity index. †Modal alleles and genotypes by populations are underlined. ‡Exact test for Hardy-Weinberg expectations. ###### Statistical parameters of forensic importance in eight Mexican-Amerindian groups for the locus D9S1120\***^†^** Tarahumara Huichol Purépecha Mazateca Tzotzil Tojolabal Lacandon Maya Global ---------------------------- --------------- --------------- --------------- --------------- --------------- --------------- --------------- --------------- ---------------- **9** [0.44]{.ul} [0.3934]{.ul} [0.4324]{.ul} [0.378]{.ul} [0.354]{.ul} 0.2115 [0.5513]{.ul} 0.2659 [0.38190]{.ul} **10** \- \- 0.0045 \- \- \- \- \- 0.0007 **13** \- \- \- \- \- \- \- 0.0040 0.0007 **14** 0.004 0.0246 0.0045 0.0122 \- \- \- 0.0159 0.0071 **15** 0.132 0.1557 0.1261 0.1585 0.1637 0.0769 0.0705 0.1508 0.1322 **16** 0.324 0.3525 0.2523 0.3415 0.2788 [0.4712]{.ul} 0.1603 [0.3413]{.ul} 0.3048 **17** 0.088 0.0574 0.1667 0.0732 0.1283 0.2404 0.1603 0.1587 0.1351 **18** 0.012 0.0164 0.0135 0.0366 0.0708 \- 0.0577 0.0595 0.0361 **19** \- \- \- \- 0.0044 \- \- 0.004 0.0014 **MAF** 0.0204 0.0406 0.0233 0.0578 0.0240 0.0499 0.0325 0.0223 **Genotype** n = 125 n = 61 n = 111 n = 41 n = 113 n = 52 n = 78 n = 126 n = 707 **9/9** 0.2080 0.1803 0.1802 [0.1951]{.ul} 0.1150 0.0769 [0.3077]{.ul} 0.0794 0.1641 **9/10** \- \- 0.0090 \- \- \- \- \- 0.0014 **9/14** 0.0080 0.0164 \- \- \- \- \- 0.0079 0.0042 **9/15** 0.0960 0.1148 0.1351 0.0488 0.0531 0.0192 0.0769 0.0794 0.0834 **9/16** [0.2720]{.ul} [0.2459]{.ul} [0.2523]{.ul} 0.1707 [0.2212]{.ul} 0.1731 0.1410 [0.1984]{.ul} [0.2178]{.ul} **9/17** 0.0880 0.0492 0.0991 0.1220 0.115 0.0769 0.1795 0.0635 0.0976 **9/18** \- \- 0.0090 0.0244 0.0796 \- 0.0897 0.0238 0.0297 **9/19** \- \- \- \- 0.0088 \- \- \- 0.0014 **13/14** \- \- \- \- \- \- \- 0.0079 0.0014 **14/15** \- \- 0.0090 \- \- \- \- 0.0159 0.0042 **14/16** \- 0.0164 \- 0.0244 \- \- \- \- 0.0028 **14/18** \- 0.0164 \- \- \- \- \- \- 0.0014 **15/15** 0.0240 0.0328 0.0180 0.0488 0.0531 \- \- 0.0238 0.0255 **15/16** 0.0960 0.0984 0.0450 0.1463 0.1239 0.0962 \- 0.1032 0.0863 **15/17** 0.0160 0.0328 0.0270 \- 0.0442 0.0385 0.0513 0.0397 0.0325 **15/18** 0.0080 \- \- 0.0244 \- \- 0.0128 0.0159 0.0071 **16/16** 0.1200 0.1475 0.0721 0.1463 0.0708 0.1731 0.0385 0.0873 0.0976 **16/17** 0.0400 0.0328 0.0541 0.0244 0.0354 [0.3269]{.ul} 0.0897 0.1270 0.0820 **16/18** \- 0.0164 0.009 0.0244 0.0354 \- 0.0128 0.0714 0.0240 **16/19** \- \- \- \- \- \- \- 0.0079 0.0014 **17/17** 0.0160 \- 0.0721 \- 0.0265 0.0192 \- 0.0397 0.02687 **17/18** \- \- 0.0090 \- 0.0088 \- \- 0.0079 0.0042 **18/18** 0.0080 \- \- \- 0.0088 \- \- \- 0.0028 **Statistical parameters** **Average** **Ho** 0.6240 0.6393 0.6577 0.6098 0.7257 0.7308 0.6538 0.7698 0.6764 **PIC** 0.6200 0.6380 0.6588 0.6582 0.7083 0.6150 0.5980 0.7237 0.6525 **PD** 0.8393 0.8557 0.8590 0.8673 0.8868 0.8099 0.8268 0.8995 0.8555 **PE** 0.3207 0.3408 0.3658 0.3028 0.4691 0.4774 0.3605 0.5443 0.3977 **Typical PI** 1.3298 1.3864 1.4605 1.2813 1.8226 1.8571 1.4444 2.1724 1.5943 ***P\**** 0.0897 0.6252 0.1203 0.2272 0.0364 0.3685 0.1952 0.3071 \*Abbreviations: MAF -- minimum allele frequency; Ho -- heterozygosity observed; PIC -- polymorphism informativity content; PD -- power of discrimination; PE -- power of exclusion; PI -- paternity index. †Modal alleles and genotypes by populations are underlined. ‡Exact test for Hardy-Weinberg expectations. Genotype distributions of D9S1120 were in agreement with the Hardy-Weinberg expectations in all Mexican populations ([Table 2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}). The only exception was the Tzotzil group, which showed a relatively low *P* value (*P* = 0.0364), however this was not significant after Bonferroni correction. In order to record representative D9S1120 genotypes for the main Mexican population groups, we obtained a ratio of genotype frequencies (RGF) between Mestizos and Amerindians and vice-versa ([Figure 2](#F2){ref-type="fig"}). Although genotypes 9/14 and 14/16 can be used to indicate the Amerindian component of Mestizo origin, their potential may be limited by low population frequencies (≤2.03 and 1.63%, respectively). In general, RGF values indicated the allele 9 was typically observed in Amerindian groups and the alleles 14 and 16 in Mestizos. {#F2} Genetic relatedness between populations --------------------------------------- We estimated genetic distances and Fst *P* values based on D9S1120 ([Table 4](#T4){ref-type="table"}) and represented them graphically ([Figure 3](#F3){ref-type="fig"}). Mexican populations were compared to the reference populations from Europe, Africa, and East Asia ([@R12]) (*P* \< 0.01) and were clearly separated in the MDS plot ([Figure 3A](#F3){ref-type="fig"}). No differentiation was observed between all five Mexican Mestizo populations but the majority of differences in pairwise comparisons between Mestizo populations and Lacandones, Tarahumaras, Purépechas, and Tzotzils were significant ([Table 4](#T4){ref-type="table"}). Among native groups, Tojolabales and Mayas showed a high frequency of allele 16 ([Table 3](#T3){ref-type="table"}), which explains their close relationship with Mexican Mestizos ([Figure 3](#F3){ref-type="fig"}). Similarly, Mulalós from Colombia ([@R12]) showed differences from all Mexican Amerindian groups (*P* \< 0.01), but were similar to most of Mexican Mestizos, except those from Yucatan (*P* = 0.0032) (data not shown). ###### Genetic distances (Fst, below diagonal) and Fst *P* values (above diagonal) between Mexican Mestizos and Amerindian populations\* Mestizos Amerindian groups ---------- ---------------- ------------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- ---------------- **Chih** \*\*\*\*\*\*\* 0.4811 0.1796 0.7476 0.7015 0.0000\* 0.0086\* 0.0000\* 0.0264 0.0000\* 0.3464 0.0000\* 0.0332 **Jal** 0.0160 \*\*\*\*\*\*\* 0.2024 0.3363 0.2158 0.0000\* 0.1468 0.0000\* 0.3628 0.0010\* 0.1958 0.0000\* 0.0389 **Ver** 0.0230 0.0339 \*\*\*\*\*\*\* 0.7544 0.6966 0.0027\* 0.0373 0.0163 0.1905 0.0251 0.0552 0.0000 0.2915 **Chis** 0.0072 0.0136 0.0079 \*\*\*\*\*\*\* 0.8938 0.0001\* 0.0826 0.0017\* 0.3423 0.0879 0.3627 0.0000\* 0.9471 **Yuc** 0.0154 0.0104 0.0075 0.0056 \*\*\*\*\*\*\* 0.0098\* 0.0359 0.0042\* 0.0706 0.0018\* 0.8602 0.0014\* 0.4635 **Tar** 0.0766 0.0548 0.0515 0.0419 0.0476 \*\*\*\*\*\*\* 0.8436 0.3577 0.3383 0.0031\* 0.0000\* 0.0000\* 0.0000\* **Hui** 0.0547 0.0381 0.0457 0.0285 0.0380 0.0069 \*\*\*\*\*\*\* 0.2822 0.8369 0.1433 0.0009\* 0.0001\* 0.1848 **Pur** 0.0798 0.0676 0.0350 0.0415 0.0444 0.0135 0.0206 \*\*\*\*\*\*\* 0.0738 0.0290 0.0000\* 0.0001\* 0.0191 **Maz** 0.0461 0.0331 0.0353 0.0205 0.0306 0.0055 0.0095 0.0161 \*\*\*\*\*\*\* 0.6686 0.0048\* 0.0009\* 0.2460 **Tzo** 0.0469 0.0438 0.0235 0.0197 0.0302 0.0161 0.0138 0.0139 0.0071 \*\*\*\*\*\*\* 0.0000\* 0.0000\* 0.1290 **Toj** 0.023 0.0116 0.0245 0.0196 0.0062 0.0724 0.0615 0.0731 0.0545 0.0567 \*\*\*\*\*\*\* 0.0000\* 0.1650 **Lac** 0.1593 0.1385 0.0904 0.1052 0.1081 0.0411 0.0609 0.0241 0.0569 0.0470 0.1484 \*\*\*\*\*\*\* 0.0000\* **MY1** 0.0186 0.0193 0.0107 0.0043 0.0101 0.0316 0.0225 0.0309 0.0150 0.0139 0.0266 0.0828 \*\*\*\*\*\*\* \*Significant Fst *P* (\<0.01). For abbreviations see [Table 1](#T1){ref-type="table"}. ![Genetic distances (Fst) represented in multidimensional scaling plots based on D9S1120: (**A**) Mexicans plus reference populations from Europe (Eu), Asia (Asia), Africa (Afr), and from Colombia: Mulaló-Mestizos (Mul), Awa (Awa), Pijao (Pij), and Coyaima (Coy) ([@R12]). (**B**) Mexican Amerindian and Mestizo populations (abbreviations are shown in [Table 1](#T1){ref-type="table"}).](CroatMedJ_53_0423-F3){#F3} A pairwise comparison between Mexican Amerindian groups, including the Mayas ([@R12]), showed that the most differentiated populations were Lacandones, Tojolabales, Tarahumaras, Tzotziles, and Purépechas ([Table 4](#T4){ref-type="table"}). Interestingly, the pooled South American groups including the Awa-Kuaikier, Pijao, and Coyaima from Colombia ([@R12]) were similar to all five Mexican Mestizo populations (*P* \> 0.01; data not shown) and to the majority of Mexican Amerindian populations, except Lacandones. Individually, the Colombian Awa-Kuaikier population was different from Tarahumaras, Purépechas, Tzotziles, and Lacandones, whereas the Pijao group was different only from Lacandones and Tojolabales. Lacandones was the most distinct and markedly differentiated Mexican Amerindian group ([Figure 3B](#F3){ref-type="fig"}). Analysis of molecular variance (AMOVA) -------------------------------------- AMOVA based on D9S1120 showed a significant inter-population differentiation among all the Mexican populations (F~ST~ = 4.03%; *P* \< 0.0001). However, Amerindians showed larger and significant inter-population differentiation (F~ST~ = 2.81%; *P* \< 0.0001) than Mestizos (F~ST~ = 0.44%; *P* = 0.187). When we clustered Mestizos vs Amerindians, the differentiation both between the groups (F~CT~ = 2.19%; *P* = 0.0098) and within the groups was significant (F~SC~ = 1.84%; *P* \< 0.000), indicating that the population clustering was not robust. Although differentiation between Amerindians from Mexico and those from Colombia ([@R12]) was not significant (F~CT~ = -0.16%; *P* = 0.409), differentiation within the groups was (F~SC~ = 2.25%; *P* \< 0.0001). Finally, when population structure was assessed between Amerindians (Mexicans plus Colombians) and populations from Europe, Asia, and Africa ([@R12]), differentiation between the groups increased substantially (F~CT~ = 8.54%; *P* \< 0.0001). Comparison of D9S1120 with CODIS-STRs ------------------------------------- The genetic diversity represented by heterozygosity (Het) and the population differentiation coefficient (F~ST~) based on D9S1120 were compared with those previously obtained using 9 CODIS-STRs from the corresponding Mexican populations (Chihuahua, Jalisco, Veracruz, Yucatan, Purépechas, and the Maya). D9S1120 showed lower genetic diversity than most of the nine CODIS-STRs in most of the populations ([Figure 4A](#F4){ref-type="fig"}), (except D5S818 and D3S1358), but it showed higher diversity in Mestizos from Veracruz and the Mayas than four and five CODIS-STRs, respectively ([Figure 4A](#F4){ref-type="fig"}). D9S1120 had the ability to differentiate Amerindians from Mestizos, and Amerindian groups between each other. However, its ability to differentiate Mexican Mestizos was limited; in fact, it was just a little higher than that of the most of the CODIS-STRs ([Figure 4B](#F4){ref-type="fig"}). ![Comparison of the genetic diversity (Het) (**A**) and genetic differentiation (F~ST~) (**B**) based on D9S1120 and 9 CODIS-STRs in 6 Mexican populations ([@R13],[@R15],[@R20],[@R21]).](CroatMedJ_53_0423-F4){#F4} Discussion ========== In this study, the STR locus D9S1120 was characterized in Mexican Amerindian groups and the admixed Mestizo population to determine its forensic and anthropological potential; importantly, our findings could apply to other Latin American countries. A total of nine alleles were found in Native Americans and Mestizos, with very similar frequency distributions as in previous reports ([@R2],[@R3],[@R12]). Five out of 14 alleles previously described in worldwide surveys were not detected: alleles 11, 12, 17.3, 18.3, and 20; however, these D9S1120 alleles are rare in all populations ([@R11],[@R12]). Interestingly, in a previous report including 24 Native American (n = 426) and 53 worldwide populations (n = 1048), the allele 10 was only detected in the Maya, Ojibwa, and Cree populations ([@R3]). In our full Mexican Native American sample (n = 1414), the allele 10 was only observed in Purépechas but not in Mayan samples, confirming its very rare frequency, at least among the studied Mexican populations. Another interesting and uncommon allele is the allele 19, which was previously found in the Mayas and whose origin is either Native American or European ([@R12]). In our study, the allele 19 (315 bp) was observed in the Tzotzil native group and three Mestizo populations (\<1%), with a higher global frequency in Mestizos than in native groups (0.607 vs 0.141%), in line with the observations of Phillips et al ([@R12]). The modal alleles 9RA and 16 displayed a prevalence of 38.2% and 39.1% in Amerindian groups and Mestizos, respectively. Therefore, elevated frequency of allele 9RA and/or low frequency of allele 16 indicate Amerindian ancestry, whereas an opposite pattern indicates an admixture or non-native ancestry (eg, European). In Mestizos, this assumption was evaluated in view of the previously described increasing northwest to southeast gradient of Amerindian ancestry and the opposite gradient of European ancestry. This pattern has been consistently obtained with different genetic systems, such as CODIS-STRs, Y-STRs, and genome-wide single nucleotide polymorphisms analysis ([@R13],[@R14],[@R22]). While the allele 16 distribution is in agreement with such ancestry distribution, the 9RA distribution confirms it only partially because its frequency in northern Mestizos from Chihuahua was low (13.7%), and higher in other Mestizo populations (22%-24%). 9RA was prevalent in the majority of Mexican Amerindian groups, except in the Mayas and Tojolobales, where the modal allele was 16 (47.1 and 34.3%, respectively). There are contrasting explanations for such allele distribution. In the Mayas, the large number of alleles and the observed heterozygosity suggests admixture, which is in agreement with previous reports on elevated gene flow ([@R15]) and on cultural practices allowing marriage with non-Mayan individuals ([@R23]). Conversely, a reduced number of alleles observed in Tojolobales, in addition to their cultural and geographic isolation, suggests genetic drift and/or founder effect. Similarly, Lacandones showed the lowest genetic diversity and largest differentiation from other populations, also suggesting genetic drift effects. This finding, although only based on a single autosomal STR, confirms the conclusions obtained with Y-chromosome markers (Rubi-Castellanos et al, unpublished data 2012) and historical records of geographic and socio-cultural isolation of Lacandones ([@R24],[@R25]). The forensic parameters estimated in this study confirm that D9S1120 can be a useful tool for human identification and molecular anthropology. On average, genetic diversity values (Ho) were larger in Mestizos than Amerindian groups ([Table 2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}). In most Mexican populations, the informativeness of D9S1120 was lower than that of the majority of CODIS-STRs, largely due to the dominant, high frequencies of 9RA and allele 16. However, given their different distributions ([@R2],[@R3],[@R11],[@R12]), D9S1120 potentially allows the discrimination of Amerindian and Mestizo biological samples from those of different origin (eg, European, Asian, or African). We would advocate the use of this STR alongside uni-parental markers (ie, Y-chromosome and mtDNA) and particularly ancestry-informative markers in commercial STR human identification kits to provide the most robust ancestry identification in forensic samples. This would be useful in multi-ethnic countries with a large proportion of Hispanic populations of Amerindian origin (eg, United States). An interesting finding was that Mestizos and Amerindian groups from Mexico could not be differentiated from their counterparts from Colombia ([@R12]), which is in agreement with the hypothesis on the common origin of Native Americans ([@R2],[@R3]). Similarly, the population structure and differentiation patterns among some Mexican populations closely agreed with previous analyses based on autosomal STRs ([@R15],[@R26]) and Y-chromosome markers ([@R27]). Additionally, MDS plot stress-values (stress \<0.1) indicated a reasonable reliability of the genetic relationships. In summary, our results emphasize the potential of D9S1120 to differentiate Native American groups from Mestizos and other population groups. Further studies of this unique STR are required, especially in view of its application in the forensic practice. Funding This work was supported by the Consejo Nacional de Ciencia y Tecnología (CONACyT) with the grant No.129693 to HR-V. Ethical approval received from the Ethical Research Committee of the Academic Center of the Ciénega, University of Guadalajara. Declaration of authorship H-RV gave the initial proposal for the study and participated in statistical analysis, data interpretation, and writing of the manuscript. VMS-G provided the technical support and participated in data analysis and writing of the discussion. MB-R provided technical support and participated in statistical data analysis and writing of the discussion. JS-F participated in the statistical analysis and data interpretation. GM-C provided advice on the experimental analysis and took part in manuscript preparation. JFM-V participated in designing of the study and writing of the manuscript. CP devised materials/protocols used for the analyses, approved the study design and scope, and helped the lead author in writing of the manuscript. Competing interests All authors have completed the Unified Competing Interest form at *[www.icmje.org/coi_disclosure.pdf](http://www.icmje.org/coi_disclosure.pdf)* (available on request from the corresponding author) and declare: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

1. Introduction {#sec1} =============== Gastrointestinal stromal tumors (GIST) are rate tumors arising in the gastrointestinal tract. Imatinib mesylate (imatinib) is frequently used in these patients for treatment. However, the resection of the tumor is the primary aim of treatment. Frequently, mutations are noted in the kit or platelet-derived growth factor receptor-A (PDGFRA). The expression of genes and their possible mutations are important for treatment of GIST. The survival is primarily correlated with the stage of disease, the tumor size, and the proliferation. Rutkowski et al. differentiated the GIST according to location and stage and reported about a 5-year survival rate in gastric GIST of 96% for stage IA, 92% for IB, 51% for II, 22% for IIIA, and 22% for IIIB \[[@B1]\]. Setoguchi et al. evaluated the prognostic aspect of gene array data in GIST \[[@B2]\]. The authors were able to show that VCAN and CD9 are prognostic markers in gastric GIST. Disease-free survival was significantly longer for patients if CD9 was enhanced and expressed, and VCAN revealed a low expression. Another study by Schmieder et al. demonstrated that p16 is also important for the identification of high-risk GIST and p16 was predictive for poor outcome \[[@B3]\]. The literature results demonstrate that quantitative data about the expression of genes may be helpful to gain prognostic information. Positron emission tomography (PET), in particular PET-CT, can be used to obtain accurate information about the distribution of a tracer over time within a given volume if a dynamic acquisition, for example, for one hour is performed. Dynamic PET (dPET) and PET-CT (dPET-CT) data can be analyzed by applying compartment and non-compartment models to achieve detailed information about the tracer kinetics. F-18-deoxyglucose (FDG) is the most commonly used tracer for PET in oncology. The tracer kinetics of FDG is closely dependent on the expression of glucose transporters and hexokinases. However, it was shown that the kinetics of FDG may be modulated by genes associated with angiogenesis and proliferation \[[@B4], [@B5]\]. Thus, gene expression may be predicted from a detailed analysis of the FDG kinetics in certain tumors. However, no data exist about the correlation of gene expression and quantitative dynamic PET data in GIST. Therefore, the aim of this study was to assess the association of FDG kinetics and gene expression in GIST. 2. Material and Methods {#sec2} ======================= The study comprises 22 patients with GIST scheduled for surgery. Dynamic PET studies were performed prior to surgery within the routine diagnostics for tumor staging. All patients had pretreatment with imatinib, and surgery was intended to remove the tumor tissue. The body volume for the dynamic study had been determined according to the clinical information where the surgery was performed. Following the intravenous injection of 200--300 MBq FDG, a dynamic data acquisition was initiated for one hour. Then whole body imaging was performed following the dynamic study. The dynamic study comprises 28 frames, including 10 frames of 30 s, 5 frames of 60 s, 5 frames of 120 s, and 8 frames of 300 s. Following the iterative image reconstruction, the images were converted to standardized uptake value (SUV) images for further evaluation. The SUV was calculated according to the following: SUV = tissue concentration (Bq/g)/(injected dose (Bq)/body weight (g) \[[@B6]\]. In 20 of the 22 patients the dynamic studies could be evaluated using a 2-tissue-compartment model. Volumes of interest (VOI) were used for the quantitative assessment of the tumors. The quantitative evaluation was performed with a dedicated software and included the calculation of SUV and the maximum SUV. Furthermore, the tracer kinetics was analyzed using a 2-tissue-compartment model. For this purpose a VOI was placed over the descending aorta to obtain an input function for the model. It was already shown by Ohtake et al. that the input data for FDG can be obtained via VOIs from a large vessel with high accuracy \[[@B7]\]. The authors noted a median error of 3.75% for the image-derived results as compared to the arterial sampling data. No partial volume correction was needed, because the diameters of the vessels exceeded 8 mm, which correlates with a recovery factor of 0.85 for our system and the image reconstruction settings used for the study. The compartment analysis software provides five parameters: *k1* and *k2*, associated with the transport of FDG; *k3* and *k4*, which are associated with the phosphorylation and dephosphorylation; vb, the fractional blood volume of a VOI, also referred to as vessel density. Furthermore, a noncompartment model was applied to the data to obtain the fractal dimension (FD) of the time activity data. The positioning of the VOIs was done according to the information provided by the surgeons about the location, where the tissue specimen was removed from the tumor. Thus, we tried to get the quantitative dPET data spatially close to the region where the tissue specimen was obtained. Tumor resection was performed in the 22 patients. A small fraction of the tumor was used for gene array analysis. The tissue specimens were immediately stored in liquid nitrogen, and total RNA was extracted for further processing. The quality of isolated RNA was evaluated photometrically using the 280/260 ratio and on an agarose gel. We used the U133A 2.0 gene array (Affymetrix Inc., Santa Clara, CA, USA), which provides quantitative information about 54675 gene probes. The processing of the RNA and gene arrays was done according to the manufacturer\'s recommendations. Gene chip expression data were normalized for the beta-2 microglobulin (Affymetrix code 34644_at, Homo sapiens mRNA for beta-2 microglobulin) using the following equation: relative expression value (REV) = 1,000 × expression value of a gene/expression value for beta-2 microglobulin \[[@B8]\]. The statistical evaluation was performed with Stata/SE 11.2 (Stata Corporation, College Station, TX, USA) on a Mac Pro 2 × 2.93 GHz, 12 Core Intel Xeon system with 24 GB RAM using Mac OS X 10.7.3 (Apple, Cupertino, CA, USA). The same system was also used for all data processing tasks, including dPET and gene array data. A dedicated software (GenPET), developed by our group, was used for the correlative evaluation of dynamic PET and gene array data \[[@B9]\]. The software provides the correlative assessment of both PET and gene array data. For the correlation analysis, a significance level of *P* \< 0.05 was used. Based on the significant results of the correlation analysis, nonlinear regression functions were calculated for the gene array and PET data. 3. Results {#sec3} ========== The statistical data for the PET parameters are provided in [Table 1](#tab1){ref-type="table"}. The mean SUV was 6.597, nearly half of the SUV~max~. The fractional blood volume (vb), associated with the vessel density, was 0.108, which reflects about 11% of the tumor volume evaluated by VOIs. The maximum vb was 0.253, which refers to a blood volume fraction of about 1/4 of the tumor volume. As expected, *k1* (FDG transport) was high, and *k3* (FDG phosphorylation) was relatively low in these tumors. Several studies are focused on the gene expression patterns in GIST. Rink et al. compared the gene expression signatures with the response to imatinib mesylate treatment in 63 patients \[[@B10]\]. Overall, the authors report about 38 genes, which were expressed lower prior to therapy and were associated with response to treatment. Interestingly, 18 of the 38 genes are related to the genes of the zinc finger group. The function of all zinc finger genes is not well known, but it is likely that they are involved in signaling and cell regulation mechanisms. Based on the results from Rink et al., we evaluated the zinc finger gene data for possible correlations with the PET data. We used a nonlinear regression approach and combinations of two or three PET variables for the correlation/regression analysis. Significant correlations (*r* \> 0.8) were found for znf43, znf85, znf91, and znf189. The PET variables used for the nonlinear regression functions are shown on the *x*-axis in [Figure 1](#fig1){ref-type="fig"}, while the *y*-axis reflects the corresponding gene expression data (unit for gene expression data: REV). Besides the zinc finger genes, we noted also a significant difference for high (\>0.441) and low *k1* values and the expression of VEGF-A, an angiogenesis-related gene ([Figure 2](#fig2){ref-type="fig"}). Interestingly, this had been already reported, for example, for colorectal tumors and gene array data \[[@B4]\]. The cyclin-dependent kinase inhibitor 1C has an impact on cell proliferation and acts as a tumor suppressor gene. A correlation (*r* = 0.85) was noted for this gene and the maximum SUV ([Figure 3](#fig3){ref-type="fig"}). 4. Discussion {#sec4} ============= In patients with GIST, PET with FDG is a common procedure for both, diagnostic purpose for staging and follow to assess the response to treatment. Apostolopoulos et al. evaluated 65 dynamic PET-FDG studies in patients with liver metastases from GIST \[[@B11]\]. He used parametric imaging for the data evaluation and report about an accuracy of 87.7% and a sensitivity of 88.2%. Gayed et al. examined 54 patients with PET-FDG and reported about a sensitivity of 86% \[[@B12]\]. The results demonstrate that PET is useful to identify GIST lesions. Dynamic PET studies provide the assessment of details of the tracer kinetics using compartment and noncompartment models. The 2-tissue-compartment model is the standard for the assessment of FDG dynamic data. One major advantage of dPET is the possibility to obtain the input function for the 2-tissue model from the images without arterial blood sampling. We calculated all five parameters of the model using a dedicated software developed by our group. The software uses a modified support vector machines algorithm to calculate the model parameters. The dephosphorylation rate of FDG may be low in tumors, but we included *k4* also in the model calculations to achieve most accurate results. One aspect of major importance is the correlation of PET results with molecular biological data. Park et al. compared in 26 patients with gastric GIST the maximum SUV (SUV~max~) with the Ki-67 index \[[@B13]\]. The authors obtained a correlation of *r* = 0.854 for the two parameters. Based on our results, the SUV~max~ was generally enhanced and above 10 SUV. The compartment data suggested a primary contribution of vb and *k1* to the global FDG uptake. However, more data are needed to evaluate the dependency of the global uptake and maximum uptake on tumor proliferation. Rink et al. identified genes, correlated with response to imatinib mesylate treatment \[[@B10]\]. Most of the genes were zinc finger genes. These genes are usually associated with the regulation of transcription. We noted significant correlations for the PET parameters and ZNF43, ZNF85, ZNF91, and ZNF189 ([Figure 1](#fig1){ref-type="fig"}). Primarily, the SUV and SUV~max~ were correlated with the zinc finger genes. However, SUV and SUV~max~ alone correlate only with ZNF189; therefore, the results from the dPET are required to predict the gene expression accurately. Rink et al. were able to demonstrate that a subgroup of zinc finger genes are predictive for the treatment outcome in patients treated with imatinib mesylate \[[@B10]\]. Furthermore, they performed knock down experiments, for example, with ZNF43, ZNF85, and ZNF91, and they showed that depletion of each of the ZNFs resulted in an improved sensitization to imatinib mesylate. Thus, predicting the expression of these genes from quantitative, dynamic PET results may be a promising approach for an improved evaluation of treatment response in GIST. Besides the zinc finger genes, other genes, for example, associated with angiogenesis, are important for the treatment results. We noted a significant difference for high and low *k1* data and the expression of VEGF-A ([Figure 2](#fig2){ref-type="fig"}). The difference of the two groups is significant with *P* \< 0.0197. The same limit for *k1* (0.441) was used as in the publication about colorectal tumors and PET \[[@B4]\]. Therefore, it can be assumed that *k1* is generally modulated by VEGF-A expression. Overall, tumors with a *k1* less than 0.441 had a lower expression of VEGF-A as compared to tumors with a higher *k1*. However, this should be shown also for other tumor types, but we can assume that the dependency of *k1* on the VEGF-A expression is most likely not dependent on the tumor type. Thus, *k1* may be used as a general classification parameter to predict VEGF-A expression in a malignant lesion. The role of the cyclin-dependent kinase inhibitor 1C (cdki 1C) is described in detail in the review paper of Kavanagh and Joseph \[[@B14]\]. cdki 1C is involved in several regulatory processes, including tumor differentiation and angiogenesis, apoptosis, cell invasion, and metastasis. The gene acts as a tumor suppressor gene. Therefore, we noted a lower expression of cdki 1C with higher SUV~max~ values ([Figure 3](#fig3){ref-type="fig"}). Actually, no publications exist about cdki 1C and FDG in PubMed, so it is not possible to compare our results with others. Actually it can be expected that inhibitors of the cell cycle usually reveal a negative correlation to PET parameters. Based on our results in colorectal carcinoma, we were able to show that cdki 2B was negatively correlated with *k3* \[[@B5]\]. In contrast, cdk2 was positively correlated with the SUV in these tumors. Overall, the correlation between cdki 1C and SUV~max~ is in agreement with the conclusions of Park et al. who found a correlation for the SUV~max~ and ki-67 \[[@B10]\]. Thus, the SUV~max~ is likely to be predictive for tumor proliferation in GIST. {#fig1} {#fig2} {#fig3} ###### Statistical data for the PET parameters. SUV SUV~max~ vb *k1* *k2* *k3* *k4* fd influx --------- -------- ---------- ------- ------- ------- ------- ------- ------- -------- Mean 6.597 13.029 0.108 0.446 0.787 0.070 0.004 1.306 0.036 Median 6.599 12.104 0.066 0.423 0.802 0.067 0.001 1.448 0.030 Minimum 1.406 2.286 0.028 0.199 0.542 0.003 0.000 1.035 0.003 Maximum 16.935 29.280 0.253 0.801 0.928 0.188 0.013 1.448 0.105 no. 20 20 20 20 20 20 20 20 20 [^1]: Academic Editors: P. Hartvig, T. Momose, and L. Roncucci

PAAR proteins are characterized by 3 proline-alanine-alanine-arginine motifs and by the ability to coordinate a zinc ion by 3 histidine and one cysteine residues. This metal binding is thought to stabilize their 3D structure. PAAR proteins are localized at the tip of T6SS where they interact with the trimeric VgrG protein. In our recent study, we identified that IglG and DUF4280 proteins are predicted to adopt a similar fold as PAAR proteins. IglG and DUF4280 proteins are characterized by 4 highly conserved cysteine residues. These 4 cysteine residues are essential for IglG function in the T6SS, for virulence in a mouse model of tularemia and were demonstrated to contribute to the coordination of a metal ion (either zinc or iron). The predicted structure of IglG and DUF4280 consensus sequences and their ability to bind metal led us to propose that IglG and DUF4280 proteins are bona fide PAAR-like proteins. Present in numerous proteins of unknown function, the DUF4280 domain (PF14107) is a domain of about 100 amino acids with 4 conserved cysteine residues. It is found in more than 250 bacterial species, including both Gram-positive and Gram-negative bacteria. A phylogenetic analysis demonstrates that PAAR-like proteins form a clade distinct from typical PAAR proteins and from their bacteriophage homologues (Gp5.4 proteins), suggesting an ancient divergence (Figure 1). The presence of PAAR-like proteins in Gram-positive bacteria (highlighted in blue in figure 1) is remarkable, given the fact that T6SS have been described so far only in Gram-negative bacteria. DUF4280 proteins from Gram-positive species appear as relatively grouped in the phylogenetic tree. Yet, their sequences do not diverge much from Gram-negative DUF4280 sequences, suggesting a common history (and a possible common function) for the PAAR-like proteins in Gram-positive and Gram-negative bacteria. This raises the fascinating possibility that Gram-positive bacteria may possess functional phage-like translocation system sharing homology with Gram-negative T6SS. {#Fig1} In the PAAR-like phylogeny, *Francisella* species (highlighted in red) appear as phylogenetic outliers in line with previous phylogenetic analyses performed by Bingle and collaborators using TssB/C (IglA/B) sequences. The divergence of *Francisella* PAAR-like protein and T6SS might be linked to its unique ability to target the eukaryotic phagosomal membrane. With the characterization of IglG as a PAAR-like protein, the three *Francisella* homologues of the canonical T6SS inner tube proteins are now identified. IglC, VgrG and IglG have structural homology with the Hcp, VgrG and PAAR proteins. Yet, experimental evidence is still lacking to demonstrate that these proteins function together to form the *Francisella* T6SS inner tube. Surprisingly, while IglG is fully required for the intracellular life cycle of *F. novicida*, it is only delaying the intracellular life cycle of a highly virulent *F. tularensis* strain. This result, although not fully understood, suggests that in some bacteria, T6SS might have a residual activity in absence of PAAR/PAAR-like proteins. Importantly, deletion of *iglG* renders all the tested strains completely avirulent in a mouse model of tularemia, indicating that a fully functional T6SS with a PAAR-like protein is required *in vivo* for virulence. As of today, it is still unknown whether the FPI encodes a T6SS effector, which would be directly responsible for the vacuole lysis. Importantly, similarly to some VgrG proteins, some PAAR and PAAR-like proteins feature N- or C-terminal extension with predicted enzymatic domains (e.g. endonuclease or lipase) suggesting that PAAR and PAAR-like domain could act as cargo to deliver effectors into the target cell. IglG possesses a short amphiphilic α-helical extension in the N-terminus of the PAAR-like domain. This IglG domain binds IglF, another FPI protein required for virulence. The role of IglF remains to be established but based on its interaction with IglG, we speculate that IglF might be translocated into the host cell together with IglG and act as an effector. Deciphering the mechanism controlling vacuole lysis in relation to the activity of this atypical T6SS is without doubt one of the most exciting challenge in the *Francisella* field. [^1]: Conflict of interest: The authors report no conflict of interest. [^2]: Please cite this article as: Claire Lays, Eric Tannier, Thomas Henry (2016). *Francisella* IglG protein and the DUF4280 proteins: PAAR-like proteins in non-canonical Type VI secretion systems? **Microbial Cell**3(11): 576-578. doi: 10.15698/mic2016.11.543

Introduction {#s1} ============ Improvements of phylogenetic methods and the availability of numerous placental mammalian genomes provide an invaluable resource to investigate controversial relationships with increasingly larger datasets and refined methods. The first phylogenomic studies utilising protein-coding sequences of mammals were optimistic to fully resolve the placental mammalian tree [@pone.0060019-Hallstrm1], [@pone.0060019-Wildman1]. However, some of the branches proved to be difficult to resolve even with these increasingly gene and taxon rich datasets and some relationships did not hold up to closer inspection [@pone.0060019-Hallstrm2]--[@pone.0060019-Hallstrm4]. Revisiting such groups using phylogenomic and network analyses provided new insights into placental mammalian evolution [@pone.0060019-Hallstrm3], [@pone.0060019-Hallstrm4]. One placental mammalian group with uncertain phylogenetic relationships among its orders is Euarchontoglires. This clade has been intensively studied through the use of concatenated genomic sequence data, retroposon insertion analyses, and multi-locus analyses of genomic data [@pone.0060019-Murphy1]--[@pone.0060019-Song1]. While some studies agree and give convincing support for internal branches [@pone.0060019-Hallstrm2], [@pone.0060019-Hallstrm3], [@pone.0060019-Kriegs1], [@pone.0060019-Song1], the major difference was the placement of the tree shrews relative to primates and Glires. In phylogenomic analyses a grouping with the primates was preferred, but the alternatives could not formally be rejected [@pone.0060019-Hallstrm2], [@pone.0060019-Hallstrm3]. The conflicting results of their relationships in earlier studies makes the Euarchontoglires an interesting group to revisit with a significantly increased genomic sequence data set using a multi-locus analysis approach. Recent coalescent-based analyses seem to have solved the question of resolving Euarchontoglires [@pone.0060019-Song1], with results being congruent with previous phylogenomic studies [@pone.0060019-Hallstrm2]. Euarchontoglires is a super-clade of placental mammals that includes primates (apes, monkeys and allies), rodents (mouse, rat and guinea pig), lagomorphs, (rabbit and hares), dermopterans (flying lemurs) and scandentians (tree shrew), and has recently been established exclusively through molecular analyses [@pone.0060019-Murphy1], [@pone.0060019-Asher1]. The original proposal of the Archonta clade [@pone.0060019-Gregory1] initially included Chiroptera (bat and flying foxes) and then Macroscelidea (elephant shrews), but these groups were later moved to other parts of the placental mammalian tree [@pone.0060019-Pumo1]--[@pone.0060019-McKenna1]. While the signal from molecular data for the monophyly of Euarchontoglires is strong, some details of the relationships within this clade remain ambiguous. In particular, the position of Scandentia differed in the numerous molecular studies, from being sister group to the primates [@pone.0060019-Murphy1], [@pone.0060019-Kriegs1], [@pone.0060019-Song1], [@pone.0060019-McKenna1], [@pone.0060019-Novacek1], sister group to lagomorphs [@pone.0060019-Schmitz1], [@pone.0060019-Arnason2], sister group to Glires (rodents plus lagomorphs) [@pone.0060019-Hallstrm3], [@pone.0060019-Waddell1] to being the first ordinal branch among the Euarchontoglires [@pone.0060019-Kjer1]. One large phylogenomic analysis using concatenated data from 2.9 million nucleotides (16 species) could reject a sister group relationship of Scandentia to primates plus Glires, but not Scandentia and primates as sister groups as an alternative to a first divergence among Euarchontoglires [@pone.0060019-Hallstrm3]. The branches of the placental mammalian tree that have been problematic to resolve are typically short with subsequent divergences occurring within 1--3 million years (Myr) [@pone.0060019-Hallstrm3], [@pone.0060019-Kumar1]. This duration is approximately the time-span establish a mammalian species [@pone.0060019-Avise1]--[@pone.0060019-vanDam1] and a time frame in which introgression can occur and incomplete lineage sorting can later complicate phylogenomic analyses [@pone.0060019-Maddison1]. Speciation processes appear to be the main reason for the incongruent results of phylogenomic attempts to resolve the mammalian species tree. Therefore, we revisit the Euarchontoglires with a new and larger set of genomic sequence data focusing on ordinal relationships within the Euarchontoglires along with the relationship of the tarsiers (haplorhine primates) relative to the other two major primate lineages, Strepsirrhini (lemurs, lorises and allies) and Anthropoidea (platyrrhines and catarrhines). The widely used concatenation method to infer the species tree can mislead inferences, especially when gene trees are in the so-called anomaly zone [@pone.0060019-Degnan1]. In addition to this, species trees generated using concatenated data can sometimes have very high bootstrap support values for incorrect relationships [@pone.0060019-Seo1]. Coalescent theory tries to trace back alleles to the most recent common ancestor [@pone.0060019-Kingman1]. Recently developed tree reconstruction methods such as STAR, species tree based on ranks of coalescent [@pone.0060019-Liu1] and MP-EST, maximum pseudo-likelihood estimates method [@pone.0060019-Liu2] estimates a species tree from multiple gene trees. STAR and MP-EST methods can be applied to large data sets because of their faster and simpler algorithms. They provide reliable results in reasonable computational time compared to more computationally intensive and exact Bayesian species tree methods (such as BEST), which are preferable for smaller data sets [@pone.0060019-Liu1]. Compared to recent phylogenomic studies [@pone.0060019-Hallstrm3], [@pone.0060019-Song1], [@pone.0060019-Jameson1] the new dataset now includes sequences of the Gibbon genome (*Nomascus leucogenys*). Genome data of the cow (*Bos taurus*) is used to root the tree. Among the possible outgroups to the Euarchontoglires [@pone.0060019-Hallstrm3] the bovine genome is arguably the one that has the best sequence coverage and annotation. The root is especially crucial for the coalescence-based species tree analyses such as STAR and MP-EST. Choosing a well-assembled genome of a closely related species facilitates the alignment and maximizes the amount of data that can be utilized. The placement of Dermoptera is also uncertain and previous studies have placed this order with the primates [@pone.0060019-Janecka1] or with Scandentia [@pone.0060019-Murphy1]. However, until a genome of this order is available, their position in the mammalian tree cannot be studied by phylogenomic approaches. Materials and Methods {#s2} ===================== The coding sequences (CDS) for the 19 species that are included in this study, listed in [Table 1](#pone-0060019-t001){ref-type="table"}, were retrieved using Biomart from the Ensembl version 65 (<http://www.ensembl.org/biomart/martview/>). The procedure for alignment and phylogenomic analyses were similar to that previously described [@pone.0060019-Hallstrm3]. Therefore it is only briefly described here, detailing only additional procedures. 10.1371/journal.pone.0060019.t001 ###### List of species included in the study and the percent coverage of alignment. {#pone-0060019-t001-1} Common name Binomial name Order Coverage of alignment (%) -------------- ------------------------------ -------------- --------------------------- Chimpanzee *Pan troglodytes* Primates 95.4 Human *Homo sapiens* Primates 99.9 Gorilla *Gorilla gorilla* Primates 92.6 Orangutan *Pongo abelii* Primates 91.5 Gibbon *Nomascus leucogenys* Primates 94.0 Macaque *Macaca mulatta* Primates 89.9 Marmoset *Callithrix jacchus* Primates 92.3 Tarsier *Tarsius syrichta* Primates 66.2 Bushbaby *Otolemur garnettii* Primates 93.6 Mouse lemur *Microcebus murinus* Primates 71.2 Tree shrew *Tupaia belangeri* Scandentia 82.5 Mouse *Mus musculus* Rodentia 96.5 Rat *Rattus norvegicus* Rodentia 88.6 Kangaroo rat *Dipodomys ordii* Rodentia 70.2 Guinea pig *Cavia porcellus* Rodentia 91.5 Squirrel *Ictidomys tridecemlineatus* Rodentia 54.4 Pika *Ochotona princeps* Lagomorpha 71.6 Rabbit *Oryctolagus cuniculus* Lagomorpha 82.7 Cow *Bos taurus* Artiodactyla 94.5 Coverage of alignment is the percent sequence coverage in 9,799,170 nucleotide long alignment. Human sequences longer than 300 nucleotides were used to find the orthologs from 18 different species using the recursive BLAST approach [@pone.0060019-Hallstrm2]. Only genes represented by at least 16 out of 19 species were kept for further analyses. All sequences were translated into corresponding amino acids and aligned using MAFFT version 6.833b [@pone.0060019-Katoh1]. Any alignment showing an overall nucleotides difference larger than 25% between any two species were discarded to ensure a conservative approach and further reduce the potential for incorrect alignments. It has been shown that the quality of multiple sequence alignment is essential in phylogenetic inference [@pone.0060019-Ogden1]. Therefore, we used BMGE, Block Mapping and Gathering with Entropy, [@pone.0060019-Criscuolo1] that utilizes similarity matrices such as BLOSUM and PAM to remove ambiguously aligned regions. We used the option of stringent trimming based on the scoring scheme BLOSUM 95. The selected amino acid alignments were then back translated to nucleotide-alignments. Both types of sequence data were analysed. Base composition analysis was done using the Treefinder (TF) version of March 2011 [@pone.0060019-Jobb1] to test for the compositional equilibrium of the bases across the species. For this we applied the sliding window approach using the default size of 500. It is prohibitive to run the model-test for large genomic scale data sets because of the large computational demands [@pone.0060019-Hallstrm2]. Similar phylogenomic studies [@pone.0060019-Hallstrm2], [@pone.0060019-Hallstrm4] estimated the preferred model to be GTR [@pone.0060019-Lanave1] and WAG2000 [@pone.0060019-Whelan1] with 4 gamma rate categories (4G) for nucleotide and amino acids respectively. These models were also estimated for a smaller dataset of 1,006 loci using the model test of TF. Assuming the remaining data set to have similar properties, these models were used subsequently for all maximum-likelihood (ML) analyses. Initially concatenated data sets for both nucleotide and amino acid sequences were used to infer the species tree using TF, reconstructing a ML tree using the GTR and WAG2000 models, respectively, and 4 gamma rate categories (4G) as rate heterogeneity parameter. ML hypothesis testing of different topologies for Scandentia and the tarsier within the Euarchontoglires were done with the approximately unbiased test (AU) [@pone.0060019-Shimodaira1] and Shimodaira- Hasegawa (SH) [@pone.0060019-Shimodaira2], as implemented in TF. These analyses are consistent with previous phylogenomic studies, thereby enabling comparisons to previous results. In the second step we analysed the nucleotide data using coalescent methods. We constructed the nucleotide species tree using the coalescent model of evolution implemented in the STAR method and with the MP-EST method implemented as an R package in Phybase [@pone.0060019-Liu3]. The STAR method is motivated by multispecies coalescent model [@pone.0060019-Rannala1], which assumes deep coalescence to be the major factor for the differences between the gene trees and species tree. STAR uses the rank of coalescence to coalesce the gene trees into species tree. For this analysis each taxon must be represented by a sequence for each gene. From the 5,875 gene data set, 1,006 gene alignments fulfilled this criterion. We also constructed the concatenated tree using the nucleotide and amino acid data. The 1,006 ([Table S1](#pone.0060019.s003){ref-type="supplementary-material"}) loci were used for the species tree reconstruction. The longest sequence length is 25,728 base pairs and average length was 1,757 base pairs ([Figure S1](#pone.0060019.s001){ref-type="supplementary-material"}). Computational constraints limited the multi-locus bootstrapping [@pone.0060019-Seo1] to 100 replicates, using the bootstrap \'mulgene\' method implemented in Phybase. Each gene was bootstrapped and the combined trees (100,600) served as input for the STAR program. Individual ML gene trees were generated for each alignment using PhyML 3.0 [@pone.0060019-Guindon1] applying the GTR model with gamma distribution. From these trees, 100 re-sampled species trees were generated using the STAR method implemented in R. The species trees were rooted with the cow and finally a consensus tree was made using the "consense" module in Phylip, version 3.69 [@pone.0060019-Felsenstein1]. A second coalescent-based method "MP-EST", which implements a pseudo maximum likelihood method under the coalescent model to estimate the species tree, was also used to generate a species tree. The MP-EST method has been shown to be more accurate than STAR when inferring short branches in a species tree [@pone.0060019-Liu2]. All 1,006 gene trees were again used as input for the MP-EST method to generate the species tree. The procedures of tree construction, including the bootstrap analyses, were similar to that described for the STAR method. We also performed network analyses to depict conflicting signal. From the 1,006 amino acid sequence alignment, individual ML trees were generated by TF. A consensus network was built from the individual gene trees using the SplitsTree4 program [@pone.0060019-Huson1], with a threshold of 10%. In addition, we selected the best ML tree from the three alternative hypotheses regarding the position of the tree shrew. Only gene trees that were separated from the second best ML tree by an arbitrary value \>0.7 standard deviations (s.d.) were retained and used for network construction. Choosing a cut-off of 0.7 s.d. allows best depicting the conflict in the networks and retains only topologies with some moderate support from single gene analyses. The three different placements of the tree shrew as earlier described were evaluated for all the three different topologies of tree shrew. A consensus network in Splits Tree4 then summarized the trees. Results {#s3} ======= The final dataset consisted of 5,875 orthologous gene alignments from 19 different species, including ten primates, seven rodents and lagomorphs, one scandentian, and one outgroup species, the cow ([Table 1](#pone-0060019-t001){ref-type="table"}). After trimming the alignments using BMGE [@pone.0060019-Criscuolo1], we generated a concatenated alignment of 9,799,170 nucleotides with an average sequence coverage of 85% for each species ([Table 1](#pone-0060019-t001){ref-type="table"}). This resulted in 71% more sites than a previous phylogenomic studies that included Euarchontoglires [@pone.0060019-Hallstrm3]. The base composition showed high homogeneity between the species both for all codon positions (NT123) and first and second codon position alone (NT12) ([Table S2](#pone.0060019.s004){ref-type="supplementary-material"}, [Table S3](#pone.0060019.s005){ref-type="supplementary-material"}). The ML consensus trees from concatenated amino acid and nucleotide data supported different topologies for the position of Scandentia. The nucleotide analysis supported a sister group relationship between Scandentia and the primates, with 100% support both including and excluding the third codon position ([Figure 1](#pone-0060019-g001){ref-type="fig"}). The amino acid ML analysis, involved 3,266,390 sites and found Scandentia as the outgroup to both the primates and Glires, albeit with negligible support ([Figure 2](#pone-0060019-g002){ref-type="fig"}). The three proposed hypotheses for the position of Scandentia were tested by ML analyses and the results are summarized in [Table 2](#pone-0060019-t002){ref-type="table"}. The support from both the resampling and comparative likelihood tests for the position of Scandentia is ambiguous when analysing the amino acid sequences, supporting the hypothesis of Scandentia either as outgroup to primates or to Glires. Yet, nucleotide data analyses unambiguously support the Scandentia and Primate sister group relationship both when tested with all the three codon positions and first two codon positions. {#pone-0060019-g001} {#pone-0060019-g002} 10.1371/journal.pone.0060019.t002 ###### ML tests statistics for different relationships of the Scandentia within the Euarchontoglires. {#pone-0060019-t002-2} Topologies pSH (AA) pAU (AA) pSH (NT12) pAU (NT12) pSH (NT123) pAU (NT123) ------------------------------------ ---------- ---------- ------------ ------------ ------------- ------------- **((Scandentia,Primates),Glires)** 0.23 \<0.001 1 0.11 1 0.23 **((Scandentia,Glires),Primates)** \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 **(Scandentia,(Primates,Glires))** 1 0.94 \<0.001 \<0.001 \<0.001 \<0.001 pSH (probability Shimodaira Hasegawa) and pAU (probability Approximate Unbiased) ML test values are shown. The concatenated tree analysis of the nucleotide sequence data from 1,006 genes was congruent with the analysis of the 5,875 genes, but lacked strong support most likely due to the reduced amount of data. The species tree topology constructed by STAR, using the 1,006 gene trees, supported the concatenated nucleotide consensus species tree using the 5,875 loci ([Figure 3](#pone-0060019-g003){ref-type="fig"}). The bootstrap support for the Scandentia-primates relationship was 94%. MP-EST yielded the same topology for the species tree with a bootstrap support of 86% for the Scandentia-primates branch ([Figure 3](#pone-0060019-g003){ref-type="fig"}). Thus, both coalescent multi gene analyses of nucleotide sequences support the Scandentia-primates relationship with high bootstrap support values. {#pone-0060019-g003} The position of the tarsier within Euarchontoglires is established with high support in all analyses. The tarsier is placed as sister taxon to the Simiformes with both the concatenated nucleotide and amino acid data analysis ([Figure 1](#pone-0060019-g001){ref-type="fig"}, [2](#pone-0060019-g002){ref-type="fig"}). Statistical ML analyses of the three different hypotheses for the position of the tarsier and for all types of sequences data reject alternative positions ([Table 3](#pone-0060019-t003){ref-type="table"}). Also, the coalescent based methods supported the same topology as the concatenated analysis with 100% bootstrap support, suggesting the tarsier as sister taxon to the Simiiformes ([Figure 3](#pone-0060019-g003){ref-type="fig"}). 10.1371/journal.pone.0060019.t003 ###### ML tests statistics for different relationships of the tarsier within the Euarchontoglires. {#pone-0060019-t003-3} Topologies pSH (AA) pAU (AA) pSH (NT12) pAU (NT12) pSH (NT123) pSH (NT123) -------------------------------------------- ---------- ---------- ------------ ------------ ------------- ------------- **((Tarsier,Anthropoidea),Strepsirrhini)** 1 0.23 1 0.23 1 0.23 **((Tarsier,Strepsirrhini),Anthropoidea)** \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 **(Tarsier,(Anthropoidea,Strepsirrhini))** \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 \<0.001 pSH (probability Shimodaira Hasegawa) and pAU (probability Approximate Unbiased) ML test values are shown. The consensus network analysis of all 1,006 genes that are represented in all species shows conflicting branches when limited to splits that are present in at least 10% of the data (10% threshold value) shown in [Figure 4](#pone-0060019-g004){ref-type="fig"}. The major signal from this single gene analyses places the tree shrew as sister group to the primates, or basal to all Euarchontoglires. Depicting the position of the tree shrew in a network based on the best ML trees from three alternative hypotheses of the tree shrew position, yielded no further resolution or insight into the evolutionary process ([Figure S2](#pone.0060019.s002){ref-type="supplementary-material"}). {#pone-0060019-g004} Discussion {#s4} ========== The increase in the availability of complete mammalian genomes has been seen as an opportunity to fully resolve all branches in the mammalian tree. Yet, numerous studies using genome scale data [@pone.0060019-Hallstrm2]--[@pone.0060019-Hallstrm4], [@pone.0060019-DosReis1] yielded the insight that species tree reconstruction is complicated by incomplete lineage sorting and possibly introgression. One major group that was problematic to resolve is the Euarchontoglires clade. In many studies the branches connecting Scandentia and tarsier to the tree were found to be difficult to place within this super-order. The reason for this is that short internal branches have been identified to be a major cause of uncertainty in most phylogenomic studies [@pone.0060019-Hallstrm2], [@pone.0060019-Hallstrm3], [@pone.0060019-McCormack1]. By increasing the taxon sampling and taking advantage of coalescent-based analytical approaches this study investigated the relationships within Euarchontoglires in more detail and with new analytical approaches. It has been shown that in multi-locus data analysis, gene tree heterogeneity and the conflicts arising due to different gene trees complicate the process of correctly inferring the species tree [@pone.0060019-Maddison1]. Gene duplication, incomplete lineage sorting and deep coalescence are obstacles in correctly inferring the species tree. Also, the entire process of inferring the species tree becomes more complicated as the amount of multi-locus data increases. To avoid analytical problems due to gene duplication we selected strictly orthologous genes. For getting a comprehensive representation of the ML tree, we began the analysis with the normal concatenation method, which has been used in numerous studies of mammalian evolution [@pone.0060019-Hallstrm1], [@pone.0060019-Wildman1], [@pone.0060019-Hallstrm3], [@pone.0060019-Song1], [@pone.0060019-DosReis1]. The resulting ML tree supported the general consensus that Scandentia is the sister group to primates. Thus, focusing on the resolution of the Euarchontoglires and increasing the dataset, yielded robust results [@pone.0060019-Murphy1], [@pone.0060019-Song1], [@pone.0060019-Novacek1]. Only the analysis of amino acid sequences failed to provide the statistical support for the position of Scandentia. Yet, testing different hypotheses for the position of Scandentia yielded high support from nucleotide sequences for the topology shown in [Figure 1](#pone-0060019-g001){ref-type="fig"} and [3](#pone-0060019-g003){ref-type="fig"}. It has been debated [@pone.0060019-Russo1]--[@pone.0060019-Fong1] whether nucleotide or amino acid data contains more reliable phylogenetic information. Generally, the use of amino acid data is advised, because amino acid sequences are expected to be less randomized than nucleotide sequences for ancient divergences [@pone.0060019-Townsend1]. However, by using a set of selected genes it has been shown that nucleotide sequence data can outperform the amino acid sequence data for phylogenetic information on time scales of less than 500 million years [@pone.0060019-Townsend1]. In this study we analyse time scales in the range of 80--90 Ma and with short divergence intervals, where coalescence and introgression can complicate phylogenetic analysis. Under these conditions amino acid sequence data may be too conserved to contain sufficient phylogenetic information. It has been observed that the approach of using concatenated sequences under certain conditions can obscure important phylogenetic signal [@pone.0060019-Kubatko1]. Incongruent gene-trees can mislead phylogenetic analyses of concatenated sequences and result in erroneous interpretations of the species relationships and sometimes the incorrect species trees receive high support values [@pone.0060019-Edwards1]. The solution to this is the analysis of individual genes and their associated evolutionary signal (gene-tree) in a coalescence based framework to recover a final species tree [@pone.0060019-Rannala1]. STAR [@pone.0060019-Liu1] is one such method for species tree reconstruction and has been successfully used for studying the phylogeny of placental mammals from protein coding genes [@pone.0060019-Song1] and from ultra conserved element sequences within mammals [@pone.0060019-McCormack1]. Similar to a recent phylogenomic study using multi-locus analyses [@pone.0060019-Song1], albeit with more than twice as much data and using a strict approach to identify orthologs, we find a clear support of Scandentia being the sister group to primates. The large data set used in this study found the same species tree in the multi locus STAR analysis as in the concatenated analysis from nucleotide sequence data. Thus, analyses of large concatenated data sets can yield the same phylogenetic results as multi-locus analyses. Congruence between the approaches increases the confidence to have identified the historic species tree, despite the conflict in individual gene analyses that is revealed by networks. For theoretical reasons a full multi-locus coalescent analysis had been preferable, but this approach is still prohibitive for large datasets. The dramatic increase in data was only possible by focussing on the phylogeny of the Euarchontoglires phylogeny and choice of a well sequenced outgroup. To further consolidate the results obtained from the STAR analysis we employed MP-EST analysis [@pone.0060019-Liu2] as an additional coalescent-based method. It has been shown to be equally reliable as STAR [@pone.0060019-Liu1], but uses a pseudo-likelihood method in the environment of coalescence theory. The results from MP-EST are congruent with the STAR method, yielding identical topology with high support values. Both methods, STAR and MP-EST require more data to reconstruct confident species trees, because they use a partial parametric method and summary statistics [@pone.0060019-Liu1], [@pone.0060019-Liu2]. However, compared to fully parametric methods, STAR and MP-EST allow analysing large and taxon-rich datasets within reasonable time. In contrast to the congruence of the concatenated and multi-locus coalescent analyses, the analysis of a smaller data set (447 genes) on the whole mammalian tree [@pone.0060019-Song1] found differences in the two approaches. The Scandentia grouped with the primates using multi-locus analyses, a result favoured by the authors for the new methodology [@pone.0060019-Song1]. The concatenated analyses grouped the Scandentia with Rodentia [@pone.0060019-Song1], however, this grouping that has been rejected previously [@pone.0060019-Hallstrm2], [@pone.0060019-Hallstrm3]. Regardless of the analytical approach, conflict in phylogenetic data needs to be shown either by careful ML analyses of alternative trees or by the phylogenetic signal from single gene analyses. The phylogenetic signal of conflicting data from single gene trees can be ideally depicted by network analyses [@pone.0060019-Huson2]. The network depicts the previous difficulties to resolve the relationship of the Scandentia by sequence analyses even from concatenating genome sequence data [@pone.0060019-Hallstrm3], [@pone.0060019-Hallstrm4], [@pone.0060019-Song1] with nearly equally long edges, but no connection to the rodents. Interestingly, retroposon insertion analyses have so far yielded a clear signal [@pone.0060019-Kriegs1] with no conflicting data for this branch. In comparison, in other studies of deep mammalian divergences some splits were problematic to resolve from this data as bifurcating tree, because of conflicting signal [@pone.0060019-Hallstrm3], [@pone.0060019-Hallstrm4], [@pone.0060019-Churakov1], suggesting that incomplete lineage sorting and/or hybridization obscure short branches [@pone.0060019-Hallstrm3], [@pone.0060019-Churakov1]. It remains to be shown, if conflicting retroposon insertions are present for Euarchontoglires. However, the high evolutionary rate in the rodents will make it difficult to study neutral sequences like that of retroposon insertions in further detail, because sequence similarity in rodents is highly eroded over the 80 million years of their evolution [@pone.0060019-Hallstrm3]. Another challenging to resolve relationship has been the tarsier's grouping with anthropoids (platyrrhines and catarrhines) or Strepsirrhini (Lemuriformes and Lorisiforms). It has remained controversial, because contrasting phylogenetic signals from molecular data support different topologies [@pone.0060019-Jameson1]. This conflict is also visible in the network analysis ([Figure 4](#pone-0060019-g004){ref-type="fig"}). Our coalescent based analyses, however, confirm tarsier as sister taxon to the Simiiformes [@pone.0060019-Jameson1], which together with the Tarsiiformes form the Haplorhini clade. This relationship was identified by analyses of concatenated data with unanimous support. Conclusions {#s4a} ----------- The presently largest data set for a multi-locus analyses of mammalian relationships resolved the long challenge of placing Scandentia as the sister group to the primates, as has been previously suggested [@pone.0060019-Murphy1], [@pone.0060019-Song1], [@pone.0060019-Novacek1]. Multi-locus analyses settled the grouping of the tarsier with anthropoids. This leaves the dermopterans as the last order to be placed in the euarchontogliran tree. New mammalian genomes, and further development of methods will soon finalize the ordinal relationships among the Euarchontoglires. Network analyses are a valuable tool to depict and evaluate conflict in gene trees that can only be identified in genome-scale phylogenetic analyses. These conflicts from multiple gene trees can now be resolved into a reliable species tree by recent implementations of coalescence-based methods into phylogenetic analysis programs [@pone.0060019-Liu1], [@pone.0060019-Liu2]. The necessity of using methods developed for population genetics for deep divergences is a surprising development, because higher-level relationships have been expected to be deeper than the coalescent times of most genes. However, phylogenomic studies have shown that this is not always the case and speciation related processes interfere with phylogenetic analyses [@pone.0060019-Hallstrm2]--[@pone.0060019-Hallstrm4]. While the use of concatenated sequences generally improves the resolution of the phylogenetic tree, the current development in mammalian evolutionary studies [@pone.0060019-Song1] show that this approach may in some cases be uncertain and multi-locus species tree analyses are preferred to yield a reliable and sound species phylogeny even for divergences as deep as that among mammalian orders. Supporting Information {#s5} ====================== ###### **Length Distribution of the 1006 longest gene trees.** (TIF) ###### Click here for additional data file. ###### **Consensus Network of selected 661 genes.** Gene selected on basis of supporting the best topology of tree shrew position with a LogL value larger than standard deviation of \>0.7 compared to best tree. (TIF) ###### Click here for additional data file. ###### **List of 1,006 loci (.xls file, Ensembl transcript id) from all the species.** (DOC) ###### Click here for additional data file. ###### **Base composition for each species for all the three nucleotide positions (NT123).** (DOC) ###### Click here for additional data file. ###### **Base Composition for each species for all the first two nucleotide positions (NT12).** (DOC) ###### Click here for additional data file. We are grateful to Dr. Nilsson for comments on the retroposon insertion analyses. [^1]: **Competing Interests:**The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: AJ BH. Performed the experiments: VK. Analyzed the data: VK. Wrote the paper: VK BH AJ.

1. Introduction {#s0005} =============== Diacylglycerol (DG) serves to activate a variety of signaling proteins, including protein kinase C (PKC) [@bib1], [@bib2], [@bib3], [@bib4], [@bib5]. PKC is involved in receptor desensitization, in modulating membrane structure events, in regulating transcription, in mediating immune responses, in regulating cell growth, and in learning and memory among many other functions. PKC is a family of closely related serine/threonine kinases, and at least ten different isoforms have been discovered to date. The isoforms can be split into three families according to their requirement for different co-factors: the conventional or classical (c) PKCs: α, βI/βII and γ; novel (n) PKCs: δ, ε, η and θ; and the atypical (a) PKCs: ζ and λ. cPKCs can be activated by Ca^2+^ and by DG. nPKCs can also be activated by DG. However, aPKCs are unresponsive to Ca^2+^ or DG. DG kinase (DGK) phosphorylates DG to produce phosphatidic acid [@bib6], [@bib7], [@bib8], [@bib9]. To date, 10 mammalian DGK isozymes have been identified, and these isozymes are subdivided into five groups according to their structural features. DGK isozymes are also involved in regulating cell growth [@bib10], [@bib11], [@bib12] and angiogenesis [@bib13], mediating immune responses [@bib14], [@bib15], [@bib16], exacerbating the severity of type-2 diabetes [@bib17], [@bib18], [@bib19] and controlling neuronal network formation [@bib20], [@bib21], [@bib22], [@bib23]. Because DGK consumes an activator of PKC, DG, DGK regulates (attenuates) cPKC and nPKC activities. Indeed, many functional linkages between PKC isozymes and DGK isozymes have been reported [@bib6], [@bib7], [@bib8], [@bib9]. Mammalian cells contain at least 50 structurally distinct DG molecular species, which are supplied from a variety of lipid metabolic pathways, such as phosphatidylinositol turnover [@bib24], phosphatidylcholine (PC) hydrolysis by PC-specific phospholipase C [@bib25], [@bib26] and PC hydrolysis by phospholipase D followed by dephosphorylation by phosphatidic acid phosphatase [@bib27] in a cell stimulation-dependent manner. The DG species generated through phosphatidylinositol turnover mainly consists of 18:0/20:4-DG. The DG species derived from PC predominantly contain saturated and monounsaturated fatty acids. DGKε has preference for 18:0/20:4-DG *in vitro* [@bib28], [@bib29]. Our recent studies reported that several isozymes of DGK selectively metabolize different DG species within the cell. For example, DGKα was suggested to phosphorylate 18:0/20:4-DG and 18:0/22:6-DG in T-lymphocytes and COS-7 cells [@bib30], [@bib31]. DGKδ utilized PC-derived 14:0/16:0-, 14:0/16:1-, 16:0/16:0-, 16:0/16:1-, 16:0/18:0-, and 16:0/18:1-DG in C2C12 myoblasts [@bib19]. Based on these results, the possibility that each DGK isozymes metabolizes limited DG species and regulates the activity of restricted PKC isozymes has been suggested. However, a comprehensive analysis of the DG dependency of all cPKC and nPKC isozymes has not been performed. It was reported that the *in vitro* activation of different PKC isozymes varies in response to different DG species [@bib32]. However, in this report, only PKCα, βI, γ, δ and ε were analyzed, and only 8:0/8:0-, which is not a natural product, 18:0/20:4-, 18:0/20:5- and 18:0/22:6-DG (50--1000 mmol%) were used. Therefore, in this study, we more comprehensively re-evaluated the effects of 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- or 18:0/22:6-DG species in concentrations ranging from 2 to 2000 mmol% on all cPKC and nPKC isozymes side by side under the same conditions and aimed to obtain fundamental knowledge to explore the relationship between PKC isozyme activation and DG-related enzymes, including DGK isozymes. The obtained results indicate that, beyond our expectations, the modes of activation of cPKC and nPKC isozymes by DG molecular species varied considerably. 2. Materials and methods {#s0010} ======================== 2.1. Materials {#s0015} -------------- 1,2-dipalmitoyl-*sn*-glycerol (16:0/16:0-DG), 1-palmitoyl-2-oleoyl-*sn*-glycerol (16:0/18:1-DG), 1,2-dioleoyl-*sn*-glycerol (18:1/18:1-DG), 1-stearoyl-2-docosahexaenoyl-*sn*-glycerol (18:0/22:6-DG) and phosphatidylserine (PS) were purchased from Avanti Polar Lipids (Alabaster, AL).1-stearoyl-2-arachidonoyl-*sn*-glycerol (18:0/20:4-DG) and the PKC substrate, the \[pGlu^4^\]-Myelin basic protein fragment 4-14 (MBP) was obtained from Sigma-Aldrich (St. Louis, MO). PKC isoforms (α (product number: 01-133), βΙΙ (01-165), γ (01-137), δ (01-135), ε (01-136), η (01-138) and θ (01-140)) were obtained from Carna Biosciences (Kobe, Japan). These PKC isoforms were expressed as N-terminal Glutathione S-transferase-fusion proteins using baculovirus expression system and were highly purified by using glutathione Sepharose chromatography. P81 phosphocellulose squares were obtained from Merck Millipore (Darmstadt, Germany). 2.2. PKC activity assay {#s0020} ----------------------- PKC activity was assayed by measuring the incorporation of ^32^P from \[γ-^32^P\] ATP into \[pGlu^4^\]-Myelin basic protein fragment 4-14 (pGlu-Lys-Arg-Pro-Ser-Gln-Arg-Ser-Lys-Tyr-Leu). The reaction mixture (12.5 μl) contained 24 mM Tris-HCl (pH 7.4), 20 mM MgCl~2~, 0.2 mM CaCl~2~ (for α, βΙΙ, and γ) or 0.5 mM EDTA (for δ, ε, η and θ), 56 mg/ml MBP, 0.05 μg of PKC, 2 mol% PS, and 2--2000 mmol% DG. PS and DG were first mixed in chloroform/methanol and then dried under nitrogen. The residue was then sonicated in a buffer solution containing 255 mM Triton X-100 and 10 mM Tris--HCl to prepare lipid vesicles. The reaction was started by the addition of 0.1 mM \[γ-^32^P\] ATP (∼300 cpm/pmol), continued for 10 min at 30 °C and stopped by spotting 10 μl of samples on 0.5×0.5 cm squares of Whatman P81 phosphocellulose paper. The papers were dried, washed four times for 15 min each time with 1% H~3~PO~4~, and transferred to a scintillation counter to determine the radioactivity of \[^32^P\] MBP. 2.3. Cell culture and transfection {#s0025} ---------------------------------- COS-7 cells were maintained in Dulbecco\'s Modified Eagle\'s Medium (Wako Pure Chemical Industries, Osaka, Japan) containing 10% fetal bovine serum at 37 °C in an atmosphere containing 5% CO~2~. The cells were transfected with pEGFP-N3-PKCη by electroporation (1×10^6^ cells/2 mm gap cuvette, 110 V, 20.0 ms pulse length, one pulse) with the Gene Pulser Xcell™ Electroporation System (Bio-Rad Laboratories, Tokyo, Japan), according to the manufacturer\'s instructions. Forty-eight hours after transfection, the cells were harvested and suspended in 500 μl of ice-cold lysis buffer (50 mM HEPES (pH 7.2), 150 mM NaCl, 5 mM MgCl~2~, 1 mM dithiothreitol, and cOmplete™ EDTA-free protease inhibitor cocktail (Roche Diagnostics, Tokyo, Japan)) and then sonicated. The mixtures were centrifuged at 10,000×*g* for 5 min at 4 °C to yield cell lysates. EGFP-PKCη expression was confirmed by Western blotting using an anti-GFP antibody (sc-9996, Santa Cruz Biotechnology, Santa Cruz, CA). 2.4. Statistical analysis {#s0030} ------------------------- Statistical comparisons were performed using one-way ANOVA followed by Tukey\'s *post hoc* test. 3. Results {#s0035} ========== 3.1. Effects of different DG species on PKCα activation {#s0040} ------------------------------------------------------- We measured the activation of highly purified cPKC and nPKC isozymes in the presence of 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- or 18:0/22:6-DG species in concentrations ranging from 2 to 2000 mmol%. The reason in choosing the concentration range of DG in the study is that DG produced by cell stimulation is approximately 20 mmol% [@bib33]. We analyzed (1) the sensitivity of cPKC and nPKC isozymes to DG (DG concentration for PKC activation), (2) their dependence on DG (fold increase of PKC activity by DG), and (3) preference for DG species. We first examined effects of different DG species on the activity of a cPKC, PKCα. PKCα activity was increased approximately 2-fold in the presence of 2 mmol% 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4-DG ([Fig. 1](#f0005){ref-type="fig"}). However, 2 mmol% 18:0/22:6-DG did not significantly activate PKCα. PKCα activity was increased by DG in a dose dependent manner and was increased approximately 2.5-, 3.5- and 7-fold by 20, 200 and 2000 mmol% DG, respectively. However, PKCα did not exhibit preference for DG species at 20--2000 mmol%. Taken together, these results indicate that PKCα has (1) a high sensitivity to DG, (2) a strong dependence on DG and (3) less of a preference for 18:0/22:6-DG at low DG concentrations ([Table 1](#t0005){ref-type="table"}).Fig. 1Effects of different DG molecular species on PKCα activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}.The results are the means±SD of four independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05).Fig. 1Table 1Summary of the effects of different DG molecular species on the activity of cPKC and nPKC isozymes.Table 1\[DG\] (mmol%)2202002000PKCαFold increase**+++++++**PreferenceDP**\>**SD**−−−**

PKCβΙΙFold increase**+++++**Preference**−−−−**

PKCγFold increase**+++++**PreferenceDP**\<**SD**−−−**

PKCδFold increase**−+++++**Preference**−**DP**\<**SDDP**\<**SD**−**

PKCεFold increase**−++++**Preference**−−−**DP**\<**SD

PKCηFold increase**−−−−**Preference**−−−−**

*PKC*η *expressed in COS-7 cells*Fold increase**−+++**Preference**−**DP**\<**SD**−−**

PKCθFold increase**+++++++++++++**PreferenceDP**\<**SD**−−−**[^1][^2] 3.2. Effects of different DG species on PKCβII activation {#s0045} --------------------------------------------------------- Differential splicing gives rise to the two forms of PKCβ, βI and βII, which only differ in their extreme C-terminal ends [@bib34]. We next tested the effects of different DG species on the activity of a cPKC, PKCβII. PKCβII activity was increased approximately 1.5-fold by 2 mmol% DG, and its activity was increased approximately 1.5-, 2- and 4-fold by 20, 200 and 2000 mmol% DG, respectively ([Fig. 2](#f0010){ref-type="fig"}). However, PKCβII did not show significant preference for either 18:0/22:6-DG or 16:0/16:0-DG at 2--2000 mmol%. Taken together, these results indicate that PKCβII has (1) a high sensitivity to DG, (2) a moderate dependence on DG and (3) no significant preference for DG molecular species ([Table 1](#t0005){ref-type="table"}).Fig. 2Effects of different DG molecular species on PKCβΙΙ activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The results are the means ± SD of three independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005).Fig. 2 3.3. Effects of different DG species on PKCγ activation {#s0050} ------------------------------------------------------- PKCγ (cPKC) activity was increased approximately 1.5-fold in the presence of 2 mmol% 18:0/22:6-DG ([Fig. 3](#f0015){ref-type="fig"}). However, 2 mmol% 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, and 18:0/20:4-DG failed to significantly activate PKCγ. PKCγ activity was increased approximately 2-, 2.5- and 6-fold by 20, 200 and 2000 mmol% DG, respectively. However, PKCγ did not exhibit a significant preference for DG species at 20--2000 mmol%. These results indicate that PKCγ has (1) a high sensitivity to DG, (2) a moderate dependence on DG and, (3) in contrast to PKCα, a moderate preference for 18:0/22:6-DG at low DG concentrations ([Table 1](#t0005){ref-type="table"}).Fig. 3Effects of different DG molecular species on PKCγ activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The results are the means±SD of five independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05).Fig. 3 3.4. Effects of different DG species on activation of PKCδ {#s0055} ---------------------------------------------------------- We next tested the effects of different DG species on the activity of an nPKC, PKCδ. PKCδ was not significantly activated by 2 mmol% DG ([Fig. 4](#f0020){ref-type="fig"}). However, the activity of this isozyme was increased approximately 2.5-fold by 20 mmol% DG, and its activity was increased approximately 5- and 5.5-fold by 200 and 2000 mmol% DG, respectively. 16:0/18:1-, 18:0/20:4- and 18:0/22:6-DG strongly activated PKCδ compared to 16:0/16:0-DG at 200 mmol%. At 20 mmol%, the activity of this isozyme was more strongly enhanced by 18:0/22:6-DG. However, PKCδ did not exhibit a significant difference in the degree of activation by DG species at 2000 mmol%. These results indicate that PKCδ has (1) a moderate sensitivity to DG, (2) a moderate dependence on DG, (3) a preference for 18:0/22:6-DG, and less of a preference for 16:0/16:0-DG at medium concentrations ([Table 1](#t0005){ref-type="table"}).Fig. 4Effects of different DG molecular species on PKCδ activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The results are the means±SD of four independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05, ^\#\#^*P*\<0.01).Fig. 4 3.5. Effects of different DG species on PKCε activation {#s0060} ------------------------------------------------------- PKCε (nPKC) was not significantly activated by 2 mmol% DG ([Fig. 5](#f0025){ref-type="fig"}). However, the activity of this isoform was increased approximately 1.5-fold by 20 mmol% DG and was increased approximately 2- and 4-fold by 200 and 2000 mmol% DG, respectively. Although PKCε did not exhibit a significant preference for DG species at 20 and 200 mmol%, this isozyme showed a preference for 18:0/22:6-DG at 2000 mmol%. These results indicate that PKCε has (1) a moderate sensitivity to DG, (2) a moderate dependence on DG and (3) a preference for 18:0/22:6-DG at high concentrations ([Table 1](#t0005){ref-type="table"}).Fig. 5Effects of different DG molecular species on PKCε activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The results are the means±SD of four independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05).Fig. 5 3.6. Effects of different DG species on PKCη activation {#s0065} ------------------------------------------------------- PKCη (nPKC) was not markedly activated by any DG molecular species at any concentrations ([Fig. 6](#f0030){ref-type="fig"}). 12-O-Tetradecanoylphorbol-13-acetate (TPA), which mimics DG, is known to strongly activate cPKC and nPKC isozymes [@bib35]. Although the effect of TPA on the activity of PKCη was examined, TPA also did not significantly activate this enzyme ([Suppl. Fig. 1](#s0090){ref-type="sec"}).Fig. 6Effects of different DG molecular species on PKCη activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The results are the means±SD of four independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity.Fig. 6 Given that PKCη expressed in insect cells is not responsive to DG, we next determined the effects of DG species on this isozyme expressed in mammalian cells (COS-7 cells) instead of insect cells. PKCη activation by DG species (approximately 1.5-fold increase) was detected at 20 mmol% ([Fig. 7](#f0035){ref-type="fig"}). This isozyme showed a weak preference for 18:0/22:6-DG at 20 mmol%. The activity of this isoform was increased approximately 1.5- and 2.5-fold by 200 and 2000 mmol% DG, respectively. However, PKCη did not show a significant preference for DG molecular species at 200 and 2000 mmol%.Fig. 7Effects of different DG molecular species on the activation of PKCη expressed in COS-7 cells. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs, as described in [Section 2](#s0010){ref-type="sec"}. The activities of vector-transfected cells were subtracted. The results are the means±SD of three independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG). The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05, ^\#\#^*P*\<0.01).Fig. 7 These results indicate that PKCη expressed in insect cells has (1) no sensitivity to DG, (2) no dependence on DG and (3) no preference to DG species, even at high concentrations ([Table 1](#t0005){ref-type="table"}). On the other hand, PKCη expressed in mammalian cells has (1) a moderate sensitivity to DG, (2) a weak dependence on DG and (3) a weak preference for 18:0/22:6-DG at medium concentrations ([Table 1](#t0005){ref-type="table"}). 3.7. Effects of different DG species on activation of PKCθ {#s0070} ---------------------------------------------------------- Finally, we tested the effects of different DG species on the activity of an nPKC, PKCθ. PKCθ activity was increased approximately 4-fold by 18:0/22:6-DG at 2 mmol% DG, and its activity was increased approximately 8-, 21- and 25-fold by all DG species at 20, 200 and 2000 mmol% DG, respectively ([Fig. 8](#f0040){ref-type="fig"}). PKCθ was more strongly activated by 16:0/18:1-DG, 18:0/20:4- and 18:0/22:6-DG than by 16:0/16:0-DG and by 18:0/22:6-DG than by 16:0/18:1-, 18:1/18:1 and 18:0/20:4-DG at 2 and 20 mmol% DG, respectively. However, PKCθ did not exhibit a significant preference for DG species at 200 and 2000 mmol%. These results indicate that PKCθ has (1) a high sensitivity to DG, (2) a very strong dependence on DG and (3) a preference for 18:0/22:6-DG at relatively low DG concentrations ([Table 1](#t0005){ref-type="table"}).Fig. 8Effects of different DG molecular species on PKCθ activation. Lipid vesicles were prepared with different DG molecular species (16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- and 18:0/22:6-DG) and PKC activity toward MBP was determined in vesicles as a function of increasing concentrations of DGs as described in [Section 2](#s0010){ref-type="sec"}. The results are the means ± SD of three independent experiments. The left axis shows the relative activity compared to the control (0 mmol% DG) and the right axis shows the specific activity. The data are significantly different from the control, 0 mmol% DG (\**P*\<0.05, \*\**P*\<0.01, \*\*\**P*\<0.005), and among the DG molecular species (^\#^*P*\<0.05, ^\#\#^*P*\<0.01).Fig. 8 Overall, these results indicate that cPKC and nPKC isozymes have different dependencies/sensitivities to DG and distinct preferences for shorter and saturated fatty acid-containing DG species (16:0/16:0-DG) and longer and polyunsaturated fatty acid-containing DG species (18:0/22:6-DG), respectively. 4. Discussion {#s0075} ============= A variety of DG molecular species are produced in stimulated cells. cPKC (α, βII and γ) and nPKC (δ, ε, η and θ) isozymes are known to be activated by DG [@bib1], [@bib2], [@bib3], [@bib4], [@bib5]. However, a comprehensive analysis of the DG dependency of these PKCs has not been performed. Therefore, we performed comprehensive comparison of the effects of DG molecular species (2--2000 mmol% 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- or 18:0/22:6-DG species) on the activities of these PKCs side by side under the same conditions. We found that beyond our expectation, the modes of activation of cPKC and nPKC isozymes by DG molecular species varied considerably. We analyzed (1) the sensitivity to DG (DG concentration for PKC activation), (2) dependence on DG (fold increase in PKC activity in the presence of DG), and (3) preference for DG molecular species of highly purified cPKC and nPKC isozymes expressed in the baculovirus-insect cell expression system. (1) Regarding sensitivity to DG, the highly sensitive isozymes are PKCα, PKCβII, PKCγ and PKCθ ([Table 1](#t0005){ref-type="table"}). (2) PKCθ was the most strongly activated and PKCα was strongly activated ([Table 1](#t0005){ref-type="table"}). The activities of cPKC and nPKC isozymes in the presence of DG species were comparable to those in the presence of TPA ([Suppl. Fig. 1](#s0090){ref-type="sec"}). Therefore, the activities at 2000 mmol% DG are roughly maximal values. (3) PKCα showed less preference to long/unsaturated fatty acid-containing DG species (18:0/22:6-DG) ([Table 1](#t0005){ref-type="table"}). PKCγ, On the other hand, PKCδ, PKCε and PKCθ showed a preference for long/unsaturated fatty acid-containing DG species (18:0/22:6-DG) ([Table 1](#t0005){ref-type="table"}). Moreover, these preferences appear at different DG concentrations. The activation capacity of saturated and polyunsaturated fatty acid-containing DG molecular species is suggested to rely on the fluidity state of the micelles and membranes [@bib36]. Taken together, these results indicate that c/nPKCs have different sensitivities and dependences on DG and a distinct preference for shorter and saturated fatty acid-containing and longer and polyunsaturated fatty acid-containing DG species, respectively ([Table 1](#t0005){ref-type="table"}). PKCη was not activated by 2--2000 mmol% DG in our assay conditions ([Fig. 6](#f0030){ref-type="fig"}). In addition, this isozyme was not activated by TPA (data not shown). Because PKCη showed relatively high specific activity in the absence of DG compared with the other isozymes ([Fig. 6](#f0030){ref-type="fig"}), this isoform may be already activated in the absence of DG/TPA. Unlike the insect cell-expressed enzyme, the mammalian cell-expressed PKCη was activated by DG ([Fig. 7](#f0035){ref-type="fig"}). Essentially the same results (activated at 2 mmol% DG and approximately 6-fold activated at 2000 mmol% DG) were obtained using PKCα expressed in mammalian cells (data not shown) compared to the insect cell-expressed enzyme ([Fig. 1](#f0005){ref-type="fig"}). The results suggest that in the case of PKCη, some modification in mammalian COS-7 cells, but not in insect cells, is needed for the enzyme to exhibit DG-dependent activation. It has been claimed that DGs with saturated chains were less effective on triggering PKC activation than those bearing polyunsaturated chains [@bib36], [@bib37], [@bib38], [@bib39], [@bib40], [@bib41]. Moreover, Madani et al. [@bib32] reported that the *in vitro* activation of different PKC isozymes varies in response to different DG species. However, in this report, only PKCα, βI, γ, δ and ε were analyzed. Moreover, only 8:0/8:0-, which is not a natural product, 18:0/20:4-, 18:0/20:5- and 18:0/22:6-DG were used. Furthermore, approximately 50--1000 mmol% of DG species were tested. In the present study, we more comprehensively analyzed the effects of a wide range of physiological DG molecular species and concentrations (2--2000 mmol% 16:0/16:0-, 16:0/18:1-, 18:1/18:1-, 18:0/20:4- or 18:0/22:6-DG species) on the activities of cPKC (α, βII and γ) and nPKC (δ, ε, η and θ) isozymes ([Table 1](#t0005){ref-type="table"}) side by side under the same conditions, and identified several additional intriguing differences, as described above. For example, we showed, for the first time, that PKCα showed less of a preference for polyunsaturated fatty acid-containing DG species (18:0/22:6-DG) at 2 mmol% and that PKCθ has a higher sensitivity and stronger dependence on DG than the other isozymes. The results obtained in the present study allow us to speculate a correlation between the PKC isozymes and DGK isozymes. DGKε has preference for polyunsaturated fatty acid-containing DG such as 18:0/20:4-DG [@bib28], [@bib29], suggesting possible linkages with PKCγ, δ, ε and θ, which have preferences for polyunsaturated fatty acid-containing DG species. DGKα was shown to phosphorylate 18:0/20:4-DG and 18:0/22:6-DG in cells [@bib30], [@bib31]. Therefore, DGKα may be associated with PKCγ, δ, ε and θ. DGKα is predominantly expressed in T-lymphocytes [@bib42] and regulates T-cell activation [@bib15], [@bib16]. Of PKCγ, δ, ε and θ, PKCθ is predominantly expressed in T-lymphocytes and is a key molecule in T-cell activation and differentiation [@bib43], implying an intimate linkage between DGKα and PKCθ. However, further studies are required to definitely indicate the relation between DGKs and PKCs. DGKδ contributes to hyperglycemia-induced peripheral insulin resistance [@bib17] and regulates the activities of PKCα [@bib18] and PKCδ [@bib17] in myotubes. Palmitic acid (16:0)-containing DG species are known to accumulate in skeletal muscle of type 2 diabetes patients [@bib44]. We showed that DGKδ indeed utilized PC-derived, palmitic acid-containing DG species (14:0/16:0-, 16:0/16:0-, 16:0/16:1-, 16:0/18:0-, and 16:0/18:1-DG) [@bib19]. The results are not inconsistent with the DG species preferences of PKCα and PKCδ ([Fig. 1](#f0005){ref-type="fig"}, [Fig. 4](#f0020){ref-type="fig"} and [Table 1](#t0005){ref-type="table"}). In the present study, we comprehensively analyzed the effects of different DG species on cPKC and nPKC isozymes side by side under the same conditions and revealed that these PKCs have different dependencies/sensitivities to DG and distinct preferences for shorter and saturated fatty acid- and longer and polyunsaturated fatty acid-containing DG molecular species, respectively. This differential regulation would be important for their physiological functions. The comprehensive *in vitro* results obtained in the present study could be a useful guide for interpreting the activation and functional significance of cPKC and nPKC isozymes *in vivo*. Funding {#s0080} ======= This work was supported by MEXT/JSPS KAKENHI Grant numbers 22370047 (Grant-in-Aid for Scientific Research (B)), 23116505 (Grant-in-Aid for Scientific Research on Innovative Areas), 25116704 (Grant-in-Aid for Scientific Research on Innovative Areas), 26291017 (Grant-in-Aid for Scientific Research (B)) and 15K 14470 (Grant-in-Aid for Challenging Exploratory Research), the Japan Science and Technology Agency (AS221Z00794F, AS231Z00139G, AS251Z01788Q and AS2621643Q), the Naito Foundation, the Hamaguchi Foundation for the Advancement of Biochemistry, the Daiichi-Sankyo Foundation of Life Science, the Terumo Life Science Foundation, the Futaba Electronic Memorial Foundation, the Daiwa Securities Health Foundation, the Ono Medical Research Foundation, the Japan Foundation for Applied Enzymology, the Food Science Institute Foundation, the Skylark Food Science Institute and the Venture Business Laboratory of Chiba University (FS). Appendix A. Supplementary material {#s0090} ================================== Supplementary material. Supplementary material. Transparency document associated with this article can be found in the online version at [10.1016/j.bbrep.2016.07.017](http://dx.doi.org/10.1016/j.bbrep.2016.07.017){#ir0005}. [^1]: Fold increase: --, less than 1.5-fold; +, more than 1.5-fold; ++, more than 3-fold; +++, more than 6-fold; ++++, more than 12-fold. [^2]: Preference: different preference for 16:0/16:0-DG (1,2-dipalmitoyl-*sn*-glycerol (DP)) and 18:0/22:6-DG (1-stearyl-2-docosahexanoyl-*sn*-glycerol (SD)).

**Suggested citation:** EFSA FEEDAP Panel (EFSA Panel on Additives and Products or Substances used in Animal Feed) , Bampidis V, Azimonti G, Bastos ML, Christensen H, Dusemund B, Kouba M, Kos Durjava M, López‐Alonso M, López Puente S, Marcon F, Mayo B, Pechová A, Petkova M, Ramos F, Sanz Y, Villa RE, Woutersen R and Anguita M, 2019 Scientific Opinion on the safety and efficacy of 3‐phytase FLF1000 as a feed additive for pigs for fattening and minor porcine species for growing. EFSA Journal 2019;17(8):5791, 8 pp. 10.2903/j.efsa.2019.5791 **Requestor:** European Commission **Question number:** EFSA‐Q‐2017‐00447 **Panel members:** Giovanna Azimonti, Vasileios Bampidis Maria de Lourdes Bastos, Henrik Christensen, Birgit Dusemund, Maryline Kouba, Mojca Kos Durjava, Marta López‐Alonso, Secundino López Puente, Francesca Marcon, Baltasar Mayo, Alena Pechová, Mariana Petkova, Fernando Ramos, Yolanda Sanz, Roberto Edoardo Villa and Ruud Woutersen. **Acknowledgments:** The Panel wishes to acknowledge the contribution of Jaume Galobart, Gloria López‐Gálvez and Jordi Tarrés‐Call to this opinion. Adopted: 3 July 2019 1. Introduction {#efs25791-sec-0002} =============== 1.1. Background and Terms of Reference {#efs25791-sec-0003} -------------------------------------- Regulation (EC) No 1831/2003[1](#efs25791-note-1005){ref-type="fn"} establishes the rules governing the Community authorisation of additives for use in animal nutrition. In particular, Article 4(1) of that Regulation lays down that any person seeking authorisation for a feed additive or for a new use of a feed additive shall submit an application in accordance with Article 7. The European Commission received a request from Fertinagro Nutrientes S.L.[2](#efs25791-note-1006){ref-type="fn"} for authorisation of the product 3‐phytase FLF1000 (3‐phytase), when used as a feed additive for pigs for fattening and minor porcine species for growing (category: zootechnical additive; functional groups: digestibility enhancers and substances which favourably affect the environment). According to Article 7(1) of Regulation (EC) No 1831/2003, the Commission forwarded the application to the European Food Safety Authority (EFSA) as an application under Article 4(1) (authorisation of a feed additive or new use of a feed additive). EFSA received directly from the applicant the technical dossier in support of this application. The particulars and documents in support of the application were considered valid by EFSA as of 7 November 2017. According to Article 8 of Regulation (EC) No 1831/2003, EFSA, after verifying the particulars and documents submitted by the applicant, shall undertake an assessment in order to determine whether the feed additive complies with the conditions laid down in Article 5. EFSA shall deliver an opinion on the safety for the target animals, consumer, user and the environment and on the efficacy of the product 3‐phytase FLF1000 (3‐phytase), when used under the proposed conditions of use. 1.2. Additional information {#efs25791-sec-0004} --------------------------- The additive 3‐phytase FLF1000 is a liquid product that contains 3‐phytase (Enzyme Commission number 3.2.1.8) produced by a genetically modified strain of *Komagataella phaffii* (CECT 13094)[3](#efs25791-note-1007){ref-type="fn"} and is authorised as a feed additive for chickens for fattening and laying hens,[4](#efs25791-note-1008){ref-type="fn"} for chickens reared for laying and minor poultry species for fattening or reared for laying or for breeding.[5](#efs25791-note-1009){ref-type="fn"} The product is also authorised in the solid form for the same species/categories.[6](#efs25791-note-1010){ref-type="fn"} The FEEDAP Panel adopted an opinion on the safety and efficacy of the additive 3‐phytase FLF1000 as a feed additive for chickens for fattening and laying hens (EFSA FEEDAP Panel, [2016](#efs25791-bib-0800){ref-type="ref"}), another one on the extension of use to chickens reared for laying and minor poultry species (EFSA FEEDAP Panel, [2018](#efs25791-bib-0003){ref-type="ref"}) and a last one on the safety and efficacy of the solid formulation of the additive (3‐phytase FSF10000) for the species assessed previously (EFSA FEEDAP Panel, [2019](#efs25791-bib-0004){ref-type="ref"}). The applicant is now requesting for an extension of the use of the additive to pigs for fattening and minor porcine species for growing. 2. Data and methodologies {#efs25791-sec-0005} ========================= 2.1. Data {#efs25791-sec-0006} --------- The present assessment is based on data submitted by the applicant in the form of a technical dossier[7](#efs25791-note-1011){ref-type="fn"} in support of the authorisation request for the use of 3‐phytase FLF1000 as a feed additive. The FEEDAP Panel used the data provided by the applicant together with data from other sources, such as previous risk assessments by EFSA. The European Union Reference Laboratory (EURL) considered that the conclusions and recommendations reached in the previous assessment regarding the methods used for the control of the active substance in animal feed are valid and applicable for the current application.[8](#efs25791-note-1012){ref-type="fn"} 2.2. Methodologies {#efs25791-sec-0007} ------------------ The approach followed by the FEEDAP Panel to assess the safety and the efficacy of 3‐phytase FLF1000 is in line with the principles laid down in Regulation (EC) No 429/2008[9](#efs25791-note-1013){ref-type="fn"} and the relevant guidance documents: Guidance on zootechnical additives (EFSA FEEDAP Panel, [2012](#efs25791-bib-0002){ref-type="ref"}) and Technical guidance: Tolerance and efficacy studies in target animals (EFSA FEEDAP Panel, [2011](#efs25791-bib-0001){ref-type="ref"}). 3. Assessment {#efs25791-sec-0008} ============= This assessment deals with a request from the applicant to extend the use of the additive 3‐phytase FLF1000 to pigs for fattening and minor porcine species for growing as a zootechnical additive (functional groups: digestibility enhancers and substances which favourably affect the environment). 3.1. Characterisation {#efs25791-sec-0009} --------------------- The additive 3‐phytase FLF1000 in its liquid formulation contains 3‐phytase (Enzyme Commission number 3.1.3.8) produced by a genetically modified strain of *K. phaffii* (CECT 13094) with a minimum phytase activity of 1,000 FTU/mL. In a previous opinion, the Panel described the additive and its manufacturing process including the production strain and its genetic modification (EFSA FEEDAP Panel, [2018](#efs25791-bib-0003){ref-type="ref"}). The additive is intended to be used in feed for pigs for fattening and minor porcine species for growing at 1,000 FTU/kg feed. 3.2. Safety {#efs25791-sec-0010} ----------- Safety aspects regarding the use of this additive in feed including the safety of the production strain, the safety for the consumer, for the users and for the environment have been previously evaluated (EFSA FEEDAP Panel, [2018](#efs25791-bib-0003){ref-type="ref"}). The FEEDAP Panel concluded that there are no concerns regarding the genetic modification of the production strain, that there are no concerns for the consumer safety and no risks for the environment are expected from the use of the product as a feed additive. Regarding the safety for the user, it was concluded that the additive is not irritant to eyes and skin and is not a dermal sensitiser; however, it should be considered a potential respiratory sensitiser. The Panel is not aware of any new information that would lead it to reconsider the conclusions drawn previously. Moreover, the FEEDAP Panel considers that the new use requested by the applicant would not modify the above conclusions. The applicant provided a combined tolerance and efficacy trial in weaned piglets to support the safety for the target species.[10](#efs25791-note-1014){ref-type="fn"} The trial included a balance trial. The piglets were under study for 48 days. From days 1 to 7 and from days 18 to 48, the piglets were kept in pens with 5 animals. From days 7 to 18, the balance study took place and four animals from each pen were placed in metabolic cages while one pig remained in the original pen and was not included in the balance. The feed intake was measured throughout the study, but the data regarding the overall feed intake was not made available by the applicant. Owing to the lack of precise data on the total feed intake of the animals, the FEEDAP Panel cannot consider further the study for the assessment and consequently no conclusion can be drawn regarding the safety of the additive for pigs for fattening nor for other minor growing porcine species. 3.3. Efficacy for pigs for fattening {#efs25791-sec-0011} ------------------------------------ A total of five efficacy trials were submitted by the applicant. Two of the trials submitted were not considered: One was the tolerance trial which was conducted with weaned piglets[11](#efs25791-note-1015){ref-type="fn"} and, therefore, cannot be considered in the assessment of the efficacy for pigs for fattening; the second one was a long‐term trial which was not considered further in the assessment due to the extensive therapeutic treatment the animals received during the study (at least three periods of five‐day medication).[12](#efs25791-note-1016){ref-type="fn"} The other three short‐term studies were considered for the assessment and are described below. The first was a balance trial in which a total of 75 castrated male pigs (PIC line 65 × PIC Camborough 22, body weight 34.3 kg) were allocated to five dietary treatments.[13](#efs25791-note-1017){ref-type="fn"} Pigs were distributed to pens in groups of three pigs each during the first 9 days of the experiment and then two pigs per pen (10 pigs per treatment) were selected and housed individually in metabolic cages for a period of 9 days to conduct the balance trial. A basal diet based on wheat, barley and soya bean meal (total phosphorus content 0.35% and calcium content 0.78%) was either not supplemented (control) or supplemented with 3‐phytase to provide 250, 500 or 1,000 FTU/kg feed. Enzyme activities were confirmed by analysis. A positive control was also considered (total phosphorus content 0.48% and calcium content 0.95%). Feed was offered on *ad libitum* basis in mash form. Health status and mortality were monitored daily. Feed intake was measured and feed to gain ratio was calculated. The balance study consisted of an adaption period of 5 days and a collection period of 4 days for total collection of faeces and urine, separated. The second was a balance trial in which a total of 60 male pigs (Piétrain × (Large White × Landrace), initial body weight 56.5 kg) were distributed to pens in groups of six pigs each and allocated to five dietary treatments.[14](#efs25791-note-1018){ref-type="fn"} Pigs were distributed to pens in groups of two pigs each during the first 7 days of the study, then after two pigs per pen were selected and individually housed in metabolic cages (12 pigs per treatment). A basal diet based on wheat, barley and soya bean meal (total phosphorus content 0.49--0.57% and calcium content 0.48--0.60%) was either not supplemented (control) or supplemented with 3‐phytase to provide 250, 500 or 1,000 FTU/kg feed. Enzyme activities were confirmed by analysis. A positive control was also considered (total P content 0.65% and Ca content 0.66%). All experimental diets contained titanium dioxide as an external marker. Feeds were offered on *ad libitum* basis in mash form. Health status and mortality were monitored daily. The balance study consisted of an adaption period of 3 days and a collection period of 3 days for total collection of faeces and urine, separated. The third trial was a digestibility trial in which measurements of the bone ash content were done.[15](#efs25791-note-1019){ref-type="fn"} A total of 18 male pigs (Duroc × (Landrace × Large White)), initial body weight 35 kg) were allocated to two dietary groups. The pigs were selected according to body weight and distributed to pens in which stayed for 18 days. Then, they were housed individually in metabolic cages for 10 days (9 pigs per treatment). A basal diet based on maize and soya bean meal (total phosphorus content 0.36% and calcium content 0.49%) was either not supplemented (control) or supplemented with 3‐phytase to provide 500 FTU/kg feed. Enzyme activities were confirmed by analysis. All experimental diets contained titanium dioxide as external marker. Feed was offered on *ad libitum* basis during the adaptation period and restricted to 2.5 times the energy requirements for maintenance according to the body weight measured during the digestibility trial. The digestibility trial consisted of a 6‐day adaptation period and 4 days of total collection of faeces. In all trials, health status and mortality were monitored daily. Body weight and feed intake were measured throughout the study period. Diets and faeces were analysed to study the digestibility of phosphorus, in trials 1 and 2 urine was also analysed to study the retention of phosphorus. In the third trial all animals were killed at the end of the study and the III and IV metacarpal bones from the left forelimb were collected from all animals at the end of the study and bone ash and phosphorus content were measured. An analysis of variance was done with the data from each study, in trial 1 group means were compared against the control using Dunnett\'s test and in trial 2 group means were compared using Tukey\'s test. The results of the phosphorus utilisation and bone ash content are presented in Table [1](#efs25791-tbl-0001){ref-type="table"}. In trials 1 and 2, pigs receiving the additive from 500 FTU/kg or 250 FTU/kg, respectively, showed a higher retention of phosphorus. In trial 3, piglets receiving 500 FTU/kg feed showed a higher digestibility of phosphorus and a higher bone content of ash and phosphorus compared to control. Therefore, the FEEDAP Panel concludes that the additive has the potential to be efficacious in improving the phosphorus utilisation in pigs for fattening at a minimum level of 500 FTU/kg feed. ###### Effect of 3‐phytase FLF1000 on the apparent phosphorus digestibility/retention and bone mineralisation Trial Group FTU/kg feed Phosphorus (%) Bone content (%) ------- ------------------- ----------------------------------------------- ------------------------------------------------ ----------------------------------------------- ----------------------------------------------- ----------------------------------------------- 1 Control 41.8 41.5 250 45.6 45.2 500 48.1[\*](#efs25791-note-0006){ref-type="fn"} 47.8[\*](#efs25791-note-0006){ref-type="fn"} 1,000 53.1[\*](#efs25791-note-0006){ref-type="fn"} 52.8[\*](#efs25791-note-0006){ref-type="fn"} Positive control 54.2[\*](#efs25791-note-0006){ref-type="fn"} 53.9[\*](#efs25791-note-0006){ref-type="fn"} 2 Control 53.7[^b^](#efs25791-note-0007){ref-type="fn"} 46.4[^b^](#efs25791-note-0007){ref-type="fn"} 52.4[^b^](#efs25791-note-0007){ref-type="fn"} 250 63.2[^a^](#efs25791-note-0007){ref-type="fn"} 51.3[^ab^](#efs25791-note-0007){ref-type="fn"} 61.8[^a^](#efs25791-note-0007){ref-type="fn"} 500 61.3[^a^](#efs25791-note-0007){ref-type="fn"} 56.7[^a^](#efs25791-note-0007){ref-type="fn"} 60.2[^a^](#efs25791-note-0007){ref-type="fn"} 1,000 62.2[^a^](#efs25791-note-0007){ref-type="fn"} 52.9[^ab^](#efs25791-note-0007){ref-type="fn"} 60.0[^a^](#efs25791-note-0007){ref-type="fn"} Positive control 59.9[^a^](#efs25791-note-0007){ref-type="fn"} 56.7[^a^](#efs25791-note-0007){ref-type="fn"} 53.7[^b^](#efs25791-note-0007){ref-type="fn"} 3 Control 45.4[^b^](#efs25791-note-0007){ref-type="fn"} 26.5[^b^](#efs25791-note-0007){ref-type="fn"} -- 37.0[^b^](#efs25791-note-0007){ref-type="fn"} 6.27[^b^](#efs25791-note-0007){ref-type="fn"} 500 55.1[^a^](#efs25791-note-0007){ref-type="fn"} 38.2[^a^](#efs25791-note-0007){ref-type="fn"} -- 39.0[^a^](#efs25791-note-0007){ref-type="fn"} 6.61[^a^](#efs25791-note-0007){ref-type="fn"} \* In trial 1, values within one column are statistically significant compared to control (p \< 0.05). ^a,b^ Values within one trial and within one column with different superscript are statistically different. John Wiley & Sons, Ltd 3.4. Post‐market monitoring {#efs25791-sec-0012} --------------------------- The FEEDAP Panel considers that there is no need for specific requirements for a post‐market monitoring plan other than those established in the Feed Hygiene Regulation[16](#efs25791-note-1020){ref-type="fn"} and Good Manufacturing Practice. 4. Conclusions {#efs25791-sec-0013} ============== The FEEDAP Panel cannot conclude on the safety of the additive for pigs for fattening nor for minor growing porcine species. The FEEDAP Panel concludes that the use of the product as a feed additive raises no concerns for the consumer safety nor for the environment. The additive should be regarded as a potential respiratory sensitiser. The FEEDAP Panel concludes that the additive has the potential to be efficacious in pigs for fattening at 500 FTU/kg feed. 5. Documentation as provided to EFSA/Chronology {#efs25791-sec-0014} =============================================== DateEvent20/04/2017Dossier received by EFSA. 3‐phytase for pigs for fattening. Submitted by Fertinagro Biotech, S.L.29/05/2019Reception mandate from the European Commission07/11/2017Application validated by EFSA -- Start of the scientific assessment11/12/2017Request of supplementary information to the applicant in line with Article 8(1)(2) of Regulation (EC) No 1831/2003 -- Scientific assessment suspended. *Issues: safety for the consumer*07/02/2018Comments received from Member States26/04/2019Reception of supplementary information from the applicant ‐ Scientific assessment re‐started03/07/2019Opinion adopted by the FEEDAP Panel. End of the Scientific assessment Abbreviations {#efs25791-sec-0015} ============= EURLEuropean Union Reference LaboratoryFEEDAP PanelEFSA Panel on Additives and Products or Substances used in Animal Feed Regulation (EC) No 1831/2003 of the European Parliament and of the Council of 22 September 2003 on additives for use in animal nutrition. OJ L 268, 18.10.2003, p. 29. Fertinagro Nutrientes S.L., Pol. Ind. La Paz parcela 185, 44195 Teruel, Spain. The name of the applicant was modified to Fertinagro Biotech, S.L. Formerly classified as *Komagataella pastoris*. The accession number for the strain presented in the mandate referred to the recipient strain and not to the production strain which is (CECT 13094). Commission implementing Regulation (EU) 2017/895 of 24 May 2017 concerning the authorisation of a preparation of 3‐phytase produced by *Komagataella pastoris* (CECT 13094) as a feed additive for chickens for fattening and laying hens (holder of authorisation Fertinagro 0014 SL). OJ L 138, 25.5.2017, p.120. Commission implementing Regulation (EU) 2019/144 of 28 January 2019 concerning the authorisation of a preparation of 3‐phytase produced by *Komagataella pastoris* (CECT 13094) as a feed additive for chickens reared for laying and minor poultry species for fattening or reared for laying or for breeding (holder of authorisation Fertinagro Biotech S.L.). OJ L 27, 31.1.2019, p.8. Commission implementing Regulation (EU) 2019/781 of 15 May 2019 concerning the authorisation of a preparation of 3‐phytase produced by *Komagataella phaffii* (CECT 13094) as a feed additive for chickens for fattening or reared for laying, laying hens and minor poultry species for fattening, for breeding and reared for laying (holder of authorisation Fertinagro Nutrientes S.L.). OJ L 127, 16.5.2019, p.1. FEED dossier reference: FAD‐2017‐0023. The full report is available on the EURL website: <https://ec.europa.eu/jrc/sites/jrcsh/files/finirep-fad-2015-0026-preparation-3phytase.pdf> Commission Regulation (EC) No 429/2008 of 25 April 2008 on detailed rules for the implementation of Regulation (EC) No 1831/2003 of the European Parliament and of the Council as regards the preparation and the presentation of applications and the assessment and the authorisation of feed additives. OJ L 133, 22.5.2008, p. 1. Technical dossier/Section III/Annex III.1. Technical dossier/Section IV/Annex IV.1. Technical dossier/Section IV/Annex IV 4. Technical dossier/Section IV/Annex IV 2. Technical dossier/Section IV/Annex IV 3. Technical dossier/Supplementary information April 2019/Annex 2. Regulation (EC) No 183/2005 of the European Parliament and of the Council of 12 January 2005 laying down requirements for feed hygiene. OJ L 35, 8.2.2005, p. 1.

Introduction {#s1} ============ The TGF-β signal transduction pathway follows an apparently straightforward downstream cascade, progressing sequentially from the interaction of ligands with transmembrane receptors, through phosphorylation of mediator Smad proteins, to transcriptional responses ([Figure 1](#pcbi-0020003-g001){ref-type="fig"}). The simple logic of this signal transduction cascade strongly contrasts with the molecular complexity of the cellular processes involved and the wide diversity of responses triggered. ![Formation of Receptor Hetero-Tetramers\ The active form of the TGF-β ligand is a dimer of two molecules held together by hydrophobic interactions and a disulfide bond \[[@pcbi-0020003-b030],[@pcbi-0020003-b031]\]. This dimer induces the formation, at the plasma membrane, of receptor hetero-tetramers that contain two type I and two type II receptors \[[@pcbi-0020003-b002],[@pcbi-0020003-b003]\]. The type II receptors phosphorylate the type I receptors; the type I receptors are then enabled to phosphorylate cytoplasmic R-Smads, which then act as transcriptional regulators.](pcbi.0020003.g001){#pcbi-0020003-g001} At the molecular level, there is an intricate signal transduction machinery that integrates signals from the 42 known ligands of the TGF-β superfamily, funnels them through the two principal regulatory Smad (R-Smad) channels (Smad1/5/8 or Smad2/3), and subsequently leads to the widespread transcriptional control of more than 300 target genes in a cell-context dependent manner \[[@pcbi-0020003-b001]\] (see [Figure 2](#pcbi-0020003-g002){ref-type="fig"}). The components of this machinery include the members of the two main receptor families (type I and type II receptors), a myriad of adaptor proteins, and the trafficking apparatus of the cell, which shuttles proteins between different subcellular compartments. Each ligand induces the formation of a receptor complex with type I and type II receptors, which then signal through one of the two Smad channels \[[@pcbi-0020003-b002],[@pcbi-0020003-b003]\]. The ability of most ligands to bind several type I and type II receptors results in a complex ligand-receptor interaction network ([Figure 2](#pcbi-0020003-g002){ref-type="fig"}). ![Interactions among the Ligands of the TGF-β Superfamily and Their Receptors\ The graphical representation lays out the specific type II/type I receptor complexes that different ligands mediate (based on data reviewed in reference \[[@pcbi-0020003-b032]\]). Each set of links drawn between a type II and type I receptor, mediated by a connecting ligand, represents a feasible ligand-receptor complex. The 14 ligands, 5 type II and 7 type I receptors shown here give rise to 50 different combinations of ligand-receptor complexes overall. Note that many of these 50 complexes share ligand and receptor species.\ The ligand-receptor complexes phosporylate the cytoplasmic R-Smads; at this point the signal is essentially funneled into two different pathways. The decision of which one is chosen depends on the particular type I receptor in the ligand-receptor complex. The type I receptors can be divided into two groups, depending on which subgroup of R-Smads they bind and phosphorylate: the first group of type I receptors (Alk1/2/3/6, shown on the bottom right) bind and activate the R-Smads Smad1/5/8, whereas the second group (Alk4/5/7, shown on the top right) act on the R-Smads Smad2/3. The phosphorylated R-Smads then form complexes with the Co-Smad Smad4.](pcbi.0020003.g002){#pcbi-0020003-g002} At the phenotypic level, the responses are extremely diverse. The members of the TGF-β superfamily act prototypically as potent negative growth regulators, but, depending on the cell type and context, they can also induce differentiation, apoptosis, cell migration, adhesion, and extracellular matrix deposition. TGF-β itself is of particular interest in cancer research. In epithelial cells, it suppresses cellular growth by inducing G1 arrest (mediated by transcriptional activation of p15 and p21) \[[@pcbi-0020003-b004]\], and its inactivation contributes to tumorigenesis. The versatility of the pathway in eliciting different types of behavior is perhaps best epitomized by the pervasive, rather paradoxical ability of TGF-β to change its function from suppressor to promoter of growth in epithelial cells during tumor progression \[[@pcbi-0020003-b004],[@pcbi-0020003-b005]\]. Current theories for explaining the variety of responses to members of the TGF-β superfamily of ligands focus mainly on the downstream transcriptional regulatory networks they activate: transcriptional cofactors of the R-Smads are expressed at different levels in a cell-specific manner, thereby modifying downstream responses. In fact, the role reversal of TGF-β from negative to positive growth regulator has been found to be associated with a phenotypic change known as epithelial-to-mesenchymal transition, in which cells change the cofactors recruited by the R-Smads and acquire motile phenotypes \[[@pcbi-0020003-b005]--[@pcbi-0020003-b007]\]. It is striking, however, that such a variety of complex responses and intricate molecular components are connected through just two Smad channels by such a simple downstream signal transduction cascade. There is a richness of experimental observations that are difficult to reconcile with this observation. In particular, whether TGF-β acts as a growth suppressor or promoter can depend on whether the tumor cells were grown in vitro or in vivo \[[@pcbi-0020003-b008]\]. In these two different situations, the extracellular context determines the way in which cells respond to TGF-β. It has been suggested that TGF-β can suppress the growth of cells around the tumor, that it can shut down locally the immune system, and that it can promote angiogenesis. All these paracrine effects would help the growth of the tumor in vivo, where it has to compete with neighboring cells. So far, although appealing, none of these mechanisms has been identified as an alternative cause of the TGF-β role reversal. The most direct way in which the extracellular context can affect the functioning of the TGF-β pathway is through signaling of other ligands of the TGF-β superfamily. As we have mentioned, ligands and receptors form a complex interaction network, where multiple ligands share receptors, potentially coupling their signaling. All these interactions are in turn coupled to *receptor trafficking*, which is known to be a mechanism that regulates signal transduction \[[@pcbi-0020003-b009],[@pcbi-0020003-b010]\]. Trafficking has been investigated in detail in many signal transduction pathways, such as the epidermal growth factor receptor (EGFR) and G protein-coupled receptor (GPCR) pathways \[[@pcbi-0020003-b011]--[@pcbi-0020003-b013]\]. The typical way in which trafficking and signaling are coupled is by the induction of receptor internalization upon ligand binding and receptor activation, as for instance in the EGFR and GPCR pathways. After internalization, receptors can activate other signaling pathways, be modified in specific ways, and be targeted for degradation or recycling back to the plasma membrane. A peculiarity of the TGF-β pathway is that receptors are constitutively internalized, even in the absence of ligand \[[@pcbi-0020003-b014],[@pcbi-0020003-b015]\]. The trafficking route that the receptors follow, however, depends on whether or not they are in a signaling complex ([Figure 3](#pcbi-0020003-g003){ref-type="fig"}). Different routes will trigger different signaling outcomes and affect how receptors are degraded. Therefore, although the explicit implementation of the coupling is different than in the EGFR and GPCR pathways, receptor trafficking and signaling are also tightly coupled in the TGF-β pathway. ![Signaling and Trafficking in the TGF-β Pathway\ Receptors in the plasma membrane interact with the signaling peptides of the TGF-β superfamily to form active complexes. Receptors and activated ligand-receptor complexes can internalize via clathrin-coated pits into endosomes, from where the active ligand-receptor complexes phosphorylate the cytoplasmic R-Smads ("receptor Smads," either the Smad1/5/8 or the Smad2/3 group) \[[@pcbi-0020003-b033]\].\ The phosphorylated R-Smads form complexes with the Co-Smad (Smad4) and then translocate into the nucleus where they act as transcriptional regulators of about 300 target genes.\ The internalized receptors recycle back to the plasma membrane (with a characteristic time of \~30 min) via a rab11-dependent, rab4-independent pathway \[[@pcbi-0020003-b014]\]. After returning to the plasma membrane, the receptors that were actively signaling can be targeted for degradation or be used for further ligand-binding or internalization \[[@pcbi-0020003-b014]\]. Receptors that did not bind ligands are simply returned to the plasma membrane. As a consequence of the trafficking processes, only about 5%--10% of receptors are present in the plasma membrane \[[@pcbi-0020003-b015]\].\ In addition to the traditional clathrin pathway, active ligand-receptor complexes can recruit Smad7-Smurf2 \[[@pcbi-0020003-b028]\], which then targets them to lipid raft--caveolar compartments (right) for degradation \[[@pcbi-0020003-b015]\].\ The ligands do not return back to the plasma membrane, but disassociate from the receptors before recycling and undergo direct degradation via the lysosomes \[[@pcbi-0020003-b014]\].\ Note that, in addition to the ligand-induced receptor degradation, we also consider a receptor degradation pathway that functions independently of ligand-binding; this represents a "constitutive" or ligand-independent degradation pathway (left).](pcbi.0020003.g003){#pcbi-0020003-g003} Taking all the current experimental data together, it is clear that many details of the underlying processes remain largely unknown. Recent experiments \[[@pcbi-0020003-b014],[@pcbi-0020003-b015]\], however, provide key information that allows computational modeling to bridge the gap between potential molecular mechanisms and experimentally observable behavior. The TGF-β pathway is at a discovery stage where modeling can help to gain "functional" intuition. Here we characterize computationally the diverse potential types of behavior that the pathway structure itself can confer on the system. The types of behavior include responses to persistent changes in ligand concentration that can be transient or sustained and simultaneous responses to multiple ligands that can be passed downstream independently of or dependently on each other. A sustained response implies that the steady-state signaling activity is a function of the ligand concentration. In this case, the higher the ligand concentration, the higher the activity of the pathway. For a transient response that precisely returns to the prestimulus level, in contrast, the steady-state activity is always the same and the pathway can only detect changes in ligand concentration. When multiple ligands signal in a dependent fashion, the extent of the coupling can be such that one ligand can suppress the effects of another one. In this regime, the pathway does not detect ligand concentrations but ratios of concentrations. As we show here, all these types of behavior can be present in the TGF-β pathway. Which specific one is selected is determined by the interplay between trafficking and signaling. Thus the pathway can be set to detect, at the receptor level, absolute levels of ligand, temporal changes in ligand concentration, and ratios of multiple ligands. Such flexibility in the pathway behavior can lead to diverse physiological outcomes that have been associated with facilitated tumor progression. Results {#s2} ======= A Concise Computational Model {#s2a} ----------------------------- In order to study the signal processing potential of the ligand-receptor network coupled with receptor trafficking, we assemble all the essential elements into a concise mathematical model that captures the logic of the underlying processes. The main goal is to represent as much complexity as possible through a small number of quantities that have direct experimental interpretation. The essential elements we consider are ([Figure 4](#pcbi-0020003-g004){ref-type="fig"}): (i) Ligands induce the formation of receptor complexes with type II and type I receptors. (ii) Receptors and ligand-receptor complexes can be present in two spatially distinct compartments: plasma membrane and internalized endosomes. (iii) The signaling activity is proportional to the number of ligand-receptor complexes that are present in the internalized endosomes \[[@pcbi-0020003-b016]\]. (iv) Receptors and ligand-receptor complexes are continuously internalized into endosomes and recycled back to the plasma membrane \[[@pcbi-0020003-b014], [@pcbi-0020003-b015]\]. (v) Receptor degradation has a constitutive contribution, which is the same for free receptors and ligand-receptor complexes \[[@pcbi-0020003-b014],[@pcbi-0020003-b015]\]. (vi) Receptor degradation has a ligand-induced contribution, which affects only receptors that have been complexed with ligands \[[@pcbi-0020003-b014],[@pcbi-0020003-b015]\]. ![Two-Compartment Model of Receptor Trafficking and Signaling\ Graphical representation and equations for a model with one ligand that forms complexes with one type I and one type II receptor. Receptors are present in two main compartments: the plasma membrane (receptors at the cell surface) and the endosomes (internalized receptors). Receptors and ligand-receptor complexes traffic between these two compartments by internalization and recycling. Only internalized ligand-bound receptors have kinase activity. Active receptors can also be internalized in a degradation pathway (right). In addition, receptors in the plasma membrane can undergo constitutive degradation, independently of whether they are ligand-bound (left). A supply of new receptors is constantly produced by gene expression.\ The concentration of the ligand is denoted by \[ *l* \]; the numbers of type I and II receptor and ligand-receptor complexes in the plasma membrane, by \[*R~I~*\], \[*R~II~*\], and \[*l R~I~R~II~*\], respectively; and the numbers of internalized type I and II receptor and ligand-receptor complexes by \[*R~I~*\], \[*R~II~*\], and \[*l R~I~R~II~*\], respectively. (Note that type II receptors are not shown in the graphical representation.) *k~α~* is the rate constant of ligand-receptor complex formation; *p~RI~* and *p~RII~* are the rates of receptor production; *k~i~, k~r~, k~cd~,* and *k~lid~* are the internalization, recycling, constitutive degradation, and ligand-induced degradation rate constants; *α* is the fraction of active receptors that are recycled back to the plasma membrane and can interact again with the ligand.\ The signaling activity of the pathway is assumed to be proportional to the number of internalized ligand-receptor complexes, \[*lR~I~R~II~*\]. This assumption is based on the observations that R-Smad proteins become rapidly dephosphorylated after inhibition of the receptor kinase activity and that nuclear Smad localization closely follows Smad phosphorylation \[[@pcbi-0020003-b016]\].](pcbi.0020003.g004){#pcbi-0020003-g004} We use these elements to develop a mathematical model based on rate equations that describe the dynamics of both how different molecular species transform into each other and how they traffic between the different cellular compartments. We assume that internalization, recycling, and degradation rates are proportional to the number of receptors or ligand-receptor complexes; and that the formation of ligand-receptor complexes is proportional to the ligand concentration and to the type I and type II receptor concentrations. In a first step toward characterizing the effects of the coupling of signaling with receptor trafficking, we consider that only a single type of ligand is present. Explicitly, we study how the components of the canonical TGF-β pathway---one ligand (TGF-β) and two receptors (Alk5 and TGFβRII), as emphasized in [Figure 2](#pcbi-0020003-g002){ref-type="fig"}---respond to changes in ligand concentration. The mathematical equations are shown in [Figure 4](#pcbi-0020003-g004){ref-type="fig"}. For typical trafficking rates (see [Materials and Methods](#s4){ref-type="sec"}), this model closely reproduces the typical time courses of Smad phosphorylation upon addition of ligand ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}A and [5](#pcbi-0020003-g005){ref-type="fig"}B). ![Typical Time Courses of the Number of Active Receptor Complexes upon Addition of TGF-β\ The typical response to sustained changes in TGF-β concentration shows partial adaptation after reaching a maximum of activity. Different values of the parameters of the model lead to this characteristic behavior. In all panels, the TGF-β concentration is increased at time 0 to saturating values and kept constant afterward, as in inset in (A).\ (A) Behavior of the model for typical trafficking rates: internalization, *k~i~ =* (3 min)^−1^; recycling, *k~r~ =* (30 min)^−1^; constitutive degradation, *k~cd~ =* (36 min)^−1^; ligand-induced degradation *k~lid~ =* (4 min)^−1^; efficiency of recycling of active receptors, *α =* 1. Note that the trafficking rate constants are given as the inverse of the corresponding characteristic times. The production of receptors is chosen to be *p~RI~ =* 8 min^−1^ and *p~RII~ =* 4 min^−1^, which leads to \~10^3^ receptors per cell under stationary conditions in the absence of ligand. The units of ligand concentration are chosen so that the association rate constant is the unit, *k~a~ =* 1. For these units, EC~50~ ≈ 2 × 10^−4^. At time 0, the ligand concentration changes from 3×10^−5^ to 0.01. The signal peaks at \~60 min.\ (B) Comparison of the model time course (upper lane) with an experimental time course of nuclear phosphorylated Smad2 (P-Smad) as reported by Inman et al. (bottom lane) \[[@pcbi-0020003-b016]\]. The model results from (A) are shown at the experimental time points and color-coded to ease comparison.\ (C) Behavior of the model with the same parameter values as in (A), with the exception of the rate constants for *internalization and recycling* that have been *decreased* to *k~i~ =* (10 min)^−1^ and *k~r~ =* (100 min)^−1^. The signal peaks at \~180 min.\ (D) Behavior of the model with the same parameter values as in (A), with the exception of the rate constants for *internalization and recycling* that have been *increased* to *k~i~ =* (1 min)^−1^ and *k~r~ =* (10 min)^−1^. The signal peaks at \~20 min.\ (E) Behavior of the model with the same parameter values as in (A), with the exception of the rate constant for *ligand-induced degradation* that has been *decreased* to *k~lid~ =* 0 and the *efficiency of recycling* of active receptors that has been *decreased* to *α =* 0.5. This implies that ligand-receptor complexes are not degraded via the caveolae pathway. In contrast, 50% of the active ligand-receptors that come back to the plasma membrane after they have signaled are degraded.\ (F) Behavior of the model with the same parameter values as in (A), with the exception of the efficiency of recycling of active receptors that has been decreased to *α =* 0.5. These parameters account for both types (caveola-dependent and recycling-dependent) of ligand-induced degradation.](pcbi.0020003.g005){#pcbi-0020003-g005} The computational model can be used also to analyze how different parameters affect the behavior of the system. For instance, the time during which the signaling activity rises is related to the time required for internalization and recycling of the receptors. Thus, the signal will peak, or stop rising, at about 30--60 min after ligand addition. If the internalization and recycling rates are changed, the position of the peak changes accordingly ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}C and [5](#pcbi-0020003-g005){ref-type="fig"}D). Other transmembrane receptor pathways, such as the EGFR pathway, have much faster kinetics; the EGFR pathway reaches peak activity as fast as 5 min after stimulation \[[@pcbi-0020003-b017]\]. The main reason for these differences is that most of the EGF receptors are present in the plasma membrane and they are ready to signal upon the addition of ligand. A similar kinetics is also observed for many GPCRs. In the TGF-β pathway, internalization occurs continuously, and only about 5%--10% of the receptors are present in the plasma membrane at a given time \[[@pcbi-0020003-b015]\]. The remaining 90%--95% of the receptors are internalized in endosomes. Receptors need to be recycled from the endosomes back to the plasma membrane in order to be able to interact with the ligand; and this process takes about 30 min on average. We can use the computational model to study the effects of different mechanisms, such as different forms of receptor degradation, on the behavior of the system. It has been observed that the addition of ligand can stimulate the degradation of the receptors in two different ways. On the one hand, receptors complexed with ligand in the plasma membrane can be internalized through a lipid raft--caveolar degradation pathway without becoming active signalers \[[@pcbi-0020003-b015]\]. The behavior obtained in this case is the one we have already described ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}A). On the other hand, ligand-receptor complexes can follow the standard internalization clathrin pathway, signal, and then be targeted for degradation upon returning to the plasma membrane \[[@pcbi-0020003-b014]\]. Interestingly, when this mechanism is explicitly modeled, the behavior obtained ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}E) is qualitatively the same as the previous one ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}A). Likewise, when both degradation mechanisms are considered together, a similar type of behavior is also obtained ([Figure 5](#pcbi-0020003-g005){ref-type="fig"}F). Thus, at this level of detail and for this range of parameter values, different mechanisms that implement ligand-induced degradation can lead to similar behavior. Control of the Signal: Transient versus Permanent Responses {#s2b} ----------------------------------------------------------- How is it possible to modify the form in which the system responds to changes in TGF-β concentration? A mathematical analysis of the model (see below and Materials and Methods for details) indicates that the key quantity that determines the qualitative behavior of the pathway is the ratio of the constitutive to the ligand-induced rate of degradation, referred to, in short, as the constitutive-to-induced degradation ratio (CIR). This quantity compares the rates of two degradation processes and, in general, does not have a simple expression in terms of rate constants. Depending on the CIR, a permanent change in ligand concentration can elicit responses between two extremes ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}). For low CIR, the ligand-induced degradation process dominates and there is a transient increase in signaling activity that returns to pre-stimulus levels ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}A and [6](#pcbi-0020003-g006){ref-type="fig"}B). For high CIR, the constitutive degradation process dominates and there is a permanently elevated level of signaling activity ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}E and [6](#pcbi-0020003-g006){ref-type="fig"}F). {ref-type="fig"}A with the exception that the *constitutive* and *ligand-induced* degradation rates have been *decreased* and *increased* by a factor three, respectively: *k~cd~ =* (3 × 36 min)^−1^; ligand-induced degradation *k~lid~ =* (4/3 min)^−1^.\ (C and D) Same parameter values as in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}A. Figure in (C) is exactly the same as [Figure 5](#pcbi-0020003-g005){ref-type="fig"}A.\ (E and F) Same parameter values as in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}A with the exception that the *constitutive* and *ligand-induced* degradation rates have been *increased* and *decreased* by a factor three, respectively: *k~cd~ =* (36/3 min)^−1^; ligand-induced degradation *k~lid~ =* (3 × 4 min)^−1^.](pcbi.0020003.g006){#pcbi-0020003-g006} For intermediate CIR, the behavior of the system is a mixture of both limiting types of behavior, with transient and permanent components ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}C and [6](#pcbi-0020003-g006){ref-type="fig"}D). The precise parameter values influence the amplitude and characteristic time of the response (see for instance [Figure 5](#pcbi-0020003-g005){ref-type="fig"}C and [5](#pcbi-0020003-g005){ref-type="fig"}D), but its qualitative shape, that is, whether the response is transient ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}A and [6](#pcbi-0020003-g006){ref-type="fig"}B) or permanent ([Figure 6](#pcbi-0020003-g006){ref-type="fig"}E and [6](#pcbi-0020003-g006){ref-type="fig"}F), depends only on the CIR. The intuitive explanation of such types of behavior is as follows (for a detailed mathematical analysis, see [Materials and Methods](#s4){ref-type="sec"}). The probability for a receptor to bind the ligand, and therefore to become active, increases with the ligand concentration. If the ligand does not induce the degradation of receptors, the number of receptors remains constant and the total activity increases when the ligand concentration increases. If the ligand induces the degradation of the receptors, the number of receptors starts to decrease after ligand addition, which will eventually attenuate the signal. At steady state, the production and degradation of receptors equal each other. In the limit of the CIR going to 0, the signal adapts completely because degradation is proportional to the activation of receptors, and therefore activation is also proportional to the production of receptors. Thus, it is the receptor production rate, not the ligand concentration, that determines the steady-state signaling activity. There are clear examples in other signal transduction pathways that show that these two limiting types of behavior can potentially lead to different physiological outcomes. For instance, transient activation of the MAPK cascade by EGF leads to cell proliferation. In contrast, permanent activation of the MAPK cascade by NGF leads to cell differentiation. In both cases, activation of the MAPK cascade induces the expression of a negative regulator that shuts down the activity of this cascade. The differences between EGF and NGF have been attributed to additional pathways activated by NGF that can prevent the inactivation of the MAPK cascade \[[@pcbi-0020003-b018]\]. Our model shows that such transient and permanent types of behavior can also be achieved by just changing the trafficking patterns, in particular by adjusting the CIR, without the need for explicitly expressing a negative regulator to shut down the cascade after signaling. Remarkably, the duration of the signaling activity also seems to affect the physiological outcomes triggered by TGF-β \[[@pcbi-0020003-b019]\]. Epithelial cells that are sensitive to the antiproliferative effects of TGF-β (HaCaT and Colo-357) have sustained activity of more than 6 h. In contrast, pancreatic tumor cell lines (PT45 and Panc-1) show short transient activity of about 1--2 h. Such a short transient confers resistance to the antiproliferative effects of TGF-β but maintains other responses to TGF-β that can lead to increased malignancy and invasiveness \[[@pcbi-0020003-b019]\]. In our model, those differentiated types of behavior arise naturally for different trafficking patterns. In particular, short transients and sustained responses imply a low and a high CIR, respectively. Control of the Signals: Coupled versus Uncoupled Channels {#s2c} --------------------------------------------------------- In vivo conditions, in contrast to those typical of in vitro experiments, expose cells to complex environments with many different growth factors. When multiple ligands of the TGF-β superfamily are present at the same time, they are likely to affect each other\'s signaling ([Figure 2](#pcbi-0020003-g002){ref-type="fig"}). To study how multiple simultaneous input signals are integrated into coordinated transcriptional responses, we extend our computational model to consider two ligands that signal through two different type II receptors and a shared common type I receptor (see [Materials and Methods](#s4){ref-type="sec"} for the mathematical equations). This is the simplest case of signal integration. Intuitively, one should expect signals to be coupled when the shared receptor is saturated with ligands and uncoupled when ligand concentrations are low. At saturation, increasing the concentration of one ligand, and thus the concentration of the corresponding ligand-receptor complex, will take the shared receptor away from the complex formed by the other ligand, thus decreasing its signaling. A mathematical analysis of the model (see [Materials and Methods](#s4){ref-type="sec"}) indicates that even when the receptors are far from ligand-saturating conditions, it is possible for signals to affect each other. The key element is again receptor trafficking. In essence, the coupling arises because the induction of degradation of the common receptor by one ligand attenuates the effects of the other ligand, which also requires the common receptor to transduce the signal. For pathways working away from receptor saturation, the interplay between trafficking and signaling determines how multiple simultaneous signals are passed downstream. As in the single ligand case, there are two extreme types of behavior ([Figure 7](#pcbi-0020003-g007){ref-type="fig"}): For low CIR, the ligand-induced degradation process dominates and signals are completely coupled ([Figure 7](#pcbi-0020003-g007){ref-type="fig"}A and [7](#pcbi-0020003-g007){ref-type="fig"}B). For high CIR, the constitutive degradation process dominates and signaling is uncoupled ([Figure 7](#pcbi-0020003-g007){ref-type="fig"}E and [7](#pcbi-0020003-g007){ref-type="fig"}F). {ref-type="fig"}, a key control quantity of the qualitative behavior of the system is the CIR. The parameter values for (A--F) are the same as in [Figure 6](#pcbi-0020003-g006){ref-type="fig"}A--[6](#pcbi-0020003-g006){ref-type="fig"}F, respectively.](pcbi.0020003.g007){#pcbi-0020003-g007} When the signals are completely coupled, the steady-state number of all ligand-receptor complexes remains constant and is independent of the ligand concentration. In this case, increasing one signal will decrease the other one by the same amount. When signals are uncoupled, the numbers of each species of ligand-receptor complexes change independently of each other. In general, for intermediate CIR, signals will show some degree of coupling ([Figure 7](#pcbi-0020003-g007){ref-type="fig"}C and [7](#pcbi-0020003-g007){ref-type="fig"}D). These results demonstrate that changes in trafficking patterns, and the corresponding degradation, can alter the way in which the pathway integrates multiple, simultaneous signals. The completely coupled case is especially interesting because it indicates that one ligand can potentially inhibit the effects of another one. Ligand-induced degradation is thus not only a mechanism for achieving transient responses, but also for coupling multiple signals. The fact that TGF-β can signal not only via Alk5 but also via Alk1 and Alk2 ([Figure 2](#pcbi-0020003-g002){ref-type="fig"}) potentially couples TGF-β signals to those of Activin A, BMP 6, BMP 7, and MIS. Thus, if TGF-β loses its growth suppressor properties, it could promote growth by inhibiting other growth suppressor pathways. For instance, there are dominant negative TGFβRII mutants that when overexpressed attenuate the response to TGF-β \[[@pcbi-0020003-b020]\]. The presence of any of these mutant receptors and TGF-β results in the formation of futile receptor complexes that can target receptors for degradation, or take receptors away, which otherwise would be available to transduce the signals of other members of the TGF-β superfamily. Context-Dependent Response to TGF-β {#s2d} ----------------------------------- The role reversal of TGF-β from negative to positive growth regulator is a widespread feature of tumor progression and is often associated with endogenous overexpression of TGF-β. As we have mentioned in the introduction, it is associated in some situations with the epithelial-to-mesenchymal transition \[[@pcbi-0020003-b006],[@pcbi-0020003-b007]\]. Under these conditions, the transcriptional program of tumor cells changes so that the Smad-activated genes promote rather than repress growth. In other situations such a transition does not seem to be present. It has been observed in breast, prostate, and colon cancer cell lines that the action of TGF-β as growth promoter or suppressor depends on whether the cells were grown in an in vitro environment or in vivo in mouse xenografts \[[@pcbi-0020003-b008],[@pcbi-0020003-b021],[@pcbi-0020003-b022]\]. The reasons for such a change remain largely unknown. It has been speculated that it could be a consequence of the effects of TGF-β on the in vivo microenvironment of the tumor cells. Another possibility is that other growth factors, such as EGF, affect how TGF-β is ultimately coupled to the cell cycle. Our model explicitly shows that the role reversal is a potentially intrinsic consequence of the design of the ligand-receptor interaction network and trafficking machineries and that it could be the result of TGF-β attenuating the effects of growth-suppressing signals of other members of the TGF-β superfamily that might be present in the in vivo cell environment. Simultaneous Perfect Adaptation and Coupled Signaling {#s2e} ----------------------------------------------------- Our model also indicates that the conditions that give rise to completely coupled integration of multiple signals are the same that, in a single-ligand system, cause the signaling activity to completely adapt to its prestimulus level. Remarkably, this concurs with observations in prostate cancer cell lines, which show that the in vivo context can not only make TGF-β a growth promoter but also that the in vitro response to TGF-β is transient \[[@pcbi-0020003-b008]\]. One should expect the extracellular environment of growth factors to be more complex in vivo than in vitro. This relationship between in vivo and in vitro behavior and its connection to the coupling between receptor trafficking and signaling underscores the importance of understanding how signal transduction pathways are embedded within the cellular microenvironment under physiologically relevant conditions. Not only mutations in the canonical pathway but also changes in trafficking patterns can move the pathway to a different functioning point. The qualitative results of our model, such as the regimes leading to transient and permanent responses as well as to completely coupled and uncoupled modes of signal integration, do not depend on the details of the model but on general properties (see [Materials and Methods](#s4){ref-type="sec"}). Thus the main ideas are also relevant to other signal transduction pathways that are coupled to receptor trafficking. In particular, revisiting the experimental data, one can see that the interplay between adaptation and signal integration (Materials and Methods) is also present in the EGFR pathway (see [Figures 4](#pcbi-0020003-g004){ref-type="fig"} and [5](#pcbi-0020003-g005){ref-type="fig"} of reference \[[@pcbi-0020003-b023]\]), in which down-regulation of erbB-2 by EGF concurs with adaptation of the signal transmitted by EGFR. Discussion {#s3} ========== Cellular functions are controlled by networks of interacting molecules that operate at different levels of organization \[[@pcbi-0020003-b024]--[@pcbi-0020003-b026]\]. Here, we have developed a concise computational model of the TGF-β pathway that shows that the receptors for the TGF-β superfamily of ligands are not just passive signal transducers. They are organized in a network that is able to process the signals before passing them downstream. Changes in receptor trafficking patterns can modify the type of behavior of the pathway in response to single and multiple ligand inputs. Already at the receptor level, the pathway can detect absolute levels of ligands, temporal changes in ligand concentration, and ratios of multiple ligands. This extra level of regulation can explain a wide variety of phenomena, such as the counterintuitive role reversal of TGF-β from suppressor to promoter of growth, and leads to a unified interpretation of seemingly disparate experimental observations. A key quantity that determines the qualitative behavior of the pathway is the CIR of the receptors. For low CIR, the pathway responds transiently to sustained changes in ligand concentration, and the signaling activities of multiple simultaneous ligands become dependent on each other. Ligand-induced degradation is thus not only a mechanism for achieving transient responses and perfect adaptation, but also for coupling multiple signals. Various experiments can be designed to test the predicted types of behavior. The most direct evidence would come from biochemical measurements of ligand-induced and constitutive degradation rates \[[@pcbi-0020003-b014],[@pcbi-0020003-b015]\]. The observed degradation rates can then be related to measurements of the levels of signaling activity. For instance, decreasing ligand-induced degradation by blocking the lipid raft--caveolar pathway with nystatin, as in reference \[[@pcbi-0020003-b015]\], should bias the system behavior toward permanent responses to step changes in ligand concentration. Molecular interventions, such as RNAi against the mRNA of proteins involved in the trafficking processes, and their effects on signaling activity would provide indirect evidence that can be related to the details of the model. Because of their well-established role in intracellular trafficking, small GTPase rab proteins \[[@pcbi-0020003-b027]\] are prime candidates for this type of approach. Certain experimental observations \[[@pcbi-0020003-b008],[@pcbi-0020003-b019]\] in conjunction with our model suggest that cells with phenotypes that favor tumor progression have low CIR. Further experiments could test the extent of this correlation by comparing trafficking rates and signaling activity between different cancerous and non-cancerous cell lines. Materials and Methods {#s4} ===================== Internalization rate. {#s4a} --------------------- It has been reported in Table 1 of reference \[[@pcbi-0020003-b015]\] that after 15 min of labeling the receptors at the plasma membrane, only 8%, 6%, 4%, and 2% of the labeled receptors remain at the plasma membrane (the different percentage values correspond to different experimental conditions). The remaining labeled receptors have been internalized in either caveolin-1 positive or caveolin-1 negative vesicles. By using the formula , where *k~i~* is the internalization rate and *f~t~* the fraction of labeled receptors that remain at the plasma membrane after a time *t,* we obtain internalization rates of 1/5.9, 1/5.3, 1/4.7, and 1/3.8 min^−1^, respectively. We have chosen *k~i~ =* 1/(3 min) for comparison with experimental data in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}B. [Figure 4](#pcbi-0020003-g004){ref-type="fig"} of reference \[[@pcbi-0020003-b014]\] shows that 1.7% of the total number of receptors is internalized per minute. When this value is rescaled by the fraction of receptors in the plasma membrane, it translates into 18% of surface receptors internalized per minute. This implies that this internalization rate is *k~i~ =* 1/(5.3 min), which is similar to the results obtained from reference \[[@pcbi-0020003-b015]\]. The details of this rescaling are as follows. Mathematically, the internalization rate constant is defined as , where *N~sur~* is the number of receptors at the plasma membrane. Mitchell et al. \[[@pcbi-0020003-b014]\] measured , where *N~tot~* is the total number of receptors. The fact that most of the receptors are internalized, so that *N~sur~*≈0.1*N~tot~*, leads to *k~i~*≈10*k~Mitchell~* The factor 0.1 results from the fact that the internalization rate is about 10× higher than the recycling rate. Therefore, under stationary conditions, the number of internalized receptors is 10× higher than the number of receptors at the plasma membrane. Ligand-induced degradation rate. {#s4b} -------------------------------- Active ligand-receptor complexes in lipid raft--caveolar compartments can recruit Smad7-Smurf2 \[[@pcbi-0020003-b028]\], which then targets them for degradation \[[@pcbi-0020003-b015]\]. Reference \[[@pcbi-0020003-b015]\] shows that receptors are internalized through the clathrin pathway and lipid raft--caveolar compartments with similar rates. We have chosen *k~lid~ =* 1/(4min) for comparison with experimental data in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}B. Constitutive degradation rate. {#s4c} ------------------------------ [Figure 3](#pcbi-0020003-g003){ref-type="fig"} of reference \[[@pcbi-0020003-b015]\] shows that when the lipid raft--caveolar pathway is blocked with nystatin, only \~30% of the initially labeled receptors remain in the cell after 8 h. This gives a characteristic degradation time of \~400 min with respect to the total number of receptors. Rescaling this number to the plasma membrane receptors, we obtain . We have chosen *k~cd~ =* 1/(36 min) for comparison with experimental data in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}B. Recycling rate. {#s4d} --------------- [Figure 3](#pcbi-0020003-g003){ref-type="fig"} of reference \[[@pcbi-0020003-b014]\] shows that after about 30 min cells stop secreting internally labeled TGF-β receptors. This recycling rate is similar to that for the EGF receptor. We have chosen *k~r~ =* 30 min for comparison with experimental data in [Figure 5](#pcbi-0020003-g005){ref-type="fig"}B. Steady and quasi-steady state analysis. {#s4e} --------------------------------------- Here we study mathematically the properties of the steady state of the system with a single ligand. By equating to 0 the derivatives in the model equations of [Figure 4](#pcbi-0020003-g004){ref-type="fig"}, we obtain that the steady-state number of internalized ligand-receptor complexes is  where the steady-state number of type I and type II receptors at the plasma membrane are obtained by solving the equations  with  The solution of these equations is  which leads to  For low values of *c,* Equation 5 reduces to  which indicates that the steady-state number of internalized ligand-receptor complexes is proportional to the ligand concentration and the production of each receptor type. For high values of *c,* in contrast, we obtain  and  Therefore, for high ligand concentration or low constitutive degradation, the steady-state number of internalized ligand-receptor complexes is controlled by the receptor with the smallest production rate and this number does not depend on the ligand concentration. The case of high *c* and low constitutive degradation is especially interesting because the steady-state signal does not depend on the ligand concentration, even when the ligand is present in small quantities. An important question to address now is: can the system detect changes in concentration in this regime? When the recycling rate is much lower than the internalization rate, the number of ligand-receptor complexes in the plasma membrane equilibrates faster than all the other variables. Therefore assuming quasi-equilibrium in this variable, , we obtain that, upon changes in the ligand concentration (Δ\[*l*\]), the changes in the number of ligand-receptor complexes in the plasma membrane (Δ\[*lR~I~R~II~*\]) follow the equation  Note that we have assumed that the number of receptors in the plasma membrane is conserved at these time scales. For small changes in ligand concentration, we obtain  This expression indicates that for high *c,* low constitutive degradation, and slow recycling (compared to internalization), the system can detect changes in ligand concentration while keeping a steady-state signal that does not depend on ligand concentration. Two-compartment model of receptor trafficking for two ligands. {#s4f} -------------------------------------------------------------- The equations for a system with two ligands with concentrations \[*l~1~*\] and \[*l~2~*\] and are:           The variables \[*R~I~*\], \[*R~II,~* ~1~\], and \[*R~II,~* ~2~\], are the numbers of type I and type II receptors in the plasma membrane; and \[*l* ~1~ *R~I~R~II,~* ~1~\] and \[*l* ~2~ *R~I~R~II,~* ~2~\] refer to the corresponding ligand-receptor complexes. The overline indicates internalized receptors and ligand-receptor complexes. The signaling activity triggered by each ligand is assumed to be proportional to the corresponding number of internalized ligand-receptor complexes. *k~α~* is the rate constant of ligand-receptor complex formation; *p~RI~*, *p~RII,~* ~1~, and *p~RII,~* ~2~ are the rates of receptor production; *k~i~, k~r~, k~cd~,* and *k~lid~*, are the internalization, recycling, constitutive degradation, and ligand-induced degradation rate constants; and α is the fraction of active receptors that are recycled back to the plasma membrane and can interact again with the ligand. Coupled signaling and perfect adaptation. {#s4g} ----------------------------------------- Computational modeling offers precise insights into the functioning of the TGF-β pathway. It is possible to go a step further and generalize the conditions that give rise to different qualitative types of behavior. Let us consider two ligands (*l* ~1~ and *l* ~2~), one type I receptor (*R~I~*), and two type II receptors (*R~II,~* ~1~ and *R~II,~* ~2~). The type I receptor is shared among the two ligand-receptor complexes \[*l* ~1~ *R~I~ R~II,~* ~1~\] and \[*l* ~2~ *R~I~ R~II,~* ~2~\]. The following conservation equations refer to the common type I receptor at steady-state. Under stationary conditions, the number of receptors produced (by gene expression) is equal to the number of receptors degraded:  where *p* is the receptor production rate; and *d~const~* and *d~lid~* are the constitutive and ligand-induced degradation rates, respectively. Assuming that a fraction δ of the activated receptors is degraded through a ligand-induced degradation process, we can express *d~lid~* as  where *i~a~* ~1~ and *i~a~* ~2~ are the rates of formation of the ligand-receptor complexes (\[*l* ~1~ *R~I~ R~II,~* ~1~\] and \[*l* ~2~ *R~I~ R~II,~* ~2~\], respectively). Therefore,  We explicitly consider two limiting cases: In the first case, there is no ligand-induced degradation (*d~const~* \> 0, δ = 0). Therefore, we have *d~lid~ =* 0, which leads to *p = d~const~*. Because *d~const~ ≡ d~const~*(*R~T~*) is a function of the total number of receptors *R~T~,* the previous condition indicates that the number of receptors remains constant *R~T~ = d~const~* ^−1^(*p*), where *d~const~* ^−1^ is the inverse function of *d~const~*. For instance, if the constitutive receptor degradation follows first order kinetics, *d~const~ =* γ*R~T~*, then *R~T~ = p*/γ. Under these conditions, if the rate of formation of complexes (*i~a~* ~1~ + *i~a~* ~2~) is small (for instance, for low ligand concentrations) compared with constitutive internalization and degradation, there is no coupling between signaling channels. In the second case, there is only ligand-induced degradation (*d~const~ =* 0, δ \> 0). Therefore, we have *p =* δ (*i~a~* ~1~ + *i~a~* ~2~), which implies that the formation of one ligand-receptor complex excludes the formation of the other one. In this case, the number of receptors in the plasma membrane does not remain constant, but is adjusted so that for a given ligand concentration the rate of formation of complexes (*i~a1~* + *i~a2~*) remains equal to *p*/δ. As an explicit example, the kinetics *i~a~* ~1~ = *l* ~1~ *R~I~ R~II,~* ~1~ and *i~a~* ~2~ = *l* ~1~ *R~I~R~II,~* ~2~ implies *R~I~ = p*/(*l* ~1~ *R~II,~* ~1~ + *l* ~2~ *R~II,~* ~2~). Under these conditions, the completely coupled mode of signal integration arises even for low ligand concentrations. The conditions that lead to the completely coupled mode of signal integration also lead to perfect adaptation for a single ligand. Consider, for example, *i~a~* ~2~ = 0. When there is only ligand-induced degradation, the fact that the rate of formation of complexes remains constant implies *p =* δ*i~a~* ~1~. The steady state of the system is fixed irrespective of the ligand concentration. Consequently, changes in ligand concentration can only elicit transient responses that completely adapt to the prestimulus level and the system exhibits perfect adaptation \[[@pcbi-0020003-b029]\]. We are indebted to Joan Massagué and Van Le for invaluable help during the early stages of this work. We are also very grateful to Gary Bader, Debora Marks, Leonor Saiz, Nikolaus Schultz, Wenying Shou, and Stas Shvartsman for numerous discussions, comments, and suggestions. CS was funded in part by the Alfred W. Bressler Scholars Endowment Fund. **Author contributions.** JMGV conceived and designed the experiments, performed the experiments, and analyzed the data. RJ and CS contributed reagents/materials/analysis tools. JMGV and RJ wrote the paper. CS critically edited the manuscript. **Competing interests.** The authors have declared that no competing interests exist. A previous version of this article appeared as an Early Online Release on December 9, 2005 (DOI: [10.1371/journal.pcbi.0020003.eor](10.1371/journal.pcbi.0020003.eor)). CIR : constitutive-to-induced degradation ratio EGFR : epidermal growth factor receptor GPCR : G protein-coupled receptor R-Smad : regulatory Smad

INTRODUCTION {#SEC1} ============ Erythroid differentiation represents an excellent model system for exploring stage-specific post-transcriptional remodeling of gene expression during terminal differentiation. Fluorescence-activated cell sorting (FACS) makes possible isolation of discrete, highly purified populations of cells as they differentiate, enucleate to form reticulocytes and ultimately mature into red cells. Early progenitors known as burst-forming unit-erythroid (BFU-E) and colony-forming unit-erythroid (CFU-E) can be highly purified by this approach, as can proerythroblasts (proE) and several stages of terminally differentiating erythroblasts termed basophilic erythroblasts (basoE), polychromatophilic erythroblasts (polyE) and orthochromatophilic erythroblasts (orthoE). We and others have analyzed RNA-seq libraries prepared from these purified populations of human erythroid cells to gain new insights into the evolving erythroid transcriptome at the level of gene-level expression, alternative splicing, non-coding RNA expression, etc. ([@B1]--[@B3]). Moreover, similar analysis of mouse erythroblast populations allows for comparisons of gene expression patterns among mammalian species ([@B1],[@B3]). Proliferating mammalian erythroblasts exhibit a robust, dynamic alternative splicing program ([@B2],[@B4]--[@B5]) enriched in genes involved in cell cycle, organelle organization, chromatin function and RNA processing ([@B2]). A prominent feature of the erythroblast splicing program is a number of alternative splicing 'switches' that increase PSI (percent spliced in) values predominantly in late erythroblasts at the polyE and orthoE stages, temporally correlated with major cellular remodeling as cells conclude their proliferation phase and prepare for enucleation. Splicing switches can alter protein function in physiologically important ways, e.g. upregulation of exon 16 splicing in protein 4.1R transcripts leads to synthesis of protein isoforms that bind spectrin and actin with high affinity, mechanically strengthening the red cell membrane prior to release into the circulation ([@B6]--[@B8]). In most cases, however, understanding the physiological functions of alternative protein isoforms generated via the erythroblast splicing program remains a challenge for future studies. Intron retention (IR) is emerging as an unexpectedly rich contributor to transcriptome diversity, providing a mechanism for gene regulation during normal differentiation and development. Recent surveys have revealed extensive IR events with distinct tissue-, developmental- and stress-specific expression patterns ([@B9]--[@B13]), suggesting precise regulation by the splicing machinery. Widespread intron retention also characterizes many cancer transcriptomes ([@B14]). Global screens across many cell and tissue types from human and mouse show surprising abundance of IR, such that 35% of multi-exon genes contain intron(s) with ≥50% retention in at least one cell type ([@B15]). IR events are also particularly abundant in plants ([@B16]). Several functions have been proposed for IR, which could provide a post-transcriptional mechanism to downregulate gene expression by inducing degradation by nuclear surveillance machinery ([@B13]) or by nonsense-mediated decay (NMD) ([@B12]). Alternatively, IR could represent a conditional block to gene expression that might be relieved to facilitate intron removal in response to appropriate signaling events ([@B17]) or developmental cues ([@B18]). Previous studies of the erythroid transcriptome entirely over-looked the IR component of the splicing program. To understand the role of IR in mammalian erythroblasts during terminal erythropoiesis, we developed custom software available at (<https://github.com/pachterlab/kma>) to analyze IR in RNA-seq data and applied these methods to study IR in populations of human erythroblasts from proE to orthoE. These new studies show that erythroblasts elaborate an extensive and diverse intron retention program encompassing numerous essential erythroid genes including those encoding splicing factors and proteins involved in iron homeostasis. Differentiation stage-specific changes in IR efficiency largely paralleled switches in splicing of cassette exons described earlier ([@B2]), reinforcing and expanding the concept that careful regulation of RNA processing plays a major role in terminal erythroid differentiation as cells mature along the path from proE to orthoE. MATERIALS AND METHODS {#SEC2} ===================== Computational methods {#SEC2-1} --------------------- RNA-seq reads were mapped using Bowtie v2.1.0 to an augmented transcriptome output by KeepMeAround (KMA) as described in our recently posted preprint (arXiv:[1510.00696](http://arxiv.org/abs/1510.00696)). Transcripts and introns were then quantified using eXpress v1.5.1 ([@B19]). We identified an unambiguous set of 186 838 quantifiable introns in the RefSeq transcriptome, but only 10 152 unique introns passed filters (below) in every condition. Intron retention values were calculated by taking the ratio of transcripts per million (TPM) ([@B20]) values of the intron (numerator) to the sum of TPM values of the overlapping isoforms and intron (denominator), resulting in IR levels on the \[0, 1\] scale. To reduce false positives, we removed introns with fewer than three uniquely mapped reads, denominator values of less than 1 TPM (excluding the intron expression) and introns with zero coverage regions longer than 20% of the intron length (Supplementary Information in ([@B2])). KMA\'s hypothesis testing feature was then used on the filtered set of introns to test whether retention levels were higher than expected given the background retention levels in each experimental condition. This test also incorporates biological replicates to further reduce the chance of false positives. Cluster analysis was performed using *k*-means clustering on a set of introns that passed the above filters in every sample and every condition. After clustering, GO analysis was performed using DAVID tools ([@B21],[@B22]) on the genes from the introns in clusters C1-C6. Code for analysis and plots can be found at <https://github.com/pimentel/erythroid_ir_analysis>. Splice site strength was calculated using MaxEntScan with the maximum entropy score and default parameters ([@B23]). RNA-seq data {#SEC2-2} ------------ RNA-seq data obtained from five highly purified human erythroblast populations---proerythroblasts (proE), early basophilic erythroblasts (e-basoE), late basophilic erythroblasts (l-basoE), polychromatophilic erythroblasts (polyE) and orthochromatophilic erythroblasts (orthoE) ([@B24])---is available at GSE53635. The data include three biological replicates of each population. Human granulocyte RNA-seq data ([@B12]) was downloaded from GSE48307. For other tissues, we imported wiggle plots, showing RNA-seq coverage along the genome, that were generated from Illumina BodyMap 2.0 data available at <http://www.ensembl.org/info/genome/genebuild/rnaseq_annotation.html>. Erythroblast cultures {#SEC2-3} --------------------- To prepare RNA and protein for further analysis of IR, CD34+ cells were purified from cord blood and differentiated into erythroblasts over the course of 16 days as described ([@B24]). Nuclear isolation {#SEC2-4} ----------------- Nuclei were prepared from ∼20 million erythroblasts according to published methods ([@B25]), with minor modifications. In brief, the erythroblast plasma membrane was lysed using 0.05% NP40, and nuclei were separated from the reddish hemoglobin-rich cytoplasmic fraction by centrifugation through a sucrose cushion at ∼2000 rpm. The whitish nuclear pellet was rinsed with ice-cold phosphate buffered saline containing 1 mM ethylenediaminetetraacetic acid and was resuspended gently to generate a turbid suspension in which nuclei were microscopically verified. Purity of the nuclear fractions was further confirmed by immunoblotting with antibodies to U1--70K protein (a kind gift from D. Black, UCLA). RT-PCR analysis of IR transcripts {#SEC2-5} --------------------------------- RNA was purified from cultured erythroblasts as described previously using RNeasy columns according to the manufacturer\'s instructions (Qiagen), but with the addition of a DNase step to eliminate potential contamination by genomic DNA. RNA from nuclear and cytoplasmic fractions was prepared using Trizol (Life Technologies). To provide additional assurance that intron-containing polymerase chain reaction (PCR) products were not derived from contaminating genomic DNA, we designed PCR assays to span at least one constitutively spliced intron as well as the candidate retained intron. PCR reaction conditions were adjusted to allow for amplification of IR products ≥3 kb in length (denaturation at 95°C for 20″, annealing at 60°C for 10″, extension at 70°C for 1′15″; 35 cycles) using KOD polymerase in the presence of betaine to enhance amplification. PCR products were analyzed on either 2% agarose gels (for products \>1.5 kb) or 4.5% acrylamide gels. All PCR products discussed in the manuscript were confirmed by DNA sequencing. RESULTS {#SEC3} ======= IR is a major feature of gene expression in differentiating human erythroblasts {#SEC3-1} ------------------------------------------------------------------------------- We mapped RNA-seq reads from highly purified human erythroblast populations to an augmented transcriptome, including introns, in order to detect all expressed regions independent of existing transcript annotations. Preliminary inspection of mapping data in the wiggle plot format, displaying RNA-seq read density along the genome, revealed that most introns were efficiently spliced in all erythroblast populations. For example, the α and β globin genes exhibited major peaks in read density over the exons and deep troughs in intronic regions due to highly efficient joining of exons and removal of introns during pre-mRNA splicing (Figure [1A](#F1){ref-type="fig"}, upper). Many housekeeping genes such as those encoding glycolytic enzymes also exhibited negligible IR (Supplementary Figure S1). In contrast, a number of important erythroid transcripts exhibited substantial IR (Figure [1A](#F1){ref-type="fig"}, lower). A very prominent IR event was found in the mitoferrin-1 gene (SLC25A37), which encodes a mitochondrial iron import protein that is critical for iron homeostasis and abundant heme biosynthesis in late erythroblasts. SLC25A37 intron 2, ∼2 kb in length, was highly retained in orthoE, while introns 1 and 3 were retained at much lower levels. Another major IR event occurs in the SPTA1 gene, encoding the structural protein α-spectrin best known for its essential role in promoting assembly of a mechanically stable erythroid membrane skeleton. Intron 20 (1.8 kb) exhibited substantial retention. We also observed moderate IR in EPOR (encoding the erythropoietin receptor), and in spliceosome-associated RNA binding proteins including UAP56 (encoded by DDX39B) and SAP155 (encoded by SF3B1). The latter is an important RNA splicing factor that is frequently mutated in the RARS (refractory anemia with ringed sideroblasts) subtype of myelodysplasia syndrome (MDS). As reported previously ([@B17]), IR also occurs in the CLK1 gene, encoding a tyrosine kinase that phosphorylates splicing factors of the SR protein family. {#F1} Intron retention was validated by RT-PCR analysis of RNA isolated from multiple independent cultures of human erythroblasts. Amplicons spanned at least one retained intron and one constitutive intron, in order to demonstrate that intron-specific retention occurs within the context of a larger transcript, and to rule out artifacts that might be caused by DNA contamination. Figure [1B](#F1){ref-type="fig"} illustrates several examples in which both the fully spliced and the intron retention products were amplified. These results confirmed that RNA-seq reads mapping to these introns were expressed in the context of intact introns retained between flanking exons in stably expressed erythroblast transcripts. Global analysis of IR in erythroblasts {#SEC3-2} -------------------------------------- Genome-wide study of IR in erythroblasts was performed using new computational tools that assign a retention value to every intron relative to expression of the flanking exons (see 'Materials and Methods' section). Applying these tools to RNA-seq data from the five erythroblast populations revealed wide variations in percent intron retention (IR), length of retained introns and number of introns retained per transcript. Hundreds of introns were retained at IR \> 0.10 in at least one erythroblast population (see Supplementary Table S1 for a complete summary of calculated IR values). Some of these represented single IR events in otherwise efficiently-spliced transcripts; however, there were also many transcripts that retained multiple introns. The distribution of IR values across the erythroblast populations showed that overall IR increases as erythroblasts differentiate, with highest IR in cells at the orthoE stage (Supplementary Figure S2). These data demonstrate that a robust IR program affects the expression of many important erythroid genes. We reasoned that dynamic regulation of IR events might be an important gene regulatory mechanism during terminal erythropoiesis, similar to stage-specific exon splicing switches executed in late erythroblasts ([@B2]). Cluster analysis of IR values for each intron at all five maturational stages revealed nine groups of introns (Figure [2](#F2){ref-type="fig"}). Clusters C1 and C2, comprising ∼470 introns, represent developmentally dynamic events that substantially increase IR in the last two differentiation stages. In contrast, clusters C3--C9 constitute a graded series of developmentally stable intron groups with differentiation-independent IR values, i.e. relatively little change from proE to orthoE. IR is relatively high in C3 but much lower in C9. {#F2} Analysis of intragenic IR patterns revealed that C3, and to a lesser extent in C4, differed qualitatively from the other clusters in that many of the highly retained introns mapped to the first or last intron of a transcript (Supplementary Figure S3). Some of these events might therefore represent alternative initiation or termination of transcription, rather than intron retention *per* *se*. However, a few high-level IR events did localize to internal introns (e.g. in SLC25A37). Computational predictions of developmentally dynamic and developmentally stable IR classes were validated by examination of RNA-seq read-mapping patterns for individual genes (Supplementary Figure S4A), and by experimental analysis of selected introns using via RT-PCR (Supplementary Figure S4B). Both approaches confirmed that some introns exhibited increased IR in late erythroblasts, while other introns maintained a more constant IR during terminal erythropoiesis. These findings mirror earlier observations of exon splicing patterns: erythroblasts alternatively splice hundreds of cassette exons, many of which maintain stable PSI values throughout terminal erythropoiesis, while a select subset undergo significant increases in PSI in late-stage polyE and orthoE ([@B2]). To evaluate the relative abundance of IR transcripts in erythroblasts, we compared IR values with gene expression levels. SLC25A37, one of the most highly expressed genes in orthoE, exhibits ∼50% retention of intron 2, so that the IR isoform is estimated to rank as the 10th most abundant polyadenylated non-globin transcript. Other genes expressing abundant IR isoforms include SPTA1 (182nd in abundance with IR ≈ 0.27) and splicing factor SF3B1 (315th in abundance with IR ≈ 0.50). IR transcripts can thus be highly expressed in late erythroblasts, potentially acting as a major post-transcriptional control point to limit functional mRNA levels. Dynamic and stable IR events modulate expression of functionally distinct gene classes {#SEC3-3} -------------------------------------------------------------------------------------- Gene ontology (GO) analysis was performed to test whether different IR clusters with might be enriched for different biological functions. We found that the dynamically increased IR events in clusters C1 and C2 were greatly enriched for terms related to RNA processing (Table [1](#tbl1){ref-type="table"}). As shown in Table [2](#tbl2){ref-type="table"}, prominent among these were spliceosome-associated factors including U1 snRNP components U1--70K and U1A (encoded by SNRNP70 and SNRPA1), the U2 snRNP subunit SAP155 (encoded by SF3B1), the U2-associated factor UAP56 (DDX39B) and its paralog DDX39A. Other related genes with introns clustering in C1 or C2 include several hnRNP proteins (HNRNPA2B1, HNRNPD, HNRNPH1, HNRNPH3, HNRNPL), RBM proteins (RBM17, RBM3, RBM39, RBM6), SR proteins (SRSF6, SRSF10) and other proteins with important functions in RNA metabolism (FUS, ACIN1, EWSR, NXF1). Importantly, maturation-associated IR events shown in Table [2](#tbl2){ref-type="table"} were confirmed by inspection of gene-specific wiggle plots, a few of which are shown in Supplementary Figure S4. ###### Biological processes enriched in IR clusters GO terms for Clusters 1 and 2 Raw *P*-value Corrected *P*-value ------------------------------------------------------------------------------------------------- ------------------- ------------------------- GO:0008380∼RNA splicing 9.11E-17 1.67E-13 GO:0006397∼mRNA processing 2.84E-13 2.14E-10 GO:0016071∼mRNA metabolic process 4.72E-13 2.37E-10 GO:0000375∼RNA splicing, via transesterification reactions 4.48E-12 1.35E-09 GO:0000377∼RNA splicing, via transesterification reactions with bulged adenosine as nucleophile 3.88E-12 1.46E-09 GO:0000398∼nuclear mRNA splicing, via spliceosome 3.88E-12 1.46E-09 GO:0006396∼RNA processing 1.68E-11 4.23E-09 GO:0006350∼transcription 1.19E-05 0.003 GO:0003723∼RNA binding 7.98E-06 0.004 GO:0016044∼membrane organization 9.07E-05 0.017 GO:0016265∼death 1.45E-04 0.024 GO:0051168∼nuclear export 1.79E-04 0.024 GO:0045449∼regulation of transcription 1.76E-04 0.026 GO:0051276∼chromosome organization 2.43E-04 0.030 GO:0009057∼macromolecule catabolic process 2.95E-04 0.034 GO:0012502∼induction of programmed cell death 4.22E-04 0.037 GO:0008219∼cell death 3.54E-04 0.037 GO:0043122∼regulation of I-kappaB kinase/NF-kappaB cascade 4.09E-04 0.038 GO:0043122∼regulation of I-kappaB kinase/NF-kappaB cascade 4.09E-04 0.038 GO:0006917∼induction of apoptosis 4.09E-04 0.040 GO:0007049∼cell cycle 5.36E-04 0.044 GO:0008270∼zinc ion binding 0.02 0.63 GO:0043167∼ion binding 0.04 0.68 GO:0043169∼cation binding 0.12 0.85 GO:0046872∼metal ion binding 0.16 0.90 GO:0046914∼transition metal ion binding 0.21 0.94 **GO Terms for Clusters 4 and 5** **Raw *P*-value** **Corrected *P*-value** GO:0046914∼transition metal ion binding^a^ 1.63E-04 0.028 GO:0008270∼zinc ion binding 6.17E-04 0.036 GO:0046872∼metal ion binding 5.16E-04 0.036 GO:0019899∼enzyme binding 1.05E-04 0.036 GO:0043167∼ion binding 4.57E-04 0.039 GO:0045449∼regulation of transcription 3.78E-05 0.040 GO:0043169∼cation binding 4.04E-04 0.046 GO:0016563∼transcription activator activity 9.46E-04 0.046 GO:0008380∼RNA splicing 0.01 0.83 ^a^Includes the mitoferrins SLC25A37 and SLC25A28. ###### Transcripts for RNA binding proteins that exhibit increased IR in late erythroblasts Gene Intron coordinates retention (proE to orthoE)^1^ Cluster Comments --------- ---------------------------- ------------------------------- --------- ----------------------------------------------------- SNRNP70 chr19:49606845--49607890 0.12--0.32 1 U1 snRNP protein U1--70K SF3B1 chr2:198283676--198285151 0.09--0.30 2 U2 snRNP protein SAP155 chr2:198283313--198283520 0.23--0.52 1 DDX39B chr6:31499183--31500556 0.08--0.18 2 U2 snRNP-associated protein UAP56 chr6:31500689--31503143 0.20--0.41 1 SRRM2 chr16:2819285--2820352 0.12--0.48 1 DDX39A chr19:14520260--14520347 0.06--0.15 2 Paralog of UAP56 chr19:14520685--14521027 0.16--0.49 1 chr19:14521147--14521800 0.04--0.22 2 chr19:14521985--14522317 0.10--0.28 2 chr19:14523491--14523824 0.07--0.25 2 SNRPA1 chr15:101826007--101826418 0.12--0.39 1 U2 snRNP protein SRSF3 chr6:36568054--36568928 0.03--0.29 2 SR protein family SRSF5 chr14:70235614--70235898 0.04--0.25 2 SR protein family chr14:70235969--70237183 0.07--0.31 2 SRSF6 chr20:42088061--42088410 0.14--0.37 1 SR protein family SRSF7 chr2:38975796--38976039 0.30--0.60 SR protein family chr2:38976489--38976670 0.26--0.40 2 SRSF9 chr12:120901926--120903429 0.03--0.17 2 SR protein family SRSF10 chr1:24301566--24304400 0.08--0.33 2 SR protein family SRSF11 chr1:70694238--70697950 0.12--0.29 6 SR protein family HNRNPD chr4:83275308--83275907 0.13--0.34 1 hnRNP family HNRNPDL chr4:83346037--83346715 0.31--0.50 4 hnRNP family chr4:83346821--83347189 0.36--0.53 4 HNRNPH1 chr5:179043220--179043869 0.06--0.17 2 hnRNP family chr5:179048037--179048242 0.35--0.49 4 chr5:179048400--179048843 0.15--0.40 1 chr5:179048979--179050037 0.06--0.19 2 HNRNPL chr19:39331200--39334481 0.03--0.16 2 hnRNP family DDX5 chr17:62496892--62498127 0.11--0.26 6 RNA helicase chr17:62498342--62498556 0.10--0.24 6 RNA helicase RBM3 chrX:48434056--48434701 0.07--0.49 1 [R]{.ul}NA [b]{.ul}inding [m]{.ul}otif (RBM) family RBM6 chr3:50098981--50099394 0.13--0.42 1 RBM family RBM15 chr1:110884891--110888160 0.54--0.30 4 RBM family RBM17 chr10:6154325--6155470 0.05--0.16 2 RBM family RBM39 chr20:34326941--34327314 0.15--0.34 1 RBM family chr20:34327471--34328446 0.11--0.22 2 NXF1 chr11:62566048--62567848 \<0.01--0.16 NA Nuclear RNA export protein ACIN1 chr14:23537881--23538684 0.08--0.24 2 Exon junction complex FUS chr16:31196501--31198122 0.15--0.45 1 RNA and DNA binding protein chr16:31198158--31199645 0.15--0.43 1 chr16:31199679--31200443 0.06--0.36 2 TRA2A chr7:23561460--23561750 0.11--0.50 Alternative splicing regulator EIF4A2 chr3:186505672--186506913 0.11--0.22 Translation factor A very different picture emerged from GO analysis of stable clusters C4 and C5, which encompass 294 IR events in 248 genes, and which maintain moderate to high IR throughout terminal erythropoiesis. RNA processing functions were not significantly enriched in C4 and C5. Instead, GO terms related to metal ion binding were over-represented in these clusters (Table [1](#tbl1){ref-type="table"}). Genes in these functional categories include several that are important for erythroid iron homeostasis, such as the mitochondrial iron importers mitoferrin-1 (SLC25A37) and mitoferrin-2 (SLC25A28) and two heme biosynthetic enzymes (PPOX and HMBS). Notably, metal ion binding functions were not significantly enriched in clusters C1 and C2. We also observed high IR in several genes with functions related to cell division (Supplementary Figure S5), although they were not as consistently enriched in any one cluster. REEP4 helps clear the ER from metaphase chromatin, thereby ensuring correct progression through mitosis and proper nuclear envelope architecture ([@B26]); three REEP4 introns that were upregulated in late erythroblasts map to C1 and C2. MICALL2 is a paralog of MICALL1, a protein that influences microtubule dynamics during early and late mitosis ([@B27]); MICALL2 transcripts exhibit widespread high levels of retention in introns distributed among clusters C1, C2, C3 and C4. In transcripts for pericentrin, a key component of the centrosome that plays a role in centriole disengagement ([@B28]), two introns that map to C4 and C5 were moderately retained across all five erythroblast populations. Developmentally dynamic IR is not due to weak splice sites {#SEC3-4} ---------------------------------------------------------- Retained introns on average are flanked by weaker splice sites than constitutively spliced introns ([@B9],[@B15]). Here we explored this relationship in greater detail and asked whether differences in splice site strength might distinguish developmentally dynamic IR from developmentally stable IR. Cluster C3 was excluded from analysis due to its small size. For stable clusters C4--C9, IR was inversely related to both 5′ and 3′ splice site strength (Figure [3A](#F3){ref-type="fig"} and [B](#F3){ref-type="fig"}), implicating splice sites in determination of IR events. This relationship appeared mostly independent of differentiation status. For C1 introns at the early proE to basoE stages, a similar relationship between splice site strength and IR was observed. In contrast, however, C1 introns at the polyE and orthoE stages increased IR values far beyond the expected range. A similar but more modest effect was observed for C2 introns. Looked at another way, introns in C1 and C2 exhibited a much greater dynamic range in IR values during erythroblast differentiation than introns in C4 and C5 that had much weaker splice sites. Together these results support a model in which splice site strength, while correlated with base line IR values in proE cells, cannot explain dynamic increases in IR in late erythroblasts .Splice site strength is thus a determinant of IR for a large group of developmentally stable introns; however, dynamic clusters C1 and C2 represent a distinct class for which features other than splice site strength must be responsible for upregulating IR during terminal erythropoiesis. {#F3} IR is associated with PTC exons {#SEC3-5} ------------------------------- Most introns are spliced co-transcriptionally; however, introns that flank alternative exons are sometimes excised post-transcriptionally ([@B29],[@B30]). To investigate whether IR events in erythroblast transcripts might be associated with properties of the adjacent exons, we examined three sets of introns derived from our previous studies of erythroblast splicing ([@B2]): (i) introns flanking alternative 'coding' exons; (ii) introns flanking alternative 'PTC' exons that introduce premature termination codons (PTC); and (iii) a control set of introns, taken from the same gene sets, filtered to contain only introns flanked on both sides by constitutive exons. Figure [4A](#F4){ref-type="fig"} illustrates examples of each group. Introns of the GAPDH gene are all members of the constitutive class; EPB41 exon 16 is a well-studied alternative exon that is upregulated in late erythropoiesis and encodes a peptide functionally implicated in membrane stability ([@B7]); and SRSF6 exon 3 is a PTC exon that is upregulated in late erythroblasts ([@B2]) and is known to induce NMD ([@B31],[@B32]). IR values for these three transcripts were low for introns flanking constitutive exons and the alternative coding exon in EPB41, but much higher for the PTC-flanking introns of SRSF6. Extending the analysis to many additional introns yielded similar results (Figure [4B](#F4){ref-type="fig"}). Introns adjacent to constitutive exons (IR^avg^ = 0.025) or coding alternative exons (IR^avg^ = 0.025) exhibited low retention values, while introns flanking PTC exons were retained to a much higher extent (IR^avg^ = 0.21). ![IR characteristics of three intron classes. (**A**) RNA-seq read mapping data for Refseq annotated gene regions with no alternative splicing (GAPDH), alternative splicing of a coding exon (exon 16 in EPB41) and alternative splicing of a PTC exon (exon 3 in SRSF6). Boxed regions indicate the alternative exons of interest and the IR values of their flanking introns. (**B**) Summary of IR results for introns adjacent to alternative exons studied in ([@B2]), Figures [3](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"} and [6](#F6){ref-type="fig"}. Constitutive exons are from the same gene sets.](gkv1168fig4){#F4} The association between PTC exons and IR was evident in a number of transcripts for RNA processing factors. Figure [5A](#F5){ref-type="fig"} shows several examples of Ensembl-annotated PTC exons (boxed) that are flanked by retained introns, while other nearby introns generally exhibit much less retention. Included in this group are major spliceosome-associated factors, SR and hnRNP proteins frequently implicated in splicing regulation, and RBM-containing proteins. However, the PTC exon-IR link was not limited to RNA processing functions since it was also an important feature of the PIEZO1 gene encoding a mechanosensitive ion channel. {#F5} A number of prominent IR events in erythroblasts were not associated with annotated PTC exons, e.g. introns in transcripts for erythropoietin receptor (EPOR), α-spectrin (SPTA1), mitoferrin1 (SLC25A37) and membrane protein KEL. Careful analysis of such cases revealed that even when PTC exons are not evident, unproductive splicing events can be mediated via cryptic 5′ or 3′ splice site(s) (Figure [5B](#F5){ref-type="fig"}). For example, the cryptic sites in KEL transcripts represented novel splices that were not annotated in the UCSC genome browser but were supported by erythroblast RNA-seq reads and/or by RT-PCR analysis of erythroblast RNA. Even after careful analysis, however, there remained introns for which no cryptic splice sites could be identified. This latter group might represent a functionally different IR class, or there could be unannotated 'decoy' sites that are difficult to detect because they splice with very low efficiency. Transcripts with highly retained introns are localized to the nucleus and are resistant to nonsense-mediated decay {#SEC3-6} ------------------------------------------------------------------------------------------------------------------ Retained introns generally introduce PTCs that can induce NMD of cytoplasmic IR transcripts ([@B12]), but stable expression of NMD-resistant nuclear IR transcripts has also been reported ([@B9],[@B15]). We isolated nuclear and cytoplasmic fractions from cultured human erythroblasts and assayed relative abundance of IR transcripts using RT-PCR. Figure [6A](#F6){ref-type="fig"} shows that IR transcripts from six different highly expressed erythroblast genes were predominantly localized to the nuclear fraction. In each case small PCR products representing spliced RNAs were detected in both nuclear and cytoplasmic fractions, but the much larger IR bands were predominantly in the nuclear fraction. Interestingly, RNA-seq data from K562 erythroleukemia cells at the UCSC genome browser (ENCODE RNA-seq Tracks) supports the preferential nuclear localization of many retained introns; data for SPTA1 and SF3B1 are shown in Supplementary Figure S6. These results show that nuclear restriction of incompletely spliced transcripts is a post-transcriptional mechanism for limiting expression of translatable mRNAs in the cytoplasm. {#F6} Nuclear-localized IR transcripts are expected to be NMD-resistant. We tested this prediction by inhibiting NMD with cycloheximide plus emetine, a treatment that enhances steady state levels of PTC-containing erythroblast transcripts from genes such as SNRNP70 (ref. ([@B2]) and Figure [6B](#F6){ref-type="fig"}). For SLC25A28, in contrast, NMD inhibition actually decreased the amount of IR product relative to the fully spliced product. Similar results were obtained when nuclear and cytoplasmic fractions were examined separately (Supplementary Figure S7) and when other IR transcripts were assayed. Interestingly, this experiment also revealed novel isoforms of SLC25A28 (Figure [6B](#F6){ref-type="fig"}) and SPTA1 (not shown) that were characterized by inclusion of previously unannotated PTC exons. IR events upregulated in late stage erythroblasts are expressed at lower levels in other tissues {#SEC3-7} ------------------------------------------------------------------------------------------------ To explore whether prominent erythroblast IR events are unique to erythroblasts or are more broadly relevant in human biology, we examined IR in a variety of tissues. Retention of introns from each erythroblast IR cluster was assessed using KMA to analyze RNA-seq data from 16 other tissues in the llumina Human BodyMap 2.0 project. Many of the upregulated introns in clusters C1 and C2 also exhibited widespread retention in other tissues, sometime comparable to levels in late erythroblasts, but more often at lower levels. These data are shown in heat map format in Figure [7A](#F7){ref-type="fig"} for C1 and in Supplementary Figure S8 for C2; retention values are summarized in Supplementary Table S2). For example, the spliceosomal components SF3B1, DDX39B, SNRNP70 and RNPA1 all showed highest retention in orthoE but also widespread retention in other tissues (Figure [7A](#F7){ref-type="fig"}; for wiggle plots see Supplementary Figure S9). In some cases, e.g. SPTA1, IR is quite erythroid specific because the gene itself is not expressed in non-erythroid tissues. {#F7} In contrast, many C4 and C5 introns were retained at similar levels across erythroid differentiation and in non-erythroid tissues (Figure [7B](#F7){ref-type="fig"} and Supplementary Figure S7B), e.g. SLC25A37 in C4 and SLC25A28 in C5. Notable exceptions to this generalization included introns in HMBS, PPOX and PIGQ, which had high IR in erythroblasts but lower IR in non-erythroid tissues. These unique IR properties might be related to their much higher expression level in erythroblasts than in other tissues (data not shown). Together these results demonstrate that erythroblasts execute a complex intron retention program that shares many aspects with other tissues but ultimately is unique to terminal erythropoiesis. DISCUSSION {#SEC4} ========== Erythroblast differentiation is an excellent model system for studying the role of RNA processing in shaping the transcriptome during development. During the final four cell divisions that comprise terminal erythropoiesis, erythroblasts execute a highly dynamic, stage-specific RNA processing program that encompasses not only a robust network of alternative exon splicing ([@B2]), but also a broad array of intron retention events (this paper). Cluster analysis revealed many developmentally-dynamic IR events and splice site analysis supported active regulation of these events rather than a general decline in splicing activity in late erythroblasts. We hypothesize that IR is a mechanistically diverse, regulated set of processes with a complexity that could parallel alternative exon splicing and that it plays an important role in human erythroblasts as they undergo extensive remodeling prior to enucleation and maturation into red cells. Further studies will be needed to investigate the various pathways and RNA binding proteins that regulate IR in erythroblasts. Consistent with studies in other cells ([@B9]), IR in erythroblast transcripts is enriched in introns flanking alternative exons. Our results further indicate that much of the enrichment is associated with the subset of cassette exons that have PTCs, and that retained introns lacking PTC exons often exhibit other unproductive splicing events. Whether these unproductive sites are causally related to IR by promoting assembly into immature spliceosomal complexes ([@B33]--[@B36]) is unknown. Another possibility is that IR and unproductive splicing are both consequences of other features that inhibit use of the normal splice sites and allow for regulation by accessory factors. Interestingly, retention of introns flanking PTC exons in the SR protein genes was noticed previously and has been shown to be conserved ([@B9],[@B31]--[@B32]). More broadly, a pool of nuclear localized IR transcripts might have several alternative fates: degradation by nuclear RNA surveillance machinery ([@B13]), non-productive splicing of the retained intron followed by degradation in the cytoplasm via NMD ([@B12],[@B32],[@B37]--[@B38]) or completion of intron excision to produce mature translatable mRNA ([@B18]). These choices might offer a flexible post-transcriptional regulatory mechanism to modulate important erythroblast pathways such as iron homeostasis, e.g. by controlling output of the mitoferrin-1 gene. Future studies will be aimed at exploring these and other potential functions for IR during terminal erythropoiesis. What might be the function of regulated IR during normal erythropoiesis? One major function might be to post-transcriptionally modulate protein output from selected genes. This interpretation is consistent with the high expression levels of intron-retaining transcripts from SLC25A37, SPTA1 and SF3B1 genes that cannot encode full length proteins due to their localization to the nucleus and the presence of multiple PTCs in the long retained introns. Given the prevalence of IR in splicing factor genes---especially in several spliceosome-associated factors---one function of IR might be to down-modulate splicing capacity of the cells in a quantitative sense as they prepare for enucleation and express less RNA per cell in late erythroblasts (Figure [8](#F8){ref-type="fig"}). In fact, IR levels for SF3B1, SNRNP70 and DDX39B co-vary not only temporally during terminal erythropoiesis, but also more or less spatially across various tissues (see BodyMap data, Supplementary Figure S8). Another possible role for IR might be to act as a post-transcriptional regulator of iron homeostasis genes, e.g. by modulating the expression of selected heme biosynthetic enzymes (PPOX and HMBS) and iron transporters (SLC25A37 and SLC25A28) in response to the iron/heme demands of erythroblasts producing huge quantities of hemoglobin. Coordinate regulation of IR in splicing factor genes or iron homeostasis genes could thus be an important cellular control mechanism (Figure [7](#F7){ref-type="fig"}). {#F8} Finally, given the capability of IR to modulate gene expression, we speculate that inappropriate regulation of IR may be an under-appreciated cause of human disease. There is precedence for mutations that induce inappropriate intron retention, for example, in the case of BRCA2 ([@B39]), and aberrant intron retention events are prominent in breast cancer ([@B40]). Recently it was shown that ZRSR2-mutated MDS patients exhibit IR specifically for introns spliced by the minor spliceosome ([@B41]), and it would not be surprising if IR is a feature of other MDS subtypes. Although intron retention in the mitoferrin gene SLC25A37 was proposed to contribute to abnormal iron accumulation in MDS erythroblasts ([@B42]), the observation that considerable IR occurs normally in erythroblast SLC25A37 transcripts suggests that this should be interpreted with caution. Supplementary Material ====================== ###### SUPPLEMENTARY DATA J.G.C. and L.P. designed the research; H.P., M.P., S.L.G. performed research and analyzed data; and J.G.C., H.P., N.M., and L.P. wrote the article. SUPPLEMENTARY DATA {#SEC5} ================== [Supplementary Data](http://nar.oxfordjournals.org/lookup/suppl/doi:10.1093/nar/gkv1168/-/DC1) are available at NAR Online. FUNDING {#SEC6} ======= National Institutes of Health \[DK094699 to J.G.C., L.P.; DK32094 and DK26263 to N.M.\]; Director, Office of Science and Office of Biological & Environmental Research of the US Department of Energy \[DE-AC02-05CH1123\]. Funding for open access charge: National Institutes of Health. *Conflict of interest statement*. None declared.

Introduction {#Sec1} ============ The aim of rational drug design is to discover new drugs faster and cheaper. Much of the effort is put into improving docking and scoring methodologies. However, most techniques assume that the exact location of binding sites -- also referred to as *pockets* or *binding cavities* -- is known. Such pockets can be located both on a surface of a single protein (and be used to modulate its activity) or at protein-protein interaction (PPI) interfaces (and be used to disrupt the interaction). This task is very challenging and we lack a method that would predict binding sites with high accuracy -- most methods are able to detect only 30%--40% of pockets^[@CR1],[@CR2]^. Traditional approaches for binding cavity detection are typically geometry-based^[@CR3]--[@CR6]^, but there are also examples of tools using binding energy to different chemical probes^[@CR7],[@CR8]^, sequence conservation (template or evolutionary methods)^[@CR9]--[@CR11]^, or a combination of these^[@CR12],[@CR13]^. For example, ProBiS^[@CR11]^ -- similarity-based tool, uses local surface alignment with sub-residue precision, allowing to find sites with similar physicochemical properties to the templates stored in the database. Such methods simultaneously detect binding sites and provide some insight into their expected properties -- they are most probably similar to the templates they were matched to. Other approaches rely on a two-step algorithm, in which potential pockets are first identified and then scored to select the most probable binding sites. For example, Fpocket^[@CR5]^ is a geometry-based method, which first finds cavities in a protein's structure and then scores them. The reverse approach is used in P2RANK^[@CR14]^, which uses a random forest (RF) model to predict \"ligandibility" score for each point on a protein's surface, to then cluster points with high scores. The latter tool is an example of applying machine learning (ML) to detect pockets -- supervised ML to score surface points and unsupervised ML to post-process these predictions. Unsupervised ML models are trained on unlabeled observations and aim to find patterns in the data in order to simplify their representation, remove the noise, and get better understanding of their nature. Supervised ML models, on the other hand, require the observations to be paired with their corresponding labels (expected output, class, etc.). The main purpose of this class of models is finding the relationship between the data and the labels that are actually desired. The data is relatively readily available (in case of P2RANK -- the structure of a protein) but the desired information is typically much harder to acquire (e.g. location of binding sites). Another axis of classification of ML models is based on their complexity, or *depth*. Deep learning (DL) is a branch of ML grouping more complex models of different types, both supervised and unsupervised. There is no clear border between \"classical" ML and DL, but in general deep models are complex, multilayer neural networks capable of finding more sophisticated and convoluted relations between input and labels than their shallow counterparts. DL models require less manual work and feature engineering, and use model's internal layers to extract features from the unprocessed data. In the context of bio- and cheminformatics DL allows to predict *in silico* properties that require much effort to establish experimentally, like detecting functional motives in sequences^[@CR15]^ or assessing binding affinity for protein-ligand complexes^[@CR16],[@CR17]^. A recent example of a deep model used for binding site detection is DeepSite^[@CR2]^. Similarly to P2RANK, DeepSite classifies each point in a 3D space based on its local environment as belonging (or not) to a binding pocket. Probabilities for all points form a 3D density, that can then be post-processed to get the most probable locations and shapes of binding sites present in the structure. Unlike P2RANK however, DeepSite uses a deep 3D convolutional neural network, with an architecture typical for image classification problems -- set of convolutional layers paired with max pooling layers, followed by fully connected layers and a final neuron with the predicted class. Such an architecture allows for the extraction of features -- 3D structural patterns -- that are immediately used by the model to make predictions. DeepSite was proven superior to two other state-of-the-art approaches at the time: Fpocket^[@CR5]^ and Concavity^[@CR9]^. The first one is a geometry-based, whereas the second -- a sequence-conservation-based method. But although DeepSite achieved better results, only approximately 50% of the predicted pockets is at most 4 Å from the actual position of the binding site. This is not an acceptable standard and calls for an improvement. In this work we present a different DL-based approach for finding binding pockets, inspired by semantic image segmentation instead of classification. Image segmentation aims at locating an object, or multiple objects, in an image. Output of such a model is a set of scores assigned to each pixel, where the score denotes the probability that a given pixel belongs to the desired object. In our case, the input is a 3D structure of a protein represented with a grid that can be analyzed in the same manner as 3D images, whereas the desired object is the binding pocket. Our model called Kalasanty is based on U-Net^[@CR18]^ -- a state of the art model for image segmentation. We adapted this model to the problem of binding cavity detection, and added functionalities that allow to easily generate predictions for protein structures. The model takes protein structure as input, automatically converts it to a 3D grid with features, and outputs probability density -- each point in the 3D space has assigned probability of being a part of a pocket. Predictions can then be saved as .cmap or .cube files, that can be later analyzed in molecular modeling software. Kalasanty can also output parts of the protein that form pockets and save them as .mol2 files. Methods {#Sec2} ======= Our approach {#Sec3} ------------ In order to solve any problem with DL, it first needs to be specified in terms of how input and output are represented. In this work we formulate the pocket detection task as a 3D image segmentation problem. This allowed us to use well established DL methods, originally developed for 2D images. With this approach both input and output are represented as 3D grids with the exact same dimensions -- each grid point in the input has a corresponding point in the output. The input is a discretized protein structure with multiple feature channels describing atomic properties. The returned output has a single channel with probability of belonging to the pocket (details in section "Data"). Known pockets, which are used for training and evaluation, are represented with binary grids, where 1s denote grid points that belong to the pocket and 0s otherwise. The model is based on an architecture called U-Net (see Fig. [1](#Fig1){ref-type="fig"}), which is a fully-convolutional, encoder-decoder model that pioneered skip connections (see text below). This kind of architecture prevents from loosing fine-grained information about the input which greatly increases the precision of the resulting segmentations.Figure 1Kalasanty's architecture. Kalasanty was built using the same ideas and architecture design as the original U-Net. Similarly to the original network, it is composed of blocks consisting of two convolutional layers and a single max-pooling or up-sampling layer (depending on the side of the "U"). However, Kalasanty works on 3D data instead of 2D images, therefore 3D versions of convolutional, pooling and up-sampling layers were used. Also, the number of layers, number of convolutional filters, and patch sizes were adjusted to match the size of the input and difficulty of the task. Kalasanty has 9 convolutional blocks -- 4 in the encoder, one in the bottleneck, and 4 in the decoder part of the network. Each block consists of two convolutional layers with the same number of filters (32, 64, 128, 256, or 512), kernel size of 3×3×3 pixels and ReLU activation function, combined either with a max-pooling layer (encoder path, left side of the Fig. [1](#Fig1){ref-type="fig"}) or with an up-sampling layer (decoder path, right side of the Fig. [1](#Fig1){ref-type="fig"}). The two first max-pooling layers and the two last up-sampling layers have 2×2×2 patch sizes, while layers in the middle have 3x3x3 patch sizes. This way, for input of 36×36×36 pixels that was used in this work, feature maps in the middle of the network (bottleneck, bottom of the Fig. [1](#Fig1){ref-type="fig"}) have spacial sizes of 1×1×1 and can be used as feature vectors. What is unique about U-Net-like models when compared with other encoder-decoder networks it that the information between the two paths is not only passed through a bottleneck, but also after each block using so-called skip connections. The final feature map from each block in the encoder is copied and concatenated with the first feature map in the corresponding decoder block (orange dashed lines). This allows to better localize features and therefore return more accurate segmentations. The model was defined with the Keras library^[@CR19]^. Apart from methods needed for ML-related tasks, we implemented custom methods for working with the molecular data: making predictions for molecules; locating amino acids forming the pockets; saving the predicted probabilities as .cmap or .cube files; and saving parts of proteins forming pockets as .mol2 files. Source code and network's parameters are freely available at <http://gitlab.com/cheminfIBB/kalasanty>. Models were trained with L2 regularization on each layer's weights ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda =1{0}^{-5}$$\end{document}$). Also, random translations and rotations were used during training to augment the dataset. As the objective function we used the negative Dice coefficient for continuous variables: $$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C(y,t)=\frac{-2{\sum }_{i,j,k}({y}_{i,j,k}\cdot {t}_{i,j,k})+\varepsilon }{{\sum }_{i,j,k}({y}_{i,j,k}+{t}_{i,j,k})+\varepsilon }$$\end{document}$$ where $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$y$$\end{document}$ and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t$$\end{document}$ are the predicted and target segmentations, $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(i,j,k)$$\end{document}$ are indices of a grid cell, and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon $$\end{document}$ is a smoothing factor ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varepsilon =0.01$$\end{document}$ was used). Minimization was performed for $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.5\cdot 1{0}^{6}$$\end{document}$ steps with 10 samples in each batch, and the Adam optimizer with a learning rate of $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1{0}^{-6}$$\end{document}$ and with default values for the remaining parameters^[@CR20]^. Data {#Sec4} ---- For training and validation of the model the sc-PDB^[@CR21]^ dataset was used. The database consists of known binding sites, accompanied with prepared protein structures. Binding sites are represented both with 3D shapes of cavities generated with VolSite^[@CR22]^ (which were used in this study) and amino-acids that form them. VolSite describes a binding cavity with a set of pharmacophoric properties arranged on a 3D grid, based on the properties of the neighboring protein atoms. Data is stored as mol2 file, with atoms encoding each property. In this project, we converted data in the grid to a binary information -- whether a point in space is part of a pocket or not. Grid cell (see section "Our approach") with such point inside was considered a part of the pocket. The database (v 2017) consists of 17594 binding sites, corresponding to 16612 PDB structures and 5540 UniProt IDs. The test set was constructed from a different dataset, one used for benchmarking by Chen *et al*.^[@CR1]^ (see below). When assessing the quality of a ML model it is important to evaluate it using a separate dataset. Using a different dataset minimizes the risk that there were some database-related artifacts, that might have been exploited by the model. However, it is easy to make a mistake and have the same protein (with slightly different structures) in the training and the test set as well. This is a common error called *data leakage* which leads to overly optimistic assessment of a model. In order to avoid data leakage, all structures of proteins from the benchmark were removed from the sc-PDB database (481 structures). Also, 304 binding sites were discarded because of errors when loading their corresponding protein structures with Open Babel. Finally, 15860 structures, corresponding to 5473 UniProt entries, were used for training. Number of structures per protein varied from 1 to 280, with median equal to 1 and mean equal to 3.26. The dataset contained protein structures originating from 952 different organisms, from which the most abundant were human (34.4%), *E. coli* (5.6%), Human immunodeficiency virus (4.2%), rat (2.9%), and mouse (2.4%). Also, diverse protein architectures were well represented, with 5171 structures of mainly alpha proteins, 2500 of mainly beta proteins, 11758 of alpha-beta proteins and 53 structures of proteins with few secondary structures^[@CR23]^. The dataset was also diverse from a sequence perspective and contained proteins from 1983 different Pfam families and 982 superfamilies, protein kinases being the most frequent. The data from sc-PDB were split into 10 folds (subgroups containing 1586 structures each) based on UniProt ID, i.e. all structures of a single protein must be in the same fold. This setup was necessary to avoid data leakage during validation. Putting different structures of the same protein in different folds would result in having almost identical training and validation examples, which might result in unnoticed overfitting, and as a consequence overoptimistic evaluation and possibly selection of incorrect hyperparameters. We also analysed binding site similarity across obtained folds (using Shaper^[@CR22]^ results provided by sc-PDB) to assure that there is no data leakage during validation. This setup was used to train the models with 10-fold cross-validation (CV). CV results were used to select model and optimization parameters and assess models' stability. The final model was trained on all 10 folds combined to achieve the best possible performance. As mentioned, for the test set we used structures from the Chen benchmark^[@CR1]^. This benchmark set contains apo and holo structures for 104 proteins (208 in total). Structures were converted to the format used in sc-PDB to evaluate the models using the following steps. First, for each structure ligand(s) and protein were split into separate files using UCSF Chimera^[@CR24]^. Solvent and ions were assigned to the protein. Then, we used VolSite^[@CR22]^ (available in IChem toolkit) to describe a cavity for each ligand. In case of apo structures, they were aligned to their holo counterparts, and then the ligands were used to select pockets. For 59 structures (31 apo and 28 holo) out of the 208 present in the benchmark, VolSite failed to find a cavity because of its insufficient buriedness. Still, the remaining part of the dataset (149 structures with 269 binding sites) offers a valuable test set because it contains diverse proteins not used for training. We used the final 149 structures with 269 binding sites to assess the performance of our model and the performance of the DeepSite model. Note, that we have no control over the dataset that was used by^[@CR2]^ and 12 proteins from the test set have been used to train DeepSite. This might result in a slightly more optimistic evaluation for DeepSite, but we have decided to keep those structures so that the test set was larger. It is important to note, that the results presented in this work cannot be directly compared to the ones presented by Chen *et al*.^[@CR1]^ because only a subset of the original dataset was used. Finally, all the resulting protein structures and segmentations were represented with 3D grids with 2 Å resolution. The grids were centered on a protein center and had 70 Å in each direction. Proteins were described with 18 atomic features used in our previous project^[@CR17]^. Pockets were transformed to 3D binary masks with same size, center, and resolutions as grids representing their corresponding protein structures. For 79 (0.5%) entries in the sc-PDB database this procedure lead to empty pocket grids. After manually inspecting several of such cases it turned out that it affects large protein complexes and less carefully prepared protein structures (e.g. 1zis, which contains two unbound protein chains, located far away from each other). Although such structures may arise in high-throughput experiments and analyses (like the one presented in this study), they are highly unlikely to occur in real-life studies aiming to discover binding cavities in a protein of interest. Also, this issue affects a neglectable fraction of the training data and none of the test examples. We therefore decided not to modify the structures, nor the procedure for data preparation. The data with empty pocket grids were used for training as negative examples, and skipped in the validation. Results evaluation {#Sec5} ------------------ Results obtained with Kalasanty and DeepSite were evaluated using two popular metrics: $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ (discretized volume overlap). $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ is the distance between the predicted and the actual center of the pocket. It is typically used to describe a success rate for the method, i.e. the fraction of sites below the given $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ threshold. Similarly to other authors^[@CR1],[@CR2],[@CR4]^, we analyzed success rates for thresholds up to 20 Å and considered pockets with $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ below 4 Å as correctly located. $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$, on the other hand, is a more strict metric comparing shapes of the predicted and actual pockets. It is the volume of the intersection of the predicted and the actual segmentations, divided by volume of their union. The two metrics complement one another, highlighting different aspects of prediction quality -- correct location ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$) and shape ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$) of predicted pockets. When used together, they provide concise yet rich description of the results, and allow to analyze them faster and more objectively than with visual investigation. In order to calculate both metrics, predicted densities were converted to binary segmentations. For Kalasanty, probability threshold of 0.5 was used, which was selected based on models' performance on the validation set. However, we note that using thresholds between 0.3 and 0.8 leads to very similar results. DeepSite predictions were obtained using the playmolecule.org web-service. Predicted binding cavities were download as .cube files and converted to binary masks using the probability threshold of 0.4, which was recommended by DeepSite's authors. Following the work of Chen *et al*.^[@CR1]^, for each structure only $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n$$\end{document}$ predicted pockets with the highest scores were considered, where $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n$$\end{document}$ is the number of pockets present in the structure. Then, each predicted pocket was matched with the closest real pocket and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ values were calculated. If no pocket was predicted, we used the worst possible values for the metrics, which were $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO=0$$\end{document}$ (no overlap) and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC\ =\ 70\sqrt{3}$$\end{document}$ Å = 121.24 Å (the biggest possible distance for a 70 Å cube). Additionally we used F1 score -- metric typically used to evaluate ML models for detection tasks. F1 score combines precision (positive predictive value) and recall (also called sensitivity) -- it is their harmonic mean. Conversely to other popular metrics used in ML, like accuracy or ROC AUC, F1 do not require notion of true negative, which is undefined for detection problems. We used $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ of 4 Å as a threshold for true positives when calculating these metrics. Results {#Sec6} ======= In this study we present two sets of results. In the first part, we describe cross-validation (CV) experiments, in which we tested Kalasanty's stability and its general properties, using nearly 16k structures from the sc-PDB database. In the second part, we show results for the model trained on the whole training set and evaluated on the test set. We also compare it to another DL-based approach -- DeepSite. For CV experiments, the training set was divided into 10 parts. Then, 10 models were trained with one fold left out for validation. This way we were able to evaluate our approach on the whole dataset, without making predictions for structures that were used to train a particular model. Also, the validation sets were used to monitor the training and to select optimal parameters defining the network architecture and optimization procedure. For each model we observed a plateau on the learning curve for the validation set. Results were stable across folds and we observed similar distributions of $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ values for each fold (see Fig. [2](#Fig2){ref-type="fig"}). The variation for $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ is so small, that 95% confidence interval for mean estimation almost melts into the mean curve (Fig. [2A](#Fig2){ref-type="fig"}). The model did not return any pockets for 4.7% of the structures (737 pockets, ranging from 31 in fold 6 to 113 in fold 2). F1 score was equal to 0.69, with recall of 0.66 and precision of 0.73.Figure 2Models' performances on the validation sets. (**A**) Success rate plot for different $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ thresholds, averaged over 10 folds. Blue area around the curve depicts 95% confidence interval for the mean estimation. 4 Å threshold is marked with gray dotted line. (**B**) $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ distribution for correctly located pockets ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC < 4$$\end{document}$ Å) in all folds combined. We have also looked for global trends in prediction accuracy and factors, that should be irrelevant for the model in order for it to generalize well. We had an unbalanced dataset and the model might have been biased in favor of the most prevalent types of proteins. Although the performance differs between different groups of proteins (see the "Discussion"), we did not observe any systemic differences between results for the most frequent types of proteins and average results. This also suggests that the model did not overfit to the training set, and that it generalizes well to new, unseen structures. Next, we used the best set of parameters and trained the final model on the whole sc-PDB dataset (15860 structures). Finally, we evaluated the performance of this model and DeepSite's on the test set. This was a more challenging dataset, containing proteins from a different source. As expected, the performance was slightly worse than the one observed in CV experiments yet still promising (see Fig. [3](#Fig3){ref-type="fig"}, panels A and C). For only 3 structures (1.3%) no pockets were detected and for 120 of the binding sites (44.6%) center of the predicted pocket was at most 4 Å from the center of the real binding site. However, 5% of correctly located pockets (compared to 1.5% in the validation set) had incorrectly predicted shape, resulting in $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ below 0.25. The F1 score was equal to 0.45 (precision=0.64, recall=0.35).Figure 3Models' performance on the test set. (**A**,**B**) Success rate plot for different $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ thresholds for Kalasanty and DeepSite, respectively. 4 Å threshold is marked with gray dotted line. (**C**,**D**) $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ distribution for correctly located pockets ($\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC < 4$$\end{document}$ Å) for Kalasanty and DeepSite, respectively. DeepSite performance on the test set was also worse than CV results reported in^[@CR2]^ (see Fig. [3](#Fig3){ref-type="fig"}, panels B and D). When threshold of 0.4 (recommended by the authors) was used, pockets were detected in all structures, but only 64 (23.8%) of them had $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ below 4 Å. What is more, only 2 pockets in the entire test set had $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ above 0.5. It is worth noting that similarly to DeepSite's authors we did not observe significant changes in the results when different thresholds between 0.3 and 0.6 were used. The overall F1 score was equal to 0.26 with precision of 0.36 and recall of 0.20. Discussion {#Sec7} ========== In order to better understand the difference between Kalasanty and DeepSite, we analyzed if the two models make similar mistakes (see Fig. [4](#Fig4){ref-type="fig"}). Although the general trends are similar and same proteins were problematic for the two models, Kalasanty correctly detected almost twice as many pockets as DeepSite (44.6% vs 23.8%, respectively). Also, for 84.6% (115 out of 136) of pockets detected by at least one of the models, Kalasanty had lower $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ than DeepSite. This difference in performance is probably caused by the fact, that DeepSite tends to return more voluminous predictions (Fig. [5](#Fig5){ref-type="fig"}). Although there is no single definition of a binding site (should it be a group of amino-acids, or a void between them?), the two models were trained on the same dataset so this comparison is justified. After converting densities into binary predictions, pockets returned by DeepSite are on average twice as big as those predicted by our model. It also explains the discrepancy between $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ and $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DVO$$\end{document}$ results for DeepSite -- even correctly located pockets are usually too big and their shape is not modeled accurately.Figure 4Comparison between $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ values achieved by DeepSite and our model: (**A**) on the whole test set and (**B**) zoomed-in on high-quality predictions. Each point corresponds to $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ values obtained with the two models. Binding sites for which U-Net did not return predictions are depicted with red crosses at the top of the image. Structures that were used to train DeepSite are colored with orange. Gray area marks 5 Å difference in both directions from the diagonal (same predictions for the two models). Predictions for L-histidinol dehydrogenase and 7,8-dihydroneopterin aldolase are marked with ellipses and annotated with proteins' short names (HDH and DHNA, respectively).Figure 5Binding site probability density for cAMP-specific 3′,5′-cyclic phosphodiesterase 4D (UniProt ID: Q08499, PDB ID: 1tb7) predicted with (**A**) Kalasanty and (**B**) DeepSite. VolSite representation of the binding site (ground truth) is depicted with points, while prediction with a mesh. Interestingly, for two proteins -- L-histidinol dehydrogenase (UniProt ID: P06988, PDB IDs: 1k75 and 1kae) and 7,8-dihydroneopterin aldolase (UniProt ID: P56740, PDB IDs: 1dhn and 2nm2) -- DeepSite correctly located all pockets in both holo and apo structures, while Kalasanty completely missed them (see Fig. [4A](#Fig4){ref-type="fig"}). We investigated what is the root of the observed differences in those two specific cases. We were interested whether poor results for the two proteins are a part of a bigger trend and can be explained by our model's inability to predict pockets for a particular group of proteins. Unfortunately such analysis is hampered by the test set size which is too small to perform statistical analysis. We therefore analyzed relationships between protein properties (superfamily, fold, source organism, length and size of the binding site) and prediction quality using CV results. From these experiments we indeed can observe higher $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$DCC$$\end{document}$ values for both superfamilies when compared to the rest of the dataset (Supplementary Fig. [S1](#MOESM1){ref-type="media"}, panels A and B). We also observed significant differences for both source organisms (Supplementary Fig. [S1](#MOESM1){ref-type="media"}, panels C and D). However, we cannot determine if those are causal relationships, or is there some other underlying factor, correlated with these two properties. To summarize, we hypothesize that poor results for the two proteins might be related to some systematic errors in predictions made for such types of proteins. Conclusions {#Sec8} =========== In this work we presented Kalasanty -- a neural network model for detecting binding cavities on protein surfaces. Kalasanty was trained and validated with the sc-PDB database and additionally evaluated on an independent test set. We compared Kalasanty with DeepSite which was proven better than Fpocket and Concavity -- one of the best conventional methods for binding site prediction. Results show that our model achieves high accuracy and is able to locate pockets more precisely than DeepSite (44.6% and 23.8% correctly located pockets from the test set, respectively). What is also important, Kalasanty is stable and cross-validation results are comparable to those obtained for new data (not used for training nor validation). However, it should be noted that the sc-PDB dataset contains only deep cavities, which tend to have better properties (druggability). Model trained on such a dataset is not able to detect binding sites that are located on flat surfaces. To obtain such a model, different datasets should be acquired which is out of the scope of this study. Kalasanty is able to find multiple binding sites for a single protein. However, if they are closely located the current post-processing procedure might merge them into one pocket. A possible extension of this approach would be to look for local probability maxima around which candidate binding sites would be constructed. Kalasanty was implemented in Python and the architecture was defined using the Keras library. Source code, together with trained model and helper scripts are freely available at <http://gitlab.com/cheminfIBB/kalasanty>. The repository can be also used to launch online demo, allowing to test Kalasanty through the web browser on a molecule of interest, without the necessity of installation nor registration. The network was supplemented with additional methods that allow for making predictions directly for molecules, and handle all necessary preprocessing under the hood. Predictions can then be saved as .cmap or .cube file and visualized in molecular modelling software. Although Kalasanty is a deep neural network, using it does not require GPU. GPU is crucial for training, but not for inference. It takes 5 seconds to load the model and a second to make a prediction on a Intel Core i7 CPU. This makes Kalasanty accessible for all researchers. Kalasanty is based on U-Net -- a state-of-the-art neural network architecture for semantic segmentation, originally developed for 2D medical images. We adapted the U-Net to process 3D protein structures and provided the model with input relevant for the task of identifying binding cavities. Deep learning methods gained popularity in the recent years because of their flexibility and potential for capturing complex relationships hidden in the data. The field of deep learning is ripe with noteworthy ideas that have already been tested in disciplines such as computer vision and sequence modelling. Therefore, this work can also be seen as an example of adapting deep learning methods developed in other fields to structural bioinformatics. Supplementary information ========================= {#Sec9} Supplementary Information. **Publisher's note** Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information ========================= is available for this paper at 10.1038/s41598-020-61860-z. The authors thank Maciej Dziubiński and Maciej Wójcikowski for their help in revising the manuscript. M.M.S.-D. and P.S. carried out the design and drafted the manuscript, M.M.S.-D. built Kalasanty and analyzed the results, and P.Z. revised the manuscript. Source code of Kalasanty is available at <http://gitlab.com/cheminfIBB/kalasanty>. The authors declare no competing interests.

Why are cows sacred? Travel anywhere in India and they have the right of way. Travel anywhere in the eastern United States and you\'ll see squirrels, more than likely, as roadkill. Yet both species serve a similar epidemiological function: they receive bites from infected vectors that might otherwise have bitten humans, and they break the chain of pathogen transmission. In the case of Indian cattle, the bites are from mosquitoes infected with malaria parasites; the squirrels, on the other hand, receive their bites from ticks infected with the spirochetes that cause Lyme disease. In both cases, the presence of a relatively inefficient host species has reduced the rate of infectious disease spread into the human host population. Recently, ecologists have uncovered several other ways in which species diversity can benefit human health. In this Essay, we describe how disease risk is influenced by biological diversity and, specifically, how some host species act to reduce the risk of transmission of virulent zoonotic pathogens to people. This represents an exciting area of study where ecologists, conservation planners, and physicians can work together to reduce disease risk and maintain biological diversity. In a world where climate change may allow vector-transmitted diseases to spread from the tropics into the temperate zone, it may be sensible to conserve biological diversity for the purely selfish reasons of protecting human health. Zooprophylaxis {#s2} ============== One of the oldest examples of biological diversity reducing disease risk occurs with malaria and domestic livestock in India, and it may partly explain why cows are regarded with deep reverence by Hindus. A variety of historical papers have suggested that sleeping in close proximity to domestic livestock, particularly cattle, may reduce the rate at which mosquitoes bite humans, and thus reduce the likelihood of infection with malaria or other vector-borne pathogens (zooprophylaxis; reviewed in \[ [@pmed-0030231-b1], [@pmed-0030231-b2]\]). We don\'t specifically know if this is why cows are sacred in parts of India (see [Figure 1](#pmed-0030231-g001){ref-type="fig"}); they have been considered sacred since the Aryans invaded in the 2nd century, B.C., but many cultural and religious taboos reflect cultural selection for activities that minimize or reduce disease risk. Certainly tribes who spent time in close proximity to cattle might have reduced their risk of malaria, particularly in regions where malaria was transmitted by Anopheles culicifacies (a cattle-biting specialist). Active zooprophylaxis was undertaken when cattle were deliberately used as a barrier between mosquito breeding sites and human settlements; it was probably most widely used in Soviet collective agriculture \[ [@pmed-0030231-b3]\] and is again being used in Tanzania today. {#pmed-0030231-g001} A slight problem with this hypothesis is that in many dry areas where malaria exhibits seasonal patterns of abundance, the by-products of cattle supply vital sources of moisture and nutriments that can contribute to the breeding success of mosquitoes. In other words, cattle may divert bites in the short term but increase mosquito abundance in the long term \[ [@pmed-0030231-b4]\]. Determining the net effects of species diversity on disease risk in other types of disease systems remains a challenging frontier. Loss of Biological Diversity and Disease Risk {#s3} ============================================= A variety of human processes contribute to the loss of biological diversity: habitat loss, habitat fragmentation, overexploitation of populations for food or other economic uses, the introduction of invasive species and diseases, climate change, and pollution \[ [@pmed-0030231-b5]\]. Habitat loss is the dominant cause; it is cited as the key cause in around 70% of the species listed as threatened or endangered by the Red List, which records global totals of imperiled species ( <http://www.redlist.org>). Habitat loss is predominantly driven by the conversion of forests and savannas into agricultural land, cities, and industrial sites. Species are lost by the interaction of two processes: the loss of habitat as conversion proceeds and the fragmentation of the remaining habitat into smaller subdivided patches. Since different species require a minimum area of habitat to meet their energetic and social needs, fragmentation creates many small populations---each of which is highly vulnerable to extinction even when quite large total areas of habitat remain. This creates the central dilemma of conservation biology: species are constantly going extinct locally, but usually only receive major attention when the remaining few individuals are threatened with total annihilation. Species with larger area requirements tend to be lost first in response to habitat fragmentation (and overexploitation). In the smallest patches, only small species or those with superior dispersal abilities persist. If the predators and competitors that determine the abundance of prey species have disappeared from the smaller patches, then the numbers of prey individuals will increase. If these prey are reservoirs for zoonotic pathogens, then the abundance of these pathogens will also increase. The classic example of this effect occurs with Lyme disease in the forests of the northeastern US---as discussed below, risk of Lyme disease is high in small patches of forest with poor species diversity. Squirrels and Lyme disease {#s4} ========================== Lyme disease is caused by a spirochete bacterium, Borrelia burgdorferi, which is transmitted through the bite of the blacklegged tick ( Ixodes scapularis) in eastern North America. These ticks feed on a wide variety of vertebrate species, including humans. Each of these host species has a different probability of infecting the ticks with the Lyme bacterium. The white-footed mouse ( Peromyscus leucopus) has the dubious distinction of being the most competent reservoir species for the bacterium---over 90% of ticks feeding on wild mice become infected with the Lyme bacterium ( [Figure 2](#pmed-0030231-g002){ref-type="fig"}). In contrast, fewer than 15% of ticks feeding on gray squirrels ( Sciurus carolinensis) become infected, even though virtually all of the squirrels carry the bacterium \[ [@pmed-0030231-b6]\]. As a consequence, the Lyme bacterium is much more prevalent in habitats with many mice than in habitats that harbor a diversity of other species \[ [@pmed-0030231-b6]\]. {#pmed-0030231-g002} Where are white-footed mice most common? Several studies have shown that very small patches of forest (less than about two hectares) contain high densities of white-footed mice. These patches are too small to support the predators and competitors that typically determine mouse numbers ( [Figure 3](#pmed-0030231-g003){ref-type="fig"}). So in small patches of forest, ticks have almost nothing to feed on except white-footed mice, and there is an overabundance of these. Small fragments of forest boast some of the highest Lyme disease risk ever documented \[ [@pmed-0030231-b7]\]. In contrast, bigger patches of forest harbor squirrels, chipmunks, foxes, weasels, and coyotes---these are all poor reservoir species for Lyme disease and also reduce white-footed mouse abundance. Here, high species diversity appears to both regulate the most competent Lyme disease reservoir (mice) and deflect tick meals away from mice and toward less competent reservoirs. {#pmed-0030231-g003} Yellow-Necked Mice and Tick-Borne Encephalitis {#s5} ============================================== Tick-borne encephalitis (TBE) is a viral infection that circulates among free-living yellow-necked mice ( Apodemus flavicollis) of Europe and the former Soviet Union. When an infectious tick bites a human, the pathogen spills over, causing serious illness. The virus attacks the human central nervous system, causing meningitis and encephalitis. Woodland workers are often at risk, but some of the "hot spots" for TBE transmission are also key holiday locations where children have become infected and died. The intriguing part of the transmission process is that the wild zoonotic hosts---the yellow-necked mice---do not become viremic, but permit transmission of the virus between co-feeding ticks through a nonviremic process. Successful transmission requires that an infectious nymph bites a mouse at the same time as a susceptible larva is feeding \[ [@pmed-0030231-b8]\]. When the density of mice is high, then the probability of two ticks feeding on the same host at the same time is very small, so transmission declines to levels where the pathogen cannot persist. In contrast, when the mice are at very low density, not enough infectious ticks are produced for the disease to persist. So the density of mice acts as a major determinant of transmission; this only allows TBE to persist when mouse density is moderate. Concomitantly, a second determinant of transmission success is mediated by the presence of noncompetent hosts in the system. Adult ticks don\'t bite mice, so the presence of large mammals (usually deer) is necessary to sustain the tick population, but deer do not permit successful transmission of TBE \[ [@pmed-0030231-b9]\]. Once again, the relative density of these hosts is important in determining the level of disease risk to humans. When deer density is high, disease levels are low because a high proportion of ticks are feeding on deer, and virus is lost from the system. When we combine these two dimensions of dilution, then high biodiversity and high densities of hosts are good for human health. House Sparrows and West Nile Virus {#s6} ================================== Similar ecological forces seem to operate with West Nile virus (WNV). This disease first appeared in the US in New York in 1999 when significant numbers of birds, particularly crows, began dropping dead in and around New York City. Quite soon afterward, the first human cases were reported, and several of these patients died \[ [@pmed-0030231-b10]\]. In the next five years, WNV spread to cover almost the entire US and parts of Canada, Mexico, and Central America. The virus is transmitted by a diversity of mosquito species and can replicate in a variety of bird species. Although in some host species the pathology is undetectable, in others, particularly crows, a rapid viremia leads to death in only a couple of days. Humans, horses, and alligators are probably dead-end hosts, meaning that viremia is either too modest or too transient to provide a source of infection for later-feeding mosquitoes. Teasing out whether species diversity affects WNV dynamics will be a thorny problem, as a large number of host and vector species are involved. Several authors have presented initial analyses that suggest there is a decline in prevalence of human WNV cases in areas with high avian diversity \[ [@pmed-0030231-b11]\]. WNV seems to pose an analogous situation to Lyme disease in that the most competent reservoirs---house sparrows, house finches, American robins, blue jays, and grackles---proliferate in heavily fragmented or otherwise degraded habitats. Consequently, where bird diversity is low, the bird community consists largely of competent reservoirs, but where natural habitats are largely intact and bird diversity is higher, many incompetent reservoirs dilute and disrupt the transmission cycle of the virus. Although a protective role for high biological diversity has been supported for directly transmitted pathogens such as hantaviruses, we expect that species diversity is most likely to reduce disease risk with vector-transmitted infections. The primary reason we get disease reductions is that the vectors that transmit the pathogen only take a limited number of bites in their lifetime; when some of these bites are taken from hosts that are not competent to amplify the pathogen, these bites are wasted. This reduces the rate at which the pathogen is transmitted. Ecologists have termed this phenomenon, the "dilution effect" \[ [@pmed-0030231-b12]\]. The primary process that produces a dilution effect is the increased diversity of host species that increase the proportion of bites that are wasted. From one crucial perspective, these transmission dilution effects may be even stronger for WNV than for Lyme disease or TBE. In the case of ticks that transmit Lyme disease and TBE, the prolonged blood meal that the tick receives when it feeds as an adult on large mammalian hosts acts as its primary form of nutrition for tick egg production. For this reason, tick abundance may be determined at least in part by the abundance of the large host species, but, at the same time, these are the species that tend to be noncompetent and unable to either amplify or transmit the pathogen. Therefore, some components of diversity, such as deer, might simultaneously reduce infection prevalence of ticks and boost tick numbers---leading to mixed effects on disease risk. In the case of tick-borne diseases, therefore, species richness alone may be only part of the story, and we need to encompass both species diversity and abundance to identify how wild animal populations influence disease risk to humans. In contrast, the abundance of mosquitoes is often independent of the abundance of the hosts that provide female mosquitoes with blood meals. Although these blood meals are crucial to egg production, the local abundance of mosquitoes may be more strongly dependent on the abundance of pools of water in which to lay eggs. For mosquito-borne zoonoses, therefore, host diversity, per se, is more likely to influence disease risk. Palliative Plants {#s7} ================= Agricultural scientists have been interested in the effects of plant diversity on agricultural diseases for many decades, and several studies have shown that crop diversity can reduce the total burden of disease in agricultural systems. For example, the transmission of plant pathogens that specialize on particular crops can be reduced simply by interspersing other crop species that the pathogen does not readily infect. In essence, the nonhost plants can act as physical barriers to the dispersal of the pathogen, absorbing them without expressing disease or transmitting them further \[ [@pmed-0030231-b15]\]. A classic example of the potential effects of crop diversity on disease was found in corn. In experimental fields, the presence of a diversity of noncorn plants reduced incidence of corn smut, a fungal pathogen \[ [@pmed-0030231-b16]\]. More recently, diverse mixtures of rice varieties showed both a lower incidence of rice blast disease and a greater yield than did monocultures of a single rice variety \[ [@pmed-0030231-b17]\]. The role of high plant diversity in reducing agricultural disease transmission provides strong analogies to zoonotic systems. Plant species that are incompetent pathogen reservoirs can reduce transmission by regulating the size of the host population (after all, space taken up by incompetent reservoirs can\'t be colonized by the competent ones). Diverse species in both types of systems can absorb but do not transmit infections. Predators and Chronic Wasting Disease {#s8} ===================================== These plant studies illustrate the potential role that adding competitors can have on disease risk, but reductions in disease prevalence also occur in situations when predators are added to an ecological community. For example, we have long known from studies of predators feeding on Dall Sheep or Serengeti wildebeest that predators selectively prey upon sick and diseased individuals that are easier to capture \[ [@pmed-0030231-b18]\]. The removal of these sick animals before they die from the infection removes the most heavily infected individuals from the host population and reduces parasite transmission rates. The classic example of this occurs with red grouse, gamebirds in northern Britain. Here, the presence of parasitic nematodes increases the vulnerability of birds to foxes and to birds of prey. When the predators remove these heavily infected birds, then the infective stages the parasites would have produced are removed from the system and the net parasite abundance in the surviving bird population is reduced, leading to a rise in the bird population. The predators are actually making the grouse population healthier and are leading to an increase in their abundance, as long as the predator is a specialist \[ [@pmed-0030231-b19]\]. Similar effects may occur with the recently emerging spongiform encephalopathies of deer (chronic wasting disease) and domestic livestock (bovine spongiform encephalopathy, or "Mad Cow Disease"), which are threatening the future of the agricultural industries worldwide. Either of these could potentially cross the species barrier to humans, causing variant Creutzfeldt-Jacob Disease or chronic brain wasting disease. The primary mode of transmission for these pathogens appears to occur when infected livestock die (usually over winter). Transmission from the infected carcasses occurs when the carcasses are gnawed upon by other creatures that are nutritionally stressed. A curious feature of the spongiform encephalopathies is that their natural range is usually in an area with very poor soil, where hosts naturally suffer bone mineral deficiencies. When scavengers, such as coyotes and buzzards, are abundant, the carcasses will not last long, and it\'s unlikely that they will be available for transmission. However, when coyote and buzzard numbers are reduced by game managers, the carcasses can persist in the environment, and transmission rates might allow the pathogens to both increase in prevalence and establish themselves in new regions. Intriguingly, dogs seem to be totally resistant to prions. Although many dogs in the United Kingdom probably ingested infected beef, there are no veterinary reports of spongiform encephalopathies in dogs; several records occur for cats. This absence of dog cases may reflect past selection for deletion of prion susceptibility in canid species that obtain significant amounts of their food from scavenging or preying upon weakened individuals. Biological Diversity and Global Climate Change {#s9} ============================================== Studies of the effects of diversity on disease are providing important insights into the major role that ecological communities play in regulating the natural abundance of zoonotic pathogens that infect humans and their domestic livestock. The problems are scientifically challenging because they involve understanding the dynamics of complex multispecies systems where birth and death rates operate on a variety of different time scales. Given that significant threats to human health may be buffered by the presence of a diversity of other species, we need to understand the dynamics of species interactions. Unfortunately, this need is increasingly urgent because we are losing biological diversity at the fastest rate ever recorded. Understanding species interactions is particularly important (or urgent) when we consider how the world of infectious diseases is likely to change in the face of ongoing climate change. At present, vector-borne diseases of humans are much more prevalent in the tropics. Tropical infections such as malaria, sleeping sickness, dengue fever, Chagas disease, leishmaniasis, and yellow fever are all diseases that worry Western tourists and military planners. The main health and economic impact of these diseases is felt by the people who live in the world\'s poorest tropical countries, and many would argue that these pathogens are the principal economic constraint on these countries \[ [@pmed-0030231-b20]\]. The warm and humid climates of the tropics provide ideal conditions for many vectors and pathogens. As the world becomes warmer, many of these pathogens may be able to spread beyond the tropics \[ [@pmed-0030231-b21]\]. Here we suddenly discover a supreme irony: although vector-transmitted diseases take a significant toll on human health in the tropics, this toll may be significantly buffered by the presence of the large diversity of other species with which tropical people coexist. Now, as we convert habitats for agriculture or with urbanization, we improve human access to food and infrastructure, but we may also reduce the ability of natural systems to buffer disease. How much worse will things get in the tropics as biodiversity declines there? Finally, we should note that as vector-transmitted diseases disperse into the current temperate zones, they will not only benefit from a wetter and warmer world, but also from one in which the natural level of biodiversity is lower than in the tropics. Certainly, some host species may also spread from the tropics into the temperate zone, but larger species typically spread at a slower rate than smaller ones, so for a significant time, vector-transmitted diseases will be moving down a gradient of biodiversity. Given a restricted choice of hosts on which to feed, they are likely to focus their attention on the most common and most abundant species: humans, and their domestic animals and plants. This provides a strong selfish motivation to conserve biological diversity---our health may depend upon it. **Funding:** The Pennsylvania State University, Center for Infectious Disease Dynamics provided logistical and financial support for the meeting at which this paper was written. APD\'s research is supported by grants under the National Institutes of Health/ National Science Foundation Ecology of Infectious Disease Program. **Citation:** Dobson A, Cattadori I, Holt RD, Ostfeld RS, Keesing F, et al. (2006) Sacred cows and sympathetic squirrels: The importance of biological diversity to human health. PLoS Med 3(6): e231. DOI: 10.1371/ journal.pmed.0030231 TBE : tick-borne encephalitis WNV : West Nile virus [^1]: **Competing Interests:** The authors have declared that no competing interests exist. [^2]: Andy Dobson is in the Department of Ecology, Evolution, and Biology, Princeton University, Princeton, New Jersey, United States of America. Isabella Cattadori, Kristle Krichbaum, Jason R. Rohr, Sarah E. Perkins, Peter J. Hudson are at the Center for Infectious Disease Dynamics, Pennsylvania State University, University Park, Pennsylvania, United States of America. Robert D. Holt is in the Department of Zoology, University of Florida, Gainesville, Florida, United States of America. Richard S. Ostfeld is at the Institute of Ecosystem Studies, Millbrook, New York, United States of America. Felicia Keesing is in the Department of Biology, Bard College, Annandale-on-Hudson, New York, United States of America.

We are pleased that our recent publication ([@bib0003]) has opened a constructive discussion ([@bib0006]). We agree with Garolla and colleagues that there are detrimental effects of human papilloma virus (HPV) on fertility and pregnancy outcomes, which is why we feel it is important to investigate some potential countermeasures. As stated in our paper, further studies should be performed. However, we would like to reply to the authors' specific concerns as follows: (i) PCR is the preferred designated method for virus detection, including clinical diagnosis of HPV ([@bib0001]), HIV ([@bib0002]) and SARS-CoV-2 in the current COVID-19 pandemic ([@bib0008]). Highly specific and highly sensitive, PCR allows HPV-detection of only a few copies in the whole ejaculate, including cell-free HPV in seminal plasma. Moreover, we believe that consideration of the findings of [@bib0009] scarcely apply to HPV detection, given that the aim there was to investigate intra-clonal heterogeneity in cancer, making bulk analyses inappropriate in that case; (ii) notoriously, HPV is still detected upon \'standard\' sperm-washing procedures ([@bib0004]), as confirmed in Figure 1 in [@bib0006]. However, we apply the \'combined\' sperm-washing procedure, which differs from the \'standard\' one ([@bib0003]; [@bib0007]); (iii) thus far, there is no universal HPV-screening in assisted reproductive technology procedures and we aim to raise awareness of the potential importance of this step. We also recommend caution and we encourage other suitable beneficial approaches to be adopted, including an anti-HPV vaccine ([@bib0005]). Indeed, we believe that a joint, double-blinded study could be helpful in defining more appropriate guidelines, as proposed at the 2019 meeting of the Italian Society for the Study of Fertility and Sterility and Reproductive Medicine (SIFES--MR).

All relevant data are within the paper and its Supporting Information files. Introduction {#sec005} ============ Owing to their physical and aesthetic properties, resin-based materials are routinely used to restore the structure and function of teeth \[[@pone.0147858.ref001], [@pone.0147858.ref002]\]. However, residual monomers released from resin restorations as a result of incomplete polymerization could have irritating effects on the oral tissues. Several dental monomers, including 2-hydroxyethyl methacrylate (HEMA), methyl methacrylate (MMA), and triethylenglycol dimethacrylate (TEGDMA), have been identified as cytotoxic molecules that disrupt the stable redox balance and result in oxidative stress \[[@pone.0147858.ref003], [@pone.0147858.ref004]\]. The imbalanced redox state of the cells, characterized by the over-production of reactive oxygen species (ROS) and depletion of glutathione (GSH), has been shown to induce cell death via apoptosis \[[@pone.0147858.ref004]--[@pone.0147858.ref006]\]. However, the exact and detailed mechanism underlying dental monomer-induced apoptosis is still largely unknown. Apoptosis can be triggered by various signals. In particular, ROS can induce oxidative DNA damage, which can subsequently upregulate p53, and thus trigger intrinsic mitochondrial apoptosis by shifting the balance in the Bcl-2 family \[[@pone.0147858.ref007]--[@pone.0147858.ref009]\]. Thus, one of the purposes of the present study is to investigate the possible involvement of mitochondrial intrinsic apoptotic pathway in dental monomer-induced cytotoxicity. Novel strategies are needed to reduce the adverse effects of dental materials. To design such strategies, it is necessary to understand the exact mechanisms by which these materials induce cell death and to find strategies to decrease or eliminate their toxicities. N-acetyl cysteine (NAC) has played a well-documented role in detoxifying dental monomers and resinous materials \[[@pone.0147858.ref010]--[@pone.0147858.ref012]\]. However, the influences of NAC on dental monomer-induced apoptosis have not been elucidated yet. Thus, the second and the most important purpose of the present study is to investigate the influences of NAC on dental monomer-induced apoptosis. Materials and Methods {#sec006} ===================== Cell cultures and cell proliferation assay {#sec007} ------------------------------------------ Human dental pulp cells (hDPCs) were derived from primary culture, as described in our previous study \[[@pone.0147858.ref012]\]. Briefly, hDPCs were isolated from the dental pulp tissues of non-carious third molars extracted from young healthy patients (18--25 years old), according to a protocol that was verbally approved by the Ethics Committee of the Fourth Military Medical University (approval number: 15--20) with written informed consent obtained from all subjects. Extracted teeth were delivered to the cell culture laboratory in isolation medium containing alpha-modified Eagle's medium (α-MEM; Gibco BRL Division of Invitrogen, Gaithersburg, MD, USA) supplemented with 10% fetal bovine serum (Gibco), 100 units/mL penicillin, and 100 mg/mL streptomycin. Upon arrival to the laboratory, the dental pulp was minced and digested in a solution containing 3 mg/mL type I collagenase and 4 mg/mL dispase (Gibco) at 37°C for 2 h \[[@pone.0147858.ref013]\]. Single-cell suspensions were obtained by passing the cells through a 70-mm strainer (BD Falcon, Franklin Lakes, NJ, USA) and cultured in α-MEM supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 mg/mL streptomycin. Media were changed every 3 days in 5% CO~2~ at 37°C. Firstly, Cell Counting Kit-8 (CCK-8) (Beyotime Biotechnology, China) was used to determine the influences of dental monomers on the viability of hDPCs. Briefly, hDPCs at the second passage were seeded into 96-well culture plates at 5 × 10^3^ cells/well and incubated at 37°C and 5% CO~2~ for approximately 24 h. When the cells reached 80% confluence, media were removed. Cells were treated with media containing 2-hydroxyethyl methacrylate (HEMA; 1, 2.5, 5, 7.5, or 10 mM), methyl methacrylate (MMA; 1, 2.5, 5, 7.5, or 10 mM), or triethylenglycol dimethacrylate (TEGDMA; 1, 2.5, 5, 7.5, or 10 mM) for 24 h. All the tested dental monomers were purchased from Sigma-Aldrich (St. Louis, MO, USA) and dissolved in complete media. CCK-8 solution was added, and the cells were incubated at 37°C for another 4 h. Absorbance of the colored solution was measured with a microplate reader (Bio-Rad Laboratories, Hercules, CA) at a wavelength of 450 nm. Cells in the control group were treated with culture medium without dental monomers. Here, we found the lowest concentration for each monomer that could induce a significant decrease in cell viability (HEMA: 1mM; MMA: 5mM; TEGDMA: 1mM). In the following study evaluating dental monomer-induced oxidative stress and apoptosis, 1 mM HEMA, 5 mM MMA and 1 mM TEGDMA were used. Human DPCs were also exposed to 1mM HEMA, 5mM MMA and 1mM TEGDMA in the absence or presence of 10 mM NAC for 24, 48, 72 or 96 h to test the influence of NAC on monomer-induced cytotoxicity. NAC (Sigma; St Louis, MO, USA) was dissolved in complete media to a final concentration of 10 mM as previously described \[[@pone.0147858.ref014]\]. After co-treatment of NAC and monomer for 24 h, the morphology of the cultured cells was observed and documented by phase contrast microscopy (1X70, Olympus, Tokyo, Japan). Results for CCK-8 assay were plotted as the mean ± SD of three independent experiments with six determinations per sample for each experiment. Measurement of intracellular reactive oxygen species (ROS) {#sec008} ---------------------------------------------------------- Levels of intracellular ROS were measured with an ROS assay kit (Beyotime Biotechnology, China). After treatment with dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) in the absence or presence of 10 mM NAC for 6 h, hDPCs were treated with dihydrodichlorofluorescein diacetate (DCF-DA), incubated in serum-free media at 37°C for 20 min, harvested in phosphate-buffered saline (PBS), and analyzed in a FACSCalibur flow cytometer (BD Biosciences). Results were plotted as the mean ± SD of three determinations per sample for each experiment, with 10,000 cells being analyzed for each sample. Biochemical assays {#sec009} ------------------ Human DPCs were seeded into a 6-well plate at a density of 1 × 10^5^ cells/well, incubated for 24 h, and treated with dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) in the absence or presence of 10 mM NAC. After 24 h, the cells were collected and resuspended in the buffer for lyticase (Sigma, USA) for 30 min at 30°C. Then the lysate was centrifugated (10,000×g, 5 min, 4°C), and the supernatant was collected. Total protein concentrations of parallel samples were measured using a BCA Protein Assay Kit (Beyotime Biotechnology, China). The supernatant was aliquoted and stored at -80°C. All experiments were repeated three times. Contents of total glutathione (total GSH), reduced glutathione (GSH) and oxidized disulfide (GSSG) were measured using a commercially available kit (Beyotime Biotechnology, China) according to the manufacturer's protocol. After the supernatant was collected, total GSH was assayed using the 5,5-dithio-bis(2-nitrobenzoic) acid (DTNB)-GSSG reductase recycling. GSSG was measured by measuring 5-thio-2-nitrobenzoic acid (TNB) which was produced from the reaction of reduced GSH with DTNB. Then the fluorescence of the samples was evaluated using a microplate reader (Bio-Rad Laboratories, Hercules, CA) at 405 nm. Total GSH and GSSG concentration were normalized to protein contents. The concentration of reduced GSH in the sample was obtained by subtracting GSSG from total GSH. The experiment was done in triplicate and repeated for three times. Lipid peroxidation was assayed by the measurement of malondialdehyde (MDA), according to the instructions of lipid peroxidation MDA assay kit (Beyotime Biotechnology, China). One hundred μL supernatant was mixed with 100 μL MDA working solution and boiled at 100°C for 15 min. Samples were cooled down to room temperature in a water bath and then centrifuged at 1,000g for 10 min. The absorbance at 532 nm was measured in a 96-well plate (200 μL/well) with a microplate reader (Bio-Rad Laboratories, Hercules, CA). The results of MDA assay were expressed as micromoles of MDA per milligram of protein (nmol/mg protein). The experiment was done in triplicate and repeated for three times. The superoxide dismutase (SOD) activity was quantified with the SOD Assay Kit (Beyotime Biotechnology, China). The 50% inhibitory activity of SOD was determined by the colorimetric method and quantified by measuring the absorbance at 450 nm using a microplate reader (Bio-Rad Laboratories, Hercules, CA). The activity of glutathione peroxidase (GPx) was quantified in the supernatant by using a GPx Assay Kit (Beyotime Biotechnology, China) and measuring the degree of reducing absorption of NADPH at 340 nm. The catalase (CAT) activity was quantified by using the CAT Assay Kit (Beyotime Biotechnology, China). The activities of SOD, GPx and CAT were normalized to protein contents. Results were plotted as the mean ± SD of three independent experiments with three determinations per sample for each experiment. Quantification of apoptotic cells by flow cytometry {#sec010} --------------------------------------------------- Dental monomer-induced cell apoptosis and the protective effects of NAC were quantified with an Annexin V-FITC apoptosis detection kit (Beyotime Biotechnology, China). Human DPCs were seeded into a 6-well plate at 1×10^5^ cells/well, incubated for 24 h, and treated with HEMA (0-1-5-10 mM), MMA (0-5-10-15 mM) or TEGDMA (0-1-5-10 mM), in the absence or presence of 10 mM NAC. Untreated cells served as the negative control. After culturing for 24 h, hDPCs were collected and washed with PBS, gently resuspended in Annexin V binding buffer, and incubated with Annexin V-FITC/propidium iodide (PI). Flow cytometry was performed using flow cytometric analysis (Becton-Dickinson FACScan). Results were plotted as the mean ± SD of three determinations per sample for each experiment, and 10,000 cells were analyzed for each sample. Analysis of caspase-3 activity {#sec011} ------------------------------ To investigate caspase-3 activation after treatment with dental monomers in the absence or presence of NAC, the Caspase-3 Colorimetric Assay Kit (Beyotime Biotechnology, China) was used. Briefly, 1×10^6^ cells treated with dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) with or without 10 mM NAC for 24 h were collected and lysed in a lysis buffer. Cell lysates were tested for protease activity by using Asp-Glu-Val-Asp-pNA (DEVD-pNA), a tetrapeptide p-nitroanilide substrate. Triple wells were used for each group. After incubation with this substrate for 2 h, the absorbance was measured at 405 nm with a microplate reader (Bio-Rad Laboratories, Hercules, CA). Caspase enzymatic activities in cell lysates were normalized to protein contents measured in a BCA Protein Assay Kit. The results were presented as means ± SD of three independent experiments. Intercellular ATP level determination {#sec012} ------------------------------------- ATP level was measured by using a firefly luciferase based ATP Assay Kit (Beyotime Biotechnology, China) according to the manufacturer's instructions. Cells were seeded into a 6-well plate at a density of 1×10^5^ cells/well. Then the cells were treated with dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) in the absence or presence of 10 mM NAC for 24 h. After rinsed with PBS, the cells were schizolysised by solution and then centrifuged at 12,000g at 4°C for 15 min and the supernatant was collected. After 100 μL of the supernatant was mixed with 100 μL of ATP detection solution, intensity readings were taken with a microplate reader (Bio-Rad Laboratories, Hercules, CA). The ATP concentrations in samples were calculated using an ATP standard curve and the protein concentration of each treatment group was determined using the BCA protein assay kit. The cellular ATP levels were expressed as nmol/mg protein. Assessment of mitochondrial membrane potential (MMP) {#sec013} ---------------------------------------------------- The fluorescent dye JC-1 (Beyotime Biotechnology, China) was used to detect cells with collapsed MMP. Briefly, cells cultured in α-MEM were exposed to dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) in the absence or presence of 10 mM NAC for 24 h and incubated with the JC-1 staining solution (5 mg/ml) for 20 min at 37°C. Cells were rinsed twice with JC-1 staining buffer. The fluorescence intensity (FI) of JC-1 aggregates was detected at an excitation/emission wavelength ratio of 525 nm/590 nm, and the FI of the JC-1 monomers was measured at an excitation/emission wavelength ratio of 490 nm/530 nm by flow cytometry (Becton-Dickinson FACScan). All results were plotted as the mean ± SD of three determinations per sample for each experiment, and 10,000 cells were analyzed for each sample. Western blot analysis {#sec014} --------------------- Human DPCs cultured in α-MEM were exposed to dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) in the absence or presence of 10 mM NAC for 24 h (*n* = 3) and lysed in RIPA buffer. Protein contents of the lysed cells were measured with the BCA Protein Assay Kit. Extracted proteins were loaded on 10% sodium dodecyl sulfate polyacrylamide gels, transferred to polyvinylidene fluoride membranes (Bio-Rad), and blocked with 5% nonfat milk powder. Membranes were incubated overnight with the following primary rabbit anti-human antibodies (Cell Signaling Technology, Danvers, MA): anti-Bcl-2, anti-Bax, anti-p53, anti-cleaved caspase-3, and actin. Membranes were incubated with goat anti-rabbit secondary antibodies. Protein signals were visualized by using the ECL Western Blotting Detection System (GE Healthcare, Piscataway, NJ, USA). Protein expression levels were normalized to actin by using Image-Pro Plus 5.0 Software (Media Cybernetics Inc., Bethesda, MD, USA). Analysis of cytosolic cytochrome c (Cyto C) release {#sec015} --------------------------------------------------- After hDPCs were exposed to dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) without or with 10 mM NAC for 24 h, the release of CytoC from mitochondria to the cytoplasm was investigated. Cells were resuspended in 1.5 mL of cold Mito-Cyto Buffer (Applygen Biotechnology, China), lysed, and centrifuged. The supernatant was centrifuged again at 12,000 g for 10 min to separate mitochondria (in pellet) from the cytoplasm (in supernatant). Cytosolic Cyto C in the supernatant was quantified with an ELISA kit (Westang Biotechnology, China), using Cyto C (0--2000 pg/mL) as a standard. Data are expressed as the total amount of intracellular solubilized Cyto C. Cyto C levels were normalized to protein content measured by a BCA Protein Assay Kit. All results were plotted as the mean ± SD of three determinations per sample for each experiment. Immunofluorescence (IF) {#sec016} ----------------------- Human DPCs were seeded into a 6-well plate at a density of 1×10^5^ cells/well and exposed to dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) without or with 10 mM NAC for 24 h. The cells were then fixed with a modified Zamboni\'s fixative (4% paraformaldehyde and 0.19% picric acid in PBS, pH 7.4) for 30 min at room temperature as described before \[[@pone.0147858.ref015]\]. Cells were washed twice with PBS, preincubated in the same buffer containing 0.3% Triton X-100 (Amresco, USA), and blocked with 0.1% BSA (Amresco, USA) in PBS for 1h. The cells were incubated with primary antibodies (rabbit anti-Bax, 1:200, Beyotime Biotechnology, China; and mouse anti-Cytochrome C, 1:50, Boster Biotechnology, China) overnight at 4°C. After three washes with PBS, cells were incubated with secondary antibodies (Alexa Fluor 594 donkey-rabbit, 1:400; and Alexa Fluor 488 donkey-mouse, 1:200; Life Technologies, Carlsbad, CA) for 30 min at 37°C. Cells were then washed in PBS and further incubated with 10 mg/ml DAPI (Sigma) at room temperature for 10 min. Fluorescent images were obtained by laser scanning confocal microscopy (Keyence Co., Osaka, Japan). Transmission Electron Microscopy (TEM) {#sec017} -------------------------------------- Human DPCs were initially collected after exposure to dental monomers (1mM HEMA, 5mM MMA and 1mM TEGDMA) without or with 10 mM NAC for 24 h, and fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer for 2 h and post fixed with a solution of 1% osmium tetroxide in 0.1 M cacodylate buffer. The cells were then embedded in epoxy resins after a graded-ethanol serial dehydration step. The embedded cells were sectioned into ultrathin slices, stained by uranyl acetate solution and lead citrate, and then observed with a transmission electron microscope Tecnai G2 Spirit BioTWIN electron microscope (FEI Company, Eindhoven, The Netherlands). Statistical analysis {#sec018} -------------------- The SPSS 18.0 software package (SPSS Inc., Chicago, IL, USA) was used to perform statistical analyses. Data was analyzed by one-way analysis of variance (ANOVA), followed by the Tukey's test. The statistical significance level was set at *P* = 0.05 for all tests. Results {#sec019} ======= Influences of dental monomers and NAC on cell viability and cell morphology {#sec020} --------------------------------------------------------------------------- The CCK-8 assay results revealed that treatment of hDPCs with dental monomers for 24 h decreased cell viability in a dose-dependent manner ([Fig 1A](#pone.0147858.g001){ref-type="fig"}). Compared to the control group, groups that were treated with HEMA (≥1 mM), MMA (≥5 mM), or TEGDMA (≥1 mM) exhibited significantly lower cell viability. In subsequent experiments to investigate influences of dental monomers on intracellular redox balance and apoptosis, hDPCs were treated with dental monomers at the lowest concentrations that can induce significant difference on cell viability. [Fig 1B](#pone.0147858.g001){ref-type="fig"} showed representative cell micrographs after treatment with various dental monomers in the absence or presence of NAC. Cells in the control group exhibited normal growth and typical spindle appearance. However, many cells became round after treatment with dental monomers, and cell population density was further reduced. Co-treatment of NAC with monomers restored cell density and preserved the normal morphology of cells. Furthermore, the long-term effects of dental monomers and NAC on cell viability were also tested. As shown in [Fig 1C](#pone.0147858.g001){ref-type="fig"}, a time-dependent decrease in cell viability was observed in cells exposed to 1 mM HEMA after 24 h, 48 h, 72 h and 96 h (all *P* \< 0.05 as compared to control group). Likewise, similar time-dependent inhibitory effects were observed in cells exposed to 5 mM MMA and 1 mM TEDGMA (all *P* \< 0.05 as compared to control group). Quite the reverse, the presence of NAC could significantly restore the cell viability (all *P* \> 0.05 as compared to control group) for all the studied time periods. {#pone.0147858.g001} Influences of dental monomers on intracellular redox balance and the protective effects of NAC on hDPCs {#sec021} ------------------------------------------------------------------------------------------------------- Compared to the control group, the intracellular ROS levels were increased in hDPCs after treatment with dental monomers for 6 h (all *P* \< 0.05, [Fig 2A](#pone.0147858.g002){ref-type="fig"}). The over-production of ROS induced by dental monomers was significantly relieved by NAC treatment (all *P* \< 0.05). {#pone.0147858.g002} The GSH levels decreased after cells were treated with dental monomers (all *P* \< 0.05, [Fig 2B--2D](#pone.0147858.g002){ref-type="fig"}). As expected, NAC alleviated monomer-induced depletion of GSH (*P* \< 0.05 vs. dental monomer-treated cells). Compared to the control group, no difference in GSSG content was found after treatment with HEMA or MMA (*P* \> 0.05, [Fig 2B and 2C](#pone.0147858.g002){ref-type="fig"}), and only a small decrease was found after treatment with TEGDMA (*P* \< 0.05, [Fig 2D](#pone.0147858.g002){ref-type="fig"}). MDA level was remarkably increased after dental monomer treatment (all *P* \< 0.05 vs. control group cells, [Fig 3A](#pone.0147858.g003){ref-type="fig"}). Compared to dental monomer-treated cells, NAC reduced intracellular MDA content level to almost the normal level (all *P* \< 0.05). {#pone.0147858.g003} SOD activity decreased after dental monomer treatment (*P* \< 0.05 vs. control group cells, [Fig 3B](#pone.0147858.g003){ref-type="fig"}). Co-treatment of NAC with HEMA restored the decreased activity of SOD (*P* \< 0.05 vs. HEMA-treated cells). For MMA and TEGDMA, NAC could also alleviate decreased activity of SOD, although there was no significant difference (*P* \> 0.05 vs. dental monomer-treated cells). Similar results were observed for GPx activity ([Fig 3C](#pone.0147858.g003){ref-type="fig"}). CAT activity increased after 24 h of dental monomer treatment (all *P* \< 0.05 vs. control group cells, [Fig 3D](#pone.0147858.g003){ref-type="fig"}). NAC co-treatment with monomers reduced CAT activity to an almost normal level (*P* \< 0.05 vs. dental monomer-treated cells). Influences of dental monomers on apoptosis and the rescuing effects of NAC in hDPCs {#sec022} ----------------------------------------------------------------------------------- According to the results of Annexin V/PI staining, a dose-dependent decrease in the percentage of viable cells was observed in hDPCs exposed to dental monomers ([Fig 4A](#pone.0147858.g004){ref-type="fig"}). It was found that 10 mM NAC alone had nearly no influence on the viability of hDPCs (*P* \> 0.05 as compared to the control group). Dental monomers greatly dose-dependently increased the percentage of cells in early apoptosis as well as in late apoptosis/necrosis. Quite the reverse, NAC exhibited restoring effects on cell viability and meanwhile decreased the percent of early apoptotic and late apoptotic/necrotic cells. Compared to untreated cells, there was no statistical difference in late apoptosis/necrosis in 1 mM HEMA, 5 mM MMA and 1 mM TEGDMA-treated cells in the presence of NAC (*P* \> 0.05 as compared to the control group). {#pone.0147858.g004} The occurrence of apoptosis was also confirmed by evaluating the activity of caspase-3, whose activation plays a central role in the execution-phase of cell apoptosis \[[@pone.0147858.ref016], [@pone.0147858.ref017]\]. As shown in [Fig 4B](#pone.0147858.g004){ref-type="fig"}, caspase-3 activity significantly increased after the treatment of dental monomers in hDPCs (*P* \< 0.05 vs. control group cells). In contrast, co-treatment with NAC significantly inhibited dental monomer-induced activation of caspase-3 (*P* \< 0.05 vs. dental monomer-treated cells). Influences of dental monomers and NAC on the functions and morphologies of mitochondria {#sec023} --------------------------------------------------------------------------------------- Cellular ATP levels of hDPCs exposed to dental monomers, in the absence or presence of NAC, were demonstrated in [Fig 4C](#pone.0147858.g004){ref-type="fig"}. After 24 h of dental monomer treatment, ATP levels were statistically lower than those of the control groups (all *P*\< 0.05, [Fig 4C](#pone.0147858.g004){ref-type="fig"}). Quite in contrast, NAC could restore the ATP levels in dental monomers-treated cells (all *P*\> 0.05 as compared to the control group). Next, we analyzed the effects of dental monomers and NAC on MMP, which was an important biomarker of intrinsic mitochondrial pathway. As can be seen in [Fig 4D](#pone.0147858.g004){ref-type="fig"}, the percent of hDPCs with collapsed MMP was significantly increased after treatment with dental monomers (all *P* \< 0.05 vs. control group cells). NAC protected against dental monomer-induced depolarization of MMP (all *P* \< 0.05 vs. dental monomer-treated cells). The morphologies of mitochondria in cells exposed to dental monomers, in the absence or presence of NAC, were examined by transmission electron microscope (TEM) here. After monomer treatment, some cells showed nuclear and cytoplasmic signs of apoptosis (condensed/peripheralized nuclear chromatin and/or cytoplasm, cytoplasmic vacuolization). As for the mitochondira, the control cells ([Fig 5A--5E](#pone.0147858.g005){ref-type="fig"}) and NAC-treated cells ([Fig 5I--5M](#pone.0147858.g005){ref-type="fig"}) showed integrity of mitochondrial membrane (inner and outer membranes), and the mitochondria cristae were not affected. However, the ultrastructures of dental monomers treated cells showed altered membrane structural integrity. Larger and elongated mitochondria with less number of cristae or deformed cristae were observed. Quite the reverse, the treatment with NAC could restore the membrane structural integrity of mitochondria in cells exposed to monomers, although some the mitochondria still showed deformed cristae ([Fig 5J--5L, 5N--5P](#pone.0147858.g005){ref-type="fig"}). {#pone.0147858.g005} Influences of dental monomers and NAC on key molecules of the intrinsic mitochondrial apoptosis pathway {#sec024} ------------------------------------------------------------------------------------------------------- Bcl-2 protein family is the key regulator of intrinsic mitochondrial apoptosis. Among the many members of Bcl-2 family proteins, Bcl-2 and Bax are two well-known molecules with anti-apoptotic, or proapoptotic effects, respectively \[[@pone.0147858.ref008], [@pone.0147858.ref018]\]. Western blot analysis revealed that in the presence of dental monomers, the expression of Bcl-2 was down-regulated (all *P* \< 0.05 as compared to the control group, [Fig 6A--6C](#pone.0147858.g006){ref-type="fig"}), while the expression of Bax was up-regulated (*P* \< 0.05 for TEGDMA as compared to the control group). The expression levels of Bax after HEMA or MMA treatment also showed a trend of increase, although there was no statistical significance compared to the control group (all *P* \> 0.05). Co-treatment with NAC resulted in enhanced Bcl-2 expression (all *P* \< 0.05 vs. dental monomer-treated cells), and reduced Bax expression (*P* \< 0.05 for TEGDMA as compared to dental monomer-treated cells, and a marginal decrease without statistical significance for HEMA and MMA). p53 is a sensor of cellular stress and is a critical activator of the intrinsic mitochondrial apoptosis pathway. The p53-dependent regulation of apoptosis occurs through transcription dependent and transcription-independent mechanisms to either activate or in activate multidomain pro- and anti-apoptotic Bcl-2 proteins \[[@pone.0147858.ref007]\]. Dental monomer treatment significantly increased the expression of p53 (all *P* \< 0.05 vs. control group cells) and NAC co-treatment significantly reversed dental monomer-induced up-regulation of p53. As for caspase-3, which is the executor of apoptosis, dental monomers greatly enhanced the intracellular level of cleaved caspase-3 (all *P* \< 0.05 vs. control group cells, [Fig 6D](#pone.0147858.g006){ref-type="fig"}), whereas the presence of NAC could block the activation of caspase-3 (*P* \< 0.05 vs. dental monomer-treated cells). {#pone.0147858.g006} Cyto C is an important intermediate in intrinsic mitochondrial apoptosis. The release of Cyto C from mitochondria could activate caspase-9, which in turn activates caspase-3 and caspase-7 to execute cell death \[[@pone.0147858.ref019], [@pone.0147858.ref020]\]. Results of ELISA showed that the cytosolic Cyto C level was elevated when cells were exposed to dental monomers for 24 h (all *P* \< 0.05 vs. control group cells, [Fig 6E](#pone.0147858.g006){ref-type="fig"}). The increase was blocked by co-treatment with NAC (*P* \< 0.05 vs. dental monomer-treated cells). These observations indicate that NAC prevented the release of Cyto C from mitochondria into the cytoplasm after dental monomer treatment. Immunofluorescence (IF) analysis revealed that Bax, which is diffusely present as fine granules in the cytosol of controls, translocated to mitochondria after dental monomer treatment with a punctate perinuclear distribution typical of mitochondria ([Fig 7](#pone.0147858.g007){ref-type="fig"}). In contrast, Cyto C, which was retained in the mitochondria in controls, redistributed diffusely in the cytoplasm ([Fig 7](#pone.0147858.g007){ref-type="fig"}). Quite the reverse, in the presence of NAC, Bax redistributed diffusely to cytosol, and meanwhile most of the strong spot signal of Cyto C was retained in the mitochondria in dental monomer-exposed cells ([Fig 7](#pone.0147858.g007){ref-type="fig"}). {#pone.0147858.g007} Discussion {#sec025} ========== In the present study, we demonstrated that dental monomers could disturb the intracellular redox balance, impair the functions of mitochondria and thus induced apoptosis through the intrinsic mitochondrial pathway in hDPCs. The presence of NAC could relieve dental monomer-induced oxidative stress and subsequently protect the cells from apoptosis. a. Dental monomer-induced oxidative stress in hDPCs {#sec026} --------------------------------------------------- In our study, the results of CCK-8 assay revealed that dental monomers dose-dependently inhibited cell viability of hDPCs. However, the tested monomers revealed different level of toxic effects on hDPCs, with MMA showing the lowest toxicity, followed by HEMA and then TEGDMA. This result is in accordance with previous studies \[[@pone.0147858.ref021], [@pone.0147858.ref022]\]. The difference in the cytotoxicity of different dental monomers might be attributed to their chemical structures, such as the number of methacrylic groups, which can solubilize cell membrane and react with GSH to cause disturbance of intracellular redox balance \[[@pone.0147858.ref023]--[@pone.0147858.ref025]\]. The dental monomer-induced cytotoxicity was paralleled by over-production of ROS and depletion of GSH in hDPCs. This result is in accordance with previous studies, which indicate that oxidative stress is the primary mechanism for the toxic effects of dental monomers \[[@pone.0147858.ref026], [@pone.0147858.ref027]\]. GSSG is the product when reduced GSH react with oxidants \[[@pone.0147858.ref028]\]. If the reduction of GSH level is a secondary result caused by ROS overproduction, an increased level of GSSG should be expected. However, in our study, it was found that dental monomers reduced GSH level without significantly increasing GSSG level. Thus, it is believed that the reduction in intracellular level of reduced GSH after dental monomer treatment is not due to the oxidation of GSH to GSSG as a secondary result of ROS over-production. Rather, GSH may be directly depleted by the monomers. Indeed, several researchers reported that the methacrylic group of dental monomers could directly react with the thiol group of GSH through Michael-type addition reaction, and thus caused the depletion of GSH \[[@pone.0147858.ref029]\]. Under physiological conditions, there is a balance between ROS and antioxidative defense system \[[@pone.0147858.ref030]\]. However, when GSH, the key component of the antioxidative defense system, is depleted by dental monomers, ROS level could subsequently be elevated as a secondary result. When the formation of ROS during monomer exposure is beyond the capacities of anti-oxidative mechanisms, the overproduced ROS can react with cellular macromolecules, such as lipids, proteins and DNA, and detrimental damages can be resulted. MDA is the results from lipid peroxidation of polyunsaturated fatty acids, and is deemed as an indicator of oxidative damage \[[@pone.0147858.ref031]\]. Our data showed that MDA generation drastically increased after treatment with dental monomers, indicating that oxidative damage was generated after hDPCs were exposed to dental monomers for 24 h. Besides the non-enzymatic antioxidant GSH, enzymatic antioxidants directly control cellular redox homeostasis by regulating the levels of particular ROS. In this study, CAT activity was increased after exposure to dental monomers, whereas SOD and GPx activities were reduced. Functionally, SOD catalyzes the breakdown of O~2~^-^ into O~2~ and H~2~O~2~, whereas CAT and GPx subsequently conduct the conversion of H~2~O~2~ to H~2~O and O~2~ \[[@pone.0147858.ref032], [@pone.0147858.ref033]\]. A recent study revealed that H~2~O~2~ was the major sort of ROS after dental monomer treatment \[[@pone.0147858.ref034]\]. Because GPx requires GSH as a substrate to reduce H~2~O~2~, the limited source of the substrate GSH as a consequence of monomer-induced GSH depletion can subsequently restrict the activity of GPx, causing the observed reduction in GPx activity \[[@pone.0147858.ref033]\]. The elevated H~2~O~2~ formation may lead the increased CAT activity and decreased SOD activity levels as a result of a feedback regulation \[[@pone.0147858.ref035]\]. b. Dental monomers impaired the functions of mitochondria and induced intrinsic mitochondrial apoptosis in hDPCs {#sec027} ---------------------------------------------------------------------------------------------------------------- In this study, a dose-dependent decrease in the percentage of viable cells, as well as an increase in apoptotic and necrotic cells was observed in hDPCs exposed to dental monomers. Furthermore, dental monomers enhanced the activity of caspase-3. These data is in accordance with many previously published researches \[[@pone.0147858.ref003], [@pone.0147858.ref034]\]. Although it is well accepted that dental monomers can cause apoptosis, its detailed pathway is still largely unknown. Apoptosis is a complex process characterized by cell shrinkage, chromatin condensation and internucleosomal DNA fragmentation \[[@pone.0147858.ref036]\]. Considering that the over-produced ROS after dental monomer treatment can attack DNA and cause DNA double strand breaks (DSBs), which may subsequently activate the stress sensor p53, we hypothesized that dental monomer-induced apoptosis may occur through a p53-related pathway. Thus we investigated the changes of p53 level after dental monomer treatment and found that p53 expression was remarkably elevated. Besides its anti-tumor activity, p53 is also a key element in the intrinsic mitochondrial apoptosis pathway \[[@pone.0147858.ref007], [@pone.0147858.ref008]\]. It regulates intrinsic mitochondrial apoptosis through transcription-dependent and transcription-independent mechanisms to either activate or inactivate multidomain pro- and anti-apoptotic Bcl-2 proteins. Therefore, we subsequently hypothesized that p53-related intrinsic mitochondrial apoptosis pathway might be involved in dental monomer-induced cell death. Therefore, the morphologies and functions of mitochondria were studied in details. TEM observation indicated that dental monomers had very destructive effects on mitochondria. Several larger and elongated mitochondria with unclear mitochondrial membrane, less cristae or deformed cristae were observed. The changes in mitochondrial morphology were accompanied with a decreased level of ATP. Furthermore, disruption in the integrity of mitochondrial membrane, inevitably leaded to depolarization of MMP and subsequent Cyto C release from mitochondrial membrane into the cytoplasm. Based on these results, we then moved on to study the expression level of other key molecules of the intrinsic mitochondrial apoptosis pathway to support our hypothesis. The results of Western Blot analysis revealed that dental monomers induced upregulation of the pro-apoptotic protein Bax and cleaved caspase-3, as well as the downregulation of the anti-apoptotic protein Bcl-2. The intrinsic mitochondrial apoptosis pathway had been well studied by numerous studies \[[@pone.0147858.ref034], [@pone.0147858.ref037]\]. Apoptosis factors acting on the mitochondria are triggered with mitochondrial membrane damage, reduction in MMP, and deficiency of respiratory chain. The apoptotic factors of Bcl-2 family play an important role in regulating the expression of anti-apoptotic and pro-apoptotic members of the Bcl-2. After cells receive apoptosis signal, Bcl-2 family proteins change their usual localization and target patterns from the cytoplasm to the mitochondrial membrane, trigger mitochondrial dysfunction, Cyto C release and caspase activation, and eventually lead to cell death \[[@pone.0147858.ref038]\]. Indeed, our immunofluorescence analysis revealed that after dental monomer treatment, Cyto C was released from the mitochondria to cytoplasm, while Bax translocated from cytoplasm to mitochondria. The observed activation of caspase-3, impaired mitochondria function, changes in p53, Bcl-2, Bax and translocation of Bax and Cyto C, supported our hypothesis that the intrinsic mitochondrial pathway might be involved in dental monomer-related cytotoxicity. c. NAC reduced dental monomer-induced oxidative stress and blocked the apoptotic effects in hDPCs {#sec028} ------------------------------------------------------------------------------------------------- NAC is a well-known free radical scavenger that can readily enter the cells. We found that NAC could alleviate dental monomer-induced oxidative stress and thus protected the cells from apoptosis. The protective effects of NAC may be attributed to three aspects. First, it can directly scavenge the over-produced ROS. Second, as a cysteine-donating compound, NAC acts as a cellular precursor of GSH to replenish the depleted GSH pool after dental monomer treatment \[[@pone.0147858.ref039]\]. Third, according to recent publications, NAC may also directly react with the methacrylic group of dental monomers through Michael-type addition reaction, and thus reduce the availability of free dental monomers \[[@pone.0147858.ref040]--[@pone.0147858.ref042]\]. In a recent review, which discussed the mechanisms underlying adaptive cell responses to dental monomer-induced oxidative stress, the authors suggested that the formation of adducts with dental monomers could only be partially responsible for the protective effects of NAC against monomer-induced cytotoxic effects \[[@pone.0147858.ref035]\]. Some researchers, using GSH synthesis-modulating substances such as GSH synthesis inhibitor buthionine sulfoximine (BSO) and, GSH synthesis promotor 2-oxo-4-thiazolidine-carboxylic acid (OTC), found that the intracellular GSH is the primary antioxidant central to the regulation of cell response towards oxidative stress induced by dental monomers \[[@pone.0147858.ref014], [@pone.0147858.ref034]\]. In addition, OTC showed similar protective effects on dental monomers-induced cytotoxicity as NAC. These results indicate that, the protective effect of NAC against monomer-related toxicity might be mainly attributed to replenish the depleted GSH pool. However, the detailed mechanism underlying NAC protection against dental monomer-induced cytotoxicity needs further investigation. Conclusions {#sec029} =========== The conclusions from the present findings are summarized in a hypothetical model ([Fig 8](#pone.0147858.g008){ref-type="fig"}). With the limitations of the present *in vitro* study, it might be concluded that dental monomers caused disturbance of intracellular redox balance, which is characterized by depletion of GSH, over-production of ROS, and differential changes of antioxidative enzymes. The oxidative stress subsequently impaired mitochondria function, up-regulated p53, and eventually initiated the intrinsic mitochondrial apoptosis pathway. NAC remarkably relieved dental monomer-induced oxidative stress and subsequently protected the cells against apoptosis. {#pone.0147858.g008} Supporting Information {#sec030} ====================== ###### (XLS) ###### Click here for additional data file. Research and Development Center for Tissue Engineering of Fourth Military Medical University is greatly gratefully acknowledged for providing equipment. The authors thank the editor and anonymous reviewers for their comments and critique. [^1]: **Competing Interests:**The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: SM JC. Performed the experiments: YJ YW. Analyzed the data: LS JL QL. Contributed reagents/materials/analysis tools: LH YL QS FY HY HL. Wrote the paper: YJ SM JC.

1. Introduction {#sec1-pharmaceutics-11-00527} =============== The number of antibiotic resistant bacteria, as well as the number of immune deficient patients, is increasing. For this reason, infections of post-operative and chronic wounds are becoming a concern for many patients and health care providers \[[@B1-pharmaceutics-11-00527]\]. Age-related immune deficiency, diabetes mellitus type 2, venous insufficiency, or immobility are the major conditions leading to chronic wounds \[[@B2-pharmaceutics-11-00527],[@B3-pharmaceutics-11-00527],[@B4-pharmaceutics-11-00527]\]. Chronic wounds are often associated with microbial biofilms, i.e., organized communities of one or more microorganism species encased and shielded by extracellular polymeric substances \[[@B5-pharmaceutics-11-00527],[@B6-pharmaceutics-11-00527]\]. It was shown that significantly higher concentrations of antibiotics are required to treat biofilm-associated infections \[[@B7-pharmaceutics-11-00527],[@B8-pharmaceutics-11-00527]\]. One of the reasons for biofilm drug-resistance is the extracellular polymeric matrix, which acts as a shield protecting bacteria from the external environment. In addition, the low concentrations of antibiotics that reach the bacteria in the biofilm favor the formation of persisters, i.e., physiologically inactive dormant cells that are less responsive to antibiotics \[[@B9-pharmaceutics-11-00527]\]. Thus, for an efficient treatment of a biofilm-associated infection, a high and sustained concentration of antimicrobial drugs should be achieved at the site of infection and within the biofilm. This is often not fulfilled by most of the available antimicrobial formulations because of unfavorable physicochemical properties of drugs like instability in different biological environments, low solubility, or high molecular weight. New solubilization strategies and innovative pharmaceutical formulations, like polymer conjugates, nanocarriers, and membranes, have the potential to improve drug delivery and thus increase their concentration at the site of infection \[[@B10-pharmaceutics-11-00527],[@B11-pharmaceutics-11-00527]\]. In this study, we investigated the drug delivery properties of polyvinylpyrrolidone (PVP)-based foils and nanofiber mats. A ciprofloxacin base was chosen as model drug because of its low water-solubility, detectability using fluorescence spectroscopy, and quorum sensing inhibiting properties in *Pseudomonas aeruginosa* (*P. aeruginosa*) at concentrations below the minimal inhibitory concentration (MIC) \[[@B12-pharmaceutics-11-00527]\]. It is a broad-spectrum antibiotic that has been approved by the U.S. Food and Drug Administration (FDA) for the treatment of urinary tract and other infections, including skin and skin-structure infections. Nevertheless, it has severe side effects, and in 2016, the FDA recognized the existence of a rare side effect, a potentially permanent syndrome called fluoroquinolone-associated disability \[[@B13-pharmaceutics-11-00527]\]. Because of the low water solubility, ciprofloxacin is formulated into tablets for oral use as monohydrochloride monohydrate salt. In this form, its absolute bioavailability is of 60% \[[@B14-pharmaceutics-11-00527]\]. Alternatively, a ciprofloxacin base has been formulated in oil-based syrups \[[@B15-pharmaceutics-11-00527]\] with a better bioavailability of up to 70% \[[@B16-pharmaceutics-11-00527]\]. Formulations for topical use (e.g., ciprofloxacin ophthalmic ointment) are made by dissolving the drug in mineral oil. However, it was shown that drug crystals can still form and even cause the blockage of the bottle nozzle \[[@B17-pharmaceutics-11-00527]\]. In general, antibiotics are not used for the topical treatment of chronic wounds because of insufficient drug bioavailability and possible local side effects. However, topical drug delivery would be an advantage in those cases, like an infected diabetic foot, where blood circulation is reduced and systemic therapies are often inefficient. In addition, it has been shown that locally applied antibiotics can reduce the risk of surgical site infections \[[@B18-pharmaceutics-11-00527]\]. Finally, topical therapies reduce systemic toxicity and spare the gut microbiome. Thus, a more efficient topical drug delivery would improve both the management and prevention of wound infections. In a previous study, we used a pH modification strategy to solubilize a ciprofloxacin base and increase the amount loaded on PVP foils and nanofiber mats. Interestingly, the residual acetic acid bound to PVP conferred a peculiar transparency and elasticity to the foils. These delivery systems were shown to be safe and to have a good anti-microbial activity in vitro \[[@B19-pharmaceutics-11-00527]\]. In this study, we tested further properties of these materials using a wound infection model based on full-thickness human ex vivo skin. Different models are available to test the antimicrobial activity of a drug. In vitro grown biofilms give the possibility to study drug efficacy toward bacteria grown in communities. Nevertheless, in vitro biofilms are often very different from those found in in vivo infections. In vivo animal models have diverse advantages, including a tissue scaffold, host immune response, and wound healing processes. However, besides ethical reasons, animal studies are expensive and results are not always reproducible in humans. To reduce the number of animal studies, ex vivo models have been proposed for preliminary studies prior to animal studies. Ex vivo human skin has been used for several years to test the skin penetration of chemicals and drugs \[[@B20-pharmaceutics-11-00527]\]. Yet, in the last few years, ex vivo porcine or human skin has been used to develop wound infection models \[[@B21-pharmaceutics-11-00527],[@B22-pharmaceutics-11-00527],[@B23-pharmaceutics-11-00527],[@B24-pharmaceutics-11-00527],[@B25-pharmaceutics-11-00527],[@B26-pharmaceutics-11-00527],[@B27-pharmaceutics-11-00527]\]. Human full-thickness skin represents not only a three-dimensional scaffold where bacteria can grow, but also a complex environment with extracellular enzymatic activity, antimicrobial peptides, and several different cell populations, including immune-active cells. Thus, human skin infection models are realistic animal-free systems; even if they cannot completely replace in vivo studies, they are useful tools to screen antimicrobial formulations \[[@B26-pharmaceutics-11-00527]\]. Using this model, we could measure the drug delivery kinetics, monitor eventual toxic effects, and measure the antimicrobial activity of the tested materials. 2. Materials and Methods {#sec2-pharmaceutics-11-00527} ======================== 2.1. Preparation of Ciprofloxacin-Loaded PVP Foils and Nanofiber Mats {#sec2dot1-pharmaceutics-11-00527} --------------------------------------------------------------------- Transparent films and nanofiber mats were prepared as described by Contardi et al. \[[@B15-pharmaceutics-11-00527]\]. Briefly, films were produced using a solvent casting method starting from aqueous solutions of PVP (3% *w/v*) with a molecular weight (MW) of 360,000 g/mol (Sigma-Aldrich, Milan, Italy), monohydrochloride monohydrate free ciprofloxacin (≥ 98.0% measured by high performance liquid chromatography (HPLC), Sigma-Aldrich, Milan, Italy) and acetic acid (≥ 99.7%, Sigma-Aldrich, Milan, Italy). Three different initial quantities of ciprofloxacin (1.2, 30, and 60 mg) were combined with the polymer and dissolved in acetic acid 30% (*v/v*) to a final volume of 30 mL to reach different drug concentrations (2.2, 44, and 88 mmol). The solutions were cast on Petri dishes (diameter 8.75 cm) for 3 days under an aspirated hood under ambient conditions (16--20 °C and 40--50% r.h.). Then, the films were placed in a vacuum desiccator for 3 more days to complete the removal of excess acetic acid. The nanofiber mats were fabricated by using a vertical electrospinning set-up. Starting solutions of PVP, acetic acid (30% *v/v*), and ciprofloxacin were prepared (final volume 6.2 mL). A higher concentration of the polymer (25% *w/v*) was used with respect to the films to allow for the electrospinning process. Three different concentrations of ciprofloxacin (2.2, 44, and 88 mM) were also prepared for the nanofiber mats. Syringes with a stainless-steel needle (18 gauge) were filled with the three different solutions and connected to a syringe pump (NE-1000, New Era Pump Systems, Inc., New York, NY, USA) working at a constant flow rate (500 μL/h). The needles were clamped to the positive electrode of a high-voltage power supply generating 26 kV at a distance of 24 cm from an aluminum disk used as a collector (diameter of 8.75 cm). Only 2 mL of each solution were electrospun in order to obtain the same amount of ciprofloxacin on both the films and nanofibers. The morphology of foils and nanofiber mats was analyzed using SEM with a variable pressure JSM-649 microscope (JEOL, Milan, Italy) equipped with a tungsten thermionic electron source working in high vacuum mode and an acceleration voltage of 10 kV. The cross-section of the films was obtained by cutting slices with a UCS ultramicrotome (Leica Microsystems, Wetzlar, Germany) equipped with a glass knife. The specimens were coated with a 10-nm thick film of gold using the sputter coater 208 HR (Cressington Scientific Instruments, Watford, U.K.). 2.2. Skin Samples and the Creation of Superficial Wounds {#sec2dot2-pharmaceutics-11-00527} -------------------------------------------------------- Abdominal skin was obtained after getting informed consent from healthy donors undergoing plastic surgery. The study was conducted according to the Declaration of Helsinki guidelines and after approval by the Ethics Committee of the Charité---Universitätsmedizin Berlin (approval EA1/135/06, renewed on January 2018). Skin explants were used within 2--4 h after surgery. Subcutaneous fat tissue was partially removed, keeping a layer of approximately 5 mm and skin pieces (1.5 × 1.5 cm) were stretched and fixed on a Styrofoam block covered with Parafilm (Bemis Company, Neenah, WI, USA) using needles. The surface of the ex vivo skin (free of injuries or redness) was cleaned with saline solution (0.9% NaCl). The epidermis was then removed with a ball-shaped milling cutter 6 mm in size (No. 28725, Proxxon, Föhren, Germany) mounted on a micro motor handpiece (Marathon N7, TPC Advanced Technology, Inc. Diamond Bar, CA, USA) and rotating at 16,000 rpm. In this way, superficial wounds of approximately 5 mm in diameter were produced \[[@B21-pharmaceutics-11-00527]\]. 2.3. Drug Penetration Kinetics {#sec2dot3-pharmaceutics-11-00527} ------------------------------ Using a punch biopsy cutter, discs of 8 mm in diameter were cut out from foils, as well as nanofiber mats containing 44 mmol ciprofloxacin, so that each disc contained approximately 250 µg of ciprofloxacin. Wounds of 5 mm in diameter were produced on 1.5 × 1.5 cm pieces of skin. The disks were applied on the top of the wounds on skin blocs of 1.5 × 1.5 cm that were stretched on a Styrofoam block, placed in a humid chamber, and incubated at 37 °C, 5% CO~2~, and 100% humidity for different time points. Thereafter, non-penetrated material was removed with a cotton swab and the treated wound was removed from the rest of the tissue by means of an 8 mm punch biopsy tool. The tissue was chopped into small pieces and placed in 2-mL tubes filled with HCl (0.1 N, 1.5 mL) to extract the ciprofloxacin. The samples were gently mixed on a shaker for 24 h at room temperature. After centrifugation for 5 min at 300× *g*, the supernatant was collected and placed in triplicate in a 96-well plate (100 µL/well). Ciprofloxacin fluorescence (excitation wavelength: 275 nm, emission wavelength: 480 nm) was measured with an EnSpire^®^ Multimode plate reader (Perkin Elmer, Akron, OH, USA). A standard curve was prepared by dissolving ciprofloxacin in HCl (0.1 N) and preparing dilutions (0.5--10 µg/mL). The amount of penetrated drug was calculated on the basis of the standard curve. Results are presented as the means and standard deviations of three independent experiments. 2.4. Metabolic Activity of Skin Cells after the Topical Application of Ciprofloxacin on Ex Vivo Skin Wounds {#sec2dot4-pharmaceutics-11-00527} ----------------------------------------------------------------------------------------------------------- Wounds were treated with 8 mm discs from PVP-foils loaded with 44 mmol of drug. Wounds treated with NaCl solution (0.9%, 20 µL) served as negative controls, and wounds treated with 50 µg/cm^2^ PVP-coated silver nanoparticles (50 nm size, nanoComposix, San Diego, CA, USA) served as positive controls. Samples and controls were incubated in six-well plates for 20 h in 2 mL RPMI-1640 medium (Gibco, Darmstadt, Germany) without phenol red supplemented with fetal calf serum (FCS) (10%, Gibco, Darmstadt, Germany), glutamine (2mM, Gibco, Darmstadt, Germany), streptomycin (100 μg/mL, Gibco, Darmstadt, Germany), and penicillin (100 I.E./mL, Sigma-Aldrich, Hamburg, Germany). Thereafter, the old medium was replaced with fresh medium added with 500 µL of 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT) (Roche Diagnostic, Berlin, Germany). After 4 h, the medium from each sample and control was collected (3 × 100 µL) and placed in a 96-well microplate. The optical density at 450 nm was read with the EnSpire^®^ Multimode plate reader using 650 nm as the reference wavelength. Results are presented as a percentage with respect to the values from the control wounds. The averages and standard deviations of values from three independent experiments are reported. 2.5. Histological Analysis of Wound Tissue after the Topical Application of Ciprofloxacin-Loaded Foils {#sec2dot5-pharmaceutics-11-00527} ------------------------------------------------------------------------------------------------------ Wounds created on ex vivo skin were treated with ciprofloxacin in solution or loaded on foils and nanofiber mats (250 µg/wound). Wounds left untreated served as controls. Skin explant cultures were grown in supplemented RPMI-1640 medium as described in [Section 2.4](#sec2dot4-pharmaceutics-11-00527){ref-type="sec"}. Every 2 days, the medium (500 mL) was removed and fresh medium (500 mL) was added. After eight days of incubation, the skin was plunge frozen in Tissue Freezing Medium (Leica Microsystems, Wetzlar, Germany) and cryosections were prepared. Hematoxylin and eosin (H&E) staining was performed following the manufacturer's manual (Roth, Karlsruhe, Germany) and images were taken using optical microscopy with an Olympus IX 50 (OLYMPUS, Hamburg, Germany). 2.6. Bacteria Inoculation and Characterization of the PAO1 Wound Infection {#sec2dot6-pharmaceutics-11-00527} -------------------------------------------------------------------------- An overnight culture of the *P. aeruginosa* strain PAO1 (ATCC 15692) was diluted with tryptic soy broth. The suspension (5 µL, 1 × 10^7^ bacteria) was injected with a 10 µL syringe (26 gauge) with a tapered tip (SGE Analytical Science, Ringwood, VIC, Australia) from the edge of the wound into the dermis. As a control, sterile saline (5 µL, 0.9% NaCl) was injected in an uninfected control wound. Wound samples were incubated in a humidified chamber at 37 °C for 20 h. Biopsies (8 mm) were taken and fixed in a solution of formaldehyde (4%) and glutaraldehyde (0.5% in 50 mmol (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (HEPES) for 48 h at room temperature. Skin samples were then washed in HEPES (50 mmol), and dehydrated in 30, 50, 70, 90, 95, and 100% ethanol. Samples were infiltrated first with a LR White/ethanol solution (1:1, 10 min), and then with pure LR White (2 × 15 min). Successively, samples were transferred to polyallomer centrifuge tubes (5 × 20 mm, Beckman Coulter, Inc., Brea, CA, USA) containing LR White with an accelerator (5 µL/mL monomer). The centrifugation tubes were capped with a gelatin capsule and the samples were left to polymerize for 1 h on ice, and then at 60 °C overnight. Five hundred nanometer sections were cut using a ultramicrotome (EM UC7, Leica, Wetzlar, Germany), mounted on poly-[l]{.smallcaps}-lysine slides, and incubated for 10 min on an 80 °C thermo-plate. Sections were stained for 4 min with Richardson's stain (1% Azure II, 1% Methyleneblue, 1% Borax, Sigma-Aldrich, Hamburg, Germany, then washed with double distilled water (ddH~2~O), and imaged using a microscope (Axiophot, Carl Zeiss Microscopy GmbH, Jena, Germany). For analysis with a scanning electron microscope (SEM), after fixation, one of the sample duplicates was cut with a scalpel in order to reveal the skin profile. The samples were then dehydrated as described above, critical-point dried, mounted on aluminum stubs, sputter-coated with a 12-nm layer of gold-palladium, and finally examined with an SEM (ZEISS 1530 Gemini, Carl Zeiss Microscopy GmbH, Jena Germany) operating at 3 kV using the in-lens electron detector. Images have been cropped, adjusted for optimal brightness, and contrasted using Photoshop Lightroom (version 6.0, Adobe Systems, San Jose, CA, USA). 2.7. Antimicrobial Activity of Ciprofloxacin-Loaded Foils and Nanofiber Mats {#sec2dot7-pharmaceutics-11-00527} ---------------------------------------------------------------------------- Each tested setting was done in triplicates with a total of at least three runs. Three different dosages of ciprofloxacin were tested (11, 250, and 500 µg/wound). The treatments with PVP foils and nanofiber mats started either 1 h or 20 h after bacteria inoculation and lasted for 20 h. Thereafter, an 8 mm punch biopsy was used to collect the wound tissue, including some of the surrounding intact skin. The tissue was placed in a 1.5 mL microcentrifuge tube containing saline (0.2 mL) and homogenized for 3 min with a sterile steel pistil mounted on a digital overhead stirrer at 150 rpm (DSL, VELP Scientifica Srl, Usmate, MB, Italy). Thereafter, samples were sonicated for 10 min in an ultrasonic bath (BactoSonic1, Bandelin, Berlin, Germany) at 40 kHz using 200 Weff to detach the bacteria. Volumes of each sample were transferred to the wells of 96-well microplates and diluted in 1:10 steps (20 µL sample + 180 µL saline) by using a multichannel pipette. A volume (5 µL) of each well was plated on square tryptic soy agar plates. After incubation overnight at 37 °C, spotting areas with 5 to 50 colony forming units (CFU) were counted. Mean values of the triplicates were calculated and bacteria number per wound was calculated considering the used dilutions. In the diagrams, bacteria counts/wound are presented as the mean and standard deviation of three independent experiments. 2.8. Data Analysis {#sec2dot8-pharmaceutics-11-00527} ------------------ Data are reported as arithmetic means and standard deviations of at least three experiments. Calculations, data processing, and graphics were prepared with Excel 2018 (Microsoft, Redmond, WA, USA). 3. Results and Discussion {#sec3-pharmaceutics-11-00527} ========================= 3.1. Preparation and Characterization of Drug-Loaded Foils and Nanofiber Mats {#sec3dot1-pharmaceutics-11-00527} ----------------------------------------------------------------------------- Starting from aqueous solutions of PVP, acetic acid, and ciprofloxacin at different concentrations, transparent films and nanofiber mats were fabricated using the solvent casting and electrospinning methods, respectively ([Figure 1](#pharmaceutics-11-00527-f001){ref-type="fig"}). The final dry transparent foils had a thickness of 150--180 μm and the absence of crystal formation was verified using SEM ([Figure A1](#pharmaceutics-11-00527-f0A1){ref-type="fig"}, [Appendix A](#app1-pharmaceutics-11-00527){ref-type="app"}). As recently demonstrated, the obtained PVP-based foils represent suitable wound dressings due to their flexibility, adhesion, and resorption properties \[[@B19-pharmaceutics-11-00527],[@B28-pharmaceutics-11-00527]\]. The nanofiber had an average diameter of 360 ± 80 nm and did not show any defects or beads ([Figure A2](#pharmaceutics-11-00527-f0A2){ref-type="fig"}, [Appendix A](#app1-pharmaceutics-11-00527){ref-type="app"}). 3.2. Foils and Nanofiber Mats Had Different Drug Delivery Profiles {#sec3dot2-pharmaceutics-11-00527} ------------------------------------------------------------------ Foils and nanofiber mats (discs of 8 mm in diameter) were applied on the top of the 5 mm in diameter wounds created on the 1.5 × 1.5 cm ex vivo skin tissue to provide an applied drug dosage of approximately 250 µg/wound. The skin explants were incubated in humidified chambers in an incubator with 100% humidity. This maintained the wetness of the skin blocks, which favored the dissolution of foils and nanofibers. After different incubation times, the non-penetrated material was removed, 8 mm biopsies were taken, and extracts were prepared to measure the amount of penetrated drug ([Figure 2](#pharmaceutics-11-00527-f002){ref-type="fig"}). Both penetration kinetics could be well-fitted to a second-degree polynomial trend line. Amounts ranging between 12 and 18 µg of ciprofloxacin/wound were detected for foils, whereas amounts ranging between 5 and 10 µg/wound were measured for nanofiber discs. Foils delivered high amounts of ciprofloxacin very quickly (approximately 12 µg after only 2 h), whereas nanofiber mats reached similar concentrations later, after 6 h. Interestingly, foils maintained the reached drug concentration in the wound over 24 h. On the contrary, in samples treated with nanofiber mats, the maximal concentration was maintained for a shorter time range (6--16 h). The decreased amounts of drug in the wound tissue observed for nanofiber mats ([Figure 2](#pharmaceutics-11-00527-f002){ref-type="fig"}b) were due to drug diffusion to the neighboring tissue ([Figure A3](#pharmaceutics-11-00527-f0A3){ref-type="fig"}, [Appendix B](#app2-pharmaceutics-11-00527){ref-type="app"}). We suppose that for nanofiber mats, a lower amount of drug penetrated the wound tissue and that, due to diffusion to the nearby skin tissue, a reduction of local drug concentration in the wound was measured after 24 h. On the contrary, foils delivered a higher amount of drug to the wound and could saturate both the wound and the nearby tissue. In a previous study, the dissolution rate and drug release for both materials was investigated in a phosphate buffer and on a mice full-thickness wound model \[[@B19-pharmaceutics-11-00527]\]. Results showed that nanofibers dissolved and released the loaded drug very quickly, within the first two hours, whereas foils took longer. The fast dissolution of nanofibers might result in a high local concentration causing drug precipitation or crystallization, which in turn can hinder drug penetration. Another important aspect to be considered is that the investigated PVP-foils contained higher amounts (1.4%) of residual acetic acid than the nanofiber mats (0.3%) \[[@B19-pharmaceutics-11-00527]\], which might also have had an influence on drug solubility and penetration. In summary, drug penetration experiments using the ex vivo wound model showed that both drug delivery systems could deliver the loaded antibiotic to the wound tissue in a controlled manner. The ability of PVP-foils to maintain a high drug concentration in the wound over 24 h is of significance, especially for the treatment of biofilm infections, which need high antibiotic concentrations to be resolved. On the other side, the more controlled delivery profile of the nanofiber mats might be more useful for the prevention of wound infections. 3.3. Local Toxicity of Ciprofloxacin in Full-Thickness Ex Vivo Skin {#sec3dot3-pharmaceutics-11-00527} ------------------------------------------------------------------- The wound model was also used to test the potential cytotoxicity of the drug and investigated materials after topical application. For these experiments, tissue blocks were cultured in supplemented RPMI 1640 medium at the air--liquid interface. After 20 h of treatment, tissue viability was tested by means of an XTT test ([Figure 3](#pharmaceutics-11-00527-f003){ref-type="fig"}a). Many cells in skin explants were still metabolically active and an evident formation of the red formazan product was observed. PVP foils did not show any toxicity. This was somehow expected with PVP being a well-tolerated, FDA-approved polymer with many uses, such as a food additive, binder in tablets, and plasma volume expander \[[@B29-pharmaceutics-11-00527]\]. Also, foils prepared from a PVP solution in 30% acetic acid resulted in having no effect on skin viability due to the fact that only residual amounts of acetic acid molecules remained within the PVP polymers after solvent evaporation during the preparation process \[[@B19-pharmaceutics-11-00527]\]. Finally, the PVP foils loaded with the three concentrations of ciprofloxacin also resulted in having no influence on the overall viability of skin cells after 20 h of incubation. Wounds treated with silver nanoparticles, which release toxic silver ions, served as positive controls. The evident reduction of formazan formation was indicative of silver toxicity toward wound cells. Thus, even if this test gave no information about the type of cells being affected by the tested substances, it was a useful method to detect overall acute toxic effects. Previous studies have shown that ciprofloxacin has toxic effects on fibroblasts \[[@B30-pharmaceutics-11-00527]\] and keratinocytes \[[@B31-pharmaceutics-11-00527]\]. Toxicity was shown to be time- and concentration-dependent. For this reason, we tested the effect of ciprofloxacin delivered by the investigated foils with regard to re-epithelialization. After 9 days of incubation in supplemented RPMI-1640 medium, a partial re-epithelialization had occurred in controls, with the re-growing of a keratinocyte layer on the edges and in the wound bed ([Figure 3](#pharmaceutics-11-00527-f003){ref-type="fig"}b,d,e). Cell nuclei and the collagen network presented a normal morphology in samples treated with empty PVP foils with residual acetic acid. Only a delayed re-epithelialization was observed, probably due to oxygen deprivation (data not shown). On the contrary, in wounds treated with ciprofloxacin, only a thin epithelial layer was observed. Cells' nuclei appeared small with an irregular morphology typical of necrotic cells and the collagen matrix was less organized than in the controls ([Figure 3](#pharmaceutics-11-00527-f003){ref-type="fig"}c,f,g). This effect was visible in all samples treated with ciprofloxacin at different concentration and was independent of the formulation (data not shown). In recent years, it has been recognized that fluoroquinolones, despite being well tolerated by a broad portion of patients, can have rare and very disabling side effects like a tendon rupture and irreversible nerve damage \[[@B13-pharmaceutics-11-00527]\]. There are several hypotheses on the mechanism of toxicity and on the reason why some persons develop these side effects more than others do. One theory is that there might be a gene variant responsible for a disrupted quinolone metabolism. Our results show that, despite ciprofloxacin inducing no changes in whole skin metabolic activity, it negatively influenced the re-epithelialization of ex vivo wounds. Even if this side effect is tolerable in the case of severe wound infections, these results stress the importance of a more local and controlled release of drugs like ciprofloxacin that have narrow therapeutic windows. 3.4. Ciprofloxacin-Loaded Foils and Nanofiber Mats Efficiently Reduced P. aeruginosa Infections {#sec3dot4-pharmaceutics-11-00527} ----------------------------------------------------------------------------------------------- Next, we investigated the antimicrobial efficacy of the investigated drug delivery systems using the ex vivo skin infection model ([Figure 4](#pharmaceutics-11-00527-f004){ref-type="fig"}). The PAO1 strain was used, which possesses several proteolytic enzymes, among those collagenases, and was shown to build biofilm-like infections on ex vivo skin \[[@B21-pharmaceutics-11-00527]\]. Bacteria inoculated in the wounds grew from 1 × 10^7^ to 1 × 10^9^ bacteria per wound after 20 h of incubation. Macroscopically, the surface of the PAO1-infected wound appeared shiny and yellowish ([Figure 4](#pharmaceutics-11-00527-f004){ref-type="fig"}c). Scanning electron microscopy images of the wound surface showed bacteria conglomerates typical of a biofilm. These three-dimensional structures are made of bacteria, extracellular materials, and probably of degraded collagen material ([Figure 4](#pharmaceutics-11-00527-f004){ref-type="fig"}d,e). The microscopic pictures of H&E-stained skin sections ([Figure 4](#pharmaceutics-11-00527-f004){ref-type="fig"}f) revealed an approximately 10-µm-thick bacteria film on the surface of the wound, but also groups of bacteria deep in the wound tissue (circles). The scanning electron microscopic analysis of the wound profile ([Figure 4](#pharmaceutics-11-00527-f004){ref-type="fig"}g,h) confirmed the formation of a bacterial biofilm within the superficial wound tissue, as well as the presence of small agglomerates or scattered bacteria deep in the wound. In the ex vivo model, most of bacteria grew on the wound's surface, despite the fact that they were inoculated deep in the connective tissue. *P. aeruginosa* can grow in both aerobic and anaerobic conditions. However, the less favorable conditions and shortage of oxygen in the deeper wound layers resulted in a slower proliferation of bacteria. Nevertheless, these small colonies can be the reason for recurrent infections if not eradicated during a treatment. For this reason, the penetration of adequate drug concentrations deep in the wound tissue and the eradication of all bacterial colonies is a crucial factor for a successful therapy. *Pseudomonas aeruginosa* has already been cultured on ex vivo skin \[[@B24-pharmaceutics-11-00527],[@B32-pharmaceutics-11-00527],[@B33-pharmaceutics-11-00527],[@B34-pharmaceutics-11-00527],[@B35-pharmaceutics-11-00527]\]. In this study, we used the ex vivo wound infection model as a three dimensional set up to test the capacity of ciprofloxacin-loaded PVP drug delivery systems to eradicate bacteria located not only on the surface, but also deep in the wound's tissue. The investigated foils and nanofiber mats were loaded with three different concentrations of ciprofloxacin, corresponding to final doses of 11, 250, and 500 µg per wound. The treatments were applied on the top of the wounds 1 h or 20 h after bacteria inoculation and kept for a further 20 h. After 1 h of inoculation, bacteria were still planktonic and thus less resistant to antibiotics. After 20 h of inoculation, the number of bacteria had increased and bacteria were organized in a biofilm, which was more difficult to treat. Samples treated 1 h after bacteria inoculation served to test the drug efficacy toward a moderate infection with approximately 1 × 10^7^ planktonic bacteria, whereas samples treated 20 h after PAO1 inoculation served to test the efficacy of foils and nanofiber mats on a severe infection with approximately 1 × 10^9^ bacteria primarily organized into a biofilm. The experiment terminated after a total of 40 h. Macroscopically, a thick, yellow biofilm was visible on the surface of untreated wounds, whereas no bacterial film was visible on the treated samples ([Figure 5](#pharmaceutics-11-00527-f005){ref-type="fig"}a). Drug crystals were visible on the wound edges, where higher concentrations of the drug were reached after the foils or nanofibers had dissolved. When applied 1 h after bacterial inoculation, both foils and nanofiber mats resulted in a complete eradication of the bacterial infection independent from the amount of drug that was applied. This result shows the potential of these drug delivery materials for the prevention of wound infections. In particular, nanofibers should be tailored to obtain a more controlled drug delivery in order to have effective antimicrobial concentrations and limit adverse side effects. In contrast, the 20-h-biofilm infection was more difficult to treat. PVP foils had the best performance ([Figure 5](#pharmaceutics-11-00527-f005){ref-type="fig"}b). Foils loaded with 500 µg ciprofloxacin achieved a complete eradication of the bacteria, and the two lower concentrations resulted in a 3-log reduction of bacteria concentration with respect to the untreated controls. Nanofiber mats had a lower antimicrobial activity with a maximal 3-log reduction achieved by the highest concentration. No effect on bacterial growth was measured for plain PVP foils or foils with residual acetic acid ([Figure 5](#pharmaceutics-11-00527-f005){ref-type="fig"}d). The most striking result was the complete eradication of the biofilm infection achieved with the 500-µg PVP foils. Foils have a higher percentage of residual acetic acid (1.4%) than nanofiber mats (0.1%) \[[@B19-pharmaceutics-11-00527]\]. Because acetic acid also has antimicrobial activity, it might have acted synergistically with ciprofloxacin. Nevertheless, the superior efficacy of the foils compared to the nanofiber mats correlated well with the delivery kinetics. Foils delivered higher amounts of the drug and over a longer time than nanofibers did. Phillips et al. assessed the efficacy of several commercially available treatments in a porcine ex vivo wound model and found that time-release silver gel and cadexomer iodine dressings were the most effective in reducing a mature biofilm with a reduction between 5 and 7 logs out of a 7-log total \[[@B24-pharmaceutics-11-00527]\]. Using an in vivo mouse model, Roy et al. found that ciprofloxacin-loaded keratin-based hydrogels with a sustained drug release profile could reduce the amount of *P. aeruginosa* in the wound bed by 99.9% without interfering with the key processes of wound healing \[[@B36-pharmaceutics-11-00527]\]. Thus, these and our results underline the importance of drug delivery systems with a sustained release profile for the efficient treatment of *P. aeruginosa* wound infections. 4. Conclusions {#sec4-pharmaceutics-11-00527} ============== In this study, we used full-thickness ex vivo human skin and a wound infection model to investigate the efficacy and tolerability of ciprofloxacin-loaded PVP foils and nanofiber mats. The model allowed us to test the antimicrobial efficacy of these materials and to correlate it to their delivery properties. Ciprofloxacin was representative of a poorly water-soluble drug to be loaded in PVP-based drug delivery systems. The use of a solubilizer (acetic acid) increased the loading capacity and drug delivery properties of nanofibers and foils. This was in turn a crucial point for the accomplishment of high local drug concentrations that were required for the successful prevention and eradication of the PAO1 biofilm infections. This strategy can be used to load PVP-based delivery systems with different types of drugs or disinfectants, including high molecular weight moieties. Besides their sustained drug release properties, such matrixes allow for a more precise dosing of the active ingredient in comparison to ointments or creams. Foils and nanofiber mats are flexible and can be easily applied in less accessible skin areas, e.g., the lower back or between the toes, which is an advantage, especially for old and disabled patients. Thus, we conclude that the combination of drug delivery systems and solubilizing agents is a promising strategy to create attractive new pharmaceutical forms for topical drug delivery to treat or prevent wound infections. Conceptualization, F.R., M.C., I.S.B. and C.S.; methodology, F.R. and C.S.; formal analysis, F.R., J.J. and M.C.; investigation, M.C., J.J. and C.S.; resources, U.B.-P., A.V. and C.S.; writing---original draft preparation, F.R.; writing---review and editing, F.R., M.C., I.S.B., U.B.-P., A.V. and C.S.; visualization, F.R. and M.C.; funding acquisition, F.R. and C.S. This research was partially funded by the German Federal Ministry of Economics (grants: KF2928204MD4 and KF2088119MD4). We acknowledge support from the German Research Foundation (DFG) and the Open Access Publication Fund of Charité---Universitätsmedizin Berlin. The authors declare no conflict of interest. Morphological Analysis of Films and Fibers ========================================== To verify the absence of crystals in the drug-loaded membranes, SEM analysis was performed. In [Figure A1](#pharmaceutics-11-00527-f0A1){ref-type="fig"}, representative images of the top-view and cross-section of ciprofloxacin-loaded foils (88 mmol) are reported. {#pharmaceutics-11-00527-f0A1} [Figure A2](#pharmaceutics-11-00527-f0A2){ref-type="fig"} shows representative SEM images of the ciprofloxacin-loaded nanofibers (2.2, 44, and 88 mmol). The diameter of the fibers was determined using ImageJ software. Approximately 100 measurements were taken to obtain the diameter distribution of each type of fiber mat. {#pharmaceutics-11-00527-f0A2} Drug Diffusion to the Surrounding Skin Tissue ============================================= In order to estimate the amount of drug that diffused to the nearby tissue, skin blocks were treated with ciprofloxacin loaded on PVP-foils or a solution of the drug in water. The wound tissue, as well as the surrounding tissue, were extracted and analyzed ([Figure A3](#pharmaceutics-11-00527-f0A3){ref-type="fig"}). The kinetics showed that the drug accumulated first in the wound tissue and diffused successively to the nearby tissue. After approximately 16 h, similar concentrations were reached in both areas of the ex vivo tissue block. {#pharmaceutics-11-00527-f0A3} {#pharmaceutics-11-00527-f001} {#pharmaceutics-11-00527-f002} {#pharmaceutics-11-00527-f003} {#pharmaceutics-11-00527-f004} {#pharmaceutics-11-00527-f005}

1. Introduction {#sec1} =============== Leukemia is the most common childhood cancer, although overall incidence is rare. Within this population, acute lymphoblastic leukemia (ALL) occurs approximately five times more frequently than acute myelogenous leukemia (AML) and accounts for approximately 78% of all childhood leukemia diagnoses \[[@B1]\]. In Brazil, the National Cancer Institute (INCA) estimated 9,370 new cases of leukemia in 2014, with the highest age incidence of 1--4 years \[[@B2]\]. The specific biological and molecular mechanisms that account for the aggressiveness and poor therapy response of some ALL cases remain to be elucidated. Once chemokines and their receptors have been discovered as essential and selective mediators in leukocyte migration to inflammatory sites and to secondary lymphoid organs, it is reasonable that they also play a critical role in tumor initiation, promotion, and progression \[[@B3], [@B4]\]. Moreover, updated research indicates that cancer cells subvert the normal chemokine system, and these molecules and their receptors become important constituents of the tumor microenvironment with very different ways to exert tumor-promoting roles \[[@B5]\]. The CC chemokine receptor 5 (CCR5) belongs to the trimeric guanine nucleotide-binding-protein-coupled seven-transmembrane receptor superfamily, which comprises the largest superfamily of human proteins \[[@B6]\]. It exerts its activity via G protein and binds to the chemokines RANTES (CCL5), MIP-1*α* (CCL3), and MIP-1*β* (CCL4) \[[@B7]\]. This receptor is involved in the chemotaxis of leucocytes to inflammation sites \[[@B8]\] and plays important function in the recruitment of macrophages, T cells, and monocytes \[[@B9]\]. The importance of CCR5 for immune response is dependent on the type of stimuli; moreover, in some cases, compensating mechanisms override the absence of CCR5 expression and function. Noteworthy, CCR5 may exert a far more important role in the immune response than in immune cells traffic regulation \[[@B10]\]. It has been shown that CCR5 expression in stromal cells as well as hematopoietic cells contributes to tumor metastasis. For instance, CCR5 is involved in chondrosarcomas metastasization \[[@B11]\] and oral cancer cells migration \[[@B12]\]. Its expression correlates with multiple myeloma cell growth, bone marrow homing, and osteolysis \[[@B13]\]. Aster and colleagues \[[@B14]\] showed that T-cell ALL is a disease primarily caused by aberrant activation of the NOTCH1 signaling pathway. In this context, expression and function of important chemokine receptors, such as CCR5 and CCR9, are partially controlled by the oncogenic NOTCH1 isoform in T-cell ALL, regulating blast malignant properties and localization of extramedullary infiltrations \[[@B15]\]. Additionally, CCR5 has been related to play a key role in metastasis of aggressive NK-cells leukemia to the liver of patients, contributing to hepatosplenomegaly and hepatic failure \[[@B16], [@B17]\]. Also, Davies et al. \[[@B18]\] showed that the G allele of rs1799987 polymorphism in*CCR5* was associated with more favorable minimal residual disease status than the A allele when comparing "best" and "worst" risk groups of B-cell ALL, adjusted for prognostic features. Considering the lower activity of*CCR5* promoter in the presence of this polymorphism \[[@B19]\], this evidence indicated that both acquired and host genetics influence response to cancer therapy. Thus, it is plausible that CCR5 might play a role in ALL pathogenesis and prognosis. It is known that the polymorphism rs333 in*CCR5*, a common 32-base pair deletion (Δ32), causes truncation and loss of this receptor on lymphoid cell surface, with complete retention in the endoplasmic reticulum within homozygous or diminished expression in heterozygous genotype \[[@B20]\]. The*CCR5* studies have demonstrated the importance of Δ32 mutation, particularly in the susceptibility to HIV infection \[[@B21]\], since CCR5 is a coreceptor in the primary stage of infection that is essential for the AIDS onset \[[@B22]\]. Our research group has reported polymorphic allelic variants related to the immune system and tumor development in different cancer types. Nevertheless, there are no data relating*CCR5*/Δ32 polymorphism to ALL population. In this context, the present work analyzed rs333 polymorphism of*CCR5* in ALL patients from the southern region of Brazil. 2. Materials and Methods {#sec2} ======================== 2.1. Human Subjects {#sec2.1} ------------------- Following approval from the Human Ethics Committee of the State University of Londrina, Paraná, Brazil (CAAE number 171231134.0000.5231), inclusion of the individuals to the study was conditioned by an obtained written informed consent form from parents regarding the use of their children and adolescents blood samples. Seventy-nine ALL patients were enrolled and diagnostic criteria were based on the guidelines proposed by Hematology Department of the University Hospital of Londrina. Recurrence risk status of ALL patients was evaluated through the GBTLI Protocol (Brazilian Group of Childhood Leukemia Treatment Protocol-99) which is based on the Cancer Therapy Evaluation Program, proposed by the National Cancer Institute, and takes into account age at diagnosis, leukocyte count, immunophenotyping, involvement of tissues other than bone marrow, and responsiveness to treatment. The control group is comprised of 80 healthy individuals free of neoplasia, matched by age and gender. 2.2. Genomic DNA Extraction {#sec2.2} --------------------------- Genomic DNA was extracted from whole blood by Biopur Mini Spin Plus Kit (Biometrix Diagnostica, Curitiba, Brazil), according to the manufacturer\'s instructions. DNA was eluted in 50 *μ*L of milliQ water and quantified by NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, USA) at a wavelength of 260/280 nm. Final preparation was stored at −20°C and used as templates in polymerase chain reactions (PCR). 2.3. Optimization of PCR for CCR5 {#sec2.3} --------------------------------- The method of genotyping (rs333) was optimized in the Laboratory of Study and Application of DNA Polymorphisms of the State University of Londrina using genomic DNA and specific primers for*CCR5*:*Primer sense*: 5′ ACC AGA TCT CAA AAA GAA 3′ and*Primer antisense*: 5′ CAT GAT GGT GAA GAT AAG CCT CA 3′ (GenBank accession AF009962). Genotyping of CCR5 was determined by PCR-SSP. The samples were amplified using 1.25 units of Taq polymerase (Invitrogen, Carlsbad, USA) in a Mastercycler Gradient Thermal Cycler (Eppendorf, Hamburg, Germany). PCR conditions were: denaturation step at 94°C for 5 min, 35 cycles of 1 min at 94°C, 1 min at 58°C and 1 min at 72°C, and 10 min of elongation at 72°C. PCR products (225 bp or 193 bp) were analyzed on polyacrylamide gel (10%), stained with silver nitrate (AgNO~3~). 2.4. Statistical Analysis {#sec2.4} ------------------------- Contingency tables and Fisher\'s exact test were used to calculate differences in genotype distributions and allele frequencies. *P* \< 0.05 was considered to indicate a statistically significant difference. Goodness of fit of Hardy-Weinberg equilibrium was tested by calculating the expected frequencies of each genotype and comparing them with the observed value using a chi-square test. 3. Results {#sec3} ========== The distribution of*CCR5* alleles in both ALL and control groups was in accordance with the assumption of Hardy-Weinberg equilibrium (*P* \> 0,05). The mean age of controls and ALL patients was 10.8 years ± 5.65 and 8.7 years ± 6.20, respectively, all of whom were predominantly from Caucasoid population, due to European colonization. The patients and controls were matched for sex and gender, although there was a modest higher frequency of males in ALL group (53.16%) than in controls (48.75%). Fifty (63.29%) ALL patients were classified in high recurrence risk group and 29 (36.71%) in low recurrence risk group. The possible observed genotypes for*CCR5* rs333 polymorphism are shown in [Figure 1](#fig1){ref-type="fig"}. Genotyping results did not show any homozygous individuals for Δ32 deletion in both groups. The heterozygous CCR5/Δ32 genotypes were observed in 8.75% (*n* = 7) of controls and 7.5% (*n* = 5) of ALL patients. To determine if there was a statistically significant increased risk of ALL development related to the *CCR*5 genotypes, we conducted logistic regression analysis ([Table 1](#tab1){ref-type="table"}), which showed that individuals with one copy of Δ32 variant allele did not exhibit ALL-associated risk. No statistical difference was observed when allelic frequency of Δ32 at rs333 in ALL patients was associated with control subjects (OR = 0.71; 95%CI = 0.22--2.30; *P* = 0.77). In addition, we compared the*CCR5* genotype distribution in ALL patients classified in high risk or low risk, according to recurrence status. From five (7.5%) ALL Δ32 carriers, three were classified as high-risk patients. However, association study between both recurrence statuses did not reach statistical significance. The*CCR5* genotypes distribution in ALL patients and controls was stratified by gender. Subgroup analyses revealed that the effect of gender was not significantly different among*CCR5* genotypes (female ALL versus female control, OR = 0.52; 95%CI = 0.09--3.07, and male ALL versus male control, OR = 0.97; 95%CI = 0.18--5.14). 4. Discussion {#sec4} ============= Chemokines and their receptors are key regulators of immune activities and in parallel could play conflicting roles in malignancy. While most combinations of these receptors and chemokines are active in cancer, many findings in the field have emphasized the chemokine CCL5 and its cognate receptor CCR5 \[[@B23]\]. The gene variants of the chemokine and chemokine receptor genes associated with inflammation may be involved in cancer initiation and progression \[[@B24]\]. Considering the remarkable difference in CCR5/Δ32 allele frequency among worldwide populations, we aimed to survey the genetic variations in*CCR5* in ALL patients and control individuals. The patients\' age in this study was the expected for ALL, which is frequent in children and younger patients. Moreover, as previously mentioned, both sample groups were composed predominantly of Caucasian individuals from Southern Brazil. However, due to high degree of miscegenation of Brazilian population and the demand to use genetic markers for correct characterization of individuals \[[@B25], [@B26]\] in our country, these data have not been explored in relation to the variants analyzed. Chemokines and chemokine receptors are among factors that may influence ALL progression and localization \[[@B15]\]. A 32-base pair nucleic acid deletion in*CCR5* exists and causes a frameshift mutation in the amino acids comprising the second extracellular loop. This deletion leads to premature truncation of the protein, disabling its ability to translocate to the membrane, impairing expression and ligand binding at the cell surface, and causing membrane receptor deficiency that may influence leukocyte trafficking \[[@B27]\]. Based on this, we hypothesized that the ability of CCR5 to bind its ligands and signal recruitment of pathogenic T cells into target tissues may be impaired, thus imparting ALL protection. Although there were no differences in the frequency of CCR5/CCR5 and CCR5/Δ32 genotypes between patient and control groups, the variant genotype had no effect on the ALL susceptibility. These results corroborated with studies in different disorders, such as leishmaniasis, breast, laryngeal, thyroid, and brain carcinoma, which also identified no differences in the frequency of these alleles among healthy subjects and patients of Southern Brazilian population \[[@B28]--[@B30]\] and worldwide \[[@B31]--[@B33]\]. Intensive multiagent chemotherapy regimens and introduction of risk-stratified therapy have substantially improved cure rates for children with ALL. Current risk allocation schemas are imperfect, as some children are classified as lower-risk and treated with less intensive therapy relapse, while others deemed higher-risk are probably overtreated \[[@B34]\]. In this context, genetic polymorphisms in chemokine receptors could predict outcome and be considered an independent risk factor to stratify and allocate therapy in ALL. More than half of the patients with ALL were classified in higher-risk group, according to the clinical and laboratorial findings at diagnosis, as defined by GBTLI LLA-99 protocol \[[@B2]\]. When the genotype data were analyzed for stratified group of ALL, the results indicated that the presence of Δ32 did not influence this clinical parameter. Similarly, a recent study conducted by our research group has not found association among tumor suppressor*TP53* and chemokine*CXCL12* polymorphisms and ALL recurrence risk status \[[@B35]\]. 5. Conclusion {#sec5} ============= The comprehension about cellular and molecular mechanisms of ALL is critically important for the development of new approaches to hematological neoplasia treatment. Although any association of rs333 polymorphism of*CCR5* was verified, we believe that the current research must lead to a better definition of the host-tumor relationship particularly with respect to immunologic response and interrelation of CCR5 and ALL development. Given the sample size of the present case-control association study, strong conclusions are not possible; however, future investigation involving much larger cases may determine the absence of clinical implications for CCR5/Δ32 alleles in relation to ALL pathogenesis. The authors would like to acknowledge the volunteers who made this study possible, the University Hospital of Londrina, and Londrina Cancer Institute, PR, Brazil, for their collaboration. This study was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Fundação Araucária do Paraná, Secretaria da Ciência, Tecnologia e Ensino Superior (SETI), Fundo Estadual para a Infância e Adolescência (FIA/PR) e Secretaria da Família e Desenvolvimento Social (SEDS), and Coordination of Undergraduate Studies of Londrina State University (PROPPG-UEL). Conflict of Interests ===================== The authors declare that there is no conflict of interests regarding the publication of this paper. {#fig1} ###### Genotype distribution of *CCR5* rs333 polymorphism and recurrence risk status analysis in ALL and control groups.   Genotypes OR 95% CI *P*\* ----------------- ------------- ----------- -------- ------------ ------ Control (80) 73 (91.25%) 7 (8.75%) 0.7 0.21--2.32 0.76 ALL (79) 74 (92.5%) 5 (7.5%)  High risk (50) 47 (94%) 3 (6%) 0.86 0.13--5.48 1.00  Low risk (29) 27 (93.1%) 2 (6.9%)  \*Fisher\'s exact test, *P* \> 0.05. OR: odds ratio; CI: confidence interval. [^1]: Academic Editor: Helen A. Papadaki

**Core tip:** Many studies investigating postoperative pancreatic fistula after gastrectomy, including measurement of drain amylase content (D-AMY) as a predictive factor. This article reviews previous studies and looks to the future of measuring D-AMY in patients after gastrectomy. Several studies have reported that the high D-AMY on postoperative day 1 or day 3 was an independent risk factor for postoperative pancreatic fistula. However, issues for clinical use remain unresolved, including the ideal timing of measurement, optimal cut-off value and intervention after prediction. Prospective clinical trials might be indispensable for D-AMY to become a common marker in clinical practice. INTRODUCTION ============ Gastrectomy with radical lymph node dissection, especially peri-pancreatic lymph node dissection, is the mainstay for resectable gastric cancers. Pancreas-related complications, especially postoperative pancreatic fistula (POPF), is one of the most common postoperative complications, and can sometimes lead to serious results, including intra-abdominal abscess, subsequent sepsis, and intraperitoneal bleeding, which usually require prolongation of hospitalization\[[@B1],[@B2]\]. Many studies predicting POPF after gastrectomy have been investigated and age, operation time, body mass index, total gastrectomy, splenectomy, anatomical position of the pancreas, and high value of drain amylase content (D-AMY) have been reported as substantial predictive factors\[[@B2]-[@B13]\]. Among those, D-AMY, which is the measurement of the amylase content in drained abdominal fluids taken through an indwelling intra-abdominal drain, is promising because it can be measured objectively regardless of the patient's preoperative condition, type of surgical procedure, and surgeon's skill\[[@B14]\]. However, several problems remain unsolved precluding the implication in common clinical practice. This article reviews previous reports and looks to the future of measuring D-AMY to predict POPF in gastric cancer surgery. MECHANISM OF PANCREATIC FLUID LEAKAGE IN GASTRIC CANCER SURGERY =============================================================== In gastric cancer surgery, the pancreatic duct is not usually transected, and the mechanism of pancreatic fluid leakage after surgery is different from that of pancreatectomy. Mainly, three mechanisms of pancreatic fluid leakage in gastric cancer surgery are reported, presumably caused by the operator and assistant surgeons. The operator can injure the surface of the pancreas by parenchymal and/or thermal injury during the dissection of the suprapancreatic lymph nodes\[[@B15]\]. Second, the assistant could compress and retract the pancreas to achieve a good view of the suprapancreatic area during suprapancreatic lymph node dissection, so called "blunt injury"\[[@B16]\]. Ida et al\[[@B16]\] conducted animal experiments using pigs and reported that blunt injury causes pancreatic necrosis and inflammatory cell infiltration, and the value of amylase content around the pancreas increases 2-4 h after the procedure. Another mechanism is that pancreatic tail mobilization during combined splenectomy or splenectomy with distal pancreatectomy can damage the pancreatic parenchyma, resulting in pancreatic fluid leakage\[[@B3]\]. However, indications for such extended lymphadenectomy for gastric cancer has recently become limited\[[@B17]\]. In this regard, pancreatic fluid leakage involving gastric cancer surgery could mostly be minor leakage, that tends to be subsided spontaneously without clinically relevant pancreas-related postoperative complications. THE BEGINNING OF STUDIES PREDICTING POPF USING DRAIN AMYLASE CONTENT IN PANCREATIC SURGERY ========================================================================================== In pancreatic surgery, the measurement of D-AMY is also used in the diagnostic criteria of POPF\[[@B18]\]. In general, pancreatectomy is recognized as a highly invasive surgery and is associated with a mortality of approximately 5% and a morbidity of 30%-60%\[[@B19]\]. Approximately 16% of patients develop POPF, making it one of the common complications of pancreatectomy\[[@B20]\]. The clinical stratification of POPFs was established by the International Study Group on Pancreatic Fistula (ISGPF) definition in 2005\[[@B18]\]. The presence of POPFs can be determined on postoperative day (POD) 3 by the amylase content in the drained fluid; therefore, the measurement of D-AMY has become common in clinical practice in the field of pancreatic surgery. After pancreatectomy, pancreatic fluid leakage is caused by a disruption of the main pancreatic duct, and D-AMY directly reflects pancreatic fluid leakage\[[@B21],[@B22]\]; therefore, the measurement of D-AMY is a reasonable prediction tool for pancreas-related complications. This concept was later extended to gastric cancer surgery; however, the mechanisms responsible for pancreatic fluid leakage are supposed to differ between pancreatectomy in which the main pancreatic duct is transected, and gastrectomy, which causes some problems. The source of pancreatic fluid leakage after gastrectomy is the seepage of pancreatic juice from the parenchymal damage of the pancreas and blunt damage by compression or retraction\[[@B15],[@B16]\]. Even if the value of D-AMY is high on POD 3, this minor pancreatic leakage seems to subside spontaneously without proceeding to clinical fistula formation\[[@B3]\]. Thus, the establishment of a gastric cancer surgery-specific definition and prediction tool for POPF is desirable. DEFINITION OF POPF IN GASTRIC CANCER SURGERY ============================================ Despite clinical importance, POPF had not been uniformly defined until 2005 when the ISGPF established the definition based on the clinical impact of POPF-related complications\[[@B18]\], and it has been well accepted in the pancreatic surgery community. The stratifications are as follows: Grade A, pancreatic fistulas with no clinical impact, although D-AMY on or after POD 3 is three times more than the upper normal serum amylase level; grade B requires a change in management or adjustment in the clinical pathway; and grade C requires a major change in clinical management and aggressive clinical intervention. This ISGPF classification is sometimes applied in the gastric cancer surgery community. However, validation of applying this definition to POPF following gastrectomy still remains unclear. As described in the previous section, the mechanisms of pancreatic fluid leakage in gastric cancer surgery are different from those in pancreatic surgery, and are shown in Figure [1](#F1){ref-type="fig"}; POPF of ISGPF grade A is not necessarily clinically significant; in other words, POPF with no clinical impact need not be defined in gastric cancer surgery. In addition, almost all POPFs in gastric cancer surgery are classified as ISGPF grade B, and ISGPF grade C is very rare. {#F1} As another definition of POPF, the Clavien-Dindo classification has been adopted, which is a comprehensive evaluation of postoperative complications and has gained widespread acceptance\[[@B23],[@B24]\]. This classification system regards grade II or higher as clinically relevant and grade III or higher as severe complications. However, this definition is sometimes inconsistent with clinical severity, although objective and simple. For example, replacement of drainage tubes under fluoroscopy is classified as Clavien-Dindo grade IIIa, despite minor changes in clinical management and regarded severe complications. An establishment of a new grading system of POPF after gastrectomy may be necessary. However, it is less frequently used and may not be familiar because the prevalence of POPF after gastrectomy is not as high compared with that after pancreatomy; it occurs in approximately 1.6% of patients who underwent distal gastrectomy\[[@B25]\] and in 2.6% of patients who underwent total gastrectomy\[[@B26]\]. At present, the ISGPF and Clavien-Dindo classification systems have each of their advantages and disadvantages, and it is desirable to use both appropriately. LITERATURE SEARCH OF STUDIES REPORTING PREDICTIVE VALUE FOR POPF USING DRAIN AMYLASE CONTENT IN GASTRIC CANCER SURGERY ====================================================================================================================== Several studies have reported associations between D-AMY and POPF after gastrectomy, and Table [1](#T1){ref-type="table"} shows these studies in the gastric cancer surgical field. In 1997, Sano et al\[[@B3]\] reported for the first time that the measurement of D-AMY was useful for POPF after gastrectomy. However, a definition of POPF had not been established at that time, and their study defined it as a condition in which the D-AMY level was more than three times the upper normal serum amylase level for more than 7 d after operation. In their study, the prevalence of POPF in patients with D-AMY ≥ 4000 IU/L on POD 1 was significantly higher compared with that in patients with D-AMY \< 4000 IU/L on POD 1, indicating the high level of D-AMY on POD 1 was retained to POD 7, regardless of clinically relevant complications. ###### Studies measuring drain amylase content for predicting postoperative pancreatic fistula after gastrectomy in patients with gastric cancer **Ref**. **Sample size** **Surgical procedure** **POD 1** **POD 3** **Definition of POPF** -------------------------------- ----------------- ---------------------------- ----------- ----------- ------------------------ --------------------------------------- ----------------------------------------------- Sano et al\[[@B3]\], 1997 102 OTG, D1 - ≥ D2 Yes 4000 IU/L No NG D-AMY \> 3 times more than S-AMY for ≥ 7 days Iwata et al\[[@B2]\], 2010 372 Gastrectomy, D1 - ≥ D2 Yes 1000 IU/L No NG ISGPF definition (grade A/B/C) Tomimaru et al\[[@B4]\], 2011 172 TG, D1 plus - D2 Yes 5000 IU/L No NG ISGPF definition (grade B/C) Miki et al\[[@B6]\], 2011 104 TG, D2 Yes 3398 IU/L No NG ISGPF definition (grade B/C) Kobayashi et al\[[@B7]\], 2015 448 Gastrectomy, D1 - ≥ D2 Yes 1949 IU/L No NG C-D classification (grade III or higher) De Sol et al\[[@B5]\], 2015 53 Gastrectomy,D2 N0 NG Yes D-AMY \> 3 times more than S-AMY ISGPF definition (grade B/C) Kanda et al\[[@B8]\], 2016 265 LDG, D1 plus - D2 Yes 904 IU/L Yes Retained at ≥ 31.2% of D-AMY on POD 1 C-D classification (grade II or higher) Taniguchi et al\[[@B9]\], 2017 591 Gastrectomy D1- ≥ D2 Yes 2900 IU/L Yes 2100 IU/L ISGPF definition (grade B/C) Kamiya et al\[[@B10]\], 2018 801 Gastrectomy D1 plus - ≥ D2 Yes 2218 IU/L Yes 555 IU/L C-D classification (grade III or higher) Wakahara et al\[[@B12]\], 2019 327 Gastrectomy D0-D2 No NG Yes 761 IU/L C-D classification (grade II or higher) POPF: Postoperative pancreatic fistula; POD: Postoperative day; OTG: Open total gastrectomy; D-AMY: Amylase content of drainage tube; S-AMY: Serum amylase content; ISGPF: International Study Group on Pancreatic Fistula; NG: Not given or reported in the study; TG: Total gastrectomy; C-D classification: Clavien-Dindo classification; LDG: Laparoscopic distal gastrectomy. The ISGPF established the definition of POPF in 2005, and Iwata et al\[[@B2]\] adapted this definition and reported associations between D-AMY and POPF after gastrectomy in 2010. They reported that the prevalence of ISGPF grade A or higher was 16.3% and that D-AMY ≥ 1000 U/L on POD 1 along with body mass index were independent risk factors for POPF. However, the study suffered from the inclusion of a broad spectrum of surgical procedures ranging from the laparoscopic approach for early-stage cancer to extended lymphadenectomy accompanied by splenectomy for advanced cancer. In 2011, there were two studies that defined ISGPF grades B or C as POPF. Miki et al\[[@B6]\] reported that the prevalence of POPF after total gastrectomy with D2 lymphadenectomy was 22.1% and that D-AMY ≥ 3398 IU/L on POD 1 was an independent predictor of POPF. Tomimaru et al\[[@B4]\] reported that the prevalence of POPF after total gastrectomy with D1 plus or D2 lymphadenectomy was 9.2% and that D-AMY ≥ 5000 IU/L on POD 1 was a predictor of POPF. The above three studies were validated by applying the ISGPF classification and the predictive ability of D-AMY on POD 1, but the surgical procedures were different among studies, which caused the prevalence of POPF and the cut-off values to be inconsistent. In 2012, Kobayashi et al\[[@B7]\] adapted the Clavien-Dindo classification and reported that D-AMY ≥ 2000 IU/L on POD 1 and C-reactive protein ≥ 20 mg/dL on POD 3 were predictive of Clavien-Dindo classification grade III or higher POPF. The Clavien-Dindo classification grade III or higher POPF is the same as the ISGPF grade B or higher (grade C) excluding antibiotic treatment. So far, we summarized the studies to evaluate the predictive value of D-AMY on POD 1. Two time point measurements of D-AMY have been developed to enhance the predictive value of POPF. In 2016, Kanda et al\[[@B8]\] reported that D-AMY on POD 1 served as a predictive factor for POPF. In addition, patients whose D-AMY level on POD 3 was retained at ≥ 31.2% of that on POD 1 were more likely to develop POPF after laparoscopic distal gastrectomy\[[@B8]\]. After that, two studies reported that the combined use of D-AMY on POD 1 and POD 3 had a higher predictive performance for POPF compared with each alone\[[@B9],[@B10]\]. The combined use had high sensitivity; however, it did not serve as an early prediction. A few studies have considered drainage volume\[[@B5],[@B27]\] and concluded that drainage volume was not significant in gastric cancer surgery. REMAINING PROBLEMS FOR FUTURE STUDIES ===================================== As described in the previous section, the definition of POPF has not been established; in addition, several problems have remained for clinical use, such as timing of measuring D-AMY, optimal cut-off value, and means of intervention after early prediction. The timing of measuring D-AMY has not been determined. There is a dilemma between early prediction and diagnostic accuracy. With a single predictive marker, there is a limitation to increasing both sensitivity and specificity, and attempts have been made to enhance the predictive accuracy of factors on POD 1 as well as on POD 3 as combined markers\[[@B8]-[@B10]\]. However, it has limited clinical use for early prediction and early intervention. In other words, the timing of measurement that has both high diagnostic accuracy and early detection has not been determined. Second, the cut-off values were different among studies. Differences in the definition of POPF, differences in the surgical procedure, and small-scale retrospective studies prevent the establishment of the optimal cut-off value. From the viewpoint of the mechanism of POPF, the differences in surgical procedures do not affect fistula formation. A large-scale prospective trial is warranted to establish an optimal cut-off value that applies to any surgical procedure. Third, it is unclear whether early intervention will improve outcomes even if early prediction is successful. Prophylactic antibiotics in gastric cancer surgery are usually administered until the next morning after surgery. Additional prophylactic antibiotic administration may be beneficial for preventing deterioration in particular POPF high-risk patients who underwent pancreaticoduodenectomy\[[@B28]\]; although so far, the benefit is unclear in gastric cancer surgery\[[@B29]\]. Currently, a prospective exploratory randomized trial to evaluate prolonged prophylactic antibacterial drug treatment for patients with high levels of D-AMY on POD 1 after gastrectomy is in progress (UMIN000012152). In addition, the benefit of the measurement of D-AMY content as an indicator of early drainage tube removal is unknown. The prophylactic drain is helpful for the detection of not only POPF but also other serious complications, including anastomotic leakage, intraoperative bleeding, and injury of the intestine. It is possible that the drainage of intra-abdominal fluids, including pancreatic juice, may prevent subsequent POPF\[[@B2]\]. From the viewpoint of the enhanced recovery after surgery program, the drainage tube should be removed as soon as it is deemed unnecessary in order to reduce drain-related complications and shorten the hospital stay after gastrectomy\[[@B30]-[@B32]\]. Additionally, unnecessary drain placement is harmful in terms of intra-abdominal fluid loss. At present, the measurement of D-AMY is not used as an indicator of early drainage tube removal, but if the level of D-AMY is low, the drainage tube can be removed with little concern for later pancreas-related complications. CONCLUSION ========== The measurement of D-AMY is promising because of its high predictive ability of POPF, even in the gastric cancer surgical field. However, many problems remain unresolved, *i.e*., definition of POPF, ideal timing for measurement, optimal cut-off value, and means of intervention after prediction. Prospective clinical trial could be imperative in order to develop D-AMY measurement in common clinical practice for gastric surgery. ACKNOWLEDGEMENTS ================ We are grateful to the nonprofit organization Epidemiological and Clinical Research Information Network (ECRIN) for providing their valuable support with this work. Conflict-of-interest statement: No potential conflicts of interest. No financial support. Manuscript source: Invited manuscript Peer-review started: December 24, 2019 First decision: January 12, 2020 Article in press: March 22, 2020 Specialty type: Gastroenterology and Hepatology Country/Territory of origin: Japan Peer-review report's scientific quality classification Grade A (Excellent): 0 Grade B (Very good): B, B Grade C (Good): C Grade D (Fair): 0 Grade E (Poor): 0 P-Reviewer: Aurello P, Chen S, Ju SQ S-Editor: Dou Y L-Editor: A E-Editor: Ma YJ [^1]: Author contributions: Nakanishi K wrote the paper; Nakanishi K, Kanda M, Sakamoto J, and Kodera Y critically revised the article. Corresponding author: Mitsuro Kanda, FACS, MD, PhD, Department of Gastroenterological Surgery (Surgery II), Nagoya University Graduate School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550, Japan. <m-kanda@med.nagoya-u.ac.jp>

{#sp1 .619} {#sp2 .620} {#sp3 .621} {#sp4 .622} {#sp5 .623} {#sp6 .624} {#sp7 .625} {#sp8 .626} {#sp9 .627} {#sp10 .628} {#sp11 .629}

Study Rationale and Context =========================== Wound suction drains have been used to decrease the rate of postoperative hematoma formation and thus wound infections for many years throughout all surgical subspecialties. Although the use of surgical drains dates back to the years of Hippocrates,[@BR1500019srsupl-1] in the orthopedic literature these drains have not been shown to be beneficial in decreasing the rates of these complications, especially in orthopedic procedures including fracture fixation or arthroplasty surgeries.[@JR1500019srsupl-2] [@JR1500019srsupl-3] [@JR1500019srsupl-4] [@JR1500019srsupl-5] However, these drains are still commonly used throughout the orthopedic community, including spine surgery. Debate in this area remains, as proponents of its use in the immediate postoperative period believe that it will prevent fluid collection in the surgical dead space and thus eliminate the media for bacterial growth. On the other hand, opponents believe that they are considered a foreign body that promotes inflammation and even sometimes provokes an infectious response. In spine surgery, the controversy is even more profound because it decreases the rare but devastating complication of postoperative epidural hematoma, but it may have a hypothetical increase in the risk of infection.[@JR1500019srsupl-6] [@JR1500019srsupl-7] In the spine literature, the incidence of epidural hematoma in the postprocedural period ranges from 0.2 to 2.9%,[@JR1500019srsupl-8] [@JR1500019srsupl-9] [@JR1500019srsupl-10] and the incidence of postoperative wound infection is 0.7 to 16%,[@JR1500019srsupl-11] [@JR1500019srsupl-12] thus identifying a method that can decrease the incidence of these complications would be of great benefit. To help address this debate, we conducted a systematic review of the use of these drains and extensively explored the efficacy and safety of closed suction wound drainage in spine surgery in the postoperative period. Clinical Question ================= In patients undergoing spine surgery for lumbar degenerative conditions, does the use of closed suction wound drains decrease the incidence of postoperative complications compared with no drain use? Materials and Methods ===================== **Study design:** Systematic review. **Search:** PubMed, Embase, the Cochrane Library, Web of Science, Scopus, and bibliographies of key articles **Dates searched:** through January 22, 2015. **Inclusion criteria:** (1) comparative studies in peer-reviewed journals; (2) adult patients undergoing spinal surgery for degenerative conditions in the lumbar spine receiving postoperative closed suction drains or no drains; (3) outcomes included at least one of the following: epidural hematoma, superficial wound infection, deep wound infection, or postoperative blood transfusion. **Exclusion criteria:** (1) skeletally immature patients (\<18 years of age); (2) surgery for intradural pathology, dural tears, tumor, trauma, fracture, or infection; (3) cervical or thoracic spine surgery; (4) nonclinical studies, case reports, and case series. **Outcomes:** (1) epidural hematoma; (2) superficial wound infection; (3) deep wound infection; and (4) postoperative blood transfusion. **Analysis:** Due to heterogeneity in patient populations (including differences in patient demographics, diagnoses, and surgical procedures) and differences in study design, a meta-analysis was not performed. We calculated the cumulative incidence by dividing the number of patients with an event by the number at risk for the event and the associated 95% confidence interval. When zero events were reported for an outcome, the confidence interval was found using the "rule of three" estimation.[@JR1500019srsupl-13] Details about the methods can be found in the online supplementary material. **Overall strength of evidence:** The overall strength of evidence across studies was based on precepts outlined by the Grades of Recommendation Assessment, Development and Evaluation Working Group. Study critical appraisals and reasons for upgrading and downgrading for each outcome can be found in the online supplementary material. Results ======= - Three randomized controlled trials (RCTs) and two cohort studies met the inclusion criteria and form the basis for this report ([Fig. 1](#FI1500019srsupl-1){ref-type="fig"}). Their characteristics are described in [Table 1](#TB1500019srsupl-1){ref-type="table"}. A list of studies excluded and the reason for exclusion can be found in the online supplementary material. - One RCT included patients undergoing multilevel decompression or fusion as a primary or reoperation procedure for herniated nucleus pulposus, spinal stenosis, degenerative spondylolisthesis, or postlaminectomy syndrome in an older population (mean age 67.4 years; [Table 1](#TB1500019srsupl-1){ref-type="table"}).[@JR1500019srsupl-14] Two RCTs included patients undergoing single-level laminectomy for herniated disk or degenerative stenosis, though one study placed the drains in the epidural space (mean age 46.7 years),[@JR1500019srsupl-15] and the second placed the drain in the lumbodorsal fascia (mean age not reported).[@JR1500019srsupl-7] - Two retrospective cohorts included patients undergoing single or multilevel laminoplasty, diskectomy, or fusion for varying diagnoses (mean age 46 to 57.3 years).[@JR1500019srsupl-16] [@JR1500019srsupl-17] {#FI1500019srsupl-1} ###### Characteristics of included studies ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ First author (year) Study type Population Treatment Diagnosis Surgery type -------------------------------------- -------------- --------------------- ---------------------- --------------------------------------------------- ----------------------------------------------------------------------------------- Payne (1996)[@JR1500019srsupl-7] RCT *n* = 200\ Drain (*n* = 103)\ • Herniated disk (% NR)\ • Singe-level hemilaminectomy (% NR)\ Age: NR\ No drain (*n* = 97) • Degenerative stenosis (% NR) • Decompressive laminectomy (% NR) % male: NR Brown (2004)[@JR1500019srsupl-14] RCT *n* = 83\ Drain (*n* = 42)\ • Herniated nucleus pulposus: *n* = 13 (16%)\ • Decompression (2--4 levels)\ Age: 67.4 y\ No drain (*n* = 41) • Spinal stenosis: *n* = 73 (88%)\ Primary: *n* = 15 (35%)\ % male: NR • Degenerative spondylolisthesis: *n* = 40 (48%)\ Reoperation: *n* = 12 (15%)\ • Postlaminectomy syndrome: *n* = 15 (18%) • Fusion with instrumentation: *n* = 16 (19%)\ • Fusion without instrumentation: *n* = 30 (36%) Mirzai (2006)[@JR1500019srsupl-15] RCT *n* = 50\ Drain (*n* = 22)\ • Herniated disk: *n* = 50 (100%) • Virgin single-level microdiskectomy with hemipartial laminectomy and flavectomy Age: 46.7 ± 7.6 y\ No drain (*n* = 28) 58% male Kanayama (2010)[@JR1500019srsupl-16] Retro cohort *n* = 560\ Drain (*n* = 298)\ NR • Single-level laminoplasty (medial facetectomy) (% NR)\ Age: 46 y\ No drain (*n* = 262) • Diskectomy (% NR) 64.3% male Walid (2012)[@JR1500019srsupl-17] Retro cohort *n* = 402\ Drain (*n* = 285)\ • Spondylosis: *n* = 31 (8%)\ • Lateral fusion: *n* = 301 (75%)\ Age: 57.3 ± 11.5 y\ No drain (*n* = 117) • Disk displacement: *n* = 145 (36%)\ • Posterior fusion: *n* = 101 (25%) 43% male • Disk degeneration: *n* = 84 (21%)\ • Disk disease with myelopathy: *n* = 23 (6%)\ • Lumbar spinal stenosis: *n* = 79 (20%)\ • Acquired spondylolisthesis: *n* = 29 (7%)\ • Spondylolisthesis: *n* = 11 (3%) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Abbreviations: NR, not reported; RCT, randomized controlled trial. Hematoma -------- - The evidence base included three RCTs and two retrospective cohorts. - The method of identification of hematoma varied among studies. One small RCT reported 89% of the patients with drains and 36% of the patients without drains had epidural hematomas as detected by magnetic resonance imaging. Most were minimal in size with only 7% prominent in the no-drain group compared with 0% prominent in the drain group.[@JR1500019srsupl-15] All other studies assessed hematomas that required drainage or reoperation. The risk in these studies ranged from 0.0 to 0.7%.[@JR1500019srsupl-7] [@JR1500019srsupl-14] [@JR1500019srsupl-16] [@JR1500019srsupl-17] - There was no difference in risk for hematoma comparing drains with no drains across studies of different design, surgeries of varying complexity, patients of different ages, or complexity of diagnosis.[@JR1500019srsupl-7] [@JR1500019srsupl-14] [@JR1500019srsupl-15] [@JR1500019srsupl-16] [@JR1500019srsupl-17] The upper 95% confident limit of risk for hematoma in the patients with drains was 13.6% compared with 16.7% for the patients without drains ([Table 2](#TB1500019srsupl-2){ref-type="table"}). ###### Hematoma, superficial wound infection, deep infection, and postoperative blood transfusion among patients receiving closed wound drains compared with those who do not receive closed wound drains following lumbar surgery ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- First author (year) Treatment Hematoma Superficial wound infection Deep infection Postoperative blood transfusion -------------------------------------- ---------------------- ------------------------------------------------ ----------------------------- ---------------- --------------------------------- ------- ----------------- ----- -------------------- Payne (1996)[@JR1500019srsupl-7] Drain (*n *= 103)\ 0[a](#FN1500019srsupl-3){ref-type="table-fn"}\ 0.0 (0.0--2.9)\ 2\ 1.9 (0.0--4.6)\ NR NR NR NR No drain (*n* = 97) 0[a](#FN1500019srsupl-3){ref-type="table-fn"} 0.0 (0.0--3.1) 1 1.0 (0.0--3.0) Brown (2004)[@JR1500019srsupl-14] Drain (*n* = 42)\ 0\ 0.0 (0.0--7.1)\ 0\ 0.0 (0.0--7.1)\ 0/42\ 0.0 (0.0--7.1)\ NR NR No drain (*n* = 41) 0 0.0 (0.0--7.3) 0 0.0 (0.0--7.3) 0/41 0.0 (0.0--7.3) Mirzai (2006)[@JR1500019srsupl-15] Drain (*n* = 22)\ 0[b](#FN1500019srsupl-4){ref-type="table-fn"}\ 0.0 (0.0--13.6)\ NR NR NR NR NR NR No drain (*n* = 28) 2[b](#FN1500019srsupl-4){ref-type="table-fn"} 7.1 (0.0--16.7) Kanayama (2010)[@JR1500019srsupl-16] Drain (*n* = 298)\ 2[c](#FN1500019srsupl-5){ref-type="table-fn"}\ 0.7 (0.0--1.6)\ 0\ 0.0 (0.0--1.0)\ 0\ 0.0 (0.0--1.0)\ NR NR No drain (*n* = 262) 0[c](#FN1500019srsupl-5){ref-type="table-fn"} 0.0 (0.0--1.1) 0 0.0 (0.0--1.1) 0 0.0 (0.0--1.1) Walid (2012)[@JR1500019srsupl-17] Drain (*n* = 285)\ 0[d](#FN1500019srsupl-6){ref-type="table-fn"}\ 0.0 (0.0--1.1)\ 10\ 3.5 (1.4--5.7)\ NR NR 68\ 23.9 (18.9--28.8)\ No drain (*n* = 117) 0[d](#FN1500019srsupl-6){ref-type="table-fn"} 0.0 (0.0--2.6) 3 2.6 (0.0--5.4) 8 6.8 (2.3--11.4) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Abbreviations: CI, confidence interval; FU, follow-up; NR, not reported. Reported as hematoma requiring drainage. "Prominent" hematoma detected with magnetic resonance imaging. Minimal hematoma was found in 31.8% of patients with drains versus 60.7% without drains, and moderate in 4.5% patients with drains versus 21.4% without drains. Reported as epidural hematoma causing neurologic compromise requiring surgical evacuation. Reported as hematoma requiring reoperation. Superficial Wound Infection --------------------------- - The evidence base included two RCTs and two retrospective cohorts. - There was no difference in risk between the patients receiving closed suction wound drains (range of risks, 0.0 to 3.5%) compared with the patients with no drains (range of risks, 0.0 to 2.6%; [Table 2](#TB1500019srsupl-2){ref-type="table"}).[@JR1500019srsupl-7] [@JR1500019srsupl-14] [@JR1500019srsupl-16] [@JR1500019srsupl-17] The upper 95% confident limit of risk for superficial wound infection in the patients with drains was 7.1% and for the patients without drains was 7.3%. Deep Infection -------------- - The evidence base included one RCT and one retrospective cohort. - There are no events recorded in the two studies that report risk for deep infection in the patients treated with versus patients treated without closed suction wound drains after lumbar surgery ([Table 2](#TB1500019srsupl-2){ref-type="table"}).[@JR1500019srsupl-14] [@JR1500019srsupl-16] The upper limit of risk for deep infection in the patients with drains was 7.1% and in the patients without drains was 7.3%. Postoperative Blood Transfusion ------------------------------- - The evidence base included one retrospective cohort. - One observation study reported a greater risk of blood transfusion after lumbar surgery in the patients with drains (24%) versus patients without drains (7%; risk ratio 3.5; 95% confidence interval 1.7 to 7.0).[@JR1500019srsupl-17] Evidence Summary ================ There was no difference in the incidence of hematoma, superficial wound infection, or deep infection in the patients with versus patients without closed suction wound drains after lumbar surgery. The overall strength of evidence for these findings is considered low or insufficient ([Table 3](#TB1500019srsupl-3){ref-type="table"}). There is insufficient evidence to draw conclusions on the effect of closed suction wound drains on the risk for postoperative blood transfusion. ###### Evidence summary: Do closed suction wound drains decrease the incidence of postoperative complications? ---------------------------------------------------------------------------------------------------------------------------------------- Outcome Overall quality of evidence Studies Effect size --------------------------------- -------------------------------------------------- ---------------------------- ---------------------- Hematoma Low due to risk of bias and imprecision 3 RCTs (*n* = 333) Drain: 2.9--13.6%\ No drain: 3.1--16.7% 2 retro cohorts (*n* = 962) Drain: 1.1--1.6%\ No drain: 1.1--2.6% Superficial wound infection Insufficient due to risk of bias and imprecision 2 RCTs (*n* = 283) Drain: 4.6--7.1%\ No drain: 3.0--7.3% 2 retro cohorts (*n* = 962) Drain: 1.0--5.7%\ No drain: 1.1--5.4% Deep infection Insufficient due to risk of bias and imprecision 1 RCT (*n* = 83) Drain: 7.1%\ No drain: 7.3% 1 retro cohort (*n* = 560) Drain: 1.0%\ No drain: 1.1% Postoperative blood transfusion Insufficient due to risk of bias 1 retro cohort (*n* = 402) Drain: 28.8%\ No drain: 11.4%\ RR: 3.5 (1.7, 7.0) ---------------------------------------------------------------------------------------------------------------------------------------- Abbreviations: RCT, randomized controlled trial; retro, retrospective cohort; RR, risk ratio. Calculated using Hanley\'s rule of three when zero events are reported for a given outcome. Clinical Guidelines =================== No clinical guidelines were found. Discussion ========== - In the setting of spinal surgery, one of the biggest concerns is an epidural hematoma; some spine surgeons use drains to decrease the incidence of postoperative hematomas and subsequently neurologic complications. However, there is no clear consensus whether is it beneficial by decreasing hematomas or harmful by increasing the risk of infection and the rate of blood transfusion. - Five studies assessing the use of closed suction wound drains in spine surgeries for lumbar degenerative conditions were identified and included in this systematic review. - One strength of our study was the thorough and systematic search of the evidence performed. The limitations included the following: the methodologies of studies included in this review were heterogeneous, and all studies had small sample sizes. There was a lack of random sequence generation or allocation concealment in the RCTs. In addition, there was a lack of observer blinding and lack of controlling for confounding factors. (See [Table 3](#TB1500019srsupl-3){ref-type="table"} in the online supplementary material.) - Contrary to previous beliefs, there is no difference in the incidence of hematoma, superficial wound infection, or deep infection in the patients with versus patients without closed suction wound drains after lumbar surgery. - As this systematic review was of low or insufficient evidence, more randomized trials of the use of closed suction wound drains in lumbar spine surgery are needed with sufficient sample sizes and appropriate methodology protocols to adequately assess the effect of drain use on these rare outcomes. Conclusions =========== Conclusions from this systematic review are limited by the quality of included studies that assessed the use of closed suction wound drains in lumbar spine surgeries for degenerative conditions. The small sample sizes, lack of random sequence generation, or allocation concealment in RCTs and lack of controlling for confounding factors may not have allowed us to detect differences between treatment groups, particularly for rare outcomes. We believe that spine surgeons should not routinely rely on the use of closed suction wound drains in lumbar spine surgery until a higher level of evidence becomes available to support its use. Editorial Perspective ===================== Surgical practice is the invasive treatment of human ailments through the application of skill and technology drawn from a combination of experiential and of learned behaviors. The placement of drains in elective posterior spine surgery is one of these time-honored learned behaviors. Use of postoperative drains is a big deal from the medical and business side of medicine as their use may influence the occurrence of certain complications and may also be a factor on length of stay and other direct expenses such as material costs and use of antibiotics as well as personnel costs. In this era of questioning everything, the use of drains should not be considered as a given but should be subjected to a rigorous review of the evidence base. Waly and coauthors received universal praise from the reviewers for their choice of the subject of the evidence-based foundation of the use of drains by performing a formal systematic review. The purported benefits of postoperative drainage are clear: a hopefully decreased infection rate and wound-healing disruption through the removal of postoperative fluid accumulation from the wound cavity and a decreased rate of epidural hematoma formation with the added risk of potential secondary neurologic deterioration. Using an extensive search strategy going all the way back to antiquity for this meritorious topic, the authors came up with two surprising results: (1) there is no evidence whatsoever to use posterior wound drains in the spine for either infection or hematoma; (2) there are remarkably few studies of higher quality that have actually examined the merits of postoperative wound drainage in spine surgery. Certainly, the evidence base for postoperative wound drainage is inverse to the commonality with which wound drainage is performed. In reviewing these studies, the fact that there were no greater differences identified is not entirely surprising. The overall number of patients in the three prospective RCTs was 333. The two retrospective cohort studies combined for 962 patients, which leaves us with a patient cohort of ∼1,300 patients as a basis for the investigation of the merits of postoperative drain use in spine surgery. For sporadic occurrences like symptomatic epidural hematomas and their reported incidence of less than 1%, this number of subjects is simply too small to be able to hope to detect any differences. Especially when "simple surgeries" such as primary microdiskectomies in low-risk patients are part of the patient cohort, the expected rate of symptomatic postoperative complications could be expected to be sporadic at best. The cohort size is insufficient even for the far more common postoperative complication of wound infections, with a reported incidence of up to 7.1%. The pool of patients includes a mixture with instrumentation and no hardware used with variable degrees of invasiveness and unclear handling of important comorbidities such as diabetes, body habitus, immune or nutritional compromise, blood thinners, and cerebrospinal fluid leak, which simply does not allow for a meaningful assessment of the influence of drains on the desired outcome of reducing infections. Other important variables, such as duration of drain placement and use of perioperative antibiotics as well as type of wound closure and incision management, are also not factored into this evaluation. In conclusion, the best we can say for now is that the use of drains does not seem to prevent an increase of postoperative wound infections. On the contrary, the claim that postoperative wound drainage would actually reduce this dreaded complication can certainly not be maintained. Given the number of variables at hand and the differences in practices and patient comorbidities, the question arises if a prospective RCT or a retrospective study format can actually ever be sufficient to actually pick up important differences in complication rates such as symptomatic epidural hematoma or wound infections. A much larger patient base, such as found in a registry, may turn out to be a preferred study format to try to better investigate this topic. Until then, the editors of the *Evidence-Based Spine-Care Journal* edition of the *Global Spine Journal* invite the commentaries of its readers around the world on their recommendations on postoperative drain use and also what pathway to use to arrive at a best practices consensus. Analytic support for this work was provided by Spectrum Research, Inc. with funding from AOSpine. **Disclosures** Feras Waly, none Mohammad Alzahrani, none Fahad H. Abduljabbar, none Tara Landry, none Jean Ouellet, none Kathryn Moran, none Joseph R. Dettori, none Supplementary Material ====================== ###### Supplementary material

Introduction {#s1} ============ Major Depressive Disorder (MDD) has a highly recurrent nature [@pone.0057510-Judd1] with relapse and recurrence rates rising up to 85% [@pone.0057510-Mueller1]. Recurrence is defined as a new Major Depressive Episode (MDE) after recovery, that is at least six months without meeting full MDE criteria, whereas relapse is the return of a MDE during remission but before recovery (1988, MacArthur Foundation Research Network on the Psychobiology of Depression) [@pone.0057510-Frank1]. Most frequently used international evidence based clinical practice guidelines i.e. National Institute for Health & Clinical Excellence (NICE) [@pone.0057510-National1], [@pone.0057510-National2], American Psychiatric Association (APA) [@pone.0057510-AmericanPsychiatric1], presume that co-morbid chronic general somatic disorders are associated with poorer outcomes of depression including an increased risk of recurrence (for readability we always refer to recurrence in case of relapse or recurrence). Depression is more prevalent in people suffering somatic illness, with prevalence rates being two [@pone.0057510-Moussavi1] to even three times as high as for people without somatic illness [@pone.0057510-AmericanPsychiatric1], [@pone.0057510-Egede1]. Depression has been associated with poorer outcomes of somatic illness [@pone.0057510-National2], [@pone.0057510-Egede1]--[@pone.0057510-Evans1], in terms of more functional disability, higher care consumption and a lower quality of life [@pone.0057510-Nol1]. Conversely, the presence of a co-morbid chronic somatic condition is perceived as an ongoing stressor that predisposes patients to depressive episodes [@pone.0057510-AmericanPsychiatric1]. The APA [@pone.0057510-AmericanPsychiatric1] describes "the presence of a chronic general medical disorder" as a risk factor for recurrence of MDD and therefore recommends that "some form of maintenance treatment will be required indefinitely". Objectives {#s1a} ---------- Remarkably, there are no reviews or meta-analyses that examined the effect of co-morbid chronic somatic illness on depressive recurrence. Given the impact of current international clinical guidelines, we aim to systematically review the evidence as to whether somatic illness is a risk factor for recurrences of MDD over a period of at least six months. If possible a meta-analysis will be performed as well (depending on the number of studies and their methodological characteristics). Methods {#s2} ======= Inclusion criteria {#s2a} ------------------ Inclusion criteria for our review were: (1) longitudinal measurement of the course of depression (2) providing absolute numbers or percentages of recurrence a) diagnosis established with an interview based on state- of the- art depression criteria (e.g. Diagnostic and Statistical Manual of Mental Disorders, DSM-III/III-R/DSM-IV) [@pone.0057510-AmericanPsychological1]--[@pone.0057510-AmericanPsychological3] or b) with standardized questionnaires that assess depressive symptoms (e.g., Inventory of Depressive Symptomatology, IDS) [@pone.0057510-Rush1] or -, Hamilton Rating Scale for Depression, HRSD) [@pone.0057510-Hamilton1] (3) with a follow-up of at least six months (4) in which data were collected for patients with and without a certain co-morbid chronic somatic illness at the same measurement intervals a) where co-morbid chronic somatic illnesses were assessed either via self-report or b) medical records or c) by a diagnosis of a medical professional. There is great overlap between self-report of somatic illnesses and diagnoses of these illnesses [@pone.0057510-Wallace1], therefore studies that used self-report as a measurement tool were included as well. If treatment effects were studied within a randomized controlled trial without a treatment-as-usual group, these were excluded. We also excluded studies on bipolar disorders. All relevant publications in English, Dutch, Spanish, Polish or German were taken into account. To assess eligibility of articles, one reviewer (GK) made the first selection based on titles. In case of doubt abstracts or full text articles were retrieved for closer reading. Thereafter, two independent reviewers made a selection based on abstracts (GK and WH); further winnowing was performed by two reviewers (GK and CB) based on full text articles. In case of inconsistencies, articles were evaluated again until consensus was reached. The kappa statistic for inter-observer variability was reported for the abstract and full-text selection. Literature search {#s2b} ----------------- In line with the APA [@pone.0057510-AmericanPsychiatric1] and NICE [@pone.0057510-National1], [@pone.0057510-National2] treatment guidelines we limited the search to chronic somatic illnesses. The chronic somatic illnesses chosen for this systematic review were: heart diseases, gastrointestinal diseases, diabetes mellitus, rheumatoid arthritis, asthma, Human Immunodeficiency Virus (HIV) and neoplasms. The choice of these chronic somatic illnesses was based on them being mentioned in the clinical practice guidelines of the APA [@pone.0057510-AmericanPsychiatric1] and NICE [@pone.0057510-National1] and their high prevalence rates brought up in additional literature [@pone.0057510-Simon1], [@pone.0057510-Koike1]. A combination of MeSH-terms and free text words was entered into the search engines PubMed, EMbase and PsycINFO. These were all screened for relevant articles through 4 December 2012. The terms 'depression or depressive disorder or major depression were combined with heart diseases or gastrointestinal diseases or diabetes mellitus or arthritis, rheumatoid or asthma or HIV or neoplasms and incidence or follow-up studies or prognos\* or predict\* or course or outcome or relaps\* or recur\* or remis\* or epidemiology. The key words regarding depression outcomes were based on Altman [@pone.0057510-Altman1] (box 2) and modified by adding relaps\*, recur\* and remis\* to fit our research purpose. Since not all studies might explicitly mention recurrence, even though they studied the impact on for example chronicity, including recurrence, we decided on the above mentioned broad search terms. Additionally, reference lists from included articles, earlier reviews and NICE [@pone.0057510-National1], [@pone.0057510-National2] and APA [@pone.0057510-AmericanPsychiatric1] clinical treatment guidelines were screened for other potentially eligible papers and experts were consulted to identify additional important papers. Data extraction and outcomes {#s2c} ---------------------------- The following data were extracted from the included articles: study site, number of participants, their age and gender distribution, information about depression assessment (method and measurement intervals), characteristics of the chronic somatic illnesses (type and assessment method) and the outcome measure (recurrence). Outcomes consisted of the differences in percentages or mean numbers of recurrence during follow-up, if applicable for multiple time intervals, between patients with and without co-morbid chronic somatic illness. Risk ratios (RR) and 95% Confidence Intervals (CI) were calculated by using Review Manager 5.1. Quality assessment {#s2d} ------------------ To allow for judgments on the quality of the included articles regarding their selection process, design, analyses and outcome measures, a modified version of the Newcastle-Ottowa Quality Assessment Scale for cohort studies was used [@pone.0057510-Wells1]. This instrument was reviewed by Deeks et al. [@pone.0057510-Deeks1] and described as one of the most usable methods for this type of study [@pone.0057510-Deeks1], [@pone.0057510-Reeves1]. All articles were judged by two reviewers (GK and WH). If information on a quality criterion was not mentioned explicitly in the article, we assigned a question mark. Results {#s3} ======= Study selection {#s3a} --------------- As shown in [figure 1](#pone-0057510-g001){ref-type="fig"}, the search engines yielded 3450 articles in total. Twenty records were identified via additional sources including APA guidelines (n = 6), experts (n = 2), reference lists from subsequently included articles (n = 10), and important reviews (n = 2). After screening based on title done by one reviewer (GK), 3121 articles were excluded. Main reasons for exclusion based on title were: studies reported about the influence of depression in a somatic illness population only or about the prevalence of depression only. A total of 349 abstracts subsequently were screened by two independent reviewers (GK and WH). After removal of 55 duplicate titles, eventually 51 full text articles were retrieved for close reading. Detailed information on reasons for exclusion is shown in the flow diagram ([figure 1](#pone-0057510-g001){ref-type="fig"}). Finally, four articles about recurrence of depression in patients with and without co-morbid chronic somatic illness were included. The Kappa statistic for the inter-observer variability was 0.91. {#pone-0057510-g001} Data extraction from included studies {#s3b} ------------------------------------- Data were extracted for the four articles with a total number of 2010 patients (n = 554 in study 1, n = 715 in study 2, n = 54 in study 3, and n = 687 in study 4) yielding information on 694 patients (34.5%) who suffered one or more chronic somatic illness. [Table 1](#pone-0057510-t001){ref-type="table"} gives an overview of the data extracted from the four studies. 10.1371/journal.pone.0057510.t001 ###### Data extraction and outcomes of the four included studies. {#pone-0057510-t001-1} Authors (publication year) Number of patients Age Gender, female (%) Depression assessment 1) Baseline 2) Follow-up Illness assessment Outcome measures Results ----------------------------------------------------------------- --------------------------------- --------- ------------------------------- --------------------------------------------------------------------------------------------------------------------------- ----------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------ ----------------------------------------------------------------------------- Wells, Rogers, Burnam & Camp (1993) 554 total (3.8% SI, 96.2% NSI) \- 73.8% (IDDM) 72.8% (non-IDDM) 1\) Eight- item depression symptom scale, DSM III criteria 2) Course of depression interview over the phone, based on DIS Screening and structured somatic history interview Percentage of patients with more than two depressive spells during follow -up 9.9 years 1 year: SI: 14.9% NSI: 29.3% 2 year: SI: 36.0% NSI: 27.8% Gerrits, van Oppen, van Marwijk, van der Horst & Penninx (2013) 715 total (44.5% SI 55.5% NSI) 42.1^a^ 66% 1)CIDI, DSM IV 2) CIDI, DSM IV and Life Chart Inventory Self-report Percentage of patients with remission with recurrence of symptoms (at least three months symptom-free interval) 2 year: SI: 6.0% NSI: 8.0% Kovacs, Goldston, Obrosky & Drash (1997) 54 total (44.4% SI, 55.6% NSI) 11.2^a^ 61.9% 1\) ISCA, DSM-III 2) ISCA, DSM-III Diagnoses at hospital Recurrence rates: cumulative proportion of developing a new episode of depression after recovery 1 year: SI: 26% NSI: 22% 2 year: SI:30% NSI: 32% 6.5 year: SI: 47% NSI: 47% Hardeveld, Spijker, De Graaf Nolen & Beekman, 2012 687 total (48.2% SI, 51.8% NSI) 40.7^b^ 68.0% 1\) CIDI, DSM-III-R 2) CIDI, DSM-III-R retrospectively and prospectively at T1 or T2 Self-report, list of 31 mostly chronic somatic conditions MDE recurrence between baseline and follow-up (one-or three years) for patients in partial or complete remission for at least six months SI: 21.1% NSI: 18.3% *Note,* SI = Somatically Ill; NSI = Non-Somatically Ill; ^a^  =  mean age; ^b^ =  median age; IDDM =  Insulin-Dependent Diabetes Mellitus; DSM III/III-R/IV criteria =  Diagnostic and Statistical Manual of Mental Disorders, third edition, third edition revised; fourth edition; DIS = Diagnostic Interview Schedule; CIDI =  Composite International Diagnostic Interview; ISCA =  Interview schedule for children and adolescents; MDE = Major Depressive Episode. Study characteristics {#s3c} --------------------- Study 1: A non-experimental longitudinal follow-up study by Wells, Rogers, Burnam, and Camp [@pone.0057510-Wells2] with adults who received care in one of the following settings: "large group-practice-style health maintenance organizations (HMOs), large multispecialty mixed prepaid and fee-for-service group practices, and small single-specialty group and solo practices". Patients with either a current depressive disorder or depressive symptoms where included (n = 554 patients). A depressive disorder was defined as meeting DSM-III criteria [@pone.0057510-AmericanPsychological1] for a lifetime diagnosis of MDD or dysthymia, an episode of MDD or dysthymia during the past 12 months, with no remission since the onset of the recent episode. Depressive symptoms were present when the cut-off score on a brief depression screener was exceeded [@pone.0057510-Burnam1]. According to the authors (personal communication), patients were asked how many episodes they experienced over follow-up with at least a two month break of depressive symptoms between episodes. The group consisted of 21 (3.8%) patients who suffered current Insulin-Dependent Diabetes Mellitus (IDDM) and 533 (96.2%) without IDDM. Two other somatic illnesses and their relation with depressive recurrence were studied by Wells et al. [@pone.0057510-Wells2] however we only used the information on the IDDM group while this was the only current chronic somatic illness mentioned. Study 2: The study of Gerrits, van Oppen, van Marwijk, van der Horst and Penninx [@pone.0057510-Gerrits1] is a cohort study (ages 18--65 years) that consisted of 1209 adult participants with a current depression or anxiety diagnosis as assessed by using the Composite International Diagnostic Interview (CIDI) [@pone.0057510-Wittchen1], based on DSM-IV criteria [@pone.0057510-AmericanPsychological3], followed across two years. Only the information on depression recurrence was used (received through personal communication with the authors) which resulted in data on 715 patients with depression. A total of 318 patients (44.5%) had one or more of the following chronic somatic diseases: cardio-metabolic, respiratory, musculoskeletal, digestive, neurological, endocrine and cancer. Study 3: Kovacs, Obrosky, Goldston, and Bonar [@pone.0057510-Kovacs1] included 54 children (ages 8--13 years) in their longitudinal follow-up study. There were 24 children (44.4%) with current IDDM and 30 children (55.6%) with no other somatic disorders from the same children's hospital. MDD was assessed with the semi structured Interview Schedule for Children and Adolescents (ISCA) [@pone.0057510-Kovacs2]. The control subjects without somatic co-morbidity and a first MDE were balanced for age of onset of first MDD, other co-morbidities and basic characteristics. Of the 24 participants with co-morbid IDDM, six (25%) already suffered from a major psychiatric disorder (i.e. n = 4 with anxiety disorder, n = 1 conduct disorder and n = 1 with functional enuresis). The follow-up period was almost ten years (mean of 9.9 years). Recovery was defined as not fulfilling criteria for MDE, i.e. the absence of symptoms or the presence of few subclinical symptoms, and persistence of this state for at least two months. Of the participants with IDDM, 21 (87.5%) recovered from a first episode as for 29 (96.7%) of the control group. Study 4: Hardeveld, Spijker, De Graaf, Nolen and Beekman [@pone.0057510-Hardeveld1] included 687 patients from the general population (ages between 18--64 years). Of the total group, 331 patients (48.2%) had a somatic disease, which was assessed by a questionnaire including 31, mostly, chronic somatic illnesses during the past 12 months. To be included in the study, patients had to be in partial or complete remission of MDD and/or dysthymia for at least six months and the amount of months of being in remission could differ between patients. Remission was defined as: not meeting the full MDE criteria and was assessed at baseline by the computerized version of the CIDI [@pone.0057510-Wittchen1]. MDE recurrence was assessed between baseline and three year follow-up. Recurrence was defined as the return of a MDE after partial or complete remission of at least six months. The authors provided us with the absolute numbers of MDE recurrence for patients with and without a somatic illness. Information on the quality criteria for all four included studies is presented in [table 2](#pone-0057510-t002){ref-type="table"}. 10.1371/journal.pone.0057510.t002 ###### Quality assessment of the included studies according to a modified version of the Newcastle-Ottowa Quality Assessment Scale for cohort studies. {#pone-0057510-t002-2} Domain Wells, Rogers, Burnam & Camp (1993) Gerrits, van Oppen, van Marwijk, van der Horst & Penninx (2013) Kovacs, Goldston, Obrosky & Drash (1997) Hardeveld, Spijker, De Graaf, Nolen & Beekman (2012) ----------------------------------------------------------------------------------------- ------------------------------------- ----------------------------------------------------------------- ------------------------------------------ ------------------------------------------------------ Representativeness of cohort **\*** **\*** **-** **\*** Selection of the non- exposed cohort **\*** **\*** **?** **\*** Ascertainment of exposure **-** **\*** **\*** **-** Comparability of groups with and without somatic illness on basis of design or analysis **\*** **\*** **\*** **\*** Assessment of depression at baseline (blinding demanded) **-** **?** **-** **-** Assessment of depression at follow-up (blinding demanded) **-** **-** **-** **-** Follow-up at least 6 months? **\*** **\*** **\*** **\*** Adequacy of follow-up of cohorts **?** **?** **?** **-** *Note,* \* = rated as meeting the quality criterion, - = rated as not meeting the quality criterion; ? = no information about quality criterion. Risk Ratios {#s3d} ----------- In [figure 2](#pone-0057510-g002){ref-type="fig"} an overview of the number of recurrences in both the patients with and without somatic illness are presented and the RR\'s of the four studies at all the follow-up intervals are shown in a forest plot. {#pone-0057510-g002} As shown in [table 1](#pone-0057510-t001){ref-type="table"} the reported recurrence rates in Wells et al. [@pone.0057510-Wells2] are lower at the one year follow-up but higher at the two year follow-up in the current IDDM group compared to the group without IDDM (14.9% vs. 29.3% at one year follow-up; 36.0% vs. 27.8% at two year follow-up*).* The calculated RR at one year follow-up is far below one, which might indicate less risk for recurrence of depression in the IDDM patients (n = 21) in comparison to the non-IDDM patients (n = 533). However at two year follow-up there is a shift to a somewhat higher than one RR, indicating more risk for depressive recurrence in the IDDM patients compared to the non-IDDM patients. In the IDDM group, the calculated RR\'s at one-and two year follow-up are based on recurrences of three and eight patients respectively. Also, the fact that the 95% CI contains the score of one in both cases leads us to conclude that there is no higher risk for two or more depressive episodes for patients with co-morbid IDDM compared to patients without co-morbid IDDM (RR = 0.49, 95% CI, 0.17--1.40; RR = 1.37, 95% CI, 0.78--2.41). Among the 715 participants in the study by Gerrits et al. [@pone.0057510-Gerrits1], 57 (8.0%) of the participants who did not have a co-morbid chronic somatic illness experienced a recurrence after remission versus 43 participants (6.0%) who did have one or more co-morbid chronic somatic illnesses. The RR of having one or more chronic somatic illnesses was around one (RR = 0.94, 95% CI, 0.65--1.36) meaning that in this sample there was no higher risk of recurrence after remission for patients with co-morbid chronic somatic illnesses. Kovacs et al. [@pone.0057510-Kovacs1] found similar recurrence rates in the illness group at one, -two -and 6.5 years of follow-up (respectively: 26% in the somatic illness group versus 22% in the control group; 30% in the somatic ill versus 32% in the control condition; 47% in both the somatically ill and the control condition). RR\'s range from 0.99--1.15, which again is around one and indicates that the presence of co-morbid IDDM therefore did not seem to heighten the risk of recurrence in this group at one-, two-, and 6.5 year follow-up respectively (RR = 1.15, 95% CI, 0.40--3.27; RR = 1.07, 95% CI, 0.48--2.42; RR = 0.99, 95% CI, 0.55--1.77). In the study of Hardeveld et al. [@pone.0057510-Hardeveld1] 135 patients (19.7%) experienced a recurrence. The median time to recurrence was six years (SD = 5.5). MDE recurrence rate was 21.1% for patients with a somatic illness compared to 18.3% for patients without a somatic illness. The RR of MDE recurrence was close to one (RR = 1.16, 95% CI, 0.86--1.57). This indicates that there is no significant heightened risk for a depressive recurrence, between baseline and three years of follow-up, for patients with a co-morbid somatic illness in comparison to patients without a co-morbid somatic illness. Discussion {#s4} ========== The aim of this systematic review was to determine whether having a co-morbid chronic somatic illness in MDD predicts a greater risk of depressive recurrence. Only four studies examined recurrence in patients with- and without co-morbid chronic somatic illnesses. Surprisingly, there was no indication that co-morbid somatic illness was associated with a greater risk for recurrence. Strengths and limitations {#s4a} ------------------------- We applied broad search terms and inclusion criteria, and acquired studies through three important databases, reference lists, guidelines and experts in the field. This led to the identification of a large number of articles that were reviewed in a systematic fashion by multiple reviewers. Most studies failed to meet the inclusion criteria, mainly because they did not include a reference group without a somatic illness. Rather, these studies presented results of depression and its course within a specific disease group only. Presence of a reference group is crucial in order to draw conclusions on the risk ratio for recurrence that is associated with co-morbid somatic illness. The included studies differed to a large extent in their methodological characteristics. Therefore it was not appropriate to do a meta-analysis and calculate a pooled risk ratio. The included studies had methodological problems; the quality assessment shows some of these limitations. Whereas baseline depression was defined by DSM criteria in all four studies that was not the case with respect to depressive recurrence at follow-up in the study by Wells et al. [@pone.0057510-Wells2]. Wells et al. [@pone.0057510-Wells2] assessed the number of patients with more than two depressive episodes during follow-up. Episodes were defined as periods of depression separated by at least two months of intervening remission as reported by the patient; no DSM criteria were applied to each separate episode. This is not in accordance with the other three studies that did identify each separate episode of depression by applying DSM criteria. Additionally, these episodes cannot be directly compared to MDE recurrences as reported in Gerrits et al. [@pone.0057510-Gerrits1], Kovacs et al. [@pone.0057510-Kovacs1] and Hardeveld et al. [@pone.0057510-Hardeveld1]. On the other hand, since Wells et al. [@pone.0057510-Wells2] reported exclusively more than two episodes during follow-up, the actual rate of recurrences could have been underestimated. Gerrits et al. [@pone.0057510-Gerrits1] used a life chart assessment to minimize the risk of missing possible recurrences [@pone.0057510-Lyketsos1]. Depressive symptomatology fluctuates over time (1) and can be easily overlooked if assessed only at fixed time points, like they did in three out of the four studies [@pone.0057510-Wells2], [@pone.0057510-Kovacs1], [@pone.0057510-Hardeveld1]. In addition, there are several differences between the studies that further complicated drawing firm conclusions on the overall effect of somatic illness on recurrence rates, such as, different follow-up times, received treatment, type and assessment method of somatic illnesses, depressive symptomatology at baseline, unequal sizes of somatic illness and reference groups (the most unequal sample had 21 somatic ill patients versus 533 patients without somatic illness in the study of Wells et al.) [@pone.0057510-Wells2]. Also there were differences in the choice and sizes of the reference groups (533, 397, 30 and 356 patients). Implications {#s4b} ------------ Current clinical guidelines, such as the APA [@pone.0057510-AmericanPsychiatric1] and the NICE [@pone.0057510-National1], [@pone.0057510-National2], identified the group of MDD patients with co-morbid chronic somatic illness as a "double trouble" group with poor prognosis. Prolonged pharmacological maintenance treatment has been recommended for these patients if they were treated with antidepressants during the acute phase of their depressive episode. Additional preventive psychological treatment was recommended for those who received psychotherapy during acute treatment. However, our review suggests that the number of studies that could provide such evidence is very small and that those few studies that are available provide no indication of any elevated risk of recurrence among depressed patients with co-morbid somatic illness. We therefore call for additional longitudinal studies about the impact of co-morbid chronic somatic illness on the course of depression. Apart from recurrence, other outcomes also are potentially relevant, such as the quality of life, severity and duration of episodes, hospitalization, sick leave from work and persistence of depression. An interesting question might be whether the prognosis of depression could deteriorate when somatic illness progresses and if so, whether treatment of the somatic illness can counter this process. Additionally, we need to study whether maintenance pharmacotherapy\'s, as well as psychological interventions, are (equally) effective for this presumed high risk group regarding recurrence. The authors wish to thank Marloes M. Gerrits, Brenda W. Penninx and their colleagues for provision of the depression recurrence data. Also we would like to thank Kenneth B. Wells, M.D., M.P.H. and his colleagues for providing us with details of their study. Finally, we want to thank Florian Hardeveld and colleagues for provision of the absolute depression recurrence data. [^1]: **Competing Interests:**The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: CB GK PC. Performed the experiments: GK WH CB. Analyzed the data: CB GK HB SH. Contributed reagents/materials/analysis tools: PC GP. Wrote the paper: GK CB HB WH GP PC SH.

2015 ACVIM Forum Research Reports Program Indianapolis, Indiana, June 3--6, 2015 Index of Abstracts {#jvim13002-sec-0001} =================================================================================================== **THURSDAY, JUNE 4**TimePresenting AuthorResearch Report Title**NEUROLOGY**3:10 PMSarah MooreA Simplified Method of Walking Track Analysis to Assess Locomotion in Dogs with Acute Spinal Cord Injury3:35 PMClare RusbridgeAttempts to Breed Out Chiari‐like Malformation: Is a Cross the Answer?4:25 PMPeter DickinsonCytogenic Alterations Define Tumor Subtypes and Key Pathways in Canine Primary Gliomas4:50 PMR. Timothy BentleyMetronomic Chlorambucil Chemotherapy for Canine Gliomas: A Phase I/II Clinical Trial5:25 PMC. Elizabeth BoudreauEvaluation of Systemic MicroRNA Administration for Immunomodulation in the Treatment of Canine Glioma5:50 PMRebecca PackerOnscreen‐Guided Brain Tumor Resection via Co‐Registration of a Resection Device and Neuronavigation System**FOOD ANIMAL**9:00 AMBenjamin NewcomerCharacterization of the Pathogenicity and Shedding of Bovine Parainfluenza 3 Genotypes9:25 AMKatharine SimpsonPercutaneous Tube Cystostomy Using a Suprapubic Catheter for Treatment of Obstructive Urolithiasis in Goats **FRIDAY, JUNE 5**TimePresenting AuthorResearch Report Title**CARDIOLOGY**2:10 PMVicky YangEffects of Canine Wharton\'s Jelly Mesenchymal Stem Cell Conditioned Media on Canine Valvular Interstitial Cells2:35 PMJoshua SternCardiac Biomarkers and Semi‐Dominance in a Genetic Model of Feline Hypertrophic Cardiomyopathy3:10 PMJeffrey SolomonSafety and Biocompatibility of the Mitrex^®^ Epicardial Annuloplasty Device in a Chronic Model3:35 PMColin SchwarzwaldEchocardiographic Assessment of Left Atrial Size in Horses: Diameter or Area, Does It Matter?4:25 PMTeresa DeFrancescoEvaluation of Point‐of‐Care Lung Ultrasound (VetBlue Protocol) for the Diagnosis of Cardiogenic Pulmonary Edema in Dogs and Cats with Acute Dyspnea4:50 PMRebecca StepienUse of NT‐proBNP and cTnI Concentrations to Detect Heart Disease in Whippets5:25 PMSimon SwiftStent Angioplasty for Treatment of Balloon Resistant Canine Valvular Pulmonic Stenosis5:50 PMVirginia Luis FuentesOnline Survey to Assess Inter‐ and Intra‐Observer Agreement on Echocardiographic Classification of Cardiomyopathy in Cats**ONCOLOGY**2:10 PMDouglas ThammAntineoplastic Effects of Auranofin in Canine Lymphoma Cells2:35 PMDouglas ThammRabacfosadine and Prednisone: Efficay of a Q21 Day Administration Schedule in Canine Lymphoma3:10 PMMario DoleraTotal Pelvic Irradiation with Simultaneous Boost in Canine Urinary Transitional Cell Carcinoma3:35 PMLuca MalfassiHypo‐Fractionated Stereotactic Brain‐Sparing Irradiation of Stage IV Canine Nasal Tumors4:25 PMJackie WypijInvestigation of the FOXM1 Transcriptional Pathway in Canine Lymphoma4:50 PMJackie WypijTubulin‐Modulating Effects of the Anthelmintic Mebendazole in Feline Cancer5:25 PMG.K. OgilvieAnti‐CD20 Monoclonal Antibody Canine Lymphoma Therapy: A Double‐Blind, Randomized, Placebo‐Controlled Study5:50 PMAmy LeBlancPreclinical Comparison of Three Indenoisoquinolines Candidates in Tumor‐Bearing Dogs**SMALL ANIMAL INTERNAL MEDICINE**8:00 AMKenny PatrickTreatment of Feline Hypersomatotropism‐efficacy, Morbidity, and Mortality of Hypophysectomy8:25 AMChen GilorDuration of Fasting But Not Diurnal Variation Affects the Response to Glucagon in Healthy Cats9:00 AMDavid GearingA Feline‐Specific Anti‐NGF Antibody is Safe and Effective for the Alleviation of Inflammatory Pain in Cats9:25 AMChristian LeuteneggerSpike Gene Mutations in Feline Coronavirus and Their Correlation to Feline Infectious Peritonitis4:25 PMLisa FreemanEvaluation of Weight Loss Over Time in Cats with Chronic Kidney Disease4:50 PMEva FurrowProteinuriaiIn Miniature Schnauzer Dogs with and without Hypertriglyceridemia5:25 PMNyssa Reine‐SalzPrevalence of Canine Infectious Respiratory Disease in Dogs in Chicago Outbreak**EQUINE**8:00 AMDaniela BedeniceClinical and Necropsy Evaluation of Acute Lung Injury and Acute Respiratory Distress Syndrome in Neonatal Foals8:25 AMErica McKenzieMetabolic Responses to a Standardized Exercise Test in Arabian Horses with Exertional Rhabdomyolysis9:00 AMAngelika SchosterLongitudinal Effect of a Multi‐Strain Probiotic on the Microbiota of Neonatal Foals9:25 AMMolly McCueSkeletal Muscle Gene Expression Profile in Type 1 Polysaccharide Storage Myopathy2:10 PMCarrie FinnoBlood and CSF Alpha‐tocopherol and Selenium Concentrations in Neonatal Foals with NAD2:35 PMCody AlcottTranscranial Magnetic Motor Evoked Potential Latencies and Axon Loss in Equine Cervical Stenotic Myelopathy3:10 PMSteeve GiguereNovel Determinant Confers Macrolide, Lincosamides, and Streptogramin B Resistance in *Rhodococcus Equi*3:35 PMAshley BoyleDetermining Optimal Sampling Site for *Streptococcus equi* Carriers Using a Loop‐mediated Isothermal PCR Assay4:25 PMBenjamin UbertiPro‐resolution Mechanisms of Inflammation in Equine Recurrent Airway Obstruction: Tamoxifen as a New Therapeutic Option4:50 PMVeronique LacombeImpact of Prolonged Hyperinsulinaemia on Insulin Signaling Pathway in Equine Striated Muscle and Digital Lamellae5:25 PMCarlos Medina‐TorresTissue Microdialysis Studies of Equine Lamellar Energy Metabolism and Microvascular Blood Flow5:50 PMMelody de LaatIncretins Potentiate the Insulin Response to Oral Glucose in Insulin‐Resistant Ponies**FOOD ANIMAL**2:10 PMDaniela BedenicePharmacokinetics of Extended Release Formulation of Eprinomectin in Adult Alpacas and Use in Alpacas with Mange2:35 PMMeredyth JonesDevelopment of a Zinc Implant‐Based Model for Urolithiais in Goats3:10 PMManuel ChamorroEfficacy of Different MLV Vaccines Administered to Calves Subsequently Challenged with BVDV or BVDV and BoHV‐13:35 PMMarie‐Eve FecteauComparison of Johne\'s Disease Prevalence on Organic and Conventional Dairy Farms in Pennsylvania4:25 PMMunashe ChigerweA Randomized Clinical Trial Evaluating Metabolism of Colostral and Plasma Derived Immunoglobulin G in Jersey Bull Calves4:50 PMChristian GerspachPlasma Lipidomic Profile in Cows During the Transition Period and in Cows with Fatty Liver5:25 PMSébastien BuczinskiBayesian Estimation of the Accuracy of Clinical Examination and Ultrasonography for Diagnosing Calves Pneumonia5:50 PMMaria PradoMilk Feeding Level and Housing Impact on Disease Incidence and Subsequent Productivity of Dairy Calves **SATURDAY, JUNE 6**TimePresenting AuthorResearch Report Title**CARDIOLOGY**8:00 AMKieran BorgeatFeline Hypersomatotropism Is a Naturally Occurring, Reversible Cause of Myocardial Remodeling8:25 AMLance VisserDiagnostic Value of Right Pulmonary Artery Distensibility Index in Dogs with Pulmonary Hypertension**NEUROLOGY**8:00 AMMario DoleraCanine Peripheral Nerve Sheath Tumors: MRI, Palliation, Surgery, and Stereotactic Radiotherapy8:25 AMMario DoleraPeripheral Glycemia in Dogs with Limb Thrombosis: A Prospective Study9:00 AMMario DoleraSurgical Stabilization of Canine Lumbosacral Spine with Stop Screws and Iliac Wing Screws9:25 AMFred WiningerFrameless Steriotaxy for Vertebral Implant10:30 AMRoger ClemmonsPLA Bead Treatment of Refractory Epilepsy in Dogs10:55 AMG. Diane SheltonThe Spectrum of Inherited Myopathies in Young Labrador Retrievers11:30 AMDennis O\'BrienMutation in Laryngeal Paralysis/Polyneuropathy with Ocular Abnormalities and Spongiform Encephalopathy11:55 AMDennis O\'BrienParoxysmal Non‐kinesogenic Dyskinesia in Soft Coated Wheaton Terriers: Mutation and Response to Acetazolamide4:25 PMByeong‐Teck KangEvaluation of Hemodynamic and Immunologic Changes in the Dog Brain After Experimental Intracerebral Hemorrhage4:50 PMStephanie ThomovskySerum Melatonin Levels in Normal Dogs and Dogs with Seizures**ONCOLOGY**8:00 AMNicola MasonCombination Lm‐LLO Immunotherapy Plus Radiation Delays Tumor Progression and Prolongs Survival in Osteosarcoma8:25 AMCailin HeinzeAssociation Between Body Condition Score and Survival in Dogs with Lymphoma and Osteosarcoma**SMALL ANIMAL INTERNAL MEDICINE**8:00 AMStephanie SmithNeutrophil Extracellular Trapp Formation is Increased in Dogs with Immune Mediated Hemolytic Anemia8:25 AMKatie TolbertFeline *T. foetus* Cytotoxicity Can Be Inhibited by Selective, Small‐molecule Cysteine Protease Inhibitors9:00 AMAllyson BerentOutcome Following Ureteral Stent Placement in Dogs for Benign Ureteral Obstructions: 44 Dogs (57 Ureters) 2009--20139:25 AMMahalakshmi YerramilliPrognostic Value of SDMA to Creatinine Ratio in Dogs and Cats with Chronic Kidney Disease10:30 AMJonathan FogleEpigenetics and CD8^+^ T Cell Dysfunction in an FIV Model10:55 AMPolina VishkautsanPharmacokinetics of Voriconazole in Healthy Cats11:30 AMRoschelle HeubergerPain Management and End of Life Care: Results of a National, Cross‐Sectional, Survey of Small Animal Owners11:55 AMHiroshi OkawaDiscovery and Clinical Effectiveness of a Composition That Promotes Hair Growth (Patent Pending)**EQUINE**9:00 AMNicola PusterlaUse of Daily Diclazuril Pelleted Top Dress for the Prevention of *Sarcocystis neurona* Infection in Foals9:25 AMMelody de LaatThe Effect of Oral and Intravenous Dextrose on C‐Peptide Secretion in Ponies10:30 AMDerek KnottenbeltThe Prevalence of Large Intestinal Mucosal Pathology in Horses10:55 AMJonathan ForemanMedical Alternatives to Conventional Cyclooxygenase Inhibitors for Treatment of Acute Foot Pain in Horses11:30 AMJulia MontgomeryInitial Characterization of the Tracheal Microbiomes in Healthy Horses and Horses with Heaves11:30 AMJulia MontgomeryInitial Characterization of the Tracheal Microbiomes in Healthy Horses and Horses with Heaves CARDIOLOGY {#jvim13002-sec-0002} ========== CANINE MITRAL VALVE INTERSTITIAL CELL GROWTH IS IMPROVED BY CANINE WHARTON\'S JELLY MESENCHYMAL STEM CELL CONDITIONED MEDIA {#jvim13002-sec-0003} --------------------------------------------------------------------------------------------------------------------------- **[Vicky Yang]{.ul}, Dawn Meola, Sarah Crain, Kristen Thane, Airiel Davis, Andrew Hoffman** Tufts University Cummings School of Veterinary Medicine, North Grafton, Massachusetts, USA Myxomatous mitral valve disease (MMVD) is the most common acquired cardiac disease in dogs and the most common cause of congestive heart failure in dogs. Chronic valvular disease will lead to valvular insufficiency, cardiac enlargement, and sometimes contractile dysfunction in the later stages of the disease. As the disease progresses, the amount of valvular regurgitation increases, eventually leading to volume overload, pulmonary venous congestion, and finally congestive heart failure with pulmonary edema. The common histologic findings seen in myxomatous valves include disarray of the collagen and elastin fibers as well as disruption of the microstructure within the valves. This disarray and disorganization is partly a result of phenotypic transdifferentiation of the valvular interstitial cells (VICs) from quiescent fibroblastic cells to myofibroblastic‐like cells, which follows the classical paradigm of fibrogenesis. As the transformation takes place, the VIC density decreases, and these cells then express alpha smooth muscle actin (α‐SMA) instead of vimentin. Similarly, TGFβ stimulation of VICs in vitro results in fibroblastic (vimentin^high+^, αSMA^low+^) to myofibrolastic (vimentin^low+^, αSMA^high+^) transition. Given that mesenchymal stem cells (MSC) can exert anti‐fibrotic effects, we investigated the effects of conditioned media (CM) derived from canine Wharton\'s Jelly MSCs (WJ‐MSC) on the growth potential of canine VICs and its ability to counter the effects of TGFβ. VICs were isolated and cultured from normal and diseased valves. Immunohistochemistry and real‐time (q‐) PCR were used to evaluate the expression of αSMA, vimentin, elastin, and collagen in VICs. CM was collected from WJ‐MSC culture, and CM exosomes were isolated by either ultrafiltration or ultracentrifugation. Cell growth and replicative capacity were evaluated using MTT assay and colony forming units (CFU). Internalization of WJ‐MSC exosomes by VICs was imaged with membrane or RNA staining. qPCR showed that diseased valve VICs had increased myofibroblastic phenotype and decreased cell growth by MTT and CFU. Culturing of VICs of either phenotype with WJ‐MSC CM or exosomes alone resulted in increased number of viable cells, and depletion of exosomes had the opposite effect. Exosomal membrane and RNA labelling confirmed exosome uptake into VIC cytoplasm and nucleus, although RNA containing exosomes were internalized by \<20% of VICs, implying selective endocytosis of RNA containing exosomes or they are less numerous. Increase in responsiveness to TGFβ1 stimulation when VICs were cultured in CM was noted. WJ‐MSC CM or its exosomes improve cell growth for VICs isolated from both normal and diseased canine mitral valves and may exert protective effects on VICs to slow the progression of canine MMVD. CARDIAC BIOMARKERS AND SEMI‐DOMINANCE IN A GENETIC MODEL OF FELINE HYPERTROPHIC CARDIOMYOPATHY {#jvim13002-sec-0004} ---------------------------------------------------------------------------------------------- **[Joshua Stern]{.ul}^1^, Kun‐Ho Song^1^, Eric Ontiveros^1^, Samantha Harris^2^** ^1^University of California Davis; School of Veterinary Medicine; Department of Medicine & Epidemiology, Daivs, California, USA, ^2^University of Arizona; College of Medicine; Department of Cellular and Molecular Medicine, Tucson, Arizona, USA A purpose‐bred colony of mixed breed cats harbors a mutation associated with development of hypertrophic cardiomyopathy (HCM) in Maine coons. HCM in Maine coon cats carrying the same A31P mutation in the cardiac regulatory protein myosin binding protein C is a disease with age‐related and incomplete penetrance. Here we hypothesized that cardiac biomarkers will reveal a semi‐dominant inheritance pattern in cats with the A31P mutation and may identify cats at risk for HCM development earlier than echocardiography. In 27 mixed‐breed cats, A31P genotype status, echocardiography, serum 2nd‐generation NTproBNP and plasma high‐sensitivity cardiac troponin‐I (CTNI) were analyzed. No cats were echocardiographically diagnosed with HCM in this study. 12 homozygous (HO), 9 heterozygous (HET) and 6 wild type (WT) cats were included. The median ages of WT, HET, and HO cats were 27, 31 and 15 months, respectively. The median and IQR for each biomarker was calculated for each group and compared for significant differences (*P* \< 0.05) with a Mann‐Whitney test. CTNI was not different between groups. NTproBNP values were as follows: WT 15 (12,16.75); HET 20 (16.5,27.0); HO 49 (32,131) and were significantly different for each genotype. In contrast to previous studies of A31P mutant cats, significant NTproBNP elevation was present in the absence of echocardiographic diagnosis of HCM. NTproBNP elevation follows a semi‐dominance model where heterozygous cats are less severely elevated than homozygous cats. This matches previous observations in clinical characterization of disease. In context of incomplete penetrance, NTproBNP may represent a tool for early identification of HCM affected cats. SAFETY AND BIOCOMPATIBILITY OF THE MITREX^®^ EPICARDIAL ANNULOPLASTY DEVICE IN A CHRONIC MODEL {#jvim13002-sec-0005} ---------------------------------------------------------------------------------------------- **[Jeffrey Solomon]{.ul}^1^, Thomas Fogarty^2^, Evan Anderson^1^, Pierluca Lombardi^3^** ^1^Infiniti Medical, Menlo Park, California, USA, ^2^Fogarty Institute for Innovation, Mountain View, California, USA, ^3^Maquet Cardiovascular, Wayne, New Jersey, USA This study evaluated the safety of the myocardial compression required to perform epicardial annuloplasty and the biocompatibility of the Mitrex^®^ device. Ten swine (seven test and three control) were used. The Mitrex^®^ device was placed in all subjects such that the septo‐lateral dimension of the mitral valve was reduced by 15--35%. Echocardiography and angiography were performed pre implant, post implant and at term. Test devices were secured in the test group and removed from the animals in the control group. Necropsy was performed at 180 days. Hearts were pressure fixed and analyzed. Test devices were placed without incident. Coronary flow, ejection fraction, left ventricular wall motion and mitral valve anteroposterior dimension were normal post implantation and at term. There were no remarkable postoperative events and all subjects survived to term with the exception of one test animal that was euthanized due to a non device related complication (refractory pleural effusion). Devices were well tolerated causing only minimal to mild fibrosis and chronic inflammation. No significant changes were observed in the myocardium except for muscle fiber atrophy near the tip of the anterior arm. There appeared to be ample tissue over the tip and no danger of perforation in all but one subject. No meaningful changes were noted in cardiac shape, ventricular wall thickness, chamber size, heart valves, and blood vessels. Myocardial compression necessary to perform epicardial annuloplasty was well tolerated. The Mitrex^®^ device was safe and biocompatible. ECHOCARDIOGRAPHIC ASSESSMENT OF LEFT ATRIAL SIZE IN HORSES: DIAMETER OR AREA, DOES IT MATTER? {#jvim13002-sec-0006} --------------------------------------------------------------------------------------------- **[Colin Schwarzwald]{.ul}, Iris Huesler** Vetsuisse Faculty, University of Zurich, Zurich, Switzerland Echocardiographic assessment of left atrial (LA) size is routinely done in horses. A variety of conventional (linear) and novel (area‐based) indices of LA size have been described, but their clinical use is not standardized and novel indices are poorly established. The goals of this study were to define reference intervals for indices of LA size in horses and to provide prove of concept for the use of area‐based indices of LA size in this species. The agreement between conventional linear measurements and novel area‐based indices of LA size was assessed in a population of healthy horses and horses with valvular regurgitation. Forty healthy horses (11 ± 4 years, 547 ± 84 kg) and 112 horses with mitral and/or aortic insufficiency (15 ± 6 years, 554 ± 76 kg) were included in this study. Echocardiographic examination was performed by a single operator (CCS) using a standardized protocol (GE Vivid 7 Dimension). The maximum LA diameter measured in a right‐parasternal long axis view (LADmax), the maximum LA diameter measured in a left‐parasternal long axis view (LADllx‐max), the maximum LA area measured in a right‐parasternal long axis view (LAAmax), and the maximum LA area measured in a right‐parasternal short axis view (LAsxAmax) were measured by the same operator. Measurements were allometrically scaled to a body weight of 500 kg. Reference intervals were calculated based on a subpopulation of 31 healthy Warmblood horses (Reference Value Advisor). Method agreement was investigated using linear regression, calculation of weighted Kappa (κ) and Bland‐Altman analyses. The level of significance was 0.05. For this study population of healthy and diseased horses, LADmax (reference interval 10.5--13.2 cm) was 0.9 (−0.9 to +2.7) cm \[mean bias (limits of agreement)\] smaller than LADllx‐max (reference interval 11.8--14.0 cm); they were significantly related (*P* \< 0.01, r^2^ = 0.45) and in moderate agreement (κ = 0.59) for classification of LA dimensions in reduced, normal, or increased size. Of the horses with a LADmax within normal limits, 10% (12/118) had a LADllx‐max above normal. Conversely, 6% (7/110) of horses with LADllx‐max within normal limits had a LADmax above normal. The LAAmax (reference interval 82--103 cm^2^) was 18 (−9 to +44) cm^2^ smaller than LAsxAmax (reference interval 84--134 cm^2^); they were significantly related (*P* \< 0.01, r^2^ = 0.51). LADmax was significantly associated with LAAmax (*P* \< 0.01, r^2^ = 0.80) and LAsxAmax (*P* \< 0.01, r^2^ = 0.50), respectively. However, out of all horses that had a LADmax within normal limits, 9% (11/121) were diagnosed with an enlarged LA based on LAAmax (κ = 0.76) and 3% (4/121) had an enlarged LA based on LAsxAmax (κ = 0.48). Of the horses with both LADmax and LADllx‐max within normal limits, 7% (7/104) had an enlarged LA based on LAAmax and 3% (3/104) had an enlarged LA based on LAsxAmax, respectively. Of the horses with both LAAmax and LAsxAmax within normal limits, 2% (2/94) had an enlarged LADmax and 6% (6/94) had an enlarged LADllx‐max. Method agreement between LAAmax and LAsxAmax was fair (κ = 0.29). Of the horses with a LAAmax within normal limits, 4% (4/110) had a LAsxAmax above normal. Conversely, 17% (23/132) of horses with LAsxAmax within normal limits had a LAAmax above normal. In conclusion, linear measurements and area‐based measurements of LA size are not always in good agreement. Therefore, LA enlargement should not be diagnosed based on a single uni‐dimensional measurement. Instead, multiple measurements of LA size, including linear and area‐based variables, should be obtained to complement subjective assessment of LA dimensions in clinical patients. EVALUATION OF POINT‐OF‐CARE LUNG ULTRASOUND (VETBLUE PROTOCOL) FOR THE DIAGNOSIS OF CARDIOGENIC PULMONARY EDEMA IN DOGS AND CATS WITH ACUTE DYSPNEA {#jvim13002-sec-0007} --------------------------------------------------------------------------------------------------------------------------------------------------- **Jessica Ward^1^, Greg Lisciandro^2^, Sandra Tou^1^, Bruce Keene^1^, [Teresa DeFrancesco]{.ul}^1^** ^1^NC State University College of Veterinary Medicine, Raleigh, North California, USA, ^2^Hill Country Veterinary Specialists & FASTVetTM, San Antonio, Texas, USA Point‐of‐care lung ultrasound (LUS) is an emerging imaging technique that can suggest the presence of cardiogenic pulmonary edema (CHF) by identifying ultrasound artifacts (B‐lines) caused by interstitial or alveolar fluid. This study was designed to determine the accuracy of a protocolized LUS technique for diagnosing CHF in dyspneic dogs and cats. Seventy‐six dogs and 24 cats were enrolled for evaluation of acute dyspnea. Exclusion criteria included trauma, pleural effusion, and the lack of a thoracic radiograph within 6 hours of LUS. Patients underwent LUS, quantifying the presence of B‐lines at 4 sites on each hemithorax. An individual site was scored as positive if \>3 B‐lines were observed. LUS with ≥2 positive sites on each hemithorax was considered positive for CHF. Medical records were then evaluated for final diagnosis. Patients with CHF had a higher number and different distribution of positive LUS sites compared to patients with noncardiac disease. Sensitivity and specificity of LUS for diagnosing CHF was 84% and 72%, respectively. When considering cats only, sensitivity and specificity of LUS was 87% and 89%, respectively. Diagnostic accuracy of LUS was similar to thoracic radiographs. LUS tended to misdiagnose CHF in cases of diffuse interstitial/alveolar disease, such as ARDS. Inter‐observer variability for quantification of B‐lines was low (kappa statistic \>0.85). In conclusion, LUS was useful in predicting CHF as the cause of dyspnea, particularly in cats. Considering the utility and rapidity of this technique, point‐of‐care LUS should be considered as a diagnostic tool for dyspneic veterinary patients. USE OF NT‐PROBNP AND CTNI CONCENTRATIONS TO DETECT HEART DISEASE IN WHIPPETS {#jvim13002-sec-0008} ---------------------------------------------------------------------------- **[Rebecca Stepien]{.ul}^1^, Virginia Luis Fuentes^2^, Heidi Kellihan^1^** ^1^University of Wisconsin School of Veterinary Medicine, Madison, Wisconsin, USA, ^2^Royal Veterinary College, London, UK Systolic heart murmurs in whippets may be due to myxomatous mitral valve disease (MMVD) or may be unassociated with any structural heart disease (HD). Echocardiography (ECHO) may be required to differentiate MMVD from functional murmurs in screening programs but is not always available. In this pilot study, we compared the ability of NT‐proBNP (BNP) and cTnI concentrations versus auscultation to predict ECHO‐documented HD in a cohort of outwardly healthy whippets screened at a national show. One hundred thirty five dogs underwent auscultation, ECHO and blood sampling for BNP and cTnI analysis. ECHO results were reviewed to identify the presence of HD based on presence of any of the following: mitral valve prolapse, mitral regurgitation, left atrial/ventricular dilation and presence of ventricular premature complexes during ECHO exam, and categorized as ECHO negative (ECHO−) or positive (ECHO+) for evidence of HD. Blood test results were analyzed (*P* \< 0.05) for all dogs as a group (n = 135), and for dogs with heart murmurs (n = 73). \[BNP\] is expressed as pmol/L; \[cTnI\] as nmol/mL. ECHO+ prevalence in this population was 47/135 (35%). 73/135 (54%) dogs had systolic murmurs (MUR), 41/73 (30%) left basilar (LB) and 32/73 (24%) left apical (LA). ECHO+ prevalence in MUR dogs was 55% (40/73). Median \[BNP\] was higher in dogs with LA (634 \[range 169--2,303\], *P* \< 0.0001) versus no murmur (304 \[112--1,048\]) or LB (279 \[164--804\]). \[BNP\] was higher in ECHO+ dogs (574 \[169--2,303\], n = 47) versus ECHO− (297 \[112--988\], n = 88, *P* \< 0.0001). In MUR dogs, \[BNP\] was significantly higher in ECHO+ dogs (578 \[169--2,303\], n = 40) versus ECHO− (364 \[164--804\], n = 33, *P* = 0.004). Median \[cTnI\] was higher in LA dogs (0.07 \[0.02--0.26\], *P* \< 0.0001) versus those with no murmur (0.03 \[0.01--0.25\]) or LB (0.03 \[0.01--0.15\]). \[cTnI\] was higher in ECHO+ dogs (0.06 \[0.01--0.26\], n = 47) versus ECHO− (0.03 \[0.01--0.25\], n = 88, *P* \< 0.0001). In MUR dogs, \[cTnI\] was higher in ECHO+ dogs (0.07 \[0.02--0.26\]) versus ECHO− (0.03 \[0.01--0.07\], *P* \< 0.0001). ROC analysis was used to predict diagnostic cut‐off values for \[BNP\] and \[cTnI\]. Test characteristics are reported below:SeSpPPVNPVAUC*P* valueAll dogs n = 135 (ECHO+ disease prevalence: 35%) \[BNP\] \> 575 pmol/L468869730.70\<0.0001 \[cTnI\] \>0.065 nmol/mL429891740.83\<0.0001 Presence of any MUR85635589NA\<0.0001Dogs with MUR n = 73 (ECHO+ disease prevalence: 55%) \[BNP\] \>575 pmol/L538883430.700.004 \[cTnI\] \>0.065 nmol/mL539795630.87\<0.0001 Presence of LA murmur81969873NA\<0.0001 The results of this pilot study suggest that in the absence of auscultation, both \[BNP\] and \[cTnI\] may be useful to distinguish between whippets with HD from those without HD. If murmur status is known, the PPV of both tests is improved. STENT ANGIOPLASTY FOR TREATMENT OF BALLOON RESISTANT CANINE VALVULAR PULMONIC STENOSIS {#jvim13002-sec-0009} -------------------------------------------------------------------------------------- **[Simon Swift]{.ul}^1^, Ivan Sosa^1^, Amara Estrada^1^, Ashley Jones^1^, Curt Fudge^2^** ^1^College of Veterinary Medicine, Gainesville, Florida, USA, ^2^Congenital Heart Center, Gainesville, Florida, USA Canine valvular pulmonic stenosis has been categorized based on the degree of leaflet thickening, leaflet fusion and pulmonary artery hypoplasia. Dogs with mild and moderate disease have normal life expectancies but those with severe disease die prematurely. Balloon dilation is often successful in severe cases with thin fused leaflets, resulting in decreased clinical signs and increased life expectancy. Dogs with thickened, dysplastic valves and pulmonary artery hypoplasia show limited response to balloon dilation. Two case reports describe the use of stents to treat obstruction to pulmonary blood flow in dogs. In one, 2 dogs with supravalvular stenosis were treated successfully with stents. In the second, 2 dogs with severe dysplastic pulmonary valve stenosis initially did well, but stenosis recurred within 6 months. We describe the use of bare metal stents to treat severe dysplastic pulmonary valve stenosis in 3 dogs. All dogs initially had a single stent implanted with a good reduction in pressure gradients. This has been maintained long term in 2 dogs. One dog suffered a stent fracture resulting in stent embolization. That dog subsequently underwent repeat cardiac catheterization with four additional stents placed in the right ventricular outflow tract to treat severe, dynamic subvalvular pulmonic stenosis. All dogs continue to receive atenolol and clopidogrel. Our findings suggest that in selected patients, stenting of severe dysplastic pulmonary valve stenosis is a viable option and can provide long term relief of the obstruction. ONLINE SURVEY TO ASSESS INTER‐ AND INTRA‐OBSERVER AGREEMENT ON ECHOCARDIOGRAPHIC CLASSIFICATION OF CARDIOMYOPATHY IN CATS {#jvim13002-sec-0010} ------------------------------------------------------------------------------------------------------------------------- **Lois Wilkie^1^, [Virginia Luis Fuentes]{.ul}^1^, Mark Rishniw^2^** ^1^The Royal Veterinary College, Hatfield, Hertfordshire, UK, ^2^Cornell University, Ithaca, New York, USA Aim: to investigate inter‐ and intra‐observer agreement on echocardiographic classification of cardiomyopathy in cats. An online survey was devised with video loops and still images of echocardiographic recordings compiled from 21 cats with a range of cardiac phenotypes. Signalment and a brief clinical history were provided. The survey was distributed via the Veterinary Information Network (VIN) to members of a cardiology list serve. A list of 13 possible diagnoses was provided and participants were asked to indicate which described each case most accurately. Selection of multiple diagnoses was not allowed and respondents were asked to include a brief justification for their choice. Intra‐observer agreement was assessed by repeating 4 of the cases. Fleiss kappa (κ) was calculated for all participants with complete responses to determine inter and intra‐observer agreement. In order to assess whether agreement was greater among board‐certified cardiologists, agreement was also calculated using only responses from ACVIM or ECVIM diplomates. The survey was attempted by 86 participants routinely involved in echocardiographic assessment of cats with myocardial disease. Participants' clinical experience in veterinary cardiology ranged from 6 months to 42 years; 49% were ACVIM diplomates and 25% were ECVIM diplomates. No case was scored with perfect inter or intra‐observer agreement. Overall, inter‐observer agreement ranged from κ0.17 to 0.96 depending on the case, and intra‐observer agreement κ0.33 to 0.67. Among ACVIM/ECVIM diplomates, inter‐observer agreement was κ0.18 to 0.95 and intra‐observer agreement κ0.29 to 0.67. The observed range of inter‐observer Fleiss kappa values demonstrates that classification of cats with cardiomyopathy was inconsistent between participants, suggesting that different criteria may be in use. Intra‐observer agreement was also poor, suggesting that irrespective of the diagnostic criteria used, they are inconsistently applied. Agreement might be improved by standardizing the echocardiographic criteria used to diagnose different cardiomyopathy phenotypes. FELINE HYPERSOMATOTROPISM IS A NATURALLY OCCURRING, REVERSIBLE CAUSE OF MYOCARDIAL REMODELING {#jvim13002-sec-0011} --------------------------------------------------------------------------------------------- **[Kieran Borgeat]{.ul}^2^, Stijn Niessen^1^, Christopher Scudder^1^, Ruth Gostelow^1^, Sophie Keyte^1^, Julia Sargent^1^, Patrick Kenny^1^, Yaiza Forcada^1^, David Church^1^, Virginia Luis Fuentes^1^, David Connolly^1^** ^1^Royal Veterinary College, London, UK, ^2^Highcroft Veterinary Referrals, Bristol, UK In humans, hypersomatotropism caused by a functional pituitary mass is recognized as a cause of increased left ventricular (LV) mass, and cardiovascular complications are a major cause of morbidity and mortality. We hypothesized that feline hypersomatotropism was associated with increased LV wall thickness and left atrial (LA) dilation, compared to control groups of non‐hypersomatotropic diabetic and age‐matched healthy cats. Cats with confirmed hypersomatotropism (IGF‐1 \>1,000 ng/mL and pituitary mass; n = 57) were prospectively recruited, as were two control groups: diabetic cats (IGF‐1 \<1,000 ng/mL; n = 27) and healthy cats with no history of diabetes or cardiovascular disease (n = 41). Cats with other endocrinopathies were excluded. Echocardiography was performed in all cases and studies were measured by one trained operator. Normally distributed continuous variables were compared using a one‐way ANOVA. Non‐normally distributed variables were compared using a Kruskal‐Wallis test. Categorical data were compared using Chi‐square tests. Paired data were compared using Wilcoxon\'s signed rank test. Significance was *P* \< 0.05, with correction for pairwise comparisons. There was no age difference between groups (*P* = 0.243). Cats with hypersomatotropism had a greater maximum LV wall thickness (6.7 mm, 4.1--10.1 mm) than diabetic (5.2 mm, 4.0--9.1 mm; *P* \< 0.001) or normal cats (5.3 mm, 3.9--6.5 mm; *P* \< 0.001). LA diameter was greater in cats with hypersomatotropism (17 mm, 14.1--29.5 mm) than in diabetic (15.3 mm, 10.1--21.3 mm; *P* \< 0.001) and healthy cats (15.8 mm, 11.2--21.5 mm; *P* \< 0.001). Aortic insufficiency was more common in both cats with hypersomatotropism and diabetes than in normal cats (*P* \< 0.001). After hypophysectomy (n = 8), echocardiographic changes were mostly reversible (Figure 1). Hypersomatotropism is a naturally occurring, reversible cause of LV hypertrophy and LA dilation in cats. DIAGNOSTIC VALUE OF RIGHT PULMONARY ARTERY DISTENSIBILITY INDEX IN DOGS WITH PULMONARY HYPERTENSION {#jvim13002-sec-0012} --------------------------------------------------------------------------------------------------- **[Lance Visser]{.ul}, Minu Im, Joshua Stern** Department of Medicine & Epidemiology, School of Veterinary Medicine, University of California Davis, Davis, California, USA We sought to determine the value of right pulmonary artery distensibility index (RPADi) for the prediction of Doppler‐derived estimates of systolic pulmonary artery pressure (sPAP) in dogs with pulmonary hypertension (PH) compared to other echocardiographic indices of PH. Dogs with tricuspid regurgitation (TR) permitting Doppler‐derived estimates of sPAP were prospectively recruited and grouped into control (n = 20; sPAP \<30 mmHg) and dogs with mild (n = 12; sPAP 30--50 mmHg), moderate (n = 12; sPAP 50--75 mmHg), and severe (n = 14; sPAP \>75 mmHg) PH. Indices of PH quantified were RPADi (percent change in diameter of the PA from systole to diastole), pulmonary artery‐to‐aortic diameter (PA:Ao), acceleration time to peak PA flow velocity (AT), and AT‐to‐ejection time of PA flow (AT:ET). Associations between indices of PH and sPAP, right ventricular fractional area change (FAC), age, gender, heart rate, and body weight were performed using linear regression. Receiver operating characteristic analysis was performed to determine the optimal cutoff values for the indices of PH in the prediction of moderately increased sPAP (\>50 mmHg). RPADi (r = −0.89) showed the strongest correlation to sPAP followed by PA:Ao (r = 0.74), AT and AT:ET (both r = −0.69), and FAC (r = −0.52). AT weakly (r = −0.42) correlated with heart rate. No other significant correlations were identified. Cutoffs to predict moderate PH were defined for RPADi (29%; sensitivity \[Sn\] 81%; specificity \[Sp\] 94%), AT:ET (0.30; Sn 65%; Sp 100%), AT (54 ms; Sn 81%; Sp 83%), and PA:Ao (1.04; Sn 100%; Sp 67%). RPADi may be useful and predictive of PH in dogs if TR is absent. NEUROLOGY {#jvim13002-sec-0013} ========= EVALUATION OF SYSTEMIC MICRORNA ADMINISTRATION FOR IMMUNOMODULATION IN THE TREATMENT OF CANINE GLIOMA {#jvim13002-sec-0014} ----------------------------------------------------------------------------------------------------- **[C. Elizabeth Boudreau]{.ul}^1^, Nasser Yaghi^2^, Padmanabh Chivukula^3^, Xiaoyang Ling^2^, Brian Porter^1^, Gwendolyn Levine^1^, Joseph Payne^3^, Amy Heimberger^2^, Jonathan Levine^1^** ^1^College of Veterinary Medicine and Biomedical Sciences, Texas A&M University, College Station, Texas, USA, ^2^University of Texas MD Anderson Cancer Center, Houston, Texas, USA, ^3^Arcturus Therapeutics, San Diego, California, USA The signal transducer and activator of transcription 3 (STAT3) pathway is a key regulator of tumorigenesis and tumor‐mediated immune suppression. MicroRNA (miR)‐124 has been shown to be an inhibitor of STAT3 signaling. In rodent models of glioma, systemic administration of nanoparticle‐encapsulated miR‐124 (designated LUNAR‐301) results in down‐regulation of STAT3, up‐regulation of anti‐tumor immune effector responses, tumor regression, and enhanced survival. Additionally, miR‐124 can reverse immune suppression and T‐cell anergy in immune cells from human glioblastoma patients. We investigated STAT3 and miR‐124 expression levels in canine astrocytomas (n = 28) and normal brain (n = 6) to determine if these tumors share key biologic features with human and rodent gliomas. We found that miR‐124 expression is absent and STAT3 is increased in canine gliomas relative to normal brain. The level of STAT3 expression and the number of glioma‐infiltrating CD3+ T cells directly correlated with astrocytic grade. Next we investigated the safety and pharmacokinetics of LUNAR‐301 in healthy dogs (n = 5). No significant adverse events or toxicity was noted during single‐dose escalation or during sustained dosing. LUNAR‐301 induced in vivo up‐regulation of immune effector responses as detected by flow cytometry. Pharmacokinetic studies confirmed delivery of miR‐124 to the immune compartment during intravenous administration of LUNAR‐301. Cumulatively, these data indicate that STAT3 is an operational therapeutic target in canine high‐grade gliomas and that systemic administration of LUNAR‐301 is safe and feasible. We are initiating a phase I/II clinical trial in client‐owned canines with spontaneously arising high‐grade gliomas to ascertain if LUNAR‐301 can induce radiographic regression and prolong survival. ONSCREEN‐GUIDED BRAIN TUMOR RESECTION THROUGH REGISTRATION OF A VARIABLE‐SUCTION TISSUE RESECTION DEVICE WITH A NEURONAVIGATION SYSTEM {#jvim13002-sec-0015} -------------------------------------------------------------------------------------------------------------------------------------- **[Rebecca Packer]{.ul}, Stephanie Engel** College of Veterinary Medicine and Biomedical Sciences, Colorado State University, Fort Collins, Colorado, USA This study describes a surgical technique in which a variable‐suction tissue resection device is linked to a neuronavigation system for direct onscreen guidance of brain tumor excision. This technique is considered the first step towards minimally‐invasive, guided neurosurgery to excise otherwise inaccessible deep intraaxial tumors. Goals of this study were 1) improved access to deep or poorly accessible masses, 2) reduced surgical trauma by using targeted approaches and improved resection tools. A retrospective evaluation of 7 dogs and 1 cat that underwent brain tumor excision using the NICO^®^ Myriad (NICO Corporation) and Brainsight neuronavigation system (Rogue Research) was performed. The patient and resection instrument were registered to the neuronavigation system. Surgery was guided by real‐time onscreen visualization of the resection instrument position relative to the pre‐operative MRI images, cross‐referenced with direct visualization where possible. In 7 of 8 cases degree of resection was evaluated on post‐operative MRI: gross‐total resection (GTR, no residual tumor evident), near‐total resection (NTR, \<100% but ≥90% resection), or sub‐total resection (STR, \<90% resection). Of these 7 cases, 2 achieved GTR (1 equivocally), 3 achieved NTR (99%, 91%, 90%), 2 achieved STR (35%, 26%). Neuronavigation‐guided resection of cortical and subcortical brain tumors using the NICO^®^ Myriad was feasible, and study goals were achieved in part. Some refinements are required. Further studies will adapt this technique to minimally‐invasive port‐based surgical approaches for deep brain tumor resection (BrainPath^™^), to improve visibility and minimize effects of brain shift. The impact on clinical outcome for minimally‐invasive approaches must also be determined. A SIMPLIFIED METHOD OF WALKING TRACK ANALYSIS TO ASSESS LOCOMOTION IN DOGS WITH ACUTE SPINAL CORD INJURY {#jvim13002-sec-0016} -------------------------------------------------------------------------------------------------------- **Rachel Song, Maureen Oldach, Ronaldo da Costa, [Sarah Moore]{.ul}** The Ohio State University, Columbus, Ohio, USA The utility of a "finger painting" technique for walking track analysis was evaluated in 20 control dogs and 29 dogs with acute thoracolumbar spinal cord injury (SCI) caused by spontaneous intervertebral disc extrusion (IVDE). Stride length (SL), base‐of‐support (BS) and the co‐variance (COV) for both parameters were measure in all four limbs at three separate time points in normal dogs and on days 3, 10 and 30 following laminectomy in dogs with SCI. SL and BS were compared between control and SCI‐affected dogs at each time point during recovery. *P* \< 0.05 was considered a statistically significant difference. Mean SL (cm) for control dogs was 43.69, 43.68, 43.43, 43.54 for the left thoracic limb (TL), right TL, left pelvic limb (PL), and right PL respectively. COV‐SL was 0.13 in each of four limbs in control dogs. Mean SL of all four limbs was significantly shorter in SCI‐affected dogs at day 3, 10 and 30 compared to normal dogs. The mean difference in SL between normal and SCI‐affects dogs declined significantly over time with recovery from injury. BS‐TL (cm) was significantly wider in SCI‐affected dogs compared to controls at days 3 and 30 after surgery. BS‐PL (cm) did not differ between groups at any time‐point. These findings support the utility of this simplified method of walking track analysis to compare differences in pelvic limb SL between normal and SCI‐affected dogs, and to assess changes in SL as a marker of recovery after SCI. ATTEMPTS TO BREED OUT CHIARI‐LIKE MALFORMATION: IS A CROSS THE ANSWER? {#jvim13002-sec-0017} ---------------------------------------------------------------------- **Susan Knowler^1^, Henny van den Berg^1^, Eric Noorman^2^, Roberto La Ragione^1^, [Clare Rusbridge]{.ul}^1^** ^1^University of Surrey, Guildford, UK, ^2^Dierenkliniek den heuvel, Best, The Netherlands Chiari‐like malformation (CM) and syringomyelia (SM) are complex inherited disorders observed most commonly in toy breed dogs and can cause a significant loss in quality of life though pain and disability. Brachycephalicism is a risk factor and it has been suggested that crossbreeding with a different breed/s then backcrossing may produce individuals free of disease. This two and a half year project took advantage of a cross between a mesaticephalic normal Australian terrier and CM affected Griffon Bruxellois (GB) with subsequent backcrossing to a GB to investigate the inheritance and phenotype of these conditions and a means of reducing the incidence of CM and risk of SM. The study cohort comprised 2 control dogs (CM affected Australian terrier and unrelated GB without CM) and a single of family of 29 dogs, 12 of which were used in 8 different mating combinations resulting in 19 progeny. T1‐weighted sagittal DICOM images were analyzed for traits (2 angles, 2 lines and a "best fit" circle diameter) shown previously to have the greatest significance for CM in the GB. The quantitative findings in this study revealed these traits to be significant for CM‐affectedness. Furthermore the external phenotypes showed that by outcrossing breed types and selecting appropriate conformation characteristics in the F1 generation, it is possible to regain the GB breed standard in the F2 generation and reduce the degree of CM. However this is dependent on careful selection of conformation and screening for CM and SM at 1 year of age. The 4 dogs affected with SM in the study all exhibited reduced caudal skull development compared to their relatives. We showed that traits on MR images were useful to distinguish the phenotype and these exhibited segregation and may be additive towards the severity of CM. It suggests such traits might be useful to quantify the condition and the risk to SM. We propose that grading of CM takes account of quantitative traits that can be used in Estimated Breeding Values (EBV) to assist breeders with their mate selections. Such a system will have to be verified to ensure appropriateness for all breeds at risk. CYTOGENETIC ALTERATIONS DEFINE TUMOR SUBTYPES AND KEY PATHWAYS IN CANINE PRIMARY GLIOMAS {#jvim13002-sec-0018} ---------------------------------------------------------------------------------------- **[Peter Dickinson]{.ul}, Daniel York, Robert Higgins, Richard Lecouteur, Danika Bannasch** UC Davis School Of Veterinary Medicine, Davis, California, USA Defining commonly occurring cytogenetic abnormalities in canine glioma is essential to an understanding of gliomagenesis, and the definition of appropriate therapeutic strategies. We used Illumina 173K HD SNP arrays to determine copy number alterations and allelic imbalances in 38 histologically confirmed spontaneous canine gliomas (11 astrocytomas, 23 oligodendrogliomas, 4 mixed oligoastrocytomas and the J3T cell line). Copy number calls and allelic events were determined by matched paired analysis using BioDiscovery Nexus Copy Number^™^ software. Hierarchical clustering by aberration profile revealed two major groups with 5/6 glioblastomas and 2/3 grade II astrocytomas comprising one group. Comparative analysis showed loss of the *INK4A* locus (*CDKN2A*/*CDKN2B* tumor suppressor genes) was significantly associated with glioblastomas. Querying for aberrant events in specified genes showed common losses/loss of heterozygosity of tumor suppressor genes and amplification of oncogenes in the TP53, RB1 and PI3K/AKT/RAS pathways. Specifically, loss of *TP53*,*CDKN2A*/P14, *CDKN1A*/P21 and gain of *MDM4* was present in the TP53 pathway; loss of *RB1*,*CDKN2A*/P16, *CDKN2B*/P15, and *CDKN2C*/P18 was present in the RB1 pathway and loss of *PTEN* and gain of *PDGFRA* and *FGFR1* was present in the PI3K/AKT/RAS pathways. Evaluation of oligodendrogliomas for deletions syntenic to the commonly described 1p loss in human oligodendrogliomas revealed a syntenic loss in CFA 5 in 36% of canine tumors. These preliminary data suggest that canine gliomas have pathway alterations in common with human tumor counterparts, however further definition of the pathways in greater detail is warranted to inform appropriate choice of targeted therapies and interpretation of future therapeutic trials. METRONOMIC CHLORAMBUCIL CHEMOTHERAPY FOR CANINE GLIOMAS: A PHASE I/II CLINICAL TRIAL {#jvim13002-sec-0019} ------------------------------------------------------------------------------------ **[R. Timothy Bentley]{.ul}^1^, Aaron Cohen‐Gadol^2^, David Jones^2^, Deborah Knapp^1^** ^1^Purdue Veterinary Medicine, West Lafayette, Indiana, USA, ^2^Indiana University, Indianapolis, Indiana, USA Gliomas in dogs are typically fatal and limited information is available to guide therapy. Spontaneous brain tumors in client‐owned dogs have been suggested as a translational model for the development of human therapies. Chlorambucil is reported to display negligible penetration of the blood‐brain‐barrier in normal animals, but penetration of the blood‐brain tumor‐barrier has not been investigated. The purpose of this translational study was to perform a safety and pharmacokinetic study of metronomic chlorambucil chemotherapy in dogs with gliomas. Four client‐owned dogs with high‐grade gliomas began receiving metronomic chlorambucil (4 mg/m^2^ PO q 24 hours) and underwent surgical resection. The chlorambucil concentrations of serum, cerebrospinal fluid and surgical brain tumor specimens were analyzed. Patients additionally received CCNU post‐operatively and were monitored with monthly examinations and laboratory work. Repeat magnetic resonance imaging was performed. One patient suffered a post‐operative cerebrovascular accident and two are currently in complete remission. A fourth was euthanized after 5 months and brain tumor was harvested for repeat chlorambucil concentration analysis. No increase in seizures was observed and no cytopenias or other major chemotherapy complications were detected. Brain tumor chlorambucil concentrations ranged from below the limit of quantification (0.024 ng/specimen) to well in excess of the serum concentration of the same patient, indicating very variable penetration of the blood‐brain tumor‐barrier. Chlorambucil could not generally be detected in the cerebrospinal fluid. Metronomic chlorambucil chemotherapy is well tolerated in dogs with glioma and may penetrate the blood‐brain tumor‐barrier in individual cases. There is preliminary evidence of activity. CANINE PERIPHERAL NERVE SHEATH TUMORS: CLINICAL ASPECTS, MAGNETIC RESONANCE IMAGING FINDINGS AND COMPARISON OF PALLIATION, SURGERY AND STEREOTACTIC RADIOTHERAPY {#jvim13002-sec-0020} ---------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Mario Dolera]{.ul}, Luca Malfassi, Simone Pavesi, Massimo Sala, Giovanni Mazza, Silvia Marcarini, Nancy Carrara, Sara Finesso** La Cittadina Fondazione Studi e Ricerche Veterinarie, Romanengo, Italy No updates for canine peripheral nerve sheaths tumor (PNST) appeared in recent literature. The aim of this study was to evaluate the correlation between clinical aspects and MRI findings of tumors involving a major peripheral nerve, plexus or root and to determine the survival time in dogs treated with palliation, surgery or stereotactic radiotherapy (SRT). Records of dogs with PNST evaluated from 2000 to 2014 were reviewed to determine signalment, duration of clinical signs, neurological examination, MRI features, treatment option (palliation, surgery, stereotactic hypo fractionated radiotherapy). Time to first event, survival times and statistical differences across categories were calculated by the Kaplan‐Meier product limit method and log‐rank test. Forty‐seven dogs (median age 9 years, male:female ratio 1.76) were included, with Labrador retrieveroverrepresented (17%). Roots lesions were the most frequent (46.8%), with C5‐T1, V nerve and left side more involved (25.5%, 19.1% and 61.7%). Presenting sings were lameness, paresis and pain. Mean duration of clinical signs was 90 days. MRI findings comprises increased diameter, hyper intense and contrast enhancing nerve roots (57.1%), plexus or peripheral nerve (42.9%), focal hypomiotropy and muscle hyper intensity (73%). The time to first event was 30 days after surgery and 240 days after SRT. Overall mean survival was 97, 144 and 371 days with palliation, surgery and SRT. A predilection for Labrador retriever is observed. Comparing our results with published data, SRT seem to promise better results than palliation or surgery and warrant further evaluation. PERIPHEAL GLYCAEMIA IN DOGS WITH LIMB THROMBOSIS: A PROSPECTIVE STUDY {#jvim13002-sec-0021} --------------------------------------------------------------------- **[Mario Dolera]{.ul}, Luca Malfassi, Roberto Vailati Facchini, Sara Finesso, Giovanni Mazza** La Cittadina Fondazione Studi e Ricerche Veterinarie, Romanengo, Italy The aim of this study was to document the pheripheral glycaemia variations in hypoperfused limbs of patients affected by Magnetic Resonance Imaging (MRI)‐confirmed arterial thrombosis. Eleven dogs were recruited. Inclusion criteria were a clinical examination supportive of limb hypoperfusion and availability of blood cell count, biochemical profile and urine analyses. Two blood samples were sampled, one from the affected limb and one from a healty limb. Plasmatic glycaemia was measured using an automated glucose analyser. All the patients underwent a total body MRI that provided the final diagnosis. The thrombus was located: in the abdominal aorta (7/11), in the subclavian artery (1/11), in the axillary artery (1/11), in the iliac arteries (2/11). Of the total abdominal aortic thrombosis, 3/7 involved also the internal iliac arteries, 2/7 the external ones and 2/7 both. The extent of the thrombosis was classified as grade 1 when the greatest portion of the thrombus did not reach half of the vessel lumen (1/11); grade 2 when the greatest portion of the thrombus was between 1/2 and 2/3 of the vessel lumen (7/11); grade 3 when the thrombus exceded 2/3 of the lumen (3/11). A substantial decrease in pheripheral glycaemia values was found in sampling arising from the affected limbs. Comparing affected limbs values with healthy limbs measurements from the same patient, the reduction was found from 17.65% to 34.41%. Accounting only the grade 3 scored patients, the percentage of reduction was found up to the 28.34%. SURGICAL STABILIZATION OF CANINE LUMBOSACRAL SPINE WITH STOP‐SCREWS AND ILIAC WINGS SCREWS {#jvim13002-sec-0022} ------------------------------------------------------------------------------------------ **[Mario Dolera]{.ul}, Luca Malfassi, Simone Pavesi, Massimo Sala, Giovanni Mazza, Silvia Marcarini, Nancy Carrara** La Cittadina Fondazione Studi e Ricerche Veterinarie, Romanengo, Italy Surgical stabilization of canine lumbosacral spine can be challenging. The aim of this research was to evaluate two surgical techniques to achieve lumbosacral stabilization in dogs either with normal or transitional vertebrae. Lumbosacral instability and degenerative stenosis were evaluated by dynamic Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). In dogs with normal vertebrae two 4.5 mm screws were bicortically inserted in S1 with the heads behind the caudal articular process of L7 to prevent the extension of the lumbosacral joint; if ventral listhesis of S1 was evident, the head of the screws were augmented by methyl methacrylate. In dogs with transitional vertebrae, two 4.5 mm screws were inserted in the iliac wings, two 3.5 mm screws were inserted in the spinous process of L6 and L7; the emerging screws were embedded in methyl methacrylate after flexion of the lombosacral spine. In cases of residual radicular compression, dorsal laminectomy and partial discectomy were accomplished. Serial clinical and imaging follow up examinations were performed. Twenty‐two large breed dogs were enrolled. In 14 dogs stop‐screws (in 4 augmented) and in 8 dogs iliac wings screws were inserted. 2 dogs required additional decompression. During a mean follow up of 36 months, clinical examination and imaging reveals amelioration of presenting complaints and reduction of radicular compression, with no surgical complications. Stop‐screws and iliac wings technique are effective methods to obtain stabilization and indirect decompression of the lumbosacral joint. Comparing with other described surgical procedures, our obtained results are better but with lesser complications. FRAMELESS STETEOTAXY FOR VERTEBRAL IMPLANT GUIDANCE IN DOGS {#jvim13002-sec-0023} ----------------------------------------------------------- **[Fred Wininger]{.ul}^1^, Nicholas Archambault^2^, Stephen Frey^3^** ^1^Veterinary Specialty Services, St. Louis, Missouri, USA, ^2^University of Missouri, Columbia, Missouri, USA, ^3^Rogue Research, Montreal, Quebec, Canada Multiple pathologies of the dog spine require stabilization with vertebral implants. Optimal implant corridors are described and vary between specific vertebrae throughout the spine. Current placement guidelines use general surface landmarks and proposed angles without opportunity to visualize vital structures at depth. Preferred corridors are often inaccessible because of soft tissue impediment. Potential complications of vertebral implant placement include neuro‐vascular compromise or insufficient bone purchase for required biomechanical loads. Trans‐articular implants present the added challenge of aligning a vertebral motion units and countering biomechanical stresses. Recent reports suggest the frequency of spinal canal violation by vertebral implants is high. Frameless stereotaxy for brain neuronavigation has been described in veterinary literature and validation studies show a high level of accuracy. These techniques utilize high resolution cross sectional imaging to create 3D data sets. Identification of fiducial markers visible on 3D models (image space) and on the subject intraoperatively (anatomic space) permits fusion for co‐localizing guidance. A position sensor is used to track, in real‐time, reflective instruments affixed to the skull and surgical tools within a virtual space. The surgeon can identify any point within that virtual space for tissue sampling or treatment. The purpose of this study was to develop a technique for implant placements in the dog spine using image‐guided intervention. Unlike the brain, which is encased within the solid calvarium and once registered with the navigation software, all structures can be reliably identified relative to the origin of the coordinate system; There are two major obstacles when considering neuronavigation of the vertebral column. First, multiple vertebral motion units require individual registration. The vertebral column is made of multiple articulating bones, making a single bone registration inapplicable for the remaining vertebrae. A novel subject tracker was created for attachment to single vertebra for individual registration. The second obstacle in adapting frameless neuronavigation to the spine is the assignment of fiducial markers. Current brain protocols require pre‐imaging procedures for placement of artificial fiducial makers rigidly affixed to the skull or teeth. Pre‐imaging surgery and affixing fiducial arrays is logistically difficult in spinal applications. For this reason, identification of natural boney fiducial markers identifiable in both image and anatomic space is preferable. Evaluated landmarks included spinous processes, accessory processes and cranial costovertebral joints. Initially a cadaveric study validating the technique in the thoracolumbar spine was performed. Qualitative measures of canal compromise and bone purchase in addition to quantitative measures of trajectory and depth were evaluated. Safety and efficacy findings of zero canal violations, statistically accurate placement, and mostly cross‐body implants warranted clinical use. Completed clinical cases included three thoracolumbar fractures with vertebral body screws, two atlantoaxial stabilizations and three lumbosacral stabilizations with transarticular screws. Stereotactic guidance enabled the surgeon to visualize cross‐sectional implantation in real time. All cases had post‐operative imaging with adequate placement of implants and satisfactory clinical outcomes. The described adaptation provides a means of safe and effective placement of vertebral implants. Multiple trajectories could be created with the computerized guidance. Subjectively less soft tissue dissection was necessary and surgical time was minimally extended. PLA BEAD TREATMENT OF REFRACTORY EPILEPSY IN DOGS {#jvim13002-sec-0024} ------------------------------------------------- **[Roger Clemmons]{.ul}** University of Florida, Gainesville, Florida, USA Gold‐bead implants have been used to help treat epilepsy, but they interfere with modern imaging techniques like MRI examinations. Polylactic acid (PLA) is a biodegradable, implantable plastic that appears to offer similar effects to gold material, but degrades over time and is invisible to modern imaging. The purpose of this study was to investigate whether PLA beads (discs 1 mm by 0.5 mm) could be useful in assisting in the control of refractory epilepsy in dogs when placed in specific acupuncture points thought to affect seizures. This is a descriptive study of a case series which was done as a non‐blinded, preliminary study. Ten patients presenting to the University of Florida Neurology Service who were experiencing frequent seizures that were refractory to more than 2 anticonvulsant medications were enrolled in the study. All patients had systemic evaluations to demonstrate no significant findings, MRI evaluations to demonstrate no significant changes in the central nervous system, and CSF analysis revealing no abnormalities. Many patients experienced cluster seizures requiring emergency interventions to control the seizures. No further adjustments to medications were made during the study, although some anticonvulsants could be reduced once the PLA bead treatment was given. An EEG was performed in dexmedetomidine‐sedated dogs (10--15 μg/kg) before and after PLA bead placement and analyzed by a statistical package (NeuroStat package of NeuroGuide by Applied Neuroscience, Inc., Seminole, FL). PLA beads were placed in transpositional and classical acupuncture points by a certified veterinary acupuncturist (Chi Institute, Reddick, FL). The injection sites were wiped with ethyl alcohol gel (Purell Advanced Hand Sanitizer, Gojo Industries, Akron, OH) and the PLA beads inserted using a modified 16 ga needle and syringe. Points used were GV20, Long hui, Nao shu, An shen, Tian ping, HT/PC7, LI4, LIV3, Bai hui, CV15 and the auricular point, Shen men. Owners were asked to keep calendars of seizure events for comparison of pre‐ and post‐treatment seizure frequencies. Data was compared by pair‐t test and ANOVA using a statistical program (SPSS ver 14.0) Of the ten patients enrolled in the study, 9 were deemed to have satisfactory results by the owners. These patients showed a reduction of seizures by \>50%, moreover, the seizure characteristics changed so that seizures were more likely to be singular rather than multiple. (*P* \< 0.5) The patient who did not respond was euthanized by the owner because of seizures that occurred 2 weeks after bead placement. We later recognized that this is a period of time when patients are most likely to seizure and it was speculated that this was a consequence of the healing from the implantation procedure before the bead took over the protection. One patient who had severe clusters every 13 days prior to bead implantation remained seizure free for 4 years (using re‐implantation of beads every 6 months). From preliminary investigations, we found that the bead implantation lasted about 8 months on average, so patients were re‐implanted every 6 months, based upon owner\'s desires. EEG examinations of patients prior to implantation showed slow‐wave, high‐amplitude activity with occasional epileptiform discharges and spike activity. The most frequent wave form was an enhanced alpha wave pattern. Following bead implantation there was a significant reduction (48 ± 13%) in amplitude of EEG activity. (*P* \< 0.05) PLA bead placement in specific points around the head and in other acupuncture points which have been found to reduce seizures appears to significantly reduce seizure activity in refractory epileptic patients. There is also evidence of an acute reduction in the electrical activity of the cortex based upon changes in the EEG following bead implantation. This may represent a promising new method to help control refractory seizures in dogs. THE SPECTRUM OF INHERITED MYOPATHIES IN YOUNG LABRADOR RETRIEVERS {#jvim13002-sec-0025} ----------------------------------------------------------------- **[G. Diane Shelton]{.ul}** University of California San Diego, La Jolla, California, USA Over the past 10 years, several inherited myopathies have been reported in young Labrador Retrievers. While some have distinct clinical phenotypes and can be easily recognized, others require in depth examinations including state of the art evaluation of muscle biopsies and DNA testing. This report describes the current status of inherited myopathies in the Labrador Retriever breed with a guide to clinical recognition. An early onset of severe and generalized muscle weakness occurring in the first several weeks of life has been described in the congenital myasthenic syndrome associated with an autosomal recessive mutation in *COLQ* (2014), which encodes the collagenous tail of acetylcholinesterase. Weakness in this disorder is fatigable and anticholinesterase drugs exacerbate the weakness. Severe generalized and progressive weakness has also been described in the X‐linked myotubular myopathy associated with a mutation in the lipid phosphatase *MTM1* (2010). In both of these disorders, creatine kinase (CK) activity is normal or only minimally elevated. Another potentially severe myopathy occurring in this age group is X‐linked dystrophin deficient muscular dystrophy (XLMD). A key finding is markedly and persistently elevated CK activity. This finding, along with a dystrophic phenotype in muscle biopsies, helps to distinguish this severe myopathy from other congenital myopathies. Other less severe myopathies can occur at several months of age including autosomal recessive centronuclear myopathy associated with a mutation in *PTPLA* (2005). The clinical presentation can appear similar to that of x‐linked myotubular myopathy, however, is not as rapidly progressive. Affected dogs can live well into adulthood. The CK activity is normal or only mildly elevated and a muscle biopsy shows a myopathic phenotype with numerous central nuclei. A very mild form of XLMD has recently been identified in several related male Labrador Retrievers with markedly elevated CK activity detected at the time of neuter. No clinical evidence of weakness or muscle atrophy is observed. Muscle biopsies show a dystrophic phenotype and dystrophin is not detectable by western blot. What is protecting these dogs from the devastating effects of dystrophin deficiency is yet to be determined. A myopathy with progressive gait abnormalities and multiple joint deformities was recently identified associated with sarcolemmal specific collagen VI deficiency (2014), similar to that of human Ulrich\'s congenital muscular dystrophy. The mutation has been confirmed in *COL6A3* but not yet published. Although not a myopathy, the syndrome of exercise‐induced collapse has been characterized and a mutation in the CNS specific *DNM1* gene described (2008). Dogs with this syndrome are clinically normal in appearance and activity, which helps to separate this disorder from all other characterized inherited myopathies. In summary, many inherited myopathies have been described in young Labrador Retriever dogs and clinicians need to be aware of the specific phenotypes. Knowledge of these myopathic disorders and recognition of the myopathic phenotype should enable a more precise clinical diagnosis and guide breeders to eliminate these disorders from their lines. A P.P248LFS3\* MUTATION IN *RAB3GAP1* IN LARYNGEAL PARALYSIS AND POLYNEUROPATHY WITH OCULAR ABNORMALITIES AND SPONGIFORM ENCEPHALOPATHY IN BLACK RUSSIAN TERRIERS {#jvim13002-sec-0026} ----------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Dennis O\'brien]{.ul}^1^, Tendai Mhlanga‐Mutangadura^1^, Gayle Johnson^1^, Diane Shelton^2^, Jeremy Shomper^1^, Nicolas Granger^3^, Jeremy Taylor^1^, Robert Schnabel^1^, Gary Johnson^1^** ^1^University of Missouri, Columbia, Missouri, USA, ^2^University of California San Diego, San Diego, California, USA, ^3^University of Bristol, Langford, North Somerset, UK Juvenile onset laryngeal paralysis/polyneuropathy has been reported in Black Russian Terriers (BRT). The purpose of this study was to further characterize the phenotype and identify the mutation responsible. Three BRT with laryngeal paralysis and weakness examined at the University of Missouri (MU) underwent electrodiagnostic testing. Medical records of 13 other BRT were reviewed and histopathology examined on 9 dogs. Whole genome sequencing (WGS) was performed on one affected BRT and rare homozygous variants ascertained by comparison to 111 other canid WGS. All affected BRT presented at 3 months of age with laryngeal paresis/paralysis. The 3 MU cases had miotic pupils, cataracts, microphthalmia, and/or persistent pupillary membranes. Neurologic examination showed sensory and motor deficits. Electromyograms showed prolonged insertional activity, fibrillations and complex repetitive discharges. Tibial nerve studies showed moderately reduce conduction velocity and compound motor unit action potential amplitudes. Muscle and nerve pathology was most prominent in the larynx, and consisted of a neurogenic pattern of muscle fiber atrophy, decreased density of large myelinated fibers, and inappropriately thin myelin sheaths. Intraneuronal vacuolation was found in the cerebellar roof nuclei and Purkinje cells. WGS identified a unique homozygous single base pair deletion: *RAB3GAP1:c.743del*. The produced frame shift predicts a premature stop codon and a truncated gene product missing 730 C‐terminal amino acids. DNA was available from 8 other affected BRT which were all homozygous for the variant, while 48 normal BRT were either heterozygous or homozygous for the wild‐type allele. Mutations in *RAB3GAP1* cause a similar phenotype in humans, Warburg micro syndrome. PAROXYSMAL NON‐KINESOGENIC DYSKINESIA IN SOFT COATED WHEATEN TERRIERS IS ASSOCIATED WITH A MISSENSE MUTATION IN *PIGN* AND RESPONDS TO ACETAZOLAMIDE THERAPY {#jvim13002-sec-0027} ------------------------------------------------------------------------------------------------------------------------------------------------------------ **[Dennis O\'Brien]{.ul}^1^, Ana Kolicheski^1^, Rebecca Packer^2^, Stephanie Thomovsky^3^, Jeremy Taylor^1^, Robert Schnabel^1^, Jason Berg^4^, Laura Vasquez^5^, Gary Johnson^1^** ^1^University of Missouri, Columbia, Missouri, USA, ^2^Colorado State University, Fort Collins, Colorado, USA, ^3^Purdue University, West Lafayette, Indiana, USA, ^4^Animal Specialty Center, Yonkers, New York, USA, ^5^Gulf Coast Veterinary Neurology & Seurosurgery, Houston, Texas, USA The purpose of this study was to characterize a hereditary, paroxysmal, non‐kinesogenic dyskinesia in Soft Coated Wheaten Terriers (SCWT) and identify the responsible mutation. Medical records and videos were reviewed for 20 SCWT or SCWT/poodle crosses that presented for a movement disorder. Whole genome sequences (WGS) were performed on 2 affected SCWT and rare homozygous variants ascertained by comparison to 111 other canid WGS. Affected dogs had episodes of dystonia and/or rapid, irregular involuntary movements of the limbs that typically began between 8 months to 3 years of age. The episodes lasted a few minutes up to several hours and became more frequent and severe over time. They occurred when the dogs were awake, but most dogs showed no clear trigger. Benzodiazepines, antiepileptic drugs and muscle relaxants showed limited benefit. Five of 8 dogs treated with acetazolamide improved and 3 had complete resolution of the dyskinesia. Histopathology of two dogs showed no pathology. WGS identified a homozygous *PIGN:c.398C\>T* variant in both affected SCWT. All 12 other affected SCWT and crosses were homozygous for the variant, while 762 normal SCWT were either heterozygous or homozygous for the wild‐type allele. Three‐hundred‐eighty‐eight randomly selected poodles and 119 dogs of other breeds were all homozygous for the wild‐type allele. *PIGN* codes for phosphatidylinositol glycans‐N, an enzyme involved in synthesis of glycosyl‐phosphatidylinositol (GPI). GPI anchors numerous proteins to lipid rafts in the cell membrane including carbonic anhydrase 4. Mutations in *PIGN* in humans cause a syndrome characterized by congenital abnormalities, neonatal hypotonia, seizures and choreoathetosis. EVALUATION OF SECONDARY NEURONAL INJURY AFTER INTRACEREBRAL HEMORRHAGE IN DOGS BY PERFUSION‐WEIGHTED MRI, PROTON MR SPECTROSCOPY AND HISTOPATHOLOGY {#jvim13002-sec-0028} --------------------------------------------------------------------------------------------------------------------------------------------------- **Sun‐Hyung Cho^1^, Dae‐Gi An^1^, Junyoung Park^2^, Chulhyun Lee^2^, Dong‐In Jung^3^, Dongwoo Chang^1^, Ji‐Houn Kang^1^, Mhan‐Pyo Yang^1^, [Byeong‐Teck Kang]{.ul}^1^** ^1^Chungbuk National University, Cheongju, Korea, ^2^Korea Basic Science Institute, Ochang, Korea, ^3^Gyeongsang National University, Jinju, Korea Intracerebral hemorrhage (ICH) is a subtype of stroke with high mortality and disability. The effects of ICH upon brain tissue are biphasic, including primary and secondary injury. Although hematoma volume is the main factor that influences ICH outcome, secondary brain injury itself results in severe neurological deficits. Therefore, the purpose of this study was to investigate the mechanisms of secondary neuronal injury by evaluating cerebral blood flow (CBF), metabolites and pathological changes in and around the intracerebral hematoma. ICH was induced in 5 healthy laboratory beagle dogs by injecting 500 U of bacterial collagenase from *Clostridium histolyticum*, which was delivered into the parietal lobe over 5 minutes with a micro infusion pump. Perfusion‐weighted magnetic resonance imaging (PW‐MRI) and proton MR spectroscopy (MRS) were performed serially at 6 different time points by using a 3T MR system: before and at 12 hours, 2.5, 4.5, 9.5 and 23.5 days after ICH. Prussian blue and immunohistochemical staining were performed to demonstrate iron, glial fibrillary acidic protein (GFAP), tumor necrosis factor‐α (TNF‐α), interleukin‐6 (IL‐6) and IL‐10 in and around the intracerebral hematoma. In hematoma, cerebral blood volume (CBV) was significantly decreased at 2.5 and 4.5 days after ICH. Additionally, significant decrease of CBF was observed at 4.5 days after ICH. In perihematoma, there were no significant changes of CBV and CBF except for the decrease of CBV at 4.5 days after ICH. Proton MRS revealed no abnormality of the levels of *N*‐acetyl‐asparate and lactate in perihematoma. A significant loss of GFAP immunoreactivity was noted in hematoma, whereas iron deposition and expressions of TNF‐α, IL‐6 and IL‐10 were significantly increased. However no pathological and immunohistochemical changes were observed in perihematoma. In conclusion, decreased CBF, iron overload and inflammation may contribute to ICH‐induced injury and could provide therapeutic targets in ICH of dogs. SERUM MELATONIN LEVELS IN NORMAL DOGS AND DOGS WITH SEIZURES {#jvim13002-sec-0029} ------------------------------------------------------------ **[Stephanie Thomovsky]{.ul}^1^, Annie Chen^2^, David Deavila^2^, Alecia Kiszonas^2^** ^1^Purdue University, West Lafayette, Indiana, USA, ^2^Washington State University, Pullman, Washington, USA Epilepsy or recurrent seizures is reported to be the most common neurologic condition in dogs. 20--30% of epileptic dogs are considered to be pharmacoresistent to one medication. Recent lab animal literature, in addition to studies in human literature, show the hormone melatonin, to have significant anti‐convulsant effects. Epileptic humans actually have significantly lower inter‐ictal serum melatonin levels compared to unaffected people. The central hypothesis of our study is that serum levels of melatonin will be lower in dogs with seizures than in normal dogs. Our aim will be to compare mean serum melatonin levels in normal dogs versus dogs with seizures. Sixty two dogs were enrolled in the study: 29 normal dogs and 33 dogs with seizures. Sampling was done during the daylight hours at three separate time points (0800 hours, 1200 hours and 1600 hours). Melatonin levels were measured in pg/mL using a radioimmunosassay. Mean melatonin values at 0800, 1200 and 1600 hours in normal dogs were 0.65, 0.55 and 0.57 pg/mL, respectively. Values for dogs with seizures were 0.64, 0.56 and 0.54 pg/mL at the same time points. No statistical significance existed between mean serum melatonin values in either group at any of the three time points. In conclusion, there were no notable differences in daytime serum melatonin values in normal dogs versus dogs with seizures. ONCOLOGY {#jvim13002-sec-0030} ======== ANTINEOPLASTIC EFFECTS OF AURANOFIN IN CANINE LYMPHOMA CELLS {#jvim13002-sec-0031} ------------------------------------------------------------ **[Douglas Thamm]{.ul}, Barbara Rose, Travis Laver, Alex Pyuen** Colorado State University, Fort Collins, Colorado, USA Lymphoma is a serious condition for which there remain unmet medical needs in humans and dogs. The gold complex auranofin has been utilized as a human therapeutic, primarily as an antirheumatic agent, and some limited information about its use in dogs has been reported. Antiproliferative and pro‐apoptotic activity has been observed in a variety of human tumor‐derived cell lines, including carcinomas of the breast, head and neck, ovary, lung, and a variety of hematopoietic tumors, including lymphoma. Putative antitumor mechanisms include inhibition of NF‐kB and STAT3 signalling, and induction of reactive oxygen species via thioredoxin reductase (TrxR1) inhibition. The dog is a well‐established model for spontaneous lymphoma in humans, owing to striking similarity in biology and gene expression. The goal of the current study was to investigate the antineoplastic effects of auranofin in a panel of canine lymphoma‐derived cell lines, and to identify potential biomarkers of drug activity for future translational studies. Four canine lymphoma ‐derived cell lines were incubated in varying concentrations of auranofin +/‐ bortezomib, elesclomol, or the conventional antineoplastic agents doxorubicin (DOX), CCNU and vincristine (VCR) for 24--72 hours. Relative viable cell number was assessed using MTS. Cell cycle distribution and apoptosis induction were evaluated using flow cytometry with propidium iodide staining and Annexin V / PI staining respectively. Induction of ROS was assessed using the redox‐sensitive fluorescent dye CM‐H2DCFDA. TrxR activity was assessed using a commercial colorimetric kit (Cayman). Changes in STAT3 and NF‐kB phosphorylation status were assessed via western analysis. Auranofin induced dose‐dependent antiproliferative effects in all canine lymphoma cell lines, with 50% inhibitory concentrations (IC50s) between 0.1 and 1 µM. These are similar to IC50s observed in human hematopoietic tumor cells, and well within clinically achievable serum concentrations. Dose‐ and time‐dependent apoptosis induction was observed, and additive to synergistic antiproliferative/propaptotic activity was observed with bortezomib and elesclomol, but not DOX, CCNU or VCR. Dose‐ and time‐dependent induction of ROS and inhibition of TRxR activity was observed following auranofin treatment, and antiproliferative activity could be partially blocked with the free‐radical scavenger N‐acetylcysteine. There was no change in STAT3 of p65 phosphorylation following auranofin treatment. In conclusion, antitumor activity is observed with auranofin treatment in vitro in canine lymphoma cells, at pharmacologically achievable concentrations. This is similar to activity observed in human lymphoma and leukemia. Cooperative activity was observed with the targeted agents bortezomib and elesclomol. These data support clinical evaluation of auranofin in canine lymphoma. Measurement of ROS accumulation and/or TRxR activity may be useful pharmacodynamic markers of drug activity. RABACFOSADINE AND PREDNISONE: EFFICACY OF A Q21 DAY ADMINISTRATION SCHEDULE IN CANINE LYMPHOMA {#jvim13002-sec-0032} ---------------------------------------------------------------------------------------------- **[Douglas Thamm]{.ul}^1^, Michelle Morges^1^, Craig Clifford^2^, Kristine Burgess^3^, Corey Saba^4^, David Vail^5^, Cheryl London^6^** ^1^Colorado State University, Fort Collins, Colorado, USA, ^2^Hope Veterinary Specialists, Malvern, Pennsylvania, USA, ^3^Tufts University, North Grafton, Massachusetts, USA, ^4^University of Georgia, Athens, Georgia, USA, ^5^University of Wisconsin ‐ Madison, Madison, Wisconsin, USA, ^6^The Ohio State University, Columbus, Ohio, USA Lymphoma is one of the most common canine cancers. While current therapies induce remission in most naïve dogs with lymphoma, drug‐resistant relapse is common and there is a distinct need for novel agents. The acyclic nucleotide phosphonate 9‐(2‐phosphonylmethoxyethyl)‐guanine (PMEG) forms an active phosphorylated metabolite, PMEGpp, in cells and causes cytotoxicity in dividing cells due to inhibition of DNA polymerases α, δ, and ε; however, PMEG\'s use as an anticancer agent is limited by poor cellular permeability and nonspecific toxicity. Rabacfosadine (VDC‐1101/GS‐9219/TANOVEA^™^), a novel double prodrug of PMEG, was designed to preferentially target lymphoid cells with significantly reduced systemic toxicity. Rabacfosadine has been administered on a variety of dosing schedules to dogs with lymphoma. Objective responses were noted in 100% of chemotherapy‐naïve dogs and 60% of refractory dogs, with a median remission duration of 128 days. Given the ease of administration and equivalent activity with a q21‐day administration schedule, we sought to generate additional data regarding efficacy of this regimen through completion of a prospective clinical trial. Additionally, we explored the potential of concomitant low‐dose prednisone to mitigate previously observed cutaneous and pulmonary adverse effects. Dogs with cytologically or histologically confirmed lymphoma were treated with rabacfosadine (0.82 mg/kg free base, as a 30‐minutes IV infusion once every 21 days). Dogs received concomitant prednisone (1 mg/kg PO QOD) throughout treatment. Dogs experiencing a complete response (CR) received 5 doses of rabacfosadine, followed by monthly rechecks. Complete clinicopathological assessment and clinical assessment of remission and adverse effects (AEs) were performed every 21 days. Response was assessed according to published VCOG criteria and AEs according to the VCOG‐CTCAE v1.1. 74 dogs were prospectively enrolled. 63 were evaluable for response assessment and 73 were evaluable for progression free interval (PFI) assessment. While 13% of evaluable dogs were treatment‐naïve and 29% had received a single line of previous treatment, the majority of dogs (59%) had received 2 or more lines of previous therapy. 50 evaluable dogs had B cell lymphoma and 13 had T cell lymphoma. The overall response rate (ORR) was 57% (25% CR, 32% PR). The ORR was 64% and 23% for B cell and T cell respectively. Degree of pre‐treatment impacted response rate: the ORR was 88% in naïve dogs, 56% in dogs treated 2nd‐line, and 51% in 3rd‐line and beyond. The median PFI was 112 days for dogs experiencing a CR and 42 days for a PR (overall median PFI 41 days). Degree of pre‐treatment significantly impacted PFI (164, 84 and 32 day for naïve, 2nd‐line and \>2nd line respectively). The majority of AEs were mild and self‐limiting: gastrointestinal (GI) and hematologic AEs were most common. Grade 3 AEs included liver enzyme elevation (4), lethargy (4), GI (2) and urinary (1). 3 dogs experienced grade 4 hematologic toxicity, and 2 developed severe hemorrhagic gastroenteritis leading to euthanasia, several weeks after the first rabacfosadine treatment and therefore of uncertain attribution. 4 dogs experienced grade 1 dermatologic AEs, and 1 dog developed grade 1 pulmonary fibrosis. In conclusion, rabacfosadine was generally well tolerated and had substantial antitumor activity in dogs with both treatment‐naïve and refractory lymphoma when administered on a q21‐day schedule. Response rates and PFI observed in this study are comparable to historical data with rabacfosadine when degree of pre‐treatment is accounted for. There was a reduction in both the frequency and severity of AEs relative to historical data; however, it is not clear whether this is a result of less frequent dosing, concurrent prednisone, or lower cumulative rabacfosadine exposures in this heavily pre‐treated population. Further studies are warranted to explore rabacfosadine at higher doses. DEFINITIVE HIGH‐DOSE HYPO‐FRACTIONATED TOTAL PELVIC IRRADIATION WITH SIMULTANEOUS BOOST IN CANINE URINARY TRANSITIONAL CELL CARCINOMA: A FEASIBILITY STUDY AND FIRST CLINICAL EXPERIENCES {#jvim13002-sec-0033} ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Mario Dolera]{.ul}, Luca Malfassi, Massimo Sala, Giovanni Mazza, Simone Pavesi, Silvia Marcarini, Nancy Carrara, Sara Finesso** La Cittadina Fondazione Studi e Ricerche Veterinarie, Romanengo, Italy The lower urinary tract transitional cell carcinoma (CCT) poses challenge in order to the appropriate radiotherapy (RT) regimen. Organs at risk (OARs) within the irradiation field (ureters, rectum) are the limiting factors in dose escalation. The primary aim of this study was to evaluate the technical feasibility of high‐dose hypo‐fractionated dynamic IMRT in non‐resectable lower urinary CCT affected dogs. The secondary goal was to evaluate the toxicity and the efficacy of the RT regimens. Three dogs with lower urinary tract CCT were treated with definitive high‐dose hypo‐fractionated RT with volumetric modulates arc therapy (VMAT) technique. The volume treatment definition include the gross tumor (GTV), the PTV1 (GTV+3 mm), limphatics (PTV2), the entire bladder, prostate in males and uretra (PTV3), the entire pelvis except the rectal volume (PTV4). Dose prescriptions were 40 Gy to PTV1, 38 Gy to PTV2, 34 Gy to the PTV3, 30 Gy to the PTV4, in 6 fractions on alternate days. A piroxicam was subministered to all dogs. Serial clinical and CT/MRI examination were performed. Disease control and toxicity effects were evaluated according to RECIST and VRTOG criteria. Three T2N0M0 urinary tract CCT were treated. Prescription goals were obtained in all three cases with V95% \>95% and V107% \>2%. During follow‐up (mean 6 months) one partial response and two complete responses were obtained. Two grade I cystitis developed. Non rectal toxicity was recognized. The initial experiences with the RT regimen adopted indicate a feasibility and effectiveness in lower urinary CCT. Longer follow‐up and larger treatment series are needed. DEFINITIVE HIGH‐DOSE HYPO‐FRACTIONATED STEREOTACTIC BRAIN‐SPARING IRRADIATION OF STAGE IV CANINE NASAL TUMORS: A FEASIBILITY STUDY AND FIRST CLINICAL EXPERIENCES {#jvim13002-sec-0034} ----------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Luca Malfassi]{.ul}, Mario Dolera, Simone Pavesi, Massimo Sala, Giovanni Mazza, Silvia Marcarini, Nancy Carrara, Sara Finesso** La Cittadina Fondazione Studi e Ricerche Veterinarie, Romanengo, Italy The prognosis for canine nasal tumours with intracranial extension is poor with an expected survival of 1 month with palliation and 6.7 months with irradiation. However, studies regarding stage IV nasal tumours treated with brain‐sparing irradiation techniques are lacking. The aim of this prospective study was to evaluate feasibility and efficacy of definitive intent stereotactic radiotherapy in dogs with nasal tumours with massive intracranial extension. Seven dogs with stage IV nasal tumours were treated with high‐dose hypo‐fractionated stereotactic radiotherapy with VMAT technique. Dose prescriptions were 32--36 Gy in four consecutive‐day fractions to the gross tumour and 30 Gy to limphatics. Adjuvant treatment included carboplatin. Serial clinical and CT/MRI examination were performed. Disease control and toxicity effects were evaluated according to RECIST and VRTOG criteria. Median survival time (MST) was evaluated using Kaplan‐Meier curves. Six carcinoma and 1 sarcoma were treated. Prescription goals were obtained in four cases with V95% \>95% and V107% \>2% whereas in 3 dogs V95% = 86--90% was accepted to limit maximum brain punctual dose \<27 Gy. Two partial response and 5 complete responses were obtained. MST was 9 months. One grade II late brain radiotoxicity and two brain ascending infections were observed. Relapse pathways involves diffuse meningeal and sphenoid invasion. The initial experiences with the RT regimen adopted indicate a feasibility and effectiveness in modified stage IV nasal tumours. The relapse pathways observed suggest to evaluate alternative adjuvant treatment in dogs treated with stereotactic radiotherapy. INVESTIGATION OF THE FOXM1 TRANSCRIPTIONAL PATHWAY IN CANINE LYMPHOMA {#jvim13002-sec-0035} --------------------------------------------------------------------- **[Jackie Wypij]{.ul}, Sara Goldschmidt, Holly Pondenis** University of Illinois at Urbana‐Champaign, Urbana, Illinois, USA Forkhead box M1 (FOXM1) is a transcription factor regulating cell cycle progression, angiogenesis, and apoptosis. Increased FOXM1 is observed in human lymphoma/leukemia and is recognized as both a prognostic factor and a therapeutic target. Thiostrepton, an antibiotic used in commercial veterinary topical products, is a putative FOXM1 inhibitor. The purpose of this study was to 1) evaluate the expression of FOXM1 pathway proteins in canine lymphoma cell lines and 2) evaluate the anti‐proliferative effects of the FOXM1 inhibitor thiostrepton on canine lymphoma in vitro. Four canine lymphoma cell lines (17‐71, GL‐1, CL‐1, and OSW) were used. Western blot was performed to assess protein expression of FOXM1 and FOXM1 pathway proteins cyclin B, survivin, and Cdc25. 17‐71 canine lymphoma cells were treated with thiostrepton (0--100 μM) for 24 hours. Cell viability (cell proliferation and cell death) were assessed with a commercial MTS assay and Trypan Blue exclusion. Canine lymphoma cell lines express FOXM1 and FOXM1 pathway proteins. Thiostrepton decreases cell viability in a dose‐dependent manner (*P* \< 0.05). In vitro results support the FOXM1 pathway as a putative target in canine lymphoma. The FOXM1 inhibitor thiostrepton has anti‐cancer activity in vitro. These results warrant further evaluation of thiostrepton in vitro for canine lymphoma. This research may lead to validation of drug repurposing of thiostrepton‐containing topical compounds for treatment of canine lymphoma, which may be particularly relevant in cutaneous lymphoma. TUBULIN‐MODULATING EFFECTS OF THE ANTHELMINTIC MEBENDAZOLE IN FELINE CANCER {#jvim13002-sec-0036} --------------------------------------------------------------------------- **[Jackie Wypij]{.ul}** University of Illinois at Urbana‐Champaign, Urbana, Illinois, USA Limited effective treatment options are available for feline cancer patients. One mechanism for new drug development is drug repositioning of low‐toxicity pharmaceuticals with inherent anti‐cancer activity. Benzimidazole anti‐parasitics such as mebendazole (MBZ) are putative novel mitotic spindle inhibitors. The purpose of this study was to 1) evaluate the potential synergism between MBZ and a traditional mitotic spindle inhibitor in vitro*,* 2) evaluate MBZ effect in vitro on tubulin polymerization and 3) evaluate serum tubulin polymerization activity in healthy cats treated with MBZ. Two feline cancer cell lines (SCCF1, oral squamous cell carcinoma and K12, feline mammary carcinoma) were incubated with MBZ at 0--100 μM and vinorelbine at various concentrations for 24 hours and cell proliferation was assessed with a commercial MTS assay. Synergism was computed utilizing a combination index. Two healthy adult cats were treated with 22 mg/kg MBZ daily. A commercial assay was used to assess direct tubulin polymerization activity of MBZ as well as the tubulin polymerization of treated cat sera. MBZ decreased cell proliferation (*P* \< 0.05) and appears synergistic with vinorelbine. MBZ decreased the rate and total polymerization of tubulin in vitro. There was measurable decreased tubulin polymerization activity in serum of the treated cats. MBZ demonstrates in vitro anti‐neoplastic effects in feline cancer cell lines. MBZ was well tolerated in vivo and resulted in measurable anti‐tubulin activity in treated cats. Given the known low toxicity, MBZ may be a potential candidate for future clinical trials in pet cats with cancer. TREATMENT OF CANINE B‐CELL LYMPHOMA WITH CHEMOTHERAPY AND A CANINE ANTI‐CD20 MONOCLONAL ANTIBODY: A PROSPECTIVE DOUBLE‐BLIND, RANDOMIZED, PLACEBO‐CONTROLLED STUDY {#jvim13002-sec-0037} ------------------------------------------------------------------------------------------------------------------------------------------------------------------ **[G.K. Ogilvie]{.ul}^1^, D.R. Proulx^1^, L. VanHorn^1^, G. Archer^1^, B. Monreal^1^, E. Moe^1^, L. Pope^1^, G. Hansen^2^, B. Alkuzweny^2^, W.M. FitzPatrick^1^** ^1^Angel Care Cancer Center, California Veterinary Specialists, Carlsbad, California, USA, ^2^Aratana Therapeutics, Inc., Kansas City, Kansas, USA Management of human B‐cell lymphoproliferative disorders, including diffuse large B‐cell lymphoma (DLBCL), includes a monoclonal antibody that targets the B‐cell antigen CD20. The aggressive form of DLBCL has many similarities to the most common form of lymphoma in dogs. Addition of the species‐specific antibody rituximab to chemotherapy for the treatment of DLBCL in humans is associated with little increase in toxicity, yet enhanced efficacy. This study evaluates a monoclonal antibody that binds specifically with the canine B‐cell antigen CD20 (canine Mab), used with chemotherapy. Dogs were enrolled in a prospective, randomized, blinded, placebo‐controlled study, treated with one 4‐week cycle of L‐CHOP chemotherapy and then randomized to either receive treatment with the canine Mab or placebo. Once remission was lost, all dogs received one dose of doxorubicin followed by treatment with the canine Mab. Efficacy was assessed by evaluation of measurable lymph nodes, cytology and clinical status. Safety was based on Veterinary Cooperative Oncology Group\'s common terminology criteria (VCOG‐CTCAE v1.1). Dogs (n = 27 enrolled, 24 randomized) all had measurable, cytologically‐confirmed and flow cytometrically immunophenotyped DLBCL. The median progression‐free and overall survival times in the canine Mab arm of the study were 167 and 325 days, respectively, compared to 93.5 and 177 days for the placebo arm. Adverse events were restricted to the L‐CHOP cycle. The canine Mab was well‐tolerated and, in conjunction with L‐CHOP, caused significant increases in survival in dogs with DLBCL. PRECLINICAL COMPARISON OF THREE INDENOISOQUINOLINES CANDIDATES IN TUMOR‐BEARING DOGS {#jvim13002-sec-0038} ------------------------------------------------------------------------------------ **[Amy LeBlanc]{.ul}^1^, Jenna Burton^2^, Chand Khanna^1^, Christina Mazcko^1^, Susan Lana^3^, Kristen Weishar^3^, Cheryl London^4^, William Kissberth^4^, Erika Krick^5^, David Vail^6^, Michael Childress^7^, Jeffrey Bryan^8^, Kimberly Selting^8^, Carolyn Henry^8^, Jennifer Wilcox^8^, Sandra Axiak‐Betchel^8^, Lisa Barber^9^, Jiuping Ji^10^, Joesph Covey^10^, Thomas Pfister^11^** ^1^Comparative Oncology Program, Center for Cancer Research, National Cancer Institute, Bethesda, Maryland, USA, ^2^UC Davis School of Veterinary Medicine, Davis, California, USA, ^3^Colorado State University, College of Veterinary Medicine, Fort Collins, Colorado, USA, ^4^Veterinary Biosciences, The Ohio State University, Columbus, Ohio, USA, ^5^School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA, ^6^School of Veterinary Medicine, University of Wisconsin, Madison, Wisconsin, USA, ^7^Purdue University College of Veterinary Medicine, West Lafayette, Indiana, USA, ^8^Department of Veterinary Medicine and Surgery, College of Veterinary Medicine, University of Missouri‐Columbia, Columbia, Missouri, USA, ^9^Department of Clinical Sciences, Cummings School of Veterinary Medicine at Tufts University, North Grafton, Massachusetts, USA, ^10^Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, Maryland, USA, ^11^Laboratory of Human Toxicology and Pharmacology, Applied/Developmental Research Support Directorate, SAIC‐Frederick, Inc., Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA Indenoisoquinolones act as topoisomerase 1 inhibitors and are known to be active anticancer agents in both humans and dogs. A clinical trial in tumor‐bearing dogs with lymphoma was conducted through the NCI Comparative Oncology Trials Consortium to define the safety, pharmacokinetics (PK) and pharmacodynamic modulation of three related novel indenoisoquinolones. Sixty‐eight dogs with lymphoma were enrolled in the dose escalation phase. Dogs were randomized to receive 5 consecutive intravenous daily doses of one agent. Each agent was escalated independently with 3 dogs/cohort. Serial pre‐ and post‐treatment tumor biopsies, tumor aspirates, and bone marrow aspirates were collected. Two 24‐hours PK curves were completed after the first and last dose of drug. Tumor and plasma PK, TOP1 levels, and gH2AX induction data were collected along with clinical toxicity and response data. The maximum tolerated dose (MTD) was defined for two of three indenoisoquinolines (NSC725776 17.5 mg/m^2^, NSC706744 100 mg/m^2^). An MTD was not reached for NSC743400. Toxicity profiles for all agents included expected gastrointestinal, hematologic and constitutional events. Objective responses were seen across all agents with a dose‐response relationship evident for NSC725776 and NSC734400. The overall response rates for evaluable dogs were 30% (7/23), 39% (3/23), and 78% (14/18) for NSC725776, NSC743400, and NSC706744, respectively. Correlative relationships between clinical response, pharmacokinetic profiles and pharmacodynamic markers were established for all 3 agents. All 3 novel agents studied herein are both tolerable and effective in tumor‐bearing dogs. Study of NSC706744 at 100 mg/m^2^ in dogs with lymphoma and measurable soft tissue sarcoma is ongoing. COMBINATION AT‐014 (ADXS31‐164) *LM*‐LLO IMMUNOTHERAPY PLUS PALLIATIVE RADIATION DELAYS TUMOR PROGRESSION AND PROLONGS OVERALL SURVIVAL IN CANINE OSTEOSARCOMA {#jvim13002-sec-0039} -------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Nicola Mason]{.ul}^1^, Josephine Gnanandarajah^1^, Ana Caceres^1^, Lilian Duda^1^, Kim Agnello^1^, Cara Blake^1^, Falon Grey^1^, Julie Engiles^1^, Anu Wallecha^2^, Yvonne Paterson^1^** ^1^University of Pennsylvania, Philadelphia, Pennsylvania, USA, ^2^Advaxis Inc, Princeton, New Jersey, USA Radiation therapy (RT) induces immunogenic tumor cell death and promotes anti‐tumor immunity that can result in regression of metastatic lesions, a phenomenon known as the abscopal effect. We hypothesized that combining RT with a recombinant HER2/neu expressing *Lm‐*LLO immunotherapy (AT‐014, ADXS31‐164) could promote anti‐tumor immunity, delay primary tumor progression and prevent metastatic disease in dogs with osteosarcoma. Ten dogs with histopathologically confirmed, treatment naïve, appendicular osteosarcoma were enrolled to this pilot study. All dogs received 16 Gy of RT in 2 fractions on consecutive days, followed by AT‐014 intravenously every 3 weeks for 8 administrations. Radiographs were performed at weeks 0, 10, 22 and every 2 months thereafter to assess primary tumor progression and development of pulmonary metastases. At these times dogs were evaluated for toxicity, lameness and quality of life (QOL) and PBMCs were collected to evaluate HER2/neu specific T cell responses. The primary endpoint was time to progression (TTP); secondary end points were safety and overall survival. Repeat AT‐014 administrations were well tolerated. Lameness and QOL improved in 7 dogs. At present, 6 dogs are still alive. After 8 doses of AT‐014, 5 dogs showed minimal radiographic evidence of primary tumor progression. 4 dogs developed pulmonary metastatic disease and 4 suffered pathological fractures. Median TTP is 204 days and median survival time is 285 days. Immunological assessment is pending. Results compare favorably to historical reports of RT alone and warrant performance of a randomized, placebo controlled clinical trial. CHARACTERIZATION OF THE RELATIONSHIP BETWEEN BODY CONDITION SCORE AND SURVIVAL IN DOGS WITH LYMPHOMA AND OSTEOSARCOMA {#jvim13002-sec-0040} --------------------------------------------------------------------------------------------------------------------- **Frank Romano^1^, [Cailin Heinze]{.ul}^1^, Lisa Barber^1^, Joel Mason^2^, Lisa Freeman^1^** ^1^Cummings School of Veterinary Medicine, North Grafton, Massachusetts, USA, ^2^USDA Human Nutrition Research Center, Tufts University, Boston, Massachusetts, USA In humans, obesity exacerbates many types of cancer by increasing the risk of cancer development, tumor aggressiveness, recurrence, and the risk of cancer death. However, this relationship in dogs has not been thoroughly evaluated. The aim of this study was to determine whether body condition score (BCS) at the time of lymphoma (LSA) or osteosarcoma (OSA) diagnosis in dogs was predictive of survival time (ST). We hypothesized that an overweight BCS at the time of cancer diagnosis would be associated with a shorter ST. Medical records of dogs diagnosed and treated for LSA or OSA between 2000 and 2010 were reviewed. Data on signalment, body weight, BCS, treatment, and survival were collected. Dogs were grouped by BCS (underweight, normal, and overweight) and STs were compared using Kaplan‐Meier survival analysis. A total of 325 dogs (LSA = 270 and OSA = 55) were included in the study. Overall, 5.5% of dogs were underweight, 54.2% were normal weight, and 40.3% were overweight. On univariate analysis, ST was significantly shorter for underweight dogs with LSA (*P = *0.017), but not for OSA (*P *= 0.518). Dogs with LSA that gained \>10% body weight after diagnosis had a longer ST (*P *= 0.003). On multivariate analysis of all dogs, change in body weight (*P* \< 0.001) and anemia (*P *= 0.026), but not BCS group, were significantly associated with ST. In conclusion, although no relationship between BCS and ST was detected, changes in body weight after diagnosis were associated with ST. SMALL ANIMAL INTERNAL MEDICINE {#jvim13002-sec-1001} ============================== TREATMENT OF FELINE HYPERSOMATOTROPISM -- EFFICACY, MORBIDITY AND MORTALITY OF HYPOPHYSECTOMY {#jvim13002-sec-0042} --------------------------------------------------------------------------------------------- **[Patrick Kenny]{.ul}, Christopher Scudder, Sophie Keyte, James Swann, Robert Fowkes, David Church, Yaiza Forcada, Stijn Niessen** Royal Veterinary College, London, UK Hypersomatotropism (HS) is an important cause of diabetes mellitus (DM) in cats. Though excision of the somatotrophinoma is generally recommended in humans with HS, surgical therapy has so far been infrequently reported in cats. We aim to describe the efficacy of hypophysectomy as a treatment for HS and the morbidity and mortality encountered in a cohort of cats. Hypophysectomy has been offered to owners who presented diabetic cats with confirmed HS (insulin‐like growth factor‐1 \[IGF‐1\] \>1,000 ng/mL, pituitary mass) to the Royal Veterinary College since 2012. All cats were operated on by one neurosurgeon. Hypophysectomy was performed by manual extirpation using fine surgical tools via a trans‐oral trans‐sphenoidal approach. All cats were rigidly positioned in a commercially available surgical head frame, and the approach to the pituitary fossa guided by referencing bony landmarks to computed tomographic scans. Cats received intensive peri‐ and post‐operative monitoring of electrolytes, glucose and blood pressure, and were initially administered conjunctival desmopressin (DDAVP), intravenous infusions of insulin and hydrocortisone, before being transitioned to conjunctival DDAVP, oral hydrocortisone and levothyroxine and subcutaneous glargine insulin. In total, 21 diabetic cats with HS underwent hypophysectomy from April 2012 -- Oct 2014 (median, range; age: 10.3 years, 5.4 to 14.8; pituitary height: 6.0 mm, 4.0 to 10.6; IGF‐1: 1,833 ng/mL, 1,138 to \>2,000, fructosamine: 574 μmol/L, 339 to 1,076). Other than mild pelvic limb weakness, no cat displayed overt neurological deficits prior to surgery. Three (14%) cats died post‐operatively. Two cats did not recover from anaesthesia and were euthanized within 24 hours; one cat developed septic meningitis and was euthanized 17 days post‐operatively. All surviving cats (n = 18) saw a reduction of serum IGF‐1, 16 cats (89%) showed IGF‐1 normalization (Median post‐op serum IGF‐1: 38 ng/mL, 15--1,955; Wilcoxon Signed Rank Test, *P* \< 0.001). Fourteen of the 18 surviving cats (78%) achieved diabetic remission; the remaining 4 achieved superior glycaemic control with lower insulin dosages (median fructosamine pre‐ and post‐operatively: 692 and 547 μmol/L respectively; median insulin dose pre‐ and post‐operatively: 20.5 and 3.5 units/kg/day respectively). Congestive heart failure was encountered as transient problem in 4/19 cats that recovered from the surgery, all 4 cases occurred prior to implementing a reduction in volume of intravenous fluid delivered as part of the post‐operative protocol. Two cats developed paresis of the left orbicularis oculi muscle, which resolved in the surviving cat. Cardiac arrest occurred in 1 cat post‐operatively at time of jugular catheter placement, which was successfully revived and made an uneventful recovery. One cat developed a left pelvic limb monoparesis, which improved but did not resolve. Palatal wound breakdown was not encountered. Hypophysectomy as a treatment for feline HS/DM resulted in a high incidence of diabetic remission and resolution of HS. DURATION OF FASTING BUT NOT DIURNAL VARIATION AFFECTS THE RESPONSE TO GLUCAGON IN HEALTHY CATS {#jvim13002-sec-0043} ---------------------------------------------------------------------------------------------- **[Chen Gilor]{.ul}, Rebecca Glock, Shir Gilor** The Ohio State University, Columbus, Ohio, USA The role of glucagon disturbances in diabetes is increasingly recognized. Glucagon stimulation tests have been described in healthy and diabetic cats previously but information is lacking on the response of healthy cats to glucagon under specific conditions. The aim of this study was to assess the effect of diurnal variation and duration of fasting on the response to glucagon in healthy cats. Five healthy cats were used in this repeated‐measures study. Cats were free‐fed regularly at 0730 and 1730 hours for 30 minutes. Glucagon stimulation tests (20 μg/kg, IM) were performed on each cat 3 times, 2 weeks apart: At 1900 hours after a 25 hours fast (PM25), at 0900 hours after a 25 hours fast (AM25), and at 0900 hours after a 15 hours fast (AM15). Glucose and insulin concentrations were measured at −15, 0, 15, 25, 35, 45 and 60 minutes post stimulation. Baseline (mean of −15 and zero minutes) and peak concentrations were compared using the Friedman test (*P* \< 0.05 considered significant). Baseline glucose and insulin did not differ significantly between treatment groups. Peak glucose concentrations occurred at 15 minutes and were significantly higher (*P* = 0.0085) at AM15 (mean ± SD = 185.2 ± 43.0 mg/dL) but did not differ between AM25 (144.4 ± 10.5 mg/dL) or PM25 (128.0 ± 18.4 mg/dL). Similarly, peak insulin concentrations occurred at 15 minutes and were significantly higher (*P* = 0.04) at AM15 (1,911 ± 1,153 ng/L) but did not differ between AM25 (739 ± 452 ng/L) or PM25 (549 ± 366 ng/L). In conclusion, prolonged fasting (25 hours) significantly blunted the glycemic response to glucagon compared to shorter fasting (15 hours) but diurnal variation had no significant effect on glucose or insulin responses. A FELINE‐SPECIFIC ANTI‐NERVE GROWTH FACTOR ANTIBODY IS SAFE AND EFFECTIVE FOR THE ALLEVIATION OF INFLAMMATORY PAIN IN CATS {#jvim13002-sec-0044} -------------------------------------------------------------------------------------------------------------------------- **[David Gearing]{.ul}^1^, Patrick Gearing^1^, Elena Virtue^1^, Duncan Lascelles^2^, Alexander Drew^1^** ^1^Nexvet Biopharma, Melbourne, Victoria, Australia, ^2^North Carolina State University, Raleigh, North California, USA Options for the management of chronic pain in cats are extremely limited ‐ in particular, the use of NSAIDs is limited by the potential for significant side effects. A safe and effective analgesic for cats is highly desirable. Neutralizing monoclonal antibodies (mAb) against Nerve Growth Factor (NGF) are analgesic in rodent models and in humans with osteoarthritis. Using a novel technique for inter‐species conversion of antibodies based on expressed cDNA sequence analysis (PETizing) we have recently described a canine‐specific anti‐NGF mAb (NV‐01) that alleviates pain due to osteoarthritis in dogs. Using the same approach, we now describe the design and development of a novel therapeutic feline‐specific anti‐NGF mAb (NV‐02) for the alleviation of pain in cats. The PETized amino acid sequences of the NV‐02 mAb heavy and light chains were converted to cDNA sequences by complete oligonucleotide‐based chemical synthesis, cloned into a mammalian cell expression vector and expressed in Chinese Hamster Ovary (CHO) cells in chemically‐defined animal‐component free media. The NV‐02 mAb was purified to homogeneity by affinity chromatography and sterile filtration. Purified NV‐02 mAb demonstrated very high affinity and potency for neutralizing NGF and did not bind complement. In pilot studies, no adverse events were observed following injection of NV‐02 mAb i.v. or s.c. in four cats. The NV‐02 mAb had an elimination half‐life of approximately 1 week and did not induce neutralizing antibodies following repeated injection. Further studies, using the kaolin model of inflammatory pain in the cat, demonstrated that NV‐02 mAb was effective at reducing signs of lameness due to inflammation in a dose‐dependent manner (*P* \< 0.05). Together, these data suggest that NV‐02 mAb has the potential to be a safe and effective therapeutic analgesic in the cat. Studies are now in progress in cats with degenerative joint disease. SPIKE GENE MUTATIONS IN FELINE CORONAVIRUS AND THEIR CORRELATION TO FELINE INFECTIOUS PERITONITIS {#jvim13002-sec-0045} ------------------------------------------------------------------------------------------------- **[Christian Leutenegger]{.ul}^1^, Nancy Sanders^1^, Jane Robertson^1^, Peter Rottier^2^** ^1^IDEXX Laboratories, Inc., West Sacramento, California, USA, ^2^Utrecht University, Utrecht, The Netherlands The feline coronaviruses (FCoV) occur as 2 pathotypes with an enigmatic, even controversial, relationship: the low virulence or nonvirulent feline enteric coronavirus (FECV) and the highly lethal feline infectious peritonitis virus (FIPV). Recently described spike gene mutations within the putative fusion region were described to correlate with the mutated form of FCoV (FIPV) leading to the clinical presentation of feline infectious peritonitis (FIP). In this presentation, the development and validation of an allelic discrimination real‐time PCR typing test which can identify each mutation separately will be described. The diagnostic sensitivity and specificity will be reported from a set of 203 European clinical samples acquired from either FIP confirmed cats or from healthy cats that previously tested FCoV positive. Of these archived samples, 187 FCoV positive samples were included into the validation. From these, 13 samples did not pass quality control and 17 had virus levels that were below the limit of detection of the PCR assay. Of the remaining 157 samples, 156 were typed correctly with an accuracy of 99.4%. One FIP characterized sample was typed FECV (diagnostic sensitivity 98.6%) while all of the healthy cats were typed FECV (100% diagnostic specificity). To confirm that these spike gene mutations are not unique to European cats with FIP, additional validation studies from US and Japanese samples were conducted. The US clinical study included 68 cases, 46 from FIP suspicious cases and 22 with non‐FIP compatible disease. The FIPV RealPCR biotyping assay was able to accurately differentiate between the FIP or non‐FIP (FECV) etiologies (*P* \< 0.0001) and did not biotype cats with confirmed non‐FIP disease as FIPV, confirming the high diagnostic specificity of the molecular test. EVALUATION OF WEIGHT LOSS OVER TIME IN CATS WITH CHRONIC KIDNEY DISEASE {#jvim13002-sec-0046} ----------------------------------------------------------------------- **[Lisa Freeman]{.ul}^1^, Marie‐Paul Lachaud^2^, Sean Matthews^3^, Linda Rhodes^4^, Bill Zollers^4^** ^1^Tufts Cummings School of Veterinary Medicine, North Grafton, Massachusetts, USA, ^2^Aratana Therapeutics, Inc, Paris, France, ^3^Statistical Consultant, Dublin, Ireland, ^4^Aratana Therapeutics, Inc, Kansas City, Kansas, USA Thin body condition, weight loss, and muscle loss are common in cats with chronic kidney disease (CKD). However, the time course and progression of weight loss before and after diagnosis have not been thoroughly evaluated. Therefore, the purpose of this retrospective analysis was to describe the weight loss experienced by cats with CKD before and after diagnosis. Cats with CKD from 6 US veterinary practices for which International Renal Interest Group (IRIS) Stage was available were eligible. Only those with age, date of CKD diagnosis, and weight measurements available in the 3 years before and after diagnosis were included in the analysis. A total of 569 cats, with a mean age at diagnosis of 14.5 ± 2.8 years, were evaluated (55.5% spayed females and 44.5% castrated males). Cats were categorized at diagnosis as IRIS Stage 1 \[n = 34 (6%)\], Stage 2 \[n = 345 (61%)\], Stage 3 \[n = 141 (25%)\], and Stage 4 \[n = 49 (9%)\]. Median body weight at diagnosis was 4.2 kg (range, 1.6--9.9 kg). Cats lost a median of 8.9% of body weight in the 12 months before diagnosis, but weight loss began as early as 3 years before diagnosis and accelerated after diagnosis of CKD. Cats below median body weight (4.2 kg) at diagnosis had a significantly shorter survival time compared to cats ≥4.2 kg at diagnosis (*P* \< 0.0001). Weight loss can be detected in cats before diagnosis of CKD, accelerates after diagnosis, and is associated with survival so careful monitoring could benefit feline health. PROTEINURIA IN MINIATURE SCHNAUZER DOGS WITH AND WITHOUT HYPERTRIGLYCERIDEMIA {#jvim13002-sec-0047} ----------------------------------------------------------------------------- **[Eva Furrow]{.ul}^1^, Valerie Parker^2^, Jordan Jaeger^4^, Susan Murdoch^3^, John Brunzell^3^** ^1^University of Minnesota, College of Veterinary Medicine, Saint Paul, Minnesota, USA, ^2^The Ohio State University, College of Veterinary Medicine, Columbus, Ohio, USA, ^3^University of Washington, School of Medicine, Seattle, Washington, USA, ^4^Carolina Veterinary Specialists, Charlotte, North California, USA Laboratory rodents with spontaneous hyperlipidemia develop glomerular injury and proteinuria. Idiopathic hypertriglyceridemia is common in Miniature Schnauzers, and a previous report demonstrated a possible link to proteinuria in the breed. The objective of this study was to evaluate the relationship between hypertriglyceridemia and proteinuria in Miniature Schnauzers. Twenty‐seven Miniature Schnauzer dogs were enrolled in the study. Fasting serum triglycerides and urine protein‐to‐creatinine ratios (UPC) were determined. The relationship between these two variables was analyzed by the chi‐square test and with simple linear regression as well as multiple regression including age, sex, and body condition score as covariates. Nine dogs had normal serum triglyceride concentrations (median = 68 mg/dL, range = 14--81 mg/dL), and 18 dogs had hypertriglyceridemia (median = 303 mg/dL, range = 87--2,089 mg/dL). There was a strong positive correlation between triglyceride concentration and UPC (r = 0.75, *P* \< 0.001). Ten of 18 dogs (56%) with hypertriglyceridemia had proteinuria (UPC ≥0.5) compared to none of the dogs with normal triglyceride concentrations (*P* = 0.009). Proteinuria was most severe in dogs with triglyceride concentrations \>400 mg/dL; 5/6 dogs (83%) with triglyceride concentrations above this level had a UPC ≥2.0. None of the dogs were azotemic or hypoalbuminemic. Endocrine disease was not diagnosed in any dog. In conclusion, significant proteinuria is present in dogs with idiopathic hypertriglyceridemia and may be due to lipid‐induced glomerular injury. Longitudinal studies and evaluation of renal pathology are warranted to further investigate these findings. NEUTROPHIL EXTRACELLULAR TRAP FORMATION IS INCREASED IN DOGS WITH IMMUNE MEDIATED HEMOLYTIC ANEMIA {#jvim13002-sec-0048} -------------------------------------------------------------------------------------------------- **[Stephanie Smith]{.ul}, Maureen McMichael, Ron Achiel, Katrina Jung** University of Illinois, Urbana, Illinois, USA It has recently been discovered that stimulated neutrophils can release chromatin fibers decorated with enzymes to form extracellular traps (NETs). Formation of NETs (NETosis) is associated with histone hypercitrullination. NETs are part of the innate immune response, but also associated with thrombosis, organ damage, and autoimmunity. In this prospective observational study, EDTA plasma was collected from 30 apparently healthy dogs, and 18 dogs with IMHA. Plasma DNA concentration was measured by diluting samples 10--160 fold in phosphate‐buffered saline/0.1% albumin, and adding 1 mM SytoxGreen. Fluorescence was recorded (Ex485/Em538) and corrected for autofluorescence using diluted samples without SytoxGreen. DNA concentrations were calculated based on a standard curve. Nucleosomes (DNA‐histone complexes) were quantified by ELISA and normalized to canine pooled normal plasma (defined as 1). The presence of citrullinated histone H3 in plasma was confirmed using western blot. Results were compared between healthy and IMHA dogs by Wilcoxon rank sum test. Plasma from dogs with IMHA (median 3.1 µg/mL, range 0--27.1) contained significantly (*P* \< 0.001) higher concentrations of free DNA than normal plasma (median 0.7, range 0.2--1.2). IMHA plasma also contained significantly (*P* \< 0.001) higher concentrations of nucleosomes (median 6.8, range 1.4--103) as compared to normal plasma (median 0.9, range 0.4--3.1). Plasma from 16 of 18 (89%) dogs with IMHA contained citrullinated histone H3, but this marker was identified in only 4 of 30 (13%) of healthy dogs. Excessive NETosis appears to be a feature of canine IMHA, and may contribute to the prothrombotic state in this disease. FELINE *T. FOETUS* CYTOTOXICITY CAN BE INHIBITED BY SELECTIVE, SMALL‐MOLECULE CYSTEINE PROTEASE INHIBITORS {#jvim13002-sec-0049} ---------------------------------------------------------------------------------------------------------- **[Katie Tolbert]{.ul}, Emily Gould, Mabre Brand** University of Tennessee, Knoxville, Tennessee, USA *Tritrichomonas foetus* (Tf) is a mucosal protozoan parasite that infects the feline distal ileum and proximal colon resulting in chronic diarrhea. Tf has a worldwide distribution with no consistently effective drugs to treat the infection. Cysteine proteases (Cp) have recently been demonstrated to promote adhesion‐dependent cytotoxicity of feline Tf to the intestinal epithelium. These results support further investigation of Cp inhibitors as potential therapeutic targets for ameliorating the pathological effects of feline trichomonosis. To reduce host toxicity, anti‐protease drugs must be tailored to inhibit only those proteases produced by pathogens that are responsible for cytopathic effects and that have redundant mechanisms of production or function with other host proteases. Thus, the aim of this study was to investigate currently available small‐molecule Cp inhibitors for their feasibility in targeting specific regions of feline Tf Cp activity and ameliorating Tf‐induced cytopathogenicity. The effect of small molecule Cp inhibitors on specific Cp activities of protein extracts obtained from 4 feline Tf isolates were identified by means of SDS‐PAGE *In gel* zymography in the presence or absence of 0.1--1.0 mM calpain inhibitor, antipain, cystatin, leupeptin, chymostatin, WRR, and K11777. The effect of inhibition of specific Cp activity on Tf cytopathogenicity was determined using feline Tf that were allowed to adhere to monolayers of porcine intestinal epithelial cells (IPEC‐J2) in co‐culture. The cytopathogenic effect of Tf following inhibition of specific Cp activity was evaluated by light microscopy, crystal violet spectrophotometric analysis and immunoblotting Tf‐infected IPEC‐J2 cells for the M30 antigen of cleaved cytokeratin 18 as a marker of apoptosis. A minimum of 8 replicates were performed for each cytotoxicity experiment. Data were analyzed using Systat software (*P* \< 0.05). Patterns of gel zymography, demonstrated by 4 different feline isolates of Tf, were similar and revealed the ability of the vinyl sulfone Cp inhibitors (K11777, WRR) and cystatin to target specific zones of feline Tf Cp activity. These inhibitors had no effect on Tf growth and significantly inhibited cytotoxicity towards the intestinal epithelium in all 4 feline Tf isolates tested (*P* \< 0.001). These studies establish that currently available small‐molecule inhibitors of cysteine proteases are capable of inhibiting specific regions of feline Tf cysteine protease activity and ameliorating Tf‐induced cytopathogenicity to the intestinal epithelium in vitro. The results of these studies provide strong evidence‐based justification for identification of the specific cysteine proteases inhibited. Moreover, examination of the effect of specific cysteine protease inhibitors on amelioration of clinical signs in cats naturally infected with Tf is warranted. OUTCOME FOLLOWING URETERAL STENT PLACEMENT IN DOGS FOR BENIGN URETERAL OBSTRUCTIONS: 44 DOGS (57 URETERS) 2009--2013 {#jvim13002-sec-0050} -------------------------------------------------------------------------------------------------------------------- **Philippa Pavia^1^, [Allyson Berent]{.ul}^1^, Chick Weisse^1^, Demetrius Bagley^2^** ^1^The Animal Medical Center, New York, New York, USA, ^2^Thomas Jefferson University, Philadelphia, Pennsylvania, USA Ureteral obstructions are a serious clinical problem in dogs and traditional surgery is associated with a high complication rate. Ureteral stenting provides a minimally‐invasive option for immediate decompression, and is considered the standard‐of‐care in human medicine. To describe the use of double‐pigtail ureteral stents in dogs for benign ureteral obstructions hypothesizing it is a safe and effective treatment option. The diagnosis of benign ureteral obstruction was made via ultrasonography, radiography, and ureteropyelography. Ureteral stents were placed endoscopically and/or surgically, with fluoroscopic‐guidance. The medical records were reviewed for pre‐, intra, and postoperative data and outcome. 44 dogs (57 ureters) underwent stent placement for ureterolithiasis (48/57 \[84%\]), stricture (5/57 \[9%\]), or both (4 \[7%\]). Endoscopic/fluoroscopic or surgical/fluoroscopic techniques were successful in 45/55 (82%) and 12/12 ureters, respectively. Pre‐operative azotemia was present in 52%(23/44) (median 2 mg/dL) and 37%(16/43) remained azotemic (median 1.3) after decompression. Urinary tract infections were present in 59%(26/44) pre‐operatively, and 58%(25/43) post‐operatively, 76%(19/25) clearing with treatment. The perioperative mortality rate was 2% (1/44), and was not procedure‐related. Three major complications occurred during stent placement. Major peri‐operative, short‐term, and long‐term complications requiring further intervention occurred in 0%, 4%, and 14%, respectively. Median follow‐up time was 1,158 days (range, 3‐\> 1,555), with 30/44 alive at last follow‐up. Findings suggest that ureteral stents are a safe and effective minimally invasive short and long‐term treatment option for benign ureteral obstructions in dogs. Complications were typically minor, but may necessitate stent exchange or use of an alternative device. PROGNOSTIC VALUE OF SYMMETRIC DIMETHYLARGININE (SDMA) TO CREATININE RATIO IN DOGS AND CATS WITH CHRONIC KIDNEY DISEASE (CKD) {#jvim13002-sec-0051} ---------------------------------------------------------------------------------------------------------------------------- **[Mahalakshmi Yerramilli]{.ul}^1^, Murthy Yerramilli^1^, Edward Obare^1^, Dennis Jewell^2^, Jean Hall^3^** ^1^IDEXX Laboratories Inc, Westbrook, Maine, USA, ^2^Hill\'s Pet Nutrition, Inc, Topeka, Kansas, USA, ^3^Oregon State University, Corvallis, Oregon, USA SDMA is one of the dimethylated derivatives of arginine and is released into cytoplasm after proteolysis. As circulating SDMA is mostly eliminated by the kidneys, plasma concentrations are affected by changes in GFR. We have previously shown that increased serum SDMA concentrations were observed in cats and dogs with reduced renal function. We have also demonstrated that serum SDMA allows for earlier detection of chronic kidney disease (CKD) in cats and dogs when compared with serum creatinine. In general, serum SDMA and creatinine concentrations correlate with each other (R^2^ ˜ 0.7--0.9). However in some CKD patients, SDMA concentrations increased disproportionately compared with serum creatinine. This discordance is seen in both cats and dogs, but the prevalence is much higher in cats. This observed discordance in CKD patients can be quantified in terms of a SDMA to creatinine ratio; the higher the ratio the greater the discordance. Although the typical SDMA to creatinine ratio is \<10 in the majority of animals with CKD, we have observed that the larger the SDMA to creatinine ratio (\>10) in CKD patients, the greater the chance of mortality. It should be noted that the ratio only applies to CKD animals; dogs and cats with creatinine and/or SDMA concentrations within normal limits can have ratios \>10. We have previously reported that the upper normal reference limit for SDMA in dogs and cats is \<14 μg/dL. For example, in a retrospective study consisting of 21 CKD cats, 15/16 that had a SDMA to creatinine ratio of \>10 were deceased within 1 year, whereas 5/5 CKD cats that had a normal ratio of SDMA to creatinine (\<10) were still alive at that time. Similar trends were observed in another study with dogs. Kaplan‐Meier survival curves were generated for both these cats and dogs based on SDMA concentrations greater than and less than 14 µg/dL. Cats with SDMA \<14 µg/dL survived approximately 1.6 times longer than cats with SDMA \>14 µg/dL. Dogs with SDMA \<14 µg/dL survived approximately 2.6 times longer compared with dogs with SDMA \>14 µg/dL. All cats and dogs were maintained with high quality care, including optimal nutrition and veterinary health care with opportunities for socialization and play time with caretakers and with daily opportunities to exercise, and play with toys. Elevated SDMA concentrations in general, and SDMA to creatinine ratios in particular, for feline and canine CKD patients could potentially provide prognostic value. These results warrant further study. THE CONVERGENCE OF EPIGENETICS AND CD8^+^ T CELL DYSFUNCTION IN THE FELINE IMMUNODEFICIENCY VIRUS MODEL OF VIRAL PERSISTENCE {#jvim13002-sec-0052} ---------------------------------------------------------------------------------------------------------------------------- **[Jonathan Fogle]{.ul}, Yan Wang** North Carolina State University College of Veterinary Medicine, Raleigh, North California, USA Generation of mature, fully‐functional CD8^+^ lymphocytes is dependent upon epigenetic changes such as DNA demethylation and histone acetylation at the promoter regions of cytokines essential for CD8^+^ T cell function and differentiation. Conversely, dysfunctional CD8^+^ T cells exhibit altered patterns of methylation and acetylation, and failure of complete differentiation, suggesting that there are early transcriptional events contributing to the induction of CD8^+^ T cell dysfunction. For many persistent viral infections such as Herpes Simplex Virus (HSV), Hepatitis C Virus (HCV), Rotavirus (RV), Human Immunodeficiency Virus (HIV), and Feline Immunodeficiency Virus (FIV), virus specific CD8^+^ T cell activation is followed by the induction of CD8^+^ T cell dysfunction, which is characterized by lack of interleukin‐2 (IL2) production and poor proliferation following T cell receptor (TCR) stimulation. Loss of IL2 production signals the onset of a progressive decline in cytokine production and antiviral function. Using the feline immunodeficiency model (FIV) for lentiviral persistence, we have clearly demonstrated that virus‐activated CD4^+^CD25^+^ T regulatory (Treg) cells provide a strong inhibitory signal and induce cell cycle arrest and down‐regulation of IL2 in *activated* CD8^+^ lymphocyte targets early and progressively during the course of infection. Further, we have shown that Treg cells are activated by FIV infection and induce the expression of the repressive transcription factor Foxp3 in CD8^+^ T cell targets following Treg cell / CD8^+^ coculture. Finally, we have recently demonstrated that Foxp3 can bind the IL2 promoter in "bulk" (virus non‐specific) CD8^+^ T cells and that blocking epigenetic rearrangements prevents Foxp3 binding to the IL2 promoter in virus non‐specific CD8^+^ T cells. Recent findings from our laboratory suggest that early epigenetic changes (histone acetylation and DNA demethylation), while essential for antiviral function, *render the CD8* ^*+*^ *T cell highly permissive to Treg‐induced, Foxp3‐mediated repression of IL2 transcription*. Our hypothesis is that activated Treg cells exploit virus‐induced epigenetic changes in CD8^+^ T cells to cause dysfunction. The purpose of these investigations was to demonstrate that Foxp3 binds the IL2 promoter in virus specific CD8^+^ T cells. Using chromatin immunoprecipitation (ChIP) and bisulfite reduction of DNA, we have assessed histone acetylation and DNA demethylation at the IL2 promoter region in virus specific CD8^+^ T cells from FIV^+^ cats. Briefly, virus specific T cells were identified by ex‐vivo proliferation in response to inactivated virus. These cells were then purified and co‐cultured with autologous Treg cells. In concert with our previous work in virus non‐specific CD8^+^ T cells, our data suggests that following coculture with autologous Treg cells, virus specific CD8^+^ T cells exhibit increased Foxp3 mRNA and decreased IL2 mRNA by RT‐qPCR (n = 2). More importantly, our most recent ChIP studies demonstrate that Foxp3 binds the IL2 promoter in virus specific CD8^+^ T cells from FIV^+^ cats following Treg cell coculture (n = 2). Collectively, these results suggest that it will be possible to block epigenetic rearrangements in virus specific CD8^+^ T cells thereby blocking Foxp3 binding to the IL2 promoter. These mechanistic studies will provide a new avenue of investigation for restoring antiviral CD8^+^ T cell function during the course of persistent viral infection. PHARMACOKINETICS OF VORICONAZOLE IN HEALTHY CATS {#jvim13002-sec-0053} ------------------------------------------------ **[Polina Vishkautsan]{.ul}^1^, George Thompson^2^, Mark Papich^3^, Jane Sykes^1^** ^1^UC Davis VMTH, Davis, California, USA, ^2^UC Davis School of Medicine, Davis, California, USA, ^3^North Carolina Unversity, Raleigh, North California, USA Voriconazole is a potent azole antifungal drug but is contraindicated in cats because of life‐threatening adverse effects. We sought to determine the pharmacokinetics of voriconazole in healthy cats after oral and IV administration so that a safe dose might be established. Six cats were administered 1 mg/kg of voriconazole IV. Plasma voriconazole concentrations were measured at multiple time points for 24 hours after administration, using high performance liquid chromatography. Subsequently, voriconazole suspension was administered to 3 groups of 2 cats at 4, 5 and 6 mg/kg PO. Plasma concentrations were measured at multiple time points for 24 hours after administration. The pharmacokinetics of tablet and suspension preparations were also compared using a range of doses from 4.0 to 4.7 mg/kg. Plasma measurements were performed over a 72‐hours period after administration. Voriconazole half‐life after IV administration was approximately 20 hours. After oral administration, maximal plasma concentrations were reached in 30--60 minutes. A dose of 4 mg/kg resulted in optimal plasma drug concentrations (1--4 µg/mL). The predicted half‐life after oral administration was 80--90 hours. Adverse effects included hypersalivation coincident with oral suspension administration (not tablets) and miosis. Miosis was associated with peak plasma voriconazole levels \>2.5 µg/mL and persisted up to 48 hours. Voriconazole has excellent oral bioavailability in cats. An oral dose of 4 mg/kg q48h may maintain adequate antifungal plasma concentrations, but multi‐dose administration studies are required. Miosis was a significant adverse effect and occurred at plasma concentrations at the high end of the target range recommended for humans. PAIN MANAGEMENT AND END OF LIFE CARE: RESULTS OF A NATIONAL, CROSS‐SECTIONAL, SURVEY OF SMALL ANIMAL OWNERS {#jvim13002-sec-0054} ----------------------------------------------------------------------------------------------------------- **[Roschelle Heuberger]{.ul}^1^, Michael Petty^2^, Page Burnia^1^, Jill Prior^1^** ^1^Central Michigan University, Mt. Pleasant, Michigan, USA, ^2^Arbor Pointe Veterinary Hospital, Canton, Michigan, USA The objective of this study was to describe knowledge, attitudes and beliefs (KAB) of a national sample of owners with regard to End of Life (EOL) care for their pets. The hypothesis included: paradigms for EOL decision‐making are few and dependent on factors such as owner education. EOL decisions are influenced by many variables, requiring large samples (n = 945) for data segmentation. Additional objectives included: describing EOL and Quality of Life (QOL) awareness, beliefs regarding Pain Management (PM), and recognition of terms such as "in home care" for pets by owners. This study was "exempt category \#2 anonymous survey" by Federal Statute on 2/12/14 by Institutional Review Board. Computerized links opened with informed consent. Instrumentation was developed through modifications of validated surveys of KAB in EOL. Veterinarians and specialty associations with known involvement in EOL and QOL were consulted, followed by owner focus groups and piloting. Designed in Survey Monkey^™^, the data were password protected. Recruitment through social media was performed (2014--2015) and data were processed in SPSS v.21. Respondent age was μ = 45.4 ± 13 years, μ = 2 ± 1.2 pets/household, pet age μ = 3 ± 1.7 years and pet losses/lifetime μ = 10 ± 1.2. Only 4% were first time pet owners, and only 14% had pet health insurance. Respondents were primarily: Female (92%) Caucasian (94%) Married (55%) Religiously Unaffiliated (41%), Education \>HS (75%), Non‐Healthcare Occupation (75%). Familiarity with "hospice" or "in home EOL care for pets" was \<33%, QOL rating scales was 20%, but willingness to use QOL scales in the future was \>70%. In terms of PM in EOL care, owners were likely or very likely to use oral medication in pill form (80%), liquid (82%), injection (70%), acupuncture (55%), acupressure/massage (64%), physical therapy (67%), hydrotherapy (61%) surgery (62%) nerve block (41%) corticosteroid injections (52%), opt for euthanasia (15%). Other (5%) included Chiropractic, transdermal patch, laser, herbals, naturopathy, aromatherapy, prayer etc. Thirty‐five percent of owners reported fiscal concerns as influential in continued EOL care. If pain could not be controlled almost 80% of owners said they would euthanize. Further data segmentation is warranted to discern differences among owners with regard to KAB in EOL, PM, QOL and euthanasia. These findings will provide a basis for educational outreach and protocol development for owners of ill or geriatric animals. DISCOVERY AND CLINICAL EFFECTIVENESS OF A COMPOSITION THAT PROMOTES HAIR GROWTH (PATENT PENDING) {#jvim13002-sec-0055} ------------------------------------------------------------------------------------------------ **Gen‐Ichiro Soma^1^, [Hiroshi Okawa]{.ul}^2^** ^1^BioMedical Research Group Inc., Tokyo, Japan, ^2^Scarecrow Inc., Tokyo, Japan To overcome alopecia related to a variety of diseases in pets, we have identified materials that can promote hair growth, particularly in bald spots, leading to the development of products that can be helpful in the field of food and medicine. We developed a composition that promotes hair growth in animals, which can be used in numerous formulations, including drugs, quasi‐drugs, dressings, adhesive bandages, supplements, foods, snacks, water, hair care supplies, dental supplies, clothes, shoes, and socks. This formulation contains lipopolysaccharide (LPS) as an active ingredient. LPS tablets and pine bark polyphenol tablets were orally administered to dogs, cats, rabbits, and hamsters with alopecia in a study conducted in Japan, with the cooperation of practitioners. The study involved no restriction on food intake and medication, and prohibited the concomitant use of other supplements. The animals were divided into 2 groups, those that received LPS tablets alone and those that received LPS tablets and pine bark polyphenol tablets. The duration of the study was approximately 30 days. Hair growth was noted in 10 of 24 animals (41.6%) administered LPS tablets alone and all 13 animals (100%) administered LPS tablets and pine bark polyphenol tablets. PREVALENCE OF CANINE INFECTIOUS RESPIRATORY DISEASE IN DOGS IN CHICAGO OUTBREAK (MARCH--APRIL 2015) {#jvim13002-sec-0056} --------------------------------------------------------------------------------------------------- **Jill A. Richardson^1^, Amy Glaser^2^, [Nyssa Reine‐Salz]{.ul}^1^, Edward J. Dubovi^2^** ^1^Merck Animal Health, Madison, New Jersey, USA, ^2^New York State Veterinary Diagnostic Laboratory, Cornell University, Ithaca, New York, USA Canine infectious respiratory disease complex (CIRDC) is a common disease complex caused by many different viruses and bacteria, including *Bordetella bronchiseptica*,*Mycoplasma cynos*, adenovirus type 2, distemper, influenza A virus, parainfluenza virus, pneumovirus and respiratory coronavirus. In March 2015, veterinarians in the Chicago area noted an increase in incidence of signs of canine infectious respiratory disease in dogs. Nasal and pharyngeal swabs from dogs showing clinical signs were submitted to the Cornell University Animal Health Diagnostic Center (AHDC). A canine respiratory polymerase chain reaction (PCR) screening panel was utilized which allows identification of the following CIRDC pathogens: *B. bronchiseptica*,*Mycoplasma cynos*, adenovirus type 2, distemper, influenza A, parainfluenza virus, pneumovirus and respiratory coronavirus. On March 16, 2015, a diagnostic sampling program was initiated which tested over 350 samples from dogs with respiratory signs using this screening panel as of April 23, 2015. Samples were collected by the reporting clinics and shipped based on the Cornell AHDC specifications. Of the over 350 screened sick dogs, 198 dogs tested positive for canine influenza using a PCR assay. None of the tested dogs were confirmed to have the H3N8 influenza strain. Further testing by Cornell and the University of Wisconsin identified the strain as Canine Influenza H3N2. The H3N2 influenza virus is of avian origin and was first isolated from clinically ill dogs in China in 2006 and South Korea in 2007. Canine H3N2 influenza virus has been associated with severe respiratory signs and other clinical signs such as fever, reduced body weight, and interstitial pneumonia. In addition, 24 dogs tested positive for parainfluenza virus, nine dogs tested positive for *B. bronchiseptica*, three dogs tested positive for adenovirus type 2, 27 dogs tested positive for pneumovirus and 29 dogs tested positive for respiratory coronavirus. Records were reviewed to determine the vaccination status of the positive dogs. None of the 198 dogs that tested positive for the H3N2 influenza virus were vaccinated against the H3N8 influenza virus. Twenty‐two of the 24 dogs that tested positive for parainfluenza virus had received an injectable distemper, adenovirus type 2, parvovirus and parainfluenza virus combination vaccine, 13 had received a monovalent *B. bronchiseptica* vaccine and five had been vaccinated with an intranasal *B. bronchiseptica* and parainfluenza combination product. Of the five dogs that received the *B. bronchiseptica* combination vaccine, timing of vaccination suggests that the virus sampled may have been of vaccine origin. The information gathered from this testing program represents the first time that the H3N2 influenza virus has been isolated from dogs in the United States. A point of origin could not be determined. In addition, data collected supports the role of parainfluenza virus as a major preventable pathogen in CIRDC and route of vaccination should be considered in vaccination protocols. EQUINE {#jvim13002-sec-0057} ====== A CLINICAL AND NECROPSY EVALUATION OF ACUTE LUNG INJURY (NALI) AND ACUTE RESPIRATORY DISTRESS SYNDROME (NARDS) IN NEONATAL FOALS {#jvim13002-sec-0058} -------------------------------------------------------------------------------------------------------------------------------- **[Daniela Bedenice]{.ul}, Samuel Jennings, Mauricio Solano, Mary Rose Paradis** Cummings School of Veterinary Medicine at Tufts University, North Grafton, Massachusetts, USA The study purpose was to evaluate the clinical and post‐mortem characteristics of neonatal foals fulfilling the criteria of neonatal ALI (nALI) and ARDS (nARDS). A retrospective cohort analysis of 786 foals (1990--2014) admitted to a university hospital under 4‐weeks old was performed. Diagnostic criteria for nALI and nARDS were applied to a preexistent database, including all patients with lateral thoracic radiographs and arterial blood gas results obtained on room air within 24‐hours of each other. Neonates born following c‐section, induction or with a diagnosis of prematurity were excluded. The diagnostic PaO2/FiO2 threshold was adjusted for age as previously described (Wilkins‐2007). Thoracic radiographs were evaluated using a pre‐existent scoring system. Histopathologic abnormalities were graded according to the distribution and frequency (0--3) of edema/congestion/hypercellularity, fibrin exudation, hyaline membranes, type‐2 pneumocyte hyperplasia, and interstitial fibrosis to establish a diagnosis of diffuse alveolar damage (DAD). All data were presented descriptively. Thirty‐four of 786 (4.3%) admitted foals fulfilled the clinical and PaO2/FiO2 threshold criteria for nALI (24/786) and nARDS (10/786). Of these, 10/24 (42%) and 4/10 (40%) survived, respectively. Histopathologic review of lung tissue was available for 10 affected foals, and showed severe, diffuse edema or congestion in all patients. Hyaline membranes, and thus DAD, were observed in 7/10 (70%), being rare in 4/7 (57%). Interstitial fibrosis was not identified. Associated pulmonary diagnoses in foals with DAD included meconium aspiration (n = 2), EHV‐1 infection (n = 2), and suppurative bronchopneumonia (n = 1). In conclusion, histopathology showed exudative rather than proliferative or fibrotic DAD in foals with nALI or nARDS. METABOLIC RESPONSES TO A STANDARDIZED FIELD EXERCISE TEST IN ARABIAN ENDURANCE HORSES WITH A HISTORY OF EXERTIONAL RHABDOMYOLYSIS {#jvim13002-sec-0059} --------------------------------------------------------------------------------------------------------------------------------- **[Erica McKenzie]{.ul}^1^, Lauren Eyrich^1^, Mark Payton^2^, Stephanie Valberg^3^** ^1^Oregon State University, Corvallis, Oregon, USA, ^2^Oklahoma State University, Stillwater, Oklahoma, USA, ^3^University of Minnesota, Saint Paul, Minnesota, USA At least 4% of Arabian endurance horses suffer from exertional rhabdomyolysis (ER) of unknown etiology. This study compared muscle histopathology and metabolic responses to a field exercise test between endurance Arabians with and without previous ER (ER: n = 10, age 15 ± 6 years; control: n = 9, 13 ± 6 years). No horses possessed the *GYS1* mutation. After 24--48 hours stall rest, paired ER and control horses, fitted with a telemetric ECG, performed 47 minutes of standardized intervals of walk and trot, and one interval of canter. Blood samples were obtained before, immediately after and 3 hours post‐exercise, and plasma was frozen immediately in liquid nitrogen. Percutaneous gluteus medius muscle samples frozen in liquid nitrogen were obtained before and 3 hours post‐exercise for glycogen analysis. Pre‐exercise muscle samples were also fixed in formalin for histopathologic analysis. Fisher\'s exact test, t‐test, and ANOVA were performed (*P* \< 0.05). No horses displayed clinical signs of ER and log‐transformed serum creatine kinase activity 3 hours post‐exercise was not significantly different between ER and control Arabians. Muscle glycogen, heart rate, PCV, and plasma total protein, glucose, lactate and electrolyte concentrations did not differ between exercising ER and control horses. ER horses had more central nuclei in mature myofibers, and higher myopathic scores compared with control horses. Arabian endurance horses with a history of ER have similar metabolic responses to a submaximal exercise test as healthy Arabians, in spite of histopathologic indicators of chronic ER. Thus, Arabian ER does not appear to be associated with a consistent metabolic myopathy. LONGITUDINAL EFFECT OF A MULTI‐STRAIN PROBIOTIC ON THE MICROBIOTA OF NEONATAL FOALS {#jvim13002-sec-0060} ----------------------------------------------------------------------------------- **[Angelika Schoster]{.ul}^1^, Henry Staempfli^2^, Miranda Abrahams^2^, Mohammad Jalali^3^, Luca Guardabassi^4^, Scott Weese^3^** ^1^University of Zurich, Vetsuisse faculty, Clinic for Equine Internal Medicine, Zurich, Switzerland, ^2^University of Guelph, Ontario Veterinary College, Departement of Clinical Studies, Guelph, Ontario, Canada, ^3^University of Guelph, Ontario Veterinary College, Departement of Pathobiology, Guelph, Ontario, Canada, ^4^University of Copenhagen, Faculty of Health and Medical Sciences, Departement of veterinary Disease Biology, Frederiksberg, Copenhagen, Denmark It is suggested that probiotics modify the intestinal microbiota but scientific proof is lacking. The objective of this study was to evaluate the effect of a probiotic containing strains selected for anti‐clostridial activity, on the intestinal microbiota of neonatal foals. 14 healthy foals were randomly allocated to receive a probiotic or placebo formulation for 3 weeks. The probiotic formulation contained two *Lactobacillus rhamnosus* strains, two *L. plantarum* strains and *B. animalis lactis*. Fecal samples were collected at 2, 4 and 6 weeks of age. Mothur analysis of next generation sequencing data was used to compare data between treatment groups. There was no effect of treatment group on alpha diversity indices, Chao richness, Simpson\'s diversity and Shannon evenness (all *P* \> 0.37). There were no significant differences in the relative abundances of any phyla, classes, or predominant genera, including *Bifidobacterium* (*P* = 0.37, 0.18 and 0.90 at 2, 4 and 6 weeks of age) and *Lactobacillus* (*P* = 0.61, 0.16 and 0.52 at 2, 4 and 6 weeks of age). Several differences between treatemnt groups were noted using LEfSe. Most notably at week 6, an unclassified Lactobacillaceae (LDA score 2.1, *P* = 0.016) was enriched in the probiotic group. There was no significant difference in beta diversity between groups, as assessed by unifrac applied to Jaccard and Yue&Clayton trees (all *P* \> 0.30). Probiotic treatment had limited effects on the composition of the microbiome. Further studies including larger numbers of foals have to be performed to assess whether this difference is significant and can be used to therapeutically modify the microbiome of foals. SKELETAL MUSCLE GENE EXPRESSION PROFILE IN TYPE 1 POLYSACCHARIDE STORAGE MYOPATHY {#jvim13002-sec-0061} --------------------------------------------------------------------------------- **Raffaella Teixeira, James Mickelson, Stephanie Valberg, [Molly McCue]{.ul}** University of Minnesota, St Paul, Minnesota, USA Type 1 Polysaccharide Storage Myopathy (PSSM1) is an inheritable neuromuscular disorder caused by a gain of function in skeletal muscle glycogen synthase that leads to excessive glycogen and abnormal polysaccharide accumulation in muscle fibers. PSSM1 horses utilize glycogen, but develop exercise intolerance and rhabdomyolysis during sub‐maximal exercise. The link between excessive glycogen, abnormal polysaccharide and rhabdomyolysis is unknown. Controlled regular exercise and a low starch/high fat diet improve PSSM1clinical signs. *We hypothesized that 1) excess glycogen and/or improper regulation of glycogen synthase results in altered gene expression in skeletal muscle energy metabolism pathways leading to rhabdomyolysis in PSSM1; and 2) that clinical improvement with daily exercise is due to normalization of gene expression in PSSM1 horses*. Gluteal muscle biopsies were collected from PSSM1 cases and controls on a high starch diet prior to training (t1), before (t2) and immediately after exercise (t3) following a standardized 3‐week training protocol. RNA was isolated and sequenced at 40‐million reads/sample; reads were mapped to EquCab2 and *de novo* assembly was performed to identify novel transcripts. Differential gene expression was measured by Edge‐R. 195, 294 and 739 genes were differentially expressed between cases and controls at t1, t2 and t3, respectively. The largest differences in expression were in cases between t1‐t3 (4,822 genes) and t2‐t3 (2,474 genes). Genes involved in glycogen metabolism, glycolysis and mitochondrial electron transport were differentially expressed between cases and controls. Understanding altered energy regulation in PSSM1 should lead to new therapies and new insights in metabolic myopathies and muscle glycogen regulation. BLOOD AND CEREBROSPINAL FLUID ALPHA‐TOCOPHEROL AND SELENIUM CONCENTRATIONS IN NEONATAL FOALS WITH NEUROAXONAL DYSTROPHY {#jvim13002-sec-0062} ----------------------------------------------------------------------------------------------------------------------- **[Carrie Finno]{.ul}^1^, Krista Estell^1^, Laramie Winfield^1^, Aaron Rendahl^2^, Jamie Textor^3^, Danika Bannasch^1^, Birgit Puschner^1^** ^1^University of California, Davis, Davis, California, USA, ^2^University of Minnesota, St. Paul, Minnesota, USA, ^3^Total Performance Equine, Martnez, California, USA Equine neuroaxonal dystrophy/equine degenerative myeloencephalopathy (NAD/EDM) is a neurodegenerative disorder affecting genetically predisposed foals maintained on an α‐tocopherol (α‐TP) deficient diet during the first year of life. We hypothesized that (1) the administration of E‐Se^®^ at 4‐days of age would have no significant effect on serum or cerebrospinal fluid (CSF) \[α‐TP\] in healthy foals and (2) serum and CSF \[α‐TP\], but not \[Se\], would be significantly decreased in NAD/EDM‐affected foals during the first year of life. Fourteen Quarter horse foals were included in the study; ten healthy foals supplemented with 0.02 mL/kg E‐Se^®^ (n = 5) or saline (n = 5) at 4 day of age and four unsupplemented NAD/EDM‐affected foals. Complete neurologic examinations, blood and CSF collections were performed before (4 day of age) and after supplementation at 10, 30, 60, 120, 180, 240 and 360 day of age. At 540 day, NAD/EDM affected foals and one unsupplemented healthy foal were euthanized and full necropsies performed. A significant decrease in blood and CSF \[α‐TP\] and \[Se\] was found during the first year of life in all foals, with the most significant changes in serum \[α‐TP\] occurring from 4--150 day. Dam \[α‐TP\] and \[Se\] significantly impacted foal concentrations through 10 day of age. An injection of E‐Se^®^ did not significantly increase CSF \[Se\], blood or CSF \[α‐TP\] in healthy foals. NAD/EDM‐affected foals had significantly lower CSF \[α‐TP\]. Despite all 14 foals remaining deficient in α‐TP, only the four genetically predisposed foals developed NAD/EDM. RELATIONSHIP BETWEEN TRANSCRANIAL MAGENTIC MOTOR EVOKED POTENTIAL LATENCIES AND AXON LOSS IN EQUINE CERVICAL VERTEBRAL STENOTIC MYELOPATHY {#jvim13002-sec-0063} ------------------------------------------------------------------------------------------------------------------------------------------ **[Cody Alcott]{.ul}^1^, Nicholas Jeffery^2^, David Wong^2^, Stephen Reed^2^, Brett Sponseller^2^, Hilary Hu^2^, Kate Hepworth‐Warren^2^, Elizabeth Whitley^2^** ^1^Veterinary Specialty Center of Tucson, Tucson, Arizona, USA, ^2^Iowa State University College of Veterinary Medicine, Ames, Iowa, USA Current diagnostic strategies for equine cervical vertebral stenotic myelopathy (CVSM) are limited to identification of structural changes within the vertebral canal and inferring spinal cord dysfunction. Delayed latencies of transcranial magnetic motor evoked potentials (TMMEPs) have been associated with spinal cord dysfunction in horses, although histopathologic confirmation and correlation with axon loss is lacking. The purpose of this study was to investigate the relationship between TMMEP latency and axonal loss in horses diagnosed with CVSM. The hypothesis is that cervical spinal cord axonal loss will correlate with TMMEP latency. TMMEP latencies from 20 horses were recorded bilaterally from the extensor carpi radialis and cranialis tibialis muscles. Ten horses had clinical signs consistent with CVSM (CVSM group) while the ten remaining horses presented for non‐neurologic disorders (control group). All horses had lateral cervical radiographs performed prior to humane euthanasia. Cerebrospinal fluid from all horses tested negative for *Sarcocystis neurona* (SAG2,4/3 ELISA). Postmortem harvesting of cervical spinal cords in all horses was via dorsal laminectomy (C1‐T2), in‐situ formalin fixation, transverse sectioning (cranial, dorsal and caudal to each intervertebral disc space). Tissue blocks were stained with luxol fast blue and an H&E counterstain. Digital images of the spinal cord transverse sections (20×) were acquired and pixel density counts were determined (Adobe Photoshop©) to determine axon density at each of the following locations within each section; dorsal lateral funiculus (DLF), dorsal funiculus (DF), ventral funiculus (VF) and lateral funiculus (LF). TMMEP latencies of the extensor carpi radialis and cranialis tibialis were compared between control and affected horses (unpaired t test). The ratio of axon densities at each site between CVSM and control horses (termed the CCPR) was plotted against the TMMEP latency of the cranialis tibialis muscle and examined for correlation. Individual funiculus axon density from the most severely affected transverse sections were plotted against TMMEP latency of the cranialis tibilais muscles to investigate the effect of location within the spinal cord. TMMEP latency showed a significant difference between groups for the cranialis tibialis muscle exclusively. Cranialis tibialis TMMEP latency linearly correlated with CCPR. Loss of axons was most severe in the ventromedial funiculi of CVSM horses. These data confirm that horses affected by CVSM have prolonged TMMEP latencies and that TMMEP latency correlates with severity of axon loss. Both of these findings indicate that TMMEP might be useful to identify and quantify severity of spinal cord damage in clinical CVSM. The severity of ventral funicular changes is a unique finding and might suggest that this area is more vulnerable to injury in horses with CVSM than previously recognized. NOVEL DETERMINANT CONFERS MACROLIDE, LINCOSAMIDES, AND STREPTOGRAMIN B RESISTANCE IN *RHODOCOCCUS EQUI* {#jvim13002-sec-0064} ------------------------------------------------------------------------------------------------------- **[Steeve Giguere]{.ul}^1^, Elisa Anastasi^2^, Londa Berghaus^1^, Mary Hondalus^1^, Jennifer Willingham^1^, Sonsiray Alvarez^2^, Noah Cohen^3^, Marilyn Roberts^4^, Jose Vazquez‐Boland^2^** ^1^University of Georgia, Athens, Georgia, USA, ^2^University of Edinburgh, Edinburgh, UK, ^3^Texas A&M University, College Station, Texas, USA, ^4^University of Washington, Seattle, Washington, USA The incidence of macrolide and rifampin resistance in *R. equi* isolated from foals has increased considerably in recent years. The objective of this study was to identify the molecular mechanism of emerging macrolide resistance in *R. equi* and to determine its transferability. Macrolide‐resistant (n = 62) and macrolide susceptible (n = 62) clinical isolates of *R. equi* from foals in the USA were studied. Whole genome sequencing of a sample of 18 macrolide‐resistant and 6 macrolide‐susceptible *R. equi* was performed. PCR was used to screen for the presence of the resistance determinant in the other isolates. Mating experiments were performed to document transfer of the determinant. The genomes of resistant isolates were virtually identical whereas there was marked chromosomal variability among susceptible isolates, suggesting expansion of a resistant clone. A novel *erm* gene, *erm*(46) was identified only in resistant isolates. There was perfect association between macrolide resistance and presence of *erm*(46) as detected by PCR in 124 isolates of *R. equi*. Expression of *erm*(46) in a macrolide‐susceptible strain of *R. equi* induced high level resistance to macrolides, lincosamides, and streptogramins B, but not to other classes of antimicrobial agents. Transfer of *erm*(46) from resistant to susceptible strains of *R. equi* was confirmed and occurred at a transfer frequency of up to 2 × 10^−3^. This is the first molecular characterization of macrolide, lincosamides and streptogramins B resistance in *R. equi*. Resistance is caused by a novel *erm* gene, *erm*(46), which is transferrable likely by conjugation. DETERMINING OPTIMAL SAMPLING SITE FOR *STREPTOCOCCUS EQUI* SUBSP *EQUI* CARRIERS USING A LOOP‐MEDIATED ISOTHERMAL PCR ASSAY {#jvim13002-sec-0065} --------------------------------------------------------------------------------------------------------------------------- **[Ashley Boyle]{.ul}^1^, Darko Stefanovski^1^, Shelley Rankin^2^** ^1^Clinical Studies New Bolton Center, University of Pennsylvania, Kennett Square, Pennsylvania, USA, ^2^School of Veterinary Medicine Pathobiology, University of Pennsylvania, Philadelphia, Pennsylvania, USA We hypothesized that samples obtained from the guttural pouch would be more sensitive than samples obtained from the nasopharynx to identify carriers of *Streptococcus equi* (*S. equi)* and that a Loop‐mediated isothermal (LAMP) PCR assay that targeted the *eqbE* gene would be more sensitive than a realtime PCR assay that targeted the *seeI* gene. Three samples were collected from each horse: nasopharyngeal flocked nylon swab (NPFS), nasopharyngeal wash (NPW), and endoscopically‐guided guttural pouch lavage (GPL). *eqbE* LAMP assay was performed on NPFS, NPW and GPL samples. The GPL sample was split into 3 aliquots; *S. equi* culture, *eqbE* LAMP and *seeI* realtime PCR. Logistic regression and area under the receiver‐operator curve (ROC) were performed using STATA 13. *P*‐values ≤0.05 were considered significant. A total of 123 samples were obtained from 40 horses (41 NPFS, 38 NPW, 44 GPL). 1/41 NPFS, 6/38 NPW and 24/44 GPL samples were positive by *eqbE* LAMP. 18/26 GPL samples were positive with *seeI* PCR. *S. equi* was isolated from 4/44 GPL samples. GPL was the best sample to detect carriers compared to NPFS (OR 48, *P* \< 0.001) and NPW (OR 6.4, *P* = 0.001). When *eqbE* LAMP results were compared to an endoscopically abnormal guttural pouch, sensitivity was 74.07% and specificity was 76.47%: LAMP ROC = 0.75, *seeI* PCR ROC = 0.78. Sensitivity and specificity of *eqbE* LAMP was 83.3% and 65.4% compared to the *seeI* PCR: ROC = 0.70. Our study demonstrates that GPL should be used to detect *S. equi* carriers and that *eqbE* LAMP assay was comparable to *seeI* PCR. PRO‐RESOLUTION MECHANISMS OF INFLAMMATION IN EQUINE RECURRENT AIRWAY OBSTRUCTION: TAMOXIFEN AS A NEW THERAPEUTIC OPTION {#jvim13002-sec-0066} ----------------------------------------------------------------------------------------------------------------------- **[Benjamin Uberti]{.ul}, Natalia Morales, Constanza Borlone, Claudio Henríquez, José Sarmiento, Gabriel Morán** Universidad Austral De Chile, Valdivia, Región De Los Ríos, Chile Equine recurrent airway obstruction (RAO) can severely limit athletic function and quality of life of adult horses. Inhalation of a variety of environmental allergens induces chronic non‐infectious respiratory inflammation, leading to hyper‐responsiveness of the lower airways, increased production of mucus and airway remodeling. Swelling and narrowing of the airways leads to decreased lung function and limited gas exchange. There is evidence that many cells and structures of the lung play an important role in the pathophysiology of the disease: the contribution of bronchial epithelial cells, vascular endothelium, lymphoid and myeloid cells has been well documented. The interaction of an allergen with resident innate immune cells induces the arrival of more inflammatory cells, via cytokines and chemokines. Once the inflammatory reaction has fulfilled its task, pro‐resolutive mechanisms take the stage: the shut‐down of chemokine signaling leads to the abrogation of inflammatory cell influx (mainly neutrophils). Neutrophils initiate their apoptotic program and release molecules that attract macrophages, which in turn perform neutrophil clearance. The phagocytosis of apoptotic neutrophils eventually reprograms macrophages towards an anti‐inflammatory phenotype. The successful completion of the pro‐resolutive mechanism leads to the restoration of tissue homeostasis. Our preliminary results demonstrate that tamoxifen, a selective estrogen receptor modulator used as treatment in all stages of estrogen‐positive human breast cancer, has a marked effect on equine neutrophil function and apoptosis. In vitro results showed that tamoxifen significantly inhibited neutrophil chemotaxis in horses and humans in response to IL‐8, from 1 μM onwards (*P* \< 0.05, one‐way ANOVA). Additionally, tamoxifen significantly inhibited production of reactive oxygen species from 2.5 μM onwards. Early neutrophil apoptosis was induced in approximately 90% of cells after 30 minutes of incubation with tamoxifen at 5 μM (Sarmiento, Perez, Morales *et al*, 2013). This induction of apoptosis coincided with in vivo results, in which neutrophils obtained from tamoxifen‐treated animals showed a marked increase of early apoptosis. IV and PO pharmacokinetic analysis of tamoxifen in horses is currently under way. A proof‐of‐concept trial was performed recently, in which 4 stalled horses with acute exacerbation of RAO were treated with tamoxifen (100 mg PO q 24 hours for 7 days). This dose was extrapolated from human and murine pharmacokinetic data. Horses received no other treatment and underwent no management changes. There was a marked decrease in bronchoalveolar lavage neutrophil count after treatment (day 0 = 28% ± 6; day 8 = 4% ± 7). Clinical parameters (spirometry, respiratory rate and pattern) improved from 3 days of treatment onwards. Tracheal mucus content did not change significantly over the course of treatment. These results suggest that tamoxifen has an effect on neutrophil survival in the pulmonary environment, and warrant further investigation. Recruitment of additional subjects with acute exacerbation of RAO is ongoing. THE IMPACT OF PROLONGED HYPERINSULINAEMIA ON THE DOWNSTREAM INSULIN SIGNALING PATHWAY IN EQUINE STRIATED MUSCLE AND DIGITAL LAMELLAE {#jvim13002-sec-0067} ------------------------------------------------------------------------------------------------------------------------------------ **Allison Campolo^1^, Lauren Keith^1^, Melody de Laat^2^, Christopher Pollitt^2^, [Veronique Lacombe]{.ul}^1^** ^1^Oklahoma State University, Stillwater, Oklahoma, USA, ^2^University of Queensland, Brisbane, Queensland, Australia Equine metabolic syndrome, an increasingly recognized disease, is characterized by obesity and insulin resistance. Although hyperinsulinemia predisposes horses to developing laminitis, the exact relationship between these conditions remains undetermined. We hypothesized that gene expression of proteins involved in the downstream insulin signaling and glucose transport pathways will be altered during prolonged hyperinsulinemia‐induced laminitis. Standardbred horses were treated with a prolonged euglycaemic, hyperinsulinaemic clamp (p‐EHC) for 48 hours or a balanced electrolyte solution. Gene expression of key proteins involved in the insulin signaling pathway was evaluated in archived striated muscle and lamellar samples using real‐time reverse transcription PCR with primers selected for insulin receptor substrate‐1 (IRS‐1), AKT‐2, and glycogen synthase kinase beta (GSK‐3β). Gene expression of the basal glucose transporter 1 (GLUT‐1) and the insulin‐sensitive GLUT‐4 were evaluated using RT‐PCR. The p‐EHC group became markedly hyperinsulinaemic while maintaining normoglycemia and developed clinical laminitis. There was no significant difference in gene expression of IRS‐1, AKT‐2, GSK‐3β, GLUT‐1 and GLUT‐4 in skeletal muscle and lamellar tissue between groups. In contrast, there was a significant upregulation of AKT‐2, GSK‐3β, GLUT‐1 and GLUT‐4 expression in cardiac tissue of the p‐EHC group compared to controls (*P* \< 0.05). These data suggested that the prolonged, euglycaemia hyperinsulinaemic clamp induced an increase in insulin sensitivity in the heart, as well as a transcriptional activation of glucose transport. In addition, the lack of downregulation of the insulin signaling pathway in both the skeletal muscle and digital lamellae supports that insulin resistance is not required for the onset of laminitis. TISSUE MICRODIALYSIS STUDIES OF EQUINE LAMELLAR ENERGY METABOLISM AND MICROVASCULAR BLOOD FLOW {#jvim13002-sec-0068} ---------------------------------------------------------------------------------------------- **[Carlos E. Medina‐Torres]{.ul}^1^, Christopher C. Pollitt^1^, Claire Underwood^1^, E. Matias Castro‐Olivera^1^, Mark P. Hodson^2^, Dean W. Richardson^3^, Andrew W. van Eps^1^** ^1^Australian Equine Laminitis Research Unit, School of Veterinary Science, The University of Queensland, Gatton, Queensland, Australia, ^2^Metabolomics Australia, Queensland Node, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Brisbane, Queensland, Australia, ^3^New Bolton Center, Department of Clinical Studies, School of Veterinary Medicine, University of Pennsylvania, Kennett Square, Pennsylvania, USA Disruption of cellular energy pathways may play a role in the different forms of laminitis. The aim of this work was to characterize the normal lamellar bioenergetic profile and establish if changes in metabolite patterns and microvascular perfusion occur as a result of modifications in limb load cycling activity and during the development of experimentally induced sepsis‐associated laminitis. First, a microdialysis technique for serial measurement of lamellar tissue energy metabolites (glucose, lactate, pyruvate, urea and glycerol) over a 24 hours period was developed, and the lamellar energy profile of 14 clinically normal horses characterized. Subsequently, 9 horses instrumented with lamellar and skin microdialysis probes were subjected to sequential interventions designed to modify tissue perfusion (4 interventions) and limb load cycling activity and weight bearing (4 interventions). Urea (20 mmol/L) was added to the perfusion fluid to investigate if urea clearance (UC; a microdialysis‐based method used to assess local blood flow) could identify lamellar perfusion changes. Energy metabolite concentrations, UC, and the standard indices of energy metabolism lactate to glucose (L:G) and lactate to pyruvate (L:P) ratios were determined during each intervention. Then, changes in lamellar bioenergetic composition and perfusion during the development of experimentally induced, sepsis‐associated laminitis were investigated in 12 horses. Treatment (n = 6) and control (n = 6) horses received oligofructose (OF) and water via nasogastric tube. Lamellar metabolite composition, L:G, L:P and UC were determined bihourly and compared. Lastly, using lamellar and skin dialysate and tissue sections from healthy horses (n = 7), and lamellar dialysate from OFT (n = 4) and CON (n = 4) horses, targeted metabolomic analysis (i.e. 44 central carbon metabolites: CCM) was performed using liquid chromatography coupled with tandem mass spectrometry (LC‐MS/MS). The differences between lamellae and skin in healthy horses and the differences between OFT and CON horses were investigated. In the intervention studies, UC alone and in combination with fluctuations in energy metabolite patterns detected profound and mild changes in lamellar perfusion, respectively. Significant changes in lamellar energy metabolite composition and UC were observed when limb load cycling activity was modified: increases in glucose and UC, compatible with increased lamellar perfusion, were observed when limb activity increased. In the OF study, glucose decreased significantly in OFT lamellar dialysate (OFT‐L), and was consistently lower in OFT than in CON horses. Lactate was higher in OFT skin (OFT‐S) compared with OFT‐L. Pyruvate decreased significantly in OFT‐L and was significantly less than CON‐L. L:G and L:P increased significantly in OFT‐L and OFT‐S, but not CON‐L. Urea decreased significantly (increased UC) in OFT‐L and remained stable in CON‐L. Hyperaemia rather than ischaemia together with some bioenergetic disturbances occurred during the developmental phase of OF‐induced laminitis. Though perfusion increased, lamellar glucose was markedly reduced, which could have compromised lamellar basal epithelial cell metabolism despite the lack of definitive evidence of lamellar bioenergetic failure. Metabolomic analysis of lamellar and skin dialysate showed no difference in metabolite composition in these two tissues in healthy horses, and the metabolome of dialysate and tissue samples was also found to be similar. A difference in CCM between OFT and CON horses was observed. Metabolomic analysis of lamellar CCM was capable of differentiating OFT from CON horses. Lamellar malate, pyruvate, aconitate and glycolate, were identified as the source of differentiation between OFT and CON groups. These studies demonstrated that lamellar microdialysis and UC can be used to assess lamellar bioenergetic and perfusion changes. Microdialysis UC suggests an increase in lamellar perfusion during the development of sepsis‐associated laminitis. The findings also suggest that lamellar perfusion and energy balance may be related to limb load cycling and are particularly affected by ambulation, supporting the hypothesis that reduced limb load cycling, but not increased weight bearing, may be associated with reduced lamellar perfusion. INCRETINS POTENTIATE THE INSULIN RESPONSE TO ORAL GLUCOSE IN INSULIN‐RESISTANT PONIES {#jvim13002-sec-0069} ------------------------------------------------------------------------------------- **[Melody de Laat]{.ul}, Jessica van Haeften, Martin Sillence** Queensland University of Technology, Brisbane, Queensland, Australia Equine insulin secretion varies between individuals and is affected by diet and metabolic status, with dysregulation resulting in hyperinsulinaemia, a significant laminitis risk. Gastrointestinal hormones may exacerbate insulin responses to feeding. This study investigated whether the incretin hormones, glucagon‐like peptide‐1 (GLP‐1) and glucose‐dependent insulinotropic peptide (GIP), augment insulin secretion following oral glucose, compared with intravenous glucose infusion. Ponies (n = 9) received D‐glucose (0.75 mg/kg) in a low glycaemic ration (0.8%BW). Assays were validated for insulin, glucose and incretin analysis of blood samples collected before and every 30 minutes for 6 hours after feeding. Seven days later the experiment was replicated however the D‐glucose was administered intravenously at a variable rate matched to individual meal consumption (\<75 minutes). Ponies were stratified for insulin sensitivity status using basal values/proxies. Although intravenous glucose elicited a larger area under the curve for insulin (AUCinsulin), the insulin response to oral glucose was larger (*P* \< 0.05) and bimodal in insulin‐resistant (compared to insulin‐sensitive) ponies and correlated well with AUCglucose. Importantly, insulin resistance did not affect insulin responses to intravenous glucose, whereas with oral glucose the AUCinsulin:AUCglucose was higher (*P* \< 0.05) in insulin‐resistant ponies. Secretion of GLP‐1active and GIP, but not GLP‐1total, was higher (*P* \< 0.05) following oral glucose (compared to intravenous) and GLP‐1active was positively correlated (*P* \< 0.05) with insulin. Further, oral glucose stimulated a larger (*P* \< 0.05) AUCGLP‐1active in insulin‐resistant ponies. Thus, the larger insulin response to oral glucose in insulin‐resistant ponies is likely due in part to incretin action. In particular, GLP‐1active may be a key factor in equine insulin dysregulation. USE OF DAILY DICLAZURIL PELLETED TOP DRESS FOR THE PREVENTION OF *SARCOCYSTIS NEURONA* INFECTION IN FOALS {#jvim13002-sec-0070} --------------------------------------------------------------------------------------------------------- **[Nicola Pusterla]{.ul}, Andrea Packham, Sarah Mackie, Philip Kass, Laszlo Hunyadi, Patricia Conrad** SVM, UC Davis, Davis, California, USA Therapeutic treatment strategies for the prevention of *Sarcocystis neurona* infection in horses have been empirical. A pelleted top dress 1.56% diclazuril anti‐protozoal drug recently introduced to the equine market and labelled for the treatment of equine protozoal myeloencephalitis (EPM) has the potential to be used for the prevention of EPM due to its convenient formulation. A low‐dose of diclazuril given at 0.5 mg/kg body weight has shown to reach plasma and cerebrospinal fluid concentrations at steady‐state in excess of the minimal concentration known to be inhibitory to *S. neurona* merozoite production in cell culture. The purpose of this study was to evaluate the temporal serological response against *S. neurona* in foals receiving daily diclazuril and in untreated herdmates. Thirty‐three foals from a farm with a high exposure rate to *S. neurona* were randomly assigned to either an untreated control group or a diclazuril treated group. Treatment consisted of the administration of 0.5 mg/kg body weight of diclazuril pelleted top dress starting at 4 weeks of age until the foals were 12 months of age. Whole blood was collected from every dam and foal 24 hours post‐foaling and monthly thereafter from every study foal for the duration of the study. The blood was tested for IgG against *S. neurona* using the indirect fluorescent antibody test. Following ingestion of colostral antibodies to *S. neurona*, there was a steady and continuous decline in seroprevalence and antibody titers to *S. neurona* until the foals from both groups reached weaning age. Thereafter, untreated foals showed an increase in monthly seroprevalence ranging from 53 to 88%. Diclazuril treated foals showed significant lower monthly seroprevalences ranging from 6 to 29%. At the end of the study, 88 and 6% of untreated and treated foals, respectively, tested seropositive to *S. neurona*. In conclusion, daily supplementation of diclazuril pelleted top dress at 0.5 mg/kg body weight demonstrated a significant reduction in seroconversion to *S. neurona* between diclazuril treated and control foals up to 12 months of age. To the author\'s knowledge this is the first report determining dose and duration of treatment for the prevention of seroconversion in foals originating from a farm with high infection rate to *S. neurona*. From an economic standpoint, supplementation of diclazuril at a low‐dose during early exposure and high‐risk periods may benefit the horse industry by reducing costs associated with EPM. THE EFFECT OF ORAL AND INTRAVENOUS DEXTROSE ON C‐PEPTIDE SECRETION IN PONIES {#jvim13002-sec-0071} ---------------------------------------------------------------------------- **[Melody de Laat]{.ul}, Jessica van Haeften, Martin Sillence** Queensland University of Technology, Brisbane, Queensland, Australia Managing equine hyperinsulinaemia is crucial for preventing laminitis. C‐peptide is secreted from the pancreas proportionally with insulin and can be used to study insulin dysregulation, but no C‐peptide assays have been validated thoroughly for horses. This study aimed to identify a suitable, non‐radioactive immunoassay for equine C‐peptide and compare C‐peptide secretion following oral and IV dextrose administration in ponies. Seven assays were assessed for precision, accuracy and specificity. Only one assay was deemed acceptable and was used to measure the response of nine ponies to oral and IV dextrose (0.75 mg/kg). The ponies were designated as insulin‐resistant (IR) or insulin‐sensitive (IS) based on fasted glucose‐to‐insulin ratios and oral glucose tests. C‐peptide concentrations increased rapidly from fasted levels following both oral (*P* \< 0.01) and IV (*P* \< 0.001) dextrose, with similar AUC for both tests. C‐peptide and insulin concentrations were correlated (*P* \< 0.05), when AUC were compared. The AUCC‐peptide was similar in IR and IS ponies for IV dextrose. However, the AUCC‐peptide for oral dextrose was 7‐fold higher (*P* \< 0.05) in IR, compared to IS ponies. Insulin clearance (fractional difference to C‐peptide) was greater for the oral test, and in IS ponies 60 minutes after dextrose administration, for both tests. Whereas C‐peptide and insulin responses to IV dextrose indicated that pancreatic capacity was similar for both groups, only IR ponies maintained this magnitude of response to oral dextrose. Increased insulin secretion is a major component of hyperinsulinaemia, although reduced clearance also appears to contribute, with increased capacity for insulin response to oral carbohydrate in IR ponies. THE PREVALENCE OF LARGE INTESTINAL MUCOSAL PATHOLOGY IN HORSES BEING EUTHANIZED FOR NON‐GASTROINTESTINAL REASONS {#jvim13002-sec-0072} ---------------------------------------------------------------------------------------------------------------- **[Derek Knottenbelt]{.ul}, Nicola Kerbyson, Timothy Parkin** University of Glasgow, Glasgow, UK The prevalence of colonic ulceration has previously been stated as being 63% in a large study of 545 horses (1) although these lesions were not defined grossly or histopathologically. Detailed gross and histological examination of the gastrointestinal tract of 36 horses euthanized for reasons unrelated to the gastrointestinal tract revealed that 35 (97%) had grossly obvious colonic mucosal pathology; of these 24 cases (67%) were considered to be of likely clinical significance. These included sand enteropathy, active cyathostominosis, right dorsal colitis and focal and diffuse colonic ulceration. Focal congestion of the caecum (15/36), ventral colon (13/36), right dorsal colon (14/36) horses and small colon (1/36) was obvious visually but the clinical significance is equivocal. This study demonstrates that a range of large intestinal mucosal pathology may be present even in the absence of overt clinical signs. Subclinical large colon / caecal disease should be considered in all cases with signs of abdominal disease although the specific diagnosis in most cases would require significant invasive investigation. Rectal biopsy is considered to be poorly correlated with most of the conditions we identified and this concurs with other recent research \[2\]. Minimally invasive diagnostic tests such as fecal occult blood, fecal pH and detailed microbiota panels need to be established to allow effective and accurate ante‐mortem diagnosis and this needs to be correlated closely with clearer definition of the range of clinical diseases identified pathologically in this study. **References** Pellegrini FL. Results of a large‐scale necroscopic study of equine colonic ulcers. J Equine Vet Sci. 2005 Mar;25(3):113--7.Sloet van Oldruitenborgh‐Oosterbaan MM. Grinwis GCM. Variations in eosinophilic infiltration within the rectal mucosa of clinically healthy horses. Proceedings of the 11th Equine Colic Symposium, Dublin, Ireland 2014. MEDICAL ALTERNATIVES TO CONVENTIONAL CYCLOOXYGENASE INHIBITORS FOR TREATMENT OF ACUTE FOOT PAIN IN A REVERSIBLE LAMENESS MODEL IN HORSES {#jvim13002-sec-0073} ---------------------------------------------------------------------------------------------------------------------------------------- **[Jonathan Foreman]{.ul}, Catherine Foreman, Benjamin Bergstom** University of Illinois, Urbana, Illinois, USA In the treatment of acute equine foot pain such as laminitis, conventional treatments include the pan‐cyclooxygenase (COX) inhibitors phenylbutazone (PBZ) and flunixin meglumine (FM). These pan‐COX inhibitors (conventional NSAIDs) are known to cause side effects such as gastrointestinal ulceration and renal disease. Veterinarians treating laminitis often search for efficacious alternatives to conventional NSAIDs. The hypothesis in this series of 4 experiments was that non‐pan‐COX‐inhibitor therapies would be shown to be similar to conventional NSAIDs for the treatment of acute equine foot pain in a reversible model of lameness in horses. 8 horses were shod with an adjustable heart bar shoe on one foot. Horses were monitored by a blinded investigator for 12‐ or 24‐hours post‐medication. Variables included HR and Lameness Score with results compared by RM ANOVA and Tukey\'s test with significance set at *P* \< 0.05. Firocoxib (0.09 mg/kg IV) was no different than saline and both were less effective than PBZ (4.4 mg/kg IV) (*P* \< 0.05). A firocoxib loading dose (0.27 mg/kg IV) was similar in efficacy to PBZ (4.4 mg/kg IV). Acetaminophen (20 mg/kg PO) and oral FM (1.1 mg/kg PO) administered separately were comparable to one another and both were better than saline (*P* \< 0.05). Acetaminophen (20 mg/kg PO BID) combined with firocoxib (0.27 mg/kg SID) was similar to PBZ (4.4 mg/kg IV BID) and both regimes were better than saline (*P* \< 0.05) in a 24‐hours study. Intravenous firocoxib (loading dose 0.27 mg/kg IV SID) and oral acetaminophen showed promise in alleviating acute equine foot pain. INITIAL CHARACTERIZATION OF THE TRACHEAL MICOBIOMES IN HEALTHY HORSES AND HORSES WITH HEAVES {#jvim13002-sec-0074} -------------------------------------------------------------------------------------------- **[Julia Montgomery]{.ul}, Katharina Lohmann, Bonnie Chaban, Lisa Johnson, Scott Dos Santos, Jordan Steedman, Janet E. Hill** Western College of Veterinary Medicine, University of Saskatchewan, Saskatoon, Saskatchewan, Canada Microbial communities colonizing the airways may play a role in the development of chronic inflammatory airway diseases. To investigate whether this may be true for equine heaves, we characterized the tracheal microbiomes of three healthy horses and five horses with historical, clinical and/or bronchoalveolar lavage cytological findings consistent with heaves. On two occasions, approximately 2 weeks apart, tracheal aspirate (TA) samples were collected transendoscopically using a two‐stage sampling catheter. Total genomic DNA was extracted from one aliquot and DNA integrity verified by PCR amplification of the horse mitochondrial *cox*1 gene. Total bacterial load was estimated by quantitative PCR targeting the 16S rRNA gene, and PCR amplicon libraries based on the universal 60 kDa chaperonin (cpn60) gene were prepared using established protocols and sequenced using the 454 GS Junior pyrosequencing platform. Sequences were assembled into operational taxonomic units (OTU) and compared to the cpnDB reference database (www.cpndb.ca) to identify the closest reference sequence (nearest neighbor) for each OTU. Phylum assignment of OTU sequences was based on the taxonomic lineage of the nearest cpnDB reference sequence. Alpha and beta diversity were calculated using QIIME, and hierarchical clustering of microbiome profiles was performed in R. A second aliquot of each sample underwent routine microbiologic analysis using aerobic culture on blood and McConkey agar plates, followed by biochemical testing and identification. Up to 12 colonies per microbiologic culture plate were amplified using cpn60 universal primer PCR, sequenced and compared to the microbiome library from the corresponding horse. A total of 15 TA samples were available for analysis. Of the five horses with heaves, two had active disease at the time of sampling and three were considered to be in remission. Total bacterial load (number of 16S rRNA copies per ml of TA) appeared higher in horses with heaves although considerable variation existed within groups and between samplings. A total of 94,800 sequence reads were analyzed (median 1,488 per sample; range 137--35,926). Consistent with the 16S rRNA copy number, microbiomes of horses with heaves appeared to have higher species richness than those of healthy horses. Microbiome profiles of horses in both clinical categories were dominated by Firmicutes, Actinobacteria and Proteobacteria. OTU sequences corresponded to 329 nearest neighbor reference sequences. OTU with similarity to *Pseudomonas fluorescens*,*Rothia* spp., *Pantoea agglomerans*, and *Streptococcus* spp. were the most prevalent among all samples and were frequently the most abundant in individual microbiomes. Both phylum level profiles and hierarchical clustering of OTU level profiles based on Bray‐Curtis dissimilarity showed that microbiome profiles of individual horses were largely similar between the two samplings, and that microbiomes did not cluster by clinical category. Sequence data were obtained from 51 bacterial colonies, which represented 35 unique sequences. Only 8/35 of these sequences were \>99% identical to OTU in the relevant microbiome libraries. Species identified by both culture and microbiome sequencing included only Firmicutes and represented only 6.7% of the sequence data generated from all horses, consistent with the more comprehensive view of the microbiome provided by the culture‐independent approach. In conclusion, this pilot study suggests that horses with heaves may have a higher bacterial load with greater species richness, but their microbiome profiles are not distinguishable from those of healthy horses at the phylum or OTU level. For most of the horses, tracheal microbiomes appeared stable during the time frame of the study. Microbiome analysis at the genetic level enhanced the assessment of microbial communities compared to conventional culture methods. Work is ongoing to confirm these initial findings in a larger group of horses. FOOD ANIMAL {#jvim13002-sec-0075} =========== CHARACTERIZATION OF THE PATHOGENICITY AND SHEDDING OF BOVINE PARAINFLUENZA 3 GENOTYPES {#jvim13002-sec-0076} -------------------------------------------------------------------------------------- **[Benjamin Newcomer]{.ul}^1^, John Neill^2^, Binu Velayudhan^3^, Paul Walz^1^** ^1^Auburn University, Auburn, Alabama, USA, ^2^National Animal Disease Center, Ames, Iowa, USA, ^3^Texas A&M Veterinary Medical Diagnostic Laboratory, Amarillo, Texas, USA Bovine parainfluenza‐3 virus (PI3) is a widespread respiratory pathogen of cattle capable of causing disease as a unique pathogen but more commonly predisposing cattle to secondary bacterial infection and development of pneumonia. Non‐A genotypes in the United States (US) have only recently been reported but little is known of their pathogenicity or shedding compared to PI3‐A. Therefore, the purpose of this pilot study was to compare the pathogenicity and shedding of clinical US isolates of all three genotypes (A, B, C). Sixteen weaned Jersey bull calves seronegative to PI3 were randomly assigned to one of four groups; calves were challenged intranasally with approximately 10^7^ CCID50 of one of the PI3 genotypes or served as negative controls. The calves were maintained in isolation rooms and daily nasal swab collection and observation for signs of disease by a blinded investigator were performed throughout the 14‐day study. Mild clinical signs were observed similarly in all challenged groups and included serous nasal discharge and coughing. Onset of nasal shedding differed between groups with all PI3‐A challenged calves shedding beginning on Day 4, 3 of 4 PI3‐B challenged calves on Day 5, and PI3‐C challenged calves on Day 6. Shedding of PI3‐C lasted 5 days; shedding of the other viruses extended 7--8 days. This study demonstrates a difference in shedding dynamics between PI3 genotypes. PERCUTANEOUS TUBE CYSTOSTOMY USING RUTNER SUPRAPUBIC CATHETER FOR TREATMENT OF OBSTRUCTIVE UROLITHIASIS IN FIVE GOATS {#jvim13002-sec-0077} --------------------------------------------------------------------------------------------------------------------- **[Katharine Simpson]{.ul}^1^, Robert Streeter^2^, Sarah Depenbrock^1^, Caitlin Jablonski^1^, Joseph Lozier^1^** ^1^The Ohio State University, Columbus, Ohio, USA, ^2^Oklahoma State University, Stillwater, Oklahoma, USA Obstructive urolithiasis in male small ruminants is a common metabolic disease that often requires surgical intervention. Temporary or permanent urinary diversion is indicated if urethral process amputation and/or retrograde urethral hydropulsion are unsuccessful in restoring urethral patency. Surgical cystostomy tube placement is often considered the treatment of choice to allow for urinary diversion and ultimately the reestablishment of urethral patency. However, the costs associated with general anesthesia, surgery and post‐operative care can be substantial. Advantages of percutaneous tube cystostomy rather than surgical tube cystostomy placement include shortened procedure time, cost‐effectiveness, and provision of rapid urinary diversion in systemically compromised patients. Complications of general anesthesia are avoided as the procedure can be performed utilizing local anesthesia with or without sedation. We describe use of a commercially available suprapubic catheter^1^ for percutaneous tube cystostomy in conjunction with urinary acidification as a primary treatment modality in cases of obstructive urolithiasis in goats with suspected or confirmed struvite urolithiasis. Five goats (four Pygmy, one French‐Alpine) were diagnosed with obstructive urolithiasis. Retrograde urethral catheterization was unsuccessful in relieving the obstructions in four of the goats. The other goat was considered to be emergent and a poor anesthetic candidate. A Rutner suprapubic catheter^1^ was percutaneously inserted into the urinary bladder utilizing ultrasound guidance in all five goats to provide urinary diversion. Patency of the urethra was reestablished in 4 of the 5 goats. Mean and median cost associated with percutaneous cystostomy was \$677.86 and \$701.54 respectively, compared to \$1406.05 (mean) and \$1393.14 (median) associated with surgical tube cystostomy during the same year. Complications noted post‐cystostomy included obstruction of the catheter (n = 1), transient uroabdomen (1), and urethral rupture (1). Based on these outcomes, we suggest that percutaneous tube cystostomy using a commercially available Rutner suprapubic catheter^1^ (16 Fr, 4 mL balloon) in conjunction with urinary acidification may be a useful non‐invasive and cost‐effective alternative to surgical interventions for treatment of obstructive urolithiasis; particularly in cases with uroliths amenable to dissolution. Additional research is indicated to fully elucidate post‐procedural complications and long‐term prognosis. ^1^Rutner Suprapubic Balloon Catheter Set, Cook^®^ Medical, 16 French, 4 mL balloon. PHARMACOKINETICS OF AN EXTENDED RELEASE FORMULATION OF EPRINOMECTIN IN HEALTHY ADULT ALPACAS AND ITS USE IN ALPACAS CONFIRMED WITH MANGE {#jvim13002-sec-0078} ---------------------------------------------------------------------------------------------------------------------------------------- **John Pollock^1^, [Daniela Bedenice]{.ul}^1^, Mark G. Papich^2^** ^1^Cummings School of Veterinary Medicine, Tufts University, Grafton, Massachusetts, USA, ^2^North Carolina State University, Raleigh, North California, USA The study purpose was to determine the pharmacokinetics (PK) and clinical effects of extended release 5% eprinomectin (Longrange^®^, Merial) following subcutaneous (SC) injection in healthy (n = 6) and mange‐infected (n = 4) adult alpacas. High‐performance liquid chromatography was performed on plasma samples obtained at regular intervals for 161‐days following a single 5 mg/kg injection SC in healthy alpacas, and for 5‐days following each dose (3 treatments, 2 months apart) in mange affected animals. Clinical monitoring included biweekly hematology and physical examination; monthly weight and fecal egg counts (FEC‐double centrifugation). Skin scrapings and biopsies were performed pre‐ and post‐treatment at two comparable sites in alpacas with mange. Four alpacas served as healthy controls. Pharmacokinetic analysis utilized a non‐compartmental model (WinNonlin‐6.0). Results were compared between time‐points using repeated‐measures ANOVA and paired samples t‐test. Eprinomectin plasma concentrations showed a biphasic peak (Cmax‐1 and 2) in all animals: Cmax‐1 (ng/mL)Tmax‐1 (days)Cmax‐2Tmax‐2AUC infinity (day\*ng/mL)T1/2 terminal (days)Mean (SD)5.72 (3.25)3.88 (5.2)6.06 (2.47)77 (12.52)456.6 (154.5)15.53 (3.16) Eprinomectin plasma concentrations remained above 1.27 ± 0.96 ng/mL for up to 120 days. Hematocrit (35.8 versus 31.3%, *P* \< 0.003) and albumin (3.5 versus 2.8 g/dL, *P* \< 0.006) reduced significantly over 6 months in multi‐dose animals, while FEC did not differ between groups. Self‐limiting injection‐site reactions occurred in 9/10 animals. Pre‐ and post‐treatment skin biopsies showed reduced hyperkeratosis, but increased fibrosis, with 1/4 alpacas remaining positive on skin scraping for mange. In conclusion, alpacas require a higher eprinomectin dose (5 mg/kg SC) than cattle, to reach comparable plasma concentrations. DEVELOPMENT OF A ZINC IMPLANT‐BASED MODEL FOR UROLITHIASIS IN GOATS {#jvim13002-sec-0079} ------------------------------------------------------------------- **[Meredyth Jones]{.ul}, Allen Roussel, W. Shawn Ramsey** Texas A&M University, College Station, Texas, USA The study of urolithiasis in small ruminants is hampered by the lack of a convenient and reliable model that would allow the quantitative assessment of the calculogenic potential of rations. The objective of this study was to develop a nonsurgical experimental model of urolithiasis in goats. Ten, 1‐year old Boer‐cross doelings were fed a calculogenic diet designed to promote struvite urolithiasis. Zinc washers, 6--13 mm in diameter, were used with some washers tied or wrapped with chromic catgut or polyester fiber suture. Goats were anesthetized and the implants were placed into the lumen of the urinary bladder retrograde through the urethra. Different combinations of washer size and suture material and pattern were used in 4 separate experiments. At the end of each experiment, 28--118 days after placement, the implants were retrieved via the urethra. Two implants required removal via cystotomy due to the large amount of crystalline precipitation. Across experiments, 25--67% of implants were spontaneously passed. All implants that remained in the bladders until the end of the experiment had grossly visible calculus accumulation. Discs with suture material accumulated 1.5--46 mg/day compared to 0.03--1.9 mg/day for those without suture material. Urolith analysis revealed that the predominant material accumulated was either struvite or amorphous magnesium calcium phosphate. This study demonstrates that uroliths can be created consistently in a nonsurgical model in goats. This model should be useful in the study of rations and to evaluate interventions that may reduce or prevent the formation of urinary calculi in small ruminants. EFFICACY OF DIFFERENT MULTIVALENT MLV VACCINES ADMINISTERED TO CALVES POSSESSING MATERNALLY‐DERIVED IMMUNITY AND SUBSEQUENTLY CHALLENGED WITH BVDV OR BVDV AND BOHV‐1 {#jvim13002-sec-0080} --------------------------------------------------------------------------------------------------------------------------------------------------------------------- **[Manuel F. Chamorro]{.ul}^1^, Paul H. Walz^1^, Thomas Passler^1^, Kay Ridell^1^, Benjamin Newcomer^1^, Roberto Palomares^2^** ^1^Auburn University, Auburn, Alabama, USA, ^2^University of Georgia, Athens, Georgia, USA The objective of these studies was to evaluate efficacy of different commercially‐available multivalent MLV vaccines to prevent clinical disease, viremia, and virus shedding in early weaned beef calves challenged with BVDV or BVDV and BoHV‐1. In the first study, 48‐male beef calves were early weaned at a median age of 72.2 days and assigned to 1 of 4 treatment groups: control (A), vaccine B, Vaccine C, and Vaccine D. Forty‐five days after vaccination calves were challenged intranasally with virulent BVDV 2. In the second study, 54‐male beef calves were early weaned (median age 93.5 days) and assigned to 1 of 5 treatment groups: control (A), vaccine B, vaccine C, vaccine D, and vaccine E. Forty‐five days after vaccination calves were simultaneously exposed to 6 cattle PI with BVDV and 8 calves acutely infected with BoHV‐1. Samples were collected in all calves for virus isolation and virus neutralization analysis before vaccination and before and after virus challenge. Calves vaccinated with vaccines B and C in the first study and calves vaccinated with vaccines B, C, and E in the second study had greater BVDV‐antibody responses and a lower proportion of viremia and BVDV shedding compared with controls. Vaccination of early weaned beef calves resulted in increased BVDV antibody responses and reduced viremia and BVDV shedding. Differences in vaccine efficacy to prevent BVDV viremia and shedding were observed. COMPARISON OF JOHNE\'S DISEASE PREVALENCE ON ORGANIC AND CONVENTIONAL DAIRY FARMS IN PENNSYLVANIA {#jvim13002-sec-0081} ------------------------------------------------------------------------------------------------- **[Marie‐Eve Fecteau]{.ul}, Helen Aceto, Terry Fyock, Raymond Sweeney** School of Veterinary Medicine, University of Pennsylvania, Kennett Square, Pennsylvania, USA Johne\'s disease (JD) affects approximately 70% of all US dairies, and can be a cause of great economic loss to dairy producers. To qualify for the label "organic", certain restrictions on management practices may predispose for the transmission of JD on the farm. The objectives were: 1‐ to compare JD prevalence between Pennsylvania organic and conventional dairy farms, and 2‐ to identify risk factors associated with differences in JD prevalence. A JD milk ELISA was performed on individual milk samples from each lactating cows in the study herds. Information regarding management practices was collected during a farm visit. Overall herd prevalence, and within‐herd prevalence were compared between groups. Logistic regression was used to identify risk factors associated with differences in JD prevalence between groups. A total of 2,739 cows from 50 herds (25 organic and 25 conventional) were included in the study. Median herd size was 58 (range 20--114 cows) for conventional farms, and 39 (range 20--211 cows) for organic farms. Of all the farms included in the study, 27/50 (54%) were positive, with 15/25 (60%) positive conventional farms and 12/25 (48%) positive organic farms. From the conventional farms, 25/1,506 (2%) cows were positive, compared with 28/1,233 (2%) cows from organic herds. After adjusting for herd size, there was no significant difference in herd prevalence (*P* = 0.56), or within‐herd prevalence (*P* = 0.41) between conventional and organic farms. Risk factors are currently being analyzed. Preliminary results indicate that there is no difference in JD prevalence between Pennsylvania organic and conventional dairy farms. A RANDOMIZED CLINICAL TRIAL EVALUATING METABOLISM OF COLOSTRAL AND PLASMA DERIVED IMMUNOGLOBULIN G IN JERSEY BULL CALVES {#jvim13002-sec-0082} ------------------------------------------------------------------------------------------------------------------------ **Kelly Pipkin, Jill Hagey, Maire Rayburn, [Munashe Chigerwe]{.ul}** UC Davis, Davis, California, USA The objective of this study was to determine the rate of catabolism of colostral derived IgG administered by oroesophageal tubing compared to IV administered plasma IgG. A randomized clinical trial was performed. Thirty newborn Jersey calves were enrolled. Fifteen were fed colostrum (CL group) and 15 were given bovine plasma IV (PL group). Calves in the CL group were fed 3 L of colostrum once, by oroesophageal tubing. Calves in the PL group were given plasma IV at a dosage of 34 mL/kg. Serum and fecal samples were collected at 0 hours, 6 hours, 12 hours, 48 hours, 5 day and 7 day. Serum and fecal IgG concentrations were determined by radial immunodiffusion. Calves in the CL group maintained serum IgG concentrations consistent with adequate transfer of immunity (≥1,000 mg/dL) throughout the study period. Calves in the PL group achieved median IgG concentrations of ≥1,000 mg/dL at 6 hours but the concentrations were \<1,000 mg/dL by 12 hours. Calves in the PL group were 5 times more likely to experience mortality compared to the CL group (hazard ratio = 5.01). Fecal IgG concentrations were not different between the 2 groups during the first 48 hours (*P* \> 0.05). Catabolism of plasma derived IgG occurs rapidly during the first 12 hours after transfusion. Fecal excretion did not explain the fate of the plasma derived IgG. PLASMA LIPIDOMIC PROFILE IN COWS DURING THE TRANSITION PERIOD AND IN COWS WITH FATTY LIVER {#jvim13002-sec-0083} ------------------------------------------------------------------------------------------ **[Christian Gerspach]{.ul}^1^, Hanspeter Naegeli^1^, Maude Gubler^1^, Claudine Bieli^1^, Endre Laczko^2^, Maja Rütten^1^, Sandro Imhasly^1^** ^1^Vetsuisse Faculty, University of Zurich, Zurich, Switzerland, ^2^Functional Genomics Center Zurich, Zurich, Switzerland The transition time around parturition and early lactation involves critical physiologic changes in dairy cows. An excessive demand for nutrients due to the increased performance required for milk production results in a negative energy balance. One major adjustment consists in the rapid mobilization of energy sources from tissue depots in the form of non‐esterified fatty acids. Cows poorly adapting to NEB are at high risk of developing disease. Many transition period diseases, including fatty liver, occur in a subclinical form, affecting milk production and reproductive performance of dairy cows. Fatty liver is interrelated with other production diseases. A noninvasive and accurate test would be helpful to diagnose fatty liver in cattle. The purpose of a large‐scale screening approach was to provide a hypothesis for the development of a novel diagnostic test based on lipidome profiles that is less invasive and more accurate. We compared the plasma lipidome of diseased dairy cows using liquid chromatography coupled to quadrupole time‐of‐flight mass spectrometry. A study on 63 cows revealed 20 masses, which could distinguish between healthy cows and cows with different stages of disease. In a study on 12 clinical healthy cows, during their transition period, 30 masses could be analyzed by MS/MS as potential biomarkers for determination of changes within the lipidome relative to calving. The main lipid groups detected, were triacylglycerides, phosphatidylcholines, and lysophosphatidylcholines. BAYESIAN ESTIMATION OF THE ACCURACY OF CLINICAL EXAMINATION AND SYSTEMATIC THORACIC ULTRASONOGRAPHY FOR THE DIAGNOSIS OF BOVINE RESPIRATORY DISEASE IN PRE‐WEANED DAIRY CALVES {#jvim13002-sec-0084} ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ **[Sébastien Buczinski]{.ul}^1^, Teresa Ollivett^2^, Nandini Dendukuri^3^** ^1^Faculté de médecine vétérinaire, Université de Montréal, St‐Hyacinthe, Quebec, Canada, ^2^Department of Medical Sciences, University of Wisconsin‐Madison, Madison, Wisconsin, USA, ^3^Department of medicine, Department of epidemiology, Biostatistics and Occupational Health, McGill University, Montréal, Quebec, Canada There is no gold standard method for the diagnosis of bovine respiratory disease (BRD) complex in Holstein pre‐weaned dairy calves. Systematic thoracic ultrasonography (TUS) has been used as a proxy for BRD that can be performed in the field by veterinarians, but cannot be directly used by producers. The TUS examination focuses on lung consolidation, a common finding in bronchopneumonia. The Wisconsin calf respiratory scoring chart (CRSC) is a simpler alternative that systematically assesses rectal temperature, cough, ear position, nasal and ocular discharge and attributes a score to each of these items as well as the decision based on the total score obtained (i.e. do nothing, monitor, or treat). However, the accuracy of CRSC is still unknown. Our objective was to estimate the accuracy of CRSC in two different populations, while adjusting for the lack of a gold‐standard to define BRD status. Two cross sectional study populations with a high BRD prevalence (n = 106 pre‐weaned Holstein calves) and an average BRD prevalence (n = 85 pre‐weaned Holstein calves) from North America were studied. All calves were simultaneously assessed using CRSC (cutoff ≥5) and TUS (cutoff ≥1 cm of lung consolidation). When using TUS as a gold standard CRSC Se were 20.0% and 55.4% and Sp were 58.0% and 100.0% respectively. Bayesian latent class models with conditional dependence were used with informative priors for BRD prevalence in both settings and TUS accuracy (sensitivity (SeTUS) and specificity (SpTUS)) and non‐informative priors for CRSC accuracy (SeCRSC/ SpCRSC). The SeCRSC (95% credible interval (CI)) and SpCRSC were 62.4% (47.9--75.8) and 74.1% (64.9--82.8). The SeTUS was 79.4% (66.4--90.9) and SpTUS 93.9% (88.0--97.6). Sensitivity analysis revealed the results were robust to prior specification. Despite their imperfect accuracy both tools are helpful for BRD management. Improvement of the accuracy of BRD detection is a key step to decrease its negative impact as well as overuse of antimicrobials for false positive cases. IMPACT OF MILK FEEDING LEVELS AND HOUSING ON THE INCIDENCE OF RESPIRATORY DISEASE AND SUBSEQUENT PRODUCTIVITY OF YOUNG DAIRY CALVES {#jvim13002-sec-0085} ----------------------------------------------------------------------------------------------------------------------------------- **[Maria Prado]{.ul}, John Wilkerson, Peter Krawczel, Chris Boyer, Arnold Saxton** University of Tennessee, Knoxville, Tennessee, USA Dairy calf pneumonia (DCP) continues to be a highly prevalent condition affecting calves during the pre‐weaning and/or post‐weaning periods. However, its impact extends beyond the actual disease episode, having a negative effect on subsequent productivity and survivability of replacement stock. The ability to identify calves during the initial stages of respiratory disease development by monitoring activity and feeding behavior would allow for earlier disease intervention and thus potentially decrease the effects that DCP has later on the calf\'s life. Therefore, the objective of this study was to determine the impact of different milk feeding levels (standard or high) and housing (individual or group) on respiratory disease development and subsequent productivity of young dairy calves. Female Holstein calves (n = 215) from a commercial dairy farm were used in this study. Calves were randomly assigned to one of three treatment groups: (1) individual hutch with standard milk (4 L/day), (2) individual hutch with high milk (8 L/day), and (3) paired hutches (2 calves) with standard milk in a randomized block design. Social interaction/activity levels as well as feeding behavior were measured by fitting calves with collars containing sensors during the pre‐weaning period. Calves were screened daily for clinical signs of disease. DCP was confirmed by PCR or serology. Body weights were collected from pre‐weaned and weaned calves. Data collected were analyzed to identify behavior changes in social interaction and/or feeding associated with the development of respiratory disease. Serum samples were collected for determining passive transfer at 48 hours after birth. Farm records are being used to evaluate the effects of management during the pre‐weaning phase on the long‐term health and productivity of these calves by a retrospective analysis. The automated sensor system successfully recorded activity levels of calves in the different treatment groups. Total activity levels were correlated with milk consumption showing significant increases during the time preceding the feedings. In addition, we established a unique time series signature from the acceleration data (activity levels) that corresponds with normal specific behaviors such as sleeping, standing, walking and/or bottle feeding. We used these data to identify any deviations from the norm. Overall, respiratory disease levels at this farm were low. All calves developed diarrhea when moved to the hutches. Calves in the high milk group gained an average of 1.28 Lbs/day versus 1.06 Lbs/day in the control group. Average daily gain was comparable between calves in the high milk group and the paired group (1.25 Lbs.). Overall, the calves fed high milk levels gained the most weight and the incidence of respiratory disease was low (0.93%, 2/215 of all enrolled calves). Milk yield, health and reproduction data will be presented.

1. Introduction {#sec1} =============== Transition-metal-catalyzed coupling reactions with aryl halides and nucleophiles have developed significantly and are widely used in fields requiring organic syntheses, such as pharmaceuticals and materials science.^[@ref1]^ For coupling reactions, aryl iodides are the most reactive substrates, followed by aryl bromides. Aryl triflates and mesylates have frequently been employed as pseudoaryl halides in coupling reactions because they are easily prepared from phenol derivatives, which are readily accessible and inexpensive compared to aryl halides.^[@ref2]^ However, they have some drawbacks, that is, the former require the use of expensive triflating agents and the latter show low activity. To address these issues, tosylates, phosphates, and carboxylates have been used as alternative pseudohalides.^[@ref3]^ Among them, aryl tosylates have several advantages: they are less expensive, readily prepared from phenol derivatives, and easy to handle because of their high hydrolytic stability and crystallinity.^[@ref4]^ To synthesize aryl alkynes, one of the most powerful tools is the Sonogashira reaction, which is a modified Heck's discovery and a palladium-catalyzed coupling reaction of terminal alkynes and aryl halides or pseudohalides in the presence of copper salts.^[@ref5]^ Since Sonogashira first reported this reaction in 1975, many studies have been carried out and remarkable improvements to the reactivity and scope of the reaction have been made. Beneficial from an environmental and economic point of view, copper-free, amine-free, high-turnover, recyclable catalysts that can operate under aqueous conditions have also been developed.^[@ref6]^ However, in most cases, the coupling partners for the terminal alkyne are aryl halides, such as iodides, bromides, and chlorides. The Sonogashira coupling of aryl tosylates has been much less explored compared to those of aryl halides. The first example of Sonogashira reaction with aryl tosylates using the XPhos ligand was reported by Buchwald's group.^[@ref7]^ This catalyst system has been used by other groups, but it requires slow addition of alkyne to obtain a high product yield. Consequently, 1-dicyclohexylphosphino-2-(di-*tert*--butylphosphino-ethyl)ferrocene (CyPF-*t*Bu) has been used as the ligand in the palladium-catalyzed Sonogashira coupling of aryl tosylates,^[@ref8]^ and *N*-heterocyclic carbine-Pd(II) complexes have been reported by Lu and Shen^[@ref9]^ in the coupling of aryl tosylates. An alternative Sonogashira-type coupling, the decarboxylative coupling of propiolic acid and aryl halides, was first reported by our group.^[@ref10]^ Compared to aryl terminal alkynes as the alkyne source, the use of arylpropiolic acids has several advantages. They are readily prepared from aryl halides via a single step without a column chromatography step.^[@ref11]^ Therefore, the range of alkyne source compounds is broad. The decarboxylative coupling reaction with aryl halides, such as iodides, bromides, and chlorides, was developed by our and other groups.^[@ref12]^ Coupling with aryl triflates was reported by Lee ([Scheme [1](#sch1){ref-type="scheme"}](#sch1){ref-type="scheme"}a).^[@ref13]^ Recently, we developed oxidative decarboxylative coupling reactions with arylboronic acids or aryl siloxanes using palladium or nickel catalysts ([Scheme [1](#sch1){ref-type="scheme"}](#sch1){ref-type="scheme"}b).^[@ref14]^ Although considerable related results have been achieved since our first report concerning decarboxylative coupling reactions, this type of reaction with aryl tosylates has not been previously reported. Very recently, Satoh's group reported the Sonogashira-type coupling of 2,2-difluorophenyl tosylate with phenylpropiolic acid to obtain difluorinated enynes.^[@ref15]^ However, their method is limited to 2,2-difluoroethynyl tosylate and the coupling with aryl tosylates was not successful. Moreover, aryl tosylates are much less reactive than vinyl tosylates. Herein, we report a palladium-catalyzed decarboxylative coupling reaction with alkynyl carboxylic acids and aryl tosylates. {#sch1} 2. Results and Discussion {#sec2} ========================= To optimize the reaction conditions, we screened the ligands first because the ligand is the most important factor in coupling reactions with aryl tosylates, which show low reactivity toward palladium-catalyzed coupling reactions. Phenyl tosylate and phenylpropiolic acid were chosen as the model substrates. Many biphenyl phosphine derivatives, which are known as Buchwald-type ligands, have been used. Unfortunately, biphenyl phosphines, which are commercially available, did not give the desired coupled products ([Table [1](#tbl1){ref-type="other"}](#tbl1){ref-type="other"}, entries 1--4). Among the bisphosphine ligands we tested, only Josiphos ligands, such as CyPF-*t*Bu, produced diphenylacetylene (**3aa**) in 45% yield (entry 8). The use of palladium sources, such as Pd(acac)~2~, Pd(PPh~3~)~2~Cl~2~, and Pd~2~(dba)~3~, resulted in moderate product yields (entries 9--11). With Pd(PPh~3~)~2~Cl~2~ and CyPF-*t*-Bu in hand, bases were tested. Organic bases such as 1,8-diazabicyclo undec-7-ene (DBU) and 1,5-diazabicyclo(4.3.0)non-5-ene (DBN) resulted in 32 and 31% yields, respectively (entries 12 and 13). The use of Cs~2~CO~3~ resulted in higher yields than other inorganic bases, such as K~3~PO~4~ and Na~2~CO~3~ (entries 14--16). Reactions in *N*-methyl pyrrolidone (NMP) or dimethyl sulfoxide (DMSO) did not give satisfactory results (entries 17 and 18). The reaction in dimethylformamide (DMF) provided the highest yield (entry 19). When the amount of both palladium and ligand decreased to 2.5 and 1.0 mol %, respectively, the corresponding product yields decreased to 76 and 54% (entries 20 and 21). However, the reaction with 1.5 mol % Pd and 5.0 mol % CyPF-*t*-Bu afforded the desired product in 94% yield (entry 22). In addition, we found that the reaction was completed in 6 h (entry 23). ###### Optimization of Decarboxylative Coupling with Phenylpropiolic Acid and Phenyl Tosylate[a](#t1fn1){ref-type="table-fn"} {#fx1} entry Pd ligand base solvent yield (%)[f](#t1fn6){ref-type="table-fn"} ------------------------------------ --------------------- ------------- ------------ --------- ------------------------------------------- 1 Pd(CH~3~CN)~2~Cl~2~ JohnPhos K~2~CO~3~ DMAc \<1 2 Pd(CH~3~CN)~2~Cl~2~ RuPhos K~2~CO~3~ DMAc \<1 3 Pd(CH~3~CN)~2~Cl~2~ *t*-BuXPhos K~2~CO~3~ DMAc \<1 4 Pd(CH~3~CN)~2~Cl~2~ CyBrettPhos K~2~CO~3~ DMAc \<1 5 Pd(CH~3~CN)~2~Cl~2~ Xantphos K~2~CO~3~ DMAc \<1 6 Pd(CH~3~CN)~2~Cl~2~ dppb K~2~CO~3~ DMAc \<1 7 Pd(CH~3~CN)~2~Cl~2~ dppf K~2~CO~3~ DMAc \<1 8 Pd(CH~3~CN)~2~Cl~2~ CyPF-*t*Bu K~2~CO~3~ DMAc 45 9 Pd(acac)~2~ CyPF-*t*Bu K~2~CO~3~ DMAc 40 10 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu K~2~CO~3~ DMAc 55 11 Pd~2~(dba)~3~ CyPF-*t*Bu K~2~CO~3~ DMAc 39 12 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu DBU DMAc 32 13 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu DBN DMAc 31 14 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu K~3~PO~4~ DMAc 20 15 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Na~2~CO~3~ DMAc 6 16 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMAc 86 17 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ NMP 5 18 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMSO 67 19 Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMF 93 20[b](#t1fn2){ref-type="table-fn"} Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMF 76 21[c](#t1fn3){ref-type="table-fn"} Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMF 54 22[d](#t1fn4){ref-type="table-fn"} Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMF 94 23[e](#t1fn5){ref-type="table-fn"} Pd(PPh~3~)~2~Cl~2~ CyPF-*t*Bu Cs~2~CO~3~ DMF 94 Reaction conditions: **1a** (0.3 mmol), **2a** (0.3 mmol), Pd (0.015 mmol), ligand (0.015 mmol), and base (0.36 mmol) reacted in solvent (1.0 mL). Pd (2.5 mol %) and CyPF-*t*Bu (2.5 mol %) were used. Pd (1.0 mol %) and CyPF-*t*Bu (2.5 mol %) were used. Pd (1.5 mol %) and CyPF-*t*Bu (5.0 mol %) were used. Reaction time was 6 h. Yield was determined by gas chromatography (GC) with an internal standard. Ligand structure:. The optimized conditions are arylpropiolic acid (1.0 equiv), aryl tosylate (1.0 equiv), Pd(PPh~3~)~2~Cl~2~ (1.5 mol %), CyPF-*t*Bu (5 mol %), and Cs~2~CO~3~ (1.2 equiv) in DMF at 130 °C for 6 h. Under these conditions, we first evaluated a variety of substituted arylpropiolic acids. The results are summarized in [Table [2](#tbl2){ref-type="other"}](#tbl2){ref-type="other"}. Monomethyl- and dimethyl-substituted phenylpropiolic acids, such as **1b**, **1c**, **1d**, **1e**, and **1f**, provided the corresponding products in excellent yields (entries 1--5). *Ortho*-, *meta*-, and *para*-methoxy-substituted phenylpropiolic acids afforded **3ga**, **3ha**, and **3ia** in 94, 93, and 91% yields, respectively (entries 6--8). 3-(Benzo\[*d*\]\[1,3\]dioxaol-5-yl)propiolic acid produced **3ja** in 88% yield (entry 9). 4′-Biphenylpropiolic acid gave rise to **3ka** in 41% yield (entry 10). Fluoromethyl- and trifluoromethyl-substituted phenylpropiolic acids, such as **1l** and **1m**, gave the desired products **3la** and **3ma** in 96 and 40% yields, respectively (entries 11 and 12). 3-(Thiophen-2-yl)propiolic acid also showed a good yield (entry 13). Arylpropiolic acids bearing ketone and ester groups gave the desired products **3oa** and **3pa** in low yields (entries 14 and 15). However, arylpropiolic acids having aldehyde and cyano groups did not provide the desired products (entries 16 and 17). In addition, we found that **1q** and **1r** were thermally decomposed under these conditions. Alkyl-substituted propiolic acids, such as octynoic and hexynoic acids, gave the coupled products **3sa** and **3ta** in 77 and 85% yields, respectively (entries 18 and 19). ###### Optimization of Decarboxylative Coupling with Phenylpropiolic Acid and Phenyl Tosylate[a](#t2fn1){ref-type="table-fn"} {#fx3} Reaction conditions: **1** (2.0 mmol), **2a** (2.0 mmol), Pd(PPh~3~)~2~Cl~2~ (0.03 mmol), CyPF-*t*Bu (0.1 mmol), and Cs~2~CO~3~ (2.4 mmol) were reacted in DMF (5.0 mL) at 130 °C for 6 h. Isolated yield. Next, numerous aryl tosylates were tested for the reaction with phenylpropiolic acid under the optimized conditions. The aryl tosylates shown in [Table [3](#tbl3){ref-type="other"}](#tbl3){ref-type="other"} were prepared from the reaction of the corresponding alcohols and tosyl chloride. *Ortho*-, *meta*-, and *para*-tolyl tosylates coupled with phenylpropiolic acid afforded the corresponding products **3ab**, **3ac**, and **3ad** in 84, 97, and 96% yields, respectively (entry 1--3). 4-Methoxyphenyl tosylate resulted in a 74% yield in the formation of **3ae** (entry 4). The 1- and 2-naphthyl-substituted tosylates provided the desired products **3af** and **3ag** in 87 and 94% yields, respectively (entries 5 and 6). Heteroaryl tosylates such as **2h** and **2i** also provided the coupled products **3ah** and **3ai** in 69 and 81% yields, respectively (entries 7 and 8). The use of 3,5-difluorophenyl tosylate **2j** resulted in an 80% yield of product **3aj** (entry 9). ###### Decarboxylative Coupling with Phenylpropiolic Acid and Aryl Tosylates[a](#t3fn1){ref-type="table-fn"} {#fx4} Reaction conditions: **1a** (2.0 mmol), **2** (2.0 mmol), Pd(PPh~3~)~2~Cl~2~ (0.03 mmol), CyPF-*t*Bu (0.1 mmol), and Cs~2~CO~3~ (2.4 mmol) were reacted in DMF (5.0 mL) at 130 °C for 6 h. Isolated yield. The range of substrates was further surveyed, and the substituted aryl tosylates and arylpropiolic acids were employed in the decarboxylative coupling reaction. As shown in [Table [4](#tbl4){ref-type="other"}](#tbl4){ref-type="other"}, unsymmetrical diaryl alkynes were formed in moderate to good yields in most cases (entries 1--7). When *meta*-methyl-substituted phenylpropiolic acid **1c** was used, **3ch** was formed in low yield (entry 8). The coupling reaction with arylpropiolic acids bearing ketone groups resulted in the poor yield of the product (entry 9). ###### Decarboxylative Coupling with Phenylpropiolic Acid and Aryl Tosylates[a](#t4fn1){ref-type="table-fn"} {#fx5} Reaction conditions: **1** (2.0 mmol), **2** (2.0 mmol), Pd(PPh~3~)~2~Cl~2~ (0.03 mmol), CyPF-*t*Bu (0.1 mmol), and Cs~2~CO~3~ (2.4 mmol) were reacted in DMF (5.0 mL) at 130 °C for 6 h. Isolated yield. From this range of substrates, we found that the electronic properties of the substituents on the aryl rings in the arylpropiolic acids were very important in this decarboxylative coupling reaction; however, the steric properties of the substituents on both arylpropiolic acids and aryl tosylates did not affect this reaction significantly. We found that the use of arylpropiolic acids bearing electron-withdrawing groups resulted in low yields or no product, and both *ortho*-tolylpropiolic acid and *ortho-*tolyl tosylate, which are sterically demanding, showed moderate yields in this transformation. To investigate the electronic effects, we carried out competitive experiments. When equal amounts of **1d**, **1l**, and **1p** were reacted with phenyl tosylate (**2a**) in a vessel under the optimized conditions, the desired products **3da**, **3la**, and **3pa** were formed in 42, 41, and 5% yields, respectively. This result suggests that methyl- and halo-substituted arylpropiolic acids showed almost same reactivity in the decarboxylative coupling reaction with phenyl tosylate, whereas electron-withdrawing groups, such as acetyl-substituted arylpropiolic acid, showed low reactivity ([Scheme [2](#sch2){ref-type="scheme"}](#sch2){ref-type="scheme"}). {#sch2} To further evaluate this catalytic system, competitive experiments between phenyl tosylate and aryl halides were conducted, as shown in [Table [5](#tbl5){ref-type="other"}](#tbl5){ref-type="other"}. We chose aryl iodides, bromides, and chlorides bearing methyl, fluoro, and acetyl groups at the para position. When 4-iodo-, 4-bromo-, and 4-chlorotoluenes were competitively reacted with phenyl tosylate, 4-iodo- and 4-bromotoluenes provided the coupled product, **5a**, in 62 and 72% yields, respectively (entries 1 and 2). However, 4-chlorotoluene showed lower reactivity than phenyl tosylate (entry 4). In the case of the reaction with 4-fluorophenylhalides, they showed similar reactivity, whereas aryl iodides and bromides provided higher reactivity than phenyl tosylate (entries 4 and 5). In contrast, the aryl chlorides showed much less reactivity than others (entry 6). In the competitive reaction with 4′-haloacetophenones and phenyl tosylate, all 4′-arylacetophenones showed higher activity than the phenyl tosylate (entries 7--9). From these results, we concluded that aryl iodides and bromides show higher activity than phenyl tosylate, but aryl chlorides are less active than phenyl tosylate, except aryl chlorides bearing an electron-withdrawing group. ###### Decarboxylative Coupling with Phenylpropiolic Acid and Aryl Tosylates[a](#t5fn1){ref-type="table-fn"} {#fx6}   aryl halides product yield (%)[b](#t5fn2){ref-type="table-fn"} --- -------------- --------------------------------------------------- ---- ---- ---- 1 **4a-I** CH~3~ I 3 62 2 **4a-Br** CH~3~ Br 2 75 3 **4a-Cl** CH~3~ Cl 74 16   aryl halides product yield (%)[b](#t5fn2){ref-type="table-fn"} --- -------------- --------------------------------------------------- ---- ---- ---- 4 **4b-I** F I 0 75 5 **4b-Br** F Br 0 68 6 **4b-Cl** F Cl 58 21   aryl halides product yield (%)[b](#t5fn2){ref-type="table-fn"} --- -------------- --------------------------------------------------- ---- ---- ---- 7 **4c-I** COCH~3~ I 0 75 8 **4c-Br** COCH~3~ Br 0 64 9 **4c-Cl** COCH~3~ Cl 19 48 Reaction conditions: **1a** (0.3 mmol), **2a** (0.3 mmol), **4** (0.3 mmol), Pd(PPh~3~)~2~Cl~2~ (0.0045 mmol), CyPF-*t*Bu (0.015 mmol), and Cs~2~CO~3~ (0.36 mmol) were reacted in DMF (1.5 mL) at 130 °C for 6 h. Yield was determined by GC with internal standard. When phenylacetylene (**1a′**) and 4-tolylpropiolic acid (**1d**) were reacted with phenyl tosylate under the optimized conditions, both coupled products **3aa** and **3da** were formed in 55 and 45% yields, respectively. This result suggests that the Sonogashira coupling and decarboxylative coupling showed similar reactivity toward phenyl tosylates under the optimized conditions. We paid attention to the greater stability of aryl tosylates, which can remain in the palladium-catalyzed decarboxylative coupling reaction ([Scheme [3](#sch3){ref-type="scheme"}](#sch3){ref-type="scheme"}). {#sch3} On the basis of our results and previous reports of coupling reaction with aryl tosylates, we propose the reaction mechanism shown in [Scheme [4](#sch4){ref-type="scheme"}](#sch4){ref-type="scheme"}. Aryl tosylates react with palladium(0) species to give the aryl palladium(II) complex **A** through oxidative addition. Arylpropiolic acid reacts with the aryl palladium complex **A** in the presence of a base to afford aryl alkynyl palladium (II) complex **B** through decarboxylation and ligand-exchange steps and finally the desired coupled product is formed through reductive elimination. {#sch4} 3. Conclusions {#sec3} ============== In summary, we developed a decarboxylative coupling reaction with alkynyl carboxylic acids and aryl tosylates. The optimized conditions are Pd(PPh~3~)~2~Cl~2~ (1.5 mol %), CyPF-*t*Bu (5.0 mol %), and Cs~2~CO~3~ in DMF solvent. Various arylpropiolic acids bearing alkyl, alkoxy, fluoro, ester, and ketone groups coupled with aryl tosylates afforded the desired diaryl alkynes in good yields. In addition, alkyl-substituted propiolic acids, such as octynoic and hexynoic acids, also gave the decarboxylative coupled products in good yields. We found that the electronic properties of the phenyl ring substituents in arylpropiolic acids are an important factor in the reactivity in this decarboxylative coupling reaction. Thus, arylpropiolic acid having electron-donating groups showed higher yields of products than those having electron-withdrawing groups. However, fluoride substituents showed similar activity to alkyl substituents. Phenyl tosylates were much less active in this decarboxylative coupling reaction than both aryl iodides and bromides. However, they provided higher product yields than aryl chlorides, except in the case of aryl chlorides bearing acetyl groups. In addition, we found that arylpropiolic acids and terminal alkynes exhibited almost same reactivity toward the coupling with phenyl tosylate. 4. Experimental Section {#sec4} ======================= 4.1. General Procedure for Decarboxylative Coupling Reactions of Alkynyl Carboxylic Acids and Aryl Tosylates {#sec4.1} ------------------------------------------------------------------------------------------------------------ Alkynyl carboxylic acid (2.0 mmol), CyPF-*t*Bu (55 mg, 0.1 mmol), bis(triphenylphosphine)palladium(II) dichloride (21 mg, 0.03 mmol), Cs~2~CO~3~ (782 mg, 2.4 mmol), aryl tosylate (2.0 mmol), and DMF (5 mL) were added to the reaction vial. The mixture was stirred at 130 °C for 6 h. Then, the mixture was extracted with Et~2~O. The organic layer was dried over magnesium sulfate. Evaporation of the solvent under reduced pressure provided the crude product, which was purified by column chromatography on silica gel. Note: It is noteworthy that the combination of DMF and Cs~2~CO~3~ at high temperature has occasionally resulted in rapid CO release. Therefore, when the reaction was carried out on gram scale, special equipment is required for safer processing with pressure control.^[@ref16]^ ### 4.1.1. 1,2-Diphenylethyne (**3aa**)^[@ref16]^ {#sec4.1.1} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1,2-diphenylethyne (**3aa**) (331 mg, 1.86 mmol, 93% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.54 (m, 4H), 7.36--7.33 (m, 6H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 131.6, 128.3, 128.2, 123.3, 89.3; MS (EI) *m*/*z*: 178 (M^+^). ### 4.1.2. 1-Methyl-2-(phenylethynyl)benzene (**3ba**)^[@ref10]^ {#sec4.1.2} 3-(*o*-Tolyl)propiolic acid (**1b**) (320 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methyl-2-(phenylethynyl)benzene (**3ba**) (377 mg, 1.96 mmol, 98% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.54 (m, 2H), 7.50 (d, *J* = 7.5 Hz, 1H), 7.37--7.33 (m, 3H), 7.23 (dd, *J* = 7.2 Hz, 2H), 7.17 (m, 1H), 2.51 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 140.1, 131.8, 131.5, 129.4, 128.3, 128.3, 128.2,125.6, 123.5, 123.0, 93.3, 88.3, 20.8; MS (EI) *m*/*z*: 192 (M^+^). ### 4.1.3. 1-Methyl-3-(phenylethynyl)benzene (**3ca**)^[@ref16]^ {#sec4.1.3} 3-(*m*-Tolyl)propiolic acid (**1c**) (320 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methyl-3-(phenylethynyl)benzene (**3ca**) (361 mg, 1.88 mmol, 94% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.52 (m, 2H), 7.36--7.22 (m, 5H), 7.23 (t, *J* = 7.6 Hz, 1H), 7.14 (d, *J* = 7.6 Hz, 1H), 2.35 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 137.9, 132.2, 131.6, 129.1, 128.7, 128.3, 128.22, 128.15, 123.3, 123.0, 89.5, 89.0, 21.2; MS (EI) *m*/*z*: 192 (M^+^). ### 4.1.4. 1-Methyl-4-(phenylethynyl)benzene (**3da**)^[@ref17]^ {#sec4.1.4} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methyl-4-(phenylethynyl)benzene (**3da**) (361 mg, 1.88 mmol, 94% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.52 (m, 2H), 7.43 (d, *J* = 8.1 Hz, 2H), 7.34--7.32 (m, 3H), 7.16 (m, 2H), 2.37 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 138.4, 131.53, 131.48, 129.1, 128.3, 128.1, 123.4, 120.2, 89.5, 88.7, 21.5; MS (EI) *m*/*z*: 192 (M^+^). ### 4.1.5. 1,2-Dimethyl-4-(phenylethynyl)benzene (**3ea**)^[@ref17]^ {#sec4.1.5} 3-(3,4-Dimethylphenyl)propiolic acid (**1e**) (348 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1,2-dimethyl-4-(phenylethynyl)benzene (**3ea**) (380 mg, 1.84 mmol, 92% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.51 (m, 2H), 7.35--7.27 (m, 4H), 7.27 (m, 1H), 7.10 (d, *J* = 7.7 Hz, 1H), 2.27 (s, 3H), 2.26 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 137.2, 136.6, 132.6, 131.5, 129.6, 129.0, 128.3, 128.0, 123.5, 120.4, 89.7, 88.5, 19.8, 19.6; MS (EI) *m*/*z*: 206 (M^+^). ### 4.1.6. 2,4-Dimethyl-1-(phenylethynyl)benzene (**3fa**)^[@ref17]^ {#sec4.1.6} 3-(2,4-Dimethylphenyl)propiolic acid (**1f**) (348 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 2,4-dimethyl-1-(phenylethynyl)benzene (**3fa**) (396 mg, 1.92 mmol, 96% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.52 (m, 2H), 7.38 (d, *J* = 7.8 Hz, 1H), 7.38--7.30 (m, 3H), 7.04 (s, 1H), 6.97 (d, *J* = 7.8 Hz, 1H), 2.47 (s, 3H), 2.31 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 140.0, 138.4, 131.7, 131.4, 130.3, 128.3, 128.0, 126.4, 123.7, 119.9, 92.6, 88.5, 21.4, 20.6; MS (EI) *m*/*z*: 206 (M^+^). ### 4.1.7. 1-Methoxy-2-(phenylethynyl)benzene (**3ga**)^[@ref17]^ {#sec4.1.7} 3-(2-Methoxyphenyl)propiolic acid (**1g**) (352 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methoxy-2-(phenylethynyl)benzene (**3ga**) (392 mg, 1.88 mmol, 94% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.56 (m, 2H), 7.50 (dd, *J* = 7.6, 1.7 Hz, 1H), 7.34--7.27 (m, 4H), 6.96 (td, *J* = 7.5, 1.0 Hz, 1H), 6.88 (d, *J* = 8.4 Hz, 1H), 3.89 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 159.9, 133.6, 131.7, 129.8, 128.3, 128.1, 123.6, 120.5, 112.4, 110.7, 93.5, 85.8, 55.9; MS (EI) *m*/*z*: 208 (M^+^). ### 4.1.8. 1-Methoxy-3-(phenylethynyl)benzene (**3ha**)^[@ref17]^ {#sec4.1.8} 3-(3-Methoxyphenyl)propiolic acid (**1h**) (352 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methoxy-3-(phenylethynyl)benzene (**3ha**) (387 mg, 1.86 mmol, 93% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.53 (m, 2H), 7.35--7.33 (m, 3H), 7.25 (t, *J* = 7.9 Hz, 1H), 7.13 (m, 1H), 7.06 (dd, *J* = 2.6, 1.4 Hz, 1H), 6.89 (ddd, *J* = 8.3, 2.6, 1.0 Hz, 1H), 3.81 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 159.3, 131.6, 129.4, 128.4, 128.3, 124.3, 124.2, 123.2, 116.3, 115.0, 89.3, 89.2, 55.3; MS (EI) *m*/*z*: 208 (M^+^). ### 4.1.9. 1-Methoxy-4-(phenylethynyl)benzene (**3ia**)^[@ref17]^ {#sec4.1.9} 3-(4-Methoxyphenyl)propiolic acid (**1i**) (352 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-methoxy-4-(phenylethynyl)benzene (**3ia**) (379 mg, 1.82 mmol, 91% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.51 (m, 2H), 7.45 (m, 2H),7.35--7.32 (m, 3H), 6.88 (dd, *J* = 8.9 Hz, 2H), 3.83 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 159.6, 133.0, 131.4, 128.3, 127.9, 123.6, 115.3, 114.0, 89.3, 88.0, 55.3; MS (EI) *m*/*z*: 208 (M^+^). ### 4.1.10. 5-(Phenylethynyl)benzo\[*d*\]\[1,3\]dioxole (**3ja**)^[@ref17]^ {#sec4.1.10} 3-(Benzo\[*d*\]\[1,3\]dioxol-5-yl)propiolic acid (**1j**) (380 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 5-(phenylethynyl)benzo\[*d*\]\[1,3\]dioxole (**3ja**) (391 mg, 1.76 mmol, 88% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.50 (m, 2H), 7.34--7.32 (m, 3H), 7.06 (dd, *J* = 8.0, 1.6 Hz, 1H), 6.98 (d, *J* = 1.6 Hz, 1H), 6.79 (d, *J* = 8.0 Hz, 1H), 5.99 (s, 2H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 147.8, 147.4, 131.5, 128.3, 128.1, 126.3, 123.3, 116.5, 111.5, 108.5, 101.3, 89.3, 87.8; MS (EI) *m*/*z*: 222 (M^+^). ### 4.1.11. 4-(Phenylethynyl)-1,1′-biphenyl (**3ka**)^[@ref17]^ {#sec4.1.11} 3-(\[1,1′-Biphenyl\]-4-yl)propiolic acid (**1k**) (444 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 4-(phenylethynyl)-1,1′-biphenyl (**3ka**) (224 mg, 0.88 mmol, 41% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.62--7.5 (m, 8H), 7.45 (t, 2H), 7.37--7.34 (m, 4H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 141.0, 140.3, 132.0, 131.6, 128.9, 128.4, 128.3, 127.6, 127.0 (2C), 123.3, 122.2, 90.0, 89.3; MS (EI) *m*/*z*: 254 (M^+^). ### 4.1.12. 1-Fluoro-4-(phenylethynyl)benzene (**3la**)^[@ref17]^ {#sec4.1.12} 3-(4-Fluorophenyl)propiolic acid (**1l**) (328 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-fluoro-4-(phenylethynyl)benzene (**3la**) (377 mg, 1.92 mmol, 96% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.53-7.50 (m, 4H), 7.35--7.34 (m, 3H), 7.04 (m, 2H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 162.5 (d, *J* = 250.0 Hz), 133.5 (d, *J*~C-F~ = 8.4 Hz), 131.5, 128.4, 128.3, 123.1, 119.3 (d, *J*~C-F~ = 3.5 Hz), 115.6 (d, *J*~C-F~ = 22.1 Hz), 89.0, 88.3; MS (EI) *m*/*z*: 196 (M^+^). ### 4.1.13. 1-(Phenylethynyl)-4-(trifluoromethyl)benzene (**3ma**)^17^ {#sec4.1.13} 3-(4-(Trifluoromethyl)phenyl)propiolic acid (**1m**) (428 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-(phenylethynyl)-4-(trifluoromethyl)benzene (**3ma**) (197 mg, 0.80 mmol, 40% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.64--7.60 (m, 4H), 7.54 (m, 2H), 7.37 (m, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 132.5, 131.8, 131.7, 129.9 (q, *J*~C-F~ = 32.6 Hz), 128.8, 128.4, 126.0 (q, *J*~C-F~ = 263.7 Hz), 128.3 (q, *J*~C-F~ = 3.8 Hz), 122.6, 91.7, 88.0; MS (EI) *m*/*z*: 246 (M^+^). ### 4.1.14. 2-(Phenylethynyl)thiophene (**3na**)^[@ref17]^ {#sec4.1.14} 3-(Thiophen-2-yl)propiolic acid (**1n**) (304 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 2-(phenylethynyl)thiophene (**3na**) (240 mg, 1.30 mmol, 65% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.51 (m, 2H), 7.35--7.34 (m, 3H), 7.29 (m, 2H), 7.01 (dd, *J* = 5.1, 3.7 Hz, 1H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 131.9, 131.4, 128.41, 128.36, 127.2, 127.1, 123.3, 122.9, 93.0, 82.6; MS (EI) *m*/*z*: 184 (M^+^). ### 4.1.15. Methyl 4-(phenylethynyl)benzoate (**3oa**)^[@ref17]^ {#sec4.1.15} 3-(4-(Methoxycarbonyl)phenyl)propiolic acid (**1o**) (408 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded methyl 4-(phenylethynyl)benzoate (**3oa**) (151 mg, 0.64 mmol, 32% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.02 (d, *J* = 8.7 Hz, 2H), 7.59 (d, *J* = 8.7 Hz, 2H), 7.55 (m, 2H), 7.37--7.36 (m, 3H), 3.93 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 166.6, 131.7, 131.5, 129.51, 129.46, 128.8, 128.4, 128.0, 122.7, 92.3, 88.6, 52.2; MS (EI) *m*/*z*: 236 (M^+^). ### 4.1.16. 1-(4-(Phenylethynyl)phenyl)ethanone (**3pa**)^[@ref17]^ {#sec4.1.16} 3-(4-Acetylphenyl)propiolic acid (**1p**) (376 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded 1-(4-(phenylethynyl)phenyl)ethanone (**3pa**) (181 mg, 0.82 mmol, 41% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.94 (d, *J* = 8.7 Hz, 2H), 7.61 (d, *J* = 8.7 Hz, 2H), 7.55 (m, 2H), 7.38--7.36 (m, 3H), 2.62 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 197.3, 136.2, 131.74, 131.70, 128.8, 128.4, 128.3, 128.2, 122.6, 92.7, 88.6, 26.6; MS (EI) *m*/*z*: 220 (M^+^). ### 4.1.17. Hept-1-yn-1-ylbenzene (**3sa**)^[@ref18]^ {#sec4.1.17} Oct-2-ynoic acid (**1s**) (280 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded Hept-1-yn-1-ylbenzene (**3sa**) (265 mg, 1.54 mmol, 77% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.39 (m, 2H), 7.28--7.24 (m, 3H), 2.39 (t, *J* = 7.2 Hz, 2H), 1.61 (m, 2H), 1.43 (m, 2H), 1.35 (m, 2H), 0.92 (t, *J* = 7.4 Hz, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 131.5, 128.1, 127.4, 124.1, 90.4, 80.5, 31.1, 28.5, 22.2, 19.3, 14.0; MS (EI) *m*/*z*: 172 (M^+^). ### 4.1.18. Pent-1-yn-1-ylbenzene (**3ta**)^[@ref19]^ {#sec4.1.18} Hex-2-ynoic acid (**1s**) (224 mg, 2.0 mmol) and phenyl tosylate (**2a**) (496 mg, 2.0 mmol) afforded pent-1-yn-1-ylbenzene (**3ta**) (245 mg, 1.70 mmol, 85% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.39 (m, 2H), 7.27--7.25 (m, 3H), 2.38 (t, *J* = 7.0 Hz, 2H), 1.63 (sextet, 2H), 1.05 (t, *J* = 7.4 Hz, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 131.5, 128.2, 127.5, 124.1, 90.2, 80.7, 22.2, 21.4, 13.6; MS (EI) *m*/*z*: 144 (M^+^). ### 4.1.19. 1-Methyl-2-(phenylethynyl)benzene (**3ab**)^[@ref10]^ {#sec4.1.19} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and *o*-tolyl tosylate (**2b**) (525 mg, 2.0 mmol) afforded 1-methyl-2-(phenylethynyl)benzene (**3ab**) (323 mg, 1.68 mmol, 84% yield). This compound is the same as **3ba**. ### 4.1.20. 1-Methyl-3-(phenylethynyl)benzene (**3ac**)^[@ref17]^ {#sec4.1.20} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and *m*-tolyl tosylate (**2c**) (496 mg, 2.0 mmol) afforded 1-methyl-3-(phenylethynyl)benzene (**3ac**) (373 mg, 1.94 mmol, 97% yield). This compound is the same as **3ca**. ### 4.1.21. 1-Methyl-4-(phenylethynyl)benzene (**3ad**)^[@ref17]^ {#sec4.1.21} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and *p*-tolyl tosylate (**2d**) (496 mg, 2.0 mmol) afforded 1-methyl-4-(phenylethynyl)benzene (**3ad**) (369 mg, 1.92 mmol, 96% yield). This compound is the same as **3da**. ### 4.1.22. 1-Methoxy-4-(phenylethynyl)benzene (**3ae**)^[@ref17]^ {#sec4.1.22} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and 4-methoxyphenyl tosylate (**2e**) (557 mg, 2.0 mmol) afforded 1-methoxy-4-(phenylethynyl)benzene (**3ae**) (308 mg, 1.48 mmol, 74% yield). This compound is the same as **3ia**. ### 4.1.23. 1-(Phenylethynyl)naphthalene (**3af**)^[@ref10]^ {#sec4.1.23} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and naphthalen-1-yl tosylate (**2f**) (597 mg, 2.0 mmol) afforded 1-(phenylethynyl)naphthalene (**3af**) (397 mg, 1.74 mmol, 87% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.45 (dd, *J* = 8.3, 1.0 Hz, 1H), 7.81 (m, 2H), 7.75 (dd, *J* = 7.1, 1.2 Hz, 1H), 7.64 (m, 2H), 7.57 (m, 1H), 7.49 (m, 1H), 7.41 (m, 1H), 7.43--7.32 (m, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 133.2, 133.1, 131.6, 130.3, 128.7, 128.4, 128.34, 128.30, 126.7, 126.4, 126.2, 125.2, 123.3, 120.8, 94.3, 87.5; MS (EI) *m*/*z*: 228 (M^+^). ### 4.1.24. 2-(Phenylethynyl)naphthalene (**3ag**)^[@ref20]^ {#sec4.1.24} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and naphthalen-2-yl tosylate (**2g**) (597 mg, 2.0 mmol) afforded 2-(phenylethynyl)naphthalene (**3ag**) (429 mg, 1.88 mmol, 94% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.06 (m, 1H), 7.84-7.81 (m, 3H), 7.60--7.57 (m, 3H), 7.50 (m, 2H), 7.38-7.36 (m, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 133.0, 132.8, 131.7, 131.4, 128.42, 128.39, 128.3, 128.0, 127.79, 127.77, 126.7, 126.5, 123.3, 120.6, 89.8, 89.7; MS (EI) *m*/*z*: 228 (M^+^). ### 4.1.25. 3-(Phenylethynyl)pyridine (**3ah**)^[@ref17]^ {#sec4.1.25} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and pyridin-3-yl tosylate (**2h**) (499 mg, 2.0 mmol) afforded 3-(phenylethynyl)pyridine (**3ah**) (247 mg, 1.38 mmol, 69% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.77 (m, 1H), 8.55 (dd, *J* = 4.9, 1.7 Hz, 1H), 7.81 (m, 1H), 7.55 (m, 2H), 7.37 (m, 3H), 7.29 (ddd, *J* = 7.9,4.9, 0.9 Hz, 1H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 152.3, 148.6, 138.4, 131.7, 128.8, 128.5, 123.0, 122.5, 120.5, 92.6, 85.9; MS (EI) *m*/*z*: 179 (M^+^). ### 4.1.26. 8-(Phenylethynyl)quinoline (**3ai**)^[@ref21]^ {#sec4.1.26} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and quinolin-8-yl tosylate (**2i**) (599 mg, 2.0 mmol) afforded 8-(phenylethynyl)quinoline (**3ai**) (371 mg, 1.62 mmol, 81% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 9.08 (dd, *J* = 4.2, 1.8 Hz, 1H), 8.18 (dd, *J* = 8.3, 1.7 Hz, 1H), 8.02 (dd, *J* = 7.2, 1.4 Hz, 1H), 7.82 (dd, *J* = 8.2, 1.4 Hz, 1H), 7.72--7.70 (m, 2H), 7.54 (m, 1H), 7.46 (dd, *J* = 8.3, 4.2 Hz, 1H), 7.40--7.36(m, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 151.1, 148.1, 136.5, 134.2, 132.0, 128.40, 128.37, 128.3, 128.2, 126.1, 123.5, 123.4, 121.6, 95.4, 87.2; MS (EI) *m*/*z*: 229 (M^+^). ### 4.1.27. 1,3-Difluoro-5-(phenylethynyl)benzene (**3aj**)^[@ref22]^ {#sec4.1.27} 3-Phenylpropiolic acid (**1a**) (292 mg, 2.0 mmol) and 3,5-difluorophenyl tosylate (**2j**) (569 mg, 2.0 mmol) afforded 1,3-difluoro-5-(phenylethynyl)benzene (**3aj**) (343 mg, 1.60 mmol, 80% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.52 (m, 2H), 7.36 (m, 3H), 7.04 (m, 2H), 6.90 (tt, *J* = 9.0, 2.4 Hz, 1H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 162.7 (dd, *J*~C-F~= 248.7, 13.4 Hz), 131.7, 128.9, 128.5, 126.0 (t, *J*~C-F~ = 11.8 Hz), 122.3, 114.5 (dd, *J*~C-F~ = 20.3, 6.5 Hz), 104.4 (t, *J*~C-F~ = 25.4 Hz), 91.3, 87.1 (t, *J*~C-F~ = 4.0 Hz); MS (EI) *m*/*z*: 214 (M^+^). ### 4.1.28. 1-Methyl-2-(*p*-tolylethynyl)benzene (**3db**)^[@ref10]^ {#sec4.1.28} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and *o*-tolyl tosylate (**2b**) (525 mg, 2.0 mmol) afforded 1-methyl-2-(*p*-tolylethynyl)benzene (**3db**) (256 mg, 1.24 mmol, 62% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.48 (d, *J* = 7.5 Hz, 1H), 7.43--7.42 (m, 2H), 7.21--7.20 (m, 2H), 7.15--7.13 (m, 3H), 2.50 (s, 3H), 2.35 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 140.1, 138.3, 131.7, 131.4, 129.4, 129.1, 128.1, 125.5, 123.2, 120.4, 93.5, 87.7, 21.5, 20.7; MS (EI) *m*/*z*: 206 (M^+^). ### 4.1.29. 1-Methyl-3-(*p*-tolylethynyl)benzene (**3dc**)^[@ref20]^ {#sec4.1.29} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and *m*-tolyl tosylate (**2c**) (496 mg, 2.0 mmol) afforded 1-methyl-3-(*p*-tolylethynyl)benzene (**3dc**) (380 mg, 1.84 mmol, 92% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.42 (d, *J* = 8.1 Hz, 2H), 7.36--7.35 (m, 2H), 7.22 (m, 1H), 7.15--7.13 (m, 3H), 2.36 (s, 3H), 2.34 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 138.3, 138.0, 132.1, 131.5, 129.1, 129.0, 128.6, 128.2, 123.2, 120.3, 89.2, 55.9, 21.5, 21.2; MS (EI) *m*/*z*: 206 (M^+^). ### 4.1.30. 1-Methoxy-4-(*p*-tolylethynyl)benzene (**3de**)^[@ref17]^ {#sec4.1.30} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and 4-methoxyphenyl tosylate (**2e**) (557 mg, 2.0 mmol) afforded 1-methoxy-4-(*p*-tolylethynyl)benzene (**3de**) (342 mg, 1.54 mmol, 77% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.46 (d, *J* = 8.9 Hz, 2H), 7.40 (d, *J* = 8.1 Hz, 2H), 7.14 (dd, *J* = 8.4, 0.6 Hz, 2H), 6.87 (d, *J* = 8.9 Hz, 2H), 3.82 (s, 3H), 2.36 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 159.4, 138.0, 133.0, 131.3, 129.1, 120.5, 115.6, 113.9, 88.6, 88.2, 55.3, 21.5; MS (EI) *m*/*z*: 222 (M^+^). ### 4.1.31. 1-(*p*-Tolylethynyl)naphthalene (**3df**)^[@ref9]^ {#sec4.1.31} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and naphthalen-1-yl tosylate (**2f**) (597 mg, 2.0 mmol) afforded 1-(*p*-tolylethynyl)naphthalene (**3df**) (291 mg, 1.20 mmol, 60% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.44 (d, *J* = 7.9 Hz, 1H), 7.83 (m, 2H), 7.74 (m, 1H), 7.60--7.50 (m, 4H), 7.44 (dd, *J* = 8.2, 7.2 Hz, 1H), 7.19 (m, 2H), 2.38 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 138.5, 133.24, 133.19, 131.5, 130.2, 129.2, 128.5, 128.3, 126.7, 126.4, 126.2, 125.3, 121.1, 120.3, 94.5, 86.5, 21.5; MS (EI) *m*/*z*: 242 (M^+^). ### 4.1.32. 2-(*p*-Tolylethynyl)naphthalene (**3dg**)^[@ref20]^ {#sec4.1.32} 3-(*p*-Tolyl)propiolic acid (**1d**) (320 mg, 2.0 mmol) and naphthalen-2-yl tosylate (**2g**) (597 mg, 2.0 mmol) afforded 2-(*p*-tolylethynyl)naphthalene (**3dg**) (424 mg, 1.64 mmol, 82% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.04 (s, 1H), 7.82--7.78 (m, 3H), 7.57 (dd, *J* = 8.4, 1.6 Hz, 1H), 7.49--7.46 (m, 4H), 7.17 (m, 2H), 2.37 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 138.5, 133.0, 132.7, 131.5, 131.3, 129.1, 128.4, 127.9, 127.7, 126.55, 126.49, 120.8, 120.2, 89.9, 89.1, 21.5; MS (EI) *m*/*z*: 242 (M^+^). ### 4.1.33. 1-((2-Methoxyphenyl)ethynyl)naphthalene (**3gf**)^[@ref23]^ {#sec4.1.33} 3-(2-Methoxyphenyl)propiolic acid (**1g**) (352 mg, 2.0 mmol) and naphthalen-1-yl tosylate (**2f**) (597 mg, 2.0 mmol) afforded 1-((2-methoxyphenyl)ethynyl)naphthalene (**3gf**) (382 mg, 1.48 mmol, 74% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 8.58 (d, *J* = 8.3 Hz, 1H), 7.82--7.76 (m, 3H), 7.59--7.56 (m, 2H), 7.49 (m, 1H), 7.41 (dd, *J* = 8.1, 7.3 Hz, 1H), 7.29 (m, 1H), 6.95 (m, 1H), 6.88 (d, *J* = 8.3 Hz, 1H), 3.91 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 160.2, 133.5, 133.4, 133.3, 130.1, 129.9, 128.7, 128.3, 126.8, 126.6, 126.4, 125.3, 121.4, 120.6, 112.8, 110.8, 91.8, 91.0, 55.9; MS (EI) *m*/*z*: 258 (M^+^). ### 4.1.34. 1-Fluoro-4-(*p*-tolylethynyl)benzene (**3ld**)^[@ref9]^ {#sec4.1.34} 3-(4-Fluorophenyl)propiolic acid (**1l**) (328 mg, 2.0 mmol) and *p*-tolyl tosylate (**2d**) (496 mg, 2.0 mmol) afforded 1-fluoro-4-(*p*-tolylethynyl)benzene (**3ld**) (353 mg, 1.68 mmol, 84% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.51 (m, 2H), 7.43 (d, *J* = 8.1 Hz, 2H), 7.43 (d, *J* = 7.8 Hz, 2H), 7.05 (m, 2H), 2.38 (s, 2H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 162.4 (d, *J*~C-F~ = 250.0 Hz), 138.5, 133.4 (d, *J*~C-F~ = 8.3 Hz), 131.4, 129.1, 120.0, 119.6 (d, *J*~C-F~ = 3.6 Hz), 115.6 (d, *J*~C-F~ = 22.1 Hz), 89.2, 87.6, 21.5; MS (EI) *m*/*z*: 210 (M^+^). ### 4.1.35. 3-(*m*-Tolylethynyl)pyridine (**3ch**)^[@ref24]^ {#sec4.1.35} 3-(*m*-Tolyl)propiolic acid (**1c**) (320 mg, 2.0 mmol) and pyridin-3-yl tosylate (**2h**) (499 mg, 2.0 mmol) afforded 3-(*m*-tolylethynyl)pyridine (**3ch**) (174 mg, 0.90 mmol, 45% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.76 (m, 1H), 8.53 (dd, *J* = 4.9, 1.7 Hz, 1H), 7.78 (m, 1H), 7.44 (d, *J* = 8.1, 2H), 7.26 (m, 1H), 7.17 (d, *J* = 7.8 Hz, 2H), 2.37 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 152.2, 148.4, 139.0, 138.3, 131.6, 129.2, 123.0, 120.7, 119.4, 92.9, 85.3, 21.6; MS (EI) *m*/*z*: 193 (M^+^). ### 4.1.36. 1-(4-(*p*-Tolylethynyl)phenyl)ethanone (**3pd**)^[@ref17]^ {#sec4.1.36} 3-(4-Acetylphenyl)propiolic acid (**1p**) (376 mg, 2.0 mmol) and *p*-tolyl tosylate (**2d**) (496 mg, 2.0 mmol) afforded 1-(4-(*p*-tolylethynyl)phenyl)ethanone (**3pd**) (112 mg, 0.48 mmol, 24% yield); ^1^H NMR (500 MHz, CDCl~3~) δ 7.92 (m, 2H), 7.59 (m, 2H), 7.44 (m, 2H), 7.17 (m, 2H), 2.60 (s, 3H), 2.37 (s, 3H); ^13^C{^1^H} NMR (126 MHz, CDCl~3~) δ 197.3, 139.1, 136.0, 131.63, 131.59, 129.2, 128.4, 128.2, 119.5, 93.0, 88.1, 26.6, 21.6; MS (EI) *m*/*z*: 234 (M^+^). The Supporting Information is available free of charge on the [ACS Publications website](http://pubs.acs.org) at DOI: [10.1021/acsomega.7b01165](http://pubs.acs.org/doi/abs/10.1021/acsomega.7b01165).^1^H and ^13^C NMR spectral data of all compounds ([PDF](http://pubs.acs.org/doi/suppl/10.1021/acsomega.7b01165/suppl_file/ao7b01165_si_001.pdf)) Supplementary Material ====================== ###### ao7b01165_si_001.pdf The authors declare no competing financial interest. This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIP) (NRF-2015R1A4A1041036 and NRF-2017R1A2B2002929). The spectral data were obtained from the Korea Basic Science Institute, Gwangju Branch.

1. Introduction =============== Melanin is responsible for skin coloration and protects the skin from damage induced by ultraviolet (UV) light. UV-induced skin hyperpigmentation causes abnormal melanin production and accumulation \[[@B1-ijms-16-10470]\]. In the melanin production pathway, tyrosinase is the rate-limiting enzyme; this enzyme participates in the hydroxylation of L-tyrosine to L-3,4-dihydroxyphenylalanine (L-DOPA). L-DOPA is further oxidized to its corresponding *o*-quinone \[[@B2-ijms-16-10470]\]. Several skin hyperpigmentation disorders such as freckles, age spots, melasma, post-inflammatory melanoderma and other hyperpigmentation syndromes are the result of abnormal melanin accumulation \[[@B3-ijms-16-10470]\]. Hence, many tyrosinase inhibitors such as kojic acid \[[@B4-ijms-16-10470]\], arbutin \[[@B5-ijms-16-10470]\] and azelaic acid \[[@B6-ijms-16-10470]\] are used in skin whitening products to prevent or treat abnormal skin pigmentation \[[@B7-ijms-16-10470]\]. However, it has been reported that whitening products with chemical skin depigmenting agents can have significant side effects, including pigmented contact dermatitis caused by kojic acid \[[@B8-ijms-16-10470]\], genotoxicity caused by arbutin \[[@B9-ijms-16-10470]\] and transient erythema or skin irritation caused by azelaic acid \[[@B10-ijms-16-10470]\]. Therefore, the search for an effective and safe skin depigmenting agent is still ongoing in the field of cosmetic research and development. In the past, antioxidants have been widely used to prevent or treat disorders related to oxidative stress in the pharmaceutical and dermatological fields. Additionally, antioxidants also have been used in the cosmetic industry to delay or prevent skin aging. It has been reported that free radicals and reactive oxygen species (ROS) are associated with several diseases such as inflammation, aging and age-related diseases \[[@B11-ijms-16-10470],[@B12-ijms-16-10470]\]. Free radical damage to the skin caused by ROS and UV-irradiation stress plays an important role in photo-aging \[[@B13-ijms-16-10470],[@B14-ijms-16-10470]\]. Antioxidants are reported to interfere with the oxidation process by chelating oxidation-catalytic metals or by scavenging free radicals and ROS \[[@B15-ijms-16-10470],[@B16-ijms-16-10470]\]. Hence, many antioxidants have been used to reduce oxidative stress or damage in the human body \[[@B17-ijms-16-10470],[@B18-ijms-16-10470]\]. However, synthetic chemical antioxidants such as *tert*-butyl hydroxyanisole (BHA) and *tert*-butyl hydroxytoluene (BHT) have been reported to be carcinogenic \[[@B19-ijms-16-10470]\]. Therefore, many studies on plant-derived antioxidants have been reported over the past decade. It has also been found that ROS can accelerate skin pigmentation. For example, nitric oxide (NO) produced by ultraviolet-irradiated keratinocytes stimulates melanin production by increasing the expression of tyrosinase and tyrosinase-related protein 1 (TRP-1) \[[@B20-ijms-16-10470],[@B21-ijms-16-10470]\]. The contribution of ROS to melanogenesis was studied by using antioxidants such as *N*-acetyl cysteine to abolish UVB-induced melanin production \[[@B22-ijms-16-10470]\]. Further, melanogenesis was reported to produce hydrogen peroxide (H~2~O~2~) and ROS, which place melanocytes under high-grade oxidative stress. The ROS scavengers and inhibitors of ROS generation may down-regulate UV-induced melanogenesis \[[@B23-ijms-16-10470]\]. Hence, inhibitors of melanogenesis, antioxidants and ROS scavengers have been increasingly used in cosmetics to prevent undesirable skin hyperpigmentation \[[@B24-ijms-16-10470]\]. The use of essential oils as functional ingredients in cosmetics is gaining momentum because of consumers' growing interest in ingredients from natural sources. Essential oils and their components are gaining increasing interest because of widespread consumer acceptance and the potential for multipurpose functional use \[[@B25-ijms-16-10470]\]. Recently, it has been reported that essential oils extracted from the leaves of *Aremisia argyi* \[[@B26-ijms-16-10470]\], *Vitex negundo* Linn. \[[@B27-ijms-16-10470]\] and *Acorus macrospadiceus* (Yamamoto) F. N. Wei et Y. K. Li. \[[@B28-ijms-16-10470]\] show de-pigmentation activity. The increased usage of essential oils has raised a number of concerns in relation to adverse health effects which need to be addressed \[[@B29-ijms-16-10470],[@B30-ijms-16-10470]\]. Eucalypt trees are evergreen and belong to the *Eucalyptus* genus and *Myrtaceae* family. The *Eucalyptus* genus is native to Australia and is one of the most widely planted genera in the world. The essential oils from *Eucalyptus* species have been used for pharmaceutical and medicinal purposes \[[@B31-ijms-16-10470],[@B32-ijms-16-10470]\], and several studies have reported that essential oils from *E. camaldulensis* leaves displayed multiple pharmacological activities, including antibacterial \[[@B33-ijms-16-10470]\] and anti-inflammatory activities \[[@B34-ijms-16-10470]\], antitermitic activity \[[@B35-ijms-16-10470]\], larvicidal and mosquito repellent activities \[[@B36-ijms-16-10470],[@B37-ijms-16-10470]\], and antioxidative and antiradical activities \[[@B38-ijms-16-10470]\]. However, so far, there have been no reports regarding potential dermatological application of essential oils from the leaves or flowers of *Eucalyptus* species. The aim of this study is to identify the chemical compositions of the essential oils extracted from *Eucalyptus camaldulensis* flowers and to determine the oils' anti-melanogenesis activities and antioxidative characteristics. 2. Results and Discussion ========================= 2.1. Chemical Compositions of E. camaldulensis Flower Essential Oil ------------------------------------------------------------------- Hydrodistillation of the flowers of *E. camaldulensis* generated a yellow oil with a yield of 2.68 mL/100 g, based on the flower dry mass. The constituents of the oil that were identified are presented in [Table 1](#ijms-16-10470-t001){ref-type="table"}, where all compounds are listed in order of their elution from a DB-5 non-polar column. In gas chromatography, Kovats index is used to convert retention times into system-independent constants. The retention index of a certain chemical compound is its retention time normalized to the retention times of adjacently eluting *n*-alkanes. Fifty-four compounds were identified ([Table 1](#ijms-16-10470-t001){ref-type="table"}), representing 100% of the oil. Oxygenated monoterpenes predominated the compounds identified (34.9%), followed by oxygenated sesquiterpenes (31.8%), monoterpene hydrocarbons (29.0%), and sesquiterpene hydrocarbons (4.3%). Among the oxygenated monoterpenes, 1,8-cineole (23.9%) was the major compound, and of the oxygenated sesquiterpenes, α-eudesmol (11.6%), γ-eudesmol (8.0%), and elemol (5.0%) were the major components. γ-Terpinene (12.9%), α-pinene (6.1%), and *p*-cymene (4.9%) were the main monoterpene hydrocarbons. Several factors are involved in determining the constituents of essential oil, including plant cultivation and/or harvesting procedures. Additionally, different analytical techniques may also result in different GC/MS data. Eucalyptol, also known as 1,8-cineole, is the major ether component in the essential oil and the major component of *E. camaldulensis* Dehn \[[@B39-ijms-16-10470]\] As some synthetic ethers have been reported to show antioxidant activities \[[@B40-ijms-16-10470]\], we hypothesized that eucalyptol may account for the antioxidant activity of the essential oil. Additionally, the concentration of *p*-cymene (4.9%) in the essential oil was lower than that in the leaf oils of *E. camaldulensis* var. *brevirostris* \[[@B41-ijms-16-10470]\]. It has been found that secondary metabolites and bioactive phytoconstituents identified by GC/MS in various plants show antimicrobial, anti-inflammatory and antioxidant activities \[[@B42-ijms-16-10470],[@B43-ijms-16-10470]\]. The chemical constituents found in *E. camaldulensis* flower essential oil may contribute significantly to the oil's biological activity, but the biological role of the individual chemical in the essential oil still remain to be elucidated. ijms-16-10470-t001_Table 1 ###### Chemical composition of the flower oil from *Eucalyptus* *camaldulensis*. Consituents K.I. ^(a)^ K.I. ^(b)^ Concentration (%) Identification ^(c)^ -------------------------------------- ------------- ------------ ------------------- ---------------------- α-Thujene 925 930 0.6 MS, K.I., ST α-Pinene 937 939 6.1 MS, K.I., ST β-Pinene 976 979 0.3 MS, K.I., ST β-Myrcene 989 990 0.3 MS, K.I., ST α-Phellandrene 1002 1002 0.2 MS, K.I., ST α-Terpinene 1015 1017 0.8 MS, K.I., ST *p*-Cymene 1020 1024 4.9 MS, K.I., ST Limonene 1025 1029 1.2 MS, K.I., ST 1,8-Cineole 1027 1031 23.9 MS, K.I., ST *cis*-β-Ocimene 1032 1037 0.1 MS, K.I. γ-Terpinene 1055 1059 12.9 MS, K.I., ST Terpinolene 1086 1088 1.3 MS, K.I., ST Linalool 1095 1096 0.1 MS, K.I., ST 3-Methyl-3-butenyl 3-methylbutanoate 1112 1114 0.5 MS, K.I. exo-Fenchol 1120 1121 0.1 MS, K.I. 3-Methyl-2-butenyl 2-methylbutanoate 1138 1141 0.2 MS, K.I. δ-Terpineol 1163 1166 0.0 MS, K.I. Borneol 1167 1169 0.0 MS, K.I., ST Terpinen-4-ol 1175 1177 5.7 MS, K.I., ST α-Terpineol 1187 1188 3.1 MS, K.I., ST Nerol 1228 1229 0.2 MS, K.I., ST Methyl geranate 1323 1324 0.3 MS, K.I. β-Elemene 1389 1390 0.2 MS, K.I., ST ( *Z*)-Jasmone 1392 1392 0.6 MS, K.I. α-Gurjunene 1409 1409 0.1 MS, K.I. β-Caryophyllene 1418 1419 0.9 MS, K.I., ST Aromadendrene 1439 1441 0.1 MS, K.I., ST *trans*-Muurola-3,5-diene 1452 1452 0.2 MS, K.I. α-Humulene 1454 1454 0.2 MS, K.I., ST *allo*-Aromadendrene 1458 1460 0.1 MS, K.I. *cis*-Cadina-1(6),4-diene 1462 1463 0.3 MS, K.I. Viridiflorene 1496 1496 0.5 MS, K.I. α-Muurolene 1500 1500 0.3 MS, K.I. γ-Cadinene 1512 1513 0.2 MS, K.I. δ-Cadinene 1522 1523 0.6 MS, K.I. *cis*-Calamenene 1528 1529 0.4 MS, K.I. *trans*-Cadina-1,4-diene 1533 1534 0.2 MS, K.I. Elemol 1549 1549 5.0 MS, K.I., ST *epi*-Globulol 1555 1556 0.1 MS, K.I. Palustrol 1567 1568 0.2 MS, K.I. Spathulenol 1577 1578 0.2 MS, K.I., ST Caryophyllene oxide 1582 1583 0.2 MS, K.I., ST Globulol 1590 1590 1.0 MS, K.I., ST Guaiol 1600 1600 0.7 MS, K.I. *cis*-Isolongifolanone 1612 1613 0.5 MS, K.I. 1,10-di- *epi*-Cubenol 1618 1619 0.2 MS, K.I. *iso*-Leptospermone 1622 1622 0.2 MS, K.I. 10- *epi*-γ-Eudesmol 1622 1623 0.5 MS, K.I. 1- *epi*-Cubenol 1628 1628 0.3 MS, K.I. γ-Eudesmol 1630 1630 8.0 MS, K.I. τ-Cadinol 1640 1640 2.6 MS, K.I. α-Muurolol 1645 1646 0.4 MS, K.I. α-Eudesmol 1652 1653 11.6 MS, K.I. (2 *z*,6*z*)-Farnesol 1698 1698 0.1 MS, K.I. Monoterpene hydrocarbons (%) 29.0 Oxygenated monoterpenes (%) 34.9 Sesquiterpene hydrocarbons (%) 4.3 Oxygenated sesquiterpenes (%) 31.8 Oil Yield (mL/100 g) 2.68 ± 0.02 ^a^ Relative retention indices experimental: *n*-alkanes (C~9~--C~24~) were used as reference points in the calculation of relative retention indices; ^b^ Kovats index on a DB-5 column with reference to *n*-alkanes \[[@B1-ijms-16-10470]\]; ^c^ MS, NIST and Wiley library spectra and the literature; K.I., Kovats index; ST, authentic standard compounds. 2.2. Cell Viability ------------------- To assess the effect of *E. camaldulensis* flower essential oil on cell viability, B16F10 mouse melanoma cells were treated with different concentrations of essential oil (0.013, 0.02075 and 0.0415 mg/mL) for 24 h. The MTT assay is a colorimetric assay for assessing cell viability. NAD(P)H-dependent cellular oxidoreductase enzymes may, under defined conditions, reflect the number of viable cells present. The results indicated that the flower essential oil had no inhibitory effect on B16F10 cell viability ([Figure 1](#ijms-16-10470-f001){ref-type="fig"}). Hence, we chose similar essential oil concentrations for our B16F10 melanoma cell experiments. {#ijms-16-10470-f001} 2.3. Inhibitory Effects of E. camaldulensis Flower Essential Oil on Melanin Production -------------------------------------------------------------------------------------- Mushroom tyrosinase is widely used as the target enzyme in screening potential inhibitors of melanogenesis. The results shown in [Figure 2](#ijms-16-10470-f002){ref-type="fig"}a revealed that *E. camaldulensis* flower essential oil inhibits mushroom tyrosinase activity. Remaining enzyme activities were 82.95% ± 3.87%, 77.73% ± 5.53% and 70.04% ± 6.78% of the control for essential oil treatments of 5.2, 13 and 26 mg/mL, respectively. Tyrosinase activity was also inhibited by kojic acid, resulting in a remaining enzyme activity of 59.14% ± 1.97% of the control. Even though the concentrations of essential oil used were higher than that of kojic acid, the enzyme inhibition effect of the oil is still less than that of kojic acid. Thus, the essential oil may be a minor inhibitor of mushroom tyrosinase. The results in [Figure 2](#ijms-16-10470-f002){ref-type="fig"}b further indicated that flower essential oil significantly decreased the intracellular melanin content. The melanin content was 85.59% ± 4.39%, 74.18% ± 2.03% and 68.9% ± 1.37% for *E. camaldulensis* flower essential oil treatments of 0.013, 0.02075 and 0.0415 mg/mL, respectively. The remaining melanin content for arbutin was 77.01% ± 3.13% of the control. After treatment, the remaining intracellular tyrosinase activity was 84.38% ± 2.43%, 70.96% ± 1.55% and 65.99% ± 1.74% for *E. camaldulensis* flower essential oil treatments of 0.013, 0.02075 and 0.0415 mg/mL, respectively. The intracellular tyrosinase activity was 81.95% ± 4.78% after the cells were treated with arbutin ([Figure 2](#ijms-16-10470-f002){ref-type="fig"}c). The results shown in [Figure 2](#ijms-16-10470-f002){ref-type="fig"}c were in accordance with the results indicated in [Figure 2](#ijms-16-10470-f002){ref-type="fig"}b, which means that essential oil inhibited B16F10 intracellular tyrosinase activity and then decreased the melanin content. Those results indicated that *E. camaldulensis* flower essential oil exhibited a potent inhibitory effect on α-MSH-induced tyrosinase activity. {#ijms-16-10470-f002} 2.4. E. camaldulensis Flower Essential Oil Inhibited the Expression Levels of Melanogenesis-Related Proteins ------------------------------------------------------------------------------------------------------------ The expression levels of intracellular melanogenesis-related proteins were examined using Western blots. The results presented in [Figure 3](#ijms-16-10470-f003){ref-type="fig"}a indicated that *E. camaldulensis* flower essential oil treatment led to a significantly reduced level of MC1R, tyrosinase, TRP-1 and TRP-2. However, the protein content of MITF was not significantly different after treatment. Furthermore, [Figure 3](#ijms-16-10470-f003){ref-type="fig"}b revealed that the expression levels of p38, p-p38; JNK, p-JNK; CREB, p-CREB; ERK and p-ERK were also decreased after treatment with essential oil at a concentration of 0.0415 mg/mL, which suggested that the MAPK, JNK, PKA and ERK signaling pathways are involved in *E. camaldulensis* flower essential oil-mediated inhibition of melanogenesis. In the present study, α-MSH was used as a cAMP inducer to stimulate melanin production. It is reported that α-MSH binds MC1R and activates adenylate cyclase, which catalyzes the conversion of ATP to cAMP and increases intracellular cAMP levels. This rise in intracellular cAMP levels activates cAMP-mediated signaling pathways and promotes pigmentation in skin melanocytes \[[@B44-ijms-16-10470],[@B45-ijms-16-10470]\]. Tyrosinase, TRP-1 and TRP-2 are three major enzymes responsible for melanin biosynthesis in mammalian melanocytes \[[@B46-ijms-16-10470]\]. MITF is the major transcriptional regulator of the tyrosinase, TRP-1 and TRP-2 genes and is known to be the most important regulator of melanocyte pigmentation \[[@B47-ijms-16-10470]\]. The results shown in [Figure 3](#ijms-16-10470-f003){ref-type="fig"}a indicated that *E. camaldulensis* flower essential oil decreased the expression levels of MC1R, tyrosinase, TRP-1 and TRP-2, inhibiting tyrosinase activity and decreasing the melanin content in B16F10 cells. However, the expression of MITF was unchanged after treatment with essential oil. Evaluation of the gene expression level of MITF will be carried out in the near future. The decreased MC1R expression that was observed suggests that the essential oil inhibited melanogenesis induced via α-MSH-mediated intracellular cAMP up-regulation. Moreover, the results shown in [Figure 3](#ijms-16-10470-f003){ref-type="fig"}b further confirm that flower essential oil inhibited cAMP-mediated PKA signaling. It has been reported that protein kinase A (PKA) signaling is involved in melanin production. The elevation of cellular cAMP levels could activate PKA, leading to activation of CREB and MITF transcriptional activity and resulting in expression of melanogenesis-related proteins \[[@B48-ijms-16-10470]\]. Our results shown in [Figure 3](#ijms-16-10470-f003){ref-type="fig"}b revealed decreased expression of p-CREB and CREB, suggesting that the essential oil inhibits melanin synthesis by blocking the PKA pathway. It is well known that the MAPK family comprises three types of protein kinases, including extracellular responsive kinase (ERK), c-Jun *N*-terminal kinase (JNK) and p38 MAPK. Earlier studies have reported that p38 activation positively contributes to melanin production by activating the cAMP response element-binding protein (CREB), which then activates MITF expression \[[@B49-ijms-16-10470],[@B50-ijms-16-10470]\]. The results in [Figure 3](#ijms-16-10470-f003){ref-type="fig"}b provide evidence that the essential oil can inactivate p38 and CREB, thereby inhibiting melanin production. In the Western blotting experiments, GAPDH was used as an internal loading control. Because the GAPDH gene is often stably and constitutively expressed at high levels in most tissues and cells, it is considered a housekeeping gene. For this reason, GAPDH is commonly used as a loading control for Western blot. {#ijms-16-10470-f003} 2.5. E. camaldulensis Flower Essential Oil Down-Regulated MAPK and PKA Signaling Pathways ----------------------------------------------------------------------------------------- To elucidate the possible action mechanisms of the essential oil on melanin production, several protein kinase regulators of melanogenesis-related signaling pathways were tested. The application of *E. camaldulensis* flower essential oil to IBMX-treated B16F10 cells significantly decreased cellular melanin content. The remaining melanin content was 51.72% ± 2.13% of the control after treatment with IBMX and essential oil (0.0415 mg/mL). The results indicated that cAMP-mediated PKA signaling was affected by the flower essential oil ([Figure 4](#ijms-16-10470-f004){ref-type="fig"}a). The addition of *E. camaldulensis* flower essential oil in PD98059-treated B16F10 cells also decreased cellular melanin content. The remaining melanin content was 45.38% ± 4.28% of the control after treatment with PD98059 and essential oil (0.0415 mg/mL). The results shown in [Figure 4](#ijms-16-10470-f004){ref-type="fig"}b indicated that the ERK-mediated signaling pathway is involved in melanin production and was affected by *E. camaldulensis* flower essential oil treatment. To investigate the role of p38 MAPK signaling on the anti-melanogenic effect of *E. camaldulensis* flower essential oil, we employed a specific inhibitor of p38, SB203580, which blocks p38 MAPK signaling. The results in [Figure 4](#ijms-16-10470-f004){ref-type="fig"}c revealed that SB203580 attenuated α-MSH-induced melanin synthesis. The remaining melanin content was 55.19% ± 6.03% of the control after treatment with SB203580 and essential oil (0.0415 mg/mL).The addition of *E. camaldulensis* flower essential oil to SP600125-treated B16F10 cells significantly decreased the cellular melanin content. The results indicated that the JNK-mediated signaling pathway, which is involved in melanin production, was affected by *E. camaldulensis* flower essential oil. The remaining melanin content was 54.82% ± 5.79% of the control after treatment with SP600125 and the essential oil (0.0415 mg/mL) ([Figure 4](#ijms-16-10470-f004){ref-type="fig"}d). ###### The effects of *E. camaldulensis* flower essential oil on melanin content in IBMX, PD98059-, SB203580- and SP600125-treated B16F10 cells. (**a**) The melanin content of IBMX-treated cells and (IBMX + *E. camaldulensis* flower essential oil)-treated cells; (**b**) The melanin content of PD98059-treated cells and (PD98059 + *E. camaldulensis* flower essential oil)-treated cells; (**c**) The melanin content of SB203580- and (SB203580 + *E. camaldulensis* flower essential oil)-treated cells; (**d**) The melanin content of SP600125- and (SP600125 + *E. camaldulensis* flower essential oil)-treated cells. The results are represented as percentages of the control, and the data are presented as the mean ± S.D. of three separate experiments. Values significantly different from the control are indicated. \*\*\* *p* \< 0.001.   2.6. Antioxidant Characteristics of E. camaldulensis Flower Essential Oil ------------------------------------------------------------------------- The antioxidant activity of *E. camaldulensis* flower essential oil was first measured in terms of radical scavenging ability using the DPPH assay. DPPH is a dark-colored crystalline powder composed of stable free-radical molecules. The DPPH assay is known to provide reliable information concerning the antioxidant capacity of specific compounds or extracts across a short time scale. In this study, vitamin C (0.05 mM; 0.53 mg/mL) and *tert*-butyl hydroxyanisole (BHA) (1 mg/mL) were used as positive antioxidant standards. The DPPH scavenging capacity of the oil was 34.22% ± 0.62%, 56.05% ± 0.74% and 83.58% ± 2.46% of control for essential oil concentrations of 17.3, 34.7 and 69.3 (mg/mL), respectively. In comparison, the scavenging capacities of vitamin C and BHA were 66.66% ± 2.09% and 91.89% ± 1.03%, respectively ([Figure 5](#ijms-16-10470-f005){ref-type="fig"}a). This ABTS radical cation is blue in color and is reactive towards most antioxidants. During this reaction, the blue ABTS radical cation is converted back to its colorless neutral form. The reaction may be monitored spectrophotometrically. The ABTS^+^ assay was employed to measure the antioxidant activity of the *E. camaldulensis* flower essential oil. The ABTS^+^ scavenging capacity of the essential oil was 20.55% ± 3.54%, 36.63% ± 0.89% and 53.58% ± 4.14% of the control at concentrations of 0.0416, 2.08, 4.16 mg/mL, respectively. In contrast, the ABTS^+^ scavenging capacity of Trolox^®^ (0.04, 2 and 4 mg/mL) was 67.74% ± 0.49%, 71.93% ± 6.25%, and 93.72% ± 0.57%, respectively. The results in [Figure 5](#ijms-16-10470-f005){ref-type="fig"}b indicated that the essential oil scavenges a significant amount of ABTS^+^ free radicals. However, the flower essential oil exhibited lower ABTS^+^ radical scavenging capacity than Trolox^®^. {#ijms-16-10470-f005} To further estimate the antioxidant capacity of *E. camaldulensis* flower essential oil in a cellular environment, intracellular ROS levels were determined. The concentration of H~2~O~2~ employed was 24 mM. After treatment, the remaining intracellular ROS induced by H~2~O~2~ was 78.42% ± 3.95% for 0.0415 mg/mL of the essential oil, which is similar to the value of 79.05% ± 5.44% for Trolox^®^ (2.0 mM; 0.5 mg/mL) ([Figure 5](#ijms-16-10470-f005){ref-type="fig"}c). The results shown in [Figure 5](#ijms-16-10470-f005){ref-type="fig"}c revealed that *E. camaldulensis* flower essential oil significantly suppressed intracellular ROS production only at the higher concentration of 0.0415 mg/mL. Therefore, *E. camaldulensis* flower essential oil was able to protect melanoma cells from oxidative injury by depletion of ROS generation; in addition, the oil may downregulate UV-induced melanogenesis. To elucidate the antioxidant characteristics of *E. camaldulensis* flower essential oil, DPPH, ABTS^+^ .radical scavenging activity and ROS-scavenging capacity of the essential oil were determined as previously described \[[@B19-ijms-16-10470],[@B51-ijms-16-10470]\]. The essential oil showed considerable antioxidant potential in all of the above analytical studies. The results demonstrated the antioxidant potential of *E. camaldulensis* flower essential oil over different ranges with distinct efficiencies. [Figure 5](#ijms-16-10470-f005){ref-type="fig"}a,b showed that the differential free radical scavenging activities of the essential oil against DPPH and ABTS^+^ radicals may result from the different mechanisms of antioxidant-radical interactions in the two assays. Furthermore, the reaction stoichiometry between the potential antioxidant chemicals in the essential oil may be different, resulting in different radical scavenging capacities \[[@B52-ijms-16-10470]\]. It has been reported that ultraviolet irradiation induces the formation of reactive oxygen species (ROS) in cutaneous tissue, provoking toxic changes including lipid peroxidation and enzyme inactivation \[[@B53-ijms-16-10470]\]. The results shown in [Figure 5](#ijms-16-10470-f005){ref-type="fig"}c suggested that the *E. camaldulensis* flower essential oil-induced decrease in melanin production may be attributed to depletion of cellular ROS. 3. Experimental Section ======================= 3.1. Chemicals and Reagents --------------------------- Kojic acid, arbutin, Folin-Ciocalteau's phenol reagent, 2,2\'-azino-bis (3-ethylbenzthiazoline-6-sulfonic acid (ABTS), 1,1-diphenyl-2-picrylhydrazyl (DPPH), [l]{.smallcaps}-ascorbic acid (vitamin C), butylated hydroxyanisole (BHA), and all other chemicals and solvents were obtained from Sigma-Aldrich (St. Louis, MO, USA). 3.2. Plant Materials and Extraction of Essential Oils ----------------------------------------------------- Flowers of *E. camaldulensis* were collected in May 2011 from Lienhuachih Research Center of the Taiwan Forestry Research Institute in central Taiwan. The samples were compared with specimen no. 59916 from the Herbarium of National Ilan University (NIU). The voucher specimen (CLH-040) was deposited in the NIU herbarium. Flowers were collected for subsequent extraction and analysis. The flowers (1 kg) were first diced, then placed in a round-bottom flask and hydrodistilled for 8 h with 3 L of distilled water. The essential oil obtained was dried with anhydrous sodium sulfate. The oil yield and all test data are the average of triplicate analyses. The essential oil was collected in a sealed glass bottle and stored in a 4 °C refrigerator. In the present study, the essential oil was diluted with dimethyl sulfoxide (DMSO), and DMSO was used as a negative control in the following experiments. 3.3. Gas Chromatography-Mass Spectrometry (GC/MS) Analysis of Essential Oil --------------------------------------------------------------------------- A Hewlett-Packard HP 6890 gas chromatograph equipped with a DB-5 fused silica capillary column (30 m × 0.25 mm × 0.25 μm film thickness, J&W Scientific, Folsom, CA, USA) and a FID detector was used for quantitative determination of oil components. Oven temperature was programmed as follows: 50 °C for 2 min, rising to 250 °C at 5 °C/min. Injector temperature: 270 °C. Carrier gas: He with a flow rate of 1 mL/min. Detector temperature: 250 °C, split ratio: 1:10. Diluted samples (1.0 μL, 1/100, *v*/*v*, in ethyl acetate) were injected manually in split mode. Identification of the oil components was based on their retention indices and mass spectra, obtained from GC/MS analysis on a Hewlett-Packard HP 6890/HP5973 equipped with a DB-5 fused silica capillary column (30 m × 0.25 mm × 0.25 μm film thickness, J&W Scientific). The GC analysis parameters listed above were used, and MS spectra were obtained (full scan mode: scan time: 0.3 s, mass range was *m*/*z* 30--500) in the EI mode at 70 eV. Data are expressed as the means ± SD of three independent experiments. Identification of the leaf essential oil constituents was based on comparisons of retention index (RI) \[[@B54-ijms-16-10470]\], retention times (RT) and mass spectra with those obtained from authentic standards and/or the NIST and Wiley library spectra as well as the literature \[[@B54-ijms-16-10470],[@B55-ijms-16-10470]\]. 3.4. Assay of Mushroom Tyrosinase Activity ------------------------------------------ To determine the inhibitory effects of *E. camaldulensis* flower essential oil on mushroom tyrosinase activity, enzyme inhibition experiments were carried out in triplicate as previously described with a slight modification \[[@B56-ijms-16-10470]\]. An aqueous solution of mushroom tyrosinase (200 units) was added to a 96-well microplate, in a total volume of 200 μL containing 5 mM [l]{.smallcaps}-DOPA dissolved in 50 mM phosphate buffer (pH 6.8) and the essential oil (5.2, 13 and 26 mg/mL) or kojic acid (200 μM; 0.028 mg/mL). The assay mixture was incubated at 37 °C for 30 min. After incubation, the amount of dopachrome produced in the reaction mixture was measured by spectrophotometric analysis of absorbance at 490 nm. 3.5. Cell Culture and Cell Viability Assay ------------------------------------------ B16F10 (ATCC CRL-6475, BCRC60031) cells were obtained from the Bioresource Collection and Research Center (BCRC), Taiwan. The cells were maintained in DMEM (Hyclone, Logan, UT, USA) supplemented with 10% fetal bovine serum and 1% antibiotics at 37 °C and 5% CO~2~ in a humidified incubator. Cell viability assays were performed using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) \[[@B57-ijms-16-10470]\]. The cells were exposed to various concentrations of *E. camaldulensis* flower essential oil (0.013, 0.02075 and 0.0415 mg/mL) for 24 h, and the MTT solution was then added to the wells. The insoluble derivative of MTT produced by an intracellular dehydrogenase was solubilized with ethanol-DMSO (1:1 mixture solution). The absorbance of the wells at 570 nm was read using a microplate reader. 3.6. Measurement of Intracellular Melanin Content ------------------------------------------------- Intracellular melanin content was measured as previously described with some modifications \[[@B58-ijms-16-10470]\]. The cells were treated with α-melanocyte stimulating hormone (α-MSH) (100 nM) for 24 h and further treated with either *E. camaldulensis* flower essential oil (0.013, 0.02075 and 0.0415 mg/mL) or arbutin (2.0 mM; 0.545 mg/mL) for another 24 h. After treatment, the cells were detached by incubation in trypsin/EDTA and subsequently centrifuged at 5000× *g* for 5 min. The cell pellets were solubilized in 1 N NaOH at 60 °C for 60 min. The melanin content was assayed by absorbance at 405 nm with spectrophotometric analysis. 3.7. Assay of Intracellular Tyrosinase Activity ----------------------------------------------- Intracellular tyrosinase activity was determined as previously described \[[@B59-ijms-16-10470]\]. The cells were treated with α-MSH (100 nM) for 24 h, then further treated with various concentrations of *E. camaldulensis* flower essential oil (0.013, 0.02075 and 0.0415 mg/mL) or arbutin (2.0 mM; 0.545 mg/mL) for another 24 h. After treatment, the cells were washed twice with phosphate-buffered saline and homogenized with 50 mM PBS (pH 7.5) buffer containing 1.0% Triton X-100 and 0.1 mM phenylmethyl-sulfonyl fluoride (PMSF). Cellular extracts (100 μL) were mixed with freshly prepared [l]{.smallcaps}-DOPA solution (5.0 mM in 50 mM phosphate-buffered saline, pH 6.8) and incubated at 37 °C for 30 min. The absorbance at 490 nm was measured with a microplate reader Gen 5™ (BIO-TEK Instrument, Winooski, VT, USA) to monitor the production of dopachrome. 3.8. Western Blotting Assay --------------------------- The cells were treated with *E. camaldulensis* flower essential oil (0.013, 0.02075 and 0.0415 mg/mL) or kojic acid (200 μM) and lysed in PBS containing proteinase inhibitors at 4 °C for 20 min. Proteins (50 μg) were resolved by SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to a polyvinylidene fluoride (PVDF) filter. The nylon filter was blocked for 1 h in 5% fat-free milk in PBST buffer (PBS with 0.05% Tween-20). After a brief wash, the filter was incubated overnight at 4 °C with several antibodies; these antibodies included anti-MC1R (1:1000), anti-MITF (1:1000), anti-tyrosinase (1:2000), anti-TRP1 (1:6000), anti-TRP2 (1:1000), anti-GAPDH (1:1500), anti-P38 (1:500), anti-p-P38 (1:500), anti-JNK (1:500), anti-p-JNK (1:500), anti-CREB (1:500), anti-p-CREB (1:500), anti-ERK (1:500), anti-p-ERK (1:500). Following incubation, the filter was extensively washed in PBST buffer. Subsequent incubation with goat anti-mouse antibody (1:10,000) conjugated to horseradish peroxidase was conducted at 25 °C for 2 h. The blot was visualized using an ECL reagent. The relative amounts of expressed proteins compared to total GAPDH were analyzed using Multi Gauge 3.0 software (Fuji, Tokyo). 3.9. Protein Kinase Regulators Assay ------------------------------------ The cells were treated with α-MSH (100 nM) for 24 h followed by a 1 h addition of different protein kinase regulators (10 μM), including IBMX, PD98059, SB203580 and SP600125. After these treatments, *E. camaldulensis* flower oil (0.0415 mg/mL) and the above kinase regulators were added to the cells, and cells were incubated for an additional 23 h. The melanin contents were assayed as described above. 3.10. DPPH Scavenging Activity Assay ------------------------------------ The antioxidant activity of *E. camaldulensis* flower essential oil was first measured by measuring the DPPH scavenging ability \[[@B60-ijms-16-10470]\]. Essential oil at various concentrations (17.3, 34.7 and 69.3 mg/mL) was added to 2.9 mL of DPPH (60 μM) solution. When DPPH reacts with any antioxidant in the essential oil that can donate hydrogen, it is reduced, and the resulting decrease in absorbance at 517 nm can be recorded using a UV-Vis spectrophotometer (Jasco, V-630, Tokyo, Japan). In this study, vitamin C (0.0088 mg/mL; 0.05 mM) and BHA (1 mg/mL) were used as antioxidant standards. 3.11. ABTS^+^ Scavenging Capacity Assay --------------------------------------- ABTS decolorization assays were carried out as previously described \[[@B61-ijms-16-10470]\]. This assay involves the generation of ABTS^+^ chromophores by oxidizing ABTS with potassium persulfate. The ABTS radical cation was produced by reacting 7 mM stock solution of ABTS with 2.45 mM potassium persulfate and allowing the mixture to stand in the dark for at least 6 h before use. Absorbance at 734 nm was measured 10 min after mixing of different concentrations of the *E. camaldulensis* flower essential oil (0.0416, 2.08 and 4.16 mg/mL) with 1 mL of ABTS^+^ solution. The ABTS^+^ scavenging capacity of the essential oil was compared with that of Trolox^®^ (2.0 or 4.0 mg/mL). 3.12. Determination of Cellular ROS Level ----------------------------------------- Cells were cultured in 24-well plates and treated with *E. camaldulensis* flower essential oil (0.013, 0.02075 and 0.0415 mg/mL) for 24 h. The cells were then incubated with 24 mM H~2~O~2~ at 37 °C for 30 min. After incubation, 2\',7\'-dichloro-fluorescein diacetate (DCFH-DA) was added to the wells, and the cells were cultured for an additional 30 min. The fluorescence intensities of DCF were measured at an excitation wavelength of 504 nm and an emission wavelength of 524 nm using a Fluoroskan Ascent fluorescent reader (Thermo Scientific, Vantaa, Finland). Data were analyzed using Ascent software (Thermo Scientific, Vantaa, Finland) \[[@B62-ijms-16-10470]\]. 3.13. Statistical Analysis -------------------------- Statistical analysis of the experimental data points was performed using the ANOVA test, which was used to compare the measured data using SPSS 12.0 statistical software (SPSS Inc., Chicago, IL, USA). Differences were considered as statistically significant for *p* \< 0.05. 4. Conclusions ============== This is the first report regarding the effect of *E. camaldulensis* flower essential oil on melanin production. In the present study, it was determined that *E. camaldulensis* flower essential oil significantly inhibits tyrosinase activity and decreases melanin synthesis. Moreover, *E. camaldulensis* flower essential oil also exhibits intracellular free radical scavenging activity. The results suggest that *E. camaldulensis* flower essential oil decreases melanin production, likely by inhibiting the signaling pathway that regulates tyrosinase activity or by depleting cellular ROS. In addition, our results demonstrate that *E. camaldulensis* flower essential oil decreases melanogenesis in melanoma cells by inactivating PKA and MAPK signaling pathways and inhibiting tyrosinase activity ([Figure 6](#ijms-16-10470-f006){ref-type="fig"}). Additionally, the inhibitory effect of *E. camaldulensis* flower essential oil on melanin production may also be mediated by the depletion of the intracellular ROS. {#ijms-16-10470-f006} The study was financially supported by the National Science Council, Taiwan, under Grant No. NSC 102-2221-E-241-019 and NSC 102-2632-B-241-001-MY3. Huey-Chun Huang conducted the tyrosinase-related studies, participated in the enzyme assays and drafted the manuscript; Ya-Chi Ho, Jia-Min Lim and Tzu-Yun Chang conducted the antioxidant and Western blotting experiments; Chen-Lung Ho harvested the *E. camaldulensis* flowers, extracted the essential oil and conducted GC/MS experiments; Tsong-Min Chang participated in design and coordination of the study, performed the statistical analysis, and drafted the manuscript. All authors read and approved the final manuscript. The authors declare no conflict of interest.

Why phenomics is key to face climate change and food security? {#s1} ============================================================== In the past decades, climatic variations related to El Niño or La Niña phenomena have brought serious challenges to the agricultural sector in LAC. While drought is the main threat to food production associated to La Niña, El Niño can cause heavy rains, flooding or extremely hot or cold weather (Allen and Ingram, [@B1]). In the last 150 years, earth\'s temperature increased at a rate of 0.045°C per decade, with almost four-fold (0.177°C) in the last 25 years (IPCC, [@B16]), and will continue to raise by another 1.1--6.4°C over the next century (Jin et al., [@B17]). This increase in temperature can lead to several agricultural associated problems such as yield reduction as a results of droughts, and the emergence and spreading of plant diseases and pests (FAO, [@B9]). Therefore, a better use of plant genetic resources and plant breeding (Borrás and Slafer, [@B5]), are key to tackling the imminent impact of climate change in food security. Further, a multidisciplinary approach that includes disciplines such as omics technologies (e.g., genomics, phenomics, proteomics, and metabolomics), plant physiology, eco-physiology, plant pathology and entomology, and soil science will be critical to increase crop resilience to climate change (Reynolds et al., [@B29]). Undoubtedly, public and private breeding programs have the challenge of producing stress tolerant cultivars whose yield potential and quality are also high. In order to increase the chances of producing desirable cultivars, breeders make a high number of crosses (e.g., Chilean wheat breeding programs generate \~800 crosses per year) and screen them under a limited number of environmental conditions (Araus and Cairns, [@B2]). Line crossing is a common experimental design for mapping quantitative trait loci (QTLs) in plant breeding. Crosses are initiated from at least two inbred lines, such as backcrosses, F2, and more derived generations (Xie et al., [@B36]). To increase the statistical inference space of the estimated QTL variance and ensure that polymorphic alleles are present in the parental gene pool, a sufficient number of parents must be sampled (Muranty, [@B24]). The number of traits measured per plot is normally limited to the size of the population. Increasing the number of traits to be measured requires additional time, resources and the use of skilled labor (Kipp et al., [@B18]). This represents a limitation toward the understanding of the interaction genotype × environment (G × E) (Furbank and Tester, [@B11]; Yang et al., [@B37]; Großkinsky et al., [@B13]; Rahaman et al., [@B26]). Although, genome sequencing has become relatively fast, cheap, and easy to produce, plant phenomics still lags behind. This unbalance has become a bottleneck in the understanding of G × E and it also limits the possibility of carrying out tests under field conditions (Lobos and Hancock, [@B20]). Therefore, there is a need to incorporate the evaluation of multiple morpho-physiological and physico-chemical traits at the high-throughput level to be able to understand for example pleiotropy or genomic variants that gave rise to a particular phenotype (Houle et al., [@B15]; Fahlgren et al., [@B8]). Due to the cost of high-throughput plant phenotyping, several international phenotyping networks have been established with the idea of joining efforts and produce research with impact. Some of the most prominent networks are: the European Plant Phenotyping Network (EPPN), Food and Agriculture COST Action FA1306, the International Plant Phenotyping Network (IPPN), the Australian Phenomics Network (APN), the German Plant Phenotyping Network (DPPN) and the U.K. Plant Phenomics Network (UKPPN). In Asia, the 1st Asia-Pacific Plant Phenotyping will be held in Beijing, China in October 2016 and the 3rd International Plant Phenotyping Symposium was held in Chennai, India in 2014. More recently in North America, the United States of America recently launched the North American Plant Phenotyping Network (NAPPN). Does latin america and the caribbean need to worry about phenomic development? {#s2} ============================================================================== Latin America is a region that includes Mexico, the Spanish/Portuguese speaking countries in Central America and the whole of South America, as well as the Caribbean (Latin America and the Caribbean---LAC). The region is highly heterogeneous in terms of climate, ecosystems, human population distribution, politics, economy and incomes, and cultural traditions. Out of a total of 17 megadiverse countries identified by the World Conservation Monitoring Centre (<http://www.unep-wcmc.org>), six are in Latin American, namely Brazil, Colombia, Ecuador, Mexico, Peru, and Venezuela. Furthermore, from the eight primary centers of origin and diversity, numbers VII (South Mexican and Central American) and VIII (South America Andes region: Bolivia, Peru, Ecuador; VIIIa The Chilean Center, and VIIIb Brazilian-Paraguayan Center) are based in the region (Vavilov, [@B32]). Due to LAC\'s diverse geography, climate change will impact the region severely. Compared to pre-industrial times, it is estimated that the mean temperature on the region will increase about 4.5°C by the end of the century (Reyer et al., [@B28]). Temperatures are expected to increase dramatically in the tropics and moderate at the subtropical regions in the north (Mexico) and south (southern Chile, Argentina and Uruguay) (Reyer et al., [@B28]). Annual precipitations are also likely to increase in Argentina, Uruguay, Brazil, Peru, Ecuador, and Colombia and decrease in the rest of the countries (Reyer et al., [@B28]). These changes have a direct impact on agricultural crop yields. It\'s expected that crops such as wheat, soybean and maize will reduce its yield potential, while others such as rice and sugar cane will increase it (Fernandes et al., [@B10]; Marin et al., [@B23]). The economic development of the regions where plant phenotyping and phenomics have been developed in the last 10 years (high-income countries) is completely different to that of LAC. According to the World Bank, around 37% of the LAC population lives under poverty or extreme poverty (World Bank, [@B35]), and near 60% of the people living in rural areas is under extreme poverty (RIMISP, [@B31]). Therefore, besides the climate change effects impacting LAC agriculture, there is also a significant knock-on the region economy, affecting particularly the lower socioeconomic strata (Ortiz, [@B25]). Although, LAC countries are wealthier, government efforts are mainly focused on priority areas such as education, health, employability, and infrastructure. Research and innovation in areas such as agriculture has been given a low priority. As a result, most Latin American farmers do not have the resources or the support to effectively adapt to a changing climate that is already showing its negative impact in agriculture (Lobos and Hancock, [@B20]). Therefore, LAC scientists and private sector must work together to develop strategies aiming at moving toward a more resilient agriculture, and one of them is the use of plant phenomics and phenotyping for breeding. Phenomics has become a powerful research tool to help breeders to generate cultivars adaptable to more challenging environmental scenarios. In the past decade, phenomics has been focused mainly on breeding of grain crops, but their application in other species of relevance for LAC (e.g., fruit, vegetables, forage and others) is almost absent (Lobos and Hancock, [@B20]). The potential of recent advances in phenomics encouraged the Plant Breeding and Phenomic Center (Dr. Gustavo A. Lobos, Universidad de Talca, Talca, Chile) and the National Plant Phenomics Centre (Dr. Anyela Camargo, IBERS, Aberystwyth University, U.K.) to organize the First Latin American Conference on Plant Phenotyping and Phenomics for Plant Breeding (November 30^st^ to December 2^nd^ 2015, Talca, Chile). This event had three main goals: (1) bring to Latin American researchers and students, international keynote speakers and plant breeding companies from around the world, to present their ongoing work on plant phenomics and phenotyping for plant breeding; (2) perform a workshop to train Latin American scientists and postgraduate students in the use of key plant phenotyping tools, the analysis of data and the mapping of traits to the genome; and (3) set up the Latin American Plant Phenomics Network (LatPPN), conceived to facilitate the training on high-throughput phenotyping and pre-breeding methodologies, scientific exchange of young/senior researchers and students, and to improve access to resources and research facilities. The conference covered a broad range of topics such as pre-breeding and breeding strategies, methods to measure and analyse trait data for plant breeding and the strategies to translate research from the bench to the field. International keynote speakers gave seminal talks and chaired the track of their expertise. Challenges and opportunities were also explored such as the handling of the high amount of data generated through high-throughput phenotyping. Multiple ideas were discussed to deal with every particular challenge. Participants also had the opportunity to attend five workshops that covered aspects such as the use of software and equipment for plant phenotyping (mainly by remote sensing), and data handling and manipulation. The LatPPN, which is chaired for two years by Chile (Dr. Gustavo A. Lobos) and Colombia (Dr. Anyela Camargo), had it first reunion during the 3^rd^ day of the conference. Representatives from LAC (Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, and Uruguay) and from other countries (Australia, Germany, Saudi Arabia, Spain, U.K., and U.S.A.) got together to discuss what LAC\'s breeding programs needed to do to become more efficiency in terms of plant phenotyping and phenomics. They also discussed the differences between phenotyping and the more complex concept of phenomics. This discussion helped to define where LAC currently stands (more focused on the phenotyping of few traits and low number of genotypes) and where it needs to be in the future (mostly oriented to the multidimensional approach of phenomics, considering a high number of genotypes assessed). For example, the wheat breeding program of INIA Chile, used to consider a classical approximation where the numbers of traits evaluated increases insofar the number of generation progresses: \~9 traits at F2--F5: susceptibility to *Puccinia triticina, P*. *graminis*, and *P*. *striiformis*, plant height, tillering capacity, type of spike, grain color, type of grain, and black point or other grain defects; \~16 at F6--F8: previous ones plus heading date, grain yield, and some grain characteristics such as test weight, protein and gluten content, sedimentation, and seed hardiness; and \~19 at F9--F10 where less than 5% of the original crosses are evaluated: previous ones plus some other required by millers such as W flour value, falling number, and some bakery aptitudes. Today, using spectrometry and thermography, this breeding program is aimed to predict some of these traits but also to consider other 30 morpho-physiological and physico-chemical characters (some examples covered in next section), screening \~800 genotypes per day. Is LAC oriented to phenomics or plant phenotyping? {#s3} ================================================== Due to resources\' availability such as equipment, skills and infrastructure, LAC has mainly focused on plant phenotyping. Although phenomics in LAC has not yet had a proper expansion, there are some good examples of institutions focusing on it: (i) The International Maize and Wheat Improvement Center (CIMMYT---Mexico) routinely uses remote sensing and high spec sensor technologies to screen for wheat and maize\'s responses to biotic and abiotic stresses, among them yield and its components, biomass, senescence (stay-green), water stress, and water use efficiency, canopy cover, photosynthetic capacity and activity (Zaman-Allah et al., [@B38]). Special emphasis is also put on 3D reconstruction for plant height, spike number and biomass determination; (ii) The Plant Breeding and Phenomic Center (University of Talca---Chile) have focused its efforts on the prediction of physiological traits by spectrometry and thermography (e.g., gas exchange, modulated chlorophyll fluorescence, pigments concentration, stem water potential, hydric and osmotic cell potential, cell membrane stability, lipid peroxidation, proline content, C and O isotopic composition) on several breeding programs (wheat, blueberries, alfalfa, strawberries, and quinoa) oriented to abiotic stresses (salt, water deficit and high temperature) (Garriga et al., [@B12]; Lobos et al., [@B21]; Estrada et al., [@B7]; Hernandez et al., [@B14]), developing also a software for exploratory analysis of high-resolution spectral reflectance data on plant breeding (Lobos and Poblete-Echeverría, [@B22]). In terms of phenotyping, most research institutes across the region have done some form of low to medium throughput phenotyping, for example: (i) The International Centre for Tropical Agriculture (CIAT---Colombia) is screening root architecture to identify markers associated to drought stress tolerance in beans and grasses (Villordo-Pineda et al., [@B33]; Rao et al., [@B27]); (ii) Embrapa (Brazil) uses traditional phenotyping to screen for root morphology in wheat (Richard et al., [@B30]); (iii) Universidade Federal de Mato Grosso, Brazil, uses traditional phenotyping tools (e.g., gas exchange measurements) to look for photosynthetic responses of tree species to seasonal variations in hydrology in the Brazilian Cerrado and Pantanal (Dalmagro et al., [@B6]); (iv) Researchers from Argentina uses conventional phenotyping equipment to investigate the response of seed weight and composition to changes in assimilate supply from leaves, to the incident solar radiation reaching the pods and to the combination of both, changes in assimilate supply from the leaves and incident solar radiation on pods of soybean plants (Bianculli et al., [@B4]), they are also trying to develop low cost tools in order to make that technology accessible to researches from LAC; (v) The International Potato Center (Peru) have improved the screening of potato breeding lines by spectroscopy (Ayvaz et al., [@B3]); and (vi) INIA (Uruguay) in collaboration with INIA (Chile) and the Plant Breeding and Phenomic Center (University of Talca---Chile), applied genotyping-by-sequencing to identify single-nucleotide polymorphisms, in the genomes of 384 wheat genotypes that were field tested in Chile under three different water regimes (Lado et al., [@B19]). How will LAC benefit from LatPPN? {#s4} ================================= The conference served as a platform to showcase LAC capabilities, investigate strengths, and weaknesses, and thereby identify where the challenges lie and what the knowledge and the technological gaps between the region and the rest of the world are. Given LAC\'s high heterogeneity in terms of climate, ecosystems and genetic diversity, as well as the differences of each country vulnerability to climate change, it was agreed how important it is for LAC\'s agri-food chain to take a more proactive role in the development of strategies leading to the selection of crops capable to withstand the impact of climate change. With the aim of identifying what LatPPN needed to do to strength LAC\'s plant phenotyping and phenomics research, the panel of participants identified the following key challenges: (i) develop LatPPN\'s own tailored identity: there is not a common crop but rather a wide diversity of them, from grasses to forest species. As previously mentioned, plant phenotyping and phenomics has been developed almost exclusively on cereal improvement, however LatPPN needs to focus on other breeding programs that are important for particular countries. For example: blueberries for Chile (Chile is the biggest exporter of fresh blueberries in the world, \~90,000 ton during 2015/16), potato for Peru (production was estimated to be 4.5 million tons for 2015), tangerines for Uruguay (production was \~6000 tons in 2014), pineapple for Costa Rica (since 2000, pineapple production has increased by nearly 300%, however production is very inefficient, each plant only produces two fruit over a period of 18--24 months, and requires significant amount fertilizer to do so) and Coffee for Colombia (exports account for \~810,000 ton in 2015) and Brazil (exports account for \~2.6 million tons in 2014). The production of these cash crops will face serious challenges (e.g., post-harvest life, or the incidence of physiological disorders, pests and diseases) in the coming decades due to the sensitivity of them to water shortages and heat stress. In this meeting, it was also highlighted: (ii) training on plant phenomics and phenotyping using strategies that allow the participation of several countries at the same time. We are aiming at finding resources to implement distance-training courses using currently available technologies such as webinars and teleconferences; (iii) learn from experienced researchers and current plant phenotyping and phenomics initiatives. In order to facilitate the interaction between researchers and institution, senior researchers on plant phenotyping and phenomics were invited to participate in the first meeting; (iv) since high-throughput phenotyping requires a broad range of capabilities (e.g., programmers, bioinformaticians, statisticians, biologists, agronomists, geneticists, physiologists), is important to promote interdisciplinary work between researchers; (v) identify the state of art of plant phenotyping and phenomics in LAC. In order to identify strengths, opportunities and weaknesses and develop targeted strategies, key information such as breeding programs, researchers, equipment and infrastructure, regional and local financial sources, and capabilities should be surveyed. All this information should be included on the future LatPPN webpage; (vi) distribute efforts on common goals (e.g., researchers from different countries working on the same species or problem), it will be necessary to standardize measurements and protocols; (vii) sharing of equipment and infrastructure; and (viii) LatPPN visibility and presence. To avoid early disenchantment, LatPPN needs to carry out activities to promote the network (e.g., events, postgraduate grants or proposal calls). In relation to weaknesses, the lack of a permanent budget to run network activities is one of LatPPN\'s main concerns. Currently, the Director and Co-Director, the executive committee (Dr. Paulo Hermann from EMBRAPA---Brazil and Dr. Gustavo Pereyra from INTA-CONICET---Argentina), and the representative members (three per country in charge of meet the local demands, thematic promotion, and economic resources leveraging) devote part of their time and resources to consolidate the network. However, they are looking into sources of support within LAC and worldwide. At the country level, there are a number of countries that have access to grants provided by their own governments. At regional level, there are a number of organizations such as PROCISUR and PROCITROPICOS, which provide regular grant support for agricultural research initiatives. At international level, there are several organizations such as FAO (the Food and Agricultural Organization), EU (the European Union), and IBS (the Inter-American Development Bank) who support agricultural research in LAC. Another weakness is LAC\'s low publication rate and the lack of accessibility of LAC institutions to main bibliographic databases. According to the World Bank, the number of publications produced by the most important economies in LAC in 2012 was 48,622 from Brazil, 13,112 from Mexico, 8,053 from Argentina, 5,158 from Chile and 4,456 from Colombia. Brazil is the only country whose output is equivalent to high-income countries where phenomics have been developing in the last 10 years; U.S.A. (412,542), Germany (101,074), U.K. (97,332), France (72,555), Spain (53,342), and Australia (47,806) (World Bank, [@B34]). In term of access to bibliographic databases, most of the institutions in the region have limited or no access to main bibliographic databases such as Scopus and Web of Knowledge. This is serious limitation to the dissemination of the work developed in LAC, especially if we are aiming at improving plant breeding programs through the use of plant phenotyping and phenomics. Despite the weaknesses, currently there are several international research institutes who are already formally collaborating with LAC on plant phenotyping and phenomics. Some of them are, Lemnatec (Germany), CSIRO (Australia), IBERS (U.K.), Universidad de Barcelona (Spain), the Julich Plant Phenomics Centre (German), and the James Hutton Institute (U.K.). The establishment of LatPPN represented a big step forward toward the consolidation of a common mind-set in the field of plant phenotyping and phenomics across LAC. Clearly there are more opportunities than disadvantages, and each weakness needs to be addressed having in mind a regional approach. Conclusions and future work {#s5} =========================== Phenomics can complement the potential of new molecular/genotyping technologies, and together with agronomy and plant breeding efforts would be a real contribution to develop new strategies to help mitigate the impact of climate change in agriculture. There are major opportunities for phenomics in LAC, not only because it has been adopted in isolated initiatives, but also as worldwide development has focused mainly on grain breeding programs. LAC researchers have identified the need to collaborate to exploit the opportunities and gathered together to organize the Latin American Plant Phenomics Network (LatPPN). Currently, LatPPN has prioritized the work on several fronts to consolidate the network (e.g., grant application to CYTED and Procisur, LatPPN\'s second meeting in April 2016 (Balcarce, Argentina) and planning a second regional conference organized by EMBRAPA during 2017, drafting of LatPPN\'s survey, drafting of LatPPN\'s white paper, and construction of LatPPN\'s webpage). What follows next is the development of strategies leading to the sustainability of the network. We are aware of the work ahead of us and know that the collaboration within LatPPN members and with other networks will be crucial to build on the foundations laid. Author contributions {#s6} ==================== Both authors contributed equally. 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. Our special gratitude to Dr. Carolina Saint Pierre (Wheat Phenotyping Coordinator at CIMMYT---Mexico) for valuable information and discussion about Latin American reality and challenges, and to Dr. Ivan Matus (INIA---Chile) for technical definitions and valuable discussion about national wheat breeding program. In Chile, this activity was supported by Universidad de Talca (research programs "Adaptation of Agriculture to Climate Change (A2C2)" and "Núcleo Científico Multidisciplinario"), and the National Commission for Scientific and Technological Research CONICYT-CHILE (FONDEF IDEA 14I10106). In the U.K., this work was supported by the "National Capability for Crop Phenotyping" grant. Award number BB/J004464/1. [^1]: Edited by: Susana Araújo, Universidade Nova de Lisboa, Portugal [^2]: Reviewed by: Sebastien Carpentier, KU Leuven, Belgium; Biswapriya Biswavas Misra, Texas Biomedical Research Institute, USA [^3]: This article was submitted to Crop Science and Horticulture, a section of the journal Frontiers in Plant Science

I. INTRODUCTION {#acm20050-sec-0001} =============== Cancer is one of the leading causes of death throughout the entire world. In the US, half of men and one‐third of women will develop cancers during their lifetimes and about half of them will receive radiation treatments.[^1^](#acm20050-bib-0001){ref-type="ref"}, [^2^](#acm20050-bib-0002){ref-type="ref"}, [^3^](#acm20050-bib-0003){ref-type="ref"} RT cancer treatments have developed from relatively simple processes into very complex procedures during the past two decades.[^(4)^](#acm20050-bib-0004){ref-type="ref"} The ViewRay System (VRS) (i.e., MRIdian system;ViewRay Inc., Oakwood Village, OH) is an integrated magnetic resonance (MR) image‐guided radiation therapy system designed to provide simultaneous MR imaging (MRI) and external‐beam radiation therapy (EBRT) treatment.[^(5)^](#acm20050-bib-0005){ref-type="ref"} The first commercial ViewRay system was installed in our institution.[^(6)^](#acm20050-bib-0006){ref-type="ref"} It has three cobalt‐60 treatment heads, 120° apart, and each head provides a nominal dose rate of 1.85 Gy / min at isocenter.[^7^](#acm20050-bib-0007){ref-type="ref"}, [^8^](#acm20050-bib-0008){ref-type="ref"}, [^9^](#acm20050-bib-0009){ref-type="ref"} Generally, the patient treatment time using ViewRay system consists of patient setup time, image acquisition time, image registration time, treatment delivery time, and finishing‐off time. The daily image acquisition time includes the time for pilot scanning, reviewing pilot, and high resolution scanning, and the "finishing‐off" time includes the time for finishing the treatment, submitting delivery record to MOSAIQ (IMPAC Medical Systems Inc., Sunnyvale, CA), and helping patient to exit the room. Based on the statistics recorded at our clinic, [Table 1](#acm20050-tbl-0001){ref-type="table-wrap"} lists the calculated averaged values and standard deviations of each time. The main motivation of this paper is to predict the total treatment delivery time at the completion of or during treatment planning in order to effectively schedule patient treatment appointments. VRS total treatment deliver time is a complex function of source decay, the order of beam deliveries, the gantry and MLC motions, and the final gantry position of the previous delivery. It is very difficult to accurately predict the total treatment time for the following reasons: (i) According to our clinical delivery records, the total treatment delivery time varies from approximately 3.28 min to 44.98 min, shown in [Fig. 1](#acm20050-fig-0001){ref-type="fig"}, weakly correlated to the prescription, and the total numbers of beams and segments (a total 120 treatment delivery log files were used and analyzed). (ii) There are three cobalt‐60 treatment heads to allow simultaneous beam delivery. The total beam‐on time attributing to the total treatment time is decided by the last finished beam at one of three treatment heads during each gantry position simultaneous delivery. (iii) Cobalt‐60 sources decay. (iv) The VRS reorders the planned gantry positions according to the current gantry angle left from the previous patient treatment delivery so that VRS rotates its gantry only in one direction to finish the whole treatment without rotating back and forth. (v) The time spent for initialization operations before the first beam‐on depends on various random situations. ###### Mean values and standard deviations (SDs) of the patient setup time, imaging and image registration time, treatment delivery time, and finishing‐off time (in min) *Patient Setup* *Imaging* *Image Registration* *Treatment Delivery* *Finishing‐off* ----------------- ----------------- ---------------------- ---------------------- ----------------- $4.95 \pm 2.78$ $6.75 \pm 1.66$ $4.82 \pm 2.31$ $15.19 \pm 6.54$ $2.89 \pm 2.06$ {#acm20050-fig-0001} Due to these complex reasons, in this paper we aim to: 1) systematically model the VRS treatment delivery, 2) statistically learn the unknown modeling parameters and constants from the recorded delivery log files, 3) predict the total treatment delivery time with acceptable accuracy based on the model after verification, and 4) aid and optimize the plan complexity using the predicted treatment time. The accuracy and feasibility of this prediction tool have been demonstrated to be a great improvement against a previous developed Microsoft Excel‐based prediction tool in our clinic. In particular, this Excel‐based tool does not support the updated source strengths after a recent cobalt source exchange, while our method considers the source decay carefully, which makes it more advantageous. Furthermore, the proposed method also indicates the treatment plan complexity, if necessary, to suggest that the treatment plan shall be improved to reduce the total treatment delivery time. This could be useful for future optimization study on the ViewRay treatment planning stage. There are few similar studies reported in the literature. The most relevant published studies were about MLC leaf sequence optimization for step‐and‐shoot IMRT plans in the treatment planning stage.[^10^](#acm20050-bib-0010){ref-type="ref"}, [^11^](#acm20050-bib-0011){ref-type="ref"}, [^12^](#acm20050-bib-0012){ref-type="ref"}, [^13^](#acm20050-bib-0013){ref-type="ref"} Comparisons of dynamic and segmental MLCs (i.e., DMLC and SMLC, respectively), were reported and discussed involving total delivery time for the purpose of deciding which method to use.[^(14)^](#acm20050-bib-0014){ref-type="ref"} The proposed work shares similar mathematics as used in previous works in computation of MLC leaf motion time from the MLC motion distance and motion speed. While considerations of gantry rotation time and treatment initialization time are important additional components in this work, the most important contributions of this work are about the MRIdian‐specific components, including considerations and modeling of source decay, the MLC motion overall and interleaf delay times, and simultaneous beam deliveries with three treatment heads. II. MATERIALS AND METHODS {#acm20050-sec-0002} ========================= A. The ViewRay treatment delivery procedure {#acm20050-sec-0003} ------------------------------------------- The VRS delivery system has three cobalt‐60 treatment heads, 120° apart, with each providing a nominal dose rate, 1.85 Gy / min, at new‐strength installation, however decaying at a half‐life time of 5.25 yrs. The three heads together provide a total dose rate comparable to that of conventional linear accelerator (linac) using simultaneous delivery. Treatment plans are created in the ViewRay treatment planning system (TPS). Each plan contains multiple treatment beam groups (i.e., gantry positions) with each containing one to three beams. Beams belonging to the same beam groups have gantry angles 120° apart and therefore could be delivered simultaneously by three treatment heads. Each beam contains one or multiple segments. Each segment is defined by a MLC formed beam aperture and a beam‐on time. There are, totally, 60 MLC leaves in 30 pairs. As one can see in [Fig. 2](#acm20050-fig-0002){ref-type="fig"}, besides that three treatment heads are at 120° apart, each head also has a limited angle range. On the treatment day, after patient is set up on the couch table, VRS will rearrange the order of the beam groups so that they will be delivered as the gantry goes either clockwise or counterclockwise. This is to avoid rotating the gantry back and forth and therefore to minimize the total treatment delivery time. The rotation direction is selected so that the starting gantry angle is as close as possible to the current gantry angle left from the last treatment delivery. VRS then rotates the gantry to the starting position, and moves the MLC leaves from open‐field positions to the planned MLC positions of the first beam segments. In VRS, beams are delivered in the step‐and‐shoot way. Radiation will be turned on for delivering one beam segment at a time. Between segments, the radiation will be turned off (the cobalt source will be moved to the off position) and the MLC leaves will move to the next position. It is important to note that the MLC leaves are not moving simultaneously due to the limitations in the software and hardware systems. The MLC control system can send only one motion command message to one leaf at a time. This leads to an interleaf motion delay. First, the MLC control system will determine which side of the leaves pairs will be moving away from the opposite side and start moving leaves on this side one by one. Once those leaves on the beginning side are all moving then the control system will move the leaves on the other side one by one. Beams belonging to a beam group are delivered simultaneously and independently by the three beam heads. After deliveries of one beam group, the VRS rotates the gantry to the next position. While rotating, MLC leaves in each head move to the next planned positions. After the gantry reaches the next position, the next group of beams will be simultaneously delivered in the same step‐and‐shoot way. {#acm20050-fig-0002} B. Prediction method {#acm20050-sec-0004} -------------------- ### B.1 The scheme of our algorithm {#acm20050-sec-0005} With the understanding of the treatment delivery procedure, we model the total treatment delivery time in four components: 1) the treatment initialization time; 2) the total beam‐on time; 3) the gantry rotation time; and 4) the MLC motion time. The gantry rotation time is calculated per beam group. The beam‐on time and MLC motion time are calculated per beam segment for each beam, and then processed per beam group. [Figure 3](#acm20050-fig-0003){ref-type="fig"} illustrates our prediction algorithm, which requires only the treatment plan data file, and each time component is predicted separately. In this study, the patient treatment delivery log files and the treatment plan data files are collected from the beam delivery and the planning stages respectively.[^(4)^](#acm20050-bib-0004){ref-type="ref"} A total 120 log files were used and formed into a log file database. Computer codes were developed in MATLAB (The MathWorks, Inc., Natick, MA) to process these data files. {#acm20050-fig-0003} ### B.2 Predicting the initialization time {#acm20050-sec-0006} The initialization time includes the time required for the gantry to rotate to the first position. This time term is rather random since it depends on the last gantry position of the last finished treatment. To estimate this component, we use the averaged value of the treatment initialization time $(T_{\textit{initial}\_\textit{log}})$ extracted from the treatment delivery log file database as:$$T_{Initial} = mean(T_{initial\_\textit{log}})$$ ### B.3 Calculating the beam‐on time {#acm20050-sec-0007} In order to predict the total beam‐on time, the actual dose rate must be determined. In VRS, the dose rates of the treatment heads decrease at half‐life (HL) time of 5.25 yrs and can be calculated as:$$DR(j) = DR_{ref}(j) \times \left( \frac{1}{2} \right)^{ndays/HL},j = 1,2,3$$where *DR(j)* denotes the dose rate of the treatment head *j* on the treatment day, $DR_{\textit{ref}}(j)$ denotes the reference dose rate of the treatment head *j* on the source strength calibration date (i.e., the reference date), and *ndays* represents the number of days between the treatment delivery day and the reference date. The beam‐on time for each beam can be calculated as:$$T_{beam - on}(b) = \frac{T_{planned - beam - on}(b) \times 1.85\, Gy/\textit{min}}{DR(j)}$$where $T_{\textit{beam} - \textit{on}}(b)$ is the calculated beam‐on time for all segments of the beam *b* with the treatment head *j*, and $T_{\textit{planned} - \textit{beam} - \textit{on}}(b)$ is the planned beam‐on time for the same beam based on the nominal dose rate 1.85 Gy / min used in the treatment planning stage.$$T_{beam - on}(b) = \frac{T_{planned - beam - on}(b) \times 1.85\, Gy/\textit{min}}{DR(j)}$$ ### B.4 Predicting the gantry rotation time {#acm20050-sec-0008} The prediction for the gantry rotation time is achieved using linear regression. The relationship between the rotation angle (i.e., the gantry angle difference between a beam group and the next beam group) and the corresponding rotation time can be modeled with a linear regression model:$$T_{G}(g) = \beta_{0} + \beta_{1} \cdot \theta_{diff}(g)$$where $T_{G}(g)$ and $\theta_{\textit{diff}}(g)$ are the gantry rotation time and the gantry angle difference between the beam groups *g* and $g + 1$, respectively, and $\beta_{0}$ and $\beta_{1}$ are the linear regression fitting parameters. The data points obtained from the log file database were used to compute $\beta_{0}$ and $\beta_{1}$. The gantry rotation time $T_{G}(g)$ in the previous treatment deliveries were obtained by computing the time between the beam‐off time of a beam group and the beam‐on time of the next gantry position. [Figure 4](#acm20050-fig-0004){ref-type="fig"} shows that the gantry angle differences (x‐axis) and the gantry rotation times (y‐axis) fit well linearly. The coefficient of determination (i.e., $R^{2}$), described as one measurement of goodness of the linear regression, is calculated to be 0.9710, which demonstrate that our linear regression predicts with the accuracy 97.1% of the variance on the gantry rotation time. Finally, to predict the total gantry rotation time $T_{G,\text{total}}$ for the given plan, 1) the beam groups are reordered so that the gantry angle positions are in the ascending/descending order, 2) $T_{G,\text{total}}$ is calculated by summing all the predicted gantry rotation time in between the reordered beam groups:$$T_{G,total} = \sum_{g = 1}^{Ng - 1}T_{G}(g)$$where $N_{g}$ is the total number of beam groups and there are a total $N_{g} - 1$ gantry rotations. It is worth noting that the beam groups in treatment deliveries would be reordered by the treatment machine in either ascending or descending order to ensure that the gantry rotates always in one direction. The gantry rotation direction will not affect the predicted gantry rotation time because the angle differences between the beam groups do not change regardless of the gantry rotation direction. {#acm20050-fig-0004} ### B.5 Predicting the MLC motion time {#acm20050-sec-0009} The MLC motion time, defined as the time for the MLC leaves to move from the position of a beam segment to the position of the next beam segment, is complicated and dependent on multiple factors including the control system delays, the MLC leaf motion speed, the MLC leaf positions and the numbers of MLC leaves in motion. Based on our communication with ViewRay engineers and our own observations of the MLC motions displayed on the treatment machine console during the treatment deliveries, we model the MLC motion using the following three terms: The overall delay $T_{\text{MLC} - \text{overall} - \text{delay}}$: defined as a) the delay after beam‐off for the previous segment and before MLC leaves start to move toward the next positions, and b) the delay after the MLC leaves arrive the target position and before the beam‐on for the next beam segment. This overall delay is due to both the cobalt sources' end effect and the delays in VRS control system.The interleaf delay $T_{\text{inter} - \text{leaf} - \text{delay}}$: defined as the time delay between one leaf and the next leaf starting to move. The delay is due to the MLC motion control system\'s being only able to send the command to one leaf at a time. There are two additional important notes: a) the MLC leaves on one side are firstly moving away from the opposite leaves one by one, and only after all the MLC leaves on one side are moving will the leaves on the opposite side start to move; b) only the MLC leaves changing positions between beam segments are involved. The MLC leaves that are not changing their positions are skipped by the inter‐MLC‐leaf delay process and therefore ignored by the MLC motion prediction calculation.The leaf motion time $T_{\text{leaf} - \text{motion}}$: defined as the time for a leaf to move from the current position to the next position. Given these three time terms, the motion time for a single MLC leaf is predicted as:$$T_{MLC,i} = T_{mlc - overall - delay} + i \cdot T_{\textit{int}er - leaf - delay} + \frac{d_{motion,i}}{v_{leaf}}$$where *i* is the MLC leaf number and $i = 1$ to 60, $d_{\textit{motion},i}$ is the motion distance (cm) of the leaf *i*, and $v_{\textit{leaf}}$ is the MLC leaf motion speed (cm/s). The MLC motion time for the beam *b* and segment s, denoted as $T_{\textit{MLC}}(b,s)$, can be obtained by calculating the maximum values of $T_{\textit{MLC},i}$ for all 60 MLC leaves for this beam segment, ignoring the leaves not moving, as:$$T_{MLC}(b,s) = \textit{max}(T_{MLC,i}),i = 1\, to\, 60$$For each beam, the MLC motion time can be calculated by summing up the MLC motion time per segment:$$T_{MLC}(b) = {\sum_{s = 1}^{N_{s} - 1}{T_{MLC}(b,s)}}$$where $N_{s}$ is the number of beam segments for this beam. It is important to note that 1) there are only $N_{s} - 1$ MLC motions between the $N_{s}$ beam segments, and 2), as we have observed at the treatment machine console, the MLC apertures of the first beam segments are formed during the gantry rotation and before the gantry rotation finishes. Thus the time for MLC leaves moving to the first beam segment positions is already covered in the gantry rotation time. The MLC motion model parameters were established by analyzing the screen capture videos recorded on the treatment machine console during treatment deliveries. $T_{\text{MLC} - \text{overall} - \text{delay}}$ was estimated to be approximately 2.2 s. $T_{\text{inter} - \text{leaf} - \text{delay}}$ was estimated to be 0.034 s. The average speed of the leaf motion was estimated to be approximately 2.2 cm / s. ### B.6 Predicting the total treatment delivery time {#acm20050-sec-0010} Due to simultaneous deliveries with three treatment heads for each beam group, in order to obtain the total treatment time, it is important to determine which beam within the beam group finishes last. The last finished beam in the beam group *g*, denoted as $b_{g}$, is determined by the maximum sum of the beam‐on time and beam MLC motion time among (up to) three beams in the same beam group. The total treatment delivery time can be obtained as follows:$$T_{total} = T_{Initial} + T_{G,total} + \sum_{g = 1}^{Ng}T_{beam - on}(b_{g}) + \sum_{g = 1}^{Ng}T_{MLC}(b_{g})$$where $N_{g}$ is the total number of beam groups. ### B.7 Verification of the prediction models with simulation {#acm20050-sec-0011} Because the prediction models we have developed in this study, especially the MLC motion model, are relatively complicated, it is very important to verify these models before applying them into prediction. Therefore we developed a computer program in MATLAB to graphically simulate the entire treatment delivery. This simulated graphic presentation was visually compared to the recorded videos on the treatment machine console to qualitatively verify the prediction models. [Figure 5](#acm20050-fig-0005){ref-type="fig"} shows a snapshot of one simulated treatment delivery. The MLC motions in three treatment heads are illustrated separately and independently in three subfigures. The title of each subfigure describes the treatment delivery status information for the corresponding treatment head index, beam number, segment number, and gantry angle. In this screen capture, beams are off at head 1 and 2, and the MLC leaves are moving towards to the next positions (the segment indices are arbitrarily set to zero in this case). Treatment head 3 is currently delivering the ninth beam and its fifth segment. The MLC aperture in white indicates that beam is on. {#acm20050-fig-0005} III. RESULTS {#acm20050-sec-0012} ============ In our simulated prediction model of the VRS treatment time, various modeling parameters were proposed and applied for predictions, which are listed in [Table 2](#acm20050-tbl-0002){ref-type="table-wrap"} with corresponding values. A total of 120 log files and their corresponding plan files were used to verify the overall prediction accuracy. The mean values and the standard deviations of each time component and total delivery time, as well as the prediction errors, for all the analyzed log files, are listed in [Table 3](#acm20050-tbl-0003){ref-type="table-wrap"}. The overall treatment delivery time prediction error is $0.21 \pm 0.28\,\text{minutes}$. [Table 4](#acm20050-tbl-0004){ref-type="table-wrap"} presents a comparison of actual and predicted treatment time for one specific patient example. By calculating the difference between the treatment delivery times predicted based on the patient treatment plan data file and the actual treatment delivery time extracted from the treatment delivery log file for the same treatment plan, the prediction errors are $0.21 \pm 0.28\,\text{minutes}$. Significantly, the mean error percentage, i.e., the percentage of the absolute prediction, is 1.82%, which is far less than the prediction error using the Microsoft Excel‐based predictor (15.75%). The max prediction error is 0.49 minutes and its corresponding maximum error percentage is equal to 7.88%, which is also a great improvement from the maximum error seen using the previous Excel‐based predictor (24.85%). It is worth noting that this prediction error includes the errors arising from a random factor, such as the initialization time. If the initialization time is eliminated from the total delivery time, the maximum prediction error of our method is 0.36 minutes. ###### Summary of the prediction parameters used in our simulation model *Modeling Parameters* *Value* -------------------------------------------------- -------------- $T_{\text{initial}}$ 27.7 (s) $\lbrack\beta_{0},\beta_{1}\rbrack$ (3.61, 0.24) $T_{\text{MLC} - \text{overall} - \text{delay}}$ 2.2 (s) $T_{\text{inter} - \text{leaf} - \text{delay}}$ 0.034 (s) $v_{\text{leaf}}$ 2.1 (cm/s) ###### Mean values and SDs (min) of four treatment time components *Components* *Mean* *SD* *Prediction Error* $(\textit{mean} \pm \textit{SD})$ ------------------------------- -------- ------- ------------------------------------------------------ $T_{\text{initial}}$ 0.463 0.149 $0.11 \pm 0.15$ $T_{\text{beam} - \text{on}}$ 10.085 7.098 $- 0.03 \pm 0.06$ $T_{G,\text{total}}$ 0.910 0.589 $0.11 \pm 0.19$ $T_{\text{MLC}}$ 2.210 2.025 $- 0.29 \pm 0.24$ $T_{\text{total}}$ 15.19 6.54 $0.21 \pm 0.28$ ###### An illustration of a patient\'s treatment time prediction result (in min) compared to the actual treatment delivery time from the corresponding recorded log file *Components* *Predicted Treatment Time* *Actual Treatment Time* *Prediction Error* ------------------------------- ---------------------------- ------------------------- -------------------- $T_{\text{initial}}$ 0.5757 0.4153 0.1604 $T_{\text{beam} - \text{on}}$ 8.1655 8.1652 0.0003 $T_{G,\text{total}}$ 0.6217 0.6251 $- 0.0034$ $T_{\text{MLC}}$ 5.9433 6.0965 $- 0.1532$ $T_{\text{total}}$ 15.3062 15.3021 0.0041 Based on the proposed prediction algorithm, we have developed a software tool using MATLAB for our dosimetrists and physicists to use for scheduling patient treatment appointments. It requires only the treatment plan data file as the input. Screenshots of this program are presented in [Fig. 6](#acm20050-fig-0006){ref-type="fig"}. One can see that the expected total treatment time is given with important time components listed and duty cycle calculated (defined in Section IV.B). This tool is currently used daily in the clinic. {#acm20050-fig-0006} IV. DISCUSSION {#acm20050-sec-0013} ============== A. Effects of moving the closed MLC pairs on the total MLC motion time {#acm20050-sec-0014} ---------------------------------------------------------------------- Use of the MLC‐shaped beam segments in IMRT requires accurate modeling of the MLC leaf transmission and interleaf leakage. Based on our observations from the recorded treatment delivery videos and the MLC position values in the treatment plans, VRS purposely moves the closed MLC leaf pairs between the beam segments even though these closed MLC leaf pairs do not contribute to the beam aperture definition. The positions of the closed MLC leaf pairs are planned by the TPS in order to avoid the leakage dose of MLC junctions at the fixed position. However, if these closed leaf pairs remained still while only those leaves forming the aperture moved, total MLC motion time could be significantly reduced. While the MLC leaf junction leakage still needs to be carefully considered, one alternative method could be moving these closed MLC junctions per treatment fraction instead of per beam segment. We analyzed the time taken to move these closed pairs for each treatment in the log file database. If the closed MLC leaf pairs do not move, the average saved MLC motion time percentage at each gantry position is 33.40%, shown in [Fig. 7](#acm20050-fig-0007){ref-type="fig"}. {#acm20050-fig-0007} B. Effects of plan complexity {#acm20050-sec-0015} ----------------------------- We have observed that the treatment delivery time for individual treatment plans vary significantly, even among the plans treating the same treatment sites with similar prescription doses. We have therefore performed an analysis on plan complexity in order to understand the relationship between plan complexity and treatment delivery time, with the intention of aiding the treatment planning process. The complexity of treatment plans arises from beam modulation (i.e., some plans require a large number of small and/or irregularly shaped beam segments). Therefore, we have analyzed a few important plan complexity metrics involving the beam modulation (BM) and the plan modulation (PM).[^(15)^](#acm20050-bib-0015){ref-type="ref"} BM can be described as the deviations of the MLC aperture shapes from a circle and BM indicates the extent of a large open field being broken into multiple small segments. PM is denoted as the plan averaged beam modulation obtained by averaging the BM values using the normalized beam‐on time (by 2 Gy nominal prescription dose) as weighing factors. Further details and definitions can be found in Du et al.[^(15)^](#acm20050-bib-0015){ref-type="ref"} The duty cycle and the total treatment time were found to correlate with BM and PM for all the treatment deliveries in the log files database. The duty cycle of each beam can be defined as the percentage of the total beam‐on time in each beam delivery, involving the sum of the beam‐on time and the MLC motion time. Similarly, the overall duty cycle of the plan is defined as the percentage of the total beam‐on time in the total treatment delivery time. [Figure 8](#acm20050-fig-0008){ref-type="fig"} shows the linear correlation between (a) the overall plan duty cycle and PM, and (b) the predicted total treatment delivery time and PM, respectively. We note that a single‐term exponential correlation method was also applied and similar trend was found to support the linear relationship between both the overall plan duty cycle and total treatment time versus the PM values. Only the dominant beam (last‐finished beam head) in each gantry position affects the total duty cycle and the total treatment time, and complexity shown only on such beams is investigated. As can be seen, the correlations indicate that increased complexity will lead to a decreased duty cycle, and more importantly result in a longer treatment delivery time. Understanding of the plan complexity versus the treatment delivery time could be used in a few different ways.[^(16)^](#acm20050-bib-0016){ref-type="ref"} In particular, the treatment delivery time could be predicted using the proposed prediction algorithm and the predicted treatment time could be compared against the mean values of previous treatment plans of the same treatment site. If the predicted treatment time of the new plan is much longer than expected, the complexity on the dominant beams of this plan could be further optimized in order to reduce the treatment delivery time. If it is not practical to control the plan complexity on the individual beam, the overall plan complexity could be controlled. {#acm20050-fig-0008} V. CONCLUSIONS {#acm20050-sec-0016} ============== In this paper, we presented a technical note for the ViewRay treatment time prediction. By simulating the VRS treatment delivery, prediction models were developed for each treatment time component. The proposed method provides accurate treatment delivery time prediction. It is useful in scheduling patient treatment appointments, and for managing plan complexity at the treatment‐planning stage. ACKNOWLEDGMENTS {#acm20050-sec-0017} =============== Research reported in this study was partially supported by the Agency for Healthcare Research and Quality (AHRQ) under award 1R01HS0222888, and partially supported by a research grant from ViewRay Incorporated. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency of Healthcare Research and Quality. COPYRIGHT {#acm20050-sec-0018} ========= This work is licensed under a [Creative Commons Attribution 4.0 International License.]{.ul}

**Suggested citation:** EFSA FEEDAP Panel (EFSA Panel on Additives and Products or Substances used in Animal Feed) , Bampidis V, Azimonti G, Bastos ML, Christensen H, Dusemund B, Kouba M, Kos Durjava M, López‐Alonso M, López Puente S, Marcon F, Mayo B, Pechová A, Petkova M, Ramos F, Sanz Y, Villa R, Woutersen R, Chesson A, Cocconcelli PS, Glandorf B, Herman L, Prieto M, Wallace RJ, Rychen G, Brozzi R and Saarela M, 2019 Scientific Opinion on the safety and efficacy of Cinergy^®^ Life B3 HiCon (*Bacillus amyloliquefaciens* NRRL B‐50508, *B*. *amyloliquefaciens* NRRL B‐50509 and *Bacillus subtilis NRRL* B‐50510) as a feed additive for pigs for fattening and minor porcine species. EFSA Journal 2019;17(3):5647, 12 pp. 10.2903/j.efsa.2019.5647 **Requestor:** the European Commission **Question number:** EFSA‐Q‐2017‐00746 **Panel members:** Giovanna Azimonti, Vasileios Bampidis, Maria de Lourdes Bastos, Henrik Christensen, Birgit Dusemund, Maryline Kouba, Mojca Kos Durjava, Marta López‐Alonso, Secundino López Puente, Francesca Marcon, Baltasar Mayo, Alena Pechová, Mariana Petkova, Fernando Ramos, Yolanda Sanz, Roberto Edoardo Villa and Ruud Woutersen. **Legal notice:** Relevant information or parts of this scientific output have been blackened in accordance with the confidentiality requests formulated by the applicant pending a decision thereon by the European Commission. The full output has been shared with the European Commission, EU Member States and the applicant. The blackening will be subject to review once the decision on the confidentiality requests is adopted by the European Commission. **Acknowledgements:** The EFSA FEEDAP Panel (EFSA Panel on Additives and Products or Substances used in Animal Feed) wishes to thank the following for the support provided to this scientific output: Jaume Galobart, Matteo Lorenzo Innocenti. Adopted: 27 February 2019 1. Introduction {#efs25647-sec-0002} =============== 1.1. Background and Terms of Reference {#efs25647-sec-0003} -------------------------------------- Regulation (EC) No 1831/2003[1](#efs25647-note-1007){ref-type="fn"} establishes the rules governing the Community authorisation of additives for use in animal nutrition. In particular, Article 4(1) of that Regulation lays down that any person seeking authorisation for a feed additive or for a new use of a feed additive shall submit an application in accordance with Article 7. The European Commission received a request from Cargill Incorporated, represented By Provimi Holding BV[2](#efs25647-note-1008){ref-type="fn"} for authorisation of the product Cinergy^®^ Life B3 HiCon (*Bacillus amyloliquefaciens* NRRL B‐50508, *B*. *amyloliquefaciens* NRRL B‐50509 and *Bacillus subtilis* NRRL B‐50510) when used as a feed additive for pigs for fattening and minor porcine species (category: zootechnical additive; functional group: gut flora stabiliser). According to Article 7(1) of Regulation (EC) No 1831/2003, the Commission forwarded the application to the European Food Safety Authority (EFSA) as an application under Article 4(1) (authorisation of a feed additive or new use of a feed additive). The particulars and documents in support of the application were considered valid by EFSA as of 8 January 2018. According to Article 8 of Regulation (EC) No 1831/2003, EFSA, after verifying the particulars and documents submitted by the applicant, shall undertake an assessment in order to determine whether the feed additive complies with the conditions laid down in Article 5. EFSA shall deliver an opinion on the safety for the target animals, consumer, user and the environment and on the efficacy of the product Cinergy^®^ Life B3 HiCon (*Bacillus amyloliquefaciens* NRRL B‐50508, *B*. *amyloliquefaciens* NRRL B‐50509 and *Bacillus subtilis* NRRL B‐50510), when used under the proposed conditions of use (see Section [3.1.4](#efs25647-sec-0013){ref-type="sec"}). 1.2. Additional information {#efs25647-sec-0004} --------------------------- The additive Cinergy^®^ Life B3 HiCon has not been previously authorised as a feed additive in the EU. 2. Data and methodologies {#efs25647-sec-0005} ========================= 2.1. Data {#efs25647-sec-0006} --------- The present assessment is based on data submitted by the applicant in the form of a technical dossier[3](#efs25647-note-1009){ref-type="fn"} in support of the authorisation request for the use of active agent as a feed additive. The technical dossier was prepared following the provisions of Article 7 of Regulation (EC) No 1831/2003, Regulation (EC) No 429/2008[4](#efs25647-note-1010){ref-type="fn"} and the applicable EFSA guidance documents. EFSA has verified the European Union Reference Laboratory (EURL) report as it relates to the methods used for the control of the active agent in animal feed. The Executive Summary of the EURL report can be found in Annex [A](#efs25647-sec-1001){ref-type="sec"}.[5](#efs25647-note-1011){ref-type="fn"} 2.2. Methodologies {#efs25647-sec-0007} ------------------ The approach followed by the FEEDAP Panel to assess the safety and the efficacy of Cinergy^®^ Life B3 HiCon is in line with the principles laid down in Regulation (EC) No 429/2008 and the relevant guidance documents: Guidance on zootechnical additives (EFSA FEEDAP Panel, [2012a](#efs25647-bib-0002){ref-type="ref"}), Technical guidance: Tolerance and efficacy studies in target animals (EFSA FEEDAP Panel, [2011](#efs25647-bib-0800){ref-type="ref"}), Guidance on studies concerning the safety of use of the additive for users/workers (EFSA FEEDAP Panel, [2012b](#efs25647-bib-0003){ref-type="ref"}), Guidance on the assessment of the toxigenic potential of *Bacillus* species used in animal nutrition (EFSA FEEDAP Panel, [2014](#efs25647-bib-0005){ref-type="ref"}), Guidance on the assessment of bacterial susceptibility to antimicrobials of human and veterinary importance (EFSA FEEDAP Panel, [2012c](#efs25647-bib-0004){ref-type="ref"}) and Guidance on the characterisation of microorganisms used as feed additives or as production organisms (EFSA FEEDAP Panel, [2018](#efs25647-bib-0006){ref-type="ref"}). 3. Assessment {#efs25647-sec-0008} ============= The additive is a preparation containing viable spores of two strains of *B*. *amyloliquefaciens* and a single strain of *B. subtilis* intended to be used in feed for pigs for fattening and minor porcine species as a zootechnical additive to improve performance. 3.1. Characterisation {#efs25647-sec-0009} --------------------- ### 3.1.1. Characterisation of the active agents {#efs25647-sec-0010} ■■■■■[■■■■■](#efs25647-note-5006){ref-type="fn"} All strains are deposited with the United States Department of Agriculture Agricultural Research Culture Collection (NRRL Collection) with the accession numbers NRRL B‐50508 and NRRL 50509 for the two *B*. *amyloliquefaciens* strains and NRRL B‐50510 for the *B*. *subtilis*.[7](#efs25647-note-1012){ref-type="fn"} None of the strains have been genetically modified.[8](#efs25647-note-1013){ref-type="fn"} ■■■■■[■■■■■](#efs25647-note-5009){ref-type="fn"} Randomly amplified polymorphic DNA (RAPD) PCR genotyping analysis supported the differentiation between the additive strains and between other similar strains. ■■■■■[■■■■■](#efs25647-note-5010){ref-type="fn"}■■■■■ ■■■■■[■■■■■](#efs25647-note-5011){ref-type="fn"}■■■■■ No cytotoxic effects were detected in any of the strains. ■■■■■[■■■■■](#efs25647-note-5012){ref-type="fn"}■■■■■[■■■■■](#efs25647-note-5013){ref-type="fn"}■■■■■(Ohmiya et al., [1989](#efs25647-bib-0008){ref-type="ref"}; Kunst et al., [1997](#efs25647-bib-0007){ref-type="ref"})■■■■■ ### 3.1.2. Characterisation of the additive {#efs25647-sec-0011} ■■■■■[■■■■■](#efs25647-note-5014){ref-type="fn"}■■■■■ The final product is a powder with a minimum specified count of *Bacillus* spp. is 2.5 × 10^9^ CFU/g additive with the three strains contributing approximately in equal numbers. Compliance with this specification was demonstrated in five batches of the additive. The counts given for the individual strains are the same for all batches ■■■■■[■■■■■](#efs25647-note-5015){ref-type="fn"} Three batches of the additive were analysed for heavy metals and arsenic, dioxins and dioxin‐like polychlorinated biphenyls (PCBs) and mycotoxins. Lead, cadmium and arsenic were detected at mean concentrations of 1.8, 0.16 and 0.31 mg/kg additive respectively, while mercury was not detected (limit of quantification (LOQ) \< 0.01 mg/kg additive).[16](#efs25647-note-1014){ref-type="fn"} The mean upper bound value for the sum of dioxins and dioxin‐like PCBs was 0.09 ng WHO‐PCDD/F‐PCB‐TEQ/kg additive. These values are substantially below the maximum levels permitted in feed materials and do not raise concerns. None of the mycotoxins included in the analysis could be detected (LOQ total aflatoxins \< 0.05 μg/kg, total fumonisins \< 30 μg/kg, total HT‐2 Toxin \< 50 μg/kg, total T‐2 toxin \< 5 μg/kg zearalenone \< 25 μg/kg, ochratoxin \< 5 μg/kg and vomitoxin \< 50 μg/kg). The same three batches were examined for evidence of microbial contamination. Coliforms, *Escherichia coli*,*Bacillus cereus* and *Staphylococcus aureus* were below the detection limits (\< 10 CFU/g additive) and no *Listeria* or *Salmonella* species could be isolated from 25 g or 50 g of the additive, respectively. Sterile culture filtrate obtained from growth of three batches of the additive did not produce zones of clearing in lawn cultures of a number of test bacterial species indicating that the strains do not produce extracellular antibacterial activity.[17](#efs25647-note-1015){ref-type="fn"} Particle size distribution using laser diffraction[18](#efs25647-note-1016){ref-type="fn"} and dusting potential by the Stauber--Heubach method were measured in three batches of additive. The mean particle size was approximately 430 μm with only around 2% (v/v) of particles with diameters \< 100 and 0.7% \< 10 μm. The dusting potential reflected both the particle size distribution and the inclusion of an unspecified mineral oil as binder and gave a mean value of 0.2 g/m^3^.[19](#efs25647-note-1017){ref-type="fn"} ### 3.1.3. Stability and homogeneity {#efs25647-sec-0012} The shelf‐life of the additive was determined by monitoring three batches stored at either 25°C/60% relative humidity (RH) or 40°C/75% RH for a period of 24 months.[20](#efs25647-note-1018){ref-type="fn"} Essentially no reduction in total *Bacillus* count was observed at either temperature. Three batches of the additive were individually mixed into a commercial vitamin‐mineral premix (containing choline chloride) at a concentration of 1 × 10^11^ CFU/kg premix and samples taken and stored in a sealed container for 6 months at 25°C.[21](#efs25647-note-1019){ref-type="fn"} Results showed that numbers of bacilli in the in the vitamin‐mineral premix after 6 months was within ±0.5 log~10~ CFU/g of the time zero count. Stability in complete feed was investigated using three batches of the additive incorporated into a typical mash feed for pigs (maize, barley and soybean) and into a pelleted feed of the same composition (pelleting conditions 95°C for 30 s).[22](#efs25647-note-1020){ref-type="fn"} Samples of the mash and pelleted feed were placed in sealed foil bags and stored for up to 3 months at 25°C. Counts of total bacilli were made at the start and after 1 and 3 months. Essentially no reduction in total counts was seen in either the mash feed or the pelleted feed after pelleting. The effect of pelleting itself, obtained by comparison of the initial counts of the mash and pelleted feed, was small and less than 0.5 log~10~ difference. A total of 10 subsamples were taken from the mash feed after 10 min mixing and analysed for total bacilli counts.[22](#efs25647-note-1020){ref-type="fn"} Based on the ten samples, the coefficient of variation was 1%, demonstrating homogeneous mixing. In the absence of any significant reduction in counts in any of the above studies, it is considered unnecessary to monitor individual strains. ### 3.1.4. Conditions of use {#efs25647-sec-0013} The product is proposed for use in feed for pigs for fattening and minor porcine species at a minimum inclusion level of 1.5 × 10^8^ CFU/kg complete feedingstuffs. 3.2. Safety {#efs25647-sec-0014} ----------- ### 3.2.1. Safety for target animals, consumers and environment {#efs25647-sec-0015} The bacterial species *B*. *subtilis* and *B*. *amyloliquefaciens* are considered by EFSA to be suitable for the qualified presumption of safety (QPS) approach to safety assessment (EFSA, [2007](#efs25647-bib-0801){ref-type="ref"}; EFSA BIOHAZ Panel, [2017](#efs25647-bib-0001){ref-type="ref"}). This approach requires the identity of the strains to be conclusively established and evidence that the strains lack toxigenic potential and do not show acquired antimicrobial resistance determinants for antibiotics of human and veterinary importance. In the view of the FEEDAP Panel, the identity of the three active agents is established and the lack of toxigenic potential confirmed. The two *B*. *amyloliquefaciens* strains do not show resistance to antibiotics of human and veterinary importance, and therefore, are presumed safe for the target species, consumers of products derived from animals fed the additive and the environment. The third strain, *B*. *subtilis* NRRL B‐50510, showed a low level of resistance to streptomycin, for which acquired resistance genes were not identified. Therefore, it also complies with the QPS qualifications and is presumed safe for the target species, consumer and the environment. Since the other components of the additive do not give rise to concerns, Cinergy^®^ Life B3 HiCon is also considered safe for the target species, consumer and the environment. ### 3.2.2. Safety for the user {#efs25647-sec-0016} No studies, other than those relating to dust formation, were provided. In the absence of data no conclusions on skin or eye irritancy or on dermal sensitisation can be made. Owing to the proteinaceous nature of the active agents, the additive is considered a potential respiratory sensitiser. The particle size distribution and dusting potential make it unlikely that additive represents a respiratory hazard. 3.3. Efficacy {#efs25647-sec-0017} ------------- ### 3.3.1. Efficacy for pigs for fattening {#efs25647-sec-0018} Four efficacy studies were performed, two in the same Member State and location and two in a non‐EU country. The first two trials were made at the same location using a common design and the same breed of pig (Table [1](#efs25647-tbl-0001){ref-type="table"}) but separated by a 3‐month period. The 96 animals (equal numbers of male and female) in each trial were distributed according to body weight at the start to a total of 32 pens each containing three animals of the same sex. Sixteen pens (eight male and eight female pens) were allocated to each of the two treatments -- a control group fed the basal diet and a test group given the basal diet supplemented with 1.5 × 10^8^ CFU bacilli/kg feed. All feeds were analysed for the presence of bacilli, but results were confused by a high background count. Animals were fed mash diets containing maize, maize dry distillers grain with solubles (DDGs), wheat middlings and soybean meal in three phases. The duration of the trials was 96 days for trial 1[23](#efs25647-note-1021){ref-type="fn"} and 97 days for trial 2.[24](#efs25647-note-1022){ref-type="fn"} The parameters recorded were body weight and total pen feed intake at days 0, 42 and the end of the trial. From these data average daily weight gain, average daily feed intake and feed to gain ratio were calculated. At the end of the trials all pigs were slaughtered and carcasses evaluated. Data were analysed by analysis of variance (ANOVA) with the pen as the experimental unit. For performance data, the model used included dietary treatment, sex, block and the interaction between treatment and sex as main effects, while for carcass data the model included dietary treatment, sex and the interaction. Group means were compared using the Dunnett\'s test. In the third trial, 120 pigs (Table [1](#efs25647-tbl-0001){ref-type="table"}) were distributed in a randomised block design across 30 pens, with 15 pens (8 of females and 7 of male barrows) per treatment with four pigs in each pen.[25](#efs25647-note-1023){ref-type="fn"} The two treatment groups were a control group given mash feeds in three phases based on maize, maize DDGs and soybean meal and a test group fed the same basal diets supplemented with the additive at 1.5 × 10^8^ CFU/kg diet (confirmed for the test group). The duration of the trial was 111 days. Body weight measurement and feed consumption per pen were made every 3 weeks or on change of diet and from these data average daily gain, average daily feed intake and feed to gain ratio calculated. Four male and four females per treatment were removed from the trial for a digestibility study (not reported) and the data arising from the pens from which pigs were removed (2, 3, 17 and 12) were excluded from the analysis of performance (in the last 2 phases). At the end of the trial 100 pigs were slaughtered and carcass composition assessed. Data were analysed by ANOVA with the pen as the experimental unit for performance measurements. For carcass composition data, the individual pig was used as the experimental unit. For all data, the model included treatment and sex as a fixed effect, and the interaction of treatment × sex. The fourth trial involved 56 pigs (Table [1](#efs25647-tbl-0001){ref-type="table"}) which were distributed in a randomised block design across 14 pens with seven pens per treatment with four pigs (2 female and 2 male barrows) in each pen.[26](#efs25647-note-1024){ref-type="fn"} Pigs were introduced into the trial in three batches at weekly intervals to ensure approximately equal start weights. The two treatment groups were a control group given mash feeds in three phases based on maize, maize DDGs and soybean meal and a test group fed the same basal diets supplemented with the additive at 1.5 × 10^8^ CFU/kg diet (confirmed for the test group). The overall duration of the trial was 109 days but, because of the different start dates, the average duration for individual animals ranged between 84 and 92 days for the control group and 84 and 93 days for the treatment group. It is presumed that individual animals were retained on trial until reaching a body weight of approximately 120 kg. Body weight and feed consumption was recorded at the start, at each dietary interval and at the end of the trial and from these data average daily gain, average daily feed intake and feed to gain ratio calculated. The data were analysed by ANOVA with the pen as the experimental unit. Initial bodyweight was included as a covariate. Group means were compared using the Tukey\'s test. ###### Details on the study design for the studies performed in pigs for fattening +--------------------------+---------------------------------------------------+--------------------------------------------------------+----------------------------------------------+----------------------------------------------+-------------------------------------------------------+ | Study (duration in days) | Breed | Total animals replicates/treatment × animals/replicate | Intended concentration in feed (CFU/kg feed) | Analysed concentration in feed (CFU/kg feed) | Basal diets (main ingredients) form | +==========================+===================================================+========================================================+==============================================+==============================================+=======================================================+ | 1 (96) | Pietrain × (Landrace × Yorkshire) | 96 | 0 | 180/2.1/6.4 × 10^6^ | (maize, maize DDGs, wheat middlings and soybean) mash | | | | | | | | | | | 16 × 3 | 1.5 × 10^8^ | 0.7/0.4/1.4 × 10^8^ | | +--------------------------+---------------------------------------------------+--------------------------------------------------------+----------------------------------------------+----------------------------------------------+-------------------------------------------------------+ | 2 (97) | Pietrain × (Landrace × Yorkshire) | 96 | 0 | 4.0/160/4.0 × 10^6^ | (maize, maize DDGs, wheat middlings and soybean) mash | | | | | | | | | | | 16 × 3 | 1.5 × 10^8^ | 0.6/0.3/0.6 × 10^8^ | | +--------------------------+---------------------------------------------------+--------------------------------------------------------+----------------------------------------------+----------------------------------------------+-------------------------------------------------------+ | 3 (111) | PIC Genetiporc 6.0 × Genetiporc F25 | 120 | 0 | 4.5/8.0/1.3 × 10^6^ | (maize, maize DDGs and soybean meal) mash | | | | | | | | | | | 15 × 4 | 1.5 × 10^8^ | 1.2/1.4/1.0 × 10^8^ | | +--------------------------+---------------------------------------------------+--------------------------------------------------------+----------------------------------------------+----------------------------------------------+-------------------------------------------------------+ | 4 (109) | (Yorkshire × Landrace) × Duroc; Yorkshire × Duroc | 56 | 0 | 6.3/1.2/1.0 × 10^6^ | (maize, maize DDGs and soybean meal) mash | | | | | | | | | | | 7 × 4 | 1.5 × 10^8^ | 0.2/2.0/0.3 × 10^8^ | | +--------------------------+---------------------------------------------------+--------------------------------------------------------+----------------------------------------------+----------------------------------------------+-------------------------------------------------------+ CFU: colony forming unit. John Wiley & Sons, Ltd ###### Effects of the additive Cinergy^®^ Life B3 HiCon on the performance of pigs for fattening Trial no Additive (CFU/kg feed) Initial body weight (kg) Feed intake (kg/day) Final body weight (kg) Daily weight gain (g/day) Feed to gain ratio ------------- ------------------------ --------------------------------------------- --------------------------------------------- ------------------------ --------------------------------------------- --------------------------------------------- 1 0 29.9 2.01 107.5 808.2 2.49[a](#efs25647-note-0009){ref-type="fn"} 1.5 × 10^8^ 29.9 1.96 108.0 813.7 2.41[b](#efs25647-note-0009){ref-type="fn"} 2 0 31.6 1.91 105.5 761.8 2.52[a](#efs25647-note-0009){ref-type="fn"} 1.5 × 10^8^ 31.5 1.89 107.8 778.6 2.43[b](#efs25647-note-0009){ref-type="fn"} 3 0 23.5 2.00 129.1 942.0 2.13 1.5 × 10^8^ 23.1 1.98 128.9 938.0 2.11 4 0 28.9 2.95[a](#efs25647-note-0009){ref-type="fn"} 122.2 1030.0 2.86 1.5 × 10^8^ 28.5 2.75[b](#efs25647-note-0009){ref-type="fn"} 121.9 1030.0 2.68 CFU: colony forming unit. ^a,b^ Means within a column and study with different superscript letters are significantly different at p \< 0.05. John Wiley & Sons, Ltd The Panel notes the abnormal high *Bacilli* counts in one control diet sample in studies 1 and 2. The applicant justified these values by a high background contamination and subjected all feed samples to confirmation of the presence of the active agents. All control feed samples proved negative and the treated feeds proved positive. Positive refers to a sample in which the majority of the colonies have a morphology similar to that of the three added Cinergy Life B3 HiCon *Bacillus* strains. The FEEDAP Panel does not consider this to be of concern. No mortalities occurred in either of the first two trials, but in trial 2 two animals from the Cinergy group had to be withdrawn because of ill‐health. There were no significant differences in growth or feed intake in either trial but, in both cases, there was a significant improvement in feed to gain ratio (Table [2](#efs25647-tbl-0002){ref-type="table"}). No significant effects were seen on carcass composition in trial 1 (liveweight, hot carcass, killing out percentage, % offals, % lean meat). In trial 2, the backfat depth was significantly reduced (15.1 mm vs. 16.7 mm) and lean meat percentage significantly increased (61.5% vs. 59.9%) in the treated group compared to control values. In the third study, two pigs died during the trial and two were removed (one from each treatment group in both cases) because of enteric disease. Overall, there were no significant differences between the control and treated groups in performance (Table [2](#efs25647-tbl-0002){ref-type="table"}) or carcass composition (not reported). In the last study, no mortalities occurred and no animals were removed from the trial. Feed intake was significantly lower in the treated group compared to the control animals without an effect on final body weight (Table [2](#efs25647-tbl-0002){ref-type="table"}). This led to an improved feed to gain ratio, although the difference did not reach significance. Since significant effects were observed in two studies only, the applicant performed an analysis pooling the data shown in Table [2](#efs25647-tbl-0002){ref-type="table"}, including the effect of the treatment, study and their interaction. No interactions were found between treatment and study, and a significant difference was found for feed to gain ratio between the treatments (control 2.50 vs. additive 2.43, p = 0.019).[27](#efs25647-note-1025){ref-type="fn"} #### Conclusions on efficacy {#efs25647-sec-0019} Two of the four individual studies showed a significant improvement in feed to gain ratio in favour of the additive. Pooling of the data from the four studies showed a potential of the additive at a minimum inclusion level of 1.5 × 10^8^ CFU/kg complete feed to improve the feed to gain ratio of pigs for fattening. ### 3.3.2. Efficacy for minor porcine species {#efs25647-sec-0020} Since a potential to improve the feed to gain ratio in pigs for fattening given the additive at the recommended inclusion level of 1.5 × 10^8^ CFU/kg complete feed was shown, the same conclusion can be extrapolated to minor porcine species at the same minimum application rate and for an equivalent growth phase. 3.4. Post‐market monitoring {#efs25647-sec-0021} --------------------------- The FEEDAP Panel considers that there is no need for specific requirements for a post‐market monitoring plan other than those established in the Feed Hygiene Regulation[28](#efs25647-note-1026){ref-type="fn"} and Good Manufacturing Practice. 4. Conclusions {#efs25647-sec-0022} ============== The three active agents comprising the additive (*B*. *amyloliquefaciens* NRRL B‐50508 and B‐50509 and *B*. *subtilis* NRRL B‐50510) meet the requirements of the QPS approach to safety assessment and are presumed safe for the target animals, consumers of products derived from animals fed the additive and the environment. Since the other components of the additive do not give rise to concerns, Cinergy^®^ Life B3 HiCon is also considered safe for the target species, consumer and the environment. In the absence of data, no conclusions on skin or eye irritancy or on the potential for dermal sensitisation can be made. Owing to the proteinaceous nature of the active agents, the additive is considered a potential respiratory sensitiser. However, the low dusting potential makes it unlikely that additive poses a risk for the respiratory system. Cinergy^®^ Life B3 HiCon showed a potential to improve the feed to gain ratio in pigs for fattening at the minimum inclusion level of 1.5 × 10^8^ CFU/kg complete feed. This conclusion is extrapolated to minor porcine species at the same application rate and for an equivalent growth phase. Documentation provided to EFSA {#efs25647-sec-0023} ============================== Cinergy^®^ Life B3 HiCon. *Bacillus amyloliquefaciens* AGTP BS918 (NRRL B‐50508), *Bacillus amyloliquefaciens* AGTP BS1013 (NRRL B‐50509), *Bacillus subtilis* AGTP BS3BP5 (NRRL B‐50510). Zootechnical additive for pigs for fattening and minor porcine species functional group: gut flora stabiliser. October 2017. Submitted by Cargill Incorporated, represented By Provimi Holding BV.Cinergy^®^ Life B3 HiCon. *Bacillus amyloliquefaciens* AGTP BS918 (NRRL B‐50508), *Bacillus amyloliquefaciens* AGTP BS1013 (NRRL B‐50509), *Bacillus subtilis* AGTP BS3BP5 (NRRL B‐50510). Zootechnical additive for pigs for fattening and minor porcine species functional group: gut flora stabiliser. Supplementary information. September 2018. Submitted by Cargill Incorporated, represented By Provimi Holding BV.Evaluation report of the European Union Reference Laboratory for Feed Additives on the Methods(s) of Analysis for Cinergy^®^ Life B3 HiCon.Comments from Member States. Chronology {#efs25647-sec-0024} ========== DateEvent30/10/2017Dossier received by EFSA16/11/2017Reception mandate from the European Commission27/2/2019Application validated by EFSA -- Start of the scientific assessment3/4/2018Reception of the Evaluation report of the European Union Reference Laboratory for Feed Additives8/4/2018Comments received from Member States31/5/2018Request of supplementary information to the applicant in line with Article 8(1)(2) of Regulation (EC) No 1831/2003 -- Scientific assessment suspended. *Issues: characterisation and efficacy*16/7/2018Reception of supplementary information from the applicant -- Scientific assessment re‐started12/9/2018Request of supplementary information to the applicant in line with Article 8(1)(2) of Regulation (EC) No 1831/2003 -- Scientific assessment suspended *Issues: characterisation*7/12/2018Reception of supplementary information from the applicant -- Scientific assessment re‐started27/2/2019Opinion adopted by the FEEDAP Panel. End of the Scientific assessment Abbreviations {#efs25647-sec-0025} ============= ANOVAanalysis of varianceBWbody weightCFUcolony‐forming unitCGchemical groupCVcoefficient of variationDDGsdry distillers grain with solublesEURLEuropean Union Reference LaboratoryLOQlimit of quantificationMICminimum inhibitory concentrationPCBpolychlorinated biphenylPCDD/Fpolychlorinated dibenzo‐*p*‐dioxins and dibenzofuransPCRpolymerase chain‐reactionQPSqualified presumption of safetyRAPDrandomly amplified polymorphic DNARHrelative humidity Annex A -- Executive Summary of the Evaluation Report of the European Union Reference Laboratory for Feed Additives on the Method(s) of Analysis for Cinergy^®^ Life B3 HiCon {#efs25647-sec-1001} ============================================================================================================================================================================= {#efs25647-sec-0026} *Cinergy* ^*®*^ *Life B3 HiCon* is the trade name of a preparation based on viable spores of the three non‐genetically modified strains *Bacillus amyloliquefaciens* AGTP BS918 (NRRL B‐50508), *Bacillus amyloliquefaciens* AGTP BS1013 (NRRL B‐50509) and *Bacillus subtilis* AGTP BS3BP5 (NRRL B‐50510). In the current application, authorisation is sought under Article 4(1) for this product under the category / functional group 4(b) 'zootechnical additives' / 'gut flora stabilisers', according to Annex I of Regulation (EC) No 1831/2003. Authorisation is sought for the use of the *feed additive* for pigs for fattening and minor porcine species. The *feed additive* is to be marketed as a powder containing a minimum content of total active substances of 2.5 × 10^9^ Colony Forming Unit (CFU)/g. The *feed additive* is intended to be used directly in *feedingstuffs* or through *premixtures* at a minimum dose of 1.5 × 10^8^ CFU/kg of complete *feedingstuffs*. For the identification of *Bacillus amyloliquefaciens* AGTP BS918, *Bacillus amyloliquefaciens* AGTP BS1013 and *Bacillus subtilis* AGTP BS3BP5, the EURL recommends for official control Pulsed Field Gel Electrophoresis (PFGE), a generally recognised methodology for genetic identification of bacterial strains. For the enumeration of all *Bacillus* spp. strains (AGTP BS918, AGTP BS1013 and AGTP BS3BP5) in the *feed additive*,*premixtures* and *feedingstuffs* the Applicant submitted the ring‐trial validated spread plate CEN method EN 15784. Based on the performance characteristics available, the EURL recommends this method for official control. Further testing or validation of the methods to be performed through the consortium of National Reference Laboratories as specified by Article 10 (Commission Regulation (EC) No 378/2005) is not considered necessary. Regulation (EC) No 1831/2003 of the European Parliament and of the Council of 22 September 2003 on additives for use in animal nutrition. OJ L 268, 18.10.2003, p. 29. Cargill Incorporated, represented By Provimi Holding BV, Veerlaan 17‐23, 3072 AN, Rotterdam, The Netherlands. FEED dossier reference: FAD‐2017‐0060. Commission Regulation (EC) No 429/2008 of 25 April 2008 on detailed rules for the implementation of Regulation (EC) No 1831/2003 of the European Parliament and of the Council as regards the preparation and the presentation of applications and the assessment and the authorisation of feed additives. OJ L 133, 22.5.2008, p. 1. The full report is available on the EURL website: <https://ec.europa.eu/jrc/en/eurl/feed-additives/evaluation-reports/fad-2017-0060?search&form-return> ■■■■■ Technical dossier/Section II/ Annex II.2.1.2.3. Technical dossier/Section II/ Annex II.2.1.2.1. ■■■■■ ■■■■■ ■■■■■ ■■■■■ ■■■■■ ■■■■■ ■■■■■ Technical dossier/Section II/Annex II.1.4.1.1. Technical dossier/Section II/Annex II.1.4.1.2. Technical dossier/Section II/Annex II.1.5.2. Technical dossier/Section II/Annex II.1.5.3. Technical dossier/Section II/Annex II.4.1.1. Technical dossier/Section II/Annex II.4.1.2. Technical dossier/Section II/Annex II.4.1.3. Technical dossier/Section IV/Annex IV.3.1. Technical dossier/Section IV/Annex IV.3.2. Technical dossier/Section IV/Annex IV.3.3. Technical dossier/Section IV/Annex IV.3.4 and Supplementary information July 18/Annex II. Technical dossier/Section IV/Annex IV.3.5. Regulation (EC) No 183/2005 of the European Parliament and of the Council of 12 January 2005 laying down requirements for feed hygiene. OJ L 35, 8.2.2005, p. 1.

INTRODUCTION {#SEC1} ============ The National Genomics Data Center (NGDC), officially approved by the Ministry of Science & Technology and the Ministry of Finance of the People\'s Republic of China in June 2019, is a national-level center dedicated to advancing life and health sciences by archiving, managing and processing a wide range of genomics related data. NGDC is established based on the BIG Data Center ([@B1]) at Beijing Institute of Genomics (BIG) of Chinese Academy of Sciences (CAS), jointly in close collaboration with two CAS institutions, namely, Institute of Biophysics (IBP) and Shanghai Institute of Nutrition and Health (SINH). Considering the rapid advancements in higher-throughput and lower-cost sequencing technologies, huge amounts of multi-omics data are generated at ever-growing rates and scales. Therefore, the primary mission of NGDC is to build archive platforms and information systems, develop advanced algorithms and tools to translate big data into big discovery, and provide open access to a suite of database resources in support of research activities of global users from both academia and industry. During the past year, NGDC has expanded, updated and enriched the amount and type of data through big data integration and value-added curation, particularly by close collaboration with IBP and SINH, with significant improvements and advances over the previous release. In terms of data attribute and curation intensity, database resources in NGDC can be generally divided into three categories: Data---raw sequence data and metadata, Information---value-added standardized information, and Knowledge---curated knowledge and knowledge graphs. Here, we provide a brief summary of new developments and recent updates, and describe the core resources and services of NGDC (Figure [1](#F1){ref-type="fig"}). All resources, along with their services, are publicly accessible through the home page of NGDC at <https://bigd.big.ac.cn>. {#F1} NEW DEVELOPMENTS {#SEC2} ================ Human genome resources {#SEC2-1} ---------------------- ***PGG*.SNV** (<http://www.pggsnv.org>) ([@B4]) is a human genome database, which gives much higher weight to previously under-investigated indigenous populations in Asia, as these genomes harbor an enormous number of variants that have not been observed in the extensively studied populations of European ancestry. In the current version, *PGG*.SNV archives 265 million single nucleotide variants (SNVs) across 220 147 present-day human genomes and 1018 ancient genomes and estimates their frequencies in 977 diverse populations, including 1009 newly sequenced genomes representing 16 indigenous populations living in unusual environments (e.g. tropical forests and highlands) in East Asia and Southeast Asia. For each variant, *PGG*.SNV provides various approaches to query SNV information and nine types of annotations. In addition, *PGG*.SNV offers users-friendly interfaces for data browsing and search and is equipped with an online tool for estimation of population genetic diversity and evolutionary parameters. ***PGG*.Han** (<http://www.pgghan.org>) (detailed in ([@B5]) in this issue) is a population genome database, which serves as the central repository of genomic data of the Han Chinese Genomes Initiative (Phase I). *PGG*.Han archives whole-genome sequencing or high-density genome-wide SNVs of 114 783 Han Chinese individuals (a.k.a. the Han100K), representing geographical sub-populations covering 33 of the 34 administrative divisions of China, as well as Singapore. *PGG*.Han provides: (i) an interactive interface for visualization of the fine-scale genetic structure of the Han Chinese population; (ii) genome-wide allele frequency of hierarchical sub-populations; (iii) ancestry inference for individual samples and controlling population stratification based on nested ancestry informative marker panels; (iv) a population-structure-aware shared control for genotype--phenotype association studies and (v) a Han-Chinese-specific reference panel for genotype imputation. Computational tools are implemented in *PGG*.Han and an online user-friendly interface is provided for data analysis and visualization. **The Chinese Genomic Variation Database** (CGVD; <https://bigd.big.ac.cn/cgvd>) (detailed in ([@B6]) in this issue) is a genomic variation database for Chinese populations. CGVD is a sub-project of the CAS Precision Medicine Initiative project (CASPMI) ([@B7]), with the aim to establish the CAS professional cohort with whole-genome deep sequencing (25--30×) and build precise reference genomes for different Chinese sub-populations. In comparison with *PGG*.Han, CGVD features high-coverage sequencing data of 991 individuals of the CASPMI cohort and 301 Chinese individuals from the 1000 Genome Project (1KGP). Accordingly, it houses genomic variations of 48.30 million SNVs and 5.77 million small indels; in contrast to dbSNP ([@B8]), 28.49 million (46.67%) SNVs and 2.25 million (31.88%) indels are novel, indicating the advantage of deeper whole-genome sequencing coverage or/and the heterogeneity of genetic background in Chinese populations. Moreover, CGVD provides star-allele frequencies of drug metabolism related genes that are essential for pharmacogenomics studies in CASPMI and 1KGP related populations. It also integrates curated knowledge of genomic variation impacts on drug absorption, distribution, metabolism, excretion and toxicity. GWAS Atlas {#SEC2-2} ---------- GWAS Atlas (<https://bigd.big.ac.cn/gwas>) (detailed in ([@B9]) in this issue) is a manually curated resource of genome-wide variant-trait associations in plants and animals. In the current version, GWAS Atlas contains 75 467 variant-trait associations for 614 traits across seven cultivated plants (cotton, Japanese apricot, maize, rapeseed, rice, sorghum and soybean) and two domesticated animals (goat and pig), which were manually extracted and curated from 254 publications. More importantly, associations and traits are annotated and presented based on a set of ontologies (Plant Trait Ontology, Animal Trait Ontology for Livestock, etc.). Taken together, GWAS Atlas integrates high-quality curated GWAS associations for animals and plants and accordingly serves as a valuable resource for genetic research of important traits and breeding application. EWAS Data Hub {#SEC2-3} ------------- Over the past decade, a large amount of epigenetic data, especially those sourced from DNA methylation array, has been accumulated as a result of numerous EWAS (epigenome-wide association study) projects. Hence, we present EWAS Data Hub (<https://bigd.big.ac.cn/ewas/datahub>) (detailed in ([@B10]) in this issue), a data hub for collecting and normalizing DNA methylation array data as well as archiving associated metadata. The current release of EWAS Data Hub integrates a comprehensive collection of DNA methylation array data from 75 344 samples. Based on an effective normalization method to remove batch effects among different datasets, EWAS Data Hub provides high-quality reference DNA methylation profiles in terms of different contexts, involving 81 tissues/cell types (that contain 25 brain parts and 25 blood cell types), six ancestry categories, and 67 diseases (including 39 cancers). iSheep {#SEC2-4} ------ iSheep (<https://bigd.big.ac.cn/isheep>) is a specialized genomics resource for sheep (*Ovis aries*), providing a wealth of information on genotype and phenotype association, domestication and climatic adaptation of domestic sheep as well as their wild relatives. The current version of iSheep houses 70 390 968 unique SNPs and 12 318 530 indels obtained from 2777 samples (including 355 samples with whole-genome sequences, 1512 samples with 50K-BeadChip and 911 samples with 600K-BeadChip) and provides comprehensive phenotypic information of 1459 worldwide sheep breeds. Meanwhile, iSheep offers an online tool to investigate the variations between individuals or among populations. Collectively, iSheep is a valuable genomics resource for the sheep research community, helpful to promote molecular breeding and farming industry for improved production traits. eLMSG {#SEC2-5} ----- eLMSG (eLibrary of Microbial Systematics and Genomics; <http://www.biosino.org/elmsg>) is a web microbial library that integrates not only taxonomic information, but also genomic information and phenotypic information (including morphology, physiology, biochemistry and enzymology). The taxonomic system of eLMSG is manually curated and composed of all validly and some effectively published taxa. For each taxon, the Latin name, taxon ID (NCBI taxonomy), etymology, rank, lineage, the dates of effective and/or valid publication, feature descriptions, nomenclature type and references for the proposal and emendations during the history of the taxon are presented. Besides these data, the species taxa contain information about 16S rRNA gene and/or genome sequences. All publicly available genome data of each type species including both type and non-type strains were collected, and if needed, re-annotated using the standardized analysis pipeline. Furthermore, pan-genomic data analyses were conducted for species with ≥5 genome sequences available. Finally, for all type species, taxonomically relevant phenotypic data were extracted and curated from literatures, which were further indexed into eLMSG as searchable and analyzable data records. Taken together, eLMSG is a comprehensive web platform for studying microbial systematics and genomics, potentially useful for better understanding microbial taxonomy, natural evolutionary processes and ecological relationships. PADS Arsenal {#SEC2-6} ------------ PADS Arsenal (<https://bigd.big.ac.cn/padsarsenal>) (detailed in this issue) is a comprehensive public database of prokaryotic defense systems related genes (PADS). To address the challenges of ever-increasing prokaryotic genomic data and the progressive discovery of novel defense systems, we develop PADS Arsenal for browsing, searching, and analyzing various defense system genes. In the current version, PADS Arsenal integrates 6 600 264 defense systems genes, which belong to 18 defense systems, 63 701 genomes and 33 390 species of archaea and bacteria. In addition, it supports defense system gene analysis by equipping with an interactive online pipeline that includes sequence homology search, multiple sequence alignment and phylogenetic analysis. Meanwhile, PADS Arsenal provides a presence-absence variation (PAV) analysis function to visualize the dynamic variation of defense system genes. Collectively, PADS Arsenal integrates a comprehensive collection of defense system genes in archaea and bacteria and thus provides valuable resources to facilitate development of novel genome editing, engineering and regulation tools. RECENT UPDATES {#SEC3} ============== BioProject and BioSample {#SEC3-1} ------------------------ BioProject (<https://bigd.big.ac.cn/bioproject>) and BioSample (<https://bigd.big.ac.cn/biosample>), designed in compliance with INSDC (International Nucleotide Sequence Database Collaboration; a joint initiative by DDBJ, EMBL-EBI and NCBI) standards, are two public repositories of biological projects and biological samples, respectively. They collect and store descriptive metadata and information about biological projects and biological materials used for experiments. By providing a centralized access to all public projects and reciprocal links to their related data, BioProject supports various projects in terms of data types, ranging from genomic, transcriptomic, epigenomic and metagenomic sequencing projects to genome-wide association studies (GWAS) and variation analyses. Similarly, BioSample serves as a centralized access to all public samples and reciprocal links to BioProject as well as other relevant database resources. In the past year, BioSample has been significantly upgraded by adding the batch submission functionality and allowing users to submit information of multiple samples in a single table, which consequently had greatly improved the efficiency of data submission. As of August 2019, BioProject houses a total of 1248 biological projects submitted by 734 users from 219 organizations and BioSample includes a total of 87 107 samples from 482 species, presenting a dramatic increase in data submission (Figure [2](#F2){ref-type="fig"}). {#F2} Genome Sequence Archive {#SEC3-2} ----------------------- As a public data repository for archiving raw sequence reads, the Genome Sequence Archive (GSA; <https://bigd.big.ac.cn/gsa>) ([@B11]) accepts data submissions from all over the world and provides free access to all publicly available data for global scientific communities. Over the past year, GSA has been significantly enhanced by upgrading the metadata submission functionality to enable batch submission of experiments and runs in a single table. Till August 2019, GSA has archived a total of 55 057 Experiments and 59 566 Runs and housed \>1200 Terabytes of submitted raw sequence data (Figure [2](#F2){ref-type="fig"}), showing the doubled volume by comparison with the previous release last August (namely, ∼580 TB). According to the statistics (<https://bigd.big.ac.cn/gsa/statistics>), data housed in GSA were submitted from 150 organizations and reported in \>100 scientific journals, including Cell, Genome Research, Genomics Proteomics Bioinformatics, Nature, Plant Cell and PNAS. More importantly, GSA has been designated as supported repository for genes and gene expression data by Elsevier. All released data in GSA are publicly accessible and downloadable at <ftp://download.big.ac.cn/gsa/>. Genome Warehouse {#SEC3-3} ---------------- The Genome Warehouse (GWH; <https://bigd.big.ac.cn/gwh>) is a public archival resource housing genome-scale data for a wide range of species. For each collected genome assembly, GWH incorporates detailed descriptive information, including metadata of biological sample, genome assembly, sequence data and genome annotation, and offers standardized quality control for genome sequence and genome annotation. Notably, in this version, the sequences of the northern Han reference genome (NH1.0; GWHAAAS00000000) has been deposited in GWH, which was *de novo* assembled with a contig N50 size of 3.6 Mb and a scaffold N50 size of 46.63 Mb (see ([@B7]) for details). In addition, GWH has been significantly upgraded by accepting updated submissions (including both genome sequence and updates of genome annotation) and improving web services for data submission, release and sharing. In particular, GWH provides data visualization for both genome sequence and genome annotation powered by JBrowse ([@B12]) and offers statistics and charts in light of assembly, genome, sequencing platform, assembly method, organization and download. Till September 2019, GWH has accepted 649 data submissions from organizations both nationally and internationally and covered a broad diversity of species, e.g. animals, plants, fungi, bacteria, archaea and viruses. Among them, 133 genome assemblies have been publicly released and reported in 19 international journals. Genome Variation Map {#SEC3-4} -------------------- The Genome Variation Map (GVM; <https://bigd.big.ac.cn/gvm>) ([@B13]) is a public database of genome variations, including single nucleotide polymorphisms (SNP) and small insertions and deletions (indel). Different from dbSNP that only accepts human data submissions, GVM collects genome variations for a wide range of species and accepts submissions of different types of genome variations from all over the world. In the current version, GVM incorporates a total of ∼8.4 billion variants for 13 animals and 19 plants, including 7.2 billion SNPs and 1.2 billion indels. By comparison with the previous version, it has been updated by integrating 47 million variants from two newly added species (diploid wheat and cat). In addition, GVM has accepted 24 genome variation data submissions involving 23 056 samples from 10 species. Non-coding RNA Resources {#SEC3-5} ------------------------ **NONCODE** (<http://www.noncode.org>) ([@B14]) is an integrated knowledgebase dedicated to the complete collection and annotation of non-coding RNAs (ncRNA). Almost all the types of ncRNAs (excluding tRNAs and rRNAs) were filtered automatically from literatures and other public databases and were later manually curated. The ncRNA sequences and their related information (such as chromosomal information, conservation, function, etc.) were collected and recorded. BLAST alignment search service and access through our custom UCSC Genome Browser were also incorporated. In the current version (v5.0), 17 species are included in NONCODE (human, mouse, cow, rat, chicken, fruit fly, zebrafish, nematode, yeast, Arabidopsis, chimpanzee, gorilla, orangutan, rhesus macaque, opossum platypus and pig). Consequently, NONCODE collects a total of 548,640 long ncRNAs (lncRNA), coupled with their expression profiles identified based on RNA-seq data for human and mouse as well as their predicted functions. Moreover, it also includes human lncRNA--disease relationships and SNP--lncRNA--disease relationships, human exosome lncRNA expression profiles and predicted RNA secondary structures of human transcripts. **NPInter** (<http://bigdata.ibp.ac.cn/npinter>) ([@B15]) is a database that documents experimentally identified functional interactions between ncRNAs (except tRNAs and rRNAs), especially lncRNAs, and protein related biomacromolecules (proteins, mRNAs or genomic DNAs). NPInter provides the scientific community with a comprehensive and integrated tool for efficient browsing and extraction of information on interactions between ncRNAs and biomolecules. With the development of high-throughput biotechnology, such as cross-linking immunoprecipitation (CLIP-seq) and Chromatin Isolation by RNA purification (ChIRP-seq), the number of known ncRNA interactions, has grown rapidly in recent years. In the current release, NPInter houses 609 020 RNA-RNA interactions, 488 315 RNA--protein interactions and 892 737 RNA--DNA interactions, and provides more user-friendly interfaces and functional modules. **piRBase** (<http://www.regulatoryrna.org/database/piRNA/>) ([@B16]) is a comprehensive database of piRNA sequences, which are a class of small RNAs that is mainly expressed in animal germ line. piRBase integrates various piRNA-related high-throughput data in multiple species, leading to the largest collection of piRNAs and their annotations. Since its launch in 2014, piRBase has incorporated 264 datasets from 21 organisms and accordingly housed a total of ∼173 million piRNAs up to now. Furthermore, piRBase provides comprehensive annotations of piRNA sequences and genomic loci as well as piRNA targets and disease-related piRNAs. In addition, epigenetic and post-transcriptional regulation data were systematically integrated to support piRNA functional study. **LncBook** ([@B17]) (<https://bigd.big.ac.cn/lncbook>) and **LncRNAWiki** ([@B18]) (<https://bigd.big.ac.cn/lncrnawiki>), are two dedicated resources of human lncRNAs, through expert curation and community curation, respectively. In the past year, LncBook has been updated by removing 1196 redundant lncRNA transcripts and updating genomic annotations of 1046 lncRNA transcripts. As a result, LncBook provides a high-quality collection of 268 848 non-redundant lncRNA transcripts and 140 356 lncRNA genes. Also, LncBook presents tissue-specific lncRNAs (TS lncRNAs) for different tissues; among the 32 tissues, testis has the largest number of TS lncRNAs (9024 lncRNAs) and the following tissue is brain (2297 lncRNAs). In addition, LncBook is equipped with an online tool for coding potential prediction, which is able to accurately identify lncRNAs in a wide range of species ([@B19]). On the other side, LncRNAWiki ([@B18]), a wiki-based platform for community curation of human lncRNAs, has been updated by curating 291 human lncRNAs with functional experiment evidence, including 149 newly added lncRNAs and 142 existing lncRNAs with updated publications. Also, 65 redundant lncRNAs based on the approved and alias symbols (<https://www.genenames.org>) were removed. Consequently, in the current release, the number of functionally validated human lncRNAs in LncRNAWiki has grown to 1951. Together, LncBook and LncRNAWiki are of great potential to achieve comprehensive integration of human lncRNAs and their annotations ([@B20]). RNA Editing Resources {#SEC3-6} --------------------- Editome Disease Knowledgebase (EDK; <https://bigd.big.ac.cn/edk>) ([@B21]) and Plant Editosome Database (PED; <https://bigd.big.ac.cn/ped>) ([@B22]) are two RNA editing resources for human and plants, respectively. In the updated version, EDK incorporates two new diseases associated with 51 experimentally validated abnormal editing events located in six mRNAs, and 10 aberrant activities involved with two editing enzymes. Furthermore, to provide an easy-to-use and downloadable reference for further functional investigation on individual RNA editing event, EDK incorporates detailed structured annotation information for each editing site, including gene, specific gene region, molecular effect, editing enzyme, associated disease and/or phenotype. As a featured database of RNA editosome in plants ([@B22],[@B23]), PED has been updated by integrating two more editing factors, which had been recently verified to be involved in RNA editing processes and related to important phenotypes in Arabidopsis and new maize variety. Collectively, EDK and PED integrate more valuable information of editing enzymes (factors) and/or editing events associated with phenotypes, so as to help users facilitate systematic investigations on RNA editing machinery in both human and plants. MethBank {#SEC3-7} -------- The Methylation Bank (MethBank; <https://bigd.big.ac.cn/methbank>) ([@B24],[@B25]) is a databank of genome-wide DNA methylomes across a variety of species, with particular focus on human health and aging, animal embryonic development and plant growth and development. In the current version, MethBank offers 43 consensus reference methylomes (CRM) for human owing to large-scale DNA methylation array data public available, which are sourced from 10 healthy human tissues including 4577 peripheral blood samples, 26 prostate samples, 241 saliva samples, 322 skin samples, 98 breast samples, 38 colon samples, 206 kidney samples, 50 liver samples, 150 lung samples and 56 thyroid samples. In addition to CRMs, MethBank provides single-base resolution methylomes (SRM) based on whole-genome bisulfite sequencing data from human, plants and animals. Up to now, MethBank includes 40 SRMs from 26 healthy human tissues, 336 from different developmental stages in five economical plants and 18 from gametes and early embryos in two model animals. In addition, MethBank provides useful information on methylation data analysis tools, helpful for users to easily find any tool of interest. EWAS Atlas {#SEC3-8} ---------- EWAS Atlas (<https://bigd.big.ac.cn/ewas>) ([@B26]) is a curated knowledgebase of epigenome-wide association studies. During the past year, it has been enriched by adding a total of 121 156 EWAS associations manually extracted and curated from 191 publications. It is noted that the MethylationEPIC (850K/EPIC) array becomes increasingly popular, so that the number of 850K-based publications in EWAS Atlas has increased accordingly. In addition, the online trait enrichment tool was further enhanced and EWAS knowledge graph (<https://bigd.big.ac.cn/ewas/network>) was newly developed to visualize and explore trait-gene networks. Till September 2019, EWAS Atlas has integrated 450 328 high-quality EWAS associations derived from 1003 studies in 401 publications, including 135 tissues/cell lines, 409 traits, 2689 cohorts and 409 ontology entities. Information Commons for Rice {#SEC3-9} ---------------------------- Information Commons for Rice (IC4R; <http://ic4r.org>) ([@B27],[@B28]) is a comprehensive resource dedicated to integrating multi-omics data for rice. To improve the completeness of gene structure and identify novel genes, the current implementation of IC4R incorporates a new gene annotation system IC4R-2.0 that is built based on a large number of 1503 public RNA-seq datasets, accordingly achieving higher integrity and quality by comparison with previous annotation systems. Specifically, IC4R-2.0 contains 56,221 protein-coding gene loci corresponding to 80 039 mRNAs, among which more than 27 000 gene loci are substantially improved with structural modification, 456 novel genes are identified, and 3215 lncRNAs and 4373 circular RNAs are annotated. In addition, although IC4R offers a high-density rice variation map of ∼18 million SNPs, these raw SNPs are not readily usable for population genetics, evolutionary analysis, association studies or genomic breeding in rice. To satisfy various needs of rice researchers on data mining of the integrated genotypic data, a committed module---SnpReady for Rice (SR4R, <http://sr4r.ic4r.org>), is developed and deployed in IC4R. SR4R features the lowest SNP redundancy and highest genetic diversity of rice populations. Currently, SR4R mainly integrates four reference SNP panels, including 'hapmapSNPs' after data filtration and genotype imputation, 'tagSNPs' selected from linkage disequilibrium (LD)-based redundancy removal, 'fixedSNPs' selected from genes exhibiting selective sweep signatures, and 'barcodeSNPs' selected from DNA fingerprinting simulation. The associated SNPs in these four panels as well as online toolkits are publicly available and downloadable. LSD {#SEC3-10} --- The leaf senescence database (LSD; <https://bigd.big.ac.cn/lsd>) ([@B29],[@B30]) is dedicated to the comprehensive collection of senescence-associated genes (SAGs) and their corresponding mutants through manual curation. In the current version (v3.0; see an update in ([@B31]) in this issue), LSD incorporates 5,853 SAGs and 617 mutants from 68 species. Notably, it integrates leaf senescence-associated transcriptome data in *Arabidopsis*, rice, soybean and poplar and identifies senescence-differentially expressed small RNAs (Sen-smRNA) in *Arabidopsis*. Moreover, LSD contains senescence phenotypes of 90 natural accessions (ecotypes) and 42 images of ecotypes in *Arabidopsis* and collects mutant seed information of *SAG*s in rice. Also, interaction pairs between Sen-smRNAs and senescence-associated transcription factors are integrated into LSD. Collectively, the updated LSD has the great potential to continue to provide useful information for the plant research community. Database Commons {#SEC3-11} ---------------- Database Commons (<https://bigd.big.ac.cn/databasecommons>), a catalog of global biological databases, provides open access to a comprehensive collection of publicly available databases and their descriptive metadata. Currently, it catalogues a total of 4615 databases, involving more than 7000 publications and ∼2000 organizations throughout the world. In the past year, Database Commons has been updated by assigning category tag(s) to each database, linking related databases and providing citation information according to Europe PMC ([@B32]). Importantly, to improve the quality of descriptive metadata for each database, we sent invitations to database owners (according to the publications) to call for community curation of their own databases. As a result, a total of 287 database owners have responded and made valuable curations to 345 databases. eGPS Cloud {#SEC3-12} ---------- eGPS Cloud (<http://egpscloud.big.ac.cn>) ([@B33]) is a multi-functional web portal that integrates comprehensive multi-omics tools and provides online data analysis services for studying evolutionary Genotype-Phenotype Systems (eGPS). In the current release, eGPS Cloud is equipped with 15 tools and 20 visualization scripts, accordingly delivering four modularized web services, that is, genomics data analysis, population data analysis, evolutionary & network data analysis, and multi-omics data visualization. It allows users to configure customized parameters for different tools and perform various data analysis online in a straightforward and friendly manner. Ongoing efforts are linking eGPS Cloud with GSA in order to provide users with seamless services for raw sequence data analysis. BIG Search {#SEC3-13} ---------- BIG Search (<https://bigd.big.ac.cn/search>) is a distributed and scalable full-text search engine built based on Elasticsearch (a highly scalable open-source search and analytics engine, <https://www.elastic.co/>). It features cross-domain search and facilitates users to gain access to a wide range of biological data almost in real-time. In the current version, BIG Search includes data indexes from all NGDC's resources and 25 partner resources (see details at <https://bigd.big.ac.cn/partners>). Additionally, EBI data resources have also been integrated into BIG Search powered by EBI Search RESTful API ([@B34]). In summary, BIG Search has been significantly updated by incorporating more data indexes from internal and external resources and displaying search results in a more user-friendly manner. BIG Submission {#SEC3-14} -------------- BIG Submission (<https://bigd.big.ac.cn/gsub>) is a one-stop submission portal that provides submission services for a series of database resources in NGDC, including BioProject, BioSample, GSA, GWH and GVM. During the past year, BIG Submission has been upgraded by optimizing the web interfaces and expanding the storage and computing resources, with the purpose to meet the needs of the rapid growth of data submissions. Importantly, it has been equipped by Aspera, a high-speed transfer tool that can greatly improve the data transfer efficiency and provide users with better submission experiences. BIG SSO {#SEC3-15} ------- BIG Single Sign-On (SSO; <https://bigd.big.ac.cn/sso>) is a user access control system that refers to systems where a single authentication provides access to multiple applications by passing the authentication token seamlessly to configured applications. In the past year, HTTPS protocols have been deployed in all web sites for security transfer, so that the BIG SSO system has been updated to be much safer and more reliable. Meanwhile, services for user registration and update have been enhanced and delivered as a micro-service. CONCLUDING REMARKS {#SEC4} ================== NGDC provides a family of database resources through big data deposition, integration and translation, with the aim to support worldwide research activities in both academia and industry. In the past year, it has been significantly updated by archiving more data submissions, performing value-added curation, and improving web interfaces and services. And most importantly, it has been enhanced as the national center by joint efforts from BIG, IBP and SINH, forming an excellent line-up of field experts from the three institutions. Ongoing and future efforts are standardization of data models and curation processes, unification of web interfaces and SSO authentication across database resources, establishment of cloud infrastructure for big data storage and transfer, and development of a variety of databases and tools to facilitate the translation of big data into big discovery. NGDC is open to worldwide collaborations, particularly seeking the possibility to collaborate with INSDC members in dealing with big data archive. In addition, NGDC promotes big data sharing at a worldwide scale by setting up the Global Biodiversity and Health Big Data Alliance (BHBD; <http://bhbd-alliance.org>); by July 2019, 20 organizational members from 11 countries have joined the BHBD Alliance, with active collaborations in organizing international meetings/symposia, training courses and joint research projects. With more stable support from the government and CAS, NGDC will continue to grow to deliver a wide range of data resources and services in aid of both domestic and international research activities. We thank a number of users for submitting data, sending suggestions, reporting bugs and getting involving in community curation. The National Genomics Data Center is indebted to its funders, including the Ministry of Science & Technology and the Ministry of Finance of the People\'s Republic of China as well as Chinese Academy of Sciences. We would like to express our sincere thanks to the late Professor Bailin Hao (1934--2018), a leading bioinformatician of his generation, who had first advocated the establishment of national center since the 1990s. FUNDING {#SEC5} ======= Strategic Priority Research Program of the Chinese Academy of Sciences \[XDA19050302, XDB13040500, XDB13040100\]; National Key Research & Development Program of China \[2018YFD1000505, 2018YFC2000100, 2018YFC1406902, 2018YFC0910400, 2018YFC0310602, 2017YFC1201200, 2017YFC0908405, 2017YFC0908404, 2017YFC0908403, 2017YFC0907505, 2017YFC0907503, 2017YFC0907502, 2016YFE0206600, 2016YFC0906403, 2016YFC0903003, 2016YFC0901904, 2016YFC0901903, 2016YFC0901702, 2016YFC0901604, 2016YFC0901603, 2016YFB0201702\]; National Natural Science Foundation of China \[91731303, 81670462, 31970565, 31871328, 31871294, 31801104, 31771465, 31771410, 31771388, 31671360, 31571358, 31525014, 1470330, 31961130380, 31711530221\]; UK Royal Society-Newton Advanced Fellowship \[NAF\\R1\\191094\]; International Partnership Program of the Chinese Academy of Sciences \[153F11KYSB20160008, 153D31KYSB20170121\]; 13th Five-year Informatization Plan of Chinese Academy of Sciences \[XXH13505-05\]; Key Program of the Chinese Academy of Sciences \[KJZD-EW-L14\]; Key Research Program of Frontier Sciences of the Chinese Academy of Sciences \[QYZDJ-SSW-SYS009\]; Key Technology Talent Program of the Chinese Academy of Sciences; The 100 Talent Program of the Chinese Academy of Sciences; K.C. Wong Education Foundation; The Youth Innovation Promotion Association of the Chinese Academy of Sciences \[2019104, 2018134, 2017141\]; The Special Project on Precision Medicine under the National Key R&D Program \[SQ2017YFSF090210\]; The Open Biodiversity and Health Big Data Initiative of IUBS. Funding for open access charge: Strategic Priority Research Program of the Chinese Academy of Sciences. *Conflict of interest statement*. None declared. **Corresponding author:** Zhang Zhang^1,2,3,10,11,\*^ **Co-corresponding authors:** Wenming Zhao^1,2,3,10,\*^, Jingfa Xiao^1,2,3,10,\*^, Yiming Bao^1,2,3,10,11,\*^, Shunmin He^1,4,10,\*^, Guoqing Zhang^1,5,\*^, Yixue Li^1,5,\*^, Guoping Zhao^1,5,6,7,\*^, Runsheng Chen^1,4,10,\*^ **NGDC MEMBERS** (Arranged by project role and then by contribution except for Team Leader (TL), as indicated) **PGG.Han:** Yang Gao^5,\#^, Chao Zhang^5,\#^, Liyun Yuan^5,\#^, Guoqing Zhang^1,5,\*^ (TL), Shuhua Xu^5,14,15,16^ (TL) **PGG.SNV:** Chao Zhang^5,\#^, Yang Gao^5,\#^, Zhilin Ning^5,\#^, Yan Lu^5,\#^, Shuhua Xu^5,14,15,16^ (TL) **CGVD:** Jingyao Zeng^1,2,3,\#^, Na Yuan^1,2,\#^, Junwei Zhu^1,2^, Mengyu Pan^1,2^, Hao Zhang^1,2,3,10^, Qi Wang^1,2,3,10^, Shuo Shi^1,2,3,10^, Meiye Jiang^1,2,3,10^, Mingming Lu^1,2,3,10^, Qiheng Qian^1,2,3,10^, Qianwen Gao^1,2,3,10^, Yunfei Shang^1,2,3,10^, Jinyue Wang^1,2,3,10^, Zhenglin Du^1,2,\#^ (TL), Jingfa Xiao ^1,2,3,10,\*^ (TL) **GWAS Atlas:** Dongmei Tian^1,2,\#^, Pei Wang^1,2,3,10,\#^, Bixia Tang^1,2,\#^, Cuiping Li^1,2,\#^, Xufei Teng^1,2,3,10^, Xiaonan Liu^1,2,3,10^, Dong Zou^1,2,3^, Shuhui Song^1,2,3,\#^ (TL) **EWAS Data Hub:** Zhuang Xiong^1,2,3,10,\#^, Mengwei Li^1,2,3,10,\#^, Fei Yang^1,2,3,10,\#^, Yingke Ma^1,2,3^, Jian Sang^1,2,3,10^, Zhaohua Li ^1,2,3,10,11^, Rujiao Li^1,2,3,\#^ (TL) **iSheep:** Zhonghuang Wang^1,2,10,\#^, Qianghui Zhu^9,10,\#^, Junwei Zhu^1,2^, Xin Li^9^, Sisi Zhang^1,2^, Dongmei Tian^1,2^, Hailong Kang^1,2,10^, Cuiping Li^1,2^, Lili Dong^1,2^, Cui Ying^1,2,10^, Guangya Duan^1,2,10^, Shuhui Song^1,2,3^, Menghua Li^9,10^ (TL), Wenming Zhao^1,2,3,10,\*^ (TL) **eLMSG:** Xiaoyang Zhi^12,\#^ (TL), Yunchao Ling^5,\#^, Ruifang Cao^5,\#^, Zhao Jiang^12^, Haokui Zhou^7^, Daqing Lv^5^, Wan Liu^5^, Hans-Peter Klenk^13^, Guoping Zhao^1,5,6,7,\*^, Guoqing Zhang^1,5,\*^ (TL) **PADS:** Yadong Zhang^1,2,3,10,\#^, Zhewen Zhang^1,2,3,\#^, Hao Zhang^1,2,3,10^, Jingfa Xiao^1,2,3,10,\*^ (TL) **BioProject & BioSample & GSA & BIG Submission**: Tingting Chen^1,2,\#^, Sisi Zhang^1,2,\#^, Xu Chen^1,2,\#^, Junwei Zhu^1,2,\#^, Zhonghuang Wang^1,2,3,10^, Hailong Kang^1,2,3,10^, Lili Dong^1,2^, Yanqing Wang^1,2,\#^ (TL) **GWH:** Yingke Ma^1,2,3,\#^, Song Wu^1,2,3,10^, Zhaohua Li^1,2,3,10,11^, Zheng Gong^1,2,3,10^, Meili Chen^1,2,3,\#^ (TL) **GVM:** Cuiping Li^1,2,\#^, Dongmei Tian^1,2,\#^, Xufei Teng^1,2,3,10,\#^, Pei Wang^1,2,3,10,\#^, Bixia Tang^1,2,\#^, Xiaonan Liu^1,2,3,10^, Dong Zou^1,2,3^, Shuhui Song^1,2,3,\#^ (TL) **NONCODE:** Shuangsang Fang^8^, Lili Zhang^4,10^, Jincheng Guo^8^, Yiwei Niu^4,10^, Yang Wu^8^, Hui Li^8^, Lianhe Zhao^8^, Xiyuan Li^8^, Xueyi Teng^4,10^, Xianhui Sun^4,10^, Liang Sun^8^, Runsheng Chen^1,4,10,\*^, Yi Zhao^8^ (TL) **piRBase:** Jiajia Wang^4,10,\#^, Peng Zhang^4,\#^, Yanyan Li^4,10^, Yu Zheng^4,10^, Runsheng Chen^1,4,10,\*^, Shunmin He^1,4,10,\*^ (TL) **NPInter:** Xueyi Teng^4,10,\#^, Xiaomin Chen^4,10,\#^, Hua Xue^4,10,\#^, Yiheng Teng^4,10^, Peng Zhang^4^, Quan Kang^4^, Yajing Hao^4^, Yi Zhao^8^, Runsheng Chen^1,4,10,\*^, Shunmin He^1,4,10,\*^ (TL) **LncBook & LncRNAWiki:** Jiabao Cao^1,2,3,10,\#^, Lin Liu^1,2,3,10,\#^, Zhao Li^1,2,3,10,\#^, Qianpeng Li^1,2,3,10^, Dong Zou^1,2,3^, Qiang Du^1,2,3,10^, Amir A. Abbasi^25^, Huma Shireen^25^, Nashaiman Pervaiz^25^, Fatima Batool^25^, Rabail Z. Raza^25^, Lina Ma^1,2,3,\#^ (TL) **EDK & PED:** Guangyi Niu^1,2,3,10,\#^, Yuansheng Zhang^1,2,3,10,\#^, Dong Zou^1,2,3,\#^, Tongtong Zhu^1,2,3,10,11^, Jian Sang^1,2,3,10^, Mengwei Li^1,2,3,10^, Lili Hao^1,2,3,\#^ (TL) **MethBank:** Dong Zou^1,2,3,\#^, Guoliang Wang^24,\#^, Mengwei Li^1,2,3,10,\#^, Rujiao Li^1,2,3,\#^ (TL) **EWAS Atlas:** Mengwei Li^1,2,3,10,\#^, Rujiao Li^1,2,3^, Yiming Bao^1,2,3,10,11,\*^ (TL) **IC4R:** Jun Yan^17,\#^, Jian Sang^1,2,3,10,\#^, Dong Zou^1,2,3,\#^, Chen Li^22^, Zhennan Wang^10,23^, Yuansheng Zhang^1,2,3,10^, Tongtong Zhu^1,2,3,10,11^, Shuhui Song^1,2,3^ (TL), Xiangfeng Wang^17^ (TL), Lili Hao^1,2,3^ (TL) **LSD**: Zhonghai Li^18,\#^ (TL), Yang Zhang^1,2,3,10,\#^, Dong Zou^1,2,3^, Yi Zhao^19^, Houling Wang^18^, Yi Zhang^18^, Xinli Xia^18,20^, Hongwei Guo^18,21^, Zhang Zhang^1,2,3,10,11,\*^ **Database Commons:** Dong Zou^1,2,3,\#^, Lina Ma^1,2,3,\#^ (TL) **eGPS Cloud:** Lili Dong^1,2,\#^, Bixia Tang^1,2,\#^, Junwen Zhu^1,2,\#^, Qing Zhou^1,2,10^, Zhonghuang Wang^1,2,10^, Hongen Kang^1,2,10^, Xu Chen^1,2^, Li Lan^1,2^, Yiming Bao^1,2,3,10,11,\*^ (TL), Wenming Zhao^1,2,3,10,\*^ (TL) **BIG Search:** Dong Zou^1,2,3,\#^ (TL) **BIG SSO:** Junwei Zhu^1,2,\#^ (TL), Bixia Tang^1,2,\#^ **BHBD:** Yiming Bao^1,2,3,10,11,\*^, Li Lan^1,2^, Xin Zhang^1,2^, Yingke Ma^1,2,3^, Yongbiao Xue^26^ (Project Leader) **Hardware & System Administration:** Yubin Sun^1,2^, Shuang Zhai^1,2^, Lei Yu^1,2^, Mingyuan Sun^1,2^, Huanxin Chen^1,2^ (TL) **Writing Group:** Zhang Zhang^1,2,3,10,11,\*^, Wenming Zhao^1,2,3,10,\*^, Jingfa Xiao^1,2,3,10,\*^, Yiming Bao^1,2,3,10,11,\*^, Lili Hao^1,2,3^ **NGDC PARTNERS** (Listed in alphabetical order by database names) **AnimalTFDB:** Hui Hu^27^, An-Yuan Guo^27^ **dbPAF & WERAM:** Shaofeng Lin^27^, Yu Xue^27^ **dbPPT:** Chenwei Wang^27^, Yu Xue^27^ **dbPSP:** Wanshan Ning^27^, Yu Xue^27^ **CellMarker**: Xinxin Zhang^28^, Yun Xiao^28^, Xia Li^28^ **CGDB:** Yiran Tu^27^, Yu Xue^27^ **circAtlas:** Wanying Wu^29^, Peifeng Ji^29^, Fangqing Zhao^29^ **DEG & DoriC:** Hao Luo^30,31,32^, Feng Gao^30,31,32^ **iEKPD:** Yaping Guo^27^, Yu Xue^27^ **GenTree**: Hao Yuan^33,34^, Yong E. Zhang^10,33,34^ **hTFtarget:** Qiong Zhang^27^, An-yuan Guo^27^ **iUUCD:** Jiaqi Zhou^27^, Yu Xue^27^ **LncRNADisease:** Zhou Huang^35^, Qinghua Cui^35,36^ **lncRNASNP:** Ya-Ru Miao^27^, An-Yuan Guo^27^ **MiCroKiTS:** Chen Ruan^27^, Yu Xue^27^ **PceRBase:** Chunhui Yuan^37^, Ming Chen^37^ **PlantTFDB:** Jin-Pu Jin^38^, Feng Tian^38^, Ge Gao^38^ **PLMD:** Ying Shi^27^, Yu Xue^27^ **PTMD**: Lan Yao^27^, Yu Xue^27^, Qinghua Cui^35,36^ **RhesusBase:** Xiangshang Li^39^, Chuan-Yun Li^39^ **SEGreg:** Qing Tang^27^, An-Yuan Guo^27^ **THANATOS:** Di Peng^27^, Yu Xue^27^ ^1^National Genomics Data Center, Beijing 100101, China ^2^BIG Data Center, Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing 100101, China ^3^CAS Key Laboratory of Genome Sciences and Information, Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing 100101, China ^4^Key Laboratory of RNA Biology, Center for Big Data Research in Health, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China ^5^Bio-Med Big Data Center, Key Laboratory of Computational Biology, CAS-MPG Partner Institute for Computational Biology, Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, Shanghai 200231, China ^6^CAS Key Laboratory of Synthetic Biology, Institute of Plant Physiology and Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200231, China ^7^Center for Quantitative Synthetic Biology, Institute of Synthetic Biology, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China ^8^Key Laboratory of Intelligent Information Processing, Advanced Computer Research Center, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China ^9^CAS Key Laboratory of Animal Ecology and Conservation Biology, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China ^10^University of Chinese Academy of Sciences, Beijing 100049, China ^11^School of Future Technology, University of Chinese Academy of Sciences, Beijing 100049, China ^12^Yunnan Institute of Microbiology, School of Life Sciences, Yunnan University, Kunming, Yunnan 650091, China ^13^School of Natural and Environmental Sciences, Ridley Building 2, Newcastle University, Newcastle upon Tyne, UK ^14^School of Life Science and Technology, ShanghaiTech University, Shanghai 201210, China ^15^Center for Excellence in Animal Evolution and Genetics, Chinese Academy of Sciences, Kunming 650223, China ^16^Collaborative Innovation Center of Genetics and Development, Shanghai 200438, China ^17^Department of Crop Genomics and Bioinformatics, College of Agronomy and Biotechnology, China Agricultural University, Beijing 100094, China ^18^Beijing Advanced Innovation Center for Tree Breeding by Molecular Design, Beijing Forestry University, Beijing 100083, China ^19^College of Life Sciences, Peking University, Beijing 100871, China ^20^College of Biological Sciences and Biotechnology, National Engineering Laboratory for Tree Breeding, Beijing Forestry University, Beijing 100083, China ^21^Institute of Plant and Food Science, Department of Biology, Southern University of Science and Technology (SUSTech), Shenzhen, Guangdong 518055, China ^22^Rice Research Institute, Guangdong Academy of Agricultural Sciences, Guangzhou 510640, China ^23^Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China ^24^College of Plant Protection, Hunan Agricultural University, Hunan 410128, China ^25^National Center for Bioinformatics, Programme of Comparative and Evolutionary Genomics, Faculty of Biological Sciences, Quaid-i-Azam University, Islamabad 45320, Pakistan ^26^Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing 100101, China ^27^Department of Bioinformatics and Systems Biology, Key Laboratory of Molecular Biophysics of the Ministry of Education, Hubei Bioinformatics and Molecular Imaging Key Laboratory, College of Life Science and Technology, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China ^28^College of Bioinformatics Science and Technology, Harbin Medical University, Harbin, Heilongjiang 150081, China ^29^Beijing Institutes of Life Science, Chinese Academy of Sciences, Beijing 100101, China ^30^Department of Physics, School of Science, Tianjin University, Tianjin 300072, China ^31^Frontier Science Center of Synthetic Biology, Key Laboratory of Systems Bioengineering, Tianjin University, Tianjin 300072, China ^32^SynBio Research Platform, Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin 300072, China ^33^Key Laboratory of Zoological Systematics and Evolution and State Key Laboratory of Integrated Management of Pest Insects and Rodents, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China ^34^CAS Center for Excellence in Animal Evolution and Genetics, Chinese Academy of Sciences, Kunming, Yunnan 650223, China ^35^Department of Biomedical Informatics, School of Basic Medical Sciences, MOE Key Lab of Cardiovascular Sciences, Center for Noncoding RNA Medicine, Peking University, Beijing 100190, China ^36^Center of Bioinformatics, Key Laboratory for Neuro-Information of Ministry of Education, School of Life Science and Technology, University of Electronic Science and Technology of China, Chengdu, Sichuan 610054, China ^37^Department of Bioinformatics, State Key Laboratory of Plant Physiology and Biochemistry, Institute of Plant Science, College of Life Sciences, Zhejiang University, Hangzhou, Zhejiang 310058, China ^38^Biomedical Pioneering Innovation Center (BIOPIC), Beijing Advanced Innovation Center for Genomics (ICG), Center for Bioinformatics (CBI), and State Key Laboratory of Protein and Plant Gene Research at School of Life Sciences, Peking University, Beijing 100871, China ^39^Institute of Molecular Medicine, Peking University, Beijing 100871, China \*To whom correspondence should be addressed: Zhang Zhang (<zhangzhang@big.ac.cn>). Correspondence may also be addressed to Wenming Zhao (<zhaowm@big.ac.cn>), Jingfa Xiao (<xiaojingfa@big.ac.cn>), Yiming Bao (<baoym@big.ac.cn>), Shunmin He (<heshunmin@ibp.ac.cn>), Guoqing Zhang (<gqzhang@picb.ac.cn>), Yixue Li (<yxli@sibs.ac.cn>), Guoping Zhao (<gpzhao@sibs.ac.cn>) and Runsheng Chen (<crs@sun5.ibp.ac.cn>). ^\#^The authors wish it to be known that, in their opinion, these authors should be regarded as Joint First Authors. [^1]: To whom correspondence should be addressed: Zhang Zhang. Tel: +86 10 84097261; Fax: +86 10 84097720; Email: <zhangzhang@big.ac.cn> [^2]: Correspondence may also be addressed to Wenming Zhao. Email: <zhaowm@big.ac.cn> [^3]: Correspondence may also be addressed to Jingfa Xiao. Email: <xiaojingfa@big.ac.cn> [^4]: Correspondence may also be addressed to Yiming Bao. Email: <baoym@big.ac.cn> [^5]: Correspondence may also be addressed to Shunmin He. Email: <heshunmin@ibp.ac.cn> [^6]: Correspondence may also be addressed to Guoqing Zhang. Email: <gqzhang@picb.ac.cn> [^7]: Correspondence may also be addressed to Yixue Li. Email: <yxli@sibs.ac.cn> [^8]: Correspondence may also be addressed to Guoping Zhao. Email: <gpzhao@sibs.ac.cn> [^9]: Correspondence may also be addressed to Runsheng Chen. Email: <crs@sun5.ibp.ac.cn> [^10]: The authors wish it to be known that, in their opinion, these authors should be regarded as Joint First Authors.

Introduction {#s1} ============ Targeting of RNA molecules to distinct subcellular destinations has emerged as a widely-used mechanism that controls the cytoplasmic distribution of a large proportion of expressed transcripts in a range of organisms ([@bib7]; [@bib34]; [@bib35]). Such RNA localization events are functionally important during development, cell-fate decisions, or physiologic responses of somatic cells, including synaptic plasticity and cell migration ([@bib11]; [@bib23]; [@bib24]; [@bib61]). Localized RNAs are thought to mediate these functional outcomes by directing and restricting protein production in specific subcellular compartments, thus contributing to the molecular compartmentalization of polarized cells. To achieve spatially-controlled protein production, translation of localized RNAs is thought to be suppressed during transport and activated at their final destination ([@bib4]; [@bib7]; [@bib30]). This model is supported by several lines of evidence. Localized RNAs are associated during transport with RNA-binding proteins (RBPs) that function as translational repressors, which block the initiation or elongation steps of translation. For example, the RBPs Khd1 and Puf6 repress the translation of the yeast ASH1 mRNA until it reaches the tip of the emerging yeast bud. Puf6 inhibits translation by binding to the initiation factor eIF5B, thus preventing ribosome subunit binding to the mRNA ([@bib17]; [@bib21]; [@bib41]). Supporting the coordinate regulation of RNA transport and translation, certain RBPs perform multiple functions and can operate to both repress translation and support RNA trafficking ([@bib1]; [@bib4]). Such multifunctional RBPs include FMRP, a known translational repressor, which can also link its target RNAs to the kinesin motor-based transport machinery ([@bib15]; [@bib18]; [@bib45]). Additionally, IMP-1/ZBP-1 is required both for transport as well as for translational repression of the beta-actin mRNA, by blocking recruitment of the 60S ribosomal subunit and initiation of translation ([@bib11]; [@bib25]). Once the RNA reaches its final destination, spatial control of gene expression is achieved by local de-repression of translation. Exemplifying this notion, at the yeast bud tip, the Puf6 and Kdh1 RBPs are phosphorylated by the Yck1 kinase, leading to release of eIF5B and translation activation ([@bib17]). Similarly, phosphorylation of ZBP1 by the Src kinase disrupts ZBP1-RNA binding leading to translation of beta-actin RNA ([@bib25]). Various other mechanisms maintain localized RNAs in a silenced state, and are relieved in a spatial manner. The Drosophila *nanos* mRNA is deadenylated and translationally repressed in the bulk cytoplasm of Drosophila embryos through the action of the RBP Smaug and the CCR4/NOT deadenylase. At the posterior pole, the Oskar protein relieves this inhibition and leads to de-repression of *nanos* translation ([@bib27]; [@bib63]). In neuronal dendrites, translation of RNAs can be suppressed by miRNAs ([@bib47]), and degradation of components of the RISC complex controls synaptic protein synthesis ([@bib2]). Transported RNAs can also be maintained in a translationally-repressed state through oligomerization or multiplexing into higher-order RNP particles or granules ([@bib8]; [@bib9]; [@bib16]). These particles (also referred to, in the case of neurons, as neuronal transport granules) share protein components as well as liquid-droplet properties with other phase-separated RNA granules, such as P-bodies and stress granules ([@bib16]; [@bib20]). Containment within such granules is thought to retain RNAs in a repressed state, inaccessible to the translation machinery. Local signals can release such 'masked' RNAs and allow their translation ([@bib6]; [@bib31]). We have been investigating a group of RNAs that are localized at protrusions of migrating cells. We refer to these RNAs as 'APC-dependent' because their localization requires the tumor-suppressor protein APC ([@bib36]; [@bib55]). Localization of APC-dependent RNAs at protrusions requires a particular subset of modified microtubules, namely detyrosinated microtubules, and is mechanically regulated in response to the stiffness of the extracellular environment ([@bib55]; [@bib60]). Specifically, increased actomyosin contractility on stiff substrates, through activation of a signaling pathway involving the RhoA GTPase and its effector formin mDia, leads to formation of a detyrosinated microtubule network, which in turn supports RNA localization at protrusions. Localization of APC-dependent RNAs at protrusions is important for efficient cell migration ([@bib55]). We hypothesize that the positive effect of APC-dependent RNAs on cell migration is mediated through local RNA translation at protrusions. Here, we use polysome association, single-molecule translation imaging reporters, and in situ imaging of endogenous nascent proteins to determine whether APC-dependent RNAs are translated at protrusions and whether their translation is affected by their location in the cytoplasm. We find that indeed, localized RNAs are translated at protrusions, but interestingly they are also translated with similar efficiency regardless of their location within the cell. Intriguingly, we observe that continuous transport to the periphery leads to coalescence of single RNAs into larger clusters that are translationally silenced. We further show that such silencing and clustering occurs at retracting protrusions. Therefore, in contrast to the model described above, APC-dependent RNAs are not locally activated solely at protrusions. Instead, after transport to the periphery, and upon protrusion retraction, they become translationally silent and segregate into multimeric RNA granules. We propose that this mechanism is used to regulate protein activities during dynamic cellular responses. Results {#s2} ======= Disrupting the localization of APC-dependent RNAs at protrusions does not affect their translation {#s2-1} -------------------------------------------------------------------------------------------------- As a first step towards assessing whether localization of APC-dependent RNAs at protrusions is coupled to their translation status, we disrupted RNA localization at protrusions and determined whether that affected the efficiency of their translation. To measure translation efficiency, we fractionated cell extracts on sucrose gradients to resolve RNAs according to the number of bound ribosomes ([Figure 1A](#fig1){ref-type="fig"}). To facilitate a larger scale analysis, we divided each gradient into four fractions based on UV absorbance traces. Fraction one includes free RNPs and the 40S and 60S ribosomal subunits, fraction 2 includes 80S monosomes, and fractions 3 and 4 include light and heavy polysomes, respectively. mRNAs in fractions 1 and 2 largely correspond to non-translated mRNAs, whereas mRNAs in fractions 3 and 4 are actively translated. To correct for variations introduced by sample manipulation during RNA purification, an equal amount of in vitro transcribed spike RNA was added to each fraction. The recovered RNA from each fraction was then used to simultaneously detect the levels of multiple RNA species along the four fractions ([Figure 1A](#fig1){ref-type="fig"}). For detection, we used nanostring analysis which allows for direct RNA counting and thus avoids biases introduced by reverse transcription and amplification. We designed probes to detect 20 protrusion-enriched RNAs, that we have previously defined as APC-dependent, 6 RNAs encoding ribosomal proteins or ribosome biogenesis factors, which we have previously shown are also enriched at protrusions but in an APC-independent manner, and eight control RNAs which based on our prior analysis are not enriched at protrusions ([@bib55]) ([Figure 1D](#fig1){ref-type="fig"}). {ref-type="fig"}.](elife-44752-fig1){#fig1} To first validate whether our experimental approach can detect changes in translation, we isolated polysome gradient fractions from control cells or cells treated with puromycin, an antibiotic that terminates translation and dissociates polysomes. Indeed, puromycin treatment led to an increase in monosomes and ribosomal subunits and a decrease in heavy polysomes ([Figure 1B](#fig1){ref-type="fig"}). (Note that NIH/3T3 cells used in these experiments have a reduced baseline amount of ribosomes engaged in translation (i.e. in light and heavy polysomes) compared to HEK293 cells shown in [Figure 1A](#fig1){ref-type="fig"}). Nanostring analysis of recovered RNA revealed that under control conditions a higher proportion of most RNAs exists in fractions 3 and 4, while treatment with puromycin significantly shifted the distribution of RNAs, in all three groups examined, towards fractions 1 and 2 ([Figure 1D](#fig1){ref-type="fig"} and [Figure 1---source data 1](#fig1sdata1){ref-type="supplementary-material"}). These results therefore indicate that indeed mRNAs present in fractions 3 and 4 correspond to actively translated transcripts, and furthermore, that our nanostring-based methodology can quantitatively detect changes in translation efficiency. To disrupt the localization of APC-dependent RNAs at protrusions, we employed two different methodologies. One is a competition-based method that relies on overexpression of an exogenous construct carrying the 3'UTR of the APC-dependent RNA, Pkp4. Our prior characterization showed that inducible expression of such competitive-UTR (cUTR) constructs leads to a preferential mislocalization of APC-dependent RNAs from protrusions (see [@bib55] for a detailed characterization of the stable cell lines used). A second method relies on the requirement for detyrosinated microtubules for localization of APC-dependent RNAs at protrusions ([@bib55]; [@bib60]). Brief treatment with the chemical compound parthenolide disrupts detyrosinated microtubules ([@bib60]) and significantly reduces RNA localization at protrusions, measured through the degree of RNA enrichment in isolated protrusion samples ([Figure 2A](#fig2){ref-type="fig"}). {ref-type="fig"}.](elife-44752-fig2){#fig2} To test whether mislocalization from protrusions is accompanied by changes in translation, polysome gradient fractions were isolated from control and cUTR- or parthenolide-treated cells and analyzed for the presence of various RNA species. In both cases, cUTR overexpression or parthenolide treatment did not affect the overall gradient profile, suggesting that no overt changes in translation resulted from either treatment ([Figures 1C](#fig1){ref-type="fig"} and [2B](#fig2){ref-type="fig"}). Furthermore, both approaches did not result in significant changes in the translation state of the control RNAs, the APC-independent RNAs or the mislocalized APC-dependent RNAs analyzed ([Figures 1D](#fig1){ref-type="fig"} and [2B](#fig2){ref-type="fig"} and [Figure 1---source data 1](#fig1sdata1){ref-type="supplementary-material"}, [Figure 2---source data 1](#fig2sdata1){ref-type="supplementary-material"}). Therefore, disrupting the peripheral localization of APC-dependent RNAs does not affect their translational efficiency, suggesting a lack of coupling between RNA transport and translational control. We note that both cUTR overexpression and parthenolide treatment resulted in a small apparent reduction in the amount of APC-dependent RNAs sedimenting in the heavy polysome fraction (fraction 4) and an apparently corresponding increase of RNA amounts in polysome fraction 3. These changes are small, they are not statistically significant and do not alter the conclusion that APC-dependent RNAs remain translationally active. Nevertheless, we make a note of this observation because it is manifested quite consistently by APC-dependent RNAs under conditions that disrupt their peripheral localization. In light of data presented below, we believe that this might reflect a real change in the organization of a small fraction of these RNPs (see discussion). Single-molecule translation reporters of protrusion-localized RNAs {#s2-2} ------------------------------------------------------------------ While the above results suggest that transport and translation of APC-dependent RNAs are not coordinated, they also raise the possibility that assessing the translation status in a whole-cell extract derived from heterogeneous cell populations might not offer the required sensitivity to detect local changes occurring on a single-cell level. To address this, we took advantage of the recently developed SunTag-based reporters that allow imaging of translation of single RNA molecules in live cells ([@bib38]; [@bib54]; [@bib57]; [@bib58]) ([Figure 3A](#fig3){ref-type="fig"}). These reporters carry a coding sequence which includes, at the 5' end, a series of SunTag peptide epitopes, which are recognized by a single chain antibody fragment fused to superfolder-GFP (scFv-GFP). The concentration of GFP-fused antibodies on a series of nascent peptides generated during translation allows the observation of translation sites as bright GFP spots, which can be distinguished from free antibody molecules or mature proteins released after translation. A second element incorporated in these reporter constructs is a series of binding sites for the PP7 bacteriophage coat protein introduced after the end of the coding sequence. These hairpin elements bind to a PP7 coat protein fused to 3 copies of mCherry fluorescent protein (3x-mCherry-PCP), thus allowing visualization of RNA molecules as bright red spots ([@bib58]) ([Figure 3B](#fig3){ref-type="fig"} and [Video 1](#video1){ref-type="video"}). To use this system, we have generated mouse NIH/3T3 fibroblast cell lines that stably express the GFP-fused antibody and the 3x-mCherry-PCP, and additionally can be induced to express reporter RNAs after addition of doxycycline ([Figure 3A,B](#fig3){ref-type="fig"} and see below). For our studies, we have been using a control reporter RNA which carries, after the PP7 repeats, a UTR sequence that doesn't direct transport to protrusions, and two localized reporter RNAs in which the PP7 repeats are followed by the 3'UTR of either the mouse Rab13 or Pkp4 RNAs, which as we have previously shown are sufficient to target reporter RNAs to protrusions ([Figure 3A](#fig3){ref-type="fig"}) ([@bib36]; [@bib55]). {#fig3} ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP (red) and the control translation reporter. Frames were acquired sequentially and with no time delay, for the duration of the movie (45 s). A merged image of the two channels is shown. Overlapping red and green spots indicate translation sites. Blue line: cell outline. Cyan line: nucleus outline. Scale bar: 5 μm. Single frames of this movie are shown in [Figure 3B](#fig3){ref-type="fig"}. 10.7554/eLife.44752.008 We first validated the RNA and translation signals detected with our implementation of the assay. To assess the specificity of the detected RNA spots, we compared the signal observed upon expression of the control PP7-containing reporter RNA to that of a similar reporter carrying a deletion of the PP7-binding sites (−PP7) ([Figure 3C](#fig3){ref-type="fig"}). mCherry intensity overlapping with translation sites (GFP spots) was measured and normalized to the intensity observed in nearby cytoplasmic regions with diffuse signal. The −PP7 construct exhibited a normalized mCherry intensity around 1, indicating that signal intensity overlapping with translation spots was similar to that of the surrounding cytoplasm. By contrast, the +PP7 containing reporter exhibited a significantly higher intensity, indicating that the mCherry signal overlapping translation sites was due to specific recognition of RNA molecules by the fused PP7 coat protein ([Figure 3C](#fig3){ref-type="fig"}). To assess whether GFP spots are indeed reflecting translation sites, images were acquired before and after treatment with puromycin ([Figure 3D](#fig3){ref-type="fig"}). RNA spots were identified in the images, and mCherry and GFP intensity in the corresponding regions was recorded. The normalized GFP intensity was calculated as a measure of the translational efficiency of each RNA spot. Puromycin treatment significantly reduced the GFP intensity of RNA spots, confirming that it reflects the presence of nascent protein chains ([Figure 3D](#fig3){ref-type="fig"}). To further determine whether this concentration of nascent chains reflects active translation, and not stalled ribosomes, we acquired images of localized translation reporters before and after treatment with harringtonine or lactimidomycin, which block initiating ribosomes but allow elongating ribosomes to run off ([@bib26]; [@bib32]). As an additional control we treated cells with cycloheximide, which stalls elongating ribosomes and prevents release of nascent chains. Indeed, brief 15 min treatment with harringtonine or lactimidomycin, but not cycloheximide, significantly reduced the observed translation signal indicating that it reflects the presence of actively translating ribosomes ([Figure 3---figure supplement 1](#fig3s1){ref-type="fig"}). RNAs targeted to protrusions are similarly translated in both internal and peripheral locations {#s2-3} ----------------------------------------------------------------------------------------------- We further imaged cells expressing either control or localized reporters (carrying the Rab13 or Pkp4 UTRs) ([Figure 4A](#fig4){ref-type="fig"} and [Videos 2](#video2){ref-type="video"}--[4](#video4){ref-type="video"}). Imaging was performed ca. 2 hr after doxycycline induction to ensure that reporter RNAs do not accumulate to high levels and do not deplete the cytoplasmic pools of scFv-GFP antibody and 3x-mCherry-PCP ([Figure 4---figure supplement 1A](#fig4s1){ref-type="fig"}). Consistent with the ability of the Rab13 and Pkp4 UTRs to target RNAs to protrusions ([@bib36]; [@bib55]), a higher proportion of localized reporter RNAs (containing the Rab13 or Pkp4 UTRs) were observed closer to the periphery compared to the control reporter ([Figure 4B](#fig4){ref-type="fig"}). Around 50% of observed localized reporter molecules were found within 3--7 μm of the cell edge, compared to 15 μm for the control reporter ([Figure 4B](#fig4){ref-type="fig"}). For all reporters, the majority of RNA particles exhibit mCherry intensities centered around a single peak indicating that they largely exist as single molecules ([Figure 4---figure supplement 2](#fig4s2){ref-type="fig"}) (but see also below). The number of particles detected by live-cell imaging is lower than the number detected in fixed cells by FISH ([Figure 4---figure supplement 1B](#fig4s1){ref-type="fig"}). The additional RNAs detected by FISH likely correspond to fast-moving molecules that cannot be discerned during live imaging with our current acquisition speed. Observing RNAs during short time-lapse imaging (\~20--30 s) reveals that the majority of RNAs are static or exhibit an oscillatory type of motion ([Videos 2](#video2){ref-type="video"}--[4](#video4){ref-type="video"}). A small subset exhibits short directed movements that might be indicative of active transport ([Videos 5](#video5){ref-type="video"} and [6](#video6){ref-type="video"}; and see below). These observations are consistent with the motion characteristics described for other localized transcripts ([@bib19]; [@bib20]; [@bib42]; [@bib62]). Given the limitations in imaging and discerning fast-moving molecules, we have not attempted to analyze the transport kinetics of our reporter RNAs. We have rather focused our analysis here on the less mobile molecules that we can confidently identify and analyze (see Materials and methods for details on image acquisition and analysis). Despite the lack of kinetic information, this analysis reflects the behavior of a substantial fraction of the existing RNA population and can provide a valuable characterization of the translation properties of individual RNA molecules in a spatial manner. {#fig4} ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP (red) and the control translation reporter. Frames were acquired sequentially and with no time delay, for the duration of the movie (13 s). A merged image of the two channels is shown. Blue line: cell outline. Scale bar: 5 μm. Single frames of this movie are shown in [Figure 4A](#fig4){ref-type="fig"} (upper panels). 10.7554/eLife.44752.013 ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP (red) and the localized translation reporter carrying the Pkp4 UTR. Frames were acquired sequentially and with no time delay, for the duration of the movie (36 s). A merged image of the two channels is shown. Blue line: cell outline. Scale bar: 5 μm. Single frames of this movie are shown in [Figure 4A](#fig4){ref-type="fig"} (middle panels). RNAs are translated both in the periphery and internal regions. 10.7554/eLife.44752.014 ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP (red) and the localized translation reporter carrying the Rab13 UTR. Frames were acquired sequentially and with no time delay, for the duration of the movie (63 s). A merged image of the two channels is shown. Blue line: cell outline. Cyan line: nucleus outline. Scale bar: 5 μm. Single frames of this movie are shown in [Figure 4A](#fig4){ref-type="fig"} (bottom panels). RNAs are translated both in the periphery and internal regions. 10.7554/eLife.44752.015 ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP and the localized translation reporter carrying the Rab13 UTR. Images of the GFP channel were acquired sequentially and with no time delay, for the duration of the movie (13 s). Blue line: cell outline. Scale bar: 3 μm. Single frames of this movie are shown in [Figure 4---figure supplement 3A](#fig4s3){ref-type="fig"}. The edge of the protrusion is towards the top. One of the observed translation sites moves in an apparently directed manner towards the edge of the protrusion. 10.7554/eLife.44752.016 ###### NIH/3T3 cell expressing scFv-GFP (green), 3x-mCherry-PCP and the localized translation reporter carrying the Pkp4 UTR. Images of the GFP channel were acquired sequentially and with no time delay, for the duration of the movie (19 s). Blue line: cell outline. Scale bar: 3 μm. Single frames of this movie are shown in [Figure 4---figure supplement 3B](#fig4s3){ref-type="fig"}. The edge of the protrusion is towards the right. One of the observed translation sites moves in an apparently directed manner towards the edge of the protrusion. 10.7554/eLife.44752.017 To address whether transport to the periphery is accompanied by changes in the translation state of the RNAs, we determined the translation efficiency of single RNA molecules as a function of their distance from the periphery. Consistent with the stochastic translation bursts reported in other systems ([@bib44]; [@bib57]; [@bib58]), for all three reporters, single RNAs exhibit a range of translation efficiencies ([Figure 4A](#fig4){ref-type="fig"}). Interestingly, however, plotting the translation efficiency of individual RNAs in relation to their position from the cell edge revealed that translation efficiency is not affected by the distance from the periphery ([Figure 4A](#fig4){ref-type="fig"}). Pearson's correlation coefficients in all cases are close to 0 (-0.07 (Rab13), −0.06 (Pkp4), −0.01 (Control)). This result is in agreement with the observations reported above showing that mislocalization of endogenous RNAs does not impact on their translation ([Figures 1](#fig1){ref-type="fig"} and [2](#fig2){ref-type="fig"}). Therefore, RNAs trafficked to the periphery through the pathway supported by the Rab13 and Pkp4 UTRs can be translated with similar efficiency in both peripheral and internal locations. We point out that our analysis is focused on less mobile molecules. We cannot currently assess if translation persists, or not, during periods of active movement. In this regard, we have observed single RNAs that undergo short-range directed movements while being translationally active ([Figure 4---figure supplement 3](#fig4s3){ref-type="fig"} and [Videos 5](#video5){ref-type="video"} and [6](#video6){ref-type="video"}). This suggests that these RNAs can be translated while in transit, an observation also noted in other systems ([@bib57]). However, due to current limitations on the speed and duration of our imaging, observation of such events is too sporadic to allow conclusions about their frequency. APC-dependent RNAs associate with heterogeneous clusters at the tips of protrusions {#s2-4} ----------------------------------------------------------------------------------- The data mentioned above have focused on single RNA molecules. However, in the course of our studies we have observed that endogenous APC-dependent RNAs exist in two states: they are found either as single molecules or as clusters made up of multiple RNAs in close spatial proximity. We have observed these clusters for virtually all APC-dependent RNAs we have detected by in situ hybridization ([Figure 5](#fig5){ref-type="fig"} and [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}). These RNA clusters are found at the very tips of cellular protrusions ([Figure 5A--E](#fig5){ref-type="fig"}). A small fraction of each individual RNA segregates into such clusters, in a 2μm-wide peripheral region ([Figure 5F--H](#fig5){ref-type="fig"}, [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}), and only a subset of protrusions exhibits such structures ([Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}; also see Figure 10 and below for further quantifications). When they form, these clusters are made up of multiple RNA molecules, as assessed by quantitation of FISH signal intensities ([Figure 5A--C](#fig5){ref-type="fig"} and [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}) and can contain different RNA species. Indeed, two distinct RNAs can be observed in the same RNA cluster ([Figure 5D](#fig5){ref-type="fig"}). Furthermore, most peripheral clusters of individual RNA species overlap with areas of accumulation of polyadenylated RNA, detected through oligo-dT hybridization ([Figure 5E,I](#fig5){ref-type="fig"}). Given that the individual detected RNAs (such as the Pkp4 RNA shown in [Figure 5E](#fig5){ref-type="fig"}) exist in relatively few copies per cell (around, or less than, a hundred copies per cell, [Figure 5---figure supplement 2](#fig5s2){ref-type="fig"}), this suggests that the visible accumulation of polyA RNA likely reflects the existence of additional RNA species at that location. Therefore, APC-dependent RNAs are found in heterogeneous RNA clusters at the tips of some protrusions. {ref-type="fig"}). Scale bars: 15 μm. (**H**) For each RNA, signal intensity histograms of all detected particles found within the 2μm-wide peripheral edge area, were used to group particles into single RNAs or RNA clusters (see [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"} ). Table lists number of particles in each category for the indicated RNAs. p-values based on Fisher's exact test against Arpc3 RNA. (**I**) Percent of overlap of the indicated RNA clusters with polyA clusters. n = number of particles observed in ca. 25 cells.](elife-44752-fig5){#fig5} This spatially-defined clustering behavior is not observed by RNAs that show a more uniform distribution in the cell body and are not targeted to protrusions ([Figure 5G,H](#fig5){ref-type="fig"} and [Figure 5---figure supplement 1](#fig5s1){ref-type="fig"}; compare APC-dependent RNAs to Arpc3 and P4hb RNAs). Furthermore, this clustering behavior is recapitulated by exogenous localized reporter RNAs. We had previously used reporter RNAs, which carry a series of MS2-binding sites for visualization in the presence of GFP-MS2 coat protein. Reporters carrying 3'UTRs of APC-dependent RNAs form clusters at protrusions whereas a reporter with a control 3'UTR does not ([@bib36]). With improved imaging systems we have now used these reporters for higher resolution visualization. MS2-tagged reporter constructs that carry the Rab13 or Net1 3'UTRs, are localized at protrusions, where they can be observed either as individual RNA particles or as clusters of particles at the tips of protrusions ([Figure 5---figure supplement 3](#fig5s3){ref-type="fig"}). Furthermore, time lapse imaging of MS2-labeled localized reporters reveals that smaller RNA particles are being trafficked towards and become incorporated into the larger peripheral clusters, further supporting the conclusion that these clusters are made up of multiple RNAs ([Videos 7](#video7){ref-type="video"} and [8](#video8){ref-type="video"}). Therefore overall, transport of APC-dependent RNAs to the periphery can be followed by clustering at the tips of protrusions and this behavior is mediated by signals within the 3'UTRs. ###### NIH/3T3 cell expressing tdMCP-GFP (green) and a MS2-reporter RNA carrying the Rab13 UTR. Images of the GFP channel were acquired sequentially and with no time delay, for the duration of the movie (29 s). A single frame of this movie is shown in [Figure 5---figure supplement 3](#fig5s3){ref-type="fig"}. Fainter spots reflect single RNA molecules. At the tip of the protrusion a brighter cluster of RNAs is observed. Single particles can be observed moving towards and incorporating into the cluster at the tip (at frames around seconds 3--4 and second 21). Scale bar: 5 μm. 10.7554/eLife.44752.022 ###### NIH/3T3 cell expressing tdMCP-GFP (green) and a MS2-reporter RNA carrying the Net1 UTR. Images of the GFP channel were acquired sequentially and with no time delay, for the duration of the movie (22 s). A single frame of this movie is shown in [Figure 5---figure supplement 3](#fig5s3){ref-type="fig"}. Fainter spots reflect single RNA molecules. At the edges of the protrusion multiple brighter clusters of RNAs are observed. Single particles from internal regions can be observed moving towards and incorporating into clusters at the tip. Scale bar: 5 μm. 10.7554/eLife.44752.023 RNA clusters at the tips of protrusions are translationally silent {#s2-5} ------------------------------------------------------------------ To investigate the translation status of RNAs within these clusters we employed the localized translation reporters. Consistent with the data described above, translation reporters carrying the Rab13 or Pkp4 UTRs could be found at the tips of protrusions in clusters containing multiple RNA molecules, as indicated by the increased mCherry intensity compared to the intensity exhibited by the single molecules in more internal regions ([Figure 6A,B](#fig6){ref-type="fig"}). Translation reporter clusters were not as pronounced as those observed for endogenous RNAs, likely because of the brief induction time. Nevertheless, strikingly, measuring the translational efficiency of localized reporter RNAs found in clusters revealed that for both reporters, carrying either the Rab13 or Pkp4 UTR, peripherally clustered RNAs are translationally silent ([Figure 6A,B](#fig6){ref-type="fig"}). Taken together all the above data suggest that, while APC-dependent RNAs are enriched in the periphery, they are translated with similar efficiency in both internal and peripheral locations. At the same time, a subset of them at the tips of protrusions coalesces into clusters which are translationally silent. {#fig6} Endogenous Rab13 RNA is translated in both internal and peripheral locations {#s2-6} ---------------------------------------------------------------------------- Our initial expectation was that RNAs at the periphery would be locally translated. While the above data show that this indeed happens, the additional observation of silencing at the periphery was rather counterintuitive. We thus first sought to validate that the above findings indeed reflect the regulation of endogenous localized transcripts. For this, we focused on Rab13 and employed the puro-PLA assay (puromycylation followed by proximity ligation amplification) to detect nascent Rab13 protein molecules in situ ([@bib53]). Puro-PLA relies on a brief (5 min) pulse of puromycin to label protein molecules that are being synthesized on ribosomes during the pulse period. Nascent Rab13 can then be visualized by PLA detection of the proximity between an anti-puromycin and an anti-Rab13 antibody ([Figure 7A](#fig7){ref-type="fig"}). Since puromycin causes chain termination and eventual release of nascent chains, even with a short pulse, it is conceivable that some of the detected signal might reflect released protein that has diffused away from the translation sites. This can be minimized by pretreatment with cycloheximide (CHX) which stalls the nascent proteins on the ribosomes without affecting puromycylation ([@bib14]; [@bib53]), thus promoting detection of in situ translation sites ([Figure 7A](#fig7){ref-type="fig"}). The specificity of the signal is validated in cells knocked down for the detected protein as well as in cells pretreated with the translation inhibitors anisomycin or harringtonine, which prevent puromycylation by interfering with the peptidyl-transferase activity or through ribosome run-off respectively. We note that we could not perform these experiments in NIH/3T3 mouse fibroblast cells, because even though the Rab13-puro PLA signal was dependent on translation (i.e. was reduced upon anisomycin treatment), it was not significantly reduced upon Rab13 knockdown, indicating that it mostly originated from non-specific binding of the Rab13 antibody in mouse cells (not shown). We thus used primary human dermal fibroblasts, in which Rab13 protein can be readily detected ([Figure 7---figure supplement 1](#fig7s1){ref-type="fig"}). Indeed, Rab13-puro PLA particles in these cells were significantly reduced upon both translation inhibition (with anisomycin or harringtonine) as well as Rab13 knockdown ([Figure 7B](#fig7){ref-type="fig"}). (Note that the majority of the remaining non-specific signal is concentrated around the nucleus). Furthermore, comparison of the Rab13-puro PLA signal between control and CHX-pretreated cells showed a small but consistent increase in control cells ([Figure 7B](#fig7){ref-type="fig"}). We interpret this increased number of particles in control cells to reflect the additional detection of a small amount of nascent Rab13 that is released and diffused away from translation sites. Inferring from these measurements, there are ca. 40 Rab13 translation sites per cell. This number is in good agreement with the ca. 40--50 Rab13 RNA molecules detected by FISH in these cells ([Figure 7C](#fig7){ref-type="fig"}). We conclude that we can specifically detect nascent Rab13 protein in human fibroblasts and that a large fraction of Rab13 RNAs are being actively translated at any moment. {#fig7} Consistent with our observations in mouse NIH/3T3 cells, primary human fibroblasts also significantly localize the Rab13 RNA to the periphery compared to the overall distribution of polyadenylated RNAs in the cytoplasm ([Figure 7C,D](#fig7){ref-type="fig"}). Visual inspection of Rab13 translation sites (Rab13-puro PLA signal) indicated that Rab13 translation occurs in both peripheral and perinuclear locations, consistent with the conclusions reached above. Since we cannot concomitantly detect Rab13 RNAs and Rab13 translation, we measured a peripheral distribution index (PDI) to describe the overall distribution in multiple cells ([@bib50]; [@bib55]). Intriguingly, this revealed that Rab13 translation signal was less peripheral compared to the Rab13 RNA signal ([Figure 7D,E](#fig7){ref-type="fig"}). While the perinuclear non-specific puro-PLA noise, mentioned above, might exaggerate this difference, it cannot fully account for it. Furthermore, we noticed that upon increased time of spreading on fibronectin-coated coverslips, even though the Rab13 RNA remained peripheral to the same extent ([Figure 7D](#fig7){ref-type="fig"}), Rab13 translation signal became even less peripheral ([Figure 7E](#fig7){ref-type="fig"}). The bias towards internal sites is not due to release and trafficking of newly-synthesized Rab13 away from translation sites, because we see the same values when puromycylation is performed after CHX pre-treatment to block nascent protein release ([Figure 7E](#fig7){ref-type="fig"}). Therefore, while these results showed that indeed endogenous Rab13 is translated in both internal and peripheral locations, they also accentuated the paradox of silencing RNAs after transporting them to the periphery and showed that this effect is enhanced over time during cell spreading. Peripheral Rab13 RNA is silenced at retracting protrusions {#s2-7} ---------------------------------------------------------- To try to understand this paradox we looked at the spreading rate of cells around the time points used for translation site imaging. We noticed that the reduction in peripherally translated Rab13 is associated with a decrease in spreading rate ([Figure 7F](#fig7){ref-type="fig"}). In other words, peripheral translation is reduced as the cells stop extending. Exploring this idea more in dermal fibroblasts was difficult since protrusions of these cells are very dynamic and at any given moment can be extending, retracting or switching behaviors ([Videos 9](#video9){ref-type="video"} and [10](#video10){ref-type="video"}), thus not allowing us to unambiguously infer the extending or retracting state of a protrusion in fixed cells, and to correlate it with Rab13 translation. ###### Primary human dermal fibroblast expressing Lifeact-GFP. Images were acquired every 6 min for a total of 3 hr. Note that each protrusion is very dynamic undergoing retracting and extending phases within minutes. 10.7554/eLife.44752.027 ###### Primary human dermal fibroblast expressing Lifeact-GFP. Images were acquired every 6 min for a total of 3 hr. Note that each protrusion is very dynamic undergoing retracting and extending phases within minutes. 10.7554/eLife.44752.028 For this reason, we turned to MDA-MB-231 breast cancer cells. These cells, when migrating on collagen-coated glass, exhibit a characteristic tail which retracts over a few minutes ([Figure 8A](#fig8){ref-type="fig"} and [Videos 11](#video11){ref-type="video"} and [12](#video12){ref-type="video"}). We define these tails as protruding regions that contain actin stress fibers and do not have detectable cortical ruffles visualized through Lifeact-GFP expression in live cells (or phalloidin staining of fixed cells). Live imaging of multiple cells revealed that, once formed, these regions are consistently retracting (90% of 130 protrusions exhibiting retraction in 51 cells). Therefore, visualization of such protrusions allows us to identify with high confidence, even in fixed cells, areas of cell retraction. {ref-type="video"}. (**B**) In situ hybridization of Rab13 and polyadenylated (polyA) RNA in MDA-MB-231 cells and PDI quantitations. Arrows point to Rab13 RNA in retracting protrusions. (**C**) Quantitation of Rab13-puro-PLA signal in MDA-MB-231 cells under the indicated conditions. (**D**) Representative images of Rab13-puro-PLA and phalloidin staining in MDA-MB-231 cells exhibiting retracting protrusions. Note that Rab13-puro-PLA signal is absent in retracting protrusions (arrows). (**E**) Rab13-puro-PLA intensity in lamellipodia or retracting protrusions. (See [Figure 9A](#fig9){ref-type="fig"} for representative outlines). (**F**) Percent of retracting protrusions positive for Rab13 RNA or puro-PLA signal based on images such as those shown in (**B**) and (**D**). Error bars: standard error. Number of cells analyzed in 2--3 independent experiments are shown within each bar. p-value: \*\*\<0.01, \*\*\*\<0.001, \*\*\*\*\<0.0001 by Student's t-test (**B, E**) or one-way ANOVA with Dunnett's multiple comparisons test, compared to control (**C**). Scale bars: 10 μm.](elife-44752-fig8){#fig8} ###### MDA-MB-231 cell expressing Lifeact-GFP. Images were acquired every 3 min for a total of 51 min. Arrow points to protrusion that retracts over a period of few minutes. Note that lamellipodial regions undergo constant dynamic retracting and extending phases. 10.7554/eLife.44752.030 ###### MDA-MB-231 cell expressing Lifeact-GFP. Images were acquired every 3 min for a total of 51 min. Arrow points to protrusion that retracts over a period of few minutes. Note that lamellipodial regions undergo constant dynamic retracting and extending phases. 10.7554/eLife.44752.031 Importantly, in MDA-MB-231 cells, the Rab13 (and other protrusion-localized RNAs) are also peripherally localized and can be found both in front lamellipodia as well as in retracting protrusions ([Figure 8B](#fig8){ref-type="fig"}). Rab13-puro PLA signal was also readily detected in these cells and was significantly reduced upon both translation inhibition (with anisomycin or harringtonine) as well as Rab13 knockdown ([Figure 8C](#fig8){ref-type="fig"}). We thus imaged Rab13 RNA, as well as nascent Rab13 protein, in cells exhibiting retracting protrusions ([Figure 8B,D](#fig8){ref-type="fig"}). We focused our analysis on retracting protrusions and front lamellipodial regions ([Figure 9A](#fig9){ref-type="fig"}), which despite being quite dynamic exhibit overall net extension ([Videos 11](#video11){ref-type="video"} and [12](#video12){ref-type="video"}). Significantly, Rab13 translation was readily detected in lamellipodia but was drastically reduced in retracting protrusions ([Figure 8E](#fig8){ref-type="fig"}). This was not due to a difference in Rab13 RNA present in these regions ([Figures 8B](#fig8){ref-type="fig"} and [9C](#fig9){ref-type="fig"}). Indeed, while almost all retracting protrusions contained Rab13 RNA, the majority of them were negative for Rab13 translation ([Figure 8F](#fig8){ref-type="fig"}), strongly indicating that Rab13 RNAs are translationally silenced at retracting protrusions. Therefore, the resolution of the paradox that we propose is that Rab13 and likely other APC-dependent RNAs are translated in extending protrusions/lamellipodia and are silenced upon retraction. {#fig9} Silenced Rab13 RNA at retracting protrusions can be found in heterogeneous clusters {#s2-8} ----------------------------------------------------------------------------------- In mouse fibroblasts we had observed an association of translationally silent RNAs with peripheral heterogeneous clusters ([Figures 5](#fig5){ref-type="fig"} and [6](#fig6){ref-type="fig"}). To determine whether such clustering is also observed for silent RNAs found in retracting protrusions of MDA-MB-231 cells, we imaged the Rab13 RNA and analyzed the observed particles either in translationally-active lamellipodial regions or in the translationally-silent retracting protrusions ([Figure 9A](#fig9){ref-type="fig"}). A frequency distribution histogram revealed that, in lamellipodia, the majority of Rab13 RNA particles exhibit intensities centered around a single peak indicating that they largely exist as single molecules. By contrast, in retracting protrusions a significant number of particles were larger and exhibited increased intensities indicating that they correspond to clusters of multiple Rab13 RNA molecules ([Figure 9B,C](#fig9){ref-type="fig"}). Therefore, corroborating our previous observations, translationally silent Rab13 RNA is found in multimeric clusters. Furthermore, since retracting protrusions exhibit a large fraction of single Rab13 RNAs, these results additionally indicate that containment within clusters is not required for silencing but might be a consequence of it. To determine whether RNA clusters in retracting MDA-MB-231 protrusions were composed of heterogeneous RNA species, we tested for the co-localization of Rab13 with polyA RNA. As mentioned above, in fibroblast protrusions, the majority of individual RNA clusters overlap with areas of visible concentration of polyA RNA ([Figure 5E,I](#fig5){ref-type="fig"}). We observed something different in MDA-MB-231 cells ([Figure 9D](#fig9){ref-type="fig"}). Examination of multiple protrusions revealed that 29.7 ± 4.6% of protrusions (n = 60) contained clusters of Rab13 RNA without any corresponding accumulation of polyA RNA signal, while in 27.9 ± 6.4% of protrusions Rab13 RNA appeared to be part of more heterogenous, polyA-positive assemblies. Considering that polyA intensity is reflective of the amount of contained RNAs, it therefore appears that MDA-MB-231 cells form a spectrum of clusters of different sizes. Given that the lifetimes of retracting protrusions are noticeably different in the two cell types (with NIH/3T3 protrusions persisting for a longer time, while MDA-MB-231 protrusions quickly retracting within a few minutes; compare [Videos 11](#video11){ref-type="video"} and [12](#video12){ref-type="video"} to [Video 13](#video13){ref-type="video"}), a likely possibility is that the observed differences reflect the dynamics of these granules (see discussion). ###### NIH/3T3 fibroblast expressing Lifeact-GFP. Images were acquired every 6 min for a period of 1 hr. Note that, in comparison to MDA-MB-231 cells, NIH/3T3 cells exhibit much slower dynamics with protrusions persisting relatively unchanged for the duration of imaging. 10.7554/eLife.44752.033 Formation of RNA clusters at protrusions is promoted by translational inhibition and requires microtubules {#s2-9} ---------------------------------------------------------------------------------------------------------- To probe more into the assembly of these peripheral clusters, we tested how their formation is affected upon translational inhibition or disruption of RNA transport to the periphery. We performed these experiments in mouse fibroblast cells, in which peripheral clusters can be readily detected through polyA accumulation ([Figure 5I](#fig5){ref-type="fig"}). Indeed, in these cells, the local concentration of polyA RNA at protrusions is a more reliable identifier of peripheral RNA clusters compared to the detection of any one particular RNA species that constitutes them. This is especially true in the case of low-abundance RNAs, such a Pkp4, for which a large proportion of protrusions exhibit visible polyA RNA clusters that either contain single Pkp4 RNA molecules ([Figure 10---figure supplement 1](#fig10s1){ref-type="fig"}; white arrows; 45.9% (n = 98)), or are devoid of Pkp4 RNA ([Figure 10---figure supplement 1](#fig10s1){ref-type="fig"}; yellow arrowheads; 41.8% (n = 98)). Therefore, at least for 3T3 cells, polyA RNA reveals more accurately the presence of peripheral clusters. {#fig10} Scoring of cell populations for the presence of polyA granules at protrusions revealed that 25--30% of protrusions contain peripheral polyA granules ([Figure 10](#fig10){ref-type="fig"}). Interestingly, inhibition of translation doesn't disrupt their formation, contrasting with the behavior exhibited by other types of cytoplasmic RNA granules, such as stress granules, which are rapidly dissolved upon inhibition of translation elongation. Instead, for peripheral polyA granules, brief treatment with cycloheximide or puromycin promotes their appearance ([Figure 10](#fig10){ref-type="fig"}), suggesting that translational inhibition is a limiting step in the formation of RNA clusters at protrusions. We further disrupted the microtubule network with nocodazole, or specifically disrupted detyrosinated microtubules using the tubulin carboxypeptidase inhibitor parthenolide ([Figure 10](#fig10){ref-type="fig"}). Of note, both treatments result in gradual retraction of protrusions and consequently in an increase in the number of cells with rounded morphology where peripheral RNA granules cannot be detected. To discount the possibility that the absence of peripheral granules was a secondary effect of the observed changes in cell shape, we focused our analysis only on cells that maintained cell protrusions and had similar morphology to the control conditions ([Figure 10](#fig10){ref-type="fig"}). Even looking at this narrower group of cells revealed that disruption of microtubules, or detyrosinated microtubules, significantly reduced the appearance of polyA RNA granules at protrusions ([Figure 10](#fig10){ref-type="fig"}). These results, together with the data presented above, are consistent with the model that peripheral RNA granules result from, and require, RNA transport to the periphery. Discussion {#s3} ========== Here, we investigate the translational regulation of APC-dependent RNAs which are targeted to cell protrusions. We find that the cytoplasmic position of APC-dependent RNAs does not affect their translation, since they can be translated similarly in both internal and peripheral locations. Rather, translation of APC-dependent RNAs is coordinated with specific peripheral cellular processes, being activated at extending protrusions/lamellipodia and suppressed upon protrusion retraction. Silencing is coupled to a change in the physical state of the RNAs manifested as single RNAs clustering into heterogenous granules at the tips of protrusions. This mode of regulation is distinct from the one proposed for several other localized transcripts, whereby RNAs are transported in a silenced state and are translationally activated only upon reaching the final destination or upon receipt of specific signals ([@bib4]; [@bib7]; [@bib30]). A reason for this latter type of regulation has been proposed to be the need to prevent protein appearance at sites, or times, where it might be deleterious. Indeed, premature appearance of the transcriptional repressor Ash1p in the mother cell during budding suppresses transcriptional programs in both mother and daughter cells and prevents mating-type switching ([@bib33]). Similarly, disrupting the timing of translational activation within neuronal axons or dendrites can lead to aberrant axonal pathfinding or synaptic responses ([@bib10]; [@bib24]; [@bib29]). Our observation of translation regardless of cytoplasmic location suggests that the proteins encoded by APC-dependent RNAs can be produced in internal regions without deleterious effects. We propose that this mode of regulation could additionally have functional implications for the encoded proteins. Specifically, translation in local environments can attribute proteins with different properties. This could result from differential protein modifications or through proximity to protein partners, which during co-translational assembly could affect the type or efficiency of multimeric complex formation ([@bib3]; [@bib30]; [@bib48]; [@bib49]). In light of these ideas, we suggest that a single mRNA, as it is being continuously translated at various stages during its transport to the periphery, could give rise to protein copies which have different properties, and therefore functional potential, depending on the local micro-environment they are translated into. We additionally show that translation of APC-dependent RNAs is specifically suppressed in retracting protrusions and this silencing is associated with the formation of multimeric heterogeneous clusters. Formation of these clusters is reduced by treatments that prevent RNA transport to the periphery and is increased upon global translational inhibition, suggesting that silencing is a limiting step in their formation. Nevertheless, silencing can occur outside of clusters, since a large proportion of RNAs are observed as single particles in retracting, translationally-silent protrusions. The appearance of these heterogenous RNA clusters at the tips of protrusions, and their relation to translation, are reminiscent of other types of RNA granules formed by liquid-liquid phase separation ([@bib13]; [@bib56]). For example, stress granules (SGs) and processing bodies (PBs) are sites of dynamic concentration of RNA molecules, but in both cases translational silencing or decay can occur regardless of RNA localization within granules ([@bib22]; [@bib40]; [@bib43]). SGs and PBs form throughout the cytoplasm under normal conditions or in response to stress. An interesting distinction of the clusters described here is that their formation is induced in particular subcellular regions associated with protrusion retraction, suggesting that their assembly/disassembly is controlled by spatial signals. The silencing of Rab13, and likely of other APC-dependent RNAs, at retracting protrusions is contrasted by their translational activation in extending lamellipodia. Given that localization of APC-dependent RNAs to the periphery is important for cell migration ([@bib55]) these data could point to a functional role for spatially segregating translation such that local protein production occurs in actively extending regions, while being suppressed in retracting areas. This would imply the existence of a dynamic regulatory mechanism that coordinates APC-dependent RNA translation with the continuous protrusion and retraction cycles that characterize cellular migration ([@bib28]; [@bib52]). A potential underlying mechanism for local silencing and granule formation could rely on spatially restricted phosphorylation/dephosphorylation events, which have been shown to affect the propensity of RNA-binding proteins to form phase-separated granules or to bind to RNAs ([@bib37]; [@bib39]; [@bib51]). In this regard, it is interesting that proteins associating with APC-dependent RNAs include the translational regulator FMRP and the RNA-binding protein FUS, one of the paradigm proteins used in understanding phase transitions ([@bib36]; [@bib60]; [@bib59]). Local modifications of FMRP or FUS could underlie the observed regulation. Additional local events, including local maturation of miRNAs, could be envisioned ([@bib46]). With regards to the assembly/disassembly dynamics of peripheral clusters and the fate of the sequestered RNAs, limitations in the duration of live-imaging we can accomplish do not permit concrete conclusions. However, our current data indicate that the extent of cluster formation might be influenced by the rate of retraction. Specifically, in NIH/3T3 cells, the majority of peripheral clusters are characterized by a visible accumulation of polyA signal, indicative of a high concentration of heterogeneous RNA species. By contrast, a substantial fraction of retracting MDA-MB-231 protrusions do not exhibit obvious polyA accumulation ([Figure 9](#fig9){ref-type="fig"}). A prominent difference between the two cell types is the speed with which protrusions retract. Contractile protrusions of NIH/3T3 cells can persist for a long time, while MDA-MB-231 protrusions quickly retract within a few minutes (compare [Videos 11](#video11){ref-type="video"} and [12](#video12){ref-type="video"} to [Video 13](#video13){ref-type="video"}). While we cannot rule out other interpretations, such as the existence of deadenylated transcripts, we favor the idea that slowly retracting protrusions allow for the build-up of large heterogenous granules, while in faster retracting protrusions peripheral granules are transient, and rapidly disassemble either through RNA degradation or release into the cytosol. The existence of translationally silent multimeric RNA clusters also offers a potential explanation for the slight decrease of APC-dependent RNAs found in the heavy RNP fraction under conditions that disrupt transport to the periphery ([Figures 1D](#fig1){ref-type="fig"} and [2B](#fig2){ref-type="fig"}). Translationally silent, higher-order RNP complexes can sediment at sucrose density gradient fractions heavier than their translated counterparts ([@bib9]). By analogy, the peripheral clusters of APC-dependent RNAs likely account for some of the RNA found at heavier fractions of the gradient. Reduction of their formation upon parthenolide treatment ([Figure 10](#fig10){ref-type="fig"}) could likely account for the apparent shift of APC-dependent RNAs towards the lighter polysome fraction of the gradient ([Figures 1D](#fig1){ref-type="fig"} and [2B](#fig2){ref-type="fig"}). We had previously reported that APC-dependent RNAs can also be found in internal cytoplasmic granules induced by expression of FUS mutants carrying ALS-associated mutations. Intriguingly, in these FUS-granules APC-dependent RNAs are translationally active ([@bib59]). Taken together, these observations raise the possibility that, at least for APC-dependent RNAs, their physical partitioning into granules is not the sole determinant, but acts in combination with the particular local environment to determine the eventual impact on their translation status. It would be interesting to further investigate how the composition of peripheral polyA granules differs from other internal cytoplasmic RNA granules with regards to both RNA and protein constituents. We note that the above study describes the regulation of APC-dependent RNAs in migrating mesenchymal cells. It would be interesting to explore how translation and transport of this RNA group is carried out in larger and more stably polarized cells such as neurons. Overall, we describe here a distinct mode of translational regulation of localized RNAs. These findings provide a different perspective towards understanding how local translation can influence protein activities and how these regulatory mechanisms could be integrated with dynamic cellular behaviors. Materials and methods {#s4} ===================== ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Reagent type\ Designation Source or reference Identifiers Additional\ (species) or\ information resource ---------------------------- --------------------------------------------------------------- ------------------------------------------------------ -------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Cell line (*Homo sapiens*) MDA-MB-231 ATCC ATCC Cat\# HTB-26, RRID: [CVCL_0062](https://scicrunch.org/resolver/CVCL_0062) Cell line (*Mus musculus*) NIH/3T3 ATCC RRID:[CVCL_0594](https://scicrunch.org/resolver/CVCL_0594) Cell line (*Homo sapiens*) Primary dermal fibroblasts provided by Dr. Ramiro Iglesias-Bartolome (NCI, NIH) Cell line (*Mus musculus*) NIH/3T3-PCFvGdelta1-PP7 (control translation reporter) This study NIH/3T3 cells expressing SunTag-based control translation reporter. Cell line (*Mus musculus*) NIH/3T3-PCFvGdelta1-deltaPP7 (control translation reporter) This study NIH/3T3 cells expressing SunTag-based control translation reporter with deletion of the PP7 repeats. Cell line (*Mus musculus*) NIH/3T3-PCFvGdelta1-Rab13U (translation reporter/Rab13 UTR) This study NIH/3T3 cells expressing SunTag-based translation reporter with the mouse Rab13 UTR. Cell line (*Mus musculus*) NIH/3T3-PCFvGdelta1-Pkp4U (translation reporter/Pkp4 UTR) This study NIH/3T3 cells expressing SunTag-based translation reporter with the mouse Pkp4 UTR. Cell line (*Mus musculus*) NIH/3T3-tdMCP-GFP_pIND20-b24bs/Net1 (MS2 reporter/Net1 UTR) This study NIH/3T3 cells expressing MS2 reporter with the mouse Net1 UTR. Cell line (*Mus musculus*) NIH/3T3-tdMCP-GFP_pIND20-b24bs/Rab13 (MS2 reporter/Rab13 UTR) This study NIH/3T3 cells expressing MS2 reporter with the mouse Rab13 UTR. Antibody anti-Rab13 rabbit polyclonal Novus Biologicals NBP1-85799 (1:1,000 WB; 1:200 PLA) Antibody anti-GAPDH (14C10) rabbit monoclonal Cell Signaling 2118 (1:2,000 WB) Antibody anti-puromycin (3RH11) mouse monoclonal Kerafast EQ0001 (1:2,000 PLA) Recombinant DNA reagent pHR-tdPP7-3x-mCherry Addgene 74926 Recombinant DNA reagent pcDNA4TO-24xGCN4_v4-kif18b-24xPP7 Addgene 74928 Recombinant DNA reagent pcDNA4TO-24xGCN4_v4-kif18b Addgene 74934 Recombinant DNA reagent pHR-scFv-GCN4-sfGFP-GB1-NLS-dWPRE Addgene 60906 Recombinant DNA reagent pHR-scFv-GCN4-sfGFP-GB1-deltaNLS-dWPRE This study Plasmid expressing scFv against GCN4 peptide of the SunTag system, fused to sfGFP, without NLS. Derived from pHR-scFv-GCN4-sfGFP-GB1-NLS-dWPRE after introduction of stop codon before the NLS sequence.  rRcombinant DNA reagent pInducer 20 Addgene 44012 Recombinant DNA reagent Phage-ubc-nls-ha-tdMCP-gfp Addgene 40649 Recombinant DNA reagent pInducer20-24xGCN4_v4-kif18b-24xPP7-Rab13 UTR This study Dox-inducible translation reporter carrying the mouse Rab13 3\'UTR Recombinant DNA reagent pInducer20-24xGCN4_v4-kif18b-24xPP7-Pkp4 UTR This study Dox-inducible translation reporter carrying the mouse Pkp4 3\'UTR Recombinant DNA reagent pInducer20-beta24bs-Rab13 UTR This study Dox-inducible MS2 reporter carrying the mouse Rab13 3\'UTR Recombinant DNA reagent pInducer20-beta24bs-Net1 UTR This study Dox-inducible MS2 reporter carrying the mouse Net1 3\'UTR Sequence-based reagent All Stars Negative control siRNA Qiagen 1027281 Sequence-based reagent Rab13 siRNA, si-Rab13 \#8 Qiagen SI02662702 target sequence: 5'-ATGGTCTTTCTTGGTATTAAA-3' Sequence-based reagent FISH probes against mouse Net1 Thermo Fisher Scientific VB1-3034209 Sequence-based reagent FISH probes against mouse Cyb5r3 Thermo Fisher Scientific VB1-18647 Sequence-based reagent FISH probes against mouse Cenpb Thermo Fisher Scientific VB1-18648 Sequence-based reagent FISH probes against mouse Rab13 Thermo Fisher Scientific VB1-14374 Sequence-based reagent FISH probes against human Rab13 Thermo Fisher Scientific VA1-12225 Sequence-based reagent FISH probes against mouse Pkp4 Thermo Fisher Scientific VB4-600264 Sequence-based reagent FISH probes against human Kif18b Thermo Fisher Scientific VA6-3170686 Sequence-based reagent FISH probes against mouse Ddr2 Thermo Fisher Scientific VB1-14375 Sequence-based reagent FISH probes against mouse Arpc3 Thermo Fisher Scientific VB1-14507 Sequence-based reagent FISH probes against mouse P4hb Thermo Fisher Scientific VB6-15898 Sequence-based reagent Custom-made codeset NanoString Technologies Item \# 116000002 Commercial assay or kit Duolink In Situ Red kit Sigma Aldrich DUO92101 Commercial assay or kit ViewRNA ISH Cell Assay kit Thermo Fisher Scientific QVC0001 ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Plasmid constructs and lentivirus production {#s4-1} -------------------------------------------- Plasmids for translation reporters: pHR-tdPP7-3x-mCherry, pcDNA4TO-24xGCN4_v4-kif18b-24xPP7 and pcDNA4TO-24xGCN4_v4-kif18b were gifts from Marvin Tanenbaum (Addgene plasmids \#74926, 74928 and 74934 respectively). pcDNA4TO-24xGCN4_v4-kif18b-24xPP7 was used to introduce different mouse UTR sequences at EcoRI/AscI sites after the PP7 repeats, and the inserts encompassing the coding sequence and UTRs were transferred into pInducer 20 lentivector (gift of Stephen Elledge, Addgene plasmid \# 44012), using the Gateway LR clonase II Enzyme mix (Thermo Fisher Scientific, cat\# 11791--020) according to the manufacturer's instructions, to generate plasmids: pInducer20-24xGCN4_v4-kif18b-24xPP7-Rab13 UTR and pInducer20-24xGCN4_v4-kif18b-24xPP7-Pkp4 UTR. pHR-scFv-GCN4-sfGFP-GB1-NLS-dWPRE was a gift from Ron Vale (Addgene plasmid \# 60906). The NLS sequence in this construct was deleted and replaced with a stop codon, to generate pHR-scFv-GCN4-sfGFP-GB1-deltaNLS-dWPRE. Plasmids for MS2-GFP-reporters: Phage-ubc-nls-ha-tdMCP-gfp was a gift from Robert Singer (Addgene plasmid \# 40649). pcDNA3-based plasmids expressing the β-globin genomic sequence followed by 24xMS2 binding sites and different 3'UTRs, were previously described ([@bib36]). Inserts from these constructs were transferred into pInducer 20 lentivector for inducible expression, to generate plasmids pInducer20-beta24bs-Rab13UTR and pInducer20-beta24bs-Net1UTR. mEGFP-Lifeact-7 (gift of Michael Davidson; Addgene plasmid \# 54610) was transferred into pCDH-CMV-MCS-EF1-Puro (System Biosciences, cat \#CD510B-1) using NheI/NotI sites for virus production. Lentiviruses were produced in HEK293T cells cultured in DMEM containing 10% FBS and Penicillin/Streptomycin. HEK293T cells were transfected with lentivectors, together with packaging plasmids pMD2.G and psPAX2 using PolyJet In Vitro DNA transfection Reagent (SignaGen) for 48 hr. Harvested virus was precipitated with Polyethylene Glycol at 4 °C overnight. Western blot {#s4-2} ------------ For Western blot detection the following antibodies were used: anti-Rab13 rabbit polyclonal (Novus Biologicals, cat\# NBP1-85799) and anti-GAPDH rabbit monoclonal 14C10 (Cell Signaling Technology, cat\# 2118). Cell culture and generation of cell lines {#s4-3} ----------------------------------------- NIH/3T3 mouse fibroblast cells (ATCC) were grown in DMEM supplemented with 10% calf serum, sodium pyruvate and penicillin/streptomycin (Invitrogen) at 37°C, 5% CO~2~. The stable NIH/3T3 cell line that inducibly expresses the Pkp4 cUTR has been described before ([@bib55]). Primary human dermal fibroblasts were kindly provided by Dr. Ramiro Iglesias-Bartolome (Center for Cancer Research, NCI, NIH) and were cultured in DMEM supplemented with 10% fetal bovine serum, sodium pyruvate and penicillin/streptomycin (Invitrogen) at 37°C, 5% CO~2~. MDA-MB-231 breast cancer cells (ATCC) were grown in Leibovitz's L15 media supplemented with 10% fetal bovine serum and penicillin/streptomycin at 37°C in atmospheric air. Cell lines have been tested for mycoplasma and are free of contamination. To generate cell lines expressing translation reporters, NIH/3T3 cells were sequentially infected with lentiviruses expressing tdPP7-3x-mCherry and scFv-GFP; a subpopulation was isolated through fluorescence activated cell sorting (FACS) and clonal lines were derived. A line expressing uniformly low levels of mCherry and GFP was used to introduce the various pInducer 20-based reporter constructs, and stably expressing cells were selected with geneticin (Thermo Fisher Scientific). The derived cell lines are: PCFvGdelta1-PP7 (control translation reporter); PCFvGdelta1-deltaPP7 (control translation reporter with deletion of the PP7 repeats); PCFvGdelta1-Rab13U (translation reporter with Rab13 UTR); PCFvGdelta1-Pkp4U (translation reporter with Pkp4 UTR). Expression of the reporters was induced by addition of 1 μg/ml Doxycycline (Fisher Scientific) approximately 2--3 hr before imaging. To generate cell lines expressing MS2-reporters, NIH/3T3 cells were infected with lentivirus expressing tdMCP-GFP and GFP-expressing cells with low level of GFP expression were sorted by FACS. This stable population was infected with pInducer20-based reporter constructs carrying 24xMS2 binding sites, and stable lines were selected with geneticin (Thermo Fisher Scientific). Derived cell lines are: tdMCP-GFP_pIND20-b24bs/Net1 (MS2 reporter/Net1 UTR) and tdMCP-GFP_pIND20-b24bs/Rab13 (MS2 reporter/Rab13 UTR). Expression of the reporters was induced by addition of 1 μg/ml Doxycycline (Fisher Scientific) approximately 2--3 hr before imaging. For translation inhibition, cells were treated with 100 μg/ml cycloheximide (Sigma Aldrich, Cat\# 239763), 2 μg/ml harringtonine (LKT Labs, product ID H0169), 1 μM lactimidomycin (Fisher Scientific, cat \# 50-629-10001), 50 μg/ml anisomycin (Sigma Aldrich, Cat\# A5862), 100 μg/ml puromycin (Thermo Fisher Scientific, Cat\# A1113803). For knockdown experiments, 40 pmoles of siRNAs were transfected into cells with Lipofectamine RNAiMAX (Thermo Fisher Scientific, cat\# 13778--150) according to the manufacturer's instructions. Cells were assayed after three days. siRNAs used were: AllStars Negative control siRNA (cat\# 1027281) and si-Rab13 \#8 (cat\# SI02662702; target sequence: 5'-ATGGTCTTTCTTGGTATTAAA-3') from Qiagen. Protrusion/cell body isolation and RNA analysis {#s4-4} ----------------------------------------------- Protrusions and cell bodies were isolated from serum-starved cells plated for 2 hr on Transwell inserts equipped with 3.0 μm porous polycarbonate membrane (Corning) as previously described ([@bib55]). Briefly, 1.5 million cells were plated per 25 mm filter and 1 or three filters were used for cell body or protrusion isolation, respectively. LPA was added to the bottom chamber for 1 hr and the cells were fixed with 0.3% paraformaldehyde for 10 min. For isolating protrusions, cell bodies on the upper surface were manually removed by wiping with cotton swab and laboratory paper. The protrusions on the underside were then solubilized by immersing the filter in crosslink reversal buffer (100 mM Tris pH 6.8, 5 mM EDTA, 10 mM DTT and 1% SDS) and gentle scraping. Cell bodies were similarly isolated after manually removing protrusions from the underside of the membrane. The extracts were incubated at 70°C for 45 min and used for RNA isolation using Trizol LS (Thermo Fisher Scientific). For nanoString analysis, RNA samples were analyzed using a custom-made codeset and the nCounter analysis system according to the manufacturer's instructions. Polysome gradient analysis {#s4-5} -------------------------- Cells were plated the day before so that they were actively growing on the day of the assay. Cells were treated with 50 μg/ml cycloheximide for 30 min at 37°C and cytoplasmic extract was collected essentially as described in [@bib5]. 10--45% sucrose gradients were prepared using BioComp gradient master (BioComp Instruments, Canada) according to the manufacturer's protocol and centrifuged at 37,000 rpm for 2 hr at 4°C in a SW41Ti rotor. After centrifugation, gradients were fractionated, and UV absorbance profiles were recorded using the BioComp piston gradient fractionator (BioComp Instruments, Canada). Based on the recorded UV profiles, fractions were pooled into the four sections described in the text. 20 ng of in vitro transcribed GFP RNA was added to each pooled fraction to correct for RNA recovery. RNA was isolated using Trizol LS (Thermo Fisher Scientific). RNA levels were assessed by nanoString analysis (nanostring Technologies, Seattle, WA) using a custom-made codeset and the nCounter analysis system according to the manufacturer's instructions. Fluorescence in situ hybridization (FISH) {#s4-6} ----------------------------------------- FISH was performed with ViewRNA ISH Cell Assay kit (Thermo Fisher Scientific) according to the manufacturer's instructions. The following Affymetrix probe sets were used: *Net1* cat\# VB1-3034209, *Cyb5r3* Cat\# VB1-18647, *Cenpb* cat\# VB1-18648, mouse *Rab13* cat\# VB1-14374, human *Rab13* cat\# VA1-12225, *Pkp4* Cat\# VB4-600264, *Kif18b* cat\# VA6-3170686, *Ddr2* cat\# VB1-14375, *Arpc3* cat\# VB1-14507, *P4hb* cat\# VB6-15898. To detect polyA RNAs, LNA modified oligodT probes (30 nucleotides) labeled with ATTO 655 were added during hybridization, pre-amplification, amplification and last hybridization steps of ViewRNA ISH Cell Assay. Cell mask stain (Thermo Fisher Scientific) was used to identify the cell outlines. Samples were mounted with ProLong Gold antifade reagent (Thermo scientific). Puromycylation and proximity ligation detection (puro-PLA) {#s4-7} ---------------------------------------------------------- Cells plated on Fibronectin (Sigma, Cat\# F1141)- or Collagen IV (Sigma, Cat\# C5533)-coated coverslips were pre-treated (or not) with the indicated translation inhibitors for 15 min and incubated for 5 min with 100 μg/ml puromycin at 37C. Subsequent steps were based on the protocol by [@bib53] with some modifications. Specifically, after puromycin incubation cell were quickly placed on ice, rinsed with ice-cold PBS, incubated for 2 min on ice with permeabilization buffer (50 mM Tris-Cl, pH 7.5, 5 mM MgCl~2~, 25 mM KCl, 100 μg/ml cycloheximide, 0.15 mg/ml digitonin, 0.5 U/μl RNasin, and Halt protease inhibitor cocktail), washed twice with ice-cold permeabilization buffer and fixed with 4% formaldehyde in PBS for 15 min at room temperature. Cells were subsequently blocked in blocking buffer (5% goat serum in PBS) for 1 hr at 37C and incubated with a pair of primary antibodies diluted in blocking buffer (in a humidified chamber for 1.5 hr at room temperature). Antibodies used were anti-Rab13 (1:200; Novus Biologicals, cat\# NBP1-85799) and anti-puromycin 3RH11 (1:2,000, Kerafast, cat\# EQ0001). After washing, PLA probes were applied in 1:10 dilution using the diluent buffer provided in the Duolink In Situ Red kit (Sigma Aldrich, cat\# DUO92101). Incubations and subsequent ligation and amplification steps were performed according to the manufacturer's instructions. After the final washes, cells were post-fixed for 10 min at room temperature with 4% formaldehyde in PBS, stained with Alexa Fluor-488 phalloidin (Thermo Fisher, cat\# A12379) in blocking buffer for 30 min and mounted using Duolink PLA Mounting medium with DAPI. Imaging and image analysis {#s4-8} -------------------------- FISH and puro-PLA images were obtained using a Leica SP8 confocal microscope, equipped with a HC PL APO 63x oil CS2 objective. Z-stacks through the cell volume were obtained and maximum intensity projections were used for subsequent analysis. Surface plot profiles, number and intensity of detected particles, were derived using the 'Surface plot' and 'Analyze particles' functions of ImageJ software (version 2.0.0-rc-69/1.52 n). Calculation of PDI index was performed using a custom Matlab script. The code is described and is available in [@bib50]. For live imaging of Lifeact-expressing cells, cells were plated on LabTek chambered coverglass, in phenol red-free media and were imaged on a Leica SP8 confocal microscope equipped with HC PL APO 63x oil CS2 objective, at constant 37C temperature and 5% CO2 (for NIH/3T3 and human dermal fibroblasts) or atmospheric air (for MDA-MB-231 cells). The 488 nm laser line was used for illumination, z-stacks through the cell volume were acquired over time and maximal intensity projections were produced. For live imaging of RNA reporters, cells were plated on LabTek chambered coverglass, in phenol red-free media, and were imaged on a Zeiss LSM 780 confocal microscope, equipped with a Plan-Apochromat 63x/1.40 Oil M27 objective, at constant 37C temperature and 5% CO~2~. 488 nm and 561 nm laser lines were used for illumination and single z-plane images were acquired over time. Two channel imaging was performed with sequential line scanning. Acquisition speed was approximately 500 ms/frame. No detectable photobleaching was observed under these conditions during a 20--40 s acquisition period. Diffraction-limited spots were identified using a custom IDL software provided by Dr. Daniel Larson ([@bib12]). The integrated fluorescence intensity of each spot was calculated using a Gaussian mask fit after local background subtraction ([@bib12]). Distance of spots from a user-defined protrusive edge were derived using a custom Matlab script, available upon request. We note that because of low signal to noise of the RNA (mCherry) channel, we were not confident of our ability to identify fast moving particles. Thus, we limited our analysis to more stationary particles defined as those particles that persist over \>6 imaging frames. Funding Information =================== This paper was supported by the following grant: - http://dx.doi.org/10.13039/100000054National Cancer Institute Intramural Research Program of the Center for Cancer Research to Stavroula Mili. This work was supported by the Intramural Research Program of the Center for Cancer Research, NCI, National Institutes of Health (SM). Additional information {#s5} ====================== No competing interests declared. Designed and performed experiments, analyzed data and edited the manuscript. Designed and performed experiments, analyzed data and edited the manuscript. Formal analysis, Investigation, Methodology, Performed experiments and analyzed data. Performed experiments analyzed data and edited the manuscript. Supervision, Funding acquisition, Designed, performed experiments, analyzed data, supervised the study, wrote and edited the manuscript. Additional files {#s6} ================ 10.7554/eLife.44752.036 Data availability {#s7} ================= All data generated or analysed during this study are included in the manuscript and supporting files. 10.7554/eLife.44752.038 Decision letter Singer Robert H Reviewing Editor Albert Einstein College of Medicine United States 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 \"Translational regulation of protrusion-localized RNAs involves silencing after transport\" for consideration by *eLife*. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and James Manley as the Senior Editor. The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. The key to the work is the use of the SunTag system to identify localized mRNAs that have been or are being translated. The authors find that mRNAs appear to be translated as they are transported through the cytoplasm on their way to being localized, but then cluster and become translationally silenced. This goes against a prevailing view that mRNAs bound to the periphery are translationally repressed until they are localized. The reviewers find the work interesting but raise several concerns. The following significant concerns must be addressed in a revised manuscript, should you wish to resubmit: The reviewers agreed that the authors make generalizations that are not supported by the data. For instance, that they see mRNAs being translated while moving, these appear to be rare cases, and do not apply to all the mRNAs without necessary statistical analysis. The non-moving mRNAs may be more likely to translate, but there is a lack of quantification to evaluate this possibility. (They should cite the Wu et al. Science paper which also noted translating mRNAs in neurons-in fact all four of the single mRNA translation papers should be cited, not just Yan et al). Second, they do not consider alternative interpretations, for example that the translation is stalled until the mRNAs reach their destination, and the proteins are released there. More quantitation is needed before conclusions can be drawn as to polysome loading, for instance using qPCR. There is a noticeable lack of free protein, this seems to be a conundrum, along with the observation that translation is silenced exactly where it is needed. Discussion of the significance of this counter-intuitive observation is lacking. Finally, no consideration is given to the nature of the cells analyzed, and whether these results are an artifact of the cells or reporter chosen, The cells used, 3T3, are not good examples of cells with protrusive activity, or of polarized cells in general where localization is essential for motility (and it is known that immortalized cells localize mRNA poorly), and the reporter is not an endogenous gene with an endogenous promoter. They should test their system on primary cells to see if their observations hold there. So while the observations are potentially interesting they must be tempered by these specific constraints. The Discussion either needs to be softened, or the authors must provide additional data to address the generality of their observations. Finally, what conclusions the authors do want to draw on their specific system needs to be backed up by quantitation. *Reviewer \#1:* This manuscript by Yasuda et al. have employed single molecule imaging of mRNA reporters to uncover a novel and very interesting mechanism for translation regulation of protrusion-localized mRNAs in cultured fibroblasts. In contrast to other reports that indicate transported mRNAs are translationally repressed during transport and unsilenced at the destination, here the data show that translational repression occurs after transport and involves the clustering of multiple mRNAs into heterogenous granules. The present experiments using fluorescent reporters is complemented by a biochemical approach to perturb mRNA localization which does not seem to alter translation. The strength of this paper is the use of SunTag reporters for live cell imaging of mRNA localization and translation, which does support a model for the formation of translationally repressed mRNPs within protrusions. While this finding does advance the state of knowledge for the field, there are concerns with some of data and resulting interpretations that suggest lack of translational repression during active mRNA transport. It does seem that new experiments could provide stronger evidence of mRNA silencing during active transport. Specific comments: Figure 1 shows that disruption of endogenous APC-dependent mRNA localization to protrusions using a competitive Pkp4 3\'UTR reporter does not result in a shift of mRNA levels from polysome to mRNP fractions on sucrose gradients, in contrast to what was observed with puromycin treatment. The data need to be quantitated by real time PCR for several localized mRNAs. Specifically, the authors should report the actual values for% mRNA in each of the four fractions for several examples of localized mRNAs. While puromycin treatment resulted in marked mRNA shifts from heavy to light fractions, the competitive 3\'UTR does appear to accumulate mRNA in an intermediate fraction \#3. Another interpretation is that these are stalled ribosomes. This possibility has not been considered or discussed. Figure 2 -- the same comment as stated above also relates to use of pharmacologic perturbation of detyrosinated microtubules. Figure 4 shows data obtained with the Sun-Tag reporter implying that the majority of single mRNAs have comparable translation efficiencies regardless of localization. The authors conclude that \"RNAs targeted to protrusions are not silenced during transport.\" One major concern with this strong interpretation is that the quantitative analysis was not done on \"directed\" particles but rather on the majority of particles that were stationary or oscillatory. A qualitative example shown does suggest translation can occur during transport. Can an experiment be done to stimulate mRNA transport and increase the% of directed RNAs, for example, serum starvations followed by stimulation as done by Singer and colleagues for b-actin? Quantitative analysis of SunTag reporter dynamics can also be done following depolymerization of detyrosinated MTs. Without evidence on quantitation of mRNAs that are actively being transported, the authors will need to soften their conclusions. Such RNAs that are not moving and not localized may be translated, but this does not imply a similar fate for those that are actively being transported to the cell protrusion. In the Discussion, the authors should not generalize their findings in fibroblasts to all polarized cells, but rather restrict their interpretation to cells that localize mRNAs as single mRNPs. In neurons, for example, some types of mRNA granules contain multiple mRNAs which do appear to be silenced during transport. These different models should be compared. *Reviewer \#2:* Many localized mRNAs are translationally silenced during transport and become activated only upon reaching their cellular destination. In their manuscript, Yasuda et al. present evidence that a specific set of mRNAs, previously characterized as APC-dependent and associated with detyrosinated microtubules, is translationally active both during transport and at the cellular periphery. However, once at the periphery, a subset of APC-dependent RNAs coalesce into translationally inactive clusters which the authors speculate are needed for dynamic regulation during cell migration. The authors use translation reporters containing the SunTag peptide array and PP7 hairpins coupled with live cell imaging to reach the conclusion that the 3′ UTRs of several APC-dependent RNAs confer translation during transport and silencing in large clusters near the periphery. The clustering effect is also seen by FISH against endogenous APC-dependent RNAs. This paper provides an interesting example that runs counter to the dominant paradigm of translationally silent mRNA transport. However, some figures seem incomplete, for example lacking comparison to the control reporter. The discussion could also be made much stronger by further explaining the authors\' hypothesis about the function of the peripheral RNA clusters. Are the clusters an intermediate in an RNA degradation pathway or are they RNA storage granules with translationally-competent RNAs that can be dispersed in a regulated fashion? These points are discussed in more detail below. Major points: 1\) Figure 5 presents one of the most interesting findings of the paper -- that APC-dependent RNAs form clusters at protrusions. However, the presentation does not come across as rigorous and should be improved by: 1) quantification of cluster sizes and statement of number of cells analyzed and 2) comparison to the control reporter in the same figure. It seems that Figure S4 makes this argument in a more rigorous way. Perhaps part of Figure S4 could be moved to Figure 5 and part of Figure 5 could be moved to S4 (the polyA and the MS2-reporter panel could be moved to supplemental). It is additionally confusing that Figure S4 suggests that most RNAs are not in clusters but Figure 5 shows examples where almost no non-clustered RNAs can be seen. It seems like the images in Figure 5 must be the more exceptional examples. Therefore, some statement or quantification of the frequency of clusters from the live cell imaging is needed. 2\) Similar to the previous point, Figure 6 should state how many cells were analyzed and include the control reporter as a comparison. 3\) It would be interesting to see more analysis of polyA RNA clusters vs. APC-dependent RNA clusters. It is shown in one example for Pkp4 RNA in Figure 5E, but not quantified and not shown for other RNAs. I believe the authors\' hypothesis is that the RNA clusters are translationally repressed, but still translationally competent (i.e. not degradation products). If the authors could show that most APC-dependent RNA clusters are polyA+, this finding would support that hypothesis. 4\) Similar to point 4, the discussion does not explain the authors\' hypothesis about the function of the RNA clusters clearly and why they are translationally silenced. There is only one sentence suggesting that they reflect \'a local regulatory mechanism used to silence APC-dependent RNAs\'. It would be good to elaborate on this point and maybe offer a brief comparison to other types of translationally silent RNA granules. 10.7554/eLife.44752.039 Author response > The following significant concerns must be addressed in a revised manuscript, should you wish to resubmit: > > The reviewers agreed that the authors make generalizations that are not supported by the data. For instance, that they see mRNAs being translated while moving, these appear to be rare cases, and do not apply to all the mRNAs without necessary statistical analysis. The non-moving mRNAs may be more likely to translate, but there is a lack of quantification to evaluate this possibility. (They should cite the Wu et al. Science paper which also noted translating mRNAs in neurons-in fact all four of the single mRNA translation papers should be cited, not just Yan et al). We completely agree with these comments and realized that some of the conclusions in the original manuscript were awkwardly stated. Our intent was not to conclude that RNAs are being translated *while* engaged in active transport, but rather that they can be translated similarly in both internal and peripheral locations. We have changed the text throughout the manuscript to clarify this. We have also specifically pointed out that the translation reporter imaging assay does not allow us to make accurate conclusions about the behavior of fast-moving molecules (subsection "RNAs targeted to protrusions are similarly translated in both internal and peripheral locations") and have referenced the papers mentioned above. Furthermore, we have now used puro-PLA to detect in situ nascent Rab13 protein distribution (new Figures 7 and 8) (subsections "Endogenous Rab13 RNA is translated in both internal and peripheral locations" and "Peripheral Rab13 RNA is silenced at retracting protrusions"). This approach, apart from looking at the translation of an endogenous RNA, also circumvents the problem of fast-moving molecules by allowing us to look at translation in situ, in fixed samples. The results are fully consistent and support the conclusion that protrusion-localized RNAs can be translated in both peripheral and internal locations. > Second, they do not consider alternative interpretations, for example that the translation is stalled until the mRNAs reach their destination, and the proteins are released there. To address this possibility, we have performed run-off experiments, by brief treatment (15min) with harringtonine or lactimidomycin, and now show that the detected translation signals largely reflect active translation. Specifically, the new Figure 3---figure supplement 1 (described in subsection "Single-molecule translation reporters of protrusion-localized RNAs") shows that the signal detected by the translation imaging reporters is reduced almost to background levels upon harringtonine (or lactimidomycin), but not upon cycloheximide treatment, showing that it reflects active translation. Additionally, detection of Rab13-puro PLA signal is reduced to background level upon harringtonine treatment (compare harringtonine with anisomycin treatment) (Figure 7B and Figure 8C) showing that endogenous Rab13 is also mostly translated by actively scanning ribosomes (new Figures 7 and 8) (subsections "Endogenous Rab13 RNA is translated in both internal and peripheral locations" and "Peripheral Rab13 RNA is silenced at retracting protrusions"). We have also seen that RNAs fractionating in polysomal fractions of sucrose gradients are shifted towards the non-translating fractions of the gradient upon harringtonine treatment. We have seen this result for a few RNAs we have detected by RT-PCR in human breast cancer cells (as detailed below we see very similar mechanisms in all cell types tested). These results are shown in [Author response image 1](#respfig1){ref-type="fig"}. {#respfig1} Unfortunately, due to technical issues we have been unable to perform this experiment in mouse fibroblasts with the full panel of RNAs analyzed in Figures 1 and 2. However, given the effect of harringtonine seen in all other assays, we believe that even without this experiment we provide strong support for the conclusion that stalled ribosomes are not a significant part of the described mechanism. > More quantitation is needed before conclusions can be drawn as to polysome loading, for instance using qPCR. We believe that we might have not clearly conveyed that the heatmaps shown in Figures 1 and 2 are indeed averages of independent experiments. The associated p-values were meant to reflect whether the underlying distributions were different or not. We have now supplied, as source data, files with the exact values of all replicates used for the generation of the heatmaps and statistics in those figures. With regards to the use of qPCR, we have opted to quantify RNA abundance using nanoString analysis instead of RT-qPCR because it allows for direct RNA counting and thus avoids biases introduced by reverse transcription and amplification. This is now stated in subsection "Disrupting the localization of APC-dependent RNAs at protrusions does not affect their translation". Indeed, when we detect the spike RNA (which is added in equal amounts in the collected sucrose gradient fractions) we find that it is much more consistently and uniformly detected by nanoString analysis. RT-PCR detection shows much larger variations in the detected spike (even though the same amount was added). This underscores the variability and biases that can result from reverse transcription and amplification, which we think are exaggerated in samples with very different RNA complexities, such as those isolated from sucrose gradients. > There is a noticeable lack of free protein, this seems to be a conundrum, along with the observation that translation is silenced exactly where it is needed. Discussion of the significance of this counter-intuitive observation is lacking. With regards to the lack of free protein, with think that this is probably a perception caused by the contrast of some images, which makes the free protein not so evident. However, this is not a real issue with the assay. Even though we specifically induce the expression of the reporters for only a short period of time (2-4hrs), precisely to avoid accumulation of both the RNA and protein to high levels, we nevertheless always detect a significant amount of free protein, seen as fast-moving GFP particles that are brighter compared to the diffuse background of the single chain GFP-labeled antibody. We think this is evident, for example, in Figures 3, figure 3---figure supplement 1 and Figure 4. With regards to the conundrum of silencing translation at the periphery, we had tried to argue before that this reflects regulation of translation during dynamic processes, however we fully understand that more and better evidence was needed. Therefore, we have now provided substantial new evidence that we believe convincingly shows that silencing occurs at retracting protrusions. Thus, the resolution of the conundrum that we propose is that RNAs are translated in extending protrusions/lamellipodia and are silenced in retracting protrusions. These data are based on the use of puro-PLA to detect nascent Rab13 protein in situ and are detailed in the new Figures 7 and 8 (subsections "Endogenous Rab13 RNA is translated in both internal and peripheral locations" and "Peripheral Rab13 RNA is silenced at retracting protrusions"). > Finally, no consideration is given to the nature of the cells analyzed, and whether these results are an artifact of the cells or reporter chosen, The cells used, 3T3, are not good examples of cells with protrusive activity, or of polarized cells in general where localization is essential for motility (and it is known that immortalized cells localize mRNA poorly)... We have now provided data in primary human dermal fibroblasts as well as in breast cancer cells (MDA-MB-231). We note that for the group of APC-dependent RNAs we are studying, we see robust localization in all cell types we have tested (primary, immortalized or transformed). We specifically show that in all cell types tested we see very similar behaviors in terms of the localization of the studied RNAs at protrusions; their translation in both internal and peripheral regions; their silencing at the periphery; and their coalescence in heterogeneous granules (new Figures 7-9 and subsections "Endogenous Rab13 RNA is translated in both internal and peripheral locations"; "Peripheral Rab13 RNA is silenced at retracting protrusions" and "Silenced Rab13 RNA at retracting protrusions can be found in heterogeneous clusters"). We therefore believe that we provide strong evidence that the reported mechanism is not an artifact of the cell line used but is generally exhibited by mesenchymal migrating cells. 3T3s are indeed different than the other two cell types used, in that they tend to form more contractile stable protrusions whose dynamics are much slower (we have provided videos to illustrate that point; compare Videos 11 and 12 to Video 13). As we briefly discuss (Discussion paragraph seven), we believe these slower dynamics might be the reason for the more pronounced formation of peripheral granules in 3T3 cells compared to the granules formed in the much more dynamic MDA-MB-231s. We believe that all cell types can provide valid information and that each one, through accentuating certain behaviors, can offer a valuable model to facilitate the study of different aspects of these structures. > ... and the reporter is not an endogenous gene with an endogenous promoter. The point of how much any reporter reflects endogenous regulation is of course very valid. Even if a reporter recapitulates certain behaviors of endogenous RNAs (like we see for our reporters in terms of their peripheral localization and clustering), there is no guarantee that other aspects are also accurately reflecting the endogenous regulation. We find however, that it would be very difficult to ever provide unequivocal evidence that an exogenous reporter accurately reflects endogenous regulation, unless the regulation of the endogenous RNAs is first known. Therefore, instead of testing the effect of adding further elements (5'UTR, promoter, introns etc.) to our translation reporters, or of engineering aptamers into the endogenous genes, we rather chose to directly assess in situ the translation of an endogenous APC-dependent RNA. For this we used the puro-PLA method (puromycylation followed by proximity ligation amplification), which allows visualization of nascent proteins and translation sites of specific endogenous transcripts. In the new Figures 7-8 we provide controls for the validation of the assay in different cell types and use it to corroborate the findings obtained with the translation imaging reporters, as well as to extend them, showing that silencing occurs at retracting protrusions. We are aware that some of the conclusions we have reached with these experiments could be further extended by live imaging of translation reporters in the more dynamic MDA-MB-231 cells. While these are certainly very interesting experiments that we intend to pursue in the future, we hope that you appreciate that generation of such stable lines, appropriate for high quality imaging, are a lengthy endeavor. We believe that the current data fully support the conclusions of the manuscript and hope you will find that the information contained in it is warranting publication at this stage. > They should test their system on primary cells to see if their observations hold there. As mentioned above we have provided data in both primary cells and transformed cancer cells that are fully consistent with our previous conclusions. > So while the observations are potentially interesting they must be tempered by these specific constraints. The Discussion either needs to be softened, or the authors must provide additional data to address the generality of their observations. Finally, what conclusions the authors do want to draw on their specific system needs to be backed up by quantitation. We have added quantitations that we believe support all the claims made. Given the new data added, we have modified sections of the Abstract, Results and Discussion to reflect the new information included in the revised version. Please see manuscript text with tracked changes for exact modifications. Reviewer \#1: > \[...\] The strength of this paper is the use of SunTag reporters for live cell imaging of mRNA localization and translation, which does support a model for the formation of translationally repressed mRNPs within protrusions. While this finding does advance the state of knowledge for the field, there are concerns with some of data and resulting interpretations that suggest lack of translational repression during active mRNA transport. It does seem that new experiments could provide stronger evidence of mRNA silencing during active transport. > > Specific comments: > > Figure 1 shows that disruption of endogenous APC-dependent mRNA localization to protrusions using a competitive Pkp4 3\'UTR reporter does not result in a shift of mRNA levels from polysome to mRNP fractions on sucrose gradients, in contrast to what was observed with puromycin treatment. The data need to be quantitated by real time PCR for several localized mRNAs. Specifically, the authors should report the actual values for% mRNA in each of the four fractions for several examples of localized mRNAs. While puromycin treatment resulted in marked mRNA shifts from heavy to light fractions, the competitive 3\'UTR does appear to accumulate mRNA in an intermediate fraction \#3. Another interpretation is that these are stalled ribosomes. This possibility has not been considered or discussed. We believe that we might have not clearly conveyed that the heatmaps shown in Figures 1 and 2 are indeed averages of independent experiments expressing the% mRNA in each fraction. The associated p-values were meant to reflect whether the underlying distributions were different or not. We have now supplied, as source data, files with the exact values of all replicates used for the generation of the heatmaps in those figures. We have detailed in point 3 our rationale for using nanoString instead of RT-qPCR for quantitation of RNA levels. We have performed ribosome run-off experiments with harringtonine to show that stalled ribosomes are not a significant part of the described mechanism. Please see response to point 2 for details. > Figure 2 -- the same comment as stated above also relates to use of pharmacologic perturbation of detyrosinated microtubules. The heatmaps in this figure are also showing the averages of independent experiments expressing the% mRNA in each fraction. We have now supplied the source data with the exact values of all replicates used for the generation of the heatmaps in this figure. > Figure 4 shows data obtained with the Sun-Tag reporter implying that the majority of single mRNAs have comparable translation efficiencies regardless of localization. The authors conclude that \"RNAs targeted to protrusions are not silenced during transport.\" One major concern with this strong interpretation is that the quantitative analysis was not done on \"directed\" particles but rather on the majority of particles that were stationary or oscillatory. A qualitative example shown does suggest translation can occur during transport. Can an experiment be done to stimulate mRNA transport and increase the% of directed RNAs, for example, serum starvations followed by stimulation as done by Singer and colleagues for b-actin? Quantitative analysis of SunTag reporter dynamics can also be done following depolymerization of detyrosinated MTs. Without evidence on quantitation of mRNAs that are actively being transported, the authors will need to soften their conclusions. Such RNAs that are not moving and not localized may be translated, but this does not imply a similar fate for those that are actively being transported to the cell protrusion. We agree with this point and acknowledge that our conclusions were awkwardly stated. Please see response to point 1 for details. > In the Discussion, the authors should not generalize their findings in fibroblasts to all polarized cells, but rather restrict their interpretation to cells that localize mRNAs as single mRNPs. In neurons, for example, some types of mRNA granules contain multiple mRNAs which do appear to be silenced during transport. These different models should be compared. It would indeed be interesting to explore how these RNAs are regulated in neuronal cells. We have added in the Discussion a specific mention to that effect and a statement emphasizing that our findings are describing behaviors of mesenchymal migrating cells (Discussion penultimate paragraph). Reviewer \#2: > \[...\] The clustering effect is also seen by FISH against endogenous APC-dependent RNAs. This paper provides an interesting example that runs counter to the dominant paradigm of translationally silent mRNA transport. However, some figures seem incomplete, for example lacking comparison to the control reporter. The discussion could also be made much stronger by further explaining the authors\' hypothesis about the function of the peripheral RNA clusters. Are the clusters an intermediate in an RNA degradation pathway or are they RNA storage granules with translationally-competent RNAs that can be dispersed in a regulated fashion? These points are discussed in more detail below. > > Major points: > > 1\) Figure 5 presents one of the most interesting findings of the paper -- that APC-dependent RNAs form clusters at protrusions. However, the presentation does not come across as rigorous and should be improved by: 1) quantification of cluster sizes and statement of number of cells analyzed and 2) comparison to the control reporter in the same figure. It seems that Figure S4 makes this argument in a more rigorous way. Perhaps part of Figure S4 could be moved to Figure 5 and part of Figure 5 could be moved to S4 (the polyA and the MS2-reporter panel could be moved to supplemental). It is additionally confusing that Figure S4 suggests that most RNAs are not in clusters but Figure 5 shows examples where almost no non-clustered RNAs can be seen. It seems like the images in Figure 5 must be the more exceptional examples. Therefore, some statement or quantification of the frequency of clusters from the live cell imaging is needed. The confusion about the prevalence of clusters probably stems from the fact that Figure 5 showed enlargements of protrusive regions only, while Figure S4 showed whole-cell images and the associated quantifications. We have now noted this more obviously in both the text, figure legends and the figures themselves We have also reorganized the data of Figure 5 and Figure S4 (now named Figure 5---figure supplement 1) and added new panels in order to present quantitations and statistical comparisons in the main figure (Figure 5H, I) (subsection "APC-dependent RNAs associate with heterogeneous clusters at the tips of protrusions"). We have tried throughout to mention sample sizes used for analyses. > 2\) Similar to the previous point, Figure 6 should state how many cells were analyzed and include the control reporter as a comparison. We have included in the legend the number of cells imaged. We had not included the control reporter because we don't see any clusters. In any case, the point of this figure is not to provide a quantitative assessment of cluster formation between control and localized reporters, but to compare the translation status of RNAs within clusters to single RNAs within the same protrusions. > 3\) It would be interesting to see more analysis of polyA RNA clusters vs. APC-dependent RNA clusters. It is shown in one example for Pkp4 RNA in Figure 5E, but not quantified and not shown for other RNAs. I believe the authors\' hypothesis is that the RNA clusters are translationally repressed, but still translationally competent (i.e. not degradation products). If the authors could show that most APC-dependent RNA clusters are polyA+, this finding would support that hypothesis. We have added further quantitation regarding the coincidence of polyA signal with APC-dependent RNA clusters (Figure 5I) to support the use of polyA staining as an identifier of peripheral clusters in 3T3 cells. Indeed, the majority of APC-dependent RNA clusters are polyA+, and polyA staining can even reveal clusters that don't contain the particular APC-dependent RNA being imaged. While it is of interest, we cannot currently make any assessment regarding the fate of RNA within granules. As mentioned in point 6, we are aware that information relevant to these questions could be provided by live imaging of translation reporters in the more dynamic MDA-MB-231 cells. While these are certainly very interesting experiments that we intend to pursue in the future, we hope that you appreciate that they are too involved to pursue in the context of the current manuscript. For now, we have added a mention to these issues in the Discussion (paragraph seven). > 4\) Similar to point 4, the discussion does not explain the authors\' hypothesis about the function of the RNA clusters clearly and why they are translationally silenced. There is only one sentence suggesting that they reflect \'a local regulatory mechanism used to silence APC-dependent RNAs\'. It would be good to elaborate on this point and maybe offer a brief comparison to other types of translationally silent RNA granules. This point is now addressed more deeply with the addition of new data showing that silencing occurs at retracting protrusions (new Figures 7-9, subsections "Endogenous Rab13 RNA is translated in both internal and peripheral locations" and "Peripheral Rab13 RNA is silenced at retracting protrusions"). [^1]: These authors contributed equally to this work.

Introduction {#s1} ============ Placental dysfunction, associated with reduced rates of nutrient uptake ([@B1]--[@B3]), is a major cause of fetal growth restriction (FGR), the failure of a fetus to reach its growth potential ([@B4]). FGR is a significant risk factor for stillbirth ([@B5], [@B6]). Additionally, FGR infants demonstrate increased incidence of childhood diseases such as cerebral palsy, and adulthood diseases including heart disease, stroke, diabetes and osteoporosis ([@B7]--[@B11]). The current lack of therapies for FGR ([@B12]) emphasizes the need for better understanding of how fetal development is normally achieved and how it is dysregulated in FGR. Placental transfer of calcium increases over gestation to match fetal demand and ensure appropriate fetal skeletal mineralization ([@B13]). Poor fetal provision of calcium *in utero* has been linked with an increased risk of developing osteoporosis later in life ([@B14]). Maternofetal transfer of calcium across the placenta involves calcium moving from maternal blood into the syncytiotrophoblast (transporting epithelium of the placenta) down an electrochemical gradient through calcium permeable cation channels (e.g., TRPV6, transient receptor potential vanilloid type 6) on the maternal-facing microvillous membrane (MVM) ([@B15]--[@B17]). Once in the trophoblast cytosol calcium is buffered to avoid overly increasing the intracellular concentration, and shuttled to the fetal-facing basal membrane (BM) by calcium binding proteins such as calbindin-D~9K~ ([@B16]). Calcium is actively transported across the BM by plasma membrane calcium ATPases (PMCA) into the fetal compartment. The actions of PMCA help to maintain calcium concentrations in the fetus above those found in maternal blood. Unlike the activity of other nutrient, and especially amino acid, transport systems that are reduced in placentas of growth restricted fetuses ([@B1]--[@B3]), the activity of PMCA is increased in human FGR ([@B18]) as is the maternofetal transfer of calcium in a rodent model of FGR, the placental-specific insulin-like growth factor *2* (P0) knockout mouse ([@B19]). Optimal fetal growth depends on adequate nutrient delivery and placental supply can be adapted to meet the metabolic needs of the developing fetus. In pregnancies with normal outcomes, adaptation of placental transport in relation to placental size appears important in both women and mice ([@B20]--[@B22]). Our previous studies of wild-type (WT) mouse litters demonstrated that maternofetal calcium transfer across the lightest placentas is adaptively up-regulated, compared to the heaviest placentas, so that all fetuses, whether with relatively lighter or heavier placentas, accrue an appropriate level of calcium relative to their size near term ([@B22]). We suggested that this increased maternofetal transfer of calcium (per gram placenta), which coincides with increased placental calbindin-D~9K~ expression at embryonic day (E)18.5, is an example of a placental adaptation that promotes fetal calcium acquisition despite a relatively small placental size. We also found a normalization of fetal calcium accretion by E18.5, following a reduction at E16.5, which may be indicative of a fetus signaling to its placenta, by as yet unknown mechanisms, to increase maternofetal transfer of calcium. The gestational timing of this adaptation in WT mice was similar to that which we previously observed in the P0 knockout mouse ([@B19]), and points to a role for fetal nutrient demand in driving this adaptation via altered expression of placental calcium binding proteins. These data showed that placental adaptations are an important feature of both normal and compromised fetal growth and help to ensure appropriate calcium acquisition relative to the size of the fetus. Nothing is yet known regarding the underlying mechanisms that affect adaptation of placental calcium transport and in particular the fetal and/or placental signals that may be important in this process. Therefore, in this study we investigated mechanisms underlying the adaptive up-regulation of maternofetal calcium transfer. Initially, parathyroid hormone-related protein (PTHrP) was investigated as a candidate fetal signal. PTHrP is produced in a number of tissues including, but not limited to, the placenta, fetal membranes and fetal brain, liver, bone and parathyroid glands ([@B23]). In these tissues, there are multiple secretory mature peptides which have a range of different biological functions that can be elicited through endocrine, paracrine, autocrine and intracrine mechanisms \[reviewed by ([@B24])\]. In women, the concentration of PTHrP in maternal serum, and PTHrP expression in amnion and choriodecidua, are both increased in late gestation in parallel with the rapid increase in fetal growth and calcium accretion ([@B25]--[@B27]). Previous studies in rodents have demonstrated the importance of PTHrP and its receptor, the PTH/PTHrP receptor, in fetal development. Deletion of these genes in mice results in neonatal death, due to skeletal dysplasia ([@B28]) or death *in utero* mid gestation due to growth restriction ([@B29]). In a spontaneously hypertensive rat model, inappropriate levels of PTHrP in the placenta, fetal plasma and amniotic fluid were associated with compromised fetal growth ([@B30], [@B31]). Enhancing endogenous levels of PTHrP, by the addition of a PTH/PTHrP receptor antagonist, improves fetal growth in this rat model ([@B31]). In PTHrP knockout mice, maternofetal calcium transfer and fetal calcium accretion are increased despite fetal hypocalcaemia and lack of a maternal fetal calcium gradient ([@B32]--[@B34]). In human pregnancies complicated by FGR, PTHrP expression in fetal membranes and placenta is increased in cases of preterm FGR ([@B35]), and concentrations are elevated in cord blood ([@B18]). PTHrP also stimulates PMCA activity in BM vesicles isolated from human placenta ([@B36]). Thus, we hypothesized that PTHrP is a candidate signal stimulating an increase in calcium transfer and would be elevated in: (1) placental tissue and tissues from fetuses of the lightest vs. heaviest (LvH) placentas in WT mice; and (2) in placentas of P0 fetuses compared to their WT littermates (WTL). Using calcium pathway-focused PCR arrays we also tested the hypothesis that placental mechanisms underpinning the adaptive increase in calcium transfer in WT mice and in P0 mice would be similar. Materials and methods {#s2} ===================== Animals ------- Experiments were performed in accordance with the UK Animals (Scientific Procedures) Act of 1986 under the authority of a UK Home office project license (PPLs 40/3385 and P9755892D) and were authorized by the Animal Welfare and Ethical Review Board of the University of Manchester. The methods stated in this study adhere to the ARRIVE guidelines ([@B37]) and comply with the animal ethical principles under which the journal operates. Wild-type C57Bl/6J (Envigo, UK) females (10--16 weeks old) and males (12--26 weeks old) were mated and discovery of a copulation plug was used to define embryonic day (E)0.5 (term = E19.5). Mice were provided with nesting material and communally housed (with the exception of stud males that were individually housed) in individually ventilated cages under a constant 12 h light/dark cycle at 21--23°C with free access to food (Beekay Rat and Mouse diet, Bantin and Kingman) and water (Hydropac, Denver, US). Pregnant female mice were euthanized (cervical dislocation appropriate under ASPA schedule 1) and a laparotomy and hysterotomy performed. All fetuses were rapidly killed by cervical dislocation. On E18.5 (*N* = 20 litters), pregnant WT females were euthanized and fetuses and placentas were rapidly harvested, blotted and wet weights measured. The lightest (*n* = 20) and heaviest (*n* = 20) placentas were identified in each litter. All placentas and fetuses were snap frozen and stored at −80°C. In 9/20 litters, brains (*n* = 18; 9 from the lightest placental group, 9 from the heaviest) and livers (*n* = 18; 9 from lightest, 9 from heaviest) from fetuses corresponding to the lightest and heaviest placentas were immediately dissected, snap frozen and stored at −80°C. Fetal weight histograms were constructed and a non-linear regression performed (Gaussian distribution) from which individualized fetal weight centiles were calculated as described previously ([@B38]). Placental specific insulin-like growth factor 2 (*Igf2*) (P0) knockout mice (*N* = 10 litters), which had deletion of the U2 exon of the *Igf2* gene, were generated as previously described ([@B39]) and were a kind gift from Dr Miguel Constância and Professor Wolf Reik. C57BL/6J female mice (8--14 weeks old) and males heterozygous for the P0 deletion (10--32 weeks old) were mated and produced mixed litters of WTL fetuses and growth restricted fetuses \[P0; reported birthweight 78% compared to WTL at E19 equivalent to E18.5 in the current study ([@B40])\]. Embryonic day was defined as above. At E18.5, placentas and fetuses (40 WTL; 38 P0 from 10 litters) were weighed, snap frozen and stored at −80°C. Fetal tail tips were collected from all fetuses and stored at −20°C for genotype determination. The aim of the study, comparing lightest vs. heaviest placentas or those from WTL vs. P0 mice within a single litter, meant that randomization or blinding of the samples was not possible. Genotyping of P0 knockout mice ------------------------------ Genotype (WTL or P0) was determined for all fetuses from P0 mice according to a previously published genotyping protocol ([@B41], [@B19]). In brief, genomic DNA was extracted from fetal tail tips using a DNeasy kit (Qiagen, Manchester, UK). *Igf2* P0^+/−^ mutants were identified with a specific primer pair to amplify a 740 bp fragment across the 5 kb deletion (P0 dF 5′-TCCTGTACCTCCTAACTACCAC−3′ and P0 dR 5′-GAGCCAGAAGCAAACT−3′) and a primer to amplify a 495 bp fragment from the WT allele (5′-CAATCTGCTCCTGCCTG−3′). PCR conditions were as follows: 4 min denaturation at 94°C; 35 cycles of 1 min at 94°C, 1 min at 56°C, 1 min at 72°C; and 10 min final extension at 72°C. Samples were loaded with bromophenol blue and run on a 1.5% agarose gel. Bands were visualized using an InGenius transilluminator (Sygene Bio, Cambridge, UK). Protein expression ------------------ The lightest and heaviest placentas from WT mice (*N* = 7 litters) and placentas of P0 and WTL (*N* = 8 litters, 1 paired P0 and WTL placenta per litter selected at random) were homogenized and processed as described previously ([@B19]). Briefly, whole homogenates were separated, by means of centrifugation, into cytosolic fractions. Due to the small amount of starting tissue, whole homogenates of fetal tissues (brains and livers; *N* = 9 litters) were used for protein expression studies. SDS-PAGE was performed followed by electrotransfer to Immobilon-FL PVDF membranes (Millipore UK Ltd., Watford, UK). Primary antibodies included: rabbit polyclonal antibodies for serum/glucocorticoid-regulated kinase 1 (SGK1; 1 μg/ml; ab43606; Abcam, Cambridge, UK) and calmodulin (CaM; 2 μg/ml; sc-5537; Santa Cruz Biotechnology c/o Insight Biotechnology Ltd, Wembley, UK); rabbit monoclonal antibody for tartrate-resistant acid phosphatase (TRAP; 0.9 μg/ml; ab191406; Abcam); and goat polyclonal antibody for PTHrP (1 μg/ml; N-19, sc-9680; Santa Cruz Biotechnology). β-actin (0.5 μg/ml; ab8227; Abcam) or β-tubulin (0.9 μg/ml; ab6046; Abcam) was used as a loading control; when used no difference was observed in β-actin or β-tubulin expression between groups. Negative controls were by omission of primary antibody. Immunoreactive species were detected with fluorescent-conjugated secondary antibodies (Li-COR Biosciences, Cambridge, UK) and membranes imaged using an Odyssey Sa Infrared Imaging System (Li-COR). Signal density was measured using Image Studio Lite (Li-COR). All signals were in the linear range of detection. Protein expression was compared separately between the lightest and heaviest placentas, fetal tissues from lightest and heaviest placentas, and WT and P0 samples. RT^2^ profiler PCR arrays ------------------------- RNA was extracted from whole placentas (*N* = 7 litters: *n* = 7 lightest, *n* = 7 heaviest; *N* = 6--7 litters: *n* = 6--7 WTL, *n* = 6--7 P0, 1 paired P0 and WTL placenta per litter selected at random) using an RNeasy Mini Kit (74104; Qiagen, Manchester, UK), RNase-Free DNase set (79254; Qiagen) and measured by a Thermo Scientific NanoDrop 2000C spectrophotometer (A~260~/A~280~ range 2.06--2.12). Any contaminating genomic DNA was removed and cDNA was synthesized from 0.5 μg RNA per sample using the RT^2^ first strand kit (330401; Qiagen; genomic DNA elimination mix for 5 min at 42°C, on ice for 1 min; reverse transcription mix for 42°C for 15 min followed by 5 min at 95°C). Expression of 168 related genes, 5 reference genes and quality controls was measured in each placenta using RT^2^ Profiler PCR arrays (PAMM-066Z mouse cAMP/calcium signaling pathway finder and PAMM-170Z mouse osteoporosis array; 96-well format; Qiagen) with RT2 SYBR® Green ROX™ qPCR mastermix (330523; Qiagen) on a Stratagene MX3005P®, according to the manufacturer\'s instructions (10 min at 95°C, 40 cycles of 15 s at 95°C followed by 1 min at 60°C; dissociation curve 1 min at 95°C, 30 s at 55°C, 30 s at 95°C). Data were analyzed using SA Bioscience PCR Array Data Analysis 3.5 Web Portal (<http://dataanalysis.sabiosciences.com/pcr/arrayanalysis.php>). Data and statistical analysis ----------------------------- Data are presented as the lightest placenta as a percentage of the heaviest in a litter (dotted line = 100%), placentas of P0 fetuses as a percentage of WTL (dotted line = 100%), or median and \[range\] where the experimental *N* = number of litters, *n* = number of placentas or fetuses. The solid line on graphs represents the median value. For P0 and WTL fetal and placental weights, average fetal weights for each genotype per litter were calculated and data are shown as a mean of these average weights. Data were analyzed by Wilcoxon matched-pairs signed-rank test or Mann Whitney test. *P* \< 0.05 was considered statistically significant. PCR array data were analyzed using SA Bioscience PCR Array Data Analysis 3.5 Web Portal (<http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php>). Fold change \[2∧(- Delta Delta Ct)\] is the normalized gene expression \[2∧(- Delta Ct)\] in the test sample (lightest placenta or placenta of P0 fetus) divided by the normalized gene expression \[2∧(- Delta Ct)\] in the control sample (heaviest placenta or placenta of WTL fetus). Fold-change values \>1 indicate an up-regulation in expression, and fold-change values \< 1 demonstrate a down-regulation (Supplementary Tables [1](#SM1){ref-type="supplementary-material"}, [2](#SM2){ref-type="supplementary-material"}). Fold regulation values are shown in **Tables 2, 3**. Results {#s3} ======= Fetal and placental weights --------------------------- As expected, lightest placentas from WT mice demonstrated significantly reduced placental weight vs. heaviest placentas at E18.5 (*P* \< 0.0001, Table [1](#T1){ref-type="table"}). Fetal weight was significantly reduced (*P* \< 0.01) and fetal weight: placental weight (F:P) ratio increased (*P* \< 0.0001), in fetuses from lightest vs. heaviest WT placentas (Table [1](#T1){ref-type="table"}). Mean fetal weight centiles were lower in the lightest compared to the heaviest placenta group (39th vs. 56th centile; *P* \< 0.05) but were not considered growth restricted (normal range 10th−90th centile). Consistent with previous studies ([@B19], [@B40], [@B41]), placentas from P0 fetuses compared to WTL fetuses were lighter, fetal weight was lower and F:P ratio higher at E18.5 (all *P* \< 0.0001, Table [1](#T1){ref-type="table"}). ###### Placental weight, fetal weight and fetal weight:placental weight (F:P) ratio in the lightest and heaviest placental groups of wild-type (WT) mice and in P0 and wild-type littermates (WTL) at embryonic day (E) 18.5. ----------------------------------------------------------------------------------------------- **Lightest** **Heaviest** **Lightest/**\ ***P*-value** **Heaviest (%)** ---------------------- ---------------- ---------------- ------------------ ------------------- Placental weight (g) 0.068\ 0.087\ 78.0\ \<0.0001 (0.052--0.080) (0.076--0.100) (64.0--89.0) Fetal weight (g) 1.151\ 1.191\ 94.5\ 0.01 (0.924--1.289) (1.018--1.441) (79.0--108.0) F:P ratio 16.8\ 14.2\ 119.5\ \<0.0001 (12.8--24.0) (11.4--16.8) (92.0--166.0) **P0** **WTL** **P0/WTL (%)** ***P*****-value** Placental weight (g) 0.065\ 0.096\ 64.0\ \<0.0001 (0.048--0.100) (0.050--0.113) (58.3--83.1) Fetal weight (g) 0.977\ 1.179\ 80.9\ \<0.0001 (0.735--1.218) (0.854--1.386) (68.1--85.6) F:P ratio 14.5\ 12.6\ 110.6\ \<0.0001 (10.91--18.4) (10.0--17.1) (100.5--146.4) ----------------------------------------------------------------------------------------------- *Data are median (range) or presented as the lightest placenta as a percentage of the heaviest in a litter (lightest/heaviest (%) column) in WT mice or as the litter average of P0 fetuses as a percentage of the litter average of their WT littermates (P0/WTL (%) column). Data are analyzed by Wilcoxon matched-pairs signed rank test (lightest vs. heaviest; P0 vs. WTL). P \< 0.05 was considered statistically significant*. PTHrP protein expression ------------------------ There were no significant differences in PTHrP protein expression between the lightest and heaviest placentas of WT mice, or within brains and livers of those fetuses from the lightest and heaviest placentas (Figures [1A--C,E](#F1){ref-type="fig"}). There was a trend toward increased PTHrP protein expression in placental tissue of P0 vs. WTL fetuses (139%; *P* = 0.06; Figures [1D--E](#F1){ref-type="fig"}). The PTHrP antibody used for these studies was discontinued during the timecourse of the project and so we were unable to assess PTHrP expression in fetal brains and livers of fetuses in P0 vs. WTL mice. Attempts to use different antibodies targeted to PTHrP failed to show reproducible amplification of signal. {#F1} RT^2^ profiler PCR arrays ------------------------- PCR arrays demonstrated significant changes in the expression of four genes (Table [2](#T2){ref-type="table"}); an increase in calmodulin 1 (*Calm1*; *P* \< 0.05) and alkaline phosphatase (*Alpl*; *P* \< 0.05) expression, and a decreased expression of tartrate resistant acid phosphatase 5 (*Acp5*; *P* \< 0.05) and heat shock protein 5 (*Hspa5*; *P* \< 0.05) in the lightest vs. heaviest WT placentas (*N* = 6). Placental protein expression of calmodulin (CaM; 107%; Figure [2A](#F2){ref-type="fig"}) and tartrate resistant acid phosphatase 5 (TRAP; 89%; Figure [2B](#F2){ref-type="fig"}) measured by Western blot was not different between lightest and heaviest placenta groups. ###### Results of the cAMP/Ca^2+^ signaling pathway finder and osteoporosis RT^2^ profiler PCR arrays in the lightest compared to the heaviest placentas from WT mice. --------------------------------------------------------------------------------------------- **Gene** **Product** **Fold regulation** ***P*-value** ---------- ---------------------- --------------------- ------------------------------------- *Calm1* Calmodulin 1 +1.6 0.04[\*](#TN1){ref-type="table-fn"} *Alpl* Alkaline Phosphatase +1.2 0.04[\*](#TN1){ref-type="table-fn"} *Acp5* Acid Phosphatase 5\ −1.6 0.04[\*](#TN1){ref-type="table-fn"} Tartrate Resistant *Hspa5* Heat Shock 70 kDa\ −1.8 0.01[\*](#TN1){ref-type="table-fn"} Protein 5 --------------------------------------------------------------------------------------------- *Fold regulation is the normalized gene expression in the lightest placentas divided by the normalized gene expression in the heaviest placentas. Fold regulation values +1 indicate an up-regulation in expression, whilst fold regulation values −1 demonstrate a down-regulation*. *P \< 0.05*. {#F2} Twenty-four genes were differentially expressed in placentas from P0 vs. WTL fetuses in the same litter as shown in Table [3](#T3){ref-type="table"}. Of note, significantly reduced expression was observed in the genes encoding calbindin-D~9K~, S100 calcium binding protein G (*S100g*; −1.9-fold, *P* \< 0.01), PTHrP (*Pthlh*; −2-fold, *P* \< 0.05) and parathyroid hormone 1 receptor (*Pth1r*; −1.7-fold, *P* \< 0.01). Expression of serum/glucocorticoid-regulated kinase 1 (*Sgk1*; 1.4-fold, *P* \< 0.05), a kinase involved in the regulation of a range of membrane transporters, ion channels and transcription factors as well as cell survival ([@B42]--[@B46]) was increased. There was no difference in serum/glucocorticoid-regulated kinase 1 (SGK1; 88%; Figure [3](#F3){ref-type="fig"}) protein expression between placentas of P0 and WT mice. ###### Results of the cAMP/Ca^2+^ signaling pathway finder and osteoporosis RT^2^ profiler PCR arrays between placentas from P0 compared to wild-type littermate (WTL) fetuses. **Gene** **Product** **Fold regulation** ***P*-value** ------------- --------------------------------------------------------- --------------------- ---------------------------------------- *Ar* Androgen Receptor +1.7 0.04[\*](#TN2){ref-type="table-fn"} *Esr1* Estrogen Receptor +1.5 0.02[\*](#TN2){ref-type="table-fn"} *Sgk1* Serum/Glucocorticoid-Regulated Kinase 1 +1.4 0.03[\*](#TN2){ref-type="table-fn"} *Eno2* Enolase 2, γ Neuronal +1.4 0.03[\*](#TN2){ref-type="table-fn"} *Tnfrsf11b* Tumor Necrosis Factor Receptor Superfamily, 11b +1.4 0.001[\*\*](#TN2){ref-type="table-fn"} *Tgfb3* Transforming Growth Factor, β3 −1.2 0.04[\*](#TN2){ref-type="table-fn"} *Rb1* Retinoblastoma 1 −1.3 0.02[\*](#TN2){ref-type="table-fn"} *Pck2* Phosphoenolpyruvate Carboxykinase 2 (Mitochondrial) −1.3 0.04[\*](#TN2){ref-type="table-fn"} *Junb* Jun-B Oncogene −1.3 0.04[\*](#TN2){ref-type="table-fn"} *Fosb* FBJ Osteosarcoma Oncogene B −1.4 0.02[\*](#TN2){ref-type="table-fn"} *Crem* cAMP Responsive Element Modulator −1.4 0.03[\*](#TN2){ref-type="table-fn"} *Per1* Period Homolog 1 (Drosophila) −1.4 0.04[\*](#TN2){ref-type="table-fn"} *Nos2* Nitric Oxide Synthase 2, Inducible −1.4 0.006[\*\*](#TN2){ref-type="table-fn"} *Brca1* Breast Cancer 1 −1.5 0.01[\*](#TN2){ref-type="table-fn"} *Acp5* Acid Phosphatase 5, Tartrate Resistant −1.7 0.02[\*](#TN2){ref-type="table-fn"} *Cyp17a1* Cytochrome P450 Family 17, a1 −1.7 0.03[\*](#TN2){ref-type="table-fn"} *Ltbp2* Latent Transforming Growth Factor β Binding Protein 2 −1.8 0.03[\*](#TN2){ref-type="table-fn"} *Gem* GTP Binding Protein (over-expressed in skeletal muscle) −1.8 0.04[\*](#TN2){ref-type="table-fn"} *S100g* S100 Calcium Binding Protein G −1.9 0.004[\*\*](#TN2){ref-type="table-fn"} *Dkk1* Dickkopf Homolog 1 −1.9 0.005[\*\*](#TN2){ref-type="table-fn"} *Wnt10b* Wingless Related MMTV Integration Site 10b −1.9 0.04[\*](#TN2){ref-type="table-fn"} *Car2* Carbonic Anhydrase 2 −2.0 0.009[\*\*](#TN2){ref-type="table-fn"} *Pthrp* Parathyroid Hormone-Related Protein −2.0 0.02[\*](#TN2){ref-type="table-fn"} *Ncam1* Neural Cell Adhesion Molecule 1 −2.0 0.002[\*\*](#TN2){ref-type="table-fn"} *Fold regulation is the normalized gene expression in placentas of P0 fetuses divided by the normalized gene expression in placentas of WTL fetuses. Fold regulation values +1 indicate an up-regulation in expression, whilst fold regulation values −1 demonstrate a down-regulation*. *P \< 0.05*; *P \< 0.01*. {#F3} There was only one similar change in gene expression between the two study groups; tartrate resistant acid phosphatase 5 (*Acp5*, −1.6-and −1.7-fold, respectively) was reduced to the same extent in the lightest placentas of WT mice and placentas of P0 fetuses (Tables [2](#T2){ref-type="table"}, [3](#T3){ref-type="table"}). Whilst TRAP protein expression was no different between lightest and heaviest WT placentas, gene expression of TRAP was increased in placentas of P0 compared to WT littermates (TRAP; 144%; Figure [2C](#F2){ref-type="fig"}). Discussion {#s4} ========== We have previously observed similar adaptive increases in maternofetal calcium transport in small placentas from WT mice, and in small pathological placentas of the P0 knockout mouse model of FGR, with accompanying changes in calbindin-D~9K~ expression ([@B19], [@B22]). Here we demonstrate that the underlying mechanisms of these adaptations in the two models appear to be distinct. Contrary to our hypothesis, expression of PTHrP was not different between lightest compared with heaviest (LvH) WT placentas, but there was a trend toward increased expression in P0 vs. WTL placentas; PTHrP has been shown to influence placental calcium transport in most ([@B32], [@B33], [@B36]), but not all studies ([@B47]). As such, any change in PTHrP expression in placentas near term may be important in the increased maternofetal calcium transfer observed in P0 vs. WTL. As shown by our previous studies, calbindin-D~9K~ is implicated as a mediator of placental adaptation in calcium transfer both in WT mice ([@B22]) and in the P0 mouse ([@B19]). However, the lack of change in placental calcium transfer in the calbindin-D~9K~ knockout mouse indicates that other candidates, including calbindin-D~28K~, TRPV5/6 and the sodium-calcium exchanger, might also be involved ([@B48], [@B49]). Thus, we adopted a holistic approach to compare the expression of genes related to calcium transfer and signaling in WT mice and P0 mice. We speculated that these experiments would identify similar calcium-specific pathways altered in the lightest and/or P0 placentas and provide insight into potential regulators of the observed placental adaptation. The number of genes showing altered expression was limited in lightest vs. heaviest WT placentas but mRNA expression of calmodulin-1 was increased in the lightest placentas at E18.5. Whilst this may act as a further indicator of the importance of calcium binding proteins in the previously observed adaptation, this altered expression was not mirrored at the protein level; the importance of the change in gene expression therefore requires further elucidation. In contrast to the lightest and heaviest placentas in WT mice, multiple genes were differentially expressed in placentas of P0 vs. WT littermates. This is perhaps unsurprising given that P0 mice represent a model of fetal growth restriction whereas lightest vs. heaviest placentas represent extremes of placental weight in a "normal" WT population. Despite the increased maternofetal calcium transport at E18.5, expression of placental calcium-related genes was generally reduced in P0 vs. WTL. Significant reductions were observed in the expression of *S100g* (encoding calbindin-D~9K~), *Pth1r* and *Pthrp*, whilst *Sgk1*, a regulator of epithelial ion transport and cell survival, was up-regulated. SGK1 is a downstream effector of the PI3K/AKT signaling pathway, and in support of the observations here, this pathway is dysregulated in the placentas of P0 knockout mice in late gestation ([@B50]). The trend for reduced gene expression near to term could be the result of timing in gestation, i.e., gene expression increased earlier in gestation to promote changes in placental nutrient transport might be downregulated nearer to term having already resulted in increased transcription of the target protein, as previously observed for calbindin-D~9K~ in lightest vs. heaviest WT placentas ([@B22]). The choice of E18.5 for these studies reflected the timepoint at which the adaptation (increased placental calcium transport) was previously observed ([@B19], [@B22]) but analyses earlier in gestation would offer further insight into the timing of these changes. With regards to the trend for reduced gene expression near term in P0 vs. WTL, altered gene expression may not be the driving force in these placentas; instead post-translational processing of binding proteins and/or receptors (e.g., TRPV6) may underlie the adaptive changes in calcium transfer. Mechanisms will need to be explored further in future experiments. Whilst the discrepancy in *S100g* gene and calbindin-D~9K~ protein expression at E18.5 will need elucidating, calbindin-D~9K~ does appear to be important in the previously reported changes in placental calcium transport in mouse models of FGR ([@B19], [@B22]). Increased *Sgk1* expression in P0 vs. WT warrants further investigation to assess whether SGK1, and its activated phosphorylated isoform, play an important role in these adaptive responses by the placenta. SGK1 influences intracellular calcium by up-regulating store operated calcium entry (SOCE), increasing calcium release-activated calcium channel (CRAC) current, and increasing the activity of TRPV5 and 6 ([@B46]). Expression and activation of SGK1 is enhanced by higher levels of cytosolic calcium thus SGK1 has been suggested as an amplifier of calcium entry; influx of extracellular calcium through SOCE combined with activation of calcium/calmodulin protein kinase signaling up-regulates levels and activity of SGK1 ([@B42]--[@B44]). Activation of SGK1 also occurs through other mechanisms, including through the phosphatidylinositol-3-kinase pathway that when stimulated by growth factors activates the mechanistic target of rapamycin complex 2 triggering the phosphorylation of 3-phosphoinositide-dependent kinase PDK1 and subsequent phosphorylation of SGK1 ([@B45], [@B51]). Identifying extracellular regulators stimulating SGK1 intracellular activity may provide potential candidate signals initiating placental adaptations. Placental *Acp5* gene expression was lower in the lightest compared to the heaviest placentas and in P0 fetuses compared to their WTLs. In contrast, TRAP5 protein expression (encoded by *Acp5*) was not different between the lightest and heaviest placentas, and significantly higher in placentas of P0 compared to WTLs. Previous studies in animal models indicate *Acp5* has an essential role in modeling, remodeling and mineralization of developing bone and cartilage ([@B52]), as well as participating in iron transfer from mother to fetus ([@B53]). Hansson et al. ([@B54]) demonstrated increased placental *Acp5* gene expression in pre-eclampsia compared to normal pregnancy and suggested that this increase might be a compensatory mechanism for poor placentation to prevent fetal malnutrition ([@B54]). The elevated expression of TRAP5 protein in placentas of P0 vs. WTL fetuses, which may be as a result of increased gene expression earlier in gestation leading to increased protein translation, supports a regulatory role for TRAP5 in this mouse model of FGR; the added complexity of the opposing direction of change in gene and protein expression, suggest post-translational modification is very important. For all of the studies described herein, there was an unequal distribution of fetal sex when considering the placental samples analyzed. For the studies comparing lightest vs. heaviest placentas in WT mice, there was a bias toward females having the lightest placentas and males having the heaviest placentas. Whilst we have previously reported that adaptive changes in maternofetal calcium transport do not appear to be influenced by fetal sex ([@B22]), future studies investigating the mechanisms underpinning these adaptations should also consider sex-dependent effects. Likewise, fetal sex should be taken into account when assessing mechanisms underpinning placental adaptations in P0 vs. WTL mice. In summary this study has shown differences in the mechanisms underlying adaptations in placental calcium transport in normal pregnancy vs. that affected by growth restriction. Our data suggest that calcium binding proteins in normal mouse pregnancy, and PTHrP and *Acp5*/TRAP in FGR (P0) pregnancy, are candidate adaptation regulatory proteins worthy of further investigation. Author contributions {#s5} ==================== KM performed research. SG and CS contributed to the conception and design of the work. CH and MD contributed to the conception and design of the work and performed research. All authors were involved in drafting the paper. 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 would like to thank the staff of the Biological Services Facility at the University of Manchester for their assistance with this project. Infrastructure support was provided by the National Institute for Health Research Manchester Biomedical Research Centre and Tommy\'s the Baby Charity. **Funding.** This study was supported by a Medical Research Council Career Development Fellowship award (Grant number MR/K024442/1), and awarded to MD. Supplementary material {#s7} ====================== The Supplementary Material for this article can be found online at: <https://www.frontiersin.org/articles/10.3389/fendo.2018.00671/full#supplementary-material> ###### Click here for additional data file. ###### Click here for additional data file. [^1]: Edited by: Elke Winterhager, Universität Duisburg-Essen, Germany [^2]: Reviewed by: Amanda Sferruzzi-perri, University of Cambridge, United Kingdom; Dana Manuela Savulescu, Wits Health Consortium (WHC), South Africa [^3]: This article was submitted to Reproduction, a section of the journal Frontiers in Endocrinology