Ethanol Impairs Intestinal Barrier Function in Humans through Mitogen Activated Protein Kinase Signaling: A Combined In Vivo and In Vitro Approach

Background Ethanol-induced gut barrier disruption is associated with several gastrointestinal and liver disorders. Aim Since human data on effects of moderate ethanol consumption on intestinal barrier integrity and involved mechanisms are limited, the objectives of this study were to investigate effects of a single moderate ethanol dose on small and large intestinal permeability and to explore the role of mitogen activated protein kinase (MAPK) pathway as a primary signaling mechanism. Methods Intestinal permeability was assessed in 12 healthy volunteers after intraduodenal administration of either placebo or 20 g ethanol in a randomised cross-over trial. Localization of the tight junction (TJ) and gene expression, phosphorylation of the MAPK isoforms p38, ERK and JNK as indicative of activation were analyzed in duodenal biopsies. The role of MAPK was further examined in vitro using Caco-2 monolayers. Results Ethanol increased small and large intestinal permeability, paralleled by redistribution of ZO-1 and occludin, down-regulation of ZO-1 and up-regulation of myosin light chain kinase (MLCK) mRNA expression, and increased MAPK isoforms phosphorylation. In Caco-2 monolayers, ethanol increased permeability, induced redistribution of the junctional proteins and F-actin, and MAPK and MLCK activation, as indicated by phosphorylation of MAPK isoforms and myosin light chain (MLC), respectively, which could be reversed by pretreatment with either MAPK inhibitors or the anti-oxidant L-cysteine. Conclusions Administration of moderate ethanol dosage can increase both small and colon permeability. Furthermore, the data indicate a pivotal role for MAPK and its crosstalk with MLCK in ethanol-induced intestinal barrier disruption. Trial Registration ClinicalTrials.gov NCT00928733


Introduction
Ethanol (ethyl alcohol) consumption is associated with several gastrointestinal (GI) and liver disorders, especially alcoholic liver disease (ALD) [1]. ALD is a progressive disease initiated by steatosis and inflammation, followed by liver fibrosis and cirrhosis [2]. Only 30% of chronic alcoholics eventually develop cirrhosis, indicating that additional factors are required [3]. Recent evidence points to a role for the gut-liver axis in the pathogenesis of ALD [4]. Ethanol is known to disrupt GI epithelial barrier integrity [5], resulting in translocation of potentially harmful bacteria and their products such as endotoxins [6] and peptidoglycans [7] into the portal circulation and consequently, liver injury. Dysfunction of GI mucosal barrier may result in increased susceptibility not only to infections [8] but also to development of ethanol-related GI cancers [9].
Human studies investigating effects of ethanol on intestinal barrier function have focused mainly on chronic heavy drinkers (. 80 g/day) [10], demonstrating that ethanol increases small intestinal permeability [11]. Effects of moderate ethanol consump-tion on GI barrier function are less well known but are especially relevant since daily moderate alcohol consumption (daily intake of 1-2 alcoholic drinks or 12-24 g ethanol), is widespread [12]. Ethanol and its metabolites can reach the distal GI tract through the vascular space [13]. Therefore, ethanol and its metabolites may be injurious not only to the small but also to the large intestinal barrier. Intestinal barrier integrity is maintained by the tight junctions (TJs), a complex meshwork of transmembrane and cytoplasmic proteins including occludin, the claudins and the zona occludens family (ZO-1,2,3), linked to the cytoskeletal protein Factin [14]. The TJs are supported by the adherens junctions proteins (AJs) E-cadherin and b-catenin, which are also required for TJ assembly [15]. Data obtained in vitro using intestinal epithelial cells (Caco-2 cell monolayers) indicate that ethanol disrupts TJs integrity via mechanisms involving oxidative stress [11], modulation of microtubules [16] and activation of the cell signaling pathway myosin light chain kinase (MLCK) [17]. The three isoforms of mitogen activated protein kinase (MAPK) isoforms including P38, extracellular signal-regulated kinase (ERK) and stress-activated protein kinase/C-Jun N-terminal kinase (SAPK/JNK) have been shown to modulate TJs integrity [18]. Furthermore, activation of MAPK has been reported to mediate intestinal epithelial barrier dysfunction [19]. Data on mechanims of ethanol-induced barrier dysfunction are mainly derived from in vitro studies [20]. In addition, insight into effects of moderate doses of ethanol on human intestinal barrier function and potential mechanisms involved is still lacking.
Our aims were to determine effects of a single moderate dose of ethanol, administered intraduodenally, on 1) small intestinal and colon permeability and 2) localization and expression of TJ in duodenal biopsies. In addition, the role of MAPK pathway as a primary signaling mechanism involved in ethanol-induced barrier disruption was investigated both in vivo in duodenal biopsies and in vitro in Caco-2 cell monolayers.

Subjects and study design
The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1 and Protocol S1. The study was performed according to a randomized placebo-controlled crossover design, in 12 healthy subjects recruited between October 2010 and July 2011. Participants were healthy men from Caucasian ethnicity, between 18 and 45 years of age to avoid genome-related differences in ethanol metabolism [21]. Exclusion criteria included BMI.35 kg/m 2 , history of GI disorders, liver diseases i.e. hepatitis and cirrhosis, abdominal surgery, excessive alcohol consumption (.20 g/day) and smoking. The study flow diagram is shown in Figure S1. The study protocol was approved by the Ethics Committee of Maastricht University Medical Center (MUMC), conducted in accordance with the Declaration of Helsinki and registered at the Clinical Trial Registry (www.clinicaltrials.gov; NCT00928733). All participants provided written informed consent. An overview of the experimental procedures is given in Figure S1. The study consisted of two test days in a random order, with a washout period of one week. Participants arrived at the MUMC at 08:00 AM after an overnight fasting. An intravenous cannula was placed in an antecubital vein for blood sampling. Then, the tube was placed intraduodenally and optimal positioning was confirmed by X-ray.
Moderate ethanol intake is difficult to define. However, despite the complexity, Department of Agriculture and Department of Health and Human Services define moderate consumption of one standard drink a day for women and two drinks men [22]. A standard drink is generally considered to be 360 ml of beer, 150 ml of wine, or 45 ml of 80-proof distilled spirits. These drinks contain roughly the same amount of absolute alcohol-approximately 12-24 g [12]. Accordingly and via the intestinal tube, either 20 g ethanol in 100 ml water (5% v/v) or water (placebo) was perfused at a rate of 7 ml/min for 15 min. Earlier observations have indicated that ethanol can reach concentrations of 2-10% in small intestinal lumen after consumption of 50 ml ethanol in 20% solution [23]. Ethanol was introduced directly into the duodenum to overcome interindividual variations in its upper GI metabolism after oral consumption [24]. Thereafter, participants received an oral multi-sugar drink to assess intestinal permeability, consisting of 1 g lactulose (Centrafarm, Etten-Leur, the Netherlands), 1 g sucralose (Brenntag, Sittard, the Netherlands), 1 g erythritol (Danisco, Copenhagen, Denmark), 1 g sucrose (Van Gilse, Dinteloord, the Netherlands), and 0.5 g Lrhamnose (Danisco) dissolved in 150 ml tap water. At t = 30 min after start of placebo or ethanol perfusion, participants underwent a gastroduodenoscopy to obtain mucosal biopsies from the second part of the duodenum (forceps diameter: 2.8 mm). Tissue samples were frozen immediately in liquid nitrogen and stored at 280uC until further analyses. Blood was collected before and 15, 30, 45, 60, 90, 120, 150 and 180 min after onset of ethanol perfusion. Urine was collected in plastic containers before, and in three fractions after ethanol perfusion, i.e. 0-2 h, 2-5 h and 5-24 h. All samples were stored at 280uC within 4 hours after collection. For safety, participants were not allowed to leave the hospital unless their breath ethanol concentration was ,0.05%, the legal limit in the Netherlands.

Measurements
Liver function, blood ethanol and fatty acid ethyl esters (FAEEs) levels. Plasma alanine aminotransferase (ALT) and cglutamyltransferas (cGT) were determined before and 60 min after ethanol perfusion. Blood ethanol concentrations were determined by headspace gas chromatography aided with Flame Ionization Detection System with a detection limit of 10 mg/dL. Plasma concentrations were converted to blood alcohol concentration (BAC) using the conversion factor 0.809. Plasma levels of fatty acid ethyl esters (FAEEs) including ethyl oleate and ethyl palmitate were analyzed by gas chromatograph with a mass selective detector GC-MS QP-2010 PLUS equipped with GCMS solutions Software (Shimadzu, Maryland, U.S.A) [25].
Assessment of intestinal barrier function. Intestinal barrier function was assessed using a multi-sugar recovery test. Urinary sugar recovery was quantified by combined HPLC (Model PU-1980 pump, Jasco Benelux, Maarsen, the Netherlands) and mass spectrometry (Model LTQXL, Thermo Electron, Breda, the Netherlands) as described by van Wijck et al. [26]. Sucrose in 0-2 h urine, Lactulose/rhamnose (L/R) ratio in 0-5 h urine and sucralose/erythritol (S/E) ratios in the 5-24 h urine fraction were determined as indicators for gastroduodenal, small and large intestinal permeability, respectively.
Localization and gene expression of ZO-1 and occludin in duodenal mucosa. Immunofluorescent staining was performed as described previously [27], with minor modifications [27]. Tissue sections were examined under Leica TCS SPE confocal laser scanning microscope (Leica Micro systems GmbH, Mannheim, Germany), images were acquired at 5126512 pixel resolutions and z-stacks were obtained using Leica Application Suite Advanced Fluorescence software (Leica Microsystems GmbH). Images were processed and quantified as described by Fisher et al. [28], with minor modifications. Briefly, fluorescent staining of ZO-1 and occludin in the TJ region from uniform Z sections perpendicular to the apical cell surface of the epithelium were subjected to projection and passed through a Gaussian smoothing filter. Plot profiles of the staining intensity along the perpendicular lines were generated using Image J software [29]. Next, grey scale values of each image were calculated and normalized against a summed maximum intensity matrix (51265126256) [28].
Gene expression of occludin, claudin 3, and claudin 4, myosin light chain kinase (MLCK) and ZO-1 in duodenal mucosal biopsies was evaluated by qPCR as described previously [30]. Primer sequences are given in the Table S1.
Assessment of MAPK isoforms phosphorylation in duodenal biopsies. Frozen tissue samples were homogenized in ice-cold PBS containing protease and phosphatase inhibitor cocktail (10 ml/ml PBS; Sigma-Aldrich). Concentrations of soluble proteins were quantified in tissue supernatants using the bicinchoninic acid assay (BCA; Bio-Rad). Lysates of Hela cells treated with Anisomycin were used as positive controls for phosphorylated P38 and JNK, whereas A431 cells treated with recombinant human epidermal growth factor were used for ERK1/2. Phosphorylated forms of P38, ERK1/2 and JNK in tissue and cell lysates were analysed using semi-quantitative sandwich ELISA according to manufacturer's instructions (Ray Biotech), and data are expressed as optical density (OD) at 450 nm.
Following subjection to the aforementioned treatments, Caco-2 monolayers were fixed on the inserts for 10 min with 4% (w/v) paraformaldehyde and permeabilized with 0.1% (v/v) Triton X-100 in PBS at RT for 40 min. Next, monolayers were processed for immunofluorescence staining of the TJ proteins ZO-1 and occludin, AJ proteins E-cadherin and b-catenin, and F-actin as we described previously [16].
Power calculation and statistical analyses. Primary outcome was intestinal permeability. Based on study by Troost et al. [31] comparing small intestinal permeability before and after intake of indomethacin in15 healthy volunteers, lactulose/rhamnose (L/R) ratio increased from 0.01360.009 to 0.03160.020 after indomethacin treatment. In the above mentioned study, the standard deviation (SD) in the indomethacin-treated group was 0.02 and the mean difference was 0.018. In our study, each subject matched to one control and to detect comperable differences in small intestinal permeability using SD of 0.02, 12 healthy volunteers were needed to be able to reject the null hypothesis that the population means of the experimental and control groups are equal using the probability power b = 0.8 and the type I error probability a = 0.05. This sample size calculation was performed with PS-Power and Sample Size Program version 3.0. Statistical analyses were performed using GraphPad Prism software (version 5, Windows, San Diego, CA, USA). Data were tested for normality by the Kolmogorov-Smirnoff test. Normally distributed data were analyzed by independent Student's t test. Wilcoxon's and Mann-Whitney U tests were used for abnormally distributed data. In vitro data, each conducted in triplicate, are presented as means 6 SD of at least three experiments. Data were analysed using one-way analysis of variance (ANOVA) followed by Bonferroni multiple comparisons and P-values adjustment (Bonferroni correction), and P,0.05 was considered statistically significant.

Subjects
Twelve male healthy volunteers [age 30.863.1 years; BMI 23.060.07 kg/m 2 ] participated in this study. None of the participants experienced any GI discomfort during the test days. No side effects occurred and no changes in plasma ALT and cGT levels after ethanol administration were observed (data not shown).

Ethanol and FAEE
Blood samples collected prior to the intervention were negative for ethanol (i.e. ,10 mg/dl plasma) and FAEEs. Following intraduodenal ethanol, plasma ethanol concentration peaked at 15 min (62 mg/dl) and gradually declined to 11 mg/dl at 180 min ( Figure 1A). Mean plasma concentrations of the FAEEs ethyl oleate and ethyl palmitate were significantly increased 30 min after ethanol administration and decreased towards basal levels at 180 min (P = 0.0003 and P,0.0001, respectively; Figure 1B).

Localization and expression of TJ proteins in duodenal mucosa
After placebo administration, ZO-1 and occludin showed strong staining in duodenal biopsies at the apical part of epithelial cells, being continuous without disruption and showed colocalization along the villous epithelium. In contrast, after ethanol, ZO-1 and occludin showed less immune reactivity, disruption and complete absence of colocalization at the apical membrane compared with placebo ( Figure 2D). These features were confirmed by measurement of normalized image intensity of immunolabeled ZO-1 and occludin in duodenal biopsies. The normalized intensity of both ZO-1 and occludin decreased significantly after ethanol versus placebo (P = 0.0008 and P,0.0001, respectively; Figure 2E). In addition, ethanol significantly down-regulated ZO-1 and upregulated MLCK mRNA expression compared to placebo (P = 0.092 and P = 0.0492, respectively; Figure 2F). No differences were found in gene expression of occludin, claudin 3 and claudin 4 (P.0.05; Figure 2F).

Phosphorylation of MAPK isoforms in duodenal mucosal biopsies
The phosphorylation of MAPK isoforms in duodenal mucosal biopsies after ethanol and placebo administration was checked quantitatively using ELISA. Results showed that ethanol significantly increased the level of phosphorylated P38 (p-P38; P, 0.0001), ERK1/2 (p-ERK; P,0.0001) and JNK (p-JNK; P = 0.0001) compared to the placebo ( Figure 3A, B and C, respectively).

MAPK activation in ethanol-induced barrier disruption in Caco-2 monolayers
To evaluate the role of MAPK in effects of ethanol on intestinal barrier function, pharmacological inhibitors of its isoforms were used. Caco-2 monolayers exposed to ethanol showed a significant decrease in TEER (P,0.0001, P,0.0001, P,0.0001; Figure 5A, C and E, respectively) and increased FITC-D4 permeation (P,0.0001, P,0.0001, P,0.0001; Figure 5B, D, and F, respectively) compared to controls. In contrast, monolayers pretreated with the MAPK isoforms inhibitors SB2035809, PD98059 and SP600125 showed no drop but significant increase in TEER (P, 0.0001; P = 0.0002, P,0.0001; Figure 5A, C and E, respectively) and significant decrease in FITC-D4 permeation (P,0.0001, P, 0.0001, P,0.0001; Figure 5B, D, and F, respectively) in response to ethanol treatment. Compared to controls, treatment of Caco-2 monolayers with 40 mM ethanol induced delocalization and internalization of ZO-1 and occludin ( Figure 6A), E-cadherin and b-catenin from apical membranes to intracellular compartments ( Figure 6B). However, inhibition of MAPK isoforms with SB2035809, PD98059 and SP600125 prevented ethanol-induced changes in localization of the TJ (Figure 6A), and AJ ( Figure 6B). Ethanol-induced intestinal barrier dysfunction may involve MLCK activation, as indicated by MLC phosphorylation [17]. As shown in Figure 6C, ethanol significantly increased MLC phosphorylation in Caco-2 monolayers compared to controls (P, 0.0001). However, inhibition of MAPK with SB2035809, PD98059 and SP600125 significantly attenuated ethanol-induced induced MLC phosphorylation (P,0.0001; P,0.0001, P,0.000, respectively; Figure 6C), which was comparable to effects of the specific MLCK inhibitor 1-(5-iodonaphthalene-1-sulfonyl)-1Hhexahydro-1, 4-diazepine (ML-7) [17]. Since MLCK activation induces contraction of the peri-junctional actomyosin ring resulting in loss of TJ integrity [32], our experiments were extended further to elucidate effects of MAPK inhibition on ethanol-induced actin cytoskeleton rearrangement by immunoflu-orescence microscopy. As shown in figure 6D, in control condition, basal F-actin is uniformly distributed and organized in ring-like structures around the cellular periphery. However, in ethanol-treated monolayers, F-actin appears disorganized with thickened stress fibers ( Figure 6D, arrowhead). Inhibition of MAPK isoforms by SB2035809, PD98059 and SP600125 attenuated ethanol-induced remodeling of actin cytoskeleton ( Figure 6D).

Discussion
This study demonstrates that administration of a single moderate ethanol dose increases both small and large intestinal permeability in healthy volunteers.The impairment of intestinal barrier function was paralleled by redistribution of ZO-1 and occludin, down-regulation of ZO-1 and up-regulation of MLCK gene expression, and activation of MAPK isoforms (i.e.p38, ERK1/2 and JNK) in duodenal mucosa. In vitro data confirmed that ethanol induces MAPK isoforms phosphorylation, together with changes in permeability, redistribution of ZO-1 and occludin, disruption of F-actin ring and a significant increase in phosphorylation of MLC. These changes were reversed by the anti-oxidant L-cysteine and by MAPK inhibitors. Previous studies reporting on effects of ethanol intestinal barrier function have mainly been performed in alcoholics [20]. We are the first to report on effects of a moderate dose of ethanol on intestinal epithelial barrier combining data at functional level with data of epithelial TJs at At functional level, our results are in line with those of Robinson et al. [33] showing that a single moderate dose of ethanol increases excretion of urinary PEG 400, pointing to increased small intestinal permeability. We did not observe any effect on gastroduodenal permeability in the present study, in contrast to previous reports [5,34]. This is not surprising given the fact that ethanol was administered via a tube directly into the distal duodenum, bypassing the stomach and proximal duodenum.
Interestingly, we found not only increase in small intestinal permeability but also in colon. While an increase in colon permeability by ethanol has been reported in animal studies, so far no data are available with respect to the human condition. In rats, ethanol (3-4.5 g/kg) gavage for 12 days was found to increase colon permeability. This could be prevented by antibiotics and also by mast cell membrane stabilizers, pointing to involvement of colon microbiota and mast cell activation [35].
After ingestion, ethanol can reach colon mucosa by diffusion from blood [13]. Intestinal bacteria exhibit alcohol dehydrogenase (ALD) and to a lesser extent aldehyde dehydrogenase (ALDH) activity [36] with a lower capacity to metabolize acetaldehyde [37]. Together with less active ALDH in colon mucosa [38], this will result in intracolonic accumulation of acetaldehyde. Increased intracolonic accumulation of acetaldehyde is considered an important factor contributing to colorectal carcinogenesis [39] and hepatotoxicity [40].
Little is known about the duration of the increments in small and large intestinal permeability by ethanol. Robinson et al. [33] found the increase in small intestinal permeability after a single ethanol dose to be transient. This does not exclude that prolonged changes may occur after repeated, chronic ethanol exposure. Ethanol consumption has been associated with elevation of plasma endotoxins [41] and increased drug absorption [42], possibly through changes in intestinal barrier function. Recent data indicate that moderate red wine consumption (i.e. 20 g ethanol per day) by patients with inflammatory bowel diseases (IBD) in remission, in whom intestinal permeability already is increased, even further increases both small and large intestinal permeability [34].
We found that ethanol-induced increase in small intestinal permeability was associated with disruption of the key TJ proteins ZO-1 and occludin, and modulated mRNA expression of ZO-1 and MLCK in duodenal biopsies. Timing of biopsy sampling, i.e. 30 min after ethanol perfusion, was based on plasma ethanol levels. Mechanistic studies in humans on effects of moderate ethanol intake are lacking. However, in chronic alcoholics, Tang et al. [43] have reported increase in microRNA 212 expression in colon mucosa with subsequent post-transcriptional suppression of ZO-1 synthesis.
In vitro, down-regulation of ZO-1 and activation of MLCK has been shown to mediate ethanol-induced TJ disruption and actin cytoskeleton rearrangement, and consequently, increased intestinal epithelial permeability [17]. Other cell signaling pathways including MAPK have also been proposed as mechanism for ethanol-induced barrier dysfunction [44]. MAPK isoforms including p38, extracellular signal-regulated kinase (ERK) and stressactivated protein kinase/C-Jun N-terminal kinase (SAPK/JNK) represent a converging point for many signaling pathways and have been linked to broad spectrum of cellular responses to extracellular stimuli such as growth factors and stress [45]. In the present study, we observed that ethanol induced phosphorylation of p38, ERK1/2 and JNK in duodenal mucosa, indicating their activation. These observations led us to further explore their role in Caco-2 cell monolayers. Our in vitro results confirmed that ethanol exposure, at a concentration that can be achieved in blood after moderate consumption, can induce MAPK activation akin to H 2 O 2 , a potent oxidative stress inducer. In addition, ethanol increased permeability and induced abnormal distribution of TJ proteins and AJ proteins, which could be attenuated by MAPK inhibition. Such effects were not found after pretreatment with the anti-oxidant L-cysteine, indicating that ethanol activates MAPK, at least in part, through an oxidative stress-dependent mechanism. This finding is in line with reports by others and by our group, identifying oxidative stress as a key factor in the mechanisms by which ethanol and its metabolites disrupt intestinal barrier function [46,47]. Furthermore, oxidative stress has been found to activate MAPK in intestinal epithelial cells [48]. Therefore, our findings have relevance for ethanol-induced gut leakiness and indicate that the deleterious effects on barrier integrity can be reduced with antioxidants. The inability of ethanol to induce changes in Caco-2 monolayers after pretreatment with MAPK inhibitors suggests that ethanol requires MAPK activation to disrupt TJ, highlighting a pivotal role for MAPK in ethanol-induced barrier disruption. Previously, it has been found that p38-dependent MLCK activation is required for burn-induced intestinal permeability [49]. Our study provided evidences that ethanol leads to MAPKdependent MLC phosphorylation, F-actin rearrangement and stress fiber formation in Caco-2 cells. Disruption of actin cytoskeleton and interaction with TJ and AJ have been implicated in ethanol and acetaldehyde-induced loss of TJ integrity in Caco-2 cell monolayers [50]. Therefore, our in vitro data support the in vivo observations that ethanol increases intestinal permeability in association with up-regulation of MLCK gene expression and activation of MAPK in duodenal mucosa.
Ethanol and FAEEs including ethyl oleate and ethyl palmitate were measured in plasma, showing that ethanol can reach high levels within 30 min after administration accompanied by increased concentrations of the FAEEs ethyl oleate and ethyl palmitate. Acetaldehyde concentrations have not been measured due to its volatile character.
Duodenal biopsies obtained at the same time point (i.e. 30 min post-administration) have shown major changes on TJ at both posttranscriptional and transcriptional levels. Prior in vitro data have shown that ethanol [46], acetaldehyde [51] and FAEEs [47] can increase paracellular permeability in Caco-2 monolayers. Since intestinal mucosa is considered a major site for FAEE synthesis [52], together with a shift in ethanol metabolism into nonoxidative pathways resulting from low ALD activity in chronic alcoholics additive or even synergistic deleterious effects of ethanol and its oxidative and nonoxidative metabolites are expected to occur in vivo [53].
Limitations of our study include the lack of assessment of TJ integrity at the colon mucosal level. In addition, being a proof of concept study, ethanol was administered directly into the duodenum to overcome interindividual variations in its upper GI metabolism. Our data allow future studies to include more physiological approaches with a major focus on the colon.
In conclusion, a moderate dose of ethanol increases small and large intestinal permeability. Oxidative stress-mediated activation of MAPK is involved in ethanol-induced TJ and AJ disruption, actin cytoskeleton reorganization and consequently, intestinal epithelial barrier dysfunction. Our findings have potential implications for pathogenesis of alcoholic-related liver and GI disorders. A better understanding of molecular mechanisms involved in ethanol-induced gut leakiness may provide clues for development of preventive strategies.  Checklist S1 CONSORT checklist.