Chronic Heavy Alcohol Use is Associated with Upregulated Paneth Cell Antimicrobials in Gastric Mucosa

Chronic heavy alcohol consumption is associated with alcoholic hepatitis, cirrhosis, and other forms of alcoholic liver disease (ALD), which have been responsible for 493,300 deaths worldwide in 2010.1 However, the pathogenesis of ALD is not completely understood.2 Apart from the widely discussed roles of the ethanol (EtOH) metabolite acetaldehyde, increased oxidative stress and epigenetic events, there is increasing evidence pointing to a diminished gut barrier integrity and microbial overgrowth and/or dysbiosis.3, 4, 5, 6, 7, 8 These mechanisms might contribute to increased translocation of bacteria or their products across the mucosal barrier into the portal circulation4, 9 and the subsequent activation of inflammatory processes in the liver.5, 10, 11, 12 The exact mechanisms that may lead to a “leaky gut” and dysbiosis have not been extensively studied so far. This investigation therefore focusses on three aspects of the mucosal integrity in gastric and duodenal mucosa: the production of antimicrobial peptides (AMP), the induction of mucosal inflammation, and the activation of the Wnt pathway in response to alcohol exposure. In gastric and duodenal biopsies obtained from two groups of individuals largely differing in alcohol consumption, we compared the messenger RNA (mRNA) expression of AMPs originating in large part from epithelial cells, such as β-defensins (HBD1, 2, 4) and elafin, and AMPs derived from Paneth cells. Paneth cells, which are normally found at the bottom of small intestinal crypts, abundantly produce the α-defensins HD5 and HD6 and numerous other AMPs, such as lysozyme or secreted phospholipase A2 with activity against a multitude of microbes.13, 14 In a rat model with experimental (non-alcoholic) cirrhosis we could previously show that a compromised small intestinal Paneth cell function predisposed some animals to bacterial translocation.15 To assess the inflammatory process in the upper gastrointestinal (GI) tract in response to alcohol exposure, we also compared the transcriptional expression of four cytokines, the pro-inflammatory interleukins IL8, IL6, IL1β, and the anti-inflammatory IL10, between the two groups. Third, we studied the mRNA expression of the Wnt pathway factors TCF4, TCF1, and Wnt3, as well as the classical Wnt target gene Axin-2. The Wnt pathway is involved in small intestinal Paneth cell maturation and governs the epithelial regenerative homeostasis within the whole GI tract.16, 17, 18 Aberration in this pathway may compromise epithelial barrier integrity.19, 20 In an additional part, we performed in vitro experiments in a human gastric epithelial cell line to examine the effect of varying EtOH concentrations on the induction of AMPs and on the activity of the Wnt pathway. With this approach, we aimed at testing the hypothesis that heavy alcohol use might promote pro-inflammatory conditions, aberrations in innate epithelial antimicrobial defense, and/or changes in the Wnt pathway, and consequently in epithelial regeneration. Such effects could contribute to the “leaky gut” and dysbiosis, which is thought to promote ALD. In addition, shifts in GI mucosal homeostasis may be linked to other alcohol-related morbidities, such as the slightly increased gastric cancer risk, which is observed in heavy alcohol users.21, 22 Exclusion criteria were signs of severe infection at the time of the study (that is, bacteremia and/or disease states meeting two or more criteria of the systemic inflammatory response syndrome, as previously defined,23 as well as increased bleeding propensity due to impaired coagulation or overt portal hypertensive gastroduodenopathy. During esophagogastroduodenoscopy, which was performed no later than 5 days after admission, three biopsy specimens were obtained from the descending duodenum, three from the gastric antrum and three from the corpus for mRNA analysis and histology (using a Radial Jaw 4 forceps, Boston Scientific, Natick, MA). Biopsies were taken from a total of 17 controls and 22 patients chosen for this study. In some instances due to low mRNA amount or quality, not all assays could be performed. The minimal number of patients includes 20 patients with heavy alcohol use at each location, 16 controls for comparisons of corpus or duodenum, and 11 controls for the antrum. From additional biopsies of the antrum and corpus, a urease test was performed to determine colonization by Helicobacter pylori. Gastric pH was determined in gastric juice aspirated at the beginning of the endoscopy using a multiple indicator strip (pH0-14, Merck, Darmstadt, Germany). Also, on the day of endoscopy, blood was obtained for routine laboratory examinations. In the antrum, the HD5 expression level of controls was at 2100±900 copies per 10 ng RNA, (mean±s.e.m.), whereas patients with heavy alcohol use exhibited considerably higher copy numbers (365500±161600). A similar difference was found for HD6 (320±130 copies in controls and 58300±32600 copies in heavy alcohol use). For both, HD5 and HD6, this was statistically significant (P value for HD5: 0.0014, Bonferroni-adjusted: 0.022; P value for HD6 0.0007, adjusted 0.011). Secreted phospholipase A2 was also increased in patients with heavy alcohol use, though this trend was not significant (Figure 1c). The mRNA expression of lysozyme, which, in the stomach, is not restricted to Paneth cells,26 showed a different pattern. The highest lysozyme transcript levels were detected in the antrum, followed by duodenum and corpus (Figure 1d). Comparing heavy alcohol users with and without IM (Supplementary Figure 2), it can be observed that HD5 expression was increased in both subgroups, whereas controls had nearly negligible HD5 transcript levels. This suggests that IM is not a prerequisite for the induction of HD5 and is present only in a minority of patients. In contrast to the data from gastric samples, the Wnt transcription factors TCF1 (Figure 6d) and TCF4 (Figure 6f) exhibited no such induction. Wnt3 (Figure 6e) showed a trend for, but no significant, upregulation after alcohol stimulation, so did IL8 and elafin (Figure 6g, h). To confirm a potential activation of the Wnt pathway by EtOH stimulation, we additionally performed reporter gene experiments using promotor constructs for HD5 and HD6 as well as the artificial Wnt responsive TOPFlash. After 20 h of stimulation, we noted a significant dose-dependent induction of the promotor activity of both studied α-defensins (Figure 7a, b) and the Wnt responsive TopFlash (Figure 7c). The inducibility of HD5 seemed to be highest compared with HD6 and the TOPFlash promotor. Consistency in the induction of all three promotors suggests that EtOH has at least partially a direct effect on β-catenin-dependent Wnt pathway activity. This is also supported by the finding that a version of the HD5 promoter, which was mutated for all three potential TCF/Lef-binding sites (Wnt response elements) showed only a minor, statistically not significant induction after EtOH stimulation in a second approach using parallel hydrogen peroxide stimulation (Figure 7d). Alcohol has previously been shown to induce oxidative stress in gastric mucosa,27 we therefore used H2O2 in these experiments to test whether this oxidative stressor can mimic the EtOH-mediated effect on HD5 promoter activity. Whereas the three different applied H2O2 concentrations had no effect on the transcriptional regulation of HD5, an additional stimulation with 5% EtOH again significantly induced the promotor, in large parts in a Wnt-dependent manner (Figure 7d). This study represents the first report on an upregulation of Paneth cell-specific products as a consequence of chronic heavy alcohol abuse. In addition, it is the first to show a direct influence of EtOH on β-catenin-dependent Wnt signaling in a human gastric epithelial cell line. Contrary to our initial hypothesis we found no evidence for defects in the production of AMPs in the stomach or duodenum. Thus, changes in the synthesis of mucosal defensins at these GI locations do not account for the translocation of bacterial products, which is thought to have a role in the pathogenesis of ALD. Also, the cytokine patterns observed in this study do not indicate a major role of inflammation. Rather, the presented data suggest that the upregulation of AMPs in patients with heavy alcohol use is restricted to Paneth cell-specific antimicrobial products (HD5, HD6) in gastric mucosa. Interestingly, the occurrence of Paneth cells in gastric mucosa of some heavy alcohol users was also accompanied by IM, a phenomenon that has previously been linked to gastritis. The observed expression of Paneth cell AMPs in gastric tissue, where those products are normally absent, was paralleled by a trend toward induction of transcription factors of the Wnt pathway and its target gene Axin-2. Even though this effect was not statistically significant, we hypothesized that a potential deregulation of this important signaling cascade could be one mechanism that might disturb the epithelial homeostasis in patients, promoting an expression of Paneth cell AMPs. As such a disturbance in cell homeostasis could have important clinical implications, we tested our hypothesis by performing different in vitro experiments in a human gastric epithelial cell line. These experiments indeed indicated a direct influence of EtOH exposure on Paneth cell α-defensin expression and β-catenin-dependent Wnt signaling. The expression of Paneth cell products is normally restricted to the small intestine, where these specialized secretory cells are found at the bottom of crypts. Paneth cells represent the only differentiated epithelial cell in the gut, which is dependent on active Wnt for its maturation and function. In addition to their most abundant products HD5 and HD6, they generate a variety of mediators forming a broad arsenal of innate epithelial AMPs that combat bacteria, viruses and other microbial threats.28 Paneth cell defects have been implicated in the pathogenesis of small intestinal Crohn’s Disease, a multifactorial inflammatory disorder.29 Multiple in vivo mouse models highlight the importance of Paneth cell AMPs in protecting the organism from pathogens and in controlling the intestinal homeostasis toward resident microbiota.30, 31, 32, 33, 34 In an animal study investigating rats with chemically induced liver cirrhosis, we could previously show that bacterial translocation across the mucosal barrier was associated with reduced expression of Panth cell-derived antimicrobials in the small intestine.15 As mentioned, the cytokine patterns in both groups indicate that mucosal inflammation has no major role in the induction of Paneth cell products observed in patients with heavy alcohol use. We also examined whether relevant comorbidities contributed to the observed phenomena. As the three patients with liver cirrhosis did not display any gastric induction of Paneth cell HD5, and duodenal expression was not majorly affected (data not shown), we conclude that the observed induction of Paneth cell antimicrobials is independent of the degree of liver damage. The gastric pathogen H. pylori is another potential inducer of HD5 expression.35 In our study, there was no significant difference regarding HD5 transcript levels in H. pylori positive and negative patients (data not shown), supporting again the assumption that heavy alcohol use is an independent mechanism for the induction of Paneth cell AMPs. Notably, achlorhydria may increase bacterial density in the GI tract.36, 37 However, in both groups only a minority of patients (13.6%) and controls (22.2%) had a pH ≥6, and in both cases, these individuals did again not differ from those with a gastric pH<6 regarding their HD5 level (data not shown). Since alcohol has been shown to have direct necrotizing effects and can promote oxidative stress in gastric mucosal cells,27, 38 the induction of Paneth cell AMPs could be a protective mechanism, aiming at restoring a weakened epithelial defense during alcohol-associated damage. On the other hand, it could also represent a first symptom of epithelial transformation.35 IM, which includes the occurrence of Paneth cells in upper GI tissues, may occur in the esophagus (Barrett’s esophagus) and in the distal stomach during gastritis. Barrett’s esophagus develops most likely due to prolonged exposure to gastric acid and bile salts, whereas gastric IM has been associated with H. pylori colonization. In the above-mentioned study,35 the gastric form of IM has been shown to include the unequivocal presence of Paneth cells in some patients, whereas others only exhibit HD5-positive cells without typical Paneth cell features in standard histology. In the present study, all patients with histologically proven IM had no indication of H. pylori infection. Also, the induction of HD5 mRNA was not only observed in heavy alcohol users with IM, but also in a subgroup without this condition (Supplementary Figure 2). Thus, it appears that IM is not a prerequisite for the production of HD5 in the stomach. The second group of AMPs studied in our investigations comprised the β-defensins HBD1, 2, and 4, which are of epithelial origin, but are also found in other cells such as macrophages, as well as the antiprotease elafin. For these factors, we found a high variability, but no significant differences. As mentioned above, one might speculate that heavy alcohol consumption could promote a disturbance in Wnt and consequently in epithelial homeostasis. In the context of inflammation39 and/or H. pylori40 associated gastric cancer development, a role for upregulated Wnt is well accepted and extensively studied. With regard to alcohol, our data on Wnt pathway activation in a gastric cell line, represent a novel mechanism. Our reporter gene assays, as well as the Axin-2 mRNA induction, demonstrated a direct activation of β-catenin-dependent Wnt in a short term in vitro setting. However, in the same approach, the Wnt transcription factors TCF1 and TCF4 exhibited no significant upregulation, suggesting that the trend for a higher expression found in the tissue of patients with heavy alcohol use may not be a direct but possibly a long-term effect. Although still speculative, the Wnt induction observed in this study may be mechanistically related to the slight increase in cancer risk, especially of non-cardia gastric cancer in heavy drinkers.21, 22 Although data on a connection between alcohol and Wnt signaling are rare, results obtained in another tissue support a link between the two. In the liver, it was recently shown that chronic alcohol feeding activates β-catenin-dependent Wnt to increase hepatocyte proliferation in mice, which ultimately promoted tumorigenesis in this model.41 If such a relationship turned out to be true in gastric tissue of heavy alcohol users, screening for HD5 expression in gastric biopsies, as a likely symptom of induced Wnt activity, could constitute a potential prognostic marker. A clear limitation of the study is the restriction to proximal GI tissue due to the lack of jejunal or ileal biopsies in our patient study. Another limitation is based on the fact that, owing to the restricted number of study participants, a further subdivision of controls into total abstainers vs. moderate alcohol consumers was not feasible. Also, as all data on alcohol intake were self-reported, we cannot exclude errors in the assessed levels of consumption. However, as according to our definition, the reported mean of daily alcohol consumption differed widely between the two groups, we are certain that such errors would not lead to a major bias. The limited number of patients and controls might be a reason why many distinct differences in mRNA expression were lacking statistical significance. This was especially the case regarding TCF1, TCF4, Axin-2, HBD2, and elafin. However, the latter two factors are also highly regulated by different pathways, including NFκB, and MAPKinases, so a high inter-individual variability is to be expected. Despite these limitations, the presented data suggest that chronic heavy alcohol consumption has the potential to promote the expression of Paneth cell AMPs, which subsequently may alter the innate gastric antimicrobial defense. Based on our mechanistic studies, this seems to be linked to changes in gastric β-catenin-dependent Wnt. Whether this is a potential protective, or rather a pathogenic mechanism, and whether similar observations can also be made in other tissues will require further studies. Guarantor of the article: Jan Wehkamp, MD. Specific Author contribution: M.J.O. and C.S. designed the study and analyzed the data. M.J.O. designed and planned the experimental work. C.S. phenotyped the patient cohort and organized sample collection. L.C. performed in vitro reporter gene experiments; S.R.D.S. performed in vitro mRNA experiments. E.F.S. and J.W. were involved in study design. M.J.O., C.S., E.F.S., and J.W. discussed the data and wrote the manuscript. Financial Support: This work was supported by the Robert Bosch Foundation and by the Wissenschaftsförderung der deutschen Brauwirtschaft (Project B70). JW is a Heisenberg grantee of Deutsche Forschungsgemeinschaft who also funded the project. Potential competing interests: This project was funded by the Wissenschaftsförderung der deutschen Brauwirtschaft (Science Fund of the German brewing industry). We thank Jutta Bader, Kathleen Siegel, and Marion Strauss for excellent technical assistance. Supplementary Information accompanies this paper on the Clinical and Translational Gastroenterology website