Department of Pathology, Southern California Research Center for Alcoholic Liver and Pancreatic Diseases and Cirrhosis, Keck School of Medicine of the University of Southern California, Los Angeles, CA
Department of Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA
Potential conflict of interest: Nothing to report.
This study was supported in part by National Institutes of Health grants K08 DK081830 and R01 AA020703 (to B. S.) and by the Alcoholic Beverage Medical Research Foundation/The Foundation for Alcohol Research (to B. S.). The study was also supported by the Pilot Project Program (to B. S.) and the Lee Summer Fellowship Award (to P. H.) of the Southern California Research Center for Alcoholic Liver and Pancreatic Diseases and Cirrhosis (grant P50AA11999) funded by the National Institute on Alcohol Abuse and Alcoholism. The study also received support from the University of California San Diego Digestive Diseases Research Development Center, US Public Health Service (grant DK080506).
The intestinal mucus layer protects the epithelium from noxious agents, viruses, and pathogenic bacteria present in the gastrointestinal tract. It is composed of mucins, predominantly mucin (Muc) 2, secreted by goblet cells of the intestine. Experimental alcoholic liver disease requires translocation of bacterial products across the intestinal barrier into the systemic circulation, which induces an inflammatory response in the liver and contributes to steatohepatitis. We investigated the roles of the intestinal mucus layer, and in particular Muc2, in development of experimental alcohol-associated liver disease in mice. We studied experimental alcohol-induced liver disease, induced by the Tsukamoto-French method (which involves continuous intragastric feeding of an isocaloric diet or alcohol) in wild-type and Muc2−/− mice. Muc2−/− mice showed less alcohol-induced liver injury and steatosis than developed in wild-type mice. Most notably, Muc2−/− mice had significantly lower plasma levels of lipopolysaccharide than wild-type mice after alcohol feeding. In contrast to wild-type mice, Muc2−/− mice were protected from alcohol-associated microbiome changes that are dependent on intestinal mucins. The antimicrobial proteins regenerating islet-derived 3 beta and gamma were expressed at significantly higher levels in the jejunum of Muc2−/− mice fed the isocaloric diet or alcohol compared with wild-type mice. Consequently, Muc2−/− mice showed increased killing of commensal bacteria and prevented intestinal bacterial overgrowth. Conclusion: Muc2−/− mice are protected from intestinal bacterial overgrowth and dysbiosis in response to alcohol feeding. Subsequently, lower amounts of bacterial products such as endotoxin translocate into the systemic circulation, decreasing liver disease. (HEPATOLOGY 2013;)
Liver cirrhosis is the twelfth leading cause of death in the United States, and 48% of all deaths from cirrhosis are alcohol-related.1 Alcoholic liver disease comprises hepatic steatosis, which may progress to alcoholic hepatitis, fibrosis, and cirrhosis.2 There is strong evidence for a gut-liver axis that is causatively related to alcohol-induced liver disease, both in patients and in experimental animal models. Gastrointestinal permeability is greater in alcoholics compared with normal subjects.3, 4 Several animal studies have demonstrated that ethanol disrupts the intestinal epithelial barrier function via a direct effect of ethanol and/or its metabolite acetaldehyde.5 Ethanol-induced gut leakiness results in elevated plasma levels of lipopolysaccharide (LPS) or endotoxin, a major component of the gram-negative bacterial outer membrane, and subsequent liver injury.6-8 Endotoxemia is more prevalent in patients with alcoholic liver disease compared with normal subjects, and plasma endotoxin levels correlate with the severity of liver damage in patients with alcoholic hepatitis.9-12 The most convincing evidence for a role of gut-derived endotoxin comes from mice harboring a genetic deletion in the LPS signaling pathway. Mice deficient in Toll-like receptor (TLR) 4 as the cellular LPS receptor, CD14 as the cellular co-receptor for LPS, or intracellular signaling molecules downstream of the LPS receptor are resistant to alcohol-induced liver injury.13-15 In addition, selective intestinal decontamination with nonabsorbable antibiotics reduces plasma endotoxin levels and prevents experimental alcoholic liver disease.16-18 Although not an established therapy, treatment with antibiotics also improves liver function in patients with alcoholic cirrhosis.19
The intestinal mucus layer forms a physical barrier between the underlying epithelium and the lumen of the gastrointestinal tract and protects the epithelium against noxious agents, viruses, and pathogenic bacteria. It consists of two separate sublayers: the inner layer is attached to the epithelial cell layer and is devoid of bacteria; the outer layer can be washed off easily and is colonized by bacteria.20, 21 The intestinal mucus layer is composed of mucins that are synthesized and secreted by intestinal goblet cells.22 Two different types of mucins exist: secreted, or gel-forming mucins, and membrane-bound mucins. There are three gastrointestinal secreted mucins (Muc2, Muc5AC, and Muc6) that are characteristically large, heavily O-glycosylated glycoproteins assembled into oligomers that contribute to the viscous properties of intestinal mucus layer.23 The intestinal membrane-bound mucins (Muc1, Muc3-4, Muc12-13, and Muc17) protect against pathogens that penetrate the inner mucus layer.24 The major and most abundant secreted mucin in the small and large intestine is mucin-2.25 Mice deficient in Muc2 are prone to colorectal cancer and appear to have a disrupted epithelial homeostasis.25 Specific clinical symptoms such as spontaneous colitis depend on their genetic mouse strain background.26
It has been reported that the intestinal mucus increases after alcohol feeding in rats.27 However, there are currently no patient data or experimental studies assessing the functional contribution of the intestinal mucus layer in alcoholic liver disease. We therefore took an unbiased approach to study the role of the intestinal mucus layer, in particular Muc2, using a mouse model of alcoholic liver injury and steatosis.
Male wild-type mice (C57BL/6J) were purchased from The Jackson Laboratory or bred in the vivarium associated with our laboratory. Male Muc2−/− mice (back-crossed to C57BL/6J for more than 10 generations) were kindly provided by Anna Velcich (Albert Einstein College of Medicine, Yeshiva University, New York, NY). Age-matched mice were used for this study. All animals received humane care in compliance with institutional guidelines. The intragastric feeding model of continuous ethanol infusion in mice has been described.28
The Lieber DeCarli diet model of alcohol feeding for 2 weeks was used to determine intestinal permeability and for an in vivo luminal killing assay. We opted to assess intestinal permeability in a complementary and noninvasive mouse model of alcoholic steatohepatitis using the Lieber DeCarli diet, because prior surgery and the implanted gastrostomy catheter could affect accurate assessment of intestinal permeability. To avoid two surgeries in the same mouse, we also chose to assess in vivo luminal killing of bacteria in mice that were fed the Lieber DeCarli diet.
Additional materials and methods are described in the Supporting Information.
Alcohol Abuse Increases the Thickness of the Intestinal Mucus Layer in Humans.
It has been reported that chronic alcohol feeding increases the total mucus content in the small intestine in rats.27 We have confirmed these data in humans. Alcoholics show a significant increase in the thickness of the mucus layer on duodenal biopsies compared with healthy humans (Fig. 1A,B).
Muc2-Deficient Mice Have Decreased Alcoholic Steatohepatitis.
To investigate the role of the intestinal mucus layer in experimental alcoholic liver disease, we used mice harboring a genetic deletion in the Muc2 gene.25 Muc2 is the most abundant secreted mucin in the gastrointestinal tract25 and its absence results in a significantly thinner mucus layer in mice as shown by Periodic acid–Schiff (PAS) staining of the small intestine (Fig. 4A). To confirm that Muc2 expression is largely restricted to the intestine, we measured Muc2 messenger RNA levels in several organs from wild-type mice. Muc2 gene expression was highest in the small and large intestine, but it was undetectable in the liver or bone marrow–derived cells (Supporting Fig. 1A). These findings were confirmed by immunofluorescent staining. Muc2 protein was abundantly expressed in the small intestine (Supporting Fig. 1B, left panel), but undetectable in the liver of wild-type mice (Supporting Fig. 1B, right panel). Small intestine from Muc2-deficient mice served as a negative staining control (Supporting Fig. 1B, middle panel).
We therefore subjected wild-type and Muc2−/− mice to the intragastric feeding model of continuous ethanol infusion for 1 week. Mice fed an isocaloric diet served as controls. Administration of ethanol lead to a comparable increase of liver weight to body weight ratio (Supporting Fig. 2A). Plasma alanine aminotransferase (ALT) levels as measures for liver injury were significantly lower in alcohol-fed Muc2−/− mice compared with wild-type mice (Fig. 2A). Micro- and macrovesicular steatosis occurred after 1 week following alcohol administration compared with wild-type mice receiving an isocaloric diet. Hepatic fat accumulation was markedly lower in Muc2−/− mice compared with wild-type mice following 1 week of continuous intragastric ethanol feeding (Fig. 2B). This was confirmed by lower hepatic triglycerides in Muc2−/− mice after alcohol administration (Fig. 2C). Plasma triglyceride levels were similar between wild-type and Muc2−/− mice fed an isocaloric and alcohol diet intragastrically for 1 week (Supporting Fig. 2B) suggesting no difference in intestinal lipid absorption. Hepatic oxidative stress was also significantly lower in Muc2−/− mice compared with wild-type mice following 1 week of intragastric alcohol feeding, as supported by thiobarbituric acid reactive substances (TBARS) assay (Fig. 2D) and by staining for 4-hydroxynonenal (Fig. 2E). Thus, Muc2 deficiency, and hence a thinner intestinal mucus layer, ameliorates experimental alcohol-induced steatohepatitis.
Alcohol Metabolism and Expression of Intestinal Mucins After Ethanol Feeding in Muc2-Deficient Mice.
To explain the different hepatic phenotype, we investigated whether Muc2 deficiency affects the intestinal absorption or hepatic metabolism of alcohol. Plasma alcohol levels were found to be comparable in wild-type and Muc2−/− mice following 1 week of intragastric alcohol feeding (Fig. 3A). Alcohol dehydrogenase (Adh) and cytochrome p450 enzyme 2E1 (Cyp2E1) are the two main hepatic enzymes to metabolize alcohol and to convert alcohol to acetaldehyde.29 Microsomal Cyp2E1 protein was similarly up-regulated in the ethanol-treated groups (Fig. 3B). Despite higher hepatic Adh activity in Muc2−/− mice compared with wild-type mice after intragastric administration of an isocaloric diet that was not observed after ethanol administration (Fig. 3C), plasma acetaldehyde levels were not different following 1 week of intragastric alcohol feeding (Fig. 3D). To investigate whether the absence of Muc2 results in a compensatory up-regulation of other intestinal mucins after ethanol administration, intestinal gene and protein expression of several mucins was assessed. Deficiency in Muc2 did not result in a compensatory increase in the thickness of the intestinal mucus layer following intragastric alcohol feeding (Fig. 4A). There was no significant difference in the gene expression of secreted mucin Muc6 or of membrane-bound mucins (such as Muc1 and Muc4) in Muc2−/− mice relative to wild-type mice after 1 week of intragastric feeding of ethanol (Fig. 4B). These findings were confirmed using immunohistochemistry for Muc1 and Muc4 in small intestinal sections of wild-type and Muc2−/− mice fed an intragastric isocaloric or alcohol diet (Fig. 4C,D).
Muc2-Deficient Mice Exhibit Lower Plasma LPS Levels and Are Protected from Microbiome Changes After Alcohol Feeding.
Alcoholic steatohepatitis is dependent on endotoxin derived from intestinal bacteria.2, 30 Since Muc2 is expressed in the intestine but not the liver, we next investigated whether translocation of bacterial products from the intestine to the systemic circulation is affected by the absence of Muc2. Indeed, systemic endotoxin levels were significantly lower in Muc2−/− mice that were fed an isocaloric diet and alcohol intragastrically for 1 week compared with wild-type mice (Fig. 5A). Altered intestinal permeability or a quantitative decrease of the intestinal microflora might allow less endotoxin to escape from the gut into the systemic circulation. We therefore assessed intestinal permeability by measuring fecal albumin following a Lieber DeCarli diet for 2 weeks.31 Fecal albumin was higher in Muc2-deficient mice at baseline and after alcohol feeding indicative of increased intestinal permeability (Fig. 5B). To confirm our findings and to directly assess intestinal permeability, we used an in vivo method by measuring recovery of ingested dextran labeled with fluorescein isothiocyanate. Isocaloric Lieber DeCarli diet or alcohol feeding for 2 weeks resulted in a significant increase of fluorescence in the plasma of Muc2−/− mice compared with wild-type mice indicative of increased intestinal permeability (Fig. 5C). Thus, despite a leakier gut barrier, Muc2−/− mice showed lower translocation of bacterial products.
Only a minority of the enteric bacteria can be cultured by conventional culture techniques.32 To assess quantitative changes in the intestinal microbiome, the total bacterial load was measured by quantitative polymerase chain reaction using universal 16S ribosomal RNA bacterial primer sets. As reported by us,28 intragastric ethanol feeding induced intestinal bacterial overgrowth in wild-type mice compared with wild-type mice fed an isocaloric diet (Fig. 5D). Interestingly, Muc2−/− mice are protected from intestinal bacterial overgrowth after alcohol feeding (Fig. 5D).
We have also shown that alcohol-associated changes in the enteric microbiome are characterized by a significant suppression of the commensal probiotic microflora, including Lactobacillus.28 We have confirmed a significant reduction of Lactobacillus in wild-type mice following intragastric ethanol feeding for 1 week compared with control animals (Fig. 5E). Muc2−/− mice are not only protected from a suppression of Lactobacillus, they actually demonstrate higher numbers of Lactobacillus after alcohol feeding compared with control Muc2−/− mice (Fig. 5E). In addition, we have previously shown and confirmed that chronic intragastric alcohol feeding for 3 weeks results in an increase of Gram-negative33Akkermansia muciniphila (Fig. 5F, left panel).28 Although no significant change was observed in wild-type mice following 1 week of intragastric alcohol feeding compared with isocaloric diet feeding, A. muciniphila was significantly lower in Muc2−/− mice compared with wild-type mice after alcohol feeding (Fig. 5F, middle panel). Growth of A. muciniphila is dependent on the presence of mucus in vitro, but not ethanol (Fig. 5F, right panel). Thus, the absence of Muc2 results in dysbiosis characterized by a decrease in gram-negative A. muciniphila that likely contributes to lower systemic levels of endotoxin. Littermate and nonlittermate wild-type mice did not show significant differences at baseline in alanine aminotransferase (ALT); intestinal permeability; intestinal bacterial burden; the quantity of the two major intestinal bacterial phyla, Bacteroidetes and Firmicutes; and Lactobacillus (Supporting Fig. 3). Taken together, Muc2−/− mice are protected from alcohol-associated quantitative and qualitative changes in the microbiome and have lower plasma levels of LPS.
Antimicrobial Protein Expression and Activity Are Enhanced in the Intestine of Muc2-Deficient Mice.
Several factors control the bacterial load of intestine including host antimicrobial molecules that are secreted by epithelial cells and Paneth cells. We have previously reported that the expression of regenerating islet-derived 3 beta (Reg3b) and gamma (Reg3g) are reduced in the small intestine of mice fed alcohol compared with control mice.28 The inhibition was pronounced in the proximal small intestine, the site with the largest relative increase in luminal bacteria and the highest intraluminal alcohol concentrations.28 We confirmed alcohol-induced inhibition of Reg3b and Reg3g protein expression in the jejunum of wild-type mice (Fig. 6A,C). Strikingly, Reg3b and Reg3g expression was much higher in Muc2−/− mice receiving an isocaloric diet or alcohol via an intragastric feeding tube for 1 week compared with wild-type mice (Fig. 6A,C). Other antimicrobial molecules such as cathelicidin antimicrobial peptide (Camp) or defensin beta 1 (Defb1) show similar responses to intragastric alcohol in wild-type and Muc2−/− mice (Fig. 6B). Interleukin-22 (IL-22) is required for the induction of intestinal Reg3b and Reg3g expression.34 IL-22 gene expression showed a trend to be higher expressed in the small intestine of isocaloric and ethanol-fed Muc2−/− mice compared with wild-type mice (Supporting Fig. 4). These results suggest that Muc2 deficiency results in a strong induction of antimicrobial factors that restrict survival or replication of the commensal microflora.
To investigate whether these findings directly translate into quantitative alterations of the commensal microflora, we used an in vivo luminal killing assay of nonpathogenic Escherichia coli in the gut of wild-type and Muc2-deficient mice as described by us.35, 36 A 4-cm loop of the proximal jejunum was ligated (without interrupting the blood supply) in anesthetized mice and injected with bioluminescent, nonpathogenic E. coli. To analyze luminal survival and killing, IVIS imaging of bioluminescent E. coli was performed at 0 minutes and 3.5 hours after injection of bacteria into ligated jejunal loops. Whereas loops of Muc2−/− mice after feeding a Lieber DeCarli isocaloric diet or alcohol for 2 weeks were essentially devoid of luminescent bacteria, bioluminescent bacteria were found in alcohol and control fed wild-type mice at a significantly higher percentage after 3.5 hours (Fig. 7A,B). This result suggests that commensal bacteria are killed more effectively in jejunal loops of Muc2−/− mice than in wild-type mice (Fig. 7C), thereby limiting intestinal bacterial overgrowth after alcohol feeding.
To demonstrate that Muc2−/− mice are protected due to intestinal changes, but not secondary to hepatic adaptations, we have chosen to administer LPS enterally. When mice were given LPS through the intragastric feeding tube daily for 1 week in addition to ethanol, increased bacterial products from gram-negative E. coli were found in the livers of Muc2−/− mice comparable to levels seen in wild-type mice (Supporting Fig. 5A). This restoration of hepatic endotoxemia exacerbated alcoholic steatohepatitis in Muc2−/− mice fed ethanol and LPS (Supporting Fig. 5B,C). This supports our finding that a decreased endotoxemia contributes to the protection of Muc2−/− mice from experimental alcoholic liver disease despite a leakier gut.
The first, and arguably best, opportunity for the body to limit toxic effects of orally administered alcohol is the gastrointestinal tract. In this study, we investigated the role of mucins and in particular intestinal Muc2 in alcoholic steatohepatitis. Alcohol increases the thickness of the intestinal mucus layer in patients with alcohol abuse. Alcoholic steatohepatitis was ameliorated in mice deficient in Muc2, which could not be explained by altered ethanol metabolism or a compensatory up-regulation of other intestinal mucins. We provide evidence that Muc2 deficiency results in altered microbiome composition and an increased expression of antimicrobial molecules. This is associated with enhanced intraluminal killing of bacteria and a decrease in the intestinal bacterial burden in Muc2-deficient mice. Less bacterial products such as LPS translocate from the intestine to the systemic circulation and cause less liver injury and steatosis (Fig. 8).
Experimental alcoholic liver disease is dependent on gut-derived bacterial products that drive liver injury and steatosis.2 There is an evolving concept that changes in the gut microflora and microbiome affect bacterial translocation, both in patients and in experimental models of alcoholic steatohepatitis. Increased plasma endotoxin and bacterial DNA have been associated with small intestinal bacterial overgrowth in patients with cirrhosis. Furthermore, small intestinal bacterial overgrowth was an independent and major risk factor for the presence of bacterial DNA in the systemic circulation in patients with cirrhosis.37, 38 Interestingly, selective intestinal decontamination decreased translocation to the mesenteric lymph nodes to the level of patients without cirrhosis, and although not an established therapy, it also benefits patients with alcoholic liver cirrhosis by improving their liver function.19, 39 Thus, intestinal bacterial overgrowth predisposes patients with liver disease to bacterial translocation.
We have recently demonstrated quantitative (overgrowth) changes in the enteric microbiome using a model of intragastric alcohol feeding in mice. Suppression of alcohol-induced intestinal bacterial overgrowth with nonabsorbable antibiotics decreases systemic levels of LPS and ameliorates alcoholic steatohepatitis in rats.18 On the other hand, if bacterial overgrowth is induced experimentally in the small intestine, this causes liver inflammation and injury.40
We speculate that activation of the mucosal innate immune system, as demonstrated by increased levels of Reg3b and Reg3g, contributes to reduced intestinal bacterial overgrowth in Muc2−/− mice. Prebiotics restore Reg3b and Reg3g expression, limit bacterial overgrowth and ameliorate alcohol-induced steatohepatitis.28 However, other antimicrobial molecules or components of the mucosal innate immune system might work in concert with Reg3b and Reg3g, and future studies are required for further investigations. Thus, based on our study, we propose a concept in which suppression of intestinal bacterial overgrowth by host antimicrobial molecules results in a decreased availability of intraluminal bacterial products. Less of these products are able to cross the intestinal barrier into the portal circulation, which eventually limits alcoholic liver disease.
We have also recently demonstrated qualitative changes in the enteric microbiome (dysbiosis) using a model of intragastric alcohol feeding in mice. Alcohol-associated dysbiosis is characterized by a profound suppression of commensal probiotic bacteria, including Lactobacillus.28 Several studies have shown that a restoration of eubiosis using supplemental probiotic Lactobacillus ameliorates alcoholic steatohepatitis in rodents.41, 42 Interestingly, Muc2−/− mice are protected from alcohol-associated changes in the microbial composition, including a suppression of Lactobacillus. In addition, Muc2 deficiency limits the proliferation of bacteria (such as gram-negative A. muciniphila) that use mucins as carbon source. Thus, the absence of Muc2 prevents alcohol-associated dysbiosis, restores intestinal homeostasis, and inhibits experimental alcoholic liver disease.
The mucus layer has a very important role in the intestine. It largely prevents the translocation of viable bacteria from the gut lumen to extraintestinal organs such as lymph nodes and the systemic circulation.43 The absence of Muc2 as a major component of the intestinal mucus layer has no obvious adverse effect for the gut-liver axis at baseline without challenge. The thickness of the mucus layer increases in alcoholics as shown in our study and by others in rodents,27 which could be interpreted as a defense against alcohol or more likely against intestinal epithelial cell injury. And indeed, enteric infections also increase the Muc2 production and the mucus layer.43 A downside of this obvious good reaction of the intestine of increasing the mucus layer is that the vigorous immune defense system of enterocytes against bacteria is impaired. We currently can only speculate how an increase in the mucus layer might affect the expression of antimicrobial molecules as part of the mucosal innate immune system. One possibility is that bacterial ligands are not as accessible to enterocytes to stimulate the expression of antimicrobials. Reg3g expression has been shown to be TLR5 and IL-22–dependent and can be induced by flagellin,34, 44, 45 but intestinal IL-22 did not correlate with Reg3 protein expression in our study. Indeed, Reg3g expression is induced through cell-autonomous MyD88-dependent TLR activation in intestinal Paneth cells.46 Thus, when the body is challenged with alcohol, the thickness of the intestinal mucus layer increases, and less antimicrobial molecules reach the lumen to control proliferation of intestinal bacteria. An apparently good reaction of the body to respond to alcohol-induced epithelial cell damage impairs the mucosal innate immune system and results in the intestinal homeostasis system to fail. One should note that this is not a general response in Muc2-deficient mice upon intestinal injury or inflammation, but is rather specific for alcohol. Other studies have shown that colitis induced by the pathogen Citrobacter rodentium is exacerbated in Muc2-deficient mice.43
Our study demonstrates that deficiency of one host gene Muc2 that is not expressed in the liver or in inflammatory cells, but largely restricted to the intestine, decreases alcoholic steatohepatitis. Our findings are consistent with the large body of evidence that experimental alcoholic liver disease is driven by the gut. Alcohol-associated changes in the microbiome, and in particular intestinal bacterial overgrowth, contributes to alcohol-induced liver injury. Taken together, our study emphasizes again the importance of the gut-liver axis. Treatment targeting the mucosal innate immune system and intestinal bacterial overgrowth might contribute to the clinical management of alcohol-induced liver disease.
We thank Akiko Ueno and Raul Lazaro from the Animal Core facility of the Southern California Research Center for Alcoholic Liver and Pancreatic Diseases and Cirrhosis, University of Southern California, for performing animal studies described in this study. We also thank Derick Han for tissue sharing and Yaron Niv and Anna Velcich for helpful discussion and careful reading of the manuscript.