Aryl Hydrocarbon Receptor Deficiency in Intestinal Epithelial Cells Aggravates Alcohol-Related Liver Disease

Abstract

Background & Aims

The ligand-activated transcription factor, aryl hydrocarbon receptor (AHR) can sense xenobiotics, dietary, microbial, and metabolic cues. Roles of Ahr in intestinal epithelial cells (IECs) have been much less elucidated compared with those in intestinal innate immune cells. Here, we explored whether the IEC intrinsic Ahr could modulate the development of alcohol-related liver disease (ALD) via the gut–liver axis.

Methods

Mice with IEC specific Ahr deficiency (Ahr^ΔIEC) were generated and fed with a control or ethanol diet. Alterations of intestinal microbiota and metabolites were investigated by 16S ribosomal RNA sequencing, metagenomics, and untargeted metabolomics. AHR agonists were used to evaluate the therapeutic potentials of intestinal Ahr activation for ALD treatment.

Results

Ahr^ΔIEC mice showed more severe liver injury after ethanol feeding than control mice. Ahr deficiency in IECs altered the intestinal metabolite composition, creating an environment that promoted the overgrowth of Helicobacter hepaticus and Helicobacter ganmani in the gut, enhancing their translocation to mesenteric lymph nodes and liver. Among the altered metabolites, isobutyric acid was increased in the cecum of ethanol-fed Ahr^ΔIEC mice relative to control mice. Furthermore, both H.hepaticus and isobutyric acid administration aggravated ethanol-induced liver injury in vivo and in vitro. Supplementation with AHR agonists, 6-formylindolo[3,2-b]carbazole and indole-3-carbinol, protected mice from ALD development by specifically activating intestinal Ahr without affecting liver Ahr function. Alcoholic patients showed lower intestinal AHR expression and higher H.hepaticus levels compared with healthy individuals.

Conclusions

Our results indicate that targeted restoration of IEC intrinsic Ahr function may present as a novel approach for ALD treatment.

Graphical abstract

Summary.

Herein, we identified that Ahr deficiency in intestinal epithelial cells enriched Helicobacter hepaticus and Helicobacter ganmani in the gut, promoted their translocation to liver, and aggravated alcohol-related liver disease (ALD) development. Dietary supplementation with AHR agonists effectively ameliorated ALD in mice, providing a new strategy for ALD treatment.

Alcohol-related liver disease (ALD) is a leading cause of liver-related morbidity and mortality worldwide, and approximately 2 million people die of liver diseases each year, of which up to 50% mortality from cirrhosis can be attributed to alcohol.^1,2 Over the past 25 years, adult per capita alcohol consumption increased by 10%.³ The exact mechanisms underpinning the pathogenesis of ALD remain unclear. Therefore, elucidating the mechanism and exploring novel therapies for ALD urgently are needed given that the current treatments are very scarce.1, 2, 3, 4

Aryl hydrocarbon receptor (Ahr) is a highly conserved, ligand-inducible transcription factor that integrates environmental, dietary, microbial, and metabolic cues to control the adaptation of multicellular organisms to environmental challenges.^5,6 The Ahr is expressed in many mammalian tissues, especially in the liver, intestine, and kidney.⁷ In the intestine, Ahr is expressed mainly by epithelial cells and innate immune cells, and plays an important role in the regulation of innate immunity.⁶ For example, it regulates the number of intraepithelial lymphocytes,⁸ controls the production of interleukin 22 by innate lymphoid cells,^9,10 and senses the bacterial virulence factors then leads to antibacterial responses.¹¹ Compared with the role of Ahr in innate immune cells in the intestine, the function of Ahr in intestinal epithelial cells (IECs) has not been studied. One previous study showed that Ahr in IECs was associated with maintenance of epithelial barrier function.¹² Given that chronic alcohol consumption can disrupt the intestinal epithelial barrier and alter gut microecology, which contributes to ALD,^13,14 we investigated whether intestinal epithelial Ahr affects the progression of ALD.

Gut dysbiosis, which contributes to the pathogenesis of ALD, could present as intestinal barrier dysfunction, gut microbiota alteration, and immune system dysregulation.14, 15, 16 Bacterial components and metabolites translocate from the leaky gut through blood and lymphatics to the liver in animal models and patients with ALD.^14,16,17 Once the microbial products translocate to the liver, they activate the innate immune receptors and induce the increased expression of hepatic inflammatory cytokines and lipogenesis-related factors, which promote the development of ALD.16, 17, 18 Not only bacterial products, but viable microbiota also can translocate to the liver. Using regenerating islet-derived 3 gamma (Reg3g) or Reg3b-deficient mice, we previously showed that the translocation of intestinal mucosa-associated bacteria to mesenteric lymph nodes and the liver could aggravate the progression of ALD.¹⁹ In addition, translocating Enterococcus to the liver could increase interleukin 1β secretion via the pathogen-recognition receptor Toll-like receptor 2 in Kupffer cells, resulting in hepatic inflammation and hepatocytes death,²⁰ while administration of bacteriophages against Enterococcus faecalis significantly reduced the severity of ALD in mice.²¹ Translocation of gut bacteria to the liver also occurred in other liver diseases such as autoimmune hepatitis.²² Because the translocation of microbial products and intestinal bacteria play such an important role in the pathogenesis of chronic liver diseases, including ALD, it is quite essential to identify new metabolites and strains of bacteria translocating to the liver, and reveal the pathogenic mechanism by which they affect the development of ALD.

Here, we found that ethanol feeding reduced the intestinal AHR expression in mice and human beings. Compared with Ahr^fl/fl mice, IEC-specific Ahr deletion mice (Ahr^ΔIEC) showed aggravated liver injury after ethanol feeding. And the levels of Helicobacter hepaticus, Helicobacter ganmani, and isobutyric acid (IBA) were enriched in ethanol-fed Ahr^ΔIEC mice compared with Ahr^fl/fl mice. In line with the increase of IBA, bacterial gene expression of ilvE, bkdA, and pdhD, which were responsible for metabolizing valine to IBA, were increased in ethanol-fed Ahr^ΔIEC mice. Further investigation showed that H. hepaticus and IBA could aggravate ethanol-induced liver injury in vivo and in vitro. Moreover, oral administration with agonists of AHR, 6-formylindolo[3,2-b]carbazole (FICZ)¹⁰ and indole-3-carbinol (I3C),⁵ could improve ethanol-induced liver damage in mice, indicating that Ahr in IECs may become a novel target for the treatment of ALD.

Results

Ahr Deficiency in IECs Aggravates Ethanol-Induced Liver Injury

We first showed that ethanol (EtOH) feeding mice subjected to the chronic-plus-binge model²³ showed decreased messenger RNA (mRNA) levels of Ahr and its downstream target genes (Cyp1a1, Cyp1a2, and Cyp1b1) in IECs (Figure 1A–D). Furthermore, ethanol exposure could directly down-regulate Ahr expression in the murine intestine-derived epithelial cell line (MODE-K) (Figure 1E). To explore the effects of Ahr deficiency on the progression of alcohol-related liver disease (ALD), the Ahr^ΔIEC mice were generated and validated by quantitative polymerase chain reaction (qPCR) and immunohistochemical staining (Figure 1F and G). Although body weight and food intake did not show significant differences between Ahr^fl/fl and Ahr^ΔIEC mice after control or ethanol feeding (Figure 1H and I), Ahr^ΔIEC mice developed more severe liver injury and steatosis relative to Ahr^fl/fl mice after ethanol feeding as shown by the increased plasma alanine aminotransferase (ALT) level, H&E staining of liver sections, Oil red O staining of liver sections, and the higher hepatic triglyceride level (Figure 1J–N). After ethanol administration, livers of Ahr^ΔIEC mice also showed significantly higher mRNA expression of inflammation-related genes, including Il1b, Il6, Cxcl5, and Cxcl10, as compared with control mice (Figure 1O). However, we found that the hepatic mRNA levels of Tnf, Adgre1 (also known as F4/80), and Ly6g in EtOH-fed Ahr^ΔIEC mice were comparable with those in Ahr^fl/fl mice (Figure 1P). Consistently, hepatic myeloperoxidase + neutrophils and tumor necrosis factor α protein expression in Ahr^ΔIEC mice did not differ from Ahr^fl/fl mice after alcohol feeding (Figure 1Q and R). To further clarify the crucial genes contributing to the increased lipid accumulation in ethanol-fed Ahr^ΔIEC mice, we evaluated the hepatic expression of genes involved in free fatty acids (FFAs) and triglyceride (TG) synthesis, elongation, and hydrolysis. Among these, we found that the expression of elongation of very long chain fatty acid protein 7 (Elovl7), which encodes an enzyme that is responsible for the elongation of FFAs, was increased nearly 5-fold in the livers of Ahr^ΔIEC mice than that in Ahr^fl/fl mice after ethanol administration (Figure 1O and P), suggestive of its potential role in promoting lipogenesis in Ahr^ΔIEC mice.

Figure 1.

Ahr deficiency in IECs of mice exacerbates ethanol-induced liver injury. (A–D) WT mice were fed a control or ethanol diet (n = 5–6 per group). (A) Body weight and food intake. (B) Plasma level of ALT. (C) Representative H&E staining images of liver sections. (D) qPCR analysis of Ahr, Cyp1a1, Cyp1a2, and Cyp1b1 in IECs isolated from control (Ctrl) and ethanol-fed mice subjected to the chronic-plus-binge model (n = 6 per group). (E) mRNA expression analysis of Ahr in murine intestine-derived epithelial cell (MODE-K) stimulated with ethanol (n = 3 independent experiments performed in 2 replicates). (F) Schematic of Ahr^ΔIEC mice and the qPCR analysis of Ahr in IECs and lamina propria (LP) cells (n = 5 per group). (G) Representative images of immunohistochemical staining for AHR expression in small intestine of Ahr^fl/fl and Ahr^ΔIEC mice. (H–N) Ahr^fl/fl and Ahr^ΔIEC mice were fed a control or ethanol diet. (H) Body weight and (I) food intake (n = 6 for Ctrl; n = 22–23 for EtOH). (J) Plasma level of ALT (n = 6 for Ctrl; n = 20–21 for EtOH). (K) Representative H&E staining images of liver sections. (L and M) Representative Oil red O–stained liver sections quantified by Image J (n = 5–6 for Ctrl; n = 8–10 for EtOH). (N) Hepatic TG content (n = 6 for Ctrl; n = 16–19 for EtOH). (O) Hepatic expression of mRNA encoded by inflammation, FFAs, and TG synthesis-related genes in ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 12–23). (P) Hepatic expression of mRNA in ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 6 per group). (Q) Myeloperoxidase staining of liver sections from ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 7–9). (R) Hepatic expression of tumor necrosis factor (TNF)α protein (n = 7–9). Scale bar: 50 μm. Data are represented as means ± SEM. ∗P < 0.05, (J and N) Mann–Whitney test; (A, B, D, E, F, H, I, M, O, P and R) unpaired t test.

To examine whether Ahr deficiency in IECs affects the absorption and metabolism of ethanol, we determined the level of ethanol in plasma and metabolism of ethanol in livers. The plasma ethanol level did not differ significantly between Ahr^fl/fl and Ahr^ΔIEC mice after ethanol feeding (Figure 2A), and the hepatic gene expression of Adh1 and Cyp2e1 (the 2 main primary enzymes that convert ethanol to acetaldehyde)^24,25 were comparable between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (Figure 2B and C). These results indicate that Ahr deficiency in IECs promoted ethanol-induced liver disease without affecting intestinal absorption and hepatic metabolism of ethanol in mice.

Figure 2.

Ahr deficiency in IECs causes the progression of ethanol-induced liver damage independent of ethanol absorption and metabolism. (A) Plasma ethanol level (n = 15 for Ahr^fl/fl; n = 14 for Ahr^ΔIEC) and (B) hepatic mRNAs of Cyp2e1 and Adh1 in Ahr^fl/fl and Ahr^ΔIEC mice fed an ethanol diet (n = 21 for Ahr^fl/fl; n = 20 for Ahr^ΔIEC). (C) Western blot of hepatic CYP2E1 and its quantification (n = 3 per group). Data are represented as means ± SEM. ∗P < 0.05, unpaired t test. Ctrl, control.

Ahr Deficiency in IECs Promotes Translocation of H.hepaticus and H.ganmani to Liver in Mice

Increasing evidence has shown that intestinal microbiota dysbiosis has been implicated in the progression of ALD,14, 15, 16 we thus used 16S ribosomal RNA (rRNA) gene sequencing to investigate the effects of IEC-specific Ahr disruption on gut microbiota. As shown by the principal coordinate analysis plot, we showed that the composition of intestinal microbiota in ethanol-fed mice clustered separately from that of pair-fed control mice (Figure 3A), while the overall microbiota composition of Ahr^ΔIEC mice did not differ significantly from Ahr^fl/fl mice either on a control or ethanol diet (Figure 3A). Despite that the total number of intestinal bacteria remained unchanged between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (Figure 3B), the cladogram (linear discriminant analysis effect size [LEfSe] analysis) showed that the abundance of Helicobacter (the family level is Helicobacteraceae, the order level is Campylobacterales) was enriched (linear discriminant analysis score > 2) in ethanol-fed Ahr^ΔIEC mice compared with Ahr^fl/fl mice (Figure 3Ci, Cj, Ck, and D). Helicobacter also was increased slightly (P = 0.076) in control-fed Ahr^ΔIEC mice in comparison with Ahr^fl/fl mice (Figure 3E). Alistipes (the family level is Rikenellaceae) was found to be enriched in the cecal content of EtOH-fed Ahr^ΔIEC mice as evidenced by the 16S rRNA sequencing results (Figure 3C and D), although this increase could not be confirmed using a qPCR assay (Figure 3F). In addition, the 3 main species of Alistipes, including Alistipes finegoldii, Alistipes timonensis, and Alistipes indistinctus, did not alter significantly between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (Figure 3F).

Figure 3.

Ahr deficiency in IECs promotes expansion of cecal H.hepaticus and its translocation to the liver in ethanol-induced mice. (A) Principal coordinate analysis (PCoA) plot of cecal microbiota of Ahr^fl/fl and Ahr^ΔIEC mice fed a control or ethanol diet (n = 6 for control [Ctrl]; n = 8 for EtOH). (B) Total bacteria in cecum of ethanol-fed mice (n = 18 for Ahr^fl/fl; n = 16 for Ahr^ΔIEC). (C) The cladogram of gut microbiome in different taxonomic levels from ethanol-fed mice. The taxa of different abundance in Ahr^fl/fl and Ahr^ΔIEC group are presented in blue and red, respectively (n = 8 per group). (D) Significantly altered bacterial taxa between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (linear discriminant analysis [LDA] score, >2 or <-2). (E) The relative abundance of Helicobacter in cecum from control-fed mice detected by 16S rRNA gene sequencing (n = 6 per group). (F) The relative levels of Alistipes and the 3 main species of Alistipes in cecum from EtOH-fed Ahr^fl/fl and Ahr^ΔIEC mice determined by qPCR (n = 6 for Ahr^fl/fl; n = 5 for Ahr^ΔIEC). (G) The relative level of H.hepaticus and H.ganmani in cecum from control or ethanol-fed mice (n = 5–6 for Ctrl; n = 18–20 for EtOH) determined by qPCR. (H and I) H.hepaticus and H.ganmani in MLNs (n = 4–6 per group) and liver (n = 8–9 per group) of ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice. (J) The relative levels of bacteria in MLNs and livers of EtOH-fed mice detected by qPCR (n = 3 per group for MLN; n = 7–9 for liver). (K) Correlation analysis between relative level of intestinal H.hepaticus with hepatic steatosis (left panel), Il6 (middle panel), and Cxcl5 expression (right panel). (L) Correlation analysis between the relative level of intestinal H.hepaticus with hepatic Il1b (upper) and Cxcl10 (lower) expression. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test.

Furthermore, we also confirmed that the species of H. hepaticus and H. ganmani (2 species of Helicobacter) were increased in the cecum content of ethanol-fed Ahr^ΔIEC mice relative to Ahr^fl/fl mice as determined by qPCR (Figure 3G). Accordingly, ethanol-fed Ahr^ΔIEC mice showed significantly higher levels of H. hepaticus and H. ganmani both in mesenteric lymph nodes (MLNs) and livers compared with the ethanol-fed Ahr^fl/fl mice (Figure 3H and I), indicating that IEC-specific Ahr deficiency might facilitate the translocation of H. hepaticus and H. ganmani from the gut to liver after ethanol exposure. In addition, we also showed that levels of most commonly studied bacteria such as Enterococcus, Bifidobacterium, Clostridium, and Prevotella did not change obviously in livers between EtOH-fed Ahr^fl/fl and Ahr^ΔIEC mice, although Firmicutes (phylum) and Enterococcus (genus) were increased in MLNs of Ahr^ΔIEC mice (Figure 3J).

More importantly, we found that the relative level of H. hepaticus in cecum was correlated positively with hepatic steatosis, Il6, and Cxcl5, but not Il1b and Cxcl10 expression (Figure 3K and L). Notably, the gene expression of gut barrier function–related proteins such as occludin (Ocln), tight junction protein 1 (Tjp1), Tjp2, and mucin 2 (Muc2) were reduced dramatically in the distal small intestine of Ahr^ΔIEC mice relative to Ahr^fl/fl mice after ethanol feeding (Figure 4A). Consistently, lipopolysaccharide levels in plasma also were increased significantly in ethanol-fed Ahr^ΔIEC mice compared with Ahr^fl/fl mice (Figure 4B), suggesting that Ahr deficiency in IECs disrupted intestinal epithelial barrier function. Of note, we failed to detect the higher mRNA levels of stem cell markers (Lrig1 and Lgr5) in the intestine of ethanol-fed Ahr^ΔIEC mice, although previous study showed that Ahr deletion in IECs promoted stem cell proliferation upon injury of infection or chemical insults,¹² the gene expression of Il22 and Il17 in the proximal small intestine did not show the obvious difference either between ethanol-fed Ahr^ΔIEC and Ahr^fl/fl mice (Figure 4C).²⁶ Taken together, Ahr deficiency in IECs enriched intestinal Helicobacter and aggravated their translocation to liver, eventually leading to enhanced alcohol-related liver injury.

Figure 4.

Ahr deficiency in IECs disrupted the intestinal epithelial barrier. (A) mRNA levels of Olcn, Tjp1, Tjp2, and Muc2 in the distal small intestine (n = 5–6 per group). (B) Plasma level of lipopolysaccharide (LPS) in EtOH-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 13 for Ahr^fl/fl; n = 11 for Ahr^ΔIEC). (C) mRNA levels of leucine-rich repeats and immunoglobulin-like domains 1 (Lrig1), leucine-rich repeat containing G-protein–coupled receptor 5 (Lgr5), Il22, and Il17 in the proximal small intestine of ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 5–16), Lrig1 and Lgr5 are the stem cell markers. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test. Ctrl, control.

IEC-Specific Ahr Deficiency Up-regulates IBA Level in the Intestines of Mice

To further decipher the underlying mechanism that IEC-specific Ahr deficiency promotes ethanol-induced liver damage, we also used untargeted metabolomics to assess the alteration of cecum metabolites between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice. The top 20 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways with significant differences are shown in Figure 5A, and pathways such as protein digestion and absorption, aminoacyl-tRNA biosynthesis, and mineral absorption differed extensively between ethanol-fed Ahr^ΔIEC and control mice. Although the tryptophan level was reduced in ethanol-fed Ahr^ΔIEC mice, the level of kynurenic acid, which was converted from kynurenine, a main metabolite of tryptophan, was comparable between Ahr^fl/fl and Ahr^ΔIEC mice (Figure 5B). Indole and its derivatives, microbe-dependent products of tryptophan, did not alter dramatically either in these 2 groups of mice (Figure 5B). Instead, 64 metabolites with significant change were identified, among them, IBA, a short-chain fatty acid (SCFA), was increased notably in ethanol-fed Ahr^ΔIEC mice compared with control mice (Figure 5C–E). The other SCFAs, including acetic acid, propionic acid, and butyric acid, were not different between the ethanol-fed Ahr^ΔIEC mice and the Ahr^fl/fl mice (Figure 5F). More importantly, we found that the increased level of IBA was correlated positively with hepatic steatosis and expression of inflammation-related genes (Figure 5G and H). Consistent with the increase of IBA, shotgun metagenomics analysis showed that the bacterial gene expression of ilvE, bkdA, and pdhD, which were responsible for metabolizing valine to produce IBA, were increased in the cecum of ethanol-fed Ahr^ΔIEC mice relative to control mice (Figure 5I and J). Accordingly, we observed the reduction of valine level in these ethanol-fed Ahr^ΔIEC mice (Figure 5C–E). To explore whether alterations of gut metabolites were associated with enrichment of Helicobacter, we performed a correlation analysis between the 64 altered metabolites and the intestinal abundance of H. hepaticus or H. ganmani. The anionic phospholipid 1-palmitoyl-2-oleoyl-phosphatidylglycerol (POPG), the secondary bile acid lithocholic acid, which was reported to cause liver injury,²⁷ nervonic acid, and the stearamide, which was increased in patients with alcoholic cirrhosis,²⁸ showed significant positive correlations with the abundance of H. hepaticus (Figure 6A and B), while the other 4 metabolites including hypoxanthine, phenylalanine-cysteine, acamprosate, and thymine showed negative correlations with H. hepaticus (Figure 6C). In addition, we found that metabolites such as adynerin, POPG, and adenosine monophosphate were correlated positively with H. ganmani, while linoleic acid was correlated negatively with H. ganmani (Figure 6D and E).

Figure 5.

Ahr deficiency in IECs increases intestinal IBA. (A) The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of intestinal metabolites from ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 6 per group). (B) Tryptophan, kynurenic acid, indole, indoleacetic acid, indoxyl acid, and indolepropionic acid levels in cecum of ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 6 per group). (C) The heatmap of the relative abundance of each significantly altered metabolite in cecum between ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 6 per group), 64 metabolites were identified with significant alteration (variable influence on projection [VIP] > 1.0; P < 0.1). (D) IBA and valine levels in the cecum content of ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 6 per group). (E) Identification of 64 metabolites with a significant increase (upward) or decrease (downward) in ethanol-fed Ahr^ΔIEC mice compared with Ahr^fl/fl mice. (F) Cecal contents of SCFAs in ethanol-fed Ahr^fl/fl and Ahr^ΔIEC mice (n = 7–10). (G) Spearman correlation of the 64 metabolites with hepatic steatosis score, ALT level, and inflammation. (H) Pearson correlation of hepatic steatosis score, Cxcl10 with intestinal IBA level. (I) Scheme for IBA-synthesis pathway. IlvE, branched-chain amino acid aminotransferase (EC:2.6.1.42); BkdA, 2-oxoisovalerate dehydrogenase E1 component (EC:1.2.4.4); and PdhD, dihydrolipoamide dehydrogenase (EC:1.8.1.4). (J) The counts of the genes involved in the IBA-synthesis pathway, including ilvE, bkdA, and pdhD, analyzed by shotgun metagenomics. n = 5 for Ahr^fl/fl; n = 4 for Ahr^ΔIEC. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test.

Figure 6.

H.hepaticus per se could produce POPG as the metabolite that enhances H.hepaticus growth. (A) The correlation analysis between stearamide, lithocholic acid, POPG, nervonic acid, and the relative abundance of intestinal H.hepaticus. (B) Spearman rank correlation between significantly altered metabolites and the intestinal relative abundance of H.hepaticus. (C) The correlation analysis between acamprosate, thymine, hypoxanthine, phenylalanine-cysteine (Phe-Cys) and the relative abundance of intestinal H.hepaticus. (D) Spearman rank correlation between significantly altered metabolites and the intestinal relative abundance of H.ganmani. (E) The correlation analysis between adynerin, POPG, adenosine monophosphate (AMP), linoleic acid, and the relative abundance of intestinal H.ganmani (Pearson correlation). (F) POPG was observed to boost the growth of H.hepaticus in vitro. H.hepaticus was cultured anaerobically in brucella agar plates containing 8% sheep blood with or without the metabolite for 6 days. Bacteria were collected and suspended in medium and their growth were determined by OD₆₀₀ measurement. n = 4 independent experiments performed in 2 replicates in each experiment. (G) The relative level of POPG in control or H.hepaticus by untargeted metabolomics (n = 2). (H) The level of POPG in the cecal content from mice fed with H.hepaticus or broth (n = 5–6). (I) Schematic representation of microinjection of H.hepaticus into the intestinal organoid. (J) POPG enhanced the growth of H.hepaticus in the intestinal organoid isolated from WT mice. H.hepaticus exposed to POPG or control medium was microinjected into intestinal organoid. After incubation for 4 days in the lumen of the organoid, bacteria were released and cultured anaerobically in brucella agar plates for 6 days and determined by OD₆₀₀ analysis. n = 3 independent experiments performed in 2 replicates in each experiment. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test.

Subsequently, we evaluated the effects of the correlated metabolites on the growth of H. hepaticus in vitro. We found that only the lipid POPG could significantly promote the growth of H. hepaticus (Figure 6F). Interestingly, the untargeted metabolomics analysis showed that H. hepaticus per se could produce POPG as the metabolite (Figure 6G), although we cannot exclude that POPG also possibly was derived from the host. Consistently, the intestinal POPG level in mice fed with H. hepaticus also was higher than that in control mice (Figure 6H). More importantly, we also showed that POPG could enhance H. hepaticus growth in intestinal organoids (Figure 6I and J and Supplementary Video). Therefore, we provided mechanistic evidence that POPG might be the most important metabolite to promote the enrichment of H. hepaticus in the intestine.

H.hepaticus Aggravates Ethanol-Induced Injury In Vivo and In Vitro

To further prove that the IEC-specific Ahr deficiency–induced increase of H. hepaticus promotes the development of ALD, we challenged wild-type (WT) mice with H. hepaticus every other day in a chronic-plus-binge model. H. hepaticus administration markedly increased levels of H. hepaticus (>500-fold) in feces of mice (Figure 7A). In line with this, mice gavaged with H. hepaticus developed more severe ethanol-induced hepatic injury, indicated by increased ALT level (Figure 7B), increased hepatic steatosis (Figure 7C), as well as enhancement of hepatic Cxcl5 and Cxcl10 expression (Figure 7D). Consistently, both cultured hepatocytes and Kupffer cells stimulated with H. hepaticus in vitro recapitulated the effects of H. hepaticus in vivo. H. hepaticus–treated mouse hepatocyte alpha mouse liver 12 (AML12) cells showed more lipid droplet accumulation (Figure 7E and F) and increased expression of inflammation-related genes such as Il1b, Il6, Cxcl5, and Cxcl10 in either basal or ethanol-stimulated conditions (Figure 7G). Notably, H. hepaticus–induced increment of Cxcl10 expression became more obvious in the presence of ethanol (Figure 7G). In addition, H. hepaticus also could induce the increase of Il1b, Il6, Cxcl5, and Cxcl10 gene expression in Kupffer cells (Figure 7H), which is comparable with that in hepatocytes. These results provided direct evidence that H. hepaticus could aggravate liver injury in vivo and in vitro.

Figure 7.

H.hepaticus promotes the progression of ALD in vivo and in vitro. (A–D) Mice were gavaged with medium or H.hepaticus every other day during ethanol feeding. (A) Relative level of H.hepaticus in the feces of mice administered medium or H.hepaticus (n = 6 per group). (B) Plasma ALT level (n = 9 per group). (C) Representative H&E staining of liver sections (n = 9 per group). (D) Il1b, Il6, Cxcl5, and Cxcl10 mRNA levels in livers (n = 9 per group). (E–G) AML12 cells were cultured in plates overnight, and subsequently were stimulated with medium or H.hepaticus and exposed to control (Ctrl) or EtOH simultaneously for 24 hours. n = 3–4 independent experiments performed in 2 replicates in each experiment. (E) Representative images of Oil red O staining, (F) quantified by ImageJ. (G) Il1b, Il6, Cxcl5, and Cxcl10 mRNA levels in AML12 cells. (H) Il1b, Il6, Cxcl5, and Cxcl10 mRNA levels in Kupffer cells stimulated with H.hepaticus for 4 hours. n = 4 independent experiments performed in 2 replicates in each experiment. Data are represented as means ± SEM. ∗P < 0.05, (A) Mann–Whitney test; (B, D, F, G and H) unpaired t test.

IBA Induces Liver Injury in Mice and Cultured Hepatocytes

IBA level was increased in ethanol-fed Ahr^ΔIEC mice compared with Ahr^fl/fl mice (Figure 5C), we thus investigated whether the increased IBA also could contribute to the progression of ALD. Administration of IBA in C57BL/6 mice showed obvious liver damage as indicated by the increased plasma ALT level, hepatic lipid accumulation, and higher hepatic TG level compared with control mice (Figure 8A–C). We also noticed that IBA could significantly increase hepatic Cxcl5 gene expression (Figure 8D). Consistent with this, AML12 cells stimulated with IBA showed more lipid accumulation and increased gene expression of hepatic Cxcl5 and Cxcl10 (Figure 8E–G). Similarly, IBA also could increase Cxcl5 and Cxcl10 expression in Kupffer cells (Figure 8H). More interestingly, we found that H. hepaticus could boost the effects of IBA on lipid accumulation (Figure 8I and J). It should be noted that IBA treatment alone or in combination with ethanol significantly up-regulated Elovl7 expression in AML12 cells (Figure 8K), which is consistent with the increased Elovl7 expression in ethanol-fed Ahr^ΔIEC mice, which had the higher level of IBA in vivo (Figures 1O and 5C–F). We thus sought to determine the effect of Elovl7 on lipid accumulation in hepatocytes. Overexpression of Elovl7 resulted in more lipid droplet accumulation in AML12 cells, either in basal or alcohol-stimulated conditions (Figure 8L and M). All of these results showed that IBA could induce hepatic steatosis and liver damage in vivo and in vitro, while H. hepaticus was able to enhance the effects of IBA on lipogenesis. We also elucidated the crucial role of Elovl7 in IBA-induced lipid accumulation.

Figure 8.

IBA induces liver damage in vivo and in vitro. (A–D) Mice were injected intraperitoneally with vehicle or IBA for 24 or 48 hours. (A) Plasma ALT level at 24 hours (n = 8 per group). (B) Representative H&E staining of liver sections at 48 hours. (C) Hepatic TG content (n = 7–8 per group) at 48 hours. (D) Hepatic Cxcl5 mRNA at 24 hours (n = 8 per group). (E) AML12 cells were seeded in 12-well plates overnight, and were treated with control (PBS) or ethanol (100 mmol/L), and stimulated with vehicle (DMSO) or IBA (1 and 4 mmol/L) at the same time for 24 hours, and the cells were stained with Oil red O. (F) Quantification of the Oil red O staining. (G and H) mRNA levels of Cxcl5 and Cxcl10 in AML12 cells (G) treated with vehicle or IBA for 24 hours and isolated Kupffer cells, or (H) treated with vehicle or IBA for 4 hours. n = 3 independent experiments performed in 2 replicates in each experiment. (I and J) AML12 cells were seeded in 12-well plates overnight and treated with medium, IBA, H.hepaticus, and IBA+H.hepaticus for 24 hours. The cells were finally stained with Oil red O. (I) Representative images of the Oil red O staining. (J) Quantification of Oil red O staining. (K) Elovl7 mRNA levels in AML12 cells exposed to EtOH (100 mmol/L), IBA (4 mmol/L), or EtOH (100 mmol/L) + IBA (4 mmol/L). (L) Cells were transfected with empty vector or Elovl7, and exposed to control or EtOH (100 mmol/L) for 24 hours. Cells were stained with Oil red O. (M) Quantification of the Oil red O staining. For cell culture experiments, at least 3 independent experiments were performed with 2 replicates in each experiment. Scale bar: 50 μm. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test. Ctrl, control.

AHR Agonists Ameliorate ALD in Mice

Two widely used agonists for AHR, I3C and FICZ,^12,29 whose activation effects on Ahr were confirmed in vitro by ourselves (Figure 9A and B), were selected to explore whether supplementation with AHR agonists could be a therapeutic approach for ALD. WT mice exposed to an ethanol diet were orally gavaged with I3C (50 mg/kg) and FICZ (50 μg/kg) daily for 15 days. Although body weight and food intake showed no significant differences compared with control mice (Figure 9C), both I3C and FICZ decreased hepatic steatosis (Figure 9D), decreased liver TG levels (Figure 9E), and reduced mRNA expression of Elovl7, Il6, Cxcl5, and Cxcl10 (Figure 9F and G). Subsequently, we confirmed that I3C and FICZ activated AHR in the intestine as shown by the increased gene expression of Cyp1a1 and Cyp1b1 (Figure 9H), whereas they did not activate hepatic AHR (Figure 9I), suggesting that selectively activating intestinal AHR was sufficient to protect mice from alcohol-induced liver damage. Intriguingly, mice treated with I3C and FICZ dramatically decreased the abundance of intestinal H. hepaticus (Figure 9J). Collectively, our findings show the therapeutic potentials of AHR agonists for ALD treatment.

Figure 9.

AHR agonists improve ethanol-induced liver disease in mice. (A) Ahr activation by I3C (50 μmol/L) and FICZ (100 nmol/L) analyzed by luciferase assay (n = 4 independent experiments). (B) Ahr, Cyp1a1, Cyp1a2, and Cyp1b1 mRNA levels in AML12 stimulated with I3C (50 μmol/L) and FICZ (100 nmol/L) (n = 4 independent experiments). (C–J) Mice were orally administered vehicle, I3C (50 mg/kg), and FICZ (50 μg/kg), and fed an EtOH diet (n = 6–8 per group). (C) Body weight and food intake. (D) Representative liver sections stained with H&E. (E) Hepatic TG content. (F) Hepatic Elovl7 mRNA level. (G) Hepatic Il6, Cxcl5, and Cxcl10 mRNA levels. (H and I) mRNA levels of Ahr, Cyp1a1, and Cyp1b1 in distal small intestine and liver. (J) The relative level of H.hepaticus in cecum. Scale bar: 50 μm. Data are represented as means ± SEM. ∗P < 0.05, unpaired t test. Ctrl, control.

Alcoholic Patients Showed Decreased Intestinal AHR Expression and Increased H.hepaticus Level

To explore whether the decreased intestinal AHR expression was relevant to alcoholic patients, we analyzed their levels in the duodenal tissues. As expected, the mRNA and protein expression of intestinal AHR was decreased significantly in alcoholic patients compared with nonalcoholic individuals by qPCR analysis and immunohistochemical staining (Figure 10A and B). Intriguingly, compared with healthy controls, patients with alcoholic liver disease showed a higher level of fecal H. hepaticus (Figure 10C). Notably, 1 patient with an alcohol-use history for more than 30 years (2–3 bottles of liquor for daily drinking, equal to 400–600 g pure alcohol/d) and very severe aspartate aminotransferase, γ-glutamyl transferase, and Fibrosis-4 index levels, had a remarkably higher abundance of H. hepaticus in the stool sample (Figure 10C).

Figure 10.

Decreased intestinal AHR level and increased H.hepaticus in alcoholic patients. (A) mRNA and (B) protein levels of AHR in duodenal tissues from controls and alcoholic patients (n = 12 controls; n = 6 alcoholic patients) determined by qPCR and immunohistochemical staining. (C) The relative level of H.hepaticus in healthy controls and patients with alcoholic liver disease (n = 13 per group). (D) A schematic diagram summarizing our findings that alcohol abuse down-regulates AHR expression in IECs, which contributes to ALD progression. Alcohol depresses the expression of Ahr in IECs, and subsequently enriches intestinal H.hepaticus, H.ganmani, and IBA levels. Ahr deficiency in IECs exacerbates intestinal H.hepaticus and H.ganmani translocation through MLNs to liver via the disrupted gut barrier. POPG, 1 metabolite of H.hepaticus, in turn boosts H.hepaticus growth. H.hepaticus and IBA augment ethanol-induced liver injury by eliciting hepatic inflammation and steatosis. Supplementation with AHR agonists, FICZ and indole-3-carbinol (abundant in cruciferous vegetables), protects mice from ALD development by activating intestinal Ahr. Data are represented as means ± SEM. ∗P < 0.05, Mann–Whitney test.

Discussion

Ahr is widely expressed throughout the body, and intestinal Ahr was reported to play an important role in enteric diseases such as inflammatory bowel disease and colitis by regulation of the immune response and intestinal barrier functions.^6,8,12 However, the association between intestinal Ahr and liver diseases still remains largely unexplored. The AHR ligands constitute a large family that generally can be categorized into 4 major sources: xenobiotics (eg, dioxin), dietary metabolites (eg, I3C), endogenous metabolites (eg, indole acetic acid), and microbial derivatives (eg, indirubin).^30,31 Although ethanol is not identified as a direct ligand of AHR, ethanol feeding was shown to reduce microbiota-dependent AHR ligand production from tryptophan in mice such as indole acetic acid.²⁶ Notably, AHR activation by microbial tryptophan metabolites were shown to improve ALD in mice,³² highlighting the association of ALD with intestinal AHR and microbiota. We thus attempt to explore the crosstalk between intestinal AHR and gut microbiota, as well as their metabolites, and clarify the role of intestinal AHR in the progression of ALD.

Considering that IECs function as the first line sensing intestinal environment change derived from dietary, microbial, and metabolic cues, we generated IEC-specific Ahr-deficiency mice in which Ahr expression in lamina propria cells remained unchanged. We showed that Ahr depletion in IECs did not directly affect intestinal absorption and metabolism of ethanol. In addition, we could not find that a lack of Ahr in IECs altered the intestinal mRNA levels of stem cell markers, although 1 previous study showed that Ahr deletion in IECs enhanced stem cell proliferation upon injury through infection or chemical insults.¹² The gene expression of Il22 and Il17 in the proximal small intestine was not different between ethanol-fed Ahr^ΔIEC mice and control mice.

Here, we observed that alcohol depressed the expression of Ahr in IECs in mice and human beings, and its deficiency enhanced the susceptibility to ethanol-induced liver injury. Compared with control mice, intestinal levels of Helicobacter (H. hepaticus and H. ganmani) and IBA were up-regulated in ethanol-fed Ahr^ΔIEC mice, which were accompanied by increased intestinal permeability, and the abundance of H. hepaticus and H. ganmani in MLNs and liver also were increased accordingly in these mice. Of note, H. hepaticus per se could produce POPG as the metabolite that, in turn, enhanced the bacteria growth and promoted its translocation from gut to liver on the basis of the intestinal barrier disruption caused by Ahr deficiency. H. hepaticus and IBA aggravated ethanol-induced liver injury by eliciting hepatic inflammation and steatosis. Importantly, oral supplementation of AHR agonists FICZ and I3C (a derivative from cruciferous vegetables) markedly ameliorated ALD by activating intestinal Ahr and reducing H. hepaticus in a mouse model, suggesting a new way for the treatment of ALD (Figure 10D). Consistently, a recent study showed that the prebiotic, pectin, could improve alcohol-induced liver injury by increasing the bacterial tryptophan metabolites, and proposed that targeting intestinal AHR activation could improve alcoholic liver disease,³² however, our study, through using IEC-specific Ahr-deficiency mice, showed elaborately that it is the IEC-intrinsic Ahr that mediated the beneficial effects of AHR activation on the progression of ALD.

In this study, we showed that a specific dysbiosis, increased intestinal Helicobacter species, contributed to the progression of ALD in mice. Previously, Helicobacter species were not regarded as pathogens for human beings because the majority of colonized individuals were asymptomatic.³³ However, increasing evidence has shown a strong association of intestinal Helicobacter species with enteric diseases such as Crohn’s disease and inflammatory bowel disease in patients, indicating its pathogenicity in the digestive diseases.^34,35 H. hepaticus, one of the main species of Helicobacter, also could be detected in patients’ bile and liver samples, and was associated with chronic liver diseases, including primary hepatocellular carcinoma and bile duct cancer.36, 37, 38, 39 Here, we identified that the H. hepaticus level was increased markedly in patients with ALD compared with that in healthy individuals. In addition, H. ganmani, another important species of Helicobacter with increased abundance in ethanol-induced Ahr^ΔIEC mice, was not explored in this study because of its current unavailability, however, the role of H. ganmani in ALD also deserves further investigation.

Besides the changes of gut microbiota, another explanation for the link between gut dysbiosis and liver diseases is the alteration of metabolites produced by the intestinal microbiota. Well-known metabolites such as SCFAs (eg, butyrate, acetate) could exert immunomodulatory effects both inside and outside of the intestine.^40,41 Here, we showed that the SCFA IBA was increased significantly in IEC-specific Ahr-deficiency mice compared with control mice after alcohol drinking. Shotgun metagenomics analysis indicated that the pathway contributing to IBA synthesis was more activated in Ahr^ΔIEC mice than that in Ahr^fl/fl mice as evidenced by the increased bacterial gene expression of ilvE, bkdA, and pdhD. Notably, the bacterial gene ilvE, coding the enzyme responsible for catalyzing valine to 3-methyl-2-oxobutanoate, also could be found in H. hepaticus. Thus, we presumed that the increased abundance of H. hepaticus in Ahr^ΔIEC mice also may contribute to the increased IBA. Regarding the role of IBA in liver disease, we showed that both IBA administration in vivo and treatment in cultured hepatocytes could induce liver injury when used alone or in combination with alcohol. Although we failed to detect IBA in mice serum, it still is possible that IBA could go to the liver from the portal vein to aggravate the alcohol-induced liver disease in Ahr^ΔIEC mice given that these mice had a disrupted intestinal barrier. Our findings also are consistent with findings in nonalcoholic steatohepatitis patients containing a higher fecal IBA level.⁴² These results therefore suggest that fecal IBA might be developed as a new biomarker to predict the progression of ALD.

Mechanistically, we used the combination of 16S rRNA sequencing, metagenomics, and untargeted metabolomics to show that IEC-specific Ahr deficiency induced the alteration of intestinal metabolites, these metabolites, such as the anionic phospholipid POPG and the secondary bile acids lithocholic acid, nervonic acid, and stearamide (increased in patients with alcoholic cirrhosis),²⁸ correlated positively with the abundance of H. hepaticus, while metabolites such as hypoxanthine, phenylalanine–cysteine, acamprosate, and thymine correlated negatively with the abundance of H. hepaticus. We further showed that the lipid POPG, which was produced by H. hepaticus, could in turn promote the growth of the bacteria itself. In addition, acamprosate, a drug for alcohol use disorder treatment,^43,44 was identified as decreased in cecal content from EtOH-fed Ahr^ΔIEC mice by untargeted metabolomics. It showed a negative correlation with H. hepaticus, although validation of this metabolite remains to be addressed by targeted metabolomics analysis. All of these changes in metabolite composition might directly promote the overgrowth of intestinal H. hepaticus, facilitating its translocation to the liver. At the gene level, we showed that the increased bacterial gene expression of ilvE, bkdA, and pdhD contributed to the increase of IBA in ethanol-fed Ahr^ΔIEC mice relative to control mice, suggesting that these genes might be good targets for regulating the intestinal IBA level. More interestingly, we showed that IBA and H. hepaticus could have a synergistic effect on lipogenesis in hepatocytes. Overall, we elucidated the balance of interactions between IEC-intrinsic Ahr, gut metabolites, and microbiota, the disruption of this balance came from IEC-specific Ahr deficiency, which would lead to the alteration of gut metabolites and microbiota as well as the development of ALD in this study.

Elovl7, one enzyme responsible for the elongation of saturated FFAs, was reported to trigger lipid accumulation in differentiated adipocytes, leading to oxidative damage and inflammation.^45,46 In our study, compared with Ahr^fl/fl mice, we found an up-regulated gene expression of hepatic Elovl7 in ethanol-induced Ahr^ΔIEC mice, which had more severe steatotic livers. This was consistent with the in vitro results that overexpression of Elovl7 resulted in lipid accumulation in cultured hepatocytes. Mechanistically, we identified that IBA, which was increased in Ahr^ΔIEC mice relative to control mice, could directly up-regulate Elovl7 mRNA expression in hepatocytes. Our observations thus proved that Elovl7 might play an important role in regulating hepatic lipid accumulation in the context of alcohol-related liver disease.

In conclusion, we showed an essential role for intestinal epithelial cell intrinsic Ahr in regulating hepatic lipid accumulation and inflammation in ALD through affecting H. hepaticus and IBA levels, which might serve as predictive biomarkers for the progression of ALD. Moreover, dietary supplementation with AHR agonists provided a new therapeutic strategy for treating ALD.

Methods

Animals

Ahr^fl/fl (stock no. 006203) and Villin-Cre (stock no. 021504) mice on a C57BL/6 background were purchased from Jackson Laboratory (Bar Harbor, ME) and bred in-house at China Pharmaceutical University. For IEC-specific Ahr disruption, Ahr^fl/fl were crossed with Villin-Cre mice to obtain Ahr^ΔIEC mice. WT mice on a C57BL/6 background were obtained from Beijing Vital River Laboratory Animal Technology Company (Beijing, China). All mice were housed in the specific pathogen-free facility and maintained under a temperature-controlled (22°C–23°C) room with a 12:12-hour light/dark cycle. All animal procedures were performed in compliance with the guidelines of the Institutional Animal Care and Use Committee of China Pharmaceutical University (Nanjing, China).

Animal Models

Age-matched female Ahr^fl/fl and Ahr^ΔIEC littermates (age, 8–10 wk; weight, 20–22 g) were subjected to the chronic-plus-binge model.²³ In this model, 8-week-old mice (female) received a Lieber-DeCarli control liquid diet (F1259SP; Bio-Serv, Flemington, NJ) for 5 days and an ethanol liquid diet (F1258SP; Bio-Serv) for 10 days, and then were administered with a single binge of 5 g/kg ethanol (459844; Sigma-Aldrich, St. Louis, MO) on day 16. Pair-fed control mice (Ahr^fl/fl and Ahr^ΔIEC littermates) received an isocaloric substitution of dextrose diet. For AHR agonist treatment, WT female mice were gavaged daily with a volume of 100 μL vehicle (dimethyl sulfoxide [DMSO] suspended in corn oil), 50 mg/kg I3C (105220; Aladdin, Shanghai, China), or 50 μg/kg FICZ (synthetized by Professor Yinan Zhang, Nanjing University of Chinese Medicine, Nanjing, China), starting on day 1 of the binge ethanol feeding model. I3C and FICZ were dissolved in a small volume of DMSO initially, and adjusted with corn oil to prepare the final concentrations before use.^12,29,47 When the mice were killed, plasma and appropriate tissues including liver, MLNs, intestine, and cecum were harvested. For IBA (I1754; Sigma-Aldrich) injection, WT female mice (age, 10 wk) were injected intraperitoneally with vehicle (phosphate-buffered saline [PBS]) or IBA (2.5% IBA diluted in PBS) at a dose of 2.5 mL/kg. Blood and liver tissues were collected at 24 or 48 hours after injection.

Human Samples

Alcoholic patients were enrolled in this study according to the pathologic examination results and inclusion criteria as described previously,^21,48 and written informed consent was signed by each patient and control. All of these participates did not take any antibiotics during the 2 weeks preceding the enrolment.

For H. hepaticus abundance in feces, patients with ALD (n = 13; 12 males/1 female; mean age, 52.8 y) were enrolled in this study. Thirteen healthy volunteers (social drinkers consuming <20 g/d; 12 males/1 female; mean age, 49.07 y) were recruited as controls. Baseline features of these subjects are shown in Table 1. Fecal samples were collected, frozen immediately, and stored at −80°C. The protocol was approved by the Ethics Committee of the Beijing Ditan Hospital, Capital Medical University (Beijing, China).

Table 1.

Demographic and Clinical Parameters of Controls and Patients With Alcoholic Liver Disease

Variables	Controls (n = 13)	Alcoholic liver disease (n = 13)
Sex, male, n (%), n = 26	12 (92.3)	12 (92.3)
Age, y, n = 26	49.07 (33–66)	52.8 (36–78)
ALT level, U/L, n = 23	19.6 (11.9–37.4)	30.5 (6.6–83.0)
AST level, U/L, n = 23	21.5 (13.0–28.0)	61.2 (19.6–237.0)
GGT level, U/L, n = 11		259.3 (18.3–932.1)
Albumin level, g/dL, n = 13		32.5 (24.5–41.5)
Bilirubin level, mg/dL, n = 13		40.4 (13.1–92.7)
Creatinine level, mg/dL, n = 13		65.9 (34.1–123.7)
INR, n = 11		1.3 (0.83–1.79)
Platelet count, 109/L, n = 13		115.6 (24.5–402.0)
MELD, n = 11		12.8 (7.0–18.0)
FIB-4, n = 11		7.2 (0.6–21.2)
FIB-4 > 3.25, n (%)		6 (46.2)
Prior length of alcohol abuse, n = 11
10–20 y		2 (18.2%)
20–30 y		4 (36.4%)
>30 y		5 (45.4%)

ALT, alanine aminotransferase; AST, aspartate aminotransferase; FIB-4, fibrosis-4 index; GGT, γ-glutamyl transferase; INR, international normalized ratio; MELD, model for end-stage liver disease.

For intestinal AHR level analysis, duodenal tissues were collected with endoscopically normal duodenum from alcoholic patients (n = 6) and individuals without alcohol consumption (controls, n = 12). Patient characteristics are summarized in Table 2. This study protocol was approved by the Ethics Committee of the Sir Run Run Shaw Hospital, Nanjing Medical University (Nanjing, China).

Table 2.

Demographic and Clinical Parameters of Controls and Alcoholic Patients

Variables	Controls (n = 12)	Alcoholic patients (n = 6)
Sex, male, n (%)	6 (50.0)	6 (100.0)
Age, y	36.7 (21–59)	51.2 (37–57)
ALT level, U/L	17.7 (8.0–25.0)	22.3 (15.0–29.0)
AST level, U/L	19.3 (14.0–29.0)	17 (15.0–20.0)
GGT level, U/L	26.1 (11.0–56.0)	30.7 (18.0–44.0)
Albumin level, g/dL	44.9 (41.2–48.8)	45.4 (43.3–51.2)
Bilirubin level, μmol/L	9.4 (5.8–13.2)	8.8 (5.3–14.3)
Creatinine level, μmol/L	69.3 (58.0–79.0)	62.2 (56.0–82.0)
INR	0.97 (0.89–1.10)	0.99 (0.89–1.04)
Platelet count, 109/L	225.7 (148.0–297.0)	215.7 (173.0–335.0)
FIB-4	0.71 (0.36–1.37)	1.19 (0.52–1.71)
Prior length of alcohol abuse, n = 6
10–20 y		2 (16.7%)
20–30 y		1 (33.3%)
>30 y		3 (50.0%)

ALT, alanine aminotransferase; AST, aspartate aminotransferase; FIB-4, fibrosis-4 index; GGT, γ-glutamyl transferase; INR, international normalized ratio.

Bacterial Cultures

H.hepaticus (ATCC51449) was a gift from Professor Quan Zhang’s laboratory (Institute of Comparative Medicine, College of Veterinary Medicine, Yangzhou University, Yangzhou, Jiangsu, China).⁴⁹ The strain was cultured anaerobically on brucella agar (8241972; BD Difco, Detroit, MI) plates with 8% sheep blood (TX0030; Solarbio, Beijing, China) for 6 days. Bacteria collected and suspended in PBS (adjusted to optical density analysis at 600 nm [OD₆₀₀] = 1.0) were used for untargeted metabolomics.

To determine the effect of metabolites for H. hepaticus growth in vitro, H. hepaticus was cultured on brucella agar plates containing 8% sheep blood with or without the metabolite. After 6 days of incubation, bacteria were collected and suspended in medium and growth was determined by OD₆₀₀ measurement.

For H. hepaticus administration, WT female mice (age, 8 wk) subjected to the chronic-plus-binge model were orally gavaged every other day during the 15 days with either H. hepaticus (suspended in 200 μL broth, adjusted to OD₆₀₀ = 1.0) or medium, respectively, as described.⁴⁹ At the end of the experiment, mice were killed and appropriate tissues were harvested.

Biochemical Assays

Blood was harvested from the inferior caval vein of mice to tubes containing anticoagulant (0.5 mol/L EDTA-Na₂), and centrifuged for 5 minutes (11,292 × g, 4°C) to obtain plasma. The levels of ALT, aspartate aminotransferase in plasma, and triglycerides in liver were determined using commercial kits (C009-2-1, C010-2-1, and A110-1-1, respectively; Nanjing Jiancheng Bioengineering Institute, Nanjing, Jiangsu, China).⁵⁰

RNA Extraction and Real-Time qPCR

The total RNA from livers of mice was isolated using RNAiso Plus (9109; TaKaRa, Dalian, Liaoning, China) according to the methods described previously.⁵⁰ The complementary DNA was synthesized from 1 μg RNA by using a High Capacity complementary DNA reverse-transcription kit (R312-02; Vazyme, Nanjing, Jiangsu, China). qPCR was performed as described in our previous work^19,50 using SYBR Premix (Q331-02; Vazyme) according to the manufacturer’s instructions on the ABI StepOnePlus real-time PCR machine (Applied Biosystems, Foster City, CA). All qPCR primers were synthesized by Sangon Biotech (Shanghai, China) and are shown in Table 3. The relative gene expression was calculated by the 2^-ΔΔCt method.

Table 3.

Primers Used for Real-Time Reverse-Transcription PCR

Name	Sequence, 5’-3’
Primers for mouse
Acaca-f	GGACAGACTGATCGCAGAGAAAG
Acaca-r	TGGAGAGCCCCACACACA
Acly-f	AGCAGACATAGTCAAAGTCCAG
Acly-r	AAGATTCAGTCCCAAGTCCAAG
Acox1-f	TCGCAGACCCTGAAGAAATC
Acox1-r	CCTGATTCAGCAAGGTAGGG
Adgre1-f	CATAAGCTGGGCAAGTGGTA
Adgre1-r	GGATGTACAGATGGGGGATG
Adh1-f	AGGCATTGTTGAGAGCGTTG
Adh1-r	CGAGGCATTAGCAGATCGC
Ahr-f	CTCCTTCTTGCAAATCCTGC
Ahr-r	GGCCAAGAGCTTCTTTGATG
Cd36-f	GCGACATGATTAATGGCACA
Cd36-r	CCTGCAAATGTCAGAGGAAA
Cpt1a-f	AGTGGCCTCACAGACTCCAG
Cpt1a-r	GCCCATGTTGTACAGCTTCC
Cxcl5-f	TGATCCCTGCAGGTCCACA
Cxcl5-r	CTGCGAGTGCATTCCGCTTA
Cxcl10-f	CCAAGTGCTGCCGTCATTTTC
Cxcl10-r	GGCTCGCAGGGATGATTTCAA
Cyp1a1-f	ACCCTTACAAGTATTTGGTCGT
Cyp1a1-r	GTCATCATGGTCATAACGTTGG
Cyp1a2-f	ATCCTGGAGATCTACCGATACA
Cyp1a2-r	TATGTAGATACAGCGCTCCTTG
Cyp1b1-f	CCAAGAATACGGTCGTTTTTGT
Cyp1b1-r	GTTAGCCTTGAAATTGCACTGA
Cyp2e1-f	CACCCTCCTCCTCGTATC
Cyp2e1-r	CGCTTTGCCAACTTGGTT
Dgat2-f	CGCAGCGAAAACAAGAATAA
Dgat2-r	GAAGATGTCTTGGAGGGCTG
Elovl1-f	CCATACATCCAGATGAGGTGAA
Elovl1-r	AGCACATGACAGCCATTCAG
Elovl5-f	CCTTGAAATAGGTACTAAGTGATGC
Elovl5-r	CTCCTTCTACATCCGCCTCT
Elovl7-f	GCAATCCTCCATGAAAAAGAACT
Elovl7-r	CCAGCCTACCAGAAGTATTTGTG
Fasn-f	CTCGCTTGTCGTCTGCCT
Fasn-r	ATGTCCACACCACCAATGAG
Il1b-f	TGTGAAATGCCACCTTTTGA
Il1b-r	GGTCAAAGGTTTGGAAGCAG
Il6-f	TAGTCCTTCCTACCCCAATTTCC
Il6-r	TTGGTCCTTAGCCACTCCTTC
Il17-f	CTGTCTCTCCCGCTACTG
Il17-r	CTAGCAGCTTCCTCTGGAA
Il22-f	AGGCATTGTTGAGAGCGTTG
Il22-r	CGAGGCATTAGCAGATCGC
Lgr5-f	GGACCAGATGCGATACCGC
Lgr5-r	CAGAGGCGATGTAGGAGACTG
Lipe-f	CCTGCAAGAGTATGTCACGC
Lipe-r	GGAGAGAGTCTGCAGGAACG
Lpl-f	CTGGTGGTCCTGGGAGTTT
Lpl-r	TCCTCAGCTGTGTCTTCAGG
Lrig1-f	TTGAGGACTTGACGAATCTGC
Lrig1-r	CTTGTTGTGCTGCAAAAAGAGAG
Ly6g-f	CATCCTTCTTGTGGTCCTACTGTGTG
Ly6g-r	TCTATCTCCAGAGCAACGCAAAATCC
Muc2-f	CCTTGCAGTCAAACTCAAAGT
Muc2-r	AAGTTTGCCCCTGGCTATGAC
Ocln-f	ATTTATGATGAACAGCCCCC
Ocln-r	CATAGTCAGATGGGGGTGGA
Srebf1-f	ACAGCGGTTTTGAACGACAT
Srebf1-r	GCTCTCAGGAGAGTTGGCAC
Tjp1-f	TGCAATTCCAAATCCAAACC
Tjp1-r	AGAGACAAGATGTCCGCCAG
Tjp2-f	TTTTTGAGCTTGTTGGCTTG
Tjp2-r	GTGATTTTCTTCAACCCGGA
Tnf-f	AGATGATCTGACTGCCTGGG
Tnf-r	CTGCTGCACTTTGGAGTGAT
18S-f	AGTCCCTGCCCTTTGTACACA
18S-r	CGATCCGAGGGCCTCACTA
Primers for bacteria
Alistipes-f	ACATAGGGGGACTGAGAGGT
Alistipes-r	GCATGGCTGGTTCAGACTTG
A finegoldii-f	GTGAGGTAACGGCTCACCAA
A finegoldii-r	GCTCCTACACGTAGAAGCGT
A Indistinctus-f	GTGAGGTAACGGCTCACCAA
A Indistinctus-r	CGATACTTTCAAACAGGTACACGT
A timonensis-f	GTGAGGTAACGGCTCACCAA
A timonensis-r	CGTACTATACTTTCAGTCAGATACACG
Bacteroidetes-f	GGCGACCGGCGCACGGG
Bacteroidetes-r	GRCCTTCCTCTCAGAACCC
Bifidobacterium-f	TCGCGTCYGGTGTGAAAG
Bifidobacterium-r	CCACATCCAGCWTCCAC
Clostridium-f	ACGCTACTTGAGGAGGA
Clostridium-r	GAGCCGTAGCCTTTCACT
Enterococcus-f	AACCTACCCATCAGAGGG
Enterococcus-r	GACGTTCAGTTACTAACG
Firmicutes-f	GGAGYATGTGGTTTAATTCGA
Firmicutes-r	AGCTGACGACAACCATGCAC
H. hepaticus-f	TCGTGTCGTGAGATGTTGGG
H. hepaticus-r	TCCACCTCGCGATATTGCTC
H. ganmani-f	TGTGGAGCTTGTCTCTGCAG
H. ganmani-r	CCCAACATCTCACGACACGA
Prevotella-f	CACRGTAAACGATGGATGCC
Prevotella-r	GGTCGGGTTGCAGACC
16S-f	GTGSTGCAYGGYTGTCGTCA
16S-r	ACGTCRTCCMCACCTTCCTC

f, forward; r, reverse.

Bacterial DNA Isolation and qPCR for 16S, H.hepaticus, and H.ganmani

Bacterial genomic DNA was extracted from mice cecum content, livers, and MLNs as previously described.^19,50 The bacterial abundance was quantified by qPCR, and the value of the 16S rRNA gene was normalized to cecum weight, and the relative abundance of H. hepaticus and H. ganmani in cecum, livers, and MLNs were normalized to 16S.

16S rRNA Gene Sequencing and Analysis

Cecal DNA samples from each group were selected randomly for 16S rRNA gene sequencing and analysis.¹⁹ The V4 region of the 16S rRNA gene was amplified and sequenced using the Illumina NovaSeq platform (Illumina, San Diego, CA) in Novogene Technology Co, Ltd (Beijing, China). Raw sequence data were analyzed with a pipeline⁵¹ based on USEARCH⁵² v10.0.240 (http://www.drive5.com/usearch) and VSEARCH⁵³ v2.13.6 (https://github.com/torognes/vsearch). Briefly, after a combination of sequences and removal of barcodes, UNOISE3⁵⁴ was performed to generate amplicon sequence variants. After sequence alignment by SILVA database⁵⁵ (Silva_123, http://www.drive5.com/usearch/manual/sintax_downloads.html), β-diversity was calculated with USEARCH. The principal coordinate analysis plot was generated based on Bray-Curtis distance. Then LEfSe⁵⁶ (http://huttenhower.sph.harvard.edu/galaxy) and the R package edgeR⁵⁷ were obtained for detecting differential taxons. All plots were drawn by R version 3.6.1 except those of LEfSe. The raw data reported here have been deposited in the National Center for Biotechnology Information Sequence Read Archive database (accession no. PRJNA663684).

Untargeted Metabolomics

Mice cecal contents were analyzed for untargeted metabolomics by using an ultrahigh performance liquid chromatography (UHPLC, 1290 Infinity LC; Agilent Technologies, Santa Clara, CA) coupled to a quadrupole time-of-flight (AB Sciex TripleTOF 6600; Applied Protein Technology, Shanghai, China). The analytes were separated on a 2.1 mm × 100 mm ACQUIY UPLC BEH 1.7-μm C18 column (Waters, Dublin, Ireland) and analyzed in both electrospray-positive and electrospray-negative ionization modes.

Metagenomics

DNA sample testing, library construction, and sequencing with an Illumina HiSeq platform were conducted at Novogene Technology Co, Ltd. The analyzing steps were described previously.⁵¹ In short, quality control was performed with Kneaddata pipeline (https://bitbucket.org/biobakery/kneaddata) to exclude the host genome. Then, these clean reads were assembled to contigs via MEGAHIT,⁵⁸ and Prokka⁵⁹ was used for contigs identification in a conda environment of MetaWrap,⁶⁰ and CD-HIT⁶¹ was used for nonredundant genes. A gene abundance table was generated through Salmon⁶² and functional annotations of Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthology by eggNOG-mapper⁶³ with its related database.

SCFA Quantification

Cecum content of SCFAs were analyzed based on previous method.⁶⁴ Briefly, samples (50 mg) were suspended in 200 μL distilled water and acidified with 50 μL (50%) sulfuric acid. Then, the solution was vortexed vigorously and extracted with diethyl ether. SCFAs were quantified by a gas chromatography Nexis GC-2030 (Shimadzu, Kyoto, Japan).

Staining Procedures

The liver samples were formalin-fixed and sectioned for H&E staining as described.¹⁹ Frozen liver sections were stained with Oil red O.¹⁹ Formalin-fixed intestinal samples were embedded in paraffin and stained with anti-AHR antibody (ab84833, 1:200; Abcam, Cambridge, MA) and antimyeloperoxidase antibody (ab9535, 1:200; Abcam). All sections were scanned by Nano Zoomer 2.0 RS (Hamamatsu).⁶⁵

Western Blot Analysis

Western blot analysis was performed as described in our previous work.¹⁹ Briefly, intestine or liver tissues were lysed and the extracted protein was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, followed by transferring into a nitrocellulose membrane. The membranes were blocked and probed with antibodies against CYP2E1 (AB1252, 1:1000; Millipore, Bedford, MA) or actin (A5441, 1:5000; Sigma-Aldrich) overnight and then incubated with the corresponding secondary antibodies. The protein bands were quantified using Image J (National Institutes of Health, Bethesda, MD), normalized to actin.

Cell Culture Experiments

MODE-K cells (C495; Shanghai Hongshun Biotechnology Co, Ltd, Shanghai, China) were treated with vehicle (PBS) or ethanol (86, 100, and 200 mmol/L) for 24 hours for extraction of RNA and qPCR analysis. For H. hepaticus co-culture experiments, AML12 cells (CRL-2254; ATCC, Rockville, MD) were seeded in 12-well plates overnight, and subsequently exposed to H. hepaticus collected from the plates (OD₆₀₀ = 1, suspended in cell culture medium, 100 μL/well). After incubation for 2 hours anaerobically, the plate was subjected to aerobic conditions for a total of 24 hours.⁶⁶ The treated cells were stained with Oil red O or extracted RNA for qPCR analysis.

Luciferase-Reporter Assay

AML12 cells were seeded in a 24-well plate and cultured for 12 hours before transfection with Ahr reporter plasmid (kindly provided by Professor Dalei Wu, Shandong University, China)⁶⁷ and pRL-TK Vector (E2241; Promega, Madison, WI). After 24 hours, cells were treated with AHR agonists (suspended in DMSO initially and diluted with medium). With another 24 hours of incubation, cells were washed with PBS and then lysed with lysis buffer (E1910; Promega). Luciferase activity was measured using a microplate reader (Tecan Spark, Männedorf, Switzerland) after adding luciferase reagent (E1910; Promega).

Isolation of IECs and Lamina Propria Cells

Small intestines were harvested immediately once mice were killed. After removal of all visible fat and Peyer’s patches, all intestines were cut open longitudinally. Then, IECs and lamina propria cells were collected based on the protocol as described previously.²⁶

Isolation of Kupffer Cells

Kupffer cells were isolated as described in our previous work.⁶⁸

Organoid Culture

Intestinal organoids were derived from WT mice and cultured as described previously.69, 70, 71 In brief, small intestine was harvested and flushed with cold Dulbecco's phosphate-buffered saline (DPBS). The intestine was cut longitudinally followed by removal of intestinal contents and villi. Intestine was washed and cut into 2- to 4-mm pieces, then tissues were collected after filtrating through a 70-μm cell strainer. After digestion in 2 mmol/L EDTA of DPBS for 30 minutes, intestine pieces were transferred into DPBS and shaken vigorously. After filtration through a 70-μm cell strainer 3 times, the crypts were harvested by centrifuging (600 × g, 5 min). Crypts were suspended in Matrigel (3432-010-01; R&D Biosystems, Minneapolis, MN) and cultured with IntestiCult OGM mouse basal medium (06000; Stemcell Technologies, Vancouver, Canada).

Microinjection of H.hepaticus Into Organoids

H.hepaticus was microinjected into the intestinal organoids as previously described.^71,72 Antibiotic-free mouse basal medium was refreshed every 3 days to remove antibiotics before injection. H. hepaticus was collected and suspended in antibiotic-free medium (OD₆₀₀ = 0.4). Then, H. hepaticus with or without POPG (5 μg/mL) was microinjected into lumens of organoids with FastGreen dye (0.05%, wt/vol, F7252; Sigma-Aldrich), which is used to track the injected organoids. After 2 hours, fresh medium with nonpermeant gentamicin (5 μg/mL) was added to prevent the overgrowth of bacteria outside the organoids. After 4 days of incubation, the injected organoids were dissociated with 200 μL broth containing 0.5% saponin (SAE0073; Sigma-Aldrich) for 15 minutes with repeated pipetting on ice. The bacteria were released and cultured on plates with 8% sheep blood for 6 days to determine OD₆₀₀ analysis.

Synthesis of FICZ

The starting materials were 1-Boc-3-(2-ethoxyl-2-oxoethyl)-indole and 1-Boc-2-chloro-3-formyl-indole, and FICZ was synthesized as described in the previous work.^73,74

Statistical Analysis

Results in this study are expressed as means ± SEM. Statistical significance between 2 groups was determined by the Mann–Whitney test or an unpaired t test. All analyses were performed using GraphPad PRISM 7.0 (San Diego, CA). P < 0.05 was considered statistically significant. For untargeted metabolomics, significance between groups was identified based on the combination of variable influence on projection values (>1.0) and P values (<0.1).

All authors had access to the study data and reviewed and approved the final manuscript.

Supplementary Material

Supplementary video

To clearly present the process that bacteria were microinjected into the intestinal organoid, Lactobacillus murinus with visible size were used instead of H.hepaticus, which is too small to be observed by video. L murinus were microinjected together with FastGreen dye (0.05%, wt/vol) to track the injected intestinal organoid.