Impaired 25-hydroxylation of vitamin D in liver injury suppresses intestinal Paneth cell defensins, leading to gut dysbiosis and liver fibrogenesis

Abstract

Vitamin D deficiency is coprevalent with various liver diseases, including cirrhosis, whereas the underlying mechanism remains elusive. Vitamin D receptor (VDR) is abundantly expressed in the distal region of the small intestine, where the Paneth cells are enriched, suggesting that vitamin D signaling may modulate the intestinal Paneth cells and their production of defensins to restrain microbiome growth in the small intestine. In this study, we found that in carbon tetrachloride-induced liver injury, hepatic 25-hydroxylation of vitamin D was impaired, leading to downregulated expression of Paneth cell defensins in the small intestine, gut dysbiosis, and endotoxinemia. Although intraperitoneal injection of endotoxin (lipopolysaccharides) alone did not elicit liver fibrosis, it exacerbated the carbon tetrachloride-initiated liver fibrogenesis. Oral gavage of synthetic Paneth cell α-defensin 5 (DEFA5) restored the homeostasis of the gut microbiota, reduced endotoxemia, relieved liver inflammation, and ameliorated liver fibrosis. Likewise, cholestyramine, a cationic resin that can sequestrate endotoxin in the intestine, attenuated liver fibrosis as well. Fecal transplant of the microbes derived from the DEFA5-treated donors improved liver fibrosis in the recipient mice. The intestinal Vdr conditional knockout mice exhibited reduction of Paneth cell defensins and lysozyme production and worsened liver injury and fibrogenesis. Thus, liver injury impairs synthesis of 25(OH)VD₃, which consequently impedes the Paneth cell functions in the small intestine, leading to gut dysbiosis and liver fibrogenesis.

NEW & NOTEWORTHY Vitamin D deficiency is coprevalent with various liver diseases, indicating the role of vitamin D in maintaining liver homeostasis. In this study, we observed that the hepatic 25-hydroxylation of VD is critical for intestinal innate immunity through VD signaling in the small intestine for maintaining Paneth cell functions. Conversely, failure of biogenesis of VD in the liver impairs intestinal immunity, leading to gut dysbiosis and endotoxemia, which promotes liver fibrogenesis.

INTRODUCTION

Tissue fibrosis is a common pathological consequence in response to extensive and persistent tissue damage, regardless of the nature of insults. Xenobiotics, alcohol, drugs, hepatic viral infections, and microbes from hepatic portal circulation are the major causes of liver stress, injury, and fibrosis. Pathogen-associated molecular patterns (PAMPs), which initiate the pathogen-induced inflammatory response in the situations of hepatic infection and bacterial translocation from portal circulation, are the causes of liver injury and fibrosis. On the other hand, damage-associated molecular patterns (DAMPs) from hepatic xenobiotic and toxin exposure may cause sterile inflammation as well. Both PAMPs and DAMPs are recognized by pattern recognition receptors that can initiate and perpetuate the inflammatory response for host defense. Moreover, excessive and prolonged inflammation, the cytokine storm, and growth factors can drive fibrogenesis for tissue repair and wound healing.

Hepatic stellate cells (HSCs) are the major source of liver fibrosis (9, 22). As hepatic pericytes and mesenchymal cells in nature, located in the space of Disse, HSCs assume their physiological functions for vitamin A storage and metabolism. HSCs are the major hepatic cells that produce soft/loosened extracellular matrix (ECM) for sinusoidal construction and functionality (10, 29). HSCs are the major cellular basis for liver fibrosis through activation in response to liver injury and microbial products from the intestine. On the other hand, HSCs also secrete ECM-degrading enzymes such as matrix metalloproteinases (MMPs) (3). MMPs produced by HSCs within the space of Disse can activate and release the latent forms of growth factors for wound healing and fibrogenesis (16). Different from other organs, the blood supply for the liver comes mainly from the portal vein. Thus, the components of food antigen and metabolites of gut microbiome can impact the liver function in wound healing and fibrosis.

In addition to maintaining calcium and phosphorus homeostasis through induction of the intestinal calcium transporter, vitamin D signaling, via the nuclear receptor (VDR) and transcriptional regulation, is known for maintaining immune systems (11, 13). In general, vitamin D signaling can activate innate immunity and simultaneously suppress adaptive immunity (7, 14). Conversely, vitamin D insufficiency or deficiency is coprevalent overwhelmingly with the incidences of a variety of degenerative diseases, including cirrhosis, type 2 diabetes mellitus, and metabolic disorders (2, 5, 18, 24). On the other hand, the gut microbiome, as a symbiotic community, is critical for maintaining the host metabolic homeostasis. Dysbiosis is related to metabolic syndrome, type 2 diabetes, fatty liver diseases, cirrhosis, and even hepatocellular carcinoma (20, 26). Intestinal innate immunity, including defensins and lysozymes secreted by the Paneth cells along with the tight junction of the intestinal epithelia, is critical to prevent bacteria from up translocalization in the small intestine and to restrain the bacterial products from entering portal circulation.

Although a massive number of microbes are presented in the colon, the small intestine has a very minimal amount of bacteria. Thus, there must be a gating system in the distal region of the small intestine, namely, ileum, to restrain microbe overgrowth spreading to the upper region of the small intestine. VDR is highly expressed in the terminal part of the ileum, suggesting that the vitamin D (VD) signal is involved in maintaining intestinal immune homeostasis (6, 25), whereas it is unknown whether liver injury can impair the Paneth cells and intestinal innate immunity. The liver is a major organ for the biosynthesis of bioactive vitamin D through 25-hydroxylation of VD. Thus, it is possible that liver injury and its resultant impairment of VD biogenesis may consequently impede gut innate immunity and the gut microbiome, which in turn may augment liver fibrogenesis.

In this study, we observed that hepatic sterile injury resulted in downregulation of the hepatic expression of Cyp27A1 and Cyp2R1, the hepatic 25-hydroxylases, leading to insufficient circulation of 25-OH VD₃ under the condition of sufficient VD dietary supplementation. Importantly, we observed that deficiency of active VD production from liver injury led to impaired Paneth cell functions, featured by downregulation of α-defensin 5 and lysozyme in the ileum, which consequently led to gut dysbiosis and endotoxemia that ultimately exacerbated liver fibrogenesis. Administration of synthetic α-defensin 5 can restrain endotoxemia. Likewise, administration of cholestyramine to sequestrate gut endotoxin attenuated hepatic inflammation and fibrosis. Moreover, fecal transplant of the microbes derived from the α-defensin 5 (DEFA5)-treated mice could relieve liver fibrosis. Conversely, Vdr conditional knockout in the small intestine impaired expression of α-defensin 5, which exacerbated liver fibrogenesis. Thus, biological synthesis of vitamin D in the liver is necessary for maintaining innate immunity of the small intestine, which ultimately protects the liver from hepatic insults and fibrogenesis.

METHODS

Induction of liver fibrosis and treatment regimens in animal models.

All procedures of the animal experiments in this study complied with the Guide for the Care and Use of Laboratory Animals (National Research Council), and the protocols were approved through the Institutional Animal Care and Use Committee of the College of Life Sciences, Sichuan University. Male BALB/c male mice at 4–5 wk of age were maintained in a controlled specific-pathogen-free (SPF) environment in housing cages (12:12-h light/dark cycle) with free access to both food and water. Liver fibrosis in the mice (n = 10) was induced by an intraperitoneal injection of carbon tetrachloride (CCl₄) in a stepwise escalating regimen: twice per week, at 0.5 mL/kg body wt for the first week, 1.5 mL/kg body wt for the second week, and 2.5 mL/kg body wt for additional 5 wk. In one experiment, synthetic human α-defensin 5 (DEFA5) was administered through oral gavage at 10 μg/mouse, twice a week for 4 wk, starting the fourth week from initial fibrotic treatment (fecal from vehicle control, n = 3 mice; DEFA5, n = 4 mice). In another experiment, during CCl₄-induced fibrogenesis, endotoxin (lipopolysaccharide, LPS, 4B Sigma-Aldrich) was additionally administered through an intraperitoneal injection, starting from the fourth week, at a dose of 0.3 mg/kg body wt for the first 2 wk, followed by an increased dose at 0.6 mg/kg for additional 2 wk (n = 6 mice for each condition). In one experiment, after induction of liver fibrosis for 4 wk, the mice were fed 3% cholestyramine (Chol) in the chow for 4 consecutive weeks (control, n = 5 mice; fibrosis, n = 4 mice; Chol, n = 7 mice). To determine the roles of gut microbiome in liver fibrosis, gut microbes were prepared from fecal pellets of the DEFA5-treated or control mice. The fecal microbes were washed in saline, and colony-forming units (CFU) were determined as described previously (31). The mice under hepatic injury for 4 wk were administered fecal bacteria at ∼10⁷ CFU through oral gavage, once per week for 4 consecutive weeks (control, n = 3 mice; fecal from CCl₄, n = 6 mice; fecal from DEFA5, n = 6 mice). After the mice were euthanized, the liver, ileal tissues, blood, and fecal pellets were collected and stored at −80°C for analysis.

Vdr conditional knockout in intestinal epithelial cells.

Intestinal epithelial cell Vdr-knockout mice, p-villin-Vdr^+/−, were generated by crossing the Vdr-flox mice with p-villin-Cre mice. Vdr-flox mice, hereinafter Vdr^f/f, were generated at Biocytogen, Beijing. The strategy and details for creating the Vdr-flox mice will be described in an upcoming report. The p-villin-Cre mice were kindly provided by Professor Zhixiong Xiao at Sichuan University. Genotypes were validated by PCR analysis with specific primers: for villin-Vdr^+/−, forward primer as AAAAGCACATATTCGCGGGCTGTTG, and reverse primer as TGTGATGGAGGAGTGTCCCATACCA; and for p-villin-Cre mice, the forward primer as CGGTCGATGCAACGAGTGAT, and the reverse primer as CCACCGTCAGTACGTGAGAT. The mice were euthanized after being treated with CCl₄ for 5.5 wk.

Histological and immunocytochemical analysis.

The liver and ileal tissues from the euthanized mice were fixed with 4% paraformaldehyde for 2–3 days at 4°C. Hematoxylin-eosin staining for general histological assessment, Sirius Red staining for liver fibrosis, and periodic acid-Schiff staining for mucous membrane of the distal region of the small intestine were described previously (25). For immunocytochemical analysis, paraffin-embedded tissue slides were dewaxed and rehydrated, followed by antigen retrieval treatment in boiling 50 mM sodium citrate, pH 6.8. After treatment with H₂O₂, the slides were treated with 0.12% Triton X-100 for 10 min. After being blocked with 5% bovine serum albumin (BSA), primary antibodies (anti-type-I collagen at 1:150 from Southern Biotechnology, 1310-01; anti-DEFA1 at 1:200, kindly provided by Professor Andre J. Ouellette at USC; anti-lysozyme from ZenBioScience at 1:400, 381103) diluted in 5% BSA were incubated at 4°C for overnight. Secondary antibodies conjugated with horseradish peroxidase (HRP; ZSJQ Biotech, PV9003/PV6000) were applied for 1 h at room temp. Slides were developed with diaminobenzidine, followed by counterstaining with hematoxylin, dehydration with alcohol, and mounting for microscopy analysis. The images were captured via the Leica TCS SP5 II system or Nikon eclipse Ti-U microscope.

Western blot analysis.

The liver and ileal tissues were homogenized and lysed in RadioImmuno Precipitation Assay buffer containing a protease inhibitor cocktail. The concentration of total protein was detected via the Pierce bicinchoninic acid assay Protein Assay Kit (Thermo). Equal-quantity protein samples were resolved via sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. The membranes were blocked in 5% nonfat milk and hybridized with primary antibodies against type-I collagen (Cell Signaling Technology), VDR (Cell Signal Transduction, D2K6W), or GAPDH (Zen BioScience, Cat. No. EE0618), followed by incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies. Images were developed via Immobilon Western Chemiluminescent substrate (Millipore).

RT-qPCR analysis.

The liver or ileal tissue was homogenized in 1 mL of TRIzol, and total RNA was extracted. The RNA was converted to cDNA via the Transcriptor First Strand cDNA Synthesis Kit (Roche, Cat. No. 04897030001). The qPCR system contained 2 μL of cDNA, 0.2 μL of forward primer, 0.2 μL of reverse primer (200 nM), and 5 μL of MIX, and diethyl pyrocarbonate (DEPC) water was added up to 10 μL. The reactions were performed with a Bio-Rad machine Cfx96. The primer sequence information is listed in Table 1. The relative mRNA expression was normalized against the mRNA of ribosomal protein L19 (RPL-19).

Table 1.

List of mouse primers for qPCR analysis

Gene	Forward 5′ → 3′	Reverse 5′ → 3′
TNF-α	TGGGACAGTGACCTGGACTGT	TTCGGAAAGCCCATTTGAGT
MUC2	GCTCGGAACTCCAGAAAGAAG	GCCAGGGAATCGGTAGACAT
RPL-19	GAAGGTCAAAGGGAATGTGTTCA	CCTTGTCTGCCTTCAGCTTGT
Arginase-1	GAACCCAACTCTTGGGAAGAC	GGAGAAGGCGTTTGCTTAGTT
Occludin	ATGTCCGGCCGATGCTCTC	TTTGGCTGCTCTTGGGTCTGTAT
ZO-1	ACCCGAAACTGATGCTGTGGATAG	AAATGGCCGGGCAGAACTTGTGTA
Claudin-2	CCTTCGGGACTTCTACTCGC	TCACACATACCCAGTCAG
IL-6	CTTCCATCCAGTTGCCTTCTTG	AATTAAGCCTCCGACTTGTGAAG
ColI-1	GCTCCTCTTAGGGGCCACT	CCACGTCTCACCATTGGGG
VDR	GCTGATGTCAAAAGAATCAGCG	TCCCTCTGGAACTGCATGGTTCC

IL-6, interleukin 6; MUC2, mucin 2 gene; TNF-α, tumor necrosis factor-α; VDR, vitamin D receptor; ZO-1, zonula occlude 1.

16s rDNA PCR-based gut microbiota analysis.

SYBR green-based qPCR analysis of 16S rRNA genes was used to measure the relative abundance of gut bacteria, the microbiota. Fecal microbe DNA was extracted using a stool DNA kit (Omega, China). The qPCR reaction system contained 2 μL of DNA, 0.2 μL of forward primer, 0.2 μL of reverse primer (200 nM), and 5 μL of MIX, and DEPC water was added up to 10 μL. Then, the reactions were analyzed with the Bio-Rad Cfx96, and the values were expressed as the percentage of a common bacterial readout as internal reference (primers for 16S rDNA analysis is summarized in Table 2). The accuracy of the qPCR-based 16 rDNA analysis was previously validated by sequencing the PCR products. The heat map was generated by using the TIGR Multi Experiment Viewer (MeV 4) software package.

Table 2.

List of primers for 16S rDNA qPCR analysis

Gene	Forward 5′ → 3′	Reverse 5′ → 3′
All bacteria	ACTCCTACGGGAGGCAGCAG	ATTACCGCGGCTGCTGG
Firmicutes	GGAGTATGTGGTTTAATTCGAAGCA	AGCTGACGACAACCATGCAC
Bacteriodetes	GGARCATGTGGTTTAATTCGATGAT	AGCTGACGACAACCATGCAGG
α-Proteobacteria	CIAGTGTAGAGGTGAAATTC	CCCCGTCAATTCCTTTGAGTT
γ-Proteobacteria ε-Proteobacteria Desulfovibrio	TCGTCAGCTCGTGTYGTGA TGGTGTAGGGGTAAAATCCG CCGTAGATATCTGGAGGAACATCAG	GGTAAGGGCCATGATG AGGTAAGGTTCTTCGYGTATC ACATCTAGCATCCATCGTTTACAGC

Determining the plasma levels of endotoxin and 25(OH)VD₃.

The plasma endotoxin (LPS) concentration was determined by using the Limulus Amebocyte Extract kit (CE80545, Chinese Horseshoe Crab Reagent Manufactory, Xiamen, China). After 1:10 dilution in sample processing buffer and heating for 10 min at 70°C, plasma LPS content was analyzed following the manufacturer’s instruction. Plasma tumor necrosis factor-α (TNF-α) levels were analyzed by using an ELISA kit (Mercodia, Uppsala, Sweden; DRE30030, China). Plasma 25(OH)VD₃ levels were measured by using an ELISA kit (OCTEIA 25-hydroxy vitamin D, AC-57F1, IDS, Ltd.).

Statistical analysis.

Data were analyzed by using SPSS software, and the statistical significance was determined with Tukey’s test or t test. Tukey’s test was mainly used for data of three groups or more, and the t test was mainly used for data of two groups. The results were presented as means ± SE. Statistical significance is presented as *P < 0.05 or **P < 0.01 or ***P < 0.001.

RESULTS

Impairment of intestinal Paneth cells in liver injury and fibrosis is related to reduced hepatic 25-hydroxylation of vitamin D.

Liver injury and fibrosis were induced by sterile hepatic insult with repeated intraperitoneal injections of CCl₄ for 8 wk (n = 10, male). The hepatic toxin was increased in a stepwise manner. Through this, the Balb/C mice could tolerate the high doses of CCl₄, and liver fibrosis was evidently generated, indicated by the formation of hepatic fibrotic septa and increased expression of type-I collagen and transforming growth factor-β1 (TGF-β1) at mRNA levels (Fig. 1, A–C). Hepatocytes are the major source of 25-hydroxylation of vitamin D, whereas nonhepatic sources such as peripheral leukocytes may contribute to local and paracrine synthesis and function. Here, we found that the plasma levels of 25(OH)-VD₃ were significantly decreased in the mice with liver injury and fibrosis (Fig. 1D). Accordingly, Cyp27A1, a hepatic 25-hydroxylase of VD, was significantly downregulated in the fibrotic liver, whereas expression of Cyp2R1 was not changed significantly (Fig. 1E).

Fig. 1.

Impairment of intestinal Paneth cells in liver injury and fibrosis is related to reduced hepatic synthesis of 25-hydroxyl vitamin D. A: male Blab/V mice were subjected to CCl₄-induced liver injury and fibrosis for a total of 8 wk (n = 10 mice). Liver fibrosis was assessed by Sirius Red staining. B: liver fibrosis was further measured by RT-PCR analysis for expression of type-I collagen. C: expression of TGF-β1 in liver fibrosis as the mRNA levels was determined by RT-qPCR analysis. D: plasma 25-OHVD₃ levels were measured by ELISA. E: hepatic expression of 25-hydroxylase, Cyp2R1, and Cyp27A1 was determined by RT-qPCR analysis. F: expression of defensins in the ileum was detected through RT-qPCR analysis. G: intestinal mucus in the ileum was measured by PAS staining and Paneth cell-specific markers, including lysozyme and MMP-7 in the ileal crypts, were measured by immunohistochemistry. H: expression of ZO-1 in the ileum was determined by immunohistochemistry. *P < 0.05; **P < 0.01. ns, not significant. Comparisons have been indicated with bars. Data show means ± SE. The data in B, C, D, E, and F were subjected to the t test. PAS, periodic acid-Schiff; CCl₄, carbon tetrachloride.

Since Vdr is highly expressed in the ileum, we then test the hypothesis that liver-injury-exerted vitamin D deficiency may consequently impede the intestinal mucosal integrity and Paneth cell functions in the distal small intestine. As shown, the mRNA levels of α-defensin 2 and α-defensin 5 were downregulated in the distal intestine of the mice with injury and fibrosis (Fig. 1F). Moreover, the Goblet cells in the villi for mucin production were substantially decreased in the distal small intestine of the mice with liver injury and fibrosis (Fig. 1G, top, pink-colored cells). Lysozyme and MMP-7, a converting enzyme for prodefensins, are the lineage markers for Paneth cells in small intestine. As shown, both lysozyme and matrix metalloproteinase-7 (MMP-7; a converting enzyme for Paneth cell defensins) in the crypts of Lieberkühn were downregulated in liver fibrosis (Fig. 1G, middle and bottom). We then measured the integrity and tight junctions in the ileum region of the mice. As shown, zonula occlude-1 staining was downregulated (Fig. 1H) along with the decreased mRNA levels in the mice with liver injury. Taken together, the experimental results support a notion that the liver-injury-initiated impairment of vitamin D biogenesis is related to the impeded intestinal innate immunity and Paneth cell functions.

Oral administration of α-defensin 5 mitigates liver fibrosis.

α-Defensin 5/6, the secretory defensins generated by the Paneth cells in the small intestine, are critical for restriction of bacterial overgrowth in the small intestine (4). We then asked if the impaired production of the Paneth cell defensins in the small intestine had a causative impact on the liver functions presumably through suppression of bacterial growth and regulation of the gut microbiome. Synthetic human α-defensin 5 (DEFA5) was administered via oral gavage to the mice with hepatic injury and liver fibrosis (Fig. 2A). As shown, the fibrotic septa and infiltration of CD3+ lymphocytes in the liver were partially but significantly decreased by DEFA5 treatment (Fig. 2B). The expression of type-I collagen protein in liver fibrosis was partially suppressed by oral gavage of DEFA5 (Fig. 2C). Inflammation in the ileum, as indicated by elevation of the mRNA levels of IL-1β and TNF-α, was also suppressed by oral administration of DEFA5 (unpublished data). The increased gut permeability from liver injury and fibrosis was improved by repeated administration of DEFA5 (Fig. 2D), which consequently reduced plasma endotoxin as well (Fig. 2E). We then measured the gut microbiome and found that the abundance of Bacteriodetes at the phyla level was downregulated in the mice with liver injury and fibrosis (Fig. 2F), similar to what we found previously in mice with vitamin D deficiency and high-fat feeding (25). Administration of DEFA5 upregulated the abundance of Bacteriodetes. Moreover, the increased abundance of Firmicutes at the phyla level and γ-Proteobacteria at the class level in liver fibrosis was partially downregulated by administration of DEFA5.

Fig. 2.

Liver fibrosis is mitigated by oral administration of α-defensin 5. A: the experimental design and treatment regimen. Liver fibrotic mice were treated by oral gavage of human α-defensin 5 (DEFA5), as mentioned in the methods. B: liver fibrosis was measured by immunofluorescent staining against type-I collagen. Infiltration of lymphocytes, CD3+ cells, in the liver was assessed by immunohistochemistry. C: Western blot analysis was used to measure liver fibrosis. D: gut permeability was assessed by oral gavage of FITC-dextran, followed by measuring the fluorescence emission in the plasma. E: plasma endotoxin (LPS) levels in the serum were measured by the Limulus Amebocyte Extract assay. F: the abundance of bacteria in the feces was determined by qPCR analysis for 16S rDNA. Sample size: vehicle control, n = 3 mice; DEFA5, n = 4 mice. *P < 0.05, **P < 0.01. Comparisons have been indicated with bars. Data show means ± SE. a-P, α-Proteobacteria; Bact, Bacteroidetes; Firm, Firmicutes; e-P, ε-Proteobacteria; γ-P, γ-Proteobacteria; Prevo, Prevotella. Data in D and E were subjected to Tukey’s test. LPS, lipopolysaccharide.

Fecal transplantation of the microbiome derived from DEFA5-treated mice can ameliorate liver fibrosis.

Given the evidence that oral DEFA5 treatment restored eubiosis, we then asked if the restored microbiome is involved in the improvement of liver fibrogenesis. Donor fecal microbes were prepared from the mice that had been treated with or without DEFA5 in the course of liver fibrogenesis (Fig. 3A). The results showed that treatments for 4 consecutive weeks with the fecal microbes derived from the DEFA5-treated mice partially improved liver fibrosis, as demonstrated by suppression of hepatic expression of type-I collagen and α-smooth actin (Fig. 3, B and C). The morphological fibrotic septa in the liver were significantly improved (Fig. 3, D and E). The fecal transplant also restored the enterotype of the gut microbiota, showing a decrease of Firmicutes and an increase of Bacteroidetes (Fig. 3F). Importantly, the results show that the dysbiosis resulting from liver injury in agreement with vitamin D deficiency had a causal role in liver fibrogenesis. On the other hand, an intervention to normalize the gut microbiome can alleviate cirrhosis.

Fig. 3.

Fecal transplant of microbiome from the DEFA5-treated mice alleviates liver fibrosis. A: experimental design and treatment regimen. The mice under hepatic injury for 4 wk, as recipients, were administered fecal microbes through oral gavage, once a week for consecutive 4 wk. Donor fecal microbes were prepared from the liver fibrotic mice that had been treated with or without DEFA5 for 4 wk. n = 6–10 mice. B and C: liver fibrosis of the recipients was quantitated by RT-qPCR analysis for expression of type-I collagen and α-smooth actin at their mRNA levels. D and E: the morphological fibrotic septa in the liver parenchyma of recipients were determined by immunocytochemical staining for type-I collagen and quantitated as the relative area of fibrotic septa. F: at the end of treatment, the gut microbiota of the recipients was determined by 16S-rDNA-based qPCR analysis. *P < 0.05; **P < 0.01; ***P < 0.001. Comparisons have been indicated with bars. Data show means ± SE. a-P, α-Proteobacteria; e-P, ε-Proteobacteria; Bact, Bacteroidetes; DEFA5, α-defensin 5; Firm, Firmicutes; γ-P, γ-Proteobacteria; Prevo, Prevotella.

Although endotoxin alone does not initiate fibrosis, it exacerbates the existing fibrogenesis.

Endotoxin derived from gram-negative bacteria from the gut is a predominant driving force for HSC activation and liver fibrosis (19, 23). The finding of increased plasma endotoxin in liver fibrosis by our animal model prompted us to examine the role of bacterial endotoxin in the context of liver injury and failed 25-OH VD synthesis. In the course of CCl₄-induced liver injury or control, additional endotoxin (lipopolysaccharide, LPS) was injected intraperitoneally, starting from the fourth week, at a dose of 0.3 mg/kg body wt for the first 2 wk, followed by an increased dose at 0.6 mg/kg for additional 2 wk. As shown in Fig. 4A, repeated injections of LPS alone did not induce liver fibrosis but could enhance the CCl₄-initiated fibrotic septa, suggesting that endotoxemia may serve as a second hit promoting fibrogenesis. Expression of type-I collagen was barely induced by LPS treatment but was significantly enhanced in conjunction with the CCl₄ hit (Fig. 4B). The results agree with the morphological improvement. Our previous work showed that IL-1 can promote HSC activation (transdifferentiation) and fibrosis through MMP-mediated activation of latent TGF-β (12, 16). Here, we noticed that the LPS-mediated promotion of liver fibrosis is related to the IL-1 expression (Fig. 4C). Importantly, we noted that the combination of LPS with CCl₄ substantially suppressed the expression of α-defensin and mucin 2 (MUC2) in the ileum (Fig. 4, D–F). Therefore, the dysbiosis-exerted endotoxemia in this context may not be sufficient to induce liver fibrosis, but it works as a second hit to enhance the liver fibrogenesis.

Fig. 4.

Endotoxin exacerbates the sterile injury-induced liver fibrosis. Mice were subjected to hepatic injury by repeated injection of CCl₄ together with or without additional administration of endotoxin (LPS) through intraperitoneal injections for a total of 5.5 wk. N = 6 mice. A: the morphological fibrotic septa in the liver parenchyma were determined by Sirius Red staining. B: liver fibrosis was quantitated by RT-qPCR analysis for collagen 1-α. C: inflammation in the liver was determined by RT-qPCR analysis. D–F: expression of DEFA5, muc2, and occludin in the ileum was determined by RT-qPCR analysis. *P < 0.05; **P < 0.01. ns, not significant. Comparisons have been indicated with bars. Data show means ± SE. CCl₄, carbon tetrachloride; DEFA5, α-defensin 5; IL-1β; interleukin-1-β; LPS, lipopolysaccharide.

Oral administration of cationic resin to sequestrate gut endotoxins ameliorates hepatic injury and fibrosis.

We previously found that oral administration of cationic resins, cholestyramine, can sequestrate the negatively charged endotoxin and profoundly reduces plasma endotoxin in the mice with high-fat diet-induced metabolic disorders (31). Given the fact that plasma endotoxin is elevated in the mice with hepatic sterile injury and its role in the enhancement of liver fibrosis, we administered cholestyramine to the mice by mixing with chow at 3% wt/wt during induction of liver fibrogenesis. As shown, parenchymal necrosis and liver fibrosis were improved through oral administration of cholestyramine (Fig. 5A). Cholestyramine administration resulted in suppression of type-I collagen expression (Fig. 5B). Also, the increased plasma levels of alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), and direct bilirubin (DBIL), the markers of hepatic injury, were all mitigated by the cholestyramine treatment (Fig. 5, C–F). These results thus demonstrate the importance of bacteria-produced endotoxins in promoting fibrogenesis through portal liver circulation.

Fig. 5.

Oral administration of cholestyramine to sequestrate intestinal endotoxin mitigates liver injury and fibrosis. Liver fibrosis was induced through repeated injection of CCl₄ for 8 wk. After the initial 4 wk of induction, a group of the mice were fed chow containing 3% cholestyramine (n = 7 mice) for additional 4 wk. A: H&E staining for the liver histology and immunohistochemical staining for type-I collagen. B: liver fibrosis, as the mRNA levels of type-I collagen, was quantitated by RT- qPCR analysis. C–F: the liver functions were measured for plasma levels of alanine transaminase (ALT), aspartate transaminase (AST), and total bilirubin (TBIL), whereas gall bladder and pancreas functions were assessed by plasma direct bilirubin (DBIL). Sample size: control, n = 5 mice; fibrosis, n = 4 mice; Chol, n = 7. *P < 0.05; **P < 0.01; ***P < 0.001. ns, not significant. Comparisons have been indicated with bars. Data show means ± SE. CCl₄, carbon tetrachloride; Chol, cholestyramine; H&E, hematoxylin-eosin.

Intestinal knockout of Vdr gene leads to Paneth cell dysfunction and exacerbation of liver fibrogenesis.

Our previous work demonstrated that Vdr is massively expressed in the distal region of the small intestine, which is ∼1,000-fold greater than the expression in the liver (25), suggesting that VD signaling plays critical functions in the ileum innate immunity, including that of Paneth cells. To test the role of Vdr in the intestine, we generated conditionally Vdr-knockout mice through crossing the flox-Vdr mice with p-villin-Cre mice. Whereas the homozygous mice showed signs of developmental defects, the F1 heterozygous mice were grossly normal. The knockout efficiency in the ileum of the p-villin-Vdr^+/− mice was evident as partial reduction of the mRNA and protein levels of Vdr in the distal region of the small intestine (Fig. 6, A and B). Immunohistochemical staining clearly revealed the distinct expression pattern of lysozyme and α-defensin 1 at the bottom of crypts of villi, as authentic markers for Paneth cells, and their expression was significantly downregulated in the ileum of Vdr^+/− mice (Fig. 6C). The expression pattern of Vdr in the epithelial cells in the ileum was confirmed by immunohistochemical staining (Fig. 6D). Importantly, the expression of α-defensin 5 at its mRNA levels was significantly downregulated in the Vdr-knockout mice (Fig. 6E). We then challenged the Vdr-c-knockout mice with CCl₄ for 6 wk. As shown, liver fibrosis as indicated by fibrotic bridges was significantly enhanced in the Vdr conditional knockout mice (Fig. 6, F and G). These results demonstrate the importance of physiological VD signaling in small intestinal epithelial cells, which consequently suppresses liver fibrogenesis.

Fig. 6.

Vdr knockout in the intestinal epithelia impairs Paneth cells and exacerbates liver fibrosis. A: the wild-type (Vdr-flox) mice and heterozygous Vrd-knockout mice in the intestinal epithelial cells (p-villin-Vdr^+/−) were described in methods. The knockout efficiency of Vdr in the ileum was determined by RT-qPCR (n = 4 or 5 mice). B: Vdr protein expressed in the ileum was determined by Western blot analysis. C: Paneth cell markers, including lysozyme and α-defensin 1 (DEFA1) expressed in the crypts of Lieberkühn of the ileum, were visualized by immunohistological staining. D: expression of Vdr in the distal region of small intestine was measured by immunohistological staining. E: expression of α-defensin 5 (DEFA5) in the distal region of the small intestine was measured by RT-qPCR analysis. D: expression of lysozyme and DEFA1 in the crypts of the distal region of the small intestine (ileum) was determined with immunocytochemical staining. F: liver fibrosis was determined by Sirius Red staining. G: liver fibrosis levels were quantitated by RT-qPCR analysis. Sample size: cont, n = 5 mice; Vdr-c-knockout, n = 3 mice. *P < 0.05; **P < 0.01. ns, not significant. Data show means ± SE.

DISCUSSION

Through portal-hepatic circulation, the liver is directly exposed to intestinal PAMPs, such as bacterial endotoxins and DNA fragments, and DAMPs, such as alcohol, food antigen, and dietary toxins. Flux of gut-derived PAMP and DAMPs to the liver through portal circulation is limited or gated by the intestinal innate immunity, including the epithelial barrier and antimicrobe agents secreted from the Paneth cells in the small intestine. Whereas the colon contains a massive amount of microbiomes including bacteria, fungi, and enteric viruses, the small intestine has a minimal amount of microbes because of the active motility of the small intestine and the ileocecal valve presenting a barrier for the colonic bacteria up translocation. Although many tissues and cells express VDR, the nuclear receptor is highly expressed in digestive organs, and it is massively expressed in the cells in the distal region of the small intestine. Such a unique pattern of VDR expression in the gut promoted us to posit that VD signaling in the intestine may maintain the gut innate immunity and even determine the enterotype of the gut microbiome. Our previous work showed that long-term dietary depletion of vitamin D in mice generates spontaneous liver fibrosis through activation of inflammation in the liver (32). A report showed that vitamin D signaling could directly suppress activation of hepatic stellate cells through inhibition of the expression of cyclin D (1), and ligand-activated VDR was found to suppress TGF-β signaling through interruption of Smad recruitment and activation of hepatic stellate cells (8). Similarly, in an animal model, VDR agonists can relieve liver injury and fibrosis (28). Nevertheless, the VDR level in the liver of mice is extremely low, whereas it is massively expressed in the ileum, indicating the possibility of vitamin D being involved in regulating a gut-liver axis that controls the homeostasis of the liver. Conversely, impairment of such an axis may promote liver diseases such as fibrosis.

The standard chow for the mice [American Institute of Nutrition 93 (AIN93)] has enough VD₃ supplementation, which is equivalent to the daily value for human beings, as recommended by the World Health Organization. At this level of dietary VD supplementation, we found the liver-injury-exerted vitamin D deficiency, as defined by reduction of plasma 25(OH)VD₃ levels (<50 nM or 20 ng/ mL), which is due to suppressed expression of 25-hyroxylase. The importance of this finding is its agreement with clinical surveys showing that vitamin D deficiency is tightly associated with the prognosis of liver injury and cirrhosis (15, 17). In agreement with our previous findings, VD signaling can upregulate intestinal innate immunity. The VDR response elements, a cis-element for transcriptional controls, are found in the 5′ promoters of the genes for innate immunity such as Paneth cell-specific α-defensins, tight junctions and intestinal mucin, and MMP-7 (Wu, unpublished results). Here, we demonstrated that in the mice with liver injury and fibrosis, the expression of α-defensins and lysozyme in the Paneth cells of the distal region in the small intestine was hampered, leading to gut dysbiosis and increased flux of endotoxin into the plasma for hepatic inflammation.

Dysbiosis of the gut microbiota is a key feature of liver diseases. Excessive flux of endotoxin into the liver is believed to be a key driving force for hepatic inflammation, which is confirmed here by carbon tetrachloride-induced liver injury. A finding in this work showed that endotoxin alone failed to elicit liver fibrosis, but it boosted sterile injury and fibrogenesis significantly. Although the underlying mechanism is not fully understood, it is physiologically reasonable, since the liver exposes itself to the food antigen and microbiological components, while immune tolerance is unique for the liver. The causative role of gut dysbiosis in worsening sterile liver injury and fibrosis is demonstrated by many lines of evidence presented in this study. First, α-defensin 5 supplementation restored the gut microbiota to a normal state, leading to improvement of liver fibrosis. Second, fecal transplant of the gut microbes derived from the defensin-treated mice was able to mitigate liver fibrosis. Third, administration of cationic resins, cholestyramine, to sequestrate intestinal bile acid can ameliorate liver fibrosis as well (27). In this study, we found that the positively charged resin can thoroughly sequestrate intestinal endotoxin, a highly negatively charged lipopolysaccharide. How the antibacterial peptides or polyamine resins impact the gut microbiome is unknown. We previously noticed that under in vivo and in vitro conditions, DEFA5 could directly suppress the growth of Helicobacter hepaticus, which belongs to the phylum of Proteobacteria (25). Using MMP-7 knockout and transgenic expression of DEFA5, a previous report demonstrated the involvement of Paneth cell-specific defensins in regulating the gut microbiome (21). A previous work showed that the absence of intestinal epithelial VDR affects microbial assemblage and increases susceptibility to dextran sodium sulfate-induced colitis in intestinal epithelial VDR (30). Importantly, these results may serve as a preclinical evidence for the potential application of polyamine resins such as cholestyramine to treat liver fibrosis through balancing of the gut microbiota and sequestration of endotoxin. In summary, liver injury and fibrosis are associated with VD deficiency due to decreased hepatic 25-hydroxylation of vitamin D in the liver. Insufficiency of VD signaling may impair gut innate immunity and integrity, including downregulation of Paneth cell functions, among others, leading to bacterial up translocation for endotoxinemia and gut dysbiosis, which may consequently promote liver fibrogenesis.

GRANTS

The work was supported by the Natural Science Foundation of China (NSFC), Grants 31571165 and 31771288 to Y.-P. Han, and the National Institutes of Health (NIH) Grants P50 AA011999 and P01DK098108 to S. J. Pandol and NIH R01DA042632 to C. Ji.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

P.W., M.L., and Y-P.H., conceived and designed research; P.W., R.Z., and T.Z. performed experiments; L.P, S.X., L.P., and R.H. analyzed data; F.R., C.J., and R.H. interpreted results of experiments; M.N., S.J.P., and Y-P.H. edited and revised manuscript; S.J.P. approved final version of manuscript.