Research ArticleLiver disease

Inhibition of ileal bile acid uptake protects against nonalcoholic fatty liver disease in high-fat diet–fed mice

See allHide authors and affiliations

Science Translational Medicine  21 Sep 2016:
Vol. 8, Issue 357, pp. 357ra122
DOI: 10.1126/scitranslmed.aaf4823

Blocking bile acids to protect the liver

Nonalcoholic fatty liver disease, which is associated with the metabolic syndrome, is becoming increasingly common, and there is no specific treatment available. Although the pathogenesis of this disorder is not yet fully understood, it is known that bile acids play key roles in lipid metabolism. Rao et al. have now identified a drug that can be given by mouth and is not systemically absorbed, but inhibits bile acid absorption from the intestine and thereby reduces the severity of fatty liver disease in a mouse model. In addition to its beneficial effects on the liver, the treatment improved glucose tolerance and cholesterol concentrations in the treated animals, suggesting that it may be useful for treating multiple components of the metabolic syndrome.

Abstract

Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease in the Western world, and safe and effective therapies are needed. Bile acids (BAs) and their receptors [including the nuclear receptor for BAs, farnesoid X receptor (FXR)] play integral roles in regulating whole-body metabolism and hepatic lipid homeostasis. We hypothesized that interruption of the enterohepatic BA circulation using a luminally restricted apical sodium-dependent BA transporter (ASBT) inhibitor (ASBTi; SC-435) would modify signaling in the gut-liver axis and reduce steatohepatitis in high-fat diet (HFD)–fed mice. Administration of this ASBTi increased fecal BA excretion and messenger RNA (mRNA) expression of BA synthesis genes in liver and reduced mRNA expression of ileal BA-responsive genes, including the negative feedback regulator of BA synthesis, fibroblast growth factor 15. ASBT inhibition resulted in a marked shift in hepatic BA composition, with a reduction in hydrophilic, FXR antagonistic species and an increase in FXR agonistic BAs. ASBT inhibition restored glucose tolerance, reduced hepatic triglyceride and total cholesterol concentrations, and improved NAFLD activity score in HFD-fed mice. These changes were associated with reduced hepatic expression of lipid synthesis genes (including liver X receptor target genes) and normalized expression of the central lipogenic transcription factor, Srebp1c. Accumulation of hepatic lipids and SREBP1 protein were markedly reduced in HFD-fed Asbt−/− mice, providing genetic evidence for a protective role mediated by interruption of the enterohepatic BA circulation. Together, these studies suggest that blocking ASBT function with a luminally restricted inhibitor can improve both hepatic and whole body aspects of NAFLD.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is one of the most common liver diseases in the Western world, with an increasing prevalence around the globe (1, 2). NAFLD encompasses a pathophysiological spectrum ranging from steatosis to nonalcoholic steatohepatitis (NASH), cirrhosis, and liver carcinoma, along with substantial liver and whole-body metabolic derangements. Effective medical therapies to slow or reverse aspects of this progression are limited. Inadequate or no response was observed for pioglitazone, vitamin E, and omega-3 fatty acid in clinical trials for NASH, although a recent trial with the farnesoid X receptor (FXR) agonistic bile acid (BA) analog obeticholic acid (OCA) showed promising results (36). Dysregulated hepatic lipid, sterol, and insulin-mediated metabolic pathways appear to be major pathophysiological contributors, but an incomplete understanding of the mechanisms underlying the development of NAFLD and its progression to NASH continue to affect the development of rational therapeutics.

BAs, their receptors (FXR, TGR5, and S1PR2), and gut luminal BA-metabolizing bacteria have emerged as important regulators of hepatic lipid and glucose metabolism (79). BAs are synthesized from cholesterol in the liver, are secreted into bile as the major solute, and function to facilitate lipid absorption in the intestine. About 95% of intestinal BAs are reabsorbed in the ileum by the apical sodium-dependent BA transporter (ASBT) and conveyed through the portal vein to the liver, where they are taken up by hepatocytes to be resecreted into bile (10). This efficient enterohepatic circulation serves to maintain the BA pool and predominantly restrict BAs to intestinal and hepatobiliary compartments. The size and composition of the BA pool is dependent on many factors, including FXR-dependent feedback regulatory pathways in both liver and ileum, enterohepatic cycling frequency, and metabolism by gut microbiota (10, 11). Changes in compartmentalization, concentration, and composition of the BA pool may have metabolic regulatory consequences, particularly in light of the spectrum of FXR agonistic and antagonistic potencies possessed by individual BAs. However, confounding our understanding of the metabolic effects of BAs through FXR and other receptors is their complex relationship with the gut microbiota, whereby the BA pool size and composition appear to be a major regulator of the microbiome community and vice versa (9).

BA signaling in the intestine and liver has a role in the regulation of lipid, glucose, and energy homeostasis and is a potential target for treatment of obesity and NAFLD (12, 13). However, the underlying mechanisms remain unclear because interventions that increase, as well as those that decrease BA (and FXR) signaling may yield metabolic benefits (8, 1420). Here, we focused on examining the effects of pharmacological and genetic inhibition of ileal BA absorption on the development of fatty liver in the American lifestyle–induced obesity syndrome (ALIOS) high-fat diet (HFD)–fed mouse model of NAFLD (21). Oral administration of SC-435, a potent luminally restricted ASBT inhibitor (ASBTi), restored glucose tolerance, reduced the steatohepatitic pathology, and altered liver gene expression in HFD-fed mice. Analysis of potential mechanisms in this dietary model and in HFD-fed Asbt−/− mice revealed that the improved features of NAFLD correlated in part with reductions in the expression of liver X receptor (LXR) target genes along with a shift toward a more hydrophobic and FXR agonistic profile of the hepatic BA pool.

RESULTS

Administration of an ASBTi impairs ileal BA uptake in HFD-fed mice

To determine whether interrupting the enterohepatic circulation of BAs with an ASBTi can improve HFD-induced hepatic steatosis, male C57Bl/6J mice were fed one of four diets for 16 weeks, as outlined in Fig. 1A: chow, HFD [modified ALIOS, 45% kcal fat plus 0.2% (w/w) cholesterol], HFD plus ASBTi for 16 weeks (HFD/ASBTi16w; 60 parts per million SC-435, ~11 mg/kg per day), or HFD for 12 weeks followed by HFD/ASBTi for the final 4 weeks (HFD/ASBTi4w; to determine whether the HFD-induced changes could be reversed). All HFD groups were given 4% sucrose water ad libitum (21). For clarity, data related to the 16-week arm are discussed first, followed by a section describing the final 4-week treatment arm. In agreement with previous studies (2224), ileal BA absorption was impaired by administration of the ASBTi, as indicated by a fourfold increase in fecal BA excretion, induction of hepatic sentinel BA synthesis target genes Cyp7a1 and Cyp8b1 [by 2.4 ± 1.0–fold (P < 0.0001) and 2.3 ± 1.0–fold (P < 0.0001), respectively; see table S1 for P values], suppression of FXR-activated Fgf15 mRNA expression in ileum [by 93 ± 7% (P < 0.0001)], and increased FXR-responsive Ibabp mRNA expression in colon [7.0 ± 2.0–fold (P < 0.0001); Fig. 1, B to E]. Asbt mRNA expression in ileum and colon was unchanged by administration of the ASBTi (fig. S1, A and B). In comparison to untreated HFD-fed mice, administration of the ASBTi induced colonic expression of Akr1b7, an FXR target gene involved in BA metabolism in mice (fig. S1C). Thus, luminally restricted inhibition of ASBT in HFD-fed mice showed the anticipated effects in ileum, colon, and liver with respect to BA metabolism. Because BAs have been reported to induce injury (25, 26), we also examined the colonic expression of genes involved in cell stress and oncogenic signaling. There were no differences between HFD and HFD/ASBTi16w mice in colonic expression of genes involved in endoplasmic reticulum stress (Atf6, Ddit3, Hspa1a, Hspa1b, Xbp1, and Xbp1p; fig. S1D), oncogenic signaling genes (Cdkn2a, Ctnnb1, Ccnd1, Myc, Mmp7, and Ogg1; fig. S1E), or genes involved in innate immunity and inflammation (Ccl2, Il1b, Il6, Il12a, Il12b, and Tnf; fig. S1F).

Fig. 1. Administration of an ASBTi increases fecal BA excretion and hepatic BA synthesis.

(A) Study design showing diets and durations of ASBTi (SC-435) treatments: chow (n = 12), HFD (n = 12), HFD/ASBTi16w (n = 16), and HFD/ASBTi4w (n = 8). (B) Fecal BA excretion in HFD and HFD/ASBTi16w feeding groups. (C) Hepatic expression of Cyp7a1 and Cyp8b1 mRNA. (D and E) Ileal Fgf15 mRNA expression (D) and colonic Ibabp mRNA expression (E). The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups; individual P values are provided in table S1.

Administration of an ASBTi restores glucose tolerance in HFD-fed mice

HFD feeding slightly increased final body weight, as compared to chow-fed mice, in agreement with the small increase in daily caloric intake (Fig. 2, A to C). However, the final body weights of the HFD/ASBTi16w mice were not different from those of the HFD mice, despite consuming more calories each day (Fig. 2, A to C, and fig. S1, G and H). Liver weight to body weight ratio was increased in the HFD mice versus chow-fed mice, which was still apparent in the 16-week HFD/ASBTi–treated group (Fig. 2D). Markers of hepatocellular injury, such as serum alanine aminotransferase and aspartate aminotransferase were not different between the groups, but a minor increase in serum alkaline phosphatase was observed in the HFD mice but not in the HFD/ASBTi16w mice (fig. S1I). To determine the effects of the ASBTi on glucose and systemic insulin sensitivity, we performed glucose and insulin tolerance tests. Feeding mice with HFD for 16 weeks resulted in glucose intolerance, whereas addition of an ASBTi to the HFD during this entire time restored glucose tolerance and improved systemic insulin sensitivity (Fig. 2, E and F, and fig. S2, A and B).

Fig. 2. Administration of an ASBTi restores glucose tolerance.

(A) Body weight gain in the indicated ad libitum–fed groups. * indicates value significantly different from WT. (B) Final body weight. (C) Average caloric intake over 16 weeks of ad libitum feeding. (D) Liver weight to body weight ratio after 16 weeks. (E) Glucose tolerance tests (GTTs). * indicates values significantly different between chow and HFD; # indicates values significantly different between chow and HFD/ASBTi4w. Individual P values are provided in table S1. (F) GTT area under the curve (AUC) (gram per liter per minute). The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups).

Administration of an ASBTi prevents hepatic steatosis in HFD-fed mice

After 16 weeks of feeding mice with the HFD, many characteristic histological features of NASH were observed in the animals’ livers, including steatosis, lobular inflammation, hepatocyte ballooning, and Mallory bodies (Fig. 3, A and B). Compared to HFD-fed mice, lipid deposition was markedly decreased in the HFD/ASBTi16w mice and to a lesser extent in HFD/ASBTi4w mice (Fig. 3, B to D). Blinded histologic assessment of slides by an experienced hepatopathologist (H.Y.) indicated that HFD/ASBTi16w treatment reduced the NAFLD activity score (NAS) (27) from 4.8 ± 1.0 to 2.8 ± 0.9, primarily driven by improvements in the steatosis score (Fig. 3, E and F; see table S2 for individual components of the NAS). Biochemical lipid measurements confirmed severe triglyceride (TG) and cholesteryl ester deposition in livers of HFD-fed mice, as compared to chow-fed animals (~12- and 15-fold increases, respectively; Fig. 3, G and H). The hepatic concentrations of TG, cholesteryl ester, and total cholesterol were markedly reduced in the HFD/ASBTi16w mice compared to those in the HFD-fed mice, to a degree whereby the liver lipid content in HFD/ASBTi16w mice was similar to the liver lipid content in chow-fed mice (Fig. 3, G and H, and fig. S2C). Only minor effects were noted for hepatic free cholesterol concentrations in diet and treatment groups when compared to the chow group (fig. S2D). Liver fibrosis was also assessed, but neither histologic assessment by Sirius Red staining nor biochemical quantitation of hydroxyproline content revealed any significant hepatic fibrosis in the HFD-fed mice. As such, there was little effect on these fibrosis markers in any of the treatment groups (fig. S3).

Fig. 3. Administration of an ASBTi reduces hepatic steatosis and accumulation of TGs and cholesterol.

(A to D) Hematoxylin and eosin–stained liver sections from (A) chow-, (B) HFD-, (C) HFD/ASBTi16w-, and (D) HFD/ASBTi4w-fed mice. (E) NAS. (F) Steatosis score. (G) Hepatic TG content. (H) Hepatic cholesteryl ester content. The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Different lowercase letters indicate significant differences between groups; individual P values are provided in table S1. Scale bars, 100 μm.

Administration of an ASBTi alters hepatic BA pool composition and FXR regulatory properties in HFD-fed mice

The total BA content was not different between chow, HFD, and ASBTi-treated livers (Fig. 4A), and HFD feeding had no discernible impact on hepatic BA pool composition when compared to chow (Fig. 4, B to D). However, blocking intestinal BA uptake altered the hepatic BA pool composition (Fig. 4B), changing its chemical and predicted FXR signaling properties. Ileal ASBT inhibition reduced the hepatic concentration of 6-hydroxylated TMCA species (ω, α, and β by > 50%) and increased the amount of the more hydrophobic BAs, TCDCA and TDCA, by greater than 10-fold (Fig. 4B). As a result, the hepatic BA content shifted from a typical ~0.8:1 balanced ratio of non–6-hydroxylated BA species to 6-hydroxylated BA species in chow-fed and HFD-fed mice to a >4:1 ratio with a predominance of the non–6-hydroxylated species in the livers of ASBTi mice (Fig. 4B). This is reflected by an increase in the calculated hydrophobicity index, which rose from −0.406 to +0.088 (Fig. 4C). In addition to changing the physicochemical properties of the hepatic BA pool, administration of the ASBTi also yielded a BA pool predicted to be, in sum, substantially more FXR agonistic (16, 2830) (Fig. 4, B and D). Because FXR signaling BAs are important physiological regulators of multiple hepatic functions including lipid metabolism, the combined amounts of FXR agonistic (TCA, TCDCA, TDCA, TLCA, and CA) and antagonistic (MCAs) BA species were compared to the hepatic TG and cholesterol content in livers of individual HFD, HFD/ASBTi16w, and HFD/ASBTi4w mice. Hepatic TG and cholesterol content directly correlated with the amount of FXR antagonistic muricholates (r = 0.4980, P = 0.0023; r = 0.5829, P = 0.0002) and inversely correlated with TCDCA+TDCA content (r = −0.5381, P = 0.0009; r = −0.5630, P = 0.0004) (fig. S4). Overall, the marked alteration of the hepatic BA pool toward FXR agonism correlated with the histologic and biochemical improvements driven by pharmacological inhibition of ileal ASBT function.

Fig. 4. Administration of an ASBTi shifts hepatic BA composition toward FXR agonism.

(A and B) Total hepatic BA content (A) and composition (B) (TMCA, tauromuricholic acid; MCA, muricholic acid, THDCA, taurohyodeoxycholic acid; TUDCA, tauroursodeoxycholic acid; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; TDCA, taurodeoxycholic acid; TLCA, taurolithocholic acid; CA, cholic acid). (C) Hydrophobicity index. (D) Pie charts for hepatic FXR antagonist/non-agonist (black) and agonist (white) BA content. (E) βTMCA inhibits TCA-induced FXR activation in HepG2 cells. (F) βTMCA inhibits TCDCA-induced FXR activation in HepG2 cells. The concentration of the individual BAs is indicated on the x axis. (G) Activation of an FXR reporter plasmid (pECRELuc) in transfected HepG2 cells by composite mixtures of the six major BAs (3 μM final concentration) modeling the in vivo hepatic BA content from the indicated groups. The vehicle (Veh) used was methanol, and positive control consisted of 3 μM TCDCA. The composition of the BA mixtures used in these in vitro studies is shown in table S3. The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups; individual P values are provided in table S1.

Previous in vitro studies using primarily individual BAs found TCDCA and TDCA to be potent FXR agonists, whereas TUDCA and TMCA species failed to activate FXR and may function as FXR antagonists (16, 3032). To begin modeling the FXR activity of the distinct hepatic BA pools in HFD and HFD/ASBTi mice, we examined the ability of BA mixtures to activate an FXR reporter plasmid in transfected human liver–derived HepG2 cells. In vitro, βTMCA antagonizes FXR activation by TCA or TCDCA (Fig. 4, E and F). However, given the in vivo findings, we thought it relevant to examine the in vitro FXR antagonistic properties of the muricholates in more complex mixtures of BAs, paralleling the composition of BAs present in liver. Mixtures of the six most abundant taurine-conjugated BAs were reconstituted at the ratios present in HFD or HFD/ASBTi livers (Fig. 4B and table S3) to model the in vivo exposure to multiple BAs and their combined effect on FXR activity. The BA composition present in HFD/ASBTi livers activated the FXR reporter plasmid to a greater extent than did the reconstituted BA mixture from HFD livers (Fig. 4G). These findings suggest that the consequences of this marked shift in liver BA composition may underlie some of the hepatic responses to ileal ASBT inhibition.

Administration of an ASBTi alters hepatic metabolic gene expression in HFD-fed mice

To explore the gene regulatory mechanisms underlying the hepatic effects of the luminally active ASBTi on HFD-fed mice, we performed RNA sequencing (RNA-seq) analyses using livers from mice fed with chow, HFD, and HFD/ASBTi16w. Comparison of the chow and HFD groups identified 5033 RNA species that were differentially expressed in livers of HFD-fed mice (3186 genes up-regulated and 1847 genes down-regulated, P < 0.05; Fig. 5A, fig. S5A, and table S4). Subsequent comparison of liver gene expression in HFD versus HFD/ASBTi16w mice identified 1483 differentially expressed genes (764 genes up-regulated and 719 down-regulated, P < 0.05; Fig. 5A and table S4). Cross-comparison of the lists of genes differentially expressed in chow versus HFD mice and in HFD versus HFD/ASBTi16w mice revealed 675 genes that are common in the two data sets. For 650 of these genes (96%), ASBTi treatment of the HFD mice changed their expression toward levels seen in chow-fed mouse livers, including several key genes involved in fatty acid and cholesterol metabolism (Srebp1c, Srebp2, Abcg5, Abcg8, and Cidec; Fig. 5, A and B, and table S4). Hierarchical clustering of RNA expression and ontological and pathway analyses revealed markedly distinct patterns between groups (Fig. 5, B and C, and fig. S5, B and C), confirming major changes in hepatic gene expression induced both by HFD and by concurrent ileal ASBT inhibition. Administration of the ASBTi changed hepatic gene expression in a variety of lipid, fatty acid, and cholesterol metabolic pathways, as well as intracellular transport and signaling (Fig. 5C and table S4). Together, these RNA-seq studies indicate marked shifts in hepatic gene transcription, intracellular signaling, and metabolism in HFD-fed mice as a consequence of ileal ASBT inhibition.

Fig. 5. Inhibition of ileal ASBT function alters global hepatic gene expression.

(A) Venn diagram of RNA-seq analysis of the two comparisons: HFD versus chow and HFD versus HFD/ASBTi16w. (B) Heat map comparing liver gene expression in HFD versus HFD/ASBTi16w groups. (C) Ontology analysis/pathway (Panther) analysis of HFD versus HFD/ASBTi16w as down-regulated (white bars) and up-regulated (black bars) pathways.

Real-time polymerase chain reaction (PCR) was used to measure the expression of critical genes in a variety of NAFLD-related pathways in the liver as well as in the ileum and colon, with the analyses focused primarily on HFD versus HFD/ASBTi16w mice (Fig. 6, A to C, and fig. S6, A to E). With regard to BA signaling, minimal effects were observed for several of the known hepatic FXR target genes including Abcb11 and Shp, whereas mRNA expression of the BA-activated G protein–coupled receptor (GPCR) S1pr2 and its downstream target Sphk2 were increased two- to threefold in HFD-fed mice and reduced with ASBT inhibition (Fig. 6A). Expression of FXR target genes was generally decreased in ileum and increased in colon (Fig. 1, D and E, and Fig. 6, B and C), reflecting the block in ileal BA absorption. Expression of the BA-activated GPCR Tgr5 was unaffected in liver, ileum, or colon by administration of the ASBTi (Fig. 6, A to C). There were no changes in ileal or colonic expression of the TGR5 target gene Gcg, the GLP-1 precursor (fig. S6, B and C).

Fig. 6. Administration of an ASBTi alters hepatic and intestinal gene expression.

(A) Hepatic mRNA expression of genes involved in BA signaling and transport, cholesterol synthesis, lipid synthesis, lipid droplet formation, fatty acid oxidation, inflammation, and fibrosis. (B and C) mRNA expression of genes involved in BA signaling and transport in the ileum (B) and (C) colon. Quantitative real-time PCR was performed on chow, HFD, HFD/ASBTi16w, and HFD/ASBTi4w groups, and expression shown relative to chow (set as 1). For clarity, gene expression in the HFD and HFD/ASBTi16w groups are displayed. Cyclophilin was used as a housekeeping gene to normalize mRNA expression. The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups; individual P values are provided in table S1.

The increases in hepatic BA synthesis with inhibition of the ileal ASBT resulted in the expected compensatory increase in expression of cholesterol biosynthetic genes (Srepb2 and Hmgr; Fig. 6A), and the reduced hepatic cholesterol content in HFD/ASBTi16w mice was associated with lower expression of Ch25oh (an endogenous source of the LXRα ligand, 25-hydroxycholesterol) and LXRα target genes (Abcg5, Abcg8, and Srebp1c). Inhibition of ileal ASBT also markedly altered the expression of hepatic genes involved in fatty acid and TG metabolism. Expression of a variety of key biosynthetic and regulatory genes (Srebp1c, Scd1, Fads1, Fads2, and Pparg) was increased two- to fourfold by HFD, and all were reduced toward chow-fed levels with ASBT inhibition (Fig. 6A and fig. S6D). Fatty acid oxidation–related gene expression was minimally affected by the ASBTi (Fgf21, Ppara, and Acaa1b; Fig. 6A and fig. S6D). In line with histological data and changes in lipogenic gene expression, RNA levels of the lipid droplet formation genes Cidea and Fsp27/Cidec were induced >50-fold by HFD and reduced by 70 ± 30% and 86 ± 10% with the ASBTi (Fig. 6A and fig. S6E).

The presence of inflammation and fibrosis with steatosis are hallmarks of progression to NASH, and HFD-induced proinflammatory gene expression (Tnf, Il1b, Ccl2, Cxcl9, and Cxcl10) was reduced in the presence of the ASBTi, whereas mRNA expression of Lcn2, a small secreted adipokine with a protective role in hepatic inflammation, was induced (Fig. 6A and fig. S6E). RNA expression of fibrosis genes Col1a1, Msln, and Gli2 were increased by HFD and reduced to baseline with ASBT inhibition, suggesting an antifibrotic effect of the ASBTi (Fig. 6A and fig. S6E), although there was no significant fibrosis detected in any of the experimental conditions in this study (fig. S3). Together, 16 weeks of ileal ASBT inhibition resulted in substantial alterations in gene expression in livers of HFD-fed mice, which is related principally to BA, cholesterol, and fatty acid metabolism, but also to a reduction in markers of inflammation and response to stress.

Administration of an ASBTi does not alter ceramides in HFD-fed mouse ileum or liver

Recent studies in a different HFD-fed mouse model (8, 18) support a role for ileal FXR signaling–associated changes in ceramide synthesis and secretion into the portal circulation in the development of fatty liver and obesity-related metabolic dysfunction. However, targeted mass spectrometry analysis of both liver and ileal ceramide concentrations in response to HFD or HFD/ASBTi16w revealed no significant changes in the amounts of total ceramide or individual ceramide species in either tissue (fig. S7). These findings suggest that different mechanisms are operative in this model, wherein the ASBTi blocks both ileal FXR signaling and ileal BA internalization.

Administration of an ASBTi partially reverses established NAFLD after 12 weeks of HFD

To determine whether inhibition of ileal BA uptake also affects established NAFLD, mice were fed the HFD for 12 weeks, followed by 4 weeks of HFD/ASBTi. The responses of sentinel FXR-responsive genes in the liver, intestine, and colon were similar in the 4- and 16-week ASBTi treatments (Fig. 1, C to E). This short treatment period induced some changes not seen in the full 16-week treatment, such as a decrease in the liver weight to body weight ratio (Fig. 2D), but glucose tolerance, NAS, and steatosis score were not significantly affected (Fig. 2, E and F, and Fig. 3, E and F) in HFD/ASBTi4w when compared to those in HFD mice. However, hepatic TG and cholesterol contents were markedly reduced compared to HFD-fed mice (Fig. 3, G and H, and fig. S2C), and visually, there appeared to be zonally restricted reductions in lipid deposition with the HFD/ASBTi4w treatment (Fig. 3, B and D). In addition, hepatic BA composition was similarly altered with 4 weeks of ASBTi as with 16 weeks of treatment (Fig. 4, B to D). Overall, these results indicate a partial reversal of the hepatic effects of 12 weeks of HFD by ASBTi treatment during the ensuing 4 weeks of HFD.

Asbt−/− mice are resistant to hepatic steatosis with short-term HFD feeding

To confirm these findings and explore potential mechanisms involving the role of ASBT in the hepatic response to HFD, short-term (1 week) HFD was administered to Asbt−/− and wild-type (WT) (Asbt+/+) mice followed by histologic, biochemical, and molecular analyses of the liver and ileum. After 1 week of HFD, when compared to WT mice, Asbt−/− mice gained less weight, despite ingesting essentially the same amount of calories as WT mice (Fig. 7, A and B). Similarly, liver weight in HFD-fed Asbt−/− mice was reduced compared to HFD-fed WT mice (Fig. 7C), but liver weight to body weight ratios were unchanged (Fig. 7D). Histological and biochemical analyses showed that the livers of HFD-fed Asbt−/− mice, but not the livers of WT mice, were protected against lipid accumulation after 1 week of HFD (Fig. 7, E to L).

Fig. 7. Hepatic accumulation of TGs and cholesterol is reduced in Asbt−/− mice.

(A) Body weight gain in the indicated ad libitum–fed groups. * indicates value significantly different from WT. (B) Average caloric intake over 1 week of feeding. (C) Liver weight after 1 week of HFD or chow. (D) Liver weight to body weight ratio after 1 week of HFD or chow. (E to H) Hematoxylin and eosin–stained liver sections from (E) WT/chow, (F) WT/HFD 1 week, (G) Asbt−/−/chow, and (H) Asbt−/−/HFD 1 week mice. (I) Hepatic TG content. (J) Hepatic total cholesterol content. (K) Hepatic free cholesterol content. (L) Hepatic cholesteryl ester content. The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups; individual P values are provided in table S1. Scale bars, 100 μm.

Asbt−/− mice have reduced total hepatic BAs, with a BA composition similar to those treated with ASBTi

Unlike in mice treated with the luminally restricted ASBTi, total hepatic BA content was reduced in Asbt−/− mice (by ~50%; Fig. 8A). However, the hepatic BA compositional changes were similar in both models (Figs. 4B and 8B), shifting toward a more hydrophobic, FXR agonistic pool (Fig. 8, C and D). Gene expression consequences of 1 week of HFD in Asbt−/− mouse livers were notable for changes similar to those seen in the longer-term ASBTi treatments (Fig. 8E and fig. S8). Notably for BA and cholesterol metabolism, Cyp7a1 mRNA was increased approximately fivefold, expression of BA-regulated genes (Shp, Abcb11, and Mafg) was reduced in HFD-fed Asbt−/− mice, and the down-regulation of Hmgr and Srebp2 mRNA observed in the HFD-fed WT mice was blocked in HFD-fed Asbt−/− mice. Increases in lipid metabolism–associated genes (Abcg8, Lxra, Srebp1c, Scd1, Ppara, and Ehhadh) in HFD-fed WT mice were abrogated in HFD-fed Asbt−/− mice (Fig. 8E and fig. S8), comparable to changes induced by ASBT inhibition. Immunoblot analysis showed that the amount of liver SREBP1 protein increased after 1 week of HFD feeding in WT mice (P = 0.0084) but not in Asbt−/− mice (P = 0.913; Fig. 8F). Thus, absence of ASBT function markedly impairs the accretion of features of NAFLD in mice in as short a period as 1 week of HFD and engages pathways associated with a more FXR agonistic hepatic BA pool. Together, the alterations in liver BA composition, ileal and colonic gene expression, hepatic TG and cholesterol content, and resolution of insulin resistance support inhibition of ileal ASBT function as a rational target for the hepatic and metabolic consequences of diet-induced fatty liver.

Fig. 8. Hepatic BA composition is altered in Asbt−/− mice.

(A) Total hepatic BA content and (B) BA composition. (C and D) Hydrophobicity index (C) and pie charts (D) for hepatic FXR antagonist/non-agonist (black) and agonist (white) BA content. (E) Hepatic gene expression in WT and Asbt−/− mice fed a HFD for 1 week. Values shown are relative to chow-fed WT mice. (F) Protein expression of full-length (fl) SREBP1 increased in HFD-fed WT mice but not in HFD-fed Asbt−/− mice. SREBP1 expression was normalized to the amount of the housekeeping protein glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and the quantitation is shown below the blot. The labeling scheme for each group is indicated in the embedded legend. Means + SD are shown. Distinct lowercase letters indicate significant differences between groups; individual P values are provided in table S1.

DISCUSSION

In the current study of a mouse model of HFD-induced NASH, the presence of a luminally restricted specific inhibitor of intestinal BA uptake (ASBTi) prevented hepatic accumulation of TGs and cholesterol and restored whole-body insulin sensitivity. Ileal ASBT inhibition in HFD-fed mice was associated with substantial hepatic transcriptional and metabolic reprogramming, suggesting that hepatocellular BA signaling was markedly altered by this gut-restricted pharmacological intervention. Ileal ASBT inhibition prevented hepatic cholesterol accumulation by inducing BA biosynthesis from cholesterol stores and shifting the liver BA pool composition toward one enriched in hydrophobic species that are more FXR agonistic in nature. This included a dramatic reduction in hydrophilic, FXR antagonistic tauromuricholates (16, 30) associated with increases in TCDCA, TDCA, and other FXR agonistic BAs. The reduction in muricholates is likely due to decreases in chenodeoxycholic acid synthesis via the alternative pathway, secondary to reductions in hepatic cholesterol content and to suppression of hepatic 6-hydroxylation. Hepatic TG and cholesterol content directly correlated with the amount of FXR antagonistic muricholates and inversely correlated with the amount of FXR agonist TCDCA+TDCA in the liver. Similar decreases in hepatic lipid accretion were observed in Asbt−/− mice versus WT mice with short-term HFD feeding and in the hepatic BA composition of the Asbt−/− mice, regardless of diet. Exposure of transfected HepG2 cells to synthetic BA mixtures modeling the hepatic BA composition found in ileal ASBT-inhibited mice increased the activation of an FXR luciferase reporter plasmid, as compared to the hepatic BA composition of HFD mice. Together, these findings suggest that inhibition of ileal BA uptake has meaningful biological and gene regulatory consequences that address multiple features of NASH in mice.

Ileal ASBT inhibition resulted in profound reduction in hepatic lipids, improvement in insulin sensitivity, and alterations in the expression of hepatic lipid, inflammatory, and BA regulatory genes. The underlying mechanisms may relate to alterations in BA, FXR, or LXR signaling or other pathways. Recent works by Jiang et al. (8, 18) demonstrated that intestinal-specific inactivation or inhibition of FXR-produced metabolic benefit by reducing the intestinal production of bioactive ceramides, which promote fatty liver and obesity-related metabolic dysfunction in HFD-fed mice. Here, inhibition of ileal FXR activation was associated with metabolic benefit without a change in ileal or hepatic ceramide content, suggesting alternative mechanisms leading to improvements in NASH. These differential results may be related to consequences of specifically targeting FXR versus BA-mediated signaling. For example, with regard to the ileal enterocyte, inhibition or inactivation of ASBT blocks BA internalization and is predicted to cause pan-inhibition of FXR as well as other intercellular BA targets and signaling pathways. This is in contrast to intestinal-specific inactivation of FXR or administration of intestinally selective FXR antagonists, both of which would block only the FXR signaling component. A relevant confounder to focusing on inhibition of intestinal FXR signaling to improve NASH is a recent work by Fang et al. (17), which demonstrated improved lipid handling and NASH in mice with a gut-restricted FXR agonist. Thus, with regard to NASH in mice, the beneficial effects associated with modifying BA signaling in ileal enterocytes may involve FXR and non-FXR targets as well as ceramide and nonceramide pathways, whose relative contributions remain to be determined.

Several metabolic pathways are involved in maintaining hepatic lipid homeostasis, whereby dysregulation of each can result in accumulation of lipid droplets. It is well established that lipid and TG syntheses are increased in NAFLD (33). In addition, changes in BA composition can elicit major effects on regulation of metabolic pathways in liver (14, 30, 34, 35). Hepatic expression of genes involved in lipid synthesis and lipid droplet formation were induced after 16 weeks of HFD and reduced in the presence of the ASBTi. In contrast, there was little effect on gene expression for β-oxidation or lipoprotein secretion pathways in this study, suggesting that they did not play a major role in the changes in hepatic TG content in this model. Note that the modified ALIOS diet used in this study contained a modest amount of added [0.2% (w/w)] cholesterol, and cholesterol accumulation is a likely contributor to NASH (3640). Cholesterol accumulation may cause increases in hepatic inflammation and oxysterol-mediated alterations in gene expression, mostly through LXRα target genes including Srebp1c. Additional studies are needed to explore the hepatic consequences of reducing cholesterol and cholesterol-derived metabolites (such as oxysterols) and to determine whether this is an important determinant of the improvements in response to an ASBTi. Thus, the mechanisms mediating the effects of ASBT inhibition may extend beyond FXR-mediated signaling in animal models of NAFLD.

Although adding an ASBTi for the final 4 weeks of the 16-week HFD treatment altered the hepatic BA composition, it did not produce a histological improvement in NASH. The effects were intermediate, with similar decreases in hepatic TG and cholesterol accumulation as observed with the 16-week treatment, but not some of the other metabolic and transcriptional effects. It is not clear whether this was a result of an insufficient duration of treatment or whether the changes in BA metabolism and signaling induced by the ASBTi are insufficient to affect NASH once it is established. Additional studies using longer periods of intervention and other models of NASH will be required to answer these questions.

In summary, treatment of mice with a luminally restricted ileal ASBTi markedly improved several facets of NASH and insulin resistance by engaging multiple components of BA signaling within the gut-liver axis. The finding that hepatic BA composition became more FXR agonistic in response to inhibition of ileal ASBT is likely to be mechanistically important but may explain only a portion of the positive effects of ASBT inhibition observed in this mouse NASH model. There is also the question of whether these BA compositional changes are relevant for humans, who largely lack 6-hydroxylated BAs under most physiological conditions. However, the targeting of ileal BA uptake to alter liver FXR agonism may alleviate some of the concerns seen in the recent human NASH trial of a potent FXR agonist, OCA, which not only improved the histological features of NASH, but also increased pro-atherogenic serum lipid concentrations in some subjects (6). Both surgical ileal exclusion and interruption of ileal BA uptake by pharmacological means have been shown to reduce serum low-density lipoprotein cholesterol by increasing hepatic cholesterol conversion to BAs (22, 41), so it is intriguing to speculate that inhibition of ASBT may provide a means to improve features of hepatic lipotoxicity and metabolic syndrome while alleviating some atherosclerotic risk factors. Mechanistic preclinical studies, in addition to well-controlled human trials, are needed to determine whether these data from a mouse NASH model will translate to humans affected by NASH and to verify the long-term effects of such an intervention.

MATERIALS AND METHODS

See the Supplementary Materials for Materials and Methods.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/357/357ra122/DC1

Materials and Methods

Fig. S1. Inhibition of ileal BA absorption does not affect intestinal and colonic gene expression, food and water consumption, and hepatic injury markers.

Fig. S2. Inhibition of ileal BA absorption improves whole-body insulin tolerance and hepatic cholesterol concentration.

Fig. S3. Inhibition of ileal BA absorption does not affect hepatic fibrosis.

Fig. S4. Inhibition of ileal BA absorption changed hepatic BA composition and correlated with TG and cholesterol concentrations.

Fig. S5. Inhibition of ileal BA absorption changed global hepatic gene expression during 16 weeks of HFD feeding.

Fig. S6. Inhibition of ileal BA absorption changed gene expression in the mouse liver.

Fig. S7. Inhibition of ileal BA absorption does not affect hepatic and ileal ceramide amount and composition.

Fig. S8. Inhibition of ileal BA absorption affects hepatic gene expression in WT and Asbt−/− mice fed with chow or HFD for 1 week.

Table S1. P values for all comparisons (provided as an Excel file).

Table S2. Individual elements of NAS.

Table S3. Composite BA mixtures of six BAs representing the major in vivo hepatic BA species in HFD and HFD/ASBTi16w mice.

Table S4. Differentially expressed genes and pathway analysis (provided as an Excel file).

Table S5. HFD composition and macronutrient information.

Table S6. Primer sequences used for real-time PCR analysis.

Table S7. Primary data (provided as an Excel file).

References (4253)

REFERENCES AND NOTES

  1. Acknowledgments: We thank Lumena/Shire for the SC-435; D. Shayakhmetov and N. C. Di Paolo (Emory University School of Medicine) for assistance and use of their microscope; the Yerkes Nonhuman Primate Genomics Core (Emory University School of Medicine) for performing RNA-seq analysis; the Children’s Healthcare of Atlanta Pathology Services for processing histology tissues; T. Osborne and P. Phelan (Sanford Burnham Prebys Medical Discovery Institute) for assistance with SREBP1 immunoblotting. Funding: This study was supported by the NIH [DK56239 (S.J.K.); DK047987 (P.A.D.); ES023485, AG038746, HL113451, ES019776, ES025632, EY022618, HD075784, HL095479, HL086773, and HL125042 (D.P.J.)]. J.E.M. was supported by an NIH Fellowship in Research and Science Teaching (FIRST) Institutional Research and Academic Career Development Award (K12 GM000680). S.J.K. and P.A.D. were also supported by the Center for Transplantation and Immune-mediated Disorders and the Children’s Healthcare of Atlanta. Author contributions: A.R., A.K., J.E.M., W.Z., K.D.R.S., T.X., B.T.K., H.Y., G.M.W., A.M.A., S.B., D.P.J., and H.W. performed the experiments and collected and analyzed the data; S.J.K. and P.A.D. managed and designed the study and conceived the experiments; A.K., A.R., A.M.A., P.A.D. and S.J.K. wrote the manuscript. Competing interests: B.T.K. was an employee of Lumena Pharmaceuticals, which is now part of Shire Pharmaceuticals. P.A.D. had previously served as a consultant for Lumena Pharmaceuticals. The other authors declare that they have no competing interests. Data and materials availability: Asbt−/− mice are available for purchase through the Jackson Laboratory. SC-435 was made available as a research gift from Lumena/Shire Pharmaceuticals. RNA-seq data are available through Gene Expression Omnibus (GSE84994).
View Abstract

Navigate This Article