|Home | About | Journals | Submit | Contact Us | Français|
Blocking intestinal bile acid (BA) absorption by inhibiting or inactivating the apical sodium-dependent BA transporter (Asbt) classically induces hepatic BA synthesis. In contrast, blocking intestinal BA absorption by inactivating the basolateral BA transporter, organic solute transporter alpha–beta (Ostα–Ostβ) is associated with an altered homeostatic response and decreased hepatic BA synthesis. The aim of this study was to determine the mechanisms underlying this phenotype, including the role of the farnesoid X receptor (FXR) and fibroblast growth factor 15 (FGF15).
BA and cholesterol metabolism, intestinal phenotype, expression of genes important for BA metabolism, and intestinal FGF15 expression were examined in wild type, Ostα−/−, Fxr−/−, and Ostα−/−Fxr−/− mice.
Inactivation of Ostα was associated with decreases in hepatic cholesterol 7α-hydroxylase (Cyp7a1) expression, BA pool size, and intestinal cholesterol absorption. Ostα−/− mice exhibited significant small intestinal changes, including altered ileal villus morphology, and increases in intestinal length and mass. Total ileal FGF15 expression was elevated almost 20-fold in Ostα−/− mice as a result of increased villus epithelial cell number and ileocyte FGF15 protein expression. Ostα−/−Fxr−/− mice exhibited decreased ileal FGF15 expression, restoration of intestinal cholesterol absorption, and increases in hepatic Cyp7a1 expression, fecal BA excretion, and BA pool size. FXR deficiency did not reverse the intestinal morphological changes or compensatory decrease for ileal Asbt expression in Ostα−/− mice.
These results indicate that signaling via FXR is required for the paradoxical repression of hepatic BA synthesis but not the complex intestinal adaptive changes in Ostα−/− mice.
Regulation of hepatic expression of Cyp7a1, the rate-limiting enzyme for the classical BA biosynthetic pathway, is complex and integrates responses to hormones, cytokines, growth factors, oxysterols, BA, xenobiotics, and diurnal rhythm . It is well established that the nuclear receptor FXR is essential for negative feedback regulation of Cyp7a1 by BA [2,3]. However, a critical role for gut-liver signaling via the endocrine polypeptide hormone FGF15 (human ortholog: FGF19) has only recently been appreciated [4,5]. In that pathway, BA act as ligands for FXR in ileal enterocytes to induce synthesis of FGF15. After its release into the circulation, FGF15 acts on hepatocytes through its cell surface receptor, a complex of β-Klotho protein and fibroblast growth factor receptor-4 (FGFR4), to repress Cyp7a1 expression and BA synthesis [4-6].
After their hepatic synthesis and secretion into the small intestine, BA are efficiently reabsorbed by active transport in the ileum and transported back to the liver . In the ileum, BA are transported across the apical brush border membrane via the Asbt (gene symbol Slc10a2) , whereas the heteromeric transporter Ostα–Ostβ (gene symbols: Ostα, Slc51a1; Ostb, Slc51a1bp) is responsible for basolateral membrane transport [8-10]. Resection of terminal ileum, ileal disease, or inactivation of Asbt in humans or mice interrupts the enterohepatic circulation of BA and results in increased fecal levels of BA, induction of hepatic Cyp7a1 expression and BA synthesis, and partial depletion of the BA pool [11-14]. Although ileal BA transport is also impaired in Ostα−/− mice, these animals do not exhibit the classical response to intestinal BA malabsorption [9,10]. The BA phenotype of Ostα−/− mice is more similar to that observed in Asbt−/− mice treated with an FXR agonist or an FGF15-expressing adenovirus . Those observations provide a possible explanation for the phenotype in Ostα−/− mice whereby blocking basolateral export causes BA to accumulate in ileal enterocytes, leading to persistent activation of FXR, increased FGF15 expression, and the unexpected repression of hepatic BA synthesis .
Portions of this work were presented at the 2010 Annual Meeting of the American Association for the Study of Liver Disease and the 21st International Bile Acid Meeting entitled “Falk Symposium 175: Bile Acids as Metabolic Integrators and Therapeutics” and published in abstract and meeting proceedings form [15,16].
The Institutional Animal Care and Use Committee approved these experiments. Fxr−/− mice (Jackson Laboratory Stock#: 004144, Strain: B6;129X(FVB)-Nr1h4tm1Gonz/J) were bred with Ostα−/− mice (C57BL/6J-129/SvEv)  to generate the required genotypes. Unless indicated, the mice were fed ad libitum standard rodent chow. In selected studies, mice were fed a basal diet  or a basal diet supplemented with 0.2% (w/w) cholic acid (CA) (Sigma–Aldrich). For analysis of ileal FGF15 protein expression, groups of non-fasted WT mice were treated with a single oral gavage of 100 mg/kg body weight of the FXR agonist, GW4064 (a gift from Dr. Timothy Willson, GlaxoSmithKline), or vehicle (Polyethylene glycol 400/Tween 80; 4:1, v/v). FGF15 protein expression was also examined in ileal extracts from WT and Asbt−/− mice maintained on the basal diet  and WT and Fgf15−/− mice (a gift from Dr. Steve Kliewer, University of Texas Southwestern Medical Center) .
The small intestine was subdivided into five equal length segments and processed for histology. Villus height and width, crypt depth, and muscle thickness were measured for all intestinal segments in at least 20 well-oriented, full-length villus units per mouse. Quantitative analyses were performed by AxioVision analysis of digitally acquired images.
Feces were collected to measure total BA content by enzymatic assay and neutral sterol content by gas–liquid chromatography. Intestinal cholesterol absorption was measured using the fecal dual-isotope ratio method; BA pool size was determined as the BA content of small intestine, liver, and gallbladder . BA composition was determined by HPLC  and used to calculate the pool hydrophobicity . Plasma BA were determined by enzymatic assay (Bio-Quant Inc.) and found to be similar between the different groups (WT, 52 ± 4 μM; Fxr−/−, 67 ± 10 μM; Ostα−/−, 52 ± 7 μM; Ostα−/−Fxr−/−, 66 ± 13 μM; n = 3–5 male mice per group). To measure ileal-associated BA, the small intestine was divided into five equal segments, flushed with PBS, and ground under liquid nitrogen using a mortar and a pestle. Aliquots of ileal tissue were then used to isolate RNA or extracted to measure tissue-associated BA . Plasma and hepatic levels of total cholesterol, free cholesterol, and triglyceride were determined by enzymatic assay (Roche Applied Science) .
Small intestinal DNA and RNA content was measured using modifications  of the diphenylamine and orcinol colorimetric methods, respectively. Total RNA was extracted from frozen tissue using TRIzol Reagent (Invitrogen). Real time PCR analysis was performed as described ; values are means of triplicate determinations and expression was normalized using cyclophilin. The primer sequences used are provided (Supplementary Table 1). Whole intestinal segment mRNA expression was calculated by normalizing to total RNA content. Unless otherwise indicated, expression levels were plotted relative to WT mice.
A rabbit polyclonal antibody was raised against a glutathione-S-transferase fusion protein (Amersham Biosciences) encompassing amino acids 171–218 of mouse FGF15, and purified by affinity chromatography using the same antigen coupled to agarose beads (Amino-Link Immobilization Kit; Pierce). Sources of other antibodies were as follows: anti-mouse Asbt , anti-mouse apoA-I , anti-β-actin (Sigma–Aldrich, A5441), HRP-conjugated anti-rabbit (Sigma–Aldrich, A9169; or Cell Signaling, 7074), HRP-conjugated anti-mouse (GE Healthcare, NXA931). Small intestinal extracts were prepared, subjected to SDS–PAGE using 8% or 4–12% gradient (Bis-Tris Midi Gel, Invitrogen) polyacrylamide gels, and analyzed by immunoblotting . Blots were stripped and reprobed with antibody to β-actin to normalize for protein load. Protein expression was quantified by densitometry using a Microtek ScanMaker i900 and FujiFilm Multiguage 3 software, and expression data was normalized to levels of the β-actin loading control.
Mean values ± SE are shown unless otherwise indicated. The data were evaluated for statistically significant differences using the two-tailed Student’s t test or by ANOVA (Tukey–Kramer honestly significant difference) (Statview; Mountain View, CA). Differences were considered statistically significant at p <0.05 and are indicated by different lowercase letters in the figures.
Deletion of FXR in Ostα−/− mice resulted in a phenotype typically observed following a block in intestinal BA absorption. Fecal BA excretion was similar between Ostα−/− and WT mice, and induced by 2 to 3-fold in Ostα−/−Fxr−/− mice (Fig. 1A; Supplementary Fig. 1A). The BA pool size was increased approximately 1.8-fold in Ostα−/−Fxr−/− vs. Ostα−/− mice (Fig. 1B; Supplementary Fig. 1B), but did not reach the levels in WT or Fxr−/− due to increased fecal BA loss. The BA pool also became enriched in taurocholate and more hydrophobic in Ostα−/−Fxr−/− vs. Ostα−/− mice, as reflected by an increase in the hydrophobicity index. The TβMC:TC ratio was also increased (Ostα−/− vs. Ostα−/−Fxr−/−; males: 0.42 ± 0.07 vs. 1.48 ± 0.20; females 0.35 ± 0.03 vs. 0.70 ± 0.10) (Fig. 1B and C; Supplementary Fig. 1B and C), suggesting a derepression of hepatic BA synthesis via the Cyp7a1 pathway.
Since intestinal cholesterol absorption is greatly affected by luminal BA concentration and BA pool hydrophobicity [14,23], fecal neutral sterol excretion and cholesterol absorption were also examined in the different genotypes. Fecal neutral sterol excretion was increased in Ostα−/− mice, reflecting a dramatic reduction in intestinal cholesterol absorption (Fig. 1D and E; Supplementary Fig. 1D). This phenotype was largely reversed by introducing FXR deficiency (Fig. 1E; Supplementary Fig. 1D) or by feeding a diet containing a more hydrophobic BA (0.2% CA) (Fig. 1F; Supplementary Fig. 1E). Plasma cholesterol and triglyceride levels were not significantly altered in male Ostα−/− mice; FXR deficiency tended to increase plasma cholesterol and triglyc-eride levels, as observed previously , and this effect was maintained in the Ostα−/− background (Supplementary material and methods). Hepatic cholesteryl ester (CE) content was increased approximately 2.5-fold in Ostα−/− mice (Fig. 1G), in agreement with the predicted decrease in hepatic conversion of cholesterol to BA. In contrast to Asbt−/− mice , hepatic cholesteryl ester content was not decreased in Ostα−/−Fxr−/− mice. Hepatic triglyceride levels tended to be higher in Ostα−/− but not Ostα−/−Fxr−/− mice (Fig. 1H).
Ostα−/− mice were indistinguishable from adult WT or Fxr−/− mice with regard to body weight or liver weight (data not shown), however, there were significant small intestinal changes that were not reversed by inactivation of FXR (Fig. 2). The small intestine was significantly heavier in both Ostα−/− and Ostα−/−Fxr−/− mice vs. WT mice. Since Ostα–Ostβ exhibits a gradient of expression along the small intestine length with the highest levels in ileum, the segmental distribution for the intestinal mass was also examined. In agreement with this pattern of expression, significant mass changes were evident in the distal small intestine of Ostα−/− mice, where intestinal weight per unit length was increased approximately 66% (Supplementary Fig. 2A). Similar changes were observed in female mice (data not shown). The ileal content of DNA and RNA was increased approximately 2-fold in Ostα−/− mice as compared with WT mice (Supplementary Fig. 2B and C), reflecting an increase in cell number. These small intestinal changes persisted in Ostα−/−Fxr−/− mice (Supplementary Fig. 2).
In Ostα−/− and Ostα−/−Fxr−/− mice, ileal but not jejunal morphology was significantly altered (Fig. 2C and D), where the villi were blunted and thickened (fused), with evidence of enterocyte dysplasia and occasionally dilated lacteals. Morphometric measurements were performed using histological sections of small intestine from the four genotypes. Muscle thickness was unchanged, but villi in ileum of Ostα−/− and Ostα−/−Fxr−/− mice were shorter and wider, with deeper crypts (Supplementary Fig. 3). There was a slight increase in lymphocytes and plasma cells in the lamina propria of Ostα−/− mice (Fig. 2C), but no associated increases in expression of selected pro-inflammatory genes or genes involved in ER stress (data not shown).
Hepatic Cyp7a1 mRNA expression was decreased by 74% in Ostα−/− mice and increased 2-fold in Ostα−/−Fxr−/− mice vs. WT mice (Table 1). Effects of Ostα inactivation were greater for hepatic Cyp7a1 than Cyp8b1 mRNA expression, consistent with previous findings demonstrating that Cyp7a1 is more strongly regulated by FGF15 signaling . Analysis of the hepatic mRNA expression of other genes involved in hepatic bile acid or cholesterol metabolism revealed an increase in FXRα3,4 isoforms in Ostα−/− mice. However, there were few other significant changes with the exception of decreases in MRP2 and SR-BI, which were not reversed by inactivation of FXR (Table 1). The decrease in hepatic Cyp7a1 expression was predicted to be secondary to an induction of ileal FGF15 expression . Measurements revealed no significant change in the ileocyte FGF15 mRNA (normalized to the housekeeping gene cyclophilin) but total ileal FGF15 mRNA expression (normalized for increased ileal RNA content secondary to intestinal morphological changes) was increased approximately 2-fold in Ostα−/− mice (Fig. 3A). The relatively small change in total FGF15 mRNA levels prompted us to examine FGF15 protein expression (Fig. 3B and C). As a control, FGF15 protein expression was determined under conditions known to alter ileal FGF15 mRNA expression [4,13]. As predicted, ileal FGF15 protein levels were increased in GW4064-treated WT mice, reduced in Asbt−/− mice, and undetectable in Fgf15−/− vs. WT mice (Fig. 3B). Remarkably, immunoblotting revealed a more than 10-fold increase for ileal FGF15 protein expression in Ostα−/− mice (Fig. 3C). There was no accompanying change in the intestinal gradient for FGF15 expression in Ostα−/− mice, nor does this increase appear to be due to a non-specific protein retention or delayed protein secretion by the ileal enterocytes, since ileal apo A-I protein levels were not increased (Supplementary Fig. 4). Accounting for the elevated ileal protein content of Ostα−/− vs. WT mice, total ileal FGF15 protein is increased almost 20-fold (Fig. 3C). The mRNA and protein expressions of FGF15 were both dramatically reduced in ileum of Fxr−/− and Ostα−/−Fxr−/− mice.
The surprising finding that ileocyte FGF15 mRNA expression (normalized to expression of the housekeeping gene cyclophilin) was not increased in Ostα−/− mice (Fig. 3A) prompted an examination of other FXR target genes. Ileal expression of Ibabp, Ostβ, PXR, Shp, and sodium-sulfate co-transporter (Slc13a1) was found to be reduced rather then increased in Ostα−/− mice (Table 2). A variety of mechanisms could modulate ileal expression of FXR target genes in Ostα−/− mice including: (1) altered FXR expression or activity, (2) decreased BA uptake by the ileal enterocyte, and (3) altered BA pool size or composition. Measurements of ileal FXR mRNA levels revealed significant decreases of individual FXR isoform and total FXR expression (Table 2). Another possible adaptation in Ostα−/− mice is a decreased BA uptake. As shown in Fig. 4, ileal Asbt protein expression is reduced almost 80% in both Ostα−/− and Ostα−/−Fxr−/− vs. WT mice. The BA pool’s decreased size and enrichment in muricholic acid, a poor FXR ligand , may also contribute to the reduced expression of FXR target genes in Ostα−/− mice. To examine this possibility, mice were fed a diet containing a 0.2% CA, an active FXR ligand, and ileal BA accumulation and gene expression were measured (Fig. 4C–F). Expression of ileal FXR target genes Ibabp and FGF15 was reduced in Ostα−/− mice vs. WT mice (Fig. 4E and F). Their reduced expression correlated with a decreased ileal BA content, which is likely to be secondary to reduced Asbt expression. Inactivation of Ostα was also associated with reduced expression of other non-FXR target genes such as ABCA1, ABCG5, HNF4α, and NPC1L1. Interestingly, their expression was not restored to WT levels in Ostα−/−Fxr−/− mice.
The major finding of this study is that FXR is required for the unanticipated changes in BA homeostasis but not the complex intestinal morphological changes observed in Ostα−/− mice. Moreover the original model , which proposed that the unexpected repression of hepatic BA synthesis in Ostα−/− mice resulted from ileal enterocyte BA accumulation and increased expression of FXR-target genes, must be revised. The phenotypic changes in Ostα−/− mice are summarized in Table 3.
Analysis of Ostα−/− mice revealed significant histological and morphometric changes in ileum, including villi that were consistently blunted and fused. These changes, which are typically associated with damage and subsequent healing, have not been reported for patients or mouse models with primary BA malabsorption, a defect in apical brush border membrane transport [14,27]. The adult Ostα−/− mice do not exhibit symptoms of persistent on-going intestinal injury, as indicated by the absence of elevated inflammatory gene expression, bleeding, or diarrhea. However, newborn Ostα−/− mice exhibit a postnatal growth deficiency [9,10], and this may coincide with the onset of injury or initiation of the adaptive response. Although the underlying mechanisms and role of BA in this process remain to be determined, the finding of a similar intestinal phenotype in Ostα−/− and Ostα−/−Fxr−/− mice argues against a role of the FXR-FGF15 pathway.
The classical physiological response to interruption of the enterohepatic circulation is increased hepatic BA synthesis [13,14,28,29]. However, despite evidence of a block in intestinal BA absorption, Ostα−/− mice have decreased hepatic Cyp7a1 expression and no increase in fecal BA excretion [9,10]. In order to further understand how inhibiting BA transport at the enterocyte apical vs. basolateral membrane produces such different hepatic responses, the contribution of the FXR-FGF15 pathway to this unexpected BA phenotype was investigated. Inactivation of FXR in Ostα−/− mice decreased ileal FGF15 expression and Ostα−/−Fxr−/− mice largely recapitulated the classical BA malabsorption phenotype. In Ostα−/−Fxr−/− mice, the BA composition and the intestinal cholesterol absorption were restored to near WT levels, consistent with the superior ability of TC vs. TβMC to solubilize and deliver cholesterol to the intestinal epithelial cells . The decreased intestinal cholesterol absorption in Ostα−/− mice is balanced by reduced cholesterol demand for hepatic BA synthesis. As a result, hepatic cholesteryl ester levels are only mildly elevated and plasma cholesterol levels are unchanged in male Ostα−/− mice. Interestingly, there were few other significant changes in the hepatic expression of genes involved in BA homeostasis in Ostα−/− mice and these changes were not reversed by inactivation of FXR.
Another goal of this study was to identify additional protective mechanisms engaged in Ostα−/− mice. Surprisingly, inactivation of Ostα did not lead to ileal BA accumulation. The mechanisms responsible for the decreased BA accumulation and reduced expression of FXR target genes appear to be multi-fold and may include decreased ileal expression of FXR . Changes in BA pool size and composition are unlikely to be major factors since expression of FXR target genes remained lower in Ostα−/− mice fed a CA-containing diet. However, a major protective mechanism in Ostα−/− mice appears to be reduced Asbt expression . This response does not require FXR, nor does it appear to be due to FGF15 acting in a paracrine fashion , since Asbt expression remains suppressed in the Ostα−/−Fxr−/− mice where FGF15 expression is markedly reduced. Another surprising finding was that the increase in ileal FGF15 production in Ostα−/− mice results from a combination of elevated FGF15 protein expression and increased number of FGF15-expressing enterocytes.
These results affirm the critical role for Ostα–Ostβ in BA absorption and metabolism and the importance of FGF15 in regulating hepatic BA synthesis. The significance of these findings with regard to human hepatic or gastrointestinal disease is unknown. However, dysregulation of FGF19 production and hepatic BA overproduction have been implicated in the etiology of idiopathic BA malabsorption  and a similar phenotype of decreased BA absorption coupled with an inability of the liver to synthesize additional BA was previously described for gallstone patients .
This project was supported by the NIH (DK047987) and an American Heart Association Mid-Atlantic Affiliate Grant-in-aid (to P.A.D.). A.R. was supported by the NIH (F32 DK079576).
The underlying research reported in this study was funded by the NIH Institutes of Health.
Conflict of interest
The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript.