Identification of a Bile Acid-responsive Element in the Human Ileal Bile Acid-binding Protein Gene

Intestinal bile acid-binding protein (I-BABP) is a cytosolic protein that binds bile acids (BAs) with a high affinity. In the small intestine, its expression is restricted to the ileum where it is involved in the enterohepatic circulation of BAs. Using the human enterocyte-like Caco-2 cell line, we have recently shown that BAs increased I-BABP gene expression. To determine whether this regulation occurs in vivo, the effect of BA depletion or supplementation was studied in mice. A dramatic drop in I-BABP mRNA levels was observed in mice treated with the BA-binding resin cholestyramine, whereas an increase was found in animals fed with taurocholic acid. BAs are physiological ligands for the nuclear farnesoid X receptor (FXR). Both FXR and I-BABP are co-expressed along the small intestine and in Caco-2 cells. To determine the role of FXR in the regulation of I-BABP expression, the promoter of the human I-BABP gene was cloned. In Caco-2 cells, cotransfection of FXR and RXRα is required to obtain the full transactivation of the I-BABP promoter by BAs. Deletion and mutation analyses demonstrate that the FXR/RXRα heterodimer activates transcription through an inverted repeat bile acid responsive element located in position −160/−148 of the human I-BABP promoter. In conclusion, we show that FXR is a physiological BA sensor that is likely to play an essential role in BA homeostasis through the regulation of genes involved in their enterohepatic circulation.

Primary BAs 1 are synthetized from cholesterol in the liver where they are conjugated with glycine or taurine prior to secretion into bile (1). In most mammals, bile is stored in the gall bladder. During a meal, BAs are released into the duode-num where they are required for the efficient absorption of dietary fat and lipid-soluble vitamins. In the distal gut, conjugated BAs may undergo bacterial modifications leading to the formation of secondary BAs. In humans, more than 90% of BAs are reabsorbed throughout the intestine and return, via the portal blood, to the liver where they are secreted again into bile. This enterohepatic circulation is essential for the maintenance of BA and cholesterol homeostasis (1).
Intestinal absorption of BAs takes place as a function of their chemical form (hydrophobicity index) through three complementary mechanisms: passive nonionic diffusion and facilitated and active protein-mediated transports. After bacterial deconjugation, passive diffusion of protonated BAs occurs in the ileum and colon. Passive absorption of glycine-conjugated BAs has also been recently reported in the jejunum of guinea pig (2). Conjugated dihydroxy-BAs are primarily absorbed at the jejunal level via facilitated transport, whereas taurine and glycine trihydroxy-BA are actively transported in the ileum (3). The relative contribution of jejunal carrier-mediated transport under physiological conditions remains to be determined. By contrast, the active ileal absorption of conjugated BAs has been the subject of extensive research. BA uptake in the ileum is mediated by an ileal sodium-dependent bile acid transporter (IBAT) located in the brush border membrane of ileocytes. This 38-kDa integral plasma membrane protein has been cloned in various species including humans (4). The physiological importance of IBAT for the maintenance of the enterohepatic circulation of BAs, and hence, cholesterol homeostasis, is supported by the observation that patients with mutations in the IBAT gene fail to absorb BAs efficiently and have reduced low density lipoprotein cholesterol levels (5,6). Once in the cell, BAs are bound to the intestinal bile acid-binding protein (I-BABP), a 14 -15-kDa cytoplasmic protein whose expression is restricted to the ileum (7). The I-BABP gene has been cloned and characterized from mouse (8) and rabbit (9). I-BABP binds BAs with a high affinity and may be involved in cellular BA uptake and trafficking because it has been found to be transiently associated with the IBAT in the brush border membrane (10). Therefore, the synthesis of I-BABP may constitute a rate-limiting step in the enterohepatic circulation of BAs. In agreement with this hypothesis, we have recently shown that BAs are potent inducers of I-BABP gene expression in enterocyte-like Caco-2 cells (11).
Recent findings have shown that BAs bind (12) and activate (12)(13)(14) the farnesoid X receptor (FXR), a member of the nuclear receptor superfamily. FXR can bind DNA sequences comprised of two inverted repeats separated by one nucleotide (IR-1) as a heterodimer with the 9-cis-retinoic acid receptor (RXR) (15,16). We report here that BAs induce expression of the human I-BABP gene through the interaction of the FXR/ RXR heterodimer with a positive BA responsive element (BARE) in the proximal I-BABP promoter. The physiological relevance of these data is supported by the fact that BA-mediated regulation of I-BABP expression is also found in vivo in the mouse.
Animals and Experimental Treatments-French guidelines for the use and care of laboratory animals were followed. Male Swiss mice (30 Ϯ 2 g) purchased from Centre d'Elevage R. Janvier (France) were used. Animals were housed individually in a controlled environment (constant temperature and humidity, darkness from 8 p.m. to 8 a.m.) and fed ad libitum a standard chow containing 3% (w/w) lipid (UAR A-04, Usine d'Alimentation Rationnelle, France). To study the effect of BA depletion or supplementation on intestinal I-BABP expression in vivo, mice were fed laboratory chow for 10 days containing either 4% (w/w) cholestyramine or 0.5% (w/w) TCA and CDCA, respectively. The efficiency of treatments were assessed by the evaluation of hepatic cholesterol 7␣-hydroxylase mRNA levels (17). After sacrifice, intestinal and hepatic samples were snap frozen in liquid nitrogen and stored at Ϫ80°C until RNAs were extracted. In adult Swiss mice, the length of the small intestine is 50 Ϯ 5 cm. To determine whether I-BABP and FXR are co-expressed in the gut, the small intestine was divided in five segments of 10 Ϯ 0.5 cm from the pylorus to the ileocecal valvula. The jejunum is known to start usually only after the Treitz's ligament. The first 10 cm (segment 1) were arbitrary considered to be the duodenum, the following 25 cm (segments 2-4) were considered to be the jejunum, and the last 15 cm (segments 4 -5) were considered to be the ileum.
Molecular Cloning of the Human I-BABP Promoter-A human genomic library (CLONTECH, HL1111j) was screened using an EcoRI fragment of human I-BABP cDNA as hybridization probe (22). Approximately 1 million plaque-forming units were screened, and several positive clones were isolated then purified. A clone containing the longest 5Ј-flanking region was characterized by subcloning in pUC119 and then sequencing on both strands by the dideoxy-chain termination method with a 373A Sequencer (Perkin-Elmer, Applied Biosystems). This clone was found to contain a large 5Ј-flanking fragment and the transcription region covering exon 1 of hI-BABP gene.
Plasmid Constructions-The 3.7-kilobase DNA genomic fragment of human I-BABP was used as a template for polymerase chain reactionbased generation of different deletion and mutation constructs of the promoter region. Polymerase chain reactions were performed with a proofreading DNA polymerase (Pfu DNA polymerase, Stratagene) and a common antisense primer that ended 44 bp downstream the transcription start site. The Ϫ2769/ϩ44 and Ϫ1204/ϩ44 bp fragments were cloned upstream the chloramphenicol acetyltransferase (CAT) gene in the pCAT3-basic vector (Promega). The -148/ϩ44 bp fragment was obtained by AvaI digestion of the -2769/ϩ44 bp fragment. Mutation and deletion of the IR-1 sequence was generated by polymerase chain reaction using the following oligonucleotides 5Ј-TCCCCAGCCTGAATA-AGGTCGGG-3Ј (mutations in the IR-1 BARE underlined) and 5Ј-GGCAATGGGGTGACAGCACTTGGGGCTTGTCCCTCCAGGT-3Ј, respectively. All constructs were confirmed prior to use by both restriction digestions and sequencing with the dideoxy chain termination method.
Cotransfection Assays-The enterocyte-like Caco-2 cells were used for the transfection studies. They were plated in 6-well plates in Dulbecco's modified Eagle's medium in the absence of phenol red supplemented with 10% charcoal-stripped fetal calf serum at 50 -60% confluency. In general, transfection mixes contained 250 ng of human FXR and/or human RXR␣ expression vectors, 4 g of I-BABP-CAT reporter plasmid, 500 ng of human IBAT expression vector (generous gift of Dr P. Dawson, Wake Forest University, Winston-Salem, NC), 500 ng of ␤-galactosidase expression vector. Cells were transfected overnight by calcium phosphate precipitation. The medium was changed by Dulbecco's modified Eagle's medium (without phenol red) supplemented with 10% delipidated calf serum and BAs, and the cells were incubated for an additional 24 h. Cell extracts were prepared and assayed for CAT and ␤-galactosidase activities.

Effects of a BA Depletion or Supplementation on I-BABP
mRNA Levels in Mice-An in vivo study was designed to assess the effect of alterations in the luminal BA levels on I-BABP gene expression. The experiments were conducted in mice rather than in rats, because their intestines are subject to intermittent fluxes of BAs released from the gall bladder. BA depletion was achieved pharmacologically by the addition of the BA sequestrant resin cholestyramine in the diet. The BA loading was performed by supplementation of the diet with either TCA or CDCA. No significant changes in body mass or in food intake were found between controls and treated mice through the course of the experiment (data not shown). The effectiveness of the treatments was demonstrated by an increase or a decrease in hepatic cholesterol 7␣-hydroxylase gene expression in response to cholestyramine and BAs, respectively (17) (data not shown). In ileum, a dramatic drop in I-BABP mRNA levels was seen in BA-depleted mice, whereas a significant increase occurred after chronic TCA feeding. By contrast, the unconjugated BA CDCA exerted no effect on I-BABP expression under these conditions (Fig. 1). These data demonstrate that changes in BA pool size result in marked alterations in I-BABP gene expression in vivo.
I-BABP and FXR Are Co-Expressed in Both Small Intestine and Caco-2 Cells-The small intestine is a heterogeneous organ characterized by variable gene expression along the gastrocolic axis at the origin of functional specialization. If the nuclear receptor FXR is involved in the regulation of I-BABP gene expression, these proteins should be coexpressed within this tissue. As shown in Fig. 2A, ileum is the exclusive intestinal segment where FXR, I-BABP, and IBAT are expressed. By contrast, L-FABP, an I-BABP-related protein known to also bind BA (23), exhibits a nonoverlapping pattern of expression with FXR ( Fig. 2A). It is noteworthy that L-FABP gene expression is not regulated by BAs (11). Consistent with previous reports (15), FXR was also found in the kidney and at a lower level in the liver (Fig. 2B). Low levels of FXR mRNA were detectable in undifferentiated Caco-2 cells used in the transfection studies. By contrast, IBAT mRNA were undetectable in these cells. As described previously (11), an induction of I-BABP mRNA levels is triggered when Caco-2 cells were sub-jected to 250 M CDCA for 24 h (Fig. 2C). Interestingly, CDCA treatment led to an apparent decrease in FXR mRNA levels (Fig. 2C), suggesting that the BAs regulate FXR expression.
FXR/RXR Heterodimer Induces Transcription of the Human I-BABP Gene-FXR regulates gene transcription as an heterodimer with RXR (15,16). To determine whether this heterodimer is involved in the BA-dependent induction of the human I-BABP gene, a Ϫ2769/ϩ44 bp fragment of I-BABP promoter was cloned upstream of a CAT reporter gene. This reporter was transiently transfected into Caco-2 cells together with human FXR and/or human RXR␣ expression vectors in the presence or absence of 100 M CDCA and/or 1 M 9-cisretinoic acid. As shown in Fig. 3A, reporter gene activity was induced efficiently in presence of FXR and CDCA. Weaker activation was seen with RXR␣ and its ligand, 9-cis-retinoic acid. Maximum transactivation was obtained when both FXR and RXR␣ expression vectors were co-transfected, and the cells were treated with CDCA and 9-cis-retinoic acid. These data demonstrate that the I-BABP gene is regulated by the FXR/ RXR heterodimer. We next evaluated the ability of different physiologically relevant BAs to activate transcription of the human I-BABP promoter. Both CDCA and the secondary bile acid deoxycholic acid were efficacious activators of the I-BABP promoter-reporter construct (Fig. 3B). CDCA-mediated induction of reporter gene expression was dose-dependent with a half-maximal effective concentration (EC 50 ) of 35 M (Fig. 3C). A significant activation occurred with concentrations as low as 5 M. In contrast to CDCA and deoxycholic acid, lithocholic acid, and the hydrophilic BA cholic acid exhibited relatively weak activation (Fig. 3B), whereas the conjugated BAs failed to activate transcription through the I-BABP promoter. These last data are paradoxical because (i) the glyco-and tauroconjugated BAs are the major forms found in vivo and (ii) TCA feeding significantly increases the I-BABP mRNA levels in the mouse (Fig. 1). We previously showed (12) that conjugated BAs require the presence of the membrane transporter (IBAT) to activate FXR in CV-1 cells. As shown in Fig. 2A, IBAT is absent in the Caco-2 cells. To overcome this limitation, Caco-2 cells were cotransfected with a human IBAT expression vector together with FXR and RXR␣. Under these conditions, the I-BABP promoter-reporter gene was strongly activated by 5 M cholic acid, TCA, or glycocholic acid (Fig. 3D). Thus, the major physiological BAs regulate expression of the human I-BABP gene promoter.
Identification of a Positive BARE in the Human I-BABP Gene Promoter-The FXR/RXR␣ heterodimer recognizes the insect ecdysone responsive element that consists of an inverted repeat of the nuclear receptor half-site sequence AG(G/T)TCA separated by 1 nucleotide (IR-1) (15). Sequence alignment of the proximal promoters of the human, rabbit, and mouse I-BABP genes revealed a highly conserved but imperfect IR-1 sequence (Fig. 4). To determine whether the FXR/RXR␣ heterodimer can bind to this IR-1 motif, electrophoretic mobility shift assays were performed using the 32 P-labeled IR-1 from the human I-BABP promoter in presence of human FXR and/or human RXR␣. Neither FXR nor RXR␣ alone bound to the probe (Fig.  5A, lanes 1 and 2). However, when mixed, the two proteins bound efficiently to the I-BABP IR-1 (Fig. 5A, lane 3). This binding was specific, as demonstrated by competition with an excess of either wild type IR-1 motif or an idealized IR-1 sequence containing two consensus half-sites (Fig. 5A, lanes 4 -7). A mutated IR-1 motif failed to compete (Fig. 5A, lanes 8 and 9).
To explore the functional role of this FXR/RXR␣-binding site in the regulation of human I-BABP gene by BAs, Caco-2 cells were cotransfected with human FXR and human RXR␣ expression vectors and I-BABP promoter-CAT reporter plasmids in which the IR-1 sequence was mutated. The Ϫ1204/ϩ44 construct containing the native IR-1 sequence was transactivated about 30-fold in the presence of CDCA (Fig. 5B, lane 1). BA transactivation of the I-BABP gene was abolished in the Ϫ148/ ϩ44 construct in which the IR-1 sequence was partially deleted (Fig. 5B, lane 2). Notably, mutation or deletion of the IR-1 motif resulted in the complete loss of reporter gene induction in response to CDCA (Fig. 5B, lanes 3 and 4). These data establish Northern blot hybridization of I-BABP, FXR, IBAT, and L-FABP mRNA. 4.5 g of poly(A ϩ ) RNA were resolved on a 1% agarose gel containing 2.2 M formaldehyde, transferred to a nylon membrane, and fixed by UV irradiation. A, the small intestine was divided into five segments of 10 Ϯ 0.5 cm from the pylorus to the ileocecal valvula as described under "Materials and Methods." Segment 1 is duodenum, from segment 2 to the middle of segment 4 is jejunum, and from middle of segment 4 to segment 5 is ileum. B, mRNA from liver and kidney. C, mRNA from Caco-2 cells treated with either vehicle alone (V) or 250 M CDCA.
this IR-1 motif in the proximal human I-BABP promoter to be a functional BARE. DISCUSSION I-BABP is a small cytoplasmic protein that belongs to the fatty acid-binding protein family (24). In the small intestine, it is found exclusively in the ileum where it binds BAs with high affinity. Although the cellular function of I-BABP is not yet fully understood, it may facilitate BA uptake and trafficking and/or serve as an intracellular buffer for protecting cells from the detergent effects of excess BAs. We have recently shown that I-BABP gene expression in enterocyte-like Caco-2 cells is tightly regulated by BAs, especially CDCA (11). To establish the physiological relevance of these data, an in vivo study was conducted in mice chronically subjected to a BA depletion or excess. Adaptative up-and down-regulation of the I-BABP expression appears to depend on the size of the BA pool because the BA sequestrant cholestyramine triggered a dramatic drop in I-BABP mRNA levels while supplementation of the diet with TCA increased I-BABP expression levels. Surprisingly, CDCA exerted no regulatory action when it was added in the diet. It is possible that CDCA does not reach the ileum in a sufficient concentration to regulate I-BABP expression under these conditions. Indeed, it is known that protonated, unconjugated BAs passively diffuse along the small intestine. The acidic microcli-mate found in the unstirred water layer overlying the microvilli of enterocytes (25) must favor the protonation of dietary CDCA and thus its passive uptake. The positive feedback reported here was not found by Arrese and co-workers (26) in the rat because neither common bile duct ligation nor pharmacological sequestration of BAs led to a change in the expression of IBAT and I-BABP. The origin of this discrepancy is likely due to differences in the regulation of BA metabolism and transport between species that have gall bladders (e.g. humans and mice), and those that do not (e.g. rats). Thus in humans and other mammals with gall bladders, it may be necessary to modulate the expression of I-BABP and other genes involved in BA homeostasis to cope with the fluxes that occur in BA levels within the enterohepatic circulation.
The nuclear receptor FXR was recently proposed to be a physiological BA sensor (12)(13)(14). FXR is found in the tissues known to have significant BA metabolism, i.e. liver, kidney, adrenals, and gut (15). Our detailed examination of its expression along the mouse small intestine shows that FXR displays an overlapping expression pattern with the BA-transporters (I-BABP and IBAT) in the ileum. This finding suggests that FXR could play a physiological role in the regulation of BA flux in the ileal segment of the small intestine. Indeed, analysis of the human I-BABP promoter reveals the existence of a specific BA responsive element, which interacts with the FXR/RXR␣ heterodimer. Because BAs exist predominantly in their conjugated form in vivo, it is noteworthy that physiological concentrations of both glyco-and tauro-conjugated BAs are able to transactivate the human I-BABP promoter when Caco-2 cells are co-transfected with the IBAT expression vector. Thus, conjugated BAs are likely to serve as natural FXR ligands in tissues that express IBAT such as the ileum. The fact that the positive BARE sequence is highly conserved in the proximal promoter of human, mouse, and rabbit I-BABP genes together with the recent finding that FXR regulates the mouse I-BABP gene (13) strongly suggest that FXR plays a critical role in the modulation of I-BABP gene expression in multiple species. Because BAs bind to both I-BABP and FXR, we speculate that the induction of I-BABP expression by FXR may provide a mechanism for feedback regulation of ileal BA-sensitive genes through modulation of intracellular BA levels.
In conclusion, we have presented the first data demonstrating that I-BABP gene expression is physiologically regulated by BA flux. The conservation of this regulatory pathway from mice to humans is consistent with the role of FXR as a physiological BA sensor in the gut. Because the fecal loss of BAs is a major way for the elimination of cholesterol, identification of FXR antagonists may provide a new therapeutic approach for the treatment of hypercholesterolemia and hence, of cardiovascular diseases. A, the human I-BABP IR-1 binds the FXR/RXR␣ heterodimer. Electrophoretic mobility shift assays were performed with in vitro translated FXR and RXR␣ and with the wild type human IR-1 as probe. Competition analysis was performed with a 5-or 25-fold excess of I-BABP IR-1, idealized IR-1 (Ideal IR-1) or mutated IR-1 (I-BABPmut IR-1). B, mutation-deletion analysis of the human I-BABP promoter. Caco-2 cells were cotransfected with FXR and RXR␣ expression vectors and the different I-BABP promoter construct as indicated. Lane 1, the Ϫ1204/ϩ44 I-BABP construct contains the IR-1 sequence located between Ϫ160 and Ϫ148; lane 2, the Ϫ148/ϩ44 I-BABP construct is a deletion of the 5Ј end of the promoter that lacks the IR-1; lanes 3 and 4, the Ϫ1204/ϩ44 I-BABPmut and I-BABPdel constructs mutate or delete the IR-1 (mutation underlined). Cells were treated with 100 M CDCA for 24 h. Data are expressed as fold activation relative to cells treated with the vehicle alone and represents mean Ϯ S.E.