Bile Acids: Natural Ligands for an Orphan Nuclear Receptor

See allHide authors and affiliations

Science  21 May 1999:
Vol. 284, Issue 5418, pp. 1365-1368
DOI: 10.1126/science.284.5418.1365


Bile acids regulate the transcription of genes that control cholesterol homeostasis through molecular mechanisms that are poorly understood. Physiological concentrations of free and conjugated chenodeoxycholic acid, lithocholic acid, and deoxycholic acid activated the farnesoid X receptor (FXR; NR1H4), an orphan nuclear receptor. As ligands, these bile acids and their conjugates modulated interaction of FXR with a peptide derived from steroid receptor coactivator 1. These results provide evidence for a nuclear bile acid signaling pathway that may regulate cholesterol homeostasis.

Cholesterol homeostasis is achieved through the coordinate regulation of dietary cholesterol uptake, endogenous biosynthesis, and the disposal of cholesterol in the form of bile acids. Bile acids are not simply metabolic by-products, but are essential for appropriate absorption of dietary lipids and also regulate gene transcription. Among the genes regulated by bile acids are cholesterol 7α-hydroxylase (Cyp7a), the rate-limiting enzyme in bile acid biosynthesis (1), and the intestinal bile acid–binding protein (I-BABP), a cytosolic protein that serves as a component of the bile acid transport system in the ileal enterocyte (2). I-BABP gene expression is induced preferentially by chenodeoxycholic acid (CDCA) relative to other more hydrophilic bile acids (3).

To examine whether CDCA mediates its transcriptional effects through an orphan member of the steroid-retinoid-thyroid hormone receptor family (4), we used a chimeric receptor system in which the putative ligand-binding domain (LBD) of the human orphan receptor is fused to the DNA binding domain of the yeast transcription factor GAL4 (5). In CV-1 cells, CDCA selectively activated FXR [NR1H4] (Fig. 1), an orphan nuclear receptor expressed predominantly in the liver, kidney, intestine, and adrenals (6, 7). This strong activation by CDCA was unanticipated because FXR responds to high concentrations of farnesoids (6) and retinoids (8).

Figure 1

CDCA selectively activates FXR. CV-1 cells were cotransfected with various nuclear receptor-GAL4 chimeras (22) and the reporter plasmid (UAS)5-tk-CAT (5). Cells were treated with 100 μM CDCA. Cell extracts were subsequently assayed for chloramphenicol acetyltransferase (CAT) activity (5). Data are expressed for each receptor as fold induction of CAT activity relative to vehicle-treated cells and represent the mean of three data points ± SD.

To further investigate the structure-activity relation of FXR activation, we tested a number of naturally occurring cholesterol metabolites, including bile acids (Fig. 2A), oxysterols and steroids, for their ability to activate full-length human or full-length murine FXR in CV-1 cells. CDCA activated both the human and mouse FXR (Fig. 2B). Dose response analysis showed that although some activation was seen at 3.3 μM, greater activation was observed at 100 μM CDCA (Fig. 2C). FXR was also activated by the secondary bile acids lithocholic acid (LCA) and deoxycholic acid (DCA), although these compounds were less efficacious than CDCA (Fig. 2, B to E). A similar activation pattern was observed in various cell lines, includingDrosophila-derived S2 cells, indicating that CDCA, LCA, and DCA do not require any specialized metabolic conversion to activate FXR (9). Urso-DCA, the 7β-hydroxy epimer of CDCA, and cholic acid (CA), which differs from CDCA by only the addition of a hydroxyl group at the 12α position, were inactive on FXR (Fig. 2B). In addition, no activation of FXR was seen with either α- or β-muricholic acid (MCA), the glycine or taurine conjugates of bile acids, oxysterols, farnesol, or other products derived from the mevalonate pathway (Fig. 2, B and F). Thus, both the 5β-cholanoic acid backbone and stereochemistry of the hydroxyl groups in CDCA are critical for optimal FXR activation.

Figure 2

Activation of FXR by bile acids. (A) Chemical structures of major human bile acids. (B) Full-length human (filled bars) and full-length murine (open bars) FXR are activated by CDCA, LCA, and DCA. CAT assays were performed with extracts of CV-1 cells transfected with expression plasmids for human or murine FXR, human RXRα, and the FXREhsp27-tk-CAT reporter plasmid (8, 23). Cells were treated with 100 μM of the indicated bile acid or farnesol, or 10 μM of the indicated steroid or TTNPB. Cholesterol, CH. (C to E) Dose response analysis of bile acids on human FXR. The assays were run as above with 1, 3.3, 10, 33, or 100 μM of CDCA, LCA, or DCA. (F) Human FXR is not activated by CDCA conjugates in CV-1 cells. The assay was run as above with 100 μM free or conjugated CDCA. (G) Human FXR is activated by conjugated bile acids in CV-1 cells expressing the human IBAT gene. The assay was run as above except cells were additionally transfected with an expression plasmid for the human IBAT gene (pCMV-HISBT) and treated with 3 μM of the indicated bile acid. Data are expressed as fold induction of CAT activity relative to vehicle-treated cells and represent the mean of three data points ± SD (5).

Bile acids are usually found conjugated to glycine or taurine, a derivative of cysteine. Cells require the presence of an active bile acid transporter for uptake of these conjugated derivatives (10). To test whether conjugated bile acids would also activate FXR, we coexpressed the human ileal bile acid transporter (IBAT) with FXR in CV-1 cells (11). FXR was strongly activated by 3 μM of the taurine or glycine conjugates of CDCA, LCA, and DCA (Fig. 2G). Weaker activation was seen with the conjugated forms of CA, and tauro-MCA was inactive (Fig. 2G). These data indicate that FXR can be activated by conjugated bile acids in tissues that express bile acid transporters such as the terminal ileum, liver, and kidney. The relation between the chemical structure of bile acids and their activation of FXR is in close agreement with the reported effects of bile acids on induction of I-BABP expression in Caco-2 cells and inhibition of Cyp7a expression in hepatocytes (3, 12)

Coactivator proteins interact with nuclear receptors in a ligand-dependent manner and augment transcription (13). A short amphipathic α-helical domain that includes the amino acid motif LXXLL (L is Leu and X is any other amino acid) serves as the interaction interface between these coactivator molecules and the ligand-dependent activation function (AF-2) located in the COOH-terminus of the nuclear receptor LBD (14). This AF-2 function of FXR was essential for response to bile acids (8,9). To test whether CDCA and its conjugates would induce a conformation of FXR that favors coactivator binding, we established a cell-free ligand-sensing assay using fluorescence resonance energy transfer (FRET) to monitor allosteric interaction of the steroid receptor coactivator–1 (SRC-1) with the receptor. The use of FRET to monitor macromolecular complex formation is well established, particularly for immunoassays (15), and this detection methodology has recently been extended to characterize ligand binding to nuclear receptors (16). The LBD of human FXR was labeled with the fluorophore allophycocyanin and incubated with a peptide derived from the second LXXLL motif of SRC1 (amino acids 676 to 700) that was labeled with europium chelate. CDCA and the corresponding glycine and taurine conjugates increased the interaction between FXR and the SRC1 peptide as determined with time-resolved FRET (Fig. 3A). Dose response analysis showed that CDCA, glyco-CDCA, and tauro-CDCA increased the amount of SRC1 peptide bound to the FXR LBD with half-maximal effective concentration (EC50) values ranging from 4.5 μM for CDCA to 10 μM for the conjugated forms (Fig. 3B and Table 1). These values are well within the physiological range of the concentrations of these bile acids in both the liver and intestine (17).

Figure 3

Free and conjugated bile acids bind to FXR. (A) CDCA and its conjugates increase SRC1 binding to FXR. The FRET ligand-sensing assay was run with 10 nM of the biotinylated FXR LBD (24) labeled with streptavidin-conjugated allophycocyanin, 10 nM of a SRC1 peptide labeled with streptavidin-conjugated europium chelate, and 100 μM of the indicated compound (25). Data are expressed as the means ± SD derived from three independent experiments. (B) Dose response analysis of CDCA binding to FXR. The FRET ligand-sensing assay was run in the presence of increasing concentrations of CDCA (open circles), glyco-CDCA (open triangle), tauro-CDCA (open boxes), or cholic acid (closed circles). (C) LCA, DCA, and TTNPB compete with CDCA for FXR binding. FRET ligand-sensing assays were run in the presence of 50 μM CDCA and increasing concentrations of LCA (open circles), DCA (open boxes), and TTNPB (closed circles).

Table 1

Potency of bile acids for binding to FXR as determined in the cell-free ligand-sensing assay (25). The indicated values for CDCA and its conjugates are EC50's derived from dose response analysis as described in Fig. 3B. The indicated values for other bile acids and their conjugates are half-maximal inhibitory concentrations (IC50's) derived from dose response analysis as described in Fig. 3C.

View this table:

Although LCA, DCA, and (E)-[(tetrahydrotetramethylnaphthalenyl)propenyl]benzoic acid (TTNPB) activate FXR in the cell-based reporter assay (Fig. 2B), they did not promote interactions between the FXR LBD and SRC1 in the FRET ligand-sensing assay (Fig. 3A). However, when these compounds were assayed in the presence of 50 μM CDCA, they disrupted the CDCA-FXR-SRC1 complex in a dose-dependent fashion (Fig. 3C and Table 1). Thus, CDCA, LCA, DCA, and TTNPB are ligands for FXR. Similarly, the taurine and glycine conjugates of LCA decreased the fluorescence signal, indicating displacement of CDCA from FXR with IC50values of 3.8 and 4.7 μM, respectively (Table 1). The conjugated forms of DCA and MCA caused a small decrease in fluorescence at the highest concentration tested, and no effect was obtained with the conjugated forms of CA and urso-DCA [Table 1 and (18)].

We conclude from several lines of evidence that the orphan nuclear receptor FXR can act as a nuclear receptor for bile acids. First, FXR is most abundantly expressed in liver, intestine, and kidney, tissues that are exposed to significant bile acid fluxes, and that express bile acid transporters. Furthermore, CDCA, LCA, DCA, and their conjugated derivatives bind to FXR at concentrations consistent with those found in tissues and known to regulate gene transcription. Finally, these bile acids are highly efficacious activators of FXR in our cell-based reporter assays. In the accompanying paper, Makishimaet al. report the effects of FXR on I-BABP and Cyp7a gene transcription (19). With the conjugated forms of the bile acids, activation is only observed in cells that express a bile acid transporter. Thus, the conjugated derivatives, which account for ∼98% of all bile acids in human bile, are likely to represent natural FXR ligands in tissues that express bile acid transporters, whereas the unconjugated forms may function as ligands in tissues that do not express these transport proteins. The identification of bile acids as natural FXR ligands suggests that these compounds may have important and unexpected functions in mammalian physiology (20). The identification of FXR target genes should provide important insights into these functions.

  • * To whom correspondence should be addressed. E-mail: tmw20653{at}


View Abstract

Navigate This Article