The nuclear receptor FXRα (NR1H4) plays a pivotal role in maintaining bile salt and lipid homeostasis by functioning as a bile salt sensor in mammals. In contrast, FXRβ (NR1H5) from mouse is activated by lanosterol and does not share common ligands with FXRα. To further elucidate FXR ligand/receptor and structure/function relationships, we characterized a FXR gene from the marine skate, Leucoraja erinacea, representing a vertebrate lineage that diverged over 400 million years ago. Phylogenetic analysis of sequence data indicated that skate Fxr (sFxr) is a FXRβ. There is an extra sequence in the middle of the sFxr ligand binding domain (LBD) compared with the LBD of FXRα. Luciferase reporter assays demonstrated that sFxr responds weakly to scymnol sulfate, bile salts, and synthetic FXRα ligands, in striking difference from human FXRα (hFXRα). Interestingly, all-trans retinoic acid was capable of transactivating both hFXRα and sFxr. When the extra amino acids in the sFxr LBD were deleted and replaced with the corresponding sequence from hFXRα, the mutant sFxr gained responsiveness to ursodeoxycholic acid, GW4064, and fexaramine. Surprisingly, chenodeoxycholic acid antagonized this activation. Together, these results indicate that FXR is an ancient nuclear receptor and suggest that FXRα may have acquired ligand specificity for bile acids later in evolution by deletion of a sequence from its LBD. Acquisition of this property may be an example of molecular exploitation, where an older molecule is recruited for a new functional role.
- nuclear receptor
- structure/function relationship
nuclear receptors play critical roles in development and physiology by sensing cellular levels of steroid hormones and dietary metabolites, although the true ligands for many of these receptors have not yet been identified or may even not exist. It has been demonstrated that nuclear receptors have shifted or gained their ligand specificity during evolution (4, 10); however, it remains to be determined how most nuclear receptors have acquired their ligand specificity and when this occurred in the course of evolution.
FXRα (NR1H4) and FXRβ (NR1H5) represent two forms of the farnesoid X receptor (FXR) in mammals, although FXRβ is a pseudogene in humans and other primates (23). FXRα and FXRβ are members of the class I nuclear receptors, which heterodimerize with the retinoid X receptors (RXR, NR2B) and regulate gene expression. FXRs share the typical structure of other members in this family, including an activation function domain 1 (AF1) at the N-terminus, followed by a conserved DNA binding domain (DBD), a ligand binding domain (LBD) that also contains an activation function domain 2 (AF2) at the C-terminus, as well as a hinge region that links the DBD and LBD (11). FXRα and FXRβ share ∼50% amino acid identity, but their ligand specificity does not overlap (23). FXRα is a bile acid-activated nuclear receptor, which plays a pivotal role in maintaining bile salt and lipid homeostasis by sensing bile salt levels and transactivating gene expression in mammals (6, 29). Its target genes include CYP7A1 (12, 19) (the rate-limiting enzyme for conversion of cholesterol into bile acids), the bile salt export pump (2) (BSEP/ABCB11), the ileal bile acid-binding protein (I-BABP) (13), and the organic solute transporter α and β (OSTα-OSTβ) (3, 18). Chenodeoxycholic acid (CDCA) is the most potent endogenous ligand for FXRα, and other bile salts activate the receptor to varying degrees (21, 25, 30). FXRα ligand specificity is also species-dependent, as noted by the observation that mouse Fxrα is less responsive to CDCA stimulation than human FXRα (hFXRα) (7), whereas mouse Fxrα is more sensitive to androsterone stimulation than is the human ortholog (31). In contrast, mouse Fxrβ (mFxrβ) is transactivated only by lanosterol, a precursor of cholesterol, although the functional role of FXRβ is not known (23). Xenopus Fxr (a β form) also does not respond to CDCA (28).
The origin of these two FXR paralogs is not known, although it has been proposed that FXR α and β differentiated from a common ancestral gene in evolution (23). To further elucidate FXR's receptor/ligand and structure-function relationships, we have used a comparative genomic approach to examine whether bile salts or lanosterol are ligands for a FXR from the marine skate, Leucoraja erinacea, whose predominant bile salt is 5β-cholestane-3α 7α,12α,24ξ,26,27-hexol sulfate, a sulfated bile alcohol (17). Interestingly, our findings indicate that bile salts only minimally activate skate Fxr (sFxr).
MATERIALS AND METHODS
Chemicals were purchased from Sigma-Aldrich, except when the source is specified. Cell culture medium DMEM, FBS, penicillin/streptomycin, trypsin, and PBS were from Invitrogen (Carlsbad, CA). TaqMan probes and primers for real-time PCR were designed using Primer Express software provided by Applied Biosystems (Foster City, CA) and synthesized by Integrated DNA Technologies (IDT, Coralville, IA). PfuUltra high-fidelity DNA polymerase (Stratagene, La Jolla, CA) was used for regular PCR. Luciferase assay kit was purchased from Promega. [35S]l-methionine was purchased from Amersham (Piscataway, NJ). Fexaramine and 6α-ethyl-chenodeoxycholic acid (6-ECDCA) were gifts from Dr. Ronald M. Evans (The Salk Institute for Biological Studies, La Jolla, CA) and Dr. Roberto Pellicciari (Universitá di Perugia, Perugia, Italy), respectively. Dr. Timothy Willson (GlaxoSmithKline, Research Triangle Park, NC) provided GW4064.
Cloning of the skate Fxr cDNA.
Human (NM_005123), rat (NM_021745), mouse (NM_009108), chicken (AF492497), zebrafish (BC092785) and Xenopus (BC061668) FXR sequences were obtained from GenBank. DNAStar software was used for sequence alignment and comparison. Two highly conserved regions with the sequence of ELCVVCGD—MDMYMRRK separated by 38 amino acids were found in the aligned FXRs. On the basis of these sequences, two degenerate oligonucleotide primers were made: forward, 5′-GARYTNTGYGTNGTNTGYGGNGA-3′; reverse, 5′-TTNCKNCKCATRTACATRTCCAT-3′. Five micrograms of skate liver total RNA was used for reverse transcription to generate single strand cDNA as PCR template. After a touch down PCR, a 162-bp band was amplified. DNA sequencing (service provided by the W. M. Keck Biotechnology Center at Yale University) confirmed that this DNA fragment encodes for a portion of the DBD of Fxr. On the basis of this sequence, the full-length cDNA of sFxr was cloned from skate liver total RNA by 5′ and 3′ RACE (rapid amplification of cDNA ends) using a kit from BD Biosciences Clontech (Palo Alto, CA). The full-length cDNA was assembled by PCR, cloned into a pCR2.1 vector (Invitrogen), and sequenced. The open reading frame (ORF) of sFxr was further confirmed using gene-specific primers directly amplified from skate liver cDNA.
To place the sFxr sequence (ABP98947.1) within a phylogenetic tree (hypothesis), a large, taxonomically diverse group of FXR sequences were retrieved from public databases. These included: Gallus gallus Nr1h4 (NP_989444.1), Macaca mulatta Nr1h4 (XP_001090069.1), Homo sapiens NR1H4 (NP_005114.1), Canis familiaris predicted Nr1h4 (XP_852959.1), Canis familiaris predicted Nr1h5 (β) (XP_848688.1), Oryctolagus cuniculus Fxr (NP_001076195.1), Mesocricetus auratus Nr1h4 (AAM53549.1), Mus musculus Nr1h4 (NP_033134.1), Mus musculus Nr1h5 (AAI19526.1), Rattus norvegicus Fxr α (NP_068513.1), Rattus norvegicus predicted Fxrβ (XP_001072587.1), Strongylocentrotus purpuratus hypothetical protein Fxr (XP_779997.2), Danio rerio hypothetical protein Fxr (NP_001002574.1), Danio rerio hypothetical protein AAH92785.1, Danio rerio hypothetical protein Fxrβ (XP_685672.2), Tetraodon nigroviridis predicted protein (EMB CAG03422), Takifugu rubripes (IMCB SINFRUG00000134981), Ciona intestinalis predicted Fxr (JGI: protein id 261841, ci0100145000). To provide outgroup polarity for rooting trees, the following LXR (NR1H3 human nomenclature) sequences were also included: Ciona intestinalis predicted Lxr (ciona4:97191), Homo sapiens NR1H3 (LXR) (NP_005684.1), Mus musculus Nr1h3 (NP_038867.2), Gallus gallus Nr1h3 (NP_989873.1), and Danio rerio hypothetical protein (NP_001017545.1). This sample of sequences provides representatives of the mammals, bony fish, the urochordate Ciona and the urchin Strongylocentrotus. To investigate the phylogenetic relationships between FXR genes, we employed parsimony as implemented in Phylogenetic Analysis Using Parsimony* program and Bayesian inference of phylogeny (BI), as implemented in MrBayes v3.1.2 (1, 27). In the Bayesian approach, the amino acid substitution model was not fixed, letting the MCMC sampling procedure visit all available models and decide which model fits the data in hand best. We ran two separate analyses starting from different random trees for 1 million generations sampling every 500 steps. After discarding the first 20% of the samples, we built a 50% majority-rule consensus tree, where clade support is expressed by the posterior probability of each clade. Three heated and one cold chain were used with heating parameter 0.05, as it was leading to better mixing. Convergence was assessed by inspecting the stationarity of the ln-likelihood in Tracer v1.3 (http://evolve.zoo.ox.ac.uk), the standard deviation of split frequencies as it approached zero (0.003028), and the successful swap frequencies between chains (0.53–0.86). The amino acid substitution model that fit the data best was the JTT (16) with posterior probability 1.00.
The full-length ORF of sFxr was cloned into a pcDNA3 expression vector (Invitrogen). Human FXRα and humanRXRα (hRXRα) expression vectors were gifts from Dr. David Mangelsdorf (University of Texas Southwestern Medical Center, Dallas, TX). A p-145Luc firefly luciferase expression plasmid, which contains a FXR response element (FXRE) in the human BSEP promoter was a gift from Dr. M. Ananthanarayanan (The Mount Sinai Medical Center, New York). A 300-bp DNA fragment from human I-BABP promoter region, which contains a FXRE was amplified from genomic DNA, confirmed by sequencing, cloned into in a pGL3-basic vector (containing firefly luciferase gene; Promega, Madison, WI), and named as pGL3-hIBABP. The two primers are: forward, 5'-CCGTTGCCATCCTGACCCTG-3'; reverse, 5'-GCCGGTGAAAGCCATGCTGCTG-3'. To generate sFxr ΔIns mutation, amino acids 349–398 of sFxr were replaced with amino acids 340–352 of hFXRα (NM_005123) using PCR. To amplify the N and C terminus of sFxr, the following primers were used: forward, 5'-GCTGAATTCAATCAACAAACAGTGCATGAATGA-3', reverse: 5'-TTCCAATAGGTCAGAATGCCCAGACGGAAGGCTCTGGTTGTACAATTGT GC-3', and forward: 5'-GGCATTCTGACCTATTGGAAGAAAGAATCCGCACATTAGGTATAGCTGA-3', reverse: 5'-GCTGAATTCTACACTTGCACTGACCAGTCC-3'. These two DNA fragments were fused by PCR using the forward primer from N-terminal and reverse primer from C-terminal portions of the sequence. The final PCR product was inserted into the pcDNA3 vector. To generate H340R mutants from sFxr wild-type and sFxr ΔIns mutation, the QuickChange kit (Stratagene) was used. The primer set is: forward, 5'-GATGTTCCTTCGCTCGGCACAATTGTACAACC-3' and reverse, 5'-TTGTGCCGAGCGAAGGAACATCACTTCCACAG-3'. All mutants were confirmed by DNA sequencing.
Quantitative real-time PCR analysis.
Total RNA was isolated from fresh skate tissues by using cesium chloride gradient centrifugation. Concentration and purity were confirmed by spectrophotometry and formaldehyde denaturing agarose gel electrophoresis. Five microgram of total RNA from each sample was used to generate cDNA by reverse transcription with the ProSTAR 1st strand synthesis kit (Stratagene). TaqMan real-time quantitative PCR assay was performed on an ABI 7500 Sequence Detection System, according to the manufacturer's protocol (ABI). For each sample, 2 μl of synthesized cDNA from reverse transcription reaction or un-reverse transcribed total RNA was detected in triplicates using a 20 μl reaction for the target gene. 28S ribosomal RNA was used as reference for normalizing the data. The sFxr cDNA fragment was purified from agarose gel, quantified by spectrophotometry, and used as template for preparation of a standard curve for absolute quantification. The sFxr primer and probe sequences are: forward, 5'-TCTCAGGCTCTCGGAGAATGC-3'; reverse, 5'-TGTCTGGAATCCTGGCAATCTT-3'; probe, 5'-ACCCTGCAAGTGGAAGTATTGGTAGAGTTTGCT-3'.
GST-sFxr construct was made by inserting the full-length ORF of sFxr into pGEX-6p-1 vector. A GST-hRXRα expression plasmid was prepared as previously described (5). Recombinant GST, GST-sFxr, and GST-hRXRα protein were expressed in E. coli DH5α, and purified using glutathione beads. Double-strand DNA oligos containing the wild-type or mutated FXRE in human I-BABP were prepared as described previously (13). EMSA was performed using the DIG-Gel Shift Kit following manufacturer's instructions (Roche Boehringer, Indianapolis, IN), where 1 ng of labeled probe was incubated with 100 ng of either GST, GST-sFxr, or GST-hRXRα protein alone or in combination.
Luciferase reporter assay.
A dual-luciferase assay was applied for functional characterization by cotransfecting either sFxr or hFXRα with the pCMX-hRXRα, pGL3-hIBABP, and phRL-CMV (constitutively expressing Renilla luciferase, Promega, Madison, WI) plasmids into HEK293T or HepG2 cells. Briefly, cells were seeded on 24-well plates and grown to 75% confluence. For each well, 250 ng pGL3-hIBABP, 100 ng hFXRα or sFxr expression plasmid, 50 ng hRXRα expression plasmid, and 2 ng phRL-CMV were mixed with 3 μl of lipofectamine2000 (Invitrogen) in Opti-MEM medium for overnight transfection. After transfection, the cells were treated with the indicated concentrations of chemicals in 0.5% charcoal stripped FBS DMEM for 24 h, and harvested by applying 100 μl of passive lysis buffer (Promega). A fraction of the cell lysate was subjected to a dual-luciferase report assay using a kit from Promega. The firefly luciferase activity was normalized by Renilla luciferase activity. Promoter activities are given as means ± SD of three independent transfections, and each transfection was performed in triplicate. The level of significance of promoter activities in the presence of regular substrates was determined using Student's t-test.
Protein three-dimensional structure modeling.
Rat Fxrα LBD crystal structure (1OT7A) was used as template for modeling the LBD structure of hFXRα, sFxr wild-type, and sFxr ΔIns mutant through SWISS-MODEL protein modeling server (http://swissmodel.expasy.org/). The predicted hFXRα LBD structure was verified by comparing with its actual crystal structure (1OSH). The three-dimensional structures were visualized with the RasMol program (www.OpenRasMol.org).
GST pull-down assay.
A 680-bp DNA fragment encoding the human steroid receptor coactivator 1 (SRC-1) amino acid 561–782 was amplified from HepG2 cell cDNA using PfuUltra DNA polymerase (Stratagene), confirmed by sequencing, and cloned into a pGEX-5X-3 expression vector (Amersham). The primers are forward: 5'-GCTGGATCCTCAGGCAGATGAGCTCAC-3', reverse: 5'-CGCACTCGAGATCCATCTGTTCTTTCTTTTCC-3'. The GST-SRC-1(AA561–782) fusion protein was expressed in E. coli DH5α, and purified using glutathione-Sepharose 4B beads. 35S-labeled hFXRα and sFxr were synthesized using the TNT T7 QuickCoupled transcription/translation kit (Promega), according to the manufacturer's instructions. Approximately, 2 μg of GST-SRC-1 protein on beads was incubated with 2.5 μl of 35S-labeled full-length hFXRα or sFxr in the presence of various FXR ligands. The mixture was incubated overnight at 4°C with gentle agitation in a total volume of 100 μl (8 mM Tris·HCl, pH 7.4, 0.12 M KCl, 8% glycerol, 0.5% wt/vol CHAPS, 4 mM DTT, and 1 mg/ml bovine serum albumin). The beads were washed four times with wash buffer (20 mM Tris·HCl, pH 8.0, 100 mM KCl, 0.5% Tween 20, and 2 mM DTT) prior to electrophoresis on a 10% SDS-PAGE, transferred to nitrocellular paper, and visualization by autoradiography.
Skate Fxr identification and tissue distribution.
The full-length sFxr cDNA contains 2,380 bp with 186 bp at the 5′ UTR, 1,557 bp for the coding region, 737 bp at the 3′ UTR, and an AATAAA sequence at 11 bp upstream of the polyA sequence (Genbank accession: EF520727). Protein sequence analysis shows that this sFxr has 131 amino acids (AA) for the AF1 domain, 70AA for the DBD as for other FXRs, 49AA for the hinge domain, and 268AA for the LBD. Sequence alignment demonstrates that sFxr have significant sequence identity to all known FXRs, including the α and β forms (Table 1).
Of note, sFxr shares 57% amino acid identity with hFXRα, which is lower than all of the α forms, but higher than all of the β forms. On the other hand, sFxr is 52% identical to mFxrβ, which is higher than all α forms but lower than all β forms. In contrast, hFXRα only shares 43% amino acid identity to mFxrβ. A Bayesian inference phylogenetic analysis indicated that sFxr is a FXR β form (Fig. 1A). Detailed sequence comparison reveals that the DBD of sFxr is the most conserved domain, sharing 91% identity with the DBD of hFXRα. However, the LBD is less conserved compared with mammalian FXR α forms. In particular, the middle of the LBD in sFxr contains an extra stretch of 37 amino acids, when compared with other FXRαs (Fig. 1B). Insertions of variable lengths are also present in other FXRβ orthologs. Interestingly, sFxr shares identical sequence to FXRα at the most distal 11 amino acids of the C-terminus, which forms helix 12 in the AF2 region (22).
sFxr and hRXRα can coactivate the FXRE from the human ileal BABP promoter.
To test whether sFxr can bind to the FXRE in the I-BABP promoter, we performed EMSA using purified recombinant proteins. As shown in Fig. 2A, the labeled FXRE probe was shifted only when both sFxr and hRXRα were added but not with either protein alone. This shift was competed by unlabeled wild-type probe, but not unlabeled mutated oligo, indicating that the binding of FXRE and sFxr/hRXRα was specific. Further, to test whether this binding can transactivate gene expression, sFxr was cotransfected with the pGL3-hIBABP construct into HEK293T cells. These cells have low endogenous expression of hFXRα (unpublished data), facilitating detection of the transfected gene. As shown in Fig. 2B, sFxr increased reporter gene activity ∼7 times more than hRXRα alone, indicating that sFxr can bind to this human FXRE and transactivate its expression. Cotransfection with hRXRα synergistically stimulated sFxr activity about 32-fold, suggesting that sFxr and hRXRα form a heterodimer in this transactivation assay. Together, these data demonstrate that sFxr can heterodimerize with hRXRα, bind to FXRE, and activate gene expression.
Bile acids are weak ligands for sFxr.
To examine whether bile salts are ligands for sFxr, CDCA was tested in varying concentrations in HEK293T cells transfected with sFxr or hFXRα. As expected, hFXRα responded to CDCA stimulation in a dose-dependent manner with 25 μM CDCA stimulating activity ∼18-fold when compared with its medium control (Fig. 3A). In contrast, sFxr was only stimulated ∼40% compared with its medium control. Minimal stimulation was also seen when these transfected cells were treated with various concentrations of scymnol sulfate, the major skate bile salt (Fig. 3B), even though a rat apical sodium-dependent bile salt transporter (Asbt, Slc10a2) expression plasmid was cotransfected to facilitate bile salt uptake. In contrast to sFxr, 25 μM scymnol sulfate was fully capable of stimulating hFXRα approximately eightfold, whereas sFxr was activated by only ∼50%. Of note, basal sFxr activity was always approximately threefold higher than basal hFXRα activity, indicating that the skate receptor may be constitutively active or may be activated by endogenous compounds in the HEK293T cells.
To examine whether other bile salts could activate sFxr, a series of bile salts was tested for sFxr and hFXRα activation, including taurine and glycine conjugated bile salts. In these experiments, a rat Asbt expression vector was cotransfected to facilitate conjugated bile salt uptake into the cells. As shown in Fig. 3C, hFXRα demonstrated variable responses to different bile salts, as previously reported (21, 25), whereas sFxr was not activated under the same conditions. To determine whether the lack of bile salt activation of sFxr was promoter-dependent, a p-145Luc construct containing the human bile salt export pump (hBSEP) FXRE(2) was tested in co-transfected cells that were treated with different concentrations of CDCA and scymnol sulfate. Similar results were seen as with this BSEP promoter (data not shown).
hFXRα and sFxr respond differently to stimulation by synthetic ligands.
To further test whether other mammalian FXR ligands could transactivate sFxr, sFxr was compared with hFXRα in cotransfected cells that were treated with different doses of 6-ECDCA, GW4064, fexaramine, and other agents using a luciferase reporter assay. In this experiment, hFXRα, but not sFxr demonstrated dose-dependent responses to the synthetic ligands (Fig. 4A), confirming previous reports (9, 26, 32) and indicating that these more potent mammalian FXR/Fxrα ligands are also not able to greatly activate sFxr. For example, 1 μM GW4064 induced hFXRα activity 18-fold, but only stimulated sFxr activity by 60% (Fig. 4A). Farnesol, androsterone, pregnenolone, and squalene were also tested in transfected HEK293T and HepG2 cells, and although 50 μM farnesol or 50 μM androsterone stimulated hFXRα and sFxr activities about twofold and 1.5-fold, respectively, pregnenolone and sequalene (up to 50 μM) did not increase hFXRα or sFxr activities under the same conditions (data not shown). In addition, 22R-hydroxycholesterol (an LXR agonist), WY14643 (a PPARα agonist), lanosterol (a mouse Fxrβ ligand), cholecalciferol [vitamin D3, a vitamin D receptor (VDR) ligand), and ergocalciferol (vitamin D2, a VDR ligand) all failed to stimulate either sFxr or hFXRα (Fig. 4B and data not shown).
It has been previously reported that polyunsaturated fatty acids, specifically arachidonic acid and docosahexaenoic acid (DHA) are ligands for hFXRα, and modulate BSEP expression (33). Because skate hepatocytes contain a high content of lipids, including DHA (24), we tested whether polyunsaturated fatty acids could activate sFxr. However, neither arachidonic acid nor DHA (up to 100 μM) significantly stimulated hFXRα or sFxr activity with or without addition of CDCA in transfected HEK293T and HepG2 cells (data not shown).
Retinoic acids activate both sFxr and hFXRα.
In contrast to these negative results, two retinoic acid receptor (RAR, NR1B) ligands, all-trans retinoic acid (atRA) and TTNPB were each capable of transactivating both hFXRα and sFxr. atRA (1 μM) stimulated hFXRα and sFxr activity approximately fivefold and threefold, respectively compared with their own medium control, whereas, 5 μM TTNPB was a more potent activator of hFXRα (∼18-fold) than of sFxr (∼1.8-fold). Rifampicin (a PXR ligand) and bilirubin (a CAR ligand) also slightly stimulated sFxr activity (Fig. 4B). To examine whether atRA is a ligand for hFXRα or sFxr, an in-vitro SRC1 recruitment assay was performed; however, the results demonstrated that neither hFXRα or sFxr were able to recruit coactivators in the presence of up to 10 μM atRA (data not shown), suggesting that atRA is not a ligand for FXR but might permissively activate FXR, possibly through RXR.
Structure modeling indicates that the extra sequence in the sFxr LBD alters its ligand-binding pocket (LBP).
To examine the influence of the extra 37 amino acids in the sFxr LBD, a structure modeling program was used to compare the LBDs of sFxr and hFXRα. The structure modeling program was initially verified by comparing the predicted and actual structures of hFXRα LBD, using the rat Fxrα LBD crystal structure as the template. As shown in Fig. 5A, the predicted structure is similar to the actual structure, except for a small difference around helix 2, which is missed in the actual structure. Of note, the LBD structure of hFXRα is in complex with fexaramine (9), whereas the LBD structure of rat Fxrα is in complex with 6-ECDCA (22).
When this modeling program was applied to sFxr, it predicted that the structure of the sFxr LBD is markedly different from the hFXRα LBD (Fig. 5B). This is particularly notable at the LBP region, where the extra amino acids in the middle of the sFxr LBD form four β-sheets between helix 5 and 6, and an additional coil structure between helix 6 and 7. Because these extra amino acids in the sFxr might explain the lack of response to classic FXR ligands, we next determined whether sFxr could gain FXRα ligand specificity when those extra amino acids were removed. A sFxr ΔIns mutant was designed by deleting those extra amino acids from sFxr and replacing this region with 13 amino acids from the same region of hFXRα. The sFxr ΔIns mutant was predicted to have an almost identical structure to the hFXRα LBD (Fig. 5C).
The sFxr ΔIns mutant gains FXRα ligand specificity.
The sFxr ΔIns mutant was characterized in HEK293T cells using luciferase reporter assays. As predicted, the sFxr ΔIns mutant demonstrated greater response to ursodeoxycholic acid (UDCA), GW4064, and fexaramine compared with its wild-type (Fig. 6A). A dose response for these activation effects was observed with these compounds. In addition, the mutant had twofold higher basal activity than the wild-type. In contrast, neither 1 μM 6-ECDCA, nor other bile salts (25 μM), including CDCA, scymnol sulfate, deoxycholic acid, cholic acid, taurocholic acid, taurochenodeoxycholic acid, or glycochenodeoxycholic acid stimulated the mutant more than its wild-type in the same assays (Figs. 3C and 6A, and data not shown). Because the only difference between CDCA and UDCA is the position of the 7-hydroxyl group, we assessed the ability of CDCA to alter UDCA responsiveness. As illustrated in Fig. 6B, CDCA (25 μM) almost abolished the stimulatory effects of UDCA (50 μM) or fexaramine (5 μM), and inhibited GW4064's activation of the sFxr mutant in a dose-dependent fashion. These results suggest that CDCA can bind to the sFxr ΔIns mutant and compete with UDCA, fexaramine, and GW4064.
Prior reports indicate that the arginine at position 328 of rat Fxrα is in direct contact with the carboxyl group of 6-ECDCA (22) and that the R331L mutation in hFXRα abolishes CDCA's activation effect (8). The corresponding residue in sFxr appears to be H340 (Fig. 1C), which is a weak basic amino acid. Because UDCA and CDCA may bind and activate sFxr ΔIns, additional studies tested whether mutation of this histidine to an arginine at position 340 in sFxr wild-type and in the ΔIns mutant would improve bile acid stimulation. However, luciferase reporter assays did not show any differential effects of these mutants when transfected cells were treated with UDCA or CDCA. However, 1 μM GW4064 induced sFxr ΔIns/H340R ∼50% more than sFxr ΔIns. The different effects of CDCA and UDCA on the sFxr ΔIns mutant suggest that UDCA and CDCA may bind to FXRα at different sites.
The function of FXRα as a bile salt sensor is well established in mammals, including human, rodents, and rabbit. However, bile salts exhibit great diversity in structure and in their ability to activate this transcription factor, and much remains to be learned about ligand structure-function relationships. The present results suggest that another vertebrate FXR, skate Fxr, is only weakly activated by bile salts, and thus the skate FXR receptor probably serves some other role in the shark and skate lineage.
Bile salts are derived from cholesterol, an important membrane lipid that is present only in eukaryotic cells. The appearance of cholesterol and other sterols such as ergosterol and phytosterols in eukaryotic membranes was a major step in the development of multicellular organisms. Cholesterol has a unique ability to increase lipid order in fluid membranes while maintaining fluidity and diffusion rates. However, cholesterol is insoluble in aqueous media, and specialized mechanisms are required to facilitate both the membrane delivery and the disposition of this compound. Disposition of cholesterol requires biotransformation to more water-soluble products, and in particular, to bile salts and steroid hormones. These metabolic products, of course, are not simply end products of cholesterol degradation but serve important physiological functions in many tissues.
Given that FXRα is a bile salt sensor in mammals, one might predict that this receptor and its ligands may have coevolved. However, the present results do not support this contention. Rather bile salts are only weak activators of sFxr, and previous studies indicate that bile salts are not effective ligands for mouse Fxrβ and Xenopus Fxr (a β form) (23, 28). The most primitive bile salts are found in reptiles, beginning with the di- and tri-hydroxylated C27 acids of Archosaurs (alligators and crocodiles) (13a). The little skate, which diverged from the other vertebrates about 400 million years ago, does not synthesize bile acids, but converts cholesterol to bile alcohols. The most abundant cholesterol degradation product found in skate bile is 5β-cholestane-3α 7α,12α,24ξ,26,27-hexol sulfate (scymnol sulfate), a sulfated bile alcohol (17).
Interestingly, the selective deletion of a stretch of amino acids in the sFxr LBD resulted in a partial gain in FXRα ligand specificity (Figs. 5 and 6). This partial gain of ligand specificity suggests that relative to FXRβ genes, FXRα genes act as bile acid sensors because of the lack of the ∼37 AA in the LBD. Thus, sFxr presumably evolved for some other yet to be determined function, but was recruited during the course of evolution into partnership with a newly evolved ligand (in this case bile salts), which first evolved in reptiles, resulting in novel interactions, namely regulation of bile salt synthesis and transport.
Although the putative endogenous ligands of sFxr are unknown, the present results indicate that retinoic acids, particularly atRA (Fig. 4B), are capable of stimulating sFxr activity compared with the other tested ligands. These results suggest that retinoic acid is a conserved activator for FXR/Fxrs. Surprisingly, however, neither hFXRα or sFxr were able to recruit coactivators in an in vitro SRC1 recruitment assay in the presence of up to 10 μM atRA (data not shown), suggesting that atRA is not a ligand for FXR but might permissively activate FXR, possibly through hRXRα. Note that atRA can be converted to 9cis retinoic acid, which is an endogenous ligand for hRXR(14); however, this possibility has not yet been tested experimentally. To further clarify whether atRA can transactivate sFxr in vivo in skate, it would be necessary to characterize the specificity of the skate RXR ligands. However, this would first require identification of the skate RXR gene to determine whether bile salts might activate sFxr if tested with its species specific heterodimer partner.
It is also important to note that the basal activity of sFxr was about three times higher than hFXRα in the luciferase reporter assays. This finding is unlikely to be due to higher expression of sFxr than hFXRα in the transfected cells because the same amount of [35S]-methionine-labeled protein was observed in in vitro transcription/translation assays (data not shown). Rather this higher basal activity and relative lack of ligand inducibility suggest that sFxr may be a constitutive activator for its target genes. At present the gene targets for sFxr are not known and must await sequencing of the skate genome.
The ability of FXR to alter gene transcription is dependent on two key components in its structure, that is, the DBD and the LBD. The DBD is the most conserved portion among FXR orthologs and binds directly to the FXRE found in the promoter of many genes, whereas the FXR LBD regulates its activity. When an agonist occupies the LBD, the peptide undergoes a conformational change, releases corepressors, recruits coactivators, and initiates transcription. The structure of the LBD determines the ligand specificity, and thus a small change in amino acid sequence could lead to a different ligand profile. The present study demonstrates that sFxr exhibits major differences in the LBD with that of human FXRα, and thus it is not surprising that its ligand specificity would also be quite distinct (Figs. 3 and 4). In contrast to mammalian FXRs, we could identify only one form of FXR from skate liver, although it is possible that our PCR approach may have missed a paralog that exhibits relatively low sequence identity with other vertebrate Fxrs. However, genome annotation indicates that there is only one FXR in Ciona intestinalis, an invertebrate chordate, which is thought to be a close ancestor of vertebrates. In Fugu (Takifugu rubripes), a teleost species, bioinformatic analyses suggest that there may be two FXR forms (20); a more detailed sequence analysis suggests that the second form (which is similar to the β form in DBD and other nuclear receptors as well) may be a pseudogene as there are at least two stop codons in its annotated sequence.
In summary, the present study has identified and characterized a divergent form of Fxr from the little skate, Leucoraja erinacea. Our results indicate that this FXR is activated by retinoids but only weakly by bile acids. Ligand sensitivity to bile acids appears to have developed after the divergence of sharks and skates and is likely to be related, in part, to the absence or presence in α or β paralogs of a 37 AA region of the LBD. Acquisition of FXR α bile acid sensitivity may be an example of “molecular exploitation” (4) created by evolutionary pressure as bile acid and bile acid metabolism developed in higher vertebrates.
This study was supported by the Mount Desert Island Biological Laboratory NIA award (to S. Y. Cai), and National Institutes of Health Grants DK-34989 and DK-25636 (to J. L. Boyer), ES01247 (to N. Ballatori), and ES-03828 (to J. L. Boyer).
David Mangelsdorf and M. Ananthanarayanan are acknowledged for providing the human FXRα/RXRα expression constructs and human BSEP promoter reporter gene construct, respectively. Ronald M. Evans, Roberto Pellicciari, and Timothy Willson provided fexaramine 6-ECDCA, and GW4064, respectively. We thank Trong Nguyen for excellent technical assistance, Ping Lam for structure modeling, and Carolyn Mattingly for the bioinformatics analysis.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society