Human FXR Regulates SHP Expression through Direct Binding to an LRH-1 Binding Site, Independent of an IR-1 and LRH-1

Background Farnesoid X receptor/retinoid X receptor-alpha (FXR/RXRα) is the master transcriptional regulator of bile salt synthesis and transport in liver and intestine. FXR is activated by bile acids, RXRα by the vitamin A–derivative 9-cis retinoic acid (9cRA). Remarkably, 9cRA inhibits binding of FXR/RXRα to its response element, an inverted repeat-1 (IR-1). Still, most FXR/RXRα target genes are maximally expressed in the presence of both ligands, including the small heterodimer partner (SHP). Here, we revisited the FXR/RXRα-mediated regulation of human SHP. Methods A 579-bp hSHP promoter element was analyzed to locate FXR/chenodeoxycholic acid (CDCA)- and RXRα/9cRA-responsive elements. hSHP promoter constructs were analyzed in FXR/RXRα-transfected DLD-1, HEK293 and HepG2 cells exposed to CDCA, GW4064 (synthetic FXR ligand) and/or 9cRA. FXR-DNA interactions were analyzed by in vitro pull down assays. Results hSHP promoter elements lacking the previously identified IR-1 (−291/−279) largely maintained their activation by FXR/CDCA, but were unresponsive to 9cRA. FXR-mediated activation of the hSHP promoter was primarily dependent on the −122/−69 region. Pull down assays revealed a direct binding of FXR to the −122/−69 sequence, which was abrogated by site-specific mutations in a binding site for the liver receptor homolog-1 (LRH-1) at −78/−70. These mutations strongly impaired the FXR/CDCA-mediated activation, even in the context of a hSHP promoter containing the IR-1. LRH-1 did not increase FXR/RXRα-mediated activation of hSHP promoter activity. Conclusion FXR/CDCA-activated expression of SHP is primarily mediated through direct binding to an LRH-1 binding site, which is not modulated by LRH-1 and unresponsive to 9cRA. 9cRA-induced expression of SHP requires the IR-1 that overlaps with a direct repeat-2 (DR-2) and DR-4. This establishes for the first time a co-stimulatory, but independent, action of FXR and RXRα agonists.

Both LRH-1 and FXR belong to the superfamily of nuclear receptors. FXR/RXRa binds to an inverted repeat sequence spaced by 1 nucleotide (IR-1) conforming to the consensus G/ AGGTCAnTGACCT [11]. Acting as a monomer, the conserved DNA binding site of LRH-1 is currently defined as (c/tCAAGGc/ tCg/a) [12,13]. In recent mouse whole-genome chromatinimmunoprecipitation (ChIP) experiments a remarkable enrichment of LRH-1-type binding sites was detected in DNA sequences precipitated with antibodies against FXR. FXR and LRH-1 were found to synergistically induce transcription of mouse Shp [14]. In line with these observations, functional LRH-1 binding sites have been identified in several genes that are also controlled by FXR/ RXRa including BSEP [2], SHP [6], organic solute transporter alpha/beta (OSTa/b) [15] and fatty acid synthase (FAS) [16,17]. FXR/RXRa-mediated transcriptional control is primarily regulated by their ligands, bile acids (in particular chenodeoxycholic acid (CDCA)) and 9-cis retinoic acid (9cRA), respectively. Earlier, we and others have shown that these ligands have opposite effects on binding of FXR/RXRa to the IR-1 and the resulting transcriptional activity of the human BSEP promoter [18,19]. FXR ligands (both CDCA and GW4064) strongly increase FXR/ RXRa-mediated expression of BSEP, while co-administration of 9cRA effectively represses this effect. 9cRA strongly reduced the binding of FXR/RXRa to the IR-1 sequences as they are present in the human BSEP and SHP promoters. In contrast to the effect on BSEP expression, however, CDCA and 9cRA synergistically activate SHP transcription, in both in vivo and in vitro experiments [19]. This suggests that the mechanisms by which these ligands control FXR/RXRa-mediated regulation of BSEP and SHP may be fundamentally different, while they are both considered to be ''typical'' FXR/RXRa target genes. Opposite effects of RXRa ligands on CDCA-induced expression of FXR/RXRa target genes have been described by others also [18,20], but the differential mechanisms remain elusive so far.
Over the last decade it has become evident that the function of FXR and SHP is not restricted to bile acid synthesis, but that these factors also play a role in liver regeneration, viral replication, tumor suppression, fibrogenesis, glucose and lipid metabolism [21,22]. It is therefore highly relevant to understand the molecular mechanisms that determine the ligand-selective regulation of FXR/RXRa target genes, in particular that of SHP.

Transfection
HepG2.rNtcp and HEK293 cells were transfected as described previously [4]. DLD-1 cells were transfected using Transfectine (Biorad, Hercules, CA) at a ratio of 3 ml Transfectine per mg DNA as recommended by the manufacturer. Expression plasmids of hFXR (pcDNA3-hFXR) and hRXRa (pSG5-hRXRa) were used at 200 ng and 100 ng, respectively, and luciferase reporter plasmids (pGL3-basic derivatives) at 1 mg per 9.6 cm 2 well. If needed, total amount of DNA was adjusted with pCMV5 plasmid to 1.3 mg per well.
Site-directed mutagenesis 59-truncated mutants of the hSHP promoter were made by PCR from the pGL3 hSHP 2569/+10. pGL3 SHP 2569/+10 FXRE KO [6] was generated via site-directed mutagenesis by full vector amplification. Deletion mutants lacking intra promoter regions, FXRE nonsense mutants and half-site nonsense mutants were generated by amplifying the two individual promoter fragments flanking the region to be mutated or deleted. Subsequently, these two fragments were fused by overlap PCR. PCR products were KpnI/BglII-ligated into pGL3 basic (Promega, Madison, USA). Oligo's (Invitrogen, Paisley, UK) used to generate these SHP promoter mutants are shown in Table S1. Endotoxin-free plasmids were isolated (Macherey-Nagel, Düren, Germany) from E. coli Top10 cells (Invitrogen, Paisley, UK). All promoter constructs were sequenced (BaseClear; Leiden; The Netherlands) to assure that the correct mutations were introduced. mRNA isolation and Q-PCR mRNA was isolated from DLD-1, HepG2.rNtcp and HEK293 cells. RT-QPCR was performed as was described before [26]. Sequences of the primer/probe sets are shown in Table S2.

Luciferase reporter assays
Cells were lysed in 500 ml passive lysis buffer (Promega, Madison, USA). After centrifugation, 20 ml of the supernatant was used to determine luciferase activity in a MPL1 Microplate Luminometer (Berthold Detection Systems). Using 50 ml luciferase substrate (Promega, Madison, USA), delay time set to 2.05 seconds and measuring time set to 10 seconds.

Statistical analysis
Data are presented as means 6 sd. Differences between conditions were determined in SPSS by Kruskal-Wallis followed by pair-wise comparison of groups by Mann-Whitney U with p# 0.05.

Results
The IR-1 at 2291/2279 is largely dispensable for FXRdependent induction of the human SHP promoter The RXRa ligand 9cRA lowers BSEP expression by inhibiting FXR/RXRa binding to the IR-1 [18,19], while transcription of other FXR/RXRa target genes (SHP, OSTb, ileal bile acid binding protein (IBABP), FGF19; see Figure S1) is super-induced. Here, we performed a detailed analysis of the human SHP promoter to obtain insight in the molecular mechanisms by which bile acids and 9cRA synergistically induce transcription of FXR/ RXRa target genes. The -569/+10 SHP promoter element described earlier [6] showed the same pattern of regulation by CDCA and 9cRA as observed for genomic SHP (Figure 1 and Figure S1). CDCA treatment resulted in a 12-fold increase of the SHP promoter activity and this was super-induced by 9cRA (+45% to 17-fold induction compared to untreated cells). An FXRE/IR-1 was previously identified at position 2291/2279 in the SHP promoter [5][6][7]. As expected, a 59-truncated SHP promoter element up to position 2303 (2303/+10) retained the CDCAinduction (9.8-fold) and 9cRA super-induction (+43% to 14-fold) characteristics. Further 59 shortening of the SHP promoter element Figure 1. The IR-1 at 2291/2279 is required for 9cRA-, but not for CDCA-mediated induction of the human SHP promoter. DLD-1 cells were transfected with hFXR and hRXRa expression plasmids and various hSHP promoter constructs as indicated. Cells were treated with or without 100 mmol/L CDCA and/or 1 mmol/L 9cRA. The synergistic effect of the FXR ligand (CDCA) and RXRa ligand (9cRA) on SHP promoter activity depends on the previously identified IR-1 located at 2291/2279. Mutation or deletion of this IR-1 sequence did not abolish SHP promoter activation by FXR/ CDCA. Luciferase activity was measured to determine the SHP promoter activity. Data are presented as means 6 SD; n$3. Vehicle-treated conditions are set to 1. p#0.05 for *) in a pairwise comparison by Mann-Whitney U test. doi:10.1371/journal.pone.0088011.g001 to 2278/+10 (deleting the IR-1) led to the loss of 9cRA superinduction. Remarkably, the CDCA-induction of the 2278/+10 SHP promoter element remained intact (8.1-fold; Figure 1). Disrupting the IR-1 sequence in the 2303/+10 promoter element (by an IR-1 knock out mutation (KO) [6], replacement by a nonsense sequence (NS) or full deletion (DEL)) led to the absence of 9cRA super-induction, but maintained the CDCA-induction (Figure 1), even in the context of the larger 2569/+10 SHP promoter element ( Figure S2). These data indicate that the 9cRA-mediated activation of the human SHP promoter depends on the IR-1 sequence at position 291/2279. However, this sequence is (largely) dispensable for the CDCA-induced activity. This latter finding was highly surprising and prompted us to study this in further detail.

CDCA-induced activation of the 2278/+10 SHP promoter elements depends on FXR
The CDCA-induced activation of the 2278/+10 SHP promoter element was comparable to the 2303/+10 SHP element (8.1-fold vs. 9.8-fold, respectively) and was fully dependent on the presence of FXR ( Figure 2). The FXR/CDCA-induction was lost when the SHP promoter was reduced to a minimal element of 269/+10. This indicates that the 2278/269 region in the human SHP promoter contains a yet unidentified sequence that is essential for FXR/CDCA-dependent regulation.
The novel FXR/CDCA-responsive element is located in the 2122/269 region of the SHP promoter Two fragments (2203/2122 or 2122/269) were deleted from the 2303/+10 and the 2278/+10 SHP promoter elements to delineate the region involved in the regulation by FXR/RXRa/ CDCA ( Figure 3A). Deletion of the 2122/269 fragment from the 2278/+10 SHP promoter element made it unresponsive to FXR/RXRa/CDCA, while deletion of the 2203/2122 did not reduce the FXR/RXRa/CDCA-activation. Importantly, the 2 122/269 deletion also strongly reduced the FXR/RXRa/ CDCA-activation of the IR-1-containing 2303/+10 SHP promoter element ( Figure 3B). Similar results were obtained when FXR/RXRa-transfected cells were treated with the synthetic FXR ligand GW4064 instead of CDCA ( Figure 3C). The FXR/ RXRa/CDCA-and FXR/RXRa/GW4064-induced regulation of the SHP promoter fragments was most pronounced in intestinal DLD-1 cells, but was also observed in hepatic HepG2.rNtcp and renal HEK293 cells ( Figure 4). These data indicate that the 2 122/269 sequence is crucial for FXR-dependent regulation of the SHP promoter. The previously identified IR-1 at position 2 291/2279 contributes only to a minor extend to the FXRdependent regulation of SHP.

An LRH-1 site is required for the FXR-dependent induction of human SHP
The 2122/269 region from the SHP promoter was screened for putative nuclear receptor binding sites. An IR-1-like sequence is detected at 2118/2106 (AtGTCtgTGtgtT) with 7 out of 12 IR-1 consensus nucleotides. Alternatively, FXR-regulation may act through a previously identified DR-1/PPARc binding site at position 290/278 [27] as it also contains the core TGACCT sequence. In addition, this region contains a binding site for LRH-1 (TCAAGGTTG at 279/271). Site-directed mutations were introduced in these 3 regions. While mutations in the IR-1-like and DR-1 sites only slightly reduced FXR-mediated induction of SHP promoter activity, it was almost completely abrogated when the LRH-1 site was mutated ( Figure 5). Previously, it was suggested that FXR and LRH-1 may synergistically induce expression of murine Shp [14]. While the human SHP promoter activity was dose-dependently induced by co-expression of mLRH-1, confirming the presence of a functional LRH-1 site, it did not enhance the FXR/CDCA-induced activity of the SHP promoter ( Figure 6). In fact, at high doses, LRH-1 rather represses FXR/CDCA-activation of the 2303/+10 hSHP promoter element. Similar results were obtained for the 2569/+10 hSHP promoter ( Figure S3). In addition, CDCA treatment of FXR/RXRa-transfected DLD-1 cells did not induce LRH-1 expression ( Figure S4). Collectively, these data indicate that the FXR-mediated induction of human SHP is independent of LRH-1, but requires the LRH-1 binding site.
FXR physically binds to the LRH-1 site in the 2122/269 region of the human SHP promoter Next, we analyzed whether FXR binds directly to the 2122/2 69 region in the human SHP promoter by applying an FXR-pull down assay [19] using a biotin-labeled 68-bp DNA fragment containing the 2122/269 region of the SHP promoter. This SHP promoter fragment efficiently precipitated FXR from nuclear extracts of CDCA-treated DLD-1 cells that stably overexpress hFXR (DLD-1.hFXR), similar as the positive control (BSEP-IR-1)  Figure 7A). Binding of FXR was abrogated when the LRH-1 site was mutated in the 2122/269 element. Moreover, the mutated sequence did not compete for FXR binding, while the wild type 2122/269 SHP promoter fragment did ( Figure 7A and B). Taken together, these data show that FXR-mediated expression of human SHP is largely controlled via direct binding of FXR to a newly-identified DNA sequence that includes an LRH-1 binding site at position 279/271.

Discussion
In this study, we show that FXR regulates human SHP expression primarily via direct binding to an LRH-1 site and not the previously identified IR-1. No synergism was detected between FXR and LRH-1 in regulation of human SHP. In contrast, the IR-1 sequence was required for 9cRA-induced expression of SHP. This is the first in-depth analysis of the co-stimulatory transcriptional regulation by the ligands of FXR and RXRa, CDCA and 9cRA. This mechanism is fundamentally different from FXR/ RXRa-mediated regulation of BSEP that acts through an IR-1 [4].
Sequencing of DNA fragments that bind FXR in the mouse genome has recently revealed that indeed the IR-1 is the most prominent binding site of this transcription factor [14,28,29]. Notably, the IR-1 in the mouse, rat and human SHP promoters deviate from the experimentally established IR-1 consensus at least at one crucial position (GAGTTAaTGACCT, where the underlined T in the SHP IR-1 is a G or C in the consensus IR-1). The second most enriched FXR-binding sequence in the genome-wide chromatin immunoprecipitation experiments appeared to confirm to an LRH-1 binding site that is in close proximity to the IR-1 [14]. Co-transfection experiments revealed a synergistic effect of FXR/RXRa and LRH-1 on the activity of the Shp promoter.
Our data show that 1) FXR-mediated expression of human SHP is largely independent of the IR-1, 2) LRH-1 does not enhance FXR/RXRa-induced SHP promoter activity, 3) FXR is precipitated with a DNA fragment containing the LRH-1 site, which is abrogated when this site is mutated; 4) FXR-mediated induction of a SHP promoter lacking an IR-1 is similar in cells without endogenous LRH-1 (HEK293) and in cells that contain intermediate (DLD-1) or high levels of LRH-1 (HepG2.rNtcp). This suggest that FXR-mediated regulation does not depend on the presence of LRH-1, which is in line with the observation that GW4064-induced Shp mRNA expression in mouse liver was hardly affected by the absence of Lrh-1 [30].
Most surprising was the fact that the IR-1 was extraneous for FXR-mediated induction of the hSHP promoter, but required a downstream sequence than harbors a LRH-1 binding site. The minimal LRH-1 binding site was, however, not sufficient to bind significant amounts of FXR in pull-down assays ( Figure S5), suggesting that sequences flanking the LRH-1 binding site are required for efficient FXR binding. This is corroborated by the observation that mutations in the 2122/269 region outside the LRH-1 consensus sequence also reduced the FXR-induced SHP promoter activity, though this was less pronounced than the mutations in the LRH-1 site. Taken together, we conclude that the LRH-1 site in the human SHP promoter is most important site for FXR-mediated expression. Importantly, this novel FXR binding site is fully conserved in the human, mouse and rat SHP promoters ( Figure S6).
Another remarkable finding was that the previously identified IR-1 was actually required for 9cRA-induced expression of SHP. Previously, it has been shown that the IR-1 overlaps with an LXRa/RXRa DR-4 response element at 2284/2269 [31]. The synthetic ligand RXRa was shown to be a potent activator of liver X receptor-alpha (LXRa)/RXRa-induced transcriptional activity, even more so than the LXRa ligand T0901317. Thus, it is very well possible that 9cRA induces SHP expression through LXRa/ RXRa. In addition, the IR-1 also contains a putative DR-2 sequence ( Figure S7). RXRa homodimers and RXRa/RAR heterodimers have been shown to bind DR-2 elements [32]. SHP promoter activity was indeed induced by 9cRA-activated RXRa, which was not affected by co-transfection with RAR (data not shown). So, alternatively, 9cRA may act via RXRa homodimers by binding the DR-2 in the 2291/2279 region in the SHP promoter.
At present it is unknown how common this alternative pathway of FXR-mediated transcription through LRH-1(-like) sequences is. The FXR/CDCA-induced activity via the LRH-1 site is insensitive to 9cRA. Together with the fact that the ''IR-1'' is required for the 9cRA-mediated induction of SHP provides the first molecular mechanism explaining how these ligands (9cRA and CDCA) lead to maximum induction of transcription, albeit via two independent sites. Maximum expression after exposure to both ligands was also observed for OSTb, IBABP, FGF19 and others observed this for phospholipid transfer protein (PLTP) [33] and SULT2A1: sulfotransferase 2A1 (SULT2A1/STD) [34]. In contrast, OSTa showed a BSEP-like pattern (9cRA blocks CDCAinduced expression). Since 9cRA was found to reduce binding of FXR/RXRa to the IR-1 sequence (as present in the human BSEP and SHP promoter), the IR-1 containing promoters of OSTb, IBABP and FGF19 need to harbor compensatory mechanism that ultimately lead to maximal transcription with both ligands.   [35] and conserved in rat and human (see Figure S6). All 4 IR-1-like sites and the LRH-1 site were mutated and analyzed for the effect on FXR/CDCA-mediated induction of the 2278/+10 hSHP promoter (B) mutating one of the IR-1 half-sites did not or only partially reduce FXR/CDCA-dependent activation of the 2122/269 hSHP promoter fragment, whereas mutations in the LRH-1 site strongly reduced the response of the 2122/269 hSHP promoter fragment to FXR/CDCA-stimulation. *) significantly different from CDCA treated 278/+10 WT (C) in the context of the 2569/210 hSHP promoter fragment the LRH-1 site is the dominant FXR/CDCA response element (over the previously identified IR-1). DLD-1 cells were transfected with hFXR and hRXRa expression plasmids and various hSHP promoter constructs as indicated. Cells were treated with or without 100 mmol/L CDCA. Luciferase activity was measured to determine the SHP promoter activity. a) significantly different from CDCA treated 569/+10 WT. b) significantly different from CDCA treated 2569/+10 IR-1 KO. Data presented as means 6 SD; n$3. P#0.05 for *), a) and b) in a pairwise comparison by Mann-Whitney U test. doi:10.1371/journal.pone.0088011.g005 hFXR-Mediated hSHP Expression Is IR-1 Independent Figure 6. No synergy between FXR and LRH-1 in human SHP regulation. LRH-1 dose-dependently induced activation of the 2303/+10 hSHP promoter fragment, confirming the presence of a functional LRH-1. In the presence of FXR a similar dose response curve is observed. However, in the presence of CDCA, FXR and LRH-1 do not synergistically activate the SHP promoter. LRH-1 rather limits the FXR/CDCA-dependent activation at a higher dose. DLD-1 cells were transfected with hFXR and hRXRa expression plasmids, the 2303/+10 hSHP promoter construct and/or increasing amounts of an mLrh-1 expression plasmid as indicated. Cells were treated with or without 100 mmol/L CDCA. Luciferase activity was measured to determine the SHP promoter activity. a) significantly different from 0 ng mLrh-1 vehicle, between vehicle treated conditions. b) significantly different from 0 ng mLrh-1 FXR/RXRa, between FXR/RXRa treated conditions. Data presented as means 6 SD; n$3. P#0.05 for a) and b) in a pairwise comparison by Mann-Whitney U test. doi:10.1371/journal.pone.0088011.g006 Our data contrast to those previously reported by Lu et al. [5] and Goodwin et al. [6], who reported that the IR-1 is essential for FXR-induced expression of human SHP. We followed the same experimental approach as these studies. Most of our data was generated using DLD-1 cells, in which we observed the most robust FXR-mediated induction of the SHP promoter. Still, we show that the IR-1 was also dispensable for FXR-mediated induction of the SHP promoter in HEK293 and HepG2 cells, which were used in the earlier studies. A putative explanation for these, seemingly opposite, observations may be that both the IR-1 and the LRH-1 site can bind FXR and activate SHP expression, but that their relative contribution may depend on the cellular and/or nuclear levels of FXR, RXRa and their ligands. These factors may vary between cell types, passage numbers and the experimental conditions in these 3 studies. The RXRa ligand (9cRA) may reduce the binding of the FXR/RXRa heterodimer to the IR-1 sequence and stimulate LXRa/RXRa or RXRa homodimer binding in the 2291/2269 region. In contrast, 9cRA does not affect the FXR-mediated regulation of SHP via the 2 122/269 element. Physiologically, this maintains SHP-mediated regulation of bile salt synthesis (CYP7A) and bile salt import (NTCP, ASBT) independent of the vitamin A/9cRA levels. In line with this, bile salt-mediated induction of Shp was maintained in vitamin A-deficient mice [19].
In conclusion, our study reveals for the first time a molecular mechanism of FXR-activated transcription that is not inhibited by the RXRa ligand 9cRA, which is the most frequent mode of regulation observed for FXR-target genes. Surprisingly, this is mediated through a non-IR-1 FXR response element that shows typical characteristics of an LRH-1 DNA binding consensus. Figure S1 Gene-and cell type-specific regulation of FXR/RXRa target genes by 9cRA. HepG2.rNtcp (A) and DLD-1 (B) cells were transfected with expression plasmids for hFXR and hRXRa and treated with or without 100 mmol/L CDCA and/or 1 mmol/L 9cRA. mRNA levels of FXR target genes were determined by Q-PCR. Data are corrected for 18S and displayed as means 6 SD; n$3. CDCA-treated conditions are set to 100. Significant differences (P#0.05) are indicated when compared to untreated condition (a), 9cRA-treated condition (b), CDCA-treated condition (c) or CDCA/9cRA-treated condition (d) in a pair-wise comparison by Mann-Whitney U test. (TIF) Figure S2 Inactivation of the IR-1 at position 2291/2279 does not abolish FXR/CDCA-mediated induction of the 2569/+10 hSHP promoter. DLD-1 cells were transfected with hFXR and hRXRa expression plasmids and the 2569/+10 SHP promoter construct (B). Cells were treated with or without 100 mmol/L CDCA and/or 1 mmol/L 9cRA. Luciferase activity was measured to determine the SHP promoter activity (A). Data presented as means 6 SD; n$3. Vehicle-treated conditions are set to 1. P#0.05 for *) in a pair-wise comparison by Mann-Whitney U test. (TIF) Figure S3 No synergy between FXR and LRH-1 in human SHP regulation. LRH-1 dose dependently induced activation of the 2569/+10 hSHP promoter fragment, confirming the presence of a functional LRH-1. In the presence of FXR a similar dose response curve is observed. However, in the presence of CDCA, FXR and LRH-1 do not synergistically activate the SHP promoter. LRH-1 rather limits the FXR/CDCA-dependent activation at a higher dose.DLD-1 cells were transfected with hFXR and hRXRa expression plasmids, the 2569/+10 SHP promoter construct and/or increasing amounts of the mLrh-1 expression plasmid as indicated. Cells were treated with or without 100 mmol/L CDCA. Luciferase activity was measured to determine the SHP promoter activity. Data presented as means 6 SD.