Ligand-activated pregnane X receptor interferes with HNF-4 signaling by targeting a common coactivator PGC-1alpha. Functional implications in hepatic cholesterol and glucose metabolism.

Previous studies show that feedback inhibition of bile acid production by bile acids is mediated by multiple mechanisms, including activation of pregnane X receptor (PXR). Consistent with these studies, the antibiotic rifampicin, a ligand for human PXR, reduces hepatic bile acid levels in cholestasis patients. To delineate the mechanisms underlying PXR-mediated suppression of bile acid biosynthesis, we examined the functional cross-talk between human PXR and HNF-4, a key hepatic activator of genes involved in bile acid biosynthesis including the cholesterol 7-alpha hydroxylase (CYP7A1) and sterol 12-alpha hydroxylase (CYP8B1) genes. Treatment with rifampicin resulted in repression of endogenous human CYP7A1 expression in HepG2 cells that was reversed by PXR small interfering RNA. The coactivator PGC-1 enhanced transcriptional activity of HNF-4, and this enhancement was suppressed by rifampicin-activated PXR. Endogenous PGC-1 from mouse liver extracts bound to PXR, and recombinant PGC-1 directly interacted with both PXR and HNF-4 in vitro. Rifampicin-dependent interaction of PXR with PGC-1 was shown in cells by coimmunoprecipitation, and intranuclear localization studies using confocal microscopy provided further evidence for this interaction. In chromatin immunoprecipitation studies, rifampicin treatment did not inhibit HNF-4 binding to the native promoters of CYP7A1 and CYP8B1 but resulted in dissociation of PGC-1 and concomitant gene repression. Most interestingly, these rifampicin effects were also observed in the phosphoenolpyruvate carboxykinase gene that contains a functional HNF-4-binding site and is central to hepatic gluconeogenesis. Our study suggests that ligand-activated PXR interferes with HNF-4 signaling by targeting the common coactivator PGC-1, which underlies physiologically relevant inhibitory cross-talk between drug metabolism and cholesterol/glucose metabolism.

tor (FXR) 1 increases transcription of small heterodimer partner (SHP), and SHP interacts with liver receptor homologue-1 and/or HNF-4, which leads to transcriptional suppression of CYP7A1 and CYP8B1 (1,4). In in vivo chromatin studies, we further demonstrated that bile acid-induced SHP recruits an mSin3A-Swi/Snf chromatin remodeling complex to the native human CYP7A1 promoter, leading to remodeling of the promoter chromatin and concomitant gene repression (5).
Although a central role for SHP in this negative regulation of bile acid production has been shown, recent knock-out mouse studies imply that there are additional bile acid-mediated regulatory mechanisms independent of SHP, including activation of mouse pregnane X receptor (PXR) by ligands such as secondary toxic bile acids (6,7). Treatment with pregnenolone 16␣carbonitrile (PCN), a specific ligand for mouse PXR, has also been shown to repress CYP7A1 transcription in mice (8,9). Furthermore, this CYP7A1 repression was no longer observed in PXR knock-out mice, indicating that PCN-activated PXR mediates suppression of bile acid production (8). Consistent with these animal studies, administration of the antibiotic rifampicin has been reported to reduce significantly hepatic bile acid levels in cholestatic patients. Thus, rifampicin has been used as a therapeutic agent for pruritus associated with intrahepatic cholestasis (10). Recently, human PXR was cloned, and rifampicin was identified as a specific ligand for human PXR, implying that activated PXR may be associated with reduced bile acid levels in liver. These clinical observations and experimental studies indicate that ligand-activated PXR reduces hepatic bile acid levels by suppressing CYP7A1 expression, but the underlying molecular mechanisms are not known. A possible mechanism for PXR-mediated repression is inhibition of signaling pathway by HNF-4, a key activator of basal expression of CYP7A1 (3,11).
The xenobiotic orphan nuclear receptor PXR and the human homologue, also called steroid and xenobiotic receptor (12,13), protect the body from harmful foreign or endogenous substances by activating a number of genes involved in drug and sterol metabolism (14 -16). PXR has a wide spectrum of foreign ligands such as antibiotics, anticancer drugs, and the anti-depressant St. John's wort, as well as endogenous ligands including sterol metabolites (15). Most interestingly, recent studies show that toxic bile acids such as lithocholic acid (LCA) activate the bile acid biosensor FXR, CAR, and the vitamin D receptor, as well as PXR and regulate transcription of genes involved in biosynthesis, transport, and metabolism of toxic bile acids in the enterohepatic system (8,(17)(18)(19)(20).
HNF-4 is a master transcriptional activator for a large number of genes in hepatocytes and pancreatic cells. By using a combination of chromatin immunoprecipitation assays and promoter microarrays, Young and co-workers (21) reported that 910 genes in hepatocytes and 758 genes in pancreatic islets were regulated by HNF-4. Also, a previous study (11) of conditional HNF-4 null mice demonstrated that HNF-4 was crucial to the expression of numerous hepatic genes and, especially, was a major in vivo regulator of genes involved in lipid homeostasis. HNF-4 is central to cholesterol and glucose/energy metabolism because it is a key activator for basal hepatic expression of CYP7A1 and CYP8B1 genes and glucose-6-phosphatase and PEPCK genes, which play central roles in bile acid biosynthesis and hepatic gluconeogenesis, respectively (1,3,(22)(23)(24). These genes all contain functional HNF-4-binding sites in their promoter, and mutation of these sites substantially disrupt promoter activity (1). 2 A DR1 motif that is a HNF-4binding site (25) is found in the bile acid responsive element (BARE) II region at Ϫ148 to Ϫ129 in the human CYP7A1 promoter and at ϩ198 to ϩ227 of the human CYP8B1 promoter (1). The human PEPCK promoter also contains a functional HNF-4-binding site at Ϫ431 to Ϫ418 (23,26).
PGC-1 is a key metabolic regulator and was originally identified as a peroxisome proliferator-activated receptor-␥-interacting coactivator in brown adipose tissue (27). Recent studies (23,24) show that PGC-1 is a versatile coactivator for numerous nuclear receptors and is implicated in diverse biological activities including lipid and carbohydrate metabolisms. PGC-1 has been shown to increase the HNF-4-mediated transactivation of CYP7A1, PEPCK, and glucose-6-phosphatase genes (23,28,29).
By using CYP7A1, CYP8B1, and PEPCK genes as model systems, we have studied molecular mechanisms by which human PXR suppresses hepatic bile acid production and gluconeogenesis. Here we show that ligand-activated PXR attenuates HNF-4 signaling by competing for the common coactivator PGC-1 in hepatic cells and thus causes dissociation of endogenous PGC-1 from HNF-4-bound promoter chromatin, resulting in concomitant gene repression. Our study implies that inhibitory cross-talk mediated by these orphan nuclear receptors may be physiologically relevant in drug and cholesterol/glucose metabolism.
Transient Transfection and Reporter Assay-Transfection was carried out by using either electroporation for ChIP and coimmunoprecipitation assays (Bio-Rad electroporator, settings at 600 microfarads and 300 V) or LipofectAMINE 2000 for functional reporter assays. Reporter assays were carried out as described (31,33) in 24-well plates. Then 16 -24 h after transfection, cells were incubated with fresh media containing rifampicin or Me 2 SO for 24 h. ␤-Galactosidase or Renilla luciferase assays were performed to normalize for transfection efficiency.
Gel Mobility Shift Assay-Six histidine-tagged (His 6 ) PXR (pET28-PXR), FLAG-tagged RXR, or 6His-HNF-4 (pET28-HNF-4) was partially purified as described previously (34). Gel mobility shift assays were performed as described previously (31,33,34). Briefly, the DNA probe was a 32 P-labeled oligonucleotide containing either the BARE II region of CYP7A1 or the PBRU region of CYP2B6/CYP2B1, and 5,000 -10,000 cpm were added to each reaction. Various amounts of HNF-4, PXR, and RXR were added to the reaction buffer (30 mM KCl, 1 mM MgCl 2 , 15 mM Tris-HCl, pH 7.9, 0.2 mM EDTA, 10% glycerol, 4 mM DTT) in the presence of 0.2 g of poly(dI-dC) in a final volume of 20 l. After 15 min of incubation at room temperature, the resulting protein-DNA complexes were fractionated on a nondenaturing acrylamide gel. For competition assays, a molar excess (6.25 to 100ϫ) of unlabeled double strand oligonucleotides containing either the DR4 motif or BARE II region was added 5 min before the addition of the probe.
GST and FLAG Pull Down Assays-GST fusion proteins, GST-HNF-4, GST-PXR, or FLAG-PXR, were expressed in Escherichia coli BL21-RIL (DE3 (pLys)) and purified by binding to glutathione-Sepharose (Amersham Biosciences) or M2-agarose. 35 S-Labeled proteins were synthesized by in vitro transcription and translation (TNT kit, Promega) according to the manufacturer's instructions. The reticulocyte lysates were precleared by incubation with GST-bound glutathione-Sepharose at 4°C for 30 min. One g of GST or GST fusion protein was incubated at 4°C for 2 h with 4 l of the precleared reticulocyte lysate containing the labeled proteins in 100 l of binding buffer (25 mM Hepes-KOH, pH 7.6, 150 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 0.5 mg/ml bovine serum albumin, 10% glycerol, and 0.25% Nonidet P-40) in the presence of protease inhibitors. After the incubation, the samples were extensively washed and eluted with 20 mM reduced glutathione, and the eluted proteins were separated by SDS-PAGE, and radioactivity was visualized by autoradiography. For FLAG affinity pull downs, 1 g of FLAG-PXR immobilized to M2-agarose, or M2-agarose was incubated with 500 g of mouse liver nuclear extracts in incubation buffer at 4°C for 2 h. After washing five times with incubation buffer, proteins associated with FLAG-PXR were eluted and detected by Western blotting using antisera against PGC-1 (sc-13067). 2 J. Miao and J. Kemper, unpublished data.
Coimmunoprecipitation Assays-HepG2 cells were transfected with either G4DBD or G4DBDPXR by using electroporation, and after 48 h, the cells were treated with either Me 2 SO or 10 M rifampicin for 2 h and subjected to coimmunoprecipitation as described (5). Briefly, harvested cells were resuspended in lysis buffer (20 mM K ϩ -Hepes, pH 8.0, 0.2 mM EDTA, 5% glycerol, 100 mM NaCl, 0.1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM NaF, 1 mM sodium orthovanadate, 1 mM DTT, and protease inhibitors). After incubation on ice for 10 min followed by sonication and centrifugation, cell extracts were precleared by incubation with 60 l of a 25% protein A-Sepharose slurry for 30 min at 4°C and were then incubated with 3-5 g of either IgG or Gal4 DBD antisera for 4 h to overnight at 4°C. Immune complexes were collected by incubation with 30 l of a 25% protein A-Sepharose slurry for 1 h followed by centrifugation. Immunoprecipitates were washed five times with lysis buffer and subjected to Western blotting. Me 2 SO or 10 M rifampicin was added during immunoprecipitation and washing steps.
Subnuclear Localization Study Using Confocal Microscopy-COS-1 cells were grown on glass coverslips in 6-well plates and cotransfected with either 0.5 g of GFP or GFPPXR along with expression plasmids for PGC-1 or empty vector. After 24 h, cells were treated with 10 M rifampicin or Me 2 SO for 3-4 h in serum-free media. Cells were fixed in 2% paraformaldehyde in phosphate-buffered saline for 20 min, mounted with mounting medium that contained propidium iodide, and subjected to confocal microscopy (Zeiss LSM 510).
Chromatin Immunoprecipitation Assays-ChIP assays were performed as described elsewhere (5). Briefly, HepG2 cells, treated with either 10 M rifampicin or Me 2 SO in serum-free media for 6 h, were incubated with 1% formaldehyde at 25°C for 10 min and subjected to sonication to reduce DNA length to 0.3-1.0 kb. Chromatin was precleared in the presence of 20 l of normal serum, 2 g of salmon sperm DNA, and 80 l of a 25% protein A-agarose slurry. Precleared chromatin samples were subjected to immunoprecipitation at 4°C overnight in the presence of 6 g of antisera for PGC-1 (sc-13067), HNF-4 (sc-8987), or normal rabbit serum. After collecting the complex by incubation with 60 l of a 25% protein A-Sepharose slurry and centrifugation, the beads were washed five times as described (35), and the chromatin immune complex was eluted. After reversing the cross-links, DNA was purified and used as a template in PCR. Semi-quantitative PCR was performed using primer sets specific for the CYP7A1 promoter (forward, 5Ј-gatatctatgcccatcttaaacagg-3Ј, and reverse, 5Ј-gaattcgggaaggatgccactg-3Ј), a control region from ϩ860 to ϩ1160 in the CYP7A1 (forward, 5Ј-gaaccacctctagagaatg-3Ј, and reverse, 5Ј-gaatctccacataaggataac-3Ј), the CYP8B1 promoter (forward, 5Ј-aggtgcaagagctgtctgaa-3Ј, and reverse, 5Ј-atggctatgctcctgcggat-3Ј), and the PEPCK (forward, 5Ј-gcacagagcagacaatcaat-3Ј, and reverse, 5Ј-gacacaacttccaactgact-3Ј).

Treatment with Ligands for PXR or FXR Resulted in Differential Expression of Endogenous CYP7A1, CYP8B1, SHP, and CYP3A4 in HepG2
Cells-Bile acid-activated FXR inhibits bile acid production by suppressing transcription of CYP7A1 and CYP8B1 in an SHP-dependent manner (1,4,36). In contrast, PCN, a specific ligand for mouse PXR, or secondary toxic bile acids such as lithocholic acid (LCA) have been shown to activate PXR or FXR and repress CYP7A1 activity in SHP-independent pathways (8,17). Thus, we first examined whether various ligands for human PXR or FXR differentially modulate expression of endogenous CYP7A, CYP8B1, SHP, or CYP3A4 mRNAs in the human hepatoma HepG2 cells. Cells were treated with a primary bile acid chenodeoxycholic acid (CDCA), a toxic secondary bile acid LCA, rifampicin, or Me 2 SO vehicle, and mRNA levels were measured by semi-quantitative RT-PCR. After treatment with LCA or CDCA, CYP7A1 mRNA levels were decreased to nearly undetectable levels and were substantially reduced by rifampicin treatment (Fig. 1A). Most interestingly, the overall expression pattern of CYP8B1 was similar to that of CYP7A1 (Fig. 1B). CDCA treatment resulted in an increase in SHP mRNA levels as reported previously (1,4,36). However, SHP expression was not induced after rifampicin treatment (Fig. 1C). Expression of the CYP3A4 gene, which is induced by PXR (14), was increased by rifampicin as expected (Fig. 1D). PCN did not modulate the expression of these endogenous genes in human HepG2 cells (data not shown). Expression of ␤-actin that served as a control was not changed by treatment of various ligands (Fig. 1E). These results indicate that the HepG2 cell line provides a suitable system to study suppression of human CYP7A1 and CYP8B1 by ligand-activated PXR, and furthermore, repression by rifampicin is not mediated by induction of SHP.
Because rifampicin is a specific ligand for human PXR, these results suggest that repression of CYP7A1 promoter activity is mediated by rifampicin-activated PXR, and presumably activation by PXR could underlie the suppression by the toxic bile acids. Because the secondary bile acid, LCA, has been reported to trigger activation of multiple nuclear receptors including FXR, CAR, and vitamin D receptor as well as PXR (8,17,18,20), in order to focus on the molecular mechanism of the PXRmediated repression, we have utilized rifampicin as a specific agonist for human PXR in this study.
If the effects of rifampicin are mediated by PXR, then rifampicin should enhance the suppression of CYP7A1 expression by exogenously expressed PXR. HepG2 cells were transfected with the CYP7A1-luciferase plasmid along with increasing amounts of human PXR and were treated with rifampicin or Me 2 SO. The ectopic expression of increasing amounts of PXR expression plasmid alone did not affect CYP7A1 promoter activity, but in cells treated with rifampicin, the promoter activity was strongly suppressed by PXR (Fig. 2B). These results suggest that the human PXR suppresses human CYP7A1 promoter activity in a rifampicin-dependent manner.
Repression of Endogenous CYP7A1 Expression Was Reversed by Overexpression of PXR siRNA-To ensure that rifampicin mediates CYP7A1 repression through the receptor PXR, we analyzed the effects of the inhibition of endogenous PXR expression on CYP7A1 transcription. Overexpression of a pSU-PER vector encoding PXR siRNA efficiently down-regulated endogenous PXR mRNA levels in HepG2 cells (Fig. 2C). Most impressively, decreased mRNA levels of endogenous CYP7A1 resulting from rifampicin treatment were substantially reversed by overexpression of PXR siRNA (Fig. 2D). These results indicate that human PXR is responsible for rifampicin-mediated CYP7A1 repression in hepatic cells.
Ligand-activated PXR Specifically Inhibits HNF-mediated Transactivation of CYP7A1-Previous studies (11,23,28,29,37) show that HNF-4 is a master hepatic activator of genes involved in the bile acid biosynthesis and gluconeogenesis including CYP7A1, CYP8B1, and PEPCK. Thus, we postulated that ligand-activated PXR might inhibit HNF-4 signaling in hepatic cells. In HepG2 cells, overexpression of HNF-4 resulted in a 2-3-fold increase in CYP7A1 promoter activity (Fig. 2E,  lanes 1 and 2). Most interestingly, increasing amounts of PXR in the presence of rifampicin progressively decreased CYP7A1 promoter activity to levels less than that without the exogenous HNF-4 expressed (Fig. 2E, lanes 3-6). In contrast, increasing amounts of PXR did not attenuate ER transactivation of the 4ERE-tk promoter activity (Fig. 2F). These results suggest that ligand-activated human PXR specifically interferes with HNF-4 transactivation of the CYP7A1 promoter activity.
PXR Directly Interacts with PGC-1 in Vitro-The thermogenic coactivator PGC-1 has been implicated in cholesterol, as well as glucose metabolism, by its ability to enhance HNF-4 transcriptional activity (23,28,29,37). Therefore, we postulated that ligand-activated PXR might interfere with HNF-4 signaling by competing for binding to PGC-1. We first tested whether FLAG-PXR could interact with endogenous PGC-1 present in mouse liver nuclear extracts. Mouse liver nuclear extracts were immunoprecipitated by FLAG-PXR. PGC-1 was present in the immunoprecipitates as detected by Western blotting using PGC-1 antisera (Fig. 4A). These results indicate that PXR can interact either directly or indirectly with endogenous PGC-1 in mouse liver in vitro.
Next, we asked whether PGC-1 could directly interact with recombinant PXR or HNF-4 by GST pull down assays. 35 S-PXR and 35 S-HNF-4 were incubated with GST-PGC-1. Although neither PXR nor HNF-4 interacted with GST, both PXR and HNF-4 interacted with PGC-1 and, most surprisingly, rifampicin treatment did not increase the interaction between PXR and PGC-1 (Fig. 4B, a and c). Conversely, we also examined whether 35 S-PGC-1 could interact with GST-HNF-4 or FLAG-PXR (bound to M2-agarose). Consistent with the results above, specific interactions of PGC with HNF-4 and PXR were ob-served (Fig. 4B, b and d). These results indicate that PGC-1 can interact directly with PXR or HNF-4 in vitro.
PGC-1 Interacts with PXR LBD in Cells in a Rifampicin-dependent Manner-Unexpectedly, the interaction of PXR and PGC-1 in vitro was independent of rifampicin. Ligand-dependent interaction may require cell-specific factors and/or posttranslational modification in cells. Therefore, we examined whether human PXR can interact with endogenous PGC-1 in HepG2 cells in a rifampicin-dependent manner. HepG2 cells were transfected with expression plasmids for either Gal4DBD or Gal4DBD full-length human PXR and then treated with Me 2 SO or rifampicin. The cell extracts were precipitated with either IgG or antisera against Gal4DBD, and association of the endogenous PGC-1 with Gal4DBD-PXR was detected by Western blotting using PGC-1 antisera. PGC-1 was not detected in the anti-Gal4DBD precipitates from the cells transfected with the Gal4DBD plasmid (Fig. 4C, a). PGC-1 was barely detectable in the anti-Gal4DBD precipitates from the HepG2 cells transfected with Gal4DBD-PXR, but the amount of PGC-1 was markedly increased after treatment of cells with rifampicin, in contrast to rifampicin-independent interaction in vitro (Fig. 4C, b).
Because the AF2 domain of the LBD of nuclear receptors is critical for interaction with coactivators (40), we further tested whether the interaction between PXR and PGC-1 occurs through this domain. HepG2 cells were transfected with an expression vector for a Gal4DBD-human PXR-LBD fusion protein (amino acids 107-434) (13) and then treated with Me 2 SO or rifampicin. The cell extracts were precipitated with either IgG or antisera against Gal4DBD, and association of PGC-1 with Gal4DBD-PXR-LBD was detected by Western blotting using PGC-1 antisera. The amount of PGC-1 in the immunoprecipitates was strikingly increased after treatment of cells with rifampicin (Fig. 4D). These results suggest that PXR interacts with PGC-1 through its C-terminal LBD domain in a rifampicin-dependent manner in cells.
Subnuclear Localization of PXR in Nuclear Speckles Is Dependent on PGC-1 and Rifampicin Treatment-PGC-1 has been reported to be localized in the nucleus and to be concentrated in nuclear speckles, where splicing factors are colocal-

FIG. 3. PXR/RXR does not inhibit binding of HNF-4 to the human CYP7A1 promoter in vitro.
A, 32 P-labeled BARE II oligonucleotide was incubated with partially purified HNF-4, PXR, or RXR as described under "Experimental Procedures." The positions of the HNF-4 with the BARE II probe and the BARE II probe are indicated by arrows. B, partially purified PXR and RXR were preincubated for 5 min with increasing amounts (6.25, 12.5, 25, 50, and 100 molar excess) of unlabeled double strand oligonucleotides containing either the DR4 motif from the PBRU region of CYP2B1 (ds DR4) or the BARE II region of CYP7A1 (ds BARE II). 32 P-Labeled PBRU probe containing the DR4 motif was added to the binding reaction, and the incubation was continued for 15 min. After incubation, complexes were analyzed by electrophoresis on native polyacrylamide gels. The positions of the PXR⅐RXR⅐probe complex and PBRU probe are indicated by arrows.
ized (41). If rifampicin-activated PXR and PGC-1 interact in cells, the distribution of PXR after rifampicin treatment might be expected to resemble that of PGC-1 in nuclear speckles. Therefore, we have examined the distribution of GFP fulllength human PXR in COS-1 cells. In control experiments with GFP alone, rifampicin treatment and expression of PGC-1 had no detectable effect on the distribution of fluorescence (Fig. 5, A  and B). Consistent with previous studies (42), GFP-PXR was predominantly found diffusely distributed in the nucleus in cultured cell lines, and treatment with rifampicin resulted in a slightly granular appearance of fluorescence (Fig. 5C). Although coexpression of PGC-1 did not substantially change the uniform distribution of PXR in the nucleus, coexpression of PGC-1 with rifampicin treatment resulted in a dramatic redistribution of GFP-PXR into nuclear speckles (Fig. 5D). To provide reliable estimates of the intranuclear localization in nuclear speckles, we randomly counted 60 cells for each experimental group. More than 60% of cells cotransfected with GFP-PXR and PGC-1 in the presence of rifampicin showed nuclear speckles, whereas only 2-5% of cells from other experimental groups showed the speckles. These results further support the conclusion that PXR and PGC-1 interact in cells in a rifampicin-dependent manner.
Inhibitory Cross-talk between PXR and HNF-4 by Targeting a Common Coactivator PGC-1-Our data suggest that rifampicin-activated PXR interacts with PGC-1 in cells and, thus, may interfere with the transcriptional activity of HNF-4 by competing for binding to PGC-1. We tested whether PXR-mediated transactivation was enhanced by overexpression of PGC-1. Expression of human PXR in HEK293 cells activated promoter activity of 3(DR4)-tk-luc in the presence of rifampicin, and increasing amounts of PGC-1 substantially enhanced the promoter activity in a dose-dependent manner (Fig. 6A,  lane, 3-6). Most interestingly, enhancement of PXR transactivation of this promoter by PGC-1 was progressively decreased by the expression of increasing amounts of exogenous HNF-4 (Fig. 6B, lanes 4 -7). Conversely, HNF-4-mediated transactivation of the CYP7A1 promoter was further enhanced by exogenous PGC-1 in HepG2 cells (Fig. 7C, lanes 1-4), and this enhancement was eliminated by increasing amounts of human PXR expression plasmid in the presence of rifampicin (Fig. 6C,  lane, 4 -7). These results show that PGC-1 is a common coactivator target for both PXR and HNF-4 and suggest that competition for binding to this coactivator may underlie the mutual antagonism of these two factors.
Ligand-activated PXR Suppresses the HNF-4/PGC-1 Pathway-To further test whether PXR specifically suppresses HNF-4 signaling by competing for the binding to PGC-1, we utilized a Gal4 reporter system. HepG2 cells were cotransfected with a reporter, 5Gal4-TATA-luc, and increasing amounts of expression plasmids, either Gal4DBD or Gal4DBD-HNF-4. Although expression of the Gal4DBD did not significantly modulate promoter activity, expression of Gal4DBD-HNF-4 substantially activated the Gal4 reporter activity (Fig.   FIG. 4

. Human PXR interacts with the coactivator PGC-1 in vitro, and the LBD of human PXR interacts with PGC-1 in a rifampicin-dependent manner in cells.
A, FLAG-PXR bound to M2-agarose was incubated with mouse liver nuclear extracts and subjected to pull down assays as described under "Experimental Procedures." Input indicates endogenous PGC-1 present in 5% of total liver nuclear extracts used in each reaction. After extensive washing, the proteins bound to FLAG or FLAG-tagged PXR were analyzed by Western blotting (WB) using PGC-1 antisera. B, GST-PGC-1 (a and c) and GST-HNF-4 (d) or GST as a negative control were immobilized on glutathione-Sepharose in a slurry. FLAG-PXR was bound to M2-agarose (b). The bound GST or FLAG proteins were incubated with 35 S-labeled human PXR (a), PGC-1␣ (b and d), or HNF-4 (c) and in the presence of either Me 2 SO (DMSO) or 100 M rifampicin (Rif) as indicated. After washing, proteins bound to the M2-agarose or glutathione-Sepharose were detected by SDS-PAGE and autoradiography. Input indicates 20% of the radioactive proteins present in each reaction. C and D, HepG2 cells were transfected with either Gal4DBD (a) or with Gal4DBD full-length human PXR (b) (C) or with Gal4DBD-PXR-LBD (D) by electroporation. Then 48 h after transfection, the cells were treated with 10 M rifampicin in serum-free media for 2 h. Cell extracts were immunoprecipitated (IP) with either IgG or antisera against Gal4DBD in the presence of Me 2 SO or 10 M rifampicin. After SDS-PAGE, association of the endogenous PGC-1 with the anti-Gal4DBD precipitate was detected by Western blotting using PGC-1 antisera. Input indicates endogenous PGC-1 present in 5% of total HepG2 cell lysates used in each immunoprecipitation, and consistent results were obtained from two independent coimmunoprecipitation assays. To ensure equal amounts of were precipitated, Western blotting using Gal4DBD antisera was also performed as indicated (D). 7A). This HNF-4-mediated transactivation was dramatically enhanced by PGC-1 (Fig. 7B). We tested whether this enhancement by PGC-1 is blunted by expression of ligand-activated PXR, and furthermore, whether overexpression of PGC-1 overcomes the inhibitory interaction between PXR and HNF-4. Enhancement of HNF-4 transactivation by PGC-1 was substantially attenuated by expression of PXR in the presence of rifampicin (Fig.  7C, lanes 3-6), and increasing amounts of PGC-1 reversed this inhibitory effect (Fig. 7C, lanes 7-9). These results further support the idea that ligand-activated PXR specifically interferes with HNF-4/PGC-1-signaling pathway in cells.
Rifampicin Treatment Does Not Inhibit Binding of HNF-4 to the Native CYP7A1, CYP8B1, or PEPCK Promoters but Results in the Dissociation of PGC-1 and Concomitant Gene Repression-Finally, we tested whether HNF-4 binding was inhibited by rifampicin-activated PXR in vivo by ChIP analyses in HepG2 cells. We also wished to determine whether PGC-1 was associated with the HNF-4-binding sites at the CYP7A1, CYP8B1, or PEPCK genes and the effect of rifampicin treatment on this association. First, we examined the effect of rifampicin on the binding of endogenous HNF-4 to the native promoters of these genes, which all contain functional HNF-4binding sites (1,23,26,43). The amount of CYP7A1 promoter sequence precipitated by antisera to HNF-4 was not changed regardless of rifampicin treatment (Fig. 8A). In contrast, little promoter sequence was present in control precipitates with IgG, and the control sequence from ϩ860 to ϩ1160 of CYP7A1 was not enriched by immunoprecipitation with HNF-4 antisera. These results indicate that rifampicin treatment did not inhibit HNF-4 binding to the native CYP7A1 promoter consistent with the inability of PXR to block HNF-4 binding in vitro. Similar results were obtained for CYP8B1 and PEPCK genes, indicating that in vivo binding of HNF-4 to these promoters was not inhibited by rifampicin treatment (Fig. 8, B and C). However, most interestingly, the amount of promoter sequence in chromatin precipitated by PGC-1 antisera was substantially decreased in rifampicin-treated cells for all three promoters (Fig. 8, A-C). Most importantly, like CYP7A1 and CYP8B1 (Fig. 1, A and B), mRNA levels of PEPCK in HepG2 cells were decreased by rifampicin treatment (Fig. 8D). These results together indicate that binding of HNF-4 to the endogenous CYP7A1, CYP8B1, or PEPCK gene promoters is not inhibited, instead PGC-1 is dissociated from these promoters after rifampicin treatment, resulting in gene repression. DISCUSSION Bile acids are synthesized from cholesterol in the liver and secreted into the small intestine to solubilize and facilitate the uptake of dietary lipids. Bile acid production in liver is also the sole means to eliminate excess cholesterol from the body. The majority of bile acids in the intestine is returned to the liver by the enterohepatic portal circulation, and about 5% of bile acids are excreted from the body (2). Despite their beneficial roles, excessive bile acids are potentially harmful. LCA is a hydrophobic secondary bile acid that is synthesized in the intestine by bacterial 7␣ dehydroxylation of a primary bile acid CDCA (15). LCA is a toxic bile acid that causes an impairment of bile flow and accumulation of bile acids and biliary toxins in the liver (10,15). Accumulation of toxic bile acids in liver and intestine is associated with intrahepatic cholestasis, liver cirrhosis, liver cancer, and colon cancer (8,17,18). The production of bile acids from cholesterol is therefore tightly regulated by inhibitory feedback mechanisms that also control cholesterol levels in the body.
Recent studies show that toxic bile acids activate PXR and regulate transcription of genes involved in biosynthesis, trans-FIG. 5. COS-1 cells were transiently cotransfected with expression plasmids for either GFP or GFP full-length human PXR and expression plasmids for either PGC-1 or empty vector. Twentyfour hours after transfection, cells were treated with Me 2 SO (DMSO, Ϫ) or 10 M of rifampicin (ϩ) for 2-4 h in serum-free media. The cells were fixed, mounted with mounting medium, and imaged by confocal microscopy. The bars inside panels A-D represent 5 m. For reliable estimate, 60 cells were randomly counted for each experiment, and representative pictures from two independent experiments are presented.
FIG. 6. Functional inhibitory crosstalk between PXR and HNF-4 by targeting the common coactivator PGC-1. A and B, HEK293 cells were cotransfected with 100 ng of 3DR4-tk-luc, 100 ng of CMV-␤-gal, 5 ng of CMV-human PXR, and increasing amounts of either pSV40-sport-PGC-1 (0, 0.1, 0.2, 0.5, and 1 g) in A or pcDNA3HNF-4 (0, 0.1, 0.5, 1, and 1.5 g) in B. C, HepG2 cells were cotransfected with 200 ng of 371h-CYP7A1-luc, 50 ng of pcDNA3HNF-4, 100 ng of pSV40-sport-PGC-1, and increasing amounts (0, 5, 10, and 50 ng) of CMV-FLAG-human PXR expression plasmids. The transfected cells were treated with 10 M rifampicin for 20 h and harvested for reporter assays. The S.E. was calculated from six independent determinations. port, and metabolism of bile acids, such as CYP7A1, oatp, or CYP3A genes, respectively (8,(17)(18)(19). In addition to PXR, the orphan nuclear receptors, FXR, CAR, and vitamin D receptor, have also been implicated as biosensors for toxic bile acids (8,(17)(18)(19)(20). Because the secondary bile acids have been reported to trigger activation of these multiple nuclear receptors, in order to delineate the molecular mechanism of the PXR-mediated suppression of bile acid biosynthesis, in this study we have utilized rifampicin as a specific ligand for human PXR. These studies also provide a molecular mechanism for the clinical effects of rifampicin in cholestasis.
The thermogenic coactivator PGC-1 is a key regulator in energy balance and carbohydrate metabolism (27). PGC-1 stimulates genes important to mitochondrial function and oxidative metabolism of fuels (23,37). In addition to oxidative phosphorylation processes, numerous studies (28, 44 -46) show that PGC-1 is a versatile coactivator for many nuclear receptors and is implicated in diverse biological activities. PGC-1 activates HNFmediated transactivation of PEPCK and glucose-6-phosphatase genes that play crucial roles in hepatic gluconeogenesis (23, 26, ChIP assays were performed as described under "Experimental Procedures." HepG2 cells were treated with Me 2 SO (Ϫ) or 10 M of rifampicin (RIF) (ϩ) in serum-free media for 6 h and subjected to formaldehyde cross-linking. Soluble chromatin was prepared by sonication. Precleared chromatin solution was immunoprecipitated by antibodies against HNF-4, PGC-1, or normal rabbit serum (NS), and precipitated DNA was further analyzed by semi-quantitative PCR using primer sets specific for the promoter or control regions as indicated by the arrows (A-C). Reproducible results for association with the promoters for HNF-4 and PGC-1 were shown from 4 and 2 independent assays, respectively. D, total RNA was isolated from HepG2 cells treated with Me 2 SO (Ϫ) or rifampicin (ϩ) in serum-free media for 24 h and subjected to RT-PCR. Semi-quantitative PCR was performed using a primer set specific for human PEPCK or ␤-actin.  29,37,47). PGC-1 has also been reported to regulate triglyceride metabolism by inducing FXR gene expression and enhancing transcriptional activity of FXR (46). A recent study (24) also suggests that PGC-1 plays a role in cholesterol/bile acid metabolism by regulating expression of the CYP7A1 gene, implying that the PGC-1 may be a coactivator for both PXR and HNF-4.
Coimmunoprecipitation assays show that PXR interacts with the coactivator PGC-1 through its C-terminal ligand binding domain in a rifampicin-dependent manner in cells. Unknown cell-specific factor(s) appear to be important for the interaction because interaction in vitro is independent of rifampicin. In confocal microscopic studies, PXR, as reported for PGC-1 (41), was localized in nuclear speckles in cells coexpressing PGC-1 and GFP-PXR in the presence of rifampicin, providing further evidence for a rifampicin-dependent interaction between PXR and PGC-1. Both gel mobility shift assays in vitro and chromatin immunoprecipitation assays in vivo showed that activated PXR did not affect binding of HNF-4 to CYP7A1, CYP8B1, and PEPCK promoters. Instead the association of PGC-1 with HNF-4 bound to these genes was inhibited, which is associated with gene repression and explains the observed decreases in mRNA levels of those genes after rifampicin treatment. These molecular effects would directly lead to reduction of bile acid levels in the liver by suppressing bile acid biosynthesis. In addition, the similar finding with the PEPCK gene suggested that glucose metabolism would also be modulated by activation of PXR.
In contrast to the antagonism between HNF-4 and PXR for CYP7A1 regulation, a previous study (48) showed that HNF-4 potentiated PXR-mediated transactivation of the CYP3A gene. This enhancement was mediated through an HNF-4-binding site at the far upstream enhancer region at the CYP3A gene. The major difference from the current study is that both PXR and HNF-4 bind to the regulatory region of CYP3A4 so that they are likely to be additive or synergistic. However, in other HNF-4-dependent genes that lack functional PXR-binding sites in their promoter or enhancer regions, PXR could be inhibitory by competing for binding to common coactivators for HNF-4, such as PGC-1. Thus, it is possible that PXR has the ability to activate or repress gene transcription depending on the promoter context. Similar dual gene activation and repression have been reported in studies of cross-talk between HNF-1 and HNF-4 (32). HNF-1 activates genes containing promoters with both HNF-1-and HNF-4-binding sites. However, HNF-1 negatively regulates expression of other HNF-4 target genes that lack HNF-1-binding sites in their promoter region, through the direct interaction with HNF-4.
In conclusion, we demonstrate that ligand-activated human PXR interferes with HNF-4 signaling by targeting a common coactivator PGC-1. Clearly, the response of hepatocytes to foreign and endogenous substances that activate PXR is more than just activation of hepatic enzymes for detoxification and metabolism of those substances. The secondary effects on HNF-4 signaling by competing for the common coactivator PGC-1 could impinge on fundamental cellular pathways such as cholesterol and glucose metabolism by modulating expression of CYP7A1 and PEPCK genes.