CYP3A4 is the most abundant cytochrome P450 enzyme expressed in the liver and small intestine and plays a crucial role in metabolism and detoxification of xenobiotics and endobiotics (Guengerich, 1999). CYP3A4 has very broad substrate specificity and metabolizes over 50% of clinical drugs. The expression levels of CYP3A4 vary widely among individuals (Lehmann et al., 1998). Compounds that activate human PXR (NR1I2) also induce CYP3A4 and drug metabolism activity (Bertilsson et al., 1998; Kliewer et al., 1998; Lehmann et al., 1998; Goodwin et al., 2002). An antibiotic, rifampicin, is a highly efficacious human PXR agonist (Kliewer et al., 2002) and has been used to treat pruritus of primary biliary cirrhosis patients for many years, although its molecular mechanism of drug action is not known (Bachs et al., 1989; Hofmann, 2002). The ligand-activated PXR forms a heterodimer with RXR and binds to the AGGTCA-like direct repeats spaced by 3, 4, or 5 bases (DR3, DR4, and DR5), and the everted repeats separated by 6 or 8 bases (ER6 and ER8) located in the PXR target genes (Kliewer et al., 1998). In the human CYP3A4 promoter, PXR binds to an ER6 located in the proximal promoter and a DR3 and ER6 in the distal xenobiotic responsive enhancer module (XREM) located approximately 8 kilobases upstream of the transcription start site (Bertilsson et al., 1998; Lehmann et al., 1998; Goodwin et al., 1999). The maximal induction of the CYP3A4 gene by PXR requires both the proximal ER6 and the distal XREM (Goodwin et al., 1999).

Lithocholic acid (LCA) is the most efficacious bile acid that activates PXR and induces CYP3A4 to catalyze 6-hydroxylation of LCA to hyodeoxycholic acid (Staudinger et al., 2001; Xie et al., 2001). Feeding LCA causes liver damage in Pxr null mice, but transgenic mice expressing a human PXR are protected from bile acid toxicity and cholestasis (Staudinger et al., 2001). PXR inhibits cholesterol 7α-hydroxylase (CYP7A1), the regulatory gene in bile acid synthesis, but induces organic anion transport protein 2, a sinusoidal Na⁺-independent bile acid transporter (Staudinger et al., 2001). Nuclear receptors interact with their coactivators such as steroid receptor coactivator-1 (SRC-1) or corepressors and activate or inhibit their target genes. We reported previously that rifampicin-activated PXR interacts with hepatocyte nuclear factor 4α (HNF4α) and blocks HNF4α recruitment of peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α), and results in inhibition of the human CYP7A1 gene (Li and Chiang, 2004). HNF4α plays a vital role in liver development and regulation of bile acid synthesis, lipid homeostasis, and xenobiotic response (Stroup and Chiang, 2000; Hayhurst et al., 2001; Kamiya et al., 2003). Both PXR and HNF4α interact with PGC-1α, which is a nuclear receptor coactivator induced by glucagon and glucocorticoid to regulate gluconeogenesis and bile acid synthesis (De Fabiani et al., 2003; Puigserver et al., 2003; Shin et al., 2003). HNF4α induces PXR gene transcription in fetal liver and enhances PXR induction of CYP3A4 (Kamiya et al., 2003; Tirona et al., 2003). It has been reported recently that SHP interacts with rifampicin-activated PXR and inhibits the CYP3A4 gene expression in vitro (Ourlin et al., 2003). SHP is a negative nuclear receptor that is induced by bile acid-activated farnesoid X receptor (FXR) and inhibits the CYP7A1 gene (Goodwin et al., 2000; Lu et al., 2000). Therefore, the bile acid-induced FXR/SHP pathway could counteract the bile acid activation of the CYP3A4 gene transcription by PXR. The physiological role of SHP in regulation of CYP3A4 needs to be further studied.

In this study, we investigated the roles of human PXR, HNF4α, and SHP on the regulation of human CYP3A4 gene transcription. Our results suggest that CYP3A4 is subjected to a complex regulation by nuclear receptors and coactivators, and that maximal induction of CYP3A4 gene expression requires concomitant inhibition of SHP gene transcription by PXR.

Materials and Methods

Cell Culture. Human hepatoblastoma HepG2 cells were obtained from The American Type Culture Collection (Manassas, VA). Primary human hepatocytes were isolated from human donors (HH1115, 22 years, male; HH1117, 68 years, female; HH1119, 29 years, female) and were obtained through the Liver Tissue Procurement and Distribution System (LTPADS) of the National Institutes of Health (S. Strom, University of Pittsburgh, Pittsburgh, PA). Cells were cultured and maintained as described previously (Li and Chiang, 2004).

Plasmids. Human CYP3A4/luciferase construct (p3A4-5′dDR3/dER6/pER6-LUC) was obtained from Marie-Jose Vilarem (Institut Federatif de Recherche 24, 34293 Montpellier, France). This reporter contains an HNF4α binding site (DR1) and PXR binding sites (DR3 and ER6) in the distal XREM (dXREM, –7800/–7600), which was linked to a 5′-flanking fragment containing a proximal ER6 (pER6) of the human CYP3A4 gene (Drocourt et al., 2002). A human CYP7A1/Luc reporter (ph-298/Luc) was constructed as described previously (Li and Chiang, 2004). Human SHP/luciferase reporter construct (SHP-luc) containing a 2-kilobase promoter sequence and expression plasmid (pCDM8-mSHP) were provided by Yoonkwang Lee (Baylor College of Medicine, Houston, TX). Expression plasmids for SHP (pcDNA3-HA-SHP), SRC-1 (pCR3.1-hSRC-1A), PGC-1α (pcDNA3/HA-PGC-1α), human RXR (pcDNA3-hRXR), rat HNF4α (pCMX-rHNF4α), and human PXR (pSG5-hPXR) were obtained from H. S. Choi (Chungnam National University, Gwang ju, Korea), M. J. Tsai (Baylor College of Medicine), A. Kralli (The Scripps Research Institute, La Jolla, CA), R. Evans (Scripps Research Institute), W. Chin (Eli Lily Research Laboratories, Indianapolis, ID), and S. Kliewer (University of Texas Southwestern Medical Center, Dallas, TX), respectively. For mammalian hybrid assays, the reporter 5XUAS-TK-Luc contains five copies of the upstream activating sequence (UAS) fused upstream of a thymidine kinase (TK) minimum promoter and the luciferase reporter gene. The GAL4 and VP16 fusion plasmids contained the ligand-binding domain (LBD) of nuclear receptors or nuclear receptor-interacting domain (RID) of coactivators fused to Gal4-DBD (DNA-binding domain) or VP16-AD (activation domain) vector. Mammalian two-hybrid fusion plasmids used were: pCMX-VP16-rHNF4 (a.a. 125–364) from D. Moore (Baylor College of Medicine); Gal4-hPXR-LBD (a.a. 107–434), VP16-SRC-1-RID (a.a. 595–780), and VP16-hPXR-LBD (a.a. 107–437) from A. Takeshita (Toranomon Hospital, Tokyo, Japan); and Gal4-PGC-1α (full-length) from A. Kralli (Scripps Research Institute). Gal4-HNF4α fusion plasmids pBx-HNF4α-FL (fulllength) was obtained from I. Talianidis (Institute of Molecular Biology and Biotechnology Foundation for Research and Technology, Hellas, Herakleion Crete, Greece). For glutathione S-transferase (GST) fusion protein expression, pGEX-4T-1 plasmid encoding the GST protein was obtained from Amersham Pharmacia Biotech (Little Chalfont, Buckinghamshire, UK). The pET23a-GST-HNF4α plasmid encoding a fusion protein of GST with full-length human HNF4 was a kind gift from Todd Leff (Wayne State University, Detroit, MI). GST-hPXR-LBD was cloned by inserting PCR-amplified human PXR ligand-binding domain (a.a. 241–434) into pGEX-4T-1 BamH1/XhoI sites. GST-mSHP containing full-length SHP was obtained from D. Moore (Baylor College of Medicine).

Quantitative Real-Time PCR Assay. Primary human hepatocytes were maintained in six-well plates. Cells were treated with increasing amounts of rifampicin (1–10 μM) (Sigma, St. Louis, MO) for 24 h. Total RNA was isolated from the cells using Tri-Reagent (Sigma) according to the manufacturer's protocol. Reverse transcription reactions were performed using a RETROscript kit according to the manufacturer's instructions (Ambion Inc., Austin, TX). For real-time PCR, samples were prepared according to the PCR Taqman Universal Master Mix 2X protocol (Applied Biosystems, Foster City, CA). Amplification of human ubiquitin C was used as an internal reference gene. Quantitative PCR analysis was conducted on the ABI 7500 System software. Relative mRNA expression was quantified using the comparative C_T (Ct) method according to the ABI manual. Assay-on-Demand PCR primers and Taqman MGB probe mix used were human CYP3A4 (Cat. No.: Hs00604506_m1), human CYP7A1 (Cat. No. Hs00167982_ml), human SHP (Cat. No. Hs00222627_ml), and human ubiquitin C (Cat. No. Hs00824723_m1).

Transient Transfection Assay. HepG2 cells were grown to ∼80% confluence in 24-well tissue culture plates. Reporter plasmid (0.2 μg) was transfected with expression plasmid (0.1 μg) using LipofectAMINE 2000 reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The pCMV-galactosidase plasmid (0.05 μg) was transfected as an internal standard for normalizing the transfection efficiency in each assay. In some samples, empty expression vectors were added to equalize the total amounts of plasmid DNA transfected in each assay. Cells were treated with vehicle (DMSO) or 10 μM rifampicin in serum-free media for 40 h and assayed for luciferase reporter activity using the Luciferase Assay System (Promega, Madison, WI). Luciferase activity was normalized by dividing the relative light units by β-galactosidase activity. Each assay was done in triplicate and individual experiments were repeated at least three times.

Site-Directed Mutagenesis. Mutations (shown in Fig. 3B) were introduced into the HNF4α binding site in dXREM of CYP3A4/luciferase construct (p3A4-5′dDR3/dER6/pER6-LUC) using the PCR-based QuikChange Site-directed Mutagenesis Kit (Stratagene, La Jolla, CA). Two complementary oligonucleotides containing the mutations were used as PCR primers. PCRs were set up according to the manufacturer's instructions using 50 ng of template DNA and 125 ng of primers. PCR cycling parameters were set as follows: denaturing at 95°C for 2 min, followed by 18 cycles at 95°C for 30 s, 52°C for 1 min, and 68°C for 18 min. The reaction mixture was digested by DpnI for 2 h to remove the template DNA and transformed into XL1-Blue super competent cells (Stratagene) for selection of mutant clones. Mutations were confirmed by DNA sequencing.

Electrophoretic Mobility Shift Assay (EMSA). Double-stranded synthetic probes containing the HNF4α-binding site were labeled with [³⁵S]dATP by filling-in reaction using the Klenow fragment of DNA polymerase I. Nuclear receptor HNF4α was synthesized in vitro using Quick-coupled Transcription/Translation Systems (Promega) programmed with receptor expression plasmids according to the manufacturer's instruction. EMSA was performed as described in detail previously (Li and Chiang, 2004).

Immunoblotting Analysis. HepG2 cells were cultured in 100-mm dishes to 80% confluence. Cells were transfected with pcDNA3-HA-SHP plasmid (5 or 10 μg) and incubated in serum-free media for 40 h. Cell lysates were analyzed by SDS-polyacrylamide gel electrophoresis. Antibody against HA-tag (Santa Cruz Biotechnology, Santa Cruz, CA) were used for Western blotting and detected by an ECL Western blotting detection kit (Amersham Biosciences, UK).

Chromatin Immunoprecipitation (ChIP) Assay. HepG2 cells were grown in 100-mm culture dishes to 80% confluence. HA-PGC-1α (10 μg) or HA-SHP plasmid was transfected into HepG2 cells. Cells were treated with 10 μM rifampicin or vehicle (DMSO) in serum-free media for 40 h. ChIP assays were performed using a ChIP Assay kit (Upstate Cell Signaling Solutions, Char-lottesville, VA) according to the manufacturer's protocol and as described previously (Li and Chiang, 2004). Ten percent of the cell lysates before immunoprecipitation were saved and used as “input”. The cross-linked DNA-protein complexes were precipitated by incubating the cell lysates with 10 μg of antibodies or nonimmune IgG (Santa Cruz Biotechnology), followed by protein A-agarose beads. The antibodies used were goat anti-HNF4α (C-19), Cat. No. sc-6556, lot no.I1203; rabbit anti-HNF4α (H171), Cat. No. sc-8987, lot no. K0204; goat anti-PXR (A-20), Cat. No. sc-7737, lot no. J0704; rabbit anti-SRC-1 (M341), Cat. No. sc-8995, lot no. C3004; goat anti-PGC-1α (P-19), Cat. No. sc-5815, lot no. I141; goat Normal IgG, Cat. No. sc-2027, lot no. E1204; and rabbit Normal IgG, Cat. No. sc-2028, lot no. F0304. The beads were washed and samples eluted with 250 μl of ChIP elution buffer. DNA fragments containing either dXREM or pER6 were PCR-amplified for 35 cycles and analyzed on a 2% agarose gel. The PCR primers used were dXREM region (dDR3/ER6, –7939 to –7576), forward-GCAGGTCCCCTGGAAAGTCAC, reverse-CACCTGGGGTCAACACAGGAC; proximal ER6 region (–277 to –62), forward-ATGCCAATGGCTCCACTTGAG, and reverse-CTGGAGCTGCAGCCAGTAGCAG; and 2nd intron region (+1460 to + 1849), forward-GTTACTAGCTTCCTTCATCAGTTG, and reverse-CATTCCAGCCTGGGCAACAAG.

GST Pull-Down Assay. GST fusion protein isolation and GST pull-down assays were described previously (Li and Chiang, 2004). The ³⁵S-labeled proteins were generated using their respective expression plasmid in the TNT rabbit reticulocyte system (Promega) in the presence of ³⁵S-labeled methionine. GST fusion proteins (∼1 μg) bound to the glutathione beads were added to 5 μl of the ³⁵S-labeled proteins. In competition assays, 50 μl of unlabeled TNT lysates incubated with pcDNA3 or the indicated expression plasmids were added to the mixture. The beads were washed and the proteins eluted for analysis on an SDS-polyacrylamide gel. One microliter of ³⁵S-labeled protein was loaded as input.

Statistical Analysis. Data were plotted as means ± standard deviation. Statistical analyses of treated versus untreated controls (vehicle or empty plasmids) were performed using Student's t test. A p < 0.05 is considered statistically significant.

Results

Rifampicin Induces CYP3A4 but Inhibits SHP mRNA Expression in Primary Human Hepatocytes. We assayed the effect of rifampicin on CYP3A4, SHP, and CYP7A1 mRNA expression in primary human hepatocytes (Fig. 1). Rifampicin dose dependently induced CYP3A4 mRNA expression by almost 150-fold at 10 μM. Interestingly, rifampicin strongly reduced SHP mRNA levels at a concentration as low as 1 μM. This is in contrast to the induction of SHP expression by bile acids via activation of FXR. As a positive control, rifampicin strongly inhibited CYP7A1 mRNA expression as previously reported (Li and Chiang, 2004). These data suggest that rifampicin activates PXR to induce CYP3A4, but reduces SHP mRNA expression.

HNF4α, SRC-1, and PGC-1α Coactivate, whereas SHP Inhibits PXR Induction of HumanCYP3A4Gene Transcription. The marked induction of CYP3A4 by rifampicin prompted us to investigate the molecular mechanism of PXR induction of the human CYP3A4 gene. Nuclear receptor activity is enhanced by coactivators and inhibited by corepressors. SRC-1 and PGC-1α are known to enhance HNF4α and PXR trans-activation activity (Li and Chiang, 2004). We reported previously that rifampicin-activated PXR inhibited CYP7A1 gene transcription by interacting with HNF4α and blocking PGC-1α recruitment to CYP7A1 chromatin (Li and Chiang, 2004). We performed reporter assays to study the effect of HNF4α, PGC-1α, and SRC-1 on PXR regulation of the human CYP3A4 gene. We used a CYP3A4 reporter that contains a well characterized distal XREM (–7800/–7600, containing an HNF4α and two PXR binding sites) linked to a 5′-flanking fragment (–262/+11, containing a PXR site). Figure 2A shows that cotransfection of this CYP3A4-Luc reporter plasmid with PXR and RXRα expression plasmids in HepG2 cells stimulated the basal reporter activity by about 3-fold. Rifampicin (10 μM) further stimulated CYP3A4 reporter activity by 4-fold. Surprisingly, we found that HNF4α did not have any effect on the CYP3A4 reporter activity. Cotransfection with both HNF4α and PXR caused a 2.5-fold stimulation of the CYP3A4 reporter activity stimulated by PXR alone when rifampicin was added. Figure 2B shows that SRC-1 and/or PGC-1α stimulated PXR induction of CYP3A4 reporter activity by 2-fold in the absence of rifampicin. Addition of rifampicin further stimulated reporter activity. The maximal induction (about 130-fold) of the CYP3A4 basal reporter activity by rifampicin was achieved when PXR, HNF4α, SRC-1, and PGC-1α were transfected. It is noted that inclusion of HNF4α in the assay greatly enhanced rifampicin induction of the reporter activity, suggesting that PXR and HNF4α synergy is important for induction of CYP3A4. These data suggest that to maximize rifampicin induction of CYP3A4 gene transcription requires HNF4α and coactivators.

HNF4α Does Not Directly RegulateCYP3A4Gene Transcription. Our result that HNF4α alone did not affect CYP3A4 reporter activity is in contrast to a previous report that HNF4α stimulated CYP3A4-Luc reporter activity (Tirona et al., 2003). EMSA, shown in Fig. 3A, confirmed that the putative HNF4α binding site in the dXREM of the CYP3A4 gene was able to bind HNF4α as the oligonucleotide probes designed according to a canonical DR1 motif, and the HNF4α binding sites in the CYP8B1 and CYP7A1 genes. Excess of unlabeled DR1 probe was able to compete out HNF4α binding to the radiolabeled 8B1, 7A1, and 3A4 probes. We then performed site-directed mutagenesis to mutate the putative HNF4α binding site in the CYP3A4 reporter plasmid. The mutant 3A4 probe did not bind HNF4α (data not shown). Figure 3B shows that mutations of the HNF4α site in the CYP3A4 reporter plasmid did not affect the basal reporter activity and the synergistic stimulation of reporter activity by PXR and HNF4α. These results suggested that HNF4α could stimulate CYP3A4 reporter activity without binding to the CYP3A4 gene. To further demonstrate that HNF4α could interact with PXR without binding to DNA, mammalian one-hybrid assay was performed. Figure 3C demonstrated that HNF4α could stimulate Gal4 reporter (5XUAS-Luc) activity induced by binding of Gal4DBD-PXR fusion protein to the UAS of the Gal4 DNA-binding domain. All these results support the hypothesis that HNF4α interaction with PXR synergistically stimulates CYP3A4 gene transcription and that the putative HNF4α binding site in the CYP3A4 gene may not be important in promoter function.

SHP Inhibits CYP3A4 Reporter Activity. A previous study shows that SHP inhibits CYP3A4 reporter activity and blocks PXR binding to DNA in EMSA (Ourlin et al., 2003). We studied the effect of SHP on CYP3A4 reporter activity stimulated by PXR and HNF4α in HepG2 cells. Figure 4A shows that SHP inhibited PXR stimulation of CYP3A4 reporter activity when activated in the presence of rifampicin. Figure 4B shows that SHP also inhibited CYP3A4 reporter activity that was synergistically stimulated by PXR and HNF4α in the presence of rifampicin. However, our EMSA failed to confirm that SHP prevents PXR binding to the PXR-binding site in the CYP3A4 gene (data not shown). This discrepancy may be due to the amount of SHP used in the assay. We used in vitro translated SHP, which typically contains nanogram quantities of translated proteins, whereas Ourlin et al. (2003) used 9 μg of GST-SHP fusion protein in EMSAs.

Mammalian Two-Hybrid Assays of SHP Inhibition of PXR Interaction with Coactivators. We then performed cell-based mammalian two-hybrid assays in HepG2 cells to study the effect of SHP on PXR interaction with HNF4α, SRC-1, and PGC-1α. Gal4-fusion proteins should bind to the Gal4 binding site in the reporter plasmid 5XUAS-TK-Luc. If a Gal4-fusion protein interacts with a VP16 activation domain-fusion protein, the reporter activity should be stimulated. Addition of a competitor protein may disrupt the interaction and reduce the reporter activity. This assay is very sensitive in detecting protein-protein interaction without interference by endogenous transcription factors. Figure 5A shows that Gal4-PXR strongly interacts with VP16-HNF4α, and rifampicin was required for the interaction. Addition of SHP strongly inhibited PXR/HNF4α interaction. Similarly, Gal4-SHP interacted with VP16-PXR in the presence of rifampicin, and HNF4α inhibited the interaction (Fig. 5B). Therefore, HNF4α and SHP competed for interaction with PXR. We did two-hybrid assays of Gal4-PXR interaction with VP16-SRC-1 or VP16-PGC-1α to test whether SHP inhibited PXR interaction with SRC-1 and PGC-1α. Figure 5C shows that Gal4-PXR interacted with VP16-SRC-1, but SHP only partially inhibited the interaction. In contrast, Gal4-PGC-1α interacted with VP16-PXR, but SHP did not affect their interaction (Fig. 5D). These results indicated that HNF4α and SHP competed for binding to PXR. However, SHP was unable to block PXR/PGC-1α interaction and only partially blocked PXR/SRC-1 interaction.

GST Pull-Down Assays of PXR Interaction with HNF4α, SRC-1, PGC-1α, and SHP. The GST pull-down assay is a cell-free assay for studying specific protein-protein interactions. GST-HNF4α (full-length) fusion protein or GST-hPXR (LBD) fusion protein strongly interacted with PXR or HNF4α, respectively, only when rifampicin was added to activate PXR (Fig. 6A). PXR interacted with SRC-1 in the absence of rifampicin, and addition of rifampicin enhanced their interaction. PXR interaction with PGC-1α does not require rifampicin. The degrees of ligand dependence for these coactivators to interact with PXR are consistent with mammalian two-hybrid assays shown in Fig. 5. We next performed GST pull-down assays to study the effect of rifampicin on SHP interaction with PXR and HNF4α. In vitro translated and ³⁵S-labeled PXR or SHP was used to interact with GST-SHP or GST-HNF4α, respectively. As shown in Fig. 6B, rifampicin strongly enhanced PXR and SHP interaction. However, rifampicin did not affect SHP interaction with HNF4α as expected. We then added in vitro translated unlabeled-SHP or PXR to study their effect on interaction of ³⁵S-labeled PXR or SHP with HNF4α, respectively. Results shown in the lower panels of Fig. 6B demonstrated that the amount of PXR or SHP that bound to GST-HNF4α was decreased. These results provided additional evidence that SHP, PXR, and HNF4α interacted and competed for binding to each other.

ChIP Assays of PXR and Coactivator Recruitment to CYP3A4 Chromatin. We then performed ChIP assays using HepG2 cells to study the effect of rifampicin on PXR recruitment of HNF4α, SRC-1, and PGC-1α to the CYP3A4 chromatin that contains PXR binding sites. The ChIP assay is considered as an in vivo assay of transcription factor interaction and coactivator recruitment in the native promoter context. Because the levels of PGC-1α are low in HepG2 cells, PGC-1α was overexpressed in HepG2 cells for ChIP assays. An antibody against PXR, HNF4α, SRC-1, or PGC-1α was used to immunoprecipitate the cross-linked DNA-protein complexes in HepG2 cells treated with rifampicin or vehicle (DMSO). PCR primers were designed to amplify the DNA fragments containing either the distal XREM or the proximal ER6 from these complexes. Figure 7A shows PCR products generated by the primer pairs for dXREM (Fig. 7A, left panel), pER6 (Fig. 7A, middle panel), and a negative control of the intron 2 region (Fig. 7A, right panel). The nonimmune IgG was used as a negative control of immunoprecipitation for ChIP assays (bottom of each panel). The results indicated that PXR was associated with both chromatins, and rifampicin did not affect PXR binding. The levels of HNF4α associated with both dXREM and dER6 chromatins were very low and similar, despite the presence of an HNF4α binding site in dXREM but not in pER6. Rifampicin strongly induced the recruitment of HNF4α to both chromatins, most likely via interaction with PXR. Rifampicin also induced PXR or HNF4α recruitment of SRC-1 to both chromatins. This is consistent with the enhanced binding of HNF4α and SRC-1 to the PXR in the presence of rifampicin shown in Fig. 6A. It is surprising that rifampicin reduced the recruitment of PGC-1α to both chromatins. This is in contrast to the GST pull-down assay results (Fig. 6A) that rifampicin had no effect on PGC-1α interaction with PXR, and the two-hybrid assay results (Fig. 5D) that rifampicin enhanced PXR interaction with PGC-1α.

We then studied the effect of SHP on PXR recruitment of coactivators to the CYP3A4 chromatin. In experiments shown in Fig. 7B, rifampicin-treated HepG2 cells were transfected with PGC-1α and SHP or pcDNA3 empty plasmid as indicated. DNA-protein complexes were immunoprecipitated with an antibody against PXR, HNF4α, SRC-1, or PGC-1α for ChIP assay. Figure 7B shows that SHP did not affect PXR binding to either chromatin. However, SHP reduced PXR recruitment of HNF4α and SRC-1 to both chromatins, whereas SHP did not affect PXR recruitment of PGC-1α. These data are consistent with those from mammalian two-hybrid assays (Fig. 5) and GST pull-down assays (Fig. 6) showing that SHP blocks PXR interaction with HNF4α or SRC-1 but does not affect PXR and PGC-1α interaction.

PXR InhibitsSHPGene Transcription. Since rifampicin strongly reduced SHP mRNA expression, we hypothesized that PXR might inhibit SHP gene transcription. We and others reported previously that HNF4α binds to the human SHP promoter and stimulates SHP gene expression (Jung and Kullak-Ublick, 2003; Jahan and Chiang, 2005). Figure 8A shows that HNF4α and PGC-1α stimulated SHP reporter activity in HepG2 cells independent of rifampicin. However, PXR strongly inhibited the SHP reporter activity stimulated by HNF4α and PGC-1α only when rifampicin was added. As a positive control, PXR inhibited the CYP7A1 reporter activity stimulated by HNF4α and PGC-1α in the presence of rifampicin (Fig. 8B) (Li and Chiang, 2004). In complete contrast, rifampicin-activated PXR markedly stimulated CYP3A4 reporter activity when HNF4α and PGC-1α also were transfected (Fig. 8C). These data support our previous finding that PXR inhibits the HNF4α target gene CYP7A1 by blocking HNF4α recruitment of PGC-1α to CYP7A1 promoter (Li and Chiang, 2004). It appears that the same mechanism also regulates the human SHP promoter. In contrast, PXR stimulates the PXR target gene CYP3A4. Taken together, our data suggest that HNF4α enhances PXR induction of CYP3A4 by protein-protein interaction.

Discussion

The ability to be strongly induced by a large variety of structurally unrelated compounds allows CYP3A4 to efficiently respond to elevated toxic metabolites such as bile acids and drugs and protects the liver against cholestasis induced by these compounds. Extensive studies in recent years have provided convincing evidence that PXR is the key regulator of phase I drug-metabolizing enzymes, phase II bile acid and bilirubin conjugation enzymes, and phase III drug and metabolite transporters (Boyer, 2005). PXR is the most important nuclear receptor that induces CYP3A4 to hydroxylate bile acids and drugs. However, most studies on PXR regulation of CYP3A4 were performed in mouse models. In this study, we used the primary human hepatocyte and HepG2 models to study the molecular mechanism of human PXR regulation of CYP3A4. Our real-time PCR assays revealed a remarkable 150-fold induction of CYP3A4 mRNA expression by rifampicin in human primary hepatocytes. In contrast, rifampicin treatment reduced SHP and CYP7A1 mRNA expression. These data prompted us to study the mechanism of PXR regulation of CYP3A4 and SHP gene transcription. Our study revealed that interaction of PXR with HNF4α and its coactivators, PGC-1α and SRC-1, contributes to the strong induction of CYP3A4 by rifampicin and that concomitant inhibition of SHP gene expression by PXR minimizes the inhibitory effect of SHP and maximizes PXR induction of CYP3A4. Several studies suggest the involvement of HNF4α in regulation of the CYP3A4, CYP2C8, and CYP2C9 genes by PXR (Kamiya et al., 2003; Tirona et al., 2003; Chen et al., 2005; Ferguson et al., 2005). This study suggests that HNF4α may act as a coactivator of PXR to recruit PGC-1α and SRC-1 to the PXR target gene, CYP3A4. PXR interacts strongly with HNF4α, SRC-1, PGC-1α, and SHP. SHP may compete with HNF4α and SRC-1 for binding to PXR and may repress CYP3A4 gene transcription (Fig. 9). Therefore, to maximize the induction of CYP3A4, PXR must inhibit SHP gene transcription.

This study also revealed a general mechanism for PXR inhibition of HNF4α-activated gene transcription. Previous studies suggest that PXR inhibited CYP7A1 by blocking HNF4α recruitment of PGC-1α (Bhalla et al., 2004; Li and Chiang, 2004). This study revealed the same mechanism for PXR inhibition of the human SHP gene. It is significant that rifampicin-activated PXR inhibits both CYP7A1 and SHP expression, whereas bile acid-activated FXR induces SHP to inhibit CYP7A1. FXR has been shown to modestly regulate the CYP3A4 gene (Gnerre et al., 2004). It was reported recently that chenodeoxycholic acid and cholic acid induce SHP and repress PXR-mediated induction of CYP3A4 in human primary hepatocytes and in mice (Ourlin et al., 2003). Chenodeoxycholic acid and cholic acid are FXR ligands. Bile acid activation of FXR, PXR, and, perhaps, also the signaling pathways in their study complicated the interpretation of their results. It is unlikely that FXR activation of CYP3A4 is a significant mechanism for bile acid induction of CYP3A4 because CYP3A4 can be induced without FXR (Schuetz et al., 2001). It seems clear that PXR and FXR regulate bile acid metabolism by somewhat different mechanisms; the secondary bile acid, LCA, activates PXR to inhibit bile acid synthesis, increase bile acid excretion, and detoxify bile acids, whereas the primary bile acid, chenodeoxycholic acid, activates FXR to inhibit bile acid synthesis and stimulate biliary bile acid excretion.

The CYP3A4 expression is affected by factors such as genetic polymorphism, hormonal status, cholestatic liver diseases, and exposure to drugs and xenobiotics. The induction of CYP3A4 varies widely among individuals and may account for different drug-metabolizing activities and drug-drug interactions. PXR agonists such as rifampicin, dexamethasone, glucocorticoids, and phenobarbital have been used for treating chronic liver diseases for several decades without knowledge of the mechanism of drug action until the discovery of PXR in 1998. A recent study of gallstone patients taking rifampicin or ursodeoxycholic acid confirmed that rifampicin and ursodeoxycholic acid induced hepatobiliary transport and detoxification systems and provided molecular mechanisms for these anticholestatic drugs (Marschall et al., 2005). Our current study provides a molecular mechanism for rifampicin induction of CYP3A4 in detoxification of bile acids and drugs during cholestasis. Drugs targeted to PXR may have potential for developing therapeutic drugs for treating cholestatic liver diseases induced by drugs.