Dual role of orphan nuclear receptor pregnane X receptor in bilirubin detoxification in mice

Authors

  • Simrat P. S. Saini,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Ying Mu,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Haibiao Gong,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • David Toma,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Hirdesh Uppal,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    3. Department of Pharmacology, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Songrong Ren,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Song Li,

    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Samuel M. Poloyac,

    1. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Wen Xie

    Corresponding author
    1. Center for Pharmacogenetics, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA
    3. Department of Pharmacology, School of Medicine, University of Pittsburgh, Pittsburgh, PA
    • Center for Pharmacogenetics, Salk Hall 656, University of Pittsburgh, Pittsburgh, PA 15213
    Search for more papers by this author
    • fax: 412-648-1664


  • Conflict of interest: Nothing to report.

Abstract

The pregnane X receptor (PXR) and the constitutive androstane receptor (CAR) are implicated in xenobiotic and endobiotic detoxification, including the clearance of toxic bilirubin. Previous studies have suggested both overlapping and preferential regulation of target genes by these receptors, but the mechanism of cross-talk remains elusive. Here we reveal a dual role of PXR in bilirubin detoxification in that both the loss and activation of PXR led to protection from hyperbilirubinemia induced by bilirubin infusion or hemolysis. The increased bilirubin clearance in PXR-null mice was associated with selective upregulation of detoxifying enzymes and transporters, and the pattern of regulation is remarkably similar to that of transgenic mice expressing the activated CAR. Interestingly, the increased bilirubin clearance and associated gene regulation were absent in the CAR-null or double-knockout mice. In cell cultures, ligand-free PXR specifically suppressed the ability of CAR to induce the multidrug resistance associated protein 2 (MRP2), a bilirubin-detoxifying transporter. This suppression was, at least in part, the result of the disruption of ligand-independent recruitment of coactivator by CAR. In conclusion, PXR plays both positive and negative roles in regulating bilirubin homeostasis, and this provides a novel mechanism that may govern receptor cross-talk and the hierarchy of xenobiotic and endobiotic regulation. PXR is a potential therapeutic target for clinical treatment of jaundice. (HEPATOLOGY 2005;41:497–505.)

Accumulation of bilirubin, the primary heme protein catabolic byproduct,1 results in hyperbilirubinemia and jaundice and is commonly seen in the genetic Crigler-Najjar diseases and Dubin-Johnson syndrome (for a review, see Jansen et al.2). Sustained hyperbilirubinemia may lead to cellular toxicity, including potentially lethal neurotoxicity (for a review, see J. Roy-Chowdhury3). Hyperbilirubinemia also can be associated with an adverse drug reaction, such as that reported for indinavir, a human immunodeficiency virus protease inhibitor.4

Detoxification of bilirubin is a multistep process. The uptake of unconjugated bilirubin into the hepatocytes is facilitated by the basolateral hepatocyte plasma membrane organic anion transporter peptides (OATPs) 4 and 2 (OATP4/SLC21A6 and OATP2/SLC21A5).5 The uptake of bilirubin by OATPs is enhanced further by the binding of bilirubin to glutathione-S-transferase (GST) A1 and A2. On entering hepatocytes, bilirubin is glucuronidated by the UDP-glucuronosyltransferase (UGT) 1A1 (UGT1A1).2 The final step of clearance is the secretion of conjugated bilirubin across the canalicular membrane into the bile mediated by multidrug resistance associated protein 2 (MRP2), an adenosine triphosphate-dependent conjugate export pump.6 Defects in the expression and/or function of these enzymes and transporters may lead to accumulation of bilirubin. For example, mutations in UGT1A1 genes have been attributed to the hereditary metabolic disorders associated with Crigler-Najjar syndrome type I,7 which are characterized by an inability to glucuronidate bilirubin.3 Gilbert syndrome, a less severe form of the disease, results from a reduced expression of UGT1A1 that is caused by a sequence abnormality in the promoter region of this gene.8 Loss of MRP2 function results in Dubin-Johnson syndrome, which is manifested by the accumulation of conjugated bilirubin.6

Recent reports have suggested that the expression of bilirubin-detoxifying enzymes and transporters are under the transcriptional control of orphan nuclear receptors pregnane X receptor (PXR) and constitutive androstane receptor (CAR). Several studies have shown that UGT1A1 is a target gene for PXR.9, 10 Transgenic mice expressing a constitutively activated hPXR (VP [herpes simplex virus virion protein VP16]-hPXR) in the liver show a marked increase in UGT1A1 activity and bilirubin clearance.9, 11 However, it is not known whether the activation of bilirubin-detoxifying genes other than UGT1A1 also contributes to the increase of bilirubin clearance in VP-hPXR mice. Having demonstrated the positive role of PXR activation in bilirubin clearance, the effect of loss of PXR on bilirubin homeostasis has not been evaluated. CAR also has been shown to induce UGT1A1.9, 10, 12–14 In animals, treatment with CAR agonists, such as phenobarbital and 1,4-bis[2-(3,5 dichloropyridyloxy)] benzene (TCPOBOP), increased the hepatic expression of UGT1A1, GSTA1 and GSTA2, MRP2, and OATP4/SLC21A6, and this induction was absent in CAR-null mice.13 Because many CAR agonists also activate PXR, it remains unclear whether activation of CAR alone is sufficient to promote bilirubin clearance. PXR and CAR are known to cross-regulate each others' target genes.15 The induction of UGT1A1 may be an overlapping mechanism of CAR- and PXR-activated bilirubin clearance, but it remains to be determined whether there are differential effects of CAR and PXR in regulating bilirubin clearance.

In this article, we show an unexpected increase of bilirubin clearance in the PXR-null mice, a phenotype also seen in transgenic mice that express the activated PXR or CAR. Our in vivo and in vitro results suggest that the ligand-free PXR functions as a CAR suppressor and that a loss of PXR caused derepression and resultant upregulation of bilirubin-detoxifying enzymes and transporters. We conclude that PXR plays both positive and negative roles in regulating bilirubin homeostasis.

Abbreviations:

OATP, organic anion transporter peptide; GST, glutathione S transferase; UGT, UDP-glucuronosyltransferase; MRP2, multidrug resistance associated protein 2; PXR, pregnane X receptor; CAR, constitutive androstane receptor; VP, herpes simplex virus virion protein VP16; TCPOBOP, 1,4-bis[2-(3,5 dichloropyridyloxy)] benzene; Dox, doxycycline; PCN, pregnenolone-16α-carbonitrile; ER-8, everted repeat spaced by eight nucleotides; tTA, tetracycline responsive transcriptional activator; CYP, cytochrome P450; FXR, farnesoid X receptor; SRC-1, steroid receptor coactivator 1.

Materials and Methods

Animals.

The creation of PXR-null,16 CAR-null,17 PXR and CAR double-knockout,18 TRE-VP-CAR/Lap-tTA transgenic,18 and Alb-VP-hPXR transgenic16 mice were previously described. The PXR-null and CAR-null mice have a mixed background of C57B and SVJ129. The VP16 transgenic mice have a mixed background of C57B and CB6F1. When necessary, the TRE-VP-CAR/Lap-tTA mice were subjected to drinking water laced with doxycycline (Dox; 2 mg/mL; Sigma, St. Louis, MO) for 7 days.

Northern Blot Analysis and Reverse-Transcriptase–Polymerase Chain Reaction.

Total RNA was prepared from tissues using the TRIZOL Reagent (Invitrogen, Carlsbad, CA). Northern hybridization was carried out as described.16 The cDNA probes for UGT1A1, GSTA2, and MRP2 were cloned by reverse-transcriptase–polymerase chain reaction using the published oligonucleotide sequences as follows. UGT1A1, 5′ TGGAAGCCACTGGCTGAGTAT 3′ and 5′ CTTCCAGAGAGGCCATAAACTC 3′; GSTA2, 5′ AGAAGGAGTGGCGGATCTGG 3′ and 5′ TATCCGAGGAAGTGATCATG 3′; MRP2, 5′ CGTGGGTGACCGACAAGAAGC 3′ and 5′ CTAGAGCTCCGTGTGGTTCAC 3′. The probe for OATP2 was described previously.19 The membranes were stripped and reprobed with GAPDH for loading control. The MRP2 reverse-transcriptase–polymerase chain reaction was performed as previously described.20

Measurement of Bilirubin Clearance and Liver Enzyme Activity.

Bilirubin and phenylhydrazine were purchased from Sigma. For bilirubin treatment, adult males were given a single tail vein injection of bilirubin (10 mg/kg body weight). Blood samples were collected 1 hour after injection. When necessary, mice were subjected to 7 days of Dox treatment before the bilirubin injection. For phenylhydrazine-induced hyperbilirubinemia, adult males were given seven daily gavages of N-acetylphenylhydrazine (0.2 mg/mouse in phosphate-buffered saline).21 Blood samples were collected 24 hours after the last treatment. Serum bilirubin and alanine aminotransferase activity was measured by ANTECH Diagnostic (Lake Success, NY).

Transfection and Reporter Gene Assay.

The tk-MRP2-Luc reporter was generated as described.22 The tk-PBRE-Luc reporter15 and Gal-hSRC-1 RID23 were described previously. In addition to the designated nuclear receptor response elements, all of the tk reporter genes contain the minimal(nt, −105 to +51) promoter of the herpes simplex virus thymidine kinase (HSV-tk) gene.24 CV-1 and 293 cell transfections were performed in 48-well plates with Lipofectamine 2000 (Invitrogen).18 Transfected cells were treated with ligands or vehicle for 24 to 40 hours before luciferase assay. The luciferase activities were normalized against the activities of cotransfected β-gal. Concentrations were: TCPOBOP, 250 nM; androstenol, 5 μM; pregnenolone-16α-carbonitrile and rifampicin, 10 μM.

DNA-Binding Analysis.

Electrophoretic mobility shift assays were performed using in vitro–transcribed and –translated proteins (Promega, Madison, WI), as described previously.15 Oligonucleotide sequences for MRP2/everted repeat spaced by eight nucleotides (ER-8) were as described previously.22

GST Pull-down Assay.

GST alone or GST-hPXR LBD (aa 108-434)25 fusion proteins on glutathione agarose beads were mixed with in vitro–translated 35S-labeled hSRC-1 protein, and the pull-down assay was performed as previously described.26

Results

Unexpected Increase of Bilirubin Clearance in PXR-null Mice.

Activation of PXR or CAR in mice has been shown to promote bilirubin clearance.9, 11–14 In this study, we examined whether PXR and CAR are required to maintain the homeostasis of bilirubin using single- or double-knockout mice.16–18 The loss of PXR and/or CAR did not significantly alter the basal level of serum bilirubin (Fig. 1). However, in contrast to the expected bilirubin sensitivity in wild-type mice, the PXR-null mice exhibited a surprisingly complete resistance to hyperbilirubinemia induced by bilirubin infusion (Fig. 1). Interestingly, the CAR-null and the double-knockout mice showed similar sensitivity as the wild-type mice (Fig. 1). The PXR-null–specific resistance to hyperbilirubinemia also was observed when hyperbilirubinemia was induced by treating the mice with phenylhydrazine, a prohemolysis agent (Fig. 1).27 We concluded that the loss of PXR alone, but not in combination with loss of CAR, was associated with an increased bilirubin clearance.

Figure 1.

Increased bilirubin clearance in pregnane X receptor (PXR)-null mice. Mouse serum was collected 1 hour after a single tail vein injection of bilirubin or 7 days of phenylhydrazine (PHZ) treatment and was measured for total bilirubin. Results represent the averages and SE. WT, wild type; KO, knockout.

PXR-null Mice Have Increased Expression of Selective Bilirubin-Detoxifying Enzymes and Transporters.

Detoxification of bilirubin involves several phase II enzymes and membrane transporters.2, 5, 6 The increased bilirubin clearance observed in PXR-null mice suggested an upregulation of bilirubin-detoxifying genes in these animals. Consistent with their resistance to hyperbilirubinemia, PXR-null mice exhibited an increase in messenger RNA expression of UGT1A1, OATP4/SLC21A6, GSTA2, and MRP2 (Fig. 2A). The expression of these genes remained largely unchanged in CAR-null and double-knockout mice (Fig. 2A), consistent with the sensitivity to hyperbilirubinemia found in both genotypes. The upregulation in PXR-null mice was gene specific, because the expression of OATP2 and cytochrome P450 (CYP) 3A11 (CYP3A11) did not increase and CYP2B10 was rather suppressed in this genotype (Fig. 2B). The intact basal expression of CYP3A11 was consistent with our previous report,16 which can be explained by the continued expression and regulatory effect of CAR and vitamin D receptor, two other positive CYP3A11 regulators.15, 28 The mechanism of CYP2B10 downregulation in the PXR-null mice remains to be elucidated. It is possible that PXRs have a positive role in maintaining the basal expression of CYP2B10, which cannot be compensated fully in the absence of this receptor. The genotype-specific upregulation of MRP2 (Fig. 2C) and MRP3 (data not shown, and Uppal et al.29) in PXR-null mice also was confirmed by reverse-transcriptase–polymerase chain reaction. MRP3 has been shown to export bilirubin conjugates into blood across the sinusoidal membrane (for a review, see Keppler and Konig30). The regulation of this transporter suggests that altered sinusoidal export also may play a role in nuclear receptor-mediated bilirubin metabolism.

Figure 2.

Pregnane X receptor (PXR)-null mice have increased expression of selective bilirubin-detoxifying enzymes and transporters. Liver total RNA was subjected to Northern blot analysis for the messenger RNA detection of (A) UDP-glucuronosyltransferase (UGT) 1A1 (UGT1A1), organic anion transporter peptide (OATP) 4 (OATP4/SLC21A6), glutathione S transferase A2 (GSTA2), and multidrug resistance associated protein 2 (MRP2); and (B) OATP2, cytochrome P450 3A11 (CYP3A11), CYP2B10, PXR, and constitutive androstane receptor (CAR). (C) Reverse-transcriptase–polymerase chain reaction analysis of MRP2 messenger RNA expression on liver total RNA. WT, wild type; KO, knockout; SLC, solute carrier.

Activation of CAR and PXR in Transgenic Mice Induces Bilirubin Clearance.

The gain-of-function of PXR and CAR in bilirubin clearance was examined in transgenic mice. We recently created transgenic mice in which the activated VP-CAR was expressed conditionally in the liver.18 In mice that carry both TRE-VP-CAR and the liver-specific Lap-tTA transgenes, the tetracycline transcriptional activator (tTA) was expected to bind to the tetracycline responsive element and induced the expression of VP-CAR in the absence of Dox. Addition of Dox displaced tTA and thus silenced VP-CAR expression (Fig. 3A). The expression and regulation of VP-CAR was confirmed by Northern blot analysis (Fig. 3B). We used the VP-CAR mice to evaluate the effect of CAR activation on bilirubin clearance. The sera from both untreated VP-CAR and wild-type mice showed no visible differences (Fig. 3C). However, on bilirubin injection, sera from wild-type mice, but not VP-CAR mice, appeared darker than those of the untreated mice, a sign of bilirubin accumulation (Fig. 3C). Serum evaluation revealed that the levels of total bilirubin in the transgenic mice were less than half of those in wild-type mice (Fig. 3D). Similar results were observed when hyperbilirubinemia was induced by phenylhydrazine (Fig. 3D). The protection was VP-CAR dependent, because no protection was seen in mice treated with Dox for 7 days before bilirubin injection (Fig. 3E), presumably because of the silencing of VP-CAR expression (Fig. 3B). Dox did not activate mouse CAR (mCAR) in transient transfection and report gene assay (data not shown). Physiologically, the VP-CAR mice were protected from bilirubin-induced hepatotoxicity. Treatment of wild-type mice with bilirubin increased activity of serum alanine aminotransferase, a liver enzyme associated with hepatotoxicity (Fig. 3F). In contrast, VP-CAR mice had normal alanine aminotransferase activity (10-35 U/L) on bilirubin treatment, but the protection was absent in transgenic mice pretreated with Dox (Fig. 3F). The effect of PXR activation on bilirubin clearance was evaluated using the liver-specific VP-PXR transgenic mice.16 Bilirubin injections induced an accumulation in the wild-type but not the VP-PXR mice, consistent with our previous observation (Fig. 3G, and Xie et al.9). Thus, genetic activation of either PXR or CAR was sufficient to promote bilirubin clearance.

Figure 3.

Increased bilirubin clearance in VP-CAR and VP-PXR transgenic mice. (A) Outline of the TRE-VP-CAR/Lap-tTA “Tet-Off' transgenic system. (B) Conditional expression of VP-CAR as revealed by Northern blot analysis. (C) Photograph of sera from mice untreated or treated with bilirubin for 1 hour. (D) Sera from mice untreated or treated with bilirubin or phenylhydrazine (PHZ) were measured for total bilirubin. (E) Mice were treated with doxycycline (Dox) for 7 days before bilirubin injection followed by measurement of total and direct bilirubin. (F) Serum alanine aminotransferase activity was measured 1 hour after bilirubin treatment. (G) Serum from wild-type or VP-PXR mice, untreated or treated with bilirubin, was measured for total bilirubin. Results in (D-G) represent the averages and standard error (n = 3-5). TRE, tetracycline response element; Lap-tTa, liver-enriched activator protein–tetracycline responsive transcriptional activator; Dox, doxycycline; VP-CAR, VP16 and CAR fusion protein; BR, bilirubin; PHZ, phenylhydrazine; ALT, alanine aminotransferase; VP-PXR, VP16 and PXR fusion protein.

Activation of CAR and PXR Induces Bilirubin Clearance via Overlapping, Yet Distinct, Mechanisms.

To delineate the mechanism of CAR- and PXR-mediated bilirubin clearance, we profiled the expression of genes encoding bilirubin-detoxifying enzymes and transporters in VP-CAR and VP-PXR transgenic mice. As shown in Fig. 4A, the expression of VP-CAR induced messenger RNA expression of UGT1A1, but not UGT1A9, consistent with previous reports.9, 12, 13 The expression of OATP4, GSTA2 (Fig. 4B), OATP2, and MRP2 (Fig. 4C) also was increased in VP-CAR mice. The upregulation was VP-CAR dependent, because a complete loss of induction of MRP2 and OATP2 was seen after Dox treatment (Fig. 4C). A similar gene profiling performed on the VP-PXR mice revealed that although the UGT1A1, OATP2 (Fig. 4D), and MRP2 (Fig. 4E) messenger RNA was induced, the expression of GSTA2 and OATP4 remained largely unchanged in this transgenic line (Fig. 4D). Our results suggest that CAR and PXR induce bilirubin clearance via overlapping, yet distinct, mechanisms.

Figure 4.

CAR and PXR induce bilirubin clearance via overlapping, yet distinct, mechanisms. Liver total RNA was subjected to Northern blot analysis. Expression of VP-CAR induced the expression of UGT1A1 but not (A) 1A9, (B) OATP4 and GSTA2, and (C) OATP2 and MRP2. Dox treatment diminished the induction of the transporters in panel C. Expression of VP-hPXR induced the messenger RNA expression of (D) UGT1A1 and OATP2, and (E) MRP2, but not (D) GSTA2 and OATP4. VP-CAR, VP16 and CAR fusion protein; UGT, UDP-glucuronosyltransferase; OATP, organic anion transporter peptide; GST, glutathione S transferase; MRP2, multidrug resistance associated protein 2.

Ligand-Free PXR Represses CAR-Mediated Activation of Bilirubin-Detoxifying Genes in Cell Cultures.

Because both the loss and activation of PXR led to resistance to hyperbilirubinemia, we hypothesize that the ligand-free PXR may function as a suppressor to repress the activation of bilirubin-detoxifying genes, such as that mediated by CAR. We also predict this repression will be relieved by the absence of PXR or the activation of PXR. To test these hypotheses, CV-1 cells were cotransfected with the tk-MRP2-Luc reporter gene and expression vectors for mCAR and/or mouse PXR (mPXR). The MRP2 reporter contained three copies of the ER-8 response element shown to be CAR, PXR, and farnesoid X receptor (FXR) responsive.22 As expected, expression of mCAR constitutively activated the MRP2 reporter, and this activation was potentiated by the agonist TCPOBOP and inhibited by the antagonist androstenol (Fig. 5A, left panel). Transfection of mPXR alone suppressed the basal expression of the reporter. The mPXR-induced MRP2 suppression was relieved by the mPXR-specific agonist pregnenolone-16α-carbonitrile but not by the hPXR-specific rifampicin. Cotransfection of mCAR and mPXR, in the absence of ligand, inhibited the constitutive activity of mCAR by more than 50%, an inhibition similar to that caused by androstenol. Again, the suppression was relieved by pregnenolone-16α-carbonitrile, but not by rifampicin. The TCPOBOP-induced reporter activity also was lower in cells transfected with both mCAR and mPXR than in cells transfected with mCAR alone (Fig. 5A, left panel). TCPOBOP is a CAR agonist, but not a PXR agonist,15, 18 suggesting that ligand-free PXR can suppress both the constitutive and ligand-inducible activity of CAR. Interestingly, the suppression was target gene specific. When a CAR-responsive CYP2B reporter gene tk-PBRE-Luc15 was used, the ligand-free PXR neither suppressed the basal expression of the reporter nor inhibited the activity of CAR (Fig. 5A, right panel). This lack of suppression was consistent with the absence of CYP2B induction in PXR-null mice (Fig. 2B). The repression of CAR-mediated activation of MRP2, but not CYP2B, by ligand-free PXR also was seen in the 293 cells (Fig. 5B). The repression was PXR specific, because FXR had little effect on the cotransfected CAR (data not shown).

Figure 5.

Ligand-free pregnane X receptor (PXR) functions as a constitutive androstane receptor (CAR) repressor in cell cultures. The tk-MRP2-Luc or tk-PBRE-Luc reporters were transiently transfected into (A) CV-1 or (B) 293 cells in the presence of CAR and/or mPXR. (C) Both PXR and CAR bind to the multidrug resistance associated protein 2 (MRP2)/everted repeat spaced by eight nucleotides (ER-8) response element as revealed by electrophoretic mobility shift assays. (D) The Gal-responsive tk-UAS-Luc reporter, Gal-SRC-RID, and VP16 and CAR fusion protein (VP-CAR) were transfected into CV-1 cells in the presence of PXR or farnesoid X receptor (FXR). (E) The tk-UAS-Luc and Gal-SRC-RID were transfected into CV-1 cells in the presence of VP16 and PXR fusion protein (VP-PXR) or VP16 and FXR fusion protein (VP-FXR). (F) Ligand-independent and dependent interaction between hPXR ligand-binding domain and hSRC-1 (left panel). Shown in the right panel is the Coomassie blue staining of the glutathione S transferase (GST) and PXR fusion proteins. Transfected cells were treated with ligands or solvent for 24 hours before luciferase assay. Results shown are fold induction over the reporter alone (A, B, D, and E). The results represent the averages and SE from triplicate assays. Ligand concentrations are 10 μM for rifampicin (RIF) and pregnenolone-16α-carbonitrile (PCN), 5 μM for androstenol, 250 nM for 1,4-bis[2-(3,5 dichloropyridyloxy)] benzene (TCPOBOP), and 100 μM for 3H–chendeoxycholic acid (CDCA) and lithocholic acid (LCA). RXR, retinoid X receptor; DMSO, dimethyl sulfoxide; Gal-SRC-RIO, gal-steroid receptor coactivator-1 fusion protein.

Compared with CAR, PXR exhibited a modestly favorable binding toward ER-8 in an electrophoretic mobility shift assay, which is more evident when an equal amount of transcribed and translated proteins in vitro–translated receptor proteins were presented (Fig. 5C). The relative DNA binding affinity may help to determine regulatory hierarchies. However, the DNA binding alone could not fully explain the repression, because a chimeric receptor composed of Gal–DNA-binding domain and PXR–ligand-binding domain was equally effective in repressing CAR activity on the tk-MRP2 reporter (data not shown). The expression of CAR at messenger RNA level was not different between wild-type and PXR-null mice (Fig. 2B). Electrophoretic mobility shift assays using liver nuclear extracts and an MRP2/ER-8 probe showed no appreciable difference in total binding between wild-type and PXR-null mice (data not shown). It is conceivable that MRP2/ER-8–responsive nuclear receptors other than CAR, such as PXR and FXR, also contribute to the total binding. A commercially available CAR antibody failed to supershift even with the transcribed and translated protein-synthesized CAR/retinoid X receptor (RXR) proteins (data not shown). Therefore, we cannot excluded the possibility that binding activity of CAR was increased in the PXR-null mice.

It has been shown that the constitutive activity of CAR was the result of the ligand-independent recruitment of p160 coactivators, such as the steroid receptor coactivator 1 (SRC-1) and the coactivator recruitment was disrupted by the antagonistic androstenol.23, 31 A coactivator recruitment assay revealed that ligand-free PXR, but not ligand-free FXR, was able to disrupt the ligand-independent interaction between CAR and SRC-1 (Fig. 5D), providing a plausible mechanism for PXR-mediated CAR repression. While investigating this interaction, we unexpectedly found that the ligand-free mPXR exhibited ligand-independent interaction with SRC-1 in a mammalian two-hybrid assay, activating the reporter gene by nearly threefold (Fig. 5E), comparable with that induced by an SRC-1–CAR interaction in the absence of ligand (3.5-fold; Fig. 5D). This interaction was enhanced by pregnenolone-16α-carbonitrile but not by rifampicin (Fig. 5E). The ligand-independent recruitment of SRC-1 was PXR specific, because FXR showed only ligand chenodeoxycholic acid (CDCA)-dependent association with this coactivator (Fig. 5E). CDCA and lithocholic acid, two weak PXR agonists,32 failed to enhance the SRC-1–PXR interaction (Fig. 5E). The ligand-independent SRC-1–PXR interaction also may have obscured the ligand effect of CDCA and lithocholic acid in this assay. The ligand-independent interaction between hPXR and SRC-1 and the ligand-induced interaction enhancement were confirmed by a GST pull-down assay (Fig. 5F).

Discussion

In this report, we show that mice deficient in PXR alone had increased bilirubin clearance that was associated with an increased expression of bilirubin-detoxifying enzymes and transporters. The increased bilirubin clearance in PXR-null mice not only was unexpected, but also had several intriguing features. First, although the expression of either activated PXR or activated CAR in the liver of transgenic mice was sufficient to increase bilirubin clearance, loss of PXR alone, but not CAR alone, was associated with increased bilirubin clearance. The other interesting observation is that loss of PXR alone, but not in combination with CAR loss, impacted bilirubin clearance. We propose that PXR has both negative and positive roles in bilirubin clearance and that this dual role of PXR is achieved, at least in part, by cross-talking with CAR, another positive regulator of bilirubin clearance.

CAR was isolated as an orphan receptor exhibiting substantial activity in the absence of ligand.33, 34 This constitutive activity was supported by the observation that CAR activity was sustained in yeast, an organism that lacks endogenous steroid ligands.23 The constitutive activity was found to be the result of ligand-independent association with coactivators.23, 31 CAR activity was later shown to be potentiated by phenobarbital and TCPOBOP15, 35, 36 and inhibited by androstane metabolites.23 Interestingly, mutations designed to block ligand binding pocket blocked the effects of both activating and inhibitory ligands but had little effect on the constitutive activity of CAR, suggesting that CAR truly has ligand-independent activity.36 Although TCPOBOP has been shown to bind to mCAR directly,37 several CAR agonists, such as phenobarbital38 and bilirubin,13 do not bind to the receptor but translocate CAR to the nucleus. Nuclear translocation alone, however, does not determine CAR activation, but rather additional regulatory events in the nucleus are required (for a review, see Swales and Negishi39).

Activation of CAR has been shown to promote bilirubin clearance. In animals, treatment with CAR agonists increased bilirubin clearance and hepatic expression of UGT1A1, GSTA1 and GSTA2, MRP2, and SLC21A6/OATP4, and this induction was absent in CAR-null mice.13 Yin Zhi Huang, a traditional Chinese remedy, has been shown to enhance bilirubin clearance by activating CAR.14 In the same study, however, treatment with pregnenolone-16α-carbonitrile, 2,3,7,8-tetracholorodibenzo-p-dioxin and WY-14,643, respective ligands for PXR, aryl hydrocarbon receptor, and peroxisome proliferator-activated receptor α, failed to induce bilirubin clearance,14 despite documented evidence that these receptors control the expression of bilirubin-detoxifying genes.9, 40, 41

Although PXR and CAR have been shown to cross-regulate each other's target genes, a remaining challenge is to determine the molecular mechanism that governs the hierarchy of xenobiotic and endobiotic detoxification in vivo. An equally intriguing question regards why CAR is equipped with a substantially higher constitutive activity than other xenobiotic receptors, despite the notion that xenobiotic induction is launched only as needed. Our results suggest that when both PXR and CAR are present, the ligand-free PXR suppresses the constitutive activity of CAR, maintaining a basal capacity of detoxification. The increased bilirubin clearance in PXR-null mice was likely the result of the derepression of CAR as a consequence of the loss of PXR. This notion is supported by the observation that the pattern of enzyme and transporter regulation in the PXR-null mice was remarkably similar to that of transgenic mice expressing the activated CAR. Meanwhile, PXR also plays a positive role in bilirubin homeostasis. The induced bilirubin clearance by PXR activation can be explained by the derepression of CAR and/or the direct activation of detoxifying genes that also happen to be PXR targets. Results from CAR-null mice suggest that the basal expression of bilirubin-detoxifying genes can be maintained by alternative receptors and/or pathways that are not targets of PXR-mediated repression. These alternative pathways also may explain why the double-knockout mice failed to exhibit increased sensitivity to hyperbilirubinemia.

The current study has revealed several unique features of PXR-mediated repression. PXR repression is gene specific and cannot be explained fully by competitive DNA binding, as the expression of CYP3A11 and CYP2B10 was not induced in PXR-null mice. The intact expression of CYP3A11 in PXR-null mice was consistent with our previous report,16 but different from an upregulation reported for another independently created PXR-null line.42 The target gene specificity of PXR-mediated repression also may explain why PXR–CAR cross-talks are a potentially physiological pathway-dependent phenomenon. We recently showed that the combined, but not an individual loss, of PXR and CAR caused markedly heightened sensitivity to lithocholic acid hepatotoxicity.29 The double-knockout–specific bile acid sensitivity is in contrast to the PXR-null–specific increase in bilirubin clearance that we report in this study. The ligand-independent recruitment of coactivators by PXR may account for, at least in part, the inhibition of the constitutive activity of CAR. However, we could not exclude the involvement of corepressors in this regulatory mechanism. PXR has been shown to interact with the corepressor silencing mediator of retinoic acid and thyroid hormone receptor (unpublished observations, 2004).43, 44 It also remains to be determined whether the ligand-free PXR inhibit the ligand-dependent CAR activation by blocking the nuclear translocation or the intranuclear regulatory pathway(s). Moreover, despite the well-documented constitutive activity of CAR, mice that are deficient in CAR fail to exhibit a systemic downregulation of CAR target genes. This lack of downregulation may be the consequence of adaptive changes as wells as receptor cross-talks. Nevertheless, this observation does argue whether CAR is truly a dominating positive regulator in vivo.

In conclusion, although paradoxical, the dual role of PXR establishes this receptor as an essential cellular factor in determining the hierarchy of function, as well as the homeostasis of xenobiotic and endobiotic detoxification. Moreover, the current study has further established PXR and CAR as therapeutic targets for clinical treatment of genetic or acquired forms of jaundice.

Acknowledgements

We thank Dr. Yanan Tian for preparation of the GST-PXR protein; Drs. Barry Forman, Hung-Ying Kao, Jun Sonoda, and Ron Evans for plasmids; Drs. Hongwu Chen, Chih-Cheng Tsai, George Michalopoulos, Yong Tae Kwon, Jeff Kudlow, and Tom Jones for discussion and comments on the manuscript. The PXR knockout line was created by Wen Xie in the laboratory of Ron Evans at the Salk Institute. The senior author would like to dedicate this paper to the memory of Dr. Li Xu, who participated in the initial characterization of VP-CAR transgenic mice.

Ancillary