Differential regulation of constitutive androstane receptor expression by hepatocyte nuclear factor4α isoforms


  • Potential conflict of interest: Nothing to report.


Constitutive androstane receptor (CAR; NR1I3) controls the metabolism and elimination of endogenous and exogenous toxic compounds by up-regulating a battery of genes. In this work, we analyzed the expression of human CAR (hCAR) in normal liver during development and in hepatocellular carcinoma (HCC) and investigated the effect of hepatocyte nuclear factor 4α isoforms (HNF4α1 and HNF4α7) on the hCAR gene promoter. By performing functional analysis of hCAR 5′-deletions including mutants, chromatin immunoprecipitation in human hepatocytes, electromobility shift and cotransfection assays, we identified a functional and species-conserved HNF4α response element (DR1: ccAGGCCTtTGCCCTga) at nucleotide −144. Both HNF4α isoforms bind to this element with similar affinity. However, HNF4α1 strongly enhanced hCAR promoter activity whereas HNF4α7 was a poor activator and acted as a repressor of HNF4α1-mediated transactivation of the hCAR promoter. PGC1α stimulated both HNF4α1-mediated and HNF4α7-mediated hCAR transactivation to the same extent, whereas SRC1 exhibited a marked specificity for HNF4α1. Transduction of human hepatocytes by HNF4α7-expressing lentivirus confirmed this finding. In addition, we observed a positive correlation between CAR and HNF4α1 mRNA levels in human liver samples during development, and an inverse correlation between CAR and HNF4α7 mRNA levels in HCC. These observations suggest that HNF4α1 positively regulates hCAR expression in normal developing and adult livers, whereas HNF4α7 represses hCAR gene expression in HCC. (HEPATOLOGY 2007;45:1146–1153.)

Hepatocyte nuclear factor (HNF4; NR2A1) controls the expression of genes involved in the transport and metabolism of nutrients including cholesterol, fatty acids, and glucose,1–4 hepatocyte polarization,5 and liver development and differentiation.6, 7 Mammals express 2 genes: HNF4α and HNF4γ.8 HNF4α gene encodes nine isoforms through alternative promoter usage and alternative splicing; their expression varies with development, differentiation, and tissue origin.9–14 Promoters P1 and P2 initiate transcripts containing exon 1A (isoforms α1-α6) and 1D (isoforms α7-α9), respectively. In murine liver, HNF4α7 is expressed mainly in the embryo and is almost absent in the adult, whereas HNF4α1 is almost exclusively expressed in the adult.13 Increase of HNF4α1 expression represses P2 activity and down-regulates HNF4α7 expression in the adult.15 Only HNF4α isoforms initiated at promoter P1 contain the activation domain AF-1 which mediates the recruitment of coactivators.16–18

The constitutive androstane receptor (CAR; NR1I3) controls the inducible expression of genes envolved in xenobiotics/endobiotics disposition, including cytochrome P450 (CYP2B/2C/3A),19, 20 UDP-glucuronosyltransferases,21 transporters,22 and others.23 CAR is activated in response to phenobarbital and many other compounds, notably bilirubin.24–26 CAR forms heterodimers with retinoid X receptor α (RXRα) which bind the promoter of target genes. CAR activity has been shown to protect against harmful effects of hyperbilirubinemia and toxic bile acids.26 Although HNF4 positively regulates genes involved in lipid homeostasis and hepatic gluconeogenesis,7 CAR negatively regulates some of these.27 Notably, CAR inhibits HNF4 binding to the DR1 motif in CYP7A1, CYP8B1 and PEPCK promoters, and HNF4 and CAR compete for common coactivators GRIP-1 and PGC-α.28

Recently, we demonstrated that CAR expression is controlled by glucocorticoids29 through a functional glucocorticoid response element (GRE),30 and that proinflammatory interleukin IL-1β or lipopolysaccharides decrease its expression in human hepatocytes.31 Moreover, liver conditional knockout of HNF4α significantly decreases CAR expression.32 However, regulatory mechanisms of CAR expression remain largely unknown.

The aim of this work was to investigate the regulation of human CAR (hCAR) in liver and in hepatocellular carcinoma (HCC). For this purpose CAR, HNF4 and other gene expression was evaluated by quantitative RT-PCR. In parallel, we analyzed the transcriptional activity of hCAR promoter by deletion analysis, mutagenesis, cotransfection, electrophoretic mobility shift assay, chromatin immunoprecipitation, and transduction of human hepatocytes by HNF4-expressing lentivirus.


AFP, α-fetoprotein; CAR, constitutive androstane receptor, NR1I3; ChIP, chromatin immunoprecipitation; CYP, cytochrome P450; DR1, direct repeat 1; EMSA, electromobility shift assay; GR, glucocorticoid receptor; HNF4α, hepatocyte nuclear factor 4 alpha; GRIP, glutamate receptor binding protein; PEPCK, phosphoenolpyruvate carboxykinase; PGC, peroxisome proliferator-activated receptor gamma coactivator; TAT, tyrosine aminotransferase.

Patients and Methods

Tissue Source.

Liver samples were obtained from patients who underwent resections for medically required purposes unrelated to our research program. The use of these specimens has been approved by the French National Ethics Committee. The use of fetal and pediatric liver samples and HCC samples has been approved by the Montpellier and Bordeaux University Hospital Centers, respectively.

Quantitative PCR.

Total RNA was extracted from cells or frozen tissues using TRIZOL reagent (Gibco BRL, Cergy-Pontoise, France). Purity was confirmed by spectrophotometry. The cDNAs were synthesized from 1 μg of total RNA using Superscript II (Invitrogen, Cergy-Pontoise, France) at 42°C for 60 minutes using random hexamers (Amersham-Pharmacia Biotech, Cambridge, UK). Two microliters of diluted reverse transcription reaction (1/10 or 1/50000 for 18S RNA) was used for real-time PCR amplification (Light Cycler, Roche Diagnostics Corp., Meylan, France): denaturation step 95°C for 8 minutes, and 50 cycles of denaturation at 95°C for 15 seconds; annealing at 67°C for 10 seconds, extension at 72°C for 20 seconds. Quality of PCR-products was assessed by monitoring a fusion step at the end of each run. Sense and reverse primers:











For each RNA, 10-fold serial dilutions of RT-PCR products were used to establish a standard curve. Relative mRNA levels were normalized with respect to 18S RNA level and values were determined by interpolation using standard curves.


The hCAR promoter constructs have been described.30 Plasmids pCMV-HNF4α1 and pCMV-HNF4α7 were kindly provided by Dr. M. C. Weiss (Paris, France). EcoRI fragments of Ear-2 and Arp-1 cDNA from pMT2-Ear-2 and pMT2-Arp-1 (kindly provided by Dr. Ladias, Boston, MA) were cloned into pSG5 (Stratagene, London, UK) to generate respective expression vectors. pSG5-hHNF4α1 and pSG5-hHNF4α7-expression vectors were obtained by cloning in pSG5 hHNF4α1 and hHNF4α7 cDNAs amplified from human liver cDNA with primers: HNF4-P2 F: 5′-CACCATGGTCAGCGTGAACGCGCCC; HNF4-P1 F:5′-CACCATGGACATGGCCGACTACAGT; HNF4-stop Rev: 5′-CTAGATAACTTCCTGCTTGGT. The FG12-hHNF4α7 vector was obtained by cloning hHNF4α7 cDNA into Hpa1-linearized FG12.

Site-Directed Mutagenesis.

Mutations of CAR HNF4RE were performed using the QuickChange kit (Stratagene) on the pGL3b-237/+144hCAR reporter construct with forward primers:



Cell Culture, Transient Transfections and Reporter Gene Assays.

HepG2 and CV1 cell lines (American Type Culture Collection, Manassas, VA) were grown without antibiotic in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO). Cotransfections were performed with Fugene-6 (Roche) using 50,000 cells/P24well, luciferase-reporter (150 ng) and expression plasmids. The pSV-β–galactosidase (50 ng) (Promega, Charbonnières France) was used as transfection control. Medium was changed after 16 hours and cells were harvested 24 hours later in reporter lysis buffer. Extracts were analysed for luciferase and β-galactosidase activities as described.33

Chromatin Immunoprecipitation (ChIP) Assay.

Human hepatocytes chromatin was prepared as described.29, 31 Diluted chromatin (1 ml) was precleared for 1 hour with 50 μl salmon sperm/bovine serum antigen/protein A-sepharose (Upstate Biotechnology) (input), and incubated for 16 hours with 10 μg anti-human HNF4α (H171) or anti-human GR (E20, Santa Cruz Biotechnology). Immune complexes were collected with salmon sperm/bovine serum antigen/protein A-sepharose, purified, and one-tenth used as PCR template (Taq DNA polymerase, Invitrogen). Two couples of primers were used. Primer set A (hCAR promoter): 5′-GCCCAGCCCCTAGTCTCTTA (−433/−414) and 5′-TCAGGAGTTGCCAGTGATTG (+13/+32), and primer set B (hCAR exon 9): 5′-CGGTTTCTGTATGCGAAGTTG and 5′-TCTTTGGTCCCAGCATTTTC. PCR parameters were 40 cycles at 60°C for primer set A and 35 cycles at 55°C for primer set B.

Western Blot Analysis.

Cell lysates were prepared from 106 cells transfected with pSG5-hHNF4α1 or pSG5-hHNF4α7-expression plasmids and stored at −80°C. Immunoblotting was performed with anti-HNF4α antibodies (H-171).

Electromobility Shift Assays.

Electromobility shift assays were performed as described29, 31 with purified rat liver nuclear extracts (Geneka Biotechnology Inc., Montréal, Canada) or in vitro–synthesized hHNF4α1 or hHNF4α7 prepared by using the TNT system (Promega). Oligonucleotides used as radiolabeled probes or competitors (sense strand, mutation in bold): hCAR HNF4RE WT, 5′-TCCCACCCAGGCCT- TTGCCCTGAGTCCA-3′; Mut-I, 5′-TCCCACCCTTTCCTTTGCCCTGAGTCCA-3′; Mut-2, 5′-TCCC- ACCCAGGCCTTTAAACTGAGTCCA-3′. Anti-HNF4α antibody (H171) was used for supershift assays.

Transduction of human hepatocytes with lentivirus expressing HNF4α7. Human hepatocytes were transduced overnight with VSV-G pseudotyped bicistronic lentiviral vector FG12 (a gift from Dr. David Baltimore), or FG12-hHNF4α7, using green fluorescent protein as a marker for transduction efficiency. Total RNA was extracted 6 days later.


Preliminary analysis revealed that during human liver development (fetal-child-adult), CAR mRNA levels increase in parallel and correlate with HNF4α1 in adults (R = 0.834) but not with HNF4α7 (Supplementary Fig. 1). In addition, HNF4α7 mRNA is overexpressed in some HCC samples and this is accompanied by a marked decrease in CAR mRNA levels (Supplementary Table 1). These data suggested that HNF4α1 and HNF4α7 differently regulate CAR expression in man.

HNF4α1 Activates CAR Promoter.

The pGL3-hCAR-LUC deletions30 (−710/+144) were cotransfected in HepG2 cells with increasing amounts of an pSG5-HNF4α1-expression vector (Fig. 1A,B). Transcriptional activity of fragment −80/+144 was not affect by HNF4α1, whereas activity of all other fragments increased in a dose-dependent manner suggesting that the −237/−80 region is necessary and sufficient for full responsiveness of CAR promoter to HNF4α1.

Figure 1.

Functional analysis of hCAR promoter in HepG2 cells. (A) pGL3-hCAR-LUC deletion constructs. (B) Constructs were cotransfected with pSG5-hHNF4α1-expression vector (values at bottom are nanograms/well) in HepG2 cells, pRSV-β-gal being used as transfection control; 48 hours later, luciferase and β-galactosidase activities were determined. Values are means ± SD (n = 3) of luciferase activity normalized to β-galactosidase activity, determined in triplicate.

Identification of an HNF4α1 Response Element in hCAR Promoter.

Computer analysis of region −237/−80 revealed a potential DR1 element exhibiting significant homology with HNF4α consensus binding site. This motif (CAR/HNF4RE) is conserved between human and rodents (Fig. 2A). To determine whether this motif binds HNF4α electrophoretic mobility shift assays (EMSA) were performed. Rat liver extracts were incubated with radiolabeled double-strand oligonucleotides including HNF-4 consensus (lanes 1-6), putative CAR/HNF4RE (lanes 7-13), or CAR/HNF4-RE mutant I (lane 14) or II (lane 15) (Fig. 2B). A single retarded complex was observed with both HNF4α consensus (lane 2) and wild-type CAR/HNF4-RE (lanes 8 and 12) but not with mutated oligonucleotides (lanes 14 and 15). This complex was destabilized by an excess of unlabeled CAR/HNF4RE or HNF4α consensus, and supershifted by an anti-HNF4α antibody lane 13). CAR/HNF4RE mutant I and II oligonucleotides competed neither with HNF4α consensus (lanes 5 and 6) nor with CAR/HNF4RE (lanes 10 and 11).

Figure 2.

HNF4α1 binds to and transactivates a proximal DR1 element of hCAR promoter. (A) Sequences of HNF4α consensus binding element; human, mouse, and rat CAR/HNF4RE (DR1); and CAR/HNF4RE mutants I and II. Half-sites are underlined and mutations are in bold. (B) EMSA of HNF4α1 binding to CAR/HNF4RE. The [32P]-radiolabeled HNF4α consensus (lanes 1-6), putative CAR/HNF4RE (lane 7-13), or mutants CAR/HNF4RE I (lane 14) and II (lane 15) were incubated in the absence (lanes 1 and 7) or presence of rat hepatocyte nuclear extracts (lanes 2-6, 8-15). For competition experiments, 100-fold molar excess of oligonucleotides was added to the binding reaction. Anti-HNF4α antibody–mediated supershift experiment (lane 13). (C,D) Wild-type or mutated pGL3-hCAR-237/+144 was (C) transfected into CV1 or HepG2 cells or (D) co-transfected into CV1 cells in the absence or presence of pSG5-hHNF4α1-expression vector (values at bottom are nanograms/well). Luciferase activities: see legend to Fig. 1. Inset to (C): immunoblot analysis of HNF4α protein in 50 μg protein extracts from human hepatocytes (H.H.), HepG2, and CV1 cells. MW, molecular weight standards. (E) Human hepatocyte chromatin (input) was immunoprecipitated or not (noIgG) by antibodies against HNF4α, GR or control IgG. Precipitated DNA was analysed by semiquantitative PCR using specific primer sets for (A) proximal −433/+32 hCAR promoter or (B) hCAR exon 9 (B).

Both half-sites of CAR/HNF4RE were submitted to PCR-mediated mutagenesis using pGL3-hCAR-237/+144 as template. Wild-type or mutated pGL3-hCAR/HNF4RE/LUC constructs were transfected into HepG2 (expressing HNF4α1) and comparatively into CV1 cells (not expressing HNF4α1). As expected, wild-type construct activity was much higher in HepG2 than in CV1 cells. Both mutations strongly reduced HNF4RE activity in HepG2 but not in CV1 cells (Fig. 2C). Immunoblot analysis confirmed HNF4α protein expression in human hepatocytes (HH) and HepG2 cells (Fig. 2C, inset). Next, wild-type and mutated pGL3-hCAR-237/+144 were co-transfected into CV-1 cells with increasing amounts of pSG5-hHNF4α1 expression vector. Both mutations strongly reduced HNF4α1-mediated pGL3-hCAR-237/+144 transactivation. The mutation localized in the first half-site exhibited the strongest inhibitory effect, suggesting that mutation of the second does not completely impair HNF4α1 binding, notably for high amount of HNF4α1 (Fig. 2D).

Finally, HNF4α1 binding to CAR promoter was investigated in the context of the native promoter in primary human hepatocytes using chromatin immunoprecipitation assays. Anti-HNF4α antibodies [but neither antiglucocorticoid receptor (GR) nor control rabbit IgGs] immunoprecipitated DNA encompassing CAR promoter −433/+32 region (A) (Fig. 2E). In control experiments, DNA encompassing CAR exon 9 (B) was not immunoprecipitated by anti-HNF4α antibodies. These results suggest that HNF4α1 binds to hCAR/HNF4RE and transactivates CAR promoter.

HNF4α1 Is a More Potent Transactivator of hCAR Promoter than HNF4α7.

In contrast to HNF4α1, HNF4α7 lacks activation function module AF-1 and therefore exhibits a markedly reduced transcriptional activity.18 To compare the relative activity of both isoforms, pGL3-hCAR-237/+144 was co-transfected in CV1 cells with increasing concentrations of pSG5-hHNF4α1 or pSG5-hHNF4α7-expression vectors (Fig. 3A). Both isoforms transactivated CAR promoter activity. However, HNF4α1 was more potent than HNF4α7. Expression of both HNF4α1 and HNF4α7 proteins in transfected cell was verifed by immunoblot (Fig. 3B).

Figure 3.

HNF4α1 and HNF4α7 differently transactivate hCAR promoter and differently cooperate with PGC1α and SRC1. (A,C) CV1 cells were cotransfected with (A) pGL3-hCAR-237/+144 and pSG5-hHNF4α1 (black bars) or pSG5-hHNF4α7 (gray bars) expression vector, and (C) with or without PGC1α-expression vector or SRC1-expression vector (values at bottom of both panels are nanograms); 48 hours later, luciferase activities were determined as indicated in legend to Fig. 1. (B) Immunoblot analysis of HNF4α in 30 μg protein extracts, 48 hours after transient transfection with pSG5-hHNF4α1 or pSG5-hHNF4α7.

Next, we investigated the effect of PCG1α and SRC1 on HNF4α1 and HNF4α7 transcriptional activity (Fig. 3C). Although PGC1α stimulated both HNF4α1-mediated and HNF4α7-mediated hCAR transactivation to the same extent, SRC1 exhibited a marked specificity for HNF4α1. Neither PGC1α nor SRC1 compensated for the lower HNF4α7-mediated hCAR transactivation.

HNF4α1 and HNF4α7 Bind to the Same HNF4-RE of hCAR Promoter and HNF4α7 Acts as a Repressor of HNF4α1-Mediated Transactivation.

In vitro–synthesized HNF4α1 and HNF4α7 proteins were tested comparatively by EMSA for binding to CAR/HNF4RE (Fig. 4A,B). The retarded bands exhibited similar intensity, irrespective of isoform or isoform combinations, suggesting a similar affinity for CAR/HNF4RE. Competitive binding experiments on the complexes formed between HNF4α consensus probe and HNF4α1 or HNF4α7, using unlabeled CAR/HNF4RE oligonucleotide as competitor confirmed these observations. The median inhibitory concentration (IC50) calculated from quantification of complexes were on average 1.2 ± 0.6 and 0.6 ± 0.2 μM for HNF4α1 and HNF4α7, respectively (Fig. 4C,D). Thus, HNF4α1 and HNF4α7 bind to CAR/HNF4RE and, because HNF4α7 is a less potent transactivator, this factor may compete with HNF4α1 to repress CAR transcription.

Figure 4.

HNF4α1 and HNF4α7 binding to CAR HNF4RE. (A) EMSA with radiolabeled hCAR/HNF4RE in the presence of in vitro–translated hHNF4α1 (lanes 3 and 8) or hHNF4α7 (lanes 7 and 9) or combinations (lanes 4-6). Antibody against HNF4α was used for supershift experiment (lanes 8 and 9). (B) Immunoblot analysis of in vitro–synthesized hHNF4α1 or hHNF4α7 proteins used in experiments (A,C). (C) EMSA with radiolabeled HNF4α consensus oligonucleotide in the presence of in vitro–translated hHNF4α1 (lanes 2-8) or hHNF4α7 (lanes 9-15). For competition experiments, CAR/HNF4RE was added to the binding reaction at increasing molar excess ratios as indicated. (D) Densitometric analysis of complexes such as those shown in Fig. 4C (n = 2).

To test this hypothesis, pGL3-hCAR-237/+144 was cotransfected in CV1 cells with pSG5-hHNF4α1 or pSG5-hHNF4α7-expression vectors, alone or in combinations (Fig. 5). In all experiments, HNF4α7 repressed HNF4α1-mediated transactivation of hCAR promoter in a dose-dependent manner. Finally, HNF4α7 was overexpressed in primary human hepatocytes via a lentiviral vector (Table 1). HNF4α7 transduction repressed CAR mRNA content, while infection with control particles had no effect. HNF4α1 mRNA levels were not affected. These results suggest that, as observed in HCC (Supplementary Table 1, when the relative amount of HNF4α7 increases with respect to HNF4α1, CAR promoter transactivation is repressed.

Figure 5.

HNF4α1-induced hCAR promoter transactivation is repressed by HNF4α7. (A) CV1 cells were cotransfected with pGL3-hCAR-237/+144 and pSG5-hHNF4α1 or pSG5-hHNF4α7-expression plasmid alone or in combinations (values shown at bottom of both panels are nanograms/well); 48 hours later, luciferase activities were determined as indicated in legend to Fig. 1. (B) Same experiments with constant total amount of pSG5-hHNF4α1 plus pSG5-hHNF4α7-expression plasmids.

Table 1. HNF4α7 Overexpression in Hhuman Hepatocytes Decreases CAR mRNA Level
  1. NOTE. Human hepatocytes (2 donors, FT264, FT266) were exposed or not (CTRL) to increasing amounts of viral particles (multiplicity of infection: 5 to 20) with control (FG12) or HNF4α7 (α7)-expressing pseudo-HIV-1 lentivirus. Six days later, total RNA was isolated and HNF4α1, HNF4α7, CAR, and 18S RNAs were quantified by real-time RT-PCR. Relative mRNA levels were determined using gene-specific standard curves, after normalization with respect to 18S RNA. N.P: not performed.

FG12 (5)N.P.N.P.1.450.93
FG12 (10)1.680.871.060.89
FG12 (20)3.711.31N.P.N.P.
FG12-HNF4α7 (5)64.450.57374.810.20
FG12-HNF4α7 (10)93.700.50328.560.19
FG12-HNF4α7 (20)59.710.40N.P.N.P.


We showed that HNF4α1 and HNF4α7 affect differently CAR gene expression in human liver. Our conclusion is based on the following points: (1) a functional HNF4RE has been identified in hCAR promoter; (2) HNF4α1 and HNF4α7 bind this element with similar affinity; (3) HNF4α1 is a stronger transactivator than HNF4α7 which represses HNF4α1-mediated transactivation; and (4) CAR mRNA levels correlate positively with HNF4α1 in human liver and negatively with HNF4α7 in HCC.

Regulation of CAR expression by HNF4α is consistent with the tissue-specific expression and function of this receptor, and with its down-regulation in HNF4α-deficient mice.32 COUP-TFs repressed α1-mediated CAR transcriptional activation (Supplementary Fig. 2) as observed with peroxisome proliferator-activated receptor,39 and several steroid receptors.40, 41 Thus, expression of HNF4 target genes is expected to be affected by COUP-TFs. Indeed, the age-dependent decrease of apoAI expression has been shown to result from decreased binding capacity and expression of HNF4α and increased concentration and binding capacity of COUP-TFII.42 Moreover, PPAR expression is diminished in aged rats compared to younger animals,43 and we observed here that CAR mRNA levels were higher in young patients (Supplementary Fig. 1C). Because COUP-TF1 levels are higher in developing fetus44 this could contribute to the lower CAR expression in fetus.

HNF4α1 and HNF4α7 are expected to affect differently the transcription of their target genes because the latter lacks the AF1 transactivation domain.28, 45–47 Indeed, we observed that while PGC1α equally potentiates HNF4α1 and HNF4α7 transcriptional activity, p160/SRC1 potentiates only HNF4α1. This could explain why, although HNF4α1 and HNF4α7 bind CAR/HNF4RE with equal efficiency, hCAR promoter transactivation by HNF4α1 is much greater. HNF4α7 represses HNF4α1-mediated transactivation of CAR most likely through competitive binding to CAR/HNF4RE. Hence, CAR expression is dependent on the relative amounts of HNF4α1 and HNF4α7. The current findings that (1) CAR expression is low in HCC exhibiting elevated HNF4α7 levels but unchanged levels of HNF4α1 (Supplementary Table 1) and (2) overexpression of HNF4α7 decreases CAR expression in primary human hepatocytes support this notion. This is consistent with the finding that in mice expressing HNF4α7 only, CAR transcripts and TCPOBOP-mediated Cyp2b10 induction were dramatically reduced, while PXR levels were unchanged.48 Moreover, in fetal, child, and adult human livers CAR and AFP mRNA levels correlated, respectively, positively and negatively with HNF4α1 mRNA levels (Supplementary Fig. 1). Collectively, these observations suggest that CAR is a marker of differentiated mature hepatocyte. Because CAR controls bilirubin elimination by activating 5 components of the clearance pathway, reduced expression of CAR in the newborn should be responsible for neonatal jaundice.25

In mouse and human adult liver, CAR mRNA expression is greater than in fetus.32, 49 During embryogenesis, glucocorticoids play a critical role in liver differentiation and maturation.3, 50–52 However, GR must cooperate with other accessory DNA-binding proteins to fully activate its target genes. For example, GR induces DNA demethylation of TAT promoter, leading to the recruitment of HNF-3 that subsequently increases glucocorticoid response.53 In human livers, although GR expression is constant during development, TAT mRNA expression roughly parallels that of CAR (Supplementary Fig. 1) suggesting similar mechanism of regulation. We therefore wondered whether and GR cooperate to control CAR expression. Preliminary experiments (M.J. Vilarem, unpublished data) revealed that CAR transactivation by α1 (but not by α7 which lacks the AF-1 domain) is enhanced by GR, thus suggesting that both factors cooperate to transactivate hCAR promoter.

Although HNF4α controls the expression of pregnane X receptor (PXR) in the mouse fetal liver,49 induction of Cyp3a11 by pregnenolone 16α-carbonitrile and PXR expression were maintained in adult HNF4α-null mice.32 Thus, HNF4α-dependent regulation of PXR expression is specific only to the fetal and perinatal but not to the adult stage and contrasts with observations on CAR expression.32 CAR and PXR are therefore regulated differently during liver development.