During the inflammatory response, intrahepatic cholestasis and decreased drug metabolism are frequently observed. At the hepatic level, the orphan nuclear constitutive androstane receptor (CAR) (NR1I3) controls phase I (cytochrome P450 [CYP] 2B and CYP3A), phase II (UGT1A1), and transporter (SLC21A6, MRP2) genes involved in drug metabolism and bilirubin clearance in response to xenobiotics such as phenobarbital or endobiotics such as bilirubin. We investigated the negative regulation of CAR, a glucocorticoid-responsive gene, via proinflammatory cytokine interleukin 1β (IL-1β) and lipopolysaccharides (LPSs) in human hepatocytes. We show that IL-1β decreases CAR expression and decreases phenobarbital- or bilirubin-mediated induction of CYP2B6, CYP2C9, CYP3A4, UGT1A1, GSTA1, GSTA2, and SLC21A6 messenger RNA. This occurs via nuclear factor κB (NF-κB) p65 activation, which interferes with the enhancer function of the distal glucocorticoid response element that we have identified recently in the CAR promoter. We demonstrate that: (1) LPSs, IL-1β, or overexpression of p65RelA inhibit glucocorticoid receptor (GR)-mediated CAR transactivation; (2) these suppressive effects can be blocked both by pyrrolidine dithiocarbamate, an inhibitor of NF-κB activation, or by overexpression of SRIkBα, a NF-κB repressor; and (3) the GR agonist dexamethasone induces histone H4 acetylation at the proximal CAR promoter region, whereas LPSs and IL-1β inhibit this acetylation as assessed via chromatin immunoprecipitation assay. In conclusion, GR/NF-κB interaction affects CAR gene transcription through chromatin remodeling and provide a mechanistic explanation for the long-standing observation that inflammation and sepsis inhibit drug metabolism while inducing intrahepatic cholestasis or hyperbilirubinemia. Supplementary material for this article can be found on the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2004.)
During inflammatory pathologies, the liver reacts by increasing the level of acute phase response proteins.1 Concomitantly, the expression of other liver-specific proteins involved in drug2 and bilirubin metabolism3 is reduced. Cytochromes P450 2B6, 2C9, and 3A4 are members of the cytochrome P450 (CYP) monooxygenase superfamily, which plays an important role in the metabolism of xenobiotics and endogenous compounds. These enzymes are responsible for the metabolic activation or inactivation of the majority of clinically used drugs and many toxins.4 The expression of these genes is regulated by a variety of factors, including drugs, hormones, development, and diet. In particular, it has been observed that inflammatory stimuli such as lipopolysaccharides (LPS) or cytokines (e.g., interleukin 1β [IL-1β] and interleukin 6) cause a marked decrease in the expression and activity of CYP2B and CYP3A enzymes.5–10 In addition, inflammatory cytokines are potent inducers of intrahepatic cholestasis and hyperbilirubinemia, which are common clinical features of hepatic inflammation due to sepsis or autoimmune and viral hepatitis.3, 11 The cholestatic effect of cytokines is believed to result from the repression of genes that normally mediate the hepatic uptake, metabolism, and biliary excretion of bile salts and various non–bile salt organic anions (e.g., bilirubin).12, 13
A number of genes, including CYP2B6,14 CYP2C9,15 CYP3A4,16 SLC21A6,17 GSTA1 and GSTA2,17 UGT1A1,17 and MRP218 are known to be transcriptionally regulated by the constitutive androstane receptor (CAR) (NR1I3). CAR is predominantly expressed in the liver and intestine19 and controls xenobiotic- (such as phenobarbital20) and bilirubin-mediated17 induction of these genes. Recently, we demonstrated that CAR expression is controlled by glucocorticoids in human hepatocytes21 and identified a functional glucocorticoid response element (GRE) in the −4447 to −4432 promoter region.22 The glucocorticoid receptor (GR) is an important transcriptional regulator involved in embryonic development, cell differentiation, and metabolic homeostasis.23, 24 Interestingly, GR function is inhibited during the inflammatory responses, most notably by nuclear factor κB (NF-κB).25, 26
The mammalian NF-κB family of proteins consists of five members, namely Rel (c-Rel), p65 (Rel A), Rel B, p50, and p52.27 The regulation of NF-κB is achieved through interaction with the inhibitory protein inhibitor κB (IκB) that binds to NF-κB and sequesters it in the cytoplasm. Upon NF-κB activation via viral infection, proinflammatory cytokines including tumor necrosis factor and IL-1β, phorbol esters, ultraviolet irradiation, or bacterial LPSs, IκB is phosphorylated by IκB kinases and then degraded by the 26S proteasome. This leads to NK-κB translocation to the nucleus and activation of target genes.28, 29
In this study, we demonstrate that NF-κB, and more specifically p65 RelA, plays a direct role in LPS- and IL-1β–mediated transrepression of CAR by interfering with the action of GR. This inhibition—which is suspected to be the consequence of a previously characterized protein–protein interaction between the hormone-activated GR and NF-κB—is believed to result from competition for nuclear coactivators such as p300/CBP or SRC-1, and/or enhanced recruitment of histone deacetylases (reviewed by Almawi and Melemedjian30). These findings provide a rationale to the long-standing observation that inflammation and sepsis induce repression of CYP expression,6, 9, 31 intrahepatic cholestasis, and hyperbilirubinemia.12, 32
Ham F-12 and William's E culture media, phenobarbital (PB), LPSs, IL-1β, vitamins and hormones, collagenase, dimethylsulfoxide, pyrrolidine dithiocarbamate (PDTC), and dexamethasone (DEX) were obtained from Sigma (St. Quentin Fallavier, France). Collagen-coated culture dishes were obtained from Corning (Iwaki, Japan). α-(32P) deoxycytidine 5′-triphosphate, α-(32P) uridine 5'-triphosphate and electrochemiluminescence-developing reagents were obtained from Amersham-Pharmacia Biotech (Buckinghamshire, United Kingdom).
Cell Culture and Transfections.
HepG2 and HeLa cells purchased from the European Collection of Cell Cultures (Salisbury, United Kingdom) were maintained in Dulbecco's Modified Eagle Medium supplemented with 10% fetal calf serum and 100 μg/mL penicillin and streptomycin (Life Technology SARL, Cergy Pontoise, France). Transfection of plasmid DNA, including 25 ng pSV-galactosidase (Promega, Madison, WI) as control, was performed as described with Fugene-6 (Roche Diagnostics Corporation, Meylan, France). After 16 hours, the medium was renewed and 0.1% dimethylsulfoxide or inducers were added. After 24 hours of incubation, cells were harvested in reporter lysis buffer (Promega) and extracts analyzed for luciferase and β-galactosidase activities as previously described.22
The following hCAR promoter reporter constructs have been previously described22: −4477/+144hCAR-pGL3 (4477-LUC), −4410/+144hCAR-pGL3 (4410-LUC), −4477/-3910hCAR-tkLUV (4477wt-thLUC), and −4477/-3910hCAR GREmut (4477GREmut-thLUC). The human GR expression vector (pSG5-hGR) was provided by Dr. J.C. Nicolas (INSERM Montpellier, France). The pSG5 vector was obtained from Stratagene (Amsterdam, The Netherlands). pBSEN-hCAR and (CYP3A4 ER6)3-tk LUC have been previously described.15 The SRIκBα expression plasmid33 containing a serine to alanine mutation at positions 32 and 36 was provided by Dr. Y. Tian (College Station, TX). pRcCMV p50 and pRcCMV p65 RelA expression vectors were provided by Dr. D. Mathieu (INSERM Montpellier, France).
Liver Samples and Hepatocyte Cultures.
Human hepatocytes were prepared and cultured as previously described.8, 9 The cells were plated into collagen-coated dishes at 0.17 × 106 cells/cm2 in a hormonally and chemically defined medium consisting of a mixture of Willians' E and Ham F12 (1:1 in volume). This medium contains 100 nmol/L DEX. Treatments, as indicated, started 48 hours after plating.
Total RNA Purification, Northern Blot, and Ribonuclease Protection Assays.
Total RNA was isolated using TRIZOL reagent (Gibco BRL, Cergy-Pontoise, France). Purity was confirmed via spectrophotometry. For Northern blotting experiments, 25 μg of total RNA were analyzed using using α-(32P) deoxycytidine 5'-triphosphate–labeled rat GAPDH complementary DNA (cDNA) probe. For ribonuclease protection assays, 25 μg of total RNA were analyzed with specific RNA probes as previously described.22 The signals were analyzed with a PhosphoImager apparatus and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
Quantitative Polymerase Chain Reaction.
cDNA was synthesized from 1 μg of total RNA using the Superscript II first-strand synthesis system for polymerase chain reaction (PCR) (Invitrogen, Carlsbad, CA) at 42°C for 60 minutes in the presence of random hexamers. One tenth was used for quantitative PCR amplification using the Light Cycler apparatus (Roche Diagnostics Corporation, Meylan, France). The following program was used: a denaturation step at 95°C for 8 minutes was followed by 45 cycles of PCR (denaturation at 95°C for 15 seconds; annealing at 65°C for 8 seconds; elongation at 72°C for 18 seconds). Sense and reverse primers, respectively, were as follows: GAPDH, 5′-GGTCGGAGTCAACGGATTTGGTCG and 5′-CAAAGTTGTCATGGATGACC; β-actin, 5′-TGGGCATGGGTCA- GAAGGAT and 5′-TCCATCACGATGCCAGTGGT; CYP2B6, 5′-GGCCATACGGGAGGCCCTTG and 5′-AGGGCCCCTTGGATTTCCG; CYP3A4, 5′-CACAAACCGGAGGCCTTTTG-3′ and 5′-ATCCATGCTGTAGGCCCCAA-3′; CYP2C9, 5′-TCCTATCATTGATT- ACTTCCCG-3′ and 5′-AACTGCAGTGTTTTCCAAGC-3′; CAR, 5′-CCGTGTGGGGTTCCAGGTAG and 5′-CAGCCAGCAGGCCTAGCAAC; GSTA1 (NM_145740), 5′-CTGCCCGTATGTCCACCTGA and 5′-GGGCTGCCAGGCTGTAGAAA; GSTA2 (NM_000846), 5′-TTCGGTTGTCCAGCCACAAA and 5′-TGCACCAGCTTCATCCCATC; SLC21A6 (NM_006446), 5′-TCAATGGGAACCAGTCTGTGGA and 5′-TGTGAGGTGCCTCCAAGTGC. UGT1A1 primers were from the human cDNA sequence M57899 according to Basten et al.34: 5′-GGTGACTGTCCAGGACCTATTGA and 5′-TAGTGGATTTTGGTGAAGGCAGTT.
Extraction of Nuclear Proteins and Western Blot Analysis.
Nuclear extracts were prepared as reported previously.21 The proteins (30 μg) were separated by SDS/8% polyacrylamide gels and transferred to nitrocellulose membranes (Millipore Corporation, Bedford, MA). Blots were immunolabeled with antibodies against human GR (Santa Cruz Biotechnology, Santa Cruz, CA), p65 RelA (Santa Cruz Biotechnology), actin (Santa Cruz Biotechnology), or CAR (Dr. M. Negishi, National Institute of Environmental Health Sciences, Research Triangle Park, NC), and developed using horseradish peroxidase-coupled immunoglobulin G and the electrochemiluminescence detection system (Amersham Pharmacia Biotech).
Chromatin Immunoprecipitation Assay.
Human hepatocytes (10.106) were cultured in the absence of DEX for 16 hours, pretreated with or without 5 μg/mL LPS or 100 U/mL IL-1β for 12 hours, and then treated with 100 nmol/L DEX for 2 hours. Formaldehyde (1%) was then added to the tissue culture media, and the plates were incubated for 10 minutes at room temperature on a rocker. The cross-linking reaction was stopped with 125 mmol/L glycine for 5 minutes. Soluble chromatin was further isolated as described previously.22 The chromatin solution was sonicated and precleared with salmon sperm DNA/protein A–sepharose (Upstate Biotechnology, Lake Placid, NY). The precleared chromatin solution was incubated for 16 hours with 5 μg of anti-acetylated histone-4 antibodies (Upstate Biotechnology). Immune complexes were collected with salmon sperm DNA/protein A–sepharose, washed, and eluted in 1% SDS/0.1 M NaHCO3. Cross-links were reversed and chromatin-associated proteins were digested with proteinase K. The DNA was recovered via phenol/chloroform extraction and ethanol precipitation. Pellets were resuspended in 50 μL of water, and 20-μL aliquots were used as a template for PCR reaction (40 cycles) with the Expand Long Template PCR System (Roche Diagnostics Corporation). The primers for PCR of CAR promoter proximal fragment (−457 to +52) were 5′-CTGGGATTACAGGTGTGAGCCACCG and 5′-TTGCTGGTTTTCCTCTGATCTCAG.
IL-1β Inhibits CAR Target Gene Expression in Human Hepatocytes.
Because inflammatory cytokines are potent inhibitors of detoxication and inducers of intrahepatic cholestasis and hyperbilirebinemia, we tested the effect of IL-1β on the CAR-mediated induction of genes involved in drug metabolism and bilirubin clearance using quantitative PCR in primary hepatocyte cultures. For this purpose, hepatocytes were pretreated with IL-1β (100 U/mL) for 24 hours and then cultured in the presence of 0.5 mmol/L PB or 100 μmol/L bilirubin for 16 hours. As shown in Fig. 1, CYP2B6, CYP2C9, CYP3A4, SLC21A6, GSTA1, GSTA2, and UGT1A1 messenger RNA (mRNA) were induced by CAR activators. In contrast to a previous observation,18 no significant induction of MRP-2 mRNA was detected. IL-1β pretreatment decreased the basal expression of all mRNAs (P < .05, data not shown), except MRP-2, and blocked their PB- and bilirubin-mediated induction.
IL-1β Represses CAR Gene Expression in Cultured Human Hepatocytes.
We previously reported that PB-mediated CYP induction and CAR expression are glucocorticoid-dependent and are strongly reduced by interleukin 6 in human hepatocytes.9 In hepatocytes treated with 25–100 U/mL IL-1β for 16 hours, mRNA levels of CAR and tyrosine aminotransferase (TAT), a glucocorticoid receptor-regulated gene, exhibited a marked decrease; in contrast, GR mRNA expression was not affected (Fig. 2A). The time-course of CAR mRNA levels exhibited a decrease (40% inhibition after 8 hours of treatment) and reached a minimum after 12 hours (80% inhibition, P < .01, assessed by real-time PCR) before returning to initial level at 24 hours (Fig. 2B). Interestingly, TAT mRNA repression exhibited similar pattern suggesting a common mechanism involving GR inactivation (Supplementary Fig. 1).
LPS and IL-1β Induce NF-κBp65 RelA Activation in Cultured Human Hepatocytes.
A common denominator of the action of LPS and IL-1β is the activation of NF-κB.27 In human hepatocytes treated with LPS (0.5 or 5 μg/mL, Fig. 3A, lanes 2 and 3) or IL-1β (50 or 100 U/mL, Fig. 3A, lanes 4 and 5), NF-κB p65 RelA protein exhibited a dose-dependent nuclear accumulation; in contrast, GR protein level was not affected.
LPS and IL-1β Affect Neither PB-Mediated CAR Nuclear Translocation nor CAR's Constitutive Transcriptional Activity.
CAR protein rapidly translocates from cytosol to nucleus in response to PB or other activators.20 We therefore evaluated the effect of IL-1β and LPS after PB treatment on the nuclear accumulation of CAR via Western blot analysis. For this purpose, human hepatocytes were pretreated or not for 2 or 20 hours with IL-1β (100 U/mL) or LPS (5 μg/mL), then treated for 2 hours with 500 mmol/L PB. PB alone provoked an increased nuclear accumulation of CAR (Fig. 3B, lanes 2 and 6) compared with control cells (Fig. 3B, lanes 1 and 5). Pretreatment for 2 hours with both IL-1β (Fig. 3B, lane 3) and LPS (Fig. 3B, lane 4) did not affect CAR accumulation in the nucleus. This suggests that neither IL-1β nor LPS affect the nuclear translocation of CAR in response to PB. In contrast, when cells were pretreated for 20 hours with both IL-1β (Fig. 3B, lane 7) and LPS (Fig. 3B, lane 8), CAR was no longer detectable in nuclear extracts. This decrease might be the consequence of either reduced CAR mRNA expression (consistent with results shown in Fig. 2) or accelerated CAR degradation, or both. However, available antibodies failed to immunoprecipitate CAR to monitor the effect of IL-1β or LPS on CAR protein degradation.
The possibility that IL-1β and LPS affect the stability of the CAR protein was then evaluated. HeLa and HepG2 cells were cotransfected with a human CAR expression vector (pBSEN-hCAR) and a CAR-responsive (CYP3A4 ER6)3-tkLUC reporter construct and treated or not for 24 hours with IL-1β (25 or 100 U/mL) or LPS (1 or 10 μg/mL) before luciferase activity monitoring. Neither IL-1β nor LPS affected the constitutive transcriptional activity of CAR, and therefore, should not alter its stability (Supplementary Fig. 2). Together these results suggest that IL-1β and LPS do not affect CAR nuclear translocation and that the decrease in CAR accumulation in the nucleus of cells results from reduced mRNA expression.
LPS and IL-1β Provoke Histone H4 Deacetylation of Proximal CAR Gene Promoter in Human Hepatocytes.
Active gene transcription is associated with chromatin structure relaxation, which is consecutive to histone tail acetylation, a process allowing wider accessibility of DNA-binding proteins.35 In contrast, histone hypoacetylation is correlated with reduced transcription or gene silencing.36 Using chromatin immunoprecipitation assay, we investigated the acetylation of histones in the CAR promoter in human hepatocytes. Hepatocytes were first maintained 16 hours in a glucocorticoid-free medium, pretreated with or without 5 μg/mL LPS or 100 U/mL IL-1β for 12 hours, then treated with 100 nmol/L DEX for 6 hours. Histone H4 acetylation was increased by DEX in the proximal CAR promoter region (Fig. 4, lane 2) which contains the TATA box. Interestingly, both LPS (Fig. 4, lane 3) and IL-1β (Fig. 4, lane 4) drastically inhibited DEX-induced histone H4 acetylation. These data suggest that LPS and IL-1β affect chromatin structure in the vicinity of CAR promoter and CAR gene transcription.
LPS and IL-1β Repress Glucocorticoid-Mediated CAR Promoter Gene Activation.
We recently characterized a functional GRE within the distal 5′-flanking region of the human CAR promoter (−4447 to −4432).22 In view of the well-known functional interferences between glucocorticoids and proinflammatory cytokines (both NF-κB activators), the mutual inhibition of GR and NF-κB provides a possible explanation for the decrease in CAR expression. To test this hypothesis, we performed transient transfection of HeLa cells (in which GR is expressed) with a −4477/+144 human CAR promoter-luciferase construct (4477-LUC). DEX produced a sixfold increase in luciferase activity, while addition of LPS or IL-1β suppressed this increase in a dose-dependent manner (Fig. 5A). DEX, LPS, or IL-1β had no measurable effect on the luciferase activity of pGL3 plasmid or of the CAR promoter construct, which does not contain the GRE (4410-LUC). To eliminate the possibility that the observed transrepression of GR by LPS or IL-1β involves other regions of the CAR promoter, same assays were performed using the minimal thymidine kinase promoter (tk) fused to the −4477/−3910 region of CAR promoter containing the wild-type (4477wt-tkLUC) or mutated GRE (4477GREmut-tkLUC).22 In HeLa cells transfected with 4477wt-tkLUC construct, DEX provoked a 17-fold increase in luciferase activity, while addition of LPS or IL-1β suppressed this increase in a dose-dependent manner (Fig. 5B). In contrast, DEX, LPS, or IL-1β had no effect on the luciferase activity of the GRE-mutated reporter construct.
GR is not expressed at a significant level in HepG2 cells, which is in contrast to HeLa cells.22 Therefore, HepG2 cells represent an interesting model in the context of the present study. HepG2 cells were cotransfected with −4477/+144 hCAR pGL3 (4477-LUC) or TAT (GRE)2tkLUC reporter construct, with the pSG5-hGR expression vector or the pSG5 empty vector as a control. Sixteen hours later, cells were pretreated with either LPS (5 μg/mL) or IL-1β (100 U/mL) for 12 hours and then cultured for 24 hours in the presence of 50 nmol/L DEX. LPS and IL-1β had no inhibitory effect on hCAR 4477-LUC reporter gene expression, but repressed the DEX-mediated 4477-LUC reporter gene transcription in the presence of GR (Fig. 5C). Similar results were obtained with the TAT GRE reporter construct. These results seem to exclude direct action of LPS or IL-1β on CAR expression and are consistent with the lack of a consensus binding site for NF-κB/c-Rel homodimeric and heterodimeric complexes (GGGGACTTTCCC) within the human CAR promoter sequences (−4711/+144). These observations confirm that CAR gene repression occurs at the transcriptional level and strongly suggest that LPS and IL-1β inhibit CAR gene expression by interfering with ligand-activated GR transactivation at the distal hCAR-GRE enhancer.
NF-κBp65 RelA Inhibits GR-Mediated CAR Promoter Gene Activation.
To determine which of the two NF-κB subunits p65RelA and p50 is responsible for the suppressive effect on CAR promoter, HepG2 cells were cotransfected with GR and either p65 or p50 expression vectors and with the 4477wt-tkLUC or 4477GREmut-tkLUC reporter constructs. Twenty-four hours later, cells were treated for 16 hours with 50 nmol/L DEX. Transactivation of the 4477wt-tkLUC reporter construct was inhibited by p65 in a dose-dependent manner, but not by p50 (Fig. 6A). In control experiments, the 4477GREmut-tkLUC construct was found not to be responsive to GR as expected, nor was it responsive to p65 or p50. These assays were repeated with two homologous CAR promoter reporter 4477-LUC and 4410-LUC constructs. The same dose-dependent repression of GR-transactivation by p65 was observed with the 4477-LUC-pGL3 construct, while neither p50 nor p65 affected luciferase activity of the 4410-LUC construct (Fig. 6B).
NF-κB Inhibitors Antagonize LPS- or IL-1β–Induced Inhibition of CAR Expression Upon Glucocorticoid Treatment.
To further confirm that NF-κB is directly involved in CAR gene repression, we used PDTC, a chemical inhibitor of LPS- and IL-1β–mediated NF-κB activation.33 When HeLa cells were transfected with 4477wt-tkLUC and treated with LPS, PDTC significantly attenuated LPS-induced suppression of CAR reporter gene activity after DEX treatment (Fig. 7A). Similar results were obtained with IL-1β (data not shown). Similarly, pretreatment of human hepatocytes with PDTC reversed the IL-1β–induced suppression of CAR mRNA (Fig. 7B). Note that CAR mRNA levels were higher in the presence of 100 μmol/L PDTC compared with control. This suggests that hepatocyte isolation and culture activate NF-κB, as observed during rat hepatocyte isolation.37
SRIκB-α is a mutant of IκB-α, with a serine to alanine mutation at residues 32 and 36. These mutations render SRIκB-α phosphorylation impossible and make the protein resistant to degradation by the proteasome, thereby causing constitutive inhibition of NF-κB.33 HepG2 cells were transiently cotransfected with −4477/−3910 hCAR-tkLUC, GR, p65, and increasing amounts of SRIκB-α expression plasmids; p65 caused a suppression of the glucocorticoid-induced reporter gene activity as expected. However, this effect was reversed by SRIκB-α (Fig. 7C). These results confirm the negative regulation of CAR expression by NF-κB through the p65 moiety of this transcription factor.
Episodes of inflammation and/or infection are frequently associated with a decreased capacity of the liver to metabolize drugs, xenobiotics, and bilirubin and to regulate bile secretion.38 Although cytokines and bacterial endotoxins are known to play a critical role in this process, the molecular mechanisms remain unclear. The data presented here suggest that CAR, a glucocorticoid-responsive gene that plays a major role in detoxification, is subjected to a transrepression process in response to proinflammatory cytokines through NF-κB–mediated inhibition of the transcriptional activity of the glucocorticoid receptor. The repressive effect of LPS or IL-1β observed in this work on the basal expression of CYP2C9, SLC21A6, UGT1A1, GSTA1, and GSTA2 genes suggests that other transcriptional activators are affected and/or that CAR controls the basal expression of these genes. In fact, UGT1A1,47 GSTA1 and GSTA2,48 and CYP2C915 are regulated by GR in addition to CAR; this accounts for the repression of their basal expression. In addition, we (manuscript in preparation) and others6 have observed that retinoid X receptor α and pregnane X receptor gene expression is also decreased by LPS and IL-1β. Collectively, these data suggest that NF-κB activation by proinflammatory molecules—through the inhibition of CAR gene expression—inhibits directly or indirectly the expression of proteins involved both in drug metabolism and bilirubin homeostasis (Supplementary Fig. 3).
The majority of papers dealing with the effects of cytokines on liver detoxification mainly focus on the down-regulation of CYP genes. Various mechanisms have been proposed to explain this effect: tumor necrosis factor α represses CYP1A1 via redox regulation of nuclear factor 1,39 interleukin 2 represses CYP2C11 and CYP3A2 via induction of the proto-oncogene c-myc,40 interleukin 1 inhibits CYP2C11 transcription via NF-κB binding to the promoter,41 and interleukin 6 inhibits CYP3A4 via induction of C/EBPβ-LIP a negative competitor of C/EBPα.42 In contrast, little is known regarding the effect of cytokines on other genes involved in liver detoxication function. Here, we show that the organic anion transporter SLC21A6 (entry of free-bilirubin in hepatocytes), UDP-glucuronosyltransferase 1A1 (UGT1A1) (bilirubin glucuronidation), and GSTA1 and GSTA2 (bilirubin-binding proteins) are also down-regulated by IL-1β. All of these genes,17 as well as MRP2 (controlling the efflux of xenobiotics and bilirubin diglucuronide across the apical membrane of the hepatocyte into the bile canaliculi),18 are controlled by CAR and pregnane X receptor. Consistent with this, CAR-null mice display elevated bilirubin levels,17 while bilirubin levels are decreased by PB in humans.43
In previous work, we demonstrated that CAR expression is controlled by the GR via a GRE located approximately 4.7 kb upstream of the promoter.22 Because IL-1β activates NF-κB in hepatocytes, and because NF-κB is known to inhibit the transcriptional activity of GR,25, 26 we hypothesized that the down-regulation of CAR by IL-1β is mediated at the transcriptional level through the negative effect of NF-κB on GR. The results presented here are consistent with this hypothesis: (1) transactivation of CAR promoter by GR was down-regulated by LPS or IL-1β, or after overexpression of the p65 RelA subunit of NF-κB; (2) the suppressive effect of LPS or IL-1β on CAR mRNA expression in hepatocytes was reversed by PDTC, a compound known to inhibit NF-κB action, and by SRIκBα, a NF-κB repressor; and (3) glucocorticoid-induced acetylation of histone H4 at the CAR promoter was decreased by LPS or IL-1β in hepatocytes.
In summary, the mechanism by which NF-κB inhibits CAR transcription is unknown. Both GR and NF-κB share transcriptional coactivators such as p300/CBP or SRC-1,44, 45 and it is conceivable that competition occurs if coactivator availability is limiting, as has been shown recently for tumor necrosis factor α– and LPS-mediated transrepression of AhR-induced Cyp1a1 expression by NF-κB.33 Because coactivators possess histone acetyl transferase activity, it is possible that NF-κB exerts its negative effect on the GR transactivation complex in the context of the CAR promoter by inhibiting histone acetyl transferase activity and/or by increasing the recruitment of histone deacetylase, which in contrast to histone acetyl transferase favors chromatin condensation. This mechanism has been proposed for the transrepression of the granulocyte-macrophage colony-stimulating factor by glucocorticoids.46 Granulocyte-macrophage colony-stimulating factor is activated by proinflammatory cytokines such as IL-1β through NF-κB activation, and DEX has been shown to inhibit IL-1β–stimulated histone acetylation by increasing the recruitment of histone deacetylase. Moreover, DEX has been shown to partially prevent the inhibition of organic anion transport between hepatocytes and bile ducts in endotoxaemic rats.50 It is therefore possible that anti-inflammatory effects of glucocorticoids and antiglucocorticoid effects of proinflammatory cytokines are mediated through similar mechanisms. Finally, and consistent with our hypothesis, Khatsenko et al. reported that LPS inhibits the PB-induced expression of CYP2B1/2 in the rat, while this effect is attenuated in rats treated with inhibitors of NO synthase, an enzyme whose activity is known to interfere with NF-κB activation.50
The authors thank Drs. M. Negishi (National Institute of Environmental Health Sciences, Research Triangle Park, NC), and J. C. Nicolas (INSERM U540, Montpellier, France) for providing various plasmids and antibodies.