Interleukin-1beta suppresses retinoid transactivation of two hepatic transporter genes involved in bile formation.

Cytokines have been implicated in the pathogenesis of inflammatory cholestasis. This is due to transcriptional down-regulation of hepatic transporters including the Na(+)/bile acid cotransporter, ntcp, and the multispecific organic anion exporter, mrp2. We have recently shown that ntcp suppression by lipopolysaccharide in vivo is caused by down-regulation of transactivators including the previously uncharacterized Footprint B-binding protein. Both the ntcp FpB element and the mrp2 promoter contain potential retinoid-response elements. We hypothesized that retinoic acid receptor (RAR) and retinoid X receptor (RXR) heterodimers would activate these two genes and that cytokines that reduce bile flow might do so by suppressing nuclear levels of these transactivators. Retinoid transactivation and interleukin-1beta down-regulation of the ntcp and mrp2 promoters were mapped to RXRalpha:RARalpha-response elements. Gel mobility shift assays demonstrated specific binding of RXRalpha:RARalpha heterodimers to the ntcp and mrp2 retinoid-response elements. The RXRalpha:RARalpha complex was down-regulated by IL-1beta in HepG2 cells. An unexpected finding was that an adjacent CAAT-enhancer-binding protein element was required for maximal transactivation of the ntcp promoter by RXRalpha:RARalpha. Taken together, these studies demonstrate regulation of two hepatobiliary transporter genes by RXRalpha:RARalpha and describe a mechanism which likely contributes to their down-regulation during inflammation.

A broader appreciation of the role of lipopolysaccharide (LPS) 1 and associated cytokines in the pathogenesis of a num-ber of diseases involving the liver has begun to emerge (1)(2)(3). Conditions such as sepsis, alcoholic, autoimmune and viral hepatitis, and parenteral nutrition-associated liver disease involve activation of hepatic inflammation. Cholestasis and hyperbilirubinemia are common clinical features of these disorders (1). As a principal target of inflammatory mediators, the liver regulates changes in hepatic protein synthesis during the acute phase response (4 -6). This can include reduction in the expression of several proteins involved in bile formation, and thus cholestasis can be categorized as a feature of the hepatic negative acute phase response (7,8). These transport proteins include sodium/taurocholate cotransporting polypeptide (ntcp) and bile salt export protein, which are the principal bile acid uptake and export proteins, respectively, as well as mrp2 (the multispecific organic anion exporter). Reduced function of the ntcp transporter leads to decreased bile acid uptake by the hepatocyte, whereas reduced function of the mrp2 transporter leads to impaired excretion of bilirubin, glutathione, and numerous organic anions, as well as to reduction of the bile salt-independent component of bile flow (7,9).
Tumor necrosis factor ␣ (TNF-␣), interleukin-1␤ (IL-1␤), and interleukin-6 (IL-6) are cytokines known to mediate acute phase changes in hepatic protein synthesis at the transcriptional level (6,10). However, the molecular mechanisms linking down-regulation of the hepatic transporters during the acute phase response are not well understood. Recent studies in rodents have begun to elucidate these mechanisms. ntcp and mrp2 gene expression are profoundly down-regulated after administration of LPS or effector cytokines to rats (3,(11)(12)(13). As recently shown in our laboratory, hepatic nuclear concentrations of two ntcp transactivators, hepatocyte nuclear factor 1 (HNF1) and Footprint B-binding protein (FpB BP), were reduced after LPS treatment of rats, providing a molecular mechanism for acute-phase suppression of ntcp expression (9). However, it was not known whether LPS-induced cytokines such as TNF-␣ or IL-1␤ would be sufficient to suppress nuclear levels of ntcp or mrp2 transactivators and associated gene promoter activity or whether additional inflammatory mediators were also contributing in vivo.
The identity of the protein(s), which comprise the ntcp transactivator, FpB BP, has been under investigation. Analysis of the ntcp FpB BP-response element, (5Ј-GGGGCAtaAGGTTA-3Ј) (nt Ϫ53/Ϫ40), demonstrates homology to a retinoid x receptor:retinoid ␣ receptor (RXR:RAR) consensus element (Table I) (14,15). The mrp2 promoter also contains several putative retinoid-response elements (16). RXR-containing heterodimers regulate a broad range of hepatic metabolic functions including bile acid synthesis, fatty acid and oxysterol metabolism, and cytochrome oxidase metabolism of medications (17)(18)(19)(20)(21)(22)(23)(24)(25). Recent reports have suggested that some nuclear hormone receptors, including RXR, and their respective target genes are suppressed by endotoxin and associated cytokines at the transcriptional level (26,27). However, little is known of retinoid regulation of hepatobiliary transport, and more specifically it was not known whether retinoids would be involved in the regulation of ntcp and mrp2 gene expression. We hypothesized that suppression of ntcp and mrp2 by cytokines might occur via suppression of retinoid transactivators, including RXR␣ and RAR␣.
In this study, we identify FpB BP as a RXR␣:RAR␣ heterodimer and characterize RXR␣:RAR␣-response elements within the rat ntcp and mrp2 gene promoters. These response elements mediate induction by retinoids and suppression by IL-1␤ of gene expression. In addition, an adjacent binding site for CAAT/enhancer-binding proteins (C/EBP) is shown to modulate RXR␣:RAR␣ transactivation of the ntcp promoter. A proposed molecular mechanism of ntcp and mrp2 acute phase transcriptional suppression via IL-1␤-induced reduction in the concentration of nuclear RXR␣:RAR␣ heterodimers is described. These mechanisms likely contribute to the reduction in transcription of these two genes during the acute phase response.

EXPERIMENTAL PROCEDURES
Materials and Plasmids-Cell culture medium was purchased from Life Technologies, Inc. HepG2 cells were obtained from the American Type Culture Collection (ATCC), Rockville, MD. Cytokines were obtained from R&D Sysytems, Minneapolis, MN. Luciferase assay reagents and the TNT TM in vitro protein synthesis kit were obtained from Promega, Madison, WI. LPS and the lactate dehydrogenase (LDH) and TACS™ DNA ladder apoptosis assay kits were obtained from Sigma. The RXR␣-specific ligand, LG364, was kindly provided by Dr. R. Heyman, (Ligand Pharmaceuticals, Inc., San Diego, CA). The RAR␣-specific ligand, TTNPB, and the peroxisome proliferator activator receptor ␣ (PPAR␣) -specific ligand, WY16143, were purchased from Biomol, Plymouth Meeting, PA. C/EBP␣, C/EBP␤, and signal transducers and activators of transcription 1 antibodies for supershift assays were obtained from Santa Cruz Biotechnology, Inc., Santa Cruz, CA. RXR␣ and RAR␣ antibodies were kindly provided by Dr. E. Allegretto (Ligand Pharmaceuticals, Inc.). The albumin promoter construct (p-787albLUC) was kindly supplied by Dr. K. Zaret (Brown University). The IL-6 promoter construct (p-IL6 -225LUC) was kindly supplied by Dr. A. Ray (Yale University). The herpes simplex virus thymidine kinase expression plasmid, p-TKLUC, was obtained from the ATCC (ATCC name: pT109luc). Expression plasmids for C/EBP␣ and C/EBP␤ were kindly provided by Dr. R. Costa (University of Illinois, Chicago). Expression plasmids for RAR␣, RXR␣, and PPAR␣ have previously been described (20).
Cell Culture-HepG2 human hepatoma cells were plated in either 12-well plates (10 5 cells/well) for transient transfections or 10-cm plates (10 6 cells/plate) for isolation of nuclear extracts. Cells were maintained in minimal essential medium supplemented with 10% fetal bovine serum and penicillin-streptomycin-glutamine.
Plasmid Constructs-Plasmids containing rat ntcp promoter sequences Ϫ6kb/ϩ47, Ϫ1237/ϩ47, and Ϫ158/ϩ47 inserted into the luciferase vector pSVoAL were used as the standard ntcp promoter test plasmids, p-6kbntcpLUC, p-1237ntcpLUC, and p-158ntcpLUC, as previously reported (14). A plasmid containing the mrp2 promoter sequence from Ϫ1044/Ϫ15, p-1044mrp2LUC, was created by subcloning an insert generated by polymerase chain reaction from rat genomic DNA into pSVoAL. It should be noted that the sequence of the resulting insert was different from the published sequence at nt Ϫ115, with a G in lieu of an A (16). This sequence was confirmed from multiple independent polymerase chain reactions using separate preparations of rat genomic DNA template. Rat ntcp and mrp2 constructs, including 5Јdeletion plasmids and internal mutations of the FpB and Footprint C (FpC) elements, were constructed as described previously (14). Five copies of the ntcp FpB element were inserted upstream of the herpes simplex virus thymidine kinase promoter to create p5xFpBTKLuc. All sequences were verified by automated sequencing (Keck Biotechnology Center, Yale University).
Transient Transfections-HepG2 cells were grown to 50% confluency prior to transfection with luciferase reporter plasmids utilizing the calcium phosphate:DNA coprecipitation technique as previously reported, with the exception of a 16-h calcium-DNA exposure and no glycerol shock (14). P-Rous sarcoma virus-␤Gal was used as an internal control for transfection efficiency.
Retinoid and Cytokine Treatments-After transfection, retinoids or cytokines were added to serum-free medium for treatment of cells for 4 -24 h (28). For nuclear hormone treatments, HepG2 cells were cotransfected with ntcp or mrp2 promoter reporter plasmids and RAR␣, RXR␣, or PPAR␣ expression plasmids together with their respective ligands. After exposure, cells were harvested, and cytoplasmic extracts were assayed for luciferase and ␤-galactosidase activities according to the manufacturer's directions. Transfections and retinoid or cytokine treatments were performed in triplicate at least twice. LDH release and the degree of apoptosis in control and treated cells were determined utilizing a colormetric LDH release assay and the TACS™ DNA apoptotic laddering kit, respectively. Cell number and trypan blue exclusion were determined for control and treated cells. HepG2 cells were also plated on 10-cm plates until reaching approximately 50% confluency. They were then treated with optimal doses of cytokines (as indicated by functional promoter assays) for times ranging from 4 to 24 h, and nuclear extracts were obtained.
Electrophoretic Mobility Shift Assays-Liver nuclei were prepared from control and cytokine-or retinoid-treated cells or control rats at the given time points according to published methods (9). Protein concentrations were determined using the Bio-Rad Bradford reagent kit. Nuclear hormone receptors were also synthesized utilizing the TNT TM kit with the respective T7 promoter-driven expression plasmids. The DNA sequence of the sense strand of each oligonucleotide is listed in Table I. Oligonucleotide probes (1 pmol/l) were end-labeled using 50 Ci of [␥-32 P]adenosine 5Ј-triphosphate (ATP) and 1 l of polynucleotide kinase T 4 . Five g of nuclear extracts or four l of in vitro synthesized proteins were incubated with 2 g of poly(dI-dC)-poly(dI-dC) in a 20-l reaction containing 25 mM HEPES, pH 7.6, 50 mM KCl, 0.5 mM dithiothreitol, 0.5 mM EDTA, 5 mM MgCl 2 , 10% glycerol on ice for 10 min. 2.5 ϫ 10 4 cpm of 32 P-end-labeled oligonucleotide (Table I) were added, and the binding reaction was allowed to proceed on ice for a total of 30 min. In competition assays, 10 -100-fold molar excess of the specific unlabeled oligonucleotide was added to the binding mixtures along with the labeled oligonucleotide. In supershift studies for RXR␣ or RAR␣, nuclear extracts were incubated with 1 l (2 mg/ml) of the relevant antibody for 2 h on ice prior to the addition of the radiolabeled probe. In supershift studies for C/EBP, nuclear extracts were incubated with 1 l (1 mg/ml) of the relevant polyclonal antibody for 1 h on ice subsequent to the addition of the radiolabeled probe. The entire sample was then electrophoresed through a nondenaturing 6% polyacrylamide gel in 0.25 ϫ Tris borate-EDTA buffer at 12 V/cm for 2 h.
Statistical Analysis-Data were expressed as the mean Ϯ S.D. or the mean Ϯ S.E. of experiments with at least three independent transfec- tions or treated HepG2 cells/group. Differences among experimental groups were analyzed by the unpaired t test.

RESULTS
ntcp and mrp2 Promoters Are Transactivated by Retinoidsntcp and mrp2 promoter constructs were transactivated by cotransfection of RXR␣ and RAR␣ expression plasmids activated by the specific ligands LG364 and TTNPB. ntcp promoter activity was increased 17-fold by combined treatment, with less activation by the individual receptors (Fig. 1A). Treatments with ligands alone or unactivated receptors had no significant effect on ntcp promoter activity (data not shown). The ntcp Ϫ36/ϩ47 construct was not activated by retinoids, placing the retinoid-response element between Ϫ158/Ϫ36. This region of the promoter includes the previously described FpB (nt Ϫ56/ Ϫ37) and FpC (nt Ϫ81/Ϫ62) elements, which contribute to basal activity in hepatocytes (14). Investigations into the identity of the FpB-and FpC-binding proteins and their transactivation by retinoids will be described below.
The rat mrp2 promoter full-length construct (p-1044mrp2Luc) was up-regulated by ligand activation of cotransfected RXR␣ and RAR␣ expression plasmids. Treatment of a series of 5Ј-deletion constructs localized the retinoid-response element between mrp2 nt Ϫ422/Ϫ398 (Fig. 1B). This element includes the sequence 5Ј-GGGTATTTAACATCTCT-GTGAACTC-3Ј, which contains a putative direct repeat (DR5)response element on the antisense strand (underlined). Treatment with PPAR␣ activated by WY16143 Ϯ ligand-activated RXR␣ had no effect on activities of either ntcp or mrp2 promoters. These data indicate that the ntcp and mrp2 promoters contain retinoid-response elements and are up-regulated by ligand-activated RXR␣:RAR␣ heterodimers.
downstream of the DR2 element did not affect activation by retinoids. These data indicate that an intact DR2 element is required to fully mediate retinoid activation of the ntcp promoter.
The ntcp FpB Element Confers Retinoid Responsiveness to a Heterologous Promoter-To confirm that the FpB element was sufficient to mediate promoter activation by retinoids, HepG2 cells were transfected with a construct containing multiple copies of the FpB-response element inserted upstream of the herpes simplex virus thymidine kinase (TK) promoter (p-5xF-pBTKLuc). The vector plasmid pTKLUC was suppressed by retinoids, whereas p-5xFpBTKLuc was activated Ϸ18-fold by ligand-activated RXR␣ and RAR␣ (Fig. 3). Moreover, it was important to determine if a CCAAT element in the herpes simplex virus TK promoter would mediate TK promoter activation via C/EBP, because subsequent experiments identified the FpC-binding protein as C/EBP and demonstrated that this element modulates retinoid activation of ntcp (29 -31). Cotransfection of C/EBP␣ did not activate the pTKLUC construct, whereas a control 3xC/EBPLuc construct was up-regulated 50-fold, arguing against the possibility that a functional C/EBP element exists in the TK promoter (data not shown). Thus, the retinoid activation observed for the p5xFpBTKLuc construct is likely due solely to the FpB element, without a contribution from the CCAAT element in pTKLUC.
The ntcp FpC Element Is Differentially Regulated by C/EBP Isoforms and Contributes to Retinoid Activation of FpB-The ntcp FpC element (nt Ϫ81/Ϫ62) bears homology to a C/EBP site and is adjacent to the RXR:RAR binding site. Because C/EBP has been shown to cooperate with nuclear hormone receptors in activating at least one other hepatic gene promoter, studies were then performed to determine if the previously identified ntcp FpC element was indeed a C/EBP element (35). The Ϫ158/ ϩ47 ntcp promoter was activated 5-fold by cotransfected C/EBP␣, whereas a mutation in FpC (FM14) essentially eliminated C/EBP␣ transactivation (Fig. 4A). The predominant basal C/EBP isoform in liver is C/EBP␣, whereas C/EBP␤ is induced in the acute phase response (32)(33)(34). To examine the effect of C/EBP␤ upon ntcp expression, increasing amounts of C/EBP␤ were cotransfected with a fixed amount of C/EBP␣. Increasing amounts of C/EBP␤ led to a dose-dependent decrease in ntcp promoter activity (Fig. 4A). EMSA using radiolabeled FpC oligonucleotide and rat liver nuclear extract revealed three specifically bound complexes (Fig. 4B). The supershift pattern obtained after coincubation with C/EBP␣ and/or C/EBP␤ antibodies indicated that these complexes were comprised of C/EBP␣ and C/EBP␤ homodimers and a C/EBP␣: C/EBP␤ heterodimer, confirming that in native liver FpC is bound predominately by C/EBP proteins. Interestingly, the internal mutation, which altered the FpC element (FM14) but retained the FpB element, reduced ntcp promoter activation by retinoids from 17-fold to approximately 7-fold (see Fig. 2). Thus, the ntcp FpB element appears to require cooperation from an adjacent C/EBP element for optimal promoter activation by retinoids.
ntcp FpB and mrp2 nt Ϫ422/Ϫ398 Response Elements Are Bound by RXR␣:RAR␣ Heterodimers-Preliminary work in our laboratory indicated that the FpB BP is a multiprotein complex. UV-cross-linking analysis demonstrated bound species of ϳ50 and ϳ100 kilodaltons (data not shown). Moreover, attempts to clone a FpB BP with yeast one hybrid screening failed to identify a single polypeptide species that bound FpB. To confirm that a RXR␣:RAR␣ heterodimer could bind the FpB element, these proteins were synthesized in vitro and then incubated with the radiolabeled FpB oligonucleotide, with the resulting complex examined by EMSA. As shown in Fig. 5A, neither RXR␣ nor RAR␣ alone bound radiolabeled FpB, whereas coincubation demonstrated a specifically shifted complex. Mutation of FpB to FM1 eliminated binding by RXR␣: RAR␣, whereas a consensus DR2 oligonucleotide provided the same binding pattern as FpB. It is important to note that coexpression of other liver-enriched RXR partners (PPAR␣ and liver X receptor) with RXR␣ did not form complexes that bound FpB (data not shown). The mrp2 Ϫ422/Ϫ398 element is homologous to a DR5 element and was also bound by RXR␣:RAR␣. These data confirmed that RXR␣:RAR␣ can bind both the ntcp FpB and mrp2 nt Ϫ422/Ϫ398 elements.
To examine binding of RXR␣:RAR␣ to the ntcp FpB and mrp2 Ϫ422/Ϫ398 elements in native rat liver nuclei, EMSA was performed utilizing rat liver nuclear extracts. Supershifting antibodies developed against yeast-expressed RXR and RAR proteins were tested against these rat liver nuclear complexes (36). As shown in Figs. 5B and 6, the ntcp-and mrp2shifted complexes migrated with the same mobility as doublets on EMSA. The anti-RAR␣ antibody significantly reduced the intensity of the doublets for both the ntcp and mrp2 elements and demonstrated a modest supershift for the mrp2 complex (Fig. 5B, arrow). The supershift for the mrp2 element was more

FIG. 4. The ntcp FpC element mediates promoter activation by C/EBP␣ and suppression by C/EBP␤.
A, the wild-type ntcp Ϫ158/ϩ47 promoterdriven luciferase reporter construct and the FM14 construct, which contained an internal mutation of the FpC element, were treated with C/EBP␣ for 24 h. The Ϫ158/ϩ47 construct was also treated with a fixed amount of C/EBP␣ (200 ng) and increasing amounts of C/EBP␤ (100 -500 ng) for 24 h. Lysates were analyzed as under "Experimental Procedures." Values are means of data normalized to control activity Ϯ S.E. *, p Ͻ 0.05 versus the untreated control; #, p Ͻ 0.05 versus the C/EBP␣-treated control. B, EMSA was performed using the radiolabeled FpC element (Ϫ81/Ϫ62) and rat liver nuclear extracts Ϯ C/EBP␣ and/or C/EBP␤ antibodies. Image analysis under "Experimental Procedures." WT, wild type; STAT, signal transducers and activators of transcription.
pronounced when the anti-RXR␣ and anti-RAR␣ antibodies were coincubated with the nuclear extracts. The anti-RXR␣ antibody significantly reduced the intensity of the mrp2 DR5 complex but did not have a significant effect upon the intensity of the shifted complex for the ntcp DR2 element. This is likely due to the spatial organization of DR2 and DR5 RXR:RAR complexes and their accessibility by these particular antibodies (see "Discussion").
ntcp and mrp2 Retinoid Elements Cross-compete for Liver Nuclear-binding Proteins-EMSA was then performed utilizing rat liver nuclear proteins to determine if native mammalian receptors bound by the ntcp and mrp2 retinoid elements could be cross-competed by the respective binding sites (Fig. 6). Densitometry demonstrated that the unlabeled FpB oligonucleotide at 100-fold excess reduced mrp2422 signal intensity by 80%, whereas the mrp2422 oligonucleotide reduced FpB signal intensity by 95%. A 100-fold excess of unlabeled mutated oligonucleotide (FM1) did not compete for binding to the native FpB BP complex. These data demonstrate that the mrp2 DR5 element and the ntcp DR2 element could cross-compete for native rat liver-binding proteins. Whether or not a hierarchy exists for relative binding affinities remains to be determined.
Cytokines Suppress the ntcp and mrp2 Promoters-HepG2 cells were transfected with the p-6kbntcpLUC construct and treated with TNF-␣, IL-1␤, or IL-6. A time course from 4 to 24 h demonstrated maximal suppression by cytokines at 16 h (data not shown). Rat ntcp promoter-driven luciferase activity was modestly, but reproducibly, reduced in cells treated with either TNF-␣ (10 ng/ml) or IL-1␤ (1 ng/ml). Treatment with IL-1␤ reduced promoter activity by 41 Ϯ 4%, whereas TNF-␣ treatment reduced activity by 40 Ϯ 5%, respectively (mean Ϯ S.E.) (Fig. 7A). Treatment with IL-6 or LPS had no effect on promoter activity at doses up to 100 ng/ml or 100 g/ml, respectively (data not shown). No significant differences in cell number, trypan blue exclusion, LDH release, or apoptotic DNA laddering were seen between control and cytokine-treated cells (data not shown).
A series of 5Ј-deletion mutants localized the principal ntcp cytokine-response element between nt Ϫ158/ϩ47. The response to TNF-␣ or IL-1␤ on the expression of the Ϫ158/ϩ47 construct was the same as that observed for the Ϫ1237/ϩ47 construct (Fig. 7A). It should be noted that the Ϫp6kbntcpLUC construct did exhibit more suppression by TNF-␣ (ϳ40%) than the p-1237ntcpLUC construct (ϳ15%), suggesting that additional TNF-␣-response elements may lie further upstream. The p-158ntcpLUC construct contains the response elements for the HNF1 and FpB BP proteins, both of which are suppressed in rat liver after LPS administration (9). By comparison, activity of the full-length mrp2 promoter construct, p-1044mrp2LUC, was reduced by 25 Ϯ 6% after treatment with IL-1␤ (Fig. 7B). Treatment with TNF-␣ or IL-6 had no significant effect upon mrp2 promoter activity (data not shown). These data indicated that a portion of the ntcp and mrp2 transcriptional suppression observed after LPS administration in vivo could be reproduced by IL-1␤ treatment of HepG2 cells. Overall cytokine potency was assessed by treatment of albumin and IL-6 promoter constructs with TNF-␣, IL-1␤, or IL-6 ( Fig. 7B). IL-1␤ suppressed albumin promoter activity by 63 Ϯ 3%, and increased IL-6 promoter activity to 350 Ϯ 30% of basal activity. TNF-␣ suppressed albumin promoter activity by 36 Ϯ 7%. Previous investigators have demonstrated that these cytokines suppress albumin mRNA expression to a similar degree (6).
Mutation of the ntcp and mrp2 Retinoid-response Elements Eliminates Promoter Suppression by Cytokines-We then examined internal mutations of FpB to determine whether this retinoid element also mediated cytokine suppression of the ntcp promoter. The mutation that was completely unresponsive to retinoids, FM1, reduced basal activity of the promoter by 37 Ϯ 5% and was not affected by IL-1␤ (Fig. 8A). Mutation of FpB also eliminated promoter suppression by TNF-␣ (data not shown). The 1-3-bp mutations within each hexad, which had retained some (3-fold) activation by RXR␣:RAR␣, also exhibited modest suppression by IL-1␤. These mutations also re-FIG. 5. RXR␣:RAR␣ heterodimers bind the ntcp FpB and mrp2 nt ؊422/؊398 retinoid-response elements. A, EMSA was performed using in vitro synthesized RXR␣ and RAR␣ proteins and radiolabeled FpB, FM1, DR2, and mrp2422 oligonucleotides. B, EMSA was performed using rat liver nuclear extracts and radiolabeled FpB and mrp2422 oligonucleotides Ϯ anti-RXR␣, anti-RAR␣, or anti-signal transducers and activators of transcription 1 (STAT1) antibodies. Sequences of these oligonucleotides are given in Table I. Image analysis provided under "Experimental Procedures." FIG. 6. Rat liver nuclear proteins bind the mrp2 and ntcp retinoid-response elements with similar affinities. EMSA was performed using rat liver nuclear extracts and the radiolabeled FpB and mrp2 422 oligonucleotides Ϯ competition with a 100-fold excess of an unlabeled FpB, mrp2422, DR2, FM1, DR5, or AP-1 oligonucleotides. Sequences of these oligonucleotides are given in Table I. Image analysis provided under "Experimental Procedures." duced basal activity by up to 20% relative to control. Similarly, as shown in Fig. 8B, removal of the mrp2 Ϫ422/Ϫ398 retinoid element eliminated IL-1␤ suppression of the promoter and reduced mrp2 promoter basal activity by Ϸ30%. These data indicated that the ntcp FpB and mrp2 Ϫ422/Ϫ398 retinoid elements contribute to basal activity and are necessary for promoter suppression by IL-1␤.
IL-1␤ Reduces Nuclear Levels of RXR␣:RAR␣-We then examined mobility shift assays from cytokine-treated HepG2 cells to determine whether functional suppression of ntcp promoter activity was associated with alterations in DNA binding of transactivator proteins. Nuclear quantities of FpB BP were significantly reduced after 12 h of IL-1␤ treatment of HepG2 cells ( Fig. 9 shows a representative EMSA). IL-1␤ treatment resulted in an average decrease of nuclear FpB BP levels of 54% Ϯ 17% (mean Ϯ S.D., n ϭ 3). FpB BP nuclear levels were not significantly affected by TNF-␣. HNF1 nuclear quantities were not consistently affected by either cytokine (three independent experiments). The proinflammatory transcription factors AP-1 and nuclear factor B were up-regulated, indicating that the reduction in FpB BP was specific and that inflammatory signals were activated in treated cells. Other than the lack of a significant change in HNF1 levels (see "Discussion"), these data were consistent with our previous in vivo studies in LPStreated rats. DISCUSSION Administration of LPS to rats reproduces the cholestasis, which is a common clinical feature of sepsis, as well as many immunological, viral, and toxic liver diseases (7). This is pri-marily due to transcriptional suppression of hepatocyte transporters, including the bile acid uptake protein, ntcp, and the multispecific organic anion pump, mrp2 (11,12). In acute inflammatory states, such as sepsis, decreased transporter function may be protective by reducing import of toxic bile acids via ntcp and enhancing retention of antioxidants such as glutathione via decreased mrp2 export (12). However, sustained suppression of these transporters (together with the bile salt export pump, bile salt export protein) in chronic liver diseases would be expected to exacerbate liver injury, secondary to the intracellular accumulation of toxic metabolites and bile acids. These complicated and somewhat contradictory phenomena require a better understanding of the factors governing hepatocyte transport in response to inflammatory mediators. In this study, we demonstrate that activation of the rat ntcp and mrp2 gene promoters by retinoids is inhibited by cytokine-mediated signaling pathways. We propose that this mechanism contributes to suppression of ntcp and mrp2 transcription during the hepatic acute phase response.
Internal mutations of the DR2 sequence within the FpB element eliminated ntcp promoter activation by RXR␣ and RAR␣, confirming that this sequence was required to mediate retinoid activation (15). An unexpected finding was that an adjacent C/EBP element was also needed to achieve optimal retinoid transactivation of the ntcp promoter. Dependence of nuclear hormone activation upon binding of another transcription factor has been demonstrated for the phosphoenolpyruvate carboxykinase (PEPCK gene) promoter, where a neighboring C/EBP binding site is required for promoter activation by thy- FIG. 7. IL-1␤ regulates ntcp, mrp2, albumin, and IL-6 promoter activity. A, HepG2 cells were transfected with a series of ntcp promoter-driven 5Ј-deletion luciferase reporter constructs and then treated with TNF␣ (10 ng/ml) or IL-1␤ (1 ng/ml) for 16 h. B, HepG2 cells were transfected with mrp2, albumin, or IL-6 gene promoter-driven luciferase reporter constructs and then treated with TNF␣ or IL-1␤ for 16 h. Lysates were analyzed as under "Experimental Procedures." Values are means of data normalized to control activity Ϯ S.E. *, p Ͻ 0.05 versus control. roid hormone or glucocorticoids (35). Cooperative activation of the PEPCK gene promoter by C/EBP and thyroid hormone may occur via enhanced recruitment of the CREB-binding protein coactivator, thus facilitating gene transcription (35). A similar mechanism may be involved in ntcp promoter transactivation by RXR:RAR and C/EBP, which to our knowledge, would be the first instance of C/EBP enhancing the activity of an adjacent RXR:RAR element. Given the involvement of C/EBP proteins and nuclear hormone receptors in the expression of a wide variety of liver genes, other examples of synergy are likely (34,37).
FpB and mrp2422 oligonucleotides, when incubated with rat liver nuclear extracts, yielded two closely migrating bands on EMSA. Both bands comprising the shifted FpB and mrp2422 complexes were significantly ablated by the anti-RAR␣ antibody. However, only the mrp2422 complex, when coincubated with both antibodies, demonstrated a notable super-shifted species. By comparision, the shifted mrp2422 complex was also significantly reduced by the anti-RXR␣ antibody, whereas the intensity of the FpB complex was not significantly affected. The pattern of these results is quite similar to that which has been demonstrated by another group utilizing these antibodies with yeast-synthesized RXR␣ and RAR␣ proteins and DR1 or DR3 elements (38). In that study, the anti-RAR␣ antibody was more effective than the anti-RXR␣ antibody in ablating the basal shifted species while demonstrating a supershift. In addition, more significant recognition by antibody was observed for a DR3 element than a DR1 element. It may be that the conformation of the RXR␣:RAR␣ heterodimer on the mrp2 DR5 element exposes epitopes more amenable to recognition by antibody than that which occurs with the more closely spaced hexads in the FpB DR2 element. This could account for the more significant signal reduction and partial supershift observed with these antibodies for the DR5 mrp2 element versus the DR2 FpB element. Prior to our studies, these antibodies have not been tested in native nuclear extracts, so it is also quite possible that additional factors in liver nuclei modulate recognition by these antibodies or that native rat receptors are not bound as well as the in vitro expressed receptors. The persistent appearance of a doublet for the FpB and mrp2422 complexes may be due to differential phosphorylation of RXR: RAR proteins, as recently reported (39,40). Whether cytokineinduced changes in the phosphorylation state of RXR␣:RAR␣ influence transactivation of ntcp or mrp2 is now being evaluated.
The liver is the main storage site for vitamin A and retinoid metabolites. Treatment of primary rat hepatocytes with transforming growth factor ␤, a cytokine which plays a central role in the pathogenesis of chronic liver disease, reduces retinyl FIG. 9. IL-1␤ treatment reduces nuclear quantities of RXR␣: RAR␣. HepG2 cells were treated with 10 ng/ml TNF␣ or 1 ng/ml IL-1␤ for 12 h. Nuclear extracts were obtained and assayed for quantities of HNF1, FpB BP (RAR␣:RAR␣), nuclear factor B, and AP-1 by EMSA. Image analysis was as given under "Experimental Procedures" (n ϭ 3, representative EMSA shown). Sp, specific competition; NSp, nonspecific competition.

FIG. 8. The ntcp FpB and mrp2
؊422/؊398 retinoid-response elements mediate promoter suppression by IL-1␤. A, HepG2 cells were transfected with reporter plasmids containing internal mutations of the FpB-response element within the ntcp nt Ϫ158/ϩ47 promoter sequence. Cells were then treated for 16 h in serum-free media containing 1 ng/ml IL-1␤, and lysates were analyzed as under "Experimental Procedures." Values are means of data normalized to the wildtype promoter basal activity ϮS.E. *, p Ͻ 0.05 versus each respective control plasmid after IL-1␤ treatment; #, p Ͻ 0.05 versus the wild-type plasmid for basal activity without treatment. B, HepG2 cells were transfected with reporter plasmids containing the mrp2 nt Ϫ422/Ϫ15 and nt Ϫ398/Ϫ15 promoter sequences. Cells were then treated for 16 h in serum-free media containing 1 ng/ml IL-1␤, and lysates were analyzed as under "Experimental Procedures." Values are means of data normalized to p-mrp2422LUC basal activity Ϯ S.E. *, p Ͻ 0.05 versus control. esterification, which may serve as a mechanism for depletion of hepatic vitamin A stores (41). Intrahepatic vitamin A stores have also been shown to be depleted in alcoholic liver injury and cholestatic liver disease (42)(43)(44)(45). Depleted stellate cell retinol stores in cholestatic fibrosis have recently been linked to transcriptional suppression of cellular retinol-binding protein I, likely via reductions in RXR and RAR gene expression (46). If hepatic retinol stores are depleted in response to inflammatory stimuli in vivo, this may provide a contributing mechanism for reducing the expression of target genes such as ntcp and mrp2.
IL-1␤ treatment of HepG2 cells led to only a modest reduction of ntcp and mrp2 promoter activities by 45 and 25%, respectively, whereas LPS treatment in vivo reduces RNA levels for each of these genes by greater than 90% (3,11). By comparison, IL-1␤ has been recently shown to reduce mrp2 RNA levels by approximately 20% in primary rat hepatocytes, with TNF-␣ having little effect (47). Consistent with prior reports, IL-6 treatment did not affect ntcp or mrp2 promoter activity and likely has little relevance for regulation of these two genes (11). Thus, the data in the present studies are consistent with other reported acute phase models but do not fully reproduce the degree of gene suppression, which is observed in vivo. There are several potential interrelated reasons for this. Compared with rat liver, hepatoma cells express lower levels of liver-enriched transcription factors such as HNF1 and C/EBP and rely more upon ubiquitous transcription factors such as AP-1 for gene regulation (48,49). HNF1, a potent activator of the ntcp promoter, was essentially unchanged in cytokinetreated HepG2 cells, as compared with a 50% decrease in the in vivo model (9,14). There are also other compounds and cytokines, such as bile acids and IL-1␣, which accumulate in response to endotoxin treatment and may contribute to the greater suppression of nuclear transcription factor levels and gene expression observed in vivo (47,50). Two recent reports have demonstrated subapical localization of both the mrp2 and bile salt export proteins within 3 h of LPS administration to rats, which could cause an early intracellular accumulation of bile acids and other organic compounds, which might then augment cytokine-induced transcriptional suppression (3,12). The cholesterol 7-␣ hydroxylase promoter is suppressed by bile acids, likely via inhibition of activating RXR heterodimers by the recently described bile acid receptor, farnesoid x receptor (25,51). Whether or not the ntcp and mrp2 promoters are transcriptionally down-regulated by bile acids is currently under study in our laboratory. All together, the greater downregulation of ntcp expression in vivo versus in vitro is likely due to a combination of features including a different and more robust inflammatory milieu, additional suppressive compounds such as bile acids, and decreased retinol stores.
In this study, we describe regulation of the rat ntcp and mrp2 promoters by RXR␣:RAR␣. C/EBP␣ is shown to potentiate ntcp transactivation by retinoids. A mechanism of ntcp and mrp2 acute phase transcriptional suppression via IL-1␤-induced reduction in nuclear RXR␣:RAR␣ heterodimers is described, whereas C/EBP␤ is shown to contribute to ntcp suppression. These studies now indicate that the expression of hepatic transporters should be included in the expanding array of diverse hepatic metabolic processes regulated by RXR heterodimers (17)(18)(19)(20)(21)(22)(23)(24)(25).