Transcriptional Regulation of the Human Sterol 12α-Hydroxylase Gene (CYP8B1) ROLES OF HEPATOCYTE NUCLEAR FACTOR 4α IN MEDIATING BILE ACID REPRESSION

Abstract Sterol 12α-hydroxylase catalyzes the synthesis of cholic acid and controls the ratio of cholic acid over chenodeoxycholic acid in the bile. Transcription of CYP8B1is inhibited by bile acids, cholesterol, and insulin. To study the mechanism of CYP8B1 transcription by bile acids, we have cloned and determined 3389 base pairs of the 5′-upstream nucleotide sequences of the human CYP8B1. Deletion analysis ofCYP8B1/luciferase reporter activity in HepG2 cells revealed that the sequences from −57 to +300 were important for basal and liver-specific promoter activities. Hepatocyte nuclear factor 4α (HNF4α) strongly activated human CYP8B1 promoter activities, whereas cholesterol 7α-hydroxylase promoter factor (CPF), an NR5A2 family of nuclear receptors, had much less effect. Electrophoretic mobility shift assay identified an overlapping HNF4α- and CPF-binding site in the +198/+227 region. The humanCYP8B1 promoter activities were strongly repressed by bile acids, and the bile acid response element was localized between +137 and +220. Site-directed mutagenesis of the HNF4α-binding site markedly reduced promoter activity and its response to bile acid repression. On the other hand, mutation of the CPF-binding site had little effect on promoter activity and bile acid inhibition. A negative nuclear receptor, small heterodimer partner markedly inhibited transactivation of CYP8B1 by HNF4α. Mammalian two-hybrid assay confirmed that HNF4α interacted with small heterodimer partner. Furthermore, bile acids and farnesoid X receptor reduced the expression of nuclear HNF4α in HepG2 cells and rat livers and its binding to DNA. Bile acids and farnesoid X receptor also inhibited mouseHNF4α gene transcription. In summary, our data revealed the critical roles HNF4α play on CYP8B1transcription and its repression by bile acids. Bile acids repress human CYP8B1 transcription by reducing the transactivation activity of HNF4α through interaction of HNF4α with SHP and reduction of HNF4α expression in the liver.

Sterol 12␣-hydroxylase catalyzes the synthesis of cholic acid and controls the ratio of cholic acid over chenodeoxycholic acid in the bile. Transcription of CYP8B1 is inhibited by bile acids, cholesterol, and insulin. To study the mechanism of CYP8B1 transcription by bile acids, we have cloned and determined 3389 base pairs of the 5-upstream nucleotide sequences of the human CYP8B1. Deletion analysis of CYP8B1/luciferase reporter activity in HepG2 cells revealed that the sequences from ؊57 to ؉300 were important for basal and liver-specific promoter activities. Hepatocyte nuclear factor 4␣ (HNF4␣) strongly activated human CYP8B1 promoter activities, whereas cholesterol 7␣-hydroxylase promoter factor (CPF), an NR5A2 family of nuclear receptors, had much less effect. Electrophoretic mobility shift assay identified an overlapping HNF4␣-and CPFbinding site in the ؉198/؉227 region. The human CYP8B1 promoter activities were strongly repressed by bile acids, and the bile acid response element was localized between ؉137 and ؉220. Site-directed mutagenesis of the HNF4␣-binding site markedly reduced promoter activity and its response to bile acid repression. On the other hand, mutation of the CPF-binding site had little effect on promoter activity and bile acid inhibition. A negative nuclear receptor, small heterodimer partner markedly inhibited transactivation of CYP8B1 by HNF4␣. Mammalian two-hybrid assay confirmed that HNF4␣ interacted with small heterodimer partner. Furthermore, bile acids and farnesoid X receptor reduced the expression of nuclear HNF4␣ in HepG2 cells and rat livers and its binding to DNA. Bile acids and farnesoid X receptor also inhibited mouse HNF4␣ gene transcription. In summary, our data revealed the critical roles HNF4␣ play on CYP8B1 transcription and its repression by bile acids. Bile acids repress human CYP8B1 transcription by reducing the transactivation activity of HNF4␣ through interaction of HNF4␣ with SHP and reduction of HNF4␣ expression in the liver.
High serum cholesterol contributes to atherosclerosis and cardiovascular diseases (1). The conversion of cholesterol to bile acids is the most significant pathway for cholesterol disposal and occurs exclusively in the liver (2)(3)(4). The neutral pathway of bile acid synthesis is subjected to bile acid feedback inhibition of the rate-limiting step catalyzed by cholesterol 7␣-hydroxylase (CYP7A1) 1 (5). CYP8B1 catalyzes the synthesis of cholic acid and controls the ratio of cholic acid (CA) over chenodeoxycholic acid (CDCA) that determines the hydrophobicity of the bile acid pool (6). CYP8B1 was purified in 1992 (7) from rabbit livers. Recently, the cDNA and the gene encoding CYP8B1 were cloned in the rabbit, rat, mouse, and human (8 -10). Interestingly, the CYP8B1 has no intron.
Cholesterol feeding or thyroid hormone repress CYP8B1 expression (8,9,11), in contrast to their stimulatory effect on CYP7A1. In streptozotocin-induced diabetic rats, the CYP8B1 activity and mRNA levels were elevated, which could be suppressed by insulin administration (12). An increase in CYP8B1 transcription may explain the increased synthesis of cholic acid in diabetes.
Recently, two overlapping CPF-binding sites have been identified in the rat CYP8B1 promoter (31). Mutation of the CPFbinding sites abolished CYP8B1 transcriptional activity such that the bile acid inhibition of the CYP8B1 promoter activity could not be determined. We analyzed nucleotide sequences of the putative BAREs in the CYP7A1 and CYP8B1 promoters of different species and unveiled a general feature that the BAREs contain the overlapping binding sites for CPF and HNF4␣ despite low sequence identity between the BAREs of these two genes (4,31). We hypothesize a general mechanism that SHP interacts either with HNF4␣ or CPF and inhibits the genes regulated by bile acids. HNF4␣ (NR2A1) is an orphan nuclear receptor that binds to the direct repeat with one base spacing (DR1) motif as a homodimer and regulates the liverspecific expression of many genes in lipoproteins and glucose metabolism. HNF4␣ has constitutive activity and is able to transactivate genes without ligand binding. HNF4 has been shown to activate CYP7A1 transcription (32).
The goal of this research was to investigate the mechanism of transcriptional regulation of the human CYP8B1 promoter by bile acids. In the present study, we characterized the promoter of the human CYP8B1. Site-directed mutagenesis, reporter gene assay, and EMSA were used to study effects of CPF, HNF4␣, and SHP. Since CPF was used in this study, the term CPF will be used in this work unless specified otherwise. It should be mentioned that FTF is the recommended name by Genomic Data Base Nomenclature Committee (accession number 9837397) (26,29). We demonstrated that HNF4␣ played a major role in regulating human CYP8B1 transcription and mediating bile acid repression by interacting with SHP. In addition, bile acids also inhibited CYP8B1 transcription by inhibiting HNF4␣ transcription.

EXPERIMENTAL PROCEDURES
Cloning of the Human CYP8B1 Gene-Based on the available human CYP8B1 sequence, the sequence from nucleotide Ϫ514 to ϩ300 relative to the transcription initiation site was amplified by polymerase chain reaction (PCR) using human liver genomic DNA as a template. One genomic clone, H8B5, was isolated from the FIX II genomic library (Stratagene, La Jolla, CA) using the end-labeled genomic fragment as a hybridization probe. Southern blot and PCR analyses confirmed that this clone (15 kb) contained the entire coding sequence of the human CYP8B1 and about 8 kb of the 5Ј-flanking sequences. A PstI/SacI restriction fragment containing 3.2 kb of the 5Ј-flanking sequences was subcloned into the pBluescript II SKϩ vector (Promega, Madison, WI) and designated as pBSK/h8B1/3.5K. Nucleotide sequencing revealed 3389 base pairs of the 5Ј-upstream sequence (GenBank TM accession number AF226627).
Construction of Human CYP8B1/Luc Reporters-The Ϫ514/ϩ300 fragment obtained by PCR was cloned into the SacI and SmaI sites of the luciferase reporter, pGL3 basic vector (Promega). The phCYP8B1Ϫ514/ϩ300Luc obtained was then digested with KpnI and SpeI to release a fragment covering the sequence from Ϫ514 to ϩ172 and replaced it with a KpnI and SpeI fragment (Ϫ3064 to ϩ172) released from pBSK/h8B1/3.5K. The resulting constructs were designated as phCYP8B1Ϫ3064/ϩ300Luc. The phCYP8B1Ϫ514/ϩ300Luc was digested with HinfI at Ϫ111, respectively. The sticky ends produced by HinfI were blunted with the Klenow polymerase. The linearized plasmid was subjected to XhoI digestion and subsequently cloned into the pGL3 vector digested with SmaI and XhoI. The phCYP8B1Ϫ514/ ϩ300Luc was also used as a template for construction of deletion mutants by PCR to generate phCYP8B1Ϫ434/ϩ300Luc, phCYP8B1Ϫ164/ϩ300Luc, and phCYP8B1Ϫ57/ϩ300Luc. All of these constructs had MluI site built in at the 5Ј-end and XhoI site at the 3Ј-end and cloned into MluI-and XhoI-digested pGL3 vector. To construct the 3Ј-deletion mutants, the phCYP8B1Ϫ514/ϩ300Luc was digested with BglII and SpeI or StuI to release fragments from ϩ180 to ϩ300 and ϩ248 to ϩ300, respectively. The sticky ends were then blunted with the Klenow polymerase, and the linearized plasmids were religated to generate phCYP8B1Ϫ514/ϩ180Luc and phCYP8B1Ϫ514/ ϩ248Luc. The phCYP8B1Ϫ514/ϩ76Luc was constructed by digestion phCYP8B1Ϫ514/ϩ300Luc with HindIII to release the fragment from ϩ76 to ϩ300 and religate the linearized plasmid. Other 3Ј-deletion mutants were obtained by PCR to construct phCYP8B1Ϫ514/ϩ220Luc, phCYP8B1Ϫ514/ϩ200Luc, and phCYP8B1Ϫ514/ϩ137Luc. These three constructs had SacI at the 5Ј-end and SmaI sites at the 3Ј-end.
Mutations were introduced into reporter constructs by PCR-based site-directed mutagenesis using QuikChange Site-directed Mutagenesis Kit (Stratagene). Two complementary oligonucleotide sets were designed as PCR primers; primer M4 (ϩ194 to ϩ234) was used to mutate the HNF4␣ site and introduce a CPF site in the reverse strand 209 tttaccttga 218 ; primer M5 (ϩ200 to ϩ245) was used to introduce a consensus 3Ј-HRE 215 aGgtCA 220 . Reaction mixtures were set up according to the manufacturer's instruction using 50 ng of template DNA and 125 ng of primers. Cycling parameter: denaturing at 95°C for 30 s, followed by 18 cycles at 95°C for 30 s, 55°C for 1 min, and 68°C for 12 min. The reaction was subjected to DpnI digestion for 2 h. The plasmids were then transformed into XL1-Blue supercompetent cells. Sequences of all constructs were confirmed by DNA sequencing.
Mammalian Two-hybrid Assay-Gal4/HNF4␣ fusion construct (provided by M. Crestani (33)) was prepared by inserting the full-length HNF4␣ coding region (amino acid residue 1-455) into plasmid pcDNA3X(Ϫ) (Invitrogen) at the BamHI site. VP16-SHP (provided by D. Mangelsdorf (24)) contained a full-length mouse SHP coding region ligated into pCMX-VP16. CheckMate TM mammalian two-hybrid assay kit was obtained from Promega. The pBIND vector contains the yeast Gal4 DNA binding domain, and the pACT vector contains the herpes simplex virus VP16 activation domain. Gal4/Id and VP16/MyoD provided in the kit were used as a positive control. The reporter plasmid pG5luc contains five copies of Gal4-binding sites fused upstream of the firefly luciferase gene (luc).
Transient Transfection Assay-Confluent cultures of HepG2 cells grown in 24-well tissue culture plates were transfected with plasmids by the calcium phosphate coprecipitation method. The reporter construct, receptor expression plasmid, and pCMV ␤-galactosidase plasmid (CLONTECH, Palo Alto, CA, one-tenth of reporter plasmid, as internal control for transfection efficiency) were transfected in each well. The pcDNA3 vector was added to normalize the amounts of DNA transfected in each assay. Cells were overlaid with serum-free media containing indicated concentrations of bile acids. Cells were lysed 40 h after transfection. Each data point is the average of triplicate assays. Each experiment was repeated three times. Luciferase activity was assayed using Luciferase Assay System (Promega) and luminescence was determined using a Lumat LB 9501 luminometer (Berthold Sys-tems, Inc., Pittsburgh, PA). Luciferase activities were normalized for transfection efficiencies by dividing the relative light units by ␤-galactosidase activity expressed from cotransfected pCMV plasmid. A human CYP7A1 reporter gene phCYP7A1Ϫ372/ϩ25Luc was constructed previously. Mouse HNF4␣/Luc reporter (pDGT43) containing 744 base pairs 5Ј-upstream sequence was provided by Dr. T. Leff (Werner Lambert/Parke-Davis, Ann Arbor, MI).
Nuclear Extract Preparation-Confluent HepG2 cells were lysed by trypsin and washed twice with cold phosphate-buffered saline (PBS). Cells were then resuspended in hypotonic buffer and swelled for 10 min on ice. The cells were broken using a Dounce homogenizer with a tightly fit pestle. One-tenth volume of 75% sucrose buffer was added and homogenized. The homogenate was spun for 30 s at 16,000 ϫ g at 4°C. Then the viscous nuclear pellet was lysed in nuclear resuspension buffer containing 0.4 M ammonium sulfate and centrifuged at 2°C for 90 min at 150,000 ϫ g to pellet nuclear debris and chromatin. Solid ammonium sulfate was added to precipitate the nuclear protein from the supernatant. The pellet was dissolved in nuclear dialysis buffer and dialyzed overnight at 4°C. Protein concentration was quantitated using Coomassie Plus Protein Assay Reagent Kit (Pierce), and the nuclear extracts were stored at Ϫ70°C in aliquots. Nuclear extracts also were isolated from the livers of rats treated with a diet supplemented with CA (1%), CDCA (1%), deoxycholic acid (DCA) (0.25%), ursodeoxycholic acid (UDCA) (1%), cholestyramine (5%), or cholesterol (1%) for 2 weeks. Animals were housed in a room with reversed dark/light cycle (3 a.m. to 3 p.m. dark, 3 p.m. to 3 a.m. light) and sacrificed at 9 am.
Electrophoretic Mobility Shift Assay (EMSA)-Double-stranded synthetic probes for EMSA were prepared by heating equal molar amounts of complementary oligonucleotides to 95°C in 2ϫ SSC (0.5 M NaCl, 15 mM sodium citrate, pH 7.0) and cooled to room temperature. The resulting double-stranded fragments were labeled by filling in the overhang incorporated in the synthetic oligonucleotides with [␣-32 P]dCTP (3000 Ci/mol) with the Klenow fragment of DNA polymerase I. Oligonucleotides filled-in with non-labeled dNTPs were used as cold competitors. Labeled fragments were purified through two G-50 spin columns. Binding reactions were initiated with the addition of 3 g of nuclear extracts to 100,000 cpm of labeled oligonucleotide probe dissolved in 20 l of the buffer containing 12 mM HEPES, pH 7.9, 50 mM KCI, 1 mM EDTA, 1 mM dithiothreitol, and 15% glycerol, and 2 g of poly(dI-dC). Samples were incubated for 20 min at room temperature. Four percent polyacrylamide gels were prepared and pre-run for 30 min at 200 V. Electrophoresis was performed at room temperature at constant 200 V for 1.5-2 h. The gel was dried and autoradiographed using Phosphor-Imager 445Si (Molecular Dynamics, Sunnyvale, CA). The images were analyzed using IP Lab Gel software (Signal Analytics Corp., Vienna, VA). Antibody supershift was carried out by adding the antibody (1-2 l) to the nuclear extract and incubated for 15 min before mixing with labeled probe. Oligonucleotides used for competition assays were HNF4 from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), SP-1 from Promega (Madison, WI), and CPF synthesized according to Gilbert et al. (30).
Immunoblot-To measure HNF4␣ protein in the nuclei, 3 g of nuclear extracts were run on 10% SDS-polyacrylamide gel electrophoresis and transferred electrophoretically to a nitrocellular membrane (Hybond ECL, Amersham Pharmacia Biotech). Membranes were blocked with 5% (w/v) non-fat milk in Tween PBS (T-PBS) overnight at 4°C and incubated with the antibody against HNF4␣ (Santa Cruz Biotechnology) at a dilution of 1:5000 in T-PBS for 2 h at 4°C. Membranes were washed three times with T-PBS and incubated with a secondary antibody (horseradish peroxidase-conjugated anti-goat IgG) at a dilution of 1:3000 at 4°C for 2 h. Immunodetection was carried out using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech). Membranes were imaged using a Kodak Imaging Station 440.

RESULTS
Functional Analysis of the Human CYP8B1 Gene Promoter-To determine the contribution of the 5Ј-flanking sequences to human CYP8B1 promoter activity, we performed transient transfection assays of CYP8B1/luciferase reporters in HepG2 cells and CHO cells. Fig. 1A shows that sequential deletion of the nucleotide sequence from the 5Ј-direction did not alter reporter activity much in HepG2 cells. In contrast, deletion of sequences from the 3Ј-direction markedly reduced reporter activity in HepG2 cells (Fig. 1B). The sequence from ϩ248 to ϩ300 apparently was very important for promoter activity, because deletion of this region reduced the activity by 75%, relative to the promoter activity of phCYP8B1Ϫ514/ϩ300Luc. Deletion of the sequence from ϩ180 to ϩ248 further reduced the promoter activities to 10%. Transfection assays of these deletion mutants were also done in CHO cells. Deletion of the sequence from 5Ј-end did not change the promoter activities as observed in the HepG2 cells (data not shown). The loss of promoter activity was much less in CHO cells than in HepG2 cells when the regions between ϩ180 and ϩ300 were deleted (Fig. 1B). It appeared that the sequence between ϩ180 and ϩ300 was important for the liver-specific transcription of the human CYP8B1. Fig. 2 shows the nucleotide sequence and putative transcription factor-binding sites of the proximal promoter of the human CYP8B1. The transcription start site is located at 325 base pairs upstream of the translation start codon (9). The TATA The 5Ј-and 3Ј-sequential deletions of human CYP8B1 upstream sequence were cloned into pGL3 basic reporter vector. Chimeric reporter plasmids (1 g) were transfected into confluent HepG2 or CHO cells. Schematic representation of the deletion constructs with numbering at the left indicates the nucleotide covered relative to the transcription start site (TS) which is indicated by a dotted line. The luciferase reporter activities of deletion constructs were expressed related to the reporter activities of plasmid hCYP8BϪ57/ϩ300LUC and hCYP8B-514/ ϩ300LUC, which were set as 1 in the 5Ј-(A) and 3Ј-deletion (B) analysis, respectively. The error bars represent the standard deviation from the mean of triplicate assays of a representative experiment. Experiments were repeated three times.
box is located at -56/-51. Sequences downstream of the transcription start site contain consensus binding sites for both ubiquitous and liver-specific transcription factors. The region from ϩ248 to ϩ300, which is critical for basal promoter activity, contains several NF-1-binding sites. The region upstream to ϩ180 contains a cluster of putative binding sequences for liver-specific factors, HNF3, CEBP, and DBP. HNF3-and CEBP-binding sites are similar to the insulin response sequence, T(G/A)TTTTG, found in the phosphoenolpyruvate carboxykinase and insulin-like growth factor-binding protein-1 that has been implicated in mediating the repression by insulin (34). The DBP plays a role in diurnal rhythm of the CYP7A1 and other clock genes (35). Interestingly, a sterol response element-3 (SRE-3)-like palindromic sequence (CACTAGTG), a SRE/Sp1 (TGCGGCCAC), and an E box (CAGGTG) are located in this region. These sequences are potential SREBP-binding sites. SREBPs are helix-loop-helix-leucine zipper transcription factors that regulate the genes in cholesterol and fatty acid synthesis (36,37). The sequence from ϩ208 to ϩ220 contains overlapping HNF4␣ and CPF-binding sites. In addition, two E boxes, an HRE half-site, AGGTCA preceded by an A/T-rich sequence (a binding site for monomeric nuclear receptor Rev-erb␣), and an SRE are located further upstream.
Mapping of the HNF4␣-and CPF-binding Sites-Sequences from ϩ198 to ϩ227 of the human CYP8B1 contain a CPFbinding site (GCAAGGTCC, Fig. 3A) which is similar to that identified in the rat CYP8B1 (31). The rat CYP8B1 contains two overlapping CPF-binding sites and a DR1 shown to be a weak PPAR␣-binding site (38). We also noticed a DR1 motif (AGGGCAaGGTCCA) overlapping with a CPF-binding site in the human CYP8B1. This feature is similar to the bile acid response element II we identified previously in the rat and human CYP7A1 (Fig. 3A) (4, 23). HNF4␣ and CPF have been shown to bind the BARE-II (27,32). To identify transcription factors bound to the sequence from ϩ198 to ϩ227 of human CYP8B1, we performed EMSA using HepG2 cell nuclear extracts (Fig. 3B). Four DNA-protein complexes were obtained. The strongest band was further identified as an HNF4␣-DNA complex by competition assay using unlabeled HNF4␣ consensus probe, in vitro synthesized HNF4␣ protein, and antibody supershift assays. A faster moving band was identified as a CPF-DNA complex using in vitro synthesized CPF and competition assay using unlabeled CPF oligonucleotides. These two complexes were competed out by 100-fold excess of unlabeled CPF or HNF4␣ oligonucleotide, respectively, but an unrelated SP-1 oligonucleotide could not compete out the complexes. Fig.  3C shows that in vitro synthesized CPF binds to this probe, but RAR␣/RXR␣, PPAR␣/RXR␣, and RXR␣ homodimer were unable to bind to this sequence. These data indicate that HNF4␣ and CPF specifically bind to the ϩ198/ϩ227 region. HNF4␣and CPF-binding sites were further studied by mutagenesis. We designed mutant oligonucleotides to alter nucleotide sequences located upstream of the putative HNF4␣ site (M1, M2, and M3), the HNF4␣ site (M3, M4, M6, and M7), and the putative CPF-binding site (M5 and M7) (Fig. 4A). Mutation in M4 altered the HNF4␣ site but created a new CPF site in the reverse strand. M5 was designed to alter the 3Ј-HRE (GGTCCA) to a consensus HRE half-site, AGGTCA, and mutated the CPF site. M6 was designed to mutate the DR1 by deleting the spacing between two HRE and mutate the downstream sequences so that the DR1 motif was altered to a DR0. M7 was designed to mutate both HNF4␣ and CPF sites.
Mutant probes were then labeled and used for EMSA to study the binding specificity with in vitro synthesized HNF4␣ and CPF (Fig. 4, B and C). Mutations of the sequences upstream of the HNF4␣/CPF site (M1, M2, and M3) reduced both HNF4␣ and CPF binding. Mutations of the HNF4␣-binding site (M4 and M7) totally abolished the HNF4␣ bindings. M5 showed stronger HNF4 binding because the 3Ј-HRE was mutated to a consensus HRE. Mutations of the core (AAGG) of the CPFbinding site (M5 and M7) abolished CPF binding. M4, which had an HNF4␣ site mutated and a CPF site created in the reverse strand, bound CPF stronger than the wild type probe. M6, which had the HNF4␣-binding site altered but not CPF site, bound CPF. Fig. 4D shows EMSA using HepG2 nuclear extracts. Essentially the same results were obtained as using in vitro synthesized HNF4␣ and CPF. These data confirmed the HNF4␣-and CPF-binding sites in this region. However, sequences upstream of this overlapping site are also involved in the binding of HNF4␣ and CPF, because mutations of these sequences reduced their binding affinity.
Transcriptional Regulation of the Human CYP8B1 by HNF4␣ and CPF-We then studied the regulation of human CYP8B1 transcription by HNF4␣ and CPF using transient transfection assay of CYP8B1/Luc reporter genes in HepG2 cells. HNF4␣ dose-dependently stimulated the reporter activity by 20-fold (Fig. 5A). Under the same assay condition, CPF had much less effect, up to 2-fold of stimulation (Fig. 5B). We did reporter assays in HEK293 cells (Fig. 6A). As in HepG2 cells, CPF (0.5 g) did not affect CYP8B1/Luc reporter activity, whereas HNF4␣ (0.5 g) strongly stimulated reporter activity by 5-fold. Cotransfection with both HNF4␣ and CPF did not potentiate reporter activity stimulated by HNF4␣. We did the same assay in 293 cells with a human CYP7A1/Luc reporter as a comparison. CPF or HNF4␣ stimulated CYP7A1/Luc reporter activity by 2-fold (Fig. 6B). Cotransfection with both CPF and HNF4␣ stimulated human CYP7A1 promoter by 4-fold. Thus these two liver-specific nuclear receptors regulate human CYP7A1 and CYP8B1 differently; both of them synergistically regulates human CYP7A1, but only HNF4␣ regulates CYP8B1.
We then introduced mutations into human CYP8B1/Luc reporter based on EMSA results. As shown in Fig. 7, mutation of the HNF4␣ site created a CPF site in the reverse strand (M4) which markedly reduced reporter activity and abolished the stimulatory effect of HNF4␣. Mutation of the 3Ј-HRE of the HNF4␣ site to a consensus HRE mutated the CPF core sequence (M5) but did not alter basal reporter activity and maintained HNF4␣ stimulation. These results suggest that the HNF4␣-binding site is critical for basal promoter activity, whereas CPF does not have much effect on human CYP8B1 transcription regardless of its ability to bind to the gene.

Suppression of Human CYP8B1 Transcription by Bile Acids
Was Mediated through HNF4␣-We then studied the effects of different bile acids and FXR on the reporter activity of the human CYP8B1/Luc reporter in transfection assay in HepG2   FIG. 3. Bile acid response elements of rat and human CYP7A1 and CYP8B1 and electrophoretic mobility shift assays of HNF4␣ and CPF binding to a probe containing nucleotide sequences of the human CYP8B1 from ؉198 to ؉227. A, alignment of rat and human CYP8B1 sequences from ϩ198 to ϩ227. Putative HNF4␣ and CPF-binding sites are indicated. Corresponding sequences of the rat and human CYP7A1 are shown for comparison. B, EMSA were performed with ␣-32 P-labeled double-stranded probe, H8Bϩ198/ϩ227. Nuclear extracts isolated from HepG2 cells and in vitro synthesized HNF4␣ and CPF were used for EMSA. Reaction mixtures contained 3 g of protein of nuclear extracts or in vitro synthesized HNF4␣ (5 l) or CPF (5 l). The unlabeled oligonucleotides were added in 100ϫ excess for competition assays. Anti-HNF4␣ antibody (2 l) was added into the reaction 30 min prior to the addition of the probe for supershift assay. C, EMSA of CPF and HNF4␣ binding to human CYP8B1 ϩ198/ϩ227 probe. TNT lysates programmed with CPF, HNF4␣, PPAR␣, and RXR␣ were used for EMSA. HNF4␣ oligonucleotide probe, CTCAGCTTGTACTTTGGTACAACA; CPF oligonucleotide probe, TAGGCCTCAAGGTCGGTCG; SP1 oligonucleotide probe, ATTCGATCGGGGCGGGGCGAG.  (Fig. 8A). Addition of DCA or CDCA (25 M) repressed the reporter activity by 50 -70%, respectively. Cotransfection with human liver Na ϩ taurocholate cotransport peptide (NTCP) was required for all taurine-conjugated bile acids to repress reporter activity in HepG2 cells, except taurolithocholic acid. Taurolithocholic acid is highly hydrophobic and may be toxic to HepG2 cells even at low concentrations. However, transfection with FXR/RXR␣ somewhat stimulated basal activity but did not enhance the inhibitory effect of bile acids on the human CYP8B1. We found previously that FXR enhanced the inhibitory effect of bile acids on CYP7A1 transcription. We interpreted that factors induced by bile acids in HepG2 cells might be sufficient for bile acid inhibition of the CYP8B1, thus the CYP8B1 may be more sensitive to bile acid inhibition than CYP7A1. This is consistent with the reported potency of bile acid inhibition as CYP8B1 Ͼ CYP7A1 Ͼ CYP27A1 (11). It is also possible that somewhat different mechanisms may be involved in bile acid repression of these genes.
We then used deletion mutants of phCYP8B1Ϫ514/ϩ300Luc to map the region conferring bile acid repression. Deletion from Ϫ514 to Ϫ57 did not affect the bile acid repression (Fig. 8B). When we deleted the sequences from the 3Ј-direction, between FIG. 5. Effects of HNF4␣ and CPF on human CYP8B1 reporter activity. A, dose-responses of HNF4␣ on phCYP8B1Ϫ514/ϩ300/Luc reporter activity. Reporter (1 g) was cotransfected with indicated amounts of HNF4␣ expression plasmid into HepG2 cells. B, dose responses of CPF on phCYP8B1Ϫ514/ϩ300/Luc reporter activity. Reporter (1 g) was cotransfected with indicated amounts of CPF expression plasmids into HepG2 cells. Reporter activities were determined as described under "Experimental Procedures" and as in Fig. 1.   FIG. 6. Effects of HNF4␣ and/or CPF on human CYP8B1 and CYP7A1 transcription. A, effect of HNF4␣ and CPF on human CYP8B1-514/ϩ300/Luc reporter activity. phCYP8B1Ϫ514/ϩ300LUC reporter (1 g) was cotransfected with 0.5 g of HNF4␣ and/or CPF into HEK293 cells. B, effect of HNF4␣ and CPF on human CYP7A1/Luc reporter activity. phCYP7A1Ϫ372/ϩ25/Luc (1 g) was transfected with HNF4␣ and/or CPF expression plasmid (0.5 g) into HEK293 cells. Control was transfected with pcDNA3 vector (0.5 g). The empty plasmid was added to compensate for the total amount of DNA transfected in each assay. The reporter activities were expressed as the relative luciferase activities normalized by ␤-galactosidase activities.  ϩ137 and ϩ220, bile acid responses as well as promoter activities were greatly reduced (Fig. 8C). This region contains HNF4␣-and CPF-binding sites. We subsequently studied the bile acid effect on mutant CYP8B1/Luc reporters in HepG2 cells. As shown in Fig. 9, the promoter activities of wild type and mutant construct M5, which had the CPF site mutated but the HNF4␣ site maintained, were suppressed by CDCA in a dose-dependent manner. When the HNF4␣ site was mutated but a CPF site was created in the reverse strand (M4), CDCA did not affect the reporter activities. These results demonstrated that HNF4␣ binding is necessary and sufficient for mediating bile acid repression of human CYP8B1 transcription, and CPF might not be involved in mediating bile acid repression of the gene.
SHP Interacted with HNF4␣ and Repressed Human CYP8B1 Transcription-Bile acids repress CYP7A1 transcription by induction of SHP which interacts with CPF and represses CYP7A1 transcription (24,25). Since HNF4␣ was found to play a major role and CPF had much lesser effect on CYP8B1 transcription, we studied the effect of SHP on HNF4␣ or CPF regulation of CYP8B1 transcription. Fig. 10A shows that cotransfection of HNF4␣ (1:5 of reporter) strongly stimulated human CYP8B1/Luc reporter activity by 8-fold, and transfection of SHP alone did not have any effect on CYP8B1 reporter activity. When cotransfected with both HNF4␣ and SHP, reporter activity was the same as the control which was transfected with pcDNA3 empty vector. Thus SHP repressed the reporter activity stimulated by HNF4␣. Fig. 10B shows that HNF4␣ dose-dependently stimulated human CYP8B1/Luc reporter activity, and cotransfection with increasing amounts of SHP strongly repressed reporter activity in a dose-dependent manner. CPF stimulated CYP8B1/Luc reporter activity by up to 2-fold when transfected with 2-fold excess of the receptor plasmid over reporter plasmid (Fig. 10C). SHP repressed the reporter activity stimulated by CPF by only about 50%, much less than its marked inhibitory effect on CYP8B/Luc reporter activity stimulated by HNF4␣.
We then employed a mammalian two-hybrid assay system to study the interaction between HNF4␣ and SHP in HepG2 cells (Fig. 10D). As a positive control of two-hybrid assays, cotransfection with Gal4/Id and VP16/MyoD hybrid constructs resulted in a strong stimulation of luciferase reporter (pG5Luc) activity. Cotransfection with both VP16/SHP and Gal4/HNF4␣ hybrid plasmids resulted in stimulation of reporter activity by 4-fold over cotransfection of Gal4/HNF4␣ with either VP16 empty vector (PACT) or VP16/MyoD. We also did two-hybrid assays with Gal4/SHP and VP16/HNF4␣ (data not shown). A strong stimulation of reporter activity by 26-fold was obtained. Thus both Gal4/HNF4␣ and VP16/SHP were required for stimulation of reporter activity in HepG2 cells indicating that HNF4␣ and SHP did interact as demonstrated by mammalian two-hybrid assays.
Bile Acids Inhibited HNF4␣ Expression-It is possible that bile acid repression of human CYP8B1 transcription may be also due to inhibition of HNF4␣ binding to CYP8B1 or inhibition of HNF4␣ expression in hepatocytes. We first examined the effect of CDCA on HNF4␣ and CPF binding to the ϩ198 to ϩ227 probe (Fig. 11A). HepG2 cells were treated with CDCA (25 M) with or without cotransfection with RXR␣/FXR. Nuclear extracts were isolated from HepG2 cells and used for EMSA. When nuclear extracts of HepG2 cells treated with CDCA (25 M) were used for EMSA, the HNF4␣-DNA complex was reduced, and the CPF-DNA complex was abolished. Interestingly, nuclear extracts of HepG2 cells cotransfected with FXR/RXR␣ generated the same gel shift pattern as using nuclear extracts isolated from untreated cells. However, when nuclear extracts isolated from HepG2 cells transfected with FXR/RXR␣ and treated with CDCA were used, the band shifts were almost completely abolished. When the HNF4␣ consensus sequence was used as a probe for EMSA using the same nuclear extract preparations (Fig. 11B), similar results were obtained. These results suggested that FXR and CDCA reduced HNF4␣ binding to DNA.
We wanted to determine whether the decreased expression level of HNF4␣ was responsible for the decreased HNF4␣ binding to DNA. We examined the nuclear HNF4␣ protein level in HepG2 cells by immunoblot assay (Fig. 12A). The HNF4␣ protein level in HepG2 cells was dramatically decreased by the CDCA treatment. Cotransfection of FXR/RXR␣ and treatment with CDCA completely eliminated HNF4␣ protein expression. We also treated rats with a diet supplemented with CDCA (1%), DCA (0.25%), CA (1%), UDCA (1%), cholestyramine (5%), or cholesterol (1%) for 2 weeks. Nuclear extracts were isolated from rat livers for EMSA. Fig. 12B shows that CDCA, DCA, and CA treatments markedly reduced the nuclear HNF4␣ protein levels. Cholestyramine and cholesterol did not alter the HNF4␣ protein levels. We then studied the effect of CDCA on mouse HNF4␣/Luc reporter activity in transfection assay in HepG2 cells (Fig. 12C). CDCA (25 M) repressed HNF4␣ reporter activity by 60%. Cotransfection with FXR/RXR␣ reduced reporter activity by 20%, and addition of CDCA further repressed reporter activity by 80%. These data revealed that bile acid repression of HNF4␣ transcription might also contribute to the inhibition of CYP8B1 transcription by bile acids, in addition to the repression by SHP/HNF4␣ interaction. DISCUSSION It has been revealed recently that FXR is a highly specific bile acid receptor that is activated by hydrophobic bile acids at physiological concentrations to directly stimulate the transcription of the genes in bile acid transport, absorption, and reverse cholesterol transport but indirectly inhibit CYP7A1 transcription (23)(24)(25). FXR may play a pivotal role in cholesterol metabolism by regulating the reverse cholesterol transport from the peripheral tissues to the liver for its conversion to bile acids. This mechanism may regulate the bile acid pool size in the liver, thus protecting liver cells from cytotoxic effect of bile acids. It appears that the bile acid-activated FXR induces a negative nuclear receptor SHP which interacts with CPF and down-regulates CYP7A1 transcription (24,25,39). It was sug- gested that the same mechanism may also regulate CYP8B1 transcription and coordinately regulates bile acid biosynthesis (24,25). However, these investigators did not provide any experimental evidence that SHP interacted with CPF and inhibited CYP8B1 transcription. We demonstrated for the first time in this study that HNF4␣ was necessary and sufficient in mediating bile acid repression of human CYP8B1 transcription. The bile acid response elements of CYP7A1 identified previously and of CYP8B1 identified here share a common characteristic, i.e. they contain an overlapping CPF and HNF4␣-binding site. Furthermore, the nucleotide sequences of the bile acid response elements of the rat and human CYP8B1 are different in that the rat CYP8B1 contains two overlapping CPF-binding sites (31), whereas the human CYP8B1 contains only one CPF site. This may explain our observation that CPF stimulated but HNF4␣ had little effect on the rat CYP8B1, 2 in contrast to the strong stimulatory effect of HNF4␣ but weak effect of CPF on human CYP8B1 transcription. Thus SHP interacts with HNF4␣ or CPF and represses human or rat CYP8B1 transcription, respectively. Therefore, CPF and HNF4␣ differentially regulate CYP7A1 and CYP8B1 transcription in a species-and gene-specific manner.
In this study we further demonstrated that SHP interacted with HNF4␣ by mammalian two-hybrid assay. Our result is consistent with the report by Lee et al. (40) that SHP interacts with HNF4␣ and stimulates reporter activity by 4-fold in twohybrid assay in HepG2 cells. In contrast, Goodwin et al. (25) observed no interaction between SHP and HNF4␣ in mammalian two-hybrid assay in CV-1 cells. SHP lacks a DNA binding domain and functions predominantly as a negative factor that heterodimerizes with many nuclear receptors (41). It has been reported that SHP either directly inhibits the transactivating activity of nuclear receptors or competes for the coactivators (40).
It has been suggested that CPF (LRH) is a competence factor that potentiates the sterol response of rat CYP7A1 transcription by oxysterol receptor, liver X receptor (LXR) (24 10. Effects of HNF4␣, CPF, and SHP on human CYP8B1 transcription. Different amounts of HNF4␣ and SHP expression plasmids were cotransfected with phCYP8B1-514/ϩ300Luc reporter (1 g) in HepG2 cells. A, effect of HNF4␣ and SHP on CYP8B1-514/ϩ300/Luc reporter activity in HepG2 cells. B, dose-dependent effects of HNF4␣ and SHP on phCYP8B1-514/ϩ300Luc reporter activity. C, dose-dependent effects of CPF and SHP on CYP8B1-514/ϩ300Luc reporter activity. Experimental conditions were the same as under Fig. 1. D, mammalian two-hybrid assay of SHP and HNF4␣ interaction in HepG2 cells. HepG2 cells were transfected with 0.5 g of the reporter vector pG5Luc and 0.5 g each of the hybrid constructs or emptyl plasmids indicated. The pBIND and pACT are GAL4 and VP16 empty vectors, respectively. Hybrid plasmids Gal4/Id and VP16/MyoD are used as a positive control of two-hybrid assay (left panel). Right panel shows two-hybrid assays of Gal4/HNF4␣ and VP16/SHP. Normalized luciferase reporter activities from triplicate samples are presented. The numbers indicate folds of induction of reporter activity over the control assay that was transfected with Gal4/HNF4␣ and pACT.
trast, CPF potentiates HNF4␣ stimulation of human CYP7A1. It is also apparent that CPF is a weak transcription factor that stimulates human CYP7A1 (27) and rat CYP8B1 reporter (31) activity when cotransfected at high levels in non-liver cells. We found previously that CPF could function as a negative factor that inhibited human CYP7A1 transcription in transfection assay in HepG2 cells (39). CPF apparently is not important in regulating human CYP8B1 transcription as shown by mutagenesis analysis reported here.
In this study we revealed that bile acids were able to repress nuclear HNF4␣ protein expression in HepG2 cells transfected with FXR and treated with CDCA and also in bile acid-treated rat livers. HNF4␣ reporter activity was strongly repressed by CDCA and FXR. The rat HNF4␣ gene is regulated by CPF (26). Therefore, interaction of SHP with CPF may repress HNF4␣ gene transcription. We reported previously (42) that HNF4␣ protein expression was reduced by PPAR␣ and ligand Wy14,643 to explain the inhibition of CYP7A1 transcription by fibrates. De Fabiani et al. (33) reported that bile acids could suppress CYP7A1 transcription by reducing the transactivation activity of HNF4␣ through a mitogen-activated protein kinase cascade. All these studies support our finding that HNF4␣ plays a pivotal role in regulating bile acid synthesis genes. FXR, HNF4␣, CPF, and SHP are liver-enriched nuclear receptors with similar tissue expression patterns. They interact with each other and regulate gene expression during liver development and differentiation. It is intriguing that a specific bile acid receptor FXR induces a nonspecific receptor SHP that then interacts with other nuclear receptors to repress gene transcription. The expression of these nuclear receptors in hepatocytes must be tightly controlled to regulate liver gene expression during development and under different physiological and pathophysiological states.
CYP8B1 transcription is strongly inhibited by bile acids, cholesterol, and insulin. The expression of CYP8B1 activity may regulate the bile acid hydrophobicity in the bile that ultimately regulates the overall rate of bile acid synthesis by feedback inhibition of CYP7A1 transcription. It has been suggested that a lithogenic diet containing cholic acid and cholesterol induces gallstone formation by facilitating the absorption of cholesterol in the intestine (43). A high cholesterol diet is known to stimulate CYP7A1 but reduce CYP8B1 transcription and result in increasing the hydrophobicity of the bile. We suggest that FXR and LXR differentially regulate CYP7A1 transcription in different species (44). When fed a high cholesterol diet to rabbits, the bile acid hydrophobicity and pool size increase such that the negative effect of FXR may dominate over the positive effect of LXR and repress CYP7A1 transcription. On the other hand, the positive effect of LXR may dominate over the negative effect of FXR and result in the stimulation of CYP7A1 transcription in rats fed a high cholesterol diet. Insulin strongly inhibits both CYP7A1 and CYP8B1 transcription and results in decreasing the conversion of cholesterol to bile acids. Insulin is known to increase the synthesis of cholesterol and triglyceride by stimulating sterol response elementbinding protein isoform 1c (SREBP-1c) transcription and may contribute to hyperlipidemia in the patients with type II diabetes (45). Mutations of the HNF4␣ gene have been identified in patients with maturity onset diabetes of the young (MODY1) (46). Interestingly, mutations of the SHP gene have been identified in obese Japanese subjects with early onset diabetes (47). These investigators suggest that SHP is a candidate MODY gene that may regulate HNF4 activity in pancreas and control energy metabolism and body weight. In diabetes, bile acid synthesis and pool size increase in association with an increase in CYP8B1 activity (48,49). This is consistent with our results that SHP and HNF4␣ play important roles in regulating human CYP8B1 transcription. Thus, understanding the molecular mechanisms of CYP8B1 transcription by bile acids, cholesterol, and insulin is important for elucidating the mechanisms of lipid metabolisms and pathogenesis of diabetes. Nuclear extracts of HepG2 cells from different treatments were separated by SDS-polyacrylamide gel electrophoresis (10%) and transferred to a nitrocellulose membrane. A goat polyclonal antibody raised against the HNF4␣ was used to detect HNF4␣ protein. B, immunoblot of nuclear proteins using liver nuclear extracts isolated from rats. Rats were treated with different bile acids, cholestyramine, or cholesterol as described under "Experimental Procedures." Each lane contained 3 g of nuclear proteins. C, effects of CDCA and FXR/RXR␣ on mouse HNF4␣/ Luc reporter activity in HepG2 cells. A mouse HNF4␣/Luc reporter (1 g) was transfected into HepG2 cells that were treated with 25 M CDCA with or without the cotransfection of FXR and RXR␣ (0.25 g). The ␤-galactosidase activity was used to normalize luciferase activities.
In summary, we have unveiled a unique mechanism of HNF4␣-mediated bile acid repression of human CYP8B1 transcription. Bile acids are signaling molecules that may regulate many genes involved in lipid metabolisms. New drugs targeted to nuclear receptors including FXR, LXR, and HNF4␣ may modulate the transcription of the genes involved in bile acid synthesis, transport, and absorption and lead to the reduction of serum cholesterol levels and the prevention of cholestasis and other liver diseases.