Regulation of bile acid biosynthesis by hepatocyte nuclear factor 4 a

Hepatocyte nuclear factor 4a (HNF4a) regulates many genes that are preferentially expressed in liver. Mice lacking hepatic expression of HNF4a (HNF4a) exhibited markedly increased levels of serum bile acids (BAs) compared with HNF4a-floxed (HNF4a) mice. The expression of genes involved in the hydroxylation and side chain b-oxidation of cholesterol, including oxysterol 7ahydroxylase, sterol 12a-hydroxylase (CYP8B1), and sterol carrier protein x, wasmarkedly decreased inHNF4amice. Cholesterol 7a-hydroxylase mRNA and protein were diminished only during the dark cycle in HNF4a mice, whereas expression in the light cycle was not different between HNF4a and HNF4a mice. Because CYP8B1 expression was reduced in HNF4a mice, it was studied in more detail. In agreement with the mRNA levels, CYP8B1 enzyme activity was absent in HNF4a mice. An HNF4a binding site was found in the mouse Cyp8b1 promoter that was able to direct HNF4a-dependent transcription. Surprisingly, cholic acid-derived BAs, produced as a result of CYP8B1 activity, were still observed in the serum and gallbladder of these mice. These studies reveal that HNF4a plays a central role in BA homeostasis by regulation of genes involved in BA biosynthesis, including hydroxylation and side chain b-oxidation of cholesterol in vivo.— Inoue, Y., A-M. Yu, S. H. Yim, X. Ma, K. W. Krausz, J. Inoue, C. C. Xiang, M. J. Brownstein, G. Eggertsen, I. Björkhem, and F. J. Gonzalez. Regulation of bile acid biosynthesis by hepatocyte nuclear factor 4a. J. Lipid Res. 2006. 47: 215–227. Supplementary key words conditional knockout mice . sterol 12ahydroxylase . oxysterol 7a-hydroxylase . sterol carrier protein x . cholic acid Hepatocyte nuclear factor 4a (HNF4a; NR2A1), an orphan member of the nuclear receptor superfamily, regulates many genes that are preferentially expressed in the liver. HNF4a target genes include several serum proteins such as apolipoproteins, blood coagulation factors, cytochromes P450, and enzymes involved in glucose, ammonia, lipid, steroid, and fatty acid metabolism (1, 2). HNF4a is also involved in human genetic diseases; mutations in the HNF4a gene cause maturity-onset diabetes of the young 1 (3), and mutations in HNF4a binding sites were found in the promoter regions of the genes encoding blood coagulation factor IX and HNF1a, causing hemophilia B Lyden and maturity-onset diabetes of the young-3, respectively (4–6). Targeted disruption of the HNF4a gene was found to be embryo-lethal (7), indicating that HNF4a is an essential factor for mammalian development. To circumvent the problem of embryonic lethality, liver-specific HNF4a-null mice, designated HNF4a, were produced using the CreloxP system (8). These mice exhibited impaired lipid homeostasis and increased serum ammonia levels as a result of decreased expression of hepatic ornithine transcarbamylase (8, 9). In addition, HNF4a mice have increased bile acids (BAs), which could have a role in the high Manuscript received 29 September 2005 and in revised form 26 October 2005. Published, JLR Papers in Press, November 1, 2005. DOI 10.1194/jlr.M500430-JLR200 Abbreviations: ACOX2, trihydroxycoprostancyl-coenzyme A oxidase 2; BA, bile acid; CA, cholic acid; CDCA, chenodeoxycholic acid; CPF, cholesterol 7a-hydroxylase promoter factor; CYP7A1, cholesterol 7a-hydroxylase promoter factor; CYP7B1, oxysterol 7a-hydroxylase; CYP8B1, sterol 12a-hydroxylase; CYP27A1, sterol 27-hydroxylase; CYP39A1, oxysterol 7a-hydroxylase; DBP, albumin D site-binding protein; DCA, deoxycholic acid; D-PBE, D-type peroxisomal bifunctional enzyme; DR1, direct repeat 1; FXR, farnesoid X receptor; HNF4a, hepatocyte nuclear factor 4a; HNF4a, liver-specific hepatocyte nuclear factor 4a-null; HNF4a, hepatocyte nuclear factor 4a-floxed; LC-MS/MS, liquid chromatography tandem mass spectrometry; LRH-1, liver receptor homologue-1; MCA, muricholic acid; NTCP, sodium taurocholate cotransporter polypeptide; OATP1, organic anion transporter polypeptide 1; PPARa, peroxisome proliferator-activated receptor a; PXR, pregnane X receptor; RXRa, retinoid X receptor a; SCPx, sterol carrier protein x; SCP2, sterol carrier protein 2; SHP, small heterodimer partner; UDCA, ursodeoxycholic acid; VLACSR, very long chain acyl-coenzyme A synthase-related gene. 1 To whom correspondence should be addressed. e-mail: fjgonz@helix.nih.gov This article is available online at http://www.jlr.org Journal of Lipid Research Volume 47, 2006 215 by gest, on O cber 9, 2017 w w w .j.org D ow nladed fom mortality of these mice as they age (8). BAs derived from cholesterol are preferentially produced in the liver, and the resulting BA pool is maintained by enterohepatic circulation. BAs secreted from liver form micelles with hydrophobic compounds, including fatty acids, sterols, and vitamins, and are absorbed from the jejunum and ileum and recycled via the portal venous system. As a result, only a small portion of BA is excreted into urine and feces, and the constant BA pool is maintained by newly synthesized BAs (10). HNF4a mice exhibited greatly increased serum BA levels and reduced expression of BA transporters, such as sodium taurocholate cotransporter polypeptide (NTCP) and organic anion transporter polypeptide 1 (OATP1) (8). Furthermore, increased levels of unconjugated and glycine-conjugated BAs in gallbladder were observed in HNF4a mice, which were associated with the reduced expression of the very long chain acyl-CoA synthase-related gene (VLACSR) and BA CoA:amino acid N-acyltransferase involved in BA conjugation (11). These results indicate that HNF4a plays an important role in the regulation of enterohepatic circulation of BA. BA production is attributable mainly to the neutral and acidic pathways of BA biosynthesis via a cascade of enzymatic reactions (12, 13). Hydroxylation of the steroid nucleus and side chain of cholesterol is catalyzed by enzymes such as cholesterol 7a-hydroxylase (CYP7A1), oxysterol 7a-hydroxylase (CYP7B1), sterol 27-hydroxylase (CYP27A1), and sterol 12a-hydroxylase (CYP8B1). The side chain b-oxidations are catalyzed by the enzymes trihydroxycoprostanoyl-CoA oxidase [acyl-coenzyme A oxidase 2 (ACOX2)], D-type peroxisomal bifunctional enzyme (D-PBE), and sterol carrier protein x (SCPx). Using gene knockout mice and transient transfection analysis, several transcription factors and/or nuclear receptors, such as the farnesoid X receptor (FXR) (14), liver X receptor (15), CYP7A1 promoter transcription factor (CPF) (16), pregnane X receptor (PXR) (17, 18), peroxisome proliferator-activated receptor a (PPARa) (19), small heterodimer protein (SHP) (20, 21), HNF1a and HNF1b (22, 23), and fibroblast growth factor receptor 4 (24), were shown to regulate genes encoding BA biosynthesis enzymes. In this study, the role of hepatic HNF4a in BA biosynthesis was investigated. In vivo and in vitro data revealed that HNF4a is critical in the regulation of

Hepatocyte nuclear factor 4a (HNF4a; NR2A1), an orphan member of the nuclear receptor superfamily, regulates many genes that are preferentially expressed in the liver.HNF4a target genes include several serum proteins such as apolipoproteins, blood coagulation factors, cytochromes P450, and enzymes involved in glucose, ammonia, lipid, steroid, and fatty acid metabolism (1,2).HNF4a is also involved in human genetic diseases; mutations in the HNF4a gene cause maturity-onset diabetes of the young 1 (3), and mutations in HNF4a binding sites were found in the promoter regions of the genes encoding blood coagulation factor IX and HNF1a, causing hemophilia B Lyden and maturity-onset diabetes of the young-3, respectively (4)(5)(6).
Targeted disruption of the HNF4a gene was found to be embryo-lethal (7), indicating that HNF4a is an essential factor for mammalian development.To circumvent the problem of embryonic lethality, liver-specific HNF4a-null mice, designated HNF4a DL , were produced using the Cre-loxP system (8).These mice exhibited impaired lipid homeostasis and increased serum ammonia levels as a result of decreased expression of hepatic ornithine transcarbamylase (8,9).In addition, HNF4a DL mice have increased bile acids (BAs), which could have a role in the high mortality of these mice as they age (8).BAs derived from cholesterol are preferentially produced in the liver, and the resulting BA pool is maintained by enterohepatic circulation.BAs secreted from liver form micelles with hydrophobic compounds, including fatty acids, sterols, and vitamins, and are absorbed from the jejunum and ileum and recycled via the portal venous system.As a result, only a small portion of BA is excreted into urine and feces, and the constant BA pool is maintained by newly synthesized BAs (10).HNF4a DL mice exhibited greatly increased serum BA levels and reduced expression of BA transporters, such as sodium taurocholate cotransporter polypeptide (NTCP) and organic anion transporter polypeptide 1 (OATP1) (8).Furthermore, increased levels of unconjugated and glycine-conjugated BAs in gallbladder were observed in HNF4a DL mice, which were associated with the reduced expression of the very long chain acyl-CoA synthase-related gene (VLACSR) and BA CoA:amino acid N-acyltransferase involved in BA conjugation (11).These results indicate that HNF4a plays an important role in the regulation of enterohepatic circulation of BA.BA production is attributable mainly to the neutral and acidic pathways of BA biosynthesis via a cascade of enzymatic reactions (12,13).Hydroxylation of the steroid nucleus and side chain of cholesterol is catalyzed by enzymes such as cholesterol 7a-hydroxylase (CYP7A1), oxysterol 7a-hydroxylase (CYP7B1), sterol 27-hydroxylase (CYP27A1), and sterol 12a-hydroxylase (CYP8B1).The side chain b-oxidations are catalyzed by the enzymes trihydroxycoprostanoyl-CoA oxidase [acyl-coenzyme A oxidase 2 (ACOX2)], D-type peroxisomal bifunctional enzyme (D-PBE), and sterol carrier protein x (SCPx).

Animals
HNF4a DL mice were generated as described previously (8).All experiments were performed with 14, 28, and 45-day-old male HNF4a-floxed (HNF4a F/F ) mice and HNF4a DL mice.Mice were housed in a pathogen-free animal facility under a standard 12 h light/12 h dark cycle with ad libitum water and chow.All experiments with mice were carried out under Association for Assessment and Accreditation of Laboratory Animal Care guidelines with the approval of the National Cancer Institute Animal Care and Use Committee.

Microarray analysis 1
cDNA fragments from selected mouse genes were obtained by PCR from cDNA clones inserted into the plasmid pBluescript II or pUC18.The majority of the cDNA fragments contained z400 and 1,000 bp.Before spotting, the PCR fragments were dissolved in 50% DMSO solution to a final concentration of 100 ng/ml.Microarrays were spotted on amino-silane-coated glass slides (Corning, London, UK) using the Affymetrix 417 TM Arrayer.Each set of PCR fragments was arranged in six grids.mRNA was prepared from total liver RNA of 45-day-old mice at 10 AM using magnetic beads (Dynal Biotech, Oslo, Norway), and 1-2 mg was subjected to reverse transcription and labeled with the fluorescent dye Cy3 (CyScribe First-Strand cDNA Labeling Kit; Amersham Pharmacia Biotech, Uppsala, Sweden).Prehybridization and hybridization of the slides were carried out in hybridization cassettes at 658C essentially as described in the Amersham Pharmacia Biosciences manual.The slides were then analyzed with a ScanArray 4000XL (Perkin-Elmer, Foster City, CA).

Microarray analysis 2
Twenty micrograms of pooled total liver RNA from 45-day-old mice (n 5 5 for each group) at 10 AM was reverse-transcribed using oligo(dT) primers with either Cy5 or Cy3 dye using the indirect labeling method (25).The raw data were obtained using a GenPix 4000A scanner (Axon, Union City, CA) and further analyzed with National Cancer Institute microarray analysis tools (http://nciarray.nci.nih.gov) for normalization of each individual array.Each experiment was replicated by reciprocally labeling with either Cy3 or Cy5.Final values for differentially expressed genes were obtained from average regression ratios of Cy5/Cy3 (HNF4a DL /HNF4a F/F ) and the reverse value of Cy5/Cy3 (HNF4a F/F /HNF4a DL ).

Northern blot analysis
Northern blot analysis was carried out as described previously (9).All probes were amplified from a mouse cDNA library using gene-specific primers and cloned into pCR II vector (Invitrogen, Carlsbad, CA).Sequences were verified using the CEQ 2000 Dye Terminator cycle sequencing kit (Beckman Coulter, Fullerton, CA) with a CEQ 2000XL DNA Analysis System (Beckman Coulter).

Analysis of oxysterols in serum
The serum levels of the oxysterols 7a-, 25-, and 27-hydroxycholestrol were determined by isotope dilution mass spectrometry and the use of deuterium-labeled internal standards as described previously (26).

Total BA analysis
Mice were anesthetized with 2.5% avertin and decapitated, and the trunk blood was collected in a serum separator tube (Becton Dickinson, Franklin Lakes, NJ).The serum was separated by centrifugation at 7,000 g for 5 min and stored at 2208C before analysis.The total BA pool was determined as described previously (24).Urine samples and gallbladder biles were diluted 10 times with water and 1,000 times with 10% methanol, respectively, before analysis.The BA content of all samples was measured by colorimetric analysis using a BA analysis kit (Sigma, St. Louis, MO).

Analysis of gallbladder bile
Bile was hydrolyzed and analyzed by GC-MS essentially as described ( 27) using a Hewlett-Packard 5973 mass-specific detection instrument equipped with a 0.33 mm phase HP-ultra 1 column.The samples were methylated with diazomethane and trimethylsilylated before analysis.The samples were first analyzed by repetitive scanning to identify different BAs.Cholic acid (CA), chenodeoxycholic acid (CDCA), deoxycholic acid (DCA), and ursodeoxycholic acid (UDCA) were quantified by isotope dilution mass spectrometry with the use of deuterium-labeled internal standards and selected ion monitoring.Other BAs were quantified from the chromatogram/total ion current obtained in the analysis of material to which no internal standard had been added.The peak of trihydroxy BAs was compared with that of CA.The peak areas of the dihydroxy BAs were compared with that of DCA.

Analysis of serum BAs
Twenty microliters of a 50 mM dehydrocholic acid solution (in methanol, used as an internal standard) was added to 60 ml of the serum sample.The mixture was deproteinized by the addition of 200 ml of acetonitrile.The sample was vortexed for 10 s and centrifuged at 14,000 g at 48C for 5 min.The supernatant was transferred to a new glass tube and extracted with 2 ml of a mixture of ethyl acetate and t-butyl methyl ether (2:1, v/v).After centrifugation at 3,000 g at 48C for 15 min, the organic phase was transferred to a new glass tube and evaporated under a stream of nitrogen gas.The residue was reconstituted in 60 ml of 50% methanol solution containing 0.2% formic acid.Ten microliters of each reconstituted sample was injected for liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis.

Identification and quantitation of intact free BAs by LC-MS/MS
LC-MS/MS analysis was performed on a PE SCIEX API2000 ESI triple-quadrupole mass spectrometer (Perkin-Elmer) controlled by Analyst software.A Phenomenex Luna C18 3 mm, 100 mm 3 2 mm inner diameter column (Torrance, CA) was used to separate free BAs.The flow rate through the column at ambient temperature was 0.2 ml/min, and optimal resolution was achieved by elution with a linear gradient of water containing 0.1% formic acid (45%!0%) and methanol (55%!100%) in 10 min at room temperature.The mass spectrometer was operated in the turbo ion spray mode with positive ion detection.The turbo ion spray tem-perature was maintained at 3508C, and a voltage of 4.8 kV was applied to the sprayer needle.Nitrogen was used as the turbo ion spray and nebulizing gas.The detection and quantification of free and conjugated BAs and the internal standard were accomplished by multiple reaction monitoring with the transition m/z 403.3/ 367.4 for dehydrocholic acid, 359.1/135.0 for lithocholic acid, 375.1/357.2for free DCA, CDCA, hyodeoxycholic acid, UDCA, and murideoxycholic acid, 373.1/355.3for free CA, hyocholic acid, and aand v-muricholic acid (MCA), and 391.1/355.2for b-MCA.All raw data were processed using Analyst software.Calibration curves were linear for free DCA, CDCA, hyodeoxycholic acid, UDCA, and murideoxycholic acid concentrations ranging from 0.05 to 10 mM and for free CA, hyocholic acid, and a-, b-, and v-MCA concentrations ranging from 0.2 to 20 mM.

Western blot analysis
Frozen livers were homogenized in a lysis buffer (9 M urea, 2% Triton X-100, 70 mM dithiothreitol, 10 mg/ml aprotinin, 10 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride) and allowed to sit on ice for 30 min.The homogenate was centrifuged at 12,000 g for 30 min at 48C, and the supernatants were used as whole cell lysates.Total protein (100 mg), determined by the Bio-Rad assay, was diluted with 33 Laemmli sample buffer, incubated at 658C for 15 min, fractionated by 10% SDS-PAGE, and blotted onto a polyvinylidene difluoride membrane (Amersham Pharmacia Biotech, Piscataway, NJ).The membrane was incubated for 1 h with PBS containing 0.1% Tween 20, 5% dry milk, 6% glycine, and 3% fetal bovine serum and then for 1 h with a 1:1,000 dilution of primary antibodies against mouse CYP7A1 (a generous gift from Dr. David W. Russell, University of Texas Southwestern Medical Center) and b-actin (Santa Cruz Biotechnology, Santa Cruz, CA).After washing, the membrane was incubated for 1 h with a 1:10,000-diluted horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology), and the reaction product was visualized using SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL).

Determination of CYP8B1 enzyme activity
CYP8B1 enzyme activity was determined by a modification of a previous report (28).7a-Hydroxy-4-cholestene-3-one (Steraloids, Newport, RI) and 7a,12a-hydroxy-4-cholestene-3-one (a gift from Dr. Mizuho Une, Hiroshima University) were used as the substrate and its metabolite, respectively.Livers from CYP8B1-null and wild-type mice were used as negative and positive controls, respectively (29).Incubation reactions were carried out in 100 mM potassium phosphate (pH 7.4) containing 0.5 mg/ml microsomal protein, 2 mM NADPH, and 100 mM 7a-hydroxy-4cholestene-3-one at 378C for 10 min with shaking.The reaction was terminated by the addition of 2 ml of ethyl acetate.6b-Hydroxytestosterone was used as an internal standard.The reaction mixture described above was centrifuged at 1,200 g for 5 min, and the upper layer was collected and dried at 408C under nitrogen gas.The extracted residue was dissolved in 100 ml of methanol, and 10 ml was injected into an LC-MS/MS system for 7a,12a-hydroxy-4-cholestene-3-one detection.The LC-MS/MS system included a PE SCIEX API2000 ESI triple-quadrupole mass spectrometer (Perkin-Elmer) and a Synergi 4 mm MAX-RP, 50 3 2 mm inner diameter column.The flow rate was 0.2 ml/min with 70% methanol and 30% water containing 0.1% formic acid.The mass spectrometer was operated in the turbo ion spray mode with positive ion detection.The turbo ion spray temperature was maintained at 3508C, and a voltage of 5 kV was applied to the sprayer needle.Nitrogen was used as the turbo ion spray and nebulizing gas.The detection and quantitation of 7a,12a-hydroxy-4-cholestene-3-one and 6b-hydroxytestosterone were accomplished by selected ion monitoring with the transition m/z 439.1 for 7a,12a-hydroxy-4-cholestene-3-one and 327.2 for 6b-hydroxytestosterone.All raw data were processed using Analyst software.The detection limit for 7a,12a-hydroxy-4cholestene-3-one is z0.5 pmol.The sensitivity of the CYP8B1 assay is z0.1 pmol/min/mg protein.

Construction of the mouse CYP8B1-luciferase reporter plasmids
The 2707, 2497, 2235, 2173, and 2104/21 fragments from the translation start site of the mouse Cyp8b1 promoter (30) containing BglII and KpnI sites were amplified by PCR and cloned into the luciferase reporter vector, pGL3/basic (Promega, Madison, WI).Mutations were introduced into the putative HNF4a binding site in the Cyp8b1-luciferase constructs by PCR-based site-directed mutagenesis using the following complementary primer pair: 59-ccgagcctctgagcaCTGTCCAAGGGCAggaacct-39 and 59-aggttccTGCCCTTGGACAGtgctcagaggctcgg-39 (Mut; mutations in the HNF4a binding site are underlined).Plasmid DNA sequences were verified with a CEQ 2000XL DNA Analysis System (Beckman Coulter).

Transient transfection assay
HepG2 and CV-1 cell lines were cultured at 378C in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal bovine serum (HyClone, Logan, UT) and 100 U/ml penicillin/ streptomycin (Invitrogen).Cells were seeded on 24-well tissue culture plates and grown to 90-95% confluence.Transfections were carried out as described previously (9).All transfections were performed using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions.After 48 h, the cells were washed with phosphate-buffered saline and assayed for dual luciferase activity using commercial kits according to the manufacturer's instructions (Promega).
a P , 0.001 compared with HNF4a F/F mice.b P , 0.005 compared with HNF4a F/F mice.c P , 0.05 compared with HNF4a F/F mice.
For competition experiments, a 25-fold excess of unlabeled oligonucleotide was added to the reaction mixture and the mixture was incubated at room temperature for 20 min before the addition of a 32 P-labeled oligonucleotide probe.For supershift analysis, 1 mg of anti-HNF4a, PPARa, and retinoid X receptor a (RXRa) antibody (Santa Cruz Biotechnology) was added to the reaction mixture, and the mixture was incubated at room temperature for 30 min after the addition of a 32 P-labeled oligonucleotide probe.

Statistical analysis
All values are expressed as means 6 SEM or SD.All data were analyzed using the unpaired Student's t-test for significant differences between the mean values of each group.

RESULTS
Disruption of BA homeostasis in HNF4a DL mice HNF4a DL mice showed age-dependent increases of serum BA levels; serum BAs in HNF4a DL mice were increased as early as postnatal day 14 and were increased dramatically by 45 days of age compared with those of HNF4a F/F mice (Fig. 1A).CA, DCA, CDCA, a-, b-, and v-MCA, and UDCA were markedly increased, ranging from 10-to 60-fold compared with HNF4a F/F mice, in 45-dayold HNF4a DL mice (Table 1).Additionally, a large number of unidentified BAs were also observed in the serum of HNF4a DL mice (data not shown).The expression of

HHF4a regulates BA biosynthesis
HNF4a was reduced z50% in the livers of HNF4a DL mice on day 14 and was not detected in 28-day-old HNF4a DL mice (Fig. 2), showing that the high serum BA levels in HNF4a DL mice were correlated with the loss of HNF4a expression.
The gallbladders of HNF4a DL mice were much larger than those of HNF4a F/F mice.This was attributable in part to an increase in the volume of biofluid and BA levels in the HNF4a DL mice (Fig. 1B, C).The total BA pool size was similar in both groups (Fig. 1D), whereas the fecal BA excretion rate was decreased (Fig. 1E) and the urinary BA excretion rate was increased (Fig. 1F) in HNF4a DL mice.The total BA excretion rate, derived from the sum of the fecal and urinary BA excretion, was still markedly lower in HNF4a DL mice.BA production in the liver is directly proportional to excreted BA; thus, the synthesis of BA in HNF4a DL mice was decreased compared with that in HNF4a F/F mice.

Expression of genes involved in BA biosynthesis in HNF4a DL mice
To determine the differences in the expression of potential HNF4a target genes involved in BA biosynthesis, microarray analysis was carried out using liver RNA collected from mice killed at 10 AM (Table 2).A marked decrease in the expression of genes encoding CYP7B1, CYP8B1, OATP1, and NTCP was found in 45-day-old HNF4a DL mice.Interestingly, there was no significant change in the expression of the Cyp7a1 gene, whereas expression of the apolipoprotein B gene was decreased and mRNAs encoding HMG-CoA synthase, scavenger receptor class B type I, PXR, RXRa, and LCAT were increased in the HNF4a DL mice.No increased expression of PXR and RXRa mRNAs was observed, as revealed in earlier studies using Northern blot analysis (8).The differentially expressed genes between HNF4a DL and HNF4a F/F mice by microarray analysis were further confirmed by Northern blot analysis and real-time PCR.By Northern analysis, the levels of CYP7A1 mRNA were not significantly different between HNF4a DL and HNF4a F/F mice at the ages tested (Fig. 2A); however, serum levels of 7a-hydroxycholesterol, which reflects the enzymatic activity of CYP7A1 (31), were significantly reduced in HNF4a DL mice (10 6 1 ng/ml) compared with HNF4a F/F mice (17 6 1 ng/ml).CYP27A1 and CYP7B1 mRNA levels were reduced significantly in the livers of HNF4a DL mice (Fig. 2A, B).However the serum levels of 27-hydroxycholesterol, the product of CYP27A1 and the substrate for CYP7B1 (32), were .2-foldhigher in HNF4a DL mice than in HNF4a F/F mice (118 vs. Pooled liver total RNA (n 5 5 for each genotype) from 45-day-old mice at 10 AM was used.Regression ratio is expressed relative to HNF4a F/F mice.
Expression of mRNAs encoding trihydroxycoprostanoyl-CoA oxidase (ACOX2) and SCPx was also reduced significantly in HNF4a DL mice (Fig. 2A, D), but expression of D-PBE mRNA was unchanged (Fig. 2A, E).Candidate genes were also detected using a cDNA microarray with liver RNA collected at 10 AM (Table 3).Decreased expression of mRNA encoding sterol carrier protein 2 (SCP2), which is an alternatively spliced variant of SCPx, and a shorter transcript derived from a different promoter from the SCPx promoter (35) was found.Expression of SCP2 was unchanged using a SCP2-specific probe (data not shown), but SCPx was decreased significantly in HNF4a DL mice using a specific probe that recognizes SCP2 and SCPx (Fig. 2A).

Expression of CYP7A1 is not induced during the dark cycle in HNF4a DL mice
The data in Table 2 and Fig. 2, derived from mice killed in the morning, revealed no difference in CYP7A1 expression between genotypes.However, the expression of CYP7A1 is known to be regulated by circadian rhythm (36,37).To determine whether the loss of HNF4a affected the response of the gene to light, CYP7A1 mRNA and protein levels were examined (Fig. 3).The expression of CYP7A1 mRNA in the livers of HNF4a DL mice was unchanged compared with that of HNF4a F/F mice in the light cycle at 10 AM (Fig. 3A), as revealed by microarray and Northern blot analyses (Fig. 2A).In marked contrast, the expression of CYP7A1 in the dark cycle at 10 PM was increased in HNF4a F/F control mice compared with HNF4a DL mice (Fig. 3B).These results were further confirmed by real-time PCR; CYP7A1 mRNA was not different at 10 AM between HNF4a DL and HNF4a F/F mice but was increased significantly in HNF4a F/F mice but not in HNF4a DL mice at 10 PM (Fig. 3C).Similarly, the expression of CYP7A1 protein was not different between HNF4a F/F and HNF4a DL mice at 10 AM (Fig. 3C) but decreased in HNF4a DL mice compared with HNF4a F/F mice at 10 PM (Fig. 3D).Thus, the increase in the serum levels of 7a-hydroxycholesterol during the light cycle could be attributable to the increased CYP7A1 mRNA found in the HNF4a F/F mice at 10 PM compared with the HNF4a DL mice.In contrast, mRNA encoding CYP8B1 was higher during the light cycle in HNF4a F/F mice but remained low in the HNF4a DL mice (Fig. 3A, B).CYP7B1 and CYP27A1 mRNA levels were not influenced by the light and dark cycles (Fig. 3A, B).BA composition in HNF4a DL and HNF4a F/F mice The expression of several genes encoding enzymes involved in BA biosynthesis was reduced in the livers of HNF4a DL mice, suggesting that the BA composition also might be altered in these mice.In particular, the fact that Cyp8b1-null mice lack synthesis of CA (29) suggests that HNF4a DL mice should also be devoid of CA.The composition of gallbladder BA was quantified by GC-MS (Fig. 4).In male control HNF4a F/F mice, the primary BAs, CA (52.2%) and b-MCA (29%), accounted for z82% of the total BAs.The remaining metabolites were DCA (2.7%), a-MCA (1.5%), v-MCA (8%), CDCA (0.7%), and UDCA (1.8%).On the other hand, CA was decreased to 44%, but its secondary metabolite DCA was increased 3-fold (8.4%), in HNF4a DL mice, indicating that the overall amount of CA-derived metabolites (CA and DCA) was unchanged in HNF4a DL mice (52.4%) versus that in control HNF4a F/F mice (54.9%).

Hepatic CYP8B1 enzyme activity is reduced in HNF4a DL mice
To further investigate the ratio of CA-derived BAs in HNF4a DL mice, the activity of hepatic CYP8B1 was measured.The expression of CYP8B1 is also regulated by circadian rhythm, and the expression pattern during the dark/light cycle is opposite from that of CYP7A1: it is highly expressed during the light cycle and minimally expressed during the dark cycle (36,37).Northern blot analysis revealed that expression of CYP8B1 in control mice at 10 AM was higher than that at 10 PM (Fig. 3A, B).Furthermore, the expression of CYP8B1 in HNF4a DL mice was reduced markedly in both cycles compared with that in HNF4a F/F mice.CYP8B1 enzyme activity in control mice was also significantly higher at 10 AM than at 10 PM, and the activity in HNF4a DL mice was again reduced markedly at both cycles (Fig. 5A).This assay was specific to CYP8B1, because no activity was observed in the livers of Cyp8b1-null mice (Fig. 5B).

HNF4a directly binds to the promoter region of the mouse Cyp8b1 and positively regulates its expression
Because CYP8B1 is expressed mainly in liver (38), the possibility was investigated that HNF4a can directly bind to and regulate the promoter of the mouse Cyp8b1 gene.The mouse Cyp8b1 promoter contains a direct repeat 1 (DR1) motif (at position 2109 to 294 from the translation start site) that is a potential binding site for nuclear receptors such as HNF4a and PPARa.To determine whether Fig. 5. Specific activities and expression of CYP8B1 in HNF4a DL and HNF4a F/F mice.A: Specific activities of CYP8B1 were measured using liver microsomal protein from HNF4a DL and HNF4a F/F mice.Data are means 6 SEM (n 5 5).Significant differences between HNF4a DL and HNF4a F/F mice: * P , 0.001 and ** P , 0.005, respectively.B: Specific activities of CYP8B1 were measured using liver microsomal protein from Cyp8b1-null and wild-type (WT) mice.Data are means 6 SEM (n 5 2).HNF4a is able to activate the mouse Cyp8b1 promoter, several Cyp8b1 promoter-luciferase reporter plasmids were constructed (Fig. 6A).When the human hepatomaderived HepG2 cells, which express HNF4a, were used for transient transfections, the promoter activity of the 2115 bp fragment containing a putative HNF4a binding site was z6-fold higher than that of the promoterless construct (pGL3/basic) and the 290 bp fragment (Fig. 6A).The activities of the longer DNA fragments were 6-to 9-fold higher than that of the 2115 bp fragment.To determine whether this activity was attributable to HNF4a, CV-1 cells, which do not express HNF4a, were used for the transfection experiments.The promoter activity of the 290 bp fragment was unchanged by cotransfection of the HNF4a expression vector, consistent with the absence of an HNF4a binding site (Fig. 6B).However, the activity of the 2115 bp fragment containing the HNF4a binding site was increased by cotransfection with an HNF4a expression vector.Similar results were obtained with the longer DNA fragments; however, the activities with these DNAs were much greater than that of the 2115 bp fragment in both CV-1 cells (Fig. 6B) and HepG2 cells (Fig. 6A).
To determine whether HNF4a can bind to the DR1-like element in the mouse Cyp8b1 promoter, gel shift analysis was performed with 32 P-labeled probes (Fig. 6C).Liver Fig. 6.Promoter analysis of the mouse Cyp8b1 gene.A: Luciferase reporter plasmids containing the mouse Cyp8b1 promoter were transfected into HepG2 cells.The normalized activity 6 SEM (n = 4) of each construct is presented as arbitrary units.B: CV-1 cells were cotransfected with the HNF4a expression vector, as indicated.The normalized activity 6 SEM (n = 4) of each construct is presented as arbitrary units.C: Nuclear extracts from liver of HNF4a F/F mice (left panel) and HNF4a DL mice (right panel) were incubated with a labeled HNF4a binding site derived from the mouse Cyp8b1 promoter in the absence (lanes 1 and 8) or presence of each unlabeled oligonucleotide for the HNF4a binding site of the mouse Cyp8b1 promoter (lanes 2 and 9), consensus HNF4a binding site (lanes 3 and 10), and consensus peroxisome proliferatoractivated receptor a (PPARa)-responsive element (lanes 4 and 11).For the supershift assay, nuclear extracts were incubated with labeled oligonucleotide probe in the presence of anti-HNF4a (lanes 5 and 12), anti-PPARa (lanes 6 and 13), and anti-retinoid X receptor a (RXRa) antibodies (lanes 7 and 14).The HNF4a-DNA complex and its supershifted complex, caused by the HNF4a-specific antibody, are indicated by the lower and upper arrows, respectively.
HHF4a regulates BA biosynthesis nuclear extracts from HNF4a F/F mice contained proteins that bound to the HNF4a binding site (Fig. 6C, lane 1, lower arrow).The HNF4a-specific binding was diminished by the addition of excess amounts of unlabeled probe and HNF4a consensus sequence but not by a PPARa-responsive element (Fig. 6C, lanes 2-4).Furthermore, these bands were supershifted by the addition of anti-HNF4a antibody, indicating that the protein bound to the DR-1like element was indeed HNF4a (Fig. 6C, lane 5, upper arrow).The specificity of binding was demonstrated by the lack of a supershift by antibodies against PPARa and RXRa (Fig. 6C, lanes 6, 7).As expected, no HNF4aspecific supershifted bands were detected when liver nuclear extracts from HNF4a DL mice were used (Fig. 6C, lane 12).These results indicate that the DR-1-like element in the mouse Cyp8b1 promoter has HNF4a-specific binding capacity.
To determine whether disruption of the HNF4a binding site in the Cyp8b1 gene affects promoter activity, mutations were introduced into the site in the (2330)luciferase promoter construct (Fig. 7A, Mut).As shown in Fig. 7B, when HepG2 cells were used for transient transfections, promoter activity of the mutated 2330 bp fragments (2330/Mut) was reduced to background levels compared with that of the wild-type promoter (2330/WT).Similarly, when CV-1 cells were used, the promoter activity of the mutated fragment was not induced by cotransfection of HNF4a expression vector (Fig. 7C).Together, the results shown in Figs. 6 and 7 suggest that HNF4a directly regulates the expression of the Cyp8b1 gene.

DISCUSSION
Cholesterol is hydroxylated by CYP7A1 and CYP27A1, the first enzymes in the classic and neutral BA biosynthesis pathways, respectively (12,13).Human Cyp7a1 has an HNF4a binding site in its promoter region (39,40).However, the expression of CYP7A1 mRNA and protein was unchanged during the light cycle between HNF4a DL and HNF4a F/F mice but higher during the dark cycle only in HNF4a F/F mice, indicating that HNF4a control of the Cyp7a1 gene may only be significant during the dark cycle.It is possible that the influence of other transcription factors during the dark cycle is decreased, allowing HNF4a to predominate.Earlier studies have revealed that the expression of DBP, which is able to transactivate the Cyp7a1 promoter, is correlated with CYP7A1 expression (37,41).A relationship between DBP and HNF4a in the regulation of Cyp7a1 has not been explored.Of interest, the expression of HNF4a was significantly increased in human liver biopsies from patients treated with cholestyramine in parallel with markedly increased expression of CYP7A1 (42), indicating that HNF4a may also regulate the expression of CYP7A1 in humans, as revealed by in vitro reporter gene studies (39,40).It is not known whether the human gene is diurnally regulated.BA concentrations may also mask the influence of HNF4a on Cyp7a1 expression.Because the enzyme activity of CYP7A1 and the mRNA encoding CYP27A1 are reduced and the BA excretion rate is also reduced, BA biosynthesis must also be lower in HNF4a DL mice.In agreement with the CYP7A1 mRNA and protein expression data, the levels of 7a-hydroxycholesterol were reduced significantly in serum of HNF4a DL mice compared with HNF4a F/F mice.A lower activity of the enzyme would be expected to result in less production of BAs and less feedback suppression of Cyp7a1 through the FXR-SHP-liver receptor homologue-1 (LRH-1) pathway (43) and an increased expression of the gene.However, SHP mRNA levels are not significantly different between HNF4a F/F and HNF4a DL mice (Table 2 and unpublished results).Clearly, the Cyp7a1 gene is under complex regulation by FXR, SHP, LRH-1, liver X receptor, HNF4a (43), and DBP (41).Which of these factors predominates in the control of Cyp7a1 will depend on the physiological situation.
Studies with the HNF4a DL mice indicate that the Cyp27a1 gene is directly regulated by HNF4a in vivo.This is in agreement with a recent report showing that HNF4a can bind to and transactivate the promoter region of the human Cyp27a1 gene (44,45).Thus, both mouse Cyp27a1 and human Cyp27a1 are direct targets for HNF4a.Interestingly, the levels of 27-hydroxycholesterol were significantly higher in HNF4a DL mice than in HNF4a F/F mice.The circulating levels of this oxysterol are regulated by the relative activities of CYP27A1 and CYP7B1.Both Cyp27a1 and Cyp7b1 were downregulated in HNF4a DL mice, but the downregulation of Cyp7b1 was almost complete.Therefore, it seems likely that the increased levels of 27-hydroxycholesterol in HNF4a DL mice reflect a markedly reduced metabolism of oxysterols.This finding was also observed in the Cyp7b1-null mice that have higher serum levels of both 25-and 27-hydroxycholesterol (46).
HNF4a binds to the promoter region of human Cyp8b1 and activates transcription, as revealed by transactivation transcription assays (47).Other studies have suggested that BAs inhibit the expression of the rat Cyp8b1 by inducing CPF/LRH-1 and inhibiting HNF4a expression (48).However, repression of Cyp8b1 in HNF4a DL mice might not be attributable to this negative feedback mechanism by BAs because CPF/LRH-1 is not induced in HNF4a DL mice (8).The work described here indicates that HNF4a positively regulates the basal expression of Cyp8b1 in vivo; thus, lower HNF4a that results from high BA levels could potentially cause a decrease in Cyp8b1 transcription.It should also be noted that CYP8B1 is needed to synthesize CA, and Cyp8b1-null mice lack synthesis of CA and its metabolites (29).However, surprisingly, CA-derived metabolites (CA and DCA) in HNF4a DL mice were not decreased compared with other BAs in the absence of CYP8B1 mRNA, and the concentration and/or total amount of CA in serum and gallbladder bile were increased along with other BAs.Furthermore, the enzyme activity of hepatic CYP8B1 was also markedly decreased, and no measurable levels of expression of CYP8B1 mRNA were detected in other tissues examined.We currently have no explanation for this discrepancy between CYP8B1 expression and CA levels.
The expression of CYP7B1 mRNA was almost extinguished in the HNF4a DL mice.However, a functional HNF4a binding site could not be located in the Cyp7b1 promoter region.Because expression of the human Cyp7b1 gene is regulated by Sp1 (49), the mouse Cyp7b1 gene might also be regulated by Sp1.Another mechanism involving Sp1 could also contribute to the loss of Cyp7b1 gene expression in the HNF4a DL mice.BA, found at high levels in HNF4a DL mice, induces hepatic cytokine expression by activation of hepatic macrophage-like Kupffer cells (50).Tumor necrosis factor-a produced by these cells represses growth hormone receptor expression by inhibiting Sp1 and Sp3 binding to their corresponding sites in the promoter (51).Thus, hepatic production of tumor necrosis factor-a and other cytokines might be increased in HNF4a DL mice, resulting in repression of the DNA binding activity of Sp1 family proteins to the Cyp7b1 promoter.As discussed above, the expression of the major genes involved in BA biosynthesis is reduced in HNF4a DL mice, including that encoding Scpx.This gene also does not contain a functional HNF4a site within proximity to its transcription start site, but it does contain GC-rich sequences that have potential Sp1 binding (data not shown).Thus, a mechanism similar to that described for the Cyp7b1 gene may have a role in the loss of Scpx gene expression in HNF4a DL mice.Finally, the presence of an HNF4a binding site at a remote distance from the Cyp7b1 and Scpx promoters cannot be ruled out.
Because side chain b-oxidation of bile alcohols is catalyzed in peroxisomes, a deficiency in this pathway has been described as a secondary feature of various peroxisome disorders (52).The gene encoding SCPx also encodes SCP2, which contains the C-terminal sequence of SCPx ( 53), but the expression of SCP2 is unchanged (data not shown) because of the use of different promoters (32).Scp2/Scpxnull mice exhibit decreased catabolism of branched fatty acyl-CoA (54) and loss of side chain b-oxidation of bile alcohol at the thiolytic step (55), indicating that SCPx is indeed critical for side chain b-oxidation of bile alcohols.Scp2/Scpx-null mice also accumulate intermediate products of BA biosynthesis (55).These intermediate products might accumulate in HNF4a DL mice as a result of the loss of SCPx expression, thus accounting for the additional unidentified peaks detected by LC-MS/MS in these mice.
After BA biosynthesis by hydroxylation of cholesterol and side chain b-oxidation, BAs are conjugated with taurine to produce bile salts in mice.The two main enzymes involved in BA conjugation are VLACSR and BA CoA: amino acid N-acyltransferase in mice.The expression of these genes was reduced in HNF4a DL mice and dependent on HNF4a binding sites and HNF4a expression (11).Thus, HNF4a regulates all steps of BA biosynthesis, transport, and conjugation.
In conclusion, characterization of liver-specific HNF4anull mice provided evidence for a critical role for hepatic HNF4a in the maintenance of BA homeostasis.Because other genes are also downregulated in the livers of HNF4a DL mice in a mechanism that does not appear to involve HNF4a, further investigation of these genes might be useful to understand the regulation of the complete pathway of BA biosynthesis and transport and possibly to establish novel therapeutic targets for diseases resulting from altered BA homeostasis.

Fig. 3 .
Fig.3.Circadian rhythm of the expression of cholesterol 7a-hydroxylase (CYP7A1) and sterol 12a-hydroxylase (CYP8B1) mRNAs in the livers of HNF4a DL and HNF4a F/F mice.A, B: Northern blot analysis of mRNA in livers from 45-day-old mice at 10 AM (A) and 10 PM (B).C: Real-time PCR of liver mRNA for CYP7A1 at 10 AM (left) and 10 PM (right).Data are means 6 SEM (n 5 8 for each group).Significant differences compared with the other three groups: * P , 0.01.D, E: Western blots of total liver proteins (100 mg) at 10 AM (D) and 10 PM (E).CYP7A1 and b-actin polyclonal antibodies were used to assess protein expression.

Fig. 4 .
Fig. 4. Composition of BA in HNF4a DL and HNF4a F/F mice.Amounts of individual BAs from 45-day-old mice were analyzed by GC-MS and are indicated as percentages of the entire pool.The position and stereochemistry of hydroxyl groups on the ring structure of each BA are indicated in parentheses.

Fig. 7 .
Fig. 7. Effects of mutations of the HNF4a binding site in the mouse Cyp8b1 promoter.A: Scheme of the wild-type (WT) and mutated (Mut) HNF4a binding site of the mouse Cyp8b1 promoter.Mutations in the HNF4a binding site are represented in boldface type and underlined.B, C: Plasmids were transfected into HepG2 (B) and CV-1 (C) cells, and the normalized activity 6 SEM (n = 4) of each construct is presented as arbitrary units.

TABLE 1 .
Analysis of free bile acids in serum of HNF4a DL and HNF4a F/F mice

TABLE 2 .
Gene expression changes in HNF4a DL mice-I

TABLE 3 .
Gene expression changes in HNF4a DL mice-II