Role of Hepatocyte Nuclear Factors in Transcriptional Regulation of Male-specific CYP2A2 *

Cytochrome P450 2A2 (CYP2A2) is an adult male-spe-cific rat liver steroid hydroxylase whose sex-dependent expression is regulated at the transcriptional level by sexually dimorphic pituitary growth hormone (GH) se-cretory patterns. In contrast to CYP2C11 and other male-specific, plasma GH pulse-inducible liver genes, CYP2A2 is highly expressed in hypophysectomized rat liver, despite the absence of GH stimulation. CYP2A2 promoter fragments 0.9–6.2 kb long exhibited unusually high basal promoter activity when transfected into the liver cell line HepG2. A further (cid:1) 2.5-fold increase in activity was obtained by cotransfection of hepatocyte nuclear factor (HNF) 3 (cid:1) or HNF4 (cid:2) . CYP2A2 promoter activity was inhibited (cid:1) 85% by transfection of HNF3 (cid:3) or HNF6, both of which are more highly expressed in female than male liver and can strongly trans -activate the female-specific CYP2C12 promoter. The male GH pulse-activated transcription factor STAT5b had no effect on CYP2A2 promoter activity, either alone or in combination with HNF3 (cid:1) and HNF4 (cid:2) , consistent with the GH pulse-independence of CYP2A2 expression. By contrast, STAT5b synergistically enhanced the transcriptional activity of HNF4 (cid:2) toward two other male-specific liver target genes, Cyp2d9

Growth hormone (GH) 1 is secreted by the pituitary gland in a highly pulsatile manner. In the adult male rat, regular plasma GH pulses of 200 -300 ng/ml occur every ϳ3.5 h, separated by periods during which GH is virtually absent from circulation. Pituitary GH release is more frequent in the adult female rat and results in a nearly continuous presence of GH in circulation at concentrations typically ranging from ϳ15 to 40 ng/ml (1). These sexually dimorphic plasma GH profiles, in turn, dictate the sex-dependent effects that GH imparts to body growth rates at puberty and to the sex-dependent expression of several hepatic cytochromes P450 (CYPs) (2,3) and other liver gene products (4). In the rat, sex-specific liver CYPs include the male-specific testosterone 16␣-and 2␣-hydroxylase CYP2C11, which is induced at puberty in male but not female rat liver, and the steroid sulfate 15␤-hydroxylase CYP2C12, which is induced in female rat liver at a similar developmental stage. Notable sex differences have also been reported for mouse (5,6) and human liver CYPs (7,8).
Liver CYP genes are regulated at the level of transcription initiation by distinct cues within sexually dimorphic plasma GH profiles (9). Continuous exposure to GH, mimicking the adult female plasma GH profile, induces the expression of female-specific CYPs, such as CYP2C12, and concomitantly inhibits the expression of male-specific CYPs, such as 2C11, 2A2, and 4A2 (10). In contrast, intermittent plasma GH pulses, characteristic of adult male rats, stimulate the expression of CYP2C11 in male rat liver (11). CYP2C11 is a "class I" malespecific liver CYP gene, i.e. one whose high level expression requires repeated stimulation by male plasma GH pulses. CYPs 2A2, 3A2, and 4A2 belong to a second, distinct class of male-specific P450s. In contrast to the class I genes, expression of these class II CYPs remains at a high level in hypophysectomized male rat liver and is induced to nearly normal male levels after hypophysectomy of females. Thus, hepatic expression of class II male CYPs is independent of plasma GH pulses (12).
STAT5b is a latent cytoplasmic transcription factor that is uniquely responsive to the male pulsatile GH pattern and is proposed to be a key mediator of the sexually dimorphic response of liver CYPs to GH stimulation (13). The importance of STAT5b in GH pulse-stimulated, sex-specific liver gene expression is consistent with the GH pulse-induced, intermittent high levels of active, nuclear STAT5b found in adult male but not female rat liver (14 -16) and is strongly supported by the loss of sexually dimorphic Cyp gene expression in STAT5b-null male mice (17)(18)(19)(20). A similar sex dependence in hepatic STAT5b activity is seen in mice (6). STAT5b binding sites have been localized in the promoters of several male-specific genes, including rat CYP genes 2C11, 2A2, and 4A2 (21), hamster CYP3A10 (22,23), and mouse Slp (24). However, STAT5b by itself is not sufficient to induce the adult male pattern of liver CYP expression, as demonstrated by the precocious activation of STAT5b, but not male CYPs, in prepubertal rats given periodic (pulsatile) injections of GH (15).
Analysis of CYP promoter sequences has revealed consensus binding sites for multiple hepatocyte-enriched nuclear factors (HNFs) (25). These liver transcription factors are characterized by structurally diverse DNA binding domains and include the variant homeodomain containing protein HNF1␣, CCAAT/enhancer binding proteins, HNF3 winged helix factors, the or-phan nuclear receptor HNF4␣, and the one-cut homeoprotein HNF6 (26). The class I male-specific CYP2C11 promoter is strongly trans-activated by HNF1␣ and HNF3␤ (21), whereas the female-specific CYP2C12 gene is in part regulated by HNF3␤ and HNF6 acting together in a highly synergistic manner (27,28). Moreover, the nuclear receptor HNF4␣ plays a central role in regulating sex-dependent mouse liver Cyp gene expression in vivo, as demonstrated in an HNF4␣-deficient mouse liver model (29) (for review, see Ref. 30).
Here we investigate the role of HNFs in regulating the CYP2A2 promoter. We demonstrate that CYP2A2 is responsive to multiple HNFs, with HNF4␣ and HNF3␥ stimulating gene expression and HNF3␤ and HNF6 inhibiting gene expression. Furthermore, CYP2A2 is shown to be unresponsive to GH pulse-activated STAT5b, which we show can synergize with HNF4␣ to activate two other male-specific HNF4␣-responsive promoters, CYP8B1 (31) and Cyp2d9 (32), but not CYP2A2. These findings are discussed in the context of a model whereby STAT5b cooperates with HNF4␣ to regulate select sex-specific liver CYP genes.

MATERIALS AND METHODS
Antibodies-Rabbit polyclonal anti-STAT5b antibody, raised against STAT5b residues 776 -786, was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Mouse polyclonal anti-V5 antibody, which detects a 14-amino acid epitope derived from the P and V proteins of the paramyxovirus SV5, was purchased from Invitrogen. Rabbit polyclonal anti-STAT5b-pY694 which is cross-reactive with STAT5b-Y694, was purchased from Cell Signaling Technology, Inc. ( GGC TTG ACT AGT GGG ATT T-3Ј; HNF1␣ antisense, 5Ј-CTG GGA  GGA GGA GGC CAT CTG GG-3Ј; HNF3␤ sense, 5Ј-CAC CAG TAT GCT  GGG AGC CGT GAA G-3Ј; HNF3␤ antisense, 5Ј-GGA CGA GTT CAT  AAT AGG CCT-3Ј; HNF3␥ sense, 5Ј-CAC CAT GCT GGG CTC AGT  GAA GAT GG-3Ј; HNF3␥ antisense, 5Ј-GGA TGC ATT GAG CAG AGA  GCG-3Ј; HNF4␣ sense, 5Ј-CAC CGC CGA CAT GGA CAT GGC TGA  C-3Ј; HNF4␣ antisense, 5Ј-GAT GGC TTC CTG CTT GGT GAT CG-3Ј; HNF6 sense, 5Ј-CAC CAG CCC GCT CAC CCG CAT C-3Ј; and HNF6 antisense, 5Ј-TGC TTT GGT ACA AGT GCT TG-3Ј. The 4-nt sequence CACC was incorporated at the 5Ј-end of all sense PCR primers (see above) to facilitate the Directional TOPO cloning reaction. Resultant PCR products were subcloned into pcDNA3.1D/V5-His-TOPO using the pcDNA3.1 Directional TOPO Expression Kit (Invitrogen). In short, blunt end PCR products for each cDNA of interest (4 l of fresh PCR product) were incubated for 5 min at 27°C together with 1 l of pcDNA3.1D/V5-His-TOPO vector and 1 l of salt solution (1.2 M NaCl, TABLE I HNF and STAT5b binding sites localized within clustered DNA binding motif regions in the 5Ј-flank of CYP2A2, CYP4A2, and CYP8B1 Proximal promoter regions encompassed by the rat Ϫ6232/CYP2A2-Luc (accession no. NM_012693), rat Ϫ2458/CYP4A2-Luc (M_57719), and human Ϫ514/CYP8B1-Luc (NM_004391) reporters were analyzed using the Web-based program Cluster Buster to identify clustered DNA binding motifs as described under "Materials and Methods." Shown are all DNA binding sites found within these clusters for STAT5b, HNF3␣, HNF3␤, HNF4␣, and HNF6 that have binding site scores Ͼ 4, as determined using binding site matrices included in the TransFac data base. STAT5b sites are those that match the consensus sequence TTC-NNN-GAA. Additional binding sites for STAT5b and the HNFs may be present outside of the clusters of transcription factor binding sites identified by Cluster Buster. Transcription factor binding site nts shown in upper case represent the core binding sequence, whereas nts in lower case correspond to the flanking sequence. Binding site locations listed are for the binding site sequences shown, including the flanking sequence. Factor binding sites shown are on the plus or minus DNA strand, as indicated. Higher binding site scores indicate a stronger match to the transcription factor binding site matrix. No HNF or STAT5b sites were found within clustered DNA binding motif regions in the short Cyp2d9 promoter fragment (112 nt) included in this analysis. Only one clustered STAT5b site, but no HNF sites, were found for CYP4A2; however, one HNF4␣ site and two HNF3 sites were found outside of the clusters. Likewise, no STAT5b sites were found in the CYP8B1 promoter fragment. No HNF6 sites were found to be present in any of the promoter sequences examined. Binding site matrices for HNF3␥ were not available in the TransFac data base. 0.06 M MgCl 2 ). The total Directional TOPO reaction was then transformed using One Shot TOP10 Escherichia coli (Invitrogen) according to the manufacturer's instructions. Correct pcDNA3.1-HNF-V5-His expression plasmid construction was confirmed by DNA sequencing, and protein size was verified by anti-V5 Western blot.
CYP2A2 and CYP4A2 Promoter Plasmids-A pUC19 bacterial plasmid containing a 7.7-kb insert of the 5Ј-flank of the rat CYP2A2 gene (33) was obtained from Dr. F. J. Gonzalez (NCI, National Institutes of Health) and used to construct the CYP2A2 promoter-luciferase reporter plasmids Ϫ6232/2A2-Luc, Ϫ2311/2A2-Luc, and Ϫ933/2A2-Luc, as follows. Plasmid Ϫ933/2A2-Luc was constructed by subcloning a 955-bp DNA fragment corresponding to nts Ϫ933 to ϩ22 of the rat CYP2A2 gene (nucleotide numbering relative to the 2A2 transcriptional start site) (accession number M33313; GenBank/EBI Data Bank). The fragment was amplified by PCR using cloned pUC19/2A2 as the template and synthetic oligonucleotides corresponding to nts Ϫ933 to Ϫ910 (upstream primer containing a 5Ј-end BglII restriction site) and ϩ22 to ϩ5 (downstream primer containing a 5Ј-end HindIII restriction site). Likewise, Ϫ2311/2A2-Luc was constructed by generating a 2333-bp DNA fragment spanning nts Ϫ2311 to ϩ22 using oligonucleotides corresponding to nt Ϫ2311 to Ϫ2292 (upstream primer containing a 5Ј-end MluI restriction site) and ϩ22 to ϩ5 (downstream primer containing a 5Ј-end XhoI restriction site). PCRs were carried out at 94°C for 1 min, 58°C for 1 min, and 72°C for 2 min for 30 cycles in a Stratagene RoboCycler. The resultant DNA fragments were double digested with BglII and HindIII (Ϫ933/2A2-Luc) or MluI and XhoI (Ϫ2311/2A2-Luc) and ligated into a similarly digested pGL3-Basic (Promega) cloning vector, which encodes a modified firefly luciferase reporter. Ϫ6232/2A2-Luc was constructed by directly ligating into Ϫ933/2A2-Luc a 5.5-kb fragment excised from pUC19/2A2 by digestion with SalI and EcoRV. All three CYP2A2 promoter-luciferase reporter constructs were verified by DNA sequencing. DNA sequencing revealed that the cloned Ϫ6232/ 2A2 promoter fragment contained 623 bp of previously unsequenced DNA at the 5Ј-end (33).
Reporter plasmids Ϫ2458/4A2-Luc and Ϫ1944/4A2-Luc were constructed as follows. The cloned genomic DNA fragment p246/4A2, obtained from Dr. F. J. Gonzalez (NCI, National Institutes of Health) (34), was used as a template in PCRs to generate CYP4A2 promoter fragments. Upstream primers corresponding to CYP4A2 nts Ϫ2458 to Ϫ2437 and nts Ϫ1944 to Ϫ1920 (containing 5Ј-end MluI restriction sites) (accession number M57719; GenBank/EBI Data Bank) were, respectively, paired with a common downstream primer corresponding to CYP4A2 nts ϩ28 to ϩ6 (containing a 5Ј-end XhoI restriction site). These PCRs generated a 2486-bp PCR product, corresponding to nts Ϫ2458 to ϩ28, and a 1972-bp fragment, corresponding to nts Ϫ1944 to ϩ28 of the rat CYP4A2 gene. The resultant DNA fragments were double digested with MluI and XhoI and ligated into MluI/XhoI-digested pGL3-Basic, generating reporter plasmids Ϫ2458/4A2-Luc and Ϫ1944/4A2-Luc. Correct CYP4A2 promoter-firefly luciferase reporter plasmid construction was confirmed by DNA sequencing.
Cell Culture and Transient Transfection-HepG2 human hepatoma and COS-1 African green monkey kidney cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin. For transient transfections, HepG2 cells were seeded at a density of 1.8 ϫ 10 5 cells/well (in 24-well plates) or 8 ϫ 10 4 cells/well (in 48-well plates). COS-1 cells were seeded at a density of 3 ϫ 10 4 cells/well in 48-well plates. Cells were transfected using FuGENE 6 reagent (Roche Applied Science). Fu-GENE 6-DNA complexes were prepared as described in the manufacturer's protocol at a ratio of 1.3:1 (FuGENE 6:DNA, v/w). Typically,

FIG. 1. CYP promoter activity in transfected HepG2 and COS-1 cells. A and B, HepG2 (left panels) and COS-1 cells (right panels)
were transfected with 20 ng each of the indicated 5Ј-deleted CYP promoter-Luc reporter constructs. Normalized firefly luciferase activities were determined (mean Ϯ S.D., n ϭ 3 separate transfections) and are shown as -fold activation relative to the activity of the empty pGL3-Basic-Luc plasmid transfected in parallel (first bar in each data set). Relative CYP2A2 promoter activities were ϳ7-8-fold higher in HepG2 compared with COS-1 cells (cf. y axis values). Fold activation values are shown above each bar in panel A. Transfections were carried out using equal amounts of DNA for each CYP promoter plasmid. each well of a 48-well tissue culture plate received a total of 350 ng of DNA, including 20 -50 ng of firefly luciferase reporter plasmid: 20 ng of CYP2A2 promoter-Luc, 20 ng of CYP4A2 promoter-Luc, 50 ng 4x-pT109-Luc, 50 ng of Cyp2d9-Luc or 50 ng CYP8B1-Luc. Where indicated, transfections also contained the following expression plasmids: 25 ng of rat GH receptor, 100 ng of STAT5b or STAT5b-Y699F, 40 ng of PGC-1␣, and 100 ng HNF expression plasmid. In all cases, 25 ng of pRL-tk-Luc (Renilla luciferase) reporter plasmid was included as an internal control for transfection efficiency. 24 h after the addition of the FuGENE 6-DNA complex, cells were stimulated for 18 h with 200 ng/ml GH in Dulbecco's modified Eagle's medium without serum or were left untreated. Cell lysates were then prepared using 80 l of 1% SDS gel sample buffer (0.07 M Tris-HCl, pH 6.8, 1% SDS, 10.6% glycerol, 5% 2-mercaptoethanol, and a trace of pyronin Y)/well of a 48-well plate. Samples were boiled for 10 min, and 20 l was loaded directly onto a 7.5% Laemmli SDS gel for Western blot analysis. Firefly and Renilla luciferase activities of samples lysed in 100 l of 1 ϫ lysis buffer (Promega, Madison, WI)/well of a 48-well plate were determined using a dual reporter assay system (Promega) and a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego). Firefly luciferase activity values were divided by Renilla luciferase activity values to obtain normalized luciferase activities (mean Ϯ S.D. values for n ϭ 3 independent transfections). Relative luciferase activities were then calculated to facilitate comparisons between samples within a given experiment.
Western Blotting-HepG2 cell extract (30 g of protein/sample) was electrophoresed for 5 h through 7.5% denaturing polyacrylamide gels and transferred overnight onto nitrocellulose membranes. Membranes were blocked for 1 h at 37°C in blocking solution containing 1% bovine serum albumin and 5% nonfat dry milk and then incubated overnight at 4°C with anti-V5 antibody (diluted 1:5,000 in blocking solution) or with anti-STAT5b-pY694 antibody (1:1,000 dilution). Antibody binding was visualized on x-ray film by enhanced chemiluminescence using the ECL kit from Amersham Biosciences. Nitrocellulose membranes were reprobed, as indicated, after incubation in stripping buffer (62.5 mM Tris-HCl, pH 7.6, 2% SDS, 50 mM 2-mercaptoethanol) for 20 min at 50°C. Membranes were then blocked for 1 h at 37°C in blocking solution containing 2% bovine serum albumin and 2% nonfat dry milk followed by incubation for 1 h at room temperature with anti-STAT5b antibody (diluted 1:2,000 in blocking solution). X-ray films were scanned using a Microtech Scanmaker V6USL scanner (Hauppauge, NY) and ScanWizard version 5.12 scanning software (Microteck, Inc.).
Promoter Analysis and Identification of HNF and STAT5 Binding Sites-The Web-based program PromoSer (biowulf.bu.edu/zlab/ promoser) (35) was used to retrieve DNA sequence information from the proximal promoter regions encompassed by the promoter sequences included in the rat CYP2A2 (6,232 nts) and CYP4A2 (2,458 nts), mouse Cyp2d9 (112 nts), and human CYP8B1 (514 nts) reporter plasmids included in this study. GenBank accession numbers, used by PromoSer to identify the genes of interest, are shown in Table I (see "Results"). The promoter sequences retrieved were evaluated for correct genomic location and orientation using the NCBI Blast program. Proximal promoters were then analyzed using the Web-based program Cluster Buster (zlab.bu.edu/cluster-buster) (36) to identify clustered DNA binding motifs in the 5Ј-flanking DNA based on a set of ϳ80 liver-expressed transcription factors using binding site matrices defined by the Trans-Fac data base (37) Table I.

RESULTS
High Basal Activity of CYP2A2 Promoter-The transcriptional regulation of the class II male CYP2A2 and CYP4A2 promoters was investigated in cells transfected with luciferase reporter plasmids containing up to 6,232 nts of CYP2A2 or 2,458 nts of CYP4A2 5Ј-flanking DNA. These CYP promoter constructs were transfected individually into HepG2 cells, a liver-derived cell line, or into COS-1 cells, which are kidneyderived. The CYP2A2 promoter constructs displayed unusually high basal expression in both cell lines (Fig. 1), with activity significantly higher in HepG2 cells (left panels) than COS-1 cells (right panels). The basal activity of Ϫ6232/2A2-Luc was substantially lower than that of Ϫ2311/2A2-Luc or Ϫ933/2A2-Luc (Fig. 1B), suggesting the presence of an upstream negative regulatory element between Ϫ2311 and Ϫ6232. The Ϫ2548/ 4A2-Luc and Ϫ1944/4A2-Luc promoter constructs both exhibited very low basal expression in HepG2 cells (activity Յ2-fold higher than that of the empty pGL3-Basic control plasmid), similar to the male-specific Ϫ1800/CYP2C11 and female-specific Ϫ1632/CYP2C12 promoters (Fig. 1A).
Liver Transcription Factors HNF4␣ and HNF3␥ Stimulate CYP2A2 Promoter Activity-Next, we investigated the effects of eight individual HNFs on the promoter activities of CYP2A2 and CYP4A2 in HepG2 cotransfection studies. Of the eight HNFs tested (HNF1␣, HNF3␣, HNF3␤, HNF3␥, HNF4␣,   FIG. 4. GH-activated STAT5b does not enhance HNF-stimulated CYP2A2 promoter activity. A, reporter plasmid Ϫ2311/2A2-Luc was transfected into HepG2 cells together with a single expression plasmid (STAT5b, HNF3␥, or HNF4␣), two expression plasmids (STAT5b with HNF3␥ or HNF4␣; and HNF3␥ with HNF4␣), or all three expression plasmids, as indicated, all in the presence of GH receptor expression plasmid. 24 h after transfection, the cells were either treated with 200 ng/ml GH for 16 h or were left untreated. Normalized firefly luciferase activity was determined, and the activity of the reporter in the absence of STAT5b or HNF factor was set at 100 (mean Ϯ S.D., n ϭ 3). B, cell extracts from the experiment shown in A were analyzed on a Western blot probed sequentially with anti-V5 followed by anti-STAT5b antibody as indicated, to verify the expression of STAT5b and each HNF. GH activation of STAT5b is indicated by the appearance of an upper, tyrosine-phosphorylated STAT5b band marked a. Each of the 16 lanes corresponds to the respective bar shown above in panel A. HNF4␣ migrated as a doublet, a major upper band (a) and a weaker lower band (b). C, cell extracts transfected with GH receptor and STAT5b were analyzed on a Western blot probed sequentially with anti-pY694-STAT5b followed by anti-STAT5b antibody as indicated, to confirm that the upper STAT5b band marked a corresponds to the tyrosine-phosphorylated form of STAT5b. D, HepG2 cells were transfected with the STAT5-responsive reporter plasmid 4x-pT109-Luc together with expression plasmids encoding STAT5b and GH receptor. Cells were treated with 200 ng/ml GH for 18 h or were left untreated. Normalized firefly luciferase activities were determined (mean Ϯ S.D., n ϭ 3), and the activity of the reporter in the absence of STAT5b was set at 1. HNF6, and CCAAT/enhancer binding proteins ␣ and ␤), only HNF4␣ and HNF3␥ stimulated CYP2A2 promoter activity, by ϳ2.5-fold each ( Fig. 2A). Cotransfection of HNF4␣ and HNF3␥ generally resulted in a small further increase in activity under conditions where both HNFs were expressed at similar protein levels ( Fig. 2A). None of the HNFs stimulated CYP2A2 promoter activity in COS-1 cells when transfected individually ( Fig. 2B and data not shown), suggesting a requirement for a liver cell factor that is present in HepG2 cells but absent in COS-1 cells. Indeed, CYP2A2 promoter activity was stimulated, by ϳ3-fold, in COS-1 cells after cotransfection of HNF4␣ and HNF3␥ (Fig. 2B). In contrast to CYP2A2, CYP4A2 promoter constructs up to 2.5 kb in length were unresponsive to transfection of any of the HNFs in either cell line (data not shown).
HNF3␤ and HNF6 Inhibit CYP2A2 Promoter Activity-In contrast to the trans-activation seen with HNF4␣ and HNF3␥, the high basal CYP2A2 promoter activity was inhibited by ϳ85% after transfection of HNF3␤ or HNF6 in HepG2 cells. This inhibition was seen with all three CYP2A2-Luc promoter constructs (Fig. 3A). Cotransfection of HNF3␤ and HNF6 resulted in a small further decrease in CYP2A2 promoter activity in the case of Ϫ933/2A2-Luc and Ϫ2311/2A2-Luc, but not Ϫ6232/2A2-Luc. The six other HNFs did not inhibit CYP2A2 promoter activity (data not shown). Furthermore, transfection of either HNF3␤ or HNF6 reversed the stimulation of CYP2A2 promoter activity by HNF4␣ (Fig. 3B), indicating that the inhibitory action of these HNFs can override the trans-activation of this promoter by HNF4␣. HNF6 and HNF3␤ are expressed at a higher level in female compared with male liver, both in rats (28) and in mice (29), suggesting that these two factors work in concert to inhibit expression of the male-specific CYP2A2 in female rat liver. HNF3␤ and HNF6 had no effect on CYP2A2 promoter activity in COS-1 cells, suggesting that the inhibitory effects of HNF3␤ and HNF6 are dependent on interactions with other liver (HepG2) factors (data not shown).
GH-activated STAT5b Does Not Enhance HNF-stimulated CYP2A2 Promoter Activity-STAT5 binding sites are present in the 5Ј-regulatory regions of CYP2C11, CYP2A2, and CYP4A2 (21) ( Table I). These sites are functional in terms of STAT5 DNA binding activity. To determine whether GH-activated STAT5b, acting through these STAT5 binding sites, can trans-activate the male-specific CYPs, the individual CYP2A2 and CYP4A2 luciferase reporter constructs were transfected into HepG2 cells together with expression plasmids coding for GH receptor and STAT5b. Treatment of the cells with GH failed to stimulate CYP2A2 (Fig. 4) or CYP4A2 reporter activity (data not shown), despite the presence of naturally occurring STAT5 binding sites in the Ϫ6232/2A2-Luc and Ϫ2311/2A2-Luc constructs and in both CYP4A2 constructs (Table I). GHactivated STAT5b also failed to stimulate CYP2A2 or CYP4A2 reporter activity in COS-1 cells (data not shown). Western blot analysis verified the expression of HNF3␥, HNF4␣, and STAT5b (Fig. 4B) and the activation of STAT5b after GH treatment (cf. upper STAT5 band in lanes 4, 12, 14, and 16, corresponding to tyrosine-phosphorylated STAT5b). Control experiments verified that the GH-induced upper STAT5b band from Fig. 4B corresponds to the tyrosine-phosphorylated form of STAT5b, as demonstrated by Western blot with antibodies specific to phosphotyrosine 694-STAT5b (Fig. 4C, upper panel,  lane 2) and total STAT5b (lower panel; cf. upper STAT5b band in lane 2 corresponding to the tyrosine-phosphorylated STAT5b). In control experiments, GH-activated STAT5b strongly activated the STAT5-responsive reporter plasmid 4x-pT109-Luc (Fig. 4D). We also examined whether the stimulatory effects of HNF4␣ and HNF3␥ on CYP2A2 promoter activity can be modulated by coexpression of STAT5b. Fig. 4A shows that cotransfection of STAT5b, followed by GH treatment, did not further increase CYP2A2 promoter activity activated by HNF4␣ and/or HNF3␥. Thus, HNF4␣ and HNF3␥ stimulate CYP2A2 expression in a STAT5b-independent manner.
Synergistic Action of HNF4␣ and STAT5b on Cyp2d9 and CYP8B1 Promoters-We next investigated whether HNF4␣ and STAT5b might act in a cooperative manner to regulate two other well established male-specific HNF4␣ target genes, CYP8B1 (31) and Cyp2d9 (32). CYP8B1 promoter activity was stimulated ϳ6-fold by HNF4␣ in HepG2 cells but not in COS-1 cells (Fig. 5A, left panel versus right panel), similar to CYP2A2 (cf. Fig. 2). Transfection of STAT5b alone had no effect on CYP8B1 promoter activity in either cell line; however, STAT5b further increased the transcriptional activity of HNF4␣ toward the CYP8B1 promoter in a synergistic manner (ϳ3.5-fold fur- ther increase in HepG2 cells and 25-fold increase in COS-1 cells after GH treatment) (Fig. 5A). Cyp2d9 promoter activity was stimulated by HNF4␣, both in HepG2 cells (ϳ3.5-fold increase) and COS-1 cells (5-fold) (Fig. 5B). Although STAT5b alone had no effect on Cyp2d9 promoter activity, HNF4␣-stimulated Cyp2d9 promoter activity was further increased ϳ2-fold upon cotransfection of STAT5b, in both cell lines. The enhancement of HNF4␣-stimulated reporter activity was abolished when STAT5b was replaced by STAT5b-Y699F, where the site of GH-stimulated tyrosine phosphorylation is mutated to phenylalanine (Fig. 5C).
Coactivation of HNF4␣-responsive CYP2A2, CYP8B1, and Cyp2d9 Promoters by PGC-1␣-Next we investigated the effect of PGC-1␣, a strong HNF4␣ coactivator (38), on the ability of STAT5b to enhance HNF4␣-mediated transcription from the CYP2A2, CYP8B1, and Cyp2d9 promoters. First, we established that PGC-1␣ enhanced HNF4␣-mediated trans-activation of CYP2A2 (1.8-fold further increase), CYP8B1 (18-fold), and Cyp2d9 (2.4-fold) in transfected HepG2 cells (Fig. 6, A-C,  left panels). PGC-1␣ also enhanced HNF4␣-mediated transactivation of CYP2A2 (4-fold), CYP8B1 (54-fold), and Cyp2d9 (2.2-fold) in COS-1 cells (Fig. 6, A-C, right panels). The transactivation of each promoter by PGC-1␣ alone seen in HepG2 but not COS-1 cells is likely to reflect the coactivation of the endogenous HNF4␣ present in HepG2 cells. PGC-1␣ substantially increased the synergistic effect of STAT5b on HNF4␣stimulated CYP8B1 promoter activity (9-fold further increase; Fig. 7B). No such enhancement was observed in the case of CYP2A2 (Fig. 7A). The corresponding PGC-1␣ stimulatory effect was much smaller with Cyp2d9 (Fig. 7C). DISCUSSION The expression of hepatic CYP genes shows sex differences in both rodents and humans. The sexual dimorphism of liver CYP gene expression is dictated by the temporal pattern of plasma GH stimulation, which is intermittent and highly pulsatile in males and nearly continuous in females. Previous studies on the cellular and molecular mechanisms whereby GH and its sexually dimorphic plasma profiles regulate the expression of male-specific liver CYPs have primarily focused on rat CYP2C11 (21, 39), a CYP gene that requires intermittent stimulation by male plasma GH pulses for full male expression. Here, we investigated the mechanism whereby GH regulates two other male-specific rat CYP genes, CYP2A2 and CYP4A2. These CYPs represent a distinct male-specific class, whose expression in females is markedly suppressed by the female plasma GH pattern, but whose expression in males does not require stimulation by plasma GH pulses. Our findings provide important new insight into the regulatory complexity that governs GH action in the liver with the identification of liverenriched transcription factors that control the male-specific expression of CYP2A2 through both positive (HNF3␥, HNF4␣) and negative regulatory mechanisms (HNF3␤, HNF6) and are themselves subject to sex-dependent GH regulation.
CYP2A2 promoter sequences 0.9 -6.2 kb long exhibited an unusually high basal promoter activity when transfected into the liver-derived cell line HepG2. CYP2A2 promoter activity was significantly higher in HepG2 cells than in the kidneyderived cell line COS-1, suggesting that endogenous liver (HepG2) factors, absent in COS-1 cells, are required for high CYP2A2 expression. By contrast, the CYP4A2 promoter displayed very low basal expression in both cell lines, as did CYP2C11 and CYP2C12 (Fig. 1). Moreover, the CYP4A2 promoter was unresponsive to all eight HNFs tested (data not shown) consistent with the absence of HNF binding sites within the transcription factor clusters identified in the 2.5-kb promoter segment used in our studies (Table I).
The high basal CYP2A2 promoter activity seen in HepG2 cells grown in GH-free culture medium is reminiscent of the high level of CYP2A2 expression seen in hypophysectomized male and female rat liver (40), where pituitary hormone stimulation and HNF6 expression (28) are both ablated. The cultured liver cell model may thus approximate the environment of the pituitary hormone-free hypophysectomized rat liver, in particular with respect to the absence of HNF6, which together with HNF3␤ is presently shown to be a strong negative regulator of the CYP2A2 promoter. Thus, the nearly complete loss of HNF6 mRNA in livers of hypophysectomized female rats (28) probably contributes to the associated strong up-regulation of CYP2A2 (40). HNF6 and HNF3␤ are both expressed at higher levels in female compared with male liver (28,29). Moreover, HNF3␤ and HNF6 RNAs are both induced to female levels in livers of male rats infused with GH in a continuous (femalelike) manner, 2 a condition under which CYP2A2 expression is strongly suppressed (10,41). These same two HNFs are also strong trans-activators of the female-specific CYP2C12 (27). Thus, HNF3␤ and HNF6 are GH-regulated feminization factors that can contribute to the sexually dimorphic expression of liver CYPs by positively regulating CYP2C12 while concomitantly inhibiting expression of CYP2A2 in female liver (Fig. 8).
Studies of a liver HNF4␣-deficient mouse model (42) demonstrate an essential role for HNF4␣ in regulating several sexually dimorphic Cyps and liver-enriched transcription factors in liver in vivo (29). HNF4␣ was shown to contribute to sex-dependent mouse liver Cyp expression by positive regulation, in males, of certain male-specific Cyps and by the concomitant inhibition of female-specific Cyps and the female-predominant HNF3␤ and HNF6 (29). In the present study, HNF4␣ was shown to trans-activate the CYP2A2 promoter. Regulation of CYP2A2 by HNF4␣ may require interactions with other HNFs, such as HNF3␥, which also stimulated CYP2A2 promoter activity. Indeed, in COS-1 cells, which are deficient in HNF4␣, trans-activation of the CYP2A2 promoter was only seen when HNF4␣ and HNF3␥ were introduced in combination. Moreover, the modest stimulation of the CYP2A2 promoter by HNF4␣ in HepG2 cells (Fig. 2) may reflect the fact that HNF4␣ is already present endogenously in these cells. Finally, the HNF4␣ coactivator PGC-1␣ enhanced the trans-activation of CYP2A2 by HNF4␣, suggesting that this coactivator may be required for the high CYP2A2 expression seen in adult male liver.
The hypothesis that HNF4␣ serves as an important regulator of male-specific liver genes (30), including CYP2A2, is supported by the finding that liver nuclear HNF4␣ protein and DNA binding activity are elevated in male compared with female rat liver and are suppressed by continuous GH treatment of the males. 2 This hypothesis is supported by the finding that HNF4␣ expression is maintained in hypophysectomized male rat liver (43), where CYP2A2 is expressed at a high level (40). Liver-specific Hnf4␣ disruption increases hepatic HNF3␤ and HNF6 RNA levels to female levels in male mice, but has no effect in females, indicating that HNF4␣ imparts negative regulation to these HNF genes in a male-specific manner (29). Conceivably, these sex-dependent effects of HNF4␣ on HNF3␤ and HNF6 expression could involve the action of GH and its sex-dependent plasma hormone profile, which in turn may dictate the lower level expression of these latter HNFs in male liver (Fig. 8). Accordingly, the high expression of CYP2A2 in male liver is proposed to result from the stimulatory effect of HNF4␣ and its male-predominant nuclear DNA-binding activity, in combination with the reduced expression of the inhibitory factors HNF3␤ and HNF6 in male compared with female rat liver, as noted above.
The transcription factor STAT5b is uniquely responsive to the male pulsatile plasma GH pattern and is proposed to be a key mediator of the sexually dimorphic response of liver CYPs to GH (13). STAT5b, which is tyrosine-phosphorylated and translocates rapidly to the nucleus in response to male plasma GH pulses (14), binds to its consensus sequence in the promoter region of STAT5b-responsive genes and activates transcription. By contrast, the more continuous female plasma GH pattern results in low levels of active STAT5b and is proposed to lead to low level expression of STAT5b target genes. The CYP2A2 and CYP4A2 promoters were presently found to be unresponsive to GH-activated STAT5b, despite the presence of strong consensus STAT5b binding sites (21) (Table I). This finding suggests that these class II, GH pulse-independent male liver CYP genes are not regulated directly by the GH pulse-activated STAT5b. To investigate further this question, we determined whether the unresponsiveness of CYP2A2 to STAT5b reflects a requirement for the collaborative participation of GH-regulated HNFs to achieve the robust male CYP transcriptional responses seen in liver in vivo. No such cooperation between STAT5b and either HNF4␣ or HNF3␥ could be detected. However, a strong, positive cooperative interaction between STAT5b and HNF4␣ was seen in the case of another male-specific gene, CYP8B1 (Fig. 5). Moreover, in the case of the female-specific CYP2C12, GH-activated STAT5b inhibits the synergistic trans-activation of that promoter by HNF6 and HNF3␤ (27). Although the absence of a cooperative interaction between HNF4␣ and STAT5b on the CYP2A2 promoter is consistent with the GH pulse-independence of this class II male CYP gene, it is alternatively possible that the CYP2A2 promoter constructs studied (extending to Ϫ6.2 kb) are missing a critical far upstream STAT5 binding site. One such potential STAT5b binding site is present at CYP2A2 nts Ϫ11269 to Ϫ11261 (data not shown).
CYP8B1, and to a lesser extent Cyp2d9, was trans-activated by HNF4␣, in agreement with earlier reports in the case of CYP8B1 (31,44) and consistent with the substantial downregulation of both genes seen in HNF4␣-deficient mouse liver (29). CYP8B1 (29) and Cyp2d9 (6) are both more highly expressed in male than in female liver, with a requirement for 2 C. A. Wiwi and D. J. Waxman, unpublished experiments.

FIG. 8. Model for sex-dependent regulation of CYPs by HNFs.
Activation of the male-specific CYP genes 8B1 and 2d9 is proposed to occur via the concerted actions of STAT5b and the nuclear receptor HNF4␣. HNF4␣ and HNF3␥ both transcriptionally activate the class II male-specific CYP2A2. Regulation of the female-specific CYP2C12 is proposed to involve the synergistic action of HNF6 and HNF3␤ (27), which are both elevated in expression in female liver in response to the nearly continuous plasma GH pattern and are presently shown to inhibit CYP2A2 expression. Expression of HNF6 and HNF3␤ is inhibited by HNF4␣ in males (29), which may contribute to the downregulation of female-specific CYP expression in male liver. For further details, see "Discussion." STAT5b, at least in the case of CYP2d9 (17). The stimulation of CYP8B1 and Cyp2d9 promoter activity by HNF4␣ was increased substantially by the HNF4␣ coactivator PGC-1␣. In the case of Cyp2d9, stimulation by HNF4␣ most likely occurs through a cryptic HNF4␣ binding site, as no HNF4␣ binding motifs were identified using the TransFac data base (Table I). The responsiveness of these promoters to HNF4␣ was also enhanced by STAT5b, despite the absence of consensus STAT5b binding sites in the promoter sequences examined. Conceivably, STAT5b may synergize with HNF4␣ by binding to a nonconsensus STAT5 binding site, such as the ones identified in other studies (45). Alternatively, the synergistic activation of these genes by HNF4␣ and STAT5b might not require direct DNA binding by STAT5b. To test this hypothesis, we probed for direct interactions between GH-activated STAT5b and HNF4␣ by anti-STAT5b supershift analysis of an HNF4 electrophoretic mobility shift analysis complex, however, no interactions were detected. 2 Admittedly, our observation that HNF4␣ and STAT5b synergistically trans-activate the CYP8B1 and Cyp2d9 promoters was made under conditions where both factors are overexpressed. As such, we acknowledge that the transfection system used in these studies limits the extrapability of our findings to the actual physiological state. Further study will be required to clarify the mechanism for the synergistic interactions between HNF4␣ and STAT5b seen on some HNF4␣ target promoters and to extend these results to rat and mouse liver in vivo.
The stimulatory effect of STAT5b on HNF4␣-dependent transcription was independent of (Cyp2d9) or only partially dependent on GH stimulation (CYP8B1) (Figs. 5 and 7). This finding was unexpected, in view of the requirement for GH to induce STAT5b tyrosine phosphorylation and nuclear translocation. This GH-independent effect of STAT5b may be the result of a low level, GH-independent STAT5b tyrosine phosphorylation in our cellular model. This possibility is consistent with the inactivity of the STAT5b tyrosine phosphorylation site mutant Y699F (Fig. 5C). Alternatively, the present findings may reflect an elevated basal nuclear level of non-tyrosinephosphorylated STAT protein, which has been shown to occur when STATs translocate to the nucleus in association with other nuclear-targeted DNA-binding proteins (46).
In conclusion, the present findings establish that the malespecific CYP2A2 is responsive to multiple HNFs, several of which are subject to sex-dependent expression and GH regulation. HNF4␣ and HNF3␥ are proposed to serve as masculinizing factors that stimulate CYP2A2 expression, whereas HNF3␤ and HNF6 serve as feminizing factors that inhibit CYP2A2 expression in female liver while concomitantly stimulating CYP2C12 expression. Finally, the male GH pulse-activated STAT5b was shown to synergize with the male-predominant, HNF4␣-responsive Cyp2d9 and CYP8B1 promoters, but not with the male GH pulse-independent CYP2A2, demonstrating that STAT5b can cooperate with HNF4␣ to regulate a subset of male-specific liver CYP genes.