Inhibitory Cross-talk between STAT5b and Liver Nuclear Factor HNF3β

STAT5b is repeatedly activated in rodent liver by the male pattern of intermittent plasma growth hormone (GH) stimulation and is required to maintain the GH pulse-regulated, male-specific pattern of liver gene expression. We presently investigate the interactions between STAT5b and hepatocyte-enriched nuclear factors (HNFs) that contribute to regulation of GH pulse-inducible, male-specific liver cytochrome P-450 (CYP) genes. STAT5 binding sites were identified in the 5′-flank of the adult male-expressed genesCYP2A2 (nucleotides −2255 to −2247), CYP4A2(nucleotides −1872 to −1864), and CYP2C11 (nucleotides −1150 to −1142). STAT5-DNA complexes were formed by eachCYP sequence with nuclear extract from GH pulse-activated male, but not female, rat liver. TheCYP2C11 STAT5 site, which is flanked by HNF3 consensus sequences, conferred STAT5b-inducible reporter gene activity in GH-treated HepG2 cells. trans-Activation of the intactCYP2C11 promoter (1.8-kilobase 5′-flank) was strongly induced by the liver nuclear factors HNF1α and HNF3β but, unexpectedly, was inhibited by GH-activated STAT5b. This STAT5b inhibitory effect could be reversed by HNF1α and reflects a functional antagonism between STAT5b and HNF3β, as evidenced by the inhibition of HNF3β DNA binding and transcriptional activity by STAT5b. HNF3β, in turn, inhibited STAT5b by a novel mechanism that leads to suppression of GH-inducible STAT5b tyrosine phosphorylation, DNA binding activity, and transcriptional activity. The potential for GH-activated STAT5b to stimulate male-specific liver CYPexpression can thus be modulated by HNF3β, highlighting the complex interrelationship between STAT5b and liver transcription factors controlling expression of GH-regulated CYPgenes.

Growth hormone (GH) 1 is secreted by the pituitary gland in a sex-dependent manner in both rodents (1) and humans (2). In male rats, GH is released into circulation every ϳ3-3.5 h, giving a repeated, pulsatile plasma hormone pattern (3), which contrasts to the more frequent, nearly continuous profile of pituitary GH secretion seen in adult females. These sexually dimorphic plasma GH patterns directly regulate the sexually dimorphic pattern of liver gene expression, in particular, liver cytochrome P-450 (CYP) gene expression (4,5). Well studied examples of sex-specific, GH-regulated liver CYPs include the male-specific androgen 16␣and 2␣-hydroxylase CYP2C11, which is strongly induced at puberty in male but not female rat liver, and the steroid sulfate 15␤-hydroxylase CYP2C12, which is exclusively expressed in adult female rat liver (6,7). The sexual dimorphism of liver with respect to expression of these and other sex-dependent liver CYPs, including CYP2A2 and CYP4A2 (both male-specific in their expression), is regulated by GH at the level of transcription initiation (8,9). GH signaling is initiated by GH receptor dimerization at the cell surface, leading to activation of the GH receptor-associated tyrosine kinase JAK2. JAK2, in turn, activates multiple intracellular signaling proteins (10 -12). One such protein, the transcription factor STAT5b, displays a unique GH pulse responsiveness in rat liver (13). The intermittent stimulation of liver cells by successive plasma GH pulses triggers a rapid and repeated tyrosine phosphorylation and nuclear translocation of STAT5b in adult male rats. By contrast, in adult female rats, the more continuous pattern of GH exposure down-regulates GH receptor-JAK2 activity, leading to a low steady-state level of the transcriptionally active, nuclear STAT5b (14 -16). A close correlation between STAT5b nuclear translocation and sex-specific liver gene expression is also seen in mouse liver (17). Targeted gene disruption studies establish that STAT5b, but not the closely related (Ͼ90% identical) STAT5a, plays a critical role in the physiological, sex-dependent responses of the liver to GH (18). Thus, STAT5b-deficient male mice exhibit a loss of sexually dimorphic liver CYP gene expression in addition to a loss of pulsatile GH-stimulated pubertal growth (19 -21). By contrast, STAT5a is a principal and obligate mediator of mammopoietic and lactogenic signaling stimulated by prolactin in the mammary gland (22).
Although the above studies demonstrate an important role for STAT5b in maintaining the male-specific pattern of liver gene expression, the molecular details of the gene regulatory mechanisms whereby STAT5b contributes to the sexual dimorphism of the liver have remained elusive. In the case of the female-specific CYP2C12 gene, GH-activated STAT5b is proposed to interfere with the strong, synergistic trans-activation of promoter activity effected by two liver-enriched transcription factors, HNF6 and HNF3␤ (23). Additional mechanisms contributing to the GH-regulated, sexual dimorphic expression of CYP2C12 have also been proposed (24 -27). In the case of the male-expressed CYP2C11, an initial promoter analysis revealed the presence of two negative regulatory regions ("silenc-ers") (28), but the identification of other functional elements and the potential role of STAT5b and specific liver transcription factors in regulating either basal on hormone-dependent transcription of this gene have not been described.
Several liver transcription factors contribute to the liverspecific expression of hepatic CYPs. These include the variant homeodomain protein HNF1, CCAAT/enhancer-binding proteins (C/EBPs), the winged helix factor HNF3, the nuclear receptor HNF4, and the one-cut homeoprotein HNF6 (29). These liver transcription factors function in unique combinations, often synergistically, to activate liver-expressed genes via a complex array of interactions. For example, HNF6 expression is stimulated by HNF4 and inhibited by C/EBP␣ (30), whereas HNF3␤ positively regulates the expression of HNF4 and HNF1␣ and their downstream targets (31). Other studies have shown that the HNF6 gene can be regulated by GH, resulting in its sex-dependent expression by a mechanism involving STAT5b and HNF4 (32). As noted above, GH-dependent and liver-specific expression of CYP2C12 in adult female rats is proposed to reflect cooperative regulation by HNF3␤ and HNF6 (23) as well as by HNF4 and HNF6 (27). Although the general importance of these transcription factor cascades in liver-specific gene expression is well established, only limited information is available regarding their particular role in the GH-regulation of sex-specific liver CYP gene transcription. Unique combinations of factors and interactions between GHregulated STAT and liver-expressed regulators appear likely and may be required to establish and maintain the liver-specific and sexually dimorphic profiles of CYP gene expression.
The present study investigates the influence of GH-activated STAT5b on the expression of CYP2C11 and its regulation by liver-enriched transcription factors. Our findings demonstrate that 2C11 gene expression is subject to regulation by GHactivated STAT5b in a manner that is modulated by two liver transcription factors that trans-activate the 2C11 promoter, HNF3␤ and HNF1␣. Moreover, novel inhibitory cross-talk between HNF3␤ and GH-activated STAT5b is described. The implications of these findings are discussed in the context of current models for the sex-dependent regulation of CYP2C11 expression by plasma GH pulses.
Expression 2C11 Promoter Plasmids-Segments of the 2C11 5Ј-flank were prepared by polymerase chain reaction amplification using Fischer 344 rat genomic DNA as template and then subcloned into the SacI and XhoI sites of the promoterless luciferase reporter plasmid pGL3-basic (Promega) with the assistance of D. Ding of this laboratory. Polymerase chain reactions were carried out for 30 cycles consisting of 94°C for 1 min, 48°C for 1 min, and 72°C for 2 min. Sense primers were as follows: 5Ј-GGC ATA AAG TGG TGG AT-3Ј (nts Ϫ1769 to Ϫ1753), 5Ј-GGA GGT GCC TGT TCT GG-3Ј (nts Ϫ1208 to Ϫ1194), 5Ј-GTC ACT TCA GAG GTT-3Ј (nts Ϫ1182 to Ϫ1168), 5Ј-GTT ATT CCC GCA TTC TC-3Ј (nts Ϫ968 to Ϫ952), and 5Ј-GGG GGT GCC TTA GTT GG-3Ј (nts Ϫ633 to Ϫ617). Sense primers were paired with a common antisense primer, 5Ј-GCA GCC TTC CTC AGG GAG-3Ј (nts ϩ22 to ϩ5), to generate five corresponding 2C11 promoter reporter constructs: Ϫ1769/ 2C11-Luc, Ϫ1208/2C11-Luc, Ϫ1182/2C11-Luc, Ϫ968/2C11-Luc, and Ϫ633/2C11-Luc. Ϫ1533/2C11-Luc was generated from Ϫ1769/2C11-Luc by digestion with SacI and SacII, followed by filling in and religation. Sequence analysis of the 2C11 promoter constructs revealed the presence of only one of the two GCTA repeats at nts Ϫ47 to Ϫ40 reported in GenBank TM (accession number XB79081). This sequence was present in amplified genomic DNA from three independent Fischer 344 rat genomic DNA polymerase chain reactions. 2C11 promoter nucleotide positions were numbered according to GenBank TM XB79081 after removing the four nucleotides corresponding to the extra GCTA sequence.
Luciferase reporters containing a single copy of the wild-type or mutated 2C11 STAT site (STAT5/2C11-Luc and STAT5mut/2C11-Luc, respectively) were prepared as follows. Complementary oligonucleotides corresponding to wild-type or mutated 2C11 nts Ϫ1159 to Ϫ1138 (see Fig. 2A) and containing XbaI and BglII site adapters were subcloned into the corresponding restriction sites of the luciferase reporter pGL3 promoter. A reporter plasmid containing two tandem copies of the 2C11 STAT5 site subcloned into pGL3 promoter 2x STAT5/2C11-Luc was kindly provided by Dr. Y. Jounaidi of this laboratory. Wild-type and mutated sequences were verified by DNA sequence analysis.
Animal Treatments and Nuclear Extract Preparation-Untreated adult Fischer 344 rats and hypophysectomized rats treated with rat GH, rat prolactin, or lipopolysaccharide were described earlier (13). Nuclear extracts were prepared from freshly excised liver tissue using standard methods and stored frozen at Ϫ80°C (33).
Cell Culture and Transient Transfection-HepG2 human hepatoma and COS-1 cells were maintained in Dulbecco's modified essential 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 1.3 ϫ 10 5 cells/well in 24-well plates. The cells were transfected with FuGENE 6 TM reagent (Roche Molecular Biochemicals). FuGENE 6-DNA complexes were prepared as described in the manufacturer's protocol at a ratio of 1.3 to 1 (FuGENE 6:DNA, v/w). Typically, each well of a 24-well tissue culture plate received a total of 800 ng of DNA: 50 -200 ng of luciferase reporter plasmid (200 ng of 2C11 promoter-Luc or 50 ng of either STAT5/2C11-Luc, 4x-pT109-Luc, or (6x-HNF3)-Cdx-2-Luc), 100 ng of GH receptor, 200 ng of STAT5b, and 200 ng HNF expression plasmid. pRL-tk-Luc plasmid (Renilla luciferase; 50 ng) was included in all transfections as an internal control for transfection efficiency. 24 h after transfection, the cells either were treated with 200 ng/ml rat GH for 18 -20 h or were left untreated. Cell lysates were prepared by shaking the cells in 150 l of 1ϫ Promega lysis buffer for 10 min at room temperature. Firefly and Renilla luciferase activities were measured using a dual reporter assay system (Promega) and a Monolight 2010 luminometer (Analytical Luminesence Laboratory, San Diego, CA). Firefly luciferase activity values were divided by Renilla luciferase activity values to obtain normalized luciferase activities (mean Ϯ SD values for n ϭ 3 separate transfections, unless indicated otherwise). Relative luciferase activities were then calculated to facilitate comparisons between samples within a given experiment. Individual group comparisons were examined for statistical significance using the two-tailed Student's t test (p Ͻ 0.05).
Cell extracts were prepared for EMSA and Western blot analysis as follows. COS-1 and HepG2 cells were seeded on 35-mm tissue culture dishes at ϳ60% confluency and incubated overnight. A total of 3.2 g of plasmid DNA containing GH receptor, STAT5b, and/or HNF3␤ expression plasmid and Renilla luciferase was transfected using FuGENE 6, as specified in each figure legend. 36 h later, the cells were either stimulated with 200 ng/ml GH for 30 min or were left untreated. The cells were washed twice with ice-cold phosphate-buffered saline and scraped with 100 l of 1ϫ Promega lysis buffer containing 1 mM sodium orthovanadate, 5 g/ml aprotinin, 5 g/ml leupeptin, 2 g/ml antipain, and 0.1 mM phenylmethylsulfonyl fluoride. The cell extracts were incubated in a cold room for 30 min with shaking and then centrifuged for 30 min at 15,000 ϫ g. The supernatants were snap-frozen in liquid N 2 and stored at Ϫ80°C. Protein concentrations were determined with Bio-Rad Dc detergent protein assay kit with bovine serum albumin as a standard.
EMSA Assays-Cell extract (10 g) was incubated for 10 min in EMSA reaction buffer, consisting of 3% glycerol, 700 M MgCl 2 , 350 M EDTA, 350 mM dithiothreitol, 7 mM Tris-HCl, pH 7.5, and 2 g of poly(dI-dC) (Roche Molecular Biochemicals). 10 fmol of 32 P-end-labeled, double-stranded oligonucleotide probe (ϳ30,000 cpm) was then added and incubated for 20 min at room temperature then 10 min on ice, followed by the addition of 2 l of loading buffer (30% glycerol, 0.25% bromphenol blue, 0.25% xylene cyanol). For STAT5b supershift analysis, rabbit polyclonal anti-STAT5b antibody was added 10 min after addition of the 32 P-labeled DNA probe, followed by an additional 10-min incubation. For HNF3␤ supershift/complex disruption analysis, goat polyclonal anti-HNF3␤ antibody was incubated with the sample for 10 min prior to addition of the labeled DNA probe. The samples were electrophoresed for 3-4 h at 4°C through a 5.5% nondenaturing gel in 0.5ϫ TBE buffer (for the 2C11 STAT5 site probe; see below) or in 0.25ϫ TBE buffer (for the 2C11 HNF3-STAT5 probe) following a 30-min pre-electrophoresis step. The gels were exposed to PhosphorImager plates overnight and visualized using a Molecular Dynamics Phosphor-Imager and ImageQuant software (Sunnyvale, CA).
EMSA analysis was performed using the following DNA probes: STAT5 site probes from rat CYP2C11, CYP2A2, and CYP4A2 genes, as listed in Fig. 1A; the ␤-casein STAT5 probe used earlier (13); rat CYP2C11 HNF3-STAT5 site, nts Ϫ1174 to Ϫ1138 (sense: 5Ј-(g)aga ggt taa tta aat gca aac aat TTC CAT GAA aaaa-3Ј; antisense: 5Ј-(g)tttt TTC ATG GAA aat gtt tgc att taa acc tct-3Ј); and mutated HNF3-STAT5 probe. Mutations were introduced either in the HNF3 binding site (aa tgc aaa catt to aa tga att gatt-3Ј) or the STAT5 binding site (TTC CAT GAA to Tat CAT GAA) of the wild-type HNF3-STAT5 probe. Nucleotides corresponding to a STAT5 consensus site are shown as capital letters; those corresponding to an HNF3 consensus site are shown in bold type, with the mutated residues underlined.
Western Blotting and Immunoprecipitation-Cell extracts (20 g) were electrophoresed on 7.5% Laemmli SDS gels, electrotransferred onto nitrocellulose membranes, and then probed with anti-Tyr(P) 699 STAT5b antibody, as described in the manufacturer's protocol, or with anti-STATb5 antibody. For immunoprecipitation with anti-STAT5b antibody, cell extract (70 g) was preimmune-cleared for 1 h at 4°C in a total volume of 200 l of IP buffer (10 mM Tris-HCl, pH 7.4, 1% Triton X-100, 0.5% Nonidet P-40, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.2 mM Na 3 VO 4 , 0.2 mM phenylmethylsulfonyl fluoride) containing 20 l of 50% protein A-Sepharose beads and 1 g of rabbit anti-mouse IgG. Protein A-Sepharose beads were removed by centrifugation, and anti-STATb5 antiserum was added to the precleared cell lysate and incubated for 3 h on ice. Immune complexes were collected by centrifugation after a further 1 h of incubation with 20 l of 50% protein A-Sepharose beads at 4°C, washed three times with 300 l IP buffer, and resuspended in 30 l of 1.5ϫ SDS gel sample buffer. The samples were then analyzed on Western blots probed with anti-phosphotyrosine antibody 4G10 using blocking and probing conditions described earlier (13). Antibody binding was visualized on x-ray film by enhanced chemiluminescence using the ECL kit from Amersham Pharmacia Biotech. To reprobe with rabbit polyclonal HNF3␤ antibody, nitrocellulose membranes were heated 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 blocked in Solution I (0.3% Tween 20 in 1ϫ phosphate-buffered saline) containing 1% bovine serum albumin and 1% nonfat dry milk for 1 h at 37°C and then incubated overnight at 4°C with anti-HNF3␤ serum (diluted 1:4000 dilution in blocking solution). The results are presented in figures prepared from gray scale scans of portions of the x-ray films of each blot. Scans were obtained using a Cannon IX-4015 scanner and Ofoto scanning software.

RESULTS
Occurrence of STAT5 Sites in 5Ј-Flanking DNA of Malespecific, GH-regulated CYP Genes-Computer analysis revealed the presence of STAT5 consensus sites (TTC NNN GAA) in the 5Ј-flanking DNA of CYP genes 2A2, 2C11, and 4A2 (Fig.  1A). Each of these CYPs is expressed in rat liver in a malespecific manner and is regulated by the temporal pattern of A, STAT5 consensus core sequences (capital letters) located in the promoters of CYP2A2, CYP4A2, and CYP2C11 at the indicated positions are shown in the context of the complete sequences of the EMSA probes used in B. B, EMSA analysis using liver nuclear extract (NE) prepared from intact or hypophysectomized (Hx) male (M) and female (F) rats, as indicated. Hypophysectomized rats were treated with GH, lipopolysaccharide (LPS), or prolactin (PRL) as noted. Male nuclear extracts were prepared from intact rats killed during a GH pulse secretory period, when STAT5b is active and nuclear (lanes 1, 4, 14, 17, and 22-25); from rats killed between plasma GH pulses, when STAT5b is cytoplasmic and inactive (lanes 6 and 19); or from rats killed late in a GH pulse, when the nuclear STAT5b pool has been partially deactivated by dephosphorylation (lanes 5 and 18). EMSA samples shown in lanes 10 -12 used extracts of GH-treated COS-1 cells that were untransfected (lane 10) or were transfected with GH receptor and either STAT5a or SAT5b (lanes 11 and 12). EMSA analysis was carried out in the presence of polyclonal anti-STAT5b antibody (1 l) to visualize the STAT5-dependent DNA binding activity as a supershifted band (lanes 3, 16, and 24 -32) or in the presence of a 100-fold excess of unlabeled probe (cc, cold competition) to confirm the specificity of the EMSA complex (lanes 2 and 15). pulsatile plasma GH stimulation (see the introduction). No STAT5 consensus sites were found in the 5Ј-flank of numerous other rat CYP genes, including CYPs 1A1, 1A2, 2A1, 2B2, 2C6, 2E1, and 4A1, which are not subject to male-specific, GH pulse regulation (data not shown). To ascertain whether the CYP STAT5 sites are functional in STAT5 binding, EMSA analyses were carried out using liver nuclear extracts prepared from adult male rats killed at the time of a plasma GH pulse when STAT5b is active and nuclear (15). Fig. 1B shows the formation of a single DNA-protein complex by the CYP2A2 (lanes 1 and 4) and CYP4A2 STAT5 site probes (lanes 14 and 17). These complexes were strongly competed by unlabeled probe (lanes 2 and 15) and were supershifted by anti-STAT5b antibody (lanes 3 and 16). Little or no specific complex was formed with liver nuclear extracts prepared from female livers (lanes 7-9 and lanes 20 and 21) or from livers of male rats killed between GH pulses (lanes 6 and 19) when STAT5b is primarily cytoplasmic and inactive (15). Protein-DNA complexes of the same mobility were formed by male liver nuclear extracts incubated with an established ␤-casein STAT5 probe and with extracts of GHtreated COS-1 cells transfected with STAT5b and GH receptor (lanes 13 and 12, respectively, and data not shown). These complexes were distinguished from the complex formed by STAT5a, which migrates more slowly (e.g. lane 11 versus lanes 1, 4, and 12). STAT5a is much less abundant than STAT5b in liver tissue (20).
In contrast to the discrete STAT5b-containing complex formed by the CYP2A2 and CYP4A2 STAT5 site probes, the CYP2C11 STAT5 site probe formed a diffuse protein-DNA complex when incubated with male rat liver nuclear extract (  and 32), neither of which activates liver STAT5 (13). Mutation of the core STAT5 binding site abolished STAT5 binding to the 2C11 promoter probe, whereas mutation of the adjacent upstream sequence was without effect ( Fig. 2A).
To determine whether the STAT5 site is functional in mediating GH-stimulated reporter gene activity, luciferase reporter constructs driven by the 2C11 STAT5 site were transfected into HepG2 cells together with expression plasmids for GH receptor and STAT5b. GH stimulated a 1.8-fold increase in luciferase reporter activity driven by the wild-type 2C11 STAT5 site but not when the STAT5 core sequence contained a mutated STAT5 site (Fig. 2B, left panel). This modest increase was observed in three independent experiments but did not reach statistical significance. However, GH-stimulated reporter activity was significantly increased, by ϳ3-fold, by adding a second copy of the isolated 2C11 STAT5 site. GH had no effect in cells transfected with the inactive STAT5b tyrosine phosphorylation site mutant, STAT5b-Y699F (Fig. 2B, right panel). We conclude that GH-activated STAT5b binds to the STAT5 site of 2C11 in a functional, transcriptionally active manner.
GH-activated STAT5b Inhibits trans-Activation of Intact 2C11 Promoter-To further characterize the GH dependence of 2C11 transcriptional activity, we prepared six luciferase reporter constructs containing various lengths of 2C11 promoter sequence, ranging from 1769 to 633 nts of 5Ј-flanking DNA and extending to nt ϩ22 relative to the transcription start site. These constructs were individually transfected into HepG2 cells together with GH receptor and STAT5b to establish a robust GH signaling pathway. Each 2C11 promoter construct exhibited significant basal expression in the absence of GH treatment, ranging from 3-to 15-fold higher than the empty pGL3-basic reporter plasmid (Fig. 3). A strong decrease in basal promoter activity was observed with Ϫ1208/2C11-Luc compared with Ϫ1182/2C11-Luc, indicating the presence of a negative regulatory element, or silencer, between nts Ϫ1208 and Ϫ1182, supporting a previous report of a silencer between nts Ϫ1226 and Ϫ1184 (28). Unexpectedly, 2C11 promoter activity was decreased by 50 -80% by GH treatment in constructs containing at least 968 nts of promoter sequence (Fig. 3). The GH-dependent inhibition of Ϫ968/2C11-Luc transcription demonstrates that the STAT5 site at nts Ϫ1150 to Ϫ1142 (Fig. 1A) is not required for this inhibition. No GH-dependent inhibition of promoter activity was observed in cells transfected with the STAT5b tyrosine phosphorylation site mutant, STAT5b-Y699F (data not shown).
Transcription Activation of 2C11 Promoter by HNF1␣ and HNF3␤-We next investigated whether the inhibitory effect of STAT5b on 2C11 promoter activity can be modulated by coexpression of one or more liver-enriched transcription factors. An initial screen of the effects of the nine liver factors (HNF1␣, HNF1␤, HNF3␣, HNF3␤, HNF4, HNF6, C/EBP␣, C/EBP␤, and DBP) on basal 2C11 promoter activity revealed a substantial trans-activation by HNF1␣ and HNF3␤ (Fig. 4, A and B). HNF6 moderately trans-activated 2C11 promoter activity (ϳ2fold increase), whereas the other factors tested, including HNF1␤ and HNF3␣, had little or no effect (data not shown). Maximal activation of the 2C11 promoter by HNF1␣ was observed with Ϫ1182/2C11-Luc (ϳ40-fold), whereas maximal activation by HNF3␤ was seen with Ϫ1769/2C11-Luc (ϳ12-fold). 5Ј-Deletion analysis revealed that the HNF1␣-responsive sequences are primarily localized to two promoter regions, spanning nts Ϫ633 to ϩ22 and Ϫ968 to Ϫ633. trans-Activation of the 2C11 promoter by HNF3␤ was first seen with Ϫ968/2C11-Luc (ϳ2-fold) and was further increased, to ϳ6-fold, with Ϫ1182/2C11-Luc. HNF3␤-stimulated promoter activity was substantially reduced by inclusion of the silencer element present in Ϫ1208/2C11-Luc. This decrease was partially reversed with Ϫ1533/2C11-Luc and was fully reversed with Ϫ1769/ 2C11-Luc, the longest construct examined, indicating a strong site of HNF3␤ trans-activation between nts Ϫ1533 and Ϫ1769. Thus, at least four promoter regions (Ϫ1769 to Ϫ1533, Ϫ1533 to Ϫ1208, Ϫ1182 to Ϫ968, and Ϫ968 to Ϫ633) contribute to HNF3␤ trans-activation of CYP2C11. Two of these regions are also the most responsive to GH-dependent STAT5b inhibition of basal promoter activity (Fig. 3).
We next examined whether HNF1␣ and HNF3␤ can cooperatively interact with each other or with STAT5b to regulate 2C11 promoter activity. Co-transfection of HNF3␤ with HNF1␣ stimulated transcription of each of the 2C11-Luc constructs additively and in a manner consistent with the activation patterns exhibited by the individual factors ( Fig. 4C and Fig. 5). Further examination of Ϫ1769/2C11-Luc revealed that when STAT5b and GH receptor were additionally present, GH treatment decreased HNF3␤-stimulated 2C11 promoter activity but had no effect on HNF1␣-stimulated promoter activity (Fig. 5). The addition of HNF1␣ to the combination of HNF3␤ and STAT5b reduced but did not eliminate, the extent to which 2C11 promoter activity was inhibited by GH-activated STAT5b (ϳ50% inhibition in the presence of HNF1␣ versus ϳ85% inhibition in its absence). This suggests that HNF1␣ may compete with HNF3␤ to block the GH-and STAT5b-dependent inhibition of 2C11 activity.
STAT5b Inhibition of HNF3␤ DNA Binding and Transcriptional Activity-The mechanism whereby STAT5b inhibits HNF3␤-stimulated 2C11 promoter activity was further characterized by examining the effects of STAT5b on HNF3␤ DNA binding activity. EMSA analysis using a 2C11 probe that encompasses a consensus HNF3 site (Ϫ1162 to Ϫ1150) and the immediately adjacent STAT5 site (Ϫ1150 to Ϫ1142) (HNF3-STAT5 probe; nts Ϫ1174 to Ϫ1138) revealed two EMSA complexes, designated complex I and complex II, in HepG2 cell extracts (Fig. 6A, lane 1). Both complexes were less intense with COS-1 cell extracts (Fig. 6B, lane 4 versus lane 1), and both were substantially decreased in intensity by 100-fold molar excess of unlabeled probe (data not shown). Complex I includes HNF3-related protein, as indicated by the partial inhibitory effect of anti-HNF3␣ and anti-HNF3␤ antibody on complex formation. This inhibition was accompanied by an increase in mobility of the residual DNA-bound protein complex (Fig. 6, A, lanes 2-4, and B, lanes 2 and 3; see quantitation in figure legend). The presence of HNF3 in complex I is supported by the substantial decrease in complex I intensity upon mutation of the core HNF3 binding site (Fig. 6C, lanes 9 and 10  versus lanes 1 and 2).
Given the substantially lower level of endogenous HNF3related proteins in COS-1 cells compared with HepG2 cells (Fig. 6B), COS-1 cells were used to investigate the effects of GH-activated STAT5b on HNF3␤ binding to the 2C11 promoter probe. COS-1 cells were transfected with HNF3␤, with STAT5b and GH receptor, or with all three factors in combination. Transfection of HNF3␤ (verified by Western blotting; see complex of similar mobility (supershiftable with STAT5b antibody; Fig. 6C, lane 5 versus lane 4). This effect of STAT5b was more apparent in experiments using the HNF3 site-mutated EMSA probe (lanes 12 and 13 versus lane 11). In control experiments, STAT5b antibody had no effect on complex I in the absence of STAT5b transfection (data not shown). Interestingly, co-transfection of STAT5b with HNF3␤ inhibited formation of both the STAT5-DNA complex and the HNF3-DNA complex in a GH-dependent manner (Fig. 6C, lane 8 versus lane 7, and data not shown). This suggests that neither factor binds efficiently to the 2C11 HNF3-STAT5 probe when HNF3␤ and GH-activated STAT5b are present simultaneously.
Because the STAT5 and HNF3 consensus binding sites are immediately adjacent on the 2C11 EMSA probe, with one overlapping nucleotide, the reduced HNF3-DNA binding activity seen in the presence of GH-activated STAT5b could, in principle, result from steric hindrance between STAT5b and HNF3␤ for binding to their respective sites. However, mutation of the STAT5 site, although leading to the expected loss of supershiftable STAT5b binding seen on the wild-type STAT5 site probes (Fig. 6C, lane 21 versus lanes 5 and 13), did not restore HNF3␤ DNA binding activity in cells co-transfected with STAT5b (lane 24). This suggests that STAT5b and HNF3␤ interact in an inhibitory manner that is unrelated to their binding to adjacent sites on the 2C11 promoter.
To test this hypothesis, we investigated whether GH-activated STAT5b inhibits HNF3␤ transcriptional activity when assayed using a reporter construct that does not contain STAT5 binding sites. HepG2 cells were transfected with HNF3␤ and a luciferase reporter driven by six copies of an isolated HNF3 binding site, derived from the Cdx-2 gene (34). This reporter is specifically trans-activated by HNF3␣ and HNF3␤ but not by HNF6 (Fig. 7A), which can bind to a subset of promoter sequences in common with HNF3␤ (35). Moreover, STAT5b inhibited HNF3␤-stimulated transcription of the reporter in a GH and dose-dependent manner (Fig. 7B), despite the absence of STAT5 binding sites. Control experiments ver-ified the dose-dependent expression of STAT5b protein (Fig.  7C) and DNA binding activity (Fig. 7D). Together, these findings suggest that the GH-and STAT5b-dependent inhibition of HNF3␤-stimulated 2C11 promoter activity (Fig. 5) results from a loss of HNF3␤ DNA binding activity and, consequently, a loss of HNF3␤-stimulated 2C11 transcription. Similarly, the inhibition by GH-activated STAT5b of basal 2C11 promoter activity (Fig. 3) is suggested to reflect inhibition by STAT5b of the endogenous HNF3␤ present in the HepG2 cells used in those studies (Fig. 6A). 5. Effect of HNF1␣ and HNF3␤ on GH-and STAT5-dependent inhibition of 2C11 promoter activity. The reporter Ϫ1769/ 2C11-Luc was transfected into HepG2 cells together with a single expression plasmid (STAT5b, HNF1␣, or HNF3␤), two expression plasmids (STAT5b with HNF1␣ or HNF3␤; HNF3␤ with HNF1␣; HNF3␤ with HNF1␣), or three expression plasmids (STAT5b, HNF1␣, and HNF3␤), all in the presence of GH receptor expression plasmid. Normalized luciferase activity was determined, and the activity of the reporter in the absence of STAT5b or HNF factor was set at 1. The extent to which HNF1␣ blocked the inhibitory action of STAT5b on HNF3␤-stimulated reporter activity varied with the ratio of HNF1␣ to HNF3␤ and STAT5b (data not shown). The lower fold activation by HNF1␣ and HNF3␤ shown in this figure compared with Fig. 4 reflects the elevated basal luciferase reporter activity seen in cells transfected with the higher total amount of DNA required for this experiment (1 g/well of a 24-well plate; total DNA normalized with sonicated salmon sperm DNA).
FIG. 6. EMSA analysis of HNF3 and STAT5 DNA binding activities using 2C11 HNF3-STAT5 probe. Expression plasmids (200 ng) encoding STAT5b or HNF3␤ (as indicated) were transfected into HepG2 cells (A) or COS-1 cells (B and C) grown in 35-mm culture dishes. 30 -36 h later, the cells were treated with GH for 30 min as indicated. The cell extracts were analyzed by EMSA using the 2C11 HNF3-STAT5 probe (A and B and lanes 1-8 of C) or the 2C11 HNF3-STAT5 probe containing mutations in the HNF3 site (panel C, lanes 9 -16) or the STAT5 site (panel C, lanes [17][18][19][20][21][22][23][24]. Where indicated, the samples were incubated with antibody (Ab) to HNF3 proteins (3␣ or 3␤, as indicated) or STAT5b (5b) prior to EMSA analysis. A and B, partial inhibition of the HNF3containing EMSA complex I by anti-HNF3 antibodies (42, 50 and 63% inhibition, respectively, in HepG2 cells the presence of HNF3␣, HNF3␤, and HNF3␣/3␤ antibodies). C, the STAT5-dependent EMSA complex co-migrates with the HNF3-containing complex I and is more easily detected when the HNF3 site is mutated (lane 12 versus lane 4). Mutation of the STAT5 site (lanes 17-24) significantly decreased the HNF3-containing complex I while increasing the intensity of complex II and revealing a new complex, complex III.

HNF3␤ Inhibits STAT5b Transcriptional Activity by Blocking STAT5b
Tyrosine Phosphorylation-We next investigated whether the inhibitory interactions between HNF3␤ and STAT5b are mutual, as judged by the effects of HNF3␤ on GH-activated STAT5b transcriptional activity. STAT5b activity was assayed in HepG2 cells co-transfected with GH receptor and a STAT5 reporter containing four tandem copies of a natural STAT5 site derived from the GH-responsive ntcp gene (36). As shown in Fig. 8A, GH stimulated a 9-fold increase in STAT5b-dependent ntcp reporter activity. Moreover, HNF3␤ FIG. 7. Inhibitory effect of GH-activated STAT5b on HNF3␤stimulated (6xHNF3)/Cdx-2-Luc reporter activity. HepG2 cells were transfected with (6xHNF3)/Cdx-2-Luc reporter plasmid in the presence of the indicated expression plasmids. The data shown are relative luciferase activities compared with samples in the absence of STAT5b or HNF factors. A, trans-activation of (6xHNF3)/Cdx-2-Luc by HNF3␣ and HNF3␤ but not by STAT5b or HNF6. B, Dose-dependent inhibition of (6xHNF3)/Cdx-2-Luc activity in the presence of increasing amounts of STAT5b expression plasmid (10 -100 ng) together with a fixed amount of HNF3␤ and GH receptor expression plasmid (100 ng each). GH treatment was for 30 min. C, plasmid dose dependence of STAT5b protein expression in HepG2 cells co-transfected with GH receptor and 300 ng of HNF3␤ expression plasmid and stimulated with GH for 30 min. Shown is a STAT5b Western blot (top panel) with quantitation of STAT5b band intensities (bottom panel; mean Ϯ range, n ϭ 2). STAT5b and its phosphorylated forms were not resolved on the Western blot shown. D, the same cell extracts shown in C were assayed for STAT5 EMSA activity using a ␤-casein STAT5 probe. Cont, control.  (49)). C, cell extracts from HepG2 cells transfected with STAT5b and HNF3␤ at the indicated plasmid weight ratios were analyzed by anti-Tyr(P) 699 -STAT5b Western blotting (lanes 1-6) or by EMSA using a STAT5 ␤-casein probe (lanes 7-11). STAT5b plasmid was fixed at 50 ng for the Western blot samples and 20 ng for the EMSA samples.
inhibited this STAT5b-stimulated transcriptional response in a dose-dependent manner, despite the absence of HNF3 binding sites in the ntcp reporter. In control experiments, transfection of another liver transcription factor, HNF1␣, had no effect on STAT5b reporter activity (data not shown).
To ascertain the mechanism for this inhibitory effect of HNF3␤, we investigated whether HNF3␤ interferes with GHstimulated STAT5b activation. Fig. 8B shows that HNF3␤ significantly inhibited STAT5b activation, as demonstrated in transfected COS-1 cells by the decreased conversion of unphosphorylated STAT5b (band 0, lane 4) to the lower mobility, tyrosine-phosphorylated form of STAT5b (band 2; compare lane 8 versus lane 4). Moreover, a large decrease in STAT5b tyrosine phosphorylation was revealed by anti-phosphotyrosine 4G10 Western blotting (lane 12 versus lane 10). HNF3␤ inhibition of STAT5b tyrosine phosphorylation was also observed in HepG2 cells, as shown by Western blotting using anti-Tyr(P) 699 -STAT5 antibody (Fig. 8C, lane 2 versus lanes 4  and 6). This inhibition resulted in a dose-dependent decrease in STAT5 DNA binding activity, assayed by EMSA using a ␤-casein probe (lane 8 versus lanes 9 -11). HNF3␤ thus inhibits STAT5b-dependent transcription by a mechanism that targets the initial, GH receptor-dependent STAT5b tyrosine phosphorylation step. DISCUSSION The present study investigates the role of STAT5b and of liver-enriched transcription factors in regulating the GH-dependent and liver-specific expression of CYP2C11. STAT5 binding sites were identified in three male-specific liver CYP promoters, and in the case of 2C11, the isolated binding site was shown to confer GH-inducible, STAT5b-dependent reporter activity when fused to a heterologous promoter. These findings support the role of STAT5b in maintaining the malespecific profile of liver gene expression that was proposed earlier, based on the repeated tyrosine phosphorylation and nuclear translocation of liver STAT5b in direct response to each male plasma GH pulse (13,15,16) and on the selective loss of male-specific CYP expression in STAT5b-deficient mice (19 -21). Analysis of the STAT5b responsiveness of the intact 2C11 promoter revealed, however, a GH-and STAT5b-dependent decrease in promoter activity (Fig. 3). This unexpected effect of STAT5b was shown to involve its mutually inhibitory crosstalk with the liver transcription factor HNF3␤, which, together with HNF1␣, can strongly trans-activate 2C11 promoter activity (Fig. 4). These findings highlight the complex interrelationship between STAT5b and liver-enriched transcription factors that contribute to the transcriptional activity of GH-regulated liver CYP genes. Conceivably, the inhibitory action of STAT5b on HNF3␤-stimulated 2C11 transcription could contribute to the silencing of the 2C11 gene in female rat liver, where STAT5b is activated in a nearly continuous manner, albeit at a low level (14).
Several liver-enriched transcription factors participate in a complex cross-regulatory network with STAT5b (30). HNF4, acting in concert with STAT5b, activates the HNF6 gene, whereas HNF6, in turn, stimulates expression of HNF3␤ and HNF4. This regulatory network contributes to the expression of GH-inducible liver CYP genes, with HNF3␤ and GH-regulated HNF6 activating the female-specific CYP2C12 by binding to distinct promoter sites (23,25), and HNF3␤ and HNF1␣, but not HNF6, strongly trans-activating CYP2C11 (Fig. 4). Our finding that STAT5b can inhibit HNF3␤-inducible 2C11 expression raises the question of whether GH pulse-activated STAT5b might repress 2C11 transcription, such that the inactivation of STAT5b at the conclusion of each GH pulse (15,16) could serve as a stimulus to de-repress and thereby activate the 2C11 gene. This possibility is not likely, however, given the positive regulatory role of STAT5b evidenced by the loss of male-specific liver CYP expression in STAT5b-deficient mice (19). An alternative hypothesis, consistent with a positive regulatory role of STAT5b, is that the high concentrations of active, nuclear STAT5b found in male rat hepatocytes early during a plasma GH pulse directly stimulate 2C11 expression. By contrast, in females, the persistence of a low level of nuclear STAT5b may serve as a negative regulatory signal by counteracting the 2C11 gene activation potential of HNF3␤. According to this model, nuclear STAT5b would reach the threshold level required to trans-activate 2C11 in male but not female liver. This activation may synergize with the action of HNF1␣, which trans-activates the 2C11 promoter (Fig. 4A) and can partially reverse the inhibitory effects of STAT5b on HNF3␤-stimulated 2C11 transcription (Fig. 5). Of note, in the present studies of the 2C11 promoter, HepG2 cells were treated with GH continuously, a treatment that mimics the female plasma GH pattern. Efforts to stimulate a pulsatile pattern of STAT5b activation in HepG2 cells were hampered by the slow deactivation of STAT5b in this cell line. This precluded a determination of whether the 2C11 promoter can be stimulated by STAT5b when the STAT is activated in a pulsatile manner, as occurs in male rat liver in vivo.
HNF3␤ and STAT5b were shown to exhibit mutual inhibitory cross-talk, as revealed by our studies on the interactions of these two factors on a 2C11 promoter fragment that contains immediately adjacent HNF3␤ and STAT5 binding sites. Further investigation revealed, however, that direct DNA binding is not required for this mutual inhibition. The inhibitory crosstalk between HNF3␤ and STAT5b could conceivably involve direct protein interactions between the two transcription factors; however, no such interaction was detectable in co-immunoprecipitation experiments. 2 In agreement with this finding, the inhibition of STAT5b transcriptional activity by HNF3␤ was shown to involve a novel mechanism whereby HNF3␤ blocks STAT5b activation at the level of STAT5b tyrosine phosphorylation rather than by inhibiting STAT5b DNA binding through direct protein-protein interactions. Possible mechanisms for this intriguing effect of HNF3␤ on STAT5b activation include HNF3␤-inducible expression of a negative regulator of GH receptor/tyrosine kinase JAK2 signaling, such as a cytosolic phosphotyrosine phosphatase (37) or a SOCS/CIS protein, several of which can strongly inhibit GH receptor-dependent signaling to STAT5b (38,39).
The present finding that STAT5b can inhibit HNF3␤ transcriptional activity in the absence of a STAT5 DNA-binding site (Fig. 7) helps explain our earlier finding that STAT5b inhibits HNF3␤-and HNF6-stimulated CYP2C12 transcription even in promoter constructs devoid of recognizable STAT5 sites (23). The mechanism for this inhibitory effect of STAT5b is unknown. Inhibitory effects of STAT5b have been observed with several other transcription factors, including the nuclear receptor PPAR␣, where inhibition is mediated by the N-terminal AF1 transcriptional domain of the nuclear receptor (40,41), and the ubiquitous factor NFB, which is inhibited by STAT5b by competition for the limiting co-activator p300/CBP (42). Interestingly, an inhibitory NFB site located immediately downstream of the TATAA box of 2C11 (Ϫ2 to ϩ8) mediates down-regulation of 2C11 promoter activity in cells stimulated with interleukin-1 (43); however, it is not known whether GHactivated STAT5b is able to counteract that inhibition and thereby stimulate 2C11 expression.
STAT5b regulates target gene expression by transcriptionally activating promoters containing ␥-interferon-activated sequences matching the consensus sequence TTC-N 3 -GAA. Promoters containing adjacent ␥-interferon-activated sequencelike motifs have been shown to bind two STAT5 dimers that interact through their N-terminal region to form a tetrameric STAT5 complex (44,45). STAT5 tetramerization may confer functional cooperativity between adjacent STAT5 binding sites by increasing the level of occupancy of both sites above a threshold level required for efficient enhancer activity. In the case of 2C11, the STAT5 site that we characterized (TTC-(N) 3 -GAA; Ϫ1150 to Ϫ1142) is flanked by two generic STAT sites (TT-(N) 5 -AA, at nts Ϫ1169 to Ϫ1161 and at Ϫ1132 to Ϫ1124), with a 9 -10-nt spacing between the central STAT5 site and each of the adjacent STAT sites. HNF3 consensus sequences at nts Ϫ1162 to Ϫ1150 and at nts Ϫ1137 and Ϫ1126 partially overlap the generic STAT sites and separate them from the central STAT5 site. Although the inhibitory cross-talk between HNF3␤ and STAT5b does not require factor DNA-binding sites, as noted above, the close spacing, indeed the overlap of the HNF3␤ and STAT5 sites in the case of 2C11, could nevertheless serve to enhance the inhibitory cross-talk by increasing competition for STAT5b DNA binding. Conceivably, high levels of active STAT5b (such as are present in a GH pulse-stimulated male liver) may be required to overcome the inhibitory action of HNF3␤, leading to STAT5b activation of 2C11 gene expression. This activation could be mediated by the STAT5 site at nts Ϫ1150 to Ϫ1142 identified in the present study, perhaps in concert with the adjacent generic STAT sites. Uncharacterized STAT5 sites elsewhere in the 2C11 promoter or elsewhere in the 2C11 gene might also be involved. STAT5b is known to bind cryptic STAT5 response elements that occur as adjacent pairs but do not match the established TTC-N 3 -GAA STAT5 consensus sequence (46).
Although STAT5b clearly plays an essential role in GH-dependent expression of male-expressed CYPs and certain other liver gene products, as demonstrated in the mouse knockout studies noted above (19,20), additional factors are likely to be required to achieve the male-specific pattern of liver CYP gene expression. This conclusion is supported by the rapid activation of liver STAT5 (within 10 -15 min) in hypophysectomized rats given a single injection of GH (13,33), in contrast to the repeated pulsatile GH stimulation (over at least 2-3 days) that is required to restore male-specific liver 2C11 gene expression in the same animal model (8,47). A further indication of the requirement for additional factors is the relatively modest stimulatory effect observed with the 2C11 STAT5 response element in the present study and that of the hamster CYP3A10 promoter in an earlier report (48). Moreover, precocious activation of STAT5b in prepubertal rats administered exogenous pulses of GH for 7 days is not sufficient to activate 2C11 gene expression, pointing to a requirement for additional liver factors that are absent in prepubertal rats (15). Further study is required to identify these factors and to establish the molecular details and underlying mechanisms whereby GH and its sexually dimorphic secretory patterns induce the sex-dependent expression of 2C11 and other liver CYP genes.