The Delayed Activation of the Prostaglandin G/H Synthase-2 Promoter in Bovine Granulosa Cells Is Associated with Down-regulation of Truncated Upstream Stimulatory Factor-2*

To elucidate the molecular mechanisms involved in the delayed induction of PGHS-2 in species with a long ovulatory process, a 1.6-kilobase fragment of the bovine PGHS-2 promoter was isolated, and its activity was characterized in primary cultures of bovine granulosa cells. Promoter activity assays performed with a series of deletion mutants revealed that the promoter region from 2 149 to 2 2 ( 1 1 5 transcription start site) confers full-length promoter activity in response to forskolin (10 m M ). Four consensus cis -elements were identified within this region, including an E-box, ATF/CRE, C/EBP, and AP2 site. Site-directed mutagenesis showed that the E-box was required for PGHS-2 promoter activity, that disruption of the C/EBP element decreased forskolin inducible activity by 29%, whereas point mutation within the ATF/CRE and AP2 element had no inhibitory effect. Electrophoretic mobility shift assays (EMSAs) performed with the 2 149/ 2 2 fragment and granulosa cell nuclear extracts obtained before (0 h) and after (18 and 20 h) human chorionic gonadotropin (hCG) revealed the regulation of multiple DNA-protein complexes. The 0-h extract generated four complexes at the E-box, whereas only one complex was produced at this site with the 18-h extract. Supershift EMSAs identified that upstream stimulatory Granulosa cells were transiently transfected with 90 fmol/well of various PGHS.LUC constructs using 2 m l of LipofectAMINE (Life Technologies) in 0.3 ml of serum/antibiotic-free minimal essential medium, and following the manufacturer’s protocol. Co-transfection with the SV40 Renilla luciferase control vector (pRL.SV40; Promega) was per- formed to normalize results. A ratio of experimental (PGHS.LUC) to control vector (pRL.SV40) of 10:1 was used. After 3 h of transfection, cells were incubated in 0.5 ml of fresh minimal essential medium supplemented with 2% fetal bovine serum in the absence or presence of forskolin (10 m M ) for variable intervals of time. At the end of the culture period, cell lysates were prepared, and firefly and Renilla luciferase activities were determined using the Promega Dual Luciferase Assay System. Granulosa Cell Nuclear Extracts and Electrophoretic Mobility Shift Assays (EMSAs)— Bovine preovulatory follicles were obtained from su-perovulated holstein heifers ovariectomized 0, 18, and 20 h after hCG (47), and granulosa cell nuclear extracts were prepared as described (29, 39). Protein concentration in each extract was determined by the method of Bradford (43) (Bio-Rad Protein Assay). EMSAs were performed as described (29, 39), with minor modifications. Briefly, extracts of nuclear proteins (1.5 m g/reaction) were incubated with 25,000– 50,000 cpm of end-labeled promoter fragment 2 149/ 2 2 and 5 m g of poly(dI/dC) (Amersham Pharmacia Biotech) in a final volume of 20 m l of buffer containing 100 m M KCl, 15 m M Tris-HCl (pH 7.5), 5 m M dithio- threitol, 1 m M EDTA, 5 m M MgCl 2 , and 12% glycerol. When antibodies were used, the nuclear extract was first preincubated for 20 min with the antiserum prior to the addition of other reagents. Binding reactions were resolved acrylamide, 0.5 3 TBE gel electrophoresis. For reference purposes, eight protein-DNA complexes were designated as bands a, b, c, d, e, f, g , and h . B , competitive EMSAs were performed in the presence of granulosa cell nuclear extract prepared 0 h post-hCG, 32 P-labeled fragment 2 149/ 2 2, and 25-molar excess of unlabeled competitor DNA fragments. Competitor DNA included wild type 2 149/ 2 2, as well as four site-directed mutants (E-box mutant [E-box mut], C/EBP mutant (C/EBP mut), ATF/CRE mutant (ATF mut), and AP2 mutant (AP2 mut)) generated in the context of the 2 149/ 2 2 PGHS-2 promoter fragment. competitive EMSAs were performed under conditions identical to those described under B , except nuclear extract was prepared from granulosa cells isolated 18 h post-hCG.

Prostaglandin G/H synthase (PGHS, 1 also known as COX) is the first rate-limiting enzyme in the biosynthetic pathway of prostaglandins from arachidonic acid (1)(2)(3). The enzyme carries two sequential catalytic functions, a cyclooxygenase reaction responsible for the conversion of arachidonic acid to PGG 2 , and a peroxidase reaction involved in the conversion of PGG 2 to PGH 2 (1). PGH 2 is the common precursor for the synthesis of all prostaglandins, prostacyclins, and thromboxanes (1,4). Two isoforms of PGHS have been characterized. The first isoform, designated PGHS-1 or COX-1, was isolated more than 20 years ago from ovine and bovine seminal vesicles (5)(6)(7), whereas the second isoform, named PGHS-2 or COX-2, was cloned from chicken and mouse fibroblasts and purified from rat granulosa cells in the early 1990s (8 -10). The two isoforms share similarities at the protein level: they are approximately the same size (600 -604 amino acids) and have conserved structural and functional domains. However, they are derived from separate genes encoding different size mRNAs (11)(12)(13). Most importantly, the two isoforms differ markedly in their expression and regulation. PGHS-1 is present in a variety of tissues and is often referred to as the constitutive isoform (3,14). In contrast, PGHS-2 is undetectable in most tissues but can be induced by several agonists, and is generally referred to as the inducible form (3,14,15). Gene-targeting experiments have underscored the distinct physiological roles of each isoform, and suggest that they are both implicated in inflammation (16 -18).
Prostaglandins are key mediators of the ovulatory process, and several studies have shown that PGHS-2 is selectively induced by gonadotropins in granulosa cells prior to ovulation (10, 19 -24). The obligatory role of PGHS-2 expression during the ovulatory process was highlighted in female PGHS-2-deficient mice, which proved to be infertile because of ovulation failure and other impaired reproductive processes (25). Interestingly, comparative studies using human chorionic gonadotropin (hCG) to induce ovulation identified a marked difference in the time course of PGHS-2 induction in species with a short versus a long ovulatory process (20,21,23). In rats, a species with a short ovulatory process (12-14 h post-hCG), PGHS-2 induction is very rapid and occurs within 2-4 h post-hCG (20). In contrast, induction of PGHS-2 in species with a long ovulatory process like cows (28 h post-hCG) and mares (39 -42 h post-hCG) is remarkably delayed and occurs only at 18-and 30 h post-hCG, respectively (21,23,26). This marked difference among species in the control of PGHS-2 gene expression could be involved in the control of the mammalian ovulatory clock (27); however, its molecular basis remains unknown.
The characterization of the mouse (28), rat (29), and human PGHS-2 promoter (13,30) has provided some of the molecular mechanisms involved in PGHS-2 transcriptional activation. Consensus cis-acting elements such as NF-B, NF-IL6 (C/ EBP), ATF/CRE, and E-box were shown to play important roles in the induction of PGHS-2 transcription (30 -38). However, the relative importance of each element varies greatly depending on the cell type and agonist involved (30 -38). Thus far, regulation of the PGHS-2 promoter in granulosa cells has been studied only in rats (29,34,39). Deletion mutant analyses have revealed that the proximal region of the promoter plays a central role in cAMP-dependent regulation (29). Site-directed mutagenesis has shown that an E-box is essential for PGHS-2 promoter activity in rat granulosa cells, and that the upstream stimulatory factor (USF) is binding to this element (34). However, regulation of USF proteins has not been observed in response to hCG, leaving the mechanisms of gonadotropin-dependent trans-activation unclear.
Although progress has been achieved on our understanding of the rapid activation of PGHS-2 gene expression in rat granulosa cells, there has been no attempt to unravel the molecular basis for the delayed induction of PGHS-2 in species with a long ovulatory process. Such studies are important as they will likely be relevant to humans, a species with a long ovulatory process (36 h post-hCG) (40). Therefore, the general objective of the present study was to develop a model to study the molecular mechanisms involved in the delayed induction of PGHS-2 in species with a long ovulatory process. The specific objectives were to clone and characterize the bovine PGHS-2 gene and promoter, to study the regulation of PGHS-2 in primary cultures of bovine granulosa cells, and to characterize the hormonal control of the bovine PGHS-2 promoter in vitro. ). An additional bovine genomic library was purchased from CLONTECH (Bio/Can Scientific, Mississauga, Ontario). TRIzol total RNA isolation reagent, 1-kb DNA ladder, synthetic oligonucleotides, LipofectAMINE, and culture media were purchased from Life Technologies Inc. (Gaithersburg, MD). Tissue culture plates were obtained from Corning-Costar (Fisher Scientific, Montreal, Quebec). Fetal bovine serum was obtained from HyClone Laboratories (Logan, Utah). RNAsin, Prime-a-Gene labeling system, DNA 5Ј-End Labeling System, avian myeloblastosis virus reverse transcriptase, Dual-Luciferase Reporter Assay, and plasmids pGEM3Zf(Ϫ), pRL-SV40, and pGL3-Basic were purchased from Promega (Madison, WI). Electrophoretic reagents were purchased from Bio-Rad. Vent DNA polymerase was obtained from New England Biolabs (Beverly, MA), whereas Taq polymerase, T4 polynucleotide kinase, and all sequencing reagents were purchased from Amersham Pharmacia Biotech.

Materials-[␣-
Isolation of Bovine PGHS-2 Gene and Promoter-A bovine genomic library (Stratagene) was screened following the manufacturer's protocol with a 5Ј 1.2-kb EcoRI fragment of the mouse PGHS-2 cDNA (41) that was labeled with [␣-32 P]dCTP using the Prime-a-Gene labeling system (Promega). Out of 900,000 phage plaques screened, six positive clones were identified, purified, and characterized by Southern blot analyses. DNA fragments were subcloned into pGEM 3Zf(Ϫ) and sequenced by the dideoxy method (41). While the six clones varied in size from 12 to 18 kb, none contained the region upstream of intron 4. To clone the promoter and the 5Ј-region of the bovine PGHS-2 gene, a second bovine genomic library (CLONTECH) was screened with the mouse PGHS-2 cDNA probe. From 100,000 phage plaques screened, one positive clone of 15 kb was isolated and shown to contain the complete bovine PGHS-2 gene as well as about 4 kb of 5Ј-flanking DNA. The entire gene and 1650 bp of the putative promoter were characterized by DNA sequencing (42).
Primer Extension Analysis-The transcription initiation site of the bovine PGHS-2 gene was determined by primer extension analysis, as described (26,29). The reaction used total RNA extracted from preovulatory follicles isolated 0 or 24 h after hCG treatment (24), and an antisense oligonucleotide 5Ј-GAGGGCGGTGCGGAGTTCCGGGGC-G-3Ј designed from bovine PGHS-2 and located 35-59 bp downstream of the transcription initiation site identified in human PGHS-2 (13,30). The extension product was analyzed by electrophoresis on a 6% polyacrylamide, 7 M urea gel, and the transcription start site was determined by comparison with an adjacent sequencing reaction that used the same oligonucleotide as primer and a PGHS-2 genomic clone that contained this region as template.
Semi-quantitative RT-PCR Analysis-Total RNA was extracted with TRIzol (Life Technologies) from granulosa cells incubated for 0, 6, 12, 24, and 36 h in the presence of forskolin (10 M). RT reactions were performed using 5 g of RNA, 20 M of an oligo poly(dT) 15 primer, and 10 units of avian myeloblastosis virus reverse transcriptase. For amplication of PGHS-2, sense and antisense primers (primers 1 and 2, Table I) specific to bovine PGHS-2 were designed to produce a 1.1-kb fragment. As an internal control, a set of sense and antisense primers (primers 3 and 4, Table I) were designed to amplify a fragment of about 0.85 kb of a constitutively expressed transcript coding for bovine glyceraldehyde-3-phosphate dehydrogenase (22). PCR reactions were performed using 1 l of Taq polymerase, 50 pmol of each primer, 5 mM dNTPs, and the following cycling conditions: 32 cycles of 30-s denaturation at 95°C, 30-s annealing at 48°C, and 1 min elongation at 72°C. A final extension at 72°C was performed for 5 min after the last cycle, and PCR products were analyzed on a 1% agarose gel.
Immunoblot Analysis-Bovine granulosa cells were collected after 0, 12, 24, 36, 48, and 72 h of culture in the presence of forskolin (10 M), and solubilized cell extracts were prepared as described (24). Protein concentration was determined by the method of Bradford (43) (Bio-Rad Protein Assay). Samples (50 g of proteins) were resolved by onedimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), electrophoretically transferred to nitrocellulose filters, and incubated with polyclonal anti-PGHS antibody 9181 and 125 I-Protein A, as described (24). Filters were exposed to film at Ϫ70°C. Immunoblot analyses were also performed with nuclear extracts prepared from granulosa cells isolated 0, 18, and 20 h post-hCG (see below), and using polyclonal antibodies (1:100 dilution) raised against USF-1, the amino terminus of USF-2, the carboxyl terminus of USF-2, C/EBP␣, C/EBP␤, and C/EBP␦ (Santa Cruz Biotechnology Inc., Santa Cruz, CA).
Prostaglandin E 2 Radioimmunoassay-Media samples were collected from primary cultures of bovine granulosa cells incubated for 0, 12, 24, 36, 48, and 72 h in the absence or presence of forskolin (10 M), and assayed for PGE 2 using a specific radioimmunoassay (44).
Deletion and Site-directed Mutants of the Bovine PGHS-2 Promoter-Five bovine PGHS-2 promoter fragments, including Ϫ1574/Ϫ2, Ϫ492/ Ϫ2, Ϫ325/Ϫ2, Ϫ149/Ϫ2, and Ϫ88/Ϫ2 (where ϩ1 ϭ transcription initiation site), were generated by PCR amplification using five specific sense primers (oligonucleotides 5, 6, 7, 8, and 9, respectively, Table I) and one common antisense primer (oligonucleotide 10, Table I). In addition, a sixth promoter fragment was produced by annealing two complementary oligonucleotides spanning the DNA region from Ϫ39 to Ϫ2 (oligonucleotides 11 and 12, Table I). Each fragment was inserted upstream of the firefly luciferase reporter gene in the vector pGL3.Basic Four site-directed mutants, including an E-box, C/EBP, ATF/CRE, and an AP2 mutant, were generated by inverse PCR (45) using Ϫ149/ Ϫ2PGHS.LUC as template and the primers listed in Table I. PCR products were digested with EcoRI and XhoI to release the Ϫ149/Ϫ2 mutated fragment, isolated on a 1% agarose gel, and ligated into pGL3.Basic. Mutations were confirmed by DNA sequencing.
Cell Culture, Transient Transfections, and Promoter Activity Assays-Paired bovine ovaries bearing a newly formed corpus luteum and a follicle of 8 -12 mm in diameter were obtained from a slaughterhouse, and transported on ice to the laboratory. The follicle, which is the putative dominant follicle of the first wave of the estrous cycle (46), was dissected from the ovary, and granulosa cells were isolated as described (24). Cells were washed twice with minimal essential medium supplemented with 2% fetal bovine serum, and seeded in 24-well plates at a density of 2-5 ϫ 10 5 cells/0.5 ml of minimal essential medium supplemented with 2% fetal bovine serum, insulin (1 g/ml), transferin (5 g/ml), and penicillin (100 units/ml)-streptomycin (100 g/ml). Cultures were incubated at 37°C in a humidified incubator gassed with 95% air and 5% CO 2 .
Granulosa cells were transiently transfected with 90 fmol/well of various PGHS.LUC constructs using 2 l of LipofectAMINE (Life Technologies) in 0.3 ml of serum/antibiotic-free minimal essential medium, and following the manufacturer's protocol. Co-transfection with the SV40 Renilla luciferase control vector (pRL.SV40; Promega) was performed to normalize results. A ratio of experimental (PGHS.LUC) to control vector (pRL.SV40) of 10:1 was used. After 3 h of transfection, cells were incubated in 0.5 ml of fresh minimal essential medium supplemented with 2% fetal bovine serum in the absence or presence of forskolin (10 M) for variable intervals of time. At the end of the culture period, cell lysates were prepared, and firefly and Renilla luciferase activities were determined using the Promega Dual Luciferase Assay System.
Granulosa Cell Nuclear Extracts and Electrophoretic Mobility Shift Assays (EMSAs)-Bovine preovulatory follicles were obtained from superovulated holstein heifers ovariectomized 0, 18, and 20 h after hCG (47), and granulosa cell nuclear extracts were prepared as described (29,39). Protein concentration in each extract was determined by the method of Bradford (43) (Bio-Rad Protein Assay). EMSAs were performed as described (29,39), with minor modifications. Briefly, extracts of nuclear proteins (1.5 g/reaction) were incubated with 25,000 -50,000 cpm of end-labeled promoter fragment Ϫ149/Ϫ2 and 5 g of poly(dI/dC) (Amersham Pharmacia Biotech) in a final volume of 20 l of buffer containing 100 mM KCl, 15 mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 1 mM EDTA, 5 mM MgCl 2 , and 12% glycerol. When antibodies were used, the nuclear extract was first preincubated for 20 min with the antiserum prior to the addition of other reagents. Binding reactions were resolved by 5% acrylamide, 0.5 ϫ TBE gel electrophoresis.

Induction of PGHS-2 in Primary
Cultures of Bovine Granulosa Cells-To establish a model in vitro for the study of de-layed PGHS-2 induction, the regulation of PGHS-2 was characterized in primary cultures of bovine granulosa cells stimulated with forskolin. RT-PCR analysis showed that no PGHS-2 mRNA was present after 0 and 6 h of forskolin stimulation, but the transcript was induced at 12, 24, and 36 h (Fig.  1A). Immunoblot analysis revealed a marked increase in PGHS-2 protein between 12 and 24 h of forskolin stimulation, followed by a progressive loss in protein expression between 24 and 72 h (Fig. 1B). To determine if the induction of PGHS-2 mRNA and protein was associated with changes in prostaglandin synthetic activities, PGE 2 concentrations were measured in culture media. No difference was observed in PGE 2 levels in cultures incubated in the absence of forskolin (Fig.  1C). However, PGE 2 concentrations were significantly higher in cultures stimulated for 24 -72 h with forskolin as compared with controls (p Ͻ 0.05; one-way ANOVA/multiple comparisons with the Fisher's LSD procedure), with most of the stimulatory effect observed at 24 and 36 h (Fig. 1C).
Characterization of the Bovine PGHS-2 Gene, Promoter, and Transcription Initiation Site-The complete primary structure of the PGHS-2 gene was determined from clones obtained from two genomic libraries. Results showed that the gene consists of 10 exons and 9 introns (Fig. 2A). The exon/intron boundaries were confirmed by comparisons with the bovine PGHS-2 cDNA (PGHS-2 cDNA GenBank accession number AF031698), and each splice site agreed with the consensus donor/acceptor (GT/ AG) sequence (Fig. 2B). Primer extension analysis was used to B. Deletion mutants of the bovine PGHS-2 promoter 5

Delayed Induction of PGHS-2 in Bovine Granulosa Cells
identify the transcription initiation site. Fig. 2C shows that a 57-nucleotide extension product was produced when the primer was hybridized to a sample known to contain PGHS-2 mRNA. The transcription start site was identified at a guanidine residue (Figs. 2C and 3). A 1.6-kb DNA fragment located immediately upstream of the bovine PGHS-2 transcription start site was isolated and sequenced. Several consensus cis-acting elements were identified within the first 600 bp upstream of the cap site, including elements for PEA3, C/EBP, AP-1, AP-2, NF-B, ATF/CRE, and E-box-binding proteins (Fig. 3). However, no consensus TATA box motif was present within the putative promoter, but the hexanucleotide 5Ј-ATAAAA-3Ј element was located 30 bp upstream of the transcription start site (Fig. 3).
Functional Analysis of the Bovine PGHS-2 Promoter-To determine if the isolated 5Ј-flanking region had functional promoter activity, a DNA fragment from Ϫ1574 to Ϫ2 (ϩ1 ϭ transcription start site) was fused upstream of the firefly luciferase reporter gene in the pGL3.Basic vector (Fig. 4A). The chimeric construct Ϫ1574/Ϫ2PGHS.LUC was transiently transfected into primary cultures of bovine granulosa cells stimulated with forskolin. Results from a time course study showed that reporter gene activity was very low after 6 h of forskolin stimulation, but gradually increased thereafter and peaked at 36 h (Fig. 4B). Thus, a period of 24 or 36 h of forskolin stimulation was selected for subsequent transient transfections.
To identify regions within the Ϫ1574/Ϫ2 fragment involved in forskolin regulation of PGHS-2 promoter activity, 5Ј-deletion mutants were designed (Fig. 4A) and transiently transfected in bovine granulosa cells. Results showed that deletions between Ϫ1574 and Ϫ149 had no effect on forskolin-stimulated luciferase activity (Fig. 4C). However, the deletion of the region between Ϫ149 and Ϫ88 resulted in a 56 and 75% decrease in basal and forskolin-stimulated activities, respectively. Further deletion between Ϫ88 and Ϫ39 resulted in a loss of reporter gene activities, with the Ϫ39/Ϫ2PGHS.LUC activities being similar to those of the promoter-less construct (Fig. 4C). Thus, the region Ϫ149/Ϫ39 appears to play a key role in the regulation of the bovine PGHS-2 promoter in granulosa cells.
To determine if consensus cis-acting elements present within the Ϫ149/Ϫ39 region were involved in promoter activity, sitedirected mutants of the AP2, C/EBP, ATF, and E-box elements were generated and tested (Fig. 5A). Results indicated that mutation of the E-box markedly decreased PGHS-2 promoter activities, with basal and forskolin stimulated activities decreasing to 8 and 9%, respectively, of the activities of the Ϫ149/Ϫ2PGHS.LUC wild type control (Fig. 5B). In contrast, mutations of other three elements had no comparable effect, but mutation of the C/EBP element resulted in a 29% decrease in forskolin-stimulated luciferase activity when compared with C, determination of the transcription initiation site of the bovine PGHS-2 gene by primer extension analysis. A labeled antisense primer was hybridized to RNA samples containing (follicle isolated 24 h post-hCG) and not containing (follicle 0 h post-hCG; negative control) PGHS-2 mRNA, and primer extension was performed as described under "Experimental Procedures." The extended product was analyzed on 6% polyacrylamide gel and its size determined by comparison with the products of a sequencing reaction using the same primer and a genomic clone spanning this region as the template. A 57-nucleotide extended product was obtained.
Protein/DNA Binding Activities within the Bovine PGHS-2 Promoter Region Ϫ149/Ϫ2-The ability of the Ϫ149/Ϫ2 PGHS-2 promoter fragment to interact with nuclear proteins was tested by EMSA using extracts prepared from granulosa cells obtained before (0 h post-hCG) and during (18 and 20 h post-hCG) induction of PGHS-2 in vivo (21,47). Results showed that all nuclear extracts generated multiple DNA-protein complexes, of which eight were selected and designated as complexes a, b, c, d, e, f, g, and h for reference purposes. Some complexes were regulated by hCG stimulation: complexes b, d, e, and f were detected only with the 0-h extract, whereas complexes c and h increased after hCG treatment (Fig. 6).
To assess the specificity of protein/DNA interactions, and to identify cis-elements involved in complex formation, competitive EMSAs were performed using 25 M excess of unlabeled wild type Ϫ149/Ϫ2 fragment, and of Ϫ149/Ϫ2 fragments containing a mutated E-box, C/EBP, ATF, or AP2 element. Results showed that all complexes generated with the 0-and 18-h extracts were competed with excess wild type Ϫ149/Ϫ2, thus demonstrating the specific nature of the interactions (Fig. 6, B  and C). Interestingly, the use of Ϫ149/Ϫ2 fragments with point mutations allowed us to assign several complexes to specific cis-elements . Four bands (b, e, f, and g; Fig. 6B) were assigned to the E-box at 0 h, but only one of them was still present at 18 h post-hCG (band g; Fig. 6C). An identical, albeit weaker, DNA/protein interaction pattern was observed with the ATFmutated DNA fragment, which likely relates to the overlapping nature of the ATF and E-box elements. Complexes a and c were localized to the C/EBP element at 0 and 18 h post-hCG, with the intensity of band c being more pronounced at 18 h. All complexes were competed by the ATF mutated Ϫ149/Ϫ2 fragment, suggesting that this element was not involved in the FIG. 3. Nucleotide sequence of the bovine PGHS-2 promoter from ؊1650 to ؉100. Numbering is relative to the transcription start site (ϩ1) determined by primer extension analysis (Fig. 2). The sequence was submitted to the transcription factor data base TFSites (GCG, Wisconsin), and putative cis-acting elements located within the first 600 bp upstream of the cap site are underlined. The nucleotide sequence was deposited to GenBank (accession number AF031699). formation of any of the observed complexes (Fig. 6, B and C). Complexes d and h could not be assigned to any of the four elements with this approach.
To confirm binding activities to the E-box and C/EBP elements, competitive EMSAs were performed with specific oligonucleotides as cold competitor DNA. Results shown in Fig. 7, A and B, confirmed those observed in Fig. 6, B and C: excess of cold E-box oligonucleotides competed bands b, e, f, and g at 0 h, and complex g at 18 h; excess cold C/EBP oligonucleotides competed complexes a and c at both time points; and competition with the combination of E-box and C/EBP oligonucleotides abolished all complexes. Interestingly, complex h was competed by the E-box or the C/EBP oligonucleotide, suggesting that both elements are required for the formation of this large complex (Fig. 7B).
USF and C/EBP Proteins Are Involved in the Formation of Multiple Protein-DNA Complexes-To determine if the identified complexes contain USF and C/EBP proteins, supershift EMSAs were performed using antibodies against USF-1, the carboxyl terminus of USF-2, and the amino terminus of USF-2, C/EBP␣, C/EBP␤, and C/EBP␦. Assays were done in the presence of 25 M excess of cold competitor mutants (E-box or C/EBP mutant in the Ϫ149/Ϫ2 context) to clearly identify bands interacting with each cis-element. Results suggest that complexes interacting at the E-box at 0 h contain USF-1 and USF-2 proteins (Fig. 8A). The complex with the fastest mobility, band b, was markedly shifted by an antibody directed against the carboxyl terminus of USF-2, but not by an antibody directed against the amino terminus of USF-2, suggesting that this complex contains the amino-truncated form of the protein.
Larger complexes are thought to contain dimers (homodimers or heterodimers) composed of a full-length USF-1 or USF-2 and a truncated UFS-2 protein (bands e and f), or of two full-length USF proteins (band g). The only complex localized to the E-box at 18 h post-hCG (band g, Fig. 8B) was shifted by all USF antibodies, in keeping with results observed at 0 h. For complexes involving the C/EBP element, band a, which is present at 0 and 18 h, contained primarily C/EBP␤ (Fig. 8, A and B), whereas band c, which is more prominent at 18 h, appeared to contain some C/EBP␣ (Fig. 8B).
To determine the potential regulation of USF and C/EBP proteins, immunoblot analyses were performed using nuclear extracts isolated before (0 h) and after (18 and 20 h) hCG treatment in vivo. Results showed that USF-1 remained relatively constant after hCG treatment, whereas full-length USF-2 and C/EBP␤ appeared lower at 18 and 20 h, as compared with 0 h (Fig. 9). No C/EBP␣ and ␦ was detected by immunoblots (data not shown). A putative amino-terminal truncated form of USF-2 (M r ϭ 18,000) was present at 0 h, but disappeared at 18 and 20 h (Fig. 9). Also, high levels of putative truncated C/EBP␤ (M r ϭ 20,000) were detected in the 0-h extract.

DISCUSSION
This study is the first to focus on the molecular mechanisms involved in the delayed induction of PGHS-2 in granulosa cells of a species with a long ovulatory process. Previous investigations have identified a marked difference in the time course of follicular PGHS-2 induction in species with a short versus a long ovulatory process, ranging from 2 to 4 h in rats (20) to 18 h in cows (21,24) and 30 h post-hCG in mares (23,26). Interestingly, this gradual delay in PGHS-2 induction in species with longer ovulatory processes was proposed as a determinant of the mammalian ovulatory clock (27). Yet, the remarkable variation in the control of PGHS-2 induction across species is intriguing, considering that it involves the same cell type (granulosa cells) and the same agonist (hCG). The present model in vitro appears to parallel the pattern of PGHS-2 induction observed in vivo (21,24), as levels of PGHS-2 mRNA and protein became maximal only at 24 h post-hCG. This is in sharp contrast with studies in rats which showed that maximal induction occurs 5-7 h post-hCG (20,48). Moreover, maximal reporter gene activity in bovine granulosa cells with Ϫ1574/ Ϫ2PGHS.LUC was observed after 24 -36 h of forskolin. In contrast, comparable studies in rats indicated that maximal PGHS-2 promoter activity was induced after 6 h of forskolin stimulation (29). The induction of PGHS-2 in the bovine model was paralleled by an increase in prostaglandin synthesis, as evidenced by the rise in PGE 2 . We predict that PGF 2␣ is also increased by forskolin in bovine granulosa cell cultures, since the induction of PGHS-2 in vivo has been associated with a significant increase in PGE 2 and PGF 2␣ (21,24,47). Thus, the delayed induction of PGHS-2 in bovine granulosa cells in vitro provides a valuable model to study the regulation of PGHS-2 gene expression in species with a long ovulatory process.
Distinct cis-acting promoter elements and/or trans-activating factors were considered as potential regulators of delayed PGHS-2 induction in this species. Deletion analysis of the bovine PGHS-2 promoter revealed that, as observed in rats (29), a proximal region of the promoter was sufficient to confer basal and forskolin inducible activities, suggesting that the same region is involved in both species. The proximal regions of the bovine and rat PGHS-2 promoters are relatively conserved

FIG. 5. Effect of site-directed mutagenesis on forskolindependent PGHS-2 promoter activity in bovine granulosa cells.
A, four site-directed mutants were generated within the context of the Ϫ149/Ϫ2 PGHS-2 promoter fragment, as described under "Experimental Procedures." B, bovine granulosa cells were transiently transfected with the promoterless plasmid pGL3.Basic (Basic), the wild type construct Ϫ149/Ϫ2PGHS.LUC (Ϫ149/Ϫ2 Wt), or with constructs containing point mutations with the AP2, C/EBP, ATF/CRE, or E-box ciselements. All cultures were co-transfected with the SV40 Renilla luciferase vector (SV40.pRL) as an internal control to normalize experimental reporter activity. After transfection, cells were incubated for 36 h in the absence or presence of forskolin (10 M). Results are presented as relative luciferase activity (firefly/renilla; mean Ϯ S.E. of triplicate cultures from three experiments).
(63% identity in nucleotide sequence), and contain four putative cis-elements, including an AP2, C/EBP, ATF/CRE, and E-box. Site-directed mutagenesis studies indicated that the E-box was essential for full promoter activity in bovine granulosa cells. Mutation of the C/EBP element resulted in a moderate reduction (29%) in forskolin induced activity, whereas mutation of ATF/CRE and AP2 elements had no inhibitory effect. The central role played by the E-box in the bovine system is in keeping with findings in rats (34). Thus, results of promoter activity assays did not provide evident clues on potential mechanisms responsible for the marked difference in the time course of PGHS-2 induction between species.
One important finding of this study is the regulation of multiple complexes involving USF proteins and the E-box. USF-1 (43 kDa) and USF-2 (44 kDa) are ubiquitous proteins that belong to the Myc family of transcription factors (49 -51). They are characterized by highly conserved basic helix loop helix and leucine zipper domains which are responsible for dimerization and DNA binding. USF proteins interact with DNA as homodimers and heterodimers, and are known to affect the expression of various genes (51)(52)(53). A recent study showed that overexpression of USF-1 and USF-2 homodimers acts as repressor of ribosomal RNA gene transcription, whereas overexpression of USF-1/USF-2 heterodimers acts as activator, sug-  Table  I. Binding reactions were resolved by 5% acrylamide, 0.5 ϫ TBE gel electrophoresis.
gesting that the relative amount of each form of dimers could play a key role in gene expression (52). Moreover, an amino terminus truncated form of USF (also called mini-USF) has been reported (50). Mini-USFs (18 kDa) lack the transcription activation domain but have the dimerization and DNA-binding domains, and would function as a dominant negative mutant (50). Expression of amino-truncated USF was used to demonstrate the role of the transcription factor in the regulation of the L-type pyruvate kinase promoter (54), but the natural occurrence of mini-USF has remained elusive. The present study identifies a physiological event during which putative mini-USF-2 could serve as repressor of gene expression. Supershift EMSAs and immunoblot analyses provided complementary evidence for the presence of amino terminus truncated USF in granulosa cells isolated prior to hCG (0 h), and their disappearance after hCG treatment. Thus, removal of repres-sive mini-USF could serve as an initial step in delayed PGHS-2 induction in species with a long ovulatory process. These results contrast with those in rats where, although granulosa cell extracts were shown to contain USF proteins binding to the E-box, there was no evidence of mini-USF (34). Moreover, hCG had no effect on protein complexes binding to the E-box, and on levels of USF proteins in rats (34). Thus, major differences exist in the nature and regulation of protein/DNA binding activities at the E-box between species with a rapid versus delayed induction of PGHS-2, and the putative repressor role of mini-USF in species with a delayed PGHS-2 induction is attractive and will require further studies.
Another important difference between the two species involves the distinct regulation of C/EBP␤ in preovulatory follicles. High levels of C/EBP␤ protein were present in bovine granulosa cells at 0 h post-hCG, and gonadotropin treatment was associated with a slight decrease in C/EBP␤. In contrast, a previous study showed that there is no C/EBP␤ in rat granulosa cells prior to hCG (0 h post-hCG), but there is a rapid induction of its mRNA and protein after gonadotropin treatment (39). The functional significance of C/EBP␤ in the regulation of the PGHS-2 promoter in granulosa cells remains unclear. In rats, recent site-directed mutagenesis experiments performed in the context of the full proximal PGHS-2 promoter revealed that the C/EBP element was not required for promoter activation in granulosa cells (34). This result is supported by the fact that the gonadotropin-dependent induction of PGHS-2 in granulosa cells is not compromised in C/EBP␤-deficient mice (55). However, there was no down-regulation of PGHS-2 mRNA subsequent to induction by luteinizing hormone/hCG in C/EPB␤ null mice, prompting the authors to speculate that C/EBP␤ induction in rodents is involved in the repression of PGHS-2 expression in normal animals (55). Overexpression of C/EBP␤ in vascular endothelial cells was shown to suppress PGHS-2 promoter activity (35). Thus, the relatively high levels of C/EBP␤ in nuclear extracts of bovine granulosa cells prior to gonadotropin could serve to repress promoter activation in species with a delayed PGHS-2 induction. Also, the presence of high levels of putative amino-terminal truncated C/EBP␤ (20 kDa; also called LIP for liver enriched inhibitor protein (56)) prior to hCG could be of functional significance. Liver-enriched FIG. 8. Effect of antisera against USF and C/EBP proteins on binding activities between granulosa cell nuclear extracts and the bovine PGHS-2 promoter fragment ؊149/؊2. To determine if USF and C/EBP proteins were involved in nuclear protein/DNA binding, EMSAs were performed in the presence of granulosa cell nuclear extracts prepared 0 h (A) or 18 h (B) post-hCG, 32 P-labeled fragment Ϫ149/Ϫ2, 25 M excess unlabeled competitor DNA fragments Ϫ149/Ϫ2 with a mutated E-box (E-box mutant) or C/EBP (C/EBP mutant), and selected polyclonal antibodies. They included antibodies raised against USF-1, the carboxyl terminus of USF-2, the amino terminus of USF-2, C/EBP␣, C/EBP␤, and C/EBP␦. Binding reactions were prepared as described under "Experimental Procedures," and were resolved by 5% acrylamide, 0.5 ϫ TBE gel electrophoresis.
FIG. 9. Immunoblot analysis of USF-1, USF-2, and C/EBP␤ protein content in bovine granulosa cell nuclear extracts prepared 0, 18, and 20 h post-hCG in vivo. Nuclear extracts were prepared from granulosa cells isolated 0, 18, and 20 h post-hCG. Proteins (50 g/lane) were analyzed by one-dimensional SDS-PAGE and immunoblotting techniques using polyclonal antibodies against USF-1, the carboxyl terminus of USF-2 (anti-USF2(C)), the amino terminus of USF-2 (anti-USF-2(N)) and C/EBP␤, as described under "Experimental Procedures." Upper arrowheads on the left indicate migration of respective full-length proteins, whereas lower arrowheads indicate putative amino terminus truncated USF-2 (mini-USF; Ref. 50) and C/EBP␤ (56). inhibitor protein, which contains the dimerization and DNAbinding domains but lacks the amino-terminal activation domain, acts as a transcriptional repressor (56). A transient decrease in full-length and amino-terminal truncated C/EBP␤ levels after gonadotropin treatment could allow activation of gene transcription through the recruitment of additional factors.
Lastly, this study characterizes the primary structure of the bovine PGHS-2 gene. The intron/exon organization of the bovine PGHS-2 gene is very similar to that of the human (13), mouse (28), and equine genes (26). The overall size of the bovine gene compares with that of other species, and the length of internal exons 2-9 and of the coding region of exons 1 and 10 are identical among bovine, murine, human, and equine PGHS-2 (13,26,28). The transcription start site of the bovine PGHS-2 gene was identified at a guanidine residue, which contrasts with the start site identified in human (cytidine; Ref. 13), mouse, rat, and equine PGHS-2 (adenosine; Refs. 26, 28, and 29). This difference in the transcription initiation site contributes, at least in part, to a small variation observed across species in the length of the 5Ј-untranslated region.
In summary, this study is the first to characterize some of the molecular mechanisms involved in the delayed induction of follicular PGHS-2 in species with a long ovulatory process. The report provides the characterization of the bovine PGHS-2 gene and promoter, as well as a model in vitro of primary cultures of granulosa cells in which the regulation of PGHS-2 closely parallels the regulation observed in vivo. The marked difference in the time course of PGHS-2 induction between rats and cattle appears associated with a distinct regulation of nuclear protein/DNA interactions at the level of the proximal promoter.
Our results indicate a novel regulation of USF proteins in granulosa cells during ovulation, with amino terminus-truncated USF-2 potentially exerting a repressor role on PGHS-2 promoter activity prior to gonadotropin treatment in bovine preovulatory follicles. Likewise, high levels of C/EBP␤ proteins in bovine follicles prior to hCG treatment, which contrasts with observations in rats, suggests that this transcription factor could also repress PGHS-2 promoter activity in bovine granulosa cells. These new findings will provide the basis of novel working hypotheses regarding the species-specific control of PGHS-2 gene expression in ovarian cells.