CCAAT Enhancer-binding Protein β and GATA-4 Binding Regions within the Promoter of the Steroidogenic Acute Regulatory Protein (StAR) Gene Are Required for Transcription in Rat Ovarian Cells*

Steroidogenic acute regulatory protein (StAR) is a vital accessory protein required for biosynthesis of steroid hormones from cholesterol. The present study shows that in primary granulosa cells from prepubertal rat ovary, StAR transcript and protein are acutely induced by gonadotropin (FSH). To determine the sequence elements required for hormone inducibility of the StAR promoter, truncated regions of the −1002/+6 sequence of the mouse gene were ligated to pCAT-Basic plasmid and transfected by electroporation to freshly prepared cells. FSH inducibility determined over a 6-h incubation was 10–40-fold above basal levels of chloramphenicol acetyltransferase activity. These functional studies, supported by electrophoretic mobility shift assays indicated that two sites were sufficient for transcription of the StAR promoter constructs: a non-consensus binding sequence (−81/−72) for CCAAT enhancer-binding protein β (C/EBPβ) and a consensus motif for GATA-4 binding (−61/−66). Western analyses showed that GATA-4 is constitutively expressed in the granulosa cells, while all isoforms of C/EBPβ were markedly inducible by FSH. Site-directed mutations of both binding sequences practically ablated both basal and hormone-driven chloramphenicol acetyltransferase activities to less than 5% of the parental −96/+6 construct. Unlike earlier notions, elimination of potential binding sites for steroidogenic factor-1, a well known tissue-specific transcription factor, did not impair StAR transcription. Consequently, we propose that C/EBPβ and GATA-4 represent a novel combination of transcription factors capable of conferring an acute response to hormones upon their concomitant binding to the StAR promoter.

Steroidogenic acute regulatory protein (StAR) is a vital accessory protein required for biosynthesis of steroid hormones from cholesterol. The present study shows that in primary granulosa cells from prepubertal rat ovary, StAR transcript and protein are acutely induced by gonadotropin (FSH). To determine the sequence elements required for hormone inducibility of the StAR promoter, truncated regions of the ؊1002/؉6 sequence of the mouse gene were ligated to pCAT-Basic plasmid and transfected by electroporation to freshly prepared cells. FSH inducibility determined over a 6-h incubation was 10 -40-fold above basal levels of chloramphenicol acetyltransferase activity. These functional studies, supported by electrophoretic mobility shift assays indicated that two sites were sufficient for transcription of the StAR promoter constructs: a non-consensus binding sequence (؊81/؊72) for CCAAT enhancer-binding protein ␤ (C/EBP␤) and a consensus motif for GATA-4 binding (؊61/؊66). Western analyses showed that GATA-4 is constitutively expressed in the granulosa cells, while all isoforms of C/EBP␤ were markedly inducible by FSH. Site-directed mutations of both binding sequences practically ablated both basal and hormone-driven chloramphenicol acetyltransferase activities to less than 5% of the parental ؊96/؉6 construct. Unlike earlier notions, elimination of potential binding sites for steroidogenic factor-1, a well known tissue-specific transcription factor, did not impair StAR transcription. Consequently, we propose that C/EBP␤ and GATA-4 represent a novel combination of transcription factors capable of conferring an acute response to hormones upon their concomitant binding to the StAR promoter.
The first and key reaction in the enzymatic cascade of steroid hormone biosynthesis is catalyzed in the mitochondria by cholesterol side chain cleavage cytochrome P450 (P450scc) 1 (1)(2)(3).
In the presence of atmospheric oxygen and reducing power provided by associated proteins, P450scc converts cholesterol substrate to the first steroid prototype molecule, pregnenolone (1). In order to do so, a supply of cholesterol is required to be transferred from cytosolic pools into the inner membranes of the mitochondrion, where P450scc resides (4 -7). Recently, it was found that cholesterol delivery into the mitochondria is enhanced by a novel protein (8,9) designated steroidogenic acute regulatory (StAR) protein (reviewed in Refs. 10 -12). More studies have established the fact that StAR is a vital protein essential for steroidogenesis in the adrenal cortex and the gonads (13,14). In rodents, StAR is also expressed in steroidogenic brain cells (15) and placenta. 2 Interestingly, StAR is not expressed in human placenta, where its role is probably assumed by a less efficient StAR substitute called MLN64 (16). Perhaps the most compelling evidence for the critical role of StAR in steroidogenesis was the discovery that various mutations of the StAR gene encoding a functionally impaired protein (14) cause a syndrome known as lipoid congenital adrenal hyperplasia (17). Affected individuals die shortly after birth in the absence of adrenal steroids, unless treated with steroid hormone replacement therapy. Similar patterns were also observed in StAR null mice (14).
Trophic hormones, such as gonadotropins and ACTH, trigger up-regulation of StAR expression by cAMP signaling (18 -20). Additionally, Ca 2ϩ changes evoke StAR expression in glomerulosa cells of the adrenal cortex (21,22). Very little is known about the factors controlling StAR expression at the transcriptional level, downstream to the signal transduction pathways. Special attention has been devoted to examine the potential involvement of the steroidogenic factor-1 (SF-1, or Ad4BP), which is a pivotal tissue-specific orphan nuclear receptor essential for regulation of many steroid hydroxylases in steroidproducing tissues (23,24). In light of the fact that StAR promoter includes several putative recognition sites for SF-1 binding, several attempts have be made to determine if the latter factor is involved in StAR regulation. At present, the available results are somewhat inconsistent. Using the human, mouse, and rat promoters, an apparent up-regulation of StAR transcription by SF-1 could be demonstrated upon co-transfection of SF-1 cDNA and promoter-reporter plasmids in nonsteroidogenic cells (25)(26)(27)(28)(29). However, other studies analyzing the activity of StAR promoter in SF-1-expressing cells did not support a role for SF-1 in a cAMP-inducible fashion (30). These data suggested that SF-1 may not confer cAMP responsiveness in authentic steroidogenic cells and, therefore, cannot be an exclusive transcription factor controlling the acute regulation of StAR in such cells.
In search for alternative regulatory elements that can mediate the acute response of StAR to hormones, we undertook to study the inducibility of the mouse promoter in ovarian gran-ulosa cells from prepubertal rats. Earlier studies have unambiguously shown that endogenous SF-1 in these cells is critical for the induction of P450scc and P450aromatase by folliclestimulating hormone (FSH) (31)(32)(33)(34). In contrast, the present study suggests that SF-1 is probably not involved in hormonal activation of StAR promoter. We also demonstrate that promoter regions capable of C/EBP␤ and GATA-4 binding are required for activation of StAR transcription in FSH-treated cells. Thus, StAR provides the first example of a steroidogenesis-associated protein that is transcriptionally controlled by C/EBP␤ and/or GATA-4.
Animals-Intact, immature female Sprague-Dawley rats (21 days old) were obtained from Harlan (Jerusalem, Israel) and maintained under 16:8 light:dark schedule with food and water ad libitum. Animals were treated in accordance with the NIH Guide for the Care and Use of Laboratory Animals. All protocols had the approval of the Institutional Committee on Animal Care and Use, Alexander Silberman Institute of Life Sciences, Hebrew University of Jerusalem. Naive granulosa cells for CAT assays were expressed form E 2 -primed rats (35). Preovulatory PMSG-hCG-treated ovaries were prepared by administration of 15 IU of PMSG (PMSG 600, Intervet, Angers, France) to 25-day-old rats, which were further treated with 4 IU of human chorionic gonadotropin (hCG, Organon Special Chemicals, West Orange, NJ) administered (subcutaneously) 50 h later. The animals were sacrificed at 8 h after hCG, and ovaries were retrieved for protein extraction. Also, postovulatory ovaries enriched with corpora lutea were similarly harvested 72 h after onset of PMSG-hCG treatment.
Cell Cultures-To obtain granulosa cells expressing low basal levels of CAT, we used a modification of a previously described method (35). Briefly, after incubation in hypertonic sucrose/EGTA-containing medium, the ovaries were incubated for additional 45 min in Dulbecco's modified Eagle's medium/F-12 medium containing 10 M indomethacin. The same medium also served for further procedures, including needle pricking of the ovaries and post-electroporation treatments. After electroporation the cells were plated onto serum-coated wells (35) (24-well plated; Nunc, Copenhagen, Denmark) and incubated at 37°C in 95% air and 5% CO 2 .
Electroporation and CAT Assay-Estradiol-primed granulosa cells (4 ϫ 10 5 ) obtained from 3-4 ovaries were electroporated in the presence of 20 g of DNA as described before (35). Cells from each cuvette were seeded into four wells and hormonal treatments were initiated after a 3-h recovery period. Following a 6-h treatment with FSH (100 ng/ml), cell lysates were prepared and CAT activity was analyzed as described previously (35). Quantitation of the CAT assay was performed using a Fuji Bio-Imaging analyzer (BAS-1000, Fuji Photo Film, Tokyo, Japan). Protein was determined by a modified method of Bradford (36) using the Bio-Rad protein assay. Data are presented as percent of [ 14 C]chloramphenicol (Amersham International, Little Chalfont, United Kingdom) converted to its acetylated products (per protein and time of assay) and the -fold induction of CAT activity over basal values measured in the absence of hormone. Data are presented as the mean Ϯ S.E. of several independent transfections as indicated in each figure.
Whole Tissue Extracts for EMSA-Whole ovarian extracts for EMSA was performed as described before (37). Briefly, ovaries were homogenized in a Dounce homogenizer using 2-3 volumes of buffer A containing 400 mM KCl, 10 mM NaH 2 PO 4, (pH 7.4) , 10% glycerol, 1 mM EDTA, 1 mM dithiothreitol, 5 M NaF, 1 mM sodium orthovanadate, 5 g/ml aprotinin, 2 M pepstatin, and 1 mM phenylmethylsulfonyl fluoride. Following homogenization, the protein slurry was freeze-thawed three times in liquid nitrogen and finally centrifuged for 2 min at 14,000 ϫ g. After determination of the protein content, the supernatant was aliquoted and kept at Ϫ70°C until use.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assay (EMSA) was performed as described before (37). Briefly, whole cell extracts (3-15 g) were incubated with 2 ng of double-stranded DNA, previously labeled by a fill-in reaction using Klenow fragment (Promega, Madison, WI) and [␣-32 P]-dCTP (Amersham International, Little Chalfont, United Kingdom). Incubation was performed using a final volume of 30 l of buffer containing 100 mM KCl, 15 mM Tris-HCl (pH 7.5), 10 mM dithiothreitol, 1 mM EDTA, 5 mM MgCl 2 , 12% glycerol, and 4.5 g of poly(dI-dC). After incubation for 35 min at room temperature, the binding reactions were resolved on pre-run 5% acrylamide gel as described previously for quantitative RT-PCR analyses (38). When competition experiments were conducted in the presence of molar excess of cold probe, the protein extracts were added last to the reaction mixture. When antibodies were used for supershift (or ablation) of a given protein-DNA complex, the protein extracts were preincubated for 25 min at room temperature with 2-8 g of the antiserum, prior to the addition of the DNA-labeled probe. The following oligonucleotide probes used for EMSA included overhanging restriction site sequences: Promoter Constructs-A previously published sequence of the 5Јflanking region of the mouse StAR gene (30), was used to clone most of the StAR promoter constructs by a PCR-based approach. 5Ј-HindIII and 3Ј-XbaI cloning sites were included in all forward and reverse primers, respectively. To generate the Ϫ1002/ϩ6 DNA fragment, the oligonucleotide sequence Ϫ1002/Ϫ982 (5Ј-GGCCAAGCTTTTCTAAGGTTC-CCTGGATCT) and Ϫ14/ϩ6 (5Ј-GGCCTCTAGAAGCTGTGGCGCA-GATCAAGT) were included in the PCR reaction (38) using mouse genomic DNA (prepared as described in Ref. 39), as template. The PCR product was digested with HindIII and XbaI (New England Biolabs) and ligated (T4-DNA ligase; Roche Molecular Biochemicals, Mannheim, Germany) into the HindIII and XbaI sites of a promoterless pCAT-Basic vector (Promega, Madison, WI). The resulting construct was designated Ϫ1002StAR. Two deletion constructs were generated from Ϫ1002StAR by StuI and AccI (New England Biolabs, Beverly, MA) to generate the Ϫ823/ϩ6 (Ϫ823StAR) and Ϫ257/ϩ6 (Ϫ257StAR) constructs, respectively.
To generate constructs with point mutations, oligonucleotides containing the point mutations of choice were used for the PCR reaction as the forward primers, using the Ϫ14/ϩ6 as the reverse primer (unless stated otherwise): Ϫ96rStAR construct was built using the Ϫ96StAR primer as the forward primer, and the Ϫ53/ϩ6mutStAR (5Ј-GGCCTCTAGATGATctcgacgtccaggac-gcAAGCATTTAAGGCAGAGCACTTGATCTGCGCCACAGCT) as the reverse primer.
Western Blot and RT-PCR Analyses-At the indicated time points, granulosa cells were extracted by lysis buffer (RIPA) and analyzed by SDS-PAGE and electro-blotting procedures as described previously (40). After a 1-h incubation with anti-C/EBP␤ (1:2000) or anti-GATA-4 (1:4000), the nitrocellulose membranes were washed and further incubated for 1 h with the appropriate peroxidase-conjugated antibodies (1:10,000 dilution). Specific signals were detected by chemiluminescence utilizing the LumiGlo substrate (New England Biolabs). Quantitation of chemiluminescence signals on x-ray films was performed as described previously (40).
Total RNA was extracted by dissolving the granulosa cells in 0.5 ml of RNAzol B (Tel-Test, Inc., Friendwood, TX) added to each culture well (16 mm). Further steps followed the manufacturer's instructions. Semiquantitative RT-PCR analysis of total RNA extracts from granulosa cells was performed exactly as described previously (40).
Statistical Analysis-Student's unpaired two-tailed t test was performed using Microsoft Excel 97 statistical analysis functions. Differences between the activities of the indicated constructs were considered statistically significant at p Ͻ 0.05.

Is SF-1 Involved in Regulation of StAR Promoter?-Aiming
to identify the regulatory elements controlling StAR expression, we have applied transient expression assays of the mouse StAR promoter by use of granulosa cells from prepubertal rat ovary. To this end, a Ϫ1002 to ϩ6 fragment of the StAR gene was cloned by PCR and ligated to a promoterless pCAT-Basic plasmid. We reasoned that the expression of the mouse promoter in rat cells is justified by the fact that the proximal regions of the rat and mouse promoters are almost identical, in particular through the first 150 base pairs upstream to the transcription start site (Fig. 1). Testing the hormonal inducibility of the promoter constructs was performed following a 6-h incubation with FSH added shortly after transfection by electroporation. Semiquantitative RT-PCR and Western blot analyses confirmed that under similar experimental conditions the levels of StAR mRNA and protein rise acutely upon the addition of FSH (Fig. 2).
At large, the activity values obtained by transfecting a series of progressive deletions of the promoter showed that hormone inducibility remained high in all constructs pruned down to Ϫ96/ϩ6 (Fig. 3). The latter region exhibited the highest -fold induction by FSH (44-fold), suggesting that two potential upstream binding sites for SF-1 (Ϫ139/Ϫ132 and Ϫ102/Ϫ95), are not necessarily required for the FSH activation of the promoter. These results did not agree with earlier reports, which strongly advocated the notion that SF-1 is implicated in regulation of StAR expression (26 -28, 30, 41). This inconsistency, together with the fact that deletion of the Ϫ139/Ϫ132 SF-1 site signifi-cantly reduced the basal activity of Ϫ152StAR (Fig. 3), urged us to cautiously reassess the importance of this element by site-directed mutations and EMSAs. To our surprise, SF-1 did not bind to a Ϫ148/Ϫ127 probe (Fig. 4), previously shown to be capable of SF-1 binding using extracts of Y-1 adrenocortical cell line (30). Instead, the rat cell extracts generated a slower migrating protein complex, which was not affected by antiserum to SF-1 (Fig. 4A, lanes 2 and 4). A closer examination of this sequence revealed a potential Sp1 site, which is overlapping the SF-1 binding element to create an "Sp1"/SF-1 motif (see Fig. 5B, probe 2). This G-rich element (Ϫ146, 5Ј-TGG-GAGGGAG, lower strand) is nearly identical to an Sp1-like binding sequence previously reported to be involved in cAMPdependent regulation of the bovine P450scc transcription (34,42). In StAR promoter, this Sp1-like site, termed "Sp1," binds a protein that is antigenically cross-reactive with Sp1 antiserum (Fig. 4B, lanes 6 and 8). Moreover, molar excess of Sp1 consensus DNA can compete for the binding of "Sp1" to its site in StAR promoter (Fig. 5A, lanes 4 and 5). Finally, a site-directed mutation replacing GG with ca (Fig. 5B, lane 16) resulted in the loss of "Sp1" band shift and rendered the SF-1 site available for a typical SF-1 binding (Fig. 5B, lane 16). Noteworthy, the "Sp1"/SF-1 element could bind both proteins, providing the extracts were prepared from the mouse MA-10 cells (Fig. 5B,  lane 14), which are highly enriched with SF-1 content. These results suggest that the Ϫ148/Ϫ127 region has a dual capacity to bind both "Sp1"and SF-1, which compete with each other depending on their relative content in a given cell type.
The apparent cooperativity of Sp1 and SF-1 in mediating cAMP-driven expression of steroidogenic genes (34, 42) could have suggested that a similar concept might be functionally relevant for FSH induction of StAR. Therefore, CAT reporter transgenes containing point mutations in the Ϫ132/Ϫ146 region were created in the context of Ϫ152StAR, as an alternative approach for the deletion strategy described before. The results in Fig. 6 show that a 5Ј-CAAGGTGG mutation to 5Ј-CAAtaTGG (Ϫ152mutSF-1), did not affect the response of StAR promoter to hormones, but instead improved it. The same mutation was previously shown to ablate SF-1 binding and transcriptional activation of the P450scc and P450aromatase promoters (31,32). The notion that binding of SF-1 is irrelevant for transcriptional activation of StAR was further strengthened by the fact that a mutant of the "Sp1" site (Ϫ152mut2"Sp1"), which capacitated SF-1 binding (Fig. 5, lane 16), did not improve any of the construct performances (Fig. 6). Finally, when the "Sp1" site was mutated (Ϫ152mut1"Sp1"), as previously done to critically impaired its activity when harboring the bovine P450scc promoter (42), no significant loss of hormonal inducibility or basal activity were noticed (Fig. 6).
FSH Responsiveness Is Determined by the Ϫ93/Ϫ51 Region: A Role for C/EBP␤-Further 5Ј deletions of StAR promoter finally resulted in a severe loss of FSH responsiveness. Fig. 7 shows that the Ϫ73/ϩ6 and Ϫ51/ϩ6 CAT constructs retained only 12% and 0.3% of the FSH-driven activity exhibited by the Ϫ96/ϩ6 construct. To verify that no additional sequences located downstream to Ϫ51 are potentially involved in hormonal regulation of this promoter, we have randomized 17 base pairs constituting the Ϫ49/Ϫ33 region. Clearly, this mutation performed within the context of the Ϫ96/ϩ6 construct did not affect its activity (Fig. 7), suggesting that no important elements reside immediately upstream to the TATA box.
Sequence analysis of the Ϫ96/Ϫ51 region depicts a putative near-consensus C/EBP␤ site (C/EBP␤-2), residing upstream to a conserved consensus GATA binding motif (Fig. 1). To test if the C/EBP␤-2 site may have a regulatory role in the rat granulosa cells, we mutated this site in the context of Ϫ96StAR, as shown in Fig. 8. However, the mutated construct (Ϫ96mut␤-2) did not affect its 55-fold response to FSH, which was not much different from the parental plasmid (Fig. 8). Further EMSA studies provided an explanation for this observation by showing that a specific C/EBP␤-2 probe (Ϫ96/Ϫ75) did not bind to any protein (data not shown). Instead, a downstream adjacent element reminiscent of an AP-1 site, if anything else (5Ј-TGACTGA), was found capable of C/EBP␤ binding. We designated this non-consensus element C/EBP␤-3 (see Fig. 1). EM-SAs showed that a C/EBP␤-3 oligonucleotide probe (Ϫ87/Ϫ70) binds C/EBP␤ in a specific fashion; antiserum to C/EBP␤ caused a supershift and ablation of the typical triplet bands  Progressive deletions of StAR promoter are schematically illustrated in the left panel. Each vector was transfected by electroporation into E 2 -primed rat granulosa cells. Cells without hormone treatment (dotted bars) and those treated with 100 ng/ml FSH (solid bars) were incubated for 6 h before extracts were prepared for CAT analysis as described under "Experimental Procedures." CAT activity was determined using 5 g of protein for a 5-h assay. The results are presented as the mean Ϯ S.E. of percent of [ 14 C]chloramphenicol converted to the acetylated products. Hatched bars represent the FSH -fold induction above basal activity. Multiple independent transfections (n) were performed for each construct. Activity levels were statistically significant when compared with the respective values obtained for the Ϫ1002StAR construct: a, p Ͻ 0.05; b, p Ͻ 0.001; b* denotes statistically different values when compared with the corresponding activities of the Ϫ152StAR construct (p Ͻ 0.001).
Despite the fact that the sequence of C/EBP␤-3 site (TGACT-GATGA) is so remotely different from a consensus C/EBP␤ motif, it is absolutely required for C/EBP␤ binding, which was lost upon a TGAtctgTGA mutation of the probe (Fig. 9B, lane  16). Yet, as could be expected, the C/EBP␤-3 motif does not necessarily exhibit the best affinity for C/EBP␤ binding, as we learn from competitive EMSAs mixing a 32 P-C/EBP␤-3 probe (Ϫ87/Ϫ70) with a 10 -100-fold molar excess of the near-consensus C/EBP␤-1 sequence (Ϫ125/Ϫ100); in doing so, a 10-fold excess of C/EBP␤-1 was enough to displace 90% of the labeled C/EBP␤-3 probe (Fig. 9, lane 14). The reason for using the C/EBP␤-1 sequence for competitor DNA lies in the curious fact that its motif (ATGAGGCAAT) specifically binds immunocross-reactive C/EBP␤ (Fig. 9C). However, mutating or deleting the C/EBP␤-1 site does not impair transcriptional activation of the StAR promoter (Fig. 3, bottom two constructs). By contrast, functional analysis of the C/EBP␤-3 sequence by sitedirected mutagenesis (Fig. 8, Ϫ96mut␤-3) suggested that an intact C/EBP␤-3 site is, indeed, required for the activity of StAR promoter.
The Role of GATA-4 -However, the moderate attenuating effect caused by mutating the C/EBP␤-3 site could not account for the severely impaired activity of the Ϫ51StAR construct (Fig. 3). Therefore, the involvement of an additional regulatory element downstream of C/EBP␤-3 was likely to be found. To test this possibility, we examined the candidacy of a perfect GATA binding site located at Ϫ61/Ϫ66 (Fig. 1). Like Ϫ96mut␤-3, modification at this GATA site (Ϫ96mutGATA) resulted in no more than a partial attenuation of the construct activity (Fig. 8). However, double mutation of both C/EBP␤-3 and GATA sites (Ϫ96doublemut) resulted in a nearly complete loss of CAT activity (Fig. 8). A similar marked impairment of the FSH responsiveness was also observed when the GATA site was mutated in the context of Ϫ73mutGATA, from which the C/EBP␤-3 motif was deleted. In our view, the residual but not negligible FSH -fold induction observed in cells transfected with the double-mutated construct (Ϫ96double mut) is probably meaningless and reflects a pleiotropic effect FSH has on the basal activity by acting as a trophic hormone. In fact, the FIG. 5. Ϫ148/؊127 StAR promoter region binds an Sp1-like ("Sp1") protein. Panels A and B include EMSA performed as described in Fig. 4 by use of the following 32 P-labeled probes: consensus Sp1 binding site (Cons. Sp1 (1); Ϫ148/Ϫ127 (2); an SF-1 binding site, SCC1(SF-1) (3); a mutated Ϫ148/Ϫ127 probe designated mut2"Sp1" (4). A, an ovarian extract from PMSG/hCG-treated rats was used as a source for DNA-binding proteins. Lanes 1-3 depict a competition experiment using unlabeled self-competitor Ϫ148/Ϫ127 DNA, lanes 4 and 5 examine the ability of molar excess of unlabeled consensus Sp1 DNA (Cons. Sp1) to compete for the labeled Ϫ148/Ϫ127 probe, and lanes 6 -10 examine reciprocal competitions using a labeled consensus Sp1 binding site probe. B, lane 11 depicts an Sp1 binding to a consensus Sp1 probe. Lanes 12 and 13 show the formation of an "Sp1" complex when a Ϫ148/Ϫ127 probe was incubated with either a rat, or a mouse ovarian extract, respectively. The same Ϫ148/Ϫ127 probe yielded both SF-1 and "Sp1" bands when tested with an extract of mouse MA-10 Leydig cell line (lane 14). Lane 15 demonstrates a typical SF-1 binding to its consensus motif (SCC1). Lane 16 depicts the effect of mutating the Ϫ148/Ϫ127 probe to mut2"Sp1" (probe 4).
FIG. 6. Site-directed mutations of the "Sp1"/SF-1 site do not affect the hormonal activation of StAR-CAT. PCR-based mutations were created within the context of Ϫ152/ϩ6 of StAR, and ligated to pCAT-Basic (Ϫ152StAR). Mutated sequences of the "Sp1" (superscript dotted line) and the SF-1 (underline) motifs are indicated by lowercase letters. Transient expression was conducted in the absence or the presence of FSH as described in Fig. 3. Hatched bars represent the FSH -fold induction above basal activity. CAT activity was determined using 5 g of protein for a 2-h assay. The results are presented as the mean Ϯ S.E. of percentage of [ 14 C]chloramphenicol converted to its acetylated products. Three independent electroporations were conducted with these constructs. a, p Ͻ 0.05 when compared with the basal activity value of Ϫ152StAR. The rest of the data were not statistically different. activity performances of the double mutant were similar to those obtained for the maximally trimmed promoter construct Ϫ51StAR, which also exhibited a significant 5-fold induction of CAT activity by FSH (Fig. 7). However, this trophic improvement of the basal activity was no higher than 3% of the FSH activity measured for the parental Ϫ96StAR construct (Fig. 7) and, therefore, could not represent other than a misleading basal activity Fig. 10 shows a single protein complex labeled by either of the two DNA probes, Ϫ73/Ϫ42, or Ϫ73/ϩ6 (Fig. 10, A and B,  respectively). These results convince that no other protein complexes can bind to the entire region residing between Ϫ73 and ϩ6. The binding specificity was demonstrated by use of antiserum to GATA-4, which ablated and supershifted the DNAprotein complex. The latter remained unaffected in the presence of anti-GATA-6 serum (Fig. 10A, lanes 1 and 2). Additionally, mutating the GATA site in the context of the Ϫ73/ϩ6 probe (5Ј-TTATCT 3 5Ј-TTAagT) associated with loss of binding capacities, as evident by competition (Fig. 10B, lanes  7 and 8) and direct binding studies using radiolabeled probe (Fig. 10B, lane 9).

Granulosa Cell Expression of C/EBP␤ and GATA-4 -
We also aimed to study by Western blot analysis the levels of C/EBP␤ and GATA-4 proteins in our tissue and cell preparations. First, we tested granulosa cell extracts prepared at identical time points previously used for determination of CAT activity. Fig. 11A shows that the C/EBP␤ antiserum crossreacted with three protein bands of 45, 39, and 22 kDa, known as C/EBP␤ isoforms (43). The levels of those proteins were barely detectable in vivo (t 0 ), but seeding the cells for a few hours in culture substantially elevated the levels of the higher molecular weight forms. The 22-kDa isoform of C/EBP␤ was not affected by seeding to culture (Fig. 11A). A 6-h treatment with FSH generated an increase of all the isoforms of C/EBP␤ up to 7-fold. In agreement with our EMSAs (Fig. 9A), Western blot analysis of the ovarian cells detected C/EBP␣ protein bands (data not shown), which were also reported by earlier studies of these proteins during follicular development (44).
By contrast to the profile of C/EBP␤ expression, neither seeding nor FSH treatment had any effect on the high level of the GATA-4 content, which is probably constitutively expressed in the rat granulosa cells (Fig. 11B). It should be noted  (Ϫ96mut␤-2, Ϫ96mut␤-3), or the GATA factors (Ϫ96mutGATA). Additionally, double-mutated constructs were prepared as either Ϫ96double mut, or Ϫ73mut-GATA. Transfection and activity assays were conducted as described in Fig. 3. The results are presented as the mean Ϯ S.E. of percentage of [ 14 C]chloramphenicol converted to the acetylated products.
Hatched bars represent the FSH -fold induction above basal activity. Multiple independent transfections (n) were performed for each construct. Activity levels were statistically significant when compared with the respective values obtained for the Ϫ96StAR construct: a, p Ͻ 0.05; b, p Ͻ 0.01; c, p Ͻ 0.001. that, in correlation to our EMSA data, we could not detect the GATA-6 protein by this Western blot analysis (data not shown). Fig. 11C shows that the C/EBP␤ and GATA-4 proteins also exist at high levels in the ovarian extracts we selected, for practical reasons, as a source for our EMSAs. Those extracts were expected to express the necessary factors since StAR expression during the post-hCG period is in its prime response (40). DISCUSSION The objective of the present study was to reveal the principal sequence elements that render the promoter of StAR gene responsive to FSH in cells of the rodent ovary. Based on previously published sequence (30), we cloned a Ϫ1002 to ϩ6 fragment of the 5Ј-flanking region of the mouse StAR gene and placed it upstream of CAT gene in a promoterless pCAT-Basic reporter plasmid. Then, by contrast to most of the previous StAR studies expressing promoter-reporter plasmids in established cell lines (26 -30, 41), we conducted our functional assays using primary naive granulosa cells from prepubertal rats. Following electroporation of these cells in the presence of plasmid DNA, a 6-h incubation in culture resulted in a robust activation (up to 50-fold) of the promoter by FSH. Having such FIG. 9. The ؊87/؊70 StAR promoter region binds C/EBP␤ proteins. A, electrophoretic mobility shift assays using PMSG/hCG ovary extract were performed as described in Fig. 4 using a Ϫ87/Ϫ70 32 P-labeled probe (lanes 1-15). The formation of the three DNA-protein complexes (arrows, lanes 1 and 4) was examined in the presence of the following antisera: C/EBP␣ (lane 2), C/EBP␤ (lanes 3 and 7), sterol-responsive element-binding protein, SREBP ( lane 5 and 6), c-Fos (lane 8), and c-Jun (lane 9). B, competition studies using the 32 P-Ϫ87/Ϫ70 as a probe (lane 11) and molar excess of unlabeled selfcompetitor DNA (lanes 12 and 13), or unlabeled C/EBP␤-1 DNA (lanes 14 and 15). Lane 16 demonstrates lack of binding to a mutated C/EBP␤-3 labeled probe. C, characterization of C/EBP␤ binding to a C/EBP␤-1 probe (Ϫ125/Ϫ100) was performed in the presence of either an antiserum to C/EBP␤ (2 g, lane 19), or anti-SF-1 serum as a negative control (4 g, lane 18).  5 and 6), or an unlabeled Ϫ73/ϩ6mutGATA probe (lanes 7 and 8). Lane 9 tests the capacity of the latter mutated probe to complex with proteins of the ovarian extract.
FIG. 11. Effect of seeding and FSH treatment on expression of C/EBP␤ and GATA-4 proteins. Except for electroporation, granulosa cells from E 2 -primed immature rats were prepared for culture exactly as described in Fig. 3. Protein extracts were prepared from cells prior to seeding (t 0 ), or after a 9-h incubation (post-seeding) in the absence of FSH. FSH treatment commenced 3 h after seeding and cells were harvested 6 h later. SDS-PAGE and Western blot analysis were performed as described under "Experimental Procedures" using 40 g of protein/lane. Chemiluminescent detection of the reactive proteins was performed using antisera to C/EBP␤ (A) or GATA-4 (B). The antiserum to C/EBP␤ detects three major isoforms, designated LAP-I, LAP-II, and LIP (43). The lower panel presents a semiquantitative analysis of the data shown in A and B. C, demonstration of C/EBP␤ and GATA-4 proteins in ovarian extracts prepared from the PMSG/hCG-treated rats. a sensitive assay in hand, we combined a progressive deletion strategy with site directed mutations aiming to validate, or eliminate, the potential involvement of candidate cis-acting regulatory elements in StAR expression. This approach led us to identification of two novel elements, C/EBP␤ and GATA-4, which were never known before to be required for transcriptional activation of genes encoding steroidogenic proteins.
High responsiveness to FSH still remained if the promoter was pruned down to position Ϫ96 from the transcription start site. However, we were interested in performing a limited analysis upstream of the Ϫ96 regions, in particular addressing those sites that were previously suggested to be functionally involved in regulation of StAR expression. For example, between Ϫ100 and Ϫ1000 of the StAR promoter reside at least two putative binding sites for the tissue-specific orphan receptor SF-1 (45), also known as Ad4BP (24). This factor is required for activation of many genes expressed in steroidogenic tissues (23). Therefore, a host of studies have recently proposed that SF-1 is also essential for expression of StAR gene (26 -30, 41). Using the naive granulosa cell model, our findings do not support this notion since the removal of all putative SF-1 sites (28,30) hardly affected the promoter performance (Ϫ96StAR and Ϫ96rStAR, Figs. 2 and 7, respectively). A possible explanation for the apparent discrepancy between the present findings and earlier works may reflect species-dependent differences between the human (26) and the murine (30) promoters. However, one cannot exclude the possibility that the StAR promoter was never examined in steroidogenically committed primary cells which contain normal levels of SF-1. Instead, the role of SF-1 was demonstrated in luteal cells (27,46) and steroidogenic cell lines expressing exceptionally high endogenous levels of the orphan receptor (30,41), or by use of nonsteroidogenic cells overexpressing transfected SF-1 plasmids (26 -29, 41). Therefore, the role of SF-1 was always examined under conditions favoring its interaction with StAR promoter, but did not necessarily simulate the physiologically compatible requirements for regulation of StAR in normal cells.
Upstream of position Ϫ96 in the mouse promoter resides an additional element of considerable interest, i.e. a nonconsensus putative binding site for Sp1 (Ϫ137/Ϫ146). A similar sequence for binding of an Sp1-like protein ("Sp1") was found important for cAMP induced transcription of the bovine P450scc gene (42). Moreover, an analogous Sp1 site seemed to be important for activation of the human StAR promoter (26). In rodents, however, this "Sp1" element is overlapping a well studied SF-1 binding motif, thus creating a mixed "Sp1"/SF-1 site. This composite sequence can bind SF-1 from Y-1 (30) or MA-10 cells (Fig. 4B), but is apparently incapable of doing so when tested with granulosa cell extracts. We may, therefore, conclude that "Sp1" and SF-1 compete for binding to this site, so that SF-1 band shift in EMSAs is demonstrable only in cell lines expressing extremely high levels of SF-1, as discussed above. However, at present it is not clear if activation of the StAR promoter in granulosa cells involves "Sp1" action, since removal of the "Sp1"/SF-1 site down to position Ϫ123 attenuated the basal activity, but retained the hormonal responsiveness intact. Interestingly, further pruning of the promoter down to Ϫ96 kept suppressing the basal activity of the promoter by removing a near-consensus C/EBP␤ (C/EBP␤-1) binding moiety at Ϫ116/Ϫ107. Thus, the C/EBP␤-1 site is irrelevant for activation of the promoter despite its apparent excellence in binding typical C/EBP␤ proteins (Fig. 9). Collectively, these results have suggested that the critical element(s) controlling the inducibility of StAR promoter must have resided further downstream, within the first 80 -90 base pairs of the promoter.
Indeed, deletion mutants and site-directed mutation within the Ϫ87 to Ϫ51 region revealed the involvement of trans-acting proteins that bind two sites located 10 base pairs apart: an upstream sequence interacting with C/EBP␤ (C/EBP␤-3) and a consensus site specifically binding a GATA-4 protein. Contrary to our earlier expectations, the sequence core subserving for C/EBP␤ binding (TGACTGA) seemed more like an AP-1 site, so that a typical c-Jun/c-Fos binding was demonstrable upon a single base substitution to TGACTcA (data not shown). Even more confusing was the fact that overlapping with this odd C/EBP␤-3 site resides another putative recognition element for C/EBP␤, designated C/EBP␤-2. However, mutating the core sequence of C/EBP␤-2 (ACAAT to tccca) did not affect the inducibility of the Ϫ96StAR construct, while a four-base mutation of C/EBP␤-3 reduced the basal and FSH-responsive activity by 50%. This shy effect of the mutated C/EBP␤-3 did not impress much until after the elucidation of the GATA binding site at Ϫ66/Ϫ61. Mutation of the GATA site resulted in a 45% drop of the promoter activity, but a double mutation of both GATA-4 and C/EBP␤-3 resulted in a dramatic 97% loss of the construct activity. These results suggest that a concurrent binding of the two regulatory proteins is required for activation of StAR promoter.
The present study shows that the prepubertal granulosa cells already express GATA-4 in vivo, and the level of this protein remained unaffected under any culture manipulations. By contrast, the C/EBP␤ proteins are absent in vivo, but their expression in culture was markedly promoted by FSH. A similar induction of C/EBP␤ expression was recently documented in testicular Leydig cells responding to cAMP (47). These results suggest that the level of C/EBP␤ might determine the rate-limiting regulatory switch of StAR transcription, while the constitutively expressed GATA-4 plays an essential, yet more permissive role for that matter. It is noteworthy that, other than StAR, members of the C/EBP family have been described as regulators of acute responses, such as the control of certain inflammatory functions (48). Moreover, the suggested involvement of C/EBP proteins in the acute response of StAR to cAMP (8,20) and trophic hormones (40,49,50) is also well accepted in light of a well studied example where C/EBP plays a critical role in transcriptional regulation of the phosphoenolpyruvate carboxykinase (PEPCK) gene (51). Similar to the emerging scenario in StAR promoter, it has been shown that cAMP and probably protein kinase A drive PEPCK transcription via a complex interactions of C/EBP with other activators and coactivators (51); also, the binding of C/EBP␤ to a non-consensus recognition motif in the StAR promoter is reminiscent of the C/EBP binding to non-dyad-symmetric sequences in the PEPCK promoter, a cyclic AMP-responsive element-like site (52) and a site termed P3(I) (53).
Our EMSAs showed that the C/EBP␤-1 and C/EBP␤-3 probes formed three DNA-protein complexes, which were identical to those obtained when binding capacities of rat granulosa cell extracts were previously examined with a C/EBP␤ motif located in the hormone-inducible prostaglandin synthase-2 promoter (54). The alternative approach testing the potential cross-reactivity of the C/EBP␤ proteins with specific antibodies tested by Western blot analysis revealed that the rat granulosa cells express three isoforms, previously identified as liver-enriched activating proteins (LAPs) and liver-enriched inhibiting protein (LIP) (43,55). Clearly, LIP and LAP-I were markedly elevated as a result of FSH treatment, while LAP-II level rose by merely seeding the cells into culture. Therefore, it is conceivable to propose that the onset of StAR transcription is controlled by a two-step mechanism: first, the seeding-induced rise of LAP-II capacitates the immediate response of the granulosa cells to FSH, and a follow-up activation of the promoter may proceed thanks to the hormone-elevated levels of LAP-I. The likelihood of such a mechanism is not too speculative since the granulosa cells express high levels of FSH-activated cyclic AMP-responsive element-binding protein (56,57), which in turn, is known to up-regulate LAP transcription (58,59). Further studies should address this hypothesis and reveal even more about the potential involvement of the C/EBP␤ isoforms in regulation of StAR expression. Also, it is not unlikely that other C/EBP proteins, such as C/EBP␣, can interact with the Ϫ87/Ϫ70StAR site. If truly so, C/EBP␣ should be considered as potential substitute for C/EBP␤ in the C/EBP␤-deficient mouse ovary (60 -63). It has been shown that the ovaries of the latter null mice do not ovulate (60, 64) despite normal production of gonadal steroids and expression of steroidogenic enzymes, probably including StAR. 3 Like C/EBP␤, GATA-4 had not been described previously as a potential regulator of genes encoding steroidogenic enzymes, or their accessory proteins. Gata4-null mice die in utero by 9.5 days postcoitum (65,66). However, recent studies have suggested that this transcription factor is involved in control of gonadal development and sex differentiation in rodents (67)(68)(69). Moreover, in adult mouse tissues, high levels of GATA-4 are observed in the ovary, testis, and heart. Interestingly, little or no GATA-4 protein was found in the cells of the mouse corpus luteum (68), which do express record levels of StAR (20,40,70,71). One way to reconcile this apparent inconstancy is provided by assuming that the role of GATA-4 during follicular phase could be fulfilled by another member of this family, like GATA-6, which is highly abundant in the corpus luteum (68). Alternatively, we may propose that the mode of StAR expression in the corpus luteum may not necessarily resemble its regulatory pattern in granulosa cells of the follicular phase. Accordingly, SF-1, which is extremely high in the corpus luteum may control the gland expression of StAR expression after all. This tempting speculation implies that activation of StAR transcription may be achieved by more than one set of cis-acting proteins, depending on the origin of the cells and tissue under study. In this regard, the present findings raise the question: what tissue-specific factor(s) can potentially replace SF-1 in our steroidogenically committed, but yet undifferentiated cell model? Alternative experimental approaches addressing this challenging question are currently under study.