CCAAT/Enhancer-binding Proteins Regulate Expression of the Human Steroidogenic Acute Regulatory Protein (StAR) Gene*

Two putative CCAAT/enhancer-binding protein (C/EBP) response elements were identified in the proximal promoter of the human steroidogenic acute regulatory protein (StAR) gene, which encodes a key protein-regulating steroid hormone synthesis. Expression of C/EBPα and -β increased StAR promoter activity in COS-1 and HepG2 cells. Cotransfection of C/EBPα or -β and steroidogenic factor 1, a transcription factor required for cAMP regulation of StAR expression, into COS-1 augmented 8-bromoadenosine 3′:5′-cyclic monophosphate (8-Br-cAMP)-stimulated promoter activity. When the putative C/EBP response elements were mutated, individually or together, a pronounced decline in basal StAR promoter activity in human granulosa-lutein cells resulted, but the fold stimulation of promoter activity by 8-Br-cAMP was unaffected. Recombinant C/EBPα and -β bound to the two identified sequences but not the mutated elements. Human granulosa-lutein cell nuclear extracts also bound these elements but not the mutated sequences. An antibody to C/EBPβ, but not C/EBPα, supershifted the nuclear protein complex associated with the more distal element. The complex formed by nuclear extracts with the proximal element was not supershifted by either antibody. Western blot analysis revealed the presence of C/EBPα and C/EBPβ in human granulosa-lutein cell nuclear extracts. C/EBPβ levels were up-regulated 3-fold by 8-Br-cAMP treatment. Our studies demonstrate a role for C/EBPβ as well as yet to be identified proteins, which can bind to C/EBP response elements, in the regulation of StAR gene expression and suggest a mechanism by which C/EBPβ participates in the cAMP regulation of StAR gene transcription.

The translocation of cholesterol from the sterol-rich outer mitochondrial membrane to the cholesterol-poor inner mitochondrial membrane is the rate-limiting step in steroid hormone synthesis (1). The steroidogenic acute regulatory protein (StAR) 1 has an integral role in this cholesterol translocation process in highly steroidogenic tissues such as the ovary, testis, and adrenal gland as evidenced in experiments of nature and mouse gene knockout studies. Congenital lipoid adrenal hyperplasia, a disease in which the production of all adrenal and gonadal steroids is severely impaired prior to the synthesis of pregnenolone, is caused by mutations that inactivate the StAR protein (1,2). Targeted disruption of the mouse StAR gene results in a phenotype identical to human congenital lipoid adrenal hyperplasia in nullizygous animals (3).
StAR gene expression as determined by in situ hybridization and Northern blot analysis in gonadal and adrenal cells revealed that the steroidogenic capacity of these cells was tightly linked to the abundance of the StAR transcripts (4,5). Transcriptional control of the human StAR gene has only recently been examined with initial studies focusing on the orphan nuclear receptor, steroidogenic factor-1 (SF-1) (6) because of its known role in the regulation of other key genes involved in steroidogenesis (e.g. P450scc, P450c17, aromatase, 3␤-hydroxysteroid dehydrogenase) (7). Indeed, the human StAR promoter was shown to have at least three SF-1 binding sites that are functionally important for basal as well as cAMP-stimulated transcription (8). SF-1 has also been shown to be important for transcriptional regulation of the mouse (9), rat (10), and bovine StAR genes (11). Sequence analysis also indicated the presence of a variety of other transcription factor binding sites including multiple putative Sp-1 binding sites, several of which were recently implicated in transcriptional control of StAR gene expression (12). Deletion analysis of the StAR promoter suggested that other transcription factors were also likely to play an important role in StAR gene expression. In addition to the known SF-1 response elements and a novel DAX-1 hairpin loop binding site (13), the StAR promoter contains several putative transcriptional response elements including a series of three sites at Ϫ80 to Ϫ69 that resemble sterol regulatory elementbinding protein (SREBP) motifs, the most proximal of which is similar to a Ying Yang 1 (YY1) transcription factor binding site. The presence of SREBP binding sites is consistent with our recent demonstration that SREBP-1a stimulates StAR reporter activity (14).
Recent observations revealed that ovarian CCAAT/enhancer-binding proteins (C/EBPs), basic leucine zipper transcription factors implicated in the regulation of a variety of genes involved in energy metabolism and cell differentiation pathways (15), are modulated by luteinizing hormone, which acts on ovarian cells through the intermediacy of cAMP, and that steroidogenic cell function is altered in mice upon loss of C/EBP␣ or -␤. Gene knockout studies (C/EBP␤) (16) and antisense experiments (C/EBP␣) (17) demonstrated that loss of either of these two members of the C/EBP transcription factor family results in the failure of differentiation of the ovulatory follicle as evidenced by the lack of ovulation and formation of the corpus luteum. The focus of the present studies was to determine if C/EBPs modulate basal and cAMP-stimulated StAR gene expression.

EXPERIMENTAL PROCEDURES
Plasmids-The cDNA for mouse SF-1, a generous gift from Dr. K.L. Parker (Southwestern Medical Center, Dallas, TX), was cloned into the pSV-SPORT-1 expression vector using standard procedures. The pMEX-C/EBP␣ and pMEX-C/EBP␤ expression plasmids were described previously (18). The pGL2-basic vector (Promega) was the source of the luciferase reporter gene for the 1.3-kilobase human StAR promoter and all other StAR promoter constructs (19). The pGL2-basic vector was also the source of the luciferase reporter gene for DEI-4, which contains 4 copies of the DEI element, a known C/EBP response element, upstream of the albumin minimal promoter (20). The ␤-galactosidase expression vector (pCH110, Amersham Pharmacia Biotech, or pCMV-␤-galactosidase) was used for normalization of luciferase data. Mutant StAR reporter constructs were prepared using the wild-type Ϫ235 StAR promoter construct. We introduced three base pair substitutions (i.e. replacing GCA in the wild type with TTT) in the C/EBP sites at Ϫ119/ Ϫ110/Ϫ110 and Ϫ50/Ϫ41 (see Fig. 1) that are known to disrupt C/EBP binding sites using the transformer site-directed mutagenesis kit (Promega, Madison, WI). The mutant StAR promoters will be referred to as the Ϫ119/Ϫ110 C/EBP mutant, the Ϫ50/Ϫ41 C/EBP mutant, and the Ϫ119/Ϫ110 ϩ Ϫ50/-41 double C/EBP mutant. Sequence analysis verified the presence of mutations. Plasmids for transfection were prepared using the Qiagen Maxiprep system.
Cell Culture and Transfection-COS-1 and HepG2 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 50 g of gentamycin/ml at 5% CO 2 and 37°C for cell propagation and plating. These cells were obtained from the American Type Culture Collection (Manassas, VA). Proliferating human granulosa cells were prepared and cultured as described previously by McAllister et al. (21). COS-1 and HepG2 cells were plated at 40,000 (FUGENE-6) or 60,000 (LipofectAMINE) cells/well in 12-well plates, respectively, on day 0. On day 1, cells were washed twice in DMEM alone and then transfected with LipofectAMINE (Life Technologies Inc.) plus plasmid DNA or were transfected with FUGENE-6 (Roche Molecular Biochemicals) plus plasmid DNA in the presence of serum as described by the manufacturer's protocol. Cells were transfected with 500 ng of each of the ␤-galactosidase expression and reporter vectors, 500 ng of the SF-1 expression plasmid, and 1-500 ng of the C/EBP expression plasmids. After a 2-h exposure to the DNA⅐LipofectAMINE complex, an equal volume of DMEM ϩ 20% fetal calf serum was added to each well and left overnight. Cells transfected with FUGENE-6 received no further manipulation until day 2. On day 2, medium was changed to DMEM ϩ 10% fetal calf serum, and 8-bromoadenosine 3Ј:5Ј-cyclic monophosphate (8-Br-cAMP; Sigma) was added to selected wells. After 24 h, cells were harvested by scraping the cells in reporter lysis buffer (Promega) followed by a single freeze/thaw cycle. Transfection of human granulosa cells utilized the FUGENE-6 protocol as described by the manufacturer. Human granulosa cells were transfected overnight with 500 ng of the StAR reporter constructs and 100 ng of pCMV-␤-galactosidase expression vectors. On day 2 cells were then cultured in DMEM/F12 with 10% fetal calf serum in the presence and absence of 8-Br-cAMP for 24 h. Cell extracts were assayed for luciferase and ␤-galactosidase activity. Human granulosa cells (60 -80% confluent) for Western blot analysis of nuclear levels of C/EBP␣ and -␤ were cultured in the absence and presence of 8-Br-cAMP for 24 h before harvesting.
Enzymatic Assays-Luciferase activity was determined in a LUMAT LB 9507 luminometer (EG&G Berthold) with Promega luciferin as substrate as described previously (14). ␤-galactosidase activity was determined by a standard colorimetric assay using 2-nitrophenyl ␤-Dgalactopyranoside as substrate. Relative luciferase activity (RLU) for each well was determined by dividing luciferase relative light units by the ␤-galactosidase activity (A 420 ). Preparation of Nuclear Extracts and Western Blot Analysis-Nuclear extracts from human granulosa cells were generated as described previously (22). Protein concentrations of the extracts were determined by the Bio-Rad dye binding assay. Equal amounts of protein (100 g) were loaded for SDS-polyacrylamide gel electrophoresis. After electrophoresis, proteins were transferred to polyvinylidene difluoride membranes for probing with antibodies to C/EBP␣ (rabbit polyclonal sc-61, Santa Cruz Biotechnology Inc., Santa Cruz, CA) and C/EBP␤ (rabbit polyclonal sc-150, Santa Cruz Biotechnology Inc.). The ECL kit (Amersham Pharmacia Biotech) was used for antigen⅐antibody complex detection.
Electrophoretic Mobility Shift Assays-Double-stranded oligonucleotide probes (see Fig. 1) 24 -32 bases in length with 4-base overhangs (CTAG) based on the wild-type StAR promoter sequence were labeled with [␣-32 P]dCTP by fill-in reaction with DNA polymerase I Klenow (large fragment, NEN Life Science Products). Mutant probes for the StAR oligonucleotides had the same base pair substitutions as indicated by the bold letters in Fig. 1 for mutant StAR reporter constructs. The C/EBP oligonucleotide (positive control) was described previously (20). The labeled probes were used in electrophoretic mobility shift assays as described previously (20). Briefly, recombinant C/EBP␣, C/EBP␤, or granulosa cell nuclear extracts (5 g) were incubated with labeled oligonucleotide probes for 20 min on ice before loading onto prerun acrylamide gels. Where indicated, antibodies to C/EBP␣ or -␤ were incubated with recombinant proteins/nuclear proteins for 20 min on ice prior to the addition of oligonucleotides. Additionally, unlabeled oligonucleotides (10 -50-fold excess) were added to selected samples to demonstrate specificity of binding.
Statistical Analysis-Results of promoter analysis studies were analyzed by analysis of variance (ANOVA) using the JMP 3.1 program (SAS Institute Inc., Cary, NC). Following the observation of a significant (p Ͻ 0.05) F-test, differences between means were determined using the Tukey-Kramer mean separation test. Fig. 1) indicated that the human, bovine, rat, and mouse StAR gene promoters exhibit a high degree of sequence similarity within the first ϳ130 bases from the transcriptional start site. Shaded nucleotides shown in Fig. 1 illustrate sequences that are conserved in three of the four species. The two known SF-1 response elements in the human promoter are shown, as are the putative SREBP-1a/YY1 binding sites. Two highly conserved sequences resembling CCAAT/enhancer-binding protein sites are located immediately 5Ј to the human SF-1 response elements. The site located at Ϫ119 to Ϫ110 contains a perfect consensus half-site "GCAAT," whereas the site at Ϫ50 to Ϫ41 diverges from the consensus motif with only 5 of 10 bases being similar to the consensus sequence. However, this region of the promoter is highly conserved across species with only the bovine sequence showing any divergence.

Identification of Motifs Resembling CCAAT/Enhancer-binding Protein Response Elements-Sequence alignment (
Promoter Analysis in COS-1 and HepG2 Cells-To determine FIG. 1. Sequence homology of the human, bovine, mouse, and rat StAR gene promoters and the known and putative response elements in the human StAR gene are depicted. Shaded bases mark those bases in the human, bovine, rat, and mouse StAR gene promoters that are identical in three of the four sequences. Known SF-1 binding sites in the human StAR promoter and the putative regions that interact with SREBP and YY1 are boxed. The DAX-1 DNA hairpin loop encompasses bases Ϫ61 to Ϫ27 (not shown). The location of the C/EBP sites and the mutations that block C/EBP binding (bold bases) are indicated below the human StAR promoter sequence. The sequences of the oligonucleotides used for the electrophoretic mobility shift assay are also shown. All oligonucleotides included 4-base pair extensions that were used for radiolabeling purposes (fill-in reaction).
if the C/EBP␣ and -␤ transcription factors are functionally important for StAR gene expression, we cotransfected COS-1 and HepG2 cells, which do not express StAR endogenously, with StAR promoter-luciferase constructs and plasmids that express these transcription factors. These cells were used to test StAR promoter activity in cells that do not express endogenous SF-1. Fig. 2 shows the results of studies in COS-1 and HepG2 cells transfected with 50 ng (maximal stimulatory dose) of either C/EBP␣ or -␤ with either the 1.3-kilobase StARluciferase reporter or the DEI-4-luciferase reporter, a control C/EBP-responsive promoter construct. In COS-1 cells, the StAR and DEI-4 promoter activities increased in a dose-dependent manner with expression of both C/EBP␣ and -␤ (data not shown). C/EBP␣ was a more effective stimulator of the promoters than C/EBP␤. In HepG2 cells, C/EBP␣ and -␤ were equally potent stimulators of the StAR and DEI-4 promoter activities (Fig. 2).
Studies with the full-length StAR promoter (1.3 kilobases) indicated that coexpression of increasing amounts of C/EBP␣ expression plasmid (1-50 ng/well) with SF-1 (500 ng/well) had an additive or greater effect on StAR promoter activity (Fig. 3). Fig. 3 depicts the results for the 50-ng doses of C/EBP␣, C/EBP␤, or the empty pMEX carrier plasmid in the absence/ presence of the SF-1 or its empty pSV-SPORT-1 carrier plasmid. Additionally, each of the six treatment groups was further subdivided, and the cells were cultured in the presence and absence of 8-Br-cAMP (1 mM). Comparison of the mean values (n ϭ 3) for the control group (empty pSV-SPORT-1 ϩ empty pMEX) Ϯ 8-Br-cAMP indicated that StAR promoter activities with and without 8-Br-cAMP were not significantly different (p Ͼ 0.1; 28,000 Ϯ 11,000; 40,800 Ϯ 15,900; mean Ϯ S.E.). The remaining data were expressed as fold increases over the mean of these control values, which were given a value of 1. The transient expression of C/EBP␣ and -␤ in the COS-1 cells increased basal StAR promoter activity 5.7-and 3.7-fold, respectively, over the control (empty pSV-SPORT-1 ϩ empty pMEX minus 8-Br-cAMP). Promoter activity was not affected by 8-Br-cAMP treatment. In contrast, transient expression of SF-1 alone had little effect on basal StAR promoter activity, whereas it potentiated cAMP responsiveness of the StAR promoter (6.3-fold). Coexpression of C/EBP␣ and SF-1 had a greater than additive effect on both basal (10.9-fold) and cAMPdependent (26-fold) StAR promoter activity (Fig. 3). Transient expression of C/EBP␤ also augmented 8-Br-cAMP-stimulated StAR promoter activity in the presence of SF-1 but to a lesser extent than C/EBP␣ (Fig. 3).
StAR Promoter Function in Human Granulosa-Lutein Cell-To investigate the role of C/EBP in cells that normally express StAR, we transfected human granulosa-lutein cells with a series of StAR promoter deletion constructs (Fig. 4). Basal StAR promoter activity increased (p Ͻ 0.05) ϳ2-fold following the deletion of the distal 400 bases of the 1.3-kilobase StAR promoter, and this level of activity was maintained in the Ϫ235 StAR construct. The Ϫ150 StAR promoter construct displayed activity similar to the full-length StAR promoter, but all further deletions caused a pronounced decline (p Ͻ 0.05) in basal StAR promoter activity with the Ϫ95, Ϫ60, and Ϫ43 constructs exhibiting similar low promoter activity levels. Cells transfected with the shortest of the StAR promoter constructs (Ϫ43 to ϩ39) displayed ϳ2-fold greater luciferase activities than the pGL2-basic plasmid.
Based on the above results, the putative C/EBP response elements were mutated in the context of the Ϫ235 promoter construct. Fig. 5 illustrates the activities of the C/EBP mutant constructs in proliferating human granulosa-lutein cells. The Ϫ119/Ϫ110, Ϫ50/Ϫ41, and the double C/EBP mutant StAR promoters contained a 3-base pair substitution in each site that disrupts the core C/EBP element. The Ϫ235 wild-type StAR promoter construct had activity well above that of the empty pGL-2 basic reporter and exhibited a 2-fold increase in response to 24 h of treatment with 8-Br-cAMP. In contrast, there was a pronounced decline (ϳ4-fold, p Ͻ 0.05) in basal promoter activity in the Ϫ119/Ϫ110 and Ϫ50/Ϫ41 C/EBP mutation constructs. However, these single C/EBP mutant constructs retained their ability to respond to cAMP as evidenced by the 2-fold increase in promoter activity over basal activity in response to 1 mM 8-Br-cAMP treatment. Mutation of both C/EBP sites resulted in a further reduction in basal StAR promoter activity (Ͼ90% of the wild type) although not affecting cAMP responsiveness.
Electrophoretic Mobility Shift Assays-Recombinant C/EBP␣ and -␤ were both able to elicit a mobility shift of the oligonucleotide probe containing the Ϫ119 to Ϫ110 and Ϫ50 to Ϫ41 elements (Fig. 6). The gel-shift bands observed for the StAR oligonucleotides were similar to those observed with a consensus C/EBP site probe. Binding of both C/EBP␣ and -␤ proteins to the probes could be competitively inhibited by the addition of increasing doses of unlabeled oligonucleotide (highest dose, 20ϫ, is shown). Additionally, the C/EBP␣ and -␤ protein⅐DNA complexes could be supershifted with the cognate antibodies (Fig.  6). These binding interactions were specific, as mutations of the putative C/EBP␣ and -␤ binding site in the Ϫ55 to Ϫ31 StAR oligonucleotide prevented the interaction of this probe with the recombinant C/EBP␣ and -␤ proteins. Furthermore, this mutant oligonucleotide failed to inhibit the formation of a protein⅐DNA complex with either of the recombinant C/EBPs. Additionally, we tested oligonucleotides representing nucleotides Ϫ89 to Ϫ64 and Ϫ96 to Ϫ64 of the human StAR promoter, which contains no C/EBP consensus binding sites, and were unable to detect any protein⅐DNA complexes with the recombinant C/EBP␣ and -␤ (data not shown).
Having established that recombinant C/EBPs bind to sequences in the StAR promoter, we next determined if human granulosa-lutein cell nuclear extracts contained proteins capable of interacting with these elements. Fig. 7A shows that a protein⅐DNA complex is formed when granulosa-lutein cell nuclear extracts are incubated with the Ϫ126 to Ϫ100 oligonucleotide. Binding specificity was demonstrated by showing that 10 -50ϫ molar excess of the cold wild-type competitor ablated the complex between nuclear proteins and the labeled oligonucleotide, whereas a mutant cold oligonucleotide containing the same 3-base mutation used for promoter assays failed to disrupt the protein-DNA interaction. The complex was supershifted by C/EBP␤-specific antibodies, but antibodies to C/EBP␣ failed to elicit a supershift, indicating that the shift in mobility was because of the interaction of C/EBP␤ with the labeled oligonucleotide probe. Fig. 7B depicts the results of the electrophoretic mobility shift assay studies with the Ϫ55 to Ϫ31 oligonucleotide and human granulosa-lutein cell nuclear extracts. A protein⅐DNA complex of greater molecular weight than that observed for the distal site was observed. Binding specificity was demonstrated by inclusion of 10 -50ϫ molar excess of cold wild-type and mutant oligonucleotides, with ablation of the radiolabeled complex with the wild-type oligonucleotide and no effect of the mutant. In contrast to our observations with the distal C/EBP response element, antibodies to C/EBP␣ and -␤ did not supershift this DNA⅐protein complex, indicating that another protein(s) with binding specificity similar to that displayed by C/EBP␣ or -␤ was interacting with the oligonucleotide probe.
Western Blot Analysis of C/EBPs in Human Granulosa-Lutein Cell Nuclear Extracts-Granulosa-lutein cell nuclear extracts were tested for the presence of C/EBP␣ and -␤ by Western blot analysis (Fig. 8). A doublet of ϳ42 kDa of C/EBP␣ was identified, and levels of these proteins were not altered by treatment of the granulosa-lutein cells with 1 mM 8-Br-cAMP. Western blot analysis also indicated that C/EBP␤ (38 kDa) was present in granulosa-lutein cell nuclei, and levels of C/EBP␤ increased 3-fold (p Ͻ 0.05) in response to a 24-h treatment with 1 mM 8-Br-cAMP.

DISCUSSION
This is the first study to demonstrate that members of the family of C/EBP proteins regulate expression of the human StAR gene. We identified two putative C/EBP binding sites within the proximal StAR promoter by sequence homology that were subsequently shown to bind recombinant C/EBP␣ and -␤. Both C/EBP␣ and -␤ increased the activity of the StAR reporter construct in COS-1 and HepG2 cells. Furthermore, mutation of the individual C/EBP binding sites markedly reduced basal StAR promoter activity in human granulosa-lutein cells, and combined mutations caused a more pronounced loss in basal StAR promoter function. Although the overall level of promoter activity in response to 8-Br-cAMP was reduced when the C/EBP binding sites were mutated, there was no change in the fold increase in promoter activity in response to 8-Br-cAMP stimulation, demonstrating that C/EBPs, and in particular C/EBP␤, and the cognate response elements regulate the basal level of StAR gene expression and boost the overall response to 8-Br-cAMP.
Our studies with COS-1 cells, a monkey kidney cell line, and HepG2 cells, a liver cell line, demonstrate that C/EBP␣ and/or -␤ transactivate the StAR promoter in cells that do not normally express StAR or significant levels of SF-1 (6). This is the second instance in which basal StAR gene promoter activity could be stimulated in SF-1-deficient cell hosts. Our previous studies showed that sterol regulatory element-binding pro-tein-1a increases StAR promoter activity in COS-1 cells in the absence of exogenous SF-1 (14). Collectively, these studies suggest that SF-1 per se is not absolutely required for basal StAR gene expression. It is possible that COS-1 and HepG2 cells contain a homolog(s) of SF-1 that is able to activate SF-1 response elements in the presence of excess exogenous C/EBPs and/or SREBP-1a. However, cAMP responsiveness appears to require SF-1, because StAR promoter activity in COS-1 cells transfected with either C/EBP␣ or -␤ did not increase when cells were treated with 8-Br-cAMP unless exogenous SF-1 was present. Consonant with these observations, we found that mutation of either C/EBP site individually caused a pronounced loss in basal StAR promoter activity in granulosalutein cells, which express the endogenous StAR gene. However, cAMP-dependent StAR transactivation in the granulosalutein cells was not ablated. Mutation of both elements further diminished basal promoter function but again did not prevent 8-Br-cAMP from stimulating promoter activity. Conversely, our previous studies demonstrated that mutation of SF-1 response elements in the StAR promoter reduced basal promoter activity but produced a more profound loss of cAMP responsiveness (8).
The nuclear factors in cells that express StAR that bind to the two motifs we identified as C/EBP response elements appear to be different. The distal element binds C/EBP␤ and apparently not C/EBP␣, even though C/EBP␣ is expressed in granulosa-lutein cells. This may reflect binding selectivity of the distal site for C/EBP␤, differences in the concentrations of the factors in the nuclear extract, or posttranslational modifications that influence binding activity. There has been one previous report suggesting selective binding of C/EBPs to a response element (23) providing a precedent for the first possibility. In contrast, the proximal element binds neither C/EBP␣ nor -␤ but rather other protein(s) that evidently have a binding specificity similar to that of C/EBPs. Although recombinant C/EBP␣ and -␤ bound specifically to the StAR oligonucleotides representing the two putative response elements, the repertoire of transcription factors and co-activators expressed in granulosa-lutein cells evidently restricts the in-teraction of C/EBP␤ and C/EBP␣ with the proximal site in favor of other binding proteins that might include other members of the C/EBP family or possibly ATF/CREB, which can bind to elements that recognize C/EBPs (24). The basis for the apparent selectivity of the distal site for C/EBP␤ and the proximal site for other factors remains to be elucidated.
The expression of C/EBPs in ovarian cells is regulated providing a potential mechanism for modulating StAR gene transcription. In the immature rat ovary, C/EBP␣ mRNA and protein are present in granulosa cells of small, antral follicles following the administration of pregnant mare serum gonadotropin (25). The levels of C/EBP␣ increased with development FIG. 6. Recombinant human C/EBP␣ and -␤ bind to the ؊126 to ؊100 (A) and ؊55 to ؊31 (B) StAR oligonucleotides. A positive control (C/EBP) oligonucleotide bound both C/EBP␣ and -␤. C/EBP␣ and -␤ protein⅐DNA complex formation to the Ϫ126 to Ϫ100 oligo was suppressed by 20-fold molar excess unlabeled probe but was not suppressed by the Ϫ55 to Ϫ31 mutant probe (data not shown). Preincubation of the C/EBP␣ and -␤ proteins with the appropriate antibody caused a supershift of the protein⅐ Ϫ126 to Ϫ100 DNA complex. The Ϫ55 to Ϫ31 StAR oligonucleotide (B) exhibited similar C/EBP␣ and -␤⅐DNA complex formations that were also supershifted by the appropriate antibodies and suppressed by the addition of a 20-fold molar excess of unlabeled oligonucleotides. The addition of excess Ϫ55 to Ϫ31 probe with the mutated C/EBP site failed to alter the C/EBP␣ and -␤⅐wild-type DNA complex formation. Furthermore, no C/EBP␣ or -␤⅐DNA complex formations were visible when the Ϫ55 to Ϫ31 C/EBP mutant oligonucleotide was used as the labeled probe. and differentiation of the antral follicle into a preovulatory follicle, at which stage granulosa cells exhibit maximal levels of C/EBP␣. Administration of an ovulatory dose of human chorionic gonadotropin caused a rapid loss of C/EBP␣ mRNA in rat granulosa cells that rebounded as the cells luteinized. In contrast to these in vivo observations in the immature rat, we observed no change in nuclear C/EBP␣ levels following a 24-h treatment with 8-Br-cAMP, suggesting that regulation of C/EBP␣ may differ in rodent and human granulosa cells. In-terestingly, Piontkewitz et al. (17) demonstrated that injection of antisense oligonucleotides that recognize C/EBP␣ prevented ovulation. This study failed to detect a difference in steroidogenic output (progesterone and estradiol) in mice given the antisense versus sense oligonucleotide. However, antisense treatment did not result in the complete loss of C/EBP␣ expression, and these observations do not rule out the possibility that other members of the C/EBP transcription factor family (e.g. C/EBP␤) could compensate for the loss of C/EBP␣ and permit certain promoters (i.e. StAR) to function at normal activity, although still preventing follicular differentiation and ovulation.
In the rat ovary, C/EBP␤ levels were also shown to be low prior to ovulation, to increase dramatically in granulosa cells after the luteinizing hormone surge, and to then decline during the later stages of luteinization (26). We found that 8-Br-cAMP treatment of human granulosa-lutein cells increased nuclear C/EBP␤ levels 3-fold, demonstrating that in vitro, the cAMP signaling pathway leads to increased C/EBP␤ expression. Cyclic AMP and Ca 2ϩ /calmodulin kinase activation are also known to stimulate C/EBP␤ expression in other cell lines (27). Recent observations document that cAMP causes the synthesis of C/EBP␤, -␦, and -in Sertoli cells (28). C/EBP␤ transactivation potential can also be augmented by extracellular signaling pathways such as inflammatory cytokines and intracellular signaling pathways such as the mitogen-activated protein kinase, cAMP-dependent protein kinase A, Ca 2ϩ /calmodulin-dependent protein kinase, and protein kinase C by the differential phosphorylation of specific amino acids (see Refs. 20 and 29 and references therein). Similar to the C/EBP␣ antisense experiments, the targeted deletion of the mouse C/EBP␤ gene did not prevent StAR gene expression in the ovary. 2 It is not known whether C/EBP␣ could compensate for the absence of C/EBP␤ in this model or whether promoter activity driven by other transcription factors is sufficient to sustain some StAR gene expression. Targeted deletion of C/EBP␣ in the ovary may provide additional insight into the role that this factor plays in mouse ovarian function. However, because of the different expression patterns of C/EBP␣ and -␤ between rodents and human granulosa cells and the known differences in SF-1 regulation of the StAR gene between the mouse and human described below, these gene knockout studies may not provide useful information regarding the control of the human StAR gene. Testing the roles each of these transcription factors plays in vivo in the regulation of StAR expression is complicated by the lack of ovaries in SF-1 knockout mice (30) as well as the perinatal lethality of the C/EBP␣ gene knockout (31).
Our findings suggest that tropic hormones that increase 2 E. Sterneck and P. F. Johnson, unpublished observation.

FIG. 7.
Proteins in granulosa-lutein cell nuclear extracts bind to the StAR promoter. Electrophoretic mobility shift assays were performed as indicated under "Experimental Procedures" using 32 Plabeled probes (A, Ϫ126 to Ϫ100; B, Ϫ55 to Ϫ31) and human granulosalutein cell nuclear extracts (5 g/lane) obtained from cells cultured under control conditions. In A, the DNA⅐protein complexes formed with the Ϫ126 to Ϫ100 oligonucleotide containing the putative Ϫ119/Ϫ110 C/EBP response element and the free probe are indicated. The addition of 10 and 50ϫ molar excess of unlabeled wild-type and mutant oligonucleotides (see Fig. 1) demonstrated competition by the wild-type oligonucleotide but no competition by the mutant probe. Supershift studies with antibodies to C/EBP␣ and -␤ indicated that the Ϫ119/Ϫ110 element interacted with C/EBP␤ present in the human granulosalutein cell nuclear extracts. Electrophoretic mobility shift assay using nuclear extracts from cells treated with 8-Br-cAMP retarded a greater amount of the oligonucleotide (data not shown). In B, DNA⅐protein complexes for the Ϫ55 to Ϫ31 oligonucleotide containing the putative Ϫ50/Ϫ41 element and the free probe are indicated. Addition of 10 and 50ϫ molar excess of unlabeled wild-type and mutant oligonucleotides (see Fig. 1) indicated specific binding to the wild-type oligonucleotide. Antibodies to C/EBP␣ and -␤ failed to supershift the Ϫ55/Ϫ31 oligonucleotide⅐protein complex. n.s., nonspecific interaction. cAMP levels can activate StAR gene transcription through a concerted mechanism that includes SF-1-dependent transactivation as well as cAMP induction of C/EBP␤, which through interactions with the distal element identified in this study raises the level of promoter activity. We have previously shown that 8-Br-cAMP-induced increases in steady state StAR mRNA levels in proliferating granulosa cells are the result of increased StAR gene transcription and that the protein synthesis inhibitor, cycloheximide, blocks the 8-Br-cAMP-provoked increase in StAR mRNA (4). Granulosa cell SF-1 expression does not dramatically change following hormone or cAMP treatment, suggesting that another factor must be involved in the increase in StAR gene transactivation (32). C/EBP␤ may be the protein factor required for 8-Br-cAMP to induce StAR gene transcription.
After this paper was submitted for publication, Reinhart et al. (33) reported that C/EBP␤ regulates mouse StAR gene transcription. These authors proposed that C/EBP␤ and SF-1 interact cooperatively to enhance StAR gene expression and provided evidence for physical interactions between SF-1 and C/EBP␤. They found that the mouse promoter retained responsiveness to cAMP despite mutations in the two C/EBP response elements they identified, a finding that parallels our observations. However, in contrast to our findings, which indicate that C/EBP response elements augment the overall response to cAMP in the presence of SF-1, Reinhart et al. (33) concluded that C/EBP response elements are required for SF-1 to transactivate the StAR promoter. This apparent discrepancy may be explained by previously described differences in the mouse and human StAR promoters. SF-1 has been reported to be important for basal but not cAMP-stimulated mouse StAR gene transcription (3), whereas SF-1 is essential for cAMP activation of the human promoter (6,8).
The more distal site identified by Reinhart et al. (33), which is homologous to the distal site we described in the human promoter, bound C/EBP␤ present in mouse Leydig cell nuclear extracts, but an antibody against C/EBP␤ did not completely supershift the complex raising questions regarding the presence of other components in the complex. This differs from our findings on the human distal element. Our proximal element, which was conserved in the mouse and rat StAR promoters, is 40 bp 3Ј from the proximal site identified by Reinhart et al. (33) in the mouse promoter. Reinhart et al. (33) did not identify, using immunologic probes, the factors binding to the more proximal element, which they suggested was a "low affinity C/EBP site." We were unable to demonstrate binding of recombinant C/EBPs to two different human StAR oligonucleotides spanning the sequence of the murine proximal site. However, it is possible that the proximal site in the murine StAR promoter recognizes proteins other than C/EBP␤, as we have shown in the case of the human proximal element.
The conserved nature of the response elements within the region of the StAR promoter containing the C/EBP response element across multiple species (Fig. 1) as well as our evidence that these response elements are bona fide regulators of StAR gene promoter function suggests a regulatory role for these factors, specifically C/EBP␤, in StAR gene expression. Regula-tion of StAR gene transcription in human steroidogenic cells plays a critical role in their primary function, as cholesterol mobilization and delivery of cholesterol to the inner mitochondrial membrane is the rate-limiting step in steroidogenesis. In this study, we have examined a small region (bases Ϫ126 to Ϫ31) of the StAR promoter that appears to be an integration site for multiple transcription responses. Our observations provide a starting point for additional studies to examine the role that each of these families of transcription factors play in regulation of StAR gene expression and ultimately steroidogenesis.