StAR

Steroidogenic Factor-1 Influences Protein-Deoxyribonucleic Acid Interactions within the CyclicAdenosine 3*,5*-Monophosphate-Responsive Regions ofthe Murine Steroidogenic Acute RegulatoryProtein Gene*

CLAVIA R. WOOTON-KEE AND BARBARA J. CLARK

Department of Biochemistry and Molecular Biology, University of Louisville School of Medicine,Louisville, Kentucky 40292

ABSTRACTDe novo synthesis of the steroidogenic acute regulatory protein

(StAR) in response to trophic hormonal stimulation of steroidogeniccells is required for the delivery of cholesterol from the mitochondrialouter membrane to the mitochondrial inner membrane and the cy-tochrome P450 side-chain cleavage enzyme. StAR expression is tran-scriptionally regulated by cAMP-mediated mechanisms, and we haveidentified a 45-bp region within the mouse promoter that is importantfor cAMP responsiveness of the gene. This region, located between2105 and 260 of the start site of transcription, contains a SF-1-binding site, a highly conserved C/EBPb 2AP-1-nuclear receptor half-

site sequences (CAN region), and a GATA-4-binding site. The SF-1element and CAN region are required for full basal activity, whereasthe GATA-4 element may account for 20% of the cAMP response inMA-10 mouse Leydig tumor cells. A cAMP-dependent protein-DNAcomplex was observed with the CAN region and mutation of a non-consensus AP-1 site within this region greatly diminished promoterstrength. Complex protein-DNA interactions within the cAMP re-sponse region (2105/260) were shown to require the SF-1 element(295), suggesting that SF-1 is required for protein-DNA interactionat the CAN (279) region and maximal activity of the promoter. (En-docrinology 141: 1345–1355, 2000)

STEROID HORMONE biosynthesis in the adrenal andgonads is stimulated by trophic hormones via a cAMP-

dependent second messenger system. Within minutes of hor-monal stimulation, cholesterol is mobilized to the outer mi-tochondrial membrane and transferred to the mitochondrialinner membrane where it is converted to pregnenolone bythe cytochrome P450 side-chain cleavage enzyme (P450scc)(1–4). This acute response is the rate-limiting step in steroi-dogenesis and is dependent upon the de novo synthesis of thesteroidogenic acute regulatory protein (StAR) (5). Without afunctional StAR protein, cholesterol accumulates in the outermitochondrial membrane and steroid production ceases.Thus, StAR functions in the transfer of cholesterol to the innermitochondrial membrane (5). This function of StAR is ele-gantly demonstrated by characterization of StAR knockoutmice that lack steroid production (6). These data confirmedthat the StAR mutations identified in patients with congen-ital lipoid adrenal hyperplasia are the underlying cause forthe disorder (7, 8).

Hormonal treatment of steroidogenic cells controls StARgene expression. Our previous studies have shown that treat-ment of MA-10 mouse Leydig tumor cells with trophic hor-mones or the cAMP analog, (Bu)2cAMP, induces StAR mes-sage within 30 min, and maximal steady state levels are

reached within 4 h (9). This induction of StAR gene tran-scription does not require de novo protein synthesis, suggest-ing that posttranslational mechanisms are involved in theacute induction of StAR gene in response to increases inintracellular levels of cAMP (10).

Several factors that participate in StAR gene activationhave been reported. One factor is the orphan nuclear receptorsteroidogenic factor-1 (SF-1). SF-1 has been shown to have acritical role in the regulation of many of the steroid hydrox-ylase genes, and gene knockout studies in mice have shownthat SF-1 is critical for development of the gonad and adrenalglands (11–14). Regulation of SF-1 may occur through directphosphorylation by a protein kinase A (PKA)-dependentpathway. This proposal is supported by studies in whichSF-1 trans-activation of the steroid hydroxylase genes waslost in PKA-deficient cell lines (15, 16). Furthermore, poten-tial PKA phosphorylation sites are present in SF-1, and invitro studies have directly demonstrated a PKA-dependentphosphorylation of SF-1 (17–19). In addition, activation of themitogen-activated protein kinase pathway has been shownto phosphorylate and active SF-1, supporting a role for phos-phorylation of SF-1 for its function (20). Our initial analysisof two SF-1 sites in the mouse promoter, located at 245 and2135 from the transcription start site, indicated that theseelements contributed to basal promoter activity but were notessential for the cAMP-dependent induction (10). The in-volvement of a SF-1 site in StAR gene activation was morerecently confirmed with the rat promoter (21). Mutation ofthe SF-1 elements at 2135 and 295, either individually or incombination, decreased SF-1-dependent reporter gene ex-

Received August 9, 1999.Address all correspondence and requests for reprints to: Dr. Barbara

J. Clark, Department of Biochemistry and Molecular Biology, Universityof Louisville School of Medicine, Louisville, Kentucky 40292. E-mail:[email protected].

* This work was supported by NIH Grant DK-51656.

0013-7227/00/$03.00/0 Vol. 141, No. 4Endocrinology Printed in U.S.A.Copyright © 2000 by The Endocrine Society

1345

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 24 April 2015. at 10:09 For personal use only. No other uses without permission. . All rights reserved.

pression in a nonsteroidogenic cell line cotransfected for SF-1expression (21). In these studies the (Bu)2cAMP-mediatedincrease in promoter activity was also depressed, suggestingthat SF-1 may mediate part of the cAMP response. However,it is not known what effect the 295 SF-1 mutation wouldhave in a steroidogenic cell line, where factors other than SF-1may be required for basal and cAMP-dependent StAR tran-scription. On the other hand, the SF-1 element located at2105/295 in the human promoter was shown to be criticalfor basal and maximal cAMP responsiveness in human gran-ulosa-luteal cells and in H295R adrenocortical cells (22, 23).Thus, it appears that the role of SF-1 in StAR expression maybe promoter and/or tissue specific.

C/EBPb has also been shown to modulate StAR basalpromoter activity. C/EBPb is a basic leucine zipper tran-scription factor that is expressed in liver, intestine, lung, andadipose tissues (24). More recently, C/EBPb has been iden-tified as the only isoform of C/EBP present in unstimulatedprimary Leydig cell cultures and MA-10 Leydig cells (25).Two putative CCAAT sites located at 2113 and 287 in themurine StAR promoter were identified, and C/EBPb inMA-10 nuclear extracts was shown to bind to the 2113 el-ement (26). A weak protein-DNA interaction was observedwith the 287 region of the promoter, but the factors were notidentified. Mutation of the 287, but not the 2113, site abol-ished SF-1-dependent trans-activation of StAR-luciferase re-porter gene expression in transfected COS-1 cells (26). How-ever, transient transfection of MA-10 cells with a StAR-luciferase reporter gene construct containing mutations ateither the 287 or 2113 C/EBPb-binding site did not affectthe cAMP-dependent response, but greatly reduced the basalactivity of the promoter. A protein-protein interaction be-tween C/EBPb and SF-1 was further demonstrated; thus,these studies suggested a possible C/EBPb-SF-1 interactionthat is required for basal promoter activity.

Although SF-1 and C/EBPb participate in basal expressionof StAR, the mechanism for cAMP-dependent induction hasnot been elucidated. Previously, we narrowed the cAMP-responsive region of murine StAR to 2254 bp of the start siteof transcription (10). Sequence analysis of this region iden-tified potential binding sites for transcription factors thathave been shown to mediate cAMP-dependent responses inother genes. Two elements that are part of the focus in thepresent study are a C/EBPb/nonconsensus activating pro-tein-1/nuclear receptor half-site located between 287/279(CAN region) and a putative binding site for members of theGATA family of zinc finger transcription factors located be-tween 268/240. Activating protein-1 is composed of eitherFos-Jun or Jun-Jun dimers that bind to the consensus se-quence TGA(G/C)TCA and regulates genes either constitu-tively or by cAMP- or Ca21-mediated mechanisms (27–29).The GATA family of transcription factors regulates geneexpression, differentiation, and cell proliferation by bindingto the consensus DNA sequence (A/T)GATA(A/G) (30).More recently, three members of the GATA transcriptionfactor family were shown to be expressed in the developinggonads: GATA-1, GATA-4, and GATA-6 (31–33). GATA-1message has been isolated in MA-10 cells, and GATA-1 wasshown to up-regulate the basal promoter activity of the ratinhibin a-subunit gene (34). GATA-4 and GATA-6 have also

been identified in testicular tissue, and transient expressionof GATA-4 was shown to trans-activate an inhibin a-promoter/reporter construct in mouse Leydig and granulosatumor cell lines (35). Thus, a new role is emerging for GATAtranscription factors in Leydig cell function.

To clarify the role of SF-1 in StAR expression in MA-10mouse Leydig tumor cells, we extended our previous studiesand demonstrated that SF-1 binds to a third element in mousepromoter at 295. We show that this SF-1 site, denoted SF1–3,is required for full basal activity of the StAR promoter. Ad-ditionally, SF-1 functions in part to stabilize protein-DNAinteractions at the C/EBPb/nonconsensus activating pro-tein-1 (AP-1)/nuclear receptor half-site (CAN) region locatedat 279. We also demonstrate that GATA-4 binds to the StARpromoter at 268 bp and contributes to 20% of the cAMP-dependent induction. We conclude that the SF1–3, CAN, andGATA-4 DNA-binding sites serve critical roles in maintain-ing full basal promoter activity, which, in turn, is necessaryfor maximal cAMP induction.

Materials and MethodsMaterials

N6,29-O-(Bu)2cAMP [(Bu)2cAMP)] was purchased from Sigma (St.Louis, MO). Restriction enzymes, Klenow enzyme, T4 DNA ligase, Lu-ciferase Assay System, and pGL2-luciferase reporter vectors were pur-chased from Promega Corp. (Madison, WI). [g-32P]ATP was obtainedfrom NEN Life Science Products (Wilmington, DE). Custom oligonu-cleotides were purchased from Genosys (The Woodlands, TX) and Gen-emed Synthesis, Inc. (San Francisco, CA). Glutathione-S-transferase(GST)-SF-1 plasmid was donated by Dr. Keith Parker, University ofTexas Southwestern Medical Center (Dallas, TX). Antibodies were pur-chased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) and Up-state Biotechnology, Inc. (Lake Placid, NY). X-ray film was obtainedfrom Eastman Kodak Co. (Rochester, NY) and NEN Life Science Prod-ucts (Boston, MA).

Cell culture

The MA-10 mouse Leydig tumor cell line was a gift from Dr. M. Ascoli(Department of Pharmacology, University of Iowa of Medicine, IowaCity, IA). The cells were grown in Waymouth’s MB/752 medium con-taining 15% horse serum and 40 mg gentamicin sulfate/ml.

Transient transfections and reporter assays

MA-10 cells were plated at 300,000 cells/well in a 24-well plate 24 hbefore transfection. Two micrograms of StAR-luciferase reporter geneplasmid and 2 mg pCMV-b-galactosidase (CMV, cytomegalovirus) ex-pression vector plasmid were cotransfected into cells using 15 mg/mlLipofectamine (Life Technologies, Inc., Grand Island, NY) and Way-mouth’s medium without antibiotics and serum as described previously(10). Twenty-four hours posttransfection, the cells were treated with 1mm (Bu)2cAMP for 16–19 h. The Promega Corp. Luciferase Assay Sys-tem was used to prepare cell lysates and measure luciferase activity(Lumat LB 9507 luminometer, Wallac, Inc., Gaithersburg, MD). b-Galactosidase activity was assayed using b-d-galactopyranosidase(Roche Molecular Biochemicals, Indianapolis, IN) as the substrate andmeasuring the absorbance at 595 nm. Relative light unit (RLU) valueswere normalized to b-galactosidase activity for each sample. Each treat-ment was performed in triplicate, and the mean 6 sem were determined.The data were expressed as fold induction relative to the 2254/135 luccontrol, which was set at 100. The mean 6 sem for the fold induction inresponse to (Bu)2cAMP for all independent transfection experimentswas calculated, and a pooled Student’s t test was performed on the data.P , 0.05 was considered statistically significant.

1346 SF-1 INFLUENCE ON PROTEIN-DNA INTERACTIONS Endo • 2000Vol 141 • No 4


StAR-luciferase constructs

PCR was used to generate the 59-deletion constructs for the StARpromoter. Linearized 2254/135 StAR promoter-pGL2-luciferase (2254StAR promoter construct) was used as the template. To amplify theregion 2150 to 135 (2150/135 luc), MluI 2150 and GL2-primers wereused in the PCR reaction, and the product was digested with MluI andHindIII and cloned into the MluI and HindIII sites of pGL2-basic. Theregions 2105 to 135 (2105/135 luc) and 268 to 135 region (268/135luc) were amplified using SF1–3 top primer and GL2 primers, and268/244 top primer and GL2 primers, respectively, and the PCR prod-uct for each clone was Klenow treated, digested with HindIII, and clonedinto the SmaI-HindIII sites of pGL2-basic. The site-specific mutants gen-erated in the 2254/135 luc plasmid were constructed by generating twooverlapping PCR products that share an engineered restriction site toalter the core sequences of interest. To produce the 59-PCR product, theGL1 primer was used to amplify the top strand, and the bottom strandwas amplified with the site-specific primer. The 39-PCR product wasproduced with the site-specific mutant primer (top strand) and GL2(bottom strand). The PCR products were digested with the appropriaterestriction enzymes and cloned into the SmaI-KpnI restriction sites ofpGL2-basic. The sequences of oligonucleotides used in the reactions arelisted under Oligonucleotides used in EMSA and PCR. The SF1–3mut/GATA-4mut-luc, GATAmut/AP-1mut-luc, and SF1–3mut/AP-1mut-luc constructs were created in a similar manner; the mutant SF1–3 sitewas engineered into GATAmut-luc using the SF1–3mut-luc primer togenerate SF1–3mut/GATAmut-luc, the AP-1mut site was engineeredinto GATAmut-luc using the AP-1mut-luc primer to generateGATAmut/AP-1mut-luc, and the AP-1mut site was engineered intoSF1–3mut-luc using the AP-1mut-luc construct to generate SF1–3mut/AP-1mut-luc. One exception for SF1–3mut/AP-1mut-luc is the 59-PCRproduct was digested with MluI and BglII, the 39-PCR product wasdigested with BglII and XbaI, and the two products were cloned into theMluI-NheI sites of pGL2 basic. All sequences were confirmed by doublestranded sequencing using the Sequenase II kit (United States Biochem-ical Corp., Cleveland, OH).

Electrophoretic mobility shift assay (EMSA)

Double stranded oligonucleotides were generated by mixing equalmolar concentrations of top and bottom strand primers in a high saltannealing buffer [10 mm Tris (pH 7.5), 50 mm NaCl, and 1 mm EDTA]and heating for 2 min at 95 C followed by slow equilibration to roomtemperature. The oligonucleotides were end labeled with T4 DNA ki-nase and [g-32P]ATP (NEN Life Science Products). To generate a probefor the 2105/244 region of StAR, linearized 2254 StAR promoter con-struct was used as the DNA template in a PCR reaction with SF1–3 (top)and 268/244 (bottom). The 2105/244 fragment containing site-specific mutations or double mutations was generated as describedabove using the corresponding mutated reporter gene constructs as thetemplate and appropriate oligonucleotides. Each PCR fragment was gelpurified and radiolabeled as described above. Restriction digestion anal-ysis was used to confirm the PCR-generated probe contained the desiredmutations. MA-10 nuclear extracts (2.5–15.0 mg) or 0.25 mg GST-SF-1were incubated with 50 fmol radiolabeled oligonucleotide for 30 min onice. The binding buffer contained the following: 50 mm Tris-Cl (pH 8.0),100 mm KCl, 12.5 mm MgCl2, 1 mm EDTA, 1 mm dithiothreitol, 20%glycerol, and 2.5 mg poly(dI:dC). Cold competition assays were per-formed with a 20- to 100-fold molar excess of oligonucleotides. Antibodysupershift assays were performed with 2.5 or 15 mg MA-10 nuclearextract proteins or 0.25 mg GST-SF-1 plus 4 mg monoclonal GATA-1,polyclonal GATA-4, polyclonal C/EBPb, or polyclonal SF-1 antibodies.For the supershift assays, all components of the binding reaction, ex-cluding the radiolabeled probe, were preincubated on ice for 30 min.Radiolabeled probe was then added, and the incubation was continuedfor 30 min. Binding reactions were resolved on a 4% nondenaturingpolyacrylamide gel. The dried gels were exposed to x-ray film, and theradioactive bands were visualized by autoradiography.

The following oligonucleotides were used in the EMSA binding re-actions. Only the top strand is listed, and changed base pairs in themutated probes are underlined.

Sequences of oligonucleotides used for PCR andEMSA probes

The following sequences were used: SF1–3, 59-CATTCCATCCTT-GACCCTCTGCA-39; SF1–3mut, 59-CATTCCATCCTCGAGCCTCT-GCA-39; 268/244, 59-TTTTTTATCTCAAGTGATGATGCAC-39; 268/244 GATAmut, 59-TTTTTCCGGACAAGTGATGATGCAC-39; 268/244 SREBPmut (SREBP, sterol regulatory binding protein), 59-TTTTT-TATCTCTCGAGATGATGCAC-39; 287/264, 59-TGCACAATGACT-GATGACTTTTT-39; 287/264 AP-1mut, 59-TGCACAATAGATCTT-GACTTTTT-39; 287/264 1/2 mut, 59-TGCACAATGACTGA-GATCTTTTT-39; GL1*, 59-TGTATCTTATGGTACTGTAACTG-39; GL2*,59-CTTTATGTTTTTGGCGTCTTCCA-39; and 2150 Mlu, 59-ACA-CACGCGTAGTCTGCTCCCTCCCACCTTGGCCA-39 (* indicates theprimers used for pGL2-basic vector).

Nuclear extract preparation

Confluent MA-10 mouse Leydig tumor cell cultures were treated withfresh Waymouth’s medium in the absence (control) or presence of(Bu)2cAMP for 4 h, and nuclear extracts were prepared following pre-viously described protocols (36). The extracts were flash-frozen in liquidnitrogen and stored at 280 C until use.

Results59-Deletion analysis localizes the cAMP-responsive regionbetween 2105 and 268 of the mStAR promoter

To more closely map the cAMP-responsive elementswithin the 2254 bp region of the murine StAR gene, a seriesof StAR promoter deletion constructs was transiently trans-fected into MA-10 cells and tested for cAMP responsiveness(Fig. 1). (Bu)2cAMP treatment of MA-10 cells resulted in an8.7-fold increase in luciferase reporter gene activity for the2254 StAR promoter construct. Deletion to 2150 caused bothbasal and (Bu)2cAMP-stimulated promoter activity to in-crease 4- and 6-fold, respectively, which represents a 70%increase in the cAMP-dependent fold induction comparedwith the 2254 StAR construct. Further deletion to 2105caused a 46% decrease in basal activity compared with 2254StAR, but the fold induction in response to the cAMP stim-ulus was not diminished. However, a 5-fold decrease in thecAMP-dependent induction was observed with the 268StAR promoter construct. These results suggest that multipleelements may be required for the cAMP response of themouse StAR promoter: a possible negative regulatory regionlocated between the 2150/2105 and a positive regulatoryregion located between the 2105/268 region of the pro-moter. The 2254/265 region of the promoter was clonedupstream of a heterologous promoter (2254/266StAR-pGL3luc), but the single copy of this region was not sufficientto confer cAMP-dependent responsiveness (data not shown).Therefore, elements 59 and 39 of 268 appear to required forthe induction of StAR gene by cAMP.

SF-1 and GATA-4 bind to elements within 2105/244region of the mouse StAR promoter

As the 59-deletion analysis indicated that the cAMP-responsive region was within 2105 bp of the transcriptionalstart site, we were interested in testing for possible cAMP-dependent protein-DNA interactions within this region. Wefirst compared the sequence of murine StAR promoter (2105to 135) to the transcription factor database and identifiedseveral potential transcription factor-binding sites that in-

SF-1 INFLUENCE ON PROTEIN-DNA INTERACTIONS 1347


cluded SF-1, C/EBP, AP-1, and GATA-1. Shown in Fig. 2Ais the sequence of the 2150 region of murine StAR, with thelocation of these elements indicated. An alignment with thehuman and rat sequences is also shown to indicate the highdegree of sequence conservation within this region of theStAR promoter. We previously characterized the SF-1–1(2135), SF-1–2 (245), and the C/EBPb (2113) elements (10,26), whereas the SF-1 site located at 295 bp has been shownto bind SF-1 in the human and rat StAR promoters (21, 37).The region between 287/264 is not as well conserved andunique to the mouse sequence are potential C/EBPb, non-consensus AP-1, and nuclear-receptor half-site elements thatwe refer to as the CAN region. Another highly conservedregion, located between 268/244, is a putative binding sitefor GATA-1 and SREBP. This region is adjacent to the pro-posed cAMP-responsive region (2105/268) identified by59-deletion analysis and was deleted in the nonresponsive39-deletion construct (2254/266StAR-pGL3luc), indicatingthat this region also may contribute to the cAMP response.

Based on the location of these elements, overlapping DNAprobes were generated to encompass the SF-1 (2105/283SF1–3), the C/EBPb-AP-1-nuclear receptor half-site se-quences (287/264 CAN), and the GATA/SREBP element(268/244 GATA) and used in EMSAs. Mutations that wereintroduced into the core binding sequences of the oligonu-cleotides are shown in Fig. 2B. The SF1–3 DNA probe (2105/283) formed a specific protein-DNA complex that was ob-served using nuclear proteins isolated from either untreated(lane 1) or (Bu)2cAMP-treated (lane 5) MA-10 cells (Fig. 3).The specificity of complex formation was shown by compe-tition with unlabeled SF1–3 probe (lanes 2 and 6), and thecomplex was not formed when an oligomer that containedtwo point mutations in the CATCCTTG core sequence (SF1–

FIG. 1. 59-Deletion analysis of the mouse StAR promoter. MA-10 cellswere cotransfected with the indicated StAR-promoter-luciferase re-porter gene constructs (denoted by the position of the base pairs5-prime to the transcriptional start site) and the pCMV-b-galactosi-dase expression vector. The transfected cells were treated in triplicatewith fresh Waymouth’s medium in either the absence (basal) or pres-ence of 1 mM (Bu)2cAMP for 16–19 h, and the luciferase activity wasdetermined and normalized to b-galactosidase expression (RLU/b-gal). For each experiment, the data were expressed as a percentageof the 2254 StAR-promoter construct (2254/135 StAR promoter-pGL2luciferase), which was set at 100%, and are graphed as themean 6 SEM values for the relative luciferase activity from three tofive experiments. The fold induction values represent the ratio of(Bu)2cAMP/basal activity for each construct. Shown are the mean 6SEM values for three to five experiments. The inset graph shows theRLU/b-gal activities of the 2254, 2105, and 268 StAR-promoterconstructs. pGL2-basic represents vector control, and NT representsnontransfected MA-10 cells. *, Statistically significant (P , 0.05)increase (2150/135) or decrease (268/135 luc) relative to the foldinduction of the 2254 StAR promoter-luciferase construct in responseto (Bu)2cAMP.

FIG. 2. A, Comparison of mouse (M),human (H), and rat (R) StAR promotersequences within 2150 bp before thetranscriptional start site. The locationof the DNA elements for SF-1, C/EBPb-,GATA-4-, and putative SREBP-bindingsites are indicated above the core se-quences by the respective name of thehighlighted factors. The putative bind-ing sites for the CAN region (287/264)include C/EBP, AP-1, and an imperfectnuclear receptor half-site (1/2). B,Shown are the region of the promoterand the name given to the introducedmutation in the 2254/135 StAR-promoter constructs and the EMSAprobes. The point mutations are indi-cated in bold.



3mut) was used in the binding reaction (lanes 4 and 8). Aprotein-DNA complex of similar mobility was observedwhen the probe was incubated with purified GST-SF-1 (lane9), and addition of an antibody specific for SF-1 greatlydiminished or abolished DNA-protein interaction (lanes 3and 7). Thus, these data verify that SF-1 also binds to thisregion of the mouse promoter. As the mobility and intensityof the SF-1 complex are the same with nuclear proteins fromeither control or (Bu)2cAMP-stimulated cells, these data in-dicate that SF-1 binding is independent of the cAMP stimulusin MA-10 cells. This element is referred to as SF1–3 to indicatethe third SF-1-binding site within 150 bp of the mouse pro-moter (Fig. 2A).

Two protein-DNA complexes were observed with theCAN (287/264) DNA probe (Fig. 4, A and B). Complex Ais a doublet that is present in binding reactions with nuclearproteins from both control and (Bu)2cAMP-treated MA-10cells, whereas complex B is a slower migrating band that ispresent only with nuclear proteins from stimulated cells (Fig.4A). As this region contains a CAAT box, and C/EBPb hasbeen proposed or shown to interact with this region (26, 38),we used a C/EBPb antibody to test for the presence ofC/EBPb in complex A or B. As shown in Fig. 4B, addition ofthe antibody did not affect protein-DNA interactions withthe CAN probe and nuclear proteins from stimulated cells

(lane 2). We verified that a protein-DNA complex is formedwith a consensus C/EBP oligonucleotide (lane 3) and thatthis complex is supershifted by antibody (lane 4), confirmingthat C/EBPb is present in the nuclear extract (25, 26). Coldcompetition with the consensus C/EBP oligonucleotide alsofailed to indicate any involvement of C/EBPb with this el-ement (lane 5). The specificity of complex formation wasdemonstrated by cold competition with the CAN probe (lane8) and a mutation in the putative nuclear receptor half-site(1/2mut) was also effective as a competitor probe (lane 6).Mutation in the putative AP-1 site (AP-1mut), on the otherhand, did not compete with the CAN probe for proteinbinding (lane 7). We next tested for Fos/Jun binding to this

FIG. 3. EMSA analysis reveals SF-1 binding to the 2105/283 regionof mouse StAR. EMSA was used to examine binding of MA-10 nuclearextract to the radiolabeled SF1–3 probe, which represents the 2105/283 region of the StAR promoter. Binding reactions containing 50fmol radiolabeled probe and 2.5 mg nuclear extracts prepared fromunstimulated (lanes 1–4) or (Bu)2cAMP-treated (lanes 5–8) MA-10cells or 0.25 mg purified GST-SF-1 protein (lane 9) were preincubatedon ice for 30 min. For cold competition or antibody interference anal-ysis, a 100-fold molar excess of cold probe (indicated as 1 in lanes 2and 6) or 4 mg SF-1 polyclonal antibody (indicated as Ab in lanes 3 and7) were preincubated with nuclear extract for 30 min; then the ra-diolabeled probe was added to all reactions, and the incubation wascontinued for 30 min on ice. The DNA-protein complexes were re-solved on a nondenaturing 4% polyacrylamide gel, the gel was dried,and the complexes were visualized by autoradiography. Mutationsintroduced into SF1–3 (SF1–3 mut) are listed in Fig. 2B. Lane 10shows incubation of SF-1 polyclonal antibody and radiolabeled probewithout nuclear extract. Free probe is not shown on the gel.

FIG. 4. A cAMP-dependent factor(s) binds to the 287/264 CAN re-gion of mouse StAR. EMSA was used to examine binding of MA-10nuclear extract to radiolabeled CAN probe, which represents the287/264 region of the StAR promoter. A, Concentration-dependentprotein-DNA complex formation. Binding reactions were preincu-bated for 30 min on ice with nuclear extracts prepared from unstimu-lated (control) or (Bu)2cAMP-treated [(Bu)2] MA-10 cells. Shown is atitration of 5, 10, and 15 mg control or (Bu)2cAMP nuclear extracts,respectively. B, Effects of cold competition and C/EBPb antibody onprotein-DNA interactions. Either a 50-fold molar excess of the indi-cated cold probe or C/EBPb polyclonal antibody (4 mg) antibody wasincluded in a preincubation with nuclear extracts for 30 min; thenradiolabeled probe was added, and the incubation was continued for30 min on ice. The protein-DNA complexes were resolved on a non-denaturing 4% polyacrylamide gel, the gel was dried, and the com-plexes were visualized by autoradiography. The sequences of theoligomer probes for the CAN region containing mutations in the po-tential AP-1 (AP-1 mut) and imperfect nuclear receptor half-sites (1/2mut) are listed in Fig. 2B. con. C/EBP, Commercially purchased con-sensus C/EBP oligomer. The arrows indicate the positions of com-plexes A and B and free probe. SS, The C/EBPb antibody supershift.



region, but addition of antibodies that recognize all familymembers did not affect complex formation (data not shown).

A prominent protein-DNA complex was apparent withthe GATA oligonucleotide (268/244 GATA) using nuclearproteins from either untreated or (Bu)2cAMP-stimulatedMA-10 cells (Fig. 5). Complex formation was abolished bycold competition with the GATA oligonucleotide (lane 2) orthe SREBPmut oligonucleotide that contains a mutation inthe putative SREBP site (lane 4). Conversely, using the SREB-Pmut as a probe did not affect complex formation (lane 3),indicating that the SREBP element is not involved in protein-DNA interactions at this region. On the other hand, no com-plex was formed with a probe that had a mutation within theGATA DNA-binding site (lane 5), and the mutant GATAprobe (GATA mut) did not compete for protein binding (lane6). A complex of similar mobility was observed using a probe

with the consensus sequence for GATA binding (lanes 7 and8) and cold competition experiments using this consensusGATA probe resulted in the lack of complex formation (lane10). Although this sequence was identified as a potentialGATA-1-binding site, GATA-1 antibodies did not affect com-plex formation (compare lanes 1 and 9). However, binding tothe consensus GATA probe appeared to be diminished byGATA-1 antibodies (compare lanes 7 and 8). Therefore, wetested for other members of the GATA family, and GATA-4antibodies resulted in a supershift of the complex (lanes11–13). The supershifted complex was apparent using nu-clear extracts from either control (lane 11) or (Bu)2cAMP-treated (lanes 12 and 13) MA-10 cells with the StAR GATAprobe (lanes 11 and 12) or the consensus GATA probe (lane13). A less prominent band appears with the 268/244 GATAprobe; however, the significance of this interaction is notknown. At present, these results indicate that GATA-4 inMA-10 cells binds to the mouse StAR promoter.

The 287/268 region of the StAR promoter is required formaximal basal promoter activity of StAR

To test the functional significance of the specific protein-DNA interactions characterized above on StAR promoteractivity, the mutations that abolished protein binding wereindividually introduced into the 2254 StAR promoter con-struct, and luciferase activity was measured in transientlytransfected MA-10 cells (Fig. 6A). Mutation of the SF1–3 siteresulted in a 43% reduction in basal luciferase activity com-pared with the 2254 StAR promoter construct (WT), but thefold induction in response to (Bu)2cAMP stimulation was notsignificantly different. Mutation of the GATA-4-binding sitealso reduced basal promoter strength (;36%); however, thecAMP response was also decreased from 8.7-fold (WT) to6.9-fold (GATA). Lastly, mutation of the putative AP-1 site(AP-1mut) or the nuclear receptor half-site (1/2mut) resultedin decreased basal activity by 63% or 57%, respectively, but,again, fold activation in response to (Bu)2cAMP was un-changed compared with that with the 2254 StAR promoterconstruct (WT). Although the fold induction by (Bu)2cAMP-treatment was not diminished with the SF1–3 and CANregion mutations, overall promoter strength was reduced.This decrease in response appeared to closely parallel theseverity of the mutation on basal promoter activity. Thesedata indicate that the SF1–3, CAN, and GATA sites all con-tribute to basal promoter strength. However, neither theSF1–3 nor the CAN region is individually required for thecAMP-dependent response of the StAR promoter. On theother hand, GATA-4 may account for 20–30% of the cAMPresponse of the murine StAR promoter.

Mutation of multiple elements affects basal activity of theStAR promoter

As mutations in the SF-1- and GATA-4-binding site indi-vidually affected StAR basal and cAMP-dependent promoteractivity, respectively, as well as the close proximity of the twobinding sites, we introduced mutations that abolished bothSF-1 and GATA-4 binding into 2254 StAR promoter con-struct and tested the effect on promoter activity in transientlytransfected MA-10 cells (Fig. 6B). The overall basal activity

FIG. 5. GATA-4 in MA-10 nuclear extracts binds the highly con-served region (268/244) of the mouse StAR promoter. EMSA wasused to examine binding of MA-10 nuclear extract to radiolabeledGATA probe, which represents the 268/244 region of the StAR pro-moter. Binding reactions were incubated for 30 min on ice with 5 mgnuclear extracts prepared from unstimulated (lanes 1–11) or(Bu)2cAMP-treated (lanes 12 and 13) MA-10 cells. For cold competi-tion or antibody interference analysis, a 20-fold molar excess of theindicated cold probe, denoted by cc (lanes 2, 4, 6, and 10), or 4 mgGATA-1 monoclonal antibody (lanes 8 and 9) or GATA-4 polyclonalantibody (lanes 11–13) were included in a preincubation with nuclearextracts; then the radiolabeled probe was added, and the incubationwas continued for 30 min on ice. Lanes 14 and 15 represent probealone or probe and antibody minus nuclear extract, respectively. Theprotein-DNA complexes were resolved on a nondenaturing 4% poly-acrylamide gel, the gel was dried, and the complexes were visualizedby autoradiography. The radiolabeled probes are denoted as GATA forthe 268/244 region, GATA mut for the 268/244 region containingmutations in the GATA core sequence, and SREBP mut for the 268/244 region containing mutations in the putative SREBP element. Thesequences for these probes are all shown in Fig. 2B. The con. GATAprobe represents the 20-mer consensus oligonucleotide for GATA thatwas used as a positive control for GATA-binding proteins. The arrowsindicate the supershift with GATA-4 Ab (SS) and the free probe; theopen arrowhead represents the specific complex.



of StAR promoter was dramatically decreased (88%) com-pared with that of the 2254 StAR promoter construct (WT),indicating that mutation of both SF1–3 and GATA had anadditive effect on promoter strength. Despite minimal basalactivity, the promoter caused a 4.8-fold increase in the cAMP-dependent response. However, this represents only 45% ofthe cAMP response for the 2254 StAR promoter construct.Thus, the double mutation has a greater effect than the singleGATA mutation on the cAMP response. Therefore, we ex-amined the possibility that either the SF1–3 or GATA-4 siteacts functionally with the AP-1 site to promote the cAMPresponse (Fig. 6B). Double mutations in the SF-1 and AP-1elements (SF1–3/AP-1) or the GATA-4 and AP-1 (GATA/AP-1) elements did not result in any greater effect on StARpromoter activity compared with an additive effect of thesingle mutations alone. The basal activities of SF1–3mut/

AP-1mut and GATAmut/AP-1mut were 16% (AP-1 plusSF-1) and 23% (SF-1 plus GATA) that of the 2254 StARpromoter construct, respectively, whereas the (Bu)2cAMP-dependent increases were 90% (SF1–3 effect) and 64%(GATA-4 effect) that of the 2254 StAR. Thus, it appears thatmutation of the AP-1 element has the single greatest effect onStAR promoter activity, due mainly to decreased basal ex-pression, whereas the SF1–3 and GATA-4 double mutationshad the most severe effect on promoter function. One pos-sibility is that SF1–3 and GATA-4 function to stabilize acommon factor(s) in the CAN (287/264) region. To test thisproposal, we examined the effects of abolishing one DNA-binding site on protein-DNA interactions at other siteswithin the 2105/244 bp of StAR promoter.

Protein-DNA interactions within the cAMP-responsiveregions of the StAR promoter

Four major protein-DNA complexes (I–IV) were detectedby EMSA analysis using a DNA probe that spans the 2105to 244 region (2105/244) of the StAR promoter (Fig. 7A).Complex formation appeared to be independent of cAMPtreatment in that similar complexes were formed with nu-clear extracts from both control and (Bu)2cAMP-treatedMA-10 cells. Antibodies to SF-1, GATA-4, and C/EBPb wereused to help identify the presence of these proteins in the fourcomplexes (Fig. 7B). The SF-1 antibody abolished complex Iand diminished complex III (lane 2), whereas GATA-4 an-tibodies caused the appearance of a supershifted complexconcomitant with diminished complex II (lane 3). C/EBPbantibodies had no effect on protein-DNA interactions withinthis region of the StAR promoter (lane 4). Cold competitionexperiments with the oligonucleotides corresponding to theSF1–3 (2105/244), CAN (287/264), and GATA (268/244)regions of the promoter were performed to verify the con-tribution of each site to complex formation. First, cold com-petition with unlabeled 2105/244 probe greatly diminishedcomplex formation (lane 5). The SF1–3 probe eliminated com-plex I and diminished complexes III and IV without affectingcomplex II formation (lane 6), whereas the GATA probecompeted for complex II formation without affecting com-plex I, III, or IV (lane 9). The CAN probe eliminated complexIV (lane 7), whereas the consensus C/EBPb oligonucleotide(lane 8) had no effect on the protein-DNA interactions withthe StAR promoter.

To further assess complex composition, a series of com-plementary experiments was performed in which the samesequence mutations used in the functional studies were in-troduced into the 2105/244 probe, and the effects on pro-tein-DNA interactions were determined (Fig. 7C). ComplexI was abolished, and complexes III and IV were diminishedby mutation of the SF1–3 site (SF1–3mut, lane 2), which isconsistent with the results of the SF-1 antibody and coldcompetition experiments. Complex II was not affected by theSF1–3 mutation, and addition of the GATA-4 antibodycaused a supershift of the remaining complex II (lane 3).Conversely, mutation of the GATA-4 site eliminated com-plex II formation (GATAmut, lane 4), and the remainingcomplexes (I, III, and IV) were greatly diminished by SF-1antibodies (lane 5). The CAN probe was an effective com-

FIG. 6. Mutational analysis reveals regulatory regions of mouseStAR. MA-10 cells were cotransfected with the indicated StAR pro-moter-luciferase reporter gene constructs and pCMV-b-galactosidaseas detailed in Materials and Methods. The transfected cells weretreated in triplicate with fresh Waymouth’s medium in either theabsence (basal) or presence of 1 mM (Bu)2cAMP for 16–19 h, andluciferase activity was determined and normalized to b-galactosidaseexpression (RLU/b-gal). For each experiment, the data were ex-pressed as a percentage of the 2254 StAR promoter basal activity,which was set at 100%. Graphed are the mean 6 SEM values fromthree to six experiments for the relative luciferase activity of StARpromoter containing mutations in either a single element (A) or twoelements (B). The fold induction values represent the ratio of(Bu)2cAMP/basal activity for each construct. Shown are the mean 6SEM values for three to five experiments. *, Statistically significant(P , 0.05) difference relative to the fold induction of the 2254 StARpromoter construct in response to (Bu)2cAMP. WT, Wild-type 2254/135 StAR promoter-luciferase construct; pGL2-basic, empty vector;NT, nontransfected.



petitor for complexes III and IV that remained with theGATAmut probe (lane 6) probe, but did not further reducecomplex formation with the AP-1mut probe (lane 7), sug-gesting that factors binding to the CAN region contribute toformation of these complexes. Indeed, mutation of the AP-

1-like site eliminated complex IV and surprisingly also re-duced all other protein-DNA interactions (lane 10). The re-maining complexes I and III were eliminated by the additionof SF-1 antibodies (lane 9), whereas GATA-4 antibodies re-sulted in a supershift of complex II (lanes 8).

FIG. 7. The SF1–3- and GATA-4-binding sites are essential for the formation of multiple complexes within the 2105/244 region of StAR. EMSAwas used to examine the binding of MA-10 nuclear extract to the radiolabeled 2105/244 region of the StAR promoter. A, Concentration-dependent protein-DNA complex formation. Fifty femtomoles of radiolabeled probe (2105/244) were incubated with 5, 10, or 15 mg nuclearextracts from untreated (control; lanes 1–3) or (Bu)2cAMP-treated [(Bu)2; lanes 4–6] MA-10 cells. The arrows indicate the positions of fourprotein-DNA complexes, denoted I–IV. Free probe is not shown on this gel. B, Cold competition or antibody interference analysis. SF-1, GATA-4,or C/EBPb antibody (4 mg) or the indicated cold competitor probes were incubated with 10 mg nuclear extracts [(Bu)2cAMP-treated]; then theradiolabeled 2105/244 probe was added, and the incubation was continued for 30 min on ice. The protein-DNA complexes were resolved ona nondenaturing 4% polyacrylamide gel, the gel was dried, and the complexes were visualized by autoradiography. The cold competitors were2105/244, SF1–3 (2105/283), CAN (287/264 region), GATA (268/244 region), and consensus C/EBP. Arrows indicate complexes I–IV, andthe open arrowhead indicates the GATA-4 supershifted complex (SS). C and D, Effects of mutation of an individual element (C) or two elements(D) on protein-DNA interactions. The sequences for the mutations that were introduced into the 2105/44 probe are given in Table 2B. Theradiolabeled probes used in the binding reaction are indicated: 2105/244, SF1–3 mut, GATA mut, and AP-1 mut, or their combination (D).Experiments in the presence of antibodies or cold competitor CAN probe were performed as described in B.



As StAR basal promoter activity was greatly reduced bymutations within two of the three SF1–3, CAN, or GATAelements (Fig. 6B), we also tested the effects of these muta-tions on protein-DNA interactions (Fig. 7D). A double mu-tation in SF-1 and GATA (SF1–3/GATA mut) resulted in theloss of all complex formation (Fig. 7D, lane 2), whereas com-plex II was still present with the probe containing a doublemutation in SF1–3 and AP-1 (SF1–3/AP-1 mut, lane 3). Dou-ble mutations of GATA and AP-1 caused a marked decreasein protein-DNA interactions for complexes III, and IV andeliminated complex II (lane 4). These data are consistent withan additive effect of the single mutations on protein-DNAinteractions, with the exception that complex I did not appearaffected by the GATAmut/AP-1mut probe. This was not anexpected result based on the decrease in complex I with theAP-1 mut probe (Fig. 7C, compare lanes 10 and 11). Theseresults indicate that complexes III and IV are dependentupon SF-1 and possibly, to a lesser extent, GATA-4.

In sum, the individual mutations in the SF1–3 or AP-1-likeregions and the SF-1 antibodies do not affect GATA-4 bind-ing; therefore, we conclude that GATA-4 binding within the2105/244 region is independent of other factors. Further,the GATA-4 mutation does not prevent SF-1-dependent com-plex formation (I, III, and IV), indicating that SF-1 binding isnot dependent on GATA-4 binding. CAN factors appear tocontribute to complex IV formation, as mutation of the AP-1-like element or cold competition with the CAN regioneliminated this complex. However, protein binding withinthe CAN region in the context of the longer probe appearsto be dependent on SF-1 binding to SF1–3 and may be in-fluenced by GATA-4 binding. Conversely, an intact AP-1element appears to be required for efficient SF-1 binding,indicating that protein binding to SF1–3 and CAN regionsmay be interdependent. Together, these data clearly dem-onstrate that complex I is due to SF-1 binding, and complexII is due to GATA-4 binding. The data are consistent withcomplexes III and IV being dependent upon SF-1, with com-plex III most likely containing SF-1, whereas CAN factorbinding contributes to complex IV.

Discussion

These studies have identified two regulatory regions of theStAR promoter: a negative region located between 2254/2150, and a cAMP-responsive region located between2105/260. Putative negative regulatory regions were pre-viously identified for the rat and porcine promoters as well(21, 39). Our promoter deletion analysis has eliminated SF-1(SF-1–1) and C/EBPb binding at 2135 and 2113 as essentialfor the cAMP response of the StAR gene and has pointed toSF1–3 (295), CAN-binding protein(s) (287/268), andGATA-4 (268) as potential factors.

Identification of the 2105 to 295 region as a SF-1-bindingsite was an expected result, as this element was previouslyshown to bind SF-1 in the human and rat StAR promoters (21,37). However, these are the first studies to show a distinctionbetween regulations of the human and mouse StAR pro-moter by cAMP in steroidogenic cell lines. Mutation of theSF-1 site (295) in the human promoter demonstrated 90%and 80% decreases in basal and forskolin-stimulated pro-

moter activity in human granulosa-lutein cells (37). Our re-cent studies in the human H295R adrenocortical cells alsohave shown that the SF-1 element at 295 is required forcAMP and angiotensin II induction of human StAR promot-er-luciferase reporter gene expression (23). Transient trans-fection of COS-1 cells with the rat StAR promoter confirmeda SF-1-dependent activation of the StAR promoter that re-quired the SF-1 site at 295 and that mutation of this siteresulted in a decrease in the cAMP response that was due tolowered basal promoter activity (21). Our results demon-strate that the mouse and rat promoters are functionally moresimilar to each other than to the human StAR promoter. Thisapparent promoter-specific difference may be reflective of adivergence within the StAR promoter sequences between thehuman and the mouse. The human sequence has an addi-tional five bases downstream of the SF-1 site and does notcontain the CAAT box, the the AP-1-like site, or the nuclearreceptor half-site that have been shown to be important forthe murine promoter in MA-10 cells (26, 38). Thus, it isapparent that SF-1 is important in StAR gene regulation butis not sufficient for the cAMP-dependent response in MA-10cells. Possibly, the sequence divergence reflects greater in-teractions of SF-1 with other factors binding in this region ofthe mouse StAR promoter.

We have confirmed the functional importance of the AP-1-like element within the 287/268 (CAN) region of mouseStAR promoter for basal transcriptional activity (26, 38). Inaddition, this study provides the first evidence that(Bu)2cAMP treatment of MA-10 cells results in the formationof a unique protein-DNA complex at this region of the StARpromoter (complex B). Identification of the CAN factor(s)should provide an important link to elucidating the mech-anism of action for cAMP-dependent regulation of StAR.Presently, our data suggest that multiple elements may con-tribute to activation of the StAR promoter. Double mutationsin either the SF1–3 and AP-1 or GATA and AP-1 elementshave a greater effect on basal promoter activity that repre-sents the combined effects of the single mutations. TheGATAmut/AP-1 mut also resulted in decreased cAMP re-sponse that is attributed to the GATA element. On the otherhand, the effect of the SF1–3 and GATA double mutation onbasal promoter activity was greater than the combined ef-fects of the single mutations. As mutations that disruptedthe AP-1-like site (AP-1mut, SF1–3mut/AP-1mut, andGATAmut/AP-1mut) have similar effects on basal activity asthe SF1–3mut/GATAmut, these data suggest that all threeelements work together to promote full basal activity of theStAR promoter that, in turn, is required for a maximal cAMPresponse. What is not clear at the time is why the cAMP-dependent response, although reduced, remains intact de-spite apparent elimination of protein-DNA binding interac-tions within the 2105/244 region of StAR promoter with theSF1–3mut/GATAmut. As we did not observe a cAMP-dependent complex with the 2105/244 probe, the cAMP-dependent CAN-binding factor(s) may not have been de-tected in this analysis. Indeed, the cAMP-dependent complexwas observed only with increased concentrations of nuclearextract, suggesting that this factor(s) may be expressed inlesser abundance. Thus, it is possible that the AP-1 region isstill functional in the SF1–3mut/GATAmut double mutation.



Previously, we demonstrated that a protein(s) in MA-10nuclear extracts binds to the CAAT box located in the 293/271 region, but with lower affinity compared with theC/EBPb site located at 2113 (26). Antibody supershift EMSAexperiments showed that C/EBPb binds to the 2113 site, butbinding to the proximal site was not tested. We now havedirectly tested C/EBPb binding to the CAN region and verifythat C/EBPb does not bind to this region of the StAR pro-moter. It is possible that our previous studies did not detectthe factor(s) we now observe because the probe used pre-viously (293/271) did not contain the entire CAN region(287/264) binding sites, which may have affected proteinbinding to this region. Indeed, complexes A and B appear tobe dependent upon an intact AP-1-like element and the flank-ing TGATGA sequence. Functionally, mutation of the puta-tive AP-1 site resulted in approximately a 63% decrease in thebasal promoter activity of StAR without affecting the cAMPresponse, a result very similar to the effects seen by mutatingthe CAAT box at 287 (26). Therefore, mutation of the CAATbox most likely affects a factor(s) binding to the AP-1-likeelement in the CAN region or vice versa.

Our results in MA-10 cells are distinct from the recentreport of murine StAR promoter activity in rat ovarian cells,which demonstrated that C/EBPb is a prominent componentin a DNA-protein complex formed with the 287/264 region(38). Interestingly, the binding site recognized by C/EBPb inovarian cell extracts is not the CAAT box core, but is theAP-1-like element. One possibility for this apparent tissue-specific difference in C/EBPb binding is that different factorsbind to the CAN region. This tissue-specific difference mayreflect the cycloheximide sensitivity of the cAMP responsefor StAR. Previously, StAR gene expression was reported tobe mediated by a cycloheximide-sensitive mechanism in hu-man granulosa-lutein cells (40). C/EBPb expression has beenshown to be increased by FSH in ovarian granulosa cells andby (Bu)2cAMP in MA-10 cells (25, 38). Subsequent to sub-mission of this paper, it was reported that C/EBPb is inducedin human granulosa cells by 8-bromo-cAMP (41). The datasupported C/EBPb in StAR basal promoter activity as wellas the proposal that the cycloheximide-sensitive, cAMP-dependent response may be linked to the induction ofC/EBPb. Thus, it is probable that a SF-1/C/EBPb or aGATA-4/C/EBPb interaction mediates the observed cyclo-heximide-sensitive response in rat ovarian granulosa cells.Consistent with this proposal, C/EBPb and SF-1 were shownto interact directly in a GST pull-down assay (26). However,StAR gene expression in MA-10 cells and Y1 mouse adre-nocortical cells is not dependent upon de novo protein syn-thesis (10). Therefore, another factor(s) that is insensitive tocycloheximide may be used in MA-10 Leydig cells and Y1cells to bind to the CAN region and promote maximal basalactivation of StAR. A difference in tissue-specific transcrip-tion factor expression may thus account for differences in theacute regulation of StAR in granulosa vs. Leydig cell cultures.

We identified GATA-4 as the factor that binds to the highlyconserved 268/244 region of the mouse StAR promoter.Functionally, GATA-4 contributes to 20% of the cAMP re-sponse. Independently, GATA-4 was recently reported tobind to the same region and was shown to be required formaximal cAMP response in rat granulosa cells (38). Mutation

of both the GATA-4 and AP-1-like elements decreased StARbasal promoter activity 97% and dramatically decreased theFSH-dependent increase in reporter gene expression in lu-teal-granulosa cells. In contrast, we demonstrate that a sim-ilar double mutation (GATAmut/AP1mut) had no greatereffect on StAR promoter activity in MA-10 cells than thecombined effects of the individual mutations within theseelements. Again, this discrepancy between our results andthose reported for rat granulosa cells indicates that the mu-rine StAR gene may be regulated in a tissue-specific manner.In MA-10 cells, we propose the GATA-4 transcription factorhas a role in stabilizing the factor(s) bound to the putativeAP-1 site and that GATA-4 and SF-1 work together to sta-bilize this common factor(s). However, these putative CAN-binding proteins and SF-1 can function in the absence ofGATA-4.

In sum, SF-1 binding is important for the basal activity ofthe StAR gene, most likely by affecting the interactionsamong the other proteins within the CAN region of thepromoter. GATA-4 binding is independent of SF-1 bindingand has an apparently minimal effect on SF-1- and CAN-protein binding, but may influence factor binding. Interac-tion between SF-1 and GATA-4 has been shown for regula-tion of the Mullerian-inhibiting substance gene; therefore,one model for SF-1 and GATA-4 is that they interact andfunction to stabilize binding of the CAN-binding factor(s)(42). Alternatively, it is possible that there is a redundancyof function for SF-1, CAN-binding protein, and GATA-4.This proposal would be consistent with our mutational anal-ysis that shows that elimination of one binding factor doesnot compromise the cAMP response of the StAR promoter.Thus, each of the DNA-binding factors may work together tostabilize the unknown cAMP-mediated transcription factorsand/or coactivators, but each factor alone may be sufficientto mediate part of the response.

Acknowledgments

We thank Dr. Keith Parker, University of Texas Southwestern Med-ical Center (Dallas, TX), for providing the GST-SF-1 expression plasmid.We also thank Drs. Carolyn M. Klinge and Keith C. Falkner, Departmentof Biochemistry and Molecular Biology, University of Louisville Schoolof Medicine (Louisville, KY), for their critical review of this manuscript.Finally, we thank Ms. Rebecca Combs for her excellent technicalassistance.

References

1. Stone D, Hechter O 1954 Studies on ACTH action in perfused bovine adrenals:site of action of ACTH in corticosteroidogenesis. Arch Biochem Biophys51:457–469

2. Karaboyas G, Koritz S 1965 Identity of the site of action of cAMP and ACTHin corticosteroidogenesis in rat adrenal and beef adrenal cortex slices. Bio-chemistry 4:462–468

3. Simpson ER, Jefcoate CR, Brownie AC, Boyd GS 1972 The effect of etheranaesthesia stress on cholesterol-side-chain cleavage and cytochrome P450 inrat-adrenal mitochondria. Eur J Biochem 28:442–450

4. Privalle CT, Crivello JF, Jefcoate CR 1983 Regulation of intramitochondrialcholesterol transfer to side-chain cleavage cytochrome P-450 in rat adrenalgland. Proc Natl Acad Sci USA 80:702–706

5. Stocco DM, Clark BJ 1996 Regulation of the acute production of steroids insteroidogenic cells. Endocr Rev 17:221–244

6. Caron KM, Soo SC, Wetsel WC, Stocco DM, Clark BJ, Parker KL 1997Targeted disruption of the mouse gene encoding steroidogenic acute regula-tory protein provides insights into congenital lipoid adrenal hyperplasia. ProcNatl Acad Sci USA 94:11540–11545

7. Lin D, Sugawara T, Strauss III JF, Clark BJ, Stocco DM, Saenger P, Rogol A,



Miller WL 1995 Role of steroidogenic acute regulatory protein in adrenal andgonadal steroidogenesis. Science 267:1828–1831

8. Miller WL 1997 Congenital lipoid adrenal hyperplasia: the human gene knock-out for the steroidogenic acute regulatory protein. J Mol Endocrinol 19:227–240

9. Clark BJ, Combs R, Hales KH, Hales DB, Stocco DM 1997 Inhibition oftranscription affects synthesis of steroidogenic acute regulatory protein andsteroidogenesis in MA-10 mouse Leydig tumor cells. Endocrinology138:4893–4901

10. Caron KM, Ikeda Y, Soo SC, Stocco DM, Parker KL, Clark BJ 1997 Charac-terization of the promoter region of the mouse gene encoding the steroidogenicacute regulatory protein. Mol Endocrinol 11:138–147

11. Lala DS, Rice DA, Parker KL 1992 Steroidogenic factor I, a key regulator ofsteroidogenic enzyme expression, is the mouse homolog of fushi tarazu-factorI. Mol Endocrinol 6:1249–1258

12. Honda S, Morohashi K, Nomura M, Takeya H, Kitajima M, Omura T 1993Ad4BP regulating steroidogenic P-450 gene is a member of steroid hormonereceptor superfamily. J Biol Chem 268:7494–7502

13. Morohashi K, Zanger UM, Honda S, Hara M, Waterman MR, Omura T 1993Activation of CYP11A and CYP11B gene promoters by the steroidogenic cell-specific transcription factor, Ad4BP. Mol Endocrinol 7:1196–1204

14. Luo X, Ikeda Y, Parker KL 1994 A cell-specific nuclear receptor is essential foradrenal and gonadal development and sexual differentiation. Cell 77:481–490

15. Wong M, Rice DA, Parker KL, Schimmer BP 1989 The roles of cAMP andcAMP-dependent protein kinase in the expression of cholesterol side chaincleavage and steroid 11b-hydroxylase genes in mouse adrenocortical tumorcells. J Biol Chem 264:12867–12871

16. Parissenti AM, Parker KL, Schimmer BP 1993 Identification of promoterelements in the mouse 21-hydroxylase (Cyp21) gene that require a functionalcyclic adenosine 39,59-monophosphate- dependent protein kinase. Mol Endo-crinol 7:283–290

17. Zhang P, Mellon SH 1996 The orphan nuclear receptor steroidogenic factor-1regulates the cyclic adenosine 39,59-monophosphate-mediated transcriptionalactivation of rat cytochrome P450c17 (17a-hydroxylase/c17–20 lyase). MolEndocrinol 10:147–158

18. Carlone DL, Richards JS 1997 Evidence that functional interactions of CREBand SF-1 mediate hormone regulated expression of the aromatase gene ingranulosa cells and constitutive expression in R2C cells. J Steroid Biochem MolBiol 61:223–231

19. Jacob AL, Lund J 1998 Mutations in the activation function-2 core domain ofsteroidogenic factor-1 dominantly suppresses PKA-dependent transactivationof the bovine CYP17 gene. J Biol Chem 273:13391–13394

20. Hammer GD, Krylova I, Zhang Y, Darimont BD, Simpson K, Weigel NL,Ingraham HA 1999 Phosphorylation of the nuclear receptor SF-1 modulatescofactor recruitment: integration of hormone signaling in reproduction andstress. Mol Cell 3:521–526

21. Sandhoff TW, Hales DB, Hales KH, McLean MP 1998 Transcriptional reg-ulation of the rat steroidogenic acute regulatory protein gene by steroidogenicfactor 1. Endocrinology 139:4820–4831

22. Sugawara T, Kiriakidou M, McAllister JM, Kallen CB, Strauss III JF 1997Multiple steroidogenic factor 1 binding elements in the human steroidogenicacute regulatory protein gene 59-flanking region are required for maximalpromoter activity and cyclic AMP responsiveness. Biochemistry 36:7249–7255

23. Clark BJ, Combs R 1999 Angiotensin II and cAMP induce human steroido-genic acute regulatory protein transcription through a common steroidogenicfactor-1 element. Endocrinology 140:4390–4397

24. Lekstrom-Himes J, Xanthopoulos KG 1998 Biological role of the CCAAT/enhancer-binding protein family of transcription factors. J Biol Chem273:28545–28548

25. Nalbant D, Williams SC, Stocco DM, Khan SA 1998 Luteinizing hormone-dependent gene regulation in Leydig cells may be mediated by CCAAT/enhancer-binding protein-b. Endocrinology 139:272–279

26. Reinhart AJ, Williams SC, Clark BJ, Stocco DM 1999 SF-1 (steroidogenicfactor-1) and C/EBPb (CCAAT/enhancer binding protein-b) cooperate toregulate the murine StAR (steroidogenic acute regulatory) promoter. MolEndocrinol 13:729–741

27. Mukai K, Mitani F, Agake R, Ishimura Y 1998 Adrenocorticotropic hormonestimulates CYP11B1 gene transcription through a mechanism involving AP-1factors. Eur J Biochem 256:190–200

28. Sun Y, Duckworth ML 1999 Identification of a placental-specific enhancer inthe rat placental lactogen II gene that contains binding sites for members of theEts and AP-1 (activator protein 1) families of transcription factors. Mol En-docrinol 13:385–399

29. Duval DL, Ellsworth BS, Clay CM 1999 Is gonadotrope expression of thegonadotropin releasing hormone receptor gene mediated by autocrine/para-crine stimulation of an activin response element? Endocrinology140:1949–1952

30. Orkin SH 1992 GATA-binding transcription factors in hematopoietic cells.Blood 80:575–581

31. Yomogida K, Ohtani H, Harigae H, Ito E, Nishimune Y, Engel JD, YamamotoM 1994 Developmental stage- and spermatogenic cycle-specific expression oftranscription factor GATA-1 in mouse Sertoli cells. Development120:1759–1766

32. Morrisey EE, Ip HS, Lu MM, Parmacek MS 1996 GATA-6: a zinc fingertranscription factor that is expressed in multiple cell lineages derived fromlateral mesoderm. Dev Biol 177:309–322

33. Heikinheimo M, Ermolaeva M, Bielinska M, Rahman NA, Narita N,Huhtaniemi IT, Tapanainen JS, Wilson DB 1997 Expression and hormonalregulation of transcription factors GATA-4 and GATA-6 in the mouse ovary.Endocrinology 138:3505–3514

34. Feng ZM, Wu AZ, Chen CL 1998 Testicular GATA-1 factor up-regulates thepromoter activity of rat inhibin a-subunit gene in MA-10 Leydig tumor cells.Mol Endocrinol 12:378–390

35. Ketola I, Rahman N, Toppari J, Bielinska M, Porter-Tinge SB, TapanainenJS, Huhtaniemi IT, Wilson DB, Heikinheimo M 1999 Expression and regu-lation of transcription factors GATA-4 and GATA-6 in developing mousetestis. Endocrinology 140:1470–1480

36. Dignam JD, Lebovitz RM, Roeder RG 1983 Accurate transcription initiationby RNA polymerase II in a soluble extract from isolated mammalian nuclei.Nucleic Acids Res 11:1475–1489

37. Sugawara T, Holt JA, Kiriakidou M, Strauss III JF 1996 Steroidogenic factor1-dependent promoter activity of the human steroidogenic acute regulatoryprotein (StAR) gene. Biochemistry 35:9052–9059

38. Silverman E, Eimerl S, Orly J 1999 CCAAT enhancer-binding protein b andGATA-4 binding regions within the promoter of the steroidogenic acute reg-ulatory protein (StAR) gene are required for transcription in rat ovarian cells.J Biol Chem 274:17987–17996

39. LaVoie HA, Garmey JC, Veldhuis JD 1999 Mechanisms of insulin-like growthfactor I augmentation of follicle-stimulating hormone-induced porcine steroi-dogenic acute regulatory protein gene promoter activity in granulosa cells.Endocrinology 140:146–153

40. Kiriakidou M, McAllister JM, Sugawara T, Strauss III JF 1996 Expression ofsteroidogenic acute regulatory protein (StAR) in the human ovary. J ClinEndocrinol Metab 81:4122–4128

41. Christenson LK, Johnson PF, McAllister JM, Strauss III JF 1999 CCAAT/enhancer-binding proteins regulate expression of the human steroidogenicacute regulatory protein (StAR) gene. J Biol Chem 274:26591–26598

42. Tremblay JJ, Viger RS 1999 Transcription factor GATA-4 enhances Mullerianinhibiting substance gene transcription through a direct interaction with thenuclear receptor SF-1. Mol Endocrinol 13:1388–1401



Documents

StAR