Steroidogenic Factor-1 and Early Growth Response Protein 1 Act through Two Composite DNA Binding Sites to Regulate Luteinizing Hormone β-Subunit Gene Expression*

Recent in vivo and in vitro studies have implicated the orphan nuclear receptor, steroidogenic factor-1 (SF-1), and the early growth response protein 1 (Egr-1) in the transcriptional regulation of the luteinizing hormone β-subunit (LHβ) gene. We have previously demonstrated the ability of SF-1 to bind to and transactivate the rat LHβ gene promoter acting at a consensus gonadotrope-specific element (GSE) located at position −127. We have now identified a second functional GSE site at position −59. In addition, based on electrophoretic mobility shift assay,in vitro translated Egr-1 is shown to bind to two putative Egr-1 binding sites (positions −112 and −50), which appear to be paired with the identified GSE sites. By transient transfection assay in pituitary-derived GH3 cells, it was seen that Egr-1 increases promoter activity of region −207/+5 of the rat LHβ gene promoter through action at both Egr-1 sites. Furthermore, LHβ gene promoter activity is markedly augmented in the presence of both factors together relative to activity in the presence of SF-1 or Egr-1 alone (150-fold versus 14-fold and 12-fold, respectively). These data define two composite SF-1-Egr-1 response-elements in the proximal LHβ gene promoter and suggest that SF-1 and Egr-1 act synergistically to increase expression of the LHβ gene in the gonadotrope.

Precise regulation of gonadotropin gene expression is required for normal reproductive function. The pituitary gonadotropins, luteinizing hormone and follicle-stimulating hormone, are composed of a common ␣-subunit linked to one of two unique ␤-subunits, LH␤ 1 or FSH␤. The common ␣-subunit can also associate with thyroid-stimulating hormone ␤-subunit in pituitary thyrotropes or, in humans, with placentally derived chorionic gonadotropin ␤-subunit.
Studies of the ␣-subunit gene promoter have identified a number of transcription factors and cognate cis-acting DNA elements that provide basal, tissue-specific, and hormonally mediated regulation of gene expression. In particular, a gonadotrope-specific element (GSE) is believed to be important for conferring gonadotrope-specific expression of the ␣-subunit gene (1,2).
The GSE, or Ad4 response element, regulates expression of a number of genes with a role in steroidogenesis, sexual differentiation, and adult reproductive function (3). The GSE/Ad4 site has been shown to interact with the transcription factor, steroidogenic factor-1 (SF-1), resulting in transcriptional activation of a variety of genes, including the steroidogenic P450, the Mullerian inhibiting substance, and the aromatase genes, among others (4 -6).
SF-1 is a member of the nuclear hormone receptor superfamily. Although it was previously considered to be an orphan member of this family, it has recently been reported that SF-1-dependent transcriptional activity is increased in the presence of cholesterol metabolites, particularly 25-OH-cholesterol (7). It is currently unknown whether this putative ligand is important for SF-1 function in nonsteroidogenic tissues, such as the pituitary gland. SF-1 expression is restricted to the steroidogenic cells of the adrenal gland and gonads and to the gonadotrope subpopulation of the pituitary gland. Thus, the pattern of SF-1 expression suggests that SF-1 may contribute to tissue-specific gene expression.
In contrast with the ␣-subunit, progress has only recently been made in the identification of transcription factors that regulate the gonadotropin ␤-subunit genes. Understanding ␤-subunit regulation at the molecular level is critical because it is the ␤-subunits that provide the functional specificity that distinguishes luteinizing hormone and follicle-stimulating hormone action.
A number of recent reports have suggested that SF-1 may play a role in the regulation of LH␤ gene expression in addition to its effects on the common ␣-subunit. Targeted disruption of the Ftz-F1 gene encoding SF-1 results in transgenic mice that lack transcripts for the gonadotrope markers LH␤, FSH␤, and gonadotropin-releasing hormone receptor and have greatly diminished levels of ␣-subunit mRNA (3). Gonadotropin-releasing hormone replacement was able to restore gonadotropin expression in four out of five of these animals, suggesting that cells from the gonadotrope lineage are present but that a specific defect in gonadotropin subunit gene expression exists (8).
By sequence homology, it has been shown that the LH␤ gene promoter contains a consensus GSE site at position Ϫ127 in the rat and Ϫ125 in the cow. In vitro studies demonstrated the ability of SF-1 to bind to and transactivate the rat LH␤ gene promoter through interaction with this putative GSE site (9). The physiologic significance of this site was confirmed using transgenic mice containing either the wild-type bovine LH␤ gene promoter or a promoter with a GSE site-specific mutation linked to a CAT reporter vector (10). Introduction of the GSE mutation substantially decreased LH␤ gene promoter activity, indicating that the SF-1 binding site is a critical regulator of LH␤ gene promoter activity in vivo. Of note, however, in both the in vitro and in vivo model systems, mutation of this GSE site failed to eliminate fully LH␤ gene expression, suggesting the presence of a second functional SF-1-response element.
Closer inspection of the rat LH␤ gene promoter sequence identified a second region with similarity to the consensus GSE site (Fig. 1). We have termed this second putative site the 3ЈGSE and refer to the previously defined site as the 5ЈGSE. The first aim of the current study was to determine the functional significance of the putative 3ЈGSE site.
In vivo observations in transgenic mice have also suggested a role for the transcription factor, early growth response protein 1 (Egr-1), in the modulation of LH␤ gene expression. Egr-1, also known as NGFI-A, zif/268, and Krox-24, is a member of the immediate early gene family, the members of which contain a zinc finger domain with a Cys 2 -His 2 motif that recognizes the nucleotide sequence CGCCC(C/A)CGC. Additional members of this family include Sp1 and the Wilms' tumor suppressor WT1 (11)(12)(13). Egr-1 is widely expressed, and in a recent report, it has been shown to be expressed in the gonadotrope, as well as in the somatotrope subpopulation of the pituitary (14). Of interest, two groups have demonstrated that targeted disruption of the Egr-1 gene results in specific loss of LH␤ gene expression, with maintenance of FSH␤ gene expression (14,15). Sequence analysis of the LH␤ gene promoter identifies two highly conserved regions with homology to the consensus Egr-1 binding site (Fig. 1). Lee et al. (15) have demonstrated that mutation of the 3ЈEgr site correlates with loss of Egr-1-mediated transactivation. However, the ability of Egr-1 to interact directly with the 3ЈEgr site and the functional importance of the putative 5ЈEgr site have not been determined.
In the study reported here, we have therefore verified the ability of in vitro translated Egr-1 to bind to both of the putative Egr-1 binding sites and have shown that both sites confer Egr-1 responsiveness to the rat LH␤ gene promoter. In addition, we have demonstrated the ability of SF-1 and Egr-1 to interact with each other and have characterized the functional importance of this interaction to the regulation of LH␤ gene expression.

EXPERIMENTAL PROCEDURES
Oligonucleotides-The oligonucleotides used for mutagenesis, electrophoretic mobility shift assays (EMSAs), and polymerase chain reactions are shown in Table I. The nucleotide sequence of the rat LH␤ gene promoter is based on newly obtained sequencing data available at GenBank TM accession number AF020505, which differ slightly from that of Jameson et al. (16). The 5ЈGSE oligonucleotide contains additional 5Ј-BamHI and 3Ј-BglII restriction sites. The Ϫ207LH-S, Ϫ82LH-S, and 5ЈLH-S primers introduced BamHI restriction sites, whereas the ϩ5LH-AS primer introduced a HindIII site, and the 3ЈLH-AS primer introduced BglII and XhoI restriction sites (restriction sites not shown).
In Vitro Translated Proteins and Antisera Used in EMSA-In vitro translated proteins were generated from plasmids containing 2.1 kilobase pairs of the mouse SF-1 cDNA (provided by Dr. K. L. Parker, Southwestern University School of Medicine) or 3.2 kilobase pairs of the mouse Egr-1 cDNA (provided by Dr. D. Nathans, Johns Hopkins University) using the TNT Coupled Reticulocyte Lysate System (Promega, Madison, WI) (11,17). The resultant product was determined to be of appropriate size by comparison with [ 35 S]methionine-labeled protein markers by SDS-polyacrylamide gel electrophoresis (PAGE).
The SF-1 polyclonal antibody, a generous gift of Dr. Parker, was generated in rabbits against a glutathione S-transferase-SF-1 fusion protein (18). The Egr-1 antiserum is a rabbit affinity-purified polyclonal antibody raised against the carboxyl terminus of mouse Egr-1 (Santa Cruz Biotechnology, Inc, Santa Cruz, CA). The Pit-1 antiserum, directed against amino acids 136 -150 of rat growth hormone pituitary factor-1/Pit-1, was provided by C. Bancroft (Mt. Sinai School of Medicine) (19). Electrophoretic Mobility Shift Assays-Region Ϫ141/Ϫ44 of the rat LH␤ gene promoter (see Fig. 5B) was produced by polymerase chain reaction using primers 5ЈLH-S and 3ЈLH-AS and subcloned into the pGEM-T vector (Promega). BamHI and BglII restriction enzymes were used to obtain the insert, which was agarose gel-purified and dephosphorylated. The remainder of the double-stranded oligonucleotides used as probes and in cold competition experiments were produced by annealing the sense oligonucleotide indicated in Table I with the corresponding antisense oligonucleotide (not shown). Probes were created by T4 polynucleotide kinase end-labeling with [␥-32 P]ATP followed by purification over a NICK column (Amersham Pharmacia Biotech).
In vitro translated protein(s) were incubated with 50,000 cpm of oligonucleotide probe in DNA-binding buffer (20 mM HEPES (pH 7.9), 60 mM KCl, 5 mM MgCl 2 , 10 mM phenylmethylsulfonyl fluoride, 10 mM dithiothreitol, 1 mg/ml bovine serum albumin, and 5% (v/v) glycerol) for 30 min on ice. For competition studies, a 200-fold molar excess of unlabeled oligonucleotide was added 5 min prior to the addition of probe. Where indicated, antiserum (1 l) was added 30 min after the addition of probe, and the incubation was continued for 2 h. Protein-DNA complexes were resolved on a 5% nondenaturing polyacrylamide gel in 0.5ϫ Tris borate-EDTA buffer and subjected to autoradiography and/or quantification using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
In Vitro Protein Interaction Assays-Protein interaction assays were performed as described previously (20). Briefly, a His 6 -containing SF-1 fusion protein was obtained by baculovirus transfection of insect cells. The fusion protein was purified by affinity chromatography with nickelnitrilotriacetic acid (NTA) agarose (Qiagen, Hilden, Germany) followed by washing with 40 mM imidazole. In vitro translation of the Egr-1, DAX-1, and RXR␣ proteins was performed with the TNT Coupled Reticulocyte Lysate System (Promega, Madison, WI) in the presence of [ 35 S]methionine. Labeled proteins were incubated with the His 6 -SF-1 fusion protein bound to Ni-NTA agarose in buffer containing 20 mM HEPES, pH 7.9, 10% glycerol, 50 mM KCl, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 40 mM imidazole for 2 h at 4°C. After extensive washing, bound proteins were eluted by boiling in SDS sample buffer followed by 10% PAGE separation and autoradiography.
The deletion mutants of Egr-1 were created by digesting the pB-SII(ϩ) vector containing the Egr-1 cDNA with either Eco47III, MscI, or SphI. The MscI-digested cDNA was treated with T4 DNA polymerase to create a blunt end. These vectors were then digested with EcoRI, yielding cDNA fragments having a 5Ј EcoRI restriction site and a blunt 3Ј-end. Next, the pCMX vector (provided by Dr. R. Evans, The Salk Institute) was digested with NheI, treated with T4 DNA polymerase, and then digested with EcoRI. The Eco47III-, MscI-, and SphI-digested cDNAs were subcloned into the altered pCMX vector to create pCMX Egr-1(1-533), Egr-1 del 486 -533, and Egr-1 del 318 -533, respectively.
Plasmids Used in Transfection Studies-Reporter constructs used for these studies were created by subcloning polymerase chain reaction products containing the LH␤ gene promoter sequence into the pXP2 vector using BamHI/HindIII restriction sites, which were introduced by primers (18). The largest construct used for these studies contained 207 base pairs of the 5Ј-flanking sequence of the rat LH␤ gene and the first 5 base pairs of the 5Ј-untranslated region generated by primers Ϫ207LH-S and ϩ5LH-AS. A 5Ј-truncated construct was obtained by subcloning the polymerase chain reaction product obtained with primers Ϫ82LH-S and ϩ5LH-AS.
Mutations in the reporter constructs were created using the Transformer Site-Directed Mutagenesis Kit (CLONTECH Laboratories, Inc., Palo Alto, CA). Generation of multiple mutations was performed sequentially and, therefore, required the use of two selection primers, one that converted a unique HindIII restriction site to a unique MluI site (pXP2) and another that reversed this mutation (pXP2-rev). The 5ЈGSE mutagenic primer eliminated a TthIII1 restriction site, in addition to introducing the desired mutation, as described previously (9). The mutagenic primers for the 3ЈGSE, 5ЈEgr, and 3ЈEgr sites introduce EcoRI, ScaI, and PstI restriction sites, respectively. All reporter constructs were confirmed by dideoxysequencing.
The SF-1 expression vector contained 2.1 kilobase pairs of the mouse SF-1 cDNA driven by cytomegalovirus promoter sequences in the vector, pCMV5 (17). The Egr-1 expression vector was created by cloning 3.2 kilobase pairs of the mouse Egr-1 cDNA into pCMV5 at BamHI and HindIII restriction sites (11).
Transfection Experiments-Monkey kidney fibroblast (CV-1) cells or rat somatolactotrope (GH 3 ) cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. CV-1 cells growing in 3.5-cm tissue culture wells (Flow Laboratories, McLean, VA) were transfected with expression (0.1 g/well) and reporter (1.65 g/ well) plasmids using the calcium phosphate precipitation method (21). GH 3 cells were cultured to 50 -70% confluence and transfected by electroporation. Approximately 5 ϫ 10 6 cells were suspended in 0.4 ml of Dulbecco's phosphate-buffered saline plus 5 mM glucose with the DNA to be transfected. The cells received a single electrical pulse of 240 V at a total capacitance of 1000 microfarads using an Invitrogen Electroporator II apparatus (Invitrogen, San Diego, CA).
For both cell types, control wells received the appropriate "empty" expression vector. Cotransfection with an Rous sarcoma virus-␤-galactosidase plasmid (1 g/well) allowed correction for differences in transfection efficiency between wells. Cells were harvested 48 h after transfection, and the cell extracts were analyzed for both luciferase and ␤-galactosidase activities (22,23). Luciferase activity was first normalized to the level of ␤-galactosidase activity. Results were then calculated as fold change relative to expression in the presence of the control empty expression vector. Data are shown as the mean Ϯ S.E. and represent a minimum of three independent experiments, with each point run in triplicate in each experiment.

SF-1 Binds Specifically to the Putative 3ЈGSE
Site-We have previously demonstrated the ability of SF-1 to increase rat LH␤ gene promoter activity specifically through action at a functional GSE site located at position Ϫ127/Ϫ119, now designated as the 5ЈGSE site (9). These studies were primarily based on transient transfection of the monkey kidney fibroblast cell line, CV-1, with preliminary confirmation in the rat pituitary-derived somatolactotrope cell line, GH 3 (9). By Northern analysis, CV-1 cells lack the mRNA that encodes the SF-1 homologue Ad4BP (24). We have also demonstrated that both CV-1 and GH 3 nuclear extracts fail to produce specific protein-DNA interactions with the 5ЈGSE probe by EMSA (data not shown).
In these earlier studies, it was noted that residual SF-1 responsiveness could be observed following mutation of the 5ЈGSE element by a CC to AA conversion at positions Ϫ124 and Ϫ123. These results suggested two possibilities: 1) the importance of additional base pairs in this GSE site to the SF-1response, or 2) the presence of additional SF-1 binding sites that contribute to the regulation of LH␤ gene promoter activity.
Sequence analysis of the rat LH␤ gene promoter identified a number of regions that resemble the consensus GSE, including region Ϫ59/Ϫ51, which was designated the putative 3ЈGSE site (Fig. 1). We therefore tested the ability of SF-1 to bind to this region of the LH␤ gene promoter using EMSA. As shown in Fig.  2A, lane 1, in vitro translated SF-1 bound to a 32 P-labeled 3ЈGSE probe to produce the complex indicated by the arrow. The specificity of this interaction was demonstrated by successful competition with unlabeled 3ЈGSE (Fig. 2A, lane 2) but not with the mutated 3ЈGSE site (3ЈGSEM) (Fig. 2A, lane 3) nor with an unrelated oligonucleotide containing two binding sites for the pituitary transcription factor, Pit-1 ( Fig. 2A, lane 4). Unprogrammed reticulocyte lysate failed to bind to this probe ( Fig. 2A, lane 5).
In order to confirm that the observed band contained SF-1, we investigated the effect of a SF-1-specific antibody on the formation of the identified complex. This antibody has previously been shown to block the ability of SF-1 to bind to the promoter element of a number of genes, including the glycoprotein ␣-subunit and 21-hydroxylase genes (2,25). Treatment with this SF-1-specific antiserum decreased the intensity of the protein-DNA complex, whereas the addition of an anti-Pit-1 antiserum, used as a negative control, had no effect ( Fig. 2A,  lanes 8 -10). The upper band, indicated by an asterisk, was observed in reactions that lacked in vitro translated protein (data not shown). This band is therefore presumed to represent a direct interaction between the antibody and the probe and does not represent supershift of the SF-1-DNA complex.
In order to localize further the SF-1-recognition site, a 3-base pair mutation was introduced into the wild-type LH␤ oligonucleotide sequence (3ЈGSE) to form mutated 3ЈGSE. As seen in Fig. 2A, lanes 6 and 7, no additional complexes were observed beyond those produced in the presence of unprogrammed reticulocyte lysate.
Parallel EMSA was performed using nuclear extract from the gonadotrope-derived cell line, ␣T3-1. This cell line has previously been shown to express both SF-1 mRNA and protein (2,3). Endogenous SF-1 present in ␣T3-1 nuclear extract was also able to bind to the 3ЈGSE region, as demonstrated by successful competition with the SF-1-specific antibody (data not shown). These data firmly establish that the putative 3ЈGSE site in the rat LH␤ gene promoter is recognized by SF-1, present as either an endogenous protein or an in vitro translated product.
The 5ЈGSE and 3ЈGSE Sites Have Similar Affinity for in Vitro Translated SF-1-We next wished to determine the relative affinities of the two GSE sites that we had identified for SF-1 binding. EMSA was performed using in vitro translated SF-1 and the 3ЈGSE oligonucleotide probe with an increasing molar excess of unlabeled 3ЈGSE or 5ЈGSE oligonucleotides  (16), cow (35), pig (36), sheep (37), horse (38), and human (39) are aligned with the core SF-1 binding site (GSE sequence) as defined in the human glycoprotein ␣-subunit and with the cognate Egr-1 site (1,31). Nucleotide numbers are relative to the transcriptional start site of the rat sequence as defined by Jameson et. al. (16). B, schematic of the rat LH␤ gene promoter illustrating the presence of two composite GSE/Egr-1 elements. (Fig. 2B). Based on this approach, the affinity of SF-1 for either of these sites is essentially identical within the sensitivity of this approach.
The Putative 3ЈGSE Site Contributes to SF-1-stimulated Increases in LH␤ Promoter Activity-In order to determine the functional significance of the interaction between SF-1 and the LH␤ gene promoter sequences, transient transfection assays were performed in CV-1 cells, the cell line used in characterization of the 5ЈGSE site (9). Consistent with our previously published results, cotransfection with the SF-1 expression vector resulted in an 80-fold increase in the luciferase activity of a wild-type reporter construct containing region Ϫ207 to ϩ5 of the rat LH␤ gene promoter (Fig. 3A). Mutation of either the 5ЈGSE or 3ЈGSE sites eliminated over 90% of the SF-1-response, whereas mutation of both sites decreased the SF-1response to the level of the empty reporter plasmid, pXP2.
Ideally, we would have pursued these investigations in a gonadotrope-derived cell line. However, at the time these experiments were performed, only the ␣T3-1 cell line was available for study. Although these cells express the endogenous gonadotropin ␣-subunit gene, they do not express endogenous or exogenous ␤-subunit genes at appreciable levels (26). Al-though we were able to detect a small increase in wild-type Ϫ207/ϩ5 LH␤ gene promoter activity following cotransfection of expression vectors for SF-1 or Egr-1, the overall low expression of the reporter construct limited the usefulness of this cell line for these studies (data not shown).
More recently, we have attempted to confirm these results using a newly isolated gonadotrope-derived cell line, L␤T2, which expresses the endogenous LH␤ gene (27). This cell line expresses SF-1 mRNA on Northern analysis and produces SF-1, or an immunologically related protein, based on EMSA studies of L␤T2 nuclear extracts (data not shown). In initial investigations, mutation of either the 5ЈGSE or 3ЈGSE site in the Ϫ207/ϩ5 LH␤-luciferase construct markedly decreased reporter expression, consistent with loss of the ability to respond to endogenous SF-1-stimulation. Furthermore, cotransfection with either the SF-1 or Egr-1 expression vectors increased LH␤ gene promoter activity by 2.5-fold and 3-fold, respectively (data not shown). Additional studies in this cell line are currently underway.
With the aim of verifying our results in a pituitary cell line, we repeated these studies in the somatolactotrope cell line, GH 3 (Fig. 3B). SF-1 markedly increased Ϫ207/ϩ5 LH␤ pro- Incubation with antiserum specific to SF-1 or Pit-1 was performed in lanes 9 and 10, respectively. Lanes 5 and 7 contain the unprogrammed reticulocyte lysate used for in vitro translation. The arrowhead indicates the specific binding complex. The asterisk indicates nonspecific binding. B, EMSA was performed using in vitro translated SF-1 and the 32 P-labeled 3ЈGSE oligonucleotide probe. Increasing molar excess of unlabeled 3ЈGSE or 5ЈGSE oligonucleotide was added to the reaction mixture, and the intensity of the SF-1-DNA complex formed was quantified by phosphorimaging. The 50% inhibitory concentration was then calculated for each of the GSE sites. moter activity in GH 3 cells, although to a lesser degree than that observed for CV-1 cells (14-fold versus 80-fold). Mutation of either of the GSE sites blunted the SF-1 response, whereas mutation and/or deletion of both sites was required to decrease the SF-1 effect to the level of the empty reporter construct.
Interestingly, although the 3ЈGSE and 5ЈGSE sites conferred similar SF-1 responsiveness in CV-1 cells, the 3ЈGSE site appears to be more important in the GH 3 cell line. Also of note, in both cell lines, the response in the presence of both sites together exceeds the additive response to either site alone, although a clearly synergistic interaction was more apparent in the CV-1 cell line. Nevertheless, these data strongly suggest that both the 3ЈGSE and 5ЈGSE sites contribute to the SF-1response in the LH␤ promoter, with tissue-specific differences in the magnitude and relative importance of each site.
Egr-1 Binds to Both Putative Egr-1 Binding Sites in the LH␤ Gene Promoter-In a recent report, specific loss of pituitary LH␤ mRNA expression was described in a transgenic mouse line deficient in Egr-1 (15). In these studies, Lee et al. (15) demonstrated that mutation of region Ϫ50 to Ϫ42 of the rat LH␤ gene promoter sequence eliminated the ability of an Egr-1 expression plasmid to increase promoter activity in both ␣T3-1 and CV-1 cells. On further sequence analysis, we identified a second putative Egr-1 site at position Ϫ112/Ϫ104 in the prox-imal rat LH␤ gene promoter with homology to the consensus Egr-1 binding site. We have designated these two regions the putative 5ЈEgr and 3ЈEgr sites (Fig. 1). Interestingly, these putative Egr-1 elements are paired with the GSE sites, suggesting a functional interaction between the two associated transcription factors.
EMSA studies were initiated to determine whether Egr-1 could bind to either of these putative elements. For these assays, probes were utilized that spanned the paired GSE and Egr sites in either the 3Ј or 5Ј regions (Fig. 4, A and B, respectively). In vitro translated SF-1 (Fig. 4, lanes 1 and 2) or Egr-1 (lanes 3 and 4) were added to the wild-type probes to produce the complexes indicated by the labeled arrows. The addition of an SF-1 blocking antibody (lane 2) or an Egr-1 supershifting antibody (lane 4) confirmed the presence of the expected protein in each of the complexes. In order to demonstrate that the Egr-1 was binding to the predicted base pairs in these probes, limited mutations were introduced into the wild-type oligonucleotides within the putative Egr-1 binding sites to form 3ЈGSE-3ЈEgrM and 5ЈGSE-5ЈEgrM (Table I). When used as probes, these mutations specifically eliminated binding by Egr-1 (Fig. 4, lane 6) while preserving recognition by SF-1 (lane 5). These results clearly support the ability of in vitro translated Egr-1 to bind to both of the identified putative Egr-1 sites.
Affinity for these two sites was determined using an EMSA approach analogous to that used to determine relative affinity  3, 4, and 6) was added to 32 P-labeled oligonucleotide probes that spanned paired GSE-Egr-1 sites, as indicated. Incubation with antiserum specific to SF-1 (lane 2) or Egr-1 (lane 4) was also performed. A, SF-1 and Egr-1 binding to a probe that spans the wild-type 3ЈGSE and 3ЈEgr sites (left panel) or a probe containing point mutations in the putative 3ЈEgr site (3ЈEgrM) (right panel). B, results utilizing a probe containing the wild-type or Egr-mutated 5ЈGSE-5ЈEgr oligonucleotides. of the GSE sites (Fig. 2B). The 5Ј GSE-5ЈEgr and 3ЈGSE-3ЈEgr oligonucleotides were used to compete for binding of in vitro translated Egr-1 to the 3ЈGSE-3ЈEgr probe. Based on this assay, the affinity of the 3ЈEgr site for Egr-1 is approximately 10-fold greater than the affinity of the 5ЈEgr site (data not shown).
EMSA Fails to Demonstrate an SF-1-Egr-1 Protein-Protein Interaction-Lee et al. (15) reported a synergistic increase in LH␤ gene promoter activity following transfection with both SF-1 and Egr-1 expression vectors. We wished to determine whether this observed functional cooperativity could be explained by direct protein-protein interaction between the two transcription factors, particularly in view of the noted pairing of their respective binding sites. EMSA was performed using a probe that spanned the 3ЈGSE and 3ЈEgr sites (Fig. 5A) or a probe containing base pairs Ϫ141 to Ϫ44 that spanned both GSE sites and both putative Egr-1 binding sites (Fig. 5B). As seen in Fig. 5A and confirmed by antibody treatment, the co-addition of in vitro SF-1 and Egr-1 (lanes 3-6) resulted in the production of two complexes that contained either SF-1 or Egr-1 but failed to demonstrate a higher order complex. Similarly, no evidence of SF-1-Egr-1 heterodimerization was detected using a probe that spanned the 5Ј and 3Ј GSE-Egr-1 pairs (oligonucleotide 5ЈGSE-3ЈEgr) (Fig. 5B, lane 5). Of note, the intensity of the SF-1 and Egr-1 DNA complexes was the same whether they were produced in the presence of one or both of the transcription factors. Neither the addition of endogenous SF-1 (present in ␣T3-1 nuclear extracts) nor changes in the binding conditions of the EMSA incubation reaction produced a detectable SF-1-Egr-1 complex (data not shown).

SF-1 Interacts Directly with Egr-1 in a Protein-Protein
Interaction Assay-Because EMSA failed to demonstrate an interaction between SF-1 and Egr-1, we attempted to detect proteinprotein binding using an alternative assay. A His 6 -SF-1 fusion protein linked to Ni-NTA agarose was tested for the ability to bind to Egr-1 that had been radiolabeled with [ 35 S]methionine during in vitro translation. This assay has been used previously to detect an interaction between SF-1 and another gonadotrope-expressed transcription factor, DAX-1 (20). As shown in Fig. 6, Egr-1 bound to the SF-1 fusion protein, whereas no binding was observed to the Ni-NTA beads alone. Relative to the amount of radiolabeled input protein, approximately 6% of the Egr-1 bound to SF-1, similar in magnitude to the interaction between SF-1 and DAX-1 (7% of input). In contrast, no binding was detected between RXR and SF-1. Deletion of the carboxyl terminus of Egr-1 had no effect on, or even increased, binding (9% of input), whereas a more extensive deletion across the Egr-1 DNA binding domain eliminated the observed interaction. The lack of interaction between SF-1 and RXR and the loss of interaction following truncation of Egr-1 both strongly suggest that the binding of SF-1 to full-length Egr-1 is specific.
Egr-1 Increases LH␤ Gene Promoter Activity through Action at Both Putative Egr-1 GSE Sites-In order to verify the functional significance of each of the putative Egr-1 binding sites, transient transfection experiments were performed in GH 3 cells. We first tested the effect of introducing mutations into either of the putative Egr-1 sites within the Ϫ207/ϩ5 region of the LH␤ gene promoter. These mutations corresponded to the mutations that eliminated binding by in vitro translated Egr-1 on EMSA. In order to optimize for detection of the response to the weaker 5ЈGSE site, cells were cotransfected with relatively high amounts of Egr-1 expression vector (5 g/well), whereas lesser amounts were used for subsequent experiments (1 g/ well). As seen in Fig. 7A, LH␤ gene promoter activity increased by approximately 12-fold in the presence of Egr-1 in the wildtype construct. Following mutation in the putative 3ЈEgr or 5ЈEgr sites, the Egr-1 response decreased to 1.7-fold or 3.5-fold, respectively, but remained significantly greater than control expression. Mutation in both putative Egr sites eliminated Egr-1 responsiveness.
SF-1 and Egr-1 Act Synergistically to Increase LH␤ Gene Promoter Activity-We next investigated whether the observed protein-protein interaction between SF-1 and Egr-1 (Fig. 6) would result in cooperative effects on LH␤ gene promoter function. As seen in Fig. 7B, cotransfection with both SF-1 and Egr-1 resulted in marked synergistic stimulation of LH␤ gene expression to over 150-fold, consistent with previous reports in CV-1 cells (15).
In order to determine whether both Egr sites are important for this SF-1-Egr-1 interaction, LH␤ gene promoter activity 5Ј-TTTCTGACCTTGTCTGTCTCGCCCCCAAAGAG-3Ј

5Ј-GAGGGGGTGGCAAGGCCACTAAGCAGAGG-3Ј
a Position derived from revised rat LH␤ promoter sequence (GenBank™ accession number AF020505), except for the Pit-1 oligonucleotide, which is based on the rat growth hormone promoter sequence (40). b EMSA and mutagenesis oligonucleotides shown as sense strand. c Includes rat LH␤ promoter sequences as shown plus sequences from pXP2 vector multiple cloning site. d PCR primers indicated by S (sense) or AS (antisense).
was measured in the presence of both transcription factors in Ϫ207/ϩ5 LH␤ gene reporter constructs containing the various combination of Egr-1 site mutations (Fig. 8A). The synergistic response to Egr-1 and SF-1 was maintained, although blunted, with mutation in either the 3ЈEgr or 5ЈEgr site. These data demonstrate that both of the Egr-1 binding sites in the LH␤ gene promoter contribute independently to the SF-1-Egr-1 synergistic effect. SF-1 Interacts Functionally with Egr-1 through Both Adjacent and Spaced Binding Sites-As noted previously, the SF-1 and Egr-1 binding sites appear to be present as pairs in the LH␤ gene promoter, albeit with slightly different spacing between the paired sites. This localization raised the possibility that the observed SF-1-Egr-1 functional interaction was dependent on the close proximity of the two binding sites, perhaps promoting heterodimer formation. Alternatively, as suggested by the mutually exclusive binding observed on EMSA, the proximity of the sites might prohibit simultaneous binding by the two transcription factors. In this situation, SF-1 and Egr-1 would presumably act through interaction between the two defined regions. Mutation of the 5ЈGSE site allowed evaluation of the 3ЈGSE site with the two Egr-1 binding sites. Additional mutation of either Egr-1 site decreased, but did not eliminate, the SF-1-Egr-1 synergistic effect on LH␤ gene promoter activ- ity. These results suggest that both of these mechanisms are functioning, i.e. the SF-1 response may be augmented by Egr-1 acting on either the adjacent or the more distant Egr-1 binding site.

DISCUSSION
The gonadotropins are critical modulators of reproductive development and function, acting on the gonads to stimulate both steroidogenesis and gametogenesis. A wide variety of studies suggest that gonadotropin biosynthesis is tightly controlled and depends, to a large degree, on regulation at the level of gene transcription. Although a number of transcription factors have been identified that modulate expression of the ␣-subunit gene, only recently have studies begun to identify the molecular mechanisms that regulate ␤-subunit gene expression.
The results reported here substantially extend our understanding of the transcription factors and DNA response elements that stimulate activity of the LH␤ gene promoter. These studies clearly demonstrate the ability of SF-1 to bind to a second region of the rat LH␤ gene promoter with homology to the consensus GSE site. This site, located at position Ϫ59, acts in conjunction with the previously identified GSE site at posi-tion Ϫ127 to confer marked SF-1 responsiveness to the LH␤ gene. Our studies also demonstrate the ability of in vitro translated Egr-1 to bind to and transactivate the LH␤ promoter through action at two sites located at positions Ϫ112 and Ϫ50. In addition, these experiments confirm the previously reported synergistic induction of LH␤ gene expression in the presence of both transcription factors, SF-1 and Egr-1 (15). As we now demonstrate, this functionally cooperativity appears to be provided by both Egr-1 binding sites and, at least in the presence of an intact 3ЈGSE site, occurs through interactions between either proximal or spaced SF-1 and Egr-1 regulatory elements.
In the functional studies reported here, we utilized a heterologous system in which expression vector(s) for SF-1 and/or Egr-1 and a reporter construct containing LH␤ gene promoter sequences were transiently transfected into a pituitary-derived somatolactotrope cell line, GH 3 . Although these studies ideally would have been performed in a gonadotrope-derived cell line, there is precedent for the study of the LH␤ gene in GH 3 cells. The GH 3 cell line has been shown to support transcription initiation from the authentic start site of the LH␤ gene and to allow cAMP-mediated increases in LH␤ promoter activity (28). These cells have also been utilized to identify an estrogenresponsive element in the LH␤ gene promoter (29). When transfected with the gonadotropin-releasing hormone receptor, GH 3 cells have been shown to support gonadotropin-releasing hormone-induced regulation of gonadotropin subunit promoter activity, which closely parallels the regulation observed in primary pituitary cells (30). As indicated under "Results," preliminary studies in the newly available, gonadotrope-derived L␤T2 cell line have supported the results obtained in the GH 3 cell line.
SF-1 is a member of the nuclear hormone receptor superfamily, which includes the thyroid hormone, estrogen, progesterone, and retinoic acid receptors. Interestingly, the consensus GSE sequence resembles a nuclear receptor binding half-site. Although members of the nuclear hormone receptor family are best known for binding to pairs of recognition half-sites, monomer binding to a single 5Ј-extended half-site has been described for both SF-1 and another orphan nuclear receptor, NGFI-B (31). Although we have identified two GSE sites in the proximal LH␤ promoter, many aspects of our data argue against the need for SF-1 multimer binding to two DNA halfsites. Although DNA bending could allow for close approximation of the two GSE sites, interaction between nuclear hormone binding sites with separations of over 70 base pairs has not been previously described. From a functional standpoint, we have demonstrated that SF-1-mediated increases in LH␤ promoter activity can be observed in the presence of a single intact GSE site (Fig. 3).
Our data demonstrate functional cooperativity between SF-1 and Egr-1 on the rat LH␤ gene promoter activity and suggest that this effect may be due to direct interaction between these two transcription factors, as demonstrated by use of a Histagged SF-1 fusion protein. Based on preliminary data using Egr-1 truncation mutants, formation of SF-1-Egr-1 heterodimers requires the presence of the Egr-1 DNA binding domain. Further studies are under way to characterize fully the SF-1 and Egr-1 protein domains that are required for protein-protein interactions and/or functional synergy.
For reasons that remain unclear, SF-1-Egr-1 protein interactions could not be detected on EMSA despite multiple attempts to optimize binding conditions. Although this result could be due to a weak interaction between the two proteins, the amount of Egr-1 bound as a percentage of input was similar to the interactions of RXR␣ and thyroid hormone receptor ␤ in analogous assays (data not shown). Precedent exists for diffi-FIG. 8. SF-1 and Egr-1 interact synergistically on the LH␤ gene promoter. GH 3 cells were transiently transfected with a luciferase reporter construct containing region Ϫ207/ϩ5 of the rat LH␤ gene promoter present as either the wild-type sequence or with various combinations of mutations in the putative GSE and Egr-1 sites. Cells were cotransfected with plasmids encoding either SF-1 and/or Egr-1, and the results were calculated as in Fig. 3, with each point representing the mean Ϯ S.E. of at least nine samples from three independent experiments. A, SF-1 and Egr-1 stimulation of LH␤ promoter activity with mutation of the Egr-1 sites. B, interaction between the intact 3ЈGSE site and either Egr-1 binding site. Ⅺ, SF-1; f, Egr-1; p, SF-1 ϩ Egr-1. culty in detecting interactions between SF-1 and other transcription factors by EMSA. For example, despite clear functional interactions, the ability of SF-1 to bind to DAX-1 required the use of the His 6 -SF-1 approach utilized for these studies (20).
SF-1 has previously been shown to bind to DAX-1, another orphan member of the nuclear hormone receptor superfamily with effects on gonadotrope function. Mutations in DAX-1 are believed to cause hypogonadotropic hypogonadism in conjunction with X-linked adrenal hypoplasia congenita (20,32). Ito et al. (20) have demonstrated that DAX-1 blunts the ability of SF-1 to increase expression of a heterologous reporter construct containing a consensus GSE site. This DAX-1 effect did not require DNA binding by DAX-1. In contrast, Egr-1 augments SF-1-mediated increases in LH␤ gene promoter activity, and this action depends upon the presence of two Egr-1 DNA binding sites in the LH␤ gene promoter. Interestingly, in studies of the CYP11A gene promoter, SF-1 has been shown to interact with Sp1, a member of the zinc finger transcription family, which includes Egr-1 (33).
It must be noted that the observed Egr-1 effects on LH␤ promoter activity are physiologically relevant only if gonadotropes express endogenous Egr-1. Topilko et al. (14) recently demonstrated that Egr-1 is, in fact, expressed in gonadotropes based on co-localization of LH␤-subunit protein and X-gal staining, which is conferred by a lacZ transgene inserted 3Ј to the endogenous Egr-1(Krox-24) promoter (14). Interestingly, we have been unable to detect Egr-1 binding to the 3ЈEgr site using untreated nuclear extracts from the gonadotrope-derived cell lines ␣T3-1 and L␤T2 (data not shown).
In studies of nonreproductive systems, it has been observed that levels of Egr-1 mRNA and protein, although low under basal conditions, are rapidly and markedly induced by a number of stimuli, including serum, phorbol 12-myristate 13-acetate, nerve growth factor, and fibroblast growth factor (34). As is well known, gonadotropin gene expression is regulated by a wide variety of hormonal factors arising from the gonads, the hypothalamus, and the pituitary itself, factors that have been postulated to increase Egr-1 levels in the gonadotrope. Experiments are currently underway to define the physiologic stimuli that regulate Egr-1 levels in the gonadotrope and thereby increase expression of the LH␤ gene.
In summary, this study has defined two composite GSE/ Egr-1 elements in the proximal rat LH␤ gene promoter, and it suggests that SF-1 and Egr-1 can act both alone and synergistically to increase expression of the LH␤ gene by the gonadotrope.