A steroidogenic factor-1 binding site is required for activity of the luteinizing hormone beta subunit promoter in gonadotropes of transgenic mice.

Analysis of luteinizing hormone (LH) beta subunit promoters from a broad range of species including teleosts and humans revealed strict conservation of a sequence homologous to the steroidogenic factor-1 (SF-1) regulatory element of cytochrome P-450 steroid hydroxylase genes. Interaction between SF-1 and this putative response element in the bovine LH beta promoter was confirmed by electrophoretic mobility shift assays. Furthermore, cotransfection of alpha T3-1 cells with an expression vector encoding SF-1 induced binding site-dependent transcription from the bovine LH beta promoter. Physiological significance of the LH beta SF-1 consensus sequence was established using transgenic mice containing either the wild type bovine promoter or a promoter with a site-specific mutation of this site. Mutation of the SF-1 binding site nearly eliminated promoter activity, and the mutant transgene remained inactive following induction of gonadotropin-releasing hormone accomplished by castrating male and female mice. Thus, increases of gonadotropin-releasing hormone within a physiological range did not compensate for the loss of the SF-1 binding site. Together, these findings indicate that the SF-1 binding site is a key regulator of LH beta promoter activity in vivo and implicate SF-1 as at least one of the transcription factors that acts through this site.

Gonadotropes within the anterior pituitary are defined, in part, by their unique ability to synthesize and secrete luteinizing hormone (LH) 1 and follicle-stimulating hormone (FSH). These glycoprotein hormones contain an identical ␣ subunit that heterodimerizes with unique ␤ subunits (1). Expression of gonadotropin ␣ and ␤ subunit genes depends on secretion of gonadotropin-releasing hormone (GnRH) from the hypothalamus (2). GnRH responsiveness, in turn, requires that gonadotropes also express GnRH receptors (3). Thus, functional integrity and physiological control of mature gonadotropes relies on regulated expression of at least four genes: ␣, LH␤, FSH␤, and GnRH receptor.
While a complex array of regulatory elements controls transcription of the ␣ subunit gene in gonadotropes (4), the ele-ments regulating the genes encoding LH␤, FSH␤, or GnRH receptor remain undefined. This lag is due partly to weak activity of LH␤ and FSH␤ promoters when examined in cell culture models (5,6). Moreover, the promoter of the GnRH receptor gene has only recently been cloned (7). Nevertheless, there is a possibility that a single transcription factor may contribute to regulated expression of all four genes.
One such candidate was revealed by gene-targeting studies directed at the Fushi tarazu factor 1 (FTZ-F1) gene that encodes steroidogenic factor-1 (SF-1) and embryonal long terminal repeat-binding protein (ELP) (8 -10). SF-1 is an orphan member of the nuclear receptor superfamily initially identified as a transcription factor that controls expression of the cytochrome P-450 steroid hydroxylase genes in gonads and adrenal cortex (8). Targeted disruption of FTZ-F1 results in an absence of the adrenal cortex (9), ovary (9), testes (9), ventromedial hypothalamic nucleus (11), and gonadotropes (10). This result was later attributed solely to the loss of the SF-1 encoding portion of the FTZ-F1 gene (12).
Although all members of the hypothalamic/pituitary/gonadal axis are influenced by removal of the SF-1 gene, the mechanism responsible for gonadotrope deficiency remains unclear. This could be explained by a direct effect of SF-1 on gonadotrope development. Alternatively, if absence of the VMH attenuates secretion of GnRH, then gonadotropes would fail to develop (11). A direct gonadotrope site of action of SF-1 is suggested by studies with the ␣ subunit gene. All mammalian ␣ subunit promoter regions examined to date contain a gonadotropespecific element (GSE) (13) that has strong homology to SF-1 binding sites in steroid hydroxylase genes. This element in the human ␣ promoter binds to SF-1 (14) and accounts for approximately 50% of its activity when analyzed after transfection in the gonadotrope-lineage cell line, ␣T3-1 (4,13). In contrast to evidence suggesting a direct pituitary effect of SF-1, an indirect hypothalamic site of action is supported by the recovery of detectable gonadotropin gene expression in GnRH-treated, SF-1 "knock-out" mice (11). The latter finding implies that restoration of GnRH secretion may compensate for loss of SF-1 in gonadotropes. Thus, the exact role SF-1 plays in regulating expression of the genes that define the functional properties of gonadotropes remains to be determined.
Herein, we address the functional significance of the SF-1 binding site found within the proximal 776-bp promoter-regulatory region of the bovine LH␤ gene. This promoter directs high level expression of reporter genes specifically to gonadotropes in transgenic mice, is fully penetrant, and is appropriately regulated by GnRH and gonadal steroids (5).

EXPERIMENTAL PROCEDURES
Materials-Acetyl coenzyme A and the GnRH antagonist, antide, were from Sigma. Radionucleotides were obtained from DuPont NEN. DNA-modifying enzymes were from either Boehringer Mannheim or Life Technologies, Inc., and Sequenase version 2.0 was from U. S. * This work was supported by United States Department of Agriculture Grant 94-37203-0718. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Biochemical Corp. Antiserum directed against mouse SF-1 (8) was the generous gift of Keith L. Parker.
Cell Culture-Transient transfection assays of ␣T3-1 cells were performed essentially as described (4), except activation vectors (CMVGH or CMVSF-1) were also included at a concentration of 60 ng/35-mm well. All wells included an RSV␤-gal vector (0.42 g/well) to control for transfection efficiency. Cells were harvested 48 h post-transfection in 150 l of reporter lysis buffer (Promega). Luciferase and ␤-galactosidase assays were performed as reported with 15 l of extract (4).
Electrophoretic Mobility Shift Assays-Nuclear extracts from ␣T3-1 cells were prepared as described (4). Electrophoretic mobility shift assays (EMSA) were performed with the following double-stranded oligodeoxynucleotides: wild type GSE, 5Ј-CCTCTCCCTGACCTTGTCT-GCCTCT-3Ј; GSE, 5Ј-CCTCTCCCGCCATGGTTCTGCCTCT-3Ј, with the mutation highlighted. Nuclear extract (8.3 mg) was incubated with unlabeled oligodeoxynucleotide competitors in a final volume of 20 l of binding buffer (12.5 mM HEPES (pH 7.8), 25 mM KCl, 10% glycerol, 0.1 mM EDTA, 0.5 mM dithiothreitol, 1.5 g of poly(dI-dC), 0.2 g of singlestranded DNA, and 0.2 g of Escherichia coli DNA). Following 5 min at 4°C, 25 fmol of end-labeled probe was added for an additional 15 min at room temperature; 2 l of rabbit anti-mouse SF-1 or nonimmune serum were then added. After 15 min, the samples were placed on ice and run on 5% polyacrylamide gels at 4°C.
Transgenic Mice-Mice containing the wild type bLH␤ promoter linked to CAT have been characterized previously (5). Liberation of the (Ϫ776/ϩ10)GSEbLH␤CAT transgene fragment was accomplished with SalI/BamHI. Transgenic mouse production, identification, and characterization of tissue-specific expression was done as reported (5). CAT assays were performed as described with 10 g of protein for 1 h (5). Each assay included control tissues from mice containing the wild type promoter.
The intact/castrate/antide treatment paradigm was accomplished as follows. Male and female mice (6 -11 weeks old) were divided into three groups for each sex. One group remained intact, received injections of antide vehicle (20% propylene glycol in normal saline) every 48 h, and animals were housed individually. The other two groups were gonadectomized under avertin anesthesia and treated with vehicle or 60 g of antide every 48 h for 10 days (5). On day 10, mice were killed and pituitaries and ventricular blood collected. CAT assays of 17.5 h duration were performed as above. All animal studies were approved by the Case Western Reserve University Institutional Animal Care and Use Committee.

RESULTS AND DISCUSSION
A Consensus SF-1 Binding Site Resides within the 5Ј Flanking Regions of the Genes Encoding ␣, LH␤, and GnRH Receptor-Sequence analysis of LH␤ promoters from multiple species including teleosts and humans revealed conservation of a core sequence highly related to a canonical SF-1 binding site (Fig. 1). In some species, including bovine, the SF-1 homolog was identical to the GSE found in the human ␣ subunit gene (13). This homolog was also present in the GnRH receptor gene, suggesting that the element may be functionally significant for genes uniquely expressed in gonadotropes. Although the GSE in the human ␣ subunit gene binds SF-1 and thus is not gonadotrope-specific, the identity between this sequence and that in the LH␤ promoter led us to retain the GSE nomenclature.
The Bovine LH␤ GSE Binds SF-1-EMSA were performed with nuclear extracts from SF-1-containing ␣T3-1 cells (14) to determine whether SF-1 binds the bovine LH␤ GSE. Three complexes were detected with the wild type probe (Fig. 2). Each was eliminated by inclusion of 100-fold molar excess of homologous competitor, but not with 500-fold molar excess of a competitor containing an 8-bp mutation disrupting the SF-1 consensus (GSE). Incubation with SF-1 antisera resulted in loss of the highest mobility band (indicated by the arrow), suggesting the presence of SF-1. Nonimmune serum had no effect, indicating that disruption of the complex was specific to the SF-1 antisera. A radiolabeled GSE probe failed to detect the highest mobility band. Together, these data establish that the bovine GSE selectively binds SF-1.
Overexpression of SF-1 Activates the LH␤ Promoter in ␣T3-1 Cells-The endogenous LH␤ gene is silent in ␣T3-1 cells, even though they have GnRH receptors and express the endogenous ␣ subunit gene (21). The LH␤ promoter is also inactive in transfection assays with this cell line (5), which is possibly due to a lack, or diminished concentration, of critical transcription factors. If SF-1 actually regulates the LH␤ promoter, its overexpression might lead to activation of the promoter in these cells. Therefore, to assess functional significance of SF-1 binding to the putative LH␤ GSE, cotransfection assays were performed in ␣T3-1 cells with the 776-bp bLH␤ promoter linked to luciferase and a vector containing the CMV promoter directing expression of SF-1. Duplicate cotransfections with a CMVgrowth hormone expression vector were used to establish specificity of an SF-1 response.
As expected, activities of the wild type LH␤ promoter (bLH␤) and a promoterless control were indistinguishable (data not shown). In contrast, overexpression of SF-1 resulted in ϳ5-fold induction of bLH␤ promoter activity relative to that obtained in the presence of the GH control (Fig. 3). Greater induction (ϳ8-fold; data not shown) was observed following addition of a constitutively active form of SF-1 (22). Although the bLH␤ promoter containing an 8-bp GSE mutation was 2-fold more  (15) genes, with mismatches in lower case. The SF-1 binding site (GSE) from the human ␣ (h␣) subunit gene has previously been characterized (14). Numbers indicate the position of the GSE homolog relative to the transcription start site (ϩ1). The LH␤ sequence was analyzed from: salmon (sGTHII␤, Ref. 16 active than the wild type promoter in the presence of GH, its response to SF-1 was negligible and equivalent to that observed with the heterologous SV40 promoter. Thus, much of the stimulatory effect of SF-1 is mediated through the GSE. While SF-1 stimulated LH␤ promoter activity only 5-fold, this was sufficient to establish a functional and statistically significant correlation between binding of SF-1 and transcriptional activation. A site-dependent effect of full-length SF-1 on other promoters has been difficult to establish. For example, Shen et al. (22) only observed activation of the Mü llerian inhibiting substance promoter with a truncated form of SF-1 devoid of the putative ligand binding domain. Similarly, the human ␣ subunit promoter with an intact SF-1 binding site is refractory to overexpression of both full-length and truncated SF-1 (data not shown), and ablation of the SF-1 binding site in the human ␣ subunit promoter reduces basal activity by only 50% (4,13). This suggests that transcription effects of SF-1 may be dependent on promoter context. The level of SF-1-dependent activation observed with the LH␤ promoter is also significant because overexpression of other proteins potentially involved in the GnRH signal transduction pathway, including constitutively active forms of Ras (23) or G␣ q (24), had no effect (data not shown). Thus, these transfection assays are the first to identify both a discrete promoter element and a transcription factor that regulate activity of the LH␤ promoter.
The GSE Is Essential for LH␤ Promoter Activity in Trans-genic Mice-We previously reported that the 776-bp bovine LH␤ promoter conferred pituitary-specific expression to a CAT reporter gene in transgenic mice (5). Activity of this promoter in mice was much stronger than that observed in transfection assays, being approximately 160-fold more active than that of the 1500-bp human ␣ subunit promoter (5). Moreover, activity was completely penetrant (observed in all mice), occurred solely in gonadotropes, and was appropriately regulated by hormones, including GnRH, 17␤-estradiol, and testosterone, which effect expression of the endogenous LH␤ gene. Transgenic mice were made with the 776-bp LH␤ promoter, containing the same GSE mutation used in the EMSA and transfection assays, linked to CAT. Integrity of the transgene in seven founders was confirmed by Southern blot analysis (data not shown), and lines were derived from five to obviate integration site effects. Activity of the GSE-LH␤ promoter in pituitary was reduced at least 10-fold when compared to activity in two lines of mice containing the wild type promoter (Fig. 4). No ectopic expression was observed in nine other tissues (data not shown).
Activity of the wild type promoter tended to be higher in females than in males (Fig. 4). This trend continued with the mutant promoter, such that the occasional mouse having detectable CAT was always female. Thus, while mutation of the GSE has a substantial effect on activity of the LH␤ promoter in both sexes, it may be even greater in male mice. From these results, we conclude that the SF-1 binding site (GSE) is required for full bLH␤ promoter activity in pituitaries of transgenic mice.
GnRH Cannot Compensate for Loss of the GSE-Qualitative rescue of LH␤ mRNA results from GnRH treatment of SF-1deficient mice maintained with exogenous adrenal steroids (11). This suggests that the loss of LH␤ gene expression in those mice is due to an effect on GnRH secretion rather than a direct effect of SF-1 in gonadotropes. To determine if an increase in GnRH could rescue activity of the mutant bLH␤ promoter, male and female transgenic mice were gonadectomized to promote a post-castration rise in GnRH that lies within a physiological range (2). An additional group were treated with the GnRH antagonist antide to confirm the efficacy of gonadectomy. To enhance detection of CAT, duration of the enzymatic assay was extended significantly from 1 to 17.5 h. This extended assay allowed detection of CAT activity in intact females containing the GSE-LH␤ promoter, although it was still 10-fold less than that observed with the wild type promoter (Fig. 5A). Activity of the mutant promoter in ovari- were assessed for CAT activity in pituitary. Each circle represents an individual mouse. All mice were at least one generation from the founder animals except the mouse from line 6, which was the founder. All assays were performed with 10 g of pituitary lysate for 1 h with CAT activity expressed as percent conversion/g of protein/h. Mouse pituitaries containing the wild type or GSE transgenes were assayed in parallel.
ectomized females was reduced to undetectable levels, suggesting that transgene was unresponsive to the expected postgonadectomy rise in GnRH. In contrast, serum LH levels did respond to gonadectomy and antide treatments, confirming their effect on GnRH secretion. Interestingly, CAT activity in ovariectomized GSE mice was reduced compared to the intact control. If the mutant promoter were simply unable to respond to GnRH, one would expect no change in expression. Thus, in the absence of a GnRH response, there may be an ovarian factor(s) that normally stimulates LH␤ promoter activity. This factor may have previously escaped detection due to the usually large positive impact of GnRH.
In contrast to females, intact male mice containing the mutant promoter lacked any detectable CAT activity (Fig. 5B). Castration also failed to induce expression, whereas an antidereversible increase in serum LH was observed. Thus, changes in GnRH were unable to compensate for loss of the GSE in the LH␤ promoter in either sex. This result contrasts with those reported by Ikeda et al. (11), where GnRH treatment of SF-1deficient mice resulted in an unquantified activation of the endogenous LH␤ gene. Several differences between the two studies render valid comparison difficult. Pharmacological doses of GnRH were used to treat the knock-out mice (11), whereas the increase in GnRH observed in our mice should more accurately approximate physiological levels. In addition, the extent of disruption of the reproductive axis in mice lacking SF-1 is unclear, as is its possible impact on gonadotropin gene responsiveness to GnRH. Finally, it is possible that additional transcription factors could compensate for SF-1 at the level of the GSE. Removal of the GSE itself, which occurred with the mutant LH␤CAT mice, would negate any potential compensatory mechanism. Precedence for such redundancy has been observed with transcription factors that are members of larger families, including MyoD (25) and cAMP response elementbinding protein (26). Insertion of the mutant transgene into the background of the SF-1-deficient mouse may resolve these differences.
In summary, the GSE within the bovine LH␤ subunit promoter binds SF-1 with high affinity, mediates activation of transcription by SF-1 in vitro and is considerably important for full activity in gonadotropes of transgenic mice. This is the first complete characterization of an element important for expression of the LH␤ subunit gene in gonadotropes and the first physiological demonstration of a functional requirement for an SF-1 binding site within any gene. Conservation of the SF-1 binding site in three of the genes that define a mature gonadotrope (␣, LH␤, and GnRH receptor) may provide a unifying mechanism for their coordinated onset and maintenance of expression.
FIG. 5. GnRH cannot compensate for loss of the GSE. Female (A) and male (B) mice from line 1 containing the bLH␤GSECAT transgene remained intact or were gonadectomized. The intact animals and a subset of gonadectomized mice were treated with vehicle, while another subset received antide. At the conclusion of the experiment, blood and pituitaries were collected for LH and CAT assays, respectively. All CAT assays were performed with 10 g of protein for 17.5 h. CAT activity is expressed as percent conversion/100 g of protein/h. Values are means Ϯ S.E. for 5-6 animals/group. The shaded region in the lower portion of the figure represents the mean activity of multiple liver samples included in this assay (1.14 Ϯ 0.14). Any values that approach this level are considered non-expressing. The activity of wild type mouse pituitaries was 49.8 Ϯ 3.0.