Pituitary Adenylate Cyclase-activating Polypeptide and Cyclic Adenosine 3′,5′-Monophosphate Stimulate the Promoter Activity of the Rat Gonadotropin-releasing Hormone Receptor Gene via a Bipartite Response Element in Gonadotrope-derived Cells*

Specific type I receptors for pituitary adenylate cyclase-activating polypeptide (PACAP) are present in gonadotrope cells of the anterior pituitary gland. By transient transfection of mouse gonadotrope-derived αT3-1 cells, which are direct targets for PACAP and express gonadotropin-releasing hormone receptor (GnRH-R), a marker of the gonadotrope lineage, we provide the first evidence that PACAP stimulates rat GnRH-R gene promoter activity. The EC50 of this stimulation is compatible with a mediation via activation of the cyclic AMP-dependent signaling pathway and, consistently, co-transfection of an expression vector expressing the protein kinase A inhibitor causes reduction in PACAP as well as cholera toxin-stimulated promoter activity. Deletion and mutational analyses indicate that PACAP activation necessitates a bipartite response element that consists of a first region (−272/−237) termed PACAP response element (PARE) I that includes a steroidogenic factor-1 (SF-1)-binding site and a second region (−136/−101) referred to as PARE II that contains an imperfect cyclic AMP response element. Gel shift experiments indicate the specific binding of the SF-1 and a potential SF-1-interacting factor to PARE I while a protein immunologically related to the cyclic AMP response element-binding protein interacts with PARE II. These findings suggest that PACAP might regulate the GnRH-R gene at the transcriptional level, providing novel insights into the regulation of pituitary-specific genes by hypothalamic hypophysiotropic signals.

The hypothalamic neuropeptide gonadotropin-releasing hormone stimulates the synthesis and release of gonadotropins, luteinizing hormone, and follicle-stimulating hormone, acting through a specific membrane receptor belonging to the family of heptahelical G protein-coupled receptors. The pituitary gonadotropins then enter the systemic circulation to regulate gonadal function, including steroid hormone synthesis and gametogenesis. The responsiveness of gonadotrope cells to GnRH 1 is dependent on the number of cell surface GnRH-R, and changes in the number of these receptors often correlate with changes in the level of receptor mRNA (1)(2)(3). To investigate this issue at the transcriptional level, namely the tissuespecific and regulated expression of the GnRH-R gene, the promoter regions of the mouse, rat, human, and ovine genes have been isolated and characterized (4 -8).
Transient transfection assays in the mouse gonadotropederived ␣T3-1 cell line have provided evidence that gonadotrope-specific activity of the mouse promoter is mediated by a tripartite basal enhancer that includes an SF-1-binding site, a consensus AP-1 element, and a novel element termed GnRH-R-activating sequence (GRAS) (9). Similarly, the interaction of SF-1 with a gonadotrope-specific element motif in the human gene has been shown to mediate gonadotrope-specific expression (10). Regarding the rat promoter, we and others have reported that full gonadotrope specific activity required a distal regulatory domain in addition to the SF-1 and AP-1 elements present in the proximal domain (7,8,11). We have found that the basal expression of the GnRH-R gene in the gonadotropederived ␣T3-1 cell line is highly dependent on a distal enhancer that is active in the context of the GnRH-R gene promoter only and therefore is termed GnRH receptor-specific enhancer. We have shown that GnRH receptor-specific enhancer activity (Ϫ1135/Ϫ753) was mediated through a functional interaction with a proximal region (Ϫ275/Ϫ226) that included the SF-1binding site (11).
In addition, the availability of these promoters has allowed the study of the hormonal regulation of the GnRH-R gene and led to the conclusion that some of the elements involved in constitutive expression were also implicated in hormonal regulation. The GRAS element was indeed demonstrated to be involved in the autocrine/paracrine stimulation of the mouse GnRH-R promoter by activin (12). Likewise, two reports based on deletion and/or mutational analysis and functional transfection studies, as well as electrophoretic mobility shift assays, have revealed the involvement of the AP-1 element in the mechanism of the homologous regulation of the mouse GnRH-R gene by GnRH in ␣T3-1 cells (13,14). An auxiliary element localized 10 bp upstream was found to be necessary for optimal activation by GnRH (13). These elements interact with members of the Fos/Jun heterodimer superfamily in agreement with PKC dependence of GnRH-R response to GnRH (13,14) with the probable implication of a mitogen-activated protein kinase pathway in this regulation (14). In contrast, the homologous, PKC-mediated desensitization of the human GnRH-R promoter activity in ␣T3-1 cells has been shown to involve a different AP-1 motif located in the distal part of the promoter (15). Much less is known regarding the regulation of the GnRH-R promoter activity by the PKA-dependent pathway. Using transient transfection in the somatolactotrope GGH3 cell line stably expressing the GnRH-R gene, the mouse GnRH-R gene was found to be responsive to cAMP (16). Similarly, treatment of ␣T3-1 cells with forskolin or a cAMP analog significantly increased luciferase activity of the transfected rat GnRH-R promoter (7). These data led us to examine the regulation of the rat GnRH-R gene transcription by cAMP as well as by a physiological activator of the cAMP-dependent signaling pathway, the pituitary adenylate cyclase-activating polypeptide (PACAP).
PACAP, a member of the vasoactive intestinal polypeptide/ secretin/glucagon family of peptides, was isolated from ovine hypothalamic extracts based upon its adenylate cyclase stimulating activity in rat pituitary cells (17). The major form of PACAP is a C-terminal amidated 38-amino acid polypeptide, but a shorter form, PACAP27, corresponding to the N-terminal 27 residues of PACAP38, is also found in the hypothalamus. Two major PACAP receptors have been identified: 1) PAC 1 receptors highly specific for PACAP that activate not only adenylate cyclase but also phospholipase C. These receptors have been shown to be expressed in the anterior pituitary, adrenal medulla, hypothalamus, testis, and ␣T3-1 pituitary cell line (18). 2) VPAC 1 and VPAC 2 receptors, which bind both vasoactive intestinal polypeptide and PACAP but activate almost exclusively adenylate cyclase (19 -21) and have been found in lung, liver, prostate gland, and seminal vesicles. At the pituitary level, PACAP stimulated the release of luteinizing hormone in vivo (22), whereas in vitro in cultured pituitary cells, it weakly stimulated luteinizing hormone and folliclestimulating hormone release. Interestingly, it notably enhanced GnRH-induced gonadotropin secretion, thus suggesting intriguing possibilities for this peptide in regulating gonadotropin secretion and reproductive function (23). Consistently, PACAP has been shown not only to stimulate glycoprotein hormone ␣-subunit synthesis and release (24,25) but also to increase ␣-subunit mRNA concentrations in primary rat pituitary cells and ␣T3-1 cells (26). Recently, Burrin and collaborators (27) have demonstrated using transfection assays in ␣T3-1 cells that PACAP regulated the expression of the human ␣-subunit gene at the transcriptional level.
In the present study, we examined the effects of PACAP and cAMP on GnRH-R promoter activity using ␣T3-1 cells. We employed deletion, mutation, and heterologous constructs of the rat GnRH-R promoter to delineate DNA sequences responsive to PACAP and cAMP. Gel shift assays were performed to assess the binding capacity of the putative elements with nuclear factors. We show that both PACAP and cAMP responsiveness of the GnRH-R gene promoter in ␣T3-1 cells is supported by two distinct proximal regions, one of which includes the SF-1-binding site localized at Ϫ245/Ϫ237 and is described previously as crucial for tissue-specific expression and the other, an imperfect cAMP response element present at position Ϫ110/Ϫ103.
A series of block replacement mutations at 8-bp intervals in the region from Ϫ260 to Ϫ221 within the pCAT0.27 GnRH-R construct (Ϫ272/Ϫ32) was generated by PCR amplification using a series of sense/antisense primers (overlapping over 14 bp) designed to place a PstI restriction site with T and C flanking bases at the 5Ј and 3Ј ends, respectively (TCTGCAGC). Overlapping fragments were generated from pCAT0.43 GnRH-R (from Ϫ433 to Ϫ32) as a template in separate PCR reactions using the mutated sense primer and antisense primer Ϫ32 Sal or the mutated antisense primer and sense primer Ϫ433 Hind. The amplified products were combined and submitted to a second round of PCR using the Ϫ272 Hind and Ϫ32 Sal primers, and the resulting products were digested with HindIII and SalI, gel-purified, and inserted into the pCAT Basic vector (Promega, Lyon, France) digested with the same enzymes. Because of a superstimulated basal activity of construct Ϫ272 Mut D, the PstI site was replaced by a KpnI site (TGGTACCC).
To subclone the artificial promoter constructs upstream of the luciferase reporter gene (Promega), the multiple cloning site of the pGL3-Basic vector was altered to provide compatible restriction sites in the appropriate orientation (11). A minimal prolactin (PRL) promoter and a single 50-bp module containing the SF-1 element were synthesized as described previously (11). The 50-bp module was introduced into the modified pGL3-Basic containing upstream either the minimal PRL promoter or a rat GnRH-R promoter fragment subcloned in place of the minimal PRL promoter (see below). The Ϫ136/Ϫ32 and the Ϫ101/Ϫ32 proximal regions of the GnRH-R promoter were generated using selected sense primers Ϫ136 Bst and Ϫ101 Bst, respectively, and antisense primer Ϫ32 Sal. The Ϫ56/Ϫ32 region was directly obtained by self-annealing two oligonucleotides with an overlap of 27 bp and thus reconstituting a double-stranded DNA, which included BstEII and SalI half-sites at each end for cloning into the BstEII/SalI sites in the modified pGL3-Basic vector. By serially replacing 8-bp segments from position Ϫ260 to Ϫ237 in the 50-bp module with the PstI restriction site flanked by T and C at the 5Ј and 3Ј ends, respectively (TCTGCAGC), three separate mutants were created and ligated upstream of the Ϫ136/ Ϫ32 promoter region in place of the 50-bp module in the modified pGL3-Basic. The series of block replacement mutations eco1 to eco4 scanning the sequence from Ϫ136 to Ϫ101 in the Ϫ136/Ϫ32 promoter region were also generated by PCR amplification using a series of sense/antisense overlapping primers designed to place an EcoRI restriction site flanked by a GC at the 3Ј end (GAATTCGC). All mutant plasmids were identified by restriction digest of midi-prep DNA and ultimately verified by nucleotide sequencing.
The PKI expression vector was prepared by subcloning the rat cDNA encoding PKI into the XbaI/EcoRI site of the pcDNA3 (Invitrogen, Leek, The Netherlands) upstream of the CMV promoter. This cDNA was initially cloned from rat pituitary by reverse transcription-PCR and ligated into pUC18 (Amersham Pharmacia Biotech) in our laboratory, 2 whereas an identical rat PKI sequence was simultaneously isolated from the brain by another group (28).
Cell Culture and Transient Transfection-The mouse gonadotrope ␣T3-1 cells were maintained in monolayer cultures in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate (Sigma) at 37°C in humidified 5% CO 2 , 95% air (29). Transfection experiments with CAT or luciferase reporter constructs were carried out using the LipofectAMINE Plus reagent-mediated procedure (Life Technologies, Inc.) as described previously (11). Following a 6-h trans-2 G. Garrel and R. Counis, unpublished data. fection, cells were subjected to appropriate treatment in Dulbecco's modified Eagle's medium supplemented with 2% fetal bovine serum, 10 units/ml penicillin, and 10 g/ml streptomycin sulfate. Thereafter, medium was aspirated, and cells were processed as described previously for ␤-galactosidase and CAT assays (8).
Preparation of Nuclear Extracts and Gel Mobility Shift Assays-The cells were seeded at 3 ϫ 10 6 cells/100-mm tissue culture dish in triplicate and cultured for 24 h in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate. The culture medium was then replaced by OptiMEM medium, and the cells were cultured for an additional 6 h. The serum-free medium was finally replaced by Dulbecco's modified Eagle's medium supplemented with 2% fetal bovine serum, 10 units/ml penicillin, and 10 g/ml streptomycin sulfate, and cells were incubated in the presence or the absence of 20 nM PACAP for 16 h. Thereafter, cells were harvested, and nuclear extracts were prepared by the method of Andrews and Faller (30).
Synthetic double-stranded oligonucleotides, designed to contain 5Ј protruding ends, were labeled (5 pmol) by filling in the recessed 3Ј termini with Klenow fragment from Escherichia coli DNA polymerase I and 50 Ci of [␣-32 P]dCTP (3000 Ci/mmol; Amersham Pharmacia Biotech). The CRE/PARE II and CREB consensus probes were labeled with 50 Ci of [␥-32 P]ATP (3000 Ci/mmol; Amersham Pharmacia Biotech) using a T4 polynucleotide kinase. All probes were then purified on a Sephadex G50 fine column. For binding reactions, nuclear extracts (9 g) and poly(dI-dC) (1 g) were incubated in binding buffer (20 mM HEPES, pH 7.9, 60 mM KCl, 1 mM EDTA, 300 g/ml bovine serum albumin, and 12% (v/v) glycerol) for 15 min at 4°C. Thereafter, 40,000 cpm DNA probe (approximately 10 fmol) was added with or without an excess of unlabeled competitor, and the incubation was continued for 30 min at 20°C. In antibody abrogation gel shift assays, nuclear extracts were incubated with either a rabbit polyclonal antibody directed against the DNA-binding domain of murine SF-1 (Upstate Biotechnology, Lake Placid, NY), a rabbit polyclonal anti-CREB antibody (Santa Cruz Biotechnology, Inc.), or an equal concentration of mouse IgG for 1 h at 4°C prior to the addition of radiolabeled probe. Free probe was separated from bound probe by electrophoresis in 5% nondenaturing polyacrylamide gels that were run at 120 V for 2 h in 1ϫ Tris-borate-EDTA buffer. Gels were then transferred to blotting paper, dried, and subjected to autoradiography for 24 -48 h with or without intensifying screens.
Statistical Analysis-The data were analyzed by one-way analysis of variance. If the F-test was significant, then the means were compared using Tukey-Kramer's method of multiple comparisons. For Figs. 1C, 5, and 6, the data were logarithmically transformed before proceeding to analysis of variance.

PACAP Stimulates
GnRH-R Promoter Activity through the PKA-signaling Pathway-To assess the potential implication of PACAP in the regulation of the rat GnRH-R gene, immortalized pituitary gonadotrope cells (␣T3-1 cells) were transiently transfected with GnRH-R promoter CAT fusion constructs then treated with 20 nM PACAP for 16 h (Fig. 1A). The effect of PACAP was compared with that exerted by activators of the PKA-and PKC-dependent pathways, cholera toxin (1 nM) and TPA (25 nM), respectively (Fig. 1B). Both PACAP and cholera toxin stimulated CAT expression of the Ϫ1257/Ϫ32 construct equivalently, with 2.5 Ϯ 0.5-fold and 2.3 Ϯ 0.1-fold increases over untreated cells, respectively. As a consequence of the presence of the enhancer in the distal part of the promoter, deletion of the sequence extending from Ϫ1257 to Ϫ515 caused a decrease in basal CAT activity (see the Introduction and Ref. 11). Nevertheless, the resulting construct Ϫ515/Ϫ32 still displayed an increased response to both PACAP and cholera toxin. A further deletion from Ϫ515 to Ϫ180 abrogated both PACAP and cholera toxin response, suggesting that response elements for PACAP and cholera toxin were localized within a proximal The GnRH receptor-specific enhancer (GnRH-R specific enhancer), as well as the GRAS, AP-1, and SF-1 elements are indicated by black boxes at their corresponding locations within the rat promoter sequence. The major transcription start sites located at positions Ϫ110/Ϫ107 and Ϫ99/Ϫ97 (relative to the ATG codon) are indicated by bent arrows. GnSE, GnRH receptor-specific enhancer. B, ␣T3-1 cells were transfected for 6 h with various GnRH-R promoter CAT constructs and then treated for 16 h with either 20 nM PACAP, 1 nM cholera toxin (Ctx), 25 nM TPA, or a combination of cholera toxin and TPA. CAT activity was calculated as CAT activity/␤-galactosidase activity and then normalized as fold induction over that of the promoterless pCAT Basic vector. The values represent the means Ϯ S.D. *, p Ͻ 0.001 compared with untreated cells. C, cells were transfected with the 500-bp 5Ј-flanking region of the rat GnRH-R gene (Ϫ515/Ϫ32) or the promoterless construct as control, followed by treatment with increasing concentrations of PACAP (2 ϫ 10 Ϫ16 , 2 ϫ 10 Ϫ12 , 2 ϫ 10 Ϫ11 , 2 ϫ 10 Ϫ10 , 2 ϫ 10 Ϫ9 , 2 ϫ 10 Ϫ8 , and 2 ϫ 10 Ϫ7 M). Measurements, which were normalized to the CMV promoter-containing vector pcDNA3, were expressed as CAT/␤-galactosidase. D, cells were co-transfected with the 500-bp 5Ј-flanking region and increasing amounts of pcDNA3PKI, an expression vector expressing PKI driven by the CMV promoter, plus pcDNA3, used as a control, followed by treatment with 20 nM PACAP or 1 nM cholera toxin (Ctx). The measurements are expressed as CAT/␤-galactosidase. All results are the means Ϯ S.D. of duplicate samples in at least three independent transfection experiments.
promoter region extending from Ϫ515 to Ϫ180. Additional experiments using similar GnRH-R promoter sequences but fused to the luciferase gene reporter gave equivalent results, even following treatments with the adenylate cyclase activator, forskolin, or the permeant cAMP analog, 8-bromo-cAMP. In response to 10 M forskolin, the luciferase constructs containing either the full-length (Ϫ1135/Ϫ32) or the enhancerless (Ϫ433/Ϫ32) promoter were stimulated 2.8 Ϯ 0.3-and 3.5 Ϯ 0.3-fold, respectively, whereas the construct containing the shortest GnRH-R promoter (Ϫ180/Ϫ32) remained unaffected (data not illustrated). Likewise, 2 mM 8-bromo-cAMP significantly stimulated the luciferase activity of both the full-length and the enhancerless constructs, by 2.0 Ϯ 0.2-and 1.9 Ϯ 0.2-fold, respectively. In contrast, unlike PACAP and activators of the PKA-dependent signaling pathway, TPA had no effect on the expression of any construct (Fig. 1B). These data are in accordance with a previous report showing that forskolin and cAMP analogs, but not TPA, increased the activity of the rat promoter transfected into ␣T3-1 (7). In addition, these results suggested the involvement of a PKA-mediated pathway for the activation of the GnRH-R promoter by PACAP.
To evaluate the specificity of the response of the rat GnRH-R promoter, increasing doses of PACAP ranging from 2 ϫ 10 Ϫ16 to 2 ϫ 10 Ϫ7 M were tested in ␣T3-1 cell cultures transfected with the Ϫ515 GnRH-R CAT construct (pCAT0.5 GnRH-R) or with the promoterless pCAT Basic vector. As shown in Fig. 1C, PACAP induced a concentration-dependent increase in GnRH-R CAT activity, with a maximal response observed at 2 nM. As expected, the activity of the promoterless vector was not affected by PACAP. More importantly, the EC 50 value for stimulation of GnRH-R promoter activity by PACAP (0.2 Ϯ 0.06 nM) was in closer agreement with signaling through the cAMP/PKA cascade (3 nM) than through stimulation of the inositol-phosphate turnover (20 nM) in ␣T3-1 cells (18). Taken together these findings implied that the stimulatory effect of PACAP on GnRH-R promoter was most probably mediated through coupling of PAC 1 receptors to the PKA-dependent signaling pathway.
To test this hypothesis, ␣T3-1 cells were transfected with pCAT0.5 GnRH-R and co-transfected with variable amounts of a vector expressing the rat PKI cDNA under the control of the CMV promoter (pcDNA3PKI). A vector containing only the CMV promoter (pcDNA3) was used as a control (Fig. 1D). Co-transfected pcDNA3PKI significantly decreased (p Ͻ 0.01) basal GnRH-R CAT activity by about 50%. Furthermore, 0.5 g of pcDNA3PKI markedly reduced (p Ͻ 0.001) both cholera toxin-and PACAP-activated GnRH-R CAT activity by 76.5 and 61.4%, respectively, whereas 1 g led to a maximal inhibition of 84 and 70.5%, respectively. Again, these findings were consistent with the implication of the cAMP-dependent signaling pathway in PACAP-stimulated GnRH-R promoter activity.
Effect of 5Ј-Deletions of the GnRH-R Gene Promoter on PACAP-and Cholera Toxin-stimulated CAT Activity-To localize the putative response elements for both PACAP and cAMP, serial 5Ј-deletion mutants with 5Ј-termini located between Ϫ515 and Ϫ180 within the GnRH-R gene promoter were designed. Transient transfection experiments were performed with ␣T3-1 cells, which were then treated with either PACAP, cholera toxin, or vehicle as above. As shown in Fig. 2 (left panel), the Ϫ515 GnRH-R CAT construct elicited a 5.3 Ϯ 0.9fold increase in the basal activity over the promoterless vector, and deletions from Ϫ515 to either Ϫ433 or Ϫ381 did not significantly affect the basal activity (6.5 Ϯ 1.1-and 6.7 Ϯ 1.1-fold over promoterless vector, respectively). However, the basal GnRH-R promoter activity was significantly decreased after 5Ј-deletion from Ϫ381 to Ϫ316, which eliminated the AP-1binding site, in agreement with a previous report from our laboratory (11). Additional deletions from Ϫ297 to Ϫ247 similarly decreased constitutive expression, whereas a further deletion from Ϫ247 to Ϫ222 that included the SF-1 element was inefficient. The latter observation was surprising because the SF-1 element is crucial for constitutive activity of the rat GnRH-R promoter. Block replacement mutagenesis of the SF-1 element in the context of the full-length (Ϫ1135/Ϫ32) or proximal (Ϫ450/Ϫ32) promoter strongly decreased basal expression (11). Because deletion of the SF-1 element in the present experiment was inefficient, this suggested that sequences upstream of Ϫ247 could be necessary for SF-1 activity.
Consistent with mediation of PACAP action through the PKA-dependent pathway, an equivalent pattern of expression was obtained in PACAP-as well as cholera toxin-treated cells (Fig. 2, right panel). Transfections using the Ϫ515 GnRH-R CAT construct resulted in a 2.6 Ϯ 0.4-or 2.8 Ϯ 0.3-fold increase in CAT activity in response to PACAP or cholera toxin stimu- lation, respectively, as compared with untreated cells. The response to either agent was unchanged following deletion of sequences extending from Ϫ515 to Ϫ297 but slightly increased with further deletions between Ϫ297 and Ϫ272 (ϳ3-4-fold in response to PACAP or cholera toxin). Finally, 5Ј-deletion from Ϫ272 to Ϫ222 caused a significant reduction (1.5-fold) in CAT activity in response to both agents, suggesting that response elements were co-localized in the Ϫ272/Ϫ222 region of the GnRH-R gene promoter. Therefore, the GRAS and AP-1 motifs that are located upstream of position Ϫ272 are unlikely to be implicated in transcriptional activation by PACAP or cholera toxin. In contrast, the SF-1-binding site at position Ϫ245/Ϫ237 as well as other potential elements in the Ϫ272/Ϫ222 region might be involved in this regulation.

PACAP Responsiveness of the GnRH-R Gene Is Dependent on an SF-1-binding Site and Additional Elements in the 16 bp
Immediately Upstream-To precisely delineate the response elements located in the Ϫ272/Ϫ222 region, we used block replacement mutagenesis to generate a series of five mutations spanning from Ϫ260 to Ϫ220 in the context of the Ϫ272 GnRH-R CAT construct. Block replacement mutations consisted of substituting the wild-type sequence with a 8-bp sequence including a restriction site (PstI or KpnI; as detailed under "Experimental Procedures"). The mutated promoter sequences were placed upstream of the CAT reporter gene in pCAT Basic as described previously and assayed for activity by transient expression in ␣T3-1 cells (Fig. 3). PACAP-stimulated activity was significantly attenuated for constructs that contained the A, B, and C mutations. The most effective mutation extended from Ϫ252 to Ϫ245 (MutB) and led to a 90% decrease (p Ͻ 0.001) in PACAP stimulation, as compared with that of the Ϫ280 wild-type promoter fragment. The mutation extending from Ϫ260 to Ϫ253 (MutA) or from Ϫ244 to Ϫ237 that included the SF-1-binding site (MutC), reduced (p Ͻ 0.001) PACAPactivated transcription by a similar extent of 62 and 64%, respectively. In contrast, E and D mutations led to insignificant changes (p Ͼ 0.05) in stimulated transcription. Therefore, we concluded that the sequence corresponding to the putative SF-1-binding site and extending over 16 bp immediately upstream (from Ϫ260 to Ϫ237) was required for PACAP stimulation of GnRH-R gene promoter activity, and it was thus designated PACAP response element I (PARE I). In this region, the Ϫ252/ Ϫ245 sequence contiguous to the SF-1-binding site appeared to be crucial for PACAP regulation because mutation B led to a maximal repression of PACAP-induced stimulation.
In Addition to SF-1, Other Protein Factor(s) Interact(s) Specifically with the PARE I Region-To analyze the binding capacity of the PARE I region (Ϫ260/Ϫ237), nuclear extracts from ␣T3-1 cells were prepared for use in gel retardation assays. Radiolabeled synthetic oligonucleotides corresponding to sequence Ϫ264 to Ϫ231, either intact (wild probe), or mutated at position Ϫ260/Ϫ253 (MutA probe), Ϫ252/Ϫ245 (MutB probe), or Ϫ244/Ϫ237 (MutC probe or SF-1 mutant) were incubated with ␣T3-1 nuclear extracts and tested for protein-DNA interactions. PACAP stimulation (20 nM for 16 h) of ␣T3-1 cells prior to preparation of nuclear extract gave results equivalent to those obtained with nonstimulated cells. As shown in Fig. 4, three DNA-protein complexes were formed when wild probe was used as a labeled oligonucleotide (lane 2, complexes I, II, and III), and all complexes were competed by the addition of an excess of homologous competitor (lanes 3-5). However, complex III appeared to be of relatively weak affinity because a marked amount of this complex was still detected in the presence of a 1000-fold molar excess of unlabeled wild probe (lane 5). In addition, neither mutation A, B, nor C seemed to affect the formation of complex III (lanes 17, 12, and 7, respectively). Taken together these data suggested that complex III was DNA sequence-independent and therefore nonspecific. More impor- Competitions for binding were conducted with increasing concentrations of the indicated unlabeled probes at a 10 -1000-fold molar excess. In antibody abrogation experiments (lanes 23-28), 9 g of nuclear proteins were incubated with an affinity-purified rabbit polyclonal antibody directed against the DNA-binding domain of murine SF-1 or an equal concentration of mouse IgG prior to the addition of radiolabeled probes consisting of the wild or the MutC probe. All binding reactions were subjected to electrophoresis through nondenaturing 5% polyacrylamide gels as described under "Experimental Procedures." tantly, radiolabeled MutC probe failed to form complex II (lane 7), indicating that the sequence corresponding to the SF-1binding site (Ϫ245/Ϫ237) was involved in the formation of this complex, which was consistent with our previous study (11). Moreover, the abundance of complex I formed with the MutC probe was significantly diminished as compared with that formed with the wild probe (lane 7), suggesting that complex I was also affected by mutation C, yet to a lesser extent than complex II. In contrast, MutA and MutB probes could form complex II but were incapable of binding protein(s) of complex I (lanes 17 and 12, respectively), which actually correlated with their inability to compete for binding with the radiolabeled MutC probe (data not shown). Thus, these data suggested that the sequences covered by mutations A and B were necessary for complex I formation, whereas the sequence covered by mutation C was required for both complex II formation and high affinity binding of the factor(s) involved in complex I. To validate further the identity of the protein(s) that interacted with the PARE I region, antibody abrogation gel shift experiments were conducted using a rabbit polyclonal antibody directed against the DNA-binding domain of the murine SF-1 protein (Fig. 4, lanes 26 -28). As is apparent from the figure, addition of the anti-SF-1 antibody to the binding reaction abrogated com-plex II formation in a dose-dependent manner, indicating that a factor immunologically related to SF-1 interacted with the SF-1 response element. Interestingly, the intensity of complex I was attenuated in the presence of increasing concentrations of anti-SF-1 antibody, which suggested that SF-1 also favored complex I formation. Collectively, these data point out the ability of nuclear protein(s) to bind the 16-bp sequence adjacent to the SF-1 element, in addition to SF-1 binding to its own site, which is in concordance with the results obtained with the transfection assays.
The Association of the PARE I Region with the Ϫ136/Ϫ32 Region Is Necessary to Confer Full Responsiveness to PACAP-To determine whether PARE I was not only necessary but also sufficient to mediate full PACAP response, a PCR-generated 50-bp module (Ϫ275/Ϫ226) encompassing the PARE I region was placed under the control of the minimal PRL promoter, and the response to PACAP stimulation was measured. Transfection with this construction demonstrated only a 2.3 Ϯ 0.2-fold increase (p Ͻ 0.001) in luciferase activity in response to PACAP stimulation (Fig. 5A), suggesting that other elements most probably localized downstream of the PARE I region in the GnRH-R promoter were required for full stimulation. The minimal PRL promoter was then replaced by the GnRH-R promoter region extending from Ϫ136 to Ϫ32. Although this downstream region alone was insensitive to PACAP stimulation, it was capable of cooperating with the PARE I-containing module. Indeed, when both elements were linked together, full response to PACAP was recovered because the fusion promoter showed an optimal 5 Ϯ 0.9-fold response under PACAP-stimulation (p Ͻ 0.001). To investigate the importance of the Ϫ136/Ϫ32 proximal region in PACAP regulation of the GnRH-R gene, two additional PCR-generated fragments, Ϫ101/Ϫ32 and Ϫ56/Ϫ32, were fused to the 50-bp module encompassing PARE I. As shown in Fig. 5A, the Ϫ101 and Ϫ56 GnRH-R LUC constructs, either alone or fused to the 50-bp module, were unresponsive to PACAP (p Ͼ 0.05). These findings suggested that element(s) located in the Ϫ136/Ϫ101 region, hereafter referred to as PARE II, participated in PACAP responsiveness and that cooperation of PARE I and PARE II was necessary for an optimal response to PACAP.
To further establish the requirement of PARE I for the PACAP response, the block replacement mutations A, B, and C were generated in the context of the 50-bp module placed upstream of the Ϫ136/Ϫ32 region, and the resulting constructs were tested for PACAP-stimulated expression (Fig. 5B). In accordance with the data in Fig. 5, mutations A, B, and C (SF-1 mutant) led to significant decreases of 82, 95, and 79%, respectively, in PACAP-activated transcription (p Ͻ 0.001). As expected, the cooperative action of PARE I and PARE II in mediating PACAP response was abrogated by targeted mutagenesis across the PARE I region.
Mutation of the cAMP Response Element within the PARE II Region Strongly Reduced PACAP-induced Stimulation-Analysis of the PARE II region revealed the presence of an imperfect CRE (5Ј-TGACGTTT-3Ј) at position Ϫ110/Ϫ103. To determine whether this CRE was critical for PACAP responsiveness of the GnRH-R gene, four block replacement mutations (eco1 to eco4) were synthesized across the PARE II region in the context of the Ϫ136/Ϫ32 region linked to the PARE I-containing module. These mutations consisted of replacing the wild-type sequence with a 8-bp sequence that included an EcoRI restriction site. As shown on Fig. 6, mutation of the imperfect CRE (eco4) elicited a weak and insignificant decrease (p Ͼ 0.05) in the basal activity of the fusion construct, whereas it resulted in a major 83% loss of PACAP-stimulated promoter activity as compared with the wild-type promoter (1.4 Ϯ 0.1-fold versus 3.6 Ϯ 0.3-

FIG. 5. Two distinct promoter regions, PARE I and PARE II, are required for PACAP-regulated activity of the rat GnRH-R gene.
A, promoter fusion constructs were designed, in which a 50-bp module corresponding to the Ϫ275/Ϫ226 SF-1-containing region was fused to either the rat PRL minimal promoter, or the Ϫ136/Ϫ32, the Ϫ101/Ϫ32, or the Ϫ56/Ϫ32 proximal region of the GnRH-R gene, placed upstream of the luciferase (LUC) reporter. The fusion constructs were transfected into ␣T3-1 cells as described previously. B, block replacement mutations A, B, and C were introduced into the context of the 50-bp module-(Ϫ136/Ϫ32)-Luc vector. The major transcription start sites located at positions Ϫ110/Ϫ107 and Ϫ99/Ϫ97 (relative to the ATG codon) are indicated by bent arrows. Luciferase activity was corrected for transfection efficiency by normalizing to the activity of TK-Renilla luciferase expression vector and expressed as fold stimulation over pLuc/PRL construct. All results shown are the means Ϯ S.D. of duplicate samples of at least three independent experiments. fold, p Ͻ 0.001). Mutations eco1 and eco3 also led to a significant but moderated decrease in PACAP stimulation (2.3 Ϯ 0.1-fold and 2.4 Ϯ 0.2-fold), and an unaltered stimulated activity was observed for mutation eco2 as compared with the wildtype construct (3.6 Ϯ 0.3-fold, p Ͼ 0.05). Therefore, the imperfect CRE located at Ϫ110/Ϫ103 appeared as the most active element that could cooperate with PARE I in mediating PACAP responsiveness. Altogether these data demonstrated that two distinct sequences were involved in the mediation of PACAPstimulated GnRH-R gene expression: PARE I, extending from Ϫ260 to Ϫ237 and including an SF-1 site, and PARE II, which contained a functional although imperfect CRE.
Gel Shift Experiments Reveal a Major Complex That Involves the CRE in the PARE II Region-To examine whether the PARE II region could bind specific factors, we designed a double-stranded oligonucleotide extending from Ϫ120 to Ϫ97 (CRE/PARE II probe) that was used in gel shift assays with the ␣T3-1 nuclear extracts. PACAP treatment of cells before the preparation of nuclear extract did not significantly change the results. As shown in Fig. 7A, a major retarded complex was observed that was not present with the probe alone. This protein-DNA interaction was specific, because there was a dosedependent reduction in the intensity of the shifted band when an increasing amount of the unlabeled competitor probe (10-, 100-, and 1000-fold molar excess, lanes [3][4][5] was added to the binding reaction. In contrast, the mutant CRE/PARE II homolog (MutCRE) failed to abolish complex formation (lanes 6 -8).
In the mutated probe, the sequence of the putative CRE was replaced by a NotI site. The formation of the complex could also be competed by an excess of CREB consensus oligonucleotide (10 -1000-fold, lanes 9 -12), indicating that the PARE II region seemed to contain a bona fide CREB/ATF binding element, most likely located at position Ϫ110/Ϫ103, even though it deviated from the consensus octameric CRE sequence (5Ј-TGACGTCA-3Ј). Interestingly, competition with increasing concentrations of unlabeled CREB consensus probe displaced the binding of the DNA-protein complex more efficiently than the homologous DNA competitor itself, probably because the CRE-like sequence of the GnRH-R promoter has a weaker affinity for CREB/ATF-related factors than the canonical probe. To ascertain the binding of a CREB/ATF protein to the PARE II region, we performed supershift assays with a rabbit polyclonal anti-CREB antibody that reacted with members of the CREB/ATF family (CREB-1, ATF-1, and cAMP-responsive element modulator). The addition of the anti-CREB antibody completely prevented binding to the radiolabeled CRE/PARE II probe (Fig. 7B, lanes 6 -8), whereas the control antibody (a mouse IgG directed against human luteinizing hormone) had no effect (lanes [3][4][5]. Similar results were obtained with the radiolabeled CREB consensus probe that was used as a positive control (lanes 9 -14). These data demonstrate that a member of the CREB family effectively binds to the PARE II region of the rat GnRH-R gene promoter. DISCUSSION In the present study, we have examined the regulation of the activity of the GnRH-R gene promoter by PACAP using transient transfection in ␣T3-1 cells, a well characterized mouse pituitary gonadotrope cell line that expresses a functional GnRH-R (31) and a PACAP-selective receptor (PAC 1 -R). It is established that in these cells, PAC 1 -R are coupled to both cAMP and inositol phosphate production as well as to increases in intracellular Ca 2ϩ concentration (18). By transient transfection in ␣T3-1 cells, we demonstrate for the first time that PACAP stimulates the activity of the rat GnRH-R gene promoter through the cAMP pathway, providing a mechanism by which this hypophysiotropic peptide may operate as a modulator of GnRH action in the anterior pituitary. This is reminiscent of previous findings showing that cAMP and PACAP enhance GnRH-induced hormone secretion in perifused rat pituitary cells (25,32).
It is noteworthy that PACAP action on the GnRH-R gene promoter activity is mediated through the sole activation of the PKA-dependent signaling pathway, whereas PACAP may act in ␣T3-1 cells via both the PKA-and PKC-dependent pathways (18). Indeed, cholera toxin-induced production of endogenous cAMP, unlike TPA, stimulates promoter activity. Moreover, PACAP effects on promoter activity are compatible with cAMP mediation based on the similarity between the dose dependence relationship determined in this study and those previously established for cAMP generation in these cells (18). Furthermore, co-transfection using a vector expressing the PKI dramatically decreased the stimulatory effect of PACAP, providing additional evidence that PACAP acts primarily through the PKA-dependent pathway. Finally, and most importantly, the cis-acting sequences that promote PACAP action are colocalized with those of cholera toxin and involve, in addition to the SF-1 element containing domain (PARE I region), a CRE that binds a protein most likely belonging to the CREB/ATF family (PARE II region).
The sensitivity of the rat GnRH-R gene to PACAP is also the property of the gonadotropin ␣-subunit gene, another important marker gene of the gonadotrope lineage (26). Interestingly, the action of PACAP on the ␣-subunit gene is similarly mediated by the PKA-dependent pathway. This conclusion was FIG. 6. The CRE-like sequence located within the PARE II region is involved in PACAP responsiveness of the rat GnRH-R promoter. A, the region of the GnRH-R promoter spanning from Ϫ142 to Ϫ90 is shown. The CRE-like sequence is boxed. The location and nucleotide sequence of the introduced block replacement mutations are indicated and labeled as eco1 to eco4. B, the series of block replacement mutations were generated by PCR amplification into the context of the chimeric 50-bp module/proximal GnRH-R promoter/Luc construct. The major transcription start sites located at positions Ϫ110/Ϫ107 and Ϫ99/Ϫ97 (relative to the ATG codon) are indicated by bent arrows. The constructs were transfected in ␣T3-1 cells as described above. The luciferase activity was corrected for transfection efficiency by normalizing to TK-Renilla luciferase expression vector and expressed as fold stimulation over pLuc/PRL construct. The results shown are the means Ϯ S.D. of duplicate samples of at least three independent experiments. based notably on results obtained in transient transfection experiments of ␣T3-1 cells stimulated with PACAP, cAMP, or PMA, in the presence or in the absence of selective inhibitors of the PKA-or PKC-dependent pathways (33). Likewise, Burrin et al. (27), using deletion and mutational analysis combined with transfection studies in ␣T3-1 cells, have localized PACAP response elements of the human ␣-subunit gene to a 50-bp sequence in the proximal promoter, which includes an SF-1binding site. In addition, full PACAP activation was shown to require the two intact CREs, located further downstream to this 50-bp region. Altogether these data suggest that the rat GnRH-R and the ␣-subunit genes can be coordinately regulated by PACAP via similar intracellular mechanisms in gonadotrope cells of the pituitary gland.
Our results showing that luciferase expression driven by the 1.2-kilobase rat GnRH-R promoter is stimulated by forskolin and cAMP analogs are consistent with the data obtained by Reinhart et al. (7). Together, these and our data contrast with those obtained with the mouse promoter, which is unaffected by forskolin treatment in transiently transfected ␣T3-1 cells (13,14). This suggests that the differential sensitivity of the rat and murine genes regarding the activation of the PKA-dependent pathways may be an intrinsic property of their respective promoters.
This is somewhat intriguing because the mouse promoter contains sequences highly homologous to the PARE regions and located at positions similar to those identified within the rat promoter. Consequently, it would be potentially able to respond to cAMP stimulation. The main difference identified to date between the two promoters is that regarding the efficiency of the GRAS element. This element was shown to be crucial for cell-specific expression of the mouse promoter (9), whereas it was much less efficient in the rat promoter context (11). This may provide a possible explanation for the differential sensitivity of the rat and mouse promoter with respect to the activation of the cAMP/PKA signaling pathway. The ␣T3-1 cells are known to produce activin, and its autocrine/paracrine stimulatory action on the mouse GnRH-R promoter is mediated through the GRAS element (see the Introduction and Ref. 12). Also, in ␣T3-1 cells, it was previously established that PACAP could activate the follistatin promoter via the cAMP-dependent PKA pathway (34). Because follistatin is a powerful inhibitor of activin action, PACAP could neutralize the activin-induced stimulation of the mouse GnRH-R promoter activity by stimulating follistatin production. The direct positive action of PACAP on the mouse promoter would therefore be masked by the indirect and opposite action of follistatin. Regarding the rat promoter, because the GRAS element is very poorly active (11), the direct effect of PACAP on promoter activity would be predominant. This hypothesis is consistent with data obtained in the GGH3 cell line, a somatolactotrope cell line stably expressing the mouse GnRH-R (16). These cells do not produce activin; the mouse GnRH-R promoter can thus be stimulated by activation of the PKA-dependent signaling pathway (see the introduction and Refs. 16 and 35) and deletion of the CRE (TGACGTTT) within the mouse promoter prevented cAMP-dependent stimulation (36).
As an initial step toward the identification and localization of PACAP-and cAMP-responsive elements within the GnRH-R promoter, serial 5Ј-deletion mutants were tested and exhibited FIG. 7. A major complex within the PARE II region (؊136/ ؊101) of the rat GnRH-R promoter involves the CRE. Electrophoretic mobility shift assays were performed using nuclear extracts prepared from ␣T3-1 cells. A, nuclear extracts (9 g) were subjected to the binding reaction along with ϳ10 fmol of wild-type (CRE/PARE II) probe. Competition for binding was conducted with increasing concentrations of the homologous unlabeled probe and the mutant probe, in which the CRE-like sequence was replaced by a NotI site or the commercial CREB consensus oligonucleotide (Promega) at a 10 -1000-fold molar excess. B, nuclear proteins (9 g) were incubated with an affinitypurified rabbit polyclonal anti-CREB antibody or an equal concentration of mouse IgG prior to the addition of radiolabeled probes consisting of the CRE/PARE II or the CREB consensus probe. All binding reactions were subjected to electrophoresis through nondenaturing 5% polyacrylamide gels as described under "Experimental Procedures." equivalent patterns of expression. This led to the demonstration of the co-localization of PACAP and cAMP responsiveness in the SF-1-containing region between Ϫ272 and Ϫ222, which excluded in this regulation the GRAS (Ϫ412/Ϫ395) and AP-1 (Ϫ352/Ϫ346) sites. Further refinement of our results by means of additional deletions as well as mutational analysis in the proximal promoter resulted in the delineation of responsive elements for both PACAP and cAMP at two distinct sites: within the region extending from Ϫ260 to Ϫ237 (PARE I), which contains the SF-1-binding site, and the region between Ϫ136 and Ϫ101 (PARE II), which contains the imperfect CRE. Intriguingly, our experiments disclose the nucleotide sequence upstream adjacent to the SF-1 element, from Ϫ252 to Ϫ245 (5Ј TTACACTT 3Ј), as the most crucial contributor to PACAP responsiveness because its disruption induced a quasi-total abrogation of PACAP response (95% inhibition). In comparison, mutation of sequences surrounding this element, notably the invalidation of the SF-1 element and mutation A, led to an important but incomplete inhibition (79 and 82%, respectively). These findings suggest the existence of an unknown factor that binds to the AB sequence and possibly interacts with SF-1. Likewise, gel mobility shift assays combined with antibody abrogation experiments suggest that the high affinity binding of the unknown factor to the Ϫ260/Ϫ245 AB sequence was not only dependent on the integrity of the adjacent SF-1 motif but was also necessary for a physical interaction with SF-1 itself. Further studies will be necessary to clarify this point and identify this factor.
It is noteworthy that SF-1 is involved in both constitutive and cAMP-regulated expression of various genes, viz. an SF-1 motif is required for both basal and cAMP-induced regulation of the rat HDL receptor promoter (37). In the human ACTH receptor promoter, two SF-1-binding sites, SF-35 and SF-98, must be present to elicit a response to cAMP, whereas full constitutive activity necessitates both sites plus a third, SF-209 (38). Furthermore, SF-1 that mediates basal and cAMP-regulated transcription of the rat steroid cytochrome P450c17 gene can be phosphorylated in vitro by PKA, which diminishes its binding and hence may play a regulatory role in transcriptional activation (39).
Alternatively, the imperfect CRE located at Ϫ110/Ϫ103 within the PARE II region appears crucial for PACAP responsiveness, because the CRE mutant (eco4) elicits a 83% decrease in PACAP stimulation. This response element binds a protein immunologically related to the CREB family, suggesting that such a factor participates together with SF-1 and the AB factor in PACAP regulation of the GnRH-R gene. As targeted mutagenesis of either the SF-1 site, the neighboring region AB or the CRE sequence significantly impairs PACAP-stimulated activity of the GnRH-R fusion construct, the cooperative action of PARE I and PARE II seems to occur in a synergistic manner. Similarly, by using selected mutants of the CREB and SF-1binding sites within the context of the rat aromatase promoter, Carlone and Richards (40) showed that CREB and SF-1 interact synergistically to confer high constitutive activity in R2C Leydig cells. Likewise, mutation in either the CRE or the SF-1 regulatory element completely eliminates synergistic stimulation of the rat inhibin ␣-promoter activity by SF-1 and the cAMP pathway in cells co-transfected with PKA and SF-1 expression vectors (41). In the same report, it was stated that SF-1 interacts directly with CREB through the likely recruitment of CBP/p300, because this co-activator further enhances transcription by these pathways. SF-1 was shown to interact with two domains of CBP/p300 that were distinct from the CREB-binding domain (42), which raised the possibility that CBP/p300 may serve as a signal integrator for both SF-1 and CREB factors. Based on these studies and our findings, we propose a model for the activation of the GnRH-R gene promoter by PACAP, which involves the contribution of CBP/p300 (Fig. 8). Under basal conditions, the AB factors together with SF-1 bind to the PARE I region within the GnRH-R gene promoter. The co-activator CBP/p300 is subsequently recruited through SF-1 interaction and promotes constitutive promoter activity. Under PACAP stimulation, CREB-related factors are phosphorylated and can then bind to CBP/p300 with high affinity. SF-1 and CREB-related factors, through their interaction with CBP/p300, are then in sufficiently close proximity to establish protein-protein interactions with a resulting increase in transcriptional activity. In contrast, the absence of one of these factors would preclude the formation of the multi-component complex, leading to nearly complete suppression of PACAP response. Such a situation occurs when the action of either AB, SF-1, or CREB-related factors is abrogated experimentally by targeted mutagenesis of their cognate elements. This model does not exclude the interaction of SF-1 with other cofactors besides CBP such as steroid receptor co-activator-1 (43,44). Further experiments will be necessary to refine the model and elucidate the precise mechanisms involved in this transcriptional regulation, which represents an important and novel aspect in the neurohormonal control of gonadotrope function.
(University of California, San Diego). We thank Marie-Claude Chenut, Danielle Duchêne, and Philippe Nguyen for contributions in the preparation of this manuscript, cell culture, and illustrations, respectively. We are grateful to Dr. Lisa Oliver (U-419 INSERM, Nantes, France) for the correction of English text and editorial assistance. We are indebted to Jean-Pierre Lagarde for help in automated DNA sequencing (Unité de Génétique Moléculaire, AP-HP Pitié-Salpêtrière, Paris, France).