Promoter Sequences Targeting Tissue-specific Gene Expression of Hypothalamic and Ovarian Gonadotropin-releasing Hormone in Vivo *

Molecular mechanisms directing tissue-specific expression of gonadotropin-releasing hormone (GnRH) are difficult to study due to the paucity and scattered distribution of GnRH neurons. To identify regions of the mouse GnRH (mGnRH) promoter that are critical for appropriate tissue-specific gene expression, we generated transgenic mice with fragments (−3446/+23 bp, −2078/+23 bp, and −1005/+28 bp) of mGnRH promoter fused to the luciferase reporter gene. The pattern of mGnRH promoter activity was assessed by measuring luciferase activity in tissue homogenates. All three 5′-fragments of mGnRH promoter targeted hypothalamic expression of the luciferase transgene, but with the exception of the ovary, luciferase expression was absent in non-neural tissues. High levels of ovarian luciferase activity were observed in mice generated with both −2078 and −1005 bp of promoter. Our study is the first to define a region of the GnRH gene promoter that directs expression to both neural and non-neural tissuesin vivo. We demonstrate that DNA sequences contained within the proximal −1005 bp of the mGnRH promoter are sufficient to direct mGnRH gene expression to both the ovary and hypothalamus. Our results also suggest that DNA sequences distal to −2078 bp mediate the repression of ovarian GnRH.

Appropriate tissue-specific expression of gonadotropin-releasing hormone (GnRH) 1 is critical for establishing and maintaining reproductive competence. It has long been recognized that hypothalamic GnRH controls gonadal steroidogenesis and ovarian follicular development by stimulating the production of gonadotropins from the pituitary. More recently, the presence of extra-pituitary GnRH has been appreciated. Low levels of GnRH expression have been found in peripheral reproductive tissues, such as placenta (1), breast, ovary, and testes (2).
The molecular mechanisms that direct the appropriate tissue-specific expression of the GnRH gene are only beginning to be elucidated. The extremely low levels of expression in the peripheral tissues along with the paucity and scattered distribution of GnRH neurons have limited the in vivo study of GnRH gene regulation. In the adult mouse brain, it has been estimated that GnRH expression is limited to only 800 neurons (3). In the adult, the vast majority of the GnRH neurons are located in the basal hypothalamus and septum, but GnRH neurons have been described along the migratory pathway from the olfactory bulbs, as well as in the cerebral cortex and limbic system (3,4). A similar anatomic organization is found in all mammals (5).
Several in vitro studies have used mouse-derived immortalized GnRH-secreting neuronal cell lines to investigate the neuronal expression of the rat and mouse GnRH gene. Transient transfection studies using the rat GnRH (rGnRH) gene promoter identified a 173-bp proximal promoter region (6) and a 300-bp region located 1.8 kb upstream from the transcription start site that conferred cell-specific expression (7). Both these sites were reported to be important for the correct neuronal expression of the rGnRH gene in vitro (8). These in vitro studies have identified several transcription factors that interact with these promoter regions to regulate rGnRH expression: C/EBP␤ (9), GATA (10,11), Oct-1 (12,13), Otx 2 (14), and SCIP/Oct-6/ Tst-1 (15). These two promoter regions are highly conserved between the rat and mouse, and in vitro studies with the mouse GnRH (mGnRH) gene promoter suggest that Oct-1 may also regulate the neuronal expression of the mGnRH gene (16).
Ultimately, these in vitro studies are unlikely to reflect the elaborate intricacy of in vivo GnRH gene regulation. In the brain, GnRH neurons are dispersed and are influenced by the growth factors, steroids, and neurotransmitters secreted by the various adjacent cell types. Our laboratory has developed previously an in vivo model to study the regulation of the human GnRH (hGnRH) gene in transgenic mice and localized a cell-specific element between Ϫ1131 bp and Ϫ484 bp of the hGnRH gene (17). A transgenic mouse study identified the critical elements for expression of the mGnRH gene between Ϫ2.1 kb and Ϫ1.7 kb of the distal 5Ј-sequence (18). Interestingly, the human and mouse cell-specific regions bear little homology to each other, and neither share homology with the critical rat promoter regions identified in the in vitro studies.
To reconcile these discrepant observations, we constructed transgenic mice with fragments of the mGnRH gene promoter fused to the luciferase (LUC) reporter gene (19). Here we report that the proximal Ϫ1005 bp of the mGnRH promoter contains the critical elements for hypothalamic expression of mGnRH. Furthermore, we demonstrate that the DNA sequences con-tained within the proximal Ϫ1005 bp are also sufficient to direct mGnRH expression to the ovary. Our study in the mGnRH-LUC mice is the first to define a region of the GnRH promoter that directs GnRH expression to both neural and non-neural tissues in vivo.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-Unless otherwise indicated, all chemicals and reagents were obtained from Sigma. Restriction enzymes were obtained from New England Biolabs (Beverly, MA) unless otherwise specified.
Construction of the mGnRH Promoter-Luciferase DNA Constructs-A GnRH promoter-luciferase construct containing Ϫ3446 to ϩ23 bp of the mouse GnRH promoter fused to pSV0aL⌬5Ј luciferase was kindly provided by Dr. Donald B. DeFranco (University of Pittsburgh, Pittsburgh, PA). The Ϫ2078/ϩ23-bp construct was produced by restriction enzyme digestion with BamHI and BglII to remove the sequences between Ϫ3446 bp and Ϫ2078 bp of the mGnRH promoter. The resulting DNA was re-ligated to generate a BamHI/BglII site. Appropriate ligation was verified by DNA sequencing analysis.
The Ϫ1005/ϩ28-bp mGnRH-luciferase construct was generated as a HindIII fragment using PCR and the Ϫ3446/ϩ23 mGnRH-pSV0aL⌬5Ј construct as template. PCR was performed in a thermocycler (PerkinElmer Life Sciences, GeneAmp PCR System 9600), and reaction mixtures contained 5 units of Taq polymerase (DisplayTaq, Display Systems Biotech, Vista, CA) and 0.5 mM dNTPs. The PCR product was restriction enzyme-digested with HindIII and ligated into a HindIIIlinearized, alkaline phosphatase (Roche Molecular Biochemicals)-dephosphorylated pA3LUC reporter vector (20,21). Orientation was checked by sequencing constructs using a primer annealing to the 5Ј end of the luciferase gene.
For the Ϫ3446mGnRH-LUC and Ϫ2078mGnRH-LUC DNA constructs, the surrounding plasmid was removed from the mGnRH-luciferase fragment by NdeI and Xmn1 double digestion. For the Ϫ1005 mGnRH-LUC construct, PvuI digestion was used to isolate it from the surrounding plasmid. The resulting DNA was electrophoresed, and the linear mGnRH-luciferase DNA fragment was excised from the gel, isolated by electroelution into a dialysis membrane (Invitrogen), and purified on an Elutip-D column (Schleicher & Schuell).
Construction and Identification of Transgenic Animals-Transgenic animals were constructed by the Beth Israel Transgenic Facility by pronuclear injection. Fertilized mouse oocytes from FVB-N mice were injected with the purified linear mGnRH-LUC DNA fragment. The resulting embryos were transferred into pseudo-pregnant foster mothers. Transgenic animals were identified with Southern blot analysis as described previously (17). Briefly, DNA was isolated from tail snips, and restriction enzyme was digested with EcoRI and separated with gel electrophoresis. DNA was then transferred to GeneScreen Plus hybridization transfer membrane (PerkinElmer Life Sciences). A 32 P-labeled 1.2-kb probe for luciferase was used to detect transgenic animals that incorporated the luciferase transgene. For identification of luciferaseexpressing transgenic lines, neonatal brains were removed from pups, homogenized, and assayed for the mGnRH-luciferase transgene as described below.
All procedures were carried out in accordance with the Animal Care and Use Committees of Children's Hospital, Boston, and The University of Chicago.
Assay of mGnRH-Luciferase Transgene-From adult mice, the hypothalamus was dissected in a single fragment consisting of tissue from 1 mm caudal to the mammillary bodies to a point just anterior of the optic chiasm, 1 mm laterally beyond the lateral aspect of the median eminence, and 3 mm dorsally. The olfactory tissue was dissected to include both olfactory bulbs and the tissue rostral to the hypothalamic section. Small representative sections were taken from the remaining tissues. In the case of the gonads, entire gonads were used due to their small size. Similarly, for examination of neonatal luciferase expression, the entire brain was removed from 2-to 3-day-old pups bearing the mGnRH-LUC transgene.
Tissues were placed in 1 ml of lysis buffer (25 mM glycylglycine, 15 mM MgSO 4 , 4 mM EGTA, 1% Triton X-100, and 1 mM dithiothreitol) and homogenized with a Polytron tissue homogenizer (Brinkmann Instruments). The homogenate was centrifuged at 15,000 ϫ g, and the supernatant was assayed for luciferase activity. Luciferase assays were done using a Lumat LB9501 luminometer (Berthold Systems Inc., Pittsburgh, PA). Samples were injected with 100 l of 0.75 mM luciferin (Molecular Probes, Eugene, OR), dissolved in lysis buffer, and 100 l of assay buffer (25 mM glycylglycine, 15 mM MgSO 4 , 4 mM EGTA, 15 mM KPO 4 , 3 mM dithiothreitol, and 3 mM ATP), and luminescence was measured for 20 s as relative light units (RLU). For some experiments, RLU were normalized for tissue size by correcting for protein content. Protein assays were done using Bio-Rad reagent and bovine serum albumin standards.
Tissue Preparation for Histology-Brains and ovaries were fixed in situ by perfusing adult mice with a 10% buffered formalin phosphate solution (Fisher). After perfusion, the brains or ovaries were removed and stored overnight in the perfusion solution with 10 -15% sucrose. Frozen sections were obtained using a Leica SM2000R sliding microtome. Brain (30 M) and ovarian (20 M) sections were collected in phosphate-buffered saline (PBS).
Histology-Immunocytochemistry using antibodies specific for luciferase and GnRH peptides was performed to co-localize these peptides in the hypothalamus. All incubations were done on a shaking platform. Brain sections were incubated overnight at 4°C with goat anti-luciferase antibody (Cortex Biochemical, San Leandro, CA) at a concentration of 1-3 g/ml diluted in PBS with 0.3% Triton X and 1% bovine serum albumin. Sections were washed in PBS, and luciferase immunoreactivity was amplified by incubating with biotinylated horse anti-goat IgG (Vector Laboratories; Burlingame, CA) at room temperature for 2 h at a concentration of 6.75 g/ml PBS with 0.3% Triton-X. Sections were washed in PBS and then incubated with streptavidin-conjugated Alexa Fluor 488 (Molecular Probes, Eugene, OR) for 2 h at a concentration of 5 g/ml PBS. After washing with PBS, sections were incubated overnight at 4°C with a 1:5,000 dilution of GnRH antibody (LR5 rabbit anti-GnRH decapeptide antibody kindly provided by Robert Benoit) in blocking solution of PBS with 0.3% Triton X and 1% bovine serum albumin. Sections were washed with PBS at room temperature and then were incubated with 3 l/ml Cy3-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch, West Grove, PA) for 2 h and followed by a PBS wash.
For the detection of luciferase protein in the ovaries, ovarian sections were incubated overnight at 4°C with rabbit anti-luciferase antibody (Cortex Biochemical, San Leandro, CA) at a concentration of 1 g/ml diluted in PBS with 0.3% Triton-X and 1% bovine serum albumin. Sections were washed in PBS and then incubated with 3 l/ml Cy3conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch) for 2 h and followed by a PBS wash.
Microscopy-Tissue sections were mounted on slides and coverslipped with vectashield mounting reagent (Vector Laboratories, Burlingame, CA). The edges of the coverslips were then sealed with standard nail polish. The double fluorescence histology sections were analyzed on a Zeiss Axiovert SS100TV inverted fluorescent microscope, and images were captured with a Zeiss video camera with a CCD chip. The Cy3 and Alexa Fluor 488 fluorescences were visualized using the appropriate filters.

RESULTS
Ϫ3446 bp of the mGnRH Gene Promoter Appropriately Targets Luciferase Expression to GnRH Neurons-A DNA construct containing the region from Ϫ3446 bp to ϩ23 bp of the mGnRH promoter fused to the luciferase reporter gene was used to generate transgenic mice (Ϫ3446mGnRH-LUC). As shown in Table I, Southern blot analysis identified four separate founder lines that incorporated the luciferase transgene. Because luciferase expression is under the control of the mGnRH promoter, mGnRH promoter activity can be detected easily by measuring the luciferase activity in tissue homogenates. To identify founder lines expressing the mGnRH-LUC transgene, whole brain homogenates were obtained from neonatal mice and assayed for luciferase activity. Luciferase activity was measured as RLU. Two Ϫ3446mGnRH-LUC founder lines were found to express luciferase. In brain homogenates from 2-to 3-day-old pups, luciferase activity was 22,309 Ϯ 1042 RLU in the lower expressing line and 43,488 Ϯ 4762 in the higher expressing line. Luciferase activity in the founder lines that did not express luciferase (19.3 Ϯ 7.4 RLU) did not differ from background levels.
In the mGnRH-LUC mice, the anatomic pattern of luciferase expression is an assay of mGnRH promoter activity. As illustrated in Fig. 1A, a similar anatomic pattern of luciferase expression was found in offspring from both founders that expressed the Ϫ3446mGnRH-LUC transgene. Data from three female offspring of founder 1 and four female offspring of founder 2 are shown. As expected, high levels of luciferase activity were seen in the hypothalamus (20,148 Ϯ 6093 and 21,263 Ϯ 8214) and olfactory lobes (10,255 Ϯ 1117 and 15,957 Ϯ 3782) when luciferase expression is under the control of the mGnRH promoter. Luciferase activity was detected in the cortex (2771 Ϯ 1505 and 12,572 Ϯ 4027), cerebellum (893 Ϯ 214 and 3807 Ϯ 1255), and midbrain (668 Ϯ 184 and 1427 Ϯ 773). With the exception of the ovary, which had low levels of luciferase activity (154 Ϯ 24 and 41 Ϯ 13), luciferase levels in the other tissues, including the testes, as shown in Fig. 2A, did not differ from background levels. These two lines of Ϫ3446mGnRH-LUC mice demonstrate that Ϫ3446 bp of the mGnRH promoter targets luciferase expression appropriately to GnRH-containing regions of the mouse brain.
To confirm that luciferase expression was confined to GnRH neurons in the hypothalamus, double-labeling immunocytochemistry experiments were performed in the Ϫ3446mGnRH-LUC mice. Brain sections from the Ϫ3446mGnRH-LUC mice were labeled with antibodies specific for luciferase and GnRH peptides. A representative neuron is shown in Fig. 1B. Luciferase expression was found only in association with GnRHcontaining neurons demonstrating that luciferase expression does faithfully reflect mGnRH expression.
The GnRH Hypothalamic Specific Element Is Located within the Proximal Ϫ1005 bp of the mGnRH Promoter-To define further the promoter element necessary to target hypothalamic luciferase expression, transgenic mice were generated using the proximal Ϫ1005 bp of the mGnRH promoter fused to the luciferase reporter gene (Ϫ1005mGnRH-LUC). As shown in Table I, six separate founder lines bearing the Ϫ1005mGnRH-LUC transgene were generated, and three transgenic lines were found to express luciferase in the neonatal brain. In brain homogenates from 2-to 3-day-old pups, luciferase levels were 7044 Ϯ 472 RLU in the lowest luciferase-expressing line and 22,259 Ϯ 5872 in the highest luciferase-expressing line. In founder lines that incorporated, but did not express the luciferase transgene, luciferase activity was 29.3 Ϯ 9.1 RLU in neonatal brain homogenates.
As shown in Fig. 1C, the anatomic pattern of luciferase expression was similar in the three luciferase-expressing lines. Data from two female offspring of founder 1 and four female offspring of founders 2 and 3 are shown. This smaller fragment of the mGnRH promoter, containing the proximal Ϫ1005 bp, also targets luciferase expression to the hypothalamus (584 Ϯ 77, 1221 Ϯ 353, and 1244 Ϯ 104), but lower levels were seen. Hypothalamic luciferase levels were 16 -36-fold higher in the transgenic mice bearing the Ϫ3446mGnRH-LUC transgene (20,148 Ϯ 6093 and 21,263 Ϯ 8214).
Although the mice generated with the Ϫ1005mGnRH-LUC transgene expressed much lower levels of luciferase, immunocytochemistry studies detected luciferase in GnRH-containing neurons, and a representative neuron is shown in Fig. 1D. As with the transgenic mice generated with Ϫ3446 bp of the mGnRH gene promoter, luciferase expression in the hypothalamus was found only in association with GnRH-containing neurons demonstrating that luciferase expression does faithfully reflect mGnRH expression.
Analysis of the Ϫ1005mGnRH-LUC mice demonstrates that the proximal Ϫ1005 bp of the mGnRH promoter targets luciferase expression to hypothalamic GnRH-containing neurons, suggesting that the mGnRH hypothalamic specific element is located within the proximal Ϫ1005 bp of the mGnRH promoter.
Deletion of the mGnRH Promoter Region Distal to Ϫ1005 bp Unmasks mGnRH Promoter Activity in the Ovary-Interestingly, as shown in Fig. 1C, offspring from the three luciferaseexpressing founder lines of the Ϫ1005mGnRH-LUC mice demonstrated very high levels of luciferase activity in their ovarian homogenates (13,641 Ϯ 652, 12,432 Ϯ 1477, and 32,893 Ϯ 5557). Luciferase activity in the other non-neural tissues was similar to background levels, indicating that ovarian luciferase was not an artifact of where the luciferase transgene incorporated into the mouse genome.
The high level of ovarian luciferase expression seen in the mice bearing the Ϫ1005mGnRH-LUC transgene was somewhat surprising because there was minimal ovarian expression (154 Ϯ 24 and 41 Ϯ 13) found in the transgenic mice bearing the Ϫ3446mGnRH-LUC transgene (Fig. 1A). No significant ovarian luciferase expression was detected in the Ϫ3446mGnRH-LUC mice even after additional female mice (n ϭ 22) from both founder lines bearing the Ϫ3446mGnRH-LUC transgene were examined (data not shown). The additional female mice ranged from 24 to 296 days in age and included lactating (n ϭ 4), pregnant (n ϭ 3), pre-pubertal (n ϭ 4), and cycling females (n ϭ 11).
These findings demonstrate that the proximal Ϫ1005 bp of the mGnRH promoter targets luciferase expression to the ovary, as well as the hypothalamus, suggesting that the GnRH ovarian-specific element is also located within the proximal Ϫ1005 bp of the mGnRH promoter. Furthermore, removal of the distal mGnRH promoter region, between Ϫ3446 bp and Ϫ1005 bp, unmasks ovarian luciferase activity in the Ϫ1005mGnRH-LUC mice. Our results suggest that the ovary may expresses GnRH at low levels because low levels of luciferase expression were detected in the Ϫ3446mGnRH-LUC mice. This finding suggests that DNA sequences in the distal mGnRH promoter, between Ϫ3446 bp and Ϫ1005 bp, TABLE I Effect of mGnRH promoter fragment on expression of luciferase transgene in the neonatal mouse brain Three different mouse GnRH promoter fragments were fused to the luciferase reporter gene and used to generate transgenic mouse lines. Transgenic lines were identified by Southern blot analysis. Luciferase expression was determined by performing luciferase assays in whole brain homogenates from neonatal mice. Luciferase activity was measured as RLU. Luciferase assays were performed in a total of 506 neonatal mice from the 27 different founder lines. Overall, luciferase activity was detected in 48% (13/27) of founder lines containing the luciferase transgene. The table shows the range of luciferase activity seen in different founder lines generated with the same mGnRH-LUC DNA construct. The mean Ϯ S.E. of luciferase activity found in whole brain homogenates from 2-3-day-old pups is shown for the lowest and highest luciferase-expressing founder lines. Luciferase activity, in brain homogenates from non-luciferase expressing lines, was similar to background levels. may normally act as an ovarian GnRH repressor element to mediate the repression of ovarian GnRH.
The mGnRH Promoter Targets Luciferase to the Granulosa Cells of Specific Ovarian Follicles-To identify the luciferaseexpressing cells in the ovary, immunocytochemistry experiments were performed using adult cycling female mice bearing the Ϫ1005mGnRH-LUC transgene. Initially, ovarian sections were labeled with antibodies specific for luciferase and GnRH peptides using the staining protocol used for the brain sections. Luciferase immunoreactivity was amplified using biotinylated secondary antibody with streptavidin-conjugated Alexa Fluor 488. GnRH immunoreactivity was visualized using Cy3 fluorescence. In these studies, no immunostaining for GnRH was detected (data not shown). Additionally, immunostaining for luciferase could not be determined because there was a great deal of nonspecific Alexa Fluor 488 fluorescence seen in our control sections, presumably from the endogenous biotin present in the ovary (data not shown).
The immunocytochemistry protocol was modified to eliminate the biotinylated secondary antibody. Ovarian sections were obtained from an adult cycling female mouse. Experimental sections were incubated with an anti-luciferase antibody, whereas control sections were incubated with nonspecific rabbit IgG. Immunoreactivity was detected by Cy3 fluorescence. Representative ovarian follicles are shown in Fig. 2. In the mice bearing the Ϫ1005mGnRH-LUC transgene, no ovarian stain-FIG. 1. Luciferase expression in ؊3446mGnRH-LUC and ؊1005mGnRH-LUC mice. A shows the anatomic pattern of luciferase expression in tissue homogenates from the Ϫ3446mGnRH-LUC mice, generated with Ϫ3446/ϩ23 bp of the mGnRH promoter fused to the luciferase reporter gene. Luciferase expression was measured in RLU. Data from founder 1 (three females) and founder 2 (four females) are shown. In these mice, luciferase expression was detected at high levels in the hypothalamus and olfactory bulbs. Low levels of luciferase activity were detected in the cortex, cerebellum, midbrain, and minimally, in the ovary. Luciferase levels in the other tissues did not differ from background levels. B shows co-localization of luciferase and GnRH in a representative neuron from the Ϫ3446mGnRH-LUC mouse. Coronal brain sections were incubated with an anti-luciferase antibody and an anti-GnRH antibody. In the left panel, the red Cy3 fluorescence represents GnRH peptide. In the middle panel, the green Alexa Fluor 488 fluorescence represents luciferase peptide. In the right panel, an overlay of the red and green fluorescence demonstrates co-localization of luciferase and GnRH. C shows the anatomic pattern of luciferase expression in tissue homogenates from the Ϫ1005mGnRH-LUC mice, generated with Ϫ1005/ϩ23 bp of the mGnRH promoter fused to the luciferase reporter gene. Luciferase expression was measured in RLU. A total of 10 female mice were used to generate this graph (two from founder 1 and four each from founders 2 and 3). Luciferase expression was measured in RLU. Luciferase activity was detected in the hypothalamus and olfactory lobes but at lower levels than were seen in transgenic mice bearing the Ϫ3446mGnRH-LUC transgene. Luciferase expression was found at very high levels in the ovaries, but luciferase activity in the other non-neural tissues was similar to background levels. These findings suggest that DNA sequences contained in the proximal Ϫ1005 bp are sufficient to direct both hypothalamic and ovarian mGnRH expression. Additionally, removal of the mGnRH promoter region distal to Ϫ1005 bp unmasks mGnRH expression in the ovary. D shows co-localization of luciferase and GnRH in a representative neuron from the Ϫ1005mGnRH-LUC mouse. In the left panel, the red Cy3 fluorescence represents GnRH peptide, and in the middle panel, the green Alexa Fluor 488 fluorescence represents luciferase protein. In the right panel, an overlay of the red and green fluorescence shows co-localization of the two proteins and demonstrates that luciferase expression faithfully reflects mGnRH expression.
ing was found in mice labeled with nonspecific IgG. In sections incubated with the anti-luciferase antibody, fluorescence was detected in the granulosa cells of some ovarian follicles. The fact that only certain follicles within the ovary contained labeled granulosa cells suggests that the luciferase expression seen in these mice is not the result of nonspecific transgene expression. Our findings suggest that increased GnRH promoter activity occurs in the granulosa cells of certain ovarian follicles, either at particular stages of follicular development or within a particular cohort of ovarian follicles.
The Ovarian GRH Repressor Element Is Located between Ϫ3446 and Ϫ2078 bp of the mGnRH Promoter-To define further the promoter element necessary to repress ovarian luciferase expression, transgenic mice were generated with the region between Ϫ2078 bp and ϩ23 bp of the mGnRH promoter fused to the luciferase reporter gene (Ϫ2078mGnRH-LUC). As shown in Table I, Southern blot analysis identified 17 separate founders bearing the Ϫ2078mGnRH-LUC transgene, and luciferase expression was detected in neonatal brains from 8 of these founder lines. In brain homogenates from 2-day-old pups, luciferase activity was 1500 Ϯ 104 RLU in the lowest luciferase-expressing line and 12,983 Ϯ 599 RLU in the highest luciferase-expressing line. In neonatal brain homogenates from founder lines that incorporated, but did not express the luciferase transgene, luciferase activity was 25.2 Ϯ 8.7 RLU. The founder line that consistently expressed the highest level of neonatal hypothalamic luciferase was examined in greater detail.
In an attempt to correct for differences in cellular content so that ovarian and hypothalamic luciferase expression could be compared more directly, the RLU values were corrected for protein content. Fig. 3 shows the anatomic pattern of luciferase expression, corrected for protein content, in mice generated with each of the three mGnRH-LUC constructs, and the data are summarized in Table II. For each of the mGnRH-LUC constructs, both male and female offspring from a representative founder were examined. Fig. 3A depicts the anatomic pattern of luciferase expression in mice generated with the Ϫ3446mGnRH-LUC transgene. Six offspring (three male and three female) from founder 1 were used to generate the data. Fig. 3B shows data from six (three male and three female) mice from the Ϫ2078mGnRH-LUC founder line that consistently expressed the highest level of neonatal hypothalamic luciferase. Fig. 3C demonstrates the anatomic pattern of luciferase expression found in transgenic mice bearing the Ϫ1005mGnRH-LUC transgene. Five offspring (two female and three male) from founder 1 of the Ϫ1005mGnRH-LUC mice were used to generate the data.
Correcting for protein content did not qualitatively change the results. Mice bearing the Ϫ3446mGnRH-LUC transgene expressed hypothalamic luciferase at high levels (8114 Ϯ 1465 RLU/mg protein), but minimal luciferase activity was detected in their ovaries (85.7 Ϯ 17.1 RLU/mg protein). Luciferase activity in the non-neural tissues, including the testes (1.67 Ϯ 1.7 RLU/mg protein), did not differ from background level (1.87 Ϯ 0.7 RLU/mg protein). When luciferase expression was under the control of the mGnRH promoter region between Ϫ1005 to ϩ28 bp, low levels of activity were detected in the hypothalamus (474 Ϯ 86 RLU/mg protein), and high levels were detected in the ovary (32,379 Ϯ 7441 RLU/mg protein). Luciferase activity was not seen in the testes (33.0 Ϯ 0.6 RLU/mg protein) or in the other peripheral tissues (7.04 Ϯ 2.0 RLU/mg protein).
The anatomic pattern of luciferase expression in the mice bearing the Ϫ2078mGnRH-LUC transgene are qualitatively similar to the Ϫ1005mGnRH-LUC mice. The Ϫ2078mGnRH-LUC mice exhibited low levels of luciferase expression in the hypothalamus (1688 Ϯ 444 RLU/mg protein) and olfactory lobes (410 Ϯ 60 RLU/mg protein) and high levels of expression in the ovary (14,066 Ϯ 1030 RLU/mg protein). Luciferase expression in the non-reproductive, non-neural tissues, including the testes (7.00 Ϯ 3.8 RLU/mg protein), was similar to background levels (7.46 Ϯ 3.2 RLU/mg protein). The high levels of ovarian luciferase expression in the Ϫ2078mGnRH-LUC mice suggested that the ovarian GnRH repressor element is located in the distal region of the mGnRH gene promoter, between Ϫ3446 and Ϫ2078 bp.
The Distal Region of the mGnRH Promoter, between Ϫ3446 and Ϫ2078 bp, May Contain a Critical Enhancer Region for the Expression of Hypothalamic mGnRH-As shown in Table I, eight founder mice were found to express the Ϫ2078mGnRH-LUC transgene in the neonatal hypothalamus. In Fig. 3, the Ϫ2078mGnRH-LUC founder line that expressed neonatal luciferase at the highest level was compared with the Ϫ3446mGnRH-LUC founder line that expressed lower levels of neonatal luciferase. Even in the Ϫ2078mGnRH-LUC transgenic line that expressed hypothalamic luciferase at the highest level, luciferase activity in the hypothalamus (1688 Ϯ 444 RLU/mg protein) was 5-fold lower than was seen in the transgenic mice bearing the Ϫ3446mGnRH-LUC transgene (8114 Ϯ 1465 RLU/mg protein).
As shown in Fig. 3 and Table II, transgenic mice bearing the 3446mGnRH-LUC transgene exhibited higher levels of hypothalamic luciferase (8114 Ϯ 1465 RLU/mg protein) than mice bearing either the Ϫ2078mGnRH-LUC transgene (1688 Ϯ 444 RLU/mg protein) or the Ϫ1005mGnRH-LUC transgene (474 Ϯ 86 RLU/mg protein). The fact that ovarian luciferase expression is found at high levels in the transgenic mice bearing these shorter fragments of the mGnRH promoter argues against a general insertion effect, chromatin configuration, or lower transgene copy number as the cause of the lower hypothalamic FIG. 2. Ovarian luciferase expression in ؊1005mGnRH-LUC mice. Ovarian sections were incubated with rabbit anti-luciferase antibody or nonspecific rabbit IgG. Immunoreactivity was detected by anti-rabbit Cy3, which is seen as red fluorescence. A shows a low power (ϫ100) magnification of a representative ovarian section. Luciferase immunoreactivity is seen specifically within two follicles. B and C show the two luciferase-containing follicles at higher magnification (ϫ200). At this higher magnification, it is clear that Cy3 fluorescence is contained within the ovarian granulosa cells. D and E show similar follicles from an ovarian section incubated with nonspecific rabbit IgG. No specific staining of the granulosa cells is seen. There does, however, appear to be some nonspecific staining of the oocyte. expression seen in these lines. These data suggest that the mGnRH promoter region between Ϫ3446 and Ϫ2078 bp contains a critical enhancer for the in vivo expression of hypothalamic mGnRH.

A Hypothalamic Specific Element for the in Vivo Expression
of mGnRH-In this study, we used various segments of the mGnRH promoter fused to the luciferase reporter gene as a marker for in vivo mGnRH gene expression. The luciferase gene is an extremely sensitive reporter, and because luciferase expression is under the control of the mGnRH promoter, the anatomic pattern of mGnRH gene promoter activity can be detected easily by measuring the luciferase activity in tissue homogenates. All three 5Ј-fragments of the mGnRH promoter that we studied, Ϫ3446/ϩ23, Ϫ2078/ϩ23, and Ϫ1005/ϩ28 bp, targeted expression of the luciferase transgene to the hypothalamus. Our study in the mGnRH-LUC transgenic mice is the first to demonstrate that sequences contained within the proximal Ϫ1005 bp of the mGnRH promoter are sufficient to target hypothalamic expression of mGnRH.
Our findings differ from a recent transgenic mouse study in which the mGnRH promoter fused to a lacZ reporter localized the critical elements for expression of the mGnRH gene between Ϫ2.1 and Ϫ1.7 kb of the promoter (18). In these mice, deletion of mGnRH promoter sequences 5Ј to 1.7 kb resulted in a complete absence of detectable ␤-galactosidase expression within the brain, whereas we detected luciferase expression in transgenic mice bearing the Ϫ1005mGnRH-LUC transgene. As shown in Table I, luciferase expression in whole brain homogenates from 2-to 3-day-old pups was as high in the Ϫ1005mGnRH-LUC mice as in the Ϫ2078mGnRH-LUC mice. In the adult, the proximal Ϫ1005 bp of the mGnRH targeted luciferase to the hypothalamus in the Ϫ1005mGnRH-LUC mice, but expression levels were lower (474 Ϯ 86 RLU) compared with the levels seen in the Ϫ2078mGnRH-LUC mice (1,688 Ϯ 444 RLU). Perhaps the mGnRH promoter region between Ϫ2.1 and Ϫ1.7 kb, although not critical for targeting hypothalamic expression, contains sequences that enhance hypothalamic expression of GnRH.
The differences observed between the mice bearing the lacZ and luciferase transgene may be due to reporter sensitivity or to the difference in the number of animals examined. Measurement of luciferase activity in tissue homogenates may be a more sensitive assay for the detection of low levels of transcription than ␤-galactosidase immunocytochemistry. Alternatively, it is possible that examination of additional mGnRH-lacZ mice would reveal ␤-galactosidase expression within GnRH neurons because the integration site into the chromosome has been shown to repress the expression of a foreign gene in transgenic mice (22). In our study, we generated six founder lines that incorporated the Ϫ1005mGnRH-LUC transgene, but only three transgenic lines expressed luciferase. As shown in Table I, only 48% (13/27) of the founder lines incorporating the luciferase transgene expressed luciferase in the neonatal mouse brain.
The mGnRH promoter (Ϫ3446 to ϩ23 bp) has been shown to results. Mice bearing the Ϫ2078mGnRH-LUC and the Ϫ1005mGnRH-LUC constructs exhibited low levels of luciferase expression in the hypothalamus and high levels of expression in the ovary. Luciferase expression in the non-reproductive, non-neural tissues (including the testes) was similar to background levels. These findings suggest that DNA sequences contained in the proximal Ϫ1005 bp are sufficient to direct both hypothalamic and ovarian mGnRH expression. The high levels of ovarian luciferase expression in the Ϫ2078mGnRH-LUC mice suggest that the ovarian GnRH repressor element is located in the distal region of the mGnRH gene promoter, between Ϫ3446 and Ϫ2078 bp.
FIG. 3. Luciferase expression, corrected for protein content, in ؊3446mGnRH-LUC, ؊2078mGnRH-LUC, and ؊1005mGnRH-LUC mice. In an attempt to correct for differences in cellular content so that ovarian and hypothalamic luciferase expression could be compared more directly, the RLU values were corrected for protein content. Fig. 3 shows the anatomic pattern of luciferase expression, corrected for protein content, in mice generated with each of the three mGnRH-LUC constructs. Please note that a different scale is used for each of the graphs. The data are also summarized in Table II. For each of the mGnRH-LUC constructs, both male and female offspring from a representative founder were examined. A depicts the anatomic pattern of luciferase expression in mice generated with the Ϫ3446mGnRH-LUC construct. Six offspring (three male and three female) from founder 1 were used to generate the data. B shows data from six (three male and three female) mice from the Ϫ2078mGnRH-LUC founder line that consistently expressed the highest level of neonatal hypothalamic luciferase. C demonstrates the anatomic pattern of luciferase expression found in transgenic mice bearing the Ϫ1005mGnRH-LUC construct. Five offspring (two female and three male) from founder 1 of the Ϫ1005mGnRH-LUC mice were used to generate these data. It appears that correcting for protein content did not qualitatively change the target faithfully reporter expression to GnRH neurons. The Ϫ3446/ϩ23 bp mGnRH promoter fragment was used to generate transgenic mice in which green fluorescent protein (GFP) was targeted to GnRH neurons (23). In the GFP mice, 99.5% of GFP-expressing neurons were found in association with immunodetectable GnRH peptide. Similarly, when we performed double-labeling immunocytochemistry studies in our Ϫ3446mGnRH-LUC and Ϫ1005mGnRH-LUC mice using antibodies specific for both luciferase and GnRH, we detected luciferase protein only within neurons containing the GnRH peptide.
In our mGnRH-LUC mice, low levels of luciferase expression were detected in tissue homogenates obtained from the cortex, cerebellum, and midbrain. The luciferase expression seen in these tissues may reflect low levels of GnRH expression rather than ectopic luciferase expression. GnRH has been described in extra-hypothalamic brain regions (24), and because the assay for luciferase activity in tissue homogenates is a very sensitive measure of mGnRH promoter activity, it would be able to detect low levels of expression in transgenic mice.
The low levels of luciferase activity seen in these tissues may actually reflect an embryological pattern of GnRH gene expression. Transgenic mice bearing a GnRH-lacZ reporter construct were found also found to express low levels of ␤-galactosidase in extra-hypothalamic regions of the brain, and further study demonstrated GnRH expression in these non-hypothalamic regions during normal embryological development (25). Alternatively, the lack of the 3Ј-sequences in the mGnRH-LUC construct may lead to aberrant regulation of the luciferase transgene so that luciferase activity persists in the absence of GnRH peptide. It has been noted that the 3Ј-element distal to exon II appears to repress GnRH expression in other neurons but has no effect in targeting the mGnRH gene (18).
Our findings also differ from observations made in studies of the rGnRH promoter in GT1-7 cells. In these studies, both a 173-bp proximal promoter region (6) and a 300-bp region located 1.8 kb upstream from the transcription start site (7) were described to be important for neuronal expression of the rGnRH gene in vitro (8). In our studies, the sequences contained in the proximal Ϫ1005 bp of the mGnRH promoter were sufficient to direct luciferase expression to the hypothalamus. Overall, our mouse GnRH hypothalamic specific element, located within the proximal Ϫ1005 bp of the mGnRH promoter, shares 90% sequence homology with the rat gene and contains the proximal Ϫ173-bp region of the rGnRH promoter that is conserved across species (6). The discrepancy between the mouse and rat data may reflect the species-specific differences in regulation of GnRH gene or may reflect the difference between data obtained in vivo versus in vitro.
Distal to the proximal Ϫ173-bp conserved region, there is ϳ50% overall homology between the proximal Ϫ1005 bp of the human and mouse GnRH promoters. Studies of the hGnRH gene in transgenic mice localized a hypothalamic specific element between Ϫ1131 and Ϫ484 bp of the hGnRH gene (17). Within the human hypothalamic specific element, there are several regions that share greater homology with the mGnRH gene, and further study of these homologous regions may elucidate the critical factors that target hypothalamic GnRH expression.
A Critical Enhancer Region for the in Vivo Expression of Hypothalamic mGnRH-The mGnRH promoter fragments containing Ϫ2078 and Ϫ1005 bp of 5Ј-sequence targeted luciferase transgene expression to the hypothalamus, but at lower levels than the transgenic mice generated with Ϫ3446 bp of mGnRH promoter. In general, comparison of transgene expression levels between different founder lines must be done cautiously because transgene expression levels are affected by the chromosomal integration site and transgene copy number (22). Nevertheless, transgenic mice derived from eight different embryos bearing the Ϫ2078mGnRH-LUC transgene were examined. It is striking that even in the transgenic line that expressed luciferase at the highest level, luciferase activity in the hypothalamus was ϳ5-fold lower than was seen in the transgenic mice bearing the Ϫ3446mGnRH-LUC transgene. Furthermore, the fact that ovarian luciferase expression is found at high levels in the transgenic mice bearing these shorter fragments of the mGnRH promoter suggests that chromosomal integration site or lower transgene copy number is unlikely to be the cause of the lower hypothalamic expression. These data would suggest that an enhancer for the in vivo expression of hypothalamic mGnRH is contained in the mGnRH promoter region between Ϫ3446 and Ϫ2078 bp.
Our in vivo observations corroborate in vitro studies performed in the GT1-7 cell line, a GnRH-secreting cell line that was developed by targeting the SV40 T antigen oncogene to GnRH neurons using the 5Ј-flanking region of the rGnRH gene in transgenic mice. Deletion analysis of the rGnRH gene promoter in the GT1-7 cell line identified a neuron-specific enhancer, located between Ϫ1863 and Ϫ1571 bp, that was found to be critical for the expression of the rGnRH gene in vitro. This region of the rat promoter shares 90% homology to the region of the mouse GnRH promoter located between Ϫ2384 and Ϫ2081 bp. Although the region between Ϫ2384 bp and Ϫ2081 bp was not found to be essential for targeting mGnRH expression to TABLE II Effect of mGnRH promoter fragment on expression of luciferase transgene in the hypothalamus and ovary Three different mouse GnRH promoter fragments were fused to the luciferase reporter gene and used to generate transgenic mice. Luciferaseexpressing lines were identified, and adult mice were examined to analyze the anatomic pattern of luciferase expression. For each mGnRH-LUC construct shown in the table, data were obtained from mice derived from a single founder. Offspring from founder 1 of the Ϫ3446 mGnRH-LUC mice and the Ϫ1005 mGnRH-LUC mice were used. For the mice generated with Ϫ2078 mGnRH-LUC transgene, the line that expressed luciferase at the highest levels in the neonatal brain was used. Luciferase activity was measured as RLU corrected for mg of protein in the sample. Hypothalamic luciferase activity was detected in transgenic mice generated with all three mGnRH-LUC constructs. In mice generated with Ϫ2078 bp and Ϫ1005 bp of the mGnRH promoter, high levels of luciferase activity were detected in the ovaries but not in the testes. Outside the central nervous system, luciferase levels in non-reproductive tissues (heart, lung, kidney, liver, and spleen) did not differ from background levels. the hypothalamus in our mGnRH-LUC mice, deletion of sequences 5Ј to Ϫ2078 bp resulted in a dramatic decrease in hypothalamic luciferase expression levels. These results would support the hypothesis that critical enhancer sequences for the in vivo expression of hypothalamic mGnRH are located between Ϫ2384 and Ϫ2081 bp of the mGnRH promoter.
Additional studies with the rGnRH promoter in GT1-7 cells identified Oct-1 (12) and GATA-4 (10, 11) as important transcription factors that bind to the neuron-specific enhancer, located between Ϫ1863 and Ϫ1571 bp, and regulate rGnRH expression in vitro. Analysis of the corresponding mouse gene sequences, between Ϫ2384 and Ϫ2081 bp, however, reveal differences in the presumed transcription factor recognition sites. There is some unpublished evidence that these changes may eliminate binding to the mGnRH promoter in vitro (26). Further study is need to determine whether Oct-1 and GATA-4 also have a role in enhancing the in vivo expression of mGnRH and whether other transcription factors are involved.
Identification of an Ovarian GnRH Repressor Element-Interestingly, when the mGnRH promoter region distal to Ϫ2078 bp was deleted, high levels of ovarian luciferase were detected, and even higher levels of ovarian luciferase were detected in mice bearing the Ϫ1005mGnRH-LUC transgene. Our studies demonstrate that sequences contained within the proximal Ϫ1005 bp of the mGnRH promoter are sufficient to target ovarian, as well as hypothalamic, expression of mGnRH. Our data also suggest that an ovarian GnRH repressor element may be located in the distal region of the mGnRH promoter between Ϫ3446 and Ϫ2078 bp, because deletion of this region unmasks luciferase expression in the ovaries of transgenic mice bearing the mGnRH-LUC transgene. We speculate that repressor proteins in the ovary normally interact with this ovarian GnRH repressor element and permit GnRH expression only in certain physiologic situations.
In the transgenic mice in which luciferase is under the control of Ϫ1005 to ϩ28 bp of the mGnRH promoter, luciferase peptide was detected in the granulosa cells of certain ovarian follicles. The fact that the luciferase protein is restricted to the granulosa cells of specific ovarian follicles suggests that ovarian luciferase expression is regulated and is not the result of nonspecific transgene expression. Our findings indicate that mGnRH promoter activity is increased in the granulosa cells of certain ovarian follicles. Perhaps under certain physiologic conditions, granulosa cells, in a particular cohort of follicles, are stimulated to release repressor proteins from the mGnRH promoter to allow mGnRH gene expression to occur.
Our finding of high levels of ovarian luciferase expression was initially surprising because no ovarian luciferase activity was detected in our previous studies (17) using the hGnRH promoter to direct luciferase expression. This discrepancy may be due to species-specific differences in the role of GnRH in follicular development or in the ovarian proteins that may interact to regulate ovarian GnRH expression. Alternatively, it is possible that an ovarian GnRH repressor element exists in the human GnRH promoter as well but has not yet been identified. Further studies may reveal the identity of the specific repressor proteins and the specific physiologic states in which ovarian GnRH expression is increased.
The function of ovarian GnRH is not known, but there is increasing evidence that GnRH may function in a paracrine or autocrine manner in the ovary. GnRH has been shown to modulate directly ovarian steroidogenesis (27)(28)(29) and follicular development (30). Ovarian GnRH is also thought to have a regulatory role in ovulation (31)(32)(33), luteinization (34), and corpus luteum function (35)(36)(37). In vitro studies also suggest that GnRH may also act as a meiosis-stimulating factor in the rat oocyte (38).
The presence of specific ovarian GnRH-binding sites was first suggested by radioligand binding studies (35,39), and GnRH receptor messenger RNA has since been demonstrated in rat (40,41) and human (42) ovaries. By using RT-PCR, GnRH messenger RNA has been detected in rat (43)(44)(45) and human ovaries (2).
The intra-ovarian location of GnRH is not yet well established in the literature. GnRH messenger RNA has been identified in human luteinized granulosa cells (46) and in human ovarian surface epithelium (47). One in situ hybridization histochemistry study localized GnRH expression to both the theca and granulosa cells of rat ovaries with GnRH messenger RNA being more abundant in the granulosa cells (48). This is in contrast to our immunocytochemistry experiments, which were unable to detect any GnRH peptide in the mouse ovary. Addi- FIG. 4. Model of tissue-specific expression of hypothalamic and ovarian GnRH. This model represents our understanding of tissuespecific expression of hypothalamic and ovarian GnRH. Our data demonstrate that the DNA sequences contained within the proximal Ϫ1005 bp of the mGnRH promoter are sufficient to direct mGnRH gene expression to both the ovary and hypothalamus. Our data also support the presence of an enhancer region for the in vivo expression of hypothalamic mGnRH in the mGnRH promoter region between Ϫ3446 and Ϫ2078 bp. We speculate that in the hypothalamus, specific enhancer proteins interact with the enhancer region to increase mGnRH expression. In the ovary, our data suggest that an ovarian GnRH repressor element may be located in the distal region of the mGnRH promoter between Ϫ3446 and Ϫ2078 bp. We speculate that repressor proteins in the ovary normally interact with this ovarian GnRH repressor element to repress ovarian GnRH expression. It is possible that this may be a mechanism for regulating ovarian GnRH expression. Under certain physiologic situations, these repressor proteins may be released to allow ovarian GnRH expression. tionally, in mice bearing the mGnRH-LUC transgene, luciferase was detected only in the granulosa cells. It is possible that in situ hybridization is a more sensitive detection method than immunocytochemistry. It is also possible, however, that the detection of GnRH messenger RNA may not reflect the pattern of GnRH protein expression.
In both the human and rat ovary, transcription of GnRH is initiated from a more upstream promoter initiation site than used for the transcription of hypothalamic GnRH messenger RNA (2,43). It also appears that a longer GnRH transcript retaining hypothalamic intron sequences is generated in the ovary (43). It is possible that the GnRH peptide produced in the ovary differs from hypothalamic GnRH and has a different protein conformation, which may explain why there are no immunocytochemistry studies of GnRH in the ovary. Our immunocytochemistry studies, using the LR5 anti-GnRH decapeptide antibody, detected GnRH in the hypothalamus, but not in the ovary, of mice bearing the Ϫ1005mGnRH-LUC transgene.
In conclusion, our model of tissue-specific expression of hypothalamic and ovarian GnRH expression is shown in Fig. 4. Our results clearly demonstrate that DNA sequences contained within the proximal Ϫ1005 bp of the mGnRH promoter are sufficient to direct mGnRH gene expression to both the ovary and hypothalamus. Our data also suggest that an enhancer for the in vivo expression of hypothalamic mGnRH is contained in the mGnRH promoter region between Ϫ3446 bp and Ϫ2078 bp. Specific proteins, in the hypothalamus, may interact with DNA sequences in the enhancer region to increase hypothalamic mGnRH expression. Additionally, we have identified a region of the mGnRH promoter, between Ϫ3446 bp and Ϫ2078 bp, that appears to mediate repression of GnRH expression in the ovary. In the ovary, specific repressor proteins interact with the ovarian GnRH repressor element to repress ovarian GnRH expression. Ovarian GnRH expression will occur only when these repressor proteins are released by specific stimuli. Our study is the first to define regions of the mGnRH gene promoter that regulate expression of GnRH in both neural and nonneural tissues in vivo. Further study is needed to elucidate the specific mechanisms by which hypothalamic and ovarian GnRH expression are differentially regulated.