Gonadotropin-Induced Expression of Messenger Ribonucleic Acid for ...

0013-7227/06/$15.00/0 Printed in U.S.A.

Endocrinology 147(10):4713– 4722 Copyright © 2006 by The Endocrine Society doi: 10.1210/en.2005-1575

Gonadotropin-Induced Expression of Messenger Ribonucleic Acid for Cyclooxygenase-2 and Production of Prostaglandins E and F2␣ in Bovine Preovulatory Follicles Are Regulated by the Progesterone Receptor P. J. Bridges, C. M. Komar, and J. E. Fortune Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca New York 14853 Follicular production of prostaglandins (PGs) is essential for ovulation, but the factors mediating gonadotropin-induced secretion of PGE and PGF2␣ remain largely unknown. We tested the hypothesis that gonadotropin-induced changes in progesterone and its receptor (PR) mediate the increase in periovulatory PGs. Heifers were treated with PGF2␣ and GnRH to induce luteolysis and the LH/FSH surge (ovulation occurs ⬃30 h after GnRH). Because there are two increases in intrafollicular progesterone/PR mRNA during the bovine periovulatory period, we first examined the temporal pattern of PG production by follicles collected at 0, 3.5, 6, 12, 18, and 24 h after GnRH. Although PGs did not increase in the follicular fluid until 24 h after GnRH, acute secretion of PGs by follicle wall (theca ⴙ granulosa cells) was initiated by 18 h and had increased manyfold by 24 h after GnRH. In vitro, FSH and LH induced dramatic transient increases in PG production by follicle wall and granulosa, but not theca, cells isolated from

O

VULATION MARKS THE culmination of follicular growth and development and involves a complex cascade of events stimulated by an estradiol-induced surge of gonadotropins. Progesterone and prostaglandins (PGs) are two key players in the ovulatory cascade. Inhibition of progesterone or PGs in vivo blocks ovulation in a variety of species (1– 8). In mice, genetic deletion of the enzyme responsible for the conversion of arachidonic acid to PGH2, cyclooxygenase (COX)-2 or deletion of the progesterone receptor (PR) inhibits or dramatically reduces ovulation (9 –11). The relationship between progesterone and PGs during the periovulatory period is unclear. In PR-deficient mice, human chorionic gonadotropin (hCG)-induced expression of mRNA for COX-2 appears normal (12), whereas in rats progesterone was reported to decrease LH-induced expression of mRNA for COX-2 and PGE secretion by preovulatory follicles in vitro (13). In contrast, experiments with sheep provided evidence that progesterone regulates PGE2-9ketoreductase, the enzyme responsible for the conversion of PGE2 to PGF2␣ (2, 14). Reports of the primary PG required for

First Published Online July 6, 2006 Abbreviations: COX, Cyclooxygenase; DEX, dexamethasone; hCG, human chorionic gonadotropin; MPA, medroxyprogesterone acetate; PG, prostaglandin; PR, progesterone receptor. Endocrinology is published monthly by The Endocrine Society (http:// www.endo-society.org), the foremost professional society serving the endocrine community.

preovulatory follicles (0 h after GnRH). PG accumulation peaked on d 2 of culture, mimicking the secretion pattern after a gonadotropin surge in vivo. In cultures of follicle wall and granulosa cells, the PR antagonist mifepristone (MIFE, 1 ␮M) inhibited LH-induced PG secretion and the progestin medroxyprogesterone acetate (1 or 10 ␮M), but not the glucocorticoid dexamethasone (1 or 10 ␮M), overcame the effect of MIFE on PGs. Semiquantitative RT-PCR revealed that MIFE inhibited LH-induced expression of cyclooxygenase-2 mRNA in granulosa cells in vitro. Again, treatment with medroxyprogesterone acetate overcame the effect of MIFE. Together these results provide strong evidence that periovulatory increases in cyclooxygenase-2 mRNA, PGE, and PGF2␣ are mediated by gonadotropin-induced increases in progesterone/ PR, indicating that in some species there is an important functional relationship between these pathways in the ovulatory cascade. (Endocrinology 147: 4713– 4722, 2006)

ovulation are also inconsistent. Ovulation rate is reduced in mice null mutant for the PGE receptor subtype EP2 (15, 16) but not in mice null mutant for FP, the PGF2␣ receptor (17), whereas experimental evidence points to PGF2␣ as the PG required for ovulation in ewes (2). It is not clear whether the differences and contradictions summarized above are due to differences in experimental conditions and/or methods or whether they reflect true species differences in key mediators of the ovulatory cascade. Cattle are an excellent model for studying the periovulatory period (i.e. the interval between the gonadotropin surge and ovulation) because luteolysis, a follicular phase, and ovulation can be induced by sequential treatment of animals that have a growing dominant follicle with PGF2␣ to induce luteal regression and GnRH (or hCG) to induce (or mimic) the LH/FSH surge (18, 19). This allows the retrieval of follicles at defined stages of the long (about 28 h) interval between the LH/FSH surge and ovulation and the large size of the preovulatory follicle permits the application of multiple treatments in vitro to cells isolated from one follicle. In the studies described herein, we used cattle as a model to investigate the mechanisms that regulate the periovulatory rise in PGs, particularly the potential role of progesterone and its receptor. Sirois and colleagues (19, 20) have shown that there is a dramatic increase in COX-2 and its mRNA in bovine granulosa cells late in the periovulatory period, coincident with increases in PGE and PGF2␣ in follicular fluid at 24 h after hCG or estrus, but the earlier stages of the periovulatory

4713

4714

Endocrinology, October 2006, 147(10):4713– 4722

interval have not been examined as carefully. Studies in our laboratory showed that at 3.5– 6 h after GnRH, shortly after the preovulatory LH/FSH surge, concentrations of progesterone in the follicular fluid and expression of mRNA for the PR in follicular tissue increase transiently, followed by a second increase late in the periovulatory period at 24 h after GnRH (21–23). Because the increases in follicular progesterone and expression of mRNA for the PR precede the dramatic increase in follicular PGs after the ovulatory surge of gonadotropins, we hypothesized that gonadotropin-induced secretion of PGs by bovine follicular cells is mediated through progesterone and the PR. The specific objectives were: 1) to examine the temporal pattern of secretion of PGF2␣ and PGE during the periovulatory period in vivo and determine whether it could be replicated in vitro, 2) to examine the relative contributions of theca vs. granulosa cells and of LH vs. FSH to PG production by preovulatory follicles, and 3) to examine the potential role of progesterone/PR in mediating the effects of gonadotropins on PG production by bovine ovulatory follicles. Materials and Methods Animals Holstein heifers were housed at the research farm and monitored for estrous behavior. Heifers that had exhibited normal and regular estrous cycles were assigned to experimental groups at estrus (d 0). A protocol developed and validated previously (18) was used to induce the development of preovulatory (before the LH/FSH surge) and periovulatory (between the LH/FSH surge and ovulation) follicles. On d 6 (afternoon) or d 7 (morning), heifers were injected im with 25 mg PGF2␣ (Lutalyse; Pharmacia & Upjohn, Kalamazoo, MI) to induce regression of the corpus luteum and thus initiate a follicular phase. Because the first follicular wave of the estrous cycle begins at a more predictable time than succeeding waves, this experimental model ensures that the development of preovulatory follicles is consistent among heifers. Follicular and luteal development were monitored in vivo by daily transrectal ultrasonography and jugular venous blood was collected by daily venipuncture for the determination of concentrations of progesterone to verify regression of the corpus luteum. At 36 h after PGF2␣, the ovary bearing the preovulatory follicle was removed by colpotomy (0 h group), or heifers were injected im with 100 ␮g of an analog of GnRH (Cystorelin; Sanofi Animal Health Inc., Overland Park, KS) to induce the ovulatory surge of gonadotropins and 3.5, 6, 12, 18, or 24 h later; the ovary bearing the periovulatory follicle was removed. In this model, the surge of LH peaks at 2 h after GnRH and follicles ovulate approximately 30 h after GnRH (18). All procedures with animals were approved by the Cornell University Animal Care and Use Committee.

Isolation and culture of follicle wall, granulosa cells, and theca interna Ovaries were transported to the laboratory (⬃10 min) in Eagle’s MEM supplemented with penicillin-streptomycin (50 U-50 ␮g/ml; Life Technologies, Inc., Grand Island, NY). The ovulatory follicle was dissected from the ovary, follicular fluid aspirated, and the follicle wall peeled away from the surrounding stroma. Depending on the experiment, follicle wall (theca interna plus attached granulosa cells), theca interna, and/or isolated granulosa cells were cultured. Within an experiment, all cell types were obtained from the same follicles. Granulosa cells were separated from the theca interna by scraping the theca with a fine glass needle, as described previously (24). The follicle wall and theca interna were cut into small pieces that were then distributed at random (three pieces/well) to 24-well culture plates (Costar, Cambridge, MA) and cultured in 0.5 ml of culture medium [modified Eagle’s MEM (25)]. Granulosa cells were collected by centrifugation, counted with a hemacytometer, and distributed to 24-well (200,000 cells/well; experiments II and IIIa) or six-well (106 cells/well; experiment IV) Primaria culture

Bridges et al. • Progesterone Regulates PG Production

plates (Falcon, Becton Dickinson, Lincoln Park, NJ) and cultured in 0.5 or 2 ml of culture medium, respectively. Treatments were applied to duplicate cultures from each follicle in experiments I-III and to single cultures from each follicle in (experiment IV. Treatments included LH (NIH LH-S26, 100 ng/ml), FSH (NIH FSHS17, 100 ng/ml), the PR antagonist mifepristone (1 ␮m), the synthetic progestogen medroxyprogesterone acetate (MPA, 1 or 10 ␮m), and the synthetic glucocorticoid dexamethasone (DEX, 1 or 10 ␮m). Mifepristone (RU-486) is a reversible PR antagonist (26); the dose used (1 ␮m) is the lowest dose that was maximally effective in pilot experiments. MIFE, DEX, and MPA were obtained from Sigma-Aldrich Corp. (St. Louis, MO). Cultures were maintained at 37 C in humidified modular incubation chambers (Billups-Rothenberg, Del Mar, CA) gassed with 5% CO2-95% air for 72 h (experiment 1), 96 h (experiments II and III), or 36 h (experiment IV). Medium was collected and replaced at 24-h intervals except in experiment I (further described in Results), IIIb, and IV (12-h intervals) and stored frozen for later measurement of PGs. Each experiment was replicated with three (experiment I), four (experiments. II, IIIb, and IV), or five (experiment IIIa) separate follicles on separate occasions.

RIAs for PGE and PGF2␣ Concentrations of PGE and PGF2␣ were determined in duplicate aliquots of culture media (0.5–25 ␮l) without extraction or in etherextracted aliquots of follicular fluid. Briefly, standards ranging from 6.25 to 800 pg/100 ␮l and pools ranging from 12.5 to 200 pg/100 ␮l were prepared from PGE2 and PGF2␣ tromethamine salt (Cayman Chemical, Ann Arbor, MI) in assay buffer (0.1 m PBS, 0.1% gelatin). The antibody to PGE2 was provided by Dr. T. G. Kennedy (University of Western Ontario, London, Ontario, Canada) and has been described previously (27). Because this antibody has 46% cross-reactivity with PGE1, total concentrations of PGE are referred to throughout the text. The antibody to PGF2␣ (Assay Designs, Inc., Ann Arbor MI) was polyclonal, produced by immunization of sheep with PGF2␣ conjugated to bovine albumin, and used at a final dilution of 1:15,000. The PGF2␣ antibody was tested for cross-reactivities and determined to cross-react with PGA1 (⬍0.5%), PGA2 (⬍0.5%), PGB1 (⬍0.5%), PGB2 (⬍0.5%), PGD2 (3.1%), PGE1 (⬍0.5%), PGE2 (1.6%), PGF1␣ (3.1%), 15-keto PGF2␣ (⬍0.5%), and 13,14dihydro-15-keto PGF2␣ (⬍0.5%) (all from Cayman Chemical). 3H(N) PGE2 (5, 6, 8, 11, 12, 14, 15) and 3H(N) PGF2␣ (5, 6, 8, 9, 11, 12, 14, 15) were obtained from NEN Life Science Products (Boston, MA) and used at a final concentration of 0.01 ␮Ci/reaction. The sensitivity of the assays for PGE and PGF2␣ was approximately 25 pg/tube. Standards, pools, and samples were incubated overnight at 4 C with 3H(N)-PGE2 or -PGF2␣ and appropriate antibody in assay buffer. Bound and free ligand were separated by a 10-min incubation at 4 C with 1 ml dextran-coated activated charcoal (Sigma-Aldrich), followed by centrifugation for 15 min at 3000 rpm. The supernatant was decanted and radiolabeled, and PGs were counted in a liquid scintillation counter. The inter- and intraassay coefficients of variation were 13.8 and 4.5% for PGE and 10.8 and 3.9% for PGF2␣.

Semiquantitative RT-PCR Total RNA was extracted from freshly isolated granulosa cells and a placentome from a pregnant cow (positive control) using TRIzol reagent (Invitrogen Corp., Carlsbad, CA) and from cultured granulosa cells using Absolutely RNA microprep kit (Stratagene, La Jolla, CA) according to the manufacturer’s directions. Samples of total RNA were treated with amplification grade DNase-I to control for possible contamination with genomic DNA and then reverse transcribed to cDNA using Superscript II RNase H⫺ reverse transcriptase and random primers (Invitrogen). Samples of first strand cDNA were amplified with Taq DNA polymerase (Invitrogen) using pairs of primers specific for COX-2 (5⬘-GTCTGGAACAACTGCTCATCGC-3⬘, 5⬘-CACAGTGCACTACATACTTACCC-3⬘) (28) and the18S subunit of rRNA (5⬘-GCTCGCTCCTCTCCTACTTG-3⬘, 5⬘-GATCGGCCCGAGGTTATCTA-3⬘) (29). For COX-2, amplification consisted of 37 cycles at 94 C for 30 sec, 59 C for 60 sec, and 72 C for 2 min, followed by a final 5 min extension at 72 C. For 18S rRNA, amplification consisted of 17 cycles at 94 C for 30 sec, 55 C for 30 sec, and 72 C for 60 sec, followed by a final 5-min extension at 72 C. The numbers


of cycles of amplification for COX-2 and 18S rRNA were based on the results of amplification curves and were within the linear range of those curves. PCR products were electrophoresed in 2% agarose/Tris borate EDTA gels containing ethidium bromide and the intensity of ethidium bromide staining was digitally quantified under UV light (Kodak Image Station 440CF; Eastman Kodak Co., Rochester, NY). Relative levels of mRNA for COX-2 were calculated by dividing the intensity of the respective band by the corresponding 18S rRNA band. PCR products were purified using a gel purification kit (QIAGEN Inc., Valencia, CA) according to the manufacturer’s directions and sequenced using BIG Dye Terminator chemistry with AmpliTaq-FS DNA polymerase (Applied Biosystems Automated 3730xl DNA Analyzer, Biotechnology Resource Center, Cornell University, Ithaca, NY). The integrity and identity of the PCR product sequences were verified using the BLAST database (www.ncbi.nlm.nih.gov/blast) (30).

Statistical analyses Data sets were first tested for homogeneity of variance by Hartley’s test (31). Heterogeneity was detected in all data sets and hence data were log (base 10) transformed before statistical analysis. However, for ease of comprehension, means ⫾ sem of nontransformed data are depicted in the figures. The concentration of PGs in follicular fluid and levels of mRNA for COX-2 were analyzed by ANOVA using the general linear model of SAS (SAS Institute, Inc., Cary, NC). The concentration of PGs in culture medium was analyzed by repeated measures ANOVA. If differences were detected, Duncan’s multiple range test was used to determine which means differed. To simplify presentation, the data shown in some of the figures (see Figs. 3 and 6) were summed over time before analysis.

Results Experiment I. Production of PGs by pre- and periovulatory follicles: in vivo vs. in vitro

In vivo. The concentrations of PGE and PGF2␣ in the follicular fluid of preovulatory follicles collected 36 h after PGF2␣ (0 h after GnRH), and periovulatory follicles collected at 3.5, 6, 12, and 18 h after GnRH were low and did not differ (P ⬎ 0.05; Fig. 1). At 24 h after GnRH the concentration of both PGE and PGF2␣ in the follicular fluid of periovulatory follicles had increased manyfold, compared with all earlier times examined (P ⬍ 0.05), as expected (19, 20). To determine whether concentrations of PGs in follicular fluid reflect secretion by follicular cells, accumulation of PGF2␣ in culture medium

FIG. 1. Concentrations of PGE and PGF2␣ in the follicular fluid of preovulatory follicles collected 36 h after PGF2␣ (0 h after GnRH) and periovulatory follicles collected at 3.5, 6, 12, 18, and 24 h after heifers were injected with GnRH to induce the LH/FSH surge. Data are means ⫾ SEM of three follicles per time point. For each PG, values with different superscripts differ (P ⬍ 0.05).


4715

after 0 – 4 and 4 – 8 h of culture (i.e. acute secretion) was assessed to provide an additional estimate of in vivo production of PGs relative to the injection of GnRH. Consistent with the results for follicular fluid, pieces of follicle wall from periovulatory follicles collected 24 h after GnRH secreted much more PGF2␣ after both 4 and 8 h of culture than follicular tissue obtained at earlier time points in vivo (Fig. 2). Interestingly, secretion of PGF2␣ by pieces of follicle wall collected at 3.5 and 18 h after GnRH was higher after 4 and 8 h of culture, compared with secretion by tissue collected at 0, 6, and 12 h in vivo (P ⬍ 0.05), suggesting that follicular cells from periovulatory follicles are producing PGs in vivo earlier than 24 h after GnRH. The higher acute secretion of PGF2␣ by pieces of follicle wall collected at 3.5 h suggests a transient period of follicular PG secretion early in the periovulatory interval that is not reflected in the follicular fluid. In vitro. Follicle cells from preovulatory follicles (0 h after GnRH) secreted very low amounts of PGF2␣ throughout the 3-d culture period. The pattern of secretion of PGs by pieces of follicle collected from periovulatory follicles was biphasic (Fig. 3), but the time of peak secretion varied, depending on the time in vivo when the follicles were obtained. Peak secretion occurred on the second day of culture when follicles were obtained early in the periovulatory period (3.5 or 6 h after GnRH) and on the first day of culture when they were obtained later (12, 18, and 24 h; Fig. 3). Overall, PGF2␣ secretion was initiated earlier in vitro and total quantities secreted were greater as periovulatory follicles were collected progressively later in vivo. During the first 24 h of culture, secretion of PGF2␣ by pieces of follicle wall from periovulatory follicles collected 24 h after GnRH was much greater than at all other in vivo time points examined (P ⬍ 0.05). During the second 24 h of culture, secretion of PGF2␣ by pieces of follicle wall collected at 6 and 12 h after GnRH was greater than by follicle wall collected at both earlier and later

FIG. 2. Secretion of PGF2␣ by pieces of follicle wall from preovulatory follicles collected at 0 h after GnRH and periovulatory follicles collected at 3.5, 6, 12, 18, and 24 h after GnRH in vivo and cultured for 8 h in medium alone. Data are the means ⫾ SEM of duplicate cultures from each of three follicles per time point. Values with different superscripts differ (P ⬍ 0.05).

4716



Experiment III. Effect of a PR antagonist on secretion of PGs induced by gonadotropins in vitro or in vivo

To test the hypothesis that gonadotropin induced production of PGs, an essential component of the bovine ovulatory cascade (7), is mediated by progesterone acting through the PR, we first treated follicular cells obtained at 0 h after GnRH (i.e. before the endogenous gonadotropin surge) with progestins. Neither progesterone nor the synthetic progestin MPA had any effect on production of PGs in vitro (data not shown). However, it seemed reasonable that the cells would not respond to progestins without exposure to luteinizing doses of gonadotropins to induce the PR. Therefore, we next examined the effects of the PR antagonist mifepristone (RU486) on PG secretion in vitro in response to treatment with LH in vitro (experiment IIIa) or in response to the LH/FSH surge in vivo (experiment IIIb). FIG. 3. Secretion of PGF2␣ by pieces of follicle wall from preovulatory follicles collected at 0 h after GnRH and periovulatory follicles collected at 3.5, 6, 12, 18, and 24 h after GnRH in vivo and cultured for 72 h in medium alone. Data are the means ⫾ SEM of duplicate cultures from each of three follicles per time point. Means were compared within each time in vitro (0 –24, 24 – 48, or 48 –72 h) across time in vivo. Within a time period in vitro, values without a common superscript differ (P ⬍ 0.05).

times in vivo (P ⬍ 0.05). From 48 –72 h of culture, maximal secretion of PGF2␣ was observed by pieces of follicle wall collected 3.5 h after GnRH (P ⬍ 0.05). Experiment II. Secretion of PGs by follicular cells of preovulatory follicles in vitro

The next objective was to determine the relative contributions of theca and granulosa cells to PG secretion and the relative roles of LH and FSH in stimulating PG production by follicle cells by culturing granulosa cells and pieces of follicle wall and theca interna from preovulatory follicles (i.e. not exposed to the gonadotropin surge in vivo) with luteinizing doses (100 ng/ml) of gonadotropins in vitro. In the absence of gonadotropins, secretion of PGs by follicle wall, theca interna, and isolated granulosa cells was very low (Fig. 4). The response to gonadotropins by follicle wall and granulosa cells was dramatic. Treatment with 100 ng/ml LH or FSH transiently, yet markedly and similarly, stimulated secretion of PGE and PGF2␣ in vitro (P ⬍ 0.05). Peak production of PGE and PGF2␣ by gonadotropin-treated pieces of follicle wall or granulosa cells was observed between 24 and 48 h in vitro. Secretion of PGE and PGF2␣ by pieces of theca interna was also transiently increased by treatment with 100 ng/ml LH (P ⬍ 0.05); however, concentrations were very low relative to LH-induced production of PGs by granulosa cells or pieces of follicle wall. These results indicate that treatment of follicle cells from preovulatory follicles with gonadotropins in vitro can be used as an in vitro model for studying the mechanisms that subserve the gonadotropin-induced production of PGs during the periovulatory period, and this model was then used for experiments III and IV, as detailed below.

Gonadotropin treatment in vitro. In experiment IIIa the effects of mifepristone on secretion of PGs by pieces of follicle wall and isolated granulosa cells were determined using tissue from preovulatory follicles collected at 36 h after PGF2␣ (i.e. time 0 after GnRH). Pieces of follicle wall (theca interna plus attached granulosa cells) were used in addition to isolated granulosa cells to compare the responses of granulosa cells cultured with their adjacent thecal layer vs. isolated granulosa cells and because mRNA for the PR has been localized to both follicular cell types in cattle (22). Because experiment II did not reveal the theca interna as a major site of PG production, pieces of theca interna were not cultured in this and the following experiments. As expected based on the results of experiment II, treatment of pieces of follicle wall with a luteinizing dose of LH induced a marked but transient increase in secretion of both PGE and PGF2␣ in vitro (P ⬍ 0.05; Fig. 5, A and B). The stimulatory effect of LH on secretion of PGE and PGF2␣ by pieces of follicle wall was completely inhibited by treatment with the PR antagonist mifepristone. Concurrent treatment with the synthetic progestogen (MPA) completely overcame the inhibition by mifepristone of LHinduced secretion of both PGE and PGF2␣ in vitro. In contrast, concurrent treatment with mifepristone and the synthetic glucocorticoid DEX had no effect, indicating the specificity of mifepristone for the PR. Treatment with MPA or DEX alone did not induce secretion of PGE and PGF2␣ in vitro (P ⬎ 0.05, data not shown). A similar response to these treatments was observed in cultures of granulosa cells (Fig. 5, C and D). The luteinizing dose of LH induced a transient increase in secretion of PGE and PGF2␣ in vitro (P ⬍ 0.05). Mifepristone inhibited LH-induced secretion of PGE and PGF2␣ and MPA, but not DEX, overcame the inhibitory effect of mifepristone on LH-induced secretion of PGs. Similar to the results for follicle wall cultures, treatment with MPA or DEX alone did not stimulate secretion of PGs by granulosa cells in vitro (P ⬎ 0.05, data not shown). Gonadotropins in vivo. The results of experiment IIIa strongly suggested that LH-induced secretion of PGE and PGF2␣ in vitro is mediated through progesterone and the PR, but the preovulatory surge of gonadotropins induces a cascade of changes that may not be fully replicated by treatment with a luteinizing dose of LH in vitro. Therefore, we next tested the



4717

FIG. 4. Secretion of PGE and PGF2␣ by pieces of follicle wall (A and B), granulosa cells (C and D), and theca interna (E and F) from preovulatory follicles collected 36 h after PGF2␣ (0 h after GnRH) and cultured for 96 h with 0 or 100 ng/ml LH or FSH (follicle wall and granulosa cells) or 0 or 100 ng/ml LH (theca interna). Data are the means ⫾ SEM of duplicate cultures from each of four preovulatory follicles. In each panel, within a time in vitro values with different superscripts differ (P ⬍ 0.05).

effects of the PR antagonist on pieces of follicle wall from periovulatory follicles collected 4 h after the peak of the LH/FSH surge in vivo (6 h after GnRH), when the concentration of progesterone in the follicular fluid and expression of mRNA for the PR are elevated (22, 23). Because follicle wall and granulosa cells were affected similarly by the treatments in experiment IIIa, only pieces of follicle wall (which represent a more intact follicular structure) were used in experiment IIIb. Pieces of follicle wall from periovulatory follicles collected 6 h after GnRH in vivo were cultured with the treatments described for experiment IIIa above, except that LH was not used. As expected (Fig. 2), secretion of PGE and PGF2␣ by pieces of follicle wall from periovulatory follicles collected 6 h after GnRH in vivo and cultured in medium alone was low during the first 24 h of culture and increased markedly between 24 and 48 h in vitro (Fig. 6). Similar to LH- or FSH-induced secretion of PGs from follicular cells collected at 0 h after GnRH (Figs. 4 and 5), secretion of PGE and PGF2␣ by pieces

of follicle wall collected at 6 h after GnRH was transient in nature, with accumulation of PGs in culture medium declining during the later times in culture. The stimulatory effect of the gonadotropin-surge in vivo on secretion of PGE and PGF2␣ in vitro was completely inhibited by treatment with mifepristone (P ⬍ 0.05). Concurrent treatment with 10 ␮m MPA, but not DEX, overcame the inhibition by mifepristone of the secretion of PGs in vitro. The lower dose of MPA (1 ␮m), which was equally effective in overcoming the effect of the PR antagonist on LH-induced PGs from preovulatory follicles (Fig. 5), was only partially able to overcome the inhibitory effect of mifepristone on PG secretion from periovulatory follicles between 24 and 48 h in vitro, the peak period of PG secretion in vitro (Fig. 6). Treatment with 1 ␮g/ml (⬃3 ␮m) progesterone was also partially effective in overcoming the inhibition of PG secretion by mifepristone, restoring PGs to a level intermediate between that observed after treatment with 1 and 10 ␮m of the synthetic progestogen (data not shown). As expected, treatment with MPA or DEX alone did

4718



FIG. 5. Secretion of PGE and PGF2␣ by pieces of follicle wall (A and B) and granulosa cells (C and D) from preovulatory follicles collected at 0 h after GnRH and cultured for 96 h in medium alone or with LH (100 ng/ml), LH ⫹ mifepristone (MIFE, 1 ␮M), LH ⫹ MIFE ⫹ MPA (1 or 10 ␮M), and LH ⫹ MIFE ⫹ DEX (1 or 10 ␮M). A–D, Accumulation of PGs during the first 72 h of culture; treatment effects from 72 to 96 h were similar to 48 –72 h, but absolute values were very low (data not shown). Data are the means ⫾ SEM of duplicate cultures from each of five preovulatory follicles. nd, Not detectable. In each panel, within a time in vitro values with no common superscript differ (P ⬍ 0.05).

not affect secretion of PGE or PGF2␣ in vitro (P ⬎ 0.05; data not shown). Experiment IV. Effect of a PR antagonist on LH-induced expression of mRNA for COX-2

In experiment IV we began to explore the mechanisms by which progesterone/PR affects PG secretion by testing the hypothesis that progesterone/PR mediates LH-induced expression of mRNA for COX-2 in bovine granulosa cells by determining the effects of a luteinizing dose of LH, LH ⫹ mifepristone, and LH ⫹ mifepristone ⫹ MPA on levels of mRNA for COX-2 in granulosa cells from preovulatory follicles (collected 36 h after PGF2␣). Expression of mRNA for COX-2 was not detectable in granulosa cells freshly isolated from preovulatory follicles (0 h control) or after 12 h in vitro (Fig. 7). Compared with control cultures without LH, treatment of granulosa cells with LH increased mRNA for COX-2 by 2.4- and 4-fold after 24 and 36 h in vitro, respectively (P ⬍ 0.05). Concurrent treatment with mifepristone completely inhibited LH-induced expression of mRNA for COX-2, and

MPA overcame the inhibitory effect of mifepristone. The effects of these treatments on COX-2 mRNA were reflected in the concentrations of PGE and PGF2␣ in culture media collected at 36 h in vitro (data not shown). As in the previous experiments, LH increased the secretion of PGE and PGF2␣ above control levels and concurrent treatment with mifepristone inhibited this effect (P ⬍ 0.05). Treatment with MPA overcame the inhibitory effect of mifepristone on LH-induced secretion of PGs. Discussion

Although ovulation is essential for successful reproduction, the mechanisms by which the gonadotropin surge induces ovulation are far from completely understood. The major focus of the current work was the relationship between progesterone and prostaglandins, two essential components of the ovulatory cascade (2, 7, 9 –11). The results provide strong evidence that gonadotropin-induced expression of mRNA for COX-2 and production of PGE and PGF2␣ are regulated by progesterone acting through the PR. To our



4719

FIG. 6. Secretion of PGE (A) and PGF2␣ (B) by pieces of follicle wall from periovulatory follicles collected at 6 h after GnRH in vivo and cultured for 96 h in medium alone or with mifepristone (MIFE, 1 ␮M), MIFE ⫹ MPA (1 or 10 ␮M), or MIFE ⫹ DEX (10 ␮M). A and B, Accumulation of PGs during the first 72 h of culture; treatment effects from 72 to 96 h were similar to 48 –72 h, but absolute values were very low (data not shown). Data are the means ⫾ SEM of duplicate cultures from each of four periovulatory follicles. In each panel, within a time in vitro values with different superscripts differ (P ⬍ 0.05).

knowledge this is the first experimental evidence showing a direct involvement of the progesterone receptor in the upregulation of PG production by follicular cells, thus causally linking two primary mediators of the gonadotropin-induced ovulatory cascade. Previous studies have provided evidence that genetic deletion of either the PR or COX-2 prevents ovulation (9 –11). Interestingly, in at least three species (rats, cattle, horses), the induction of mRNA for COX-2 occurs at around 10 h before follicular rupture, rather than at a specific time after the gonadotropin surge (reviewed in Ref. 32), whereas in rhesus monkeys mRNA for COX-2 is induced much earlier relative to the time of ovulation (around 28 h before rupture), but COX-2 protein and an increase in PGs in the follicular fluid were not detected until 4 –16 h before ovulation (33). Previous experiments in our lab revealed two separate increases in both progesterone in the follicular fluid and levels of mRNA for PR, with one peak occurring early in the periovulatory period at 3.5– 6 h after GnRH and the second at 24 h after GnRH, much closer to ovulation (21, 22, 34). Thus, in experiment I we examined temporal patterns of PG production in vivo, between the injection of GnRH and the anticipated time of ovulation, to determine their relationship to temporal changes in progesterone/PR. The dramatic increase in both PGF2␣ and PGE in the follicular fluid collected 24 h after GnRH, about 6 h before the expected time of ovulation, is consistent with previous reports on PGs in follicular fluid (19, 20). Because it is not clear to what extent concentrations of substances in the follicular fluid reflect their production by granulosa and theca cells, secretion of PGs by follicular wall tissue was examined during the first few hours in vitro, when the cells are still under the influence of the in vivo environment, and those results suggest that the

periovulatory increase in PG secretion begins between 12 and 18 h after GnRH. The fact that secretion of PG by follicular tissue collected 18 h after GnRH tripled during the second 4 h of culture, compared with the first 4 h, suggests that this increase is indeed the initial stage of the periovulatory rise. It is intriguing that there was a smaller, but significant, transient increase in acute PG secretion by follicular tissue collected 3.5 h after GnRH because previous studies revealed 7- and 5-fold increases in mRNA for PR in follicular tissue and in progesterone in the follicular fluid at 3.5 h vs. 0 h after GnRH (21, 22). This is consistent with results of Acosta et al. (35) showing a small, transient increase in PGF2␣ in the ovarian venous effluent before a large increase closer to the time of ovulation. The physiological significance of the early transient increase in PG in the follicular fluid at 3.5 h after GnRH, and any potential functional relationship to the contemporaneous transient increases in progesterone and PR, remain to be determined. Studies in our laboratory have shown a complex pattern of expression of mRNA for receptors for PGE (EP2, EP3, and EP4) and PGF2␣ (FP) in both theca and granulosa cells, including up-regulation of mRNA for EP2 in granulosa cells and for FP and EP4 in theca cells early in the periovulatory period at 6 h after GnRH (36). Those results suggest a complex pattern of PG action throughout the periovulatory period. Analysis of the patterns of secretion of PGs by pre- and periovulatory follicular tissue over 3 d of culture suggested that prior exposure to the LH/FSH surge in vivo is sufficient to reproduce the effects of the surge in vitro qualitatively but not quantitatively. Regardless of the time that the tissue was isolated, there was always a robust but transient increase in PG accumulation in the culture medium. As would be ex-

4720


FIG. 7. Semiquantitative RT-PCR showing that the inhibition of LHinduced mRNA for COX-2 by mifepristone is overcome by concurrent progestogen treatment. A, Photographs of ethidium bromide staining in representative agarose gels show the pattern of expression of COX-2 mRNA and 18S rRNA in granulosa cells of preovulatory follicles collected at 0 h after GnRH and cultured for 0, 12, 24, or 36 h in medium alone or with LH (100 ng/ml), LH ⫹ mifepristone (MIFE, 1 ␮M), LH ⫹ MIFE ⫹ MPA (10 ␮M). na, Not applicable. B, Relative levels of mRNA for COX-2, calculated by correcting for the intensity of the band for the 18S subunit of rRNA for each sample. Data are the means ⫾ SEM of cultures from each of four preovulatory follicles. nd, Not detectable. Within a time in vitro values with different superscripts differ (P ⬍ 0.05).

pected if secretion in vitro mirrors that in vivo, the later during the periovulatory period the follicle was removed from the animals, the earlier was the rise in PGs and the earlier the termination of the rise. However, the quantity of PGs secreted in vitro was affected by the time of isolation of the follicle, with follicular tissue obtained at 3.6 and 6 h, 12 and 18 h, and 24 h secreting low, intermediate, and large amounts of PG, respectively, over the 3-d culture. Whether these quantitative differences are due to the absence in vitro of optimal levels of substrates for PG synthesis and/or additional regulatory signals that modify overall expression of PG synthetic enzymes is yet to be determined. Comparisons of follicular wall tissue with isolated follicular cell types suggested that the granulosa cells are the source of the PGs produced by follicular wall preparations containing both cell types. This is not surprising because the laboratory of Sirois and colleagues reported induction of COX-2 and its mRNA in granulosa, but not theca, cells late in the periovulatory period, at 24 h after an endogenous surge of LH/FSH (20), or after injection of hCG as an ovu-


latory stimulus (19). However, Murdoch et al. (37) reported an increase in mRNA for COX in the theca interna of periovulatory ovine follicles 8 –24 h after injection of GnRH, suggesting that theca interna cells of ruminant follicles may produce PGs early in the periovulatory interval. In the current studies, secretion of PGs by theca interna in vitro was very low, compared with granulosa cells or follicle wall. Although there was a small stimulatory effect of LH on the secretion of PGs by theca interna, a similar effect was observed in additional cultures in which theca was treated with a luteinizing dose of FSH (data not shown), suggesting that the increased concentration of PGs was due to very minor contamination of the theca preparations with granulosa cells. In contrast, immunostaining for COX-2 in rhesus money periovulatory follicles was positive in theca as well as granulosa cells (33). Ovulation can be induced in cattle by treatment with either GnRH, which induces an LH/FSH surge, or with hCG, which has only LH bioactivity (38). An increase in intrafollicular prostaglandin production has been observed in cattle after injection of either hCG (19) or GnRH (current study) and after a spontaneous LH/FSH surge (20). Thus, it was not known whether FSH, as well as LH, plays a role in the periovulatory increase in PG production. The results of the current in vitro studies showed that luteinizing doses of the two gonadotropins have equivalent ability to stimulate secretion of both PGE and PGF2␣ by follicle wall and granulosa cells. Over 4 d of culture, both gonadotropins stimulated a large transient increase in PG production that peaked during the second day of culture and then declined. Periovulatory production of PGs is an essential component of the ovulatory cascade in cattle (1, 7) as well as in other mammalian species that have been studied (3– 6, 8). The hypothesis that gonadotropin-induced production of PGs is mediated by progesterone acting through the PR was strongly supported by the finding that addition of exogenous progestins (MPA or progesterone) overcame the inhibitory effects of the PR antagonist mifepristone on PG production in vitro by both follicle wall preparations and granulosa cells. Whether follicular cells were isolated before the endogenous surge and exposed to LH in vitro or were exposed to an LH/FSH surge in vivo and isolated 6 h later, mifepristone completely inhibited the positive effects of gonadotropins on PG production and the synthetic progestin MPA restored PGs to levels not significantly different from controls. In the experiments with granulosa cells obtained 6 h after GnRH, progesterone was as effective as MPA at reversing the inhibitory effects of mifepristone on PG production, indicating that this effect was not specific to the synthetic progestin. The specificity of the effects of mifepristone was demonstrated by the failure of the glucocorticoid DEX to mimic the effects of progestin in overcoming the inhibition of PR effects. It is interesting that, in the experiments with follicular cells obtained before the endogenous gonadotropin surge, both doses of MPA (1 and 10 ␮m) completely overcame the inhibition of PGs by mifepristone, whereas when follicular cells were isolated 6 h after GnRH administration, only the higher dose of MPA restored PGs to control levels. This suggests that exposure of follicular cells to the LH/FSH surge in vivo in advance of their exposure to mifepristone and


MPA in vitro modulates some aspects(s) of their environment that affects their responses in vitro. Understanding how this occurs will require further experimentation. Whether follicular cells were obtained before or after the LH/FSH surge, neither progesterone nor the progesterone agonist MPA alone stimulated the production of PGs by follicular wall tissue or isolated granulosa cells. This may seem contradictory to the conclusion that progesterone/PR is an obligatory mediator between gonadotropins and periovulatory PG production. However, when follicular cells are obtained before the gonadotropin surge, they lack PR (22) and thus would not be capable of responding to progestin treatment. When bovine follicular cells were obtained at 3.5 or 6 h after GnRH, induction of a transient increase in mRNA for PR had already occurred (22), coincident with a transient rise in progesterone in the follicular fluid (21). In fact, the concentrations of progesterone in the follicular fluid during this initial transient increase (about 175 ng/ml or 0.6 ␮m) are not significantly different from the concentration observed during the second increase at 24 h after GnRH, close to the time of ovulation (21, 23). Taken together, these results suggest that by 6 h after the LH/FSH surge maximally effective concentrations of endogenous progesterone are present, and this may explain why addition of exogenous progestins produces no additional effects. When granulosa cells were obtained before the LH surge (0 h after GnRH) levels of mRNA for COX-2 were nondetectable, but a luteinizing dose of LH induced the mRNA between 12 and 24 h of culture. This is consistent with the timing of the increase in PGs observed in cultures treated with LH and with the induction of mRNA for COX-2 in preovulatory follicles of cattle treated with hCG in vivo or exposed to an endogenous LH surge (19, 20). In the current experiments, the PR antagonist mifepristone inhibited the LH-stimulated rise in COX-2 mRNA and concurrent treatment with the synthetic progestogen MPA completely overcame the inhibition of expression of COX-2 mRNA, restoring it to the levels observed with LH alone. These results show that progesterone/PR mediate the effects of LH on PG production by increasing mRNA for this key enzyme in the pathway of prostaglandin synthesis. In summary, the induction of the intrafollicular periovulatory increase in PGE and PGF2␣ by the gonadotropin surge in vivo was replicated in bovine follicular wall and granulosa cells in vitro by luteinizing doses of LH or FSH. Experimental evidence presented herein supports an obligatory role for progesterone acting through the PR in the induction of mRNA for COX-2 and the increase in PGs. Because ovulation is of such critical importance to the survival of a species, different mammalian species may well have evolved different variations on the complex cascade of intrafollicular changes that allow ovulation to occur successfully. Indeed, hCG induced an increase in mRNA for COX-2 in the ovaries of pregnant mare serum gonadotropin-primed mice null mutant for the PR (12), suggesting that up-regulation of progesterone/PR and of COX-2/prostaglandins by gonadotropins occurs via separate intracellular pathways in that species. The current results illustrate the importance of using various animal models to study periovulatory events, espe-


4721

cially models of practical significance, such as domestic animals and primates. Acknowledgments We thank D. Bianchi for care of animals and surgical procedures, K. Sierzega for technical assistance, Dr. M. Roberson for critical reading of the manuscript, Dr. T. G. Kennedy for kindly providing the antibody to PGE, and the National Hormone and Pituitary Program (supported by the National Institute of Diabetes and Digestive and Kidney Diseases) for providing LH and FSH. Received December 13, 2005. Accepted June 26, 2006. Address all correspondence and requests for reprints to: Dr. J. E. Fortune, Department of Biomedical Sciences, Cornell University, Ithaca, New York 14853. E-mail: [email protected]. This research was supported by National Institutes of Health Grant HD41592 (to J.E.F.) and a Lalor Foundation Fellowship (to P.J.B.). Current address for P.J.B.: Department of Clinical Sciences, University of Kentucky, Lexington, Kentucky 40536. Current address for C.M.K.: Department of Animal Science, Iowa State University, Ames, Iowa 50011. Author disclosure summary: the authors have nothing to declare.

References 1. Algire JE, Srikandakumar A, Guilbault LA, Downey BR 1992 Preovulatory changes in follicular prostaglandins and their role in ovulation in cattle. Can J Vet Res 56:67– 69 2. Murdoch WJ, Peterson TA, Van Kirk EA, Vincent DL, Inskeep EK 1986 Interactive roles of progesterone, prostaglandins, and collagenase in the ovulatory mechanism of the ewe. Biol Reprod 35:1187–1194 3. Barbosa I, Maia HJ, Lopes T, Elder MG, Coutinho EM 1979 Effect of indomethacin on prostaglandin and steroid synthesis by the marmoset ovary in vivo. Int J Fertil 24:142–144 4. Killick S, Elstein M 1987 Pharmacologic production of luteinized unruptured follicles by prostaglandin synthetase inhibitors. Fertil Steril 47:773–777 5. Mikuni M, Pall M, Peterson CM, Peterson CA, Hellberg P, Bra¨nnstro¨m M, Richards JS, Hedin L 1998 The selective prostaglandin endoperoxide synthase-2 inhibitor, NS-398, reduces prostaglandin production and ovulation in vivo and in vitro in the rat. Biol Reprod 59:1077–1083 6. Priddy AR, Killick SR, Elstein M, Morris J, Sullivan M, Patel L, Elder M 1990 The effect of prostaglandin synthetase inhibitors on human preovulatory follicular fluid prostaglandin, thromboxane, and leukotriene concentrations. J Clin Endocrinol Metab 71:235–242 7. De Silva M, Reeves JJ 1985 Indomethacin inhibition of ovulation in the cow. J Reprod Fertil 75:547–549 8. Duffy DM, Stouffer RL 2002 Follicular administration of a cyclooxygenase inhibitor can prevent oocyte release without alteration of normal luteal function in rhesus monkeys. Hum Reprod 17:2825–2831 9. Dinchuk JE, Car BD, Focht RJ, Johnston JJ, Jaffee BD, Covington MB, Contel NR, Eng VM, Collins RJ, Czerniak PM, Gorry SA, Trzaskos M 1995 Renal abnormalities and an altered inflammatory response in mice lacking cyclooxygenase II. Nature 378:406 – 409 10. Lim H, Paria BC, Das SK, Dinchuk JE, Langenbach R, Trzaskos JM, Dey SK 1997 Multiple female reproductive failures in cyclooxygenase 2-deficient mice. Cell 91:197–208 11. Lydon JP, DeMayo FJ, Conneely OM, O’Malley BW 1996 Reproductive phenotypes of the progesterone receptor null mutant mouse. J Steroid Biochem Mol Biol 56:66 –77 12. Robker RL, Russell DL, Espey LL, Lydon JP, O’Malley BW, Richards JS 2000 Progesterone-regulated genes in the ovulation process: ADAMTS-1 and cathepsin L proteases. Proc Natl Acad Sci USA 97:4689 – 4694 13. Hedin L, Eriksson A 1997 Prostaglandin synthesis is suppressed by progesterone in rat preovulatory follicles in vitro. Prostaglandins 53:91–106 14. Murdoch WJ, Farris ML 1988 Prostaglandin E2-9-ketoreductase activity of preovulatory ovine follicles. J Anim Sci 66:2924 –2929 15. Kennedy CRJ, Zhang Y, Brandon S, Guan Y, Coffee K, Funk CD, Magnuson MA, Oates JA, Breyer MD, Breyer RM 1999 Salt-sensitive hypertension and reduced fertility in mice lacking the prostaglandin EP2 receptor. Nat Med 5:217–220 16. Hizaki H, Segi E, Sugimoto Y, Hirose M, Saji T, Ushikubi F, Matsuoka T, Noda Y, Tanaka T, Yoshida N, Narumiya S, Ichikawa A 1999 Abortive expansion of the cumulus and impaired fertility in mice lacking the prostaglandin E receptor subtype EP2. Proc Natl Acad Sci USA 96:10501–10506 17. Sugimoto Y, Yamasaki A, Segi E, Tsuboi K, Aze Y, Nishimura T, Oida H, Yoshida N, Tanaka T, Katsuyama M, Hasumoto K, Murata T, Hirata M,

4722

18.

19. 20. 21.

22. 23. 24. 25. 26. 27. 28.


Ushikubi F, Negishi M, Ichikawa A, Narumiya S 1997 Failure of parturition in mice lacking the prostaglandin F receptor. Science 277:681– 683 Komar CM, Berndtson AK, Evans ACO, Fortune JE 2001 Decline in circulating estradiol during the periovulatory period is correlated with decreases in estradiol and androgen, and in messenger RNA for P450 aromatase and P450 17␣-hydroxylase, in bovine preovulatory follicles. Biol Reprod 64:1797–1805 Sirois J 1994 Induction of prostaglandin endoperoxide synthase-2 by human chorionic gonadotropin in bovine preovulatory follicles in vivo. Endocrinology 135:841– 848 Liu J, Carrie`re PD, Dore´ M, Sirois J 1997 Prostaglandin G/H synthase-2 is expressed in bovine preovulatory follicles after the endogenous surge of luteinizing hormone. Biol Reprod 57:1524 –1531 Komar CM 1998 Effects of the luteinizing hormone surge on the capacity of preovulatory follicles to secrete steroids and oxytocin and the molecular mechanisms that mediate periovulatory changes in steroidogenesis in cattle. Dissertation, Cornell University Jo M, Komar CM, Fortune JE 2002 Gonadotropin surge induces two separate increases in messenger RNA for progesterone receptor in bovine preovulatory follicles. Biol Reprod 67:1981–1988 Jo M, Fortune JE 2003 Changes in oxytocin receptor in bovine preovulatory follicles between the gonadotropin surge and ovulation. Mol Cell Endocrinol 200:31– 43 Fortune JE, Hansel W 1979 The effects of 17␤-estradiol on progesterone secretion by bovine theca and granulosa cells. Endocrinology 104:1834 –1838 Voss AK, Fortune JE 1991 Oxytocin secretion by bovine granulosa cells: effects of stage of follicular development, gonadotropins, and coculture with theca interna. Endocrinology 128:1991–1999 Sarkar NN 2002 Mifepristone: bioavailability, pharmacokinetics and use-effectiveness. Eur J Obstet Gynecol Reprod Biol 101:113–120 Evans CA, Kennedy TG, Patrick JE, Challis JRG 1981 Uterine prostaglandin concentrations in sheep during late pregnancy and adrenocorticotropin-induced labor. Endocrinology 109:1533–1538 Sayasith K, Bouchard N, Sawadogo M, Lussier JG, Sirois J 2004 Molecular


29.

30. 31. 32. 33. 34. 35.

36. 37. 38.

characterization and role of bovine upstream stimulatory factor 1 and 2 in the regulation of the prostaglandin G/H synthase-2 promoter in granulosa cells. J Biol Chem 279:6327– 6336 Madan P, Bridges PJ, Komar CM, Beristain AG, Rajamahendran R, Fortune JE, MacCalman CD 2003 Expression of messenger RNA for ADAMTS subtypes changes in the periovulatory follicle after the gonadotropin surge and during luteal development and regression in cattle. Biol Reprod 69:1506 –1514 Altschul SF, Madden TL, Scha¨ffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ 1997 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25:3389 –3402 Neter J, Wasserman W, Kutner MH 1985 Applied linear statistical models: regression, analysis of variance, and experimental designs. 2nd ed. Homewood, IL: R. D. Irwin Sirois J, Sayasith K, Brown KA, Stock AE, Bouchard N, Dore N 2004 Cyclooxygenase-2 and its role in ovulation: a 2004 account. Hum Reprod Update 10:373–385 Duffy DM, Stouffer RL 2001 The ovulatory gonadotrophin surge stimulates cyclooxygenase expression and prostaglandin production by the monkey follicle. Mol Hum Reprod 7:731–739 Jo M, Fortune JE 2002 Oxytocin inhibits LH-stimulated production of androstenedione by bovine theca cells. Mol Cell Endocrinol 188:151–159 Acosta TJ, Ozawa T, Kobayashi S, Hayashi K, Ohtani M, Kraetzl WD, Sato K, Schams D, Miyamoto A 2000 Periovulatory changes in the local release of vasoactive peptides, prostaglandin F2␣, and steroid hormones from bovine mature follicles in vivo. Biol Reprod 63:1253–1261 Bridges PJ, Fortune JE, Regulation, action and transport of prostaglandins during the periovulatory period in cattle. Mol Cell Endocrinol, in press Murdoch WJ, Slaughter RG, Ji TH 1991 In situ hybridization analysis of ovarian prostaglandin endoperoxidase synthase mRNA throughout the periovulatory period of the ewe. Dom Anim Endocrinol 8:455– 457 Freidman S, Gurevich M, Shemesh M 1995 Bovine cyclic endometrium contains high-affinity luteinizing hormone/human chorionic gonadotropin binding sites. Biol Reprod 52:1020 –1026

Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving the endocrine community.