3Correspondence: Lawrence Espey, Ph.D.

Differential display is a PCR-based procedure that uses random primers to amplify and isolate cDNAs of genes that are uniquely expressed in clearly defined physiological and pathological processes. In the present study, four different primer sets happened to generate cDNA fragments of ovarian carbonyl reductase genes that were uniquely expressed during the ovulatory process in eCG-primed immature rats. This gene transcription was further characterized by extracting ovarian RNA at 0, 2, 4, 8, 12, and 24 h after induction of ovulation by injecting the primed animals with hCG. The results show that there are at least three homologous forms of this gene that are transcribed during ovulation. Northern blot analyses indicate a 14-fold increase in ovarian mRNA for carbonyl reductase, with expression reaching a peak at 8 h after hCG treatment and then declining to negligible levels during the next 16 h. In situ hybridization revealed that most of the transcription was in the thecal connective tissue of the ovary and was absent from the granulosa layer of ovarian follicles. Treatment of the animals with ovulation-blocking doses of epostane (an inhibitor of progesterone synthesis) or indomethacin (an inhibitor of prostanoid synthesis) did not reduce the expression of ovarian carbonyl reductase. Nevertheless, the temporal pattern of expression of carbonyl reductase following the induction of ovulation suggests that this enzyme activity is at least indirectly associated with the ovulatory process.

It is well established that there are fundamental changes in follicular steroid and prostanoid metabolism during the ovulatory process in the mammalian ovary. For example, in the gonadotropin-primed immature rat model, b -estradiol, androstenedione, testosterone, and to a lesser extent 17a -OH progesterone decrease in the ovary before the follicle ruptures, while progesterone, prostaglandin E₂, and prostaglandin F_2a increase enormously during the ovulatory process [1]. The regulation of these follicular steroids and prostanoids by various enzymes and enzyme inhibitors has been studied extensively during the past three decades [2, 3]. However, there is only limited information on the relationship of these ovarian compounds to gonadotropin-induced ovarian carbonyl reductase activity during ovulation. There is evidence that several isoenzymes of carbonyl reductase increase in the ovary in response to various dosages of gonadotropin(s) [4-9], and there is more recent evidence of ovarian expression of several homologous genes for carbonyl reductases [10, 11]. Yet, the ovarian functions of these aldo-ketoreductases are not clearly understood.

In our recent use of differential display PCR [12] to identify ovarian gene expression during ovulation, we have happened to isolate three slightly different isoforms of genes for carbonyl reductases on four different occasions (i.e., using four different primer sets for the RT-PCR reaction in the differential display procedure). Northern blotting has confirmed that these carbonyl reductase genes are distinctly expressed in the rat ovary during ovulation. The purpose of this report is to further characterize carbonyl reductase gene expression during the ovulatory process in the gonadotropin-primed immature rat model.

Immature Wistar rats were selected from litters in a breeding colony consisting of 80 adult females and 40 males. Animals were housed at 23° C under a 14-h daily light cycle that began at 0700 h. Immature rats were chosen because, in this experimental model, it is easy to induce superovulation and extirpate ovaries that consist mainly of gonadotropin-responsive follicular tissue. In addition, ovaries of immature rats do not contain the atretic follicles or corpora lutea that are characteristic of cycling adult animals. Therefore, the ovarian response to gonadotropic stimulation can be attributed primarily to preovulatory follicles.

The young females in the litters were weaned at ~ 23 days of age. The animals selected for experimentation were within the weight range of 40-50 g. (From our experience, body weight is more important than age in order to obtain a consistent ovulation rate upon stimulating the animals with specific doses of gonadotropins.) At 0800-0900 h on the day after weaning, the immature rats were injected sc with 10 IU eCG (G4877, Sigma Chemical Co., St. Louis, MO). At 0800-0900 h, 2 days later, the ovulatory process was initiated by 10 IU hCG (CG-5, Sigma) sc. Using this protocol, the animals begin ovulating approximately 12 h after the hCG injection. Ovulation rate was determined by counting the number of ova in the oviducts at 24 h after hCG was injected. The specified doses of eCG and hCG yield approximately 60-70 mature follicles per pair of ovaries (i.e., per animal), and approximately 60-70% of the ovarian mass consists of mature follicles.

Epostane (courtesy of Sterling-Winthrop Research Institute, Rensselaer, NY) was suspended in distilled water, and 1.0 N NaOH was added dropwise during vigorous stirring until the agent dissolved. The pH of the epostane solution was adjusted to 7.8-8.0 by adding 0.1 N HCl during stirring. This synthetic steroid derivative was given subcutaneously in a dose of 5.0 mg per rat at 3 h after hCG was administered. Indomethacin (I7378; Sigma) was dissolved in sodium carbonate solution containing 4 mg of Na₂CO₃ for every 10 mg of indomethacin. After ~ 15 min of stirring, the indomethacin solution was adjusted to pH 7.4 - 7.6 by slowly adding 0.1 N HCl during vigorous stirring. This nonsteroidal anti-inflammatory drug was given sc in a dose of 1.0 mg per rat at 3 h after hCG was administered to induce the ovulatory process. Ovarian RNA was extracted for Northern blotting according to the extraction procedure described below. Control ovaries were extracted at 0 h and 8 h after the administration of hCG, and the ovaries from animals treated with indomethacin or epostane were extracted at 8 h after inducing the ovulatory process with hCG.

Total RNA was extracted from whole ovaries that were extirpated at 0, 2, 4, 8, 12, and 24 h after the ovulatory process had been initiated by a dose of 10 IU of hCG. (Since follicles begin rupturing approximately 12 h after exposure to exogenous hCG, the six different extracts of RNA represented one preovulatory control group, four experimental groups taken at the specified intervals during the ovulatory process, and one postovulatory control group that contained newly developing corpora lutea.) At each of the six designated intervals after hCG, the ovaries from 12-20 rats were immediately frozen in liquid nitrogen and then pooled together to provide a total of approximately 1 g of ovarian tissue for RNA extraction based on the procedure of Chomczynski and Sacchi [13]. Each gram of ovarian tissue was placed in 3.0 ml of 4 M guanidine isothiocyanate solution and homogenized for 20-30 sec using a Tissue Tearor homogenizer (985-370, Biospec Products, Inc., Bartlesville, OK). The homogenized samples were layered onto 1.5 ml of 5.7 M cesium chloride and centrifuged over-night at 126,000 x g. Following extraction in phenol/chloroform/isoamyl alcohol, the total RNA in each extract was diluted 100-fold and quantitated at 260 and 280 nm to estimate the concentration of RNA in the original extract and to be sure there was not excessive protein in the samples. To reduce the amount of DNA in the nucleic acid extracts prior to the RT-PCR reactions, 50 m g aliquots of RNA were exposed to DNAse from a MessageClean Kit (M601, GenHunter Corporation, Nashville, TN) available for this specific purpose. After this cleaning step, the RNA was quantitated by spectrophotometry once again, and aliquots were diluted to a concentration of 0.1 m g RNA/m l H₂O for use in the RT step, below.

The RT-PCR reactions for differential display were carried out using RNAimage Kits (G501, G508, GenHunter). Briefly, for reverse transcription, 0.2 m g of the clean total RNA was primed with one-base-anchored oligo-dT primers containing a HindIII site (H = AAGCTT) attached to the 5’-end (i.e., 5’-HTTTTTTTTTG-3’ and 5'-HTTTTTTTTTA-3'), and the reaction was carried out with Moloney murine leukemia virus reverse transcriptase. The thermocycler was programmed at 65° C for 5 min, 37° C for 60 min, and 95° C for 5 min, followed by a 4° C hold. The first-strand cDNAs were amplified subsequently by PCR using selected 13-mer arbitrary up-stream primers (as designated in the first paragraph of the RESULTS) from the two RNAimage Kits in conjuction with the oligo-dT primers. The amplification was carried out with Taq DNA polymerase (M1861, Promega, Madison, WI), and the reaction mixture included ³⁵S-dATP (NEG034H, NEN Life Science Products, Boston, MA) for labelling the PCR products. The thermocycler (Thermolyne Temp Tronic model #DB66925, Barnstead/Thermolyne Corp., Dubuque, IA) was programmed for an initial 2-min denaturation at 94° C, followed by 40 cycles of amplification with denaturation at 94° C for 30 sec, annealing at 38° C for 2 min, and extension at 72° C for 30 sec, followed by a 72° C dwell for 5 min and a 4° C hold.

The PCR products were separated by electrophoresis on a sequencing gel consisting of 6% acrylamide and 48% urea in 1% TBE buffer. Before loading onto the gel, 3.5 m l of each PCR sample was mixed with 2.0 m l of loading dye and heated at 80° C for 3 min. Electrophoresis was carried out at a gel temperature of 50-55° C. The gels were dried onto Whatman 3M paper and then stapled to HyperFilm b MAX (Amersham Life Science, Arlington Heights, IL) for 1-3 days. Bands of cDNA that were differentially displayed during the six time intervals during the peri-ovulatory period were visually selected for further analysis.

The unique cDNA bands that were identified on the autoradiographs were located in the acrylamide gel (on Whatman paper) by aligning the staple holes in the paper with the staple holes in the corresponding autoradiograph. Upon comparing the six parallel lanes of electrophoretically separated cDNA, only the single most intense band was excised from the dried gel. The excised band was rehydrated in 100 m l of H₂O and boiled for 15 min before recovering the cDNA by ethanol precipitation. The recovered cDNA was reamplified using the same original primer set and initial PCR conditions, except that the dNTP concentration in the reaction mixture was 20 m M and radioactive dATP was not added. Prior to using this PCR product as a probe for Northern blot analysis, the size, purity, and amount of the amplified cDNA was estimated by electrophoresis of a 10 m l aliquot of the sample on a 2% agarose minigel. The cDNA was visualized by ethidium bromide staining, and the size and amount of each PCR product was estimated by correlating it to a ladder of standards consisting of 50, 150, 300, 500, 750, and 1000 bp (G3161, Promega). If the amount of cDNA in the 10 m l aliquot of PCR product was equal to, or greater than, the amount in the bands of the standards (i.e., ³ 300 ng), then the PCR product was used for Northern blotting. However, if the PCR product band was less intense than the nearest band of cDNA standard, then the first-round PCR sample was diluted 1:10, and 4 m l of this dilution was used as the template for another 40-cycle PCR amplification before performing Northern blotting.

Minigels consisting of 1.2% low melting point agarose in 6.7% formaldehyde and 1 M HEPES/NaPO₄ (pH 7.0) were used for electrophoresis of the total RNA to prepare Northern blots. Aliquots containing 20 m g of the original total RNA extracted from each of the six stages of ovaries were loaded into adjacent wells of a minigel, and electrophoresis was carried out at 38 V. After electrophoresis, the RNA in the minigel was blotted onto a nylon membrane by standard procedures, and the Northern blot was vacuum baked at 80° C for 100-120 min. Radiolabeling of the PCR product (i.e., of the differential display amplicon) with ³²P-dCTP (NEG013H, NEN) was facilitated by a Prime-a-Gene Labeling System (U1100, Promega). Unincorporated dNTPs were removed from the labeling mixture by centrifugation through NICK-Spin Columns (17-0862-02, Pharmacia Biotech, Uppsala, Sweden). The hybridization mixture contained approximately 2,000,000 cpm of probe/1.0 ml of solution. After 15-18 h of hybridization, the blots were rinsed to a stringency of 0.5X SSC/0.1% SDS at 55° C and exposed to Hyperfilm MP (Amersham) for an appropriate amount of time. To confirm that the different lanes on the Northern blots had equivalent amounts of RNA, the blots were stripped and then re-probed with b -actin cDNA (7323, Ambion, Inc., Austin, TX) that were isolated from the differential display gels.

The density of the signals from the Northern blots was analyzed by the NIH-image program (http://rsb.info.nih.gov/nih-image/) using the download "NIH-image 162 fat.hqx". Each Northern analysis of a uniquely expressed cDNA was scanned simultaneously with the b -actin control for the same Northern and the image was saved in Adobe Photoshop. The corresponding lanes on the experimental and control blots were aligned, and the image was rotated 90° in order to use the rectangular selection tool of the NIH-image program to select simultaneously an area encompassing a corresponding cDNA experimental lane with its b -actin control lane. After digitizing all of the bands on the Northerns, the ratio of the density of each experimental band to its corresponding b -actin control band was calculated for each lane on the Northerns. Having obtained these ratios, the value of the 8-h lane was arbitrarily established as 100%, and the densities of all of the other lanes were expressed as percentages of the 8-h value.

On the basis of the Northern blot patterns, cDNA fragments that appeared to be unique to ovulation were cloned. This step was facilitated by a pCR-TRAP Cloning System (P404, GenHunter). PCR products from re-amplification of a potentially unique cDNA were blunt-end ligated into a specific cloning site of the plasmid vector that endowed competent E. coli with tetracycline resistance. The E.coli were grown on LB agar plates that contained 10 m g tetracycline/ml. To identify a colony that contained plasmids with the desired cDNA insert (rather than plasmids with stray nucleotides or primers) 6-8 colonies were scored and small amounts of the colony cells were transferred by a pipet tip to a microfuge tube containing lysis buffer. The cDNA inserts were PCR-amplified out of each of the colony plasmids using a primer set (available in the pCR-TRAP Cloning System) that flanked each side of the plasmid insert site by 62 bp. Subsequently, the sizes of the PCR products were analyzed by electrophoresis on a 2% agarose minigel containing ethidium bromide for visualization of the cDNA. Only amplified cDNAs that were of a size comparable to the size initially estimated at the time of extraction from the differential display gel were analyzed further. (In this instance, the cDNA fragments on the minigel were selected only if they were the original size plus 124 bp, which takes into account the flanking segments of plasmid DNA that were amplified simultaneously with the insert.)

When a potentially unique amplicon was originally cut out of the differential display gel, it was likely that the extirpated Whatman 3M paper contained additional cDNA fragments that were approximately the same size as the unique cDNA. Therefore, although a given colony might have contained a cDNA insert that appeared to be the appropriate size, the insert was not necessarily the same cDNA fragment that exhibited differential hybridization during the initial Northern blot analysis. To identify colonies that actually contained cDNA fragments that were unique to ovulation, aliquots of all of the PCR samples from the previous step that contained the predicted size of cDNA insert were used to probe additional Northern blots. This secondary analysis by Northern blotting re-confirmed the uniqueness of a given cDNA fragment while simultaneously identifying specific cloning colonies that contained the unique cDNA.

Samples of colonies containing unique cDNAs were cloned further in LB medium containing 10 m g tetracycline/1.0 ml medium. The harvested plasmid DNA was purified using a Wizard Minipreps DNA Purification System (A7500, Promega). The double-stranded plasmid DNA was prepared for sequencing by an AidSeq Kit C (P203, GenHunter). Manual sequencing was performed using a Sequenase Version 2.0 DNA Sequencing Kit (US70770, Amersham) utilizing ³⁵S-dATP (NEG034H, NEN). The nucleic acid sequences were analyzed by a BLASTn search of the database server at the NCBI (http://www.ncbi.nlm.nih.gov).

Ovaries were removed at the six peri-ovulatory intervals described in the above section on RNA extraction and were fixed overnight in 4% paraformaldehyde. The in situ hybridization protocol was adapted from a published method [14]. The procedure was facilitated by using a Riboprobe Transcription System (P1450, Promega).

Numerical data are presented as means ± SE. The significances of the differences were determined by Duncan's multiple range tests after a completely randomized one-way analysis of variance of the means of the groups. The probability value used as the cutoff between "significant" and "not significant" was P = 0.05.

During this study, 144 different primer sets were used in the PCR reactions to amplify the cDNA generated from the ovarian mRNA that was extracted at the 6 designated intervals during the peri-ovulatory period. Four of those primer sets resulted in differential display autoradiographs that happened to reveal ovulation-specific expression of cDNAs associated with carbonyl reductase (Figure 1). The particular pairs of primers that yielded differentially expressed cDNAs that were homologous to carbonyl reductase genes were (1) 5’-HTTTTTTTTTG-3’ and 5’-HCTCAACG-3’ (H-AP4, GenHunter); (2) 5’-HTTTTTTTTTG-3’ and 5’-HAACGAGG-3’ (H-AP7); (3) 5’-HTTTTTTTTTG-3’ and 5’-HAACTGAG-3’ (H-AP58); and (4) 5'-HTTTTTTTTTA-3' and 5’-HAACTGAG-3’ (H-AP58). The differentially expressed cDNAs are referred to tentatively as G4, G7, G58, and A58, respectively, on the basis of the primer sets that amplified them (Figure 1).

Each of the cDNA fragments that were generated by the above four sets of primers hybridized in a similar pattern to Northern blots. The cDNA probes generated minimal signals in the 0-h and 24-h lanes of the Northerns (Figure 2), while the strongest signal was at 8 h after hCG. The strongest signal (i.e., the 8-h lane) was arbitrarily set at 100%, and the densities at the other intervals of the ovulatory process were expressed as fractions of that maximum. Accordingly, the signal densities (normalized against the b-actin control) at 0, 2, 4, 8, 12, and 24 h after hCG were 7.4%, 25.9%, 71.9%, 100%, 57.2%, and 8.2%, respectively (Figure 2). Thus, during the first 8 h of the ovulatory process, there was a 14-fold increase in carbonyl reductase gene expression. Subsequently, it declined, resulting in expression at 24 h that was not significantly different (p = 0.94) from the 0-h control level.

Sequencing revealed that the cDNA fragments that were differentially displayed in Figure 1 were all approximately 300 bp in length (Figure 3). However, there were slight variations in the lengths of the cDNAs for three distinctly different reasons: First, as expected, the number of bp in a given fragment varied depending on the site where the random primer annealed to the first strand cDNA (Figure 3). Second, the poly-T primers exhibited slight variations in the manner in which they annealed to the mRNA (or, the cDNA). Third, four of the six different clones that were sequenced had a 7-bp deletion located approximately 1/3 of the distance from the poly-T tail. In addition, there were three other positions along this @ 300 bp segment of carbonyl reductase genes that exhibited polymorphism (Figure 3). Close comparison of the sequences of the six different clones reveals that they are fragments of three different isoforms of an ovarian carbonyl reductase gene. Clones G4-2 and G7-2 (NCBI Accession #AF023086) are basically identical to one another, clones G4-1 and G7-1 (NCBI Accession numbers) are essentially the same, and clones G58-1 and A58-1 (NCBI Accession numbers) are identical except that the poly-T end is slightly different (Figure 3). Furthermore, the existence of these three different isoforms explains why, in at least one instance, two different lengths of unique cDNAs are visible in the same differential display autoradiograph. In Figure 1, the autoradiograph corresponding to the G7 primer set reveals a longer cDNA (i.e., G7-2) and a shorter cDNA (i.e., G7-1) that are both uniquely expressed during ovulation. Also, sequence analyses of several different clones from the slightly distinguishable double-band in the G4 autoradiograph (Figure 1) verified that at least two of the three different cDNA isoforms were produced by this primer set.

In view of the degree of homology among the six cDNAs, only one (i.e., the 305 bp G7-2 fragment) was used for in situ hybridization analysis. The results confirm the temporal pattern of carbonyl reductase mRNA expression observed by Northern blotting. There was very limited signal from the probe at 0 h and 24 h into the ovulatory process, while the strongest signal was at 8 hours after the rats were stimulated by hCG (Figure 4). Hybridization was intense in the thecal layers of mature follicles, as well as in extra-follicular connective tissue throughout the ovary, while there was no significant hybridization in the granulosa layer (Figure 5).

For these tests, Northern blots were prepared from RNA that was extracted from control ovaries at 0 h and 8 h into the ovulatory process, and extracted from experimental ovaries that were taken at 8 h after hCG from rats that had been treated 5 hours earlier with ovulation-inhibiting doses of epostane or indomethacin. As in the Northern blotting tests (described above) at the six different intervals during ovulation, the signal density (normalized against the b-actin control) of the 8-h lane was arbitrarily set at 100% (Figure 6). This intensity at 8 h was 13-fold greater than the 0-h control value for carbonyl reductase gene expression. The anti-ovulatory agent epostane, which blocks progesterone synthesis, increased insignificantly (p = 0.35) the signal density to 110.6% of the 8-h control value. Likewise, indomethacin, which blocks prostanoid synthesis, also increased insignificantly (p = 0.09) the signal density to 120.5% of the 8-h control value. Thus, neither of these two well-known ovulation-inhibiting agents reduced the ovarian expression of carbonyl reductase genes.

Carbonyl reductase is an NADPH-dependent, cytosolic enzyme that is widely distributed in human and animal tissues [5, 6, 15-20]. Actually, it is a group of monomeric aldo-ketoreductases with broad specificity for converting carbonyl compounds such as aromatic, aliphatic, and cyclic aldehydes and ketones into alcohols. More recently, it has been suggested that there may be two different categories of enzymes, one subgroup being more specific for steroids and the other with broader specificity [17]. The 20b -hydroxysteroid dehydrogenase found in pig testicles [15] and the carbonyl reductase found in rat ovaries [4, 5, 10, 11] are gonadal enzymes that appear to belong to the former category. This report provides further characterization of the timing and nature of such gene expression in ovarian tissue during ovulation.

The G4-2/G7-2 cDNA fragments of a carbonyl reductase gene are not identical to any subjects registered in the NCBI gene database. In contrast, the G4-1/G7-1 cDNA fragments are identical to segments of a 981 bp mRNA (NCBI Accession #D89069) that is reportedly inducible in immature rat ovaries that have been stimulated by eCG, alone [11]. Similarly, the G58-1/A58-1 cDNA fragments are identical with a 997 bp mRNA (NCBI Accession #X84349) found in rat testes [17]. Also, all six of the clones are highly homologous to a 1053 bp mRNA (NCBI Accession #U31966) for carbonyl reductase in the mouse cerebellum [18]. Likewise, they are similar to a segment of a non-inducible carbonyl reductase gene (NCBI Accession #D89070) in the rat ovary [11]. Thus, the cDNA fragments presently discovered by the randomness of the differential display procedure are polymorphic representations of the superfamily of genes that translate into aldo-ketoreductases that have been associated with steroid and eicosanoid metabolism in a number of tissues.

The physiological role of these aldehyde- and ketone-reducing enzymes is not clear [17]. They were first associated with ovarian function a decade ago [4]. However, the subsequent efforts to clarify their role in ovarian follicular events has led to confusing conclusions. Diverse results have suggested that ovarian gene expression for carbonyl reductase is important "in ovarian follicular development" [11], but "is not always necessary for follicular maturation" [9]. In similar contradiction, other reports conclude that carbonyl reductase "is closely involved…in the ovulatory process" [6], but "is not always necessary for…the ovulatory process" [8]. It has been further suggested that, rather than contributing to follicular maturation or ovulation, carbonyl reductase may be "involved in the formation of neo corpora lutea after ovulation" [8]. This uncertainty about the function of ovarian carbonyl reductase may be the consequence of using a variety of unusual protocols for treating rats with gonadotropins in diverse studies during the past decade [4-9, 11]. On the other hand, the present study provides the first clear analysis of the temporal pattern of expression of three isoforms of ovarian carbonyl reductase genes during ovulation in a common experimental model of the gonadotropin-primed immature rat.

The present study also demonstrates that the temporal pattern of expression of carbonyl reductase mRNA in eCG-primed immature rats is the same, regardless of whether one examines the differential display gels, the Northern blots, or the in situ hybridization data. There was a slight increase in carbonyl reductase mRNA at 2 h after ovulation was induced in the animals by treatment with hCG. The expression reached a peak at 8 h after hCG, and it remained relatively high at 12 h after hCG, i.e., when follicles first begin to rupture. Subsequently, the ovarian mRNA for carbonyl reductase had returned to control levels by 24 h after hCG, i.e., by 10-12 h after completion of the ovulatory process (an observation that does not support a role for this enzyme in luteal development). This post-ovulatory down-regulation is not consistent with earlier reports [4, 8] that ovarian carbonyl reductase enzyme activity is the highest after ovulation, i.e., in neo-natal tissue.

This report contains the first in situ hybridization data on the localization of ovarian carbonyl reductase mRNA expression. Most of the signal was from the thecal layers of the larger follicles, along with additional expression in the extra-follicular connective tissue in the ovary. In contrast, there was negligible expression of carbonyl reductase gene in the granulosa layer of the follicles. These observations are in spatial (but not temporal) conformity with earlier immunohistochemical data that localized the actual enzyme in ovarian thecal tissue [4].

It remains to be determined why carbonyl reductases are expressed so strongly in ovarian thecal connective tissue during ovulation. The temporal pattern of their expression suggests they must be related in some way to the ovulatory process. However, it is not possible to establish from the present data that carbonyl reductase activity contributes in some direct way to the mechanism of ovulation, because doses of epostane and indomethacin that inhibited ovulation did not block the transcription of mRNA for this enzyme. This finding suggests that neither progesterone nor prostanoids, respectively, are required for the induction of ovarian carbonyl reductase expression. Yet, it is not possible to exclude any role for this enzyme in the ovulatory process, because the gonadotropin-induced expression of carbonyl reductase could precede the increases in ovarian steroids and prostanoids within the chronology of ovarian events that lead to follicular rupture. Conversely, since neither epostane nor indomethacin completely block the ovulatory increase in progesterone [21, 22], it is still possible that this steroid may be associated with carbonyl reductase gene expression in the ovary (although there is no evidence of progesterone receptor expression in ovarian thecal tissue).

Finally, studies in other biological systems have led to the hypothesis that aldo-ketoreductases may have a physiologic role in the detoxification of steroids, prostanoids, and pterins in tissues where their endogenous production is excessive [16, 20, 23]. Therefore, it is possible that carbonyl reductase may be serving as a local protective response to the substantial increases in ovarian steroids and prostanoids during ovulation. It would be interesting to know whether the peri-ovulatory expression of this enzyme might also influence the mobilization of follicular fibroblasts that is known to occur at the time of ovulation. In any event, the significant expression of carbonyl reductase mRNA in the fibroblast-rich thecal layers of the follicle suggests that ovarian carbonyl reductases have some significant ovarian function at the time of ovulation. Their specific role in the function of thecal fibroblasts during ovulation is worthy of further analysis.

Figure 1. Four autoradiographs of differentially displayed carbonyl reductase cDNAs (arrows) using the primer sets G4, G7, G58, and A58. Note that G7 contains two distinct, parallel bands, while G4 and A58 contain barely discernible parallel bands of cDNA that are differentially expressed during the peri-ovulatory period.

Figure 2. Intensity of Northern blot signals at the six intervals of the peri-ovulatory period following hCG administration. The most intense signal (at 8 h after hCG) was arbitrarily set at 100%. The actual Northerns for cDNA fragments G4-1 and G7-2 are shown along with the respective b -actin controls for the same Northern blots. The linear graph of the image density analysis is based on six different Northerns, with one probe prepared from each of the six different cDNA fragments identified in Figure 3.

Figure 3. Comparison of the sequence of six slightly different cDNA fragments generated by four different primer sets. Initial brackets indicate portions of the poly-T primers, and final brackets indicate the random primers, including the HindIII (H) restriction site. The four polymorphous regions are indicated by bold letters within horizontal lines. The open rectangles indicate the position of a 7-bp deletion in four of the cDNAs.

Figure 4. Change in intensity of the in situ hybridization signal from probe G7-2 during the six different peri-ovulatory intervals following hCG administration. Lighter micrographs on the left show ovarian sections stained with eosin and hematoxylin, while micrographs on the right are autoradiographs from serial sections containing hybridized G7-2 probe labeled with ³⁵S-UTP.

Figure 5. Closer view of the distribution of probe in the thecal connective tissue of the ovary. Arrows point to the granulosa layer of mature follicles where there is negligible hybridization.

Figure 6. Comparison of the % of signal from Northern blots containing RNA extracted at 8 h after hCG from animals that were also treated with either 5 mg of epostane (Epo), or 1 mg of indomethacin (Indo) administered at 3 h after hCG. Bar graph data is based on NIH-image analyses of four different Northerns, with two each being probed with cDNA fragments G4-1 and G7-2. The signal from the 8-h control lane was arbitrarily set at 100% optical density. In parallel groups of animals, the ovulation rate was determined at 24 h after hCG.