Xenopus laevis ovarian CYP17 is a highly potent enzyme expressed exclusively in oocytes. Evidence that oocytes play a critical role in Xenopus ovarian androgen production.

Progesterone has long been considered the primary mediator of Xenopus oocyte maturation. We have recently shown, however, that androgens, which are equal or more potent promoters of maturation and are present at higher levels in ovulating frogs, may also be playing an important physiologic role in mediating maturation. Here, we examined the role of CYP17, a key enzyme mediating sex steroid synthesis, in Xenopus ovarian androgen production. We found that the 17,20-lyase activities of Xenopus CYP17 exceeded the 17alpha-hydroxylase activities in both the Delta4 and Delta5 pathways; thus, Xenopus CYP17 rapidly converted pregnenolone and progesterone to dehydroepiandrosterone (DHEA) and androstenedione, respectively. This remarkably robust activity exceeds that of CYP17 from most higher vertebrates, and likely explains why virtually no progesterone is detected in ovulating frogs. Additionally, ovarian CYP17 activity was present exclusively in oocytes, although all other enzymes involved in sex steroid production were expressed almost entirely in surrounding follicular cells. This compartmentalization suggests a "two-cell" model whereby Xenopus ovarian androgen production requires both follicular cells and oocytes themselves. The requirement of oocytes for ovarian androgen production further introduces the unusual paradigm whereby germ cells may be responsible for producing important steroids used to mediate their own maturation.

The phenomenon of steroid-induced maturation of Xenopus oocytes has served as a model for studying cell cycle and nongenomic steroid signaling for several decades (1)(2)(3)(4)(5). During this time, progesterone has been considered the primary physiologic mediator of oocyte maturation, perhaps through interactions with classical progesterone receptors expressed in the oocyte (6,7). We have recently shown, however, that testosterone, rather than progesterone, may be the primary physiologic mediator of Xenopus oocyte maturation (8). This dominant role of testosterone is evidenced by the observation that testosterone is a substantially more potent activator of oocyte maturation than progesterone, and that testosterone levels are significantly higher than progesterone (which is nearly unde-tectable) in the serum and ovaries of ␤-HCG 1 -stimulated frogs. In addition, we have shown that, in vitro, isolated oocytes rapidly convert exogenously added progesterone to the androgen androstenedione (AD), which is an equally potent promoter of maturation when compared with progesterone. These data suggest that both AD and progesterone are likely inducing maturation under in vitro conditions typical for "progesteronemediated" maturation.
The high levels of ovarian testosterone in the absence of detectable progesterone imply that Xenopus ovaries are extremely efficient at metabolizing progesterone. Additionally, the ability of isolated oocytes to rapidly convert progesterone to androgens suggests that the oocytes themselves may be important contributors to progesterone metabolism in the ovary. To explain the lack of in vivo progesterone accumulation, as well as to characterize the role of oocytes in ovarian androgen production, we separated oocytes from surrounding follicular cells and examined the ability of both cell types to metabolize progesterone and other androgen precursors. We focused on the function of the cytochrome P450 enzyme CYP17 in ovarian androgen production, as it is known to play a pivotal role in the synthesis of androgens (Fig. 1). CYP17, an endoplasmic reticulum membrane-bound multifunctional enzyme (9 -11), exhibits two enzymatic activities. Its 17␣-hydroxylase activity converts pregnenolone and progesterone into their respective 17␣-hydroxylated products 17␣-hydroxypregnenolone (17OHPreg) and 17␣-hydroxyprogesterone (17OHProg). The second 17,20-lyase activity cleaves the steroid side chains of 17OHPreg and 17OHProg to yield dehydroepiandrosterone (DHEA) and AD, respectively. CYP17 is expressed in several steroidogenic tissues (12,13), including adrenal cortex, ovary, and testis. Whereas earlier work implicates the presence of CYP17 in oocytes (8), the kinetic characteristics of the Xenopus CYP17 (XeCYP17), as well as its expression levels in other ovarian cell types, is not known. Furthermore, expression levels of other important steroidogenic enzymes in various ovarian cell types have not been examined.
We show here that XeCYP17 has high 17␣-hydroxylase, and even more pronounced 17,20-lyase, activities in both the ⌬5 (pregenenolone) and ⌬4 (progesterone) pathways (Fig. 1). This differs from most known CYP17 isoforms, which generally favor one pathway over the other, and rarely have 17,20-lyase activity that rivals their 17␣-hydroxylase activity. Furthermore, we show that ovarian CYP17 activity is present exclu-sively in oocytes, whereas other important steroidogenic enzymes, including 3␤-HSD and 17␤-HSD, are located primarily in the surrounding follicular cells. Finally, we propose a "twocell" model for androgen synthesis in the Xenopus ovary that involves both oocytes and follicular cells. This model implies that germ cells themselves are critical for Xenopus ovarian androgen production, which in turn may play an important physiologic role in promoting their own maturation.

EXPERIMENTAL PROCEDURES
Oocyte and Follicular Cell Preparation-Ovaries were harvested from Xenopus laevis (Nasco, Fort Atkinson, WI) and oocytes were isolated by incubation with 1 mg/ml collagenase A (Roche Molecular Biochemicals) at room temperature for 4 h in modified Barth's solution (MBSH) as previously described (14,15). Oocytes were then washed and incubated overnight at 16°C in MBSH with 1 mg/ml bovine serum albumin, 1 mg/ml Ficoll, 100 units/ml penicillin, and 0.1 mg/ml streptomycin. Stage V-VI oocytes were selected and examined microscopically to confirm the absence of follicular cells. Follicular cells were isolated by incubating ovaries in collagenase as above for 1 h. Oocytes were allowed to settle by gravity for ϳ5 min, and the supernatant containing follicular cells was then centrifuged at 800 ϫ g for 5 min. The pelleted cells were washed 3 times with MBSH, and any remaining oocytes were removed manually under a dissecting microscope. Similar numbers of follicular cells were used in all of the metabolism experiments. In addition, nearly identical results were obtained using follicular cells removed from oocytes after up to 4 h of treatment with collagenase, or using follicular cells separated from oocytes by incubation with trypsin (16).
Preparation of Oocyte Membranes-Crude oocyte membranes were prepared as previously described (17). In short, stage V-VI oocytes were homogenized in membrane buffer (83 mM NaCl, 1 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, 0.5 g/ml leupeptin, 0.5 g/ml pepstatin, 10 mM Hepes, pH 7.6) at 4°C. The homogenate was centrifuged at 800 ϫ g for 5 min and the supernatant containing the membranes was removed and centrifuged two more times at 800 ϫ g. The supernatant was then centrifuged at 15,000 ϫ g for 15 min, and the membrane pellet was resuspended in membrane buffer. Membranes were centrifuged two more times at 15,000 ϫ g and resuspended in MBSH. Protein concentrations were then measured using the BCA kit (Pierce), and samples were frozen at Ϫ80°C until needed. We saw no significant drop in CYP17 activity after a single freeze/thaw cycle of the membranes.
CYP17 Enzyme Assays in Oocyte Membranes-17␣-Hydroxylase and 17,20-lyase activities in oocyte membranes were performed as previously described (18). Membranes were assayed under initial rate kinetics by incubation in 50 mM potassium phosphate buffer (pH 7.4) with 0.03-1 M steroid (Sigma and Steraloids, Newport, RI) and 1 mM of the cofactor NADPH in a 500-l total volume at 16°C for 30 min. Each reaction contained 50 g of membranes and 50,000 cpm of [7-3 H]pregnenolone (19), [ 3 H]17␣-hydroxypregnenolone, [1,2,6, H]progesterone, or [1,2,6,7-3 H]17␣-hydroxyprogesterone (PerkinElmer Life Sciences). Steroids were extracted with 3 ml of 3:2 ethyl acetate:hexane. Amounts of radioactivity were measured using a scintillation counter, with Ͼ90% recovery from medium. Steroids were concentrated under nitrogen and separated by TLC using 3:1 chloroform:ethyl acetate. Quantitation of steroids was measured by cutting out the steroid spots on the TLC plate and measuring radioactivity using liquid scintillation (20). Kinetic behavior was approximated as a Michaelis-Menten system for data analysis. The identities of all steroids in these and the other metabolism experiments were confirmed by high performance liquid chromatography.
CYP17 Enzyme Assays in Transfected HEK-293 Cells-HEK-293 cells were grown in complete medium consisting of Dulbecco's modified Eagle's medium, 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin (Invitrogen). Experiments were performed on 12-well plates. Cells were transfected by calcium phosphate precipitation (8,21) with either pcDNA3.1 (mock) or pcDNA3.1 containing cDNA encoding the Xenopus or human CYP17 protein (8). After 48 h, cells were placed in fresh complete medium containing 5% charcoal-stripped fetal bovine serum and 60,000 cpm/well of radiolabeled progesterone, 17␣-hydroxyprogesterone, pregnenolone, or 17␣-hydroxypregnenolone at the indicated concentrations. Cells were incubated at 37°C, and medium was removed at the specified times. Steroids were extracted with 5 ml of 3:2 ethyl acetate:hexane, the organic layer was concentrated, and TLC followed by autoradiography and quantitation was performed as described above.
Steroid Metabolism Assays in Ovaries, Oocytes, and Follicular Cells-Ovaries (ϳ500 mg/sample), isolated oocytes (30 oocytes per sample), follicular cells, or recombined follicular cells and oocytes (20 oocytes per sample) were incubated with ϳ500,000 cpm of [ ␤-HCG-mediated Maturation of Oocyte in Ovarian Fragments-Ovarian fragments of ϳ100 -200 mg were washed in MBSH and treated for 1 h in 2 ml of MBSH with either ethanol or 100 nM VN/85-1 (a gift from A. Brodie, University of Maryland). Ethanol concentrations were kept constant. ␤-HCG was then added at a concentration of 100 units/ ml, and the ovarian fragments were incubated at 16°C for ϳ12 h. The MBSH was removed, and steroids were extracted and analyzed by radioimmunoassay (8). Oocytes were manually removed from the ovarian fragments and maturation was determined by visualization of a white spot on the animal pole.

Sex Steroid Precursors Pregnenolone and Progesterone Are Rapidly Metabolized to Testosterone by Xenopus Ovaries-Pre-
vious work in our laboratory demonstrated that ␤-HCG or pregnant mare serum gonadotropin (PMSG) stimulation of Xenopus ovaries in vivo and in vitro produced high levels of testosterone and small to moderate amounts of androstenedione. In contrast, little to no androgen precursors, including pregnenolone, 17OHPreg, progesterone, and 17OHProg, were detected (8). To determine whether Xenopus ovaries could convert sex steroid precursors to androgens in the absence of gonadotropins, we examined pregnenolone and progesterone metabolism by incubating ovarian tissue with the respective radiolabeled steroids. Both pregnenolone and progesterone were rapidly converted to testosterone (Fig. 2), with more than 90% loss of each steroid by 2 h. Virtually no intermediates or further metabolites of testosterone (e.g. estrogen or dihydrotestosterone) were detected in both the pregnenolone-and progesterone-treated ovaries (Ͻ10% of total counts at all time points). These data suggest that Xenopus ovaries contain all of the enzymatic machinery necessary for the conversion of sex steroid precursors to testosterone independent of gonadotropin stimulation. Furthermore, this ability to rapidly metabolize pregnenolone and progesterone likely explains why both steroids are nearly undetectable in gonadotropin-stimulated ovaries both in vivo and in vitro (8).

CYP17 in Xenopus Oocyte Membranes Possesses High 17␣-Hydroxylase and 17,20-Lyase Activity in Both the ⌬5 and ⌬4
Pathways-The rapid conversion of progesterone and pregenenolone to androgens implies that the Xenopus ovary contains CYP17, the first key enzyme in the conversion of these sex steroid precursors to androgens (22,23). Indeed, we previ- ously cloned the XeCYP17 enzyme from an oocyte cDNA library and demonstrated its ability to catalyze two enzymatic reactions: the hydroxylation of progesterone to 17OHProg, and the conversion of 17OHProg to AD (8). To further characterize the XeCYP17 enzyme, and to determine its role in Xenopus ovarian androgen production, we used membrane preparations from isolated oocytes to measure the kinetic characteristics of XeCYP17. Representative Lineweaver-Burk plots for the 17␣hydroxylase and 17,20-lyase activities in both the ⌬4 and ⌬5 pathways are shown in Fig. 3, with the calculated average K m and V max values for five separate experiments shown in Table  I. The K m values for the 17␣-hydroxylase and 17,20-lyase activities in both the ⌬4 and ⌬5 pathways ranged between 26 and 117 nM. The maximal velocities (V max ) for 17␣-hydroxylase activity in both the ⌬4 and ⌬5 pathways were similar in magnitude, ranging from 1.3 to 5.2 pmol/mg of protein/min. Surprisingly, XeCYP17 contained remarkably high 17,20-lyase activity in both the ⌬4 and ⌬5 pathways, with V max ϭ 2.3 pmol/mg of protein/min for the conversion of 17OHPreg to DHEA (⌬5 pathway), and V max ϭ 12.3 pmol/mg of protein/min for the conversion of 17OHProg to AD (⌬4 pathway). This relatively high 17,20-lyase activity could be seen more clearly by examining the ratio of V max values for the 17,20-lyase versus 17␣-hydroxylase reactions in both pathways (Table I). The lyase/hydroxylase values were 1.8 in the ⌬5 pathway, and 2.5 in the ⌬4 pathway, indicating that XeCYP17 catalyzed the 17,20-lyase reaction even more quickly than the 17␣-hydroxylase reaction in both pathways. Furthermore, these results suggest that, in the Xenopus ovary, androgens can be rapidly and efficiently produced via both the ⌬4 or ⌬5 pathways.
Xenopus CYP17 Expressed in HEK-293 Cells Completely Metabolizes Sex Steroid Precursors in Both the ⌬4 and ⌬5 Pathways, whereas Human CYP17 Favors the ⌬5 Pathway-The high 17,20-lyase activity of XeCYP17 in both the ⌬4 and ⌬5 pathways is rather unusual, as most CYP17 isoforms favor one pathway over the other (12,18,24). For example, the human CYP17 (HuCYP17) has little to no 17,20-lyase activity in the ⌬4 pathway; thus, human sex steroid synthesis is felt to involve primarily the ⌬5 pathway (18). To directly compare the kinetic profiles of the Xenopus and human CYP17 enzymes, and to confirm that the observed enzymatic activities in the Xenopus oocyte membranes were in fact mediated by the cloned Xe-CYP17 enzyme, we expressed the human and Xenopus CYP17 proteins individually in HEK-293 cells. Fig. 4, A and B, shows that XeCYP17 converted progesterone to 17OHProg, and then to AD (40% of total counts by 8 h), confirming that both 17␣hydroxylase and 17,20-lyase activities are carried by this protein. In contrast, HuCYP17 contained high 17␣-hydroxylase, but very little 17,20-lyase, activity in the ⌬4 pathway (Fig. 4, A and C, 6% conversion to AD by 8 h). Calculated K m values for HuCYP17-mediated 17␣-hydroxylase reactions in the ⌬4 and ⌬5 pathways were ϳ0.54 and 0.28 M, respectively (data not shown), which correlate well with published values (18). This confirms that the enzyme is functioning as expected in our HEK-293 expression system.
Comparison of the calculated lyase/hydroxylase V max ratios in HEK-293 cells confirmed that XeCYP17 had high 17,20lyase activity in both pathways, with lyase/hydroxylase V max ratios of approximately unity (Fig. 4D). In contrast, the 17␣hydroxylase reaction was dominant in both pathways for the HuCYP17, with ratios of ϳ0.3 in the ⌬5 pathway and Ͻ0.05 in the ⌬4 pathway. The lower lyase/hydroxylase ratios of Xe-CYP17 expressed in the HEK-293 cells when compared with the oocyte membranes could be because of many factors, including the availability of important cofactors such as NADPH and the flavoprotein reductase(s), or the species compatibility of the 17,20-lyase cofactor cytochrome b 5 in human versus Xenopus tissues. The 17,20-lyase activity in HEK-293 cells was still relatively high, however, thus these experiments appear to confirm that the cloned XeCYP17 enzyme is indeed responsible for the potent 17,20-lyase activity seen in oocyte membranes. Notably, similar results were qualitatively seen in COS cells; however, the presence of endogenous CYP17 in these cells precluded their use for quantitative studies.
Xenopus Oocytes Possess High CYP17 Activity, but Little to No 3␤-HSD or 17␤-HSD Activity-Having established that Xenopus CYP17 can metabolize sex steroid precursors equally well in both the ⌬4 and ⌬5 pathways, we next determined which cells within the ovary contained CYP17 activity. As mentioned, we had previously shown that isolated Xenopus oocytes possess high CYP17 activity in the ⌬4 pathway (8). To confirm the presence of CYP17 in Xenopus oocytes, we separated oocytes from surrounding follicular cells and performed steroid metabolism experiments. Notably, isolated oocytes were examined very carefully both under the dissecting microscope and by staining of oocyte sections to exclude the presence of follicular cell contamination in our preparations. Fig. 5 represents one of over 50 different hematoxylin/eosin-stained oocyte sections, with no detectable follicular cell contamination.
As expected, isolated oocytes contained high CYP17 activity in both the ⌬5 and ⌬4 pathways, as radiolabeled pregnenolone and progesterone were rapidly metabolized to only DHEA and AD, respectively (Fig. 6A). Interestingly, the isolated oocytes did not appear to significantly express any of the other enzymes responsible for the various stages in the production of sex steroids. For example, 3␤-HSD is responsible for the conversion of ⌬5 to ⌬4 steroids (Fig. 1). The rapid conversion of radiolabeled pregnenolone to DHEA in the absence of detectable ⌬4 steroid production (Fig. 6A, left panel) argues strongly that oocytes lack significant 3␤-HSD activity. Likewise, 17␤-HSD is necessary for the conversion of DHEA to androstenediol, and of androstenedione to testosterone (Fig. 1). The lack of detectable androstenediol in the pregnenolone-treated cells, as well as the near absence of testosterone in progesterone-treated cells (Ͻ10% of the total counts at all time points measured), sug- gests that oocytes contain little 17␤-HSD activity (Fig. 6A). Finally, stimulation of isolated oocytes with ␤-HCG did not promote significant steroid production by radioimmunoassay (data not shown), suggesting that they contain little CYP11A1 or steroidogenic acute regulatory protein (stAR) activity, both of which are necessary to augment formation of pregnenolone (Fig. 1).
Xenopus Ovarian Follicular Cells Contain 3␤-HSD and 17␤-HSD Activities, but No CYP17 Activity-If oocytes have no significant 3␤-HSD and 17␤-HSD activity, then these enzymes must be present in the surrounding ovarian follicular cells to complete ovarian androgen synthesis. In support of this hypothesis, small amounts of radiolabeled pregnenolone were converted to progesterone by isolated ovarian follicular cells, indicating the presence of 3␤-HSD in these cells (Fig. 6B, left  panel). Similar numbers of follicular cells converted radiolabeled DHEA to AD at a significantly higher rate (Fig. 6B, middle panel), thus confirming the presence of 3␤-HSD and suggesting that DHEA may be preferred over progesterone as a substrate for Xenopus ovarian 3␤-HSD. Finally, radiolabeled AD was very efficiently converted to testosterone (Fig. 6B, middle and right panels), thereby demonstrating the presence of 17␤-HSD in the follicular cells as well. Surprisingly, follicular cells did not hydroxylate progesterone or pregnenolone at all (Fig. 6B, left panel, and data not shown), indicating that ovarian CYP17 activity is contained exclusively in the oocytes themselves.
Because separation of oocytes and follicular cells revealed differential expression of the steroidogenic enzymes, the isolated cell types were recombined to determine whether the complete steroidogenic pathway from pregnenolone to testosterone could be reconstituted. The recombination of follicular cells and oocytes resulted in testosterone production (Fig. 6C), confirming that these two populations of cells were indeed

TABLE I Kinetic parameters of XeCYP17 in Xenopus oocyte membranes
The kinetic parameters of XeCYP17 expressed in Xenopus oocyte membranes were measured as described under "Experimental Procedures." The 17␣-hydroxylase and 17,20-lyase activities in both the ⌬4 and ⌬5 pathways are indicated, as are the steroid substrates and end products for each reaction (parentheses). The V max ratios of lyase: hydroxylase activity for the two pathways are indicated in the right column. Values are expressed as mean Ϯ S.D. (n ϭ 5). sufficient to mediate complete steroidogenesis. Several intermediate steroids were also produced, including progesterone, DHEA, and AD, suggesting that the reconstituted system was less efficient than the intact ovary in producing testosterone.

Blockade of Androgen Production Attenuates ␤-HCG-stimulated Maturation of Xenopus Oocytes in Intact Ovarian
Follicles-To confirm the physiologic importance of CYP17-mediated androgen production in Xenopus oocyte maturation, intact ovarian follicles were stimulated with ␤-HCG alone or in combination with the potent CYP17 inhibitor VN/85-1. The maxi-   (Table II, ethanol), which is consistent with more than 8 similar studies using identical conditions (data not shown). This amount of maturation is also similar to that seen when ovarian follicles were treated with 500 nM AD (Table II), 2 suggesting that ϳ50% maturation is likely the maximum attainable under these conditions. Inhibition of CYP17 with VN/85-1 significantly decreased ␤-HCG-mediated oocyte maturation by 28%. As expected, VN/85-1 also decreased, but did not eliminate, both testosterone and AD production. Interestingly, although progesterone production increased slightly in the presence of VN/85-1, progesterone levels still remained quite low, suggesting that very little was being produced in vivo, even the presence of a CYP17 inhibitor. This result confirms the data in Fig. 6, where pregnenolone was shown to be inefficiently converted to progesterone by follicular cells. Together, these data suggest that CYP17-mediated androgen production may indeed be important for maturation in vivo. Incomplete inhibition of maturation by VN/85-1 was likely because of induction by the remaining AD and testosterone (which is 10-fold more potent than AD and progesterone (8)), as well as perhaps by the slightly higher amounts of progesterone. Addition of 500 nM AD rescued the inhibitory effects of VN/85-1 by increasing AD and testosterone levels ( Table II). Addition of exogenous progesterone also rescued the effects of VN/85-1 by increasing progesterone levels while androgen levels remained low. These data are consistent with earlier work where progesterone promoted maturation of isolated oocytes treated with ketoconazole (8), and suggest that, although low endogenous progesterone production precludes its importance relative to androgens in vivo, exogenous progesterone is still capable of promoting maturation in vitro.

DISCUSSION
Ovarian sex steroid production is essential for follicular growth and subsequent ovulation in nearly every animal (25)(26)(27). In frogs and fish, these steroids also appear to be critical regulators of oocyte maturation (28,29), which is defined as the resumption of meiosis from prophase I to metaphase II. Fish oocyte maturation is regulated by various hydroxylated progesterone metabolites (30,31), whereas testosterone may be an important physiologic mediator of Xenopus oocyte maturation (8). Sex steroids may be involved in higher vertebrate oocyte maturation as well; however, evidence for or against such a role is still minimal at this point in time.
Taken together, these data lead us to speculate that the level of CYP17-mediated 17,20-lyase activity, as well the preferential use of one or both (⌬4 and ⌬5) of the steroidogenic path-  ways, may reflect the need of the enzyme to produce reproductive steroids versus cortisol in various organisms. In humans, for example, cortisol is essential for normal growth and development. Cortisol is derived primarily from the 21-hydroxylation of 17␣-hydroxyprogesterone; thus, the relatively low 17,20lyase activities of human CYP17, which leads to the accumulation of 17␣-hydroxylated steroids, may be necessary to permit cortisol production in human adrenal glands. In contrast, frogs, as well as sharks, eels, and most fish, do not appear to need cortisol; instead, they rely on corticosterone or corticosterone metabolites as their primary glucocorticoids (39 -41). CYP17 is therefore not required for glucocorticoid production in these lower vertebrates; its sole function appears to be to generate sex steroids. In short, CYP17 may have evolved from favoring primarily sex steroid production in lower vertebrates, which do not require cortisol, to promoting both cortisol and sex steroid production in higher vertebrates. Accordingly, the lower vertebrate CYP17s appear to form a closely related phylogenetic family by sequence homology, whereas the more evolved higher vertebrate CYP17s appear to be part of a distinctly separate family of proteins (Fig. 7).
The mechanism behind the unusually high 17,20-lyase activity of XeCYP17 relative to other CYP17 isoforms is an intriguing issue. One clearly important factor that regulates 17,20-lyase activity of human CYP17 is the presence of the co-factor protein cytochrome b 5 (18,42). Perhaps the high 17,20-lyase activity of XeCYP17 is because of differences in the influence of or dependence on cytochrome b 5 as a cofactor for this reaction. Alternatively, Xenopus CYP17 may contain sequences that enhance its ability to bind to 17-hydroxylated steroids or other important cofactors. Given its relatively high 17,20-lyase activity, the Xenopus isoform may serve as a useful tool in teasing apart the mechanisms controlling the two enzymatic activities of the CYP17 enzyme.
The exclusive expression of XeCYP17 in oocytes suggests an unusual mechanism of sex steroid production driving oocyte maturation in the frog ovary. Taking into account the results reported in this study, we propose a model for steroid biosynthesis and steroid-induced maturation of oocytes (Fig. 8). In this model, pregnenolone would be produced in the follicular cells. Because pregnenolone is inefficiently converted to progesterone (Fig. 6), even in the presence of a CYP17 inhibitor (Table II), very little progesterone is likely being produced by 3␤-HSD at any time. Because the follicular cells do not express CYP17, pregnenolone must then enter the surrounding oocytes to be converted to DHEA. Additionally, because CYP17 is equally active in the ⌬4 pathway, any small amounts of progesterone that are produced by the follicular cells would be rapidly converted to AD in the oocyte, thus further preventing significant accumulation of progesterone. DHEA and AD would then be transported back to the follicular cells, where 17␤-HSD and 3␤-HSD would complete testosterone synthesis. Finally, testosterone from the follicular cells would re-enter the oocyte to promote its maturation. Because AD is also capable of promoting maturation, one cannot rule out the possibility that AD also plays a role in oocyte maturation in vivo; however, given the significantly higher potency and ovarian concentrations of testosterone relative to AD (8), testosterone is most likely the primary physiologic mediator of maturation in Xenopus oocytes. Although progesterone appears capable of promoting maturation in vitro, the lack of significant progesterone production at all times by ␤-HCG-stimulated frog ovaries, even in the presence of a CYP17 inhibitor, argues against a major role for progesterone in oocyte maturation in vivo.
Our model bears some similarity to the two-cell models put forth to explain sex steroid biosynthesis in other systems (43,44). In this case, we do not actually know how many different cell types exist within our follicular fraction; however, it is quite clear that these cells contain all of the important steroid synthetic enzymes except CYP17, whereas oocytes contain CYP17 but no other relevant activities. It is still possible that our oocyte preparations contain a small population of follicular cells that contain all of the detected CYP17 activity; however, this explanation seems less likely for the following reasons. First, this population of cells would have to be so tightly associated with oocytes that it was completely resistant to separation by both collagenase and trypsin, as the follicular cell preparations contained no detectable CYP17. Second, this contaminating population of cells would have to be very small, as it was undetectable by both stereoscopic and histologic examination. Third, in order for such a small contamination to be mediating the high velocities recorded in Table I, which are very similar to the velocities of other CYP17 enzymes overexpressed in fibroblast or COS cells, as well as with those seen in cultured thecal cells (45)(46)(47), the turnover rate of the Xenopus CYP17 would have to be extremely high. We know that this is not the case, as comparison of the human and Xenopus CYP17 17␣-hydroxylase activities in HEK cells (Fig. 5), and recently in yeast microsomes (data not shown), reveals that they have nearly identical turnover rates (ϳ6 min Ϫ1 ). Finally, progesterone injected directly into oocytes is immediately converted to AD (8), arguing that the enzymatic activity is within the oocyte itself.
We are left with the intriguing notion that, in the frog ovary, germ cells, or oocytes, play a critical role in the production of the steroid used for their own maturation. To our knowledge, this is the first example of germ cells being directly involved in steroid production, although it has not been carefully examined in other lower vertebrates, such as fish. Because oocytes make up Ͼ90% of the ovarian volume in frogs and fish, one could speculate that such lower vertebrate animals might require their oocytes to contribute to ovarian sex steroid production. In contrast, higher vertebrates, in which the ovarian volume primarily consists of follicular cells, may no longer need oocytes to subserve this function.
Interestingly, this concept of oocytes actively participating in their own maturation (in our case through the production of androgens) is consistent with the recently described work in mammalian systems, where oocytes have been shown to communicate with surrounding somatic cells to promote granulosa cell proliferation and differentiation (48). Oocytes may therefore utilize many different mechanisms to assist in orchestrating proper follicular development and subsequent ovulation.
Finally, these studies further explain and support earlier work suggesting that, although progesterone is capable of promoting maturation in vitro, androgens may be the primary mediators of maturation in vivo, where very little progesterone is ever produced. Steroid-induced maturation of Xenopus oocytes has been a puzzling field of investigation for many years, as it appears to occur independent of transcription and may involve signaling via classical steroid receptors acting outside of the nucleus (6 -8, 49). Further studies of Xenopus oocyte maturation by androgens, in addition to progesterone, may aid in finally determining the details behind this complex and fascinating process.