The induction of oocyte maturation: transmembrane signaling events and regulation of the cell cycle

The induction of oocyte maturation is known to involve an initial action of agonists at the oocyte surface. This leads to activation of a cytoplasmic maturation-promoting factor (MPF) which induces the observable events associated with maturation (reviews by Smith and Ecker, 1970; Smith, 1975; Wasserman and Smith, 1978b; Baulieu et al. 1978; Masui and Clark, 1979; Mailer and Krebs, 1980; Mailer, 1983; Gerhart et al. 1985; Masui and Shibuya, 1987). Recently, major advances have been made in the characterization of MPF and its mode of action in regulating cell cycle events. New evidence also has been obtained concerning the transmembrane signaling events that lead to MPF activation. The purpose of this article is to review and integrate these advances, partly in the context of earlier work in the field.

Oocyte maturation has been studied in a variety of vertebrate and invertebrate organisms, but the process has been investigated most intensively in amphibians. Full-grown amphibian oocytes are arrested in late G₂ of meiosis I and must progress to the second meiotic metaphase before fertilization is possible. The resumption of meiosis in vivo is brought about by the action of a gonadotropic hormone which acts on ovarian follicle cells, causing them to produce progesterone which acts directly on the oocyte to initiate the process of oocyte maturation. Similarly, progesterone induces maturation in vitro in oocytes dissected from their ovarian follicles (review by Wasserman and Smith, 1978b).

A few hours after steroid treatment, the oocyte nucleus (germinal vesicle, GV), situated near the center of the oocyte, starts to migrate towards the animal hemisphere surface and begin the process of dissolution. The arrival of the GV at the cortex causes pigment to be displaced, producing a whitish circular spot which is later delineated by a dark ring of the displaced pigment. This white spot is the first visible indication that oocyte maturation is proceeding. After dissolution of the nuclear membrane (GVBD), the condensed chromosomes align on the first metaphase spindle, complete meiosis I, and realign on the second spindle where they remain until the mature egg is fertilized or parthenogenetically activated.

Strictly speaking, an oocyte is not mature until it has progressed to the second meiotic metaphase and can be activated. However, since GVBD is the easiest event to score, it frequently has been used as the major if not sole criterion that maturation is underway. An example of normal GVBD in oocytes induced to mature with progesterone is shown in Fig. 1A. However, in oocytes from some females, and in oocytes treated with certain substances, the GV will rise to the surface, displacing pigment, but it does not undergo GVBD. An example of this phenomenon has been reported by Coleman et al. (1981) after injection of cytochalasin or colchicine into oocytes and by Muramatsu et al. (1989) who injected oocytes with cloned DNA. The latter authors refer to this ‘white spot’ as indicative that maturation has been initiated. While it is not clear what process(es) allow GV migration to the oocyte surface (see Lessman, 1987), migration per se is not indicative of maturation. At the other extreme, several instances have been reported in which GVBD actually occurs in response to a stimulus (Deshpande and Koide, 1982) but the oocytes also exhibit obvious degenerative changes which precede the onset of GVBD (Fig. 1B). The physiological significance of GVBD in such oocytes is difficult to evaluate.

The timing of GVBD can vary considerably in oocytes from different females. This can result in part from differing environmental conditions under which animals are maintained in different laboratories. In addition, a variety of diverse media have been used to culture oocytes and the time at which GVBD occurs after agonist treatment in oocytes in different media can vary as much as 2-fold (Varnold and Smith, unpublished data). Finally, the injection of gonadotropins into females, either to induce ovulation (human chorionic gonadotropin) or to improve the synchrony of response to progesterone (pregnant mare serum gonadotropin) can dramatically alter the time of GVBD. Reynhout et al. (1975) reported that oocytes from females not injected with hCG for at least 4-6 weeks (unstimulated females) usually exhibit GVBD beginning about 6h after progesterone treatment. In animals injected only a few days earlier, GVBD occurred within 2h of progesterone exposure and occasionally oocytes from such females matured spontaneously as a result of dissection from their follicles (no progesterone exposure). At the other extreme, oocytes from animals maintained for some time in the absence of obvious stimulation may exhibit GVBD at times greater than 10 h after progesterone exposure. Because of such variability, data frequently have been normalized such that the time from progesterone addition to that at which 50 % of the oocytes exhibit GVBD has been set at 1.0 (Wasserman and Masui, 1975). While this normalization allows a comparison of data from several different laboratories, it obscures a more fundamental point.

If one assumes that oocytes must progress through a certain sequence of events in order for GVBD to occur, then in those cases in which the time of GVBD is relatively soon after progesterone exposure, it seems reasonable to suggest that critical early events either have been bypassed or have already occurred prior to progesterone treatment, i.e. oocytes are downstream in the pathway that results in maturation. Conversely, in cases in which GVBD is very slow in response to stimulation, it would appear that the agonist acts upstream in the pathway. Stated differently, there may be more than one point in the putative pathway leading to GVBD and subsequent meiosis at which oocytes can be arrested. How, then, does one determine at which point oocytes are arrested? Short of defining all of the steps involved in the initial induction process and those involved in MPF activation, this question cannot be answered definitively. However, we have tended to view oocytes taken from unstimulated females as the benchmark.

In addition to progesterone which is the physiological inducer of maturation (Schuetz and Glad, 1985), several other steroid hormones as well as a large number of seemingly diverse drugs and chemicals are reported to induce oocyte maturation. A partial list of these various agonists is compiled in Table 1, which also indicates which inducers are effective when microinjected into oocytes. Additional lists could be compiled of agents that inhibit progesterone-induced maturation as well as those that both speed up and retard the timing of GVBD in response to progesterone. It should be emphasized that, in most cases, the only assay for maturation which has been used is GVBD. Furthermore, in many cases, the time of GVBD relative to progesterone controls has not always been monitored. Thus, the various ‘agonists’ could act at various points in the putative pathway leading to GVBD. Nevertheless, a variety of the nonhormonal compounds listed are known to be active at the surface of cells (Baulieu et al. 1978). These could function non-specifically to alter the activity of membrane-bound enzymes, possibly by modifying hydrophobic lipid-protein interactions. In this sense, several of the nonhormonal compounds listed in the table are reported to decrease membrane adenylate cyclase activity (Schorderet-Slatkine et al. 1982) as well as membrane protein kinase C activity (Mori et al. 1980). Thus, they could mimic the effects of progesterone on the oocyte surface. The validity of such a suggestion could be tested more directly if the nature of the progesterone target in the oocyte membrane was well established.

Reports that a specific progesterone receptor exists within a melanosome fraction derived from oocytes (see Coffman et al. 1979) are provocative since melanosomes are localized in the oocyte cortex. The equilibrium dissociation constant (K_d) for such receptors was about 10⁻⁸M, consistent with estimates of the ovarian progesterone concentration in animals stimulated to ovulate with gonadotropins. Similarly, the minimum concentration of progesterone that induces maturation in vitro is about 10⁻⁸ M. However, oocytes of the albino mutant of Xenopus, which lack melanosomes as well as premelanosomes, are induced to mature by progesterone. Further, the melanosome component that binds progesterone is melanin itself (Coffman et al. 1979). Thus, the melanosome ‘receptors’ play no physiological role in oocyte maturation.

Sadler and Mailer (1982) reported identification of a steroid receptor on the surface of Xenopus oocytes which could be photoaffinity-labeled with the synthetic progestin R5020, a relatively weak agonist for maturation. In this case, a contribution from melanosomes was eliminated by manually isolating the plasma membrane-vitelline envelope complex which does not contain pigment. They identified a 110x10³ M_r protein which bound the steroid with a K_d of 10⁻⁵M to 10⁻⁶M, depending on the procedure, and which existed at a concentration of 4.2×10¹¹ sites/oocyte (6.5×10⁴ sites um²). While this K_d is close to the effective concentration of R5020 that induces maturation, both the μM K_d and the relatively high concentration of putative receptors are more characteristic of non-specific association of steroids with a membrane protein. Thus, in the absence of more compelling evidence, it is difficult to conclude that steroids (or other agonists) induce maturation by binding to a specific high-affinity receptor.

Transmembrane signaling events usually involve interaction of an agonist with membrane receptors which, mediated by GTP-binding proteins (G proteins), leads to changes in the intracellular concentration of second-messenger molecules. The evidence concerning membrane receptors was discussed above and data pertaining to G proteins in the oocyte is evaluated in several of the subsequent pages.

The two second messenger systems studied most intensively in the oocyte have been those involving cAMP and [Ca²⁺]₁. Changes in cAMP regulate the activity of cAMP-dependent protein kinase (PKA) (Krebs, 1972) while alterations in [Ca²⁺]₁ lead to modulation of several enzymes via the calcium-binding protein calmodulin (Blackshear et al. 1988). Intracellular calcium levels are regulated by the second message inositol trisphosphate (IP₃), produced by hydrolysis of membrane-bound phosphatidylinositol 4, 5-bisphos-phate (PIP2). Phosphodiesterase cleavage of PIP₂ also generates yet another second message, 1,2-diacylglycerol (DAG), which is involved in regulation of protein kinase C activity (Berridge, 1986; Nishizuka, 1986). The PKA, Ca-calmodulin, and PKC pathways all have been implicated in the initial response of oocytes to agonists which induce maturation. As discussed below, current evidence suggests that only two of these, PKA and PKC, are actually involved in the induction process, and both appear to be affected by progesterone.

Several lines of evidence originally led to the hypothesis that an elevation in [Ca²⁺], might be necessary and sufficient to induce oocyte maturation (reviews by Cork et al. 1987; Cicirelli and Smith, 1987). First, iontophoresis of calcium into the oocyte cortex or incubation in the ionophore A23187 have been reported to induce maturation. Second, tracer flux studies showed that the rate of ⁴⁵Ca²⁺ efflux from preloaded oocytes increased within minutes after progesterone exposure, suggesting a release of bound calcium in response to the hormone. Third, progesterone was observed to induce a transient rise in [Ca²⁺]_i within minutes after exposure to progesterone in a significant percentage of oocytes injected with the calcium photoprotein aequorin. Finally, several laboratories reported that oocytes injected with the calcium-binding protein calmodulin were induced to undergo maturation in the absence of steroid treatment, although the percentage of oocytes that responded was quite variable.

In contrast to the above experiments, Robinson (1985) was unable to detect any changes in [Ca²⁺]₁ during oocyte maturation using calcium-sensitive electrodes. In a subsequent repeat of the aequorin experiment (Cork et al. 1987), a small transient increase in [Ca²⁺]₁ was observed as an early response to progesterone, but in only one oocyte. In all other cases, no change was observed. Nevertheless, maturation always occurred in response to progesterone. Thus, while a transient elevation in [Ca²⁺]₁ could be observed, it was not an essential response to progesterone.

It now seems clear that an elevation in [Ca²⁺]₁per se is not sufficient to induce maturation. For example, in the experiments described above with A23187, maturation was obtained only when the concentration of Ca²⁺ (and Mg²⁺) in the medium was relatively high (10−20mM). Treatment of oocytes incubated in lmM-Ca²⁺ with A23187 does not induce maturation although it does cause an increase in [Ca²⁺]₁ (Cicirelli and Smith, 1987). Also, injection of oocytes with IP3 which releases bound Ca²⁺ in Xenopus oocytes (Berridge, 1988), does not induce maturation (Picard et al. 1985). While the calcium efflux data referred to above were interpreted as indicating an increased calcium ion concentration in the cytoplasm, an alternative interpretation is that increased efflux results from an increased efficiency of the calcium extrusion pump induced by progesterone (O’Connor et al. 1977). This would imply an actual decrease in [Ca²⁺]₁ in response to the steroid.

There is general consensus that a transient decrease in cAMP levels, resulting from an inhibition of adenylate cyclase activity, is an obligatory first step in the mechanism by which progesterone induces oocyte maturation. This in turn is thought to lead to a decrease in cAMP-dependent protein kinase activity (PKA) which would result in the dephosphorylation of a putative maturation-inhibiting phosphoprotein. As with the calcium story, the evidence supporting this sequence of events falls into four categories (review by Mailer, 1983). First, activators of adenylate cyclase, such as cholera toxin, which elevate cAMP levels in Xenopus oocytes, inhibit progesterone-induced maturation. Second, oocyte cAMP levels are reported to decrease within minutes after progesterone exposure. Third, progesterone is reported to partially inhibit adenylate cyclase activity. Finally, injection of the regulatory subunit of PKA, expected to reduce activity of the catalytic subunit in oocytes, induces maturation, while injection of the catalytic subunit inhibits progesterone-induced maturation.

A scheme depicting the events described above is shown in Fig. 2. Additional data supporting the model comes from the observation that injection of the heatstable inhibitor protein (PKI) of the catalytic subunit into oocytes also induces maturation. Finally, injection of phosphatase inhibitor proteins 1 and 2, which might be expected to prevent dephosphorylation of the putative inhibitory phosphoprotein, delays GVBD in progesterone-stimulated oocytes (see Mailer, 1983). In this context, inhibitor-1 is activated by PKA, suggesting that a decrease in cAMP could also lead to increased phosphatase-1 activity. This would enhance déphosphorylation of the putative meiosis-inhibitory protein (see Cyert and Kirschner, 1988). In contrast to the above, several recent studies have indicated that a decrease in cAMP content is not always an obligatory or even sufficient step in the induction of maturation (see later section). This has led to the suggestion that oocytes may contain a pathway independent of that involving cAMP which can result in maturation. In view of this, it seems desirable to discuss in more detail certain aspects of the data supporting the involvement of the cAMP pathway in oocyte maturation.

Reports of cAMP changes in oocytes as a response to progesterone have been quite variable, but most have shown a modest decrease of about 20 % within minutes after progesterone exposure (Mailer, 1983; Cicirelli and Smith, 1985; Gelerstein et al. 1988). Since the endogenous cAMP pool in stage 6 oocytes averages between 1.5−2.5 pmole oocyte^-1, these data suggest a progester-one-induced decrease of 0.3−0.5 pmole within minutes after agonist stimulation.

The mechanism by which progesterone could cause such decreases in unclear. An increase in the phosphodiesterase activity, which hydrolyzes cAMP, is reported not to occur in response to progesterone (Sadler and Mailer, 1987). On the other hand, several laboratories have observed that adenylate cyclase activity in isolated oocyte membranes is altered by progesterone (see Mailer, 1983) in that the steroid can prevent stimulation of activity by cholera toxin. However, the effects of progesterone on basal activity were quite variable. This is due in part to the very low level of basal activity (0.05 pmol oocyte^-1 h^-1) observed in control oocytes (no cholera toxin or progesterone exposure) (Sadler and Mailer, 1981). Thus, even if progesterone resulted in complete inhibition of this activity, several hours would be required to generate the reported decreases in cAMP content.

The most thoroughly characterized GTP-binding proteins (G proteins) involved in transmembrane signaling events are those associated with regulation of adenylate cyclase activity (Freissmuth et al. 1989). Xenopus oocytes contain the two guanine nucleotidebinding proteins, G_s and G_i, normally associated with regulation of adenylate cyclase. However, progesterone does not appear to regulate cAMP levels through these proteins in a conventional manner. For example, pertussis toxin, which inhibits G_i, elevates oocyte adenylate cyclase activity but does not prevent inhibition of adenylate cyclase by progesterone (Sadler et al. 1984). Curiously, the toxin is reported not to affect oocyte cAMP levels (Mulner et al. 1985), but is reported to both delay (Sadler et al. 1984) and accelerate (Mulner et al. 1985) the time course of progesterone-induced GVBD. One solution to these apparent discrepancies is the suggestion by Sadler and Mailer (1983) that the mechanism by which progesterone acts on oocyte adenylate cyclase is novel, with properties in common both with the P site described for adenosine agonists as well as the more conventional receptor-mediated systems. They suggest that progesterone decreases the ‘tum-on’ reaction of G_s, i.e. decreases the rate of guanine nucleotide exchange on the G_s protein (also Jordana et al. 1984). However, even if correct, this mechanism again would not account for the magnitude of the decrease in cAMP content observed after progesterone treatment.

R₂C₂ + n(cAMP) R(cAMP)_n + 2C (1) Several lines of evidence suggest that PKA is fully dissociated in the oocyte (Masaracchia et al. 1979; Huchon et al. 1981). Thus, elevating cAMP levels should have little effect on catalytic activity of the kinase. Accordingly, it is not clear why agents such as cholera toxin, that increase oocyte cAMP should also inhibit maturation, unless inhibition is unrelated to effects on cAMP levels.

Based on the equilibrium reaction in equation (1), injecting R subunits into oocytes would be expected to drive the reaction to the left, decreasing catalytic activity and releasing cAMP; cAMP levels would increase temporarily. In the experiments reported by Mailer and Krebs, R subunit of the type II isozyme was injected into oocytes as the subunit-cAMP complex. Assuming no degradation of the protein after injection, sufficient subunit was injected to result in a decrease in catalytic activity of 20−50 %. This point was not tested directly, although Masaracchia et al. (1979) reported a decreased level of free catalytic subunit in extracts from progesterone-treated oocytes. In contrast, Cicirelli et al. (1988) reported no change in PKA activity during the entire course of oocyte maturation. Masaracchia et al. (1979) also reported that the regulatory subunit from type I PKA does not induce maturation when injected into oocytes. This would be somewhat surprising since RI also binds to C. However, we have observed that the type I regulatory subunit does induce oocyte maturation, albeit less efficiently than that from type II, when injected into oocytes (Varnold and Smith, unpublished data).

One might expect that injection of excess catalytic subunit also would drive the reaction in equation (1) to the left, resulting in a decrease in catalytic activity. This appears not to be the case. There are suggestions that injection of C into oocytes alters the phosphorylation of numerous proteins, both quantitatively (see Mailer, 1983) and qualitatively (Boyer et al. 1987). However, identification of these substrates and their role in maturation has not been accomplished. Mailer and Krebs (1980) state that injected catalytic subunit severely depresses protein synthesis; this alone would inhibit maturation. They suggest that the putative inhibitory phosphoprotein might function to prevent the synthesis of essential proteins early in the maturation response. In summary, while the data presented by Mailer and Krebs (1977) provides strong support for the sequence of events depicted in Fig. 2, many questions remain unanswered. Clearly, identification and characterization of the putative maturation-inhibitory phosphoprotein would go a long way towards answering these questions.

There exist several examples in which oocyte maturation has been obtained without obvious involvement of the cAMP pathway. For example, Birchmeier et al. (1985) reported that the oncogenic protein encoded by H-ras^{val 12} induced maturation when injected into Xenopus oocytes without any corresponding change in cAMP levels. It should be pointed out that the oncogenic ras protein did not induce GVBD in cholera toxin-treated oocytes indicating that ras protein does not overcome the effects of elevating cAMP levels. Nevertheless, Birchmeier et al. (1985) suggested that oocytes contain an alternate pathway able to trigger meiosis, which bypasses changes in intracellular cAMP levels. In support of this, the injection of a monoclonal antibody against the oncogenic ras protein is reported to inhibit insulin-induced but not progesterone-induced maturation (Korn et al. 1987; Deshpande et al. 1987).

Allende et al. (1988) reported that activated ras protein (H-ras^{val 12}) induces GVBD in Xenopus oocytes incubated with cycloheximide or puromycin, as does MPF, while both inhibitors block maturation induced by progesterone. This suggests the possibility that ras protein acts downstream of the protein synthesis requirement in oocyte maturation, possibly at the level of MPF, rather than at a point in the initial induction process. This observation has not been confirmed and is suspect on other grounds. If the oncogenic protein was acting downstream of the protein synthesis requirement, it seems reasonable to anticipate that maturation (GVBD) would be faster in ras-injected oocytes than in those exposed to progesterone. That is not the case (Birchmeier et al. 1985). In addition, while both the protein and ras^val12 transcript induce maturation in stage 6 oocytes, neither activates MPF (induces GVBD) in smaller stage 4 oocytes (Johnson and Smith, unpublished data) which contain preMPF (Taylor and Smith, 1987). This implies an action by ras protein early in the sequence of events induced by progesterone, i.e. prior to MPF activation.

Gelerstein et al. (1988) demonstrated that treatment of oocytes with acetylcholine (ACh) shortly after progesterone exposure caused GVBD sooner than in oocytes exposed to progesterone alone, and also rapidly lowered intracellular cAMP levels. However, ACh alone did not induce maturation suggesting that a decrease in cAMP per se is not sufficient to trigger maturation. In contrast, addition of adenosine to oocytes elevated endogenous cAMP levels and abolished the progesterone-induced decrease in cAMP. Nevertheless, adenosine alone induced GVBD although the time of GVBD was slower than that observed in response to progesterone. Gelerstein et al. (1988) suggested the coexistence in oocytes of different and parallel mechanisms for the induction of maturation.

The tumor-promoting phorbol ester TPA (12-o-tetra-decanoylphorbol 13-acetate) can substitute for DAG in activating protein kinase C and Stith and Mailer (1987) reported that oocytes treated with TPA undergo GVBD in the absence of hormone treatment. This result suggests that activation of the PKC pathway can induce oocyte maturation. However, when PKC isolated from bovine tissue was injected into oocytes it neither induced maturation nor changed the time of GVBD in oocytes treated at the time of injection with progesterone. It did result in GVBD within a shorter time after treatment of oocytes also with insulin compared to oocytes induced to mature with insulin alone. Based on this observation and other data, Stith and Mailer suggested that while the PKC pathway is not involved in progesterone-induced maturation, insulin might work by this route.

With respect to PKC injections, Muramatsu et al. (1989) have reported that a mutant construct of PKC which expresses activity in the absence of phorbol ester activation initiates maturation when injected into Xenopus oocytes. However, that effect involved only migration of the oocyte GV to the animal hemisphere surface and not GVBD. We have confirmed the observation of Stith and Mailer (1987) that injection of a mixture of PKC isozymes (rat brain) at the time of progesterone treatment has no effect on maturation. However, when PKC is injected into oocytes 30-60 min prior to progesterone exposure, the time of GVBD is significantly delayed (Varnold and Smith, unpublished data). It should be pointed out that injected proteins must diffuse through the cytoplasm to potential sites of action before exerting an effect; injecting prior to progesterone exposure may provide sufficient time for diffusion. Thus, one interpretation of these experiments is that elevated PKC activity is inhibitory during early times after progesterone treatment.

The observation that TPA induces GVBD has not been confirmed. In fact, Bement and Capeo (1989) have observed that treatment of oocytes containing follicle cells with TPA results in cytolysis. We also have observed that such treatment leads to extreme mottling of the pigmented animal hemisphere (Fig. 1C) but that such oocytes usually contain an intact GV (Varnold and Smith, unpublished data; see also Stith and Mailer, . Bement and Capeo further observed that, in oocytes with no follicle cells, the phorbol ester induced cortical granule breakdown and elevation of the vitelline envelope, a typical response to activation stimuli. In no case was GVBD obtained.

Maturation in starfish oocytes is induced by the action of 1-methyladenine acting on the oocyte surface, resulting in the activation of MPF (Kishimoto and Kanatani, 1976). In essentially all aspects, the process parallels that observed with amphibian oocytes. In this case, phorbol esters inhibit the induction of maturation by 1-MA (Kishimoto et al. 1985). In contrast, surf clam oocytes are shed with an intact GV and GVBD is induced by fertilization; TPA is reported to induce GVBD is these oocytes (Bloom et al. 1988).

In making comparisons among diverse organisms, it should be emphasized that, while all of the oocytes mentioned contain an intact GV at the time maturation is induced, it is less clear that they are arrested in meiosis at the same point relative to the G2/M transition. For example, full-grown Xenopus oocytes contain active lampbrush chromosomes while, in the surf clam, oocytes contain chromosomes already partially condensed, i.e. they are downstream in the sequence of events leading to meiosis I compared with Xenopus. Thus, if the TPA data overall is taken at face value, it might suggest that both a decrease and increase in PKC activity are involved in maturation, depending on the point in the pathway at which oocytes are arrested.

Recently, Cicirelli and Krebs (unpublished data) observed that sphingomyelinase is a potent inducer of maturation in Xenopus oocytes and we have confirmed this observation (Varnold and Smith, 1989, unpublished data). Sphingomyelinase acts on membrane sphingolipids to produce ceramide, which in turn is hydrolyzed to sphingosine, a potent inhibitor of PKC (Hannun et al. 1986) as well as Ca-calmodulin-dependent enzymes (Jefferson and Schulman, 1988). We have observed further that sphingosine induces oocyte maturation, as does staurosporine, also an inhibitor of PKC (Varnold and Smith, 1989, unpublished data). Taken together, these observations again suggest the possibility that inactivation of PKC, similar to the situation proposed for PKA, could represent a mechanism for the induction of oocyte maturation. The question then becomes, is the PKC pathway affected by progesterone or is it a separate pathway?

In response to extracellular signals, the increase in DAG which activates PKC is brought about by increased phosphodiesterase activity (phospholipase C) which hydrolyzes PIP₂. That enzyme is coupled to a G protein (Cockcroft and Stutchfield, 1989), which presumably acts like the G_s protein coupled to adenylate cyclase. By analogy, we have made the assumption that extracellular signals could act to decrease DAG levels, possibly via an inhibitory G protein that reduces phospholipase C activity. This would result in a decrease in PKC activity. On that basis, Varnold and Smith (1989; unpublished data) measured the mass of DAG in Xenopus oocytes as a function of time after progesterone exposure. The results are shown in Fig. 3. Within 15 s after exposure of full-grown (stage 6) oocytes to progesterone, DAG levels decreased by about 30% compared to control oocytes. The DAG levels remained low for at least the first 2min, returned to control levels by about 15 min, and then begin to continuously increase up to the time of GVBD. Also shown in the figure are the effects of progesterone on DAG levels in stage 4 oocytes (800/rm in diameter), which are not responsive to progesterone. In this case, there is no change in DAG mass, arguing against a totally nonspecific effect of steroids on the oocyte membrane. A similar decrease in IP₃ mass was observed in response to progesterone (data not shown). Thus, the data imply that progesterone acts rapidly to reduce hydrolysis of PIP₂, resulting in a decrease in the level of both second messengers.

The data described above were obtained by extracting lipids from whole oocytes and thus would have measured DAG changes in both the cytoplasm and oocyte membrane. In additional experiments, we measured changes in DAG mass in oocyte plasma membranes isolated manually (Sadler and Mailer, 1981). The membrane DAG averaged about 6% of total DAG mass in the oocyte. However, when these membranes were exposed to progesterone, DAG levels also began to decrease with 15 s and continued to decrease, reaching a level 41% of controls by 5 min (Fig. 3). The continued decrease would be expected since replacement of membrane inositol phospholipids from the cytosol would not occur in isolated membranes. This early decrease in DAG in oocyte membranes supports the hypothesis that a decrease in PKC activity, while not yet measured directly, is a very early response of oocytes to progesterone.

A summary of data discussed on the several preceding pages is as follows. Progesterone induces an early transient decrease in the content of a known second messenger molecule, cAMP, which then is presumed to result in decreased PKA activity. The same agonist also induces an early transient decrease in another known second messenger molecule, DAG, which is presumed to result in decreased activity of a second protein kinase, PKC. Unlike cAMP, DAG levels then increase as oocytes approach GVBD, implying that increased PKC activity might be involved in later events of maturation.

At least two models can be considered in which both the PKA and PKC pathways would be involved in the induction of oocyte maturation by a single agonist. These are shown schematically in Fig. 4. The first suggests that the putative maturation-inhibitory protein is phosphorylated at multiple sites by both kinases. This idea is not without precedent as several examples exist in the literature in which a single substrate is phosphorylated by both protein kinases (Nishizuka, 1986). Thus, meiotic arrest could be released by decreasing the total phosphate content of the protein via inhibition of either (or both) of the kinase pathways. The second model is based on many observations indicating extensive cross-talk between the PKC and PKA pathways (Nishizuka, 1986). For example, changes in PKC activity have been linked to modulation in adenylate cyclase activity (Yoshimasa et al. 1987; Rozengurt et al. 1987) . Reciprocally, examples exist in which changes in cAMP alter phospholipid hydrolysis and PKC activity (Supattapone et al. 1988; Kato et al. 1989). In postulating interactions within the oocyte, one is influenced by the observation that the progesterone-induced decrease in DAG mass appears to precede the decrease in cAMP levels. This suggests the possibility that interaction between the two pathways is sequential, i.e. a decrease in PKC activity would lead to a decrease in PKA activity. This could be mediated by PKC effects on cAMP levels due in part to inactivation of adenylate cyclase.

Both models clearly are speculative at this point. However, the essential features of each model appear to be testable. Furthermore, the involvement of two pathways in the induction of maturation, especially one (PKC) that can be regulated by several products of membrane lipid catabolism (Blackshear et al. 1988) would help explain how a relatively large number of diverse ‘agonists’ can induce oocytes to mature.

Early experiments on Rana pipiens and Xenopus laevis oocytes established that MPF activity appears in the cytoplasm prior to GVBD, appears at the same time in manually enucleated oocytes and can be continually amplified through serial cytoplasmic transfers into recipients not exposed to hormone (review by Wasserman and Smith, 1978b; Masui and Clark, 1979). The latter observation in particular suggested that oocytes contain a store of inactive MPF which can be activated and amplified by small amounts of active MPF. Reynhout and Smith (1974) and Wasserman and Masui (1975) further demonstrated that injection of cytoplasm containing MPF activity always induced precocious GVBD in recipient oocytes. Since the initial appearance of MPF activity, but not amplification, requires protein synthesis (Wasserman and Masui, 1975; Gerhart et al. 1984), this suggests that much of the time lag between progesterone treatment and GVBD is involved in production of active MPF, i.e. injected MPF can bypass early progesterone-induced steps. The ability of MPF to induce maturation in the absence of protein synthesis is unique and has been used as a diagnostic assay for MPF activity (Gerhart et al. 1984).

Wasserman and Smith (1978) first reported that MPF activity cycles during the mitotic divisions of early cleavage in both Rana pipiens and Xenopus laevis embryos. The peak of MPF activity coincided approximately with M-phase of the cell cycle, and activity disappeared when cells completed mitosis. They showed further that protein synthesis was necessary for reappearance (but not disappearance) of MPF activity at each division, suggesting either that an activator of MPF, or MPF itself, must be synthesized during each cell cycle. Shortly thereafter, Sunkara et al. (1979) showed that synchronized HeLa cells contain MPF activity, assayed using Xenopus oocytes, which appeared in G2, peaked at metaphase, and disappeared in Gi of the cell cycle. These key experiments and several additional studies on MPF in diverse cells (Nelkin et al. 1980; Weintraub et al. 1982; Doree et al. 1983; Kishi-moto et al. 1982, 1984; Gerhart et al. 1984; Sorensen et al. 1985) established the generality of MPF as an M-phase promoting factor in mitotic as well as meiotic cells (Gerhart et al. 1984, 1985).

Attempts to purify MPF began about 15 years ago, but in spite of efforts in several laboratories (Wasserman and Masui, 1976; Drury, 1978; Wu and Gerhart, 1980; Adlakha et al. 1985; Nguyen et al. 1986), progress until very recently has been very slow. The two major difficulties have been the extreme lability of MPF and the fact that the only assay involved microinjection into Xenopus oocytes. The development of systems to monitor nuclear membrane breakdown in vitro greatly facilitated assays of MPF activity (Lohka and Mailer, 1985; Miake-Lye and Kirschner, 1985). Concerning lability, Wu and Gerhart (1980) developed procedures in which MPF activity could be obtained by gentle homogenization and in which MPF activity survived limited dilution. They purified MPF about 100-fold and suggested it might be a phosphoprotein with an apparent relative molecular mass of about lOOxlO³. The breakthrough in purification was achieved by Lohka et al. (1988). By using essentially the conditions developed by Wu and Gerhart (1980) and a very large amount of material, they obtained small amounts (1 % yield) of a fraction purified approximately 3000-fold. That preparation contained two predominant proteins, with relative molecular masses of 34 and 45X10³, respectively.

The 34K protein has been identified as the Xenopus homolog of a fission yeast protein encoded by a gene (cdc2⁺), which is required for the G2/M transition in the mitotic cell cycle (Gautier et al. 1988; Dunphy et al. 1988) . It has now been implicated as a component of MPF in starfish (Labbe et al. 1988, 1989; Arion et al. 1988) and clam oocytes (Draetta et al. 1989) as well as in mouse (Morla et al. 1989) and human (Draetta and Beach, 1988; Brizuela et al. 1989) tissue culture cells. The cdc2 protein (p34^cdc2) is a serine/threonine protein kinase, which can phosphorylate a number of substrates under varying conditions but exhibits a strong preference for histone H1 as cells progress from G₂ to M. The 45K protein, which is also a substrate for p34^cdc2, has not been identified definitively, but is thought to be analogous to a protein called cyclin.

Cyclins were first identified in sea urchin and clam embryos as products of maternal mRNA which are synthesized and accumulated during interphase and then rapidly degraded near the end of each mitosis (Evans et al. 1983). They have now been identified in several eukaryotic cells including yeast (Goebl and Byers, 1988; Solomon et al. 1988), Xenopus (Minshull et al. 1989; Murray and Kirschner, 1989), Drosophila (Lehner and O’Farrell, 1989; Whitfield et al. 1989), starfish (Standart et al. 1987) and probably in mammalian tissue culture cells (Draetta and Beach, 1988). Based on the nucleotide sequence of cDNAs from clams, they fall into two classes, A and B, with predicted relative molecular masses of 42K and 48K, respectively (Swenson et al. 1986; Westendorf et al. 1989) . However, they usually display apparent relative molecular masses on SDS gels of about 55K; the putative cyclin in mammalian cells is a protein of 62K.

Swenson et al. (1986) originally reported that the mRNA for clam cyclin A induces the resumption of meiosis when injected into Xenopus oocytes. Similar results now have been obtained with the sea urchin cyclin mRNA (Pines and Hunt, 1987) and the mRNA for clam cyclin B (Westendorf et al. 1989). These results suggested that cyclin either is a component of active MPF or that it functions as an activator of MPF. Additional observations suggest that interaction of the cyclins with p34^cdc2 is necessary for the histone H1 kinase associated with MPF to be active (Draetta et al. 1989; Meijer et al. 1989; Brizuela et al. 1989). For example, both cyclins A and B in clam oocytes are found in association with p34^cdc2 as assayed by immunoprecipitation with either anti-cyclin A, anti-cyclin B, or anti-cdc2 sera (Draetta et al. 1989). Meijer et al. (1989) also have demonstrated that anti-cyclin antibodies precipitate p34^cdc2 in sea urchin eggs. Finally, anti-cdc2 sera coprecipitates p34^cdc2 and a protein (p62) thought to be cyclin in mitotic HeLa cell extracts (Draetta and Beach, 1988).

Protein synthesis is required for the appearance of active MPF in mitotically dividing cells (see Swenson et al. 1989). Since p34^cdc2 is constitutively present in dividing cells, this would imply that synthesis and degradation of cyclin regulates the activity of MPF. Recently, this view has been confirmed with the demonstration that cyclin is the only newly synthesized protein necessary to induce MPF activity and drive the cell cycle in a Xenopus egg extract (Minshull et al. 1989; Murray and Kirschner, 1989; Murray et al. 1989). Several additional lines of evidence show that the rapid degradation of cyclin at the end of each M phase is required for loss of MPF activity (Draetta et al. 1989; Murray et al. 1989; Luca and Ruderman, 1989). This view of cyclin regulation of the cell cycle is shown in Fig. 5.

Since the induction of maturation in Xenopus oocytes also requires protein synthesis, one might suggest that progesterone stimulation leads to the synthesis of cyclin, which then results in active MPF. This possibility remains uncertain. Frog, starfish, clam and mouse oocytes all are arrested at the G₂/M border of meiosis I. In response to the appropriate stimulus, MPF activity appears, induces meiosis I, and then cycles between meiosis I and II (Gerhart et al. 1984; Hashimoto and Kishimoto, 1988; Arion et al. 1988; Draetta et al. 1989). Protein synthesis is required in all cases for entry into meiosis II but, with the exception of frog oocytes, not for meiosis I. Nevertheless, at least in clam oocytes, both cyclins A and B are normally synthesized during the transition from G₂ to meiosis I (Westendorf et al. 1989) . A resolution to this discrepancy is provided by the observation that clam oocytes contain a store of cyclin B protein which is present as large aggregates. Shortly after the induction of meiosis, cyclin B protein appears in a more soluble disperse form, which allows it to interact with stored cdc2 protein, generating active MPF (Westendorf et al. 1989).

Westendorf et al. (1989) speculate that Xenopus oocytes also might contain a pool of cyclin protein, which is released in response to progesterone and associates with stored cdc2 protein. To explain the protein synthesis requirement, they suggest that progesterone stimulation leads to synthesis of a non-cyclin protein which somehow unmasks the stored cyclin. One candidate for such a protein is the product of the c-mos proto-oncogene (Freeman et al. 1989; Sagata et al. 1989). Thus, the rate-limiting step in MPF activation would be release of preexisting cyclin rather than synthesis and accumulation of new cyclin. While Xenopus oocytes contain maternal mRNA for cyclin B (and probably cyclin A), cyclin synthesis has not been observed in the oocyte in response to progesterone (Minshull et al. 1989); it has been observed after activation of mature eggs (Murray and Kirschner, 1989). On the other hand, the existence of stored cyclin protein in Xenopus oocytes has not been documented. These possibilities remain open questions.

Resolution of the sequence of steps involved in activation of MPF in oocytes would be greatly facilitated by a clearer understanding of the interaction between p34^cdc2 and cyclins, i.e. what is MPF? On this point, two schools of thought have developed. On the one hand, active MPF is viewed as a complex of p34^cdc2 and cyclin (Draetta et al. 1989). Thus, neither component would exhibit MPF activity independently. The evidence for this comes largely from studies referred to earlier in which antisera against one or the other of the proteins precipitates both, and the complex exhibits histone H1 kinase activity. It is not clear from such studies whether the antisera quantitatively precipitated all of the respective proteins as the complex. At least in sea urchin eggs, the content of p34^cdc2 is in large excess relative to cyclin, and the majority of the p34 is not complexed with cyclin (Meijer et al. 1989). On the other hand, Gautier et al. (1989) have observed that only about 10% of the p34^cdc2 actually functions as MPF during Xenopus oocyte maturation.

The second view is that cyclin activates p34^cdc2 but is not necessarily a stably bound component of MPF; p34^cdc2 alone could exhibit MPF activity once activated (Murray et al. 1989). This is based in part on the report that histone H1 kinase (MPF) purified from starfish oocytes contains only p34^cdc2 (Labbe et al. 1988). In addition Murray and Kirschner (personal communication) have observed that MPF activity induced by cyclin activation of pre-MPF in a Xenopus egg extract is maintained after removal of the cyclin with anti-sea urchin cyclin B antibody (Murray et al. 1989). Crucial to interpretation of this latter experiment is the complete removal of cyclin from activated MPF by immunoprecipitation. Murray and Kirschner (personal communication) found that the antibody precipitation removed only 80% of the cyclin, but with no loss of MPF activity in the supernatant. Since the assay for MPF involved microinjection into cycloheximidetreated oocytes, the results imply that p34^cdc2 alone can function as MPF in the oocyte. Further, since amplification in response to MPF can occur in cycloheximidetreated oocytes the results imply that p34^cdc2 alone can activate preMPF.

Draetta and Beach (1988) reported that the phosphorylation of p34^cdc2 and cyclin (p62) both are subject to cell cycle regulation in HeLa cells and that such regulation affects the activity of cdc2 kinase. In G₁ cells, cdc2 was unphosphorylated, not associated with p62, and inactive as a kinase. In G₂ cells, p34^cdc2 became phosphorylated on both tyrosine and threonine/serine residues and formed a complex with p62 which itself became phosphorylated, presumably by the cdc2 kinase; the complex exhibited kinase activity. By the time cell division was completed, p34^cdc2 was dephosphorylated, dissociated from p62, and protein kinase activity was no longer observed.

Since only the hyperphosphorylated form of cdc2 was observed in association with p62, Draetta and Beach (1988) suggested that complete phosphorylation of p34^cdc2 is a necessary precondition for complex formation. These observations have been extended with the report that p34^cdc2 is inactive as a histone H1 kinase until it associates with p62, although the p34^cdc2 alone does exhibit casein kinase activity (Brizuela et al. 1989). They suggest that one role of cyclin might be to confer ‘M-phase specificity’ on the cdc2 kinase.

Draetta et al. (1988) have shown that cdc2 is a major phosphotyrosine-containing protein in HeLa cells and the level of tyrosine phosphorylation is subject to cell cycle regulation. Morla et al. (1989) have shown further that while increasing tyrosine phosphorylation of cdc2 correlates with formation of the cdc2/p62 complex in mouse 3T3 fibroblasts, the complex is inactive as a histone H1 kinase. Quantitative tyrosine déphosphorylation occurs during entry into mitosis and this correlates with maximal H1 kinase activity. Further, in vivo inhibition of tyrosine dephosphorylation correlates with G₂ arrest. These observations are not restricted to mammalian tissue culture cells.

Recently, Gautier et al. (1989) have shown that maximal histone H1 kinase activity of p34^cdc2 in maturing Xenopus oocytes correlates with dephosphorylation of the protein. Conversely, phosphorylation of p34^cdc2 led to inactivation of H1 kinase activity in egg extracts that contained active MPF. Dunphy and Newport (1989) have extended these observations in showing that tyrosine phosphorylation of Xenopus cdc2 is high in interphase (oocytes) but absent during M phase (unfertilized eggs). Activation of preMPF in oocyte extracts resulted in dephosphorylation of tyrosine on cdc2 which correlated with activation of H1 kinase activity. Furthermore, the product of the yeast gene sucl (pl3), which inhibits entry into mitosis in the extracts, blocks tyrosine phosphorylation and kinase activation. While these findings support the view that tyrosine dephosphorylation of cdc2 is an important step in MPF activation, this step alone may not be sufficient. At least in HeLa cells, entry into mitosis is associated with both tyrosine and threonine dephosphorylation of p34^cdc2

The possibility considered earlier that active p34^cdc2 alone can activate preMPF appears to pose a dilemma when one considers the data discussed above. Thus, if p34^cdc2 is active as MPF only when dephosphorylated and if only phosphorylated p34^cdc2 is able to bind cyclin, what role does cyclin play in the activation of preMPF in the oocyte? One solution to this apparent dilemma could be obtained by minor modification in the sequence of events depicted above. For example, if one speculates that phosphorylation of cyclin facilitates binding to cdc2 and that cyclin binding in turn facilitates cdc2 dephosphorylation, then injection of active cdc2 would amplify MPF activity via cyclin phosphorylation, which in turn would generate more active cdc2. This suggestion necessitates that some cyclin preexists in the oocyte and that it be available as substrate for the cdc2 kinase. Obviously, this idea could be tested for by injection of purified p34^cdc2 into oocytes.

The initial interaction of steroids at the oocyte surface results in a transient decrease in two intracellular second messages, one (cAMP) involved in regulating the protein kinase A pathway and a second (DAG) which regulates a membrane-bound protein kinase (PKC). Both are serine/threonine protein kinases. Thus, the initiation of maturation, like the activation and functioning of MPF, appears to be controlled by negative regulation of protein kinases and/or activation of the appropriate phosphatases. The molecular details of this initial event remain to be fully elucidated. However, based on the discussion in the several preceding pages, a model that integrates the transmembrane signaling events with activation of MPF activity can be suggested.

Crucial to understanding the mechanism by which progesterone releases oocytes from G₂ arrest is the hypothesis, proposed by Mailer and Krebs (1977), that oocytes contain a putative phosphoprotein which maintains meiotic arrest. Progesterone-induced déphosphorylation of this putative protein would then lead to activation of MPF and subsequent meiotic events. The simplest model that integrates these events is one in which the putative phosphoprotein is p34^cdc2. Thus, within minutes after progesterone treatment, transient inactivation of one or more protein kinases would initiate the process of preMPF activation by dephosphorylating the cdc2 kinase.

Both serine/threonine and tyrosine kinases appear to be involved in cdc2 kinase activation, and both could be inactivated concurrently in response to progesterone. However, to be consistent with available data, activation of MPF is viewed as occurring by two déphosphorylation steps. First, progesterone would result in partial dephosphorylation of cdc2 by inactivation of a serine/threonine protein kinase(s). Concurrently, either synthesis of cyclin or a protein that releases cyclin would occur. The cdc2 protein at this point, still phosphorylated on tyrosine residues, would not exhibit histone H1 kinase activity but would be able to phosphorylate other substrates including cyclin (Brizuela et al. 1989; Morla et al. 1989; Meijer et al. 1989). Thus, the major function of the first step would be to phosphorylate cyclin which would facilitate binding to cdc2 kinase. The role of cyclin phosphorylation relative to its interaction with cdc2 is not clear. Murray et al. (1989) suggest that phosphorylation of cyclin is a prerequisite for cyclin degradation. However, cyclin breakdown does not appear to require phosphorylation, at least during cleavage in sea urchin embryos (Neant et al. 1989). On the other hand, Meijer et al. (1989) have demonstrated that phosphorylation of cyclin correlates well with maximal histone H1 kinase activity, suggesting that cyclin phosphorylation might be a crucial step in MPF activation. At any rate, the model suggests that once bound, phosphorylated cyclin prevents additional phosphorylation of the cdc2 by a tyrosine kinase, resulting in dephosphorylated and fully active p34^cdc with histone H1 kinase activity (Fig. 6).

The model clearly is speculative at this point, and appears to be at odds with the observations of Cyert and Kirschner (1988) that activation of preMPF in a stage 6 Xenopus oocyte extract involves two components, preMPF and an inhibitor of activation (INH), separable in different ammonium sulfate fractions. They suggest that preMPF is activated by phosphorylation and that INH inhibits the activation step, but only when it also is phosphorylated; INH could be the putative phosphoprotein. This certainly remains a possibility. On the other hand, their studies did not involve highly purified preparations and there is no direct evidence that INH is functional in vivo. More relevant, the original premise that MPF is active only when phosphorylated appears to be incorrect.

Because of the intense interest in MPF as a regulator of mitosis in actively dividing cells, less attention has been paid in the past few years to molecular details of the progesterone-dependent release from G₂ arrest in the oocyte. This also traces in part to the view that the mechanism of release might be specific to only those cell types that arrest in G₂, and that the progesteronedependent release might be specific to amphibian oocytes (Gerhart et al. 1984). However, the protein products of two oncogenes, c-mos (Freeman et al. 1989; Sagata et al. 1989) and ras (Birchmeier et al. 1986), both induce maturation when injected into oocytes, and both appear to act very early in the sequence of events induced by progesterone. These observations suggest that studies on the initial response(s) of oocytes to progesterone may have wider applicability to understanding cell-cycle regulation than previously suspected.

I thank my colleagues Peter Bryant, Hans Bode and John Scott for helpful suggestions in preparing this manuscript, and Robert Varnold for technical assistance. The original research presented in this paper was supported by NIH grant HD 04229.