Maturation (M‐phase) promoting factor (MPF) plays a pivotal role in oocytes during their maturation. This review concentrates on its function at three important time‐points. First, its activation during meiotic progression from prophase I arrest at germinal vesicle breakdown. Second, its role during the transition from meiosis I to meiosis II, a defining feature of meiosis involving segregation of homologous chromosomes. Third, maintenance of its activity at metaphase II arrest and the necessity for its destruction during oocyte activation. An understanding of how oocytes switch it on and turn it off underpins much of the basic cell biology of oocyte maturation.
What makes meiosis unique is that two consecutive cell divisions occur in the absence of an intervening period of DNA synthesis (S‐phase). In mitotic cells, S‐phase always precedes M‐phase in order to maintain euploidy. What makes meiosis interesting is that much of the basic cell cycle machinery employed is analogous to that used in mitosis. The devil therefore is in understanding the detail and defining what actually may be unique to meiosis.
This review concentrates on what is known about the events controlling the two meiotic divisions in mammalian oocytes during their maturation, taking the mouse as the model mammal. It therefore overlooks some interesting aspects of oocyte biology, such as the transition from mitosis to meiosis in primordial germ cells and the atretic loss of the large majority of oocytes during pachytene I, when the process of recombination is monitored. More is known about the female than the male gametes due to ease of study and the ability to examine the process in real time. However, recent developments of spermatogonial cell lines that undergo meiosis in culture may help to redress this balance (Feng et al., 2002; Sakai, 2002).
This review concentrates on the activity of maturation promoting factor (MPF) during the two meiotic divisions following arrest from the dictyate stage of prophase I in fully grown oocytes. MPF activity oscillates up and down in time with entry to, and exit from, M‐phase respectively during the mitotic cell cycle; and its activity is critically regulated during the two cell divisions of meiosis.
Turning it on when it matters: MPF, the universal M‐phase kinase
MPF activity drives somatic cells into mitosis and oocytes into meiosis (recent reviews: Pines, 1999; Masui, 2001; Dorée and Hunt, 2002). Initially purified by its ability to induce nuclear envelope breakdown (NEB) in frog oocytes, it has been proposed to phosphorylate: (i) condensin, a large protein complex, needed to supercoil DNA during mitosis (Swedlow and Hirano, 2003); and (ii) several protein components of the nuclear envelope, including nucleoporin, a component of the nuclear pore complex, whose phosphorylation may be one of the first steps in NEB (Burke and Ellenberg, 2002; Lénárt and Ellenberg, 2003; Lénárt et al., 2003).
MPF is a heterodimer, composed of a catalytic CDK1 (p34cdc2) subunit and a regulatory cyclin B subunit. There are at least three types of cyclin B (B1, B2 and B3), and in mammals it appears that cyclin B1 is primarily responsible for MPF activity. Although binding of CDK1 to cyclin B1 is necessary, it is not sufficient for kinase activity. Switching on of MPF in all cells is further governed by the balance in the regulatory activity of Wee1/Myt1 kinases, which cause an inhibitory phosphorylation of CDK1 at Thr14 and Tyr15 (and hold the heterodimer in an inactive state—so‐called pre‐MPF); and the CDC25 phosphatases, which cause an activating dephosphorylation of CDK1 at the same sites. High CDC25 and low Wee1/Myt1 activity are needed for switching on the CDK1 component of MPF. In addition, before entry into mitosis, cyclin B1 (and so MPF) is spatially restricted to the cytoplasm, through a cytoplasmic retention sequence, containing a nuclear export signal. When cells commit to mitosis, cyclin B1 has to become phosphorylated within its cytoplasmic retention sequence, leading to rapid accumulation of cyclin B1 and MPF within the nucleus, and ensuing NEB (Hagting et al., 1999).
Turning it on in oocytes: MPF activation during prophase I release
Mammalian oocytes arrest at the dictyate stage of prophase I for most of their lives. It is following a mid‐cycle gonadotrophin surge that oocytes resume meiosis. Morphologically, NEB [so‐called germinal vesicle breakdown (GVBD) in oocytes] occurs within a few hours, and this event is preceded by a rise in MPF activity. The follicular environment is thought to inhibit oocyte maturation because spontaneous cell cycle resumption is observed when fully grown oocytes are removed from their follicles and cultured in vitro. Small oocytes (<60 µm) are not competent to resume meiosis when released from their follicular environment. As CDK1 concentration is lower than that of cyclin B1 in mouse oocytes (Kanatsu‐Shinohara et al., 2000), and as CDK1 levels are higher in competent compared with incompetent oocytes, it is attractive to speculate that meiotic competence is correlated with the attainment of threshold levels of CDK1. However, expression of exogenous CDK1 in incompetent oocytes fails to make them competent (de Vantery et al., 1997). Much more likely is that alterations in how MPF is regulated spatially (nucleus versus cytoplasm) and temporally (CDC25 and Wee kinases) will underlie the phenomenon of oocyte competency.
The spontaneous resumption of meiosis occurring when competent oocytes are released from their inhibitory follicular environment is believed to be triggered by a fall in oocyte levels of cyclic AMP (cAMP: Eppig, 1989; Conti et al., 1998). In vivo meiotic arrest is achieved by a stimulatory G protein (Gs) acting on adenylyl cyclase (Mehlmann et al., 2002), and this activity cannot be maintained when oocytes are released from their intrafollicular environment. The decrease in intra‐oocyte cAMP during release from prophase I arrest must directly or indirectly activate MPF. The most likely mechanism by which cAMP maintains arrest has been studied in frog oocytes, and is through activation of protein kinase A, which in turn phosphorylates CDC25 (Duckworth et al., 2002). Phosphorylated CDC25 is sequestered in the cytoplasm by 14‐3‐3, a family of small acidic proteins, preventing its accumulation in the nucleus, a process believed to be required for switching on of MPF activity (Yang et al., 1999). It will be of interest to determine if the same mechanism is responsible for cAMP‐mediated arrest in mammalian oocytes.
In frog oocytes there is much debate as to the role of MAP kinases (MAPK) in the initiation of oocyte maturation (e.g. Nebreda and Ferby, 2000). There is still contention as to the need for MAPK activation in order for MPF to be activated following progesterone application, the physiological stimulator of oocyte maturation in the frog. In the mouse this issue is neatly side‐stepped, since in the c‐mos knockout mouse, in which there is no MAPK activity, oocytes undergo GVBD with normal kinetics (Colledge et al., 1994; Hashimoto et al., 1994).
Although the changes in the balance of the phosphatases and kinases involved in release from prophase I arrest could also occur in mitosis, there may be differences. For example, CDC25b (there are three members of the CDC25 family in mammals: a, b and c) is both solely and absolutely required for MPF activation at GVBD in oocytes, but is dispensable in male meiosis and somatic mitosis (Lincoln et al., 2002). The reason for this is currently unclear. Furthermore, it may be that novel activators of MPF are observed in oocytes. Indeed such activators (Speedy/RINGO) have been observed in frog oocytes where they induce oocyte maturation (Ferby et al., 1999; Lenormand et al., 1999; reviewed by Nebreda and Ferby, 2000) by acting on CDK1 directly, so bypassing the need for cyclin B1. Interestingly RINGO can also induce oocyte maturation in mouse oocytes (Terret et al., 2001), but the physiological significance of this remains to be examined.
In mice, at least, GVBD can proceed in the absence of any de‐novo protein synthesis. Indeed at the GV stage, cyclin B1 mRNA is translationally repressed. In its 3′ untranslated region, cyclin B1 mRNA contains several cytoplasmic polyadenylation elements (CPE) that bind CPEB, an RNA‐binding protein (Tay et al., 2000) conferring translational dormancy. The CPEB becomes phosphorylated during oocyte maturation by IAK1/Eg2 kinase (Hodgman et al., 2001) which effectively switches on cyclin B1 synthesis and such kinase activity peaks at metaphase I.
Turning it off, but only for a while: the first meiotic division
Following GVBD and activation of MPF, recombined homologous chromosomes converge at the spindle equator forming a metaphase configuration. The goal of first meiosis is a reductional division in which recombined homologous chromosomes are segregated, whilst sister chromatids remain attached. Contrast this with mitosis and the second meiotic division in which sisters are separated. Segregation at meiosis I is therefore treated as a unique division, errors in which are believed to result in many of the aneuploidies observed in early human conceptuses (Nicolaidis and Petersen, 1998; Pellestor et al., 2003).
During mitosis there exists a well‐characterized surveillance mechanism called the spindle checkpoint which monitors the alignment of sister chromatids on the mitotic spindle during prometaphase. It is not currently defined if a similar mechanism operates to monitor homologous chromosomes during meiosis I. During mitosis, checkpoint components, comprised of the Mad and Bub family of proteins, monitor attachment and tension at kinetochores, which attach sister chromatids to the spindle microtubules. The Mad and Bub proteins orchestrate the spindle checkpoint by preventing premature anaphase prior to full chromosome alignment (Yu, 2002; Musacchio and Hardwick, 2002). When metaphase is reached, with both sisters under tension and fully aligned, the checkpoint proteins become inactive. The target for the spindle checkpoint action is the anaphase promoting complex or cyclosome (APC/C). The APC/C is an E3 ubiquitin ligase complex that polyubiquitinates cyclin B1, thereby targeting it for proteolysis by the 26S proteasome (Irniger, 2002). Therefore during exit from mitosis, MPF activity declines due to loss of cyclin B1. The APC/C requires accessory proteins for its ligase activity and during the metaphase–anaphase transition it is CDC20/fizzy that is essential for activity. Before metaphase, CDC20 is held in a ternary complex with the spindle checkpoint protein Mad2 and the APC/C, preventing APC/C activation (Fang et al., 1998).
Currently there is much interest in determining what role, if any, the spindle checkpoint and the APC/C play in vertebrate meiosis I; in yeasts meiosis I is monitored by a spindle checkpoint (Shonn et al., 2000; Bernard et al., 2001). In the Xenopus oocyte the APC/C appears dispensable for the metaphase I–anaphase I transition, since activators of the spindle checkpoint and inhibitors of the APC/C all fail to arrest oocytes at metaphase I (Peter et al., 2001; Taieb et al., 2001). This has led to the hypothesis that in vertebrates, at least, meiosis I is characterized by a poor spindle integrity surveillance mechanism that may underlie the high disjunction rates observed in humans. It is, however, true that mammalian oocytes have been delayed in meiosis I by brief incubations, and arrested by longer incubations, with microtubule‐disrupting agents that induce a spindle checkpoint (Jones et al., 1995; Brunet et al., 1999). This observation in itself lends support to the idea that the APC/C needs to function in mammalian oocytes for them to progress beyond meiosis I. Furthermore increasing cyclin B1 levels in oocytes during their maturation delays polar body formation, again suggesting that cyclin B1 needs to be efficiently degraded in oocytes for homologous chromosome segregation (Polanski et al., 1998; Ledan et al., 2001). Also a truncated cyclin B1 that cannot be degraded by the APC/C prevents meiosis I in both human and mouse oocytes (M.Herbert, personal communication), suggesting that APC/C‐mediated cyclin B1 degradation is necessary for homologous chromosome segregation. Contrary to these observations are studies on the female XO mouse, which despite possessing a misaligned, univalent X chromosome undergoes meiosis I with normal kinetics, whereas sperm of the XO, sex‐reversed male mouse rarely successfully complete the first meiotic division (Hunt and LeMaire‐Adkins, 1998).
No S‐phase after meiosis I
One unique feature of meiosis is that sister chromatids remain condensed and re‐align on the second meiotic spindle following meiosis I; they do not decondense as would occur at the end of mitosis. There is much evidence to suggest that c‐mos, a MAP kinase kinase kinase, plays a critical role in this during the meiosis I–meiosis II transition. c‐Mos activity, through the MAPK pathway, regulates in a positive way both the migration of the spindle to the cortex (Verlhac et al., 2000a), in order to set up an asymmetric cell division, and the condensation of chromatin (Araki et al., 1996; Verlhac et al., 1996).
The c‐mos–MAPK pathway cannot be wholly responsible for continued chromatin condensation during meiosis I–meiosis II. This is because the majority of oocytes progress through the first meiotic division and reach metaphase II in c‐mos knockout mice. In contrast, complete inhibition of protein synthesis in the few hours before the first meiotic division leads to nuclear reformation for the vast majority of oocytes following first polar body extrusion (Clarke and Masui, 1983). These data suggest that continued expression of an unstable protein during meiosis I is needed for the maintenance of chromatin condensation.
The second meiotic division
The long‐term stability (several hours) of MPF levels in mammalian oocytes is an interesting feature of metaphase II arrest. Oocytes arrest with a fully formed spindle and aligned chromosomes. If this were mitosis the APC/C would be active and cyclin B1 degraded. To some extent this does indeed occur during metaphase II arrest since the APC/C is active and cyclin B1 is turned over in a manner that is totally dependent on the APC/C (Kubiak et al., 1993; Nixon et al., 2002). Cyclin B’s half‐life during metaphase II arrest has been estimated at 1.2 h (Nixon et al., 2002) to 1.9 h (Winston, 1997). However, the oocyte has the capacity to match cyclin B1 synthesis with its degradation, in order that MPF levels do not drop. Thus in mammalian oocytes cyclin B1 needs to be continually synthesized to maintain arrest.
What is limiting cyclin B1 destruction, and so maintaining metaphase II arrest? The cytoplasmic activity present in oocytes to maintain metaphase arrest has been coined cytostatic factor (CSF; Masui and Markert, 1971) and the most characterized pathway believed to be involved in CSF‐induced arrest is c‐mos–MAPK. Parthenogenesis results in oocytes from c‐mos knockout mice (Colledge et al., 1994; Hashimoto et al., 1994) suggesting that this pathway is responsible for maintaining metaphase II arrest. Interestingly in the frog it is a component of the spindle checkpoint pathway, Bub1, that is ultimately responsible for the c‐mos‐mediated cell arrest (Gross et al., 2000; Schwab et al., 2001). It is an attractive hypothesis that oocytes are held at metaphase II by a spindle checkpoint. In the mouse this would be further supported by the finding that Mad2, another spindle checkpoint component, is localized to the sister chromatid kinetochores at metaphase II arrest, but is lost following oocyte activation (Kallio et al., 2000). Such localization of spindle checkpoint components is normally observed when the spindle checkpoint pathway is active. However, addition of microtubule‐disrupting agents that activate the spindle checkpoint pathway are able to greatly stabilize cyclin B1 levels at metaphase II (with a half‐life of >9 h; Winston, 1997) and furthermore prevent the oocyte from being activated by sperm or by parthenogenetic stimulation (Winston et al., 1995). Therefore if a spindle checkpoint pathway does operate at metaphase II to maintain cell cycle arrest, it cannot be the same as that which is induced by pharmacological microtubule‐disrupting agents. This is because the former is patently overcome by the sperm at fertilization, but the latter prevents the oocyte from activating.
There are two further confusions to address in understanding how oocytes maintain a metaphase arrest. First, if it is the c‐mos–MAPK (–Bub1) pathway responsible for maintaining a metaphase arrest, then it is not entirely clear why oocytes from c‐mos knockout mice should re‐enter a metaphase state after completing meiosis II (Verlhac et al., 1996). Oocytes from c‐mos knockout mice parthenogenetically activate but instead of entering interphase of the first embryonic cell cycle, they re‐arrest in a metaphase state with a monopolar spindle and remain in this state for several hours, before going on to activate. This begs the question, what is inducing the chromatin to remain condensed in the absence of c‐mos? Microinjection of c‐mos or constitutively active MAPKK into the blastomeres of 2‐cell mouse embryos arrests the blastomeres, as would be predicted for an arresting activity. However, this arrest is in interphase, not mitosis (Verlhac et al., 2000b).
In conclusion, it is clear that the c‐mos–MAPK pathway has an important role to play in meiosis, especially in the meiosis I–meiosis II transition. What is still unclear is its true contribution to maintaining a metaphase arrest during meiosis II. Two other candidates that may contribute to metaphase arrest are noteworthy. Firstly is cdk2, which, with its partner cyclin E, is important in the G1/S cell cycle transition. In frog oocytes, cdk2 antisense constructs abolish metaphase II arrest, which can then be re‐established by microinjection of cdk2 (Gabrielli et al., 1993). However, other studies that utilized p21cip, a potent cdk2 inhibitor, failed to find any effect of cdk2 inhibition on metaphase II arrest (Furuno et al., 1997). Therefore further studies are needed to resolve the role of cdk2 in metaphase II arrest. The second candidate worthy of discussion is Emi1, the most recent addition to the group of proteins believed to play a role in metaphase II arrest in frog oocytes. Interestingly it acts independently of the MAPK pathway (Reimann and Jackson, 2002). Emi1 binds to CDC20, and in so doing prevents the APC/C from becoming active. In somatic cell division it has a role in preventing APC/C activity before entry into mitosis (Reimann et al., 2001). Depletion of Emi1 from frog oocyte extracts prevents them from arresting at metaphase and excess Emi1 arrests oocytes at metaphase. It will be interesting to determine if Emi1 plays an analogous role in mammalian oocytes.
Calcium: nature’s way of turning it off
An intracellular Ca2+ signal triggered by the fertilizing sperm is the endogenous signal for cyclin B1 destruction in metaphase II arrested oocytes. Ca2+ is the necessary and sufficient trigger for oocytes to complete meiosis and so enter the first embryonic cell cycle. The Ca2+ signal may be generated by an oocyte receptor on the oolemma, activated by a sperm ligand on gamete binding. However, increasing evidence suggests that a soluble sperm factor is introduced into the oocyte following sperm fusion (e.g. Jones et al., 1998). The most likely candidate is phospholipase C (PLC) zeta, a new member of the PLC family (Saunders et al., 2002) or a truncated form of the c‐kit receptor that activates PLC through the Src‐family kinase Fyn (Sette et al., 2002). Whatever the exact mechanism, Ca2+ appears to activate calmodulin‐dependent protein kinase II (CaMKII; Markoulaki et al., 2003). The ultimate targets of this kinase are still unknown, despite its central importance in the physiological process. One can only speculate, for example, that Emi1 has a consensus sequence for CaMKII phosphorylation.
It is interesting that in mammals, in contrast to frogs (which also arrest at metaphase II) the Ca2+ signal is oscillatory and lasts several hours (rather than monotonic and lasting for minutes). A single Ca2+ rise can be an effective parthenogenetic stimulus, but only in aged oocytes, which have a reduced capacity to synthesize cyclin B1. In fresh oocytes, the second polar body is extruded but the chromatin remains condensed and reorganizes on a new monopolar spindle (Kubiak, 1989). The ability of the oocytes to continually synthesize new cyclin B1 probably accounts for this phenomenon. Indeed direct measurement of cyclin B1 destruction relative to the oscillatory Ca2+ signal has demonstrated that an oscillatory signal is needed to obtain prolonged cyclin B1 destruction (Nixon et al., 2002). Interestingly peaks in CaMKII activity are associated with the peaks in Ca2+ (Markoulaki et al., 2003).
What is the target of CaMKII action? Since the Ca2+ spikes speed up (6‐fold) cyclin B1 destruction (Nixon et al., 2002) it would seem safe to assume that the target is either the APC/C or the 26S proteasome, which degrades polyubiquitinated proteins such as cyclin. If one extrapolates from other species, the likelier target is the proteasome rather than the APC/C. In both ascidian (Kawahara and Yokosawa, 1994) and frog oocytes (Aizawa et al., 1996) Ca2+ has been found to increase 26S proteasome activity. Similarly in sand‐dollar oocytes, fertilization results in increased activity (Chiba et al., 1999), although in this case it is a fertilization‐induced rise in intracellular pH, rather than Ca2+, that triggers increased proteolysis. However, since most metaphase‐arresting proteins are believed to play a role in decreasing APC/C activity (see above) during metaphase II arrest, then the APC/C would also seem a very logical CaMKII target. No study has yet directly addressed this question of where Ca2+ acts in mammalian oocytes.
Understanding of the cell cycle has proceeded at a fast pace over recent years, thanks to the genetic manipulations possible in yeasts and the biochemistry of cycling frog oocyte extracts. Some of the interesting and unique features of meiosis, however, remain unresolved, for example, what stabilizes cyclin B1 during metaphase arrest and precisely how CaMKII mediates cyclin B1 destruction. It is hoped that within the next few years this pathway, central in our understanding of the basic cell biology of fertilization, will be resolved.
I would like to thank Tom Ducibella, Hayden Homer, Mark Levasseur and Alex McDougall for critical reading of the manuscript, and Mary Herbert for communicating unpublished work. I would like to acknowledge the continued support of the Wellcome Trust to my laboratory.