A new role for Mos in Xenopus oocyte maturation: targeting Myt1 independently of MAPK

In animal kingdom, oocytes arrest the cell cycle at the G2 prophase boundary of the first meiotic cycle. Then, they grow to their maximal size. G₂ arrest is terminated by specific signals, often hormones, that cause release from cell cycle arrest and progression into the meiotic cell cycles until fertilization. These events define the oocyte maturation. In Xenopus oocytes, progesterone stimulation induces Mos mRNAs polyadenylation and translation (Gebauer et al., 1994; Sheets et al., 1995). Then, the proto-oncogene Mos kinase activates the MAPK cascade by phosphorylation of the MAPK-activating kinase MEK (Posada et al., 1993). The Mos-MEK-MAPK cascade is thought to constitute a positive-feedback loop (Howard et al., 1999; Matten et al., 1996). Even if Hsp90 function is required for activation and phosphorylation of Mos (Fisher et al., 2000), the early process of Mos activation remains unclear. MAPK activation is required for efficient activation of the maturation-promoting factor (MPF), a complex formed by cyclin B and Cdc2 kinase (for reviews, see Doree, 1990; Nurse, 1990), probably through activation of a target of MAPK, p90rsk (for reviews, see Abrieu et al., 2001; Ferrell, 1999; Nebreda and Ferby, 2000; Sagata, 1997).

While it is accepted that Cdc2 kinase activation is essential for germinal vesicle breakdown (GVBD) in all species, differences in the timing of, and requirement for, MAP kinase activation occur between different species (for a review, see Yamashita et al., 2000). MAPK activation is not required for the initial activation of MPF in oocytes of species such as starfish or mouse, which activate MAPK only after GVBD. By contrast, in Xenopus, MAPK is activated before GVBD and facilitates Cdc2 kinase activation. Until recently, MAPK activation was thought to be required for GVBD, as injection of constitutively activated MAPK (Haccard et al., 1995) or MEK (Huang et al., 1995) into immature oocytes induced GVBD. Moreover, experimental treatments that block activation of MAPK, significantly inhibit resumption of meiosis (Gotoh et al., 1995; Kosako et al., 1994; Sagata et al., 1988). However, recent reports from different laboratories have shown that the MAP kinase pathway was not essential for entry into meiosis I (Fisher et al., 1999; Gross et al., 2000). Together, these data suggest that an alternative, MAP-kinase-independent, pathway may exist to initiate MPF activation in response to progesterone. Mos has been reported to be the candidate initiator of Xenopus oocyte maturation (Sagata et al., 1989), and the use of Mos antisense oligonucleotides inhibits GVBD (Sagata et al., 1988; Sheets et al., 1995), suggesting that Mos is necessary for MPF activation. Because the highly specific target of Mos is MEK, which activates MAPK, if Mos is still required to trigger GVBD independently of MAPK cascade, it must have a target other than MEK.

In addition to activating MEK, it has been suggested, in mouse oocytes, that Mos can inhibit a phosphatase whose activity inactivates MAPK (Verlhac et al., 2000). However, this second function of Mos still involves MAPK activation and cannot answer the question of whether Mos is implicated or not in a MAPK-independent pathway.

In this study, we show that Mos is required for entry into meiosis and controls the G2 arrest exit independently of MAPK cascade pathway by facilitating the conversion of pre-MPF into MPF. In this function, the target of Mos is probably the Myt1 kinase, a direct inhibitor of Cdc2 kinase.

Oocytes manipulations in MMR buffer (100 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 0.1 mM EGTA, 5 mM Hepes pH 7.7) and homogenates in oocyte buffer (50 mM β-glycerophosphate, 10 mM MgCl₂, 7.5 mM EGTA, 1 mM DTT, pH 7.3) were performed as described (Peter et al., 2001). Progesterone was used at a final concentration of 1 mM. MEK inhibitor U0126 (Promega) was dissolved in dimethylsulfoxide (DMSO) at a concentration of 10 mM and diluted in MMR to 50 μM. Cycloheximide was used at 100 μg/ml. For microinjections, the usual volume injected was 50 nl.

Antisense Mos oligonucleotides were based on those already tested in Xenopus oocytes (Sheets et al., 1995); in this case, CATATCCTTGCTTGTATTTTCAGTGC. The antisense Eg2 oligonucleotide was CAAGTAGCGTTGTACGGTGACAGCC. Oligonucleotides were dissolved in water at 2 μg/μl and were microinjected into stage VI oocytes 1 hour before progesterone treatment.

Eg2 cDNA isolated from the pKS-Eg2 plasmid described by Roghi et al. (Roghi et al., 1998) was subcloned in pCS2 vector and Myt1 cDNA was cloned in psp64T vector. To allow Eg2 protein synthesis before stimulation, mRNAs were injected 16 hours before progesterone addition.

Mos wild-type and kinase-dead genes were cloned into the Xenopus expression vector pXen1 (MacNicol et al., 1997) to allow production of GST-tagged Mos mRNA by SP6 in vitro transcription. Capped mRNA were prepared with mMESSAGE mMACHINE^TM kit (Ambion) and used at 1 mg/ml.

Constructions of the sea urchin non-degradable GST-cyclin B and the Xenopus MBP-Mos have been described previously (Abrieu et al., 1996).

Recombinant proteins were prepared according to standard procedures. They were stored at 1 mg/ml and diluted just before use, MBP-Mos generally to 20 μg/ml, GST-cyclin B to different concentrations, depending on the experiment, as indicated.

The anti-Mos, Anti-ERK and anti-pY ERK were obtained from Santa Cruz (SC-086, SC-94 and SC-7383, respectively). The anti-pTpY ERK and pY Cdc2 were obtained from New England Biolabs (9106S and 9111S, respectively). Other antibodies used were rabbit polyclonal antisera against Xenopus Cdc2 C terminus, Myt1 C terminus, MEK C terminus, Plx1 C terminus and full-length recombinant Cdc25. The anti-β tubulin antibodies were obtained from E7 (Iowa hybridoma bank) and the monoclonal anti-Eg2 antibody was kindly provided by C. Prigent (Roghi et al., 1998). All the polyclonal antibodies were used affinity purified. Western blots were probed with primary antibody at 50 ng/ml, and the appropriate secondary antibody horseradish peroxidase (HRP) conjugate diluted according to recommendations (Amersham) and revealed by ECL (New England Nuclear).

For GST pull-down, oocytes were homogenized in oocyte buffer (20 μl per oocyte), and after centrifugation (13,000 rpm for 3 minutes at 4°C), the clear supernatant was incubated with glutathione-sepharose beads (Pharmacia) for 1 hour at 4°C. GST pull-downs were washed twice in RIPA buffer (10 mM NaH₂PO₄ pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.5% deoxycholate, 80 mM β-glycerophosphate, 50 mM NaF, 1 mM DTT) then in 50 mM Tris pH 7.5.

Myt1 phosphorylation was tested on GST pull-down in phosphorylation buffer (100 μM ATP, 5 μCi ³²PγATP, 10 mM MgCl₂ and 1 μM microcystin)

In-gel MBP kinase assays were performed exactly as described by Shibuya et al. (Shibuya et al., 1992). Total histone H1 kinase activities were assayed as described (Labbe et al., 1988).

In order to prevent Mos synthesis, antisense oligonucleotides directed against the 3′-UTR of Mos mRNA (Sheets et al., 1995) were microinjected 1 hour before progesterone stimulation. The antisense oligonucleotides were very efficient in suppressing polyadenylation of Mos mRNA, as no accumulation of Mos protein could be detected by Western blotting as late as 7 hours after hormonal stimulation, yet oocytes underwent GVBD at that time. By contrast, accumulation of Mos was already detected 45 minutes after progesterone addition and its level increased with time at least until GVBD in control oocytes (Fig. 1A, part 1).

As a faint signal was detected with antibodies Mos at the time of GVBD, we were concerned that a faint amount of Mos protein, not detectable by western blotting, could be synthesized before GVBD in oocytes injected with Mos antisense oligonucleotides. Accordingly, we supposed that MAPK cascade might be activated, as Mos kinase has been reported to induce MAPK activation (Posada et al., 1993). The activation of MAPK, which corresponds to its dual phosphorylation of the Tyr190 and the Thr188 residues on Xenopus ERK (Fig. 1B) (Anderson et al., 1990; Haystead et al., 1992; Posada, 1993), was monitored using the antibody pYpT ERK, which recognizes the diphosphorylated form of MAPK. In oocytes injected with Mos antisense oligonucleotides (+AS Mos), the signal was faint and only detectable in oocytes that underwent GVBD (Fig. 1A, part 2). By contrast, in control oocytes (–AS Mos), it was already detectable before GVBD and strongly increased at GVBD. Moreover, the characteristic electrophoretic shift of MAPK/ERK that is associated with its activation was not detected by the polyclonal anti-ERK antibody in Mos antisense-injected oocytes, showing that a very low proportion of MAPK molecules were activated after GVBD (Fig. 1A, part 3). These results show that Mos antisense oligonucleotides are not sufficient to suppress Mos expression completely. However, this low level of expression is not enough to activate MAPK significantly.

There is experimental evidence that the activating dual phosphorylation of MAPK by MEK proceeds by an ordered mechanism, Y190 being phosphorylated before T188 (Haystead et al., 1992). In an effort to detect whether such an early event occurred before GVBD in the oocytes injected with Mos antisense oligonucleotides, we probed the same membrane with an antibody specific for MAPK phosphorylated only on the tyrosine 190 (Ab pY ERK). As shown in Fig. 1A (part 4), no significant delay was detected in early tyrosine phosphorylation of MAPK in Mos antisense-injected oocytes, when compared with control oocytes, suggesting that at least a low level of MEK activation occurs as soon as 15 minutes after progesterone stimulation, as a consequence of the low level of Mos protein synthesis that still proceeds in such oocytes. As Mos protein accumulation is greater than background levels, and dual MAPK phosphorylation is only detected after GVBD in Mos antisense-injected oocytes, these events could be a consequence of Mos stabilization after GVBD (Castro et al., 2001; Nishizawa et al., 1992).

In order to estimate the low amount of Mos protein still present in mature Mos antisense-injected oocytes, we performed successive dilutions of homogenates from non injected matured oocytes and compared the signal detected by anti-Mos antibodies in Mos antisense-injected oocytes at the different dilutions. To ensure identical loading of sample of each dilution, we used an homogenate from non injected immature oocytes as a vehicle for serial dilutions. We estimated that the amount of Mos protein present in matured oocytes injected with Mos antisense oligonucleotides does not exceed 5-10% of total Mos protein in matured control oocytes (Fig. 1C). Thus, an even lower level of Mos protein is enough to trigger the early process of monophosphorylation of MAPK, but not sufficient for its complete activation (Fig. 1A, part 3) (see Discussion). Even though only a very small amount of Mos protein was synthesized in oocytes injected with Mos antisense oligonucleotides, a high proportion (50 to 90%, depending of the experiment) of these oocytes underwent GVBD, although with a delay when compared with control oocytes (Fig. 2B, Fig. 3B). Taken together, these experiments show that a faint amount of Mos protein is sufficient to trigger GVBD, but is not enough to activate MAPK.

If Mos is required to control the exit of oocytes from G₂ arrest, it does so independently of MAPK activation, as oocytes can undergo GVBD without MAPK activation (Fisher et al., 1999; Gross et al., 2000), which requires double phosphorylation of its T loop. However, the small amount of Mos protein still observed in Mos antisense-injected oocytes, was still sufficient for MAPK monophosphorylation, which could thus be essential for GVBD. To test this possibility, we looked for resumption of meiosis in oocytes injected with Mos antisense oligonucleotides in the presence or absence of the MEK inhibitor U0126. As expected, monophosphorylation of MAPK did not occur in the presence of U0126 (Fig. 2A, lanes 2,4). Yet, in this experiment, the Mos antisense-injected oocytes, treated with U0126, underwent GVBD with the same timecourse as oocytes treated with only U0126 (Fig. 2B). However, depending on the experiment, the Mos antisense-injected oocytes treated with U0126 can undergo GVBD with a timecourse that is slower than oocytes treated only with U0126. These results demonstrate that the faint amount of Mos protein produced in oocytes injected with Mos antisense oligonucleotides is sufficient to trigger GVBD independently of both MEK and MAPK activation.

Nevertheless, the above experiments failed to provide a definitive answer to the question as to whether the Mos protein is required, or not, for progesterone to trigger GVBD. We therefore tried to find a way to suppress Mos protein synthesis completely. The Eg2 kinase has been reported to facilitate maturation when oocytes are stimulated by low concentration of progesterone (Andresson and Ruderman, 1998), and to control Mos protein synthesis by phosphorylating CPEB (Mendez et al., 2000), a protein involved in cytoplasmic polyadenylation (Hake and Richter, 1994). Consistent with these previous results, we found that synthesis of Mos protein could be increased by overexpression of the Eg2 kinase in Mos antisense-injected oocytes, although it did not significantly increase Mos level in control oocytes (Fig. 3A), suggesting that overexpression of Eg2 increases Mos synthesis when Mos translation is limited. Thus, we thought that inhibition of Eg2 synthesis could possibly suppress the low level of Mos synthesis in Mos antisense-injected oocytes. Eg2 mRNA is polyadenylated and its translation increases in response to progesterone (Frank-Vaillant et al., 2000; Paris and Philippe, 1990). Mos and Eg2 antisense oligonucleotides (directed against the 3′ UTR of each mRNA) were co-injected in stage VI oocytes 1 hour before progesterone addition and oocytes scored for GVBD 20 hours after hormone stimulation. Although nearly all the oocytes injected with Mos antisense or Eg2 antisense alone underwent GVBD (75% for Mos and 95% for Eg2), only 5-15% of co-injected oocytes did so (Fig. 3B). To verify that Mos protein was not synthesized in co-injected oocytes that failed to undergo GVBD 20 hours after progesterone stimulation, we tested for the presence of Mos protein, either directly, by western blotting, or indirectly, using detection of tyrosine-only phosphorylated MAPK molecules as a highly sensitive marker for Mos kinase activity in homogenates prepared from stimulated oocytes under the different conditions (Fig. 3C). In contrast to oocytes injected with Mos antisense oligonucleotides alone (Fig. 3C, lanes 2-3), the monophosphorylated form of MAPK was not detected in oocytes injected with both antisense oligonucleotides (Fig. 3C, lane 4; the faint signal detected is also present in stage VI oocytes), showing that MEK was not activated at all in these oocytes. We cannot exclude the possibility that antisense Eg2 oligonucleotides, which phosphorylate CPEB, could act on polyadenylation of another mRNA in addition of Mos. To verify that antisense Eg2 oligonucleotides do not act on an essential protein other than Mos, we looked for resumption of meiosis, in presence of U0126, in antisense co-injected oocytes in which a subthreshold amount of Mos protein was injected 2 hours after progesterone addition. As shown in Fig. 3D, 70% of oocytes co-injected with antisense Eg2 and Mos antisense underwent GVBD when they were also injected with Mos protein, whereas none of them re-entered meiosis when they were not supplied with Mos protein. Moreover, the oocytes supplied with Mos protein that underwent GVBD in presence of U0126 did not activate MAPK (data not shown), proving that the amount of Mos protein injected was too low to overcome the effect of the MEK inhibitor. Together, these experiments show that synthesis of a small amount of Mos protein (less than 10% of the control level) is both necessary and sufficient for the induction of GVBD by progesterone independently of MAPK activation in Xenopus oocytes.

Even though MAPK activity is dispensable, we have now shown that Mos expression is still required for Xenopus oocytes to be released from G2 arrest and progress to GVBD. As expected, Mos is also required for activation of cyclin B-Cdc2 detected by H1 kinase assay (Fig. 3C).

The fact that no cyclin synthesis is apparently required for GVBD in Xenopus oocytes (Hochegger et al., 2001; Minshull et al., 1991) implies that activation of cyclin B-Cdc2 kinase before GVBD can entirely rely on the conversion of stockpiled cyclin B-Cdc2 complexes (pre-MPF) from an inactive into an active form (MPF).

To investigate whether Mos targets the conversion of pre-MPF into MPF, we chose to bypass the signal transduction pathway triggered by progesterone stimulation. Indeed, re-initiation of meiosis by Mos protein injection required either early process(es) induced by progesterone, or MAPK activation, because the MEK inhibitor U0126 prevents GVBD when oocytes are not stimulated by progesterone (data not shown) (Gross et al., 2000). By contrast, injection of cyclin B1 protein at sufficient concentration did not require the MAPK cascade pathway to induce GVBD, as oocytes injected with GST-cyclin B1 in presence of cycloheximide underwent GVBD without Mos synthesis and MAPK activation (Fig. 4).

Since the investigations of Solomon et al. (Solomon et al., 1990), it has been known that low amounts of cyclin B form inactive pre-MPF complexes with endogenous Cdc2 in Xenopus oocytes, which can be converted into active MPF complexes by increasing cyclin B levels. In the following experiment, we looked for an amount of either GST-cyclin B1 or MBP-Mos proteins that was slightly too low to allow stage VI oocytes to undergo GVBD. Then we tested whether the co-injection of such an amount of the two proteins allowed the resumption of meiosis in the presence of U0126, i.e. in the absence of MAPK cascade activation. This was indeed the case: injection of subthreshold amount of cyclin B1 or MBP-Mos protein alone did not trigger GVBD, but GVBD occurred in more than 40% of the oocytes when these proteins were injected together, even in the presence of U0126 (added in the medium 1 hour before injection). We repeated the experiment several times (Fig. 5A) and consistently verified that oocytes undergoing GVBD did not activate MAPK (Fig. 5B). Then, we looked for the state of tyrosine phosphorylation of Cdc2 in GST-cyclin B1-Cdc2 complexes by monitoring electrophoretic mobility of Cdc2 and observed that, under conditions where oocytes underwent GVBD, Cdc2 kinase was dephosphorylated as soon as 1 hour after injection of both Mos and cyclin B1 proteins (Fig. 5C).

We also injected a subthreshold amount of GST-cyclin B1 with or without a subthreshold amount of MBP-Mos protein into immature oocytes incubated in medium containing both U0126 and cycloheximide. As shown in Fig. 5A, the cooperative effect of Mos was observed even with cycloheximide, demonstrating that Mos facilitates, in the absence of MAPK cascade activation, the conversion of pre-MPF into MPF at a post-translational level.

The equilibrium between pre-MPF and MPF is universally controlled at the post-transcriptional level by the balance between double specificity kinases that phosphorylate residues Thr14 and Tyr15 of Cdc2 and maintain cyclin B-Cdc2 complexes in the inactive pre-MPF form, and the dual specificity phosphatase Cdc25c, which dephosphorylates these inhibitory residues and converts pre-MPF into active MPF (for a review, see Palmer and Nebreda, 2000). During meiotic maturation and early embryonic development, the inhibitory kinases are Myt1 and Wee1 (Mueller et al., 1995; Murakami and Vande Woude, 1998). In Xenopus G₂ arrested stage VI oocytes, Myt1 is exclusively expressed. Wee1 expression starts after GVBD and becomes the major inhibitory kinase in the first embryonic cell cycle (Nakajo et al., 2000).

Formally, Mos could facilitate conversion of pre-MPF into MPF in the first meiotic cell cycle either by activating Cdc25c or by inhibiting Myt1. In the first case, a facilitating effect of Mos should also be observed in the second meiotic cell cycle and in early embryogenesis. However, this does not appear to be the case for early embryogenesis. Indeed, ectopic expression of Mos (or prevention of Mos inactivation) has rather been reported to suppress post-translational conversion of pre-MPF into MPF in the early embryonic cell cycle (Abrieu et al., 1997).

In the next experiment, we investigated whether Mos facilitates or inhibits this conversion in the second meiotic cell cycle. Oocytes were induced to undergo GVBD by progesterone, then arrested at interphase by suppressing protein synthesis with cycloheximide. Such arrested oocytes were injected with various amounts of either GST-cyclin B1 alone or both GST-cyclin B1 and MBP-Mos, then formation of active MPF was assessed by monitoring H1 histone kinase of GST-cyclin B1 pull-downs. As shown in Fig. 6, we failed to detect a facilitating effect of Mos, contrasting with results obtained in the first meiotic cell cycle. Rather, an inhibitory effect of Mos was systematically observed. Thus, the facilitating effect of Mos is restricted to the period when Myt1 is the only kinase antagonizing MPF activation. Hence, Myt1 is a good candidate as a target for Mos in this process. This is consistent with the previous report that p90rsk, a target of the MAPK cascade, negatively controls Myt1 inactivation in the first meiotic cell cycle (Palmer et al., 1998). However, we found in the above section that Mos also controls MPF activation independently of MAPK cascade. This could involve a direct interaction between Mos and Myt1.

We therefore examined the possibility that Mos may associate with Myt1 in vivo, using GST pull-down (MacNicol et al., 1997) to detect protein-protein interactions of in vivo synthesized proteins. We injected stage VI oocytes (incubated or not in U0126) with either GST-Mos WT (wild-type) or GST-Mos KD (kinase dead mutant) mRNAs. After 12 hours (to allow Mos expression to occur), progesterone was added and oocytes were collected 4 hours later. After GST pull-down, the Mos complexes were analyzed by western blot (Fig. 7A). Mos expressed in this way indeed interacted with Myt1 (Fig. 7A, lanes 5-8). This interaction was specific, as no significant interaction of GST-Mos was detected with other proteins tested, including polo like kinase (Plx1) or MAP kinase. Moreover, the only identified direct target of Mos kinase besides tubulin (Zhou et al., 1991), the MEK protein, was recovered in absence of U0126 in the pull-down of WT Mos but only faintly on pull-down of KD Mos. This is in agreement with the fact that Mos-MEK interaction requires Mos kinase activity, as previously described in a two-hybrid system (Chen and Cooper, 1995). It is noteworthy that the MEK inhibitor U0126 strongly decreases the Mos-MEK interaction. By contrast, both WT and KD Mos proteins interacted with Myt1, to nearly the same extent. Depending on whether the oocytes underwent GVBD or not, the hyperphosphorylated (inactive) form of Myt1 was associated with the WT Mos protein (Fig. 7A, lanes 5,6), and the hypophosphorylated (active) form of Myt1 interacted with the kinase-dead Mos protein (Fig. 7A, lanes 7,8). WT Mos protein was produced to a high level from microinjected GST-Mos mRNAs, and consequently activated the MAPK cascade to some extent, even in the presence of U0126, and induced GVBD. Thus, it was not possible in this experiment to establish unambiguously that Mos kinase could induce Myt1 phosphorylation independently of the MAPK cascade or of MPF activation.

As Mos associates with Myt1 in oocyte extracts we examined whether Mos could phosphorylate Myt1 in vitro. Mos wild-type kinase requires post-translational events to gain activity when translated in vitro. Thus, we tested if Myt1 translated in vitro could be phosphorylated by active Mos isolated from mature oocyte previously injected with GST-Mos WT mRNA. As shown in Fig. 7B, Myt1 translated in reticulocyte extract is able to bind efficiently Mos WT and Mos KD (Fig. 7A, lanes 5,6). Moreover, a low level of Myt1 phophorylation by Mos WT (Fig. 7A, lane 6) when compared with Mos KD (Fig. 7A, lane 5) is observed, and a faint shift is detectable on the autoradiogram and on the western blot (detected by anti-Myt1 antibodies). These results suggest that Mos could phosphorylate Myt1 on some sites in vivo. Indeed, it is probably that binding is not sufficient, and some chaperon(s) is (are) necessary to allow the kinase to be active.

We then examined the phosphorylation state of Myt1 when oocytes were induced to mature by progesterone in conditions where the MAPK cascade or/and MPF activation could be efficiently blocked. It has been reported that phosphorylation of Myt1 by p90rsk could be responsible for Myt1 inactivation (Palmer et al., 1998). As shown in Fig. 8A, in the presence of U0126 and the absence of MAP kinase activity (Fig. 8A, panels 2,3), a subset of Myt1 underwent a mobility shift typical of hyperphosphorylation in progesterone treated oocytes (Fig. 8A, panel 1). This is consistent with its inactivation, as the same oocytes underwent GVBD and Cdc2-tyrosine 15 residue was dephosphorylated (Fig. 8A, panel 4, lane 11). p90rsk was certainly not responsible for this phosphorylation, as p90rsk activation has been demonstrated to be mediated by MAPK activation in Xenopus oocytes.

In the above experiment, phosphorylation of Myt1 on some sites could still be due to Cdc2 kinase (Booher et al., 1997), although Myt1 has not been shown unambiguously to be a target for Cdc2 kinase in vivo. Thus, we tried to find conditions were MAPK was inactive and Cdc2 kinase not still activated. As shown in Fig. 8B, oocytes treated with U0126 presented Myt1 phosphorylation on some sites, while Cdc2 kinase was not apparently still activated (lane 15). However, the time during which Myt1 is partially phosphorylated and Cdc2 is inactive is probably very short, as cyclin B-Cdc2 kinase becomes activated as soon as Myt1 is inactivated. In the same experiment, oocytes injected with Mos antisense oligonucleotides and treated with U0126 did not produce Myt1 phosphorylation at the same time (lane 10), because either the faint amount of Mos synthesized was not sufficient to trigger Myt1 phosphorylation without Cdc2 activation, or the samples were not taken at the right time.

These results are consistent with the view that Mos protein may be necessary and sufficient to trigger Myt1 phosphorylation on some sites, independently of MAPK activation when cyclin B-Cdc2 kinase is not still activated. Yet, p90rsk and/or Cdc2 kinase activation is required for complete Myt1 hyperphosphorylation and, presumably, its full inactivation.

Fully grown, stage VI oocytes are arrested in prophase of meiosis I and are induced to mature by exposure to progesterone. Progesterone stimulates the synthesis of Mos protein, which has been shown to be the candidate initiator of Xenopus oocyte maturation, as microinjection of Mos mRNA into oocytes activated MPF and induced GVBD in absence of progesterone (Sagata et al., 1989) and, possibly, in absence of protein synthesis (Yew et al., 1992). However, Nebreda and Hunt (Nebreda and Hunt, 1993) showed that Mos protein consistently activated MAPK in oocyte extracts, whereas the activation of MPF was variable, except when oocytes were stimulated by progesterone, suggesting that one role of Mos kinase was to maintain high MAPK activity in meiosis and that protein synthesis was needed to completely activate MPF.

Several studies, using either Mos antisense oligonucleotides, or anti-MEK antibodies or MAPK phosphatase, have suggested that the Mos-MAPK pathway is necessary for the resumption of meiosis in Xenopus oocytes (Gotoh et al., 1995; Kosako et al., 1994; Sagata et al., 1988). MAPK activates p90rsk, which phosphorylates and inactivates Myt1, the inhibitory kinase of Cdc2 (Palmer et al., 1998). This pathway may explain why the activation of MAPK leads to MPF activation. However, recent results using the Hsp90 inhibitor geldanamycine or the MEK inhibitor U0126, indicate that, in the absence of detectable MAPK activation, oocytes still undergo MPF activation and GVBD upon progesterone stimulation, although with a delay when compared with control oocytes (Fisher et al., 1999; Gross et al., 2000). As the only identified in vivo target of Mos is the highly specific MAPK kinase MEK (Posada et al., 1993), which activates MAPK pathway, these results lead us to question whether Mos is required for triggering GVBD and, in this case, what might be its target in addition to MEK.

Using Mos antisense oligonucleotides, we showed that a faint amount of Mos protein was still synthesized in injected oocytes and that 75% of the oocytes underwent GVBD, even though delayed when compared with control oocytes. MAPK must be phosphorylated on the Tyr190 and the Thr188 of its T-loop to be activated (Anderson et al., 1990). Dual phosphorylation of MAPK by MEK proceeds by an ordered mechanism (Haystead et al., 1992), leading to a short inactive monophosphorylated intermediate. In maturing oocytes injected with Mos antisense oligonucleotides, only a faint proportion of MAPK was diphosphorylated at the time of GVBD, while the process of monophosphorylation occurred, demonstrating that MEK kinase was activated (because that process did not exist in the presence of the MEK inhibitor U0126). Why do MAPK diphosphorylation and activation not occur while MEK kinase is activated? If in Xenopus, as in mouse, Mos also inhibits a phosphatase activity that inactivates MAPK (Verlhac et al., 2000), we can imagine that the faint amount of Mos proteins (5-10%) is sufficient to activate MEK kinase but not enough to inhibit MAPK phosphatase (Sohaskey and Ferrell, 1999).

The use of Mos antisense oligonucleotides alone, in our hands, was not sufficient to suppress Mos synthesis completely. It has been shown that Eg2 kinase controls Mos mRNA translation, by phosphorylating CPEB, a protein required for cytoplasmic polyadenylation (Mendez et al., 2000). We used antisense Eg2 oligonucleotides as a tool to inhibit Mos synthesis in oocytes injected with Mos antisense oligonucleotides. We verified that antisense Eg2 nucleotides did not act on an essential protein other than Mos, without excluding the possibility that they could have an effect on protein(s) upstream of and necessary for Mos synthesis (Barkoff et al., 1998), or that Eg2 synthesis may become essential when Mos protein is absent. Nevertheless, the mechanism by which the antisense Eg2 oligonucleotides can suppress Mos synthesis in such oocytes requires further investigations. Using co-injection of Mos and Eg2 antisense oligonucleotides, we succeeded in completely suppressing Mos synthesis in 90% of oocytes; no monophosphorylation of MAPK (i.e. complete ablation of MEK activation) was observed in these oocytes, while oocytes injected with Eg2 antisense alone readily activated the MAPK cascade and underwent GVBD. The co-injected oocytes did not undergo GVBD, confirming that Mos is essential for re-entry into meiosis after progesterone stimulation. These results raise a paradox: on the one hand, MAPK activation seems to be dispensable for MPF activation in Xenopus, as in other species, such as starfish or the mouse (Colledge et al., 1994; Hashimoto et al., 1994; Picard et al., 1996); on the other hand, Mos, whose only known function is to activate MAPK cascade, is essential for triggering GVBD. As Mos is still required in oocytes that undergo GVBD in the absence of MAPK activation, at least another target for Mos should exist, in addition to MEK, to trigger MPF activation.

It has recently been shown in mouse oocytes that Mos exerts control of MAPK activation independently of MEK activation (Verlhac et al., 2000), by inhibiting a phosphatase activity that inactivates MAPK. This new function of Mos still involves the MAPK pathway and does not explain why Mos is still required for Xenopus oocytes to enter into metaphase of meiosis I when MAPK is not activated. Therefore, what might the Mos target be in this case?

To search for the unknown Mos target, we decided to develop an approach that by-passed the transduction pathway triggered by progesterone. First, we showed that Mos facilitates the conversion of inactive pre-MPF into active MPF form at a post-translational level. Given that this effect was specific to the first meiosis, we looked for an effect of Mos on a direct Cdc2 regulator that acts predominantly during this period. We found that Mos associated with Myt1 in oocyte extracts, regardless of whether oocytes underwent GVBD. As Myt1 has been reported to be phosphorylated by p90rsk, a target of MAPK, this result might be considered surprising. However, we found that Myt1 was highly but not completely phosphorylated in oocytes stimulated by progesterone in the absence of MAPK activation, demonstrating that Myt1 could be phosphorylated, at least partially, by another kinase in addition to p90rsk. Moreover, we observed a low level of Myt1 phosphorylation by Mos in vitro, suggesting that Mos could phosphorylate Myt1 on some sites in vivo.

We have shown that Mos triggers GVBD by facilitating MPF activation, independently of MAPK activation, and that Mos associates with Myt1. However, the mechanism by which this interaction decreases Myt1 activity was not elucidated. Verlhac et al. (Verlhac et al., 2000) have suggested that Mos controls a okadaic acid (OA)-sensitive phosphatase negatively involved in the MAPK pathway. However, it has been shown, particularly in starfish (where MPF is activated upstream MAPK activation), that MPF activation is also negatively controlled by a OA-sensitive phosphatase, because injection of okadaic acid prevents MPF inactivation by Myt1 (Picard et al., 1991). The same OA-sensitive phosphatase possibly controls both dephosphorylation of MAPK and the MAPK-independent dephosphorylation of Myt1. One hypothesis to explain MAPK pathway-independent Mos function might that the interaction between Mos and Myt1 restricts the access of a OA-sensitive phosphatase to Myt1 kinase, and thus facilitates its inactivation (Fig. 9). The fact that Mos triggers GVBD in presence of U0126 only when the oocytes are stimulated by progesterone (this study) (Gross et al., 2000) suggests that at least one unidentified protein is required besides Mos to activate MPF independently of MAPK cascade.

In summary, the main discovery of this work is that Mos, whose only function was previously believed to activate the MAPK cascade, is also involved, independently of this pathway, in activation of MPF by targeting Myt1.

We thank D. Fisher and the members of the laboratory for helpful discussion and C. Prigent for providing us the Eg2 antibodies. M. P. was supported by a LNCC fellowship. J. C. L. is supported by the CNRS and LNCC, and E. M. is supported by the CNRS and ARC (grant number 4469).