Control of Cdc6 accumulation by Cdk1 and MAPK is essential for completion of oocyte meiotic divisions in Xenopus

In proliferative cells, Cdc6 is an important protein for the initiation of DNA replication by allowing the assembly of pre-replicative complexes (pre-RCs) during G1 on replicative origins (Diffley, 2004). Independently of this licensing activity, Cdc6 is required for spindle formation during mitosis and meiosis (Anger et al., 2005; Illenye and Heintz, 2004; Narasimhachar et al., 2012; Yim and Erikson, 2010). Cdc6 further regulates M-phase progression by activating checkpoint mechanisms at the G2/M transition and by inhibiting Cdk1 activity during mitosis exit (Borlado and Mendez, 2008; Yim and Erikson, 2010,, 2011). Intriguingly, little is known regarding the role and the regulation of Cdc6 during the meiotic cell cycle besides its requirement for the acquisition of competence to replicate DNA in female germ cells.

Xenopus laevis oocytes are naturally arrested in prophase of the first meiotic division (meiosis I) and resume meiosis at the time of ovulation following progesterone stimulation. Oocytes progress through the first and then second meiotic division (meiosis II), where they halt at the metaphase (M)II stage, awaiting fertilization. Importantly, DNA replication is repressed between the two meiotic divisions in order to generate haploid gametes. Prophase oocytes have developed a similar strategy to that of quiescent cells arrested in G0 to avoid unscheduled DNA replication during their long-lasting arrest. In both cases, the inability to replicate is due to the absence of Cdc6. As cells re-enter the cell cycle, either in G1 for somatic cells or in meiosis I for oocytes, the ability to replicate is restored through Cdc6 accumulation (Cook et al., 2002; Duursma and Agami, 2005; Lemaître et al., 2002,, 2004; Mailand and Diffley, 2005; Stoeber et al., 1998; Whitmire et al., 2002). In Xenopus oocytes, Cdc6 is translated from maternal mRNAs and starts to accumulate 45–60 min after germinal vesicle breakdown (GVBD, which marks entry into the first meiotic division) (Lemaître et al., 2002; Whitmire et al., 2002). Once Cdc6 is present, the replicative machinery becomes immediately functional, but DNA replication is inhibited (Lemaître et al., 2002; Whitmire et al., 2002). Because Cdc6 accumulation coincides with entry into meiosis II, the timing of Cdc6 accumulation should be essential to prevent unscheduled DNA replication between the two meiotic divisions. The mechanisms regulating the timing of Cdc6 accumulation and the functional consequences of its misregulation on meiotic divisions remain unknown.

Two master kinases orchestrating meiotic divisions, the Cdk1–Cyclin-B complex and mitogen-activated protein kinase (MAPK; only one MAPK protein is expressed in Xenopus oocytes), are good candidates for controlling the timing of Cdc6 accumulation. The first one, called M-phase promoting factor (MPF), is a complex between Cdk1 and Cyclin B, and a potent inhibitor of DNA replication in somatic cells (Diffley, 2004; Gautier et al., 1990,, 1988). In Xenopus prophase oocytes, Cdk1–Cyclin-B complexes are inactive owing to inhibitory phosphorylation at residues T14 and Y15 on Cdk1 (Karaiskou et al., 2001). Upon progesterone stimulation, Cdk1 is dephosphorylated and activated, leading to GVBD and entry into meiosis I (Gautier et al., 1989; Karaiskou et al., 2001). Once oocytes reach MI, the degradation of Cyclin B inactivates Cdk1 within 1 h of GVBD to promote anaphase I and exit from meiosis I. At this stage, although degradation of Cyclin B is not yet complete, newly synthesized Cyclin B reaccumulates and reactivates Cdk1 to initiate entry into meiosis II. In Xenopus oocytes, inhibiting Cdk1 reactivation abolishes entry into meiosis II and triggers the formation of interphasic replicative nuclei (Furuno et al., 1994; Huchon et al., 1993; Iwabuchi et al., 2000; Ohsumi et al., 1994). Therefore, the MI–MII transition is a sensitive period during which the activity of Cdk1, the main replicative inhibitor, decreases, but Cdc6 synthesis has already started. Importantly, DNA must not replicate during the transition to reduce oocyte ploidy.

The second kinase network that might regulate DNA replication is the Mos–MAPK pathway. Mos is a germ-line specific kinase that is synthesized in response to progesterone from maternal mRNA that is indirectly responsible for the activation of MAPK at GVBD (Dupré et al., 2011). The Mos–MAPK pathway remains active from GVBD to MII, where it plays an essential role in maintaining MII arrest until fertilization (Bhatt and Ferrell, 1999; Dupré et al., 2011,, 2002; Gross et al., 2000; Haccard et al., 1993; Sagata et al., 1989). As for Cdk1, inhibiting this pathway during the MI–MII transition promotes DNA replication in oocytes (Dupré et al., 2002; Furuno et al., 1994; Gross et al., 2000), clearly demonstrating that both Cdk1 and the Mos–MAPK pathway negatively control DNA replication in oocytes.

Our results provide new insights into the mechanisms regulating Cdc6 accumulation during the meiotic divisions. We show here that Cdc6 becomes unstable very early upon entry into meiosis I, being degraded by a mechanism that is dependent on the proteasome and a Skp1–Cullins–F-box complex in which Cdc4 is the F-box protein (SCF^Cdc4; Cdc4 is also known as Fbxw7). Thereafter, both the Cdk1–Cyclin-B complex and the Mos–MAPK pathway control the turnover of Cdc6 in opposing ways, leading to the gradual accumulation of Cdc6 during entry into meiosis II. Tight regulation of Cdc6 accumulation is essential for meiotic maturation because the precocious expression of Cdc6 in prophase or in MI impairs meiosis resumption or drives the oocyte into a pseudo-interphasic replicative state by inhibiting Cdk1. Therefore, the accumulation of Cdc6 needs to be strictly synchronized with Cdk1 reactivation after meiosis I to ensure the success of the meiotic cell cycle progression.

To analyze Cdc6 turnover during meiotic maturation, prophase oocytes were injected with mRNA encoding Xenopus histidine-tagged wild-type Cdc6 (His–WT-Cdc6) and then stimulated with progesterone to induce resumption of meiosis (Fig. 1A). Cdk1 and MAPK activation was examined by immunoblotting for Cyclin B2 (also known as Ccnb2), of which the phosphorylation and up-shift by SDS-PAGE depends on Cdk1 activation, and MAPK phosphorylation (Fig. 1A). Cdc6 protein was visualized using an antibody against histidine (Fig. 1A). As expected, both Cdk1 and MAPK were activated at GVBD and in MII (Fig. 1A). His–WT-Cdc6 was efficiently translated in prophase but became undetectable 2 h after progesterone addition (Fig. 1A). Cdc6 was detected 30 min later at GVBD and was still present in MII (Fig. 1A). To confirm that loss of His–WT-Cdc6 before GVBD was due to proteolytic degradation, prophase oocytes were injected with recombinant GST-tagged Xenopus wild-type Cdc6 protein (GST–WT-Cdc6) and Cdc6 stability was verified using an antibody against GST. GST–WT-Cdc6 remained stable for at least 5 h following its injection in prophase (supplementary material Fig. S1A). Oocytes were further induced to mature by using progesterone (Fig. 1B). GST–WT-Cdc6 was stable up to 180 min following hormonal stimulation but became undetectable in oocytes, at the time when 50% of oocytes reached GVBD (GVBD₅₀), independently of GVBD status (Fig. 1B). Therefore, progesterone induces degradation of Cdc6 just before GVBD. Because the proteasome inhibitor MG132 or the prior injection of the Cdk inhibitor p21^Cip1 (also known as Cdkn1a) abolished degradation of GST–WT-Cdc6 at GVBD (Fig. 1C), the underlying mechanism relies on the proteasome and on the initial activation of Cdk1.

In various cell lines, Cdc6 turnover depends on multiple domains, including one TPxxS binding site for the F-box protein Cdc4, which mediates SCF^Cdc4-dependent degradation (Clijsters and Wolthuis, 2014; Perkins et al., 2001; Petersen et al., 2000). We mutated the F-box-binding motif TPxxS of Xenopus Cdc6 into APxxS, and the corresponding mRNA or recombinant GST–APxxS-Cdc6 were injected into prophase oocytes (Fig. 1A,C). Following progesterone addition, overexpressed His–APxxS-Cdc6 and recombinant GST–APxxS-Cdc6 proteins were no longer degraded during meiosis resumption (Fig. 1A,C). Therefore, ectopic Cdc6 becomes unstable and is degraded shortly before GVBD by a Cdk1- and proteasome-dependent mechanism involving its F-box-binding domain and presumably SCF^Cdc4.

We then investigated whether the absence of endogenous Cdc6 is due to its degradation or the inhibition of its translation in prophase or during meiosis resumption. Prophase oocytes were selected at various steps of oogenesis from stage I to full-grown stage VI (Dumont, 1972), and lysates were immunoblotted for Cdc6. The protein was absent from stage-II oocytes onwards (Fig. 1D). Moreover, inhibiting either Cdk1 activation with p21^Cip1 or the proteasome with MG132 was not sufficient to promote Cdc6 stabilization and detection in prophase oocytes, or in response to progesterone (Fig. 1E; supplementary material Fig. S1B), suggesting that the synthesis of endogenous Cdc6 is repressed during prophase arrest of growing oocytes and is re-initiated around GVBD following Cdk1 activation.

Cdc6 stability was further analyzed during the MI–MII transition by injecting GST–WT-Cdc6 or GST–APxxS-Cdc6 at GVBD (Fig. 2). Following progesterone stimulation, Cyclin B2 was phosphorylated at GVBD and degraded during the MI–MII transition, and then re-accumulated in MII (Fig. 2). Cdk1 activity was estimated by monitoring the phosphorylation at residue T320 of a peptide derived from the protein phosphatase 1 subunit PPP1CC (GST–PP1C-S), as described in Lewis et al. (2013) (Fig. 2). In agreement with the Cyclin B2 pattern, Cdk1 underwent two successive waves of activation, corresponding respectively to the GVBD–MI period and entry into MII (Fig. 2). MAPK was phosphorylated at GVBD and remained active until MII (Fig. 2). When injected at GVBD, both GST–WT-Cdc6 and GST–APxxS-Cdc6 remained detectable up to 4 h after GVBD, and oocytes proceeded normally through the MI–MII transition (Fig. 2). Therefore, a mechanism that stabilizes Cdc6 becomes effective during the MI–MII transition following Cdk1 activation in meiosis I. The same experiment was performed in the presence of an inhibitor of protein synthesis, cycloheximide, added at GVBD. As expected, cycloheximide prevented the accumulation of Cyclin B2 and Wee1, a kinase that accumulates during entry into MII (Fig. 2). Under these conditions, GST–WT-Cdc6 and GST–APxxS-Cdc6 protein levels decreased substantially 180 min after GVBD (Fig. 2), revealing that Cdc6 is unstable in meiosis I and is degraded by a SCF^Cdc4-independent mechanism. Taken together, these results demonstrate that once Cdk1 is activated, Cdc6 is stabilized in a protein-synthesis-dependent manner during the MI–MII transition. Because the addition of cycloheximide at GVBD prevents Cyclin B2 reaccumulation, and consequently leads to Cdk1 and MAPK inactivation, this opens up the possibility that one of these activities, or both of them, control Cdc6 stability after GVBD.

In yeast and somatic cells, Cdc6 can be stabilized by Cdk-dependent phosphorylation at serine or threonine residues [(S/T)-P sites], or by direct interaction with mitotic cyclins (Borlado and Mendez, 2008; Mimura et al., 2004). We first generated a mutant of Cdc6 (GST–7A-Cdc6) in which the (S/T)-P sites were mutated into alanine residues. Although GST–WT-Cdc6 and GST–APxxS-Cdc6 were efficiently thiophosphorylated in vitro by Cdk1 and MAPK (supplementary material Fig. S2A,B), the 7A mutant was not (supplementary material Fig. S2C,D). Furthermore, GST–7A-Cdc6 remained stable when injected at GVBD and was degraded after cycloheximide addition (Fig. 2B). Therefore, Cdc6 phosphorylation at (S/T)-P sites is not required to protect Cdc6 from degradation after MI.

To address the role of B-type cyclins in Cdc6 stabilization, low levels of a non-degradable form of a truncated mutant termed Δ90-Cyclin-B were injected together with GST–WT-Cdc6 at GVBD, and cycloheximide was concomitantly added (Fig. 2A). Although the amount of injected Δ90-Cyclin-B was too low to reactivate Cdk1 and MAPK in the presence of cycloheximide, GST–WT-Cdc6 was no longer degraded following GVBD (Fig. 2A). Non-degradable Cyclin B is thus sufficient to stabilize Cdc6 independently of Cdk1 activity. We next asked whether GST–WT-Cdc6 interacts with Cdk1–Cyclin-B complexes. GST–WT-Cdc6 was added to prophase or MII extracts, and GST was pulled down. Both Cyclin B2 and Cdk1 were recovered with GST–WT-Cdc6 in MII extracts, as seen by immunoblotting (Fig. 3A). Control MII-arrested oocytes or oocytes injected at GVBD with GST–WT-Cdc6 were then fractionated using gel filtration chromatography on Superose-12 columns (Fig. 3B,C). Cdk1 and Cyclin B2 eluted in distinct fractions; fractions 7–9 correspond to high-molecular-mass complexes, and fractions 24–26 correspond to the dimeric Cdk1–Cyclin-B complex, as previously described in De Smedt et al. (2002). Interestingly, endogenous Cdc6 was recovered together with Cdk1 and Cyclin B2 in the high-molecular-mass complexes (Fig. 3B,C). In Cdc6-injected oocytes, the recombinant protein was also recovered in the fraction containing high-molecular-mass complexes (Fig. 3C), suggesting that GST–WT-Cdc6 interacts with Cdk1–Cyclin-B complexes during the MI–MII transition.

To determine whether this interaction exists in ovo, Cdk1–Cyclin-B complexes were affinity-purified from prophase or MII-arrested oocytes using p13^suc1 beads (beads coated with the product of the yeast gene suc1). Importantly, endogenous Cdc6 was detected by immunoblotting on p13^suc1 beads from MII oocytes (Fig. 4A). To further confirm the interaction between endogenous Cdc6 and Cdk1–Cyclin-B, MII extracts were incubated with an antibody against Cdc6 and then fractionated on a Superose-6 column (Fig. 4B). As previously observed, endogenous Cdc6 eluted with Cdk1 and Cyclin B2 in control oocytes (Fig. 4B). In the presence of an antibody against Cdc6, Cdc6 still co-eluted with Cdk1–Cyclin-B in fraction 3 (Fig. 4B). IgGs were then recovered from fraction 3 by adding protein-G-coupled beads and Cyclin B2, Cdk1 and Cdc6 were detected on the beads (Fig. 4C). Therefore, Cdk1–Cyclin-B and Cdc6 form a trimeric complex in meiosis-II oocytes. Altogether, these data strongly suggest that the interaction between Cdc6 and Cdk1–Cyclin-B complexes contributes to its stabilization after GVBD.

To investigate the roles of B-type cyclins and the Mos–MAPK pathway on Cdc6 stabilization, the accumulation of endogenous Cdc6 was monitored in oocytes that had been injected with antisense oligonucleotides targeting B-type cyclins or Mos mRNAs in order to abolish Cdk1 reactivation and MAPK activation, respectively (Fig. 5). In control oocytes, Cdk1 activity followed the usual two-wave pattern, whereas MAPK was activated at GVBD and remained active until MII (Fig. 5A). As already reported by Lemaître et al. (2002) and Whitmire et al. (2002), Cdc6 accumulated progressively, starting 45–60 min after GVBD until MII (Fig. 5A). According to Haccard and Jessus (2006) and Hochegger et al. (2001), injecting antisense oligonucleotides against B-type cyclins did not block the first wave of activation of Cdk1 and GVBD, but did abolish Cdk1 reactivation and entry into meiosis II (Fig. 5B). As a consequence of Cdk1 inactivation, MAPK was inactivated 180 min after GVBD (Fig. 5B). Under these conditions, Cdc6 accumulation was strongly delayed compared with that in controls – it was only detectable 150 min after GVBD (Fig. 5B) – suggesting that B-type cyclins positively regulate the accumulation of endogenous Cdc6, in agreement with our previous hypothesis. It is noteworthy that Cdc6 accumulation started when MAPK was inactivated in oocytes that had been injected with antisense oligonucleotides against B-type cyclins (Fig. 5B), suggesting that MAPK could negatively regulate Cdc6 accumulation. We then impeded the synthesis of Mos by injecting morpholino antisense oligonucleotides and, as a consequence, MAPK was neither phosphorylated nor activated following hormonal stimulation (Fig. 5C). Interestingly, the accumulation of endogenous Cdc6 occurred earlier than in control oocytes, almost at the same time as GVBD (Fig. 5C). The accumulation of endogenous Cdc6 was also advanced in progesterone-treated oocytes that had been incubated with the MEK inhibitor U0126, which prevents MAPK activation during meiotic maturation (supplementary material Fig. S3). Altogether, our results demonstrate the positive role of the accumulation of B-type cyclins on endogenous Cdc6 accumulation, counterbalanced by a negative regulation that is exerted by the Mos–MAPK pathway.

Interestingly, Cdc6 can inhibit Cdk1 during the G2/M transition (Archambault et al., 2003; Calzada et al., 2001; Clay-Farrace et al., 2003; El Dika et al., 2014; Elsasser et al., 1996; Murakami et al., 2002; Oehlmann et al., 2004). If such effect were to operate in oocytes, deregulation of Cdc6 accumulation is expected to disturb meiotic divisions. As seen in Fig. 1, injection of 0.05 µM GST–WT-Cdc6 or GST–APxxS-Cdc6 did not prevent meiosis resumption. To determine whether Cdk1 inhibition by Cdc6 is dose-dependent, increasing amounts of GST–WT-Cdc6 were added to extracts from MII-arrested oocytes, and Cdk1 activity was assayed in vitro using GST–PP1C-S (supplementary material Fig. S4A). Cdk1 was fully active in MII extracts and was inhibited in a dose-dependent manner by GST–WT-Cdc6 (supplementary material Fig. S4A). Cdk1 activity was inhibited by 70% with 0.5 µM of GST–WT-Cdc6 in MII extracts (supplementary material Fig. S4A). This concentration is ten times higher than that used in Fig. 1, but corresponds to the physiological concentration of Cdc6 in MII-arrested oocytes (supplementary material Fig. S4B).

Next, the amount of GST–WT-Cdc6 or GST–APxxS-Cdc6 that was injected into prophase oocytes was increased to 0.5 µM (Fig. 6). Following the addition of progesterone, GVBD induction was slightly delayed in the presence of GST–WT-Cdc6 and strongly impaired with GST–APxxS-Cdc6 (Fig. 6A). Moreover, the levels of Cdk1 activity at GVBD were reduced in GST–WT-Cdc6-injected oocytes compared with those of control oocytes, as determined by using a Cdk1 activity assay and phosphorylation of Cyclin B2 and Greatwall (also known as Mastl) (Fig. 6B). Following injection of GST–APxxS-Cdc6, Cdk1 activity was strongly reduced in oocytes that reached GVBD and was not detectable in oocytes that failed to mature (Fig. 6B). Therefore, the presence of a stable form of Cdc6 in prophase impairs Cdk1 activation in response to progesterone. Oocytes that performed GVBD after injection of GST–WT-Cdc6 or GST–APxxS-Cdc6 were further analyzed 4 h later. Cdk1 activity was strongly reduced, Cyclin B2 and Greatwall were dephosphorylated, and MAPK was inactivated (Fig. 6B), suggesting that the oocyte did not enter meiosis II. The external morphology of these oocytes was altered (Fig. 6C), with pigment re-arrangements that were similar to those at egg activation, as well as at exit from meiosis and entry into interphase. Therefore, depending on the amount of Cdc6 present in prophase, Cdc6 has the potential to inhibit Cdk1 activation before GVBD but also after MI.

We then examined whether deregulating Cdc6 expression levels after the first meiotic division prevents entry into meiosis II. Oocytes were injected at GVBD with 0.5 µM of GST–WT-Cdc6, and Cdk1 activity was monitored at the indicated times (Fig. 7A). In control oocytes, Cdk1 activity increased at GVBD, then decreased 1 h later and rose again when oocytes entered MII (Fig. 7A). In oocytes that had been injected with 0.5 µM of GST–WT-Cdc6 at GVBD, Cdk1 activity decreased after GVBD, as in control oocytes, but did not increase thereafter (Fig. 7A). Furthermore, these oocytes did not enter the second meiotic division, as determined by immunoblotting for several molecular markers of meiosis II entry – phosphorylation of Greatwall, accumulation of Wee1 and Cyclin B2, as well as activation of MAPK (Fig. 7B). These results indicate that oocytes injected at GVBD with high amounts of GST–WT-Cdc6 exit from meiosis I but do not proceed through meiosis II. Instead, they are driven into a pseudo-interphasic state owing to inhibition of Cdk1 reactivation.

In somatic cells and in Xenopus egg extracts, overexpression of Cdc6 prevents Cdk1 activation by activating checkpoints that are dependent on ATM and/or ATR, or Chk1 (Borlado and Mendez, 2008). To determine whether this pathway mediates the negative effect of Cdc6 towards Cdk1, prophase oocytes were incubated with an inhibitor of ATM and ATR, caffeine, and then stimulated with progesterone to resume meiosis (Fig. 7). Oocytes were further injected at GVBD with 0.5 µM of GST–WT-Cdc6 (Fig. 7). As expected, caffeine prevented Chk1 phosphorylation in MII-arrested oocytes that had been injected with a DNA substrate that contained double-strand breaks (Fig. 7A, inset), but neither affected meiosis resumption nor Cdk1 activation in progesterone-stimulated oocytes (Fig. 7). Moreover, caffeine was unable to restore Cdk1 reactivation when GST–WT-Cdc6 was injected at GVBD, as seen by using GST–PP1C-S in assays (Fig. 7A) and by immunoblotting for Greatwall, Wee1, Cyclin B2 and phosphorylated MAPK (Fig. 7B). Therefore, the inhibitory effect of Cdc6 on Cdk1 activation is unlikely to be mediated by a DNA-damage-related checkpoint in Xenopus oocytes. Because 0.5 µM of GST–WT-Cdc6 interacts with Cdk1–Cyclin-B complexes without promoting their dissociation during the MI–MII transition (Fig. 3C), this direct interaction probably accounts for the inhibition of Cdk1 through Cdc6 after meiosis I.

Our results prompted us to analyze DNA replication in oocytes that had been injected at GVBD with GST–WT-Cdc6 by monitoring the in vivo incorporation of biotin–dUTP into genomic DNA using horseradish peroxidase (HRP)-coupled streptavidin. Cdc6 accumulation starts 45 min after GVBD, conferring the ability to replicate DNA (Lemaître et al., 2002; Whitmire et al., 2002). However, this competence is repressed and DNA is not replicated between the two meiotic divisions. Adding cycloheximide 45 min after GVBD, when Cdc6 is already expressed at low levels, induces DNA replication (Furuno et al., 1994; Lemaître et al., 2002) (Fig. 8A). Treatment with cycloheximide is thus a convenient way to test the competence of oocytes to replicate DNA. As seen in Fig. 8A, no incorporation of biotin–dUTP was detected in control oocytes, whether they were treated or not with cycloheximide at GVBD; however, 45 min later, adding cycloheximide promoted DNA replication (Fig. 8A). To study the effect of Cdc6 on DNA replication, prophase oocytes were first injected with biotin–dUTP, stimulated with progesterone and then injected at GVBD with 0.05 µM or 0.5 µM of GST–WT-Cdc6 (Fig. 8B). Oocytes were collected 4 h 45 min after GVBD to monitor Cdk1 activity and DNA replication, as well as levels of GST–Cdc6 and endogenous Cdc6 (Fig. 8C). In control oocytes, Cdk1 was active, and biotin–dUTP was not incorporated (Fig. 8C). As expected, adding cycloheximide 45 min after GVBD prevented Cdk1 reactivation, and oocytes underwent DNA replication (Fig. 8C). Injecting a low concentration of GST–WT-Cdc6 (0.05 µM) at GVBD did not affect entry into meiosis II – Cdk1 was efficiently reactivated, and no DNA replication was observed (Fig. 8C). By contrast, the injection of 0.5 µM GST–WT-Cdc6 at GVBD abolished Cdk1 reactivation, and DNA replication was promoted after meiosis I (Fig. 8C).

We then tested whether 0.05 µM of GST–WT-Cdc6, which does not prevent entry into meiosis II as shown above, is sufficient to support DNA replication. For this purpose, the synthesis of endogenous Cdc6 was prevented by using antisense oligonucleotides before injection of biotin–dUTP (Fig. 8B). Oocytes were then stimulated with progesterone, injected at GVBD with 0.05 µM of GST–WT-Cdc6 and, 45 min after GVBD, were further incubated with cycloheximide to monitor DNA replication (Fig. 8B). The inhibition of endogenous Cdc6 synthesis abolished the competence to replicate DNA, but injection of 0.05 µM of GST–WT-Cdc6 at GVBD rescued DNA replication in oocytes (Fig. 8C). Therefore, GST–WT-Cdc6 injected at a concentration that is too low to inhibit Cdk1 is sufficient to support DNA replication after meiosis I.

To ascertain whether DNA replication following GST–WT-Cdc6 injection at GVBD is a consequence of Cdk1 inhibition by Cdc6, the same experiment was performed in the presence of aphidicolin (APD), an inhibitor of DNA replication (Fig. 8D). Prophase oocytes were injected with biotin–dUTP, stimulated with progesterone and injected at GVBD with 0.5 µM of GST–WT-Cdc6 in the presence or absence of APD (Fig. 8D). APD efficiently inhibited DNA replication in progesterone-stimulated oocytes that had been incubated in cycloheximide 45 min after GVBD (Fig. 8D). As previously shown, GST–WT-Cdc6 inhibited the activity of Cdk1 by 60% and led to DNA replication (Fig. 8D). As expected, no DNA replication was observed in the presence of APD, but Cdk1 activity was still inhibited by GST–WT-Cdc6 under these conditions (Fig. 8D). This demonstrates that precocious expression of Cdc6 at the beginning of the first meiotic division inhibits Cdk1 reactivation, which forces the cell into a replicative interphasic state.

In this paper we show that the regulation of Cdc6 turnover during meiotic maturation is essential to coordinate the levels of Cdc6 accumulation with entry into the second meiotic division, avoiding Cdk1 inhibition and unscheduled DNA replication.

Our data revealed that recombinant Cdc6 is stable in prophase and becomes unstable in response to progesterone, being degraded by the proteasome shortly before entry into meiosis I. This degradation mechanism relies on a SCF^Cdc4-dependent proteolytic system, controlled by initial activation of Cdk1, either directly or indirectly. In prophase, transcription is dormant, and Cdc6 expression is shut down during the long period of oocyte growth. Inhibiting the proteasome with MG132 does not permit precocious expression of endogenous Cdc6 following progesterone addition. Although we cannot exclude that our antibody is not sensitive enough to detect very small amounts of Cdc6, this result supports the idea that Cdc6 translation is repressed during prophase arrest. Moreover, because endogenous Cdc6 does not accumulate in p21^Cip1-injected oocytes following progesterone stimulation, this further indicates that the translation of endogenous Cdc6 is specifically activated downstream of the first Cdk1 activation, possibly by the polyadenylation of its mRNA, as proposed by Lemaître et al. (2002). Therefore, Cdk1 activation concomitantly initiates Cdc6 synthesis and activates SCF^Cdc4. Hence, endogenous Cdc6 protein is absent until GVBD. The protein starts to be translated at GVBD depending on Cdk1 activation but cannot accumulate at that time owing to SCF^Cdc4-dependent degradation. This intricate regulation of the stability of Cdc6 in response to progesterone is important because the presence of high Cdc6 protein levels before GVBD can inhibit the first activation of Cdk1 in a dose-dependent manner. Although overexpressing Cdc6 through injection of mRNA does not delay meiosis resumption in Xenopus (our results) and mouse oocytes (Anger et al., 2005), we show here that injection of recombinant Cdc6 protein, at concentrations corresponding to endogenous protein levels in MII (0.5 µM), decreases Cdk1 activation following hormonal stimulation. This inhibitory effect of Cdc6 over Cdk1 is stronger when a non-degradable mutant of Cdc6 is injected, because meiosis resumption is abolished in the majority of oocytes. Hence, the SCF^Cdc4-dependent proteolytic system that is responsible for degradation of Cdc6 acts as a safety mechanism to prevent any precocious accumulation of Cdc6 before GVBD, which would compromise meiosis resumption.

Starting from GVBD, the expression levels of Cdc6 and thus its accumulation, result from two mechanisms regulating its turnover, one being responsible for its stabilization and the other for its degradation. The stabilizing mechanism is set up when Cdk1 activity is established in MI and does not rely on the Cdk1-dependent phosphorylation of Cdc6, as demonstrated here with the 7A mutant of Cdc6 that remains stable after GVBD. Interestingly, the binding of mitotic cyclins stabilizes Cdc6 in yeast (Drury et al., 2000; Elsasser et al., 1999; Mimura et al., 2004; Perkins et al., 2001; Weinreich et al., 2001). This mechanism is probably conserved in Xenopus oocytes. When oocytes are treated with cycloheximide at GVBD, the degradation of ectopic Cdc6 is blocked by the presence of a non-degradable Cyclin B when it is injected at a dose too low to reactivate Cdk1. Moreover, the accumulation of endogenous Cdc6 is strongly delayed when synthesis of B-type cyclins is prevented. Because recombinant and endogenous Cdc6 both interact with Cdk1–Cyclin-B complexes, this mechanism certainly accounts for Cdc6 stabilization in MI and during entry into meiosis II. However, Cdc6 does not accumulate during the first hour following GVBD owing to partial degradation of B-type cyclins and a second effector, the Mos–MAPK pathway, which control Cdc6 turnover. The suppression of the Mos–MAPK pathway promotes precocious accumulation of endogenous Cdc6. Furthermore, endogenous Cdc6 appears unexpectedly in oocytes that have been injected with antisense oligonucleotides against B-type cyclins when the Mos–MAPK pathway is inactivated. This is the first demonstration that the Mos–MAPK pathway acts independently of B-type cyclin turnover to negatively regulate Cdc6 accumulation, certainly by inducing its degradation. Interestingly, DNA is decondensed and a nucleus is reformed after meiosis I, when the Mos–MAPK module is inactive (Furuno et al., 1994). Whether these nuclei or the state of DNA compaction are involved in controlling the turnover of endogenous Cdc6 warrants further analysis.

This tight regulation of Cdc6 turnover through B-type cyclins and the Mos–MAPK pathway after meiosis I synchronizes the timing of Cdc6 accumulation, which is essential for proper succession of the two meiotic divisions without DNA replication. The precocious expression of 0.5 µM Cdc6 in MI impairs Cdk1 reactivation. As a consequence, oocytes enter a replicative interphase-like stage. This process involves the direct interaction between Cdc6 and Cdk1–Cyclin-B, a mechanism already known to contribute to exit from M-phase in Xenopus extracts, yeast and human cells (Archambault et al., 2003; Calzada et al., 2001; El Dika et al., 2014; Elsasser et al., 1996; Perkins et al., 2001; Weinreich et al., 2001). Interestingly, Cdc6 specifically associates with Cyclin B2 and Cdk1 in high-molecular-mass complexes, which are distinct from the canonical dimeric Cdk1–Cyclin-B complexes (De Smedt et al., 2002). Whether these high-molecular-mass complexes play a pivotal role in the activation of Cdk1 is unknown. However, these results highlight the existence of a feedback loop between Cdc6 and Cdk1–Cyclin-B complexes that operates during the MI–MII transition – Cdc6 inhibits Cdk1 activity, whereas accumulation of B-type cyclins stabilizes Cdc6. Because ectopic Cdc6 is stabilized at GVBD, this explains why high levels of injected Cdc6 inhibit Cdk1 reactivation. Hence, the MI–MII transition is a crucial period for oocytes where Cdc6 starts to be synthesized and stabilized, with the potential to prevent Cdk1 reactivation and entry into meiosis II.

Why does endogenous Cdc6 not perturb meiotic divisions under physiological conditions? At the time of entry into meiosis I, although stabilized by binding to B-type cyclins, Cdc6 is present at very low levels because, at the start of its synthesis, the levels are at zero. During the MI–MII transition, B-type cyclins are degraded and, as a consequence, no longer protect Cdc6 from degradation. At the same time, the Mos–MAPK pathway, the activity of which is maintained at a high level, negatively regulates Cdc6 accumulation, probably by destabilizing the protein. For this reason, Cdc6 levels at the end of meiosis I are insufficient to inhibit Cdk1 reactivation and entry into meiosis II, although they are sufficient to support DNA replication. The Mos–MAPK pathway is therefore key to synchronize the MI–MII transition by coordinating Cdc6 expression levels with Cdk1 reactivation. Indeed, this pathway stimulates the reaccumulation of a positive regulator of Cdk1, namely B-type cyclins, and postpones the accumulation of a Cdk1 inhibitor, Cdc6, to enable entry into meiosis II. Furthermore, it prevents unscheduled DNA replication by ensuring that no pre-RCs are assembled when Cdk1 activity is low. As the oocyte proceeds through the second meiotic division and then arrests at MII, the mechanism stabilizing Cdc6 through its interaction with newly synthesized B-type cyclins bypasses the negative effect that is exerted by the Mos–MAPK pathway. Cdc6 is therefore stabilized and can accumulate. The balance between the Mos–MAPK pathway, which remains constantly active from GVBD to MII, and Cdk1–Cyclin-B complexes, the quantity and activity of which decrease and then rise again during the same period, results in a progressive and gradual accumulation of Cdc6. Importantly, this dual regulation of Cdc6 turnover by B-type cyclins and the Mos–MAPK pathway tightly controls the expression levels of endogenous Cdc6, which thus does not disturb normal progression of the meiotic divisions. Hence, the competence to replicate the DNA that is necessary for the embryonic cell cycles following fertilization is established but is repressed by the high and stable activity of Cdk1 in MII. Cdc6 is therefore a focal point that can control the correct order of events during meiotic maturation by coordinating the competence to replicate and the succession of the two meiotic divisions.

Xenopus laevis adult females [Centre de Ressources Biologiques Xenopes, UMS3387, Centre National de la Recherche Scientifique (CNRS), France] were bred and maintained under laboratory conditions (Animal Facility Agreement no. B75-05-13; Charles Darwin ethics committee, no. 5). Reagents, unless otherwise specified, were from Sigma.

Xenopus prophase oocytes were obtained as described in Haccard et al. (2012). Oocytes of different sizes (stage I to full-grown stage VI) were sorted on binocular using a micrometer and according to Dumont (1972). The microinjected volume per oocyte was 50 nl. Meiosis resumption was triggered with 2 µM progesterone. Cycloheximide, MG132, aphidicholin (APD) and caffeine were used at concentrations of 355 µM, 100 µM, 60 µM and 100 µM, respectively, 1 h before progesterone addition. Prophase oocytes were injected with antisense oligonucleotides directed against mRNAs encoding all B-type cyclins (Cyc8/B2-5 cocktail; Hochegger et al., 2001), Mos (Dupré et al., 2002) or Cdc6 (Lemaître et al., 2002). The non-degradable sea urchin Δ90-Cyclin was purified as described in Frank-Vaillant et al. (1999). Oocytes referred to as GVBD were collected at the time of the first pigment rearrangement at the animal pole. Oocytes were homogenized in ten volumes of extraction buffer (80 mM β-glycerophosphate, pH 7.3; 20 mM EGTA; 15 mM MgCl₂), centrifuged at 10,000 g for 15 min, and the supernatants were used for further analysis.

An equivalent of half or one oocyte was loaded onto 10% or 12% SDS-polyacrylamide gels (Laemmli, 1970) and immunoblotted as described in Dupré et al. (2002). Antibodies against the following proteins were used – phosphorylated MAPK (1:1000, Cell Signaling Technology, 9106); total MAPK (1:1000, Abcam, SC-154); Cyclin B2 (1:1000, Abcam, ab18250); total Cdk1 (1:1000, Abcam 3E1, ab20151); Wee1 (1:1000, Zymed, 51-1700); PP1C phosphorylated at T320 (1:30,000, Abcam, ab62334); alkylated thiophosphates (1:30,000, Abcam, ab92570); Cdk1 phosphorylated at Y15 (1:1000, Cell Signaling Technology, 9111) and Chk1 phosphorylated at S345 (1:1000, Cell Signaling Technology, 2348). HRP-conjugated antibodies directed against either GST (1:5000, Sigma, A-7340) or Histidine (1:10,000, Sigma) were also used. The antibody against Greatwall has been described in Dupré et al. (2013). The antibody directed against Xenopus Cdc6 was raised by immunizing rabbits with recombinant GST–WT-Cdc6; it was then purified by retro-elution as described in Burke et al. (1982). Appropriate HRP-labeled secondary antibodies (Jackson ImmunoResearch) were used in chemiluminescent analysis (Pierce).

Full-length wild-type Xenopus Cdc6 cDNA was purchased from Open Biosystem (clone ID, 6634013; accession no., BC070554). DNA encoding histidine-tagged WT-Cdc6 was generated by PCR amplification of full-length Cdc6 cDNA and pRN3 vector (containing a T3 promoter, kindly provided by Dr J. Moreau, CNRS, Paris) using the following primers: 5′-CATCACCACCACCACCACGGTATGCCAAGCACCAGGTCTCGGTC-3′ (encoding six histidine residues) and 5′-CACCCTCACAACGCTAAATCCCTGAATTGAGAACATTCC-3′ for Xenopus Cdc6; 5′-CACCCTCACAACGCTAAATCCCTGAATTGAGAACATTCC-3′ and 5′-GTGGTGGTGGTGGTGATGCATGAATTCAGATCTGCCAAAGTTGA-3′ for pRN3.

Xenopus histidine-tagged Cdc6 was cloned into pRN3 vector using the ligation-independent cloning technique (Aslanidis and de Jong, 1990) and was then transformed into JM109 bacteria, the resulting construct was sequenced. cDNA encoding histidine-tagged APxxS-Cdc6 was generated by PCR amplification of full-length Cdc6 in pRN3 using the following primers: APxxS Fw, 5′-AAAGGGGCAAGAGGCCCCACCCAGCTC-3′; APxxS Rev, 5′-GAGCTGGGTGGGGCCTCTTGCCCCTTT-3′.

mRNAs encoding histidine-tagged WT- and APxxS-Cdc6 were produced using the AmpliCap-Max T3 high-yield message maker kit (Epicentre).

cDNAs encoding histidine-tagged WT- or APxxS-Cdc6 were subcloned into pGex4T1 (GE Healthcare) within EcoR1 and Not1 sites. GST–7A-Cdc6 was generated by mutating residues S45, S54, S74, S88, S108, S120 and S411 into alanine residues within GST–WT-Cdc6 using the primers described in Pelizon et al. (2000). The expression of recombinant GST–Cdc6 proteins was induced in BL21 pLys bacteria (Promega) by autoinduction for 3 days at 17°C (Studier, 2005). Bacteria were lysed in lysis buffer (20 mM Hepes, pH 7.8; 0.5 M NaCl; 1 mM EDTA; 1 mM EGTA) supplemented with 0.1% lysozyme, 0.1% Triton X-100 and 0.1% Igepal. After sonication, lysates were centrifuged and incubated for 1 h 30 min at 4°C with glutathione–agarose beads (Sigma, G4510) that had been previously equilibrated overnight at 4°C in lysis buffer. Glutathione beads were washed in washing buffer (20 mM Hepes, pH 7.8; 0.5 M NaCl; 0.1 mM EDTA; 0.1 mM EGTA; 5 mM DTT), and Cdc6 was eluted with 8 ml of washing buffer supplemented with 5 mM glutathione. The purification of Cdc6 proteins in each fraction was examined by using Coomassie-stained SDS-polyacrylamide gels. Fractions containing a high amount of Cdc6 were pooled and concentrated against polyethyleneglycol in the presence of 0.5 M NaCl to prevent precipitation. The final concentration of GST–Cdc6 was 10 µM. In experiments using 0.5 µM GST–Cdc6, control oocytes were injected with the corresponding purification buffer containing 0.5 M NaCl (buffer 1). In experiments using 0.05 µM GST–Cdc6, GST–Cdc6 was diluted ten times with the same buffer without salt to reach 0.05 M NaCl (buffer 2).

Recombinant GST–PP1C-S peptide was expressed from a pGEX plasmid (a kind gift from R. Golsteyn, University of Lethbridge, Canada). GST–p21^Cip1 and GST–PP1C-S were produced by using autoinduction (Studier, 2005) and purified as previously described (Frank-Vaillant et al., 1999; Lewis et al., 2013).

Inactive or active Cdk1 was affinity purified from four prophase- or MII-arrested oocytes using p13^suc1 beads and then incubated with 0.6 µg of GST–Cdc6 for 30 min at 30°C in kinase buffer (20 mM Hepes, pH 7.4; 2 mM EGTA; 10 mM β-mercaptoethanol; 0.1 mM ATPγS; 10 mM MgCl₂). Active MAPK (100 ng; Sigma, E7407) was incubated with 0.6 µg of GST–Cdc6 for 30 min at 30°C in kinase buffer (25 mM MOPS, pH 7.2; 12 mM glycerol 2-phosphate; 25 mM MgCl₂; 5 mM EGTA; 2 mM EDTA; 0.1 mM ATPγS and 0.25 mM DTT). Each reaction was stopped by adding 20 mM EDTA and then incubated for 1 h with 12.5 mM p-nitrobenzyl mesylate (Abcam, Ab138910) for alkylation. Incorporated thiophosphates were immunoblotted using an antibody directed against alkylated thiophosphates.

An extract equivalent to 0.1 oocyte was incubated with 0.2 µg of GST–PP1C-S peptide for 30 min at 30°C in kinase buffer (20 mM Hepes, pH 7.4; 2 mM EGTA; 10 mM β-mercaptoethanol; 0.1 mM ATP; 10 mM MgCl₂; 10 µM okadaic acid). The reaction was stopped with Laemmli buffer (Laemmli, 1970) and then immunoblotted for GST and PP1C that had been phosphorylated at residue T320. The signals were quantified using ImageJ software.

GST pull-down experiments were performed with 1 µg of GST–WT-Cdc6 that had been pre-coupled to 30 µl of glutathione–agarose beads (Sigma, G4520), which was then incubated for 1 h at 4°C with 100 µl of extracts of prophase or MII oocytes. Beads were then recovered, and the resultant proteins were immunoblotted.

For p13^suc1 pull down, 20 µl of control Sepharose or p13^suc1-coupled beads were incubated for 2 h with 50 µl of lysate from prophase or MII oocytes. Beads were then recovered by centrifugation, and the resultant proteins were immunoblotted.

Extracts from MII oocytes were prepared as described in De Smedt et al. (2002), and were incubated or not with an antibody against Cdc6 for 1 h. The extracts were then fractionated on either a Superose-12 or a Superose-6 gel filtration column (Amersham Biosciences). Fractions of 0.25 ml or 0.5 ml, respectively, were collected and subjected to immunoblotting. In the experiment using Superose-6 columns, 400 µl of fraction 3 was further incubated with 25 µl of protein-G-coupled beads for 2 h at 4°C; immunoblotting was then performed.

Prophase oocytes were injected with 50 µM of biotin-11–dUTP (Jena Bioscience). DNA was extracted from ten oocytes as described in Dupré et al. (2002), separated on 0.8% agarose gel and transferred overnight onto a nitrocellulose membrane by southern blotting, as described in Southern (1975). The membrane was saturated in 5% milk for 90 min and incubated overnight with streptavidin–HRP (1:10,000, Thermo Scientific) to detect incorporation of biotin–dUTP into genomic DNA.

MII-arrested oocytes were incubated for 1 h in buffer M [10 mM HEPES, pH 7.6; 88 mM NaCl; 1 mM KCl; 0.33 mM Ca(NO₃)₂; 0.41 mM CaCl₂; 0.82 mM MgSO₄] that had been supplemented with 0.1 M EGTA to prevent oocyte activation. Oocytes were then injected with HindIII-λ-digested DNA (Fermentas, SM0101), which mimics DNA double-strand breaks, and collected 1 h later for further analysis.

We thank all members of our laboratory for helpful discussions and Dr K. Wassmann for the critical reading of the manuscript.