Control of G2/M Transition in Xenopus by a Member of the p21-activated Kinase (PAK) Family: A Link Between Protein Kinase A and PAK Signaling Pathways?*

X-PAKs are involved in negative control of the process of oocyte maturation in Xenopus (1). In the present study, we define more precisely the events targetted by the kinase in the inhibition of the G2/M transition. We show that microinjection of recombinant X-PAK1-Cter active kinase into progesterone-treated oocytes prevents c-Mos accumulation and activation of both MAPK and maturation-promoting factor (MPF). In conditions permissive for MAPK activation, MPF activation still fails. We demonstrate that a constitutive truncated version of X-PAK1 (X-PAK1-Cter) does not prevent the association of cyclin B with p34 cdc2 but rather prevents the activation of the inactive complexes present in the oocyte. Proteins participating in the MPF amplification loop, including the Cdc25-activating Polo-like kinase are all blocked. Indeed, using active MPF, the amplification loop is not turned on in the presence of X-PAK1. Our results indicate that X-PAK and protein kinase A targets in the control of oocyte maturation are similar and furthermore that this negative regulation is not restricted to meiosis, because we demonstrate that G2/M progression is also prevented in Xenopus cycling extracts in the presence of active X-PAK1.

In Xenopus laevis, oocytes are naturally arrested at the G 2 /M transition of meiosis I. Upon release of progesterone by the follicle cells surrounding the oocytes, maturation occurs and the cell cycle resumes. Translation of the proto-oncogene c-Mos (a MAPK 1 kinase kinase), is an essential preliminary event (2,3) in turning on the MAPK cascade, which eventually results in almost simultaneous activation of MAPK and maturation-promoting factor (MPF), a heterodimer composed of p34 cdc2 kinase and cyclin B (4,5). Eggs eventually arrest again at metaphase of second meiosis with high MPF and cytostatic factor (in which c-Mos participates (6)) activities until fertilization. Conversion of inactive pre-MPF complexes, stored in oocytes, into active MPF requires dephosphorylation of residues Thr 14 and Tyr 15 on p34 cdc2 kinase (7). Dephosphorylation of these amino acids is crucial for MPF activation, and this process is highly regulated by an activating phosphatase, Cdc25 (for review see Ref. 8), and two inhibitory kinases, Myt1 (9) and Wee1 (10).
Although oocyte maturation has been extensively studied over the past 10 years as a model to understand the mechanisms involved in reentry into the cell cycle, the signal transduction pathway between progesterone binding to its receptor and the activation of the MAPK cascade and MPF is ill understood. Protein kinase A (PKA) appears to be a crucial player in these events, because activation of progesterone receptor is followed by a sudden drop in cAMP concentration (11) and likely a subsequent inactivation of protein kinase A. Indeed, microinjection of the PKA catalytic subunit prevents progesterone-induced maturation in oocytes, whereas the expression of the PKA regulatory subunit is sufficient to induce maturation (12). Matten et al. (13) reported that PKA negatively regulates maturation by controlling both c-Mos translation and Cdc25 activation. How PKA acts on c-Mos de novo translation is, however, unclear, and we have demonstrated that in conditions in oocytes in which activation of MAPK is allowed, PKA does not inhibit the c-Mos translation (14), raising the possibility that PKA could act on a single target in the negative regulation of maturation.
A number of studies demonstrate the importance of the MAPK cascade in MPF activation (6,15) during Xenopus oocyte maturation. Indeed, microinjection of recombinant c-Mos is itself sufficient to induce oocyte maturation in the absence of hormonal treatment (16,17), whereas CL100 (MAPK phosphatase-1) inhibits progesterone-induced maturation (18). Important work in Saccharomyces cerevisiae toward the understanding of the control of the MAPK cascade has been undertaken in the last few years. The response to the mating pheromone signal is mediated by the direct interaction of the G␤ subunit of G protein heterotrimers to Ste20, a serine threonine kinase (19). Activation of Ste20 results in activation of the MAPK kinase kinase Ste11, which allows the MAPK cascade to be turned on and results in MAPK (Kss1 and Fus3) activation (reviewed in Ref. 20). In Xenopus oocytes, Gotoh et al. (18) showed that microinjection of constitutively active Ste11 is able to efficiently induce the maturation process. Moreover, addition of the N-terminal regulatory domain of Ste11 to Xenopus extracts prevents a constitutive form of Ste20 from activating MAPK (21). p21-activated kinases (PAKs) are Ste20 homologues. Understanding of their functional importance in signal transduction and in regulation of the actin cytoskeleton is rapidly growing (see Refs. [22][23][24]. We have cloned Xenopus PAKs and recently demonstrated that a member of this family is involved in regulation of the oocyte maturation process (1). Indeed, we showed that microinjection of a constitutive truncated version of X-PAK1 (X-PAK1-Cter) into oocytes completely blocks both insulin and progesterone mediated maturation, and conversely, the dominant negative version of the kinase facilitated this process. In this present report, we focus on how X-PAK1 negatively regulates maturation. Our results demonstrate that X-PAK1-Cter prevents c-Mos synthesis and MAPK activation in progesterone-treated oocytes. Even if MAPK is activated by injection of recombinant c-Mos, no H1 kinase activity can be detected. All the enzymes involved in regulation of p34 cdc2 kinase are affected by X-PAK1-Cter expression in progesterone-treated oocytes. We show that in the presence of X-PAK1 kinase activity, pre-MPF complexes can efficiently form but activation of pre-MPF fails. Furthermore, in the presence of active MPF, X-PAK1 kinase activity prevents the amplification loop from being switched on. We have analyzed a possible link of transduction pathways used by PKA and X-PAK in controlling maturation and show that X-PAK1-Cter is still capable of inhibiting maturation induced by the regulatory domain of PKA. Finally, using cell-free Xenopus egg cycling extracts, which mimic the alternation of S and M phases of the first embryonic cycles, we can reproduce the inhibition of entry into mitosis with the catalytic domain of X-PAK1. Thus the involvement of Xenopus PAK in G 2 /M transition is not restricted to oocyte meiotic maturation and can be extended to the first embryonic cell cycles.

EXPERIMENTAL PROCEDURES
Preparation and Handling of Oocytes-Xenopus females were obtained from the CNRS breeding center located in Montpellier, France. Fully grown oocytes were prepared, free from follicle cells, by collagenase treatment in Ca 2ϩ -free OR 2 (12), then transferred in MMR (100 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 0, 1 mM EGTA, and 5 mM HEPES, pH 7.7) in which all experiments were performed. The usual volume microinjected was 40 nl; injections were performed near the equator of the oocytes. Progesterone was used with a final concentration of 1 mM; cycloheximide (100 g/ml) was added 1 h before further treatment (progesterone or microinjection).
Cycling Extracts-Cycling extracts were prepared essentially as described (25). Briefly, fresh eggs were dejellied in 2% cysteine (pH 7.8) washed and parthenogetically activated. Eggs were crushed at 12,000 ϫ g for 15 min, and the cytoplasmic layer was supplemented with an ATP regenerating system, cytocholasin B and protease inhibitors, before use. Sperm chromatin (500/l) and [ 35 S]methionine (1 mCi/ml) were added to the extract at time 0. Recombinant proteins (2 l, at 0.2 mg/ml per 50 l extract) were added after 10 min. Nuclei assembled from demembranated sperm nuclei and NEBD (26) were observed with a fluorescent microscope using Hoechst 33342 to stain chromosomes. Aliquots were withdrawn at indicated times and directly frozen in liquid nitrogen. All samples were thawed together and treated simultaneously, either immunoprecipated with a mix of anti-Xenopus cyclin B1 and B2 or with anti-Xenopus cyclin A; H1 kinase assays were performed on the different immunoprecipitates and analyzed by15% SDS-PAGE. Total extracts were also analyzed on Western blots with anti-p34 cdc2 antibodies directed against the C terminus.
Plasmids and Recombinant Proteins-Cloning of X-PAK1-Cter in pMal vector has been reported previously (1). The c-Mos pMal plasmid was kindly provided by A. Nebreda and T. Hunt (27).
Both recombinant c-Mos-MalBP and X-PAK1-CterMalBP proteins were produced as soluble proteins and purified on amylose-Sepharose matrix according to standard proctocols (28). The sea urchin glutathione S-transferase-cyclin B construct was kindly donated by Dr. J. Gautier (glutathione S-transferase is fused to sea urchin type B cyclin (from amino acids 13 to 409)); the recombinant protein was expressed in Escherichia coli and affinity purified as described previously (29).
Construction of pSP64T plasmid encoding histidine-tagged fulllength Xenopus Myt1 is described in Ref. 30. The pSP64 plasmid encoding full-length 5Ј-His-tagged Wee1 was constructed as follows. The pET construct encoding full-length Wee1 (kindly provided by W. G. Dunphy) was used (1 ng) in a PCR reaction with both BATG primer (5Ј-CCGTCGACGGCATATGAGGACGGCCATGTCATG) containing a SalI restriction site (underlined) and CTERWEE primer (5Ј-CCAAGCT-TGCAGTCCCAGCTCCTGAATCCACA-3Ј) containing a HindIII restriction site (underlined). The resulting PCR product was cloned in frame in pQE30 vector (Qiagen) linearized by SalI and HindIII. Full-length clone were screened and sequenced and then used in a PCR reaction (1 ng) with both pQE31 primer (5Ј-CAATTGTGAGCGGATAACAATT-3Ј) and CTERWEE primer described above. The resulting PCR product was EcoRI blunt inserted in pSP64 vector. In vitro translations of both pSP64 Myt1 and pSP64T Wee1 were performed using TNT coupled rabbit reticulocyte lysate kit (Promega) and [ 35 S]methionine-labeled according to the manufacturer's instructions.
Antibodies, Immunoprecipitation, and Western Blot Analysis-Polyclonal antibodies directed against Xenopus c-Mos and MAPK were obtained from Santa Cruz (Santa Cruz, CA). The c-Mos antibody was raised against peptide 314 -335 within the C-terminal domain of Xenopus c-Mos; the MAPK antibody was raised against peptide 305-327 of rat ERK1 and recognizes both ERK1 and ERK2 in all vertebrates tested. Anti-Plx1 polyclonal antibody raised against a peptide encoding the 17 C-terminal amino acids of Plx1, the Xenopus anti-Cdc25 antibody raised against the recombinant protein, and the X-PAK1-Cter antibodies against the C-terminal peptide of the protein were previously described (1,30). Anti-Xenopus cyclin B1, B2, and A polyclonal antibodies have been made in the laboratory by immunizing rabbits against recombinant fusion proteins.
Cyclin B and cyclin A immunoprecipitates were performed by incubation of the extracts with the indicated antibodies linked to protein A-Sepharose in modified RIPA buffer for 1 h at 4°C and washed three times in modified RIPA buffer and twice in kinase buffer (25 mM HEPES, pH 7.5, 25 mM MgCl 2 , 25 mM ␤-glycerophosphate, 2 mM dithiothreitol, and 0.1 mM orthovanadate).
Kinase Activities of MPF-Total Histone H1 kinase activities were assayed according to the protocol described by Labbé et al. (31). Immunoprecipitates were incubated for 15 min in 20 l of kinase buffer containing 1 l of histone H1 at 5 mg/ml, 1 l of 1 mM ATP, and 0.5 l of [␥-32 P]ATP, loaded on 15% SDS-PAGE, and visualized by autoradiography.
In gel MBP kinase assays were performed exactly as described by Shibuya et al. (32). Briefly, samples were run in a 15% polyacrylamide gels containing 0.5 mg/ml MBP. After denaturation and renaturation steps, gels were incubated in the presence of [␥-32 P]ATP. Gels were washed and dried; in gel MBP kinases were visualized by autoradiography.
For Plx-1 activities, immunoprecipitates were incubated for 15 min in a mix containing 1 mg/ml of dephosphorylated ␣-casein, 100 mM ATP, 2 mM MgCl 2 , and 50 mM Tris, pH 7.5. Immunocomplexes were boiled in Laemmli buffer and run on SDS-PAGE; the phosphorylation of casein substrate was visualized by autoradiography.
Cyclin B-p34 cdc2 kinase used for microinjection was prepared from starfish oocytes by affinity chromatography on p13 suc1-Sepharose and ion exchange chromatography on MonoQ and MonoS columns as described (31).

X-PAK1-Cter Prevents
Progesterone-induced MAPK Activation-We previously observed (1) that catalytically active X-PAK1-Cter inhibits germinal vesicle breakdown (GVBD) in progesterone-treated oocytes. We have demonstrated the physiological relevance of our results since a dominant negative (X-PAK1-CterK/R) mutant reproducibly facilitates hormoneinduced maturation. To investigate how X-PAK1-Cter exerts its inhibitory function, we first examined, as shown in Fig. 1 (A-D), the fate of the well known events triggered by progesterone treatment or c-Mos microinjection during oocyte maturation in the presence or in the absence of active X-PAK1-Cter. Possible effects of X-PAK1-Cter in the absence of maturation inducing factor were controlled in parallel. H1 kinase activity, reflecting MPF activation, is shown in Fig. 1A. When GVBD occurred in c-Mos microinjected or progesterone-treated oocytes, a strong activity was measured; by contrast, when X-PAK1-Cter was injected 1 h before c-Mos microinjection or progesterone treatment, no GVBD was observed and H1 kinase activation was completely abolished (Fig. 1A). Because p21activated kinases are involved in MAPK regulation, we analyzed the accumulation of the c-Mos proto-oncogene (  (Fig. 1B), MAPK was phosphorylated resulting in its mobility shift upon activating tyrosine phosphorylation on Western blot (Fig. 1D) and activated as observed against the MBP substrate (Fig. 1C).
In contrast X-PAK1-Cter efficiently blocked c-Mos accumulation in progesterone-treated oocytes, which further prevented any activation of MAPK ( Fig. 1, C and D). However, in c-Mos microinjected oocytes (Fig. 1B), X-PAK1-Cter failed to prevent MAPK activation (Fig. 1, C and D), and because MAPK was activated, endogenous c-Mos accumulation was observed (Fig.  1B), but GVBD did not occur. Thus X-PAK1-Cter is likely to target not only MAPK activation but also MPF activation in progesterone-treated oocytes and therefore mimics the effects of PKAc on oocyte maturation (13,14).
Catalytically Active X-PAK1 Does Not Interfere with Pre-MPF Complex Formation-Free p34 cdc2 is believed to be in large excess in stage 6 oocytes because only a limited amount of endogenous cyclin B is available (33), and microinjection of recombinant sea urchin cyclin B is sufficient to induce GVBD and MPF activation (34). We took advantage of these observations to test whether X-PAK1 is involved in regulation of cyclin B-p34 cdc2 complex formation. To do so, we examined the association of the exogenous cyclin B to the endogenous p34 cdc2 subunit in the presence of X-PAK1-Cter. X-PAK1-Cter was injected into oocytes 1 h before cyclin B injection, five oocyte aliquots were withdrawn at different times after injection as indicated, and the exogenous cyclin B was recovered from crushed eggs on glutatione-Sepharose beads. Bound proteins were tested in parallel, for associated H1 kinase activity and by Western blot for the presence of cyclin B associated p34 cdc2 kinase. In control oocytes injected with recombinant cyclin, H1 kinase was readily detected on glutathione-beads (Fig. 2, lower panel), but when oocytes were injected with recombinant cyclin B in the presence of X-PAK1-Cter, no H1 kinase activity (Fig. 2, lower panel, ϩX-PAK1-Cter) was obtained, even upon longer incubation (data not shown). An equivalent amount of Cdc2 subunit was bound to exogenous cyclin B in these oocytes (Fig.  2, upper panel, ϩX-PAK1-Cter) compared with controls ( Fig. 2, upper panel, ϪX-PAK1-Cter), but the Cdc2 subunit migrated as a slower band, indicating that residues Thr 14 and Tyr 15 are phosphorylated. Therefore, pre-MPF complex formation occurs but is not activated in the presence of X-PAK1-Cter.
Upon Progesterone Treatment, p34 cdc2 Kinase Regulatory Enzymes Are Affected by X-PAK1-Cter Activity-Activation of MAPK by c-Mos microinjection is not sufficient to prevent the inhibition of maturation by X-PAK1-Cter. Cyclin B and p34 cdc2 subunits can associate in the presence of X-PAK1-Cter, but these pre-MPF complexes fail to activate. Thus catalytically active X-PAK1 must target a component involved in the activation/inactivation status of p34 cdc2 upstream regulators. During the first embryonic cycle, both Myt1 and Wee1 kinases are inactivated by hyperphosphorylation at the G 2 /M transition (9,10), and at least one of the kinases responsible for Wee1 phosphorylation is p34 cdc2 itself (10), which participates in this way to the MPF amplification loop. Hyperphosphorylation is easily visualized in gel by a dramatic mobility shift of the kinases. Murakami and Vande Woude (35) recently showed that Wee1 is not yet present in stage 6 oocytes; Myt1 may thus be the only Cdc2 inhibitory kinase present in Xenopus oocytes. Nevertheless, microinjection of in vitro transcribed Xe-Wee1 RNAs in oocytes inhibits progesterone-induced maturation probably by changing the kinase/phosphatase ratio, thus impeding activation, due to the limited amounts of endogenous regulatory proteins available. As shown in Fig. 3A, we injected very small amounts (so as not to interfere with the physiological balance of proteins present in oocytes) of in vitro translated [ 35 S]methionine-labeled kinases to assess their phosphorylation state in progesterone-treated oocytes and in the presence of active X-PAK1-Cter. As expected, in control progesterone-treated oocytes, both Myt1 and Wee1 are hyperphosphorylated; but in contrast, no shift is observed when progesterone-treated oocytes are microinjected with X-PAK1-Cter, indicating that neither kinase is inactivated by progesterone treatment if X-PAK1-Cter is present.
Inactivation of p34 cdc2 inhibitory kinases is coincident with phosphorylation of the p34 cdc2 activating Cdc25 phosphatase. Cdc25 phosphatase is positively regulated by at least two kinases, Plx (36) and MPF itself. To test whether Cdc25 activation is also targetted in the X-PAK1 inhibition of maturation, we performed Western blot analysis using anti-Cdc25 antibodies of stage 6 oocytes and progesterone-treated oocytes in the presence or absence of X-PAK1-Cter. As previously shown, Cdc25 phosphatase was shifted in mature progesterone-treated oocytes (37,38), compared with control stage 6 oocytes. No shift was obtained when X-PAK1 had been injected in progesteronetreated oocytes, demonstrating that X-PAK1 prevents progesterone-induced activation of Cdc25c. Cdc25c is a target of Plx, a Xenopus Polo-like kinase (36), and our group has reported that Plx is a component of the MPF amplification loop required for full MPF activation at the onset of mitosis. Thus our next experiment was to verify the level of activation of Plx in progesterone-treated oocytes in the presence of catalytically active X-PAK1. As shown on Fig. 3C, no activity is detected on Plx immunoprecipitates in stage 6 oocytes using casein as a substrate, whereas as early as 3 h after progesterone treatment we detected a very strong Plx activity. In contrast when progesterone-treated oocytes had been injected with X-PAK1-Cter, no Plx activity is associated with immunoprecipitates. Thus X-PAK1 prevents Plx activation induced by progesterone treatment of the oocytes by an as yet unknown mechanism.
X-PAK1-Cter Prevents the Amplification Loop from Being Turned On-Our results indicate that X-PAK1 is likely to exert its control over the maturation process by negatively regulating an early event required to activate the small amount of MPF necessary to turn on the positive feedback loop. To further confirm the role of X-PAK1-Cter, we injected subthreshold concentration of active cyclin B-p34 cdc2 complexes purified from starfish 1 h after a preliminary injection of MalBP or X-PAK1-Cter (Fig. 4). In the absence of X-PAK1-Cter, 50% of the oocytes presented a white spot, characteristic of GVBD, as soon as 1 h after MPF injection, whereas in the presence of X-PAK1-Cter activity, no maturation spot was observed. We examined the effect of X-PAK1-Cter injection on the active cyclin B/p34 cdc2 kinase complex in terms of MAPK activation and H1 kinase activity and by Western blot analysis of XPAK1 and Cdc25. Western blot analysis of XPAK1 (Fig. 4, upper   FIG. 3. p34 cdc2 kinase regulatory enzymes are affected by X-PAK1-Cter activity during progesterone-induced meiotic maturation. A, stage 6 Xenopus oocytes were divided into three groups. In group 1, they were simultaneously microinjected with in vitro translated 35 S-labeled MYT1 (upper panel) or WEE1 (bottom panel) and X-PAK1-Cter 1 h before progesterone treatment (Pg ϩ, X-PAK1-Cter ϩ). In group 2 X-PAK1-Cter microinjection was omitted (Pg ϩ, X-PAK1-Cter Ϫ). The third group of control oocytes were only microinjected with in vitro translated 35 S-labeled MYT1 or WEE1 (Pg Ϫ, X-PAK1-Cter Ϫ). 5 h after progesterone treatment, oocytes were homogenized and monitored for changes in electrophoretic mobility of WEE1, MYT1 directly by autoradiography (A). B, oocytes were treated exactly as in A except microinjection of the in vitro translation product was omitted. The shift of endogenous Cdc25 was analyzed by Western blotting. C, Xenopus stage 6 oocytes were microinjected with X-PAK1-Cter protein 1 h before progesterone treatment (Pg ϩ, X-PAK1-Cter ϩ), and control progesterone-treated oocytes (Pg ϩ, X-PAK1-Cter Ϫ) and stage 6 oocytes (Pg Ϫ, X-PAK1-Cter Ϫ) were analyzed in parallel. At the indicated time after progesterone treatment, oocytes were homogenized, and Plx1 activities were measured on immunoprecipitates (3 oocyte equivalents/reaction) using casein as a substrate (upper panel). H1 kinase activities were measured in parallel on the same homogenates (lower panel). panel) demonstrates that the recombinant protein is stable throughout the experiment. The zero time (cyclinB-p34 cdc2 ) represents the subthreshold H1 kinase activity injected in oocytes and measured immediately after the injection (Fig. 4,  bottom panel). The kinase amplification loop is rapidly turned on because H1 kinase activity reached a maximum as early as 1.5 h after MPF injection. This occurred in parallel with both MAPK activation and a shift of Cdc25 as revealed by Western blot. In contrast, the injected H1 kinase was not amplified in the presence of X-PAK1-Cter; the MPF activity is therefore not sufficient by itself in the presence of X-PAK1-Cter to phosphorylate the first element of the positive amplification loop.

X-PAK1-Cter Blocks G 2 /M Progression in the First Mitotic
Cycle-To test whether X-PAK1-Cter inhibition of MPF activation is specific for the G 2 /M transition during Xenopus oocyte maturation or can be extended to the first mitotic cell cycles, we took advantage of Xenopus cycling egg extracts, which spontaneously oscillate between interphase and mitosis. Cycling extracts were prepared as described previously (25). Purified recombinant MalBP or X-PAK1-Cter fusion proteins were added 10 min after sperm chromatin addition (t ϭ 0) to the extract, [ 35 S]methionine was added in parallel to follow endogenous cyclin synthesis and degradation (Fig. 5). Aliquots were withdrawn at various times for analyses, and sperm chromatin swelling, nuclei formation, and nuclear envelope breakdown (NEBD) were monitored microscopically. NEBD was routinely observed in control extracts containing purified MalBP protein at around 60 -70 min after the start of the experiment, and by 80 -90 min mitosis had ended (data not shown). However, NEBD was never observed in extracts supplemented with X-PAK1-Cter. Instead, nuclei grew larger in size and remained stable for at least 2 h. We first analyzed cyclin A-Cdc2 activation on cyclin A immunoprecipitates using histone H1 as a substrate. Fig. 5A shows the kinetic of cyclin A-p34 cdc2 kinase activation. In control extracts with MalBP added, kinase activity increases during interphase to reach a maximum at 60 min, just before NEBD, and then disappears due to cyclin A degradation (Fig. 5A, MAlBP). In contrast, when X-PAK1-Cter kinase is present in the extract (Fig. 5A, X-PAK1-Cter), cyclin A-p34 cdc2 activity increases to finally stabilize at a very high level, but no drop in the kinase activity can be detected during the entire length of the experiment, indicating that cyclin A degradation did not occur in these extracts.
Identical samples were immunoprecipitated with a mix of antibodies directed against cyclin B1 and cyclin B2 to analyze cyclin B-p34 cdc2 activation (Fig. 5B). As expected, in control extracts (Fig. 5B, MalBP) cyclin B-p34 cdc2 activity increased to maximum at 70 min, activation of the kinase (H1) correlated with phosphorylation of the radiolabeled cyclin B1/B2 synthesized in the extract (CB 1 ), as visualized by a shift in gel. At 80 min, cyclin B-p34 cdc2 activity decreased dramatically, and the radiolabeled cyclin disappeared, indicating that cyclin degradation had occurred. Cyclin synthesis (CB 1 ) was observed again at the beginning of the next cycle (100 min). When catalytically active X-PAK1-Cter was added to the extract, no activation of cyclin B-Cdc2 could be detected (Fig. 5B, X-PAK1-Cter, H1). This result correlates with the absence of shift of [ 35 S]methionine-labeled newly synthesized cyclin B, as well as by stability of the labeled immunoprecipitated cyclin B band over the time of the experiment (Fig. 5B, X-PAK1-Cter, CB 1 ). We thus confirm the previous finding (30) that active cyclin A-p34 cdc2 complexes are not sufficient for the activation of cyclin B-p34 cdc2 kinase.
We further analyze, using Western blot with anti-Cdc2 antibodies, that during the course of incubation (between 20 and 90 min) of the cycling extract with X-PAK-Cter, p34 cdc2 is progressively shifted, reflecting phosphorylation and consequent inactivation on Tyr 15 and Thr 14 residues (Fig. 5C, X-PAK1-Cter). The phosphorylation on tyrosine was confirmed using anti-phospho-tyrosine antibodies (data not shown). At 90 min, in the presence of X-PAK-Cter, roughly half of the p34 cdc2 subunit, is shifted, and the lower band probably represents p34 cdc2 associated with cyclin A (Fig. 5C, X-PAK1-Cter). In contrast, in the MalBP supplemented extract, all p34 cdc2 subunit is present as a faster migrating form upon NEBD, corresponding to full activation of cyclin B-p34 cdc2 complexes (Fig.  5C, MalBP). Therefore, the inhibitory effect of X-PAK1-Cter on G 2 /M progression is not restricted to meiosis.
X-PAK1 Prevents Oocyte Maturation Induced by PKA Regulatory Subunit-Protein kinase A is a tetramer composed of two regulatory and two catalytic subunits (39), and cAMP binding to the regulatory subunit of protein kinase A induces its release from the catalytic subunit and subsequent activation of the kinase. Microinjection into oocytes of an excess of PKA regulatory subunit inhibits endogenous PKA activity and induces oocyte maturation (40). Because X-PAK1-Cter targets the same components to bring about the block of the G 2 /M transition as those previously reported for the catalytic subunit of protein kinase A, we next wanted to study a possible link between protein kinase A and X-PAK transduction pathways in the control of the G 2 /M transition. Thus, we injected oocytes first with X-PAK1-Cter and then with PKA/R. The amount of PKA/R we injected is itself sufficient to bind all endogenous PKA/C and inactivate PKA, because we observed a maturation spot on these oocytes and confirmed that indeed maturation occurred by H1 kinase activation and c-Mos accumulation (Fig.  6). However, in the presence of X-PAK1-Cter, PKA/R was no longer able to induce c-Mos synthesis because no accumulation of the proto-oncogene c-Mos was detected by Western blot (Fig.  6) and no activation of pre-MPF occurred, because H1 kinase Samples were withdrawn at indicated times and observed microscopically for nuclei formation and NEBD. Later on, aliquots of the samples were immunoprecipitated with anti-cyclin A antibodies, and the immunoprecipitates were tested for H1 kinase activity (A). Other aliquots of the same samples were analyzed for H1 kinase activity on anti-cyclin B1 and B2 immunoprecipitates (B, H1). The upper part of the same gel was exposed longer to visualize endogenous cyclin B synthesis, shift by phosphorylation, and degradation. Aliquots of the same samples taken between 20 and 90 min during the time course of the experiment were directly loaded on SDS-PAGE, and the total p34 cdc2 subunits were analyzed by Western blot. In this particular experiment, NEBD was detected at 60 min in MalBP supplemented extract, and NEBD never occurred in X-PAK1-Cter supplemented extracts. activity was low and the oocytes remained morphologically indistinguishable from stage 6 oocytes.
Considering the above result, we addressed the question whether X-PAK could be an effector downstream of PKA on the same transduction pathway. If this is the case we should measure the increased activity of the endogenous X-PAK1 when PKA is activated. However, because the three cloned Xenopus PAKs present a high level of homology in their kinase domain (1), 2 we tested a possible enhancement of kinase activity by active PKA for all three X-PAKs. We thus isobutylmethylxanthine-treated or mock treated oocytes for various lengths of time (1-5 h) and immunoprecipitated the endogenous PAKs with three different specific antibodies. Kinase assays using histone 2B as a substrate were performed on the immunoprecipitates. No significant changes in the three Xenopus PAKs kinase activity could be measured between the oocytes presenting an increased PKA activity and the control oocytes (data not shown).
We tested yet another potential link between the two pathways, a potential regulatory phosphorylation of PKA/R and/or PKA/C by X-PAK1-Cter. X-PAK1-Cter was microinjected in oocytes for either 1 or 2 h, and the active recombinant kinase recovered on amylose-beads was then tested for its ability to use PKA/C and PKA/R subunits as substrates in in vitro kinase assays. No phosphorylation of PKA/C nor PKA/R was observed under these conditions (data not shown).

DISCUSSION
Oocyte maturation is a highly regulated process, and a complete understanding of the many intertwining regulations involved is far from being a reality. GVBD is always associated with high cyclin B-p34 cdc2 activity, which results from the activation of stockpiled complexes, occurring by a well described amplification loop. Although experimental conditions have been described in which GVBD can occur in the absence of MAPK activity (34), it is widely accepted that the MAPK cascade participates in normal maturation in the activation of MPF. Understanding the event required for activation of the first cyclin B-p34 cdc2 complex, which in turn will start the amplification loop, is crucial to explain the maturation process, and in this regard identification of new regulatory proteins will certainly help to reconstitute the sequence of activation of the different pathways.
In this report we investigate how constitutive Xenopus p21activated kinase negatively controls the oocyte maturation process. Our results indicate that X-PAK1 mimics the effects previously reported for the catalytic subunit of protein kinase A by preventing c-Mos accumulation as well as MAPK and MPF activation (13,14). We show, in a context in which MAPK activation is activated by injection of c-Mos, that accumulation of the endogenous proto-oncogene does occur, in the presence of X-PAK1. Thus X-PAK1 does not interfere with the amplification loop in which active MAPK controls c-Mos synthesis (42) but rather prevents the initial MAPK activation. Even on c-Mos microinjection, no H1 kinase activity is detected, indicating that X-PAK1 also affects MPF activation. We used recombinant cyclin B, and active MPF, to demonstrate that formation of new pre-MPF complexes is not affected but that they are not activated in the presence of active X-PAK1.
Indeed, we show that the regulatory kinases and Cdc25 phosphatase identified in MPF amplification during the maturation process remain in their initial state. A well accepted concept is that activation of the first MPF molecules is the result of activation and/or inactivation of Cdc25 and Myt1, respectively, by an upstream regulating element different from MPF itself. Indeed, in SDS-PAGE, Cdc25 displays a complex phosphorylation pattern and can also be phosphorylated and activated by kinases other than MPF itself, Plx-1 (36), and possibly Raf-1 (43).
Raf-1 may be a Cdc25-activating kinase (43), and it also acts as a MAPK kinase kinase (44) whose involvement in regulation of maturation is subject to controversy. Using a dominant negative mutant approach it has been reported (45) that Raf-1 activation is required for progesterone-induced maturation, but this result was challenged (46). It was recently shown that autophosphorylation of Raf-1 on serine 621 (47) is a way to down-regulate its activity, and this same site can also be phosphorylated by PKA. We discussed in a previous paper a possible transient activation of Raf-1 in Xenopus oocytes, upon downregulation of PKA, during normal maturation induced by progesterone (14). Because Raf-1 mobility shift observed during maturation by progesterone is usually equated to the kinase activation, we verified that Raf-1 is indeed prevented from shifting upon progesterone treatment of oocytes injected with X-PAK1-Cter (data not shown). As X-PAK1 belongs to the well known family of PAK proteins that were first isolated as upstream regulators of MAPK cascades, involvement of X-PAK in Raf-1 kinase regulation is an interesting hypothesis.
We show that X-PAK1-Cter negatively targets the same events in inhibition of maturation as PKA/C and moreover that X-PAK1-Cter prevents maturation induced by the regulatory subunit of protein kinase A, but we have been unable to demonstrate any direct activation of endogenous X-PAK1 or two other closely related family members X-PAK2 and X-PAK3 by PKA/C. Both pathways could also be linked in a different manner; for example, X-PAK could interfere with the ability of PKA to form a tetramer (for example by phosphorylation of PKA/R) and consequently inhibit its inactivation. Indeed it has previously been demonstrated that PKA/RII can be phosphorylated in vitro by different kinases, and recently Keryer et al. (41) reported that MPF can also phosphorylate it. We verified that neither PKA/R nor PKA/C are substrates of X-PAK1-Cter. At the moment we cannot rule out other possible cross-links in the pathways used by the two kinases; however, another possible explanation of our results would be that both pathways are parallel and that active X-PAK and PKA target an identical substrate that allows the MPF amplification loop to be turned on.
Both PKA and X-PAK kinases are also capable of inhibiting the G 2 /M transition during the first mitotic cycle, at which FIG. 6. X-PAK1-Cter prevents meiotic maturation induced by the microinjection of the PKA regulatory subunit. Xenopus stage VI oocytes were microinjected or not with X-PAK1-Cter protein and injected again 1 h later with the purified R1 regulatory subunit of PKA. 5, 7, and 9 h after the second injection, oocytes were homogenized, and total cell extracts were analyzed for H1 kinase activities (upper panel, H1) or c-Mos accumulation monitored by Western blotting (bottom panel, c-mos). The appearance of a white spot, indicating that maturation had occurred, was obvious in R1 injected oocytes as soon as 4 h after the injection but was never observed when X-PAK1-Cter was injected before PKA/R. point cyclin B-p34 cdc2 is still regulated by phosphorylation on Tyr 15 , but control by MAPK activity is absent. We show that X-PAK1-Cter does not prevent cyclin A-p34 cdc2 activation in cellfree egg cycling extracts, and in fact the kinase reaches an abnormally high level of activity compared with a control cycling extract supplemented with MalBP fusion protein. Nevertheless, under these conditions, high cyclin A-p34 cdc2 activity does not allow cyclin B-p34 cdc2 activation, and as a consequence the cyclin degradation pathway is not turned on. We show that cyclin B-p34 cdc2 complexes are indeed formed, but no H1 activity is associated with them, and the p34 cdc2 subunit displays a slow electrophoretic mobility indicative of phosphorylation on Tyr 15 . We conclude that X-PAK1-Cter activity is very specific because it does not interfere with the activity of cyclin A-p34 cdc2 complexes, which are not subject to a regulation by phosphorylation on Tyr 15 (30,48), but it does interfere specifically with the activation of cyclin B-p34 cdc2 complexes by preventing dephosphorylation of Tyr 15 . In 1994 Grieco et al. (49) showed that PKA activity oscillates during the first embryonic cycles in Xenopus extracts, being high at interphase and falling at the onset of mitosis. They demonstrated that sustained PKA activity inhibits cycle progression in a dose-dependent manner and that extracts eventually arrest in interphase with a high level of inactive pre-MPF complexes. In this context, X-PAK1-1Cter and PKA probably act independently of a control over MAPK activity, because MAPK is inactive in the cycling extract, and so this cannot be compared with the G 2 /M transition block described previously (50), in which the proto-oncogene c-Mos is added to cycling extracts. The work of Grieco et al. (49) indicates that PKA could stimulate an okadaic acid-sensitive serine/threonine phosphatase pathway that dephosphorylates Cdc25.
Plx kinase is another Cdc25 positive regulatory kinase (36) recently shown (30) to be required for the G 2 /M transition during the first mitotic cell cycle in Xenopus eggs. Cyclin B-p34 cdc2 enhances Plx activation, which triggers Cdc25 and MPF activation in a positive feedback loop. Adding blocking anti-Plx antibodies to an egg cycling extract suppressed entry into mitosis, although a strong cyclin A-p34 cdc2 activity is detected, as we observed ourselves in the X-PAK1-Cter block in cycling extracts. Moreover, pre-MPF complexes also remain inactive and phosphorylated on Tyr 15 , if Plx is blocked with these antibodies. We did not investigate the status of Plx in cycling extracts incubated with X-PAK1-Cter, but we show that during the maturation process Plx activation is prevented by active Xenopus p21-activated kinase.
In summary, our data demonstrate that PKA and X-PAK pathways target the same components in their inhibition of G 2 /M progression during meiosis as well as in the first mitotic cycle in Xenopus and that their transduction pathways are likely to be connected at the point where the MPF amplification loop is turned on. We are currently in the process of isolating physiological substrates of X-PAK active kinase in Xenopus extracts to get new insights into factors regulating the G 2 /M transition.