Intra-M Phase-promoting Factor Phosphorylation of Cyclin B at the Prophase/Metaphase Transition*

Activation of Cdc2-cyclin B (or M phase-promoting factor (MPF)) at the prophase/metaphase transition proceeds in two steps: dephosphorylation of Cdc2 and phosphorylation of cyclin B. We here investigated the regulation of cyclin B phosphorylation using the starfish oocyte model. Cyclin B phosphorylation is not required for Cdc2 kinase activity; both the prophase complex dephosphorylated on Cdc2 with Cdc25 and the metaphase complex dephosphorylated on cyclin B with protein phosphatase 2A display high kinase activities. An in vitro assay of cyclin B kinase activity closely mimics in vivo phosphorylation as shown by phosphopeptide maps of in vivo and in vitro phosphorylated cyclin B. We demonstrate that Cdc2 itself is the cyclin B kinase; cyclin B phosphorylation requires Cdc2 activity both in vivo(sensitivity to vitamin K3, a Cdc25 inhibitor) and in vitro (copurification with Cdc2-cyclin B, requirement of Cdc2 dephosphorylation, and sensitivity to chemical inhibitors of cyclin-dependent kinases). Furthermore, cyclin B phosphorylation occurs as an intra-M phase-promoting factor reaction as shown by the following: 1) active Cdc2 is unable to phosphorylate cyclin B associated to phosphorylated Cdc2, and 2) cyclin B phosphorylation is insensitive to enzyme/substrate dilution. We conclude that, at the prophase/metaphase transition, cyclin B is mostly phosphorylated by its own associated Cdc2 subunit.

Simultaneously with Cdc2 dephosphorylation, cyclin B becomes phosphorylated as described in a variety of models such as yeast (6), starfish oocytes (13), sea urchin eggs (11), goldfish oocytes (29), Xenopus oocytes (30), and human cells (31). The residues phosphorylated in cyclin B 1 have been identified in Xenopus: Ser-2, Ser-94, Ser-96, Ser-101, and Ser-113 (32,33). Mutational studies (29,32,33) have suggested that cyclin B phosphorylation is required neither for activity of the Cdc2 kinase, for Cdc2 binding, nor for cyclin B destruction in anaphase. More recently, Li et al. (34) showed that if the residues were mutated to nonphosphorylatable residues, the Cdc2-cyclin B complex does not migrate from the cytoplasm to the nucleus and therefore loses its MPF activity. The kinase(s) responsible for cyclin B phosphorylations has not been identified so far. Autophosphorylation of the cyclin B by the Cdc2 kinase itself has been suggested in sea urchin eggs (11) and in Xenopus oocytes (35). Furthermore, mitogen-activated protein kinase is able to phosphorylate Ser-94 or Ser-96 of Xenopus cyclin B 1 (32). Cyclin B 2 has also been found to be a substrate for the Xenopus c-mos proto-oncogene product in vitro (36), but not in vivo (37). Recently, a kinase (Cyk) phosphorylating only one cyclin B 2 residue (Ser-53) in vitro has been partially characterized in Xenopus (38).
Taking advantage of the high and natural synchrony of starfish oocytes, we have investigated the regulation of cyclin B phosphorylation. Cyclin B phosphorylation is easily detected on an immunoblot following SDS-PAGE; compared with the unphosphorylated isoform, phosphorylated cyclin B displays a retarded migration ("shift"). We first clearly show in vitro that cyclin B phosphorylation is not required for Cdc2 kinase activity. Second, in vivo as well as in vitro, the cyclin B phosphorylation shift requires Cdc2 activation (dephosphorylation of both Thr-14 and Tyr-15 inhibitory residues). Then in vitro experiments using chemical inhibitors of CDKs, like olomoucine (39), roscovitine (40), and purvalanol (41), and cyclin B phosphopeptide mapping demonstrate that Cdc2 is the kinase responsible for the cyclin B phosphorylation shift observed at metaphase in starfish oocytes. Furthermore, two experiments clearly show that Cdc2 phosphorylates its own associated cyclin B subunit, rather than the cyclin B subunit from another complex. Cyclin B phosphorylation thus occurs in an intra-MPF complex fashion.

Preparation of Starfish Oocytes
Starfish Oocyte Maturation-The starfish Marthasterias glacialis were collected in Northern Brittany and kept under running sea water until use. The gonads were dissected out and gently torn open in ice-cold CaFASW. Oocytes were then filtered through cheesecloth and washed four times in CaFASW to remove the 1-MeAde-producing follicle cells. They were resuspended in the same medium as a 10% (v/v) suspension. Oocyte maturation was triggered by the addition of 1-MeAde to a final concentration of 1 M. During maturation time course experiments, 1-ml aliquots of the oocyte suspension were withdrawn at regular intervals after hormonal stimulation and centrifuged in microtubes, and the oocyte pellets were frozen in liquid nitrogen.
Vitamin K 3 Treatment-Vitamin K 3 is a powerful inhibitor of the Cdc25 phosphatase (24,43). It inhibits hormone-induced oocyte maturation (24). Oocytes were treated with vitamin K 3 (0 -250 M final concentration) for 15 min prior to the 1-MeAde addition. After a 30-min incubation, aliquots were centrifuged, and oocyte pellets were frozen in liquid nitrogen.
Preparation of Prophase and Metaphase Oocytes-1-ml aliquots of an oocyte suspension before (prophase oocytes) or 20 min after the 1-MeAde addition (metaphase oocytes) were rapidly centrifuged, the supernatant was removed, and the oocyte pellets were frozen in liquid nitrogen.

Preparation of Xenopus Oocytes
Adult Xenopus laevis females were purchased from CRBM (Centre de Recherche en Biologie Moléculaire, Montpellier, France). Frogs were primed with 500 international units of human chorionic gonadotropin and stored overnight in laying tanks containing HSB. Mature oviposited eggs were dejellied in 2% cysteine in HSB medium during 10 min under agitation. They were washed three times in HSB medium and frozen immediately at Ϫ80°C until use.

Purification of Cdc2-cyclin B on p9 CKShs1 -Sepharose Beads
The Cdc2-cyclin B complex was purified from starfish oocytes by affinity chromatography on p9 CKShs1 -Sepharose beads, prepared as described in Azzi et al. (44). 400 l of homogenization buffer were added per 100 l of prophase or metaphase oocyte pellets. After sonication, extracts were centrifuged at 14,000 ϫ g for 10 min at 4°C. The supernatant was then incubated at 4°C for 30 min and under constant rotation, with 10 l of p9 CKShs1 -Sepharose beads in the presence of 400 l of bead buffer. After removal of the supernatant, the beads were washed three times with ice-cold bead buffer. The complex was eluted or not eluted by free p9 CKShs1 (2 mg/ml) for 30 min under constant rotation and used for further analysis. The same protocol was used for the preparation of Xenopus MPF.

Microsequencing of Starfish M. glacialis Cyclin B
Starfish cyclin B from prophase and metaphase oocytes were prepurified first on Green A-agarose beads (loading in 10-fold diluted homogenization buffer, extensive washing with the same buffer, and elution with 0.2 M NaCl in the same buffer). This procedure leads to rapid concentration and approximately 20-fold purification of Cdc2-cyclin B. 2 The concentrated kinase was then affinity-purified on p9 CKShs1 -Sepharose beads and analyzed by SDS-PAGE and Amido Black staining. The prophase and metaphase cyclin B bands were digested and partially microsequenced at the Institut Pasteur (Paris).

Preparation and Purification of GST-Pyp3 and GST-Cdc25A Fusion Proteins
Bacterial growth and purification of the fusion proteins were as described by Borgne and Meijer (24).

In Vitro Dephosphorylation of Cdc2 by Purified Phosphatases
p9 CKShs1 -Sepharose beads, loaded with prophase oocyte extracts, were prepared as described above. Following the bead buffer step, the beads were washed three times with Tris buffer A prior to incubation for 30 min at 30°C with 100 l of recombinant phosphatases and/or 2 l of PP2A 1 . The dephosphorylation reaction was stopped by the addition of 1 ml of bead buffer. The beads were washed three times with bead buffer before cyclin B phosphorylation assays.

In Vitro Cyclin B Phosphorylation Assay
The prophase Cdc2-cyclin B complex purified on p9 CKShs1 -Sepharose beads and treated or not treated with different phosphatases was incubated for 10 min at 30°C with 15 M ATP in a final volume of 30 l. The phosphorylation reaction was stopped by the addition of 1 ml of bead buffer. The beads were washed three times with bead buffer before the addition of 50 l of 2ϫ Laemmli sample buffer and analysis by SDS-PAGE and Western blotting.

In Vitro Cyclin B Dephosphorylation
The metaphase Cdc2-cyclin B complex was treated with 100 l of GST-Pyp3 or 10 l of PP2A 1 , after washes with Tris buffer A, for 30 min at 30°C. The beads were then washed with bead buffer and recovered with 50 l of 2ϫ Laemmli sample buffer prior to analysis by SDS-PAGE and Western blotting.

Cyclin B Phosphopeptide Mapping
For in vivo cyclin B labeling, 30 ml of prophase starfish oocytes (a 10% suspension in CaFASW) were incubated at room temperature with [ 32 P]phosphate (40 mCi/ml) for 3 h. 1-MeAde was then added for a 20-min incubation. After washes of the oocytes with CaFASW, an extract was prepared for Cdc2-cyclin B purification on p9 CKShs1 -beads. In vitro labeled cyclin B was obtained in the conditions of the phosphorylation assay (see above) in the presence of [␥-32 P]ATP. In vivo and in vitro 32 P-labeled cyclin B were resolved by SDS-PAGE. Labeled cyclin B was detected by autoradiography, and the bands were excised from the gel. Cyclin B was subjected to complete digestion with trypsin or ther-

FIG. 2. In vivo dephosphorylation of Cdc2/phosphorylation of cyclin B during starfish oocyte maturation.
Prophase oocytes were treated with 1 M 1-MeAde. At regular intervals, aliquots of the oocyte suspension were withdrawn and frozen in liquid nitrogen. The Cdc2cyclin B kinase was purified from oocyte extracts by affinity chromatography on p9 CKShs1 -Sepharose beads and was analyzed by SDS-PAGE and Western blotting with anti-cyclin B (A) and anti-PSTAIRE (B) antibodies. Nuclear envelope breakdown occurred 20 min after the 1-MeAde addition, in this experiment.
FIG. 3. Histone H1 kinase activity of different phosphorylation states of the Cdc2-cyclin B complex. First, the Cdc2-cyclin B kinase was purified from prophase (P) or metaphase oocyte extracts (M) by affinity chromatography on p9 CKShs1 -Sepharose beads and treated or not treated with different phosphatases. The complex was next assayed for its histone H1 kinase activity. [ 32 P]phosphate incorporation in histone H1 was measured by direct counting (C) or by autoradiography (F). The p9 CKShs1 -bound proteins were then analyzed by SDS-PAGE and Western blotting with anti-cyclin B (A and D) and anti-PSTAIRE (B and E) antibodies. molysin as described by Walaas et al. (45). The phosphopeptides were separated by electrophoresis at pH 3.5 (first dimension) and chromatography on silica plates (second dimension).

Histone H1 Kinase Activity Assay
The kinase activity of Cdc2-cyclin B was measured after its purification on p9 CKShs1 -Sepharose beads and various treatments. Assays were performed by incubation of 10 l of packed p9 CKShs1 -Sepharose beads for 5 min at 30°C with 1 mg/ml histone H1 and 15 M [␥-32 P]ATP in a final volume of 30 l. Assays were terminated by transferring the tubes into ice. After a brief centrifugation, 25 l of supernatant were spotted on 2.5 ϫ 3-cm pieces of Whatman p81 phosphocellulose paper. Filters were washed five times in 1% phosphoric acid, dried, and transferred in plastic scintillation vials with 1 ml of ACS (Amersham Pharmacia Biotech) scintillation fluid. [ 32 P]Phosphate incorporation in the histone H1 substrate was measured in a Packard counter. 50 l of 2ϫ Laemmli sample buffer were added to the remaining beads and supernatant prior to SDS-PAGE and analysis by autoradiography.

Inhibition of Cyclin B Phosphorylation and of Histone H1 Kinase Activity Assay
We have used three chemical inhibitors of CDKs, olomoucine (39), roscovitine (40,46), and purvalanol (41) which are purine derivatives and act as competitive inhibitors for ATP binding. Inhibition experiments were performed on prophase Cdc2-cyclin B complex, previously dephosphorylated by GST-Cdc25A, and eluted from p9 CKShs1 -Sepharose beads by free p9 CKShs1 (2 mg/ml). Histone H1 kinase activity was assayed by incubation of 2.5 l of soluble Cdc2-cyclin B for 10 min at 30°C with 15 M [␥-32 P]ATP and various concentrations of inhibitors, in the presence or absence of 1 mg/ml histone H1, in a final volume of 30 l. The histone H1 kinase activity of Cdk2-cyclin E was assayed using 0.5 l of purified kinase incubated for 5 min at 30°C with 1 mg/ml histone H1, 15 M [␥-32 P]ATP, and various concentrations (0 -100 nM) of p27 Kip1 (a GST fusion protein provided by Dr. B. Ducommun) in a final volume of 30 l. The histone H1 kinase activities were measured in a Packard counter. The cyclin B phosphorylation assay was monitored as described above in the presence of ATP and increasing concentrations of inhibitors and analyzed by autoradiography or Western blotting.

Electrophoresis and Western Blotting
Proteins bound to p9 CKShs1 -Sepharose beads were recovered with 2ϫ Laemmli sample buffer. Samples were run in 10% SDS-polyacrylamide gels. For detection of 32 P-labeled proteins, gels were stained with Coomassie Blue and exposed overnight to Hyperfilm MP. For Western blotting, proteins were transferred from the gel to a 0.1-m nitrocellulose sheet (Schleicher and Schuell) in a milliblot-graphite electroblotter system (Millipore Corp.) for 30 min at 1.5 mA/cm 2 in transfer buffers. Subsequently, the filter was blocked with 5% low fat milk in TBST for 1 h. The filter was then washed with TBST and incubated for 1 h with the first antibodies (anti-PSTAIRE (1:2000) or anti-cyclin B (1:1000)). After four washes (1 ϫ 20 min, 3 ϫ 5 min) with TBST, the nitrocellulose sheet was treated for 1 h with horseradish peroxidase-coupled secondary antibodies diluted in TBST (1:1000). The filter was then washed five times (1 time for 20 min, 4 times for 5 min) with TBST and analyzed by enhanced chemiluminescence with ECL detection reagents and hyperfilm MP.

In Vivo Cyclin B Phosphorylation at the Prophase/Metaphase Transition in Starfish
Oocytes-The addition of 1-methyladenine to prophase-arrested starfish oocytes triggers Cdc2-cyclin B kinase activation, nuclear envelope breakdown, and entry into meiotic division (reviewed in Ref. 47). The complex is inactive in prophase, Cdc2 being phosphorylated on Thr-14 and Tyr-15 and cyclin B being unphosphorylated. When cells enter in metaphase, Cdc2 is sequentially dephosphorylated first on Thr-14 and then on Tyr-15 (24), while cyclin B becomes phosphorylated.
Cyclin B phosphorylation results in a change in mobility (shift) on SDS-PAGE as detected by silver staining (Fig. 1) and with monoclonal anti-cyclin B antibodies (Fig. 2). The two bands corresponding to prophase and metaphase cyclin B were partially microsequenced and compared with the M. glacialis cyclin B known sequence (10). This comparison shows that our prophase and shifted metaphase bands correspond to bona fide M. glacialis cyclin B (Fig. 1).
The changes in electrophoretic mobility of the two MPF subunits can be monitored during the oocyte maturation time course following stimulation by 1-MeAde (Fig. 2). The phosphorylation state of the complex was analyzed after purification of Cdc2-cyclin B on p9 CKShs1 -Sepharose beads, SDS-PAGE, and Western blotting with anti-PSTAIRE antibodies for Cdc2 and with anti-cyclin B antibodies for cyclin B. At metaphase entry, cyclin B phosphorylation occurs simultaneously with Cdc2 dephosphorylation (Fig. 2).
To ensure that the observed cyclin B shift is due to a phosphorylation reaction, we have treated the purified metaphase cyclin B with different phosphatases (Fig. 3, A, B, and C). Treatment with PP2A 1 leads to cyclin B dephosphorylation, while treatment with the tyrosine phosphatase GST-Pyp3 does not (Fig. 3A). After treatment with PP2A 1 , cyclin B presents the same electrophoretic mobility as the prophase cyclin B.
Cyclin B Phosphorylation Is Not Required for the Catalytic Activity of Cdc2-Cyclin B-We next investigated the importance of cyclin B phosphorylation for the catalytic activity of the Cdc2-cyclin B complex. The dephosphorylated Cdc2-unphosphorylated cyclin B complex can be obtained in vitro by dephosphorylation of metaphase cyclin B by PP2A 1 (Fig. 3, A,  B, and C) or by dephosphorylation of prophase Cdc2 by recom- binant GST-Cdc25A (Fig. 3, D, E, and F). This complex displays the same histone H1 kinase activity as the metaphase complex (dephosphorylated Cdc2-phosphorylated cyclin B) (Fig. 3, C and F). This directly demonstrates that cyclin B phosphorylation is not necessary for Cdc2 kinase activity.
Cyclin B Phosphorylation Requires Cdc2 Activation-Prior to the 1-methyladenine addition, oocytes were treated with increasing concentrations of vitamin K 3 , an inhibitor of Cdc25 phosphatase (24). A dose-dependent inhibition of germinal vesicle breakdown (GVBD; Fig. 4A) and of Cdc2 dephosphorylation was observed (Fig. 4C). A dose-dependent inhibition of cyclin B phosphorylation was also observed (Fig. 4B). These results suggest that Cdc2 activation (by dephosphorylation of Thr-14 and Tyr-15) is necessary for cyclin B phosphorylation in vivo.
We next set up an in vitro cyclin B kinase assay (Fig. 5, A and  B). Prophase oocyte Cdc2-cyclin B was first purified and immobilized on p9 CKShs1 -Sepharose beads. The Cdc2 subunit was then dephosphorylated with GST-Cdc25A phosphatase. Cyclin B phosphorylation was initiated by the addition of 15 M ATP and incubation at 30°C. Incubation with ATP without prior Cdc25A phosphatase treatment did not lead to cyclin B phosphorylation (Fig. 5, A and B). Pretreatment with Cdc25A and incubation without ATP did not lead to cyclin B phosphorylation either. When the complex was dephosphorylated by Cdc25A and then incubated with ATP, cyclin B phosphorylation occurred. The phosphorylated cyclin B obtained in vitro could be dephosphorylated by phosphatase 2A (data not shown).
The Cdc25 phosphatase effect is mimicked by the successive treatments of Cdc2-cyclin B with the Ser/Thr phosphatase 2A and the tyrosine phosphatase Pyp3, which together completely dephosphorylate Cdc2 (Fig. 5, C and D). Treatment with only one of these enzymes does not allow cyclin B phosphorylation to occur upon exposure to ATP. These results show that both Thr-14 and Tyr-15 residues of Cdc2 should be dephosphorylated, allowing full Cdc2 activation (24) for cyclin B phosphorylation. Therefore, Cdc2 activation is also necessary for cyclin B phosphorylation in vitro.
Cyclin B Phosphorylation Is Sensitive to Chemical Inhibitors of CDKs-To further analyze the role of Cdc2 in cyclin B phosphorylation, we have performed (Fig. 6) the cyclin B shift assay in the presence of roscovitine, a selective inhibitor of CDKs (40,46). The soluble Cdc2-cyclin B complex from prophase oocytes was previously dephosphorylated by Cdc25A and incubated with ATP in the presence or absence of 10 M of roscovitine, a concentration that inhibits the histone H1 kinase activity of Cdc2 (Fig. 6A). The presence of roscovitine inhibits cyclin B phosphorylation, as shown by anti-cyclin B immunoblotting (Fig. 6B). These results further suggest that cyclin B phosphorylation is dependent on Cdc2 kinase activity.
We next tested other specific inhibitors of CDKs on the in vitro phosphorylation of cyclin B. Soluble Cdc2-cyclin B was prepared as in Fig. 6, and both Cdc2-cyclin B kinase activity  6. Sensitivity of the in vitro cyclin B shift to roscovitine. The Cdc2-cyclin B kinase was purified from prophase (P) oocyte extracts by affinity chromatography on p9 CKShs1 -Sepharose beads, treated with GST-Cdc25A, and eluted by free p9. The histone H1 kinase activity assay was performed in the presence of 15 M ATP, 1 mg/ml histone H1 with or without 10 M roscovitine. The complex was assayed for its histone H1 kinase activity by direct counting (A). The cyclin B phosphorylation was performed in the presence of 15 M ATP with or without 10 M roscovitine. The complex was analyzed by SDS-PAGE and Western blotting with anti-cyclin B antibodies (B) (control Cdc2cyclin B, metaphase (M)). and cyclin B phosphorylation were simultaneously assayed in the presence of increasing concentrations of purvalanol, roscovitine, or olomoucine (Fig. 7). Histone H1 kinase activities were measured by direct counting (Fig. 7A). Cyclin B phosphorylation was detected by autoradiography (Fig. 7B). The same dose-dependent inhibitions of histone H1 kinase activity and of cyclin B phosphorylation were observed with the three inhibitors, purvalanol being more efficient than roscovitine, itself being more efficient than olomoucine. These results show that both activities are inhibited in a very similar fashion by the three inhibitors of CDKs. Most if not all of the histone H1 kinase activity recovered on p9 CKShs1 -Sepharose beads can be attributed to Cdc2-cyclin B (24,44). The p9 CKShs1 -Sepharose beads also present affinity for Cdk2. We therefore tested the effect of p27 kip1 , a powerful inhibitor of Cdk2 but not of Cdc2, on cyclin B phosphorylation (Fig. 8). The Cdk2 inhibitor was unable to inhibit in vitro cyclin B phosphorylation, excluding a possible involvement of Cdk2 in cyclin B phosphorylation. The histone H1 and cyclin B kinase activities measured in our assays (Figs. 6 and 7) can thus be attributed to the same kinase, Cdc2-cyclin B, i.e. MPF.
We next compared in vivo and in vitro cyclin B phosphorylation by phosphopeptide mapping. Cyclin B was labeled in vivo, by incubating maturing starfish oocytes with [ 32 P]phosphate, or in vitro under the conditions of the shift assay in the presence of [␥-32 P]ATP. Both in vivo and in vitro phosphorylated cyclin B were purified on p9 CKShs1 -Sepharose beads and on SDS-PAGE. The labeled cyclins were then digested by trypsin (Fig. 9, A, B, and C) or thermolysin (Fig. 9, D-F). The twodimensional phosphopeptide maps of in vivo and in vitro phosphorylated cyclin B were similar, but not entirely identical, in each experimental procedure (Fig. 9, compare panels A and B  and panels D and E). This shows first that most of the sites phosphorylated in our shift assay and most sites phosphorylated during maturation are the same. Second, these results are consistent with the fact that Cdc2, which phosphorylates cyclin FIG. 7. Inhibition of cyclin B phosphorylation and histone H1 kinase activity by chemical inhibitors of Cdc2. The Cdc2-cyclin B kinase was purified from prophase oocyte extracts (P) by affinity chromatography on p9 CKShs1 -Sepharose beads, treated with GST-Cdc25A, and eluted by free p9. Histone H1 kinase activity assay was performed by incubation of the soluble complex with 15 M ATP, 1 mg/ml histone H1, and increasing concentrations of purvalanol, roscovitine, or olomoucine. The complex was assayed for its histone H1 kinase activity by direct counting (A). Cyclin B phosphorylation was assayed similarly in the presence of 15 M ATP and increasing concentrations of purvalanol (1), roscovitine (2), or olomoucine (3) and detected by autoradiography following SDS-PAGE (B). B in our in vitro kinase assay, is the major physiological cyclin B kinase. Third, other kinases may additionally phosphorylate cyclin B in vivo.
Cyclin B Phosphorylation Occurs as an Intra-MPF Complex Event-We next wondered if Cdc2-cyclin B was phosphorylating itself on cyclin B (intra-MPF phosphorylation) or phosphorylating the cyclin B subunit of other complexes (extra-MPF phosphorylation). In a first experiment, we prepared soluble and active Cdc2-cyclin B complex from both starfish and Xenopus metaphase oocytes. Each of these MPF was added to soluble prophase complex dephosphorylated or not dephosphorylated by Cdc25, in the presence or absence of ATP (Fig.  10). In the intraspecific experiment using active MPF from starfish (Fig. 10A), the prophase cyclin B does not shift in the presence of ATP. It shifts only after Cdc25 treatment, which activates prophase Cdc2. Therefore, the addition of active starfish MPF is unable to induce phosphorylation of the prophase cyclin B in the presence of ATP. The signals are similar in the presence or absence of ATP; prophase cyclin B remains in its lower position on the immunoblot. The same result was obtained in the interspecific experiment using active Xenopus MPF, which is unable to induce starfish prophase cyclin B phosphorylation (Fig. 10B). The absence of effect is plainly seen in the interspecific experiment, because Xenopus cyclin B is not recognized by our anti-cyclin B antibodies. These results demonstrate that Cdc2 is able to phosphorylate its own associated cyclin B subunit but not cyclin B from another complex.
We have performed another type of experiment to demonstrate the intra-MPF nature of cyclin B phosphorylation. We reasoned that if cyclin B phosphorylation results from intra-MPF rather than extracomplex phosphorylation, it should be insensitive to dilution, i.e. independent of enzyme/substrate concentration, or in other words independent of the probability of enzyme/substrate encounter. We incubated a constant quantity of soluble and activated prophase Cdc2-cyclin B with [␥-32 P]ATP in increasing reaction volumes from 15 to 150 l (Fig. 11). The ability of Cdc2 to phosphorylate cyclin B was not affected by volume changes; the level of phosphorylated cyclin B remained constant independently of the dilution (Fig. 11A). The same experiment was made in the presence of a low histone H1 concentration (Fig. 11B). Despite the presence of competing histone H1, cyclin B phosphorylation remained unaffected by the dilution. In contrast, phosphorylation of the exogenous substrate decreased according to the reaction volume. This result clearly demonstrates the intracomplex nature of the phosphorylation reaction.

DISCUSSION
In Vivo Cyclin B Phosphorylation at the Prophase/Metaphase Transition-Cyclin B phosphorylation at the G 2 /M transition of the cell cycle has been described in numerous models from yeast to human cells (6, 11, 13, 29 -31). We have characterized the electrophoretic shift of cyclin B occurring at entry in metaphase in starfish oocytes as a phosphorylation reaction (Fig. 3A). During the maturation time course (Fig. 2), cyclin B phosphorylation occurs simultaneously with the Cdc2 dephosphorylation. The dephosphorylated Cdc2-dephosphorylated cyclin B complex (obtained either by treatment of the prophase Cdc2 with Cdc25 or by treatment of the metaphase cyclin B with PP2A 1 ) displays a kinase activity equivalent to that of the metaphase complex (dephosphorylated Cdc2-phosphorylated cyclin B) (Fig. 3). This shows that cyclin B phosphorylation is not essential for the catalytic activity of the complex and confirms similar data obtained indirectly (mutation of phosphorylation sites) in goldfish (29) and Xenopus (33) oocytes. Studies from Li et al. (33,34) have shown the requirement of cyclin B 1 phosphorylation for nuclear localization and Xenopus oocyte maturation. Nuclear localization of cyclin B 1 is regulated by its phosphorylation at sites within the cytoplasmic retention signal domain (34,48). Phosphorylation of cyclin B 1 masking the cytoplasmic retention signal would therefore permit migration into the nucleus, allowing phosphorylation of nuclear substrates by MPF. Ookata et al. (49) have demonstrated that in starfish oocytes, Cdc2-cyclin B complex is in -FIG. 8. p27 kip1 does not inhibit cyclin B phosphorylation. Purified cdk2cyclin E was assayed for its histone H1 kinase activity in the presence of increasing concentrations of p27 kip1 (0 -100 nM). [ 32 P]phosphate incorporation in histone H1 was measured by direct counting (A). The Cdc2-cyclin B kinase was purified from prophase oocyte extracts (P) by affinity chromatography on p9 CKShs1 -Sepharose beads and pretreated with GST-Cdc25A. The cyclin B phosphorylation reaction was performed in the presence of 15 M [␥-32 P]ATP with increasing concentrations of p27 kip1 (0 -100 nM). Cyclin B phosphorylation was detected after SDS-PAGE and Western blotting with anti-cyclin B antibodies (B) (control Cdc2cyclin B, prophase (P)). Signal intensities were measured to evaluate the cyclin B phosphorylation (A).
deed activated in the cytoplasm prior to its translocation to the nucleus.
Cyclin B Phosphorylation Is Carried Out by the Cdc2-Cyclin B Complex-We have shown by various types of experiments that cyclin B phosphorylation is only possible when Cdc2 is dephosphorylated. First, when Cdc25 is inhibited in vivo by vitamin K 3 , no cyclin B phosphorylation is observed (Fig. 4).
Second, the in vitro cyclin B kinase assay consisting in incubation of the prophase complex with ATP only works when Cdc2 has previously been treated with Cdc25 or with PP2A 1 plus Pyp3 successively (Fig. 5). It clearly appears that Cdc2 activation by dephosphorylation of its inhibitory Thr-14 and Tyr-15 residues is required for both in vivo and in vitro phosphorylation of cyclin B. Knowing the importance of Cdc2 activation, we next examined the involvement of Cdc2 kinase activity in in vitro cyclin B phosphorylation. The cyclin B kinase is sensitive to chemical inhibitors of CDKs (Figs. 6 and 7). The level of sensitivity of cyclin B phosphorylation to purvalanol, roscovitine, and olomoucine has been tested (Fig. 7). Both histone H1 kinase activity and cyclin B phosphorylation, assayed simultaneously, were similarly and dose-dependently inhibited by the three inhibitors. Phosphorylation is also inhibited by the presence of histone H1, a competitive substrate (Fig. 3). Although p9 CKShs1 -Sepharose beads bind Cdc2, Cdk2, and Cdk3 in mammalian cell extracts (50,51), in starfish oocytes the extreme and natural synchrony eliminates potential contaminations by CDKs from other cell cycle phases. Furthermore, an unusually large amount of p34 cdc2 -cyclin B kinase is accumulated in oocytes (8,13) and the histone H1 kinase activity associated to the p9 CKShs1 -bound proteins essentially corresponds to p34 cdc2cyclin B (24,44). As a precaution, we tested p27 kip1 , an inhibitor of Cdk2 (Fig. 8). The noninhibition of cyclin B phosphorylation by p27 kip1 eliminates Cdk2 as a candidate for the cyclin B kinase. Therefore, in vitro, the cyclin B kinase could be nothing other than Cdc2 itself. Furthermore, according to the phosphopeptide map analysis (Fig. 9), in vivo and in vitro cyclin B phosphorylated sites are very similar. Since Cdc2 phosphorylates cyclin B in vitro, Cdc2 seems to be the major physiological cyclin B kinase responsible for the cyclin B shift observed in vivo. The consensus sequence for a phosphorylation site by Cdc2 is (S/T)PX(K/R) as reviewed by Nigg (52) and Kishimoto (53), but other distantly related sites are probably equally phosphorylated. The M. glacialis cyclin B contains the sequence 87 SPEP as a potential candidate for phosphorylation by Cdc2. From sequence alignments, Ser-87 corresponds to the phosphorylated Ser-94 from X. laevis cyclin B 1 . The sites corresponding to X. laevis cyclin B 1 Ser-2, Ser-96, and Ser-113 are not conserved in starfish. Ser-94, corresponding to X. laevis Ser-101, is not a consensus Cdc2 phosphorylation site, and it might be phosphorylated by another kinase.
In this study, we have used cyclin B complexed to Cdc2 as a substrate for the cyclin B kinase, presumably the most physiological substrate. Other studies using recombinant cyclin B or cyclin B peptides have proposed other kinases as potential cyclin B 1 kinases, such as mitogen-activated protein kinase (32), and cyclin B 2 kinases, such as c-Mos (36) and Cyk (38). The physiological relevance of these kinases remains unknown. However the incomplete identity of the in vivo and in vitro cyclin B phosphopeptide maps suggests the existence of additional kinases, besides Cdc2, which may be active in vivo on cyclin B. Nevertheless, even the identification of the individual phosphorylation sites would not allow the definitive identification of the kinases, since some kinase sites overlap (mitogenactivated protein kinase and Cdc2, for example).
Existence of an Intra-MPF Phosphorylation of Cyclin B-Autophosphorylation of cyclin B by the Cdc2 kinase itself was suggested, but not demonstrated, several years ago, in sea urchin eggs (11) and in Xenopus oocytes (35). Our results now demonstrate the existence of an intra-MPF phosphorylation of cyclin B (Fig. 10). Active MPFs purified from starfish and Xenopus metaphase oocytes are unable to induce phosphorylation of the prophase cyclin B associated with prophase (inactive) Cdc2. We cannot rule out the possibility that the Xenopus MPF does not recognize starfish cyclin B. This is certainly not the case in the intraspecific experiment. The dilution experiment (Fig. 11) clearly demonstrates that cyclin B phosphorylation occurs in an intracomplex fashion; dilution of the enzyme/ substrate has no consequence on the phosphorylation reaction. Since cyclin B phosphorylation is only intracomplex, phosphorylation of prophase cyclin B by the active complex is obviously impossible, as shown in Fig. 10. This intracomplex mechanism may have a physiological significance in immediately coupling cyclin B phosphorylation to Cdc2 dephosphorylation (and activation). In this case, enzyme and its substrate are close to each other and present in an equimolar ratio. Such a mechanism explains why cyclin B phosphorylation is concomitant with Cdc2 dephosphorylation preceding nuclear translocation in response to 1-methyladenine (Fig. 2). We propose that activated Cdc2 rapidly phosphorylates its own cyclin B subunit in the cytoplasmic retention signal domain, allowing nuclear translocation of the complex as discussed above. Furthermore, intra-MPF phosphorylation of cyclin B is possible because the complex is highly active in its unphosphorylated cyclin B form (Fig. 3).
Conclusion-This study has allowed us to confirm directly that 1) cyclin B phosphorylation is not necessary for the kinase activity of the MPF, 2) Cdc2 is the main kinase responsible for the cyclin B phosphorylation-associated shift observed at the prophase/metaphase transition in starfish oocytes, and 3) an intra-MPF reaction is the mechanism of starfish cyclin B phosphorylation. This unique intracomplex cyclin B phosphorylation belongs to the regulatory pathway controlling MPF function.