Differential Regulation of Cdc2 and Cdk2 by RINGO and Cyclins* 210

Cyclin-dependent kinases (Cdks) are key regulators of the eukaryotic cell division cycle. Cdk1 (Cdc2) and Cdk2 should be bound to regulatory subunits named cyclins as well as phosphorylated on a conserved Thr located in the T-loop for full enzymatic activity. Cdc2- and Cdk2-cyclin complexes can be inactivated by phosphorylation on the catalytic cleft-located Thr-14 and Tyr-15 residues or by association with inhibitory subunits such as p21Cip1. We have recently identified a novel Cdc2 regulator named RINGO that plays an important role in the meiotic cell cycle of Xenopus oocytes. RINGO can bind and activate Cdc2 but has no sequence homology to cyclins. Here we report that, in contrast with Cdc2- cyclin complexes, the phosphorylation of Thr-161 is not required for full activation of Cdc2 by RINGO. We also show that RINGO can directly stimulate the kinase activity of Cdk2 independently of Thr-160 phosphorylation. Moreover, RINGO-bound Cdc2 and Cdk2 are both less susceptible to inhibition by p21Cip1, whereas the Thr-14/Tyr-15 kinase Myt1 can negatively regulate the activity of Cdc2-RINGO with reduced efficiency. Our results indicate that Cdk-RINGO complexes may be active under conditions in which cyclin-bound Cdks are inhibited and can therefore play different regulatory roles.

Cyclin-dependent kinases (Cdks) are key regulators of the eukaryotic cell division cycle. Cdk1 (Cdc2) and Cdk2 should be bound to regulatory subunits named cyclins as well as phosphorylated on a conserved Thr located in the T-loop for full enzymatic activity. Cdc2-and Cdk2cyclin complexes can be inactivated by phosphorylation on the catalytic cleft-located Thr-14 and Tyr-15 residues or by association with inhibitory subunits such as p21 Cip1 . We have recently identified a novel Cdc2 regulator named RINGO that plays an important role in the meiotic cell cycle of Xenopus oocytes. RINGO can bind and activate Cdc2 but has no sequence homology to cyclins. Here we report that, in contrast with Cdc2cyclin complexes, the phosphorylation of Thr-161 is not required for full activation of Cdc2 by RINGO. We also show that RINGO can directly stimulate the kinase activity of Cdk2 independently of Thr-160 phosphorylation. Moreover, RINGO-bound Cdc2 and Cdk2 are both less susceptible to inhibition by p21 Cip1 , whereas the Thr-14/Tyr-15 kinase Myt1 can negatively regulate the activity of Cdc2-RINGO with reduced efficiency. Our results indicate that Cdk-RINGO complexes may be active under conditions in which cyclin-bound Cdks are inhibited and can therefore play different regulatory roles.
Cyclin-dependent kinases (Cdks) 1 are essential regulators of the eukaryotic cell division cycle. Cdks are catalytically inactive as monomers, and their enzymatic activity is modulated by association with other proteins as well as by both inhibitory and activating phosphorylations (1)(2)(3)(4).
Activation of Cdks requires association with regulatory subunits named cyclins that are synthesized and degraded in a cell cycle-dependent manner (5,6). Crystallographic studies have revealed that cyclin binding results in important structural modifications on the catalytic cleft of the Cdk molecule, allowing the correct folding for a better accessibility to ATP and substrates (7). In addition, cyclins have been shown to provide Cdks with targeting domains, responsible for substrate selec-tion and thus biological specificity (8,9). The Cdk-cyclin complex is fully active only after phosphorylation of a conserved Thr residue (Thr-161 in Cdc2 and Thr-160 in Cdk2) located on the T-loop, close to the catalytic cleft of the kinase (10 -14). This phosphorylation is catalyzed by the Cdk-activating kinase (CAK), identified as a trimeric complex (Cdk7, cyclin H, and MAT1) in higher eukaryotes (15)(16)(17)(18)(19) or as a monomeric enzyme, named Cak1/Civ1, in budding yeast (20 -22). CAKs appear to be constitutively active during the cell cycle (12,17,23).
The activity of Cdk-cyclin complexes can be inhibited by phosphorylation of residues present within the ATP-binding region of the Cdk, Thr-14 and Tyr-15 in higher eukaryotes (24,25). Two kinases, Wee1 and Myt1, have been shown to phosphorylate Cdc2 either only on Thr-14 or Tyr-15 or on both residues, depending on the species (26 -28). Both kinases are down-regulated by phosphorylation just before entry into the M phase of the cell cycle (28 -30). Cdk-cyclin complexes can also associate with small inhibitory molecules, the CKIs (Cdk inhibitors) (31). CKIs interact both with the amino-terminal lobe of the Cdk and with the cyclin subunit (32), and they can inhibit cyclin-bound Cdks by disrupting the correct folding of the catalytic cleft of the kinase and by interfering with CAKinduced phosphorylation (reviewed in Ref. 4).
During the progesterone-induced meiotic maturation (G 2 /M phase progression) of Xenopus oocytes, two different Cdc2 populations need to be activated. In the G 2 -arrested oocyte, about 10% of the Cdc2 is associated with cyclin B and phosphorylated on Thr-161 but maintained inactive by phosphorylation on Thr-14 and Tyr-15. The remaining 90% of the Cdc2 is free, not complexed with cyclins, but it also appears to be required for progesterone-triggered M phase entry. Thus, the G 2 /M progression in oocytes involves dephosphorylation of Thr-14/Tyr-15 residues of the cyclin B-associated Cdc2, which is preceded by the activation of monomeric Cdc2 by newly synthesized proteins (reviewed in Ref. 33). This free Cdc2 recruited to newly formed complexes during maturation should be able to escape the mechanisms that normally ensure the inhibition of Cdc2cyclin complexes in the G 2 -arrested oocyte. A candidate for this triggering role is a recently identified protein named RINGO (rapid inducer of G 2 /M progression in oocytes) (34) or Speedy (35). This protein is required for the meiotic G 2 /M progression in Xenopus oocytes and can bind to and stimulate the kinase activity of Cdc2 (34) as well as bind to Cdk2 (35). Here we have shown that RINGO binding also activates Cdk2 and have analyzed the biochemical mechanisms that regulate Cdc2 and Cdk2 activation by RINGO. Our results indicate that RINGObound Cdc2 and Cdk2 are active in the absence of the conserved T-loop located Thr phosphorylation and at the same time are more resistant to negative regulation via inhibitory phosphorylations and association with CKIs.

Bacterial Expression and Purification of Recombinant Proteins-
MalE-RINGO was expressed in Escherichia coli and purified as described (34). The GST fusion proteins of Xenopus Cdc2 wild type, K33R, and T161A, Xenopus Cdk2 wild type, K33R, and T160A, human p21 Cip1 , and human cyclin A were bacterially expressed and purified as described (36). Human His-cyclin B1 was expressed in baculovirusinfected insect cells and purified as described (37). Human Myt1 purified from baculovirus-infected insect cells was provided by R. Booher (38). Cak1/Civ1-His (both wild type and the kinase-dead mutant K31A) was purified from yeast and provided by C. Mann (20).
Microinjection and Fractionation of Xenopus Oocytes-Fully grown stage VI Xenopus oocytes were microinjected with 50 nl of the indicated recombinant proteins or in vitro transcribed capped mRNAs prepared using the MEGAscript kit (Ambion). The RINGO and Cdc2 constructs for in vitro transcription have been described previously (34,39). The Cdc2 T161A mutant was prepared by polymerase chain reaction using as a template the same Cdc2 construct and the QuikChange sitedirected mutagenesis kit (Stratagene). Oocyte lysates were prepared in 4 volumes of EB buffer (80 mM ␤-glycerophosphate, 20 mM EGTA, 15 mM MgCl 2 , 1 mM dithiothreitol, pH 7.3, and protease inhibitors) and fractionated exactly as described in Ref. 40 to obtain the 40 -60% ammonium sulfate fraction F60, which was stored in aliquots at Ϫ80°C.
Kinase Activation Assays in Oocyte Extracts (F60)-Aliquots of the F60 cytosolic fraction corresponding to 80 oocytes were thawed, and the ammonium sulfate pellet was dissolved in 400 l of EB buffer. The reconstituted F60 was then concentrated on a Microcon-10 filter device (Amicon-Millipore) to a final volume of 80 l, supplemented with 10 l of 1 mM ATP and 10 l of 0.5 mM protein kinase A inhibitor peptide (Sigma). In a typical reaction, 15 l of freshly reconstituted F60 were preincubated with 10 l of purified Cak1/Civ1 or p21 Cip1 for 15 min at 30°C followed by a further incubation with 10 l of His-cyclin B1 or MalE-RINGO for 15 min at 30°C. Myt1 (10 l) was added after cyclin B1 or RINGO. In some cases, Myt1 was added in the F60 in the presence of RINGO or cyclin B1 and 100 M ATP. When any of these components were not included in the reaction, 10 l of EB buffer were added. The histone H1 kinase activity was measured in a final volume of 50 l in the presence of 100 M ATP, 3 Ci of [␥-32 P]ATP (3,000 Ci/mmol, Amersham Pharmacia Biotech), 10 mM MgCl 2 , and 0.3 mg/ml of histone H1 (Sigma). After 15 min at 30°C, the kinase reactions were analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. For quantification, the histone H1 bands were excised from the gel, and radioactive phosphate incorporation was determined by liquid scintillation counting (Wallac 1409 counter). In some experiments, phosphorylated bands were directly quantified on the gels using a Fuji phosphorimaging device.
Kinase Activation of Recombinant Cdc2 and Cdk2-Bacterially expressed GST⅐Cdc2 proteins could not directly be activated in vitro but require prior incubation in concentrated Xenopus oocyte extracts (34), perhaps for the correct folding of the proteins. The GST⅐Cdc2 proteins were then recovered on glutathione-Sepharose beads (10 l, Amersham Pharmacia Biotech) and washed three times with EB buffer before being used for kinase reactions. In contrast, purified bacterially produced GST⅐Cdk2 was directly used for the Cdk2 activation assays. Typically 1-2 M GST-fused Cdc2 or Cdk2 were used per assay following the same protocol as indicated above for the F60 cytosolic fraction except that we used a final volume of 12 l.
Immunoprecipitation, Immunoblotting, and Glutathione-Sepharose Pull-down-Immunoprecipitations and immunoblottings were performed as described in Ref. 36, using the anti-Cdc2 monoclonal antibody A17 coupled to beads (Santa Cruz Biotechnology, Inc.), the anti-Cdc2 monoclonal antibody 3E1 (a kind gift from Julian Gannon and Tim Hunt), and affinity-purified rabbit anti-RINGO antibodies (34).
To recover GST⅐p21 Cip1 and its associated proteins, 45 l of the F60 fraction were diluted to 100 l with EB buffer, and then 10 l of glutathione-Sepharose beads were added. After a 45-min incubation, the beads were recovered by centrifugation and washed three times. Both supernatant and bead fractions were analyzed by immunoblotting using the anti-Cdc2 antibody 3E1.

RINGO-induced Cdc2 Activation Is Independent of Thr-161
Phosphorylation-We have previously shown that purified recombinant RINGO can stimulate the kinase activity of Cdc2 expressed in insect cells or in Xenopus oocytes as well as bacterially produced GST⅐Cdc2 (34). As full activation of Cdc2-cyclin complexes requires phosphorylation of Cdc2 on Thr-161 (3, 4), we investigated whether this phosphorylation was also required for activation of Cdc2 by RINGO. We incubated purified RINGO or cyclin B1 proteins with the F60 fraction prepared from G 2 -arrested oocyte extracts by ammonium sulfate precipitation, which contains only monomeric Cdc2 but not Aand B-type cyclins or Cdk2 (40) (see supplementary data). The experiment was performed in the absence or presence of the monomeric yeast Cdk-activating kinase Cak1/Civ1, which can efficiently phosphorylate Thr-161 in Cdc2 (20 -22). When the F60 was preincubated in the presence of Cak1/Civ1, we observed a strong increase in the H1 kinase activity triggered by purified cyclin B1, as expected ( Fig. 1, A, B, and E). The addition of RINGO to the F60 induced H1 kinase activity ( Fig. 1, C-E). However, Cak1/Civ1 did not increase but rather partially inhibited the RINGO-induced H1 kinase activity in the same F60 fraction (Fig. 1, C-E). Cak1/Civ1 had the same MalE-RINGO or His-cyclin B1 was then added either at 2 M (corresponding to 100 g/ml and 125 g/ml for MalE-RINGO and His-cyclin B1, respectively) (B, D, and E) or the indicated concentrations (A and C), and the samples were incubated for an additional 15 min at 30°C prior to the histone H1 kinase assay. F, F60 fraction incubated with wild type or a kinase-dead (kd) Cak1/Civ1 mutant (0.25 M), MalE-RINGO (2.5 M), or His-cyclin B1 (1 M), as indicated above. 100% histone H1 kinase activity corresponds to 1,500 and 10,800 cpm for RINGO and cyclin B1, respectively. effect when added to the extract before or after the addition of RINGO (data not shown). This suggests that T-loop phosphorylation is not needed for (and may even impair) Cdc2 activation by RINGO.
The reduction of RINGO-induced Cdc2 activity in the presence of Cak1/Civ1 may be attributable to an inhibitory effect of the phosphate bound to Thr-161 or to the interference between RINGO and Cak1/Civ1 for Cdc2 interaction. To distinguish between these two possibilities, we compared the effect of wild type Cak1/Civ1 and a kinase-inactive Cak1/Civ1 mutant on the RINGO-induced H1 kinase activity of the F60 fraction (Fig. 1F). Both wild type and kinase-dead Cak1/Civ1 proteins inhibited Cdc2 activity to the same extent, indicating that the inhibitory effect of Cak1/Civ1 on Cdc2-RINGO probably involves steric interference.
To confirm these results, we used bacterially produced GST⅐Cdc2 preincubated in concentrated Xenopus oocyte extracts (34). RINGO was able to stimulate the H1 kinase activity of wild type GST⅐Cdc2 but not of an inactive GST⅐Cdc2 K33R mutant protein processed in parallel ( Fig. 2A). We then incubated wild type GST⅐Cdc2 or the GST⅐Cdc2 T161A mutant in the absence or presence of Cak1/Civ1 and assayed their kinase activity on histone H1 (Fig. 2B). As expected, cyclin B1 was able to activate the H1 kinase activity of GST⅐Cdc2, and this was strongly enhanced by the presence of Cak1/Civ1 in the case of Cdc2 wild type but not the Cdc2 T161A mutant (Fig. 2B). In contrast, RINGO activated both Cdc2 wild type and the T161A mutant to similar extents (Fig. 2B), and their kinase activities were not further increased in the presence of Cak1/Civ1. In fact, Cak1/Civ1 partially inhibited the RINGO-dependent activation of Cdc2 but did not inhibit the nonphosphorylable T161A mutant (Fig. 2B). This suggests that the inhibition might not require Cak1/Civ1 catalytic activity but is apparently abolished by replacing the Thr-161 of Cdc2 with an Ala residue. Altogether, the results indicate that the activation of Cdc2 by RINGO, either endogenous Cdc2 present in oocyte extracts or purified recombinant GST⅐Cdc2, can occur in the absence of Thr-161 phosphorylation. Furthermore, we reproducibly observed that Cdc2-RINGO complexes exhibit a lower specific activity toward histone H1 than Thr-161-phosphorylated Cdc2-cyclin B complexes.
Consistent with these in vitro results, we also found that overexpression in Xenopus oocytes of either wild type Cdc2 or the T161A Cdc2 mutant resulted in similar levels of RINGOtriggered H1 kinase activity (Fig. 3A). In contrast, the level of H1 kinase activity triggered by ectopic cyclin B1 was strongly reduced in Cdc2 T161A-expressing oocytes (Fig. 3A). These results indicate that RINGO is also able to induce Cdc2 activation independently of Thr-161-phosphorylation in oocytes. Moreover, RINGO binds Cdc2 under these conditions since we could co-immunoprecipitate Cdc2 and RINGO from oocyte extracts but only upon overexpression of the proteins (Fig. 3B).
RINGO-induced Cdk2 Activation Is Independent of Thr-160 Phosphorylation-RINGO has been shown to bind to Cdk2 upon overexpression in Xenopus oocytes, although it was not reported whether this resulted in the activation of Cdk2 by RINGO (35). We found that the kinase activity of bacterially produced GST⅐Cdk2 was directly activated (without any prior incubation in oocyte extracts) by recombinant RINGO, as determined by using histone H1 as an in vitro substrate (Fig. 4A). Moreover, the addition of Cak1/Civ1 did not modify the extent of RINGO-triggered Cdk2 activation, whereas Cak1/Civ1 was absolutely required for the cyclin A-induced activation of GST⅐Cdk2 (Fig. 4A). Under these conditions, we confirmed that Cak1/Civ1 phosphorylated Cdk2 with similar efficiency in the presence of either cyclin A or RINGO and that RINGO did not induce any Cdk2 autophosphorylation in the absence of Cak1/ Civ1 (Fig. 4B). These results indicate that RINGO can directly activate Cdk2 and that this activation occurs in the absence of Thr-160 phosphorylation, as in the case of Cdc2. However, in contrast to Cdc2-cyclin B, the level of activation of Cdk2 was similar using either RINGO alone or cyclin A plus Cak1/Civ1. Interestingly, preincubation of Cdk2 with RINGO led to poor subsequent phosphorylation by Cak1/Civ1 in comparison with Cdk2 alone or Cdk2 preincubated with cyclin A (Fig. 4C, upper  panel). This lower level of Cdk2 phosphorylation did not affect the H1 kinase activity of the Cdk2-RINGO complex (Fig. 4C,  lower panel). These results indicate that RINGO binding may interfere with the accessibility of Cak1/Civ1 to Cdk2. To confirm that Cdk2 activation by RINGO is truly independent of Thr-160 phosphorylation, we used the Cdk2 T160A mutant. We found that RINGO stimulated the H1 kinase activity of the T160A mutant protein as efficiently as (or perhaps even slightly better than) in the case of the wild type Cdk2 (Fig. 4D). As a control, RINGO did not activate the Cdk2 K33R mutant.
Myt1 Inhibits Cdc2-RINGO with Reduced Efficiency-Because the phosphorylation of the conserved Thr located in the T-loop was not required for the activation of Cdc2 and Cdk2 by RINGO, we investigated whether the inhibitory phosphorylation on Thr-14/Tyr-15 could affect the activity of the Cdc2-RINGO complex. We used the human Myt1 kinase, which has been shown to phosphorylate Cdc2-cyclin complexes preferentially on Thr-14 and to inhibit their kinase activity (26,38). We found that the addition of 1 M Myt1 to F60 oocyte extracts significantly reduced the H1 kinase activity triggered by either cyclin B1 or RINGO, suggesting that the Cdc2-RINGO complex is also subjected to inhibition by Myt1 (Fig. 5A). However, lower concentrations of Myt1 (below 0.3 M) were more efficient inhibitors of histone H1 kinase activity of Cdc2-cyclin B than of Cdc2-RINGO. Using recombinant GST⅐Cdc2, we confirmed the reduced inhibitory effect of Myt1 on Cdc2-RINGO (Fig. 5B). The different concentrations of recombinant GST⅐Cdc2 (1 M) versus endogenous Cdc2 present in the F60 (about 0.05 M) could explain the differences in the degree of inhibition between both systems. We also found that the reduced inhibitory effect of Myt1 on Cdc2-RINGO correlated with less efficient phosphorylation of Cdc2 in the presence of RINGO than in the presence of cyclin B1 (data not shown).
p21 Cip1 Is a Poor Inhibitor of Cdc2-RINGO and Cdk2-RINGO-Another important mechanism that down-regulates the activity of Cdks is their association with small proteins termed CKIs (4). We investigated the effect of the CKI family member p21 Cip1 on the RINGO-induced activation of Cdc2. As A, oocytes injected with in vitro transcribed mRNAs encoding either Cdc2 wild type (wt) or Cdc2 T161A and incubated for 10 h prior to the injection of 50 ng of recombinant MalE-RINGO or His-cyclin B1. As a control, some oocytes were only injected with the Cdc2 mRNAs. After a further incubation for 2 h (the injected Cdc2 was overexpressed about 3-fold over the endogenous level), the oocyte lysates were prepared and assayed for histone H1 kinase activity, which was quantified using a phosphorimaging device. B, oocytes injected with in vitro transcribed Cdc2 mRNA and incubated overnight prior to the injection of RINGO mRNA or stimulation with progesterone (15 M), as indicated. Lysates were prepared from 10 oocytes at the time of GVBD and immunoprecipitated with anti-Cdc2 antibodies followed by immunoblotting with anti-RINGO antibodies. Aliquots of the same lysates were also directly analyzed by immunoblotting using anti-Cdc2 antibodies. expected, when the F60 oocyte extract was preincubated with 1 M p21 Cip1 (a concentration sufficient to inhibit most of the known Cdks (41)), the cyclin B1-induced H1 kinase activation was completely blocked. However, the H1 kinase activity triggered by RINGO was only partially inhibited (Fig. 6A). This indicates that p21 Cip1 has lower affinity and/or inhibitory efficiency for the Cdc2-RINGO complex than for the Cdc2-cyclin B complex. We also investigated the ability of p21 Cip1 to inhibit the Cdk2-RINGO complex. We found that 1 M p21 Cip1 had no inhibitory effect on the GST⅐Cdk2-RINGO complex assembled in vitro using purified recombinant proteins, although the GST⅐Cdk2-cyclin A complex was efficiently inhibited by the same concentration of p21 Cip1 (Fig. 6B). We also found evidence that Cdc2-RINGO complexes may be more resistant to p21 Cip1 inhibition in Xenopus oocytes. Injection of p21 Cip1 in oocytes prevented Cdc2 activation induced by progesterone. In contrast, RINGO was still able to induce significant Cdc2 activation in p21 Cip1 -injected oocytes (Fig. 6C).
It therefore appears that both Cdc2-and Cdk2-RINGO complexes are less susceptible to inhibition by p21 Cip1 . As the interaction mediated by the MRAIL motif of cyclins is essential for the ability of p21 Cip1 to inhibit Cdks, a simple interpretation of our results is that RINGO lacks the MRAIL hydrophobic patch, leading to a weak Cdk-CKI association (32,42). To test this possibility, GST⅐p21 Cip1 was added to F60 (which contains only monomeric Cdc2) in the presence of either RINGO or cyclin B1 and then recovered on glutathione-Sepharose beads to investigate binding of Cdc2 to p21 Cip1 by immunoblot (Fig. 6D). In these experiments, the addition of cyclin B1 led to the recovery of a significantly higher amount of Cdc2 associated with p21 Cip1 than the addition of RINGO (Fig. 6D), suggesting that p21 Cip1 has lower affinity for Cdc2-RINGO complexes.

DISCUSSION
Cdks are catalytically inactive as monomers and, for full activity, should be bound to a cyclin as well as phosphorylated on a Thr located on the T-loop, whereas the catalytic cleftlocated Thr-14 and Tyr-15 residues should be dephosphorylated. We have investigated the mechanism of activation of Cdks by a new protein named RINGO that has no similarity to cyclins at the amino acid sequence level. We found that Cdk- RINGO complexes respond to Cdk regulatory mechanisms in a different way than Cdk-cyclin complexes. Thus, the RINGOinduced kinase activity of Cdc2 and Cdk2 can bypass the requirement for phosphorylation of Thr located on the T-loop both in vitro and in oocytes. Moreover, Cdk-RINGO complexes appear to be less susceptible to inhibition by Thr-14/Tyr-15 phosphorylation and CKI binding.
How can the activity of Cdc2-and Cdk2-RINGO complexes be independent of Thr-160/161 phosphorylation, which is essential for full activation of these Cdks by cyclins? Crystallographic studies have revealed that phosphorylation of the Thr located in the T-loop induces subtle but important conformational changes; this phosphate acts as an organizing center, stabilizing the T-loop region (which probably participates in substrate recognition) as well as the Cdk-cyclin complex (14). In fact, one of the functions of the T loop phosphorylation may be to improve the binding of protein substrates independently of the catalytic activity per se (reviewed in Ref. 4). A hypothesis consistent with our results is that RINGO binding to Cdc2 and Cdk2 induces the conformational changes in the T-loop region that are normally induced upon Thr-160/161 phosphorylation. It should be noted that, in contrast to Cdc2 and Cdk2, other Cdks have been reported to be active in the absence of phosphorylation in the T-loop (for example the Cdk7-cyclin H-Mat1 complex (reviewed in Ref . 4)). This is also the case for Cdk5, which associates with and is activated by a brain-specific 35-kDa subunit, p35, as well as by its 25-kDa truncated form (p25) produced by the action of proteases (43,44). Interestingly, a recent crystallographic study of the neuronal Cdk5-p25 complex reveals that the interaction between the p25 activator and the T-loop of Cdk5 results in a T-loop conformation closely resembling that of the Thr-160-phosphorylated Cdk2 associated with cyclin A. 2 These data suggest a common mechanism by which non-cyclin Cdk activators could bypass the requirement for T loop phosphorylation.
RINGO-bound Cdc2 and Cdk2 are also less susceptible to two mechanisms that negatively regulate the activity of Cdkcyclin complexes: interaction with CKIs and phosphorylation on Thr-14/Tyr-15. This appears to correlate with weaker interaction with p21 Cip1 and less efficient phosphorylation by Myt1 of the Cdk-RINGO complexes. Interestingly, the similarity between RINGO and other non-cyclin Cdk activators also extends to the negative regulators as p25 allows Cdk5 to escape inhibition by the Tyr-15-specific kinase Wee1 (44) and by CKIs such as p27Kip1 (45). The absence of the MRAIL hydrophobic motif, typical of cyclins, could be responsible for the reduced inhibitory efficiency of p21 Cip1 , as well as Myt1, on Cdc2-RINGO. It will be interesting to test the activity of Cdk-RINGO complexes on other substrates that require the MRAIL motif (for example the retinoblastoma family (42)). It is also conceivable that RINGO may provide Cdks with a substrate recognition and recruitment patch, perhaps also of a hydrophobic nature as in the MRAIL motif of cyclins (8).
Our biochemical findings confirm the properties expected for RINGO to have a triggering role in the activation of the meiotic cell cycle in Xenopus oocytes. Previous results indicated that newly synthesized RINGO protein is likely to be implicated in the activation of monomeric Cdc2 during the progesteroneinduced oocyte maturation (34). About 90% of the Cdc2 present in the oocyte is not associated with cyclins (46), and only part of this monomeric Cdc2 appears to be already phosphorylated on Thr-161. 3 During the progesterone-triggered G 2 /M transition, newly synthesized cyclin and RINGO proteins could target the two different Cdc2 populations, phosphorylated and nonphosphorylated on Thr-161, respectively. The Cdc2-RINGO complexes would be excellent triggers of the G 2 /M progression of the oocyte. Firstly, they are poor substrates for Myt1, the kinase that phosphorylates Cdc2 on Thr-14/Tyr-15 and is probably responsible for the maintenance of Cdc2-cyclin B complexes inactive in G 2 -arrested oocytes (36,47). Secondly, Cdc2-RINGO complexes can also be active in the absence of Thr phosphorylation in the T-loop and can therefore bypass negative regulation by Thr-161 phosphatases potentially present in the oocyte, such as PP2C (48). Thus, the RINGO-activated Cdc2 generated in response to progesterone would not be affected by negative regulators of Cdc2 activity present in the oocyte and would be able to trigger the maturation process. Our results are also consistent with the possibility that newly synthesized RINGO may directly activate Cdk2, which is present in G 2 -arrested oocytes and accumulates in response to progesterone (49). Previous work based on the use of the p21 Cip1 inhibitor had concluded that Cdk2 activity might be dispensable for the meiotic cell cycle in oocytes (50). However, these results need to be re-evaluated in light of our observations that p21 Cip1 is not an efficient inhibitor of Cdk2-RINGO. What the actual substrates of the RINGO-activated Cdc2 and Cdk2 could be in oocytes remains to be investigated. However, it should be noted that either the activation of Cdc25 or the inhibition of Myt1 should suffice to convert the inactive Cdc2-cyclin B complexes into active M-phase promoting factor and thus start the auto-amplification feedback loops that would induce entry into M-phase of meiosis of the oocyte.