AN ESSENTIAL PHOSPHORYLATION-SITE DOMAIN OF HUMAN CDC25C INTERACTS WITH BOTH 14-3-3 AND CYCLINS

Human cdc25C is a dual-specificity phosphatase involved in the regulation of cell cycle progression in both unperturbed cells and in cells subject to DNA-damage or replication checkpoints. In this study, we describe the structure-function relationship of an essential domain of human cdc25C which interacts with 14-3-3 proteins. We show that this domain is a bi-functional interactive motif, which interacts with cyclins, primarily through their P-box motif, in addition to 14-3-3 proteins. Characterization of the structural features of this domain by NMR spectroscopy and circular dichroïsm reveals two distinct alpha helical moieties interconnected by a loop carrying the 14-3-3 binding site. Moreover, the helical folding is induced upon binding to 14-3-3, suggestive of a conformational regulation of this domain of cdc25C through interactions with partner proteins i n vivo . Combining our structural and biochemical data, we propose a detailed model of the molecular mechanism of cdc25C regulation by differential association with 14-3-3 and cdc2-cyclin B.


4
of cdc25 by enhancing the nuclear export of cdc25 thanks to its nuclear export signal (NES) (26). In Xenopus and in human cells, in contrast, 14-3-3 proteins seem to promote the increased cytoplasmic retention of cdc25 by preventing recognition of the bipartite nuclear localization sequence (NLS) of cdc25 by the nuclear import machinery (27)(28)(29)(30).
In spite of these numerous studies, the molecular basis underlying the mechanism of regulation of cdc25 by 14-3-3 has still not been resolved. With the aim of investigating this issue, we used fluorescence spectroscopy, gel filtration and affinity chromatography to examine the interactive properties of a 51 amino acid peptide, MP51, derived from residues 195 to 244 of human cdc25C which encompass the 14-3-3 binding site and the putative bipartite NLS (31) . We found that this domain of cdc25C not only interacts with 14-3-3ζ in vitro and in vivo, but also with human cyclins A and B1 with similar high affinity. We show that the interaction of MP51 with cyclins is primarily mediated through their P-box motif and is independent on MP51 phosphorylation, in contrast to the interaction between MP51 and 14-3-3ζ. Moreover we found that phosphorylation of both MP51 and full-length human cdc25C by cds1 favours the formation of stable MP51 and cdc25C / 14-3-3 complexes over that of MP51 or cdc25C / cyclin B1 complexes, both in vitro and in vivo. Finally, we examined the structural properties of MP51by circular dichroism and NMR spectroscopy, and propose the first structural model of an essential N-terminal regulatory domain of human cdc25C.
The consequences of the dual-functionality of this domain of human cdc25C are discussed with respect to its structural characteristics, as a function of its phosphorylation status, and in the context of cell cycle regulation.

EXPERIMENTAL PROCEDURES
by guest on March 23, 2020 http://www.jbc.org/ Downloaded from Synthesis, phosphorylation and purification of MP51 -MP51 was synthesized by solid phase peptide synthesis according to the Fmoc/tBuc strategy, then purified by ion exchange chromatography and semi-preparative HPLC as described previously (31). MP51 was identified by electrospray mass spectrometry and amino acid analysis. Stoichiometric phosphorylation of MP51 on Ser 22 was performed using 1-5 µg recombinant human GST-cds1 (kindly provided by Dr. C. McGowan) for 30 minutes at 37°C in kinase buffer (50mM Tris-HCl pH 7.5, 100mM NaCl, 10mM MgCl 2 , 2mM ATP). Following phosphorylation, MP51 was purified by gel filtration chromatography on a Superose 12 column and identified by mass spectrometry. . Circular Dichroism -CD spectra were recorded on a Jasco 800 dichrograph using 1 mm thick quartz cells. MP51 or MP51P was initially diluted to a final concentration of 0.1 mg ml -1 in water or phosphate buffer (pH : 7.0) and structural variations were measured as a function of changes in the initial CD spectrum upon addition of increasing concentrations of methanol or 2,2,2-trifluoroethanol (TFE). CD spectra were measured between 185 nm and 260 nm, and ellipticity values were calculated at 220 nm as previously described (32). For binding experiments, 14-3-3ζ was diluted to a final concentration of 8 µM in phosphate buffer pH:7.0 , incubated for 15 min with a two-fold excess of MP51 or MP51P. Data collection and analysis were performed with the package software from Jasco. Bruker AMX 600 spectrometer, equipped with q z-gradient triple resonance ( 1 H, 15 N, 13 C) probe, at 6 and 70 ms mixing times and NOESY (34) with 200 and 300 ms mixing times were used for proton assignments. 2D heteronuclear inverse detected 13 C-1 H HSQC (35) and HSQC-TOCSY spectra were recorded at natural abundance for carbon identification. The carrier frequency was centered on the water signal and the solvent was suppressed by continuous low power irradiation during the relaxation delay and during the mixing time for NOESY spectra. 2D spectra were obtained using 2048 and 4096 points for each t1 value, and 512 t1 experiments were acquired. The spectral width was 10 ppm for proton and 165 ppm for 13 C. Prior to Fourier transform, the time domain data were multiplied by a phase-shifted sine bell or square-sine-bell window function in both dimensions. 1 H chemical shift assignments were achieved by standard two-dimensional sequence-specific methods (36). Assignments of α-carbons were determined by direct heteronuclear correlation (HSQC) and ambiguities were solved through remote connectivities obtained by homonuclear transfers in HSQC-TOCSY. The random coil chemical shift values used to calculate the chemical-shift index were those obtained in aqueous solution containing 30% v/v TFE (37). The methylene carbon of TFE was taken as reference at 61.70 ppm in 13 C spectra. Expression and purification of proteins -Human cdk1, cdk2 and human cyclins A and B1 were overexpressed in E. coli and purified as described previously (39,40). pmal-human 14-3-3 ζ (kindly provided by Dr. D. Fisher) was prepared as a maltose-binding fusion protein, and maltosebinding protein (MBP) was removed with factor Xa. Human 14-3-3 ζ was further purified by ion exchange chromatography on a Mono Q column. The P-box domain of human cyclin B1 (residues 183 to 237), was overexpressed in E. coli and purified as a maltose-binding fusion protein. Fulllength human-cdc25C was prepared as described previously (40). GST-Nter-cdc25C (residues 1 to 302) was overexpressed in E coli and purified as a GST-fusion protein. For fluorescence experiments, the GST tag was cleaved with thrombin and Nter-cdc25C was futher purified by ion exchange chromatography on a Resource Q column (Pharmacia). Both full-length cdc25C and Nter-cdc25C were phosphorylated with recombinant human GST-cds1 for 30 min at 37°C in a kinase buffer (50 mM Tris-HCl pH 7.5, 100 NaCl, 10 mM MgCl 2 , 2 mM ATP). Following phosphorylation, cdc25C and Nter-cdc25C were purified by affinity chromatography on a GST-Trap column and by gel filtration chromatography on a Superose 12 column. Protein concentrations were routinely determined using the BCA protein Assay Reagent kit (Pierce), using bovine serum albumin as a standard .

Secondary structure prediction and modeling of MP51 -
Complexes formed between MP51 or MP51P and cyclin B1 or 14-3-3ζ were purified by size exclusion chromatography using either a Bio-Sil TSK 125 column (BioRad) or a Superose 12 column (Pharmacia). Cyclin B1 and/or 14-3-3ζ were incubated with a 2-fold excess of peptide for 30 min at 25°C in 50 mM potassium phosphate buffer (pH 7.5) containing 150 mM NaCl.
Chromatographies were performed in the same buffer at a flow rate of 1 ml per minute and peak fractions were analyzed by SDS-PAGE. A similar procedure was used to purify complexes formed between phosphorylated full-length cdc25C and cyclin B1 or 14-3-3ζ. Binding experiments were performed by tryptophan fluorescence spectroscopy essentially as described previously (39,40). Intrinsic tryptophan fluorescence was excited at 290nm and emission spectra were recorded between 310 and 400 nm. Recombinant cdks, cyclins, cdk/cyclin complexes and human 14-3-3ζ were incubated in fluorescence buffer for 30 minutes and titration experiments were initiated by addition of increasing concentrations of peptide (MP51, MP51P) or proteins (cdc25C, Nter-cdc25) to a fixed concentration of these proteins (100 nM). Measurements were corrected and fitted according to a standard quadratic equation using the Grafit software (Erathicus software Ltd) as previously described (39,40) .

Binding of MP51 to 14-3-3 -
With the aim of dissecting the mechanism of cdc25C regulation by 14-3-3 at the molecular level, we took advantage of the absence of tryptophan groups in MP51 and of the presence of highly conserved tryptophan residues in 14-3-3 proteins to monitor and quantify their interactions by intrinsic fluorescence spectroscopy. 14-3-3 proteins indeed contain two well conserved tryptophan residues : Trp 59 located in helix C and Trp 228 in helix I, which form respectively the floor and the wall of the channel in the X-ray structure of 14-3-3 (25,41,42). As previously documented (31) and as shown in Fig.1a , addition of increasing concentrations of unphosphorylated MP51 to 14-3-3ζ induced quenching of its intrinsic fluorescence by 40 % and curve fitting yielded a Kd value of 132 nM ± 25. Moreover, we were able to purify stable complexes of unphosphorylated MP51/14-3-3ζ by gel filtration chromatography, in the presence of a high concentration of salt, revealing that the interactions are mainly hydrophobic (Fig.2a). The same experiment performed with MP51 phosphorylated on Ser 22 , denoted MP51P, induced a similar quenching of fluorescence, but a 6-fold higher affinity for 14-3-3ζ, with a Kd value of 21 ± 9 nM, revealing the pivotal role of phosphorylation of Ser 22 in this interaction. However, the observation that MP51 peptide is able to interact with 14-3-3ζ in vitro in the absence of phosphorylation suggests that initial interactions occur through residues other than phosphoSer 22 . The significant quenching of fluorescence observed in these experiments indicates that direct contacts between residues of MP51 and the Trp groups of 14-3-3ζ are involved. Moreover, curve fitting reveals that two molecules of MP51 or MP51P bind per 14-3-3ζ dimer, indicating that the two binding sites in the 14-3-3ζ dimer are accessible and that binding of MP51 peptide to one site does not exclude binding to the other, as already documented for short 14-3-3 binding peptides containing tandem phospho-serine motifs (25). In comparison, as reported in Table 1, unphosphorylated recombinant full-length cdc25C and the N-terminal domain of cdc25C, denoted N-ter cdc25C, interacted very poorly with 14-3-3ζ , whereas their phosphorylation by cds1 promoted their high affinity interaction with 14-3-3ζ with Kd values of 57 ± 5 nM and 48 ± 7 nM, respectively and a stoichiometry of 1 per 14-3-3ζ. dimer. involves mainly hydrophobic contacts. Interestingly, we have previously shown that full-length recombinant human cdc25C interacts with cyclins A and B1 with very similar Kd values, i.e. 72 ± 4.5 nM and 85 ± 9 nM, respectively (40). As such, these data suggest that MP51 is the main domain in cdc25C involved in high affinity interactions with cyclins. At least two studies have shown that a well-defined motif in cyclin B1, the P-box, is an essential domain required for the cdc25-dependent dephosphorylation of cdc2-cyclin B at the G2/M transition, and which may serve as an activator of cdc25 phosphatases " in trans " (8,43). We have previously suggested that human cdc25C might interact physically with this motif (40). Like cyclins, the P-box is characterized by the presence of a conserved tryptophan group, of which we took advantage to investigate whether MP51 could interact with the P-box, by fluorescence spectroscopy. As shown in Fig.1c, addition of increasing concentrations of MP51 to an MBP-P-box fusion protein induced quenching of its intrinsic tryptophan fluorescence by 20%, whereas MBP alone did not interact with MP51. These results indicate that the P-box motif and the MP51 motif of human cdc25C interact directly , and curve fitting yielded a dissociation constant for this interaction of ca 300 nM. As for full-length cyclin B1, binding of MP51 to the P-box was not regulated by phosphorylation on Ser 22 (Table I). We also demonstrated that recombinant full-length and Nter-cdc25C formed stable complexes with cyclin B1 and with its P-box, independently of phosphorylation (TableI).These results show, for the first time, that the P-box domain of cyclin B1 interacts physically with human cdc25C and that MP51 is likely to be the main domain of cdc25C involved in this interaction. However, comparison of the dissociation constants for the different interactions indicates that other residues within the Nterminal, regulatory domain of cdc25C, as well as other domains of cyclin B1 must be involved in secondary interactions between these two proteins.

Binding of MP51 to cdks and cyclins -
We have previously shown that in vitro, recombinant full-length cdc25C interacts with monomeric cyclins with high affinity, but with monomeric cdks with low substrate-enzyme affinity, and with preformed cdk-cyclin complexes with intermediate, cooperative affinity, suggesting that in vivo the interaction between cdc25C and cdk-cyclin complexes is primarily mediated by the interaction between cdc25C and the cyclin (40). We therefore examined to what extent MP51 could interact with cdks and cdk-cyclin complexes. As shown in Fig.1d, addition of MP51 to cdk2 or cdk1 (data not shown for the latter) induced an increase in their intrinsic fluorescence by a factor of 1.5 to 2.2. Best fits for titration curves yielded Kd values of 568 ± 100 nM for cdk1 and 510 ± 95nM for cdk2, in the same range as those reported for full-length cdc25C (40), indicative of a similar substrate-enzyme interaction. In contrast, the affinity of MP51 for preformed cdk/cyclin complexes is relatively low compared to that of full-length cdc25C (40), with Kd values of 1.2 µM for cdk2/cyclin A and 0.8 µM for cdk1/cyclin B1 (data not shown). These results indicate that although MP51 peptide is able to interact with monomeric cyclins and monomeric cdks in vitro, it is unable to interact cooperatively with both components in the cdk-cyclin complex, in the absence of other domains of full-length cdc25C. The micromolar affinity measured between MP51 and cdk/cyclin complexes suggests a typical transient substrate-enzyme interaction, as previously determined for histone H1 (44). In support of this hypothesis, we were indeed able to phosphorylate MP51 on Ser 20 using active cyclin-dependent kinase immunoprecipitated from cell extracts (data not shown). respectively. As shown in Fig.2c, when MP51/cyclinB1 complexes were incubated with 14-3-3ζ, subsequent chromatographic separation yielded two main peaks, the one containing 14-3-3ζ, the other MP51/ cyclinB1 complexes. When the same experiment was performed with purified MP51P/cyclin B1 complexes, 14-3-3ζ disrupted this interaction and displaced most of MP51P from the cyclin, to form highly stable MP51P/14-3-3ζ complexes. These results clearly show that phosphorylation of Ser 22 favours the interaction between MP51 and 14-3-3ζ over that with cyclin B1, as expected from the respective Kd values, and is essential for displacing MP51 from cyclins to 14-3-3 proteins. Moreover, they reveal that phospho-Ser 22 is still accessible for 14-3-3ζ when MP51P is bound to cyclin B1. In contrast, when the converse experiments were performed with either MP51 or MP51P bound to 14-3-3ζ, cyclin B1 (even in five-fold excess) was unable to displace them, suggesting that most of the peptide is masked and inaccessible for cyclin B1 once bound to 14-3-3ζ (data not shown). To address the relevance of these results with respect to fulllength cdc25C, we performed essentially the same experiments using full-length cdc25C phosphorylated by cds1. As shown in Fig.2d, addition of a five-fold excess of cyclin B1 was not sufficient to promote the disruption of highly stable pcdc25C/14-3-3ζ complexes. In contrast, addition of a two-fold excess of 14-3-3ζ lead to the displacement of cyclin B1 from preformed pcdc25C/cyclin B1 complexes to produce pcdc25C/14-3-3ζ complexes. These data indicate that full-length cdc25C behaves essentially like MP51 in its ability to associate differentially with both

Physiological binding of MP51 and cdc25C to 14-3-3 and cdc2-cyclin B -To address the
physiological relevance of the interactions described above, we examined the ability of MP51 and of N-ter cdc25C to bind 14-3-3 and cdk-cyclin complexes from total extracts of human HS68 fibroblasts. MP51 and MP51P were crosslinked to epoxy-activated Sepharose resin, whilst unphosphorylated and phosphorylated GST-Nter cdc25C were coupled to GST Sepharose resin. We first verified the ability of these beads to retain either recombinant 14-3-3, cyclins or cdks. As shown in Fig. 3a, 14-3-3ζ was preferentially retained by the phosphorylated forms of MP51 and Nter cdc25C, and bound the unphosphorylated beads with a lower affinity. In contrast, monomeric recombinant cyclins A and B were similarly retained by both phosphorylated and unphosphorylated beads. Finally, in the absence of cyclins, monomeric recombinant cdc2 was not retained by any of these beads, consistent with the idea that the interaction between cdc25C and cdks is of a weak enzyme-substrate type, and requires the presence of cyclin for stabilization. In all cases, the interactions observed were consistent with the Kd values obtained by fluorescence spectroscopy.
We next examined whether a differential binding behaviour could be observed between endogenous 14-3-3 and cdk-cyclin complexes upon incubation of the phosphorylated or unphosphorylated beads with total cell lysates of HS68 fibroblasts. As shown in Fig. 3b, 14-3-3ε bound exclusively to phosphorylated N-ter cdc25C, and preferentially to MP51P, although a residual fraction also bound to MP51. In contrast, cyclin B and cdc2 were preferentially retained by the unphosphorylated beads. A variable, residual fraction of cyclin B could also be detected associated to the phosphorylated beads, suggesting that all 14-3-3ε had been titrated out of the extract and that an excess of cdc25C protein or peptide was available to bind some cyclin B.
Interestingly, although cdc2-cyclin B complexes were specifically retained by MP51 and N-ter cdc25C, cdk2-cyclin A complexes were not, reflecting the in vivo specificity of the cdc25C isoform for mitotic cdc2-cyclin B complexes. These results clearly demonstrate that in vivo, phosphorylation of cdc25C on Ser 216 favours the interaction between cdc25C and 14-3-3 over that with cdc2-cyclin B complexes. The structure of MP51 and MP51P were determined by circular dichroïsm and NMR. We have previously shown that MP51 does not adopt any typical secondary folds in aqueous solution, but instead multiconformational states (31). Moreover we did not observe any detectable changes in the α-helical content of MP51 following stoichiometric phosphorylation on Ser 22 by recombinant cds1 (Fig 4c). However, addition of increasing concentrations of trifluoroethanol (TFE) stabilized αhelical structures in both MP51 and MP51P, as revealed by the characteristic CD spectra with two minima, at 208 and 222 nm, associated with a maximum at 193 nm ( Fig. 4b and c). TFE induced conformational transitions which stabilized in a final plateau, with a first transition at a concentration of 30%, characterized by an α-helical content of about 50%, essentially as expected from the secondary structure predictions. That the peptide is ordered in α-helices when the hydrophobicity of its environment is increased is a good indication that it may fold when surrounded by the rest of cdc25C, or when interacting with partner proteins as has been described for several proteins involved in important cellular functions (45). To address this possibility, we examined the structural status of MP51 and MP51P bound to 14-3-3ζ by CD. As shown in Fig 4d, the spectrum of 14-3-3ζ is characteristic of a structured protein with a high α-helical content . The Unfortunately no inter-helix NOEs could be detected, probably because of the presence of organic solvent which precludes hydrophobic interactions within the peptide. Given that TFE has been shown to stabilize α-helices in peptides which have an intrinsic potential to form αhelices, but not to induce new, non-existant α-helical structures (47), and that MP51 adopts an α−helical structure when bound to 14-3-3ζ, we propose that similar helical domains are likely to exist in full-length human cdc25C. Essentially the same structural data were obtained for MP51P, including the trans conformation found for all residues in MP51.

Structural characterization of MP51 -
Based on the data obtained by NMR and by secondary structure predictions, a three-dimensional model of the structure of MP51 was elaborated (Fig. 6a). This model highlights the 2 α-helices determined by NMR (α1 from Leu 5 to Tyr 18 ; α2 from Arg 31 to Ile 43 ), as well as phosphoSer 22 and Pro 24 located in the interconnecting loop. Residues 33 to 37 and 46 to 50 of MP51, corresponding to the bipartite NLS of human cdc25C, are located in part in the amphipathic α2-helix and in part in the unfolded C-terminus of the peptide. Interestingly, projection of α2-helix shows that the lysine residues in the first moiety of the NLS ( Lys 33 and Lys 37 ), are located on the same side of the helix, whereas hydrophobic residues (Leu 32 , Val 35 , Ile 43 ) are located on the other. In addition, α2-helix is located between two proline residues, Pro 30 and Pro 44 , which appear to induce a breakage in the peptide structure, thereby positioning all the residues of the NLS on one same side of the structure, as shown in the surface representation in Fig.6b. Interestingly, alignment of the primary sequence and secondary structure prediction of the three human cdc25 isoforms suggests that cdc25B and cdc25C, but not cdc25A, share this common structural organization in the 14-3-3 binding domain, with two α-helices interconnected by a loop carrying the canonical 14-3-3 binding site ( Fig. 6c and   d). shown to play a critical role in cell cycle progression and checkpoint control and it is now common knowledge that cdc25 proteins interact with 14-3-3 isoforms both in vitro and in vivo (26 -31). In this study, we have shown that the MP51 domain of cdc25C interacts and forms stable complexes with 14-3-3ζ in vitro and in vivo, in the absence of phosphorylation, suggesting that residues other than phosphoSer 22 are involved in this interaction, although phosphorylation of Ser 22 dramatically increases its stability, consistent with previous in vitro studies with Raf-derived peptides (25,48,49). In vivo, however, mutation of this serine to alanine completely prevents coimmunoprecipitation of 14-3-3 proteins with cdc25 phosphatases from cell extracts (26 -30) and in vitro we similarly observe a very weak interaction between 14-3-3ζ and unphosphorylated recombinant N-ter or full-length cdc25C, whereas the phosphorylated counterparts form highly stable complexes. This discrepancy most likely reflects conformational constraints present in the full-length protein, which impede formation of stable 14-3-3/cdc25C complexes, in the absence of a negatively charged phosphate group on Ser 216 .
MP51 also interacts with cyclins A and B, with an affinity similar to that for 14-3-3ζ. We have shown that this interaction is is primarily mediated through the P-box motif of cyclins, providing the first direct biochemical evidence of an interaction between cdc25 phosphatases and the P-box of cyclins. This interaction was not affected by phosphorylation of Ser 22 , suggesting that the residues in MP51 involved in this interaction are different from those which interact with 14-3-3ζ. However, phosphorylation of MP51, of N-ter, and of full-length cdc25C clearly favours their interaction with 14-3-3 over that with cyclins both in vitro and in vivo. This suggests that phosphorylation induces conformational changes in the MP51 domain of cdc25C which favour 14-3-3 binding, which in turn renders the residues involved in interactions with cyclin B1 inaccessible. From these data, we conclude that MP51 is a bi-functional domain of human cdc25C, capable on the one hand of interacting with 14-3-3 proteins, through which control of subcellular localization is exerted, and capable of interacting with cyclins, on the other hand, thereby targeting cdc25C to cyclin-dependent kinases and promoting its mitotic function.
Structural characterization of MP51 -Cdc25 phosphatases are composed of a highly conserved C-terminal catalytic domain, and an N-terminal regulatory domain, specific to each cdc25 isoform.
Although the structures of the C-terminal domain of human cdc25A and cdc25B have been solved (50,51), no structural information is available yet for the N-terminal domain of these phosphatases.
Based on NMR data and on structural predictions, we have elaborated a structural model of MP51, which constitutes the first structural data of a subdomain of the N-terminus of human cdc25C.
MP51 can be described as an elbow-shaped entity formed by two independent α-helical domains, separated by a non-structured intervening loop which carries the two critical phosphorylation sites Ser 20 and Ser 22 (corresponding to Ser 214 and Ser 216 in human cdc25C), as well as the consensus 14-3-3 binding site. This structure is induced upon binding of 14-3-3ζ , and can be mimicked by a hydrophobic environment, using organic solvents. Our data reveal that residues of the bipartite NLS of human cdc25C, are located in part in the amphipathic α2-helix and in part in the unfolded Cterminus of the peptide, and are all located on the same side of the structure of MP51. In vivo, this NLS is therefore presumably accessible for binding to the α-importin receptor. In contrast, using the structure of 14-3-3/peptide complexes (25,48,49) as a template for docking MP51 to 14-3-3, the α2-helix of MP51 appears to be buried inside the cleft of the 14-3-3 dimer, thus rendering the NLS completely inaccessible for recognition by the nuclear import machinery, consistent with the recent observation by Kumagai and Dunphy (28) that 14-3-3 binding reduces the ability of Xenopus cdc25 to associate with α-importin.
Phosphorylation of MP51 on Ser 22 did not induce any dramatic modifications in its overall structure. However, this phosphorylation significantly increased the affinity of MP51 for 14-3-3 ζ and enabled 14-3-3 ζ to displace MP51P from preformed MP51P/cyclin B1 complexes, suggesting that some conformational changes must occur. In Raf-1-derived phosphopeptides complexed with 14-3-3, Pro +2 in the RSXSXP motif (in bold) adopts a cis conformation, which reinforces the interaction with 14-3-3 (25,48). NMR analysis of MP51 and MP51P reveal that in both peptides all the residues are in trans, indicating that phosphorylation of Ser 22 is not responsible for an induced, stabilizing isomerization of Pro 24 . Hence, we propose that if such an isomerization takes place in this mode 1 phosphopeptide, it must be induced by 14-3-3 binding. phosphorylation of Ser 216 is required for binding of 14-3-3ζ to cdc25C, presumably by disrupting sterical hindrances in the full-length protein, but additionally by inducing a stabilizing isomerization of Pro 218 . High affinity binding of 14-3-3 to cdc25C would prevent mitotic activation of this phosphatase in a dual fashion: on the one hand by preventing the nuclear localization of cdc25C, by burying its NLS within the channel of the 14-3-3 dimer ; on the other hand by precluding interactions with other proteins, in particular cyclins. We propose that in vivo, during interphase, a large fraction of cdc25C is phosphorylated on Ser 216 by C-TAK1 and rapidly complexed to 14-3-3, which masks the NLS as well as residues required for interaction with cyclin B, and thus prevents its nuclear localization and cyclin binding. As Ser 216 is deeply buried within the 14-3-3 dimer, we suggest that dissociation of 14-3-3, and not phosphoSer 216 dephosphorylation, is the first step in the mechanism of cdc25C activation at mitosis. This would on the one hand render the NLS accessible for recognition by the nuclear import machinery, and on the other enable interactions with cdc2cyclin B complexes. Dephosphorylation of Ser 216 is probably required before cyclin B1 binds cdc25C, so as to prevent competitive, re-binding of 14-3-3. What promotes disruption of the highly stable 14-3-3/cdc25C complex still remains to be determined. Our data clearly exclude the possibility of cyclins displacing 14-3-3 from cdc25C. As the in vivo activation of cdc25C at the G2/M transition is dependent on phosphorylation at mitotic sites, we suggest that disruption of the 14-3-3/cdc25C complex may be regulated by this phosphorylation.